Helicoverpa and Heliothis species include some of the world’s most significant crop pests, causing billions of dollars of losses globally. As such, a number are regulated quarantine species. For quarantine agencies, the most crucial issue is distinguishing native species from exotics, yet even this task is often not feasible because of poorly known local faunas and the difficulties of identifying closely related species, especially the immature stages. DNA barcoding is a scalable molecular diagnostic method that could provide the solution to this problem, however there has been no large-scale test of the efficacy of DNA barcodes for identifying the Heliothinae of any region of the world to date. This study fills that gap by DNA barcoding the entire heliothine moth fauna of Australia, bar one rare species, and comparing results with existing public domain resources. We find that DNA barcodes provide robust discrimination of all of the major pest species sampled, but poor discrimination of Australian Heliocheilus species, and we discuss ways to improve the use of DNA barcodes for identification of pests.
Citation: Mitchell A, Gopurenko D (2016) DNA Barcoding the Heliothinae (Lepidoptera: Noctuidae) of Australia and Utility of DNA Barcodes for Pest Identification in Helicoverpa and Relatives. PLoS ONE 11(8): e0160895. https://doi.org/10.1371/journal.pone.0160895
Editor: Daniel Doucet, Natural Resources Canada, CANADA
Received: May 6, 2016; Accepted: July 26, 2016; Published: August 10, 2016
Copyright: © 2016 Mitchell, Gopurenko. 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.
Data Availability: The complete DNA alignment (all that is required to replicate our results) is supplied as supplementary data. Additional data such as sequence trace files, specimens collection data and specimen images, are available from BOLD as public project HELAU. All of the new DNA sequences produced for this paper have been submitted to GenBank and Accession numbers appear in the paper.
Funding: This work was supported by NSW BioFirst Initiative and Australian Museum Research Institute.
Competing interests: The authors have declared that no competing interests exist.
The Heliothinae is a cosmopolitan subfamily of Noctuidae containing some 365 described species worldwide . The larvae feed on flowers and fruits of herbaceous plants and include a number of the world's worst agricultural pests, such as Heliothis and Helicoverpa species. Pest control is dependent on rapid and accurate species identification, however, closely related species of Heliothinae may be impossible to distinguish without genitalic dissections. While the phylogeny of Heliothinae has been relatively well studied using both molecules and morphology [2, 3, 4, 5, 6], providing a robust framework for research on heliothine genomics and biology, species diagnostics has lagged behind. Identifying Helicoverpa species, for example, is at best a highly specialized task, and all too often impossible for the immature stages. The obvious solution is DNA-based diagnostics, developed within a rigorous taxonomic framework. Published molecular diagnostic studies of heliothine pests have each focussed only on pairs of species. This is surprising given the global economic importance of this group and its diversity of species of agricultural significance which are found in six genera.
Helicoverpa armigera Hübner (1809), known as the Cotton Bollworm in Australia and as the Old World Bollworm in the Americas, was estimated to cost the Australian cotton industry $225 million annually  despite the successful introduction of transgenic cotton engineered to resist this pest in Australia ten years prior. H. armigera is highly polyphagous and has been recorded from around 50 different families of plants and many major crops including cotton, soybeans, corn, tobacco, tomatoes and others. Adults of H. armigera are morphologically very similar to H. zea (Boddie 1850) which is its likely sister species . There are no known morphological characters to separate larvae of these species  and consequently there have been no resources for non-specialist identification of these species, until very recently. What prompted recent research in this area was the detection of H. armigera in Brazil, however the species seems to have been present in the country for some time when it was detected, as it had spread widely and was reported to have reduced crop yields in the 2012–2013 season by 35%, causing $1 billion in damage . This is a timely reminder of the need for species diagnostic methods that work on all life stages. Clearly there remains a global need for molecular diagnostic methods that can reliably distinguish species of Helicoverpa and other heliothine pests.
Modern systematics studies of heliothines began with the morphological work of Hardwick , refined by Matthews , Poole  and others. Matthews  revised the Australian heliothine fauna, describing eight new species. Cho et al.  initiated molecular systematics studies of the higher-level phylogeny of the group, culminating in a study based on DNA sequence data for two nuclear genes (EF-1α and DDC) and one mitochondrial gene (COI), resolving many of the outstanding questions in heliothine higher-level systematics . Diagnosis and identification of heliothine species is less well developed, despite their enormous economic importance, as many of the species are difficult to distinguish, there being little morphological variation among species, especially for larvae. This is particularly true of Australian Heliocheilus to the extent that “it is not possible to distinguish the species on the basis of the male genitalia” (Matthews 1999) and identifications require a series of specimens.
In addition to 24 species of Heliocheilus, the Australian Heliothinae include two species of Adisura, three Australothis, three Heliothis and five of the world’s approximately 20 species of Helicoverpa, for a total of 37 species. Three Australian Helicoverpa species are pests, including H. armigera, the Native Budworm H. punctigera (Wallengren, 1860) and the Oriental Tobacco Budworm H. assulta (Guenée, 1852) while Heliothis punctifera (Walker 1857) is a polyphagous minor pest species. Australothis rubrescens (Walker 1858) has a broad diet which overlaps with Helicoverpa species and its larvae may be confused with Helicoverpa. Other pest heliothines around the world include H. zea (Boddie, 1850), H. gelotopoeon (Dyar, 1921), Chloridea virescens (Fabricius, 1777) (previously Heliothis virescens, but the genus Chloridea was reinstated ) and various species of Adisura, Heliothis, Heliocheilus and Masalia.
