Zoologica Scripta

Deep cox1 divergence and hyperdiversity of Trigonopterus weevils in a New Guinea mountain range (Coleoptera, Curculionidae)

ALEXANDER RIEDEL,DAAWIA DAAWIA &MICHAEL BALKE

Submitted: 31 March 2009 Riedel, A., Daawia, D. & Balke, M. (2010). Deep cox1 divergence and hyperdiversity Accepted: 8 July 2009 of Trigonopterus weevils in a New Guinea mountain range (Coleoptera, Curculionidae).— doi:10.1111/j.1463-6409.2009.00404.x Zoologica Scripta, 39, 63–74. Trigonopterus is a little-known genus of flightless tropical weevils. A survey in one locality, the Cyclops Mountains of West New Guinea, yielded 51 species, at least 48 of them undescribed. In this study, we show that mtDNA sequencing, or DNA barcoding, is an effective and useful tool for rapid discovery and identification of these species, most of them morphologically very difficult to distinguish even for expert taxonomists. The genus is hyperdiverse in New Guinea and different species occur on foliage and in the litter layer. Morphological characters for its diagnosis are provided. Despite their external similarity, the genetic divergence between the species is high (smallest interspecific divergence 16%, mean 20%). We show that Trigonopterus are locally hyperdiverse and genetically very strongly structured. Their potential for rapid local biodiversity assessment surveys in Mela- nesia is outlined (a-diversity); providing a regional perspective on Trigonopterus diversity and biogeography is the next challenge (b-diversity). Corresponding author: Alexander Riedel, Staatliches Museum fu¨r Naturkunde, Erbprinzenstr 13, D-76133 Karlsruhe, Germany. E-mail: [email protected] Daawia Daawia, Jurusan Biology, FMIPA-Universitas Cendrawasih, Kampus Waena-Jayapura, Papua, Indonesia. E-mail: [email protected] Michael Balke, Zoologische Staatssammlung, Mu¨nchhausenstr. 21, D-81247 Mu¨nchen, Germany. E-mail: [email protected]

Introduction many of the species are still undescribed and it is almost Beetles form a major part of global biodiversity and with impossible to identify the described ones without compari- 62 000 described species and possibly more than 150 000 son with the historic type material. A commonly applied undescribed ones, the weevils (Curculionoidea) are the remedy in all sorts of biodiversity studies is sorting to dominating group among them (Oberprieler et al. 2007). morphospecies by parataxonomists. This approach has its Weevils are especially diverse in tropical forests. Although merits, but the results depend strongly on the experience they constitute a high proportion of terrestrial biodiver- of the parataxonomist, regardless of the academic level or sity, they are largely ignored during surveys and decision cultural background. Not surprising, sorting to morpho- making in conservation programmes. For example, during species (and the authors include themselves here when it a Rapid Biodiversity Assessment (RAP) of Conservation comes to taxa they are not particularly familiar with) has International in the Cyclops mountains of West New Gui- be shown to suffer often from high error rates (Krell nea, only four entomologists (covering butterﬂies and 2004). aquatic insects) out of 30 scientists participated, whereas DNA sequencing has recently been promoted as a tool the remainder studied vertebrates or plants (Richards & to help overcome this impediment, DNA barcoding being Suryadi 2002). One of the reasons behind this imbalance currently the most popular approach (Hebert et al. 2003; is certainly the ‘taxonomic impediment’ (Hoagland 1996): Schindel & Miller 2005; Stoeckle & Hebert 2008). How- the lack of taxonomic expertise for many of the megadi- ever, surprisingly few studies have to date sought to use verse taxa such as weevils (Wheeler et al. 2004). In fact, DNA sequence data, or barcode data, for rapid biodiversity

ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters d Zoologica Scripta, 39, 1, January 2010, pp 63–74 63 Hyperdiversity of Trigonopterus weevils d A. Riedel et al. assessment. Initial studies by Smith et al. (2005) (Madaga- examined for safe identification. However, not a single scan ants) and Borisenko et al. (2008) (Surinamese small illustration of these structures has ever been published. mammals) found good congruence between morphological Preparation of specimens, especially extraction of their identification and barcode clusters. Among the ants, some- genitalia, is difficult and time-consuming. The fusion of times more conservative morphospecies than molecular body sutures and the extensive defense adaptations make a ‘species’ delineation (i.e. genetically diverse morphospecies fast dissection, comparable with that of other beetles, with up to 16% cox1 p-distance), and higher b-diversity impossible even for an expert. Thus, the routine identifica- from molecular taxonomic units vs. morphospecies were tion of a large number of specimens on the basis of mor- found. This means that different localities have very simi- phological characters alone is impossible for Trigonopterus. lar looking species which parataxonomists tend to lump together into the same morphospecies, a well-documented The area problem with morphospecies sorting, cf. Krell (2004). The Cyclops Mountains are one of the most clearly While sequence data can thus assist in the rapid delivery defined and isolated New Guinea mountain ranges, situ- of biodiversity assessment of relevant data, it has also been ated on the North coast of Indonesian Papua (formerly shown that incomplete lineage sorting and organelle cap- Irian Jaya), close to the border with Papua New Guinea ture might lead to erroneous results (e.g. Monaghan et al. (Fig. 1). They extend over 45 km from Tanahmerah Bay 2006). Consequently, taxa require individual evaluation. In in the West, to Jayapura, the capital of Papua, in the East. this study, we explore the potential of DNA sequencing or To the South, they are bordered by Lake Sentani; to the barcoding for rapid surveys of a very diverse but neglected North, they drop to the Pacific Ocean. The highest peak, group of phytophagous beetles in Melanesia. Mt. Rafeni, is 1880 m high (van Royen 1965; Polhemus & Richards 2002). The taxon The Cyclops Mountains represent a former island that Trigonopterus Fauvel is a genus of flightless, apterous wee- was accreted to northern New Guinea in the Tertiary. vils placed in the Cryptorhynchinae of Curculionidae There are uncertainties about the precise timing of events, (Morimoto 1978; Alonso-Zarazaga & Lyal 1999), distrib- but the formation of the island arc may have begun in the uted from Eastern Sumatra to Samoa and from New Cale- Eocene, while the collision occurred between Miocene and donia to the Philippines. The centre of its diversity Pliocene (Monnier et al. 1999). Probably, the same island appears to be in New Guinea. Trigonopterus inhabit tropi- arc also included all or some of the following terranes: Ar- cal forests, both on foliage and in the litter layer. In New fak, Biak, Japen, Van Rees, Foja, Torricelli, Prince Alexan- Guinea, the foliage-frequenting species are a prominent der, Adelbert, Finisterre and Saruwaged (Polhemus 2007). element of the local arthropod fauna. Sometimes more The vegetation of the Cyclops Mountains was studied than 50% of the phytophagous beetles collected by beat- in detail by van Royen (1965). Forest types change quite ing forest undergrowth are Trigonopterus. They possess a markedly with elevation and range from tropical lowland pectoral receptacle into which the rostrum and the anten- forest to montane forest dominated by Araucaria and Not- nae can fit when the weevil is alarmed. Trigonopterus wee- hofagus in the upper elevations. Such montane habitats are vils attain a typical defense position, thanatosis, by folding well-isolated from others in the region. The nearest area the legs over the ventral side of the body. Also, the elytra would be the Bewani Mountains in the East (c. 100 km), are fused with each other and with the thorax, thus form- the Central Range in the South (c. 130 km) and the Foja ing a solid capsule. In combination with an extraordinary Mountains in the West (c. 150 km). These montane habi- well-sclerotized cuticle, the body of these weevils can tats are likely to harbour a variety of endemic plant and withstand considerable pressure. animal species. Sir David’s Long-beaked Echidna (Zaglos- Trigonopterus weevils are excellent indicator organisms sus attenboroughi) is so far known to inhabit the upper ele- for biodiversity monitoring in Melanesia, being present vations of the Cyclops Mountains only. It was described as and easily collected in the majority of natural habitats. recently as 1998 by Flannery and Groves. There are no Their a-diversity is high and our unpublished data also endemic bird or butterfly species recorded, which can be suggest very high b-diversity. Trigonopterus’s ability to dis- accounted for by their high mobility. A look at other, less perse over degraded habitat is presumably very limited mobile species would be rewarding to identify endemism because of their flightlessness and small size. Yet, Trigon- on species level. opterus remained poorly known. The majority of the 90 Apart from its rich biodiversity, its importance as a named species were described before 1930 (Pascoe 1885; water catchment area for nearby Sentani and the provincial Faust 1898; Heller, 1916). The species are usually very capital Jayapura is one of the reasons why the area is offi- similar to each other externally, and genitalia need to be cially protected in the 22 500 ha Cyclops Mountains

