# Source paper archive - Source title: Dissecting the genetic basis of DUS characters in Pakistani wheat varieties - Author(s): Haq, I. U., Qureshi, I. A., Hussain, M. A., Javaid, R. A., Sheikh, F., Abbas, Z., Azam, M. - Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.01.763533v1 - PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.01.763533v1.full.pdf - PDF SHA-256: 82ba02e6e937f4259e651ee9f73f36c5826cbd7bc4b11b00165abd6e18fa8d9b - Converter: pymupdf4llm 0.0.27 - Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version. --- [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) # 1 Dissecting the genetic basis of DUS characters in Pakistani wheat 2 varieties ## 3 Ihtisham Ul Haq¹ [⁎], Ijaz Ahmed Qureshi¹, Muhammad Ans Hussain [2], Rana Arsalan 4 **Javaid¹, Fawad Sheikh¹, Zeeshan Abbas** **[3]** **, Muhammad Azam¹** 5 1-Plant Breeders’ Rights Registry, Ministry of National Food Security and Research, Islamabad, 6 Pakistan 7 2-National Key Laboratory of Crop Genetic Improvement, Huazhong Agricultural University, 8 Wuhan 430070, China 9 3-National Institute for Genomics & Advanced Biology, National Agricultural Research Center, 10 Islamabad, Pakistan ## 11 ⁎ Corresponding author: ihtisham.aheer@pbrr.gov.pk 12 Abstract 13 Distinctness, Uniformity and Stability (DUS) testing is fundamental to varietal 14 characterization and the protection of Plant Breeders’ Rights (PBRs), but conventional DUS 15 assessment relies predominantly on morphological characteristics whose expression may be 16 influenced by environmental conditions, developmental stage and observer-related variation. 17 Integrating character-specific molecular markers with conventional DUS assessment may 18 therefore improve the reproducibility and genetic resolution of varietal identification. This study 19 investigated the genetic architecture of DUS characteristics in a panel of 102 historical and 20 modern Pakistani wheat varieties using genome-wide association studies (GWAS). Phenotypic 21 variation was evaluated using distribution, correlation and hierarchical clustering analyses, while 22 the germplasm was genotyped using a wheat genotyping-by-targeted sequencing platform. After 23 quality filtering, 21,670 high-quality single-nucleotide polymorphisms (SNPs) were retained for 24 analyses of genetic diversity, population structure, linkage disequilibrium (LD) and marker-trait 25 associations. Population structure analysis indicated extensive admixture, with cross-entropy 26 supporting K = 9 ancestral components, whereas hierarchical genetic clustering separated the 27 germplasm into five broad groups. The first three principal components explained 8.09%, 7.20% 28 and 5.39% of the total genetic variation, respectively. GWAS using BLINK and FarmCPU 29 identified significant associations for several diagnostically important DUS characteristics. 30 Growth habit was strongly associated with SNP 1B_129176704, with an additional signal on 1 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 31 chromosome 5B. Shattering showed its strongest association at 1D_445256324, accompanied by 32 several closely linked SNPs in the same region and a secondary significant locus at 33 7B_703502819. Flag leaf length was associated with a cluster of SNPs on chromosome 6A, 34 while tiller density showed a major association with 5D_205488820. These findings demonstrate 35 substantial phenotypic and genetic diversity within Pakistani wheat germplasm and identify 36 genomic regions that merit validation as character-specific molecular tools to complement 37 conventional DUS testing and PBR implementation. 38 **Keywords:** Wheat, DUS, Plant Breeders’ Rights, GWAS ## 39 INTRODUCTION 40 Food insecurity remains a major global challenge and is increasingly intensified by climate 41 related stresses. United Nations Sustainable Development Goal 2 aims to end hunger, achieve 42 food security, improve nutrition and promote sustainable agriculture by 2030 (UN, 2015). The 43 development and deployment of climate-resilient crop varieties can contribute to this objective; 44 however, plant breeding requires substantial long-term investment. Effective systems for 45 protecting new plant varieties are therefore important for encouraging continued innovation and 46 supporting returns on breeding investment. The International Union for the Protection of New 47 Varieties of Plants (UPOV) Convention provides an international framework for the protection 48 of new plant varieties (UPOV, 1991). In Pakistan, the Plant Breeders’ Rights Act, 2016 (GoP, 49 2016) and the Plant Breeders’ Rights Rules, 2018 (GoP, 2018) provide the national legal 50 framework for Plant Breeders’ Rights (PBRs), broadly following the principles of the UPOV 51 Convention 1991. 52 Distinctness, Uniformity and Stability (DUS) testing is central to the grant of PBRs and is 53 conventionally based on the expression of qualitative, pseudo-qualitative and quantitative 54 morphological characteristics across prescribed testing conditions and seasons. Because 55 phenotypic expression may be influenced by environmental conditions, developmental stage and 56 scoring subjectivity, character-specific molecular information could provide an additional layer 57 of evidence for varietal characterization. The UPOV Technical Working Group on Biochemical 58 and Molecular Techniques and DNA Profiling has considered approaches in which molecular 59 markers linked to specific characteristics can support DUS testing (UPOV, 2020). More recently, 60 genome-wide association analysis has been used to identify loci controlling DUS characteristics 2 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 61 in wheat (Zanella _et al._, 2026), and genomic prediction has been proposed as a complementary 62 strategy for managing DUS reference collections and supporting trait-specific distinctness 63 decisions (Roberts _et al._, 2026). 64 Wheat is one of the world’s most important food crops and contributes substantially to global 65 calorie and protein intake (Erenstein _et al._, 2022). Its broad adaptation, storability and nutritional 66 importance make it central to global food security (Iqbal _et al._, 2022). Wheat is also of particular 67 importance to Pakistan, where it occupies a major proportion of the cultivated area and 68 contributes substantially to national food supply (GoP, 2025). Genome-wide association studies 69 conducted in Pakistani and related wheat germplasm have identified significant loci for several 70 agronomic and adaptive traits. Under heat stress, significant marker-trait associations have been 71 reported for kernel-weight reduction (Hussain _et al._, 2026). Allelic and haplotypic variation at 72 _TaHST1_ has been associated with heat tolerance (Khan _et al._, 2022), while loci and haplotypes 73 related to reduced plant height have been investigated in relation to _Rht_ genes (Suleman _et al._, 74 2025). Additional studies have reported candidate genes related to glucose metabolism, ion 75 transport, hormone responses and starch synthesis (Zhou and Irshad, 2026), as well as 76 quantitative trait nucleotides associated with canopy architecture, flag leaf area and flag leaf 77 width (Farhan _et al._, 2026). 