Genetic mapping of a chickpea (Cicer arietinum L.) diversity panel for mineral biofortification towards human nutrition

Abstract

Chickpea (Cicer arietinum L.) is a cool‐season legume with low fats and high concentrations of protein, carbohydrates, and minerals. As such, chickpea can contribute to alleviate hidden hunger due to mineral deficiencies and malnutrition, particularly in Asian and African countries. A chickpea germplasm panel composed of 256 accessions with both desi and kabuli types was evaluated for calcium (Ca), potassium (K), magnesium (Mg), phosphorus (P), copper (Cu), iron (Fe), manganese (Mn), selenium (Se), and zinc (Zn) concentrations. High mean concentrations were found for Ca (181.2 mg/100 g), K (971.9 mg/100 g), Mg (135.2 mg/100 g), P (383 mg/100 g), Cu (0.73 mg/100 g), Fe (5.3 mg/100 g), Mn (4.1 mg/100 g), Se (0.01 mg/100 g), and Zn (2.41 mg/100 g), with a broad concentration range and high percent recommended daily allowance noted for each element. Significant positive correlations were found among all minerals except for a negative correlation between Ca and K. Genomic association mapping detected a total of 14 single‐nucleotide polymorphisms (SNPs) for Ca, Mg, P, Mn, and Zn. The candidate genes associated with these SNPs are pivotal to physiological pathways for plant stress response, health, and ionic homeostasis. Admixture analysis found nine diverse subpopulations in the panel based on ancestral diversity. Principal component analysis revealed population structure consistent with admixed ancestral subpopulations, chickpea types, and country of origin. Together, these findings show chickpea mineral biofortification can be achieved using genomic techniques and classical approaches to breed for high mineral concentrations to improve human health and plant defense.

Core Ideas

• Chickpea has high concentrations of macrominerals (Ca, K, Mg, and P) and microminerals (Cu, Fe, Mn, Se, and Zn), showing positive correlations with each other.

• Significant single‐nucleotide polymorphisms identified for Ca, Mg, P, Mn, and Zn indicate genes regulating several functions in plants particularly governed by the presence/involvement of these minerals.

• Genetic diversity of accessions is revealed by population structure studies.

Plain Language Summary

Minerals are vital in human nutrition, regulating the structural and physiological systems in the body. Mineral malnutrition has been an alarming issue over the globe, which can be prevented by biofortifying staples for minerals. Chickpea is widely used and also a staple in several parts of the world. This study aimed at evaluating a chickpea germplasm accession panel for important minerals (Ca, K, Mg, P, Cu, Fe, Mn, Se, and Zn) and identify single‐nucleotide polymorphisms (SNPs) for minerals. The analysis revealed high concentrations, percent recommended daily allowance, a broad range, and significant positive correlations for most minerals. Genome‐wide study indicated significant SNPs for Ca, Mg, P, Mn, and Zn, associated with several plant regulatory genes. Population structure studies uncovered the genetic diversity for the candidate accessions in this panel. Notably, several superior accessions were identified with high concentrations of two or more minerals, which can be suitable candidate parents for mineral biofortification breeding programs.

Abbreviations

ANOVA

analysis of variance

BLINK

Bayesian‐information and linkage‐disequilibrium iteratively nested keyway

BLUP

best linear unbiased predictor

GAPIT

genome association and prediction integrated tool

GWAS

genome‐wide association study

LD

linkage disequilibrium

MAF

minor allele frequency

PCA

principal component analysis

PVE

phenotypic variance explained

QQ

quantile–quantile

RDA

recommended daily allowance

SNP

single‐nucleotide polymorphism

1. INTRODUCTION

Plant and animal food sources are composed of macronutrients such as carbohydrates, proteins, fats, and fiber as well as micronutrients including minerals and vitamins. Macronutrients meet human energy requirements for sustaining structural and functional systems (Venn, 2020). While micronutrients (minerals and vitamins) are trace elements that are required in lesser concentrations (Ritchie & Roser, 2018) and have several indispensable roles in efficient growth, development, and physiological functions, staple foods, such as wheat, rice, corn, tubers, and so forth, provide sufficient macronutrients; however, inadequate micronutrient availability, or hidden hunger, is an ongoing global concern and is estimated to affect one‐third of the global population (HLPE, 2017). Notably, some populations in Asia, Africa, and Latin America are severely deficient in calcium (Ca), magnesium (Mg), iron (Fe), and zinc (Zn) (Bueckert et al., 2011).

Fe, Zn, vitamin A, and folate deficiencies have been reported in 50% of children aged 3–5 years and 67% of all reproductive‐age women globally (World Health Organization, 2023). Micronutrient mineral deficiencies can lead to severe diseases or disorders that impair physical and mental health. Inadequate Zn results in a suppressed immune system, insufficient iodine (I) leads to poor brain development in children, and limited Fe causes anemia, especially in young children and pregnant women (World Health Organization, 2023). Low levels of Ca in the diet negatively impact bone strength and lead to stunting, while a diet low in Mg can increase weakness and fatigue, result in loss of appetite, and be associated with attention‐deficit/hyperactivity disorder (ADHD) in children (Hemamy et al., 2021).

Mineral deficiencies can be avoided by biofortifying common staple foods with micronutrient minerals using breeding approaches. Popular staples include pulse crops, which are legume grains that are consumed in the form of fresh or dried seeds. These are inexpensive daily diets for African and Asian populations, yet Americans and Europeans seldom include pulses in their diets (OECD‐FAO, 2023). Chickpeas (Cicer arietinum L.) are widely consumed either as dry, cooked seeds or processed in various forms to make hummus, snacks, and so forth. Chickpea seeds are nutritionally dense in carbohydrates (50%–58%), protein (18%–22%), and fats (3.8%–10%) and have significant concentrations of important minerals such as Fe, Zn, Ca, Mg, and so forth (Jukanti et al., 2012; N. Wang et al., 2010). Recently, chickpeas have gained a niche in the Western protein market with a range of products, including pasta, bread, spreads, protein powder, beverages, and so forth, due to increased awareness of sustainable health goals and preferences for vegetarian/vegan diets (Salaria et al., 2023). This highlights the importance of biofortification to improve the nutritional profile of chickpea, including mineral concentrations, using breeding approaches (Salaria et al., 2023).

