DNA fingerprinting, fixation-index (Fst), and admixture mapping of selected Bambara groundnut (Vigna subterranea [L.] Verdc.) accessions using ISSR markers system

As a new crop in Malaysia, forty-four Bambara groundnut (Vigna subterranea L. verdc.) genotypes were sampled from eleven distinct populations of different origins to explore the genetic structure, genetic inconsistency, and fixation index. The Bambara groundnut, an African underutilized legume, has the capacity to boost food and nutrition security while simultaneously addressing environmental sustainability, food availability, and economic inequalities. A set of 32 ISSRs were screened out of 96 primers based on very sharp, clear, and reproducible bands which detected a total of 510 loci with an average of 97.64% polymorphism. The average calculated value of PIC = 0.243, RP = 5.30, H = 0.285, and MI = 0.675 representing the efficiency of primer set for genetic differentiation among the genotypes. The ISSR primers revealed the number of alleles (Na = 1.97), the effective number of alleles (Ne = 1.38), Nei's genetic diversity (h = 0.248), and a moderate level of gene flow (Nm = 2.26) across the genotypes studied. The estimated Shannon’s information index (I = 0.395) indicates a high level of genetic variation exists among the accessions. Based on Nei’s genetic dissimilarity a UPMGA phylogenetic tree was constructed and grouped the entire genotypes into 3 major clusters and 6 subclusters. PCA analysis revealed that first principal component extracted maximum variation (PC1 = 13.92%) than second principal component (PC2 = 12.59%). Bayesian model-based STRUCTURE analysis assembled the genotypes into 3 (best ΔK = 3) genetic groups. The fixation-index (Fst) analysis narrated a very great genetic diversity (Fst = 0.19 to 0.40) exists within the accessions of these 3 clusters. This investigation specifies the effectiveness of the ISSR primers system for the molecular portrayal of V. subterranea genotypes that could be used for genetic diversity valuation, detection, and tagging of potential genotypes with quick, precise, and authentic measures for this crop improvement through effective breeding schemes.

PCR amplification. Initially a total of 96 ISSR primers from "Integrated DNA Technologies Inc. ", Singapore was screened, of which 32 sets of primers were preferred based on their efficacy to detect clear and sharply distinct polymorphic bands across all the 44 Bambara groundnut genotypes (Table 1). Selected 32 ISSR primers used in this study and their properties (sequences, base pair, GC content, melting temperature, and annealing temperature can be found as "Supplementary Table S1" online. PCR reaction mixture contained "2 × Power Taq PCR Master Mix" is 2 × concentration mixture of DNA polymerase, buffer, and dNTP mixture, MgCl, and Taq (BIOTEKE Corporation), primer, template DNA, and nuclease-free water. PCR was performed in T100 Thermal Cycler from BIO-RAD (Hercules, California 94547, USA). The DNA amplification mixture of 25 μL contained: Master mix 12.5 μL, Nuclease free water 8.5 μL, DNA template (40 ng), 2.0 μL, and primer 2.0 μL. The PCR amplification program was carried out using the following conditions of an initial denaturation at 95 °C for 3 min followed by 35 cycles of denaturation at 94 °C for 45 s, annealing at the primer specific temperature for 1 min, extension at 72 °C for 2 min and final extension was adjusted at 72 °C for 10 min and followed by saturated at 4 °C. The amplified products were separated on 1.5% (w/v) agarose gel with 1 × TBE buffer and stained with "GREEN VIEW Nucleic Acid Gel Stain (5.0 μl/100 ml)" by electrophoresis at 80 V for 75 min adjusted with 400A using horizontal gel electrophoresis system (Bio-Rad Wide Mini Sub-Cell GT and Bio-Rad Sub-Cell GT). The gels were photographed and developed under UV light using the Gel Doc XR + documentation system (BIO-RAD, Hercules, CA, USA). We repeated PCR reaction twice with each primer to ensure the primer ISSR marker's reliability and reproducibility and discarded the primers that showed weakly and no banding pattern. The size of amplified fragments was measured by running a 100 bp DNA ladder (BIOTEKE Corporation) in the gel as a standard size marker.