78 The present study evaluated DUS characteristics in 102 Pakistani wheat varieties representing 79 historical and modern germplasm. The primary objective was to identify SNP associations for 80 DUS characteristics using high-throughput GWAS models and to assess whether robust marker 81 trait associations could provide a basis for future character-specific molecular tools. Such tools 82 are intended to complement, rather than replace, conventional DUS phenotyping and may 83 ultimately strengthen varietal identification and PBR enforcement in Pakistan. ## 84 MATERIALS AND METHODS 85 _**Plant material and DUS phenotyping**_ 86 Identity records and DUS phenotypic data for 102 Pakistani wheat varieties, including 87 historical lines and modern cultivars, were obtained from the DUS Office, Federal Seed 88 Certification and Registration Department (FSC&RD), Ministry of National Food Security and 89 Research, Islamabad, Pakistan. Qualitative and pseudo-qualitative characteristics were converted 90 to numerical DUS notes for analysis, as described in Supplementary Table S1. 3 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 91 Phenotypic data were analysed in R using the packages dplyr, tidyr, ggplot2, dendextend and 92 corrplot. Trait distributions were visualized using violin plots, while hierarchical clustering was 93 used to evaluate phenotypic similarity among varieties. Correlation analysis was performed to 94 examine the direction and magnitude of relationships among characteristics. Where traits were 95 measured on different scales, standardized _z_ -scores were used to facilitate direct comparison. 96 _**Genotyping and genetic analyses**_ 97 The wheat panel was genotyped using a 16K wheat genotyping-by-targeted sequencing (GBTS) 98 platform (Rasheed and Fayyaz, 2023). The initial dataset contained approximately 37,000 SNPs. 99 Quality filtering was performed using BCFtools, and markers with a minor allele frequency 100 (MAF) < 0.05 were removed. After filtering, 21,670 high-quality SNPs were retained for 101 downstream analyses. 102 Linkage disequilibrium (LD) was evaluated as pairwise marker correlation ( _r_ ²) using PLINK 2.0. 103 LD decay was assessed by plotting _r_ ² against physical distance in the Chinese Spring wheat 104 reference genome (Consortium _et al._, 2018) for the A, B and D subgenomes. An _r_ ² threshold of 105 0.20 was used as a reference for LD decay. Principal component analysis (PCA) was performed 106 on the filtered SNP dataset, and the first three principal components were visualized in three 107 dimensional space. 108 Population structure was evaluated using a model-based ancestry framework implemented in the 109 LEA package in R across _K_ = 2-10 ancestral components. Model fit was compared using cross 110 entropy with the optimal value of _K_ selected from the minimum cross-entropy. A hierarchical 111 dendrogram based on genome-wide marker distances was also generated. Together, PCA, 112 ancestry analysis and hierarchical clustering were used to assess relatedness, genetic 113 differentiation and the potential influence of population structure on association analysis. 114 _**Genome-wide association studies and post-GWAS analysis**_ 115 Genome-wide association analyses were conducted using Bayesian-information and Linkage 116 disequilibrium Iteratively Nested Keyway (BLINK) and Fixed and random model Circulating 117 Probability Unification (FarmCPU) implemented in GAPIT in R. For each characteristic, raw _P_ 118 values were adjusted using the Benjamini-Hochberg false discovery rate (BH-FDR) at α = 0.05 119 and the Bonferroni family-wise error threshold at α = 0.05. With 21,670 markers, the nominal 120 Bonferroni threshold was approximately _P_ = 2.31 × 10⁻⁶. Manhattan and quantile-quantile (QQ) 4 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 121 plots were examined to evaluate the distribution of association signals and model calibration. 122 Genomic inflation factors were evaluated from the QQ-plot outputs. 123 Significant SNPs identified by GWAS were positioned relative to the IWGSC Chinese Spring 124 reference genome (Consortium _et al._, 2018). Candidate genes were considered within LD 125 defined intervals surrounding significant SNPs. Candidate regions were interpreted with 126 reference to gene position and published functional information. To evaluate the phenotypic 127 effects of significant marker-trait associations, post-GWAS genotype-class comparisons were 128 performed using appropriate parametric or non-parametric statistical procedures following the 129 general approach described by (Nawade _et al._, 2025). Kruskal-Wallis tests were used for the 130 genotype-class comparisons reported for the principal loci. ## 131 RESULTS 132 _**Phenotypic variation and trait relationships**_ 133 The evaluated wheat varieties showed substantial phenotypic variation across the DUS 134 characteristics. Standardization to _z_ -scores enabled direct comparison among traits measured on 135 different scales. The violin plots revealed marked differences among characteristics in 136 distribution width, central tendency and skewness, indicating substantial trait-specific variation 137 within the germplasm. Qualitative and pseudo-qualitative characteristics generally showed more 138 discrete or concentrated distributions, as expected from their note-based scoring system. 139 Hierarchical clustering further demonstrated phenotypic differentiation among the varieties. 140 Based on their combined standardized trait profiles, the genotypes were separated into five major 141 phenotypic clusters (Figure 1). Two clusters contained a relatively large proportion of the 142 germplasm, whereas the remaining groups contained smaller and more distinct sets of varieties. 143 The separation of these groups at comparatively large linkage distances indicates meaningful 144 differences in overall phenotypic profiles. 145 The correlation matrix showed both positive and negative relationships among traits, although 146 most pairwise correlations were relatively weak (Figure 2). This pattern indicates that many DUS 147 characteristics capture largely independent dimensions of varietal morphology. Collectively, the 148 distribution, clustering and correlation analyses demonstrate broad phenotypic diversity within 149 the panel and provide an appropriate phenotypic basis for subsequent association analysis. 