DNA barcoding has been a major advancement in the field of taxonomy, seeing much effort put into the barcoding of wide taxa of organisms, macro and microalgae included. The mitochondrial-encoded cox1 and plastid-encoded rbcL has been proposed as potential DNA barcodes for rhodophytes, but are yet to be tested on the commercially important carrageenophytes Kappaphycus and Eucheuma. This study gauges the effectiveness of four markers, namely the mitochondrial cox1, cox2, cox2-3 spacer and the plastid rbcL in DNA barcoding on selected Kappaphycus and Eucheuma from Southeast Asia. Marker assessments were performed using established distance and tree-based identification criteria from earlier studies. Barcoding patterns on a larger scale were simulated by empirically testing on the commonly used cox2-3 spacer. The phylogeny of these rhodophytes was also briefly described. In this study, the cox2 marker which satisfies the prerequisites of DNA barcodes was found to exhibit moderately high interspecific divergences with no intraspecific variations, thus a promising marker for the DNA barcoding of Kappaphycus and Eucheuma. However, the already extensively used cox2-3 spacer was deemed to be in overall more appropriate as a DNA barcode for these two genera. On a wider scale, cox1 and rbcL were still better DNA barcodes across the rhodophyte taxa when practicality and cost-efficiency were taken into account. The phylogeny of Kappaphycus and Eucheuma were generally similar to those earlier reported. Still, the application of DNA barcoding has demonstrated our relatively poor taxonomic comprehension of these seaweeds, thus suggesting more in-depth efforts in taxonomic restructuring as well as establishment.
Citation: Tan J, Lim P-E, Phang S-M, Hong DD, Sunarpi H, Hurtado AQ (2012) Assessment of Four Molecular Markers as Potential DNA Barcodes for Red Algae Kappaphycus Doty and Eucheuma J. Agardh (Solieriaceae, Rhodophyta). PLoS ONE 7(12): e52905. https://doi.org/10.1371/journal.pone.0052905
Editor: Sofia Consuegra, Aberystwyth University, United Kingdom
Received: September 7, 2012; Accepted: November 23, 2012; Published: December 20, 2012
Copyright: © 2012 Tan et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Project funding is by the E-Science Fund, Project No. 14-02-03-4027, Department of Fisheries Malaysia (53-02-03-1062) and University of Malaya Malaya, PPP (Phylogenetic analysis of Kappaphycus spp. and Eucheuma spp. with the application of various molecular markers; grant number: PV014/2011A). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The introduction of DNA barcodes by Herbert and co-workers , ,  has brought about large impacts on the advancement of systematic biology; where short, easily amplified regions of DNA exhibiting large variation among species, yet sufficiently variable within species, are constantly used for species delineation, identification as well as archiving with reference to known, established species , , . The Barcode of Life Data System (BOLD) is notably the largest initiative in establishing a worldwide DNA barcode library, signifying its importance and popularity for the scientific community , , . The usefulness of DNA barcoding is evident when dealing with taxa displaying phenotypic plasticity throughout diphasic or triphasic life cycles as well as taxa involving cryptic species. These phenomena are generally predominant in marine macroalgae, thereby enticing the application of DNA barcoding, as reported in numerous studies encompassing the order Gelidiales , Gigartinales, , , , Graciliariales , , Laminariales , and Fucales . Studies on DNA barcoding over broader taxa of rhodophytes have also been reported with promising results , , .
The rhodophytes Kappaphycus and Eucheuma, commercially known as “cottonii” and “spinosum”, respectively are widely established as lucrative sources of carrageenan, with Indonesia and the Philippines being the largest carrageenophyte producers worldwide . Despite being extensively farmed, the morphologically diverse nature of Kappaphycus and Eucheuma still poses difficulties in species identification , , , , , , , even leading to the cultivation of mixed populations that inevitably reduces overall yield . These have resulted in the subsequent employment of molecular phylogenetic studies which all share one main objective – to infer and understand the phylogenetic relationships between Kappaphycus and Eucheuma congeners. As of now, various molecular markers have been introduced for the molecular taxonomy of these carrageenophytes, namely the mitochondrial-encoded partial cox1 and cox2-3 spacer, nuclear-encoded ribosomal Internal Transcribed Spacer (ITS) and 28S large subunit (LSU), plastid-encoded rbcL, RuBisCO spacer and the 23S Universal Plastid Amplicon (UPA) , , , . However, the suitability of these genetic markers as DNA barcodes are to date, unassessed.
Having high quality DNA barcodes for Kappaphycus and Eucheuma would hasten the identification and recording of these seaweeds throughout the globe, particularly those within Southeast Asian waters, where it is believed that many species are yet to be described. Accurate identification of undescribed species would be most useful in developing improved commercial varieties, preferably ones with higher carrageenan yield, faster growth as well as better disease resistance. Apart from species diversity, data on the distribution of Kappaphycus and Eucheuma would also be useful in the detection of invasive strains , . The potential benefits of DNA barcoding for Kappaphycus and Eucheuma underline the main purpose of the present study, which is to gauge the efficacy of four molecular markers, including the commonly used mitochondrial cox1 and plastid rbcL barcodes, the mitochondrial cox2-3 spacer and a newly designed cox2 molecular marker as potential DNA barcodes for selected Kappaphycus and Eucheuma specimens found in Southeast Asia. Marker assessment and comparison were carried out using the tree-based DNA identification technique as well as the distance technique- Best Match (BM), Best Close Match (BCM) and All Species Barcodes (ASB) criteria , . Additionally, the effect of larger datasets on the accuracy and robustness of DNA barcoding was assessed by incorporating sequences of the widely used cox2-3 spacer marker which includes specimens collected from different parts of the world, to that generated in the present study.
Materials and Methods
Kappaphycus and Eucheuma specimens were not collected from any national parks or protected areas, thus not requiring any specific permits for sampling. Specimens were collected from open coastal areas as well as from aquaculture sites with consents from the respective owners. Members of the two genera are not endangered or protected species.
Details of samples used in this study are summarized in Table 1. Specimens were either sampled directly from farms, retrieved via snorkeling or scuba diving. Sample preservation and genomic DNA extraction were based on protocols described by . Details on molecular markers and primers utilized in this study are summarized in Table 2. The mitochondrial encoded cytochrome c oxidase subunit 2 (cox2) primers were designed based on the cox2 region of the complete mitochondrial genome of Chondrus crispus (GenBank no. NC_001677) . PCR parameters of the cox2 molecular marker are similar to that of rbcL, but with 30 amplification cycles. Amplicons were electrophoresed using a SYBR®Safe (Invitrogen, USA) stained 1.0% agarose gel, prior to purification using LaboPassTM (Cosmo Gentech, Korea) gel and PCR purification kits. Purified products were sent to Lucigen (Taiwan) for ABI SoLiD sequencing.
