Geographical variation in the distributional constraints along a gradient of population aggregation

Factors determining species distributions have frequently been shown to vary geographically, yielding spatially variable species–environment relationships when developing species distribution models. Using occurrence data for four bird species, we analysed to what extent position within the distribution range along a gradient of population aggregation determined geographical variation of distributional constraints. We built generalised linear models for the whole dataset (global models) and separately for each region within the species range with decreasing population aggregation (core, middle and peripheral). For both global and regional models, we compared species responses to habitat conditions showing an overall higher model fit in peripheral regions than in middle and core regions. Better model fit in peripheral regions was also found after including a spatial factor (i.e. an autocovariate estimating the proportion between the presences and total observations within a 10 km radius) into the global and regional models. attered availability of suitable habitat patches in a predominantly hostile landscape matrix seems to be the limiting factor for species distributions in peripheral regions. Conversely, the larger number of spatially dependent occurrences in core and middle region may favour source and sink population dynamics, yielding poorer relationships between habitat conditions and species occurrence. The role of the spatial factor after removing the co-variation with habitat conditions separately for each region suggests that endogenous processes such as dispersal vary throughout the species range but differently for each species. Geographical variation in distributional constraints can be interpreted as a consequence of the inherent spatial character of ecological processes and their interaction with varying environmental conditions throughout the species range. Overlooking the effects of spatial variation in distributional constraints may lead to erroneous management conclusions and inadequate estimation of the species’ response to environmental changes.