Scopus24:21-32,July2004 A preliminary survey of the avifauna of Eminit Forest, Loima Hills, western Turkana District Marc Baker Eminit Forest, located in the Loima Hills of northern Kenya, is an essential part of the grazing lands of the Turkana people. The Turkana people are sub-divided into 17 groups, with five of these groups, the Ngisir, Ngimonia, Ngikamatak, Wayakwara and Lukumong, depending on the Loima ecosystem for dry season grazing (Oba 2001). Through tribal agreements the Turkana people have created an indigenous land- use system which is enforced by the tribal chiefs and adhered to the majority of the time. Eminit forest not only provides dry season grazing but also a refuge during periods of tribal instability and sporadic clashes that occur between the Turkana and Karamajong. As reported by Barrow (1988) and Oba (2001) the conservation of biodiversity at this site is not related to external management but indigenous knowledge and land-use, which has evolved out of a need to protect the natural resource for use during times of environmental stress and tribal instability. In terms of their effect on localised rainfall patterns and water catchment, the Loima Hills, and Eminit Forest in particular, are extremely valuable to the local pastoralist inhabitants as well as resident and migratory fauna. The main aims of this survey were to establish a bird species list and note any species of conservation concern that may occur in the forest. An assessment of the threats to those species was made as part of a report for the UNDP/GEF Crossborder Biodiversity Project which was conducting a forest conservation project in the district. Study site The Loima Hills (also known as Murua Ngithigerr) are situated at 03°30'N, 35°00'E covering an area of approximately 160 kml Combined with the Puch Prasir plateau they cover a total area of 3000 km^ representing one of the most extensive highland areas in Turkana District. Eminit Forest is situated on the Loima Hills plateau, from an m altitude of 2050 to a forested peak on its southern edge at 2286 m. The km km forested plateau is 9 at its widest point (east to west axis), 15.8 at its longest point (north to south axis) and covers an area of approximately 83 km^ (all values measured from 1:250 000 map, series Y503, sheet NA-36-4, 1975). The lower slopes of the Loima Hills are accessed by bush road, a distance of 55 km east-north-east of Lodwar, 22 Marc Baker the provincial capital of Turkana District. Approximately 9 km (10%) of the western edge of Eminit Forest is located in Kenya bird atlas square 12D, the remainder being located in square 13C (Lewis & Pomeroy 1989). The forested plateauis relatively flatwith an altitudinal range of only m 200 over an area of 83 km^ but the edges of the Loima Hills consist of steep slopes and cliffs which create an irregular drainage pattern over bare rock. Although there are more established drainage lines, no flowing water was observed in May 2001. The west and north-west slopes of the Loima Hills drain into the upper Tarach catchment system, the south and south-east areas into the Kosipur-Turkwell catchment and the Kaulathe-Turkwell catchments system drains the north-eastern slopes of the mountain (Barrow 1988). The vegetation of the Loima Hills area is related to elevation more than terrain. The lower slopes largely comprise Aristida mutabilis and Acacia mellifera with Acacia tortilis dominating the vegetation along seasonal water courses and drainage lines. The soils are generally rocky and composed of euteric outcrops and calcic cambisols. The dry montane forest is dominated by Juniperus procera, Olea europea, Olea capensis, Tech nobilis and Podocarpus falcatus, while the sparse undergrowth is composed of Maytenus heterophylla, Domheya goetzenii and Pistacia aethiopica (Barrow 1988). The forest interior comprises mainly Podocarpus falcatus with Juniperus procera, with the latter dominating along ridges. These dominant species form a loose canopy at an estimated 25-30 m, with Olea europea and O. capensis forming a sub-canopy at an estimated 10-15 m. In approximately 65% of the forest visited, there was