Characterization and identification of annual wild Cicer species for seed protein and mineral concentrations for chickpea improvement

Assigned to Associate Editor Luigi Guarino. Abstract Developing nutrient-rich crop cultivars is the most economic strategy to combat malnutrition resulting from protein and mineral deficiencies. Chickpea (Cicer arietinum L.) is an important staple grain legume source of good quality dietary protein around the world, particularly in southern Asia, northern Africa, and the Middle East. In the present investigation, the genetic variability for protein and mineral concentrations was studied in 41 accessions of cultivated chickpea and eight annual wild Cicer species of primary, secondary, and tertiary gene pool. Large variability was observed between and within Cicer species for seed protein, Fe, Zn, Cu, Mn, Ca, and Mg concentration with high heritability. C. chorassanicum (Bunge) Popov was found to be the most promising species for high seed protein and Ca; C. judaicum Boiss. for high seed Fe, Cu. and Mg; C. yamashiatae Kitam. for high seed Zn and Fe; and C. pinnatifidum Jaub. & Spach for high seed Mn concentrations. All the wild Cicer accessions except ICC20190 (C. echinospermum P. H. Davis) had high concentration of at least one or more seed nutrients. Wild Cicer accessions such as ICC17141 (C. chorassanicum), ICC17269 and ICC17303 (both C. pinnatifidum), ICC17261 and ICC17262 (C. reticulatum), ICC20236 (C. chorassanicum), and ICC17117 and ICC17281 (C. yamashitae) were found promising for multiple seed nutrients. As C. reticulatum Ladiz. and C. echinospremum accessions are crossable with cultivated chickpea, promising C. reticulatum accessions identified in the present study can be used in crossing program for developing new nutrient-rich chickpea cultivars.


Crop Science
especially among people having vegetarian diets. Among the micronutrients, iron (Fe) and zinc (Zn) deficiencies are the major health problems prevalent in Asia, Africa, and Latin America affecting about 3 billion people (Darnton-Hill, Bloem, & Chopra, 2006;Welch & Graham, 2004). The Fe and Zn are the cofactors of important proteins such as hemoglobin, and cytochrome and transcription factors. Inadequate intake of these minerals in the diet of about 50% of the world population may affect their proper growth and development (Brown, Peerson, Rivera, & Allen, 2002;Welch, 2002). The calcium (Ca) deficiency causes rickets in children and osteomalacia, a generalized demineralization of bones, in adults and may also contribute to osteoporosis (Hays & Swenson, 1985;Malhotra, 1998;Murray, Granner, Mayes, & Rodwell, 2000). The magnesium (Mg) deficiency increases neuromuscular hyperirritability, which may lead to irregular heartbeat and tremors (Murray et al., 2000). Similarly, copper (Cu) deficiency may also lead to impaired cardiovascular system development, neurological and immunological abnormalities, and alterations in cholesterol metabolism (Gambling & McArdle, 2004;Georgieff, 2007;Klevay et al., 1984;Reiser, Yang, & Canary, 1987). Manganese (Mn) is an essential micronutrient for metabolic regulation, bone mineralization, and formation of glycosaminoglycans (Wedler, 1994).
To combat nutritional deficiencies, the conventional efforts such as micronutrient supplementation and food fortification, are not affordable in developing countries. Hence, one of the most important and economic approaches is to biofortify common staple crops with macro-and micronutrients by breeding for nutritionally dense, high-yielding crop cultivars (Bouis, 2003). Staple cereals alone do not meet the daily requirement of essential micronutrients (Shi et al., 2008). In contrast, grain legumes or pulses, having high protein, vitamin, and mineral contents, are essential for alleviating micronutrient deficiencies and malnutrition and thus complement the conventional cereal-based diets (Broughton et al., 2003;Ufaz & Galili, 2008).

