bale mountains rodent communities and their relevance to the ethiopian wolf (canis simensis)

Afr. J. Ecol. 1995, Volume 33, pages 301-320

Bale Mountains rodent communities and their relevance to the Ethiopian wolf (Canis simensis)

CLAUD10 SILLERO-ZUBIRI*, FRANCOISE H. TATTERSALL and DAVID W. MACDONALD Wildlije Conservation Research Unit, Zoology Department, South Parks Road, Oxford OX1 3PS, U. K.

Summary Rodent communities in the Bale Mountains, Ethiopia, form the prey base for the endangered Ethiopian wolf (Canis simensis) and therefore a knowledge of their ecology and biomass is an important management tool. The rodent communities are also of intrinsic interest, being composed of a high proportion of little-known endemic species. Data are presented on the habitat preferences, abundance and biomass of rodents in the Afroalpine belt, the ericaceous belt and montane grasslands of the Bale Mountains. Three species - Lophuromys Jlavopunctatus, Stenocephalemys griseicauda and Otomys typus - characterize the montane grasslands. Stenocephalemys albocaudata, Arvicanthis blicki and Lophuromys melanonyx characterize the Afroalpine belt, and data are presented on their activity and population dynamics. All three species for which data were available bred during the wet season. Minimum density estimates ranged between 32-89/ha for L. melanonyx, 3.2-127Iha for A . blicki and 16-60/ha for S. albocaudata. The mean monthly biomass of three diurnal rodent species was estimated at 25.7 kg/ha and 23.8 kg/ha in two Afroalpine regions. Ethiopian wolf abundance correlated positively with an index of diurnal rodent biomass and total biomass, but not with nocturnal rodent biomass. Ecological constraints on Afroalpine rodents, their population ecology and their importance in maintaining the Bale Mountains Ethiopian wolf population are discussed.

Key words: Bale, Ethiopia, Arvicanthis, Lophuromys, Stenocephalemys, Canis

RbumC Les communautes de rongeurs des Bale Mountains, en Ethiopie, constituent les proies de base du menace loup d’Abyssinie (Canis simiensis), c’est pourquoi la connaissance de leur Ccologie est un outil important pour une bonne gestion. Les communautes de rongeurs prksent aussi un intCrGt en soi, car elles sont composees pour une large part d’espbces endimiques peu connues. On prksente des donnies sur les preferences en matiere d’habitat, I’abondance et la biomasse dans l’etage afroalpin, l’etage a ericacees et les prairies de montagne des Bale Mountains. Trois especes (Lophuromys jlavopunctatus, Stenocephalemys griseicauda et Oto- mys typus) caracterisent les prairies de montagnes. Stenocephalemys albocaudata, Arvicanthis blicki et Lophuromys melanonyx caracterisent l’etage afroalpin, et I’on

*Corresponding author

302 C. Sillero-Zubiri et al.

presente des donntes sur leurs activitis et la dynamique de leur population. Les trois especes pour lesquelles on dispose de donnkes se reproduisaient pendant la saison des pluies. Les estimations minimales de densite de situaient entre 32 a 89ha pour L. melanonyx, 3,2 a 127ha pour A. blicki et de 16 st 60/ha pour S. albocaudata. Les biomasses mensuelles moyennes pour trois especes diurnes etaient estimeies A 25,7 kgha et 23,8 kgha dans deux regions afroalpines. L’abondance des loups d’Abyssinie etait directement like a un index caracterisant la biomasse des rongeurs diurnes et la biomasse totale, mais pas avec la biomasse des rongeurs nocturnes. On discute les contraintes Ccologiques qui pesent sur les rongeurs afroalpins, l’tcologie de leur population et leur importance dans le maintien de la population des loups d’Abyssinie dans les Bale Mountains.

Introduction The Bale Mountains National Park (BMNP), in the southeastern Ethiopian Highlands, is one of the last strongholds of the Ethiopian wolf (Canis simensis Ruppell), a specialist rodent predator found only in open Afroalpine habitat above 3000 m a.s.1. (Gottelli & Sillero-Zubiri, 1992; Sillero-Zubiri & Gottelli, 1991, 1994; Sillero-Zubiri, 1994). The rodent communities of the Bale Mountains are of great importance to the wolf; Tachyoryctes macrocephalus Ruppell, Arvicanthis blicki Frick and Lophuromys melanonyx Petter together account for 86.4% of prey occurrences in the wolfs diet (Sillero-Zubiri & Gottelli, 1995a), but little is known of their ecology, or of the ecology of other rodents in the Bale Mountains.

Ethiopia is notable for the extent of its high ground. The country contains 50% of the Afrotropical region’s land above 2000m a d . , and 80% of that region’s land above 3000 m a.s.1. (Yalden, 1983). The Bale Mountains form the largest continuous area above 3000 m a.s.1. in Africa, and are situated within the 2470 km2 BMNP. As a result of its unusual geography, Ethiopia has many endemic plants and animals (Kingdon, 1990; Yalden & Largen, 1992), with 30 species (1 IYo) from a total of 263 mammals listed as endemics.

Yalden & Largen’s (1992) list of endemic mammals includes the endangered Ethiopian wolf and fourteen species of rodents. Four of these endemic rodents ( Tachyoryctes macrocephalus, Stenocephalemys albocaudata Frick, Lophuromys melanonyx and Megadendromus nikolausi Dieterlen & Rupp) are confined to areas above 3000 m a.s.l., and six others (Dendromus lovati De Winton, Arvican- this blicki, Stenocephalemys griseicauda Petter, Praomys albipes (Riippell), Praomys ruppi Van der Straeten & Dieterlen and Mus mahomet Rhoads) extend into areas higher than 3000 m a.s.1. There are at least a further sixteen species of non-endemic rodents, although only four of these (Tachyoryctes splendens Ruppell, Otomys typus Heuglin, Lophuromys Jlavopunctatus Thomas and Arvi- canthis abyssinicus (Ruppell)) are found above 3000 m (Dorst, 1972; Yalden, Largen & Kock, 1976; Hillman, 1986; Yalden, 1988; Yalden & Largen, 1992).

Hitherto our understanding of the structure and basic habitat requirements of the Bale Mountains rodent community was sketchy. For example, two common species, L. melanonyx and S. griseicauda have only recently been described (Petter, 1972), and there has only been one short-term survey of the rodent fauna (Yalden, 1988). This lack of information has restricted our ability to assess Ethiopian wolf prey availability and habitat quality, both factors critical when assessing in situ conservation action, and vital to selection of release areas should

Bale Mts rodents and the Ethiopian wolf 303

there ever be a captive breeding and reintroduction scheme for the wolf (Gottelli & Sillero-Zubiri, 1992; Sillero-Zubiri, 1994). However, the Bale Mountains rodent community is also interesting in its own right as it contains many little studied endemics with restricted distributions.

