Environmental Variation

For each planting, environmental conditions were characterised based on the local temperature profile. Temperature is known to be an important environmental cue regulating Arabidopsis development (Granier et al., 2002; Springate & Kover, 2014) and warming is a primary consequence of climate change. On-site ground temperature was initially modelled at hourly resolution for 203 days following field transplantation using daily minimum and maximum temperature (the maximum number of days of data available for all plantings; temperature modelling as described in Wilczek et al. (2010), providing a high-resolution description of temperature conditions. However, lowering the resolution to daily minimum and maximum temperature yielded equally accurate trait predictions (Appendix S3). We adopted this resolution in our model because this is the typical resolution available for historical records and future projections (Cornes et al., 2018; Gent et al., 2011; Thornton et al., 2016). We used all 203 days of data to describe environmental variation, resulting in 406 (203 x 2) environmental predictors.