Table 6. Prediction accuracy for winter survival of switchgrass populations in Validation data set #2. Predictions were based on GBLUP methods applied to training data sets from Spring 2014 or 2015 and applied to independent populations evaluated in 2017 at Arlington or Madison, WI.

				All individuals		Putative tetraploids only
Training set	GBLUP prediction methoda	Number of families	Number of SNP markers	Spring 2014	Spring 2015	Spring 2014	Spring 2015
All Populations	Individual	368	200,933	0.05	0.19	0.17	0.01
	Arlington			0.34	0.14	0.46	0.17
	Madison			0.05	0.14	0.15	0.00
All Lowland	Individual	257	176,581	0.01	0.13	0.20	0.01
	Arlington			0.34	0.14	0.46	0.17
	Madison			0.05	0.14	0.15	0.00
Southern Lowland	Individual	168	141,344	0.51	0.19	0.51	0.25
	Arlington			0.79	0.46	0.84	0.49
	Madison			0.65	0.24	0.64	0.28

^aIndividual: Prediction of individuals that were exactly sequenced. 178 Individuals representing 23 populations from location Madison were sequenced and All individuals comprised all 178 individuals and putative tetraploids comprised 138 individuals. Arlington: prediction of population GEBV at the Arlington location (23 populations for All individuals and 18 populations for putative tetraploids only); Madison: prediction of population GEBV at the Madison location (23 populations for All individuals and 18 populations for putative tetraploids only). GEBV = genomic estimated breeding value.