A total of 45 purslane accessions collected from seven states of Western Peninsular Malaysia were used in this study. Among 45 samples, 11 samples were collected from Selangor, five from Melaka, five from Negeri Sembilan, nine from Kedah, five from Perak, five from Penang and the remaining five from Perlis. The collected purslane samples were transplanted in glass house of Field-2 at the Faculty of Agriculture, Universiti Putra Malaysia, and the whole molecular study was done in the Plant Breeding and Molecular Genetic Laboratory, AgroBio Complex, Faculty of Agriculture, Universiti Putra Malaysia. Brief morphological descriptions and collection details of purslane accessions have been given in Table 1 and Fig. 1.

Genomic DNA from each accession was isolated from fresh leaf tissue of 20-day-old plants using the cetyltrimethylammonium bromide (CTAB) method slightly modified from Doyle and Doyle [29]. About 0.5 g of fresh leaf tissue was finely milled using mortar and pistil within liquid nitrogen. About 0.5 g of fresh leaf tissue was ground in liquid nitrogen and taken into a 2-mL micro-centrifuge tube and then combined with 1000 μL of CTAB buffer [100 mM Tris-HCl pH8.0, 20 mM ethylenediaminetetraacetate (EDTA) pH 8.0, 1.4 mM NaCl, 2% (w/v) CTAB, 2% (w/v) polyvinylepyrrolidone (PVP) and β-mercaptaphenol and 3 μL of mercapethanol. The samples were incubated at 500 rpm at 65 °C for 1 h (60 min), and the tubes were gently shaken at 10-min intervals to ensure the mixture was homogeneous. After the incubation step, the samples were centrifuged at 13,000 rpm for 5 min to precipitate the solid parts of the cell. Afterwards, the total amount of supernatant was transferred into a new 2-mL Eppendorf tube. Then, an equal volume of chloroform:isoamyl alcohol (24:1, v/v) was added to each sample and the tube was gently inverted to homogenize the mixture and centrifuged at 13,000 rpm for 5 min to precipitate polysaccharides. The upper phase was then transferred into a new 2-mL Eppendorf tube, excluding the green and white lower phases, and again, 600 μL of cold isopropanol were incorporated and mixed gently by inverting the tube at least 50 times. All samples were then incubated at –20 °C for 30 min. The samples were removed from the ambience at –20 °C and were centrifuged for 10 min at 13,000 rpm to get a DNA pellet. The resultant supernatant was carefully removed and the pellet was allowed to dry in the air. The DNA pellet was then washed and dried out 1–2 times by adding 600 μL of 75% ethanol at 4 °C until white a DNA fiber appeared. Finally 50 μL of TE buffer (10 mM tris-HCl pH 8.0, 1 mM EDTA pH 8.0) and 1 μL of Rnase (Dnase free) were added and vortexed with the finger to ensure the complete dissolution of DNA in the TE buffer. The samples were later incubated at 37 °C for 1 h to break down RNA, and then stored at –4 °C.

One microliter of each DNA sample was put on a NanoDrop spectrophotometer (ND-1000, NanoDrop Technologies Inc., Wilmington, DE, USA), and the relative purity with the concentration of the extracted DNA was displayed. The final concentration of each DNA sample was diluted with 1 × TE buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, pH 8.0) to get the required concentration and kept in a refrigerator at –20 °C for PCR analysis.

A total of 30 ISSR primers obtained from various publications were screened with P. oleracea DNA. Out of these, 28 primers that formed clear and reproducible polymorphic bands were chosen and used for genetic diversity analysis among 45 purslane accessions. The ISSR primers’ sequence with annealing temperature is presented in Table 2.

Polymerase chain reaction for amplification of ISSR fragments was performed in 15 μL of reaction volume containing 1 μL of a mixture of 70 ng of template DNA, 2.0 μL of ISSR primer, 7.4 μL of master mix (Dream Taq green PCR master mix-2X containing green buffer, dNTPs and 4 mM of MgCl₂, supplied by Fermentas-Fisher Scientific, UK) and 4.6 μL of nuclease-free water. PCR amplification was carried out in a thermocycler (T100TM, Bio-Rad) following the 3-phase Touchdown PCR protocol [30], with slight modifications. In the first phase, the initial temperature was adjusted to 95 °C for 3 min and the initial denaturation temperature was set at 95 °C for 30 s, annealing at T_m = 62 °C for 60 s; the elongation temperature was set at 72 °C for 30 s. In the second phase, the denaturation temperature was set at 95 °C for 30 sec, annealing at T_m = 52 °C for 60 s and the elongation at 72 °C for 2 min; finally the elongation temperature was set at 72 °C for 10 min in the third phase. The first and second phases were repeated 9 and 30 times, respectively, followed by cooling; finally, the sample was hold at 4 °C. For electrophoresis, 1.5% of molecular grade Agarose (Nacalai Tesque Inc., Kyoto, Japan) gel was prepared containing 1 μL of Midori green in 1X TBE buffer (0.05 M Tris, 0.05 M boric acid, 1 mM EDTA, pH 8.0). The gel was run at a constant voltage of 80 V for 80 min and visualized under Molecular Imager^® (GelDocTM XR, Bio-Rad). However, most of the patterns with extremely good polymorphism and useful information were often accompanied with a background smear. To reduce this smear, 2% formamide was used in the reaction. All the patterns generated were repeated at least three times in order to obtain reproducible data.

