Introduction

There are several cases known of leafmining Lepidoptera developing into important agricultural pests, such as Phyllocnistis citrella Stainton, 1856 (Gracillariidae) on citrus, now a worldwide problem (Heppner and Dixon 1995) and Leucoptera coffeella Guérin-Méneville, 1842, Leucoptera meyricki Ghesquière, 1940and related species (Lyonetiidae) on coffee, that are amongst the more important coffee pests (Le Pelley 1973). Leafmining moths apparently often disperse easily, possibly due to their small size, and some have shown rapid invasions over large areas, e.g.: Cameraria ohridella Deschka & Dimić, 1986, Phyllonorycter leucographella (Zeller, 1850), Phyllonorycter issikii (Kumata, 1963)and Macrosaccus robiniella (Clemens, 1859) (Šefrová 1999; Hellrigl 2001; Šefrová 2002a, b; Šefrová and Laštůvka 2002; Davis and De Prins 2011).

Lepidopteran leafminers of grapevine (Vitis vinifera L.) have not yet developed into serious pests in Europe, although one North American species did recently invade European vineyards; Phyllocnistis vitegenella Clemens, 1859 (Lepidoptera: Gracillariidae) became established in Italy and elsewhere in Europe around 1995 (Posenato et al. 1997). The only native European leafminer of grape is Holocacista rivillei (Stainton, 1855) (Lepidoptera: Heliozelidae) (Hering 1957), a minor pest in vineyards in southern Europe and western Asia (see references below). Holocacista rivillei was described from Malta and later reported from Italy. It develops two to three generations annually (Mariani 1942; Marchi 1956; Camporese and Marchesini 1991; Dal Rì and Delaiti 1992; De Tomaso et al. 2008; Baldessari et al. 2009). Infestations leading to damage are infrequent, probably because pest populations are controlled by a complex of eulophid parasitoids (Hymenoptera) (Camporese and Marchesini 1991; Alma 1995). European populations of Phyllocnistis vitegenella occur in northern Italy (Marchesini et al. 2000; Villani 2002; Reggiani and Boselli 2005; Duso et al. 2011), Slovenia (Seljak 2005) and Switzerland (Cara and Jermini 2011). Itcan produce up to four generations annually and has given rise to local outbreaks in northeastern Italy (Posenato et al. 1997; Marchesini et al. 2000). The larvae of both moths produce characteristic mines in grapevine leaves; in Holocacista rivillei a narrow initial gallery leads subsequently to an oval full-depth blotch, from which the larva cuts out an oval pupal case or shield, in which it pupates, leaving an oval hole in the leaf. Phyllonorycter vitegenella makesa long tortuous gallery mine in the upper epidermis, with a distinct dark central frass line that ends in a pupal chamber. Both species can easily be detected in a vineyard based on the presence of their diagnostic leafmines.

In the summer of 2007, leafmines similar to those caused by Holocacista rivillei were observed in a vineyard in northeastern Italy (Borgo Valsugana, Trento province). However, the initial gallery mine was immediately enlarged into a larger blotch, indicating a different species. Adults reared from these mines differed from Holocacista rivillei in size and wing pattern. On their external characters they were identified as belonging to the genus Antispila Hübner, [1825], but not to one of the two species currently known in Europe, i.e., Antispila treitschkiella (Fischer von Röslerstamm, 1843) and Antispila metallella ([Denis & Schiffermüller], 1775) (Karsholt et al. 1995; Ellis 2010; van Nieukerken 2011) which both feed on dogwood (Cornus spp.). Thus, we determined that we had either an undescribed species or an alien species introduced from another continent. This pest has been reported previously as Antispila sp. (Baldessari et al. 2009; Duso et al. in press). Because taxonomic knowledge of the family Heliozelidae is poor, very few species being described up to modern standards, and because many of the known Antispila species are associated with Vitis species or related Vitaceae, it took some time to establish that this species was an undescribed, but common, North American species, hitherto confused with the North American Antispila ampelopsifoliella Chambers, 1874, described from Virginia creeper Parthenocissus quinquefolia. Unfortunately this confusion has already led to the incorrect introduction of the name Antispila ampelopsifoliella into European literature (Laštůvka 2009; van Nieukerken 2011; van Nieukerken et al. 2011a). In this paper we describe the species as Antispila oinophylla van Nieukerken & Wagner, sp. n., provide a diagnosis for its identification, and characterize its geographic distribution and life cycle. We also sequenced a part of the cytochrome C oxidase subunit I (COI) gene (DNA barcode) (Hebert et al. 2003b; Hebert et al. 2003a) as well as those of a selection of other Heliozelidae and the other two European grape leaf-mining micro-moths (Holocacista rivillei and Phyllonorycter vitegenella). DNA barcode data played important roles in revealing the original source of the infestation and unravelling the taxonomy of the new grape pest.

