Belg. J. Zool., 135 (supplement) : 3
December 2005
PREFACE
The present issue of the Belgian Journal of Zoology contains the proceedings
of the 9th International African Small Mammal Symposium (ASMS),
held at the Sokoine University of Agriculture, Morogoro,
Tanzania from 14-18 July 2003
The African Small Mammal Symposia are organised
We apologise for the long delays, we thank all the authors
every fourth year and this meeting was the first to be held
for their patience and hope that the result is satisfying.
in tropical Africa. There were 114 participants from 31
countries and more than half of the attending scientists
During the final preparations of these proceedings, we
were from Africa. This, as well as the many contributions
lost one of the great people in African small mammal
from Africa in this proceedings issue, are clear indica-
biology. Em. Prof. dr. Walter N. VERHEYEN, from the Uni-
tions of the growth of the community of African small
versity of Antwerp, Belgium died in December 2005 at
mammal scientists. The large attendance to the meeting,
the age of 73 years. He was a well known specialist of
and the publication of these proceedings, was made possi-
African rodent taxonomy and through several projects the
ble by the much appreciated financial support from the
driving force behind rodent research projects, in Tanzania
Flemish Interuniversity Council (VLIR-UOS) and the
and elsewhere. He was also one of the co-organisers of
European Commission (STAPLERAT-ASMS ICA4-CT-
the first African small mammal meeting. Although he had
2002-50029). Together, they contributed to travel and
been retired for 7 years, he continued to be active at the
participation fees for 49 scientists and students from
laboratory every day, writing papers, guiding students and
Africa and Europe.
colleagues and collaborating with foreign visitors who
came to benefit from his knowledge. He was a mentor and
The meeting came towards the end of the EC-sup-
friend to many of us. We dedicate these proceedings to his
ported project "STAPLERAT", on the biology and man-
memory.
agement of rodent problems in staple crops in Africa. Ear-
lier in 2003, another project "RATZOOMAN", also
Herwig LEIRS, Rhodes H. MAKUNDI & Stephen DAVIS
funded by the EC had begun to study the role of rodents
as carriers of pathogens in and around growing cities in
Africa. Both projects demonstrate the heavy burden
GUEST EDITORS
placed by rodents on the food security and health of
human communities in Africa. An ecologically-based
We appreciated the help of the reviewers listed below
approach to management, which is highly desirable,
that assessed and improved the papers in these proceed-
requires a thorough knowledge of the animals basic biol-
ings and often spent considerable time correcting manu-
ogy and ecology. Moreover, though some species in some
scripts and interacting with authors that were not yet so
parts of their range are considered pests, many rodents
experienced in publishing scientific papers in English.
and other small mammals are valuable parts of biodiver-
sity. The range of topics and species covered by the con-
L.ARNTZEN, A.ASSOGBADJO, N.AVENANT, P.BARRIERE,
tributions to this issue illustrates the variety of research
A.BEKELE, S.BELMAIN, N.BENNETT, G.BRONNER,
that is carried out.
P.BROWN, C.CHIMIMBA, M.CORTI, J.F.COSSON, C.DENYS,
The publication of these proceedings was a lengthy
J.M.DUPLANTIER, J.FAHR, C.FAULKES, L.FIEDLER,
project, stalled now and then by slow communication
L.GRANJON, D.HAPPOLD, M.HAPPOLD, B.S.KILONZO,
with authors in places where access to the internet still is
B.KRASNOV, E.LECOMPTE, J.LEHTONEN, H.LEIRS,
not that common, changing people that assisted in keep-
A.LINZEY, J.LODAL, R.MAKUNDI, A.MASSAWE, A.MONAD-
ing track of all papers and the usual burden of other com-
JEM, S.MORAN, V.NICOLAS, N.OGUGE, M.PERRIN,
mitments. We appreciated very much the assistance of
B.SICARD, G.SINGLETON, W.STANLEY, P.TAYLOR, H.TRIS-
N.WOUTERS, V.SLUYDTS and A.VLAEMINCK in the process.
TIANI, S.VIBE-PETERSEN, P.WITHERS


Belg. J. Zool., 135 (supplement) : 5-10
December 2005
Comparative and functional morphology of the middle
ear in Zambezian mole-rats (Coetomys ­ Cryptomys,
Bathyergidae)

Simone Lange and Hynek Burda
Department of General Zoology, Faculty of Biology and Geography(FB 9), University of Duisburg Essen, D-45117 Essen, Germany
Corresponding author : Simone Lange, e-mail : lange_simone@web.de
ABSTRACT. Within the family Bathyergidae, the genus Coetomys (Cryptomys) is the most speciose. However, an
unambiguous morphological or morphometrical species diagnosis has not been feasible to date. The middle ear
structures involved in sound transmission were examined and measured in six species of Zambezian mole-rats of
varying body sizes : Coetomys amatus, C. anselli, C. darlingi, C. kafuensis, C. mechowi and C. whytei. Compari-
sons revealed many differences in the shape of the middle ear ossicles allowing species differentiation. Bullar vol-
ume, eardrum area and cross-sectional area of the bony meatus were positively correlated with body size (as repre-
sented by condylo-basal length) whereas the size of middle ear ossicles was rather independent of body size. Middle
ears shared typical features with those found in low-frequency hearing mammals. Contrary to the findings in heter-
omyids and the Spalax ehrenbergi species complex, within the genus Coetomys, those species occupying more
mesic habitats had more efficient tympano-ossicular systems (suggesting more sensitive hearing) than species from
drier habitats.
KEY WORDS : middle ear, functional morphology, Coetomys, Cryptomys, species diagnosis.
INTRODUCTION
functional aspects of the middle ear are considered to
reflect adaptations to the species' environment (e. g. WEB-
Among the five genera of the Bathyergidae (African
STER & WEBSTER, 1975). The study of the ear has a great
mole-rats), representatives of the genus Coetomys (Cryp-
potential for both comparative taxonomic and functional
tomys) can be easily recognized, yet to date it is not possi-
interpretations.
ble to determine different species within the genus by
Taking these facts into account we decided to study the
classical morpho-taxonomic traits like skull, pelage or
middle ear structures in Coetomys species and to test their
size. The genus Coetomys is the most speciose among the
diagnostic value. Furthermore, since the selected Zam-
bathyergid genera. At least 16 species distributed from
bezian species also represent forms occupying different
West through Central to South Africa have been described
climatic regions, they provide unique opportunity to test
in literature (cf., INGRAM et al., 2004). Most of the species
general applicability of conclusions derived from the
to date have been identified within the Zambezian phyto-
study of the ear in blind mole-rats (see above).
chorion. All the species have been identified on the basis
of karyological, allozyme, or DNA analyses.
The existence of cryptic ­ yet morphologically practi-
MATERIAL AND METHODS
cally indistinguishable ­ sibling species parallels the find-
ings in some other subterranean rodents like blind mole-
Middle ears of six species of African mole-rats of vary-
rats (Spalax) (NEVO et al., 2001). Nevertheless, various
ing body sizes (between 60 g in Coetomys amatus and
structures and parameters of the ear in the Spalax ehren-
300 g in C. mechowi) representing different clades of
bergi species complex have proved to be of use for mor-
Coetomys were examined. (Coetomys represents a newly
phological species diagnosis (BURDA et al., 1990) sup-
described genus, INGRAM et al., 2004, which has been pre-
porting thus also the experience of palaeontologists and
viously referred to as Cryptomys). They all originate from
taxonomists as reflected by LAVOCAT & PARENT (1985)
the Zambezian phytochorion : Coetomys kafuensis (adults
who noted that the auditory region serves as an excellent
N = 11, juveniles N = 1) from Itezhi-Tezhi (Zambia), C.
guide for following the evolution of the dentition. More-
anselli (adults N = 9, juveniles N = 1) from Lusaka and
over, it has been shown that in the blind mole-rat in par-
surroundings (Zambia), C. amatus (adults N = 1) from
ticular (like in other subterranean mammals in general),
Chibale (Zambia), C. darlingi (adults N = 2) from Chi-
middle ear morphology reflects also the species' habitat
manimani (Zimbabwe), C. mechowi, (adults N = 10, juve-
and way of life (BURDA et al., 1989, 1990, 1992). VON
niles N = 1) from the Copperbelt province (Zambia) and
BEKESY (1974) stated that the physical laws served as
C. whytei (adults N = 5, juveniles N = 3) from Karonga
guidelines for the evolution of the structures and func-
(Malawi). All the specimens were preserved in 70%-etha-
tions of the middle and inner ear. Additionally, morpho-
nol for at least 4 weeks.

6
Simone Lange and Hynek Burda
Comparative morphometry
the difference in the area of the eardrum and the stapedial
footplate leading to the area ratio and the difference in the
Condylo-basal length (anterior face of the upper inci-
length of mallear and incudal levers leading to the lever
sors to the posterior edge of the occipital condyles) and
ratio (R
bullar dimensions (length of the longest axis of the bulla,
ELKIN, 1988). The product of both ratios (transfor-
mation ratio) expresses the middle ear efficiency in sound
width and height from the top of the auditory meatus to
transmission and thus middle ear sensitivity (F
the bottom of the bulla tympanica) were measured with
LEISCHER,
1978).
digital callipers. Middle ear structures were prepared,
examined under a stereoscopic binocular and drawn using
a drawing tubus at different magnifications. Measure-
Climatic data
ments were taken from drawings considering the respec-
tive magnifications (15x ­ 40x). The following variables
The mean annual precipitation was calculated from
were measured : longer and shorter radius (perpendicular
monthly values from about an average of 55 years recorded
to the longest radius) of the eardrum, longer and shorter
by the nearest climatological station according to the Glo-
radius of the cross-section of the bony meatus, length of
bal Historical Climatological Network database : <http ://
the mallear lever, length of the incudal lever, longer and
www.ncdc.noaa.gov/ol/climate/research/ghcn/ghcn.html>.
shorter radius of the stapedial footplate. Levers were
measured perpendicular to the axis of rotation from the
RESULTS
axis of rotation to the umbo of the eardrum and to the
point of action of the incus. The eardrum, meatus and sta-
pedial footplate area were calculated as ellipses (p x
Comparative morphology
longer radius x shorter radius), bullar volume was calcu-
In the studied species, manubrium mallei and crus
lated as the product of length x height x width. Bullar vol-
longum of the incus were rather parallel to each another
ume, eardrum area, meatus area, mallear lever, incudal
(Fig. 1). The eardrum was nearly circular, no pars flaccida
lever and stapedial footplate area were correlated with
was apparent. Middle ear ossicles were not fused with the
condylo-basal length (linear regression with ANOVA,
tympanic bone, so the middle ears were of the "freely
SPSS 11.0).
mobile" type. Malleus and incus were fused whereas the
Functional morphology
incudo-stapedial joint was rather loose and the malleo-
incudal complex separated easily from the stapes. A
Movements of the eardrum are transmitted to the mal-
gonial was missing, the footplate area was quite large and
lear-incudal complex and then to the stapes which is
the middle ear (stapedial and tensor tympani) muscles
attached to the vestibular window of the cochlea. Through
were reduced or missing. The anterior and posterior crura
the arrangement of this ossicular chain, pressure and force
of the stapes were asymmetric, no stapedial artery was
are amplified. This is accomplished by two mechanisms :
found.
Fig. 1. ­ Shape of the middle ear ossicles (left - anterior, right ­ posterior).

Middle ear in mole-rats
7
The shape of the middle ear ossicles (Fig. 1) was rather
species. Within particular species, no differences were
similar; nevertheless some species-specific differences
found in shape of middle ear ossicles between adult and
were identified. Thus, the head of the malleus was partic-
juvenile specimens.
ularly massive in C. whytei. The transitional part between
manubrium and head of the malleus where the tensor
Comparative morphometry
tympani muscle attaches was rather straight in C. whytei
and C. amatus. In the other species, there was a processus
The mean values of particular parameters (Table 1) do
which was more (C. anselli) or less (C kafuensis.) con-
not include juveniles. Bullar volume, eardrum and meatus
spicuous. In C. kafuensis and C. darlingi, crus stapedis
area tended to be correlated with condylo-basal length.
posterior was broader than crus stapedis anterior whereas
The stapedial footplate area was absolutely and relatively
the opposite was true in C. anselli, C. whytei, C. amatus
noticeably small in C. amatus, C. darlingi and C. whytei.
and C. mechowi. The attachment of the stapedial muscle
Mallear and incudal lever seemed to be independent of
was clearly differently located and shaped in the studied
body size.
TABLE 1
Measurements (mean values and standard deviations) of middle ear structures. in adult specimens. (LCB = condylo-basal length, Bulla
= bullar volume, AM = cross-sectional area of bony meatus, MT = area of membrana tympani, BS = area of basis stapedis, ML = mal-
lear lever, IL = incudal lever)
LCB
Bulla
AM
MT
BS
ML
IL
N
(mm)
(mm³)
(mm²]
(mm²)
(mm²)
(mm)
(mm)
C. kafuensis
11
27.8 ±1.8
321 ± 46
1.0 ± 0.2
6.8 ± 1.4
0.59 ± 0.28
2.1 ± 0.1
1.1 ± 0.2
C. anselli
9
33.1 ± 1.9
380 ± 68
0.9 ± 0.3
8.6 ± 1.1
0.53 ± 0.10
2.1 ± 0.3
1.1 ± 0.1
C. amatus
1
32.5
208
0.5
8.9
0.46
1.9
1.0
C. darlingi
2
36.2 ± 3.9
331 ± 11
0.7 ± 0.5
9.7 ± 5.8
0.48 ± 0.3
2.0 ± 0.1
1.0 ± 0.1
C. mechowi
10
45.5 ± 6.4
913 ± 302
1.4 ± 0.3
14.1 ± 2.4
0.62 ± 0.08
2.5 ± 0.2
1.3 ± 0,1
C. whytei
5
36.9 ± 1.8
469 ± 110
0.9 ± 0.1
11.6 ± 1.4
0.50 ± 0.04
2.4 ± 0.1
1.0 ± 0,1
When juvenile specimens were included into the analy-
Functional morphology
sis, volume of the tympanic cavity was strongly positively
correlated with the condylo-basal length (R² = 0.797, p <
The calculated functional parameters (Table 2) could
0.001, ANOVA). Eardrum area and cross-sectional area
of the bony meatus were less dependent upon the body
be related to the mean annual precipitation in the area of
size (R² = 0.299, p < 0.001, ANOVA and R² = 0.464, p <
occurrence of the particular species : With increasing
0.001, ANOVA) (Fig. 2). The length of mallear and incu-
rainfall the transformation ratio (expressing middle ear
dal levers (R² = 0.148, p = 0.01 and R² = 0.008, p = 0.557,
sensitivity) increased from 25 in C. kafuensis occurring in
ANOVA) and the area of the stapedial footplate (R² =
the driest habitat through C. anselli (33), C. amatus (37),
0.205, p < 0.005, ANOVA) were rather independent of
C. darlingi (43), C. mechowi (42) to 52 in C. whytei living
the body size (Fig. 3).
in the most mesic habitat.
TABLE 2
Precipitation in the area of occurrence and morpho-functional parameters (means and standard deviations) of the middle ear in adult
Coetomys mole-rats of different species. (LCB = condylo-basal length AR = area ratio, LR = lever ratio, TR = transformation ratio
(AR x LR))
annual precipitation
LCB
N
AR
LR
TR
(mm/year)
(mm)
C. kafuensis
11
787
27.8 ±1.8
13.1 ± 2.2
1.3 ± 0.3
25.2 ± 6.1
C. anselli
9
822
33.1 ± 1.9
16.4 ± 4.4
2.0 ±0.4
33.2 ± 10.1
C. amatus
1
1132
32.5
19.3
1.9
36.8
C. darlingi
2
1150
36.2 ± 3.9
20.3 ± 2.7
2.1 ± 0.0
42.7 ± 5.4
C. mechowi
10
1185
45.5 ± 6.4
21.0 ± 4.6
2.0 ± 0.3
42.3 ± 14.4
C. whytei
5
1603
36.9 ± 1.8
23.2 ± 2.0
2.2 ± 0.3
52.2 ± 7.7

8
Simone Lange and Hynek Burda
Fig. 2. ­ Linear regression of condylo-basal length and size of middle ear structures : bullar
volume, meatus and eardrum area.

Middle ear in mole-rats
9
Fig. 3. ­ Linear regression of condylo-basal length and size of the middle ear ossicles : stapedial footplate area, mallear and
incudal lever.
DISCUSSION
species-specific trait with minimum individual variability
(BURDA, 1979). Actually, in some cases it is not necessary
Comparative morphology
to consider middle ear morphology if the species can be
easily distinguished on the basis of the geographic origin
Morphology of the middle ear ossicles can be applied
and/or the general body size.
to enable species diagnosis. Although the sample sizes in
C. amatus and C. darlingi are very small, there is good
Since the morphology and morphometry of the middle
reason to assume that morphology described on the basis
ear ossicles (which derive from the endocranium) is
of just one or few individuals is representative of a spe-
almost adult-like at birth, even juvenile specimens can be
cies. Middle ear morphology is known to be a reliable
reliably determined.

10
Simone Lange and Hynek Burda
Freely mobile middle ears are considered as a typical
BURDA, H., V. BRUNS & G.C. HICKMAN (1992). The ear in sub-
trait characterizing low-frequency hearing mammals. The
terranean insectivores and rodentia in comparison with
studied species shared with low-frequency hearing forms
ground-dwelling representatives. I. Sound conducting sys-
also other traits like fused ossicles, lacking gonial, large
tem of the middle ear. J. Morph., 214 : 49-61.
stapedial footplate and reduced middle ear muscles (cf.,
BURDA, H., V. BRUNS & E. NEVO (1989). Middle ear and coch-
B
lear receptors in the subterranean mole-rat, Spalax ehren-
URDA et al., 1992). These traits are considered as adapta-
bergi. Hear. Res., 39 : 225-230.
tions to optimize low-frequency hearing (MASON, 2001;
BURDA, H., E. NEVO & V. BRUNS (1990). Adaptive differentia-
FLEISCHER, 1978, 1973) in underground burrows where
tion of ear structures in subterranean mole-rats of the Spalax
low frequencies of around 500 Hz are best transmitted
ehrenbergi superspecies in Israel. Zool. Jb. Syst., 117 : 369-
(HETH et al., 1986) whereas lower and higher frequencies
382.
are absorbed more rapidly.
FLEISCHER, G. (1973). Studien am Skelett des Gehörorgans der
Säugetiere, einschließlich des Menschen. Säugetierkundl.
Functional morphology
Mitt., 21 : 131-239.
FLEISCHER, G. (1978). Evolutionary principles of the mammalian
In the studied species, the effectiveness of the middle
middle ear. Adv. Anat. Embryol. Cel. Biol., 55 : 1-70.
ear in sound transmission increased with increasing
HETH, G., E. FRANKENBERG & E. NEVO (1986). Adaptive optimal
humidity of their particular habitats. This relationship is
sound for vocal communication in tunnels of a subterranean
surprising because in Spalax ehrenbergi (BURDA et al.,
mammal (Spalax ehrenbergi). Experientia, 42 : 1287-1289.
1990) and several heteromyids (WEBSTER & WEBSTER,
INGRAM, C.M., H. BURDA & R.L. HONEYCUTT (2004). Molecular
1980), the opposite correlation was found and interpreted
phylogenetics and taxonomy of the African mole-rats, genus
as adaptation to the environment. Sound transmission is
Cryptomys and the new genus Coetomys Gray, 1864. Molec-
influenced by several factors from which relative humid-
ular Phylogeny and Evolution, 31 : 997-1014.
L
ity is one (B
AVOCAT, R. & J.P. PARENT (1985). Phylogenetic analysis of
ASS et al., 1995) : with increasing humidity
middle ear features in fossil and living rodents. NATO ASI
attenuation first increases and after a turning point it
Series A, Life Sciences 92 : 333-354.
decreases again. Spalax and heteromyids are adapted to
MASON, M.J. (2001). Middle ear structures in fossorial
the attenuation which increases with increasing aridity in
mammals : a comparison with non-fossorial species. J.
their habitat. At least for heteromyids, acoustic communi-
Zool., Lond., 255 : 467-486.
cation over long distances and predator avoidance is of
NEVO, E., E. IVANITSKAYA.& A. BEILES (2001). Adaptive radia-
great importance (WEBSTER & WEBSTER, 1980, 1992).
tion of blind subterranean Mole Rats. Backhuys Publishers,
Leiden.
RELKIN, E.M. (1988). Introduction to the analysis of middle ear
REFERENCES
function. In : JAHN & SANTOS-SACCHI (eds), Physiology of
the ear
, Raven Press, New York : 103-123.
BASS, H.E., L.C. SUTHERLAND, A J. ZUCKERWAR, D.T. BLACK-
WEBSTER, D.B. & M. WEBSTER (1975). Auditory system in
STOCK & D.M. HESTER (1995). Atmospheric absorptions of
Heteromyidae : functional morphology and evolution of the
sound : further developments. J. Acoust. Soc. Am., 97 : 680-
middle ear. J. Morph., 146 : 343-376.
683.
WEBSTER, D.B. & M. WEBSTER (1980). Morphological adapta-
BEKESY, G. V. (1974). Introduction. In : KEIDEL & NEFF (eds),
tions of the ear in the rodent family Heteromyidae. Amer.
Handbook of sensory physiology. Vol. V/1, Auditory system,
Zool., 20 : 247-252.
anatomy physiology (ear), Springer, Berlin : 1-8.
WEBSTER, D.B. & M. WEBSTER (1992). Parallel evolution of
BURDA, H. (1979) : Morphology of the middle and inner ear in
low-frequency sensitivity in old world and new world desert
some species of shrews (Insectivora, Soricidae). Acta Sc.
rodents. In : WEBSTER, FAY & POPPER (eds), Evolutionary
Nat. Brno, 13(4) : 1-48.
biology of hearing, Springer, Berlin : 633-636.

Belg. J. Zool., 135 (supplement) : 11-15
December 2005
Importance of rodents as a human food source in Benin
A.E. Assogbadjo, J.T.C. Codjia, B. Sinsin, M.R.M. Ekue and G.A. Mensah
Faculté des Sciences Agronomiques, Université d'Abomey-Calavi, 05 BP 1752 Cotonou (Akpakpa-Centre), Benin
Corresponding author : A.E. Assogbadjo, e-mail : assogbadjo@yahoo.fr
ABSTRACT. Rodents are an important food source for villagers near the Lama forest reserve, located in the south
of Benin between 6°55 - 7°00N and 2°04 - 2°12 E. This study was designed to look at the consumption of rodents as
a food source combined with a survey of rodents sold in markets. Data was collected on : rodents species consumed,
frequencies of consumption and food preferences. Some animals were captured in order to confirm the species.
Rodents were a major part of diet included 10 species : grasscutter (Thryonomys swinderianus), giant rats (Criceto-
mys gambianus
), Gambian Sun-squirrel (Heliosciurus gambianus), crested porcupine (Hystrix cristata), ground
squirrel (Xerus erythropus), grass rat (Arvicanthis niloticus), slender gerbil (Taterillus gracilis), Kempi's gerbil
(Tatera kempii), multimammate rats (Mastomys spp.) and grass mouse (Lemniscomys striatus venustus). On aver-
age, young people and children consumed rodents 6 times per person per month. The preferences of local popula-
tions were grasscutter and giant rats which were sold in local markets at relatively high prices US$8-10 and US$2-4
respectively. It is important to conduct further studies to look at the impact of this hunting on the rodent populations
and to ensure sustainable harvesting.
KEY WORDS : Rodents, Human consumption, Lama forest, Benin.
INTRODUCTION
composed of about 173 plant species and belongs mainly
to the soudano-guinean and guineo-congolian flora.
Little attention has been given to the beneficial effects
Accordingly, the natural vegetation of the forest is charac-
of rodents to human food security (MENSAH, 1991; JORI et
terized as dense humid semi-deciduous forest. In spite of
al., 1994; HANOTTE & MENSAH, 2002). In Africa, rodents
intensive poaching, it contains a rich and fairly abundant
are a significant source of animal protein for humans,
fauna that is maintained by protection activities.
especially in tropical Africa (AJAYI & OLAWOYE, 1974;
MALEKANI & PAULUS, 1989; FALCONER, 1996; MALAISSE,
1997; NTIAMOA-BAÏDU, 1998). In Benin, there have been
MATERIAL AND METHODS
few studies conducted to show how important rodents are
in helping to ensure the food security of the populations
This study comprised two phases : (1) food consump-
(BAPTIST & MENSAH, 1986; CODJIA & HEYMANS, 1988;
tion and socio-economic investigation in the bordering
HEYMANS & CODJIA, 1988; MENSAH, 1991; ASSOGBADJO,
villages and (2) the captures of rodents in various vegeta-
2000). Therefore, a better understanding of the way
tion groups and villagers' farms. A total of 126 villagers
rodents contribute directly to the diet of local populations
were classified into three age classes (young, adult and
is required. This study, carried out in Lama forest reserve
old) and two genders (male or female) (Table 1). Villagers
(Bénin), describes a case study on the consumption of
between 5 to 25 years old were considered as young, an
rodents by forest-bordering human populations.
informant aged 26 to 50 years old were considered as
Study site
adults, and an informant aged above 50 years old were
considered as old. A structured questionnaire was used to
Lama forest reserve is located in south Benin from
interview individuals or a group of informants by com-
6°55' to 7°00' N and between 2°04' and 2°12' E (Fig. 1).
bining retrospective method with direct observations. For
It covers 16,250 ha, including 2,290 ha of dense forest as
examining the relative importance of rodents in the diet,
censused in 1999. The bordering populations of this forest
data were collected from information on the consumption
comprise 20 rural villages with an estimated number of
of other mammals to pair them with these obtained on
41,500 individuals (1998) belonging mainly to the Holli
rodents. Data were collected on the frequency of con-
and Fon ethnic groups. The altitude of the forest averages
sumption and the food preferences of each species of
60 m. Soils are vertisols of a clay-sandy type. The water
mammals (including rodents). Practical handbooks were
network is exclusively composed of ponds and seasonal
used to help the informant in identifying the animal spe-
streams. The climate is a transitional guinean type, falling
cies and also to obtain some useful information on it (DE
between the bimodal and unimodal rainfall distribution.
VISSER et al., 2001; KINGDON, 1997; SINSIN et al., 1997;
The annual average rainfall is 1,112 mm. The annual
HEYMANS, 1986). The frequencies of consumption were
average temperature varies between 25°C and 29°C. Rel-
obtained by averaging the number of times a given spe-
ative humidity remains very high throughout the year,
cies was consumed per week and per informant. Three
even in the dry season. The vegetation of the forest is
level of consumption frequency were defined as :

12
A.E. Assogbadjo, J.T.C. Codjia, B. Sinsin, M.R.M. Ekue and G.A. Mensah
Fig. 1. ­ Location of Lama reserve forest in Benin
· Less consumed species : rodent species consumed by
RESULTS
1 to 25% of the informants
Diversity and habitats of rodents consumed
· Fairly consumed species : rodent species consumed
by bordering populations of Lama forest reserve
by 26 to 50% of the informants
Ten rodent species were consumed by local populations
· Highly consumed species : rodent species consumed
(Table 2) : Thryonomys swinderianus, Cricetomys gambi-
by more than 50% of the informants.
anus, Heliosciurus gambianus, Hystris cristata, Xerus
erythropus, Arvicanthis niloticus, Taterillus gracilis, Tat-
To capture rodents, we used a combination of several
era kempii, Mastomys natalensis and Lemniscomys stria-
methods. We lay during 2 weeks traditional traps cur-
tus venustus. These species belong to 4 rodent families
rently used by local population in four types of
namely Murideae (6 species), Sciurideae (2 species),
vegetation : fallow, dense and degraded forests, planta-
Thryonomideae and Hystricideae (1 species for each).
tions and farms. In addition, we employed indigenous
Although rodents were trapped in different vegetation
hunters to use their traditional rodent hunting techniques.
types, villagers' farms and forests were the preferred hab-
Hunting takes place between 6h and 12h in the morning
itats for most of rodent species consumed (Table 2)
and between 15h and 18h in the afternoon. Regarding
giant rats, villagers dug them from their burrows with a
Hunting techniques for rodents
hoe before killing them and used chasing and bush fires
and other collecting strategies in the study area
methods for other rodents. This enabled us to survey the
different traditional hunting techniques and to understand
The hunting techniques varied according to the type of
how rodents were collected from the wild.
animal, vegetation and season. The most common hunt-
ing techniques were chasing, trapping and using bush
fires, especially at the end of the dry season before the
TABLE 1
land preparation for agriculture. Bush fires were the most
Number of local villagers (informants)
frequently used technique for hunting rats.
who were interviewed for the study
The grasscutter (Thryonomys swinderianus) is the
rodent species most collected by local villagers, due to the
Gender
Young
Adult
Old
Total
quality of its meat and the income that can be gained. Vil-
lagers hunt grasscutters in small groups of young people,
Male
6
25
22
53
Female
3
41
29
73
by lighting bush fires to disturb the animals and flush it
Total
9
66
51
126
from the bush to be chased by dogs. Hunting takes place
between 6h and 12h in the morning and between 15h and
NB : The age classes were defined as : young, 5-25 years old; adults,
18h in the afternoon. Giant rats (Cricetomys gambianus),
26-50 years old; and old, >50 years old.
were dug from their burrows with a hoe.

Importance of rodents as a human food source in Benin
13
TABLE 2
Mammal species consumed by local villagers.
Habitat type
Proportion
Scientific name
Common name
Family
For
Pla
Far
Fal
Consumed
Thryonomys swinderianus
Grasscutter
Tryonomideae
+
+
+
+

> 75%
Hystrix cristata
Crested porcupine
Hystricideae
+
+
+
> 75%
Heliosciurus gambianus
Gambian sun-squirell
Sciurideae
+
+
51-75%
Xerus erythropus
Ground squirell
Sciurideae
+
+
51-75%
Cricetomys gambianus
Giant rat
Murideae
+
+
+
> 75%
Arvicanthis niloticus
Grass rat
Murideae
+
+
> 75%
Taterillus gracilis
Slender gerbil
Murideae
+
+
+
51-75%
Tatera kempi
Kempi's gerbil
Murideae
+
+
51-75%
Mastomys natalensis
Multimammate rat
Murideae
+
51-75%
Lemniscomys striatus venustus
Grass mouse
Murideae
+
+
+
51-75%
Habitat types : For = Forest; Pla = Plantation; Far = Farm; Fal = Fallows.
Grasscutters and giant rats account for most of the
cases sold after they had been captured (Table 3). Apart
2
from these two highly preferred species, other rodent spe-
1,8
cies are hunted by using bush fires, dogs and hunting.
1,6
on per week
1,4
People can buy rodents in local markets for their con-
1,2
Children
per pers
sumption. However, this was uncommon in the study
y
1
Young
area. Hunting is still the main way for villagers to obtain
0,8
Old
r
equenc
0,6
rodents for animal protein. Table 3 outlines the average
n f
i
o
0,4
numbers of rodents killed per week and per hunter. This
umpt
0,2
gives also the sale prices for rodent meats.
0
Cons
k
r
e
ke
cutter
hbuc
dui
TABLE 3
ass
Giant rat
us
h
iver hog
B
crub har
Gr
us
B
S
Red r
Average number of rodents killed per week per hunter, and aver-
Species
age sale prices per individual animal.
Fig. 2. ­ Consumption frequencies for children, young and old
Number of
Sale price in
villagers per individual per week for the most hunted mammals
Species
Part sold
rodents killed
1999 (US $)
in Lama reserve forest according to different age classes.
Grasscutter
4
The whole
8 to 10
Giant rat
10
The whole
2 to 4
Other rodent species
15
Not sold
-
1,8
1,6
on per week
1,4
Consumption frequency for mammal
1,2
1
per pers
Men
and rodent species
y
0,8
Women
0,6
r
equenc
0,4
In villages around Lama forest reserve, any kind of
i
on f
0,2
bush meat is considered as edible by local villagers,
umpt
0
k
e
despite the governmental restrictions on hunting. More
Cons
at
utter
ker
har
r hog
iant r
hbuc
than 75% of the village population ate grasscutter, giant
assc
G
us
rub
B
rive
Gr
Bush dui
Sc
ed
R
rat, grass rat and crested porcupine, while the other rodent
Species
species were consumed by 51-75% of the village popula-
Fig. 3. ­ Consumption frequencies for male and female villag-
tion (Table 2).
ers per individual per week for the most hunted mammals in
Lama reserve forest according to informant gender.
Rodent meats were consumed at least 6 (Fig. 3) times
per month per person. This rate was at least twice the
meat consumption of other mammal species (Figs 2 & 3).
Local populations' consumption preference
The frequency of meat consumption in men is much
for various mammal and rodent species
higher than for women (x² = 1.16, p < 0.05) (Fig. 3), and
More than 53% (made up of giant rat 5%, grasscutter
the frequency of meat consumption for young people was
40%, common rat 8%) of the villagers preferred rodent
mush higher than old people (x² = 0.56, p < 0.05) (Fig. 2).
meat than the meat of other mammal (Fig. 4). The red

14
A.E. Assogbadjo, J.T.C. Codjia, B. Sinsin, M.R.M. Ekue and G.A. Mensah
river hog (Potamochoerus porcus) also was a significant
many rodent species populations were able to cope with
source of meat (45%), with bushbuck (Tragelaphus scrip-
recurrent hunting without extinction (MALAISSE, 1997).
tus) accounting for 2% of consumption.
The limiting factor is much more the lack of thorough
knowledge on their ecology and the density of their popu-
lations. Therefore, it would be desirable to undertake a
Bushbuck
population study in these regions to look at the impact of
2%
Giant rat
this hunting on the rodent populations. The hunting of
5%
All other rats
wildlife, in particular rodent species as found here, pro-
8%
vide important sources of animal proteins and incomes
for local populations, and therefore should be integrated
in the concept of sustainable development. The consump-
tion of large rodents for their meat (grasscutter and giant
rat) is not only a consequence of lack of meat, but also a
Grasscutter
response to a set of complex factors including cultural
40%
constraints, preferences and values. Such factors may
explain why older people consumed more rodent meat
Red river hog
than younger people in our study area and highlight the
45%
importance of these resources for rural populations of
Africans.
Fig. 4. ­ Mammals species
CONCLUSION
Rodents will continue to be a considerable source of
DISCUSSION
animal protein and income for villagers of Lama reserve
forest. Rodents are the animal species most frequently
Wildlife constitutes an important food resource, which
consumed and preferred by the local populations. How-
cannot easily be replaced or removed without causing
ever, wild animals, including rodents, are not always
negative socio-economic disturbances. To understand the
taken into account in the national programmes for food
contribution of wildlife in the food of local populations
security. Management of these resources should be
we should not consider only the big game. Most of the
included in sustainable resource initiatives that are
meat consumed by forest bordering populations in this
already part of the cultural values of poor rural popula-
study came from small mammals, which could be cap-
tions. Production in the wild and production in the exten-
tured in any time of the year. If rodents were not availa-
sive and intensive domestication of wild fauna can be
ble, bush meat would not be consumed by more than 60%
integrated into national programmes for protected areas
of local populations (ASSOGBADJO, 2000). Rodents were
management. This is necessary to take into account the
and still are the main source of animal food for rural pop-
concerns of local populations and concurrently to satisfy
ulations and provide an important dietary quantity of ani-
requirements for keeping the balance within ecological
mal proteins (COLYN & DUDU, 1987; MALEKANI & PAU-
communities.
LUS, 1989; WETCHI et al., 1988). The grasscutter and giant
rat were most consumed by villagers in our study.
LITERATURE CITED
MALAISSE (1997) showed that 100 g of grasscutter or
giant rat's fresh or smoked muscle provided 28 g and 42.6
g of protein, 10 mg and 20 mg of iron, and 936 Kj and
AJAYI & OLAWOYE, (1974). Some indications of the social
acceptance of the African rat (Cricetomys gambianus Water-
1132 Kj energy respectively. MALAISSE (1997) also indi-
house) in southern Nigeria. Nigerian J. Forestry, 4(1) : 36-
cated that the nutritional value of rodents is similar to that
41.
of beef and chicken. In addition, these two rodent species
ASIBEY, E.O.A. & G.S. CHILD (1990). Aménagement de la faune
are sold in local market at high prices, providing them
pour le développement rural en Afrique subsaharienne. Una-
with a source of income. In Lama forest reserve, the sell-
sylva, 161 : Vol 41.
ing price for a grasscutter was US$10-12 (Table 3). In this
ASSOGBADJO, A.E. (2000). Etude de la biodiversité des res-
area, a hunter killed an average of 4 individual grasscutter
sources forestières alimentaires et évaluation de leur contri-
rats per week. This is equivalent to an income of US$40-
bution à l'alimentation des populations locales de la forêt
48 per week (US$160-192 per month). This pattern is
classée de la Lama. Thèse d'ingénieur Agronome. FSA/
characteristic in West Africa as NTIAMOA-BAIDU (1998)
UNB, Bénin. P. 131 + Annexes.
reported that the incomes resulting from the sales of bush
BAPTIST, R. & G.A. MENSAH (1986). Benin and West Africa : the
meat enable households not only to buy less expensive
cane rat-farm animal of the future? World Animal Review,
other source of protein such as fish, but also it helps sat-
60 : 2-6.
isfy other needs for the families. For example, in Ibadan
COLYN, M. & A. DUDU (1987). Exploitation du petit et moyen
gibier des forêts ombrophiles du Zaïre. Nat. Fau., 3(4) : 22-
(Nigeria), in 1975, when the meat of sheep and beef were
39.
sold respectively at US$2.80-4.20 /kg, grasscutter meat
CODJIA, J.T.C. & J.C. HEYMANS (1988). Problématique liée à
cost UD$9.60 (ASIBEY & CHILD, 1990). The hunting pres-
l'utilisation du gibier et écoéthologie de quelques rongeurs
sure on wildlife led to a progressive reduced availability
consommés au Bénin. Nat. Fau., 4(4) : 4-21.
of animal products in nearby cities where poaching was
DE VISSER, J., G.A. MENSAH, J.T.C. CODJIA & A.H. BOKONON-
common. However, due to their high rate of reproduction,
GANTA (eds), (2001). Guide préliminaire de reconnaissance

Importance of rodents as a human food source in Benin
15
des rongeurs du Bénin. C.B.D.D./Ecooperation/ReRE/VZZ -
MALAISSE, F. (1997). Se nourrir en forêt claire africaine :
République du Bénin/Royaume des Pays-Bas. ISBN :
Approche écologique et nutritionnelle. P.384. CTA.
99919-902-1-6, 253 p.
MALEKANI, M. & J. PAULUS (1989). Quelques aspects de la con-
FALCONER, J. (1996). Sécurité alimentaire des ménages et forest-
sommation du cricétome, Cricetomys (Rongeur), par les
erie. Analyse des aspects socio-économiques. FAO, Rome,
populations zaïroises. Tropicultura, 4 : 60-67.
1996.
MENSAH, G.A. (1991). Élevage des espèces de gibier : cas de
HANOTTE, O. & G.A. MENSAH (2002). Biodiversity and domesti-
l'aulacodiculture. (Élevage de l'aulacode : Thryonomys
cation of 'non-conventional' species : a worldwide perspec-
swinderianus). Mémoire spécial. Chapitre D : Gestion du
tive. 7th World Congress on Genetics Applied to Livestock
patrimoine forestier. Gestion de la faune sylvestre. 10ème
Production, 19-23 August, 2002, Montpellier, France, 30 :
Congrès Forestier Mondial "La forêt, Patrimoine de l'Ave-
543-546.
nir", Paris 17 au 26 septembre 1991.Revue Forestière Fran-
HEYMANS, J.C. & J.T.C. CODJIA (1988). L'élevage des rongeurs :
çaise, Hors série, 5 : 301-309.
une possibilité pour résoudre le problème alimentaire en
NTIAMOA-BAIDU, Y. (1998). La faune sauvage et la sécurité ali-
Afrique. Rised-Bulletin, 7 : 9-12.
mentaire en Afrique. P. 109. Rome.
HEYMANS, J.C. (1986). Petit guide des mammifères du Nord-
SINSIN, B., I. DAOUDA, E. AHOKPE, A. TEHOU, P. COUBEOU, I.
Bénin. Bibliothèque FSA/UAC, Campus d'Abomey-Calavi,
TOKO & S. FANDOHAN (1997). Faune sauvage des forêts
Bénin.
classées de Goungoun, de la Sota et des environs. Labora-
JORI, F., G.A. MENSAH & E. ADJANOHOUN (1994). Grasscutter
toire d'Ecologie Appliquée FSA/UNB.
production : an example of rational exploitation of wildlife.
WETCHI, L., M. BIYA & J. RUELLE (1988). Observations sur
Biodiversity and conservation, 4 : 257-265.
l'importance relative des voies d'approvisionnement du
KINGDON, J. (1997). The kingdon field guide to African mam-
gibier mammalien vendu au marché central de Kisangani
mals. (ed) Academic press. ISBN 0-12-408355-2. P. 465.
(Zaïre). Ann. Fac. Sci. Kis., 5 : 105-114.


Belg. J. Zool., 135 (supplement) : 17-19
December 2005
Pcr detection of Leptospira DNA in rodents and insecti-
vores from Tanzania

G.F. Mgode1, G. Mhamphi1, A. Katakweba1, E. Paemelaere2, N. Willekens2, H. Leirs2,
R.S. Machang'u
1 and R.A. Hartskeerl3
1 Pest Management Centre, Sokoine University of Agriculture, PO Box 3110, Morogoro, Tanzania
2 Evolutionary Biology Group, University of Antwerp, Groenenborgerlaan 171, B-2020, Antwerp, Belgium
3 Department of Biomedical Research, Royal Tropical Institute, Meibergdreef 39, 1105 AZ, Amsterdam, the Netherlands
Corresponding author : G.F. Mgode, e-mail : gfmgode@lycos.com, gfmgode@hotmail.com
ABSTRACT. The true prevalence of leptospirosis in Tanzania is unknown or underestimated. In this study we
report on the prevalence of leptospirosis in Morogoro, Tanzania, by PCR detection of leptospiral DNA in 27 kid-
neys of rodents (Mastomys spp, Rattus spp, and Mus spp) and insectivores (Crocidura spp). The PCR study comple-
mented previous attempts to isolate the leptospires and to perform seroprevalence by the microscopic agglutination
test (MAT). Results of this study indicated an overall detection rate of 11% by PCR, 7.4% by isolation and 0% by
the MAT. Based on our analysis, it is recommended to use PCR and isolation for the detection of leptospires in
potential host animals.
KEY WORDS : Leptospirosis, prevalence, microagglutination, rodents, Tanzania
INTRODUCTION
zania, by PCR detection of leptospiral DNA from kidney
tissues of rodents and insectivores captured in this town.
Pathogenic Leptospira causes leptospirosis in a wide
This is the first report on the molecular prevalence of
range of mammalian hosts. Rodents are considered the
leptospires in an urban/periurban setting in Tanzania.
primary natural reservoirs of leptospirosis in many parts
of the world (ALSTON & BROOM, 1958; FAINE, 1982). In
MATERIAL AND METHODS
Tanzania, the true prevalence of leptospirosis is unknown
or underestimated due to limited knowledge on this dis-
DNA extraction from kidney tissues and PCR :
ease, and hence it is neglected in clinical diagnosis. The
gold standard assay in leptospirosis diagnosis is the
Kidneys of 20 rodents : Mastomys spp (18), Rattus spp
microscopic agglutination test (MAT) described by WOLF
(1) and Mus spp (1), and seven insectivores or shrews (Cro-
(1954). Due to the diversity of antigens within the Lept-
cidura spp) were used. The kidney was ground in a sterile
ospira species, this method may fail to reveal infection
mortar containing 500 µl sterile distilled, de-ionized water.
with certain leptospiral serovars especially in a newly
The kidney homogenate (200 µl) was used to extract DNA
studied area where the prevalent (endemic) serovars are
using the Anansa® Fast `n' Easy Genomic DNA Purifica-
unknown, or in cases where antibody titres are low or
tion kit (Tebu-Bio Laboratories, Cedex, France).
absent. Especially in the case of animals, MAT may be
The PCR was carried out using specifically designed
specific to an infecting serovar or to antigenically closely
primers for the detection of pathogenic and saprophytic
related serovars. The chance of detecting leptospiral anti-
leptospiral DNA as described by MURGIA et al., (1997)
bodies in a new environment, therefore, increases with
with slight modifications. Briefly, the PCR consisted
the number of serovars included in the antigen panel.
Lepat 1 (5'-GAG-TCT-GGG-ATA-ACT-TT-3') and Lepat
Leptospires can be isolated from pathological materials
2 (5'-TCA-CAT-CG(CT)-TGC-TTA-TTT-T-3') primer
(blood, cerebral spinal fluid, urine and kidney tissues), by
pair for pathogenic Leptospira; and Sapro 1 (5'-AGA-
culturing the primary specimen in selective culture media
AAT-TTG-TGC-TAA-TAC-CGA-ATG-T-3') and Sapro 2
containing neomycin sulphate, sodium sulphathiozole,
(5'-GGC-GTC-GCT-GCT-TCA-GGC-TTT-CG-3') prim-
cyclohexamide and 5-Fluorouracil to reduce contamina-
ers for saprophytic Leptospira.
tion (ADLER et al., 1986; ALEXANDER, 1991; FAINE, 1982).
DNA of known pathogenic Leptospira species (serovar
Molecular diagnosis of leptospirosis has been greatly
Kenya, serogroup Ballum), and saprophytic species (sero-
facilitated by PCR detection of specific leptospiral DNA
var Patoc, serogroup Semaranga) was used as the control.
with specific primers that enable amplification of all
The reactions mix consisted 10 mM Tris-HCl, pH 8.3,
saprophytic and pathogenic leptospires, as well as lepto-
50 mM KCl, 200 mM of each deoxynucleoside triphos-
spires of ambiguous classification (MURGIA et al., 1997;
phates (dNTP), 0.5 µM of each primer, 5 µl template
MERIEN et al., 1992; GRAVEKAMP et al., 1993).
DNA with modified MgCl concentration (2 mM) and
The aim of this study was to obtain molecular data for
DNA polymerase (1 U). The PCR condition for patho-
assessing the prevalence of leptospires in Morogoro, Tan-
genic Leptospira were : initial denaturation at 93oC for 3


18
G.F. Mgode, G. Mhamphi, A. Katakweba, E. Paemelaere, N. Willekens, H. Leirs, R.S. Machang'u and R.A. Hartskeerl
min then 35 cycles of denaturation at 93oC for 1 min,
Isolation of live leptospires
primer annealing at 48oC for 1 min, DNA extension at
from kidney homogenates :
72oC for 1 min, and further 10 min extension after the last
cycle. Saprophytic PCR condition were : heat denatura-
Kidney cultures of the 27 animals yielded two Lept-
tion at 93oC for 3 min, then 35 cycles of heat denaturation
ospira isolates from Crocidura spp., derived from the
at 93oC for 1 min, primer annealing at 63oC for 1.5 min,
same two animals that were scored positively with the
DNA extension at 72oC for 2 min and after the last cycle
PCR method.
extension continued for further 10 min.
Seroprevalence of leptospiral antibodies
Leptospira isolation from kidney tissues :
in rodents :
Tissue samples were prepared by grinding the freshly
No agglutination was found in the MAT of the 27 sera
obtained kidneys of Mastomys spp (18), Rattus spp (1), Mus
against the five-leptospiral serovars used. Table 1 summa-
spp (1) and Crocidura spp (7) in sterile phosphate buffered
rises all of these results.
saline (pH 7.2). About 0.5 ml of the kidney homogenate
was inoculated in Fletcher's Leptospira medium containing
TABLE 1
as selective growth inhibitor, 5-Fluorouracil (200 µg/ml).
Comparison of leptospires detection rates by PCR, isolation and
The cultures were incubated at ambient temperature (26-
microagglutination test (MAT) in rodents and insectivores.
30oC) and examined for leptospiral growth at seven-day
intervals by dark field microscopy (FAINE, 1982, 1988).
Test
Sample tested
Positive sample
Percent positive
Seroprevalence of leptospires in rodents
PCR
27
3
11%
Isolation
27
2
7.4%
and insectivores :
MAT
27
0
0%
The microscopic agglutination test (MAT) was used to
Rodents (n=20) and insectivores (n=7)
detect leptospiral antibodies in the sera of the same rodents
and insectivores used in the PCR and isolation studies.
The MAT was carried out as described by COLE et al.,
DISCUSSION
(1973) using live antigen of five Leptospira serovars. The
serovars were; a previously identified/proposed serovar
The results of this study indicated a detection rate of
Sokoine (MGODE et al., in Press) (serogroup Icterohaemor-
leptospires of 11% (3/27) by PCR on the investigated
rhagiae); serovar Hebdomadis (serogroup Hebdomadis);
cases, compared to 7.4% (2/27) by isolation and 0% by
serovar Hardjo (serogroup Sejroe); serovar Kenya (sero-
serology (MAT). The rate of Leptospira detection by PCR
group Ballum) and serovar Pomona (serogroup Pomona).
per animal species was 6% for Mastomys spp. and 29%
for Crocidura spp. These relatively high infection rates
RESULTS
could represent a hazard to public health. As there was
only one individual of Rattus spp. and Mus spp. analysed
with the different detection methods, it is not possible to
PCR of kidney tissues :
comment on the potential infection rates of these two spe-
Out of 20 rodents and seven insectivores tested, three
cies. Our data suggest that leptospire detection will be
were PCR positive, generating a 330 base pair product
highest using PCR and isolation in a situation where sero-
with the Lepat 1 and Lepat 2 primers (Fig. 1). The PCR
logical antibody detection (MAT) fails, particularly in a
positive samples were from Crocidura spp., 2 of 7 (29%)
new study area.
and Mastomys spp., 1 of 18 (6%). None of the samples
Serological survey using MAT requires use of known
were PCR positive with the Sapro1 and Sapro2 primers.
or closely antigenically related prevalent leptospiral sero-
This finding was consistent with the presence of patho-
vars. The use of multiple antigens from different serovars
genic leptospires.
in MAT makes this test time consuming. In our study, the
negative MAT results found may be due to the limited
number of serovars (5) employed as antigen. This also
Marker
Kenya 133
114 146
suggests that the leptospires detected by PCR (n=3) and
isolation (n=2) might posses a variant antigenic pattern,
which is unrelated to that of the new putative serovar
Sokoine, serovar Hebdomadis, serovar Hardjo, serovar
Pomona and serovar Kenya used in the MAT. However,
330 bp
this is somewhat unexpected as both serovars Machang'u
(isolate RM1) and Kenya (isolates Sh9 and Sh25) were
recently isolated from cattle and Cricetomys gambianus
rats in periurban Morogoro, respectively (MACHANG'U et
al., 2003). These serovars were thus anticipated to be gen-
Fig. 1. ­ PCR products of DNA from kidneys of Crocidura spp
erally prevalent among rats in urban/periurban Morogoro.
(133 and 114) and Mastomys spp (146) with Lepat1 and Lepat2
An alternative and more likely explanation is that the
primers. Serovar Kenya is the control pathogen, and M is the
Crocidura spp. and Mastomys spp. are natural hosts for
DNA ladder.
The products were separated by electrophoresis in 3% agarose
one or more of the serovars included in the MAT panel
and stained with ethidium bromide.
displaying low antibody titres below the detection thresh-

Pcr detection of Leptospira DNA in rodents and insectivores from Tanzania
19
old of the MAT. Lower leptospiral antibody (usually IgG)
BLACKMORE, D.K., L.M. SCHOLLUM & K.M. MORIARTY (1984).
levels encountered in natural hosts may indeed give MAT
The magnitude and duration of titres of leptospiral aggluti-
negative results, especially in the case of heterologous
nins in human sera. New Zealand medical journal, 97 : 83-
serovars (B
86.
LACKMORE et al., 1984; EVERARD & BENNETT,
1990; P
C
ALIT et al., 1991).
OLE, J.R., C.R. SULZER & A.R. PURSELL (1973). Improved
microtechnique for the leptospiral agglutination test. Applied
The success in isolation of leptospires was supported
Microbiology, 25 : 976-980.
by the PCR analysis. The percentage of successful isola-
EVERARD, C.O.R. & S. BENNETT (1990). Persistence of leptospi-
tions may, however, not reflect the true percentage of car-
ral agglutinins in Trinidadian survey subjects. European
riership because a number of cultures (5) were lost due to
journal of epidemiology, 6 : 40-44.
contamination. Indeed contamination with less fastidious
FAINE, S. (1982). Guidelines for the control of leptospirosis.
and faster growing microorganisms forms a major limita-
World Health Organization, Geneva.
tion of Leptospira isolation in spite of the fact that con-
FAINE, S. (1988). Leptospirosis. In : BALOWS A, W.J. HAUSLER,
tamination can be kept at minimum by deploying selec-
E.H. LENNETTE, M. OHASHI & A. TURANO (eds), Laboratory
tive growth inhibitors such as 5-Fluorouracil in the
diagnosis of infectious diseases. Principles and practice.
culture medium.
Vol. 1. Bacterial, mycotic and parasitic diseases. Springer-
Verlag, New York.
As MAT can fail to reveal leptospires carriership in
GRAVEKAMP, C., H. VAN DE KEMP, M. FRANZEN, D. CARRING-
some host animals, it is recommended that PCR and isola-
TON, G.J. SCHOONE, G.J.J.M. VAN EYS, C.O.R. EVERARD,
tion methods are also used, where feasible, to investigate
R.A. HARTSKEERL & W.J. TERPSTRA (1993). Detection of
infection sources. These methods are particularly impor-
seven species of pathogenic leptospires by PCR using two
tant when carrying out studies in area with unknown lept-
sets of primers. J. Gen. Microbiol., 139 : 1691-1700.
ospirosis prevalence. Rodents and insectivores should be
MACHANG'U, R., G. MGODE, J. ASENGA, G. MHAMPHI, R. HARTS-
considered important reservoirs of leptospires in the
KEERL, M. GORIS, C. COX, B. WEETJENS & R. VERHAGEN
Morogoro area and can serve as indicators of leptospiro-
(2003). Characterisation of Leptospira isolates from captive
giant pouched rats, Cricetomys gambianus. In : S
sis prevalence.
INGLETON,
G.R., L.A. HINDS, C.J. KREBS & M.D. SPRATT (eds), Rats,
Mice and People, Rodent Biology and Management.
Aus-
ACKNOWLEDGEMENTS
tralian Centre for International Agricultural Research, Can-
berra 2003.
We thank the Pest Management Centre, Sokoine University
MERIEN, F., P. AMOURIAUX, P. PEROLAT, G. BARANTON & I.
of Agriculture (Morogoro, Tanzania), the Evolutionary Biology
SAINT GIRONS (1992). Polymerase chain reactions for detec-
Group, University of Antwerp (Antwerp, Belgium) and the
tion of Leptospira species in clinical samples. Journal of
Department of Biomedical Research, Royal Tropical Institute
clinical microbiology, 30 : 2219-2224.
(Amsterdam, the Netherlands) for supporting this work.
MGODE, G.F., R.S. MACHANG'U, M.G. GORIS, M. ENGELBERT, S.
SONDY § R.A. HARTSKEERL (in press). New Leptospira sero-
var Sokoine of serogroup Icterohaemorrhagiae from cattle in
REFERENCES
Tanzania. Int. J. Syst. Epidemial Microbiol.
MURGIA, R., N. RIQUELME, G. BARANTON & M. CINCO (1997).
ADLER, B., S. FAINE, W.L. CHRISTOPHER & R.J. CHAPPEL (1986).
Oligonucleotides specific for pathogenic and saprophytic
Development of an improved selective medium for isolation
leptospira occuring in water. FEMS Microbiol. Lett., 148 :
of leptospires from clinical material. Veterinary Microbiol-
27-34.
ogy, 12 : 377-381.
PALIT, A., C. HOSKING, L. CREATE, J. MORTON & S. FAINE
ALEXANDER, A.D. (1991). Leptospira. In : BALOWS, A., W.J.
(1991). Leptospirosis in dairy farmers of Western Victoria,
HAUSLER, K.L. HERMANN, H.D. ISENBERG & H.J. SHADONY
Australia. In : KOBAYASHI Y. (ed.), Leptospirosis. Proceed-
(eds), Manual of clinical microbiology 5th ed. American
ings of the Leptospirosis Research Conference (1990).
society for microbiology, Washington.
Hokusen-Sha, Tokyo.
ALSTON, J.M. & J.C. BROOM (1958). Leptospirosis in man and
WOLF, J.W. (1954). The laboratory diagnosis of leptospirosis.
animal. E & S. Livingstone, Edinberg and London.
Thomas publishers, USA.


Belg. J. Zool., 135 (supplement) : 21-29
December 2005
Investigating the role of natural gallery forests outside
the Congolese rainforest as a refuge for African forest
shrews

Patrick Barrière1,2, Rainer Hutterer3, Violaine Nicolas4,2, Sophie Quérouil5,2 and Marc
Colyn
1,2
1 Laboratoire Ecobio : Ecosystèmes, Biodiversité et Evolution, UMR 6553 - CNRS, Université de Rennes 1, Station Biologique,
F-35380 Paimpont, France.
2 Laboratoire EVE : Ethologie, Evolution et Ecologie, UMR 6552 - CNRS, Université de Rennes 1, Station Biologique, F-35380
Paimpont, France.
3 Zoologisches Forschungsmuseum Alexander Koenig, Adenauerallee 160, D-53113 Bonn, Germany.
4 Museum National d'Histoire Naturelle, Département de Systématique et Evolution, UMR 5202 - CNRS, USM 601, Labora-
toire Mammifères et Oiseaux, 55 rue Buffon, F-75005 Paris, France.
5 Instituto do Mar (IMAR), Departamento de Oceanografia e Pescas, cais Santa Cruz, 9901-862 Horta, Portugal.
Corresponding author : P. Barrière, e-mail : Patrick.Barriere@laposte.net
ABSTRACT. Conditions that prevailed in rainforest faunal refuges during glacial periods of the Pleistocene, partic-
ularly their size, position and habitat characteristics, remain little investigated. After the peak of the last interglacial
period of the Holocene (7000 years B.P.), the Guineo-Congolese rainforest has been reduced in size and isolated
gallery forests emerged in the peripheral Northern Congo forest-savanna mosaic, mainly because of the reduction in
rainfall. In the north of the Central African Republic, 400 km north of the present rainforest zone, up to the Sudano-
Sahelian savannas, such gallery forests harbour several forest species of plants, birds and mammals related to West
and/or East Congo faunal regions. This suggests that since the catastrophic destruction of central African rainfor-
ests, that culminated about 2500 years ago, these galleries could have mimicked the conditions that occurred in the
Pleistocene refuges. We tested whether these natural gallery forests, outside the Congolese rainforest, could act as
refuges for small forest mammals such as shrews. Composition and structure of shrew communities were studied in
three main regions belonging to three river basins and two distinct phytoregions. They were compared to two other
shrew communities located within the main Congolese rainforest, also in C.A.R.. None of the typical rainforest
shrew species was collected within the studied isolated gallery forests. Thus, climatic and habitat characteristics
within these gallery forests were presumably not suitable for these forest patches to act as climatic refuges for the
forest shrew fauna.
KEY WORDS : Soricidae, Crocidura, Suncus, Central African Republic, Community structure, Biogeography, Refuge
theory.
INTRODUCTION
the north, up to the present Sudano-Sahelian savanna
zone. Since 5000 years B.P., the reduction in rainfall
In the Quaternary period (from 1.8 My B.P. to present),
(BERTAUX et al., 2000), associated with climatic distur-
Africa underwent climatic oscillations resulting in several
bance, resulted in contraction of the central African rain-
phases of forest fragmentation and extension. During the
forests that culminated about 2500 years ago (MALEY,
last glacial maximum of the Pleistocene (between about
2001). In the north of the Central African Republic
20000 and 15000 years B.P.), as a result of the very cold
(C.A.R.), since 7000 years B.P., the extension of the
and dry climate, tropical rainforest decreased in size and
savannas (MALEY & BRENAC, 1998) resulted in the isola-
became fragmented. Paleontological and biogeographic
tion of gallery forests outside the Congolese rainforest.
data suggest that it was limited to lowland forest patches
Despite a new phase of forest extension that began 2000
in the downstream zone of the large rivers and on lower
years ago (MALEY, 2001), it is probable that these
slopes of mountains (HUTTERER et al., 1987; MALEY,
Holocene gallery forests remained isolated from the Con-
1987; COLYN, 1991; COLYN et al., 1991; MALEY, 1996;
golese rainforest and could have acted as forest refuges. It
MALEY & BRENAC, 1998). According to HAFFER (1969),
is hypothesised that Pleistocene and Holocene refuges
these forest patches may have acted as refuges for the for-
probably formed a network of forest isolates within a for-
est flora and fauna. In contrast, during the inter-glacial
est-savanna mosaic zone rather than small homogenous
periods, the warmer and wetter climate favoured an
forest isolates (LEAL, 2000; MALEY, 2001). The principal
extension of the rainforest. The last inter-glacial period
refuges of the Pleistocene were probably located in the
culminated at 7000-8000 years B.P., during the Holocene
downstream zone of the large rivers, within the present
(MALEY, 2001). The central African rainforest extended to
rainforest zone. However, the conditions that prevailed in

22
Patrick Barrière, Rainer Hutterer, Violaine Nicolas, Sophie Quérouil and Marc Colyn
these forest refuges, particularly their size and habitat
MATERIAL AND METHODS
characteristics remain poorly investigated.
Study area
In the North of the C.A.R., more than 400 km north of
The study area (80000 km2) is located in the north of
the present northern boundary of the rainforest, Holocene
C.A.R., more than 400 km north of the present northern
gallery forests are known to harbour typical Congolese
boundary of the Congolese rainforest (Fig. 1). It has been
forest plants, birds and mammals (FAY, 1988; CHRISTY,
managed until the end of the year 2000 by the "Pro-
1999; ECOFAC, 2001). These taxa also occur in the West
gramme de Développement de la Région Nord"
Central or East Central regions (sensu COLYN, 1999;
(P.D.R.N.). Since then, it has been associated with the
COLYN & DELEPORTE, 2002b) and these galleries repre-
ECOFAC project under the name "Zones Cynégétiques
sent their northernmost range limit. In particular, FAY
Villageoises" (Z.C.V.). Shrews were collected at eleven
(1988) showed that three strictly arboreal species of forest
sites (Fig. 1) located within National Parks (Bamingui-
primates (Cercopithecus pogonias, C. nictitans and C.
Bangoran N.P., sites 5-7; Manovo-Gounda-St Floris N.P.,
ascanius), known to be widely distributed in the Congo-
8-9 and its periphery, 10), within a hunting zone (Sangba,
lese rainforest, can be found in the northern C.A.R. gal-
sites 3 and 4) or in neighbouring zones (sites 1, 2 and 11)
within the project's action area (T
lery forests, in the south-western sector of Manovo-
ÉLLO, 2000; ECOFAC
ONLINE). All these sites, which are located in the upstream
Gounda-St. Floris National Park. The aim of this study is
zone of several rivers, were pooled in three main regions :
to test whether these natural gallery forests, located out-
Bohou (sites 1 and 2), Bamingui-Bangoran (sites 3-7) and
side the Congolese rainforest, may act as refuge for ter-
Manovo-Mara (sites 8-11) They belong to three river
restrial small forest mammals such as shrews. This may
basins (Bahr Aouk, Bamingui and Kotto) and two distinct
help to understand the conditions that could have pre-
phytoregions (medio-Sudanian savanna and Sudano-
vailed in the Pleistocene refuges.
Sahelian savanna; (Table 1; Fig. 1).
N
W
E
SUDAN
CHAD
S
100 km
.10 .
C
11
hari
Bahr Aouk
I
II
Ba
m
III
in
.7 8 .
gu
.
9
i
.6 . .
. 1
IV
5
4
2
3
CENTRAL AFRICAN REPUBLIC
Kotto
Lobaye Bangui
Oubangui
Batouri
Sa ²
CAMEROON n
Kongana
gha
DEMOCRATIC REPUBLIC OF CONGO
REP. OF
CONGO
Fig. 1. ­ Study sites in the north of the Central African Republic : 1, Kpata; 2, Bohou; 3, Bamingui-Brendja; 4, Sangba; 5, Kaha
pond, near Bamingui; 6, Kivou pond; 7, Bangoran; 8, Manovo; 9, Koumbala; 10, Gordil; 11, Délembé. Phytoregions sensu
BOULVERT (1986) : I, Sudano-Sahelian savanna; II, medio-Sudanian savanna; III, Sudano-Guinean savanna and peri-forest sec-
tor of the Guineo-Congolian rainforest and savanna domain; IV, rainforest sector of the Guineo-Congolian domain. The two
sites located within the Congolese rainforest and selected for comparison (Batouri and Kongana) are also plotted.

Shrews from gallery forests outside the Congo River basin
23
The Bohou region is located in the upstream zone of
cal medio-Sudanian sector is of Encephalartos septentri-
the Kotto River within the Congo River basin, while the
onalis type at the sites 1-4 and 8-9, with included bamboo
two other regions are located in the upstream zone of the
savannas of Oxytenanthera abyssinica type (e.g. site 3),
Chari River, within the Chad River basin. A savanna zone
and of Butyrospermum paradoxum parkii type at sites 5-7
of ca. 5 km has been separating the two river basins for
(BOULVERT, 1986). It harbours gallery forests of Anogeis-
the past few hundred years (FAY, 1988). The southern-
sus leiocarpus type at all sites (1-9). The northernmost
most sites (1-9) belong to the medio-Sudanian savanna
sites (10, 11) belong to the Sudano-Sahelian domain and
domain sensu BOULVERT (1986), while the other two (10
are of Terminalia lexiflora type.
and 11) belong to the Sudano-Sahelian savanna. The typi-
TABLE 1
Characteristics of each study site. Site numbers refer to Figure 1. Main habitat types : BS, bamboosian
savanna; GF, gallery forest; RS, riparian savanna; S, undifferentiated savanna.
Main habitat
Annual
Site
Study Site
Coordinates
Main Region
Sub-climate
Rainfal
No
GF
S
(mm)
1
Kpata
08°03'N; 21°24'E
Bohou
GF
sub-Sudanian
1200-1400
2
Bohou
07°43'N; 21°20'E
Bohou
GF
sub-Sudanian
1200-1400
3
Bamingui-Brendja
07°47'N; 20°57'E
Bamingui-Bangoran
GF
BS
sub-Sudanian
1200-1400
4
Sangba
07°45'N; 20°43'E
Bamingui-Bangoran
RS
sub-Sudanian
1200-1400
5
Kaha pond near Bamingui
07°26'N; 20°08'E
Bamingui-Bangoran
GF
S
sub-Sudanian
1000-1200
6
Kivou pond (= Kaga Yara)
07°51'N; 20°11'E
Bamingui-Bangoran
RS
sub-Sudanian
1000-1200
7
Bangoran
08°05'N; 20°21'E
Bamingui-Bangoran
S
sub-Sudanian
1000-1200
8
Manovo
08°22'N; 20°57'E
Manovo-Mara
RS
sub-Sudanian
1200-1400
9
Koumbala
08°19'N; 21°17'E
Manovo-Mara
GF
sub-Sudanian
1200-1400
10
Gordil
09°36'N; 21°41'E
Manovo-Mara
RS
Sudano-Sahelian
1200-1400
11
Délembé
09°44'N; 22°39'E
Manovo-Mara
RS
Sudano-Sahelian
1000
All the sites have a Sudano-Guinean climate, with sub-
plastic drift fence. All the lines were constructed as
Sudanian or Sudano-Sahelian sub-climates (Table 1).
described in NICOLAS et al. (2003). Trapping period,
Gallery forests were surveyed in each of the three regions,
number of pitfall lines and cumulated trapping effort var-
and the surrounding savanna was also surveyed within the
ied between sites (Table 2). The thirty-five pitfall lines
Chad River basin (Table 1). Sampling in different river
totalled a pitfall trapping effort of 8581 bucket-nights.
basins and phytoregions, rather than focusing on sam-
Mixed Sherman traps and metal snap-traps were also
pling within a single gallery forest, was considered the
used in transects, mainly to capture rodents, and totalled
best way to have a good representation of the shrew com-
32844 trap-nights. In addition, a few specimens were
munity in the study area. Human populations in the area
incidentally collected by J.L. TÉLLO since June 1997.
have a density lower than 0.5 persons/km2 and have a low
Most of the small mammals captured were weighed,
disturbance effect on the small mammal fauna, especially
measured, autopsied and preserved in 10% formalin. Spe-
on shrews that are never hunted nor eaten.
cies identification was performed by two of the authors
Two sites located within the Congolese rainforest were
(R.H. and P.B.) on the basis of morpho-anatomical analy-
selected for comparison (Fig. 1). The Batouri River site,
ses, and in some instances supported by molecular analy-
in the Ngotto forest, consists of a mixed-species semi-
ses (QUÉROUIL et al., 2001; in press). The taxonomic
deciduous rainforest at the northern limit of the Guineo-
nomenclature follows HUTTERER (1993), except for Sun-
Congolese rainforest (more details in BOULVERT, 1986).
cus megalura, previously considered as a Sylvisorex spe-
This primary rainforest is often affected by storms and
cies, but now considered as a Suncus species according to
numerous tree falls have created openings in the canopy
QUÉROUIL et al. (2001). Three problematic Crocidura
(BARRIÈRE et al., 2000). With a typical forest Guinean cli-
taxa, belonging to species complexes still in need of revi-
mate, the rainfall, occurring mainly from May to October,
sion, were named Crocidura cf. denti, C. cf. hildegardeae
averages 1600 mm per year. The Kongana study area con-
and C. cf. poensis as they could not be definitively identi-
sists mainly of a mixed-species semi-deciduous rainforest
fied. The material was deposited at the Station Biologique
and a mono-dominant Gilbertiodendron dewevrei forest,
de Paimpont, University of Rennes 1, France and the
and has a climate similar to that of Ngotto (RAY & HUT-
Museum Alexander Koenig, Bonn, Germany.
TERER, 1996).
Data analysis
Sampling
The trapping effort (TE, in trap-nights) was defined as
Trapping was mainly performed from May to August
the number of traps (or buckets) set for a 24-hour period,
1998 and from June to August 1999 (Table 2), i.e. during
and the trap success (TS) as the number of individuals
the early wet season, by J.L. TÉLLO and/or a member of
captured per 100 trap-nights; i.e. TS = (N/TE)*100,
the University of Rennes 1. Our major method of collect-
where N is the number of shrews captured. Genus and
ing shrews was by dry, not-baited pitfall traps, with 10-
species richness (G and S) were defined as the number of
litre buckets positioned at 5 m intervals along a linear
distinct genera and species identified, respectively. Spe-

24
Patrick Barrière, Rainer Hutterer, Violaine Nicolas, Sophie Quérouil and Marc Colyn
cies relative abundance (pi, %) was defined as the number
but for the estimation of trap success and species relative
of individuals (ni) of species i captured per 100 individu-
abundance, only the individuals captured by pitfall traps
als of all species, i.e. pi = (ni/N)*100. For the estimation
were selected.
of species richness, all shrew captures were considered,
TABLE 2
Trapping characteristics (trapping period, number of lines and trapping effort, mainly in pitfall (P)), shrew trap success, identity and
number of shrews captured. Site numbers refer to Table 1 and Figure 1. Number of shrews collected in mixed Sherman and metal
snap-traps line (L) are in brackets and those collected by hand (H) in square-brackets. Main habitat type codes refer to Table 1. A spe-
cies code (A-N) was attributed to each shrew taxon.
Main region
Bohou
Bamingui-Bangoran
Manovo-Mara
Total
Site number
1
2
3
4
5
6
7
8
9
10
11
Mean of collect
P
L
P
P
P
P
H
P
L
L
P
P
P
P
L
H
Total
Main Habitat type
GF
GF
GF
BS
RS
GF
S
RS
S
RS
GF
RS
RS
Jun-
May- Aug- Aug-
Jul-
Jul-
Jul-
Jun-
Jun-
Trapping period
98
98
98
99
99
99
98
99
99
No of p
itfall lines
4
5
3
4
4
4
6
3
2
35
Trapping effort (TE)
1,584
1,500 648
700
900
700
1,584 600
365
8,581 32,844
No of shrews captured (N)
58
(1)
41 (1)
26
29 [4]
23
[1]
16
(4)
(4) 47 (2) 23 [4] 7 (1)
270
(13)
[9]
292
Trap success (TS)
3.66
2.73
4.01
4.14
2.56
2.29
2.97 3.83
1.92
3.15
0.04
A Crocidura cf. denti Dollman, 1915
4
(1)
15
4
4 [1]
5
2
18 (1) 1 [1]
53
(2)
[2]
57
B Crocidura cf. hildegardeae Thomas, 1904
23 (1)
4
16
1
[1]
6
2
1
1
54
(1)
[1]
56
C Crocidura cf. poensis (Fraser, 1843)
2
(1)
12 [1] 3 (1)
17
(2)
[1]
20
D Crocidura fuscomurina (Heuglin, 1865)
1
1
1
E Crocidura lamottei Heim de Balsac, 1968
1
(2)
1
(2)
3
F Crocidura littoralis Heller, 1910
4
9
5
18
18
G Crocidura ludia Hollister, 1916
41
41
41
H Crocidura nanilla Thomas, 1909
1
1
2
2
I Crocidura olivieri (Lesson, 1827)
5
1
4
2 [3]
1
7
(2)
(3)
3 (1)
1
1
25
(6)
[3]
34
J Crocidura roosevelti (Heller, 1910)
1
2
1
4
4
K Crocidura turba Dollman, 1910
[2]
[2]
2
L Crocidura yankariensis Hutterer & Jen-
7
7
7
kins, 1980
M Suncus infinitesimus (Heller, 1912)
2
2
7
2
6
18
1
1
39
39
N Suncus megalura (Jentink, 1888)
2
2
2
Specimens not seen
5
1
6
6
G
enus rich
ness (G)
2
1
2
2
2
2
1
1
1
1
2
2
2
2
1
1
2
Sp
ecies rich
ness (S)
7
1
4
5
7
5
1
4

2
2
6
7
5
13
5
5

14
RESULTS
shrew C. yankariensis also constitute the first known
record of the species in C.A.R.
Overall community composition
Community structure
within the gallery forests
A total of 292 shrews representing two genera and
fourteen species were collected. One genus (Suncus) was
In the gallery forests, with a pitfall trapping effort of
only represented by two species and the other one (Croci-
5568 bucket-nights, 169 shrews (TS=3.04) representing
dura) by 12 species (Table 2). More than ninety-two per-
two genera and eight species were recorded (Fig. 2a). The
cent of shrews were collected in pitfall traps (N=270;
two genera (Crocidura and Suncus) were captured in each
TS=3.15) while only 13 individuals were captured in
of the three main regions. The number of species varied
Sherman traps (TS=0.04), nine were collected by hand
from five to seven according to the region. It is noteworthy
and none in metal snap-traps. While only five species
that the shrew community had a higher species richness
were trapped in Sherman traps, thirteen species were cap-
and trap success at Bohou (S=7 and TS=3.66, respec-
tured in pitfalls. Crocidura turba was only collected by
tively), within the Congo River basin. All the species
hand. Two species were numerous (C. cf. hildegardeae
recorded are either savanna species or species occurring
and C. cf. denti, each comprising about 20% of all
within the rainforest and at its margin (see Table 3) but
shrews), five were common (C. ludia, S. infinitesimus, C.
preferentially in open habitat. Crocidura ludia, the most
olivieri, C. cf. poensis and C. littoralis) and the other
forest-dependent species amongst the recorded species,
seven were captured infrequently. The three specimens of
was highly dominant at Bohou, while it was absent from all
the savanna species C. lamottei, described from Lamto
pitfall traps in the Chad River basin. Within the latter, the
(Ivory Coast) and since then widely recorded within
structure of the community differed between the two stud-
Sudanian and Guinean savannas from Senegal to western
ied regions : the medium-sized C. cf. hildegardae was
Cameroon, constitute the first known record of the spe-
more abundant at Bamingui-Bangoran (pi=37.5%) while
cies in C.A.R. and the easternmost limit of its distribution
the medium-sized C. cf. denti and the tiny S. infinitesimus
range. The seven specimens of the uncommon savanna
co-dominated in the Manovo-Mara region (pi=38.3%).

Shrews from gallery forests outside the Congo River basin
25
(a)
(b)
pi (%)
Mean body weight (g)
pi (%)
Mean body weight (g)
80
40
80
40
TE=1,584; N=47; TS=2.97
TE=965; N=30; TS=3.11
60
G=2; S=6
30
60
G=2; S=8
30
Man 40
20
40
20
(sites 8-11)
20
10
20
10
X
80
80
TE=2,400; N=64; TS=2.67
TE=2.048; N=71; TS=3.47
60
G=2; S=5
60
G=2; S=10
Bam 40
40
(sites 3-7)
20
20
X
80
TE=1,584; N=58; TS=3.66
I K C A E B J N L D H M
60
G=2; S=7
Boh
40
(sites 1-2)
20
I F A E B J G M
Fig. 2. ­ Pitfall trapping effort (TE); number of individuals (N), of genera (G) and of species (S) pitfall-trapped; trap success
(TS) and distribution of relative abundance (pi) of the shrew species within (a) the gallery forests and (b) the surrounding
savanna of the three main regions studied : Boh (Bohou), Bam (Bamingui-Bangoran) and Man (Manovo-Mara). Species codes
(A-N) are defined in Table 2 and codes of the two Suncus species (M and N) are in bold. Species are ordered by decreasing
mean body weight (as indicated by the triangles). Crosses indicate records by another mean than pitfall.
Community structure
Differences in the community structure
within the savanna
between gallery forest and savanna
Species richness and trap success were higher in the two
In the savanna zone of the two regions belonging to the
savanna habitats (S=12 and TS=3.35, respectively) than in
Chad River basin, with a pitfall trapping effort of 3013
the gallery forests (S=8 and T=3.04), despite a lower pitfall
bucket-nights, 101 shrews (TS=3.35) representing two
trapping effort (TE=3013). The difference in term of trap-
genera and ten species were recorded (Fig. 2b). Two addi-
success was mainly due to the capture of one additional
tional species were collected by other means than pitfall
species (C. cf. poensis), absent from the gallery forests and
trapping. The two genera (Crocidura and Suncus) were
dominant at Manovo-Mara. Suncus megalura was captured
captured in each of the two regions. The number of spe-
only in the riparian savanna of the Bamingui-Bangoran
cies and the trap success were higher at Bamingui-Ban-
region and the other species that were not captured in gal-
goran (S=10 and TS=3.47, respectively) than at Manovo-
lery forest are four savanna species (C. fuscomurina, C.
Mara (S=8 and TS=3.11). While the medium-sized Croci-
nanilla, C. turba and C. yankariensis) and the problematic
dura cf. hildegardae was dominant at Bamingui-Ban-
C. cf. poensis. The latter is taxonomically closely related to
goran (pi=36.6%), the shrew community of Manovo-
the savanna species C. turba and both belong to the poensis
Mara was dominated by C. cf. poensis (pi=50%) of higher
species complex, still in need of revision (QUÉROUIL et al.,
size.
in press). It could refer to a species different from the true

26
Patrick Barrière, Rainer Hutterer, Violaine Nicolas, Sophie Quérouil and Marc Colyn
C. poensis, which could mainly occur in rainforest. For a
forest shrews, comparison of the shrew community within
given region, the shrew community structure varied
these galleries were made with two other communities in
between gallery forest and savanna, especially at Manovo-
the northernmost part of the Congolese rainforest (RAY &
Mara where C. cf. denti and S. infinitesimus co-dominated
HUTTERER, 1996; BARRIÈRE et al., 2000; see Fig. 1 and
in gallery forest (pi=38.3% each), while C. cf. poensis and
Table 3), also situated in C.A.R. and within the West Cen-
C. yankariensis were the most abundant species in savanna
(pi=50% and 23.3%, respectively). Whatever the habitat
tral faunal region. In the Ngotto forest (Batouri River
type the shrew community structure and the identity of the
site), shrews were mainly pitfall-trapped (BARRIÈRE et al.,
dominant species varied also between regions.
2000) while in Kongana, shrew remains were collected in
carnivore scats (RAY & HUTTERER, 1996). For the two
DISCUSSION
communities combined, five genera and 20 species were
collected (Table 3). In each of these two localities, rain-
In order to test whether the natural gallery forests out-
forest species were the most numerous and represented
side the Congolese rainforest could act as refuge for rain-
altogether more than 88% of the whole collection.
TABLE 3
Comparison between the structure of shrew communities (pi, relative abundance) within gallery forests outside the Congolese rainfor-
est and within the rainforest zone at two sites. "Forest I" means primary forest and "II" means secondary. When absent in the gallery
forests, the presence of the species in the adjacent savanna is indicated by an "X" for each of the three regions. Region abbreviations :
Boh (Bohou); Bam (Bamingui-Bangoran); Man (Manovo-Mara). Species in bold are known to occur within the rainforest zone. Habi-
tat preferences (HP) : F, predominantly dense forest species; FS, species mainly occurring in transition zones such as secondary forest
or forest-savanna mosaic; S, predominantly savanna species. Mean body weight (MBW) is given in grams.
Within the Congolese rainforest
Outside the rainforest
Study location
Batouri
Kongana
Boh
Bam
Man
Total
Main habitat type
Forest I
Forest I & II
Gallery forests
Period
Oct98 ­ Nov99
May92 ­ May94
May98 ­ Jul99
Mean of sampling
Pitfall
carnivore scats
Pitfall
Pitfall trapping effort (bucket-nights)
43,240
­
1,584
2,400
1,584
5,568
Genus richness (G)
5
4
2
2
2
2
Species richness (S)
18
16
7
5
6
8
No of captures (N)
1,339
311
58
64
47
169
Trap success (TS)
3.1
3.66
2.67
2.97
3.05
HP
MBW
Sylvisorex johnstoni
F
2,9
25.0
36.7
Sylvisorex ollula
F
16,1
9.3
3.5
Crocidura nigrofusca
F
11,2
0.1
13.5
Sylvisorex konganensis
F
4,9
0.8
1.3
Paracrocidura schoutedeni
F
7,4
17.6
24.1
Congosorex verheyeni
F
7,1
1.8
Crocidura crenata
F
7,6
8.1
Crocidura dolichura
F
6,1
8.4
3.9
Suncus remyi
F
1,9
3.6
2.3
Crocidura batesi
F
16,0
11.1
3.2
Crocidura goliath
F
70,0
0.1
0.3
Sylvisorex pluvialis
F
5,0
1.0
Crocidura grassei
F
11,0
0.5
Crocidura poensis
F
14,0
2.1
Crocidura cf. m utesae
FS
16,1
0.3
2.3
G
Crocidura ludia
FS
6,5
0.6
1.3
70.7
24.3
F
Crocidura littoralis
FS
21,0
2.3
6.9
14.1
10.6
10.7
B
Crocidura cf. hildegardeae
FS
10,3
0.1
1.3
37.5
4.3 15.4
A
Crocidura cf. denti
FS
10,5
0.9
2.9
6.9
31.3
38.3
24.9
I
Crocidura olivieri
FS
32,8
9.4
0.3
8.6
3.1
6.4
5.9
N
Suncus megalura
FS
4,5
x
x
M
Suncus infinitesimus
FS
2,5
3.5
12.5
38.3
16.6
J
Crocidura roosevelti
FS
7,6
1.7
x
2.1
1.2
E
Crocidura lamottei
S
10,5
1.7
x
0.6
D
Crocidura fuscomurina
S
3,0
x
x
H
Crocidura nanila
S
2,5
x
x
x
K
Crocidura turba
S
17,0
x
x
L
Crocidura yankariensis
S
4,4
x
x
C
Crocidura cf. poensis
S
14,0
x
x
x
In comparison, among the five genera occurring within
However, it is important to note that, at Batouri, when the
the West Central region, only two (Crocidura and Suncus)
cumulated pitfall trapping effort reached the value
were recorded in the gallery forests. A lower number of
obtained at the northern galleries (5568 bucket-nights),
species was also recorded in the gallery forests (S=8).
four species had not been collected yet. None of the typi-

Shrews from gallery forests outside the Congo River basin
27
cal forest species of the genera Sylvisorex, Paracrocidura,
ests and the Congolese rainforest. Being amongst the
Suncus and Congosorex was recorded in the gallery for-
smallest mammals, shrews have high metabolic rates
ests. Suncus was represented by S. infinitesimus, previ-
resulting in high energy requirements and water loss
ously recorded from forest patches in forest-savanna
(CHURCHFIELD, 1990), especially the smallest species
mosaic, in eastern Congo and by S. megalura, which was
(VOGEL et al., 1981). These characteristics may be con-
only recorded within the savanna zone and is actually not
straining life histories to a greater extent than occurs in
a typical rainforest dweller but a widely distributed spe-
larger mammals (SYMONDS, 1999) and could explain the
cies (HUTTERER et al., 1987). Among the five Crocidura
differences observed between small mammals and larger
species recorded in both Congolese rainforest and gallery
ones. It is then possible that the typical rainforest shrews
forest, all have ecological preferences directed towards
would not have survived in the gallery forests because the
open rainforest or forest patches within forest-savanna
conditions would have been too dry and too hot in com-
mosaic, and are medium or large sized. Three species,
parison to the more clement conditions of the Congo
absent from the rainforest, were recorded within the gal-
River basin.
lery forests : C. roosevelti occurs at the margin of the
Differences in food resources could also have an influ-
rainforest in forest-savanna mosaic, and C. lamottei and
ence. Afro-tropical shrews, closely dependent on the
S. infinitesimus occur only in savannas or in savanna-for-
understorey leaf litter, feed mainly on small arthropods.
est mosaic. Despite the absence of the typical forest shrew
These shrews have a high level of food niche overlap and
fauna within the gallery forests, it is noteworthy that C.
very small prey is of greatest importance to the smaller
ludia, which is listed as vulnerable (B1+2c) by the IUCN
species (CHURCHFIELD et al., 2004; DUDU et al., in press).
(HILTON-TAYLOR, 2000), was common at Bohou.
Therefore, shrews would be more dependent on the cli-
In Afro-tropical primary lowland rainforest, shrew
matic conditions and food availability than larger mam-
communities, when surveyed by an adequate trapping
mals, and they could not have been able to survive in the
protocol (i.e. pitfall traps with drift fence, NICOLAS et al.,
warm and dry gallery forests in comparison to the Congo-
2003) appear to be dominated by one of the smallest spe-
lese rainforest. As previously suggested, central African
cies, such as Sylvisorex johnstoni in West Central Africa
endemic shrews are highly dependent on primary forest
(e.g. in Gabon : BARRIÈRE, 1997; GOODMAN et al., 2001;
environments (HUTTERER et al., 1987). The unsuitability
NICOLAS et al., 2004; in C.A.R. : BARRIÈRE et al., 2000; in
of these galleries to act as refuge for terrestrial small for-
Republic of Congo : BARRIÈRE, 1997; in Equatorial
est mammals could also be attributed to the small size of
Guinea : LASSO et al., 1996; and in Cameroon : HUTTERER
the patches, their peripheral location outside the present
& SCHLITTER, 1996); or by Crocidura obscurior (3.6 g of
Congolese rainforest, more than 400 km from its margin,
mean body weight) in West Africa (Ivory Coast : BAR-
and the time elapsed since their isolation from the Congo-
RIÈRE et al., 1999; CHURCHFIELD et al., 2004). By contrast,
lese rainforest. This is in agreement with hypotheses
in secondary forest or open habitats within the Congo
which suggest that refuges could have been much larger,
River basin, the dominant shrews are usually larger, such
close to one another, on the margins of the rainforest zone
as C. cf. hildegardeae (more than 10 g) in included
and in the downstream zone of large rivers (HUTTERER et
savanna at Odzala National Park, Republic of Congo
al., 1987; MALEY, 1996; DELEPORTE & COLYN, 1999).
(Marc Colyn & Patrick Barrière, unpublished data) and at
Ngotto forest, C.A.R. (BARRIÈRE et al., 2000), or such as
CONCLUSION
C. buettikoferi (11 g) in cacao-coffee plantations included
in the Taï National Park, Ivory Coast (BARRIÈRE et al.,
Altogether, these findings support the hypotheses that
1999). In the gallery forests of the Z.C.V., the tiny S.
Pleistocene refuges would have been composed of a net-
infinitesimus co-dominated (with C. cf. denti) the shrew
work of forest patches within a forest-savanna mosaic
community of Manovo-Mara and the small C. ludia
rather than of small homogenous forest isolates and local-
highly dominated the community of Bohou. Nevertheless,
ised in the downstream zone of large rivers. The knowl-
it is noteworthy that the medium-sized C. cf. hildegar-
edge of the location and size of refuges is essential for the
deae and C. cf. denti were co-dominant in Bamingui-Ban-
understanding of evolutionary scenarios. The refuges
goran and that C. cf. denti was co-dominant (which S.
could probably not maintain the whole mammalian fauna,
infinitesimus) in Manova-Mara.
but rather a limited number of taxonomic units with suita-
Among the 169 shrews, representing eight species, col-
ble anatomy, physiology and ecological preferences. The
lected in gallery forests within three main regions, none
knowledge of the history of tropical African floras and
of the typical rainforest shrews occurring within the
faunas will progress thanks to the multiplication of com-
Congo River basin was recorded. Similarly, no typical
parative phylogeographic studies (QUÉROUIL, 2001;
rainforest muroid rodent species was identified among a
QUÉROUIL et al., 2002; 2003) using relevant biological
collection of 449 individuals, at the present state of analy-
markers (COLYN & DELEPORTE, 2002a).
sis (Marc Colyn & Violaine Nicolas, unpublished data).
The gallery forests harbour several plant, bird, primate
and other larger mammal species, typical of the Congo-
ACKNOWLEDGEMENTS
lese rainforest, suggesting that these galleries may pres-
ently act as refuge for these forest taxa. However, it is not
Field studies were supported by EU-DGVIII-PDRN, "Pro-
gramme de Développement de la Région Nord"
the case for terrestrial small mammals. At first sight,
(NORCADEV); EU-DGVIII-Ecofac program, "Conservation et
these contrasting results are surprising. The observed dif-
Utilisation Rationnelle des Ecosystèmes Forestiers en Afrique
ferences may be attributed to distinct climatic and habitat
centrale", http.www.ecofac.org, (managed by AGRECO, GEIE,
characteristics as food resources, between the gallery for-
BDPA, SCETAGRI, SECA, CIRADFORET) and EU-DGVIII-

28
Patrick Barrière, Rainer Hutterer, Violaine Nicolas, Sophie Quérouil and Marc Colyn
Biofac program, "Origine et maintien de la biodiversité en Afri-
The Biology of the Soricidae II, Carnegie Museum of Natu-
que centrale" (University of Rennes 1-CNRS, UMR 6552).
ral History Special Publication.
P. Barrière and V. Nicolas received a fellowship from the Minis-
ECOFAC online : (http ://www.ecofac.org).
tère de l'Education Nationale de l'Enseignement Supérieur et de
ECOFAC (2001). Composante RCA, Projet de développement
la Recherche (France). We thank J. Lobão Téllo (PDRN) and
des zones cynégétiques villageoises. Rapport d'activités
A. Pénelon (ECOFAC) for their logistic support in the field,
juillet-décembre 2001 (http ://www.ecofac.org/Composan-
Olivier Perpète and the local teams of Sangba and Ngotto. We
tes/ZonesCynegetiquesVillageoises.htm).
ECOFAC-CEE
are also grateful to E. Verheyen and M. Dillen for their contribu-
report, Libreville : 116-137.
tion to molecular analyses.
FAY, M. (1988). Faunal monkey populations in the Central Afri-
can Republic : the northern limits. A census in Manovo-
REFERENCES
Gounda-St. Floris National Park. Mammalia, 52 : 57-74.
GOODMAN, S.M., R. HUTTERER & P.R. NGNEGUEU (2001). A
B
report on the community of shrews (Mammalia : Soricidae)
ARRIÈRE, P. (1997). Ecologie des Peuplements de Soricidae
Forestiers d'Afrique Centrale : considérations nouvelles sur
occurring in the Minkebe Forest, northeastern Gabon. Mam-
base de la méthode de piégeage Pitfall. Report to the Univer-
mal. Biol., 66 : 22-34.
sity of Rennes 1, Rennes : 26 pp.
HAFFER, J. (1969). Speciation in neotropical birds. Science, 165 :
BARRIÈRE, P., P. FORMENTY, R. HUTTERER, O. PERPÈTE & M.
131-137.
COLYN (1999). Shrew community from Taï forest (Côte
HILTON-TAYLOR, C. (2000). 2000 IUCN Red List of Threatened
d'Ivoire); ecological vigil searching the Ebola virus reser-
Species. IUCN, Gland, Switzerland and Cambridge, UK :
voir. Abstracts of the 8th International African Small Mam-
xviii + 61 pp.
mal Symposium, MNHN, Paris : 57.
HUTTERER, R., E. VAN DER STRAETEN & W.N. VERHEYEN (1987).
BARRIÈRE, P., V. NICOLAS, R.K. MARO, G. YANGOUNDJARA & M.
A checklist of the shrews of Rwanda and biogeographical
COLYN (2000). Ecologie et structuration des peuplements de
considerations of African Soricidae. Bonn. Zool. Beiträge,
micromammiferes musaraignes et rongeurs. ECOFAC-CEE
38 : 155-172.
report, Libreville : 55 pp.
HUTTERER, R. (1993). Order Insectivora. In : WILSON & REEDER
BERTAUX, J., D. SCHWARTZ, A. VICENS, A. SIFFEDINE, H.
(eds), Mammal species of the world, a taxonomic and geo-
ELENGA, M. MANSOUR, A. MARIOTTI, M. FOURNIER, L. MAR-
graphic reference, second edition, Washington, D.C. : 69-
TIN, D. WIRRMANN & M. SERVANT (2000). Enregistrement de
130.
la phase sèche d'Afrique centrale vers 3000 ans B.P. par la
HUTTERER, R., & D.A. SCHLITTER (1996). Shrews of the Korup
spectrométrie IR dans les lacs Sinnda et Kitina (sud-Congo).
National Park, Cameroon, with the description of a new Syl-
In : SERVANT & SERVANT-VILDARY (eds), Dynamique à long
visorex (Mammalia : Soricidae). Spec. Publ. Mus. Texas
terme des écosystèmes forestiers intertropicaux, Mémoires
Tach. Univ., : 57-66.
UNESCO, Paris : 43-49.
LASSO, C., R. HUTTERER & A. RIAL (1996). Records of shrews
BOULVERT, Y., (1986). République centrafricaine, carte phy-
(Soricidae) from Equatorial Guinea, especially from Monte
togéographique. Editions de l'ORSTOM, Paris : 131 pp.
Alen National Park. Mammalia, 60 : 69-76.
CHURCHFIELD, S. (1990). The natural history of shrews. CHRIS-
LEAL M. (2000). Microrefugia, small scale ice age forest rem-
TOPHER HELM (ed), London : 175 pp.
nants. In : ROBBRECHT & DEGREEF (eds), Abstracts XVIth
CHURCHFIELD, S., P. BARRIÈRE, R. HUTTERER & M. COLYN
AETFAT Congress. Scripta Bot. Belg. : 3.
(2004). First results on the feeding ecology of sympatric
M
shrews (Insectivora : Soricidae) in the Tai National Park,
ALEY J. (1987). Fragmentation de la Forêt Dense Humide
Africaine et extension des biotopes montagnards au Quater-
Ivory Coast. Acta Theriologica, 49(1) : 1-15.
naire récent : nouvelles données polliniques et chronologi-
CHRISTY, P. (1999). La République Centrafricaine, carrefour
ques. Implications paléoclimatiques et biogéographiques.
biogéographique. Canopée, Bulletin sur l'Environnement en
Palaeoecology of Africa 18 : 307-334.
Afrique Centrale, 14 : 6.
C
MALEY, J. (1996). The African rain forest - main characteristics
OLYN, M. (1991). L'importance zoogéographique du bassin du
fleuve Zaïre pour la spéciation : le cas des primates simiens.
of changes in vegetation and climate from the Upper Creta-
Ann. Mus. R. Afr. Centr., Tervuren, Belgique : 264.
ceous to the Quaternary. Proc. R. Soc. Edinburgh 104B : 31-
C
73.
OLYN, M. (1999). Un nouveau statut biogéographique pour
l'Afrique centrale. Canopée, Bulletin sur l'Environnement
MALEY, J. (2001). La destruction catastrophique des forêts
en Afrique Centrale, 14 : 3-5.
d'Afrique centrale survenue il y a environ 2500 ans exerce
C
encore une influence majeure sur la répartition actuelle des
OLYN, M., & P. DELEPORTE (2002a). Marqueurs biologiques en
biogéographie historique : problèmes de taxonomie, de dis-
formations végétales. Syst. Geogr. Pl., 71 : 777-796.
tribution et de filtre écologique. Biosystema, 20 : 41-48.
MALEY, J., & P. BRENAC (1998). Vegetation dynamics, palaeoen-
COLYN, M., & P. DELEPORTE (2002b). Biogeographic analysis of
vironments and climatic changes in the forests of west Cam-
central African forest Guenons. In : GLENN & CORDS (eds),
eroon during the last 28,000 years. Rev. Palaeobot. Palyno.,
The Guenons : Diversity and Adaptation in African Mon-
99 : 157-188.
keys., Kluwer Academic/Plenum Publishers, New-York : 61-
NICOLAS, V., P. BARRIÈRE & M. COLYN (2003). Impact of
78.
removal pitfall trapping on the community of shrews
COLYN, M., A. GAUTHIER-HION & W. VERHEYEN (1991). A re-
(Mammalia : Soricidae) in two African tropical forest sites.
appraisal of the paleoenvironment history in Central Africa :
Mammalia, 67 : 133-138.
evidence for a major fluvial refuge in the Zaire basin. J. Bio-
NICOLAS, V., P. BARRIÈRE, S. GUIMONDOU & M. COLYN (2004).
geogr., 18 : 403-407.
Variabilité structurale des peuplements forestiers de ron-
DELEPORTE, P., & M. COLYN (1999). Biogéographie et dynami-
geurs (Muridae) et musaraignes (Soricidae) dans les Monts
que de la biodiversité : application de la PAE aux forêts pla-
Doudou, Gabon. In : FISCHER (ed), A floral and faunal inven-
nitiaires d'Afrique centrale. Biosystema, 17 : 37-43.
tory of Monts Doudou : Gabon with reference to elevation
DUDU, A., S. CHURCHFIELD & R. HUTTERER (in press). Commu-
variation, Proceedings of the California Academy of Sci-
nity structure and food niche relationships of coexisting rain-
ences, Memoir 28, San Francisco : 107-127.
forest shrews in the Masako Forest, north-eastern Congo.
QUÉROUIL, S. (2001). Intérêts et limites de l'approche molécu-
In : MERRITT, CHURCHFIELD, HUTTERER & SHEFTEL (eds),
laire pour aborder la biogéographie et la spéciation :

Shrews from gallery forests outside the Congo River basin
29
quelques exemples chez les mammifères d'Afrique tropicale.
into the systematics of African Crocidura (Crocidurinae,
Ph.D. thesis, University of Rennes 1, France : 310 pp.
Soricidae) using 16s rRNA sequences. In : MERRITT,
QUÉROUIL, S., R. HUTTERER, P. BARRIÈRE, M. COLYN, J.C. KER-
CHURCHFIELD, HUTTERER & SHEFTEL (eds), The Biology of
BIS PETERHANS & E. VERHEYEN (2001). Phylogeny and evo-
the Soricidae II, Carnegie Museum of Natural History Spe-
lution of African Shrews (Mammalia : Soricidae) inferred
cial Publication.
from 16s rRNA sequences, Mol. Phylogenet. Evol., 20(2) :
RAY, J.C. & R. HUTTERER (1996). Structure of a shrew commu-
185-195.
nity in the Central African Republic based on the analysis of
QUÉROUIL, S., M. COLYN, E. VERHEYEN & P. DELEPORTE (2002).
carnivore scats, with the description of a new Sylvisorex
De la phylogéographie comparée à la biogéographie
(Mammalia : Soricidae). Ecotropica, 1 : 85-97.
historique : l'exemple de quelques mammifères forestiers
d'Afrique centrale. Biosystema 20, Systématique et
SYMONDS, M.R.E. (1999). Life histories of the Insectivora : the
role of phylogeny, metabolism and sex differences. J. Zool.
Biogéographie : 131-140.
London, 249 : 315-337.
QUÉROUIL, S., E. VERHEYEN, M. DILLEN & M. COLYN (2003).
Patterns of diversication in two African forest shrews : Sylvi-
TÉLLO, J.L. (2000). Inquiétante recrudescence du braconnage
sorex johnstoni and Sylvisorex ollula (Soricidae, Insectivora)
des éléphants dans le nord de la centrafrique. Canopée, Bul-
in relation to paleo-environmental changes. Mol. Phylo-
letin sur l'Environnement en Afrique Centrale, 17 : 11-12.
genet. Evol., 28 : 24-37.
VOGEL, P., M. GENOUD & H. FREY (1981). Rythme journalier
QUÉROUIL, S., P. BARRIÈRE, M. COLYN, R. HUTTERER, A. DUDU,
d'activité chez quelques Crocidurinae africains et européens
M. DILLEN & E. VERHEYEN (in press). A molecular insight
(Soricidae, Insectivora). Rev. Ecol. (Terre Vie), 35 : 97-108.


Belg. J. Zool., 135 (supplement) : 31-38
December 2005
The fibrovascualar ring : A synapomorphy of hystricog-
nath Rodentia newly described in Petromus typicus and
Octodon degus

Andrea Mess
Institut für Systematische Zoologie, Museum für Naturkunde, Humboldt-Universität zu Berlin, Invalidenstr. 43, D-10115 Ber-
lin, Germany
Corresponding author : Andrea Mess, e-mail : andrea.mess@rz.hu-berlin.de
ABSTRACT. Hystricognathi is a higher monophyletic taxon within Rodentia that is supported particularly by char-
acters of early ontogeny and placentation. Since the original findings are based on a small sample of species, further
information on character distribution is desirable. The present paper provides new insight into the morphology of
one of the most significant characters in two additional species of Hystricognathi, i.e. Petromus typicus A. SMITH,
1831 and Octodon degus MOLINA, 1782. In both species an arterial ring within the inverted yolk sac splanchno-
pleura that is associated with a network of capillaries is present throughout pregnancy. Consequently, the existence
of the so called "fibrovascular ring" is confirmed for these species and distributional data suggest that it can be
regarded as a synapomorphic feature of Hystricognathi. Thus, two of the three original diagnostic characters of
Hystricognathi are now confirmed for a larger set of taxa.
KEY WORDS : Rodentia, Hystricognathi, phylogeny, ontogeny, fetal membrane structures, yolk sac vessels, sinus ter-
minalis, fibrovascular ring.
INTRODUCTION
2002, 2005; MESS et al., 2002) were obtainable. Publica-
tions on fetal membrane structure resulting from this
Hystricognathi is a well established taxon within
work focussed on the structural organisation of the chorio-
Rodentia (TULLBERG, 1899/1900; SIMPSON, 1945; EISEN-
allantoic placenta (cf. MESS, 2003), whereas other struc-
BERG, 1981; LUCKETT & HARTENBERGER, 1985, 1993;
tures such as the yolk sac are currently under investiga-
WILSON & REEDER, 1993; MCKENNA & BELL, 1997; NED-
tion. At present, one of the original defining characters of
BAL et al., 1994; HUCHON et al., 1999, 2000). It is sup-
Hystricognathi is found in the two investigated species :
ported as a monophylum particularly by characters of
the subplacenta as a distinct region within the chorioal-
early ontogeny and placentation as summarised by
lantoic placenta (MESS, 2003). The present work provides
PATRICK LUCKETT in 1985 and widely accepted by the
insights into another one of the defining characters of
scientific community. However, these important findings
Hystricognathi according to LUCKETT'S review, i.e. the
are based on only 8 species of Hystricognathi, which have
fibrovascular ring (following the nomenclature of PER-
been more or less well studied and often not including
ROTTA, 1959) or capillary band (according to LUCKETT &
detailed descriptions of the morphological context (cf.
MOSSMAN, 1981). This structure represents a network of
LUCKETT, 1985; MOSSMAN, 1987). Moreover, members of
capillaries within the inverted yolk sac splanchnopleura
several larger groups of Hystricognathi are not included
that is associated with a ring-like artery. It is known only
in that sample (see below). Thus, further investigations
for Hystricognathi. The main focus herein is a description
are necessary in order to reveal the distribution of fetal
of the morphology of the ring system in the two species.
membrane characters. Accordingly, in the last couple of
Finally, its evolutionary and phylogenetic significance is
years investigation of fetal membrane structures in two
discussed too.
additional species of Hystricognathi have been con-
ducted, based on breeding groups of the South African
dassie rat (Petromus typicus) and the South American
MATERIAL AND METHODS
degu (Octodon degus). Both belong to subgroups of Hys-
tricognathi that have been suggested to represent most
To reveal the occurrence of the fibrovascular ring and
probably basal offshoots of the African and South Ameri-
the associated ring-like artery, either vascular injections
can hystricognaths, respectively (see MESS, 1999a for
of the yolk sac vessels or else histological sections have
background information). Thus, investigation of their
to be conducted. The later method is more often used, and
fetal membrane structures is essential to a fuller under-
thus the present study is based on histological serial sec-
standing of the evolutionary differentiation of Hystricog-
tions of Petromus typicus and Octodon degus. All mate-
nathi. Especially for Petromus, it is only recently that
rial was obtained from the breeding groups of both spe-
information on placentation (e.g., MESS, 1999b, 2001,
cies, housed at the Humboldt-University, Berlin.
2003) or even basic data on reproductive biology (MESS,
Information on the examined stages is given in the text or

32
Andrea Mess
else is provided in MESS (2003). The study focuses on
rams by using MacClade (Version 4.0)1. Since opinions
that stages that includes the ring in total in order to follow
on the phylogenetic relationships of Hystricognathi to
the course of the vessels and to reveal if the fibrovascular
other Rodentia varies, members of each of the main
ring and the artery possess a ring-like structure. In some
rodent subgroups are included as out-groups, and two
of the later stages that are also used for this study, only
independently established trees (derived from NEDBAL et
one half of the placenta was prepared for light micros-
al., 1994, 1996, and MCKENNA & BELL, 1997) are used
copy, whereas the other parts have been used for electron
(see discussion). Accordingly, the stem species pattern of
microscopy or other applications. Thus, in some of these
Rodentia (= the character set of the last common ancestor
cases additional information is derived from manually
of all rodents) is reconstructed as a first step towards rec-
prepared specimens. The histological sections have been
ognising the evolutionary transformations on the stem lin-
analysed with Zeiss Axioskop or Axioplan microscopes
eage of Hystricognathi. Data on the character distribution
and the structures of interest have been documented by
in relevant taxa is summarised in Table 1, and the results
using a Camera Lucida. The reconstruction of character
of the MacClade analysis (including character treatment,
evolution was done on the basis of pre-existing cladog-
CI and RI values) are given by Fig. 3.
TABLE 1
Distribution of the fibrovascular ring system within the inverted yolk sac splanchnopleura and its development
in rodent species and some members of other eutherian orders, derived from the literature and own material.
Short names (*) are given for each species, which are used to demonstrate the results of the analysis in Fig. 3.
Character I ­ Ring-like artery : 1 = absent, 2 = present.
Character II ­ Network of capillaries : 1 = absent, 2 = present.
Character III ­ Extent of the ring system : 1 = absent, 2 = not prominent, 3 = prominent.
Species
(*)
I
II
III
Main citation
Petromus typicus
Pet
2
2
2
MESS (2003), present study
Octodon degus
Oct
2
2
2
MESS (2003), present study
Cavia porcellus
Cav
2
2
3
Several studies, see MOSSMAN (1987); own material
Erethizon dorsatum
Ere
2
2
3
PERROTTA (1959)
Myocastor coypus
Myo
2
2
3
HILLEMANN & GAYNOR (1961)
Chinchilla lanigera
Chin
2
2
3
TIBBITS & HILLEMANN (1959)
Dasyprocta spec.
Dasy
2
2
3
BECHER (1921a, b); MIGLINO (pers. comm.)
Thryonomys swinderianus
Thry
2
2
3
ODUOR-OKELE & GOMBE (1982, 1991)
Hystrix africaeaustralis
Hyst
2
2
3
LUCKETT & MOSSMAN (1981)
Bathyergus janetta
Bath
2
2
3
LUCKETT & MOSSMAN (1981)
Rattus norvegicus
Rat
1
1
1
e.g. BRIDGEMAN (1948); own material
Mus musculus
Mus
1
1
1
e.g. THEILER (1972); own material
Jaculus jaculus
Jac
1
1
1
KING & MOSSMAN (1974)
Aplodontia rufa
Aplo
1
1
1
HARVEY (1959a)
Citellus tridecemlineatus
Cite
1
1
1
e.g. MOSSMAN & WEISFELDT (1939)
Sciurus vulgaris
Sci
1
1
1
e.g. SCHOOLEY (1934); own material
Geomys bursarius
Geo
1
1
1
MOSSMAN & HISAW (1940); MOSSMAN & STRAUSS (1963)
Castor canadensis
Cast
1
1
1
WILLEY (1912); FISCHER (1971)
Pedetes capensis
Ped
1
1
1
FISCHER & MOSSMAN (1969); OTIANG'A-OWITI et al. (1992)
Oryctolagus cuniculus
Ory
1
1
1
MOSSMAN (1926, 1987); own material
Ochotona spec.
Och
1
1
1
HARVEY (1959b)
RESULTS
running in the yolk sac splanchnopleura along the
umbilical cord (Figs 1b, c). At the outer border of the
1. The fibrovascular ring in Petromus typicus
chorioallantoic placenta the branches are fused to each
other (Figs 1a, d), assuming a ring-like structure of the
In the youngest investigated stage of Petromus typi-
cus of about 5 weeks of pregnancy, an arterial ring is sit-
artery. From the arterial ring, smaller arteries branch out
uated within the yolk sac splanchnopleura where the lat-
in their way to supply the extended area of the yolk sac
ter is attached to the chorioallantoic placenta (Figs 1a-
splachnopleura. In Fig. 1c, such an offshoot of the arte-
d). According to the inverted nature of the yolk sac in
rial ring is given on the right side. The artery is accom-
Hystricognathi and other rodents, the yolk sac vessels
panied by a dense network of capillaries, the fibrovascu-
are situated on the inner side, whereas the yolk sac
lar ring or capillary band (Figs 1a-d). As, for instance, in
endoderm is on the outer side in close vicinity to the
the guinea pig, the ring-like artery is intermingled
uterus (Figs 1a-d). The artery possesses two branches
within these capillaries.
1 The number of characters is too small to develop an own phylogeny, e.g. by applying Paup.

