Genetic Diversity Analysis of Walnut (Juglans regia L.) from Kashmir Valley Using RAPD and ISSR Markers

The genus Juglans is characterized by monoecious and heterodichogamous character, counting approximately 40 species, caterogised into three sections, Dioscaryon Dode, Cardocaryon Dode and Thyoscaryon Dode. Juglans regia L. belongs to Dioscaryon Dode Section and is the most economically important nut and oil species. Walnut grows well in regions having temperate climate and is naturally distributed from east of Turkey to north of Pakistan, Iran, Afghanistan, mountains of Nepal and Central Asia [1]. In year 2014/15, worldwide production of walnuts reached 655,651 metric tons (kernel basis), 14 percent high from the previous year, and 87 percent high when compared with 2004/05, which confirms the upward drift over the last decade [2]. The China and United States clearly led global production, accounting for the 67 percent of total world production. India contributes 2% to the world’s total walnut production. Worldwide production of walnut has been rising speedily in recent years mainly in Asian countries due to its high nutritional value and antioxidant properties [3]. In India, walnuts are grown in Arunachal Pradesh, Himachal Pradesh, Uttarakhand and Jammu and Kashmir. J&K is the main centre for commercial walnut production in India and contributes around 98% of the country’s output and the state has been declared as the “Agri. Export zone for Walnuts”. Most walnuts produced for export by India are produced in this state. According to the United Nations Development Programme Special Unit for South-South Cooperation (UNDP SU/SSC), 63,000 hectares of Jammu and Kashmir are presently under walnut cultivation. These plantings produce around 60,000 tonnes of walnuts worth an approximated 25 million rupees. Moreover, the plant has medicinal importance for human health which is because of its high antioxidant capacity [4] and ω-3 fatty acid concentration [5].

A thorough understanding of population diversity is must for the valuable utilization of the genetic variability accessible to breeders [6].
In order to raise the walnut production and exportation, new genotypes should be introduced and old genotypes should be preserved in gene banks [7]. For achieving this goal, the foremost step is detection of superior genotypes and then characterization of their genetic variety. For this, the conventional methods are now being complemented by molecular methods, enabling breeders to take better decisions while choosing the germplasm to be used in breeding programs [8]. Therefore molecular markers are suggested to explore genetic diversity in walnut [9]. Reports are available concerning evaluation of genetic diversity among walnut genotypes using Restriction Fragment Length Polymorphism (RFLP) [10], Random Amplified Polymorphic DNA (RAPD) [11], Inter-Simple Sequence Repeat (ISSR) markers [12,13], Microsatellite markers [9,14] and Amplified Fragment Length Polymorphism (AFLP) [15]. Ultimately, high-quality genotypes could be used in crossing programs to produce new commercial genotypes.
RAPD markers have the advantages of being easy to use, low cost of experiments and they also cover the entire genome. Moreover, they can be used as a primary method of choice for screening the genotypes to find duplicates in collections. RAPDs have been used to evaluate the level of polymorphism in Juglans [7] as well as in other important fruit tree species, with attractive results [16]. In Jammu and Kashmir, India, a high level of genetic diversity was found within walnut populations Jammu and Kashmir with different molecular markers.

Plant material
A total of ninety six walnut genotypes were obtained from the research farm of Central Institute of Temperate Horticulture, Srinagar, India representing 94 local selections from eight geographically isolated districts of Jammu And Kashmir State and two exotic varieties (Table  1). Maximum number (32) of genotypes were represented by district exhibited by Morphological, RAPD and SSR markers [17,18].
ISSRs are important tool to assess the genetic diversity in a plant collection or wild species, and are helpful to identify genotypes, even in highly related individuals [12,[19][20][21]. In addition to the reproducibility and low cost of the technique [22], the most important advantage of this method is its universality and ease of development [6], ISSRs have been extensively used to evaluate genetic diversity of populations of economic forest tree species [23]. To our knowledge, ISSR have never been used in walnut genetic diversity studies in India. So we are reporting it for the first time. The aim of this study was to find out the genetic relationships among 96 genotypes of Juglans regia from different selected districts of Budgam followed by Ganderbal (26), Pulwama (11), Baramulla (11), Anantnag (07), Shopian (03) Srinagar (03) and Bandipora (01).