64 Zoologica Scripta, 39, 1, January 2010, pp 63–74 d ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters A. Riedel et al. d Hyperdiversity of Trigonopterus weevils

Fig. 1 Map of New Guinea with enlarged selection of the Cyclops Mountains. The sampled transect is indicated by dots.

Reserve (Ratcliffe 1984). However, in recent years, popula- is very hard to reach and was visited only once when tion pressure increased markedly, especially from the Sen- weather conditions were favourable. tani area. Illegal gardening using slash and burn Two collecting methods were applied, based on previ- agriculture is evident, in areas up to an elevation of about ous experience that different species of Trigonopterus occur 700 m. In December 2006, following heavy rainfalls, land- on the foliage and in the litter layer. The former were col- slides occurred on the Southern slopes that caused major lected by hand and by using a beating sheet, beating the damage to environment and infrastructure. The area of lower vegetation along the trail. Foliage higher than about the Cyclops Mountains is severely threatened and effective 2.5 m was not covered in this survey. Specimens were col- protection measures are urgently needed. lected in a tube of absolute ethanol. The tube was changed after an elevational interval of 200–300 m. Elevations were Materials and methods read from the altimeter of a Suunto X-Lander. Field sampling Edaphic species were collected by sifting the litter layer. Field work was conducted on 6 days within 2 weeks in A beetle sieve (after Reitter, as sold by H. Winkler, Vienna) November ⁄ December 2007. A transect was sampled was used, which consists of 1 cm hardware cloth ﬁxed in a between 300 and 1520 m elevation (Fig. 1). Specimens ripstop nylon sifter. Sampling consisted of vigorous sifting collected from foliage in the upper elevations (1200– of leaf litter and twigs down to, but not including, soil. Sift- 1520 m) may be somewhat under-represented as the area ing for each sample was performed for about 30–45 min

ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters d Zoologica Scripta, 39, 1, January 2010, pp 63–74 65 Hyperdiversity of Trigonopterus weevils d A. Riedel et al. until 5–10 L of sifted litter had accumulated. Precise sample scribed genus of wingless Cryptorhynchinae (ARC600). locations were selected based on suitable terrain and the The latter two species possess a reduced subtriangular presence of good litter accumulations. It was attempted to metanepisternum and tarsi of the plesiomorphic type with cover the full elevational range of the transect evenly with larger divaricate claws. litter samples. It was attempted to cover a variety of litter Previous experience had shown that many Trigonopterus types (near big trees ⁄ between undergrowth ⁄ on ridge ⁄ on species differ by more or less subtle external morphologi- level ground) within each sample location within a distance cal characters, such as size, sculpture, dorsal outline and of about 25 m from the fixed sampling locality. Coordinates lustre. Based on these characters, a pre-selection was were measured with Garmin GPSmap 60CSx. Samples were made. It was attempted to include at least three specimens carried back to the hotel in the evening. Altogether, 16 litter of each ‘morphospecies’ if series were available. These samples were collected. were selected from a maximum range, i.e., different speci- The litter samples were processed to maximize specimens were chosen from samples of the lowest ⁄ the highest mens obtained within a short period. First of all, sample elevation. Genital characters could not be consulted at this bags were carefully opened and specimens found on top of stage because the genitalia are held inside the abdomen by the litter ⁄ on the cloth of the bag were taken and preserved extensive muscle tissue. It is only after maceration that the in absolute ethanol. During the first night after sampling, genitalia can be safely extracted from the specimen. Thus, this method yielded a relatively large number of speci- these preliminary ‘morphospecies’-hypotheses needed revi- mens. On the following day, portions amounting 1–2 sion in a second step after genital characters had been handsful of sifted litter were placed on a white rubber studied and specimens were properly mounted. Altogether, cloth and placed in the sun. Specimens usually started 222 Trigonopterus were selected, plus 3 outgroups. walking and became visible after 5 min. Subsequently, the litter was placed into ecclectors (Winkler and Konzelman Preparation of specimens ecclectors) for another 2–3 days. When no additional spec- DNA was extracted nondestructively to allow subsequent imens appeared, the dry litter was disposed. A total of examination of genital morphology and photographic doc- 1265 specimens of Trigonopterus were collected. umentation, followed by museum storage of voucher specimens. Standard procedures of beetle dissection had to be Diagnostic characters of Trigonopterus and selection of modified: the beetles were placed individually in an specimens embryo dish containing absolute ethanol. The prothorax Although Trigonopterus is morphologically and phylogenet- was carefully pulled from the mesothorax, rupturing the ically clearly defined, published information on how to thin connecting membrane between the two and thus separate it from similar weevils is insufficient. A study to splitting the body into an anterior and a posterior part. identify autapomorphies of the genus is in progress (by A. Both parts were used for DNA extraction. After the Riedel). Characters that were used in this study to diag- extraction, the digested specimens were stored in ethanol nose Trigonopterus are outlined below, as a number of and softened in 10% acetic acid for 1 day before further other weevil genera occurring in the area could easily be preparation. After the extraction of the genitalia, the ante- confounded with Trigonopterus. rior and posterior body parts were put together again. Trigonopterus species are usually subglabrous, without or This step works surprisingly well in Trigonopterus, because with only few scales, whereas the majority of other Cryp- the mesothorax forms a distinct anterior neck which fits torhynchinae have a dense vestiture of scales or setae. into the prothorax, forming a natural plug connection. Exceptions occur in both groups; therefore, this character Winged beetles of comparable size would usually fall is only useful for presorting. Furthermore, only apterous apart after such a preparation as the elytra get detached Cryptorhynchinae with a reduced or absent metanepister- from the thorax easily. Instead, in such groups, the abdo- num were considered to be closely related to Trigonopterus. men is removed while slightly opening up the elytra. In This excludes fully winged genera such as Imathia (Curcu- Trigonopterus, this standard procedure cannot be used lioninae: Storeini) or Semiathe (Cryptorhynchinae), which because the elytra are fused with each other and to the possesses an elongate metanepisternum. Trigonopterus was base of the abdomen. Thus, removal of the abdomen diagnosed based on the complete loss of the metanepister- would inflict irreparable damage to the specimen. num and by an apomorphic structure of the tarsus with However, because of the fusion of elytra and thorax in minute claws and a deeply incavated articulation of tarso- Trigonopterus, it is safe to split these specimens between mere 4. One fully winged species of Semiathe (code meso- and prothorax. This is also useful for the subse- ARC689 on the tree) was included as outgroup, as well as quent preparation of the genitalia. The abdominal tip is one wingless species of Telaugia (ARC602) and an unde- often deeply retracted dorsally into the elytra. It can be