The molecular weight for each band was measured by using Alfa Imager software version 5.5. Each amplification product position was considered as a locus. The amplified products for ISSR marker loci at a specific position in a gel were scored visually as “1” for the presence and “0” for the absence of a band to generate a binary data matrix. A 50-bp DNA ladder was added with the first load of the gel to score of the bands in each gel. The only bands that were scored are those that are > 50 bp in length and reproducible. The scoring of the visual bands was done according to the size of the DNA ladder bands.

From the data generated through binary scoring of the gel photo, genetic distance and similarity coefficient were estimated using Jaccard's similarity coefficients [31]. Cluster analysis was performed to construct a dendrogram based on the similarity matrix data using the unweighted pair group method using arithmetic averages (UPGMA) using NTSYS-pc 2.10 software [32]. Polymorphism information content (PIC) values were calculated to evaluate diverse levels of each ISSR primer according to Anderson et al. [33] using the formula: PIC = 1–ΣP²_ij, where P_ij is the frequency of the jth allele (marker) for the ith ISSR locus. The binary data were also subjected to PCA (Principal Component Analysis) to investigate the structure of our collection. The PCA of the 45 purslane accessions were calculated by EIGEN module of NTSYS-pc 2.10 [32] software. The genetic diversity parameters of each geographic group, which included the percentage of polymorphic loci (PLP), the observed number of alleles per locus (n_a), the effective number of alleles per locus (n_e), Nei's gene diversity (h) and Shannon's information index (I) were calculated by POPGENE software version 3.2 [34].

After obtaining the optimum number of subpopulations, an analysis of molecular variance (AMOVA) was performed using Arlequin software version 3.1 (Excoffier et al., 2005) [35] to estimate the genetic variance components within populations and between populations.

Twenty-eight ISSRs were used to estimate the genetic diversity of 45 accessions of P. oleracea. A total of 167 alleles were identified, and 163 (97.6%) of them were polymorphic. The sizes of the amplicons ranged from 200 to 1500 bp (Fig. 2). The number of alleles for each ISSR ranged from three (ISSR20) to nine (ISSR15 and ISSR24) with a mean of 5.96. The PIC values for the evaluated 28 ISSR primers were moderately high (0.223 to 0.375), with a mean value of 0.337, indicating that the ISSR primers used in this study were effective and polymorphic (Table 3). The maximum numbers of effective alleles (1.766) were found in ISSR2 and the lowest alleles were detected in ISSR10 (Table 3). Furthermore, the highest expected heterozygosity was estimated in ISSR2 (0.499) and the lowest was recorded in ISSR10 (0.257) with the mean value of 0.433. The Nei's index varied from 0.215 to 0.427 with the highest value in ISSR2 and the lowest in ISSR10 and the mean value was 0.346, whereas the Shannon's information index varied from 0.330 to 0.616 with the highest value in ISSR2 and the lowest in ISSR20 with the mean value of 0.513 (Table 3).

The ISSR marker data from the selected primers were subjected to cluster analysis using NTSYS program. The similarity was constructed using Dice coefficient method. Cluster analysis was done to group the genotypes into a dendrogram. From this dendrogram, the 45 purslane accessions were grouped into seven major clusters at a coefficient of 0.76 and the similarity coefficient value ranged from 0.53 to 0.91.

Cluster VII consisted of 22 accessions and is the biggest group among seven clusters, followed by cluster VI which contained nine accessions; cluster III comprised five accessions; cluster II four accessions; cluster I and IV were both composed of four accessions and only Ac18 formed single group in cluster V (Fig. 3 and Table 4). The genetic differences between 45 P. oleracea accessions ranged from 0.91 to 0.38 where the highest distance was observed between Ac24 and Ac23 and the lowest distance was found between Ac42 and Ac10 (see Supplementary data). The pairwise population matrix of Nei's genetic distance according to seven states has been presented in Table 5. The highest genetic distance (0.29) was observed between the Penang and Negeri Sembilan populations, while the least (0.08) was found in the Negeri Sembilan one in relation to Selangor.

PCA analysis was done to see the displacement of the accessions and to further confirm the clustering pattern obtained from the dendrogram. The result of the three-dimensional (3D) and two-dimensional (2D) plots of PCA for ISSR marker analysis was also similar to that of clustering in the dendrogram (Figs. 4 and 5). Eigenvectors help to show the contribution from each axis (pc1–pc3) to the total variation present in the germplasm. Principal components analysis (PCA) indicated the first three axes accounted for 57.7% of total variation among all the accessions studied. The first, second and third axes represented 36.2, 16.7 and 4.8% of the variation respectively (Table 7). From PC1, the highest value was 0.196 followed by 0.195 and 0.194 while the least were–0.035,–0.014, and–0.013, respectively. The highest (0.196) was found in Ac25, while the least was found in Ac43.

The analysis of molecular variance was carried out to enable differentiation of purslane population. This analysis revealed that variation within population accounted for 89% of the total variation, while among the populations was found to be 11% (Table 6). The AMOVA also expressed that variation among populations and within populations were highly significant (at P < 0.010).