Genomic selection of 1,000 biomass sorghum accessions and empirical validation

Authors: YU, XIANQING - Iowa State University; LI, XIANRAN - Iowa State University; WU, YUYE - Kansas State University; MITCHELL, SHARON - Cornell University; ROOZEBOOM, KRAIG - Kansas State University; WANG, DONGHAI - Kansas State University; BERNARDO, REX - University Of Minnesota; Wang, Ming; Pederson, Gary; TESSO, TESFAYE - Kansas State University; YU, JIANMING - Iowa State University

Submitted to: Plant and Animal Genome Conference
Publication Type: Abstract Only
Publication Acceptance Date: 11/4/2014
Publication Date: 1/11/2015
Citation: Yu, X., Li, X., Wu, Y., Mitchell, S.E., Roozeboom, K.L., Wang, D., Bernardo, R., Wang, M.L., Pederson, G.A., Tesso, T.T., Yu, J. 2015. Genomic selection of 1,000 biomass sorghum accessions and empirical validation.[abstract] Plant and Animal Genome Conference. p.194.

Interpretive Summary: Substantial genetic diversity exists in sorghum, a key lignocellulosic biofuel species in the United States. The implementation of genomic, genetic tools to select and enhance current germplasm will greatly accelerate new variety development. Several key questions should be considered in implementing genomics-assisted biomass feedstock improvement: 1) How to tap into the vast plant germplasm collections for biomass crop improvement? 2) How to increase the information contained in genotypic and phenotypic data for the selected germplasm? 3) How to leverage a high-throughput phenotyping method such as Near Infrared (NIR) to facilitate plant biomass composition investigation? 4) How robust are the various genomic prediction models on biomass traits? In this study, genotyping by sequencing (GBS) was conducted for 1,000 sorghum accessions sampled from the germplasm bank and generated 125k SNPs after quality filtering. A set of 300 accessions, selected to be most representative from this panel were extensively phenotyped for biomass yield, plant height, stem diameter, stalk number, stalk lodging, and root lodging at Kansas in 2013 and Texas in 2012 and 2013. Cellulose and lignin content were investigated through both wet chemical and high throughput NIR methods. Genome-wide prediction models were established with several common methods. Biomass traits of the 700 untested accessions were predicted by using the optimal genomic selection model. Predictions for accessions with high and low biomass potential plus 100 random accessions are being validated with field phenotyping. This research highlights the integration of genomic selection, selective phenotyping, and genotyping-by-sequencing in realizing the potential of gene banks.

Technical Abstract: Substantial genetic diversity exists in sorghum, a key lignocellulosic biofuel species in the United States. The implementation of genomic, genetic tools to select and enhance current germplasm will greatly accelerate new variety development. Several key questions should be considered in implementing genomics-assisted biomass feedstock improvement: 1) How to tap into the vast plant germplasm collections for biomass crop improvement? 2) How to increase the information contained in genotypic and phenotypic data for the selected germplasm? 3) How to leverage a high-throughput phenotyping method such as Near Infrared (NIR) to facilitate plant biomass composition investigation? 4) How robust are the various genomic prediction models on biomass traits? In this study, genotyping by sequencing (GBS) was conducted for 1,000 sorghum accessions sampled from the germplasm bank and generated 125k SNPs after quality filtering. A set of 300 accessions, selected to be most representative from this panel were extensively phenotyped for biomass yield, plant height, stem diameter, stalk number, stalk lodging, and root lodging at Kansas in 2013 and Texas in 2012 and 2013. Cellulose and lignin content were investigated through both wet chemical and high throughput NIR methods. Genome-wide prediction models were established with several common methods. Biomass traits of the 700 untested accessions were predicted by using the optimal genomic selection model. Predictions for accessions with high and low biomass potential plus 100 random accessions are being validated with field phenotyping. This research highlights the integration of genomic selection, selective phenotyping, and genotyping-by-sequencing in realizing the potential of gene banks.