Winter Hardiness and Freezing Tolerance of Winter Faba Bean (Vicia faba L.): From QTL Identification and Validation to Genome-Wide Predictions of Genotypic Values
Doctoral thesis
Date of Examination:2025-03-28
Date of issue:2025-06-11
Advisor:Prof.Dr. Wolfgang Link
Referee:Prof. Dr. Wolfgang Link
Referee:Dr. Christian Möllers
Files in this item
Name:PhD_Thesis_Alex_Windhorst_23.05.2025.pdf
Size:7.58Mb
Format:PDF
Abstract
English
The growing interest in plant-based proteins and a more sustainable agricultural system, with wider crop rotations, reduced fertilizer input, and increased crop and fauna diversity, has intensified in response to climate change. Faba bean (Vicia faba L.), as locally adapted cool-season grain legume, is a viable alternative to unsustainable soybean imports, and its cultivation area is continuously increasing in Europe countries. Although autumn sown winter faba bean offers a significant yield advantage of up to 47% over spring faba bean, large scale cultivation is currently limited to France and the UK. The primary constraint is insufficient winter hardiness, as cold winters with frost below -12 °C pose a serious risk, often resulting in winterkill and substantial yield losses. To date, breeding progress for improved winter hardiness has been limited. Field-based selection depends on well-differentiating winters, which are rare. Consequently, artificial screening methods for assessing freezing tolerance have been developed to facilitate QTL identification and support selection decisions. Freezing tolerance is the major component of winter hardiness. Despite the broad genetic variation for freezing tolerance and winter hardiness present in the genuine winter faba bean gene pool, research and breeding efforts have been hindered by a lack of genomic tools and methods. Accordingly, QTL identification and SNP marker development for marker-assisted selection (MAS) could markedly enhance genetic gain for these traits in winter faba bean breeding. The recently published V. faba reference genome, along with the availability of low-cost, high-density genotyping platforms, could further accelerate genetic gain by enabling genome-wide predictions of genotypic values. As climate change progresses, milder winters with less predictable but severe frosts and late-frost events are likely to become more frequent. Therefore, tolerance to late-frost should be considered equally important as tolerance to frost in winter. The objective of this dissertation was to decipherer the quantitative genetic architecture of winter-frost and late-frost tolerance in European winter faba bean using state-of-the-art genomic approaches and methods. In Chapter 2, the genetic variation for late-frost tolerance in winter faba bean was investigated. For this purpose, the screening method, which was developed for the investigation of winter-frost tolerance, had to be adapted to enable sufficient dehardening and thus an accurate assessment of the late-frost stress response. A set of 188 winter faba bean inbred lines was studied in a series of seven late-frost treatment experiments, revealing significant genetic variation for late-frost tolerance. This genetic variance provides opportunities for breeding to enhance the level of late-frost tolerance in winter faba bean. The high but incomplete correlation among the three freezing-related sub-traits, combined with the trait-specific marker-trait associations identified via GWAS, confirmed the highly quantitative nature of late-frost tolerance. However, large-effect QTLs were identified, suggesting the potential for substantial genetic gain through MAS when applied in breeding populations for selection or allele introgression. In this regard, identification of at least two pleiotropic QTLs affecting multiple freezing-related sub-traits offers promising MAS targets. Moreover, prediction of genotypic values based on GWAS-estimated marker effects proved effective for multi-marker-based MAS, achieving prediction abilities of up to 0.57. To further dissect the genetic architecture of freezing tolerance, the comprehensive set of 13 freezing-related sub-traits and additional morphological traits were investigated for both late-frost tolerance and winter-frost tolerance in Chapter 3. Previously published detailed phenotypic data on winter-frost tolerance were used to facilitate an in-depth comparison of these traits in the same set of 188 inbred lines. Several large-effect QTLs were identified as treatment-unspecific, indicating their involvement in both winter-frost and late-frost tolerance. However, most of the phenotypic variance for winter-frost and late-frost tolerance was explained by treatment-specific QTLs. Apparently, tolerance to winter-frost, i.e., in hardened plants, is controlled by both distinct and overlapping genetic mechanisms as the tolerance to late-frost, i.e., in dehardened plants. Furthermore, both frost stress symptoms, such as loss of turgor and color in leaves, and the survival under severe frost seem to be affected by a combination of trait-specific and major pleiotropic QTLs. In addition, the reliability of GWAS results was evaluated through a two-step validation approach in an independent set of 64 winter faba bean lines, which were also phenotyped for winter-frost and late-frost tolerance. Several marker-trait associations were verified in this independent dataset, highlighting associated QTLs as high-confidence targets for candidate gene identification. Indeed, credible candidate genes for freezing tolerance involved in vernalization were identified within the major pleiotropic QTLs, corroborating previous findings in other legumes. However, the two-step validation approach also exposed limitations of the GWAS study design and emphasized the need for validating GWAS results in independent genetic background such as a breeding population. Despite the identification of numerous large-effect QTLs based on high-quality phenotypic data, a substantial proportion of the observed phenotypic variance for both winter-frost and late-frost tolerance remained unexplained. To circumvent the need for significance testing of marker-traits associations and QTL identification, the potential of genomic prediction for marker-based selection for freezing tolerance was evaluated in Chapter 4. This evaluation also includes several agronomically important traits, such as yield, derived from a comprehensive 16-year historical field trial dataset, as genomic prediction has not yet been extensively studied in faba bean. A simple genomic best linear unbiased prediction (GBLUP) model achieved reasonable prediction abilities (>0.35) for almost all freezing-related sub-traits and agronomic traits, even with relatively small training population sizes. These results suggest a high potential for implementing genomic selection in (winter) faba bean breeding. The number of SNPs used in the prediction model had only limited effect on the prediction ability if not reduced below 5,000 SNPs. However, a clear decline in prediction ability was observed when genetic relatedness between the training population and the predicted genotypes decreased. Both findings provide important information for consideration of practical implementation. Additionally, predicting a genotype’s performance in "new" environments (i.e., environments not included in the training dataset) remains a major challenge, as genotype-by-environment interactions and limited heritability values significantly affect prediction ability in such "across-cycle" prediction scenarios. Finally, genomic prediction of freezing-related sub-traits could also be integrated into an indirect genomic selection strategy, using these sub-traits as proxies for winter hardiness in the field. Preliminary evaluations of this indirect genomic selection approach predicted a promising acceleration of genetic gain for winter hardiness. In conclusion, this dissertation represents a significant contribution to the understanding of the genetic architecture of freezing tolerance in winter faba bean. Artificial screening for freezing tolerance played a key role in this research. Nevertheless, its advantages and limitations for both research and breeding were critically discussed in Chapter 5. In addition, utilization of genome-wide markers enabled QTL identification for freezing tolerance at a resolution which has not been possible in faba bean before. A large set of verified high-confidence markers was identified, offering valuable resources for candidate gene search and MAS. However, application of GWAS-derived markers in MAS has certain limitations that should be considered. With both GWAS and genomic prediction now available, the optimal integration of these approaches into (winter) faba bean breeding programs must be evaluated to drive genetic gain. Chapter 5 provides an outlook on how the performance of genomic prediction could be improved by integrating the verified marker into the prediction model. Additionally, several concepts and ideas for effectively incorporating genomic prediction into a line breeding program are proposed to enable accelerating genetic gain in (winter) faba bean breeding.
Keywords: Faba bean; GWAS; Winter hardiness; MAS; QTL; Breeding; Genomic Prediction; Genomic Selection; Freezing tolerance; Marker score; Abiotic stress