Aim: Global climate change has been linked to shifts in species' geographic and elevational distributions, with taxa varying in responsiveness. This variation may be due to a time lag of the response or climate alone not being a simple determinant of distribution limits. To tease apart the role of climate in distribution, we compared the temperature response of predicted occurrence revealed by ecological niche modelling (ENM) on historical climate with that of performance in a multi-population transplant experiment. Congruence would support that climate is a main driver of distribution limits of a species.

Location: North America.

Taxon: Arabidopsis lyrata subsp. lyrata.

Methods: Seeds of 20 populations of North American Arabidopsis lyrata from across the entire range were collected, propagated and then sown along a latitudinal transect across and beyond the species' range. Lifetime performance was related to niche-and range-determining climatic variable revealed by ENM.

Results: Lifetime performance did not consistently decline towards the high-latitude range limit, but it did so towards the low-latitude range limit. This decline was slightly weaker for low-latitude populations, indicating divergent climate adaptation. The overall performance curve on the field-measured minimum temperature in early spring was fairly congruent with the curve of ENM-predicted suitability on this important niche-determining variable. ENM-based projections revealed that the southernmost populations were vulnerable under climate warming.

Main Conclusions: Results verified that ENM based on species occurrences can well-predict plant performance under field conditions. Congruence teaches us that with the climate change so far, the species exhibits a colonisation deficit in the north. Furthermore, the southernmost populations are vulnerable to extinction. A likely outcome is the shrinking of the species' range.

The Arabidopsis lyrata occurrence data set was compiled with data from Lee-Yaw et al. (2018, Ecography, doi: 10.1111/ecog.02869), the GBIF database with occurrence data entries after 1960 (GBIF.org, September 14, 2022), and various herbaria collections. The thinned data set contains 433 occurrences/entries, each with information on: sample identity (point_ID), latitude, longitude, and the source of information.