417-16 Exploring GxE and Genomic Prediction in a Two-Row Barley Pilot Study.

Poster Number 614

See more from this Division: C01 Crop Breeding & Genetics
See more from this Session: Crop Breeding and Genetics: III

Wednesday, November 18, 2015
Minneapolis Convention Center, Exhibit Hall BC

Jeffrey Neyhart, University of Minnesota, St. Paul, MN and Kevin P. Smith, 1991 Upper Buford Circle, University of Minnesota, St. Paul, MN
Poster Presentation
  • jneyhart_CSSA_2015_poster.pdf (1.9 MB)
  • Abstract:
    Rapid changes in the malting and brewing industries and the accompanying demand for locally-grown barley have stimulated a nationwide interest in producing barley across a range of different environments. Breeding for such a large and diverse target population of environments is constrained by the tendency of breeding programs to impose more intense selection during earlier generations and in few environments. This is also true for genomic selection, where the line performance is predicted using a training population that may be tested in few environments. In the pilot year of a nationwide, multi-environment genomic selection project, we evaluated a training population (TP) consisting of 183 two-row barley breeding lines. The TP was tested in 11 environments across New York, Minnesota, and Montana for traits including height and heading date. A panel of 50 randomly selected progeny lines (C1R) was evaluated in a subset of environments. All lines were genotyped using GBS, and a novel variant-calling pipeline discovered 4,365 SNP markers. For heading and height, we estimated marker effects via RR-BLUP and performed cross-validation within environments and between pairs of environments. Prediction accuracy within environments was high for heading (r = 0.780) and moderate for height (r = 0.383). Accuracy between environments depended on both the training and validation environments for height, but only the validation environment for heading. For height, environment interaction effects measured by AMMI were associated with between-environment prediction accuracy. Prediction accuracy of the progeny set was lower than the cross-validation accuracy, but followed trait patterns. Although limited in size, we expect that this pilot will provide insight into the analysis of a larger, more environmentally-diverse trial.

    See more from this Division: C01 Crop Breeding & Genetics
    See more from this Session: Crop Breeding and Genetics: III