Comparison of locations used in cotton-breeding trials

Cotton-breeding trials are conducted annually throughout the commercial growing regions of eastern Australia. Accumulated yield data for the period 1974—85 were assembled into an incomplete cultivar × location × year table. This table was then analysed in order to compare test locations. The method involved analysing cultivar × location tables separately for each year, using symmetric joint regression analysis. Results were then collected into location × year tables and further analysed. Four criteria for comparing test locations were developed. The discrimination criterion is important when locations are evaluated in terms of their ability to display cultivar differences. The representation criterion measures the ability of a location to mirror the relative performance of cultivars over all locations. The other 2 criteria are concerned with the mean yield at test locations and the stability of location yields over years. Based on the 4 criteria, preferred test locations are recommended.