AUTHOR=Beyene Yoseph , Gowda Manje , Pérez-Rodríguez Paulino , Olsen Michael , Robbins Kelly R. , Burgueño Juan , Prasanna Boddupalli M. , Crossa Jose 

TITLE=Application of Genomic Selection at the Early Stage of Breeding Pipeline in Tropical Maize

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 12 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2021.685488

DOI=10.3389/fpls.2021.685488

ISSN=1664-462X

ABSTRACT=Doubled haploid (DH) line production capacity currently exceeds the capacity to phenotypically evaluate the complete set of test cross candidates in multi-location trials. The ability to partially select the lines based on genotypic data, while maintaining or improving genetic gains achieved using phenotypic selection alone, can result in significant resource savings. The objective of the present study was to evaluate genomic selection (GS) prediction scenarios for grain yield (GY), anthesis date (AD) and plant height (PH) based on multi-year empirical data for designing GS strategies in the early stages of a maize breeding pipeline. We used field data from 3,068 tropical maize DH lines genotyped using rAmpSeq markers and evaluated as testcrosses in well-watered (WW) and managed water-stress (WS) environments in Kenya from 2017 to 2019. Three prediction schemes were compared using 1) one year of performance data to predict a second year, 2) two years of pooled data to predict a third year, and 3) using individual or pooled data plus converting a certain proportion of individuals from the testing set (TST) to training set (TRN) to predict the next year’s data. Employing five-fold cross-validation, we observed that the average prediction accuracies, estimated from the independent validation schemes for GY, ranged from 0.19 to 0.29 under WW and 0.22 to 0.31 under WS, when the one-year datasets were used as TRN to predict a second year’s data as TST. The average prediction accuracies increased to 0.32 under WW and 0.31 under WS when the two-year datasets were used as TRN to predict the third-year data set. In a forward prediction scenario, good predictive abilities (0.53 to 0.71) were observed when the TRN consisted of previous year breeding data and converting 30% of the next year’s data from TST set to TRN set. Further implications of these results for breeding program design are discussed.