Ten date palm cultivars (five males and five females) from four different locations in Saudi Arabia were examined (Table 1). Two male Moshwaq (MOS-A and MOS-H) cultivars originated from different geographic locations. Leaf samples for each of these cultivars were kindly provided by Hada Al-Sham Station, King Abdulaziz University, KSA.

Extraction of total DNA was performed separately from leaves of individual plants using the modified procedure of Gawel and Jarret [25]. To remove RNA contamination, RNase A (10 mg/mL, Sigma, USA) was added to the DNA solution and incubated at 37 °C for 30 min. Estimation of the DNA concentration in different samples was done by measuring optical density at 260 nm.

Thirteen primers were utilized for ISSR analyses (Table 2). PCR was performed in a total reaction volume of 25 μL and amplification (Perkin Elmer 2400 thermocycler, Germany) was programmed to 40 cycles after an initial denaturation cycle for 4 min at 94 °C. Each cycle consisted of denaturation at 94 °C for 1 min, annealing at 40 °C for 2 min, and extension at 72 °C for 2 min, followed by a final extension cycle for 7 min at 72 °C. Reactions were done in three replicates per sample (e.g., bulk of five DNA extracts) to ensure reproducibility of the data.

AFLP analysis was performed using the AFLP Analysis System I (Invitrogen, Life Technologies, Grand Island, NY, USA) according to the manufacturer's protocol. Genomic DNA samples were digested with the restriction enzymes EcoRI and MseI, followed by ligation of adapters to the digested DNA fragments. Pre-amplification was carried out using EcoRI primer plus one extension base at the 3′ position (A) and MseI primer plus one extension base at the 3′ position (C) to amplify fragments that contain complementary sequences. Six combinations of EcoRI primers (plus three extension bases) and MseI primers (plus three extension bases) were successfully used to selectively amplify the DNA fragments matching the primer-extension sequences. The six combinations were: E-AAC/M-CAA, E-ACA/M-CAG, E-ACC/M-CAT, E-ACT/M-CTC, E-AGC/M-CTG and E-AAG/M-CTT. Reactions were done in three replicates per sample and non-repeatable data were removed.

ISSR products were visualized using agarose gel electrophoresis (1.2% in 1X TBE buffer) followed by staining with ethidium bromide (0.3 ug/mL). Amplicons were visually examined with an UV transilluminator and photographed using a CCD camera (UVP, UK). AFLPs were separated by capillary electrophoresis and amplicon sizes were estimated on ABI 3500 DNA sequencer (Applied Biosystems, Foster City, California, USA). Using the program Genemapper 4.1 (Applied Biosystems), a genetic fingerprint was produced for each sample by scoring the presence (1) or absence (0) of a standardized set of markers between 50 and 600 base pairs in size [26].

Fragments recovered by both techniques were considered reproducible and scorable based on the dataset generated from the three separate amplifications for each primer and primer combination. Data were scored as (1) for the presence and (0) for the absence of a given fragment, and sizes were estimated by comparison with a 100-bp ladder (Bioron, Germany) using Gel Works 1D advanced gel documentation system (UVP, UK). Binary data matrices were entered into TFPGA (version 1.3) and analyzed using qualitative routine to generate a similarity coefficient. Dissimilarity coefficients were used to construct dendrograms using unweighted pair group method with arithmetic average (UPGMA) and sequential hierarchical and nested clustering (Neighbor-Joining or NJ) routine using NTSYSpc (version 2.10, Exeter software). Principle component analysis (PCA) was performed using NTSYSpc (version 2.10) and 3D-plotted using GraphPad Prism (version 5.0).

Similarity matrices from ISSR and AFLP dataset were compared based on the TFPGA, the normalized Mantel statistics [27], and the PIC (polymorphism information content) was calculated using the following formula [28,29]:

ISSR analysis is based on inter tandem repeats of short DNA sequences proven to be highly polymorphic even among closely related genotypes due to the lack of functional genetic constraints in these non-coding DNA regions [33]. Among 30 prescreened ISSR primers, 13 were selected for their scorability and reproducibility. Analysis of the 13 primers generated 135 amplicons across ISSR primers with a mean number of 10 amplicons per primer. The size of the ISSR amplified fragments ranged from 203 bp (for primer HB10) to 4596 bp (for primer HB15). The highest number of amplicons (15) was produced by primer HB10, whereas the lowest number (9) was revealed by primers 844B, 17898A, HB14 and HB15 (Table 3). The number of polymorphic amplicons was as high as 114 (85% polymorphism) and the average number of polymorphic fragments were 8.8 per primer (Table 3). The level of polymorphism for different ISSR primers ranged from 20% (for primer HB8) to 100% (for primers 814 and HB15). In this context, Hussein et al. [34] reported a low level of polymorphism (28.6%) in ISSR analysis of Egyptian date palm cultivars, while Adawy et al. [4] and Moghaieb et al. [24] reported high polymorphism in ISSR analysis (64.1% and 73%, respectively). In addition, Marsafari and Mehrabi [35] revealed a higher level of ISSR polymorphism (95.67%) when characterizing some Iranian date palm cultivars. The low level of polymorphism in ISSR analysis generated by Hussein et al. [34] was suggested to be due to a narrow genetic background of the Egyptian date palm cultivars analyzed.

