Assessment of genetic diversity in cowpea (Vigna unguiculata L.) germplasm using morphological and molecular characterisation

Cowpea (Vigna unguiculata L.) is the most important grain legume crop grown in tropical and subtropical regions. Cowpea grain has a high nutritional value containing high amount of protein (23–29%). A total of 32 cowpea genotypes were selected for characterization at molecular and morphological markers under normal irrigation and drought stress conditions separately, as an assisting tool for a reliable varietal selection in breeding programs. In this study 17 morphological characters and multivariable statistical methods were studied and followed by using a set of 22 Simple Sequence Repeat (SSR) primer pairs for molecular characterizations. The analysis of variance for morphological traits revealed significant differences among accessions for all measured traits. In molecular (SSR) analysis, a total of 186 alleles were detected with an average of 2 alleles for each locus, and genetic distance between genotypes was estimated 0.0066. The average genetic distance based on Nei’s index among genotypes was 0.116, and the polymorphism information content value for SSR loci varied from 0.625 for primer Vm5 to 0.25 for primer Vm25 with an average of 0.445. Results of factor analysis determined 5 and 6 factor in drought stress and normal irrigation condition explaining 81.17 and 88.20% of the total variation respectively. The average genetic similarity observed across all the genotypes was 75.8%. *Corresponding author: Khosro Mafakheri, Plant Breeding and Biotechnology Department, University of Tabriz, Tabriz, Islamic Republic of Iran E-mail: Kh.mafakheri@tabrizu.ac.ir Reviewing editor: Manuel Tejada Moral, University of Seville, Spain Additional information is available at the end of the article ABOUT THE AUTHORS Khosro Mafakheri is currently PhD candidate of Plant Breeding at the Plant Breeding and Biotechnology Department, Agriculture faculty of the University of Tabriz, Iran. His research area is on the study of the reaction of plant spp. to biotic and (a) biotic stresses. Mohammad Reza Bihamta is a professor and lecturer in the Department of Agronomy and Plant Breeding in the College of Agriculture and Natural Resources, University of Tehran, Karaj, Iran. His research area is on evaluating and breeding of crop plants under biotic and abiotic stress. Her research interest is in modification and breeding new varieties of crops. Ali Reza Abbasi is an associated professor and lecturer in the Department of Agronomy and Plant Breeding in the College of Agriculture and Natural Resources, University of Tehran, Karaj, Iran. He is mainly working on genetic diversity of crop plants using molecular markers. Her research interest is in plant biotechnology. PUBLIC INTEREST STATEMENT Cowpea is the most important crop which is grown in tropical and subtropical regions. In this study, molecular and morphological characterizations of 32 cowpea genotypes were studied under drought stress condition. The indicator criterions associated with drought stress such as biological yield, 100 weight grain, number of seeds per pod, pods per plant and harvest index were examined under field conditions. The results showed that the genotypes of 7, 210, 291, and 313 along with the controls (Mashhad and Parasto) were identified as tolerant genotypes with the highest yield. Molecular analysis using 22 SSR primer pairs detected a total of 186 alleles and a genetic distance of 0.0066 between genotypes. The highest gene flow and PIC were detected for primers of Vm68 (14.88) and Vm25 (0.625) respectively. Analysis showed that the primers related to economic and biological yield in normal and drought conditions were Vm70, Vm33 and Vm3, Vm26 respectively. Received: 02 March 2017 Accepted: 28 April 2017 First Published: 11 May 2017 © 2017 The Author(s). This open access article is distributed under a Creative Commons Attribution (CC-BY) 4.0 license. Page 2 of 20 Khosro Mafakheri


PUBLIC INTEREST STATEMENT
Cowpea is the most important crop which is grown in tropical and subtropical regions. In this study, molecular and morphological characterizations of 32 cowpea genotypes were studied under drought stress condition. The indicator criterions associated with drought stress such as biological yield, 100 weight grain, number of seeds per pod, pods per plant and harvest index were examined under field conditions. The results showed that the genotypes of 7, 210, 291, and 313 along with the controls (Mashhad and Parasto) were identified as tolerant genotypes with the highest yield. Molecular analysis using 22 SSR primer pairs detected a total of 186 alleles and a genetic distance of 0.0066 between genotypes. The highest gene flow and PIC were detected for primers of Vm68 (14.88) and Vm25 (0.625) respectively. Analysis showed that the primers related to economic and biological yield in normal and drought conditions were Vm70, Vm33 and Vm3, Vm26 respectively.

