Integrative Taxonomy for Species Delimitation: A Case Study in Two Widely Accepted Yet Morphologically Confounding <i>Rosa</i> Species Within Sect. <i>Pimpinellifoliae</i> (Rosaceae)

The use of morphological traits as a practical approach for delimiting taxa at various ranks has long been regarded as a reliable basis for taxonomy. However, its efficacy has been increasingly called into question in many taxonomic groups due to its inherent limitations, such as failing to account for phenotypic plasticity, ecologically driven variation (e.g., ecotypes), and parallel evolution. These factors often introduce ambiguity or misleading similarities, thereby obscuring the true evolutionary relationships among taxa, particularly in the context of species delimitation. In the present study, we employ an integrated methodology that combines quantitative morphological analyses, whole-genome data, and ecological measurements to resolve the species boundaries of two morphologically similar roses, Rosa sericea and Rosa hugonis, which have long been considered as two distinct species but lack clear morphological boundaries. Our findings reveal that the unbiased analysis of morphological data based on a large and representative sample size was insufficient to identify effective diagnostic traits. However, when complemented with genome-wide population-level sequencing data or integrated with geographic and ecological niche assessments, the delineation of species boundaries was significantly improved. Furthermore, ecological data provide additional insight into the abiotic factors driving interspecific and intraspecific divergence. By integrating multiple lines of evidence-spanning genomic (intrinsic) and phenotypic (extrinsic) traits-and incorporating the interaction between species and their environments, species boundaries can be delineated with greater confidence. A well-defined species can thus be established through the mutual corroboration of diverse datasets, thereby ensuring a more rigorous and comprehensive taxonomic framework.