Core Idea
• Large variability present within and between cultivated and wild Cicer species for grain nutrients • Promising wild accessions having high concentration of protein and mineral contents identified • Most diverse pair of accessions identified for use in crossing program for chickpea improvement neni, Sobhan, & Chibbar, 2015). Chickpea is a potential staple food crop in ∼60 countries, and improving its nutritional quality will help in reducing protein and mineral deficiencies for people in these countries. To achieve this objective, new and diverse sources of variations are needed to further improve its nutritional quality. Sufficient variability is present in genus Cicer, which contains 43 wild Cicer species for chickpea improvement. Wild Cicer species possess many useful traits and can be exploited as important sources of variability for improving nutritional quality including protein and mineral content of cultivated chickpea. Limited information is available regarding variability for seed nutrient concentration in the wild Cicer species compared with the cultivated chickpea. The present investigation was carried out to study the genetic variability for seed protein and mineral concentrations among different accessions-originating from or collected from different countries and belonging to eight annual wild Cicer species and cultivated chickpea-and to identify promising wild Cicer accessions having high seed protein and mineral concentration for use in chickpea improvement programs.  (Table 1). Six chickpea varieties, belonging to desi (ICCV96029, JG 11, ICCV10, and G130) and kabuli (KAK 2 and L 550) types, are extensively cultivated in India and were used as checks (Table 1). ICCV96029 (Kumar & Rao, 2001), JG 11 (Dattatri, Mahadeva, Sudhakar, Dhanalakshmi, & Sudhakar, 2010), ICCV10 (Gowda et al., 1995), and G130 (Singh, 1987) are super early-, early-, medium-, and late-flowering  (Zope, Wanjari, Kumar, Van, & Rao, 2002) is a medium-flowering, kabuli-type variety, and L 550 (Dua, Chaturvedi, & Sewak, 2001) is late-flowering, kabuli-type variety.

Methodology
The study was carried out under the controlled greenhouse conditions maintained at 22˚C air temperature in 2014, 2015, and 2016 at ICRISAT, Patancheru, India. The hard seed coat of wild Cicer accessions was scarified to initiate germination. Before sowing, the seeds were treated with a combination of fungicides (2 g thiram plus 1 g carbendazim kg −1 seed) for prevention from seed and soil-borne fungal diseases. Nonscarified seeds of chickpea varieties and scarified seeds of wild Cicer accessions were kept for germination in wet Petri plates at room temperature for 3 d. The germinated seedlings were transplanted (one seedling per pot) in pots containing a mixture of sterilized black soil, sand, and farmyard manure in 2:1:1 proportion. Plants were maintained in three replications containing two pots per replication. After 1 mo of germination, the seedlings were grown under 18-h extended daylength till maturity. At maturity, pods from individual plants were harvested and the healthy matured seeds free from dust and metal particles were used for analysis. About 20 g of seeds of each accession wase analyzed at Charles Renard Analytical Laboratory, ICRISAT, Patancheru to estimate seed protein and mineral (Fe, Zn, Cu, Mn, Ca and Mg) concentrations. Total N was estimated colorimetrically on continuous-flow auto-analyzer by using sulfuric acid and selenium digestion (Sahrawat, Kumar, & Murthy, 2002). Protein concentration was calculated by using conversion factor 6.25 (total N × 6.25). Concentration of micronutrients (Fe, Zn, Cu, Mn) and macronurtients (Ca and Mg) were estimated on inductively coupled plasma-optical emission spectrometry method (Wheal et al., 2011).
The seeds were washed with distilled water followed by drying in the oven at 60˚C for 48 h before grinding. About 20 g of dried seed sample from each accession was powdered in a stainless steel grinder and the powered seed samples were kept overnight in the oven at 55˚C. The samples and the standards were simultaneously digested with appropriate blanks in triplicate (three independent analyses). About 0.3 g of powered sample was taken in a 50-ml polypropylene tube, and 2 ml of concentrated nitric acid and 0.5 ml hydrogen peroxide was added. The contents were left overnight for cold digestion in a digestion chamber. Initially, the samples were digested at 80˚C for 0.5 h and afterwards at 125˚C for ∼2 hr to get colorless and clear digest. Upon cooling of the digests, the contents were dissolved in distilled water and the final volume made up to 25 ml followed by mixing on vortexer. Digested samples were filtered and analyzed for seed Fe, Zn, Cu, Mn,