5 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 150 _**Genotyping and genetic analyses**_ 151 Model-based population structure was evaluated across K = 2-10 ancestral components. At all 152 values of K, most accessions contained contributions from more than one ancestry component, 153 indicating extensive admixture among the evaluated wheat varieties. Cross-entropy decreased 154 from approximately 0.59 at K = 2 to a minimum of approximately 0.518 at K = 9, followed by a 155 slight increase to approximately 0.524 at K = 10 (Figure 3). Accordingly, K = 9 was selected as 156 the best-supported model for population structure. The progressive reduction in cross-entropy up 157 to K = 9 indicates that the panel contains substantial substructure that is not adequately 158 represented by only a few ancestral groups. At K = 9, several sets of accessions showed 159 relatively high contributions from specific ancestry components, while extensive admixture 160 remained evident, consistent with historical gene flow and shared breeding backgrounds. 161 The circular hierarchical dendrogram provided complementary evidence of genetic 162 differentiation. Based on genome-wide marker distances, the accessions were broadly separated 163 into five major genetic clusters. One cluster contained a substantial proportion of the accessions 164 and showed considerable internal branching, whereas the smaller clusters comprised more 165 closely related sets of genotypes (Figure 4). The five broad dendrogram groups and nine ancestry 166 components represent different levels of population structure rather than conflicting results. 167 Hierarchical clustering summarizes broad genome-wide relatedness, whereas the model-based 168 ancestry analysis resolves finer substructure and admixture within the major groups. Together, 169 these analyses indicate a hierarchical genetic organization within the Pakistani wheat panel. 170 The first three principal components explained 8.09%, 7.20% and 5.39% of the total genetic 171 variation for PC1, PC2 and PC3, respectively. Most accessions occupied a relatively compact 172 region of the three-dimensional PCA space, consistent with shared ancestry and genetic 173 exchange among breeding materials. Nevertheless, differentiation was evident along all three 174 axes, and at least one accession was clearly separated from the main group. The moderate 175 proportion of variance explained by each component suggests that population differentiation is 176 distributed across several dimensions rather than being dominated by a single major axis. 177 Genome-wide SNP-density analysis showed marker coverage across all 21 wheat chromosomes 178 of the A, B and D subgenomes. SNP abundance, assessed in 1-Mb windows, varied among 179 chromosomes and genomic regions. Some intervals contained relatively few SNPs, whereas 6 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 180 others contained more than 50 markers per 1-Mb window. Despite this heterogeneity, the filtered 181 marker set provided broad genome-wide coverage for population and association analyses. LD, 182 expressed as pairwise _r_ ², declined with increasing physical distance in all three subgenomes, 183 although the decay trajectories differed. The D genome showed comparatively high LD at very 184 short distances followed by a relatively rapid decline, whereas the A and B genomes showed 185 more gradual decay and retained higher LD over longer distances. These subgenome-specific 186 patterns indicate differences in the extent of allelic association across the wheat genome. 187 Overall, the genetic analyses demonstrated substantial diversity, pronounced admixture and 188 hierarchical population structure in the study panel. The combined ancestry, clustering and PCA 189 results support the use of models that account for population structure when interpreting marker 190 trait associations. 191 _**Genome-wide association studies**_ 192 GWAS for growth habit (TGH) identified two principal loci, with the strongest and most 193 consistent signal on chromosome 1B and a second signal on chromosome 5B. In both BLINK 194 and FarmCPU, the associated regions showed clear departures from the genomic background. 195 The QQ plots showed that deviation from the expected null distribution was concentrated 196 primarily in the upper tail, consistent with a limited number of true association signals rather 197 than widespread inflation. The lead SNP on chromosome 1B, 1B_129176704, was highly 198 significant (P = 3.756 × 10⁻⁹; BH-FDR = 8.140 × 10⁻⁵; Bonferroni-adjusted P = 8.140 × 10⁻⁵). 199 The locus had a positive allelic effect of 1.610 and a low MAF of 0.053, indicating that the effect 200 allele was relatively uncommon in the study panel (Figure 5). A second marker, 5B_586820792, 201 was also associated with growth habit (P = 2.333 × 10⁻⁶; BH-FDR = 2.528 × 10⁻²). Its 202 Bonferroni-adjusted P value of 0.05056 placed the association at the boundary of the strict 203 family-wise significance criterion. Genotype-class comparisons supported phenotypic 204 differentiation at both loci. For 1B_129176704, the Kruskal-Wallis test was significant (P = 205 0.00878), and the A/A class showed the highest growth-habit scores, the G/G class showed 206 intermediate values and the G/A class showed the lowest values. For 5B_586820792, genotype 207 classes also differed significantly (Kruskal-Wallis P = 0.00376), with the C/C genotype showing 208 higher scores than the T/T and T/C classes (Figure 6). These results identify chromosome 1B as 7 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 209 the major growth-habit region in this panel, with additional evidence for a secondary locus on 210 chromosome 5B. 211 Shattering (SH) showed a highly significant major association on chromosome 1D and a 212 secondary significant locus on chromosome 7B (Table 1). The chromosome 1D peak exceeded 213 both BH-FDR and Bonferroni thresholds, and the QQ plots showed that most markers closely 214 followed the null expectation except for the strongest associations. The most significant SNP was 215 1D_445256324 (P = 5.207 × 10⁻¹¹; BH-FDR = 1.128 × 10⁻⁶; Bonferroni-adjusted P = 1.128 × 216 10⁻⁶). Several neighboring SNPs, including 1D_445256334, 1D_445256342, 1D_445256363 and 217 1D_445256389, also showed very small P values, suggesting that these markers represent a 218 shared LD block or a major QTL. A second significant association was identified at 219 7B_703502819 (P = 1.414 × 10⁻⁷; BH-FDR = 5.106 × 10⁻⁴; Bonferroni-adjusted P = 3.064 × 220 10⁻³). Genotype-class comparisons were consistent with the GWAS signals. At the representative 221 1D marker 1D_445256389, the Kruskal-Wallis test was significant (P = 0.00361). The C/C 222 reference class showed shattering scores near 3.0, whereas the C/T class clustered near 1.0. A 223 homozygous alternate class was not represented in the displayed data. At 7B_703502819, 224 genotype classes also differed significantly (Kruskal-Wallis P = 0.00561); the C/C class showed 225 the highest shattering scores, whereas the G/G and G/C classes remained lower. Together, the 226 results indicate a major shattering-associated region on chromosome 1D with an additional 227 contribution from chromosome 7B. 228 Flag leaf length (FLL) was associated primarily with a cluster of SNPs on chromosome 6A. Both 229 