SusCrop - Knowledge-driven genomic predictions for sustainable disease resistance in wheat

Progress in plant breeding is based on the ability to do precise selection of new offsprings with desired traits. Traditionally, this is done by testing of large populations in field trials over multiple years and locations. This is both costly and time consuming. Genomic selection makes it possible to use maker data to predict the breeding value of new breeding lines based on statistical models. There is a large potential to save both time and money if the models are reliable. The prediction models are built by genotyping a training population with thousands of markers and modelling the effects of these markers based on available phenotypic data for the same lines. The core idea behind WheatSustain was to incorporate the knowledge of known genes and their mode of action in the genomic prediction models that are used in plant breeding to make them more reliable. As cases for the project, we chose wheat, which is the largest cereal crop in Europe, and resistance to two important diseases, fusarium head blight (FHB) and stripe rust, which each illustrates important challenges for making genomic selection into an effective and reliable selection tool in plant breeding. The project was based on cross-disciplinary international collaboration among leading research groups and wheat breeding programs in Norway, Ireland, Germany, Austria, Mexico, USA and Canada. As training population for genomic prediction modelling for Norwegian wheat breeding, we used our existing panels of 300 spring wheat lines and 100 winter wheat lines (MASBASIS). These were genotyped with the 25K SNP chip together with 300 new breeding lines from Graminor that were used for validation of the developed prediction models. These materials were all tested in field trials in the 2019, 2020 and 2021 seasons in Norway and by selected project partners in 2020 and 2021. Through genetic analyses of the data from inoculated Fusarium nurseries in Norway, Austria and Canada we obtained a good overview of the genetic architecture of FHB resistance in the Norwegian spring wheat breeding material. The main genetic factors governing resistance was detected and we validated an important QTL on chromosome 7A that will be useful for future resistance breeding. In a similar manner, we obtained a good overview of the most important genetic factors for stripe rust resistance in the Norwegian breeding material based on data from field trials in Norway, Germany, Austria and China. Moreover, a major gene on chromosome 6A for adult plant resistance was validated by use of data from new breeding lines. In 2022 and 2023 we worked on integrating this genetic information into improved prediction models for FHB and stripe rust resistance in Norwegian spring wheat breeding. To find optimal way of applying the improved genomic prediction models in actual wheat breeding programs, a simulation program was developed evaluate plant breeding schemes in silico using stochastic simulations. By simulating a conventional spring wheat breeding program based on actual phenotypic and genotypic data we could show that integration of speed breeding with genomic selection will lead to faster genetic gains for FHB resistance than genomic selection or phenotypic selection alone. The consortium also worked on a common winter wheat training population consisting of 230 winter wheat lines from the collaborating breeding companies and research institutions. This common training population was tested for FHB and stripe rust resistance in field trials at selected locations in Germany, Austria and Norway in 2020 and 2021. Based on this testing, solid multi-environmental data sets were obtained for both diseases, which were used in genetic analyses to reveal the genetic architecture of the traits and validate resistance genes in new breeding lines from the collaborating breeding programs. For yellow rust we found two major QTL on chromosomes 2B and 6A, and validation of the results on new breeding lines from the breeding programs in Germany showed that these two QTL in combination give good protection against the disease. Further, our genomic prediction modeling showed that genomic selection could be made more effective by including markers for the two QTL in the prediction models. For FHB in winter wheat, the association mapping highlighted that the disease resistance is closely associated with plant height, anthesis date and anther retention. By including these factors in the prediction models, the accuracy of genomic selection can be improved. The close collaboration with the plant breeders in this project was very successful and resulted in validated molecular markers and genomic prediction models that are already being utilized both in public and commercial plant breeding. Results from the project was presented at many conferences several scientific papers have already been published while the remaining ones will be completed during the coming months.

- Mer effektiv foredling av resistens mot aksfusariose og gulrust i hvete ved hjelp av genomisk seleksjon - Mer effektiv planteforedling generelt gjennom forbedrede statistiske modeller for genomisk prediksjon - Lavere mykotoksininnhold i framtidig norsk og europeisk hveteproduksjon ved bruk av nye sorter med bedre Fusariumresistens - Redusert behov for sprøyting mot fungicider i norsk og europeisk hveteproduksjon ved bruk av nye sorter med bedre resistens mot gulrust - Økt lønnsomhet og mindre klimaavtrykk i norsk og europeisk hveteproduksjon gjennom større arealproduktivitet og mindre behov for soppsprøyting

Genomic selection (GS) enables the prediction of breeding values of progeny lines without costly phenotyping, saving time and money, increasing intensity of selection as well as accuracy of trait prediction. The core idea behind this project is to make use of biologically relevant data, quantitative trait loci (QTL) and marker-trait relationships to improve prediction accuracy. We seek to bridge the gap between marker-assisted selection (MAS) based on known markers for well-characterized genes and GS based on anonymous markers by incorporating prior knowledge into genomic prediction models. To highlight the potential of this approach, we have chosen two of the most devastating plant diseases affecting European and North American wheat production - FHB and stripe rust. The WheatSustain project will develop new multi-trait models and methods for GS of disease resistance by incorporating disease resistance loci with known effects, insight into host-pathogen interactions, race specificity of resistance genes, and genetic correlations among traits. The developed models will be tested and validated on actual wheat breeding material from collaborating breeding programs, and continuously improved by close interaction with the respective breeders. WheatSustain will establish a close collaboration among world leading experts on genomic prediction modeling in plants and animals, bioinformatics, wheat genomics and leaders in the field of plant pathology and host-pathogen relationships for stripe rust and FHB resistance in wheat, involving groups from Norway, Ireland, Germany, Austria, Mexico, USA and Canada. Plant breeders will provide germplasm with phenotypic and genotypic data, take part in disease valuations and test out the developed breeding methodologies in their breeding programs. This dynamic research environment will also provide excellent research training for PhD students and postdocs, thereby advancing the education of future researchers and plant breeders.