Genomic selection and reproducibility: are complex models distracting us from true scientific validity in the presence of genotype-by-environment interaction?

Genomic selection is poised to remain an exemplary plant breeding tool if reproducibility is maximized. Defined as the consistency of results when using the same data and approaches, reproducibility is as crucial as predictive ability when assessing the performance of genomic selection. Striving for reproducibility instead of model complexity will bolster efforts to make genomic selection applicable to more situations, as well as understandable to a wider range of end-users. The promises of high-dimensional, multi-environment data to contribute to accurate genomic-estimated breeding values are more likely to be attained if reproducibility is maximized. A focus on reproducibility will underscore trust among plant breeders and other practitioners using genomic selection.