DNA barcoding could provide an efficient way to identify heliothine species but it remains to be tested and implemented in a comprehensive manner. Previous studies developing DNA sequence data for Helicoverpa species have focussed on local needs, distinguishing usually just two species although sometimes with up to two other species included as outgroups [9, 12, 13, 14, 15, 16, 17, 18, 19]. On the other hand, Cho et al.  sequenced the DNA barcode region of the COI gene for some 70 heliothine species, including about 10 pests, however they sequenced only a single individual for most species as their aim was a phylogenetic analysis of species, not species delimitation and diagnostics. Furthermore, none of the above studies produced data that complies with the “BARCODE” data standard, which requires deposition of voucher specimens in a collection, archiving of raw sequence trace files, and sequence length and quality standards. This study aimed to fill that gap, building a comprehensive DNA barcode data set to aid identification of Australian heliothines and testing the utility of DNA barcoding for quarantine identifications.
Materials and Methods
We first present DNA barcode data obtained from decades-old museum specimens of Australian heliothines, then add published data from GenBank to examine the utility of DNA barcoding for identification of both Australian heliothines and exotic species, particularly those of quarantine significance.
Collecting fresh specimens for a DNA study would have been an expensive and time consuming proposition since many of these species are distributed across the remote and relatively inaccessible arid zones of northern and central Australia. To circumvent both this difficulty and the challenging task of identifying freshly collected specimens, it was decided instead to build a core data set for Australian Heliothinae using only material examined by Matthews . These specimens are housed in the Australian National Insect Collection (ANIC) and most were collected in the 1990s. Four additional specimens of Australothis tertia collected in 2000–2003 and not examined by Matthews  were also sampled. In total there were 139 specimens, with mean and median ages at the time of DNA extraction of 18.3 and 16 years, respectively (see Table 1). Given the age of these specimens this required the development of a PCR primer set and amplification strategy that would allow the routine DNA barcoding of decades-old insect specimens .
DNA extractions, polymerase chain reaction (PCR) amplifications and DNA sequencing was performed using the PCR primer set and amplification strategy described previously . In short, the PCR strategy targets decades-old museum specimens, amplifying two short overlapping PCR fragments of approximately 300 bp each, that are subsequently reamplified using an internal primer on one end and the M13 primer on the other end. Together the two short fragments yield 559 bp of contiguous COI sequence within the DNA barcode region, fulfilling the requirements of the BARCODE standard .
Sequence trace files were assembled and consensus sequences constructed, aligned and trimmed using Geneious 7.1.9 . Consensus sequences, specimen collection data, specimen images and sequence trace files were uploaded to the Barcode of Life Data System (BOLD, and are available for download as public project Heliothinae of Australia (HELAU). Sequences were also submitted to GenBank as accession numbers KP688427—KP688435 and KX422482—KX422604. BOLD was also used for some analyses, including calculation of intra- versus interspecies distances.
Two data sets were constructed for phylogenetic analysis. Data set 1 comprised the 132 sequences derived for this study as described above. Data set 2 was composed of 1,553 sequences and was made by adding sequences retrieved from GenBank on 5 April 2016. The new data comprised 161 sequences derived from specimens collected in Australia and 1,260 sequences from non-Australian material. The latter comprised mostly exotic species but also species that occur in Australia, such as Helicoverpa armigera. The 161 Australian sequences comprised seven sequences from unpublished BOLD projects of the authors, 137 sequences produced from material also in the Australian National Insect Collection  and 17 earlier published sequences . Sequences from the 3’-half of COI (not the DNA barcode region) were discarded, and the alignment was trimmed to 559 nt. Further sequences were eliminated after preliminary analyses suggested they were misidentified, e.g. GenBank accession JX509812.1  ostensibly Heliocheilus fervens, clearly belongs in the Pyralidae (BIN BOLD:AAL8596) and was tagged on BOLD by author AM. The data set is provided as supplementary material in FastA format.
FABOX v. 1.4.1  was used to edit sequence names. MEGA v.6.06  was used to calculate Kimura 2-Parameter genetic distances (that distance model was chosen only to facilitate comparison with distances calculated by BOLD) and to test models of sequence evolution for phylogenetic analysis, the preferred model being the one with the lowest Bayesian Information Criterion (BIC) score. For data set 1 this proved to be the General Time Reversible model with Gamma-distributed rates (GTR+G), while for data set 2 this was the General Time Reversible model with Gamma-distributed rates and Invariable sites (GTR+G+I). Phylogenetic analyses utilized Maximum Likelihood (ML) methods and were performed in Geneious v.7.1.9  using the plugins available for Fasttree 2 , PhyML 3.0  and RAxML 7.2.8 . Partitionfinder v.1.1.1  was used to select a partitioning scheme for RAxML analysis.