Resulting electropherograms were viewed, truncated at terminals and contigs generated using ChromasPro V1.5 (Technelysium Pty Ltd). Multiple sequence alignments were computed using ClustalX V2.0  and converted into NEXUS format prior to subsequent analyses.
Considering the poor taxonomic status of Kappaphycus and Eucheuma, certain molecular marker assessments were conducted using two different criteria, where (i) species were named and processed based on Operational Taxonomic Units (OTU) (Table 1) , or (ii) original species names were used with no alterations whatsoever throughout the analyses, herein termed non-OTU. The 2nd criterion is only applicable under the Large Dataset Assessment. An OTU in this context refers to a cluster of species-specific specimens constituting a monophyletic clade which is (i) sufficiently variable genetically from the sister taxa such that bifurcating patterns are observed or (ii) geographically distinct. Morphological criteria were not applied due to the extensive morphological plasticity of these red algae.
Distance analyses were performed using TaxonDNA's Species Identifier v1.7.7  for each molecular marker dataset. The pairwise distances for intra- and interspecific frequencies; calculated using both the uncorrected and the Kimura 2-parameter (K2P) corrected pairwise distances, were plotted to observe overlaps in genetic variability, if present. These pairwise distances were computed using the (i) total overlap range and (ii) 90% overlapping range- the largest 5% of the intraspecific and lowest 5% of the interspecific samples excluded. Minimum base pairs in common for distance calculations were set at 300 bp for all molecular markers assessed. The effectiveness of respective molecular markers was tested based on the Best Match (BM) and Best Close Match (BCM) criteria comprehensively described by Meier and co-workers . In short, the BM criterion assigns a species name to the query sequence based on its best barcode match, regardless of the magnitude of similarity between the query and the barcode sequences. BCM, on the other hand involves the initial identification of the best-matched barcode, determination of whether the barcode is sufficiently similar to the query, followed lastly by the assignment of a species name . The smallest interspecific distance generated using the Pairwise Summary module for each molecular marker was used as threshold value for BM and BCM computations.
Neighbor-Joining (NJ) trees were generated based on the Kimura 2-parameter model using default PAUP 4.0b10  settings for each DNA marker to provide visual displays of genetic variation within and between species. Tree nodal supports were generated via 1,000 bootstrap replicates. Resulting NJ trees were analyzed and processed using Figtree v1.3.1 . For all trees, the success of species identification was determined based on the criteria proposed in  and . Under Hebert and co-workers , identification was considered a success when the query clusters along with all conspecific sequences; and considered a failure when the query matches conspecific sequences occurring in multiple clusters or clades within the tree. Ambiguous were all singletons within the dataset. Meier and co-workers  used a revised, and more stringent identification criteria for the tree-based approach. Queries were considered correctly identified when in polytomy with conspecifics, or at least one node into a clade of conspecifics. Misidentification were those queries in polytomy with only allospecific sequences or those at least one node into an allospecific clade. Queries without conspecific sequences or queries sister to conspecifics were considered as ambiguous or unidentified. Additionally, the phylogeny of Kappaphycus and Eucheuma were interpreted and briefly described based on the resulting NJ, Maximum Likelihood (ML), Maximum Parsimony (MP) and Bayesian Inference (BI) trees.
Large Dataset Assessment
The cox2-3 spacer, currently the most widely used genetic marker for Kappaphycus and Eucheuma, were used to simulate empirically the effectiveness of a molecular marker in relatively larger datasets. Cox2-3 spacer sequences from recent studies , , ,  were obtained from the GenBank. Specimens of ambiguous or uncertain identity were excluded from the dataset. Multiple sequence alignments were generated and subjected to similar K2P distance and tree-based analyses. Datasets were analyzed based on the application of the OTU and non-OTU criteria. An additional distance-based identification criterion termed All Species Barcodes (ASB) was also employed using Species Identifier v1.7.7 . This relatively more subtle approach compares the query to barcodes generated using the same threshold for BM and BCM. Queries were considered as successfully identified when matched with at least two conspecific barcodes of the species in question. Ambiguous were all queries followed by only one conspecific barcode or only a portion of the conspecific sequences. Queries were designated misidentified when matched with allospecific barcodes.
Phylogenetic trees were also inferred for the cox2-3 spacer based on Maximum Likelihood (ML), Maximum Parsimony (MP) and Bayesian (BI) algorithms. Solieria was used as the outgroup for each analyses . Parsimony analyses were conducted using PAUP 4.0b10  as heuristic searches using 1,000 bootstrapping replications; with 100 stepwise random sequence addition and tree bisection reconnection (TBR) branch swapping. All characters were assigned unordered and unweighted. Retention indices (RI) and Consistency indices (CI) were also generated.
ML analysis was performed using raxmlGUI  and BI via Mr. Bayes v3.2.1 , . Best fit nucleotide substitution models were determined for each DNA region using Kakusan 4 , generating command files for RAxML and input files for MrBayes. ML trees were inferred using the ML + thorough bootstrap algorithm, with the GTR+GAMMA model over 20 independent searches and 1,000 non-parametric bootstrap replicates. Identical sequences were not omitted for the analysis. For the BI analyses, two sets of four Monte Carlo Markov Chains (MCMC) were performed in parallel over 2,000,000 generations, with trees sampled every 500th generation. Tracer v1.5 (http://196 tree.bio.ed.ac.uk/software/tracer/) was used to assess convergence of log likelihood values, where 100,000 generations were discarded as burn-ins, well after stationarity was achieved. Results were used to generate a 50% majority-rule consensus tree. All trees were again analysed and processed using Figtree v1.3.1 .
All 29 sequences were continuous and easily aligned for all four molecular markers, generating datasets of different final lengths: cox1 (1,411 bp), cox2 (575 bp), cox2-3 spacer (365 bp) and for rbcL (1,464bp). Cox1 exhibited the most phylogenetically informative sites of 277 (19.6%) characters, followed by rbcL with 126 (8.6%) characters, cox2 with 115 (20%) characters and the cox2-3 spacer with 76 (20.8%) characters. All sequences were deposited into GenBank (see Table 1). The cox1 sequences amplified in this study encompasses the ones by Saunders et al.  and Robba et al. .
