Niche differentiation between deeply divergent phylogenetic lineages of an endemic newt: implications for species distribution models

Abstract

Species distribution models (SDMs) treat species as a single unit, neglecting intraspecific variation. Few studies address the impact of intraspecific variation on SDM performance, and none of them account for the well-known inter-algorithm variability in prediction performance. The endemic Iberian amphibian Lissotriton boscai comprises two geographically highly structured phylogenetic lineages, which allowed us to explore how intraspecific variation affects the overall performance of SDMs and the predicted ecological niche. We built species and lineage distribution models using three different presence-only algorithms. We also tested for niche overlap, niche equivalency and niche similarity, using an ordination technique. We found differences in the predicted potential distribution of the two lineages and the underlying environmental factors. Moreover, intraspecific differences in model predictive capacity existed irrespective of which algorithm was used to build the distribution models. This was coupled with lineages showing a low degree of niche overlap and occurring in relatively different environmental niches spaces. The intraspecific variation observed in L. boscai led to an improved intraspecific predictivity of the lineage level based-distribution models. There was partial spatial agreement between the niche overlap and independently reported secondary contact zones. Thus SDMs built only at the species level may be too naive to predict impacts of global change on species distributions.