Fasttree was used for preliminary analyses because of its speed, while PhyML and RAxML were used for final analyses. Fasttree analyses utilized the “pseudocounts” option recommended when alignments contain non-overlapping sequences. PhyML analyses optimized topology, branch lengths and rates, used the “BEST” topology search option and calculated “SH-like” support values. RAxML analyses used the ML search convergence criterion, implemented two data partitions: nucleotide positions 1 and 2 combined versus position 3, and performed 500 fast bootstrap replicates.
Data set 1 (132 sequences)
We obtained DNA barcode data from 37 species, including 36 of Australia’s 37 species of Heliothinae, plus the New Zealand endemic species Australothis volatilis Matthews & Patrick (1998), with a mean of 3.6 sequences per species (S.D. = 1.22). The only Australian species we could not obtain DNA barcode data for was Heliothis hoarei Matthews (1999), known from only four specimens, with the two ANIC specimens collected in 1938 and 1956.
Of the 139 specimens sampled from Matthews’  material examined we obtained COI sequence data for 132 specimens or 95% of samples (Table 1), which had a mean age at DNA extraction of 17.4 years. BARCODE standard compliant sequences (>486 nt in length, less than 3 N’s) were recovered from 107 specimens (77%), with a mean age of 16.4 years, and minimum and maximum ages of 8 and 38 years, respectively. Partial barcode sequences, with a mean length of 299 nt, were recovered from a further 25 specimens (18%), with a mean age at DNA extraction of 21.8 years, and minimum and maximum ages of 15 and 47 years, respectively. No sequence could be obtained from the remaining seven specimens sampled (5%) which had a mean age of 33 years, and minimum and maximum ages of 16 and 55 years, respectively.
The ML tree derived using PhyML is shown in Fig 1 with the species that were recovered as unique clusters collapsed to single terminal nodes (triangles). RAxML bootstrap values and ML-based SH-like support values are displayed on each branch if ≥ 0.5. All species of Adisura (n = 2), Heliothis (n = 2) and Helicoverpa (n = 5) were recovered as unique clusters in both ML analyses, with strong support. Australothis species (n = 4) were each recovered as unique clusters except for A. rubrescens. Adisura marginalis and Australothis exopisso were each divided into two distinct barcode clusters, with low levels of variation within them, but with distances of 5.1% and 4.8% between the two clusters, respectively. Other intraspecific distances were less than 1% and interspecific (nearest-neighbour) distances generally were greater than 2–3%.
Well-supported terminal clusters (species or species groups) collapsed. Numbers in parentheses following names are number of sequences within that group. Asterisk indicates species not recovered as monophyletic. Numbers on branches are RAxML bootstrap values followed by SH-like support values from PhyML expressed as a percentage, both shown only if ≥ 50.
For Heliocheilus a very different picture emerged, with only 11 of 24 species recovered as unique clusters in the PhyML analysis. Of the remaining 13 species, five were recovered as paraphyletic in one or both phylogenetic analyses (H. cladotus, H. neurota, H. pallida, H. aleurota, H. eodora) and a terminal group of eight species were recovered as a grade. Seven of the eight species in this terminal grade shared one or more haplotypes with another species. The maximum interspecific distance among the eight species peaked at 1.4%, which was roughly equal to the maximum intraspecific distance. Only one third (8 of 24) of Heliocheilus species displayed an obvious barcode gap (low intraspecific versus large interspecific distances).
Data set 2 (1,553 sequences)
Data set 2 contained sequences for a number of genera not found in Australia, including the Pyrrhia group (Pyrrhia, Heliothodes, Eutricopis) (6 species), Schinia-group (Schinia, Heliolonche, Psectrotarsia and one sequence lacking identification beyond subfamily) (56 species), Protoschinia (1 species), Masalia (4 species) and Chloridea (2 species), the latter genus formerly known as the Heliothis virescens species group. Data set 2 also contained an additional 11 species and 54 sequences of Heliothis, five additional species and 148 sequences of Heliocheilus and an additional six species and 658 sequences of Helicoverpa, mostly from H. armigera and H. zea. Taxon sampling density for Australian species, excluding Helicoverpa armigera, was increased to a mean of 7.7 sequences per species (S.D. = 2.96).
The full ML tree derived using PhyML for this expanded dataset including all public barcode region data for Heliothinae is shown in two parts as S1 Fig and S2 Fig. The ML tree was rooted with the entire Pyrrhia group , comprising six species in three genera. All Pyrrhia-group species and 39 of 42 Schinia-group species with multiple sequences were recovered as unique clusters. The ML tree is redrawn in Figs 2 and 3 with many single-taxon clusters collapsed to aid in visualizing species recovery for the remaining taxa.
Some clusters collapsed. Numbers in parentheses following names are number of sequences within that group. Numbers on branches are RAxML bootstrap values followed by SH-like support values from PhyML expressed as a percentage, both shown only if ≥ 50. Asterisk indicates species not recovered as monophyletic.
Species clusters collapsed. Numbers in parentheses following names are number of sequences within that group. Numbers on branches are RAxML bootstrap values followed by SH-like support values from PhyML expressed as a percentage, both shown only if ≥ 50. Asterisk indicates species incongruence due to misidentifications (see discussion).
