Genomic selection (GS) can accelerate plant breeding gains by reducing breeding cycle times, reducing phenotyping costs, or improving selection accuracy. GS is especially promising for perennial crops such as intermediate wheatgrass (IWG, Thinopyrum intermedium) that may require multiple years of evaluation under phenotypic recurrent selection. A major obstacle in implementing GS is the need for an affordable, high-density, genetic marker system that is scalable to thousands or tens-of-thousands of samples in breeding programs, especially in emerging or minor crop species. As sequencing costs continue to decrease, low-coverage whole genome skim-sequencing (skim-seq) has become an attractive method for GS. Using commercial laboratory products and open-source software, we implemented whole genome prediction at breeding program scale using ultra-low coverage (0.01x- 0.05x, 100-125 million reads per sample) whole genome skim-seq. Using STITCH (Sequencing to Imputation Through Constructing Haplotypes) imputation software, we evaluated optimization of imputation parameters including sequence coverage and number of assumed ancestral haplotypes. Finally, we evaluated whole genome prediction cross-validation accuracies using reduced representation genotyping-by-sequencing (GBS) versus skim-seq data for IWG, an outcrossing, heterozygous, large genome (12.7 Gb), polyploid perennial species. Our results indicate correlations between cross-validation accuracies across five traits in IWG using skim-seq data (r=0.29-0.61) can be used as effectively as GBS (r=0.29-0.55) while generating low-coverage archival sequence data that will be robust to technological advances. These methods will be applicable to a wide range of crops and scale to breeding program size, allowing for more tractable implementation of GS within breeding programs.

Faba bean (Vicia faba L.) is one of the oldest cultivated crops in the world. It is the third most important feed grain legume globally, after soybean and lupin. It is an autogamous plant with a partial outcrossing rate ranging from 20% to 80%. The objective of most faba bean improvement programs is to enhance yield; however, yield is a complex trait influenced by many other traits. Therefore, in this study, we focused on seed traits that are related to faba bean yield. A set of 110 faba bean genotypes was tested across two different locations (Germany and Egypt) to investigate the effect of genotype-environment interactions and identify single-nucleotide polymorphism (SNPs) related to seed characteristics. This study revealed that there is high genetic variation among genotypes in all traits in each location, and the genotype × location interactions were significant. There was a strong positive correlation among the seed characteristics within each location, but the correlations between the two locations were weak or not significant. FB-231 and FB-227 performed very well in both countries based on the selection index (SI) values, whereas FB-193 and FB-185 had the lowest SI values in both countries. All genotypes were genotyped via single primer enrichment technology, which resulted in 33,165 SNP markers. The association mapping revealed 162 and 31 significant SNPs in seed traits scored in Germany and Egypt, respectively. A set of seven SNPs was associated with more than one seed trait in Germany, whereas only one SNP was associated with two traits in Egypt. No shared markers were found between the two locations for any of the seed traits. These markers represent potential targets for future breeding programs to enhance seed size and understanding of its genetic control in the faba bean.

We conducted simulations of common bean (Phaseolus vulgaris L.) breeding programs to better understand the interplay between different choices a breeder must make when launching a genomic selection (GS) pipeline. We complement preceding studies on optimizing model parameters and training set makeup by exploring the practical implementation of GS in a common bean breeding program aimed at increasing seed yield. We simulated 24 GS implementation pathways on (1) what generation to train a new prediction model, (2) what generation to select parents for the next cycle, (3) which generation to collect training data, and (4) whether to use a parametric (ridge regression best linear unbiased predictor) or a nonparametric model (artificial neural network) for estimating breeding values. We found that early generation parent selections (also called rapid-cycle GS) generally resulted in higher gain over three breeding cycles compared to late-generation parent selections. When implementing a new parametric genomic prediction model, training data should be as diverse as possible, while also matching testing data in terms of genetic makeup and allele frequency. Parametric models showed more consistent genomic estimated breeding value prediction accuracy, while nonparametric models fluctuated, showing both the highest and the lowest prediction accuracy across all pathways. Despite the trade-off between gains and genetic variance, nonparametric models showed greater balance of allelic diversity and gains. We observed that the key to sustained gains over time is the renewal of genetic variance. Our results indicate a potential for the use of nonparametric models, but more investigation will be required to stabilize their performance.