Here data are presented on the species composition and relative abundance of rodents in various habitats, which complement and extend Yalden's (1 988) data, and allow the calculation of an index of rodent biomass. Rodent activity patterns are also examined to investigate which rodent species are most easily available to the wolves. For the majority of species the transect snap-trapping data are insufficient to examine population dynamics, but grid live-trapping and snap- trapping were used to obtain more detailed data on population dynamics for the three commonest species, A. blicki, L. melanonyx and S. albocaudata. Details of T. macrocephalus ecology and distribution have been published elsewhere (Sillero-Zubiri, Tattersall & Macdonald, 1995a).

Study sites All study sites were within the BMNP, and covered three main vegetation zones, named after Hillman (1986) and Miehe & Miehe (1993, 1994) as: the Afroalpine belt (3400-4200 m a.s.l.), the ericaceous belt (3400-3800 m a.s.1.) and the montane grasslands (3000-3100 m a.s.1.). Annual rainfall varied from 800 to 1150 mm at different altitudes and was characterized by a warm April-October rainy season and a November-March dry season, with warm days and sub-zero night temperatures (Sillero-Zubiri, 1994).

Most of the study sites are in the Afroalpine belt, which covers nearly 700 km2, including the main plateau and peaks of the Bale Mountains. Afroalpine habitats are characterized by short, sparse vegetation, heavy frosts and low rainfall. There were three main study areas within the Afroalpine belt: Web Valley in the west (3450 m a.s.1.; 7'N 39'42'E); Sanetti Plateau in the east (3800-4050 m a.s.1.; 6'52" 39'55'E); and Tullu Deemtu, a drier southern section of Sanetti (3700-4377 m a d . ; 6'50" 39'50'E).

In the Web Valley study sites were situated in three vegetation sub-divisions: Alchemilla pasture (divided into seasonally waterlogged swamp shores, valley floors and rocky ridges); sedge swamps; and rocky, thin-soiled mesas (sub- sampled as short-grass mesa tops, and mesa slopes, which were steep rocky cliffs dominated by dense Artemesia and Helichrysum shrubs). Alchemilla pasture was a short herb community, which included Helichrysum and Artemesia shrubs, kept in an early successional stage by regular livestock grazing. On the Sanetti Plateau trapping was carried out in three vegetation sub-divisions: herbaceous communities dominated by Alchemilla (divided into swamp shores and valley floor); rocky grasslands (either <20% vegetated or >20% vegetated with Helichrysum shrub); and sedge swamps. Tullu Deemtu was a large area of uniform, 10-50cm tall Helichrysum heathland, which was drier than the main plateau and the Web Valley as it was positioned in the rain shadow of Tullu Deemtu peak (4377m), the highest mountain in southern Ethiopia. This area was trapped in two sub-divisions: low plant diversity Helichrysum dwarf-shrub, and drainage lines, where moisture allowed growth of grass and herbs.

The ericaceous belt (6"55'N 39'57'E and 7"03'N 39'45'E) encircled the Afroalpine belt above the tree-line, at 3400-3800 m a d . ; it was uniform Erica


Table 1. Monthly number of trap nights in each main habitat type, summed from 1989 to 1992.

Tullu Montane Web ,Sanetti Deemtu Erica Grassland

January February March April May June July August September October November December

340 440 360 320 220 140 120 440 220 55 1 I10 420

- - - 540 300 - 240 120 60 60 240 I20

120 60

120

- - -

- - -

- - -

- 240 - - - - - -

I20 40 260 20 242 120 330 440

- -

- 120 -

- - -

Total 4341 2270 320 842 480

trimera scrub (<2m tall), kept from climax by periodic fire (Miehe & Miehe, 1994). The montane grasslands (6'52" 39"44'E), represented by BMNP's Northern grasslands in the Gaysay valley and a corridor connecting it to the upper Web Valley, were a mosaic of grassland tracts on seasonally flooded flat ground. Further details of the vegetation and geography of the Bale Mountains are given in Miehe & Miehe (1993), Hillman (1986), Kingdon (1990), Mohr (1971) and Sillero-Zubiri (1994).

Materials and methods Snap- trapping transects Between 1988 and 1992 the relative abundances of rodents in all sub-habitats and both seasons were assessed using transects of twenty metal snap traps (20 cm x 10 cm; Fox Traps, Germany) set 10 m apart. During this period trapping was carried out opportunistically, and as a result there was considerable variation between years, months and habitats in the number of traps set. Summing across years, monthly numbers of trap nights in each main habitat zone are given in Table 1; at least one wet season and one dry season transect was run in each habitat sub-division. Traps were baited with a paste of peanut butter and wheatharley flour, and they were checked in the early morning, at noon and in the evening for three consecutive days.

Relative abundance of each species was assessed as the percentage trap success, that is, the number of animals caught per 100 trap days or nights. Neither snap-trapping nor live-trapping (see below) caught the large (370-930 g), fossorial T. rnacrocephalus, which is, nonetheless, an important member of the Bale Mountains rodent community. Details of its ecology are given in Yalden (1975, 1985) and Sillero-Zubiri et al. (1995a).


Rodents caught in snap traps were sexed, weighed and identified with refer- ence to a vast collection of specimens available from the BMNP Museum, identified by Hillman (1986) and Yalden (1988). Reproductive condition (signs of lactation, pregnancy or hymen perforation; testes size and position) was recorded for all rodents caught in snap traps. The proportion of captured adult females which were pregnant was assessed by determining pregnancy using macroscopic detection of foetuses in the uterus (Hanny, 1964); this method overlooks foetuses in the first third of pregnancy and therefore underestimates total reproductive output. Mean numbers of implanted foetuses were calculated by dissection of sub-samples of female L. melanonyx, A. blicki and S. albocaudata. All carcasses obtained from snap traps were utilized to bait wolf traps and for food preference trials (Sillero-Zubiri, 1994; Sillero-Zubiri & Gottelli, 1995a), and for morphometric analyses (Sillero-Zubiri, Tattersall & Macdonald 1995b).

Live-trapping Densities of A. blicki and L. melanonyx, at a site of Alchemilla pasture in Web, and a herbaceous community site at Sanetti, were estimated using a 0.16 ha grid of 50 collapsible aluminium Sherman traps (Sherman, Tallahassee, FL, U.S.A.) set at 10 metre intervals, with two traps per point. A. blicki and L. melanonyx live together in mixed colonies (Yalden, 1988) and were both present at all trapping sites. Traps were pre-baited for two days with peanut buttedflour bait and then set at dawn and cleared at noon and dusk for three consecutive days. At Web live-trapping occurred in November 1989, January, March and June 1990, and then monthly from October 1990 until March 1992, with the exception of December 1991. At Sanetti live- trapping took place monthly from November 1989 to February 1990, from October 1990 until February 1991, and during May 1991, October and December 1991, and January and March 1992. Animals were marked by fur clipping.