Scoring ISSR band. A set of 32 ISSR primers produce a sharp and reproducible band during DNA fingerprinting of 44 Bambara groundnut accessions. Polymorphic ISSR bands were scored by using the UVIDOC software version 99.02 on the manually detecting method for the actual band sizing based on the standard weight of the DNA 100 bp Ladder. The electrophoretic profiles were coded according to the present visible and repeatable bands on the gel-electrophoresis map as "1" and absent of band at the same loci were coded as "0". www.nature.com/scientificreports/ Statistical analysis. Genetic diversity and frequency analysis. For primer data analysis multiple software was used based on the accounted band profiles. Only repeatable, distinct, and well-resolved fragments were coded as presence (1) or absence (0) for each marker and presented as part of a binary matrix. POPGENE version 1.32 37 was used to calculate genetic diversity for each population such as percent polymorphic bands (PPB), observed number of alleles (Na), the effective number of alleles (Ne), Nei's genetic diversity (H), Shannon's information index (I). To analyse the genetic diversity in segmented populations, we estimated the total genetic diversity (Ht), genetic diversity within a population (Hs), Nei To measure the gene frequencies as well as genetic divergence between individuals of Bambara groundnut accessions were also investigated using Nie's unbiased genetic distances matrix and genetic identities 39 performing by GENALEX 6.5 software 40 . The analysis of PCA for the 44 V. subterranea was carried out using the same data of ISSR primers. The graphical representation based on Euclidian measures of PCA was revealed by NTSYS PC ver. 2.02; PCA biplot was generated using JMP version 16.0 from SAS program, PCA 3D, and pie chart for graphical visualization of eigenvalues and variation ratio for all PCs were illustrated by NCSS 2021 software. Moreover, the scatter matrix with density and box plot for correlation regression study among marker efficiency index (MI, PIC, RP, EMR, H) was visualized by NCSS 2021 software.
Genetic relationship analysis. Clustering was performed to determine the relative genetic distance between individuals and to check the consistency of population genetic differentiation. The Nei's unbiased genetic distance was used to construct a dendrogram or phylogenetic tree for the 11 Bambara groundnut population using UPGMA (Unweighted Pair Group Arithmetic Mean) method in POPGENE program version 1.32 followed by MEGA (Molecular Evolutionary Genetics Analysis) version 6.10 for Windows reported by Tamura et al. 41 and Nilkanta et al. 42 .
Marker efficiency analysis. The performance of the primers was measured by calculating different parameters including polymorphic Information Content (PIC), Resolving Power (RP), and Discriminating Power (DP), expected heterozygosity (H), and arithmetic mean heterozygosity (H avp ) for each primer by program iMEC (https:// irsco pe. shiny apps. io/ iMEC/) 43 . This program calculates PIC using (Botstein et al. (1980) formula PIC = 1 -Σ p i 2 -Σ Σ p i 2 p j 2 where p i and p j are the population frequency of the ith and jth allele. The first summation is over the total number of alleles, whereas the two subsequent summations denote all the i and j where i = j. EMR was calculated using Powell et al. 44 formula EMR = n β, where n is the average number of fragments amplified by an individual to a specific system marker (multiplex ratio) and β is estimated from the number of polymorphic loci (np) and the number of non-polymorphic loci (nnp); β = n p /(n p + n np ). The resolving power (RP) of each primer was calculated as Prevost and Wilkinson 45 formula; R = Σ I b , where I b represents the informative fragments. where Ib or band informativeness is represented on a scale of 0-1 and is defined as Ib = 1 -(2 ×|0.5 -p|), where pi is the proportion of individuals containing the ith band. Discriminating Power (DP) estimated by Tessier et al. 46 as D = 1 -C; where C is the confusion probability is C = Σ ci = Σ p i N pi − 1 /N − 1 where for N individuals, C is equal to the sum of all ci for all of the patterns generated by the primer. Expected heterozygosity as per formula of H = 1 -Σ p i, where pi is the allele frequency for the ith allele, and the summation is over all available alleles. Arithmetic means heterozygosity (H avp ) the formula given by Powell et al. 44 is H avp = Σ Hn/np, where Hn is the heterozygosity of the polymorphic fraction of markers and the summation is over all the polymorphic loci np. To characterize the capacity of each primer to detect polymorphic loci among the genotypes, we also calculated de Marker Index (MI) for each primer as a product of PIC and EMR 47 .