very little undergrowth, with saplings forming much of the vegetation structure between ground level and 5 m. There were occasional shrubs where breaks in the canopy had allowed light to penetrate, forming islands offorest edge habitat within the forest. Large amounts of dead wood and leaf litter covered the forest floor providing a micro-environment for large numbers of invertebrates. Much of this dead wood seemed to be very old and the rate of decompositionslow which is possibly linked to the lack of moisture. Methods A combination of mist-netting, timed species counts (TSCs) and ad hoc observation were used to survey the avifauna in Eminit Forest over a period of ten days from the 30 April to 9 May 2001. Mist netting was carried out from 06:00 to 12:00 and 16:00 to 18:30. All birds caught were fitted with a standard Museum of Nairobi ring and the following biometric measurements were taken: tarsus, wing, bill, tail length and weight. These data, along with brood patch (six point scale from O=none to 5=full) and moult data were deposited with the National Museum of Kenya, Nairobi. Avifauna of Eminit Forest, Loima Hills 23 Timed species counts were carried out in the mornings between 07:00 and 09:00 in forest, and between 18:00 and 19:00 at forest edge. Two counts were carried out each morning, the first being at a randomly m selected point along a path, at least 500 into forest. The second count m was carried out 300 (and no more than 15 minutes walk) from the firstpoint in a directline walked on a easterly compass bearing. m Only birds within a fixed radius of 25 (estimate of error is 3 m, m 12%) and 3 above the ground were recorded. Each point was sampled for 40 minutes with observations being allocated to 5 minute time intervals giving scores of between 0 and 8 over the sample period. All other methods followed Bennun and Waiyaki (1992) All counts were madeby the same two observers (Marc Baker and Njano Mbilinyi) The results from Loima were compared with those from other northern dry montane forests, in particular Kulal and Njiru (Borghesio pers. comm.) and Moroto (Davenport et al. 1996). The main aim was to look at the similarity of these sites in terms of the avifauna known to occur, using Bray-Curtis cluster analysis. This uses group-average clustering to give a hierarchy of clusters indicating similarity as a percentage. The analysis concentrates on the forest dependant avifauna (both forest specialist and forest generalist) as defined by Bennun et al 1996. Results and discussion A total of 87 species were recorded, representing 40 families and including three species of conservation concern (Bennun & Njoroge 1996). A full list is shown in the appendix. Of these 87 species, 48 were recorded in Eminit Forest and at forest edge (above 2050m), whilst the m remainder were recorded in the dry Acacia woodland between 800 (at the foot of the Loima Hills) and 2050 m. Mist netting A total of 2616 net-metre-hours (nmh) were completed over a five-day period, in which 13 birds were caught and processed (Table 1). The low number of captures reflects the low effort but also the low capture rate of 0.004 birds per nmh. This was probably due to the lack of a shrub layer within the forest which would normally help conceal the nets. Table 1. Results frommist-nettingin EminitForest. Species Ecol.Type No.caught Rank White-eyedSlatyFlycatcher F 3 1 Scaly-throated Honeyguide f. 2 2 Grey-backedCamaroptera f. 2 2 BrownWoodlandWarbler F 3 YellowWhite-eye f. 3 Brown-headedTchagra Fe 3 Cape-RobinChat f. 3 AmethystSunbird f. 3 VariableSunbird f. 3 24 Marc Baker Almost all of the birds caught and ringed were forest species: two forest generalists (F), six forestusers (f.) and one forest edge (fe) species. Timed species counts A total of 16 TSCs were carried out in forest interior and forest edge, giving a species list and index of relative abundance for each habitat type. The results for the timed species counts are shown in Tables 2 and 3 which indicate the commonest canopy and mid-stratum species in both the forest interior and forestedge. In the forest interior, ten TSCs were carried out during which 