Crop Science
Ca, and Mg concentrations by inductively coupled plasmaoptical emission spectrometry (Teledyne Leeman Labs). The concentration of micro-and macronutrients were expressed in mg kg −1 and g kg −1 seed, respectively. Protein concentration was expressed as a percentage.

Statistical analysis
The ANOVA was carried out using data recorded on five traits (seed protein, Fe, Zn, Cu, and Mn concentration) in three replications during 3 yr (2014, 2015, and 2016) and on two additional traits (seed Ca and Mg concentration) in 2016. and The replicate-wise values of protein and mineral concentrations recorded in 41 accessions from nine Cicer species were used for year-wise statistical analysis following general ANOVA for various factors and their interactions using Gen-Stat (15th ed.) (http://www.genstat.co.uk). Bartlett's homogeneity of variance test (Bartlett, 1937) revealed that the error variances between 3 yr were homogeneous for most traits, and therefore, pooled analysis was performed. For pooled ANOVA, data on five traits (seed protein, Fe, Zn, Cu, and Mn concentration) in 41 accessions belonging to nine Cicer species was used. For analysis, the accessions were nested within species (species/accessions) and the significance of differences were tested within and between species, species/accessions, and interaction means by using respective least significant differences. Broad-sense heritability (h 2 ) was calculated using pooled data for seed protein, Fe, Zn, Cu, and Mn and using 2016 data for seed Ca and Mg. Broad-sense heritability was categorized as low (<0.30), moderate (0.30-0.60), and high (> 0.60). Pearson correlation coefficients were estimated to identify the useful associations between different seed nutrients.
Pooled data on five seed nutrient traits-protein, Fe, Zn, Cu and Mn concentration-was used to create a phenotypic distance matrix for each trait by calculating the differences between each pair of entries. The diversity index was calculated for each trait by averaging the differences in the phenotypic values divided by the respective range (Johns et al., 1997). The mean, minimum, and maximum diversity were calculated, and the pair of species and accessions showing the maximum and minimum diversity were identified. Cultivated and wild Cicer species and accessions were clustered using Euclidean phenotypic distance matrix following hierarchical clustering (Ward, 1963) in R package 'cluster.' The performance of wild Cicer accessions over years for seed nutrient concentration was compared with the best cultivated chickpea, used as checks. The promising accessions with higher seed Fe, Zn, Cu, Mn, and protein concentrations were identified based on the pooled analysis. For seed Ca and Mg concentration, the promising accessions were identified based on the performance of wild accessions compared with cultivated chickpea in 2016.

RESULTS
The year-wise ANOVA showed significant differences among species and species/accessions for all the traits (P ≤ .001). Pooled analysis showed significant differences between years, species × years, and species/accessions × year ( Table 2). Partitioning of the total sum of squares revealed the greater importance of species/accessions followed by species toward total variability in seed protein concentration in individual years (30-39% variations attributed to species/accessions and 28-34% attributed to species) and equal contribution of species/accessions (16%) and species (17%) in pooled analysis. Variability in all seed mineral concentrations was mainly due to species in year-wise and pooled analyses (Table 2).

Crop Science
T A B L E 2 Analysis of variance (ANOVA) for protein and mineral concentration in annual wild and cultivated Cicer accessions evaluated over three seasons (2014)(2015)(2016)  Overall, C. chorassanicum was superior for all the seven seed nutrients (protein, Fe, Zn, Cu, Mn, Ca, and Mg) and C. pinnatifidum for six seed nutrients (Fe, Zn, Cu, Mn, Ca, and Mg).