Introduction
Cowpea (Vigna unguiculata L. Walp), a member of the family Fabaceae is mainly grown in the tropical and sub-tropical areas including Africa, Asia, South America, and part of Southern Europe and United States (Singh, Chambliss, & Sharma, 1997). Total production and cultivated area of legumes have been estimated 69 million tons and over 78.5 million hectare worldwide (Faostat, 2013(Faostat, , www. faostat.fao.org.2013. Cowpea is used as a vegetable crop and green or dry fodder. It is also used as green manure, a nitrogen-fixing or soil erosion controlling crop (Davis et al., 1991). Dry seeds of cowpea contain 20-25% Protein, 1.8% Fat, 60.3% Carbohydrate and are rich sources of iron and calcium (Majnoon Hoseini, 2008). Moreover, atmospheric nitrogen fixing ability is extremely valuable when it is cultivated with cereal crops in crop rotation system (Timko, Ehlers, & Roberts, 2007). Cowpea crop increases soil nitrogen up to 40-80 kg per hectare (Quin, 1997).
Drought is one of the most important abiotic stress factors which leads to reduce in crop yield and potential grain production in cowpea (Bruce, Edmeades, & Barker, 2002), and also can affect growth and productivity of most agricultural plants (Aslam, Khan, Saleem, & Ali, 2006). Plant growth under the effect of environmental conditions may be divided into enforced damage effects caused by the environment (stress) and adaptive responses controlled by the plant (resistance) (Fitter & Hay, 1987). Drought stress or imbalance between water supply and demand is not only limited to arid or semi-arid areas, but sometimes irregular distribution of rainfall in temperate regions causes unfavorable conditions for plant breeding and improvement which results in significant decrease in plant yield (Kumar & Singh, 1998).
Genetic diversity is the most important factor limiting the average number of alleles identified per SSR locus during screening program (Legesse, Myburg, Pixley, & Botha, 2007).
Assessment of local and regional plant genotypes is important for identifying diversities among germplasm which can help breeder to improve some local varieties and solving production constraints.
Accordingly, as main objectives of this study were to investigate the extent of genetic diversity and relationship between cowpea genotypes belonging to Iranian germplasms and assess the significant correlation among distances approximated based on morphological character and molecular (SSR) markers.

Germplasms
The 32 cowpea genotypes (Vigna unguiculata L.) were collected from cowpea germplasm gene bank, department of agronomy and plant breeding, University of Tehran, Karaj, Iran. Codes, names and country of origin for each accession are shown in Table 1.

Field and greenhouse assessment
Field trial (located at 1,112.5 meters above sea level, latitude: 35°-56′N and longitude: 50°-58′E with mean annual rainfall of 242 mm) was conducted at Experimental Research Farm of University of Tehran, Iran and Greenhouse trial at College of Agriculture and Natural Resources in Karaj, Tehran, Iran in 2014. All cowpea genotypes selected were sown in two separate experiments including normal irrigation and water stress conditions with a randomized complete block design in three replications . Water stress treatment was applied at 6th leaf stage, when the risk of seedlings decline is reduced. Irrigation intervals were employed once each two weeks for water stress conditions and once each one week for normal irrigation.

Morphological study
From each accession, five plants were randomly selected to record morphological traits. Morphological characterizations consisted of 17 traits including economical yield, biological yield, harvest index, weight of 100 grains, number of grain per pod, number of pod per plant, grain length, grain width, grain thickness, plant height, date to 50% germination, date to 50% flowering, date to 50% pod emerging, date to 50% maturing, pod length, pod thickness, and pod width according to set standards for morphological characters recommended by IPGRI descriptor (Darwin, Brungart, & Simpson, 2003).