13 species were recorded (Table 2). African Goshawk Accipiter tachio, Olive Pigeon Columha arquatrix, Narina Trogon Apaloderma Narina and Brown Woodland Warbler Phylloscopus umbrovirens were recorded only in forest interior. Table 2. Results oftimed species countsinforestinterior, EminitForest. Species Ecol.Type TSCIndex Rank BrownWoodlandWarbler F 4.0 1 OliveThrush F 2.0 2 Grey-backedCamaroptera f. 1.9 3 DuskyFlycatcher F 1.4 4 White-eyedSlatyFlycatcher F 0.9 5 GreyCuckooShrike FF 0.8 6 AmythstSunbird f. 0.8 6 YellowWhite-eye f. 0.8 6 Olive Pigeon F 0.7 7 DuskyTurtle Dove f. 0.7 7 AfricanGoshawk F 0.6 8 Common Bulbul f. 0.5 9 NarinaTrogon F 0.2 10 At forest edge, six TSCs were carried out in which 15 species were recorded (Table 3). Variable Sunbird Cinnyris venusta, Cape Robin-Chat Cossypha cajfra, Baglafect Weaver Ploceus haglafect, Northern Puffback Dryossvopus gambensis, Paradise Flycatcher Terpsiphone viridis and Yellow-crowned Canary Serinus canicollis were all recorded only at forest edge. Table 3. Results oftimed speciescounts along forestedge,EminitForest. Species Ecol.Type TSCIndex Rank VariableSunbird f. 5.3 1 AmethystSunbird f. 4.5 2 YellowWhite-eye f. 3.6 3 Common Bulbul f. 3.6 3 Cape Robin Chat F 3.5 4 BaglafectWeaver f. 2.6 5 OliveThrush F 2.6 5 Northern Puffback F 1.3 6 Grey-backedCamaroptera f. 1.3 6 GreyCuckooShrike FF 1.1 7 Paradise Flycatcher F 1.1 7 YellowCrownedCanary f. 1.1 7 DuskyFlycatcher F 0.6 8 White-eyedSlatyFlycatcher F 0.6 8 DuskyTurtle Dove f. 0.3 9 Avifauna of Eminit Forest, Loima Hills 25 Species accounts Great Sparrowhawk Accipiter melanoleiicus (dark morph) was recorded four times carrying food in the same direction at dusk and thus suspected to be breeding. As all observations of this species were of a dark morph in the same location at approximately the same time it was presumed that this was the same individual. Peregrine Falcon Falco peregrinus was recorded daily at dusk hunting over grassland. The presence of cliffs which this species requires for breeding make it likely that this was a residentbird. Heavy streaking on the breast indicated that this was a sub-adultbird. White-crested Turaco Tauraco leucolophus was recorded in forest edge and riverine forest as low as 1800 m. It was absent from the forest interior and the dry woodlandbelow 1800 m. African Hill Babbler Alcippe abyssinica was recorded throughout the forested plateau. Cunningham-van Someren and Schifter (1981) described a separate race for Loima (loima) based on collections madeby Joseph Mwaki in 1962. Olive Thrush Turdus olivaceus was observed throughout habitat types above 2000 m, where it was recorded feeding in the ground layer, sub- canopy and canopy. This is usually a bird of the ground and shrub layer, so it is thought that the absence of a shrub layer combined with a lack of resource competition from greenbuls and hornbills (see below) may have allowed this species to occupy the canopy layer. Cunningham-van Someren and Schifter (1981) make reference to the sharp division between the extremely pale birds from Loima, Morongole, Karissia, Nguruman Range and those from other areas. Brown Woodland WarblerPhylloscopus umbrovirens was the commonest bird in the forest interior (see TSC results. Table 2). A bird usually associated with shrub layer and small trees, this species predominantly occupied the sub-canopy and canopy, and was often recorded crawling through theJuniperus-Podocarpus canopy at over 30 m. Cinnamon Bracken Warbler Bradypterus cinnamomeus was recorded twice in the undergrowth of forest clearings created by felled trees and imdergrowth along drainage lines within the forestinterior. Baglafect Weaver Ploceus baglafect emini, is distributed in woodland and m m marsh between 800 and 1200 from north-west Uganda east to the Loima Hills. It was recorded in forest edge and clearings where it was relatively abundant (see TSC results. Table 3); several juvenile birds were also recorded. Large numbers of swifts mainly African Black Swift Apus barbatus (estimates of flock size over 250 individuals) but also significant numbers of Alpine A. melba and Mottled Swift