Fig 2 shows relationships among species of Heliothis, Chloridea, Masalia and Heliocheilus. All Heliothis and Chloridea species are recovered as unique clusters, however H. acesias and C. virescens consisted of two clusters separated by > 5% and >2% sequence divergence, respectively. Masalia species were represented by only one sequence each, but were separated by >2% from their nearest neighbours. For Heliocheilus, results were comparable with data set 1, except that five exotic species were sampled, only seven of the 24 species were placed in unique clusters, and the terminal grade that contained eight species in data set 1 comprised 14 species.
Fig 3 shows relationships among Australothis and Helicoverpa. Both genera were recovered as unique clusters, with strong support. A. rubrescens samples were divided among four distinct clusters, each separated by more than 3.5% sequence divergence. All species of Australothis were well separated from each other but there was little support for relationships among species. Eleven species of Helicoverpa were represented in Fig 3 and all were recovered as unique clusters and well separated from other species, except for H. assulta and H. fletcheri, which shared haplotypes. In addition, four sequences within the H. armigera cluster, which contained 419 sequences, were identified as either H. assulta or H. punctigera.
DNA barcode sequence recovery
Specimen age affected our ability to recover DNA barcode sequences. Using the PCR primers and amplification strategy developed for older specimens  we were able to get BARCODE standard compliant sequences (i.e. >486 nt of contiguous sequence with two or fewer ambiguous sites) from 107 of 139 (77%) samples with a mean age at DNA extraction of 18.1 years, and partial barcode data from 132 of 139 (95%) of samples.
Only two thirds of the Australian species sampled were recovered in unique barcode clusters by the ML analyses. As a result, during the early stages of this project we could not rule out the possibility of cross-contamination of samples during DNA extraction or PCR. This possibility was of concern because the PCR procedure we employed relies on reamplification of initial PCR products using hemi-nested primers . However, a number of factors subsequently convinced us of the veracity of our data. The first was our use of copious negative controls in our experiments, with empty wells in tissue sample plates being subjected to DNA extraction and two rounds of PCR without detection of PCR products. A second factor was the lack of sequence variation detected for each sample in the region of overlap between the two half-length barcode fragments. Despite being only 58bp in length this region contains 3–4 highly variable sites that often differ even between closely related species. The third factor was the independent publication of DNA barcode data  for much of the Australian Lepidoptera fauna, including Heliothinae. Inclusion of this data  in our expanded data set 2 resulted in almost complete congruence, with their sequences either clustering with, or being >99% similar to, conspecific sequences in our data set 1.
Do DNA barcodes track species boundaries?
The ML tree derived from data set 2, the expanded set of all public data for Heliothinae (Figs 2 and 3, S1 and S2 Figs) gave very similar results for Australian taxa to that presented in Fig 1. Many of the additional sequences in data set 2 were from non-Australian taxa such as Pyrrhia, Schinia and related genera. With minor exceptions all of these species were recovered as unique clusters and they were not considered further in this study. We note however that DNA barcoding is known to fail in some species of Schinia which were not included in our data set: DNA barcode data from 35 specimens failed to track species boundaries in the six species of the S. volupia species complex . Unfortunately these sequences were not captured by our search term “Heliothinae” on GenBank because the sequences were suppressed (e.g. see GenBank accession GU702778) when their identification was not updated beyond “Lepidoptera”, thus they were not included in this study.
Increased sampling of heliothine species and greatly increased geographic sampling of some of the non-endemic Australian species, resulted in a much higher rate of species recovery in single clusters (“monophyly”) overall, despite further reducing the already poor rate of recovery of Australian Heliocheilus species. Below we discuss the most significant results by genus.
Based on our ML analyses, only one third (eight of 24) of Australian Heliocheilus species were recovered as unique clusters and had discernible barcode gaps between species: H. aberrans, H. albivenata, H. ferruginosa, H. halimolimnus, H. ionola, H. melibaphes, H. ranalaetensis, and H. rhodopolia. The last species was included in the terminal grade of 14 species only because the relationships among it and other species were too complex to illustrate in Fig 2 (but see S1 Fig). Four species that were recovered as unique clusters in data set 1 were placed in multiple clusters in data set 2 due to the inclusion of new haplotypes in the expanded dataset, i.e., H. cladotus, H. abaccheutus, H. vulpinotatus and H. canusina. For H. cladotus, H. abaccheutus, H. vulpinotatus, H. eodora and H. pallida, paraphyly seems to result only from the very low levels of sequence variation and a resulting lack of differentiation among species, and it is probable that sequencing a larger portion of the mitochondrial genome would provide enough informative characters to recover those species as unique clusters and distinct from closely related species. However, H. aleurota and H. canusina each had a single distant sequence (>1.8% divergent from the others for that species) added from published data , resulting in those species being represented in the terminal grade of 14 species in data set 2. For the latter two species we cannot exclude the possibility of cross-contamination or misidentification of samples. For H. neurota there are two distinct clusters of sequences which are up to 1.8% divergent from each other, and neither is included in the terminal grade of 14 species.