Genetic structure and admixture analysis. To infer the profile of the population structure and admixture detection, a Bayesian model clustering algorithm by STRU CTU RE version 2.3.4 48 was performed based on ISSRs binary data sets of 44 BG genotypes. Before performing the structure analysis, 44 genotypes were categorized into 1 to 11 distinct population groups (i.e., each population the group comprised of 4 individuals genotype) ( Table 1). The Bayesian admixture analysis is one of the most perfect approaches for diploids and polyploids 49 to sense the patterns of population genetic structure using dominant markers due to it does not assume prior information of inbreeding and Hardy-Weinberg equilibrium even can be executed with a comparatively low population and loci 50 . No pre-data on population origin is necessary to determine the most likely number of populations (K) under the admixture model and correlated allele frequencies 51 . To estimate the best genetic unit, K value, a burn-in period of 5.0 × 10 4 followed by 1.0 × 10 6 m Markov Chain Monte Carlo (MCMC) simulations at 4 iterations 52 with ten autonomous runs were performed with K value was pre-set from 1 to 10 53 . For this purposes the output files of structure analysis were squeezed into a single "Zip-Rar" file then upload online "Structure Harvester 0.6.93 version" (http:// taylo r0. biolo gy. ucla. edu/ struc ture) 54 to determine the average Loglikelihood, Ln P(D), probability by K-graph, the most provable K value using ΔK method by Evanno et al. 53 and Q value (standard Q > 0.60 < Q admixture) showing membership coefficient (%) value. This value assigned accessions to a certain population, finally allocate the accessions into a specific cluster based on the maximum (K) likelihood value was used 49 . The bar plot for best K was documented by STRU CTU RE software: Version 2 48 . Moreover, STRU CTU RE software: Version 2 also used for calculating the fixation index (Fst), is the proportional increases of homozygosity. However, the value of the Fst index of a group of the population can range from "0" (no different) to "1" (completely different) i.e., no alleles held in common 55

Results
Polymorphism quantification by ISSR primers. The gel image is taken from each primer, based on the result, the gained DNA fingerprinting pattern was very distinct and repeatable ( Fig. 1) although detecting the banding pattern and its clarification of specific gel images is always challenging. The range of the amplified band was noted from 100 to 1580 bp (  Genetic distance (Nei's measure) analysis. Table 4 disclosed the genetic distances (GD) among 44 V.
Genetic relationship. The genetic relationship of 44 V. subterranea accessions was attained from ISSR primers scoring data set using Nei's (original) genetic dissimilarity coefficient. The magnitude of relatedness and disparity among the accessions are demonstrated in (Fig. 4 and Table 4). The clustering pattern depicted the accessions into a phylogenetic tree or dendrogram displayed the existence of significant genetic divergence among the evaluated genotypes. The branching pattern of a phylogenetic tree is its topology in which each branch is a line connecting either two internal nodes to each other or an external node to an internal node and the length of a branch denotes the genetic distance. The accessions were expediently grouped into six definite sub-cluster (node remark with yellow-colored circular sign) under three distinct major clusters (node remarked with red-colored diamond shape sign), collected from different geographical origin. The largest among the three major clusters Table 3. Efficacy of primer polymorphism calculated with iMEC of Bambara groundnut genotypes. SB scored bands, PB polymorphic bands, D discriminating power, Emr effective multiplex ratio, H expected heterozygosity, Havp mean heterozygosity, MI marker index, PIC polymorphism information content, Rp resolving power. * and **Both (UBC 835) name is same as per the source of the collection, but the sequences are not same and collected from two different sources. www.nature.com/scientificreports/  Heatmap analysis. Based on Euclidian cluster distance and Ward (unsquared distances) linkage clustering method using ISSR data set illustrated three distinct groups of 44 Bambara groundnut accessions (Fig. 5). The genetic relationship study among the accessions revealed by Nei's distance generated clustering pattern of three major groups which resemble the clustering pattern developed by heatmap analysis. Hence, ISSR linked current research leads to investigating the genetic relatedness among the accessions and identifying the actual genetic distance to avoid any pseudo-diversity. In horizontal dendrogram (rows) represent the accessions and the vertical dendrogram (column) represent the ISSR loci. The red and blue square plots of the heatmap indicate the presence (1) and absence (0) of loci in each accession, respectively. Both rows and columns are clustered using Euclidean distance and Ward (unsquared distances) linkage. Zimisuhara et al. 51 reported four clusters based on heatmap cluster analysis using ISSR binary data in Ficus deltoidei Jack.