AFLP analysis is based on the PCR amplification of selected restriction fragments from a total genomic DNA digest and this approach combines the reliability of RFLPs with the advantages of PCR methods. Therefore, AFLP permits the development of more accurate and comprehensive fingerprints [36]. In the present study, AFLP analysis of the six primer combinations generated a total of 700 amplicons with a mean number of 117 amplicons per primer combination. The size of the fragments ranged from 40 bp (for primer combinations E-AAC/M-CAA, E-ACA/M-CAG and E-AAG/M-CTT) to 591 bp (for primer combination E-AGC/M-CTG). The number of polymorphic amplicons was 534 (76% polymorphism) and the average number of polymorphic fragments was 89 per primer combination (Table 3). The level of polymorphism ranged from 63% (for primer combination E-AAG/M-CTT) to 84% (for primer combination E-AGC/M-CTG). The highest number of amplicons (210) was detected by primer combination E-ACC/M-CAT, whereas the lowest number (83) was revealed by primer combinations E-ACA/M-CAG and E-ACT/M-CTC (Table 3). Cao and Chao [37] screened 32 primer combinations using two date palm cultivars and found that different primer combinations produced 50–70 fragments ranging in size from 50–700 bp. Diaz et al. [19] used five AFLP primer combinations on three date palm cultivars and generated 310 amplicons, 220 of which (71%) were polymorphic. El-Khishin et al. [38] profiled five Egyptian date palm cultivars using six AFLP primer combinations that generated a total of 433 amplicons with the highest level of polymorphism at 59.02%. Adawy et al. [5] employed 28 AFLP primer combinations to examine relationships among five Upper Egypt date palm cultivars and produced 1135 fragments with as low as 41.59% polymorphism. These discrepancies in the number of AFLP amplicons and the percentage of polymorphisms are likely due to the levels of divergence among different date palm cultivars and the use of different primer combinations.

The total number of cultivar-specific markers scored across species and marker type was as high as 241, 208 from AFLPs and 33 from ISSRs (Tables 4 and 5). The highest number of cultivar-specific markers from ISSRs (6) was scored for primer HB10, while no cultivar-specific markers were detected for primer HB9. The highest number of cultivar-specific markers from ISSRs for an individual cultivar (AJW) was seven, while the lowest was two for cultivars DEK, PER, SUK-Q, SHA and MOS-H (Table 4). The highest number of cultivar-specific AFLP markers for an individual primer combination was 61 (primer combination E-ACC/M-CAT), whereas the lowest number (24) was for primer combination E-ACT/M-CTC. The highest number of cultivar-specific AFLP markers for a single cultivar was 45 (RAB), while the lowest (3) was for cultivar SHA (Table 5). We recommend the use of primer combinations E-ACC/M-CAT and E-AGC/M-CTG in estimating distances among date palm species due to the generation of high number of amplicons, high percentage of polymorphism and high number of cultivar-specific markers.

Genetic similarities among the 10 date palm cultivars based on Nei's method [39] within and across both types of markers are shown in Table 6 and Fig. 1. The highest pairwise similarity indices for ISSR, AFLP and combined datasets were 84% between DEK (female) and PER (female), 81% between AJW (female) and TAB (female), and 80% between AJW (female) and TAB (female), respectively. However, the highest pairwise similarity indices for ISSR, AFLP and combined datasets among male cultivars only were 76, 80 and 79%, respectively, all of which were between SUK-Q and RAB. The latter represents the most consistent results of ISSR, AFLP and combined data analysis. The lowest similarity indices across the three datasets were 65% between TAB (female) and SUK-Q (male), 67% between SUK-A (female) and SUK-Q (male), and 67% between SUK-A (female) and SUK-Q (male), respectively.

