DNA barcoding confirms polyphagy in a generalist moth, Homona mermerodes (Lepidoptera: Tortricidae)

Introduction

Short sequences of cytochrome c oxidase I (COI), or animal DNA barcodes, can illuminate species boundaries when combined with other sources of information including taxonomic knowledge, morphology, and ecology (Rubinoff & Powell 2004; Schindel & Miller 2005; Dasmahapatra & Mallet 2006; DeSalle 2006; Rubinoff 2006; Miller 2007). Recent studies of mitochondrial DNA (mtDNA) variation in polyphagous insect species have revealed host races and cryptic species with ecological and/or morphological correlates of the DNA markers (Waring et al. 1990; Berkov 2002; Blair et al. 2005). Other molecular studies have favoured broader species concepts than those based on traditional morphological characters (Uechi et al. 2003; Vahtera & Muona 2006).

Hebert et al. (2004) used genetic distances to establish species limits in a morphologically and ecologically diverse hesperiid butterfly based on unique haplotypes corresponding to morphologically and ecologically distinct caterpillars. The correlation between molecular markers and ecology supported the delimitation of several new and relatively uniform species within a single hypervariable species. Re-analysis of these data by Brower (2006) using phylogenetic methods also indicated cryptic species although fewer in number than Hebert et al. (2004).

We took a similar approach to investigate whether the highly polyphagous and widespread Homona mermerodes Meyrick (Lepidoptera, Tortricidae) was one or more species. H. mermerodes was commonly reared in ecological studies of herbivorous insects in Papua New Guinea (Novotny et al. 2004) and was suspected of containing cryptic species because of its polyphagous habit, highly variable wing colouration, and because a closely related species, Homona coffearia, was found to be a species complex (Whittle et al. 1987). DNA barcodes provide an efficient molecular test of species concepts underpinning our previous ecological studies (e.g. Miller et al. 2003). Homona and related genera such as Adoxophyes are notoriously difficult to identify by wing pattern because of the extreme variability, which has driven the use of alternative characters such as pheromones and molecular genetics to diagnose species (Whittle et al. 1987; Lee et al. 2005).

Homona currently includes some 30 species distributed throughout Southeast Asia and Australia (Brown 2005) as the several African species should be placed in other genera (Razowski 2004: 174). Homona includes widespread polyphagous pest species, notably H. coffearia (Nietner) and H. magnanima Diakonoff (Dugdale et al. 2005). Diagnosis of many species in the genus is complicated by their sexual dimorphism and highly variable wing pattern. Six species are known from New Guinea (Table 1). Although many species were reviewed in a series of studies by Diakonoff (1948, 1953, 1968, 1983), the genus is in need of comprehensive revision to clarify its taxonomic status and the species involved.

Materials and methods

During our studies of caterpillar host specificity in Papua New Guinea rainforests (Madang, East Sepik, and West Sepik provinces), caterpillars of Homona mermerodes were collected, photographed, and reared to adults. Collecting was conducted throughout all months of the year at multiple sites throughout the northern lowlands of Papua New Guinea (PNG) during 1994–2006 (Novotny et al. 2002a; Miller et al. 2003). All individuals were collected by local assistants and brought to a field rearing station. Sampling effort was standardized across all hosts at all localities (Novotny et al. 2002a). Caterpillars were fed fresh foliage of the plant species on which they were collected and reared to adults whenever possible. Caterpillars and adults were first sorted to morphospecies by experienced parataxonomists and later matched to museum specimens at USNM (Smithsonian Institution) and BMNH (Natural History Museum, London). In addition, adults were light-trapped at four sites in northern Queensland.

DNA extraction and analysis

DNA sequences were obtained from 102 specimens of Adoxophyes sp. nr. marmarygodes Diakonoff, Homona aestivana, H. mermerodes, Homona spargotis, Homona trachyptera, and Homona sp. nr. salaconis from four localities in PNG and 24 specimens of Homona mermerodes and Homona spargotis collected by light trapping in Queensland, Australia. Food plants were recorded for PNG moths but were unavailable for Australian moths. PNG vouchers are deposited at USNM and the PNG National Agricultural Research Institute (NARI). Australian vouchers are deposited at USNM and Australian National Insect Collection (ANIC).

Sequences were produced at the University of Minnesota and at the University of Guelph. In Minnesota, DNA was extracted from legs of adult individuals and amplified using the QIAGEN DNeasy Tissue Kit. Primers LepF1 and LepRI (Hebert et al. 2004) were used to obtain a 661-bp fragment of COI by cycle sequencing in both directions. Data were collected by polyacrylamide gel electrophoresis using an Applied Biosystems ABI 377 automated DNA sequencer. In Guelph, DNA was extracted from single legs and the same 661 bp fragment was amplified with a standard thermocycling regime (Hajibabaei et al. 2006). PCR products were subsequently sequenced on an ABI 3730 capillary sequencer. Sequences are available at the NCBI GenBank database (Accession nos EF070743–EF070863 and EF432740–EF432746) and at the Barcode of Life Database (see online Supplementary material).

