Distinguishing Anuran species by high‐resolution melting analysis of the COI barcode (COI‐HRM)

Abstract Taxonomic identification can be difficult when two or more species appear morphologically similar. DNA barcoding based on the sequence of the mitochondrial cytochrome c oxidase 1 gene (COI) is now widely used in identifying animal species. High‐resolution melting analysis (HRM) provides an alternative method for detecting sequence variations among amplicons without having to perform DNA sequencing. The purpose of this study was to determine whether HRM of the COI barcode can be used to distinguish animal species. Using anurans as a model, we found distinct COI melting profiles among three congeners of both Lithobates spp. and Hyla spp. Sequence variations within species shifted the melting temperature of one or more melting domains slightly but do not affect the distinctness of the melting profiles for each species. An NMDS ordination plot comparing melting peak profiles among eight Anuran species showed overlapping profiles for Lithobates sphenocephala and Gastrophryne carolinensis. The COI amplicon for both species contained two melting domains with melting temperatures that were similar between the two species. The two species belong to two different families, highlighting the fact that COI melting profiles do not reveal phylogenetic relationships but simply reflect DNA sequence differences among stretches of DNA within amplicons. This study suggests that high‐resolution melting analysis of COI barcodes (COI‐HRM) may be useful as a simple and rapid method to distinguish animal species that appear morphologically similar.


| INTRODUC TI ON
Taxonomic identification can be difficult when two or more species appear morphologically similar. For example, the frogs Hyla versicolor and Hyla chrysoscelis share breeding ponds (Harding, 1997) but appear identical and can only be differentiated by their mating calls and karyotype (Collins & Conant, 1998;Jaslow & Vogt, 1977). Among five species of Radix snails, intraspecific variability in shell morphology and anatomy overlap among species (Glöer, 2002). For many species, the difficulty is more acute for early life stages such as embryos and juveniles when definitive morphological characteristics are not yet established. DNA-based species identification, however, is extremely reliable. One widely used approach is DNA barcoding using a region of the mitochondrial cytochrome oxidase subunit I (COI) gene for animal species (Hebert, Cywinska, & Ball, 2003). The gene is PCR-amplified, sequenced, and then compared to a sequence database for species identification.
An alternative to DNA sequencing for detecting variations among PCR amplicon sequences is high-resolution melting analysis (HRM). During HRM, the fluorescence of PCR amplicons in the presence of a dsDNA saturating dye is continuously monitored as temperature increases at predetermined increments. Because there is a loss in fluorescence when double-stranded DNA denatures and the temperature at which DNA melts is sequence-dependent, unique amplicons produce a unique melting curve. The first derivative of each melting curve can then be plotted against temperature to produce unique melting peak profiles that may be useful for species identification.
The effectiveness of HRM for distinguishing related species is unclear. Of concern is the effect of amplicon size. With primers LCO1490 and HC02198, frequently used to DNA barcode animals, a 658-bp region of interest within the COI gene is amplified (Folmer, Black, Hoeh, Lutz, & Vrijenhoek, 1994). For screening single-nucleotide polymorphism, sensitivity and specificity is greater for PCR products of 300 bp or less compared to larger products (Reed & Wittwer, 2004). However, large PCR products can produce melting peak profiles with multiple melting peaks (Lilliebridge, Tong, Giffard, & Holt, 2011;Rasmussen, Saint, & Monis, 2007) that aid in differentiating sequence variants. Indeed, detection of virus serotypes using HRM analysis of a ~600 bp PCR product has been reported (Steer, Kirkpatrick, O'Rourke, & Noormohammadi, 2009).
The goal of the present study was to test the usefulness of high-resolution melting analysis of the COI barcode (COI-HRM) to distinguish species using anurans as a model. The sequence of the COI gene has been shown to be effective for identifying anurans to species (Perl et al., 2014;Smith, Poyarkov, & Hebert, 2008;Vences, Thomas, Bonett, & Vieites, 2005). Using DNA collected from eight different species in Hattiesburg, Mississippi, we found that PCR-HRM of the COI barcode produced melting peak profiles that were highly reproducible and useful for distinguishing species.

