Taxonomic Identification of Mediterranean Pines and Their Hybrids Based on the High Resolution Melting (HRM) and trnL Approaches: From Cytoplasmic Inheritance to Timber Tracing

Fast and accurate detection of plant species and their hybrids using molecular tools will facilitate the assessment and monitoring of local biodiversity in an era of climate and environmental change. Herein, we evaluate the utility of the plastid trnL marker for species identification applied to Mediterranean pines (Pinus spp.). Our results indicate that trnL is a very sensitive marker for delimiting species biodiversity. Furthermore, High Resolution Melting (HRM) analysis was exploited as a molecular fingerprint for fast and accurate discrimination of Pinus spp. DNA sequence variants. The trnL approach and the HRM analyses were extended to wood samples of two species (Pinus nigra and Pinus sylvestris) with excellent results, congruent to those obtained using leaf tissue. Both analyses demonstrate that hybrids from the P. brutia (maternal parent) × P. halepensis (paternal parent) cross, exhibit the P. halepensis profile, confirming paternal plastid inheritance in Group Halepensis pines. Our study indicates that a single one-step reaction method and DNA marker are sufficient for the identification of Mediterranean pines, their hybrids and the origin of pine wood. Furthermore, our results underline the potential for certain DNA regions to be used as novel biological information markers combined with existing morphological characters and suggest a relatively reliable and open taxonomic system that can link DNA variation to phenotype-based species or hybrid assignment status and direct taxa identification from recalcitrant tissues such as wood samples.


