Optimizing the test locations and replicates in multi‐environmental cotton registration trials in southern Xinjiang, China

Abstract

AbstractMulti‐environment trials are routinely conducted around the world to select superior cultivars; the numbers of replicates and locations remains insufficiently studied despite their effects on trial efficiency and cost. The objective of the present study was to compensate for the mentioned lack by dealing with the specific case of cotton (Gossypium hirsutum L.) in arid conditions of China and by estimating the optimal numbers of locations and/or replicates in a more comprehensive way than implemented so far, that is, optimizing with regard jointly to three criteria, namely, genotype ranking, location evaluation, and environment investigation. Trial heritability and the optimal numbers of locations and replicates were estimated through adapted genotype plus genotype × environment biplot analysis from data of cotton variety registration trials in the fringes of the Taklimakan Desert in Southern Xinjiang in China over the 2011–2020 period when three replicate trials were generally conducted in nine locations. Despite the identification of two mega‐environments (MEs) through location grouping biplot analysis, our study showed robust results of genotype ranking, correlation among test locations, and location grouping that were not affected by reducing the number of replicates. It was concluded that two replicates (instead of three) were sufficient for reliable genotype evaluation, test location evaluation, and ME classification in the studied trial scheme. The implied savings justifies similar study on other schemes.