Family Heliozelidae

The family Heliozelidae (superfamily Adeloidea) comprises 123 described species in 12 genera (van Nieukerken et al. 2011b), with the greatest diversity in North America and Australia. Larvae of the Heliozelidae produce leafmines (rarely galls) in various trees and vines, rarely herbs, and typically cut-out an oval case or shield from the leafmine, in which they moult once into a non-feeding final instar or prepupa, and finally pupate in the leaf litter or on plant parts. All are thought to overwinter in temperate regions as prepupae. Eight species of Heliozelidae occur in Europe (van Nieukerken 2011), belonging to four genera: Antispila, Antispilina Hering, 1941, Heliozela Herrich-Schäffer, 1853 and Holocacista Walsingham & Durrant, 1909. Four species of Heliozelidae were previously known from Italy: Heliozela lithargyrellum (Zeller, 1850), Heliozela sericiella (Haworth, 1828), Holocacista rivillei and Antispila treitschkiella (Fischer von Röslerstamm, 1843) (Karsholt et al. 1995), but it is likely that the fauna is incompletely sampled, and that most European species occur in Italy as well. In addition to Antispila oinophylla, we record here Antispila metallella ([Denis & Schiffermüller], 1775) as new from Italy (see Appendix B).

The grapevine family Vitaceae comprises an important group of hosts for the genus Antispila worldwide; out of 32 Antispila species for which host plants are known, 13 feed on Vitaceae, of which at least ten are associated with the genus Vitis (Table 1). Several more unnamed species are also associated with Vitaceae. In North America, Heliozela aesella Chambers, 1877 makes galls in leaves and shoots on Vitis (McGiffen and Neunzig 1985). There are only a few previous records of Heliozelidae as minor pests on grape: in addition to Heliozela rivillei, as mentioned above, Antispila uenoi has been recorded as a pest in Japan (Kuroko 1987; Ueno et al. 1987). Antispila viticordifoliella Clemens, 1860 is listed by McGiffen and Neunzig (1985) as occurring on bunch grape leaves, but not as a pest.

Material and methods Material

Antispila oinophylla adults were collected from Borgo Valsugana for sequencing and larvae were collected and reared for morphological studies. Holocacista rivillei and Phyllocnistis vitegenella adults were also collected from northeastern Italy (Appendix B). To obtain material of the new speciesand related species from North America for comparison, various Antispila mines and larvae were collected by EJvN and CDo during a field trip September-October 2010 in the states of Georgia and Tennessee and by EJvN in September 2011 (partly with DLW) in Connecticut, Massachusetts, Vermont and New York state. Other material included in the taxonomic and DNA analyses was collected by DLW, who has been collecting and rearing Antispila and other leafminers from across North America for three decades (Appendix B). Further material was studied or borrowed from the following collections.

Abbreviations for depositories:

ANSP Academy of Natural Sciences in Philadelphia, Pennsylvania, USA

CNC Canadian National Collection of Insects, Arachnids and Nematodes, Ottawa, Ontario, Canada

DLW Research collection of David L. Wagner, Storrs, Connecticut, USA

MCZ Museum of Comparative Zoology, Harvard University, Cambridge, Massachusetts, USA

RMNH Netherlands Centre for Biodiversity Naturalis, former Leiden Zoology collections, Leiden, Netherlands.

UMDC University of Maryland, College park, USA

UPI University of Padova, Department of Environmental Agronomy and Crop Science, Italy

ZMUO Zoological Museum University of Oulu, Finland.

Rearing

Collected leaves were kept in polystyrene jars or bags, with some moss and or tissue added, until the larvae had prepared the shields. It was often necessary to remove the cut/out shields from the leaves, which were then removed from the breeding jars and dried as vouchers. Breeding jars were kept during winter in an outbuilding, and brought indoors in March, where they were kept until emergence of adults. Specimens collected during fall 2011 were still in hibernation diapause when this manuscript was accepted.

