Genomic Prediction for Winter Survival in Lowland Switchgrass.

Genomic Prediction for Winter Survival in Lowland Switchgrass Hari P. Poudel1, Guillaume P. Ramstein1, C. Robin Buell2, Michael D. Casler1,3; (1) Department of Agronomy, University of Wisconsin-Madison, Madison, WI; (2) Department of Plant Biology, and DOE Great Lakes Bioenergy Research Center, Michigan State University, East Lansing, MI; (3) USDA Dairy Forage Research Center, Madison, WI Switchgrass (Panicum virgatum L.) is a North American native perennial warm season grass and a promising cellulosic bioenergy feedstock. There are two ecotypes of switchgrass, lowland and upland. The lowland ecotype has generated considerable interest because of its higher biomass yield compared to the upland ecotype. However, lowland ecotypes planted in northern latitudes exhibit very low winter survival. Winter survival can be improved by selectively saving surviving plants, but this approach requires several years for the completion of one generation and its success is highly dependent on the occurrence of winter conditions that generate the appropriate selection pressure. Genomic selection (GS) has the potential to enhance switchgrass breeding by reducing selection time and eliminating the dependence on weather. In this study, genomic prediction was assessed on a total of 264 lowland half sib (HS) families and 121 lowland × upland HS families representing multiple geographic regions. We screened winter survival using a score based on visual assessment of the percentage of living shoots after initial spring growth on a scale of 0-20. The maternal parent of each HS family was genotyped using the exome capture sequencing and aligned to Version 1.1 of AP13 switchgrass reference genome. Prediction accuracies using all 385 HS families as a single training population were 0.57 and 0.53 using field data from 2014 and 2015 respectively. This model was tested for validation on a different dataset consisting of 1146 HS families representing 132 lowland and upland populations. The genomic estimated breeding values (GEBV) for winter survival decreased with a decrease in 30-year normal minimum temperature at the site-of-origin of the source population (p < 0.05). Our results suggest that winter survival of lowland switchgrass in Wisconsin can be predicted using genomic DNA markers and our GS model will be useful in both breeding lowland switchgrass for increased winter survivorship and in identifying target sites for additional lowland switchgrass germplasm collection.