Population sizes on trap grids were estimated by Capture Mark Recapture techniques using numbers of individuals and the Bailey Triple Catch estimate (Begon, 1979). In order to obtain an estimate of absolute density for L. melanonyx and A. blicki we added a boundary strip equivalent to half the average distance moved between traps by individuals captured on consecutive nights. Half of the average distance moved by consecutively captured L. melanonyx, using data from both sites combined, was 10.8 m, giving a trap area of 0.38 ha. On average, half the nightly distance moved by A. blicki was 8 m, giving an effective trap area of 0.314 ha.

Snap- trapping grids Density estimates for S. albocaudata were obtained using five snap-trapping grids at Sanetti (herbaceous community) and eight at Web (Alchemilla pasture). Each grid was trapped only once. At both Sanetti and Web two grids were trapped during December 1989, and one grid was trapped in each of November 1989 and January, February and March 1990; in addition at Web a grid was trapped during April and June 1990. Each grid covered 0-16 ha, consisting of 50 snap traps with two traps at points 10 m apart. Densities were estimated as number of individuals caught. The absolute trapping area for S. albocaudata was estimated by adding an arbitrary 5 m boundary strip, giving an area of 0.25 ha.


Activity Between 1989 and 1991, at valley floor and swamp shore at Web, and herbaceous communities at Sanetti, twelve 50 x 50 m plots containing A . blicki and L. melanonyx were observed throughout the day for up to seven days each, totalling seventeen observation days during the dry season and ten during the wet season. The plot was scanned with binoculars every 10 minutes throughout the day, and the number of individuals present on the surface was counted.

Development of predictors of Ethiopian worf density Absolute biomass (kgha) provided by the three major Ethiopian wolf prey species (L. melanonyx, A. blicki and T. macrocephalus; Sillero-Zubiri & Gottelli, 1995a) was estimated monthly for valley floor at Web, and herbaceous communities at Sanetti by multiplying the mean weight of snap-trapped animals by an estimate of species density. For L. melanonyx and A. blicki density was estimated as the mean monthly number of individuals live-trapped per hectare; and for T. macrocephalus it was the mean minimum number of individuals observed per hectare, using data obtained from Sillero-Zubiri et al. (1995a).

A biomass index, incorporating all snap-trapped species and weighted for sub-habitats, was calculated by multiplying transect snap-trap success by the mean weight for each species (Sillero-Zubiri, et al., 1995b); this effectively gave an estimate of the biomass contributed per 100 snap-traps per night. This biomass index is, of necessity, crude, and is intended only for general comparisons of habitat zones relative to each other.

Estimates of Ethiopian wolf densities were obtained from Gottelli & Sillero- Zubiri (1995b) and Sillero-Zubiri (1994). Densities were estimated by direct observation of nineteen packs across the BMNP during four years (Sillero-Zubiri & Gottelli, 1994), and by road counts in all main habitat types, following standard sampling methodology (e.g. Hirst, 1969; Norton-Griffiths, 1978). Eighty-four transects, of between 20 and 40 points at 25 m intervals, were surveyed in the late afternoon to assess the numbers of rodent burrows, rodent alarm calls and wolf signs (droppings and signs of wolf diggings) within a 5 m radius of each transect point.

Results Habitat selection Almost all of the rodents caught in the snap-trap transects belonged to one of six species: Lophuromys melanonyx (n = 10 1 1 ), Stenocephalemys albocaudata (n = 994), Arvicanthis blicki (n= 680), Lophuromys Javopunctatus (n= 203), Steno- cephalemys griseicauda (n= 160) and Otomys typus (n=25). In addition ten Dendromus lovati were trapped, five at Tullu Deemtu and five on Sanetti Plateau valley floor.

For all species, trap success in the Web Valley was lowest in sedge swamp areas (Table 2) where only A. blicki and S. albocaudata were occasionally trapped. In the Web Valley A. blicki and L. melanonyx were commonest on valley floors and low ridges, and least abundant in sedge swamp and mesa tops.


Table 2. Percentage trap success for rodents within Web Valley sub-habitats. Percent total area refers to the percentage of the total area covered by a sub-habitat within Web Valley. Percentage trap success for SA and SG was calculated using night-time trapping effort; percentage trap success for the remaining species was calculated using daytime trapping effort

Trap success (Yn) Trap effort YO Total

Habitat daylnight area AB LM LF SA SG OT

Alchemilla pasture Valley floor 201 111951 45 16.7 22.4 0.2 14.9 0.05 0

Swamp shore 4201420 6 4.8 12.4 1.4 6.2 0.24 0 Low ridges 3601360 15 15.6 17.5 0 13.3 0 0

Slope 7401740 2 4.9 11.5 10.1 13.9 6.8 1.1 Mesas

TOP 5 1915 19 20 1.4 1.5 0.8 18.1 0.8 0 Sedge swamp 3401340 12 0.9 0 0 2.4 0 0

~~ ~ ~ ~~

AB, Arvicanthis blicki; LM, Lophuromys melanonyx; LF, Lophuromys flavopunctatus; SA, Steno- cephalemys albocaudata; SG, Stenocephalemys griseicauda; OT, Otomys typus.

S. albocaudata, in contrast, was most abundant on mesa tops, although mesa slopes, ridges and valley floor were also favoured. L. Jlavopunctatus, S. griseicauda and 0. typus were all most commonly trapped on mesa slopes, and were rarely-or never, in the case of 0. typus-found in other Web Valley sub-habitats.

Only A. blicki, L. melanonyx and S. albocaudata were trapped at Sanetti. Overall trap success was greatest in valley floor, which was the preferred habitat for A. blicki and L. melanonyx (Table 3). S. albocaudata was the commonest species in rocky grassland (its most preferred habitat) and sedge swamp. As at Web, trap success was lowest in sedge swamp, where only S. albocaudata was found.

Of the five main areas studied, overall trap success, weighted for the relative contribution made by each sub-habitat, was highest at Sanetti (434%), Web (38.8%) and the montane grasslands (33.5%), and lowest in the ericaceous belt (21 *6%) and at Tullu Deemtu (13.5%). In montane grassland L. Jlavopunctatus and S. griseicauda were the most frequently trapped species, and 0. typus was the least frequently trapped, although montane grassland was its preferred habitat (Table 4). In the three Afroalpine belt zones (Web, Sanetti and Tullu Deemtu) A. blicki, L. melanonyx and S. albocaudata were the three commonest species, while L. flavopunctatus, S. griseicauda and 0. typus were poorly represented, or not present at all (Table 4). All six rodent species were present in the ericaceous belt, with S. albocaudata the most common and L. melanonyx the least commonly trapped.

