Skip to main content
ARS Home » Pacific West Area » Albany, California » Western Regional Research Center » Crop Improvement and Genetics Research » Research » Publications at this Location » Publication #349091

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

item FIELDER, JASON - North Dakota State University
item LANZATELLA, CHRISTINA - Former ARS Employee
item Edme, Serge
item Palmer, Nathan - Nate
item Sarath, Gautam
item Mitchell, Robert - Rob
item Tobias, Christian

Submitted to: BMC Plant Biology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/1/2018
Publication Date: 7/9/2018
Citation: Fielder, J.D., Lanzatella, C., Edme, S.J., Palmer, N.A., Sarath, G., Mitchell, R., Tobias, C.M. 2018. Genomic prediction accuracy for switchgrass traits related to bioenergy within differentiated populations. Biomed Central (BMC) Plant Biology. 18:142.

Interpretive Summary: Switchgrass is a native tallgrass prairie species that produces large quantities of biomass and requires few inputs. These attributes make it attractive as a forage, for conservation planting, and as a feedstock for biofuel and bioenergy production. Traditional switchgrass breeding requires 3-4 years to complete one cycle of trait improvement. In order to more efficiently breed switchgrass for bioenergy production, the long breeding cycle needs to be shortened. This can be accomplished by a process called genomic selection, which uses dense DNA-based genetic information to predict traits without actually assessing them in each generation, thus reducing the generation interval. In this report, we use four different methods to predict different switchgrass traits related to bioenergy from genetic information. We also measure the impact of population, breeding history, and degree of relatedness on prediction accuracy. Depending on the trait and population, prediction accuracy ranged from very high to insignificant. Not accounting for origin, breeding history, and degree of relatedness increased apparent accuracy, but most of this increase was due to structural differences between populations rather than differences among them. Breeders can use this type of information to make informed decisions about the best candidates to propagate for improvement of their targeted traits.

Technical Abstract: Switchgrass 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. Using five distinct populations comprised of three lowland, one upand and one hybrid-based accession, the accuracy of genomic predictions under different cross-validation strategies and prediction methods was investigated. kin-BLUP, partial least squares, sparse partial least squares, and BayesB were employed to predict yield, morphological, and composition data collected in 2012-2013 from a replicated Nebraska field trial. The individual genotypes were determined using a genotyping by sequencing strategy. Population structure was assessed by F statistics which ranged from 0.3952 between lowland and upland populations to 0.0131 among the lowland populations. Prediction accuracy was significantly affected by both cross-validation strategy and trait. Accounting for population structure by using a cross-validation strategy constrained by population resulted in accuracies that were 69% lower on average than apparent accuracies in the presence of population structure. Accuracy ranged from 0.57 and 0.52 for cell wall soluble glucose and fructose, respectively, to insignificant for traits with low repeatability or where most of the variation was between populations, such as spring regreening and yield. These results demonstrate that genomic prediction can be used with some success for predicting genetic gains in switchgrass. However, factors such as population structure, training set composition, trait genetic architecture, and marker density influence prediction accuracy.