Ordination: principal component (PCA) analysis. Ordination is a collective term for multivariate analysis which adapts a multidimensional group of data in such a way that the similar species or samples are plotted close together while the dissimilar one has placed far apart 56 also known as multivariate gradient analysis. PCA is used for similarities which starts with the binary data matrix (e.g., presence versus absence of alleles in molecular marker data). When there are no missing data, the output of PCA and PCoA will be similar 57 . To lead the clustering investigation eigenvalues and total percentages of principal component case scores were used. The graphical distribution of eigenvalues, percent of genetic variation, and cumulative percent of genetic variation based on all axes (PCs) were displayed by pie chart in Fig. 6. The first three principal components covered 31.42% (PCA) of cumulative variation (Table 5) and which is accounted for greater than the total variation exposed in the populations. However, 51.12% total variation was captured by 1 st nine principal components as shown in Table 5. The PCA analysis revealed that first three principal components captured PC1 = 13.92%, PC2 = 12.59% and PC3 = 4.91% of total variation. Moreover, the PCA analysis has 44 principal components (PCs) out of which the first 25 PCs and 10 PCs contributed 80% and 53% of the total variation (Fig. 6). In the case of PCA analysis from the principal component one (PC1), the highest value was 0.25 for the accessions (G28, G29, G35, G38) followed by 0.24 for the accessions (G30, G31, and G37) while the least values (0.00) were found for the accessions G10 and G19 which have no contribution to total diversity (Table 5). Furthermore, in PC1, most of the accessions contributed positively toward the variation of one group than another except accessions G12, G13, G14, G15, and G20 bearing negative values. In PC2, 17 accessions and 22 accessions in PC3 had a positive contribution to diversity (Table 5). Two dimensional (2D: Fig. 7A) and three dimensional (3D: Fig. 7B) visual illustration of PCA analysis exposed that the entire accessions were distinctly grouped into three genetic components based on Euclidian distance which is the evidence of findings of clustering pattern analysis. In the PCA plot, we observed that within and among the accessions genetically associated genotypes were placed closer to each other while the distant genotypes were positioned far apart. Most of the accessions exhibited similar values of Shannon diversity (Hˊ indices) with a range from 1.86 to 2.01. The highest value was 2.01 for the accessions (G4 and G5) afterward 2.0 for accession (G10) while the least was 1.86 recoded for accessions G22 and G44 (Table 5).