A. aequatorialis (estimates of flock size over 50 individuals) were recorded daily, feeding above grassland and along the edges of cliffs. Nyanza Swift A. niansae was observed daily in mixed species flocks, but in considerably lower 26 Marc Baker numbers than other members of this family (usually 3-4). Common House Martin Delichon urbica were present in low numbers, often in mixed flocks. Comparison with other northern montaneforests A comparison was made of species that have been recorded in the similar northern dry montane forests of Kulal and Njiru (Borghesio pers. comm.) and Moroto (Davenport et al. 1996). This comparison was based on the forest dependant species (forest specialist FF, and forest generalist F) for each site. Loima had 19 species, Moroto 28, Kulal 16 and Njiru 26. When looking at the similarity of these sites, all mountains showed a high percentage in the similarity of their species composition, the highest being shown by Moroto and Loima (68%) and the least by Moroto and Kulal (55%, Table 4). Table 4. Similarity matrix showing the percentage similarity between the montaneforest sites. SimilarityMatrix Moroto Njiru Kulal Loima 68.0851 66.6667 51.4286 Kulal 54.5455 66.6667 Njiru 62.963 Moroto * * * The only species so far recorded at all of these sites were: Mountain Buzzard Buteo oreophilus, Red-chested Cuckoo Cuculus solitarius, Narina Trogon, African Hill Babbler Pseudoalcippe abyssinica, Olive Thrush, Brown Woodland Warbler and Northern Puffback. The similarity between Moroto and Loima is influenced by the presence of two species in particular that have not been recorded at Kulal and Njiru, namely Tambourine Dove Turtur tympanistria and African Wood Owl Ciccaha woodfordii. Clearly, Moroto has the greatest diversity of forest dependant species (n=28), with eight occurring only at that site. At Njiru (n=26) there are four species unrecorded from Kulal and Loima, all species typical of dry montane forests. There were several groups of birds that were conspicuous by their absence from Eminit. No members of the genus Apalis were recorded. Yellow-breasted Apalis Apalis flavida, a bird of woodland and forest edge, was expected to occur, and this absence maybe linked to isolation, as it has not been recorded in the arid woodland of north-western Kenya (Zimmerman et al. 1996). The absence of Grey Apalis Apalis cinerea, an active bird found in the canopy of similar montane forest, cannot be linked to habitat variables. It would almost certainly have been recorded were it present. Chestnut-throated Apalis Apalis porphyrolaema has been recorded in Moroto and in isolated montane forests south of Loima. The absence of any greenbuls, Cabanis's Greenbul Phyllastrephus cabanisi in particular, is odd. Resource Avifauna of Eminit Forest, Loima Hills 27 availability does not appear limited for an insectivorous bird such as this, which does occur on Moroto, Kulal and Njiru. This species is commonly caught in mist-nets so should have been recorded if present. Yellow-whiskered Greenbul Andropadus latirostris is another likely species that was not recorded. This bird is predominately frugivorous and very few sub-canopy fruiting shrubs or trees were noted. Northern Brownbul Phyllastrephus strepitans, a bird of woodland and forest edge, was also expected to occur since the habitat and resource availability would seem to suit this species. Silvery-cheeked Hornbill Bycanistes hrevis was not recorded despite the presence of fruiting figs and the large trees required for nesting by this species. The absence of White- starred Forest Robin Pogonocichla stellata is unusual because this species is found in the lower stratum of these forest types and occurs on many isolated forest islands in Kenya and Tanzania, supporting the fact that this species has a good ability to disperse. Abyssinian Ground Thrush Zoothera piaggiae occurs on Kulal, Njiru and Moroto but is absent at Loima. There are several reasons why these species may be missing including: 1) they were missed by the survey team, although this is unlikely for those species which are vocal or gregarious (e.g. some of the apalises); 2) some aspect