Beyond range size: drivers of species' geographic range structure in European plants

AimTo assess if and how species range size relates to range structure, if the observed geographic range properties can be retrieved from predicted maps based on species distribution modeling, and whether range properties are predictable from biogeophysical factors. LocationEurope Time periodCurrent Major taxa studied813 vascular plant species endemic to Europe MethodsWe quantified the size and spatial structure of species geographic ranges and compared ranges currently occupied with those predicted by species distribution models (SDMs). SDMs were constructed using complete occurrence data from the Atlas Florae Europaeae and climatic, soil and topographic predictors. We used landscape metrics to characterize range size, range division and patch shape structure, and analysed the phylogenetic, geographic and ecological drivers of species range size and structure using phylogenetic generalized least squares (pGLS). ResultsRange structure metrics were mostly decoupled from species range size. We found large differences in range metrics between observed and predicted ranges, in particular for species with intermediate observed range size and occupied area, and species with low and high observed patch size distribution, geographic range filling, patch shape complexity and geographic range fractality. Elevation heterogeneity, proximity to continental coasts, Southerly or Easterly geographic range positions and narrow ecological niche breadth constrained species observed range size and range structure to different extents. The strength and direction of the relationships differed between observed and predicted ranges. Main conclusionsSeveral range structure metrics, in addition to range size, are needed to adequately describe and understand species ranges. Species range structure can be well explained by geophysical factors and species niche width, albeit not consistently for observed and predicted ranges. As range structure can have important ecological and evolutionary consequences, we highlight the need to develop better predictive models of range structure than provided by current SDMs, and we identify the kinds of species for which this is most necessary.