Unlocking historical phenotypic data from an ex situ collection to enhance the informed utilization of genetic resources of barley (Hordeum sp.).

Key message Historical data generated during seed regeneration are valuable to populate a bio-digital resource center for barley (Hordeum sp.). Precise estimates of trait performance of genetic resources are considered as an intellectually challenging, complex, costly and time-consuming step needed to exploit the phenotypic and genetic diversity maintained in genebanks for breeding and research. Using barley (Hordeum sp.) as a model, we examine strategies to tap into historical data available from regeneration trials. This is a first step toward extending the Federal ex situ Genebank into a bio-digital resource center facilitating an informed choice of barley accessions for research and breeding. Our study is based on historical data of seven decades collected for flowering time, plant height, and thousand grain weight during the regeneration of 12,872 spring and winter barley accessions. Linear mixed models were implemented in conjunction with routines for assessment of data quality. A resampling study highlights the potential risk of biased estimates in second-order statistics when grouping accessions for regeneration according to the year of collection or geographic origin. Based on rigorous quality assessment, we obtained high heritability estimates for the traits under consideration exceeding 0.8. Thus, the best linear unbiased estimations for the three traits are a valuable source to populate a bio-digital resource center for the IPK barley collection. The proposed strategy to leverage historical data from regeneration trials is not crop specific and can be used as a blueprint for other ex situ collections.

Large-scale genotyping and phenotyping of a worldwide winter wheat genebank for its use in pre-breeding.

Plant genetic resources (PGR) stored at genebanks are humanity's crop diversity savings for the future. Information on PGR contrasted with modern cultivars is key to select PGR parents for pre-breeding. Genotyping-by-sequencing was performed for 7,745 winter wheat PGR samples from the German Federal ex situ genebank at IPK Gatersleben and for 325 modern cultivars. Whole-genome shotgun sequencing was carried out for 446 diverse PGR samples and 322 modern cultivars and lines. In 19 field trials, 7,683 PGR and 232 elite cultivars were characterized for resistance to yellow rust - one of the major threats to wheat worldwide. Yield breeding values of 707 PGR were estimated using hybrid crosses with 36 cultivars - an approach that reduces the lack of agronomic adaptation of PGR and provides better estimates of their contribution to yield breeding. Cross-validations support the interoperability between genomic and phenotypic data. The here presented data are a stepping stone to unlock the functional variation of PGR for European pre-breeding and are the basis for future breeding and research activities.

Genomics-informed prebreeding unlocks the diversity in genebanks for wheat improvement.

The great efforts spent in the maintenance of past diversity in genebanks are rationalized by the potential role of plant genetic resources (PGR) in future crop improvement-a concept whose practical implementation has fallen short of expectations. Here, we implement a genomics-informed prebreeding strategy for wheat improvement that does not discriminate against nonadapted germplasm. We collect and analyze dense genetic profiles for a large winter wheat collection and evaluate grain yield and resistance to yellow rust (YR) in bespoke core sets. Breeders already profit from wild introgressions but PGR still offer useful, yet unused, diversity. Potential donors of resistance sources not yet deployed in breeding were detected, while the prebreeding contribution of PGR to yield was estimated through 'Elite × PGR' F1 crosses. Genomic prediction within and across genebanks identified the best parents to be used in crosses with elite cultivars whose advanced progenies can outyield current wheat varieties in multiple field trials.

Assessing the Suitability of Elite Lines for Hybrid Seed Production and as Testers in Wide Crosses With Wheat Genetic Resources.

The use of genetic resources in breeding is considered critical to ensure future selection gain, but the absence of important adaptation genes often masks the breeding value of genetic resources for grain yield. Testing genetic resources in a hybrid background has been proposed as a solution to obtain unbiased estimates of breeding values for grain yield. In our study, we evaluated the suitability of European wheat elite lines for implementing this hybrid strategy, focusing on maximizing seed yield in hybrid production and reducing masking effects due to susceptibility to lodging, yellow rust, and leaf rust of genetic resources. Over a 3-year period, 63 wheat elite female lines were crossed with eight male plant genetic resources in a multi-environment field experiment to evaluate seed yield on the female side. Then, the resulting hybrids and their parents were tested for plant height, lodging, and susceptibility to yellow rust and leaf rust in a further field experiment at multiple locations. We found that seed yield was strongly influenced by the elite wheat line choice in addition to environment and observed substantial differences among elite tester lines in their ability to reduce susceptibility to lodging, yellow rust, and leaf rust when the hybrid strategy was implemented. Consequently, breeders can significantly increase the amount of hybrid seed produced in wide crosses through appropriate tester choice and adapt genetic resources of wheat with the hybrid strategy to the modern cropping system.

