Trait analysis
We analyzed five traits for phenotypic differentiation: fall bud set, spring bud flush, specific leaf area (SLA), height, and trunk basal diameter. Phenology of bud set and bud flush were measured in the fall of 2015 and the spring of 2016, respectively. Bud set was recorded as the initiation of bud formation, where internode elongation had ceased and the newly emerged, bundled leaves were clustered at the same level on the stem, offset from the shoot axis (Frewen et al. 2000). Bud set was measured at 6-10 day intervals from September through December of 2015 on three replicates of all 12 genotypes per population in each garden. We scored trees based on the bud stage exhibited by 50% or more of the apical meristems. There was little within-plant variation in apical bud development, so we felt this was a good approximation of whole plant progression towards dormancy. Spring bud flush was recorded as the date when all leaves had flushed from the tree, specifically when a leaf had expanded from all buds on the branch. Bud flush was measured every two weeks from February through the end of April in the Yuma and Agua Fria gardens, and through the end of May in the Canyonlands garden, where colder temperatures persist later into the spring. Bud set, bud flush, and height were assessed the full first year of growth (2015-2016) for every genotype in each of the common gardens.
Specific leaf area was measured using the average of three to six fully expanded leaves that were free of or had minimal herbivore damage (if no leaves without damage could be found). To standardize for light availability, all leaves were collected from a single south-facing branch, collected approximately at breast height in May and June of 2020. Due to mortality in the gardens over the five years since planting, and the time required for sampling, SLA was only measured on 12 populations in Yuma and Agua Fria, and seven populations in Canyonlands, with 3-7 genotypes measured within each population. Although early SLA analysis from 2015 showed similar patterns, concern over residual maternal and greenhouse effects cautioned against using this first year of data. Leaves were scanned and leaf area was measured using ImageJ software (Schneider et al . 2012). After the area scans, leaves were dried using silica beads and then weighed. Lastly, we used trunk basal diameter recorded at the end of the fourth year of growth (2018), which was measured as the diameter at root collar (DRC), ~10cm from the soil, on every live tree in the gardens. Diameter at root collar was used instead of diameter at breast height because it allows us to track tree growth consistently from planting up to their current stature.