Identification of candidate loci for leafing date in Iranian walnut germplasm collection using GBS-GWAS approach and validation by SNaPshot Authors: Mehdi Fallah, Mohammad M Arab, Harun Karcı, Saadat Sarikhani, Mousa Rasouli, Darab Hassani, Salih Kafkas, Kourosh Vahdati Published: 2026 ABSTRACT AbstractLate-spring frost is increasingly damaging walnut orchards worldwide, underscoring the urgent need for late-leafing cultivars adapted to climate instability. Despite extensive genomic resources in Juglans regia, additional robust markers for leafing date are still needed to enhance breeding efficiency. Here, we combine high-density genotyping-by-sequencing (GBS) with multi-model genome-wide association analysis (GWAS) to dissect the genetic architecture of leafing date in a uniquely diverse panel of 85 Iranian walnut genotypes and commercial cultivars evaluated over three consecutive years. Leafing date was defined as the stage when 50% of terminal buds reached budburst, quantified as the cumulative number of days since March 1st. Population structure analysis revealed three major genetic clusters corresponding to geographic origin, reflecting the deep evolutionary history of the Iranian walnut. Using 68 K high-quality Single-nucleotide polymorphisms )SNPs(, GWAS consistently identified significant loci on chromosomes 1, 4, 6, 8, 11, and 13. Notably, the most robust associations were consistently detected across the 2017 and 2018 growing seasons, as well as in the multi-year average data, using MLMM, Blink, and FarmCPU models. Based on these stable associations, a SNaPshot marker panel was developed and validated in an independent set of 24 Iranian and Turkish genotypes. Among these, a single marker—JRgwLDch4pos20189439 on chromosome 4—exhibited strong and repeatable discrimination between early- and late-leafing phenotypes across populations. This locus represents a key experimentally validated leafing-date marker for Iranian walnut germplasm and provides a powerful tool for marker-assisted selection (MAS). Our findings reveal the multi-allelic nature of leafing date and demonstrate that markers may vary among geographically distinct germplasm pools. The validated chr4-associated marker offers a promising genomic resource that may support efforts to accelerate breeding of climate-resilient walnut cultivars in Iran and potentially other regions with related genetic backgrounds.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12870-026-09026-3.Keywords: Juglans regia L., Budbreak timing, Late-spring frosts, Quantitative trait dissection, Temperate tree genomics, Linkage disequilibrium Introduction Persian walnut (Juglans regia L.) originated in Asia, especially Iran, where it has been cultivated for centuries and remains highly esteemed among nut fruit growers [1]. China, the USA, Iran, and Türkiye are the leading global walnut producers, accounting for 35.1%, 18.7%, 9.6%, and 9.0% of world production, respectively (collectively contributing over 72%). Between 2013 and 2023, Iran’s average annual production was approximately 350,000 tonnes [2]. However, recent climate anomalies, particularly late-spring frosts (LSFs), have led to significant production fluctuations. The spring frost damage is influenced by the leafing date [3]. After leafing, trees lose their frost resistance, and if the temperature drops back to or near freezing, frost damage might happen [4]. Delayed leafing reduces damage from late-spring frost, and leafing date is an important consideration in cultivar selection [5]. Therefore, breeding for late-leafing cultivars is effective for stabilizing walnut production [6, 7]. Traditional breeding such as hybridization, genetic diversity exploitation and germplasm screening are common strategies to obtain late-leafing genotypes resistant to LSFs. Additionally, recent advancements in molecular breeding approaches such as genome-wide association studies (GWAS) and marker-assisted selection (MAS) have provided us novel avenues to boost the effectiveness of fruit tree breeding, leading to the release of promising cultivars resistant to LSFs [7–11]. By utilizing MAS and genomic selection, breeders can overcome the constraints of walnut’s long juvenile phase and high maintenance costs [12, 13]. These molecular tools allow for the rapid identification of essential commercial traits at an early developmental stage, streamlining a process that traditionally takes many years [14–16]. GWAS is an association mapping approach that exploits historical recombination in diverse germplasm panels, enabling marker–trait association discovery without the need for bi-parental segregating populations. In plant breeding, association mapping is particularly valuable because it can reveal loci and closely linked markers that are directly usable for marker-assisted selection [17, 18]. In GWAS, the first step is to align the reads to reference genome sequences to mine variants to establish the correlations among genotypic and phenotypic data. Morevoer, strongly linked loci inherited together in meiosis can be detected thanks to linkage disequilibrium (LD), which provides the opportunity to predict putative genes associated with traits. As selected loci are associated with a target trait, GWASs can be preferred for the detection of QTL (Quantitative Trait Loci) genomic regions and strongly linked molecular markers within the QTL to be used for future breeding programs [19]. In the last decade, several studies have focused on identifying the genomic regions governing phenological and morphological traits and abiotic stress responses in walnuts utilizing the GWAS approach [20–24]. The genomic age for research on walnut biology began with the publication of the first reference genome of J. regia, which was assembled from short reads by de novo assembly [25], followed by the introduction of the first high-density Axiom™ J. regia 700 K SNP genotyping array [21]. This permitted genetic dissection of walnut traits including nut-related traits and water use efficiency [20, 26, 27]. Findings from the combination of GWAS and quantitative trait locus (QTL) mapping revealed significant loci for lateral bearing on Chr11, and leafing date and harvest dates on Chr 1 [21]. Furthermore, GWAS and QTL mapping analyses for walnut pellicle color revealed loci on Chr 1, Chr 10, Chr 14, and Chr 16 to be associated with this trait [24]. To further investigate the phenological traits of walnuts, the genetic control of leafing date and female/male flowering dates, using the Axiom™ J. regia 700 K SNP array, was performed on 170 geographically diverse walnut accessions and an F1 progeny [22]. Using GWAS, researchers identified strong associations of single-nucleotide polymorphisms (SNPs) located at the beginning of chr 1 with leafing (physical position: 6,514,832 bp) and female flowering dates, leading to the development of a Kompetitive Allele Specific polymerase chain reaction (PCR) (KASP) marker for leafing date, which was validated using a walnut population of the University of California, Davis [22]. In Türkiye, a study was carried out on the genetic control of phenological traits of walnuts in 2019. Kefayati et al. [28] discovered a major QTL for leafing date on linkage group 4 (LG4) contributing to 52.0-68.8% of the phenotypic variance. Then, SSR markers JRHR209732 and CUJRB012 were introduced as the molecular markers for leafing date in Türkiye’s walnut population. JRHR209732 and CUJRBO12 SSR markers were aligned with the walnut genome (Chandler v2.0) and mapped with 100% matching on the chr1 within the chromosomal position at 