DNA extraction
In order to in order fresh leaf samples to extract high quality, seeds of 32 genotypes selected were initially sown in small pots in greenhouse and fourteen-day-old leaves from plants were then sampled, frozen immediately in liquid nitrogen and stored in −80 C until DNA extraction. DNA was extracted from three young leaves at 4 to 5 leaves stage using CTAB method according to Doyle and Doyle (1990). Quality and quantity of extracted DNA were checked using agarose gel (1%) and spectrophotometer respectively suitable dilutions were then prepared (50 ng μl −1 ). In polymerase chain reaction assay, a set of 22 primer pairs of SSR markers were selected and used in this study according to previous reports (Diouf & Hilu, 2005;Fathi, 2010;Hong, Schroth, Matthews, Yau, & Bradbury, 1993;Li et al., 2001;Nei, 1973) (Table 2). PCR was performed using a Bio-Rad thermo cycler device (Bio-Rad Laboratories Inc., Hercules, CA, USA) in a final volume of 15 μl containing 2 μl DNA (50 ng μl −1 ), 1.5 μl buffer DNA extraction (10x) pH 8.3, 1.2 mM Mgcl 2 (25 mM), 1 mM of dNTPs (10 mM), 1 mM of forward primer (10 pico-mols μl −1 ), 1 mM reverse primer (10 pico-mols μl −1 ), and 2.0 μM (5 U/μl) Taq DNA polymerase enzyme. Thermal cycles were started with an initial denaturation of 94°C for 4 min, followed by 35 cycles of 40 s at 94°C, 40 s at different annealing temperature corresponding to each prime set (Table 2) and at 72°C for 2 min. An additional cycle of 5 min at 72°C was used for final extension. After PCR, amplicoms were visualised on 1% agarose gel. And then the samples were mixed together an equal volume of 98% DMF and denatured 5 min at 94°C. The 8 microliter PCR reaction product after opened double-stranded were separated on a sequencing 6% polyacrylamide gel electrophoresis solution with ammonium persulpite (APS) and 70 μl TEMED (Bassam, Caetano-Anollés, & Gresshoff, 1991).

SSR statistical analysis
All genotypes were scored for the presence and absence of SSR Bands. Only clear and repeatable amplification products were scored as 1 for present bands and 0 for absent ones, and this data matrix was subjected to further analysis. Polymorphism was calculated based on the presence or absence of bands. Data matrix was created and used to calculated the genetic distance and similarity using Gen Alex, NTSYS-pc and Pop Gene software analysis. The related genetic parameters were computed as the number of polymorphic bands, and average alleles per locus, heterozygosity (H e ), polymorphism information content (PIC), and other genetic parameters. SAS var 9.2 and SPSS var.21 statistical softwares were used for statistical analysis. Biplot analysis was conducted using STATGRAPHICS. NTSYS-pc computer program (Rohlf, 2000) was used for generating the similarity matrixes and UPGMA clustering.
The information of each pair primer was deduced using the polymorphic information content (PIC) as described by Weir (1996), using the Power Marker ver. 3.25 program. The genetic structure of the accessions were investigated by Analysis of Molecular Variance (AMOVA). The fixation index was used to analyze the genetic structure. Unweight Pair Group Method using Arithmetic averages (UPGMA) on the similarity indices and principal coordinate analysis (PCA) was performed to identify genetic variation patterns among cowpea genotypes using NTSYS-pc version 2.11. Cluster analysis was carried out based on genetic distance. The resulting clusters were represented as dendrograms and printed in MEGA software. To estimate K value, the clusterdness statistic (Rosenberg et al., 2005) was calculated and averaged it over all repeats per K. The highest clusterdness was achieved for K = 2 (result not shown). Furthermore, CLUMPP (Jakobsson & Rosenberg, 2007) was used to merge different Q matrices and cluster assignment was plotted with destruct (Rosenberg, 2004). F-statistics, gene flow, genetic identity and distance were estimated using POPGENE 1.32 software (Yeh, Yang, & Boyle, 1999). Number of alleles observed per locus, and effective number of alleles per locus (N e ), were determined according to Kimura and Crow (1964). Shannon index of diversity were calculated to determine the diversity in populations according to Shannon and Weaver (1949). F ST or G ST statistics were also performed to measure heterozygosity within populations. F ST or G ST were estimated according to Wright (1978). Gen flow was estimated from F ST or G ST according to Nei (1973) as follow: N m = 0.25(1 − F ST )/F ST . The genetic identity and distance between population across all loci was estimated using Nei (1978) unbiased genetic distance coefficient.
