Harnessing landrace diversity empowers wheat breeding

The study addresses how to systematically harness the underutilized genetic diversity preserved in wheat landraces to improve modern cultivars amid mounting food security challenges. Wheat provides roughly 20% of global caloric intake, yet climate change, geopolitical instability, and a narrow export base threaten supply. Modern breeding has narrowed genetic diversity, slowing yield gains. Historically, locally adapted landraces tolerated environmental stresses but generally yielded less, and their diversity has seen limited use in contemporary breeding due to technical and resource constraints, including the lack of sequence-linked genomic resources and appropriate multi-environment phenotyping datasets. The authors propose leveraging the genetically and geographically diverse A. E. Watkins landrace collection through integrated whole-genome sequencing, haplotype mapping, and multi-year, multi-environment field phenotyping to discover, design, and deploy landrace alleles and haplotypes of value for modern wheat improvement.

Prior work established the Watkins landrace collection as a gene discovery resource and documented wheat landrace genome diversity and the annotated wheat reference genome (IWGSC RefSeq v1.0), enabling genome-wide variant discovery and mapping. The 10+ Wheat Genomes and subsequent whole-genome sequencing efforts revealed diversity patterns and breeding footprints in modern wheat. Breeding history includes the Green Revolution RHT1 semi-dwarfism genes and alternative semi-dwarfing loci such as RHT8. Tools like the Wheat Breeders’ Array and statistical frameworks such as ADMIXTURE, GWAS, and nested association mapping (NAM) have been applied in cereals, but comprehensive integration of whole-genome landrace sequencing with high-resolution haplotype maps and extensive multi-environment phenotyping remained limited. The authors identify gaps in accessible, sequence-resolved haplotypes, diagnostic markers, and validated pre-breeding materials that bridge landrace diversity with elite backgrounds for complex traits, which this work seeks to address.

• Germplasm and sequencing: Whole-genome resequencing (mean 12.73×) of 827 A. E. Watkins bread wheat landrace accessions and 208 modern cultivars (plus 15 previously described) aligned to IWGSC RefSeq v1.0. Variant discovery yielded ~262 million high-quality SNPs, short indels, and CNVs. Population structure was inferred via t-SNE and ADMIXTURE, defining seven Watkins ancestral groups (AG1–AG7).
• Comparative diversity and haplotypes: Identity-by-state long-range haplotypes in modern cultivars were visualized using k-mer-based IBSpy. Linkage disequilibrium (LD)-based haplotype analysis with PLINK defined 71,282 haploblocks, cataloguing Watkins-unique versus shared haplotypes. Functional impact was annotated with SnpEff, emphasizing Watkins-unique SNPs in genes monomorphic in modern wheat.
• Phenotyping and mapping populations: Extensive field phenotyping was performed over 10 years and 10 environments in the UK and China, aggregating >717,000 observations for 137 traits spanning yield, quality, adaptation, and stress tolerance. The Watkins panel supported GWAS; 73 Paragon × Watkins recombinant inbred line (RIL) populations (6,762 RILs) supported bi-parental QTL mapping and NAM-GWAS via imputation, combining historical and experimental recombination to detect both common and rare alleles.
• Genetic analyses: GWAS, QTL mapping, and NAM-GWAS identified marker-trait associations (MTAs) and QTL at haplotype resolution. Useful alleles were defined by trait-improving directions under breeding selection (including both directions for traits under stabilizing/disruptive selection). Bi-parental mapping was emphasized for rare haplotypes (e.g., yellow rust ‘Warrior’ race resistance), complementing GWAS/NAM.
• Near isogenic lines (NILs) and introgression: Two backcrosses of Paragon to selected Paragon × Watkins RILs produced 738 NILs (isogenic families) carrying 127 prioritized QTL alleles (mainly from AG1, AG3, AG4, AG6, AG7). These introgressions encompassed 44,338 Watkins-unique LD haplotypes. Multi-location/year replicated field trials in large plots (6 m²) quantified allelic effects for 11 core traits; environmental stability was assessed using AMMI.
• Case studies and fine mapping: A height/biomass QTL on 7BL was validated for yield benefits without adverse HI effects. The semi-dwarfing locus RHT8 (Mara-like) was fine-mapped from a 0.82-Mb region down to 6.7 kb containing two genes (TraesCS2D02G057800 and TraesCS2D02G057900) with tightly anti-correlated expression during tillering; novel haplotype-specific markers were developed.
• Marker resources: Diagnostic tag SNPs and corresponding molecular markers were identified for ~1.7 million haplotypes. KASP assays for two example QTL (5A B1 awn inhibitor and 7A Rc coleoptile colour) were validated on 382 accessions, achieving 87.4% and 93.7% correct phenotype predictions, respectively.
• Data and resource sharing: All germplasm, genomic and phenotypic data, haplotype maps, and markers are made available via the Watkins Worldwide Wheat Genomics to Breeding portal (https://wwwg2b.com/).

