Determination of efficient test sites for evaluation of peanut breeding lines using the CSM-CROPGRO-peanut model

Abstract

Efficient testing of crop breeding lines requires a set of complementary test sites that adequately sample the target environments with minimal duplication. Such test sites have been derived from actual multi-environmental trial (MET) data, which often have a limitation with respect to their environmental coverage. However, this limitation can be overcome using a crop simulation model. The goal of this study was to determine the efficient test sites for METs of peanut breeding lines in Thailand using the CSM-CROPGRO-Peanut model. The model was used to simulate pod yield for 17 peanut lines at all peanut production areas in Thailand that included 76 locations in the early-rainy season, 39 locations in the mid-rainy season and 47 locations in the dry season for 30 years. The simulated data were used to sub-divide the locations for each season into groups using cluster analysis and the genotype plus genotype × environment (GGE) biplot method. Six sets of test sites were obtained based on different scenarios for site selection that included combinations of geographical distribution and representation of location-groups as determined by the two methods. Set 1 was based on geographic distribution. Sets 2–4 were based on location grouping by cluster analysis, but with the sites distributed in all regions (Set 2), or only in the north (Set 3) or northeast (Set 4). Set 5 consisted of the sites currently used, and Set 6 was based on location grouping by the GGE biplot. Although Sets 2 and 6 appeared to capture more genotype × location interaction than the others, performance rankings of the test genotypes were almost the same for all sets. They were, therefore, considered equally effective for breeding line evaluation. The final selection was then based on the convenience, and consequently the cost, for conducting the METs. Set 4 was considered most preferable in this regard. This study demonstrated the usefulness of a crop simulation model as a tool in determining the most efficient test sites for the evaluation of peanut breeding lines.