of the habitat was unsuitable. The stratification of the vegetation seemed to have a large impact on avian diversity, and the feeding ecology of many of the bird species. Many of the species normally associated with undergrowth were notably absent, while insectivorous and omnivorous species one would normally associate with undergrowth were regularly observed in the sub-canopy and canopy 3) they have been unable to colonise the site due to poor dispersal ability and/or lack ofhistorical habitat connectedness. White-browed Robin-Chat Cossypha heuglini was collected in the Loima Hills in 1962 by Joseph Mwaki for A.D. Forbes-Watson, but was not recorded during the current survey. Threats and conservation recommendations The threats to biodiversity at this site are varied. Firstly, the presence of fire arms has almost certainly greatly reduced wildlife populations. Although dikdik are fairly common, these were the only "large'' mammals that were observed in the acacia woodland. It is unlikely that opportunistic hunting with firearms would affect the avifauna as the cost of a single round would not be justified by the meat gained. However, gunfire was heard several times during this two week survey suggesting that ammunition was nothard to obtain. The large amounts of fuel wood still available around the Loima Hills and within Eminit Forest would indicate that felling trees for firewood by local people is not a cause of concern. Within Eminit Forest there were no signs of felling for fuelwood or timber, though several trees had beenburnt for wild honey. In a couple of instances, Podocarpus 28 Marc Baker falcatus were felled for seemingly no apparent reason other than to allow light through to the forest floor and create grazing for future use. The major conservation concern encountered was charcoal production. This activity is mainly restricted to the Acacia tortillis riverine woodland running from the Loima Hills east to Lodwar. The demand for charcoal is generated in Lodwar where the increasing populationhas led to an increased demand for fuel. The indigenous land-use system previously outlined, is born of an understanding of the need for dry season grazing (for herbaceous production levels see Barrow 1988). The land-use systems that are in place are designed around a model of sustainability to ensure that this resource is protected in perpetuity for times of environmental stress. In terms of the conservation of biodiversity, the current land-use system would seem to be beneficial with the current land-users also being its guardians. However, as stated by Barrow (1988), the presence of a legal framework that acknowledges the current land-use system as well as allowing a measure of protected status is essential. Whereas definitions of protected areas such as Game Reserve, National Reserve and Biosphere Reserve are based on the exclusion of people and preservation of the natural resource to varying degrees, the best option for Loima would be a Forest Conservation Reserve with its own rules and regulations (Barrow 1988). Conclusions This preliminary survey shows Eminit Forest to have a relatively poor avifaunal diversity with only 19 forest dependent species of which four are forest specialists. Due to its northerly location it does represent an isolated outlier of this montane forest type, and as with other isolated mountains, Eminit represents an important refuge for the movement of birds as well as other wildlife and people. Although only four forest specialist species have so far been recorded at this site, Eminit Forest's effect on local rainfall patterns, water drainage and thus vegetation biomass over a large area is clear (Barrow 1988, Amuyunzu & Oba 1991). As such, Eminit Forest is of significance for many species of birds and wildlife that, although they are not truly forest dependant species, are very much dependant on the presence of the forest. Acknowledgments Many people aided this field project. Thanks to all the staff at the UNDP/GEF Crossborder Biodiversity Project in Nairobi and Lodwar, in particular Mr. George Ondenge, Mrs. Rose Antibo, Mr. Moses Lekong, Mr. Richard Musee