Morphology

Methods for preparation of the genitalia follow Nielsen (1980a) and van Nieukerken (1985), with some minor changes. Nielsen’s unrolling technique does not work well for Heliozelidae, so we usually embedded the total genitalia in dorso-ventral position. For staining male genitalia we used (Mayers) haemaluin or phenosafranin. Wings were stained with phenosafranin and mounted in euparal. Photographs of moths, leafmines, genitalia slides and wing slides were taken with a Zeiss AxioCam digital camera attached, respectively, to a Zeiss Stemi SV11 stereo-microscope, a motorized Zeiss SteREO Discovery.V12 (only Figs 1, 40, 41) or a Zeiss Axioskop H, using Carl Zeiss AxioVision software.

The Distribution Map for North America was prepared with DMap 7.0 (Morton 2000).

Molecular analysis

DNA was extracted destructively from larvae or adult specimens preserved in 96% or 100% ethanol or extracted in a non-destructive fashion from the abdomen of voucher specimens, which were then used to prepare genitalic dissections (protocol in Knölke et al. 2005). From some larvae used for DNA extractions, the cuticle was also cleared and saved. In Padova, total DNA was extracted applying a salting-out protocol (Patwary et al. 1994). In Leiden extractions were carried out with the Qiagen DNeasy Blood and Tissue kit (QIAGEN), using the protocol “purification of total DNA from animal tissues (spin‐column protocol).”

A 665 bp or a 658 bp fragment of the mitochondrial COI gene was amplified using the following primers: in Padua LCO1490 and HCO2198 (Folmer et al. 1994), in Leiden the Lep primers (Hebert et al. 2004), often tailed with T7 promotor and T3 tails in the shorter (amplifying 665 bp) and longer versions (amplifying 658 bp): T‐LepF1-short and T‐LepR1-short or T‐LepF1 and T‐LepR1, or when not tailed LepF1-short and LepR1-short. For some older museum specimens, the DNA was too degraded for amplifying sections over 400 bp long. For these we used internal primers (Hajibabaei et al. 2006). For details of primers see the BOLD site (http://www.barcodinglife.com/).

In Padova, amplification was carried out in 20ml volumes containing 2ml from the nucleic acid extract, 200mM dNTPs, 0.5mM of each primer, 4mM 10x PCR buffer, 2.5 mM MgCl₂ and one unit of Taq polymerase (Promega). The reaction was performed in an INC PTC-100 thermal controller (MJ Research Inc.). Amplification conditions were as follows: the first period of denaturation was 94°C for 5 min, followed by 38 cycles of denaturation at 94°C for 1 min, annealing at 48°C for 1 min, and extension at 72°C for 1 min; the final extension cycle had a step at 72°C for 5 min. A negative control with no template was included for each series of amplifications, to detect instances of contamination. The amplified products were separated on a 1% agarose gel and visualized under UV following staining with Sybr Safe (Invitrogen). PCR products were purified with the ExoSAP-IT kit (Amersham Biosciences).

In Leiden, amplification was performed in volumes of 25 µl. The PCR cycle consisted of 3 min initial denaturation at 94°C, 15 sec cycle denaturation at 94°C, 30 sec cycle at 50°C, 40 sec cycle extension at 72°C for 40 cycles. After all cycles had finished, a final extension was performed at 72°C for 5 min. The amplified products were separated on a 1% agarose gel and visualized under UV following staining with ethidium bromide.

The sequencing at Padova was performed at the BMR Genomics Service (Padova, Italy) in an ABI PRISM automatic DNA sequencer (Applied Biosystems), in both forward and reverse direction, but for some samples only in forward direction. In Leiden PCR clean-up and sequencing was outsourced to MACROGEN on an ABI 3730XL, all samples were sequenced in both forward and reverse direction. The chromatograms were checked with Sequencher (Gene Codes Corporation) and the resulting sequences were aligned by eye in BIOEDIT 7.0.9.0 (Hall 2004).

