Reevaluation of taxa included in genus Onobrychis using genomic and phenomic approaches

The genus Onobrychis is one of the most taxonomically complex genera of the Fabaceae family. Uncertainties in the number of species and high polymorphism in morphological characters make it necessary to re-evaluate this genus from a systematically. In this study, a multi-faceted approach was adopted to elucidate the taxonomic relationships of the genus Onobrychis. Firstly, the morphological diversity of the genus was measured manually and by digital image analysis on leaflet and fruit characters. Twelve traits were determined for the leaflets and 15 for the fruits. With the phenotypic data obtained, Principal Component Analyses (PCA) and cluster analyzes were performed and dendrograms were drawn. In addition, the assessment of species with common measurements from the phenotypic data obtained were classified using machine learning algorithms with an aim to reveal the distinctions between species more reliably. For molecular level evaluation, genetic diversity and taxonomic relationships were analyzed using Inter-Primer Binding Site (iPBS) markers. PCA analysis was performed by calculating the genetic distance from the data obtained by iPBS analysis and it was aimed to determine the taxonomic relationship by drawing a dendrogram. The iPBS marker results indicated that there was high polymorphism among the taxa included the genus and marker profiling could not clearly distinguish subgenera and sections. Leaf morphological analyses revealed that Onobrychis species largely reflect taxonomic distinctions at the subgenus level, but there is high morphological diversity in Sisyrosema subspecies. PCA and clustering analyses based on fruit morphology largely reflected subgenus-level distinctions in the genus Onobrychis, but higher morphological diversity was also detected, especially in the subgenus Sisyrosema. Machine learning analyses have shown that fruit and leaf morphological data have high differenciation capacity among Onobrychis species. Classifications made with Random Forest (RF), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN) algorithms achieved accuracy rates of up to 97%, demonstrating that classical morphological data carry a strong taxonomic signal when supported by modern computational methods. In conclusion, phenotypic analyses based on leaf and fruit morphology and molecular data obtained using iPBS markers revealed that taxonomic relationships can be largely reflected in the genus Onobrychis. Machine learning analyses have shown that morphological data have high differenciation power among species and that classical systematic approaches could be more reliable when supported by modern computational methods. However, it is thought that chloroplast and mitochondrial genome analyses, as well as further phenomic analyses could be employed to accurately reveal the boundaries at the subgenus.