Germplasm collections of cultivated plants constitute the source for further genetic progress. These collections gained further interest with approaches for tracking allelic variants associated to phenotypic variations within core collections. In order to explore the structure of genetic variation in pepper (Capsicum spp.) and to select core-collections maximizing the genetic and the phenotypic diversity, a pepper collection including 1,352 non redundant accessions from 11 Capsicum species from 89 different countries was genotyped using 28 microsatellite (SSR) markers spanning the whole genome. Model-based analysis structured the collection into 6 clusters, with 3 clusters separating the main species complexes, including cultivated species and wild relatives, according to taxonomic classification (C. frutescens/C. chinense, C. baccatum, C. pubescens), and 3 additional clusters for C. annuum. The relationships between the cultivated C. annuum species and the wild relative (C. annnuum var. glabriusculum) was precised. The 3 C. annuum clusters were significantly distinct for plant and fruit descriptors corresponding to cultivar types, showing that the genetic structure of this cultivated species was strongly impacted by the long term human selection of landraces in primary as well as secondary diversification centres. We settled nested core-collections of 8, 16, 32, 64 and 128 C. annuum accessions capturing from 37 to 90 % of the genetic diversity for further sequencing efforts and establishment of high throughput genotyping assays. By compiling phenotypic and genotypic data, a larger core-collection of 332 accessions was established, capturing 97 % of the C. annuum genetic and phenotypic diversity for further genetic association studies.
