Introduction
Savannah monkeys Chlorocebus spp. are the most widely distributed non-human primates in Africa, occurring from Senegal to Ethiopia and southwards to South Africa. Although the taxonomy of savannah monkeys has been the subject of debate (Grubb et al., Reference Grubb, Butynski, Oates, Bearder, Disotell, Groves and Struhsaker2003), Groves (Reference Groves, Wilson and Reeder2005) divides Chlorocebus into six species. The least known and most range-restricted of these is the Bale monkey Chlorocebus djamdjamensis, endemic to the southern Ethiopian highlands.
Other savannah monkeys, such as the grivet Chlorocebus aethiops and vervet monkey Chlorocebus pygerythrus are habitat generalists, inhabiting a wide range of habitat types including savannah, woodland, forest, grassland and riverine and gallery forests (Kingdon, Reference Kingdon1997; Zinner et al., Reference Zinner, Pelaez and Torkler2002; Barrett, Reference Barrett2005). The occurrence of Bale monkeys has been documented in the Bale Mountains National Park, Harenna Forest, Kacha and Rirra areas (Carpaneto & Gippoliti, Reference Carpaneto and Gippoliti1994; Kingdon, Reference Kingdon1997; Butynski et al., in press), and in Odobullu Forest (Butynski et al., in press; Mekonnen et al., Reference Butynski, Atickem, De Jong, Butynski, Kingdon and Kalina2010). In addition, Bale monkeys have been recorded in bamboo forest 23 km north-west of Dodolla and in the Djam-Djam Mountains near Abera, east of Lake Abaya at c. 3,000 m (Carpaneto & Gippoliti, Reference Carpaneto and Gippoliti1994; Butynski et al., in press). However, the distribution of the species has not previously been adequately documented.
The Bale Mountains, which harbour a diverse range of endemic fauna and flora, are part of the Eastern Afromontane Biodiversity Hotspot (Myers et al., Reference Myers, Mittermeier, Mittermeier, Fonseca and Kent2000). Despite this diversity, however, habitat loss and fragmentation are intensifying because of the increasing human population and associated agricultural expansion (Philip et al., 2001; WWF, 2001). The Bale Mountains National Park, which was established in 1970 to conserve the endemic mammals of the region, currently provides limited protection because of lack of law enforcement. Studies of the distribution patterns and habitat preferences of wildlife in this human-modified landscape are thus crucial to focus conservation efforts in the region.
The Bale monkey is one of Africa’s least known primates. At the time of our study it was categorized as Data Deficient on the IUCN Red List but has now been categorized as Vulnerable (Butynski et al., Reference Butynski, Gippoliti, Kingdon and De Jong2008). The combination of the species’ thickly forested mountainous habitat, a tendency to flee rapidly upon encountering humans, and quiet behaviour may explain why no detailed field study of the species has previously been undertaken.
The objectives of the study described here were to determine the population size and habitat preferences of the Bale monkey in one of its known strongholds, Odobullu Forest, and to conduct surveys across the Bale Mountains to document further the species’ distribution.
Study area
The study of habitat preferences and population size was carried out in the 71 km2 Odobullu Forest, south-eastern Ethiopia (Fig. 1). Odobullu Forest is controlled by a private organization, Rift Valley Hunting Safaris. The area is protected from livestock grazing and tree cutting but bamboo harvesting is allowed. Hunting of wildlife is allowed based on quota-managed trophy hunting but hunting of the Bale monkey is not permitted. The topography of the area is mountainous, with cliffs and valleys intersected by streams at altitudes of 2,250–3,022 m. The broader distribution survey was conducted across the Bale Mountains in an area of 5,551 km2 over altitudes of 2,400–3,250 m, in the bamboo forest (Fig. 1).
The study area in the Bale Mountains, with the five areas of bamboo forest in which we located the Bale monkey Chlorocebus djamdjamensis, and known historical records of the species.