At the end, based on scoring banding patterns (present "1" and absent "0") and phenotypic traits evaluated in the field as independent and dependent variables respectively, stepwise regression analysis was performed using statistical software of SPSS ver. 19.

Result
Genetic variation in morphological quantitative traits showed a high variability in 32 cowpea genotypes. Analysis of variance (ANOVA) showed a total of 23% variation among population while 77% variation within population with significant value (p < 0.01) ( Table 7). Basic statistical values for each quantitative trait in 32 cowpea genotypes of the four evaluated populations under two normal irrigation and stress conditions are separately presented in Tables 3 and 4. In both normal irrigation and drought stress conditions, the most variable characters were economical yield, biological yield, date to 50% germination and weight of 100 grains. The characters grain width, grain thickness, pod thickness and pod width were with less variation in both conditions.
The principal component analysis for total of 17 morphological traits in both normal and stress conditions, without rotation varimax for all genotypes were estimated. In normal condition the six principal components analysis explained 88.2% of total phenotypic variation of evaluated traits, having eigenvector equal to 32.99, 14.79, 13.69, 11.93, 8.78 and 6.02% of the total variance respectively. The first traits in component predominant were 100-grain weight, grain length, grain width, grain thickness, number of days to 50% maturity pod, pod length, pod width and thickness of the sheath with positive loadings having a large impact, The second component was dominated by traits such as economic yield, harvest index with a negative impact and the number of seeds per pod, number of pods per plant, plant height, pod length, number of days to 50% flowering, number of days to 50% maturity pod, and number of days per 50% pod emerging with negatively charged but positively charged pod thickness are the most important. Under drought stress condition the five principal components analysis explained 81.17% of total variation for trait evaluated, which had eigenvector equal to 29.03, 21.86, 13.04, 10.49, and 6.74% of the total variance respectively. In the first component, the most important traits were 100-grain weight, grain width, thickness grain, plant height, pod length, the thickness of the pod with a lot of positive impact, The second component consisted of harvest index, with a negative charge and the number of pods per plant, plant height, days up to 50% flowering, days to 50% pod emerging, days to 50% maturity with the positive charge are the most important role. The first two parameters at were used to determine the suitability of data for factor analysis of KMO (Kaiser-Mir-Olkine) and Bartlett test. In both conditions, KMO value was greater than 0.6, which means that correlations among the obtained data was suitable for principal component analysis, and Bartlett test was also significant in terms of both enough correlation between the variables. Therefore it can be concluded that the data are suitable for principal component analysis. Due to the fact that in both normal and drought stress conditions, the first and second components (PC1 and PC2), explained the most of the variance of the data for main factors, Yield and yield components were recognized as the two proper factors to coordinate in determining dispersion and superior genotypes. Observations under normal conditions (Figure 1), revealed that genotypes 313 and 37, along with control genotypes Mashhad (998) and Parasto (999) with the first and second factors were positive and higher seed yield per plant also respectively.
Under drought stress conditions, according to the first and second principal components, genotypes 294,141,220,222,107,175,232,246,17 and 229 were determined to be more positive, showing more seed yield under drought stress conditions (Figure 3). Biplots were drawn for 32 genotypes evaluated to morphological traits using principal component analysis according to the first and second components. As shown in Figures 1 and 3, in both normal and drought stress conditions, genotypes were broadcasted in four areas A, B, C and D. Genotypes located in area A, showed highest values for both factors. Density curves on both conditions, based on the first and second components for normal irrigation and drought stress conditions are shown in Figures 2 and 4, respectively. In this graphs, two main genotypes (right side) were consistent with the first cluster and in the cluster analysis is that most of the genotypes into its place, The other clusters that make up a smaller percentage of genotypes are indivisible in the diagram (left side). So according to the results, the density plot of cluster analysis was justified.