Traditional chickpea mineral biofortification approaches have been popular in breeding programs (Dhaliwal et al., 2021; Mehboob et al., 2022; Thavarajah & Thavarajah, 2012; Vandemark et al., 2018). While the role of genomic approaches has not been emphasized until a few recent studies (Diapari et al., 2014; Fayaz et al., 2022; Roorkiwal et al., 2022; Srungarapu et al., 2022; Upadhyaya et al., 2016). Advancements in genomics technology have played a critical role in mineral biofortification in nutritional breeding programs for cereals (Kotla et al., 2019) and millets (Sharma et al., 2022; Singhal et al., 2021). Recognizing this role, the present study was conducted to evaluate a chickpea germplasm panel for mineral concentrations, along with the use of genetic data to provide insights into mineral biofortification in chickpea. The study hypothesis states that a vast genetic diversity may be present in the chickpea germplasm panel for minerals and significant marker‐trait associations may be detected. The study objectives, thus, aimed to evaluate a chickpea germplasm panel for macromineral (Ca, potassium [K], phosphorus [P], and Mg) and micromineral (copper [Cu], Fe, manganese [Mn], selenium [Se], and Zn) concentrations, identify significant single‐nucleotide polymorphisms (SNPs) associated with candidate genes, and study population structure using phenotyping and genomic association mapping approaches.

2. MATERIALS AND METHODS

2.1. Composition of plant material

A chickpea germplasm panel with 256 accessions was assembled and included both desi (91) and kabuli (165) types (Salaria et al., 2023). The accessions were diverse in origin and were collected from different continents, including Asia (199), North America (55), Europe (1), and South America (1) (Table S1). The panel aimed to represent a comprehensive chickpea genetic diversity by including desi and kabuli types where desi types could be utilized to explore favorable traits such as hardiness, flavor, quality, earliness, adaptivity, and so forth. This information can be deployed into the future chickpea breeding programs in the United States, which are mostly focused on kabuli types.

2.2. Experimental conditions

The experimental trials were conducted in 2020 at the Washington State University Spillman Agronomy Farm (Pullman, WA; 46.73° N, 117.18° W). The soil at the experimental farm has characteristics of Mollisols (Palouse series, fine‐silty, mixed superactive, mesic, Pachic Ultic Haploxerolls). The seeds were treated before planting using a chemical formulation with fungicides fludioxonil (0.56 g/kg), mefenoxam (0.38 g/kg), and thiabendazole (1.87 g/kg) (Syngenta, Greensboro, NC), along with thiamethoxam (0.66 mL/kg; Syngenta, Greensboro, NC) for insect control and molybdenum (0.35 g/kg) as a micronutrient.

Core Ideas

• Chickpea has high concentrations of macrominerals (Ca, K, Mg, and P) and microminerals (Cu, Fe, Mn, Se, and Zn), showing positive correlations with each other.

• Significant single‐nucleotide polymorphisms identified for Ca, Mg, P, Mn, and Zn indicate genes regulating several functions in plants particularly governed by the presence/involvement of these minerals.

• Genetic diversity of accessions is revealed by population structure studies.

The two experiments, referred to as DTST1 and DTST2, were planted on April 30, 2020, and May 29, 2020, respectively. The experiments were conducted using an α‐lattice design with three replicates/entries. The mechanical planting was done in four‐row plots of 1.5 m length and 20 cm row‐to‐row spacing using a seed rate of 25 seeds/plot. A single application of post‐plant/pre‐emergence herbicide metribuzin (0.42 kg/ha; Bayer Crop Science) and linuron (1.34 kg/ha; NovaSource) was used to suppress or control weeds. Plots were mechanically harvested between September 23, 2020, and October 21, 2020. Seed samples from each plot were cleaned and provided to the Pulse Quality and Nutrition Laboratory at Clemson University (Clemson, SC) for mineral analysis.

2.3. Mineral extraction

The extraction of minerals was performed by using a modified method described by Thavarajah et al. (2007). Seeds of each plot were ground into fine powder and stored at 4°C. Analysis‐ready samples were prepared by digesting 250 mg ± 1% of seed powder overnight with 4 mL concentrated nitric acid (70% HNO₃). Samples were then placed in a hot water bath at 90°C for 2 h, followed by the addition of 4 mL of hydrochloric acid (38% HCl) and an additional 1 h in the hot water bath. The sample solution was filtered using Whatman paper (20–25 µm) and diluted with deionized nanopure water to a final volume of 10 mL. Concentrations of Ca, P, K, Mg, Cu, Mn, Fe, Zn, and Se were determined using inductively coupled plasma emission spectrometry (ICP‐6500 Duo, Thermo Fisher Scientific). A standard stock solution of macrominerals (10,000 ppm) and microminerals (1000 ppm) was used for serial dilutions to make standard calibration curves with concentration ranges from 5 to 400 ppm for macrominerals and 0.05 to 10 ppm for microminerals. Standard references for lentil (CDC Redberry) and peach (NIST 1547) were used for validation of calibration measurements. A set of random samples was analyzed for moisture content in each replicate and averaged to estimate replication‐specific moisture content.

2.4. Phenotypic data analysis

Phenotypic data were collected for minerals to determine means, ranges, correlation coefficients, and frequency distributions using JMP Pro 16.2.0 (SAS Institute Inc., 2021). Cumulative means were calculated by averaging across replicates and then experiments. Frequency distribution histograms with box plots, standard error bars, and normality density curves were generated using JMP software. The goodness of fit for normality was tested with the Shapiro–Wilk test, and a non‐normal distribution of minerals was found. The data were subjected to square root, logarithmic, and rank scale for transformation. The results obtained from the transformed data were very similar to the outcomes of the original data, so the original dataset was ultimately used to report further results. Fe concentration data for the chickpea accessions showed several outliers constricting the whole dataset and therefore were excluded from downstream analysis.

