CHARACTERIZING RESISTANCE TO WHEAT STEM SAWFLY (CEPHUS CINCTUS NORTON) IN HARD WINTER WHEAT (TRITICUM AESTIVUM L.)

Abstract

Wheat stem sawfly (WSS) is a native grass feeding pest of winter wheat (Triticum aestivum L) which is difficult to control since most of its life cycle occurs within the stem of wheat plants. The only well-characterized genetic resistance to WSS is the solid stem locus (Sst1) on chromosome 3B, which exhibits environmental variability. It is critical to further characterize the solid stem trait and identify novel forms of genetic resistance outside of Sst1 to improve the overall resistance of wheat to WSS. In the second chapter, genome-wide association studies (GWAS) were performed on lines in the Colorado State University wheat breeding program grown between 2014 and 2025 field seasons for three traits of interest: heading date (HD), WSS damage in the form of stem cutting (CUT), and stem solidity (SOLID). Significant marker-trait associations (MTA) from these GWAS were used to identify beneficial allelic combinations (AC) for WSS resistance. The stem cutting ACs which had the lowest damage estimates were those in which lines possessed the resistant haplotype at every locus assessed (CUT = 2.35, error = 0.21, N = 56). Lines that had all resistant alleles in the stem solidity ACs showed the same superior estimate (SOLID = 14.7, error = 0.78, N = 22). MTA for GWAS and AC models did not overlap, indicating the presence of WSS resistance in the population that is not associated with stem solidity. Understanding these different AC groups and their effects on stem solidity and cutting can improve efficiency in breeding for WSS resistance. Despite the large effect of SSt1, additional loci associated with these WSS resistant phenotypes have previously been identified. This oligogenic genetic control lends itself to genomic prediction. In chapter three, best linear unbiased prediction (GBLUP), GBLUP+SSt1, Bayesian (Bayes A and BayesLASSO), and random forest models were assessed for predictive ability using cross- and forward-validation. Additionally, environments were culled based on heritability to evaluate the effect of training population heritability on predictive ability. Overall, random forest models performed well in cross- and forward-validation for cutting (r ̅_CV = 0.67, r ̅_FV = 0.45) and GBLUP performed well for solidity (r ̅_CV = 0.68, r ̅_FV = 0.64). Increasing heritability of the training population did not improved cross-validation predictive ability for cutting (r ̅_BayesA = 0.75, r ̅_(random forest) = 0.77) or solidity (r ̅_GBLUP = 0.85, r ̅_BayesA = 0.86). For forward-validation, increasing heritability did not improve predictive ability for cutting (r ̅_BayesA = 0.39, r ̅_(random forest) = 0.43) and decreased predictive ability for solidity (r ̅_GBLUP = 0.54, r ̅_BayesA = 0.55). These results highlight the need to identify models best suited to the genetic architecture of individual traits of interest. In chapters four and five, two pair of near isogenic lines (NILs) were grown over two field seasons to understand the metabolome, proteome, and phytohormone profile differences in response to WSS infestation in hollow stem and solid stem wheat. It has been reported that Clearfield®(CL) wheat varieties exhibit lower WSS infestation. To further understand potential molecular differences between CL non-CL varieties, chapter four utilized the herbicide tolerant line ‘Byrd CL Plus’ and non-herbicide tolerant line ‘Byrd’. Stem samples were collected for three replicates from each NIL during pre-, peak, and post- infestation time points. Five differentially expressed proteins were identified across all three time points. At peak infestation, two protein accessions A0A3B6JCK1 (Log2FC = -3.48, p-valueFDR < 0.01) and A0A3B6NUD3 (Log2FC = -2.30, p-valueFDR < 0.01) were found to be associated with the flowering gene FLOWERING LOCUS C (FLC) and glycolysis, respectively. One unannotated metabolite was upregulated during peak infestation (Log2FC = 3.20, p-valueFDR = 0.02). Additionally, exploratory analysis of significant metabolites (Log2FC ≥ |6|, p-valueunadjusted < 0.05) revealed strong correlations with metabolites and phytohormones, abscisic acid and jasmonate iso-leucine, which were associated with plant defense pathways. While these results potential low levels differences in plant responses and defenses, it also showed the complexity of the of plant responses to infestation and underlying defense mechanisms. To further investigate molecular differences between hollow stem and solid stem wheat varieties, chapter 5 focuses on a NIL pair containing the hollow stem line ‘CO20SFD047R.4’ and solid stem line ‘CO20SFD047R.1’ were utilized in the same study as chapter 4. Post-infestation, two protein accessions A0A3B6NPX7 (Log2FC = -3.32, p-valueFDR = 0.01) and A0A3B6SBN0 (Log2FC = -2.45, p-valueFDR = 0.02) were found to be associated with lipid transport and galactose transport and cell wall component synthesis, respectively. Additionally, the defense signaling phytohormone jasmonate isoleucine (JA-Ile) was significantly differentially expressed at peak infestation (Log2FC = 1.59, p-valueFDR = 0.02) and post-infestation (Log2FC = -1.46, p-valueFDR = 0.04). These results highlight the complex downstream effects associated with SSt1 resistance to WSS. Solid stem expression is variable and the tradeoff with yield component traits is not well understood. In chapter six, eight sets of near isogenic lines (NIL) were grown across six environments to investigate the impact of the SSt1 locus on 11 yield component traits and solidity. There were little to no consistent effects of the SSt1 locus on biomass traits, plant height, and heading date. Analysis supported the identification of three NIL groups, NIL1, NIL6, and NIL8 that were consistently solid across all environments with solidity estimates of 14.5, 14.5, and 12.9, respectively. However, grain yield in solid stem lines was 205 kg ha-1 greater than the hollow stem line for NIL1 and 747 kg ha-1, and 260 kg ha-1 lower than the hollow stem allele for NIL6 and NIL8. These results highlighted NIL1 as being an ideal genetic background for the solid stem trait due to its reliable solidity and minimal yield-related penalties. Furthermore, these results indicate no consistent trade off in agronomic traits associated with the SSt1 locus across different genetic backgrounds. Further understanding the relationship between sawfly resistance and agronomic trade-offs could be beneficial for developing more elite WSS resistant cultivars.