Biplot analysis. The biplot-based representation of PCA showed the association of evaluated 44 accessions of Bambara groundnut along with ISSR loci loaded in the same plot. The principal component analysis assembled the total tested accessions in a diverse group based on the ISSR data set. Using the JMP version 16.0 analytical tools from the SAS program, we generated PCA sample (accessions) loading (Fig. 7C)  Admixture analysis. The structure is a population analysis tool used to assess the patterns of genetic structure from a set of samples. To identify subsets of the whole sample by detecting allele frequency differences within the data and can assign individuals to those sub-populations based on analysis of likelihoods. The structure uses data from individuals in a population to identify allele frequency differences. The genetic structure of accessions was estimated based on Bayesian (theorem) clustering analysis using the STRU CTU RE program of Evanno et al. 53 method followed by Structure harvester. The structure analysis of V. subterranea accessions was initially performed based on the maximum number of (K = 1 to 10) as the original population order displayed in Fig. 8. However, the most probable value of population was calculated to the maximum peak at ΔK = 3 (K value = 104.97; Lnprob (K) = − 8053.2) (Fig. 9B) with a rate of change of the likelihood distribution (mean) (Fig. 9C); Absolute value of the second order rate of change of the likelihood distribution (mean) (Fig. 9D) and mean of estimated Ln probability in Fig. 9E. Based on best K = 3 determining that all the evaluated accessions might be positioned into three major clusters visualized with three distinct colors of red, yellow, and purple (Fig. 9A). Determination of delta K is an ad hoc quantity related to the second-order rate of change of the log-likelihood of data related to the number of clusters 51 . Nevertheless, regarding the membership likelihood G23  www.nature.com/scientificreports/ (Q) > 0.60, some of the accessions showed unique standards which lead to the pure population while Q < 0.60 regarded as the admixture populations 49 . In the bar plot (Fig. 9A), 22 accessions under the red color zone were recorded as highly pure ones whereas the other 6 accessions and 16 accessions were assembled in the yellow and purple color zone, respectively represented as admixture units. Based on Q > 0.60 as the purity standard, in the yellow zone, 6 accessions (G40, G39, G37, G43, G41, and G44) declared as pure ones though G43, G41, and G44 received genetic material from the population of the red zone. Moreover, out of 16 accessions of purple color zone 9 accessions (G42, G38, G36, G35, G30, G34, G33, G31, and G32) were noted as admixture units based on Q < 0.60 (Fig. 9A).

Fixation index (F ST values) analysis.
Measuring gene flow (also called gene migration or allele flow) can further be accelerated by the estimation of Fst (also known as Fixation index) 58 . Genetic differentiation among the accessions due to genetic structure is measured by the fixation index (Fst) using genetic polymorphism data. It is one of the most frequently applied statistics in explaining the population genetic structure 59 . Using the STRU CTU RE program, the fixation index can be measured. Considering the best delta K value (ΔK = 3), resultant the entire genetic component is grouped into three clusters. The Structure output can be displayed as a "triangle plot" in which two clusters were plotted at two vertices and all others were at the third. In the triangle plot (Fig. 10D), individuals are represented by a colored dot that corresponds to the different populations. Individuals who are in the corners are assigned to one population or another. The distance between each cluster was shown in Table 6 and the maximum distance was recorded for cluster 1 vs cluster 2 (0.066) followed by cluster 1 vs cluster 3 (0.065), whereas cluster 2 and 3 (0.05) were closely associated with each other displayed by the tree plot (Fig. 10C). Average distances (expected heterozygosity) between individuals in the same cluster were recorded highest for cluster 1 (0.2478). The graphical distribution of the Fst means the value of three clusters was displayed in Fig. 10B. The estimated mean Fst value (Table 6)

Discussion
Polymorphism quantification by ISSR primers. The use of a molecular marker is a very common phenomenon to investigate the population structure and genetic diversity as well as distinguishing one genotype from another as a prerequisite for pre-breeding and breeding of crops improvement. Molecular genetic diversity is very crucial as it gives a greater precise measure of polymorphism related to morphological characterizations. Most of the markers showed complete polymorphism suggesting the efficacy of these ISSR markers for the assessment of genetic variation among the V. subterranea species. Usually, the efficiency of a certain primer for evaluating the population genetic structure is extremely subjected to the level of polymorphism that could be generated among the accessions. Our results represented a moderate to a higher level of genetic diversity among the studied accessions. A similar trend of diversity is very common as self-pollinated members of Bambara groundnut from the genus Vigna, recommend its medium genetic base, which is perhaps assembling of novel gene incorporation due to dynamic forces of natural selection. In