Scatterplots were generated to visualize Pearson's correlation coefficients estimated with a restricted maximum likelihood approach among different minerals. Density plots and regression lines were fit on scatterplots with 95% confidence intervals. Percent recommended daily allowance (%RDA) was calculated for each mineral using an average recommended daily allowance of 1000 mg/day for Ca, 410 mg/day for Mg, 3000 mg/day for K, 700 mg/day for P, 13 mg/day for Fe, 0.9 mg/day for Cu, 9.5 mg/day for Zn, 2.05 mg/day for Mn, and 0.06 mg/day for Se for adult males and females 19–50 years of age as a reference according to National Institute of Health (NIH) survey reports by the U.S. Department of Health and Human Services (https://www.ncbi.nlm.nih.gov/books/NBK545442/table/appJ_tab3/?report=objectonly). The formula used for %RDA calculation was as follows:

$$\begin{array}{ccc}\hfill \%\mathrm{RDA}& =& \left(x/\mathrm{average}\mathrm{RDA}\mathrm{for}\mathrm{a}\mathrm{particular}\mathrm{mineral}\right.\hfill \\ & & \left.\mathrm{by}\mathrm{NIH}\mathrm{survey}\mathrm{reports}\right)\times 100,\hfill \end{array}$$

where x = lowest and highest concentrations of a particular mineral, with ranges given in Table 1.

TABLE 1.

Range, mean, percent recommended daily allowance (%RDA) (recommended daily allowance), and broad‐sense heritability of chickpea minerals.

Mineral	Range (mg/100 g)	Mean (mg/100 g) a	%RDA b	Percent broad‐sense heritability (H 2)
Ca	86.08–279.49	181.21 ± 3.16	8.61–27.95	69.6
K	832.66–1287.09	971.99 ± 4.07	27.76–42.90	9.6
Mg	112.80–166.32	135.17 ± 0.58	27.51–40.37	11.9
P	316.78–478.24	383.01 ± 1.66	45.25–68.32	8.2
Cu	0.51–1.24	0.73 ± 0.01	56.67–137.78	5.8
Fe	3.83–8.78	5.32 ± 0.05	29.46–67.54	0
Mn	2.75–6.06	4.05 ± 0.04	134.15–295.61	24.3
Se	0–0.06	0.01 ± 0.00	0–100	0
Zn	1.71–3.15	2.41 ± 0.02	18–33.16	16.4

^a Mean concentration of chickpea minerals for replicates over experiments ± standard error of mean.

^b %RDA calculated using average recommended daily allowance of 1000 mg/day for calcium (Ca), 410 mg/day for magnesium (Mg), 3000 mg/day for potassium (K),700 mg/day for phosphorus (P), 13 mg/day for iron (Fe), 0.9 mg/day for copper (Cu), 9.5 mg/day for zinc (Zn), 2.05 mg/day for manganese (Mn), and 0.06 mg/day for selenium (Se) for adult males and females within age group of 19–50 years according to National Institute of Health (NIH) survey reports by U.S. Department of Health and Human Services (https://www.ncbi.nlm.nih.gov/books/NBK545442/table/appJ_tab3/?report=objectonly).

Analysis of variance (ANOVA) was conducted for the α‐lattice design to find the effects of genotype, replication, experiment, sample batch processed, and their interaction in R version 4.0.5. Batch effects were used to calculate best linear unbiased predictors (BLUPs) with Bayesian inference using the rstanarm package version 2.21.3 (Stan Development Team, 2021) in R according to the following model:

$$\mathrm{Mineral}\sim (1|\mathrm{GenotypeNum})+(1|\mathrm{Batch})+(1|\mathrm{Rep})+(1|\mathrm{Exp})$$

The downstream analysis was conducted using BLUPs to account for batch‐to‐batch variation (batch effects) due to processing with an analysis instrument. Broad‐sense heritability was calculated for each mineral by using the variance components estimated in ANOVA as per the formula given below:

$$\begin{array}{ccc}\hfill \mathrm{Heritability}\left(h^2\right)& =& {{\mathrm{\sigma }}^2}_{\mathrm{GenotypeNum}}/\left({{\mathrm{\sigma }}^2}_{\mathrm{GenotypeNum}}+{{\mathrm{\sigma }}^2}_{\mathrm{Batch}}\right.\hfill \\ & & \left.+{{\mathrm{\sigma }}^2}_{\mathrm{Rep}}+{{\mathrm{\sigma }}^2}_{\mathrm{Exp}}+{{\mathrm{\sigma }}^2}_{\mathrm{Residual}}\right)\hfill \end{array}$$

2.5. Population structure analysis

The genetic structure of the chickpea germplasm panel was found using SNP data to conduct admixture and principal component analysis (PCA). PLINK files were generated from VCF files to perform admixture analysis using ADMIXTURE version 13.0 (Alexander et al., 2009). The admixture analysis was performed for multiple K‐values (K = 3 to K = 10) with fivefold cross‐validation. The lowest cross‐validation error was found for K = 9, showing the expected number of subpopulations. The Q‐matrix for K = 9 contained ancestry coefficients for accessions in different subpopulations. The Q‐matrix was used to generate admixture plots using ggplot2 (Wickham, 2016). Genome association and prediction integrated tool (GAPIT) generated the principal components (PCs) for the chickpea germplasm panel. The first two PCs (PC1 and PC2) were plotted according to the admixture subpopulations, country of origin, and chickpea types using ggplot2 (Wickham, 2016).

2.6. Genome‐wide association study

Plant leaf tissue for a total of 253 accessions was harvested from healthy seedlings planted in the greenhouse to isolate DNA. Three accessions, CDC‐Frontier, Royal, and Spanish White, were not sequenced. Extracted DNA was used to detect SNPs by following the genotyping by sequencing pipeline employed by LGC Biosearch Technologies (https://www.biosearchtech.com/). Paired‐end reads (150 bp) were aligned to the chickpea reference genome of CDC Frontier (Cicer arietinum v1.0) (Varshney et al., 2013) using the Burrow–Wheeler aligner (Li & Durbin, 2010). The Genome Analysis Toolkit (Van der Auwera & O'Connor, 2020) (https://gatk.broadinstitute.org/) was used for SNP calling. SNP filtering (<20% missing data and >5% minor allele frequency [MAF]) was performed using VCFtools (Danecek et al., 2011) for quality control, resulting in 15,927 high‐quality SNPs. After filtering, missing genotypes were imputed using Beagle version 5.4 (Browning et al., 2018).