Introduction
Forest trees constitute about 82% of continental biomass and harbour more than 50% of terrestrial biodiversity. The first seed plants to have evolved were Gymnosperms, which today occupy about 25% of the planet's forests. Pinus is the most important genus within the Gymnosperms and within the Pinaceae family, taking into account the number of species (109), [1]and their contribution to forest ecosystems [2]. Pine trees (Pinus L.) are important elements of the Mediterranean landscape. They have played a major role in the evolution of Mediterranean flora and vegetation [3] and have been widely used by its inhabitants since prehistoric times [4]. The genus is divided into subgenera, sections and subsections. Various classifications have been proposed in this genus [5], [6] with the most recent efforts focusing on DNA phylogenetics to identify related species [7], [8].
Comprehensive classifications of the genus Pinus were established earlier using morphological and anatomical traits [6], [9]; later, molecular data were also included [8], [10], [11]. Congruence between classical and molecular taxonomy has been generally observed, nevertheless variations are noticed between classical and molecular approaches. Some of the first molecular studies in pines involved restriction patterns of the plastid genome [12], which is paternally inherited in the genus Pinus (e.g. [13]). A number of studies have focused on particular taxa within the genus Pinus, for instance in subgenus Pinus [14] and Eurasian species [15]. More recent classifications, including a large number of species, were established from nuclear internal transcribed spacer (ITS) [16] and plastid sequences (rbcL) [8], [16] while molecular technologies were used to identify natural hybrids, detect introgression and study hybrid speciation [17], [18]. Plastid sequences being uniparentally inherited can be part of a hybrid identification protocol. Unequivocal species and hybrids identification in pines is important for both basic research and applied forestry.
The use of DNA sequences to identify forest tree species and hybrids coupled with advances in wood DNA isolation requiring minute amounts of DNA for PCR-based approaches, offer an innovative opportunity for tracing wood and wood products throughout the chain-of-custody from the forest to the consumer. Wood DNA is stable and cannot be manipulated [19]; therefore the development of reliable and efficient tracing methods can be incorporated in forestry applications. Two early examples of relevant DNA marker applications are the identification of wood origin used for the production of wine barrels in France [20] and the DNA fingerprinting of rootstock and logs of the endangered tropical species Intsia spp. (Merbau) to control and restrain illegal logging [21]. To our knowledge there has not been a pertinent systematic study using these approaches with regard to Pinus spp. yet.
The trnL approach is a method of identifying plant species using short DNA sequences [22], [23]. The method is extremely useful in species identification, including cryptic species, biodiversity studies, forensic analysis and phylogenetics. Different short length regions of the plastid genome have been used as DNA molecular identification sites primarily for species identification [22], [23]. The CBOL (Consortium for the Barcode of Life) plant working group recommended using the 2-locus combination of rbcL+matK as a plant barcode [22]. Nevertheless, [24] had already shown the suitability of the trnL intron as a barcode for plants due to its well analyzed and thoroughly understood evolutionary patterns, conserved secondary structure, significant discrimination power, availability of universal primers and wide application. The use of trnL (UAA) intron as a supplementary locus has been advocated for those projects that involved PCR amplification of DNA from highly degraded tissues such as wood [25]. In this communication, we consider the trnL locus as a DNA identification marker. Molecular identification using plastid DNA regions have been extended already in the food industry, evolution studies and forensics [26], [27], [28], [29]. In gymnosperms, the trnL approach has been applied in sparse samples of Cycadales (cycads) and Pinales (conifers) [5], [30], [31]. Gymnosperm life history characteristics (incomplete reproductive isolation, primarily paternal plastid inheritance, large effective population sizes, long generation cycles; [32]) complicate the use of the trnL approach despite a relatively restricted number of extant species (,1000).
The typical trnL approach is time consuming and dependent upon highly experienced personnel and considerable resources, which can hinder large-scale application for forest tree populations. High-resolution melting (HRM) analysis [33] allows genotyping and fingerprinting by discriminating DNA sequence variants such as single nucleotide polymorphisms (SNPs) and small insertion and deletions (indels) based on the shape of melting transitions (Tm) of real-time PCR products [33], [34], [35]. HRM analysis can be applied not only for allele discrimination by targeting well-characterized SNPs, but also for screening of the existence of unknown sequence variations without a sequencing process. HRM is a powerful and accurate technique which is cheaper, faster and simpler than alternative approaches requiring post-PCR processing enzyme restriction and electrophoresis, labelled probes for SNP detection sequencing or TaqManprobe-based real-time PCR [36]. HRM has been used as a molecular diagnostic for species discrimination in higher plants [27], [37], [38] including trees and shrubs, such as Prunus [37] and Vaccinium [39].
The objectives of the present study were as follows: (a) to analyze sequence variation of the trnL plastid region and test its usefulness in the identification of Mediterranean Pinus species (including the endemic to Transcaucasia but introduced to the eastern Mediterranean P. eldarica), (b) to develop a rapid, simple, and stable HRM real-time PCR assay targeting the trnL region for Pinus species identification as an alternative and efficient approach that can be used in the molecular taxonomy studies of pines, (c) to apply HRM analysis of the trnL plastid region for the investigation of plastid inheritance using interspecific hybrids of Group Halepensis (P. halepensis and P. brutia) as an case study, and (d) to trace wood origins by HRM analysis for the highly commercial pine species P. sylvestris and P. nigra.

Ethics Approval
No specific permissions were required for sampling from the Aristotle University Botanic Garden for which there is open access for members of the Faculty of Forestry and Natural Environment of the University (in this case Professor Aravanopoulos F., coauthor). Furthermore, no specific permissions were required for sampling from the Aristotle University Laboratory of Forest Genetics and Tree Breeding, Experimental Plantation in Triadi, Greece for which there is open access for members of the Laboratory (in this case Professor Aravanopoulos F., co-author). No specific permits were required for the described field studies. The locations are not privately-owned or protected in any way, and the field studies did not involve endangered or protected species.

