Genetic Support for Phenotype-based Racial Classification in Sorghum

The classification of domesticated sorghum [Sorghum bicolor (L.) Moench] into five major races is based on phenotypic characteristics of the spikelet and panicle. Presumably, this classification also reflects genetic relatedness. In this study, we apply both model-based (STRUCTURE) and model-free (principal components analysis) approaches to a genetic data set of 434 single nucleotide polymorphisms and simple sequence repeat alleles in a panel of 216 exotic sorghum lines to study the congruence between genetic relatedness and phenotype-based racial classification. Both model-based and model-free approaches clearly identify genetic groups that correspond closely to the caudatum, durra, kafir, and guinea races of sorghum. Genetic groups differ from racial groups in several significant ways, including the classification of the West African Kaura group with guineas rather than with caudatums, and the identification of a novel genetic group affiliated with both kafirs and guineas. These results highlight that high-throughput genotyping now provides curators with an additional, critical tool to better characterize and quantify diversity found in collections. Because of technological advances and associated cost reductions of unit assays, high-throughput genotyping is now applicable to all crop species.