Using GGE biplot and comprehensive selection index to investigate mega-environments of cotton cultivar

The application of cultivar selection index in crop variety breeding program could improve simultaneous selection efficiency of multiple target traits. Also genotypes derived from explorations of interactions with the environment and investigations of mega-environments using selection index contribute to rational utilization of specific adaptations of certain cultivars and environments, which could eventually enhance the reliability of variety breeding and multi-trait applications. As the most useful statistical and visualizing tool for mega-environment investigation, GGE biplot technique has been extensively used in the analysis of regional crop-trial datasets. Nevertheless, reports on cotton mega-environment identification using comprehensive cultivar selection index have to date remained scarce. The objective of this study was: 1) to construct a set of practicable cultivar selection index in line with national cotton variety registration criteria and 2) to investigate mega-environments using multi-trait selection index based on GGE biplot analysis. Datasets were collected from 39 sets of regional trials of cotton varieties, including 585 single-site cultivar comparison tests in the Yangtze River Valley (YaRV) in 2000-2013. Based on the results, the constructed cotton cultivar selection index (SI) was SI = 0.40 × lint cotton yield + 0.13 × fiber strength + 0.09 × (fiber length + micronaire value) + 0.11 × Fusarium wilt + 0.09 × Verticillium wilt + 0.10 × harvesting ratio of seed cotton before frost. Based on GGE biplot analysis, cotton planting region in YaRV was divided into four mega-environments for selection index. The four mega-environments included Sichuan Basin, Nanxiang Basin, Zhejiang Province Coastal Region and YaRV Middle/Lower Reaches. YaRV Middle/Lower Reaches mega-environemnt was most representative of the entire region. It covered the main cotton planting regions in YaRV, including the area around Dongting Lake in Hunan Province, the Jianghan Plains, the Southeast Downland in Hubei Province, the area around Poyang Lake in Jiangxi Province, the area along Yangtze River in Anhui Province, the Ningzhen Hilly Region, the area along Yangtze River and the Coastal Region in Jiangsu Province. However, the mega-environments of Sichuan Basin, Nanxiang Basin and Zhejiang Province Coastal Region were identified as special sub-regions with distinct ecological conditions. This set of environments was less representative of cotton planting region in YaRV. Subsequently, it was beneficial to promote breeding efficiency in order to realize broad adaptation selection of multi-trait across the entire cotton planting region in YaRV via preferential arrangements of breeding locations in the YaRV Middle/Lower Reaches maga-environment. Although the other mega-environments were not conducive for selection to represent the entire region for broad breeding adaptation programs, they were suitable for focusing on specific adaptive cultivar selections. This study showed the effectiveness of GGE biplot analysis in ecological regionalization. It was successfully used to divide the mega-environments in RaRV based on cultivar selection index. The study provided the scientific basis for decision-making on multi-trait cotton selections and recommendation of new cultivar policies in YaRV. It also provided a good example for implementation of similar ecological analyses in other cotton planting regions or even other crops.