The genus Phoenix L. (Arecaceae) comprises 14 species (Govaerts and Dransfield 2005), distributed from the E Atlantic (Macaronesia), through Africa, the Mediterranean region, S Asia to islands in the Indian Ocean (Madagascar, Andaman) and the NW Pacific (Taïwan and N Philippines). Phoenix is morphologically and phylogenetically highly divergent from the other palm genera, and constitutes the monogeneric tribe Phoeniceae within the subfamily Coryphoideae (Asmussen et al. 2006, Dransfield et al. 2008). The position of Phoenix within the subfamily Coryphoideae has been confirmed by a generic-level phylogenetic analysis of the entire palm family (Arecaceae) that included plastid and nuclear DNA sequences, cpDNA RFLPs and morphological data (Baker et al. 2009).

The taxonomy, phylogeny and evolution of the genus itself have been assessed using morphological and molecular approaches. According to Barrow (1998), both morphological, and molecular data of the 5S intergenic spacer of the nuclear ribosomal 5S DNA unit supported the existence of two clades of closely related species. The first clade included Phoenix dactylifera, Phoenix sylvestris, Phoenix theophrasti and Phoenix canariensis –the so-called “date-palm complex”–, and Phoenix atlantica (Pintaud et al. 2010). The second group comprised the sister species Phoenix paludosa and Phoenix roebelenii. However, Barrow’s (1998) molecular analysis included only 11 out of the 13 species recognized at that time, since Phoenix atlantica was left as an insufficiently known taxon. Its status as a valid species was confirmed later by Henderson et al. (2006). Using one plastid and 16 nuclear microsatellite markers, Pintaud et al. (2010) demonstrated that all members of the “date-palm complex” are distinct species. Moreover, their data suggested that Phoenix atlantica and Phoenix dactylifera were sister species. Unfortunately, Phoenix paludosa and Phoenix andamanensis were not included in their analyses. Combining sequence data of the chloroplast psbZ-trnfM and rpl16-rps 3 loci, Pintaud et al. (2013) depicted five distinct phylogenetic lineages within Phoenix (Phoenix loureiroi-acaulis-pusilla, Phoenix roebelenii-paludosa, Phoenix caespitosa, Phoenix reclinata, and Phoenix rupicola-theophrasti-canariensis-dactylifera-atlantica-sylvestris), and restricted the “date palm complex” to Phoenix dactylifera-atlantica-sylvestris. This complex could be distinguished by the presence of a 3-repetitions haplotype of a 20 bp minisatellite motif at the rpl 16-rps 3 locus, that was absent in all other species. Phoenix andamanensis was the only taxon not included in their study.

The cultivated date palm Phoenix dactylifera L. is the most important fruit crop in the Middle East and North African countries. This species was probably domesticated around 4 000 B.C. in the Mesopotamia-Arabic Gulf area (Nesbitt 1993, Zohary and Hopf 2000, Tengberg 2012) and is nowadays distributed worldwide.

Phoenix species are largely interfertile and many interspecific hybrids have been recognized or suspected (Greuter 1967, Wrigley 1995). The spread of the domesticated Phoenix dactylifera resulted in situations of sympatry with wild species, promoting interspecific gene flow, in particular with the endemic Phoenix canariensis in the Canary Islands (González-Pérez et al. 2004), and possibly with Phoenix theophrasti in Turkey (Boydak and Barrow 1995), Phoenix atlantica in the Cape Verde Islands (Henderson et al. 2006), and Phoenix sylvestris in NW India (Newton et al. 2013). Moreover, spontaneous and directed hybridization between species is an important aspect of Phoenix ornamental cultivation (Tournay 2009).

Added to the common hybridization process between Phoenix species, the paucity of systematically useful morphological and anatomical characters within the genus (Barrow 1998), makes it difficult to establish a comprehensive taxonomy of the genus Phoenix. Because of this confusing situation, a reliable DNA marker set (barcode) to discriminate among Phoenix species and hybrids would be extremely useful.

Hebert et al. (2003) introduced the concept of “DNA barcode” as a new approach to taxon recognition, assuming that a short standardised DNA sequence can distinguish individuals of a species because genetic differentiation between species exceeds that within species. Since then, DNA barcoding has become increasingly important as a tool in taxonomic studies and species delimitation, as well as in the discovery of new (cryptic) species (e.g. DeSalle et al. 2005, Hebert et al. 2004, Hebert and Gregory 2005, Savolainen et al. 2005, Hajibabaei et al. 2007). A consortium of scientists suggested the two-locus combination of rbcL + matK plastid genes as the universal plant barcode (CBOL 2009), while other authors (Chen et al. 2010, Yao et al. 2010) proposed the ITS2 region as a more efficient barcode. The China Plant BOL Group (2011) highlighted the importance of both sampling multiple individuals and using markers with different modes of inheritance, and suggested to incorporate the ITS1/ITS2 region into the core barcode for seed plants.

However, despite all efforts, no locus (alone or in combination), has proven to be 100% efficient as universal DNA barcode in plants at the species level.