Activity Three species, D. lovati, S. albocaudata and S. griseicauda were nocturnal (Table 5) , while the remainder were caught predominantly during the morning. L. Jlavopunctatus showed the most widespread activity: 29Y0 of captures occurred during the afternoon and 21% during the night (Table 5 ) .


Table 3. Percentage trap success for rodents within Sanetti Plateau sub-habitats. Percent total area refers to the percentage of the total area covered by a sub-habitat within Sanetti Plateau. Percent- age trap success for SA and SG was calculated using night-time trapping effort; percentage trap success for the remaining species was calculated using daytime trapping effort

Trap effort YO Total Habitat daylnight area

Trap success (YO)

AB LM SA

Herbaceous community Valley floor 7901740 65.0 Swamp shore 2601260 2.5

>20?40 Vegetated 731/731 15.0 <200/0 Vegetated 4401440 12.5

Sedge swamp 80180 4.0

Rocky grassland

12.4 21.6 17.9 3.1 7.7 8.1

9.3 12.1 18.9 2.7 1.6 8.2 0 0 1.3

AB, Arvicunthis hlicki; LM, Lophuromys melanonyx; SA, Sieno- cephulemys ulbocuuduiu.

Table 4. Percentage trap success from snap-trapping transects in five main vegetation zones, weighted for sub-habitat area. Percentage trap success for SA and SG was calculated using night-time trapping effort; percentage trap success for the remaining species was calculated using daytime trapping effort

Trap success (“h)

Habitat Altitude daylnight AB LM LF SA SG OT Trap effort

Afroalpine belt Tullu Deemtu 3700-4377 3201320 2.4 0.8 0 10.2 0.1 0 Sanetti Plateau 38004000 227112251 10.0 17.5 0 16.0 0 0 Web Valley 345g3500 4270/4210 10.6 14.0 0.5 13.3 0.3 0.02

Ericaceo us be1 t 34W3800 842/842 0.8 0.3 6.3 7.9 5.0 1.3 Montane grassland 300&3 100 480/480 3.7 4.2 14.3 1.9 7.9 1.5

AB, Arvicunrhis hlicki; LM, Lophuromys melanonyx; LF, Lophuromys jluvopunciuius; SA, Sieno- cephulemys ulbocuudutu; SG. Sienocephulemys griseicuudu; OT, Oiomys typus.

Observation of A. blicki and L. melanonyx showed that few of these rodents were present above ground before about 0900 h or after 1530 h (Fig. 1). Peak mean numbers (up to 13.5 f 1.6) were observed between about 1030 h and 1200 h. Although highest mean temperatures were recorded at 1300 h (Fig. l), there was a significant positive correlation between mean temperature and mean number of rodents observed ( ~ 0 . 7 8 , n=30, P<O*OOl).

Population dynamics of L. melanonyx Live-trapping Similar numbers of individual L. melanonyx were captured during trap sessions at Sanetti and Web (Fig. 2): the range per session at Sanetti was 15-34 individuals (39-5-89.5ha) with a mean of 23.1 f 6.2 (60Wha), while at Web the range was 12-29 individuals (31.6-76.3ha) with a mean of 20.9 f 5-4

Bale kits rodents and the Ethiopian w o y 309

Table 5. Timing of activity of small mammals expressed as percentage of captures occurring during the morning (0700-1230 h), afternoon (123&1830 h) and at night (183M700 h)

YO captures occurring at

Species captures morning afternoon night No.

A. blicki 584 79.7 20.2 0.1 L. melunonyx 794 76.2 23.3 0.5 L. puvopunctutus 80 50.0 28.8 21.2 0. typus 7 71.4 - 28.6 S. griseicuudu 52 - 1.9 98.1 s. ulbocuudutu 752 0.3 - 99.7 D. lovati 10 - - 100.0

-s- Temperature --o-- Rodents

I I I

?

I .( I 1 I I 1 I I I I I

, I I

b

I I I

.( I 1 I I 1 I I I I I

- 1 5

- 10

- 5

- 0

m U d 0 0 0 w

w 0

Y 0 a 6

2 a

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 r m m o d ~ m ~ m w r m O O O d d d d d d d d d

Time (h)

Fig. 1. Temperature ('C) and activity pattern of diurnal rodents (A. blicki and L. melanonyx). Rodent activity was measured as mean number of rodents observed at 10 minute intervals on 50 m x 50 m plots.

(55/ha). Bailey Triple Catch estimates for population sizes of L. melanonyx were considerably greater than the number of individuals caught: at Web estimated population sizes ranged from 38.7 f 34-5 to 250 f 235-3, equivalent to 102-658/ ha; at Sanetti estimates ranged from 44.9 f 29 to 228.4 f 147.9, equivalent to 118-601/ha. Standard deviations were large despite the fact that 79% of individuals at Sanetti and 68% at Web were recaptured within 24 hours. No clear seasonal trends were apparent at either site using either Bailey's Triple Catch estimate or numbers of individuals caught. The overall ratio of ma1es:females


Fig. 2. Number of individual L. mehnonyx live-trapped in Sanetti and Web between November 1989 and March 1992. The trapping grid covered 0.16 ha, but estimated actual trapping area was 0.38 ha.

live-trapped did not differ significantly from equality in either Sanetti (178: 146) or Web (209: 188).

Transect snap-trapping There were large variations in the abundance of L. melanonyx in different habitats (Tables 2, 3 and 4), and there were differences in the number and timing of transects trapped in each habitat. The only habitats with sufficiently similar abundances and large enough sample sizes to allow analysis of seasonality were valley floor and low ridge habitats at Web. Monthly trap success data summed from a three year period (Fig. 3) show no obvious seasonal trend.

Although pregnant and lactating females were found throughout the year, the greatest proportion of trapped females in all habitats were pregnant or lactating during the first half of the wet season (Fig. 3), and fewest females showed signs of pregnancy from December to February. For 52 pregnancies, the mean number of foetuses was 1.88 f 0.38. Proportions of juveniles in the population, by contrast, were highest from the late wet season (October) to late dry season (February), and lowest during the wet season (Fig. 3).

Population dynamics of A. blicki Live-trapping The density of A . blicki was higher at Web than at Sanetti. Numbers of individuals caught ranged from 1 to 17 (3-54/ha) with a mean of 6.6f44.5 (21ha) at Sanetti, and from 7 to 40 (22-127Iha) with an average of 16.8 k 7-9 (53*5/ha) at Web (Fig. 4). There was a slight trend at both sites for a decline during the wet season and an increase in abundance during the dry season, although data from the wet season were scanty. Estimated population size (Bailey's Triple Catch) on the Sanetti grid ranged from 0 to 39 f 32.2 (0-124/ha), and on the Web grid estimates were between 11.7 f 10.5 and 75 f 67 (37-239/ha).

