Evaluation of models for utilization in genomic prediction in the Louisiana sugarcane breeding program

Authors: SATPATHY, SUBHRAJIT - LSU Agcenter; SHAHI, DIPENDRA - LSU Agcenter; BLANCHARD, BRAYDEN - LSU Agcenter; PONTIF, MICHAEL - LSU Agcenter; GRAVOIS, KENNETH - LSU Agcenter; KIMBENG, COLLINS - LSU Agcenter; Hale, Anna; Todd, James; RAO, ATMAKURI - Indian Agricultural Research Institute; BAISAKH, NIRANJAN - LSU Agcenter

Submitted to: Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/24/2022
Publication Date: 8/29/2022
Citation: Satpathy, S., Shahi, D., Blanchard, B., Pontif, M., Gravois, K., Kimbeng, C., Hale, A.L., Todd, J.R., Rao, A.R., Baisakh, N. 2022. Evaluation of models for utilization in genomic prediction in the Louisiana sugarcane breeding program. Agriculture. 12(9):1330. https://doi.org/10.3390/agriculture12091330.
DOI: https://doi.org/10.3390/agriculture12091330

Interpretive Summary: Sugarcane is used for the production of sugar and biofuels. Varieties are bred for optimal yields of sugar and biomass, but the breeding process is slow and labor intensive. In other crops, DNA markers are heavily utilized, but the genetics of sugarcane are complex and have made the development of new techniques for variety improvement difficult. In this study, the use of various statistical methods to determine the ability of DNA markers to predict yield in sugarcane more quickly than traditional breeding methods are evaluated. The results showed that some methods worked better than others, but there is potential to improve the speed in which new sugarcane varieties are selected.

Technical Abstract: Sugarcane (Saccharum spp.) is an important perennial grass crop for both the sugar and biofuel industries. Sugarcane breeding programs focus on the improvement of economic traits through incremental genetic gain. With the advancement in high-throughput genotyping and phenotyping techniques, genomic selection emerged as a promising marker-assisted breeding tool. In this study, we assessed ridge regression best linear unbiased prediction (rrBLUP) and Bayesian models to evaluate genomic prediction accuracy using a 10-fold cross validation on 95 commercial and elite parental clones from the Louisiana sugarcane variety development program. We constructed datasets of 3,906 SNPs based on soil type (light – silty loam, heavy – clay) and crop type (plant cane, ratoon) to predict the genomic estimated breeding values (GEBVs) of the clones for stalk sucrose concentration and sugar and cane yield per hectare. We used Spearman’s rank correlation and Pearson’s correlation between phenotypic breeding values and GEBVs to measure prediction accuracy. Traits exhibited moderate to high heritability. Prediction accuracy based on rank correlation was high in all cross-effect prediction models with soil and crop types as fixed effects. In general, Bayesian models demonstrated a higher correlation than rrBLUP. Pearson’s correlation without soil and crop type as fixed effects was lower with no clear pattern among the models. The results demonstrate potential of genomic prediction in the Louisiana sugarcane variety development program.