The BOLD BIN database  includes 19 species in a single BIN (BOLD:ACE4297) which corresponds closely to our terminal grade of 14 species (noting that species may occur in multiple BINs). Three species that we regard as diagnosable through ML-based analysis of DNA barcode data are included in this BOLD BIN, i.e., H. cladotus, H. ferruginosa and H. neurota, but this difference just reflects our differing methodologies. Other species names included in this BOLD BIN in error are: H. epigrapha (a junior synonym of H. ferruginosa), H. venata and H. neurias (both junior synonyms of H. cramboides), H. clathrata (a junior synonym of H. neurota) (Matthews 1999), H. confundens (see discussion below on H. confundens) and Rivula niphodesma. The latter species bears a superficial resemblance to H. cramboides but belongs in a different family, Erebidae, and is clearly a misidentified specimen.
The remaining eight species in the terminal grade are the same eight species recovered in the terminal grade of data set 1 (Fig 1). These eight species are completely intermingled in the trees, and seven of eight species share at least one haplotype with other species. Given that there is little morphological variation among Australian Heliocheilus species, this raises the question whether all the named species indeed warrant species status. However, it turns out that some of the most similar looking species pairs have quite distinctive DNA barcodes (e.g. H. aberrans versus H. albivenata) while some of the species sharing haplotypes with other species have very distinctive wing patterns (e.g. H. cistella and H. flavitincta). In fact, H. flavitincta is the most distinctive of all Heliocheilus with a yellow-orange ground colour to the forewings, overlaid by dark brown to black lines, whereas most of its congeners are a dull pale brown with indistinctive markings. It is possible that introgression is involved here, or a Wolbachia-mediated mtDNA selective sweep.
Many Heliocheilus species are difficult to differentiate from other species. H. ferruginosa males cannot be separated from H. thelycritus males as the only diagnostic characters are in the female genitalia . Therefore some of the male specimens sampled were initially labelled “H. ferruginosa-thelycritus” to indicate that they could be either species. Fortunately, positively identified females of each species were well separated in the trees, with H. thelycritus being placed in the terminal group of eight species indistinguishable by barcodes but H. ferruginosa females forming a distinct group of their own. Males of the two species associated with one or the other cluster and were secondarily labelled with the appropriate species name.
Heliocheilus confundens, although treated in the revision of Australian Heliothinae, is known only from Indonesia  and was not sampled by us in this study (data set 1). Five specimens supposedly of this species were sampled in a previous study  and are included in BIN BOLD:ACE4297, however the specimens were collected in north-western Australia and were examined by Matthews as part of his revision, therefore they cannot be H. confundens. We treated these specimens as Heliocheilus sp. although we have retained the name “confundens” in parentheses in S1 Fig, however the specimens appear to us to be H. cramboides, as the photographs and collection data for some of these specimens matches those of H. cramboides specimens examined by Matthews .
Suitable nuclear gene data could shed light on the reasons for DNA barcode failure in Australian Heliocheilus, however, because all the material sequenced for this study is decades old it was not possible to sequence nuclear DNA from these specimens using conventional PCR-based Sanger sequencing approaches. We note that a previous study  also found low levels of variation among the Australian Heliocheilus species in both nuclear genes examined, and this likely represents a recent, rapid radiation.
Heliothis, Australothis and Adisura.
DNA barcoding proved successful in distinguishing all species of these genera for which barcode data was obtained, including four species of Adisura, 12 species of Heliothis, given that H. adaucta is a synonym of H. maritima, and four species of Australothis. The two distinct sequence groups (BOLD BINs) recovered for Adisura marginalis and the four groups of Australothis rubrescens warrant further investigation as they could represent cryptic species, Wolbachia infected lineages, genetically diverged populations and/or ancestral mitochondrial lineages retained within populations.
Two very distinct barcode clusters were recovered for Chloridea virescens which is considered the second worst pest in the western hemisphere after Helicoverpa zea , and is a major pest of cotton, tobacco and soybeans. Interestingly, the 114 sequences from Brazilian specimens  formed a distinct group, indeed a different BOLD BIN, 2.1–3.6% distant from the remaining C. virescens, which were collected from North, Central and South America. This raises the possibility that the Brazilian populations sampled previously  belong to a different species and have been misidentified. Apart from C. virescens and C. subflexa, the only Chloridea species currently on BOLD is C. molochitina, however searching BOLD with the Brazilian sequences does not result in a BOLD match to that species. The only other species of Chloridea known to be a minor pest is C. tergemina . The possibility that cryptic sibling species of C. virescens exist has been raised before  and this hould be investigated further for the specimens from Brazil .
DNA barcodes readily distinguish 10 of the 11 species of Helicoverpa for which barcode data exists in the public domain. The exception is sequences identified as being from H. fletcheri (GenBank accessions KF492623—KF492626) which are identical to sequences of H. assulta. A tropical African species, H. fletcheri was placed it in the H. zea-group and regarded as most similar to H. toddi , but H. toddi sequences were not available for comparison. These sequences were deposited in GenBank in 2013 but are not yet associated with a publication that confirms the species identifications, therefore the identity of these sequences should be treated with caution.