Finally, the VCF file was converted to HapMap format in Tassel software version 5.0 (Bradbury et al., 2007). A genome‐wide association study (GWAS) was conducted with the GAPIT version 3 (J. Wang & Zhang, 2021) package using the Bayesian‐information and linkage‐disequilibrium iteratively nested keyway (BLINK) model in R. Due to limitations in population size on the power of the GWAS, a minor allele count of 20 was used to filter alleles. This translates to a more stringent MAF of 0.078125 and resulted in 10,602 quality SNPs remaining for GWAS. The top five PCs were used to control population structure in association mapping using the BLINK model, while in MLM, both kinship and PCs were used. GAPIT generated Manhattan plots and quantile–quantile (QQ) plots for all minerals. Significant SNPs were indicated by using the Bonferroni threshold (p‐value < 0.05/10,602). GWAS was also conducted by using different approaches including for each experiment separately using GAPIT and GEMMA (Zhou & Stephens, 2012), combined GWAS using GEMMA, meta‐GWAS using METAL software (Willer et al., 2010), and multivariate analysis using GAPIT. Population structure was controlled in experiment‐wise GWAS with GAPIT using PCs (PC = 5), with GEMMA by using the kinship matrix, and in multivariate analysis by using both PCs (PC = 5) and the kinship matrix. The size of linkage disequilibrium (LD) blocks containing significant SNPs was determined by computing LD estimates using PLINK (Purcell et al., 2007) software (Table S2). Jbrowse (Skinner et al., 2009) was used to find candidate genes in local LD with significant SNPs. The combined GWAS using GAPIT has been discussed in detail in the manuscript.

3. RESULTS

3.1. Phenotypic variation of minerals in the chickpea germplasm panel

The chickpea germplasm panel demonstrated large phenotypic variation for both macrominerals (Ca, Mg, K, and P) and microminerals (Fe, Cu, Zn, Mn, and Se). Histograms indicated the spread of various minerals in the panel, and box plots show the mean values and possible outliers for each mineral (Figure 1). The fitted normal distribution curves were tested for goodness of fit and indicated a non‐normal distribution for all minerals except Mg, P, and Zn. Descriptive statistics for chickpea mineral phenotypic data are shown in Table 1. Broad concentration ranges were evident for both macrominerals (Ca, Mg, K, and P) and microminerals (Fe, Cu, Zn, Mn, and Se). Maximum %RDA values were <50% for Ca, K, and Mg, in the mid‐range for P and Fe, and very high for Se, Mn, and Cu (Table 1). Chickpea accessions with the highest mean concentration of macrominerals Ca, K, Mg, and P were W6 25977, CA15940141C, W6 26045, and W6 25897, respectively. The concentrations of microminerals Cu, Mn, Se, and Zn were highest in accessions W6 26044, PI 239859, W6 25879, and W6 25897, respectively.

FIGURE 1.

Histograms of chickpea accessions with mean values for minerals (mg/100 g of seeds). In the boxplots, the position of mean values is indicated by a rhombus (◊), the median is represented by a small line in the box, and possible outliers in the chickpea panel are shown as points. The green lines indicate the normal curves fitted based on mean values for chickpea minerals, and goodness of fit was tested using the Shapiro–Wilk test of normality. Standard error bars are also shown.

ANOVA showed significant genotypic effects for all minerals (p < 0.001) except Fe and Se (Table 2). The experiment effects and interaction effects (genotype × experiment) vary for different minerals, while batch effects were significant for all minerals except Fe. Broad‐sense heritability estimates indicated high heritability for Ca (69.6%), while for other minerals, heritability was low, ranging from 0 to >25% (Table 1). Pearson's pairwise correlation coefficients estimated between different minerals (Figure 2) indicate significant positive correlations for Ca and P (0.15; p < 0.05) as well as Cu and Se (0.20) and K and Mg (0.20) (p < 0.01). Highly significant positive correlations (p < 0.001) were evident between several minerals (Figure 2). Only Ca and K (−0.42; p < 0.001) demonstrated a significant negative correlation.

TABLE 2.

Analysis of variance (ANOVA) for chickpea minerals.

Source of variation
Minerals	Genotypes (Gen) df = 255	Experiment (Exp) df = 1	Batch (Exp, Rep) df = 6	Gen × Exp df = 255
Ca	***	**	***	**
K	***	NS	***	NS
Mg	***	***	***	NS
P	***	NS	***	**
Cu	***	**	***	NS
Fe	NS	NS	NS	NS
Mn	***	NS	***	***
Se	NS	NS	**	**
Zn	***	NS	***	***

***, **, * denote statistical significance at p < 0.001, p < 0.01, p < 0.05, respectively. NS, not significant.

FIGURE 2.

Correlation analysis of chickpea minerals. Ca, calcium; Cu, copper; K, potassium; Mg, magnesium; Mn, manganese; P, phosphorus; Se, selenium; Zn, zinc. Significance codes: ***p < 0.001, **p < 0.01, *p < 0.05.

3.2. Population structure analysis

The results of the admixture analysis indicate nine diverse subpopulations in the chickpea germplasm panel with a cross‐validation error of 0.58816 (Figure 3a). An earlier study on this panel reported seven subpopulations instead of nine, and this could probably be due to the variable number of SNPs used for admixture analysis in both studies (Salaria et al., 2023).

FIGURE 3.

(a) Admixture analysis of chickpea germplasm panel (n = 253) for K = 9 where the x‐axis indicates the number of individuals in different colors corresponding to their proportion of ancestry shown as the y‐axis and (b) admixture coefficients observed in chickpea germplasm accessions categorized in nine subpopulations, where more orange indicates higher admixture and more blue indicates minimal admixture.

