Molecular methods and tools for managing and assessing genetic diversity in rose germplasm collections

Polyploidy, extensive hybridization, and centuries of human selection have made roses one of the most genetically complex ornamental crops. This dissertation addresses the challenge of dissecting genetic diversity in polyploid roses through an integrative framework that connects genebank management, molecular characterization, and breeding applications. In the first results chapter, a robust, genome-wide distributed SNP marker set optimized for tetraploid cultivars was developed based on SNP chip data. The marker set was subsequently converted into PACE markers and used to screen more than 4,000 rose accessions, followed by further validation of marker performance. These markers enable reliable genotyping, cultivar verification, and the detection of redundancies within genebank collections, thereby improving data quality and curation practices. Building on this molecular foundation, the second study employed both unsupervised and supervised machine learning approaches on a subset of the characterized accessions to reveal genetic clusters that extend beyond traditional horticultural classifications. These data-driven groupings provide a refined view of population structure and might be a solid basis for constructing representative subsets for downstream analyses such as selective sweep detection and genome-wide association studies (GWAS). The third results chapter focused on identifying genomic regions under selection, shedding light on how breeding has shaped the rose genome. Using selective sweep analyses based on SNP chip data, regions enriched for GO terms and known genes were identified. While limited by SNP density and the absence of haplotype-level data, these analyses revealed clear signatures of selection related to ornamental traits such as recurrent flowering. In the fourth manuscript, already known markers for specific traits were improved, and potential regions associated with new traits were identified through GWAS. In contrast to previous studies, the association panel was expanded to 285 genotypes. Significant markers were identified for six traits (shoot anthocyanin, prickle number, leaf glossiness, petal number and length, fragrance) and subsequently validated using PACE assays in up to 192 genotypes, confirming the robustness of the associations. In conclusion, this work highlights how molecular marker approaches can deepen our understanding of genetic diversity in polyploid crops and lay the foundation for more data-driven and efficient rose breeding.