Pairwise distances (based on the corrected Kimura 2-parameter) showing the intra- and interspecific genetic variability for cox1, cox2, cox2-3 spacer and rbcL were plotted in Figure 1. Both the corrected and uncorrected pairwise distances generated similar results for cox2, cox2-3 spacer and rbcL. No overlapping in terms of intra- and interspecific genetic divergence was observed for all four molecular markers. However, the rbcL marker showed the least distance (0.06% differences for total and 90% overlaps) between the smallest pairwise distance among interspecific but intrageneric samples and the largest pairwise distance among intraspecific sequences- smaller “barcoding gap”. This was followed by cox2-3 spacer (0.06% for total and 90% overlaps), cox2 (0.52% for total and 90% overlaps) and cox1 (K2P: 0.64% for total overlaps and 0.71% for 90% overlaps; uncorrected pairwise distance: 0.63% for total overlaps and 0.70% for 90% overlaps).
Numeric values are arranged according to: the difference between the smallest interspecific but intrageneric distance and the largest intraspecific distance|smallest pairwise distance between interspecific but intrageneic sequences|largest pairwise distance between intraspecific sequences; followed by the number of observations affected (in brackets).
Results based on the Best Match (BM) and Best Close Match (BCM) criteria are as follows: All four genetic markers were able to correctly identify all 29 species based on the BM criteria. For the BCM criteria, rbcL showed the highest success in identification with 100%, followed by cox2 (96.6%), cox2-3 spacer (93.1%) and cox1 (79.3%). No matches were recorded at values 3.44%, 6.89% and 20.68% for cox2, cox2-3 spacer and cox1 respectively.
Resulting NJ phylogenetic trees are shown in Figure 2. Similar tree topologies were observed based on the cox1, cox2, cox2-3 spacer and the rbcL molecular markers, all showing an apparent phylogenetic delineation between Kappaphycus and Eucheuma i.e. Neighbor Joining bootstrap supports (NJ-BS) of 100% respectively. Better resolution was observed for the mitochondrial-encoded genetic markers compared to the plastid-encoded rbcL counterpart on the species level, with relatively higher discrimination in the clustering of specimens.
(NJ) trees of selected Kappaphycus and Eucheuma from Southeast Asia based on (a) cox1; (b) cox2; (c) cox2-3 spacer; (d) rbcL molecular markers. Numbers at node indicate corresponding bootstrap values over 1,000 replicates. Clade annotations represent Operational Taxonomic Units (OTU), where KA = Kappaphycus alvarezii; KS = Kappaphycus striatum; KAr = Kappaphycus sp. “Aring-aring”; ED = Eucheuma denticulatum.
Tree-based DNA identification using Hebert et al.'s  criteria showed 100% identification success across all species for all four genetic markers. Apart from the plastid rbcL DNA marker, identification success saw a decrease with the application of Meier et al.'s  revised identification criteria. For the cox1 marker, identification success was recorded at 89.7%, with the remaining 10.3% sequences assigned as ambiguous. Both the cox2 and cox2-3 spacer showed identification success of 96.6%, with 3.4% ambiguity.
Tree topologies based on ML, MP and BI analyses were identical to that of the NJ trees and were not shown (to be distributed upon request). For the monophyletic Kappaphycus clade, four subclades were generated, denoted as OTUs KA1, KS1, KS2 and KAr; corresponding to K. alvarezii, K. striatum 1, K. striatum 2 and Kappaphycus sp. “Aring-aring” respectively. All K. alvarezii specimens were inferred to be the same species with no significant genetic variation by all four markers (NJ-BS = 99–100%). Two genotypic strains of K. striatum-KS1 (1, 31, 59, 460, GUI4, SIT5 and V6) and KS2 (83, 98, 105 and SIT4) were observed with high support (NJ-BS = 99–100%) for each marker, although the clustering was less obvious for rbcL. Cox1 demonstrated more specific delineation within species, where the Philippine GUI4 specimen was inferred to be genetically different from the rest of the KS1 specimens (NJ-BS = 99%). Based on all four markers, Kappaphycus sp. “Aring-aring”, first reported by Tan and co-workers (2012), formed a highly supported monophyletic clade (NJ-BS = 95–100%) sister to that of K. alvarezii and K. striatum (NJ-BS = 100%). Specimens 49 and 115, which were both from Sabah, were shown with moderate support (NJ-BS = 72% and 75%) to be more closely related to one another compared to that of sample 93 by the cox2-3 spacer and cox2 genetic marker respectively.
Eucheuma denticulatum specimens formed a monophyletic clade composed of subclades termed as ED1 (44, 45, 56, 454 and BOH5) and ED2 (41, 42 and 97) for all four molecular markers (NJ-BS = 100%). Similar to that observed for K. striatum, the assortment of ED1 and ED2 was not that discrete for the rbcL marker. Conversely, the cox1 marker exhibited relatively better intraspecific delineation, where the Indonesian 454 specimen was shown to be different from the remaining ED1 samples (NJ-BS = 100%). Although moderately supported (NJ-BS = 77%), samples 42 and 97 were inferred by cox1 to be distinct from that of sample 41.
Despite the equal ability of all four markers to cluster species consistently and somewhat accurately, the cox1, cox2 and rbcL NJ trees showed increased robustness in terms of nodal supports, of which the cox1 marker exhibited more specific resolution in terms of intraspecific genetic variations.
Large Dataset Assessment
Multiple sequence alignments with a length of 341bp were generated from the dataset incorporating all selected 76 cox2-3 spacer sequences (including the Solieria outgroup), representing ten OTUs or five non-OTUs. Cox2-3 spacer sequences generated from this study were truncated to allow better comparisons with shorter GenBank counterparts. 118 phylogenetically informative and 192 constant characters were recorded.
Figure 3 illustrates the intraspecific and interspecific genetic divergence based on the K2P model between OTUs and non-OTUs. For the former, a pairwise distance overlap of 0.29% was observed between the smallest interspecific and the largest intraspecific sequences; whereas no overlaps were observed for the latter.
Numeric values are arranged according to: the difference between the smallest interspecific but intrageneric distance and the largest intraspecific distance|smallest pairwise distance between interspecific but intrageneic sequences|largest pairwise distance between intraspecific sequences; followed by the number of observations affected (in brackets).
Identification successes based on the BM and BCM criteria are shown in Figure 4. All species were correctly identified under BM for OTU and non-OTU analyses. Under BCM selection, 90.66% of samples were correctly identified when species were considered as OTUs, and 97.33% when species were not considered as OTUs. No matches were found for the remaining queries. Under the All Species Barcodes (ASB) identification criteria, an identification success of 90.66% and 61.33% were observed for OTU and non-OTU species categorization respectively. 36% of sequences were considered as ambiguous under the non-OTU criteria whereas the remaining queries were designated matchless.