As described in the previous study, the panel mainly had accessions that originated in India (78.66%), followed by the United States (21.74%), with <1 % accessions from other countries such as Canada, Iran, Pakistan, and Peru (Table S1). This influenced the composition of subpopulations in terms of the country of origin. Admixture Subpopulations 3 (green; n = 50) and 4 (purple; n = 12) contained all the accessions from India. Subpopulation 2 (blue; n = 23) was composed entirely of accessions of US origin. The largest proportion of accessions (>70%) in admixture Subpopulations 7 (brown; 9 of 12), 8 (pink; 32 of 33), and 9 (grey; 37 of 46) were of Indian origin. Subpopulation 1 (red; n = 44) includes 27 accessions from India (61%), 16 from the United States (36 %), and only one from Peru. The accessions in Subpopulation 5 (orange; n = 20) were mostly from the United States (10), followed by India (6), Iran (3), and Canada (1). Subpopulation 6 (yellow; n = 13) was comprised of accessions from Iran (3) and the United States (10). Accessions from Iran were also found in Subpopulations 7 (3), 8 (1), and 9 (4). The remaining accessions from the United States (10) were found in Subpopulation 9. Accessions from Canada (1) and Pakistan (1) were found in Subpopulations 5 and 9, respectively. Subpopulation 3 (n = 50) was the largest, and Subpopulations 4 and 7 were the smallest (n = 12). Subpopulation 9 was the most diverse admixture group.

Admixture coefficients for accessions show their composition either as pure or admixed types from other subpopulations. The admixture coefficient varied from 0 to 1, where 0 indicated an admixture from several subpopulations, and 1 indicated a pure type with no contribution from other subpopulations (Figure 3b). Subpopulation 7 has the highest number of admixed individuals, followed by Subpopulations 3 and 1. Subpopulation 6 had no accessions with low admixture coefficients. In total, 135 accessions were either pure or closer to pure type, while 118 accessions were admixed.

Genetic PCA indicated that PC1 (17.22%) and PC2 (7.32%) explained the highest proportions of the total variance (Figure 4). The accessions in admixture Subpopulations 2, 4, 5, and 6 were distinctively clustered as indicated by blue, purple, orange, and yellow circles, respectively (Figure 4a). These subpopulations had accessions of Indian and US origin (Figure 4b). The accessions from India show clustering, while other accessions were scattered (Figure 4b). The two different chickpea types, desi and kabuli, were also tightly clustered (Figure 4c).

FIGURE 4.

Principal components (PCs) analysis: PC1 and PC2 plotted with chickpea germplasm accessions as colored points according to (a) admixture subpopulation, (b) country of origin, and (c) chickpea types (desi [D] and kabuli [K]).

3.3. SNP discovery using GWAS

Significant SNPs were only identified for Ca, Mg, P, Mn, and Zn using the BLINK model in GAPIT (Figure 5). MAF, p‐values, and percentage phenotypic variance explained (PVE) for significant SNPs were provided in Table 3. Seven significant SNPs for Ca were located on chromosomes 1, 2, 4, 5, and 6 with MAF values ranging from 12.99% to 47.40%, respectively. The percentage PVE by these SNPs varied from 1.07% to 77.11%. One significant SNP each for P (MAF: 9.96%; PVE: 41.38%) and Mn (MAF: 28.14%; PVE: 13.82%) were located on chromosomes 2 and 1, respectively. Three significant SNPs for Mg were detected and located as follows: one on each of chromosomes 1 (MAF: 13.85%; PVE: 21.58%), 6 (MAF: 32.47%; PVE: 9.75%), and 7 (MAF: 21.21%; PVE: 23.16%), and two SNPs were found for Zn on chromosomes 2 and 6 (MAF: 48.05%; PVE: 16.59% and MAF: 42.86%; PVE: 9.55%). QQ plots for Ca, Mg, P, Mn, and Zn with the BLINK model are also shown in Figure 5. The results for other GWAS approaches can be found in Supporting Information Files S3 and S4 and Table S2. A summary of SNPs found in various GWAS approaches has been provided in Table 4. Several common SNPs were identified for different minerals when compared among these approaches except for copper. A single SNP was found common to each “K and Mn” and “Ca and Zn” (Table 4).

FIGURE 5.

Manhattan plots and quantile–quantile (QQ) plots with significant single‐nucleotide polymorphisms (SNPs) for chickpea minerals (Ca, Mg, P, Mn, and Zn) using the Bayesian‐information and linkage‐disequilibrium iteratively nested keyway (BLINK) model. Significant threshold lines for the Bonferroni threshold (p‐value < 0.05/10,602) are shown.

TABLE 3.

Minor allele frequency (MAF), p‐values, and percent phenotypic variance explained (PVE) for significant single‐nucleotide polymorphisms (SNPs) identified for chickpea minerals.

SNP	Chromosome number	MAF (%)	p‐value	PVE (%)
Ca
SCM001764.1_3076716	Chr 1	16.67	1.11E‐06	1.46
SCM001764.1_32039334	Chr 1	49.78	5.63E‐10	77.11
SCM001765.1_182658	Chr 2	12.99	7.87E‐07	1.07
SCM001767.1_1793224	Chr 4	22.94	9.70E‐07	3.27
SCM001767.1_24230971	Chr 4	30.74	9.73E‐08	4.20
SCM001768.1_32591343	Chr 5	47.40	4.47E‐07	4.27
SCM001769.1_33922656	Chr 6	43.07	4.22E‐06	1.57
Mg
SCM001764.1_736638	Chr 1	13.85	5.73E‐07	21.58
SCM001769.1_7550832	Chr 6	32.47	4.07E‐06	9.75
SCM001770.1_42748122	Chr 7	21.21	1.10E‐07	23.16
P
SCM001765.1_3611738	Chr 2	9.96	4.21E‐07	41.38
Mn
SCM001764.1_35092003	Chr 1	28.14	1.81E‐06	13.82
Zn
SCM001765.1_31891906	Chr 2	48.05	1.23E‐06	16.59
SCM001769.1_55962024	Chr 6	42.86	4.38E‐06	9.55

TABLE 4.