The fibrovascular ring in hystricognath rodents
33
Fig. 1. ­ The fibrovascular ring in different ontogenetic stages of the African Petromus typicus.
A-D : about 5 weeks of pregnancy, early placental differentiation stage
(corresponding to Petromus 26 in MESS (2003)).
E : about 7 to 8 weeks of pregnancy, mid gestation
(corresponding to Petromus 19b).
F : about 12 weeks of pregnancy, near-term stage
(corresponding to Petromus 1).
*Abbrevations see Fig. 2.

34
Andrea Mess
In older stages of Petromus of either 7 to 8 weeks of
every specimen varies within the yolk sac splanchnop-
pregnancy (mid gestation) as well as in several near-term
leura, the capillary band or fibrovascular ring is, even in
stages (about 12 weeks of pregnancy), the arterial system
near-term stages, not very prominent at all (Fig. 1f shows
is in a similar position (Figs 1e, f). Most of the investi-
a portion of the object with a quite well differentiated
gated material allows no clear decision about whether it is
capillary network). Laterally the artery and the associated
ring shaped. However, at least in one near-term stage, a
fibrovascular ring now reach to that region of the yolk sac
ring-like structure of the artery and the associated fibrov-
splanchnopleura where the infolding took place (Fig. 1f).
ascular ring or capillary band is clearly established. A
ring-like structure of the artery also appears in a macro-
2. The fibrovascular ring in Octodon degus
scopically prepared specimen. The later stages are charac-
terised by the following : In the mid-term stage a mod-
As in Petromus, Octodon has an arrangement of an
estly prominent artery is situated within the yolk sac
artery and a network of capillary in the inverted yolk sac
splanchnopleura near to the attachment of the chorioal-
splanchnopleura throughout pregnancy. The first indica-
lantoic placenta (Fig. 1e). It is in close association with
tion of the structures of interest can be noticed in an early
the relatively small, undifferentiated network of capillar-
stage of 25 to 26 days of pregnancy, where the vasculari-
ies (Fig. 1e). This capillary band (or fibrovascular ring) is
sation of the yolk sac splanchnopleura begins2. In the next
more laterally extended along the yolk sac splanchnop-
stage of 35 days of pregnancy, the arterial system can
leuric surface than in the earlier stage. The artery and its
clearly be recognised (Fig. 2a). The artery is quite small
offshoots are interrelated to this network of capillaries as
and is situated very near to the attachment of the yolk sac
described above. Since in the mid-term stage an infolding
splanchnopleura to the chorioallantoic placenta. It pos-
of the yolk sac splanchnopleura begins, it should be
sesses a ring-like structure, and supplies smaller arterial
assumed that the above described arterial system does not
vessels that run along the surface of the yolk sac splanch-
reach far laterally to that region of the splanchnopleura
nopleura (Fig. 2a). The accompanying capillaries are like-
(cf. Fig. 1e). Finally, in near-term stages the artery is more
wise relatively small and not markedly differentiated, but
distant to the chorioallantoic placenta than in former
clearly intermingled with the artery (Fig. 2a). The artery
stages (Fig. 1f). Also in these term stages the artery is
and the related capillary band or fibrovascular ring is in
accompanied by the network of capillaries. Although the
some distance from the region where the infolding of the
distinctness and lateral extension of the capillaries in
yolk sac splanchnopleura begins (Fig. 2a).
Fig. 2. ­ The fibrovascular ring in the South American hystricognath rodent Octodon degus.
A : about 5 weeks of pregnancy, placental differentiation stage (corresponding to Octodon 17
in MESS (2003)).
B : late pregnancy, 64 days (corresponding to Octodon 10).
Abbreviations in Figs 1-2 (scale bar = 1 mm) :
1 : yolk sac splanchnopleura
2a : ring-like artery in the splanchnopleura
2b : lateral offshoot of the artery
3 : capillary band or fibrovascular ring associated with the artery
4 : chorioallantoic placenta
5 : uterus
6 : umbilical cord
2 However, the specimen was fixed only with formaldehyde in total, and thus the structures are not very distinctly preserved, and thus a detailed
description of this stage is impractical.

The fibrovascular ring in hystricognath rodents
35
In older stages of Octodon of either 64 days of preg-
MOSSMAN, 1987, original description by BECHER, 1921a,
nancy or 84 days (near-term stage), a quite similar picture
b). Finally, in the material of some newly described South
is recognisable : The artery is still small, as well as the
American species the fibrovascular ring system seem also
associated capillary band which is only formed by a lim-
be present, e.g. in Kerodon rupestris and Agouti paca
ited number of individual capillaries (Fig. 2b). These
(MIGLINO, pers. comm.). In all these species, the arterial
structures are situated very near to the attachment to the
ring and its associated capillary band are characteristi-
chorioallantoic placenta even in advanced developmental
cally similar to that of the guinea pig. Moreover, as far as
stages (Fig. 2b). In macroscopically studied material sim-
described by the authors, it reached laterally at least to the
ilar in age, a ring-like structure of the artery is likely.
beginning of the villous region of the yolk sac splanchno-
pleura. The only difference within the taxa sample is the
DISCUSSION
fact, that the ring system is not fully circular in some spe-
cies such as Hystrix (LUCKETT & MOSSMAN, 1981).
In Petromus typicus as well as in Octodon degus an
Petromus and Octodon clearly fit into the described pat-
artery within the inverted yolk sac splanchnopleura associ-
tern, and thus the presence of the fibrovascular ring is
ated with a network of capillaries, the capillary band or
established for the species investigated herein. In non-
fibrovascular ring is established throughout pregnancy.
hystricognath rodent taxa, an arterial and capillary ring
Both species possess ring-like arteries : In the youngest
system has not been found in similar position so far (see
stage of Petromus, a nearly complete ring-like structure of
Table 1, e.g. WILLEY, 1912; MOSSMAN, 1926, 1987;
the artery appears to be present since its two branches are
SCHOOLEY, 1934; MOSSMAN & WEISFELD, 1939; MOSS-
fused to each other in the histological serial section. At
MAN & HISAW, 1940; BRIDGEMAN, 1948; HARVEY, 1959a,
least in one of the near-term stages as well as in a manually
b; MOSSMAN & STRAUSS, 1963; FISCHER & MOSSMAN,
prepared specimen, a ring-like structure is present, sug-
1969; FISCHER, 1971; THEILER, 1972; KING & MOSSMAN,
gesting that it is present throughout pregnancy. In Octodon
1974; LUCKETT, 1985; OTIANG'A-OWITI et al., 1992).
the ring-like structure of the artery and its capillary net-
The functional meaning of this structure is completely
work occurs in histological sections of the 5 week old stage
unknown, as well as its origin during vascularisation of
and in manually prepared specimens of various ages. In
the yolk sac splanchnopleura or the possible evolutionary
both species the fibrovascular ring is situated near the
precursors (LUCKETT & MOSSMAN, 1981; MOSSMAN,
attachment site towards the chorioallantoic placenta, dis-
1987). However, the presence of the fibrovascular ring is
tinct from the villous splanchnopleura that has differenti-
essential for phylogeny, since it is one of three synapo-
ated during later ontogeny. However, in all investigated
morphies of Hystricognathi given by LUCKETT in 1985,
serial sections of near-term stages, the capillary band is not
especially as characters that are uniquely derived in Afri-
as prominent as is the case in late pregnancy stages of other
can and American hystricognath rodents (LUCKETT &
Hystricognathi such as Cavia or Hystrix (see Table 1).
MOSSMAN, 1981). With the discovery of these characters,
A fibrovascular ring was first described as a dense net-
the long-lasting dispute about the independent develop-
work of capillaries that is accompanied by a complete or
ment of Hystricognathi from different continents was sett-
nearly complete ring-like artery supplied by the yolk sac
led at least in the mid 1980's (LUCKETT & HARTENBERGER,
artery in the guinea pig Cavia porcellus by RUTH JACK-
1985). Information on the distribution of the fibrovascular
SON, later MOSSMAN (R.J. MOSSMAN, 1927, cited after
ring within hystricognaths is now accessible for a larger
LUCKETT & MOSSMAN, 1981; MOSSMAN, 1987). Hence,
set of taxa, i.e. more than ten species altogether compris-
this ring-like structure within the yolk sac splanchnop-
ing also members of subgroups that are supposed to rep-
leura surrounds the attachment of the umbilical cord and
resent basal offshoots.
radiates into finer vessels that supply the yolk sac surface.
A reconstruction of character evolution was carried out
Following the original discovery, the ring system was
by applying MacClade on the basis of pre-existing hypoth-
noticed in illustrations derived from former studies on
eses of rodent systematic. Though several molecular phyl-
fetal membranes in Cavia, although not mentioned in the
ogenies are available (e.g., HUCHON et al., 2000 (nuclear
text (relevant citations in MOSSMAN, 1987). Moreover, the
gene, i.e. vWF); ADKINS et al., 2001 (combined analysis on
existence of such an arterial ring system appears to be
several nuclear and rRNA genes)), most of these studies
present in other members of Hystricognathi that have
cannot be used, because they consider small taxa samples
been investigated in regard to their fetal membrane struc-
that usually do not include Petromus (or even Octodon).
tures, i.e. in Erethizon dorsatum (PERROTTA, 1959 :
Thus, following MESS (2003), a cladogram based on 12S-
"Fibrovascular ring"), Chinchilla lanigera (TIBBITTS &
rRNA genes (NEDBAL et al., 1994, 1996) is chosen in pref-
HILLEMANN, 1959 : "Sinus terminalis", network of capil-
erence to reconstruct character evolution. Moreover, an
laries and multiple anastomosis), Myocastor coypus (HIL-
independently established, morphology-based classifica-
LEMANN & GAYNOR, 1961 : "Sinus terminalis"), Hystrix
tion (MCKENNA & BELL, 1997) is considered too. Rodentia
africaeaustralis and Bathyergus janetta (LUCKETT &
and members of Lagomorpha as an additional out-group
MOSSMAN, 1981 : "Sinus terminalis or arterial circle with
are used in this tree, because the evolution of yolk sac
capillary band), Thryonomys swinderianus (ODUOR-
characters is not resolved by considering only rodents (see
OKELO & GOMBE, 1982, 1991 : "Fibrovascular ring",
MESS, 2003)3. In both of the cladograms under considera-
ODUOR-OKELO, pers. comm), and Dasyprocta azarae (cf.
tion, only those species can be utilised for which sufficient
3 MCKENNA & BELL (l.c.) present Lagomorpha and Macroscelidea as nearest relatives of Rodentia. The results from the analysis do not change, if a
member of Macroscelidea had been chosen instead of the lagomorphs.





36
Andrea Mess
data about fetal membrane structures are available, result-
ing in a restricted number of species and relationships
compared to the original trees (see Table 1 and Fig. 3)4.
Characters 1 and 2 are respectively the presence or absence
of the ring-like artery and the capillary network. Addition-
ally, the extent of the fibrovascular ring system is included
(character 3), i.e. the dorsolateral extent of the capillary
band and the intermingled artery in the yolk sac splanchno-
pleura based on a subjective impression. The characters are
treated as unordered. Accordingly, the presence of the ring-
like artery (character 1) and the associated network of cap-
illaries (character 2) in the yolk sac splanchnopleura
resulted as evolutionary transformations towards Hystri-
cognathi in both of the underlying trees, and thus the
fibrovascular ring system appears to be a synapomorphic
or derived character state for hystricognaths (Figs 3a, c).
Since two independently established trees support the same
character polarity, the presumed character evolution
appears relatively stable. Finally, in regard to character 3,
the molecular tree indicated a prominent fibrovascular ring
system in the stem species pattern of Hystricognathi as a
derived condition within Rodentia, and independent trans-
formations from the hystricognath stem species pattern
towards the inconspicuous conditions in the newly investi-
gated Petromus and Octodon (Fig. 3b). In contrast, the
morphology-based tree allows no establishment of this
character polarity within hystricognaths (Fig. 3d). Thus,
Fig. 3. ­ Results of the MacClade analysis on the two selected
the available results only tentatively suggest that Petromus
cladograms, i.e. a molecular phylogeny (NEDBAL et al., 1994,
and Octodon have (independently) reduced the extent of
1996) and morphology-based relationships (MCKENNA & BELL,
their fibrovascular ring system, although the presumed
1997). The characters are treated as unordered. CI = 0.8, RI =
character polarity appears not to be very stable.
0.96 in both trees. See Table 1 for abbreviations and full names
of the selected taxa.
A : Evolution of characters 1 and 2 after NEDBAL et al.
B : Evolution of character 3 after NEDBAL et al.
C : Evolution of characters 1 and 2 after MCKENNA & BELL
D : Evolution of character 3 after MCKENNA & BELL
CONCLUSION
In conclusion, the fibrovascular ring is revealed for a
larger set of taxa within hystricognaths and the distribu-
tional data suggest that it can be regarded as homologous
within hystricognaths and as a synapomorphic feature of
that group. The present findings consolidate LUCKETT'S
hypothesis. Thus, up to now two of the three defining
characters of Hystricognathi ­ the fibrovascular ring and
the subplacenta ­ have been confirmed for more taxa.
Very early ontogenetic stages of Petromus and Octodon
have not been studied so far, and thus, no information is
available on the third synapomorphy of Hystricognathi
according to LUCKETT (1985), the primary interstitial
implantation.
ACKNOWLEDGEMENTS
The present work is derived from a contribution to the 9th
International African Small Mammal Symposium which was
held in Morogoro / Tanzania in July 2003. Above all, I wish to
4 Compared to the above mentioned study (MESS, 2003), some more Hystricognathi are included, and some non-hystricognath species are omitted (the
data are more homogenous than that of the chorioallantoic placenta).

The fibrovascular ring in hystricognath rodents
37
thank the organisers for carrying out this stimulating meeting as
tionary Relationships among Rodents, Plenum Press, New
well as for the chance to publish symposiums proceedings. The
York, NATO ASI-Series 92 : 227-276.
establishment of the breeding groups of Petromus typicus and
LUCKETT, P. & J.-L. HARTENBERGER (1985). Evolutionary rela-
Octodon degus as well as the gathering and processing of pla-
tionships among rodents : comments and conclusions. In :
cental material took place at the Humboldt-University, Berlin
LUCKETT & HARTENBERGER (eds), Evolutionary Relation-
within the Institute of Systematic Zoology of the Museum of
ships among Rodents, Plenum Press, New York, NATO ASI-
Natural History. Consequently, I would like to thank the director
Series 92 : 685-712.
of that institution, Prof. U. Zeller, and all colleagues that are
LUCKETT, W.P. & J.-L. HARTENBERGER (1993). Monophyly or
involved in the different stages of this project. Prof. H. Hoch and
Polyphyly of the Order Rodentia : Possible Conflict
her team enabled the use of MacClade. Moreover, I am grateful
Between Morphological and Molecular Interpretations. J.
to Manfred Ade for comments on earlier versions of the manu-
Mamm. Evol., 1 : 127-147.
script, and to Patrick Luckett for discussions on the Petromus
LUCKETT, W.P. & H.W. MOSSMAN (1981). Development and
material during his stay in Berlin a few years ago. Finally, I want
phylogenetic significance of the fetal membranes and pla-
to thank Jason Dunlop for helping with the English, and an
centa of the African hystricognathous rodents, Bathyergus
unknown referee for helpful comments on a former version of
and Hystrix. Am. J. Anat., 162 : 266-285.
the manuscript.
MCKENNA, M.C. & S.K. BELL (1997). Classification of mam-
mals above the species level. Columbia University Press,
LITERATURE CITED
New York.
MESS, A. (1999a). The rostral nasal skeleton of hystricognath
A
rodents : evidence on their phylogenetic interrelationships.
DKINS, R.M., E.L. GELKE, D. ROWE & R.L. HONEYCUTT
(2001). Molecular Phylogeny and divergence time estimates
Mitt. Mus. Naturk. Berlin, Zool. Reihe, 75 : 19-35.
for major rodent groups : evidence from multiple genes.
MESS, A. (1999b). Plazentation von Petromus typicus, ein Bei-
Mol. Biol. Evol., 18 : 777-791.
trag zur Reproduktionsbiologie und evolutiven Differen-
B
zierung der Hystricognathi. Courier Senckenberg, 215 : 153-
ECHER, H. (1921a). Die Entwicklung des Mesoplacentariums
und die Placenta bei Agouti (Dasyprocta azarae Schl.). Z.
157.
Anat. Entw.-gesch., 61 : 337-364.
MESS, A. (2001). Evolutionary differentiation of placental
B
organisation in hystricognath rodents. In : DENYS, GRANJON
ECHER, H. (1921b). Der feinere Bau der reifen Plazenta von
Agouti (Dasyprocta azarae Schl.). Z. Anat. Entw.-gesch.,
& POULET (eds), African Small Mammals, IRD Éditions, col-
61 : 439-454.
lection Colloques et séminaires. Paris : 279-292.
B
M
RIDGEMAN, J. (1948). A morphological study of the develop-
ESS, A. (2002). Petromus typicus : reproductive biology of a
ment of the placenta of the rat. I. An outline of the develop-
poorly known animal. Research Committee Newsletter, 8 :
ment of the placenta of the white rat. J. Morph., 83 : 61-85.
38-40.
E
M
ISENBERG, J.F. (1981). The mammalian radiation. An analysis
ESS, A. (2003). Evolutionary transformations of chorioallan-
of trends in evolution, adaptation, and behaviour. Athone
toic placental characters in Rodentia with special reference
Press, London.
to hystricognath species. J. Exp. Zool., 299A : 78-98.
FISCHER, T.V. (1971). Placentation in the American beaver (Cas-
MESS, A. (2005) : Felsenratte. In : GANSLOßER, U. (ed), Die
tor canadensis). Am. J. Anat., 131 : 159-184.
Sozialsysteme der Säugetiere, Filander Verlag, Fürth : 156-
FISCHER, T.V. & H.W. MOSSMAN (1969). The fetal membranes of
158.
Pedetes capensis, and their taxonomic significance. Am. J.
MESS, A., J. PINT & K. FAUST (2002). Ontogenetic differentia-
Anat., 124 : 89-116.
tion in Petromus typicus : a basis for more detailed interpre-
HARVEY, E.B. (1959a). Placentation in Aplodontidae. Am. J.
tation of results on ultrasonic approaches in a precocial
Anat., 105 : 63-89.
rodent. Advances in Ethology, 37 : 145.
HARVEY, E.B. (1959b). Placentation in Ochotonidae. Am. J.
MOSSMAN, H.W. (1926). The rabbit placenta and the problem of
Anat., 104 : 61-85.
placenta transmission. Am. J. Anat., 37 : 433-497.
HILLEMANN, H.H. & A.I. GAYNOR (1961). The definitive archi-
MOSSMAN, H.W. (1987). Vertebrate Fetal Membranes. Compar-
tecture of the placentae of Nutria, Myocastor coypus
ative Ontogeny and Morphology, Evolution, Phylogenetic
(Molina). Am. J. Anat., 109 : 299-317.
Significance, Basic Function Research Opportunities. Rut-
HUCHON, D., F.M.CATZEFLIS & E.J.P. DOUZERY (1999). Molecu-
gers University Press, New Brunswick, New Jersey.
lar Evolution of the Nuclear von Willebrand Factor Gene in
MOSSMAN, H.W. & F.L. HISAW (1940). The fetal membranes of
Mammals and the Phylogeny of Rodents. Mol. Biol. Evol.,
the pocket gopher illustrating an intermediate type of rodent
16 : 577-587.
membrane formation. I. From the unfertilized tubal egg to
HUCHON, D., F.M. CATZEFLIS & E.J.P. DOUZERY (2000). Vari-
the beginning of the allantois. Am. J. Anat., 66 : 367-391.
ance of molecular datings, evolution of rodents and the phyl-
MOSSMAN, H.W. & F. STRAUSS (1963). The fetal membranes of
ogenetic affinities between Ctenodactylidae and Hystricog-
the pocket gopher illustrating an intermediate type of rodent
nathi. Proc. R. Soc. Lond. B, 267 : 393-402.
membrane formation. II. From the beginning of the allantois
KING, B.F. & H.W. MOSSMAN (1974). Fetal membranes and unu-
to term. Am. J. Anat., 113 : 447-477.
sual giant cell placenta of the jerboa (Jaculus) and jumping
MOSSMAN, H.W. & L.A. WEISFELDT (1939). The fetal mem-
mouse (Zapus). Am. J. Anat., 140 : 405-432.
branes of a primitive rodent, the thirteen-striped ground
LUCKETT, W.P. (1985). Superordinal and intraordinal affinities
squirrel. Am. J. Anat., 64 : 59-109.
of rodents. Developmental evidence from the dentition and
MOSSMAN, R.J. (1927). The yolk-sac placental circulation of the
placentation. In : LUCKETT & HARTENBERGER (eds), Evolu-
guinea-pig. Thesis, University of Wisconsin.

38
Andrea Mess
NEDBAL, M.A., M.W. ALLARD & R.L. HONEYCUTT (1994).
SIMPSON, G.G. (1945). The principles of classification and a clas-
Molecular systematics of hystricognath rodents : Evidence
sification of mammals. Bull. Am. Mus. Nat. Hist., 85 : 1-350.
from the mitochondrial 12S rRNA gene. Mol. Phylogenet.
SCHOOLEY, J.P. 1934. Early development in some Sciuridae. J.
Evol., 3 : 206-220.
Morph., 56 : 477-511.
NEDBAL, M.A., R.L. HONEYCUTT & D.A. SCHLITTER (1996).
Higher-level systematics of rodents (Mammalia, Rodentia) :
THEILER, K. (1972). The House Mouse. Development and Nor-
Evidence from the mitochondrial 12S rRNA gene. J. Mamm.
mal Stages from Fertilisation to 4 Weeks of Age. Springer,
Evol., 3 : 201-237.
New York.
ODUOR-OKELO, D. & S.G. GOMBE (1982). Placentation in the
TIBBITTS, F.D. & H.H. HILLEMANN (1959). The development and
cane rat (Thryonomys swinderianus). Afr. J. Ecol., 20 : 49-
histology of the chinchilla placenta. J. Morph., 105 : 317-
66.
366.
ODUOR-OKELO, D. & S.G. GOMBE (1991). Development of the
TULLBERG, T. (1899/1900). Ueber das System der Nagethiere :
foetal membranes in the cane rat (Thryonomys
eine phylogenetische Studie. Akademische Buchdruckerei,
swinderianus) : a re-interpretation. Afr. J. Ecol., 29 : 157-
Uppsala.
167.
O
W
TIANG'A-OWITI, G.E., D. ODOUR-OKELO & S.G. GOMBE (1992).
ILLEY, A. (1912). Foetal membranes of the American beaver
Foetal membranes and placenta of the springhare (Pedetes
(Castor canadensis). Zool. Jahrb. (Suppl 15), 2 : 191-218.
capensis larvalis Hollister). Afr. J. Ecol., 30 : 74-86.
WILSON, D.E. & D.M. REEDER (1993). Mammal species of the
PERROTTA, C.A. (1959). Fetal membranes of the Canadian por-
world. A taxonomic and geographic reference. Smithsonian
cupine, Erethizon dorsatum. Am. J. Anat., 130 : 35-59.
Institution Press, Washington, London.

Belg. J. Zool., 135 (supplement) : 39-43
December 2005
Barn owl pellets : a useful tool for monitoring small
mammal communities?

N.L. Avenant
P.O. Box 266, National Museum, Bloemfontein, 9300, South Africa
Corresponding author : N.L. Avenant, e-mail : Navenant@nasmus.co.za
ABSTRACT. Monthly fluctuations in the diet of Barn owl Tyto alba were compared to prey availability in a typical
South African dry sandy highveld grassland over a 12 month period. Mice, shrews, bats, birds and insects were all
major prey items, and their contribution in pellets fluctuated significantly over months. Barn owl proved to be very
efficient samplers of the small mammal prey group : not only was the owl more successful than museum personnel
in sampling the variety of species present during a specific time of year, but peaks in prey utilization were also more
characteristic of actual fluctuations than that found by traps. Owl pellet analysis is a valuable asset during small
mammal monitoring studies, and is especially useful for sampling small mammal indicator species during environ-
mental impact assessments. However, owl pellet analysis should never be seen as an alternative for small mammal
trapping when small mammal community structure is the focus of study.
KEY WORDS : Barn owl, small mammal, monitoring, sampling method
INTRODUCTION
as it increases from (a) to (b) in Fig. 1. Generalist species
(or species with a wide habitat tolerance) were generally
Lately, small mammal communities have been used as
found to dominate small mammal numbers on the lower
indicators of habitat integrity (see AVENANT, 2000a, 2003;
part of the curve, with the opposite true for specialist spe-
AVENANT & KUYLER, 2002; AVENANT & WATSON, 2002).
cies, which increase in number towards the end of the
A growth curve (Fig. 1) has been postulated for a number
curve. Together, these findings led to the idea that the
of animal & plant groups (see WANG et al., 1999),
direct monitoring of small mammals be used as a rela-
whereby highest species richness increase with succes-
tively quick and inexpensive method of indicating eco-
sion (= primary productivity) up to a point of climax, and
logical disturbance/habitat integrity, and therefore a use-
then decrease to a point where equilibrium is reached.
ful tool for wildlife managers and environmental
Species number fluctuates around this point until distur-
consultants.
bance takes place. Depending on the measure and speed
of disturbance the number of species for the specific ani-
mal group may follow the curve backwards, with highest
Climax
species numbers found at intermediary disturbance, and
b
s
lowest species number found at/after extreme distur-
i
e
c
bance. "Relatively few ruderal species dominate when
e
p
disturbances are frequent, and relatively few highly com-
s
f
petitive species dominate when disturbances are rare;
o
intermediate levels of disturbance allow succession to
r
e
proceed but limit the ability of competitive species to
b
m
dominate the community" (VALONE & KELT, 1999). The
number of microhabitats and primary productivity is also
Nu
a
high at the point of climax, and able to sustain a number
of individuals from different species. The data of a
Succession
number of longer term small mammal studies in southern
Primary productivity
Africa can be fitted to this curve (e.g. ROWE-ROWE &
LOWRY, 1982; ROWE-ROWE, 1995; FERREIRA & VAN
Fig. 1. ­ Correlation between number of small mammal spe-
AARDE, 2000), while our relatively short-term studies
cies and succession / primary productivity. See text for a
(AVENANT, 2000a, b; KUYLER pers. comm.; AVENANT &
description of letters a, b.
KUYLER, 2002; AVENANT & WATSON, 2002), where small
mammal communities were correlated with the abun-
dance of pioneer plant species and/or ecological value of
Identifying small mammal remains from animal scats
the veld in the Free State province, indicate that a similar
and pellets is an indirect method of monitoring small
curve can be expected. The latter studies have also indi-
mammals. In the past it has been a useful tool for animal-
cated that small mammal Shannon-diversity and Even-
ecologists and wildlife managers : it does not only reflect
ness (E ) may be good indicators of ecosystem integrity,
the hunting and feeding behaviour of the predator, but are
var

40
N.L. Avenant
also useful for studying the systematics, geographical dis-
The standardized method prescribed to EIA consultants
tribution, population ecology and craniology of prey ani-
working in the Free State grassland (AVENANT, 2000b;
mals. Fossil remains derived from owl pellets have also
FERREIRA & AVENANT, 2003) were used to sample small
made an important contribution in the reconstruction of
mammal communities : One hundred snap traps were
palaeo-environments (see DAVIS, 1959; AVERY, 1982,
placed per transect. These were spaced 5m apart and left
1987, 1991, 1992, 1999).
open, checked and re-baited in the early morning and late
In the present study the indirect method of the analysis
afternoon for four consecutive days and nights. Bait used
of modern Barn owl, Tyto alba affinis Blyth, 1862, pellets
was a mixture of peanut butter, rolled oats, sunflower oil
were compared to our direct method of monitoring small
and marmite (yeast extract). Rodents and shrews trapped
mammals, and the relevance of these results to the study
were sexed, weighed, measured, dissected and study skins
of habitat integrity considered.
and skulls deposited in the National Museum (Bloemfon-
tein, South Africa) collection. Trap success (or percentage
success) is the number of small mammals captured per
MATERIAL AND METHODS
100 trap nights. Variety is the number of species found,
while diversity, calculated using the Shannon index
Study area
(MAGURRAN, 1988), is a measure of both the number of
species and equality of representation of the individuals
Between March 1998 and September 2000 more than
of all species.
400 pellets of the Southern African Barn Owl, Tyto alba,
were collected on a daily basis from a single locality at
To increase our species list, 100 PVC live traps (on
Florisbad Research Station (28°46'S; 26°04'E), central
separate, 10 m distant parallel transects), 10 pitfall traps
interior of South Africa. The vegetation is typical Dry
(15 cm diameter, 20 cm deep; at random trap stations on
Sandy Highveld Grassland (Veld type 37 - L
each transect), and spades (to search in disused termitaria)
OW &
R
were also used.
EBELO, 1996). Mean annual precipitation in this summer
rainfall area is c. 450mm, and mean daily maximum and
minimum temperatures ranges from c. 31°C and 14°C in
RESULTS & DISCUSSION
January to c. 16°C and -1°C in July (WEATHER BUREAU,
1986).
The main prey items (i.e. contributing > 40% to the
Fresh pellets were individually placed in paper bags,
volume of the majority of pellets in which it occurs)
air-dried and later teased apart. Large, easily diagnosed
were : mice (mean monthly IV = 68.0), birds (IV = 1.4),
fragments of prey were macroscopically identified while
insects (IV = 0.7; orders Coleoptera, Lepidoptera,
hair, teeth and feathers were identified under a stereo
Orthoptera & Mantodea), shrews (IV = 0.4) and bats (IV
microscope at 25x or 50x magnification. Prey items were
= 0.01) (Table 1). Molerat and plant material were present
identified to species level where possible by comparing
in < 0.5% of pellets. No reptile remains were found. All
undigested remains with a reference collection and from
main prey items, except bats, fluctuated significantly
published results, e.g. scales on hair imprints (KEOGH,
(p<0.05) over months (Fig. 2). The fluctuation pattern of
1983 a, b) and tooth form (DE GRAAFF, 1981; PERRIN,
mice, the major prey item in the diet of the Barn owl, did
1982; BOWLAND & BOWLAND, 1989).
not simply follow their densities (density significantly
Both percentage volume (a prey item's percentage con-
highest in late-autumn/early-winter, and lowest just
tribution to total volume ingested) and percentage occur-
before the start of their breeding season at the end of
rence (a percentage of the number of pellets in which a
winter : see AVENANT, 2000a, b), but were, nevertheless,
prey item was present) of prey items in pellets were deter-
more accurate than our small mammal trapping over the
mined, and an Importance Value calculated (IV = percent-
past six years in indicating real densities (sampling with
age volume x percentage occurrence / 100). The compu-
traps have, among others, been found to be influenced by
ter programme Statistica for Windows (Statsoft Inc.,
differences in age group structure and the availability of
1995) was used to do the statistical analyses. Analysis of
natural food). The other prey items generally follow the
Variance (ANOVA) tests were used to detect inter-group
inverse pattern of mice, being highest from c. October to
differences. The 95% level (p<0.05) was regarded as sta-
May.
tistically significant for all tests.
Small mammals were sampled during late-summer and
TABLE 1
late-autumn at the six most diverse habitats in the Floris-
Mean monthly Percentage Occurrence, Percentage Volume, and
bad Research Station grounds :
Importance Value (IV) of prey items of the southern Barn Owl
(1) Exotic Kikuju sp. grass & Eucalyptus sp. trees
Tyto alba at Florisbad Research Station, 1998 ­ 2000.
(28°46.052'S; 26°04.248'E); (2) Open eroded area
Prey item
% Occurrence
% Volume
IV
(28°46.044'S; 26°04.303'E); (3) Low bushes on post-cli-
max grassland (28°45.893'S; 26°04.243'E); (4) Vegeta-
Mice
85.4
79.6
68.0
tion (mostly sedges) around a swampy area (28°46.045'S;
Shrews
10.7
4.1
0.4
26°04.268'E); (5) "Open" Themeda triandra grassland
Molerats
0.5
6.1
0.03
(28°46.039'S; 26°04.352'E); (6) "Dense" Themeda trian-
Bats
2.3
0.6
0.01
dra grassland (28°45.910'S; 26°04.186'E). Sites 1 and 2
Birds
13.4
10.6
1.4
were considered the most influenced by man, and sites 5
Insects
14.2
4.6
0.7
and 6 the least.
Plants
0.3
0.1
0.0003

Barn owl pellets : a useful tool for monitoring small mammal communities?
41
% Volume
% Occurrence
Dec
Dec
Nov
Nov
Oct
Oct
Mice
Sep
Mice
Sep
Aug
Aug
Birds
Jul
Birds
Jul
Jun
Jun
May
May
Insects
Apr
Insects
Apr
Mar
Mar
Shrews
Feb
Shrews
Feb
Jan
Jan
Bats
Bats
Fig. 2. ­ Monthly fluctuations of the main prey items in the diet of southern African Barn
owl Tyto alba at Florisbad Research Station.
The nocturnal small mammal species Mastomys
important as prey. Does this mean that the owl forages at
coucha and Tatera spp. contributed by far the highest per-
different times during the warmer and colder months, or
centage to the volume of prey ingested (Fig. 3). From
is it that the crepuscular and diurnal prey become inac-
September to April, however, the crepuscular Rhabdomys
tive earlier during the winter days as soon as the temper-
pumilio and diurnal Otomys irroratus became more
ature drops?- no answer could be found in the literature.
50
e 40
m
u
l
o
v
30
e
g
a
t
n
e 20
c
r
e
P 10
0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
(n=19)
(n=30)
(n=54)
(n=59)
(n=42)
(n=42)
(n=37)
(n=29)
(n=33)
(n=17)
(n=30)
(n=22)
Fig. 3. ­ Monthly fluctuations of the main small mammal prey species (i.e. contributing > 3% to the
Figure 3: Monthly fluctuations of the main small mammal prey species (i.e. contributing > 3%
volume of pellets in at least one month) in the diet of southern African Barn owl Tyto alba at Floris-
to the volume of pellets in at least one month) in the diet of southern African barn owl
Tyto alba
bad Research Station.
at Florisbad Research Station.
, R. pumilio ;
, M. coucha ;
, C. cyanea ;
, R. pumilio;
, M. coucha;
, C. cyanea;
, O. irroratus;
M. albicaudata;
, O. irroratus ;
, M. albicaudata ;
, Tatera spp .;
, R. rattus
, Tatera spp.;
, R. rattus
In the present study, analysis of owl pellets proved use-
owls has been shown in a number of other studies (e.g.
ful when determining small mammal species present in
HAPPOLD & HAPPOLD, 1986; DENYS et al., 1999; BA et al.,
this grassveld ecosystem. Despite its preference for some
2000; GRANJON et al., 2002). As in our small mammal
species, the Barn owl nevertheless sampled more species
studies (see AVENANT, 2003), however, the owl has found
than our extensive trapping efforts indicated (Table 2).
the "scarcer" c. 15 percent of the total number of small
The presence of Tatera brantsi, Myosorex varius, and the
mammal species sampled only in the late autumn - early
red listed Mystromys albicaudatus, and difficult to sample
winter months, while more than 25 percent of species
Saccostomus campestris and Desmodillus auricularis
were not preyed upon in the spring months.
were only indicated by its presence in owl scats. Although
The multimammate mouse Mastomys coucha, as well
the owl may forage outside of the Florisbad Research Sta-
as species richness and diversity, has been used as an indi-
tion grounds, all species found in owl pellets are expected
cator of disturbance (see AVENANT, 2000a, b; AVERY,
to occur within the grounds. Also, four other species sam-
1991, 1992). In the relatively disturbed habitats at Floris-
pled by the owl could not be found by traditional trapping
bad this mouse has been found to dominate in four of the
methods (see Table 2). This effective sampling of Barn
six habitats sampled by museum personnel. It was also a

42
N.L. Avenant
TABLE 2
AVERY, D.M. (1982). Micromammals as palaeoenvironmental
indicators and an interpretation of the late quarternary in the
Terrestrial small mammal species sampled by the southern Afri-
southern Cape Province, South Africa. Ann. S. Afr. Mus., 85 :
can Barn owl Tyto alba versus small mammal species sampled
183-374.
by personnel of the National Museum. X, sampled by traps; x,
AVERY, D.M. (1987). Late Pleistocene coastal environment of
sampled by alternative methods (see text).
the Southern Cape Province of South Africa : micromam-
mals from Klasies River mouth. J. Archaeol. Sci., 14 : 405-
Small mammal species
Owl
Personnel
421.
AVERY, D.M. (1991). Micromammals, owls and vegetation
Rhabdomys pumilio
X
X
change in the Eastern Cape Midlands, South Africa, during
Mastomys coucha
X
X
the last millennium. J. Archaeol. Sci., 20 : 357-369.
Otomys irroratus
X
X
AVERY, D.M. (1992). Man and/or climate? Environmental deg-
Mystromys albicaudatus
X
radation and micromammalian community structure in
Saccostomus campestris
X
South Africa during the last millennium. S. Afr. J. Science,
Desmodillus auricularis
X
88 : 483-489.
Tatera leucogaster
X
X
AVERY, D.M. (1999). A preliminary assessment of the relation-
Tatera brantsi
X
ship between trophic variability in southern African Barn
Mus minutoides
X
X
Owls Tyto alba and climate. Ostrich, 70(3&4) : 179-186.
Mus musculus
X
x
Rattus rattus
X
x
BA, K., L. GRANJON, R. HUTTERER & J-M. DUPLANTIER (2000).
Myosorex varius
X
Les micromammifères du Djoudj (Delta du Sénégal) par
Crocidura cyanea
X
X
l'analyse du régime alimentaire de la chouette effraie, Tyto
Suncus varilla
X
x
alba. Bonn. Zool. Beitr., 49 : 31-38.
Cryptomys hottentotus
X
x
BOWLAND, A.E. & J.M. BOWLAND (1989). An identification to
rodent prey in carnivore scats and pellets. Lammergeyer, 40 :
Total number of species
15
10
8-9.
DAVIS, D.H.S. (1959). The Barn owl's contribution to ecology
major prey item of the owl. Commensal species, such as
and palaeoecology. Ostrich sup., 3 : 144-153.
the house mouse and house rat, have being used to
DE GRAAFF, G. (1981). The Rodents of Southern Africa. Butter-
describe changes in vegetation / presence of man in palae-
worth & Co., Pretoria.
ontological times. In the present study these two species
DENYS, C., W. CHITAUKALI, J.K. MFUNE, M. COMBREXELLE & F.
contributed throughout the year to owl diet ­ a low contri-
CACCIANI (1999). Diversity of small mammals in owl pellet
bution, with no significant difference between months or
assemblages of Karonga district, northern Malawi. Acta
zool. Cracov.,
42 : 393-396.
seasons (p>0.1) ­ but were not found in our traps.
FERREIRA, S.M. & N.L. AVENANT (2003). Influences of trap-
spacing on descriptors of hypothetical small mammal com-
ACKNOWLEDGEMENTS
munities in grasslands at Tussen-die-Riviere Nature
Reserve. Navors. Nas. Mus., Bloemfontein. 19 : 21-30.
FERREIRA, S.M. & R.J. VAN AARDE (2000). Maintaining diver-
James Brink and Lloyd Rossouw are thanked for their inter-
sity through intermediate disturbances : evidence from
est, support and making the facilities at Florisbad available to
rodents colonizing rehabilitating coastal dunes. Afr. J. Ecol.,
us. Piet Mdala consistently collected all fresh pellets. Pieter Wil-
38 : 286-294.
liamson analysed the owl pellets and, together with Isak Sek-
G
huni and Jacob Senoge, assisted with the trapping of small
RANJON, L., C. BRUDERER, J.F. COSSON, A.T. DIA & F. COLAS
(2002). The small mammal community of a coastal site of
mammals. The Council and Directors of the National Museum,
south-west Mauritania. Afr. J. Ecol., 40 : 10-17.
Bloemfontein are thanked for the permission to carry out this
project.
HAPPOLD, C.D. & M. HAPPOLD (1986). Small mammals of
Zomba Plateau, Malawi, as assessed by their presence in pel-
lets of the Grass owl, Tyto capensis, and by live-trapping.
REFERENCES
Afr. J. Ecol., 24 : 77-87.
KEOGH, H.J. (1983a). A photographic reference system of the
microstructure of the hair of Southern African Cricetidae
AVENANT, N.L. (2000a). Small mammal community characteris-
and Muridae. S. Afr. J. Wildl. Res., 13 : 1-51.
tics as indicators of ecological disturbance in the Willem
Pretorius Nature Reserve, Free State, South Africa. S. Afr. J.
KEOGH, H.J. (1983b). A photographic reference system of the
Wildl. Res., 30 : 26-33.
microstructure of the hair of southern African bovids. S. Afr.
J. Wildl. Res.,
13 : 89-131.
AVENANT, N.L. (2000b). Terrestrial small-mammal diversity in
LOW, A.B. & A.G. REBELO (1996). Vegetation of South Africa,
Korannaberg Conservancy, Free State, South Africa. Navors.
Lesotho and Swaziland. Department of Environmental
Nas. Mus., Bloemfontein.16(4) : 69-82.
Affaires and Tourism. Pretoria.
AVENANT, N.L. (2003). The use of small-mammal community
MAGURRAN, A.E. (1988). Ecological diversity and its measure-
characteristics as an indicator of ecological disturbance in
ment. Cambridge University Press, Cambridge.
the Korannaberg Conservancy. In : SINGLETON, HINDS,
PERRIN, M.R. (1982). Prey specificity of the Barn owl, Tyto
KREBS & SPRATT (eds), Rats, Mice & People : Rodent Biol-
alba, in the Great Fish River valley of the Eastern Cape
ogy and Management, ACIAR Monograph No 96.
Province. S. Afr. J. Wildl. Res., 12(1) : 14-25.
AVENANT, N.L. & P. KUYLER (2002). Small mammal diversity in
ROWE-ROWE, D.T. (1995). Small-mammal recolonization of a
the Maguga area, Swaziland. S. Afr. J. Wildl. Res., 32 : 101-
fire-exclusion catchment after unscheduled burning. S. Afr.
108.
J. Wildl. Res., 25(4) : 133-137.
AVENANT, N.L. & J.P. WATSON (2002). Mammals recorded in
ROWE-ROWE, D.T. & P.B. LOWRY (1982). Influence of fire on
the Sandveld Nature Reserve, Free State Province, South
small-mammal populations in the Natal Drakensberg. S. Afr.
Africa. Navors. Nas. Mus., Bloemfontein. 18(1) : 1-12.
J. Wildl. Res., 12 : 130-139.

Barn owl pellets : a useful tool for monitoring small mammal communities?
43
VALONE, T.J. & D.A. KELT (1999). Fire and grazing in a shrub-
productivity of arid and semi-arid grasslands in north China.
invaded arid grassland community : independent or interac-
J. Arid. Env., 43 : 467-475.
tive ecological effects? J. Arid. Env., 42 : 15-28.
WANG, G., Z. WANG, Q. ZHOU & W. ZHONG (1999). Relationship
WEATHER BUREAU (1986). Climate of South Africa. Climate sta-
between species richness of small mammals and primary
tistics up to 1984. Government Printer, Pretoria.


Belg. J. Zool., 135 (supplement) : 45-51
December 2005
Feeding biology of the dassie-rat Petromus typicus
(Rodentia, Hystricognathi, Petromuridae) in captivity

Andrea Mess1(*) and Manfred Ade2(*)
1 Humboldt-Universität zu Berlin, Institut für Systematische Zoologie, Invalidenstr. 43, D-10115 Berlin, Germany. and-
rea.mess@rz.hu-berlin.de
2 Joachim-Karnatz-Allee 41, D-10557 Berlin, Germany
Corresponding author : Andrea Mess, e-mail : andrea.mess@rz.hu-berlin.de
ABSTRACT. We examined the feeding biology of the poorly known dassie-rat Petromus typicus. External morphol-
ogy indicates that the digging for soil-inhabiting invertebrates as food is unlikely. Animals in captivity refuse to eat
insect larvae and data from field studies indicate that invertebrates play no major role with regard to the intake
quantity. Observations on jaw movements and occlusion patterns of the cheek teeth indicate that Petromus is not
restricted to high-fibre plant matter as food. This matches the catholic diet of Petromus in captivity and in the wild,
where e.g. flowers and fruits are consumed when available. The rooted and moderately hypsodont cheek teeth sug-
gest limited adaptation to abrasive plant material in comparison to other grass feeding hystricognaths. However,
captive specimens consume high fibrous graminoid material during all activity phases, even when energetically
more rewarding food is available. This suggests that fibre is an important food component. The stomach has no
proventriculus or similar structure. Therefore, fermentation of plant matter in that region and/or rumination is
unlikely. The caecum is large and haustrated, indicating the ability to process cellulose by micro-organisms. The
morphology of the proximal colon indicates the presence of the so-called colon separating mechanism (CSM). It is
therefore likely that the animals are able to produce vitamin and protein-rich faeces. This is confirmed by the occur-
rence of coprophagy by Petromus. The great variety of food sources hints at the ability of Petromus to cope with
unstable environments, as is the case in xeric areas.
KEY WORDS : Rodentia, Hystricognathi, Petromus, feeding behaviour, nutrition, digestive system, coprophagy,
rumination.
INTRODUCTION
characterised by a k-selective or precocial strategy (MESS
et al., 2001). They are mainly herbivorous. Since
African hystricognath rodents are well known as por-
Petromus is often suggested to have retained a large
cupines (Hystricidae), cane rats (Thryonomyidae), as well
number of plesiomorphic conditions of Hystricognathi
as by the group of subterranean species of mole rats
(cf. MESS, 1999a), this species is important for recon-
(Bathyergidae). In contrast, the monotypic family
structing the evolution of mammals in Africa. Particu-
Petromuridae (TULLBERG, 1899/1900; cf. MCKENNA &
larly, according to the dassie-rats limited distribution and
BELL, 1997; WILSON & REEDER, 1993) and its only mem-
tolerance to xeric conditions, it could serve as a model for
ber, the dassie-rat or noki Petromus typicus A. SMITH,
understanding how mammals use strategies to cope with
1831 (Fig. 1a), is less well known. Petromus is endemic
the increasing aridity of the Southern African Subregion.
to the Southern African Subregion, confined to the arid to
Data on Petromus is remarkably "few" : Only some
semi-arid zone in the southernmost parts of Angola, in
basic information on their natural habitat, nutrition and
Namibia, and in the north-western part of the Cape Prov-
reproduction is available (e.g., WITHERS et al., 1980; DE
ince in RSA (SKINNER & SMITHERS, 1990; COETZEE,
GRAAFF; 1981; SKINNER & SMITHERS, 1990; COETZEE,
2002). It appears to be the geologically oldest rodent
1983, 2002; NOWAK, 1999), including a few field studies
inhabitant of the Namib desert (MEESTER, 1965). It lives
(WITHERS, 1979, 1983; GEORGE, 1981; GEORGE &
in rocky habitats, for instance in the crevices of the kopjes
CROWTHER, 1981; COETZEE 2002; RATHBUN & RATHBUN,
in the Namibian escarpment (SKINNER & SMITHER, 1990).
this volume). It had been assumed that Petromus is able to
Accordingly, Petromus possesses features regarded as
ruminate (COETZEE, 1983). According to GEORGE (1981)
adaptations for living in rock crevices, i.e. a flattened
the diet is dominated by graminoids and the species is
skull and flexible ribs (VAUGHAN et al., 2000; see also
considered a herbivore (see also COETZEE, 1983). How-
TULLBERG, 1899/1900; ELLERMAN, 1940; NOWAK, 1999).
ever, one report found that Petromus feeds to a significant
Hystricognathi, including Petromus, differ conspicu-
degree on insects (WITHERS, 1979; apparently not known
ously from other rodents : Derived characters (apomor-
by GEORGE loc. cit.). Today a variety of food is reported
phies) are associated with their reproduction which is
to be taken (RATHBUN & RATHBUN, this volume) and
(*) Both authors equally contributed to this study.

46
Andrea Mess & Manfred Ade
rumination (including regurgitation and remastigation)
1998, 1999; ADE et al., 2001) and general biology (e.g.,
would demand an elaborate gastro-intestinal apparatus,
MESS, 2002, 2005; MESS et al., 2000, 2002). Here, quali-
i.e. at least a proventriculus. The feeding capabilities of
tative observations derived from the animals in captivity
Petromus are still enigmatic. More information on these
will be presented with special reference to nutrition, feed-
aspects, affecting the biology as well as the ecological
ing behaviour and morphology of the digestive system.
significance of this species, is required. In the last couple
of years, a breeding group of Petromus has been estab-
We will review scattered information about this poorly
lished and maintained successfully. Research so far deals
known species. Morphological data will help to integrate
with placentation and their evolutionary history (e.g.,
these data to a functional picture in the sense of "whole
MESS, 1999b, 2001, 2003), external morphology (ADE,
organism biology" (NOVACEK, 1998).
Fig. 1. ­ Habits, environment and feeding conductance of Petromus from a breeding colony.
A : Petromus 43, a male individual that was given to the Tierpark Berlin.
B : Interior of the animal house at the Humboldt-University of Berlin.
C : Petromus 4, a male individual feeding on his daily food from the feeding dish.
D : Petromus 27, a female engaged with reingestation of faecal droppings.
MATERIAL AND METHODS
ard food pellets for chinchillas (Sniff ® Chi, based on
plant fibres with 14% raw fibre content) or a mixed food
The breeding group, started in 1995, is based on 8 ani-
for guinea pigs (Sniff ® Ms Müsli) is given. These dry
mals from the RSA. The group is now housed at the Hum-
foods contain a considerable amount of vitamin C. This
boldt-University, Berlin, and was formerly bred at the
industrially produced food is supplemented by a mixture
Universities of Tübingen and Göttingen. Currently, the
of fresh plant material. Most frequently used are carrots
animals are housed in primate cages : 1 x 0.7 x 0.8m for
(roots as well as green parts), tomatoes, paprika, cucum-
single individuals or pairs, and double this space for fam-
bers, radishes, kohlrabi, broccoli, maize, and more rarely
ily groups (Fig. 1b). Inside, several resting places and a
apples, pears, grapes and other fruits are given. (Fig.
nest box are offered, as well as the possibility for sand
1c)Occasionally dried bread, sunflower seeds or nuts are
bathing. The animal house is characterised by the follow-
given in addition. Finally, mineral supply is provided by
ing parameters : about 25°C air temperature, 50% humid-
small pieces of standard pet limestone and salt stones.
ity, 12 hour light with additional daylight spots provided 4
This study is based on qualitative observations on vari-
to 6 hours each day (Fig. 1b).
ous animals from the breeding group. Occasionally, a
The animals are fed with hay, i.e. hard-pressed pellets
time lapse video camera using infrared spectrum was
containing dried graminoids with about 26% raw fibre
used to observe the behaviour during the dark phase.
material (Sniff ® Heucobs). Moreover, once a day stand-
Behaviour in captivity is compared with data derived

Feeding biology of Petromus typicus
47
from field studies. The dassie-rats were conditioned to
edly abraised at the buccal side (Fig. 2c), whereas, corre-
accept close observations by rewarding them with food
spondingly, the lower interacting gnawing teeth show
and allogrooming by humans when approaching the ani-
abrasion on the lingual side. The morphology of the jaw
mal keeper or student. This way it was possible to explore
joint indicates that propalineal (back- and forth) move-
the direction of chewing movements by touching the
ments are likewise possible, as in all rodents (see BUTLER,
working jaws.
1985; THENIUS, 1989).
Anatomical examination was conducted by gross mor-
phological analysis using 60 fold magnification of the
The gastro-intestinal tract
gastrointestinal tract of one female individual (Petromus
Nr. 19 of the breeding colony) as well as of the morphol-
The stomach is large. Moreover it is markedly curved,
ogy of the teeth and head region by using skulls and mate-
almost U-shaped (Fig. 2d). It has no transversely-running
rial from the wet collection that has been built up during
the years. In particular, tooth structure is described on the
folds or septa producing proventriculus-like structures
basis of animals that have been born in the wild
without glands as, e.g. in murids (Fig. 2d). There is a con-
(Petromus SZ 7499, Zoologische Sammlung, University
tinuous layer of glands present as judged from gross mor-
of Tübingen) as well as in captivity (Petromus Nr. 61 of
phological analysis. The caecum is large in diameter and
the colony).
strongly subdivided or haustrated (Fig. 2e). The transition
area from the caecum towards the colon is inflated (Fig.
2e). The proximal part of the colon is moderately large in
RESULTS
diameter. Dissection of the proximal colon region reveals
that longitudinally-running ridges are present (Fig. 2f).
Two prominent ridges run distally. Proximally, they are
DESCRIPTION OF FEATURES ASSOCIATED
associated with some low oblique ridges in the transition
WITH NUTRITIONAL BIOLOGY
zone between the caecum and the colon. The longitudinal
ridges possess a transversely ridged surface structure. The
two main ridges are closely apposed to each other, enclos-
1. Morphology of the digestive system
ing a distinct groove (Fig. 2f).
and associated structures
2. Behaviour associated with feeding
Oral head region
Feeding
The oral cavity, as in all rodents, is bipartite. A gnaw-
ing compartment and a chewing compartment are present.
The compartments are produced by inwardly projecting
The hands are used to hold the food during gnawing
"lips" (inflexa pellita) provided with micro vibrissae, sepa-
action (Fig. 1c). This is a remarkable process by which
rating the front or gnawing teeth, respectively, from the
Petromus adjusts the position of the food item to the
cheek teeth (Fig. 2a). These oral rim projections meet
gnawing tooth (Fig, 1c; see also LANDRY, 1970). The food
nearly at the median plane, separated by a well devel-
pieces lay between a groove formed by the reduced
oped, longitudinally extending papilla palatina. The
thumbs and the proximal and distal pads (for terminology
gnawing teeth are easily exposed by an upper lip cleft
of hand morphology see ADE & ZIEKUR, 1999). It appears
(Fig. 2a). The rhinarium of Petromus is strongly reduced;
that Petromus eats repeatedly during the whole day, espe-
not even small narial pads are present (Fig. 2b). Instead,
cially on hay which is provided without restriction. In
there are only small, reduced, inconspicuous cushions at
between feeding activities, extended resting phases take
the entrance of the nares (Fig. 2b). The chewing compart-
place, using the warm day-light spots. The animals are
ment proximal of the diastema ("filled" with the inflexa)
active every few hours during the night or dark phase,
consists of 4 cheek teeth (dP4, M1-3). The teeth possess
which is usually linked with feeding on hay or other
deep transverse infolding of the enamel, referred to as
available food. Both during day and night, the hay pellets
bilophodont condition with an anterior protoloph and a
are eaten either at the place where they have been depos-
posterior metaloph (sensu THENIUS, 1989, see Fig. 2c.)
ited by the animal keepers or they are transported to
The borders of the lophs below (buccal side) and above
where the dassie-rats prefer to sit down and rest. Fresh
(lingual side) form distinct cusps (Fig. 2c). As judged
from the occlusion pattern, the grinding and shearing
food and food pellets are more often eaten directly from
actions are not produced in the horizontal plane as in
the feeding dishes without transporting them (Fig. 1c).
mainly grass-consuming hystricognaths (T
Within pairs or family groups, a female has first access to
HENIUS, 1989).
Instead, there is a more strongly developed vertical tooth
the feeding dish, especially when she is pregnant or lac-
relief. The relief indicates a "mortar- and- pestle" action
tating. Typically, conflicts at the feeding dish are settled
(see LUCAS, 1979) as in dilambdodont teeth (e.g. Tupaia,
by vocal dispute between the individuals and not by phys-
THENIUS, 1989). This means that there is a marked trans-
ical attacks. If such attacks occur they are usually not vio-
verse component of action during chewing. Reflecting
lent (our animals have been carefully accustomed to each
this transverse component, the upper tooth row is mark-
other before putting them together).

48
Andrea Mess & Manfred Ade
Fig. 2. ­ Morphology of oral head and gastro-intestinal tract in Petromus typicus.
A : The mouth cavity and gnawing teeth. Scale bar = 0.2 cm. (ADE, 1998).
B : The rhinaric region in a subadult individual. Scale bar = 0.1 cm. (ADE, 1998).
C : The left upper cheek teeth row (dP4, M1-3) of Petromus SZ 7499 from above (full-size photo) and from the buccal side (inlet) to
demonstrate the distinct cusps. Accordingly, mesial is on the left hand side and buccal on top. Scale bar = 0.2 cm.
D : The stomach of Petromus 19 after macroscopic preparation. Scale bar = 0.5 cm.
E : Transition from the haustrated caecum into the proximal colon. Scale bar = 0.5 cm.
F : The proximal colon after preparation with longitudinal folds. Scale bar = 0.5 cm.
Abbreviations in Fig. 2 :
1
: upper inflexa pellita
11 : opening of the oesophagus into the stomach
2 : lower inflexa pellita
12 : pylorus, i.e. transition from stomach into proximal gut
3 : the tongue
13 : internal region of the stomach
4 : upper and lower gnawing teeth (dI2)
14 : duodenum
5 : external opening of the nares
15/15* : the caecum : external view and lumen of caecum
6 : inconspicuous rudiments of narial pads
16 : transition zone between caecum and colon
7 : hairy parts in the rhinaric region
17 : the proximal colon
8 : protoloph of M1
18 : area of the colon possessing longitudinal folds inside
9 : metaloph of M1
19 : longitudinal ridges inside the proximal colon
10 : cusp of M1, present at the lingual side
20 : groove between the ridges

Feeding biology of Petromus typicus
49
Coprophagy
not feed exclusively on them. Instead they switched
between the extra food and hay pellets. Thus far,
The animals are able to produce two different kinds of
Petromus has never been observed to eat insects or other
faecal droppings, dark brown ones which are considera-
animals offered. Feeding trials have been conducted by
bly dry and a second type that is more greenish in colour
using meal-worms or crickets. Even during pregnancy or
and wet. It was never observed that a specimen consumes
lactation, the animals refused such food. Moreover, trials
the brown pellets, but the greener ones were eaten fre-
to feed them with pellets for hamsters and mice were not
quently (Fig. 1d). It appears that a transitional production
successful. Cheese or small amounts of meat products
from brown to greenish faecal pellets occur. The two dif-
were not accepted.
ferent sorts of pellets are easily recognised by the
animals : When the droppings are changing their colour
towards greenish, Petromus pick them up with the mouth
DISCUSSION AND CONCLUSION
and bite into it. If they are not appropriate ­ usually when
the colour is still brownish ­ the droppings are immedi-
Petromus possesses characteristics of the naso-labial
ately thrown away, and the next droppings that appear at
and oral region linked with a flexible diet. The bipartite
the anus are tested again. Pellets of distinctly greenish
organisation of the oral mouth cavity allows the gnawing
colour are eaten, often chewing them a while before swal-
teeth to be easily exposed and used for exploration, e.g. of
lowing. The activity related to reingestation mostly
food consistency or texture (A
occurs during the extensive resting periods.
DE, 1998; LUCAS, 1979),
while the inner part of the mouth is protected. In terms of
Rumination, "jack knife behaviour" and "tail stand"
evolution, the exploratively used gnawing teeth have
replaced functions of the rhinarium, i.e. the originally tac-
Although hundreds of hours have been spent observing
tile region of the head in rodents (ADE, 1998). According
different individuals of Petromus, it was not possible to
to the almost complete reduction of the rhinarium in
find any indication for the occurrence of rumination as
Petromus and the fact that the animal projects the dorsum
suggested in literature. Neither after the animals had eaten
nasi rostrally and not the rhinaric region when exploring
their daily amount of vegetables and food pellets, nor
its surroundings, it can be concluded that the rhinaric
after feeding on hay at other times has an indication of
region is not extensively used for specific exploration of
rumination been found (Our Petromus have been accom-
the environment as i.e. in insectivorous terrestrial mam-
modated to close sight contact by the observer). Rumina-
mals. This is supported by behavioural observations com-
tion-related "jack-knife behaviour" as described by
paring Rattus norvegicus, Petromus typicus, Cavia por-
COETZEE (1983), i.e. bending down of the head toward the
cellus and Octodon degus (see observations of MESS &
abdomen, seems to be restricted to male individuals, and
ADE described in ADE, 1998). Thus, extensive rummaging
associated with the cleaning of the genitals. During the
in the soil for insects is very unlikely. However, morphol-
bending down action the penis is elongated to about dou-
ogy does not preclude consumption of non-soil insects,
ble its normal length. Afterwards the penis is taken into
i.e. insects from higher strata of the vegetation or surface
the mouth and cleaned by moving the mouth up and
running forms. Petromus has not been observed to feed
down. Finally, after interrupting the close contact
on insects when we offered them to our captive animals.
between mouth and penis, the individual jerks up upright,
On the basis of observations on feeding behaviour,
often chewing or smacking with its lip region. Such
COETZEE (1983) came to the conclusion that Petromus is
cleaning activities in males occur frequently, distributed
mainly herbivorous in the field. DE GRAAFF (1981), refer-
throughout the day, indicating that the jack-knife action is
ring also to stomach contents, classifies the species as
a comforting behaviour. The newly described "tail stand
strictly feeding on plant matter (leaves, berries, seeds,
behavioural pattern", which means that an animal stand
flowers of compitae). GEORGE (1981) and COETZEE
on its front feet while propping up the hind feet and drum-
(1983) claim a preference for grasses. RATHBUN & RATH-
ming them against the abdomen for several seconds has
BUN (this volume) did not observe dassie-rats searching
been suspected to be important for digestive efficiency
for, or eating, invertebrates. However, WITHERS (1979)
(see Fig. 3 in RATHBUN & RATHBUN, this volume). This
has found a significant contribution of insects in stomach
behaviour has been observed from time to time in captive
contents. The latter reference has not been cited in any
animals from the breeding group too. It is more rare than,
paper except for RATHBUN & RATHBUN (loc. cit.).
for instance, coprophagy or the jack-knife movements.
The puncture-crushing mode with high and sharp cusps
Food preferences
of the teeth, enabling similar sized mammals to use inver-
tebrates as food (LUCAS, 1979; PFRETSCHMER, 1997), is
Petromus has been frequently observed to drink water,
not present in Petromus. The cusps of the cheek teeth of
using the outlets of the water bottles. The tongue is used
Petromus are blunt. During the mortar- and- pestle action,
during water uptake. Hay is given ad lib. and the animals
compressive forces should prevail. Referring to LUCAS
feed on the pellets repeatedly during the day. Standard pet
(1979), our tentative conclusion is that the cheek teeth are
limestones as well as salt stones are used sporadically.
more adapted to fracture plant material than invertebrate
According to the consumption of fresh plant material,
material. However, it cannot be excluded that the gnaw-
Petromus accepts a variety of various vegetables and
ing teeth may serve to puncture and crush exoskeletons,
fruits (see Material & Methods). It appears that Petromus
especially when bearing in mind the sophisticated ability
have individual preferences with regard to the food
to use the hands during the gnawing process. The cheek
offered. When extra food is given, it appears that the ani-
teeth may then also serve as crushing devices. In fact,
mals show a clear preference for seeds and nuts, but do
crushing is the presumed major function of these kind of

50
Andrea Mess & Manfred Ade
teeth with blunt cusps (see THENIUS, 1989). Constraints
of chymus into the caecum as the basis for the production
on the amount of microfaunivory may come from the
of specialised pellets which contain a significantly higher
need to preserve a sufficiently dense population of cellu-
amount of vitamins and protein (BJÖRNHAG & SNIPES,
lose processing micro-organisms in the intestinal tract
1999; HOLTMAIER, 2002). Thus, the ability to use high
(see below), i.e. a sufficient amount of cellulose has to be
fibre matter for energy and protein production has to be
ingested to enable the animal to live solely from plant
assumed for Petromus. The fact that captive specimens
matter. This may be crucial when facing the dryness of
consume high fibrous graminoid material all day long,
the habitat, which is negatively correlated to the amount
even if energetically more rewarding food is available
of invertebrates (SCHULTZ, 2000). Thus, we suspect that
suggests that there is conspicuous dependency on this
physiological regulation might suppress the consumption
kind of food.
of invertebrates. Even WITHERS (1979) points out that
The newly described "tail-stand behaviour" (RATHBUN
Petromus is predominantly herbivorous, despite the fact
& RATHBUN, this volume) is seen from time to time in the
that he notes a high proportion of insects in the stomachs
caged animals. It is suspected by RATHBUN & RATHBUN
of some specimens. The point is that it uses cellulose rich
that this behaviour is related to digestive efficiency by
material.
mechanical stimulation and support of peristaltic move-
The occlusion pattern of the cheek teeth, resembling
ments of the intestine.
omnivorous types of teeth (see THENIUS, 1989), indicates
Judged from the data on morphology and behaviour,
a less specialised mode of chewing compared to other
Petromus is 1) a hindgut fermenter of cellulose with a
Hystricognathi. This fits into the picture that Petromus
special mechanism to utilise micro-organisms as a source
utilises a variety of (plant) material ranging from stems,
of protein and vitamin supply, 2) depending mainly on
leaves to fruits and even insects. Strict herbivory, e.g.
plant matter, but the variety of food in this regard is large,
using high fibrous plant material is related to enamel
3) not confined to high fibrous plant material, but able to
ridges working in a more or less common horizontal
use this kind of material successfully in various combina-
grinding plane (BUTLER, 1985; PFRETSCHMER 1997). This
tions, and 4) not micro-faunivorous to a large degree. The
is not the case in Petromus. Moreover, the cheek teeth are
CSM enables the animals to produce foreign protein
hypsodont but not continuously-growing, i.e. protection
within themselves which hints that they are potentially
of the teeth from rapid abrasion by silicate containing
independent from animal protein. In summary, a great
material such as grasses is not well developed.
variety of food is used. This hints at the ability of
The stomach is large and curved, but internally undi-
Petromus to cope with unstable environments as is the
vided. This uniform cavity indicates that the stomach con-
case in xeric areas, such as in the Southern African Subre-
tent will be exposed to an acid and enzymatic milieu. This
gion.
implies two important consequences. 1) The milieu for
the stomach content has a low pH preventing the effective
establishment of micro-organism populations that could
ACKNOWLEDGEMENTS
serve as fermenters and protein donors (proventriculus-
function possibly in rodents (S
The establishment of a breeding group of Petromus typicus
TARCK, 1995) and rumi-
nants). 2) The chymus is itself acidic and contains
was possible with support from several sides : In particular we
would like to thank Prof. W. Maier (University of Tübingen) for
enzymes. Both could strongly affect the mucosa. This
establishing contacts in order to get animals from RSA and to
means that rumination would be detrimental if there is no
enable us to start this project. After staying for about one year at
extensive buffering by mucus in the oesophagus and oral
the universities of Tübingen and Göttingen, respectively, the
cavity. The latter is unlikely, and completely unknown,
group is now housed at the Humboldt-University Berlin. Thanks
for mammals. Furthermore, the data derived from the lab-
to responsible persons (Profs. W. Maier, H.-D. Kuhn, U. Zeller)
oratory group suggest that the so-called "jack-knife
and to colleagues that have been involved in maintaining the
behaviour" described by COETZEE (1983) can not be con-
colony, e.g. A. Billepp, F. Postlep and J. Pohl (Humboldt-Uni-
firmed to be associated with rumination. A similar behav-
versity). The writing of the manuscript profited markedly from
iour occurs restricted to male individuals when cleaning
discussions with Galen Rathbun and Neels Coetzee. Finally,
the penis. The animals regularly show chewing move-
many thanks to Jason Dunlop for helping with the English.
ments afterwards without any indication of food matter
inside the mouth. Thus, it appears likely that this behav-
LITERATURE CITED
iour belongs to comfort behaviour (including masturba-
tion which might explain the chewing and smacking
ADE, M. (1998). Zur Evolution des Rhinariums der Glires. Eine
afterwards, G. RATHBUN, pers. comm.). The ingested plant
Rekonstruktion auf phylogenetisch-systematischer Grund-
material is most likely fermented by micro-organisms in
lage. Wissenschaft & Technik Verlag, Berlin.
the intestinal tract. The caecum is large and haustrated in
ADE, M. (1999). Macroscopic study on the rhinarium of the
comparison to hystricognaths with marked herbivorous
Lagomorpha. With special reference to the glires hypothesis.
and high fibrous diet, such as the chinchilla (T
Mitteilungen aus dem Museum für Naturkunde zu Berlin,
ULLBERG,
1899/1900). Large caecae are typical for herbivorous
Zoologische Reihe, 75 : 191-216.
rodents (H
ADE, M., S. FRAHNERT & A. MESS (2001). Die Rekonstruktion
ESSE & DOFLEIN, 1935; WITHERS, 1979). Longi-
der Evolution des Nahrungserwerbsapperates der Glires
tudinal folds in the proximal colon reveal that a colon
(Rodentia + Lagomorpha) als Beitrag für die Analyse ökolo-
separating mechanism (CSM) as the structural prerequi-
gischer Fähigkeiten der Rodentia. Mammalian Biology,
site of coprophagy is present. This is supported by the
Sonderheft Bd., 66 : 6-7.
production of special feacal droppings and their ingestion
ADE, M. & I. ZIEKUR (1999). The forepaws of the rodents Cryp-
(also WITHERS, 1979). The CSM creates a selected reflux
tomys hottentotus (Bathyergidae) and Nannospalax ehren-

Feeding biology of Petromus typicus
51
bergi (Muridae, Spalacinae) : Phylogenetic and functional
MESS, A. (2003). Evolutionary transformations of chorioallan-
aspects. Mitteilungen aus dem Museum für Naturkunde zu
toic placental characters in Rodentia with special reference
Berlin, Zoologische Reihe, 75 : 11-17.
to hystricognath species. Journal of Experimental Zoology,
BJÖRNHAG, G. & R.L. SNIPES (1999). Colonic separation mecha-
299A : 78-98.
nism in lagomorph and rodent species. Mitteilungen aus dem
MESS, A. (2005) : Felsenratte. In : GRANSLOßER, U. (ed.), Die
Museum für Naturkunde zu Berlin, Zoologische Reihe, 75 :
Sozialsysteme der Säugetiere. Filander Verlag, Fürth : 156-
275-281.
158.
BUTLER, P.M. (1985). Homologies of molar cusps and crests,
MESS, A., T.B. HILDEBRANDT, M. FAßBENDER, R. HERMES & F.
and their bearing on assessments of Rodent phylogeny. In :
GÖRITZ (2000). Placentation and embryonic development of
LUCKETT & HARTENBERGER (eds), Evolutionary relation-
Petromus typicus (Rodentia, Hystricognathi, Petromuridae) :
ships among rodents. A multidisciplinary analysis. Nato ASI
results on ultrasonic examination and selected morphologi-
Series A, Vol. 92, Plenum Press, New York : 381-401.
cal data. Advances in Ethology, 35 : 98.
COETZEE, C.G. (1983). The feeding behaviour of Petromus typi-
MESS, A., B. MOHR & T. MARTIN (2001). Transformations in the
cus A. Smith, 1831 (Rodentia : Petromuridae). Annals
stem species pattern of hystricognath Rodentia and the cli-
Musee Royal De L'Afrique Centrale Tervuren ­ Belgique,
matic change in the Eocene to Late Oligocene time interval.
Sciences Zoologiques, 237 : 203-206.
Mitteilungen aus dem Museum für Naturkunde zu Berlin,
COETZEE, C.G. (2002). The distribution and breeding season of
Zoologische Reihe, 77 : 193-206.
the dassie-rat, Petromus typicus (Petromuridae, Rodentia).
MESS, A., J. PINT & K. FAUST (2002). Ontogenetic differentia-
Folia Zoologica, 51 Suppl. 1 : 23-35.
tion in Petromus typicus : a basis for more detailed interpre-
ELLERMAN, J.R. (1940). The families and genera of living
tation of results on ultrasonic approaches in a precocial
rodents. Volume I. Rodents other than Muridae. Trustee of
rodent. Advances in Ethology, 37 : 145.
he British Museum of Natural History, London.
NOVACEK, M.J. (1998). In : BUTLER, D. Museums research
GEORGE, W. (1981). The diet of Petromus typicus (Petromuri-
comes off list of endangered species. Nature, 394 : 115-117.
dae, Rodentia) in the Augrabies Falls National Park. Koedoe,
NOWAK, R.M. (1999). Walker's Mammals of the World. Johns
24 : 159-167.
Hopkins University Press, Baltimore.
GEORGE, W. & G. CROWTHER (1981). Space partitioning between
PFRETSCHMER, H.-U. (1997). Biomechanik des Zahnschmelzes.
two small mammals in a rocky desert. Biological Journal of
In : ALT & TÜRP (eds), Die Evolution der Zähne. Phylogenie
the Linnean Society, 15 : 195-200.
­ Ontogenie ­ Variation, Quintessenz Verlag, Berlin : 391-
GRAAFF, G. DE (1981). The rodents of Southern Africa. Butter-
400.
worths, Durban & Pretoria.
RATHBUN, G.B. & C.D. RATHBUN (this volume). Dassie-rat
HESSE, R. & F. DOFLEIN (1935). Tierbau und Tierleben in ihrem
(Petromus typicus) feeding ecology and petrophily. Belgian
Zusammenhang betrachtet. Erster Band : Der Tierkörper als
Journal of Zoology.
selbständiger Organismus. Fischer, Jena.
SCHULTZ, J. (2000). Handbuch der Ökozonen (S. 364-403 : Tro-
HOLTMEIER, F.-K. (2002). Tiere in der Landschaft. Einfluss und
pische/Subtropische Trockengebiete). Eugen Ulmer, Stutt-
ökologische Bedeutung. Ulmer, Stuttgart.
gart.
LANDRY, S.O. Jr. (1970). The Rodentia as omnivores. Quarterly
SKINNER, J.D. & R.H.N. SMITHERS (1990). The mammals of the
Review of Biology, 45 : 351-372.
Southern African subregion. University of Pretoria, Pretoria,
L
RSA.
UCAS, P.W. (1979). The dental-dietary adaptations of mam-
mals. Neues Jahrbuch für Geologie und Paläontologie,
STARCK, D. (1995). Teil 5 : Säugetiere. In : STARCK (ed), Bd. 2.
Monatshefte, 8 : 486-512.
Wirbeltiere. Lehrbuch der Speziellen Zoologie (begr. v.
M
K
CKENNA, M.C. & S.K. BELL (1997). Classification of mam-
AESTNER, A.), Fischer, Jena, Stuttgart, New York : 1-1241.
mals above the species level. Columbia University Press,
THENIUS, E. (1989). Zähne und Gebiß der Säugetiere. In : NIET-
New York.
HAMMER, SCHLIEMANN & STARCK (eds), Handbuch der
M
Zoology, Volume VIII Mammalia, de Gruyter, Berlin.
EESTER, J. (1965). The origins of the southern African mam-
mal fauna. Zoologica Africana, 1 : 87-95.
TULLBERG, T. (1899/1900). Ueber das System der Nagethiere :
M
eine phylogenetische Studie. Akademische Buchdruckerei,
ESS, A. (1999a). The rostral nasal skeleton of hystricognath
rodents : evidence on their phylogenetic interrelationships.
Uppsala.
Mitteilungen aus dem Museum für Naturkunde zu Berlin,
VAUGHAN, T.A., J.M. RYAN & N.J. CZAPLEWSKI (2000). Mamm-
Zoologische Reihe, 75 : 19-35.
alogy. Sanders College Publ., Fort Worth.
MESS, A. (1999b). Plazentation von Petromus typicus, ein Bei-
WILSON, D.E. & D.M. REEDER (1993). Mammal species of the
trag zur Reproduktionsbiologie und evolutiven Differenzie-
world. A taxonomic and geographic reference. Smithsonian
rung der Hystricognathi. Courier Senckenberg, 215 : 153-
Institution Press, Washington, London.
157.
WITHERS, P.C. (1979). Ecology of small mammal community
MESS, A. (2001). Evolutionary differentiation of placental
and a rocky outcrop in the Namib desert. Madoqua, 2 : 229-
organisation in hystricognath rodents. In : DENYS, GRANJON
246.
& POULET (eds), African Small Mammals, IRD Éditions, col-
WITHERS, P.C. (1983). Seasonal reproduction by small mammals
lection Colloques et séminaires. Paris : 279-292.
of the Namib desert. Mammalia, 47, 105-204.
MESS, A. (2002). Petromus typicus : reproductive biology of a
WITHERS, P.C., G.N. LOUW & J. HENSCHEL (1980). Energetics
poorly known animal. Research Commitee Newsletter, 8 :
and water relations of Namib desert rodents. South African
38-40.
Journal of Zoology, 15 : 131-137.


Belg. J. Zool., 135 (supplement) : 53-56
December 2005
Reproductive rhythm of the grass rat, Arvicanthis abys-
sinicus
, at the Entoto Mountain, Ethiopia

Tilaye Wube
Department of Biology, Addis Ababa University, P.O.Box 1176, Addis Ababa Ethiopia
Current address : Beit Margolin, Department of Biology, Haifa University (Oranim Campus), 36006, Tivon, Israel
Corresponding author : Tilaye Wube, e-mail : tilayewube@yahoo.com
ABSTRACT. Data on the reproductive rhythm of the grass rat, Arvicanthis abyssinicus, were collected in a bush
area and a crop field at the Entoto Mountain for a one-year period from August 2000 to July 2001. Very few individ-
uals of Arvicanthis were captured from the bush area. The number of captures from the crop field, however, was
much higher. Higher percentages of reproductively active males and females were recorded during the rains and the
couple of months that followed. Breeding decreased in the dry months. The presence of cereals on the field did not
seem to augment breeding during the dry months. The outcome of the present study is consistent with most previous
results where breeding in Arvicanthis sp. primarily occurs during the rains and the few months that follow. It is rec-
ommended that physiological studies should be conducted for a better understanding of the reproductive timing of
Arvicanthis in an eco-physiological context.
KEY WORDS : Entoto, Breeding rhythm, Arvicanthis abyssinicus, Ethiopia.
INTRODUCTION
(MULLER, 1977; SICARD et al., 1996). NEAL (1981)
reported a year round reproduction in Western Uganda
Arvicanthis (Lesson, 1822) is an endemic genus to
where the climate is described as uniform and less sea-
Africa and distributed along the equatorial belt from Sen-
sonal.
egal to Somalia and across the Nile Delta from Egypt to
Tanzania (D
Several explanations have been proposed to explain the
ELANY & MONRO, 1986). It is identified as
one of the major pest rodents whenever occurring around
various patterns of reproductive timing in Arvicanthis.
cultivated lands (T
SICARD et al. (1996) suggested that Arvicanthis stopped
AYLOR & GREEN, 1976). The taxonomy
of the genus is controversial. With an understanding of
breeding during the rainy season in Burkina Faso due to
the need for further revision of its taxonomy, populations
the gonadoinhibitory effect of long day photoperiod that
of the Nile Delta and West Africa are conventionally
prevails during the season. MULLER (1977), on the other
grouped under Arvicanthis niloticus (Ruppell, 1842).
hand, explains his observation on Ethiopian Arvicanthis
About 5 species are described from East African popula-
as one happening due to physiological stress from the
tions. These are A. blicki (Frick, 1914), A. abyssinicus, A.
cold in the Simien Mountain Highlands of the country. He
dembeensis (Ruppell, 1842), A. somalicus (Thomas,
suggested that the rats could be physiologically too
1903) and A. lacernatus. The first 4 species occur in Ethi-
stressed to reproduce under the prevailing cold. NEAL
opia (Y
(1981) did not observe any correlation between quality of
ALDEN et al., 1976). Some authorities tend to lump
A. dembeensis with A. abyssinicus (M
food and reproduction in central Kenya. Rather, he
USSER & CARLETON,
1993).
attributes the cessation of breeding at the end of the dry
season to increased temperature. He also concludes,
The breeding rhythm of the genus is well documented
based on his results from Western Uganda, that Arvican-
in various parts of Africa. However, results show that fur-
this is capable of continuous breeding and as such the
ther investigations are always needed to understand its
question that shall be addressed is "What makes Arvican-
breeding rhythm more comprehensively. Most studies
this to stop breeding?" and not "What initiates it to?"
show that breeding in Arvicanthis begins with the start of
GHOBRIEL & HODIEB (1982) on the contrary, emphasized
the rains (DELANY, 1964; GHOBRIEL & HODIEB, 1982;
the importance of nutritious food (cereals, seeds and ani-
FISHER, 1991; BEKELE & LEIRS, 1997). Even then, the
mal matter) in playing crucial role to determine the timing
exact timing of breeding during the rainy season showed
of breeding in Arvicanthis. The gonadostimulatory effect
inconsistencies. In Central Kenya, Arvicanthis started
of green stuffs and the availability of drinking water were
breeding few weeks after the start of the rains (NEAL,
also suggested by other researchers (References in NEAL,
1981). While in Nakuru, another Kenyan locality, breed-
1981) to explain the reproductive timing of Arvicanthis.
ing began about a month later (DELANY & MONRO, 1986).
In Nigeria, breeding began a month before the rains
The present study documents data on the breeding
(RABIU & FISHER, 1989). In Ethiopia and Burkina Faso,
activity of A. abyssinicus at the Entoto Mountain, Ethio-
however, Arvicanthis bred only during the dry season
pia from August 2000 to July 2001.

54
Tilaye Wube
METHODS
ear) were taken. Females with perforate vagina, large nip-
ples or well vascularized and distended uteri were consid-
The Entoto Mountain is located about 8 km North of
ered to be reproductively active. If implanted embryos
Addis Ababa. Most part of the mountain is covered with
were observed, they were identified as pregnant. Males
eucalyptus trees. The forest is under government protec-
with scrotal testes, very well visible epididymal tubules
tion. Part of it is inhabited by farmers who cultivate cereal
and large seminal vesicles were considered to be sexually
crops like wheat, barley, and teff. Since eucalyptus trees
active. Individuals that weighed less than 35 g were iden-
discourage growth of ground cover, rodents were not
tified as juveniles, since there were no individuals below
expected to be found in the forest proper of the mountain.
that weight, which showed any sign of sexual maturity.
Within the forest there were clearings covered with dense
bushes which were thought more suitable for rodents.
TABLE 1
One area (100 m x 50 m) with this kind of vegetation
Captured rodents from the two study sites at Entoto Moun-
(height up to 175 cm) was selected to be one of the two
tain.
study sites. It was dominated by plant species like Heli-
chrysum shimperi, Chilocephalum shimperi, Echinops

Species
Bush
Crop
macrochaetus and Conyza schimperi. The grass species
Andropogon amethystinus was also found abundantly.
Arvicanthis abyssinicus
24
167
The second study site was established inside and around a
Desmomys harringtoni
126
7
cultivated land. This site was planted with barley. The
Praomys albipes
74
32
two sites were distantly separated (about 4 km) with few
Lophyromys flavopantatus
34
17
chances of migration of rodents between one another.
Total
258
223
The study area has one rainfall peak during the months
of June-August. During the study period, there was also a
small rainy period in March and May. The smallest
RESULTS
amount of rainfall was recorded during the months Octo-
ber to February. February was the month with the highest
A total of 481 rodents belonging to four species were
average maximum temperature while December and Jan-
captured (Table 1). Of these, 258 were captured in the
uary recorded the lowest (Fig. 1a).
bush habitat and the other 223 in the crop field. The bush
In the bush, trapping started at the beginning of August
habitat rodent fauna was dominated and heavily infested
and continued for a year on a monthly interval. In the crop
by Desmomys harringtoni (Thomas, 1903) with a total
field trapping started from November. Each trapping ses-
capture of 126 individuals (48.8%) while Arvicanthis
sion of the month lasted for three consecutive days and
abyssinicus was the least abundant (n=24) making up
nights. Victor Mouse Snap Traps were used (small and
only 9.3% of the total. On the other hand, the crop field
large versions). Traps baited with peanut butter were set
was completely dominated by A. abyssinicus (n=167 or
near rodent pathways, burrows or sites with good bush
74.9 %) and D. harringtoni was the least abundant spe-
cover under rocks and in crevices. Traps were checked in
cies (n=7 or 3.1%). The other two species, Praomys albi-
the morning (9 :00 a.m.) and late afternoon (5 :00 p.m.).
pes (Ruppell, 1842) and Lophyromys flavopunctatus
After retrieving the catches, standard body measurements
(Thomas, 1888), were also more abundant in the bush
(body weight and length of head-body, tail, hind foot and
habitat than in the crop field (Table 1).
TABLE 2
Seasonal variation of captured Arvicanthis abysinicus from the bush habitat (light font) and the crop field (bold font)
2000
2001
Aug.
Sept.
Oct.
Nov.
Dec.
Jan.
Feb.
Mar.
Apr.
May
Jun.
Jul.
Females
0
3
2
2
0
1
0
0
0
0
0
0
3
11
15
16
8
6
5
8
4
Males
2
5
2
2
0
0
0
0
0
0
1
0
4
3
13
18
13
8
6
9
2
Juveniles
0
0
0
0
1
3
0
0
0
0
0
0
3
2
3
2
3
1
0
0
1
Total
2
8
4
4
1
4
0
0
0
0
1
0
10
16
31
36
24
15
11
17
7
Table 2 summarizes the catch distribution of A. abys-
ruary (n=36), the lowest in July (n=7). The male to female
sinicus in both study sites. In the bush area, the highest
sex ratio was 1 :1.
number of A. abyssinicus was recorded in September
Figure 1B shows the percentage of reproductively
while between February and July none was captured
active males and females from the crop field. Higher pro-
except for the single specimen in June. The adult male to
portions of reproductively active males and females were
female sex ratio was 1.4 :1. Only 4 juveniles were cap-
obtained during the rains and the first two months that
tured, in December and January. In the crop field, the
followed. In the crop field, there was a high percentage of
highest numbers of A. abyssinicus were recorded in Feb-
reproductively active males and females in November

Reproductive rhythm of the grass rat, Arvicanthis abyssinicus, at the Entoto Mountain, Ethiopia
55
and December. During these two months, all the captured
In the crop habitat several individuals of A. abyssinicus
males were sexually active and except for one individual
were captured. The observed reproductive activity rhythm
in December, all the females were pregnant. Reproductive
was in conformity with most previous observations that
activity declined in both sexes from January onwards. For
report peak reproduction in A. abyssinicus during and
the bush area, data were too scarce to discuss reproduc-
shortly after the rains (TAYLOR & GREEN, 1976; NEAL,
tion.
1981; GHOBRIEL & HODIEB, 1982). Of course, in the
present study it is not clear what happened during August
(a) 400
25
- October 2000 because the data collection started only in
November. Even then, the fact that the reproductive activ-
20
)
ity started to increase together with the start of the rain in
300
C
(o

the 2001 rainy season i.e. June, it could be reasonable to
mm)
15
r
e

speculate that there had been the same trend during
ll ( 200
f
a

atu
August - October 2000.
er
in
10
a
p
Reproductive activity was highly reduced during the
R
m
100
e
5
T
dry months, despite the presence of cereals (nutritious
food) in that season. This was against Taylor and Green's
0
0
observation in Western Uganda where artificially sup-
A
S
O
N
D
J
F
M
A
M
J
J
plied cereals modified the breeding peak (TAYLOR &
GREEN, 1976). Even though the barley harvest was com-
(b) 100
pletely collected in April and May, the crop was on field
during all the previous months. If the presence of cereals
80
had any gonadostimulatory influence, the percentage of
e
g

60
reproductively active A. abyssinicus should not have sig-
t
a

nificantly decreased in the dry months. However, this
r
cen

40
needs to be substantiated by data from a longer study
e
P

period.
20
0
ACKNOWLEDGEMENTS
A
S
O
N
D
J
F
M
A
M
J
J
2000
2001
The Aklilu Lemma Research Grant Fund (ALF) is duly
Months
acknowledged for providing the research fund. The Ethiopian
Fig. 1. ­ (a) Monthly
Heritage Trust also receives my sincere appreciation for allow-
total rainfall (bars) and average maximum and minimum tem-
ing me to conduct the research in part of the Entoto Mountain. I
perature (lines) and (b) percentage of reproductively active
am also grateful to the Ethiopian Meteorological Service
males and females from the crop field (white bars : females;
Agency for providing the rainfall and temperature data of the
bl k b
l )
study area for the study period. Afework Bekele is duly
acknowledged for the provision of the rodent traps. I also would
DISCUSSION
like to sincerely acknowledge Mesfin Kebede who has been my
assistant during all months of the data collection period. My
appreciation also goes to all the people who helped me in the
A. abyssinicus was always rare or absent in the bush
research process one way or another and particularly the guards
area. Earlier studies showed that Arvicanthis prefers habi-
at the seedling preparation station of the Ethiopian Heritage
tats which have good hiding places and aerial cover
Trust at Entoto. I am very grateful to the anonymous referees
(WUBE & BEKELE, 2001). The bush habitat of the present
and editors for constructive comments and suggestions on ear-
study was also such a habitat. However, it can be sug-
lier drafts of the manuscript.
gested that Arvicanthis here gave priority to cultivated
fields where cereals can be obtained abundantly while
REFERENCES
avoiding the less nutritious and crowded bush area. Dur-
ing the study period, the bush habitat harboured a number
BEKELE, A. & H. LEIRS (1997). Population ecology of rodents of
of other rodent species while A. abyssinicus dominated
maize fields and grassland in central Ethiopia. Belg. J. Zool.,
the crop field. Generally it can be hypothesized that, 1)
127 : 39-48.
there might have been competition among rodents of the
DELANY, M.J. (1964). A study of the ecology and breeding of
region for the most nutritious crop fields and the competi-
small mammals in Uganda. Proc. Zool. Soc. Lond., 142 :
tion ended in favour of A. abyssinicus while the rest of the
347-370.
species were relegated to the bush habitat or 2) the other
DELANY, M.J. & R.H. MONRO (1986). Population Dynamics of
rodent species did not show interest for the crop habitat
Arvicanthis niloticus (Rodentia Muridae) in Kenya. J. Zool.,
even though it was nutritious. Rather they gave priority to
Lond., 209 : 85-103.
the presence of cover and hiding places while A. abyssini-
FISHER, M. (1991). A reappraisal of the reproductive ecology of
Arvicanthis in Africa. Afr. J. Ecol., 29 : 17-27.
cus managed to survive in the more exposed crop fields
F
by digging nests and taking refugee under the available
OX, B.J. & J.H. BROWN (1993). Assembly rules for functional
groups in north American desert communities. Oikos, 67 :
rocks and crevices. It has been suggested earlier that there
358-370.
is always one dominant species in an area which is shared
GHOBRIEL, I. & A.S.K. HODIEB (1982). Seasonal variations in
by different populations that have similar niche (FOX &
the breeding of the Nile rat (Arvicanthis niloticus). Mamma-
BROWN, 1993).
lia, 46 : 320-333.

56
Tilaye Wube
MULLER, J.P. (1977). Population okologie von Arvicanthis abys-
region of Burkina Faso (formerly Upper Volta). Can. J.
sinicus in der grassteppe des Semien Mountains National
Zool., 72 : 201-214.
Park (Athopien). Z.Saugetierk., 42 : 145-172.
TAYLOR, K.D. & M.G. GREEN (1976). The influence of rainfall
MUSSER, G.G. & M.D. CARLETON (1993). Family Muridae. In :
on diet and reproduction in African rodent species.
WILSON, D.E. & D.H. REEDER (eds), Mammal species of the
J.Zool.Lond., 180 : 367-389.
world. A taxonomic and geographic reference. Washington
D.C. : Smithsonian Press : 501-755.
WUBE, T. & A. BEKELE (2001). Habitat preference of the African
N
grass rat, Arvicanthis dembeensis (Rodentia, Muridae), in
EAL, B.R. (1981). Reproductive biology of the unstriped grass
rat, Arvicanthis, in East Africa. Z. Saugetierk., 46 : 174-189.
Koka, Central Ethiopia. In : DENYS, C., L. GRANJON & A.
R
P
ABIU, S. & M. FISHER (1989). The breeding season and diet of
OULET (eds), Proceedings of the 8th African Small Mam-
Arvicanthis in northern Nigeria. J. Trop. Ecol., 5 : 375-386.
mals Symposium, Paris, France. 499-510.
SICARD, B., D. MAUREL, F. FUMINIER & J. BOSSIN (1996). Cli-
YALDEN, D.W., M.J. LARGEN & D. KOCK (1976). Catalogue of
mate, trophic factors and breeding patterns of the Nile grass
the mammals of Ethiopia. 2. Insectivora and Rodentia.
rat (Arvicanthis niloticus solatus) : a 5 year study in Sahlian
Mon.Zool.Ital. (N.S.) (Suppl.).

Belg. J. Zool., 135 (supplement) : 57-61
December 2005
Comparative physiology of heat production in rodents
under increasing salinity : The effects of habits and hab-
itat

Abraham Haim, Uri Shanas and Michael Scantlebury*
Department of Biology, University of Haifa, Oranim, Kiryat Tivon 36006, Israel, *Present address : Mammal research Insti-
tute, Department of Zoology and Entomology, University of Pretoria, Pretoria 0002, South Africa
Corresponding author : Abraham Haim, e-mail : ahaim@research.haifa.ac.il
ABSTRACT. Small mammals inhabiting environments that are either seasonally or perpetually dry, such as Medi-
terranean or desert ecosystems respectively, commonly have physiological capabilities that enable them to deal with
water shortage. We compared results of thermoregulatory responses of rodent species inhabiting different habitats
and having varying activity periods, when salinity increases in their water source, as often occurs in their natural
habitats during the dry period.
Experimental animals were maintained on a diet of dry soy-beans and an increased salinity of their water source (2%
agar gel), from 0.9% to 3.5% NaCl in mesic species and up to 7% in xeric species. While desert species could cope
with high salinities in their water source, mesic species could not. Desert-adapted species depending on their preferred
micro-habitats differ in their thermoregulatory responses. Rock dwellers, such as the golden spiny mouse Acomys rus-
satus
and the bushy tailed gerbil Sekeetamys calurus, reduce their resting metabolic rates (RMR) and increase nonshiv-
ering thermogenesis (NST) capacity in response to increasing salinity. In contrast, the deep burrowing fat jird Meri-
ones crassus
, increases RMR and only slightly increases NST-capacity.
Our study suggests that species occupying different habitats vary their thermoregulatory capabilities, in relation to
dehydration and increasing salinity in the water source. This may be a consequence of adaptation from the original
ecosystem to the current environment in which a species inhabits.
KEY WORDS : Nonshivering thermogenesis, resting metabolic rate, aridity, thermoregulation, kidney function
INTRODUCTION
concentrated urine, and efficient heat economy is typified
by resting metabolic rates (RMR) that are lower than the
Small mammals must find their food and water
expected from body mass, according to allometric equa-
resources in their immediate surroundings. Therefore,
tions (KLEIBER, 1961; HART, 1971; HAIM & BORUT, 1981;
they can be used as indicators of habitat quality. Israel,
HAIM, 1987; HAIM & IZHAKI, 1993). The relative medul-
being a transition zone for biogeographically different
lary thickness (RMT) is an anatomical variable that can
regions (T
be used for predicting kidney function (SPERBER, 1944;
CHERNOV & YOM-TOV, 1988), inhabits rodent
species of different evolutionary origins. Furthermore, the
SHKOLNIK, 1988; WEISSENBERG & SHKOLNIK, 1994).
landscape changes in relation to precipitation and altitude,
Therefore, it is expected that species with high RMT indi-
from a sub-alpine ecosystem on Mount Hermon, through
ces will be able to increase their urine concentration dur-
the Mediterranean and steppe ecosystems and finally to
ing spells of drought that lead to increased salinity of
the extreme arid ecosystem. Rodent species show differ-
water sources.
ent distribution patterns, ranging from the occupation of a
Deserts can be cold at nights, and endotherms such as
single ecosystem as the golden spiny mouse Acomys rus-
rodents that possess low RMR's have to increase heat
satus (arid ecosystem), through to species with wide dis-
production over a short period to maintain their body tem-
tributional ranges as in the case of the common spiny
perature (HAIM & LEVI, 1990). Nonshivering thermogene-
mouse A. cahirinus (arid, steppe and Mediterranean eco-
sis (NST) is an important mechanism for heat production
systems).
in small mammals such as rodents (JANSKY, 1973). Fur-
thermore, this mechanism has been found to compensate
A comparison of physiological variables, such as water
for the lower RMR values in species like desert rodents
economy and heat production, between different species
(H
from disparate environments, or divergent populations
AIM & IZHAKI, 1993).
within the same species from distinct habitats, is of signif-
The objectives of this study were to compare the ther-
icant importance for the understanding of adaptation to
moregulatory responses of different rodent species sub-
the environment. Studies (SCHMIDT-NILSEN, 1964; SHKOL-
jected to osmolarity challenges. Specifically, to examine :
NIK & BORUT, 1969; WEISSENBERG & SHKOLNIK, 1994)
(1) If such a challenge will have an impact on heat pro-
have demonstrated that efficient water economy can be
duction, by means of NST? (2) If NST values can be
examined through the ability of the kidney to produce a
related to the pattern of activity and habitat? This paper

58
Abraham Haim, Uri Shanas and Michael Scantlebury
compares five different species of rodents, some of an
Resting metabolic rate (RMR) was measured for each
African origin, that occur in Israel, while others are of a
species as the minimal oxygen consumption (VO Min) at
2
Palearctic origin but occur also in Africa.
1oC below its lower critical temperature. Body tempera-
ture (T Min) was measured at the end of VO Min meas-
b
2
urements, by inserting a copper-constantan thermocouple
MATERIAL AND METHODS
3cm deep into the rectum of the experimental individual.
The thermocouple was connected to a TH-65 Wescor dig-
Animals
ital thermometer.
VO NA was measured as the maximal response of VO
Data were collected on the following rodents : Com-
2
2
to a noradrenalin (NA) injection (Sigma) 1.5mg/Kg
mon spiny mouse (Acomys cahirinus), Golden spiny
mouse (Acomys russatus), Bushy tailed gerbil (Sekeeta-
(HELDMAIER, 1972; HAIM et al., 1995). Approximately
mys calurus), Fat jird (Meriones crassus), Tristram's jird
20min. after VO NA values were achieved and VO lev-
2
2
(Meriones tristrami) (Table 1). In all instances the experi-
els started to decline, the experimental individual was
mental animals were fed crude soybeans that were dried
removed from the metabolic chamber and its body tem-
for 48h at 60oC to a constant weight. Water was supplied
perature was measured once again and presented as
in the form of 2% agar gel (20g of dry agar dissolved in
T NA. NST-capacity was calculated as the ratio of
b
1000ml of de-ionized water) to which desired salinity lev-
VO NA to VO Min (R
2
2
ON & HAIM, 2001; SCANTLEBURY
els were achieved by dissolving appropriate amounts of
et al., 2002).
NaCl. Body mass was measured every second day, during
the acclimation periods. When a loss of more than 20% in
Statistics : All values are given as mean ± SD for n = 7.
body mass was recorded, the experimental individual was
Results showed a normal distribution and therefore Stu-
removed from the experiment. Following each acclima-
dent t-test was used for statistical analysis.
tion period (14 days) to a given salt concentration, urine
volume, urine osmolarity and nonshivering thermogene-
sis (NST) variables were measured.
RESULTS
TABLE 1
Characteristics of the studied species, data are taken from HAR-
A marked difference was noted between species from
RISON & BATES (1991).
mesic and xeric habitats. The xeric species comprising S.
calurus
, A. russatus and M. crassus, survived on agar
Species
Region
Habitat
Activity
with a 7% salinity, whereas the mesic species M. tristrami
A. russatus
Xeric
Rock dweller
Diurnal
and A. cahirinus were only able to withstand 3.5% (Table
A. cahirinus
Mesic
Rock dweller
Nocturnal
2). Meriones crassus showed the lowest drop of body
S. calurus
Xeric
Rock dweller,
Nocturnal
mass (10.1%) when acclimated to 7% salinity (Table 2).
M. crassus
Xeric
Burrow
Nocturnal
The highest osmolarity values in urine were recorded in
M. tristrami
Mesic
Burrow
Nocturnal
A. russatus and M. crassus (Table 3), and among the
desert species the lowest values were found in S. calurus.
Urine collection and variables analysis
TABLE 2
For each level of salinity, the animal was placed in a
The response of body mass to increased salinity in the water
mesh net cage (19.5 x 11.5 x 9cm) above a sheet of Para-
source of dehydrated individuals, of different studied rodent
film for a period of 24h. Urine was collected with a Pas-
species. Values presented as total body mass (gr) for 0.9%; 3.5%
teur pipette and subsequently placed into Eppendorff
(mesic species) and 7% (xeric species) salinity of the water
tubes and stored at 4oC every 6h. Urine volume was
source and as the % difference between salinities. Data for A.
measured using a Gilson pipette to the accuracy of 1µl.
russatus is from RON & HAIM (2001); for A. cahirinus from
Urine Osmolarity was measured using a Wescor 5500
SHANAS et al. (2003); for S. calurus from PALGI & HAIM (2003);
Vapor Pressure Osmometer (PALGI & HAIM, 2003).
for M. Crassus, from HAIM (UNPUBLISHED) and for M. tristrami,
from NEUMAN et al. (2000).
NST variables
Species
Salinity (%)
Wb (gr)
Wb (%)
Oxygen consumption (VO ) was measured using an
A. russatus
0.9
50.4
±6
22.6
2
open flow system (D
7
39.0
±8.2
EPOCAS & HART, 1957). The air was
A. cahirinus
0.9
43.1
±7
19.5
pumped into the metabolic chamber using a pump (Aqua-
3.5
34.7
±7.6
Serene). Oxygen concentrations were measured from the
S. calurus
0.9
59.7
±9.2
19.1
air exiting the metabolic chamber using, an oxygen ana-
7
48.9
±5.5
M. crassus
0.9
100.1
±15.8
10.1
lyzer (Servomex 750A) connected to a multimitter
7
90.0
±9.3
(Tabor). The air was dried with a silica-gel column at the
M. tristrami
0.9
79.2
±8.3
17.7
entrance and exit ports of the metabolic chamber.
3.5
65.2
±7.9

Heat production in rodents
59
ues decreased (1.7oC) with the increase of salinity in the
water source (Table 3).
Fig. 1. ­ Resting metabolic rates RMR (mlO /g.h) of each
2
studied species under control conditions (0.9% salinity of the
Fig. 2. ­ The maximal VO (mlO /g.h) response to noradrena-
water source) and of maximal salinity (3.5% for mesic species
2
2
lin injection (VO NA) of each studied species under control
and 7% for the xeric ones). Data for A. russatus RON & HAIM
2
conditions (0.9% salinity of the water source) and of maximal
(2001); data for S. calurus PALGI & HAIM (2003); data for M.
salinity (3.5% for mesic species and 7% for the xeric ones).
crassus HAIM (UNPUBLISHED); for M. tristrami NEUMAN et al.
Data sources are as in Fig. 1.
(2000) and data for A. cahirinus SCANTLEBURY et al. (2002).
Apart from M. crassus, the RMR of all species showed
a decrease as the salinity of the water increased (Fig. 1).
The VO response to NA increased significantly (P<0.01)
2
only in M. crassus, when the salinity in the water source
was increased (Fig. 2). NST-capacity increased signifi-
cantly (P<0.001) with the rise in salinity only in A. russa-
tus
and in S. calurus (Fig. 3). T Min values generally
b
decreased in all species, apart from S. calurus (P<0.05 for
M. crassus and P<0.01 for A. russatus) as a response to
the increase in salinity (Table. 3). The sharpest decrease
(1.5oC) was observed in A. russatus and the lowest
decrease (0.6oC) in M. crassus and in M. tristrami. The
lowest T Min values were measured for A. russatus
Fig. 3. ­ Nonshivering thermogenesis NST-capacity (VO NA/
2
b
(34.8oC) under a salinity of 7% of the water source. The
RMR) of the studied species under control conditions (0.9%
highest increase of T NA as a response to increase in
salinity of the water source) and of maximal salinity (3.5% for
b
mesic species and 7% for the xeric ones). Data sources are as
salinity was observed in M. crassus, (1.5oC) while in
in Fig. 1.
some of the studied species as in A. cahirinus T NA val-
b
TABLE 3
Minimal body temperature (T Min) and the maximal body temperature response to a
b
noradrenalin injection (T NA), maximal urine concentration of each species at control and in
b
maximal salinity of the water source. RMT relative medullary thickness. Values for T are
b
taken from RON & HAIM (2001) for A. russatus; from SCANTLEBURY et al. (2002) and SHANAS
et al. (2003) for A. cahirinus; from PALGI & HAIM (2003) for S. calurus; from HAIM (UNPUB-
LISHED) for M. crassus and from NEUMAN et al. (2000) for M. tristrami. RMT values for A.
cahirinus
are taken from WEISSENBERG & SHKOLNIK (1994) for all other species values are
taken from BROSH (1971).
Species
Salinity (%)
T Min
T NA
Osmolarity
RMT*
b
b
A. russatus
0.9
36.3±0.3
38.2±0.8
5353±725
11.4
7
34.8±1.1
37.8±1.3
9123±3292
A. cahirinus
0.9
36.7±0.8
37.5±0.3
3006±485
9.3
3.5
35.0±0.9
35.8±0.7
3389±623
S. calurus
0.9
37.3±0.2
39.6±0.9
3197±1167
9.03
7
37.4±0.6
39.7±0.7
7091±1729
M. crassus
0.9
35.8±0.5
36.8±0.7
2450±830
10.3
7
35.2±0.3
38.3±0.3
9642±2066
M. tristrami
0.9
36.8±0.6
38.0±0.5
1500±120
8.2
7
36.2±0.5
37.7±0.4
3210±523
* Values are taken from BROSH, 1971; for A. cahirinus from WEISSENBERG & SHKOLNIK, 1994.