BLINK and FarmCPU detected a pronounced 6A signal, and the QQ plots showed acceptable 230 calibration, with departure from the expected distribution concentrated among the most 231 significant markers. Several SNPs between approximately 77.9 and 78.2 Mb, including 232 6A_77909780, 6A_77913131, 6A_77913187, 6A_77960267, 6A_77999760, 6A_77999868, 233 6A_77999899, 6A_78065495, 6A_78065597, 6A_78152531, 6A_78152577, 6A_78230010 and 234 6A_78230072, shared essentially identical association statistics (P = 6.037 × 10⁻⁶; BH-FDR = 235 6.541 × 10⁻³) and a positive allelic effect of 4.057. The similarity among these statistics is 236 consistent with strong LD among markers tagging the same underlying region. Additional SNPs 237 at approximately 79.36, 81.24, 86.49, 92.35 and 96.51 Mb showed stronger evidence of 238 association, and some passed Bonferroni correction. For the representative marker 239 6A_77909780, the Kruskal-Wallis test was significant (P = 0.0452). The G/G class showed a 8 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 240 higher median FLL than the A/A class and the A/G class was intermediate. The direction of this 241 descriptive trend was consistent with the positive allelic effect estimated by GWAS. Overall, the 242 results support a chromosome 6A region associated with flag leaf length, but validation in larger 243 populations and haplotype-based analyses will be important for refining the causal interval. 244 GWAS for tiller density (TM; tillers m⁻²) identified a major locus on chromosome 5D. Both 245 BLINK and FarmCPU showed a distinct association peak in this region, and the QQ plots 246 displayed upper-tail deviation consistent with a true association signal. The lead SNP, 247 5D_205488820, was highly significant (P = 8.773 × 10⁻¹⁰; BH-FDR = 1.901 × 10⁻⁵; Bonferroni 248 adjusted P = 1.901 × 10⁻⁵). The marker had a large negative estimated effect (−190.459 tillers 249 m⁻²) and a MAF of 0.488, indicating that the associated polymorphism was common in the study 250 panel. The genotype-class comparison was also significant (Kruskal-Wallis P = 0.022). The C/C 251 reference genotype showed the highest tiller densities, with a median of approximately 320 tillers 252 m⁻², whereas the C/T class showed substantially lower values, centred around 120-130 tillers 253 m⁻². The T/T homozygous class was not represented in the displayed genotype-class plot. The 254 direction of the phenotypic difference was consistent with the negative allelic effect estimated by 255 GWAS. Taken together, these results identify 5D_205488820 as a strong candidate marker for 256 tiller density that warrants independent validation. ## 257 DISCUSSION 258 This study demonstrates that several morphologically defined DUS characteristics in Pakistani 259 wheat germplasm have a detectable genetic basis and can be associated with discrete genomic 260 regions. Significant marker-trait associations were identified for growth habit, shattering, flag 261 leaf length and tiller density. These findings are relevant to DUS testing because conventional 262 characterization remains predominantly phenotype-based, while the expression of some 263 characteristics can be influenced by environment, developmental stage and observer-related 264 variation. Recent studies have shown that historical DUS records can be integrated with genome 265 wide marker data to investigate the genetic architecture of DUS characteristics in wheat (Zanella 266 _et al._, 2026), while genomic prediction has been proposed as a complementary tool for reference 267 collection management and trait-specific distinctness decisions (Roberts _et al._, 2026). The 268 present study extends these approaches to Pakistani wheat germplasm and provides candidate 269 genomic regions for future character-specific marker development. 9 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 270 Substantial phenotypic variation was evident across the evaluated DUS characteristics, as shown 271 by the contrasting trait distributions and the separation of varieties into five broad phenotypic 272 clusters. Most pairwise trait correlations were relatively weak, indicating that many descriptors 273 captured distinct dimensions of morphology rather than redundant information. This is desirable 274 for varietal characterization because highly correlated descriptors provide less independent 275 discriminatory information. Similar analyses of wheat DUS datasets have also reported 276 substantial variation in the diversity and genetic architecture of individual DUS characteristics 277 (Zanella _et al._, 2026). The use of two complementary multi-locus GWAS methods further 278 strengthened the analysis. FarmCPU iteratively combines fixed- and random-effect models to 279 improve power while reducing confounding between tested markers and kinship (Liu _et al._, 280 2016), whereas BLINK improves computational efficiency and models associated loci without 281 relying on arbitrary fixed LD bins (Huang _et al._, 2019). 282 Growth habit was associated most strongly with 1B_129176704, while a secondary signal was 283 detected at 5B_586820792. The chromosome 5B association is biologically plausible because it 284 lies in the broader genomic context of vernalization-related genes. Growth habit and seasonal 285 adaptation are closely related, and winter wheat generally exhibits a more prostrate vegetative 286 growth habit than spring wheat (Hyles _et al._, 2020). _VRN-B1_ is a major MADS-box regulator of 287 vernalization response and developmental transition in wheat (Milec _et al._, 2013). The 288 chromosome 5B signal was reported to lie approximately 10 Mb from _VRN-B1_ 289 (TraesCS5B02G396600), placing it within the broad LD scale observed in the present 290 germplasm. The homoeologous _VRN1_ loci on chromosomes 5A, 5B and 5D are major 291 determinants of seasonal growth habit and the transition from vegetative to reproductive 292 development. Dominant _VRN1_ alleles reduce or eliminate the vernalization requirement and 293 promote spring growth habit, whereas recessive alleles at major _VRN1_ loci are characteristic of 294 winter growth habit (Santra _et al._, 2009). (Zhang _et al._, 2008) also demonstrated a strong 295 relationship between allelic variation at _Vrn-A1_, _Vrn-B1_, _Vrn-D1_ and _Vrn-B3_ and variation in 296 wheat growth habit. (Guedira _et al._, 2014) further showed that variation in the _VRN-B1_ region 297 affected vernalization requirement, winter dormancy release and heading date. Thus, the 5B 298 association has both positional and functional plausibility. In contrast, the major 1B_129176704 299 signal could not be confidently assigned to a functionally validated wheat gene controlling 300 vegetative growth habit based on the literature considered here. This locus may therefore 10 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 301 represent a population-specific or previously unresolved region and should be prioritized for 302 interval-level annotation and independent validation. 