Within-species variability
Based on the narrow range, it is evident that low variability was present within cultivated and different wild Cicer species for seed protein concentration (Table 3; Figure 1). When the accessions were compared with the respective species means, only one accession, ICC17303 from C. pinnatifidum showed significantly higher seed protein concentration (24.2%). For seed Fe concentration, the highest vari-ability was observed within C. judaicum (161.2-232.7 mg kg −1 ), wherein three accessions, ICC17204, ICC17271, and ICC17316, showed significantly higher seed Fe concentration (204-233 mg kg −1 ) when compared with species mean (199 mg kg −1 ). Though there were only two accessions, ICC17141 and ICC20236, of C. chorassanicum, a large variation was observed between them for seed Fe and Mn concentrations (57 mg kg −1 Fe and 64 mg kg −1 Mn concentration in ICC17141 and 78 mg kg −1 Fe and 51 mg kg −1 Mn concentration in ICC20236). Similar pattern was observed between two accessions of C. yamashitae wherein seed Fe and Zn concentration was significantly higher in ICC17117 (∼172 mg kg −1 Fe and 50 mg kg −1 Zn) and seed Mn concentration in ICC17281 (67 mg kg −1 ). Variability for seed Fe concentration was also observed within C. reticulatum and C.  (Table 3), and three accessions, ICC17123, ICC17124, and ICC17164, showed significantly higher seed Zn concentration (39-41 mg kg −1 ) than the species mean ( Figure 1). The highest variability for seed Cu and Mn concentration was observed within accessions of C. judaicum (Table 3). Within C. judaicum accessions, three accessions, ICC17204, ICC17271, and ICC17316, had high seed Cu concentration (8.6-8.9 mg kg −1 ) and ICC17188, ICC17274, and ICC17316 had high seed Mn concentration (92-99 mg kg −1 ). High variability for seed Ca and Mg concentration was observed within both cultivated and wild species (Table 3). Within C. arietinum, ICGV 96029 showed high seed Ca, and G130 showed high seed Ca and Mg concentration (Figure 1). Amongst the wild accessions, C. echinospermum accession ICC20192 (3.58 g kg −1 ), C. judaicum accession ICC17204 (5.96 g kg −1 ), and two C. pinnatifidum accessions, ICC17200 (4.47 g kg −1 ) and ICC17276 (4.48 g kg −1 ), had significantly higher seed Ca concentration than the respective species means (Figure 2). Similarly, two C. pinnatifidum accessions, ICC17126 (1.80 g kg −1 ) and ICC17303 (1.97 g kg −1 ), and one C. reticulatum accession, ICC17262 (2.01 g kg −1 ), had high seed Mg concentration (Figure 2).

Correlation between seed nutrients
Based on accession means, the significant positive correlation was observed between seed Fe and Cu (r = .816), seed Fe and Mn (r = .377), and seed Mn and Cu (r = .401) concentration (Table 4), whereas seed protein concentration had significantly negative association with seed Fe (r = −.492) and Mn (r = −.419) concentration. At the species level, protein concentration was negatively associated with seed Fe (r = −.684) concentration. Significantly positive association was observed between seed Fe and Cu (r = .800) concentration (Table 4). Further, significantly positive correlation of seed Ca was observed with seed Fe and Cu concentration, both at species and accessions levels. Positive but nonsignificant correlation was observed between seed Ca and Mg concentration based on accession means (r = .426) and species mean (.330) in 2016 (data not given).