Bale Mts rodents and the Ethiopian wolf 31 1

a IA 0 Y 8 a U Y 0 a

1 0 0

8 0

6 0

4 0

2 0

0

-4- p r e g n a n t or l a c t a t i n g f e m a l e s -0- J u v e n i l e s -m- Trap s u c c e s s

1

month Fig. 3. Percentage trap success, percentage of females which were lactating or pregnant, and percentage of juvenile L. rnelanonyx. Percentage trap success was measured for snap-trap transects in valley floor and ridge habitats in the Web valley. Percentage of pregnant or lactating females and percentage of animals caught which were juvenile were measured using data from all snap-trap transects.

Overall, proportions of A. blicki recaptured within 24 hours were higher in Sanetti (68%) than in Web (48*3%), indicating that the Sanetti population was sampled more intensively than the Web population. However, in Sanetti a significantly greater proportion of females (79%) than males (47.4%) was recaptured within 24 hours (x2=5.68, df= 1, P<0.05), introducing further potential bias. Approximately equal numbers of individual male and female A. blicki were captured at Sanetti (3850) and Web (164:169).

Transect snap-trapping As with L. melanonyx, seasonal changes in abundance of A. blicki were analysed using transect snap-trapping data only from valley floor and ridges in Web Valley. A. blicki showed a clear decrease in trap success through the second half of the wet season, an increase through the dry season, and a period of high trap success during the first half of the wet season (Fig. 5 ) . Reproduction in A. blicki, as judged from the proportion of females which were pregnant or lactating (using data from all habitats), was most intense in the mid and late wet season (June to October), and, although reproduction continued throughout the year, it was lowest from January to March (Fig. 5). The mean number of foetuses, from 25 pregnancies, was 3.24 f 0.93. As with L. melanonyx, proportions of juveniles in the population tended to follow an inverse pattern to the percentage of females which were pregnant or lactating. Proportions of juveniles peaked at the beginning of the dry season (November) and declined through the remainder of the dry season and the wet season (Fig. 5 ) .


4 0 ” $ a

Fig. 4. Number of individual A. blicki live-trapped in Sanetti and Web between November 1989 and March 1992. The trapping grid covered 0.16 ha, but actual trapping area was estimated at 0.314 ha.

Population dynamics of S . albocaudata Grid snap-trapping Densities of S. albocaudata were slightly higher on snap-trap grids at Web than at Sanetti, both as numbers of individuals caught and as estimated population. At Web an average of 10.4 individuals (range=4-15) were caught on each grid (equivalent to an average of 42/ha, and a range of 1&60/ha) while at Sanetti an average of 6.6 individuals (range=4-11) were caught (equivalent to an average of 26/ha, and a range of 16-44/ha). Approximately equal numbers of males and female were caught at Sanetti (17:19) and Web (41:42).

Transect snap-trapping As with the previous two species, only snap-trapping data from valley floors and ridges at Web were used for analysis of seasonal changes. No obvious seasonal trends were found for S. albocaudata, although trap success was lowest in May and September (wet season) and highest in January and March (dry season) (Fig. 6). The greatest proportion of females were pregnant in May and June, and the lowest proportion of pregnancies was during the dry season and very early wet season, from December to April (Fig. 6). From 27 pregnancies the mean number of foetuses was 3-59*0.93. Once again there was a trend for highest proportions of juveniles in the population during the dry season, although there was also a peak in August (Fig. 6).

Rodents as predictors of Ethiopian wolf distribution Absolute biomass of Ethiopian wolf prey species, as kilograms per hectare, was estimated monthly in the predominant habitat type at Sanetti (herbaceous communities) and Web (valley floor) for L. melanonyx, A. blicki and


PI 0

U cl 0 U Y 0 a

a

-4- pregnant or l a c t a t i n g f e m a l e 6 --a- J u v e n i l e . -m- Trap mucce66

loo 1 8 0

6 0

4 0

2 0

0

Month Fig. 5. Percentage trap success, percentage of females which were lactating or pregnant, and percentage of juvenile A. blicki. Percentage trap success was measured for snap-trap transects in valley floor and ridge habitats in the Web valley. Percentage of pregnant or lactating females and percentage of animals caught which were juvenile were measured using data from all snap-trap transects.

Tachyoryctes macrocephalus (Table 6). T. macrocephalus contributed most in biomass. Overall, mean monthly biomass from the three species together was estimated at 23.8 kgha for herbaceous communities at Sanetti, and 25.7 kg/ha for valley floor at Web. Highest monthly biomass estimates were obtained during March, April and May at Web, and during March and May at Sanetti.

The biomass index (kilograms per 100 transect snap-trap nights) for snap- trapped rodents was highest on Sanetti Plateau (Table 7), followed, in order, by Web Valley, montane grasslands, the ericaceous belt and Tullu Deemtu. Ethiopian wolf density (Table 7), measured both from observation and road counts, correlated positively with the total biomass index (r=0*959, df= 3, P ~ 0 . 0 1 for observation; ~ 0 . 9 6 5 , df=3, P ~ 0 . 0 1 for road counts) and the biomass index for diurnal species ( ~ 0 . 9 4 , df= 3, P<0.02 for observation; ~ 0 . 9 1 , df=3, P<0*05 for road counts), but not for nocturnal species.

There was also a positive correlation between rodent signs and Ethiopian wolf signs, droppings or diggings, along habitat assessment transects. The percentage of transect points at which wolf signs were found correlated positively with the average number of rodent burrows per transect point ( ~ 0 . 6 6 6 , n=83, P<O.OOl), the percentage of transect points including rodent burrows ( ~ 0 . 6 2 , n=83, P<O-OOl), and the percentage of transect points at which rodent alarm calls were recorded (r=0-585, n=83, P<O.OOl). A similar correlation was found between wolf signs and average fresh giant mole rat signs ( ~ 0 . 3 8 3 , n=83, P<O.OOl; Sillero-Zubiri et al., 1995a).

314 C. Sillero-Zubiri et a].

P r e g n a n t or l a c t a t i n g females -0- J u v e n i l e s -C Trap muccesm

0) m Q LI e Q U Y 0) a

2 0 0

8 0

6 0

4 0

2 0

0

n o n t h Fig. 6. Percentage trap success, percentage of females which were lactating or pregnant, and percentage of juvenile S. ulbocuudutu. Percentage trap success was measured for snap-trap transects in valley floor and ridge habitats in the Web valley. Percentage of pregnant or lactating females and percentage of animals caught which were juvenile were measured using data from all snap-trap transects.