303 Flag leaf morphology is physiologically important because the flag leaf contributes substantially 304 to assimilate production during grain filling, while its dimensions are quantitatively inherited and 305 environmentally responsive (Wang _et al._, 2022). In the present study, FLL was associated with 306 an extended set of markers on chromosome 6A. SNPs between approximately 77.9 and 78.2 Mb 307 had nearly identical association statistics, indicating strong LD, while additional signals occurred 308 at approximately 79.36, 81.24, 86.49, 92.35 and 96.51 Mb. This pattern could reflect one 309 extended LD region or multiple linked loci affecting flag leaf development. Independent support 310 is provided by (Zhang _et al._, 2026), who identified a leaf-length locus, QLfL_C&H-6A, spanning 311 approximately 93.0-99.0 Mb and proposed TraesCS6A03G0284400 as a principal candidate 312 gene. The present 96.51-Mb marker falls within that reported interval, and the 92.35-Mb marker 313 lies immediately proximal to it. This physical correspondence strengthens the biological 314 credibility of the distal 6A association. Haplotype analysis across the broader 6A region will be 315 important for determining whether the 77-81-Mb and 92-97-Mb signals belong to a single 316 extended haplotype or represent distinct loci. The lack of a significant Kruskal-Wallis test for 317 6A_77909780 also emphasizes that the broader locus or haplotype may be more informative than 318 that individual SNP. 319 Shattering produced the strongest statistical association in the study, centred on 1D_445256324. 320 The presence of several highly significant neighbouring SNPs around 445.26 Mb strongly 321 suggests that these markers tag a common LD or haplotype block rather than multiple 322 independent causal variants. A secondary association was detected at 7B_703502819, and 323 genotype-class analyses supported phenotypic differentiation at both chromosome 1D and 7B 324 loci. Shattering in wheat is closely related to rachis brittleness and spike disarticulation 325 (Pourkheirandish and Komatsuda, 2022). Importantly, the present associations do not correspond 326 to the canonical _BTR_ loci responsible for the classical brittle-rachis domestication phenotype. 327 The principal brittle-rachis loci are located on the short arms of homoeologous group-3 328 chromosomes; Br1/BTR-D occurs on 3DS, with related loci on 3AS and 3BS (Liu _et al._, 2022). 329 Wheat _BTR1/BTR2_ homologues have likewise been placed on chromosomes 3A and 3B 330 (Akpinar _et al._, 2018). Consequently, the chromosome 1D association should not be annotated as 331 _BTR1_ or _BTR2_ . 11 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 332 Rachis brittleness is, however, genetically more complex than the canonical _BTR_ pathway alone. 333 (Jiang _et al._, 2014) identified major brittle-rachis QTLs on chromosomes 2DS, 2DL, 3DS and 334 5AL in Tibetan semi-wild hexaploid wheat, demonstrating that loci outside chromosome group 3 335 can modify rachis disarticulation. The domestication gene _Q_ on chromosome 5A provides 336 another example: (Jiang _et al._, 2019) showed that a novel _Q_ allele containing a 161-bp 337 transposon insertion could generate brittle rachis through altered rachis cell-wall properties. 338 (Peleg _et al._, 2011) also reported QTL affecting domestication-related spike characteristics, 339 including a locus on chromosome 7B. Although older genetic maps do not allow confident 340 physical colocalization with 7B_703502819, these studies support the existence of additional 341 genetic modifiers of spike strength and disarticulation. The present 1D and 7B associations may 342 therefore represent modifier loci affecting rachis strength, cell-wall composition, abscission-zone 343 development or mechanical resistance. Because the 1D signal combines very strong statistical 344 support with multiple neighbouring SNPs and clear genotype-phenotype differentiation, it is a 345 high-priority target for conversion into a breeder-friendly assay and validation across 346 independent DUS seasons. 347 Tiller density showed a strong association with 5D_205488820 that remained significant after 348 both FDR and Bonferroni correction. The marker had a MAF of 0.488 and a large negative 349 allelic effect, and the genotype-class comparison provided complementary phenotypic support. 350 The involvement of chromosome 5D in wheat tillering has been reported previously. (Huang _et_ 351 _al._, 2003) identified tiller-number QTL on several chromosomes, including 5D, using advanced 352 backcross material. (Li _et al._, 2010) subsequently identified a major chromosome 5D QTL 353 affecting tiller number at different developmental stages, and (Ren _et al._, 2018) identified the 354 stable locus cqTN-5D.1 across multiple tillering stages. Collectively, these studies establish 355 chromosome 5D as a recurrent genomic contributor to wheat tillering. 356 Nevertheless, previously reported 5D QTL cannot be assumed to be identical to 5D_205488820. 357 (Ren _et al._, 2018) showed that 5D tillering QTL detected in different mapping populations were 358 not necessarily the same locus, and older SSR-based studies lack the physical resolution required 359 for direct correspondence with a SNP at 205.49 Mb. The present association therefore represents 360 strong trait- and chromosome-level support for the importance of 5D, while the exact causal gene 361 remains unresolved. Fine mapping of the LD-defined interval around 5D_205488820, followed 362 by annotation of high-confidence genes and evaluation of their expression and functional 12 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 363 relevance to branching pathways, would help identify plausible causal candidates. The genotype 364 count data underlying this locus should also be checked before publication because the reported 365 MAF is high, whereas the displayed genotype-class plot does not contain a T/T homozygous 366 class. 367 Taken together, the detected loci differ in the degree of external biological support. The distal 368 chromosome 6A FLL region has particularly strong positional support because it overlaps an 369 independently reported leaf-length interval. The chromosome 5D tiller-density signal is 370 supported at the chromosome and trait level by several previous mapping studies. The 1D and 7B 371 shattering loci are statistically strong but do not correspond to the classical group-3 _BTR_ genes, 372 while the major chromosome 1B growth-habit locus remains an unresolved candidate region. 