Kappaphycus and Eucheuma samples were categorized under either Operational Taxonomic Units (OTU) or non-OTU conditions. Results were generated based on the Best Match (BM), Best Close Match (BCM) and All Species Barcodes (ASB) criteria.
Due to the similarity in tree topologies based on the NJ, ML, MP and BI algorithms, and for simplicity, resulting data are compiled and depicted as Figure 5. The NJ, MP and BI phylogenetic trees are provided as supporting data Figure S1, S2 and S3 respectively. Tree-based DNA identification is assessed using the NJ tree without taking into account the Solieria outgroup not used in the NJ dataset. In accordance to identification criteria by Hebert and co-workers , 100% identification success is reported when tree-based identification is based on the OTU concept. When taxa are queried using conventional taxonomic naming (non-OTU), percentage of successful identification is reduced to 94.6%, where the remaining 5.4% indicated misidentifications. Application of Meier et. al.'s  identification criteria returned comparatively lower successful identification rates, where 95.9% success and 4.1% ambiguity was recorded for OTUs; and 67.6% successful identification, 27% ambiguity and 5.4% misidentification for non-OTUs.
Number at nodes is arranged according to NJ bootstrap support/ML bootstrap support/MP bootstrap support/Bayesian posterior probabilities. Large dataset assessment: blue lines indicate Operational Taxonomic Units (OTU), whereas red lines represent non-OTU clusters. The Solieria outgroup was excluded in the NJ analysis.
Phylogenetic analyses of the entire cox2-3 spacer dataset is inclusive of the Solieria outgroup. A total of 32 most parsimonious trees were produced by Parsimony analysis, with a Retention Index (RI) and Consistency Index (CI) of 0.9748 and 0.7934 respectively. The tree topology is generally similar to the NJ trees generated for molecular marker assessment, with most specimens clustering into existing clades. Remaining GenBank samples, mostly of different species or locality, were inferred as additional, discrete monophyletic clades. Nevertheless, the overall tree topology is still congruent to that of phylogenetic trees reported in earlier studies , ,  and will not be discussed in detail. Nodal supports of the phylogenetic tree (Figure 5) were arranged in the order of Neighbor Joining bootstrap supports (NJ-BS), Maximum Likelihood bootstrap supports (ML-BS), Maximum Parsimony bootstrap supports (MP-BS) and Bayesian Inference posterior probabilities (BI-PP). Major nodes were mostly moderate to highly supported despite the addition in taxa. Fairly high support were observed for the monophyly of Kappaphycus (NJ-BS = 100%; ML-BS = 99%; MP-BS = 100%; BI-PP = 1.00) and Eucheuma (NJ-BS = 100%; ML-BS = 78%; MP-BS = 100%; BI-PP = 0.92) used in this study. Although strongly supported (NJ-BS = 91%; ML-BS = 96%; MP-BS = 94%; BI-PP = 1.00), polytomy is observed between K. alvarezii (KA1, KA2 and KA3) and K. striatum; where KA2 (NJ-BS = 67%; ML-BS = 74%; MP-BS = 63%; BI-PP = 1.00) and KA3 (NJ-BS = 99%; ML-BS = 100%; MP-BS = 98%; BI-PP = 1.00) represent K. alvarezii specimens from the African and Hawaiian regions respectively. Kappaphycus sp. “Aring-aring” (NJ-BS = 97%; ML-BS = 93%; MP-BS = 92%; BI-PP = 1.00) and Eucheuma denticulatum (NJ-BS = 100%; ML-BS = 100%; MP-BS = 100%; BI-PP = 1.00) were inferred to be monophyletic as well. Clade EP (NJ-BS = 100%; ML-BS = 100%; MP-BS = 100%; BI-PP = 1.00) comprises of Eucheuma platycladum collected from around Southeast Africa.
The dataset in the present study represents a small conglomerate of selected and commonly available Kappaphycus and Eucheuma samples from Southeast Asia. Sampling size is restricted to an amount supposedly cost-effective for molecular marker assessments as potential DNA barcodes. Owing to the relatively scarce records of Kappaphycus and Eucheuma in the GenBank, this study will also serve as a preliminary work to increase the amount of reference data using the potential DNA barcode, which can eventually be established as a barcode library.
Molecular Marker Assessment
The accuracy of a DNA barcode is largely determined by the magnitude of overlap between intraspecific variation and interspecific divergence. Ideally, the absence of an overlap would render species identification straightforward but this condition is virtually non-existent in very large datasets, in which the less the overlap, the more accurate it is for species identification , , . Results indicated that no overlaps were observed for the cox1, cox2, cox2-3 spacer and the rbcL genetic markers and have minimal impact on the accuracy of species identification for Kappaphycus and Eucheuma in this study (Figure 1). However, the absence of overlaps may be attributed to the small taxa limited to Southeast Asia (particularly so the genus Eucheuma) at this time, and is expected to change as more samples of different species or from different geographical locations are included in the future , . The presence of a “barcoding gap” (the absence of overlaps between intra- and interspecific genetic variations) may also be explained by the relatively lower genetic diversity amongst red algae as compared to those observed in arthropods.
All four molecular markers collectively showed an identification success of 100% for the BM criteria. Under the BCM criteria, the plastid encoded rbcL gene exhibited the highest identification success of 100% whereas the mitochondrial encoded cox2, cox2-3 spacer and cox1 spacer recorded slightly lower success of 96.6%, 93.1% and 79.3% respectively. Misidentification of species was not reported for all four molecular markers. The eventual increase in conspecific DNA sequences for Kappaphycus and Eucheuma is expected to reduce the probability of queries not meeting any matching sequences. All these results reflect the relatively less variable rbcL region as compared to mitochondrial counterparts, which was also reported by Geraldino et al.  and Yang et al. . Although widely championed as a good potential DNA barcode, the relatively higher intraspecific variation of the cox1 marker (<0.43% for Kappaphycus; <0.07% for Eucheuma) requires caution in avoiding misidentifications. Similar or higher intraspecific patterns were pointed out in previous studies associated with rhodophytes , , , , , .
Tree-based DNA identification approach returned 100% success in species identification for all four molecular markers using the criteria by Hebert and co-workers . Although the application of the relatively stricter identification criteria by Meier and co-workers  generally lowered the successful identification scores, they were still higher compared to the results derived from the distance-based approach. For instance, cox1 and cox2 showed a higher identification success of 89.7% and 96.6% respectively. Ambiguous identification in this study was mostly caused by queries that formed a sister group to a cluster of conspecific sequences. This is also expected to decrease as more reference sequences are deposited in the GenBank for Kappaphycus and Eucheuma. Contrary to the better results obtained by the tree-based method in this study, empirical studies involving much larger taxa coverage have reported the preference of distance-based assessment over tree-based ones in terms of accuracy and robustness , .