Discussion Ecological constraints on Afroalpine rodents L. melanonyx, S. albocaudata and A. blicki were characteristic of the Afroalpine belt, as was the large T. macrocephalus, which could not be trapped but was censused (Sillero-Zubiri et al., 1995a). Yalden (1988) suggested that 0. typus might be an Afroalpine moorland specialist, but although the species was well represented in the diet of Ethiopian wolves at Sanetti and Tullu Deemtu (Sillero-Zubiri & Gottelli, 1995a) it was never trapped in these zones. L. Javopunctatus and S. griseicauda were characteristic of the montane grasslands of the Gaysay Valley and, along with 0. typus, decreased in abundance in trap returns as altitude increased through the ericaceous and Afroalpine belts. None of these three species was found at Sanetti, although S. griseicauda was occasionally found at Tullu Deemtu suggesting that it was limited by factors other than altitude.

The temperature extremes experienced in the high altitude regions of the Bale Mountains could place severe limitations on all species of mammal and may well have been a major structural force on the mammal community. In the Bale Mountains thermo-regulatory constraints imposed by soil depth are thought to limit the distribution of the giant molerat, T. macrocephalus (Sillero-Zubiri et al., 1995a). At around 50 cm soil depth the temperature remains constant despite large fluctuations in air temperature (Sillero-Zubiri et al., 1995a). Small


Table 6. Estimates of absolute monthly biomass (kgha) of the three main Ethiopian wolf prey species in the predominant sub-habitats of Sanetti Plateau and Web Valley (herbaceous communities and valley floor, respectively). Biomass estimates of A. blicki and L. melunonyx were estimated using the mean monthly number of individuals live-trapped on a grid, and the monthly mean weight of snap-trapped animals. Densities of T. mucrocephalus were estimated using the minimum number of individuals observed on a grid (Sillero-Zubiri et ul., 1995a), and a mean weight of 618 g

Web Valley Sanetti Plateau

Month AB LM TM Total AB LM TM Total

Jan Feb Mar APr May Jun Jul Aug SeP Oct Nov Dec

Mean

7. I 3.9 12.4 4.1 4.3 18.5 5.1 4.8 21.0 4.0 4.5 19.8 6.3 4.9 17.3 5.4 4.6

- 19.8 7.6 4.6 9.9 5.3 6.6 12.4

- 9.9 5.2 5.2 12.8 6.3 5.9 11.7

-

-

-

5.6 4.9 15.0

23.4 3.1 26.9 3.1 30.9 3.7 28.3 - 28.5 12.6 - -

- -

22.0 -

24.3

23.3 2.0 23.9 2.2

-

- -

25.7 4.5

6.0 2.5 4.4 4.4 29.7

8.3 12.4

-

- -

- 9.9 - 9.9 - -

- -

4.4 14.8 5.4 7.4

5.5 12.4

-

37.8

33.3 -

-

-

21.2 15.0

23.8

AB, Arvicunthis blicki; LM, Lophuronijs i i i d i i r i u i i y . ~ : TM, Tuchyoryctes mucrocephalus.

Table 7. Ethiopian wolf density (individualdkm') and biomass index, weighted for sub-habitat area, for diurnal (Arvicunthis blicki, Lophuroniys rnuktnorijs and L. fluvopunctutus) and nocturnal (Stenocephulemys griseicuudu, S. ulbocuudutu and 0tomy.s typus) snap-trapped rodent prey species. The biomass index represents the biomass (kg) contributed per 100 transect snap-trap nights using data from all months. Mean weights used were as follows: Arviccmthis blicki, 126 g; Lophuromys melunonyx, 94 g; L. jluvopunc- tutus, 49 g; Stenocephulemys griseicuudu, 101.5 g; S. ulbocuudutu, 129.5 g; Oiomys typus, 100 g

Web Sanetti Montane Tullu Ericaceous Valley Plateau grassland Deemtu belt

Biomass index diurnal rats nocturnal rats TOTAL

2.68 2.9 1.56 0.38 0.44 1.76 2.07 1.2 1.4 1.66 4.44 4.97 2.76 1.78 2.1

Ethiopian wolf density road counts I .05 1.2 0.1 0.16 0.09 observation I .2 1.2 0.3 0.25 0.1

mammals such as rodents can, therefore, avoid extreme cold by going under- ground, in a way that larger mammals such as ungulates cannot.

Behavioural characteristics may protect rodents from cold. Many Afroalpine rodents show diurnal behaviour: only three were nocturnal out of the eight species recorded. A preference for protective habitats might also play a part in thermoregulation. For example, in the Web Valley L. Jlavopunctatus, a small,


partly nocturnal species, and the nocturnal S. griseicauda, were found mostly on mesa slopes. These areas are dominated by tall (up to 50 cm) Artemesia and Helichrysum shrubs and rocky boulders which may protect the rodents from the more extreme temperatures of the open valleys. S. albocaudata was particularly interesting because it appeared not to use deep soils or diurnal activity to escape cold temperatures. S. albocaudata was the dominant species in the ericaceous belt, Tullu Deemtu, mesas and rocky grassland; these areas all have a mean soil depth of less than 40cm (Sillero-Zubiri et al., 1995a). Improved thermoregulation through its large size (Sillero-Zubiri et al., 1995b), and presumably slow metabolic rate, might enable it to utilize environments which require it to have shallow burrows, either because of shallow soil, or because of the danger of flooding its burrow.

D. lovati, which was the least abundant rodent trapped and was only caught in the Afroalpine belt, was also notable for its apparent lack of adaptations to cold. The species was nocturnal, weighing only 16 g with a tail equal in length to its body (Sillero-Zubiri et al., 1995b), yet was found up to 3900 m a.s.1. at Tullu Deemtu (a record which markedly expands its altitudinal range). Furthermore it has been found nesting underneath a boulder (Yalden & Largen, 1992), and so apparently does not escape cold in deep burrows. Clearly this species is of interest as it must use some other mechanism of coping with cold, such as hibernation, extremely slow metabolic rate, or high energy food. Seven out of ten captures of D. lovati were during the warmer wet season, particularly during June and July, at a time when relatively few trapping transects occurred. The remaining three captures occurred at the end of the dry season. The pattern of captures suggests that D. lovati hibernates or otherwise avoids activity during the coldest months.