373 These differences emphasize the need to distinguish statistical association from causal-gene 374 assignment and to validate each locus independently before regulatory or breeding application. ## 375 IMPLICATIONS FOR MOLECULAR-ASSISTED DUS TESTING AND 376 PLANT BREEDERS’ RIGHTS 377 The present findings have direct relevance to the emerging use of molecular information in DUS 378 systems. (Zanella _et al._, 2026) demonstrated that loci associated with wheat DUS characteristics 379 can be identified and that trait-associated SNPs can be converted into breeder-friendly KASP 380 assays. Their work also showed that the predictive value of molecular markers varies 381 substantially among traits. This distinction is important because markers intended to support 382 DUS testing should remain demonstrably linked to the expression of recognized DUS 383 characteristics rather than serving only as measures of overall genetic distance. (Roberts _et al._, 384 2026) similarly emphasized the complementary role of molecular information in phenotype 385 based DUS decisions. Their genomic-prediction framework uses historical DUS phenotypes and 386 marker data to support reference-collection management and predict differences in individual 387 DUS characteristics. The present study complements this approach by identifying discrete loci 388 that may be suitable for a character-specific marker strategy. 389 For practical implementation in Pakistan, the chromosome 1D shattering region, chromosome 1B 390 growth-habit locus, chromosome 5D tiller-density locus and distal chromosome 6A flag-leaf 391 length region represent priority targets for validation. Before any regulatory use, lead SNPs 392 should be converted into low-cost assays such as KASP, tested in a larger independent collection 13 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 393 of Pakistani varieties, and evaluated across the same environments and seasons used for official 394 DUS characterization. Marker alleles should show repeatable correspondence with DUS notes, 395 and their sensitivity, specificity, reproducibility and predictive error should be quantified. 396 Haplotype-based assays may be more informative where several significant SNPs tag the same 397 LD block, particularly for the chromosome 1D shattering and chromosome 6A flag-leaf-length 398 regions. Molecular markers should therefore be regarded as complementary evidence that 399 strengthens, rather than substitutes for, conventional multi-season DUS phenotyping. ## 400 Conclusion 401 The present study provides a genome-wide assessment of the genetic basis of diagnostically 402 important DUS characteristics in Pakistani wheat germplasm. The panel displayed substantial 403 phenotypic diversity, extensive genetic admixture and hierarchical population structure, 404 providing a suitable framework for association analysis. Robust genomic signals were detected 405 for growth habit on chromosomes 1B and 5B, shattering on chromosomes 1D and 7B, flag leaf 406 length on chromosome 6A and tiller density on chromosome 5D. Among these, the chromosome 407 1D shattering locus and chromosome 5D tiller-density locus showed particularly strong 408 statistical support, while the distal chromosome 6A flag-leaf-length region also received 409 independent positional support from published wheat studies. The major chromosome 1B 410 growth-habit signal may represent a previously unresolved region and merits detailed functional 411 investigation. These loci should be considered candidate regions rather than immediately 412 deployable diagnostic markers. Independent validation across larger germplasm panels, 413 environments and official DUS seasons is required, together with haplotype analysis, candidate 414 gene annotation and conversion of informative variants into breeder-friendly assays. If validated, 415 character-specific molecular markers could improve the reproducibility and efficiency of varietal 416 identification and provide an additional evidence layer for DUS decisions. Their most 417 appropriate role is as a complement to conventional phenotyping, thereby strengthening the 418 scientific basis of wheat variety protection and Plant Breeders’ Rights implementation in 419 Pakistan. 420 _**Conflict of Interest:**_ The Authors declare no competing interests. 14 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 421 _**Contribution Statements:**_ MA, RAJ and FS collected DUS characters data; IUH and MAH did 422 GWAS and post-GWAS analysis; IAQ and ZA wrote manuscript; IUH and RAJ proof-read 423 manuscript. 424 _**Acknowledgement:**_ Authors acknowledge the Dr. Awais Rasheed, Quaid-e-Azam University, 425 Islamabad, Pakistan for providing genotyping assistance under the Science and Technology 426 Partnership Program from the Ministry of Science and Technology (MoST) of China 427 (KY202201001) and Higher Education Commission (HEC) NRPU grant number 15269. ## 428 REFERENCES 429 Akpinar, B.A., S. Biyiklioglu, B. Alptekin, M. Havránková, J. Vrána, J. Doležel, A. Distelfeld, 430 P. Hernandez, IWGSC and H. Budak. 2018. Chromosome‐based survey sequencing reveals the 431 genome organization of wild wheat progenitor _Triticum dicoccoides_ . Plant Biotechnology 432 Journal. 16(12):2077-2087. 433 Consortium, I.W.G.S., R. Appels, K. Eversole, N. Stein, C. Feuillet, B. Keller, J. Rogers, C.J. 434 Pozniak, F. Choulet and A. Distelfeld. 2018. Shifting the limits in wheat research and breeding 435 using a fully annotated reference genome. Science. 361(6403):eaar7191. 436 Erenstein, O., M. Jaleta, K.A. Mottaleb, K. Sonder, J. Donovan and H.J. Braun. 2022. Global 437 trends in wheat production, consumption and trade. In Wheat improvement: food security in a 438 changing climate (47-66). Springer International Publishing Cham. 439 Farhan, M., M.K. Naeem, A. Muhammad, M. Kiedrzyński, M. Uzair, B. Saleem, Z. Mahmood, 440 A. Mir, M.R. Khan and A.H. Khan. 2026. Genome-wide association analysis reveals natural 441 genetic variations controlling canopy architecture traits in bread wheat. Scientific Reports. 16(1): 442 6433. 443 Government of Pakistan. (2016). Plant Breeders' Rights Act, 2016. 444 Government of Pakistan. (2018). Plant Breeders' Rights Rules, 2018. 445 Government of Pakistan. (2025). Economic Survey of Pakistan, 2024-25. 446 Guedira, M., P. Maloney, M. Xiong, S. Petersen, J.P. Murphy, D. Marshall, J. Johnson, S. 447 Harrison and G. Brown‐Guedira. 2014. Vernalization duration requirement in soft winter wheat 448 is associated with variation at the VRN‐B1 locus. Crop Science. 54(5):1960-1971. 449 Huang, M., X. Liu, Y. Zhou, R.M. Summers and Z. Zhang. 2019. BLINK: a package for the next 450 level of genome-wide association studies with both individuals and markers in the millions. 451 Gigascience. 8(2):giy154. 452 Huang, X., H. Cöster, M. Ganal and M. Röder. 2003. Advanced backcross QTL analysis for the 453 identification of quantitative trait loci alleles from wild relatives of wheat ( _Triticum aestivum_ L.). 454 Theoretical and Applied Genetics. 106(8):1379-1389. 455 Hussain, S., F. Kong, I. Naureen, M. Habib, N. Ghori, F.S. Awan, H.F. Ghuman, K. Hameed and 456 X. Zhang. 2026. GWAS on yield-related traits of Pakistani wheat under post-anthesis heat stress. 457 BMC Genomics. 27(1):351. 458 Hyles, J., M.T. Bloomfield, J.R. Hunt, R.M. Trethowan and B. Trevaskis. 2020. Phenology and 459 related traits for wheat adaptation. Heredity. 