How does forest species specialization affect the application of the island biogeography theory in fragmented landscapes?

Aim: Overall species richness in habitat remnants is seldom explained by the island biogeography theory (IBT). In this study, we tested the effectiveness of the IBT in explaining species richness of forest birds with or without considering the effect of the different forest specialization of species (generalist, edge and interior; community analysis). We also identified single species and groups of species that could serve as indicators of different fragmentation degrees (indicator species analysis). Location: Broadleaved forests in western Lombardy (northern Italy). Methods: We evaluated bird species richness in 344 forest fragments. Fragment area (FA), distance from the nearest source area (> 1000 ha) (DSA) and number of neighbouring fragments (NF) were calculated for each fragment. Using Poisson generalized linear models, we compared a model that evaluated the relationships between fragment covariates and species richness to a model that also considered the effect of forest species specialization on each relationship. Moreover, we investigated the association between each species and particular fragmentation conditions by calculating the IndVal index. Results: The application of the IBT to species richness performed significantly better if we took into account forest species specialization when evaluating the effect of fragment covariates. In particular, the positive effects of FA and NF and the negative effect of DSA were significantly stronger on interior species rather than on generalist and edge species. Using the IndVal index we identified six species in three groups strongly associated with specific fragmentation degrees. Main conclusions: The results showed the strong influence of the specialization of species on their distribution in fragmented landscapes. Interior species were the best candidates as a proxy of fragmentation effects. However, the indicator species analyses revealed that not all interior species showed the same sensitivity to fragment covariates.