4,097,966–4,098,222 bp and 4,545,922–4,546,022 bp [29]. Kefayati et al. [28] results are in line with Marrano et al. [21] findings, which identified a SNP significantly linked to leafing date within the same regions on Chr1 (physical position: 4,650,186 bp). Furthermore, by using DArT‑seq for genotyping, Bükücü et al. [23] identified a marker associated with leafing date with significant association in linkage group (LG) 4 (physical position: 28418795 bp). Despite global reports on walnut QTLs, the genetic potential of Iranian germplasm remains unexplored within a high-resolution framework. We hypothesized that integrating ~ 68 K SNPs with multi-year phenotyping and multi-model GWAS would identify stable loci governing leafing date. Here, we investigated 85 diverse individuals evaluated over three years. Our objectives were (i) to identify candidate genes via multi-model GWAS and (ii) to validate these findings using a SNaPshot marker system. This study addresses the existing knowledge gap by bridging high-density discovery with explicit validation, providing a robust tool for marker-assisted selection (MAS) and the development of late-leafing walnut genotypes. Materials and methods Plant material A panel of 85 genotypes, including native Iranian genotypes and commercial cultivars, was used as to identify markers associated with the leafing date using the GWAS based on GBS data. The plant samples were gathered from the walnut research orchard at the Horticultural Science Research Institute (HSRI), located in Karaj, Iran (35°75′ N, 50°95′ E, Fig. S1a-c). Fig. S1a–c presents the climatic profiles of Karaj, including temperature parameters, cloud cover and humidity, as well as rainfall and the number of rainy days during 2017, 2018, and 2019. The study was conducted using a Completely Randomized Design (CRD) with three replicates (individual trees) per genotype. The population (85 12-year-old individuals) consisted of 74 superior native Iranian genotypes, which were originally collected as scions from diverse geographical regions across Iran. They were originally collected as scions from diverse geographical regions across Iran. These genotypes were derived from mother trees of seedling origin and have undergone years of evaluation as part of the Iranian walnut breeding program. Additionally, three Iranian cultivars (‘Damavand’, ‘Jamal’, and ‘Alvand’), which were developed through selection programs, were included. The panel also featured several international commercial cultivars to ensure broad genetic diversity: four of US origin (‘Serr’, ‘Chandler’, ‘Pedro’, and ‘RDM’), three of French origin (‘Ferjean’, ‘Franquette’, and ‘Fernor’), and one Hungarian cultivar (‘Alsószentiváni 117’). To ensure data consistency and minimize measurement bias, phenotypic evaluations for the leafing date were performed on the same individual trees across all study years. All observations were recorded by a single trained researcher. Phenological stages were recorded according to the International Plant Genetic Resources Institute (IPGRI) walnut descriptor. To facilitate statistical and GWAS analysis, the observed calendar dates were converted into numerical values relative to March 1st (Day 0). This conversion was performed to transform phenological observations into a continuous numerical scale suitable for linear modeling and association mapping. This walnut germplasm has sufficient genetic variability regarding important breeding traits especially leafing date since walnut breeding field was set up with various walnut seedlings collected from different regions of Iran. Phenotyping of the populations The phenological observation of the leafing date in walnut individuals was recorded based on IPGRI descriptor for three successive years, 2017, 2018 and 2019. In addition to the three-year data, the three-year average of leafing date was also calculated and included to analysis. Therefore, four sets of phenotypic data were prepared for GWAS analyses. The leafing date was the date when 50% of terminal buds have enlarged and the bud scales have split exposing the green leaves. The leafing date was reported according to March 1st [30, 31] (Table S1). DNA extraction, GBS library construction, sequencing and SNPs calling Dry leaf segments (~ 50 mg) were placed in a 96-well plate (1.2 ml, 8-tube strip format) and ground using a GenoGrinder 2000 (Spex CertiPrep Inc., Metuchen, NJ, USA) at 650 strokes/min for 15 to 30 s. DNA extractions were performed using 96-well plates of silica columns with the modified cetyltrimethylammonium bromide (CTAB) method following a standard procedure as described previously (Doyle and Doyle, 1987). The DNA concentration was measured with a Qubit dsDNA device, and after preparing DNA with a concentration of 50 ng/µl, 10 µl was used for library preparation. GBS was performed on 85 walnut samples using a modified GBS protocol with two restriction enzymes, HindIII-HF (3U) and MseI (1.5U), followed by the ligation of individual-specific barcode and common adapters using T4 DNA ligase [32]. Subsequently, PCR amplification was performed on the ligated samples, and the genomic library was prepared. The libraries were sequenced using 90 bp single-end reads. Sequencing was performed using the Illumina HiSeq4000 at the University of California Davis Genome Center (https://dnatech.genomecenter.ucdavis.edu/illumina-library-sequencing) [33]. Raw sequence data were processed for SNP calling using the TASSEL 5 GBS v2 pipeline (v5.2) against the Juglans regia cv. Chandler v2.0 reference genome [34]. BWA (v0.7.17) [35] was used to align 64 bp tags to the reference genome, retaining only tags that occurred at least ten times in the dataset and had a BWA MAPQ score of at least 20. Initially, SNPs with a read depth < 5 were set as missing. To ensure high data quality before imputation, we discarded taxa and SNPs with > 70% missing data, resulting in a raw dataset of 115,420 SNPs. Missing data were then imputed using Beagle 5.0 [36]. The imputation quality was monitored using the Beagle R2 (DR2) values, and only imputed genotypes with a high confidence score (R2 > 0.8) were retained. Following imputation, a final rigorous filtering step was applied to exclude SNPs with a Minor Allele Frequency (MAF) < 0.05. This process yielded a final high-quality dataset of 68,671 SNPs across 85 individuals. The impact of imputation on the dataset was assessed by comparing the Linkage Disequilibrium (LD) decay and MAF distribution before and after the process, showing no significant bias in the genetic structure. Next, population structure and LD of the dataset were estimated and visualized using the GAPIT R package [37]. Genetic diversity and population structure analysis To characterize the genetic background of the population, genetic diversity parameters including observed (Ho) and expected (He) heterozygosity were calculated using the adegenet R package (v2.1.3). Population structure and individual relatedness were estimated using the GAPIT R package (v3.0; [37]). Principal Component Analysis (PCA) was performed to determine the main axes of genetic variation, and the optimal number of clusters was assessed using a Scree plot. A kinship matrix (K) was also calculated using the VanRaden algorithm implemented in GAPIT to account for internal family structures. Genome-wide association analysis GWA analysis was performed on the 85 walnut individuals for leafing date using phenotypic data from three consecutive years (2017, 2018, and 2019) and their mean values. Single trait GWAS was conducted by applying three different