We present the first reference genome of the highly heterozygous autotetraploid Allium porrum (leek). Combining long-read sequencing with single-nucleotide polymorphism (SNP)-array screening of two experimental F1 populations, we generated a genetic map with 11,429 SNP markers across eight linkage groups and a chromosome-scale assembly of A. porrum (leek) totaling 15.2 Gbp in size. The high quality of the reference genome is substantiated by 97.2% BUSCO completeness and a mapping rate of 96% for full-length transcripts. The linkage map exposes the recombination landscape of leek and confirms that crossovers are predominantly proximal, located to the centromeres, contrasting with distal recombination landscapes observed in other Allium species. Comparative genomics reveals structural rearrangements between A. porrum and its relatives (Allium fistulosum, Allium sativum, and Allium cepa), suggesting a closer genomic relationship to A. sativum. Our annotated high-quality reference genome delivers crucial insights into the leek genome structure, recombination landscape, and evolutionary relationships within the Allium genus, with implications for species compatibility in breeding programs, facilitating marker-assisted selection and genetic improvement in leek.

Root rot caused by Aphanomyces euteiches is a major concern in pea (Pisum sativum L.). The lack of other effective control strategies makes crucial the development of resistant varieties. Although partial resistance has been reported, its quantitative inheritance, the association of resistance‐linked genomic regions with unfavorable agronomic traits, and the limited understanding of soil pathogen populations hinder its progress in breeding programs. To search for alternative genomic regions associated with this partial resistance, a genome‐wide association study (GWAS) was performed on a pea collection not yet explored for A. euteiches resistance in genetic studies. The 323 accessions of the collection were inoculated with RB84 isolate, and foliar and root symptoms were assessed 20 days after inoculation. The performed GWAS revealed 27 significantly associated markers among 26,045 SilicoDArT and 7033 single‐nucleotide polymorphism marker datasets. Detected markers were distributed along the seven pea chromosomes, with 12 within previously described quantitative trait loci (QTLs). Chromosomes 2 and 5 harbored a significant number of associated markers, identified here for the first time, highlighting promising regions for future investigation. Twenty‐one candidate resistance genes were identified. This study uncovers new genomic regions linked with A. euteiches resistance and provides molecular markers and candidate genes to support precision breeding. Newly identified QTL may be more effective against specific isolates than known QTL, enabling improved QTL rotation in the field.

An ensemble of multiple genomic prediction models has grown in popularity due to consistent prediction performance improvements in crop breeding. However, technical tools that analyze the predictive behavior at the genome level are lacking. Here, we develop a computational tool called Ensemble AnalySis with Interpretable Genomic Prediction (EasiGP) that uses circos plots to visualize how different genomic prediction models quantify contributions of marker effects to trait phenotypes. As a demonstration of EasiGP, multiple genomic prediction models, spanning conventional statistical and machine learning algorithms, were used to infer the genetic architecture of days to anthesis (DTA) in a maize mapping population. The results indicate that genomic prediction models can capture different views of trait genetic architecture, even when their overall profiles of prediction accuracy are similar. Combinations of diverse views of the genetic architecture for the DTA trait in the teosinte nested association mapping study might explain the improved prediction performance achieved by ensembles, aligned with the implication of the Diversity Prediction Theorem. In addition to identifying well-known genomic regions contributing to the genetic architecture of DTA in maize, the ensemble of genomic prediction models highlighted several new genomic regions that have not been previously reported for DTA. Finally, different views of trait genetic architecture were observed across subpopulations, highlighting challenges for between-population genomic prediction. A deeper understanding of genomic prediction models with enhanced interpretability using EasiGP can reveal several critical findings at the genome level from the inferred genetic architecture, providing insights into the improvement of genomic prediction for crop breeding programs.

Genomic selection (GS) has revolutionized breeding practices by integrating genotype and phenotype data to predict genomic estimated breeding values, offering the potential to accelerate breeding cycles and intensify and enhance early‐stage selections. This approach utilizes the concept of linkage disequilibrium (LD) between genetic markers and quantitative trait loci within populations. LD, the nonrandom association between alleles at different loci, provides valuable insights into historical recombination patterns, although it can change over time under strong selection or genetic drift. This study aimed to investigate the influence of recombination on haplotype sizes and LD, assess the impact of additive (A) versus additive + epistasis (A+I) genetic models on GS predictive ability (PA), and demonstrate how haplotype resolution in the training set (TS) impacts the PA of GS. For this, we used biparental (MP2) and multiparent (MP6–8) populations, where the main difference between them was the recombination rate. As expected, a strong correlation between LD decay and the number of recombination opportunities within populations was observed, with smaller haplotype blocks in populations experiencing more recombination. The use of A+I models increased heritability but did not improve PA. Finally, populations with smaller haplotype sizes in the TS exhibited enhanced PA. This study demonstrates the effect of haplotype size on GS accuracy, and its uniqueness lies in its focus on populations where the primary differentiating factor is haplotype size. It offers an important tool for breeders in designing GS strategies, providing valuable guidance for future breeding efforts.