60
Abraham Haim, Uri Shanas and Michael Scantlebury
DISCUSSION
result of the decrease in heat production with or without
any change in heat dissipation. In S. calurus the decrease
Many desert rodent species show low RMR values
in heat production with the increase in salinity is accom-
(DEGEN, 1997). This physiological trait enables them to
panied by a decrease in heat dissipation and as a result
conserve water and to keep a balanced heat exchange. In
T Min does not decreases. However, in the case of M.
b
addition, for species that do not engage in reproduction
crassus, T Min decreased by 0.6oC although heat produc-
b
annually as a result of the harsh unpredictable desert con-
tion increased. Therefore, it is suggested that in M.
ditions, lower RMR values could result as an adaptive
crassus, the desert burrow dwelling species, heat dissipa-
longevity trait (HAIM, 1987; HAIM & IZHAKI, 1993). How-
tion increases with the increasing of salinity.
ever, in deserts, nights can be cool even during summer.
As both spiny mice are rock dwellers with a poor ther-
Therefore, increased NST is an important and efficient
mal refuge that use evaporative mechanism for heat dissi-
mechanism for heat production and thermoregulation in a
pation (SHKOLNIK & BORUT, 1969; WEISSENBERG &
cold environment (HAIM & LEVI, 1990). It was also noted
SHKOLNIK, 1994), the decrease in T Min is assumed to be
that acclimation to heat or dehydration increases NST-
b
an important contribution for thermoregulation, as it con-
capacity (Fig. 3), compared with control groups at 0.9%
serves water. In contrast, the bushy tailed gerbil S. calu-
salinity (HOROWITZ & SAMUELOFF, 1989; YAHATA et al.,
rus, shows the same distribution pattern as A. russatus
1999; RON & HAIM, 2001).
and the same habitat, but is nocturnal and digs shallow
We show that rock dwelling, desert rodent species such
burrows, which may be used as a thermal refuge (HARRI-
as, A. russatus and S. calurus can further decrease their
SON & BATES, 1991; PALGI & HAIM, 2003). Under such
RMR values (RON & HAIM, 2001; PALGI & HAIM, 2003),
conditions, it can maintain its T Min values even under
b
in response to increasing salinity (of dehydrated individu-
dehydration conditions, whereas the diurnal A. russatus
als). In contrast, the soil form burrow dwelling M.
low T values, will enable it to forage for longer periods
b
crassus, a desert adapted species (HAIM & TCHERNOV,
under the hot conditions during day time in its habitat
1974), did not reduce it's RMR under the same conditions
(HAIM et al., 1998).
(Fig. 1). As the thermal refuge for rock dwellers is less
In conclusion, the results of our study indicate that
efficient than a deep burrow (HAIM et al., 1998), it is pro-
thermoregulatory mechanisms in different rodent species,
posed that the difference in habitat may play an important
respond differently to dehydration caused by a high pro-
role in the thermoregulatory response to dehydration. A
tein diet and increasing salinity of the water source.
decrease in RMR values was noted also in the two mesic
Desert species can tolerate higher salinity values, twice
species at 3.5% salinity, but these RMR values are much
those tolerated by mesic species. Thermoregulatory
higher than those of the rock dwelling desert adapted spe-
response varies between the different species and is
cies at 7% salinity and are close to those of M. crassus at
affected by the time of activity (habits) and by the species
7% salinity (Fig.1).
habitat. The spiny mice of the genus Acomys are depend-
The absence of a significant change in VO NA, apart
ent on their kidneys for survival in the desert since they
2
from the increase observed in M. crassus, suggests that
have a poor thermal refuge and use water for evaporative
although the response to increased salinity is a reduction
cooling. This latter phenomenon may indicate that the ori-
in RMR, it seems to have no effect on the response to
gin of this genus is from an environment where water was
noradrenalin (Fig. 2). These results tend to suggest that
not limited. Therefore, when facing water shortage in
the number and sensitivity of the adrenalin receptors in
xeric environments, spiny mice decrease their T Min val-
b
the brown adipose tissues (BAT) do not change under the
ues and as consequence conserve water.
current conditions of dehydration, as was suggested by
REDLIN et al. (1992) for the difference in thermogenic
REFERENCES
capacity of juvenile rats.
Food consumption (apparent digestible dry matter
BROSH, A. (1971). Correlation between concentration abilities
intake ­ ADDMI ­ and digestible energy) responds to
and kidney structure. Oranim, School of Education of the
photoperiodic manipulations (HAIM & LEVI, 1990). Yet,
Kibbutz Movement (In Hebrew).
the observation that VO NA increases in M. crassus as a
2
DEGEN, A.A. (1997). Ecophysiology of small desert mammals.
result of dehydration may indicate that decreased food
Springer, Berlin.
quality, increases the number or affinity of the adrenergic
DEPOCAS, F. & J.S. HART (1957). Use of Pauling oxygen ana-
receptors in this species. In addition to a secure thermal
lyser for measurement of oxygen consumption of animals in
refuge, M. crassus has an efficient kidney and among the
open circuit system and in short lag, closed circuit apparatus.
tested desert species, has the highest ability to concentrate
J. Appl. Physiol., 20 : 388-392.
its urine at 7% salinity (Table 2). This finding supports
HAIM, A. (1987). Metabolism and thermoregulation in rodents :
the kidney anatomy where the relative medullary thick-
are these adaptations to habitat and food quality? S. Afr. J.
ness (RMT) is high (10.3, B
Sci., 83 : 639-642.
ROSH, 1971).
HAIM, A. & A. BORUT (1981). Heat production and dissipation
In S. calurus, T Min does not decrease in response to
b
in golden spiny mouse Acomys russatus from two extreme
an increase in salinity. Body temperature is the outcome
habitats. J. Comp. Physiol., 142B : 445-450.
of two opposite physiological processes, namely heat pro-
HAIM, A. & I. IZHAKI (1993). The ecological significance of rest-
duction and heat dissipation. In S. calurus heat production
ing metabolic rate and nonshivering thermogenesis for
(as reflected by VO ) decreases and has the same value as
2
rodents. J. Therm. Biol., 18 : 71-81.
A. russatus. Yet, in A. russatus T Min drops at 7% salinity
b
HAIM, A. & G. LEVI (1990). Role of body temperature in sea-
by 1.5oC. This drop of T Min (Table 2) may be a direct
b
sonal acclimatization : Photoperiod-induced rhythms and

Heat production in rodents
61
heat production in Meriones crassus. J. Exp. Zool., 256 :
RON, E. & A. HAIM (2001). How dehydration affects the ther-
237-241.
moregulatory and osmoregulatory abilities of the golden
HAIM, A., R.M. MCDEVITT & J.R. SPEAKMAN (1995). Daily var-
spiny mouse Acomys russatus. Isr. J. Zool., 47 : 15-20.
iations in the response of wood mice Apodemus sylvaticus to
SCANTLEBURY, M., D. AFIK, U. SHANAS & A. HAIM (2002).
noradrenaline. J. Exp. Biol., 198 : 561-565.
Comparative nonshivering thermogenesis in adjacent popu-
HAIM, A. & TCHERNOV, E (1974). The distribution of myomorph
lations of the common spiny mouse (Acomys cahirinus)
rodents in the Sinai peninsula. Mammalia, 38 : 201-223.
from opposite slopes : the effects of increasing salinity. J.
HAIM, A., R.J. VAN AARDE & N. ZISAPEL (1998). Body tempera-
Comp. Physiol., 172B : 1-5.
ture daily rhythms in the striped mouse Rhabdomys pumilio :
S
The effects of alpha and beta blockade. Physiol & Behav.,
CHMIDT-NIELSEN, K. (1964). Desert animals. Oxford Univer-
sity Press, Oxford.
63 : 889-893.
H
S
ARRISON, D.L. & P.J.J. BATES (1991). The mammals of Arabia.
HANAS, U., D. AFIK, M. SCANTLEBURY & A. HAIM (2003). Dif-
2nd. Ed. Harrison Zoological Museum, Lakeside Printing,
ferential osmoregulatory capabilities of common spiny
Sevenoaks, Kent.
mouse (Acomys cahirinus) from adjacent microhabitats. J.
H
Zool. Lond., 261 : 7-13.
ART, J.S. (1971). RODENTS. In : WHITTOW, G.C., (ed), Compar-
ative physiology of thermoregulation, Academic Press, New
SHKOLNIK, A. (1988). Physiological adaptations to the
York.
environment : the Israeli experience. In : YOM-TOV, Y. & E.
HELDMAIER, G. (1972). Cold adaptive changes of heat produc-
TCHERNOV (eds), The Zoogeography of Israel, Dr. Junk Pub-
tion in mammals. In : SMITH, R.E., (ed), Bioenergetics Inter-
lishers, The Hague : 487-496.
national Symposium on Environmental Physiology, FASEB,
SHKOLNIK, A. & A. BORUT (1969). Temperature and water rela-
MD : 79-82.
tions in two species of spiny mice (Acomys). J. Mammal.,
HOROWITZ, M. & S. SAMUELOFF (1989). Dehydration stress and
50 : 245-255.
heat acclimation. Prog. Biometeorol., 7 : 91-99.
SPERBER, I. (1944). Studies on the mammalian kidney. Zool.
JANSKY, L. (1973). Nonshivering thermogenesis and its ther-
Bidr. Uppsala, 22 : 249-431.
moregulatory significance. Biol. Rev., 48 : 85-132.
K
TCHERNOV, E. & Y. YOM-TOV (1988). Zoogeography of Israel.
LEIBER, M. (1961). The fire of life. Wiley, New York.
N
In : YOM-TOV, Y. & E. TCHERNOV (eds), The Zoogeography
EUMAN, A., E. RON & O. RAZ-EL (2000). Heat production by
nonshivering thermogenesis (NST) in the jird (Meriones
of Israel, Dr. Junk Publishers, The Hague : 1-6.
tristrami) : is it affected by dehydration? Isr. J. Zool., 46 :
WEISSENBERG, S. & A. SHKOLNIK (1994). Metabolic rate and
381 (abstract).
water economy in the desert and Mediterranean populations
PALGI, N. & A. HAIM (2003). Thermoregulatory and osmoregu-
of the common spiny mouse (Acomys cahirinus) in Israel.
latory responses to dehydration in the bushy-tailed gerbil
Isr. J. Zool., 40 : 135-143.
Sekeetamys caslurus. Arid Envir. (In press)
YAHATA, T., F. FURUYAMA, T. NAGASHIMA, M. MORIYA, K.
REDLIN, U., B. NUESSLEIN & I. SCHMIDT (1992). Circadian
KIKUCHI-UTSUMI, T. KAWADA & A. KUROSHIMA (1999).
changes of brown adipose tissue thermogenesis in juvenile
Thermoregulatory responses of the inbred heat tolerant FOK
rats. Am. J. Physiol., 262 : R504-R508.
rat to cold. Am. J. Physiol., 46 : R362-R367.


Belg. J. Zool., 135 (supplement) : 63-67
December 2005
Spatial distribution of commensal rodents in regions
with high and low Lassa fever prevalence in Guinea

Elisabeth Fichet-Calvet1, Kékoura Koulémou2, Lamine Koivogui2, Barré Soropogui2, Oumar
Sylla
2, Emilie Lecompte1, Stéphane Daffis3, Allali Kouadio4, Stéphane Kouassi4, Chantal
Akoua-Koffi
4, Christiane Denys1 and Jan ter Meulen3
1 Laboratoire Mammifères & Oiseaux, Museum National d'Histoire Naturelle, 55 rue Buffon, 75 005 Paris, France
2 Projet de Recherches sur les Fièvres Hémorragiques en Guinée, Centre Hospitalier Donka, CHU Donka, BP 5680, Conakry,
Guinée
3 Institute of Virology, Philipps University, Robert-Koch-Str. 17, 35037 Marburg, Germany
4 Institut Pasteur d'Abidjan, BP490, 01 Abidjan, Côte d'Ivoire.
Corresponding author : Elisabeth Fichet-Calvet e-mail : ecalvet@club-internet.fr
ABSTRACT. Lassa fever is a hemorrhagic fever caused by an arenavirus, which affects approximately 150,000 per-
sons per year in West Africa. It is principally transmitted by rodents of the genus Mastomys, which serve as both
reservoir and vector of the virus. This study tested the hypothesis that human Lassa fever in Guinea is related to the
occurrence rate of the multimammate rat, Mastomys spp, inside houses. The analysis was based both on Lassa virus
antibody surveys in the human population and the commensal rodent distribution in the same prefectures. The ana-
lysis took into account several data sets compiled from the literature (LUKASHEVICH et al., 1993; DEMBY et al., 2001)
and data from our own ongoing rodent trapping activities in Guinea. The analysis revealed a probable regional gra-
dient of introduced rodent species in houses, with the black rat, Rattus rattus, predominating on the coast and the
house mouse, Mus musculus, predominating approximately 200 km southeast to the coast. The autochthonous spe-
cies, Mastomys spp, were present deep inside the country, from High to Forest Guinea. The regression analysis of
Mastomys occurrence on human Lassa virus antibody prevalence showed a positive correlation in six administrative
regional districts (prefectures) between an increasing Mastomys occurrence and increasing seroprevalence. A sev-
enth prefecture, where mainly M. musculus occurred, was discordant with this trend, and possible explanations for
this divergence are discussed. The partial replacement of Mastomys by other species, especially R. rattus and Myo-
mys
daltoni, is discussed as a potential explanation for the low-prevalence of Lassa fever in certain areas.
KEY WORDS : Lassa fever, risk, prevalence, rodent-borne disease, Mastomys, occurrence.
INTRODUCTION
by KEENLYSIDE et al. (1983) and MCCORMICK et al. (1987)
in Sierra Leone. The principle modes of infection are
Lassa fever is a hemorrhagic fever caused by an arena-
thought to be exposure to highly infectious urine of
virus, which affects approximately 150,000 persons per
chronically infected rodents (rev. in MCCORMICK, 1999)
year in West Africa. In the Republic of Guinea, the activ-
and the direct handling of animals captured for human
ity of Lassa fever varies according to different geographi-
consumption (TER MEULEN et al., 1996).
cal zones, as determined by human seroprevalence stud-
ies. In the northern and coastal regions human antibody
Some species such as M. erythroleucus, M. huberti and
prevalences are generally low (3 ­ 5%), whereas they
M. natalensis are distributed in West Africa (BRAMBELL
may be as high as 40% in southern Guinea, close to the
& DAVIS, 1941; DUPLANTIER et al., 1990; BRITTON-DAV-
borders with Sierra Leone and Liberia (L
IDIAN et al., 1995 ; GRANJON et al., 1997). Their presence
UKASHEVICH et
al., 1993; T
has also been confirmed in Senegal (DUPLANTIER et al.,
ER MEULEN et al., 1996; BAUSCH et al., 2001).
1990) and in Mali (GRANJON, pers. com.). As their pre-
The disease is principally transmitted by rodents of the
cise identification is difficult, the multimammate rats
genus Mastomys, the multimammate rats, which serve
recently investigated in our study for Lassa virus and
both as reservoirs and vectors of the virus (SALAZAR-
antibodies in Guinea were classed as Mastomys spp.
BRAVO et al., 2002), and they might be infected by up to
according to DEMBY et al. (2001) in order to avoid false
30% in Lassa-endemic areas (MONATH et al., 1974;
identifications.
KEENLYSIDE et al., 1983; MCCORMICK et al., 1987;
DEMBY et al., 2001). Lassa antibodies have been detected
We hypothesised that a high occurrence of Mastomys in
in a small number of other rodent species; however, Lassa
houses is a possible risk factor of human Lassa fever. To
virus has rarely been isolated from other genera (WULFF
this end, we performed a combined analysis of the litera-
et al., 1975; DEMBY et al., 2001) The primary human
ture data and our own data generated during ongoing eco-
infections occur in villages, particularly in houses
logical and genetic studies on commensal rodents to eval-
infested by many rodents belonging to this genus as noted
uate the eco-epidemiology of Lassa fever in Guinea.

64
Fichet-Calvet, Koulémou, Kovoigui, Soropogui, Sylla, Lecompte, Daffis, Kouadio, Kouassi, Akoua-Koffi, Denys and ter Meulen
MATERIAL AND METHODS
An additional data set obtained from Kedougou in
southeastern Senegal by BA (2002) was included in the
Background on Lassa virus activity in humans
map presentation. From July 1998 to April 2001, the cap-
tures in houses were mainly composed by Mastomys spp.
From 1990 to 1992, Lukashevich and his collaborators
(90%; 380/432) and R. rattus (9%; 38/432).
conducted a large epidemiological survey of Lassa virus
activity in human populations in the Republic of Guinea
(LUKASHEVICH et al., 1993). They sampled 25 villages,
Own data on rodent community
distributed in different prefectures and established the
Lassa virus antibody prevalences based on ELISA. Their
Some rodents were collected in houses, distributed in 3
results are shown in table 1, where the mean seropreva-
villages on the coastal region, Bamba (10°00'N, 13°53'W,
lences by prefecture were calculated.
20 m a.s.l., ±650 inhabitants) and Yafraya (10°05'N,
13°40'W, 20 m a.s.l., ±1600 inhabitants) in October 2002,
TABLE 1
and Gayebombo (10°08'N, 13°35'W, 86 m a.s.l., ±80
inhabitants) in May 2003. The rodents were trapped with
Prevalence of Lassa Virus-specific antibodies by prefecture in
BTS and Ugglan traps in 3-days sessions during October
Guinea. Adapted from LUKASHEVICH et al., 1993.
2002 and with Sherman traps during May 2003. Between
four and six traps were set at 30 households. The location
Prevalence in %
Prefecture
of traps was partly directed by the household and varia-
(N°positive/N°tested)
tion in building structures. Usually, two traps were set by
Kindia (Madina Oula)
34 (59/171)
room, e.g. bedroom, corridors, food storage, or an exter-
Boffa
4 (6/160)
nal kitchen and were baited with a mixture of peanuts,
Boké
5 (5/102)
dried fish and wheat flower. Thirty houses were sampled
Pita
6 (10/165)
Labe
7 (8/111)
with a total of 456 trap-nights.
Mali
5 (9/176)
Faranah
35 (149/420)
In May 2003, an additional village in the Fouta Djal-
Siguiri
11 (45/418)
lon, Gagal (11°05'N, 12°17'W, 1050 m a.s.l., ±760 inhab-
Guékédou
37 (226/604)
itants) was similarly sampled using Sherman traps, where
Yomou
27 (119/441)
Lola
28 (102/358)
16 houses were sampled, giving 210 trap-nights.
Background on rodent community
Data analysis
From 1996 to 1997, a team from the Centers for Disease
The analysis proposed here is the combination of data
Control (CDC), Atlanta, tested rodent populations in differ-
compiled from 1) the human prevalence based on the
ent regions of Guinea for the presence of Lassa virus and
paper of Lukashevich, 2) the commensal rodent commu-
antibodies (DEMBY et al., 2001). They investigated 26 vil-
nity based on the paper of Demby and 3) our own data
lages, distributed in different regions which partially over-
based on two recent rodent trapping sessions. To access to
lap with those of Lukashevich. They captured rodents at
the correlation between human Lassa virus prevalence
444 house sites and 7 bush sites. As the major part of the
and Mastomys occurrence, a simple linear regression
rodent collection was made in houses, we considered their
analysis was used with the dependent variable being the
results as mainly commensal rodents. Table 2 gives the
human prevalence and the independent variable being the
occurrence rate of Mastomys, Rattus and Mus (percent of
Mastomys occurrence. Seven prefectures entered into the
total captures) in their trapping. To be concordant with the
regression, which was performed with Statview 5, SAS
human data, the rodent data are partitioned by prefecture.
Institute Inc. (1998).
In the Kindia prefecture, DEMBY et al. (2001) sampled both
the town (60,000 inhabitants) and three villages belonging
to the Madina Oula district. To account for the bias of the
RESULTS
high occurrence of Mus musculus in large towns (DUPLAN-
TIER et al., 1991; DUPLANTIER et al., 1997), the correlation
analysis was performed without the data from Kindia town.
Commensal rodents on the coast
and in the Fouta Djallon
TABLE 2
The distribution of the commensal rodents according to
Occurrence of the three main commensal species in villages by
each sampled village showed that on the coast, Rattus rat-
prefecture in Guinea, adapted from DEMBY et al., 2001. * Data
tus was very abundant in large villages such as Bamba
from Kindia prefecture showing the captures from the town
and Yafraya, whereas Mastomys spp remained the only
itself were excluded, see text for explanation.
species in small villages such as Gayebombo (Table 3).
Prefecture
Mastomys spp. Rattus rattus
Mus musculus
To include these data in the analysis, the Mastomys occur-
rence was calculated by combining the collection of the
Kindia*
8 (24/295)
9 (26/295)
81 (240/295)
three coastal villages, indicating that Mastomys com-
Faranah
94 (271/289)
0
0
prised 25% (15/61) of the rodent community. In Fouta
Kissidougou
92 (71/77)
8 (6/77)
0
Guékédou
94 (505/536)
5 (25/536)
0
Djallon, the Mastomys community in houses was 45% (4/
Yomou
94 (505/536)
5 (25/536)
0
9), with the majority of captures identified as Myomys
Lola
94 (505/536)
5 (25/536)
0
daltoni.

Spatial distribution of commensal rodents in Guinea
65
TABLE 3
Trapping results from the coastal (Boffa and Dubreka) and the Fouta Djallon (Pita) regions in Guinea. Bamba
and Yafraya were sampled in October 2002 whereas Gayebombo and Gagal were sampled in May 2003.
Prefecture
Village
Mastomys
Rattus
Myomys
Praomys
Cricetomys
Mastomys occurrence
Boffa
Bamba
3
20
0
1
0
25% (15/61)
Dubreka
Yafraya
0
24
0
0
1
Gayebombo
12
0
0
0
0
Pita
Gagal
4
0
5
0
0

45% (4/9)
This is also supported by the studies performed in the
Tongo Field area in Sierra Leone where 26% (248/953) of
humans were Lassa antibody positive and Mastomys spp.
constituted approximately 80% (311/383) of the captures
(KEENLYSIDE et al., 1983). However, this correlation was
not observed in the prefecture of Kindia, where three rural
villages (Madina Oula, Kagbele and Dar es Salaam) had a
high human prevalence of Lassa antibodies (34%) and a
low occurrence of Mastomys spp. (8%). One explanation
for this contradiction could be due to the fact that a Lassa
fever outbreak occurred in this region in 1982-83, leading
to 137 deaths (BOIRO et al., 1987). The majority of these
cases may have originated from human-to-human trans-
mission from the potential introduction of a highly infec-
Fig. 1. ­ Linear regression of percent Mastomys occurrence
tious Lassa virus variant. It is, therefore, possible that a
and percent human Lassa seroprevalence in seven prefectures
high percentage of survivors could still be seropositive in
in Guinea. The question mark is related to the unknown epide-
the absence of Lassa virus infected Mastomys in the
miological situation when the Mastomys occurrence is null.
houses. Alternatively, Mastomys might have been
replaced by M. musculus, between the occurrence of the
epidemic in the 80s and the study performed by Demby
Human seroprevalence
more than 10 years later. This could be due to the intense
related to Mastomys occurrence
population movements and traffic between the two coun-
tries, particularly during the civil war in Sierra Leone
The regression analysis showed a non-significant coef-
from 1991 to 2002. In such a situation, we can predict that
ficient (r2 = 0.226, p = 0.280) when the seven prefectures
the Lassa fever risk in the Madina Oula zone will
were included (Fig. 1). However, the regression was sig-
decrease if M. musculus persists in colonizing the house-
nificant (r2 = 0.904, p = 0.004) when Kindia was excluded
holds. It is also possible that contact with infected rodents
from the analysis. These results were mapped, indicating
does not take place in houses but in the fields and bush
missing data related to human seroprevalence or com-
during hunting of Mastomys and other rodents as a food
mensal rodents (Fig. 2).
source (TER MEULEN et al., 1996). It is expected that our
ongoing investigations will help to understand the eco-
epidemiology of Lassa fever in regions with a low preva-
DISCUSSION
lence of Mastomys spp.
Our study on commensal rodents in coastal Guinea and
in the Fouta Djallon, combined with the data of Demby,
ACKNOWLEGEMENTS
showed that there could be a gradient of introduced spe-
cies such as Rattus rattus and Mus musculus, from the
This study was supported by a grant from the Howard
coast to the highlands. It is, therefore, suggested to survey
Hughes Medical Institute, USA, to Jan ter Meulen and a grant
inside the country, from the Fouta Djallon to Forest
from the European Community, INCO-DEV program (grant
Guinea, to observe autochthonous species such as Masto-
ICA4-Ct2002-10050). We are grateful to Amadou Doré and
mys spp. and Myomys daltoni in households. A Rattus
Fodé Kourouma for their technical assistance in the field and to
rattus gradient has been described by DUPLANTIER et al.
Dr Steven Belmain for his useful comments on the earlier ver-
(1991, 1997) in Senegal, particularly on the axis between
sion of this article.
Casamance and the Southeastern region near Kedougou.
It is suggested that R. rattus is moving further inland with
REFERENCES
humans and increasing its prevalence with increasing
transportation activity and improved infrastructure.
BA, K. (2002). Systématique, écologie et dynamique de popula-
The hypothesis that the Lassa fever risk is directly cor-
tions de petits rongeurs potentiellement réservoirs ou hôtes
related to the magnitude of the occurrence of Mastomys
de virus au Sénégal. Diplôme de l'Ecole Pratique des Hautes
spp. inside houses was partially verified in our analysis.
Etudes, Montpellier, 126 p.

66
Fichet-Calvet, Koulémou, Kovoigui, Soropogui, Sylla, Lecompte, Daffis, Kouadio, Kouassi, Akoua-Koffi, Denys and ter Meulen
Fig. 2. ­ Distribution of Lassa virus human prevalence and Mastomys occurrence by prefecture in Guinea.
BAUSCH, D.G., A.H. DEMBY, M. COULIBALY, J. KANU, A. GOBA,
Mastomys in Senegal. Zeitschrift für Zoologische Systematik
A. BAH, N. CONDE, H.L. WURTZEL, K.F. CAVALLARO, E.
und Evolutionsforschung, 28 : 289-298.
LLOYD, F. BINTA BALDET, S.D. CISSE, D. FOFONA, I.K.
DUPLANTIER, J.M., L. GRANJON, F. ADAM, & K. BA (1991).
SAVANE, R. TAMBA TOLNO, B. MAHY, K.D. WAGONER, T.G.
Répartition actuelle du rat noir (Rattus rattus) au Sénégal :
KSIAZEK, C.J. PETERS & P.E. ROLLIN (2001). Lassa fever in
facteurs historiques et écologiques. In : LE BERRE M. & L.
Guinea : I. Epidemiology of human disease and clinical
LE GUELTE (eds), Le rongeur et l'espace, Editions Chabaud,
observations. Vector Borne and Zoonotic Diseases, 1 : 269-
R. : 339-346.
281.
B
D
OIRO, I., N.N. LOMONOSSOV, V.A. SOTSINSKI, O.K. CONSTANTI-
UPLANTIER, J.M., L. GRANJON & K. BA (1997). Répartition
biogéographique des petits rongeurs au Sénégal. J. Afr.
NOV, E.A. TKACHENKO, A.P. INAPOGUI & C. BALDE (1987).
Eléments de recherches clinico-épidémiologiques et de labo-
Zool., 111 : 17-26.
ratoire sur les fièvres hémorragiques en Guinée. B. Soc.
GRANJON, L., J.M. DUPLANTIER, J. CATALAN, & J. BRITTON-DAV-
Pathol. Exot., 80 : 607-612.
IDIAN (1997). Evolutionary systematics in the genus Masto-
BRAMBELL, F.W.R. & D.H.S. DAVIS (1941). Reproduction of the
mys (Thomas, 1905) (Rodentia : Muridae) : a review. Belg.
multimammate mouse (Mastomys erythroleucus Temm.) of
J. Zool., 127 : 7-18.
Sierra Leone. Proc. Zool. Soc., 111B : 1-11.
KEENLYSIDE, R.A., J.B. MCCORMICK, P.A. WEBB, E. SMITH, L.
BRITTON-DAVIDIAN, J., J. CATALAN, L. GRANJON & J.M.
ELLIOTT & K.M. JOHNSON (1983). Case-control study of
DUPLANTIER (1995). Chromosomal phylogeny and evolution
Mastomys natalensis and humans in Lassa virus-infected
in the genus Mastomys (Mammalia, Rodentia). J. Mammal.,
households in Sierra Leone. Am. J. Trop. Med. Hyg., 32 :
76 : 248-262.
829-837.
DEMBY, A.H., A. INAPOUGUI, K. KARGBO, J. KONINGA, K. KOUR-
L
OUMA, J. KANU, M. COULIBALY, K.D. WAGONER, T.G.
UKASHEVICH, L.S., J.C. CLEGG & K. SIDIBE (1993). Lassa virus
K
activity in Guinea : distribution of human antiviral antibody
SIAZEK, C.J. PETERS, P.E. ROLLIN & D.G. BAUSCH (2001).
Lassa fever in Guinea : II. Distribution and prevalence of
defined using enzyme-linked immunosorbent assay with
Lassa virus infection in small mammals. Vector Borne and
recombinant antigen. J. Med. Virol., 40 : 210-217.
Zoonotic Diseases, 1 : 283-296.
MCCORMICK, J.B. (1999). Lassa fever. In : J.F. SALUZZO & B.
DUPLANTIER, J.M., J. BRITTON-DAVIDIAN & L. GRANJON (1990).
DODET (eds), Emergence and control of rodent-borne viral
Chromosomal characterization of three species of the genus
diseases, Elsevier, Paris : 177-195.

Spatial distribution of commensal rodents in Guinea
67
MCCORMICK, J.B., P.A. WEBB, J.W. KREBS, K.M. JOHNSON &
TER MEULEN, J., I. LUKASHEVICH, K. SIDIBE, A. INAPOGUI, M.
E.S. SMITH (1987). A prospective study of the epidemiology
MARX, A. DORLEMANN, M.L. YANSANE, K. KOULEMOU, J.
and ecology of Lassa fever. J. Infect. Dis., 155 : 437-444.
CHANG-CLAUDE & H. SCHMITZ (1996). Hunting of perido-
MONATH, T.P., V.F. NEWHOUSE, G.E. KEMP, H.W. SETZER & A.
mestic rodents and consumption of their meat as possible
CACCIAPUOTI (1974). Lassa virus isolation from Mastomys
risk factors for rodent-to-human transmission of Lassa virus
natalensis rodents during an epidemic in Sierra Leone. Sci-
in the Republic of Guinea. Am. J. Trop. Med. Hyg., 55 : 661-
ence, 185 : 263-265.
666.
SALAZAR-BRAVO, J., L.A. RUEDAS & T.L. YATES (2002). Mam-
WULFF, H., A. FABIYI & T.P. MONATH (1975). Recent isolation
malian reservoirs of Arenaviruses. Curr. Top. Microbiol.,
of Lassa virus from Nigerian rodents. B. World Health
262 : 25-63.
Organ., 52 : 609-613.


Belg. J. Zool., 135 (supplement) : 69-75
December 2005
Noki or dassie-rat (Petromus typicus) feeding ecology
and petrophily

Galen B. Rathbun1 and Carolyn D. Rathbun2
1 Department of Ornithology and Mammalogy, California Academy of Sciences (San Francisco), c/o PO Box 202, Cambria,
California 93428, U.S.A.
1 Piedras Blancas Field Station, Western Ecology Research Center, United States Geological Survey, San Simeon, California
93452, U.S.A.
2 PO Box 202, Cambria, California 93428, U.S.A.
Corresponding author : Galen B. Rathbun, e-mail : grathbun@calacademy.org
ABSTRACT. The noki or dassie-rat (Petromus typicus) is a rupicolous diurnal herbivore that is endemic to the
southwestern arid biogeographical zone of Africa. It is the only representative of the hystricognath family Petromu-
ridae. During our study of the social structure of nokis, we gathered data on their feeding behaviours during four
periods totaling nearly seven months from 2000 through 2003.
Nokis fed on a wide variety of plants with a preference for fresh leaves and stems, fruits, and flowers when available.
They also ate a wide variety of dry leaves and stems. Relatively few plants and plant parts were avoided. There was no
indication that food was cached or stored. There was no evidence that nokis and rock hyraxes (Procavia capensis)
competed for food resources, despite often feeding together. We speculate that a previously undescribed and unusual
noki behaviour (the "tail-stand") is associated with coprophagy.
In hot and arid regions of Australia, some herbivorous mammals feed on the rich flora in rocky areas that is associated
with unique water regimes often found in these habitats. A similar relationship may partially explain why nokis are
endemic to rocky habitats in the southwestern arid zone of Africa. Other features contributing to the adaptive syn-
drome include their phylogeny and historical zoogeography, a need for dietary water, a low metabolic rate, a flexible
and diverse diet, and an unusual suite of behaviours associated with digestion. The result is an obligate petrophile.
KEY WORDS : Diet, Dassie-rat, Feeding, Namibia, Noki, Petromus, Petrophily, Rupicolous
INTRODUCTION
Nokis superficially resemble ground squirrels, includ-
ing their largely diurnal activity. Apart from general natu-
ral history observations (e.g., SKINNER & SMITHERS, 1990)
The noki or dassie-rat (Petromus typicus A. Smith,
and reports based largely on opportunistic observations
1831) belongs to the monospecific hystricognath family
(e.g., COETZEE, 1983), there are only two field studies of
Petromuridae. We prefer the common name "noki"
noki ecology (WITHERS, 1979; GEORGE & CROWTHER,
because it avoids the confusion by many people between
1981). Recently, reproduction and behaviour of captive
dassie-rats and rock dassies (rock hyraxes in the mamma-
nokis have been studied (MESS, 2002).
lian order Hyrcoidea) and true rats (species in the rodent
families Muridae and Cricetidae). Noki is derived from a
Because the noki has a limited distribution, there is
Hottentot dialect (S
considerable interest in this near-endemic Namibian fam-
HORTRIDGE, 1942) and was used by
G
ily of rodents (GRIFFIN, 1998). In this paper we report
EORGE & CROWTHER (1981).
information on noki feeding ecology that we gathered
Nokis are endemic to Africa in the southwest arid bio-
while studying the social structure and behaviours of free-
geographical region (M
ranging nokis in Namibia.
EESTER, 1965), where they are
closely associated with rocky habitats, especially the
Namibian escarpment zone with its numerous mountains,
METHODS
cliff faces, and inselbergs or kopjes (COETZEE, 2002).
They occur from extreme southwestern Angola south
Our study was near the Erongo Wilderness Lodge (21o
through Namibia, and into northwestern Cape Province of
27.679 S, 15o 52.523 E) on Okapekaha Farm, about 10
South Africa. The aridity of the escarpment and closely
km west of Omaruru town in the foothills of the Erongo
related Namib Desert is at least 15 million years old
Mountains. The site is 1240 m above sea level and is
(WARD & CORBETT, 1990) and the noki has had an ancient
characterised by huge rounded granite dikes and domes
association with these biomes (MEESTER, 1965), as dem-
that rise about 100 m above the surrounding peneplain
onstrated by several morphological adaptations to living
and smaller 10-20 m high granite outcrops or kopjes (Fig.
in rock crevices (GEORGE & CROWTHER, 1981; SKINNER &
1) surrounded by intruding fingers of the surrounding
SMITHERS, 1990). These include a flattened cranium, flex-
bushveld. The vegetation at the study site is composed of
ible ribs, and dorso-lateral mammae.
low trees and bushes interspersed with seasonally dense

70
Galen B. Rathbun and Carolyn D. Rathbun
annual and perennial forbs and perennial bunch grasses.
Annual mean rainfall at Omaruru Prison is 292.9 mm,
The dominant trees include Combretum apiculatum, Ster-
with virtually all of this falling during the months of
culia africana, Terminalia prunoides, and Boscia albi-
November through April (Fig. 2). Annual average mini-
trunca and the more dominant bushes included several
mum and maximum temperatures are 11.4 and 31.0o C,
species of Grewia, Croton gratissimus, Dichrostachys
with May through August being the coolest as well as dri-
cineria, and Mundulea sericea.
est months (Fig. 2).
Fig. 1. ­ Single kopje (3-m-high cluster of boulders in foreground above road) at the Erongo Mountains, Namibia, study site
where many of our observations occurred. Note the dense concentration of food plants (mostly Grewia spp. and bunch grasses)
at the base of the granite rock.
200
40
175
35
150
30
Co
mm
125
ure,
25
Maximum
100
Minimum
perat
20
75
15
Mean Rainfall,
50
Mean Tem
10
25
5
0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
0
Month
Jan
Feb Mar
Apr May Jun
Jul
Aug Sep
Oct
Nov Dec
Month
Fig. 2. ­ Left : Average (40 years) monthly rainfall at Omaruru Prison, located about 10 km from the noki study site. Right : Aver-
age (10 years) maximum and minimum monthly temperatures at Omaruru Prison.

Noki feeding ecology
71
It is particularly difficult to catagorise the phenology of
rence of freezing temperatures and the quantity and tim-
arid-land plants because they respond very differently to
ing of the seasonal rainfall. To draw some generalizations
the high year-to-year climatic variation (HÜSER, 1976).
about noki feeding ecology, we have subjectively
For example, whether and when some perennial plants
assigned some phenological traits to the common plants
lose their leaves (generally those classified as "inconsist-
at out study site (Table 1), realizing that in some instances
ently deciduous" in Table 1) is closely tied to the occur-
this is probably an oversimplification.
TABLE 1
Common plants found at our noki study site (alphabetical order by genus). Those
that are especially associated with boulder habitats (e.g., noki habitat) in the
Erongo area (personal observations and P. Carven, personal communication) are
indicated by an asterisk (*) in the "Traits" column, while those without an asterisk
also are commonly found in the surrounding bushveld. Plants not observed eaten
by nokis during the 2000-2003 study period are indicated with a dash (--) in the
"Parts Eaten" column. In the "Traits" column, we assigned one feature from each
of the following groups (separated by commas): perennial (P) or annual (A), decid-
uous (D) or inconsistently deciduous (I), tree (T) or bush (B) or forb (F) or grass
(G). We subjectively ranked each food plant in importance (most = 1 and least = 3)
to noki diet based on our observations.
Scientific Name
Parts Eaten
Traits
Abutilon angulatum
Green leaves & stems
P, D, B, 3, *
Abutilon fruticosum
Green leaves & stems
P, D, B, 3, *
Abutilon ramosum
Green leaves & stems
P, D, B, 3, *
Acacia erubescens
Green leaves
P, D, T, 3
Adenolobus garipensis
Flowers
P, I, B, 3, *
Barleria lancifolia
--
P, D, B, *
Barleria sp.
--
P, D, B
Bidens biternata
Entire green plant, dry stems
A, D, F, 2
Blepharis obmitrata
Green leaves
P, D, B, 3
Boscia albitrunca
Flowers
P, I, T, 3
Cardiospermum pechuelii
--
P, D, B
Combretum apiculatum
Dry leaves
P, D, T, 1
Commiphora glaucescens
Dry leaves & stems
P, D, T, 3, *
Croton gratissimus
--
P, D, T, *
Cyphostemma omburense
Green & dry leaves
P, D, F, 1
Dichrostachys cinera
Green leaves
P, D, B, 3
Dombeya rotundifolia
--
P, D, B, *
Dyerophytum africanum
--
P, D, B, *
Enneapogon scoparius
Dry stems
P, D, G, 2
Erythrina decora
--
P, D, T, *
Ficus ilicina
--
P, I, T, *
Forsskaolea viridis
Entire green plant
A, D, F, 1
Grewia bicolor
Green & dry leaves & fruit
P, D, B, 1
Grewia flava
Fruit
P, D, B, 2
Grewia flavescens
Green leaves
P, D, B, 1
Grewia tenax
Green leaves
P, D, B, 1
Helinus integrifolius
Green leaves
P, D, B, 3, *
Hibiscus micranthus
Green leaves & stems
P, D, B, 3
Hibiscus castroi
--
P, D, B
Indigofera filipes
--
P, I, B
Jamesbrittenia pallida
--
P, D, B, *
Lycium basciifolium
Green leaves
P, I, B, 3
Montinia caryophyllacea
Dry leaves
P, D, B, 3 *
Mundulea sericea
--
P, D, B, *
Obetia carruthersiana
--
P, D, T, *
Portulaca sp.
Green leaves
A, D, F, 1
Schmidtia kalahariensis
Dry stems
P, D, G, 2
Solanum rigescentoides
Green leaves
P, D, B, 3, *
Steganotaenia araliacea
--
P, D, T, *
Sterculia africana
Dry leaves, flowers
P, D, T, 2, *
Stipagrostis uniplumis
Dry stems
P, D, G, 2
Talinum arnotii
Green leaves
A, D, F, 1
Terminalia prunioides
--
P, D, T
After determining the suitability of the study site in
15 inch folding aluminium or 16 x 5 x 5 inch single-door
June 2000 we captured, tagged, and observed four to six
wire mesh live traps set during daylight hours and baited
nokis during each of four periods : 25 December 2000
with pieces of raw carrots or apples. To prevent hyper-
through 5 January 2001, 5 September through 21 Novem-
thermia in captured animals we positioned traps in the
ber 2001, 24 April through 7 July 2002, and 10 May
shade or avoided trapping during mid-day.
through 26 July 2003. We caught animals with 4.5 x 4 x

72
Galen B. Rathbun and Carolyn D. Rathbun
We attached radio transmitters with collars made of
of the other plants that we did not see nokis feed on (e.g.,
antenna wire inside Tygon tubing (Holohil Systems Ltd.,
Barleria lancifolia and Hibiscus castroi,), the former
Carp, Ontario, Canada; model MD-2C, 2.2 g weight, 120-
three also showed no evidence of being browsed by the
day battery life, 20-pound test 10-cm-long wire whip
other rupicolous mammals on our study site, such as rock
antenna). We radio-located each of the nokis several
hyrax (Procavia capensis Pallas, 1780), klipspringer
times a day between 0430 and 2230 hours. When air tem-
(Oreotragus oreotragus Zimmermann, 1783), and Jame-
peratures were below about 30o C we sat on top of granite
son's Rock Rabbit (Pronolagus randensis Jameson,
boulders and with 8 x 40 binoculars watched tagged as
1907). Although we often observed nokis feeding on dry
well as untagged nokis. Even with the advantage of being
grass stems (Table 1), which were abundant in some areas
able to always find the radio-tagged animals, observation
at the bases of the kopjes (Fig. 1), we did not see them eat
often was difficult because of obstructing rock, the ani-
or harvest grass seed-heads.
mals' wariness, and their use of narrow and deep rock
The upper portions of the kopjes that the nokis occu-
crevices for shelter.
pied were virtually devoid of growing plant matter, which
required the animals to descend to the base of the rocks to
RESULTS
find food (Fig. 1). They often carried single leafed twigs
or grass stems (up to about 20 cm long) from the bases of
the kopjes to favoured basking and resting spots higher in
The nokis exhibited a catholic diet of plants and plant
the rocks, where they fed on the material, including later
parts (Table 1), including dry fragments of unidentifiable
in the day or on a subsequent day if it was not initially
leaves and stems when fresh plant matter was available.
consumed. For example, in late June 2003 at 0655 hrs.
However, some plants appeared to be particularly impor-
one of the collared male nokis moved from the crevice
tant in the diet of nokis, perhaps because they were espe-
where he spent the night to a crevice at the lower edge of
cially common, nutritious, or moist (importance category
the kopje and began to harvest the green leaves and stems
1 in Table 1). For example, at the end of the dry season
from a Grewia flaviscens bush that was about 1.5 metres
and prior to the rains (September and October, Fig. 2),
from his crevice. During the 35-minute feeding bout he
when nearly 90% of the common plants at our study site
made nine trips to the bush, each time bringing back to his
were leafless (Table 1), several trees flowered. The nokis
crevice a leafed stem, which he ate in his crevice before
often foraged on the surfaces of boulders and the ground
returning for more. After the feeding bout, he started a
under Sterculia africana and Boscia albitrunca trees
session of basking in the sun, which lasted most of the
where they gleaned fallen flowers. After the main rains
morning. Even though nokis often harvested plants, we
(April and May), they foraged on the green leaves, flow-
found no evidence that this material was actually cached
ers, and fruits of bushes, especially several species of
or stored for later consumption.
Grewia (Table 1). With the approach of the dry season
(June and July) nokis focused on plants that still con-
We never observed or radio-tracked the animals further
tained moisture, particularly the vine Cyphostemma
than about 10 m away from the base of kopjes and rock
omburense with its fleshy leaves, and annual forbs (e.g.,
crevices, where they immediately retreated if disturbed.
Forsskaolea viridis and Portulaca sp.) that grew in the
Although they often climbed into bushes and out onto tree
deep shade and moist soil at the base of granite boulders.
limbs to forage, they usually remained within leaping dis-
At the height of the dry season (August and September)
tance (ca. 1 m) of rocks and safety. The high risk of pre-
nokis fed mostly on dry leaf and stem detritus that accu-
dation while foraging was illustrated by three of the
mulated at the bases of rock faces and in rock crevices.
radio-tagged nokis being killed and eaten at favoured for-
Although it was difficult to identify these dry plants, we
aging sites. Even though these three different sites were
suspect they were the same species that the nokis fed on
only 2-3 m from the safety of rock crevices at the bases of
during other parts of the year.
kopjes, we suspect that the diurnal and solitary black
mongooses (Galerella nigrata Thomas, 1928), which we
Another aspect of their habitat is that the plants were
often saw hunting in and around our kopjes, surprised and
highly clumped and the clumps often were composed of
captured the foraging nokis before they could reach the
different species (Fig. 1). Thus, nokis on one kopje had
safety of a rock crevice.
access to different food plants than nokis on a nearby
kopje. For example, the kopje in Fig. 1 lacks several spe-
We tallied radio locations associated with nokis forag-
cies, most notably the trees Boscia albitrunca and Com-
ing or harvesting (we did not include instances of plants
miphora glaucescens and the bushes Adenolobus garip-
being ingested after they had been harvested and carried
ensis and Mundulea sericea. This spatial variation in
up into the kopje) by daylight quarters between 0600 and
species composition made it difficult to determine which
1800 hours in 2001 and 2002. We used the proportion of
plants nokis avoided; we suspect that they actually fed on
our radio-tracking effort in each quarter and total feeding
most plants, even if we only documented them eating 28
bouts to calculate the expected foraging bouts per quarter.
out of the 43 (65.1%) most common plants found associ-
These data, starting with 0600 to 0900 hrs., were 20
ated with kopjes in our study area (Table 1). Some plants,
observed and 10 expected, 4 and 10.5, 4 and 11, and 22
however, were only eaten at specific stages in their phe-
and 18.5. The observed distribution is significantly differ-
nology. For example, the leaves and stems of Montinia
ent from the expected (X2 = 9.82, df = 3, P=0.02), indicat-
caryophyllacea were only eaten once they had dried in
ing that foraging was concentrated in early morning and
late July and August. Other plants seemed to be com-
late afternoon.
pletely avoided, including Croton gratissimus, Jamesbrit-
Three times we observed an unusual behaviour by
tenia pallida and Indigofera filipes. Indeed, unlike some
adults that we call a "tail stand" (Fig. 3). These occurred

Noki feeding ecology
73
on flat basking sites and entailed standing on the front
WITHERS (1979) found little seasonal variation in the
feet and propping up the hindquarters with the down-
diet of nokis, and like nokis at Augrabies Falls (GEORGE,
turned and stiffened tail while vigorously kneading or
1981), they ate predominately the stems and leaves of
scratching the abdomen simultaneously with both rear
grasses and dicotyledonous plants. Nokis at Erongo sea-
feet. Each "tail stand" lasted about 10 seconds. We sus-
sonally ate different species, as illustrated by their focus
pect this behaviour may be related to their feeding habits
on forbs during the wet season (these plants disappeared
(see discussion).
during the dry season). Also, the diet of the Erongo nokis
closely followed the phenology of most plants -- focusing
on flowers, then fruits, and then leaves and stems accord-
ing to the season. It is possible that seasonal differences
were not found at Tumasberg with faecal analysis because
some tissues (e.g., flowers and fruits) are difficult to
detect and there is unlikely to be any significant differ-
ence between green and dry leaves of the same species
after being digested.
The obvious avoidance by nokis of several plants at our
study site (see Results section) is likely due to their con-
taining secondary defence compounds, as evidenced by
their strongly aromatic leaves. It is not clear, however,
why nokis at all three study sites seemed to avoid grass
seed-heads, with their presumed higher energy content
compared to leaves and stems.
Nokis, unlike many desert rodents, are not able to rely
only on metabolic water (WITHERS et al., 1980) and thus
Fig. 3. ­ Adult female noki performing a "tail stand." Drawing
need free-standing water (personal observations; MESS &
based on a photograph of a free-ranging female at a basking
ADE, this volume) or moisture in plants. Indeed, the
site.
importance of water in their diet was shown at Augrabies
Falls by their preference for the bases of grass stems,
Rock hyraxes and nokis often foraged simultaneously
which have a higher water content than tops (GEORGE,
in the same Grewia bushes, sometimes within 10 cm of
1981). Our study site was more mesic compared to
each other, but we never saw any agonistic behaviour
Augrabies Falls and Tumasberg, which average about one
between the two species. They also used the same basking
half and one third the annual rainfall of our site. Because
spots and on 11 occasions we observed single rock
noki metabolic rate is about 25% lower than the predicted
hyraxes displace single nokis from these sites. In these
weight-specific rate (WITHERS et al., 1980), perhaps their
cases the approaching rock hyrax either seemed oblivious
diet at Erongo was influenced more by the nutritional
to the presence of the noki, or it displaced the noki after
quality of food plants than by moisture content.
slowly and cautiously approaching it in a posture that
The nokis at Tumasberg ingested a significant amount
suggested curiosity. In several cases the approach even
of insect material (WITHERS, 1979), whereas insectivory
included an attempt to sniff the basking rodent. In all
was not documented at Augrabies Falls (GEORGE, 1981).
instances, the noki fled only when the rock hyrax
We did not observe nokis searching for or eating inverte-
approached very closely, often to within a few centime-
brates, and captives do not eat insects or meat (MESS &
tres. Although we observed nokis responding to rock
ADE, this volume). It is not clear if insects were ingested
hyrax alarm calls by bolting for cover, they did not react
inadvertently at Tumasberg, or if they were purposefully
to rock hyrax territorial cries. We never had an opportu-
eaten in relation to optimal foraging or water needs. In
nity to determine whether rock hyraxes responded to the
any case, it further demonstrates the flexible diet of nokis.
noki "cheeeeeee" alarm call.
WITHERS (1979), COETZEE (1983), and MESS & ADE
(this volume) describe coprophagy in nokis, and Coetzee
DISCUSSION
also describes captive nokis remasticating food after sit-
ting up on their rear legs and bending their head sharply
The diet of nokis was quantified using faecal analyses
down to the abdomen and then jerking upright. This "jack
in the Augrabies Falls National Park on the Orange River
knife" motion apparently induces regurgitation prior to
in South Africa (GEORGE, 1981) and at Tumasberg, an
remastication. We observed coprophagy and the jack
inselberg in the Namib Desert of Namibia (WITHERS,
knife action, but only a very few times and we were una-
1979), which are about 200 km and 950 km south of our
ble to clearly distinguish the two. Indeed, MESS & ADE
Erongo study site. In general, the diet at these two sites
(this volume) have not observed regurgitation and remas-
was unremarkable; Petromus ate a wide variety of plants
tication in captives and believe that the jack knife action
in rough proportion to their occurrence. However, the
is actually related to male autogrooming of the genitals or
dominant plants and diet at the two sites and our study
possibly masterbation. In any case, if coprophagy and the
site were different with little overlap in species eaten,
jack knife behaviours had not been previously described
which demonstrates the catholic and flexible diet of
(COETZEE, 1983) we probably would not have recognised
nokis. Captive nokis also show a wide tolerance for dif-
them. It is possible that coprophagy occurred more fre-
ferent plant foods (MESS & ADE, this volume).
quently than our observations indicate, especially if it was

74
Galen B. Rathbun and Carolyn D. Rathbun
performed mainly while animals were hidden from view
their leaves. Nokis have several peculiar morphological
in rock crevices. Another possibility is that coprophagy
adaptations to living in rock crevices that suggest a long
and remastication (if it indeed occurs) are related to
association with rocky areas. All these features result in
increasing the efficiency of digesting plant material with a
the noki being an obligate petrophile.
high fiber content (COETZEE, 1983), and are thus more
common where coarser plants and plant parts dominate
ACKNOWLEDGEMENT
the diet, as may be the case at the more arid Augrabies
Falls and Tumasberg study sites. Perhaps the "tail stand"
We are grateful for the logistical support and fellowship pro-
is also related to digestive efficiency -- the aggressive
vided by our colleagues in Namibia, including Mike Griffin and
kneading of the abdomen with the rear feet somehow aid-
Rob Simmons of the Ministry of Environment and Tourism, Tim
ing in coprophagy by manipulating or stimulating the
and Laurel Osborne of Windpoort Farm, and Seth Eisab of the
digestive track.
National Museum. We thank our friends at the Erongo Wilder-
At Augrabies Falls, rock hyraxes and nokis do not
ness Lodge for enthusiastically hosting us, especially owner and
builder Danie Holloway and his sons Sean and Justin, managers
compete for shelters because they use different sites based
Roger and Romilly Fussell and Mike and Helen Warren, and the
on their dissimilar body sizes and the food plants they
entire staff, especially Lindy van den Bosch. Thelma van Ryn
both use apparently are plentiful enough to avoid compe-
assisted us with access to climate data. Patricia Craven gener-
tition (GEORGE & CROWTHER, 1981). The lack of food
ously identified the more "obscure" plants for us. Our work was
caches by either species also suggests that this resource is
completed under a research permit from the Ministry of Envi-
not limited. During our study, we observed no agonistic
ronment and Tourism. We appreciate the assistance of Emcé
behaviours between the two species at feeding or basking
Bouwer of the Marietha Bouwer Agencies in obtaining research
sites, which further supports the absence of competition
visas from the Ministry of Home Affairs. We benefited from the
between the two. Indeed, nokis responded to the alarm
comments and suggestions made on early versions of this paper
calls of rock hyraxes, suggesting that their close spatial
by Cornelius Coetzee, Patricia Craven, Mike Griffin, Andrea
Mess, and an anonymous reviewer.
and temporal association was mutually beneficial because
of increased vigilance for predators.
LITERATURE CITED
In hot and dry regions of Australia, some herbivorous
mammals are closely associated with rocky habitats
BURKE, A. (2002). Plant communities of a central Namib insel-
because they feed on the particularly diverse and produc-
berg landscape. Journal of Vegetation Science, 13 : 483-492.
tive flora at these sites (FREELAND et al., 1988). The rich
BURKE, A. (2003). The role of Namibian inselbergs in contribut-
flora is the result of the water concentrating and retaining
ing to local and regional plant species richness. Biodiversity
characteristics of the rocky areas. A similar moisture-
and Conservation, 12 : 469-486.
related explanation for the vegetative richness of riparian
COETZEE, C.G. (1983). The feeding behaviour of Petromus typi-
zones is more widely recognised. The bases of kopjes and
cus A. Smith, 1831 (Rodentia :Petromuridae). Annales du
rock faces at our study site also supported a relatively rich
Musee Royal de l'Afrique Centrale, Tervuren, Belgique,
flora. Not only were plants denser at the bases of kopjes,
Sciences Zoologiques, 237 : 203-206.
but they seemed to grow taller and remain green longer
COETZEE, C.G. (2002). The distribution and breeding seasons of
the dassie-rat, Petromus typicus (Peteromuridae, Rodentia).
than in the surrounding bushveld. In addition, there was
Folia Zoologica, 51(Supplement 1) : 23-35.
some evidence of greater species diversity. For example,
FREELAND, W.J., J.W. WINTER & S. RASKIN (1988). Australian
of the 43 common species found at our noki study site, 19
rock-mammals : A phenomenon of the seasonally dry trop-
(44.2%) were especially associated with our kopje habi-
ics. Biotropica, 20 : 70-79.
tats (Table 1), while the remainder were more widespread
GEORGE, W. (1981). The diet of Petromus typicus (Petromuri-
in the surrounding bushveld. Similar patterns in plant
dae, Rodentia) in the Augrabies Falls National Park. Koedoe,
communities have been found on other Namibian insel-
24 : 159-167.
bergs (B
G
URKE, 2002, 2003). We believe the rich flora
EORGE, W. & G. CROWTHER (1981). Space partitioning between
associated with the bases of kopjes is an important factor
two small mammals in a rocky desert. Biological Journal of
in providing an abundant, reliable, and seasonally rich
the Linnean Society, 15 : 195-200.
GRIFFIN, M. (1998). The species diversity, distribution and con-
source of food for nokis as well as other rupicolous her-
servation of Namibian mammals. Biodiversity and Conser-
bivorous mammals.
vation, 7 : 483-494.
HÜSER, K. (1976). Der Niederschlagsgang und die Nieder-
CONCLUSIONS
schlagsverteilung im Gebiet des Erongo mittleres Südwest-
afrika. Journal of the South West African Scientific Society,
30 : 7-21.
Nokis exhibit a rupicolous adaptive syndrome charac-
MEESTER, J. (1965). The origins of the southern African mam-
terised by several remarkable and inter-related features of
mal fauna. Zoologica Africana, 1 : 87-93.
their feeding ecology. They are restricted to harsh rocky
MESS, A. (2002). Petromus typicus : reproductive biology of a
habitats where their flexible and catholic herbivorous diet
poorly known animal. University of Erlangen, Berlin, Ger-
is well suited to the rich but clumped, variable, and highly
many, Research Committee Newsletter, 8 : 38-40.
seasonal vegetation. In addition, their low metabolic rate
MESS, A. & M. ADE (2005). Clues on the feeding biology of the
and need for non-metabolic water probably relates to their
dassie rat Petromus typicus (Rodentia, Hystricognathi,
Petromuridae) derived from a breeding colony. Journal of
variable and flexible diet. The unusual behaviours associ-
Belgian Zoology, 135 (Supplement) : 43-49.
ated with digestion enable them to feed efficiently on
SHORTRIDGE, G.C. (1942). Field notes on the first and second
plant parts with a high fibre content, which is likely criti-
expeditions of the Cape Museums mammals survey of the
cal during the dry season when nearly all plants have lost
Cape Province, and description of some new subgenera and

Noki feeding ecology
75
subspecies. Annals of the South African Museum, 36 : 27-
Namib Research. Transvaal Museum Monograph No. 7.
100.
Transvaal Museum, Pretoria, South Africa. 17-26.
W
S
ITHERS, P.C. (1979). Ecology of a small mammal community
KINNER, J.D. & R.H.N. SMITHERS (1990). The mammals of the
Southern African Subregion. University of Pretoria, Pretoria,
on a rocky outcrop in the Namib Desert. Madoqua, 11 : 229-
Republic of South Africa. 769 pp.
246.
WITHERS, P.C., G.N. LLOUW & J. HENSCHEL (1980). Energetics
WARD, J.D. & I. CORBETT (1990). Towards an age for the
and water relations of Namib desert rodents. South African
Namib. In : SEELY M.K. (ed), Namib Ecology : 25 years of
Journal of Zoology, 15 : 131-137.


Belg. J. Zool., 135 (supplement) : 77-82
December 2005
When to live alone and when to live in groups : ecologi-
cal determinants of sociality in the African striped
mouse (Rhabdomys pumilio, Sparrman, 1784)

Carsten Schradin
Ecophysiological Studies Research Group, School of Animal, Plant and Environmental Sciences, University of the Witwa-
tersrand, Private Bag 3, Wits 2050, Johannesburg, South Africa. e-mail : schradin@mweb.co.za
Corresponding author : Carsten Schradin, e-mail : carsten@schradin.com
ABSTRACT. One aim in animal behaviour is to explain why and when animals live in groups. The main approach
has been to compare closely related gregarious and solitary species. Here, I discuss data of a medium sized, diurnal
muroid rodent, the striped mouse, which demonstrates a high level of intraspecific variability of its social system. In
the arid Succulent Karoo, the social structure of the striped mouse is best described as a territorial group living sol-
itary forager with communal breeding and helpers at the nest. Groups can consist of up to 30 adult mice, i.e. four
breeding females, one breeding male and their adult offspring. In contrast, the striped mouse is solitary in the mesic
grasslands of South Africa, with females inhabiting intrasexually exclusive territories and male territories overlap-
ping those of several females. Association between the sexes is limited to mating, and offspring leave their mother's
territory as juveniles. Home ranges in the grasslands are much larger than in the Succulent Karoo. I suggest that the
main ecological reasons for these differences in social organization are food abundance, the availability of suitable
nesting sites, and the possibility of sun-basking. Whether these ecological differences acted as selection pressures in
the past that caused genetic differences and finally speciation (as proposed by a recent study), or whether these eco-
logical differences lead to behavioural differences via an ontogenetic pathway, remains a topic for further research.
KEY WORDS : social flexibility, Rhabdomys, striped mouse, ecological determinants of sociality.
INTRODUCTION
where groups are even larger and more complex (SCHRA-
DIN & PILLAY, 2004). In the Succulent Karoo, groups nor-
The striped mouse (Rhabdomys pumilio) is a medium
mally consist of one breeding male, up to four breeding
sized (adult body weight 30-70g) muroid rodent, which is
females and their non-reproducing adult offspring of both
active mainly during mornings and afternoons (KRUG,
sexes which remain in their natal territory (SCHRADIN &
2002; SCHRADIN & PILLAY, 2004). It is widely distributed
PILLAY, 2004). Males are permanently associated with
in southern Africa, inhabiting many different habitats,
groups of breeding females and participate in parental
such as grasslands, marshes, forests, semi-deserts and
care (SCHRADIN & PILLAY, 2003). However, whereas mice
deserts (KINGDON, 1974). The social organization of the
of one group sleep in the same nest, have the same group
striped mouse differs dramatically in correlation with the
territory and interact highly amicably with one another,
habitat it occupies. In moist grasslands, both sexes have
they forage alone (SCHRADIN, published online May 2005,
territories that overlap with territories of several individu-
DOI : 10.1007/s10164-005-0158-2) and react highly
als of the opposite but not the same sex (BROOKS, 1974;
aggressive towards mice from other groups (S
C
CHRADIN,
HOATE, 1972 ; PERRIN, 1980a; SCHRADIN & PILLAY, 2005
2004; S
b; W
CHRADIN & PILLAY, 2004). These differences in
ILLAN, 1982; WILLAN & MEESTER, 1989; WIRMING-
social organization between striped mice from the xeric
HAUS & PERRIN, 1993), females raise their young alone,
and associations between males and females are only for
areas and the moist grasslands lead to the question of
mating (B
whether there is only one single species, R. pumilio (WIL-
ROOKS, 1974; WILLAN, 1982). In sum, the
striped mouse in the grasslands is a solitary species
SON & REEDER, 1993). In fact, there appears outbreeding
(S
depression between different populations, which also
CHRADIN & PILLAY, 2005 b). In contrast, studies con-
ducted in xeric habitats indicate that the striped mouse is
show assortative mate choice decisions in captivity, i.e.
a social species here, e.g. in the Kalahari (NEL, 1975; own
females prefer males of their own population (PILLAY,
observ.). A detailed study revealed that the striped mouse
2000a; PILLAY, 2000b). An allozyme study of 23 different
is social in the Namib desert, with groups consisting of
populations revealed significant differences between pop-
one breeding female, her offspring that sometimes remain
ulations, with genetic distance being correlated with geo-
within the maternal territory even after reaching adult-
graphical distance (MAHIDA et al., 1999). Whereas this
hood, and sometimes one adult breeding male that is per-
study suggested the existence of different subspecies, no
manently associated with one female and her offspring,
evidence for different species was found. However, recent
(KRUG, 2002). This is similar to the situation in the Suc-
studies using mitochondrial DNA proposes the existence
culent Karoo, a desert in the north west of South Africa,
of two different species, with R. pumilio representing the

78
Carsten Schradin
social species living in the xeric deserts and semi-deserts,
other newly-emerged plant material and insects are
and R. dilectus representing the closely related solitary
important protein sources. Since protein is essential for
sister species living in the mesic grasslands (RAMBAU &
the onset and maintenance of reproduction (PERRIN,
ROBINSON, 2003). These two branches have separated less
1980a), breeding occurs during spring and lasts for three
than three millions years ago and further studies will have
months (SCHRADIN & PILLAY, 2005a). Striped mice reach
to test whether the new species R. dilectus will be recog-
sexual maturity at two months of age (BROOKS, 1982),
nized (RAMBAU & ROBINSON, 2003).
such that the first pups born during a particular breeding
season could reproduce during the season of their birth.
To what extent can genetic differences between social
However, this would be at the end of the breeding season,
and solitary striped mouse populations explain the
with already declining food abundance, and offspring sur-
observed social differences? An answer to this question
vival may be compromised then. Also, investing energy
is not apparent, but males from the solitary populations
in the grasslands show highly-developed paternal care in
into reproduction would reduce the survival probability of
captivity (S
young parents, since energy could alternatively be
CHRADIN & PILLAY, 2003), for which no evi-
dence exists from the field (S
invested in somatic development or stored as fat to buffer
CHRADIN & PILLAY, 2005
b; W
the effects of poor food supply during summer. (S
ILLAN, 1982). Also, there is no difference in pater-
CHRA-
nal response in captivity between males from the Succu-
DIN & PILLAY, 2005 a). This could explain why adult off-
lent Karoo, where the striped mouse is highly social, and
spring stay at home and invest in personal survival until
the grasslands (SCHRADIN & PILLAY, 2003). In this paper
the next breeding season rather than into reproduction
I discuss how ecological differences between the Succu-
during the season of their birth.
lent Karoo and the grasslands can explain the extreme
differences in social organization between striped mice
2) Whereas protein rich food occurs primarily during
from the two localities. Whether these ecological differ-
spring, overall food abundance is high throughout the
ences lead to genetic or to ontogenetic differences caus-
year. The dominant plant species, such as Zygophyllum
ing population typical social structures remains hereby
retrofractum shrubs and several succulents are available
unknown.
year round and provide a stable food supply. Neverthe-
less, the mice show considerable loss in body weight dur-
The Ecological Model
ing the hot, dry summer (SCHRADIN & PILLAY, 2005 a).
Plant growth starts again in autumn when the rain falls
again (COWLING et al., 1999), and food availability
Basic ecological differences
improves and reaches a peak in spring. It appears that
striped mice in the Succulent Karoo do not need large ter-
The main difference between the two habitats is the
ritories and can share their territories including resources
pattern of rainfall. The grasslands in the eastern part of
with up to 30 other group members (SCHRADIN & PILLAY,
South Africa are a mesic habitat, obtaining more than
2004), without experiencing severe competition for food.
1000 mm of rainfall per annum, which occurs mainly dur-
ing summer (ACOCKS, 1988). In contrast, the Succulent
3) The patchily distributed plant cover makes it possi-
Karoo is an arid habitat, situated in a winter rainfall
ble for mice to sun bask. In the morning and afternoon,
region, and receives only 50-400 mm rain per annum
the mice of one group sit together in front of their nest,
(ACOCKS, 1988; COWLING et al., 1999) and 160 mm at my
which is typically situated in a large Zygophyllum bush,
field site. Differences in rainfall pattern lead to dramatic
and warm themselves up in the sun (SCHRADIN & PILLAY,
difference in plant cover. In grasslands, the entire area is
2004). The time when the sun starts shining on their nest
covered by a sea of grasses and herbs, whereas shrubs are
has a significant effect on the initiation of activity
the dominant growth form in the Succulent Karoo, with in
(unpubl. data; KRUG, 2002). Sun basking might work as a
between open areas inhabited by succulents and in spring
method of energy saving and as thus leads to reduced
wildflowers (COWLING et al., 1999). The Succulent Karoo
demand of food intake, reduced foraging activity and as
is rich in endemic plant species (COWLING et al., 1999)
thus small territory size.
and regarded as one of 20 global biodiversity hotspots
(MYERS et al., 2000). In the following sections, the eco-
4) Striped mice preferably nest inside dense and thorny
logical consequences of these differences in rainfall and
Zygophyllum shrubs (S
by this vegetation are discussed. The ecological model for
CHRADIN & PILLAY, 2004). How-
ever, the number of bushes of this species that are big
the Succulent Karoo is shown in Fig. 1a, for the grass-
lands in Fig. 1b. Different critical points are marked
enough for a striped mouse group is limited, and there is
within the figures, and their importance is outlined below
strong competition between striped mice and syntopic
in a chronically order for both habitats.
bush karoo rats (Otomys unisulcatus) for access to these
nesting sites (SCHRADIN, published online May 2005,
DOI : 10.1007/s10164-005-0158-2). As bush karoo rats
Succulent Karoo (Fig. 1a)
weigh more than double that of striped mice, they typi-
cally win all encounters. The limited number of nesting
1) Food abundance is high in spring after the winter
sites might also promote staying at home at a good nest-
rains. Wildflowers are particularly important food
ing site instead of leaving the natal nest and nesting alone
resources during spring (unpubl. data), and together with
at a suboptimal place.

Ecological determinants of sociality in the striped mouse
79
few rain in winter
Succulent Karoo
food
patchily distributed
cover (shrubs)
protein rich food
basic food
3
4
potential of sun basking
limited amount of good
(conserve energy)
nesting sites
short season (spring)
much
2
short breeding
small territories
season (3 months)
1
offspring delay
low mortality
reproduction until
next year
5
high population density
GROUPS
much rain in summer
Grasslands
food
grass cover over entire
area
protein rich food
basic food
3
4
no potential of sun basking
unlimited amount of good
nesting sites
long season (summer)
few, widely distributed
2
long breeding season
large territories
(7 months)
1
offspring start
high mortality
reproduction at once
5
low population density
SOLITARY
Fig. 1. ­ A model proposed for the connections between rainfall, protein content of food, food abundance and cover avail-
ability, and the social structure of the striped mouse in (a) the Succulent Karoo, and (b) the grasslands of southern Africa.
For detailed descriptions see text.

80
Carsten Schradin
5) Individuals that are present in early summer after the
territories to become even bigger, thereby increasing the
breeding season have a chance of 30% to survive until the
energetic costs of traveling.
next breeding season (SCHRADIN & PILLAY, 2005 a). This
3) In the grasslands, the vegetation covers the entire
survival probability is very high compared to that of
area. Thus, here it is impossible for mice to come out of
grasslands, where it is less than 3% (BROOKS, 1982). It
the cover and perform sun basking to reduce energetic
can be explained by the investment of energy into sur-
demands by passive warming up. Instead, mice have to
vival instead of reproduction (1), the overall good food
increase their body temperature by metabolic heat, for
supply (2), and benefits of group living. As striped mice
which they have to find more food, and thus need larger
in the Succulent Karoo forage alone, but share one nest,
territories to fulfill this need.
benefits of group living must be related to nest sharing :
1. Communal infant care (S
4) Nest sites in the grasslands are abundant, particu-
CHRADIN & PILLAY, 2004)
including paternal care (S
larly in areas of dense grass (own observ.).
CHRADIN & PILLAY, 2003) ; 2.
Thermoregulatory benefits through huddling during the
5) Survival probability of juveniles in the grasslands is
night (ANDREWS & BELKNAP, 1986; CONTRERAS, 1984;
only about two months and annual survival probability is
temperatures can drop below zero in winter and spring
only 2.3% (BROOKS, 1974). Low food abundance (PER-
and are even in summer typically below 15 degrees C and
RIN, 1980b) and low ambient temperatures during winter
thus clearly below the thermal neutral point which might
probably lead to high mortality. Cold weather is a critical
be around 30 degrees, CANALS et al., 1998); 3. Increased
factor, as mice are solitary and thus do not benefit from
vigilance during the night towards potential predators
the advantage of huddling in a group. Early dispersal of
approaching the nest (which is built aboveground using
juveniles is probably another important factor influencing
hay). Support for points 2 and 3 is available through
survival probability, as dispersal into unknown habitat is
unpublished results of videotaping inside two natural
likely to reduce survival probability. Furthermore, young
nests, in which mice were sleeping closely huddled
adults do not invest in survival (accumulating resources
together, and quickly left the nest during the middle of the
such as fat to survive the winter), but immediately invest
night after a disturbance. In conclusion, the high survival
into reproduction. In the grasslands, mice of both sexes
probability leads to a high population density of over 50
start reproducing with a body weight below 30g, whereas
mice/ha at the start and 200 mice/ha after the breeding
in the Succulent Karoo offspring of both sexes remain at
season (SCHRADIN & PILLAY, 2005 a). The resulting habi-
home without reproducing, reaching body weights above
tat saturation might then force offspring to stay at home,
40g before reproducing (SCHRADIN & PILLAY, SUBM-B).
as no vacant territories are available for emigration.
The low population density, which results from the low
survival probability, means that territories are vacant into
Grasslands (Fig. 1b)
which offspring can immigrate when reaching adulthood.
1) The main protein-rich food sources in the grasslands
DISCUSSION
are grass seeds and insects. These food sources are availa-
ble throughout the entire spring and summer, and the
The model described here is no more than a plausible
breeding season stretches over this period of six to seven
explanation for the observed patterns of sociality in free-
months (BROOKS, 1974; PERRIN, 1980a; PERRIN et al.,
living striped mice. However, it shows associations
2001). Thus, mice born at the beginning of the breeding
between abiotic variables (level and season of rainfall),
season can potentially reproduce in the season of their
the biotic environment (plant cover, food availability and
birth for several months. In the grasslands, mice do not
protein content of food) and social organization in one
stay at home, but emigrate as juveniles and breeding
species (Rhabdomys pumilio). Other factors than the ones
occurs in young individuals weighing less than 30g (WIL-
described in the model might also have effects. One such
LAN, 1982); this does not occurs in the Succulent Karoo
factor could be predation pressure, which is very difficult
(SCHRADIN & PILLAY, 2005 b).
to estimate.
2) Whereas the green grasslands give the impression of
A model is only good if it can do two things : First it
high food abundance, this may not be the case. In contrast
has to describe the phenomena observed in nature. Above
to other syntopic rodents like vlei rats (Otomys irroratus),
I outlined how the model describes the patterns observed
the striped mouse does not feed on grass (PERRIN, 1980b).
in the Succulent Karoo and the grasslands. One test would
Its main food comprises seeds, berries and herbs (CURTIS
be to determine if it also can describe patterns observed in
& PERRIN, 1979; PERRIN, 1980b). These food sources are
the Namib (described by KRUG, 2002). The main social
patchily distributed and scarce. The low abundance of
difference between striped mice in the Succulent Karoo
food sources may explain why striped mice have 6 times
and the Namib is that groups are smaller in the Namib and
(females) or even 10 times (males) larger home ranges in
no cooperative breeding occurs. This is in accordance
the grasslands than in the Succulent Karoo (SCHRADIN &
with the ecological difference that availability of protein
PILLAY, SUBM-B). This difference is even greater when one
rich food is lower (1 in Fig. 1a), but more constant over
takes into account that home ranges are exclusive in the
time (2 in Fig. 1a), such that a clear breeding season is
grasslands, i.e. overlap only to a small extent with other
absent (1 in Fig. 1a; KRUG, 2002). As in the Succulent
individuals, but overlap with 5 to 30 other group members
Karoo, good possibility for basking exists (3 in Fig. 1a)
in the Succulent Karoo (SCHRADIN & PILLAY, 2005 b).
and does occur (KRUG, 2002), good nesting sites are
The low food abundance in the grasslands would make it
extremely limited (4 in Fig. 1a), and the habitat is satu-
more costly to live in groups, as sharing the territory with
rated in the Namib with sometimes extremely high popu-
other mice that use the same food resources would force
lation densities (5 in Fig. 1a), which can explain group

Ecological determinants of sociality in the striped mouse
81
living as a result of the lack of emigration possibilities
sberg) because of a lower energy need due to the higher
(KRUG, 2002). Thus, it seems that the model is also suita-
ambient temperature.
ble for the Namib, although the unique environmental
4. Providing super-optimal nesting sites should lead to
parameters for this habitat would have to be included.
increased sociality. The obvious option for this would be
The second prerequisite for the usefulness of a model is
to present nest boxes, but pilot studies with artificial nest
that it makes predictions that can be tested. Below I state
boxes were not successful (unpubl. data).
the predictions made by the model for both habitats, again
5. Increased survival probability by providing food in
referring to the important aspects pointed out in Fig. 1.
the field should lead to a higher degree of sociality, as
Hereby it is not expected that a single factor will "cause"
pointed out in point 3.
group living or a solitary lifestyle, but that it is the combi-
nation of factors and predictions that are important, as
illustrated in Fig. 1.
CONCLUSIONS
Predictions in Succulent Karoo
The striped mouse is a convenient model for studying
the environmental determinants of sociality. Hereby, it is
1. Offspring born at the end of the breeding season
not of crucial importance, whether the striped mouse is
have a lower survival probability. Heavier mice (i.e.
one or two closely related sister species (see Introduc-
greater body fat) of the same age group have a higher sur-
tion). In this paper, the intention was to point out the eco-
vival probability.
logical differences that explain differences in sociality.
2. Territory size should increase during seasons with
Whether these ecological differences acted as selection
poor food supply. In areas with a lower food supply,
pressures in the past that caused genetic differences and
smaller groups are expected.
finally speciation, or if these ecological differences lead
3. The body temperature of mice increases significantly
to behavioural differences via an ontogenetic pathway,
when basking (to be measured by implants) and energy
remains a topic for further highly interesting and impor-
expenditure is lower during periods of good sun basking
tant research. Further studies should experimentally test
opportunities (summer compared to winter; to be meas-
the predictions outlined in this paper to test the model
ured by doubled labeled water).
described.
4. Removal of bush karoo rats should lead to striped
mice occupying their abandoned nest sites.
ACKNOWLEDGEMENTS
5. Low population density leads to a more solitary life-
style. Removal of groups should lead to adult offspring of
I would like to thank Northern Cape Department of Agricul-
other groups leaving their group and taking over these
ture, Land Reform, Environment & Conservation and KwaZulu
Natal Wildlife for their assistance and Mr Klaas van Zyl, Enrico
vacant territories.
Oosthuysen and their staff at Goegap Nature Reserve as well as
Many of these predictions might be testable in future.
Charl Brummer and his staff in Kamberg Nature Reserve for
The Succulent Karoo is known to have a low, but highly
their support during my field studies. I am grateful to important
predictable rainfall pattern (COWLING et al., 1999) and as
comments by N. Pillay that improved the manuscript signifi-
such predictable food abundance for the striped mouse.
cantly. This study was supported by the Deutsche Forschungsge-
However, the Succulent Karoo is currently (2003) experi-
meinschaft (German Science Foundation DFG), the Swiss
encing the severest drought since many years. Thus,
National Science Foundation, the Schweizerische Gesellschaft
while my model indicates high survival probability and
für Naturwissenschaften, the Fonds zur Förderung des akadem-
ischen Nachwuchses (FAN) of the University of Zurich and the
high population density in the Succulent Karoo, these are
University of the Witwatersrand.
unlikely in 2003. This dramatic drought will thus offer the
opportunity to study the effects of a reduced population
density on the social organization of the striped mouse. It
REFERENCES
will be possible to test which individuals survived (pre-
diction 1), and it will be interesting to see if there are
ACOCKS, J.P.H. (1988). Veld Types of South Africa, 3rd edition.
changes to the social structure (i.e. are the mice solitary,
Botanical Research Institute, Pretoria.
A
prediction 2 and 5). At the same time, bush karoo rats
NDREWS, R.V. & R.W. BELKNAP (1986). Bioenergetic benefits
of huddling by deer mice (Peromyscus maniculatus). Comp.
became nearly locally extinct at my field site, making it
Biochem. Physiol., 85 : 775-775.
possible to test prediction 4.
BROOKS, P.M. (1974). The Ecology of the Four-Striped Field
Mouse, Rhabdomys pumilio (Sparrman, 1984), with particu-
Predictions in grasslands
lar Reference to a Population on the Van Riebbeeck Nature
Reserve, Pretoria
. PhD thesis, University of Pretoria, South
1. Mice born at the end of the breeding season should
Africa.
stay longer in their natal territory.
BROOKS, P.M. (1982). Aspects of the reproduction, growth and
2. Experimental increase of food availability should
development of the four-striped mouse, Rhabdomys pumilio
lead to larger population density (shown by PERRIN &
(Sparrman, 1784). Mammalia, 46 : 53-64.
JOHNSON, 1999), smaller home ranges, increased survival
CANALS, M., M. ROSEMANN, F.F. NOVOA & F. BOZINOVIC (1998).
probability, and eventually to habitat saturation and
Modulating factors of the energetic effectiveness of huddling
finally to a higher level of sociality.
in small mammals. Acta Theriologica, 43 : 337-348.
CHOATE, T.S. (1972). Behavioural studies on some Rhodesian
3. Striped mice in warmer grassland habitats (e.g. in the
rodents. Zoologica Africana, 7 : 103-118.
area of Pretoria) should have smaller home ranges than
CONTRERAS, L.C. 1984. Bioenergetics of huddling : test of a
mice in colder grassland habitats (e.g. high in the Draken-
psycho-physiological hypothesis. J. Mamm., 65 : 256-262.

82
Carsten Schradin
COWLING, R.M., K.J. ESLER & P.W. RUNDEL (1999). Namaqua-
Muridae) : interpopulation breeding studies. Mammalia, 64 :
land, South Africa - an overview of a unique winter-rainfall
461-470.
desert ecosystem. Plant Ecology, 142 : 3-21.
RAMBAU, R.V. & T.J. ROBINSON (2003). Molecular genetics of
CURTIS, B.A. & M.R. PERRIN (1979). Food preferences of the
Rhabdomys pumilio subspecies boundaries : mtDNA phylo-
vlei rat (Otomys irroratus) and the four-striped mouse
geography and karyotypic analysis by fluorescence in situa
(Rhabdomys pumilio). S. Afr. J. Zool., 14 : 224-229.
hybridization (FISH). Molec. Phyl. Evol., 28 : 564-575.
KINGDON, J. (1974). East African Mammals. Academic Press,
SCHRADIN, C. (2004). Territorial defense in a group living soli-
London.
tary forager : who, where against whom? Behav. Ecol.
KRUG, C.B. (2002). The behavioural ecology of Rhabdomys
Sociobiol., 55 (5) : 439-446.
pumilio in the Namib Desert. PhD thesis, Rheinische Frie-
SCHRADIN, C. (published online May 2005, DOI : 10.1007/
drich-Wilhelms-Universität Bonn, Germany.
s10164-005-0158-2). Whole day follows of the striped
M
mouse. J. Ethol; 24.
AHIDA, H., G.K. CAMPBELL & P.J. TAYLOR (1999). Genetic var-
iation in Rhabdomys pumilio (Sparrman, 1784) : an alloz-
SCHRADIN, C. & N. PILLAY (2003). Paternal care in the social
yme study. S. Afr. J. Zool., 34 : 91-101.
and diurnal striped mouse (Rhabdomys pumilio) : laboratory
M
and field evidence. J. Comp. Psychol., 117 : 317-324.
YERS, N., R.A. MITTERMEIER, C.G. MITTERMEIER, G.A.B.D.
F
SCHRADIN, C. & N. PILLAY (2004). The striped mouse (Rhab-
ONSECA & J. KENT (2000). Biodiversity hotspots for conser-
vation priorities. Nature, 403 : 853-858.
domys pumilio) from the succulent karoo of South Africa : A
territorial group living solitary forager with communal
NEL, J.A.J. (1975). Aspects of the social ethology of some Kala-
hari Rodents. Z. Tierpsychol., 37 : 322-331.
breeding and helpers at the nest. J. Comp. Psychol., 118 (1) :
37-47.
PERRIN, M.R. (1980a). The breeding strategies of two co-exist-
SCHRADIN, C. & N. PILLAY (2005 a). Demography of the striped
ing rodents, Rhabdomys pumilio (Sparrman, 1784) and Oto-
mouse (Rhabdomys pumilio) in the Succulent Karoo. Mam-
mys irroratus (Brants, 1827). Acta Oecol., 1 : 383-410.
mal. Biol., 70 : 84-92.
PERRIN, M.R. (1980b). The feeding habits of two co-existing
SCHRADIN, C. & N. PILLAY (2005 b). Intraspecific variation in
rodents, Rhabdomys pumilio (Sparrman, 1784) and Otomys
the spatial and social organization of the African striped
irroratus (Brants, 1827), in relation to rainfall and reproduc-
mouse. J. Mammal., 86 : 99-107.
tion. Acta Oecol., 1 : 71-89.
WILLAN, B.R. (1982). Social ecology of Otomys irroratus, Rhab-
PERRIN, M.R., C. ERCOLI & E.R. DEMPSTER (2001). The role of
domys pumilio and Mastomys natalensis. PhD thesis, Uni-
agonistic behaviour in the population of two syntopic Afri-
versity of Natal.
can grassland rodents, the striped mouse Rhabdomys pumilio
WILLAN, K. & J. MEESTER (1989). Life-history styles of southern
(Sparrman, 1784) and the multimammate mouse Mastomys
African Mastomys natalensis, Otomys irroratus and Rhab-
natalensis (A. Smith, 1834) (Mammalia Rodentia). Trop.
domys pumilio (Mammalia, Rodentia). In : M.N. BRUTON
Zool., 14 : 7-29.
(ed), Alternative Life-History Styles of Animals, Kluwer
PERRIN, M.R. & S.J. JOHNSON (1999). The effect of supplemen-
Academic Publishers, Dordrecht, pp. 421-439.
tal food and cover availability on a population of the striped
WILSON, D.E. & D.M. REEDER (1993). Mammal Species of the
mouse. S. Afr. J. Wildl. Res., 29 : 15-18.
World. A Taxonomic and Geographic Reference, 2nd edition.
PILLAY, N. (2000a). Female mate preference and reproductive
Smithsonian Institution Press, Wahingtons D.C.
isolation in populations of the striped mouse Rhabdomys
WIRMINGHAUS, J.O. & M.R. PERRIN (1993). Seasonal-changes in
pumilio. Behaviour, 137 : 1431-1441.
density, demography and body-composition of small mam-
PILLAY, N. (2000b). Reproductive isolation in three populations
mals in a Southern temperate forest. Journal of Zoology,
of the striped mouse Rhabdomys pumilio (Rodentia,
229 : 303-318.

Belg. J. Zool., 135 (supplement) : 83-89
December 2005
Movements and spatial patterns of Mastomys erythroleu-
cus
in maize cropping systems in the Kenyan Rift Valley

Christopher Odhiambo1, Nicholas Otienoh Oguge2 and Herwig Leirs3,4
1 Mpala Research Centre, PO Box 555, Nanyuki, Kenya
2 Earth Watch Institute, Samburu Conservation Research Initiative, PO Box 60033, Nairobi, Kenya
3 University of Antwerp, Department of Biology, Groenenborgelaan 171, B-2020 Antwerpen, Belgium
4 Danish Pest Infestation Laboratory, Danish Institute of Agricultural Sciences, Department of Integrated Pest Management,
Skovbrynet 14, DK-2800 Kongens Lyngby, Denmark
Corresponding author : Christopher Odhiambo, e-mail : codhiambo@mpala.org
ABSTRACT. We studied movements and spatial patterns of Mastomys erythroleucus in four permanent capture-
mark-recapture grids in maize cropping systems in the Kenyan Rift Valley. Mean daily movements were affected by
maize crop phenology. There was also a significant interaction effect between sex and crop phenology on daily
movements. Ranging distances varied by sex with males ranging further than females between successive captures.
The majority of marked individuals remained within 20 m of their point of first capture for the three consecutive
days in each trapping period. Rodent captures were highly clustered around the grid center or the periphery in three
grids but were distributed randomly in the fourth grid. The clustered dispersions could suggest habitat preferences
by M. erythroleucus within these fields and such areas could provide targets for ecologically based management.
Changes in movement patterns in response to environmental factors allow for colonization and use of available
resources in cultivated areas. Rodent control measures may focus on, among other approaches, limiting dispersal in
farms.
KEY WORDS : Movements; spatial patterns; Mastomys erythroleucus; maize cropping systems.
INTRODUCTION
community structure and its biological diversity (HOLT,
1997).
In Kenya, maize is a subsistence crop grown in small-
Mastomys erythroleucus (Temminck, 1853), a multi-
holdings of 0.5 to 20 hectares under partial mechanization
mammate mouse, is an important murid pest of maize
(GOVERNMENT OF KENYA, 1997). Such land tenure sys-
crops in the Kenyan Rift Valley (ODHIAMBO & OGUGE,
tems are often under different management practices that
2003). Earlier reports from the area mentioned it as M.
lead to the formation of habitat patches across agricultural
natalensis (TAYLOR, 1968; TAYLOR & GREEN, 1978), but
landscapes (MAKUNDI et al., 1999). In these patches, the
that species has a different chromosome number.
qualities of resources including levels of their productiv-
Although documentation of current rodent damages is not
ity are frequently ephemeral and follow an annual crop
available, earlier reports have indicated 20% loss of
cycle. Species respond to these changes in resource levels
maize and 34-100% loss of wheat and barley during the
over space and time by adopting different life history
1951 and 1962 rodent outbreaks in western Kenya (TAY-
strategies, especially in their choice of diet, preferred hab-
LOR, 1968). In the eastern Africa region, the economic
itat and mating systems (FLEMING, 1979).
importance of this species has been reported from the
farmlands in Ethiopia (BEKELE & LEIRS, 1997) while the
Habitat heterogeneity created from resource patches
congeneric M. natalensis is the most important pest of
has profound influence on the intensity of the effects of
maize in Tanzania (MWANJABE et al., 2002). Different
density-dependent factors (such as competition, diseases
strategies for success of pest populations have implica-
and predation) on various demographic parameters
tions for control. One of the fundamental goals in ecology
(GOLDWASSER et al., 1994). Rodents in African fields are
is to understand the distribution and abundance of organ-
highly motile animals (LEIRS et al., 1997) and through
isms and to use this knowledge for the management of
dispersal, may connect between patches to reduce the lev-
populations in a variety of natural and managed ecosys-
els of competition for limited resources (STENSETH & LID-
tems (GUTIERREZ, 1994). Our study assessed for (i) effects
ICKER, 1992; LAMBIN, 1994; MANSON et al., 1999). Such
of crop phenology on movement patterns of M. erythro-
movements are reflected in their demographic traits such
leucus; and (ii) its distribution patterns in maize crop
as mortality, recruitment (LARSEN & BOUTIN, 1994), den-
fields. We tested the following hypotheses; first, that
sities, distribution, persistence, extinctions and colonisa-
movement patterns in this species remain unaffected
tions (DIFFENDORFER et al., 1995; FERRERAS 2001; THO-
under changing crop phenology in cultivated areas. Sec-
MAS et al., 2001; BRITO & FERNANDEZ, 2002; JOHNSON et
ondly, that individuals in this population as in the major-
al., 2002; KIE et al., 2002). This may ultimately influence
ity of species are distributed at random.

84
Christopher Odhiambo, Nicholas Otienoh Oguge & Herwig Leirs
MATERIAL AND METHODS
Study site
The study was conducted in Rongai Division, Nakuru
District, in the Kenyan Rift Valley (35o 28' - 35o 36' E and
0o 13' - 1o 10' S), at altitudes of between 1520-2400m
above sea level (Fig. 1). The climate in this area falls
between semi-arid (annual rainfall less than 760 mm) in
the lower areas and dry sub-humid (annual rainfall of
1270mm) regions in the higher altitudes. Mid-day tem-
perature ranges between approx. 24o-30oC. The area is
important for maize production mainly for subsistence
but also as a cash crop. The region contributes substan-
tially to the country's production of maize, beans and
potatoes (GOVERNMENT OF KENYA, 1997).
In this study area, the original vegetation of moist for-
est has been replaced largely by a mosaic of cropped and
fallow areas with exotic tree species Grevillea robusta
and Cyprus spp. on the high slopes. In the sub-humid
region, the current vegetation has resulted from repeated
Fig. 1. A map of Kenya showin
Fig.
g the location of Ron
1. ­ A map of Kenya showing
gai Division and the st
the locatio
udy grids.
n of Rongai Divi-
burning, grazing and cultivation except along moist
sion and the study grids.
ravines. The most common ravine tree species were
Podocarpus latifolius, Syzygium guineense, Ficus stuhl-
mannii, Albizia coriaria and Acacia nilotica. In the inter-
Data analyses
mediate zone, vegetation is dominated by species of Aca-
Movements during the trapping periods were deter-
cia xanthophloea, Acacia seyal, Acacia nubica and
mined from the secondary capture histories. The XY
Acacia senegal. The common grasses are Themeda trian-
reduced capture histories were used in the data input file
dra and Themeda diplandra. In the driest part of the Divi-
of the program CAPTURE (REXSTAD & BURNHAM, 2002).
sion, the study was carried out near the River Rongai,
Range lengths were determined by calculating the linear
where the riverine vegetation is dominated by shrubs of
maximum distance between capture stations in consecu-
Lantana camara. Other common species included scat-
tive months (DELANY & MONRO, 1985), using the for-
tered Acacia species, succulents such as Euphorbia can-
mula;
delabrum, Euphorbia tirucalli and Aloe ballyi. The domi-
nant grasses in fallow areas there were Panicum
d= (x -x )2 + (y -y )2............................(1)
maximum, Cynodon dactylon and Themeda triandra.
2
1
1
1
Where
d = linear maximum distance estimate (m)
Sampling methods
x = X-co-ordinate of the initial trap station
1
x = X-co-ordinate of the final trap station
2
Rodents were live-trapped using Sherman's LFA traps
y = Y-co-ordinate of the initial trap station
1
(Sherman Traps Inc., Tallahassee, Florida, USA) on four
y = Y- co-ordinate of the final trap station
permanent one-hectare grids coded (i) Mugo, (ii) Beth,
2
(iii) Kurt, and (iv) Moto. The grids were established in
Daily movement and range length data were first stand-
April 2000 along an altitudinal gradient from a sub-humid
ardized using log (x + 1) transformation (ROHLF &
(Mugo grid), over intermediate ares (Beth and Kurt grids)
SOKAL, 1984); then General Linear Model (GLM) analy-
to semi arid (Moto grid) conditions (Fig. 1). Each grid
sis of variance in program SYSTAT Version 9.0 was used
consisted of trapping stations laid at 10 m within and
to test the effects of sex, maize crop phenology and grid
between rows giving a total of 100 stations and marked
location on net movements during the trapping period.
using white painted bricks. One trap was placed at each
We did not make appreciable rodent recaptures in year
station with traps being opened for three consecutive days
2000; therefore, these estimates and analysis were based
and nights every 28 days between May 2000 and Decem-
on data obtained between January and December 2001.
ber 2001. Trapping in Kurt grid commenced in April
Distribution of individuals within maize fields was
2001. Fried coconut cubes mixed with peanut butter and
mapped out using frequency of primary captures in the
corn oil were used as bait.
100 trapping stations per grid. Only the first capture of
each individual in a month was used as the point of refer-
Animals were sampled using the Capture-Mark-Recap-
ence to eliminate biases. To enable further analysis, the
ture (CMR) technique. Captured animals were identified
100 trapping locations were divided into three distinct
to species level, weighed, sexed and their trapping station
concentric zones : an outer periphery with 36 trapping
and other general remarks were recorded. Each animal
stations, middle zone with 28, and a central one with 36
was individually marked by toe clipping. The animals
for use in analysis of spatial patterns. Thereafter, a coeffi-
were then released at the point of capture and traps re-
cient of dispersion (CD) was calculated for each of the
baited.
three zones (KREBS, 1989). The significance of the depar-

Movements and spatial patterns of Mastomys erythroleucus
85
ture from randomness was assessed statistically by
movements (F = 2.42, P = 0.016, n = 204) as ranging dis-
computing :
tances varied between planting/seedling (23.6 ± 3.1 m),
harvesting (20.1 ± 1.8 m), vegetative (17.6 ± 2.3m), rip-
t = s²/X-1.0
ening (16.5 ± 4.7 m) and fallow (14.2 ± 2.1 m) stages,
2/(n-1)......................................(2)
respectively. There was also a significant interaction
effect (F = 2.76, P = 0.012, n = 204) between sex and
where s2 is the sample variance, X is the sample mean
crop phenology on these movements during the capture
capture frequencies and n is the sample size, respectively
periods. A high proportion (Mugo grid 60.43%, Beth
(CLAPHAM, 1936). The calculated t was compared to the
critical values on mathematical tables for n-1 degrees of
79.27%, Kurt grid 54.72% and Moto grid 60.00%) of ani-
freedom (R
mals moved shorter distances (<20m) between successive
OHLF & SOKAL, 1984).
capture sessions (Fig. 3).
RESULTS
Range length was significant between sexes (F =
18.972, P < 0.001, n = 136), with males (44.19 ± 3.8 m)
Movements and range lengths
ranging further than females (27.54 ± 2.4 m) (Fig. 4).
Daily movement in M. erythroleucus was similar (F =
Ranging distances were similar across the grids (F =
1.39, P = 0.24, n = 204) between sexes across the four
1.302, P = 0.276, n = 136). No interactive effect was
grids (mean ± SEM : males= 19.54 ± 1.87 m; females=
detected between sex and grids (F = 1.231, P = 0.301, n =
17.76 ± 1.41 m) (Fig. 2). Maize phenology affected
136).
a) Mugo Grid
Male
b) Be
Fem
th Grid ale
Male
Female
80
80
Fallow
Seedling
Vegetative
Ripening
Harvest ing
Fallow
Seedling
Vegetative
Ripening
Harvesting
Fallow
Fallow
clearing
clearing
70
70
&
& ploughing
ploughing
60
60
50
) 50
)
m
t
s
(
s

(
m
en
40
40
e
nt
m
o
v
em
o
ve
m
30
30
e
a
n
m
ean
M
M
20
20
10
10
0
0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Capture period (month)
Capture period (month)
M
c) K a
u le
F
rt Grid emale
Male
d) Moto Grid Female
80
80
Fallow
Seedling
Vegetat ive
Ripening
Harvesting
Fal ow
Seedling
Vegetative
Ripening
Harvesting
Fallow
Fal ow
clearing
70
70
clearing
&
& ploughing
ploughing
60
60
50
50
)
)
t
s
(
m
e
n
t
s
(
m
40
en 40
m
e
m
ov
ove
30
30
a
n
m
e
a
n
m
Me
M
20
20
10
10
0
0
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Capture period (month)
Capture period (month)
Fig. 2. ­ Mean (± SEM) daily movement patterns of male (black squares, full lines) and female (open circles, dashed lines) M.
erythroleucus
under different maize crop phenology in; a) Mugo, b) Beth, c) Kurt and d) Moto grids as estimated at 95% confi-
dence interval from XY reduced secondary Capture histories using the program CAPTURE. Data collected from January to
December 2001. The different stages of the maize crop fields are indicated.

86
Christopher Odhiambo, Nicholas Otienoh Oguge & Herwig Leirs
40
Male
Female
40
Male
Female
a) Mugo Grid
b) Beth Grid
35
35
t
t
e
n
30
30
e
m
v
v
e
men
25
25
f
mo
o
f
mo
20
20
) o
)
(%
e (%
15
15
g
e
t
a
t
a
g
10
10
ercen
ercen
P
P
5
5
0
0
0-10
11-19
20-29
30-39
40-49
50-59
60-69
70>
0-10
11-19 20-29 30-39 40-49 50-59 60-69
70>
Distance class (m)
Distance class (m)
40
Male
Female
40
Male
Female
c) Kurt Grid
d) Moto Grid
35
35
t
n
30
30
e
me
e
nt
v
m
25
25
o
ve
f
mo
m
20
20
)
o
)
of
%
e

(
%
15
15
g
t
a
n
n
t
a
g
e
(
10
10
ce
ce
er
P
er
P
5
5
0
0
0-10
11-19
20-29
30-39
40-49
50-59
60-69
70>
0-10
11 -19
20-29
30 -39
40-49
5 0-5 9
60-69
70>
Dist ance class (m)
Distance class (m)
Fig. 3. ­ Class frequencies of daily movements by adult male (dark) and female (light) M. erythroleucus in a) Mugo, b) Beth,
c) Kurt and d) Moto grids between successive capture positions following release. Only animals caught within the three
sampling sessions are included.
only clustered peripherally (CD = 2.3, t = 5.3, P < 0.01, n
80
= 36) while it was random at Moto. In Mugo grid, 68.1%
)
m 70
of the captures were made from the whole of A-line and
(
th 60
along the 5th, 6th, 7th and 8th vertical lines. In the Beth
g
n 50
grid, 75.1% of captures were made from the edges of the
e
e l 40
farm on the J-line and along the 2nd, 3rd, 9th and 10th verti-
g
n 30
a
cal lines. In Kurt grid captures were mainly from one
r 20
edge of the farm thus on line-1 and constituting 43.6% of
ean 10
the captures. In Moto grid, no particular trapping stations
M
0
could be associated with significant proportions of cap-
Mug
M o
ug
Be
B t
e h
Ku
K rt
u
Mo
M t
o o
tures.
Gr
Gi
r d
i
DISCUSSION
Fig. 4. ­ Mean range lengths for adult male (dark) and
female (light) M. erythroleucus between successive capture
Our study shows that male and female Mastomys eryth-
positions following release.
roleucus cover similar distances in their daily activities,
Distribution of captures within grids
i.e. 19. 5 (± 1.9 m) and 17.8 (± 1.4 m), respectively (Fig.
2). However, the long daily movements made during
Patterns in rodent distribution were not entirely con-
seedling and harvesting stages of maize crop may be due
sistent within the maize fields (Fig. 5). Rodent captures
to habitat alterations caused by weeding and trampling on
exhibited both peripheral (CD = 1.7, t = 3.1, P < 0.01, n =
vegetation, respectively, which could have led to reduced
36) and central (CD = 1.8, t = 3.6, P <0.01, n = 36) clus-
ground cover forcing rodents to seek new habitat patches.
tering at Mugo grid. Similar peripheral (CD = 2.5, t = 6.5,
Manual harvesting could have further reduced food
P < 0.01) and central (CD = 1.7, t = 3.1, P < 0.01) cluster-
resources in the farms. The short movements during the
ing were noted at Beth grid. Distribution in Kurt grid was
vegetative, ripening and fallow stages of the crop (Fig. 2),

Movements and spatial patterns of Mastomys erythroleucus
87
coupled with the majority of captures (60.43%, 79.27%,
ments. Increases in food supplies lead to small home
54.72% and 60.00% in Mugo, Beth, Kurt and Moto grids,
ranges, declined space use and immigration into the area,
respectively) less than 20m from point of original capture
while the reverse is observed when food supplies are
(Fig. 3) suggests that M. erythroleucus settled temporarily
reduced (BOUTIN, 1984). The reasons for the interaction
within the perimeter of the resource rich habitat patches
effect between sex and crop phenology, however, remain
of the maize farms. Earlier experimental work has shown
unclear in this study.
an inverse relationship between food and rodent move-
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
A
i i i
i i i i ii
A
i
i
i
i i i
B
i
i
i i i i i i
B
i i i
i
i i i
C
i
i i i i i i i i
C
i i i
i i
D
i
i
D
i
i
i
i
i
E
i
i i i i i
E
i i i
i
i i
F
i
i
i i
i
F
i
i i
G
i
i
i i
i i
G
i i i i
i
i i i i
H
i i
i i i i i i
H
i i i i
i i i i
I
i
i i i i i i
I
i i i i i i i i i i
J
i
i i
i
i i i
J
i ii i i ii ii i
(a) Mugo grid
(b) Beth grid
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
i
A
i i i
i
i i
A
i
i
i
i
i
i i i
i
B
i
i i
i
i i i i
B
i
i
i i
i
C
i i i
i
i
C
i
i
i
i
i
i
i
i
D
i
i i i i i
i
D
i
i
i
E
i
i
i i i
E
i
i i i
i
i
F i i i
i
i
i
i
F
i
i
i i
i
i
i
G
i i
i i i i i
G
i
i
i i
i
i i
i
H
i
i
i
i i
H
i
i i
i
i
i
I
i
i i i i i i i i
I
i
i
i i i
i i
J
i i i i i i i i i i
J
i
i
i i i i i i i
(c) Kurt grid
(d) Moto grid
Fig. 5. ­ Distributions of total capture frequencies per trapping station for M. erythroleucus in a) Mugo, b) Beth, c) Kurt and
d) Moto grids between January and December 2001.
Although ranging distances were similar between
roleucus up to 50 m from the edges of newly sown vege-
grids, males ranged longer distances than females (44.2 ±
tation-bare maize fields. The clustering of rodent captures
3.8 m and 27.5 ± 2.4 m), respectively (Fig. 4). This would
at certain trapping positions (Fig. 5), suggests that some
suggest larger home range in males whose larger body
parts of the habitat were utilised more or a possibility of
size (ODHIAMBO, 2003) suggests a polygynous social sys-
some social attraction that enhanced presence of other
tem (OSTFELD, 1986). Similarly, a study by MARTIN et al.
individuals. Crops in these loci, for example, at the
(1989) in Kitale area in western Kenya captured M. erth-
peripheral and central zones of the farm are likely to suf-

88
Christopher Odhiambo, Nicholas Otienoh Oguge & Herwig Leirs
fer increased risk to depredation from rodents leading to
GOLDWASSER, L., J. COOK & E.D. SILVERMAN (1994). The
the patchy nature of damage. Similar results indicating
effects of variability on Metapopulation dynamics and rates
site fidelity and edge effects have also been reported from
of invasion. Ecology, 75(1) : 40-47.
the studies of Arvicanthis niloticus in Kenyan grassland
GOVERNMENT OF KENYA (1997). Economic Review. Government
(DELANY & MONRO, 1985), and for other small rodents in
Press. Nairobi. Kenya.
forest and adjoining farmland habitats (MANSON et al.,
GUTIERREZ, P.A. (1994). A Physiological Based Tritrophic Per-
1999). Conversely, random distribution of captures in
spective in Bottom-Up down regulation. Ecology, 75 : 2227-
Moto may be attributed to pure and uniform stands of
2242.
fodder grass surrounding the trapping stations. Other
HOLT, R.D. (1997). From meta-population dynamics to commu-
nity structure : some consequences of spatial heterogeneity.
studies have reported the influence of microhabitat on
In : HANSKI I. and M. E. GILPIN (eds). Metapopulation
food, abundance and distribution of Mastomys in Kenyan
biology : ecology, genetics and evolution. Academic Press,
grasslands (MARTIN & DICKINSON, 1986; OGUGE, 1995).
San Diego, California, USA : 149-164.
From our study, we can conclude that the crop phenol-
JOHNSON, R., J.W.H. FERGUSON, A.S. VAN JAARSVELD, G.N.
ogy strongly influenced movements of M. erythroleucus.
BRONNER & C.T. CHIMIMBA (2002). Delayed responses of
Males generally made wider field excursions than
small mammal assemblages subject to afforestation-induced
females, but once settled most rodents in the farm pre-
grassland fragmentation. Journal of Mammalogy, 83 (1) :
290-300.
ferred to temporarily stay near their home ranges.
Changes in movement patterns in response to environ-
KIE, G.J., R.T. BOWYER, M.C. NICHOLSON, B.B. BOROSKI & E.R.
LOFT (2002). Landscape heterogeneity at differing scales :
mental factors allow for colonisation and use of emerging
effects on spatial distribution of Mule Deer. Ecology, 83(2) :
resources in cultivated areas. Control measures may focus
530-544.
on, among other approaches, limiting dispersal during
KREBS, C.J. (1989). Ecological Methodology. Hapre & Row,
periods of high precipitation. The clustered dispersions
Publishers, Inc, New York, NY (664 pp).
would suggest habitat preferences by M. erythroleucus in
LAMBIN, X. (1994). Natal philopatry, competition for resources
these fields and this may provide targets for ecologically-
and inbreeding avoidance in Townsends' voles (Microtus
based rodent management.
townsendii). Ecology, 75(1) : 224-235.
LARSEN, K.W. & S. BOUTIN (1994). Movement, survival and set-
tlement of young red squirrels. Ecology, 75(1) : 214-223.
ACKNOWLEDGEMENTS
LEIRS, H., R.VERHAGEN, C.A. SABUNI, P.S. MWANJABE & W.N.
VERHEYEN (1997). Spatial dynamics of Mastomys natalensis
We would like to thank the Officers in the Ministry of Agri-
in a field-fallow mosaic in Tanzania. Belgian Journal of
culture especially Mr. Mbuba, Mr. Korir, Mr. Kimwele and Mr.
Zoology, 127 : (Suppl.) : 29-38.
Mugo for their cooperation and provision of logistic support. We
MANSON, H.R., R.S. OSTFELD & C.D. CANHAM (1999).
are also grateful to local farmers who allowed us to use their
Responses of a small mammal community to heterogeneity
farms for sampling our rodent data. This study was fully funded
along forest-old-field edges. Landscape Ecology, 14 : 355-
by the Rockefeller Foundation's Forum on Agricultural
367.
Resource Husbandry programme.
MAKUNDI, R., N. OGUGE & P. MWANJABE (1999). Rodent Pest
Management in East Africa - an ecological approach. In :
REFERENCES
SINGLETON, G., H. LEIRS, Z. ZHANG, and L. HINDS (eds). Eco-
logically-based Rodent Management. ACIAR Publications,
Canberra : 460-476.
BEKELE, A. & H. LEIRS (1997). Population ecology of rodents in
M
maize fields and grasslands in central Ethiopia. Belgian
ARTIN, G.H.G. & N.M. DICKINSON (1986). Small mammal
abundance in relation to microhabitat in a dry sub-humid
Journal of Zoology, 127 (Suppl.) : 39-48.
grassland in Kenya. African Journal of Ecology, 23 : 223-
BRITO, D. & F.A.S. FERNANDEZ (2002). Patch relative impor-
234.
tance to metapopulation viability : the neotropical marsupial
MARTIN, G.H.G., N.O. OGUGE & J.M. GATIMU (1989). Technical
Micoureus demerarae as a case study. Animal Conservation,
Report for the Development of Rodent Control Program in
5 : 45-51.
the Crop Protection Branch, Ministry of Agriculture, Kenya.
BOUTIN, S. (1984). Effect of late winter and food addition on
FAO/UNDP Project Ken/85/009.
numbers and movement of Snowshoe hare. Oecologia, 62 :
MWANJABE, P.S., F.B. SIRIMA & J. LUSINGU (2002). Crop losses
393-400.
due to outbreaks of Mastomys natalensis (Smith, 1834)
CLAPHAM, A.R. (1936). Over dispersion in grassland communi-
Muridae, Rodentia, in the Lindi region of Tanzania. Interna-
ties and the use of statistical methods in plant ecology. Jour-
tional Journal of Biodeterioration and Biodegradation, 49 :
nal of Ecology, 24 : 232-251.
133-137.
DELANY, M.J. & R.H. MONRO (1985). Movement and spatial
ODHIAMBO, C. (2003). Population dynamics and habitat use by
distribution of the Nile rat (Arvicanthis niloticus) in Kenya.
murid rodents in Maize cropping systems in Rongai Divi-
Journal of Tropical Ecology, 1 : 111-130.
sion, Nakuru District, Kenya. M.Sc Thesis, Kenyatta Uni-
DIFFENDORFER, E.J., M.S. GAINES & R.D. HOLT (1995). Habitat
versity, Nairobi.
fragmentation and movements of three species of small
ODHIAMBO, C. & N.O. OGUGE (2003). Patterns in rodent pest
mammals (Sigmodon, Microtus and Peromyscus). Ecology,
distribution in a maize cropping system in the Kenyan Rift
76(3) : 827-839.
Valley. In : SINGLETON, G., L.A. HINDS, C.J. KREBS & D.M.
FERRERAS, P. (2001). Landscape structure and asymmetrical
SPRATT (eds). Ecologically-Based Rodent Management.
inter-patch connectivity in metapopulation of endangered
ACIAR Publications, Canberra : 217-219.
Iberian lynx. Biological Conservation, 100 : 125-136.
OGUGE, N.O. (1995). Diet, seasonal abundance and microhabi-
FLEMING, H.T. (1979). Life-history strategies. In : STODDART
tats of Praomys (Mastomys) natalensis (Rodentia : Muridae)
D.M. (ed). Ecology of small mammals. Chapman and Hall,
and other small rodents in a Kenyan sub-humid grassland
London : 1-52.
community. African Journal of Ecology, 33 : 211-223.

Movements and spatial patterns of Mastomys erythroleucus
89
OSTFELD, R.S. (1986). Territoriality and mating systems of Cali-
TAYLOR, K.D. (1968). An outbreak of rats in agricultural areas
fornia voles. Journal of Animal Ecology, 38 : 341-359.
of Kenya in 1962. East African Agricultural and Forestry
REXSTAD, E. & K. BURNHAM (2002). Users' guide for interactive
Journal, 34 : 66-77.
program CAPTURE. Abundance Estimation of Closed Ani-
mal Populations. Internet : ftp ://pwrcftpr.er.usgs.gov/
TAYLOR, K.D. & G. M. GREEN (1978). The influence of rainfall
research/software/capture.zip.
on diet and reproduction in four African rodent species. Lon-
ROHLF, F.J. & R.R. SOKAL (1984). Statistical tables. 2nd ed. W.
don Journal of Zoology, 174 : 453-471
H. Freeman and Co., San Franscisco.
STENSETH, N.C. & W.Z. LIDICKER JR. (1992). Pre-saturation and
THOMAS, J.A., N.A.D. BOURN, R.T. CLARKE, K.E. STEWART, D.J.
saturation dispersal 15 years later : some theoretical consid-
SIMCOX, G.S. PEARMAN, R. CURTIS & B. GOODGER (2001).
erations. In : STENSETH and W. Z. LIDICKER JR. (eds). Animal
The quality and isolation of habitat patches both determine
dispersal : Small mammals as a model. Chapman and Hall,
where butterflies persist in fragmented landscapes. Proceed-
London : 199-220.
ings of the Royal Society of London, 268 : 1791-1796.