There are four anomalous sequences nested within the H. armigera cluster. One sequence was identified as H. punctigera in a study analysing the diets of invertebrate predators using COI sequences (GenBank accession JQ240198.1, ). The other three sequences (GenBank accessions JX509775 –JX509777) were identified as H. assulta in a study which concluded that H. assulta and H. armigera had identical DNA barcodes. However, our data demonstrates a minimum of 2.4% divergence between the latter two species, and at least 4% distance between H. punctigera and any other species. We conclude that these sequences are almost certainly misidentified to species, however neither of the cited papers provides any information about voucher specimens or the basis on which the identifications were made. Therefore there is no way to check the identifications and the sequences should be disregarded.
The two most economically important species of Helicoverpa are H. zea and H. armigera, and our data set contains 202 sequences of the former species and 419 of the latter, with most of the data from published data sets [9, 12, 13, 19, 36]. The specimens were collected from 22 countries, covering most of the known distributions of both species, including North America (for H. zea), South America (for both species), Australia, Asia and Europe (for H. armigera). The two species were each recovered as unique clusters and as sister-groups, separated by a minimum genetic distance of 1.9%. Thus DNA barcoding holds up to global sampling and can be used to distinguish these species reliably.
Identification of heliothine pest species
The incursion of H. armigera in Brazil went undetected for about five years  giving the species time to establish on corn, soybean and cotton and spread throughout the country, reducing crop yields by 35% and resulting in economic losses of about $1 billion . Early detection of H. armigera might have prevented this biological invasion. However, distinguishing H. armigera and H. zea adults is a difficult and specialized task and larvae of H. armigera cannot be distinguished from those of H. zea using morphology, despite efforts to find morphological characters. For H. armigera and H. zea, the head chaetotaxy, mandibles, hypopharyngeal complex, body coloration and markings, body chaetotaxy, pinacula size and shape, setal color, cuticle texture, and crochet counts and arrangement for various instars do not bear any morphological characters that reliably separate larvae of these two species . Instead a nuclear ribosomal DNA-based Real Time PCR assay based on ITS2 sequences was proposed to identify immature stages of these species , and a similar PCR assay based on ITS1 has also been proposed .
Although rapid molecular diagnostic tests now exist for distinguishing H. armigera from H. zea, there remains an urgent need for molecular diagnostics methods which can distinguish other Helicoverpa species, Heliothis species and other heliothine pest species. DNA barcodes provide an ideal platform for such identifications because there is no limit to the number of species that can be detected with a single assay. This study demonstrates that DNA barcodes also can be used to reliably distinguish the economically important species of Helicoverpa (with the possible exception of the minor pest H. fletcheri, which unpublished data on GenBank suggests may have identical DNA barcodes to H. assaulta), Heliothis, Chloridea, and likely most other species of Heliothinae. Australian Heliocheilus species are a notable exception with less than half of the species being diagnosable using DNA barcodes. Those species that cannot be diagnosed using DNA barcodes form a single cluster. In addition, none of the Australian Heliocheilus are pests, thus quarantine agencies using barcode data would easily be able to tell native from exotic species and pests, such as the African species H. albipunctella (the Millet Head Miner), from non-pests.
Standards for quarantine identifications
DNA databases used for quarantine identifications require high levels of data integrity and data redundancy . While BOLD is an enormously useful tool for species identification, it is not without errors resulting from incorrectly identified specimens and/or cross-contamination of samples. While there are advantages to having all the non-BARCODE compliant COI gene sequences from GenBank stored on BOLD, it can also be a source of error. It is also not unprecedented, in our experience, to encounter publically released data on BOLD that has been incorrectly identified to species. This ultimately undermines the usefulness of BOLD for applications such as quarantine identifications. We note that BOLD has a facility for community third-party annotation of barcode records, and the Barcode Index Number (BIN) system  provides an efficient mechanism for detection of taxonomic misassignments. However, we argue that these facilities alone are insufficient to ensure a high standard of species identifications. Instead we advocate for a second, higher barcode standard that should be applied to regulated species such as those of quarantine importance, commercial fish species, IUCN red-listed species, etc. The second standard would have more rigorous criteria for species identification, shifting the onus to data submitters to demonstrate unequivocally that their voucher specimens have been accurately identified, for example by providing photographic evidence of genitalia dissections or other necessary diagnostic characters, and/or having identifications vetted by independent taxonomic experts. Other criteria that might also been considered in the higher standard include whether all closely related species that need to be distinguished have been sampled.
DNA barcodes were assembled for the entire heliothine moth fauna of Australia, bar one rare species. The data revealed deep mtDNA divergences in two Australian species, which may represent cryptic species, but very shallow divergences among about half of the Australian fauna of Heliocheilus, which consequently cannot be identified using this method. Of the 91 species remaining in the expanded global data set after excluding all Heliocheilus, 87 species (96%) were readily identifiable with DNA barcodes. Thus DNA barcoding can provide a powerful solution to quarantine identifications of Helicoverpa, Heliothis and other Heliothinae. While real time PCR methods for identifying Helicoverpa species are faster, current methods are useful only for distinguishing between H. armigera and H. zea, and may give misleading results if other species are processed unwittingly. Such assays are therefore best suited to high-throughput screening once identifications have been narrowed down to a few choices through other means. A much more powerful approach is to derive DNA sequence data which can be used to query an extensive database of reference sequences for many species. However, more emphasis is needed on distinguishing true “reference” sequences from others. Reference sequences should not only be of highest quality and derived from properly vouchered specimens, their species identifications should be backed by scientific data such as images of diagnostic morphological characters, and they should be performed or vetted by taxonomic experts.
