Conclusions
Our results show that populations are significantly differentiated with respect to growth, morphology, and phenology traits, supporting a multi-trait hypothesis of divergent selection (QST> FST). Phenology traits were the most differentiated among populations with the largest QSTvalues, while SLA and height supported the divergent selection hypothesis in some gardens, but not others. In contrast, basal trunk diameter was the only trait to show evidence of stabilizing selection and only in the coldest common garden (Fig. 4). In addition, we found source climate is significantly correlated with trait differences across the gardens, suggesting the large climatic gradient experienced by these Arizona populations is an agent of selection. Interestingly, the magnitude of trait variation detected among populations depended, in part, on their growing environment. We found most traits had the greatest population differences with highest QST values in the warmest garden and declined as the trees were planted in cooler environments. Specific leaf area was the only trait measured with the opposite response of higher population differentiation in the cold garden (Fig. 4). Populations exhibited local adaptation in growth and phenology traits, with many populations growing largest in the gardens that most closely matched their home climates. This study demonstrates that experimental common gardens simulating climate change, across even a portion of a species range, can have a substantial impact on how important functional traits are differentially expressed among populations. The gradient of climate-driven selection may lead to the identification of a geographic mosaic of local adaptation that may also cascade to affect associated species and communities (e.g., Thompsonet al . 2005; Smith et al. 2011; Wooley et al.2020). Importantly, we found that the detection of past selection on population-level trait differences, as measured by QST-FST analysis, is modified by growing environment. This finding suggests past climate can interact with the current and future climates to affect population responses. Strategies for management of widespread species like Fremont cottonwood would benefit from considering the climatic selection pressures of source locations to anticipate their performance under changing environmental conditions. Acknowledgements
This research was supported by NSF-IGERT and NSF GK-12 Fellowships (HF Cooper), NSF Bridging Ecology and Evolution grant DEB‐1914433 (RJ Best, GJ Allan, R Lindroth, TG Whitham), NSF MacroSystems grant DEB-1340852 (GJ Allan, TG Whitham, CG Gehring, & KC Grady), NSF Macrosystems grant DEB-134056 (KR Hultine), NSF DBI-1126840 (TG Whitham), which established the Southwest Experimental Garden Array. We thank our agency partners for helping to facilitate use of the common gardens: Dana Warnecke and Kelly Wolf at Arizona Game and Fish (Agua Fria), Erica Stewart at the Bureau of Land Management (Yuma), and Barry Bakker, Phil Adams, and the Redd family at The Nature Conservancy’s Canyonlands Research Center at Dugout Ranch. We acknowledge Christopher Updike, Zachary Ventrella, Davis Blasini, Dan Koepke, and Matthew McEttrick, along with many volunteers for help establishing and maintaining the common gardens. We thank Helen Bothwell for her help developing and troubleshooting the SNP library. Lastly, thanks to Jacob Cowan, Michelle Hockenbury, Teresa Reyes, Michelle Bem, and Jackie Parker for assistance with data collection in the field, and the Cottonwood Ecology and Community Genetics Lab for their constructive comments and reviews.