Using Finlay-Wilkinson Regression to Analyze Genotype–Environment Interaction for Biomass from Switchgrass Field Trials
Authors:
Li Zhang1* ([email protected]), Jason Bonnette1, Philip A. Fay2, Robert B. Mitchell3, Arvid R. Boe4, Felix B. Fritschi5, David B. Lowry6, Francis M. Rouquette Jr.7, Yanqi Wu8, Julie D. Jastrow9, Roser Matamala9, Thomas E. Juenger1
Institutions:
1Department of Integrative Biology, University of Texas–Austin; 2U.S. Department of Agriculture, Agricultural Research Service, Grassland Soil and Water Research Laboratory; 3Wheat, Sorghum, and Forage Research Unit, Agricultural Research Service, U.S. Department of Agriculture; 4Department of Agronomy, Horticulture, and Plant Science, South Dakota State University; 5Division of Plant Science and Technology, University of Missouri–Columbia; 6Department of Plant Biology, Michigan State University; 7Texas A&M AgriLife Research; 8Department of Plant and Soil Sciences, Oklahoma State University; 9Environmental Science Division, Argonne National Laboratory
Abstract
Switchgrass is a perennial warm season C4 grass native to much of North America and a promising biofuel feedstock candidate. It is common in most prairies and exhibits extensive variability and adaptation across its range, especially related to latitude and precipitation gradients. Much of this variability is associated with evolved southern lowland and northern upland ecotypes.
This study utilized switchgrass biomass data from a structured mapping population and a diversity panel collected across 10 common gardens over multiple years. These field trial datasets were analyzed using the Finlay-Wilkinson regression (FW) approach to explore the genetic architecture of general vigor and environmental sensitivity. The FW method involves regressing the performance of each genotype against environmental means in a two-step procedure. The first step computes average plant performance at each site–year combination as a metric of environmental quality. The second step estimates the intercept and slope of each genotype regressed against the ordered environmental mean. The slope of the regression is a measure for adaptability and the intercept is a measure for general performance. The research team obtained an intercept, slope, and posterior standard deviation associated with the two parameters for each genotype of the two populations.
For the genetic mapping population, R/qtl2 software was used for quantitative trait locus (QTL) mapping while taking into consideration relatedness and individual weights. Individual weights in this case were calculated as n/(SD2), where n is the number of occurrences of that genotype across all the environments and SD is the posterior standard deviation for that genotype obtained from the FW regression. The team identified 23 QTLs associated with the intercept with the most significant QTL on chromosome 5N at marker position 84.03623 centimorgans (cM), and 11 QTLs associated with slope with the most significant QTL on chromosome 3N at marker position 77.91787 cM. For the Atlantic diversity panel, ASRgwas software was used to fit a linear mixed genome-wide association study model including a single nucleotide polymorphism (SNP)-based kinship matrix to control for population structure. A total of 4,128 SNPs were identified (p<5e-04) for the intercept with most significant SNPs showing signal on chromosomes 1N and 7K. For the slope, 2,725 SNPs were identified with the significant SNP primarily localized on chromosomes 5K and 7K. The extensive field trial dataset and analyses reveal several genomic regions and candidate genes impacting general vigor and environmental sensitivity. These data indicate direct strategies for improving high performing switchgrass cultivars across continental scale environmental variation.
Funding Information
This research is supported by the DOE Office of Science, BER Program, award no. DE-SC0021126.