The dendrograms based on ISSR, AFLP and combined datasets were congruent with the similarity indices. The Neighbor-Joining (NJ) tree generated from ISSR data was not well resolved and bootstrap support for resolved nodes in the tree was low, except the node joining the two cultivars SUK-Q and RAB (94%). However, AFLP or combined dataset was resolved and bootstrap support for resolved nodes in both trees was high. The highest bootstrap for resolved nodes joining two cultivars was also scored for SUK-Q and RAB (99 and 100%, respectively), followed by AJW and TAB (93%) (Fig. 1). The topology of the ISSR tree is largely incongruent with the AFLP or the combined tree. The ISSR tree consisted of four clusters. Cluster I included four cultivars, all of which are females (SUK-A, DEK, PER and TAB). Cluster II consisted of two subclusters, the first included SUK-Q and RAB and the second included MOS-H. Cluster III included AJW and MOS-A, while cluster IV included SHA. The AFLP tree consisted of three clusters. Cluster I included the five female and two male cultivars, while cluster II included MOS-A, and cluster III included the two cultivars SUK-Q and RAB. Cluster I was divided into four subclusters, the two male cultivars (SHA and MOS-H) were in subcluster 4. The combined tree was congruent with the AFLP tree, largely due to the fact that most of the data in the combined analysis is derived from AFLPs. However, subcluster I of cluster I of the combined tree was further divided into two groups, one included SUK-A, DEK and PER, while the second involved the other two female cultivars AJW and TAB. The dendrograms from AFLP and combined markers indicated a partial separation of cultivars based on sex, which is also reflected in the pairwise similarity indices. There is no correspondence between the tree topology and fruit shape or color. The high pairwise relatedness between SUK-Q (male) and RAB (male) (79%) may be explained by the similarity in sex (male), fruit shape (oval) and fruit color (brown), while the similarity between AJW and TAB (80%) can be explained by the similarity in sex (female) and fruit shape (oval). Unexpectedly, pairwise similarity between MOS-A and MOS-H was low (71%), although they are the same sex (male), fruit shape (cylindrical) and fruit color (yellow). These results suggest that more detailed genomic and genetic analysis of these cultivars is required to characterize their relatedness.

The results of pairwise similarity indices and dendrograms constructed for different cultivars from ISSR, AFLP and combined datasets were consistent with those of PCA analyses. Percentages of variation in PCA plots of ISSR, AFLP and combined datasets were 30.11, 34.26 and 33.13% on the X axis, 26.84, 23.25 and 24.25% on the Y axis and 21.99, 21.99 and 22.00% on the Z-axis (Fig. 2a). PCA indicated that pairs of cultivars SUK-Q and RAB and AJWA and TAB are closely related. Based on geographic location, PCA plots of ISSR, AFLP or combined data (Fig. 2b) consistently indicated relatedness among two cultivars from Al-Madinah (SHA and RAB) and three cultivars from Al-Riyadh (DEK, TAB and PER). The other cultivars showed no relatedness based on their geographic locations. The PCA plot of ISSR data based on fruit shape showed the lowest level of consistency with that of AFLP or combined dataset. Accordingly, the PCA plot of combined dataset was partially inconsistent with that of AFLP dataset (Fig. 2c). SUK-Q and RAB in ISSR plot were the most closely related cultivars, as they share the same fruit shape (oval) and color (brown). In addition, SUK-A, PER and TAB share the same fruit shape (oval) and TAB and DEK share fruit color (yellow) (Fig. 2c, d). In AFLP or combined PCA plots, two sets of three cultivars (AJW, TAB and PER with oval fruits, and SHA, MOS-A and MOS-H with cylindrical fruits) showed relatedness based on fruit shape (Fig. 2c). In the PCA plots of AFLP or combined data based on fruit color, three sets of cultivars; i.e., SHA, MOS-A and MOS-H (yellow), TAB and DEK (yellow) and PER and SUK-A (brown) were most closely related (Fig. 2d).

The polymorphism information content (PIC), average heterozygosity (He), effective multiplex ratio (E), and marker index (MI) were computed based on experimental data (Table 7). AFLP data revealed higher PIC, He, E and MI values (Table 7) compared to ISSR indicating that AFLP is more effective in detecting polymorphism among date palm cultivars. Forty-eight of the 114 ISSR markers (42%) and 267 of the 534 AFLP markers (50%) exhibited PIC values ranging from 0.4 to 0.6 (Fig. 3a, b). Effective multiplex ratio (E) of the AFLP depends on the fraction of polymorphic markers (β) as many of the fragments obtained by one primer combination are polymorphic across the examined date palm cultivars. These results highlight the distinctive nature of the AFLP marker compared to ISSR marker as a powerful procedure to survey the genetic diversity of date palm cultivars.

These results are in agreement with those of Powell et al. [28] (for both types of markers in soybean). In the present study, ISSR markers were found less reliable for detecting genetic relatedness among date palm cultivars than AFLP markers. The use of more ISSR primers may improve the reliability of this approach for characterizing cultivars at the molecular level. AFLP markers have recently been considered suitable for genomic diversity and cultivar fingerprinting [40–42]. In a number of studies, AFLP markers were also identified for economically important traits [43–45], some of which can be utilized in marker-assisted selection programs.

More recently, a genomic sequence analysis interestingly demonstrated that P. dactylifera experienced a genome-wide duplication and genetic diversity analysis indicated that stress resistance and sugar metabolism-related genes tend to be located in chromosomal regions with low density of single-nucleotide polymorphisms [46].

In conclusion, AFLP and ISSR markers differed in their ability to differentiate individuals and for detecting polymorphisms. They can complement each other, although this was not the case in the present study. However, these markers did provide sufficient variation to identify date palm cultivars from Saudi Arabia. Some of the AFLP markers generated through this work (e.g., sex marker) can be utilized in the future in breeding programs of date palm.