A total of 19 parental palms (15 duras and four pisiferas) and 16 F₁ generations from 16 crosses of Dura (D) × Pisifera (P) were employed in this study (Table 1). Each parental palm and F₁ generation consisted of 19 to 20 individuals or plant samples. The Deli dura materials originated from Sabah Breeding Programme (SPB) were used as female parents. The male parents were the AVROS pisiferas, the descendents of BM119 from Oil Palm Research Station, Banting, Selangor, Malaysia. The dura palms were identified as “D” (D1–D15) whereas; pisifera palms were identified as “P” (P1–P4). The 16 F₁ generations were coded as “DP” (DP1–DP16).

Young leaves of selected parental palms were sampled from Malaysian Palm Oil Board (MPOB) Research Station, Kluang, Johore while 16 F₁ (D × P) generations were collected from MPOB Research Station Keratong, Pahang. Samples were cut into small pieces, labeled and stored in liquid nitrogen during transportation to MPOB Headquarters. All the samples were stored at −80 °C until used for further analysis.

DNA from oil palm leaf tissue was extracted following a modified method of Dellaporta et al. [8]. Two grams of leaf tissue were ground to fine powder in the presence of liquid nitrogen using mortar and pestle. The powder was then transferred into 30 ml Oak Ridge tube, mixed with 15 ml of extraction buffer (100 mM Tris pH 8.0, 20 mM EDTA pH 8.0, 1.4 M NaCl, and 10 mM merchapto-ethanol) and incubated at 60 °C for 30 mins. An equal volume of chloroform/isomyl alcohol (24:1 v/v) was then added and the mixture was centrifuged at 10,000 rpm for 20 min to separate leaf residues. The supernatant was collected into a new tube and mixed with two volume of isopropanol in order to precipitate DNA. The DNA was separated by centrifugation at 10,000 rpm for 20 mins. After discarding the supernatant, the DNA pellet was washed twice with 70% ethanol containing 10 mM ammonium acetate, dried and dissolved in TE buffer (10 mM Tris, 1 mM EDTA pH 8.0). RNase treatment was then carried out by adding RNase (50 μg/ml) and incubated at 37 °C for 20 mins. Ammonium acetate (7.5 M) and two volumes of absolute ethanol were added. The DNA was pelleted by centrifuging at 10,000 rpm for 20 mins. After washing with 70% ethanol, the pellet was dissolved in 1 ml of TE buffer.

Two restriction enzymes, EcoR1 and HaeIII were used for digestibility test of DNA. Both digested and undigested DNA was loaded into 0.8% agarose gel and electrophoresis was done at 100 V in 1X TAE buffer. The gel was stained in ethidium bromide and visualized under UV light. The image was captured using a Polaroid camera. DNA concentration was determined using spectrophotometer. Optical density (OD) was noted at wavelength 260 and 280 nm. The concentration of DNA was calculated from OD reading at 260 nm while DNA purity was calculated by the ratio of absorbance obtained at 260 and 280 nm. The ratio of good DNA quality ranged from 1.8 to 2.2.

A sum of nine microsatellite primers, out of 12 primers tested earlier, were developed in MPOB and derived from expressed sequence tag database [9]. These primers were chosen based on their ability to generate polymorphic and scorable amplification products. The annealing temperature of the primers ranged from 52 to 54 °C (Table 2).

PCR reaction mixture contained primer mix (T₄ polynucleotide kinase, (⁻³³P-ATP, microsatellite primers and kinase buffer), 5U/μl Taq DNA polymerase, 10 mM dNTPs, 50 mM MgCl₂, 10X PCR buffer and 1 μl of template DNA. DNA template was diluted to final concentration of 50 ng/μl. PCR was performed in a thermocycler (Perkin Elmer 9600) essentially as described by Billotte et al. [10]. The PCR amplification was done as follows: denaturation at 95 °C for 30 seconds, annealing at 52 to 54 °C (depending on the primer used) for 30 seconds, extension at 72 °C for 30 seconds; all of these steps were repeated for 35 cycles.

Mixture of final PCR product, bromophenol blue and xylene cyanol were denatured at 90 °C. Bromophenol blue and xylene cyanol were used to show migration because they give color to the PCR product. The samples were run in 6% PAGE at 1600 V for 2 to 3 h. The polyacrylamide gels (30 cm × 30 cm) were then vacuumed dried for an hour and exposed against X-ray film (Kodak) for 3 to 4 days at –80 °C depending on the radioactive signal on the gel. Microsatellite banding patterns observed in 16 F₁ generations and their parental palms were transformed into alleles and loci. Each band within each locus was identified according to alphabetical order and beginning with the most anodal migrating band designated as allele A, the next band was allele B or otherwise (Table 3). With microsatellites, one can usually identify more than two alleles per locus.

For field performances, 16 F₁ generations were evaluated. In each generation, a total of 19 to 20 individuals were used in measurement of 29 quantitative characters such as yield, yield components and vegetative characters (Appendix 1). Yield data were collected for 7 years. Oil palm usually starts fruiting about 27 months after field planting. The first fruit normally ripen during 3rd year after field planting. Thus, yield recording only commenced at 36 months [11]. Each palm was inspected after every 10 days and any ripen fruit bunch present was harvested and weighed using a simple spring balance. The weight and number of bunches were recorded for each palm. The vegetative measurement is the measure of physiological and vegetative performance of a palm. The vegetative data were scored using non-destructive method proposed by Corley and Breure [12].