S1 Fig. ML tree from PhyML for data set 2 (1,553 taxa).
Subtree containing Australothis and Helicoverpa collapsed (see S2 Fig).
S2 Fig. Subtree for Australothis and Helicoverpa from ML tree from PhyML for data set 2 (1,553 taxa).
The Australian National Insect Collection provided access to their collections and sampling of legs for DNA extraction, Marianne Horak, Ted Edwards and You Ning Su provided generous assistance with the process.
- Conceived and designed the experiments: AM.
- Performed the experiments: AM DG.
- Analyzed the data: AM.
- Contributed reagents/materials/analysis tools: AM.
- Wrote the paper: AM DG.
- 1. Matthews M. Heliothine moths of Australia: a reference guide to pest bollworms and related noctuid groups. CSIRO Publishing, Melbourne; 1999.
- 2. Matthews M. Classification of the Heliothinae. Natural Resources Institute Bulletin No. 44. Natural Resource Institute, Kent; 1991.
- 3. Mitter C, Poole RW, Matthews M. Biosystematics of the Heliothinae (Lepidoptera: Noctuidae). Annu Rev Entomol 1993; 38: 207–225.
- 4. Cho S, Mitchell A, Regier JC, Mitter C, Poole RW, Friedlander TP, et al. A highly conserved nuclear gene for low-level phylogenetics: Elongation Factor-1α recovers morphology-based tree for heliothine moths. Mol Biol Evol 1995; 12: 650–656. pmid:7659020
- 5. Cho S, Mitchell A, Mitter C, Regier J, Matthews M, Robertson R. Molecular phylogenetics of heliothine moths (Lepidoptera: Noctuidae: Heliothinae), with comments on the evolution of host range and pest status. Syst Entomol 2008; 33: 581–594.
- 6. Pogue MG. Revised status of Chloridea Duncan and Westwood, 1841, for the Heliothis virescens species group (Lepidoptera: Noctuidae: Heliothinae) based on morphology and three genes. Syst Entomol 2013; 38: 523–542.
- 7. Cleary AJ, Cribb BW, Murray DAH. Helicoverpa armigera (Hübner): can wheat stubble protect cotton from attack. Australian Journal of Entomology 2006; 45: 10–15.
- 8. Gilligan TM, Passoa SC. LepIntercept, An identification resource for intercepted Lepidoptera larvae. Identification Technology Program (ITP), USDA/APHIS/PPQ/S&T, Fort Collins, CO. 2014. Available: http://www.lepintercept.org.
- 9. Mastrangelo T, Paulo DF, Bergamo LW, Morais EGF, Silva M, Bezerra-Silva G et al. Detection and genetic diversity of a heliothine invader (Lepidoptera: Noctuidae) from North and Northeast of Brazil. J Econ Entomol 2014; 107: 970–980. pmid:25026655
- 10. Hardwick DF. The corn earworm complex. Memoirs of the Entomological Society of Canada 1965; 40: 1–247.
- 11. Poole RW, Mitter C, Huettel M.A revision and cladistic analysis of the Heliothis virescens species-group (Lepidoptera: Noctuidae) with a preliminary morphometric analysis of Heliothis virescens. Mississippi Agricultural and Forestry Experiment Station Technical Bulletin 185, Mississippi Entomological Museum; 1993.
- 12. Behere GT, Tay WT, Russell DA, Heckel DG, Appleton BR, Kranthi KR, et al. Mitochondrial DNA analysis of field populations of Helicoverpa armigera (Lepidoptera: Noctuidae) and of its relationship to H. zea. BMC Evol Biol 2007; 7: 117. pmid:17629927
- 13. Behere GT, Tay WT, Russell DA, Batterham P. Molecular markers to discriminate among four pest species of Helicoverpa (Lepidoptera: Noctuidae). Bull Entomol Res 2008; 98: 599–603. pmid:18631420
- 14. Gilligan TM, Tembrock LR, Farris RE, Barr NB, van der Straten MJ, van de Vossenberg BTLH, et al. A Multiplex RealTime PCR Assay to diagnose and separate Helicoverpa armigera and H. zea (Lepidoptera: Noctuidae) in the New World. PLoS One 2015; 10, e0142912. pmid:26558366
- 15. Li QQ, Li DY, Ye H, Liu XF, Shi W, Cao N, et al. Using COI gene sequence to barcode two morphologically alike species: the cotton bollworm and the oriental tobacco budworm (Lepidoptera: Noctuidae). Mol Biol Rep 2011; 38: 5107–5113. pmid:21181271
- 16. Orui Y, Matsuzawa H, Koike Y, Yoshimatsu S. Discrimination of Helicoverpa armigera (Hübner) and H. assulta (Guenée) (Lepidoptera: Noctuidae) by PCR-RFLP analysis, and application to surveying occurrence of H. armigera in tobacco fields of Japan. Japanese Journal of Applied Entomology and Zoology 2000; 44: 73–79.