As well as environmental constraints, interspecific competition and predation may have been important in determining the structure of the rodent community. Some separation on the basis of habitat has already been shown, and presumably further trophic separation occurs, although the diets of most species are poorly known. Two species which merit special attention are A . blicki and L. melanonyx, which live together. Although both are predominantly herbivores, there is evidence that they concentrate on different sectors of the vegetation, with A . blicki eating more monocotyledonous material than L. mefanonyx; in addition, L. melanonyx eats a small amount of invertebrate material (Yalden, 1988; Yalden & Largen, 1992). Both L. mefanonyx and A. blicki are important in the diet of the Ethiopian wolf, and the adaptation to living together might be a response to predation, whereby the predator is confused by large numbers of prey, or predator detection is improved. High pitched whistling cries are given by colony members when a potential predator approaches (Yalden & Largen, 1992; Sillero-Zubiri, pers. obs.).

Population dynamics of A. blicki, L. melanonyx and S . albocaudata Happold & Happold (1992) predict that because high altitude environments have higher rainfall, longer wet seasons, cooler temperatures and lower productivity than environments at lower altitude, rodent populations at high altitude will be characterised by low numbers, low turnover, small litter size and small seasonal fluctuations. More specifically, Happold & Happold (1989) concluded that on


.African mountains rodents generally reproduce during the wet season and decline in numbers during the dry season, and that maximum population size does not exceed 2 4 times the minimum population size. Some of these predictions are true for the three commonest Bale Mountains rodents.

All three species-A. blicki, L. melanonyx and S. albocaudata-bred during and immediately after the rainy season (April-October), during which time they had two, or, in the case of L. melanonyx, three, reproductive peaks. In the tropics, where rainfall is seasonal, many rodent species, including Lophuromys spp., show a peak in their reproductive activity towards the end of the rainy season (Delany, 1972). Rain is accompanied by vegetation growth which in turn influences reproductive output through diet and cover. In the Bale Mountains the vegetation becomes green soon after the onset of the rainy season, not only influencing fodder quantity, but probably also quality.

In the Simien Mountains, an ecosystem similar to that of the Bale Mountains, with 75% of the annual rainfall falling between June and September, Muller (1977) found that the largest proportion of pregnant females and small S. albocaudata, L. jlavopunctatus and 0. typus occurred during and just after the rainy season, a pattern similar to that found at the BMNP. No pregnancies were reported for January to April, the driest months. Arvicanthis abyssinicus was an exception, showing a different reproductive periodicity to other Simien small and large mammals as it concentrated most reproductive effort during the first half of the dry season (Muller, 1977). This pattern was not found for A. blicki.

Arvicanthis species usually breed during the rainy season, and the length of the breeding season is related to the number of humid months, presumably via plant growth (Fisher, 1991). Although a small number of female A. blicki were reproductively active throughout the year, the reproductive season proper started in the early wet season and ended with the end of the rains. During this time, however, capture rates on snap-traps decreased to a low coinciding with the end of the breeding period, as did the proportion of juveniles. Thus the first peak in reproductive activity seemed to be largely unsuccessful. At the end of the dry season capture rates peaked, but very few females were pregnant. An apparently similar situation occurs in the European species Apodemus sylvaticus as a result of density dependent processes (Wilson, Montgomery & Elwood, 1993).

In general Arvicanthis species are opportunistic breeders, capable of rapid population expansion under the right conditions of diet and vegetative cover (Delany, 1986), both factors which rely on good rainfall (Delany & Monro, 1986). If A. blicki populations were similarly volatile there could be large fluctuations in the prey biomass available to wolves. Fluctuations in A. blicki on live-trapping grids were larger than fluctuations of L. melanonyx, and there was greater variation at Sanetti that at Web. Furthermore, A . blicki was the only species analysed which showed clear seasonal differences in trappability. How- ever, except in May at Sanetti the effect of these fluctuations on overall biomass available to the wolves was negligible in comparison with the effects of variation in T. macrocephalus populations

Contrary to the predictions of Happold & Happold (1992) the rodent population on the Bale Mountains was high in comparison with many other studies at lower altitudes (Delany, 1986; Happold & Happold, 1992), although with the exception of A. blicki maximum population size was never more than


four times the minimum. Mean numbers of foetuses carried by pregnant female A. blicki, L. melanonyx and S. albocaudata were small in comparison to other studies of African rodents (Delany, 1986), particularly for L. melanonyx. Slow reproductive output might explain the apparently large lags between the start of reproductive activity and the appearance of juveniles in the population. There were few juveniles in the trappable populations of S. albocaudata and L. melanonyx, and for both species no juveniles under 28 g were caught, although snap-traps caught other species as small as 11 g.

Rodents as indicators of suitable Ethiopian wolf habitat Ethiopian wolves are diurnal predators, and the diurnal rodent species, with the probable exception of L. jlavopunctatus, compose up to 97% of prey volume in the wolves’ diet (Sillero-Zubiri & Gottelli, 1995a). These rodents are therefore of vital importance in maintaining wolf populations, and an understanding of rodent distribution, abundance and biomass, is essential in planning for the future conservation of the Ethiopian wolf.

The role of the Afroalpine rodent community in limiting the distribution of the Ethiopian wolves can be seen by the relationship between wolf density and diurnal rodent biomass index. Areas with low rodent biomass index-the Helichrysum dwarf-scrub of Tullu Deemtu, and the ericaceous belt-were areas with lowest wolf densities. The remaining three zones, Sanetti, Web and the montane grasslands, had more than twice the biomass available in Tullu Deemtu and the ericaceous belt, and, particularly in Web and Sanetti, had higher densities of Ethiopian wolves. Although our use of varied transects rather than one or two plots precludes any detailed study of population dynamics, the technique means that our biomass indices were estimated using data from a wide area and as such are probably more representative of what was actually available to the wolves than detailed data from a limited study area would have been.

Sillero-Zubiri et al. (1995a) positively correlated wolf density with T. macrocephalus abundance in four areas (Tullu Deemtu, Sanetti, Web and the ericaceous belt), and suggested that the species was vital in determining the presence of the wolf. Because they are roughly six times the weight of any other rodent, hunting T. macrocephalus is likely to be considerably more efficient than hunting a smaller species, and indeed, where they are present, they constitute 37% of the wolves’ diet (Sillero-Zubiri & Gottelli, 1995a). Nonetheless, the positive correlation between wolf abundance and an index of biomass of smaller rodents shows that the giant molerat is not the only rodent important to the wolf. Activity patterns of A. blicki and L. melanonyx have shown that these species are readily available to the wolves during the day.

The Ethiopian wolf is currently the world’s most endangered canid, with viable populations probably remaining only in the Bale Mountains National Park and the Simien Mountains National Park. Gottelli & Sillero-Zubiri (1992) suggest further conservation strategies for the wolf, including captive breeding and reintroduction. Should this take place, it will be necessary to evaluate the suitability of habitats for reintroduction of wolves, and the structure of the rodent community will play an essential part in this assessment. The presence of T. macrocephalus or T. splendens, and the biomass contributed by smaller rodents, should be used as an index of the suitability of a habitat for wolves.

