Molecular Ecology Resources (2008) 8, 247–255 doi: 10.1111/j.1471-8286.2007.01996.x DNABlackwell Publishing Ltd BARCODING Barcoding corals: limited by interspecific divergence, not intraspecific variation T. L. SHEARER* and M. A. COFFROTH† *School of Biology, Georgia Institute of Technology, 310 Ferst Dr., Atlanta, GA 30332-0230, USA, †Department of Geology, State University of New York at Buffalo, 109 Cooke Hall, Buffalo, NY 14260, USA Abstract The expanding use of DNA barcoding as a tool to identify species and assess biodiversity has recently attracted much attention. An attractive aspect of a barcoding method to iden- tify scleractinian species is that it can be utilized on any life stage (larva, juvenile or adult) and is not influenced by phenotypic plasticity unlike morphological methods of species identification. It has been unclear whether the standard DNA barcoding system, based on cytochrome c oxidase subunit 1 (COI), is suitable for species identification of scleractinian corals. Levels of intra- and interspecific genetic variation of the scleractinian COI gene were investigated to determine whether threshold values could be implemented to discriminate conspecifics from other taxa. Overlap between intraspecific variation and interspecific diver- gence due to low genetic divergence among species (0% in many cases), rather than high levels of intraspecific variation, resulted in the inability to establish appropriate threshold values specific for scleractinians; thus, it was impossible to discern most scleractinian species using this gene. Keywords: COI, DNA barcoding, interspecific divergence, intraspecific variation, scleractinian coral, species identification Received 24 January 2007; revision accepted 15 August 2007 DNA barcoding has become an attractive, yet controversial et al. 2003a, b; 2004a; Hajibabaei et al. 2006; Smith et al. 2006; means to identify species over a wide taxonomic range. Witt et al. 2006) and invasive species (e.g. Siddall & Budinoff Such methods utilize a short DNA sequence, rather than 2005), and to assess biodiversity (Janzen et al. 2005; Smith multiple traditional taxonomic characters to characterize et al. 2005). This barcoding system relies on the observation species. Hence, the attraction — rapid and inexpensive gen- that COI sequence divergence between most congeneric eration of sequences with little expertise required, and the species is generally greater than 2% (Hebert et al. 2003b), controversy — species identification determined by only a whereas intraspecific variation is often lower than 1% single complex character rather than multiple taxonomic (Avise 2000). This gap between inter- and intraspecific characters which require expert knowledge of the species variation, or the ‘barcoding gap’ is presumed to be suffi- in question (e.g. Seberg et al. 2003; Will & Rubinoff 2004; ciently wide such that threshold values can be implemented Dasmahapatra & Mallet 2006). to distinguish conspecifics from other taxa (Meyer & Paulay The DNA barcoding system proposed by Hebert et al. 2005). Although this holds true for some taxa (e.g. birds, (2003a, b), based on ~650 bp sequence of the mitochondrial Hebert et al. 2004b; spiders, Barrett & Hebert 2005; butterflies, cytochrome c oxidase subunit 1 (COI) gene, can readily Hajibabaei et al. 2006; fish, Ward et al. 2005), one criticism of discriminate among closely related species across most the barcoding system has been that intraspecific variation animal phyla. In several cases, this system has been utilized may overlap interspecific divergence, resulting in unreliable to distinguish known species (Hebert et al. 2003a, b; Ward taxonomic designations (Meyer & Paulay 2005). Substantial et al. 2005), to identify new or cryptic species (e.g. Hebert overlap has been observed in some taxonomic groups as a result of high levels of intraspecific COI sequence diver- Correspondence: TL Shearer, Fax: (404) 385 4440; E-mail: gence (e.g. cowries, Meyer & Paulay 2005; amphibia, Vences [email protected] et al. 2005). Overlap of inter- and intraspecific variation is © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd 248 DNA BARCODING less frequently due to low levels of interspecific sequence and 0.2 mm of each primer. The forward primer was a divergence. This variability in the rate of evolution among shortened version of the Folmer et al. (1994) universal taxa indicates that no standard level of sequence diver- primer, LCOI1490: 5′-CAAATCATAAAGATATTGG-3′. The gence can be implemented to discern species (Will & reverse primer was the Folmer et al. (1994) universal primer, Rubinoff 2004) and the utility of this barcoding system HCOI2198: 5′-TAAACTTCAGGGTGACCAAAAAATCA-3′. should be evaluated on a taxon-by-taxon basis. These