Understanding the evolutionary processes that shape the landscape of genetic variation and influence the response of species to future climate change is critical for biodiversity conservation. Here, we sampled 27 populations across the distribution range of a dominant forest tree, Quercus acutissima, in East Asia, and applied genome-wide analyses to track the evolutionary history and predict the fate of populations under future climate. We found two genetic groups (East and West) in Q. acutissima that diverged during the Pliocene. We also found a heterogeneous landscape of genomic variation in this species, which may have been shaped by population demography and linked selections. Using genotype-environment association analyses, we identified climate-associated SNPs in a diverse set of genes and functional categories, indicating a model of polygenic adaptation in Q. acutissima. We further estimated three genetic offset metrics to quantify genomic vulnerability of this species to climate change due to the complex interplay between local adaptation and migration. We found that marginal populations are under higher risk of local extinction because of future climate change, and may not be able to track suitable habitats to maintain the gene-environment relationships observed under the current climate. We also detected higher reverse genetic offsets in northern China, indicating that genetic variation currently present in the whole range of Q. acutissima may not adapt to future climate conditions in this area. Overall, this study illustrates how evolutionary processes have shaped the landscape of genomic variation, and provides a comprehensive genome-wide view of climate maladaptation in Q. acutissima.

Fresh leaves were collected from mature trees located at least 50 m apart in each population. Genomic DNA was extracted from the silica gel-dried leaves using the Plant DNA Kit (Bioteke, Beijing, China) and sequenced on the Illumina NovaSeq 6000 platform with a target coverage of 30×. A total of 84 individuals were sequenced in this study, while the remaining 33 individuals were sequenced in a previous study (Liang et al., 2021).  Raw sequence reads were trimmed using Trimmomatic v0.39 and mapped to the Quercus robur reference genome assembly using BWA v0.7.15. Subsequently, genotypes were called using HaplotypeCaller implemented in GATK v4.1 (DePristo et al., 2011), and variants at sites with mapping quality ≥30 and base quality ≥30 were called using HaplotypeCaller. Next, strict criteria were applied to reduce the number of low-quality genotype calls, and 22 interspecific hybrids were discarded. We called all available sites and filtered the sites according to the request of specific analyses.