When habitat is lost so are species. One way of investigating the processes underlying this pattern is to pay attention to the identity (not only the number) of species. What happens to between-site differences in species composition when habitat loss transforms formerly continuous habitat into habitat fragments?

Who consults widely applied theoretical frameworks (e.g. theory of island biogeography) to answer this question will come to the conclusion that between-site differences in species composition – i.e. beta-diversity – should increase following habitat loss due to a strong influence of chance on the extinction process. Species are assumed to be ecologically equivalent (all have the same chance of getting extinct) and ecological drift (stochastic changes is species abundance) to increase in importance when populations are small. Further, chance makes it unlikely that populations surviving in different habitat remnants belong to the same species, and homogenization is hindered by isolation.

Who, on the other hand, consults empirical work will find that for various groups of plants and animals it is common to observe that, of the diverse set of species in continuous habitats, it is frequently the same small set of species that persists after habitat loss. Apparently, only certain resistant species are able to survive in fragments, thereby making the species composition in fragments deterministically more (and not less) similar, indicating – in contrast to theoretical models – low influence of chance on species extinction.

In our study “Ecological filtering or random extinction? Beta-diversity patterns and the importance of niche-based and neutral processes following habitat loss “ we investigated how the importance of different processes changes with habitat loss relying on a large database of small mammals in the Brazilian Atlantic Forest. We used a null model approach to quantify beta-diversity and make inferences about the relative importance of niche-based (deterministic) and neutral (stochastic) processes on community assembly at landscapes with varying degree of habitat loss.

Our results did not support a positive relationship between beta-diversity and habitat loss, as predicted by commonly-used theoretical frameworks. Rather, when considering exclusively species composition (disregarding their abundance), beta-diversity was independent from habitat loss, with small mammal communities being more similar than expected by chance in deforested as well as continuously-forested landscapes. However, when species abundance was taken into consideration, we observed a drastic decrease in beta-diversity with habitat loss (i.e. biotic homogenization), thereby indicating an increase (rather than a decrease) in the importance of deterministic processes at landscapes with high degrees of habitat loss. Finally, we observed a drastic change in species composition in a highly deforested landscape, with communities being not just a rarefied sample but rather disproportionately dissimilar to the communities in continuously-forested landscapes.

These results indicate that habitat loss can be seen as a strong ecological filter and species extinction is clearly more influenced by deterministic than by stochastic processes. Against this background, the incorporation of relevant species traits into theoretical models seems to be a useful step forward for the practical relevance of these models. Moreover, pro-active measures seem to be essential to prevent tropical landscapes to go beyond critical levels of habitat loss.

The authors through Thomas Püttker