primers amplified 705 base pairs (bp) of the COI gene A genetic identification system would be particularly in the coral mitochondrial genome. Thermal cycling protocol useful for organisms such as scleractinian corals, whose conditions consisted of a denaturing step of 2 min at 95 °C identification may be uncertain due to phenotypic plastic- followed by 40 cycles of 95 °C for 30 s, 47 °C for 90 s and ity, morphologically cryptic species or lack of identification 72 °C for 90 s. The amplified products were purified through tools at various life stages (Hebert et al. 2003a). This system a 2.0% agarose gel in 1× TAE buffer. Reactions for cycle would be an extremely useful tool to characterize morpho- sequencing were performed using the SequiTherm EXCEL logical variants of a species, identify cryptic or endemic II DNA Sequencing Kit-LC (Epicentre Technologies) accord- species and rapidly assess and compare biodiversity of coral ing to modifications specified by LI-COR Biotechnology reefs. A genetic method to identify newly settled Caribbean Division. COI sequences were read in both directions on a scleractinian recruits has recently been described, in part, LI-COR Gene ReadIR 4200 automatic DNA sequencer. COI to overcome the difficulty in morphological identification sequences were submitted to GenBank (Table 1, Reference 1). of this early life stage (Shearer & Coffroth 2006). In that In addition to the sequences obtained in this study, study, the COI gene was useful in resolving species for 97 COI sequences from Caribbean, western Atlantic and Indo- some coral genera with high recruitment rates on artificial Pacific scleractinians were downloaded from GenBank substrate (agaricids and poritids). This was possible due to (Table 1). All sequences were aligned manually using seqapp the relatively small number of Caribbean scleractinian spe- 1.9a (Gilbert 1992) and trimmed to the minimum number of cies and unusually low levels of intraspecifc mitochondrial corresponding nucleotides. Pairwise Kimura’s 2-parameter sequence variation observed within corals often resulting (K2P) genetic distances among all species, within families, in a species-specific genetic signal (Shearer et al. 2002). Low within genera and within species were calculated using levels of COI sequence differentiation within coral species paup* (version 4.0b10 Swofford 2002). K2P is the best DNA should improve the capacity of this gene to distinguish substitution model for low genetic distances (Nei & Kumar species by minimizing the ‘noise’ created by overlapping 2000). To determine levels of intraspecific nucleotide vari- intra- and interspecific genetic distances. Limited taxonomic ation, sequences from multiple colonies of 27 species were comparisons within the phylum Cnidaria, within which analysed (Table 2). scleractinians are classified, indicate that nucleotide poly- morphism in COI is insufficient to discern some closely Results related species (Shearer et al. 2002; Hebert et al. 2003b); however, several species can be discerned using this gene COI sequences generated from this study were obtained from (Shearer & Coffroth 2006). Levels of interspecific diver- 49 samples, comprising 30 Caribbean species representing gence must be more thoroughly examined to assess the 17 genera in 11 families (Table 1, Reference 1). The 705 bp potential for DNA barcoding in corals. This study evaluates COI sequences were aligned with scleractinian sequences whether intra- and interspecific variation in COI of Carib- downloaded from GenBank (Table 1) and trimmed to a bean, western Atlantic and Indo-Pacific scleractinian corals 615 bp sequence containing no indels. In total (newly gener- is amenable to establishing threshold values critical to ated sequences plus those downloaded from GenBank), distinguish species using this barcoding gene. 40 Caribbean, four western Atlantic and 46 Indo-Pacific scleractinian coral species, representing 44 genera in 14 families (Table 1) were analysed. To ensure COI sequences Materials and methods were cnidarian in origin and not from symbiotic dino- Tissue from 30 Caribbean coral species (Table 1, Reference 1) flagellates associated with corals, homology of nucleotide was collected and preserved in high salt solution (250 mm sequences with azooxanthellate cnidarian COI (Metridium EDTA, 20% DMSO, saturated NaCl, pH 7.5), 95% ethanol senile GenBank Accession no. U36783) was confirmed (data or frozen in liquid nitrogen. DNA was isolated following not shown). the protocol described in Shearer et al. (2005) or Coffroth Intraspecific nucleotide variation was low (mean 0.048%, et al. (1992) and frozen at –20 °C until amplified. SE ± 0.017) even at third codon positions, with sequence Amplifications were performed in 10 μL volume