Premium
Custer Analysis for Genotype × Environment Interaction with Unbalanced Data
Author(s) -
Ouyang Z.,
Mowers R. P.,
Jensen A.,
Wang S.,
Zheng S.
Publication year - 1995
Publication title -
crop science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.76
H-Index - 147
eISSN - 1435-0653
pISSN - 0011-183X
DOI - 10.2135/cropsci1995.0011183x003500050008x
Subject(s) - homogeneous , selection (genetic algorithm) , biology , principal component analysis , cluster (spacecraft) , hybrid , cultivar , statistics , measure (data warehouse) , adaptation (eye) , missing data , yield (engineering) , mathematics , computer science , agronomy , data mining , artificial intelligence , combinatorics , materials science , neuroscience , metallurgy , programming language
From the viewpoint of a seeds business, classification of locations into homogeneous groups based on genotype × environment interaction (GE) facilitates selection of testing sites and proper placement hybrids or cultivars. Our principal objective was to develop methods to cluster ICI Seeds' strip‐test locations into homogeneous groups with respect to GE. Difficulties encountered in these cluster analyses included missing values in the distance matrix used for cluster analysis (caused by unbalanced data), selection of a proper distance measure, and graphical presentation of clusters. We propose use of a modified distance measure to resolve the first two difficulties. Dendograms are used to present clusters, and bar charts are used to show relative effects of individual hybrids or cultivars within clusters. Classification of counties in Iowa is into four groups: northern, central, southeastern, and southwestern regions. The results of the GE analyses compare favorably with breeder and agronomist observations on hybrid adaptation and show some similar groupings with divisions of the state specified in the Iowa State University yield test report.