The first DNA barcoding analysis in palms (Jeanson et al. 2011) achieved a 92% success in species discrimination by applying a combination of three markers (the plastid matK and rbcL, and the nuclear ITS2) to the tribe Caryoteae (subfamily Coryphoideae).

Investigating the taxonomic status of Phoenix atlantica, in comparison with its close relatives Phoenix dactylifera, Phoenix canariensis and Phoenix sylvestris, Henderson et al. (2006) identified a polymorphic cpDNA minisatellite locus, situated within the trnG (GCC)-trnfM (CAU) intergenic spacer. Its structure was based on the 12 bp motif CTAACTACTATA repeated in tandem 2-6 times. Four haplotypes were observed: one specific of Phoenix canariensis, one restricted to some individuals of Phoenix sylvestris, and two shared between Phoenix dactylifera, Phoenix atlantica and Phoenix sylvestris. Pintaud et al. (2010) studied this locus in 12 Phoenix species, identifying five haplotypes, whose pattern of variation was strongly associated with species. The maximum number of haplotypes per species was three (Phoenix roebelenii). Yet, most of the haplotypes were shared between species, viz. the 3-repetitions haplotype was the most common haplotype within the genus, and was shared by eight out of the 12 species. Phoenix canariensis was the only taxon characterised by the 5-repetitions haplotype. Hence, despite the promising information obtained, the minisatellite alone did not allow to distinguish all Phoenix species.

Given the potential of the trnG (GCC)-trnfM (CAU) spacer for barcoding in Phoenix, we examined a wider cpDNA region, viz. a ~700 bp sequence psbZ-trnfM (CAU) (Figure 1), in search of an efficient DNA barcode locus for species delimitation and identification of female parents in hybrids in the genus Phoenix.

One hundred and thirty-six individuals, belonging to 13 Phoenix species, with emphasis on Phoenix dactylifera, were analysed in this work (Appendix). Phoenix andamanensis was not included in the analysis due to a lack of material.

For each sample, genomic DNA was extracted from 40 mg of freeze-dried leaf tissue which was first grinded using a bead-mill homogenizer Tissuelyser (Qiagen, France). Extraction was performed using the DNeasy Plant Mini Kit protocol along with the QIAcube robotic workstation for DNA automated purification (Qiagen, France). Extracted DNA was quantified by means of a Nanodrop ND1000 spectrophotometer (Thermo Fisher Scientific Inc., USA) and visualized on 1% agarose gels stained with ethidium bromide.

The PCR amplification was carried out using the monocotyledoneous universal primers psbZ-IGS-F: GGTACMTCATTATGGATTGG, and trnfM-IGS-R: GCGGAGTAGAGCAGTTTGGT (Scarcelli et al. 2011). The amplified cpDNA fragment was approximately 700 bp long. PCR reactions were prepared in 25 µl of total volume, containing the following reagent concentrations: 5 ng/µl DNA template, 0.2 μM each of forward and reverse primers, 2X Failsafe PCR PreMix E (Epicentre Biotechnologies, Madison USA), 2.5 U/µl Failsafe Enzyme Mix (Epicentre Biotechnologies, USA), and DNase-free sterile water. PCR parameters were the following: an initial denaturation step at 94 °C for 3 min, then 35 cycles at 94 °C for 30 s, 60 °C for 30 s, and 72 °C for 1 min, and a final elongation step at 72 °C for 10 min. PCR products were controlled on 1% agarose gels stained with ethidium bromide and then purified using Ampure Agencourt kit (Agencourt Bioscience Corporation, USA). Their quantification was done by means of a Nanodrop ND1000 spectrophotometer (Thermo Fisher Scientific Inc., USA). Cycle sequencing was carried out using the Big Dye Terminator v3.1 kit (Applied Biosystems, USA). Cycle sequencing products were purified using the CleanSeq Agencourt Kit (Agencourt Bioscience Corporation, USA) and were then analysed on an ABI 3130 automated DNA Sequencer (Applied BioSystems, USA).

The chromatograms obtained with the forward and reverse primers were combined and edited with SeqMan II 5.00 software (DNASTAR Inc., USA), to generate consensus sequences, which were aligned in BioEdit (Hall 1999), using the Clustal W algorithm. The obtained alignment was further improved manually with MESQUITE v2.75 (Maddison and Maddison 2007). The observed polymorphisms were positioned in reference to the complete chloroplast genome sequence of Phoenix dactylifera cv. Khalas, available in GenBank (accession NC_013991.2).

To assess the potential of the psbZ-trnfM region as a barcode for accurate species identification, we evaluated the proportion of correct identifications using TaxonDNA (Meier et al. 2006). The Best Match and Best Close Match tests were run for species with > 1 individual and with nearly complete sequences, which resulted in a reduced dataset of 11 species (excluding Phoenix acaulis and Phoenix atlantica) and 121 individuals. Because of this constraint, the two species represented by only one individual were analysed by direct comparison of their sequences. Moreover, direct sequence comparison included not only nucleotide substitutions as in the TaxonDNA analysis, but also indels, minisatellites and homopolymers.