125(6):417-430. 460 Iqbal, M.J., N. Shams and K. Fatima. 2022. Nutritional quality of wheat. In Wheat. IntechOpen 461 Rijeka, Croatia. 15 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 462 Jiang, Y.F., X.J. Lan, W. Luo, X.C. Kong, P.F. Qi, J.R. Wang, Y.M. Wei, Q.T. Jiang, Y.X. Liu 463 and Y.Y. Peng. 2014. Genome-wide quantitative trait locus mapping identifies multiple major 464 loci for brittle rachis and threshability in Tibetan semi-wild wheat ( _Triticum aestivum ssp._ 465 _tibetanum Shao_ ). PLoS One. 9(12):e114066. 466 Jiang, Y.F., Q. Chen, Y. Wang, Z.R. Guo, B.J. Xu, J. Zhu, Y.Z. Zhang, X. Gong, C.H. Luo and 467 W. Wu. 2019. Re‐acquisition of the brittle rachis trait via a transposon insertion in domestication 468 gene Q during wheat de‐domestication. New Phytologist. 224(2):961-973. 469 Khan, M.I., Z. Kainat, S. Maqbool, A. Mehwish, S. Ahmad, H.M. Suleman, Z. Mahmood, M. 470 Ali, A. Aziz and A. Rasheed. 2022. Genome-wide association for heat tolerance at seedling stage 471 in historical spring wheat cultivars. Frontiers in Plant Science. 13:972481. 472 Li, Z., T. Peng, Q. Xie, S. Han and J. Tian. 2010. Mapping of QTL for tiller number at different 473 stages of growth in wheat using double haploid and immortalized F2 populations. Journal of 474 Genetics. 89(4):409. 475 Liu, J., Y. Yao, M. Xin, H. Peng, Z. Ni and Q. Sun. 2022. Shaping polyploid wheat for success: 476 Origins, domestication, and the genetic improvement of agronomic traits. Journal of Integrative 477 Plant Biology. 64(2):536-563. 478 Liu, X., M. Huang, B. Fan, E.S. Buckler and Z. Zhang. 2016. Iterative usage of fixed and 479 random effect models for powerful and efficient genome-wide association studies. PLoS 480 Genetics. 12(2):e1005767. 481 Milec, Z., T. Sumíková, L. Tomková and K. Pánková. 2013. Distribution of different Vrn-B1 482 alleles in hexaploid spring wheat germplasm. Euphytica. 192(3):371-378. 483 Nawade, B., K. Do Min, M.B. Lee, D.Y. Hyun, C. Choi and J.Y. Kim. 2025. Integrative multi484 locus GWAS and SNP effect analysis reveal the genetic basis of heading, maturity, and grain485 filling duration in a bread wheat. Plant Physiology and Biochemistry. 110729. 486 Peleg, Z., T. Fahima, A.B. Korol, S. Abbo and Y. Saranga. 2011. Genetic analysis of wheat 487 domestication and evolution under domestication. Journal of Experimental Botany. 62(14):5051488 5061. 489 Pourkheirandish, M. and T. Komatsuda. 2022. Grain disarticulation in wild wheat and barley. 490 Plant and Cell Physiology. 63(11):1584-1591. 491 Rasheed, A. and M. Fayyaz. (2023). Pakistani historical wheat panel 37K SNP data. 492 Ren, T., Y. Hu, Y. Tang, C. Li, B. Yan, Z. Ren, F. Tan, Z. Tang, S. Fu and Z. Li. 2018. 493 Utilization of a Wheat55K SNP array for mapping of major QTL for temporal expression of the 494 tiller number. Frontiers in Plant Science. 9:333. 495 Roberts, A.M., K. Neugebauer, E.O. Ewaoluwagbemiga, D. Milbourne, S. Byrne, J. Cockram, 496 C.M. Zanella, M. Wallace, M. Niedermayer and L. Andreani. 2026. Integrating genomic 497 prediction into crop DUS testing: new approaches in support of reference collection management 498 and distinctness assessment. Theoretical and Applied Genetics. 139(4):93. 499 Santra, D.K., M. Santra, R. Allan, K. Campbell and K. Kidwell. 2009. Genetic and molecular 500 characterization of vernalization genes Vrn‐A1, Vrn‐B1, and Vrn‐D1 in spring wheat germplasm 501 from the Pacific Northwest region of the USA. Plant Breeding. 128(6):576-584. 502 Suleman, H.M., H. Qayyum, S. Ur Rehman, K. Majeed, M. Mukhtar, S. Zulfiqar, Z. Mahmood, 503 A. Aziz, M. Fayyaz and A. Mehvish. 2025. Genetic dissection of plant height in spring wheat 504 from Pakistan using genome-wide SNPs and allele-specific Rht markers. Molecular Breeding. 505 45(11):90. 506 United Nations. 2015. Transforming Our World: The 2030 Agenda for Sustainable 507 Development. 16 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 508 UPOV. 1991. International Convention for the Protection of New Varieties of Plants. Retrieved 509 [from https://upovlex.upov.int/en/convention](https://upovlex.upov.int/en/convention) 510 Wang, Y., L. Qiao, C. Yang, X. Li, J. Zhao, B. Wu, X. Zheng, P. Li and J. Zheng. 2022. 511 Identification of genetic loci for flag-leaf-related traits in wheat ( _Triticum aestivum_ L.) and their 512 effects on grain yield. Frontiers in Plant Science. 13:990287. 513 Zanella, C.M., R. Horsnell, B. Love, T.I. Wright, J. Taferner-Kriegl, C. Flamm, L. Andreani, C. 514 Delogu, V. McMillan and M. Wallace. 2026. Genome-wide association study (GWAS) identifies 515 genetic loci controlling Distinctness, Uniformity, and Stability (DUS) traits in wheat. Theoretical 516 and Applied Genetics. 139(2):49. 517 Zhang, J., W. Xu, R.S. Gowda, J. Johnstone, M. Alinsug, A. Bohra, V. Garg, A. Chitikineni, D. 518 Bennett and M. Zhou. 2026. Genomic and Physiological Insights Into Heat–Drought Tolerance 519 in Wheat Through GWAS and Phenotypic Evaluation. Plant, Cell & Environment. 520 Zhang, X., Y. Xiao, Y. Zhang, X. Xia, J. Dubcovsky and Z. He. 2008. Allelic variation at the 521 vernalization genes Vrn‐A1, Vrn‐B1, Vrn‐D1, and Vrn‐B3 in Chinese wheat cultivars and their 522 association with growth habit. Crop Science. 48(2):458-470. 523 Zhou, B. and A. Irshad. (2026). A Genome-Wide Association Analysis of Resistant Starch and 524 Related Traits in 383 Populations of Wheat ( _Triticum aestivum_ L.) from China and Pakistan. 525 Proceedings of the 2026 6th International Conference on Bioinformatics and Intelligent 526 Computing, 527 528 17 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 529 530 **Figure 1. Phenotypic variation and hierarchical clustering of DUS characteristics in** 531 **Pakistani wheat varieties.** 532 Distribution of standardized phenotypic values and hierarchical relationships among the 102 533 wheat varieties evaluated for DUS characteristics. The violin plots (left) show the distribution, 534 density and central tendency of individual characteristics after standardization to _z_ -scores. The 535 hierarchical dendrogram (right) was generated from the combined phenotypic profiles and 536 separates the evaluated varieties into five major phenotypic clusters, illustrating the extent of 537 morphological diversity within the germplasm panel. 538 18 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 539 **Figure 2. Correlation matrix of DUS characteristics evaluated in Pakistani wheat varieties.** 540 Pairwise correlation coefficients among the evaluated DUS characteristics are presented as a 541 lower-triangular heatmap. The colour scale ranges from −1 to +1, with red indicating positive 542 correlations, blue indicating negative correlations and values approaching white indicating weak 543 or negligible relationships. 544 19 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 545 546 547 **Figure 4. Model-based population structure of Pakistani wheat varieties across different** 548 **numbers of ancestral populations.