Methods
Habitat classification and vegetation mapping
We classified the vegetation of Odobullu Forest into four types (tree-dominated forest, bamboo forest, bushland and grassland). We used the supervised classification method in Erdas Imagine v. 8.6 (ERDAS, Norcross, USA) to analyse 10-m resolution Spot images of 160 randomly selected locations (Dean & Smith, Reference Dean and Smith2003; Irwin et al., Reference Irwin, Johnson and Wright2005). To calibrate the image we surveyed vegetation composition within a 4 × 5 m quadrat at each of the 160 points, located with a global positioning system (GPS), recording plant species and their frequency, and canopy cover. Quadrats that contained trees > 30 cm diameter at breast height with dominant canopy cover were classified as tree-dominated forest. Quadrats with > 60% cover of Euphorbiaceae (Erythrococca spp.) and/or other shrubs were classified as bushland. Quadrats with > 50% cover of bamboo Arundinaria alpina were classified as bamboo forest. Quadrats with primarily grass species were classified as grassland. Following classification, the areas of each of the four habitat types were calculated using the geographical information system (GIS) ArcGIS v. 9.1 (ESRI, Redlands, USA).
Habitat preference and population estimate
In Odobullu Forest transect sampling (Peres, Reference Peres1999) was used to determine the habitat preference of the Bale monkey and to estimate the species’ population size. Transects were established based on a stratified random sampling approach within each habitat type (Chapman et al., Reference Chapman, Fedigan and Fedigan1988; Plumptre, Reference Plumptre2000; Lacher, Reference Lacher2003). A total of nine transects of 1.8–3.0 km in length were censused in bamboo forest, tree-dominated forest and bushland using both existing and newly established trails. We did not survey grassland because we found that it comprises only 7% of the area and is distributed in many discontinuous patches. Each transect was censused 12–16 times in total, during both the wet (September–October 2007) and dry seasons (December 2007–January 2008). Censuses were conducted on foot by AM and a field assistant, together with villagers who were familiar with the area, from 06.30–06.45 to 10.30–10.45 and 14.00–18.00 (Peres, Reference Peres1999; Timmuck & Vaughan, Reference Timmuck and Vaughan2002) at an average speed of 1 km h-1 in forest and 2 km h-1 in bushland (Wallace et al., 1998; Peres, Reference Peres1999; Chapman et al., Reference Chapman, Gillespie, Skorupa and Struhsaker2000).
When Bale monkeys were encountered the dominant habitat type was noted (Wallace et al., 1998), along with location (with a GPS), group size, distance from observer and sighting angle (Chiarello, Reference Chiarello2000; Fashing & Cords, Reference Fashing and Cords2000; Bennet et al., Reference Bennet, Leonard and Carter2001; Anderson et al., Reference Anderson, Rowcliffe and Cowlishaw2007). Distance was estimated visually by AM, who previously practiced estimating distance, and sighting angle was measured with a compass.
Encounter rates of groups per km were calculated by habitat type (Wallace et al., 1998; Bobadilla & Ferrari, Reference Bobadilla and Ferrari2000), and sightings were summarized as the total number of groups and individuals observed in each habitat type (Anderson et al., Reference Anderson, Rowcliffe and Cowlishaw2007). We estimated density using the animal–observer distance method to the first animal seen (Chapman et al., Reference Chapman, Fedigan and Fedigan1988; Chapman et al., Reference Chapman, Gillespie, Skorupa and Struhsaker2000; Fashing & Cords, Reference Fashing and Cords2000; Marshall et al., Reference Marshall, Lovett and White2008). We created a histogram of all animal–observer distances, in 10-m interval classes (Müller et al., Reference Müller, Velo, Raheliarisoa, Zaramody and Curtis2000). The fall-off distance or maximum reliable sighting distance was determined by applying the 50% cut-off rule, i.e. the point at which sighting frequency decreased by at least 50% from the previous sighting class interval (Fashing & Cords, Reference Fashing and Cords2000; Chapman & Chapman, Reference Chapman and Chapman2002; Teelen, Reference Teelen2007; Weghorst, Reference Weghorst2007). Group sightings lying beyond the fall-off distance were excluded from density estimation (Twinomugisha & Chapman, Reference Twinomugisha and Chapman2006).
Group density was calculated as the total number of groups sighted within the fall-off sighting distance divided by the total transect length multiplied by transect width. Population density was estimated by multiplying group density estimates with the observed mean group size (Rosenbaum et al., Reference Rosenbaum, O’Brien, Kinnaird and Supriatna1998; Fashing & Cords, Reference Fashing and Cords2000). Sightings of solitary individuals were excluded from calculations of group density estimates and encounter rates (Fashing & Cords, Reference Fashing and Cords2000; Worman & Chapman, Reference Worman and Chapman2006). The total population was estimated by multiplying the average group density estimates by the total area of suitable habitat (Chiarello, Reference Chiarello2000).