Phenotypic diversity, cluster analysis, and identification of promising accessions
At the species level, the mean phenotypic diversity index was 0.220. The maximum diversity was found between C. yamashitae and chickpea (0.534) and the least diversity was between C. reticulatum and C. echinospermum (0.029). For accessions, the mean phenotypic diversity index was 0.135. The maximum diversity was found between chickpea genotype ICCV96029 and C. judaicum accession ICC17271 (0.444) and minimum diversity was between C. cuneatum accession ICC20215 and C. cuneatum ICC20175 (0.001). The five most similar and diverse pairs of species accessions were identified (Table 5).
Hierarchical cluster analysis based on five seed nutrients over 3 yr was performed to separate the cultivated and wild species and accessions into distinct groups. The cluster analysis following Ward's method grouped nine cultivated and wild Cicer species into five clusters (Figure 3). Cluster 1

Crop Science
T A B L E 4 Correlation between protein and mineral concentrations in annual wild and cultivated Cicer accessions evaluated over years (2014)(2015)(2016)  is comprised of chickpea, C. reticulatum, C. echinospermum, C. bijugum, and C. cuneatum. The remaining four wild Cicer species were clustered separately with each species in individual cluster (Figure 3). Based on accession means, 41 cultivated and wild Cicer accessions were grouped into five clusters ( Figure 4). Cluster 1 was the largest cluster, consisting of 21 accessions that included all cultivated chickpea genotypes, all the accessions of C. bijugum (3 accessions), C. echinospermum (5 accessions), C. cuneatum (4 accessions), and three accessions of C. reticulatum. The remaining four accessions of C. reticulatum and both the accessions of C. chorassanicum were clustered into Cluster 2. All the accessions of C. judaicum (7 accessions), C. pinnatifidum (5 accessions), and C. yamashitae (2 accessions) were clustered into Cluster 3, Cluster 4, and Cluster 5, respectively. Cluster 2 had the highest mean for seed protein concentration followed by Cluster 1. Cluster 3 had the highest mean seed Fe concentration and Cluster 4 had the highest seed Mn concentration.
The performance of wild Cicer accessions was compared with the cultivated chickpea genotypes (used as checks). Among the cultivated chickpea, G130 had significantly higher seed protein and mineral concentration and hence this genotype was selected as the best check for comparing the wild Cicer accessions. For seed protein concentration, only one accession, ICC17141 (C. chorassanicum, 25.5%), was significantly better than G130 (24.1%). A total of 27 wild Cicer accessions for seed Fe concentration (52-233 mg kg −1 ), 12 accessions for seed Zn concentration (36-50 mg kg −1 ), 23 accessions for seed Cu concentration (4.3-8.9 mg kg −1 ), 22 accessions for seed Mn concentration (41.4-191.8 mg kg −1 ), 25 accessions for seed Ca concentration (3.245-6.122 g kg −1 ) and 15 accessions for seed Mg concentration (1.699-2.007 g kg −1 ), were significantly better than the best check, G130 (46.7 mg kg −1 seed Fe, 34.6 mg kg −1 seed Zn, 3.7 mg kg −1 seed Cu, 36.5 mg kg −1 seed Mn, 2.78 g kg −1 seed Ca, and 1.59 g kg −1 seed Mg) ( Table 6) Cluster 2 (1) Cluster 3 (1) Cluster 4 (1) Cluster 5 (1) F I G U R E 3 Cluster diagram depicting different clusters formed using nine annual wild and cultivated Cicer species following Ward's method based on five seed nutrient concentrations. Cluster 1 consisted of cultivated chickpea, progenitor species C. reticulatum, cross-compatible secondary gene pool species C. echicospermum, and cross incompatible tertiary gene pool species, C. bijugum, and C. cuneatum. Remaining cross-incompatible tertiary gene pool species were clustered separately in different clusters five accessions for each seed nutrient are given in Table 7. Overall, on pooled analysis, only one accession, ICC17141 (C. chorassanicum), was found promising for all five seed nutrients (protein, Fe, Zn, Cu, and Mn) followed by two accessions, ICC17269 and ICC17303 of C. pinnatifidum, ICC17261 of C. reticulatum, ICC20236 of C. chorassanicum, and two accessions, ICC17117 and ICC17281, of C. yamashitae were found promising for all four minerals (Table 6).

