Modeling genotype × environment interaction for single- and multi-trait genomic prediction in potato (Solanum tuberosum L.)

Abstract

In this study we extend research on genomic prediction (GP) to polysomic polyploid plant species with the main objective to investigate single trait (ST) versus multi-trait (MT) for multi-environment (ME) models for the combination of three locations in Sweden (Helgegrden [HEL], Mosslunda [MOS], Ume [UM]) over two year-trials (2020, 2021) of 253 potato cultivars and breeding clones for five tuber weight traits and two tuber flesh quality characteristics. This research investigated the GP of four genome-based prediction models with genotype environment interactions (GE): (1) single trait reaction norm model (M1), (2) single trait model considering covariances between environments (M2), (3) single trait M2 extended to include a random vector that utilizes the environmental covariances (M3) and (4) multi-trait model with GE (M4). Several prediction problems were analyzed for each of the GP accuracy of the four models. Results of the prediction of traits in HEL, the high yield potential testing site in 2021, show that the best predicted traits were tuber flesh starch (%), weight of tuber above 60 or below 40 mm in size, and total tuber weight. In terms of GP, accuracy model M4gave the best prediction accuracy in three traits, namely tuber weight of 4050 or above 60 mm in size, and total tuber weight and very similar in the starch trait. For MOS in 2021, the best predictive traits were starch, weight of tuber above 60, 5060, or below 40 mm in size, and total tuber weight. MT model M4 was the best GP model based on its accuracy when some cultivars are observed in some traits. For GP accuracy of traits in UM in 2021, the best predictive traits were weight of tuber above 60, 5060, or below 40 mm in size and the best model was MT M4 followed by models ST M3 and M2.

Inbreeding effects on the performance and genomic prediction for polysomic tetraploid potato offspring grown at high Nordic latitudes

“…The  estimates also differed for total tuber yield among the four S1 offspring, suggesting that genomic predictions depend on the genetic background and may be effective for purging harmful alleles after inbreeding. Genomic prediction may be further improved in potato when using released cultivar and advanced breeding clones, by using a multi-trait, multi-environment GEBV modeling [36]. Modeling genotype × environment interaction in the multi-environment analyses may further exploit the information on the relationship between the site-year combinations, thereby leading to larger  than those from the single-environment analyses [33], while multi-trait genomic prediction may maximize genetic gain with respect to a focal trait while controlling the variation in multiple secondary traits in potato [36].…”

“…Genomic prediction may be further improved in potato when using released cultivar and advanced breeding clones, by using a multi-trait, multi-environment GEBV modeling [36]. Modeling genotype × environment interaction in the multi-environment analyses may further exploit the information on the relationship between the site-year combinations, thereby leading to larger  than those from the single-environment analyses [33], while multi-trait genomic prediction may maximize genetic gain with respect to a focal trait while controlling the variation in multiple secondary traits in potato [36]. Furthermore, deep learning [37] for genomic prediction of complex multigenic traits such as tuber yield may be worth pursuing in potato breeding.…”

Partial least squares enhance multi-trait genomic prediction of potato cultivars in new environments

“…MT models have more problems of convergence than ST models, and implementing MT models for genomic prediction is more challenging due to the size and complexity of the underlying data sets 25 . However, in a recent potato study on GS prediction, Cuevas et al 26 investigated ST versus MT in ME models for the combination of six environments for five tuber weight traits and two tuber flesh quality characteristics. The best predictive model was the MT for predicting several traits of potato observed in some environments and predicted in other environments.…”

Partial least squares enhance multi-trait genomic prediction of potato cultivars in new environments

“…MT models have more problems of convergence than ST models, and implementing MT models for genomic prediction is more challenging due to the size and complexity of the underlying data sets (Montesinos-López et al 2019c). However, in a recent potato study on GS prediction, Cuevas et al (2022) investigated ST versus MT in ME models for the combination of six environments for five tuber weight traits and two tuber flesh quality characteristics. The best predictive model was the MT for predicting several traits of potato observed in some environments and predicted in other environments.…”