On the relative importance of space and environment in farmland bird community assembly.

The relative contribution of ecological processes in shaping metacommunity dynamics in heavily managed landscapes is still unclear. Here we used two complementary approaches to disentangle the role of environment and spatial effect in farmland bird community assembly in an intensive agro-ecosystem. We hypothesized that the interaction between habitat patches and dispersal should play a major role in such unstable and unpredictable environments. First, we used a metacommunity patterns analysis to characterize species co-occurrences and identify the main drivers of community assembly; secondly, variation partitioning was used to disentangle environmental and geographical factors (such as dispersal limitation) on community structure and composition. We used high spatial resolution data on bird community structure and composition distributed among 260 plots in an agricultural landscape. Species were partitioned into functional classes, and point count stations were classified according to landscape characteristics before applying metacommunity and partitioning analyses within each. Overall we could explain around 20% of the variance in species composition in our system, revealing that stochasticity remains very important at this scale. However, this proportion varies depending on the scale of analysis, and reveals potentially important contributions of environmental filtering and dispersal. These conclusions are further reinforced when the analysis was deconstructed by bird functional classes or by landscape habitat classes, underlining trait-related filters, thus reinforcing the idea that wooded areas in these agroecosystems may represent important sources for a specific group of bird species. Our analysis shows that deconstructing the species assemblages into separate functional groups and types of landscapes, along with a combination of analysis strategies, can help in understanding the mechanisms driving community assembly.

Abstract licence: CC BY

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.