Sequences were edited using sequencher version 4.2 and manually aligned. Haplotype diversity was examined prior to distance analysis with paup 4.0b10 (Swofford 2001). When identical haplotypes were obtained from different individuals, only a single individual was retained for phylogenetic analysis. Phenetic methods often used in studies of DNA barcoding are less explicit about distinguishing evolutionarily relevant information than character-based parsimony or likelihood-based methods (Hillis & Huelsenbeck 1992; DeSalle & Siddall 2005; Cognato 2006). Therefore, species limits and relationships between H. mermerodes and similar species were evaluated using maximum parsimony. Parsimony analysis was conducted with paup 4.0b10 and heuristic searches of 1000 random addition sequence replicates with a maximum of 10 000 trees per replicate. To illustrate haplotype diversity, a neighbour-joining tree showing branch lengths was generated (Fig. 3) with a model of nucleotide substitution selected by modeltest 3.7 (Posada & Crandall 1998) according to the Akaike information criterion (AIC). The relative strength of clade support was assessed using nonparametric bootstrap resampling with 1000 replicates and 10 random addition sequence replicates per bootstrap replicate.

Adoxophyes sp. nr. marmarygodes, a morphospecies superficially similar to Homona, was used to root the phylogenetic trees. The morphospecies is near Adoxophyes marmarygodes Diakonoff, but lacks the female signum illustrated by Diaknoff (1952: 161), although Diaknoff's matching of sexes needs to be tested. Species limits within the complex of highland New Guinea Adoxophyes (A. marmarygodes, A. tetraphracta Meyrick (= A. acropeta Diakonoff), and A. aniara Diakonoff) are unclear and in need of revision.

Taxonomy

Homona mermerodes Meyrick (1910: 213) is known from the Solomon Islands, New Guinea, and Australia (Queensland). The Sumatra record by Diakonoff (1947: 342 and 1952: 404), was a misidentification, and was not repeated by Diakonoff (1983: 104). Museum collections and field experience indicate that it is widespread in the lowlands of New Guinea. It was not found in our study at 1700 m a.s.l. (Novotny et al. 2005), and mid-elevation specimens are uncommon in museum collections. The highest record known to us is 1900 m (Tapini, Loloipa River, Bome, PNG, in ANIC), and there are several records from PNG localities between 1500 m and 1200 m in ANIC, USNM, and BMNH.

The full bibliography of H. mermerodes is: Meyrick (1910): 213–214 [original description]; Meyrick (1912): 15 [checklist]; Meyrick (1913): 19 [checklist]; Meyrick (1938): 505 [distribution records]; Durrant (1916): 153 [distribution records]; Froggatt (1940): 13 [Clerodendron host record]; Diakonoff (1941): 38 [distribution record]; Diakonoff (1947): 342–343 [distribution records]; Diakonoff (1948): 509 [mentions possible relationships]; Diakonoff (1952): 404–405 [distribution records]; Diakonoff (1953): 18 [key]; Diakonoff (1983): 104 [checklist]; Simon Thomas (1958): 228; 1962: 89 [serious Citrus pest]; Basset et al. 1996: 180 [misidentification of H. aestivana]; Horak et al. (1996): 129 [checklist, including two new synonyms from Australia]; Novotny et al. 2003: 708 [Piper spp. host records]; Brown (2005): 388 [checklist]. Despite the fact that it is an occasionally important agricultural pest, the wings and genitalia of this species have never been illustrated and no diagnostic characters for identification have been published.

The wing shape and pattern are sexually dimorphic, and wing pattern is extremely variable (Fig. 1). Homona mermerodes is distinguished from all but one other species of Homona in New Guinea by its brown or orange brown hind wings (not orange or yellow). The other species with brown hindwings is H. salaconis (Meyrick) (Diakonoff 1983: 107) or a species close to it, which can be distinguished by the dark bar along the middle of the forewing costal edge in both sexes (longer than ever occurs in H. mermerodes) and the strongly sinuate costa in females. The male genitalia of H. mermerodes are characteristic: nearly the entire ventral half of the valva is sclerotized as a sacculus which culminates in a long, slender digit in the centre of the valva; the uncus is long and thin (Fig. 2). No variation in genitalia beyond species limits was found by extensive dissections of our reared specimens and of material in the ANIC, BMNH, Naturalis (Leiden), and USNM. General morphological variability excluding genitalia was also reviewed in the collection of Papua New Guinea National Agriculture Research Institute. General taxonomic methods are described in Holloway et al. (2001) and Miller et al. (2003).

Homona mermerodes was described (Meyrick 1910) from 26 specimens from Solomon Islands, New Guinea and Australia. Ian Common labelled a lectotype in 1956, but never published the designation. In order to be clear about the application of the name, we designate that female from Gizo, Solomon Islands, collected by A.S. Meek in 1905 in BMNH, with genitalia slide 3370, as lectotype. Three males from New Guinea in BMNH are labelled as paralectotypes, including genitalia slide 3498; we have not attempted to account for the remaining specimens.