For molecular analysis, scorable fragments were transformed to genotypic data and arranged as a data matrix. Data were analyzed using Biosys1 computer program to generate distance coefficients. Genetic distances among populations were computed according to the method described by Rogers [13].

Mean data of 29 characters of 16 F₁ generations, obtained during field evaluation, were analyzed by principle coordinate analysis using GENSTAT 5.13 software program (copyright 1987, Lawes agricultural Trust, Rothamasted Experimental station, UK). Estimation of distance between the generations was determined and a dendrogram was constructed using distance matrix. Correlation analyses were performed using Pearson's coefficient to determine the correlation level among the microsatellite-based genetic distance and hybrid performance [7] using the SAS program.

Nine microsatellite primers showed 29 polymorphic bands in parental palms and their F₁ generations. In general, two to six bands were scored per primer (Table 3). Primer MF233056 produced the highest alleles (6) compared to other primers. Four alleles were produced by primer CNH 00887 and MF233033. Primer CNH 00938, CNH 01617 and MF2331019 produced three alleles. Primer CNI 01937, EAP 03160 and EO 02978 produced only two alleles (Fig. 1 and Table 3).

In the case of parental lines, genetic variability was calculated and presented in Table 4. The mean number of alleles per locus ranged from 2.0 to 2.2, with a grand mean of 2.1. The percentage of polymorphic loci at 0.95 criterion ranged from 88.5 to 100% with average of 97.0%.

The grand means observed (H_o) and expected (H_e) heterozygosities across populations were 0.613 and 0.462, respectively. The mean expected heterozygosity values ranged from 0.414 (D3) to 0.498 (D6). The mean observed heterozygosity was the lowest for D2 (0.557) and the highest for D15 (0.696).

Genetic variability of F₁ generations are presented in Table 5. The mean number of alleles per locus ranged from 1.9 to 2.6, with an average of 2.3. The percentage of polymorphic loci at 0.95 criterion ranged from 88.9 to 100% with an average of 94.5%. The mean observed (H_o) and expected (H_e) heterozygosities across populations were found to be 0.621 and 0.455, respectively. The mean expected heterozygosity values ranged from 0.387 (DP8) to 0.498 (DP6 and DP1). The mean observed heterozygosity was the lowest for DP3 (0.512) and the highest for DP11 (0.722).

Table 6 showed the values of genetic distance among the dura and pisifera parental palms. Cluster analysis indicated a clear separation of dura and pisifera parental palms (Fig. 2). The Rogers's distance values for dura parental palms ranged from 0.050 to 0.573. The lowest genetic distance corresponded to dura parental palms, D9 and D10 with value of 0.05. On the other hand, the highest genetic distance was found between palms, D2 and D14 with a value of 0.573. A value of 0.573 indicated that there were 57.3% dissimilarities among the populations in the portion of genome surveyed by the nine microsatellite markers. The genetic distance between dura and pisifera parental palms showed that the highest genetic distance was found between palms, D11 and P4 (0.746). On the other hand, the lowest genetic distance was detected between palms, D2 and P1 (0.444) (Table 6).

The genetic distance between 16 F₁ (D × P) generations is presented in Table 7. The values ranged from 0.089 (between DP10 and DP11) to 0.313 (between DP8 and DP16). There was no 0 (zero) value indicating that no population was identical. From the pedigree information, DP10 and DP11 had common male parental palms (P4). Besides that, the female parental palms of these progenies were found to be the siblings. DP8 and DP16 had the highest genetic distance value.

The results of cluster analysis in the form of dendrogram are presented in Fig. 3. Four clusters were formed. Two clusters had single population (DP4 and DP6), which were separated from other clusters. The first cluster was divided into 3 sub-clusters. Sub-cluster 1 contained DP2 and DP16, which shared one male parent, P1. Sub-cluster 2 consisted of DP12, DP13 and DP14. DP13 and DP14 had a common male parental palm, P1. Meanwhile, in sub-cluster 3, DP9 and DP7 had a same male parent, P3. The second cluster consisted of DP3, DP5, DP15, DP8 and DP1. These DPs were inter-related because they were produced from the same male and female grandparents. In addition, DP3 and DP5 had common pisifera parent where as DP8 and DP1 had come from the same male grandparent.

Quantitative genetic distance between different pairs of different progenies was calculated. The pair-wise genetic distance indicated that the highest genetic distance was obtained between DP16 and DP14 (2.2026) (Table 8) followed by DP4 and DP14 (2.1809), DP16 and DP13 (2.0974), DP4 and DP12 (2.0846), DP4 and DP15 (2.0700), DP4 and DP8 (2.0381), DP4 and DP1 (2.0330) and DP4 and DP2 (2.0175).

The genetic distance analysis using UPGMA clustering system generated 5 genetic clusters with correlation coefficient of 1.26. Here, DP4 and DP16 formed 2 separate clusters. The progenies, DP7 and DP9 grouped together while DP3 and DP5 produced another cluster (Fig. 4). Ten progenies of oil palm (DP15, DP14, DP13, DP6, DP2, DP10, DP8, DP, DP12, DP11 and DP1) were clustered distinctly in the same group.

The results of correlation analyses showed that there was no relationship between genetic distance of parental palms and the performance of the progenies for yield components and vegetative characters except mean nut weight and leaflet number (Table 9). At p ≤ 0.05, significant negative correlation was found for mean nut weight (−0.51) and leaflet number (−0.55).