Plant Material and DNA Isolation
Plant material (needle and wood samples) was obtained from the Aristotle University of Thessaloniki Botanic Garden [40], from the Aristotle University of Thessaloniki Forestry Herbarium (TAUF) (http://sciweb.nybg.org/science2/IndexHerbariorum.asp) and the Triadi experimental plantation, of the Laboratory of Forest Genetics and Tree Breeding, Aristotle University [41]. These are located in the greater Thessaloniki area, Greece ( Table 1). The P. brutia x halepensis interspecific hybrids are full-sib F 1 hybrids: interspecific hybridization success has been verified by isoenzyme species-specific genetic markers [41]. We have analysed eight samples per pure species originating from different natural populations and 10 samples of the full-sib F 1 hybrids.
DNA isolation from fresh tissue was performed using 0.1 g of pine needle as starting material in the form of fine powder by employing the Qiagen DNeasy plant mini kit according to manufacturer's instructions. DNA concentration was estimated by standard spectrophotometric methods at 260 nm and 280 nm UV lengths using an Eppendorf BioPhotometer. DNA integrity was tested by gel electrophoresis in a 0.8% agarose gel. Samples were then diluted to a 20 ng/mL concentration.
DNA from wood was extracted using the DNeasy Plant Mini Kit (Qiagen) and applying the same modifications and optimizations as reported elsewhere [40]. Prior to extraction, the surface tissues of wood samples were removed using a saw to avoid contamination with other plant DNA. For DNA extraction, 50-100 mg of shavings produced by drilling of the clean inner part (sapwood) of wood samples was used. We used 80-90 mg of wood shavings for each experiment. Polyvinylpyrolidone (PVP40000, Roth) was added into 800 mL of the AP1 lysis buffer from Qiagen (Hilden, Germany; step up to 2.6% w/v). For all samples, 800 mL AP1 lysis buffer was added to the ground wood samples instead of 500 mL because of sample high absorption capacity. The mixture was incubated overnight at 65uC under a 60 rpm vertical rotation [42], [43]. Blank control extractions were performed simultaneously, starting out with an empty reaction tube and were treated identically for the rest of the analysis. We used individually isolated triplicate samples for all species.

Verification of the DNA Isolation Method
DNA samples isolated from pine needles and wood from the same tree were analyzed by PCR amplification, HRM genotyping Table 1. Taxonomical inference, sample origin, code of voucher specimen, collector, place of Voucher deposition and associated references of the Pinus species and hybrids employed in this study.

PCR Amplification for trnL Analysis
PCR amplification was performed in a total volume of 25 mL in a MJ research thermocycler. The reaction mixture contained 20 ng genomic DNA, 1X PCR buffer, 2.5 mM MgCl 2 , 0.2 mM dNTP, 300 nM forward and reverse primers (Table 2), and 0.5 U Kapa Taq DNA polymerase (Kapa Biosystems, USA). Initial denaturing step of 95uC for 3 min followed by 30 cycles of 95uC for 20 s, 54uC for 40 s and 72uC for 40 s, then a final extension step of 72uC for 2 min.

PCR Amplification for HRM Analysis
The primers used are presented in Table 2. PCR amplification, DNA melting and end point fluorescence level acquiring PCR amplifications were performed in a total volume of 15 mL on a Rotor-Gene 6000 real-time 5P HRM PCR Thermocycler (Corbett Research, Sydney, Australia). A third generation DNA intercalating dye, SytoH-9, which at high concentrations can saturate all available sites within double stranded DNA, was used. The reaction mixture contained 20 ng genomic DNA, 1X PCR buffer, 2.5 mM MgCl 2 , 0.2 mM dNTP, 300 nM forward and reverse primers (Table 2), 1.5 mM SytoH-9 green fluorescent nucleic acid stain and 0.5 U Kapa Taq DNA polymerase (Kapa Biosystems, USA). SytoH-9 fluorescence provides a more accurate assessment of DNA melting status compared to SYBR Green I and can be used to monitor the accumulation of the amplified product during PCR and the subsequent product melting on the RotorGene 6000 (software version 2.0.2, Corbett Life Science, Cambridge, UK).
A rapid PCR protocol was conducted in a 36-well carousel using an initial denaturing step of 94uC for 3 min followed by 30 cycles of 95uC for 20 s, 54uC for 30 s and 72uC for 40 s, then a final extension step of 72uC for 2 min. The fluorescent data was acquired at the end of each extension step during PCR cycles. Before HRM, the products were denatured at 95uC for 5 s, and then annealed at 50uC for 30 s to randomly form DNA duplexes. Specifically, to trace the origin of wood with HRM we used the DNA sequence information obtained for trnL ( Figure S1) and we designed specific primers (Table 2) which can discriminate DNA samples between P. sylvestris and P. nigra by a 1 bp polymorphism. The PCR products were separated by electrophoresis in 3.5% agarose gel (not shown).
HRM was performed as follows: pre-melt at the first appropriate temperature for 90 s, and melt at a ramp of 10uC in an appropriate temperature range at 0.1uC increments every 2 s. The fluorescent data were acquired at the end of each increment step. End point fluorescence level was acquired following the melting process by holding at 60uC for 5 min. In order to further increase the reproducibility and reliability of the HRM curve analysis (by obtaining similar amplified quantities of final PCR products before melting), finer adjustments by diluting were made to the genomic DNA templates obtained from needles and woods of each of the species. A Ct parameter of 2264 cycles at a threshold of 0.01 of the normalized fluorescence was established. All samples were examined in duplicate.

