B1764
Title: Predictions in multi-environment agricultural trials
Authors: Aniruddha Pathak - Iowa State University (United States) [presenting]
Somak Dutta - Iowa State University (United States)
Abstract: The additive main effects and multiplicative interaction (AMMI) model is widely used for analyzing genotype-by-environment interaction in multi-environment field trials. However, the ordinary AMMI model does not allow predictions for untested genotypes. A novel hierarchical AMMI mixed-effects model is developed that uses the kinship information among the genotypes and accommodates the missing data. A scalable stochastic expectation-maximization algorithm is developed for likelihood-based inference with large multi-environment trial datasets and is further accelerated by the squared extrapolation method. Simulation studies and maize data from the genomes to fields initiative are used to illustrate the prediction improvements and detection of non-linear genotype-by-environment interactions.
