A total of 160 Ubi gadong accessions were collected from two states of Malaysia: Terengganu and Kelantan. Samples were collected from five different districts and 11 villages and weighing about 423.5 kilograms of corms (yellow and white, two different varieties of same species) obtained during sampling. Then, selections were done on all samples based on the desired characteristics of a high tree with high yield containing lot of corms. Then, as many as 48 samples from 160 Ubi gadong germplasms were selected as “primary” samples, which were then grown and maintained in Germplasm plot, Research Farm in Gong Badak campus, Universiti Sultan Zainal Abidin. Complete herbarium specimens with male and female flowers were also stored in the herbarium for future identification in Universiti Sultan Zainal Abidin, Kuala Terengganu and Universiti Kebangsaan Malaysia, Bangi. Brief collection details of Ubi gadong accessions are displayed in Table 1.

As Ubi gadong usually produce corms instead of normal seeds like other crops, vegetative propagation was followed using corms/bulbs. Sampled bulbs were planted in polybags (15 × 18 inches), weighing 200–300 g each using a prepared soil (soil:sand:manure = 3:2:1) with five replications. Growing areas were covered with 50% shade and light using black shade cloths until proper maturity of the Ubi gadong plants. Another method of preparation of propagating materials is; to cut corm with respect to some parts of the form and number of buds that form on the surface of the skin of the corm. Transplanted bulbs were tagged and labeled properly. All other agronomic practices (weeding, irrigation, fertilizer application etc.) were done following standard methods. Data on tree height and corm weight were observed every two months (Fig. 1). The plant height was measured from the tree root or the rhizome to the shoot.

The young leaves from each accession of 18 selected D. hispida samples were the main source of DNA. The collected leaf samples were stored in the CTAB buffer to improve the process of solving the cell. DNA extraction was performed using Geneaid Genomic DNA Mini Kit (Plant) to produce high-quality DNA extracts. One microliter of each DNA sample was put on NanoDrop spectrophotometry (ND-1000, NanoDrop Technologies Inc., Wilmington, DE, USA), and the relative purity with concentration of the extracted DNA were 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 12 ISSR primers obtained from various publications were screened with D. hispida DNA; out of these, only three primers, which formed clear and reproducible polymorphic bands, were chosen and used for genetic diversity analysis among 48 Ubi gadong accessions. ISSR primers with sequence are displayed in Table 2.

Polymerase chain reaction for amplification of ISSR fragments was performed in 15 μl of reaction volume containing 1 μl of 70 ng template DNA, 2.0 μl of ISSR primer, 7.4 μl master mix (Dream Taq green PCR master mix-2X containing green buffer, dNTPs, and 4 mM MgCl₂, supplied by Fermentas-Fisher Scientific, UK) and 4.6 μl nuclease-free water. PCR amplification was carried out in a thermocycler (T100TM, Bio-Rad). The initial temperature was adjusted to 94 °C for 2 min and the initial denaturation temperature was set at 94 °C for 1 min. Later on, the elongation temperature was 72 °C for 1 min and finally the elongation temperature was set at 72 °C for 10 min. Finally, the infinitive holds at 10 °C. For electrophoresis, 1.5% molecular-grade Agarose (Nacalai Tesque Inc., Kyoto, Japan) gel was prepared containing 1 μl of Midori green in 1 × 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 70 V for 2 h 30 minutes 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.

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 present and “0” for absent of a band to generate a binary data matrix. Data analysis was performed using Phylogenetic Analysis Using Parsimony (PAUP) software. Comparative data was assessed using a score Neighbour-Joining analysis. A 100-bp DNA ladder was added with the first load of the gel to score for the bands in each gel. The only bands that were scored are those that are > 100 bp in length and reproducible. The scoring of the visual bands was done according to the size of the DNA ladder bands.

Our results showed that plant height and corm yield increased at the beginning to the middle intervals (1 to 3 from February 12 to June 17), but no significant changes were observed in the following two intervals (3 and 4 from August 30 to October 16). These findings were consistent with those of Bahera et al. [6], who reported that the growth increased in the first 5- to 6-month growing periods. This is because, in the early stages of growth, a majority of the plants, especially tuber crops, use more nutrients and water contained in the corm to produce new corms [7].

