Inter-site transferability of crop varieties: another approach for analyzing multi-locational variety trials

Abstract

Improved crop varieties developed at experimental stations are intended for release for production in farmers' fields which are not necessarily the test sites. Therefore, it is important to measure the inter-site transferability of the crop variety. This study was undertaken to establish a statistical measure of inter-site transferability of a variety. Data on grain yield from two sets of multi-locational trials on barley and four sets of the wheat trials conducted by the Cereal Program, International Center for Agricultural Research in the Dry Areas, Syria, were analyzed to evaluate the inter-site transferability of the varieties. Genotype × environment interaction was significant and experimental errors were heterogeneous in each set of trials. Correlations between the slope (of linear regression of variety mean on location mean) and variety mean were significant (P < 0.01) in all trials. Correlations between the developed inter-site transferability statistic and the variety mean were significant (P < 0.05) in only two trials. Correlations between mean yield and probability of the inter-site transferability statistic were insignificant in all trials. The inter-site transferability statistic and the associated probability of the transferability may be used to select varieties which are high yielding as well as transferable to a new environment. The strong association of slope with mean yield restricted the scope of varietal selection but the introduced statistic did not suffer from this drawback.