and Mr. Japheth Katembo. I amextremely grateful to Dr. Leon Bennun and Mr. Luca Borghesio for their prompt email responses and to Dr. Alan Rodgers for his technical advice. Special thanks go to my field team, Mr. Njano Mbilinyi, Avifauna ofEminit Forest, Loima Hills 29 Mr. David Gitau and Miss. Lauren Persha for their valuable contributions and Mr.Jo Andersonforproofing the final drafts. References Amuyunzu, C.L. and Oba, G.1991. Vegetation Resources of Central Turkana District, Kenya. UNESCO. — Barrow, G.C. 1988, Loima Forest, Turkana Its present management andfuture con- servation issues. TurkanaRural DevelopmentProgram (NORAD). Bennun, L.A. and Waiyaki, E.M. 1992. Mau Forest Complex Ornithological Survey Report. NationalMuseums ofKenya. Bennun, L.A., Dranzoa, C. & Pomeroy, D. 1996. The forest birds of Kenya and Uganda.Journal oftheEastAfrican NaturalHistory 85: 23-48. Bennun, L.A. and Njoroge, P. 1996. Birds to watch in East Africa: A preliminary Red Data list. Nairobi: The NationalMuseums ofKenya. Britton, P.L. 1980. (ed) Birds of East Africa. Nairobi: The East Africa Natural HistorySociety. Cunningham-van Someren, G.R. & Schifter, H. 1981. New races of montane birds from Kenya and southern Sudan.Bw/Zefm of the British Ornithologists' Club EANHS 1996. Check-list of the birds ofKenya. Nairobi: Ornithological sub-com- mittee. EastAfricaNaturalHistorySociety. Davenport, T. and Howard, P. 1996. Moroto Forest Reserve, Biodiversity Report. Kampala: UgandaForestDepartment. Lewis, A. & Pomeroy D. 1989. A bird Atlas ofKenya. Rotterdam: A. A. Balkema Publishers. Oba, G. 2001. Reporton theevaluation ofbiodiversity in crossborders ofeastAfricafor F.A.O. The Turkanasites ofKenya. Noragric. Zimmerman, D. A., Turner, D. A. & Pearson, D. J. 1996. Birds ofKenyaand north- ern Tanzania. SouthAfrica: RusselFriedmanBooks. 30 Marc Baker Appendix Bird species recordedfrom the foot ofLoimaHills at800mto theforested plateau at2100m. Taxonomic order follows EANHS (1996) withtheirr\umber indicated. TheKenyanbird atlas squarenumberis given for each species following Lewis & Pomeroy (1989). Those species whichare ofparticular conservationsignificance are indicated as: RNT=RegionallyNearThreatened, RVu=RegionallyVulnerable and RR=RegionallyRestricted (Bennunand Njoroge 1996). Ecological type for forestbirdsfollows Bennunetal. (1996); FF=forestspecialist, F=forest generalist and f.=forestvisitor. The remaining non-forestbirds are categorised as follows: W=Widespread,Wd=Woodland, G=Grassland and Fe=Forestedge. Breeding records are indicated as: SusBr^Suspectedbreeding andBr=Confirmedbreeding. No Family/English name Scientificname Ecol.type Atlassq. AccipitridaG 1ui bwailow-1aiieaMie OfielictiniBnocouni Wd 130 \Vd DiacKi\iie Milvusmigrdns G 130 12D 1 1/ narnerriaWK roiyuoroioeslypnus f. 130 12D 125 AfricanGoshawk Accipitertachio F 130 127 Shrika Accipiterbadius f. 130 104 urealoparrowriaWK MccipiieriTieidnuieucus F 130 12D 139 Mountain Buzzard Buteooreophilus FF 130 12D W 142 AugurBuzzard Buteoaugur 130 W 158 Martial Eagle Polemaetusbellicosus 130 159 PygmyFalcon Polihieraxsemitorquatus Wd 130 162 PeregrineFalcon Faicoperegrinus Fe 130 12D iDD Mincan noDDy ralCOcuvieri F 130 12D 1/4 roxi\esTrei FaIcoalop6x Wd 130 Phasianidae lyu uresiearrancoiin FrancolinusSBphaena Wd 130 Glareolidae 257 SomaliCourser Cursoriussomalsnsis Wd 130 ririliimhirlap 357 TambourineDove Turturtympanistria F 130 12D 359 Emerald-spottedWood Dove Turturchalcospilos f. 130 365 OlivePigeon Columbaarquatrix FF 130 366 Speckled Pigeon Columbaguinea Wd 130 371 MourningDove Streptopeliadecipiens Wd 130 376 DuskyTurtleDove Streptopelialugens f. 130 12D Musophagidae 399 White-crestedTuraco Tauracoleucolophus f. 130 12D 401 White-belliedGo-away-bird Corythaixoidesleucogaster Wd 130 Cuculidae 409 Red-chestedCuckoo Cuculussolitarius F 130 12D 417 EmeraldCuckoo Chrysococcyxcupreus F 130 12D 419 Klaas'sCuckoo Chrysococcyxklaas F 130 12D 422 White-browedCoucal Centropussuperciliosus Fe 130 Strigidae 444 WoodOwl Ciccaba woodfordii F 130 Caprimulgidae 449 MontaneNightjar Caprimulquspoliocephalus F 130 12D 460 Slender-tailedNightjar Caprimulqusclarus Wd 130