Analysis of adaptation through segmented linear regression. 2. Application

Abstract

The method of simple linear regression of Finlay & Wilkinson (1963), extended by Eberhart & Russell (1966), assumes a simple linear relation between the response of a genotype and the environmental index. Thus, it is not able to identify the genotype with the desirable sensibility of response to the environment, that is, the genotype responsive to favorable or improved environments and which maintain reasonable productivity in adverse environments. Silva & Barreto (1986) propose a segmented linear model which allows to distinguish the several patterns of response which comprise the combinations of low, average and high responsiveness in the unfavorable and favorable environments. Silva (1995) approaches this model in the context of the general linear model and presents the method of segmented linear regression as an extension of the simple linear regression method to provide more possibilities for inferences to characterize and compare genotypes with regards to the adaptability to the environment. This paper illustrates this method through its application to data of an experiment for comparison of upland rice cultivars.