models implemented in the R package Genome Association and Prediction Integrated Tools (GAPIT v3.0; [37]): (i) the Fixed and Random Model Circulating Probability Unification (FarmCPU); (ii) the Multiple Loci Linear Mixed Model (MLMM); and (iii) Linkage-disequilibrium Iteratively Nested Keyway (BLINK). To control for spurious associations, population structure (Q) and genetic relatedness (K) were explicitly incorporated into the models. Specifically, the first three principal components (PCs) were included as fixed effects to account for population stratification, with the number of PCs determined based on the Scree plot analysis. Additionally, a kinship matrix (K) was included as a random effect to account for individual relatedness among the genotypes. Quantile–quantile (QQ) plots were applied to check if the model was correctly fitted. To account for the multiple-testing burden arising from the use of multiple models and datasets, a dual reporting framework distinguishing genome-wide significant and suggestive associations was adopted. While the individual genome-wide significance threshold was set at 7.28 × 10− 7 (Bonferroni correction, α = 0.05/68,671), We classified SNPs as genome-wide significant or suggestive based on the Bonferroni threshold (P ≤ 7.28 × 10⁻⁷). For loci that did not reach genome-wide significance, consistency across models and years was considered as additional supporting evidence to highlight potentially stable associations. This cross-validation approach significantly reduces the likelihood of false-positive associations and ensures the robustness of the identified genomic loci. Furthermore, the results from each year and each model were reported separately to maintain transparency regarding the consistency of the associations. Development of SNaPshot markers for the leafing date Significant and suggestive SNPs identified in the GWAS analysis were selected to validate and detect putative candidate genes controlling leafing date. Based on GWAS results regarding the leafing date, SNaPshot primers were developed [38] to target the most significantly associated SNP with this trait. Six SNPs were most strongly associated with the leafing date. SNaPshot markers were designed using the online software BatchPrimer3 using default settings. First, based on the Chandler v2 genome, a 300-base sequence from SNP flanking regions was extracted for each highly linked locus (Table S2). Then, (Single-Base Extension) SBE-SNP forward and reverse primers were designed from near the SNP genomic regions (Table S3). In addition to the six SNPs, a primer was also designed to test an SNP associated with the leafing date reported by Bernard et al. [22] and to assess its efficiency in the Iranian walnut population. Therefore, a total of seven SNP primers was selected for SNaPshot analysis (Table 1). Table 1SNaPshot primers were designed for the leafing date based on GWAS results from the Iranian walnut population, as well as primer design for loci associated with the leafing date in walnut populations from other countriesWalnut populationChrPhysical positionTypePrimer name5’ − 3’Iran11,427,218SNPJRGWLDch1p1427218FCCATGATACAGGAAATCTACATACAJRGWLDch1p1427218RTCTTGTTGGTGGTGCAGAGTASBEJRGWLDch1p1427218SBEFACAGGAAATCTACATACAGAGTCCTJRGWLDch1p1427218SBERACAAAAGTAGCATAAAAAGCTTCAA420,189,439SNPJRGWLDch4p20189439FAAGCCTGGATAAATTCCCTATTJRGWLDch4p20189439RTTCTCAGCGGAAACAAACTTACSBEJRGWLDch4p20189439SBERTCATGGTCAACTTCCGTCC69,541,059SNPJRGWLDch6p9541059FCCTTGCAGAAATTGTGGAGTAGJRGWLDch6p9541059RTTCTTGGAGGGTTGGGTAGTSBEJRGWLDch6p9541059SBEFGCATAATTTTGAGGAGGAGACATGJRGWLDch6p9541059SBERATTGGCACATCTCTTCATTCAGT69,488,618SNPJRGWLDch6p9488618FCCAAAGGTCCAACCTCATTACTTJRGWLDch6p9488618RAAGTGGGATCGAGGGTATCASBEJRGWLDch6p9488618SBEFCAAACATTTCCCTAACTTCCATAACJRGWLDch6p9488618SBERCTTTAATCTTGGTGTGATAAGCTTG815,615,923SNPJRGWLDch8p15615923FGCTTTCCCCATTTTGCTTACJRGWLDch8p15615923RCATCTCTTCATTTGGCTCCTTATSBEJRGWLDch8p15615923SBEFGCCAAGCAAACCAAAAAAAAJRGWLDch8p15615923SBERGACCAATGCACTTTCAAAGCTT1116,581,922SNPJRGWLDch11p16581922FCTGCAAAGCTTGAATGTGATTAJRGWLDch11p16581922RTTTCTCTATTTTGGTAAACGGTATCSBEJRGWLDch11p16581922SBEFTTCCATTATGAAAACAAAAAGTAACGJRGWLDch11p16581922SBERGCAAAGCGATCCCTTGTCTFrance*16,514,832SNPJRGWLDch1p6514832FAGCATAAATGCCGAGCTATCJRGWLDch1p6514832RGAAGGTTCTGCCAAGCTACAGSBEJRGWLDch1p6514832SBEFGGACAGCAATAAACTCAATCACACJRGWLDch1p6514832SBERCACTGTCCAGCGACGAGATürkiye**14,097,966–4,098,222SSRJRHR209732-FTGTTCAGATGGATCGATTAGGAGSSRJRHR209732-RCCTCTTTCTTTAGATGCTTGTGG4,545,922–4,546,022SSRCUJRBO12-FACTCATCAAGATCCCCGACTACSSRCUJRBO12-RCCACATCGTCTGGGTTCAT*Candidate loci associated with the leafing date in the French walnut population, as reported by Bernard et al. [22]**SSR primers for the leafing date in the Turkish walnut population, introduced by Kefayati et al. (2019) [28] SNaPshot primers were designed for the leafing date based on GWAS results from the Iranian walnut population, as well as primer design for loci associated with the leafing date in walnut populations from other countries *Candidate loci associated with the leafing date in the French walnut population, as reported by Bernard et al. [22] **SSR primers for the leafing date in the Turkish walnut population, introduced by Kefayati et al. (2019) [28] Validation of leafing date markers Validation processes were carried out in the Molecular Genetics Laboratory of the Horticulture Department in Çukurova University, Adana, Türkiye. To validate and evaluate the efficiency of the primers in distinguishing the walnut population based on leafing date, a validation panel consisting of a population of 24 individuals (Table S4) was used, including 7 very early-leafing and 7 very late-leafing Iranian cultivars/genotypes, as well as 5 very early-leafing and 5 very late-leafing Turkish cultivars/genotypes. For this purpose, the extracted genomic DNAs of 24 walnut accessions were diluted to 5 ng DNA concentrations for the first PCR. First, PCR analyses were performed and then loaded to gel electrophoresis prepared with 3% agarose to check the amplification before the cleaning. For the validation of highly linked loci, the SNaPshot analysis was carried out according to manufacturer’s protocol with minor modifications using an Applied Biosystems SNaPshot Multiplex Kit [Applied Biosystems (ABI) Inc., UK]. Reactions were carried out according to Kafkas et al. [38]. One µL of cleaned PCR product was mixed with 9.8 µL Hi-Di formamide and 0.2 µL GeneScan-120 LIZ Size Standard [Applied Biosystems (ABI) Inc., Foster City, CA]. Products were denatured at 95 °C for 5 min and capillary electrophoresis was employed in an ABI PRISM 3130xl Genetic Analyzer [Applied Biosystems (ABI) Inc., Tokyo-Japan], which has a 36-cm-length capillary array and POP-7 polymer [Applied Biosystems (ABI) Inc., Foster City, CA]. The output data from the ABI device were analyzed using GeneMapper v4.0 software, and based on the output charts, the success rate of each primer in separating early-leafing from late-leafing samples was determined [38]. Results Phenotypic diversity of leafing date In the present study, the leafing date of 85 individuals of Iranian walnut genotypes was evaluated for three consecutive years (2017–2019). Following the method of Bükücü et al. [23], in addition to the three-year phenotypic data, the three-year average of leafing date was calculated. Ultimately, four sets of phenotypic data, as shown in Table S1, were used for GWAS analyses. The phenotypic evaluation revealed consistent and significant variation in leafing date within the population across all study years. The range of leafing dates (from