Belg. J. Zool., 135 (supplement) : 91-96
December 2005
Influence of variations in land use intensity on species
diversity and abundance of small mammals in the Nama
Karoo, Namibia

Anke Hoffmann and Ulrich Zeller
Museum für Naturkunde, Institut für Systematische Zoologie, Invalidenstr. 43, 10115 Berlin, Germany
Corresponding author : Anke Hoffmann, e-mail : anke-hoffmann@gmx.net
ABSTRACT. The influence of the intensity of land use on small mammals in the ecoregion Nama Karoo, Namibia
was investigated within the biodiversity programme BIOTA. Changes in species diversity and abundance were
investigated across a fence separating heavily grazed communal and lightly grazed government owned rangeland.
Assessing and monitoring of the small mammal populations were done seasonally from 2001-2003 on each of 2ha
plots by using capture-mark-recapture methods. In total, 311 individuals representing nine species were caught
within 5760 trap nights. Species richness, total abundance, species diversity and settlement was lower in the over-
grazed area. The most abundant species were the Gerbillinae, Gerbillurus vallinus and Tatera leucogaster. T. leu-
cogaster
did not occur in the overgrazed area. Due to the loss of grass cover, smaller bush diversity, bush encroach-
ment and smaller arthropod abundance in the overgrazed area, changes in the small mammal community were most
likely caused by the loss of food resources, available dew, disruption of habitat structures, cover and shelter and by
increased predation risk. Only the `desert' species, G. vallinus, was favoured by the degraded land. It is also obvious
that the uncontrolled grazing in the communal lands has affected the biodiversity and the regeneration potential,
thus leading to land degradation.
KEY WORDS : Namibia, Nama Karoo, fence-line contrast, land use, rodent ecology, Gerbillurus, Tatera.
INTRODUCTION
on vegetation (NOY-MEIR et al., 1989; OLSVIG-WHIT-
TAKER et al., 1993; TODD & HOFFMAN, 1999); with only a
Small mammals are important components of arid and
few studies on how the extent of grazing influences
semi-arid ecoregions : as consumers (KERLEY, 1992a),
arthropod assemblages (DEAN & MILTON, 1995; RIVERS-
predators and dispersers of seeds (PRICE & JENKINS,
MOORE & SAMWAYS, 1996; SEYMOUR & DEAN, 1999;
1986), burrowers and as prey for carnivores and raptors
HOFFMANN et al., 2003). Investigations on the effect of
(KOTLER, 1984; HUGHES et al., 1994). Changes of habitat
different land use practises on small mammal assem-
structure and complexity are associated with changes in
blages in Africa are scanty (KERLEY, 1992b; NYAKO-
small mammal community structure and species richness
LARTEY & BAXTER, 1995; MONADJEM, 1999).
(DELANY, 1964; ROSENZWEIG & WINAKUR, 1969; BOND et
al., 1980; G
The aim of this study was to determine the influence of
RANT et al., 1982; ROWE-ROWE & MEESTER,
1982; K
different land use intensities on the diversity of small
OTLER, 1984; ABRAMSKY, 1988; KERLEY, 1992b;
E
mammals. Within the interdisciplinary biodiversity pro-
LS & KERLEY, 1996; HOFFMANN, 1999; AVENANT, 2000).
gramme BIOTA (Biodiversity Monitoring Transect Anal-
Large herbivores can modify the vegetation layer in
ysis in Africa; cf. ZELLER, 2003) a study on the population
terms of structure and species composition to a level
ecology of small mammals has been carried out in the
where small mammals are affected (BOWLAND & PERRIN,
Nama Karoo, southern Namibia. This study was carried
1989; KEESING, 1998; HOFFMANN, 1999). An ecological
out in the heavily grazed communal rangeland and a
disturbance of the habitat is often associated with
neighbouring moderately grazed rangeland, used for Kar-
decreases in small mammal diversity. Therefore biodiver-
akul breeding purposes, to address the following
sity of small mammals can be used as an indicator of dis-
questions :
turbance in an ecosystem.
Most of the Karoo ecoregion in Namibia is rangeland
(1) Does small mammal species richness, abundance
for livestock grazing (H
and diversity differ between overgrazed and moderately
OFFMAN et al., 1999) and heavy
grazing has left parts seriously degraded (L
grazed areas?
LOYD, 1999).
Livestock grazing has been identified as the major threat
(2) Does heavy grazing influence the composition of
to biodiversity in that region, but also mining, agriculture
small mammal assemblages?
and alien invasive plants are significant threats (LOVE-
GROVE, 1993; LLOYD, 1999). In general most investiga-
(3) Which species, if any, are most affected by the hab-
tions on the effects of grazing in rangelands have centred
itat changes? Does it favour any species?

92
Anke Hoffmann and Ulrich Zeller
MATERIAL AND METHODS
in each plot. The investigations took place at different
seasons per year. Traps were baited with a mixture of pea-
Study area
nut butter, oats, mashed bananas and bird seeds. These
were set before sun set, checked at night and in the morn-
The Nama Karoo occurs on the central plateau of the
ing. In the second study year, day trapping was done in
Cape Province in South Africa, and extends over the
addition. Captured animals were weighed, sexed and
Orange River into Namibia in the northwest, where the
body measurements and reproductive status were
study was conducted. This ecoregion is described as a
recorded. Each animal was individually marked by using
vast, open, arid region dominated by grassy dwarf shrub-
a subcutaneous tattoo on the underside of the tail's base
land with summer rain and climatic extremes, where
(HUGO, 1990; cf. HOFFMANN, 1999).
droughts are common (VENTER et al., 1986; DEAN & MIL-
TON, 1999). Most of this ecoregion is rangeland for live-
Data analysis
stock grazing (HOFFMAN et al., 1999); less than one per-
cent of the Nama Karoo is protected (COWLING, 1986;
The term `trap night' is used to describe one trap which
BARNARD et al., 1998). The region is characterized by
was set for a 24h period (ROWE-ROWE & MEESTER, 1982).
fence-line-contrasts caused by varying land use practises.
Trap success was calculated as the number of captured
individuals/100 trap nights. Abundance was used because
it differed only by 1.2% from the Minimum Number
Alive (MNA) (cf. BRONNER & MEESTER, 1987). For spe-
cies diversity calculations the Shannon Wiener diversity
index (Hs) was chosen. For survival calculations, new
individuals of the last trapping session were not consid-
ered. Those which were trapped only in one trapping ses-
sion, `survived' at least for one week.
Assessing and monitoring environmental features
For understanding the interrelation between small
mammal coenosis and their habitat different environmen-
tal features were assessed and monitored. Data on rainfall
and temperature were given by a BIOTA computer con-
trolled weather station near the Gellap-Ost observatory.
Fig. 1. ­ Fence-line-contrast of the study sites. Left : over-
Bushes were counted, their sizes estimated in four catego-
grazed communal farming area of Nabaos; right : moderately
ries between 0.5 - >2.0m and mapped. Ground vegeta-
grazed governmental farming area of Gellap-Ost.
tion cover per plot was estimated within ten 4m²-frames
by using the Londo-scale (LONDO, 1975) and then extrap-
The study was conducted on two neighbouring areas
olated. Also plant phenology was monitored over the
with different land use practises (Fig. 1), approximately
study period. Arthropod sampling was done using 10 pit-
20km northwest of Keetmanshoop. One study plot was
falls per plot over 8 days, to assess and monitor changes
highly overgrazed, mainly by goats within Nabaos com-
of epigaeic arthropod abundance (cf. VOHLAND et al.,
munal area (26°23'26''S, 17°59'43' E). The other plot (dis-
2005). Small mammal burrows were counted and mapped
tance 1.5km) was within the government Karakul sheep
once in both plots in October 2002. Observations of
breeding farm in Gellap-Ost (26°24'04''S, 18°00'17''E). In
potential small mammal predators were recorded.
contrast to the uncontrolled grazing in Nabaos, Gellap-
Ost uses a rotating grazing system with a lower stocking
RESULTS
rate. Free-ranging ungulates like Kudu (Tragelaphus
strepsiceros) and Steenbok (Raphicerus campestris) were
Habitat features of the study plots
rarely observed.
In Nabaos, no grass layer existed all year round. In
There are three main seasons : hot/wet (January-April)
Gellap-Ost the dominant grass species Stipagrostis unip-
cold/dry (May-August) and hot/dry (September-Decem-
lumis (height app. 50cm) covered the ground by up to
ber). Rainfall occurs in summer from January to April and
10%. Herbs occurred mainly after the rain. Bush cover
averages 150mm per year (cf. MENDELSOHN et al., 2002).
was generally low (Nabaos : 2.5%, 488m2, 1382 ind.;
In 2002 rainfall averaged 178mm, but in 2003 there was
Gellap-Ost : 2.1%, 415m2, 685ind.). More large bushes
drought, with only 55mm of rainfall. The mean monthly
( >1.5m) were found in Gellap-Ost, where bushes cov-
temperature range was 2002 22°-37°C (maximum) and
ered an area 1.6 times larger than that in Nabaos. Rhigo-
8°-20°C (minimum).
zum trichotomum was the dominant bush in both plots,
whereas Catophractes alexandri and Calicorema capitata
Trapping
were the subdominant species in Gellap-Ost, and Nabaos
Trapping was conducted on two 2ha grids which were
respectively. Both Boscia foetida and Phaeoptilum spino-
separated by a fence and 1.5km apart. Each grid consisted
sum were also abundant in both plots. Bush diversity was
of 90 Sherman standard live traps spaced by 15m inter-
lower in Nabaos (Hs 0.96) than in Gellap-Ost (Hs 1.28).
vals. Capture-mark-recapture methods (CMR) were used
All burrows were located in bushes in Nabaos, whereas
during trapping sessions of 4 consecutive trapping nights
only 61% were in bushes in Gellap-Ost. The rest of the

Influence of variations in land use intensity in the Nama Karoo, Namibia
93
burrows were in the open grassland. Twelve percent (out
(Canis mesomelas), Bat-eared Fox (Otocyon megalotis),
of 25) and 37% (out of 90) of the burrows were occupied
Cape Fox (Vulpes chama), Caracal (Felis caracal), mon-
by small mammals in Nabaos and Gellap-Ost respec-
gooses and various species of snakes.
tively.
In total, within 1280 trap nights 16713 epigaeic arthro-
Species richness
pods in 19 orders (without mites and collembola) were
Between October 2001 and August 2003, eight trap-
collected and were composed mainly of ants, with 9466
ping sessions per plot were conducted. Out of a total of
specimens, beetles with 1673 specimens, and termites,
5760 trap nights, 311 individuals (911 captures) repre-
with 747 specimens. Most animals were trapped in May
senting nine species were caught and marked (Table 1).
in both years, after the rainy season. In Nabaos the arthro-
The mean species richness in Nabaos was 3.3 (range : 2-
pod activity was lower and only 38% of the ground active
6) and in Gellap-Ost 5.4 (range : 4-6) (Fig. 2). The mean
arthropods were trapped.
monthly trapping success was 4.79 + 2.33 for Nabaos and
Potential predators of small mammals in the area were
10.42+ 6.32 for Gellap-Ost. The overall trap success in
the Spotted Eagle Owl (Bubo africanus), Pale Chanting
Nabaos out of 2880 trap nights was 3.75 and 7.05 in Gel-
Goshawk (Melierax canorus), Black-backed Jackal
lap-Ost.
TABLE 1
Species richness
Listed are all captured species and the total of recorded individuals within 5.760 trap nights.
Nabaos
Gellap-Ost
per plot 2880 trap nights
ind.(n)
%
ind.(n)
%
Macroscelididae
Elephantulus intufi (A. Smith, 1836)
1
0.92
9
4.46
Muridae >Gerbillinae<
Desmodillus auricularis (A. Smith, 1834)
10
9.17
3
1.49
Gerbillurus vallinus (Thomas, 1918)
80
73.39
44
21.78
Gerbillurus paeba (A. Smith, 1836)
1
0.92
Tatera leucogaster (Peters, 1852)
118
58.42
>Murinae<
Aethomys namaquensis (A. Smith, 1834)
1
0.92
13
6.44
Mus indutus (Thomas, 1910)
2
0.99
Rhabdomys pumilio (Sparrman, 1784)
12
11.01
4
1.98
Saccostomus campestris (Peters, 1946)
4
3.67
9
4.46
total individuals
109
202
S captures
282
629
The overall species richness and abundance was lower
T. leucogaster was the dominant species in Gellap-Ost,
in Nabaos than in Gellap-Ost, which is also expressed by
with a composition of 33-72% of the total abundance
the diversity index (Hs) : Nabaos (Hs 0.95; 7 species, 108
(Fig. 2b). In August 2003, when the abundance of T. leu-
individuals.), Gellap-Ost (Hs 1.29; 8 species, 203 individ-
cogaster was lowest, the proportion of other species was
uals). The main species were the gerbils, Gerbillurus val-
highest, which is also shown by a high diversity index
linus (Thomas, 1918) and Tatera leucogaster (Peters,
(1.47). Although fewer species were recorded per trap-
1852). It was striking, that T. leucogaster was not
ping session in Nabaos (Fig. 2) compared to Gellap-Ost,
recorded on the Nabaos plot. Mus indutus (Thomas,
there was an overlap in the species occurring in the two
1910) was only trapped in Gellap-Ost, but was also
plots (Table 1).
observed once in Nabaos. Crocidura sp. was found in owl
pellets of Bubo africanus, collected in the farming area of
Settlement and survival
Gellap-Ost during the study period, (pers. communication
M
Considering all individuals which had been trapped at
IKE GRIFFIN, Namibia 2003).
least over 2 trapping sessions (>11 weeks), we found a
Abundance and diversity
lower overall recapture rate in Nabaos (19.3%, n=109)
than in Gellap-Ost (31.8%, n=198). Five species were
In Nabaos, species diversity and total abundance (3-29
recaptured in Gellap-Ost. These were T. leucogaster
individuals/2ha) was lower than in Gellap-Ost (12-75
(37.9%), G. vallinus (11.4%), Aethomys namaquensis
individuals/2ha). In both plots, the highest recruitment
(38.5%), Elephantulus intufi (66.7%), Saccostomus
was found in August 2002 due to the high reproduction
campestris (33.3%). In Nabaos only G. vallinus (22.5%)
activity during the rainy season, followed by decrease of
and Desmodillus auricularis (20.0%) were recaptured.
total abundance in October 2002. G. vallinus was the
The mean minimum `survival' rate in weeks (w) shows
dominant species in Nabaos, and only subdominant in
the longest survival period for E. intufi in Gellap-Ost
Gellap-Ost, where population density fluctuated highly
(mean 20.0 w, range 1-64 w, n=9), followed by A.
(Fig. 2).
namaquensis (mean 9.6 w, range 1-41 w, n=13), T. leu-

94
Anke Hoffmann and Ulrich Zeller
cogaster (mean 8.5 w, range 1-53 w, n=116), S. campes-
more sensitive to habitat disturbance. Their requirements
tris (mean 4.3 w, range 1-11 w, n=9), G. vallinus (mean
in the microhabitat in relation to their nest building
3.6 w, range 1-42 w, n=44). In Nabaos G. vallinus has
behaviour are much higher than for precocial species
`survived' longer (mean 5.5 w, range 1-44 w, n=80) than
(ZELLER, 2003).
in Gellap-Ost, but D. auricularis was recorded for only
Many small mammal species are able to successfully
the mean of 3.0 weeks (range 1-10 w, n=11). In total ten
tolerate and exploit changes in their physical and biologi-
individuals had been trapped over a period of >40 weeks :
cal environments (DELANY & HAPPOLD, 1979). Grazing is
two specimens of G. vallinus in Nabaos and one in Gel-
one example for such an environmental change, which
lap-Ost, six specimens of T. leucogaster (one female 53
influences stratification of grass, plant species composi-
weeks) and one male E. intufi, which was still trapped in
tion and the standing crop biomass of grassland ecosys-
the last trapping session in Gellap-Ost after a `survival' of
tems (BOWLAND & PERRIN, 1989). The variety and abun-
about 64 weeks.
dance of small mammal communities might be dependent
on how grazers have utilized the grassland (GRANT et al.,
Hs 0.36
0.36

0.69
0.69
0.71
71 0.69

0.58
58
0.94
0.94 0.64
94
1982). Overgrazing affects the food (LACK, 1954) and
80
8
70
7
shelter of small mammals (BOWLAND & PERRIN, 1989).
60
6
The reduction of vegetation cover exposes them to preda-
50
5
tion (PEARSON, 1971). NYAKO-LARTEY & BAXTER (1995)
40
4
found that sites under rotation farming, such as the gov-
30
3
20
2
ernmental area in Gellap-Ost, and those grazed by cattle,
individuals / 2ha
no. of species
individuals / 2ha
10
1
support more rodents than constantly grazed areas and
0
0
those grazed by sheep. Therefore it is assumed that the
O N D J F M A M J J A S O N D J F M A M J J A
a)
2002
200
2003
20
habitat differences across the fence-line is not only caused
by the intensity of grazing, but also by the differences in
Hs 1.07
07

0.93 1.26 1.08 0.99
08

1.12
12
1.
33 1.47
33
feeding behaviour of goats (primarily browsers) and
80
8
sheep (selective grazers).
70
7
60
6
In contrast to a study in the Succulent Karoo (JOUBERT
50
5
& RYAN, 1999), the small mammal composition in the
40
4
overgrazed area of Nabaos was never just a subset of the
30
3
species composition encountered in the moderately
20
2
no. of species
individuals / 2ha
grazed areas of Gellap-Ost. Nevertheless, a high overlap
10
1
0
0
of occurring species was found in the year round study.
O N D J F M A M J J A S O N D J F M A M J J A
The mean species richness recorded in Nabaos was low
2002
200
2003
20
b)
(3.3), although it is comparable to the species richness
T.
T leuc
.
ogas
leuc
t
ogas e
t r
G.
G v
. a
v llinus
oth
ot er species
no. of
no.
s
of pecies
pec
(3.8) recorded in the semi-arid Karoo of southern Africa
(KERLEY, 1992b) and in different Fynbos habitats in South
Fig. 2. ­ a, b : Abundance and diversity of small mammals in
Africa (BOND et al., 1980; NEL et al., 1980; ELS & KER-
neighbouring study sites.
LEY
a : Nabaos, b : Gellap-Ost. Pictured is the abundance of
, 1996). Species richness was much higher in the
G. vallinus and T. leucogaster and other species (pooled), the
lightly grazed area of Gellap-Ost (mean : 5.4 species)
species number and the Shannon-Wiener-diversity-index for
across the fence-line. The dominance of small mammal
each trapping session.
communities also varied. The total disappearance of T.
leucogaster in the overgrazed area and the dominance of
G. vallinus was conspicuous. The higher `survival' rate of
G. vallinus indicates that this xeric adapted species, which
DISCUSSION
is confined to the western sector of the South West Arid
Zone and is known to prefer surface sand (DEGRAAFF,
In this study the intensive and uncontrolled grazing by
1981; DEMPSTER et al., 1999), found a more suitable habi-
livestock in the communal area had a clear negative
tat in the degraded land than in the grassy area of Gellap-
impact on species richness, diversity and settlement/sur-
Ost. This confirms the results of a biodiversity study in
vival of small mammals. This suggests that the total loss
rangelands of South Africa (FABRICIUS et al., 2003),
of ground vegetation cover leads to a reduced food supply
where the communal grazing area was characterized by
(plants, arthropods) and available dew for small mam-
xeric adapted reptiles and predatory arthropods, whereas
mals. Also disruption of habitat structure, cover and shel-
the nature reserve and commercial farms supported more
ter leads to a higher predation risk. Bushes represent areas
mesic-adapted species. In contrast, T. leucogaster is
of high food density and low predation risk (KOTLER,
found in a wide range of savannas and open woodlands of
1984). It is assumed that the decrease of large bushes in
southern Africa (DEGRAAFF, 1981; SMITHERS, 1983),
the communal area due to the intensive browsing by goats
where they generally occur in areas with mean annual
also reduces protection from predation. Burrows in the
rainfall of 250 mm and upwards (SKINNER & SMITHERS,
overgrazed area were found exclusively in bushes, which
1990). This gerbil occurs in drier areas like the Nama
is probably because of the loss of ground cover, and the
Karoo where it is assumed to depend on an adequate
direct disturbance by livestock. Intensive and continuous
ground vegetation layer not only because of food availa-
trampling by ungulates is described as a disruptive factor
bility and cover, but also due to dew water availability.
to small mammals (KEESING, 1998; HOFFMANN, 1999;
CHRISTIAN (1980) assumes that water availability may
KEESING & CRAWFORD, 2001). Altricial species are much
play an important role in coexistence and resource alloca-

Influence of variations in land use intensity in the Nama Karoo, Namibia
95
tion in desert rodents. Because T. leucogaster were found
REFERENCES
to be omnivorous (PERRIN & SWANEPOEL, 1987; NEAL,
1991; MONADJEM, 1997) with a high proportion of plant
ABRAMSKY, Z. (1988). The role of habitat and productivity in
items like seeds, rhizomes and bulblets (DEGRAAFF,
structuring desert rodent communities. Oikos, 52 : 107-114.
1981), might also explain their exclusive presence in Gel-
AVENANT, N.L. (2000). Small mammal community characteris-
lap-Ost.
tics as indicators of ecological disturbance in Willem Preto-
rius Nature Reserve, Free State, South Africa. South African
Several workers in southern Africa have reported cor-
Journal of Wildlife Research, 30 (1) : 26-33.
relations between the distributions of individual small
BARNARD, P., C.J. BROWN, A.M. JARVIS, A. ROBERTSON & L.
mammal species and measured habitat parameters, espe-
VAN ROOYEN (1998). Extending the Namibian protected area
cially ground cover (B
network to safeguard hotspots of endemism and diversity.
OND et al., 1980; KERLEY, 1992b;
M
Biodiversity and Conservation, 7 : 531-547.
ONADJEM, 1997). The quantity of cover is of prime
importance to the density and diversity of small mam-
BOND, W., M. FERGUSON & G. FORSYTH (1980). Small mammals
and habitat structure along altitudinal gradients in the south-
mals, but when cover reaches threshold levels the degree
ern Cape mountains. South African Journal of Zoology, 15 :
of plant species diversity becomes important (BOWLAND
34-43.
& PERRIN, 1989). A number of studies have shown that
BOWLAND, A.E. & M.R. PERRIN (1989). The effect of overgraz-
small mammal community structure is a function of plant
ing on small mammals in Umfolozi Game Reserve.
architecture (ROSENZWEIG & WINAKUR, 1969; BOND et al.,
Zeitschrift für Säugetierkunde, 54 : 251-260.
1980; KERLEY, 1992b; ELS & KERLEY, 1996). BOND et al.
BRONNER, G. & J. MEESTER (1987). Comparison of methods for
(1980) and ELS & KERLEY (1996) maintain that microhab-
estimating rodent numbers. South African Journal of Wildlife
itat features such as vegetation structure, cover and
Research, 17 (2) : 59-63.
height, relative humidity, litter depth, and foliage height
CHRISTIAN, D.P. (1980). Vegetative cover, water resources, and
diversity are directly related to the life form and growth
microdistributional patterns in a desert rodent community.
pattern of plant species within a plant community and
Journal of Animal Ecology, 49 : 807-816.
these factors are important floristic variables affecting
COWLING, R.M. (1986). A description of the Karoo Biome
small mammal community structure. The mechanism for
Project. South African National Scientific Programmes
Report
122, Pretora.
this relationship is generally thought to be that niche
DeGraaff, G. (1981). The rodents of Southern Africa. Durban,
availability (more specifically foraging microhabitat
Butterworths.
availability) is a function of habitat complexity : a more
DEAN, W.R.J. & S.J. MILTON (1995). Plant and invertebrate
complex habitat will contain more niches which may be
assemblages on old fields in the arid southern Karoo, South
exploited by more species (ROSENZWEIG & WINAKUR,
Africa. African Journal Ecology, 33 : 1-13.
1969).
DEAN, W.R.J. & S.J. MILTON (1999). The Karoo. Ecological pat-
terns and processes. Cambridge University Press, Cam-
Until now small African mammals have not enough
bridge.
considered adequately in both the planning and manage-
DELANY, M.J. (1964). An ecological study of the small mam-
ment of conservation areas, largely due to the difficulty in
mals in the Queen Elizabeth Park, Uganda. Revue de Zoolo-
assessing this group. According to AVENANT (2000), bio-
gie et de Botanique Africaines, 70 (1-2) : 129-147.
diversity of small mammals can be used as an indicator of
DELANY, M.J. & D.C.D. HAPPOLD (1979). Ecology of African
disturbance in an ecosystem, whereas the domination of
mammals. Longman, London and New York, 434pp.
an indicator species, low species richness and low diver-
DEMPSTER, E.R., M.R. PERRIN & C.T. DOWNS (1999). Gerbillu-
sity are useful tools for indicating disturbance on the pri-
rus vallinus. Mammalian Species, 605 : 1-4.
mary producer level. An understanding of determinants of
ELS, L.M. & G.I.H. KERLEY (1996). Biotic and abiotic correlates
small mammal community structure is therefore impor-
of small mammal community structure in the Groendal Wil-
tant for practical development of arid and semi-arid
derness Area, Eastern Cape, South Africa. Koedoe, 39 (2) :
121-130.
rangeland management guidelines (KERLEY, 1992b). BAR-
FABRICIUS, C., M. BURGER & P.A.R. HOCKEY (2003). Comparing
NARD et al., (1998) ascertains that the establishment of
biodiversity between protected areas and adjacent rangeland
wildlife conservation in commercial and communal farm-
in xeric succulent thicket, South Africa : arthropods and rep-
lands could improve the current protection status of the
tiles. Journal of Applied Ecology, 40 : 392-403.
Nama Karoo, with rural communities responsible for the
GRANT, W.E., E.C. BIRNEY, N.R. FRENCH & D.M. SWIFT (1982).
ecological management of large areas in habitats other-
Structure and productivity of grassland small mammal com-
wise overlooked for conservation.
munities related to grazing-induced changes in vegetative
cover. Journal of Mammalogy, 63 : 248-260.
HOFFMAN, M.T., B. COUSINS, T. MEYER, A. PETERSEN & H. HEN-
ACKNOWLEDGEMENTS
DRIKS (1999). Historical and contemporary agricultural land
use and the deserfication of the Karoo. In : DEAN, W.R.J. &
S.J. MILTON (eds), The Karoo. Ecological patterns and proc-
The study was carried out in the framework of the BIOTA
esses, Cambridge University Press, Cambridge : 257-273.
project financed by BMBF. We appreciate the support of the
HOFFMANN, A. (1999). Habitatnutzung und Populationsdynamik
Ministry of Environment and Tourism of Namibia, National
von Kleinsäugern im Grasland des Queen Elizabeth
Museum of Namibia, the Gellap-Ost Research Station and the
National Park, Uganda. Dissertation, Technische Univer-
village people from Nabaos. We wish to thank Mike Griffin for
sität Braunschweig, (http ://www.biblio.tubs.de/ediss/data/
his valuable advices and for the analysis of the owl pellets. We
20000114a/20000114a.html).
thank Seth Eiseb, Walter Hauwanga, Erik Maluleke and Mathew
HOFFMANN, A., K. VOHLAND, P. GIERE & U. ZELLER (2003).
Shuuya for their excellent field assistance. We are grateful to
Influence of grazing intensity on biodiversity of small mam-
Katrin Vohland for providing her entomological expertise.
mals and arthropods in the Nama Karoo of southern

96
Anke Hoffmann and Ulrich Zeller
Namibia. Abstract. In : Arid Zone Ecology Forum Confer-
NOY-MEIR, I., M. GUTMAN & Y. KAPLAN (1989). Responses of
ence 2003, Kathu, South Africa.
Mediterranean grassland plants to grazing and protection.
HUGHES, J.J., D. WARD & M.R. PERRIN (1994). Predation risk
Journal of Ecology, 77 : 290-310.
and competition affect habitat selection and activity of
NYAKO-LARTEY, Q. & R.M. BAXTER (1995). The effects of dif-
Namib desert gerbils. Ecology, 75 (5) : 1397-1405.
ferent grazing regimes on the population dynamics of small
HUGO, A. (1990). Wiederfangerfolg bei Kleinsäugern mit einer
mammals in the Eastern Cape. Transvaal Royal Society of
neuen Markierungsmethode. Zeitschrift für Säugetierkunde,
South Africa, 50 (2) : 143-151.
55 : 421-424.
OLSVIG-WHITTAKER, L.S., P.E. HOSTEN, I. MARCUS & E. SHO-
JOUBERT, D.F. & P.G. RYAN (1999). Differences in mammal and
CHAT (1993). Influence of grazing on sand field vegetation in
bird assemblages between commercial and communal range-
the Negev Desert. Journal of Arid Environment, 24 : 81-93.
lands in the Succulent Karoo, South Africa. Journal of Arid
PEARSON, O.P. (1971). Additional measurements of the impact
Environment, 43 : 287-299.
of carnivores on California voles (Microtus californicus).
KEESING, F. (1998). Ecology and behavior of the pouched
Journal of Mammalogy, 52 : 41-49.
mouse, Saccostomus mearnsi, in central Kenya. Journal
PERRIN, M.R. & P. SWANEPOEL (1987). Breeding biology of the
Mammalogy, 79 (3) : 919-931.
bushveld gerbil Tatera leucogaster in relation to diet, rainfall
K
and life history theory. South African Journal of Zoology, 22
EESING, F. & T. CRAWFORD (2001). Impacts of density and
large mammals on space use by the pouched mouse (Saccos-
(3) : 218-227.
tomus mearnsi) in central Kenya. Journal of Tropical Ecol-
PRICE, M.V. & S.H. JENKINS (1986). Rodents as seed consumers
ogy, 17 : 464-472.
and dispersers. In : MURRAY, D.R. (ed), Seed dispersal, Aca-
K
demic Press, Australia : 191-235.
ERLEY, G.I.H. (1992a). Small mammal seed consumption in the
Karoo, South Africa : further evidence for divergence in
RIVERS-MOORE, N.A. & M.J. SAMWAYS (1996). Game and cattle
desert biotic processes. Oecologia, 89 : 471-475.
trampling, and impact of human dwellings on arthropods at a
K
game park boundary. Biodiversity and Conservation, 5 :
ERLEY, G.I.H. (1992b). Ecological correlates of small mammal
community structure in the semi-arid Karoo, South Africa.
1545-1556.
Journal of Zoology, London, 227 : 17-27.
ROSENZWEIG, M.L. & J. WINAKUR (1969). Population ecology of
K
desert rodent communities : habitats and environmental
OTLER, B.P. (1984). Risk of predation and the structure of
desert rodent communities. Ecology, 65 : 689-701.
complexity. Ecology, 50 (4) : 558-572.
R
L
OWE-ROWE, D.T. & T. MEESTER (1982). Habitat preferences
ACK, D. (1954). The natural regulation of animal numbers.
and abundance relations of small mammals in Natal Draken-
Oxford, Clarendon Press.
sberg. South African Journal of Zoology, 17 : 202-209.
LLOYD, J.W. (1999). Nama Karoo. In : J. KNOBEL (ed), The mag-
SEYMOUR, C.L. & W.R.J. DEAN (1999). Effects of heavy grazing
nificent natural heritage of South Africa, Sunbird Publish-
on invertebrate assemblages in the Succulent Karoo, South
ing, Cape Town, South Africa : 84-93.
Africa. Journal of Arid Environment, 43 : 267-286.
LONDO, G. (1975). De decimale schaal foor vegetatiekundlige
SKINNER, J.D. & R.H.N. SMITHERS (1990). The mammals of the
opnamen van permanente Kwadraten. Gorteria, 7 (7) : 101-
southern African subregion. Pretoria, University of Pretoria.
106.
SMITHERS, R.H.N. (1983). The mammals of the southern African
LOVEGROVE, B. (1993). The living deserts of southern Africa.
subregion. Pretoria, University of Pretoria.
Fernwood Press, Cape Town.
TODD, S.W. & M.T. HOFFMAN (1999). A fence-line contrast
MENDELSOHN, J., C. ROBERTS & T. ROBERTSON (2002). Atlas of
reveals effects of heavy grazing on plant diversity and com-
Namibia. A portrait of the land and its people. Publisher :
munity composition in Namaqualand, South Africa. Plant
Ministry of environment and Tourism of Namibia, Cape
Ecology, 142 : 169-178.
Town.
VENTER, J.M., C. MOCKE & J.M. DE JAGER (1986). Climate. In :
MONADJEM, A. (1997). Habitat preferences and biomasses of
CROWLING, R.M., P.W. ROUX & A.J.H. PIETERSE (eds), The
small mammals in Swaziland. African Journal of Ecology,
Karoo Biome : a preliminary synthesis. Part 1. Physical
35 : 64-72.
Environment. South African National Scientific Programmes
MONADJEM, A. (1999). Population dynamics of Mus minutoides
Report 124, Pretoria.
and Steatomys pratensis (Muridae : Rodentia) in a subtropi-
VOHLAND, K., M. UHLIG, E. MARAIS, A. HOFFMANN & U. ZELLER
cal grassland in Swaziland. African Journal of Ecology, 37 :
(2005) : Impact of different grazing systems on diversity,
202-210.
abundance and biomass of beetles (Coleoptera), a study from
NEAL, B.R. (1991). Seasonal changes in reproduction and diet of
Southern Namibia. Mitteilungen aus dem Museum für
the Bushveld gerbil, Tatera leucogaster (Muridae : Roden-
Naturkunde in Berlin, Zoologische Reihe, 81 (2) : 131-143.
tia), in Zimbabwe. Zeitschrift für Säugetierkunde, 56 : 101-
ZELLER, U. (2003). Functional zoodiversity in biomes of south-
111.
ern Africa and its changes by land-use and environment. In :
NEL, J.A.J., I.L. RAUTENBACH & G.J. BREYTENBACH (1980).
BIOTA Southern Africa. Towards sustainable use of biodi-
Mammals of the Kammanassie Mountains, southern Cape
versity. Published by University of Hamburg, Botanical
Province. South African Journal of Zoology, 15 : 255-261.
Institute and Botanical Garden, Volume I : 231-264.

Belg. J. Zool., 135 (supplement) : 97-102
December 2005
Morphological and morphometrical analyses of three
cryptic species of Tatera Lataste, 1882 (Rodentia : Muri-
dae) from West Africa

Laurent Granjon
Laboratoire de Mammalogie (CBGP, UR 022), IRD, BP 2528, Bamako, Mali & UMR "Origine, Structure et Evolution de la
Biodiversité", Dept "Systématique et Evolution", Museum National d'Histoire Naturelle, 55 rue Buffon, 75005 Paris, France

Corresponding author : Laurent Granjon, e-mail : granjon@ird.fr
ABSTRACT. Three chromosomally well-differentiated but morphologically similar species of the Gerbilline rodent
genus Tatera occur in West Africa, namely T. gambiana, T. guineae and T. kempi. In order to find reliable diagnostic
characters, morphological and morphometrical analyses were performed on samples of these three species from
Mali, the only country where they are known to occur sympatrically. Body measurement comparisons show that T.
guineae
has a longer tail and hindfoot, and T. gambiana a shorter ear, relative to the two other species. Skull meas-
urements are less variable between species as only a larger upper tooth row characterizes T. kempi, and a shorter
tympanic bulla length T. guineae. Conversely, geographically distant samples of T. kempi from Mali and Chad differ
mainly for skull measurements. There is a wide overlap between the ranges of values recorded in the three species.
Principal Component and Discriminant Analyses of skull measurements highlight the distinctiveness of T. guineae,
whose sample is clearly separated from those of T. kempi and T. gambiana. All these results validate the status of
morphologically cryptic species for T. kempi and T. gambiana, long treated as synonymous. Tatera guineae is the
most differentiated, but the diagnosis of an individual specimen may prove difficult due to the observed interspe-
cific overlap in all the characters considered.
KEY WORDS : Gerbilline rodents, Tatera, skull measurements, multivariate analyses, phenetic similarity.
INTRODUCTION
lighted the difficulty in finding and using unambiguously
diagnostic morphological characters to recognize species
The Gerbilline rodent genus Tatera Lataste, 1882 is
in the genus. This problem has been repetitively men-
widely distributed in all sub-Saharan Africa, occupying a
tioned by those who tried to organize the variability
variety of habitats in northern and southern savannas to
observed in Tatera into a coherent taxonomic arrange-
the margins of the rainforest. MUSSER & CARLETON
ment (PIRLOT, 1955; DAVIS, 1966; ROSEVEAR, 1969; MAT-
(1993) recognized twelve species in the genus, the major-
THEY & PETTER, 1970; TRANIER, 1974; BATES, 1985 &
ity of which occur in southern or eastern Africa. Only two
1988). Since the first karyotypic data obtained by MAT-
of these can be considered as having an exclusive West
THEY (1954), cytogenetics has proven to be of great help
African range, namely T. guineae Thomas, 1910 and T.
in species identification, and chromosome data have
kempi Wroughton, 1906. The first one is characteristic of
accumulated for this genus (see QUMSIYEH & SCHLITTER,
the Sudan savannas from Senegal, Guinea and Sierra
1991 for a review until 1990; COLANGELO et al., 2001;
Leone to Burkina Faso and Togo, whereas the second
VOLOBOUEV et al., in prep.). As for West African species,
ranges from Sudan to Guinean savannas and from Sierra
there is now a general agreement on karyotypic character-
Leone to eastern Chad and northern Central African
istics of T. gambiana (2n=52), T. guineae (2n=50) and T.
Republic. A third species, T. gambiana Thomas, 1910,
kempi (2n=48; COLANGELO et al., 2001; VOLOBOUEV et al.,
should now be re-evaluated as present in West Africa :
in prep.).
Long treated as synonym of T. kempi, T. gambiana has
proven to be a distinct biological species, which main
These three species have been found in southern Mali,
range covers Sahel savannas of Senegal and Mali (MAT-
which as such represents a unique crossroads for this
THEY & PETTER, 1970; VOLOBOUEV et al., in prep.), with a
genus in West Africa. Based on series of karyotyped and
proposed eastward extension to eastern Niger (DOBIGNY
therefore unambiguously determined specimens, we try
et al., 2002a). On the other hand, the presence of T.
here to find morphological and/or morphometrical char-
robusta (Cretzschmar, 1830) in West Africa should still
acteristics enabling their determination in the field or
be considered doubtful, as it only relies on one single
from museum specimens. Interspecific variability of the
specimen from Burkina Faso (BATES, 1985).
characters used was assessed in T. kempi through the
Tatera taxonomy and systematics has always been a
comparative study with a sample from Chad, i.e. the east-
matter of debate. As early as 1906, WROUGHTON high-
ern extreme of the species range.

98
Laurent Granjon
MATERIALS AND METHODS
specimens and/or they were highly correlated with other
characters. PCA was run using the covariance matrix on
All the specimens used in this study were karyotyped
log-transformed data, whereas DA gave better results
following standard procedures (see G
based on non-transformed data.
RANJON & DOBIGNY,
2003 for instance). They are currently deposited as skin
and skull specimens in the small mammal collection of
RESULTS
the Institut de Recherche pour le Dévelopement, Bamako
and will ultimately be housed at the Museum National
d'Histoire Naturelle, Paris. The samples from Mali are as
The mean and standard deviation values, as well as
follows : 7 T. kempi (M177, M178, M179, M180, M196,
ranges of all measurements taken on the 4 Tatera sam-
M197, M198) from Founé (12°50'N - 4°42'W); 15 T.
ples, are summarized in Table 1. Distributions of individ-
gambiana from various localities of west (Sekokoto :
ual dental wear data were compared between samples,
13°31'N ­ 10°45'W, N=1 : M4178; Maréna : 14°38'N -
prior to other statistical tests : None of the differences
10°36'W, N=2 : M4183, M4190) and central south
observed was found significant via non parametric Wil-
(Katibougou : 12°30'N ­ 8°6'W, N=1 : M4412;
coxon tests (P=0.083 between T. kempi from Mali and
Samanko : 12°32'N ­ 8°5'W, N=7 : M4909, M4912,
from Chad, P=0.194 between T. kempi and T. gambiana
M4916, M4918, M4954, M4955, M4978; Samaya :
from Mali, P=0.059 between T. kempi and T. guineae
12°34'N ­ 8°4'W, N=4 : M194, M195, M199, M202);
from Mali, and P=0.924 between T. gambiana and T.
and 24 T. guineae from various localities along the Guin-
guineae). Based on the hypothesis that age is one of the
ean border, mainly in the Mts Mandingues region
important determinants of dental wear in rodents, this
(Kolobo : 13°15'N ­ 11°31'W, N=1 : M4170; Gainsoa :
suggests that the samples compared can be considered of
12°27'N ­ 10°15'W, N=1 : M4723; Balamansala :
similar age, which makes the following comparisons
12°11'N ­ 8°48'W, N=8 : M4670, M4694, M4709,
more relevant. Sexual dimorphism was tested for all body
M4753, M4796, M4858, M4864, M4902; Kigniélendo :
and skull measurements in the three largest samples (T.
12°20'N ­ 8°32'W, N=1 : M4664; Kangaba : 11°58'N ­
kempi from Chad, T. gambiana and T. guineae from
8°25'W, N=1 : M188), in the central south (Niamana :
Mali), using Mann-Whitney U test. Significant differ-
13°01'N ­ 8°14'W, N=4 : M4015, M4019, M4093,
ences (0.01<P<0.05) were found in only 5 (of 48)
M4111; Kalifabougou : 12°57'N ­ 8°11'W, N=7 :
instances, males being in all cases larger : Wt, HB and T
M4037, M4044, M4048, M4099, M4104, M4107,
in T. kempi from Chad, Wt and IC in T. gambiana from
M4114) and near the border with Burkina Faso in the
Mali. According to these results, sexual dimorphism was
south east (Mamouroubougou : 11°12'N ­ 5°29'W, N=1 :
considered negligible in the samples here used and males
M4879). Additionnally, 23 specimens of T. kempi from
and females were pooled in subsequent analyses.
the Zakouma National Park in south eastern Chad
Given the relatively small sample sizes, the differences
(10°41'N ­ 19°29'E) were included (I16, H2, O2, O5,
recorded by student t tests were conservatively quoted as
O9, O26, O27, O28, O29, O38, O45, O46, P'3, Q10, Q17,
significant only when P<0.01. T. gambiana and T. kempi
Q19, Q20, Q27, Q43, R6, R7, R8, R13). All specimens
from Mali are the two samples showing the fewer number
were adult, most (60 of 67) belonging to age classes 2 or 3
of differences (n=3), whereas T. gambiana and T. guineae
following the 4-class system of BATES (1985) based on
on the one hand, T. kempi from Mali and T. kempi from
dental wear.
Chad on the other hand, display significant differences for
The following classical body measurements were
7 measurements each. Regarding body measurements in
taken, to the nearest gram or millimetre : Weight (Wt),
the Malian samples, T. guineae is characterized by a
head and body length (HB), tail length (T), ear length (E)
longer tail and longer hindfoot than both T. gambiana and
and hindfoot lenth, excluding claw (HF). A selection of
T. kempi, the latter sharing a long ear with T. guineae rela-
11 skull measurements was made, based on their pre-
tive to T. gambiana. Skull measurements do not show
sumptive usefulness as revealed by previous studies,
many significant differences between species. The main
especially those of BATES (1985, 1988) and COLANGELO
one concerns UTR, significantly longer in T. kempi than
et al. (2001) : Greatest length of skull (GLS), condyloba-
in the two other species, and TBL, significantly shorter in
sal length (CBL), greatest width of skull (GWS), breadth
T. guineae than in the two other species. Overlap between
of braincase (BB), interorbital constriction (IC), rostral
ranges of values is the rule, with the exception of tail
width (RW), rostral length (RL), trans molar width
length in the sample of T. guineae when compared with
(TMW), tympanic bulla length (TBL), mandible length
the samples of the two other species from Mali. Differ-
(ML) and upper toothrow length (UTR). They were taken
ences between the two geographically distinct samples of
as described in BATES (1988), to the nearest 0.1mm.
T. kempi often concern measurements that are not signifi-
Skulls were examined for possible diagnostic characters.
cantly variable between species in Malian samples, and
Special attention was paid to the first lower molar, the
are mainly skull ones.
first lamina of which was characterized following the pat-
terns illustrated in B
The distribution of individuals of the four study sam-
ATES (1988).
ples according to the morphology of their first lower
Measurements were compared by means of student t
molar first lamina is summarized in Table 2. Most of the
tests. Eight skull measurements were used to run Princi-
T. gambiana and T. kempi specimens are characterized by
pal Component (PCA) and Discriminant (DA) analyses,
a M first lamina opening posteriorly, vs only half of the
1
using SYSTAT 8.0 (1998). GLS, RL and ML were aban-
T. guineae specimens, the other half having their first
doned because they could not be taken in a number of
lamina divided into 2 islands.

Tatera morphometrics
99
R
9
6
5
3
9 3
4-
5.9
6
18 6.1
S *
S
UT
1
4 5.7
23
2
3
0
.
1
6
4
5.
5.6 0.178 3-
3-
N
*
**
N
5.
6.0 0.175 5.7-6.2
5.8 0. 5.
.4
.6
.8
29 13 26
93 15 25
6 95 43 24.3
44
25
* S S
S
ML
1
5
20
2
1
24. 1.4
22. 1.2
23. 0.2 .6-
23. 0.931
N N
N
2
0
.
9-
2
0
.
3-
23
2
1
.
1-
2
L
9
.17 15 10.7
68 93 1.6
.67 16
.50
1.
-1
* S
S
TB
1
5
24
2
3
*
*
10 0.4
9.
N
9
.2-
0.5 9-1
10 0.2
10 0.38
1
0
.
4
-
1
0
.9
9
.6
S SN S
1
3
-
8
.
7
24
-
8
.
5
3
-
8
.
6
N
N
*
*
TMW
8
.
16 0.312 7.8
7
.
94 0.395 7.1
8
.
25 0.226 8-8.5
7
.
73 0.268 7.3
5 .7
.8
.1
14
2 15
5
72
04 14
SN S
RL
1
1
90
1
9
12
32 -14.2
*
*
*
1
3
.
1
4
0.
1
4
.
2
8
1.
1
3
.
6
8
0.
1
2
.
5
9
0
.6
N N
1
1
.
8-
1
1
.
5-
1
3.3
1
1
.
5-
9
9
-
5
.
9
-
6
.
9
7666
1
1 -5.4
-
5
.
7
S S
*
RW
1
4 5.27
2
4
2
31
2
6
0
0
.28
NS N N
4.9
5
.56 0.39 4.9
5
.33 0.1 5.1
5
.03 0. 4.5
2 .0
8
4 .7
3
S S
IC
1
5
-7
7
-6
6
.35
2
4
0
.
30
NS
N
NS
6
.0
6
.55 0.30 6-7.3
6
.39 0.34 5.9
6
.32 0.223 6-6.9
5
3
4
4
1)
0 0 16.
0 4
2 2
5
6
16.
SN S
000
BB
1
5
2
4
2
32
4
1
3
1
5
.7 0.40 1-
1
5
.5 0.43
1
5
.9 0.39 4-
1
5
.6 0.
NS N N
NS
<0.
1
5.
1
4
.
9
-
16.
1
5.
1
4
.
7
-
16.
*p
.2
.5
;
**
10 98 20.7
56 75 21
6 83 80 20.3
37
20
S S
001
1
2
19
2
2
GWS
19. 0.9 .5-
18. 1.1
19. 0.6 .6-
18. 0.672
NS N N
*
**
0.
17
1
6
.
7-
18
1
6
.
9-
*p<
.7
1
;
*
88 36
37 36.4
2
6
1 37.2
S S
1
1
4
24
6
2
3
N N *
*
E
0.0
CBL
3
4
.
63
1
.3
3
3
.
87
1
.7 .4-
3
5
.
90
0.6
3
4
.
00
1
.
33 .4-
L
3
1
.
8-
30
3
5
-
3
6
.7
30
B
p<
TA
s
(*
lt
0
su
0 3 1.2
5 4
2
-4
9 22
S
re
.2 52 -4
.8 86 -40.6
5 .2
.7
*
GLS
1
2
20
.1
1
9
38 1.
37 1.
39 0.712
36 1.4
NS N NS
3
6
.5
3
3.5
38
3
3
.
1
-
4
0
t
t
e
s
t
s
5
5
ent
0 7
1 1
6 9
7
E
5
2
4
23.
7
2
3
2
3
7 21.
*
*
*
t
ud
1
9.7 1.20 18-22
2
1.2 0.87
*
*
*
NS
2
0-
2
1.8 1.06 20-23
2
0.0 1. 17-
a
t
e
d
s
3 7 5
7 6 7
7 2 4
7
* S
s
o
ci
H
1
51
-3
2
3
-3
7
-3
2
3
*
*
3
0
.8 1.39 29
3
4
.8 1.23 32
3
2
.5 1.51 31
3
1
.8 1.245
** N
NS
2
8
-
3
3
.
5
n
d
as
82 2
20 6
20 96
26
5
9
S
t
e
r
a
, a
TL
1
1
9
3
7
*
**
*
*
NS
1
44. 8.54 128-160
1
73. 6.49 163-179
1
38. 12.6 120-150
1
49. 8. 127-170
f

Ta
e
s
o
8
5
5 33 19 18
00 57
14 22
96
75
15
S SN
mpl
HB
9-
2
4
0-
2
31
9
.
8
83
N N NS
NS
sa
1
53. 13.8 13
1
43. 12.8 116-164
1
38. 12.5 12
1
38.
1
1
4
-
1
5
9
4
of
7 08
4
6
3
4
7
90
Wt
1
51
2
4
7
2
3
* * NS
NS
mm)
1
1
4.4 30.2 78-180
8
7
.
1
3
2
2.0 50-136
7
7
.
7
1
1
2.1 56-
7
1
.
9
6
1
2
.
51
5
1
-
101
n
n
t
s

(i
ion
i
a
t
i
o
n
i
a
t
i
o
n
i
a
t
i
o
n
at
eme
e
e
e
e
vi e
ur
an
ng
an
ng
an
ng
an
ng
as

N Me St. Dev Ra
N Me St. Dev Ra
N Me St. Dev Ra
N Me St. D Ra
l
me
kul
h
ad
n
d
s
e
C
i
y
a
l
i
ea
in

mpi
li
e
mp

Bod

ke
li
mpi
Ma
ad
s
gu

vs
e
vs k ke
b
i
a
n
a
Ma
ea
p
i
Ma
p
i
Ch
na
ia

a
m

in
e
m

e
m

b
i
a
n
a
v

p
i
Mali
g
k
k
m
m

T.
T
.
gu

T.
T.
ga gamb guineae vs
ke

100
Laurent Granjon
TABLE 2
4
Morphology of the first lower molar first lamina (see Bates,
3
T. gambiana, Mali
1988 for an illustration)
2 2
T. kempi T. kempi
T. kempi, Mali

T. gambiana
T. guineae
(Mali)
(Chad)
1
T. guineae, Mali
1st lamina open
12
8
5
22
0
posteriorly
Discriminant Vector -1
1st lamina open
0
0
0
0
anteriorly
T. kempi, Chad
1st lamina divided
1
8
1
0
-3
into 2 islands
-5 -4 -3 -2 -1 0 1 2 3 4 5
Other
0
0
1
1
Discriminant Vector 1
Fig. 2
Fig. 2. ­ Scatter plot of the first two Discriminant Vectors
2
from the DA of 8 raw skull measurements of the four samples
of Tatera from Mali (T. gambiana, T. guineae & T. kempi) and
Chad (T. kempi). Confidence ellipses drawn using "Sample"
1
option with P=0.683.
The hierarchical classification (Single Linkage
26.5% 0
Method) based on Mahalanobis distances between group
PC 2 ­
means (Fig. 3) shows that the two T. kempi samples are
phenetically the most similar, T. guineae being the most
-1
divergent. When both T. kempi samples are pooled in DA,
the overall percentage of well-classified individuals
remains 94%, (2 T. kempi and 1 T. gambiana misclassi-
-2-3
-2
-1
0
1
2
3
fied). When only samples from Mali are considered, this
PC 1 ­ 49%
Fig. 1
value raises to 97%, only one T. gambiana being mis-
taken for a T. kempi.
Fig. 1. ­ Scatter plot of the first two Principal Components (F1
x F2) of the PCA run on 8 log-transformed skull measure-
ments of the four samples of Tatera from Mali (T. gambiana -
T. guineae
squares, T. guineae - triangles & T. kempi ­ open circles) and
Chad (T. kempi ­ black circles).
T. gambiana
T. kempi (Mali)
The scatter plot of the first two principal components is
T. kempi (Chad)
represented in Fig. 1 (representing more than 75% of the
Fig. 3. ­ Hierarchical classification (single linkage method)
total variance of the original matrix). All the characters
based on Mahalanobis distances between the four samples of
correlate positively with PC 1, which thus represents an
Tatera studied in DA (cf Fig. 2).
overall size axis. The main loadings on this axis are from
RW, GWS, CBL, TMW and IC. The dispersion of indi-
viduals of the four study samples along this factor
matches quite well the age-related differences in size
DISCUSSION
observed within each sample. Most of the inter-samples
differentiation occurs along PC 2, which mainly separates
The validity of the various characters that have been
T. guineae from the other three samples. The variables
used to distinguishing between species and species
which contribute the most to this discrimination are TBL
groups in Tatera has been critically discussed by DAVIS
(positive loading) and RW (negative loading). Overlap
(1966) and ROSEVEAR (1969). DAVIS (1966) distinguished
between the two T. kempi and the T. gambiana samples is
an afra and a robusta group based on a series of charac-
important, and stay so along PC 3 or 4 (not shown).
ters showing a relatively large degree of overlap. Among
the 3 species studied here, T. kempi and T. gambiana are
The scatter plot of the first two discriminant vectors
considered to belong to the afra group, and T. guineae to
represents more than 95% of the total dispersion among
the robusta one. As a matter of fact, the general coloration
the study samples (Fig. 2). The associated classification
of T. kempi and T. gambiana is usually quite dull, their
matrix indicates that all T. guineae (14/14) and T. kempi
pelage is rather harsh and their tail shorter or subequal
from Mali (6/6) are correctly classified by the analysis, vs
than head and body, without a densely haired terminal
only 9 of 10 T. gambiana and 19 of 21 T. kempi from
pencil while the pelage of T. guineae is more silky with a
Chad, yielding an overall value of 94% of well-classified
brighter coloration, and its tail is usually much longer
individuals.
than head and body, with a marked terminal pencil of

Tatera morphometrics
101
hairs. These tail characteristics appear to be reliable
Multivariate analyses confirmed the distinctiveness of
whenever the tail is complete, while pelage aspect is more
T. guineae, which even appeared well distinct in PCA
subject to individual variation.
ordination. Conversely, T. gambiana and T. kempi still
Similarly, skull characters are difficult to associate to
show some overlap in the distribution of their individuals
one or another species. No one has been found to unam-
in the DA space. This confirms the validity for West Afri-
biguously characterize any of the three species here stud-
can species of the distinction between the afra and
ied. Even the morphology of the zygomatic plate, said to
robusta groups made by DAVIS (1966). Within the species
project less forward in T. guineae than in other West Afri-
of the afra group, the percentage of misclassified individ-
can species by ROSEVEAR (1969), can be of ambiguous
uals between T. kempi and T. gambiana was weak, which
interpretation. The morphology of the M first lamina
makes it possible to envisage successful assignation of
1
used by BATES (1988) to distinguish between T. valida
unknown individuals with rather great confidence based
valida and T. valida kempi also diplayed variability within
on this kind of analyses. For that purpose, sample sizes of
each sample. However, it proved to be less variable in
unambiguously determined individuals should be
both the T. kempi samples studied here than in the one,
enlarged, so that age-related (and possibly sex-related)
mainly from Benin, analysed by COLANGELO et al. (2001).
variation can be taken into account more precisely. But
Here, and as in T. gambiana, the first lamina opens poste-
even with such precautions, one should keep in mind that
riorly in the vast majority of individuals, as supposed to
this kind of use should be restrained to geographically
be the rule in T. kempi according to BATES (1988).
well-defined regions : Geographical variation, here illus-
Measurements give useful information on species char-
trated between samples of T. kempi from Mali and Chad,
acteristics, but they can hardly be used on an individual
is always important and counfounding in such groups of
basis to identify a given specimen unambiguously, as the
cryptic species, a qualifier that certainly applies to these
range of values overlap widely in most cases. The same
West African Tatera, as it does for East African ones
was observed by B
(BATES, 1985) or for the related genus Taterillus Thomas,
ATES (1985) in his comparative analy-
sis of cryptic species from East Africa, namely T. robusta,
1910 (DOBIGNY et al., 2002b).
T. nigricauda (Peters, 1878) and T. phillipsi (de Winton,
1898). This casts doubt on the absolute diagnostic value
ACKNOWLEDGEMENTS
of some of the criteria proposed by ROSEVEAR (1969) and
resumed by MATTHEY & PETTER (1970) to distinguish
Many thanks are due to S. Ag Atteynine and D. Abdoulaye
between Tatera species in West and Central Africa (e.g.
for their careful work at skull and skin preparation, and for their
molar row length or bullae length, but see hereunder). As
participation to the capture and karyotyping of the correspond-
said above, tail length is especially helpful in discriminat-
ing specimens, to which B. Sicard and C. Koné also contributed.
ing T. guineae from the other species, as is hind foot
I am also grateful to C. Denys for access to specimens from
length. Combining these two measurements, in bivariate
Chad, and to J.M. Duplantier, V. Volobouev and an anonymous
referee for their comments on an earlier draft of this paper.
plots, clearly separate T. guineae from the two other spe-
cies (not shown here due to scarcity of data on tail
length). As for skull measurements, the short bullae of T.
REFERENCES
guineae and the long molar row of T. kempi were already
used by ROSEVEAR (1969) in his determination key for
BATES, P.J.J. (1985). Studies of gerbils of genus Tatera : The
West African species of Tatera. The large maxillary tooth
specific distinction of Tatera robusta (Cretzschmar, 1826),
row in T. kempi was also mentioned by COLANGELO et al.
Tatera nigricauda (Peters, 1878) and Tatera phillipsi (De
(2001) in their sample from Benin, when compared with
Winton, 1989). Mammalia, 49 : 37-52.
data on T. nigrita W
B
ROUGHTON, 1906 (= T. valida Bocage,
ATES, P.J.J. (1988). Systematics and zoogeography of Tatera
1890 in M
(Rodentia : Gerbillinae) of north-east Africa and Asia. Bonn.
USSER & CARLETON, 1993) from Chad pub-
Zool. Beitr., 39 : 265-303.
lished by TRANIER (1974). However, comparing skull
C
measurement data of T. kempi from Mali or Chad with
OLANGELO, P., M.V. CIVITELLI & E. CAPANNA (2001). Morphol-
ogy and chromosomes of Tatera Lataste, 1882 (Rodentia
those on T. kempi from Benin presented by COLANGELO et
Muridae Gerbillinae) in West Africa. Tropical Zoology, 14 :
al. (2001) shows that the latter are systematically larger.
243-253.
This trend is not associated with an older mean age of the
DAVIS, D.H.S. (1966). Contribution to the revision of the genus
sample from Benin, as the differences of dental wear dis-
Tatera in Africa. Ann. Mus. Roy. Afr. Centr. Zool., 144 : 49-
tribution between this sample and the two T. kempi ones
65.
of our study are not significant (Wilcoxon non parametric
DOBIGNY, G., A. NOMAO & J.C. GAUTUN (2002a). A cytotaxo-
test, P=0.339 with Malian sample, P=0.837 with Chadian
nomic survey of Rodents from Niger : implications for sys-
sample). Conversely, it could be linked with the way in
tematics, biodiversity and biogeography. Mammalia, 66 :
which the measurements were taken (experimental bias),
495-523.
but may also illustrate a true geographical variation. The
DOBIGNY, G., M. BAYLAC & C. DENYS (2002b). Geometric mor-
latter interpretation is reinforced by the differences
phometrics, neural networks and diagnosis of sibling Tateril-
observed here between the samples from Chad and Mali.
lus species. Biological Journal of the Linnean Society, 77 :
319-327.
It has also to be underlined that none of the variables used
G
in bivariate representations by B
RANJON, L. & G. DOBIGNY (2003). The importance of cytotax-
ATES (1985, 1988) to sep-
onomy in understanding the biogeography of African
arate between East African species proved to be of use
rodents : Lake Chad murids as an example. Mammal Review,
here (GLS vs RW or BB; RW vs RL), showing again the
33 : 77-91.
great similarity between the skulls of the samples under
MATTHEY, R. (1954). Nouvelles recherches sur les chromosomes
study.
des Muridae. Caryologia, 6 : 1-44.

102
Laurent Granjon
MATTHEY, R. & F. PETTER (1970). Etude cytogénétique et taxo-
QUMSIYEH, M.B. & D.A. SCHLITTER (1991). Cytogenetic data on
nomique de 40 Tatera et Taterillus provenant de Haute-Volta
the rodent family Gerbillidae. Occ. Papers Mus. Texas Tech
et de République Centrafricaine (Rongeurs, Gerbillidae).
Univ., 144 : 1-20.
Mammalia, 34 : 585-597.
ROSEVEAR, D.R. (1969). The rodents of West Africa. Trustees of
the British Museum (Natural History), London, 604p.
MUSSER, G. & M.D. CARLETON (1993). Family Muridae. In :
SYSTAT (1998). SYSTAT for Windows, Version 8.0, SPSS Inc.,
WILSON, D.E. & D.M. REEDER (eds), Mammal species of the
Chicago.
world : a taxonomic and geographic reference, Smithsonian
TRANIER, M. (1974). Dimorphisme chromosomique dans une
Inst. Press, Washington : 501-755.
population de Tatera du Sud du Tchad (Rongeurs, Gerbilli-
dés). Mammalia, 38 : 224-233.
PIRLOT, P.L. (1955). Variabilité intra-générique chez un rongeur
WROUGHTON, R.C. (1906). Notes on the genus Tatera with
africain (Tatera Lataste). Ann. Mus. Congo, Tervuren, Zool.,
descriptions of new species. Ann. Mag. Nat. Hist., 7 (17) :
39 : 9-66.
474-499.

Belg. J. Zool., 135 (supplement) : 103-107
December 2005
Survival and roost-site selection in the African bat Nyc-
teris thebaica
(Chiroptera : Nycteridae) in Swaziland

Ara Monadjem
All Out Africa Research Unit, Department of Biological Sciences, University of Swaziland, Private Bag 4, Kwaluseni, Swazi-
land
Corresponding author : Ara Monadjem, e-mail : ara@uniswacc.uniswa.sz
ABSTRACT. Survival and mortality of African bats is poorly known. A banding study of a population of Nycteris
thebaica
, roosting in 15 road culverts in north-eastern Swaziland, was initiated in 1998. Since then, a total of 799
bats have been banded including five cohorts of same-aged individuals of known age. Cohort life-tables and survi-
vorship curves were calculated using these data. For both males and females, survivorship was low in the first year,
but increased thereafter. Approximately 15% of females and 10% of males banded as juveniles in 1998 and 1999
survived to three years of age. Of female bats banded as adults in July 1998 (n = 39), 23% had survived to January
2003 (4.5 years). The corresponding value for males (n = 6) was zero. Of 28 male bats banded in 1998, only one
(4%) was recaptured after four years. Female values probably reflect true survival and mortality, whereas, dispersal
and movement complicate the values for males. Female Nycteris thebaica did not randomly select roosting sites.
The 15 culverts were occupied by six discrete groups of female bats. The presence of male bats at the study site was
irregular, with movements of 9 km having been recorded by one particular individual.
KEY WORDS : Nycteris thebaica, survival, roost selection, Africa.
INTRODUCTION
is not known whether these individuals regularly move
between culverts or use one or a few culverts exclusively.
The ecologies and life histories of African bats are rela-
Nycteris thebaica often uses different day and night
tively poorly known. For many species, even basic distri-
roosts (TAYLOR, 1998). Night roosts are generally associ-
butional information is highly fragmented and far from
ated with feeding (BOWIE et al., 1999; SEAMARK &
complete. To date, most bat studies in Africa have
BOGDANOWICZ, 2002), while day roosts appear to function
focused on distribution, taxonomy, reproduction and to a
as resting sites.
lesser extent on diet, domiciles and echolocation. Limited
The main objectives of this study were to : 1) deter-
information is available on other biological aspects of
mine the age-specific survival of Nycteris thebaica in
African bats including longevity, survival and the behav-
Swaziland, and 2) investigate daytime roost-site selection
ioural and ecological aspects of roost-site selection. Some
in this species.
notable exceptions are VAN DER MERWE (1989) who
showed that Miniopterus schreibersii natalensis (A.
METHODS
Smith, 1834) could survive up to 13 years in South
Africa, LAVAL & LAVAL (1977), O'SHEA (1980) and HAP-
This study was conducted at Mlawula Nature Reserve
POLD & HAPPOLD (1990, 1996) who investigate roost-site
(26E 11'S; 31E 59'E, 160 m above sea level) in the
selection and other aspects of social behaviour in Pipist-
lowveld of northeastern Swaziland. Mlawula Nature
rellus nanus (Peters, 1852). Other species of Sub-Saharan
Reserve typically experiences hot, wet summers (October
Africa microchiropterans whose roosting behaviour have
to March) and cool, dry winters (May to August). Mean
been studied include Myotis bocagii (Peters, 1870) (BRO-
annual rainfall is approximately 500-600 mm, but rainfall
SSET, 1976), Coleura afra (Peters, 1852) (MCWILLIAM,
can vary dramatically between years.
1987), Lavia frons (E. Geoffroy, 1810) (WICKER & UHRIG,
1969) and Tadarida pumila (Cretzschmar, 1830-1831)
The road leading to the main entrance of the reserve is
(M
tarred and is situated within the reserve. It is approxi-
CWILLIAM, 1987).
mately 3 km in length and passing beneath this road are
The microchiropteran bat Nycteris thebaica E. Geof-
15 culverts 60-100 cm wide and up to 50 m long. Nycteris
froy, 1818 is widespread in Africa, and tolerates a wide
thebaica has been known to be using these culverts as
range of environmental conditions (SMITHERS, 1983),
daytime roosts for, at least, the past 15 years (MONADJEM,
roosting in caves, mine adits and various other hollow
2001). The bats roosting within these 15 culverts formed
sites (CHURCHILL et al., 1997; TAYLOR, 1998; TAYLOR,
the basis of this study. A small number of Nycteris theba-
2000). In Swaziland, it regularly roosts in road culverts
ica was also captured in culverts passing under the rail-
where adult females significantly outnumber adult males
way track within 10 km of the main study area. Bats were
(MONADJEM, 1998; MONADJEM, 2001). Females are
captured in the culverts by pushing a "shield" (a piece of
present in significant numbers throughout the year, but it
chipboard with the diameter of the culvert) through the

104
Ara Monadjem
culvert into a large sweep net placed over the entrance.
The sex ratio of juveniles did not deviate significantly
The other end of the culvert was blocked to prevent bats
from parity. In total, 196 female and 181 male juveniles
that had manoeuvred past the shield from leaving the cul-
were banded, with three unsexed individuals. Juvenile sex
vert. The process was repeated until all bats had been cap-
ratio also did not differ significantly between years (÷2 =
tured. In the first year of study this technique was not
4.359, degrees of freedom = 4, P > 0.05; Table 1).
fully developed and occasionally many bats escaped.
From August 1999, however, very few bats escaped. All
TABLE 1
captured bats were sexed, aged and were fitted with metal
Numbers of juveniles banded per year, including juvenile sex
bat-bands which were attached around the forearm. Three
ratio and productivity of adult females. Totals do not add up for
age classes were identified : juveniles, sub-adults and
the years 1999, 2001 and 2002 as single unsexed individuals
adults. Juveniles were defined as being dependent on their
were banded in these years.
parents and were easily identified by size and pelage,
which was greyer than that of the adults. Sub-adults were
Number of juveniles banded
Juvenile
identifiable by pelage colour (which was still greyer than
Offspring
sex ratio
that of the adults) only in February and probably March,
Year
per adult
(male/
Male
Female
Total
female
and represented individuals born in the previous breeding
female)
season i.e. November. By July the bats were more than 7
months of age, and could no longer be differentiated from
1998
13
6
19
2.0
0.8
older bats (M
1999
39
47
87
0.8
1.1
ONADJEM, 2001).
2000
53
55
108
1.0
1.0
Culverts were surveyed in July, October and December
2001
36
41
78
0.9
0.8
1998. From August 1999, culverts were surveyed four
2002
40
47
88
0.8
0.8
times per year in July/August, October, December/Janu-
ary and February. In 1998, only two culverts were sur-
veyed. From August 1999 all 15 culverts were surveyed,
Survivorship curves were similar in shape for male and
and all bats captured, during each survey.
female bats banded as juveniles (Fig. 2). However, male
survival was lower than that of females. Juvenile survival
Cluster analysis, conducted by the computer program
was lowest in the first six months, thereafter levelling off
"Primer" (CLARKE & GORLEY, 2001), was used to identify
and remaining similar throughout adult life. Approxi-
clusters of roost sites based on the adult males and
mately 15% of females and 10% of males banded as juve-
females utilizing them. Dendrograms were generated
niles in 1998 and 1999 survived to three years of age. Of
based on Bray-Curtis similarities computed on the
female bats banded as adults in July 1998 (n = 39), 23%
number of times each bat was recorded roosting in each
had survived to January 2003 (4.5 years). The corre-
culvert. Counts were square-rooted so as to down-weight
sponding value for males (n = 6) was zero. Of 28 male
the contributions of a few individuals recorded many
bats banded in 1998, only one (4%) was recaptured after
times in relation to individuals recorded just once.
four years. Adult bats captured in 1998 would have been
born at the latest in November 1997 making them over
RESULTS
five years old when recaptured in 2003.
A total of 799 bats have been banded since 1998 and
100
there have been 1835 recaptures. Of these, 380 were
90
banded as juveniles. Numbers of juveniles banded varied
80
between years, but the ratio of juveniles to adult females
70
did not differ significantly between the years (x2 = 3.930,
60
degrees of freedom = 4, P > 0.05; Table 1). Between 1998
50
and the end of 2000, adults formed a significant propor-
40
% survival
tion of new bats banded. From 2001 onwards, however, a
30
decreasing number of bats were first captured as adults
20
(Fig. 1). Juveniles comprised only 45% of all new bats in
10
the first three years of the study, but 57% in 2001 and
0
0
0.5
1
1.5
2
2.5
3
3.5
4
81% in 2002.
Age (years)
400
Fig. 2. ­ Survivorship
350
curves for male (square) and female (diamond) Nycteris the-
baica banded as juveniles.
300
250
Approximately a third of all juvenile females survived to
200
one year, while only a quarter of juvenile males survived to
150
Cumulative total
the same age (Table 2). However, this was not statistically
100
different (x2 = 1.301, degrees of freedom = 1, P > 0.05).
50
Survival to one year varied between years (males : x2 =
0
1998
1999
2000
2001
2002
2003
21.416, degrees of freedom = 3, P < 0.05; females : x2 =
Year
9.582, degrees of freedom = 3, P < 0.05), with highest sur-
Fig. 1. ­ Cumulative total
vival of females and males recorded in 1998 and 1999,
number of new (i.e. unbanded) adult Nycteris thebaica cap-
respectively. Survival was lowest for both sexes in 2001.

Survival and roost-selection in Nycteris thebaica
105
TABLE 2
Fig 3 (a)
Minimum estimates of first year survival of juvenile Nycteris
thebaica
.
Year
Proportion Proportion
Total
Overall
juvenile
juvenile
number
proportion
females
males
surviving
surviving
surviving
(n)
(n)
1998
0.50
0.15
5
0.26
1999
0.38
0.54
38
0.44
2000
0.38
0.19
31
0.29
2001
0.15
0.11
10
0.13
Mean ± SE
0.35 ± 0.07
0.25 ± 0.10
0.28 ± 0.06
Fig 3 (b)
Bats were recorded roosting in 13 of the 15 culverts,
with some culverts regularly supporting large numbers of
bats (Table 3). Adult bats did not utilize these day roost
sites randomly. Adult females tended to be captured in the
same culvert on consecutive surveys, occasionally being
recorded in neighbouring culverts. This is illustrated in
the cluster analysis presented in Fig. 3. Neighbouring
roosts cluster together in this analysis, demonstrating sim-
ilarity in the female "community" using these roosts. Six
major groupings are apparent, suggesting that females
Fig. 3. ­ Cluster analysis showing relationships between cul-
live in groups whose adult female composition is fairly
verts based on : a) female and b) male Nycteris thebaica roost-
stable (Fig. 3a). A similar pattern is shown by male bats,
ing in the 15 culverts. The numbers refer to culvert numbers
presented in Table 3. The six female groups are as follows :
however, neighbouring roosts do not necessarily cluster
group 1 (culvert 1), group 2 (culverts 3 & 5), group 3 (culverts
together (Fig. 3b), suggesting greater movement between
6, 7, 8 & 9), group 4 (culvert 10), group 5 (culverts 11, 12 &
nonadjacent culverts for males than for females.
13), group 6 (culverts 14 & 15).
TABLE 3
Total number, mean number and adult sex ratio (male :female) of bats recorded in each of the 15 culverts, and
rates of occupancy. Culverts 6 and 7 were surveyed 19 times each; all remaining roosts were surveyed 16 times
(see Methods).
Culverts
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Proportion of vis-
0.50
0
0.31
0
0.25
0.26
0.42
0.69
0.81
0.25
0.38
0.31
0.31
0.31
0.88
its occupied.
No. of bats.
48
0
65
0
92
216
210
345
373
33
19
13
16
41
364
No. of bats/visit
6
-
13
-
23
43
26
32
29
8
3
3
3
8
27
when occupied.
Adult sex ratio.
0.21
-
0.45
-
0.62
0.18
0.14
0.24
0.23
0.18
0.20
0.25
0.83
0.44
0.23
DISCUSSION
small sample size, which in turn was due to the fact that
only two out of 15 culverts were surveyed in that year.
The proportion of new (unbanded) adults decreased
The female-biased adult sex ratio (MONADJEM, 2001),
over the duration of the study, demonstrating that levels
therefore, must be a result of differences in survival and/
of immigration into the study population are low. This
or dispersal after weaning. Rates of recapture were higher
suggests that emigration from the study area may be cor-
for females than for males. Whether this is due to greater
respondingly low. Similar low levels of dispersal have
been reported for other species of microchiropteran bats
mortality among males or whether a greater proportion of
(O'DONNELL, 2000; SWIFT, 1998). Thus, the recapture
males disperse from natal roost sites is not known. How-
rates of banded bats in this study are thought to reflect
ever, the sex ratio of immigrant (unbanded) bats is not
survival.
different from that of resident (banded) bats
Juvenile sex ratio did not deviate from parity. The sex
(male :female = 0.41 and 0.47, respectively; 2 = 0.830, P
ratio of the 1998 cohort, though biased toward males, was
> 0.05), suggesting that the skewed sex ratio may be due
not statistically significant and probably a result of the
to differential survival rates. Greater female survival has

106
Ara Monadjem
also been reported for Chalinolobus tuberculatus
culverts. This was illustrated by the fact that a male Nyc-
(O'DONNELL, 2002) and Plecotus auritus (SWIFT, 1998).
teris thebaica originally banded at a different location
Nycteris thebaica is long-lived with relatively high
was recaptured at this study site, representing a distance
rates of adult survival. More than a fifth of the adult
of approximately 9 km.
females captured at the beginning of the study survived at
least five years. Comparable results for other species of
ACKNOWLEDGEMENTS
African bats are severely limited. Miniopterus schreiber-
sii natalensis
has been shown to survive at least 13 years
This is the 3rd Communication of the All Out Africa
but survival rates were not estimated (VAN DER MERWE,
Research Unit (www.all-out.org). I would like to thank all the
1989).
various students at the University of Swaziland who assisted
Mortality rates, in contrast, were high in the first year
with data collection. My sincere thanks to referees Drs. M. Hap-
pold and J. Fahr for greatly improving this paper.
after birth and varied between years. The high survival
rate of females born in 1998 may have been due to the
small number of juveniles sampled in that year. The
REFERENCES
apparently low survival of males and females born in
2001 is probably an artifact of methodology. Survival to
ALDRIDGE, H.D.J.N., M. OBRIST, H.G. MERRIAM & B. FENTON
one year was indicated not only by recaptures after
(1990). Roosting, vocalizations, and foraging by the African
exactly one year, but also subsequent recapture of indi-
bat, Nycteris thebaica. J. Mammal., 71 : 242-246.
viduals one-and-a-half, two or more years later. Thus an
BOWIE, R.C.K., D.S. JACOBS & P.J. TAYLOR (1999). Resource
individual recaptured for the first time after two years,
use by two morphologically similar insectivorous bats (Nyc-
teris thebaica
and Hipposideros caffer). S. Afr. J. Zool., 34 :
would have been registered as being alive after one year,
27-33.
even though it was not captured at that time. For this rea-
BROSSET, A. (1976). Social organization in the African bat,
son, the survival values for this cohort may increase with
Myotis bocagei. Zeitschrift fur Tierpsychologie, 42 : 50-56.
subsequent sampling. Excluding 2001 values, average
CHURCHILL, S., R. DRAPER & E. MARAIS (1997). Cave utilization
first year survival was 0.42 for females and 0.29 for
by Namibian bats : population, microclimate and roost selec-
males. This is slightly lower compared with estimates for
tion. S. Afr. J. Wildl. Res., 27 : 44-50.
Chalinolobus tuberculatus from New Zealand (O'DON-
CLARKE, K.R. & R.N. GORLEY (2001). Primer v5 : user manual.
NELL, 2002), but in the range of estimates for microchi-
Plymouth Marine Laboratory, Plymouth.
ropterans from the Northern Hemisphere such as Eptesi-
FENTON, M.B. (1983). Roosts used by the African bat, Sco-
cus fuscus (MILLS et al., 1975), and Myotis myotis (ZAHN,
tophilus leucogaster (Chiroptera : Vespertilionidae). Bio-
1999).
tropica, 15 : 129-132.
FENTON, M.B., R.M. BRIGHAM, A.M. MILLS & I.L. RAUTENBACH
ALDRIDGE et al., (1990) and CHURCHILL et al., (1997)
(1985). The roosting and foraging areas of Epomophorus
provided limited data on the roosting habits of Nycteris
wahlbergi (Pteropodidae) and Scotophilus viridis (Vespertil-
thebaica, while FENTON et al., (1987), based on a sample
ionidae) in Kruger National Park, South Africa. J. Mammal.,
of five adults, showed that Nycteris grandis may exhibit
66 : 461-468.
roost fidelity. In contrast, this paper is the first to report
FENTON, M.B., D.H.M. CUMMING, J.M. HUTTON & C.M.
on roost occupancy and fidelity for any Nycteridae based
SWANEPOEL (1987). Foraging and habitat use by Nycteris
on a large number of banded individuals surveyed over
grandis (Chiroptera : Nycteridae) in Zimbabwe. J. Zool.,
several breeding seasons. Female Nycteris thebaica
Lond., 211 : 709-716.
exhibited roost site fidelity returning to the same culvert
FENTON, M.B. & I.L. RAUTENBACH (1986). A comparison of the
or neighbouring set of culverts. This is despite the relative
roosting and foraging behaviours of three species of African
insectivorous bats (Rhinolophidae, Vespertilionidae, and
proximity of the fifteen culverts. Nycteris thebaica, there-
Molossidae). Can. J. Zool., 64 : 2860-2867.
fore, appears to be a sedentary species with females form-
HAPPOLD, D.C.D. & M. HAPPOLD (1990). The domiciles, repro-
ing groups of fairly stable composition.
duction, social organization and sex ratios of the Banana
The formation of groups of stable composition has also
Bat, Pipistrellus nanus (Chiroptera : Vespertilionidae) in
been observed in other African bats including Myotis
Malawi, Central Africa. Zeitschrift fur Saugetierkunde, 55 :
bocagii (B
145-160.
ROSSET, 1976), Coleura afra (MCWILLIAM,
1987), Tadarida pumila (M
HAPPOLD, D.C.D. & M. HAPPOLD (1996). The social organisa-
CWILLIAM, 1988) and in non-
African species such as Chalinolobus tuberculatus
tion and population dynamics of leaf-roosting banana bats,
Pipistrellus nanus (Chiroptera : Vespertilionidae), in
(O'DONNELL, 2002). Roost fidelity has been observed in
Malawi, east-central Africa. Mammalia, 60 : 517-544.
several African microchiropterans including Hipposi-
LAVAL, R.K. & M.L. LAVAL (1977). Reproduction and behav-
deros commersoni (VAUGHAN, 1977), Rhinolophus hide-
iour of the African Banana Bat, Pipistrellus nanus. J. Mam-
brandti and Tadarida midas (FENTON & RAUTENBACH,
mal., 58 : 403-410.
1986), Coleura afra (MCWILLIAM, 1987), Tadarida pum-
MCWILLIAM, A.N. (1987). The reproductive and social biology
ila (MCWILLIAM, 1988) and male Pipistrellus nanus
of Coleura afra in a seasonal environment. In : FENTON,
(HAPPOLD & HAPPOLD, 1996). In contrast, Scotophilus
M.B., P. RACEY & J.M.V. RAYNER (eds), Recent advances in
leucogaster (FENTON, 1983) and S. viridis/S. borbonicus
the study of bats. Cambridge University Press : 324-350.
(FENTON et al., 1985) regularly switch day roosts.
MCWILLIAM, A.N. (1988). Social organization of the bat Tadar-
ida (Chaerephon) pumila (Chiroptera : Molossidae) in
Male Nycteris thebaica exhibited less roost site fidelity
Ghana, West Africa. Ethology, 77 : 115-124.
than females. Movement between roosts was greater in
MILLS, R.W., G.W. BARRETT & M.P. FARRELL (1975). Population
males than females. Males, therefore, appeared not to be
dynamics of the big brown bat (Eptesicus fuscus) in south-
tied down to one particular culvert or group of adjacent
western Ohio. J. Mammal., 56 : 591-604.

Survival and roost-selection in Nycteris thebaica
107
MONADJEM, A. (1998). Mammals of Swaziland. The Conserva-
SWIFT, S.M. (1998). Long-eared bats. T & AD POYSER, Natural
tion Trust of Swaziland and Big Game Parks, Mbabane.
History, London.
MONADJEM, A. (2001). Sexual dimorphism, sex ratio and pre-
TAYLOR, P.J. (1998). The smaller mammals of KwaZulu-Natal.
liminary recapture rates of Nycteris thebaica (Nycteridae :
University of Natal Press, Pietermaritzburg.
Chiroptera) in Swaziland. Durban Mus. Novit, 26 : 49-52.
O'D
T
ONNELL, C.F.J. (2000). Cryptic local populations in a tem-
AYLOR, P.J. (2000). Bats of southern Africa. University of Natal
perate rainforest bat Chalinolobus tuberculatus in New Zea-
Press, Pietermaritzburg.
land. Anim. Conserv., 3 : 287-297.
VAN DER MERWE, M. (1989). Longevity in Schreibers' long-fin-
O'DONNELL, C.F.J. (2002). Timing of breeding, productivity and
gered bat. S. Afr. J. Wildl. Res., 19 : 87-89.
survival of long-tailed bats Chalinolobus tuberculatus
VAUGHAN, T.A. (1977). Foraging behaviour of the giant leaf-
(Chiroptera : Vespertilionidae) in cold-temperate rainforest
nosed bat (Hipposideros commersoni). East African Wildlife
in New Zealand. J. Zool., Lond., 257 : 311-323.
Journal, 15 : 237-249.
O'SHEA, T.J. (1980). Roosting, social organization and annual
cycle in a Kenya population of the bat Pipistrellus nanus.
WICKER, W. & D. UHRIG (1969). Verhalten und okologische
Zeitschrift fur Tierpsychologie, 53 : 171-195.
nische der gelbflugelfledermaus Lavia frons (Geoffroy)
S
(Chiroptera, Megadermatidae) Zeitschrift fur Tierpsycholo-
EAMARK, E.C.J. & W. BOGDANOWICZ (2002). Feeding ecology
of the common slit-faced bat (Nycteris thebaica) in Kwa-
gie, 26 : 726-736.
Zulu-Natal, South Africa. Acta Chiropterologica, 4 : 49-54.
ZAHN, A. (1999). Reproductive success, colony size and roost
SMITHERS (1983). The mammals of the southern African subre-
temperature in attic-dwelling bat Myotis myotis. J. Zool.,
gion. University of Pretoria, Pretoria.
(Lond.), 247 : 275-280.


Belg. J. Zool., 135 (supplement) : 109-112
December 2005
Effects of habitat fragmentation on diversity of small
mammals in Lulanda Forest in Mufindi, Tanzania

Alex W. Kisingo1, Christopher A. Sabuni2, Lisette Coiffait3, Becca Hayhow3 and Britt Larsen3
1 College of African Wildlife Management, Mweka, P.O. Box 3031, Moshi, Tanzania
2 SUA Pest Management Centre, P.O. Box 3010, Morogoro, Tanzania
3 University of Newcastle, UK
Corresponding author : e-mail : awk@mwekawildlife.org or csabuni@hotmail.com
ABSTRACT. The Lulanda forest cover a portion of the Udzungwa mountains in Mufindi district, Tanzania, ranging
from 1480 ­ 1640 meters above sea level. The forest consists of three forest patches dominated by Parinari excelsa
and a corridor between two of them that is being regenerated to a forest under the help of the Tanzania Forest Con-
servation Group (TFCG). A capture-mark-recapture study was carried out to document the small mammal species
found in Lulanda forest patches and corridor. There is a considerable difference in small mammal species composi-
tion between the montane forest and the corridor with a higher diversity in the corridor.
KEY WORDS : Small mammals, Habitat fragmentation, Forest patches and Corridor
INTRODUCTION
small mammals are important contributors to biodiversity
of woodland-savannah ecosystem in sub-Saharan Africa
and good ecological indicators (L
The Udzungwa Mountains are part of the Eastern Arc
INZEY & KESNER, 1997).
Mountains (EAM), a chain of isolated mountain groups
Although many researchers have worked on the overall
that run from Taita hills in Kenya to Udzungwa Moun-
biodiversity richness of the EAM, little is known on the
tains in southern Tanzania (LOVETT & CONGDON, 1990).
small mammal diversity of specific forest patches and
The mountains have been recognized as one of the 25-
corridors between them in the Udzungwa Mountains.
biodiversity hotspots in the world (MYERS et al., 2000).
This paper presents part of a wider study aimed at docu-
Rapid increase of the human population, acquired needs
menting the animal species found in the Lulanda forest
from forests such as farmland, timber, firewood, and
patches and corridor. Our interest was to observe the
medicinal plants cause an overall loss of forest habitat
effect of habitat fragmentation to the diversity of small
and fragmentation of the remaining area (RODGERS,
mammals and the role of corridor rehabilitation on sup-
1998). Anthropogenic alteration of habitat is therefore
porting the diversity of forest patches.
affecting whole ecosystems and biota, particularly forest
around the equator where hotspots are centred (MYERS et
al., 2000). When habitats such as forest undergo fragmen-
METHODS
tation, remnant patches of the habitat are increasingly iso-
lated in a matrix of altered and often heavily used lands
(G
Study area
ROOMBRIDGE, 1992). Habitat loss and fragmentation are
major threats to the viability of populations and have been
shown to be good predictors of extinction threats in biodi-
The Lulanda forest is located in the southern
versity hotspots (BROOKS et al., 1999, FERRERAS, 2001).
Udzungwa mountains (5km east of Mufindi Escarpment
Fragmentation of these habitats isolates also small mam-
East Forest Reserve) in two valleys on the edge of the
mal populations (BROOKS et al., 1999). For example, the
east-facing escarpment from 1480 ­ 1640 meters above
high degree of anthropogenic forest fragmentation in the
sea level. The forest consists of three forest patches and a
Taita Hills, also part of the Eastern Arc Mountains, most
corridor between two of them. Fufu forest patch has an
likely added to differential extinction (CORDEIRO, 1998).
area of 89.3 ha (approximately 1000m x 600m), Magwila
an area of 82.6 ha (approximately 1100m x 400m), and
In this study we define small mammals as non-flying
the corridor linking Fufu and Magwila is 54 hectares. The
mammals weighing less than 1kg when adult (BARNETT &
Tanzania Forest Conservation Group is regenerating the
DULTON, 1995). Small mammals provide food for avian
corridor by replanting the indigenous trees in the area that
and mammalian predators while, at the same time being
was formerly covered by farmland, and the replanted
important consumers of seeds and herbage. Therefore,
trees were six years old at the time of the study.

110
Alex W. Kisingo, Christopher A. Sabuni, Lisette Coiffait, Becca Hayhow and Britt Larsen
Fig. 1. ­ A map showing the location of Lulanda forest (after DOODY, 2002)
The forest patches are dominated by Parinari excelsa
minimized a possible edge effect of boundaries between
with swampy open areas in valley bottoms. Canopy is up
forest, corridor and adjacent cultivated land.
to 30m, intact in parts but generally much disturbed fol-
Two prebaiting nights were conducted before the com-
lowing extraction of timber species in the past (LOVETT &
mencement of trapping in order to reduce trap shyness.
CONGDON, 1990). All areas have undergone disturbances
There was a total of 120 bucket-pit fall trap-nights and
but there is no evidence that the forest patches have ever
240 Sherman trap-nights. A mixture of peanut butter and
been completely cleared, so the majority is primary for-
maize flour was used as bait. All traps were checked both
est. The surrounding habitat is farmland with the major
at dawn and dusk. However traps in the corridor were left
crop being maize. There is encroachment along the edges
closed during the day as it was felt that they were exposed
of the forest for cultivation, and the collection of building
to high temperatures.
poles, firewood and medicinal plants. A footpath through
For surveying small primates (bushbabies), suitable
the forest links the village with cultivated areas.
habitats containing vines and climbers were identified.
In order to study the effect of habitat fragmentation on
Three wire mesh traps (30 cm x 30 cm x 45 cm) were
the diversity of small mammals, trapping programs were
placed approximately 1.5m above ground level more than
carried out for these animals in two trapping sites in each
500m apart. Bananas and bamboo wine were used as bait;
study area (i.e. Fufu, Magwila and the corridor).
bananas were also smeared on surrounding branches and
vines. Sixty trap-nights were conducted in two areas
Trapping design
(Fufu and Magwila forest patches). The captured mam-
mals were fur clipped for capture-mark-recapture (BAR-
Each trap site (Fufu, Magwila and corridor) consisted
NETT & DUTTON, 1995).
of four bucket pit-fall lines running parallel 50m apart
For the purpose of this study we used species richness
following a methodology used by STANLEY et al., (1996).
given by the total number of species occurring in an area
Each line consisted of 10 buckets buried in soil so that the
and local diversity as expressed by the Shannon-Wiener
rim was flush with the ground trapping points were 5m
and Simpson indices (KREBS, 1989).
apart. A drift fence of clear plastic sheet of approximately
The Shannon Index (H) is given by,
50cm was erected bisecting each bucket along the length
of the line. Twenty medium Sherman traps (23 cm x 9.5
s
cm x 8 cm) were placed at each side of the fence at 5m
H = - (P )(lnP )
i
i
intervals. The lines were laid out in such away that they
i=1

Effect of habitat fragmentation on diversity of small mammals in Lulanda Forest in Mufindi, Tanzania
111
Where S= Number of species,
TABLE 1
P =Proportion of individuals of the total sample
Captures of small mammals in pitfall, by genus and site.
i
belonging to the ith species.
Genera
Corridor
Fufu
Magwila
The Simpson index of diversity (D) is equal to the
Dendromus
7
0
3
probability that two randomly picked organisms belong to
Crocidura
9
6
4
the same species. It is given by :
Mus
3
5
4
Praomys
0
2
2
s
Lophuromys
0
0
0
D = 1- (P )2
i
TABLE 2
i=1
Captures of small mammals in Sherman traps, by genus
where P = Proportion of individuals belonging to the
and site.
i
ith species in the community.
Genus
Corridor
Fufu forest
Magwila
The index of similarity between areas was calculated as
Dendromus
0
0
0
2z/(x+y), where x and y are the number of species occur-
Crocidura
12
0
0
ring in each patch and z the number of species occurring
Mus
21
1
0
in both patches.
Praomys
1
69
37
Lophuromys
4
10
8
Grammomys
10
0
0
RESULTS
Hylomyscus
0
5
0
Beamys
0
0
1
A total of 45 small mammals were caught in bucket pit
falls, 124 in Sherman traps and 8 in bushbaby traps
TABLE 3
(tables1-3). Mammals caught were from 10 genera (seven
Captures of small mammals in bushbaby
rodents, two insectivores and one primate). Exact species
traps, by genus and site.
identification was not possible since the animals were
released after trapping. The most commonly trapped
Genus
Fufu
Magwila
mammal in both forest patches was Praomys sp, with
about 49.5% of all individuals recorded. However, only
Galagoides
6
0
one individual of this species was trapped in the corridor.
Beamys
1
1
Mus sp. was the most encountered genus in the corridor
accounting for 42.1% of individuals trapped in the corri-
DISCUSSION
dor. Grammomys sp. and Crocidura sp. were also repre-
sented in higher numbers (36.8%) within the corridor;
Grammomys sp. was not recorded at all in either forest
The TFCG corridor regeneration program is expected
patch. Certain genera were caught only in the forest
to have a positive effect on the Lulanda forest reserve. At
patches, including Hylomyscus and Beamys sp.. Two
this stage of succession six years after replanting work
Beamys were caught in bushbaby traps (one in Fufu, one
started, the corridor already supports a diverse small
in Magwila) and one in Sherman trap in Magwila. As
mammal community. Although five mammal genera were
noted in the methodology, Sherman traps in the corridor
represented in both the forest and the corridor, the genera
were closed during the day; therefore diurnal species
composition of the corridor differed from that of the for-
were only caught in the forest patches. Six Grant's galago
est patches. Praomys appeared to be the dominant small
(Galagoides grantii) were trapped in the Fufu forest patch
mammal genus in both forest patches, in contrast to the
in the wire mesh traps and none in the Magwila forest
corridor where Praomys was only caught once and there
patch (however they were excluded in diversity indices
were no other apparent dominant genera. Several species
calculations since they are at the upper weight level of
were met in only one of the forest patches (e.g. Hylomy-
what should be considered small mammals.
scus sp., Galagoides grantii) and it is possible that frag-
mentation still blocked the migration route for this spe-
In terms of local species richness, Magwila was the
cies from Fufu to Magwila, e.g. because of the arboreal
highest in richness with seven species followed by both
life of the bushbabies.
Fufu and the corridor with six species each. The Shannon
Index of the three sites were : 2.145, 1.421 and 1.275 for
The Shannon indices suggested that the corridor is
the corridor, Fufu and Magwila respectively. The highest
more diverse followed by Fufu then Magwila forest
dominance/Simpson index was found in the corridor
patches. The inverse results between species richness and
(D=0.731) followed by Fufu forest patch (D=0.42). The
diversity is attributed to the fact that species diversity
lowest dominance index was recorded in the Magwila
considers both richness and evenness of species in a par-
forest patch (D=0.369). The obtained similarity indices
ticular area. The Simpson index of diversity (D) measures
were between corridor and Fufu 0.667, between the corri-
the distribution of the individuals among the species in a
dor and Magwila 0.769 and between Fufu and Magwila
community (MALIMBWI et al., 1998). Magwila, which had
0.769.
the highest number of species, is the least when consider-

112
Alex W. Kisingo, Christopher A. Sabuni, Lisette Coiffait, Becca Hayhow and Britt Larsen
ing species dominance; this is attributed to the imbalances
BROOKS, T.M., S.L. PIMM & J.O. OYUGI (1999). Time lag
in species distribution/species evenness.
between deforestation and bird extinction in tropical forest
fragments. Conservation Biology, 13 : 1140­1150.
The three sites show fairly high similarity indices
CORDEIRO, N.J. (1998). A preliminary survey of the Montana
among them, perhaps because of the relatively small area
avifauna of Mt. Nilo, East Usambaras, Tanzania. Scopus,
over which the study was carried out.
20 : 1-18.
DOODY, K. (2002). An assessment of a reforestation programme
Our findings suggest that the corridor is in the early
in the Southern Udzungwa Mountains, Tanzania. Tanzania
stages of succession when compared to the rather climax
Forest Conservation Group Technical Report 3.
communities of the forest patches. Similar studies needs
FERRERAS, P. (2001). Landscape structure and asymmetrical
to be carried out periodically in the study area and in
inter-patch connectivity in a metapopulation of endangered
other parts of the eastern arc mountains, paying attention
Iberian lynx. Biological Conservation, 100 : 125-136.
also to agricultural and settled areas surrounding the
GROOMBRIDGE, B. (ed) (1992). Global biodiversity assessment.
remaining forests, so as to come out with the effect of
Status of the Earth's Living Resources. Chapman and Hall,
London, 585 pp.
habitat fragmentation on the diversity of small mammals
KREBS, C.J. (1989). Ecological methodology. Harper Collins
in the entire archipelago.
Publishers, New York, 554 pp.
LINZEY, A.V. & M.H. KESNER (1997). Small Mammals of Wood-
land Savannah Ecosystem in Zimbabwe, Density and Habi-
ACKNOWLEDGEMENT
tat occupancy Patterns. Journal of Zoology, London 243 :
137-152.
The authors are most grateful to the University of Newcastle
LOVETT, J.C. & T.C.E. CONGDON (1990). Notes on Lulanda For-
Expedition Council for organizing and funding the expedition,
est, sourthern Udzungwa mountains. East Africa Natural
Tanzania Forest Conservation Group (TFCG) for providing
History Society Bulletin, 20 : 21.
equipments, accommodation, the village of Lulanda, Pest Man-
MALIMBWI, R.E., J. KIELAND-LUND, J. NDUWAMUNGU (1998).
agement Centre for providing trapping techniques. We are grate-
Species diversity and standing crop development in four
ful to Mr. E.K. Msyani of CAWM, Dr. L. Mulungu and Richard
miombo vegetation communities. In : CHAMSHAMA, S.A.O.
Odhiambo of SUA for reviewing this paper and also for their
(ed), Proceedings of the First Annual Forest Workshop. Fac-
valuable suggestions. Numerous other individuals and institu-
ulty of Forestry, SUA, Morogoro. pp. 201-212.
tions are responsible for the success of this paper. We thank
MYERS, N., R.A. MITTERMEIER, C.G. MITTERMEIER, G.A.B. DA
them all and hope we have not misinterpreted them.
FONSECA & J. KENT (2000). Biodiversity hotspots for conser-
vation priorities. Nature, 403 : 853-858.
RODGERS, W.A. (1998). An introduction to the conservation of
REFERENCES
the Eastern Arc Mountains. J.East Afr. Nat. Hist., 87 : 7-18.
STANLEY, W.T., S.M. GOODMAN & R. HUTTERER (1996). Notes
on the insectivores and elephant shrews of the Chome For-
BARNETT, A. & J. DUTTON (1995). Expedition Field
est, South Pare Mountains, Tanzania. (Mammalia : Insec-
Techniques : Small Mammals. Royal Geographic Society
tivora et Macroscelidea). Zool. Abh. (Dresden), 49 : 131-
(with the Institute of British Geographers), London, 126 pp.
147.

Belg. J. Zool., 135 (supplement) : 113-118
December 2005
Community structure and seasonal abundance of
rodents of maize farms in Southwestern Tanzania

Richard Odhiambo1, Rhodes Makundi1, Herwig Leirs2,3 and Ron Verhagen2
1 Sokoine University of Agriculture, Pest Management Centre, P. O. Box 3110, Morogoro, Tanzania
2 University of Antwerp (RUCA), Department of Biology, Groenenborgerlaan 171, B-2020, Antwerpen, Belgium
3 Danish Pest Infestation Laboratory, Danish Institute of Agricultural Sciences, Department of Integrated Pest Management,
Skovbrynet 14, DK-2800 Kongens Lyngby, Denmark
Corresponding author : Ron Verhagen, e-mail : ron.verhagen@ua.ac.be.
ABSTRACT. Community characteristics and seasonal abundance of rodents were investigated in a small-scale
maize field-fallow land mosaic in southwestern Tanzania between February 2001 and May 2002. During the study,
a total of 2568 rodents were captured in 9150 trapnights giving a 28% trap success. Also shrews of the genus Croci-
dura
(Soricidae) were recorded. Mastomys natalensis comprised the highest proportion of rodents caught account-
ing for 82.9% of all captures. Other rodent genera captured included : Tatera, Saccostomus, Graphiurus, and Steat-
omys
. Relative densities as measured by both trap success and the number of rodents per hectare, and biomass
varied between seasons with and without crop in the field but not between habitat types. The work reports seasonal-
ity in breeding for the two most commonly trapped species, Mastomys natalensis and Tatera leucogaster.
KEY WORDS : rodents, community structure, seasonal abundance, refuge habitat, Africa
INTRODUCTION
while grass and weed seeds provide supplementary food
(MWANJABE, 1993). Fallow land matrix has also been
Outbreaks of rodents over large areas have been
assumed to act as refuge for rodents in the maize field
reported in many areas in Africa; however, considerable
during unfavourable conditions. As such it is expected
damage to agriculture has also been reported in non-out-
that rodents that inhabit the maize fields will dissipate
break years (FIEDLER, 1988, 1994). In eastern Africa,
into the fallow land and return only during the attractive
especially in small-scale farms and marginal landscapes,
crop stages. This means that during the unfavourable peri-
field rodents are a very serious concern after drought and
ods, rodent densities will be higher in the surrounding
poor soils as a major constraint to improved yield of sta-
matrix than in the maize field and vice versa during the
ple crops. Rodent damage particularly to cereal crops
attractive periods. A sound understanding of community
remains a chronic problem among small-scale farmers in
dynamics in the farm lands and the surrounding matrix
this region. However, quantitative information on the
may allow to predict changes in rodent densities and com-
type and level of damage remains descriptive. Earlier
munity structure, which is of prime importance for the
reports of maize losses due to rodents shows damages of
development of species and/or community specific man-
up to 20% annually in Kenya (TAYLOR, 1968; also
agement strategies.
reviewed in OGUGE et al., 1997) and Ethiopia (GOOD-
In the present paper, we investigate community struc-
YEAR, 1976). TAYLOR (1968) reported that during rodent
ture and seasonal abundance of rodents of maize farms
outbreaks in Kenya, maize losses could be as high as 34
and their surrounding matrix in southwestern Tanzania.
to 100%.
We hypothesize that, rodents in the maize field move to
For Tanzania, the average annual yield loss of maize is
the surrounding matrix during the unfavourable crop
estimated to be around 5 to 15% (MAKUNDI et al., 1991).
stages and vice versa during the attractive stages leading
This corresponds to more than 400,000 tonnes of maize,
to a marked difference in densities in the two habitats at
equivalent to an amount that could feed 2.3 million peo-
the different crop stages.
ple for a whole year (LEIRS, 2003). Besides the usual
annual losses in Tanzania, irregular rodent outbreaks
occur during which damage to crops can increase to over
MATERIAL AND METHODS
80% (LEIRS et al., 1996; MWANJABE & LEIRS, 1997).
In Tanzania, agricultural fields are situated in a matrix
The study took place in Chang'ombe village of Chunya
of surrounding habitats which in smallholder settings is
district, which is located in southwestern Tanzania
most often fallow land. These habitats change seasonally
between 07° 58'S ­ 33° 18'E and 08° 46'S - 33° 18'E.
leading to a spatial component of the community and
The climate of the area is dry subhumid with annual rain-
population dynamics of organisms living in these fields
fall of about 900mm. Rainfall is unimodal with peak pre-
(LEIRS et al., 1997b). The fallow lands are said to provide
cipitation between November and March. Although rain-
a suitable ground for shelter and breeding for rodents
fall is variable between years, it is relatively reliable.

114
Richard Odhiambo, Rhodes Makundi, Herwig Leirs and Ron Verhagen
We selected two sites for the placement of our removal
and, 2) density estimates; rodent captures per given area.
grids. These sites were denoted as Chunya1 and Chunya2.
A chi-square (x2) test was used to inspect any differences
At each of the two removal grids, 75 snap traps and 75
in sex ratios. Biomass is based on the total weights of the
Sherman traps were used, giving a total of 150 trap sta-
different rodents captured and is expressed in relation to
tions. Of the 150 trap stations per grid, 50 trap stations
the studied area.
were placed within the target crop (maize) while the other
100 trap stations were placed in the matrix habitat. The
traps were placed in lines on which traps were spaced
10m apart. Within the target crop each of the 5 trap lines
RESULTS
(10 traps/line) were spaced 10m apart while the 5 trap
lines outside the target crop (20 traps/line) were spaced
Rodent species composition
40m apart up to 200 m from the edge of the crop. During
trapping sessions, snap traps and Sherman traps were
placed alternating in each trap line. The trap positions
A total of 2568 captures of five species of rodents were
were marked with painted bricks so that the locations
made in 9150 trap nights. Also shrews of the genus Croci-
could be used during subsequent trapping. The traps were
dura were caught during this study. The five species of
baited with a mixture of peanut butter and maize scrap.
rodents recorded were Mastomys natalensis (82.9%), Tat-
The two grids were about 3km apart. Animals were
era leucogaster (15.9%), Saccostomus campestris
trapped during the different crop stages : a) at least one
(0.8%), Graphiurus murinus (0.2%) and Steatomys sp.
month before planting b) planting/after planting c) seed-
(0.2%) (Table 1).
ling stage d) vegetative/middle of growth stage e) before
harvesting/ripe stage and f) at least one month after har-
80
vest. During these sessions, trapping was carried out for
Fallow
three consecutive days.
Maize field
70
Entire field
60
Trapped animals were identified, weighed, sexed and
measured. The body condition (visible injury, ectopara-
50
sites) and reproductive condition were also noted. In addi-
tion, the point of capture, date and weather conditions
40
were recorded. Processed specimens were preserved in
r
ap success
T
30
10% buffered formalin. Live-trapped animals were sacri-
%
ficed under diethyl ether before processing.
20
10
For purposes of description and analysis, rodent com-
munity is defined as all trappable rodent species occur-
0
.
.
.
.
.
v
v
ring in the study sites. Description of seasonal abundance
Veg
ar
ar
Veg
Veg
Harv.
and community composition is based on trapped individ-
Seedl.
Plant.
Plant.
Seedl.
A/H
A/H
B/Plant.
B/Plant.
uals. Rodent abundance is expressed as 1) percent trap-
Crop stage/season
ping success i.e. the proportion of captures relative to the
Fig. 1. ­ Trap success for the different crop stages/seasons in
number of traps set over a given period (TELFORD, 1989)
southwestern Tanzania.
TABLE 1
Rodent species composition in Chunya, southwestern Tanzania
Species
Chunya1
Chunya2
Totals
% contribution
Mastomys natalensis
946
1176
2122
82.9
Tatera leucogaster
300
103
403
15.9
Steatomys sp.
6
0
6
0.2
Graphiurus murinus
4
2
6
0.2
Saccostomus campestris
1
20
21
0.8
Totals
1257
1301
2558
100
Trap success
low land than for those in the maize field. However, the
differences were small, inconsistent and insignificant (t =
The mean trap success for Chunya1 and Chunya2 com-
-0.166, df = 22, P = 0.869). Trap success differed signifi-
bined was 28%. Trap success varied greatly throughout
the study being lowest in February/March during the veg-
cantly between seasons with and without crops on the
etative stage and highest in July/August, three months
field (F
= 20.99, p < 0.001), but not between both
5, 120
after harvesting and before planting (Fig. 1). In most trap-
habitat types and there was no interaction between season
ping sessions, trap success was higher for rats in the fal-
and habitat type.

