About Us: Amber Hoffstetter, PhD Exit Seminar

Nov 13, 2015, 9:00am - 10:00am

Amber Hoffstetter, PhD Exit Seminar

Advisor: Dr. Clay Sneller

Assessing genome wide breeding strategies for economic traits in soft winter wheat and their impact on gain and genetic architecture

With next generation sequencing technology, such as genotyping-by-sequencing (GBS), breeders can now genotype large populations with thousands of markers.  This technology can be coupled with statistical methods such as genome-wide association studies (GWAS) and genomic selection (GS) to identify marker-trait associations and estimate marker effects.  Where GWAS studies estimate each marker separately and use a p-value to determine significance, GS ignores significant thresholds and uses a training population (TP) with phenotypic and genotypic data to estimate all markers simultaneously.  These effects are then used to predict the genomic estimated breeding values (GEBV) of other individuals.  We performed a GWAS analysis using an elite population of soft red winter wheat lines and identified 14 QTL for grain yield (GY), four for Fusarium Head Blight (FHB) index, four for flour yield (FY), and five for softness equivalence (SE) Across all traits the R2 values ranged from 1.8 to 3.5%.  We also determined the prediction accuracy GS for these four traits.  Using all markers and lines we found the prediction accuracies ranged from 0.35 (FHB) to 0.57 (GY, Wooster, Ohio).  In general using only data from TP lines with low GEI or marker subsets increased the GS accuracy.  When using the TP to predict the performance of the 23 parental lines, accuracies using weighted correlations based on the parent’s contribution to the TP produced the highest prediction accuracies (r = 0.08 to 0.85).  The accuracy of the TP model for predicting the phenotypes of the validation population was low (r = -0.25 to 0.22), especially for GY, but improved when using a subset of VP lines more related to the TP (r = 0.01 to 0.71).  When analyzing the impact of GS on diversity and linkage disequilibrium (LD) we found that there was a loss of diversity across the two cycles of GS and that the second cycle of GS (GC1) is more inbred than the TP.  LD for most marker pairs remains low across all three populations.  The correlation of R2 values across the three populations ranged from 0.46 to 0.65.  As LD between markers in the TP increases, a similar or higher LD is found with the F2 individuals comprising the two cycles of GS (GC0 and GC1).  The frequency of the 1 allele for majority (46%) of markers associated with GY in Wooster, Ohio decreases while the remaining markers have either the 1 allele increasing or remaining unchanged.  The preferred allele for these two trends is increasing for 95% and 80% of the markers respectively. The frequency of the 1 allele for individuals in the top 10% (best) and bottom 10% (worst) of the GC0 and GC1 individuals relative to the TP indicates that in the first cycle the majority (53%) of markers show signs of genetic drift while in the second cycle the majority (60%) show signs of direction selection.  The results of this work show that these two breeding strategies could be useful for the SRWW program here of Ohio State.  And indicates that GS impacts genetic diversity, LD, and allele frequencies.

Friday, November 13, 2015 9:00-10:00 AM
130 Research Services (Wooster) Video Linked to 202F Kottman Hall (Columbus)