Morphology-based genetic diversity analysis reveals introgressive hybridizations obscure species boundaries of three wild roses endemic to Taiwan Island

Interspecific hybridization in the genus Rosa (Rosaceae) is a common natural phenomenon. Hybrids often exhibit heterosis and new combinations of traits, which can provide raw materials for horticultural breeding. DNA barcodes and microsatellites have been proposed to facilitate species discrimination and hybrid detection. However, most SSR markers developed for roses have been found unapplicable to Rosa sect. Synstylae because of null alleles or failed amplification. In this study, we designed 15 pairs of microsatellite primers, along with four previously developed primers specifically for Rosa sect. Synstylae; we then analyzed 174 individuals of three closely related and sympatrically distributed Rosa species as a test case to evaluate the consistency between morphological and genetic hybrid identifications and to compare the discrimination efficiency of the DNA barcodes versus SSRs in detecting admixture. Principle coordinate analysis identified several individuals with intermediate phenotypes among the three rose species. Hybridization, intraspecific morphological polymorphism, and sample collection at different growth stages or phenological phases may have hindered species identification based on morphology and distorted the morphological clustering results. The molecular analyses showed that 12 (6.8%), 13 (7.4%), and 15 (8.6%) individuals were identified as admixed by STRUCTURE, NewHybrids, and nrITS sequences, respectively, of which only seven hybrids showed signs of admixture across all three methods. About 81% of the morphologically identifiable hybrids exhibited admixture based on SSRs. Meanwhile, approximately 69% of morphologically identifiable hybrids were detected, but four morphologically pure species individuals were identified as genetically admixed based on nrITS sequences. Some morphologically pure species individuals were genetically identified as hybrids while some morphological hybrids were identified as pure individuals based on certain molecular markers. Overall, EST-SSRs discriminated morphological hybrids more accurately than nrITS. We inferred that there is ongoing interspecific gene exchange among the three wild Rosa species that obscures morphospecies boundaries. Combining multiple data types and analytical approaches offers powerful utility for hybrid detection, regardless of the level of hybridization.