Using differences in abiotic factors between seed origin and common garden sites to predict performance of Pacific madrone (Arbutus menziesii Pursh)

Abstract Climate shifts and concomitant weather extremes such as extended droughts make it difficult to determine the best seed sources for reforestation. Seed source selection based on current climate conditions avoids further disrupting biotic interactions that may confer ecosystem resilience, but seed source selection could also draw from populations adapted to anticipated future climate conditions. Knowledge of relationships between seed sources and adaptation to climatic variables helps evaluate options. This study used four common gardens of Pacific madrone (Arbutus menziesii Pursh) to model tree performance in height growth, dieback, mortality, and, at one site, phenology, using differences in abiotic variables from the tree’s geographic origins and the planting location. While the degree of variation explained by even the best models was relatively low, there were some trends. Patterns of growth and mortality differed across common garden sites, but the distance seeds were moved to the common garden site, seed source ecoregion, and the difference in the length of frost-free period between a seed source’s site of origin and the common garden site were in the best models for growth and mortality. Differences in mean summer precipitation, an indication of potential drought stress, were only in the best models for growth at the wettest and driest common garden sites. Trees more local to a common garden site generally had higher growth, and sometimes lower mortality. However, some Willamette Valley trees not local to a common garden site had relatively greater growth and low mortality so could be good candidates for broader seed movement. The abiotic factors that best explain growth and mortality were similar in less stressful and more stressful years, although models for stress years included more abiotic variables. Abiotic variables were better at explaining timing of spring leaf flush than the performance metrics, and spring flush did not correlate well to growth, dieback, or mortality. Our results support the importance of abiotic factors in predicting growth, mortality, and phenology in Pacific madrone, but their predictive capacity and relative importance depends on the site and may vary even for nearby sites.