Bale Mts rodents and the Ethiopian wolf 3 19

Simple, inexpensive field techniques, such as transect sampling for T. macrocephalus fresh signs (Sillero-Zubiri et al., 1995a) and abundance of rodent burrows could be used for such an evaluation.

Acknowledgments This study was part of the Bale Mountains Research Project, a long term programme funded since 1983 by NYZS/The Wildlife Conservation Society, under the auspices of the Ethiopian Wildlife Conservation Organisation. We would like to thank Edriss Ebu, who almost single-handedly did most of the trapping. We also thank Eshetu Bayena, Menassie Gashaw, Helen Pedersen and Dada Gottelli for their assistance with trapping, activity observations and running transects. We are grateful to Derek Yalden and Chris Hillman for advice and support. We were also funded by the Peoples’ Trust for Endangered Species under the aegis of the IUCN/SSC Canid Specialist Group.

References BEGON, M. (1979) Investigating Animal Abundance. Edward Arnold, London. DELANY, M.J. (1972) The ecology of small mammals in tropical Africa. Mamm. Rev. 2, 142. DELANY, M.J. (1986) Ecology of small rodents in Africa. Mamm. Rev. 16, 1 4 1 . DELANY, M.J. & MONRO, R.H. (1986) Population dynamics of Arvicanthis niloticus (Rodentia: Muridae)

DORST, J. (1972) Notes sur quelques rongeurs observCs en Ethiopie. Mammalia 36, 182-192. FISHER, M. (1991) A reappraisal of the reproductive ecology of Arvicanthis in Africa. Afr. J. Ecol. 29,

GOTTELLI, D. & SILLERO-ZUBIRI, C. (1992) The Ethiopian wolf-an endangered endemic canid. Oryx 26,

HANNY, P. (1964) The harsh-furred rat in Nyasaland. J. Mammal. 54, 345-358. HAPPOLD, D.C.D. & HAPPOLD, M. (1989) Demography and habitat selection of small mammals on

HAPPOLD, D.C.D. & HAPPOLD, M. (1992) The ecology of three communities of small mammals a t

HILLMAN, J.C. (1986) Bale Mountains National Park: Management Plans. Wildlife Conservation Inter-

HIRST, S.M. (1969) Road-strip census techniques for wild ungulates in African woodlands. J. Wild/.

KINGDON, J. (1990) Island Africa. Collins, London. MIEHE, G. & MIEHE, S. (1993) On the physiognomic and floristic differentiation of ericaceous vegetation

MIEHE, S. & MIEHE, G. (1994) Ericaceous Forests and Heathlands in the Bale Mountains of South Ethiopia.

MOHR, P.A. (1971) The Geology ofEthiopia. 2nd edn. Addis Ababa University Press, Addis Ababa. MCILLER, J.P. (1977) Populationsokologie von Arvicanthis abyssinicus in der Grassteppe des Semien

NORTON-GRIFFITHS, M. (1978) Counting Animals. 2nd edn. Handbook 1. African Wildlife Foundation,

PETTER, F. (1972) Deux rongeurs noveaux d’Ethiopie: Stenocephalemys griseicauda sp. nov. et Lophuromys

SILLERO-ZUBIRI, C. (1994) Behavioural Ecology ofthe Efhiopian WOK Canis simensis. DPhil thesis, Oxford

SILLERO-ZUBIRI, C. & GOTTELLI, D. (1991) Conservation of the endemic Ky Kebero (Canis simensis).

SILLERO-ZUBIRI, C . & GOTTELLI, D. (1994) Canis simensis. Mamm. Species 485, 14.

in Kenya. J. Zool. Land. 209,85-103.

17-27.

205-214.

Zomba Plateau, Malawi. J. Zool. Lond. 219, 581-605.

different altitudes in Malawi, Central Africa. J. Zool. Land. 228, 81-101.

national, New York Zoological Society.

Manage. 33,4@48.

in the Bale Mountains, SE Ethiopia. Opera Botanica 121, 85-1 12.

Ecology and Man’s Impact. Stiftung Walderhaltung in Africa, Hamburg.

Mountains National Park (Athiopien). Z. Suugerierk. 42, 145-1 72.

Nairobi.

melanonyx sp. nov. Mammafia 36, 171-181.

University, Oxford.

Walia 13, 35-46.


SILLERO-ZUBIRI, C. & GOTTELLI, D. (1995a) Diet and feeding behaviour of Ethiopian wolves (Canis simensis). J. Mammal. 76, 531-541.

SILLERO-ZUBIRI, C. & GOTTELLI, D. (l995b) Spatial organization in the Ethiopian wolf Canis simensis: large packs and small stable home ranges. J. Zool. Lond. 237, 65-81.

SILLERO-ZUBIRI, C., TATTERSALL, F.H. & MACWNALD, D.W. (1995a) Habitat selection and daily activity of giant mole-rats (Tachyorycies macrocephalus): significance to the Ethiopian wolf (Canis simensis) in the Afroalpine ecosystem. Biol. Conserv. 72, 77-84.

SILLERO-ZUBIRI, C., TATTERSALL, F.H. & MACWNALD, D.W. (l995b) Morphometrics of endemic rodents from the Bale Mountains, Ethiopia. J. Afi. Ecol. 109, 387-391.

WILSON, W.L., MONTGOMERY, W.I. & ELWOOD, R.W. (1993) Population regulation in the wood mouse Apodemus sylvaricus (L.). Mamm. Rev. 23, 73-92.

YALDEN, D.W. (1975) Some observations on the giant mole-rat Tachyorycres macrocephalus (Riippell 1842) (Mammalia: Rhizomyidae) of Ethiopia. Monir. Zool. Iial. N.S. Suppl. 6 , 275-303.

YALDEN, D.W. (1983) The extent of high ground in Ethiopia compared to the rest of Africa. Siner: Eihiopian J. Sci. 6, 35-39.

YALDEN, D.W. (1985) Tachyorycres macrocephalus. Mamm. Species 237, 1-3. YALDEN, D.W. (1988) Small mammals of the Bale Mountains, Ethiopia. Afr. J. Ecol. 26, 281-294. YALDEN, D.W. & LARGEN, M.J. (1992) The endemic mammals of Ethiopia. Mamm. Rev. 22, 115-150. YALDEN, D.W., LARGEN, M.J. & KOCK, D. (1976) Catalogue of the mammals of Ethiopia. 2. lnsectivora

and Rodentia. Monii. Zool. Iial. N.S. Suppl. 8, 1-1 18.

(Manuscript accepted 31 August 1994)

bale mountains rodent communities and their relevance to the ethiopian wolf (canis simensis)

Documents