earliest to latest genotype) was 31 days in 2017, 33 days in 2018, and 34 days in 2019, with a three-year average of 32 days. Detailed annual summary statistics, including mean values distributions for each year, are provided in Table S1. This substantial and stable genetic diversity in leafing date within the Iranian walnut germplasm provided a robust foundation for the successful identification of associated genomic markers via GWAS. PCA analysis The genetic structure of the 85-member population of Iranian walnut cultivars and genotypes was analyzed using PCA, based on 68,671 high-resolution SNPs retained after filtering (Fig. 1a/b). The results indicated that the studied walnut population exhibited high genetic diversity, which were divided into three main groups based on their genetic structure. The heterozygosity frequency distribution of the population showed that over 80% of the cultivars and genotypes exhibited heterozygosity levels between 0.2 and 0.3. Given the high rate of cross-pollination in walnuts, these results were expected. Additionally, the heterozygosity frequency of the 68,671 SNP markers ranged from 0 to 0.6, with SNP markers with heterozygosity between 0.1 and 0.2 being the most frequent (Fig. 1c/d). Population structure and genetic diversity of 85 Iranian walnut genotypes based on 68,671 GBS-derived SNPs. a Two-dimensional (PC1 vs. PC2) and (b) three-dimensional (PC1, PC2, and PC3) Principal Component Analysis (PCA) plots, revealing the distribution of genotypes into three distinct genetic clusters. The clusters in panel (a) are highlighted with colored ellipses to represent the primary genetic groupings identified through the PCA decomposition. c Frequency distribution of heterozygosity across 85 individuals and (d) distribution of heterozygosity for the 68,671 SNP markers. All analyses were performed using the GAPIT R package, and the population stratification was inferred based on the first three principal components, which captured the majority of the genetic variance A clustering analysis of 85 Persian walnut individuals, based on genetic data from 68,671 SNP markers, categorized the cultivars and genotypes into three groups (Fig. 2). The first group consisted of 11 cultivars and genotypes, mostly including late-leafing cultivars originating from the U.S. and France. The second group comprised 21 cultivars and genotypes, characterized predominantly by early-leafing traits, including samples from the northwest of Iran, such as Tabriz and Urmia, along with some samples from Tehran and Alborz provinces. The third group, the largest, included 53 cultivars and genotypes, mostly mid- to early-leafing types with origins in Zanjan and Qazvin provinces. Additionally, some other genotypes from Tehran and Alborz provinces were also present in this group. a Heat map plot of kinship matrix using average linkage clustering based on SNP markers depicts the existence of three different groups among walnut genotypes, b Determination of the optimal value of K = 3 Linkage disequilibrium analysis The LD decay analysis reveals a sharp decline in linkage disequilibrium (r²) with increasing physical distance between genetic markers, followed by a gradual plateau (Fig. 3). This trend, captured by a fitted nonlinear regression curve, reflects underlying recombination dynamics and the extent of local genetic correlation. A commonly applied threshold of r² = 0.2 delineates the point beyond which LD becomes negligible, providing a basis for determining optimal marker density in genome-wide studies. The rate of LD decay is a fundamental determinant of the resolution and statistical power of association mapping, as it reflects the historical recombination landscape within the population. In this Iranian walnut germplasm, the relatively rapid LD decay indicates that the panel possesses sufficient recombination to enable high‑resolution GWAS, facilitating the fine‑mapping of candidate genomic regions. However, such rapid decay also implies the need for a dense marker set to ensure adequate coverage of functional variants. Conversely, genomic regions exhibiting slower LD decay (extended LD blocks) suggest that significant markers may represent broader haplotypic segments rather than single causal variants. Based on our global LD patterns, the marker density used in this study effectively captures the existing LD structure, providing adequate mapping resolution for identifying loci associated with leafing date. Linkage Disequilibrium (LD) Decay Across Physical Distance in the Iranian Walnut (Juglans regia) Population Genome-Wide Association Study (GWAS) GWA analyses using MLMM, Blink, and FarmCPU models were performed with phenotypic data for the leafing date from 2017 to 2019, as well as the three-year average leafing date. Genome-wide significant and suggestive genomic loci associated with leafing date were identified (Tables 2 and 3). The analyses identified signals including both genome-wide significant and suggestive associations on Chr 4, 6, 8, 11, and 13 related to the leafing date (Fig. 4). The GWAS results using the MLMM model with the 2018 phenotypic data showed the most significant genome-wide association with leafing date on Chr 4 (Fig. 4). The most important associated SNP was located at the physical position 20,189,439 bp with a p-value of 9.37E-19. In addition, a genome-wide significant association was observed on Chr 6 (physical position: 9,541,059 bp; p-value: 7.43E-12), Chr 8 (physical position: 15,615,923 bp; p-value: 5.48E-11), Chr 11 (physical position: 16,193,288 bp; p-value: 7.44E-15) and Chr13 (physical position: 11,217,089 bp. p-value of 3.13E-16). We also plotted boxplots of the allelic effects for SNP S4-20189439 on leafing date, using phenotypic data from 2017 to 2019 and a three-year mean (Fig. 5). Table 3Summary of GWAS results for leafing date using the MLMM modelGWAS using MLMM modelDatasetChrPhysical positionP-valueSignificance category R 2 20171116,193,2888.48E-08GWS0.08323,840,1652.60E-05SUG0.09420,189,4393.963E-05SUG0.09420,189,4733.963E-05SUG0.09815,615,9234.921E-05SUG0.09136,308,2025.776E-05SUG0.03136,308,2195.776E-05SUG0.03112,663,2157.35E-05SUG0.05716,267,1617.66E-05SUG0.05420,222,3668.771E-05SUG0.07420,223,0538.771E-05SUG0.072018420,189,4399.373E-19GWS0.091311,217,0893.138E-16GWS0.03111,941,0951.6E-10GWS0.09718,231,4484.683E-09GWS0.07310,979,7272.118E-08GWS0.0416,142,8901.873E-06SUG˂0.01411,396,4322.14E-05SUG0.01517,160,0234.318E-05SUG0.01111,423,2935.206E-05SUG0.0355,808,9625.959E-05SUG0.02111,403,9077.761E-05SUG0.031014,742,8788.333E-05SUG0.09201969,541,0593.297E-07SUG0.121323,241,8372.791E-05SUG0.041323,241,8852.791E-05SUG0.04314,554,4403.873E-05SUG0.04166,813,3216.497E-05SUG0.05629,048,0436.94E-05SUG0.03Mean69,541,0591.015E-08SUG0.12420,200,6961.302E-06SUG0.071326,091,5173.377E-05SUG0.071322,724,5765.379E-05SUG0.011326,091,4965.661E-05SUG0.081116,539,4588.338E-05SUG0.09GWS = genome-wide significant (P ≤ 7.28 × 10⁻⁷); SUG = suggestive (P ≤ 10 − 4) Summary of GWAS results for leafing date using the MLMM model GWS = genome-wide significant (P ≤ 7.28 × 10⁻⁷); SUG = suggestive (P ≤ 10 − 4) GWAS results displaying Manhattan plots, Quantile–quantile plots and Boxplots for leafing date trait. a MLMM model, b Blink model, c FarmCPU model Box plots of the allele effects for the SNP S4-20189439 associated with leafing date, using phenotypic data from 2017 to 2019 and the three-year mean Table 2Summary of GWAS results for leafing date using the FarmCPU and BLINK modelsGWAS using FarmCPU modelGWAS using Blink modelDatasetChrPhysical PositionP-valueSignificance