Yield–trait performance landscapes: from theory to application in breeding maize for drought tolerance

TLDR: The existence of rugged yield-performance landscapes with multiple peaks and intervening valleys of lower performance supports the proposition that phenotyping strategies, and the directions emphasized in genomic selection can be improved by creating knowledge of the topology of yield-trait performance landscapes.

Abstract: The effectiveness of breeding strategies to increase drought resistance in crops could be increased further if some of the complexities in gene-to-phenotype (G/P) relations associated with epistasis, pleiotropy, and genotype-byenvironment interactions could be captured in realistic G/P models, and represented in a quantitative manner useful for selection. This paper outlines a promising methodology. First, the concept of landscapes was extended from the study of fitness landscapes used in evolutionary genetics to the characterization of yield–trait-performance landscapes for agricultural environments and applications in plant breeding. Second, the E(NK) model of trait genetic architecture was extended to incorporate biophysical, physiological, and statistical components. Third, a graphical representation is proposed to visualize the yield–trait performance landscape concept for use in selection decisions. The methodology was demonstrated at a particular stage of a maize breeding programme with the objective of improving the drought tolerance of maize hybrids for the US Western Corn-Belt. The application of the framework to the genetic improvement of drought tolerance in maize supported selection of Doubled Haploid (DH) lines with improved levels of drought tolerance based on physiological genetic knowledge, prediction of testcross yield within the target population of environments, and their predicted potential to sustain further genetic progress with additional cycles of selection. The existence of rugged yield-performance landscapes with multiple peaks and intervening valleys of lower performance, as shown in this study, supports the proposition that phenotyping strategies, and the directions emphasized in genomic selection can be improved by creating knowledge of the topology of yield–trait performance landscapes.