Approaches to implement genomic selection in three Swiss apple breeding programs

M. Jung, S. Bühlmann-Schütz, M. Hodel, M. Kellerhals, N. Bolliger, M. Köhle, M. Kobelt, H. Muranty, B. Studer, G.A.L. Broggini, A. Patocchi
Apple (Malus × domestica) production is challenged by the changing climate, market demands, consumer preferences and the pressure to minimize the use of plant protection products. These factors require apple breeding to be flexible, fast, and efficient. Genomic selection has proven useful to increase the efficiency of selection for suitable traits. Three breeding institutions in Switzerland (Agroscope, Poma Culta, Lubera) united efforts to implement genomic selection in their breeding programs. This implementation is built upon the genomic and phenotypic data set of the multi-national apple reference population (apple REFPOP) that was previously deemed suitable for the application of genomic selection in breeding. The 534 apple REFPOP genotypes (269 accessions and 265 progeny from 27 parental combinations) were combined with 547 genotypes of three Swiss breeding programs (340 progeny from 23 parental combinations and 207 advanced selections). The available SNP data set of the apple REFPOP was complemented with genomic data of the three programs obtained using the Illumina Infinium® 20K SNP genotyping array. To study the impact of a temporal overlap between phenotypic information available for the apple REFPOP from years 2018-2020 and phenotypes of the Swiss breeding germplasm collected in 2021, the Swiss copy of the apple REFPOP was phenotyped again in 2021. These phenotypic data were connected to the genomic data set in a genomic prediction analysis, with progeny of the breeding programs as validation set and the training set composed of: i) the apple REFPOP with phenotypes from 2018 to 2020, ii) the apple REFPOP with phenotypes from 2018 to 2021, iii) and iv) the apple REFPOP (2018-2021) and advanced selections or progeny, and v) a set based on all available genotypes and phenotypes. Identification of the most suitable prediction scenario(s) using the reported genomic predictive abilities for key quantitative traits will provide guidance on application of genomic selection in apple breeding.
Jung, M., Bühlmann-Schütz, S., Hodel, M., Kellerhals, M., Bolliger, N., Köhle, M., Kobelt, M., Muranty, H., Studer, B., Broggini, G.A.L. and Patocchi, A. (2023). Approaches to implement genomic selection in three Swiss apple breeding programs. Acta Hortic. 1362, 131-138
DOI: 10.17660/ActaHortic.2023.1362.18
https://doi.org/10.17660/ActaHortic.2023.1362.18
genomic prediction, quantitative traits, Malus × domestica, reference population, training set design
English

Acta Horticulturae