category R 2 ChrPhysical positionP-valueSignificance category R 2 20171116,193,2887.44E-15GWS0.081116,193,2881.91E-10GWS0.08815,615,9235.48E-11GWS0.06323,840,1652.6E-05SUG0.09323,840,1654.38E-07GWS0.09420,189,4393.96E-05SUG0.091616,009,5146.39E-07GWS0.03420,189,4733.96E-05SUG0.09125,825,2108.81E-07SUG0.03815,615,9234.92E-05SUG0.06420,200,6965.73E-06SUG0.07136,308,2195.78E-05SUG0.0312713,9137.53E-06SUG˂0.01136,308,2025.78E-05SUG0.03114,100,7814.74E-05SUG˂0.01112,663,2157.35E-05SUG0.051037,181,9688.81E-05SUG˂0.01716,267,1617.66E-05SUG0.05420,222,3668.77E-05SUG0.07420,223,0538.77E-05SUG0.072018812,918,4632.87E-06SUG0.06812,918,4632.87E-06SUG0.06420,189,4394.46E-06SUG0.09420,189,4394.46E-06SUG0.09420,189,4734.46E-06SUG0.09420,189,4734.46E-06SUG0.09813,424,6454.82E-06SUG0.07813,424,6454.82E-06SUG0.07420,251,5808.71E-06SUG0.08420,251,5808.71E-06SUG0.0845,955,6451.54E-05SUG0.0645,955,6451.54E-05SUG0.06432,553,6851.92E-05SUG0.06432,553,6851.92E-05SUG0.06135,077,7302.16E-05SUG0.07135,077,7302.16E-05SUG0.07815,206,0342.48E-05SUG0.05815,206,0342.48E-05SUG0.05815,206,0522.48E-05SUG0.05815,206,0522.48E-05SUG0.05633,875,9283.49E-05SUG0.05633,875,9283.49E-05SUG0.05633,875,9203.49E-05SUG0.05633,875,9203.49E-05SUG0.05420,223,0533.5E-05SUG0.07420,223,0533.5E-05SUG0.07420,222,3663.5E-05SUG0.07420,222,3663.5E-05SUG0.07426,945,0515.62E-05SUG0.08426,945,0515.62E-05SUG0.081110,919,7256.63E-05SUG0.071110,919,7256.63E-05SUG0.071136,552,8036.89E-05SUG0.071136,552,8036.89E-05SUG0.071116,581,9227.75E-05SUG0.091116,581,9227.75E-05SUG0.091116,581,9547.75E-05SUG0.091116,581,9547.75E-05SUG0.09415,085,5837.79E-05SUG0.07415,085,5837.79E-05SUG0.07825,359,0217.92E-05SUG0.05825,359,0217.92E-05SUG0.0569,541,0599.69E-05SUG0.1269,541,0599.69E-05SUG0.12812,908,9669.93E-05SUG0.05812,908,9669.93E-05SUG0.05201969,541,0597.34E-07SUG0.1269,541,0593.78E-09GWS0.1269,497,5808.05E-06SUG0.091323,241,8372.79E-05SUG0.041313,426,4058.57E-06SUG0.071323,241,8852.79E-05SUG0.0469,488,6181.09E-05SUG0.09314,554,4403.87E-05SUG0.05427,706,6191.59E-05SUG0.07166,813,3216.5E-05SUG0.0569,484,8122.01E-05SUG0.09629,048,0436.94E-05SUG0.03128,456,5513.26E-05SUG0.0469,484,8455.91E-05SUG0.081622,345,9145.97E-05SUG0.061313,482,7389.27E-05SUG0.05Mean69,541,0591.02E-06SUG0.1269,541,0597.44E-12GWS0.1269,484,8129.52E-06SUG0.10420,200,6964.02E-09GWS0.071116,539,4582.18E-05SUG0.091326,091,5173.59E-05SUG0.0769,497,5803.3E-05SUG0.091322,724,5765.01E-05SUG0.011116,581,9223.34E-05SUG0.091116,539,4585.11E-05SUG0.091116,581,9543.34E-05SUG0.091326,091,4965.75E-05SUG0.08420,189,4394.1E-05SUG0.09815,277,3968.9E-05SUG0.04420,189,4734.1E-05SUG0.091313,108,8409.02E-05SUG0.08323,840,1655.32E-05SUG0.0969,488,6185.44E-05SUG0.091116,535,5126.09E-05SUG0.0869,484,8456.55E-05SUG0.08737,350,2657.5E-05SUG0.081313,108,8407.69E-05SUG0.08426,945,0518.08E-05SUG0.081116,193,2888.44E-05SUG0.081116,580,4439.06E-05SUG0.08GWS = genome-wide significant (P ≤ 7.28 × 10⁻⁷); SUG = suggestive (P ≤ 10 − 4) Summary of GWAS results for leafing date using the FarmCPU and BLINK models GWS = genome-wide significant (P ≤ 7.28 × 10⁻⁷); SUG = suggestive (P ≤ 10 − 4) Identification of genome-wide significant and suggestive chromosomal positions associated with leafing date Three criteria were used to identify the chromosomal positions with the strongest association to leafing date: (1) chromosomes that showed higher significance levels in Manhattan plots; (2) suggestive associations defined as −log10(P) ≥ 4; and (3) chromosomal positions that not only met the first two criteria but also appeared most frequently across analyses with Blink, FarmCPU, and MLMM models using four sets of phenotypic data (2017, 2018, 2019, and the three-year average). On chr 1, a physical position of 1,427,218 bp with a p-value of 8.70568E-05 was identified by the Blink and FarmCPU models using phenotypic data from 2017, 2018, and the three-year average. On Chr 4, the locus at 20,189,439 bp with a p-value of 9.3734E-19 was identified, a genome-wide significant association (P = 9.37 × 10⁻¹⁹), the strongest signal detected in this study. The association of this chromosomal position with the leafing date was confirmed by all three models (Blink, FarmCPU, and MLMM) using phenotypic data from 2018, 2019, and the three-year average. On Chr 6, physical position at 9,541,059 bp and 9,488,618 bp were identified with p-values of 7.43538E-12 and 9.51713E-06, respectively. The first locus was identified by the Blink, FarmCPU, and MLMM models using the 2018, 2019, and three-year average data, while the second locus was identified by the FarmCPU model using 2019 and three-year average data. On Chr 8, the physical position at 15,615,923 bp with a p-value of 5.48259E-11 was identified by all three models (Blink, FarmCPU, and MLMM) using the 2017 phenotypic data. Finally, on Chr 11, the physical position at 16,581,922 bp with a p-value of 3.33916E-05 was identified by the Blink and FarmCPU models using the 2018 and three-year average data. From this study, a total of six SNP markers (including both genome-wide significant and suggestive loci) were selected (Table 4). To confirm the final selection and efficiency of the identified markers, primers were designed for the above chromosomal positions. Table 4Genome-wide significant and suggestive genomic regions associated with leafing date in walnutPhenotypic dataChrPhysical position (bp)p-valueSignificance categoryModels2017/2018/ Average11,427,2188.70568E-05SUGBlink/ FarmCPU2017/2018/ Average420,189,439*9.3734E-19GWSMLMM/Blink/ FarmCPU2018/2019/ Average69,541,0597.43538E-12GWSMLMM/Blink/ FarmCPU2019/ Average69,488,6189.51713E-06SUGFarmCPU2017815,615,9235.48259E-11GWSMLMM/Blink/ FarmCPU2018/ Average1116,581,9223.33916E-05SUGBlink/ FarmCPUGWS = genome-wide significant (P ≤ 7.28 × 10⁻⁷); SUG = suggestive (P ≤ 10− 4)*Indicates the SNP exhibiting the most significant -log_10(P) peak in the Manhattan plot, confirmed as a candidate marker for leafing date in the Iranian walnut germplasm Genome-wide significant and suggestive genomic regions associated with leafing date in walnut GWS = genome-wide significant (P ≤ 7.28 × 10⁻⁷); SUG = suggestive (P ≤ 10− 4) *Indicates the SNP exhibiting the most significant -log_10(P) peak in the Manhattan plot, confirmed as a candidate marker for leafing date in the Iranian walnut germplasm The distribution of significant SNPs across the walnut genome is visualized in the Manhattan plots (Fig. 4). A highly prominent peak was consistently identified on Chr 4 at position 20,189,439 bp across all three models (MLMM, Blink, and FarmCPU), reaching a maximum significance level of P = 9.37 × 10− 19 (Table 4). This specific locus, annotated in Fig. 4, exhibited the strongest association with the leafing date across multiple years (2017/2018 average). The alignment between these visual signals and the genomic regions listed in Table 4 confirms this position as a robust candidate marker for breeding programs focused on phenological traits in the Iranian walnut germplasm. Validation of Iranian leafing date markers To assess the accuracy of synthesized Iranian leafing date markers, their effectiveness was evaluated on a population of 24 walnut trees. The results demonstrated that all primers produced high-quality bands across the entire population, indicating the successful efficiency of the synthesized primers in amplifying DNA from both early- and late-leafing walnut populations (Fig. S2). Separation of early and late leafing walnut genotypes The ABI Genetic Analyzer was used to distinguish early- and late-leafing walnut genotypes, and the primer JRgwLDch4pos20189439 showed a strong ability to