Challenges of G x E and How to Overcome Them.

Genotype by Environment interaction is a long-standing issue in biology and agriculture. It hampers our ability to accurately predict plant traits under diverse environments for multiple genotypes. Three major areas of research can be identified as the key elements to establish an Integrated Modeling Approach for Performance Prediction (IMAPP): 1, quantitative and population genetics framework and associated developments; 2, crop physiological modeling and agriculture production system; and 3, systems biology. Each of these three areas has its advantages and disadvantages in modeling and predicting performance at different levels (e.g., number of genotypes, context of performance, and input and output). Recent advances in genomic technologies have greatly empowered research to conduct genome-wide prediction within the quantitative and population genetics framework. Genome-wide prediction has been shown as a practical tool that outperforms the traditional quantitative trait locus (QTL) identification and marker assisted prediction. Through genome-wide prediction, all individual genotypes can be cross-referenced for prediction. Systems biology and comparative genomics allow the construction of gene regulatory networks (GRN) that facilitate both pathway- and network-based mapping, which may be incorporated into overall prediction of phenotypes. Crop models have the capacity to cross-reference tested and untested environments. Advances in crop modeling enabled the inclusion of multiple cultivar parameters to accommodate the genetic diversity, which makes crop models the best potential framework to synthesize all genetic and non-genetic information for prediction under diverse environments.  However, targeted research is urgently needed to incorporate these new advances into a broader crop modeling framework that deals directly with environmental and production conditions.