this current study, the magnitude, and pattern of genetic variation within 44 V. subterranea accessions using 32 ISSR primers exposed the availability of polymorphism. We detected a total of 510 DNA fragments with an average of 15.93 bands per prime. In the present study, comparatively high percentages of polymorphism (97.64%) were observed using the ISSR primer serve as a high potential tool for genetic discrimination among the closely related V. subterranea species. Relative studies in vigna species particularly in V. subterranea based on RAPD, AFLP, SSR, DArT array, and ISSR primer systems were effectively used and reported by researchers (Massawe et al. 16  Primer efficiency analysis. According to Amiryousefi et al. 43 , there are two major dimensions of genomic marker polymorphism excellency and informativeness as heterozygosity (H) and the polymorphic information content (PIC). These indices were measured based on data gained from ISSR primers using the iMEC (online marker efficiency calculator). The range of H and PIC value for a binary or dominant marker is maximum as 0 (monomorphic) to 0.5 (highly judicial, with multiple alleles in an identical frequency) due to assume of two alleles per locus and both are influenced by the number and frequency of alleles 61 . Estimation of PIC value delivers a projection of discriminatory power of a locus by considering not only alleles numbers but also the relative frequencies of those alleles 62  www.nature.com/scientificreports/ PIC index indicated better sources of variation that will assist plant breeders to assess genetic diversity and inter or intra relationships among genotypes. Resolving power (RP) is an index of the separating ability of a certain marker and an effective multiplex ratio (EMR) is a matrix which highly depends on the polymorphic extent of markers. Considering the range High PIC and H values indicate the advanced discriminatory capacity of both marker systems. MI highlights the distinctive power of the primer. A higher value of Discriminating (D) power (closer to 1) indicates a lower possibility of a mix-up between V. subterranea accessions 43 . There was a positive correlation was observed among PIC, RP, MI, and EMR which is supported by Kayis et al. 63 and Ramzan et al. 64 . Most of the ISSR primers were highly polymorphic and informative, suggested for genetic discrimination analysis of this studied genus. The mean of primer efficiency index was comparatively high (RP = 5.30, MI 0.675, D = 0.96) and this matrix indicates the overall efficacy of the tested primers which provides exact differentiation among the accessions. The greater the RP and MI indices refer to the greater efficacy of the respective primer 65 17 reported genetic distance varied from GD = 0.10 to 0.68 for 100 Bambara genotypes using AFLP markers. There was high similarity covered from 0.83 to 0.94 recorded for 12 Bambara groundnut genotypes using RAPD by Fatimah and Ardiarini 60 . The accessions with high relatedness from two different geographical zones suggesting the involved genotypes may have the common origin and/or mechanical mixture of seeds from one agro-ecological zone to another across Nigeria. The genotypes with less relatedness indicating the presence of extreme divergence among the evaluated accessions. Typically, the genetic diversity of a population in a species is influenced by several evolutionary factors, such as geographic distance, natural selection, reproductive system, gene flow, seed dispersal as well as the center of diversity 67 . However, a significant extent of genetic diversity is predictable in V. subterranean accessions due to geographic dispersal of the genus but differentiation among the collected germplasm is inadequate at growing areas due to mixing of germplasms within the regions and the fact is that farmers either produce their seeds or collected seeds from unauthorized ways. As a result, the existence of close relatedness was noted among some of the accessions used in this study due to the accessions collected from similar locations or origins or names of different landraces.
Genetic relationship. Based on the ISSR banding profile we discovered three major clusters with six subclusters of the accessions evaluated in this study. The findings in our study illustrated the efficacy of ISSR markers partitions the accessions into closely related genetic groups than another marker system. The major cluster I and II occupied the maximum number of accessions whereas the smallest cluster was major cluster III which had only three genotypes from different origins. The accessions that positioned the same cluster with different regions of collection, reflecting a close genetic association despite their diverse origins. Our finding was validated by Mohammed et al. 3 stated the seven clusters of 50 Bambara groundnut species and Odongo et al. 22 reported three clusters of 105 Bambara groundnut genotypes using SSR primers. Conversely, the genomic grouping of Bambara groundnut accessions related to geographical distribution based on RAPDs and AFLP reported by Amadou et al. 13 and Ntundu et al. 17 , respectively. Generally, the genotypes were positioned more closely in our generated phylogenetic tree suggesting that they were genetically more similar having identical genes. On the other hand, accessions that possess the distant group suggesting that they were genetically dissimilar even though they come