distinguish early- and late- leafing genotypes in the Iranian and Turkish populations within the tested panel (Figs. 6 and 7). This primer corresponds to an SNP located at position 20,189,439 on Chr 4 showing the highest association with the leafing date among all identified markers. Other primers, including the one designed for the chromosomal region identified by Bernard et al. (2020), failed to fully separate the studied population in terms of leafing date. In our validation test on 24 cultivars and genotypes, the allele‑specific ABI electropherograms clearly separated early‑ and late‑leafing individuals based on the presence of A or G nucleotides at the SNP position, resulting in two non‑overlapping allele clusters that largely corresponded to the observed phenotypic groups in the 24‑genotype validation set. The output charts from the ABI genetic analyzer using the SNaPshot method with the Iranian primer JRgwLDch4pos20189439, which corresponds to the genomic region 20,189,439 on chromosome 4 of walnut, were analyzed in the very early-leafing Iranian (1:‘7t’, 2:‘8t’, 3:‘35t’, 4:‘Jamal’, 5:‘Damavand’, 6:‘Alvand’, 7:‘Chaldoran’) and Turkish (8:‘Maras12’, 9:‘KSU12’, 10:‘KSU5’, 11:‘C6S4’, 12:‘B41’) walnut populations (Table S4). All early-leafing cultivars and genotypes showed a green peak with a size between 37–38 nucleotides, indicating the presence of adenine (A) in the SNP region. This green peak, derived from the SNaPshot primers, was common to all early-leafing varieties and genotypes The output charts from the ABI genetic analyzer using the SNaPshot method with the Iranian primer JRgwLDch4pos20189439, which corresponds to the genomic region 20,189,439 on chromosome 4 of walnut, were analyzed in the very late-leafing Iranian (1:‘RDM’, 2:‘Chandler’, 3:‘Pedro’, 4:‘Franquette’, 5:‘Alsozentivani117’, 6:‘Fergean’, 7:‘Fernor’) and Turkish (8:‘M324’, 9:‘M283’, 10:‘M243’, 11:‘Howard’, 12:‘Wck225’) walnut populations. Among the analyzed samples, several cultivars with foreign backgrounds were included as benchmarks, comprising those of US origin (‘RDM’, ‘Chandler’, ‘Pedro’, and ‘Howard’), French origin (‘Franquette’, ‘Ferjean’, and ‘Fernor’), and Hungarian origin (‘Alsószentiváni 117’) (Table S4). According to the images, all late-leafing cultivars and genotypes displayed, in addition to the green peak with a size between 37–38 nucleotides, a blue peak with a size between 35–37 nucleotides. The green and blue peaks, resulting from the SNaPshot primers, indicate the presence of adenine (A) and guanine (G) at the SNP locus, respectively. The presence of guanine (G) was common to all late-leafing cultivars and genotypes Discussion Phenotypic data and genetic diversity Leafing date is a complex phenological trait influenced by a range of physiological and environmental factors, exhibiting significant variation among different cultivars and genotypes [29, 39]. Our walnut collection included 85 individuals, comprising common cultivars from the USA and France, along with Iranian walnut cultivars and genotypes known for their exceptional traits with an extensive genetic diversity. Kinship analysis (Fig. 2) confirmed the presence of substantial diversity within this germplasm. Additionally, LD decayed rapidly, suggesting a large effective population size for this walnut population. Previous studies have already documented the genetic diversity of various Juglans regia L. germplasm [20, 31, 40–43]. A study of a natural J. regia population in the eastern Italian Alps reported moderate genetic diversity [44]. Using the Axiom 700 K SNP array, Arab et al. [20] divided a set of 95 Iranian accessions adapted to diffirent climate zone into four subpopulations according to geographic grouping. Similarly, Bernard et al. [22] genotyped the INRAE walnut germplasm collection with the same SNP array and found that genotypes from Western Europe and America formed one cluster, while those from Eastern Europe and Asia constituted another. However, in our study, French, American, and native walnut genotypes were not clearly distinguished from each other. This lack of clear stratification in our PCA and kinship results likely stems from the limited size of the association panel (85 individuals) and its specific composition. Unlike broader germplasm collections, our panel consists of elite genotypes and selected commercial cultivars that may share common ancestry or have undergone similar selection pressures for horticultural traits. Furthermore, the extensive gene flow historically facilitated by the Silk Road and the subsequent use of foreign cultivars in local breeding programs might have diluted the genetic boundaries between these geographic groups. Therefore, the overlapping clusters in our results reflect a high level of genetic admixture within this specialized breeding panel rather than a lack of overall diversity. In interpreting our GWAS results, we acknowledge that the population size of 85 accessions represents a potential limitation. However, this panel constitutes a curated core collection of Iranian walnut germplasm, selected to maximize genetic diversity while maintaining high phenotypic quality. The wide variation observed for leafing date, together with the inherently high heterozygosity and historical recombination in walnut, provided sufficient power to detect major-effect loci even within a moderate sample size. Notably, the consistent identification of a significant marker on chromosome 4 across three consecutive years supports the robustness of the association signal in this elite germplasm. Nevertheless, we emphasize that future studies using larger and more globally representative populations will be valuable for validating minor-effect loci and further refining the genetic architecture of leafing date. Finalizing the findings on genetic control of leafing date through GWAS For the first time, SNPs associated with leafing date were identified in the Iranian walnut population. SNPs on Chr 1, 4, 6, 8, 11, and 13 were found to be associated with leafing date. Recently, similar studies on leafing date have been conducted on walnut populations from the USA, France, and Türkiye, identifying markers on various chromosomes [21–23, 28]. Marrano et al. (2019b) utilized phenotypic data from the UC Davis walnut germplasm, which included 896 individuals, and genetic data obtained from sequencing. Using both QTL mapping and GWAS methods, they identified markers associated with phenological traits in walnut. By employing the SUPER, FarmCPU, and MLMM models for association mapping, they were the first to identify SNPs related to leafing date, harvest date, and lateral bearing in the U.S. walnut genetic improvement program. They identified signals on Chr 1, 4, 8, 10, 11, 12, 13, 16, and 17 associated with leafing date, with the SNP with physical position 4,650,186 on Chr 1 showing the strongest association (p-value = 2.72E-31), as detected by all three models (SUPER, FarmCPU, and MLMM). Additionally, they identified Chr8, at position 3,137,543, as another locus associated with leafing date through QTL mapping [45]. Bernard et al. [22], in their effort to identify markers related to leafing date in the INRAE walnut germplasm, gathered two years of phenotypic data in 2018 and 2019, then used GWAS and QTL mapping to identify SNPs associated with this trait. They used the MLMM and FarmCPU models and identified signals on Chr1, 2, 4, 6, 7, 8, 10, 15, and 16 related to leafing date, with the physical position 6,514,832 on Chr 1 (p-value = 8.95E-19) showing the highest significance for leafing date. They subsequently