** 549 Ancestry proportions of the 102 wheat varieties estimated for K = 5–10 ancestral components. 550 Each vertical bar represents an individual variety and different coloured segments indicate the 551 estimated proportional contribution of each ancestral component. Cross-entropy analysis 552 identified K = 9 as the best-supported population-structure model. 20 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 553 554 **Figure 3. Genome-wide genetic diversity, principal component analysis, SNP distribution** 555 **and linkage disequilibrium decay in Pakistani wheat varieties** 556 **(a)** Circular hierarchical dendrogram based on genome-wide SNP data, showing the genetic 557 relationships among the 102 wheat varieties and their separation into five broad genetic clusters. 558 **(b)** Three-dimensional principal component analysis (PCA) based on the filtered SNP dataset; 559 PC1, PC2 and PC3 explain 8.09%, 7.20% and 5.39% of the total genetic variation, respectively. 560 **(c)** Genome-wide distribution of SNPs across the 21 wheat chromosomes, expressed as the 561 number of markers within 1-Mb windows. **(d)** Linkage disequilibrium (LD) decay for the A, B 562 and D subgenomes, expressed as pairwise _r_ ² against physical distance (Mb). The horizontal 563 dashed line represents the reference LD threshold of _r_ ² = 0.20. 564 21 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 565 **Figure 5. Genome-wide association analysis of growth habit, shattering, flag leaf length and** 566 **tiller density using BLINK and FarmCPU models.** 567 Manhattan and quantile–quantile (QQ) plots generated for growth habit (TGH), shattering (SH), 568 flag leaf length (FLL) and tiller density (TM) using the BLINK and FarmCPU GWAS models. 569 For each trait, the BLINK Manhattan and QQ plots are shown alongside the corresponding 570 FarmCPU Manhattan and QQ plots. Manhattan plots display marker significance as −log₁₀( _P_ ) 571 across the 21 wheat chromosomes. Horizontal dashed lines indicate the Bonferroni and 572 Benjamini–Hochberg false discovery rate (BH-FDR) significance thresholds at α = 0.05. QQ 573 plots compare observed and expected −log₁₀( _P_ ) values, with genomic inflation factors (λ) shown 574 for assessment of model calibration and residual population-structure effects. 22 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 575 576 **Figure 6. Genotype-specific phenotypic effects of representative GWAS-associated loci for** 577 **tiller density, flag leaf length, shattering and growth habit in Pakistani wheat varieties.** 578 Boxplots show the phenotypic distributions of genotype classes for six significant or 579 representative marker–trait associations: **(a)** tiller density (TM) at 5D_205488820, **(b)** flag leaf 580 length (FLL) at 6A_92352012, **(c)** shattering (SH) at 1D_445256389, **(d)** shattering at 581 7B_703502819, **(e)** growth habit (TGH) at 1B_129176704, and **(f)** growth habit at 582 5B_586820792. Genotypes are grouped as homozygous reference, heterozygous and 583 homozygous alternate classes, where present. The nucleotide substitution and Kruskal–Wallis _P_ 584 value are shown above each panel. Significant genotype-class differences were observed for TM 585 at 5D_205488820 ( _P_ = 0.022), FLL at 6A_92352012 ( _P_ = 0.0452), SH at 1D_445256389 ( _P_ = 586 0.00361), SH at 7B_703502819 ( _P_ = 0.00561), TGH at 1B_129176704 ( _P_ = 0.00878), and TGH 587 at 5B_586820792 ( _P_ = 0.00376), supporting the phenotypic effects of these GWAS-identified 588 loci. 589 23 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.01.763533; this version posted November 3, 2026. The copyright holder for this preprint](https://doi.org/10.64898/2026.11.01.763533) (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made [available under a CC-BY 4.0 International license.](http://creativecommons.org/licenses/by/4.0/) 590 **Table 1. Significant marker-trait associations identified for flag leaf length, shattering,** 591 **growth habit and tiller density in Pakistani wheat varieties.** 592 |Trait|SNP|Alleles*|Pvalue|BH-FDR|Bonf adj P<br>. .|Effect|MAF| |---|---|---|---|---|---|---|---| |FLL|6A_77909780|G/A|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_77913131|T/C|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_77913187|T/G|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_77960267|T/C|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_77999760|C/A|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_77999868|A/C|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_77999899|C/A|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_78065495|G/A|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_78065597|A/G|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_78152531|A/G|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_78152577|T/C|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_78230010|T/C|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_78230072|A/C|6.037e-06|6.541e-03|1.308e-01|4.057|0.102| |FLL|6A_79357043|G/C|2.155e-06|6.541e-03|4.669e-02|4.057|0.102| |FLL|6A_79357055|A/C|2.155e-06|6.541e-03|4.669e-02|4.057|0.102| |FLL|6A_81237733|A/T|2.155e-06|6.541e-03|4.669e-02|4.057|0.102| |FLL|6A_81237750|C/G|2.155e-06|6.541e-03|4.669e-02|4.057|0.102| |FLL|6A_86485742|C/T|2.155e-06|6.541e-03|4.669e-02|4.057|0.102| |FLL|6A_92352012|A/G|1.647e-06|6.541e-03|3.568e-02|3.503|0.124| |FLL|6A_96508113|T/G|1.647e-06|6.541e-03|3.568e-02|3.503|0.124| |SH|1D_445256324|A/G|5.207e-11|1.128e-06|1.128e-06|-1.927|0.492| |SH|1D_445256334|A/G|8.307e-10|3.600e-06|1.800e-05|1.109|0.492| |SH|1D_445256342|T/C|8.307e-10|3.600e-06|1.800e-05|1.109|0.492| |SH|1D_445256363|T/G|8.307e-10|3.600e-06|1.800e-05|1.109|0.492| |SH|1D_445256389|T/C|8.307e-10|3.600e-06|1.800e-05|1.109|0.492| |SH|7B_703502819|C/G|1.414e-07|5.106e-04|3.064e-03|0.245|0.144| |TGH|1B_129176704|A/G|3.756e-09|8.140e-05|8.140e-05|1.610|0.053| |TGH|5B_586820792|C/T|2.333e-06|2.528e-02|5.056e-02|1.125|0.067| |TM|5D_205488820|T/C|8.773e-10|1.901e-05|1.901e-05|-190.459|0.488| 593 Significant SNPs detected by genome-wide association analysis are presented with their 594 chromosomal position-based marker identifier, observed alleles, raw _P_ value, Benjamini– 595 Hochberg false discovery rate (BH-FDR), Bonferroni-adjusted _P_ value, estimated allelic effect 596 and minor allele frequency (MAF). Multiple closely linked SNPs on chromosome 6A were 597 associated with flag leaf length (FLL), while shattering (SH) showed a major association block 598 on chromosome 1D together with a secondary locus on chromosome 7B. Growth habit (TGH) 599 was associated with loci on chromosomes 1B and 5B, whereas tiller density (TM) showed a 600 strong association with 5D_205488820. Positive and negative effect values indicate the direction 601 of the estimated allelic effect according to the allele coding used in the GWAS analysis. 602 - **Alleles** are reported as the two nucleotide states observed at each SNP. 603 24