Rodent community structure and abundance
115
Density patterns
60
The total rodent densities varied in the different crop
Mastomys
Tatera
50
seasons. Densities ranged between 4 animals per hectare
Other species
during the vegetative period of February/March and 54 ani-
40
mals per hectare just before land preparation and planting
in September/October (Fig. 2). This same trend was
30
recorded for animals caught both in the maize field and fal-
low land. However, relative densities were always higher
20
in maize compared with fallow. Densities in maize ranged
between 12 and 156 animals per hectare while those in the
10
fallow area were 3 to 41 animals per hectare. In Chunya1,
the overall densities ranged from 4 to 49 animals per hec-
0
tare. Fallow recorded lower densities (3-37/ha) than in the
r
v
r
v
Veg
a
a
Veg
Veg
Plant
Plant
Harv
maize (6-84/ha) (Figs 3a&b). In Chunya2, the overall rela-
Seedl
Seedl
A/H
A/H
B/plant
B/plant
tive density was 5-59 animals per hectare. Again here fal-
Crop stage/season
low recorded lower densities (4-45/ha) than maize (12-162/
ha) (Figs 3c&d). Relative densities differed significantly
Fig. 2. ­ Seasonal relative densities of the different rodent taxa
between seasons with and without crop on the field (F 5, 120
in maize fields in southwestern Tanzania.
= 18.03, p < 0.001) but not between both habitat types.
100
10
3b
40
3a
Mastomys
3a
Mastomys
Tatera
Tatera
80
Other species
80
Other species
tare 30
60
20
40
10
20
No. of animals per hec
0
0
.
.
.
.
.
t.
t.
t.
t.
.
.
.
rv.
edl.
eg.
edl.
arv
arv
edl.
eg.
eg.
edl.
Veg
an
an
an
an
Veg
Veg
V
V
V
Harv
Ha
Seedl.
Pl
Pl
Seedl.
Se
Plant.
Plant.
Se
Se
A/Harv
A/Harv
B/Pl
B/Pl
A/H
A/H
B/Plant.
B/Plant.
50
180
3c
3d
160
tare 40
140
120
30
per hec
100
mals
80
20
60
40
10
No. of ani
20
0
0
.
e
.
e
ng
ng
atativ
esting
atativ
esting
Planting
Planting
Planting
Planting
Planting
Planting
Planting
Planting
Planting
Planting
f
ter Harv
Seedli
f
ter Harv
Seedli
A
B/
B/
A/
Veg
Harv
A
B/
B/
A/
Veg
Harv
Crop stage/season
Crop stage/season
Fig. 3. ­ Seasonal relative densities of rodents
a) in Chunya 1 fallow land; b) in Chunya 1 maize farm; c) in Chunya 2 fallow land; d) in Chunya 2 maize farm
Biomass
mys, Mastomys natalensis formed the largest biomass on
The seasonal weight of the total number of rodents cap-
all trapping occasions accounting for as high as 86.2% of
tured in the 4.5 ha grids varied from 923 to 9246g (205-
the total biomass in April (Table 2). In most seasons, bio-
2055g/ha). Except for February 2001 (vegetative stage)
mass was higher for rodents in the fallow land than those
when Tatera recorded equal biomass with that of Masto-
in the maize crop. However, again here as with trap suc-

116
Richard Odhiambo, Rhodes Makundi, Herwig Leirs and Ron Verhagen
cess the differences were small, inconsistent and insignif-
field (F
= 24.98, p < 0.001), but not between both
4, 120
icant (t = -1.13, df = 32, P = 0.26). Biomass differed sig-
habitat types and there was no interaction between season
nificantly between seasons with and without crops on the
and habitat type.
TABLE 2
Total biomass (g/4.5ha) of rodents and other small mammals in Chunya (% contribution are given in parentheses).
Month
Mastomys
Tatera
Steatomys
Graphiurus
Saccostomus
Crocidura
All species
February 2001
1612 (43.5)
1970 (53.1)
63 (1.7)
38 (1.0)
0
24 (0.6)
3709
March 2001
1257 (62.9)
710 (35.5)
0
28 (1.4)
0
0
1999
July 2001
2945 (60.6)
1854 (38.1)
0
24 (0.5)
0
41 (0.8)
4862
August 2001
4915 (53.1)
4331 (46.8)
0
0
0
0
9246
September 2001
4887 (62.2)
2964 (37.7)
0
0
0
0
7851
October 2001
5637 (65.0)
3038 (35.0)
0
0
0
0
8675
November 2001
3975 (58.7)
2796 (41.3)
0
0
0
0
6771
December 2001
1959 (69.0)
799 (28.1)
62 (2.2)
0
0
19 (0.7)
2839
January 2002
1606 (58.5)
1141 (41.5)
0
0
32 (1.2)
0
2747
February 2002
951 (78.5)
238 (19.6)
0
0
0
0
1211
April 2002
796 (86.2)
127 (13.7)
0
0
0
0
923
May 2002
2804 (82.8)
539 (15.9)
0
0
0
42 (1.2)
3385
Reproduction
Reproductive activity was highly seasonal in M.
natalensis and T. leucogaster, which were the only spe-
Of the rodents trapped, 1177 (46.0%) were males, 1063
cies caught in appreciable numbers. This seasonality in
(41.6%) were females while 318 (12.4%) were unsexed.
breeding was similar in both habitat types. Females of M.
The sex ratios (male :female) of Tatera leucogaster
natalensis had a high proportion of females with perfo-
(170 :191), Saccostomus campestris (12 :9), Graphiurus
rated vagina around March during the vegetative stage of
murinus (2 :4) and Steatomys sp. (2 :4) were all of the
crop; this is followed by a long period of non-reproduc-
expected one-to-one (1 :1) ratio. In overall more males of
tive activity (Fig. 4a). The same reproductive pattern was
Mastomys natalensis (991 :855) were caught than females
witnessed in T. leucogaster (Fig. 4b). Females of M.
(x2 = 5.0097, p < 0.05) (Table 3). However, the seasonal
natalensis and T. leucogaster are pregnant (Figs 4c&d) or
sex ratio of M. natalensis was not significantly different
lactating (Figs 4e&f) around February, March and April
from the expected 1 :1 ratio.
thereafter reproduction is suspended.
120
4a
120
4b
Fallow
Maize
100
100
80
80
60
60
40
40
Perforated vagina
%
20
20
Percent perforated vagina
%
0
0
60
80
4c
4d
50
60
40
40
30
20
20
% Pregnant females 10
% Pregnant females
0
0

Rodent community structure and abundance
117
60
60
4e
4f
50
50
40
40
30
30
20
20
% Lactating females 10
% Lactating females 10
0
0
2/2001
3/2001
7/2001
8/2001
9/2001
1/2002
2/2002
4/2002
5/2002
2/2001
3/2001
7/2001
8/2001
9/2001
1/2002
2/2002
4/2002
5/2002
10/2001
11/2001
12/2001
10/2001
11/2001
12/2001
Month
Month
Fig. 4. ­ Reproductive
characteristic of M. natalensis and T. leucogaster in fallow land and maize field
a) Proportion of female M. natalensis with perforated vagina; b) Proportion of T. leucogaster with perforated vagina; c) Pro-
portion of visibly pregnant M. natalensis; d) Proportion of visibly pregnant T. leucogaster; e) Proportion of lactating M.
natalensis;
f) Proportion of lactating T. leucogaster.
TABLE 3
Sex ratio of rodent species caught in southwestern Tanzania (NS = not significant)
Species
Males
Females
2
Significance
Mastomys natalensis
991
855
5.0097
P < 0.05
Tatera leucogaster
170
191
0.6108
NS
Steatomys sp.
2
4
0.3333
NS
Graphiurus murinus
2
4
0.3333
NS
Saccostomus campestris
12
9
0.2143
NS
DISCUSSION
up to a thousand animals per hectare have been reported
in outbreak periods, with normal peaks of several hun-
On a seasonal (crop stage) scale, we found no differ-
dreds in studies involving Mastomys spp. TELFORD (1989)
ences in trap success, relative densities and biomass of
reported a density estimate of 1125 animals/ha while
rodents in fallow land and maize field. There were ani-
LEIRS (1995) reported densities of 900 animals/ha and
mals in the maize fields during the different crop seasons
CHRISTENSEN (1996) reported densities of 384 animals/ha
and their relative abundance varied seasonally just like
in Morogoro, Tanzania. Our densities of 3 ­ 162 animals/
that in the surrounding matrix. Likewise, we did not find
ha compare well with those obtained in studies in Kenya
any evidence of the fallow land acting as a refuge for
(ODHIAMBO & OGUGE, 2003) and Ethiopia (BEKELE &
rodents during the non-attractive crop stages nor did the
LEIRS, 1997; BEKELE, et al., 2003). These findings are
data report of animals leaving the fallow land to the maize
however not surprising given the fact that M. natalensis
field during the attractive stages. The same density pat-
comprised the highest percentage of rodents captured.
terns were recorded for both fallow land and maize field.
The population dynamics of this species is known to be
There is a large turnover in the densities of rodents across
influenced by both density dependent and density-inde-
time in both the maize field and fallow land. This type of
pendent factors occurring simultaneously (LEIRS et al.,
scenario has also been reported in a study of Mastomys
1997a). Moreover, its breeding characteristics are
natalensis in a maize field-fallow mosaic in Morogoro,
strongly dependent on the amount of rainfall (LEIRS et al.,
Tanzania (LEIRS et al., 1997b)
1989).
The lowest densities were recorded in February/March
during the vegetative period of the crop. Densities start
Breeding activity was clearly seasonal in M. natalensis
rising steadily in May and peak densities are recorded in
and T. leucogaster in our study with breeding females
September/October, two to three months before planting.
(visibly pregnant or lactating) occurring in April and
This density pattern appeared not to be related with the
May. This is consistent with literature findings from pop-
crop stage rather it could be explained by the fact that the
ulations of these or related species where reproductive
first pregnant females were observed by the end of Febru-
activity appears to be linked in some way to the pattern of
ary (vegetative stage) so that an increase in numbers is
rainfall in areas which have well-defined wet and dry sea-
not expected before the second half of April. Indeed num-
sons. Breeding commences a few months after the onset
bers start increasing by May. Our density estimates are
of the rains and cease during the dry season (e.g. TAYLOR
rather low compared to reported densities for Tanzania
& GREEN, 1976; DELANY & MONRO, 1986; LEIRS et al.,
and elsewhere (LEIRS, 1995 and references therein) where
1994; MONADJEM & PERRIN, 1997). In our study, the

118
Richard Odhiambo, Rhodes Makundi, Herwig Leirs and Ron Verhagen
peaks of pregnancies and lactations followed each other
LEIRS, H., R. VERHAGEN, W. VERHEYEN, P. MWANJABE & T.
closely with a short time lag.
MBISE (1996). Forecasting rodent outbreaks in Africa : an
ecological basis for Mastomys control in Tanzania. J. Appl.
Ecol.
, 33 : 937-943.
ACKNOWLEDGEMENTS
LEIRS, H., N. STENSETH, J.D. NICHOLS, J.E. HINES, R. VERHAGEN
& W. VERHEYEN (1997). Stochastic seasonality and non-lin-
The study was carried out in the framework of the STA-
ear dependent factors regulate population size in an African
PLERAT project financed by the European Union (ICA4-CT-
rodent. Nature, 389 : 176-180.
2000-30029). We appreciate the support of Prof. R. S. Mach-
LEIRS, H., R. VERHAGEN, C.A. SABUNI, P. MWANJABE & W. VER-
ang'u and the staff of Pest Management Centre, Sokoine Uni-
HEYEN (1997). Spatial dynamics of Mastomys natalensis in a
versity of Agriculture. We extend our thanks to Ramadhan Iddi,
field-fallow masaic in Tanzania. Belg. J. Zool., 127(1) : 29-
Khalid Kibwana, Omari Kibwana, Kevin Kessy and Claud Kav-
38.
ishe for their excellent field assistance.
LEIRS, H. (2003). Management of rodents in crops : the pied
piper and his orchestra. In : SINGLETON, G.R., L.A. HINDS,
REFERENCES
C.J. KREBS & D.M. SPRATT (eds), Rats, mice and people :
rodent biology and management
. ACIAR Monograph
No.96. Canberra, Australian Centre for International Agri-
BEKELE, A. & H. LEIRS (1997). Population ecology of rodents of
cultural Research, pp. 183-190.
maize fields and grassland in central Ethiopia. Belg. J. Zool.,
127(1) : 39-48.
MAKUNDI, R.H., T.J. MBISE, & B.S. KILOZO (1991). Observa-
tions on the role of rodents in crop losses in Tanzania and
BEKELE, A., H. LEIRS & R. VERHAGEN (2003). Composition of
control strategies. Beit. Trop. Land. Vet., 4 : 465-474.
rodents and damage estimates on maize farms at Ziway,
Ethiopia. In : SINGLETON, G.R., L.A. HINDS, C.J. KREBS &
MONADJEM, A. & M.R. PERRIN (1997). Population dynamics of
D.M. SPRATT (eds), Rats, mice and people : rodent biology
Lemniscomys rosalia (Muridae : Rodentia) in a Swaziland
and management. ACIAR Monograph No.96. Canberra,
grassland : effects of food and fire. S. Afr. J. Zool., 32(4) :
Australian Centre for International Agricultural Research,
129-135.
pp. 262-263.
MWANJABE, P.S. (1993). The role of weeds on population
CHRISTENSEN, J.T. (1996). Home range and abundance of Masto-
dynamics of Mastomys natalensis in Chunya (Lake Rukwa)
mys natalensis (Smith, 1834) in habitats affected by cultiva-
valley. In : MACHANG'U, R.S. (ed), Economic importance
tion. Afr. J. Ecol., 34 : 298-311.
and control of rodents in Tanzania. Workshop proceedings,
D
6-8 July 1992. Morogoro, Sokoine University of Agricul-
ELANY, M.J. & R.H. MONRO (1986). Population dynamics of
Arvicanthis niloticus (Rodentia : Muridae) in Kenya. J.
ture, pp. 34-42.
Zool., 209 : 85-103.
MWANJABE, P. & H. LEIRS (1997). An early warning system for
F
IPM-based rodent control in small holder farming systems in
IEDLER, L.A. (1988). Rodent problems in eastern Africa. In :
P
Tanzania. Belg. J. Zool., 127(1) : 49-58.
RAKASH, I. (ed), Rodent pest management. CRC Press, Inc.,
Boca Raton, pp. 35-65.
ODHIAMBO, C.O. & N.O. OGUGE (2003). Patterns of rodent pest
F
distribution in a maize cropping system in the Kenyan Rift
IEDLER, L.A. (1994). Rodent Pest Management in eastern
Africa. FAO plant production and protection. Bulletin 123,
Valley. In : SINGLETON, G.R., L.A. HINDS, C.J. KREBS &
Rome, Italy.
D.M. SPRATT (eds), Rats, mice and people : rodent biology
and management
. ACIAR Monograph No.96. Canberra,
GOODYEAR, J.J. (1976). Population fluctuations of the rat-like
Australian Centre for International Agricultural Research,
rodents of importance in agricultural fields in Kaffa prov-
pp. 217-219.
ince, Ethiopia. MA thesis, Bowling Green State University
pp. 72.
OGUGE, N.O., D. NDUNG'U & P. OKEMO (1997). Effects of neem
plant (Azadirachta indica Juss, Maliaceae) products on
LEIRS, H., W. VERHEYEN, M. MICHIELS, R. VERHAGEN & J.
maize grain consumption by three common rodent pests in
STUYCK (1989). The relation between rainfall and the breed-
Kenya. Belg. J. Zool., 127(1) : 129-135.
ing season of Mastomys natalensis (Smith, 1834) in
Morogoro, Tanzania. Ann. Soc. Roy. Zool. Belg., 119 : 59-64.
TAYLOR, K.D. (1968). An outbreak of rats in agricultural areas
L
of Kenya in 1962. E. Afr. Agric. For. J., 34 : 66-77.
EIRS, H., R. VERHAGEN & W. VERHEYEN (1994). The basis of
reproductive seasonality in Mastomys rats (Rodentia : Muri-
TAYLOR, K.D. & M.G. GREEN (1976). The influence of rainfall
dae) in Tanzania. J. Trop. Ecol., 10 : 55-66.
on diet and reproduction in four African rodent species. J.
L
Zool., 180 : 367-389.
EIRS, H. (1995). Population ecology of Mastomys natalensis
(Smith, 1834). Implication for rodent control in Africa. Bel-
TELFORD, S.R. Jr. (1989). Population biology of multimammate
gian Administration for Development Co-operation, Agri-
rat, Praomys (Mastomys) natalensis at Morogoro, Tanzania
culture Edition No. 35, 256pp.
1981-1985. Bull. Fl. State Mus. Biol. Sc., 34(6) : 249-288.

Belg. J. Zool., 135 (supplement) : 119-125
December 2005
The role of rodents and small carnivores in plague ende-
micity in Tanzania

Bukheti Kilonzo1, Julius Mhina2, Christopher Sabuni1 and Georgies Mgode1
1 Pest Management Centre, Sokoine University of Agriculture, P.O. Box 3110, Morogoro, Tanzania
2 National Institute for Medical Research, P.O. Box 4, Amani, Tanzania
Corresponding author : Bukheti Kilonzo, e-mail : kilonzo@suanet.ac.tz; kilonzo_bs@yahoo.com
ABSTRACT. Between 1974 and 2003, blood samples were collected from wild and commensal rodents, and wild
and domestic small carnivores in selected villages of seven districts in Tanzania that have experienced human
plague outbreaks and seven districts that have not experienced any outbreak of the disease. The samples were tested
for antibodies against Yersinia pestis Fraction I antigen, using passive haemagglutination (PHA) or ELISA tests. Of
the 3354 rodents and 558 small carnivores from the plague infected districts, 122 (3.6%) rodents (captured in
Mbulu and Lushoto districts) were plague positive; 29 (5.2%) small carnivores from Mbulu, Arumeru, Hai and
Lushoto districts were plague positive, 28 of these were domestic dogs (Canis familiaris). PCR tests showed that
17.5% of 211 rodents tested from Lushoto contained Y. pestis DNA. In the non-infected districts, 1545 rodents and
171 domestic dogs were tested. 11 (0.7%) of the rodents (captured in Monduli, Chunya and Masasi districts) were
plague-positive. In Masasi district, 10.4% (7/67) of the rodents and 43.6% (17/39) of the dogs were positive for
anti-Y. pestis IgG. It was concluded that wild and commensal rodents as well as wild and domestic small carnivores
play a potential role as reservoirs and/or carriers of sylvatic plague in Tanzania, and that the disease exists in areas
where human plague outbreaks have not occurred before. In order to update the distribution of the disease it is pro-
posed that further epidemiological surveillance activities are established.
KEY WORDS : Rodents, small carnivores, plague, passive haemagglutination, ELISA. PCR.
INTRODUCTION
Over the years, outbreaks of the disease have occurred
in various parts of the country and involved large num-
Plague has been endemic in Tanzania for more than a
bers of human cases and substantial case-fatality rates.
century. The first authentically recorded epidemic
During the past half century (1953 ­ 2003), a total of
occurred at Image, Iringa in 1886. At the time of this out-
8956 plague cases of whom 731 (8.2%) were fatal, were
break, however, it was noted that the local people were
reported from ten districts in the country. Since 1980,
quite familiar with the disease which was locally known
however, only three districts (Lushoto, Singida and
as "Chambafu/Shaambafu" and that they knew it was
Karatu) have experienced outbreaks of the disease, and
associated with rodents. It was also noted that communi-
involved 8298 and 646 (7.8%) reported cases and deaths,
ties, under the guidance of their leaders, were burning
respectively (KILONZO, 2003).
houses as a means of controlling rodents and fleas and
Prior to the studies reported in this paper, limited inves-
consequently controlling the disease (ROBERTS, 1935;
tigations were made to understand the species and ecol-
MSANGI, 1968). The second authentically recorded epi-
ogy of rodents involved in the epidemiology of plague in
demic occurred at Kiziba, Bukoba in 1897 (DAVIS et al.,
the country. HUBBARD (1973) incriminated many rodent
1968). Likewise, the local people were already familiar
species in Tanzania as suitable reservoirs of plague in
with the disease that was referred to as "Rubunga", and
view of their hosting of flea species known to be efficient
were isolating plague patients as a means of controlling
vectors of the disease elsewhere. Some observations
its spread. Based on available information, Yersinia pestis
made in plague ­ endemic areas in the country revealed
was isolated for the first time in Tanzania during this epi-
that Mastomys natalensis, was the most frequent natural
demic (DAVIS et al., 1968). The Kiziba focus is probably
reservoir of plague and that it played an important role in
the oldest in the country as plague was introduced to this
maintaining the disease as it is partly refractory to the
area from Uganda as far back as 1883 (CLYDE, 1962).
infection, and hence, it is not eradicated during plague
Since then, the disease spread and established itself in
epizootics (GUGGISBERG, 1966; HUBBARD, 1973). MSANGI
many parts of the country especially the Central, North-
(1968) demonstrated the presence of haemagglutination
eastern, Northern and South-western regions (Fig. 1). The
plague antibodies in 0.8% and 1.5% of clinically healthy
spread was facilitated by slave and ivory caravans that
M. natalensis and Arvicanthis abyssinicus, respectively.
mostly moved across the hinterland to the coast and
Many other wild rodent species including Tatera robusta,
through the Kilimanjaro region to Mombasa in Kenya.
Grammomys dolichurus; Rhabdomys pumilio and Oto-
Indeed, most established plague foci today are found
mys angoniensis have been suggested as suitable reser-
along the ancient slave and ivory trade routes (MSANGI,
voirs of the disease. This has been argued on the basis of
1968; KILONZO, 1981).
seropositive assessments in Kenya where the ecological

120
Bukheti Kilonzo, Julius Mhina, Christopher Sabuni and Georgies Mgode
and climatic features are similar to those in Tanzania, and
ilar flea species which are found on known rodent reser-
the fact that these rodent species are abundant in areas
voirs (DAVIS et al., 1968; HUBBARD, 1973; SIONGOK et al.,
where outbreaks of plague occur frequently and host sim-
1977).
Fig. 1. ­ Distribution of plague in Tanzania (1953-2003).
Many outbreaks of human plague in the country have
populations during the years 1978 ­ 1979 which
been associated with or preceded by large increases and/
prompted the use of zinc phosphide for their control in
or mortalities of rodents in the infected area. The Wahehe
view of the severe damage caused to agricultural crops
people in Iringa, for example, reportedly observed that
(MKAMI, 1980; KILONZO & MHINA, 1982).
remarkable rat mortalities were associated with "Cham-
bafu"
(plague) outbreaks long before the arrival of Ger-
The establishment of plague foci and distribution pat-
man administrators in 1884 (MSANGI, 1968). LURZ (1913)
terns in Tanzania has been based on outbreaks of the dis-
similarly reported large numbers and deaths of rodent
ease among human populations, rather than on substantia-
populations prior to the 1912 outbreak of plague in
tion of the disease among natural reservoirs in the
Rombo district. Likewise, the 1948 outbreak of the dis-
particular area. In the past, very limited investigations
ease in Iramba district which involved 312 and 178
(57.1%) recorded cases and deaths, respectively, was pre-
were carried out to substantiate natural reservoirs, sec-
ceded by large mortalities of rodents in late 1947 (A
ondary reservoirs and/or carriers of the disease. In order
NON-
to know its actual distribution, and hence be able to fore-
YMOUS, 1948). Similar population build-ups and plague
epizootics were reported prior to plague outbreaks in
cast outbreaks, adequate information on the reservoirs
Hanang (then Mbulu) and Same districts in 1951 and
and carriers as well as its endemicity level and the popu-
1964, respectively (ANONYMOUS, 1951 and 1964). The
lation densities of its efficient vectors is desirable. The
first outbreak of human plague in Lushoto district in
purpose of the present study was to partly fulfill this
April, 1980 was preceded by large increases of rodent
objective.

Rodents, small carnivores & plague in Tanzania
121
MATERIALS AND METHODS
study. The first category of districts (infected) comprised
Singida, Mbulu, Arumeru, Hai, Rombo, Same and Lush-
Time and areas of study :
oto. The second (un-infected) category of districts com-
These studies were conducted at different times of the
prised Masasi, Chunya, Igunga, Monduli, Muheza, Kil-
year, between 1974 and 2003. Seven districts that have
ombero and Morogoro-Rural (Fig. 2). At least two villages
experienced at least one recorded outbreak of human
in each district were selected for the surveys. In plague-
plague, and the same number of districts which have never
infected districts, the selected villages included the ones
recorded any outbreak of the disease, were selected for the
where the most recent outbreaks occurred.
Fig. 2. ­ Areas surveyed for plague in Tanzania.
Trapping, collection and processing
were sorted, counted and preserved in 70% ethanol for
of specimens :
their subsequent identification. Blood samples were left at
room temperature overnight for spontaneous separation
Sylvatic and commensal rodents were live-trapped
of serum with a minority of samples separated by centrif-
using Sherman, Chauvancy or Box traps baited with pea-
ugation. Sera were preserved at 0-4ºC while in the field
nut butter or roasted sardines. Traps were inspected in the
and at ­20ºC after returning to the laboratory. In a few
morning with captures removed and taken to a central
occasions, sera were preserved in liquid nitrogen.
processing location. Live-captured animals were anaes-
In Hai, Rombo, Arumeru and Mbulu districts, small
thetized with ether and brushed with ether-soaked cotton
wild carnivores were live-captured with steel cage traps
wool to kill its arthropod ectoparasites which were then
or killed by handgun. Venous blood was aseptically col-
removed by scrubbing the fur of the animal with a small
lected from the captured/shot animals and similarly proc-
brush. Each animal was bled from the heart using a dis-
essed. Collection and processing of blood from domestic
posable syringe and needle or from the orbital vein using
dogs and cats was effected after obtaining informed con-
capillary tubes and capped microtubes. Flea ectoparasites
sent from their owners.

122
Bukheti Kilonzo, Julius Mhina, Christopher Sabuni and Georgies Mgode
Testing for plague infection :
were also subjected to PCR tests and 17.5% of them con-
tained Y. pestis DNA (Table 1c). The seropositive rodents
A total of 4899 sera were tested against Yersinia pestis
were captured in Mbulu and Lushoto districts, where
Fraction I (FI) antigen, using the Passive haemagglutina-
tion (PHA) test and controlled by the Passive haemagglu-
recent and active human plague cases, respectively, have
tination inhibition (PHAI) test. Furthermore, 289 serum
been recorded. A total of 1545 rodents were evaluated in
samples from Lushoto and Masasi districts were also
the districts where outbreaks of the disease have not been
tested by the Enzyme Linked Immunosorbent Assay
reported. Of these animals, 15 (1.5%) were plague posi-
(ELISA) and Polymerase Chain Reaction (PCR) tech-
tive and had been captured in the Monduli, Chunya and
niques for detection of current and past infections.
Masasi districts (Table 2a).
RESULTS
A total of 729 small carnivores were examined from
six plague-infected and two un-infected districts (Table
A total of 3354 rodents were tested for anti-plague anti-
3). Of these, 47 (6.4%) were positive for plague antibod-
bodies in the seven districts with established plague
ies. The majority (95.7%) of the positive carnivores were
endemic foci. Of these, 122 (3.6%) were positive for spe-
domestic dogs from the Arumeru, Mbulu, and Lushoto
cific haemagglutination antibodies against Y. pestis (Table
districts (plague-infected), and the Masasi district (un-
1a). Of the rodents tested in Lushoto district, 222 were
infected). Other plague positive carnivores identified
also tested with the ELISA technique. Of these, 7.7% and
were one wild cat (Felis lybica) from the Hai district and
11.3% were positive for anti-plague IgG and IgM respec-
one domestic cat (Felis catus) from the Masasi district
tively (Table 1b). A total of 211 rodents from Lushoto
(Table 3)..
TABLE 1a
Species and infection rates of rodents in districts with previous records of plague outbreaks :
(1a) Results of PHA tests
Nos. tested and (%) positive in each district
Rodent species tested
Singida
Mbulu
Arumeru
Hai
Rombo
Same
Lushoto
Total
Rattus rattus
96 (0)
37(8.1)
12 (0)
118 (0)
37 (0))
131 (0)
804 (3.4)
1235
Mastomys natalensis
355 (0)
43 (9.3)
78 (0)
19 (0)
84 (0)
35 (0)
552 (4.2)
1166
Arvicanthis nairobae
7 (0)
3 (0)
20 (0)
2 (0)
39 (0)
-
292 (2.2)
363
Lophuromys sp.
-
3 (0)
-
3 (0)
-
-
133 (6.8)
139
Pelomys fallax
-
-
-
-
-
-
60 (20)
60
Grammomys dolichurus
-
1 (0)
-
-
-
-
55 (3.7)
56
Otomys spp.
-
-
-
-
-
8 (0)
124 (11.3)
132
Heliosciurus sp.
-
-
-
-
-
-
17 (0)
17
Praomys spp.
-
-
-
-
-
-
3 (0)
3
Tatera robusta
21 (0)
-
3 (0)
-
8 (0)
16 (0)
3 (0)
51
Lemniscomys striatus
-
26 (11.5)
-
12 (0)
1 (0)
-
-
39
Rattus norvegicus
-
1 (0)
-
-
-
-
-
1
Cricetomys gambianus
-
-
-
9 (0)
28 (0)
-
1 (0)
38
Rhabdomys pumilio
-
-
35 (0)
-
4 (0)
-
-
39
Tachyoryctes daemon
-
-
1 (0)
-
2 (0)
-
-
3
Acomys spinossismus
-
-
-
-
-
8 (0)
-
8
Aethomys spp
4 (0)
-
-
-
-
-
-
4
Total
483 (0)
114 (8.3)
149 (0)
163 (0)
203 (0)
198 (0)
2044 (5.5)
3354 (3.6)
TABLE 1b
(1b) Results of ELISA tests on rodent sera from Lushoto district
(Range of titres :IgG : 1 :4 - 1 :128; IgM : 1 :8 - 1 :512; Minimum specific titre = 1 :4)
Rodent species
No. tested for antibodies
No. &% positive for IgG
No. &% positive for IgM
Mastomys natalensis
142
15 (10.6)
13 (9.2)
Arvicanthis nairobae
25
6 (23.1)
7 (26.9)
Lophuromys sp
14
1 (7.7)
2 (15.4)
Rattus rattus
19
2 (10.5)
2 (10.5)
Grammomys dolichurus
14
-
-
Praomys spp.
6
-
-
Mus (L) minutoides
1
-
-
Petrodromus sp.
1
1 (100)
1 (100)
Total
222
17 (7.7)
25 (11.3)

Rodents, small carnivores & plague in Tanzania
123
TABLE 1c
(1c) : Yersinia pestis DNA in rodents captured in Lushoto district :
results of Polymerase Chain Reaction (PCR) tests (Minimum specific titre =Lowest
dilution of the test serum that produces positive reaction with specific Fraction I plague
antigen).
Rodent species
Number tested
Number positive
% positive
Mastomys natalensis
131
25
19.1
Arvicanthis nairobae
25
7
28.0
Lophuromys sp
14
1
7.1
Rattus rattus
19
3
15.7
Grammomys dolichurus
14
0
0
Praomys sp.
6
0
0
Mus (Leggada) minutoides
1
0
0
Petrodromus sp. (Elephant shrew)
1
1
100
Total
211
37
17.5
TABLE 2a
(2a) Species, numbers and infection rates of rodents and insectivores tested in districts with no records of
plague outbreaks. Figures in brackets refer to percentage of animals positive for plague. In Masasi, sera were
tested by ELISA technique for detection of antibodies, in other districts PHA tests were used.
Numbers tested and% infection in each district
Animal species tested
Masasi
Chunya
Igunga
Monduli
Muheza
Kilombero
Morogoro
Total
Rural
Rattus rattus
7 (0)
105 (2.9)
22 (0)
5 (0)
14 (0)
15 (0)
133 (0)
301
Mastomys natalensis
56 (12.5)
379 (0.8)
105 (0)
104 (1.9)
35 (0)
109 (0)
262 (0)
1050
Arvicanthis nairobae
-
6 (0
-
4 (0)
-
-
-
10
Aethomys sp.
2 (0)
-
-
-
-
-
-
2
Tatera robusta
2 (0)
17 (0)
-
7 (0)
6 (0)
-
1 (0)
33
Saccostomus campestris
-
2 (0)
-
-
-
-
-
2
Crocidura hirta
-
-
5 (0)
-
3 (0)
1 (0)
35 (0)
44
Lemniscomys griselda
-
-
-
-
-
-
3 (0)
3
Mus sp.
-
-
-
-
-
-
100 (0)
100
Total
67 (10.4)
509 (1.2)
132 (0)
120 (1.7)
58 (0)
125 (0)
534 (0)
1545 (1.5)
TABLE 2b
(2b) : Observation of Y. pestis F1 by ELISA tests of rodent sera collected in Masasi
district
Species
No. tested for F1
No. &% positive for F1
Mastomys natalensis
45
14 (55.5)
Rattus rattus
7
1 (14.3)
Tatera sp.
2
0 (0)
Aethomys sp.
2
0 (0)
Total
56
15 (26.8)
DISCUSSION AND CONCLUSIONS
that M. natalensis was the major reservoir of the disease
in Kenya and Tanzania and concluded that it was respon-
The present observations broadly indicated that many
sible for maintaining and passing the infection to the
species of rodents are suitable and serve as natural reser-
house rat, R. rattus, and to humans. Our observations are
voirs of plague in Tanzania. All these species could play
consistent with these reports; however, our data also indi-
an important role in the epidemiology of the disease.
cated that other field rodent species including Arvicanthis
These observations are consistent with those reported
nairobae, Lemniscomys striatus, Lophuromys spp., Pelo-
from Kenya (DAVIS et al., 1968). As the examined rodents
mys fallax, Grammomys dolichurus, Otomys spp. and
were clinically healthy when live-trapped, these animals
Rattus rattus could play similar roles in disease mainte-
are at least partly refractory, and hence, potentially capa-
nance. Current studies in the Lushoto plague focus sug-
ble of maintaining the disease enzootically for long peri-
gest close interaction and exchange of flea ectoparasites
ods. GUGGISBERG (1966) and HUBBARD (1973) suggested
between sylvatic and commensal rodents, thus facilitating

124
Bukheti Kilonzo, Julius Mhina, Christopher Sabuni and Georgies Mgode
TABLE 3
Species and infection rates of small carnivores in Tanzania. Dog sera from Masasi District were tested by
ELISA technique.
Animal species examined and% infected
Districts
Canis
Felis
Genetta
Civettictis
Crocuta
Felis
Otocyon
Ichneumia
Total
familiaris
catus
genetta
civetta
crocuta
lybica
megalotis
albicauda
Arumeru
35 (11.4)
-
-
-
-
-
-
2 (0)
37 (10.8)
Hai
39 (0)
-
2 (0)
-
-
2 (50)
-
-
43 (2.3)
Rombo
-
-
-
1 (0)
-
-
-
1 (0)
2 (0)
Mbulu
55 (3.6)
-
-
-
2 (0)
2 (0)
3 (0)
-
62 (3.2)
Singida
19 (0)
-
-
-
-
-
-
-
19 (0)
Lushoto
388 (5.7)
7 (0)
-
-
-
-
-
-
395 (5.6)
Chunya
129 (0)
-
-
-
-
-
-
-
129 (0)
Masasi
39 (43.6)
3 (33.3)
-
-
-
-
-
-
42 (42.9)
Total
704 (6.4)
10 (10)
2 (0)
1 (0)
2 (0)
4 (25)
3 (0)
3 (0)
729 (6.4)
transfer of the disease causative agents (MAKUNDI et al.,
such animals should, be involved in epidemiological stud-
2003).
ies aimed at establishing endemic foci of plague. Epide-
Our observations further indicated that only two of the
miological surveillance services for plague should be
seven plague-active foci districts (Lushoto and Mbulu)
established and regularly carried out in many districts;
demonstrated detectable haemagglutination antibodies.
especially those, which experience frequent outbreaks of
This was probably attributable to the fact that specimen
rodent populations. Such services will facilitate updating
collection in these districts was carried out during or soon
the distribution of natural foci of the disease in the coun-
after outbreaks of the disease, whereas in other districts
try, the forecasting of outbreaks and allowing the prompt
the study was done several years after the occurrence of
application of appropriate preventive measures. This
the last reported outbreak. These observations could sug-
improved information network will also enable research-
gest that the use of rodents alone is not enough for the
ers, extension personnel, community members and other
detection of plague endemicity over long periods of time.
people handling rodents in various parts of the country, to
The presence of both IgG and IgM immunoglobulins
take the necessary precautions.
among rodent populations in Lushoto district was an indi-
cation of past and current infections but does not suggest
ACKNOWLEDGEMENTS
how long ago the animals carrying IgG were infected.
Furthermore, our data showed that areas where out-
Our heartfelt gratitude is due to the Sokoine University of
breaks of human plague have not occurred before can also
Agriculture, the National Institute for Medical Research (Amani
harbour animal reservoirs of the disease. Surveys in Mon-
Centre) and the International Development Research Centre
duli, Chunya and Masasi districts were carried out at
(IDRC), for their financial and material support, without which
times of epidemics of rodent populations, and where
accomplishment of these studies would have been difficult. We
some rodent species were found to have been exposed to
also wish to appreciate the invaluable assistance offered by the
infection with Y. pestis. Considering the limited home
technical personnel of the Amani Medical Research Centre and
range of the infected rodents, it is expected that the ani-
the SUA Pest Management Centre during the course of imple-
mals contracted the disease locally, thus suggesting the
menting the studies, the brotherly co-operation of community
members and leaders in districts and villages where we collected
existence of endemic foci in these areas despite the
our data, and the SPMC Personal Secretary, Ms T. Sembua, for
absence of reported human plague outbreaks.
typing the manuscript of this paper.
The presence of specific anti-plague antibodies and/or
antigens in small carnivores (domestic dogs, domestic
REFERENCES
and wild cats) in districts where plague has occurred
before (Arumeru, Mbulu, Lushoto and Hai) and in a dis-
trict (Masasi) where the disease has not occurred before,
ANONYMOUS (1948). Annual Report of the Medical Department,
suggests involvement of these animals in the epidemiol-
1948. Government Printer, Dar es Salaam, pg. 25.
ogy of plague in Tanzania as observed elsewhere
ANONYMOUS (1951). Annual Report of the Medical Department,
(P
1951. Gouverment Printer, Dar es Salaam, pg. 12-13.
OLAND & BARNES, 1979 ; TAYLOR et al., 1981; ANONY-
ANONYMOUS (1964). Annual Report of the Medical Department,
MOUS, 1984). These animals are known to be efficient car-
riers of the disease and not more than 2% are killed by the
1964. Gouverment Printer, Dar es Salaam, pg. 19.
pathogen (K
ANONYMOUS (1984). Morbidity Weekly Report of 31 August,
ARIMI, 1974 ­ Pers. Comm.). Anti-plague
1984. Veterinary Plague : Medicine Digest, 10 : 27.
antibodies are also known to persist in small carnivores
for long periods and the animals can occasionally serve as
BARNES, A.M. (1990). Plague in the United States of America :
Present and Future. In : D
sources of human infection (R
AVIS,.L.R., R.E. MARSH ; (eds).
UST et al., 1971 ; BARNES,
Proceedings of the 14th Vertebrate Pest Conference, 43-46.
1990).
CLYDE, D.F. (1962). Plague and Polio (1948-1956). In : History
Despite the low sampling and infection rates of carni-
of the Medical Services of Tanganyika. Government Press,
vores, other than dogs, in our survey, the data suggest that
Dar es Salaam, 174-176.

Rodents, small carnivores & plague in Tanzania
125
DAVIS, D.H.S., R.B. HEISCH, D. MCNEILL & K.F. MEYER (1968).
TON, G.R., L.A. HINDS, C.J. KREBS & D.M. SPRATT (eds),
Serological survey of plague in rodents and other small
Rats, Mice and People; Australian Centre for International
mammals in Kenya. Trans. Roy. Soc. Trop. Med. Hyg., 62 :
Agricultural Research, Canberra : 20-24.
838-861.
MKAMI, D.L. (1980). Taarifa ya Tauni tarehe 12/4/80 hadi 18/9/
GUGGISBERG, C.A.W. (1966). Ecological studies of sylvan
80, Lushoto. Unpublished report from the District Health
rodent plague reservoirs in Kenya. WHO Vector Control /
Officer (Lushoto) to the District Medical Officer (Lushoto).
66.217, pg. 43. Proceedings at the WHO Seminar on
Kumb. Na. HQ/41/LGC/176 of 24 August, 1980.
Rodents and Rodent Ectoparasites, Geneva, October 1966.
MSANGI, A.S. (1968). Observations on the endemicity of plague
HUBBARD, A.C. (1973). Observations on the life histories and
in Tanzania. Ph.D. thesis, University of London.
behaviour of some small rodents from Tanzania. Zool. Afri-
MSANGI, A.S. (1975). The surveillance of rodent populations in
cana, 7 : 419.
East Africa in relation to plague endemicity. Dar es Salaam
KILONZO, B.S. (1981). The origin, dissemination and present
University Science Journal, 1 : 8-15.
status of plague in Tanzania. Dar. Med. Jour., 8 : 130-142.
P
K
OLAND, J.D. & A.M. BARNES (1979). Plague. CRC Handbook
ILONZO, B.S. & J.I.K. MHINA (1982). The first outbreak of
human plague in Lushoto district, North-east Tanzania.
series in zoonoses. STEELE, J.H. (ed), Inc. Boca Ralon, Flor-
Trans. Roy. Soc. Trop. Med. Hyg., 76 : 172-177.
ida, CRC Press.
K
ROBERTS, J.I. (1935). The endemicity of plague in East Africa.
ILONZO, B.S. (2003). The Ecology and Epidemiology of
Plague in Africa, with special reference to Tanzania. Pro-
E. Afr. Med. J., 12 : 200-219.
ceedings of an International Workshop on the Epidemiology
RUST, J.H., B.R. MILLER, M. BAHMANYAR, J.D, MARSHALL, JR
and Ecology of Q-fever, Tularemia and Plague. Fogerty
PURNAVEJA, D.C. CAVANAUGH & S.T. HLA (1971). The role
International Centre, National Institute of Health, Washing-
of domestic animals in the epidemiology of plague (iii).
ton, USA.
Antibody to Yersinia pestis in sera of dogs and cats. J. Infect.
LURZ, R. (1913). An epidemic of plague on Kilimanjaro in 1912.
Dis., 124 : 527-531.
Arch. Schiffs. Chd. Trop. Hyg., 17 : 593-599.
SIONGOK, T.K.A., A.M. NJAGI & S. MASABA (1977) : Another
MAKUNDI, R.H., B.S. KILONZO & A.P. MASSAWE (2003). Inter-
focus of sylvatic plague in Kenya. E. Afr. Med. J., 54 : 964.
action between rodent species in agro-forestry habitats in the
TAYLOR, P., D.H. GORDON & M. ISAACSON (1981). The status of
western Usambara Mountains, north-eastern Tanzania, and
plague in Zimbabwe. Ann. Trop. Med. Parasitol., 75 : 165-
its potential for plague transmission to humans. In : SINGLE-
173.


Belg. J. Zool., 135 (supplement) : 127-131
December 2005
The population structure of four rodent species from a
tropical region (Kisangani, D. R. Congo)

D. Amundala1, A. Bapeamoni1, W. Iyongo1, J. Kennis2, M. Gambalemoke1, N. Kadange1,
P.G.B. Katuala
1 and A. Dudu1
1 Laboratoire d'Ecologie et de Gestion des Ressources Animales (LEGERA), Faculté des Sciences, Université de Kisangani, BP
2012 Kisangani (R.D. Congo)
Corresponding author : E-mail : nicaisedrazo@yahoo.fr
2 Evolutionary Biology Group, University of Antwerp, Groenenborgerlaan 171, B-2020 Antwerpen, Belgium
Corresponding author : J. Kennis, e-mail : jan.kennis@ua.ac.be
ABSTRACT. This study summarizes the data on the population structure of Deomys ferrugineus, Hybomys lunaris
Lophuromys dudui on the mainland and Praomys jacksoni on the mainland as well as on islands on the Congo River.
All species at these three localities show a stable population structure without seasonal variation. Reproduction on
the mainland populations is probably continuous year-round, with subadult presence during the whole year. The
island populations of P. jacksoni have a different population structure from the mainland population, probably
caused by the periodic inundations. The sex-ratio is even for L. dudui and H. lunaris. For P. jacksoni and D. ferru-
gineus
however, more males were captured on the mainland. On the islands, the sex-ratio pattern is not clear cut and
differs between years. No clear seasonal variation in sex-ratio has been found in our populations.
KEY WORDS : population structure, sex ratio, Lophuromys, Praomys, Deomys, Hybomys.
INTRODUCTION
mainland, except for Lophuromys dudui, where data from
fallow lands near Kisangani city also exist and were
Since 1979, the Faculty of Science, University of
pooled together with data from Masako. At Masako, 300
Kisangani conducts studies on Rodents in the region of
Victor snap traps have been used from 1984 to 1986 to
Kisangani. During these studies, the structure of the
totalize 13300 trap nights during a total of 27 months. The
rodent population and their ecological distribution was
trapping effort was approximately evenly distributed dur-
recorded. This work is a synthesis on existing data about
ing this period at Masako. In 1984, two months at the end
the population structure of H. lunaris, D. ferrugineus, L.
of the wet season and one month of a `drier' season were
dudui and P. jacksoni from the region around Kisangani,
sampled. During 1997, 60 traps were used at Masako to
from 1984 to 2000.
totalize 780 trap nights. Sampling was only conducted
during one wet season from September to November. In
Kisangani city, 80 Victor traps were used with an
MATERIAL AND METHODS
unknown number of trap nights. Traps were baited with
palm nut pulp (Elaeis guineensis), cassava, salty fish or
Kisangani is located at 0°31' N, 25°11' E and lies
peanut butter.
between 360m and 460m above sea level. The area is situ-
Mbiye and Mafi Islands were prospected together
ated in an equatorial climate type of the A zone, accord-
fi
because of their vicinity. The study site was prospected in
ing to the classification of Köppen (DUVIGNEAUD, 1974).
1996 using 80 snap traps totalizing 2101 trap nights and
The climate is characterized by the lack of a prolonged
again in 2000 using 83 Sherman live traps (2075 trap
dry season. Precipitation is abundant with a monthly aver-
nights in total). At the Kungulu Island, 40 Sherman traps
age of 152 mm, but rainfall is irregularly distributed. The
were used totalizing 960 trap nights in 1999. At all these
mean relative humidity is 85% without much variation
localities, palm nut pulp was used as bait.
during the year. However, two short drier periods
(December - February and June - August) and two wet
The trapping took place on transect lines with trap sta-
periods (March - May and September - November) exist.
tions separated 10m from one another. The distance
between the traplines varies between 500m to 1000m. We
The survey is based on existing material collected from
trapped across different habitats, including primary for-
October 1984 to December 2000 from the city of Kisan-
ests, secondary forests, fallow lands, periodically inun-
gani and its surroundings (Masako Forest Reserve and the
dated secondary forests and along water courses. Within
islands Kungulu, Mbiye and Mafi on the Congo River).
the city of Kisangani, we trapped in different types of fal-
We used Victor snap traps and/or Sherman LFA live traps
low lands (older and younger fallow lands, concession of
on line transects for our rodent trappings.
the university, ...). The results from all habitats were
The results from October 1984 to December 1986 and
pooled together since data on trap placement regarding to
1997 are obtained at the Masako forest reserve on the
the habitat is not available.

128
Amundala, D., A. Bapeamoni, W. Iyongo, J. Kennis, M. Gambalemoke, N. Kadange, P.G.B. Katuala, and A. Dudu
The population structure is given for the two drier sea-
during this year. For the island populations surveyed dur-
sons and the two wetter seasons combined for the years
ing 1996, 1999 and 2000, P. jacksoni shows a high
1985, 1986 and 2000. 1984 data represent part of the wet
number of adults compared to subadults and juveniles.
and one month of the `drier' season (October ­ Decem-
Total number of captures for each species and season is
ber). Data for 1997 and 1999 regarding to the seasons are
not available.
given in Table 1.
Animals are categorized in groups based on reproduc-
TABLE 1
tive characters as follows :
Total number of captures per season (if available) for all species
·Juvenile : females with non-perforated vagina and
and years. DS = dry season, WS = wet season.
invisible teats or males with abdominal, non-developed
testes
Year
84
DS85
WS85
DS86
WS86
97
·Subadults : females with small visible nipples but a
non-perforated vagina or males with non-scrotal testes
H. lunaris
13
47
75
262
193
6
L. dudui
10
65
99
138
142
25
(internally developed but not externally visible)
D. ferrugineus
19
50
89
127
105
34
·Adults : females with large nipples and/or a perforated
P. jacksoni1
42
59
138
235
219
42
vagina or males with scrotal, externally visible testes
Year
DS96
WS96
99
DS00
WS00
Body weight classes per age class were also calculated.
Data were not available for the years 1984 and 1997. Data
P. jacksoni2
43
80
28
72
61
were grouped together per locality (for the years 1985 and
1 data from Masako on the mainland
1986 at Masako). There was no variation in body weights
2 data from the different islands on the Congo River
between these two years (not shown).
The sex-ratio per year and combined over all years was
calculated for each species at Masako on the mainland
Figures
(except for 1997 and for the sampled months during
Fig. 1
1984) and on the islands combined. Chi square tests were
used to test for significant sex ratio bias within each spe-
300
cies between years, as well as bias in the combined sex-
ratio for all species over all years were data exist for the
250
species. We expect an even sex ratio in small mammals
(HARDY, 1997), but sex ratio bias have already been found
200
in other small mammals (for instance for wild Mus minu-
toides individuals, K
150
RACKOW, 1997) and linked to ecolog-
ical factors, for instance in the rodent species Mastomys
natalensis
in Tanzania (K
100
ENNIS et al., submitted).
The material was fixated using 10% formaldehyde or
50
70% ethylic alcohol. The identification of the specimens
was performed in the "Laboratoire d'Ecologie et de Ges-
0
tion des Ressources Animales (LEGERA)" by comparing
84
DS85
WS85
DS86
WS86
97
morphometric and craniometric data with those proposed
by MEESTER & SETZER (1971), HOLLISTER (1916), HUT-
Fig. 2
TERER & DUDU (1991), HUTTERER & HAPPOLD (1983).
160
140
RESULTS
120
Population structure in relation
100
to the season
80
The population structure of H lunaris, D. ferrugineus, L.
60
dudui and P. jacksoni in and around Kisangani is similar. In
all species the population structure does not vary with the
40
season (Figs 1-4), based on seasonal data from the years
20
1985, 1986, 1996 and 2000. For the years 1997 and 1999,
information about the season of the captures is not availa-
0
ble. Moreover, there were no records of D. ferrugineus, H.
84
DS85
WS85
DS86
WS86
97
lunaris, and L. dudui in 1996, 1999 and 2000 at the pros-
Fig. 1. ­ Population structure for Hybomys lunaris. Black =
pected island sites, and only P. jacksoni was captured there.
juvenile, shaded = subadult and white = adult. DS = dry
season; WS = wet season. The Y-axis shows the number of
The amount of subadults is larger than the amount of
captures.
adults and juveniles from 1984 to 1986; except for L.
Fig. 2. ­ Population structure for Lophuromys dudui. Black =
dudui, where adults are more common. In 1997 however,
juvenile, shaded = subadult and white = adult. DS = dry
adults seem dominant for our four study species. For H.
season; WS = wet season. The Y-axis shows the number of
lunaris, subadults are not represented in our trappings
captures.

The population structure of four rodent species from a tropical region (Kisangani, D. R. Congo)
129
TABLE 2
Fig. 3
Comparison between body weight W (in g) for all study spe-
140
cies.Weight class data for 1884 and 1997 are not available. Data are
120
grouped per locality. Lower limits are given for juveniles and upper
limits are given for adults. For subadults, the range is given with
100
outer bounds. Number of captures per weight class is given between
80
brackets.
60
85-86
96
99
2000
Year
W (g)
W (g)
W (g)
W (g)
40
P. jacksoni
20
Juveniles
<21 (91)
<22 (12)<23 (2) <20 (8)
Subadults
20-41 (423) 21-31 (11)
23-28 (3)
20-26 (10)
0
Adults
>40 (137)
>30 (100)
>29 (23)
>24 (117)
84
DS85
WS85
DS86
WS86
97
L. dudui
Juveniles <41
(89)
Fig. 4
Subadults 40-51
(240)
250
Adults >50
(115)
D. ferrugineus
200
Juveniles
<41 (33)
Subadults 40-71
(312)
Adults >70
(26)
150
H. lunaris
Juveniles
<31 (116)
100
Subadults 30-51
(271)
Adults >50
(190)
50
0
Sex ­ ratio
84
DS85
WS85
DS86
WS86
97
Table 3 gives an overview of the sex-ratio calculations
Fig. 5
for the years 1985 and 1986 at Masako. Lophuromys
dudui
and Hybomys lunaris show an even sex-ratio year-
90
round at Masako, without seasonal variation. For Deomys
80
ferrugineus at Masako, the sex-ratio is significantly
70
biased towards males in both the `drier' and wet seasons
60
(X² = 5.5, p = 0.019). When combining the data from
50
1985 and 1986, the overall sex-ratio is also significantly
40
biased towards males (X² = 11.4, p = 0.001). For Praomys
30
jacksoni on the mainland (Masako), the overall sex-ratio
combined over the two study years is significantly biased
20
towards males (X² = 22.3, p < 0.001). Only during the wet
10
season of 1985, there was no significant difference in cap-
0
DS96
WS96
99
DS00
WS00
tures of males and females. All the other seasons during
the two-year period show a sex-ratio which is signifi-
Fig. 3. ­ Population structure for Deomys ferrugineus. Black =
juvenile, shaded= subadult and white = adult. DS = dry season;
cantly biased towards males (X² > 6.1, p < 0.014).
WS = wet season. The Y-axis shows the number of captures.
Data collected on the islands show a different sex-ratio
Fig. 4. ­ Population structure for the mainland populations of
pattern (Table 4). More males are caught than females
Praomys jacksoni (Masako, 1984-1986 and 1997). Black =
during the wet seasons of 1996 (x² = 5, p = 0.025) and the
juvenile, shaded = subadult and white = adult. DS = dry
`edrier' seasons of 2000 (x² = 4.76, p = 0.03). However,
season; WS = wet season. The Y-axis shows the number of
during the wet seasons of 2000, a significantly female
captures.
biased sex-ratio exists (x² = 3.99, p = 0.047). The overall
Fig. 5. ­ Population structure for the island populations of
sex-ratio, combined over all study years, does not show a
Praomys jacksoni (1996, 1999 and 2000). Black = juvenile,
shaded = subadult and white = adult. DS = dry season; WS =
significantly biased sex-ratio (x² = 2.39, p = 0.12) but
wet season. The Y-axis shows the number of captures.
shows a tendency towards more male captures (54.4% of
total captures).
Age class composition
DISCUSSION
Age structure
Body weight classes are given in Table 2. Upper limit
The regular presence of all age classes on the mainland
of the body weight is given for juveniles and lower limit
in and around Kisangani, indicates a stable population
is given for adults. For subadults, ranges are given with
structure, even in the different seasons. Rodent reproduc-
outer bounds. The number of captures per age class is also
tion is thus continuous in and around Kisangani. Only the
indicated.
year 1997 shows another pattern, but this is probably due

130
Amundala, D., A. Bapeamoni, W. Iyongo, J. Kennis, M. Gambalemoke, N. Kadange, P.G.B. Katuala, and A. Dudu
TABLE 3
Sex-ratio of H. lunaris, L dudui, D. ferrugineus and P. jacksoni collected at Masako during the years 1985 and
1986. Significant deviations from an equal sex-ratio are indicated with an asterisk. DS = dry season; WS = wet
season.
Total
DS 85
WS 85
DS 86
WS 86
N
%
N
%
N
%
N
%
N
%
H. lunaris
Female
327
51.1
29
40.8
33
40.2
164
53.6
101
55.8
Male
313
48.9
42
59.2
49
59.8
142
46.4
80
44.2
Total
640
71
82
306
181
²
0.31
2,4
3,1
1,5
2,4
L. dudui
Female
107
44
9
40.9
23
43.4
40
47.6
35
41.7
Male 136
56
13
59.1
30
56.6
44
52.4
49
58.3
Total
243
22
53
84
84
²
3.46
0,7
0,9
0,2
2,3
P. jacksoni
Female
298
41
41
35
75
47.2
94
41.8
88
39.6
Male
425
59
76
65
84
52.8
131
58.2
134
60.4
Total
723
117
159
225
222
²
22.3*
10,5*
0,5
6,1*
9,5*
D. ferrugineus
Female
168
41.6
36
46.1
39
43.8
53
39.2
40
38.5
Male
236
58.4
42
53.9
50
56.2
80
60.8
64
61.5
Total
404
78
89
133
104
²
11.4*
0,5
1,4
5,5*
5,5*
TABLE 4
Sex-ratio of Praomys jacksoni captured on the islands. Significant deviations from an equal sex-ratio are indi-
cated with an asterisk. DS = dry season; WS = wet season.
Total
DS96
WS96
99
DS00
WS00
N
%
N
%
N
%
N
%
N
%
N
%
Female
139
45.6
30
47.6
30
37.5
12
42.8
22
36
45
61.6
Male
166
54.4
33
52.4
50
62.5
16
57.2
39
64
28
38.4
Total
305
63
80
28
61
73
²
2.39
0.14
5.00*
0.57
4.76*
3.99*
to the low trapping effort that year and the short period of
SHEPPE (1972) and ANADU (1979) in Zambia, HUBERT
trapping (3 months in one wet season) compared to the
(1977) in Senegal, NEAL (1977) in Uganda, LEIRS et al.
year-round trapping conducted at Masako during the
(1989; 1990) in Tanzania. The population structure of
other years reported here.
savannah rodent populations thus changes considerably
DIETERLEN (1986) found however that the age compo-
during the year.
sition of rodents in primary forest changed significantly
Our data from the prospected islands show that the pop-
after reproduction. DUPLANTIER (1989) has also noted an
ulation structure of Praomys jacksoni on the islands differs
influx of immature animals during the month of April for
from the population structure on the mainland. Although
populations in tropical rainforest in Gabon, where the
trapping effort and total number of individuals is lower on
reproduction period extends from January to March. On
the islands than on the mainland, we think these results are
the other hand, HAPPOLD (1974, 1979) found that the pop-
real population effects because all the prospected years on
ulation structure of Praomys tullbergi, Hylomyscus stella,
the different islands show the same pattern. Adults form a
Thamnomys rutillans, Lophuromys sikapusi and Graphiu-
much larger part of the population as opposed to the popu-
rus sp. in the rainforest of Nigeria is composed of adults
lations on the mainland (Masako and Kisangani). It is pos-
and subadults all year round, suggesting that the rodent
sible that these island populations have a different popula-
populations there reproduce thus year-round. The contin-
tion structure because these islands are periodically
uous reproduction of rodents in and around Kisangani
inundated. The effect on the species composition is also
could be related to the absence of a real `dry' season. We
clear, only the arboreal P. jacksoni was captured. During
did not find any evidence to support the existence of well-
the inundation periods, animals have to find refuge in the
defined reproductive periods in L. dudui, P. jacksoni, D.
trees. The arboreal Praomys jacksoni adults certainly can
ferrugineus and H. lunaris.
survive periodical, short inundation that last at most one
In contrast to the rainforest, it is well known that most
week. Perhaps intraspecific competition for food during
savannah species show distinct reproductive periods as
the inundations has an impact on the survival of juveniles
documented by PIRLOT (1954) and DIETERLEN (1967) in
and subadults or perhaps the juveniles (and maybe even
Congo-Kinshasa, COETZEE (1965) in South Africa,
subadults) have difficulties finding suitable trees to climb

The population structure of four rodent species from a tropical region (Kisangani, D. R. Congo)
131
in and find refuge. Perhaps there is a competition for
CROSS, R.M. (1977). Population studies on Praomys tullbergi
`good' trees with abundant food between adults and sub-
(Thomas) and other rats of forest regions of Sierra-Leone.
adults/juveniles. Further studies are needed to clarify the
Rev. zool. Afri., 91(2) : 345-367.
causes of this population structure difference.
DIETERLEN, F. (1967). La dynamique des populations de
Muridés dans les forêts centrafricaines (Région du Kivu).
Chronique de l'IRSSAC, 2 (2-3) : 33-34.
Sex -ratio
DIETERLEN, F. (1986). Seasonnal reproduction and population
D
dynamics in Rodents of the African lowland rain forest.
UPLANTIER (1989) and HAPPOLD (1983) observed a
Cimbebasia, (A)8(1) : 1-7.
male-biased sex-ratio for some Muridae and Cricetidae,
DUPLANTIER, J.M. (1989). Les Rongeurs myomorphes forestiers
similar to our observations for the mainland population of
du Nord-Est du Gabon : Structure du peuplement, démogra-
P. jacksoni and D. ferrugineus. Studies conducted by
phie, domaines vitaux. Rev. Ecol., 44 : 329-346.
HAPPOLD (1977), CROSS (1977), RAHM (1967) have also
DUVIGNEAUD, P. (1974). La synthèse écologique : Populations,
found a preponderance of males for Praomys jacksoni,
Communautés, Ecosystèmes, Biosphère, Noosphère. Ed.
Hylomyscus stella and Deomys ferrugineus.
Doin, Paris VI, 286p.
For Hybomys lunaris and Lophuromys dudui however,
HAPPOLD, D.C.D. (1974). The small Rodents of Forest-savannah-
the sex-ratio is equal during the year.
farmland association near Ibadan, Nigeria, with observation
on reproduction biology. Rev. Zool. Afr., 88(4) : 814­836.
For these species, similar results were also obtained by
HAPPOLD, D.C.D. (1977). A population study on small rodents in
SHEPPE (1972; 1973), HAPPOLD (1977) and DUPLANTIER
the tropical rain forest of Nigeria. Terre & Vie, 31 : 285-458.
(1989).
HAPPOLD, D.C.D. (1979). Age structure of a population of Prao-
mys tullbergi (Muridae, Rodentia) in Nigeran rain forests.
The island populations of Praomys jacksoni show varia-
Terre & Vie, 33 : 253-274.
tions in sex-ratio depending on the season. This variation
HAPPOLD, D.C.D. (1983). Rodents and Lagomorphs.
is different in our two different study years. During the wet
Trop.Savanas, 16 : 363-400.
season of 1996, more males were captured at the islands
HARDY, I.C.W. (1997). Possible factors influencing vertebrate
Mbiye and Mafi. In 2000 however, more females were
sex ratios : an introductory overview. Applied Animal
captured during the wet season at island Kungulu. The
Behaviour Science, 51 : 217-241.
populations on these islands are not stable and the periodic
HOLLISTER, N. (1916). Shrews collected by the Congo Expedi-
inundations can also have an effect on the sex-ratio. At the
tion of The American Museum. Bull. Amer. Mus. Nat. Hist.,
mainland, the sex-ratio does not show seasonal variation.
35 : 663-680
These results could be influenced by the trapping efforts
HUBERT, B. (1977). Ecologie des populations de Rongeurs de
(although similar on the islands), by the different trapping
Bandia (Sénégal), en zone
methods and by differences in trappability between the
Sahelo-Soudanniene. Terre & Vie, 31 : 33-100.
sexes. Sex-specific differences in home ranges and disper-
HUTTERER, R. & A.M. DUDU (1991). Redescription of Crocidura
sal can possibly influence the trappability of the different
caliginea, a rare shrew from northeastern Zaïre (Mammalia,
Soricidae). J. Afr. Zool., 104 : 305-311.
sexes. HUBERT (1977) found for instance that in savannas,
rodent home ranges of males and females increase during
HUTTERER, R. & D.C.D. HAPPOLD (1983). The Shrews of
Nigeria, genus Sylvisorex from Rwanda and Zaïre (Insec-
the reproduction period. HAPPOLD (1983) noted that the
tivora, Soricidae). Z. Säugetierk., 50 : 266-271.
home range of Muridae and Cricetidae varies from 100 to
KRACKOW, S. (1997). Maternal investment, sex-differential
300 m² and does not differ much according to the sex.
prospects, and the sex ratio in wild house mice, Behav. Ecol.
DUPLANTIER (1989) however, found that the average home
Sociobiol., 41 : 435-443.
range of Muridae can measure more than 1500 m² for
LEIRS, H., W. VERHEYEN, M. MICHIELS, R. VERHAGEN & J.
females whereas for males the home range could be even
STUYCK (1989). The relation between rainfall and the breed-
larger, especially in species like Praomys tullbergi, Hybo-
ing season of Mastomys natalensis (Smith, 1834) in
mys univittatus and Deomys ferrugineus. He also found
Morogoro, Tanzania, Ann. Soc. r. Zool. Belg., 119 (1) : 59-61.
that the home range is inversely proportional to the density
LEIRS, H., J. STUYCK, R. VERHAGEN, W. VERHEYEN (1990). Sea-
of the rodent population. If home ranges are larger for one
sonal variation in growth of Mastomys natalensis (Rodentia,
sex, this can produce differences in trappability because of
Muridae) in Morogoro, Tanzania. Afr. J Ecol., 28 : 325-341.
the higher mobility of one sex. Trappability can also be
MEESTER, J. & H.W. SETZER (1971). The Mammals of Africa :
influenced by pregnancy and the care for offspring (by the
an Identification Manual. Smithsonian Institution Press
females, males or both). No data on sex-specific trappabil-
Washington.
ity and sex-specific mobility exist but capture-mark-recap-
NEAL, B.R. (1977). Reproduction of the multimammate rat
Praomys (Mastomys) natalensis (Smith) in Uganda. Z.
ture studies and population genetic studies will be under-
Säugetierk., 42 : 221-231.
taken in the near future.
PIRLOT, P.L. (1954). Pourcentage de jeunes et périodes de repro-
duction chez quelques rongeurs du Congo Belge. Ann. Mus.
REFERENCES
Congo, Tervuren, Zool. 1 : 41-46.
RAHM, U. (1967). Les Muridées des environs du lac Kivu et des
régions voisines de l'Afrique centrale et leur écologie. Rev.
ANADU, P.A. (1979). A study of bait selection in small African
Suisse Zool., 74 : 439-520.
Rodents. Acta Theriol., 24 : 501-511.
SHEPPE, W.A. (1972). The annual cycle of small mammal popu-
COETZEE, C.G. (1965). The breeding season of the multimmate
lations of a Zambian floodplain. J. Mamm., 53 (3) : 445-460.
mouse Praomys (Mastomys) natalensis (A. Smith) in the
SHEPPE, W.A. (1973). Notes on Zambian Rodents and Shrews.
transvaal highveld. Zool. Afri., 1(1) : 29-39.
Puku, 7 : 167-190.


Belg. J. Zool., 135 (supplement) : 133-140
December 2005
Preliminary data on the biodiversity of rodents and
insectivores (Mammalia) in the periphery of Kisangani
(D. R. Congo)

I. Mukinzi1, P.G.B. Katuala1, J. Kennis2, M. Gambalemoke1, N. Kadange1, A.M. Dudu1, M.
Colyn
3 and R. Hutterer4
1 Laboratoire d'Ecologie et de Gestion des Ressources animales, Faculté des Sciences, Université de Kisangani, R.D. Congo
2 Evolutionary Biology Group, University of Antwerp, Groenenborgerlaan 171, B-2020 Antwerpen, Belgium
3 Laboratoire d'Ethologie, Evolution et Ecologie Station Biologique, CNRS-UMR, 6552 Université de Rennes I, F-35380 Paim-
pont, France
4 Section of Mammalogy, Zoologisches Forschungsmuseum Alexander Koenig, Adenauerallee 160, D-53113 Bonn, Germany
Corresponding author : I. Mukinzi, e-mail : hjcmukinzi@yahoo.fr, dudakaibe@yahoo.fr
ABSTRACT. This study presents the species diversity of rodents and insectivores s.l. as observed in the forests
around Kisangani (D.R. Congo) between 1979 and 2003. 7736 specimens were collected using Victor snap traps,
Sherman live traps and pitfalls. In total, 49 small mammals species (36 rodents and 18 shrews) were identified of
which 42 were actually captured and 7 more were observed around Kisangani.
The number of species varied between habitats : 16 species were found (12 rodents, 4 shrews) in primary forest, 36
species (21 rodents, 15 shrews) in secondary forest, 31 species (20 rodents, 11 shrews) in fallow lands and 16 species
(15 rodents and 1 shrew) in wetlands. On the right bank of the Congo River, 40 species were collected against 19 on
the islands and 15 on the left bank of the Congo River. At this stage, the right bank seems to be more diverse as far as
the small mammal fauna is concerned, but more studies on the left bank need to be conducted.
KEY WORDS : biodiversity, Rodentia, Insectivora, Kisangani, Democratic Republic of Congo.
INTRODUCTION
This work is a synthesis of the main results from the
study on the diversity of small mammals around Kisan-
In the Democratic Republic of Congo, small mammal
gani in different types of forests and the derived, anthro-
phylogeny and zoogeography are thoroughly studied in
pogenic habitats. We will establish the specific richness
the eastern national parks and their peripheral zones, with
and distribution of rodents and shrews in all habitats by
syntheses by HOLLISTER (1916), HATT (1940) and
combining the available data from 1979 to February
SCHOUTEDEN (1948).
2003.
Ecological studies of small mammals in the Demo-
cratic Republic of the Congo (DRC) forest zones are rare
however. According to C
MATERIAL AND METHODS
OLYN (1986), most of the rain
forest region between the Congo River and the Rift Val-
ley, although recognized as containing some endemic spe-
Study areas
cies, remains unstudied. DUDU (1991) also noted that the
small mammal species from eastern Kivu (Albertine Rift)
The forest zones studied lie within a radius of about 50
was studied much better than that of other regions of the
km around the city of Kisangani (0°31'N, 25° 11'E, alti-
DRC, more particularly the lowland rainforests.
tude 396-425 m above sea level). Previously covered by
In Kisangani, small mammal studies (shrews and
primary rainforest, this area contains at present different
rodents) started in 1979 in order to determine their spe-
forest ecosystems (primary and secondary forests) on the
cific diversity and ecology. Some results were already
two banks of the Congo River and its islands and is also
published by DUDU et al. (1985, 1997, 2005); COLYN &
characterised by peri-urban degraded areas (fallow lands
DUDU (1986); DUDU & GEVAERTS (1986, 1987); DUDU
and fields). It includes localities that, according to the dis-
(1989), HUTTERER & DUDU (1990); KADANGE et al.
tribution stated by COLYN (1991), are part of the East
(1998). Data on other collections are still unpublished, for
Central faunistic zones (islands and right bank of the
instance the shrew collections of which specimens remain
Congo River) and South Central faunistic zone (left
to a large extent unidentified.
bank).

134
I. Mukinzi, P.G.B. Katuala, J. Kennis, M. Gambalemoke, N. Kadange, A.M. Dudu, M. Colyn and R. Hutterer
1
7
3
4
6
9
5
8
10
2
Fig. 1. ­ Map of the Kisangani area (after H. Gevaerts) indicating the different study sites. 1 : Masako forest reserve; 2 :
Yoko forest reserve; 3 : Zoo Kisangani; 4 : Linoko; 5 : Lula; 6 : Kisangani town; 7 : Kungulu island; 8 : Mbiye island;
9 : Mafi island; 10 : Tundulu island. Inlay : map of D.R. Congo showing the position of Kisangani.
The main study sites (Fig. 1) are the protected areas
Cyathogyna viridis and Pycnocoma insularis respectively
(Masako, Yoko, forest of the Kisangani Zoo), islands
on Mbiye and Kungulu islands (MOSANGO, 1991;
within the Congo River (Mbiye, Mafi, Kungulu and Tun-
NSHIMBA, unpublished data).
dulu) and the surroundings of Kisangani city (Kabondo,
Grand-seminaire, Linoko, concession of the Science Fac-
Secondary forest :
ulty, Plateau medical, Plateau Boyoma, Kikongo and
finally Lula on the left bank of the Congo River).
The secondary forests are very diverse, comprised of a
mixture of trees also occurring in old fallow lands and
Masako Forest Reserve (2.105 ha) is located 15 Km
primary forest. On the right bank of the river, it is charac-
north-east of Kisangani, on the old Buta's road. One third
terised by Zanthoxylum gilletii, Cynometra hankei,
of the reserve is occupied by primary forest; the remain-
Peterstiantus macrocarpus, Musanga cecropioides, Ter-
der consists of old-growth secondary forests, fallow lands
minalia superba, etc., and on the left bank by Scorodo-
and crops. The Yoko Forest Reserve (6.975 ha) lies south
phloeus zenkeri, Albizia adiantifolia, Uapaca guineensis,
of Kisangani, on the left bank of Congo River, between
Cynometra alexandrii, Panda oleoza, Musanga cecropio-
21 and 38 km, towards Ubundu. It consists of young and
ides, etc. (LOMBA & NDJELE, 1998)
old-growth secondary forests. The Zoo of Kisangani (84
ha), an artificial forest, is located on the right bank of
Wetland forest :
Tshopo river, 4 km from the city, on the road towards
Buta. The islands Mbiye (5600 ha), Tundulu (76 ha) and
The wetland forests are constantly or periodically inun-
Mafi (20 ha) are located upstream on the Congo River to
dated and this habitat is composed out of various trees
the south-east of Kisangani. Only Kungulu island (100
depending on whether they are located in the forest or on
ha), at the convergence of the Lindi and the Congo rivers,
banks of watercourses. They include : Uapaca guineen-
lies downstream from Kisangani city. All islands are cov-
sis, Uapaca heudolotii, Mitragyna stipulosa, Trichilia
ered by primary forests, secondary forests, fallow lands
retusa, Coelocaryon botryoides, Alchornea cordifolia,
and wetland forests.
Costus lucanisianus, etc.
Primary forest :
Fallow lands :
This type of habitat is essentially composed of Gilber-
Formed essentially by herbaceous groupings consisting
tiodendron dewevrei and Scaphopethalum thonneri as
of Panicum maximum, Pennisetum purpureum, P. polys-
undergrowth, whereas for the wetter areas on the different
tachyon, Spermacoce latifolia and of shrub associations
islands, it is heterogeneous and semi deciduous with Gil-
of Cnestis ferruginea, Craterispemum cerinanthum, Afro-
bertiodendron dewevrei and Coelocaryon bothryoides on
momum laurentii and Costus lucanisianus, Triumpheta
Mbiye and Piptadeniastrum africanum and Celstis mild-
cordifolia and Selaginella myosurus. (NYAKABWA, 1982
braedtii on Kungulu. The undergrowth is dominated by
and BAELONGANDI, unpublished data)

Rodents and insectivores of Kisangani
135
Kisangani region within the equatorial climate zone of
May ­ August 1998;
the Afi type according to the Köppen classification. Pre-
February, March and July 1999;
cipitation is abundant year-round but not uniformly distrib-
February 2003.
uted, with a monthly average around 152mm (1970-1990).
Tundulu :
Even during the driest month there is more than 60mm
December 1980 to April 1981;
rainfall and a relative humidity averaging 85%. Therefore
May to July 1994
there is no real dry season, but two `drier' seasons with
weak precipitation exist : December - February and June-
Mbiye :
August with a minimum in January. Two wet seasons exist
October 1982;
with heavy precipitation : March-May and September-
December 1994 to September 1996;
November with maxima in April and in October.
January to December 2000
February 2002.
Capture-removal studies were carried out in different
biotopes with the use of various trapping techniques, such
Kisangani city :
as pitfalls, capture-removal grids and capture-removal
August 1993 to July 1994;
lines with different trap systems. Mostly the sampling
August to November 1997;
was conducted during one-year periods, with captures
June 1998 to February 1999;
each month. The sample years were not always consecu-
May 1994 to August 1994
tive. In rare cases, the sampling period was shorter (3 to 6
Zoo :
months). Specimens were sampled using removal trap-
January to June 1980;
ping in lines and in grids, with Victor Rat traps (175 x 85
June 1985 to May 1986;
mm) and Sherman LFA live traps (76 x 89 x 229 mm),
December1995 to October 1996
both baited with palm nut, cassava bread or cassava car-
Yoko :
rot. Traps were placed at 10m distance; the distance
June 1995 to May 1996
between the lines varied from 500 to 1000m. The number
Masako :
of traps and the length of the trapping line varied accord-
October 1984 to November 1985;
ing to the dimensions of the different habitats and the
December 1985 to December 1986;
study aspects. Grids have only been used at Masako from
April 1985 to April 1986;
1985 to 1988. Twelve grids were used in fallow lands, 12
June 1986 to April 1988;
grids in the secondary forest and 4 in the primary forest.
May 1996;
The grids (100 x 100m) consisted of 100 signposted trap
September 1997 to November 1997;
stations, placed at 10m distance, with two traps per sta-
June 1999 and November 2001.
tion within a radius of 1m from the signpost. A grid was
thus covered by 200 traps and the grids were separated by
The specimens were fixed in formalin 10% and biop-
minimum 200m distance.
sies were preserved in ethylic alcohol 85%. The determi-
nation of rodents was effected at first at the University of
Since November 2001, the pitfall capture technique
Antwerp, Belgium (Evolutionary Biology Group, late
permitted to increase the specific richness of rodents and
Prof. Verheyen) then later at the Laboratoire d'Ecologie et
shrews of certain localities. Each pitfall line had a length
de Gestion des Ressources Animales (LEGERA, Kisan-
of 105 m with a station each 5 metres, thus the pitfall line
gani) by comparing morphometric and cranio-dental
consisted of 20 plastic buckets, buried so that the rim was
characters as described by DELANY (1975), HOLLISTER
level with the ground. The buckets were 290 mm high
(1916), MEESTER & SETZER (1971), HUTTERER & HAP-
with a superior diameter of 290 mm and a lower diameter
POLD (1983), HUTTERER & DUDU (1990). Shrew speci-
of 220 mm. The bottom was pierced with small holes to
mens were identified at the Alexander Koenig Museum
permit the infiltration of water during rain showers. The
(Germany) and the University of Rennes (France) with
pitfalls were installed following protocols in STANLEY et
morphological (skull morphometrics) and karyological
al. (1998) and NICOLAS et al. (2003). A canvas or plastic
techniques.
drift fence of 0.5m height was constructed that bisected
the rim of the buckets. At each end of the pitfall line, the
We calculated different indices of diversity for the dif-
fence was prolonged for 2.5 metres beyond the last
ferent study localities and habitats. We calculated the
bucket. Generally, we constructed three pitfall lines
Shannon ­Wiener diversity index using the following
within different habitats at each locality. The distance
formula :
between two lines varied according to the explored
H = -S(pi log pi) (B
2
ARBAULT, 1981).
localities : from 200m to 2000m for Masako and from
We also calculated the equitability E using the formula
300 to 1000m for the islands (Mbiye and Kungulu). The
H/Hmax and the Simpson's index
trapping continued for 3 to 6 days (Victor and Sherman
D = 1 - S(pi)².
traps) or for 21 consecutive days (pitfall lines) during a
Finally, we calculated the trapping success (the number
sampling period that varied from 1 to 12 months per study
of individuals captured per 100 trap nights).
year.
Sampling periods per locality
RESULTS
are as follows :
The materiel collected sums up to 7736 specimens
Kungulu :
(rodents and insectivores combined), with in total 25 spe-
December 1978 to April 1979;
cies of rodents (4 families) and 17 species of shrews or

136
I. Mukinzi, P.G.B. Katuala, J. Kennis, M. Gambalemoke, N. Kadange, A.M. Dudu, M. Colyn and R. Hutterer
TABLE 1
Diversity indices for the Rodentia and Insectivora captured at all mainland study sites, per habitat and per study
site. Masako, the zoo and Kisangani town lie on the right bank of the Congo River, Yoko and Lula are situated on
the left bank. The number of specimens is indicated, together with the specific richness SR (number of species),
the Shannon-Wiener Index H, the equitability E and the Simpson Index. PF = primary forest, SF = secondary for-
est, FL = fallow land, WF = wetland forest, total = total per study site, if more than one habitat was prospected.
Masako
Zookis.
Ki. Town
Yoko
Lula
PF
SF
FL
WF
Total
SF
FL
WF
Total
FL
SF
FL
WF
Total
FL
Rodentia
# spec.
168
1809
1772
410
4159
274
157
44
475
766
64
54
73
191
100
SR
12
19
19
15
23
6
9
6
11
10
7
9
7
12
8
H
2.18
2.75
3.04
3.04
3.06
1.62
2.03
0.51
2.04
1.74
1.63
2.73
2.17
2.34
2.29
E
0.61
0.65
0.71
0.78
0.68
0.63
0.64
0.20
0.59
0.52
0.58
0.86
0.77
0.65
0.77
Simpson
0.87
0.80
0.85
0.85
0.84
0.60
0.61
0.99
0.66
0.54
0.57
0.81
0.71
0.63
0.76
Insectivora

# spec.
24
78
39
9
150
35
3
7
10
20
3
SR
4
15
10
4
16
4
2
1
1
2
2
H
1.52
2.93
2.90
1.01
2.96
0.79
0.92
0.29
0.92
E
0.76
0.75
0.87
0.51
0.74
0.39
0.29
Simpson
0.60
0.81
0.84
0.34
0.82
0.26
0.44
0.10
0.44
TABLE 2
Diversity indices for the Rodentia and Insectivora captured at all island study sites, per habitat and per study
site. The number of specimens is indicated, together with the specific richness RS (number of species), the
Shannon-Wiener Index H, the equitability E and the Simpson Index. PF = primary forest, SF = secondary for-
est, FL = fallow land, WF = wetland forest, total = total per study site, if more than one habitat was prospected.
Mbiye
Kungulu
Tundulu
Mafi
PF
SF
FL
WF
Total
PF
SF
FL
Total
SF
FL
Total
FL
Rodentia
# capt.
305
223
298
298
1124
36
100
72
208
86
104
190
83
SR
6
8
9
6
11
4
8
7
10
2
3
4
8
H
0.84
0.87
1.19
1.21
1.58
1.88
2.04
2.44
2.39
0.09
0.95
0.69
1.96
E
0.32
0.29
0.38
0.47
0.34
0.94
0.68
0.87
0.72
0.09
0.60
0.35
0.65
Simpson
0.26
0.25
0.36
0.42
0.33
0.71
0.70
0.79
0.75
0.02
0.38
0.24
0.62
Insectivora
# capt.
67
12
44
29
152
9
45
6
60
20
SR
3
1
2
2
4
1
4
1
4
1
H
1.04
0.77
0.41
0.87
0.86
0.66
E
0.66
0.77
0.41
0.44
0.43
0.33
Simpson
0.48
0.35
0.98
0.37
0.31
0.22
TABLE 3
Total number of species, trap nights, number of specimens and trap success per locality (rodents and insecti-
vores combined). * Total number of trap nights not available.
Masako
Zookis
Kisangani
Yoko
Lula
Mbiye
Kungulu
Tundulu
Mafi
# Species
39
11
14
14
10
15
14
4
9
Trap nights
19465
2212
2994
2880
670
9756
3016
750
*
Total # captures
4309
475
801
211
103
1276
268
190
103
Trap success
22.14
21.47
26.75
7.33
15.37
13.08
8.89
25.33
*
elephant shrews (3 families). The following Rodentia
Praomys mutoni, Praomys misonnei, Paraxerus boehmi,
were found : Colomys goslingi, Dendromus mystacalis,
Rattus rattus and Stochomys longicaudatus.
Deomys ferrugineus, Funisciurus.pyrrhopus, Funisciurus
anerytrus, Grammomys kuru, Graphiurus lorraineus,

The following Insectivora (sensu lato, including Chryso-
chloridae and Macroscelidae) were found : Amblysomys leu-
Graphiurus surdus, Hybomys lunaris, Hylomsycus stella,
corhinus, Crocidura olivieri, C. cfr. hildegardeae, C. denti,
Hylomyscus aeta, Hylomyscus parvus, Lophuromys
C. dolichura, C. jacksoni, C. latona, C. littoralis, C. cfr.
dudui, Lophuromys flavopunctatus, Lemniscomys stria-
ludia, C. congobelgica, C. caliginea, Petrodromus tetradac-
tus, Malacomys longipes, Mastomys natalensis, Mus
tylus, Rynchoncyon cirnei, Scutisorex somereni, Sylvisorex
minutoides, Oenomys hypoxanthus, Praomys jacksoni,
jonhstoni, Sylvisorex cf oriundus and Suncus infinitesimus.

Rodents and insectivores of Kisangani
137
TABLE 4
Trap nights, number of specimens and trapping success per habitat for the localities and years were these data
are available. Several years were thus omitted. The year 2001 at Masako was omitted here due to unfinished
determination work. PF = primary forest, SF = secondary forest, FL = fallow land, WH = wet habitat.
Masako (1985-1987, 1997, 1999)
Mbiye island (1996, 2000, 2002)
Kungulu island (1999, 2003)
PF
SF
FL
WH
PF
SF
FL
WH
PF
SF
FL
# captures
141
1705
1691
362
220
109
223
171
32
198
25
Trap nights
830
6480
5790
1480
1936
1060
2141
2168
280
2280
200
Trap. success/hab.
16.99
26.31
29.21
24.46
11.36
10.28
10.42
7.89
11.43
8.68
12.50
Eight other species were not captured but often
Dendromys mysticalis appear to be strictly confined to for-
observed in the wild or found at game meat markets and
ests of the right bank of the Congo River. In contrast,
are thus added to the species list : Anomalurus beecrofti
Lophuromys flavopunctatus occurs abundantly on the left
and A.derbianus (Anomaluridae), Protoxerus stangeri
bank of the Congo River in forests as well as in fallow
and Heliosciurus rufobrachium (Sciuridae), Atherurus
lands. The highest diversity is recorded on the right bank
africanus (Histricidae), Thryonomys swinderianus (Thry-
with 40 species of which Masako alone is populated by 39
onomidae), Cricetomys emini (Cricetidae) and Pota-
species and represents 55.7% of all captures. On the
mogale velox (Potamogalidae).
islands, 19 species (14 rodents, 5 shrews) are present and
on the left bank, although few studies have taken place
50
there, 15 species are present (12 rodents, 3 insecti-
vores).The specific diversity varies in the different habi-
40
tats. In the forests (primary and secondary), 45 species (38
captured and 7 observed, 30 rodents and 15 insectivores)
e
are found, against 38 species in fallow land (31 captured
g 30
t
a

and 7 observed, 26 rodents and 12 insectivores).
r
cen
e
20
40
P
35
10
30
0
25
ecies
PF
SF
FL
WF
f
sp

Habitats
20
o
er

Fig. 2. ­ Total number of captures per habitat for rodents
b
m
15
(black), insectivores (shaded) and rodents and insectivores
u
N

combined (white).
10
5
In total, 99% of captures were rodents. Masako yielded
the highest diversity (H=3.06, E=0.68, D=0.84 for
0
rodents and H=2.96, E=0.74, D=0.82 for shrews), as indi-
Victor (1979-1997)
Sherman (1996-2003)
Pitfall (2001-2003)
cated by Tables 1 and 2. The highest trapping success was
Trap type
observed in degraded areas (in town and on Tundulu
Fig. 3. ­ Number of species per trap type for rodents (black),
island); the lowest trap success values were found at the
insectivores (shaded) and rodents and insectivores combined
Yoko reserve (7.33%) and the island Kungulu (8.89%), as
(white).
indicated by Table 3. Trap success per habitat is given in
Table 4 for all localities where these data are available.
Species distribution differs per habitat (Fig. 2). Fallow
The most common species captured during our study in
lands contain the highest relative abundance of rodents
the different habitats are Praomys jacksoni, Lophuromys
dudui and Lemniscomys striatus. Praomys jacksoni, Lem-
and shrews with respectively 47% and 35% of captures;
niscomys striatus, Hylomyscus stella, Hybomys lunaris,
followed by secondary forests (35% and 31%). The low-
Oenomys hypoxanthus, Mus minotoïdes, Stochomys longi-
est numbers of rodents were captured in primary forest
caudatus, Gramnomys kuru (formerly Thamnomys ruti-
for rodents and in wet habitats for insectivores. The com-
lans), Rattus rattus, Malacomys longipes and Lophuromys
parison of different trapping techniques illustrated in Fig.
dudui occur almost everywhere. A few species are particu-
3 shows a higher capture rate for Soricidae in the pitfalls
lar to one bank of the Congo River. Praomys misonnei, P.
whereas rodents have a higher capture rate with Victor
mutoni, Hylomyscus aeta, Hylomyscus parvus, Paraxerus
and Sherman traps. All species are captured regularly in
boehmi, Funisciurus pyrrhopus, Deomys ferrugineus and
all months (Table 5).

138
I. Mukinzi, P.G.B. Katuala, J. Kennis, M. Gambalemoke, N. Kadange, A.M. Dudu, M. Colyn and R. Hutterer
TABLE 5
TNumber of captures per species per season for the localities and during the years were these data are available
(DS = dry season, WS = wet season).
s
us
s
u
s
datus
pu
t
hr
eu
s
e
u
s
s
is
ho
vus
i
atus
i
icau
ngi
rr
i
s
ta
dui
xanth
ni
n
ei
m
u
ru
li
rrain
str
r
ugin
anery
u
r
dus
.lo
longipe
ypo
cksoni
l
ong
u
s py
u
s.
ys du
us
mys
h
r
u
s
s
scys stella
scys par
ys natalen
s muto
s mison
s boeh
mys k
ys fer
ys lunar
om
co
mys .
h
i
ur
my
o
mys
mys
r
on
nisciur
nisciur
aphiu
bom
l
o
my
o
myscys ae
lo
a
xeru
o
p
hur
e
mnis
a
stom
aomy
aomy
ammo
t
a
l
Sites
yea
Seas
Colomys gos
Deom
Fu
Fu
Grap
Gr
Hy
Hy
Hyl
Hy
L
L
Malac
M
Mus minutoides
Oeno
Praomys ja
Pr
Pr
Par
S
t
ocho
Gr
To
Masako 1985
DS
2
69
0
0
0
0
60
36
8
0
13
4
8
2
4
0
101
2
19
8
16
5
369
WS
2
87
1
0
0
2
75
72
6
3
49
28
31
0
9
1
138
9
0
6
25
19
563
1986
DS
0
127
1
0
1
0
259
127
4
14
87
21
19
1
3
0
235
17
3
5
37
39
1000
WS
0
105
0
0
0
1
193
98
6
16
82
22
12
0
11
0
219
3
7
4
31
13
823
Mbiye
1996
DS
1
0
0
1
2
0
0
0
0
0
8
4
9
0
0
1
63
0
0
0
0
3
92
WS
1
0
0
1
0
0
0
0
0
0
8
1
12
0
0
0
80
0
0
0
0
5
108
2000
DS
1
0
0
0
0
0
0
18
0
0
11
0
0
0
5
0
61
0
0
0
0
0
96
WS
1
0
0
0
0
0
0
13
0
0
14
0
0
0
4
0
73
0
0
0
0
0
105
Yoko
1996
DS
2
0
0
0
0
0
20
7
0
0
0
5
2
0
2
0
26
0
0
0
5
71
WS
2
0
1
0
0
0
32
5
0
0
0
0
0
0
4
2
61
0
0
0
5
112
DISCUSSION
Praomys mutoni also on the left bank of the Congo River,
although it was missing in our collections from that area.
The highest abundance was noted in fallow lands
The presence of Thryonomys swinderianus might be
whereas the lowest is noted in primary forest and in wet
recent (starting from 1992). It occurs around Kisangani as
habitats. This distribution probably depends on food
a result of forest destruction (DUDU, 1994).
resource availability in these different habitats. Almost all
Crocidura olivieri (3.70%) and Scutisorex somereni
species were captured in the dry as well as in the wet sea-
(0.87%) are the most abundant insectivore species occur-
son as noted by DUDU et al. (2005). The highest diversity
ring in all study sites and habitats. The study of Insec-
of rodents was noted in Masako, followed by the island
tivora with pitfalls has only been conducted on the right
Kungulu and the Yoko forest reserve on the left bank.
bank of the Congo River and on some islands. Our study
Some species such as Suncus infinitesimus, Sylvisorex
does not cover areas situated beyond 50 km of Kisangani
johnstoni, Sylvisorex cfr oriundus, Dendromus mystacalis
and trapping effort was different at the different sites.
are found for the first time during our studies at Masako
This explains partly the absence of some species such as
forest. According to COLYN & DUDU (1986), habitats with
Crocidura nigrofusca, C. goliath, and C. turba collected
a high floral diversity usually support a high fauna diver-
previously in the forests between Maiko and Tshopo riv-
sity. Except for Masako, diversity data on shrews are still
ers (COLYN, 1986) in our collections. Future collections
lacking. This prevents us from extending our analysis to
from the left bank of the Congo River should help to bet-
include the Insectivora. More details on dietary analysis,
ter understand shrew biodiversity and only part of the
feeding habit, coexistence, occurrence and food overlap
shrew collection data at Masako have been used in the
of shrews in Masako forest are given in DUDU et al.
present article.
(2005). STANLEY et al. (1998) reported that pitfalls are
effective for surveying insectivore fauna but are less suc-
We compared the number of terrestrial small mammals
cessful with larger rodents. This is confirmed by our
species from a number of studies carried out in different
results. The combination of snap traps, Sherman life traps
parts of the DRC (Table 6). Although the small mammal
and pitfalls should be more efficient to collect mid-sized
survey reported in the present paper is not exhaustive, the
small mammals.
tropical lowland forests around Kisangani may be already
be classified among the most diverse equatorial forests of
The most abundant species in the region are Praomys
central Africa as far as rodent and shrew ´species diver-
jacksoni (34% of total captures), Lophuromys dudui
sity is concerned.
(16%) and Hybomys lunaris (12%). According to DUDU
and GEVAERTS (1987), Praomys jacksoni and Lophuromys
dudui
occur everywhere in Kisangani and its surround-
ACKNOWLEDGEMENTS
ings. Lemniscomys striatus has been collected at many
sites, but occurs everywhere with a low abundance
We are grateful to the following members of our team
(3.56% of total captures). Deomys ferrugineus (7.79%)
assisting us with the field work : Gembu T., Danadu M.,
was mostly captured in Masako forest reserve and has
Amundala D., Iyongo W.M., Kaswera K. and Bapeamoni
been signalled once only at Linoko (Left bank of the
A. We also thank the late Prof. dr. Walter Verheyen for his
Lindi river) in 1998. The absence of some species on
collaboration on the identification of the rodent species
either bank does not mean that they do not occur there
and his friendship. We regret to have lost his presence at
because our small mammals surveys are still incomplete.
the moment we would benefit the most from our collabo-
For example COLYN (personal communication), found
ration.

Rodents and insectivores of Kisangani
139
TABLE 6
Comparison between the number of species per genus from different studies in rainforest
habitat, inside and outside the DRC. A = present study, B = DUDU, 1991, C = COLYN,
1986 (A, B, C all in the Kisangani area); D = DIETERLEN & DE BALZAC, 1979 in the
Kivu; E = LEIRS et al., 1999 in Kikwit; F = ECOFAC II 2000 (Ngotto, Central African
Republic; unpublished report); G = ECOFAC 2001 (Mt. Doudou, Gabon; unpublished
report)
Family
Genus
A
B
C
D
E
F
G
Anomaluridae
Anomalurus
2
2
3
1
Cricetidae
Cricetomys
1
1
1
1
Deomys
1
1
1
1
Dendromys
1
3
Tatera
2
Gliriidae
Graphiurus
2
2
1
Muridae
Colomys
1
1
1
1
Dasymys
1
Gramomys
1
2
Heimyscus.
1
Hybomys
1
1
1
1
1
Hylomyscus
3
3
1
1
1
3
Lemniscomys
1
1
1
1
Lophuromys
2
1
1
2
1
1
Mastomys
1
1
1
Mus
1
1
1
2
Malacomys
1
1
1
1
1
1
Oenomys
1
1
1
1
Pelomys
3
Praomys
3
3
2
1
1
2
Rattus
1
1
1
1
Steatomys
1
Stochomys
1
1
1
1
1
Thamnomys
1
1
1
1
Sciuridae
Funisciurus
2
2
2
2
Protoxerus
1
1
1
Paraxerus
1
1
1
1
Heliosciurus
1
Hystricidae
Atherurus
1
1
1
Thryonomidae
Thryonomys
1
1
Potamogalidae
Potamogale
1
1
1
Chrysochloridae
Amblysomus
1
1
1
Chlorotalpa
1
Macroscelidae
Petrodromus
1
1
Rynchoyon
1
1
1
Soricidae
Congosorex
1
Crocidura
10
9
6
9
10
12
5
Myosorex
1
Paracrocidura
1
1
Sylvisorex
2
4
3
2
Suncus
1
1
1
Scutisorex
1
1
Total species
49
40
33
15
44
25
21
REFERENCES
DIETERLEN, V.F. & H. HEIM DE BALZAC (1979). Zur Ökologie
und Taxonomie der Spitzmäuse (Soricidae) des Kivu ­
B
Gebiete. Saügetierk. Mitt., 27(4) : 241-287.
ARBAULT, R. (1981). Ecologie des populations et des peuple-
ments. Masson, Paris.
DUDU, A.M. (1989). Données écoéthologiques préliminaires des
Rongeurs myomorphes et sciumorphes de l'île Kungulu.
COLYN, M. (1986). Les Mammifères de forêts ombrophiles entre
Ann. Fac. Sci. Unikis : 67-73.
les rivières Tshopo et Maiko (Région du Haut-Zaïre). Bull.
Inst. r. Sci. Nat. Belg.
, 56 : 21-26.
DUDU, A.M (1991). Etude du peuplement d'Insectivores et de
Rongeurs de la forêt ombrophile de basse altitude du Zaïre
COLYN, M. (1991). L'importance zoogéographique du bassin du
(Kisangani, Zaïre). Thèse doct. inédite, Université
fleuve Zaïre pour la spéciation : le cas des primates simiens.
d'Anvers : 171p.
Paimpont,Université de Rennes, France, pp : 1-13.
DUDU, A.M. (1994). Note de la présence de Thryonomys swind-
COLYN, M. & A.M. DUDU (1986). Relevé systématique des
erianus (Temminck, 1827), le grand aulacode (Rongeurs,
Rongeurs (Muridae) des îles forestières du fleuve Zaïre entre
Thryonomidae) à Kisangani, Zaïre. Ann. Fac. Sci. Unikis,
Kisangani et Kinshasa. Rev. Zool. Afr., 99 : 353-357.
10 : 159-161.
DELANY, M.J. (1975). The rodents of Uganda.Trustees Brit.
DUDU, A.M., M. KAZADI & H. GEVAERTS (1985). Données
Mus. Nat. H., 8 : 165pp.
préliminaires sur la distribution des Rongeurs des îles Kun-

140
I. Mukinzi, P.G.B. Katuala, J. Kennis, M. Gambalemoke, N. Kadange, A.M. Dudu, M. Colyn and R. Hutterer
gulu, Tundulu, et d'un biotope sur la rive gauche du fleuve
KADANGE, N., T. GEMBU, G.B. KATUALA & A. DUDU, (1998).
Zaïre à Kisangani. Ann. Fac. Sci. Kisangani, 3 : 77-91.
Distribution des Rongeurs et Insectivores (Mammalia) dans
DUDU, A. & H. GEVAERTS (1986). Notes de l'évolution des cap-
les habitats naturels du jardin zoologique de Kisangani (R.D.
tures des rongeurs à Masako (Kisangani, Zaîre). Ann. Fac.
Congo). Ann.Fac.Sci. Unikis, 11 : 201-209.
Sci. Kis., sp. : 19-22.
LEIRS, H., N.J. MILLS, J.W. KREBS, J.E. CHILDS, A. DUDU, N.
DUDU, A.M. & H. GEVAERTS (1987). Examen de la distribution
WOOLEN, G. LUDWIG, J.C. PETERS, T.S. KSIAZEK and other
des Muridae (Rodentia, Mammalia) dans quelques associa-
study group members (1999). Search for thee Ebola virus
tions végétales des environs de Kisangani (Zaïre). Ann. Fac.
reservoir in Kikwit, democratic Republic of Congo. Reflec-
Sci. Unikis, 4 : 103-110.
tion on Vertebrate collection. The Journal of Infectious dis-
DUDU,A., R. VERHAGEN, H. GEVAERTS & W. VERHEYEN (1997).
eases, 179, Suppl. 1 : 155-163.
Population structure and reproduction cycle of Praomys
LOMBA, B. & M. NDJELE (1998). Utilisation de la méthode du
jacksoni (De Winton, 1897) and first data on the reproduc-
transect en vue de l'étude de la phytodiversité dans la
tion of P.misonnei Van Der Straeten and Dudu 1990 (Muri-
réserve de la Yoko (Ubundu, RD Congo). Ann. Fac. Sci.
dae) from Masako forest (Kisangani, Zaïre), Belg. J. Zool.,
Unikis, 11 : 35-46.
27 (1) : 67-70.
MEESTER, J. & H.W. SETZER (eds) (1971). The Mammals of
DUDU,A., S. CHURCHFIELD & R. HUTTERER (2005). Community
Africa, an Identification Manual. Smithsonian Institution
structure and food niche relationships of coexisting rain-for-
Press. Washington, D.C. pp. 1-29.
est shrews in the Masako Forest, north-eastern Congo. In :
MOSANGO, M. (1991). Contribution à l'étude botanique et bio
MERRITT, J.F., S. CHURCHFIELD, R. HUTTERER & B. SHEFTEL
géochimique de l'écosystème forêt en région équatorial (île
(eds), Advances in the Biology of Shrews II New York, Spe-
Kongolo Zaîre). Thèse inédite Ulb.Fac.Sci. laboratoire
cial Publ. Int. Soc. Shrew Biol., No. 1 : 229-239.
d'Ecologie végétale : 446p.
HATT, R.T. (1940). Lagomorpha and Rodentia others than Sciu-
NICOLAS, V., P. BARRIERE & M. COLYN (2003). Impact of
ridae, Anomaluridae and Idiuridae collected by the Ameri-
removal pitfall trapping on the community of shrews
can museum Congo expedition. Bull. Amer. Mus. Nat. Hist.,
(Mammalia : Soricidae) in two African tropical forest sites.
76 : 457-604.
Mammalia, 67 : 133-138.
HOLLISTER, N. (1916). Shrews collected by the Congo Expedi-
NYAKABWA, M. (1982). Phytocénose de l'écosystème urbain de
tion of the American Museum. Bull. Amer. Mus. Nat. Hist.,
kisangani, (1ère partie) Ph.D. Theses Inédit Unikis : 418p.
35 : 663-680.
SCHOUTEDEN (1948) Faune du Congo Belge et du Rwanda-
HUTTERER, R. & D.C.D. HAPPOLD (1983). The Shrews of
Urundi. Mammifères. Ann Mus. Congo, Tervuren, Zool., 1 :
Nigeria (Mammalia : Soricidae). Bonn. Zool. Monogr. :18 :
331pp.
79.
STANLEY,W.T., P.M. KIHAULE, K.M. HOWELL & R. HUTTERER
HUTTERER, R. & A.M. DUDU (1990). Redescription of Crocidura
(1998). Small mammals of the Eastern Arc Mountains, Tan-
caliginea, a rare shrew from north eastern Zaïre (Mammalia,
zania. Journal of East African Natural History, 87(1/2) : 91-
Soricidae). J. Afr. Zool., 104 : 305-311.
100.

Belg. J. Zool., 135 (supplement) : 141-144
December 2005
A morphological assessment of Myosorex zinki, an
endemic shrew on Mount Kilimanjaro

W. T. Stanley1, M. A. Rogers1 and R. Hutterer2
1 Department of Zoology, Field Museum, 1400 Lake Shore Drive, Chicago, Illinois 60605, U.S.A.
2 Zoologisches Forschungsmuseum Alexander Koenig, Adenauerallee 160, D-53113 Bonn, Deutschland, r.hutterer.zfmk@uni-
bonn.de
Corresponding author : W.T. Stanley, e-mail : bstanley@fmnh.org
ABSTRACT. The specific status of Myosorex zinki is analyzed based on recently collected material from Mt. Kili-
manjaro. M. zinki historically has been viewed as a subspecies of M. blarina, but is distinct from M. blarina in sev-
eral cranial dimensions. The recently collected series exhibits no sexual dimorphism. Contrary to what past records
have indicated, M. zinki is not restricted in occurrence on the mountain and is found in many habitats on Mt. Kili-
manjaro including forest, heathland, and moorland near the edge of the alpine desert. This species remains the only
endemic mammalian species to Mt. Kilimanjaro.
KEY WORDS : Insectivora, Soricidae, Myosorex zinki, Kilimanjaro, biodiversity, systematics, Africa.
INTRODUCTION
length, hindfoot length, ear length and weight. All meas-
urements were in mm except weight, which was in grams.
Myosorex zinki was described as a subspecies of Myos-
Measurements of the cranium and mandible were made
orex blarina by HEIM DE BALSAC & LAMOTTE (1956)
with digital calipers to the nearest 0.1 mm. These
based on a skin with incomplete skull collected at 3700 m
included condylo-incisive length (CI), basal length (BL),
on Mt. Kilimanjaro. HEIM DE BALSAC (1967) later sug-
post-palatal length (PPL), length of entire upper tooth row
gested that M. zinki deserved specific rank, but others
(UTRL), length of complex teeth in upper tooth row (P4-
retained it as a subspecies of M. blarina (HEIM DE BALSAC
M3), width of third upper incisor (I3W), width of canine
& MEESTER, 1977; HONACKI et al., 1982; SHORE & GAR-
(CW), length of third upper molar (M3L), width of third
BETT, 1991). The need for additional specimens was often
upper molar (M3W), length from first upper incisor to,
cited as required to resolve the taxonomic status of the
and including, the upper canine (I-C), least interorbital
form of Myosorex on Mt. Kilimanjaro. In 2002, an eleva-
width (LIW), bimaxillary width (BW), nasal width (NW),
tional survey of the small mammals of Mt. Kilimanjaro
greatest width of the braincase (GW), height of the brain-
produced several specimens on this unique shrew, which
case (HBC) (measured by placing skull on microscope
allow us to conduct a morphological assessment of this
slide, measuring from the ventral surface of the slide to
taxon compared to M. blarina specimens from the Rwen-
the highest point of the cranium and then subtracting the
zori Mountains.
thickness of the slide from that measurement), length of
mandible including the incisor (M+I), length of mandibu-
MATERIAL AND METHODS
lar tooth row including first incisor (LTR), and length of
the lower molars 1-3 (m1-3).
Specimens of Myosorex were collected during a survey
Standard descriptive statistics (mean, range, standard
of the small mammals of the south-eastern slope of Mt.
deviation and coefficient of variation) were derived for
Kilimanjaro conducted July and August, 2002. Five sites
each taxon. We used a one-way analysis of variance to
were surveyed at 2043, 2470, 2897, 3477 and 4000 m
test for significant variation between sexes within M.
respectively. Both pitfall lines (for insectivores, prima-
zinki and between M. zinki and M. blarina. Principal com-
rily) and standard breakback traps (for rodents) were used
ponents were extracted from a variance-covariance
for the collection of voucher specimens. Details on the
matrix using the cranial variables converted to natural
methodology (which was identical to other small mam-
logarithms. All statistical analyses were done with Excel
mal surveys of montane areas of Tanzania) can be found
and SPSS.
in STANLEY et al. (1996, 1998). The Myosorex blarina
examined in this study were all collected in the Rwenzori
RESULTS
Mountains and are housed at the Field Museum of Natu-
ral History (Chicago, USA). The holotype of M. blarina
Summaries of external and cranial measurements for
zinki (SMNS 4505) was examined in the Stuttgart Natural
male and female M. zinki are presented in Table 1 and 2,
History Museum (Germany).
respectively. There were no significant differences
Standard measurements were taken of the specimen in
between sexes in either external or cranial measurements
the field including total length, head and body length, tail
(Tables 1 and 2). All measurements were subsequently

142
W. T. Stanley, M. A. Rogers and R. Hutterer
TABLE 1
External measurements of Myosorex zinki. Differences between sexes were not significant (P > 0.05, see text).
External
Total length
Head &
Tail length
Hindfoot
N
Ear (mm)
Weight (g)
measurement
(mm)
body (mm)
(mm)
(mm)
Male
6
129.7 ± 6.3
93.7 ± 5.4
35.8 ± 1.6
16.3 ± 1.2
7.5 ± 0.8
16.5 ± 2.3
118-137
84-100
33-37
15-18
6-8
14.5-19.5
Female
6
130.3 ± 4.7
93.0 ± 2.4
36.2 ± 2.5
15.8 ± 0.4
7.2 ± 1.0
15.7 ± 1.2
125-137
89-95
33-40
15-16
6-9
14-17
TABLE 2
Cranial measurements (mm) of male and female Myosorex zinki (mean value ± standard devia-
tion, range) with results of a one-way ANOVA to test for sexual dimorphism.
Measurement
Males (n = 5*)
Females (n = 6)
F value
p
CI
22.9 ± 0.64 (22.1-23.7)
22.8 ± 0.61 (22.0-23.6)
0.20
0.66
BL
20.3 ± 0.63 (19.5-21.0)
20.1 ± 0.59 (19.2-20.8)
0.22
0.65
PPL
10.5 ± 0.36 (10.0-10.9)
10.4 ± 0.35 (9.7-10.7)
0.17
0.69
UTR
9.5 ± 0.20 (9.2-9.7)
9.4 ± 0.22 (9.1-9.7)
0.43
0.53
P4-M3
5.4 ± 0.18 (5.1-5.6)
5.3 ± 0.14 (5.1-5.4)
0.16
0.70
I3W
0.6 ± 0.03 (0.5-0.6)
0.5 ± 0.03 (0.5-0.6)
5.05
0.05
CW
0.7 ± 0.05 (0.6-0.7)
0.6 ± 0.02 (0.6-0.6)
2.58
0.14
M3L
1.5 ± 0.05 (1.4-1.5)
1.5 ± 0.07 (1.4-1.6)
0.05
0.83
M3W
0.9 ± 0.03 (0.8-0.9)
0.8 ± 0.05 (0.8-0.9)
0.77
0.40
I-C
4.3 ± 0.12 (4.1-4.4)
4.2 ± 0.14 (4.1-4.4)
1.48
0.25
LIW
4.9 ± 0.17 (4.6-5.0)
4.7 ± 0.10 (4.6-4.9)
2.04
0.19
BW
6.5 ± 0.17 (6.3-6.8)
6.5 ± 0.10 (6.3-6.6)
1.02
0.34
NW
2.3 ± 0.07 (2.2-2.4)
2.2 ± 0.05 (2.2-2.3)
2.93
0.12
GW
11.6 ± 0.18 (11.3-11.8)
11.5 ± 0.33 (11.1-12.0)
0.50
0.50
HBC
7.1 ± 0.11 (6.9-7.2)
7.0 ± 0.23 (6.7-7.4)
0.08
0.79
MI
14.1 ± 0.31 (13.8-14.5)
14.0 ± 0.44 (13.4-14.6)
0.33
0.58
LT
8.6 ± 0.19 (8.4-8.9)
8.5 ± 0.22 (8.3-8.8)
0.51
0.49
m1-3
4.26 ± 0.08 (4.2-4.4)
4.2 ± 0.11 (4.0-4.3)
0.34
0.57
* Sample size for HBC was 4 for males.
TABLE 3
Cranial measurements (mm) of Myosorex blarina and Myosorex zinki (mean value ± standard devia-
tion, range) with results of a one-way ANOVA to test for significant differences between species.
Measurements in bold represent those that exhibit significant differences.
Measurement
M. blarina (n = 4)
M. zinki (n = 11*)
F value
p
CI
22.0 ± 0.35 (21.5-22.3)
22.8 ± 0.59 (22.0-23.7)
6.44
0.02
BL
19.6 ± 0.36 (19.1-19.9)
20.2 ± 0.58 (19.2-21.0)
3.57
0.08
PPL
9.6 ± 0.18 (9.4-9.8)
10.4 ± 0.33 (9.7-10.9)
20.49
0.001
UTR
9.6 ± 0.24 (9.3-9.9)
9.4 ± 0.19 (9.1-9.7)
3.09
0.10
P4-M3
5.7 ± 0.19 (5.4-5.9)
5.3 ± 0.14 (5.1-5.6)
14.00
0.002
I3W
0.7 ± 0.06 (0.6-0.8)
0.6 ± 0.03 (0.5-0.6)
50.5
0.000
CW
0.9 ± 0.04 (0.9-0.9)
0.6 ± 0.04 (0.6-0.7)
165.7
0.000
M3L
1.8 ± 0.07 (1.7-1.9)
1.5 ± 0.06 (1.4-1.6)
88.9
0.000
M3W
1.1 ± 0.02 (1.0-1.1)
0.8 ± 0.04 (0.8-0.9)
118.0
0.000
I-C
4.2 ± 0.08 (4.1-4.3)
4.2 ± 0.13 (4.1-4.4)
0.001
0.97
LIW
5.4 ± 0.08 (5.2-5.4)
4.8 ± 0.14 (4.6-5.0)
53.6
0.000
BW
7.2 ± 0.13 (7.1-7.4)
6.5 ± 0.13 (6.3-6.8)
76.3
0.000
NW
2.5 ± 0.07 (2.4-2.6)
2.3 ± 0.06 (2.2-2.4)
32.0
0.000
GW
11.7 ± 0.19 (11.5-12.0)
11.5 ± 0.27 (11.1-12.0)
3.35
0.08
HBC*
7.0 ± 0.18 (6.8-7.2)
7.0 ± 0.18 (6.7-7.4)
0.07
0.79
MI
14.1 ± 0.21 (13.8-14.4)
14.1 ± 0.36 (13.4-14.6)
0.07
0.79
LT
8.8 ± 0.14 (8.6-8.9)
8.6 ± 0.19 (8.3-8.9)
4.85
0.04
m1-3
4.6 ± 0.12 (4.4-4.7)
4.2 ± 0.09 (4.0-4.4)
37.3
0.000
* Sample size for HBC for M. zinki was 10.
combined for a comparison of crania from M. zinki to
measurements for the two taxa are presented in Table 3.
samples of M. blarina. Coefficients of variation for cranial
Out of the 18 measurements taken, 12 were significantly
characters were all less than 9%. Summaries of cranial
different (P < 0.05) between groups (one-way ANOVA;

Morphological assessment of Myosorex zinki
143
Table 3). The initial principal components analysis indi-
width and length of teeth (I3W, CW, M3W, M3L). The
cated that four variables : I-C, GW, HBC, and MI did not
second component was most heavily influenced by length
contribute meaningfully to the extraction of components
variables (CI and BL), but these loadings were less than
in multivariate space (communalities < 0.7), so these were
0.90 and specimens are not clearly separated along the
not included in the principal components analysis. The
first two components derived explained 87.5 and 5.9% of
second component. A plot of individual specimen scores
the variation, respectively. The dimensions that had the
is presented in Fig. 1. The differences between the two
highest loadings (greater than 0.90) on the first principal
taxa in both breadth of the skull and the dimensions of the
component were those of breadth (LIW and BW) and
third upper molar are exhibited in Figs 2 and 3.
Fig. 1. ­ Projection of individual specimen scores on the first two principal com-
ponents.
HONACKI et al., 1982). With the specimens now available,
the recognition of Myosorex zinki as a species is now war-
ranted, supporting statements by HEIM DE BALSAC &
LAMOTTE (1956), HEIM DE BALSAC (1970), HUTTERER (in
GRIMSHAW et al., 1995) and others (STANLEY & HUT-
TERER, 2000).
Fig. 2. ­ Crania of M. zinki and M. blarina.
DISCUSSION
Several authors have pointed out the distinctiveness of
M. zinki (H
Fig. 3. ­ Upper tooth row of M. zinki and M. blarina.
EIM DE BALSAC & LAMOTTE, 1956; GRIMSHAW
et al., 1995), but the small number of specimens available
until now caused some to retain the taxon as a subspecies
Because of the small number of specimens previously
of M. blarina (HEIM DE BALSAC & MEESTER, 1977;
collected and the localities of those collections, M. zinki

144
W. T. Stanley, M. A. Rogers and R. Hutterer
was thought to be rare and restricted in distribution on Mt.
duct research in Tanzania was granted by the Tanzania Commis-
Kilimanjaro (HEIM DE BALSAC, 1967; GRIMSHAW et al.,
sion of Science and Technology and the Tanzania National
1995). A recent survey of the small mammals along the
Parks. Funding for field work came from the Marshall Field,
south-eastern slope of Mt. Kilimanjaro (W.T. S
Ellen Thorne Smith and Barbara Brown Funds. Sheila Reynolds
TANLEY,
unpubl. data) documented the presence of this shrew in
kindly provided some translations.
many different habitats on the mountain including forest,
heathland, moorland, and the edge of the alpine desert.
LITERATURE CITED
While it was not as abundant as other shrews at each site
surveyed, M. zinki was the most widely distributed insec-
BURGESS, N.D., D. KOCK, A. COCKLE, C. FITZGIBBON, P. JENKINS
tivore along the elevational gradient, ranging from 2470
& P. HONESS (2000). Mammals. In : BURGESS, N.D., & G.P.
to 4000 m (Fig. 4). While M. zinki was not documented at
CLARKE (eds), Coastal forests of Eastern Africa. IUCN,
the lowest site (2043 m), it is too early to say it does not
Gland and Cambridge : 173-190; 401-406.
occur in the lower forests of Kilimanjaro.
GRIMSHAW, J.M., N.J. CORDEIRO & C.A.H. FOLEY (1995). The
mammals of Kilimanjaro. Journal of East African Natural
History, 84 : 105-139. (appeared 1997)
HEIM DE BALSAC, H. (1967). Faits nouveaux concernant les
Myosorex (Soricidae) de l'Afrique orientale. Mammalia, 31 :
610-628.
HEIM DE BALSAC, H. (1970). Precisions sur la morphologie et les
biotopes de deux Soricidae d'altitude, endemiques du Kili-
mandjaro. Mammalia, 34 : 478-483.
HEIM DE BALSAC, H. & M. LAMOTTE (1956). Evolution et phylo-
génie des Soricidés africains. Mammalia, 20 : 140-167.
HEIM DE BALSAC, H. & J. MEESTER (1977). Order Insectivora.
Part 1. In : MEESTER, J. & H.W. SETZER (eds), The mammals
of Africa : an identification manual
, Smithsonian Institution
Press, Washington, D.C. : 1-29.
HONACKI, J.H., K.E. KINMAN & J.W. KOEPPL (eds) (1982).
Fig. 4. ­ Graph of distribution of Myosorex along an eleva-
Mammal species of the world : A taxonomic and geographic
tional gradient on the south-eastern slope of Mt. Kilimanjaro.
reference. Allen Press, Inc. and The Association of System-
The numbers above each bar represent the percentage of total
atics Collections, Lawrence, Kansas, 694 pp.
shrews captured at each elevation.
SHORE, R.F. & S.D. GARBETT (1991). Notes on the small mam-
mals of the Shira Plateau, Mt. Kilimanjaro. Mammalia, 55 :
Myosorex zinki remains the only endemic mammal on
601-607.
Mt. Kilimanjaro so far, but detailed systematic study of
STANLEY, W.T., S.M. GOODMAN & R. HUTTERER (1996). Notes
the small mammals in eastern Africa may reveal other
on the insectivores and elephant shrews of the Chome For-
taxa endemic to this mountain. B
est, South Pare Mountains, Tanzania. (Mammalia : Insec-
URGESS et al. (2000) hold
tivora et Macroscelidea). Zoologische Abhandlugen Staatli-
that Crocidura monax is restricted to Mt. Kilimanjaro
ches Museum fur Tierkunde Dresden, 49 : 131-147.
while other authors (STANLEY et al., 2000) list this species
STANLEY, W.T., S.M. GOODMAN & P.M. KIHAULE (1998). Results
from various Eastern Arc mountains. The problem
of two surveys of rodents in the Chome Forest Reserve,
requires careful examination. Further studies are also
South Pare Mountains, Tanzania. (Mammalia : Rodentia).
needed to both expand our knowledge of the ecology and
Zoologische Abhandlugen Staatliches Museum fur Tierkunde
distribution of M. zinki on Kilimanjaro and determine if it
Dresden, 50 : 145-160.
or closely related forms occur on nearby mountains, such
STANLEY, W.T. & R. HUTTERER (2000). A new species of Myoso-
as Mt. Meru.
rex Gray, 1832 (Mammalia : Soricidae) from the Eastern Arc
Mountains, Tanzania. Bonner Zoologische Beiträge, 49 : 19-
29.
ACKNOWLEDGEMENTS
STANLEY, W.T., P.M. KIHAULE, K.M. HOWELL & R. HUTTERER
(2000). Small mammals of the Eastern Arc Mountains, Tan-
We thank I. Lejora, N. Mafuru, A. Makwetta, P.M. Kihaule,
zania. Journal of East African Natural History, 49[for
and M.A. Munissi for assistance in the field. Permission to con-
1998] : 91-100.