DISCUSSION
Identification and exploitation of wild species harboring new and diverse genetic variability for seed protein and mineral concentration is the most economic strategy to alleviate pro-tein and nutrient deficiency. Enormous variability for seed protein and nutrients was observed between and within eight annual Cicer species used in this study. Thus, nutrient-rich accessions can be selected for use in chickpea breeding programs. The previous attempts to study the genetic variability for grain nutrient concentration in chickpea were mainly based on cultivated germplasm (Aliu et al., 2016;Diapari et al., 2014;Upadhyaya et al., 2016;Vandemark et al., 2018) or on a few accessions belonging mostly to the primary, secondary, or a few tertiary gene pool species (Kaur, Grewal, Gill, & Singh, 2019;Wettberg et al., 2018) mostly limited to estimation of protein concentration (Ocampo, Robertson, & Singh, 1998). The present study provides a comprehensive analysis of variability for protein and mineral concentrations in eight annual wild and cultivated Cicer species originating or collected from seven countries. Significant variations

Crop Science
KAK 2 IC C 17 18 7 IC C 1 7 1 5 6 I C C 1 7 2 8 9 Cluster 2 (6) Cluster 3 (7) Cluster 4 (5) Cluster 5 (2) F I G U R E 4 Cluster diagram depicting different clusters formed using 41 annual wild and cultivated Cicer accessions following Ward's method based on five seed nutrient concentrations. Cluster 1 consisted of all accessions of cultivated chickpea, C. echinospermum, C. bijugum, C. cuneatum, and three out of seven accessions of C. reticulatum for seed protein concentration in annual wild Cicer species has been reported, which ranged from 16.8% in C. cuneatum to 26.8% in C. pinnatifidum amongst eight wild Cicer species (Ocampo et al., 1998) and 24.3% in C. judaicum to 25.3% in C. pinnatifidum amongst three wild Cicer species: C. judaicum, C. pinnatifidum, and C. echinospermum (Kaur et al., 2019). In the present study, the highest seed protein concentration was found in C. chorassanicum (24.9%). Seed protein concentration in C. cuneatum accessions was found higher (22.8-23.8%) in the present study than the previous report (Ocampo et al., 1998). In contrast, the seed protein concentration in C. pinnatifidum accessions was lower (21.3-24.2%) in the present study than the previous reports (Kaur et al., 2019;Ocampo et al., 1998). Differences were observed for most of the seed nutrient concentrations when compared with the previous report (Kaur et al., 2019). For example, the range of seed Fe and Zn concentrations in C. pinnatifidum was almost similar but large differences were observed in C. echinospermum (35.1 mg kg −1 Zn and 48.6 mg kg −1 Fe in the present study) when compared with the previous report (15.6 and 92.4 mg kg −1 Zn and Fe, respectively) (Kaur et al., 2019). Similarly, average seed Fe concentration in C. judaicum was reported to be 83.2 mg kg −1 (Kaur et al., 2019) compared with 199 mg kg −1 in the present study. These C. pinnatifidum), ICC17261 and ICC17262 (both C. reticulatum), ICC20236 (C. chorassanicum), and ICC17117 and ICC17281 (both C. yamashitae) were found to be promising for multiple seed nutrients. The highest seed Fe concentration was found in the accessions of C. judaicum. These accessions were originated in different countries; for example, ICC17316, having 233 mg kg −1 seed Fe concentration, is from Ethiopia; ICC17271 (209 mg kg −1 ) is from Lebanon; ICC17204 (204 mg kg −1 ) is from India; ICC17149 (201 mg kg −1 ) is from Israel; ICC17148 (201 mg kg −1 ) is from Israel; ICC17188 (183 mg kg −1 ) is from Syria; and ICC17274 (161 mg kg −1 ) is from Syria. These accessions also have high seed Cu, Mn, Ca and Mg concentration compared with other wild Cicer accessions.