Tree analysis

Neighbor-joining (NJ) trees based on DNA barcode sequences of all available specimens were reconstructed with Paup* 4.0b10 (Swofford 2003). Genetic distance calculations were performed both using the Kimura two-parameter (K2P) model and uncorrected P distance (Srivathsan and Meier 2011). After initial analyses with barcodes of Italian Phyllocnistis vitegenella, we excluded this gracillariid from subsequent analyses (because it was so divergent from focal Heliozelidae: minimum K2P distance being greater than 18%). A Genbank sequence of Incurvaria masculella (Denis & Schiffermüller, 1775) (Incurvariidae), another member of the superfamily Adeloidea, was used as the outgroup. Bootstrap values were calculated with 10, 000 replicates.

Phylogenetic trees based on maximum parsimony were generated with PAUP using a heuristic search, 1, 000 replicates, with tree-bisection-reconnection (TBR) as the branch-swapping algorithm. A bootstrap analysis was run with TNT (Goloboff et al. 2008), a program made available with sponsorship of the Willi Hennig Society, for 10, 000 replicates. From the dataset we selected one sequence for all barcode clusters with less than 2% intraspecific distance, but we included four specimens of our target species Antispila oinophylla, two from Italy and two from the USA.

A Bayesian Analysis was carried out with the same dataset. Model selection was performed using jModeltest 0.1.1 (Posada 2008). The best-fit model was chosen based on AIC value (Posada and Buckley 2004). Bayesian analyses were run in MrBayes 3.1.2 (Ronquist and Huelsenbeck 2003). Each analysis was run twice, starting from random starting trees, for 20 million generations and sampling every 1000 generations. Two partitioning schemes were explored: first, each codon position was given a separate partition and rate multipliers, while the second scheme combined first and second codon positions into a single partition with respect the third codon positions (Shapiro et al. 2006). Convergence of the Markov Monte Carlo chains was assessed by plotting the likelihood scores in Tracer v1.5 (Rambaut and Drummond 2007). A conservative burn-in of 5 million generations was chosen.

The sequence data generated and used in this study have been deposited in the public BOLD database (project “Antispila Vine introduction” [ANTVI] and GenBank (Appendix B).

Field observations

Surveys were carried out from 2007 to 2011 to investigate the Antispila oinophylla distribution in northeastern Italy. We sampled commercial vineyards but also isolated vine rows and plants of Virginia creeper, Parthenocissus quinquefolia.

Observations on Antispila oinophyllaphenology and behaviour were carried out in 2008 and 2009 in Borgo Valsugana (Trentino Regione). The vineyard was planted with a Chardonnay cultivar and was trained with the local “pergola” system. The vineyard received a number of fungicide treatments but insecticides were not applied. In 2008, a total of 180 leaves (30 plants, six leaves per vine) were sampled six times during the season, from May to September. In 2009, a total of 100 leaves (five replicates of ten plants, two leaves per plant) taken from the mid part of the shoots were sampled across ten dates, from May to September. In both years the number of mines produced by Antispila oinophyllalarvae was assessed on each leaf. In 2009, active mines containing living larvae were distinguished from those vacated by the larvae (mines with larval cut-outs).

Results Identification

To identify the new Italian Antispila, we checked all descriptions of the Vitaceae miners, as well as all other known Antispila species. Unfortunately, outside Europe, genitalia have been illustrated and described only for Japanese species of Antispila, including all five Vitaceae miners (Kuroko 1961; Kuroko 1987). For the North American fauna, only a revision for three Cornaceae-feeding Antispila (with genitalia illustrations), has been published (Lafontaine 1973). The genitalia of the Italian populations (Fig. 9) did not match any published illustrations. An important external character of the moths is the silver apical spot on the forewing (Figs 1–2), a feature found in just a few members of the genus, whereas the other pattern elements that we examined are more general across the genus. Similarly-sized subapical spots were only noted in descriptions of some Antispila from the New World, although larger subapical patches occur in Japanese species, such as Antispila orbiculella Kuroko, 1961. After excluding a poorly known species from Brazil as a less likely candidate, two North American Vitaceae miners with this spot were studied in more detail: Antispila voraginella Braun, 1927, occurring in Arizona and southern reaches of the Rocky Mountain area, and Antispila ampelopsifoliella, which occurs widely acrosseastern North America. The genitalia of the male holotype of Antispila voraginella did not match, but several specimens identified as Antispila ampelopsifoliella and reared from Vitis, had almost identical genitalia as the Italian populations. However, all specimens of Antispila ampelopsifoliella reared from Parthenocissus, were consistently different (Antispila ampelopsifoliella was described by Chambers from leafmines that he collected on Parthenocissus in Kentucky). Leafmines that we collected in 2011 in northeastern United States on Parthenocissus further showed that at least two species with different mines occur on that host. DNA barcoding results discussed below demonstrated that the Italian and North American examples from Vitis belong to the same species, and that American Parthenocissus feeders belong to two different barcode clusters, supporting our morphological and biological findings that two Antispila species, co-occurred on Parthenocissus in eastern North America. Material from the Chambers collection (see below) was insufficient to confirm the identity of Antispila ampelopsifoliella. Here we restrict the name Antispila ampelopsifoliella to one of the two species feeding on Parthenocissus. The Vitis miner from North America, previously misidentified in collections as being Antispila ampelopsifoliella, is unnamed, morphologically identical to the Italian population, and described below.