Belg. J. Zool., 135 (supplement) : 145-152
December 2005
Movements, habitat use and response of ricefield rats to
removal in an intensive cropping system in Vietnam

Peter R. Brown1,2, Nguyen Phu Tuan3 and Peter B. Banks2
1 CSIRO Sustainable Ecosystems, GPO Box 284, Canberra, ACT, 2601, Australia
2 School of Biological, Earth and Environmental Sciences, The University of New South Wales, NSW, 2052, Australia
3 National Institute of Plant Protection, Chem, Tu Liem, Hanoi, Vietnam
Corresponding author : Peter Brown, Phone : + 61 2 6242 1562, Fax : + 61 2 6242 1505, e-mail : peter.brown@csiro.au
ABSTRACT. Rapid post-control reinvasion typically hampers attempts to manage rodent pests, yet little is known
about the demography or behaviours of re-invaders. Here we study the habitat use and movement of Rattus argen-
tiventer
using radio-telemetry during a non-breeding season (tillering growth stage of rice) and a breeding season
(ripening stage of rice) in lowland irrigated rice in Vietnam. On two treatment sites, farmers removed rats by hunt-
ing, digging up burrows and by using trap barrier systems (early planted field of rice surrounded by a plastic fence
set with multiple capture rat traps), and on two control sites, farmers conducted their normal control practices. The
95% minimum convex polygon home range size of rats during the non-breeding period was 2.4 ha (n = 12) and sig-
nificantly smaller than during the breeding period (9.8 ha; n = 10). There was no difference in home range size
between treatment (removal sites) and control sites. During the non-breeding period, rats preferred to use the bank/
channel habitat during the day, and preferred vegetable habitats at night. During the breeding period, rats preferred
using rice habitats both during the day and at night. This preference during the breeding period was strongly influ-
enced by the availability of abundant cover and food offered by the mature rice crops. Rats were moving about the
rice fields in random directions and were not influenced by the removal of rats at nearby locations. We conclude that
even at low population densities, rodent control would need to be conducted over large areas to prevent recolonisa-
tion through random dispersal events and that rodent burrows should be destroyed during the non-breeding season
when little cover is provided by crops.
KEY WORDS : habitat use, movements, ricefield rat, removal, trap barrier system, bounty system.
INTRODUCTION
data on how rat populations respond to such control
actions or how quickly reinvasion occurs. Also of interest
Rodents are a significant problem for agriculture in
is how rats respond to control at different stages of crop
Vietnam. They are considered the number one pre-harvest
development when the availability of food and cover
pests of lowland irrigated rice crops, especially in the
changes, and whether there are differences in response in
Mekong and Red River Deltas (BROWN et al., 1999,
breeding and non-breeding seasons of the rats.
2003b). In particular, the ricefield rat, Rattus argentiv-
enter (Robinson & Kloss, 1916), is the most common
A key strategy for animals successfully reinvading
rodent found in rice crops in Vietnam, and it is an impor-
areas is to have high rates of dispersal. One method for
tant pest of rice crops in other parts of Southeast Asia
measuring rates of dispersal of small mammals into
including Malaysia and Indonesia. In Indonesia, it causes
vacant areas is to experimentally remove animals from
annual pre-harvest losses of around 17% (GEDDES, 1992;
grids. SCHIECK & MILLAR (1987) studied the response of
LEUNG et al., 1999). Other rodent species inhabiting rice
red-backed voles (Clethrionomys gapperi) to removal
fields in Vietnam include R. losea, R. rattus and Bandi-
trapping in a mountain fir forest in Alberta Canada and
cota indica (BROWN et al., 1999, 2003b). Little is known
found that about 80% of the voles caught in the removal
about how these rodent pest species interact with each
area originated from a distance of less than two home
other within the rice growing areas or how management
ranges away. NAKATA & SATOH (1994) studied the
should be implemented in a palliative manner to reduce
response of individual grey-backed voles (Clethrionomys
damage to rice crops. Currently, most farmers are reactive
rufocanus bedfordiae) to removal trapping to determine
in their control actions, only implementing management
the source of animals moving into the removal grid and
once the rat problem is moderate to severe.
the distance that these animals moved from source areas.
Methods for controlling damage caused by rodents in
After 2 weeks, over 90% of the voles initially located
rice agro ecosystems include application of rodenticides
within 30 m of the edge of the removal grid were making
(BUCKLE, 1999), hunting, fumigation, physical barriers
single-direction movements towards the removal grid.
such as the trap barrier system (TBS, SINGLETON et al.,
Conversely, BOUTIN et al. (1985) found that only 28% of
1998, 1999), and cultural practices such as synchronised
snowshoe hares (Lepus americanus) dispersed into
cropping, sanitation of fields and encouraging predators
removal areas and that most animals died on their home
(such as barn owls) (LEUNG et al., 1999). There are few
range rather than dispersing, while SULLIVAN & SULLIVAN

146
Peter R. Brown, Nguyen Phu Tuan & Peter B. Banks
(1986) found that the colonization rate was 25 ­ 58% per
site families manage small plots of land, each 0.04 ha, and
four-week period.
each family generally owns a total of 0.5 ­ 0.7 ha of land.
There are few examples where researchers have moni-
The principal crop grown in the area is rice. There are two
tored changes in movements of pest small mammals in
main rice-growing seasons each year, the spring rice sea-
natural field conditions using contemporary control meth-
son (transplanting late February and harvested mid June),
ods. E
and the summer rice season (transplanted mid July and
FFORD et al. (2000) looked at home range changes
in feral brushtailed possums (Trichosurus vulpecula) in
harvested late September). Rice is not grown in winter
New Zealand after applying an 80% control in one half of
because it is too cold. Other crops are vegetables (broc-
their experimental plot. They found that possums on the
coli, cabbage, kohlrabi, onion, pumpkin, tomato) and
edge of the control area moved their home ranges towards
flowers (chrysanthemum, rose). Summers are hot and
the removal area and that the "vacuum effect" in the pos-
wet, and winters are cool and dry. The annual average
sums was largely confined to home range adjustments by
rainfall is approximately 1600 mm, most falling during
individuals that had ranges overlapping the area of
May to September. Farmers irrigate their crops from
reduced density. L
channels using water supplied from large storage dams in
EIRS et al. (1997) found that recolonisa-
tion of maize fields by the multimammate rat (Mastomys
nearby hills. The soil type is heavy red clays.
natalensis) occurred very rapidly after a rodent control
The radio-tracking study was conducted during the
operation.
spring rice season in 2002. Rice was sown in February
Despite the economic and social costs caused by R.
2002 and then harvested in late June and early July 2002.
argentiventer, its habitat use and movements in the rice
Two sessions of radio-tracking were conducted to coin-
agro-ecosystems of Vietnam is not well understood. In
cide with the non-breeding season of rats (during the till-
West Java, Indonesia, there are two lowland irrigated rice
ering stage of the rice crop; March) and during the breed-
crops produced each year corresponding with the wet and
ing season of rats (during the ripening stage of the rice
dry seasons. Rattus argentiventer accounted for >95% of
crop; June).
rodent species captured (LEUNG et al., 1999) and were
found to have home ranges of 1-3 ha, with little differ-
Trapping and radio-tracking
ences between males and females, with smaller home
At each site, rats were caught using single-capture wire
range in the breeding season compared to the non-breed-
cage traps. Traps were baited with fresh vegetables and
ing season (BROWN et al., 2001). They mostly utilised
set strategically at sites where there was obvious rodent
banks (burrows) during the tillering stage of the rice crop
activity to catch as many rats as possible over an area of
(non-breeding period), but switch to daytime use of rice
250 x 250 m. At each site, fifty traps were set per night
paddies throughout the ripening stage of the rice crop
for eight consecutive nights in March and six consecutive
(breeding period) (BROWN et al., 2001). Research is there-
nights in June. All adult female R. argentiventer rats were
fore required in Vietnam because the cropping system and
collared on treatment and control sites and all adult male
composition of rodent species are different.
R. argentiventer were collared on control sites only.
As part of a large project examining the population
Resources and labour were limited so we chose not to
response to a range of rodent control methods at a village-
monitor males on treatment sites. Traps were checked
level (> 100 ha), we examined how individual rats used
hourly during the first few hours after sunset and early
their environment and how they might respond to
each morning. At capture, each rat was weighed (± 2 g),
removal of other rats through the application of control
sexed, and breeding condition determined and to confirm
techniques conducted by farmers. Specifically, we consid-
species identification and condition. Females with raised
ered whether rats moved in a random pattern (classical
teats and perforated vagina were classified as adults, and
diffusion) or directed their movement towards areas of
males with descended testes were classified as adult. Prior
lower population density ("vacuum effect" EFFORD et al.,
to release, at point of capture, each rat was fitted with a
2000). This was done by radio-tracking individual rats
single-stage radio transmitter (Sirtrack, New Zealand)
occupying rice fields on sites where farmers conducted a
attached to a nylon cable tie which functioned also as a
range of recommended rodent management practices
collar around the animal's neck.
(treated sites) and sites where farmers were not influ-
A 250 x 250 m grid of bamboo poles set 25 m apart
enced in their rodent management techniques (control
was used to provide reference points for locating radio-
sites).
collared rats. Radio tracking at all sites was conducted for
up to 14 days in both March and June. Four locations or
MATERIAL AND METHODS
"fixes" were sought each day : one during daylight hours
(0800-1400 hrs) for location of rat nests; and three after
Study site
dusk (1900-2400 hrs) when rats were most active. Night
fixes were 1 to 1.5 hours apart. It was not always possible
The study was conducted in Vinh Phuc Province, in
to obtain three fixes for each rat after dusk. Collared rats
northern Vietnam, 40 km north of Hanoi (21°08' N;
were tracked with a hand-held 3-element Yagi antenna
105°45' E). Four study sites were selected to comprise
connected to a radio receiver. More than 80% of location
part of a main village or sub-villages. Each site was 0.5 to
fixes were tracked to within 1 m of their actual location,
1 km apart and about 100 - 150 ha in size. The sites were
based on sightings of collared rats. For others, it was not
set up in March 1999 to monitor the population dynamics
possible to obtain more accurate fixes, because rats were
of rats before implementation of ecologically based
moving around in rice paddies and would swim away
rodent management (BROWN et al., 2003b). Within each
before we could obtain an accurate fix. The habitat type

Movements of rats in rice fields
147
(large roadside bank or channel bank, rice paddy, vegeta-
first two days of tracking (calculated by averaging X- and
ble crop, fallow, flower crop) and activity (e.g. sighted in
Y- coordinates of the first 5 -8 fixes) and the average loca-
field or known to be in a burrow) were recorded for each
tion from the last two days of tracking (last 5-8 fixes)
fix.
were calculated for each rat. Each period of tracking con-
tained at least two daytime locations. The distances (m)
Home ranges were calculated from 95% and 100%
and directions (bearings) moved from the first two days to
minimum convex polygons (MCP) using RANGES V
the last two days were then determined. On Control sites,
(KENWARD & HODDER, 1996). We calculated 95% and
distances and directions towards the principal compass
100% MCP because the 100% MCP may include forays
points (± 45° of each of N, E, S, and W) were calculated,
from their core areas to explore new areas and thus rele-
and on Treatment sites, distances and directions towards
vant to our hypotheses. Analyses were performed on rats
the removal area (± 45° of N for Treatment 1, ± 45° of E
that had >15 fixes, the minimum number of fixes required
for Treatment 2) and away from removal areas, were cal-
to estimate 80% of the home range size as found by
culated.
BROWN et al. (2001) and confirmed with these data. Home
ranges were ln-transformed to reduce the skewed distri-
bution for statistical analysis. The range span was also
RESULTS
calculated using RANGES V, and is defined as the largest
distance across the MCP.
In March, we trapped 51 rats from 2800 trap nights
The habitat use for each rat was determined within
(trap success = 1.8%) in total from all sites, and in June
each individual animal's home range by examining the
we trapped 21 rats from 2400 trap nights (trap success
proportion of fixes within each habitat type (OTIS &
0.9%). Twenty-one adult R. argentiventer rats were col-
WHITE, 1999). Log ratios of usage/availability were cal-
lared for radio-tracking in March, and ten adult R. argen-
culated for each habitat for each rat as the basis for com-
tiventer rats were collared in June (Table 1). The regional
positional analysis of proportional habitat use (AEBISCHER
abundance of rats at this time (spring) of year was gener-
et al., 1993). Habitat availability and use was compared
ally low (mean trap success of 0.5% in March and 1.9%
between months (March and June) and time of day (Day
in June from our regular trapping locations as part of the
or Night). During each tracking session, the crop types
village-level study being conducted, Fig. 1), and we
grown in each field within the 250 x 250 m grid (6.25 ha)
believed we captured the majority of R. argentiventer
area were recorded by walking through each site. The
present in the area. In March, 12 and three rats were
area of channels, banks and paths was estimated by meas-
removed by farmers by hunting and digging burrows
uring their widths and lengths. The area of each habitat
from removal areas on Treatment 1 and Treatment 2
type was then calculated and converted to a proportion of
respectively, and in June, 63 and 50 rats were removed by
habitat available.
farmers from removal areas on Treatment 1 and Treat-
ment 2 respectively, most of which were juvenile animals
Implementation of treatments
dug from nests (evidence of active breeding on the sites).
On Treatment sites (T1 and T2), two areas were set up :
In March, 12 rats had > 15 fixes (57% of rats captured),
1) where rats were captured and collared for radio-track-
whereas ten rats in June had > 15 fixes (100%). Nine rats
ing (non-removal area, as described above), and 2) where
collared and released in March died from suspected poi-
rats were removed. Each area was 6.25 ha in size. The
soning (small movements, lethargic behaviour observed,
removal areas were 225-250 m from the non-removal
or were found lying dead on the ground), one died from
areas based on average home range sizes and distances
predation (found radio collar lying with remains of inter-
that rats would travel and get caught in a TBS (B
nal organs) and two were thought to be hit by farmers
ROWN et
al., 2001, 2003a). Rats were removed by the use of a tacti-
(fatally wounded by blow to body), whereas in June, there
cal bounty system (SINGLETON et al., 1999), where farm-
was no mortality of radio-collared rats during the tracking
ers were paid 200 dong (USD$0.02 per rat) to hunt and
period (Table 1). If we combine deaths due to rodenticide
dig rats from the removal area at a stage when rat popula-
and injury, we find that in March, farmers caused a mor-
tions abundance was low. The bounty system operated
tality rate in collared rats of 20%, 85%, 67%, and 33% for
during both March and June on both Treatment sites. In
C1, C2, T1, and T2 respectively (52% overall).
addition, on T2 in June, two trap-barrier systems (TBS;
SINGLETON et al., 1998, 1999) were present with sticky
The average home range size of rats (estimated using
rice as the lure crop (variety Khang Dan, 140-150 days
the 95% minimum convex polygon method) in March
duration, established in late March and harvested after we
was 2.40 ha (± 0.47 SE) and in June was 9.79 ha (± 3.31
concluded field work in June). On Control sites (C1 and
SE) (Fig. 2). The ln-transformed home range size for
C2), farmers conducted their normal rodent control prac-
female rats was significantly larger in June than in March,
tices.
(F
= 4.781; P = 0.048), but there was no difference
1,13
between Treated and Control sites (F
= 0.005; P =
1,13
To measure the distance and direction of movements of
0.947;). We could not test for differences between males
rats from non-removal areas the average location from the
and females, because no males were captured in June.

148
Peter R. Brown, Nguyen Phu Tuan & Peter B. Banks
TABLE 1
Summary of radio-collared rats in March (non-breeding season) and June (breeding season) at Vinh Phuc,
Vietnam. Shown are Control (C1 and C2) and Treatment (T1 and T2) sites, radio-collar frequency, sex, the
number of days each animal was tracked, the number of fixes obtained, the fate of the animal, the home
range sizes (ha, calculated using 95% and 100% minimum convex polygon, MCP) and home range span
(m).
Tracking
No. days
95%
100%
Site
Rat No.
Sex
No. fixes
Fate
Span
period
tracked
MCP
MCP
March
C1
37
M
8
28
Alive
1.01
1.15
187
64
M
8
28
Alive
4.99
5.07
496
47
M
12
45
Alive
3.13
3.96
360
241
F
7
21
Poisoned
2.32
2.33
329
15
F
9
30
Alive
1.28
1.85
309
C2
3
M
2
5
Poisoned
-
-
-
6
M
2
6
Poisoned
-
-
-
55
M
11
42
Alive
0.93
0.94
186
60
M
6
24
Poisoned
2.10
2.27
265
24
M
1
2
Poisoned
-
-
-
54
M
1
2
Fatally
-
-
-
injured
41
M
2
5
Poisoned
-
-
-
T1
75
F
9
28
Poisoned
4.18
4.59
342
68
F
2
5
Poisoned
-
-
-
60
F
2
3
Predation
-
-
-
T2
25
F
12
43
Poisoned
0.14
1.33
199
39
F
9
39
Alive
3.79
4.43
381
43
F
2
6
Fatally
-
-
-
injured
44
F
14
52
Alive
4.19
4.83
716
45
F
13
50
Alive
0.79
0.91
217
49
F
1
3
Missing
-
-
-
June
C1
66
F
10
34
Alive
16.75
18.28
827
26
F
9
30
Alive
3.83
3.85
370
79
F
10
30
Alive
1.04
1.07
193
C2
53
F
13
46
Alive
4.89
9.90
487
T1
33
F
14
48
Alive
1.44
1.49
256
73
F
13
47
Alive
2.56
2.74
287
20
F
12
43
Alive
15.26
16.03
908
T2
23
F
14
45
Alive
2.08
4.16
355
71
F
13
45
Alive
17.07
19.28
755
57
F
9
32
Alive
32.97
43.91
1433
6
age (16.3% increase) than 95% MCP. There was no sig-
Breeding season
Control 1
nificant difference in size of home ranges between the
5
Control 2
Treatment 1
95% and 100% MCP (paired t-test; t = -0.508, P =
19
4
Treatment 2
0.618), therefore the 100% MCP did not provide addi-
3
tional information for home range analysis.
2
r
ap success (%)
March - non-breeding season
June - breeding season
T
1
35
5
2
5
0
4
6
0
30
Spring rice crop
25
Jan 02
Feb 02
Mar 02
Apr 02
May 02
Jun 02
Jul 02
20
Month
15
Fig. 1. ­ Abundance of rats (number of rats captured per 100
10
trap nights) on four sites used for a separate village-level
Home range size (ha)
study, Vinh Phuc, Vietnam from January to July 2002. Shown
5
is approximate timing of the spring rice crop and rat breeding
0
season (based on pregnant and lactating adult females) (P. R.
Male
Female
Female
Male
Female
Female
Brown and N. P. Tuan unpublished data).
Control
Treatment
Control
Treatment
Fig. 2. ­ Box plot of 95% minimum convex polygon home
range sizes (ha) for males and females, in treatment and con-
The home range size of rats calculated using the
trol sites for March (non-breeding season) and June (breeding
100% MCP in March was 2.81 ha (± 0.48 SE) and in
season). The box encloses the 25th and 75th percentiles; the
June was 12.07 ha (± 4.18 SE). The 100% MCP home
solid line shows the median and the dotted line the mean
range size in March was 0.40 ha larger on average
home-range size. Vertical lines span the 10-90th percentiles.
(17.8% increase) and in June was 2.28 ha larger on aver-
Sample sizes are shown at the top.

Movements of rats in rice fields
149
100
had constructed a burrow there and were feeding within
a) March
Available
the field.
80
Day
Night
In June rats were spending more time in rice habitats
60
with 73.2% of day fixes and 73.9% of night fixes in rice
cent (%)
paddies compared to the 40.0% available (Fig. 3b). Rats
40
Per
had reduced their use of bank/channel habitats to 25.0%
20
and 18.8% for day and night fixes respectively (compared
to 10% available). Very few fixes occurred in vegetable,
0
flower or fallow fields. The availability of crops changed
b) June
Available
80
Day
between March and June because of changes in the types
Night
of crop grown.
60
The ratios of usage/availability confirm that in March,
40
rats preferred to use bank habitats, and in June, rats pre-
Percent (%)
ferred to use rice habitats and banks to a lesser extent
20
(Table 2).
0
Vegetable
Bank/channel
Flower
Rice
Fallow
The distances moved by rats from the average of the
Habitat type
first 2 days to the last 2 days were generally twice as large
on treated sites as they were on control sites (Control
March = 88.0 m ± 30.6 SE, n = 7; Treatment March =
Fig. 3. ­ Habitat use of ricefield rats (sexes and sites com-
bined) showing the percentage of habitats available to rats and
190.9 m ± 94.0 SE, n = 5; Control June = 184.9 m ± 65.8
the percentage of radio-telemetry location fixes within each
SE, n = 4; Treatment June = 411.8 m ± 166.1 SE, n = 6)
habitat type for day and night fixes, Vinh Phuc Province, Viet-
(Fig. 4), but the distances moved were not significant
nam 2002. (a) March, the non-breeding period for rats during
(Time F
= 2.86; P = 0.108; Treatment F
= 0.278; P
1,18
1,18
the tillering stage of rice, and (b) June, the breeding period for
= 0.604; interaction F
= 0.06; P = 0.811).
1,18
rats during the ripening stage of rice. Error bars represent
standard error of means from habitat use of individual rats.
The directions moved by rats on control and treatment
sites were proportional with the directions available. On
control sites, there were three rats that moved towards
The home range span of rats was used as an estimate of
North, three to East, five to South and zero to West, with
possible linear movements (Table 1). There was no signif-
no preference for direction moved (the association
icant difference between home range span between
between the observed directions used and directions
months (F
= 1.409; P = 0.256) or treatment (F
=
expected was not significant : 2 = 4.636; P = 0.2004).
1,13
1,13
3
0.253; P = 0.624). The average home range span in March
On treatment sites, there was one rat that moved towards
was 332.3 m (± 44.0 SE, n = 12) and in June was 587.1 m
the removal area and 10 rats that did not move towards
(± 123.4 SE, n = 10). This confirms that the distance
the removal area (1/5 rats in March and 0/6 rats in June),
between non-removal and the removal areas was set at the
with no preference for direction moved (2 = 1.778; P =
1
right distance (225-250 m).
0.1824). Therefore, the direction of movements were
essentially random on both control and treatment sites.
In March, rats spent most time during the day in the
To confirm that rats were indeed breeding in June, the
bank/channel habitat (82.8% compared to 10% available)
animals that could be retrieved were assessed for breed-
and at night most fixes occurred in the vegetable habitat
ing condition (presence of embryos or litter of pups in the
(37.7% compared to 51.3% available) (Fig. 3a). Rats
burrow). Of the three female rats recaptured, one was
were not located in rice fields during the day at any stage
pregnant and two had young pups in their burrow, con-
during March. Rats used the flower habitat roughly in
firming that they were indeed breeding (100%). The
proportion with availability (day fixes = 10.3%; night
breeding condition of the other females could not be
fixes = 15.91; available 16.7%). Some rats consistently
ascertained because it was not possible to recapture the
had day and night fixes in flower fields suggesting they
animals.
TABLE 2
Habitat selectivity of ricefield rats during March (tillering stage of rice crop; non-breeding season)
and June (ripening stage of rice crop; breeding season) for day and night fixes for each habitat,
Vinh Phuc province, Vietnam. The selectivity index is calculated by dividing the proportion of
observations of rats in each habitat type by the proportion of habitat available. A selectivity value
of > 1 implies preference while a value of < 1 implies avoidance.
Month
Time
Vegetable
Bank
Flower
Rice
Fallow
March
Day
0.12 ± 0.04
8.42 ± 0.81
0.57 ± 0.45
0.00 ± 0.00
0.01 ± 0.00
Night
0.76 ± 0.16
3.07 ± 0.65
0.91 ± 0.43
0.71 ± 0.16
0.53 ± 0.52
June
Day
0.02 ± 0.02
2.37 ± 0.48
0.05 ± 0.05
3.65 ± 0.21
0.01 ± 0.00
Night
0.09 ± 0.05
1.89 ± 0.48
0.04 ± 0.02
3.65 ± 0.26
0.86 ± 0.45

150
Peter R. Brown, Nguyen Phu Tuan & Peter B. Banks
1200
that home range sizes were larger during the breeding sea-
7
5
4
6
son. In house mice in Australia, for example, home ranges
1000
were significantly smaller during the breeding season
(CHAMBERS et al., 2000). The home range size of Rattus
800
rattus in macadamia nut orchards in Hawaii did not vary
ed (m)
between males and females and did not vary through dif-
600
ferent stages of nut development (TOBIN et al., 1996).
CHRISTENSEN (1996) found no seasonal variation in home
400
range sizes of Mastomys natalensis in Tanzania as deter-
Distance mov
mined by capture-mark-release data. However, both male
200
and female Calomys venustus in Argentine agroecosys-
tems had larger home ranges during the breeding season
0
compared to the non-breeding season (PRIOTTO et al.,
March C
March T
June C
June T
2002). It is not clear why R. argentiventer might have a
Time and Treatment
larger home range during the breeding season (June), but
Fig. 4. ­ Box plot of distances (m) moved by rats from first 2
it could be related to the farming activities or farmers pre-
days to last 2 days of tracking on Treatment (T) and Control
paring for harvest. We expected that adult female rats, if
(C) sites in March and June 2002. The box encloses the 25th
they are actively breeding, would have stable, small home
and 75th percentiles, the solid line shows the median and the
ranges particularly if they are caring for young in the nest.
dotted line the mean distance travelled, and the vertical lines
The recaptures of rats in June confirmed that the rats were
span the 10-90th percentiles. Sample sizes are shown at the
indeed breeding (pregnant or suckling new born pups).
top.
We could not prevent farmers from undertaking extra-
neous rodent control on our study sites. On our control
sites in March, farmers poisoned nine radio-collared rats
DISCUSSION
with rodenticide and two other rats died through farmers
causing fatal injury. This reflects the rat control efforts
The removal of ricefield rats during a low-density
employed by farmers during the tillering stage of the rice
phase in an intensive cropping system in Vietnam did not
crop. Farmers are generally busy with preparations for
induce movements of neighbouring rats towards the
harvest in June, so they have little time for undertaking
removal area. Rats on non-removal areas were moving
rodent control. No deaths of radio-collared rats occurred
randomly with regard to directions, and we believe that
on any site in June. The impact of these activities on this
for R. argentiventer, recolonisation events during low-
study is difficult to determine. Rats in this intensive rice
density phases occur through random dispersal events
growing agroecosystem are subject to a wide array of dis-
(classical diffusion). We could not support the "vacuum
turbances including ploughing of fields, harvesting of
effect" proposed by EFFORD et al. (2000) for brushtail
crops, irrigation of crops, and application of chemicals for
possums in New Zealand, however the density of rats in
weed or insect control. Rat populations have developed
this study was low. KREBS et al. (1976) found that recolo-
strategies for survival under these conditions through
nisation rates were higher when densities of Microtus
high reproductive output (LAM, 1983; TRISTIANI et al.,
townsendii were higher because of competition for space,
1998) and through their ability to recolonise areas.
so this study should be repeated at higher densities (>10%
Rats were using a range of habitats that were available
trap success) to test this hypothesis.
to them, and their choice of habitat was related to cover
It is likely that populations of ricefield rats are made up
and availability of food. When cover from tillering rice
of predominantly transient animals with high rates of dis-
was low (March), rats were spending time in burrows in
persal, as found for multimammate rats (Mastomys
the bank/channel habitat, and when rice was ripening
natalensis) in Tanzania (LEIRS et al., 1996). We found rats
(June), rats were spending their time in the rice fields. We
moved around a great deal, and in some cases rats had
could not measure availability of food for rats, but obser-
very large home ranges (> 5 ha) and did not consistently
vations made at the time showed that abundant food was
use a particular burrow or nest site. In studies conducted
always available through ripening vegetable crops such as
in both Vietnam and Indonesia, ricefield rats have recap-
kohlrabi, tomatoes, cabbage and broccoli, and particularly
ture rates of less than 1% (BROWN et al., 1999, 2003b;
in June, abundance of maturing rice. Food was therefore
LEUNG et al., 1999; JACOB et al., 2003b), and part of this
probably not a limiting resource. In March, rats preferred
reason may be because of the high proportion of transient
to use the bank/channel habitat during the day, but pre-
animals. We therefore predict that rats inhabiting these
ferred vegetable habitats at night. In June, rats preferred
highly modified and intensive rice production systems
using rice habitats during the day and at night. This pref-
would have higher rates of dispersal than rats living in
erence in June was strongly influenced by the availability
stable environments.
of abundance cover and food offered by the maturing rice
The home range size of rats during the non-breeding
crops. These findings are similar to that found for R.
season (tillering stage, March; 2.7 ha) was of the same
argentiventer in Indonesia (BROWN et al., 2001).
order as that found for ricefield rats in Indonesia (2-3 ha,
These results suggest that there would be little point in
BROWN et al., 2001). However, the home range size was
destroying rat burrows along channels and bank habitats
much larger during the breeding season (ripening stage,
during the later stages of crop growth (after maximum til-
June; 10 ha) than in Indonesia. We were surprised to find
lering stage of rice) because rats were predominantly uti-

Movements of rats in rice fields
151
lising rice crops (BROWN et al., 2001). It would be inter-
domesticus) in the wheatlands of Northwestern Victoria,
esting to monitor the changes in habitat use and
Australia. J. Mammal., 81 : 59-69.
movements of rats after harvest of the rice crop to see
CHRISTENSEN, J.T. (1996). Home range and abundance of Masto-
whether they revert back to using the channel/bank habi-
mys natalensis (Smith, 1834) in habitats affected by cultiva-
tat or disperse to other habitats offering sufficient food
tion. Afr. J. Ecol., 34 : 298-311.
and cover. J
EFFORD, M., B. WARBURTON & N. SPENCER (2000). Home-range
ACOB et al. (2003a) found that the home range
size of R. argentiventer in Indonesia decreased by 67%
changes by brushtail possums in response to control. Wildl.
Res.,
27 : 117-127.
after harvest. The findings from the current research will
GEDDES, A.W.M. (1992). The Relative Importance of Pre-har-
help in refining appropriate management practices that
vest Crop Pests in Indonesia. Natural Resources Institute,
farmers can use on a large scale (e.g. village level) (SIN-
Chatham, UK.
GLETON, 1997; BROWN et al., 2003b; LEIRS, 2003; JACOB
JACOB, J., D. NOLTE, R. HARTONO, J. SUBAGJA & SUDARMAJI
et al., 2003b).
(2003a). Pre- and post-harvest movements of female rice-
Further research is required to examine recolonisation
field rats in West Javanese rice fields. In : SINGLETON, G.R.,
when population densities are higher and to look at other
L.A. HINDS, C.J. KREBS & D.M. SPRATT (eds), Rats, Mice
compensatory mechanisms such as breeding performance
and People : Rodent Biology and Management, ACIAR
Monograph 96, ACIAR, Canberra : 277-280.
and recruitment.
JACOB, J., SUDARMAJI & G.R. SINGLETON (2003b). Ecologically-
based management of rice-field rats on a village scale in
ACKNOWLEDGEMENTS
West Java : experimental approach and assessment of habitat
use. In : SINGLETON, G.R., L.A. HINDS, C.J. KREBS & D.M.
We sincerely thank the farmers who assisted us with the
SPRATT (eds), Rats, Mice and People : Rodent Biology and
implementation of the treatments and for their help with the
Management, ACIAR Monograph 96, ACIAR, Canberra :
radio tracking. We thank M. Davies (CSIRO), D.T Hue, P.T.T.
191-196.
Ha, P.T. Hoa, L.T. Hoa, P.T. Lien, P.V. Kien, and Ms Yen (NIPP)
KENWARD, R.E. & K.H. HODDER (1996). RANGES V - An Analy-
for assistance with this research. We thank D. Grice and G. Sin-
sis System for Biological Location Data. Institute of Terres-
gleton for comments on early drafts. The research was sup-
trial Ecology, Dorset.
ported by the Australian Centre for International Agricultural
KREBS, C.J., I. WINGATE, J. LEDUC, J.A. REDFIELD, M. TAITT &
Research (AS1/98/36) and CSIRO. The research was conducted
R. HILBORN (1976). Microtus population biology : dispersal
in accordance with the Australian code of practice for the care
in fluctuating populations of M. townsendii. Can. J. Zool.,
and use of animals for scientific purposes (SEAEC No. 01/02 -
54 : 79-95.
15).
LAM, Y.M. (1983). Reproduction in the ricefield rat Rattus
argentiventer. Malayan Nature Journal, 36 : 249-282.
LEIRS, H. (2003). Management of rodents in crops : the Pied
REFERENCES
Piper and his orchestra. In : SINGLETON, G.R., L.A. HINDS,
C.J. KREBS & D.M. SPRATT (eds), Rats, Mice and People :
AEBISCHER, N.J., P.A. ROBERTSON & R.E. KENWARD (1993).
Rodent Biology and Management, ACIAR Monograph 96,
Compositional analysis of habitat use from animal radio-
ACIAR, Canberra : 193-190.
tracking data. J. Wildl. Manage., 74 : 1313-1325.
LEIRS, H., W. VERHEYEN & R. VERHAGEN (1996). Spatial patterns
BOUTIN, S., B.S. GILBERT, C.J. KREBS, A.R.E. SINCLAIR &
in Mastomys natalensis in Tanzania (Rodentia, Muridae).
J.N.M. SMITH (1985). The role of dispersal on the population
Mammalia, 60 : 545-555.
dynamics of snowshoe hares. Can. J. Zool., 63 : 106-115.
LEIRS, H., R. VERHAGEN, C.A. SABUNI, P. MWANJABE & W.N.
BROWN, P.R., N.Q. HUNG, N.M. HUNG & M. VAN WENSVEEN
VERHEYEN (1997). Spatial dynamics of Mastomys natalensis
(1999). Population ecology and management of rodent pests
in a field-fallow mosaic in Tanzania. Belg. J. Zool., 127 : 29-
in the Mekong River Delta, Vietnam. In : SINGLETON, G.R.,
38.
L.A. HINDS, H. LEIRS & Z. ZHANG (eds), Ecologically-Based
LEUNG, L.K.P., G.R. SINGLETON, SUDARMAJI & RAHMINI (1999).
Management of Rodent Pests, Australian Centre for Interna-
Ecologically-based population management of the rice-field
tional Agricultural Research, Canberra : 319-337.
rat in Indonesia. In : SINGLETON, G.R., L.A. HINDS, H. LEIRS
BROWN, P.R., L.K-P. LEUNG, SUDARMAJI & G.R. SINGLETON
& Z. ZHANG (eds), Ecologically-Based Management of
(2003a). Movements of the ricefield rat, Rattus argentiv-
Rodent Pests, Australian Centre for International Agricul-
enter, near a trap-barrier system in rice crops in West Java,
tural Research, Canberra : 305-318.
Indonesia. Int. J. Pest Manage., 49 : 123-129.
NAKATA, K. & K. SATOH (1994). Distance of movements of
BROWN, P.R., G.R. SINGLETON & SUDARMAJI (2001). Habitat use
voles after removal in Clethrionomys rufocanus bedfordiae
and movements of the rice-field rat, Rattus argentiventer, in
(Rodentia : Cricetidae). Appl. Ent. Zool., 29 : 97-103.
West Java, Indonesia. Mammalia, 65 : 151-166.
OTIS, D.L. & G.C. WHITE (1999). Autocorrelation of location
BROWN, P.R., N.P. TUAN, G.R. SINGLETON, N.V. TUAT, T.Q. TAN
estimates and the analysis of radiotracking data. J. Wildl.
& L.T. HOA (2003b). Impact of village-level rodent control
Manage., 63 : 1039-1044.
practices on rodent populations and rice crops in Vietnam.
PRIOTTO, J., A. STEINMANN & J. POLOP (2002). Factors affecting
In : SINGLETON, G.R., L.A. HINDS, C.J. KREBS & D.M.
home range size and overlap in Calomys venustus (Muridae :
SPRATT (eds), Rats, Mice and People : Rodent Biology and
Sigmodontinae) in Argentine agroecosystems. Mammalian
Management, ACIAR Monograph 96, ACIAR, Canberra :
Biol., 67 : 97-104.
197-202.
SCHIECK, J.O. & J.S. MILLAR (1987). Can removal areas be used
BUCKLE, A.P. (1999). Rodenticides - their role in rodent pest
to assess dispersal of red-backed voles? Can. J. Zool., 65 :
management in tropical agriculture. In : SINGLETON, G.R.,
2575-2578.
L.A. HINDS, H. LEIRS & Z. ZHANG (eds), Ecologically-Based
SINGLETON, G.R. (1997). Integrated management of rodents : A
Management of Rodent Pests, Australian Centre for Interna-
Southeast Asian and Australian perspective. Belg. J. Zool.,
tional Agricultural Research, Canberra : 163-177.
127 : 157-169.
CHAMBERS, L.K., G.R. SINGLETON & C.J. KREBS (2000). Move-
SINGLETON, G.R., SUDARMAJI & S. SURIAPERMANA (1998). An
ments and social organization of wild house mice (Mus
experimental field study to evaluate a trap-barrier system

152
Peter R. Brown, Nguyen Phu Tuan & Peter B. Banks
and fumigation for controlling the rice field rat, Rattus
SULLIVAN, T.P. & D.S. SULLIVAN (1986). Resiliency of snow-
argentiventer, in rice crops in West Java. Crop Prot., 17 : 55-
shoe hares to population reduction. J. Appl. Ecol., 23 : 795-
64.
806.
TOBIN, M.E., R.T. SUGIHARA, A.E. KOEHLER & G.R. UEUNTEN
SINGLETON, G.R., SUDARMAJI, JUMANTA, T.Q. TAN & N.Q. HUNG
(1996). Seasonal activity and movements of Rattus rattus
(1999). Physical control of rats in developing countries. In :
(Rodentia, Muridae) in a Hawaiian macadamia orchard.
Mammalia, 60 : 3-13.
SINGLETON, G.R., L.A. HINDS, H. LEIRS & Z. ZHANG (eds),
TRISTIANI, H., J. PRIYONO & O. MURAKAMI (1998). Seasonal
Ecologically-Based Management of Rodent Pests, Austral-
changes in the population density and reproduction of the
ian Centre for International Agricultural Research,
ricefield rat, Rattus argentiventer (Rodentia : Muridae), in
Canberra : 178-198.
West Java. Mammalia, 62 : 227-239.

Belg. J. Zool., 135 (supplement) : 153-157
December 2005
Farmer's perceptions of rodents as crop pests : Knowl-
edge, attitudes and practices in rodent pest management
in Tanzania and Ethiopia

Rhodes H. Makundi1, Afework Bekele2, Herwig Leirs3,4, Apia W. Massawe1, Winnie
Rwamugira
1 and Loth S. Mulungu1
1 Pest Management Centre, Sokoine University of Agriculture, P.O. Box 3110, Morogoro, Tanzania
2 Department of Biology, Addis Ababa University, Addis Ababa, Ethiopia
3 University of Antwerp, Department of Biology, Groenenborgerlaan 171, B-2020 Antwerpen, Belgium
4 Danish Pest Infestation Laboratory, Danish Institute of Agricultural Sciences, Department of Integrated Pest Management,
Skovbrynet 14, DK-2800, Kgs. Lyngby, Denmark
Corresponding author : Rhodes H. Makundi, e-mail : makundi@suanet.ac.tz; makundi@yahoo.com
ABSTRACT. A study was conducted using a structured questionnaire to obtain information about the nature and
extent of rodent damage to crops, farmer's perceptions of crop pests and their knowledge, attitudes and practices to
their management in Tanzania and Ethiopia. The study was carried out in five localities (Makuyu -Central Tanzania;
Chunya-Southwest Tanzania; Ziway and Adami Tulu (south of Addis Ababa) and Gumer/Limmo-South-west of
Addis Ababa, both in Central Ethiopia). In Tanzania, maize is the major crop, both for food and sale. Other crops
are sorghum, rice, simsim, groundnuts and millet. In Central Ethiopia, farmers grow maize, sorghum, teff, beans,
barley, wheat, potatoes and enset. The study showed that farmers in Tanzania and Ethiopia are well aware of rodent
problems and considered them to be number one pest. Rodent problems are regular and maize is the most affected
crop in Tanzania. In Ethiopia, maize, enset and potatoes are the most affected crops. Maize in Ethiopia and Tanzania
is susceptible to rodent damage, most seriously at planting and seedling stage. Although different rodent control
techniques are practiced in Tanzania, farmers prefer using rodenticides (68.7%) to other strategies. In Ethiopia, trap-
ping, hunting and rodenticides are the most practised techniques for rodent control. Farmer's attempts to control
rodents in both countries are based on economic reasons and generally, rodent control is not undertaken when there
are no crops in the fields. Farmers are responsible for rodent control activities in their individual fields. The study
shows that farmers in Tanzania and Ethiopia are concerned with rodent infestation and are also aware of the critical
growth stage when the crops are most susceptible. A lack of multiple rodent management methods and inadequate
knowledge of appropriate and sustainable techniques appeared to be the main reasons for the over dependence on
rodenticides, particularly in Tanzania. Therefore, this suggests that farmers require a strong extension input to man-
age rodent problems.
KEY WORDS : Rodent management, knowledge, attitudes, practices, Tanzania, Ethiopia.
INTRODUCTION
control is considered the responsibility of farmers who
conduct control activities individually, and rarely on a
collective basis. However, in many situations, farmers
In East Africa, rodent pests are considered a major
have few effective technologies that can be used to reduce
problem in agriculture and public health (MAKUNDI et al.,
the impact of rodents on their crops. It has been reported,
1999). Rodents cause considerable economic losses in
however, that the socio-economic conditions and culture
staple crops, particularly tuber crops and cereals.
of farmers influence the rodent pest management prac-
MAKUNDI et al. (1991) reported losses of approximately
tices used (SUDARMAJI et al., 2003). Rodent pest manage-
15% in cereals in Tanzania. Damage of maize at sowing
ment therefore will also be influenced by the farmer's
and seedling stage was estimated to be 40-80% in
knowledge on variables affecting crop damage, the level
Morogoro, Tanzania (MWANJABE & LEIRS, 1997). Wide-
of crop susceptibility, the rodent pest population during
spread crop damage and losses were reported in agricul-
the most susceptible crop stage and how much farmers
tural land during rodent outbreaks in Kenya (TAYLOR,
are prepared to control the pests. We conducted a study in
1968). Rodent outbreaks are regularly experienced in
Ethiopia and Tanzania to establish farmer's perceptions
Tanzania (HARRIS, 1937, MKONDYA, 1977; MWANJABE,
and knowledge of rodents as crop pests.
1990) and are associated with severe crop losses. Accord-
ing to SICHILIMA et al. (2003), considerable losses of tuber
crops occur in Zambia due to infestation by mole rats.
MATERIALS AND METHODS
Estimates of maize damage and losses in experimental
fields in Central Ethiopia indicated losses of about 26%
The study was carried out using a structured question-
(BEKELE et al., 2003). In Tanzania and Ethiopia, rodent
naire developed at Sokoine University of Agriculture,

154
Rhodes H. Makundi, Afework Bekele, Herwig Leirs, Apia W. Massawe, Winnie Rwamugira & Loth S. Mulungu
Morogoro, Tanzania. It was administered through inter-
and Ethiopia, the frequency of occurrence of rodent out-
views with farmers in five localities in Tanzania and Ethi-
breaks was high. Regular outbreaks were reported by
opia. Two of the five localities were in Tanzania
66.6% and 59.7% of farmers in Chunya and Mvomero,
(Mvomero District and Chunya District, in Central and
respectively, in Tanzania. In Ziway, Adami Tulu and
Southwest Tanzania, respectively). Three study locations
Gumer/Limmo, in Ethiopia, 48.7, 26.1 and 34.3% of
were in Central Ethiopia (Ziway, Adami Tulu and Gum-
farmers, respectively, reported regular rodent outbreaks
mer/Limmo). Sixty farmers were interviewed in each
(Table 1).
locality. The questionnaire consisted of a series of struc-
tured questions focussing on the following :
­ The size of cultivated field/family
­ Ranking of the status of different pests affecting crops
100
90
­ Ranking of crop damage according to crop phenology
80
70
Rodents
­ The proportions of fields damaged
60
Insects
50
­ The rodent management techniques/approaches, and
Birds
40
Nematodes
­ The most appropriate time to control rodents.
30
Others
20
The field staff in the respective areas identified the spe-
10
Percentage responding farmers
0
cies of rodents involved in crop damage. The data were
Very
Important
Less
Very
Important
Less
analysed using the SPSS computer software and are pre-
Important
Important Important
Important
Tanzania
Ethiopia
sented as percentages.
Fig. 1. ­ Ranking of the importance of different pests organ-
isms in maize in farmer questionnaires in Tanzania (N = 120)
RESULTS
and Ethiopia (N = 120)
The majority of farmers in Tanzania and Ethiopia are
small landholders cultivating fields that are 1 ­ 2 ha in
Asked about the most susceptible crop stages, 70.4%
size (96.4% and 99.3% in Tanzania and Ethiopia, respec-
and 82.5% of respondents in Tanzania and Ethiopia,
tively). Most of the respondents regarded rodents as the
respectively, considered seed retrieval to be most serious
number one pest they were least able to control. In Tanza-
(Fig. 2). About 60.6% of farmers in Tanzania and 50.8%
nia, 93.9% of respondents considered rodents as number
in Ethiopia indicated rodent damage to maize by seedling
one pest compared to only 3% who considered insects to
cutting to be serious. However, comparatively insignifi-
be number one pest. Comparative figures for Ethiopia
cant proportion of farmers in Ethiopia (1.6%) considered
show that 75% of the farmers considered rodents as very
rodent damage to maize cobs as severe, compared to 2%
important pests in their crops (Fig. 1). In both Tanzania
of the respondents in Tanzania (Fig. 2).
TABLE 1
Frequency of occurrence of rodent outbreaks in study sites
Percentage of farmers responding
Tanzania
Ethiopia
Gumer/
Frequency
Chunya
Mvomero
Ziway
Adami Tulu
Limmo
Regular (occurs every season)
66.6
59.7
48.7
26.1
34.3
Occasional (Occurs every few seasons)
33.3
30.3
48.7
73.9
45.7
Rare (Occurs every few years)
0
10.0
2.60
0
20.0
TABLE 2
Rodent control techniques practised by farmers in Tanzania and Ethiopia
Technique
Percent respondents practising specific rodent control technique
Tanzania
Ethiopia
Chunya
Mvomero
Adami Lulu
Ziway
Gumer/Limmo
Rodenticides
73.5
63.9
70.8
82.1
88.0
Field sanitation
4.0
2.0
12.5
2.6
17.3
Trapping
2.3
0
70.8
84.6
96.0
Hunting
2.0
0
66.7
64.1
0
Trapping and sanitation
6.1
2.0
-
-
-
Sanitation and rodenticides
17.2
0
17.2

Perceptions of rodents as crop pests
155
The majority of farmers ranked maize as the most
cides, the other control approaches were also more widely
affected crop in Tanzania, with 94.9% of respondents
practised than in Tanzania (Table 2). For maize, farmers
indicating that rodent damage to maize was very impor-
applied control measures more often before planting
tant, compared to 11.0, 3.0, 4.0 and 3.0% for sorghum,
(Table 3). In Ethiopia more farmers continued to apply
millet, cassava and pulses, respectively (Fig. 3). In Ethio-
control measures after planting maize (Table 3a) than was
pia, 42.9, 90.7, 4.5 and 98.7% of the respondents reported
the case in Tanzania (Table 3b).
rodent damage as very important in pulses, enset, barley
and potatoes, respectively, but not in teff (Fig. 3).
TABLE 3A
Time most appropriate for rodent control in Ethiopia
80
70
Maize
Percentage of farmers controlling rodents
Sorghum
Village
60
Cassava
Before planting
After planting
50
Pulses
Adami Tulu
91.7
8.3
40
Ziway
97.4
2.6
30
Gumer/Limm
96.3
3.7
20
Percent responding farmers
10
TABLE 3B
0
Time most appropriate for rodent control in Tanzania
Planting
Seedling
Vegetative
Harvesting
Percent of farmers controlling rodents
Village
90
Before planting
After planting
80
Maize
70
Chunya
98.3
1.7
Sorghum
Mvomero
97.6
2.4
60
Teff
50
40
DISCUSSION
30
20
Percent responding farmers
The damage to crops by rodents and the subsequent
10
yield losses at harvest is economically significant since
0
farmers in both countries are small landholders with little
Planting
Seedling
Vegetative
alternative incomes, other than from their staple and cash
Fig. 2. ­ Rank
crops. Farmers ranked rodents as number one pest, proba-
ing of rodent damage to crops at different crop growth stages
bly because they are least able to control them compared
in farmer questionnaires in Tanzania (top, N = 120) and in
to the other pests. In similar studies conducted in South
Ziway and Adami Lulu, Ethiopia (bottom, N = 120)
East Asia, SUDARMAJI et al. (2003) in Indonesia and TUAN
et al. (2003) in Vietnam reported that farmers perceived
100
rodents as the most important pests in their crops. The
Maize
90
Sorghum
regular occurrence of rodent outbreaks reported by the
80
70
Millet
majority (64.6%) of respondents in Tanzania is consistent
60
Cassava
with earlier reports (HARRIS, 1937; CHAPMAN et al., 1959;
50
Pulses
40
Enset
MKONDYA, 1977; MWANJABE, 1990; unpublished reports
30
Barley
Rodent Control Center, Morogoro, Tanzania, 1990-1998).
20
Potatoes
Occurrences of rodent outbreaks in East Africa were also
10
Percentage responding farmers
0
recorded by TAYLOR (1968) and KEY (1990). The farmer's
Very
Important
Less
Very
Important
Less
knowledge on manifestations of rodent damage to cereals
Important
Important
Important
Important
is also consistent with previous field observations
Tanzania
Ethiopia
reported by MAKUNDI et al. (1999); MWANJABE & LEIRS
Fig. 3. ­ Ranking of the importance of rodent damage to spe-
(1997) and recently by MULUNGU et al. (2003). MWAN-
cific crops in farmer questionnaires in Tanzania (N = 120) and
JABE & LEIRS (1997) reported maize damage of 40-80% in
Ethiopia (N = 180)
the seedling stage in Tanzania, while 20% damage was
reported in Kenya (TAYLOR, 1968). In experimental fields
In Tanzania, the majority of farmers indicated using
in Ziway, Ethiopia, where the study was conducted,
rodenticides (73.5 and 63.9% of respondents for Chunya
rodent damage to maize seedlings averaged 12.6% lead-
and Mvomero, respectively) to control rodents, with
ing to 26.4% yield loss (BEKELE et al., 2003). In the cur-
much lower proportions of respondents using field sanita-
rent study, 82.5% of farmers in Ethiopia experienced seed
tion/ and or field sanitation and rodenticides (17.1% of
retrieval by rodents, while 50.8% reported damage to
respondents for Chunya and none for Mvomero), hunting
seedlings. However, BEKELE et al. (2003) reported that an
and other strategies (Table 2). Although more respondents
important part of the damage to maize occurred after the
(70.8, 82.1 and 88.0% for Adami Tulu, Ziway and Gum-
seedling stage. It is therefore possible that seed retrieval
mer/Limmo, respectively) in Ethiopia were using rodenti-
by rodents has no serious consequences on final yields

156
Rhodes H. Makundi, Afework Bekele, Herwig Leirs, Apia W. Massawe, Winnie Rwamugira & Loth S. Mulungu
compared to seedling cutting since farmers replant to
Measures taken to control rodents are much more inte-
replace missing seedlings. As the rain season progresses,
grated in Ethiopia than in Tanzania, but generally they
it becomes impossible to replace cut seedlings and there-
rely on rodenticides to a greater extent than on the other
fore, any further seedling cut by rodents affects the poten-
control measures. The study also shows that M. natalensis
tial yield of the crop (MULUNGU, 2003). Most farmers
and Tatera leucogatser are the major pest species in the
applied control measures before rather than after planting,
study localities in Tanzania, whereas Arvicanthis dem-
which reduces the rodent infestation in the fields. From a
beensis and M. erythroleucus are the major pests in Ethio-
management point of view, this is the most appropriate
pia.
time particularly for a maize crop because the necessity to
replant, which is costly in terms of labour and the cost of
ACKNOWLEDGEMENT
replacing the seeds, will no longer be there.
Farmers practise a range of techniques to control
This study was financed by the European Union through the
rodents in both countries but control is based on eco-
STAPLERAT Project (ICA4-CT-2000-30029). We wish to thank
nomic reasons. The reliance on rodenticides appears to be
the various parties who were involved in the field studies in
related to effectiveness of this technique, particularly
localities in Ethiopia and Tanzania. Sokoine University of Agri-
when rodent population densities are high (M
culture (Pest Management Center) (Tanzania) and Universtity of
AKUNDI et
Addis Ababa (Department of Biology) (Ethiopia) provided the
al., 1999). However, the costs involved may be prohibi-
logistics that enabled perfection of the work. The authors also
tive for farmers to use this technique. In Tanzania, how-
wish to thank all the farmers who provided the information on
ever, government intervention in the form of free supplies
wich the publication is based.
of rodenticides, distribution and supervision of bait appli-
cation is done only during outbreaks of rodents but not for
routine rodent control. A comparison of the rodent control
REFERENCES
measures shows that there is more integration of
approaches in Ethiopia than in Tanzania, partially attrib-
BEKELE, A., H. LEIRS & R. VERHAGEN (2003). Composition of
rodents and damage estimates on maize farms at Ziway,
uted to lack of free supplies of rodenticides in Ethiopia.
Ethiopia. In : SINGLETON, G.R., L.A. HINDS, C.J. KREBS and
Farmers also apply control measures more intensively
D.M. SPRATT (eds). Rats, Mice and People : Rodent Biology
before planting than after planting. In the study localities
and Management. ACIAR Monograph 96 : 262-263.
in Tanzania, a good proportion of the farmers also contin-
CHAPMAN, B.M., R.F. CHAPMAN & D. ROBERTSON (1959). The
ued to apply rodent control measures after planting indi-
growth and breeding of the multimammate rat, Rattus
cating that rodents were considered a threat to the maize
natalensis (Smith) in Tanganyika Territory. Proceedings of
crop in the early stages of growth. This indicates farmer's
Zoological Society, London, 133 : 1-9.
awareness and knowledge on when reduction of damage
HARRIS, W.V. (1937).The grey field mouse. The East Aafrican
by rodent control is most effective and is consistent with
Agricultural Journal, 2 : 315-318.
earlier reports that more severe damage of the crop occurs
KEY, G.E. (1990). Pre-harvest crop losses to the African stripped
at planting and seedling stage (M
ground squirrel, Xerus erythropus, in Kenya. Tropical Pest
AKUNDI et al., 1999). In
Management, 36 : 223-229.
the study sites in Tanzania and Ethiopia, different rodent
M
species were responsible for the reported damage to
AKUNDI, R.H., N.O. OGUGE & P.S. MWANJABE (1999). Rodent
Pest Management in East Africa ­ An Ecological Approach.
crops. In both countries farmers have local names to dif-
In : SINGLETON, G., L. HINDS, H. LEIRS & Z. ZHANG (eds).
ferentiate between species. In Central Tanzania, Masto-
Ecologically-based Rodent Management. Australian Center
mys natalensis was the dominant and most abundant pest
For International Agricultural Research, Canberra,
species, where as in Southwest Tanzania, two species,
Australia : 460-476.
namely M. natalensis and Tatera leucogaster were
MAKUNDI, R.H., T.J. MBISE & B.S. KILONZO (1991). Observa-
responsible for the reported damage. In Central Ethiopia,
tions on the role of rodents in crop losses in Tanzania and
BEKELE et al. (2003) reported that Arvicanthis dembeensis
control strategies. Beitrage zur Tropischen Landwirtschaft
and M. erythroleucus were the major pest species whereas
und Veterinarmedizin (Journal of Tropical Agriculture and
Mus mohamet and Tatera robusta were minor pest spe-
Veterinary Science), 29(4) : 465-474.
cies. In Tanzania, 46.5% of farmers considered maize cob
MKONDYA, C.B. (1977). Preliminary proposals and hints for
approaches to outbreak evaluation and control strategies
damage to be serious, compared to only 1.6% in Ethiopia.
against rodent infestation in the Shinyanga outbreak foci in
This is possibly due to differences in the species found in
Tanzania. Dar es Salaam, Ministry of Agriculture, Crop
the study localities. In Tanzania, M. natalensis is a good
Development Division.
climber and has been reported to damage cobs on stand-
MULUNGU, L.S. (2003). Assessment of maize (Zea mays L.)
ing maize stems (MULUNGU, 2003). Similar findings have
damage and yield loss due to rodents in the field. PhD The-
not been reported for the species found in Ethiopia.
sis, Sokoine University of Agriculture, Morogoro, Tanzania.
178p.
MULUNGU, L.S., R.H. MAKUNDI & H. LEIRS (2003). Robustness
CONCLUSIONS
of techniques for estimating rat damage and yield loss in
maize fields. In : SINGLETON, G.R., L.A. HINDS, C.J. KREBS
The study shows that farmers in Tanzania and Ethiopia
& D.M. SPRATT (eds). Rats, Mice and People : Rodent Biol-
ogy and Management
. ACIAR Monograph 96 : 224-228.
experience severe crop damage due to rodents and that
MWANJABE, P.S. & H. LEIRS (1997). An early warning system
rodent control is undertaken for economic reasons.
for IPM based rodent control in small holder farming sys-
Rodents are considered the number one pest species in
tems in Tanzania. Belgian Journal of Zoology, 127 : 49-58.
both countries. Further, farmers are aware of the most
MWANJABE, P.S. (1990). Outbreaks of Mastomys natalensis in
vulnerable stages when severe attack of crops occurs.
Tanzania. African Small Mammal Newsletter, 11 : 1.

Perceptions of rodents as crop pests
157
SICHILIMA, A.M., M.S. ZULU & H. LEIRS (2003).The effects of
TAYLOR, K.D. (1968). An outbreak of rats in agricultural areas
Tephrosia vogelii and land preparation methods on mole rat
of Kenya in 1962. East African Agricultural and Forestry
activity in cassava fields. In : SINGLETON, G.R., L.A. HINDS,
Journal, 34 : 68-71.
C.J. KREBS & D.M. SPRATT (eds). Rats, Mice and People :
Rodent Biology and Management. ACIAR Monograph 96 :
TUAN, N.P., SS.J. WILLIAMS, P.R. BROWN, G.R. SINGLETON, T.Q.
254-255.
TAN, D.T. HUE, P.T.T. HA & P.T. HOA (2003). Farmer's per-
SUDARMAJI, G.R. SINGLETON, N.A. HERAWATI, A. DJATIHARI &
ceptions and practices in rat management in Vinh Phuc Prov-
RAHMINI (2003). Farmer's perceptions and practices in rat
management in West Java, Indonesia. In : S
ince, northern Vietnam. In : SINGLETON, G.R., L.A. HINDS,
INGLETON, G.R.,
L.A. HINDS, C.J. KREBS & D.M. SPRATT (eds). Rats, Mice
C.J. KREBS & D.M. SPRATT (eds). Rats, Mice and People :
and People : Rodent Biology and Management. ACIAR
Rodent Biology and Management. ACIAR Monograph 96 :
Monograph 96 : 389-394.
399-402.


Belg. J. Zool., 135 (supplement) : 159-165
December 2005
Rodent population fluctuations in three ecologically
heterogeneous locations in northeast, central and south-
west Tanzania

Rhodes H. Makundi, Apia W. Massawe and Loth S. Mulungu
Pest Management Center, Sokoine University of Agriculture, P.O. Box 3110, Morogoro, Tanzania
Corresponding author : Rhodes H. Makundi, e-mail : rmakundi@suanet.ac.tz; rmakundi@yahoo.com
ABSTRACT. Rodent population fluctuations and breeding patterns were investigated at localities in South-west,
Central and North-east Tanzania. The three localities are ecologically heterogeneous in vegetation types, rodent
species diversity, rainfall pattern and altitude. Capture-Mark-Release studies were conducted in 2001-2003 to com-
pare rodent species composition and population trends. In North-eastern Tanzania, species composition is diverse
and includes Mastomys natalensis, Lophuromys flavopunctatus, Grammomys dolichurus, Arvicanthis nairobae,
Praomys delectorum and Mus sp.. Five species were recorded in South-western Tanzania namely, M. natalensis,
Graphiurus murinus, Saccostomus elegans, Tatera leucogaster and Steatomys pratensis. In Central Tanzania M.
natalensis was dominant, but a few Lemniscomys griselda were captured. Rodent abundance fluctuations were dis-
tinctively seasonal, especially for M. natalensis in the three localities and T. leucogaster in South-west Tanzania. In
North-eastern Tanzania, L. flavopunctatus, G. dolichurus, A. nairobae and P. delectorum had low, but relatively sta-
ble populations throughout the year. In South-west Tanzania, population peaks of M. natalensis and T. leucogaster
were reached in the dry season (June-September). In Central Tanzania, breeding of M. natalensis was seasonal, with
highest population abundance during July-November. Female M. natalensis were reproductively active in January-
May and males had scrotal testes in December-June. No males were sexually active during July-November. Female
T. leucogaster in South-western Tanzania were reproductively active during November-April/May whereas sexually
active males appeared in the population during November-March. In view of the observed rodent population fluctu-
ations and breeding patterns, recommendations are given for pragmatic rodent control in South-west and Central
Tanzania and for plague in North-eastern Tanzania.
KEY WORDS : Mastomys natalensis, Tatera leucogaster, Grammomys dolichurus, Lophuromys flavopunctatus sp.,
Praomys delectorum, Saccostomus elegans, Graphiurus, Mus, Arvicanthis nairobae, Tanzania, population fluctuation,
crop damage, plague.
INTRODUCTION
nisms underlying such fluctuations are little known. For
this reason, ecologically heterogeneous areas may be
expected to exert different influences on resident rodent
In Tanzania, rodent populations exhibit a range of den-
species leading to varying levels of population fluctua-
sities within and between seasons and years. Temporal
tions. Changes in population density of rodents, particu-
variations in rodent density have been reported for vari-
larly the occurrence of high numbers at the most suscepti-
ous species, including the most common pest, Mastomys
ble stage of crop growth, may have severe consequences
natalensis. TELFORD (1989), LEIRS (1992) and CHRIS-
on crop damage and losses. In plague outbreak foci, these
TENSEN (1996) reported densities of 1125, 900 and 384
changes could also influence the severity of disease out-
animals/ha, respectively, in Morogoro, Tanzania. For a
break and dissemination. We therefore investigated popu-
species whose breeding characteristics are strongly
lation fluctuations of different species of rodents in three
dependent on rainfall patterns, such fluctuations are
ecologically heterogeneous localities in Tanzania in view
expected to be the rule rather than the exception
of providing a pragmatic approach for effecting control
(MAKUNDI & MASSAWE, 2003). Population fluctuations of
measures to reduce rodent damage to maize crop at sow-
M. natalensis are influenced by density dependent and
ing and seedling stage and for plague control.
density independent factors occurring simultaneously,
which regulate population size (LEIRS et al., 1997).
MATERIALS AND METHODS
Although much emphasis has been directed towards
understanding the effect of weather on rodent population
Study sites
dynamics in Tanzania (TELFORD, 1989; LEIRS, 1992;
MWANJABE & LEIRS, 1997; etc), the intrinsic characteris-
The study was conducted in northeast, central and
tics of the species and nature of habitats have received
southwest Tanzania. In Northeast Tanzania (NET), the
much less attention. For example, it is more common for
study was carried out at two locations, Manolo and Mag-
certain species populations to fluctuate more widely in
amba, in Shume Ward, Lushoto District, in the western
certain types of habitats than in others, but the mecha-
Usambara Mountains. Shume ward is located north of

160
Rhodes H. Makundi, Apia W. Massawe and Loth S. Mulungu
Lushoto town (04º 42' 16"S, 38º 12'16'E). The area has a
in Makuyu (MKA and MKB) had several species of
single, but asymmetrical rainy season, extending from
grasses, dominated by Pennisetum spp., R. cochinchinen-
October to May. November/December and March/April
sis and Cymbopogon spp. These study sites were more
are the wettest months (Fig. 2). The dry season is from
uniform in vegetation type, and therefore were less heter-
July to September. There were two permanent trapping
ogeneous. The rain patterns shows two rainy seasons; the
sites in the western Usambara Mountains. At the Mag-
short rainy season is from November to end of December,
amba locality, a grid was set adjacent to the montane rain
sometimes extending to January (Fig. 3b). In some years,
forest in an area reserved for agro-forestry at an altitude
no rains, or very little rains are received in this season.
of 1730 m above sea level (a.s.l.). The grid was located on
The long rains season (March-May) is characterized by
a steep slope and was planted with trees including Gravil-
heavy downpour except in some years when it is mar-
lea robusta and evergreen bushes. The Manolo study area
ginal.
was at an altitude of 1826 m a.s.l. The grid was set in per-
Although the three study sites in NET, CTZ and SWT
manent fallow land and bushes, surrounded by fields of
are ecologically heterogeneous, there is a strong human
maize, fruit trees and Gravillea robusta. Much of the grid
impact on vegetation and climate. The study sites in NET
had bushes whose vegetation was dominated by Rumex
were part of the tropical moist forest covering most of the
usambarensis. The Manolo and Magamba study sites are
Usambara Mountains. However, a large proportion of the
about 15 km apart.
natural montane forest has been cleared for agriculture and
In Chunya District, South-west Tanzania (SWT), the
pine plantations. The extended wet season, supplemented
climate is characterized by a long dry season from April to
by irrigation, allows intensive cultivation of various crops
November (Fig. 5). There is a single rainy season from
including cereals, beans, various vegetables, and fruits.
December to March, but the amount of rainfall varies con-
Temperatures are on average 18-220C and frost is regu-
siderably between years. Chunya District has a hilly land-
larly experienced on the ground when the temperature
scape, with vegetation characterised by miombo wood-
falls below 100C at night, particularly in July and August.
lands, scattered acacia trees and bush thickets. However,
In CTZ, the area is extensively cultivated, but bushes
the study area was in the flat low lands in the Lake Rukwa
and fallow patches of land intersperse between individual
basin within the Rukwa Rift Valley. It is characterised by
small fields, creating a vegetation mosaic. Flourishing
miombo woodlands opened for agriculture, although the
opportunistic weeds in and around crop fields are com-
soils are of low fertility. Wooded savannah grasslands of
mon during the rainy season, but these are ploughed into
Acacia commiphora bushlands and Brachystegia julber-
the soil or burned during land preparation in the following
nadia woodlands dominate the uncultivated areas. Crops
season. When the short rains are adequate, the fields are
in cultivated fields included sorghum, sunflower, maize
cultivated with maize and sorghum, but the main crop-
and cassava. The farms are generally small in size and fal-
ping season is during the long rains. In SWT, cultivated
low patches between fields are common, which increases
fields and grasslands for grazing dominate the landscape,
the heterogeneity of the habitats. Overgrazing, particularly
but large patches of woodland are scattered in the Lake
in the long dry season, leaves most of the landscape virtu-
Rukwa Valley.
ally bare of vegetation. The study sites consisted of two
grids, located in Chang'ombe village (08º 46'S 33º 18'E),
Rodent trapping
at an altitude of 600 m a.s.l. The two grids, coded CHB
and CHC, were about 4 km apart and are within the Lake
Two 100 x 100 m grids were set in each locality, except
Rukwa basin. Grid CHB was under maize cultivation for
in CTZ where there were a total of four grids (coded
several years before the study, but was maintained fallow
MKA, MKB, MLA and MLB). In each grid there were
throughout the study period. Grid CHC was communal
one hundred trapping stations at 10 m apart, with a single
land that had been maintained fallow, with many acacia
Sherman trap set per station. Trapping of rodents was
trees and bushes before and during the study. In the dry
conducted for three consecutive nights every month from
season, it was subjected to grazing by cattle and goats. The
May 2002 and April 2003 in NET, January 2001 to April
grids were dominated by grass species, particularly the
2003 in SWT and January 2001 to May 2003 in CTZ.
guinea fowl grass, Rotthboelia cochinchinensis and the
Captured animals were identified, marked by toe clip-
bobbin weed, Leucas martinnicensis.
ping, weighed and the breeding conditions were recorded.
In Central Tanzania (CTZ), the study was conducted at
In males, the breeding condition was determined by the
two sites in Mvomero District. Each of the localities
position of the testes, whether scrotal or abdominal. In
(Makuyu and Milama) (06º 22'S 37º 38'E) had two grids
females, the breeding condition was determined either by
(MKA and MKB for Makuyu; MLA and MLB for Mil-
signs of pregnancy by palpation, lactation, and/or perfo-
ama). The two localities were about 5 km apart and grids
rate vagina. Animals were released at the point of capture
were approximately 300-400 m apart. The grids in
soon after the data were recorded. Population density esti-
Makuyu were fields under maize cultivation before the
mates were determined in the programme CAPTURE.
study but were maintained fallow throughout the study
period. Fields under maize cultivation surrounded the
RESULTS
grids. The grids in Milama were located in a large land
block (>25 ha) owned by a local Catholic Mission. The
Table 1 shows the principal habitat type and species
land, which had not previously been under crop cultiva-
richness in the three study localities. Species richness was
tion, was fallow and had several tree species dominated
higher in the forest/agro-forestry habitats in NET and the
by kapok, but was also occasionally used for grazing cat-
savannah grassland, woodland and cultivated fields in
tle and goats by the neighbouring villagers. The study site
SWT, than in the extensively cultivated fields in CTZ.

Rodent population fluctuations in Tanzania
161
TABLE 1
Location of study sites, principal habitat type and rodent species richness.
Western Usambara
South west Tanzania
Central Tanzania
Mountains (NET)
(SWT)
(CTZ)
Species captured
Forest and agro-
Savanna grasslands, wood-
Cultivated fields, crop-fal-
forestry habitats
lands and cultivated fields
low mosaics
Mastomys natalensis
X
X
X
Mus sp.
X
Grammomys dolichurus
X
Arvicanthis nairobae
X
Lophuromys flavopunctatus
X
Praomys delectorum
X
Graphiurus sp.
X
Tatera leucogaster
X
Saccostomus elegans
X
Steatomys sp.
X
10 animals/ha. In NET, A. nairobae, a savannah species,
(a)
120
was relatively abundant in February-May compared to
M. natalensis
P. delectorum
other months. It is also obvious that more P. delectorum
100
G. dolichurus
were captured in Magamba, where the grid was adjacent
A. nairobae
to the natural forest than in the Manolo site. L. flavopunc-
80
L. flavopunctatus
tatus was found more abundantly in the secondary bush/
forest fallow land inter-phase. Fig. 2 shows the rainfall
60
pattern in the western Usambara Mountains. July-Sep-
tember were the driest months. Populations of M.
No. of animals/ha
40
natalensis peaked in August during the dry season and
gradually declined towards January. Breeding individuals
20
of M. natalensis occurred in the population in higher pro-
portions during March-August and March-June for males
0
and females, respectively (Figs 3a and 3b).
May June July Aug Sept Oct Nov Dec Jan Feb Mar Apr
2002
2003
700
600
500
3 5
M .n a t a le n s is
(b)
400
3 0
P . d e le c t o r u m
G . d o lic h u r u s
A . n a ir o b a e
300
2 5
Rainfall (mm)
a
L . f la v o p u n c t a t u s
a
l
s
/
h
200
2 0
m
1 5
100
.
of ani
o
N
1 0
0
5
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
2000
2001
2002
2003
0
M a y J u n e
J u l y
A u g
S e p t
O c t
N o v
D e c
J a n
F e b
M a r
A p r
2 0 0 2
2 0 0 3
Fig. 2. ­ Rainfall pattern in Lushoto, Western Usambara
Mountains, northeast Tanzania.
Fig. 1. ­ Rodent population density fluctuations at Manolo (a)
and Magamba (b) in the Western Usambara Mountains, north-
east Tanzania.
Fig. 4 shows the rodent population fluctuations in the
Rukwa Valley (SWT). High populations of M. natalensis
occurred, particularly between July and September, when
Fig. 1 shows the rodent population trends for different
densities were close to 200 animals/ha. The population
species in NET. In both study sites, the population density
density of Tatera leucogaster in SWT remained below 40
of M. natalensis was higher than for the other rodent spe-
animals/ha throughout the year. The abundance and popu-
cies. Populations of M. natalensis reached 100 animals/ha
lation densities of Graphiurus murinus, Saccostomus ele-
and 33 animals/ha in Manolo (Fig. 1a) and Magamba
gans and Steatomys sp. were relatively low. Fig. 5 shows
(Fig. 1b) study sites, respectively, in August. The lowest
the rainfall pattern in SWT, with a long dry season from
population density was in January, with less than 5 ani-
May to October. Figs 6a and 6b show the population
mals/ha in both sites. Praomys delectorum, Grammomys
trends of female and male M. natalensis and the propor-
dolichurus, Arvicanthis nairobae and Lophuromys flavop-
tions of individuals in breeding condition. Sexual activity
unctatus showed less marked fluctuations with less than
in females extended from February to May, with peak

162
Rhodes H. Makundi, Apia W. Massawe and Loth S. Mulungu
activity in March/April. Breeding activity was associated
the onset of the short rain season (Fig. 10). Breeding
with the onset of the rains and population peaks were
activity reached peak between September and February
reached in the dry season. For males, reproductively
for females (Fig. 11a) and December-March for males
active individuals appeared in the population in January
(Fig. 11b).
until end of April. T. leucogaster also showed a seasonal
activity in breeding, which was concentrated in the wet
300
season (November-April). Females with perforated
250
vagina, in lactating or in pregnant condition were found in
the population from November to April (Fig. 7). Repro-
200
ductive activity for males followed a similar pattern, with
150
sexually active individuals appearing in the population
from November to March (Fig. 8).
100
Rainfall (mm)
50
25
(a)
0
Total capture
ha 20
Abdominal testes
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Scrotal testes
2001
2002
2003
ed per 15
ur
Fig. 5. ­ Rainfall pattern in Chunya, Lake Rukwa Valley,
southwest Tanzania.
l
es capt 10
a
m
.
of
o
5
N
70
0
(a)
Total capture
May
June
July
Aug
Sept
Oct
Nov
Dec
Jan
Feb
Mar
Apr
60
Closed
Perforated
2002
2003
50
40
30
20
25
(b)
10
Total capture/ha
ha
No. captured females per ha
20
Perforated vagina
0
ed per
Closed vagina
ur
Jan
Oct
15
Feb
Mar
Mar
Apr
May
Jun
July
Aug
Sept
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
July
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
s capt
2001
2002
2003
l
e
a 10
m
e
f
.
of
5
o
N
120
(b)
Total capture
0
100
Abdominal
May
June
July
Aug
Sept
Oct
Nov
Dec
Jan
Feb
Mar
Apr
Scrotal
2002
2003
80
60
Fig. 3. ­ Reproductive conditions of male (a) and female (b)
Mastomys natalensis at Manolo, Western Usambara Moun-
40
tains, northeast Tanzania.
20
No. of captured males per ha
0
Jan
Feb
Mar
Mar
Apr
May
Jun
July
Aug
Oct
Sept
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
July
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
200
2001
2002
2003
180
M. natalensis-CHB
160
M. natalensis-CHC
Fig. 6. ­ Breeding condition of female (a) and male (b) Masto-
T. leucogaster-CHB
140
mys natalensis in Chunya, Lake Rukwa Valley, southwest
T. leucogaster-CHC
a
l
s
/
ha
120
i
m
Tanzania.
n 100
a
80
.
of
o
N
60
40
20
18
0
ha 16
b
r
y
n
l
p
b
r
y
n
l
p
b
r
Total capture
Jan
ug
ug
Fe
J
u
J
u
Ma
Apr
Ma
J
u
A
Se
Oct
Nov
Dec
Jan
Fe
Ma
Apr
Ma
J
u
A
Se
Oct
Nov
Dec
Jan
Fe
Ma
Apr
per 14
Closed
2001
2002
2003
es 12
Perforated
10
Fig. 4. ­ Rodent population density fluctuations in Chunya,
ed femal
8
Lake Rukwa Valley, southwest Tanzania.
6
aptur
4
2
No. of c
Fig. 9 shows rodent population density fluctuations in
0
n
b
n
l
y
n
b
n
l
y
n
CTZ. In this locality, only M. natalensis were captured in
Ja
Fe
Mar
Mar
Apr
May
Ju
Ju
Aug
Oct
Sept
Nov
Dec
Ja
Fe
Mar
Apr
May
Ju
Ju
Aug
Sep
Oct
Nov
Dec
Ja
Mar
Apr
large numbers, and showed dramatic fluctuations with
2001
2002
2003
highest peaks of population density in July- November.
Fig. 7. ­ Breeding condition in female Tatera leucogaster in
This coincided with the beginning of the dry season and
Chunya, Lake Rukwa Valley, southwest Tanzania.

Rodent population fluctuations in Tanzania
163
that have occurred (including vegetation and climate) as a
35
result of human activity, including opening of the wood-
30
Total capture
Abdominal
lands and wooded grasslands in SWT and clearing of the
25
Scrotal
natural forest in NET for agricultural development, have
ed males 20
affected the spatial distribution and temporal fluctuations
in density of rodent species. In NET, colonization by
15
savannah species of rodents, mainly M. natalensis and A.
. of captur 10
o
N
nairobae, is probably due to habitat changes brought
5
about by agriculture (MAKUNDI et al., 1999). In all the
0
three localities, a much more detailed study is required to
n
n
l
y
n
b
n
l
y
n
Ja
Oct
elucidate the species specific habitat requirements, which
Feb
Mar
Mar
Apr
May
Ju
Ju
Aug
Sept
Nov
Dec
Ja
Fe
Mar
Apr
May
Ju
Ju
Aug
Sep
Oct
Nov
Dec
Ja
Mar
Apr
not only determines the species distribution, but also tem-
2001
2002
2003
poral fluctuations.
Fig. 8. ­ Breeding condition in male Tatera leucogaster in
100
Chunya, Lake Rukwa Valley, south-west Tanzania.
(a)
90
a
80
h
250
per
70
s
l
e
a
60
m
Total capture
200
e
50
Closed vagina
ed f
/ha
ur
Perforated vagina
s
40
Milama A
150
capt
30
mal
Milama B
.
of
o
20
N
Makuyu A
100
10
Makuyu B
0
No. of ani
n
b
r
r
y
n
l
g
p
t
v
n
b
r
r
y
n
l
g
p
t
v
c
n
b
r
r
y
n
l
g
50
Ja
Fe
Ma
Ap
Ju
Ju
Ju
Ma
Ju
Au
Se
Oc
No
Ja
Fe
Ma
Ap
Ma
Ju
Au
Se
Oc
No
De
Ja
Fe
Ma
Ap
Ma
Ju
Au
2001
2002
2003
0
n
b
r
y
n
l
t
v
n
b
r
y
n
l
t
v
c
n
b
r
y
Ja
Fe
Ma
Apr
Ju
Ma
Ju
une
Ju
Ja
Fe
Ma
Apr
Ju
Ja
Fe
Ma
Apr
J
Aug
Sep
Oc
No
Ma
Aug
Sep
Oc
No
De
Ma
2001
2002
2003
70
(b)
Fig. 9. ­ Rodent population density fluctuations in Mvomero,
60
a
Morogoro, central Tanzania.
h 50
per
les
a 40
Total capture
450
ed m
Abdominal testes
ur 30
pt
Scrotal testes
a
400
c
.
of 20
o
350
N
10
300
)
m
0
r
r
l
r
r
l
r
r
l
m
y
v
y
v
y
250
n
b
n
g
p
n
b
n
g
p
c
n
b
n
g
Ja
Fe
Ma
Ap
Ju
Ju
Ju
Ma
Ju
Au
Se
Oct
No
Ja
Fe
Ma
Ap
Ma
Ju
Au
Se
Oct
No
De
Ja
Fe
Ma
Ap
Ma
Ju
Au
Rainfall (mm)
2001
2002
2003
infall ( 200
a
R
150
Fig. 11. ­ Breeding conditions of female (a) and male (b) Mas-
100
tomys natalensis in Mvomero, Morogoro, central Tanzania.
50
0
t
v
c
n
b
r
r
y
n
l
g
p
t
v
c
n
b
r
r
y
n
l
g
p
t
v
c
n
b
r
r
y
Oc
No
Ju
Ju
De
Ja
Fe
Ma
Ap
Ma
Ju
Au
Se
Oc
No
De
Ja
Fe
Ma
Ap
Ma
Ju
Au
Se
Oc
No
De
Ja
Fe
Ma
Ap
Ma
2000
2001
2002
2003
The relationship between rainfall, breeding activity and
Fig. 10. ­ Rainfall pattern in Mvomero, Morogoro, central
fluctuations in numbers is obvious in the three localities.
Tanzania.
Breeding occurred during and after the rains in all the
localities. The population peaks were reached in the dry
season. The influence of rainfall on breeding of M.
DISCUSSION
natalensis has been widely reported (e.g. LEIRS, 1992;
DELANY, 1974; TELFORD, 1989). In the three study sites,
Populations of M. natalensis underwent drastic
food resources probably influenced to a great extent the
increases and declines in numbers, particularly in SWT
fluctuations in rodent population density, but with greater
and CTZ. In SWT, M. natalensis and T. leucogaster were
influence in SWT and CTZ than in NET. It is also notice-
relatively abundant, but T. leucogaster maintained a con-
able that there were some local effects on breeding of
sistently low population density. The fluctuations of the
rodents, probably mediated through food resources in the
population of M. natalensis in CTZ followed a trend
three localities. Similar observations were made by
reported by LEIRS (1992). Tropical rodent species popula-
SWANEPOEL (1978) who reported that in an agricultural
tions are generally influenced by many factors, but rain-
area, P. natalensis were breeding during winter in irri-
fall is considered as a principal factor determining the
gated wheat fields, but not in the natural vegetation. It has
onset of breeding activity and reproduction (DELANY,
also been suggested that abundant high quality food in the
1986; LEIRS, 1992 and references therein). Temporal
absence of predation can induce population fluctuations
eruptions and extinctions of populations of M. natalensis
to outbreaks proportions (HUBERT & ADAM, 1985), proba-
were observed in the three localities in primarily fallow
bly due to among the factors, the effects on breeding. In
and agricultural land. It is apparent that land subjected to
most Muridae in Africa breeding occurs during the most
agriculture is a more variable habitat for M. natalensis
favourable time when resources are most abundant
and populations may fluctuate greatly as resources
(DELANY, 1972; CHEESEMAN & DELANY, 1979). The abun-
change in quality and quantity. The ecological changes
dance and quality of these resources most certainly

164
Rhodes H. Makundi, Apia W. Massawe and Loth S. Mulungu
increases survival and recruitment of young leading to
fluctuation patterns suggest that rodent management will
increases in population density.
be more effective in reducing seed depredation and seed-
The study sites in SWT and NET supported more spe-
ling damage if carried out before and during planting and
cies of rodents than in CTZ. Apart from the variations in
early during the seedling stage of the maize crop. TEL-
rainfall patterns in the three localities, differences in habi-
FORD (1989) suggested that rodent control should be con-
tat types were very pronounced. Habitat heterogeneity in
centrated between January and the onset of the long rains
terms of vegetation structure was more pronounced in
in CTZ, a duration of 2-3 months. However, this is not
SWT and NET than in CTZ, where extensive cultivation
practical for poor resource farmers, with only about 0.5-
was practised. The wooded savannah grasslands of Aca-
1.0 ha of maize fields. Since maize is most susceptible to
cia commiphora bushlands and Brachystegia julbernadia
rodent damage in the first 2 weeks after planting
woodlands and cultivated fields with fallow patches in the
(MAKUNDI et al., 1999), a single treatment with bromadi-
uncultivated areas were prominent habitats in SWT. This
olone or zinc phosphide, as currently practised in Tanza-
probably accounted for the higher species richness than in
nia may not be effective enough as fields are soon
CTZ. In NET the agro-forest fields adjacent to the natural
invaded by rodents 1-2 weeks after saturation baiting. A
moist forest and the forest itself, were the dominant habi-
more pragmatic approach arising from observations from
tats for rodent species. These two habitats were more
this study, and a general recommendation for all parts of
complex in vegetation structure and microhabitats, and
Tanzania experiencing similar problems, will be to carry
therefore could also explain the higher number of species
out saturation baiting with either broamadiolone or zinc
recorded.
phosphide or other recommended rodenticide three times;
the first a week before planting, the second during plant-
It is known that environmental factors can influence
ing and the third at the beginning of the second week after
populations of the same species at different locations
planting. This recommendation assumes that most or all
(KREBS, 1999). The more heterogeneous habitat com-
farmers shall carry out control activities simultaneously.
plexes in NET supported fewer individuals and had much
lower fluctuations of population density. The forest
In NET, human plague outbreaks occur in October-
dwelling species in NET showed less dramatic density
March when populations of rodents are low (Fig. 1a).
fluctuations indicating occupation of a much more pre-
This can be attributed to presence of "free" fleas, seeking
dictable and stable habitat. In general, species that inhabit
for alternative hosts at low rodent densities (MAKUNDI &
relatively stable habitats show less dramatic changes in
MASSAWE, 2003; MAKUNDI et al., 2003). A different
numbers compared to those inhabiting unstable habitats
approach for plague control in this locality is recom-
(O
mended. To avoid increasing mortality pressure on
DUM, 1966). These species have probably reached a
stable equilibrium in which fluctuations only occur within
already declining populations of rodents and also not to
limits set by the available resources, which do not show
increase the population of `free' fleas without a host in
marked seasonal variations. M. natalensis appeared to
the environment, the most practical approach will be to
respond to increased food resources in the aftermath of
intensify control of fleas with insecticides and applying
the rains by fast breeding and greater recruitment of
rodent control only in houses where Rattus rattus is dom-
young than the other species in SWT and NET. Mastomys
inant. This kind of approach is different from current
natalensis was found predominantly in the cultivated land
practices in which rodent control is intensified during
and in fallow land, which are much more unstable habi-
plague outbreaks within and outside houses.
tats compared to forest and woodlands. In NET, the inten-
sive cultivation throughout the year, sometimes with low
ACKNOWLEDGEMENT
or little ground cover and few fallow patches between
individual fields is probably not favourable for a large
We wish to acknowledge with gratitude the financial
build-up of rodent populations, including M. natalensis.
support from Sokoine University of Agriculture-Flemish
In SWT, T. leucogaster generally maintained a consist-
Inter-University Council (SUA-VLIR) Programme for the
ently low population with no major fluctuations in den-
work in the Usambara Mountains, NET and partial sup-
sity. This species occupied the same habitat as M.
port in Chunya, SWT; the European Commission (EU)
natalensis and experienced similar environmental condi-
funded STAPLERAT Project (ICA4-CT-2000-30029)
tions and yet the fluctuations were low. This could proba-
supported the work in CTZ and SWT. The support of
bly be attributed to competition for seeds and other
technical and field staff of the Pest Management Centre,
resources with the numerically dominant M. natalensis.
Sokoine University of Agriculture, is highly appreciated.
The same speculation could have accounted for the low
numbers of A. nairobae in NET where M. natalensis was
REFERENCES
relatively dominant numerically.
CHEESEMAN, C.L. & M.J. DELANY (1979). The population
Implications on rodent control
dynamics of small rodents in a tropical Africa grassland.
In SWT and CTZ, rodents are associated with severe
Journal of Zoology London, 188 : 451-475.
C
crop losses, particularly maize at sowing and seedling
HRISTENSEN, J.T. (1996). Home range and abundance of Masto-
mys natalensis (Smith, 1834) in habitats affected by cultiva-
stages. At the onset of the long rains in CTZ and SWT,
tion. African Journal of Ecology, 34 : 298-311.
rodent populations are still high for maize crop damage to
DELANY, M.J. (1986). Ecology of small rodents in Africa. Mam-
occur at sowing and seedling stage. Therefore, it is impor-
mal Review, 16(1) : 1-41.
tant to control rodents in both CTZ and SWT to prevent
DELANY, M.J. (1974). The ecology of small mammals. Edward
maize crop damage. The observed population density
Arnold, London.

Rodent population fluctuations in Tanzania
165
DELANY, M.J. (1972). The ecology of small rodents in tropical
D.M. SPRATT (eds). Rats, mice and people : Rodent biology
Africa. Mammal Review, 2 : 1-42.
and management. ACIAR Monograph 96 : 242-245.
HUBERT, B. & F. ADAM (1985). Outbreaks of Mastomys erythro-
MAKUNDI, R.H., B.S. KILONZO & A.W. MASSAWE (2003). Inter-
leucus and Taterillus gracilis in the Sahelo-Sudanian zone in
action between rodent species in agro-forestry habitats in the
Senegal. Acta Zoology Fennica, 173 : 113-117.
western Usambara mountains, northeast Tanzania, and its
KREBS, C.J. (1999). Ecological Methodology. Addison Wesley
potential for plague transmission to humans. In : SINGLETON,
Longman, Menlo Park, California.
G.R.; L.A. HINDS, C.J. KREBS & D.M. SPRATT (eds). Rats,
LEIRS, H. (1992). Population ecology of Mastomys natalensis
mice and people : Rodent biology and management. ACIAR
(Smith 1834) : Implication for rodent control in Africa. PhD
Monograph 96 : 20-24.
Thesis, University of Antwerp, Antwerpen, Belgium.
MWANJABE, P.S. & H. LEIRS (1997). An early warning system
LEIRS, H., N. STENSETH, J.D. NICHOLS, J.E. HINES, R. VERHAGEN
for IPM-Based rodent control in smallholder farming sys-
& W. VERHEYEN (1997). Stochastic seasonality and non-lin-
tems in Tanzania. Belgian Journal of Zoology, 127 : 49-58.
ear dependent factors regulate population size in an African
ODUM, E.P. (1966). Ecology. HOLT, RINEHART and WINSTON,
rodent. Nature, 389 : 176-180.
New York-Chicago-San Francisco-Toronto-London : 152p.
MAKUNDI, R.H, N.O. OGUGE & P.S. MWANJABE (1999). Rodent
S, E.P. (1966). Ecology. Holt, Rinehart and Winston, New York-
Pest Management in East Africa ­ An Ecological Approach.
Chicago-San Francisco-Toronto-London : 152 p.
In : SINGLETON, G., L. HINDS, H. LEIRS, H. & Z. ZHANG (eds).
SWANEPOEL, C.M. (1978). Demographic changes associated
Ecologically-based Rodent Management. Australian Center
with a Praomys natalensis population which is transient
For International Agricultural Research, Canberra,
between wheat field and grassland habitats. M.Sc. Thesis,
Australia : 460- 476.
University of Rhodesia, Salisbury.
MAKUNDI, R.H. & A.W. MASSAWE (2003). Review of recent
TELFORD, S.R. JR. (1989). Population biology of the multimam-
advances in studies of the ecology of Mastomys natalensis
mate rat, Praomys (Mastomys) natalensis at Morogoro, Tan-
(Smith 1834) Rodentia (Muridae) in Tanzania, eastern
zania 1981 ­ 1985. Bulletin Florida State Museum Biologi-
Africa. In : SINGLETON, G.R.; L.A. HINDS, C.J. KREBS &
cal Series, 34(6) : 249-287.


Belg. J. Zool., 135 (supplement) : 167-173
December 2005
Social organization of the Eastern Rock Elephant-shrew
(Elephantulus myurus) : the evidence for mate guarding

David O. Ribble1 and Michael R. Perrin2
1 Department of Biology, Trinity Univ., One Trinity Place, San Antonio, TX 78212
2 School of Biological and Conservation Sciences, University of Kwazulu - Natal, Pietermaritzburg, R.S.A.
Corresponding author : David O. Ribble, e-mail : dribble@trinity.edu
ABSTRACT. Understanding the costs and benefits of defending solitary females, or mate guarding, may be the key
to understanding the evolution of monogamy in most mammals. Elephant-shrews, or sengis, are a unique clade of
small mammals that are particularly attractive for studies of mate guarding. We studied the spatial organization of
Eastern Rock Sengis (Elephantulus myurus) in KwaZulu-Natal, South Africa, from August ­ December 2000. Our
objectives were to describe the home ranges of males and females using radiotelemetry, noting the sizes and overlap
of adjacent ranges and how the spatial organization changes through time. Males and females were spatially associ-
ated in monogamous pairs despite the fact that males contributed no obvious direct care to offspring. These monog-
amous associations persisted despite the fact that some males had home ranges large enough to encompass multiple
females. Males also had more variable ranges, perhaps because they spent more time at the periphery of their ranges
exploring for the presence of additional females. There was likely competition for females, as range shifts were
observed when male territory holders died or disappeared. It seems likely that this species is a model study organ-
ism to investigate the costs and benefits of mate guarding.
KEY WORDS : Social organization; Elephantulus myurus; mate guarding, monogamy.
INTRODUCTION
al., 2001). All studies have indicated that elephant-shrews
represent a monophyletic group (CORBET & HANKS, 1968;
Elephant-shrews or sengis (KINGDON, 1997) are a
TOLLIVER et al., 1989), and there exists no other group of
unique group of small mammals with no ecological or
closely related mammals that are all suspected to be
behavioral equivalents outside of Africa. All species feed
monogamous.
largely on invertebrates (RATHBUN, 1979; CHURCHFIELD,
Monogamy is one of the more evolved forms of social
1987; KERLEY, 1995), and all are highly cursorial and
organization in mammals and is found in fewer than 10%
capable of very fast locomotion (RATHBUN, 1979). The
of mammalian species (KLEIMAN, 1977; KLEIMAN &
smaller species usually produce only one or two offspring
MALCOLM, 1981). Monogamy in mammals has tradition-
that are born in a very precocial state. These life histories
ally been proposed to be due to either the necessity for
are more similar to small-bodied cursorial herbivores than
male care (obligate monogamy) or due to female disper-
similar-sized small mammals. Behaviorally, all of the 15
sion (facultative monogamy; KLEIMAN, 1977, 1981; WIT-
species of sengis from 4 genera are suspected to be
TENBERGER & TILSON, 1980; SLOBODCHIKOFF, 1984; BAR-
monogamous (RATHBUN, 1979). Of the species studied in
LOW, 1988). There is no evidence to suggest sengi males
detail, male and female pairs have overlapping territories
engage in any direct paternal care activities, especially
that result in monogamous associations, probably for life
since the young are so precocial.
(SAUER, 1973; RATHBUN, 1979; FITZGIBBON, 1995, 1997).
The objectives of this study were to describe the spatial
Territory defense is same-sex specific, and despite their
organization of the Eastern Rock Sengi (Elephantulus
nearly congruent territories, males and females spend lit-
myurus, Thomas and Schwann 1906) to determine if this
tle time together except during estrus, when the male con-
species exhibits monogamous association patterns in nat-
tinuously attends and follows the female (RATHBUN,
ural populations. Elephantulus myurus is distributed in
1979). Scent-marking appears to be an important compo-
southern Zimbabwe, western Mozambique, eastern Bot-
nent of pair bond maintenance (LUMPKIN & KOONTZ,
swana, and eastern South Africa on rocky outcrops in
1986; KOONTZ et al., 1999). Males are also known to
semi-arid savannahs (NEAL, 1995). Unlike other elephant-
occasionally visit neighboring territories, typically result-
shrews, E. myurus do not travel along a network of trails;
ing in intrasexual aggressive interactions (RATHBUN,
rather they use their swift cursorial gait to travel from
1979).
rock to rock (RIBBLE, personal observation). The primary
Recent molecular work indicates that sengis are a part
breeding season of E. myurus in southern Africa is
of an early radiation of African mammals that is repre-
August-March, with minimal breeding from April-July
sented by the extant golden moles, tenrecs, the aardvark,
(STOCH, 1954; WOODALL & SKINNER, 1989; NEAL, 1995).
hyraxes, sea cows, and elephants (HEDGES, 2001; MUR-
Females are typically anestrus from May to July (VAN
PHY et al., 2001). Consensus is building to place all of
DER HORST & GILLMAN, 1941). We described the social
these mammals in the Superorder Afrotheria (MURPHY et
organization of E. myurus by determining the home

168
David O. Ribble and Michael R. Perrin
ranges of males and females, noting the size and overlap
adults. Individuals were radiotracked for 4 ­ 8 days (mean
of adjacent ranges and if the spatial organization changes
= 6 + 1.8 days) during each session each month. At the
through time. Since no studies on the social organization
start of the study, radiotransmitters were removed after
of this species had been previously conducted, we were
each session. It became apparent that these elephant-
also interested in noting any features of the social organi-
shrews handled the radiotransmitters with no apparent
zation that would provide insight into the evolution of
problems. Pregnant females gained their expected weight
monogamy in elephant-shrews.
and successfully weaned offspring while radiocollared.
Towards the end of the study radiotransmitters were left
METHODS
on individuals for as long as 50 days. On average, indi-
viduals actually gained 0.14 ± 0.09 g per day while carry-
ing radiotransmitters (range ­0.03 to 0.83 g per day).
We studied the social organization of E. myurus on a
Radiotracking observations were taken at all hours of the
10-ha rock outcrop at Weenen Nature Reserve, located in
night and day because preliminary observations indicated
the KwaZulu-Natal province of South Africa
E. myurus could be active at any time. Individual loca-
(S28°52.5398' E030°00.2193'), from August through
tions were separated by at least one hour to avoid autocor-
December 2000. Weenen is a 4183-ha game reserve with
relation of data (SWIHART & SLADE, 1985).
habitats characterized by open, acacia savannahs with tall
grasses (e.g. Hyparrhenia spp. and Themeda triandra)
We collected home-range data on each radiotagged E.
and thicker woodlands (e.g. Acacia karoo) along valley
myurus during 2 to 4 (mean = 2.7) monthly sessions dur-
bottoms and riparian corridors (PERRIN & TAOLO, 1999a,
ing this study. Animals were trapped at the beginning of
1999b).
each radiotelemetry session to check their reproductive
status and replace radiotransmitters that quit working.
Individuals were trapped on the outcrop with Elliot alu-
The last radiotracking session was conducted in Novem-
minum traps baited with peanut butter and oats, and occa-
ber, and animals were trapped in December in order to
sionally supplemented with chopped-up insect parts (DU
remove their radiotransmitters. The minimum convex
TOIT & FOURIE, 1992). Captured elephant-shrews were
polygon (MCP) of all radiolocations and trap locations
recorded, ear-tagged, and a streak of hair dye was applied
during a monthly session was recorded as the home range
to either their back or sides for visual recognition. In the
for each individual for that month. We accepted statistical
early morning, E. myurus were readily observed basking
significance at P < 0.05.
on rocks, which made it easy to confirm that we had
marked all individuals in the population.
To document home ranges of individuals, we attached
RESULTS
"mouse-style" radiotransmitters (SM-1, AVM Instrument
Company, Ltd., Colfax, CA) around the necks of sengis
From 10 to 14 adult E. myurus were observed on the
with plastic cable-ties. This was accomplished by physi-
study outcrop each month (Table 1). The number of males
cally restraining the animals, avoiding the use of anesthe-
and females was most often equal except in September
sia. Radiotransmitters weighed on average 3.20 + 0.07
when the sex ratio was 7males :4 females. Females were
(1SE) g, which was 5.3 + 0.15 % of their average weight of
first observed lactating in September, and the first juve-
60.5 + 1.1 g. Individuals were radiotracked with an AVM
niles were observed and trapped in October.
receiver attached to a 3-element Yagi antenna. Many of the
Radiotelemetry indicated that individuals were active
radiolocations were confirmed with visual sightings
at any hour (Fig. 1), although activity was reduced in the
(33%). During the night individuals were visible with a
middle of the night (ca. 2300 ­ 0500h) and the middle of
strong headlamp and seemed unconcerned with our pres-
the afternoon (ca. 1200 ­ 1700h). Individuals were most
ence. Some locations were determined by removing the
active and furthest from their home range centre between
antenna coaxial cable from the receiving antenna and wav-
1800 and 2300h. During the morning activity period (ca.
ing the lead over the boulder where the elephant-shrew
0600 ­ 1100h), E. myurus spent most of their time bask-
was taking refuge. Individuals were recorded as "active" if
ing on the tops of boulders, presumably warming their
they were moving about or "resting" if they were station-
body temperature (MZILIKAZI et al., 2002).
ary. The Universal Transverse Mercator (UTM) coordi-
The mean monthly home-range size for males (3958 +
nates of locations were ascertained with a Garmin GPS 12
625 m2) was larger than females (2011 + 130 m2; P <
receiver (Garmin International, Inc.). The receiver was left
0.05). Across all monthly radiotelemetry sessions, seven
in place at the radiolocation for 10 min to calculate the
males had home ranges that were at least twice the size
average position determined from satellites during the
(range 8204 ­ 13487 m2) of the mean monthly female
entire 10-min interval. We conducted experiments that
home range of 2011 m2. The average number of intrasex-
indicated this 10 min point-averaging feature resulted in a
ual overlaps each month was 0.9 and 0.4 for males and
reading that was within 1.8 + 0.3 (1SE) m of subsequent
females, respectively, which was not significantly differ-
readings at the same spot (RIBBLE, unpublished data).
ent. Intrasexual overlap was greater for males than
After trapping the rock outcrop and conducting prelim-
females (18 vs. 2%; P < 0.05). The home-range data from
inary radio-tracking on 5 individuals in August, we
November 2000 are representative of the monthly pat-
attempted to radiotrack all adult individuals located in the
terns (Fig. 2), showing the lack of overlap between adja-
outcrop the next 3 months. Since individuals were marked
cent females. Female ranges tended to be overlapped by
with hair dye and visible in the morning hours basking on
only one male, but there were cases where one female
rocks and no individuals were observed outside the rock
range was overlapped by more than one male (see female
outcrop, we were confident that we were tracking all
777 overlapped by males 724 and 738; Fig. 2).

Social organization of the Eastern Rock Sengi
169
TABLE 1
Summary of spatial and temporal relationships of Elephantulus myurus at Weenen Game Reserve, South
Africa, 2000. Horizontal bars represent life span and location of individuals. A dotted line indicates individual
was alive, but not paired. Adult individuals (boldface numbers) within same boxes were presumably paired
and their offspring (italic numbers) are included in the same box. ( arrow = Home-range shift; d = disappear-
ance; X = mortality; O = offspring)
Individuals Gender August
Sept
Oct
Nov
Dec
719
d




718
766

783
O
d
784
O
720



d
724
775
O
d
776
O
806
777
738
805
770



d


773

d



721
774
771
723



d


772
O
d
736
810

100%
0
0
0
0
1
4
1
1
14
10
3
6
80%
3
17
21
17
6
7
8
13
13
8
3
13
e
60%
g
t
a

Resting
4
8
3
11
15
Active
r
cen

22
6
e
40%
P
3
27
20
6
10
4
22
30
24
7
7
7
10
9
20%
5
1
4
0%
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour of Day
Fig. 1. ­ Activity patterns of radio-collared Elephantulus myurus during the entire study. Sample sizes are indi-
cated for each bar.