DNA extraction
Total DNA was extracted from fresh, young leaf tissue (0.5 g), by a Cetyltrimethylammonium bromide (CTAB) method as described by Doyle [24]. The extract was further purified by standard method of Sambrook [25]. Concentration and purity of DNA was determined by electrophoresis in 0.8% agarose gels and spectrophotometrically on Nano Drop (Thermo). This extraction method yielded an average DNA content of 500 μg g -1 of leaf tissue, and purified DNA revealed A260/ A280 = 1.80 ± 0.22, revealing the high purity of DNA samples.

RAPD and ISSR amplification
DNA from each walnut genotype was screened with 13 RAPD primers and 19 ISSR primers (Tables 2 and 3). For RAPD, the PCR reaction (50 μl) consisted of: 1X reaction buffer (20 mM

Molecular data analysis
For qualitative analysis, the RAPD and ISSR data were analyzed as dominant markers. Presence or absence of the band was scored as 1 or 0 respectively, thus obtaining the molecular identification profile for each genotype. The capacity of each primer to distinguish among the genotypes was evaluated by the number of amplicons per primers, Percent Polymorphism, Polymorphic Information Content (PIC) [26], Effective Multiplex Ratio (EMR), Resolving Power (Rp) [27] and Marker Index (MI) [28].
The dendrogram was constructed using NTSYS-pc version 2.02e software [29] (Applied BioStatistics, Inc., Setauket, NY, USA) package to compute pairwise Jaccard's similarity coefficients [30]and the similarity matrix thus obtained was used in cluster analysis using an UPGMA (unweighted pair-group method with arithmetic averages) and sequential, agglomerative, hierarchical and nested (SAHN) clustering algorithm. In order to estimate the goodness of fit of the UPGMA cluster analysis between similarity matrix and the dendrogram of RAPD and ISSR, the coefficient of correlation was calculated using the Mantel test [31] using the MxComp Module of NTSYS PC software (version 2.02e) (Applied BioStatistics, Inc., Setauket, NY, USA) package.

RAPD analysis
Thirteen RAPD primers used in present study produced 54 scorable fragments out of which 15 (27.78%) were polymorphic across 96 walnut genotypes. The number of polymorphic alleles found at each RAPD locus ranged from 0 to 7 with an average of 1.15 loci per primer. The most prolific RAPD primer was OPC-04 that amplified 7 polymorphic loci. Significantly higher level of polymorphism (43%) was observed in walnut through RAPD markers by Fakhraddin [33]. Our study revealed higher number of average alleles per loci, on an average 4.15 alleles was observed per loci which are higher than previous reports of Nicese [11] and Potter [34]. Resolving power (Rp) values for RAPD primers ranged from 0 to 12 with an average of 6.99. MI value for RAPD ranged from 0 to 1.36 with an average of 0.32. The PIC values of the primer combinations in RAPD varied from 0 to 0.29 with an average of 0.07. Our average values of MI and RP are higher than those which were observed by Fatahi [35] and Ahmed [17]. The respective values for overall genetic variability for PIC, Rp and MI in all the 96 genotypes are summarized in Table 2. The RAPD derived data was subjected to calculate the genetic similarity which revealed a varying degree of genetic relationship for genotypes belonging to different geographically isolated districts. The Jaccard's similarity coefficient ranged from 0.78 to 1.00 with an average of 0.91 amongst all the 96 walnut genotypes. The lowest pair-wise similarity coefficient value was between CITH-W3 and PTS-10 (0.78). This indicates a quite high degree of genetic variability within the species which may be because of different original parents. Cluster analysis defined four main groups (Figure 1). First Cluster consisted of 26 genotypes with an average similarity of 94%. Second cluster, the major cluster comprised of 45 genotypes with an average similarity of 95%. Third and Fourth Cluster consisted of 4 and 21 genotypes respectively with an average similarity of 89% and 93% respectively. The principal coordinate analysis (PCoA) showed that the first three axis accounted for 94.59% (91.36%, 2.12% and 1.08% by 1 st , 2 nd and 3 rd co-ordinate respectively) of total variation. The grouping shown in dendrogram was at par with that shown in 3D scatter of PCoA ( Figure 2).
The number of alleles (na), the effective number of alleles (ne) [36], Nei's gene diversity (h) and Shannon's Information index (I) [37] are given in Table 4. The average effective number of alleles for all the RAPD primers was 0.76 with locus OPU-03 possessing highest number of effective alleles (1.48) followed by OPJ-04 (0.45) and OPC-04 (0.27). Similarly, Shannon's Information index ranged from 0 to 0.43 (OPU-03) with an average of 0.12 and Nei's gene diversity ranged from 0 to 0.29 (OPU-03) with an average of 0.08.