Table 3 shows that accession 25 (DH0051) showed the highest plant height (288 cm), while accession 24 (DH0050) showed the lowest (84 cm). The largest amount of corms (1309 g) was recorded in accession 2 (DH0028) with the lowest (62 g) was noted in accession 48 (DH0080), respectively (Table 3). These results indicated that a significant morphological diversity was observed among the Ubi gadong germplasm collections. Correlation analysis was also done on the two parameters used by Pearson correlation (R²) on recorded data once every two months: the cumulative analysis results of Records I, II III, IV and Record V are displayed in Table 4.

Pearson values above 0.3 (R² > 0.3) show a strong correlation between the two parameters. The correlation between tall plants and bulbs in February, April, June and August was not significant when compared to the month of October, when the corm weight and height data were significant in the early stages of growth, the plants began to adapt to the new environment, and the chemical productivity or physiological activity was low (Table 4). The average of all measured data readings and Pearson values equal or same with 0.367 are significant. This shows that the corm weight and the height of the plants have a strong and positive correlation. The relationship between measured value and accessions plotted in a scatter graph, using SPSS statistical software 17, is shown in Fig. 2.

The proficient and consistent use of molecular markers such as ISSR for the study of genetic diversity in any food crop or tree crop requires the selection and application of primers that will give clear, distinct, reliable, and sufficient information needed to study the divergence that occurs within the crop [8]. In our research, the number of polymorphic loci detected per primer combination varies according to the primer. ISSR markers amplified distinct band patterns among the 48 D. hispida accessions and revealed polymorphism.

In this study, only three primers (ISSR4, ISSR5 and ISSR8) out of 12 ISSR primers showed polymorphism; this has proved the suitability of those ISSR markers for the diversity analysis of Ubi gadong germplasm collections. Table 5 shows the primer selections made based on the temperature of molecular markers on different plates. The sizes of the amplicons ranged from 300 to 1610 bp, among which the ranges of amplicons for ISSR4 was 415–1538, for ISSR5 was 306–1500 and for ISSR8 was 320–1610, respectively (Fig. 3 and Table 6). A total of 39 alleles were identified, and 22 (56.41%) of them were polymorphic (Fig. 3 and Table 6). Individually, the percentage of polymorphic bands produced by ISSR4 was 38.5%, by ISSR5 83.3%, and by ISSR8 50% (Table 6). Overall, the diversity of amplification paths for ISSR4 and ISSR8 was lower than for ISSR5 markers (Table 6).

The study of genetic diversity is a critical component of applied plant breeding for optimizing the choice of parents in a crop-breeding program [9,10]. Effective germplasm assessment provides the scientific basis for the selection of parents/donors for recombination breeding or hybrid breeding, and to breed for specific agro-ecological conditions and situations [11]. Diversity analysis at the molecular level using PCR-based markers is the efficient and rapid method for identifying the relationships and/or differences among the genotypes [12]. Among the PCR-based markers, microsatellites are becoming more popular and suitable for large-scale analysis, both for genetic diversity and breeding research [13,14].

The ISSR marker data from the selected primers were subjected to cluster analysis using NTSYS program. The similarity was constructed using the Dice coefficient method. Cluster analysis was done to group the genotypes into dendrogram. From this dendrogram, the 48 Ubi gadong accessions were grouped into 10 major clusters (a cluster analysis using only ISSR3 has been presented here), which clearly represents the highest genetic diversity among Ubi gadong accessions (Fig. 4). Among the 10 clusters, cluster VII, consisting of 19 accessions, is the biggest group among 10 clusters, followed by cluster IX containing 11 accessions, cluster X comprised of eight accessions, cluster VIII with four accessions and cluster I, II, III, IV, V and VI all are composed of one accession only (Fig. 4 and Table 7). Very interestingly, both accessions DH27 and DH28 were collected from the same district and even from the same village, but they group far distant clusters, i.e. they are genetically different. On the other hand, in accessions DH29 to DH64, all those samples were also collected from the same villages under the same district, but due to their vast genetic variation, they are also grouped within different clusters. For other samples, the same results have also been observed (Fig. 4 and Table 7), whereas ISSR5 grouped all 48 accessions into seven different clusters and ISSR8 into eight different clusters, respectively (data not shown).