170
David O. Ribble and Michael R. Perrin
Fig. 2. ­ Minimum convex polygons of male (thin-lined polygons) and female (thick-lined
polygons) Elephantulus myurus during November 2000.
Fig. 3. ­ Density contours for the fixed-kernel-density estimates of males during the entire
study. The thin-lined contours represent the 95% contours, and the thick-shaded contours
represent the 50% contours.
Despite the cases where female ranges were over-
female's home range during successive months. Based on
lapped by multiple males, each month there tended to be
these assertions, we assigned "putative" pairs each month
one primary male who 1) overlapped a majority of the
(Table 1). It was apparent that when a male or female dis-
female's range, and 2) consistently overlapped the
appeared or died, another individual would quickly move

Social organization of the Eastern Rock Sengi
171
into or shift their home range to occupy the abandoned
example, males may defend a territory containing a
home range. For example, in October male 736 was adja-
female and her offspring that could increase resource
cent to male 723 who was paired to female 771 (Table 1).
availability (KLEIMAN, 1977; RUTHBERG, 1983), provide
In November, male 723 had disappeared and male 736
protection from infanticide (VAN SCHAIK & DUNBAR,
shifted his home range to coincide with female 771. In
1990), and provide protection from predators (BARASH,
December, male 736 was found dead, likely due to preda-
1975; DUNBAR & DUNBAR, 1980). Any of these factors
tion. Another new male, 810, was trapped within the
could affect offspring and mother survivorship, and hence
home range of female 771. Four cases of range shifts by
be a benefit to males in defending and mating with a soli-
adults were observed during this study, 3 by males and
tary female. Sengis are very cursorial, often behaving
one by a female (Table 1).
more like small antelopes than typical small mammals
Based on their locations when first observed, we identi-
(RATHBUN, 1984), and some species build and maintain
fied offspring from 3 lactating females (Table 1). Two of
elaborate networks of trails through the ground litter
these offspring, 783 and 784 (Table 1), were observed by
within their territories. RATHBUN, (1979) proposed that
flashlight around 2200 h one evening with their mother,
the trail-building activities of male E. rufescens (Rufous
female 766. While each of the 3 females were lactating
elephant-shrew) may indirectly benefit his female and
we never observed any interactions between the lactating
offspring by providing efficient access to the territory for
female and her presumptive mate, nor did we detect the
foraging and predator escape. Since E. myurus do not use
two to be near each other with radiotelemetry.
trails, it is unlikely that females and their offspring benefit
from trail maintenance activities, although there could be
At the end of the study, all trapping and radiotelemetry
some other indirect benefits of the male's presence. It
locations were combined to estimate the overall home-
does seem clear, however, that the evolution of monog-
range size and intrasexual overlap. The fixed-kernel-den-
amy in E. myurus is not due to the necessity of male care
sity estimator (SEAMAN & POWELL, 1996) was also used
(obligate monogamy).
to calculate home-range size using all the location data.
The MCP estimates of home-range size using all data
In contrast to the necessity of male care in cases of
were significantly different between genders (male mean
obligate monogamy, facultative monogamy results when
= 9901 + 2593 m2; female mean = 3623 + 367 m2; P <
females exist at very low densities due to the dispersion
0.05) and significantly larger than the monthly averages
and quality of food resources, and males can subsequently
(P < 0.05 and P < 0.01 for males and females, respec-
monopolize only one female (KLEIMAN, 1977, 1981; WIT-
tively). The overall home-range size for males was 150%
TENBERGER & TILSON, 1980; SLOBODCHIKOFF, 1984; BAR-
larger, whereas females were 80% larger than the monthly
LOW, 1988). The essential feature of facultative monog-
average. The 95% fixed-kernel estimates were also signif-
amy is that both sexes are constrained by resource quality
icantly larger for males (mean = 11065 + 2576 m2; Fig. 3)
and distribution so that monogamy is the only option
than females (mean = 3132 + 220 m2; P < 0.05). For
available. If female ranges are widely dispersed, then
males, the mean number of same-sex overlaps (3.7) and
individual males may only be able to access one female
the mean percentage of intrasexual overlap (67%) were
and mate monogamously.
significantly greater for the overall combined data than
If the density of females affects the strategies of males,
the monthly averages for males (P < 0.01). There were no
a clear prediction of the facultative monogamy theory is
differences for females.
that the mating strategies of males should be responsive
to the density and availability of unpaired females. Recent
studies of so-called facultatively monogamous species
DISCUSSION
have indicated that males are not responsive to the availa-
bility of unpaired females (e.g. KOMERS, 1996), but rather
These data indicate that male and female E. myurus are
males remain faithful due to the benefits of mate guard-
spatially associated in monogamous pairs, yet males were
ing. The evolutionary principle of mate guarding is that
never observed in the same vicinity of females with off-
defending and mating with a single female during succes-
spring supporting the presumption that males contributed
sive reproductive events is a better option than roving to
no direct care to offspring. Similar results have been
mate with, or defending multiple females (PARKER, 1974;
observed with other species of sengis, including E. rufes-
WITTENBERGER & TILSON, 1980; BROTHERTON & KOMERS,
cens (RATHBUN, 1979), Rhynchocyon chrysopygus
2003). The benefits to mate guarding in mammals likely
(RATHBUN, 1979), Petrodromus tetradactylus (RATHBUN,
are due to the high costs of searching and or defending
1979; FITZGIBBON, 1995), and Macroscelides proboscides
multiple females. Recently, mate guarding has been pro-
(SAUER, 1973). These studies have led to the conclusion
posed to account for monogamy in Madoqua kirkii
that all 15 species of Macroscelidea may be monoga-
(Kirk's dik-dik). M. kirkii are socially and genetically
mous, making the sengis a very unique clade of mammals
monogamous (KRANZ, 1991; BROTHERTON et al., 1997),
in which every species is monogamous. Why all sengis
yet males exhibit no direct or indirect paternal behaviors
are monogamous is not clear.
that increase juvenile survivorship (BROTHERTON &
There is no evidence that male sengis engage in any
RHODES, 1996). The dispersion of females does not
direct parental care activities, in part because the young
appear to account for monogamy in this species either, as
are so precocial. Thus, it does not appear that direct male
many males have territories large enough to encompass
care explains monogamy in elephant-shrews. The benefits
multiple females yet do not (BROTHERTON & MANSER,
of the presence of the male to offspring survival and
1997), and mated males fail to respond to the presence of
female reproductive success may be more subtle than the
unmated females in adjacent territories (KOMERS, 1996).
obvious direct benefits of male care of offspring. For
The reported costs to males of mating with multiple

172
David O. Ribble and Michael R. Perrin
females include increased predation, male-male competi-
BARLOW, G. (1988). Monogamy in relation to resources. In :
tion, and the risk of being cuckolded by other males.
SLOBODCHIKOFF (ed), The ecology of social behavior. Aca-
Thus, the benefits of remaining with a single female and
demic Press, San Diego, CA : 55-79
assuring access to one female during estrus outweigh the
BROTHERTON, P.N.M. & P.E. KOMERS (2003). Mate guarding and
costs of mating with multiple females (K
the evolution of monogamy in mammals. In : R
OMERS, 1996;
EICHARD &
B
BOESCHE (eds), Monogamy : Mating Strategies and Partner-
ROTHERTON & MANSER, 1997). For females, it may be
there are costs associated with harassment by extra-pair
ships in Birds, Humans, and Other Mammals, Cambridge
University Press, Cambridge : 42-58.
males, which results in females accepting monogamy
BROTHERTON, P.N.M. & M.B. MANSER (1997). Female disper-
(BROTHERTON et al., 1997).
sion and the evolution of monogamy in the dik-dik. Animal
Mate guarding may also explain the social organization
Behaviour, 54 : 1413-1424.
of sengis. F
BROTHERTON, P.N.M., J.M. PEMBERTON, P.E. KOMERS & G.
ITZGIBBON (1997) demonstrated that male R.
chrysopygus that attempted to defend an additional
MALARKY (1997). Genetic and behavioural evidence of
monogamy in a mammal, Kirk's dik-dik (Madoqua kirkii).
female lost weight and could only temporarily defend the
Proceedings of the Royal Society of London B, 264 : 675-
larger territory. This was presumably due to the costs of
681.
defending the extra female, not the extra space, as some
BROTHERTON, P.N.M. & A. RHODES (1996). Monogamy without
males had territories twice the size of others (FITZGIBBON,
biparental care in a dwarf antelope. Proceedings of the Royal
1997). In this study, monogamous associations of E. myu-
Society of London B, 263 : 23-29.
rus persisted despite the fact that some males had home
CHURCHFIELD, S. (1987). A note on the diet of the rock elephant
ranges large enough to encompass multiple females. The
shrew, Elephantulus myurus, in Zimbabwe. Journal of Zool-
home ranges for the entire study were larger with more
ogy, London, 213 : 743-745.
overlap in part due to an increase in sample locations over
CORBET, G.B. & J. HANKS (1968). A revision of the elephant-
a longer time period. However, intrasexual overlap was
shrews, Family Macroscelididae. Bulletin of the British
significantly greater for males (67%) than females (18%;
Museum of Natural History, Zoology Series, 16 : 1-111.
P < 0.01) for these home ranges. This indicates that over
DUNBAR, R.I.M. & P. DUNBAR (1980). The pairbond in the
time, males have more variable and larger home ranges in
klipspringer. Animal Behaviour, 28 : 219-229.
which the boundaries are explored and expanded in dif-
DU TOIT, J.S. & L.J. FOURIE (1992). Bait preferences of the rock
elephant shrew. South African Journal of Wildlife Research,
ferent directions, supporting observations in other sengis
22 : 121-123.
that males make forays to monitor surrounding females
FITZGIBBON, C.D. (1995). Comparative ecology of two elephant-
(RATHBUN, 1979). There was likely competition for
shrew species in a Kenyan coastal forest. Mammal Review,
females, as range shifts were observed when male terri-
25 : 19-30.
tory holders died or disappeared. It seems likely there-
FITZGIBBON, C.D. (1997). The adaptive significance of monog-
fore, that this species is a model study organism in which
amy in the golden-rumped elephant-shrew. Journal of Zool-
to investigate the costs and benefits of mate guarding.
ogy, London, 242 : 167-177.
HEDGES, S.B. (2001). Afrotheria : Plate tectonics meets genom-
KOMERS & BROTHERTON (1997) argued that under-
ics. Proceedings of the National Academy of Sciences, 98 :
standing the costs and benefits of defending solitary
1-2.
females, or mate guarding, may be the key issue to under-
KERLEY, G.I.H. (1995). The round-eared elephant-shrew Macro-
standing the evolution of monogamy in mammals. Data
scelides proboscides (Macroscelidea) as an omnivore. Mam-
from this study and other studies would further indicate
mal Review, 25 : 39-44.
that the monogamous members of the Macroscelidea are
KINGDON, J. (1997). The Kingdon Field Guide to African Mam-
model organisms in which to experimentally manipulate
mals. Academic Press, London.
the costs and benefits of defending solitary females and
KLEIMAN, D.G. (1977). Monogamy in mammals. The Quarterly
therefore advance our understanding of the evolution of
Review of Biology, 52 : 39-69.
monogamy in mammals in general.
KLEIMAN, D.G. & J. R. MALCOLM. (1981). The evolution of male
parental investment in mammals. In : GUBERNICK &
KLOPFER (eds), Parental care in mammals, Plenum Press,
ACKNOWLEDGEMENTS
New York : 347-387.
KLEIMAN, D.G. (1981). Correlations among life history charac-
teristics of mammalian species exhibiting two extreme forms
The senior author was supported by an academic leave from
of monogamy. In : ALEXANDER & TINKLE (eds), Natural Sex-
Trinity University, San Antonio, Texas. The KwaZulu-Natal
ual and Social Behaviour. Recent Research and New Theory,
Nature Conservation Service sponsored this research and in par-
Chiron Press, New York : 332-344
ticular Peter Thompson and John Llewelyn made sure all went
KOMERS, P.E. (1996). Obligate monogamy without paternal care
well. The generous help in the field from Helen Ballew, Sibusio
in Kirk's dikdik. Animal Behaviour, 51 : 131-140.
Mncube, James Harvey, and Bev Eichbauer was appreciated.
KOMERS, P.E. & P.M.N. BROTHERTON (1997). Female space use
We would also like to thank Galen Rathbun for generously shar-
is the best predictor of monogamy in mammals. Proceedings
ing his enthusiasm and wealth of information and insight into
of the Royal Society of London B, 264 : 1261-1270.
the ecology and behavior of elephant-shrews.
KOONTZ, F.W., J.L. WELLINGTON & P.J. WELDON (1999). The
sternal gland of the rufous elephant-shrew, Elephantulus
REFERENCES CITED
rufescens (Macroscelidea, Mammalia) : anatomy, matura-
tion, and secretion composition. In : JOHNSON, MULLER-
SCHWARZE & SORENSON (eds) Advances in Chemical Signals
BARASH, D.P. (1975). Ecology of paternal behavior in the hoary
in Vertebrates, Plenum Publishers, New York : 163-171.
marmot (Marmota caligata) : an evolutionary interpretation.
KRANZ, K.R. (1991). Monogamy in the dik-dik. Applied Animal
Journal of Mammalogy, 56 : 613-618.
Behaviour Science, 29 : 87-105.

Social organization of the Eastern Rock Sengi
173
LUMPKIN, S. & F.W. KOONTZ (1986). Social and sexual behavior
SAUER, E.G.F. (1973). Zum sozialverhalten der kurzohrigen ele-
of the rufous elephant-shrew (Elephantulus rufescens) in
fantenspitzmaus, Macroscelides proboscideus. Zetschrift fur
captivity. Journal of Mammalogy, 67 : 112-119.
Saugetierkunde, 38 : 65-97.
MURPHY, W.J., E. EIZIRIK, S.J. O'BRIEN, O. MADSEN, M. SCALLY,
SEAMAN, D.E. & R.A. POWELL (1996). An evaluation of the
C.J. DOUADY, E. TEELING, O.A. RYDER, M.J. STANHOPE,
accuracy of kernal density estimators for home range analy-
W.W. DEJONG & M.S. SPRINGER (2001). Resolution of the
sis. Ecology, 77 : 2075-2085.
early placental mammal radiation using Bayesian phyloge-
SLOBODCHIKOFF, C.N. (1984). Resources and the evolution of
netics. Science, 294 : 2348-2351.
social behavior. In : PRICE, SLOBODCHIKOFF & GUAD (eds), A
MZILIKAZI, N., B.G. LOVEGROVE & D.O. RIBBLE (2002). Exoge-
New Ecology. Novel Approaches to Interactive Systems,
nous passive heating during torpor arousal in free-ranging
John Wiley and Sons, New York : 227-251.
rock elephant-shrews, Elephantulus myurus. Oecologia,
STOCH, Z.G. (1954). The male genital system and reproductive
133 : 307-314.
cycle of Elephantulus myurus jamesoni (Chubb). Philosoph-
NEAL, B.R. (1995). The ecology and reproduction of the short-
ical Transactions of the Royal Society of London, B, 238 :
snouted elephant-shrew, Elephantulus brachyrhyncus, in
99-126.
Zimbabwe with a review of the reproductive ecology of the
SWIHART, R.K. & N.A. SLADE (1985). Testing for independence
genus Elephantulus. Mammal Review, 25 : 51-60.
of observations in animal movements. Ecology, 66 : 1176-
PARKER, G.A. (1974). Courtship persistence and female guard-
1184.
ing as male time investment strategies. Behaviour, 48 : 157-
TOLLIVER, D.K., L.W. ROBBINS, I.L. RAUTENBACH, D.A. SCHLIT-
184.
TER & C.G. COETZEE (1989). Biochemical systematics of ele-
PERRIN, M.R. & C. TAOLO (1999a). Diet of introduced roan ante-
phant shrews from southern Africa. Biochemical Systematics
lope at Weenen Nature Reserve, South Africa. South African
and Ecology, 17 : 345-355.
Journal of Wildlife Research, 29 : 43-51.
VAN DER HORST, C.J. & J. GILLMAN (1941). The menstrual cycle
PERRIN, M.R. & C. TAOLO (1999b). Habitat use by a herd of
in Elephantulus. South African Journal of Medical Sciences,
introduced roan antelope in KwaZulu-Natal, South Africa.
6 : 27-47.
South African Journal of Wildlife Research, 29 : 81-88.
VAN SCHAIK, C.P. & R.I.M. DUNBAR (1990). The evolution of
RATHBUN, G.B. (1979). The social structure and ecology of ele-
monogamy in large primates : a new hypothesis and some
phant-shrews. Advances in Ethology, Supplement to Journal
crucial tests. Behaviour, 115 : 30-62.
of Comparative Ethology, 20 : 1-77.
WITTENBERGER, J.F. & R.L. TILSON (1980). The evolution of
RATHBUN, G.B. (1984). Elephant-shrews, Order Macroscelidea.
monogamy : hypotheses and evidence. Annual Review of
In : MACDONALD (ed), The Encyclopedia of Mammals. Facts
Ecology and Systematics, 11 : 197-232.
on File Publications, New York : 730-735.
WOODALL, P.F. & J.D. SKINNER (1989). Seasonality of reproduc-
RUTHBERG, A.T. (1983). The evolution of monogamy in pri-
tion in male rock elephant shrews, Elephantulus myurus.
mates. Journal of Theoretical Biology, 104 : 93-112.
Journal of Zoology, London, 217 : 203-212.


Belg. J. Zool., 135 (supplement) : 175-177
December 2005
Density and cover preferences of Black-and-rufous
elephant-shrews (Rhynchocyon petersi) in Chome Forest
Reserve, Tanzania

Stephanie Coster and David O. Ribble
Department of Biology, Trinity University, One Trinity Place, San Antonio, TX 78212
Corresponding author : David O. Ribble, e-mail address : dribble@trinity.edu
ABSTRACT. The objective of this study was to determine the density and habitat preference of the Black-and-
rufous elephant-shrew (Rhynchocyon petersi) in Chome Forest Reserve, Tanzania. Chome Forest (143km2) is
located in the South Pare Mountains and provides critical habitat for endangered R. petersi. Twelve 300m transects
were cut through the centre of the forest in an east-west direction and the number of elephant-shrew nests within 2.5
meters on each side of the transects was recorded. The mean number of nests per 100m transect (0.39 ± 0.47 [1SE])
translated to a density estimate of 19 elephant-shrews per km2 (SE=23). Nest sites tended to be found in areas with
greater than expected cover at the low (<5m) levels. These results indicate the population density of R. petersi is
lower in the Chome Forest Reserve than in most populations in the Eastern Arc Mountains. The reasons for this dif-
ference and the conservation implications are discussed.
KEY WORDS : elephant-shrew, sengi, conservation, density.
INTRODUCTION
ries, their presence is an indication of a healthy forest eco-
system.
Africa's tropical forests are home to large diversity of
Of the giant elephant-shrews, the most is known about
species, many of which are endemic to the African conti-
the Golden-rumped elephant-shrew (R. chrysopygus) and
nent. With increases in both human population and defor-
there are few records about R. petersi. The objective of
estation, more and more animals are becoming threatened
this study was to estimate the density of elephant-shrews
(MYERS, 1988). The elephant-shrews or sengis (order
based on nest counts and analyze habitat preferences in an
Macroscelidea) are one such group. There are 15 species
undisturbed forest reserve where R. petersi were known
in this strictly African mammal group, three of which are
to occur (STANLEY et al., 1996).
referred to as "giant" elephant-shrews and are of the
genus Rhynchocyon (Peters 1847). All three Rhynchocyon
species are considered threatened due to habitat destruc-
MATERIALS & METHODS
tion and fragmentation, including the species that is the
focus of this study, the Black-and-rufous elephant-shrew
Chome Forest Reserve is located within the Eastern
(R. petersi, Bocage 1880) (NICOLL & RATHBUN, 1990).
Arc Mountains (37o 58' 0.12" E, 4o 17' 60" S), a range on
the southeast coast of Kenya and the eastern coast of Tan-
The giant elephant-shrews share similar life histories in
zania. Chome Forest is made up of mostly wet montane
that they are diurnal insectivores that live in lowland and
forests (submontane, montane, and upper montane) with
montane forests and dense woodlands (RATHBUN, 1984).
elfin forest on high ridges and heathlands on rocky, acidic
They can be found in altitudes ranging from sea level-
soils. It covers approximately 142.8km2 and is situated on
2300m. While foraging they use their long proboscis to
the ridges and plateau of the South Pare Mountains in the
turn over leaf litter and dig up beetles, termites, other
district of Same in Kilimanjaro Region, Tanzania. The
insects and centipedes. Once the arthropods are exposed,
reserve was established in 1951 under the National Forest
the sengi's long tongue extends and scoops them up
Policy and Draft Act to ensure ecosystem stability
(KINGDON, 1997).
through conservation of forest biodiversity, water catch-
For shelter, the giant elephant-shrews build nests. The
ment and soil fertility. Because of the high annual rainfall
dimensions of their shelters are typically one meter wide
and pristine forest cover, the forest has a high water
with a body-sized bowl of 20cm long, 15cm wide and
catchment value and is an important resource for the 22
10cm deep (RATHBUN, 1979). Giant elephant-shrews live
surrounding villages in the catchment area. The altitude in
in monogamous pairs with defined territories and there-
the reserve ranges from 1250-2400m. The estimated
fore each animal can make and maintain up to ten nests in
annual rainfall is 1400mm. During the dry season fire is a
one territory with several nests in use at one time (FITZ-
problem because it replaces dry and lower forests with
GIBBON & RATHBUN, 1994). Their territories are typically
heath land. Historically fire was not a threat but fires have
about 1-1.7 hectare (RATHBUN, 1979). Because of their
increased with human activity near the forest. STANLEY et
dependence on undisturbed forest and their large territo-
al. (1996) noted the presence of R. petersi in the Chome

176
Stephanie Coster and David O. Ribble
Forest, but no study has documented the densities of these
layer (< 5m) elephant-shrews selected higher shade
elephant-shrews in this forest.
classes for nests than expected (2 = 8.14, d.f.= 2, P <
This study was conducted in the rainy season between
0.05; Fig. 1). In particular the shade class of 36-55% was
April 11th and 29th 2001. Chome Forest is accessible by
higher than expected (Fig. 1).
road, but the forest itself is navigated only by footpaths.
One such trail that is heavily trafficked bisects the middle
DISCUSSION
of the forest from a west to east direction and is used by
locals trekking from Mhero to Kanza or Mhero to Bombo.
The estimated population of R. petersi in Chome Forest
Because the path was established and because there had
was 19 (SE=23) elephant-shrews per km2. This estimate
been sengi sightings in the area, this trail was chosen to be
is most interesting when compared to density estimates of
a reference path for transects. In order to estimate ele-
R. petersi from other forest reserves in the Eastern Arc
phant-shrew density and habitat preference, transects
Mountains. HANNA & ANDERSON (1994) estimated popu-
were cut through the forest starting from the forest edge
lation density of R. petersi for seven study sites in the
on the western side (near Mhero boundary). Twelve
Eastern Arc Mountain range using similar techniques to
transects were cut perpendicular from the path, each
this study (Table 1). The population density of Chome
300m long and paced 500m apart. The first transect was
Forest Reserve is low when compared to these other sites
500m from the forest edge.
where R. petersi was found. According to HANNA &
Nest frequencies within 2.5 meters on each side of the
ANDERSON (1994), the available habitat for R. petersi
transect were tallied. Both newer (in use) and older nests
tends to be fragmented given its location at higher eleva-
were recorded. For more qualitative data, the number of
tions, which is typically on isolated mountains. Current
scraping/digging sites was also tallied. Using a density
logging and hunting pressures on these forests have fur-
conversion factor from FITZGIBBON & RATHBUN (1994)
ther exacerbated the lack of habitable areas for R. petersi.
study on R. chrysopygus, the population density per km2
Though Chome Forest is closed to timber harvesting and
was estimated.
hunting, pit saws and traps were sighted in the forest and
To examine if giant elephant-shrews were selecting
therefore human activity in the forest could be limiting
specific shade classes for their nests, percent canopy
the numbers of R. petersi. Also because of the proximity
cover was estimated every 20m along each transect at
of the village and the lack of a buffer zone, the forest is
each of three layers of canopy : < 5m, 5-15m, and > 15m.
isolated which could prevent immigration into the exist-
Percent canopy cover was divided into four shade classes,
ing population. FITZGIBBON (1994) suggested that for R.
of 0-15%, 16-35%, 36-55% and >56%. Canopy cover was
chrysopygus in Kenya, selective tree felling and pole cut-
also recorded at each observed nest site and compared to
ting in protected areas have little effect on elephant-shrew
available cover with 2 analysis.
densities but she warns that in unprotected areas, human
pressure may be more of a threat.
RESULTS
TABLE 1
The average number of nests found per 100m of
Estimated population densities of Rhynchocyon petersi in forest
transect was 0.39 (SE= 0.47). Using the density conver-
sites throughout Eastern Tanzania (from HANNAH & ANDERSON,
sion factor from the FITZGIBBON & RATHBUN (1994)
1994).
study, the estimated density was 19 per km2 (SE=23). By
extrapolation a liberal estimate for the whole reserve
Population density in pristine
Area
would be approximately 2700 R. petersi.
Forest site areas
areas of forest reserve (No./
(km2)
km2)
100
Pugu
11
79.3
90
Kazimzumbwi
29
67.1
80

Ruvu
98
42.7
Habitat Available
tal
Kiono
20
42.7
70
Nest Sites
f
to

Kisiju
2
0
60
e o
Kwamkoro
Unknown
> 0 *
50
tag
Kiwanda
Unknown
> 0 *
40
Chome ­ This study
143
19.0
rcen
30
e
P

* Not enough animals were captured for a density estimate, although
20
a few animals were observed.
10
0
0-15
16-35
36-55
>56
Rhynchocyon petersi chose nesting sites in areas of
Shade Class (%)
greater canopy cover than expected (Fig. 1), probably to
Fig. 1. ­ Shade classes for nest sites of Rhynchocyon petersi
avoid predators and to find sufficient leaf litter to con-
compared to the available habitat data collected every 20m.
struct nests. Nests were observed frequently at the base of
trees, and typically wild coffee was the predominant
shrub of the understory. Scraping and digging sites were
There was no evidence that elephant-shrews were
often found near coarse woody debris perhaps due to the
selecting specific shade classes in the middle or upper
higher proportion of prey found living in this substrate.
canopy layers (P > 0.10). However, in the lowest canopy
Finally, the forest edge seemed to have more nests (aver-

Status of Black-and-rufous elephant-shrew
177
age of 2.7 nests/100m) than the other areas of the forest
further help from Jo Anderson. We would like to thank the
(0.3 nests/ 100m), indicating the Black-and-rufous ele-
authorities at Chome Forest Reserve for granting access. The
phant-shrew may perhaps forage in both the forest as well
Department of Biology at Trinity supported our travel to the 9th
as in the surrounding heathland.
International African Small Mammal Symposium, Sokaine Uni-
versity of Agriculture, Tanzania, where we first presented these
The population density conversion factor we used was
results and we are grateful for that support. We would lastly like
determined from data collected on R. chrysopygus (FITZ-
to thank Galen Rathbun for his continued help and encourage-
GIBBON & RATHBUN, 1994). HANNA & ANDERSON (1994)
ment in our studies of sengis.
argued that R. petersi exist at lower densities than R.
chrysopygus. If this is the case, then the conversion esti-
REFERENCES
mator results in a liberal density estimate for R. petersi.
When extrapolating the population estimate for the entire
forest, we assumed that the whole forest area provided
FITZGIBBON, C.D. (1994). The distribution and abundance of the
golden-rumped elephant- shrew Rhynchocyon chrysopygus
adequate habitat for R. petersi, but there is evidence of
in Kenyan coastal forests. Biological Conservation, 67 :
disturbed areas of the forest which include burnings and
153-160.
heathland habitat which has not been known to support
FITZGIBBON, C.D. & G.B. RATHBUN (1994). Surveying Rhyn-
elephant-shrew nests. All of these surveys of R. petersi
chocyon elephant-shrews in tropical forest. African Journal
were conducted over short periods of time and ideally
of Ecology, 32 : 50-57.
longer studies should be conducted. However, in general
KINGDON, J. (1997). The Kingdon Field Guide to African Mam-
Rhynchocyon populations do not vary substantially over
mals. Academic Press. San Diego, CA : 142-152.
time (RATHBUN, 1979) so we are confident in our relative
HANNA, N. & J. ANDERSON (1994). Njule 92, Final Report,
comparisons between forests.
assessing the status and distribution of the black-and-rufous
The results of this study indicate that the population of
elephant-shrew. Unpublished report. Oxford University
expedition to Tanzania. Oxford, England : 67pp.
R. petersi in Chome Forest Reserve is low and isolated
MYERS, N. (1988). Tropical Forests and their Species : Going,
when compared to other populations in the Eastern Arc
going, ...? In : WILSON (ed), Biodiversity. National Academy
Mountains and thus long term conservation plans must
Press. Washington DC : 28-35.
safeguard the future of the forest. This study also showed
NICOLL, M. & G. RATHBUN (1990). African Insectivora and Ele-
that forest cover is essential to the elephant-shrew, pre-
phant-shrews : an action plan for their conservation. IUCN,
sumably to avoid predation, while leaf litter is crucial for
Gland.
nesting materials. With the proposed community conser-
RATHBUN, G.B. (1979). The social structure and ecology of ele-
vation agreement and the re-opening of the forest to
phant-shrews. Advances in Ethology, Supplement to Journal
selective timber harvesting, the elephant-shrew popula-
of Comparative Ethology, 20 : 1-77.
tion should be closely monitored.
RATHBUN, G. (1984). Elephant-shrews, Order Macroscelidea.
In : MACDONALD (ed), The Encyclopedia of Mammals. Facts
on File Publications, New York : 730-735.
ACKNOWLEDGEMENTS
StANLEY, W., S.M. GOODMAN & R. HUTTERER (1996). Notes on
the insectivores and elephant shrews of the Chome Forest,
The senior author of this study completed the project under
South Pare Mountains, Tanzania. Zoologische Abhandlun-
the guidance of the School for International Training (SIT) with
gen, 49 8 : 132-147.


Belg. J. Zool., 135 (supplement) : 179-181
December 2005
Evaluation of thiram and cinnamamide for protection of
maize seeds against multimammate mice, Mastomys
natalensis
, in Tanzania

Victoria Ngowo1, Loth S. Mulungu2, Jens Lodal3, Rhodes H. Makundi2, Apia W. Massawe2 and
Herwig Leirs
3,4
1 Rodent Control Centre, P.O. Box 3047, Morogoro, Tanzania
2 Pest Management Centre, Sokoine University of Agriculture, P.O. Box 3110, Morogoro, Tanzania
3 Danish Pest Infestation Laboratory, Skovbrynet 14, DK-2800, Kgs. Lyngby, Denmark
4 University of Antwerp Dept. Biology, Groenenborgelaan 171, B-2020 Antwerpen, Belgium
Corresponding author : Victoria Ngowo, e-mail : vngowo@hotmail.com or ngowov@yahoo.com
ABSTRACT. Farmers in Tanzania consider rodents to be the major vertebrate pest of maize, especially at planting
and seedling stages and annual losses are high. We evaluated the potential of two seed-dressing compounds, thiram
and cinnamamide, as rodent repellents to protect maize against damage by multimammate rats, Mastomys natalen-
sis
. In laboratory tests, the two compounds showed a strong repellent effect against M. natalensis and thus the
potential to protect maize seeds. The two compounds were evaluated in maize fields using Randomized Complete
Block Design (RCBD) with three replications. The results show that these repellents are effective for protecting
maize seeds against multimammate rats in the field, but in locations with high population of Tatera leucogaster,
seedlings are still damaged after emergence. Therefore, in such locations, other control measures, including applica-
tion of rodenticides just before seedling emergence may be necessary.
KEY WORDS : cinnamamide, economic loss, Mastomys natalensis, repellents, Tatera leocogaster, thiram, seed pre-
dation.
INTRODUCTION
damage to seeds by multimammate rats, M. natalensis. In
general, repellents may be classified as either primary or
In Tanzania, farmers consider rodents to be the main
secondary, according to their site of activity in the target
vertebrate pest (LEIRS et al., 2003). It has been estimated
species (ROGERS, 1978). Primary repellents provoke
that the annual economic loss due to rodents in maize
instantaneous responses through taste, olfaction, or irrita-
fields may amount to 42.5 million dollars (MULUNGU,
tion of the buccal cavity. Secondary repellents produce
2003), a loss that may be preventable by poisoning and
distressing effects after eating (e.g. gastrointestinal
trapping (STENSETH et al., 2001). However, poisoning and
malaise or other illness) which, if associated with a novel
trapping techniques are frequently ineffective, environ-
cue, may cause the subject to develop a conditioned aver-
mentally hazardous and socially unacceptable or uneco-
sion to a given food (GILL et al., 1995). Some repellent
nomic (MYLLYMÄKI, 1987). Thus alternative methods to
compounds have both primary and secondary activity
prevent rodent damage are needed.
(GILL et al., 1994). For example, the cinnamamide used in
The deterrence approach to rodent control is not new
the current study is considered bitter and does not smell
(NOLTE & BARNETT, 2000, CAMPBELL & EVANS, 1985),
good (GILL et al., 1995). Thiram has a bad strong smell
although emphasis on chemical repellents as a means of
which probably has olfactory repellence in rodents. The
reducing damage by rodents and other animals has
present study, therefore, reports the results of field tests
increased in recent years. The need for materials to pro-
with thiram and cinnamamide.
tect maize at planting and seedling stage is generally rec-
ognized (NGOWO et al., 2003). Ideal repellent seed dress-
MATERIAL AND METHODS
ing would prevent rodents from damaging the seed
(SIMMS et al., 2000). The toxic effect on rodent should be
Study locations
minimal; otherwise they will act as rodenticides and basi-
cally create vacant space that will attract other rodents.
Two field experiments were conducted in December,
Moreover, the repellents must not have phytotoxic effects
2002 and March, 2003 in Chunya (South -west Tanzania)
that would reduce germination rates (NOLTE & BARNETT,
and Mikese - Morogoro, (Eastern Central Tanzania),
2000, MYLLYMÄKI, 1987, CAMPBELL & EVANS, 1985).
respectively, during the maize cropping seasons. In
Preliminary laboratory studies from a wide range of
Chunya, maize is planted in November or December
botanic and synthetic repellents suggest that dressing
depending on the onset of rainfall, while in Mikese, it is
maize seeds with thiram and cinnamamide can reduce
planted in March. Initial trapping was carried out for three

180
Victoria Ngowo, Loth S. Mulungu, Jens Lodal, Rhodes H. Makundi, Apia W. Massawe and Herwig Leirs
consecutive nights one week before ploughing using 300
Y = differences in maize seed damage due location
ijk
and 200 Sherman live traps per night at Mikese and
different
Chunya, respectively, in order to determine the species
µ = Overall mean of maize seed damage due to location
composition and abundance. Therefore, there were a total
difference
of 900 and 600 trapping nights for Mikese and Chunya,
R = Replication effect
respectively. The traps were placed in 100 x 100m grids, on
i
10 trap lines, 10m apart, each with 10 trapping stations also
L = Location effect
j
10m apart. Peanut butter mixed with maize bran was used
(RL) = Main plot error
ij
as bait. Traps were inspected each morning and captured
A = Treatment effect
k
animals were identified and counted according to species.
(LA) = Location and treatment interaction effect
jk
In both locations, the experimental set up was a Random
(RLA) = Experimental error
Complete Block Design (RCBD) with three replications.
ijk
The replicates were 70 * 70 m maize fields. Untreated
seeds were planted in three control plots at each site. All
RESULTS AND DISCUSSION
plots were 100m apart. Other cultivated maize plots sur-
rounded the experimental plots. In Chunya, only thiram
The effect of seed dressing on maize seed depredation
was used to treat maize seeds. At Mikese, thiram and cin-
at Mikese is shown in Table 1. The results show that there
namamide were used for seed dressing separately and each
were highly significant differences (F = 203.5, df = 2, p =
was, tested in three individual fields. Maize seeds (STAHA
0.001) between treated and untreated plots at Mikese : the
variety is commonly used by farmers in the study areas) not
treated maize seeds were less predated compared to
formally treated with chemicals (fresh from a farmer) were
untreated maize seeds. For Chunya, the results were not
used in this study. Eighty grams of maize seeds were mixed
significantly different (F = 1.42, df = 1, p = 0.36) between
thoroughly with 0.8 grams of the respective chemical
treated and untreated maize seeds.
repellent (i.e. thiram and cinnamamide). The treated seeds
were left in the laboratory for 24 hours before planting and
TABLE 1
thereafter were planted in rows, 90cm apart and 60cm
The effect of dressing with Thiram and Cinnama-
between planting holes, with three seeds per planting hole.
mide on maize seed depredation by rodents in
Mikese, Morogoro and Chunya, Mbeya, Tanzania
Assessment of crop damage
Locations
Crop damage assessment was carried out at seedling
Treatments
stage, 10 days after planting. We used a non-stratified sys-
Mikese
Chunya
tematic row sampling technique to assess damage as
described by M
Control
52.10 ± 3.51a
38.4 ± 9.03a
WANJABE & LEIRS (1997) and MULUNGU et
al. (2003). The sampling units were maize rows; four rows
Thiram
27.53 ± 4.55b
46.8 ± 3.24a
Cinnamamide
26.41 ± 1.13b
-
apart, leaving out the two outer rows. The assessor walked
along maize rows across the plot, counting seedlings at
Means followed by the same letter are not significantly
each hole. Since three seeds had been planted per hole, we
different from one another at the 95% probability level.
calculated the difference between observed and expected
number of seedlings based on two assumptions, viz. the
In Chunya, the amount of rainfall was low and erratic,
germination is 100% and other factors remain constant. The
causing sporadic germination. During the evaluation, the
difference, therefore, was expressed as percentage damage.
distribution and amount of rainfall was an important factor
Data Analysis
that influenced rodent damage to maize seed germination
(MULUNGU, 2003). Therefore, rodent damage to seeds and
The data were analyzed in a general linear model with
seedlings appeared to depend on the duration of germina-
maize seed damage as the dependent variable and treat-
tion, particularly in Chunya. Similar observations were
ment as the factor interest, with field (and for thiram also
reported by KEY (1990) on the effect of rainfall on maize
site) as random factor (SAS, 1990). The damage data were
damage by squirrels during the seedling stage. In areas
subjected to Analysis of Variance (ANOVA) (SAS, 1990).
with erratic rainfall germination is sporadic and hence,
The data were analyzed according to the following statis-
seeds and seedlings were available at intervals spreading
tical model at each location :
over several days. We compared depredation of untreated
Y = µ + R + A + (RA)
seeds with maize seeds treated with thiram in Chunya and
ij
i
j
ij
Mikese. The results show that depredation of untreated
Where :
maize seeds at both locations did not differ significantly (F
Y = Differences in maize seed damage
ij
= 5.05, df = 1, p = 0.09), suggesting that the extent of
µ = Overall mean of maize seed damage
rodent damage to untreated and treated seeds was similar.
R = Replication effect
However, the interaction between treatment and location
i
A = Treatment effect
was significantly different (F = 20.86, df = 1, p = 0.01).
j
This suggests that thiram treatment at Mikese was more
(RA) = Experimental error
ij
effective in preventing rodent damage than at Chunya.
Since thiram was tested in both locations, the combined
The differences between these two locations were
analysis was done by using the following model :
probably due to the rodent species present. In Chunya,
Y = µ + R + L + (RL) + A + (LA) + (RLA)
two rodent species, Tatera leucogaster and M. natalensis
ijk
i
j
ij
k
jk
ijk

Evaluation of thiram and cinnamamide for protection of maize seeds
181
TABLE 2
Rodent species composition at Mikese and Chunya (one week before planting).
Locations
Mikese
Chunya
Species
Captured
Trap nights
Composition Species
Captured
Trap nights
Composition
individuals
(%)
individuals
(%)
Mastomys natalensis
688
900
98.01
Mastomys natalensis
287
600
93.79
Tatera leucogaster
10
900
1.42
Tatera leucogaster
19
600
6.21
Lemniscomysspp
4
900
0.57
-
-
-
-
were most abundant (Table 2) and both predated on maize
blackbirds from eating rice. International journal of Pest
seeds and seedlings. In this location, maize seedlings
Management, 40(2) : 195-198.
were cut, probably by T. leucogaster.
LEIRS, H., A. SKONHOFT, N.C. STENSETH & H. ANDREASSEN
(2003). A bioeconomics model for the management of Mas-
Similar observations were reported in India where T.
tomys natalensis mice in maize fields. In : SINGLETON, G.R.;
indica caused damage to seedlings immediately after ger-
L.A. HINDS, C.J. KREBS & D.M. SPRATT (eds), Rats, mice
mination (RAO, A.M.K.M. personal communication).
and people : Rodent biology and management. ACIAR
Therefore, in locations with high populations of T. leu-
Monograph 96 : 358-361.
cogaster much higher damage will be expected in addition
KEY, A. (1990). Pre-harvest crop losses to the African striped
to that caused by M. natalensis. The initial trapping before
ground squirrel, Xerus erythropus, in Kenya. Tropical Pest
planting indicated that the population of rodents at Mikese
Management. 36(3) : 223-229.
was dominated by M.natalensis (98%) while at Chunya, it
MULUNGU, L.S. (2003). Assessment of maize (Zea mays L.)
was composed of M. natalensis (93%) and T. leucogaster
damage and yield loss due to rodents in the field. PhD Thesis
Sokoine University of Agriculture, Morogoro, Tanzania.
(6%). Other species occurred in relatively low numbers. At
280pp.
Mikese, therefore, there were fewer depredations of seed-
MULUNGU, L.S., R.H. MAKUNDI & H. LEIRS (2003). Robustness
lings in treated plots, most probably due to the absence of
of techniques for estimating rat damage and yield loss in
T. leucogaster. The discrepancies between these two loca-
maize fields. In : SINGLETON, G.R.; L.A. HINDS, C.J. KREBS
tions suggest that it is unlikely that a single repellent will
& D.M. SPRATT (eds), Rats, mice and people : Rodent biol-
be effective against all seed and seedling depredating
ogy and management. ACIAR Monograph 96 : 224-228.
rodent species. The results suggest that thiram and cin-
MWANJABE, P.S. & H. LEIRS (1997). An early warning system
namamide are effective against M. natalensis after seedling
for IPM-Based rodent control in smallholder farming sys-
emergence and that they can protect damage to maize
tems in Tanzania. Belgian Journal of Zoology, 127 : 49-58.
seeds and seedlings in the absence of T. leucogaster.
MYLLYMÄKI, A. (1987). Control of rodent problems by the use
of rodenticides : rationale and constraints. In : RICHARDS,
C.G.J. & T.Y. KU (eds), Control of mammal pests. Taylor &
ACKNOWLEDGEMENTS
Francis, London. New York Philadelphia : 83-111.
NGOWO, V., J. LODAL; L.S. MULUNGU; H. LEIRS; R.H. MAKUNDI
& A.W. MASSAWE (2003). Evaluation of Thiram and Cin-
Much of the work presented here was made possible through
namamide as potential repellents against maize seed depre-
assistance and cooperation of staff from Rodent Control Center
dation in the field by the multimammate rat, Mastomys
and Sokoine University Pest Management Center. The work was
natalensis, in Tanzania. In : SINGLETON, G.R.; L.A. HINDS,
carried out with financial support from the European Union
C.J. KREBS & D.M. SPRATT (eds), Rats, mice and people :
funded STAPLERAT (INCO-DEV ICA4-CT-2000-30029).
Rodent biology and management. ACIAR Monograph 96 :
260-261.
REFERENCES
NOLTE, D.L. & J.P. BARNETT (2000). A repellent to reduce
mouse damage to longleaf pine seed. International Biodete-
rioration & Biodegradation
, 45 : 169-174.
CAMPBELL, D.L. & J. EVANS (1985). Evaluation of seedling pro-
ROGERS, J.B. (1978). Repellents to protect crops from vertebrate
tection materials for reducing wildlife damage on BLM
pests : some considerations for their use and development.
lands in western Oregon : 1983 and 1984 : Olympia, Wash-
American Chemical Society Symposium Series, 67 : 150-
ington, Report.
165.
GILL, E.L., R.W. WATKINS, J.E. GURNEY, J.D. BISHOP, C.J.
SAS (1990). SAS Statistics Users Guide, Statistical Analysis Sys-
FEARE, C.B. SCANLON & D.P. COWN (1995). Cinnamamide :
tem, Fifth ed. SAS Institute Inc. Carry, pp 1028.
A nonlethal chemical repellent for birds and mammals. In :
SIMMS, L.C., C.E. MULLINS & M.J. WILSON (2000). Seed dress-
MASON, J.R. (ed) 2nd Repellents in Wildlife Management
ings to control slug damage in oilseed rape. Project report :
Symposium : Proceedings of a Symposium. Denver, Colo-
pp22.
rado, August 8-10 1995 : 43-51.
STENSETH, N.C.H., H. LEIRS, S. MERCELIS & P. MWANJABE
GILL, E.L., M.B. SERA, S.B. CANAVELLI, C.J. FEARE, M.E. ZAC-
(2001). Comparing strategies of controlling African pest
CAGNINI, A.K. NADIAN, M.L. HEFFERNAN & R.W. WATKINS
rodents : an empirically based theoretical study. Journal of
(1994). Cinnamamide prevents captive chestnut-capped
Applied Ecology, 38 : 1020-1031.


Belg. J. Zool., 135 (supplement) : 183-185
December 2005
Spatial patterns and distribution of damage in maize
fields due to mastomys natalensis in Tanzania

Mulungu, Loth S.1*, Rhodes H. Makundi1, Apia W. Massawe1, Robert S. Machang'u1, Victoria
Ngowo
2 and Herwig Leirs3,4
1 Pest Management Centre, Sokoine University of Agriculture, P.O. Box 3110, Morogoro, Tanzania.
2 Rodent Control Center, P.O. Box 3047, Morogoro, Tanzania
3 University of Antwerp, Department of Biology, Groenenborgerlaan 171, B-2020 Antwerpen, Belgium
4 Danish Institute of Agricultural Sciences, Department of Integrated Pest Management, Danish Pest Infestation Laboratory,
Skovbrynet 14, DK-2800, Kgs. Lyngby, Denmark
Corresponding author : mulungu@suanet.ac.tz or lothmulungu@yahoo.co.uk
ABSTRACT. We describe the spatial distribution of rodent damage to maize seedlings in field studies in Morogoro,
Tanzania. The distribution of damage was assessed at the level of the planting hole (with three seeds per planting
hole) and at the level of the maize field (where the assessed units were plots of 10x10 planting holes). The most
abundant rodent species in the fields were the multimammate rat, Mastomys natalensis. At the planting hole level,
damage was fairly regular or random. At the field level, damage to seedlings was clustered irrespective of whether
the fields were situated in mosaic or monoculture surroundings, but the clusters were not more concentrated near
the edges or the near the centre of the field. We conclude that M. natalensis does not exhibit specific exhaustive
searching behaviour when feeding on seeds and seedlings in maize fields and that several local factors determine
the distribution of the damage.
KEY WORDS : Mastomys natalensis, spatial distribution, rodent damage, maize.
INTRODUCTION
within maize fields in Tanzania, and establish whether
these differ depending on the type of vegetation surround-
Rodents have the potential to breed quickly and infest
ing the fields.
crops leading to serious economic damage (FIEDLER,
1994). In Tanzania, damage to maize crop is largely
MATERIALS AND METHODS
attributed to Mastomys natalensis, and the Nile rat, Arvi-
canthis sp. (MAKUNDI et al., 1991). In one study, more
Locations and seasons
than 98% of the rodents found in maize fields were M.
natalensis (MASSAWE et al., 2003). Little is known on the
Field experiments were carried out during the cropping
spatial distribution of the damage caused by M. natalensis
season of 2000 and 2001, in two farms at Sokoine Univer-
in maize fields, although the spatial population patterns in
sity of Agriculture, Morogoro, Tanzania. The first farm is
Tanzania are known (LEIRS et al., 1996).
located at 6º50'S, 37º38'E at an altitude of 510 m above
KEY (1990) and REDHEAD & SAUNDERS (1980) reported
sea level (a.s.l.) and the second at 6º46'S, 37º37'E at 480
a strong correlation between rodent damage caused to
m a.s.l. The two areas have a bimodal rainfall pattern
maize and sugar cane and the presence of surrounding
(with a long and a short rainy season). The study was con-
uncultivated land. BUCKLE et al. (1985) and SCHAEFER
ducted during the long rains which is also the main maize
(1975) reported that at low population densities of Rattus
growing season. The seeds were sown in early March.
sp., damage in rice fields was variable, sometimes clus-
tered or sometimes evenly distributed over the field. At
Treatments
high population densities, the centre of the field was dam-
The study was carried out in ten plots of 70 x 70 m
aged, while border rows sustained little or no attack.
each. Six of the maize fields were located in mosaic land-
Proper sampling is essential for pest monitoring, sur-
scape of maize fields surrounded by fallow land; four
veillance and forecasting damage levels. Sampling meth-
were part of larger monoculture maize field. All fields
ods have to be simple and unequivocal and must find a
received similar standard agronomic treatments, i.e. early
compromise between costs and desired precision (KRANZ,
ploughing, application of Triple Super Phosphate ferti-
1993). Sampling methods, sample size and sampling pro-
lizer (20 kg P 0 /ha) before planting, and nitrogen ferti-
2 5
cedure, therefore, should be based on the spatial distribu-
lizer (40 kg N/ha) twice as a top dressing, three weeks
tion of rodent damage in order to ensure that a sample is
after sowing and again after booting stage. Three maize
representative for the entire population in a particular
seeds (of the local variety Staha®) were planted per hole,
field (APLIN et al., 2003). The aim of the current study
at a planting space of 90 x 60 cm between planting holes.
was to describe the spatial distribution of rodent damage
Weeding was carried out twice.

184
Mulungu, Loth S., Rhodes H. Makundi, Apia W. Massawe, Robert S. Machang'u, Victoria Ngowo and Herwig Leirs
Sampling procedures
with the larger sampling units (10 x 10 planting holes),
the distribution of damage appeared to be highly clustered
Crop damage assessment was carried out at seedling
(variance­to-mean values of 3.6 and 3.7 for mosaic and
stage, ten days after planting, by sampling every individ-
monoculture fields, respectively), regardless of whether it
ual planting hole in each field. The assessor walked
was in mosaic or monoculture fields. Fig. 1 shows, as an
across the field and recorded the number of seedlings at
example, a schematic representation of the spatial distri-
each sampled hole in a row. Since three seeds were
bution of damage in one mosaic field. The nature of dam-
planted per hole, damage was expressed as the proportion
age over the field can be readily seen, with areas of heavy
of missing emerged seedlings. At this stage, there were no
damage and other areas with hardly any damage at all.
other pests causing damage to the seedlings and all miss-
However, the figure does not suggest clustering of dam-
ing seedlings were therefore attributed to rodent damage.
age in the centre or at the edges of the field. Maps for
Germination failure due to drought or seed quality was
other fields show similar results.
assumed to be evenly distributed, but was also considered
of low importance in the experimental fields.
TABLE 1
Determination
Variance to mean ratio and spatial distri-
of rodent damage distribution pattern
bution of rodent damage in Mosaic and
Monoculture maize fields
The variance-to-mean ratio (s2/mean) of damage inten-
sity at sampling points was calculated in order to estimate
Field categories
the distribution of rodent damage in each field. When the
Mosaic fields
Monoculture
variance to mean ratio is large, the variation of damage
fields
distribution increases, meaning that damage is more
aggregate. A small variance to mean ratio indicates a
Individual planting hole
more regular damage distribution. KRANZ (1993) sug-
gested that damage with a variance-to-mean ratio from
Mean
0.49
0.83
0.7 - 1.3 would be classified as random, with a ratio >1.3
Standard Error
0.03
0.07
Median
0.50
0.73
aggregate or clustered and with a ratio <0.7 regular.
Minimum
0.39
0.70
We analyzed the distribution of damage in the field at
Maximum
0.56
1.00
two levels. At a first level, the planting hole was used as
Confidence level (95.0%)
0.07
0.19
the sampling unit. The number of missing seedlings was
Number of fields (N)
6
4
used as an indicator of damage at each planting hole. This
10 x 10 planting holes
could vary from 0 (indicating no seed removal by
rodents) to 3 (indicating 3 seeds removed/damaged). A
Mean
3.65
3.72
regular distribution at this sampling unit level would indi-
Standard Error
0.99
0.33
cate that rodents removed an equal number of seeds from
Median
3.70
3.73
each planting hole, while a clustered distribution would
Minimum
1.27
2.92
mean that the damage was higher in some planting holes
Maximum
7.90
4.52
while others were left untouched. It should be pointed out
Confidence level (95.0%)
2.55
1.04
that this level of analysis does not provide any informa-
Number of fields (N)
6
4
tion about the spatial distribution of damaged planting
Scale used for variance : mean ratio; <0.7 = regular, 0.7 - 1.3 = ran-
holes in relation to each other. For the second level of
dom, and >1.3 = cluster. Adopted from KRANZ (1993).
analysis, the field was divided into small areas of 10 x 10
planting holes. Damage was then calculated as the total
number of missing seedlings in each area. An aggregate
The clustered distribution at the field level indicates
distribution at this level would indicate that seeds were
that rodents are more active in some parts of the field than
removed or damaged in planting holes that are close to
in others. This corresponds to observations that also
each other. A regular distribution would indicate that
rodent captures in those same fields are spatially clustered
damage is spread uniformly over the field. At both sam-
(MASSAWE, 2003). Small within-field variation in soil and
pling levels, the mean to variance ratio of damage was
vegetation cover may contribute to such clustering, and
calculated for each field. Summary statistics of this ratio
this could be affected by e.g. land preparation methods.
were then calculated for all fields in a mosaic landscape
As observed in another study, M. natalensis can adjust its
and for all fields in a monoculture landscape. The spatial
feeding behaviour depending on prevailing local circum-
distribution of damage was also plotted on a map to visu-
stances such as cover and predation risk (MOHR, 2001).
ally verify where in the field any clusters would occur.
Our study showed no obvious edge effect with more or,
conversely, less damage near the field edges as observed
RESULTS AND DISCUSION
in other crops with other rodent species (e.g. BUCKLE et
al., 1985; SCHAEFER, 1975).
The mean variance-to-mean ratios at the planting hole
The random or regular distribution at the planting hole
level were 0.5 and 0.8 in mosaic and monoculture fields,
level is informative about the rodents' searching behav-
respectively (Table1). These results show a fairly regular
iour. The rodents do not necessarily dig up all seeds from
distribution of damage in the mosaic fields and a more
single planting holes, rather they seem to move between
random distribution in the monoculture fields. However,
planting holes without spending a long time searching at

Spatial patterns and distribution of damage in maize fields due to mastomys natalensis in Tanzania
185
each of them. In this way, the rodents may be actually
REFERENCES
thinning the seedling density but leaving one or more
seedlings at each hole. As seen at the field level, however,
APLIN, K., P. BROWN, J. JACOB, C. KREBS & G. SINGLETON
such thinning is not done evenly throughout the field.
(2003). Field Methods for Rodent Studies in Asia and the
Indo-Pacific. ACIAR Monograph, No.100 : 136pp.
60
BUCKLE, A.P., Y.C. YONG & H.A. RAHMAN (1985). Damage by
rats to rice in South-east Asia with special reference to an
integrated management scheme proposed for peninsular.
Acta Zoological Fennica, 173 : 139-144.
FIEDLER, L. (1994). Rodent Pest Management in East Africa.
Rome, Italy : 83pp.
40
KEY, A. (1990). Pre-harvest crop losses to the African striped
ground squirrel, Xerus erythropus, in Kenya. Tropical Pest
Management
, 36(3) : 223-229.
KRANZ, J. (1993). Introduction to sampling in crop protection.
In : KRANZ, J. & F. HOLZ (eds), Basics of Decision - Making
Row number
and planning for integrated pest management (IPM). Deut-
20
sche Stiftung für internationale Entwicklung (DSE), Federal
Republic of Germany : 33-39.
LEIRS, H., W. VERHEYEN & R. VERHAGEN (1996). Spatial patterns
in Mastomys natalensis in Tanzania (Rodentia, Muridae).
Mammalia, 60(4) : 545-555.
MAKUNDI, R.H., T.J. MBISE & B.S. KILONZO (1991). Observa-
0
tions on the role of rodents in crop losses in Tanzania and
0
20
40
60
80
control strategies. Beitrage zur Tropischen Landwirtschaft
Hole number in row
und Veterinarmedizin (Journal of Tropical Agriculture and
Veterinary Science)
, 29(4) : 465-474.
Fig. 1. ­ Spatial distribution of rodent damage along rows of a
MASSAWE, A.W. (2003). Effect of cropping systems and land
maize field in a mosaic landscape. Big bubbles indicate three
management practices on rodent population characteristics.
seeds were removed by rodents, medium size bubbles indicate
PhD Thesis, Sokoine University of Agriculture, Morogoro,
two seeds were removed, small bubbles indicate one seed
Tanzania : 176pp.
removed, and no bubble (empty) indicates no seeds were
removed. For clarity, we show on this figure only the observa-
MASSAWE, A.W., H. LEIRS, W.P. RWAMUGIRA & R.H. MAKUNDI
tions for every 5th row, starting from row 3.
(2003). Effect of land preparation methods on spatial distri-
bution of rodents in crop fields. In : SINGLETON, G.R., L.A.
HINDS, C.J. KREBS & D.M. SPRATT (eds), Rats, mice and
From our observations we conclude that M. natalensis
people : Rodent biology and management. ACIAR,
does not exhibit specific exhaustive searching behaviour
Canberra : 229-232.
when feeding on seeds and seedlings in maize fields and
MOHR, K. (2001). Feeding decision as an anti-predation strategy
that several local small-scale factors determine the distri-
in the African multimammate rat (Mastomys natalensis).
MSc Thesis, University of Copenhagen, Denmark.
bution of the damage.
REDHEAD, T.D. & I.W. SAUNDERS (1980). Evaluation of thallium
sulphate baits against rats in Queesland sugar-cane fields
ACKNOWLEDGEMENT
adjacent to different vegetation types. Journal of Protection
Ecology
, 2 : 1-19.
SCHAEFER, J. (1975). Field rat control as implemented by the
This research was supported by SUA-VLIR. We also appreci-
Philippine-German crop protection program. In : Proceed-
ate the excellent field assistance from E.Y. Mshuza from the
ings of All India Rodent Seminar. Ahmedabad, India : 345-
Rodent Control Center, Ministry of Agriculture, Morogoro.
351.


Belg. J. Zool., 135 (supplement) : 187-190
December 2005
Influence of land preparation methods and vegetation
cover on population abundance of Mastomys natalensis
in Morogoro, Tanzania

Apia W. Massawe1, Winifrida Rwamugira2, Herwig Leirs3, 4, Rhodes H. Makundi1 and Loth S.
Mulungu
1
1 Pest Management Centre, Sokoine University of Agriculture, P.O.Box 3110, Morogoro, Tanzania
2 Crop Science and Production, SUA, Morogoro, Tanzania
3 University of Antwerp, Department of Biology, Groenenborgerlaan 171, B-2020 Antwerpen, Belgium
4 Danish Pest Infestation Laboratory, Danish Institute of Agricultural Sciences, Department of Integrated Pest Management,
Skovbrnet 14, DK-2800 Kongens Lyngby, Denmark
Corresponding author : Apia W. Massawe, e-mail : massawe@suanet.ac.tz or apiamas@yahoo.com
ABSTRACT. A Capture-Mark-Release study was carried out in Morogoro, Tanzania, from April 1999 to April 2001
to investigate the effects of land preparation methods and cropping systems on population abundance of Mastomys
natalensis
in crop fields. Two land preparation methods (tractor ploughing; slash and burning) and two cropping
systems (mono-cropping with maize; inter-cropping with maize and beans) were included in the study. The experi-
mental design was a Complete Randomized Design with 2x2 factors, with two replicates. In slash and burn fields,
rodent population abundance and distribution were strongly influenced by vegetation cover regardless of the type of
cropping system. Higher rodent population peaks occurred in dense vegetation cover in slash and burn relative to
tractor ploughed fields. In contrast, there were no obvious associations between vegetation cover and population
abundance in the tractor ploughed fields, particularly in the mono-cropping system. A negative correlation between
vegetation cover and population abundance of M. natalensis was obtained in fallow land surrounding the crop fields
(r = -0.63; p 0.05). The results show that the effect of vegetation cover on population abundance of M natalensis
in crop fields is strongly influenced by the type of land preparation methods. Tractor ploughing and clearance of fal-
low land surrounding crop fields could be a useful method to reduce the invasion of crops by M. natalensis.
KEY WORDS : Mastomys natalensis, vegetation cover, land preparation methods, population abundance, Tanzania.
INTRODUCTION
logical characteristics of rodent populations. In this study
we tested the hypothesis that rodent population character-
Burning of vegetation in order to destroy rodent habi-
istics are influenced by land preparation methods and
tats has been a common practice in East Africa (G
land management practices. Thus we investigated the
REEN &
T
relationship between rodent population abundance and
AYLOR, 1975). In Tanzania, many farmers burn their
fields in the aftermath of the harvest or immediately
vegetation cover in different cropping systems and land
before ploughing. This probably changes the habitat for a
preparation methods.
short duration, but most likely it has no detrimental effect
on the future population size of rodents because burnt
MATERIALS AND METHODS
areas soon have new vegetation and are re-invaded rap-
idly. The type of farming practices affects the nature of
the habitat, shelter and population density of rodents
The study area was located at Solomon Mahlangu
(M
Campus (Mazimbu), Sokoine University of Agriculture,
AKUNDI et al., 1999). A mosaic of small plots of vari-
ous crops, intermingled with patches of fallow and per-
Morogoro, Tanzania (6o46'S 37o37'E, 480m above sea
manent grass-land, combined with minimum land prepa-
level). A capture-mark-recapture (CMR) study was con-
ration and subsequent flourishing of weeds, creates
ducted during the 1999 - 2000 cropping seasons. Eight
favourable conditions for rodents species especially M.
70x70m grids were prepared, consisting of 7 parallel
natalensis and results in a high degree of damage (T
lines, 10m apart, and 7 trapping stations per line (total of
AY-
49 trapping stations/grid), also 10m apart. One Sherman
LOR, 1968; MWANJABE, 1993; MYLLYMÄKI, 1989).
live trap was placed on each trapping station. A 200m-
Little attempt has so far been made to determine inter-
300m wide zone of fallow land separated the grids from
actions of rodents with the various cropping systems
each other. The grids were subjected to two types of crop-
found in many agricultural areas under different land
ping systems (mono-cropping and inter-cropping) and
management practices (YEBOAH & AKYEAMPONG, 2001;
two land preparation methods (tractor ploughing; slash
WHISSON, 1996). One of these interactions, for example,
and burning). The mono-cropping system consisted of a
is the influence of agricultural practices on certain eco-
monoculture of maize and the inter-crop consisted of a

188
Apia W. Massawe, Winifrida Rwamugira, Herwig Leirs, Rhodes H. Makundi and Loth S. Mulungu
mixture of maize and beans. The experimental design was
80
6
a Completely Randomized Design (CRD) with 2x2 fac-
)
tors replicated twice. The grids were ploughed in Novem-
70
Tractor ploughed monocrop
5
ber 1999 and February 2000 during the short and long
60
rain seasons, respectively. Tractor ploughing was done
4
using a disc plough at a depth of 30cm, a normal rooting
50
depth for most annual crops. Slashing was done manually
40
3
and the vegetation was left to dry and subsequently
30
burned. Maize sowing followed a standard procedure
2
(planting lines 90cm apart, plant holes 60cm apart, and
20
Vegetation cover classes
three seeds per planting hole). The bean crop was sown 3
1
10
weeks after the maize, at a spacing of 50cm x 10cm. All
Rodent population size (animals/0.5ha
necessary agronomic practices such as fertilizer applica-
0
0
tion and weeding were carried out in all the plots. Triple
Oct.7
April
Aug.
Oct.3
Jan.3
Super Phosphate (20kg/ha) and Nitrogen (40kg N/ha)
April.24
June.17
Aug.12
Nov.30
Dec.28
Feb.26
June.15
Nov.21
Feb.27
were applied before sowing and 3-4 weeks after sowing,
respectively.
Trapping was conducted in each grid for three consecu-
tive nights at intervals of four weeks. Traps were baited
80
6
)
with peanut butter mixed with maize bran and were
70
Slash and burn-monocrop
inspected early in the morning. Animals were marked by
5
toe-clipping. The trapping station, sex, weight, and repro-
60
ductive status of captured animals were recorded. Ani-
4
50
mals were later released at the station of capture. Plant
cover estimations were done during the monthly capture
40
3
session and were used to assess the effect of cover on
30
population size. In each grid, the assessor moved diago-
2
nally across the grid from point 1A to 7G and from 1G to
20
Vegetation cover classes
7A. At each point a qualitative estimation of ground cover
1
10
(other than maize crop) was made using a scale of 1-5, in
Rodent population size (animals/0.5ha
0
0
quadrate measuring 5m*5m. The corresponding values
were : 1 = no cover (< 15%); 2 = sparse cover (15-40%);
Oct.7
April
Aug.
Oct.3
Jan.3
3 = moderate cover (41-65%); 4 = dense cover (66-90%);
April.24
June.17
Aug.12
Nov.30
Dec.28
Feb.26
June.15
Nov.21
Feb.27
5 = very dense cover (>90%). In the surrounding fallow
land, cover estimation was done on all the four sides of
Fig. 1. ­ Rodent population abundance (bars) and vegetation
the grids. The relationship between vegetation cover and
cover (lines) in tractor ploughed fields (monocrop) and slash
rodent population abundance was investigated. Three
and burn fields (monocrop). Data were collected at intervals
parameters were used in the fittings : population size,
of four weeks.
vegetation cover in the field and vegetation cover in the
fallow land. Correlation analysis was performed between
the different factors using Pearson­moment product cor-
Figs 1 and 2 show that higher population peaks were
relation. Population data were log transformed to normal-
found in dense vegetation cover in slash and burn field
ize them before the analysis.
than in the tractor ploughed fields. There was no obvious
association between vegetation cover and population
abundance in the tractor ploughed fields, particularly in
RESULTS AND DISCUSSION
the mono-crop. A negative correlation between vegeta-
tion cover and population abundance of M. natalensis was
The population dynamics of M. natalensis in the study
obtained in the fallow land (Pearson Product ­ Moment
area followed an already established pattern (T
correlation; r = - 0.63, p 0.05).
ELFORD,
1989; LEIRS, 1995), but showed marked variations
Population sizes increased with increasing cover in the
between individual fields brought about by land prepara-
slash and burn fields and decreasing cover in the fallow
tion methods and cropping systems. Slashing and burn-
land (r2 = -0.62; p 0.05). In the tractor ploughed fields
ing, tractor ploughing, monoculture and intercropping
population size remained low as cover increased (r2 = -
resulted in differences in the habitats available to the
0.51; p 0.05).
rodents. Shelter and production of plant biomass were
specifically altered by the land preparation methods.
In the mono-cropped fields, rodent population size
Slashing and burning took place in November and new
increased with decreasing cover in the fallow land (N =
vegetative growth occurred immediately after the onset of
76; r2 = - 0.54; p 0.05), while in the inter-cropped fields
the short rains. This was followed by an increasing popu-
rodent population increased with decreasing cover in the
lation size, probably due to an invasion from the fallow
fields. A high rodent population size occurred in the inter-
land (MERCELIS & LEIRS, 1999) and early breeding, which
cropped fields when cover was low. Seasonal variation in
for M. natalensis occurs with the onset of short rains
population size in relation to vegetation cover was
(LEIRS et al., 1993).
observed. During the short rains and non-cropping season

Influence of land management on rodent populations
189
(dry season), population size increased with increasing
population abundance in the fallow land. MAKUNDI et al.
cover in the fields.
(2000) reported that agriculture is a major disturbing fac-
tor in any ecosystem, and further commented that the tim-
80
6
ing and intensity of this activity may affect the species
a
diversity and richness. This suggests that animals
70
Tractor ploughed-intercrop
5
migrated from the fallow land to the crop fields and estab-
60
lished new home ranges.
4
50
The opportunistic behaviour enables M. natalensis to
take advantage of changes in habitats, particularly in rela-
40
3
tion to food resources. According to TAYLOR & GREEN
30
(1976), when cereals and weed seeds were abundant, both
2
grass and dicotyledonous plants (as found in the fallow
20
Vegetation cover classes
land) were eaten sparingly or were absent in the diet of M.
1
10
natalensis. It has been suggested that the fallow land at
Rodent population size (animals/0.5h
certain stages during the growth of the crop is a less suita-
0
0
ble habitat compared to the crop fields.
Oct.7
April
Aug.
Oct.3
Jan.3
It is apparent that agricultural activities may increase
April.24
June.17
Aug.12
Nov.30
Dec.28
Feb.26
June.15
Nov.21
Feb.27
species richness (M. natalensis) whereas in the undis-
turbed fallow land the dominance of this species is
reduced. This observation conforms to general theories in
species succession (ODUM, 1971). In Australia, STICKEL
80
6
(1979) reported that in a crop field ­ hay mosaic (analo-
)
gous to crop ­fallow land mosaic in our study area) the
70
Slash and burn-intercrop
5
entire population of house mice moved from their long
60
established home ranges in a hay field to a field of ripen-
4
ing wheat where they established new home ranges. Other
50
studies have also shown the importance of farming prac-
40
3
tices on movements of populations of rodents. According
30
to NEWSOME (1969a, b) the growth and harvest of wheat
2
in Australia had major influence on the migration of
20
Vegetation cover classes
house mice.
1
10
Our study shows a strong association between popula-
Rodent population size (animals/0.5ha
0
0
tion size and vegetation cover in slash and burn fields. It
is apparent that these fields were less disturbed than the
Oct.7
April
Aug.
Oct.3
Jan.3
tractor ploughed fields. This suggests that populations of
April.24
June.17
Aug.12
Nov.30
Dec.28
Feb.26
June.15
Nov.21
Feb.27
M. natalensis build-up faster in slash and burn fields
(mono and intercrop fields) than in the tractor ploughed
Fig. 2. ­ Rodent population abundance (bars) and vegetation
fields probably due to higher survival and recruitment.
cover (lines) in the tractor ploughed fields (intercrop) and
Since the distribution of animals in the tractor ploughed
slash and burn fields (intercrop). Data were collected at inter-
fields was not random but was restricted to the edges
vals of four weeks.
(MASSAWE et al, 2003), it is an indication that there was
less migration and colonization of these fields irrespec-
The selection for suitable habitat by M. natalensis is
tive of the cover.
viewed to be a behavioural process, which maximizes fit-
It is apparent that surrounding fallow lands in crop
ness. Vegetation, apart from providing food resources,
fields are an important consideration in rodent pest man-
acts as cover for protection from predators. M. natalensis
agement. For example, studies in the Victoria Mallee,
generally avoid exposed places to reduce the risks of pre-
Australia, showed that fence-lines were the most impor-
dation (MOHR et al., 2003). The habitat changes were an
tant donor habitats because they provided abundant grass
important factor in the abundance of M. natalensis in the
seed early in the breeding season (SINGLETON, 1989;
different fields. In crop fields, the changes are usually
TWIGG & KAY, 1994). Rodent management in such fields
drastic and occur over a short period of time, which also
should aim at destruction of ground cover which affects
brings about changes in the rodent population densities.
rodents immediately by exposing them to predators and,
The different types of treatments (tractor ploughed versus
more slowly, by removing their food supplies. Popula-
slash and burn and mono versus inter-crop) were associ-
tions of M. natalensis, have been observed to increase
ated with a sequence of habitat changes both temporally
after cover removal in adjacent fields (GREEN & TAYLOR,
and spatially, and these are reflected in variation in the
1975). GREEN & TAYLOR (1975) therefore suggested that
rodent population abundance in the different fields.
any attempts to reduce rodent numbers over wide areas by
The fallow land with dense grass and weed cover
means of cover destruction would have to be coordinated
became more and more unfavourable for M. natalensis
so that all harbourage is removed at more or less the same
particularly when new vegetation got established in the
time.
slash and burn and tractor ploughed fields. This is
Our study also shows that following land preparation,
reflected in the negative correlation between cover and
animals escape into the fallow lands adjacent to crop

190
Apia W. Massawe, Winifrida Rwamugira, Herwig Leirs, Rhodes H. Makundi and Loth S. Mulungu
fields. Therefore, removal of the fallow patches and field
patch-dependent decisions and density-dependent GUDS.
sanitation measures, when conducted by all or the major-
Oikos, 100 : 459-468.
ity of farmers will reduce rodent population size in crop
MWANJABE, P.S. (1993). The role of weeds on population
fields.
dynamics of Mastomys natalensis in Chunya (Lake Rukwa)
valley. In : MACHANGU, R.S. (ed), Economic importance and
control of rodents in Tanzania
. Workshop proceedings, July
ACKNOWLEDGEMENT
6-8, 1992, Sokoine University of Agriculture, Morogoro :
34-42.
This research was supported by the Flemish Interuniversity
MYLLYMÄKI, A. (1989). Population dynamics of Mastomys
Council (Belgium) through the SUA-VLIR program.
natalensis (Smith) in relation to agricultural systems, inci-
dence of damage and implications in control strategies. Final
report. Denmark-Tanzania Rodent control Project, DANIDA
REFERENCES
Mission- Dar-es-Salaam, Tanzania.
NEWSOME, A.E. (1969a). A population of house mice temporar-
GREEN, M. & K.D. TAYLOR (1975). Preliminary experiments in
ily inhabiting a South Australian wheat fields. Journal of
habitat alteration as a means of controlling field rodents in
Animal Ecology, 38 : 341-360.
Kenya. In : HANSSON, L. & B. NILSON, (eds), Biocontrol of
NEWSOME, A.E. (1969b). A population of house mice temporar-
rodents, Ecological Bulletin, 19 : 175-282.
ily inhabiting a reedbed in South Australia. Journal of Ani-
L
mal Ecology, 38 : 361-377.
EIRS H., R.VERHAGEN & W.VERHEYEN (1993). Productivity of
different generations in a population of Mastomys natalensis
ODUM, E.P. (1971). Fundamentals of Ecology (3rd Edition).
rats in Tanzania. Oikos, 68 : 53-60.
W.B. SAUNDERS Co. London. 574pp
L
SINGLETON, G.R. (1989). Population dynamics of an outbreak of
EIRS, H. (1995). Population ecology of Mastomys natalensis
(Smith, 1834). Possible implication for rodent control in
house mice (Mus domestica) in the Mallee wheatlands of
Africa. Agricultural Edition No : 35. Belgium Administra-
Australia-hypothesis of plague formation. Journal of Zool-
tion for Development Cooperation, Brussels, 268pp.
ogy (London), 219 : 495-515.
STICKEL, L.F. (1979). Population ecology of house mouse in
MAKUNDI, R.H., N. OGUGE & P.S. MWANJABE (1999). Rodent
unstable habitats. Journal of Animal Ecology, 48 : 871-188.
pest management in East Africa - an ecological approach.
T
In : S
AYLOR, K.D. (1968). An outbreak of rats in agricultural areas
INGLETON, G.R., L.A. HINDS, H. LEIRS & Z. ZHIBIN,
of Kenya in 1962. East African Agricultural and Forestry
(eds), Ecologically- based rodent management (Edited by).
Journal, 34 : 66-77.
ACIAR Monograph No. 59 : 460-476.
TAYLOR, K.D. & M.D. GREEN (1976). The influence of rainfall
MAKUNDI, R.H., A. KALININ, A. PRZETAKIEWICZ & S.J. AHN
on diet and reproduction in four African rodent species.
(2000). Influence of land use on tenebroid beetle diversity.
Journal of Zoology, London, 180 : 367-389.
Proceedings of the Third Mashav course on the conservation
TELFORD, S.R. (1989). Biology of the multimammate rat, Prao-
of Biodiversity in Desert Ecosystem. May-June, 2000, Uni-
mys (Mastomys) natalensis at Morogoro, Tanzania, 1981-
versity of Ben Gurion of the Negev : 97-107.
1985. Bulletin of Florida State Museum Biological Science,
MASSAWE, A.W., H. LEIRS, W.P RWAMUGIRA & R.H. MAKUNDI
34(6) : 249-288.
(2003). Effect of land preparation methods on spatial distri-
TWIGG, L.E. & B.J. KAY (1994). The effect of microhabitat and
bution of rodents in crop fields. In : SINGLETON, G.R., L.A.
weather on house Mouse (Mus domestica) numbers and the
HINDS, C.J KREBS & D.M SPRATT (eds), Rats, Mice and
implications for management. Journal of Applied Ecology,
People : Rodent Biology and Management. Australian Cen-
31 : 657-665.
tre for International Agriculture Research, Canberra : 229-
WHISSON, D. (1996). The effect of two agricultural techniques
232.
on population of canefield rat (Rattus sordidus) in sugarcane
MERCELIS, S. & H. LEIRS (1999). The recolonization of "depop-
crops of North Queensland. Wildlife Research, 23 : 589-604.
ulated" maize fields by Rodent : implication for rodent con-
YEBOAH, S. & S. AKYEMPONG (2001). Factors influencing the
trol in Tanzania. War against rats, 8 : 11.
distribution of the mole rat, Crptomys zechi (Rodentia, Bath-
MOHR, K. (2003). Foraging of multimammate rats rat (Masto-
yergidae) in Ghana. African Journal of Ecology, 39 : 233-
mys natalensis) under different predation pressure : cover,
240.

Belg. J. Zool., 135 (supplement) : 191-196
December 2005
Biodiversity and ecology of small mammals (Rodents
and Shrews) of the "Réserve de Faune à Okapis", Demo-
cratic Republic of the Congo

Pionus G.B. Katuala1*, John A. Hart2, Rainer Hutterer3, Herwig Leirs4,5 and Akaibe Dudu1
1 Laboratoire d'Écologie et de Gestion des Ressources Animales (LEGERA), Université de Kisangani, Faculté des Sciences,
R.D. Congo
2 Wildlife Conservation Society, NY, USA
3 Zoologisches Forschungsinstitut und Museum Alexander Koenig, Section of Mammals, D-531113 Bonn, Germany.
4 Evolutionary Biology Group, University of Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium
5 Danish Pest Infestation Laboratory, Danish Institute of Agricultural Sciences, Department of Integrated Pest Management,
Skovbrynet 14, DK-2800 Kongens Lyngby, Denmark
Corresponding author : e-mail : Pionus G.B. Katuala, e-mail : pionuskatuala@gmail.com
ABSTRACT. We carried out a small mammal (rodents and shrews) inventory in the Okapi Fauna Reserve in the
Ituri Forest, Democratic Republic of Congo. Using snap traps and life traps we collected 1577 specimens of small
mammals belonging to 7 species of Soricomorpha and 23 species of Rodentia. Rodents included Hylomyscus par-
vus
, which previously only was known in D.R.Congo near Kisangani. The record of Crocidura congobelgica (Sori-
comorpha) is the second record after the description of the species.
Hybomys cf. lunaris (27.58%) and Praomys jacksoni (21.81%) were the most abundant and ubiquitous species on the
prospected habitats (mixed forest, monodominant forest, swampy forest, hill forest, secondary forest and fallows)
except in grassy fallows where Lophuromys dudui was the most abundant species. Other common species in the
Reserve included Deomys ferrugineus, Hylomyscus stella, H. aeta, Malacomys longipes and Praomys misonnei.
Shrews included Crocidura congobelgica, C. hildegardeae, C. olivieri and Scutisorex somereni. The Shannon-Wiener
and Simpson diversity indices show a large diversity and high equitability in small mammal communities. The sex
ratio for the principal species was near parity. All the species seemed to breed throughout the year but for most repro-
duction was less intensive in the dry season.
KEY WORDS : rodents, shrews, biodiversity, ecology, R.D. Congo, Réserve de Faune à Okapi.
INTRODUCTION
Study areas
The biodiversity of the rainforest of Congo remains
The RFO (13500 km2) stretches from 1° to 3° N and
poorly documented. This is particularly true for small
28° to more than 29° E. The elevation varies between 700
mammals. We carried out a study on small mammals
m a.s.l. in the extreme west and 1350 m on the highest
(rodents and shrews) in the Réserve de Faune à Okapis,
rocky hills between the Epulu-Nepoko Rivers (HART &
which is situated in the Ituri Forest, in the Northeast of the
BENGANA, 1997).
Democratic Republic of Congo. The Ituri forest (± 70000
Vegetation can roughly be classified as a mosaic of
km2) and Epulu locality are well known as the main habitat
tropical rain forests including primary forest, swamps or
of Okapi, Okapia johnstoni (Sclater, 1901). This forest also
marsh forests, fallows and secondary forests. Primary for-
is the habitat of a diversity of other mammals, and one of
ests are mixed forests dominated by Julbernadia seretii
the world's greatest diversity hotspots for forest ungulates
and Cynometra alexandri (Caesalpiniaceae) and mono-
(HART, 1985). The protection of the okapi's natural habitat
dominant forests dominated by Gilbertiodendron dew-
led in 1992 to the establishment of a Natural Reserve
evrei (Ceasalpiniaceae). Descriptions of these forests can
named "Réserve de Faune à Okapis" here "RFO" (Fig. 1).
be found in HART (1985), EWANGO (1994), NDJANGO
Since the middle of the 1980's, studies were under-
(1994), MAKANA et al., (1995) and MAKANA (1999).
taken in this forest in order to document the biodiversity
There are no detailed climatic records for the RFO as a
as a basis for management (HART, 1985; HART, 1985;
whole. However, BURTOT (1971) estimated the mean
SIKUBWABO, 1987; KATEMBO, 1990; CONWAY, 1992;
annual temperature as about 24 °C and annual rainfall
EWANGO, 1994; NDJANGO, 1994; TSHOMBE, 1994;
between 1700 and 1800 mm. April, May, August and
MAKANA et al., 1998; MAKANA, 1999; HART & CARRICK,
September are wettest while December, January and Feb-
1996; HART & BENGANA 1997). The present small mam-
ruary receive less than 100 mm of precipitation. HART &
mal survey was carried out in July-August 1993 and
CARRICK (1996) confirmed this climatic tendency in the
March 1994.
research stations of Afarama, Epulu and Lenda (Fig. 2).

192
Pionus G.B. Katuala, John A. Hart, Rainer Hutterer, Herwig Leirs and Akaibe Dudu
28°
29°


Fig. 1. ­ Map of the Réserve de Faune à Okapis, D.R.Congo, showing prospected localities
(adapted from HART & BENGANA, 1997).
MATERIAL AND METHODS
250
25
24.5
Temperature (°C)
Small mammals were collected during two periods : 28
200
June to 8 August 1993 and 2 to 28 March 1994 in five
24
localities : Afarama (AF) 1° 33' N, 28° 31' E, 800 m a.s.l.;
150
23.5
Bapukele (BA) Epulu (EP) 1° 24'N, 28° 35 E, 760 m, Lenda
(LE) 1°19' NS, 28° 38'E, 750 m and Njaro (NJ) 2°03'N,
Rainfall (mm)
100
23
28°50'E 960 m with Aketu hill 2° 5' N 28° 48' E, 1200 m.
22.5
50
22
Trap lines with trapping stations approximately 10 m
apart were set following paths, trails, streams or small riv-
0
21.5
ers. Three types of traps were used : Sherman LFA live
J
F
M
A
M
J
J
A
S
O
N
D
traps, and "Victor" snap traps and "Museum Special" snap
Month
traps. Two traps baited with the pulp of palm nut (Elaeis
guinensis) were placed at each trap station for tree consec-
Fig. 2. ­ Annual variation of rainfall and temperature in RFO :
utive nights. Traps were checked in the morning from
Epulu sector (after data from HART & CARRICK, 1996).
8 :30 am, and sometimes once again toward 5 :00 pm.

Small mammals of the Okapi Fauna Reserve
193
Each captured small mammal was preliminary identi-
together 62.65% of all rodent specimens. Moderately
fied, sexed, weighed and measured externally, sacrificed
abundant species Praomys misonnei (6.53%), Malacomys
and then fixed in 10% formaldehyde solution. Identifica-
longipes (5.96%) Hylomyscus aeta (5.77%), Deomys fer-
tion and collection of reproductive condition data were
rugineus (5.07%) and Lophuromys dudui (4.57%) total
completed in the laboratory. Diversity was evaluated with
27.90%. Occasional species Mus minutoides and
the Shannon-Wiener index, Simpson index and Equitabil-
Lophuromys luteogaster with respectively 2.28% and
ity index. Local diversity indices were compared between
1.84%. The last group includes twenty rare species (<1%
localities using the sectoral index H (see K
each) that together total only 5.33% of all specimens.
b
REBS, 1994;
RAMADE, 1984).
Among the rare species we found Hylomyscus parvus,
which grants RFO the status of being the second site in
RESULTS AND DISCUSSION
D.R. Congo where H. parvus is recorded, the first site
being Masako near Kisangani. Other noteworthy rare spe-
Species composition
cies are Dendromus mystacalis, Graphiurus lorraineus,
Colomys goslingi
and Praomys verschureni. With 23 spe-
We collected 1577 small mammals belonging to 30
cies recorded, the rodent species diversity in RFO seems
species. The collection includes 33 shrews (2.09%, 7 spe-
to be high and comparable to of the other forest blocks in
cies) and 1544 rodents (97.91%, 23 species). Seven
the Congo-basin. RAHM (1966) and DUDU (1991) men-
shrew species were caught. The most common species are
tioned 26 and 28 species respectively for Irangi and
Crocidura denti (0.57%) and C. olivieri (0.44%). The
Masako. Around Kisangani, new data (MUKINZI et al.,
record of C. congobelgica is the second record after the
2003) counted 30 species. The number of species in the
description of the species in 1916 (HOLLISTER, 1916).
RFO, reported here, is only a minimum value and could
Shrew diversity in RFO seems to be poorer than in
increase by surveys that are extended time and space, by
Masako Reserve (12 species) (DUDU, 1991). This might
including species that we have seen but not caught, as well
be due to the short survey period of our study and the use
as species reported by other authors as HATT (1940) and
of Victor and Museum Special snap traps that are not very
CARPANETO & GERMI, (1989). These species are Thryomo-
appropriate to catch shrews.
mys swinderianus, Atherurus africanus, Anomalurops
Among the rodents four groups occur. Dominant spe-
beecrofti, Anomalurus derbianus, A. pusillus, Idiurus
cies Hybomys cf lunaris (27.58%), Praomys jacksoni
macrotis and l. zenkeri; Funisciurus pyrrhopus, F. alexan-
(21.81%) and Hylomyscus stella (13.25%) represent
dri, Heliosciurus rufobrachiu, and Protoxerus stangeri.
TABLE 1
Number of specimens collected for different species in the Réserve de Faune à Okapis, DR Congo, in 1993 and
1994. The lower part of the table indicates a number of diversity indices.
Species
1993
1994
Total
Soricomorpha
Crocidura congobelgica Hollister, 1916
3
1
4
Crocidura denti Dollman, 1915
3
6
9
Crocidura hildegardeae Thomas, 1904
-
3
3
Crocidura latona Hollister, 1918
1
2
3
Crocidura olivieri (Lesson, 1827)
4
3
7
Crocidura sp.
2
2
Scutisorex somereni (Thomas, 1910)
2
2
4
Rodentia
-
-
Colomys goslingi Thomas & Wroughton, 1907
-
2
2
Cricetomys emini Wroughton, 1910
-
1
1
Dendromus mystacalis Heuglin, 1863
-
1
1
Deomys ferrugineus Thomas, 1888
44
36
80
Funisciurus anerythrus (Thomas, 1890)
2
1
3
Grammomys kuru
1
2
3
Graphiurus lorraineus Dollman, 1910
1
-
1
Hybomys cf. lunaris Thomas, 1906
173
262
435
Hylomyscus aeta (Thomas, 1911)
75
16
91
Hylomyscus parvus Brosset et al.,1965
-
1
1
Hylomyscus stella (Thomas, 1911)
88
121
209
Lemniscomys striatus (Linnaeus, 1758)
2
6
8
Lophuromys dudui Verheyen et al., 2002
39
33
7
Lophuromys luteogaster Hatt, 1934
4
25
29
Malacomys longipes Milne-Edwards, 1877
46
48
94
Mus minutoides A. Smith, 1834
5
31
36
Oenomys hypoxanthus (Pucheran, 1855)
1
9
10
Paraxerus boehmi (Reichenow, 1886)
1
5
6
Praomys jacksoni (De Winton, 1897)
118
226
344
Praomys misonnei Van der Straeten & Dieterlen, 1987
57
46
103
Praomys verschureni (Verheyen & Van der Straeten, 1977)
1
2
3
Rattus rattus (Linnaeus, 1758)
-
3
3
Stochomys longicaudatus (Thomas, 1915)
6
4
10
Number of specimens
677
900
1577
Number of species
23
29
30
Shannon-Weiner diversity index Ha
3.161
3.121
3.205

194
Pionus G.B. Katuala, John A. Hart, Rainer Hutterer, Herwig Leirs and Akaibe Dudu
TABLE 1
Number of specimens collected for different species in the Réserve de Faune à Okapis, DR Congo, in 1993 and
1994. The lower part of the table indicates a number of diversity indices.
Species
1993
1994
Total
Ha maximum (Hmax)
4.523
4.858
4.907

Equitability (E)
0.698
0.643
0.653
Simpson diversity index (D)
0.855
0.823
0.800
Sectoral index Hb
0.063
A lower number of species was recorded in at Afarama
TABLE 2
(11) and Lenda (12) where undisturbed primary forests
Sex ratio of the most common species Réserve de Faune à Oka-
prevail than in Bapukele (16), Epulu (22) and Njaro (24)
pis (RFO, this study) and Masako Reserve (DUDU, 1991)
where the regenerating habitats (fallows and secondary
forest) are common. Indeed, in these anthropogenous
RFO
Masako
environments, crops and rich herbaceous stratum provide
Species
small mammals with food and necessary shelter as sug-
F
M
M/F
M/F
gested by DIETERLEN (1989) and DUDU (1991).
Deomys ferrugineus
33
41
0.80
0.67
The community of small mammals in the localities of
Hybomys cf. lunaris
196
231
0.85
0.98
the RFO is much diversified : H = 2.766; 2.779; 3.018;
a
Hylomyscus aeta
50
41
1.22
-
3.054; 3.156; and D = 0.843; 0.825; 0.799; 0.818; and
Hylomyscus stella
81
122
0.66
0.47
0.838 and E = 0.800; 0.775; 0.684; 0.671; and 0.789
Lophuromys dudui
34
38
0.89
0.85
respectively for Afarama, Lenda, Epulu, Njaro, and Bap-
Lophuromys luteogaster
13
11
1.18
-
ukele. The same species are encountered in different
Malacomys longipes
46
47
0.98
1.15
localities and habitats as shown by the Shannon-Wiener
Mus minutoides
12
11
1.09
-
sectoral index (H ). In all pairs of localities, H
Oenomys hypoxanthus
5
3
1.67
-
b
b
Paraxerus boehmi
1
5
0.20
-
approaches zero, indicating very strong similarity
Praomys jacksoni
143
186
0.77
0.70
between the localities (RAMADE, 1984) (the lowest value
Praomys misonnei
38
64
0.59
-
was 0.026 for Bapukele-Epulu and the highest value
Stochomys longicaudatus
5
5
1.00
1.08
0.233 for Afarama-Epulu) (Fig. 3).
wet season, there were 28 to 81% pregnant females,
0.2500
whereas in dry season this percentage declined slightly
0.213
0.233
(23 to 55%) according to the species (Fig. 4).
0.2000
0.189
0.191
30
0.1500
0.150
0.133
rity indexH
25
a
imil 0.1000
0.116
20
S
15
0.068
0.0500
0.054
10
0.026
5
0.0000
0
BA-
BA-
BA-
AF-
EP-
AF-
EP-
AF-
LE-
AF-
Deomys
Hybomys
Hylomyscus
Hylomyscus
Lophuromys
Malacomys
Praomys
Praomys
EP
NJ
LE
LE
NJ
BA
LE
NJ
NJ
EP
ferrugineus
lunaris
aeta
stella
dudui
longipes
jacksoni
misonnei
Pairs of localities
Fig. 3. ­ Pairwise similarity between trapping localities in the
Réserve de Faune à Okapis, D.R.Congo
80
70
60
Sex ratio
50
40
The sex ratio (males/females) of 13 main species was
30
20
not far from parity (Table 2). However, in most species a
10
slight advantage in favour of males was observed. Our
0
results are similar to those obtained by DUDU (1991) and
Deomys
Hybomys
Hylomyscus Hylomyscus
Lophuromys
Malacomys
Mus
Praomys
Praomys
ferrugineus
lunaris
aeta
stella
dudui
longipes
minutoides
jacksoni
misonnei
DUDU et al. (1997) in Masako Reserve.
Fig. 4. ­ Proportion of female reproductive classes in common
species during the wet season (upper graph) and the dry season
Reproduction
(lower graph) in the Réserve de Faune à Okapis, D.R.Congo.
Reproduction occurred in both trapping periods (sea-
Black columns = juveniles, barred columns = adults, white col-
umns = pregnant.
sons) although it seems more intensive in the wet season.
All three age classes (juveniles, sub-adults and adults)
were present in the wet and dry seasons. The percentage
The presence of sexually active individuals in each
of pregnant females varied between 25 and 62%. In the
trapping session is an indication of reproductive activity

Small mammals of the Okapi Fauna Reserve
195
during the wet and the dry season. Several authors
DUBOST, G. (1968). Aperçu sur le rythme annuel de reproduction
reported that the reproduction activity of small mammals
des Muridés du Nord-Est du Gabon. Biol. Gabonica, 4 : 227-
in the tropical rain forest is continuous throughout the
239.
year, but with peaks in the wet season and throughs in dry
DUDU, A.M. (1991). Etude du peuplement d'Insectivores et de
season (RAHM, 1970; DUBOST, 1968; DIETERLEN, 1986;
Rongeurs de la Forêt ombrophile de basse altitude au Zaire
D
(Kisangani, Masako). Unpublished Ph.D. thesis, Antwerpen
UPLANTIER, 1989; DUDU, 1991). In the RFO, food
resources are always available during the course of the
University.
year (H
DUDU, A., R.VERHAGEN, H. GEVAERTS & W. VERHEYEN (1997).
ART, 1985) and breeding activity can take place in
Population structure and reproductive cycle of Praomys
all seasons, but is still linked to the seasonal rainfall dis-
jacksoni (De Winton, 1897) and first data on the reproduc-
tribution (Fig. 2). This resembles the observations of
tion of P. misonnei Van der Straeten & Dieterlen, 1987 and P.
DUBOST (1968) in Gabon, HAPPOLD (1977, 1978) in the
mutoni Van der Straeten & Dudu, 1990 (Muridae) from
Nigeria rain forest, and DUDU (1991) in Masako Reserve
Masako forest (Kisangani, Zaire). Belg. J. Zool., 127 (suppl.
(RD Congo).
1) : 67-70.
DUPLANTIER, J.M. (1989). Les rongeurs myomorphes forestiers
Litter size
du Nord-Est du Gabon : Structure du peuplement, démogra-
phie, domaines vitaux. Rev. Ecol (Terre et Vie), 44 : 329-
The litter size (Table 3) is generally small (1-4
346.
embryos) with the average that varies in the same range
EWANGO, N.E. (1994). Contribution à l'étude structurale de la
as reported by RAHM (1970) HAPPOLD (1978) DIETERLEN
fôret monodominante à Gilbertdendron dewevrei de la
(1989) and DUDU (1991) for the same species.
Réserve de Faune à Okapi (Ituri, haut-Zaïre). Mémoire de
licence inédit, 66 p. & annexes.
TABLE 3
HAPPOLD, D.C.D. (1977). A population study on small Rodents
in tropical rain forest of Nigeria. Terre et la Vie, 31 : 385­
Observed litter sizes of some common rodent species in the
458.
Réserve de Faune à Okapis, D.R. Congo. N=sample size;
HAPPOLD, D.C.D. (1978). Reproduction, growth and develop of
Max=maximum observed litter size; Avg=average observed lit-
a West African mouse, Praomys tullbergi (Thomas). Mam-
ter size; Literature=reported values for this species in literature
malia, 42 : 73-95.
HART, J. & P. CARRICK (1996). Climate of the Réserve de Faune
Species
N
Max
Avg
Literature
à Okapis : Rainfall and temperature in the Epulu sector
1986-1996. CEFRECOF Working paper n° 2.
Deomys ferrugineus
12
3
1.84
1.69
HART, J.A. & F. BENGANA (1997). Distribution and status of
Hybomys cf. lunaris
74
4
2.24
2.02-2.8
large Mammals and Human activities in the Réserve de
Hylomyscus aeta
14
4
3.14
2.9-3.4
Faune à Okaps, Démocratic Replublic of Congo. Summary
Hylomyscus stella
25
4
3.00
-
Report with Recommendations for Conservation zoning and
Lophuromys dudui
11
3
2.50
1.83-3.0
managment; 14 p + fig.
Malacomys longipes
13
4
2.30
2.5-3.19
HART, J.A. (1985). Comparative dietary ecology of a commu-
Praomys jacksoni
65
4
3.06
2.19-3.8
nity of frugivorous forest ungulates in Zaire. Ph.D. thesis,
Praomys misonnei
20
4
3.15
-
Michigan State University : 170 p.
HART, T.B. (1985). The ecology of a single-species dominant
ACKNOWLEDGEMENTS
forest and a mixed forest in Zaïre, Africa. Ph.D. thesis,
Michigan State University : 215 p.
H
We are very grateful to the CEFRECO/Epulu Directors Ter-
ATT, R.T. (1940). Lagomorpha and Rodentia other than Sciuri-
dae, Anomaluridae and Idiuridae, collected by the American
esa B. and John A. Hart for their hospitality and collaboration.
Museum Congo Expedition. Bull. Am. Nat. Hist., 76 : 457­
We appreciate the material support from Julian Kerbis of the
604.
Field Museum of Natural History, Chicago, (USA). Many
thanks to the Mbuti pygmy team conducted by "the mangasa"
HOLLISTER, N. (1916). Shrews collected by the Congo expedi-
Kenge who took us around their estate of Ituri forest.
tion of the Amecican Museum. Bull. Amer. Nat. Hist., 35 :
663-680.
KREBS, C.J. (1994). Ecology. The Experimental Analysis of Dis-
REFERENCES
tribution and abundance. 4th edition, Harper Collins College
Publishers. 801 p.
BULTOT, F. (1971). Atlas climatique du basin congolais. Publica-
KATEMBO, M. (1990). Contribution à l'étude du régime alimen-
tions de l'institut pour l'Etude Agronomique au Congo
taire du léopard Panthera pardus (Linnée, 1758), Carnivora
(I.N.E.A.C.). Hors série.
dans la forêt de l'Ituri (Zaire). Mémoire de licence, UNIKIS.
CARPANETO, G.M. & F.P. GERMI (1989). The mammals in the
MAKANA, J.R., T. HART & J. HART (1995). Forest structure and
zoological culture of the Mbuti Pygmies in North-Eastern
diversity of liana and understory treelets in monodoninant
Zaire. Hystrix (n.s.) 1 : 1-83.
and nixed forest in the Ituri Zaire. SI/MAB Symposium : 23-
CONWAY, D. (1992). A comparison of soil parameters in mono-
25.
dominant and mixed forest in Ituri forest, Zaire. University
MAKANA, J.R. (1999). Forest structure, species diversity and
of Aberdeen Tropical environment Science. Honour Project,
spatial patterns of trees in monodominant and mixed forests
Scotland.
strands in the Ituri forest, Democratic Republic of Congo.
DIETERLEN, F. (1986). Seasonal reproduction and population
MS thesis, Oregon State University.
dynamics in Rodents of an African Lowland rain forest.
MAKANA, J.R., T. HART & J. HART (1998). Forest structure and
Cimbebasia, Ser (A), 8 H (1) : 1-7.
diversity of liana and understory treelets in monodoninant
DIETERLEN, F. (1989). Rodents. In : LIETH, H. & M.J.A. WERGER
and nixed forest in the Ituri, Zaire. In : DALLMEIER F. & J.A.
(ed), Tropical Rain Forest Ecosystems. Elsevier Science :
COMISKEY (eds), Forest biodiversity research, Monitoring
383-400.
and Modeling. Conceptual Background and old World Case

196
Pionus G.B. Katuala, John A. Hart, Rainer Hutterer, Herwig Leirs and Akaibe Dudu
studies. Vol. 20, Man and Biosphere Series. The Parthenon
RAHM, U. (1970). Note sur la reproduction des Sciuridés et
Publishing Group, Pearl River, N.Y. : 429-446.
Muridés dans la forêt équatoriale du Congo. Rev. Suisse de
MUKINZI, I., N. KADANGE, A. BAPEAMONI, D. AMUNDALA, T.
Zool., 77 : 635-646.
GEMBU, K. KASWERA, M. DANADU, M. GAMBALEMOKE,
P.G.B. K
R
ATUALA, M. COLYN, R. HUTTERER & A. DUDU
AMADE, F. (1984). Élément d'Écologie. Écologie Fondamen-
(2003). Preliminary data on rodents and shrews (Mammalia)
tale. MACGRAW-HILL, Paris, 404 p.
biodiversity in the Kisangani periphery. 9th International
SIKUBWABO, K. (1987). Contribution à l'étude de l'écoéthologie
African Small Mammals Symposium, Sokoine University of
de l'Okapi (Okapia johnstoni Sclater, 1901) femelle dans la
agriculture, Morogoro, Tanzania. Abstracts : p 56.
forêt de l'Ituri : rythme d'activités journalières, fréquenta-
NDJANGO, N. (1994). Contribution à l'étude de la forêt mixte de
tion du milieu, distances parcourues et domaine vital.
la Réserve de Faune à Okapi (Ituri, Haut-Zaïre). Mémoire de
Mémoire, Fac. Sci. UNIKIS, inédit, 50 p.
licence inédit, Fac. Sci., UNIKIS, 48p.
RAHM, U. (1966). Les Mammifères e la forêt équatoriale de l'Est
TSHOMBE, R. (1994). Census report of humain populations in
du Congo. Ann. Mus. Roy. Afr. Cent., Tervuren, Série 8, 149 :
and around the Réserve de Faune à Okapis. For Wildlife
38-121.
Conservation Society.

Belg. J. Zool., 135 (supplement) : 197-216
December 2005
Cytotaxonomy of rodent species from Ethiopia, Kenya,
Tanzania and Zambia

Marco Corti1, Riccardo Castiglia1, Paolo Colangelo1, Ernesto Capanna1, Francesca Beolchini1,
Afework Bekele
2, Nicholas O. Oguge3, Rhodes H. Makundi4, Alfred M. Sichilima5, Herwig
Leirs
6, 7, Walter Verheyen6* and Ron Verhagen6
1 Dipartimento di Biologia Animale e dell'Uomo, Università di Roma "La Sapienza", Via Borelli 50, 00161 Roma, Italy
2 Biology Department, Addis Ababa University, P.O. Box 1176, Addis Ababa, Ethiopia
3 Department of Zoology, Kenyatta University, P.O. Box 43844, Nairobi, Kenya
4 Pest Management Centre, Sokoine University of Agriculture, P.O. Box 3110, Morogoro, Tanzania
5 Soil and Crops Research Branch, Mutanda Research Station, P.O. Box 110312, Solwezi, Zambia
6 Evolutionary Biology Group, University of Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium
7 Danish Pest Infestation Laboratory, Danish Institute of Agricultural Sciences, Department of Integrated Pest Management,
Skovbrynet 14, DK-2800 Kongens Lyngby, Denmark
Corresponding author : Marco Corti, e-mail : marco.corti@uniroma1.it, phone: 0039-0649918092, fax: 0039-064457516
ABSTRACT. An extended survey of taxa belonging to two genera of Cricetomyinae (Cricetomys and Saccos-
tomus), one Gerbillinae (Gerbilliscus), eight Murinae (Acomys, Aethomys, Arvicanthis, Lemniscomys, Mus (Nanno-
mys
), Mastomys, Grammomys, Stenocephalemys) and one Myoxidae (Graphiurus) was carried out as part of the EU
programme "Staplerat" involving Ethiopia, Kenya, Tanzania and Zambia. Here we report the diploid and autosomal
fundamental numbers of these rodent taxa. Seventeen of them were unknown, for four we report chromosomal var-
iants and for another 16 new localities where they occur. We discuss their specific status taking into consideration
our results together with data from literature and highlight the problems in taxonomy and systematics that are yet to
be solved, due do their extended range and the occurrence of species complexes. We highlight cases for which there
should be a re-evaluation of specific names that were not included in the last rodent checklist.
KEY WORDS : Cytogenetics, Rodents, taxonomy, East Africa.
INTRODUCTION
The second part of the twentieth century saw the devel-
opment of new approaches, such as cytogenetics, genet-
ics, molecular genetics and morphometrics which first
Current knowledge of African rodent taxonomy has
separately and later combined into a multidisciplinary
been largely influenced by the history of colonization and
approach have been employed to study African rodent
by several expeditions organised independently by major
taxonomy, and often in collaboration with African Institu-
museum Institutions carried out during the early part of
tions. This approach was based on the biological species
the twentieth century. As a result, Natural History muse-
concept derived from the Modern Synthesis
ums in Europe and in the United States posses the largest
(DOBZHANSKY, 1937). However, most of this work was
collections of African rodents, which include most spe-
carried out on a local scale which has limited its full
cies types so constituting a unique reference for correct
application. Much of the information on African Rodents
taxonomic identification and classification purposes. It is
is scattered over wide areas, and taxonomic revision for
worthy of note that, although essential from a scientific
many of the genera is far from compete. This is also a
point of view, this resulted in a proliferation of names, as
consequence of the occurrence of sibling and cryptic spe-
most descriptions were carried out with little or limited
cies. Therefore, there is a need for continuing collabora-
comparison to the large series, which is the only way to
tion in order to allow investigations to cover larger areas.
arrive at the correct identification of variation (e.g., the
The Staplerat project1, founded by the European Union
long list of taxa reported by ALLEN in 1939).
and involving Ethiopia, Kenya, Tanzania and Zambia as
* Walter Verheyen deceased late 2005 after acceptance of this paper. He was a great example for all other authors of this paper and for African rodent
taxonomy in general.
1 STAPLERAT, Protecting staple crops in eastern Africa: integrated approaches for ecologically based field rodent pest management. An international
collaborative project, supported by the European Union's 5th Framework RTD Program.

198
Corti, Castiglia, Colangelo, Capanna, Beolchini, Bekele, Oguge, Makundi, Sichilima, Leirs, Verheyen and Verhagen
African partners, constituted an excellent opportunity to
dividing the Kenyan and Tanzanian savannas, up to the
investigate rodent taxonomy on a larger scale. The study
deciduous bushland and thicket which characterizes the
concerned an area which extended from the Ethiopian
bottom of the Rift which penetrates and bisects the Ethio-
plateau south to northern Zambia across the eastern side
pian plateaux.
of the Great Rift Valley of Kenya and Tanzania. The pro-
gramme involved the identification of integrated
During the three-year project there were several collec-
approaches for rodent pest control in agricultural areas.
tions, both in fields and surrounding areas. Samples of
This naturally implied a need for a correct taxonomy of
each taxon found at each locality were studied using
these captured species.
multidisciplinary approaches, including cytogenetics,
molecular genetics and morphometrics. Here we present
The areas studied formed two main biotic zones typical
the cytotaxonomic results, providing the diploid (2n) and
of the eastern part of Africa, i.e. the Somali-Maasai and
the Zambezian (M
fundamental numbers (NFa) only. A description of each
ENAUT, 1983; WHITE, 1986). These
karyotype by means of differential staining will appear
areas represent independent cradles of speciation and
elsewhere for each genus as a separate paper.
evolution for different rodent faunas, from the Late
Miocene to the present day (DENYS, 1999). Our study
Karyotype descriptions constitute the primary tool for
reports the genera and species which, in addition to those
rodent species identification, as it has been established and
in the dry and wet Miombo woodland savannas, provide
generally accepted that the reason behind the high diversity
evidence of this recent history.
shown by this mammalian order is related to its high rate of
The limit between the Somali-Maasai and the Zambez-
chromosomal mutation (CORTI, 2002; KING, 1993). We
ian biomes occurs in south western Tanzania and north-
present here the karyotypes for two genera of Cricetomyi-
west Zambia (Fig. 1), and represents a crucial area where
nae (Cricetomys and Saccostomus), one Gerbillinae (Ger-
independently evolved rodent faunas converge. Further-
billiscus), eight Murinae [Acomys, Aethomys, Arvicanthis,
more, our sampling localities extend to the northwest of
Lemniscomys, Mus (Nannomys), Mastomys, Grammomys,
the Somali-Maasai, across the Pare and Uzambaras range
Stenocephalemys] and one Myoxidae (Graphiurus).
Fig. 1. ­ Map of Ethiopia, Kenya, Tanzania and Zambia (as indicated in the inset) with
the collection localities (see Table 1 for the locality names and description). Elevation
contour intervals are also shown. The dotted line represents the border between the
Somali-Maasai (North) and the Zambezian (South) biotic zones.