• Origins and diversity: Watkins landraces resolve into seven ancestral groups (AG1–AG7). Modern cultivars largely derive from two groups (AG2/AG5 of Western/Central European origin), supported by independent datasets. Five groups (AG1, AG3, AG4, AG6, AG7) are phylogenetically isolated from modern wheat, representing untapped diversity.
• Variant discovery: ~262 million high-quality SNPs across Watkins + modern; variants unique to Watkins include 162 million SNPs (62%), 9.7 million indels (57%), and ~57,000 CNVs (53%), predominantly in AG1/3/4/6/7. Among Watkins-unique SNPs, 325,915 affect gene function; 13,902 genes monomorphic in modern wheat carry Watkins-unique SNPs, implicating traits such as yield, stress tolerance, nutrition, and disease resistance.
• Long-range haplotype structure: Modern wheat shows extensive IBS with AG2/5-derived landraces, with mean intact centromeric IBS tracts ~159.78 Mb and shorter distal segments. As few as 26 Watkins accessions can model donors of IBS segments to reconstitute >50% of modern genomes.
• LD haplotypes: 71,282 haploblocks identified; 69.6% (49,626) contain only Watkins-unique haplotypes (median 5, mean 11.85 haplotypes per block). A small proportion (~2.5%) of unique variants in Watkins were also present in modern germplasm due to targeted introgressions from wild relatives (e.g., 1BL/1RS, RHT1, Pch1).
• Genetic effects: In total, 8,253 genetic effects were detected (3,280 QTL; 1,428 GWAS MTAs; 3,545 NAM-GWAS MTAs). Based on allelic effect direction, 1,696 have potential to improve modern cultivars like Paragon. Approximately 36% of top SNP associations are genic.
• Disease resistance: Bi-parental mapping identified 15 new loci for yellow rust (Puccinia striiformis) resistance in the UK and Australia, including to the ‘Warrior’ race. Twelve loci lie outside AG2/5, with five from Iranian landraces; GWAS alone did not capture these rare alleles.
• Trait trade-offs and NIL validation: Near isogenic line testing showed that several positive-effect QTL for one trait are neutral or positive for correlated traits (e.g., grain weight vs grain number; yield vs protein), enabling decoupling of antagonistic relationships.
• 7BL height/biomass QTL: A Watkins haplotype on 7BL increased plant height by 9.37 cm (P=0.002) and grain yield by 0.39 t ha⁻¹ (P<0.002) without reducing harvest index, across multi-location trials with breeding partners.
• RHT8 fine mapping and markers: RHT8 was narrowed to a 6.7-kb interval containing TraesCS2D02G057800 and TraesCS2D02G057900; expression is tightly anti-correlated (Pearson r = -0.979, P<0.0001). Haplotype-specific markers were developed, providing breeders with tools for height control with minimal yield penalty, especially in Mediterranean environments.
• Deployment via NILs: 127 prioritized QTL alleles were introgressed into Paragon, represented by 738 NILs, including 44,338 Watkins-unique haplotypes. AMMI-based evaluation across ≥3 locations/years (totaling 9 years of testing for the 127 loci) showed substantial, environmentally robust effects: heading date from -6 to +2 days; height from -5 to +13 cm; yield increases up to 0.91 t ha⁻¹ (significant cases). Arithmetic sums of significant component effects suggest potential gains of +4.5 t ha⁻¹ grain yield, +11,500 grains m⁻², and +55.6 mg thousand grain weight (not accounting for epistasis).
• Marker platform: Tag SNPs and markers for ~1.7 million haplotypes enable haplotype-led selection. KASP markers for B1 (awn inhibitor, 5A) and Rc (coleoptile colour, 7A) predicted phenotypes with 87.4% and 93.7% accuracy, respectively.

The work demonstrates a complete framework to discover, validate, and deploy landrace diversity into modern wheat breeding. By showing that modern wheat derives predominantly from AG2/5, the study highlights the large, untapped reservoir of useful alleles in the five isolated ancestral groups. Integrating whole-genome haplotype maps with NAM-GWAS, bi-parental mapping for rare alleles, and NIL-based field validation addresses the traditional bottleneck of translating landrace variation into elite backgrounds for complex traits. Case studies show that antagonistic trait relationships (e.g., biomass/height vs yield; yield vs grain protein) can be uncoupled at specific loci, enabling selection of favorable alleles without historical trade-offs. Fine mapping and marker development for RHT8 provide practical, diagnostic tools to manage plant height and adaptation with minimal pleiotropic penalties. The extensive NIL resource with quantified, environmentally stable effects and the publicly accessible haplotype-tag markers establish immediate pathways for breeders to stack beneficial Watkins haplotypes into modern pedigrees across environments.

This study delivers a genomics-to-breeding pipeline that unlocks the value of wheat landrace diversity. Key contributions include: (1) whole-genome sequencing and population structure of 827 Watkins landraces revealing five ancestral groups largely absent from modern wheat; (2) a high-resolution LD- and IBS-based haplotype map linking genotypes to thousands of QTL/MTAs; (3) validation and deployment of 127 prioritized landrace QTL alleles via 738 NILs, encompassing 44,338 Watkins-unique haplotypes, with robust field-measured benefits for yield, quality, adaptation, and disease resistance; and (4) fine-mapped loci and diagnostic markers (including RHT8) and a comprehensive set of tag SNPs for ~1.7 million haplotypes to drive haplotype-led breeding. Future directions include incorporating long-read sequencing to capture large structural and copy-number variation, advancing breeding technologies to overcome linkage drag and facilitate allele stacking, and validating combined allele effects in diverse elite backgrounds and target production environments. Open-access sharing of resources via the Watkins portal is intended to catalyze global breeding efforts and foster collaborative innovation.

• Sequencing modality: Short-read resequencing limited the detection and resolution of large structural variants and complex copy-number variation; long-read, chromosome-scale assemblies would enhance variant discovery and functional interpretation.
• Breeding deployment: Combining multiple novel Watkins alleles in a single elite variety faces linkage drag and recombination constraints; new breeding technologies and strategies are needed to efficiently stack alleles while preserving existing favorable haplotypes.
• Validation background: Most allelic effects were quantified as NILs in the UK spring wheat Paragon background, which has lower yield potential than modern UK winter wheats; transferability and additive gains require validation across diverse elite backgrounds and environments, and epistatic interactions remain to be comprehensively assessed.
• Policy and access: Restrictions on international germplasm exchange can impede broad deployment; sustained open data and germplasm sharing are needed to realize global impact.