How Does Population Size Affect the Estimation of Barley Stripe Rust Resistance QTL?.
M. Isabel Vales1, Chris C. Schön2, Flavio Capettini3, Xianming M. Chen4, Ann E. Corey1, Diane E. Mather5, Christopher C. Mundt1, Kelley L. Richardson1, J. Sergio Sandoval-Islas6, H. Friedrich Utz2, and Patrick M. Hayes1. (1) Oregon State University, Dept. of Crop and Soil Science, 107 Crop Science Bldg., Corvallis, OR 97331-3002, (2) University of Hohenheim, State Plant Breeding Institute, Stuttgart, 70593, Germany, (3) ICARDA/CIMMYT, Barley Program, Apdo. Postal 6-641, Mexico D.F., 6600, Mexico, (4) USDA ARS, 209 Johnson Hall, WSU, Pullman, WA 99164, (5) University of Adelaide / Molecular Plant Breeding Cooperative Research Centre, School of Agriculture and Wine, Glen Osmond SA, 5064, Australia, (6) Colegio de Postgraduados, Instituto de Fitosanidad, Montecillo, Mexico D.F., 56230, Mexico
The limited population sizes used in many quantitative trait locus (QTL) detection experiments can lead to underestimation of QTL number, overestimation of QTL effects, and failure to quantify QTL interactions. We used the barley/barley stripe rust pathosystem to evaluate the effect of population size on the estimation of QTL parameters. We generated a large (n = 409) population of doubled haploid lines derived from the cross of two inbred lines, BCD47 and Baronesse. This population was evaluated for barley stripe rust severity in the Toluca Valley, Mexico, and in Washington State, USA, under field conditions. BCD47 was the principal donor of resistance QTL alleles, but the susceptible parent also contributed some resistance alleles. The major QTL, located on the long arm of chromosome 4H, close to the Mlo gene, accounted for up to of 34% of the phenotypic variance. Subpopulations of different sizes were generated using three methods - resampling, selective genotyping, and selective phenotyping - to evaluate the effect of population size on the estimation of QTL parameters. In all cases, the number of QTL detected increased with population size. QTL with large effects were detected even in small populations, but QTL with small effects were detected only by increasing population size. Selective genotyping and/or selective phenotyping approaches could be effective strategies for reducing the costs associated with conducting QTL analysis in large populations. The method of choice will depend on the relative costs of genotyping vs. phenotyping.