Integrative morphological and genomic analyses reveal diversity, reticulate evolution, and adaptation in diploid and tetraploid Rosa species from Xinjiang
Xinjiang, with extensive mountain systems and desert basins, forms a major biogeographic corridor between East and Central Asia and harbors taxonomically challenging wild Rosa species. We analyzed 252 accessions (159 diploids, 93 tetraploids) and integrated morphometrics, ploidy estimates, and whole-genome resequencing. PCA of 14 quantitative traits revealed extensive overlap, especially between diploid R. beggeriana and tetraploid R. laxa. In contrast, mixed-trait clustering based on Gower distance was consistent with the genomic backbone, while Random Forest and linear discriminant analysis (LDA) identified ten diagnostic traits that improved discrimination (86.2% accuracy). This integrated framework helped delineate lineage boundaries and attribute persistent phenotypic overlap to reticulate evolution. Diploid resequencing yielded 4.77 million SNPs and revealed deep lineage structure and pervasive introgression, including a multi-species admixed lineage with mosaic ancestry. Genome scans and trait–environment associations highlighted stress-related candidate loci (including signals overlapping DRS1 and DCL4). They also linked dense pedicel hair to hot–arid habitats in an arid lineage of R. beggeriana, pending functional validation. Phylogenies inferred from 240 plastomes and 5,641 single-copy nuclear genes revealed strong cytonuclear discordance and gene-tree discordance, consistent with rapid divergence and incomplete lineage sorting and further shaped by introgression and plastid capture. For tetraploids, integrative evidence supported recent autopolyploidization and ongoing gene exchange, although short-read data limited dosage-aware inference. Overall, our results clarify reticulate evolutionary histories of Xinjiang roses and provide a lineage-informed basis for conservation and germplasm utilization.