Summary of various genome‐wide association study (GWAS) approaches used for minerals analysis in the chickpea germplasm panel.

GWAS approach	Mineral
	Ca	K	Mg	P	Cu	Mn	Zn
GAPIT
Experiment 1	SCM001764.1_32039334 SCM001766.1_7443960 SCM001767.1_1793224 SCM001767.1_24230971 SCM001767.1_44309048 SCM001767.1_6590949 SCM001768.1_32591343 SCM001768.1_43367169 SCM001769.1_34116646 SCM001769.1_4459925 SCM001770.1_15703125 SCM001770.1_34379524	SCM001765.1_17919805 SCM001767.1_33401539 SCM001771.1_13727593		SCM001765.1_3611738 SCM001767.1_36475924	SCM001764.1_29712504 SCM001768.1_29539792 SCM001768.1_29539804	SCM001764.1_19382383	SCM001766.1_38033315
Experiment 2	SCM001764.1_3255353 SCM001765.1_13878731 SCM001767.1_13641148 SCM001767.1_13672841 SCM001768.1_15998666 SCM001768.1_20433203 SCM001768.1_32364616 SCM001769.1_33922656 SCM001769.1_4459925 SCM001769.1_48277780 SCM001769.1_55962024*	SCM001770.1_37316445				SCM001764.1_6265483 SCM001767.1_28590000
Combined	SCM001764.1_3076716 SCM001764.1_32039334 SCM001765.1_182658 SCM001767.1_1793224 SCM001767.1_24230971 SCM001768.1_32591343 SCM001769.1_33922656		SCM001764.1_736638 SCM001769.1_7550832 SCM001770.1_42748122	SCM001765.1_3611738		SCM001764.1_35092003	SCM001765.1_31891906 SCM001769.1_55962024*
Multivariate	SCM001764.1_35256946
METAL
Meta_GWAS	SCM001764.1_32039334 SCM001764.1_3255353 SCM001767.1_1793224 SCM001767.1_24230971 SCM001767.1_44309048 SCM001768.1_15998666 SCM001768.1_32364616 SCM001768.1_32591343 SCM001769.1_33922656 SCM001769.1_34116646 SCM001769.1_4459925 SCM001769.1_48277780 SCM001769.1_55962024* SCM001770.1_15703125 SCM001770.1_34379524	SCM001770.1_37316445∆	SCM001770.1_42748122	SCM001765.1_3611738 SCM001767.1_36475924		SCM001770.1_37316445∆ SCM001767.1_28590000 SCM001769.1_58029053	SCM001765.1_31891906
GEMMA
Experiment 2	SCM001766.1_14964439 SCM001768.1_15998666 SCM001769.1_47745516						SCM001771.1_2351536
Combined	SCM001764.1_37414139 SCM001764.1_32039334 SCM001764.1_36719702 SCM001765.1_13878731 SCM001766.1_14964439 SCM001767.1_12676452 SCM001769.1_51769101 SCM001769.1_38746552 SCM001769.1_47745516 SCM001771.1_10932470

Note: Common single‐nucleotide polymorphisms (SNPs) for a mineral across various approaches have been shown in green font, and symbols “*” and “∆” represent SNPs common between minerals.

Abbreviation: GAPIT, genome association and prediction integrated tool.

4. DISCUSSION

Mineral biofortification in staple foods gained attention in pulse breeding programs to improve the intake of the recommended daily allowance of minerals for target populations around the world. Chickpea mineral biofortification can contribute to alleviating mineral deficiencies, especially in Asian and African countries (Srungarapu et al., 2022). Lifestyle shifts among the Western population for vegetarian/vegan preferences and sustainable health goals further expanded the consumer market for chickpea and its products (Alcorta et al., 2021; Foyer et al., 2016). This study shows the potential for chickpea mineral biofortification indicated by the distribution, wide range, acceptable %RDA limits for various macrominerals and microminerals, and significant genotypic effects seen in the chickpea germplasm panel (Tables 1 and 2). The range found for minerals was comparable to previously published reports on chickpea (Hall et al., 2016; Rachwa‐Rosiak et al., 2015).

The mean concentrations in this panel were higher than USDA (2025) estimates for Ca (181.21 vs. 57 mg/100 g), K (971.99 vs. 718 mg/100 g), Mg (135.17 vs. 79 mg/100 g), P (383.01 vs. 252 mg/100 g), Cu (0.73 vs. 0.66 mg/100 g), Fe (5.32 vs. 4.31 mg/100 g), and Se (0.01 mg/L vs. 0 µg/100 g) (USDA, 2025). Conversely, the panel had lower mean concentrations than USDA (2025) estimates for Mn (4.05 vs. 4.15 mg/100 g) and Zn (2.41 vs. 2.76 mg/100 g). Percent RDA for minerals ranged from low to medium for Ca, K, Mg, P, Fe, and Zn to extensively higher for Cu, Mn, and Se. The large continuous genetic variation in mineral concentrations reported across various studies suggests these traits were quantitatively inherited (M. Zhang et al., 2020). This was further indicated by the significant effects noted, including interaction effects for different minerals. The quantitative and complex nature of chickpea minerals suggests polygenic control of these traits, with a few large and several small‐effect genes contributing to the total phenotypic variance of the population under study (Beavis et al., 2016; M. Zhang et al., 2020).