Reciprocal Recurrent Genomic Selection Is Impacted by Genotype-by-Environment Interactions.

Reciprocal recurrent genomic selection is a breeding strategy aimed at improving the hybrid performance of two base populations. It promises to significantly advance hybrid breeding in wheat. Against this backdrop, the main objective of this study was to empirically investigate the potential and limitations of reciprocal recurrent genomic selection. Genome-wide predictive equations were developed using genomic and phenotypic data from a comprehensive population of 1,604 single crosses between 120 female and 15 male wheat lines. Twenty superior female lines were selected for initiation of the reciprocal recurrent genomic selection program. Focusing on the female pool, one cycle was performed with genomic selection steps at the F2 (60 out of 629 plants) and the F5 stage (49 out of 382 plants). Selection gain for grain yield was evaluated at six locations. Analyses of the phenotypic data showed pronounced genotype-by-environment interactions with two environments that formed an outgroup compared to the environments used for the genome-wide prediction equations. Removing these two environments for further analysis resulted in a selection gain of 1.0 dt ha-1 compared to the hybrids of the original 20 parental lines. This underscores the potential of reciprocal recurrent genomic selection to promote hybrid wheat breeding, but also highlights the need to develop robust genome-wide predictive equations.

Unlocking big data doubled the accuracy in predicting the grain yield in hybrid wheat.

The potential of big data to support businesses has been demonstrated in financial services, manufacturing, and telecommunications. Here, we report on efforts to enter a new data era in plant breeding by collecting genomic and phenotypic information from 12,858 wheat genotypes representing 6575 single-cross hybrids and 6283 inbred lines that were evaluated in six experimental series for yield in field trials encompassing ~125,000 plots. Integrating data resulted in twofold higher prediction ability compared with cases in which hybrid performance was predicted across individual experimental series. Our results suggest that combining data across breeding programs is a particularly appropriate strategy to exploit the potential of big data for predictive plant breeding. This paradigm shift can contribute to increasing yield and resilience, which is needed to feed the growing world population.

Suitability of Single-Nucleotide Polymorphism Arrays Versus Genotyping-By-Sequencing for Genebank Genomics in Wheat.

Genebank genomics promises to unlock valuable diversity for plant breeding but first, one key question is which marker system is most suitable to fingerprint entire genebank collections. Using wheat as model species, we tested for the presence of an ascertainment bias and investigated its impact on estimates of genetic diversity and prediction ability obtained using three marker platforms: simple sequence repeat (SSR), genotyping-by-sequencing (GBS), and array-based SNP markers. We used a panel of 378 winter wheat genotypes including 190 elite lines and 188 plant genetic resources (PGR), which were phenotyped in multi-environmental trials for grain yield and plant height. We observed an ascertainment bias for the array-based SNP markers, which led to an underestimation of the molecular diversity within the population of PGR. In contrast, the marker system played only a minor role for the overall picture of the population structure and precision of genome-wide predictions. Interestingly, we found that rare markers contributed substantially to the prediction ability. This combined with the expectation that valuable novel diversity is most likely rare suggests that markers with minor allele frequency deserve careful consideration in the design of a pre-breeding program.

Aroma and quality of breads baked from old and modern wheat varieties and their prediction from genomic and flour-based metabolite profiles.