developed a KASP marker for this region and validated it using a population of 90 early- and late-leafing walnuts at the UC Davis, confirming the marker’s association with leafing date in the French walnut population [22]. In addition, Kefayati et al. [28] studied a walnut population derived from a cross between Chandler and Kaplan-86 in Türkiye, conducted quantitative trait locus (QTL) mapping and identified markers associated with leafing date on LG4, which explained 52% to 68.8% of phenotypic variance. They then introduced the SSR markers JRHR209732 and CUJRB012 as molecular markers for leafing date in the Turkish walnut population [28]. In another study on the Turkish walnut population, Bukucu et al. [23] used a germplasm of 188 walnuts and three years of phenotypic data from 2016, 2017, and 2018. They performed GWAS analyses using the MLM model, revealing markers associated with leafing date on LG1, 4, and 16. Among them, the location 28,418,795 on LG4, with a p-value = 2.71E-12, showed the strongest association with leafing date in the Turkish walnut population [23]. Considering the research conducted on leafing date in walnut germplasm from the USA, France, and Türkiye, each study introduced markers associated with leafing date, with some variation in the identified genomic regions. While there are similarities in some cases, the differences in chromosomal loci associated with leafing date could be due to the use of different populations with distinct genetic structures. In addition to population-specific genetic backgrounds, factors such as panel size, marker density, LD structure, and differences in phenotyping protocols may also contribute to the observed variation in associated loci. Taken together, these findings support the interpretation that leafing date is influenced by multiple alleles distributed across the genome, with probable major loci located on Chr 1 and 4. However, the complementary explanations mentioned above should also be considered when comparing results across studies [46–48]. Researchers have successfully extended the use of the GBS-GWAS approach to uncover marker-trait relationships in other fruit trees [49–51]. GWAS analyses on the UC Davis almond population revealed SNPs on Chr 2, 5, 7, and 8 associated with traits of shell thickness, shell weight, crack-out percentage, and log maximum force. Five SNPs were selected for KASP assays, and the marker designed for SP02_23368051 showed an association with shell hardness [49]. In a study of the French apple population, GWAS model GEMMA and MLMM analyses revealed the strongest association for the trait of leafing date on chr 9 (SCh09_680633). Additionally, they identified a strong association between a peroxidase gene, MdPRX10, and leafing date [50]. Another study aimed to decipher the genetic control of flowering date in a Moroccan olive population. GWAS analyses revealed three genetic loci on chr 1 and 4 associated with marker-trait associations. Additionally, their results suggested a polygenic architecture underlying flowering date in olive [51]. Despite the location of the validated SNP within an uncharacterized protein, its functional significance likely stems from its genomic context and linkage with neighboring functional loci within the same Linkage Disequilibrium (LD) block. In deciduous trees like walnut, leafing date is a complex trait governed by a network of genes involved in chilling requirement, circadian clock regulation, and phytohormone signaling (particularly gibberellins and abscisic acid) [52, 53]. The presence of a significant SNP in this region suggests that it may be linked to nearby regulatory elements or cis-acting factors that modulate the expression of genes involved in meristem activity or seasonal dormancy release. Given that many uncharacterized genomic regions in Juglans regia have been recently associated with stress responses and phenological transitions, this SNP could potentially act as a marker for a larger regulatory hub controlling the transition from endodormancy to leafing date [21, 26]. While the specific function of this protein remains to be elucidated, its genomic proximity to pathways regulating the dormancy-growth cycle provides a plausible biological framework for its association with leafing date, warranting further functional studies to pinpoint the exact causal mechanism [54]. A SNaPshot marker related to leafing date is released for the first time in Iranian walnut population SNaPshot is a primer extension-based mini-sequencing method developed for the analysis of SNPs [38]. The separation of these primers is based on the color and size of the peaks generated by the ABI Genetic Analyzer. The presence of each color blue, black, green, or red indicates the existence of a specific base at the SNP site. Specifically, blue represents guanine (G), black represents cytosine (C), green represents adenine (A), and red indicates thymine (T) at the SNP location [55]. Based on the results obtained and the identification of bases corresponding to the peak charts from early-and late-leafing population, it was determined that early-leafing cultivars and genotypes at the SNP region on Chr 4, physical position 20,189,439, possess only the adenine (A) base. In contrast, late-leafing cultivars and genotypes at the SNP region have both adenine (A) and guanine (G) bases. The presence of guanine (G) in all late-leafing cultivars and genotypes, and its absence in early-leafing ones, indicates an association between the guanine base and late leafing date at the 20,189,439 physical position (Fig. 7). Most late-leafing cultivars and genotypes displayed two peaks at the SNP region, indicating heterozygosity (G/A). Some other late-leafing cultivars and genotypes showed a single peak with the G/G base at the SNP region, indicating homozygosity. In our validation panel, the G allele was consistently associated with late leafing, although larger datasets will be needed to confirm its reliability across broader germplasm. Although the complex nature of walnut phenology implies that other minor-effect loci also contribute to the trait, the S4-20189439 (G) allele provides a robust diagnostic tool for identifying late-leafing individuals within Iranian and Turkish germplasm, distinguishing them from the predominantly ‘A’ carrying early-leafing population. The SNaPshot marker developed in this study offers a significant advantage over existing markers. Given that our candidate locus was identified within a diverse Iranian germplasm, this marker may have potential for broader applications, although further testing in larger and more diverse populations is required. Furthermore, its validation in the Iranian and Turkish samples suggests promising predictive value, but the limited size of the validation panel (24 genotypes) means that additional evaluation is necessary before generalizing its performance. However, given the complex genetic foundation of phenological traits in perennials [12], additional markers, particularly those associated with minor loci, will be required to enhance selection accuracy. Finally, this SNP is situated within a gene that encodes an uncharacterized protein, and it would be intriguing to learn more about the functional role of this gene [22]. While the marker can predict a substantial portion of the phenotype, the regulatory networks associated with this complex trait remain unknown. While previous studies, such as Bernard et al., identified leafing date markers on Chromosome 1, our study revealed a distinct locus (S1-1427218) in the Iranian germplasm. The lack of validation for previously reported markers in our population highlights the unique genetic architecture of Iranian walnuts. Furthermore, our most stable locus, S4-20189439, was consistently detected by MLMM, BLINK, and FarmCPU models across consecutive years. To broaden the utility of these markers, we propose that the markers developed in this study be evaluated in European and American populations to determine their effectiveness across different genetic backgrounds. Validating the identified markers across European and American germplasm represents an important next step. Such broader evaluation would help confirm their transferability and strengthen their utility for international walnut breeding programs. Furthermore, as observed with the marker on chromosome 1 reported by Bernard et al.