66 Zoologica Scripta, 39, 1, January 2010, pp 63–74 d ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters A. Riedel et al. d Hyperdiversity of Trigonopterus weevils pushed ventrad and opened up with a pin that is inserted ware (written by Gaurav Vaidya, Singapore. http://code. anteriorly through the thoracic opening. The genitalia are google.com/p/taxondna/; http://taxondna.sourceforge.net/) then removed from the abdominal apex. They can be was used to study genetic divergences in our data set, to pulled out in better condition from the extracted specimen cluster sequences at different set thresholds (Meier et al. than possible in fresh ones: extensive muscle tissue sur- 2006) and to calculate the smallest interspecific distance rounding the long ductus ejaculatorius respectively ductus (Meier et al. 2008). spermathecae is cleared by enzyme digestion. In specimens According to Hebert et al. (2003), cox1 divergence even with intact and somewhat hardened tissue, these parts between closely related hexapod species is roughly often tear off and remain in the specimen when the genita- between 6% and 15%, depending on the order. We thus lia are removed. evaluated clustering at such thresholds as a potentially fast Thus, with the preparation outlined above, the DNA tool to estimate diversity with our data set. We have extracted specimens remain without visible damage and shown elsewhere that divergence between recent sister can be used as type material without any problem. The species might be much smaller (Monaghan et al. 2006; genitalia of all the extracted specimens were macerated Balke et al. 2007), but understand that the magnitude of with 10% KOH, rinsed in diluted acetic acid, stained with genetic divergence between prospective species has to be an alcoholic Chlorazol Black solution and stored in glyc- evaluated strictly on a case-by-case basis. We also use erol in microvials attached to the pin of the specimens. nested cladistic phylogenetic analysis as implemented in Photographs were taken with an automated Leica Z6 APO the program TCS 1.13 (Clement et al. 2000). The obtained (Leica Microsystems, Wetzlar, Germany), a JVC KY70 statistical networks subdivide the variation on the basis of camera (JVC Professional Products) and Automontageª the level of homoplasy within the data themselves, i.e. dis- software (Syncroscopy, Cambridge, UK). tinguish between long (homoplastic) and short (nonhomo- plastic) branches. This provides a relative measure of DNA sequencing and sequence analysis divergence within a given data set, rather than a predeter- Genomic DNA was extracted from individual beetles using mined phenetic cut-off value (Monaghan et al. 2006). the DNeasy tissue kit (Qiagen, Hilden, Germany). Using standard PCR protocols and double-stranded sequencing Results with PCR primers, we sequenced the 5¢ end of cyto- Initial morphospecies count chrome c oxidase 1 (cox1). At first, standard primers were On the basis of the initial study of the external morphol- used: LCO1490: 5¢-GGTCAACAAATCATAAAGA- ogy, we recognized 40 morphospecies. TATTGG-3¢, HCO2198: 5¢-TAAACTTCAGGGTGAC- CAAAAAATCA-3¢ (Folmer et al. 1994). For about 30 Second morphospecies count samples in which PCR amplification with these primers After examination of genital morphology of all specimens, failed, primers adjusted for Cryptorhynchinae were we recognized a total of 51 Trigonopterus species. used: LCO1490-JJ CHACWAAYCATAAAGATATYGG, HCO2198-JJ AWACTTCVGGRTGVCCAAARAATCA Sequence data (Astrin & Stu¨ ben 2008). Sequences were edited using the The sequences had no indels and were thus easily aligned Sequencher 4.2 software package (GeneCodes Corp., Ann on the nucleotide level. Total cox1 uncorrected p-distance Arbor, MI, USA) and submitted to GenBank [accession in the 222 Trigonopterus sequences was 0.0–27.5%. Both numbers FN429126 - FN429350 (will be added in proof)]. neighbor-joining and maximum-likelihood analysis pro- duced strongly structured trees, with deeply structured Sequence data analysis monophyletic clusters. These visually obvious clusters A data matrix was generated in Sequencher and exported were congruent with the 51 morphospecies based on exter- to MacClade (Maddison & Maddison 2000) to inspect the nal plus genital morphology. The smallest interspecific amino acid sequence to confirm the nucleotide alignment morphospecies distance was 16.5% (mean 20.5%) in Trig- and for quality control. As no stop codons or other suspi- onopterus. Intraspecific distances ranged from 0.0% to cious features were detected, we assembled a 658-bp data 8.8% (mean 1.2%). set in PAUP* 4.0b10 (Swofford 2002), trimming the sequence ends to some degree to avoid too many missing Species recognition based on DNA sequence data data. The data were then exported, and maximum-likeli- Clustering the data at thresholds of 8–12% revealed 51 hood analysis ran in GARLI (Zwickl 2006). Data structure clusters that corresponded to the 51 recognized morpho- was also visualized by compiling a neighbor-joining tree in species (Fig. 2). Clustering at 13% revealed 50 clusters, PAUP*, using uncorrected p-distances. Taxon DNA soft- but only 49 clusters contained only one morphospecies,

ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters d Zoologica Scripta, 39, 1, January 2010, pp 63–74 67 Hyperdiversity of Trigonopterus weevils d A. Riedel et al.

