RESUMEN
Because of its wide distribution, high yield potential, and short cycle, the potato has become essential for global food security. However, the complexity of tetrasomic inheritance, the high level of heterozygosity of the parents, the low multiplication rate of tubers, and the genotype-by-environment interactions impose severe challenges on tetraploid potato-breeding programs. The initial stages of selection take place in experiments with low selection accuracy for many of the quantitative traits of interest, for example, tuber yield. The goal of this study was to investigate the contribution of incorporating a family effect in the estimation of the total genotypic effect and selection of clones in the initial stage of a potato-breeding program. The evaluation included single trials (STs) and multi-environment trials (METs). A total of 1,280 clones from 67 full-sib families from the potato-breeding program at Universidade Federal de Lavras were evaluated for the traits total tuber yield and specific gravity. These clones were distributed in six evaluated trials that varied according to the heat stress level: without heat stress, moderate heat stress, and high heat stress. To verify the importance of the family effect, models with and without the family effect were compared for the analysis of ST and MET data for both traits. The models that included the family effect were better adjusted in the ST and MET data analyses for both traits, except when the family effect was not significant. Furthermore, the inclusion of the family effect increased the selective efficiency of clones in both ST and MET analyses via an increase in the accuracy of the total genotypic value. These same models also allowed the prediction of clone effects more realistically, as the variance components associated with family and clone effects within a family were not confounded. Thus, clonal selection based on the total genotypic value, combining the effects of family and clones within a family, proved to be a good alternative for potato-breeding programs that can accommodate the logistic and data tracking required in the breeding program.
RESUMEN
Breeding for dry matter yield and persistence in alfalfa (Medicago sativa L.) can take several years as these traits must be evaluated under multiple harvests. Therefore, genotype-by-harvest interaction should be incorporated into genomic prediction models to explore genotypes' adaptability and stability. In this study, we investigated how enviromics could help to predict the genotypic performance under multiharvest alfalfa breeding trials by evaluating 177 families across 11 harvests under four cross-validation scenarios. All scenarios were analyzed using six models in a Bayesian mixed model framework. Our results demonstrate that models accounting to the enviromics information led to an increase of genetic variance and a decrease in the error variance, indicating better biological explanation when the enviromic information was incorporated. Furthermore, models that accounted for enviromic data led to higher predictive ability (PA) in a reduced number of harvests used in the training data set. The best enviromic models (M2 and M3) outperformed the base model (GBLUP model-M0) for predicting adaptability and persistence across all cross-validation scenarios. Incorporating environmental covariates also provided higher PA for persistence compared with the base model, as predictions increased from 0 to 0.16, 0.20, 0.56, and 0.46 for CV00, CV1, CV0, and CV2. The results also demonstrate that GBLUP without enviromics term has low power to predict persistence, thus the adoption of enviromics is a cheap and efficient alternative to increase accuracy and biological meaning.