- 17. Perera OP, Allen KC, Jain D, Purcell M, Little NS, Luttrell RG.Rapid Identification of Helicoverpa armigera and Helicoverpa zea (Lepidoptera: Noctuidae) using ribosomal RNA Internal Transcribed Spacer 1. J Insect Sci 2015; 15: 155. pmid:26516166
- 18. Ranjith MT, Chellappan M.DNA barcoding compliments morphological identification in tomato fruit borer, Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae). Entomon 2015; 40: 125–138.
- 19. Tay WT, Soria MF, Walsh T, Thomazoni D, Silvie P, Behere GT, et al. A Brave New World for an Old World Pest: Helicoverpa armigera (Lepidoptera: Noctuidae) in Brazil. PLoS One 2013; 8: e80134. pmid:24260345
- 20. Mitchell A. Collecting in collections: a PCR strategy and primer set for DNA barcoding of decades-old dried museum specimens. Mol Ecol Resour 2015; 15: 1102–1111. pmid:25644663
- 21. Hanner R. Proposed standards for BARCODE records in INSDC (BRIs). Technical report, Database Working Group, Consortium for the Barcode of Life. 2009. Available: http://barcoding.si.edu/PDF/DWGdatastandards-Final.pdf.
- 22. Kearse M, Moir R, Wilson A, Stones-Havas S, Cheung M, Sturrock S, et al. Geneious Basic: an integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics 2012; 28: 1647–1649. pmid:22543367
- 23. Ratnasingham S, Hebert PDN.BOLD: The Barcode of Life Data System (www.barcodinglife.org). Mol Ecol Notes 2007; 7, 355–364. pmid:18784790
- 24. Hebert PDN, deWaard JR, Zakharov EV, Prosser SWJ, Sones JE, McKeown JTA, et al. A DNA ‘Barcode Blitz’: Rapid Digitization and Sequencing of a Natural History Collection. PLoS One 2013; 8: e68535. pmid:23874660
- 25. Li J, Han H-L, Gao Q, Jin Q, Chi M-Y, Wu C-S, et al. Species identification of Noctuidae (Insecta: Lepidopera) with DNA barcoding of support vector machine and neighbor-joining method. Journal of Biosafety 2012; 04.
- 26. Villesen P. FaBox: an online toolbox for FASTA sequences. Mol Ecol Notes 2007; 7: 965–968.
- 27. Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. MEGA6: Molecular Evolutionary Genetics Analysis Version 6.0. Mol Biol Evol 2013; 30: 2725–2729. pmid:24132122
- 28. Price MN, Dehal PS, Arkin AP. FastTree 2–Approximately Maximum-Likelihood Trees for Large Alignments. PLoS One 2010; 5: e9490. pmid:20224823
- 29. Guindon S, Dufayard JF, Lefort V, Anisimova M, Hordijk W, Gascuel O. New algorithms and methods to estimate Maximum-Likelihood phylogenies: assessing the performance of PhyML 3.0. Syst Biol 2010; 59: 307–21. pmid:20525638
- 30. Stamatakis A.RAxML-VI-HPC: Maximum Likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 2006; 22: 2688–2690. pmid:16928733
- 31. Lanfear R, Calcott B, Ho SYW, Guindon S. PartitionFinder: combined selection of partitioning schemes and substitution models for phylogenetic analyses. Mol Biol Evol 2012; 29: 1695–1701. pmid:22319168
- 32. Pogue MG, Ouellette GD, Harp CE.2013 A revision of the Schinia volupia (Fitch) species complex (Lepidoptera: Noctuidae: Heliothinae). Zootaxa 2013; 3716: 157–191.
- 33. Ratnasingham S, Hebert PDN. A DNA-Based Registry for All Animal Species: The Barcode Index Number (BIN) System. PLoS One 2013; 8: e66213. pmid:23861743
- 34. Albernaz KC, Silva-Brandão KL, Fresia P, Cônsoli FL, Omoto C. Genetic variability and demographic history of Heliothis virescens (Lepidoptera: Noctuidae) populations from Brazil inferred by mtDNA sequences. Bull Entomol Res 2012; 102: 333–343. pmid:22126989
- 35. Juen A, Hogendoorn K, Ma G, Schmidt O, Keller MA. Analysing the diets of invertebrate predators using terminal restriction fragments. J Pest Sci 2012; 85: 89–100.
- 36. Leite NA, Alves-Pereira A, Correa AS, Zucchi MI, Omoto C. Demographics and Genetic Variability of the New World Bollworm (Helicoverpa zea) and the Old World Bollworm (Helicoverpa armigera) in Brazil. PLoS One 2014; 9: e113286. pmid:25409452
- 37. Sosa-Gómez DR, Specht A, Paula-Moraes SV, Lopes-Lima A, Yano SAC, Micheli A, et al. Timeline and geographical distribution of Helicoverpa armigera (Hübner) (Lepidoptera, Noctuidae: Heliothinae) in Brazil. Revista Brasileira de Entomologia 2016; 60: 101–104.
- 38. Frewin A, Scott-Dupree C, Hanner R. DNA barcoding for plant protection: applications and summary of available data for arthropod pests. CAB Reviews 2013; 8: 1–13.