Bread aroma is the principal characteristic perceived by the consumer yet it is mostly disregarded in the product chain. The main aim of this study was to evaluate the potential to include bread aroma as a new target criterion into the wheat product chain. The objectives of our study were to (i) quantify the influence of genetic versus environmental factors on the bread aroma and quality characteristics, (ii) evaluate whether bread baked from modern wheat varieties differ in terms of aroma from those baked from old varieties, and (iii) compare genomic and metabolomic approaches for their efficiency to predict bread aroma and quality characteristics in a wheat breeding program. Agronomic characters as well as bread aroma and quality traits were assessed for 18 old and 22 modern winter wheat varieties evaluated at up to three locations in Germany. Metabolite profiles of all 120 flour samples were collected using a 7200 GC-QTOF. Considerable differences in the adjusted entry means for all examined bread aroma and quality characters were observed. For aroma, which was rated on a scale from 1 to 9, the adjusted entry means varied for the 40 wheat varieties between 3 and 8. In contrast, the aroma of bread prepared from old and modern wheat varieties did not differ significantly (P < 0.05). Bread aroma was not significantly (P < 0.05) correlated with grain yield, which suggested that it is possible to select for the former character in wheat breeding programs without reducing the gain of selection for the latter. Finally, we have shown that bread aroma can be better predicted using a combination of metabolite and SNP genotyping profiles instead of the SNP genotyping profile only. In conclusion, we have illustrated possibilities to increase the quality of wheat for consumers in the product chain.

Historical phenotypic data from seven decades of seed regeneration in a wheat ex situ collection.

Genebanks are valuable sources of genetic diversity, which can help to cope with future problems of global food security caused by a continuously growing population, stagnating yields and climate change. However, the scarcity of phenotypic and genotypic characterization of genebank accessions severely restricts their use in plant breeding. To warrant the seed integrity of individual accessions during periodical regeneration cycles in the field phenotypic characterizations are performed. This study provides non-orthogonal historical data of 12,754 spring and winter wheat accessions characterized for flowering time, plant height, and thousand grain weight during 70 years of seed regeneration at the German genebank. Supported by historical weather observations outliers were removed following a previously described quality assessment pipeline. In this way, ready-to-use processed phenotypic data across regeneration years were generated and further validated. We encourage international and national genebanks to increase their efforts to transform into bio-digital resource centers. A first important step could consist in unlocking their historical data treasures that allows an educated choice of accessions by scientists and breeders.

Grain number and grain yield distribution along the spike remain stable despite breeding for high yield in winter wheat.

Two winter wheat (Triticum aestivum L.) populations, i.e. 180 genetic resources and 210 elite varieties, were compared in a field trial to analyse how grain number and grain yield distribution along the spike changed during the breeding process and how this associates to yield-related traits. Elites showed in average 38% more yield compared to resources. This breeding improvement mainly derived from an increase in grains and yield per spike in addition to grains and yield per spikelet. These increments corresponded to 19, 23, 21 and 25%, respectively. Not much gain in thousand grain weight (4%) was observed in elites as compared to resources. The number of spikelets per spike was not, or even negatively, correlated with most traits, except of grains per spike, which suggests that this trait was not favoured during breeding. The grain number and grain yield distributions along the spike (GDAS and GYDAS) were measured and compared by using a novel mathematical tool. GDAS and GYDAS measure the deviation of a spike of interest from the architecture of a model spike with even grain and yield distribution along all spikelets, respectively. Both traits were positively correlated. Elites showed in average only a 1% improvement in GDAS and GYDAS values compared to resources. This comparison revealed that breeding increased grain number and yield uniformly along the spike without changing relative yield input of individual spikelets, thereby, maintaining the general spike architecture.

Leveraging the Use of Historical Data Gathered During Seed Regeneration of an ex Situ Genebank Collection of Wheat.

Genebanks are a rich source of genetic variation. Most of this variation is absent in breeding programs but may be useful for further crop plant improvement. However, the lack of phenotypic information forms a major obstacle for the educated choice of genebank accessions for research and breeding. A promising approach to fill this information gap is to exploit historical information gathered routinely during seed regeneration cycles. Still, this data is characterized by a high non-orthogonality hampering their analysis. By examining historical data records for flowering time, plant height, and thousand grain weight collected during 70 years of regeneration of 6,207 winter wheat (Triticum aestivum L.) accessions at the German Federal ex situ Genebank, we aimed to elaborate a strategy to analyze and validate non-orthogonal historical data in order to charge genebank information platforms with high quality ready-to-use phenotypic information. First, a three-step quality control assessment considering the plausibility of trait values and a standard as well as a weather parameter index based outlier detection was implemented, resulting in heritability estimates above 0.90 for all three traits. Then, the data was analyzed by estimating best linear unbiased estimations (BLUEs) applying a linear mixed-model approach. An in silico resampling study mimicking different missing data patterns revealed that accessions should be regenerated in a random fashion and not blocked by origin or acquisition date in order to minimize estimation biases in historical data sets. Validation data was obtained from multi-environmental orthogonal field trials considering a random subsample of 3,083 accessions. Correlations above 0.84 between BLUEs estimated for historical data and validation trials outperformed previous approaches and confirmed the robustness of our strategy as well as the high quality of the historical data. The results indicate that the IPK winter wheat collection reveals an extraordinary high phenotypic diversity compared to other collections. The quality checked ready-to-use phenotypic information resulting from this study is the first brick to extend traditional, conservation driven genebanks into bio-digital resource centers.