Cluster analysis (UPGMA) using squared euclidian distance was done in order to determine the affinity of genotypes and clustering them based on quantitative data for normal irrigation (Figure 5), and drought stress condition ( Figure 6). Under normal conditions, cowpea genotypes were clustered into four groups based on morphological traits ( Figure 5), each of them divided into different subgroups. The first group was mainly consisted of genotypes 196, 43, 215, 186, 246, 192, 8, 210, 76 and 141, are located used to genotype interactions in the group having a mean value of 100 grains weight, biological weight and economic weight, equal to 13.25, 282.23 and 83.22 g respectively. Genotypes 313,998,7,291,229,37,307,17,107,193,203,9,30,162,294,232,175,222 and 220 were clustered in the second group, having a mean value of 100 grains weight, biological weight and economic weight, equal to 14.1, 361.10 and 109.153 g respectively. The third group included two genotypes 49 and 174, with mean  values for weight of 100 grains, biological and economical weight are 12.93, 445.32 and 100.3 g respectively. The fourth group accommodated only the genotype 999 from USA, which had a 100 grain weight of 20.98 g, biological, and economic weights equal to 626.33 and 151.33 g respectively ( Figure 5).   Dendrogram for drought stress based on UPGMA method, clustered the total 32 cowpea genotypes into four main groups. The first group included cowpea genotypes 222,192,220,193,8,215,141,17,246,7,37,186,76,9,175,999,162,34 and 107, having mean 100 grain weight biological and economic weight values equal to 12.04, 263.22 and 70.52 g respectively. The second group was divided into two sub-clusters. These two sub-clusters included genotypes 232, 291,294,49,203,307,196,229,30 and 313, with mean 100 grain weight of 12.51 g, and biological and economical weights of 372.76 and 87.68 g. The third group had only two genotypes 174 and 99, with mean 100 grain weight of 10.27 g, biological and economical weights of 609.02 and 167.23 g respectively, and the fourth group only included one genotype 998, with mean100grain weight biological and economical weights equal to 18.06 g, 755.13 g and 123.02 g respectively ( Figure 6).
In this study, 22 pairs of SSR primers sets (Table 2) were used to assess the genetic diversity and estimate genetic polymorphism in 32 cowpea genotypes based on codominant marker system. Using these markers showed a high level of diversity in the present study. All 22 pair SSR primers used in this study were polymorphic, a total of 186 polymorphism bands were amplified using these SSR markers. Among the studied primers, the lowest number of alleles per locus ranged from 4 to 5 alleles for Vm25 and Vm4, vs. highest value of 13 and 14 alleles for Vm39 and Vm26 primers respectively. The average number of alleles in this study (8.45) is comparable to those from previous studies of genetic diversity (Cholastova & Knotova, 2012), finding 8.5 alleles per locus. And number of alleles per microsatellite locus ranging from 4 to 14, is in contrast to the results obtained by Cholastova and Knotova (2012), which had reported an average of 22.3 alleles per locus using microsatellite primers in AlfaAlfa (Medicago sativa L.). Since the higher number of alleles for each microsatellite markers (SSR), is a more suitable indicator for estimating genetic diversity (Röder et al., 1998), among 22 primers studied, those showing a high number of alleles, were more suitable for estimating genetic diversity. A total of 740 bands were observed for all populations, of which 346 bands were polymorphic, including 92 (47.57%) band for population 1 (of India), 136 bands (73.51%) for population 2 (of America's population), 66 bands (30.27%) for population 3 (of Latin America), and 66 bands (27/03%) for population 4 (of Asia). The average polymorphism of 44.59% was observed for bands in total population. The average number of alleles per locus was estimated to be 35 alleles ranging from 14 to 67 alleles and PIC was calculated to 22 pair primers SSR representing the allelic diversity for a specific locus, varying from 0.25 to 0.625 with a mean of 0.445. The SSR primers Vm5 and Vm25 showed the highest and lowest allelic frequencies respectively. The SSR primer Vm5  gave the highest allele frequency with maximum content of PIC, more promising than the other markers for determining the genetic distance.