Taxonomy Antispila Hübner

Antispila Hübner, [1825]: 419. Type species Antispila stadtmuellerella Hübner, [1825]: 419 (a junior synonym of Antispila metalella ([Denis & Schiffermüller], 1775), subsequent designation by ICZN (1988).

DNA barcoding and species relationships

Barcode analysis

Neighbor-joining trees of all sequenced barcodes, both based on Kimura 2P distances and uncorrected distances give highly similar results in topology and branch lengths, we illustrate here the last one (Fig. 30). All species clusters have a bootstrap value of 100, and within-species variation is usually low or absent. We caution, however, that for several species, such as Antispila treitschkiella or Holocacista rivillei most sequences are from just one or two populations. Two species clusters show large intraspecific distances: the two specimens of Antispila hydrangaeella have 5.22% K2P distance and 4.99% uncorrected pairwise distance, and the species tentatively named Antispila cf viticordifoliella forms two clusters with around 4% distance in both methods. Although the mines of these clusters look superficially the same we have not studied the adults of one cluster, so it is possible that these clusters represent separate species.

We have 20 sequences representing Antispila oinophylla, seven of which are 100% identical, five from Italy (including one from Parthenocissus) and two from North America (RMNH.INS.18392 from Georgia and RMNH.INS.18558 from New York). The others are very similar, with at most five nucleotides differing from those of the core group (RMNH.INS.18394 from Georgia). The genetic distance varies from 0 to 1.23% K2P distance (1.22% uncorrected). The differences occur in 16 different positions, of which six cases are found in more than one specimen (e.g., a G instead of A in position 82 combined with a T in 316; the seven specimens forming a “clade” in Fig. 30 with RMNH.INS.18533 and BVS04; position 550: C instead of T; four specimens forming the “clade” in Fig. 30 with RMNH.INS.18533, position 634 a T instead of A in RMNH.INS.18298 and RMNH.INS.18300, both from Tennessee). Several haplotypes are found both in Italy and North America. The largest distance is between two North American specimens, one from Georgia and one from Tennessee (RMNH.INS.18394 and LGSM035–06). The genetic distance to the closest congeneric species Antispila voraginella is large: more than 10%.

Phylogenetic analyses

The maximum parsimony analysis of the barcode sequences resulted in three shortest trees, of which the 50% majority rule tree is illustrated (Fig. 31). The semi-strict tree differs only in the position of Heliozela aesella, which forms a polytomy with the three main heliozelid clades in Figure 31. Of the 658 characters, 243 characters are parsimony informative. Bootstrap values are taken from the TNT analysis. The two Bayesian analyses of the same dataset showed few differences, we here illustrate the consensus tree based on three partitions (Fig. 32).

Both cladograms are rather similar. Antispila oinophylla forms a highly supported clade. Clades for Heliozela, a core Antispila grouping and a clade with several smaller genera and the Antispila ampelopsifoliella group were recovered, with strong support in the Bayesian analysis for thelatter clade (0.97) and for Heliozela (1) and less support for core Antispila (0.74). Within the core Antispila clade, the two Vitaceae species form a clade, well supported in the Bayesian tree, nested in or sister to the Cornaceae-feeding species.

