Predicted climate response and potential applications

We predict temporally and spatially heterogeneous responses to climate change in the coming decades. Despite high interannual variability, the general trends were consistent with previous findings (Cook et al., 2012; Daele et al., 2012; E. Post et al., 2018; Wu et al., 2017; Yu et al., 2010; X. Zhang et al., 2007). In Scandinavia, the predicted delay in DTB likely reflects the disruptive effect of winter warming on vernalization (Cook et al., 2012; Wu et al., 2017; X. Zhang et al., 2007). Delayed flowering would increase the period of vegetative growth, causing a downstream increase in reproductive output (Choe et al., 2001; Daele et al., 2012; Tienderen et al., 1996). Conversely, earlier bolting predicted for spring and summer cohorts in Central Europe and South Mediterranean were consistent with expectations from thermal time models of accelerated development at higher temperatures (Chew et al., 2012; Wilczek et al., 2009). The corresponding decline in fecundity may reflect a shortened period of growth and decreased flower production (Scheepens & Stöcklin, 2013), although this is unclear because accelerated development has also been suggested to increase reproductive output (Cook et al., 2012).
While we could not infer the biological causes underlying our predictions, they clearly show a breakdown of current local adaptation and increase in genetic offset caused by climate change. For the majority of Europe, our predictions suggest local provenancing is less effective in the long-term since fecundity is predicted to decline by 2099. Indeed, we identified a source of genetic variation that could help establish climate-resilient populations of A. thaliana from an unexpected origin: Spain. The Spanish genotype Ll-2 had higher predicted fecundity than local genotypes throughout parts of Europe well beyond its native range, being potentially suitable for boosting climate resilience in locations as disparate as the Balkans, Finland, and Northwestern Russia. Importantly, our findings are supported by empirical reports that southern genotypes outperformed local genotypes across Europe (Wilczek et al., 2014) and field trials suggesting northern Spanish genotypes like Ll-2 are better-equipped to survive future warming (Exposito-Alonso et al., 2018). Our example clearly demonstrates the value of moving beyond local provenancing as a strategy for sourcing seeds.
Nevertheless, our findings should not be taken as an absolute sign of future maladaptation. In this study, the predicted germination date was determined from a small number of plantings and assumed to remain constant between years. In reality, germination timing is another climate-responsive trait that relies on environmental cues (Finch‐Savage & Leubner‐Metzger, 2006) and would likely vary between years. Sensitivity to germination cues can also differ between genotypes, adding another layer of complexity to germination timing (Martínez-Berdeja et al., 2020). This would require field experiments to train the model, although in silico tests on the effect of varying germination date using AraCast indicate climate response can be seasonally and geographically differentiated. Such predictions suggest the potential for A. thaliana to shift germination time in response to climate change, a phenomenon that has been observed in alpine species (Mondoni et al., 2012). If this seasonal shift occurs, the species has the potential to avoid maladaptation and persist in the face of climate change.