The average number of effective alleles was variable from 1.205 to 1.936 for primer Vm3 and 1.059 for primer Vm25. The Shannon information index (I) values also showed an identical trend variable, with a range of 0.409 for primer Vm34 and 0.122 for primer Vm25, and an average of 0.265. Shannon index sham as PIC show polymorphism primer.
To investigate the relationship between the populations, demographic variables were examined separately for each primer (Table 5). The first indicator, of the total genetic diversity (H T ), ranged from 0.05 for primer Vm25, to 0.236 for primer Vm34 with an average of 0.146. It should also be noted that the primer Vm70 with an average of 0.225, showed the highest amount of total genetic variation. Also the lowest amount of genetic diversity within populations (H s ) ranged from (0.048) for primer Vm25 to (0.192) for primer Vm40. Coefficient of genetic differentiation between populations (G ST ) with an average of 0.0758, ranged from 0.2025 to 0.0325 for primers Vm34 and Vm68 respectively, as the highest and lowest values of the index. G ST statistic is the ratio diversity between populations versus to the ratio total variety of is shows. Amount of gene flow (N m ) was calculated for all primers with a mean 6.09. The amount of gene flow between populations and genetic differentiation coefficient is negatively correlated with those. Among the primers studied for gene flow in Table 6    populations, maximum and minimum gene flow were observed in primer Vm68 (14.879) and primer Vm12 (1.801). High genetic similarity observed in this study, may be due to the integration and overlap in populations as a result of gene flow among populations. Most likely, the genotypes in different populations in nearby areas Probably the result of mechanical or genetic mixing in a long time and with different genotypes within populations, moreover it is assumed that genotypes are even moved in geographically separated populations (Table 5).
In order to evaluate the genetic diversity within populations, average indices of the intra population that included the observed number of alleles (N a ) were calculated. Results showed an average of 1.47 for Latin America as highest and 0.55 for America population as the lowest. High average number of effective alleles was observed in all populations, with the Asian population showing the highest value with an average of 1.238. Nei index unbiased distance and genetic similarity matrices were estimated over all SSR markers, gene diversity (h), and Shannon index (I), population Asia and Latin America showing the highest distance (with an average of 0.146 and 0.233 respectively). To elucidate the relationships between populations, Nei unbiased measure of genetic distance matrices was used in cluster analysis based on the UPGMA algorithm to construct a dendrogram. Number of polymorphic bands in Latin America population with 136 bands was the highest and America's population with the lowest number of 52 bands showed the lowest distance (Table 6). Shannon information index (I) and the percentage of polymorphic information content (%PIC), revealed a high genetic diversity between geographic areas because genotypes with high genetic similarity regions close to the geographic areas away from other genotypes (Konan, Rhee, & Haddad, 2011). In this study, the ability of SSR markers confirmed by distinguishing closely related genomes resulting the minimum genetic distance (Smith et al., 1997).
Analysis of molecular variance (AMOVA) showed that the variation among populations is greater than the variation within population; in other words, genetic structure of cowpea genotypes by AMOVA showed no significant genetic variation (p = 0.226) among populations. (Table 7). The genetic differentiation index value (G ST = 0.0758), indicates a very low genetic differentiation among populations. The variance within populations accounted for the highest portion (77%) of the total variance, against 23% of the variance observed among populations. This represents that the difference between the populations is high, so that the plants have high diversity within populations and populations are heterogeneous. Low PhiPT value shows the distinction between large populations (Table 7).
Similarity matrix and genetic distance population was formed using GenAlex software based on genetic similarity coefficient and genetic distance. The genetic distance between populations varied from 0.0066 for population of India (1), with America population (2), to 0.0282 for population of Latin America (3) with a population of Asia (4) ( Table 8).