The Bayesian analysis recovered a monophyletic Antispila ampelopsifoliella group. In both analyses this group clusters with the small genera Coptodisca, Holocacista and Antispilina. These all share the reduced venation as described here for Antispila oinophylla. Relative positions of these small genera and the two clades of Antispila vary amongst various analyses. In the Bayesian tree there is low support for a clade of Antispilina and Holocacista. In none of the analyses was Heliozelidae recovered as a monophyletic group.

Vitaceae-feeding taxa are indicated in the cladograms by a purple colour. If these cladograms correctly represent the phylogenetic history of the Heliozelidae, it appears that Vitaceae were the ancestral hosts for the family.

Comparative notes to other species

Below we will briefly treat the other Vitaceae miners amongst North American and European Heliozelidae and one other closely related species, in order to distinguish them from Antispila oinophylla. As there are several more Antispila species in North America than currently described, this is a preliminary treatment until a thorough revision can be completed. Because we have not yet been able to link some larval barcode clusters to their associated adults, the number of leafmine types described below is higher than the number of adult “species”. Material examined for each of these “taxa” is listed in the Appendix A.

Antispila “vitis1”

Fig. 40

From this barcode cluster we have just two females from Florida (Fig. 40, one barcoded) and one larva from Connecticut, of which it is unclear to what type mine it belongs. The female is indistinguishable externally from Antispila oinophylla. Almost certainly this represents another new species.

Distribution of Antispila oinophylla in Italy (Fig. 62)

In Italy, Antispila oinophylla was detected for the first time in the summer of 2007 in a vineyard located in Valsugana (Borgo Valsugana, Trento province, Trentino-Alto Adige Region). Additional surveys conducted in the late summer of 2007 revealed its occurrence also in the neighbouring Vicenza and Belluno provinces (Veneto Region), particularly in neglected vineyards. In 2008, the distribution of the species did not differ greatly in the Trento province; elsewhere the insect was recorded in commercial vineyards of three provinces of the Veneto Region (Vicenza, Belluno and Treviso), sometimes at significant densities. In a number of vineyards Antispila oinophylla occurred together with Phyllonorycter vitegenella, rarely with Holocacista rivillei. In 2009 and 2010, a dense infestation was detected in commercial vineyards located in the Vicenza province (Breganze), about 80 km south of Borgo Valsugana. In this area severe symptoms had been detected as early as 2006 but they were misidentified as being caused by Holocacista rivillei. Since viticulture of this area is much more extensive than that around Borgo Valsugana, it is likely that Antispila oinophylla was introduced first in the Vicenza province and dispersed from there to the other areas. Also in 2010 the species was recorded in the Verona province, 90 km west of Breganze (E. Marchesini, pers. comm.). The distribution of Antispila oinophylla in Italy in 2010 is presented in Fig. 62.

Field observations in Italy

Observations carried out in winter 2008 showed that fully fed, final instar larvae of Antispila oinophylla overwintered inside their cases, fixed to the vine trunks or training stakes. Most larvae pupated in May and the first adults were seen in early June. Mines were detected first in the second half of June. Larvae of the penultimate instar cover the internal surface of the mine with a thin layer of silk, cut away an oval leaf section from both the upper and lower leaf surfaces, and then formed a case by joining the excised leaf sections with silk. Case-bearing larvae move slowly on the leaf surface and then descend with a silken thread until they contact a trunk, training stake, or other solid object to which they affix their case. In the experimental vineyard, the first cases were observed in the first half of July. An additional generation occurred from the second half of August onwards. In 2008, 86.9% of the leaves were infested with a density of 3.26 ± 0.25 (mean ± standard error) mines per leaf by the end of the first generation. In late summer, by the end of the second generation, 95.6% of leaves were infested with an average of 5.44 ± 0.37 mines per leaf.

Observations carried out during 2009 in the same vineyard, confirmed the existence of two generations. Adults were detected from early June to early July. The first mines were observed in mid-June and the first cases in late June (Fig. 63). Larval densities of the first generation peaked in early July, and by late July most mines had been abandoned by the larvae. Mines of the second generation were visible beginning in the second half of August. In the first generation, 96% of leaves were infested with an average of 4.6 ± 0.53 mines per leaf. In the second generation 97% of leaves were mined with an average density of 6.67 ± 0.72 mines per leaf. Active larvae were found until mid-October.

