Premium
Spatial Dependence of Growth Attributes and Local Control in Wheat and Oat Breeding Experiments
Author(s) -
Samra J. S.,
Anlauf R.,
Weber W. E.
Publication year - 1990
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/cropsci1990.0011183x003000060008x
Subject(s) - avena , biology , coefficient of variation , selection (genetic algorithm) , agronomy , kriging , statistics , field experiment , spatial variability , variance (accounting) , trait , mathematics , artificial intelligence , computer science , programming language , accounting , business
Adjustment for microenvironmental heterogeneity in inadequately replicated field experiments is desirable for maximizing genetigains through selection. Moving means, empirically weighted means, and a model‐based optimal predictor (kriging) have been used to estimate an index of the local heterogeneity of 276 wheat ( Triticum aestivum L.) and 336 oat ( Avena sativa L.) plots. The adjustments of the data based on this index have been compared. The wheat experiment was unreplicated and that on oat had two replications. Trend variation along rows and columns was about 25% of the variance in wheat and 9 to 18% in the oat trial. Depending on the trait, 39 to 54% of the remaining variability of wheat and 7 to 20% of oat was stochastically isotropically spatially structured. Kriging reduced the coefficient of variation (CV) in all the traits, including yield, and never made overcorrections for the local variation, whereas adjustments based on moving means and empirically weighted means frequently increased the CV. Local control simulated only from check plots was less practical in wheat as compared to oat. Wheat selections from the unadjusted data were paired or clustered in localized parts of the field. These selections became randomly distributed across the entire field after the microenvironmental variation was removed by the kriging method.