from the same population as well as similar origins. This circumstance may prompt by some factors such as the mixture of seeds, mating system, natural selection, spontaneous mutation, additionally the local farmer produces their seeds or purchase from neighboring markets. A similar trend of the result was proposed in their study of Bambara groundnut by Massawe et al. 14 using RAPD, Somta et al. 19 using SSR, Mukakalisa et al. 15  www.nature.com/scientificreports/ dimensional space, e.g., as a 2D or 3D graphic. PCA attempts to discover the principal axes through a matrix of eigenanalysis and eigenvectors. Eigenvalues are usually ranked from the greatest to the least. The first eigenvalue is often called the "dominant" or "leading" eigenvalue. Eigenvalues are also often called "latent values". The values recorded in principal axes of PCA arebeing advocated in the similar trends of result noted by Arolu et al. 69 . In our study, the PCA analysis showed maximum variation captured by PC1 (13.92%) and PC2 (12.59%) which is supported by Rungnoi et al. 1 and stated 90.3% variation led by the first three PCs using ISSR and RAPD in PCoA analysis of V. subterranea; Odongo et al. 22 zanalyzed PCoA and concluded that 84.30% of total variability spanned by the first three PCs using SSR primer in V. subterranea which is higher than our findings; Molosiwa et al. 21    www.nature.com/scientificreports/    74 . Our estimated Fst was 0.1896, 0.3684, and 0.3997 for cluster 1, cluster 2, and cluster 3, respectively. Considering the above scale, suggested that the population under cluster 1 showed great genetic differentiation whereas the population under cluster 2 and 3 showed very great genetic diversity. Frequent gene flow led to a low level of genetic differentiation with a small genetic distance among them. Oppositely, low gene flow governs the plant's adaptation to different growing regions influencing the higher level of genetic differentiation with greater genetic distance. Genetic enhancement of crops depends on the extent of genetic differentiation among accessions. The currently estimated fixation index using ISSR is higher than the report published by Kumar et al. 75   www.nature.com/scientificreports/ Figure 8. The population membership of the studied Bambara groundnut species group for a priori distinct number of K = 1-10 inferred by the STRU CTU RE software (PRITCHARD LAB, CA, USA). Each accession is signified by a vertical column divided into colored segments that represent the individual's estimated association fractions in K clusters and black vertical lines isolated the 44 accessions. The numeric number beneath the adjacent bar graph indicates the accession IDs and source of sampled population (11) code in parenthesis that is mentioned as in Table 1.

Conclusion
This is the forerunner initiative on the valuation of genetic differentiation and population structure of V. subterranea genotypes using ISSR primers in Malaysia. Genetic relatedness and population structure are crucial for plant breeding schemes for this crop improvement as well as its conservation. Considering this intent, to conduct this study, ISSR primer was used and exhibited a moderate to high level of efficiency in assessing genetic differentiation and genetic structure in V. subterranea populations. The amplification of many polymorphic loci indicated the used set of ISSR primers have the potential to the assessment of genetic diversity among the existing accessions. However, the combination and a large number of molecular markers (dominant and co-dominant) to further assessment of genetic variation is highly advocated. In terms of diversity indices and genetic relationships, a significant proportion of variation was accounted for among the evaluated accessions and the diverse genotypes are suggested to use in a breeding system. Oppositely, the genotypes with low average diversity indicated the potential risk of declining genetic variation due to limited genetic basis, which alarming or enlightening the implication of biodiversity, assembling, and conserving their wild genetic resources. Moreover, the Structure, PCA, UPMGA, and Nei's analysis divulged the entire accessions into three distinct genetic components based  www.nature.com/scientificreports/ on ISSR amplified genomic data sets. Furthermore, fixation index (Fst) and genetic structure with admixture analysis revealed the persistent genetic drift among the gene pool of V. subterranea accessions. Typically, this investigation provides an initial scientific basis of genetic data for this crop enhancement and conservation policies in the future. The result of this study will assist in more accurate portrayal, classification, preservation, and maximum utilization of genetic resources and may have real implications in future breeding schemes to broaden the genetic diversity of V. subterranea species.

Data availability
All data are available in the text body of the manuscript. We also confirm that, a voucher specimen of the identified species has been deposited in a publicly available herbarium and GenBank, ITAFoS, Universiti Putra Malaysia (UPM). The deposition number-Bambara groundnut (Vigna subterranea) /ITAFoS/UPM/S4-2020.  Table 6. Allele-freq. divergence among pops (Net nucleotide distance), computed using point estimates of P through structure analysis. a Average distances (expected heterozygosity) between individuals in same cluster.