Two cycles of recurrent selection for seed yield in common bean

Seed yield of common bean (Phaseolus vulgaris L.) can be improved by exploi ting the genetic variability available in the crop's cultivated races and g ene pools. The objectives of this study were to (i) compare yield gains obt ained through two cycles of recurrent selection, based on S-1 family tests, in two types of populations [interracial Middle American (MA) and inter-ge ne-pool Andean x Middle American (AMA)], and (ii) estimate genetic variance , heritability, and expected gains from selection. The MA population compri sed 11 parents from race Mesoamerica, three from Jalisco (Mexico), and two from Durango (Mexico). The AMA population had six parents from race Mesoame rica, one from Jalisco, five from Nueva Granada (northern Andes), and two f rom Chile (also Andean). From the initial cycle (C-0) of each population, 4 5 S-2 families were evaluated at three locations in Colombia in 1992 to sel ect the 10 highest yielding families to be intermated in a diallel fashion (excluding reciprocals) to generate 45 S-1 families for cycle 1(C-1). Simil arly, 45 S-1 families from C-1 and cycle 2 (C-2) were evaluated in 1993 and 1995, respectively. The 10 highest yielding selected families from each of C-0, C-1, and C-2 and a corresponding bulk of all 45 S-1 families from eac h of the three cycles were evaluated at three sites in 1997. Data were reco rded for seed yield, seed weight, and maturity. Genetic variance among 45 S -1 families in C-1 and C-2 was greater in the AMA population than in the MA , diminishing from C-1 to C-2 in MA and increasing in AMA. Heritability for seed yield ranged from 0.31 to 0.46, for seed weight from 0.75 to 0.86, an d for maturity from 0.50 to 0.81. Expected gains from selection for seed yi eld was lower (3-4%) in C-2 compared with C-1 (7.9-13.2.%). The mean yields of C-0, C-1, and C-2 of the MA were significantly (P < 0.05) higher than t hose of the AMA. The C-2 of both MA and AMA out yielded their respective C- 1 and C-0 families. The average realized yield gain per selection cycle was 15% or higher in both populations. These results justify the use of recurr ent selection in interracial and inter-gene-pool populations for improving seed yield of common bean.