Identification of PCR Products HRM Analysis
The Rotor-Gene 6000 proprietary software (vs. 2.0.2) was used to genotype species, subspecies and hybrids. The negative derivative of fluorescence (F) over temperature (T) (dF/dt) curve primarily displaying the Tm, the normalized raw curve depicting the decreasing fluorescence vs. increasing temperature, and difference curves [33] were primarily used. Furthermore, a twostep procedure was followed [38] to assess similarity of unknown HRM curves with a known one. Each species was set as a 'genotype' (reference species) and the average HRM Genotype Confidence Percentages (GCPs) (value attributed to each species being compared to the genotype, with a value of 100 indicating an exact match) for the replicates (disregarding the most outlying replicate) were tabulated [44]. GCPs were re-coded from a 1-100 to a 1-20 range of values to decrease the number of different genotypes causing small differences in the shape of the melting curves giving slightly different GCPs [37]. The means of the confidence percentage of the species replicates assigned to a representative genotype, together with the standard deviation, were thus obtained. In order to assess whether Pinus species would theoretically be distinguishable from each other corresponding sequences of trnL amplicons from all eight Pinus species were analysed for their melting pattern using the computer program uMELT [45].

Sequence Analysis
PCR products were directly sequenced in two directions for each product with Big Dye terminator v3.1 Cycle sequencing kit (PE Applied Biosystems, Foster City, CA, USA) in an automated ABI 3730 sequencer (PE Applied Biosystems). The sequences were aligned with the MAFFT multiple sequence alignment web service implemented in JalView 2.6.1 [46].

Data Analysis
The sequence character-based method [47] was used with DnaSP [48] and the information from each site was treated as a character to distinguish the species. Assembly of the final sequences was performed using the combination of Phred/ Phrap/Consed software (http://www.phrap.org/). Sequence quality was determined by Phred base caller [49]. Bases with Phred quality scores less than 30 were manually inspected and edited where appropriate using Consed [50]. The B-index cut-off has been set according to [51].

Barcoding of PCR Products by HRM Analysis
The amplicons of oligonucleotide trnL sets were subjected to HRM curve analysis. Optimum ramps for this oligonucleotide set were selected based on the highest confidence percentages of the normalized curves and the ability to visually distinguish between the conventional melt curve peaks obtained from each reference species. The optimum ramp was 0.1uC s 21 for the trnL oligonucleotide set.
To investigate whether the polymorphism in the trnL region of different Pinus species was detectable in conventional melting curves, DNA melting profile analysis was performed. Figure 1A-C depicts the melting peaks (and DNA melting curves) generated for each Pinus species. The peaks represent the temperatures at which the maximal fluorescence decay occurs and are indicative of the dissociation temperature of the amplified product(s). The melting peaks temperature for all pine species are presented in Table 3. Peaks were evident for each species within the range 80.98uC to 83.55uC. All species profiles produced two maxima.
Analysis of the normalized HRM curves with the marker trnL ( Figure 1A, B) revealed that most of the species could be distinguished visually, for example, P. halepensis and P. nigra, as the HRM curves obtained are highly characteristic for each amplicon. Despite the fact that some species peaks present overlapping, the HRM melting curves are quite characteristic for each species based on shape, being dependant on the interplay between GC content, length of amplified product and sequence, even when they define the same T m values. Furthermore, closer examination of the HRM difference curves, with the mean P. pinea curve as the baseline, revealed part of the curve sitting outside the 90% confidence interval curve, suggesting that all the examined taxa via the HRM curves are different species ( Figure 1A) based on the samples analyzed. Assigning species P. sylvestris as a reference genotype we were able to estimate the confidence value of similarity between P. sylvestris and the other Pinus species used in the study. This was achieved by subtracting the area of P. sylvestris melting curve (difference graph, Figure 1B) from the rest of the produced melting curves by the other species. We have showed that trnL was a potential region for distinguishing the species studied ( Figure 1B). The average genotype confidence percentages (GCPs) resulting from HRM analysis of the trnL region of eight pine species are shown in Table 4. Moreover, GCPs were calculated and a cut off value of 90% was used to assign a genotype for the trnL region. The highest GCP (88.83) was found between the P. pinea and P. brutia species, while the lowest (0.0) was between P. eldarica and all other Pinus species.

