Analysis of yield stability and test site representativeness of sugarcane trials using combined AMMI and HA-GGE biplot models

Knowledge on yield stability of sugarcane varieties and representativeness of test sites during national regional trials of sugarcane varieties is critical for rationally regional distribution and guiding sugarcane breeding. Additive main effects and multiplicative interaction (AMMI) and heritability adjusted GGE (HA-GGE) biplot models are the two widely used statistical methods in analyzing data on crop variety trials. Using experimental data from regional trials, the interactions between crop genotype and environment can be analyzed to determine yield potential of tested varieties and representativeness of pilots. In order to assess the representativeness and discrimination ability of 13 test locations of sugarcane variety trail, the AMMI model was used in combination with HA-GGE bipliot to analyze cane and sugar yields of 11 sugarcane varieties planted at 10 national regional trial sites of sugarcane varieties in China. The results showed that cane and sugar yields were significantly different for different varieties and test sites. There was significant interaction effect between variety and test site. The ‘FN40’ sugarcane variety had the best performance, with high yield, good productivity and high stability of cane yield and sugar yield. The yield of ‘YZ08-2060’ sugarcane variety was slightly lower than that of ‘FN40’, while the stability of its cane and sugar yields was slightly higher than that of ‘FN40’. Compared with ‘ROC22’ variety, ‘YG43’, ‘YG46’ and ‘MT02-205’ had higher cane and sugar yields, but had lower stability. ‘FN40’, ‘YG43’, ‘YG46’ and ‘YZ08-2060’ had stronger adaptability, and were recommended for cultivation at suitable planting sites. The result further indicated that three test locations, Suixi of Guangdong Province, Kaiyuan of Yunnan Province and Fuzhou of Fujian Province, had higher discrimination and better representativeness based on comprehensive analysis of AMMI and HA-GGE biplot models. Here, two sugarcane varieties (‘FN40’ and ‘YZ08-2060’), both with high yield and good stability, were recommended for commercial production in all sugarcane planting areas in China. In conclusion, the combined used of AMMI and GGE biplot model provided more accurate assessment for the productivity, stability and adaptability of sugarcane varieties. It was also very reliable in terms of demonstration of discrimination and representativeness of test sites. This study provided valuable reference base for the identification and promotion of new sugarcane varieties.