| Tissue collection/DNA extraction
Tissue samples from three adult frogs of each species (Table 1)

| PCR/HRM
PCR and HRM, using the Rotor-Gene 6000 (Corbett Life Sciences, now Qiagen), was performed in a 25 μl reaction volume containing 12.5 μl of EconoTaq PLUS 2X Master Mix (Lucigen), 5 μl of 10 ng/ μl extracted DNA, 2 μl each of 5 μM forward and reverse primers Initial melting took place at 95°C for 5 min followed by 35 cycles of melting at 94°C for 1 min, annealing at 46°C for 1 min, extension at 72°C for 1 min, and a final extension at 72°C for 10 min. High-resolution melting analysis was performed immediately after completion of PCR. Sample fluorescence was acquired from 65°C to 95°C at 0.2°C increments, five seconds after each temperature increase had been reached. To better visualize the effect of temperature on DNA melting and the possible presence of multiple melting domains, first derivatives of the change in sample fluorescence over time (−dF/dT) were calculated using the Rotor-Gene 6000 Series Software (version 1.7) at each 0.2°C increment and plotted against temperature.

| PERMANOVA/NMDS
To compare the DNA melting profiles statistically, the relative first derivative of sample fluorescence at each temperature increment between 75°C and 90°C was used to construct dissimilarity matrices based on the Euclidean distance metric. Permutation analysis of variance (PERMANOVA) (Anderson, 2001) with 10,000 permutations was used to test for significant differences among the melting peak profiles of members of both the genus Lithobates and Hyla.
PERMANOVAs were performed using the adonis function of the vegan package (Oksanen et al., 2015) in R. p-values < .05 were considered significant. Nonmetric multidimensional scaling (NMDS), retaining two dimensions, was performed using the metaMDS function of the vegan package in R. Confidence ellipses were drawn at the 50% level to aid in visualizing clusters using the car package (Fox & Weisberg, 2011) in R.

| DNA sequencing
PCR products were sequenced commercially (Eurofins MWG Operon LLC) using the Sanger sequencing. Unincorporated nucleotides and primers remaining in PCR products were digested prior to sequencing using ExoSAP-IT (Affymetrix) according to the manufacturer's protocol. PCR products were sequenced in both directions with primers used during PCR. Primer sequences were TA B L E 1 Anuran taxa used in the study

| RE SULTS
A single amplicon, 706 bp in length, was amplified from each species collected (data not shown). The amplicons contained up to three melting domains that produced melting profiles that can be used to clearly distinguish congeneric Lithobates (Figure 1a) and

| D ISCUSS I ON
Accurate species identification is essential in fields such as community ecology and conservation biology. For some animals, the task is not possible based on appearance either because morphological differences overlap or because of morphological variability.
For example, the earthworms Lumbricus terrestris and Lumbricus herculeus appear to differ in segment number, body mass, and body length but identification is not reliable because their measurements overlap (James et al., 2010). Snails in the genus Radix cannot be distinguished based on shell morphology because it is variable, overlapping among species and is phenotypically plastic depending on environmental conditions (Pfenninger, Cordellier, & Streit, 2006). The problem is especially acute among embryonic and larval forms before the development of morphological characteristics that distinguish adults.
To distinguish morphologically similar species, DNA barcodes based on the sequence of the mitochondrial cytochrome oxidase subunit I (COI) gene (Hebert et al., 2003)  The melting peak profiles of the COI amplicon appeared highly distinguishable among species for both Lithobates (Figure 1a) and Hyla (Figure 1b) suggesting that HRM analysis may be a useful way to distinguish related species. Different regions of the COI amplicon melted at different temperatures, forming melting domains F I G U R E 3 Nonmetric multidimensional scaling ordination plot, using Euclidean distance as the distance metric, summarizing the difference in melting peak profiles among the eight anuran species. Euclidean distance was calculated after technical duplicates were averaged and the relative first derivative values of fluorescence (−dF/dT) were obtained for each specimen. Squares denote sequence identity within the species The melting profile of amplified DNA depends on sequence, length, GC content, and heterozygosity (Farrar & Wittwer, 2017;Reed, Kent & Wittwer, 2007;Steer et al., 2009)  In conclusion, we used anurans to show that HRM analysis of the COI barcode can be used to distinguish related species. COI-HRM analysis offers no advantage over traditional sequencing. However, it is much faster because there is no need to do sequencing. It is a closed-tube method by which the PCR and HRM steps are performed in the same reaction tube thereby eliminating the need for any post-PCR manipulations. Although the cost of DNA sequencing is now relatively low, it can be a barrier for laboratories that are not well funded or for laboratories with large numbers of samples to study. After DNA extraction, the analysis only takes a few hours to complete and species differentiation relies only on visual comparison of the melting peak profiles. The only requirement is that the user has access to a thermal cycler capable of real-time fluorescence measurements.

ACK N OWLED G EM ENTS
The authors would like to thank Dr. Jennifer Lamb for her assistance in collecting and confirming the identity of anuran specimens.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
S.E. and S.Y.W. conceived the study. S.E. obtained specimens and per-