Plastid DNA Inheritance with the DNA-barcoding and HRM Approaches
Partial sequences of the gene plastid trnL revealed differences in the nucleotide composition between P. halepensis and P. brutia. To confirm the direction of parental inheritance, paternal relationships between the two parental species and their hybridized offspring were constructed. The NJ tree of the plastid trnL indicated that the F 1 hybrids were grouped together with P. halepensis (Figure 2A). The exclusive presence of the paternal parent's P. halepensis sequence in progeny hybrids indicates the paternal inheritance of cpDNAs in these Group Halepensis species.
HRM analysis of the trnL region detected two different genotypes corresponding to P. halepensis and P. brutia ( Figure 2B). All artificial P. brutia x halepensis hybrids had the same genotype as their paternal parent P. halepensis. These results also confirmed the paternal inheritance of plastid DNA in the Group Halepensis species. Furthermore, the HRM results were verified through sequencing of the PCR amplification products for the trnL intron region ( Figure S1). Although the sequences are well conserved, we have identified mainly single nucleotide polymorphisms. In addition, we have detected two insertions (CCAGT in P. peuce and TC in P. eldarica and P. brutia), as well as one ATTCA deletion in P. nigra and P. sylvestris.

Pinus Identification with the trnL Marker
We obtained 74 trnL sequences in total from eight different Pinus species (Table S1) and 10 artificial P. brutia x halepensis hybrids. All sequences have been deposited in NCBI database (Table S1). A very good amplification success was evident in all species studied. The analyzed loci exhibited high PCR success with trnL primers reaching a success rate of 99.8%. All PCR products corresponding to the trnL DNA marker were successfully sequenced and high quality bidirectional sequences were obtained. Sequence quality  Table 3. Mean 6 standard deviation (SD) of the points for the melting peaks of the amplicons resulted from the eight Pinus species in several runs of trnL PCR followed by high resolution melt curve analysis at a ramp of 0.1 o s 21 . was very high as determined by the Phred base caller and Consed. The B-index cut-off has been set to 0.89. The trnL matrix presented a sequence of 549 bp and indels with an average length of 3.714 bp. The average number of indels was 9.09. The mean sequence divergence in Pinus species was 1.390. The discriminating power of the trnL marker at the species level was 100%.

Species
To establish character-based molecular identification sites for all Pinus species studied, eight nucleotide positions of the trnL nuclear region were chosen. These particular nucleotide positions revealed the highest numbers of diagnostic characters (Table 5). Using only these chosen positions, all species could be distinguished by at least one diagnostic site. P. brutia and P. eldarica in particular, shared similar first peaks in both uMELT and HRM curve analyses, but nevertheless the position of the second peaks differed, making them distinct from one another. Overall, we identified eight diagnostic positions at sites 69, 115, 122, 130, 35, 140, 298 and 547 (Table 5). They included all Character Attributes (CAs) for closely related sequences such as P. brutia and P. eldarica, and further sites with a high number of CAs (i.e., more than one clade had a unique character state or polymorphism at that site). P. brutia differed only at one diagnostic site from P. eldarica.

