PREDICT: Omics-basierte Modelle zur Vorhersage der Ertragsleistung von Rapshybriden
Zusammenfassung der Projektergebnisse
Hybrid vigour, or “heterosis”, is an important phenomenon for improving the yield performance of outcrossing crops. In oilseed rape, the most important vegetable oil crop in Europe and the second-most important worldwide, hybrids today make up the major share of the international seed market. However, in comparison to other important hybrid seed crops like maize, oilseed rape shows relatively low levels of hybrid vigour. The oilseed rape breeding industry therefore has considerable interest in a more optimal utilisation of heterosis for improvement of hybrid performance. Heterosis is a very complex trait which is genetically difficult to select for in breeding. The PREDICT project aimed to test and develop new methods for prediction of hybrid performance based on data from modern genomics techniques, which can circumvent the need to evaluate all breeding lines or hybrids in expensive, time-consuming multi-environment field trials. The project implemented comprehensive phenotypic datasets for elite spring oilseed rape (canola) hybrids from parental lines that showed contrasting suitability as parents for hybrid breeding. Plant materials and phenotype data were generated and provided by two cooperating (non-funded) commercial partners. Genome-wide single-single-nucleotide polymorphism (SNP) markers were surveyed in 475 elite parental lines, using a high-high-throughput commercial genotyping tool, to elucidate population structure in the parental hybrid pools and obtain detailed information about genome-wide diversity patterns. Large-scale automated phenotype data for early developmental traits, along with global metabolite and gene expression profiles, were generated from seedlings of the same parental lines grown in a climatised greenhouse. A set of 950 F1 hybrids generated from different combinations of these thoroughly characterised parental lines were subjected to detailed field-based yield evaluations of agronomic traits, important seed quality traits and grain yield. Different statistical models were tested and adapted for prediction of hybrid performance based on different combinations of morphological, sequence-based (SNP), metabolite and/or transcriptome markers in seedlings of parental lines. The underlying hypothesis was that gene and/or metabolite expression should better capture the complex interactions between all genes of an individual in response to environmental factors experienced during plant growth, hence such datasets should help improve the accuracy of performance predictions to a greater extent than predictions based only on genome-wide DNA profiles. For the economically important target traits seed yield and oil content, a significant but small increase in prediction accuracy was obtained by adding parental mRNA data to the SNP datasets. However, this improvement is unlikely to be relevant for commercial breeding, because predictions using only genomewide DNA marker profiles also gave robust, stable and accurate predictions for most traits. Hence the added cost of mRNA profiling (using present-day technologies) is unlikely to be offset by sufficient commercial gains from higher breeding progress though the use of these more sophisticated methods. This means that presently-available genome-wide platforms are able to be effectively used by commercial breeders for improvement of complex traits in oilseed rape. The results further suggest that SNP markers are very efficient at capturing different complexities of genetic trait regulation in oilseed rape. We found that SNP-based hybrid performance predictions using Reproducing Kernel Hilbert Space (RKHS) prediction models provide an accurate, cost-effective alternative to large-scale phenotypic selection for various important traits. The methods and tools developed in the project are already being applied in the breeding activities of the associated partners, underlining the rapid transfer of these groundbreaking research methods into a commercial setting. A surprising, highly relevant finding of the genome-wide association analyses was the discovery of the temporal dynamics of developmentally-related traits. This knowledge opens the important new research concept “systems mapping” for analysis of the genetic architecture of trait expression considering the temporal succession of molecular plant developmental processes.
Projektbezogene Publikationen (Auswahl)
- PREDICT: Omics-based models for prediction of hybrid performance in Brassica napus (oilseed rape). Plant Organ Growth Symposium, Gent, Belgium, 10-12 March 2015
Knoch D, Abbadi A, Jan H, Klukas C, Leckband G, Meyer RC, Micic Z, Snowdon R, Ulrich M, Altmann T
- (2016) Genomic prediction of testcross performance in canola (Brassica napus). PLoS ONE 11: e0147769
Jan HU, Abbadi A, Lücke S, Nichols RA, Snowdon RJ
(Siehe online unter https://doi.org/10.1371/journal.pone.0147769) - Imprints of divergent evolution and artificial selection in oilseed rape breeding pools. Plant and Animal Genome XXIV, San Diego, CA, USA, 2016
Werner CR, Hatzig S, Jan HU, Qian L, Abbadi A, Leckband G, Snowdon R
- DFG-Predict: Omics-based prediction of hybrid performance and systems genetic analyses in spring oilseed rape. GPBC 2018 - German Plant Breeding Conference, 28.02.-02.03.2018, Wernigerode, Germany
Knoch D, Abbadi A, Bräutigam A, Grandke F, Himmelbach A, Meyer RC, Riewe D, Samans B, Snowdon R, Altmann T
- Dynamic growth QTL in Arabidopsis thaliana and Brassica napus. In: Book of Abstracts: ICAR 2018 - 29th International Conference on Arabidopsis Research, 25-29 June 2018, Turku, Finland. 342
Meyer RC, Knoch D, Weigelt-Fischer K, Abbadi A, Snowdon R, Altmann T
- (2019) Genome-facilitated breeding of oilseed rape. In: Liu S, Chalhoub B, Snowdon R (eds). The Brassica napus genome. Springer
Werner C, Snowdon R
(Siehe online unter https://dx.doi.org/10.1007/978-3-319-43694-4)
