The forest avifauna of Arabuko Sokoke Forest and adjacent modified habitats

Arabuko Sokoke Forest (ASF) is the largest area of coastal forest remaining in East Africa and a major Important Bird Area in mainland Kenya. The study analysed data from point count surveys over 15 months in three land use types; primary forest (PF), plantation forest (PL), and farmlands (FM), and compared these to the first comprehensive bird checklist for the forest, as well as recent surveys from other studies. Avifaunal diversity and abundance were compared using multivariate analysis to determine bird responses to different land use characteristics. The primary forest held a distinctive bird community, while the bird communities of farmlands and plantation forest were more similar to each other. Land use had a significant effect on overall avian diversity and abundance. The current forest avifauna was divided into forest specialists (16 species), forest generalists (26 species) and forest visitors (30 species). Seven species of forest specialist and generalists recorded prior to 1980 may no longer occur in the forest. Of 38 specialists and generalists recorded in our point counts, 19 were also recorded on farmland and 28 in plantations. One forest specialist, the Green Barbet, was most encountered outside the forest. Future research should focus on habitat use by these bird species, and the extent of movement by forest birds between the remaining patches of coastal forest. Patterns of habitat use by birds in the area suggest that vegetation heterogeneity and habitat complexity are especially significant in sustaining diverse and abundant bird populations. The management of plantations and farmland will be critical for the conservation of forest generalists and forest visitors.


INTRODUCTION
Since protected areas alone will be inadequate to conserve global biodiversity including birds, the management of human-modified habitats is critical for tropical bird conservation (1). Predicting the response of bird communities to habitat disturbances is still a challenge (2). Past studies have indicated a strong relationship between forest area and the 23 number of species of forest birds (3). While agroforests with a mix of cultivated and natural shade trees will attract a bird feeding guilds (15). In this paper, we focus on the species level and examine the effects of land use on the randomized separately. The process was repeated every evening before the next day of the survey until all points were 84 surveyed in every month of the field visits. This limited movement to short distances between points in one land use   index (16,21). Birds recorded were classified into three categories according to forest utilization: forest specialists (FF), 104 forest generalists (F) and forest visitors (f) according to (22).

106
Data analysis 107 Species richness was calculated as the cumulative number of species recorded at each point count station. ANOVA and 108 Tukey HSD post hoc tests were used to test for differences in bird diversity and abundance among the land use types.
Species diversity at each point was calculated based on the Shannon diversity index (H) (23)   Over the whole survey period, a total of 97 bird species was recorded at 81 points (Appendix 1). A Jacknife procedure 129 estimated the expected bird species richness for the area covered by the counts to be 114.82. Thus, the number of species 130 obtained was close to the expected, suggesting that the counts had sampled the whole bird community.

131
Of the 29 species in the East African Coastal forest which occur in Kenya, 26 have been recorded in Arabuko Sokoke 132 (10,14). During the point counts 17 of these species were recorded (Table 1) (27). Thirteen species were never recorded on farmland, and only one, the Scaly Babbler (Turdoides 138 squamulatus), was common in this habitat and in fact not recorded in the primary forest during point counts. Six of the 139 seventeen species were not recorded from the plantation areas, but some appeared to use this habitat regularly for bellied Starling (Notopholia corusca).  167 2). We found a highly significant difference in species numbers between primary forest and farmland, Tukey test (P< 168 0.001), a significant difference between primary forest and farmland (P<0.05) but no significant difference between 169 plantation and farmland (ns) (Fig 1).

214
Response of the avian community to land use changes 215 As expected, bird species diversity was highest in the primary forest while the farmlands supported fewer species. This 216 has been reported in many other surveys, e.g. (9,30,31). High bird species diversity in the primary forest can be attributed 217 in part to the complexity of habitat structure that provided resources for feeding and nesting (32) nesting, food resources and refuge from predators. Forest interior species have been reported to be highly sensitive to 232 disturbance and will even shun clearings or gaps resulting from treefalls (36) logged areas may take many years to be 233 recolonised (37) and species breeding in holes or crevices are particularly adversely affected (38).

234
The decline in bird species diversity in plantation forest and farmlands can be linked to the extent of disturbance and 235 reduction of native vegetation in these two land use types. In farmlands around the forest, there are few native fruiting 236 plant species on which birds and other animals may depend, so that farmlands support fewer species (9,39). Low canopy cover in plantations and farmlands due to intensive disturbance leading to low vertical vegetation heterogeneity, could 238 be a major contributor to the low number of species and the reduced density and abundance of many bird species.

239
However, we found evidence of increased diversity at points in farmlands with remnant native vegetation patches. Other 240 studies have also shown that even small changes in the structure and composition of tree cover may have a significant 241 impact on bird assemblages (5,32), leading to changes in bird diversity and community composition, with fewer species 242 and foraging guilds present in more intensively managed landscapes (7). Logging for timber in the plantation forest 243 resulting in clear-felled areas could also be a contributing factor to low species counts. Excessive disturbance brought 244 about by logging operations usually reduces total avian diversity (40) or promotes an influx of non-forest species which 245 replace forest specialists (37). Such changes particularly affect less abundant, range-restricted birds and rainforest 246 specialists (8). The trees in the plantations, mainly Eucalyptus sp. and Casuarina sp. provide few food resources. In East 247 Usambara, Tanzania, (41) also recorded a significantly lower diversity of forest birds in Eucalyptus plantations 248 compared to primary forest, which they attributed to limited nesting opportunities and reduced understorey cover.

249
Plantation forest in this study was associated with reduced variation in the vegetation structure, which lacks an 250 intermediate stratum which many bird species require.

251
Planting of native tree species is unlikely to be adopted by farmers without either financial incentives, or clear evidence

265
While conservation of the primary forest is important for bird conservation in this area, breeding areas outside the forest 266 may be equally significant. The breeding of Clarke's Weaver had been a mystery for many years and the nest had never conservation group, and the site is used for harvesting thatching material (49). Formal protection of this woodland and 271 the adjacent wetland is thus essential for preserving the breeding sites of this endemic species, which at other times is 272 apparently restricted to Arabuko Sokoke Forest.
Some species of special conservation concern were forest restricted including the Sokoke Pipit, Clarke's Weaver and 275 the Kenya Crested Guineafowl (Guttera pucherani) while others showed strong forest dependence (Appendix 1).

276
Similarly in the Solomon Islands endemic species were largely restricted to the forest, whereas birds with a wider 277 distribution were more common in other land-use types (50). Farmland and plantations around Arabuko Sokoke Forest 278 are currently subject to frequent large scale disturbance and a decline of forest specialists in such degraded farmland has 279 also been reported by (51). Forest specialists species are typically the first to be lost from degraded areas, whereas

290
Such trees improve soil fertility and are useful in the short term as fuel wood, hence providing multiple benefits to the 291 farmers (54). Farmers can also be influenced to adopt on-farm tree planting by supplying them with quality tree seeds 292 and involving them in experimental agroforestry alongside researchers (55). Redesigning agricultural landscapes to be 293 more diverse could promote both crop production, ecosystem services such as pollination (56)(57)(58) and use by forest