Based on both the distance and tree-based DNA identification approaches, the relatively conserved nature and the absence of overlaps between inter- and intraspecific genetic variability of the rbcL gene serves as a better potential DNA barcode for Kappaphycus and Eucheuma. However, the reduced genetic variation would also imply the incapability of the rbcL marker to detect incipient speciation or genetic diversity within species , . This will not only result in an underestimate of the actual genetic richness of these seaweeds, but will most likely overestimate interspecific variation as well due to the unavailability of closely related species . The drawbacks of the plastid encoded rbcL marker can be accounted for using supplementary molecular data generated from the relatively more variable, mitochondrial derived cox1, cox2 or cox2-3 spacer. The concept of combined molecular data, in this case of molecular markers, is not new considering the occasional failures of the DNA barcodes in correctly and consistently identifying species, as observed in Karner blue butterflies , seagrasses  and flowering plants . Utilization of genomic DNA from different origin i.e. mitochondrial, plastid or nuclear with different evolutionary rates would offer a better picture with respect to gene genealogy and evolutionary lineage.
The cox1 and cox2 DNA markers hold better potential as supplementary DNA barcodes compared to the cox2-3 spacer because of their protein coding properties. These markers are more conserved across taxa with less indel mutations and are much simpler to check for errors through amino-acid translation. Lack of recombination and uniparental inheritance of mitochondrial markers are added advantages . The relatively high mitochondrial copy numbers also enable ease of amplification. The cox2 genetic marker, with its moderately high interspecific divergences within generic level and non-existent intraspecific variation, has demonstrated relatively successful results in terms of species identification in this study and is thus regarded as a potential mitochondrial DNA barcode for Kappaphycus and Eucheuma. Large scale application of this molecular marker is expected to bring about advancements mainly in the field of agriculture, taxonomy and biomonitoring. However, considering the already widely utilized cox2-3 spacer exhibited almost similar barcoding traits as that of the mitochondrial cox2 marker, continual implementation is recommended, in spite of its non-coding nature. The implementation of the cox2 molecular marker across taxa may pose a problem as well since the full potential of the cox2 marker has yet to be tested in other rhodophytes. Additionally, the massively abundant and readily available cox1 sequences within the GenBank (although not extensively tested on Kappaphycus and Eucheuma at that time, cox1 has been proposed as the potential universal DNA barcode for red algae) is apparently more practical and economical to work with, despite the relatively higher intraspecific variations which may reduce the accuracy for species identification , , , , , , . Based on the results of this study, the authors have come to a consensus that the mitochondrial cox2-3 spacer was in overall the best potential DNA barcode for Kappaphycus and Eucheuma; whereas the combined mitochondrial-encoded cox1 and the plastid-encoded rbcL markers serve better as DNA barcodes encompassing the entire rhodophyte taxa.
The robustness and efficiency of DNA barcoding tends to increase with increased reference sequences and taxonomic scrutiny , , , . This includes genetically distinct individuals within a species' range to account for molecular markers with high variability such as the cox1 . Although this would greatly increase GenBank data and hence lead to an inevitable increment in terms of computational demand, larger and properly annotated datasets would expedite phylogeography, evolutionary biology, biodiversity or population genetic studies in the future.
Large Dataset Assessment
An additional 46 GenBank cox2-3 spacer sequences of Kappaphycus and Eucheuma were compiled to simulate the effectiveness of a molecular marker in large, empirical datasets. Certain specimens of unknown or uncertain identity i.e. Kappaphycus cottonii, Eucheuma isiforme etc. were not used to avoid data confounding. Morphological plasticity of these seaweeds has rendered species identification and description challenging, even to seasoned taxonomists. As of now, distinctive morphological characters are still undiscovered despite ample of DNA evidence supporting the possible existence of new, or perhaps cryptic species. This has led to the amplification of Operational Taxonomic Units (OTU) in this context; representing genotypic diversity possibly overlooked via conventional morphological traits.
Distance-based results on the cox2-3 spacer dataset have shown that the incorporation of more sequences decreases the “barcoding gap” (when not overlapping), and to the extent of forming overlaps between inter- and intraspecific divergence. When Kappaphycus and Eucheuma species were regarded as non-OTUs, the “barcoding gap” for the entire length of cox2-3 spacer genetic marker decreased from 0.27% to 0%. When under OTU assortment, the intra- and interspecific genetic divergences formed an overlap of 0.29%. These observations were not surprising considering the larger sample size would eventually lead to higher occurrence of specimens with varying genetic composition , . This would undoubtedly affect the identification accuracy of DNA barcoding. BCM results saw a slight dip from 93.1% to 90.7% when samples were regarded as OTUs. Similar patterns were observed for tree-based DNA identification under Meier's (2006)  criteria, where identification success reduced from 96.6% to 95.9% for non-OTUs, caused mainly by ambiguous sequences. These “singleton” sequences or query sequences sister to known species can be avoided with increased data and taxonomic rectification or reformation.
The distance-based All Species Barcodes (ASB) assessment, being relatively stricter compared to BM and BCM; has reflected our poor taxonomic comprehension on Kappaphycus and Eucheuma, in which at our current state (represented by non-OTU species assortment); a mere 61.3% of queries could be identified correctly using DNA barcoding. This figure is probably overestimated considering the large amount of Kappaphycus and Eucheuma yet to be discovered and described. Misidentifications under non-OTU conditions were caused by the African (KA2) and Hawaiian (KA3) K. alvarezii specimens, where the latter remained unresolved using the cox2-3 spacer and cox1 (results not shown) molecular markers. Still, combined cox2-3 spacer and RuBisCO spacer data has shown with moderate support that the KA3 specimens were more closely related to K. alvarezii than K. striatum . This implies the possible limitation in resolving power of individual markers at intraspecific levels although the rbcL and cox2 molecular markers are yet to be tested. On the other hand, ASB has returned relatively higher identification successes (90.7%) when Kappaphycus and Eucheuma species were regarded as OTUs, thus providing invaluable insights into the possible taxonomic structure of these rhodophytes in the future.