Results

During a decade of sampling herbivorous insects in the northern lowlands of Papua New Guinea, we collected ~143 000 individual Lepidoptera (Novotny et al. 2002a, b; Weiblen et al. 2006). Among them, 1411 were identified as Homona mermerodes, of which 739 were reared to the adult stage. H. mermerodes was among the more widespread generalist species being successfully reared from about 40% of the host plant species we studied (44 out of 111 species) representing at least 19 plant families (Table 2). H. mermerodes was the 28th most common species in our sample (705 reared species, median species abundance 9 individuals). Species (and numbers of individuals) sequenced during the study included: Adoxophyes sp. nr. marmarygodes (5), Homona aestivana (7), Homona mermerodes (76), Homona spargotis (4), Homona trachyptera (22), and Homona sp. nr. salaconis (12). Sequences from these 126 moths yielded 34 unique haplotypes (Table 3). The best fitting model of molecular evolution according to the Akaike information criterion (AIC) was general time reversible with estimated base frequencies, a parameter for the proportion of invariant sites and a shape parameter for the distribution of rate heterogeneity among sites (GTR + I + G). A neighbour-joining phylogram fitted with GTR + I + G model parameters is shown in Fig. 3. Homona phylogeny was rooted with two sequences from the Adoxophyes sp. nr. marmarygodes. There were 158 variable sites (23.9%) of which 113 were parsimony-informative (17.9%). The neighbour-joining tree was identical to one of > 10 000 most parsimonious trees. Although the strict consensus of these equally parsimonious trees did not conflict with the neighbour-joining tree, quite a number of branches collapsed. Monophyly of each Homona species was strongly supported (93–100% bootstrap value) whereas resolution and support for intraspecific clades was generally lacking (Fig. 3). There were only 11 parsimony-informative sites (1.7%) among H. mermerodes haplotypes and pairwise distances averaged 0.010 ± 0.001 (X ± SD). Intraspecific distances were much lower on average than interspecific distances (0.107 ± 0.020) and were consistent with general patterns of variability of COI sequence in Lepidoptera (Hebert et al. 2003). Named species boundaries were confirmed by pairwise genetic distances forming a bimodal distribution with nearly discontinuous ranges of interspecific and intraspecific divergence (Fig. 4).

Homona mermerodes haplotype diversity was Poisson-distributed (Fig. 5) and the most common haplotype was encountered in 42 moths (Table 3). This haplotype was collected at multiple localities across northern New Guinea covering a distance of more than 500 km from Madang to West Sepik Province. A less common haplotype, distinguished by a C to T transition at position 277 out of 661, was encountered in seven instances. Four other New Guinea specimens exhibited unique mutations. Although sequences were double-checked for accuracy, it is still possible that some singleton haplotypes could be artefacts of DNA polymerase errors. Two out of four Australian haplotypes were encountered repeatedly. The paraphyly of New Guinea haplotypes with respect to the Australian material (Fig. 3) suggests a phylogeographical scenario in which Australia was colonized by H. mermerodes from New Guinea but clade support for this hypothesis is lacking.

Discussion

In contrast to a previous study identifying host-associated cryptic species in widespread generalist butterfly (Hebert et al. 2004), evaluation of species limits by DNA barcoding confirmed the polyphagous habit in Homona moths. The most common Homona mermerodes haplotype, for example, was reared from 14 different plant species including 11 phylogenetically diverse families (Table 2). The second most-common haplotype was reared from five plant species including as many families. Although inferences about host specificity cannot be drawn from singleton haplotypes, the pattern of generalized feeding exhibited by the more common haplotypes and observations of H. mermerodes caterpillars feeding on 44 out of 111 host plant species sampled in New Guinea strongly support the polyphagous habit of this species (Weiblen et al. 2006). We found no evidence of host-associated mtDNA haplotype differentiation, which might be expected from the matrilineal inheritance of mtDNA and the fact that ovipositing females choose host plants. Intraspecific differentiation mediated by female choice and the tendency to respond to chemical cues similar to the larval host plants may sometimes lead to sympatric speciation (Corbet 1985; van Klinken 2000) but no host-associated differentiation was evident in H. aestivana, which fed on at least four different plant species in as many families, H. trachyptera, which fed on 10 plant species, or Homona sp. nr. salaconis, which fed on seven species.

Variation in wing colouration among the sampled individuals was substantial (Fig. 1), which is typical for H. mermerodes in museum collections. Variable wing patterns from the same mtDNA haplotype suggest a genetic polymorphism or phenotypic plasticity in this trait. On the contrary, genitalia of all dissected H. mermerodes showed little morphological variability, confirming that genital morphology is a legitimate source of data for evaluating species concepts in Lepidoptera (e.g. Miller 1994).

Our findings indicate that broad host range, morphological variability, and regionally distributed populations need not be associated with mtDNA divergence. DNA barcoding may lead to the discovery of cryptic species in some cases but in others the same technique may validate accepted species concepts and reinforce other information such as ecological data on diet breadth and specialization. In the case of Homona mermerodes, DNA barcoding confirms the polyphagous habit of a generalist moth.