Discussion Taxonomy and identification

Identification of the unknown leafminer proved to be difficult. Many groups of Microlepidoptera remain poorly studied taxonomically. Even in North America, Powell and Opler (2009) estimated that at least one third of the microlepidopteran fauna is still undescribed. There seems to be little chance of overcoming this situation, and even groups feeding on economically important plants such as Vitis species, remain unstudied. Although we had assembled substantial material of Antispila, the morphological similarity across the genus was confusing, and only after checking several genitalia slides and COI barcodes did it become clear that what was previously called “Antispila ampelopsifoliella” was composed of at least two cryptic species on different hostplants. Finding COI barcode matches, in order to rule out the possibility of a non-American sibling species, took more time, because of lack of fresh material and because the Vitis miners in North America are more diverse than previously thought. An initial matching of the Italian pest’s barcode with that of an Antispila record in the BOLD identification system collected in the Great Smoky Mountains, helped focus our research efforts, and underscored the importance of a public DNA barcode reference database. In 2010 and 2011, with increased geographic and taxonomic sampling, we were able to confirm initial results and match additional sequences to those of the introduced Italian Antispila populations. The facts that several North American specimens show a 100% identical barcode to the majority of Italian specimens, the overall small genetic distances across all Italian specimens, and that the largest COI distance found was between two North American specimens, corroborate our position that the Italian populations represent a recent introduction from North America. All Heliozelidae species in this study differed sufficiently in their barcodes to allow reliable identification. The barcode data of North American material in addition showed us that the groups of Vitis and Parthenocissus miners are more diverse than currently recognized and that we cannot identify all taxa with certainty based strictly on morphological grounds. We also note that the North American Vitaceae-feeding Antispila exhibit important differences in male secondary characters and genitalia. A revision of the genus is much needed, but was not possible in the context of this study, where a name was urgently needed for a pest of grapevines. Elsewhere, for example in mainland Asia, the group of Vitis miners is completely unworked, and in need of taxonomic study (before new outbreaks occur).

We emphasize here the importance of combining traditional morphological descriptions with the additional dataset of DNA sequences for taxonomic groups whose identification is particularly difficult and mainly based on the description of genitalia.

An interesting observation is that we did not find any occurrence of Antispila oinophylla on Parthenocissus in its natural habitat in North America, although it utilizes that host in Italy. We found Antispila oinophylla mines on Vitis growing intertwined with Parthenocissus vines, that harboured two different species of Antispila, all occurring within a few centimetres of each other. Despite this sympatry, we did not find any indication of host shifts.

Phylogeny

While it is generally inadvisable to rely solely on DNA barcodes for phylogenetic inferences, several recent studies suggest that some phylogenetic information could be taken from both the sequences themselves or translated amino acids (Wilson et al. 2011). Our phylogenetic results show that on the basis of the COI barcode, Antispila is a paraphyletic genus in relation to the genera Holocacista, Antispilina and Coptodisca. A generic revision of Heliozelidae has not been published, but the late Ebbe Nielsen made a primer to such a revision in his unpublished thesis (Nielsen 1980b) that has been examined by the senior author. Nielsen recognised three clades, one with Heliozela and some related genera, one with Antispila and Antispilina and a final one with Ischnocanaba Bradley, 1961 (from the Solomon Islands), Holocacista, Coptodisca and a new South American genus. The only difference with our findings is the position of Antispilina. Interestingly the clade of the Antispila ampelopsifoliella group with Holocacista, Antispilina and Coptodisca, as we find it, is characterised by the very similar reduced venation. A reduced venation has been reported before from some exotic Antispila (Kuroko 1961), but for instance all Japanese species seem to share the complex venation of the core Antispila as illustrated here (Kuroko 1961; Kuroko 1987). Another character noted by Nielsen to group Holocacista and Coptodisca is the habit of larvae to attach their cases to stems rather than the soil. This behavioural character is shared with Antispila oinophylla and other members of the Antispila ampelopsifoliella group. Despite the poor support for this clade on the basis of barcodes, the mitochondrial and behavioural data collectively suggest that Nielsen’s groups could be good and, should such prove to be the case, the genus Antispila will need to be subdivided into at least two genera. Alternatively, many of the smaller genera would need to be synonymised into one large Antispila, or the Antispila ampelopsifoliella group and the smaller genera should be combined in one genus. In the latter case, the generic name would become Coptodisca, which is unfortunate, since the genus in its current circumscription is well recognisable both in morphology and biology. In any case, such decisions are outside the scope of the present paper and should be made after a careful phylogenetic generic revision.

