STUDIES ON THE STRUCTURAL MODEL OF COTTON YIELD AND ITS IMPLICATIONS ON COTTON BREEDING
Chen Zhongfang Xie Qilin Chen Hongliang Teng Chuanyuan Shen Duanzhuang Xiao Yngfang Hu Yanxin Gu Limei (Institute of Economic Crops,Jiangsu Academy of Agricultural Sciences)
Three yibeld structural models of two cotton finite populations ar(?)investigated by multiple regression.(1)Results indicated that modle Ⅰ,lint yield/unit are a(?)=boll num-bers/unit area(B/M)×single boll weight(Y_1/B)×lint percentage(Y_2/Y_1);model Ⅱ,lint yield/unit area(y_2)(?) numbers/unit area(PN/M)×boll number/plant(B/PN)×single boll weight(Y_1/B)×lint Pe-rcentage(Y_2/Y_1);and model Ⅲ,lint yield/plant(?)=boll number/plant(B/PN)single boll weight(Y_1/B)lint percentage(Y_2/Y_1)areagreeable to cotton biological entities.Because the multiple regressionequation on these 3 models reach significance of 1% leavel of probabi-lity and the partial regression coefficients also reach significance at 1%leavel of probability.(2)Boll number is the most important component contributing toyield,followed by single boll weight and lint percentage.(3)Selection pressure should be put on boll number in the cottonbreeding program as far as to increase the yield is concerned.Becauseof correlation between boll number,boll weight and lint percentage beingsignificantly negative,selection for bolls must be appropriate.Theproblem for selecting boll number effectively needs to be studied.