Crop Science
Of these selected accessions, ICC17261 and ICC17262 (both from Turkey) hold a great potential for improving nutritional quality of chickpea as these accessions belong to the primary gene pool species C. reticulatum. C. reticulatum is easily crossable with cultivated chickpea (Singh & Ocampo, 1997), and the two promising accessions identified in this study can play an important role in improving seed Fe, Zn, Cu, and Mn concentration of cultivated chickpea. Another crossable species is C. echinospermum (Singh & Ocampo, 1997), and the two accessions, ICC20218 and ICC20244, of this species (both from Turkey) were found to have high seed Fe concentration. Another potent C. echinospermum accession having high seed Zn and Ca concentration was ICC20257. These accessions can be used in chickpea improvement programs for improving seed protein, Fe, Zn, and Ca concentration of cultivars. The promising C. reticulatum and C. echinospermum accessions can also be used to develop multiparent populations following complex cross approach such as three-way and four-way crosses with a view to combine genes and alleles from different wild species into a common cultivated genetic background (Sharma, 2017). The complex cross approach may lead to the generation of desirable transgressive segregants as a result of unexpected epistatic effects as has already been reported in Cicer (Ocampo et al., 1998).
Use of promising accessions belonging to tertiary gene pool species C. chorassanicum, C. pinnatifidum, C. judaicum, C. bijugum, C. cuneatum, and C. yamashiatae is hindered because of cross-incompatibility barriers. Several efforts were made in the past to use tertiary gene pool species for chickpea improvement (Ahmad & Slinkard, 2004;Ahmad, Slinkard, & Scoles, 1988;Badami, Mallikarjuna, & Moss, 1997;Clarke et al., 2006;Clarke, Kumari, Khan, & Siddique, 2011;Croser, Ahmad, Clarke, & Siddique, 2003;Mallikarjuna, 1999;Mallikarjuna & Muehlbauer, 2011;Mallikarjuna, Jadhav, Nagamani, Amudhavalli, & Hoisington, 2007). In all these studies, efforts were made to generate interspecific hybrids between cultivated chickpea and different cross-incompatible tertiary gene pool species following embryo rescue techniques. However, no success was obtained in these studies except in C. arietinum × C. bijugum (Mallikarjuna et al., 2007). Hence, there is a need to develop novel techniques and strategies for accessing genes and alleles from these important and unexploited sources. One of the strategies could be to study the crossability relationships of primary and secondary gene pool species, C. reticulatum and C. echinospermum, respectively, with six tertiary gene pool species. Based on these studies, the potential of C. reticulatum and C. echinospermum as 'bridge species' can be explored to access variability from the crossincompatible tertiary gene pool species.
Highly positive association observed between seed Fe and Cu as well as seed Fe and Mn concentration provides an opportunity for simultaneous improvement of these seed nutrient traits. Vandemark et al. (2018) also found significant positive correlation between Fe and Cu content in chickpea and lentil and between Fe and Mn in chickpea. However, negative correlation between seed protein and Fe concentration will hinder the progress in improving both seed protein and Fe concentration simultaneously in chickpea cultivars. Nonsignificant association of seed protein with other nutrients such as Zn, Cu, Ca, and Mg indicates the possibility of combining higher seed Zn and other nutrient concentrations with higher seed protein.
This study led to the identification of wild Cicer accessions for multiple seed nutrient traits as well as the most diverse and similar pair of accessions. The five most diverse pairs of accessions or species indicated that the cultivated chickpea is highly diverse from C. judaicum and C. yamashitae, whereas it was found to be closely related to C. echinospermum and C. bijugum. In contrast, C. cuneatum accessions ICC 20215 and ICC 20175 (both from Ethiopia), C. echinospermum accessions ICC 20257 and ICC 20192 (both from Turkey), and C. reticulatum accessions ICC17262 and ICC17261 (both from Turkey) were found to be the most similar accessions. These results indicate the presence of duplicate accessions of these species in the gene bank.
Further, the cluster analysis grouped 41 cultivated and wild accessions into five clusters, wherein similar accessions were placed in the same cluster. Interestingly, all cultivated chickpea genotypes along with all accessions of C. echinospermum, C. bijugum, C. cuneatum, and three of the seven C. retictulaum accessions were clustered together in one cluster, whereas remaining four accessions of C. reticulatum were grouped in a separate cluster. This helps in selecting the most diverse accessions within cross-compatible gene pool for immediate use in chickpea improvement program. It is interesting to note that the accessions of C. choarssanicum, C. judaicum, C. pinnatifidum, and C. yamashitae were clustered in the species-specific groups and clusters. This shows that the variability between species is higher than the within species variability for seed nutrient concentrations. Systematic hybridization between promising accessions for seed protein and mineral concentrations chosen from different clusters will help to create new genetic variability for improving the seed nutrient concentrations of cultivated chickpea. Further, the contrasting pair of accessions within or between cross-compatible species such as C. reticulatum accession ICC17261, having significantly higher concentrations of seed Fe, Zn, Cu, and Mn, and C. echinospermum accession ICC 20190, having the lowest concentration of all the seed nutri-ents, can be used to develop mapping populations for the mapping of genes and quantitative trait loci for multiple seed nutrients. Overall, the selected wild Cicer accessions with high seed nutrient concentrations hold promise in improving the nutritional content of high-yielding and well-adapted chickpea cultivars.

