A network modeling approach provides insights into the environment-specific yield architecture of wheat

Authors: De Witt, Noah; GUEDIRA, MOHAMMED - North Carolina State University; MURPHY, J. PAUL - North Carolina State University; MARSHALL, DAVID - Retired ARS Employee; MALTECCA, CHRISTIAN - North Carolina State University; Brown-Guedira, Gina

Submitted to: Genetics
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/1/2022
Publication Date: 5/10/2022
Citation: De Witt, N., Guedira, M., Murphy, J., Marshall, D., Maltecca, C., Brown Guedira, G.L. 2022. A network modeling approach provides insights into the environment-specific yield architecture of wheat. Genetics. 221:3. https://doi.org/10.1093/genetics/iyac076.
DOI: https://doi.org/10.1093/genetics/iyac076

Interpretive Summary: Wheat (Triticum aestivum) yield is impacted by a diversity of developmental processes which interact with the environment during plant growth. This complex genetic architecture complicates identifying quantitative trait loci that can be used to improve yield. Trait data collected on individual processes or components of yield have simpler genetic bases and can be used to model how quantitative trait loci generate yield variation. The objectives of this experiment were to identify quantitative trait loci affecting spike yield, evaluate how their effects on spike yield proceed from effects on component phenotypes, and to understand how the genetic basis of spike yield variation changes between environments. A 358 F5:6 recombinant inbred line population developed from the cross of LA-95135 and SS-MPV-57 was evaluated in 2 replications at 5 locations over the 2018 and 2019 seasons. The parents were 2 soft red winter wheat cultivars differing in flowering, plant height, and yield component characters. Data on yield components and plant growth were used to assemble a structural equation model to characterize the relationships between quantitative trait loci, yield components, and overall spike yield. The effects of major quantitative trait loci on spike yield varied by environment, and their effects on total spike yield were proportionally smaller than their effects on component traits. This typically resulted from contrasting effects on component traits, where an increase in traits associated with kernel number was generally associated with a decrease in traits related to kernel size. In all, the complete set of identified quantitative trait loci was sufficient to explain most of the spike yield variation observed within each environment. Still, the relative importance of individual quantitative trait loci varied dramatically. Path analysis based on coefficients estimated through structural equation model demonstrated that these variations in effects resulted from both different effects of quantitative trait loci on phenotypes and environment-by-environment differences in the effects of phenotypes on one another, providing a conceptual model for yield genotype-by-environment interactions in wheat.

Technical Abstract: Wheat (Triticum aestivum) yield is impacted by a diversity of developmental processes which interact with the environment during plant growth. This complex genetic architecture complicates identifying quantitative trait loci (QTL) that can be used to improve yield. Trait data collected on individual processes or components of yield have simpler genetic bases and can be used to model how QTL generate yield variation. The objectives of this experiment were to identify QTL affecting spike yield, evaluate how their effects on spike yield proceed from effects on component phenotypes, and to understand how the genetic basis of spike yield variation changes between environments. A 358 F5:6RILpopulation developed from the cross of LA-95135 and SSMPV-57 was evaluated in two replications at five locations over the 2018 and 2019seasons. The parents were two soft red winter wheat cultivars differing in flowering, plant height, and yield component characters. Data on yield components and plant growth were used to assemble a structural equation model (SEM) to characterize the relationships between QTL, yield components and overall spike yield. The effects of major QTL on spike yield varied by environment, and their effects on total spike yield were proportionally smaller than their effects on component traits. This typically resulted from contrasting effects on component traits, where an increase in traits associated with kernel number was generally associated with a decrease in traits related to kernel size. In all, the complete set of identified QTL was sufficient to explain most of the spike yield variation observed within each environment. Still, the relative importance of individual QTL varied dramatically. Path analysis based on coefficients estimated through SEM demonstrated that these variations in effects resulted from both different effects of QTL on phenotypes and environment-by-environment differences in the effects of phenotypes on one another, providing a conceptual model for yield genotype-by-environment interactions in wheat.