Some accessions had high concentrations of different minerals, such as CA15940141C, with high mean concentrations of K (1287.09 mg/100 g), Mg (161.82 mg/100 g), Mn (5.56 mg/100 g), P (455.42 mg/100 g), and Zn (3.09 mg/100 g). Likewise, W6 25897 had high mean concentrations of Mg (158.96 mg/100 g), P (478.24 mg/100 g), and Zn (3.15 mg/100 g), and W6 26044 had high mean concentrations of Cu (1.24 mg/100 g) and Se (0.04 mg/100 g). High broad‐sense heritability was found only for Ca, while moderate to low heritability was estimated for other minerals. This revealed that mineral concentrations were largely under the environmental influence, with a negligible genetic basis. Similar findings were reported by Powers et al. (2021) in peas (Pisum sativum L.). However, in a chickpea study, high heritability for Mn and Zn concentrations and moderate to low heritability for other minerals was reported (Farida Traoré et al., 2022). Several other studies in chickpea conducted for Fe and Zn reported high heritability estimates contrary to our findings (Jayalakshmi et al., 2019; Samineni et al., 2022; Srungarapu et al., 2022; Upadhyaya et al., 2016). Low‐heritability traits can be improved by leveraging the correlation information of high‐heritability traits. Selection for low‐heritability traits in advanced generations was using breeding methods such as pedigree and single‐seed descent (Casali & Tigchelaar, 1975). Employing advanced genomic tools with multiple different breeding populations further increases the scope of high predictability and improvement of low‐heritability traits (Budhlakoti et al., 2022). The multi‐trait genomic selection approach was evident to improve a low‐heritable trait, that is, cooking time in Phaseolus species when compared in different breeding populations (Diaz et al., 2021).

Also, the genotypic effects were not found to be significant for Fe and Se concentrations. Further testing by increasing the chickpea accessions and their evaluation across multiple years and locations could be useful to accurately determine mineral heritability. The correlation coefficients among these minerals were positive and significant (Figure 2), which suggested the possibility of selecting simultaneously for high concentrations of several minerals. The significant negative correlation between Ca and K implies that selection for either Ca or K can inversely impact the concentration of the other mineral. Earlier correlation studies in chickpea report similar results for most of the minerals considered here (Dragicevic et al., 2018; Erbas Kose & Mut, 2020; Vandemark et al., 2018). These findings revealed low to high correlation coefficients but strong significance among different mineral concentrations in this population.

Admixture analysis revealed nine subpopulations with the lowest cross‐validation error, which aligns with the results of a recent study in chickpea (Rahman et al., 2024). However, several studies identified a very high number of subpopulations, such as 12–15 (Ahmed et al., 2021; Kalve et al., 2022), while other studies indicated a small number (e.g., up to 2–3) (Channale et al., 2023; Roorkiwal et al., 2022; Srungarapu et al., 2022). Admixture population structure is a very population‐dependent grouping of accessions based on their ancestral history. The nine subpopulations noted here represent the distinct ancestral backgrounds for each group. Genotypic PCA was informative for distinctively classifying the chickpea types (desi and kabuli) and corresponded to the country of origin (Figure 4b,c), and clearly indicated the independent evolution of both chickpea types influenced by their domestication or utilization in a particular country (J. Zhang et al., 2024). Genetic segregation of accessions in origin‐based PCA suggests that population structure may impact the marker trait associations, especially if there are high frequencies of specific alleles in accessions of Indian origin. Notably, desi types in the panel are predominantly originated from India and in general have found limited use in US breeding programs. Thus, this germplasm could be a source to new alleles for economic, adaptive, and quality traits relevant to US chickpea breeding efforts. Correlation and heritability studies for various traits using accessions from diverse groups could provide a precise picture of how the traits behave in different groups.

Likewise, the knowledge of population structure for the chickpea germplasm accessions based on admixture and PCA could be useful to identify accessions with diverse ancestral and genetic backgrounds within a cultivated genome, despite having shared places of origin or chickpea types for accessions. This provides a chance to make an informed decision for selecting the appropriate parents in terms of diversity and ancestral relations to breed for specific minerals by efficiently using phenotypic and genetic data in mineral biofortification programs in chickpeas.

In pulse crops, genomic association mapping was used as a powerful tool for identifying SNPs associated with nutritional traits of interest (Johnson et al., 2021, 2023; Roorkiwal et al., 2022; Salaria et al., 2023; Srungarapu et al., 2022). A total of 14 SNPs were found for five chickpea minerals, namely, Ca, Mg, P, Mn, and Zn (Figure 5; Table 3). LD blocks were defined for each SNP based on the correlation coefficient (r ²) between different markers, and genes were identified within these LD blocks (Table S2). The seven SNPs identified for Ca were found on chromosomes 1, 2, 4, 5, and 6 with a varying number of genes located within their respective LD blocks. The identified genes within LD blocks and their annotations were based on the chickpea reference genome database, Jbrowse; however, the information about functional roles and pathways linked to these genes was derived from pre‐existing annotations found in the database. The genes Ca_v2.0_01546, Ca_v2.0_02963, and Ca_v2.0_02947 were linked to pathways for cellulose degradation (Xin et al., 2023), pectin lyase/glycoside hydrolase activity (He et al., 2024), and sphingolipid biosynthesis (Thuleau et al., 2013), respectively where Ca plays a key role. Another gene, Ca_v2.0_12639, encodes proteins for regulating mitochondrial functions, which can be directly impacted by the Ca²⁺ influx and concentrations in the cell mitochondria (Duchen, 2000). Three SNPs were found for Mg, having varying numbers of genes found within their LD blocks (Table S2). Among these genes, Ca_v2.0_00091 was associated with mRNA splicing and stabilization, an important cell function regulated by Mg ions homeostasis in the cell (Horlitz & Klaff, 2000), while another gene, Ca_v2.0_16266, was found to be regulating sorbitol synthesis in plants with Mg as a key contributor to directing its production and transport in roots (Tian et al., 2022).

A single gene, Ca_v2.0_03277, was found in an LD block corresponding to only one SNP, SCM001765.1_3611738, identified for P. This gene encodes NAC (NAM: No apical meristem, ATAF, CUC: Cup‐shaped cotyledon) transcriptional regulator protein superdomains, which channel signaling pathways to plant stress response including plant survival in low phosphorus levels or P starvation mode and maintaining P homeostasis (Nuruzzaman et al., 2013). Similarly, only one SNP (SCM001764.1_35092003) was found for Mn, but the LD block for this SNP had six genes. An important gene, Ca_v2.0_01628, was regulating a key putative membrane‐bound O‐fucosyltransferase enzyme utilized in glycosylation and is strongly activated by manganese (Palma et al., 2004). For Zn, a gene, Ca_v2.0_18608, located within an LD block for an SNP (SCM001769.1_55962024) was known to regulate numerous zinc finger proteins, which play roles in DNA recognition (Wolfe et al., 2000), yield performance (Xu et al., 2020), plant development, stress regulation, and homeostasis (Han et al., 2021, 2022). As stated before, these functions of genes associated with SNPs were based on the gene annotations available in the public chickpea genome database; however, gene functional and validation studies should be conducted to further understand the physiological roles and effects of these genes.

