Low density marker-based effectiveness and efficiency of early-generation genomic selection relative to phenotype-based selection in dolichos bean (Lablab purpureus L. Sweet).

Genomic prediction has been demonstrated to be an efficient approach for the selection of candidates based on marker information in many crops. However, efforts to understand the efficiency of genomic selection over phenotype-based selection in understudied crops such as dolichos bean (Lablab purpureus L. Sweet) are limited. Our objectives were to (i) explore the effective marker density for achieving high prediction accuracy and (ii) assess the effectiveness and efficiency of genomic selection over phenotype-based selection on seed yield at early segregating generations in dolichos bean. In this study, the training population, which consisted of F recombinant inbreds, had a shared common parent with the breeding population, which consisted of F generation breeding population. The populations were genotyped with newly synthesized genomic simple sequence repeat-based markers. The effective marker density for genomic prediction was assessed by using a varying number of markers in predictions using 11 different models. Furthermore, the effectiveness of genomic selection was assessed by comparing the genetic gains in progenies between genotypes selected based on predicted seed yield and phenotypically selected genotypes. Our results indicate that low-density markers that are evenly distributed throughout the genome are sufficient for the integration of genomic selection in dolichos breeding programs. The genomic selection was proved to be two times more effective than phenotypic selection in early-generation selection in dolichos beans. The results have a significant impact on adopting genomic selection in regular breeding programs of Dolichos beans at a low cost.