Early detection of plant community responses to climate warming along mountain roads

Global warming is causing species to shift their ranges towards higher latitudes and elevations, leading to a reassembly of plant communities and associated community thermophilization (i.e. an increasing number or cover of thermophilic species, sometimes at the expense of mesic or cold‐adapted species). Given the large variation typically observed in the magnitude and direction of range shifts, quantifying community thermophilization might provide a more sensitive method to detect climate change impacts within short time periods and across limited spatial extents, as it integrates range shifts across multiple species while also accounting for changes in abundance. Here, we combined an assessment of (i) species‐level range shifts and (ii) changes in community‐inferred temperatures (thermophilization) along three mountain roads in Switzerland to ask whether plant communities have responded to a warming climate over a 10‐year period, and whether community thermophilization is a sensitive metric for early detection of these changes. We found a community thermophilization signal of +0.13°C over the 10‐year study period based on presence‐absence data only. Despite significant upwards shifts of species' upper range limits in the lower part of the studied elevational gradient and a decrease in species richness at high elevations, significant thermophilization was not detectable if community‐inferred temperatures were weighted by species' covers. The low cover values of species that were gained or lost from local communities over the study period, together with their similar species‐specific temperatures to resident species, explained the discrepancy between the thermophilization detected in cover‐weighted versus unweighted models. Synthesis. Our work shows that plant species are rapidly shifting to higher elevations along roadsides in the western Swiss Alps and that this translates into a detectable warming signal in plant communities within 10 years. However, the species‐level range shifts and the community‐level warming effect are mostly based on gained/lost species with low cover values, preventing the detection of community thermophilization signals when incorporating cover changes. We therefore recommend including unweighted approaches as an additional option for early detection of community‐level responses to changing climate, ideally in combination with assessments of species‐level range shifts.