Distribution across the Bale Mountains
Based on the altitudinal preference of Bale monkeys, (2,400–3,250 m; Mekonnen, Reference Mekonnen2008), the species’ potential range was demarcated using a 200-m digital elevation model in ArcGIS. This area was then classified into three habitat types (agriculture and human settlement, tree-dominated forest and bamboo forest) using a 10-m resolution satellite image with Erdas Imagine (Rappole et al., Reference Rappole, King and Leimgruber2000; Irwin et al., Reference Irwin, Johnson and Wright2005).
Based on the results of the habitat preference survey, the bamboo forest of the Bale Mountains was then surveyed, from east to west, covering c. 40% of the total area of this forest, during January–April 2008. We surveyed by walking along existing paths, besides rivers and on trails that we cut. When Bale monkeys were sighted we noted the location, with a GPS, group size, altitude and habitat type (Lehman, Reference Lehman2004; Isbell & Chism, Reference Isbell and Chism2007). Locations of the species recorded during our surveys and locations obtained from the literature (Butynski et al., in press) were incorporated into ArcGIS to create a map of the species’ distribution (Baumgarten, Reference Baumgarten2006). This survey was supplemented with informal interviews (Iwanaga & Ferrari, Reference Iwanaga and Ferrari2002) of the people who were familiar with, and reside in the vicinity of, the bamboo forest, during which we showed photographs of the Bale monkey.
Results
Tree-dominated forest, bamboo forest and bushland accounted for 56, 20 and 17% of the area of Odobullu Forest, respectively. A total of 136 transect censuses were conducted over a cumulative distance of 280 km in the three habitat types. Bale monkey groups were sighted on 43 occasions, all in bamboo forest (Table 1).
Sightings of Bale monkey Chlorocebus djamdjamensis groups along transects in bamboo forest, tree-dominated forest and bushland in Odobullu Forest (Fig. 1).
The mean group size of Bale monkeys encountered along the two transects in bamboo forest was 19.1 ± SD 5.43 (range 13–41). The fall-off distance was 40 m and not significantly different between transects (Mann-Whitney U test, P > 0.05). Of the sightings, 41 were used to estimate density (two were beyond the fall-off distance). The mean group encounter rate was 0.58 ± SD 0.08 km-1, mean group density was 7.0 ± SD 1.3 km-2 (range 6.0–7.8) and mean individual density 132 ± SD 14 km-2 (range 121–141). Based on the range of mean individual density we estimated that the total population size of the Bale monkey in the bamboo forest of Odobullu is 1,718–2,002.
We sighted the Bale monkey in bamboo forest at five localities across the Bale Mountains (Fig. 1), of which Shedem, Woge and Gamma are newly discovered locations for the species. The highest altitude at which we observed the species was 3,200 m, in the Woge area. The Odobullu and Shedem populations were not in the vicinity of villages but rather surrounded by tree-dominated forest in mountainous terrain, whereas the Woge and Gamma populations were within the vicinity of people. In the Shedem population we observed two groups, of c. 18 and 32 each, but local people estimated that > 10 groups occur in the area. In the Woge and Gamma populations group sizes were 15–30 and we estimate that there are > 300 Bale monkeys in each locality.
Discussion
The results of our study suggest that the Bale monkey exclusively inhabits bamboo forest, irrespective of season, and avoids tree-dominated forest and bushland. As with the golden monkey Cercopithecus mitis kandti at Mgahinga Gorilla National Park, Uganda (Twinomugisha & Chapman, Reference Twinomugisha and Chapman2006) the habitat preference of the Bale monkey is almost certainly because of its dietary specialization on bamboo, which accounts for 77% of its diet at Odobullu (Mekonnen, Reference Mekonnen2008; Mekonnen et al., Reference Mekonnen, Bekele, Fashing, Hemson and Atickem2010). The Bale monkey is thus a specialist compared to its close relatives the grivet C. aethiops and vervet C. pygerythrus, both of which are habitat generalists (Kingdon, Reference Kingdon1997; Zinner et al., Reference Zinner, Pelaez and Torkler2002; Barrett, Reference Barrett2005). Species with a narrow ecological niche are particularly susceptible to extinction because of habitat destruction (Yu & Dobson, Reference Yu and Dobson2000; Harcourt, Reference Harcourt2006). The legal harvesting of bamboo by local people for commercial purposes could pose a serious future threat to the Bale monkey, and needs to be either discouraged or closely managed. The survival of the species clearly depends on planning and implementing conservation strategies for the management of the bamboo forest.