ISSR analysis
ISSR markers are helpful in fields of genetic diversity, phylogenetic studies, gene tagging, genome mapping and evolutionary biology in a wide range of plant species [6]. They have been effectively used in many tree species, including walnut (J. regia L.) [34,38], olive (Olea europaea L.) [39], mulberry (Morus L.) [40] and plum (Prunus L. spp.) [41]. The genetic variation observed through RAPD was further analyzed by ISSR to obtain a more accurate molecular characterization of the studied walnut genotypes. ISSR analysis of 19 markers produced 72 scorable fragments out of which 67 (93.05%) were polymorphic. The number of alleles observed at each ISSR locus varied from 1(p819; p847) to 9 (p811) with an average of 3.78 alleles per locus and polymorphic loci also range from 1 (p 847; p819; p830) to 9 (p811) with an average of 3.52 polymorphic loci per primer. The most prolific ISSR primer was p811 amplifying 9 polymorphic bands, while the least productive markers amplifying single monomorphic loci were p819 and p847. Hence we observe that ISSR markers have higher potential to discriminate walnut genotypes from each other due to capability of generating higher number of polymorphic loci per primer. PIC values range from 0.04 (p830) to 0.49 (p847) with an average of 0.27. MI values ranged from 0.12 (p830) to 3.77 (p847) with an average of 1.08. Primer p811 exhibited the highest Rp value (9.64) and the lowest value (1.10) was recorded for the primer p847 with an average of 4.20. Prevost and Wilkinson (1999) reported a linear relationship between the ability of a primer to distinguish genotypes and Resolving power. The percentage of polymorphic bands (PPB), the average number of polymorphic bands per primer, PIC, MI and Rp obtained in our present results were higher than those previously found in ISSR studies in walnut by Potter [34] Christopoulus [12] Mahmoodi [13] and Ji [42]. The Jaccard's similarity coefficient ranged from 0.22 to 0.85 with an average of 0.58 among all the 96 walnut genotypes used which is lower than the findings of Ji [43]. The lowest pair-wise similarity (22%) was observed between genotype HCB1 and Chinovo while as highest pair-wise similarity (85%) was observed between genotype DKT2 and BWS19. The reason for this higher similarity is that the two genotypes may have the same parents. In the present study, 94 genotypes were separated into ten distinct clusters and two genotypes (HCB1 and PSB2) do not form part of any cluster (Figure 3). First cluster consisted of 2 genotypes with an average similarity of 58.1%. Second cluster, the largest one, consisted of 25 genotypes with an average similarity of 68.1%. Third, fourth and fifth cluster consisted of 21, 13 and 2 genotypes with an average similarity of 68.6%, 62.4% and 63.5% respectively. Similarly sixth, seventh and eighth cluster was represented by 11, 6 and 2 genotypes with an average similarity of 66.4%, 66% and 62.7% respectively. Ninth and tenth cluster comprised of 10 and 2 genotypes with an average similarity of 63% and 60.5% respectively.
Furthermore individuals of two main clusters (Cluster 2 and Cluster 3) have an average similarity of 65.9%. The principal coordinate analysis (PCoA) showed that the first three axis accounted for 65.65% (59.39%, 3.34% and 2.90% by 1 st , 2 nd and 3 rd co-ordinate respectively) of total variation. The grouping shown in dendrogram was at par with that shown in 3D scatter (Figure 4). The na, ne, h and I are summarized in Table 5  showed that all the genotypes formed one major cluster and five minor clusters and one genotype PTS21 does not form the part of any cluster. Cluster I, the major cluster, comprising of 63 genotypes with 77% average similarity represented the selections