Unbalanced historical phenotypic data from seed regeneration of a barley ex situ collection.

The scarce knowledge on phenotypic characterization restricts the usage of genetic diversity of plant genetic resources in research and breeding. We describe original and ready-to-use processed data for approximately 60% of ~22,000 barley accessions hosted at the Federal ex situ Genebank for Agricultural and Horticultural Plant Species. The dataset gathers records for three traits with agronomic relevance: flowering time, plant height and thousand grain weight. This information was collected for seven decades for winter and spring barley during the seed regeneration routine. The curated data represent a source for research on genetics and genomics of adaptive and yield related traits in cereals due to the importance of barley as model organism. This data could be used to predict the performance of non-phenotyped individuals in other collections through genomic prediction. Moreover, the dataset empowers the utilization of phenotypic diversity of genetic resources for crop improvement.

Genome-Wide Prediction of the Performance of Three-Way Hybrids in Barley.

Predicting the grain yield performance of three-way hybrids is challenging. Three-way crosses are relevant for hybrid breeding in barley ( L.) and maize ( L.) adapted to East Africa. The main goal of our study was to implement and evaluate genome-wide prediction approaches of the performance of three-way hybrids using data of single-cross hybrids for a scenario in which parental lines of the three-way hybrids originate from three genetically distinct subpopulations. We extended the ridge regression best linear unbiased prediction (RRBLUP) and devised a genomic selection model allowing for subpopulation-specific marker effects (GSA-RRBLUP: general and subpopulation-specific additive RRBLUP). Using an empirical barley data set, we showed that applying GSA-RRBLUP tripled the prediction ability of three-way hybrids from 0.095 to 0.308 compared with RRBLUP, modeling one additive effect for all three subpopulations. The experimental findings were further substantiated with computer simulations. Our results emphasize the potential of GSA-RRBLUP to improve genome-wide hybrid prediction of three-way hybrids for scenarios of genetically diverse parental populations. Because of the advantages of the GSA-RRBLUP model in dealing with hybrids from different parental populations, it may also be a promising approach to boost the prediction ability for hybrid breeding programs based on genetically diverse heterotic groups.

Genomic Prediction of Barley Hybrid Performance.

Hybrid breeding in barley ( L.) offers great opportunities to accelerate the rate of genetic improvement and to boost yield stability. A crucial requirement consists of the efficient selection of superior hybrid combinations. We used comprehensive phenotypic and genomic data from a commercial breeding program with the goal of examining the potential to predict the hybrid performances. The phenotypic data were comprised of replicated grain yield trials for 385 two-way and 408 three-way hybrids evaluated in up to 47 environments. The parental lines were genotyped using a 3k single nucleotide polymorphism (SNP) array based on an Illumina Infinium assay. We implemented ridge regression best linear unbiased prediction modeling for additive and dominance effects and evaluated the prediction ability using five-fold cross validations. The prediction ability of hybrid performances based on general combining ability (GCA) effects was moderate, amounting to 0.56 and 0.48 for two- and three-way hybrids, respectively. The potential of GCA-based hybrid prediction requires that both parental components have been evaluated in a hybrid background. This is not necessary for genomic prediction for which we also observed moderate cross-validated prediction abilities of 0.51 and 0.58 for two- and three-way hybrids, respectively. This exemplifies the potential of genomic prediction in hybrid barley. Interestingly, prediction ability using the two-way hybrids as training population and the three-way hybrids as test population or vice versa was low, presumably, because of the different genetic makeup of the parental source populations. Consequently, further research is needed to optimize genomic prediction approaches combining different source populations in barley.