COMPARISON OF SELECTION METHODS FOR DRY BEAN POPULATIONS DERIVED FROMCROSSES BETWEEN GENE POOLS

New approaches are needed to broaden the genetic base and improve the yield potential of large-seeded ed beans (Phaseolus vulgaris L.). The objectives of this research were to compare the effectiveness of two r ecurrent selection methods to develop indeterminate red-seeded bean po pulations that yield more than determinate red-seeded varieties. The o riginal base population was produced from crosses between genotypes fr om the Andean and Middle American gene pools. The performance of the p opulations developed by recurrent selection was tested in field trials conducted over three growing seasons in Puerto Rico. Both recurrent s election with F-2 evaluation and recurrent selection with F-5 evaluati on were successful in developing tropically adapted indeterminate bean populations having seed type and seed size suitable for the Caribbean . Recurrent selection with F-5 evaluation produced a larger number of indeterminate populations with greater seed yields than the determinat e check cultivars. The greater seed yield of these indeterminate popul ations was attributed to longer reproductive periods and a greater num ber of days to physiological maturity. Heterosis for seed yield, the l arge amount of segregation for seed type and agronomic traits, and the lack of evidence of a linkage between small-seed size and erect, inde terminate plant architecture favor the use of recurrent selection with advanced generation evaluations to improve bean populations derived f rom different gene pools.