Timber Tracing with HRM Analysis
In order to test the DNA isolation method, needle and wood DNA extracts from the same tree were amplified and genotyped with a Pinus specific plastid trnL primer. In addition, no amplification product was noted both for isolation and for PCR negative controls throughout the whole study. After the confirmation that single species can be identified by HRM analysis, we applied the same approach for species identification using fresh wood samples. We report herein results from two species (P. nigra and P. sylvestris), although similar results were obtained from all Pinus species studied, as well as from old wood specimens (Ganopoulos et al., in preparation). Figures 3A and 3B depict the HRM melting peaks and curves of wood and needles that originated from the same tree have the same profile. The normalized HRM curves for the amplicons of the two wood products (one per individual tree and corresponding species), based on HRM analysis with the trnL marker PinustrnL2 are shown ( Figure 3A). Each genotype produced a unique melting plot that was easily distinguishable from the other and consistent with the observed nucleotide differences among them. The melting profile of the trnL PCR amplicon from P. sylvestris produced a single melting peak, whereas melting of the P. nigra amplicon produced an additional very distinct peak ( Figure 3A). The melting profiles within species were consistent in all samples and replicates. The results show that the trnL sequence obtained from wood and needles of the same tree were identical and the sequences differed among P. sylvestris and P. nigra in one position (283: C.A) allowing species discrimination.

Pinus Identification with the trnL Marker and HRM Analysis
The PCR amplification success rate was effective since there was no apparent relation between amplicon size and PCR success rate. An increase in amplicon size may result in an adverse effect on the PCR amplification success rate [19]. However, such effects were not observed, potentially due to the relatively short fragment length of the trnL region (80 bp) used. We have demonstrated that sequence variation within the analyzed region of the trnL gene enabled the use of HRM analysis for the differentiation of even closely related Pinus species. The eight samples used originated from different natural populations, presented the same haplotype and showed with statistical confidence the absence of intraspecific variation. Nevertheless, the above results should be further verified by future analysis of a larger sample size both in terms of species, populations and individuals within populations. Unknown samples could be analyzed using the above standard species samples as reference controls. Unknown samples may be analyzed in the absence of reference controls by the comparison of the unknown and reference samples HRM curves (e.g. evaluation of the difference curve GCP values, DMelt, or its derivatives). This is possible as HRM curves can be exported in the form of numerical data (Table S2) and imported in analysis programs (such as Microsoft Excel), thus permitting the comparison of the results from different laboratories. Nevertheless, this procedure should be adopted with caution as results may be influenced by several factors such as DNA quantity and quality, Taq polymerase, PCR buffer, chromophore used, template quantity and quality and finally the instrument used.
The range of trnL amplicons in size (536-542 bp) and variation (1-5%) in nucleotide sequence (not shown) are in agreement with previous studies, in which featured amplicons of 400-500 bp in size that differ by only 1 bp could be consistently differentiated by HRM curve analysis [52], [53]. The HRM method is a closed  tube post-PCR method, which permits the rapid analysis of genetic variation in Pinus species via the use of plastid molecular identification regions. HRM measures the rate of double stranded DNA dissociation to single stranded DNA with increasing temperature [54]. This gradual denaturation of PCR amplicons is monitoring real-time subtle changes in fluorescent signal over temperature by including a fluorescent dye in the PCR reaction that intercalates homogenously into DNA and fluoresces when bound to dsDNA [55]. The change in fluorescence measures the thermally-induced DNA dissociation by HRM and the observed melting behaviour is characteristic of the particular DNA product as determined by sequence length, GC content, complementarity, and nearest neighbour thermodynamics [54]. The latest progress of this technique enabled the increased resolution and precision of the instruments and the development of saturating DNA dyes, thus allowing the use of HRM for genotyping. Differentiation down to genus and -in many cases -species level is possible based on melting temperatures (Tm) of specific PCR products [56]. HRM analysis has already been used for the identification of other plant species [37], [57], [58]. In our case, accurate melting curves were generated, allowing us to determine whether different amplicons have the same or different sequence [33]. This study has capitalized on the HRM advantage that melting curves from different amplicons can be differentiated on the basis of shape, even when they define the same T m values [59]. The discrimination of Mediterranean Pinus spp. based on the sequenced portion of the plastid trnL gene and HRM analysis was effective in depicting the taxonomic status of the species employed. Taxa concordance to species was apparent. The sequence variation between taxa also resulted in the differing shapes of the HRM melting curves that separated clearly at all taxonomic levels. Results provide supporting evidence towards the notion of P. brutia being a separate species, not a subspecies of P. halepensis (P. halepensis subsp. brutia Ten.) as the current botanic designation holds [60]. This outcome is in agreement to earlier results providing evidence regarding the separate species status of P. brutia based on nuclear (isoenzyme; [41], [61]) and cytoplasmic (cpDNA; [62], [63]) genetic markers. Hence, the pertinent section of Flora Hellenica [60]may need to be re-examined. The species status of P. eldarica is less clear. This taxa has been considered as a subspecies of P. brutia (P. brutia subsp. eldarica Medwed; [64]), or a separate species [6]. Even though differentiation at the species level by a single diagnostic site has been reported in the literature [65] the differentiation of P. eldarica in our study is less apparent than that observed among other Halepensis species studied.

