Sebastian Michel, Franziska Löschenberger, Christian Ametz & Hermann Bürstmayr 

Theoretical and Applied Genetics April 2023; vol. 136, Article number: 79

Key message

A novel genomic selection strategy offers the unique opportunity to develop qualitative race-specific resistant varieties that possess high levels of the more durable quantitative race-nonspecific resistance in their genetic background.

Abstract

Race-specific qualitative resistance genes (R-genes) are conferring complete resistance in many pathosystems, but are frequently overcome by new virulent pathogen races. Once the deployed R-genes are overcome, a wide variation of quantitative disease resistance (QDR) can be observed in a set of previously race-specific, i.e., completely resistant genotypes—a phenomenon known as “vertifolia effect.” This race-nonspecific QDR is considered to be more durable in the long term, but provides merely a partial protection against pathogens. This simulation study aimed to detangle race-specific R-gene-mediated resistance of pending selection candidates and the QDR in their genetic background by employing different genomic selection strategies. True breeding values that reflected performance data for rust resistance in wheat were simulated, and used in a recurrent genomic selection based on several prediction models and training population designs. Using training populations that were devoid of race-specific R-genes was thereby pivotal for an efficient improvement of QDR in the long term. Marker-assisted preselection for the presence of R-genes followed by a genomic prediction for accumulating the many small to medium effect loci underlying QDR in the genetic background of race-specific resistant genotypes appeared furthermore to be a promising approach to select simultaneously for both types of resistance. The practical application of such a knowledge-driven genomic breeding strategy offers the opportunity to develop varieties with multiple layers of resistance, which have the potential to prevent intolerable crop losses under epidemic situations by displaying a high level of QDR even when race-specific R-genes have been overcome by evolving pathogen populations.

See https://link.springer.com/article/10.1007/s00122-023-04312-2

Figure 4: Average disease severity in the simulated recurrent selection schemes for the genetic foreground comprising both race-specific and race-nonspecific disease resistance, i.e., when race-specific R-genes were considered to be effective in the training populations as well as in the progeny populations (solid lines; closed circles) as well as the genetic background comprising solely the race-nonspecific quantitative disease resistance, i.e., when race-specific R-genes would have been overcome by the pathogen and not be effective anymore in the progeny populations (dashed lines; open circles). It was assumed that the race-specific R-genes are still effective in the entire training population when fitting genomic best linear prediction models with a mixed training population of race-specific and race-nonspecific resistant genotypes (GBLUPMIXGBLUPMIX), which were compared with models including fixed effects for the most significant marker–trait associations that were either upweighted (WBLUPMIXWBLUPMIX) or unweighted (mWBLUPMIXmWBLUPMIX) in the computation of the genomic estimated breeding values. The potential of training population devoid of race-specific resistant lines was furthermore tested by removing the respective lines based on mapped marker–trait associations (GBLUPQDR(MTA)GBLUPQDR(MTA)) or with a disease severity smaller than x = 10%, x = 20%, x = 30%, x = 40%, or x = 50% (GBLUPQDR(CULLx)GBLUPQDR(CULLx)). The best 20 crosses among all 400 parents within each selection cycle were selected based on the genomic estimated mid-parent values obtained by the different model-by-training population combinations (A–C; baseline strategy) or after a marker-assisted preselection for the absence (D–F;−R-gene strategy) or presence (G–I; + R-gene strategy) of race-specific R-gene-mediated resistance among the potential parents. Result are shown for a resistance allele frequency of 0.2–0.3 at the simulated R-genes in the founder population