Past climate‐driven range shifts and population genetic diversity in arctic plants

Abstract

AbstractAimHigh intra‐specific genetic diversity is necessary for species adaptation to novel environments under climate change, but species tracking suitable conditions are losing alleles through successive founder events during range shift. Here, we investigated the relationship between range shift since the Last Glacial Maximum (LGM) and extant population genetic diversity across multiple plant species to understand variability in species responses.LocationThe circumpolar Arctic and northern temperate alpine ranges.MethodsWe estimated the climatic niches of 30 cold‐adapted plant species using range maps coupled with species distribution models and hindcasted species suitable areas to reconstructions of the mid‐Holocene and LGM climates. We computed the species‐specific migration distances from the species glacial refugia to their current distribution and correlated distances to extant genetic diversity in 1295 populations. Differential responses among species were related to life‐history traits.ResultsWe found a negative association between inferred migration distances from refugia and genetic diversities in 25 species, but only 11 had statistically significant negative slopes. The relationships between inferred distance and population genetic diversity were steeper for insect‐pollinated species than wind‐pollinated species, but the difference among pollination system was marginally independent from phylogenetic autocorrelation.Main conclusionThe relationships between inferred migration distances and genetic diversities in 11 species, independent from current isolation, indicate that past range shifts were associated with a genetic bottleneck effect with an average of 21% loss of genetic diversity per 1000 km−1. In contrast, the absence of relationship in many species also indicates that the response is species specific and may be modulated by plant pollination strategies or result from more complex historical contingencies than those modelled here.