Decoding the genomic basis of adaptive capacity and vulnerability in the high-altitude Saussurea obvallata complex

Decoding how adaptation and vulnerability are distributed across rugged landscape is essential for anticipating biodiversity responses to climatic change. We investigated the Saussurea obvallata complex, a group of closely related lineages distributed across the Himalayan-Hengduan Mountains (HHM), to ask how climatic heterogeneity and historical isolation shape genomic variation, ecological divergence, and the speciation continuum. To address these questions, we integrated plastome and RAD-seq based nuclear SNPs with genotype-environment association (GEA) analyses, gradient forest (GF), generalized dissimilarity modeling (GDM), and ensemble species distribution models (SDMs). Projected genomic offset under future climate scenarios (2070) and ensemble SDMs were used to map genomic vulnerability and forecast habitat shifts. Nuclear SNPs resolve shallow divergence and cytonuclear discordance consistent with incomplete lineages sorting and episodic introgression. After formally accounting for background population structure and spatial effects, the pure environmental fraction remains modest but significant, with within-population label-permutation nulls falling well below observed values. Concordant signals across partial redundancy analysis, latent factor mixed model (notably June cloud metrics), and GF/GDM (temperature seasonality, slope/elevation, cloud regime) indicate a genuine environment-linked component of allele-frequency turnover, while acknowledging that reduced-representation data may under detect polygenic architecture. Genomic offset maps highlight vulnerability hotspots along the southern Western and Eastern Himalayas and parts of the Hengduan Mountains, whereas the central HHM shows lower offset and potential refugial stability. Together, results support a speciation continuum shaped by both isolation and environmental selection and motivate a dual conservation strategy: safeguard diversity-rich, low-offset refugia while mitigating risk in high-offset peripheral regions through enhanced connectivity, microrefugia, and genomic monitoring.