Cytotaxonomy of East Africa Rodents
199
MATERIAL AND METHODS
Giemsa standard method (pH7). Pictures of metaphases
were collected using the digital camera Photometrics Sen-
The following eight localities were sampled (STAPLERAT,
sys 1600 and the Iplab software (Scanalytics, Inc, version
2003): Mugo and Zeway (Ethiopia); Rongai, Kitale and
2.420).
Nairobi (Kenya); Chunya (Tanzania); Meheba and
Taxonomic definitions for Muroidea followed the most
Mutoma (Zambia) (Fig. 1; Table 1). These are prevalently
updated check-list by MUSSER & CARLETON (2005) and,
arable fields (for the original habitat description, see Table
for Myoxidae, HOLDEN (1993). Specimen identification
1). To achieve a more realistic representation of species
was based on comparisons with type material and rele-
range outside of the project areas, captures in surrounding
vant series. We discuss the taxonomic problems encoun-
localities were included, particularly in the geographic gaps
tered in species identification and description. For species
between the field study areas (Fig. 1; Table 1).
whose taxon is in doubt, chromosomal comparisons and
Specimens were live-trapped using Sherman Folder
analysis of the sequences of mitochondrial genes were
traps and then transported alive to the following laborato-
carried out in an integrated approach in order to make
ries: Biology Department, Addis Ababa University (Ethi-
identification possible (such results will appear else-
opia); Department of Zoology, Kenyatta University
where). However, it was not possible to reach a definitive
(Kenya); Pest Management Centre, Sokoine University of
taxonomic definition for some. These are provisionally
Agriculture, Morogoro (Tanzania); Mutanda Agricultural
indicated in the following sections as "cf.", "cfr.", "sp.",
Research Station, Solwezi (Zambia).
or with an acronym.
Chromosome preparations were obtained from the
A total number of 187 specimens representing 37 puta-
bone marrow following the air-drying method of HSU &
tive species were analysed. Specimens are preserved at
PATTON (1969). Cell suspensions in fixative were then
the permanent collection of Musée Royal de l'Afrique
transported to the Dipartimento di Biologia Animale e
Centrale, Tervuren (codes starting with a "T") and of the
dell'Uomo, Università di Roma 'La Sapienza', where
Museo di Anatomia Comparata dell'Università di Roma
slides were prepared. Metaphases were stained by the
"La Sapienza" (codes with "ET", "ZM", "KE", "TZ").
.
TABLE 1
Collection localities, with the locality code for Fig. 1, latitude and longitude, and the current and original habitat (see, for habitat refer-
ence, White, 1983)
Locality
Latitude and
Country
Locality
Habitat
Original habitat
code
Longitude
Ethiopia
Mugo
1
07°50'N - 37°59'E
Enset fields
Highland with some alpine vegetation
and moorland
Zeway
2
07°55'N - 38°43'E
Maize fields
Savannah woodland with acacia trees
Kenya
Kitale
3
01°01'N - 35°00'E
Maize fields
Mosaic of lowland rainforest and sec-
ondary grassland
Rongai
4
00°10'S - 35°51'E
Maize fields
Mosaic of East African evergreen
bushland and secondary Acacia
wooded grassland
Nairobi
5
01°16'S - 36°49'E
Grassland around buildings
" "
Tanzania
Kitundu
6
01°53'S - 31°39'E
Rain forest
Rain forest
Forest
Jipe
7
03°41'S - 37°42'E
Savannah bushes with scattered trees
Savannah bushes with scattered trees
Lwami
8
03°41'S - 37°32'E
Bushland with scattered trees
Bushland with scattered trees
Kisiwani
9
04°07'S - 37°57'E
Scattered bushes and grassland
Scattered bushes and grassland
Ngasumet
10
04°29'S - 37°10'E
Grassland with scattered bushes
Grassland with scattered bushes
Matongolo
11
04°31'S - 36°28'E
Bushland
Bushland
Mkomazi
12
04°39'S - 38°05'E
Wooded grassland
Wooded grassland
Singida
13
04°49'S - 34°44'E
Wooded grassland
Wooded grassland
Mombo
14
04°54'S - 38°13'E
Grassland with scattered bushes
Grassland with scattered bushes
Ndaleta
15
05°12'S - 36°30'E
Grassland
Grassland
Zoissa
16
05°40'S - 36°25'E
Bushland
Bushland
Mvumi
17
06°20'S - 35°50'E
Wooded grassland
Wooded grassland
Mission
Inala
18
05°25'S - 32°49'E
Grassland with scattered bushes
Grassland with scattered bushes
Morogoro
19
06°49'S - 37°40'E
Cultivated areas, with fallow, scattered Wooded grassland
trees
Itigi
20
05°41'S - 34°28'E
Bushland
Bushland
Dakawa
21
06°26'S - 37°34'E
Miombo woodland
Miombo woodland
Chunya B
22
07°58'S - 33°18'E
Cultivated fields, grassland with scat-
Wetter Zambezian Miombo woodland
tered trees
Chunya A
23
08°31'S - 33°24'E
Wetter Zambezian Miombo woodland
" "
Zambia
Mutanda
24
12°22'S - 26°16'E
Maize fields
" "
Res. Sta-
tion
Meheba
25
12°33'S - 25°41'E
Maize fields
" "
Mutoma
26
13°45'S - 24°57'E
Maize fields
" "

200
Corti, Castiglia, Colangelo, Capanna, Beolchini, Bekele, Oguge, Makundi, Sichilima, Leirs, Verheyen and Verhagen
Karyotype descriptions
rence of small arms, which do not occur in the other kary-
otypes. This karyotype is described here for the first time.
CRITETOMYNAE (Roberts, 1951).
­ Saccostomus (Peters, 1846).
The subfamily Cricetomyinae comprises three genera,
Beamys (Thomas, 1909) [with two species B. hindei
Saccostomus is common and widespread in savannahs,
(Thomas, 1909) and B. major (Dollman, 1914)], Criceto-
scrubby areas, and cultivated fields from South Ethiopia
mys (Waterhouse, 1840) (with four species), and Saccos-
and Somalia through East Africa down to the Cape Prov-
tomus (Peters, 1846), [with the two species S. campestris
ince. Despite this commonness, taxonomy has long been
(Peters, 1846) and S. mearnsi (Heller, 1910)].
a source of debate (DELANY, 1975; ELLERMAN et al., 1953;
MISONNE, 1974). In fact, the most recent rodent checklist
­ Cricetomys (Waterhouse, 1840).
(MUSSER & CARLETON, 2005) includes two species only,
i.e. S. campestris (Peters, 1846) and S. mearnsi (Heller,
There are currently four recognised species of giant
1910). However, a wide karyotypic polymorphism within
pouched rats (GENEST-VILLARD, 1967): Cricetomys
the genus has been described, from the extreme south of
ansorgei (Thomas, 1904), C. emini (Wroughton, 1910),
the range (GORDON, 1986) to the north (CAPANNA et al.,
C. gambianus (Waterhouse, 1840), and Cricetomys
1985; CORTI et al., 2004) through the central part of
kivuensis (Lönnberg, 1917). Previously it was suggested
Africa (FADDA et al., 2001; HUBERT, 1978). Recently, a
that there existed six (ALLEN, 1939) or one (ELLERMAN et
combined approach using both cytochrome b sequences
al., 1953) species, while GENEST-VILLARD (1967)
groups and cytogenetics (CORTI et al., 2004) established
described predominantly savannah-dwelling (C. gambi-
that mearnsi and campestris represent two distinct natural
anus) and lowland forest (C. emini) species. There are
groups and species complexes and was able to suggest a
limited data regarding possible variation, so taxonomy
partial taxonomic resolution of the genus.
must be considered as provisional for East Africa. Chro-
mosomal descriptions are available for West African
­ Saccostomus cf. elegans (Thomas, 1897).
specimens only. MATTHEY (1954) described a karyotype
Three different karyotypes have been found. Two of
for C. gambianus (unknown origin) with 2n=78. In Sen-
them occur in two nearby localities (approximately 15 km
egal, GRANJON et al. (1992) found a karyotype with 2n=80
apart) in south Tanzania, named Chunya A and B; a third
and NFa=82, and in Benin, CODJA et al. (1994) described
is typical of northwest Zambia (Fig. 1, Tab. 1).
a karyotype with 2n=82 and NFa=88 for C. gambianus,
and a karyotype of 2n=80 and NFa=88 for C. emini.
Tanzania, Chunya A (GT50664). The diploid number is
2n=42 and the NFa is 46. The X chromosome is a
medium-size submetacentric. The karyotype includes two
pairs of large submetacentrics and two pairs of very small
metacentrics, the remaining chromosomes being acrocen-
trics decreasing in size (see CORTI et al., 2004, for a
Giemsa stained karyotype).
Tanzania, Chunya B 2 (GTZ522, GTZ524, FTZ502,
FTZ519, FTZ520). The diploid number is 2n=44 and
NFa=46. The karyotype consists of three pairs of biarmed
autosomes, one of which is a large subtelocentric and two
are small metacentrics; the remaining autosomes are acro-
centrics decreasing in size. The X and Y chromosomes
are a large metacentric and a medium size submetacen-
tric, respectively (see CORTI et al., 2004, for a Giemsa
Fig. 2. ­ The karyotype of Cricetomys cf. gambianus, 2n=80
stained karyotype).
and NFa=84, X Y.
Zambia Saccostomus sp., Mutanda Research Station
(FZM3, FZM9, FZM15, FZM18). The diploid number is
2n=44 and the NFa is 48. The autosomal complement is
­ Cricetomys cf. gambianus (Thomas, 1904).
identical to the one described for Chunya B, except for a
One male from Morogoro (TZ), no voucher specimen.
pair of small metacentric chromosomes (No 20; in CORTI
No morphological and DNA comparisons are possible for
et al. 2004) that in Chunya B are acrocentrics. Further-
the moment. Therefore, we refer to this cytotype as "cf.
more, the X chromosome is a large submetacentric and Y
gambianus". The diploid number is 2n=80 and the
is a medium size metacentric.
NFa=84. The autosomes are composed of seventy-four
On the basis of the entire sequence of the cytochrome b
acrocentrics decreasing in size and by four medium and
mitochondrial gene, CORTI et al. (2004) have shown that
small metacentrics (Fig. 2). The X chromosome is one of
these three karyotypic forms belong to the campestris
the largest chromosomes of the whole karyotype and it is
species group and that they all form a monophyletic clade
biarmed. The Y chromosome is a medium-size submeta-
typical of the Zambezian domain, the northern limits of
centric. This karyotype differs from those in Senegal
which occur in southern Tanzania. There is a low propor-
(GRANJON et al., 1992) and Benin (CODJA et al., 1994):
tion of nucleotide substitutions between them, suggesting
not only are the diploid and NFa different, but also sev-
that the ongoing chromosomal differentiation has not led
eral pairs of acrocentrics are characterized by the occur-
to full speciation. The southern Tanzanian specimens

Cytotaxonomy of East Africa Rodents
201
(Chunya A and B) occur north (230 km) of Karonga, the
attributed to G. nigricaudus. However, there is evidence
locality on the northern shores of Lake Nyasa from which
that this karyotype would be better attributed to another
S. elegans (Thomas, 1897) was described. However, a
species (COLANGELO et al., 2005). G. robustus (2n=36,
craniological comparison of our specimens with the type
NFa=64; QUMSIYEH et al., 1987; FADDA et al., 2001), but
of S. elegans (BMNH 97.18.1.207) and of two "co-types"
this should be attributed to G. vicinus (DOBIGNY et al.,
of S. campestris (BMNH 7.1.1.181 and 58.6.18.19)
2002; GRANJON & DOBIGNY, 2003; see later), occurring
revealed that the Chunya specimens could also be allo-
from Chad to the Horn and East Africa; two specimens
cated to typical campestris, whereas the northern Zam-
from Central African Republic with 2n=46 and NFa=64
bian specimens closely resemble the elegans type skull.
were identified by MATTHEY & PETTER (1970) as G.
However, the skull differences between the two groups
robustus, but this was probably incorrect, as later they
could also be due to age differences (indeed, the elegans
were ascribed by QUMSIYEH et al. (1987) to G. phillipsi
type skull is of an old individual, and the "co-types" of
(Somalia, Kenya, and Ethiopia); MATTHEY & PETTER
campestris are young animals). It is obvious that this tax-
(1970) described a karyotype with 2n=36 and NFa=62
onomic problem will only be solved by undertaking an
from a Central African Republic specimen and attributed
adequate study of 1) the sexual dimorphism and skull
it to G. kempi, but this would be better referred to as G.
growth in a statistically relevant population of Saccos-
robustus. G. kempi (2n=48, NFa=62-64; CODJA et al.,
tomus from the region concerned and of 2) the karyology
1994; COLANGELO et al., 2001); GAUTUN et al. (1986)
and genetics of Saccostomus individuals from the Tette
referred to a specimen from Guinea with 2n=46 as G.
region in Mozambique (topotypical for S. campestris).
kempi. Samples with 2n=48-50 and NFa=52-66, attrib-
One must remember that a new undescribed species
uted to G. nigrita (now included in G. kempi), were
occurs in Tanzania north of the border with the Zambez-
reported from Chad, Zaire, Zambia and Angola. MAT-
ian domain in the Maasai Steppe (FADDA et al., 2001). It
THEY & PETTER (1970) attributed specimens from Burkina
has been shown (C
Faso and Ivory Coast to G. hopkinsoni (2n=48, NFa=62-
ORTI et al., 2004) that this new taxon
belongs to the Saccostomus cf. mearnsi complex.
64), now synonymous of G. kempi. G. gambianus (synon-
ymous of G. kempi) (2n=52, HUBERT et al., 1973); MAT-
GERBILLINAE (Gray, 1825).
THEY (1969) described in a specimen from Senegal 2n=52
and NFa=64 and this was ascribed to G. validus, but prob-
­ Gerbilliscus (Thomas, 1897).
ably it should be considered G. gambianus (MATTHEY &
PETTER, 1970). DOBIGNY et al. (2002) reported the same
Gerbilliscus has been proposed recently as a distinct
karyotype from Nigeria. G. guineae (2n=50, NFa=64;
genus from Tatera (PAVLINOV, 1999; CHEVRET &
MATTHEY & PETTER, 1970; BENAZZOU et al., 1984; GAU-
DOBIGNY, 2005). It includes exclusively all the African
TUN et al., 1985). More recently, BULATOVA et al. (2002)
species of the former Tatera (Lataste, 1882), an Asian
reported a karyotype from Ethiopia with 2n=52 and
monospecific with only T. indica (MUSSER & CARLETON,
NFa=62 that was attributed to G. validus.
2005). According to this checklist, the genus now
Two different karyotypes were found in the present
includes 10 species. However, recent molecular data
study. One shows 2n=36 and NFa=68, which is shared by
(COLANGELO et al., 2005; COLANGELO et al., submitted)
three different species distinguishable on the basis of
suggest the occurrence of cryptic species which cannot be
skull morphology and molecular analyses (COLANGELO et
recognized through cytogenetics and/or morphometrics
al., 2005), i.e. G. nigricaudus, G. robustus and G. vicinus
(i.e. G. vicinus). Moreover, there is molecular (CHEVRET
(for a discussion of their systematic status see COLANGELO
& DOBIGNY, 2005; COLANGELO et al., sub.) and cytoge-
et al., 2005). The second karyotype shows 2n=40 and
netic evidence indicating a close relationship between
NFa=66 and characterizes G. leucogaster.
Gerbilliscus and Gerbillurus. An extensive systematic
revision would probably indicate Gerbillurus as synony-
­ Gerbilliscus leucogaster (Peters, 1852).
mous of Gerbilliscus (CHEVRET & DOBIGNY, 2005; COL-
ANGELO et al., sub.).
Dakawa (FT50545, FT50546); Tanzania. The diploid
number is 2n=40 and NFa=66. The karyotype is com-
There have been a high number of karyotypic studies
posed of fourteen pairs of biarmed chromosomes decreas-
performed on the genus, but they all originate from scat-
ing in size and five pairs of medium size acrocentrics. The
tered areas, and no serious attempt has been made to sum-
X chromosome is a large metacentric, and the Y is a
marize and include them in a general framework. Data
medium size submetacentric (see COLANGELO et al.,
available for the African species are as follows: G. afra
2005). Phylogenetic analyses based on the cytochrome b
(2n=44, NFa=66; MATTHEY, 1954; QUMSIYEH, 1986), a
and 16S mitochondrial genes showed a marked diver-
South African endemic. G. brantsii (2n=44, NFa=66;
gence between these specimens and those attributed to the
MATTHEY, 1954; QUMSIYEH, 1986), ranging from South
robustus species group (i.e. G. robustus, G. nigricaudus
Africa to Zambia. G. leucogaster (2n=40, NFa=66; GOR-
and G. vicinus (COLANGELO et al., 2005).
DON & RAUTENBACH, 1980; QUMSIYEH, 1986), ranging
from South Africa to Southwest Tanzania; MATTHEY
­ Gerbilliscus robustus (Cretzscmar, 1826).
(1954) identified a specimen from the Central African
Republic as G. schinzi but this species is now considered
Zeway (FET107, FET119, FET127); Ethiopia. The
synonymous of G. leucogaster. G. nigricaudus (2n=36,
diploid number is 2n=36 and NFa=68. All autosomes are
NFa=68, COLANGELO et al., 2005), occurring in Kenya
biarmed decreasing in size. The X chromosome is a large
and Tanzania. QUMSIYEH et al. (1987) reported a karyo-
metacentric, and the Y chromosome is a small acrocentric
type with 2n=40 from Kenya (one specimen) which was
(see COLANGELO et al., 2005). This same chromosomal

202
Corti, Castiglia, Colangelo, Capanna, Beolchini, Bekele, Oguge, Makundi, Sichilima, Leirs, Verheyen and Verhagen
formula has already been reported from several other
was in striking contrast with the XX/XY typical of A. spi-
localities from west Africa along the arid sub-Saharan
nosissimus and the other species of the genus. Further-
belt to East Africa down to Tanzania across the arid
more, MATTHEY (1965a, b, and unpublished) also showed
savannahs (QUMSIYEH et al., 1987; FADDA et al., 2001),
an inter- and intra-individual variability in chromosome
suggesting that this karyotype is highly represented in this
constitution in A. selousi. Later, DIPPENAAR & RAUTEN-
biome.
BACH (1986) found the same 2n=60 and NFa=68, but with
a submetacentric X in Transvaal (South Africa). All
­ Gerbilliscus nigricaudus (Peters, 1878).
together, these data do not support the idea of maintaining
the two taxa in synonymy, and they should be definitively
Mkomazi (GT50216, GT50217), Lwami T50230, Jipe
considered as separate valid species.
(GT50453, GT50475, GT50476, T50456F); Tanzania. The
diploid number is 2n=36 and NFa=68. All autosomes are
The analysis of a partial fragment (715 bp) of the gene
biarmed decreasing in size. The X chromosome is a large
for cytochrome b in the Tanzanian samples, for which we
metacentric, and the Y chromosome is a small acrocentric
report here the karyotypes (unpublished data), confirmed
(see COLANGELO et al., 2005). We found this species in a
that A. spinosissimus and A. cf. selousi, although mono-
restricted area across the border of Tanzania and Kenya.
phyletic, constitute two well separated species. This
MATTHEY (1969) and QUMSIYEH et al. (1987) reported for
genetic and chromosomal distinction is also reported for
G. nigricaudus a diploid number of 40 with NFa respec-
the Tanzanian locality of Berega by BAROME et al. (2001).
tively 66 and 68, but probably these specimens are refera-
For these reasons, we decided to provisionally maintain
ble to G. leucogaster.
the name A. cf. selousi for these specimens until the
"selousi" ­ like group has received adequate investiga-
tion. The karyotypes known for other species are as fol-
­ Gerbilliscus vicinus (Peters, 1878).
lows: A. cahirinus, 2n=36 and NF=68 (VOLOBOUEV et al.,
Itigi (
1996); A. dimidiatus (cf. A. airensis, Agades, Niger,
GT50339), Matongolo (FT50062), Ndaleta
(
2n=42, NFa=66), 2n=38 and FN=70 (V
FT50144, GT50153, GT50158), Inala (FT50579), Ngas-
OLOBOUEV et al.,
umet (
1991); A. ignitus, 2n=50 and NF=66-68 (M
FT50190), Mombo (GT50214, GT50215, GT50226,
ATTHEY,
1956); A. russatus, 2n=66 and NF=66 (W
FT50227), Jipe (GT50473); Tanzania. Nairobi (GKE135);
AHRMAN &
Kenya. The karyotype appears to be the same as G. robus-
ZAHAVI, 1953).
tus from Ethiopia, but these samples show a marked
Following the 1999 expedition in the Maasai Steppe,
genetic divergence from the Ethiopian specimens sug-
FADDA et al. (2001) reported three different karyotypes
gesting that G. vicinus should be considered a separate
for this part of Tanzania, corresponding to A. spinosis-
species (COLANGELO et al., 2005). BATES (1988) reported
simus (2n=60, NFa=70), A. wilsoni (2n=62, NFa=76) and
a significant geographical variation in skull and body size
A. ignitus (2n=36, NFa=68). Here we add further infor-
in G. robustus, suggesting the possible occurrence of a
mation from a higher number of localities and for a higher
distinct race in central Tanzania. He identified G. swayth-
number of specimens for these species.
lingi (Kershaw, 1921), type locality Morogoro, as the hol-
otype for this race. However, an extensive morphometric
and molecular genetic comparison would probably
­ Acomys spinosissimus (Peters, 1852).
ascribe this race to G. vicinus.
Chunya (GT50676, GT50673), Tabora-Inala (GT50676),
MURINAE (Illiger, 1815).
Zoissa (GT50119, GT50088, FT50087, FT50202), Maton-
golo (GT50003); Tanzania. The species occurs in a typical
­ Acomys I. (Geoffroy, 1838).
Zambezian domain (Southern Tanzania to Southern
Transvaal). However, it extends north to the Maasai
These spiny mice are widespread throughout all of
Steppe, which is a typical Somali-Maasai arid domain.
Africa, the near and Middle East, and some Mediterra-
The diploid number in the localities investigated is 2n=60
nean islands. The rodent checklist by MUSSER & CARLE-
and the NFa=70. The karyotype is represented by 5 pairs
TON (2005) lists 19 species, but as for most of the other
of metacentrics and submetacentrics, 22 pairs of acrocen-
African rodent genera, their taxonomy and systematics
trics decreasing in size, and one pair of small metacen-
are yet to be established. Cytogenetics, morphology and,
trics. The X chromosome is a large acrocentric and the Y-
recently, molecular genetics (BAROME et al., 1998; 2001)
chromosome is a small subtelocentric. This karyotype is
have been widely used to shed some light on the taxon-
the same as the one described by FADDA et al. (2001) in
omy and systematics of the genus.
the Maasai Steppe and by DIPPENAAR & RAUTENBACH
(1986) in South Africa.
Since the first descriptions of the karyotype of Acomys
by MATTHEY (1956; 1963; 1965a, b; 1968), further data
concerning chromosomal variation in the genus have
­ Acomys cf. selousi (Roberts, 1951).
been added by several authors. Originally there was con-
siderable interest in the karyotype of A. selousi, long con-
Dakawa (GTZ521); Tanzania. The diploid number is
sidered synonymous of A. spinosissimus (see MUSSER &
2n=59, NFa=68. The karyotype is constituted by five
CARLETON, 1993). The peculiar chromosomal sex deter-
pairs of medium size biarmed chromosomes, 24 pairs of
mination found in the former, with a single and excep-
acrocentrics decreasing in size (Fig. 3). The single X
tionally large X chromosome in both males and females,
chromosome is very large.

Cytotaxonomy of East Africa Rodents
203
loid numbers (2n=44-50). Recent molecular analyses
based on cytochrome b sequences (DUCROZ et al., 2001;
RUSSO et al., 2001; CASTIGLIA et al., 2003a) confirmed the
high genetic difference between the two subgenera and
provided evidence for the paraphyly of the genus in a
wider phylogenetic context involving several other Afri-
can species of Murinae and Otomynae (DUCROZ et al.,
2001; CASTIGLIA et al., 2003a). These data suggest the
need for a definitive splitting of the genus.
There is also evidence indicating the occurrence of
cryptic species. For example, it has been recognised that
A. chrysophilus, widespread from Kenya to South Africa,
must be separated into two species, corresponding to dif-
ferent cytotypes previously identified (GORDON & RAU-
Fig. 3. ­ The karyotype of Acomys cf. selousi, 2n=59 and
TENBACH, 1980; GORDON & WATSON, 1986; VISSER &
NFa=68. Note the single X chromosome of very large size.
ROBINSON, 1986): the true A. chrysophilus (2n=50) and A.
ineptus (2n=44). These two species occur sympatrically
­ Acomys wilsoni (Thomas, 1892).
and differ in gross sperm and bacular morphology (VISSER
& ROBINSON, 1986; BREED et al., 1988) as well as in their
Jipe (GT50467, FT50463, GT50466, FT50439,
quantitative cranial traits (CHIMIMBA, 1998; CHIMIMBA et
GT50440, GT50447, GT50446, GT50438, FT50437),
al., 1999).
Ngasumet (FT50246, FT50247, GT50247); Tanzania. The
range of the species includes Sudan, Ethiopia, Somalia,
­ Aethomys kaiseri (Noack, 1887).
Kenya and Tanzania, but its limits are unknown. The first
description of the karyotype of the species was first
Meheba, Solwezi (FZM13, FZM32); Zambia. The dip-
described by FADDA et al. (2001). The diploid number is
loid number is 2n=50 and NFa=60. The karyotype is
2n=62 and NFa=76. The autosomal set is composed of
composed of forty-five pairs of acrocentric chromosomes
eight pairs of meta- and submetacentric chromosomes,
decreasing in size and five pairs of small meta/submeta-
three of which are the largest of the set and five of which
centrics. The sex chromosomes are very large and
range from medium to small, and of 22 pairs of acrocen-
biarmed with the Y being slightly larger that the X chro-
trics decreasing in size. The X chromosome is acrocentric
mosome. The length of each sex chromosome is approxi-
(the largest amongst the acrocentrics) and the Y chromo-
mately double the length of autosome pair 1. The karyo-
some is a small acrocentric.
type is reported in CASTIGLIA et al. (2003a).
­ Acomys ignitus (Dollman, 1910).
­ Arvicanthis (Lesson, 1842).
Lwami (FT50232, FT50527, GT50529, FT50528,
Arvicanthis probably represents one of the African
FT50517, FT50499, FT50505), Ngasumet (GT50185);
rodent genera that has received most attention over the
Tanzania. The range includes Tanzania and Kenya, but its
last decade. As a result, the number of species recognised
limits are unknown. Samples were collected at the
increased to seven after the latest rodent checklist
extreme North of the Maasai Steppe, where a corridor
(MUSSER & CARLETON, 2006). Previously, CORBET &
with Southern Kenya is present through the Uzambaras
HILL (1991) and MUSSER & CARLETON (1993) recognized
and the Pare mountain range. The first description of the
five species only (although with some disagreement
karyotype was given by FADDA et al. (2001) and shows 2n
regarding their systematics and taxonomy). The numer-
= 36 and NFa = 68. The karyotype resembles A. cahirinus
ous studies performed on a chromosomal (BASKEVICH &
and, in fact, the species was included by ELLERMAN
LAVRENCHENKO, 2000; CAPANNA & CIVITELLI, 1988;
(1941) in the Cahirinus group.
CAPANNA et al., 1996; CASTIGLIA et al., 2003b, 2006; CIV-
­ Aethomys (Thomas, 1915).
ITELLI et al., 1995; FADDA et al., 2001; VOLOBOUEV et al.,
1987; 1988; 2002a; GRANJON et al., 1992), allozymic
The genus has traditionally been divided into two sub-
(CAPULA et al., 1997), mtDNA sequencing (DUCROZ et al,
genera (DAVIS, 1975): Micaelamys, including A.
1997; CORTI et al., submitted), and morphometric (AFE-
namaquensis and A. granti, and Aethomys, for which nine
WORK BEKELE et al., 1993; CORTI & FADDA, 1996; FADDA
species are currently recognised (MUSSER & CARLETON,
& CORTI, 1998; 2001) basis have shown that the genus is
2005; VISSER & ROBINSON, 1986). Following the original
represented by a higher number of species.
hypothesis by DAVIS (1975), the two subgenera can be
Two major clades occur within Arvicanthis, roughly an
distinguished on the basis of tail length and colour of ven-
eastern and a western one, although some of the western
tral parts.
species extend into the East and vice versa (DUCROZ et al.,
Chromosomal studies (MATTHEY, 1954; VISSER & ROB-
1997; CORTI et al., submitted). This result is based on
INSON, 1986; BAKER et al., 1988) have shown that the
molecular studies (and congruent with karyotypic analy-
karyotypes of the "Micaelamys" A. namaquensis (2n=24)
ses). Taxonomic confusion in the past, part of which
(from Southeast Zambia to South Africa) and A. granti
remains unsolved, is probably due to the high level of
(2n=32) (Cape Province) are divergent from the other
convergence shown in morphology, which obscures spe-
species studied, which are all characterized by higher dip-
cies differences (FADDA & CORTI, 2001).

204
Corti, Castiglia, Colangelo, Capanna, Beolchini, Bekele, Oguge, Makundi, Sichilima, Leirs, Verheyen and Verhagen
The karyotypes of the taxa recognized so far are as fol-
better addressed through a cytogenetic and molecular
lows: A. cf. somalicus, 2n=62, NFa=62-63 (Ethiopia;
approach. Moreover, also the karyotype and the DNA
BASKEVICH & LAVRENCHENKO, 2000). A. abyssinicus,
analyses on A. niloticus were from samples of breeding
2n=62, NFa=62 (Ethiopia; CORTI et al., 1996). A. blicki,
colonies outside the topotypical area. A. dembeensis has
2n=48, NFa =64 (Ethiopia; CORTI et al., 1995; 1996). A.
been described as endemic to Ethiopia, and is considered
neumanni, 2n=54, NFa=62 (Tanzania; FADDA et al., 2001;
a relatively `lowland' taxon, occurring from sea level to
CASTIGLIA et al., 2003b). A. nairobae, 2n=62 NFa=78
2000 mt a.s.l. (YALDEN et al., 1976). However, there are
(Tanzania; FADDA et al., 2001; CASTIGLIA et al., 2003b).
problems regarding the taxonomic status of the samples
A. ansorgei, 2n=62, NFa=74/76 (Senegal, Mali, Burkina-
Faso; previously named ANI-3; V
so far analysed in different studies. The first one concerns
OLOBOUEV et al.,
2002a). A. rufinus (Benin; previously named ANI-4; C
the type specimen. To date, the most complete study was
IV-
ITELLI et al., 1995; VOLOBOUEV et al., 2002a). "A. niloti-
carried out through multivariate morphometrics (AFE-
cus" complex, 2n=62, NFa=62/64, described from Egypt,
WORK BEKELE et al., 1993) on eight localities (four of
Sudan, Ethiopia, northern Senegal and northern Burkina
them from the bottom of the Rift along the Ethiopian
Faso, southern Mauritania, Mali, Niger, Chad; VOLO-
lakes), which confirmed a clear morphological distinction
BOUEV et al., 1988; 2002a; PHILIPPI, 1994; DUCROZ et al.,
from the other major taxon occurring in the country, i.e.
1997; CIVITELLI et al., 1995).
A. abyssinicus. One of the populations examined was
from Lake Tana, on the western side of the Rift Valley, i.e.
There are further karyotypes described for which there
the topotypical area from which Rüppel described the
is still no species assignment or evidence supporting the
type specimen in 1842 (Deraske, 12° 25' N - 37° 20' E).
original allocation made by authors. This is the case for
This population clustered with the others that occurred on
the karyotype with 2n=44 and FN=72 described in Soma-
the eastern side of the Rift at altitudes below 2000 mt
lia by CAPANNA & CIVITELLI (1988), which was attributed
to A. niloticus, which evidently characterizes a different
a.s.l. This morphological relatedness was considered suf-
unidentified species. In their recent review, C
ficient to include them all under A. dembeensis. On this
ASTIGLIA et
al. (2006) defined this taxon with the acronym ANI-8.
morphological basis, CORTI et al. (1995) attributed the
Furthermore, the situation is particularly confusing in
karyotype of specimens from Koka to A. dembeensis, and
Ethiopia, where a karyotype with 2n=60 and NFa=76 has
so far this has been considered the typical karyotype of
been described in Konso (Gamo-Gofa, South Ethiopia;
this species. However, the species was included by
ORLOV et al., 1992) and another one with 2n=56 and
MUSSER & CARLETON (2005) in A. niloticus. The compar-
NFa=78 in Gambella (BULATOVA et al., 2002).
ison of the cytochrome b sequences of samples of A. dem-
beensis
and A. niloticus from Egypt (although from a
The karyotypic results following the field studies of the
breeding colony, but with very low genetic differences
Staplerat project are presented here for A. nairobae and
from Ethiopian samples) (DUCROZ et al., 1998; CORTI et
for a number of taxa for which there is yet to be any
al., submitted) and the fact that the two share the same
definitive taxonomic solution. The latter highlight a com-
karyotype (C
plex pattern of sibling species and speciating taxa. This is
ORTI et al., 1996) do not support a specific
particularly true for the A. niloticus complex that will be
distinction so that its synonymy with A. niloticus should
discussed first. Two other unknown karyotypes are indi-
be definitively accepted. On the other hand, due to the
cated as ANI-5 and ANI-6.
variability characterizing this complex, CORTI et al. (sub-
mitted) still suggest this taxon should be referred to as A.
cf. niloticus 1.
­ Arvicanthis niloticus (Desmarest, 1822 complex).
Zeway (FET110, FET114, FET125, FET126,
The two taxa A. niloticus and A. dembeensis (Rüppel,
FET127); Ethiopia. Kitale (FKE119); Kenya. The diploid
1842) have been considered either as synonymous (see
number is 2n=62 and the NFa is 62. The autosomal set is
MUSSER & CARLETON, 1993) or as separate species (see
characterized by 58 acrocentrics decreasing in size and a
YALDEN et al., 1976). A. niloticus (type localities: Upper
pair of small metacentrics. The X chromosome is a large
Egypt, Fayum and Giza areas, part of the Nile delta), sur-
submetacentric (the largest chromosome in the set) and
prisingly, has received limited attention in all studies.
the Y chromosome is a metacentric of medium size.
FADDA & CORTI (1998) examined, through three dimen-
Although these Ethiopian and Kenyan specimens seem to
sional geometric morphometrics, the geographic variation
all have the same karyotype, they show a genetic differen-
in A. niloticus along the Nile Valley from Cairo down to
the extreme south of Sudan, also including A. testicularis
tiation in progress (CORTI et al., submitted). Therefore,
(Dollman, 1911) (synonymous of A. niloticus; M
these authors referred to the specimen from Kenya as A.
USSER &
C
cf. niloticus 5 and to those from Ethiopia as A. cf. niloti-
ARLETON, 2005). They evidenced in A. niloticus a north-
south clinal variation in morphometric traits which some-
cus 2. Furthermore, this karyotype is identical to the one
how contrasts in direction that characterizing A. testicula-
described for A. cf. niloticus 1 (Koka, Ethiopia) by CORTI
ris. This was considered by the Authors as sufficient to
et al. (1996) and for A. somalicus (Awash National Park,
keep them as separate taxa, but the problem should be
Ethiopia) by BASKEVICH & LAVRENCHENKO (2000).

Cytotaxonomy of East Africa Rodents
205
opian Rift Valley at lower altitudes. Furthermore, karyo-
typic variants have been found in the Konso (2n = 60 and
NFa = 76) and in the Gambella (2n = 56 and NFa = 78)
regions by BASKEVICH & LAVRENCHENKO (2000), all
resembling morphologically A. niloticus. It should be
noted that these areas are more than 5° latitude south of
the type locality, so that it is questionable whether these
results highlight a chromosomally polytypic or different
species. BASKEVICH & LAVRENCHENKO (2000) provision-
ally named the specimens from Konso and Gambella
Arvicanthis sp.1 and Arvicanthis sp. 2, respectively.
FADDA & CORTI (2001) morphometrically examined the
specimens from Konso, Gambella, and Omo (plus a
Somalian population) and found that they share unique
morphometric characteristics, so that they were provi-
sionally named A. sp. 3.
­ Arvicanthis nairobae (Allen, 1909).
Nairobi, type locality (FKE145); Kenya. The karyo-
type is characterized by 2n=62 and NFa=78. The auto-
somal set is composed of 9 pairs of biarmed autosomes
and 21 pairs of acrocentric chromosomes decreasing in
size. The biarmed chromosomes are represented by two
large submetacentrics, four pairs of medium size submet-
Fig. 4. ­ A) The karyotype of Arvicanthis ANI-5, 2n=56,
acentrics and three pairs of small metacentrics. The X and
NFa=62, X Y. B) The karyotype of Arvicanthis ANI-6,
Y chromosomes are subtelocentric, of large and small
2n=60, NFa=72, X X.
size respectively. This karyotype is identical to the one
described in Tanzania (FADDA et al., 2001; CASTIGLIA et
al., 2003b).
Arvicanthis sp ANI-5. Rongai (FKE100, FKE103,
FKE133, GKE127, GKE147); Kenya. The diploid number
­ Lemniscomys (Trouessart, 1881)
is 2n=56 and the NFa is 62. The autosomes are composed
by three pairs of large metacentrics, one pair of small
The striped grass mice genus Lemniscomys is widely
metacentrics, and 23 pairs of acrocentrics decreasing in
distributed throughout the African sub-Saharan savan-
size (Fig. 4A). The X chromosome is a large submetacen-
nahs and in Northwest Africa. The taxonomy and system-
tric and the Y chromosome is a medium size metacentric.
atics of the genus is complex as it includes twelve recog-
This karyotype is described here for the first time. A
nised species (MUSSER & CARLETON, 2005) which are
banding study to assess the karyotype relationships with
probably at different levels of morphological differentia-
the other Arvicanthis in east Africa will appear elsewhere
tion (ELLERMAN, 1941; VAN DER STRAETEN & VERHEYEN,
(CASTIGLIA et al., 2005). A phylogenetic tree based on the
1980) and are included in different groups. These are the
entire sequence of the cytochrome b gene (CORTI et al.,
"barbarus" group (striped species), including L. barba-
submitted) suggests that this species is a member of the
rus, and L. hoogstrali; the striatus group (spotted spe-
A. niloticus sensu lato group.
cies), including L. bellieri, L. macculus, L. mittendorfi
Arvicanthis sp ANI-6. Zeway (GET129); Ethiopia. The
and L. striatus; the rosalia group (plain coloured species),
diploid number is 60 and the NFa is 72. The karyotype is
which includes L. griselda, L. rosalia, L. linulus, and L.
composed by four pairs of large metacentric chromo-
roseveari. VAN DER STRAETEN & VERHEYEN (1980) found
somes, three pairs of small-sized metacentrics, and 22
morphometric differences within the "striatus" group,
pairs of acrocentrics decreasing in size (Fig. 4B). The X
with L. striatus being similar to L. linulus (Rosalia group)
chromosome is a large submetacentric (Fig. 4B). This
and well differentiated with respect to L. bellieri and L.
karyotype is described here for the first time. A banding
macculus, which instead clusters with L. barbarus. On the
study to assess the karyotype relationships with the other
basis of multivariate morphometrics of skull linear meas-
Arvicanthis in east Africa will appear in print elsewhere
urements, CARLETON & VAN DER STRAETEN (1997)
(CASTIGLIA et al., 2005). ORLOV et al. (1992) found a
recently split L. barbarus into two species, i.e. L. barba-
karyotype in Konso (Gamo-Gofa region, South of Zeway,
rus restricted to scrub vegetation along a narrow coastal
along the Rift Valley) which is very similar to this one,
strip in Morocco, Algeria and Tunisia, and L. zebra with a
with the exception of two additional pairs of small meta-
sub-Saharan distribution. Relationships between species
centrics. It is still questionable whether all these South
still remain unresolved, except for the close relationship
Ethiopian cytotypes represent a chromosomally polytypic
between L. bellieri and L. macculus (DUCROZ et al.,
species. A phylogenetic tree based on the entire sequence
2001).
of the cytochrome b gene (CORTI et al., submitted) has
There is a conspicuous number of karyotypic data
shown that the Zeway karyotype constitutes the sister
showing the occurrence of several different diploid and
group of A. ansorgei and belongs to the West African
fundamental numbers. Considerable chromosomal varia-
clade of Arvicanthis, which extends along the south Ethi-
tion has been shown in West and Central Africa for L.

206
Corti, Castiglia, Colangelo, Capanna, Beolchini, Bekele, Oguge, Makundi, Sichilima, Leirs, Verheyen and Verhagen
striatus by VAN DER STRAETEN & VERHEYEN (1985;
size submetacentric, with a polymorphism occurring in
Burkina Faso; 2n=44, NFa= 58), GAUTUN et al. (1985;
the former (CASTIGLIA et al., 2002b). This karyotype has
Burkina Faso; 2n=43), MATTHEY (1959; Congo; 2n=48,
already been reported and described by FADDA et al.
NFa= ~56), CAPANNA et al. (1997; Benin; 2n=44;
(2001) for other localities from the Maasai steppe.
NFa=72-74), CASTIGLIA et al. (2002b; 2n=44, NFa=68),
DUCROZ (1998) and VAN DER STRAETEN & VERHEYEN
­ Lemniscomys cf. zebra (Heuglin, 1864).
(1978) (Ivory Coast; 2n=44, NFa=58). The large differ-
ences in NFa in these specimens may be due, in part, to
Matongolo (FT50030), Mvumi Mission (GT50333),
different interpretations of small heterochromatic chro-
Ngasumet (G50179, GT50178, FT50174), Kisiwani
mosomal arms. L. bellieri seems to be characterized by a
(GT50429), Mbugani-Chunya (GT50699), Itigi
constant 2n=56 and NFa=60 karyotype in Burkina Faso
(GT50348), Lwami (FT50492, FT50518, FT50524,
and Ivory Coast (V
FT50525, GT50502, GT50523, GT50526, GT50507,
AN DER STRAETEN & VERHEYEN, 1978;
D
GT50501) Ndaleta (GT50152); Tanzania. The 2n is 54,
UCROZ, 1998; TRANIER & GAUTUN, 1979). L. barbarus
also seems to have the same constant karyotype across
and the NFa is 58. Autosomes are represented by one pair
Morocco (S
of large submetacentrics (the largest of the autosomal
TITOU et al., 1997; 2n=54, NFa=58), Ivory
Coast (M
set), by two pairs of very small metacentrics and by acro-
ATTHEY, 1954; 2n=54) and the Algerian Coast
(F
centrics decreasing in size. The X chromosome occurs in
ILIPPUCCI et al., 1986; 2n=54, NFa =58). L. macculus
has been studied in Central African Republic (D
two forms, both submetacentric, differing in the length of
UCROZ,
1998; 2n=56, NFa=62) and L. mittendorfi in Cameroon
the short arm. CASTIGLIA et al. (2002b) studied the occur-
(F
rence of the two forms and found similar frequencies. The
ÜLLING, 1992; 2n=56). Recent findings in Tanzania
(C
Y chromosome is a medium size metacentric. This karyo-
ASTIGLIA et al, 2002b; FADDA et al., 2001) have shown
a karyotype with 2n=54, NFa=68 for L. zebra, and 2n=54,
type has already been reported and described by FADDA et
NFa=62 for L. rosalia. The latter presents striking differ-
al. (2001) for other localities from the Maasai steppe.
ences with the 2n=48, NFa =62 form described by
D
­ Lemniscomys cf. striatus massaicus (Pagenstecher,
UCROZ et al. (1999) for L. rosalia in Kwazulu Natal. The
two differ due to several rearrangements and, on this
1885).
basis, CASTIGLIA et al. (2002b) argued that the two taxa
Rongai (FKE105, GKE116), Nairobi (FKE110,
should be considered as separate species. The type local-
FKE112, GKE125, GKE137, GKE146); Kenya. These
ity of L. rosalia is Monda, Nguru Mtns., Tanzania and,
specimens are characterised by 2n=48 and NFa=54. The
therefore, the name should be kept for the Tanzanian
karyotype is composed of four pairs of meta-submetacen-
specimens, and it was suggested that L. calidior (Thomas
trics, the remaining autosomes being acrocentrics
and Wroughton, 1908) should be used as the oldest avail-
decreasing in size (Fig. 5). The chromosome X is a large
able name for the South African taxon.
submetacentric, the Y is a small metacentric. This karyo-
Furthermore, CASTIGLIA et al. (2002b) maintained that
type is presented here for the first time. The chromosomal
species relationships based on karyotypes contrast with
differences between these specimens and L. striatus from
the views that consider the plain coloured species as
West Africa are significant. A description of chromo-
primitive and the multi-striped ones as derived. L. bel-
somal differences based on differential staining will be
lieri, L. macculus and L. zebra have an ancestral karyo-
described elsewhere.
type and are characterised by a multi-striped or spotted
pelage, while L. rosalia has a derived karyotype and is a
plain coloured species. This hypothesis is also in agree-
ment with the morphometric relationships outlined by
VAN DER STRAETEN & VERHEYEN (1980), who found a
strict morphological similarity between L. barbarus and
L. bellieri /L. macculus.
Here we present the karyotype of L. cf. striatus massai-
cus from Kenya and additional material for L. rosalia and
L. cf. zebra from Tanzania (these two occur sympatrically
in several of the localities studied).
­ Lemniscomys rosalia (Thomas, 1904).
Zoissa (F50083), Jipe (FT50458, FT50460, FT50461,
Fig. 5. ­ The karyotype of Lemniscomys cf. striatus massai-
FT50462, FT50479, GT50442, GT50459, GT50477,
cus, 2n=48, NFa=54, X Y.
GT50478), Kisiwani (GT50428, T50427, FT50391,
T50429), Mbugani-Chunya (T0672), Morogoro (T50201,
FT50208, FT50547, GT50490, GT50491, GT50548),
­ Mus (L., 1758), subgenus Nannomys (Peters, 1876).
Ngasumet (FT50172, FT50186), Dakawa (T50551,
T50552); Tanzania. The species is characterised by 2n=54
The African species of the subgenus Nannomys, known
and NFa=62. The autosomes comprise three pairs of large
also as Leggada, constitute one of the major taxonomic
subtelocentrics, two pairs of small metacentrics and
puzzles and an emblematic group due to their fast rate of
twenty-one pairs of acrocentrics decreasing in size. The X
speciation often associated with chromosomal rearrange-
chromosome is a submetacentric and the Y is a medium
ments. They form a monophyletic group, the ancestor of

Cytotaxonomy of East Africa Rodents
207
which migrated from Asia through Iraq, Iran, and Saudi
The high chromosomal diversification of the Zambian
Arabia into Ethiopia (JOTTERAND, 1972). CORBET (1990)
samples together with the occurrence of "all-acrocen-
has highlighted that although there can be no doubt
trics" and Rb populations in other areas confirm the
regarding the dichotomy between Nannomys and Mus
extensive chromosomal variability of this taxon (MAT-
sensu strictu, the recognition of the former as a separate
THEY, 1964). As the type locality for the species is Cape
genus is still far from being solved. MUSSER & CARLETON
Town (South Africa), the attribution of these specimens
(2005) listed 19 species in the subgenus, but this number
(including those reported by JOTTERAND, 1972) to this
is likely to increase due to the occurrence of cryptic and
species is therefore questionable.
chromosomal species (VEYRUNES et al., 2004). Further-
more, VEYRUNES et al. (2005) recognised seven clades on
the basis of cytochrome b sequences, probably corre-
sponding to more species since some of these may corre-
spond to species complexes.
Previous cytogenetic studies divided the species into
two large cytotaxonomic groups, differing in the mor-
phology of the sex chromosomes (JOTTERAND, 1972). The
sex chromosomes of the first group are primitively acro-
centric. In the second group, both primitive X and Y chro-
mosomes have been translocated onto a pair of auto-
somes. Subsequent analyses showed that three different
pairs of autosomes were involved in the Robertsonian
translocation event (JOTTERAND-BELLOMO, 1986; 1988).
A definite taxonomic revision will require multidiscipli-
nary studies integrating molecular, cytogenetics, morpho-
logical and morphometrics analyses. In addition, the dis-
tributional extent of chromosomal polymorphism and of
the different cytotypes through differential staining and in
situ
hybridisation still needs to be elucidated to provide a
clear assessment of patterns of chromosomal evolution in
this group.
Nannomys sp. Rongai ­ A (minutoides complex), Nai-
robi (FKE106, FKE111, FKE144); Kenya. Species iden-
tification was possible by analysing the skull and cyto-
chrome b phylogeny (unpublished data). The karyotype
of this species is 2n=22, FN=36. The autosomes are con-
stituted by six pairs of metacentrics and four pairs of
acrocentrics; the X chromosome is a large metacentric
and the Y a large submetacentric (Fig. 6A). This karyo-
type is presented here for the first time. The morphology
of the sex chromosomes strongly suggests the fusion of
the original sex chromosomes with a pair of autosomes,
thus indicating the strict similarity of this karyotype to of
N. minutoides from Zambia. Therefore, further banding
analyses are needed to define relationships between these
species within a wider taxonomic framework.
Nannomys sp. (minutoides complex), Mutanda
Research Station, Mutoma, Solwezi (FZM8, GZM16,
Fig. 6. ­ A) The karyotype of Nannomys sp. Rongai ­ A
GZM24); Zambia. The chromosomal complement was
(minutoides complex), 2n=22, FN=36, X Y. B) The karyotype
2n=25 in a male and a female, while another female had
of Nannomys proconodon, 2n=36 and NFa=34, X Y. C) The
2n=24. However, the NF is 36 in all specimens. The kary-
karyotype of Nannomys cf. emesi, 2n=36, FN=36. D) The
otype is composed of five large metacentric chromo-
karyotype of Nannomys sp. Rongai ­ B, 2n=36, NFa=34, X Y.
somes, and six pairs of acrocentrics decreasing in size.
The heterochromosomes are fused with an autosome. In
­ Nannomys proconodon (Rhoads, 1896).
both females studied, one of the X chromosomes showed
a partial deletion, thus resembling the Y in size. This
Zeway (FET105, FET113); Ethiopia. The karyotype is
karyotype differs significantly from the one reported by
characterized by 2n=36 and NFa=34. All chromosomes
JOTTERAND (1972) from Kafue (Zambia, Lusaka region),
are acrocentrics decreasing in size. The X chromosome is
characterized by a 2n=34 with all-acrocentric autosomes.
a large acrocentric with Y being small (Fig. 6B). This
The reduction in diploid number (2n=24-25), together
species was identified in Ethiopia by YALDEN et al. (1976)
with the maintenance of the FN and the occurrence of
on the basis of skull measurements. It was considered a
large biarmed chromosomes is due to the presence of Rb
synonym of M. setulosus by MUSSER & CARLETON (1993)
fusions (for identification, see CASTIGLIA et al., 2002a).
and confirmed hesitantly by YALDEN et al. (1996). The

208
Corti, Castiglia, Colangelo, Capanna, Beolchini, Bekele, Oguge, Makundi, Sichilima, Leirs, Verheyen and Verhagen
type locality is Sheikh Hussein, Ethiopia (East of the Rift
LET (1979) reported additional NFa variability (54 ­ 56).
Valley, at the same latitude of Zeway), and the species
M. erythroleucus ranges from Senegal to Ethiopia and
seems to occur in the same habitats as N. tenellus, i.e.
Uganda and shows 2n=38; the NFa varies from 40 to 60.
from lowland forests to arid savannas.
However, the comparison of the G-banded karyotypes
Nannomys cf. emesi. Kitale (GKE117); Kenya. The
and cytochrome b based phylogeny suggests that the
chromosomal complement is 2n=36, FN=36. All chromo-
specimens showing NFa=40 represent a separate species
somes are acrocentrics decreasing in size (Fig. 6C). The
(VOLOBOUEV et al., 2002b), while the taxonomic status of
sex chromosomes have been not yet identified. This kary-
the other cytotypes (NFa=50, NFa=52-53 and NFa=60)
otype is presented here for the first time, and is very simi-
remains uncertain (VOLOBOUEV et al., 2001; 2002b). M.
lar in morphology N. proconodon. The identification of
shortridgei occurs in the extreme northwest of Botswana
the species was based on skull measurements. N. emesi
and in northeast Namibia. GORDON (1985) reported in
was considered synonymous of M. mahomet by MUSSER
Namibia a karyotype with 2n=36 and NFa=50.
& CARLETON (2005), who underlined that the series of
QUMSIYEH et al. (1990) reported a karyotype with
emesi could not be distinguished from the large samples
2n=32 and NF=50-54 from Kenya which they considered
of mahomet collected by Osgood in Ethiopia.
to be different from M. natalensis and that attributed to
Nannomys sp. Rongai ­ B. (GKE107, FKE136,
M. hildebrandtii. This karyotype was considered by QUM-
GKE140); Kenya. This species carries the same "all acro-
SIYEH et al. (1990) to be similar to the one described by
centric" karyotype (2n=36, NFa=34; Fig. 6D) as N. cf.
CAPANNA et al. (1982) in Somalia as M. huberti, although
emesi, but there are important differences in the external
there was no comparison with type material. QUMSIYEH et
body morphology, the specimens from Kitale being much
al. (1990) argued that M. hildebrandtii (type locality Ndi,
smaller than those from Rongai. No species assignment
Tahita Hills, in Kenya) is an older name for M. huberti
for the moment has been attempted for these specimens.
(type locality Zungeru, N. Nigeria). However, the karyol-
This karyotype is presented here for the first time. The X
ogy for the West African specimens assigned to huberti
and the Y chromosomes are large and medium size acro-
(2n=32, FN=44) is different from that of the Kenyan
centrics, respectively.
specimens that QUMSIYEH et al. (1990) considered to be
hildebrandtii (2n=32, FN=50-54). Therefore, GRANJON et
­ Mastomys (Thomas, 1915).
al. (1997) claimed that these taxa are not synonyms and
M. huberti is a different species, even though specimens
The species of this genus are widespread in the African
from the type region have not yet been investigated kary-
continent where they represent an important component
ologically and the type locality lies outside the distribu-
of the rodent fauna and constitute a serious problem for
tion range of what is currently called M. huberti (GRAN-
agriculture and human plague (SINGLETON et al., 1999).
JON et al., 1997). For the time being there is no indication
However, their taxonomy has been subject to discussion
that hildebrandtii (with a type locality in southern Kenya)
for a long time. Several papers (GREEN et al., 1980;
is different from M. natalensis. Therefore, while waiting
HUBERT et al., 1983; CHEVRET et al., 1994; BRITTON-DRA-
for a definitive comparison with the karyotypes from type
VIDIAN et al., 1995; GRANJON et al., 1996; GRANJON et al.,
localities, M. hildebrandtii should be considered only as a
1997; LECOMPTE et al., 2002; VOLOBOUEV et al., 2002b;
younger synonym of M. natalensis.
COLANGELO et al., in prep.), employing cytogenetic, mor-
phological and molecular approaches, have assessed the
Finally, LAVRENCHENKO et al. (1998) described a new
monophyly of the genus and clarified the phylogenetic
species from the Awash Valley (Ethiopia) and named it
relationships within the genus and what was considered
M. awashensis; its diploid number is 32 and the NFa is
the Praomys sensu lato group. However, the most recent
54, but it differs form M. natalensis in chromosome mor-
morphological and cytogenetic studies have indicated the
phology and C-banding pattern.
possible occurrence of sibling species and/or new cryptic
Three different chromosomal formulas have been
species (LAVRENCHENKO et al., 1998; VOLOBOUEV et al.,
found in our samples. The more common karyotype
2001; 2002b).
shows 2n=32 and NFa=52-54 and here has been attrib-
The four species M. natalensis, M. huberti, M. coucha,
uted to M. natalensis. However, one specimen collected
M. erythroleucus are phylogenetically closely related,
in Ethiopia (Zeway) shares the same 2n=32 and NFa=54
while the taxonomical and systematic position of the
karyotype, but was found not to be closely related phylo-
other taxa included in the genus is still uncertain (G
genetically to M. natalensis on the basis of the analysis of
RAN-
cytochrome b sequences (COLANGELO et al., in prep).
JON et al., 1997; LECOMPTE et al., 2002).
Therefore, in absence of further comparisons, it is here
The available karyotypic data are as follows. The dip-
provisionally referred to M. awashensis described in Ethi-
loid number of M. natalensis is 32 and the NFa varies
opia by LAVRENCHENKO et al. (1998).
from 52 to 54 across the distribution range; this is the
most representative species, and occurs in sub-Saharan
­ Mastomys natalensis (Smith, 1834).
Africa. M. huberti has the same diploid number as M.
natalensis
(2n=32) but the NFa ranges from 44 to 46
Nairobi (FKE142), Kitale (GKE123); Kenya. Maton-
(DUPLANTIER et al., 1990; GRANJON et al., 1997); the
golo (FT50012), Morogoro (GT50198, GT50199,
range is restricted to Mauritania, Mali, Burkina Faso and
FT50200), Singida (GT50197); Tanzania. Mutoma,
Senegal (GRANJON et al., 1997). M. coucha has been
Meheba (GZM2, GZM5, FZM34, GZM36, GZM38);
described from southern Africa (RSA and Zimbabwe)
Zambia. All the karyotypes show 2n=32 and both
and presents 2n=36 and NFa=52-54 (LYONS et al., 1980;
NFa=52 and 54 were found. The most common karyotype
LEE & MARTIN, 1980; GREEN et al., 1980); however, HAL-
is composed of 12 pairs of biarmed chromosomes of

Cytotaxonomy of East Africa Rodents
209
decreasing size and three pairs of acrocentrics. The X
with numbers ranging from 40 up to 60 (MATTHEY,
chromosome is a large submetacentric, and the Y chro-
1965b; 1966a; 1967; KRAL, 1971; TRANIER, 1974;
mosome is a large acrocentric. This karyotype is compa-
HUBERT et al., 1983; VIEGAS-PÉQUIGNOT et al., 1987;
rable to those described for M. natalensis from other
ORLOV et al., 1989; BRITTON-DAVIDIAN et al., 1995;
localities (MATTHEY, 1955, 1966a, b; CAPANNA et al.,
CODJA et al., 1996; BULATOVA et al., 2002; VOLOBOUEV et
1982; HUBERT et al., 1983; ORLOV et al., 1989; DUPLAN-
al., 2001; 2002a). According to the analysis of cyto-
TIER et al., 1990; BASKEVICH & ORLOV, 1993; LEIRS,
chrome b (COLANGELO et al., in prep.) the genetic diver-
1994; BRITTON-DAVIDIAN et al., 1995; CODJA et al., 1996;
gence between these two cytotypes it is very low in spite
FADDA et al., 2001).
of the remarkable NFa variation.
­ Stenocephalemys (Frick, 1914).
The genus now includes the three Ethiopian endemic
species S. albocaudata, S. albipes and S. griseicauda
(MUSSER & CARLETON, 2005). It has been recognised
recently, on the basis of cytogenetic, molecular and mor-
phometric data (CHEVRET et al., 1994; CORTI et al., 1995,
1999; FADDA et al., 2001; LAVRENCHENKO et al., 1999;
LECOMPTE et al., 2002) that they constitute a mono-
phyletic assemblage phylogenetically related to species of
Praomys, Mastomys, Myomyscus, Heimyscus, and Hylo-
Fig. 7. ­ The karyotype of Mastomys erythroleucus, Zeway,
myscus (LECOMPTE et al., 2002). These studies have also
Ethiopia 2n=38, NFa=53-54, X Y. Note the heteromorphic
shown that S. albipes should not be referred to Myomys
condition of one of the smallest metacentrics.
albipes or Praomys (MISONNE, 1969; QUMSIYEH et al.,
1990; YALDEN et al., 1976).
­ Mastomys cfr. awashensis (Lavrenchenko et al., 1998).
This genus has long been considered taxonomically
confused, as it was included in a group of genera together
Zeway (ET102F); Ethiopia. The karyotype presents the
with Praomys, Mastomys, Hylomyscus, Colomys and
same chromosomal formula as M. natalensis (2n=32 and
Stenocephalemys which are closely related (CHEVRET et
NFa=54) and is composed of thee pairs of acrocentrics
al., 1994). There are available cytogenetic data for S. albi-
and twelve pairs of biarmed chromosomes, six of which
pes, 2n=46 and NFa=50-53 (various localities in Ethio-
are metacentrics and six subtelocentrics. The X and the Y
pia), S. albocaudata (2n=54, NFa=60) and S. griseicauda
chromosomes are submetacentrics, the former being
(2n=54, NFa=54) (CORTI et al., 1999).
medium-large and the latter medium-small. A preliminary
G-banding analysis (the complete comparison will appear
elsewhere) has shown differences probably due to dele-
­ Stenocephalemys albipes (Rüppell, 1842).
tions and/or additions in the autosomal complement.
According to the cytochrome b sequences (COLANGELO et
Mugo (FET108, FET109, FET111, GET121, GET123);
al., in prep), this specimen is not related to the M.
Ethiopia. The species is widespread all over Ethiopia in
natalensis clade. Therefore, there is strong evidence sug-
forest blocks at altitudes ranging from 800 m up to 3300
gesting that this taxon represents a separate species from
mt a.s.l. (YALDEN et al., 1976; AFEWORK BEKELE & CORTI,
M. natalensis. Further analysis will probably confirm the
1997). The diploid number is 46 and the NFa is 50-53
attribution of this taxon to M. awashensis.
(Fig. 8). The X is a large submetacentric and the Y (not
shown) a medium size metacentric. All karyotypes are
composed of 4 pairs of large biarmed chromosomes, 17
­ Mastomys erythroleucus (Temminck, 1953).
pairs of acrocentric chromosomes decreasing in size (on
some of them very small arms are visible), and of one pair
Rongai (FKE101, FKE115, GKE132); Kenya. Zeway
of small biarmed chromosomes. The chromosome pair
(GET128, FET122, FET128); Ethiopia. The Kenyan
numbers 1 and 3 are polymorphic (the latter not shown),
karyotype consists of 12 pairs of biarmed and six pairs of
i.e. they are found either as acrocentrics or submetacen-
acrocentric chromosomes. The X chromosome is a
trics due to a pericentric inversion. CORTI et al. (1999)
medium size submetacentric and the Y is a small submet-
described the karyotype of this species from other locali-
acentric. The Ethiopian karyotype is represented by 17-18
ties in Ethiopia, and found the same diploid number and
biarmed and 19-20 acrocentric chromosomes. The X is a
the same variation in NFa (50-53), due to the occurrence
large submetacentric and the Y is a small subtelocentric
of a polymorphism. It is interesting to note that the poly-
(Fig. 7). All the samples show the typical M. erythroleu-
morphism found in Zeway involves the same autosomes
cus diploid number (2n=38). However, the Ethiopian and
as those described by CORTI et al. (1999) except for one
Kenyan specimens differ in NFa which is 52-53 in the
specimen (ET123) that showed a polymorphism involv-
former and 60 in the latter. A wide variability in NFa has
ing the largest chromosome pair (n° 1) of the entire auto-
been reported from other East and West African localities,
somal set (Fig. 8).

210
Corti, Castiglia, Colangelo, Capanna, Beolchini, Bekele, Oguge, Makundi, Sichilima, Leirs, Verheyen and Verhagen
karyotype is presented here for the first time, but it resem-
bles the one described from Katanga by MATTHEY (1971).
Fig. 8. ­ The karyotype of Stenocephalemys albipes. 2n=46,
NFa=50-53, X X. Note the heteromorphic condition of the
largest autosome.
­ Grammomys (Thomas, 1915).
The genus includes 12 species (MUSSER & CARLETON,
2005). The species cytogenetically studied so far show a
wide variation in B chromosomes. CIVITELLI et al. (1989)
described specimens of G. gazellae (synonym of G. mac-
millani)
from Central Africa 2n=56-71, with 2-17 B chro-
mosomes. The occurrence of B-chromosomes as well as
Robertsonian fusions has been documented by ROCHE et
al. (1984) in five specimens (a mother and four pups) of
G. dolichurus from Somalia (2n=54-61, NFa=70-75).
MATTHEY (1971) described the karyotype of G. surd-
aster from Katanga (2n=52; NFa=62). FADDA et al.
(2001) found a karyotype in the Maasai steppe with
2n=27 and NFa=39 that could not be allocated to any spe-
cies. They maintained that, after craniological and cranio-
metrical comparison with extensive Tanzanian material
including most of the relevant type-specimens, the taxo-
nomic situation of the Grammomys-Thamnomys complex
Fig. 9. ­ A) The karyotype of Grammomys sp. Surdaster
of this part of Africa is yet to be established. More
(Zambia), 2n=50-51, NFa=61, X X. Note the heteromorphic
recently, on the basis of 16S rRNA gene analyses of
condition of one of the smallest metacentrics an of the X. B)
The karyotype of Grammomys sp. surdaster (Tanzania),
extensive samples, VERHEYEN et al. (2003) reported the
2n=42, NFa=64, X Y. C) The karyotype of Grammomys sp.
occurrence, in this part of Africa, of at least three species
butingi, 2n=20, NFa=31, X Y. Note the heteromorphic condi-
complexes, i.e. G. macmillani, G. dolichurus and G. surd-
tion of one of the medium size chromosomes.
aster. Taxonomic attribution made here is based on mor-
phological comparison with larger series and mtDNA
analysis (VERHEYEN et al., 2003), but apart from including
Grammomys sp. surdaster complex Kitundu Forest,
the cytotype in a species group, no definitive allocation or
Uluguru Mt. range (FT50235, GT50237); Tanzania. The
new classification was possible.
karyotype is 2n=42, and the NFa is 64 (Fig. 9B). The
karyotype comprises 12 pairs of biarmed chromosomes
­ Grammomys sp. surdaster (Thomas and Wroughton,
decreasing in size, from large to small, and 8 pairs of
1908) complex.
acrocentrics of medium to small size. The X is a large
Mutanda Research Station (FZM25, GZM14); Zambia.
metacentric and the Y is the smallest acrocentric. This
The diploid number is 2n=51 in the male and 2n=50 in the
karyotype is presented here for the first time.
female and the NFa is 61 (Fig. 9A). The difference
between the male and female depends on a structural
­ Grammomys sp. butingi (Thomas, 1911 complex).
Robertsonian fusion in the former concerning the largest
metacentric (not shown in the figure). The autosomes are
Jipe (FT50485); Tanzania. The karyotype is character-
represented by one pair of large metacentric chromo-
ized by 2n=20 and NFa=31. The autosomal set is repre-
somes, a heteromorphic pair formed by a small metacen-
sented by two pairs of very large metacentrics, two pairs
tric and a small acrocentric, five pairs of small metacen-
of medium size subtelocentrics, a pair of medium size
trics, and 17 pairs of acrocentrics decreasing in size (Fig.
heteromorphic chromosomes (a metacentric and a subte-
9A). There are two different X configurations: a submeta-
locentric) and four pairs of small metacentrics (Fig. 9C).
centric (in the male) and a subtelocentric occurring
The X chromosome is a medium size subtelocentric and
together with the submetacentric (in the female). The Y
the Y a small metacentric. This karyotype is presented
chromosome is a small subtelocentric (not shown). This
here for the first time.

Cytotaxonomy of East Africa Rodents
211
MYOXIDAE (Gray, 1821).
Graphiurus sp. 2. Chunya (FTZ506); Tanzania. Unfor-
tunately the quality of the preparations is only sufficient
­ Graphiurus (Smuts, 1832).
for a general description of the karyotype. The 2n is 50
and the NFa is 66. The karyotype is composed of nine
The systematic relationships and the taxonomy within
pairs of metacentrics decreasing in size and 15 pairs of
the genus are at present not fully resolved (BENTZ &
acrocentrics decreasing in size (Fig. 10C). Both the X and
MONTGELARD, 1999; GENEST-VILLARD, 1978; SCHLITTER
the Y are acrocentrics but the former is the largest one of
et al., 1985; HOLDEN, 1993; 1996). Revisions made over
the complement and the latter is very small. This karyo-
the last twenty years, all of which are exclusively based
type is presented here for the first time.
on size and morphology, have reached different conclu-
sions: the number of species described varies from five
(MEESTER & SETZER,1971) and six (GENEST-VILLARD,
1978), to fourteen (HOLDEN, 1993). More recently, MON-
GELARD et al. (2003) examined six taxa of Graphiurus
through nuclear and mitochondrial genes, amongst which
G. microtis from Tanzania. They showed that the low res-
olution found for the genus most probably accounts for
their rapid diversification.
In general, the karyotypes of Graphiurus are character-
ized by the prevalence of biarmed autosomes (ZIMA et al.,
1994), but diploid numbers are known for only a few spe-
cies: G. cf. parvus, 2n=70 (DOBIGNY et al., 2002); G. muri-
nus, 2n=70 (TRANIER & GAUTUN, 1979); G. hueti, 2n=40
(TRANIER & DOSSO, 1979). One should recall, however,
that chromosomal variation is common in Myoxidae, as
shown in the European Eliomys (FILIPPUCCI et al., 1988;
1990) and it would not be surprising to find an even larger
one in Graphiurus.
Five individuals from three different localities were
analyzed and their karyotypes are presented here.
Because their taxonomy is still under discussion (the most
recent checklist by HOLDEN, 1993, highlighted that taxo-
nomic assignment must be considered provisional), we
provide no definite specific allocation for our specimens.
For the moment, we refer to them as belonging to
Graphiurus cf. murinus, Graphiurus sp. 1 and Graphiu-
Fig. 10. ­ A) The karyotype of Graphiurus cf. murinus,
rus sp. 2.
2n=60, NFa=76, X Y. B) The karyotype of Graphiurus sp. 1
(Zambia), 2n= 54, NFa=78, X Y. C) The karyotype of
Graphiurus cf. murinus. Zeway (GET100, GET 104,
Graphiurus sp. 2 (Tanzania), 2n=50, NFa=66, X Y.
FET106); Ethiopia. The karyotype is characterized by
2n=60 and NFa=76. The autosomal set is composed of a
large pair of metacentrics, one medium size pair of meta-
centrics, four pairs of subtelocentrics of medium size,
DISCUSSION
three pairs of medium to small metacentrics and 20 pairs
of acrocentrics decreasing in size (Fig. 10A). The X chro-
mosome is a large acrocentric and the Y a very small one.
The number of karyotypes discussed here totals 37.
This karyotype is presented here for the first time. Y
Seventeen are described here for the first time. For four
AL-
species or species complexes, we report chromosomal
DEN et al. (1976) attributed to this species the "relative
small number of records" for Ethiopia. The type locality
variants and for sixteen karyotypes additional localities of
of the species is the Cape of Good Hope, South Africa,
occurrence. Some of the karyotypes characterize taxa that
but it is questionable whether this wide distribution
have been either fully accepted by the most recent check-
reflects the occurrence of a single species.
lists or need taxonomic re-evaluation or a new descrip-
tion. The latter are Cricetomys cf. gambianus, Saccos-
Graphiurus sp. 1. Meheba (FZM6); Zambia. The dip-
tomus cf. elegans, Gerbilliscus nigricaudus, G. cfr.
loid number is 2n=54 and NFa=78. The karyotype is
muansae, Arvicanthis (ANI-5 and ANI-6), A. nairobae,
composed of nine pairs of metacentric chromosomes
Acomys selousi, Lemniscomys zebra, L. cf. striatus mas-
decreasing in size (from the largest of the entire set to
saicus, Nannomys proconodon, Nannomys cf. emesi,
nearly the smallest), four pairs of subtelocentrics, and 13
Nannomys sp., Grammomys surdaster, Grammomys sp.1
pairs of acrocentrics decreasing in size (from the second
and sp. 2, Graphiurus cf. murinus, Graphiurus sp. 1 and
in size of the entire set to the smallest one) (Fig. 10B).
sp. 2. In all other cases, as the karyotypes have been or
The X chromosome is a medium size metacentric, and the
will be described elsewhere, we have reported the occur-
Y is the smallest element of the set. This karyotype is pre-
rence of chromosomal variants, new sampling sites or
sented here for the first time.
have provided a more precise taxonomic attribution. Fur-

212
Corti, Castiglia, Colangelo, Capanna, Beolchini, Bekele, Oguge, Makundi, Sichilima, Leirs, Verheyen and Verhagen
thermore, we have highlighted problems in taxonomy,
be involved in this project. The Director of the Pest Manage-
due to the occurrence of sibling and cryptic species or
ment Centre (SPMC) (Prof. R.S. Machang'u) provided both
species complexes, for which further detailed analyses are
moral and logical support in setting up project activities. Many
needed.
individuals were involved in the field work, in particular Dr.
Apia W. Massawe, Dr. Loth Mulungu, Mr. Georgies Mgode and
There is a general pattern emerging from this report
Mr. Kevin Kessy (Morogoro) and in laboratory work Alessandra
that confirms all recent suggestions regarding African
Scanzani (Rome), Protas P. H. Tesha (Morogoro). We appreciate
rodent taxonomy and systematics researchers, i.e. biodi-
their help. We also wish to thank Dr Carlo Fadda who played a
versity is much higher than has been previously esti-
fundamental role in the collection of specimens from various
mated, at least that revealed by cytogenetics, so that the
parts of Tanzania and who helped in the laboratory work in
list of 386 species (MUSSER & CARLETON, 1993) will
Rome and Antwerp, with support from the SUA-VLIR program.
increase rapidly. This is true for the all genera investi-
Patrick Mwanjabe (Rodent Control Centre, Morogoro, Tanza-
gated, with the exception of Aethomys, for which there
nia) participated initially in the project. Sadly, he died after a
was no apparent problem of taxonomic attribution at spe-
serious illness. He will be fondly remembered for his spirit,
competence and ability to establish friendship through real
cies level, but for which karyotypes are known from a
human contacts with all partners. For this reason we dedicate
limited number of localities (one only for A. kaiseri) and
this paper to his memory.
variants cannot be excluded a-priori.
However, it is difficult to state whether the chromo-
REFERENCES
somal variants detected warrant a different species attri-
bution in all cases. For many this is undoubtedly true.
Acomys cf. selousi, Arvicanthis ANI-5, Arvicanthis ANI-
AFEWORK BEKELE, E. CAPANNA, M. CORTI, L.F. MARCUS & D.A.
SCHLITTER (1993). Systematics and geographic variation of
6, Nannomys tenellus, N. proconodon, N. cf. emesi, Nan-
Ethiopian Arvicanthis (Rodentia, Muridae). J. Zool., Lond.,
nomys sp. (Rongai), Grammomys sp.1 and sp. 2 represent
230: 117-134.
cases that would strongly suggest a specific attribution.
AFEWORK BEKELE & M. CORTI (1997). Forest blocks and alti-
For others, it would be premature to do so until more
tude as indicators of Myomys albipes (Mammalia: Rodentia)
extensive analyses have been performed, possibly includ-
distribution in Ethiopia. Trop. Zool., 10: 287-293.
ing estimates of gene flow between cytotypes. For exam-
ALLEN, G.M. (1939). A checklist of African mammals. Bulletin
ple, the different karyotypes of Cricetomys cf. gambianus
of the Museum of Comparative Zoology at Harvard College,
found in West (GRANJON et al., 1992; CODJA et al., 1994)
83: 1-763.
and East Africa could correspond to different biological
BAKER, R.J., M.B. QUMSIYEH & I.L. RAUTENBACH (1988). Evi-
species or represent extremes of some form of chromo-
dence for eight tandem and five centric fusions in the evolu-
somal variation. Nonetheless, these data and parallel anal-
tion of the karyotype of Aethomys namaquensis A. Smith
yses suggest that at least seven specific names that were
(Rodentia: Muridae). Genetica, 76: 161-169.
not included in M
BAROME, P.-O., M. MONNEROT & J.-C. GAUTUN (1998). Intrage-
USSER & CARLETON's (1993) checklist
neric phylogeny of Acomys (Rodentia, Muridae) using mito-
need to be restored and accepted at the full species rank.
chondrial gene cytochrome b. Mol. Phyl. Evol., 9: 560-566.
These are Saccostomus cf. elegans, Gerbilliscus cfr.
BAROME, P., V. VOLOBOUEV, M. MONNEROT, J. MFUNE, W. CHI-
muansae, Acomys cf. selousi, Lemniscomys cf. zebra,
TAKAULI, J.C. GAUTUN & C DENYS (2001). Phylogeny of
Nannomys proconodon, N. cf. emesi and Grammomys
Acomys spinosissimus (Rodentia, Muridae) from north
surdaster.
Malawi and Tanzania: evidence from morphological and
Within species chromosomal variation is a common
molecular analysis. Biol. J. Linn. Soc., 738: 321-340.
phenomenon in rodents (recent literature is rich in refer-
BASKEVICH, M.I. & L.A. LAVRENCHENKO (2000). Review of
ences; see, for example, those cited here for Arvicanthis)
karyological studies and the problems of systematics of Ethi-
and the cases we report here of the Stenocephalemys albi-
opian Arvicanthis Lesson, 1842 (Rodentia: Muridae). Bonn.
zool. Monogr
., 46: 209-215.
pes and Saccostomus cf. elegans are evident examples.
B
These could represent instances of raciation but, for the
ASKEVICH, M.I. & V.N. ORLOV (1993). Karyological differenti-
ation of rats of the genus Mastomys (Rodentia, Muridae)
moment, this is impossible to establish. On the contrary,
from the central part of Ethiopian Rift Valley. Zool. Zhur.,
the cytotypes of the Arvicanthis niloticus complex
72: 112-121.
(Zeway, Kitale, ANI-5) clearly suggest a process of diver-
BATES, P.J.J. (1988). Systematics and zoogeography of Tatera
gence in action that might have reached full speciation as
(Rodentia: Gerbillinae) of north-east Africa and Asia. Bon.
suggested by the analysis of the cytochrome b mitochon-
Zool. Beitr. 39: 265-303.
drial gene (CORTI et al., submitted). However, we are cau-
BENAZZOU, T., E. VIEGAS-PÉQUIGNOT, M. PROD'HOMME, M.
tious when considering most of the taxa studied here, as
LOMBARD, F. SETTER, B. DUTRILLAUX (1984). Chromosomal
we believe that extensive comparisons with type material
phylogeny of Gerbillidae. III. Species study of the genera
using a multidisciplinary approach is needed before
Tatera, Taterillus, Psammomys and Pachyuromys. Ann.
reaching a definite taxonomic conclusion.
Genet. 27: 17­26.
BENTZ, S. & C. MONTGELARD (1999). Systematic position of the
African dormouse Graphiurus (Rodentia, Gliridae) assessed
ACKNOWLEDGEMENTS
from Cytochrome b and 12S rRNA mitochondrial genes. J.
Mamm. Evol., 6: 67-83.
This work was made possible thanks to European Union's 5th
BREED, W.G., G.A. COX, C.M. LEIGH & P. HAWKINS (1988).
Framework RTD Program "STAPLERAT", grants MURST 40%
Sperm head structure of a murid rodent from southern
(E.C.), Fondo Musei Naturalistici Accademia Nazionale dei
Africa: the red veld rat Aethomys chrysophilus. Gamete Res.,
Lincei (E.C.), Progetti d'Ateneo (M.C.), and SUA-VLIR Project
19: 191-202.
(H.L., R.V., W.V.). We wish to acknowledge the permission
BRITTON-DAVIDIAN, J., J. CATALAN, L. GRANJON & J.M.
given by Sokoine University of Agriculture to permit its staff to
DUPLANTIER (1995). Chromosomal phylogeny and evolution

Cytotaxonomy of East Africa Rodents
213
in the genus Mastomys (Mammalia, Rodentia). J. Mamm.,
CIVITELLI, M.V., P. CONSENTINO & E. CAPANNA (1989). Inter-
76: 248-262.
and intra-individual chromosome variability in Thamnomys
BULATOVA, N., L. LAVRENCHENKO, V. ORLOV & A. MILISHNIKOV
(Grammomys) gazellae (Rodentia, Muridae). B-chromo-
(2002). Notes on chromosomal identification of rodent spe-
somes and structural heteromorphisms. Genetica, 79: 93-
cies in western Ethiopia. Mammalia, 66: 128-132.
105.
CAPANNA, E., M.V. CIVITELLI & A. CERASO (1982). Karyotypes
CIVITELLI, M.V., R. CASTIGLIA, J.-Cl. CODJA & E. CAPANNA
of Somalian rodent populations. 3. Mastomys huberti
(1995). Cytogenetics of the genus Arvicanthis (Rodentia,
(Wroughton, 1908) (Mammalia Rodentia). Mon. Zool. It., 5:
Murinae). 1. Arvicanthis niloticus from Republic of Benin
141-152.
(West Africa). Z. Säugetierkunde, 60: 215-225.
CAPANNA, E., M.V. CIVITELLI & M.G. FILIPPUCCI (1985). Karyol-
CODJA, J.T.Cl., C. CHRYSOSTOME, M.V. CIVITELLI & E. CAPANNA
ogy of 11 species of African Rodents. IV Int. Ther. Cong.,
(1994). Les chromosomes des Rongeurs du Bénin (Afrique
Edmonton, Canada, Aug. 13-20.
de l'Ouest): 1. Cricetidae. Rend. Fis. Acc. Lincei, s. 9, v. 5:
CAPANNA, E. & M.V. CIVITELLI (1988). A cytotaxonomic
277-287.
approach to the systematics of Arvicanthis niloticus (Des-
CODJA, J.T.Cl., M.V. CIVITELLI, D. BIZZOCO & E. CAPANNA
marest 1822) (Mammalia Rodentia). Trop. Zool., 1: 29-37.
(1996). Les chromosomes de Mastomys natalensis et Masto-
CAPANNA, E., AFEWORK BEKELE, M. CAPULA, R. CASTIGLIA,
mys erythroleucus du Sud Bénin (Afrique de l'Ouest): nouv-
M.V. CIVITELLI, J-Cl. CODJA, M. CORTI & C. FADDA (1996).
elles précisions sur la variabilité chromosomique. Mamma-
A multidisciplinary approach to the systematics of the genus
lia, 60: 299-303.
Arvicanthis, Lesson 1842 (Rodentia, Muridae). Mammalia,
COLANGELO, P., M.V. CIVITELLI & E. CAPANNA (2001). Morphol-
60: 677-696.
ogy and chromosomes of Tatera Lataste 1882 (Rodentia
CAPANNA, E., J.C. CODJIA, C. CHRYSOSTOME & M.V. CIVITELLI
Muridae Gerbillinae) in West Africa. Trop. Zool. 14: 243­
(1997). Les chromosomes des rongeurs du Benin (Afrique
253.
de L'Ouest): 3. Murinae. Rend. Fis. Acc. Lincei, 8: 25­37.
COLANGELO, P., M. CORTI, E. VERHEYEN, F. ANNESI, N. OGUGE,
CAPULA, M., M.V. CIVITELLI, M. CORTI, AFEWORK BEKELE & E.
R.H. MAKUNDI & W. VERHEYEN (2005). Mitochondrial phyl-
CAPANNA (1997). Genetic divergence in the genus Arvican-
ogeny reveals differential modes of chromosomal evolution
this (Rodentia: Murinae). Biochem. Syst. Ecol., 25: 403-409.
in the genus Tatera (Rodentia: Gerbillinae) in Africa. Mol.
CARLETON, M.D. & E. VAN DER STRAETEN (1997) Morphologi-
Phylogenet. Evol., 35: 556­568.
cal differentiation among Subsaharan and North African
COLANGELO, P., L. GRANJON, P.J. TAYLOR & M. CORTI. Evolu-
populations of the Lemniscomys barbarus complex (Roden-
tionary systematics in African Gerbilline rodents (g. Gerbil-
tia: Muridae). Proc. Biol. Soc. Washington, 110: 640­680.
liscus and Gerbillurus): inference from mitochondrial genes.
CASTIGLIA, R., E. GORNUNG & M. CORTI (2002a). Cytogenetic
Submitted.
analyses of chromosomal rearrangements in Mus minutoides
CORBET, G.B. (1990). The relevance of metrical, chromosomal
/ musculoides from North-West Zambia through mapping of
and allozyme variation to the systematics of the genus Mus.
the telomeric sequences (TTAGGG)n and Banding tech-
Biol. J. Linn. Soc., 41: 5-12.
niques. Chromosome Res., 10: 399-406.
CORBET, G.B. & J.E. HILL (1991). A world list of mammalian
CASTIGLIA, R., C. FADDA, M. CORTI, A. SCANZANI, W. VERHEYEN
species. Oxford: Oxford Univ. Press.
& E. CAPANNA (2002b). Chromosomal evolution in the Afri-
C
can Arvicanthine rats (Murinae, Rodentia): comparative
ORTI, M. (2002). Rodent taxonomy at the end of the 20th cen-
tury. Delectatur musica. Bull. Inst. Roy. Sci. Nat. Belg., Biol-
cytogenetics of Lemniscomys (L. zebra, L. rosalia, L. stria-
ogie, 71: 17-33.
tus) and Arvicanthis dembeensis. J. Zool. Syst. Evol. Res.,
40: 223-231.
CORTI, M., M.V. CIVITELLI, AFEWORK BEKELE, R. CASTIGLIA &
C
E. CAPANNA (1995). The chromosomes of three endemic
ASTIGLIA, R., M. CORTI, P. COLANGELO, F. ANNESI, E.
C
rodents of the Bale mountains, South Ethiopia. Rend. Fis.
APANNA, W. VERHEYEN, A. SICHILIMA & R. MAKUNDI
(2003a). Chromosomal and molecular characterization of
Acc. Lincei, s. 9, v. 6: 157-164.
Aethomys kaiseri from Zambia and Aethomys chrysophilus
CORTI, M., M.V. CIVITELLI, R. CASTIGLIA, AFEWORK BEKELE &
from Tanzania (Rodentia, Muridae). Hereditas, 139: 81-89.
E. CAPANNA (1996). Cytogenetics of the genus Arvicanthis
C
(Rodentia, Muridae). 2. The chromosomes of three species
ASTIGLIA, R., M. CORTI, TESHA PROTAS, A. SCANZANI, C.
F
from Ethiopia: A. abyssinicus, A. dembeensis and A. blicki.
ADDA, E. CAPANNA & W. VERHEYEN (2003b). Cytogenetics
of the genus Arvicanthis (Rodentia, Muridae). 3. Compara-
Z. Säugetierkunde, 61: 339-351.
tive cytogenetics of A. neumanni and A. nairobae from Tan-
CORTI, M. & C. FADDA (1996). Systematics of Arvicanthis
zania. Genetica, 118: 33-39.
(Rodentia, Muridae) from the Horn of Africa: a geometric
C
morphometrics evaluation. It. J. Zool., 63: 185-192.
ASTIGLIA, R., AFEWORK BEKELE, R. MAKUNDI, N. OGUGE & M
CORTI (2006). Chromosomal diversity in the genus Arvican-
CORTI, M., A. SCANZANI, A.R. ROSSI, M.V. CIVITELLI, AFEWORK
this from the horn and east Africa. A taxonomic and Phylo-
BEKELE & E. CAPANNA (1999). Karyotypic and genetic
genetic evaluation. J. Zool. Syst. Evol. Res. In press.
divergence in the Ethiopian Myomys-Stenocephalemys com-
CHEVRET, P., L. GRANJON, J.M. DUPLANTIER, C. DENYS & F.
plex (Mammalia, Rodentia). It. J. Zool., 66: 341-349.
CATZEFLIS (1994). Molecular phylogeny of the Praomys
CORTI, M., R. CASTIGLIA, F. ANNESI & W. VERHEYEN (2004).
complex (Rodentia: Murinae): a study based on DNA/DNA
Mitochondrial sequences and karyotypes reveal hidden
hybridization experiments. Zool. J. Linn. Soc., 112: 425-442.
diversity in African pouched mice (subfamily Cricetomyi-
CHEVRET, P. & G. DOBIGNY (2005). Systematics and evolution of
nae, genus Saccostomus). J. Zool., London, 262: 1­12.
the subfamily Gerbillinae (Mammalia, Rodentia, Muridae).
CORTI, M., F. ANNESI, E. VERHEYEN & W. VERHEYEN Mitochon-
Mol. Phyl. Evol., 35: 674­688.
drial phylogeny and chromosomal evolution of Arvicanthis
CHIMIMBA, C.T. (1998). A taxonomic synthesis of southern Afri-
(Rodentia, Muridae). Submitted.
can Aethomys (Rodentia: Muridae) with a key to species.
DAVIS, D.H.S. (1975). Genus Aethomys. In: MEESTER, J. & W.H.
Mammalia, 62: 427-437.
SETZER (eds), The Mammals of Africa: an identification
CHIMIMBA, C.T., N.J. DIPPENAAR & T.J. ROBINSON (1999). Mor-
manual. Part 6.6. Smithsonian Institution Press, Washington:
phometric and morphological delineation of southern Afri-
1-5.
can species of Aethomys (Rodentia: Muridae). Biol. J. Linn.
DELANY, M.J. (1975). The rodents of Uganda. London: Trustees
Soc., 67: 501-527.
of the British Museum (N.H.).

214
Corti, Castiglia, Colangelo, Capanna, Beolchini, Bekele, Oguge, Makundi, Sichilima, Leirs, Verheyen and Verhagen
DENYS, C. (1999). Of mice and men. Evolution in the East and
FILIPPUCCI, M.G., M.V. CIVITELLI & E. CAPANNA (1988). Evolu-
South Africa during Plio-Pleistocene times. In: BROMAGE,
tionary genetics of the dormouse, Eliomys quercinus Wag-
T.G. & F. SCHRENK (eds), African biogeography, climate
ner, 1840. 1. Karyotype divergence. Boll. Zool., 55: 35-45.
change, and human evolution. Oxford University Press New
FILIPPUCCI, M.G., F. KATZEFLIS & E. CAPANNA (1990). Evolu-
York: 226-252.
tionary genetics and systematics of the garden dormouse,
DIPPENAAR, N.J. & I.L. RAUTENBACH (1986). Morphometrics
Eliomys Wagner, 1840 (Gliridae, Mammalia): 3. Further
and karyology of the Southern Africa species of the genus
karyological data. Boll. Zool., 57: 149-152.
Acomys I. Geoffroy Saint-Hilaire, 1838 (Rodentia: Muri-
FÜLLING, O. (1992). Ergänzende Angaben über Gaumenfalten-
dae). Ann. of the Transvaal Museum, 34: 129-183.
muster von Nagetieren (Mammalia: Rodentia) aus Kamerun.
DOBIGNY, G., A. NOMAO & J.-C. GAUTUN (2002). A cytotaxo-
Bonn. Zool. Beitr., 43: 415­421.
nomic survey of Rodents from Niger: implications for sys-
GAUTUN, J.-C., M. TRANIER & B. SICARD (1985). Liste Prelimi-
tematics, biodiversity and biogeography. Mammalia, 66:
narie des rongeurs du Burkina-Faso (ex Haute-Volta). Mam-
495-523.
malia, 49: 537­542.
D
G
OBZHANSKY, T. (1937). Genetics and the origin of species.
AUTUN, J.C., I. SANKHON & M. TRANIER (1986). Nouvelle con-
Columbia University Press.
tribution à la connaissance des rongeurs du massif guinéen
des monts Nimba (Afrique occidentale). Systématique et
DUCROZ, J.F. (1998). Contribution des approches cytogenetique
aperçu quantitatif. Mammalia, 50: 205-217.
et moleculaire a l'etude systematique et evolutive de genres
de rongeurs Murinae de la `division' Arvicanthis. PhD Dis-
GENEST-VILLARD, H. (1967). Revision du genre Cricetomys
sertation. Museum National d'Histoire Naturelle, Paris.
(Rongeurs, Cricetidae). Mammalia, 31: 390-432.
GENEST-VILLARD, H. (1978). Révision systématique du genre
DUCROZ, J.F., L. GRANJON, P. CHEVRET, J.M. DUPLANTIER, M.
Graphiurus (Rongeurs, Gliridae). Mammalia, 42: 391-426.
LOMBARD & V. VOLOBOUEV (1997). Characterization of two
GORDON, D.H. (1985). Cytogenetics of Shortridge's Mouse
distinct species of Arvicanthis (Rodentia, Muridae) in West
(Mastomys shortridgei) (Rodentia: Muridae) from southwest
Africa: cytogenetic, molecular and reproductive evidence. J.
Africa/Namibia. IV Ther. Cong. Edmonton, Canada, August
Zool., Lond., 241: 709-723.
13-20, 1985.
DUCROZ, J.F., V. VOLOBOUEV & L. GRANJON (1998). A molecular
GORDON, D.H. (1986). Extensive chromosomal variation in the
perspective on the systematics and evolution of the genus
pouched mouse, Saccostomus campestris (Rodentia, Criceti-
Arvicanthis (Rodentia, Muridae): Inferences from complete
dae) from Southern Africa: a preliminary investigation of
cytochrome b gene sequences. Mol. Phylogenet. Evol., 10:
evolutionary status. Cimbebasia, 8: 37-47.
104-117.
GORDON, D.H. & I.L. RAUTENBACH (1980). Species complexes
DUCROZ, J.F., L. GRANJON, M. LOMBARD & V. VOLOBOUEV
in medically important Rodents: chromosome studies of
(1999). Comparative chromosome analysis (R.- and C-
Aethomys, Tatera, and Saccostomus (Rodentia: Muridae,
Bands) of two South African murid species, Lemniscomys
Cricetidae). S. Afr. J. Sc., 76: 559-561.
rosalia and Rhabdomys pumilio (Rodentia, Murinae).
GORDON, D.H. & C.R.B. WATSON (1986). Identification of cryp-
Cytogenet. Cell Genet., 87: 69­74.
tic specie of rodents (Mastomys, Aethomys, Saccostomus) in
DUCROZ, J.F., V. VOLOBOUEV & L. GRANJON (2001). An Assess-
the Kruger National Park. South. Afr. J. Sci., 21: 95-99.
ment of the Systematics of Arvicanthine Rodents using
GRANJON, L., J.M. DUPLANTIER, J. CATALAN & J. BRITTON-DAV-
Mitochondrial DNA sequences: Evolutionary and biogeo-
IDIAN (1992). Karyotypic data on Rodents from Senegal. Isr.
graphical implications. J. Mammal. Evol., 8: 173-206.
J. Zool., 38: 106-111.
DUPLANTIER, J.M., J. BRITTON-DAVIDIAN & L. GRANJON (1990).
GRANJON, L., J.M. DUPLANTIER, J. CATALAN, J. BRITTON-DAVID-
Chromosomal characterization of three species of the genus
IAN & G.N. BRONNER (1996). Conspecificity of Mastomys
Mastomys in Senegal. Zeit. Zool. Syst. Evolut. forschung, 28:
natalensis (Rodentia: Muridae) from Senegal and South
289-298.
Africa: evidence from experimental crosses, karyology and
ELLERMAN, J.R. (1941). The families and genera of living
biometry. Mammalia, 60: 697-706.
rodents. Vol. II. Family Muridae. British Museum (Natural
GRANJON, L., J.M. DUPLANTIER, J. CATALAN & J. BRITTON-DAV-
History), London.
IDIAN (1997). Systematics of the genus Mastomys (Thomas,
E
1915) (Rodentia: Muridae): a review. Belgian J. Zool., 127:
LLERMAN, J.R., T.C.S. MORRISON-SCOTT & R.W. HAYMAN
(1953). Southern African mammals 1758 to 1951: a reclassi-
7-18.
fication. London: British Museum (N.H.).
GRANJON, L. & G. DOBIGNY (2003). The importance of chromo-
somally-based identifications for correct understanding of
FADDA, C. & M. CORTI (1998). Geographic variation of Arvi-
African rodent zoogeography: Lake Chad murids as an
canthis (Rodentia, Muridae) in the Nile Valley. Z. Säuge-
example. Mamm. Rev., 33: 77-91.
tierkunde, 63: 104-113.
GREEN, C.A., H. KEOGH, D.H. GORDON, M. PINTO & E.K.
FADDA, C. & M. CORTI (2001). Geographic variation in Arvican-
HARTWIG (1980). The distribution, identification, and nam-
this (Rodentia, Muridae) in the Nile Valley: a three dimen-
ing of the Mastomys natalensis species complex in southern
sional geometric morphometrics evaluation. Z. Säuge-
Africa (Rodentia: Muridae). J. Zool., London, 192: 17-23.
tierkunde , 63: 104-113.
HALLET, J.M. (1979). Chromosome polymorphism in Praomys
FADDA, C., M. CORTI & E. VERHEYEN (2001). Molecular phylog-
(Mastomys) natalensis in Southern Africa: diploid studies.
eny of Myomys/Stenocephalemys complex and its relation-
South. Afr. J. Sci., 75: 413-415.
ships with related African genera. Biochem. Syst. Ecol., 29:
HOLDEN, M.E. (1993). Family Myoxidae. In: WILSON, D.E. &
585­596.
D.M. REEDER D (eds), Mammal species of the world. A taxo-
FADDA, C., R. CASTIGLIA, P. COLANGELO, R. MACHANG'U, R.
nomic and geographic reference. Smithsonian Institution
MAKUNDI, A. SCANZANI, P. TESHA, W. VERHEYEN & E.
Press, Washington D.C.: 763-770.
CAPANNA (2001). The Rodent fauna of Tanzania 1: a cyto-
HOLDEN, M.E. (1996). Systematic revision of Sub-Saharan Afri-
taxonomic report from the Maasai Steppe (1999). Rend. Fis.
can dormice (Rodentia: Myoxidae: Graphiurus) Part 1: an
Acc. Lincei, s. 9, v12: 5-18.
introduction to the generic revision, and a revision of
FILIPPUCCI, M.G., M.V. CIVITELLI & E. CAPANNA (1986). The
Graphiurus surdus. Am. Mus. Novitates, 3157: 1-44.
chromosomes of Lemniscomys barbarus (Rodentia, Muri-
HUBERT, B. (1978). Revision of the genus Saccostomus (Roden-
dae). Boll. Zool., 53: 355­358.
tia, Cricetomyinae), with new morphological and chromo-

Cytotaxonomy of East Africa Rodents
215
somal data from specimens from the lower Omo Valley,
MATTHEY, R. (1964). Evolution chromosomique et speciation
Ethiopia. Bull. Carnegie Mus. Nat. Hist., 6: 48-52.
chez les Mus du sous-genre Leggada Grey 1837. Experien-
HUBERT, B., F. ADAM & A. POULET (1973). Liste préliminaire
tia, 20: 1-9.
des Rongeurs du Senegal. Mammalia, 37: 76-87.
MATTHEY, R. (1965a). Le problème de la détermination du sexe
HUBERT, B., F. MEYLAN, F. PETTER, A. POULET & M. TRANIER
chez Acomys selousi De Winton ­ Cytogénétique du genre
(1983). Different species in genus Mastomys from Western,
Acomys (Rodentia-Murinae). Rev. Suisse Zool., 72: 119­144.
Central and Southern Africa (Rodentia, Muridae). Ann. Mus.
MATTHEY, R. (1965b). Etudes de cytogènètique sur des Murinae
Roy. Afr. Centr., Sc. Zool., 237: 143.148.
Africains appartenant aux genres Arvicanthis, Praomys,
H
Acomys et Mastomys (Rodentia). Mammalia, 29: 228-249.
SU, T.C. & J.L. PATTON (1969). Bone marrow preparations for
chromosome studies. In: B
M
ENIRSCHKE, K. (ed), Comparative
ATTHEY, R. (1966a). Cytogénétique et taxonomie des rats
Mammalian Cytogenetics. Springer-Verlag, Berlin: 454-460.
appartenant au sous-genre Mastomys Thomas (Rodentia ­
J
Muridae). Mammalia, 30: 105-119.
OTTERAND, M. (1972). Le polymorphisme chromosomique des
Mus (Leggadas) africains. Cytogenetique, zoogeographie,
MATTHEY, R. (1966b). Une inversion péricentrique à l'origine
evolution. Rev. Suisse Zool., 79: 287-359.
d'un polymorphisme chromosomique non-Robertsonienne
dans une population de Mastomys (Rodentia ­ Murinae).
JOTTERAND-BELLOMO, M. (1986). Le genre Mus africain, un
Chromosoma, 18: 188-200.
exemple d'homogeneité caryotypique: etude cytogenetique
M
de Mus minutoides/musculoides (Cote d'Ivoire) de M. setu-
ATTHEY, R. (1967). Note sur le cytogénétique de quelques
muridés africains. Mammalia, 31: 281-287.
losus (Republique Centrafricaine), et de M. mattheyi (Bur-
kina Faso). Cytogenet. Cell Genet., 42: 99­104.
MATTHEY, R. (1968) Cytogénétique et Taxonomie du genre Aco-
mys ­ A. percivali Dollmann et A. wilsoni Thomas, espèces
JOTTERAND-BELLOMO, M. (1988). Chromosome analysis of five
d'Abyssinie. Mammalia, 32: 621­627.
specimens of Mus bufo-triton (Muridae) from Burundi
MATTHEY, R. (1969). Chromosomes de Gerbillinae. Genres Tat-
(Africa): three cytogenetic entities, a special type of chromo-
era et Taterillus. Mammalia, 33: 522-528.
somal sex determination, taxonomy, and phylogeny.
M
Cytogenet. Cell Genet., 48: 88­91.
ATTHEY, R. (1971). Dimorphisme sexuel du chromosome X et
caryotype de Grammomys surdaster Th. et Wrough. (Mam-
KING, M. (1993). Species evolution, the role of chromosome
malia: Muridae). Boll. Zool., 38: 183-186.
change. Cambridge: Cambridge University Press.
MATTHEY, R. & F. PETTER (1970). Etude cytogénétique et tax-
KRAL, B. (1971). Non-Robertsonian variability of karyotype in
onomique de 40 Tatera et Taterillus provenant de Haute-
rats of the subgenus Mastomys. Zool. Listy, 20: 39-40.
Volta et de République Centrafricaine (Rongeurs, Gerbilli-
LAVRENCHENKO, L.A., O.P. LIKHNOVA, M.I. BASKEVICH & AFE-
dae). Mammalia, 34: 585-597.
WORK BEKELE (1998). Systematics and distribution of Mas-
MEESTER, J. & H.W. SETZER (1971). The mammals of Africa, an
tomys (Muridae, Rodentia) from Ethiopia, with the descrip-
identification manual. Smithsonian Institution Press, Wash-
tion of a new species. Z. Säugetierkunde, 63: 37-51.
ington D.C.
LAVRENCHENKO, L.A., A.N MILISHNIKOV, V.M. ANISKIN & A.A.
MENAUT, C. (1983). The vegetation of African savannas. In:
WARSHAVSKY (1999). Systematics and phylogeny of the
BOULIÈRE, J. (ed), Ecosystems of the world. Elsevier,
genus Stenocephalemys Frick, 1914 (Rodentia, Muridae): a
Amsterdam: 109-149.
multidisciplinary approach. Mammalia, 64, 475­494.
MISONNE, X. (1969). African and Indo-Australian Muridae:
LECOMPTE, E., L. GRANJON, J.K. PETERHANS & C. DENYS (2002).
Evolutionary trends. Ann. Mus. Roy. Afr. Cent., Tervuren,
Cytochrome b-based phylogeny of the Praomys group
Serie IN-8, Sci. Zool.: 172: 1-219.
(Rodentia, Murinae): a new African radiation? C.R. Biolo-
MISONNE, X. (1974). Order Rodentia. Part 6. In: MEESTER J. &
gies, 325: 827-840.
H.W. SETZER (eds), The mammals of Africa: and identifica-
LEE, M.R. & L.K. MARTIN (1980). Mastomys (=Praomys)
tion manual. Washington, D.C. Smithsonian Institution
natalensis is not a Rattus (Mammalia: Rodentia): karyologi-
Press: 1-39.
cal evidence. Cytogenet. Cell Genet., 28: 95-103.
MONTGELARD, C., C.A. MATTHEE & TJ. ROBINSON (2003).
L
Molecular systematics of dormice (Rodentia: Gliridae) and
EIRS, H. (1994). Population ecology of Mastomys natalensis
(Smith, 1834). Implications for rodent control in Africa. Bel-
the radiation of Graphiurus in Africa. Proc. R. Soc. Lond.,
gian administration for Development Cooperation, Agricul-
270: 1947­1955.
tural Ed. n° 35, Brussels.
MUSSER, G.G. & M.D. CARLETON (1993). Family Muridae. In:
L
WILSON, D.E.& D.H. REEDER (eds), Mammal species of the
YONS, N.F., D.H. Gordon & C.A. Green (1980). G-banding
chromosome analysis of species A of the Mastomys natalen-
world. A taxonomic and geographic reference. Smithsonian
sis complex (Smith, 1834) (Rodentia Muridae). Genetica,
Institute Press, Washington D.C.
54: 209-212.
MUSSER, G.G. & M.D. CARLETON (2005). Superfamily Muroi-
dea. In: W
M
ILSON, D.E. &, D.M. REEDER (eds), Mammal spe-
ATTHEY, R. (1954). Nouvelles recherches sur les chromosomes
cies of the world: a taxonomic and geographic reference, 3rd
des Muridae. Caryologia, 6: 1­44.
edition. Baltimore: Johns Hopkins University Press.
MATTHEY, R. (1955). Noveaux documents sur les chromosomes
ORLOV, V.N., N.SH. BULATOVA & A.N. MILISHNIKOV (1989).
des Muridae. Problemès de cytologie comparée et de taxon-
Karyotypes of some mammalian species (Insectivora,
omie chez les Microtinae. Rev. Suisse Zool., 62 : 163-206.
Rodentia) in Ethiopia. In: SOKOLOV V.E. (ed), Ecological
MATTHEY, R. (1956). La formule chromosomique de quelques
and Faunistic studies in South-Western Ethiopia. USSR
Murinae (Muridae-Rodentia-Mammalia). Archiv. der Julius
Committee for the UNESCO Programme of Man and Bio-
Klaus-Sitftung fur Vererbungsforschung. Sozialanthr.
sphere (MAB) and Institute of Evolutionary Morphology
Rassen., 31: 294-306.
and Animal Ecology of the USSR Academy of Science,
MATTHEY, R. (1959). Formules chromosomiques de Muridae et
Moscow: 72-94 (in Russian).
de Spalacidae. La question du polymorphisme chromo-
ORLOV, V.N., M.I. BASCKEVICH & N.SH. BULATOVA (1992).
somique chez les mammifères. Rev. Suisse Zool., 66: 175­
Chromosomal sets of rats of the genus Arvicanthis from
209.
Ethiopia. Zool. Zh., 71: 103-112.
MATTHEY, R. (1963). Polymorfisme chromosomique intraspèci-
PAVLINOV, I.J. (1999). Current concepts of Gerbillid phylogeny
fique et intraindividuel chez Acomys minous Bate (Mamma-
and classification. In: DENYS, C., L. GRANJON & A. POULET
lia-Rodentia-Muridae). Rev. Suisse Zool., 72: 119-144.
(eds), African Small Mammals. Proceedings of the 8th Inter-

216
Corti, Castiglia, Colangelo, Capanna, Beolchini, Bekele, Oguge, Makundi, Sichilima, Leirs, Verheyen and Verhagen
national Symposium on African Small Mammals. IRD Edi-
The phylogeny and identification of African thicket rat spe-
tions, Paris: 141­149.
cies (Muridae, Grammomys, Thamnomys) based on 16S
PHILIPPI, T. (1994). Zur innerartlichhen Variabilität von Arvican-
mtDNA sequences. African Small Mammals, 9th Interna-
this niloticus Desmarest 1822 in Ägypten und im Sudan
tional Symposium, Morogoro, Tanzania, 14-18 July: 76.
(Mammalia: Rodentia: Muridae). Sencken. biol., 73: 39-44.
VEYRUNES, F., J. CATALAN, B. SICARD, T.J. ROBINSON, J.M.
QUMSIYEH, M.B. (1986). Phylogenetic studies of the rodent fam-
DUPLANTIER, L. GRANJON, G. DOBIGNY & J. BRITTON-DAVID-
ily Gerbillidae: chromosomal evolution in the southern Afri-
IAN (2004). Autosome and sex chromosome diversity among
can complex. J. Mamm., 67: 680­692.
the African pygmy mice, subgenus Nannomys (Muridae;
QUMSIYEH, M.B., M.J. HAMILTON & D.A. SCHLITTER (1987).
Mus). Chromosome Res., 12: 369­382.
Problems in using Robertsonian rearrangements in determin-
VEYRUNES, F., J. BRITTON-DAVIDIAN, T.J. ROBINSON, E. CALVET,
ing monophyly: example from the genera Tatera and Gerbil-
C. DENYS & P. CHEVRET (2005). Molecular phylogeny of the
lurus. Cytogenet. Cell Genet., 44: 198­208.
African pygmy mice, subgenus Nannomys (Rodentia, Muri-
QUMSIYEH, M.B., S.W. KING, J. ARROYO-CABRALES, I.R.
nae, Mus): Implications for chromosomal evolution. Mol
AGGUNDEY, D.A. SCHLITTER, R.J. BAKER & K.J. MORROW JR.
Phylogenet. Evol., 36: 358-369.
(1990). Chromosomal and protein evolution in morphologi-
VIEGAS-PÉQUIGNOT, E., M. RICOUL, F. PETTER & B. DUTRILLAUX
cally similar species of Praomys sensu lato (Rodentia, Muri-
(1987). Cytogenetic study of three forms of Mastomys
dae). J. Hered., 81: 58-65.
(Rodentia, Muridae). Bras. J. Genet., 10: 221-228.
ROCHE, J., E. CAPANNA, M.V. CIVITELLI & A. CERASO (1984).
VISSER, D.S. & T.J. ROBINSON (1986). Cytosystematics of the
Caryotypes des rongeurs de Somalie. 4. Premiere capture de
South African Aethomys (Rodentia: Muridae). S. Afr. J.
rongeurs arboricoles du sous-genre Grammomys (genre
Zool., 21: 264-68.
Thamnomys, murides) en Republique de Somalie. Mon.
VOLOBOUEV, V., E. VIEGAS-PÉQUIGNOT, F. PETTER & B. DUTRIL-
Zool. It., 19: 259-277.
LAUX (1987). Karyotypic diversity and taxonomic problems
RUSSO, I.M., C.T. CHIMIMBA & P. BLOOMER (2001). Geographic
in the genus Arvicanthis. Genetica, 72: 147-150.
mtDNA variation in Aethomys namaquensis (Rodentia:
VOLOBOUEV, V., E. VIEGAS-PÉQUIGNOT, M. LOMBARD, F. PETTER,
Muridae). Proceedings of 8th International Theriological
J.M. DUPLANTIER & B. DUTRILLAUX (1988). Chromosomal
Congress, sun City, South Africa, 12-17 August, 2001.
evidence for a polytypic structure of Arvicanthis niloticus
SCHLITTER, D.A., L.W. ROBBINS & S.L. WILLIAMS (1985). Taxo-
(Rodentia, Muridae). J. Zool. Syst. Evol. Res., 26: 276-285.
nomic status of dormice (genus Graphiurus) from west and
V
central Africa. Ann. Carn. Mus., 54: 1-9.
OLOBOUEV, V.T., M. TRANIER & B. DUTRILLAUX (1991). Chro-
mosome evolution in the genus Acomys (Rodentia, Muri-
SINGLETON, G.R., L.A. HINDS, H. LEIRS & Z. ZHANG (1999).
dae): chromosome banding analysis of Acomys cf. dimidia-
Ecologically-based management of rodent pest. ACIAR
tus. Bonn. Zool. Beitr., 42: 253-260.
Monograph No. 59, Canberra.
S
VOLOBOUEV, V.T., J.-C. GAUTUN & M. TRANIER (1996). Chromo-
TAPLERAT, (2003). Protecting staple crops in eastern Africa:
integrated approaches for ecologically based field rodent
some evolution in the genus Acomys (Rodentia, Muridae):
pest management. Final report. Contract Number ICA4-CT-
chromosome banding analysis of Acomys cahirinus. Mam-
2000-30029, an European Union Shared-Cost Project.
malia, 60: 217-222.
S
VOLOBOUEV, V.T., A. HOFFMANN, B. SICARD & L. GRANJON
TITOU, S., M. BURGOS, F. ZURITA, R. JIMENEZ, A. SANCHEZ & R.
D
(2001). Polymorphism and polytypy for pericentric inver-
IAZ DE LA GUARDIA (1997). Recent evolution of NOR-bear-
ing and sex chromosomes of the North African rodent Lem-
sions in 38-chromsomae Mastomys (Rodentia, Murinae) and
niscomys barbarus. Chromosome Res., 5: 481­485.
possible taxonomic implications. Cytogenet. Cell Genet., 92:
T
237-242.
RANIER, M. (1974). Parenté des Mastomys du Maroc e du Séné-
gal (Rongeurs, Muridés). Mammalia, 38: 558-560.
VOLOBOUEV, V.T., J.-F. DUCROZ, V.M. ANISKIN, J. BRITTON-DAV-
TRANIER, M. & H. DOSSO (1979). Recherches caryotypiques sur
IDIAN, R. CASTIGLIA, G. DOBIGNY, L. GRANJON, M. LOMBARD,
les rongeurs de Cote d'Ivoire: résultats préliminaires pour
M. CORTI, B., SICARD & E. CAPANNA (2002a). Chromosomal
les milieux fermés. Mammalia, 43: 254-256.
characterization of Arvicanthis species (Rodentia, Murinae)
T
from western and central Africa: Implications for taxonomy.
RANIER, M. & J.C. GAUTUN (1979). Recherches caryologiques
sur les rongeurs de Cote d'Ivoire: resultats preliminaires
Cytogenet. Gen. Res., 96: 250­260.
pour le milieu ouvert. Le cas d'Oenomys hypoxanthus orna-
VOLOBOUEV, V.T., V.M. ANISKIN, E. LECOMPTE & J.-F. DUCROZ
tus. Mammalia, 43: 252­254.
(2002b). Patterns of karyotype evolution in complexes of
VAN DER STRAETEN, E. & W.N. VERHEYEN (1978). Karyological
sibling species within three genera of African muroid
and morphological comparison of Lemniscomys striatus
rodents inferred from comparison of cytogenetic and molec-
(Linnaeus, 1758) and Lemniscomys bellieri Van der Straeten,
ular data. Cytogenet. Gen. Res., 96: 261-275.
1975, from Ivory Coast (Mammalia, Muridae). Bull. Carne-
WAHRMAN, J. & A. ZAHAVI (1953). Intra-generic differences in
gie Mus. Nat. Hist., 6: 41­47.
chromosome numbers of spiny-mice (Rodentia, Murinae).
VAN DER STRAETEN, E. & W.N. VERHEYEN (1980). Relations
Bull. Res. Council Israel, 3: 265
biomètriques dans le groupe spécifique Lemniscomys stria-
WHITE, F. (1986). La vegetation de l'Afrique. UNESCO, Paris.
tus Mammalia, Muridae). Mammalia, 44: 73­82.
YALDEN, D.W., M.J. LARGEN & D. KOCK (1976). Catalogue of
VAN DER STRAETEN, E. & W.N. VERHEYEN (1985). Premiere cap-
the mammals of Ethiopia. 2. Insectivora and Rodentia. Mon.
ture de Lemniscomys striatus au Burkina-Faso. Mammalia,
Zool. It., suppl., 8: 1-118.
49: 576­577.
ZIMA, J., M. MACHOLÁN & M.G. FILIPPUCCI (1994). Chromo-
VERHEYEN, E., M. DILLEN, C. FADDA, M. COLYN, L.S.
somal variation and systematics of Myoxids. Hystrix, 6: 63-
MULUNGU, N.O. OGUGE, J. KERBIS & W. VERHEYEN (2003).
76.