Genomic prediction accuracy for switchgrass traits related to bioenergy within differentiated populations

Abstract

BackgroundSwitchgrass breeders need to improve the rates of genetic gain in many bioenergy-related traits in order to create improved cultivars that are higher yielding and have optimal biomass composition. One way to achieve this is through genomic selection. However, the heritability of traits needs to be determined as well as the accuracy of prediction in order to determine if efficient selection is possible.ResultsUsing five distinct switchgrass populations comprised of three lowland, one upland and one hybrid accession, the accuracy of genomic predictions under different cross-validation strategies and prediction methods was investigated. Individual genotypes were collected using GBS while kin-BLUP, partial least squares, sparse partial least squares, and BayesB methods were employed to predict yield, morphological, and NIRS-based compositional data collected in 2012–2013 from a replicated Nebraska field trial. Population structure was assessed by F statistics which ranged from 0.3952 between lowland and upland accessions to 0.0131 among the lowland accessions. Prediction accuracy ranged from 0.57–0.52 for cell wall soluble glucose and fructose respectively, to insignificant for traits with low repeatability. Ratios of heritability across to within-population ranged from 15 to 0.6.ConclusionsAccuracy was significantly affected by both cross-validation strategy and trait. Accounting for population structure with a cross-validation strategy constrained by accession resulted in accuracies that were 69% lower than apparent accuracies using unconstrained cross-validation. Less accurate genomic selection is anticipated when most of the phenotypic variation exists between populations such as with spring regreening and yield phenotypes.Electronic supplementary materialThe online version of this article (10.1186/s12870-018-1360-z) contains supplementary material, which is available to authorized users.

Genome‐wide association and genomic prediction for biomass yield in a genetically diverse Miscanthus sinensis germplasm panel phenotyped at five locations in Asia and North America

“…Because we have previously identified genetic groups within M. sinensis (Clark et al, , ) and evaluated phenotypic differences among these groups in this set of field trials (Clark et al, ), we corrected for population structure by removing variance attributable to genetic groups by (a) performing genomic prediction within individual groups (Tables and ), (b) estimating BLUPs for genotype‐within‐genetic‐group from Equations and ( G ( D ); Table ), or (c) analyzing residuals of genotype BLUPs fitted to genetic group ( R ; Equation ). Prediction accuracies were typically lower for the methods that accounted for population structure relative to analyses that did not account for structure, which was consistent with expectations because failure to account for population structure has been shown to bias estimates upwards (Fiedler et al, ; Guo et al, ; Riedelsheimer et al, ). For biomass yield, genomic prediction accuracies were mostly moderate for each of the three methods we used to control for population structure (Tables ).…”

“…The analyses were based on 46,177 SNP markers and 568 accessions. Genetic groups were excluded from analysis when fewer than 50 genotypes had phenotypic data typically lower for the methods that accounted for population structure relative to analyses that did not account for structure, which was consistent with expectations because failure to account for population structure has been shown to bias estimates upwards (Fiedler et al, 2018;Guo et al, 2014;Riedelsheimer et al, 2012). For biomass yield, genomic prediction accuracies were mostly moderate for each of the three methods we used to control for population structure (Tables 3-5).…”

“…However, if Equations and are applied to all entries in the germplasm panel, they do not account for population structure. Prior studies have found that population structure can bias estimates of genomic prediction accuracies upwards (Fiedler et al, ; Guo et al, ; Riedelsheimer et al, ). Because we previously found that the genetic groups within M. sinensis differed from each other with respect to yield and yield‐component traits, and because it was possible to predict from marker data which genetic group a genotype was in, we expected genomic prediction accuracy across the entire germplasm collection to be higher than genomic prediction accuracy within groups.…”

“…Azevedo et al (2017) used the first several eigenvectors of the relationship matrix as covariates in the prediction model. Fiedler et al (2018) assigned entire genetic groups to the training or validation set. Arruda et al (2015) used k-means clustering to identify closely related individuals and to make sure they were not in both the training and validation sets.…”

Winter hardiness of Miscanthus (III): Genome‐wide association and genomic prediction for overwintering ability in Miscanthus sinensis

“…Population structure affected genomic prediction of overwintering ability for M. sinensis . For whole panel analysis, Method 1, which did not account for population structure, resulted in prediction accuracies that were biased upward relative to Methods 2 (by 8 points) and 3 (by 10 points) which did account for population structure; the observed bias was consistent with prior studies on other crops (Fiedler et al, ; Riedelsheimer et al, ; Spindel et al, ). To effectively differentiate the overwintering potential of individuals within the Northern Japan and Korea/N China groups, from which nearly all genotypes survived the first winter, higher resolution phenotyping schemes such as a 1–10 ordinal system to capture relative vigor, rather than the current binary system that captured only survival data could be helpful; additionally, challenging these genotypes in colder winter environments would be another promising strategy.…”