Climate data
We obtained 30-year (1921–1950) averages of climate data for each genotype source location from the 270 m resolution California Basin Characterization Model (BCM) (Flint et al. 2013). These mid-20th-century values were used instead of more recent climate data because they more closely resemble the conditions when the genotypes were establishing as seedlings.
For the GEA, we chose to focus on raw environmental variables rather than environmental PCA axes, as several previous studies have done (Eckert et al. 2010, 2015). This is because PCA associations can be challenging to interpret if, for example, the axes include both temperature and moisture variables. Instead, we used PCA (Fig. S4) to select five environmental variables that have low correlation with one another across tree source locations: mean climatic water deficit (CWD, a measure of evaporative demand exceeding soil moisture); mean minimum winter (December-February) temperature (TMIN); mean maximum summer (June - August) temperature of summer (TMAX); mean monthly winter precipitation (PPTW); and mean April 1st snowpack (PCK4). Other climate variables considered but not included in the analysis were actual evapotranspiration (AET), potential evapotranspiration (PET), excess water, recharge, runoff, snowfall, snowmelt, soil water storage, and snow sublimation.