The inclusion of GenBank sequences did not cause significant alterations to the phylogenetic tree topology (Figure 2 and Figure 5), apart from the additional clade KA3 composed of Hawaiian Kappaphycus alvarezii and clade EP consisting of Eucheuma platycladum. The monophyly of both Kappaphycus (NJ-BS = 100%; ML-BS = 99%; MP-BS = 100%; BI-PP = 1.00) and Eucheuma (NJ-BS = 100%; ML-BS = 78%; MP-BS = 100%; BI-PP = 0.92) was moderate to highly supported. Taxonomic statuses of specimens excluded in this study i.e. Kappaphycus cottonii, Kappaphycus procrusteanus, Betaphycus philippinensis and Eucheuma isiforme remained unaddressed at this time. Solid inferences can only be made when type specimens are sequenced, or when more reference sequences are available.
Similar patterns of biodiversity as to that earlier reported were observed , , . K. alvarezii KA1, being the most widely distributed genotype, was reported throughout Southeast Asia, Africa, Columbia, Panama as well as Hawaii. Cox2-3 spacer genealogy suggested that the African and Hawaiian strains of K. alvarezii KA1 may possibly be introduced strains, presumably traceable back to the Philippines. This is not surprising as there were efforts to introduce foreign, good strains of Kappaphycus for cultivation in the past , , . Phylogenetic trees generated using the cox1, cox2 and rbcL DNA markers (Figure 2) have collectively displayed similar topologies constituted of similar specimens for clade KA1. This supports the report by Tan and co-workers (2012)  that local varieties of cultivated K. alvarezii do not differ genetically despite the distinctive morphologies, at least for Malaysian specimens. K. alvarezii KA2 and KA3 represent genotypes unique to Africa and Hawaii respectively. The taxonomic position of the Hawaiian K. “alvarezii” specimens (KA3) remained unresolved using the cox1 (results not shown) and cox2-3 spacer marker, where the Hawaiian samples occur as a polytomy to both K. alvarezii and K. striatum. Although GenBank sequences are limited, combined analysis using both the cox1 and cox2-3 spacer suggested that clade KA3 is sister to that of K. alvarezii and K. striatum (Results not shown). This contradicts the phylogenetic results based on the combined cox2-3 spacer and RuBiSCO spacer dataset, where the Hawaiian Kappaphycus specimens were inferred to be sister to K. alvarezii (KA1 and KA2 in this context) with moderate support (MP-BS = 70%) . Application of the mitochondrial cox2 and plastid rbcL marker may better elucidate the current confusion associated with clade KA3. As earlier reported, two genotypes were observed for Kappaphycus striatum , , denoted here as clades KS1 and KS2. No distinctive differences in terms of gross morphology were identified between these two genotypes as of now and it is unsure whether cryptic species may apply in this situation. However, it appears that cultivated K. striatum mostly reside within clade KS1 and are largely common in Southeast Asia, although a recent study has also reported its occurrence in Uroa, Tanzania as a result of strain introduction . Kappaphycus sp. “Aring-aring” from Malaysia was earlier shown to be phenotypically and genotypically different from K. alvarezii and K. striatum ; comparative studies against type specimens, particularly that of Kappaphycus cottonii (Weber-van Bosse) Doty ex P.C.Silva are currently underway in order to determine its validity as a new species.
Eucheumatoids are relatively poorly studied compared to Kappaphycus, possibly due to identification difficulties as well as lower economic value. Eucheuma denticulatum, being the more popularly cultivated species, were clustered into three genotypically distinct subclades. Subclade ED1 represents specimens from Southeast Asia and Hawaii, subclade ED2 from Southeast Asia, Hawaii and Tanzania  whereas samples within ED3 were exclusively from Africa. Despite coexisting in the South China and Celebes seas, Southeast Asian Eucheuma denticulatum ED1 and ED2 does not share similar morphological characteristics. This was shown by Ganzon-Fortes and co-workers , demonstrating the differences between the “Endong/Spaghetti” variety (ED2) from that of the usual Spinosum variety (ED1) of Eucheuma denticulatum. The “Endong” variety, thence named E. denticulatum (Burman) Collins & Hervey var. endong Trono & Ganzon-Fortes var. nov. exhibited smooth, slender terete axes with whorls of determinate branchlets at predictable intervals . Recent samplings in Malaysia have also revealed that the local Eucheuma denticulatum “Cacing” variety fits the morphological and biochemical descriptions of the “Endong” variety, and its genotypic affinity also supported by cox1 and rbcL molecular data (data not shown). The apparent and distinctive morphological characters of the “Endong” variety does not fit the original descriptions for E. denticulatum, thus suggesting that it may be a new species instead of a rare variety. With reference to Figure 2, this was shown to be potentially true using the full length cox1 genetic marker, where the monophyly of clades ED1 and ED2 were highly supported. Relatively lower nodal supports were displayed by the cox2-3 spacer and cox2 DNA markers, followed lastly by the rbcL marker which fails to clearly indicate monophyly of ED1 and ED2. These patterns are reflective of the genetic variability of each molecular marker and suggest that clades ED1 and ED2 are probably undergoing divergence or have recently diverged. E. denticulatum ED3 was inferred to share a common ancestry with E. denticulatum ED1 and ED2, and are to date only reported in Africa. Considering the significantly different morphologies reported for ED2, it would be interesting to relook into the detailed anatomy of the African E. denticulatum.
Neighbor-Joining (NJ) tree based on the cox2-3 spacer marker. Numeric values at nodes indicate NJ bootstrap supports. Large dataset assessment: blue lines indicate Operational Taxonomic Units (OTU), whereas red lines represent non-OTU clusters. The Solieria outgroup was omitted from the analysis to enable implementation of tree-based identification criteria.
Maximum Parsimony (MP) phylogenetic tree based on the cox2-3 spacer. Number at nodes indicates MP bootstrap supports. Large dataset assessment: blue lines indicate Operational Taxonomic Units (OTU), whereas red lines represent non-OTU clusters.
Bayesian (BI) phylogenetic tree based on the cox2-3 spacer DNA marker. Number at nodes indicates BI posterior probabilities. Large dataset assessment: blue lines indicate Operational Taxonomic Units (OTU), whereas red lines represent non-OTU clusters.
We would like to thank the Department of Fisheries Sabah, Prof. Dr. Suhaimi Md. Yasir, En. Adibi Rahiman bin Md. Nor, Dr. Yeong Hui Yin, Fiona Keng Seh Lin and Japson Wong for helping in seaweed collection. We thank Gaurav Vaidya for his support on using TaxonDNA. We would also like to thank Song Sze Looi, Ng Poh Kheng and Poong Sze Wan for their advice and guidance on molecular taxonomy. We are grateful to the University of Malaya and the Institute of Earth and Ocean Sciences (IOES) for research facilities.
Conceived and designed the experiments: PL JT. Performed the experiments: PL JT. Analyzed the data: PL JT. Contributed reagents/materials/analysis tools: PL JT SP DH HS AH. Wrote the paper: PL JT SP DH AH.
