Premium
REML‐Based Diallel Analysis
Author(s) -
Möhring J.,
Melchinger A. E.,
Piepho H. P.
Publication year - 2011
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/cropsci2010.05.0272
Subject(s) - diallel cross , covariance , mixed model , a priori and a posteriori , statistics , variance (accounting) , analysis of covariance , variance components , computer science , random effects model , econometrics , mathematics , biology , meta analysis , medicine , philosophy , botany , accounting , epistemology , hybrid , business
Diallel analyses can give different results depending on the model and the experimental design. A unified framework to fit and compare different models by mixed‐model packages is lacking. We, therefore, present a general diallel model and the required additional restrictions for the genetic variance–covariance structure to become equivalent to other commonly used diallel models. We discuss the definitions and requirements of the commonly used models compared with the general model and provide an analysis for three real datasets. If the assumptions regarding the genetic variance–covariance structure are not fulfilled, biased variance‐component estimates for general and specific combining ability effects are obtained when using random genetic effects. We implemented a general diallel model in the statistical software packages SAS and ASReml and give detailed program codes. If no a priori information about the genetic model for genotypes is available, the general model can be used to analyze all diallel designs easily with standard statistical software.