Evaluation of locations for testing dry bean cultivars in western Canada using statistical procedures, biological interpretation and multiple traits

Abstract

Genotype × environment (GE) effects in regional registration trials for dry beans in western Canada were studied to determine whether geographic distribution of sites could be rationalized. The structure of the GE effects on the maturity, seed weight and seed yield of seven dry bean check cultivars grown at eight locations during 4 yr was investigated using GE mean squares decomposition, cluster analysis and the additive main effect and multiplicative interaction (AMMI) method. The analyses revealed a high level of redundancy in the locations which suggested that most GE effects can be captured with fewer testing sites. The partition of the GE mean squares demonstrated the possibility of reproducing the GE structure of the entire data sets with as few as three sites; however, more locations may be needed to compensate for unpredictable environmental effects. Based on biological interpretation of groupings and visual assessment of the AMMI displays, a five-location set fully represented the GE effects on maturity, seed weight and seed yield and accounted for the inconsistent clustering of the Brooks site for the three traits. Also, the set should allow for site losses due to unpredictable environmental events. The dry bean industry in western Canada is expanding to non-traditional growing areas and the establishment of trials in these areas should be considered as they may create GE effects not considered to date. Key words: Phaseolus vulgaris L., genotype × environment interactions, cluster analysis, AMMI analysis