—which appeared ineffective in our collection—the chromosome 4 marker identified here may also require validation in more diverse genetic backgrounds. Acknowledging these potential limitations is crucial, as the performance of molecular markers can vary significantly depending on the specific genetic background and population structure. Conclusions The genetic diversity of walnut in Iran is remarkably high. Harnessing this diversity, together with identifying markers linked to key breeding traits, is an efficient strategy for developing new cultivars with desirable characteristics. In this study, GWAS was used to identify and validate markers associated with leafing date in the Iranian walnut population. Genomic regions on chromosomes 1, 4, 6, 8, 11, and 13 were found to be strongly associated with leafing date. The design and development of SNaPshot primers demonstrated that the JRgwLDch4pos20189439 primer, corresponding to position 20,189,439 on chromosome 4, effectively differentiated early- and late-leafing genotypes among the validated samples. As breeding programs in Iran prioritize late leafing together with desirable nut and kernel traits, this primer could provide a valuable tool for molecular breeding aimed at developing walnut cultivars with improved resilience to late-spring frost. Comparisons with previous studies conducted in the USA, France, and Türkiye show that distinct SNPs associated with leafing date have been identified in each country’s walnut populations. This indicates that the genomic regions underlying leafing date may differ among populations with diverse genetic structures. Consequently, leafing date appears to be a multi-allelic trait, frequently represented on chromosomes 1 and 4 in walnut. Further studies using broader walnut populations, or using larger and more genetically diverse materials, are essential to fully provide deeper insights into the molecular mechanisms governing leafing date. Future research should focus on validating the identified chromosome 4 marker across diverse European and American germplasm. This will confirm its transferability and ensure its reliability for international marker-assisted selection (MAS) in walnut breeding. Supplementary Information Supplementary Material 1: Fig. S1a-c. Diagram of maximum, minimum and average temperature (a), annual cloud and humidity (b) and rainfall and rainy days (c) of Karaj in 2017, 2018 and 2019. Fig. S2. Gel electrophoresis of PCR products from the SNaPshot primer test for leafing date traits in populations of early- and late-leafing Iranian and Turkish walnut cultivars and genotypes. Supplementary Material 2: Table S1. Phenotypic data of 85 Iranian walnut cultivars/genotypes used for GWAS analysis. Supplementary Material 3: Table S2. SNPs used for primer design and 300-bp upstream and downstream sequences flanking the SNP regions based on the Chandler v2 genome. Supplementary Material 4: Table S3. Characteristics of SBE-SNP primers designed for genomic regions identified through GWAS analysis of 85 Iranian walnut individuals. Supplementary Material 5: Table S4. 24 Early-and late-leafing cultivars and genotypes used for validation of the Iranian walnut leafing date markers. Supplementary Material 6. Acknowledgements We would like to thank the Genome Center at the University of California, Davis, and Dr. Patrick James Brown, for their assistance in sequencing the Iranian walnut population. We also would like to thank Iran National Science Foundation (INSF), Center of Excellence for Walnut Improvement and Technology of Iran, and the University of Tehran and Cukurova University for their supports. Authors’ contributions MF, MMA, DH and KVdesigned the research, MF, MMA, HK, DH and SK collected relevant literature data and conducted the data analysis; MF wrote the manuscript, MF, MMA, SK and KV discussed and interpreted the data. MF, MMA, HK, SS, MR, SK and KV revised the manuscript. All authors read and approved the manuscript. Funding The authors declare that no direct funds, grants, or monetary support were received during the preparation of this manuscript. However, the Iran National Science Foundation (INSF), the Center of Excellence for Walnut Improvement and Technology of Iran, the University of Tehran, and Cukurova University provided institutional and in-kind support by facilitating and covering the costs of specific parts of the study, including laboratory materials. In addition, sequencing costs for the Iranian walnut population were supported through the Genome Center at the University of California, Davis. Data availability The datasets generated and/or analysed during the current study are available in the Genome Variation Map (GVM) [56, 57] repository, under accession number GVM001270 ( [https://bigd.big.ac.cn/gvm/getProjectDetail? Project=GVM001270](https:/bigd.big.ac.cn/gvm/getProjectDetail? Project=GVM001270) ) Declarations Ethics approval and consent to participate This plant materials don’t include any species at risk of extinction. We declare that all the experimental plants were collected with permission from local authorities of agricultural department and the plant materials in the experiment. We comply with relevant institutional, national, and international guidelines and legislation for plant study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Mehdi Fallah, Email: fallah.mehdi@ut.ac.ir. Kourosh Vahdati, Email: kvahdati@ut.ac.ir. References Associated Data Supplementary Materials Supplementary Material 1: Fig. S1a-c. Diagram of maximum, minimum and average temperature (a), annual cloud and humidity (b) and rainfall and rainy days (c) of Karaj in 2017, 2018 and 2019. Fig. S2. Gel electrophoresis of PCR products from the SNaPshot primer test for leafing date traits in populations of early- and late-leafing Iranian and Turkish walnut cultivars and genotypes. Supplementary Material 2: Table S1. Phenotypic data of 85 Iranian walnut cultivars/genotypes used for GWAS analysis. Supplementary Material 3: Table S2. SNPs used for primer design and 300-bp upstream and downstream sequences flanking the SNP regions based on the Chandler v2 genome. Supplementary Material 4: Table S3. Characteristics of SBE-SNP primers designed for genomic regions identified through GWAS analysis of 85 Iranian walnut individuals. Supplementary Material 5: Table S4. 24 Early-and late-leafing cultivars and genotypes used for validation of the Iranian walnut leafing date markers. Supplementary Material 6. Data Availability Statement The datasets generated and/or analysed during the current study are available in the Genome Variation Map (GVM) [56, 57] repository, under accession number GVM001270 ( [https://bigd.big.ac.cn/gvm/getProjectDetail? Project=GVM001270](https:/bigd.big.ac.cn/gvm/getProjectDetail? Project=GVM001270) )