from Budgam (31.7%), Ganderbal (30.15%), Pulwama (11.11%), Anantnag (6.34%), Srinagar (4.76) and Shopian (3.17%). Thus the majority of genotypes in Cluster I are from district Budgam followed by Ganderbal. Within the first cluster, two subgroups consisting of 52 and 11 genotypes were found. The second cluster comprised of only two genotypes each from Budgam and Ganderbal district with an average similarity of 78%. The third cluster comprising of eleven genotypes with an average similarity of 77% represented the genotypes from district Budgam (45.45%), Pulwama (18.18%) and 9.0% each from district Ganderbal, Baramulla, Shopian and exotic collections. Therefore the majority of genotypes in this cluster are from district Budgam. Fourth cluster consisted of 7 genotypes with an average similarity of 81%. This cluster was represented by genotypes from Budgam (42.85%) and Baramulla (42.85%). Fifth cluster consisted of 9 genotypes with an average similarity of 76%. This cluster was represented by the genotypes from district Ganderbal (33.35%), Budgam (22.2%), Anantnag (22.2%), Pulwama (11.1%) and      [Lewontin (1972). Baramulla (11.1%). Sixth cluster consisted of 3 genotypes each from district Ganderbal, Budgam and Anantnag with an average similarity of 76%. The out-crossing nature and seedling origin of walnut may be responsible for such diverse clustering. No relationship was found between the spatial and genetic proximity of the germplasm apart from few genotypes which may be originated from similar regions. Since the genotypes have seedling origin and were grown as nuts which shows high degree of segregation and thus leading to development of entirely new genotypes. Thus the genotypes showing similarity may be derived from the genotypes from the same cluster. The common practice in the region was to raise the walnut trees from nuts of genotypes having superior quality and thus leading to development of new walnut genotypes with some degree of superior nut qualities. The principal coordinate analysis (PCoA) showed that first three axis accounted for 78.53% (74.82, 2.01 and 1.69 by I st , 2 nd and 3 rd co-ordinate respectively) of total genetic variance. The grouping shown in dendrogram was at par with that shown in 3D scatter.

RAPD and ISSR combination
Two way Mantel test [31] was done between the RAPD and ISSR data matrices to compare the goodness of the fit of each similarity matrix. The correlation coefficient (r) was found to be 0.11 between matrices generated by RAPD and ISSR markers using the Mantel Test (t=2.39, p=0.99) supporting the validity of the dendrogram.
Both the marker techniques of RAPD and ISSR proved to be extremely effective in discriminating the 96 walnut genotypes analyzed, since the majority of bands were polymorphic amongst genotypes. The number and percent of polymorphic loci, polymorphic information content (PIC), and marker index (MI) of primers were higher for ISSR markers. In this study, we found a high degree of polymorphism (93.05%) in the 96 genotypes studied. This validates that ISSR markers are useful markers in genetic divergence studies allowing an unequivocal identification for genotypes. The high level of polymorphism possibly reflects the outcrossing character of walnut because almost similar results have been obtained in fruit and nut tree species like pistachio [42] or olive [44].

Conclusion
Present study provides the base for association mapping studies and linkage map development in walnut. Since the walnut population we studied is heterogeneous with respect to number of traits and those traits can be linked with markers through association mapping studies. Present study clearly identified the markers with potential for utilization in walnut breeding programs.