ABSTRACT
The Guadalquivir Valley produces 90% of Spain's cotton. Currently, about 30 cultivars are available to cotton growers. Most of these are upland cultivars grown in the USA. It is critical to establish the adaptability and stability of the leading cotton cultivars grown in Spain. Data from the Andalusian Agrarian Experimentation network were utilized in this study. Tests were conducted over a fouryear period, 19871990, at 13 locations in four provinces. Eighteen cultivars were tested; 12 originated in the USA and 6 were developed in Spain. Seedcotton yield ranged from 3367 kg/ha for Acala SJC1 to 4218 kg/ha for Tabladilla 16. Coker 310 and Deltapine 90 produced yields above 3850 kg/ha. The linear regression coefficient (b), the mean square deviation (msd) and the coefficient of determination (r2) were used in analyzing the data. Principal factor analysis was applied to the residual (genotype x environment interaction) matrix of the 18 genotypes and 13 sites. We have found a negative significant correlation between the regression coefficient and minimum yield. This result confirms the conclusion that cultivars with higher regression coefficients are, in general, the least productive in less favorable environment, and those with lower regression coefficient are the most productive. Some exceptions exist, such as Tabladilla 16, a Spanish cultivar, and Deltapine 90 an American cultivar, which having different regression coefficients had high seedcotton production in all environments. Some cultivars were very adapted, based on the results of b, r2 and msd. The regression coefficient, average and minimum yield were the most related parameters with factor 1. Factor 2 may be related to r2 and msd, and factor 3 to average yield. The Factor Analysis is a useful tool to establish adaptability of cultivars and can help in interpreting some results which on individual basis, would be difficult to realize.