Another interesting result from our provisional phylogenetic analyses is the hypothesis that Vitaceae may form the ancestral hostplants for modern Heliozelidae. For the basal genus Plesiozela Karsholt & Kristensen, 2003 no host plant information is available (Karsholt and Kristensen 2003). Vitaceae occupy a rather isolated position in the angiosperm phylogeny, as sister to all core rosids (Wang et al. 2009). Other Heliozelidae feed on a wide variety of angiosperm families, but most on “eudicots”. Heliozela species feed mostly on rosids (Fagaceae, Betulaceae, Myrtaceae), Antispila species usually on asterids (Cornaceae, Rubiaceae), Coptodisca species on both rosid and asterid trees or shrubs (Davis 1998, van Nieukerken unpublished). Still, these results should be regarded as provisional hypotheses, and should be vigorously tested by analysing additional taxa and genes, as well as using morphological characters.

Introduction in Italy

Antispila oinophylla is the first alien species of Heliozelidae introduced into Europe (Lopez-Vaamonde et al. 2010). Since our manuscript was finished a second species of Heliozelidae from North America was reported as introduction to Italy: a Coptodisca species on Juglans (Bernardo et al. 2011).

Factors leading to the introduction of Antispila oinophylla in Italy are unknown. Antispila oinophylla is the most recent Nearctic insect species reported to be damaging grapevines in Italy (first in Europe). Its invasion follows those of Phyllocnistis vitegenella (Posenato et al. 1997) and Erasmoneura vulnerata Fitch, 1851 (Hemiptera: Cicadellidae) (Duso et al. 2005). Trade of vines from North America to Italy is limited while that of the alternative host Parthenocissus quinquefolia seems to be more intense. However, the absence of records of Antispila oinophylla on Phyllonorycter quinquefolia in North America makes introduction with Virginia creeper a less likely pathway. Anyway, because the caterpillars routinely attach their cocoons to debris, stems or stakes, transport of Antispila cases is probably common, and thus not unlikely to have happened. With the frequency of modern air traffic even the transport of adults, and in particular gravid females is not impossible. The fact that Antispila oinophylla is an abundant and widespread species in eastern North America, together with its life history, makes such a possibility even more likely. However, it is also a warning that other species with similar life styles could be the next introduction, with an unpredictable outcome. Introduction from North America apparently occurs rather commonly; 16.5% of alien Lepidoptera species in Europe originate from North America (Lopez-Vaamonde et al. 2010).

The presence of several North American haplotypes of the DNA barcode in Italian material of Antispila oinophylla may indicate that the introduction could have involved more than a single introduction event.

Infestation

Early observations, carried out during 2007 and 2008 in the Trento province, showed that the incidence of infestation by Antispila oinophylla was significant in vineyards not treated with insecticides. By 2009 significant infestation levels were observed in several commercial vineyards in the Trentino and Veneto Regions despite the application of insecticides. Phyllocnistis vitegenella is also increasingly important in commercial vineyards in northeastern Italy. Native parasitoids showed some effects in keeping Phyllonorycter vitegenella below economic thresholds (Marchesini et al. 2000). Local outbreaks could be associated with the use of broad-spectrum insecticides, probably because they knock out many egg and larval parasitoids, thereby disrupting the interactions between the pest and its natural enemies. Similar mechanisms could affect the relationships between Antispila oinophylla and its parasitoids. Knowledge of such relationships will be required to understand fully what pest status Antispila oinophylla might reach in the future. In Trento province, presently, the role of predators and parasitoids in controlling Antispila oinophylla appears to be negligible. However, in the Veneto the situation is different, with 32 to 48% of the larvae and pupae in late summer being parasitized (C. Duso and A. Pozzebon, unpublished data). The identification of parasitoids from Italian vineyards is in progress. Phyllocnistis vitegenella and Holocacista rivillei share a number of parasitoid species (Mariani 1942; Camporese and Marchesini 1991; Alma 1995; Marchesini et al. 2000). It is therefore likely that some of these will also be found to attack Antispila oinophylla populations.