Premium
Methodology for Identifying Wide Adaptability in Crops 1
Author(s) -
Nor Khalid M.,
Cady Foster B.
Publication year - 1979
Publication title -
agronomy journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.752
H-Index - 131
eISSN - 1435-0645
pISSN - 0002-1962
DOI - 10.2134/agronj1979.00021962007100040009x
Subject(s) - adaptability , multivariate statistics , regression analysis , index (typography) , regression , statistics , stability (learning theory) , principal component analysis , cultivar , linear regression , mathematics , crop yield , yield (engineering) , range (aeronautics) , computer science , agronomy , engineering , ecology , machine learning , biology , world wide web , materials science , metallurgy , aerospace engineering
When several cultivars are grown over a range of environments, regression measures of stability or wide adaptability can be estimated. The most commonly used measures include Eberhart and Russell's and Finlay and Wilkinson's. These measures use the average yield of the cultivars at each site as an index of the site's productivity. Because of conceptual and statistical problems associated with using the site mean yields, emphasis is now on a principal component method, based on measurements of environmental variables at each site, for formulating an environmental index. We wished to retain the regression approach of the earlier developers. The objective of this paper, therefore, to develop multivariate regression methodology for providing an alternative environmental index not dependent on the cultivar responses. The index is based on physical measurements of the environments affecting crop yield. The resulting data analysis, based on a beta response model, is straightforward and easy to apply in the same general situations as previous stability and adaptability studies. The methodology is applied to a set of corn ( Zeo mays L.) data from several New York environments. Adaptability rankings using the new methodology compare favorably with those based on site means (Spearman rank correlation of 0.69 with 16 degrees of freedom). With improved measurement techniques and understanding of site variable and yield relationships, the environmental index methodology can be an alternative regression measure of stability or wide adaptability.