Elevational rear edges shifted at least as much as leading edges over the last century

Aim Range shifts along elevational gradients are considered a major response of mountain species to climate change. However, empirical studies have so far mainly focused on leading edges or on species' optima, and evidence of rear edge shifts remains scarce. Yet, the balance between leading and rear edge shifts has important consequences for conservation and co-determines species' extinction risk. Here, we present a comparative synthesis of range dynamics observed at both range limits. Location Global. Time period 1850-present. Major taxa studied Plants, invertebrates, vertebrates. Methods From the literature, we compiled elevational leading and rear edge shifts of 1,026 species observed at the same localities over the same time period. We used linear mixed-effects models to analyse whether both range limits shifted upslope, whether leading edges shifted faster than rear edges and elevational range sizes have thus changed, whether observed shifts were linked to temperature changes, and whether shifts lagged behind temperature changes. Results Despite pronounced species-specific variation, both range limits shifted upslope on average. Rates of shift did not differ between rear and leading edges, elevational range sizes thus did not change. Regional differences in temperature trends were only related to dynamics at rear edges. Yet, the stronger climate warmed regionally, the more species' responses lagged behind expectations at both range limits. Main conclusions Our results demonstrate that extinctions at rear edges of mountain species have at least been as common as colonizations at leading edges. The drivers of observed range limit shifts are not deducible from our data, but weak relationships with temperature trends suggest that other factors than climate warming played an additional role. These results do not relax concerns about possible detrimental effects of environmental change on mountain biodiversity and point to the importance of refocusing monitoring towards a better representation of rear edge dynamics