Our density estimates depend on the encounter rates and group size, and the method used to estimate transect width (Shah, Reference Shah2003; Weghorst, Reference Weghorst2007). To reduce errors in group size estimation, group sizes were always recounted by the accompanying assistants during transect walks. The maximum reliable sighting distance was used for estimation of transect width, a technique that has proven effective for density estimation in many forested sites (Chapman et al., Reference Chapman, Fedigan and Fedigan1988; Chapman et al., Reference Chapman, Gillespie, Skorupa and Struhsaker2000; Fashing & Cords, Reference Fashing and Cords2000; Shah, Reference Shah2003; Irwin et al., Reference Irwin, Johnson and Wright2005; Marshall et al., Reference Marshall, Lovett and White2008). We therefore believe that our estimate of the total population size of the Bale monkey in the bamboo forest of Odobullu is reasonably accurate.
The density and biomass of many primates are primarily related to food quality and availability (Chapman & Chapman, Reference Chapman and Chapman2002; Fashing et al., Reference Fashing, Dierenfeld and Mowry2007). The relatively high density of Bale monkeys in Odobullu Forest may be related to the high abundance and productivity of their main food source, bamboo (Mekonnen, Reference Mekonnen2008). It is likely, however, that other features, such as soil quality, climate, forest history, competition, predation and disease (Oates et al., Reference Oates, Whitesides, Davies and Mole1990) also influence the density of the Bale monkey at Odobullu.
Surveys in c. 40% of the bamboo forest in the Bale Mountains located the Bale monkey in three new localities. These discoveries suggest that the range of the species is broader than suggested in previous reports (Carpaneto & Gippoliti, Reference Carpaneto and Gippoliti1994; Kingdon, Reference Kingdon1997; Butynski et al., in press). The Bale monkey is now categorized as Vulnerable on the IUCN Red List (Butynski et al., Reference Butynski, Gippoliti, Kingdon and De Jong2008), partly based on the results of our surveys. This is the first time the species has received a designation other than Data Deficient. Although extensive, our surveys did not cover all of the species’ potential habitat and therefore the Vulnerable categorization should remain, pending the results of further surveys, which we plan to conduct, across all the bamboo forest of the Bale Mountains and Sidamo region (the western extension of the Bale Mountains).
Acknowledgements
This research was funded by the Ethiopian Wolf Conservation Programme, Primate Conservation, Inc. and Pittsburgh Zoo and PPG Aquarium. Satellite imagery was provided by FARM-Africa and SOS Sahel Ethiopia. We thank Addis Ababa University (Department of Biology) and Frankfurt Zoological Society for logistical support, the Federal Ethiopian Wildlife Conservation Authority and Oromia Region Wildlife Department for permission to conduct this research, and Peter Fashing, Nga Nguyen, Colin Groves, Tom Butynski and two anonymous reviewers for their helpful comments and for providing access to relevant literature. We are also grateful to our field assistants and camp attendants and to the local villagers across the region.
Biographical sketches
Addisu Mekonnen’s research interests include the ecology, conservation and taxonomy of the Bale monkey and other primates and large mammals. Afework Bekele is working on a wide range of wildlife species in Ethiopia in collaboration with students, in particular in the general area of mammal ecology. Eyob Teshome is working on land-use cover change in the Bale Mountains in relation to wildlife conservation and is also interested in GIS and remote sensing. Graham Hemson is working on the conservation of the Ethiopian wolf and is also interested in the conservation and ecology of carnivores and other wildlife. Anagaw Atickem is working on the ecological genetics of mountain nyala and Bale monkey and is interested in GIS-based ecological studies and the use of population genetics in wildlife research.
