Tropical forests suffer severe habitat destruction. Thus, tropical forests frequently consist today of only a few small remnants that are often embedded within a matrix of agricultural fields and tree plantations. Forest specialist species have experienced severe population declines under these circumstances. We studied bird communities based on census plots set up in a near-natural forest block, as well as degraded forest patches, tree plantations, and agricultural fields, across the Taita Hills in southern Kenya. We classified each bird species according its ecology and behavior. We quantified the land cover and landscape configuration around each census plot. Typical forest species were mainly observed in the near-natural forest block, and to a lower extent in degraded forest patches. Plantations were almost devoid of birds. Bird communities of small forest fragments were more similar to that of agricultural land than the near-natural forest block. Most frugivorous, insectivorous and nectarivorous birds occurred in forest habitats, while granivorous bird species dominated the bird communities of agricultural land. The surrounding landscape had a marginal impact on bird species composition at local sites. Our study showed that the preservation of near-natural cloud forest, including small forest patches, is essential for the conservation of forest-dependent species, and that plantations do not serve as surrogate habitats.

Species respond differently to landscape structures and environmental changes. In nature conservation, however, responses of a few indicator groups are often generalised to the ecosystem level. In this study, we analyse how birds and butterflies respond to identical landscape structures and environmental gradients across a habitat mosaic in southern Kenya. The study area represents natural coastal forest (core and margins) as well as different agro-environments (orchards and pastures), which partly may also be suitable surrogate habitats for forest species. We assessed birds and butterflies during the same time and along identical transects, covering the dry and rainy season. The obtained results indicate that both indicator groups depict habitat types in some aspects similarly but in others in somewhat different ways. Thus, strongest differences in community similarity were visible between forest interior and open landscape (i.e. pastures) for both taxa. The forest community strongly overlapped with orchards for birds but less so for butterflies. Thus, orchards in close geographic context with natural forests might be possible surrogates for certain forest bird species but less so for the little mobile forest butterfly species. The temporal variation in species richness, abundances, and community structures was much stronger for butterflies compared to birds. Thus, seasonality in tropical ecosystems has to be considered when interpreting community structures of butterflies but much less so in birds. In general, birds and butterflies proved to be suitable indicator groups to ecologically assess landscape structures in East Africa, with birds more representing the landscape level and butterflies more the fine-grained habitat scale.