- 1. Hebert PDN, Cywinska A, Ball SL, de WaardJR (2003a) Biological identifications through DNA barcodes. Proceedings of the Royal Society B: Biological Sciences 270: 313–321.
- 2. Hebert PDN, Ratnasingham S, de WaardJR (2003b) Barcoding animal life: cytochrome c oxidase subunit 1 divergences among closely related species. Proceedings of the Royal Society B: Biological Sciences 270: S96–S99.
- 3. Hebert PDN, Stoeckle MY, Zemlak TS, Francis CM (2004) Identification of birds through DNA barcodes. PLoS Biol 2: e312.
- 4. Ellegren H, Hubert N, Hanner R, Holm E, Mandrak NE, et al. (2008) Identifying Canadian Freshwater Fishes through DNA Barcodes. PLoS ONE 3: e2490.
- 5. Hollingsworth PM, Graham SW, Little DP (2011) Choosing and Using a Plant DNA Barcode. PLoS ONE 6: e19254.
- 6. Jinbo U, Kato T, Ito M (2011) Current progress in DNA barcoding and future implications for entomology. Entomological Science 14: 107–124.
- 7. Ratnasingham S, Hebert PDN (2007) Molecular Ecology Notes. 7: 355–364BOLD: the Barcode of Life Data System (www.barcodinglife.org).
- 8. Sarkar IN, Trizna M (2011) The Barcode of Life Data Portal: Bridging the Biodiversity Informatics Divide for DNA Barcoding. PLoS ONE 6: e14689.
- 9. Wong LL, Peatman E, Lu J, Kucuktas H, He S, et al. (2011) DNA Barcoding of Catfish: Species Authentication and Phylogenetic Assessment. PLoS ONE 6: e17812.
- 10. Freshwater DW, Tudor K, O'Shaughnessy K, Wysor B (2010) DNA barcoding in the red algal order Gelidiales: comparison of COI with rbcL and verification of the “barcoding gap”. Cryptogamie, Algologie 31: 435–449.
- 11. Clarkston BE, Saunders GW (2010) A comparison of two DNA barcode markers for species discrimination in the red algal family Kallymeniaceae (Gigartinales, Florideophyceae), with a description of Euthora timburtonii sp. nov. Botany 88: 119–131.
- 12. Le Gall L, Saunders GW (2010) DNA Barcoding Is a Powerful Tool to Uncover Algal Diversity: A Case Study of the Phyllophoraceae (Gigartinales, Rhodophyta) in the Canadian Flora. Journal of Phycology 46: 374–389.
- 13. Saunders GW (2008) A DNA barcode examination of the red algae family Dumontiaceae (Gigartinales, Florideophyceae) in Canadian waters reveals substantial cryptic species diversity. 1. The foliose Dilsea-Neodilsea complex and Weeksia. Botany 86: 773–789.
- 14. Kim MS, Yang MY, Cho GY (2010) Applying DNA barcoding to Korean Gracilariaceae (Rhodophyta). Cryptogamie, Algologie 31: 387–401.
- 15. Saunders GW (2009) Routine DNA barcoding of Canadian Gracilariales (Rhodophyta) reveals the invasive species Gracilaria vermiculophylla in British Columbia. Molecular Ecology Resources 9: 140–150.
- 16. McDevit DC, Saunders GW (2010) A DNA barcode examination of the Laminariaceae (Phaeophyceae) in Canada reveals novel biogeographical and evolutionary insights. Phycologia 49: 235–248.
- 17. Kucera H, Saunders GW (2008) Assigning morphological variants of Fucus (Fucales, Phaeophyceae) in Canadian waters to recognized species using DNA barcoding. Botany 86: 1065–1079.
- 18. Robba L, Russell SJ, Barker GL, Brodie J (2006) Assessing the use of the mitochondrial cox1 marker for use in DNA barcoding of red algae (Rhodophyta). American Journal of Botany 93: 1101–1108.
- 19. Saunders GW (2005) Applying DNA barcoding to red macroalgae: a preliminary appraisal holds promise for future applications. Philosophical Transactions of the Royal Society B: Biological Sciences 360: 1879–1888.
- 20. Sherwood AR, Kurihara A, Conklin KY, Sauvage T, Presting GG (2010) The Hawaiian Rhodophyta Biodiversity Survey (2006-2010): a summary of principal findings. BMC Plant Biology 10: 258.
- 21. Bixler HJ, Porse H (2010) A decade of change in the seaweed hydrocolloids industry. Journal of Applied Phycology 23: 321–335.
- 22. Bindu MS, Levine IA (2010) The commercial red seaweed Kappaphycus alvarezii–an overview on farming and environment. Journal of Applied Phycology.
- 23. Conklin KY, Kurihara A, Sherwood AR (2009) A molecular method for identification of the morphologically plastic invasive algal genera Eucheuma and Kappaphycus (Rhodophyta, Gigartinales) in Hawaii. Journal of Applied Phycology 21: 691–699.
- 24. Doty MS (1985a) Eucheuma alvarezii, sp. Nov. (Gigartinales, Rhodophyta) from Malaysia. In: Abbott IA, Norris JN, editors. Taxonomy of economic seaweeds: with reference to some Pacific and Caribbean spcies. La Jolla, California: California Sea Grant College Program. 37–45.
- 25. Doty MS, Norris JN (1985b) Eucheuma species (Solieriaceae, Rhodophyta) that are major sources of carrageenan. In: Abott IA, Norris JN, editors. Taxonomy of economic seaweeds: with reference to some Pacific and Caribbean species. La Jolla, California: California Sea Grant College Program. 47–61.
- 26. Paula EJ, Pereira R, Ohno M (1999) Strain selection in Kappaphycus alvarezii var. alvarezii (Solieriaceae, Rhodophyta) using tetraspore progeny. Journal of Applied Phycology 11: 111–121.
- 27. Phang SM, Yeong HY, Lim PE, Adibi Rahiman Md Nor, Gan KT (2010) Commercial varieties of Kappaphycus and Eucheuma in Malaysia. Malaysian Journal of Science 29: 214–223.
- 28. Zuccarello GC, Critchley AT, Smith J, Sieber V, Lhonneur GB, et al. (2006) Systematics and genetic variation in commercial Kappaphycus and Eucheuma (Solieriaceae, Rhodophyta). Journal of Applied Phycology 18: 643–651.
- 29. Tan J, Lim PE, Phang SM (2012) Phylogenetic relationship of Kappaphycus Doty and Eucheuma J. Agardh (Solieraceae, Rhodophyta) in Malaysia. Journal of Applied Phycology DOI: 10.1007/s10811-012-9833-1.