SUMMARY AND CONCLUSION
Developing nutrient-rich chickpea cultivars is the most economic strategy to combat malnutrition resulting from protein and mineral deficiencies. In this study, efforts were made to identify new and diverse sources of variations for seed protein and mineral (Fe, Zn, Cu, Mn, Ca, and Mg) concentrations among 41 accessions of cultivated chickpea and eight annual wild Cicer species belonging to primary, secondary, and tertiary gene pools. Overall, large variability was observed both between and within species for seed protein, Fe, Zn, Cu, Mn, Ca, and Mg concentration with high heritability. Most of the wild Cicer accessions were found to have high concentration of at least one or more seed nutrients. Two C. reticulatum accessions, ICC17261 and ICC17262; two C. chorassanicum accessions, ICC17141 and ICC20236; two C. pinnatifidum accessions, ICC17269 and ICC17303; and two C. yamashitae accessions, ICC17117 and ICC17281, were found promising for multiple seed nutrients. Though promising C. reticulatum accessions can be used in crossing program for improving cultivated chickpea, the use of tertiary gene pool species for chickpea improvement needs concerted efforts in standardizing protocols and strategies to overcome crossincompatibility barriers.

AU T H O R C O N T R I B U T I O N S
SHS planned the study; SHS and SAL raised the material under green house; CN analyzed the seed samples for seed nutrient concentrations in quality lab; SS analyzed the data; SHS and SAL prepared the manuscript; SAL, CN and SS provided their inputs. All the authors reviewed and approved the final manuscript.

A C K N O W L E D G E M E N T S
The partial finding support provided by the CGIAR Research Program on Grain Legumes and Dryland Cereals (GLDC) is duly acknowledged. This work was undertaken as part of the initiative 'Adapting Agriculture to Climate Change: Collecting, Protecting and Preparing Crop Wild Relatives,' which is supported by the Government of Norway. The project is managed by the Global Crop Diversity Trust with the Millennium Seed Bank of the Royal Botanic Gardens, Kew, and implemented in partnership with national and international genebanks and plant breeding institutes around the world. For further information, see the project website: http://www. cwrdiversity.org.

C O N F L I C T O F I N T E R E S T
Authors have no conflict of interest.