Additionally, several SNPs were shared among various approaches, highlighting the potential SNPs, which could be used for trait validation. Furthermore, the genotyping assays could be designed to use the SNPs for marker‐assisted selection in early generations in the chickpea breeding populations. The genotypes with the most favorable allele combinations could be advanced for evaluation of minerals in multi‐environment experiments. Also, genomic prediction models could be developed by using phenotypic data and significant SNPs as predictors along with genome‐wide marker data. These genomic prediction models can then be used to select promising genotypes based on their genomic estimated breeding values without performing phenotyping.

However, the SNPs not consistent across experiments were indicative of the influence of different environmental factors varying within the experimental site. Yet, common SNPs between two different minerals—“K and Mn” and “Ca and Zn”—suggest that the genes located within their LD blocks may have a pleiotropic role in mineral metabolism (Burstin et al., 2007). These SNPs could be used for simultaneously targeting the improvement of minerals.

Prior chickpea association studies were conducted for Ca, Mg, Cu, Mn, Fe, and Zn, where Fe and Zn were commonly studied as compared to other minerals (Fayaz et al., 2022; Roorkiwal et al., 2022; Samineni et al., 2022; Srungarapu et al., 2022). They identified SNPs for these minerals on various chromosomes. In our study, a total of nine minerals were included (Ca, K, Mg, P, Cu, Fe, Mn, Se, and Zn); however, we identified SNPs for Ca, Mg, P, Mn, and Zn. SNPs found in prior findings were different than the SNP association found in our study. This can possibly be due to the distinct and diverse experimental material and phenotypes observed for the minerals evaluated in our study.

Previous studies have reported PVE (%) for SNPs associated with mineral concentrations in chickpea (Diapari et al., 2014; Fayaz et al., 2022; Roorkiwal et al., 2022; Samineni et al., 2022; Upadhyaya et al., 2016), common beans (Delfini et al., 2021), and wheat (Bhatta et al., 2018). PVE estimated for both Mn and Zn ranged between 2% and 14.5% in the common bean (Delfini et al., 2021), wheat (Bhatta et al., 2018), and chickpea (Mn and Zn: Fayaz et al., 2022; Roorkiwal et al., 2022; Zn: Diapari et al., 2014; Samineni et al., 2022; Upadhyaya et al., 2016), which aligns with our findings (Table 3). Magnesium (9.75%–23.16%) and P (41.38%) were found with higher PVE values in our study compared to reports for Mg (0%–13.49%) and P (2%–>9%) in common beans (Delfini et al., 2021) and for Mg (1.4%–14.6%) in wheat (Bhatta et al., 2018) and chickpea (4.58%–8.56%) (Roorkiwal et al., 2022). However, PVE for Ca (1.07%–77.11%) was found to be quite higher for an SNP (SCM001764.1_32039334) in contrast to phenotypic variance (>21.5%) estimated in the previous studies (Bhatta et al., 2018; Cu et al., 2021; Delfini et al., 2021). PVE estimates were moderate to high for these minerals and variable compared to previous studies in different crops; nonetheless, conducting further studies for minerals in chickpea using large populations and multiple locations/years experiments can provide a true and broader picture for effective genomic regions explaining phenotypic variance.

5. CONCLUSIONS

Chickpea mineral biofortification can be a crucial breakthrough in addressing hidden hunger among populations in many nations. The chickpea germplasm panel evaluated in this study demonstrated high concentrations of both macrominerals and microminerals, as well as positive correlations among minerals. Superior accessions identified can be effectively used in breeding programs. The SNP discovery for various minerals showed important pathways (plant stress response, plant health, homeostasis, cellulose and sphingolipid pathways, etc.) governed by the associated candidate genes. These results urge researchers to explore mineral biofortification not only for human health but also from the perspective of plant health. The findings of this study may allow chickpea breeders to effectively utilize knowledge of phenotyping and genomic regions for breeding stress‐resilient, mineral‐rich chickpea cultivars.

AUTHOR CONTRIBUTIONS

Sonia Salaria: Data curation; formal analysis; methodology; software; validation; visualization; writing—original draft; writing—review and editing. Lucas Boatwright: Data curation; formal analysis; funding acquisition; methodology; project administration; software; supervision; validation; visualization; writing—review and editing. George Vandemark: Conceptualization; data curation; formal analysis; funding acquisition; investigation; methodology; project administration; resources; software; supervision; validation; visualization; writing—review and editing. Dil Thavarajah: Conceptualization; data curation; formal analysis; funding acquisition; investigation; methodology; project administration; resources; software; supervision; validation; visualization; writing—review and editing.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Table S1. Origin of chickpea accessions comprising the panel for mineral studies.

Table S2. Significant SNPs with potential LD blocks and candidate genes with their annotations.

File S3. Manhattan plots and QQ‐plots for different GWAS approaches.

File S4. List of SNPs found in different GWAS approaches.

Salaria, S. , Boatwright, L. , Vandemark, G. , & Thavarajah, D. (2025). Genetic mapping of a chickpea (Cicer arietinum L.) diversity panel for mineral biofortification towards human nutrition. The Plant Genome, 18, e70152. 10.1002/tpg2.70152

DATA AVAILABILITY STATEMENT

Data are available as a supplementary document and are attached to the manuscript. Phenotypic data for mineral concentrations in the chickpea germplasm panel, the VCF file (unfiltered for high MAF), and the scripts used for data analysis can be found at a public GitHub repository (https://github.com/SSalaria5/Chickpea_minerals_03_08_2025).