- 30. Fredericq S, Freshwater DW, Hommersand MH (1999) Observations on the phylogenetic systematics and biogeography of the Solieriaceae (Gigartinales, Rhodophyta) inferred from rbcL sequences and morphological evidence. Hydrobiologia 398/399: 25–38.
- 31. Zhao S, He P (2011) Molecular identification based on ITS sequences for Kappaphycus and Eucheuma cultivated in China. Chinese Journal of Oceanology and Limnology 29: 1287–1296.
- 32. Halling C, Wikström SA, Lilliesköld-Sjöö G, Mörk E, Lundsør E, et al.. (2012) Introduction of Asian strains and low genetic variation in farmed seaweeds: indications for new management practices. Journal of Applied Phycology In press.
- 33. Meier R, Shiyang K, Vaidya G, Ng P (2006) DNA Barcoding and Taxonomy in Diptera: A Tale of High Intraspecific Variability and Low Identification Success. Systematic Biology 55: 715–728.
- 34. Leblanc C, Boyen C, Richard O, Bonnard G, Grienenberger JM, et al. (1995) Complete sequence of the mitochondrial DNA of the rhodophyte Chondrus crispus (Gigartinales). Gene content and genome organization. Journal of Molecular Biology 250: 484–495.
- 35. Larkin MA, Blackshields G, Brown NP, Chenna R, McGettigan PA, et al. (2007) ClustalW and ClustalX version 2.0. Bioinformatics 23: 2947–2948.
- 36. Meyer CP (2004) Toward comprehensiveness: Increased molecular sampling within Cypraeidae and its phylogenetic implications. Malacologia 46: 127–156.
- 37. Swofford DL (2003) PAUP*. Phylogenetic Analysis Using Parsimony (* and other methods). Version 4. Sinauer Associates, Sunderland, Massachusetts.
- 38. Rambaut A, Drummond A. FigTree v1.3.1. Institute of Evolutionary Biology, University of Edinburgh, Edinburgh, United Kingdom 2010. http://tree.bio.ed.ac.uk/software/figtree/.
- 39. Zuccarello GC, West JA (2006) Molecular phylogeny of the subfamily Bostrychioideae (Ceramiales, Rhodophyta): subsuming Stictosiphonia and highlighting polyphyly in species of Bostrychia. Phycologia 45: 24–36.
- 40. Silvestro D, Michalak I (2011) raxmlGUI: a graphical front-end for RAxML. Organisms Diversity and Evolution. 10.1007/s13127-011-0056-0.
- 41. Huelsenbeck JP, Ronquist F, Nielsen R, Bollback JP (2001) Bayesian inference of phylogeny and its impact on evolutionary biology. Science 294: 2310–23114.
- 42. Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572–1574.
- 43. Tanabe AS (2011) Kakusan4 and Aminosan: two programs for comparing nonpartitioned, proportional and separate models for combined molecular phylogenetic analyses of multilocus sequence data. Molecular Ecology Resources 11: 914–921.
- 44. Meyer CP, Paulay G (2005) DNA Barcoding: Error rates based on comprehensive sampling. PLoS Biol 3: e422.
- 45. Meier R, Zhang G, Ali F (2008) The Use of Mean Instead of Smallest Interspecific Distances Exaggerates the Size of the “Barcoding Gap” and Leads to Misidentification. Systematic Biology 57: 809–813.
- 46. Geraldino PJL, Yang EC, Boo SM (2006) Morphology and Molecular Phylogeny of Hypnea flexicaulis (Gigartinales, Rhodophyta) from Korea. Algae 21: 417–423.
- 47. Yang EC, Kim MS, Geraldino PJL, Sahoo D, Shin J-A, et al. (2007) Mitochondrial cox1 and plastid rbcL genes of Gracilaria vermiculophylla (Gracilariaceae, Rhodophyta). Journal of Applied Phycology 20: 161–168.
- 48. Geraldino PJL, Yang EC, Kim MS, Boo SM (2009) Systematics of Hypnea asiatica sp. nov. (Hypneaceae, Rhodophyta) based on morphology and nrDNA SSU, plastid rbcL, and mitochondrial cox1. Taxon 58: 606–616.
- 49. Wiriyadamrikul J, Park JK, Lewmanomont K, Boo SM (2010) Additional records of Gelidiella fanii (Gelidiales, Rhodophyta) from the western Pacific based on morphology, rbcL and cox1 analyses. Botanica marina 53: 343–350.
- 50. Virgilio M, Backeljau T, Nevado B, De Meyer M (2010) Comparative performances of DNA barcoding across insect orders. BMC Bioinformatics 11: 206.
- 51. van Velzen R, Weitschek E, Felici G, Bakker FT (2012) DNA Barcoding of Recently Diverged Species: Relative Performance of Matching Methods. PLoS ONE 7: e30490.
- 52. Gompert Z, Nice CC, Fordyce JA, Forister ML, Shapiro AM (2006) Identifying units for conservation using molecular systematics: the cautionary tale of the Karner blue butterfly. Molecular Ecology 15: 1759–1768.
- 53. DeSalle R, Lucas C, Thangaradjou T, Papenbrock J (2012) Development of a DNA Barcoding System for Seagrasses: Successful but Not Simple. PLoS ONE 7: e29987.
- 54. Kress WJ (2005) Use of DNA barcodes to identify flowering plants. Proceedings of the National Academy of Sciences 102: 8369–8374.
- 55. Zuccarello GC, Burger G, West JA, King RJ (1999a) A mitochondrial marker for red algal intraspecific relationships. Molecular Ecology 8: 1443–1447.
- 56. Pickering TD, Skelton P, Sulu RJ (2007) Intentional introductions of commercially harvested alien seaweeds. Botanica Marina 50: 338–350.
- 57. Ganzon-Fortes ET, Trono GC, Villanueva RD, Romero JB, Montaño MNE (2011) ‘Endong’, a rare variety of the farmed carrageenophyte Eucheuma denticulatum (Burman) Collins & Hervey from the Philippines. Journal of Applied Phycology DOI: 10.1007/s10811-011-9740-x.
- 58. Freshwater DW, Rueness J (1994) Phylogenetic relationships of some European Gelidium (Gelidiales, Rhodophyta) species, based on rbcL nucleotide sequence analysis. Phycologia 33: 187–194.
- 59. Gavio B, Fredericq S (2002) Grateloupia turuturu (Halymeniaceae, Rhodophyta) is the correct name of the non-native species in the Atlantic known as Grateloupia doryphora. European Journal of Phycology 37: 349–359.