The dendrogram provided a basic overview of diversity analysis among cowpea genotypes (Figure 7). No clustering according to region could be observed in the dendrogram, and also no specific clustering was observed for the genotypes. Stepwise regression analysis was performed for each trait separately, to determine the relationship between quantitative traits and molecular data and to identify markers associated with traits. Every quantitative trait was considered as the dependent variable (y) and all molecular markers as the independent variables (x). According to the results, the first X (the first primer) was the most popular, and so entered into the model and then were the other variables (X) (other primers) within in the equation. Based on regression analysis under normal irrigation and drought stress condition (Table 9), a large number of alleles for most of the traits showed a high percentage of variation, suggesting that this model can be used in modifying depending on the marker. In normal condition maximum number of markers were identified for traits grain thickness and width of the pod (33 markers) and the lowest number for the number of days to 50% germination (2 markers). In the normal condition, the highest and lowest rate of regression coefficient were observed for 100-grain weight (R 2 = 68%), and pod length (R 2 = 16.9%) respectively, and for most of the traits, marker Vm68, justified the regression model and a high percentage of the variation. According to primer Vm68, traits including number of grain per pod, date to 50% germination, date to 50% pod emerging, and date to 50% maturing were the first variables put into the regression model. Primer Vm70 was a suitable marker for economic yield as moderators for the regression model.
Under drought stress conditions, a number of primers that justify a higher percentage of morphological traits were identified based on an analysis of regression (Table 9). Maximum and minimum number of markers were identified for pod length (35 markers) and pod width (3 markers) respectively. The highest and lowest regression coefficients (R 2 ) were observed for traits grain thickness (67.4%) and days to 50% flowering (18.9%) respectively. In these conditions the alleles from markers Vm68 and Vm26 were the first variables into the model for most of traits. And therefore high rates of change in traits were justified. So that traits primers Vm68 and Vm26 markers were the first to enter the model for 100 grain weight, number of pod per plant, pod width, biological yield, number of grain per pod, date to 50% germination, and pod length.

Discussion
Genetic diversity is a prerequisite for genetic improvement of agricultural crops. But proper use of genetic diversity within germplasm collections needs a good knowledge about their characteristics. Characterization of accessions is traditionally based on morphological and agronomic traits, which is of high interest for plant breeders. Results of this study confirmed the existence of a high morphological variation in cowpea genotypes collection assessed in this study revealing an interesting starting point for plant development programs in order to introduce new and hybrid varieties. The most important traits of cowpea for breeders and farmers are seed size and produced, high 100 grain Weight and biological yield. In this study, genotypes with highest seed production in normal irrigation condition were 220 (137.83 g) and 291 (193.73 g), and in drought stress condition were 210 (195.44 g) and 291 (132.73 g) (more than 13.98gr and 12.26 gr/100 seeds weight respectively). Phonological stages (days to 50% germination, flowering, poding, and days to maturity) showed a lower/higher coefficient of diversity, in agreement with results reported by Hedge and Mishra (2009), and Stoilova and Pereira (2013). In some genotypes, date to flowering and maturity was lower, as the earliest genotype (186) started flowering 82 days after germination compared with latest ones (genotypes 313 and 998) needing 131 days for beginning of this latent. Shortened flowering period is an advantage at high temperatures and low air humidity may be avoided (Stoilova & Pereira, 2013). In our study, genotypes of 186 and 37 with earlier flowering in American and Asian populations, produced 110 and 434 gr of economical yield and 91 and 255 gr of biological yield, in group C of normal irrigation ( Figure 2) and in group C of drought stress condition (Figure 3) respectively. Most of quantitative traits studied are influenced by environment conditions (Stoilova & Pereira, 2013). Most important yield components in cowpea crops are number of pods, seeds per plant, and pods per plant. In normal irrigation condition high values for these traits were found for two genotypes (220 and 30), and then the genotype of 294. But under drought stress condition high values for these traits were found for two genotypes (203 and 162), and many genotypes were similar to this genotypes (215). In normal irrigation condition genotypes with higher weight of seeds/plant were 220 and 291and genotypes with higher weight of 100 seeds were 998 and 999. In drought stress condition, genotypes with higher weight of seeds/plant were 210 and 291, and for weight of 100 seeds 30, 998, and 999 proved to be superior genotypes.
Presence of genetic diversity in crop populations is not simply detected by morphological characteristics. Use of molecular markers in plant diversity studies is increasingly increased for detection of differences in crop populations at the DNA level (Meng et al., 1998).