Plastid DNA Inheritance
Taking advantage of the trnL polymorphisms and the capacity of the HRM analysis, we were able to confirm the paternal inheritance of the plastid genome in one artificial interspecific P. brutia x halepensis cross. Results are in concordance to the initial preliminary results by [62] who used the plastid matK gene and a small portion of the 39-flanking region within the trnK intron. The combination of universal primers and high polymorphism of the trnL region reported herein is very useful for the study of cytoplasmic inheritance and the identification of closely related species. For instance, the parental inheritance patterns of the plastid genomes of shortleaf pine (P. echinata Mill.), loblolly pine (P. taeda L.) and slash pine (P. elliottii Engelm.) through the trnL-trnF intergenic spacer polymorphism analysis was investigated and confirmed [66]. The parental inheritance of cpDNA in conifers has been verified in many species and is widely accepted today as a rule [67].

Timber Tracing with HRM Analysis
DNA extraction and isolation from fresh wood samples and subsequent species identification by the trnL approach and HRM analysis were entirely successful. The corresponding analysis of leaf DNA from each individual showed identical results, therefore indicating that wood can be a reliable DNA source. Especially the success of HRM analysis provides an efficient and reliable tool that greatly facilitates wood identification by simple molecular means. In this respect, HRM analysis further advances wood DNA analysis and extend previous works [68], [69]. Therefore, HRM analysis can be very useful in tracing wood origins, for example regarding illegal logging where logs can be traced back to stumps at least at the species level. hybrids using the oligonucleotide set Ph-trnL. (C) Comparison of the partial nucleotide sequence of the trnL from Pinus brutia and P.halepensis. doi:10.1371/journal.pone.0060945.g002 Table 5. Character-based DNA identities for eight Pinus species and one interspecific hybrid for the trnL plastid region.

Conclusions
This is the first comprehensive study describing the application of HRM curve analysis for differentiation of Pinus species and hybrids in leaf and wood samples. One trnL marker was able to differentiate among the eight Pinus species, following HRM curve analysis (Figure1A, 1B). The trnL region may not be regarded as the most variable region in the plastid genome [70]. Nevertheless, it has been proved a suitable region for molecular identification based on the plastid genome [24] and this study points towards its application as a candidate marker. The usefulness of this approach is demonstrated by its successful application in the identification [71], [72] of even closely related species including pines [8]. Besides basic research in phylogeography, taxonomy and evolution, the unambiguous and straightforward Pinus species identification has multi-faceted practical applications. The presence of species-specific profiles permits the unequivocal identification of natural interspecific hybrids and monitoring in hybrid seed orchards, for instance in P. brutia x halepensis. It also permits the evaluation of the success of artificial pollinations, for example in P. nigra x sylvestris and P. brutia x halepensis hybrids. Forest seed certification, dispersal of forest reproductive material and management of seed source storage and deployment can thus be greatly facilitated. Forensic applications (i.e. in archaeology) can also be advanced through the application of this approach. These results have been extended to the use of wood as a DNA source coupled with the advantages of the application of HRM analysis. Timber tracing with HRM analysis is expected to contribute to future forest certification schemes, forest industry product management, state agencies monitoring, customs offices, illegal trading and false marketing information. Figure S1 DNA sequence alignment analysis of pine species shows differences in the DNA level.

(TIF)
Table S1 GenBank accession code of samples subjected to the species identification test using the trnL region.