Testing methods and statistical models of genomic prediction for quantitative disease resistance to Phytophthora sojae i
- PDF / 1,861,258 Bytes
- 14 Pages / 595.276 x 790.866 pts Page_size
- 17 Downloads / 193 Views
ORIGINAL ARTICLE
Testing methods and statistical models of genomic prediction for quantitative disease resistance to Phytophthora sojae in soybean [Glycine max (L.) Merr] germplasm collections William R. Rolling1,2 · Anne E. Dorrance1,3 · Leah K. McHale1,4 Received: 28 May 2020 / Accepted: 4 September 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract Key message Genomic prediction of quantitative resistance toward Phytophthora sojae indicated that genomic selection may increase breeding efficiency. Statistical model and marker set had minimal effect on genomic prediction with > 1000 markers. Abstract Quantitative disease resistance (QDR) toward Phytophthora sojae in soybean is a complex trait controlled by many small-effect loci throughout the genome. Along with the technical and rate-limiting challenges of phenotyping resistance to a root pathogen, the trait complexity can limit breeding efficiency. However, the application of genomic prediction to traits with complex genetic architecture, such as QDR toward P. sojae, is likely to improve breeding efficiency. We provide a novel example of genomic prediction by measuring QDR to P. sojae in two diverse panels of more than 450 plant introductions (PIs) that had previously been genotyped with the SoySNP50K chip. This research was completed in a collection of diverse germplasm and contributes to both an initial assessment of genomic prediction performance and characterization of the soybean germplasm collection. We tested six statistical models used for genomic prediction including Bayesian Ridge Regression; Bayesian LASSO; Bayes A, B, C; and reproducing kernel Hilbert spaces. We also tested how the number and distribution of SNPs included in genomic prediction altered predictive ability by varying the number of markers from less than 50 to more than 34,000 SNPs, including SNPs based on sequential sampling, random sampling, or selections from association analyses. Predictive ability was relatively independent of statistical model and marker distribution, with a diminishing return when more than 1000 SNPs were included in genomic prediction. This work estimated relative efficiency per breeding cycle between 0.57 and 0.83, which may improve the genetic gain for P. sojae QDR in soybean breeding programs.
Introduction
Electronic supplementary material The online version of this article (https://doi.org/10.1007/s00122-020-03679-w) contains supplementary material, which is available to authorized users. * Leah K. McHale [email protected] 1
Center for Applied Plant Science and Center for Soybean Research, The Ohio State University, Columbus, OH 43210, US
2
Present Address: Vegetable Crop Research Unit, USDA-ARS, Madison, WI 53706, US
3
Department of Plant Pathology, The Ohio State University, Wooster, OH 44691, US
4
Department of Horticulture and Crop Science, The Ohio State University, Columbus, OH 43210, US
Breeding to improve yield and quality of crop products has been beneficial to providing food, fiber, feed, fuel,
Data Loading...