A relatively high level of variation was found among cowpea genotypes in present study using SSR markers with 44.59% polymorphism. All SSR loci analyzed in this study exhibited a high degree of polymorphism with 4 to 14 alleles per locus with an average of 8.45 alleles per primer. Fatokun, Stone, and Smith (2008) detected a range of 4 to 13 alleles among 48 wild cowpea lines with an average of 7.5 alleles per primer. Asare et al. (2010) reported 4 to 13 alleles in cowpea assessions collected from Ghana, while Sawadogo, Ouedraogo, Gowda, and Timko (2010) reported 5 to 12 alleles in cowpea assessions collected from Burkina Faso using cross species SSRs from Medicago. In contrast, Diouf and Hilu (2005) reported 1 to 9 alleles in cowpea germplasm. In another study, Vm27 primer showed the lowest number of alleles among 90 cultivated cowpea lines and a wild cross compatible relative (Li et al., 2001). This inconsistency could probably be attributed to the number of genotypes and diversity of germplasm used in above mentioned studies. In this study, PIC showed a range of 0.25 and 0.625 with a mean of 0.45. Kuruma, Suzuki, and Ueda (2010) observed PIC ranging from 0.09 to 0.87 with a mean of 0.34, while Fatokun et al. (2008), observed PIC ranging from 0.29 to 0.87 with a mean of 0.68. One of the most important indicators for the comparison of different markers of differentiation is their PIC. High PIC values indicate high polymorphism or represents a rare allele or alleles at indicator position, which plays an important role in the differentiation and individuals (Agrama & Tuinstra, 2003). Padulosi, Hoeschle-Zeledon, and Bordoni (2007) suggested that an area with intense variation may probably be the one where the crop must have been cultivated for a long time as a result of inbreeding and introgression among different varieties.
The high genetic diversity at molecular level is in accordance with the variation observed in total morphological traits among cultivated genotypes. Total genetic diversity (H T ) was 0.146 on average and ranged from 0.05 to 0.236 and also primer Vm70 exhibited the highest gene diversity (H T ) of 0.225. Genetic diversity interpolation (H S ) levels of heterozygosity was observed ranging from 0.048 (Vm25) to 0.192 (Vm40). Lack of heterozygosity can be attributed to the improved nature of cowpea, where the proportion of heterozygosity is likely to be low. The high level of resemblance reported among cowpea varieties may be due to self-pollination (Padulosi, 1993). The index G ST was low 0.076 indicating low and high differentiation of G ST ranging from 0.0325 to 0.2025. Lack of differentiation among populations and cowpea genotypes indicate the high levels of gene flow between population as well as lack of sufficient time for significant genetic differentiation across populations. The high discriminating power of SSR is also an important factor in the analysis of variation in the gene pool of crops. Overall, relatively high level of similarity was observed among cowpea genotypes for most of morphological characteristics.
There was not found stable trait in our study in total of the 32 cowpea genotypes analysed showing the existence of high variation in morphological traits. In fact, populations from different countries showed a high range of morphological and molecular differences. The higher level of polymorphism showed for these 32 cowpea genotypes may be attributed to a strong genetic diversity of them.
In our study no positive correlation detected between countries and genotypes presenting a high degree of variation between four sampling eareas. This result suggest a high rate of genetic exchange among populations, probably due to interchange between plant materials. Because cowpea, like other legumes and vegetables crops, is managed by farmers based on morphologic and phenotypic traits. The existence of high polymorphism distinguishing local accessions and genotypes has been presented in other studies but it reflects the absence of geographic spatial structure among them, which strongly supports the main result of our work (Marcos et al., 2013).
The knowledge of genetic relationships among genotypes provides useful information to address breeding programs and germplasm resource management (Roldán-Ruiz et al., 2001). In this study, morphological data combined with molecular analysis (SSR markers) characterised the genetic relationships among cowpea genotypes for finding informative markers, thus a significant relationship was made between these two different markers. The range of genetic distance observed for morphological traits was normally higher than SSR markers, which may reflect the influence of the environment on phenotypic features.
Generally, it can be claimed that the combination of microsatellites and morphological data can be highly informative so that they could be efficient and accurate tools for detecting genetic diversity in cowpea germplasm.