Development and evaluation of common-garden-based and genomic-based ecological models for climate change adaptation in interior spruce

Forests, as a fundamental part of ecosystems, are increasingly challenged by rapid climate change, raising critical questions about the accuracy of ecological models used to guide forest practices. This dissertation evaluates, applies, and advances ecological modeling frameworks used to characterize local adaptation and predict population responses in forest tree species under climate change, using the interior spruce hybrid complex in western Canada as the model group. First, I assessed the reliability of climate data sources commonly used in genecology research by comparing on-site micro station observations and ClimateNA, a scale-free climate data source, with standard weather stations. Using linear regression and response functions based on 126 provenances planted at 17 sites, I quantified differences in predictive performance and selected a data source for downstream ecological modeling. Results demonstrated that ClimateNA is broadly useful, unless high-quality precipitation data are required. Second, I applied common-garden-based Universal Response Functions to model population-level growth responses using the same populations. I established a systematic framework for model evaluation, including spatial validation with independent productivity data. The optimized model was further used to examine local adaptation patterns and predict optimal seed transfer under contemporary and future climates. Third, I employed landscape genomics to investigate climate adaptation using Gradient Forest modeling with 41,253 SNPs genotyped in 1,684 natural-origin interior spruce individuals from 252 open-pollinated populations. Winter temperature and moisture-related variables were identified as potential key drivers of genomic variation. I further delineated 11 seed zones based on genomic variation and estimated genetic offsets under future climate scenarios, revealing potential maladaptation risks in British Columbia. Finally, I developed a genomic-informed modeling framework by incorporating genomic-informed predictors derived from Gradient Forest analyses into an extended Genomic-informed Universal Response Function. By replacing provenance climate variables with genomic-informed variables, this approach relaxed assumptions of linear climate-genetic relationships in provenances and improved representation of adaptive genetic structure, which provided refined predictions of adaptive potential and seed sources under future climates. Together, these studies demonstrate a methodological progression from climate data validation to genomic-informed adaptive modeling. This work contributes to improving predictive accuracy in forest management in the context of anthropogenic climate change.