Nested cladistic phylogenetic analysis with TCS software set to 95% connection limit identiﬁed 62 statistical networks, in which nine of the 51 morphospecies were split into two or three networks: 2, 8, 11, 12, 30, 38, 39, 48 and 51. Number of ‘perfect clusters’ was thus 42.

Taxonomy ⁄ new species The 51 Cyclops species were compared with our complete image database of type specimens for the Trigonopterus described from the Papuan region. Three of them could be tentatively assigned to an available species name as they agreed relatively well with characters of the type specimens. However, no ﬁnal identiﬁcation could be given as the type series of Trigonopterus obnixus (Faust) possibly contain more than one species and as the type localities of Trigonopterus oblongus (Pascoe) and Trigonopterus illex (Faust) are quite distant from the site of this study and minor differences do exist. Thus, our survey uncovered 48 new species.

Habitat specialization Of the 51 species, 20 are edaphic, while 31 are foliage- frequenting species (Figs S1–S51). Both habitats appear clearly separated in Trigonopterus: none of the edaphic species was collected while beating foliage, whereas only two specimens of foliage-frequenting species were found in litter samples. Such a clear separation is remarkable in Cryp- torhynchinae that are usually xylophagous and may be found occasionally as ‘tourists’ on foliage.

Discussion Species count The first morphospecies hypotheses which were merely based on external morphology showed two different types Fig. 2 Two Trigonopterus species from the Cyclops Mountains. of conflict with molecular data. First, externally similar spe- Habitus of male (left) and its aedeagus (right). Note that species cies had been overlooked. In two cases, three species had externally very similar usually can be safely separated by morphology of male genitalia. been lumped (species 19 ⁄ 20 ⁄ 21; sp. 26 ⁄ 27 ⁄ 30); in seven cases each, two species were lumped (sp. 7 ⁄ 8; sp. 9 ⁄ 10; sp. one cluster contained members of species 7 and 8; cluster- 16 ⁄ 17; sp. 18 ⁄ 22; sp. 23 ⁄ 25; sp. 33 ⁄ 34; sp. 46 ⁄ 47). ing at 14%: 48 clusters, 45 of them corresponding to rec- Second, a small number of errors occurred when assign- ognized morphospecies; 15%: 46 ⁄ 42; 16%: 31 ⁄ 30 (i.e., ing specimens to the initial 40 ‘morphospecies’. Eleven there was one cluster lumping 21 morphospecies), etc. specimens were assigned to the wrong ‘morphospecies’. Lowering the threshold to 7% has the opposite effect, This can be clearly attributed to the difficult working con- retrieving 52 clusters, 50 of which correspond to the given dition, sorting unmounted specimens in alcohol, which morphology based taxonomy, members of species 12 were does not allow viewing all details and also impedes com- split into two clusters; 6%: 52 ⁄ 50 (species 12 split); 5%: paring a number of specimens side by side. 53 ⁄ 49 (species 11 and 12 both split); 4%: 56 ⁄ 47 (species However, after studying the genital morphology and 30, 39 and 11 split in two each, species 12 split into three revising our morphospecies delineations, we found congru- species); 3%: 56 ⁄ 47 (species 30, 39 and 11 split in two ence between morphological and molecular data. This each, species 12 split into three species); 2%: 59 ⁄ 45 (spe- means that all members of any given morphospecies also cies 30, 39 and 11 split in two each, species 12 and 51 form a monophyletic group on the cox1 tree (Fig. 3). In split into three species each). this study, the morphospecies were supported by long

68 Zoologica Scripta, 39, 1, January 2010, pp 63–74 d ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters A. Riedel et al. d Hyperdiversity of Trigonopterus weevils

Fig. 3 Trigonopterus species from the Cyclops Mountains. Problems in sorting morphospecies based on external characters, caused by sexual dimorphism and intraspeciﬁc variation. Upper row: Difference in habitus is bigger between female (left) and male (middle) of species 10 than between males of species 10 (middle) and species 9 (right). Lower row: Species 18; female (left); smooth male (middle); striate male (right).

branches and were intuitively obvious from the tree topol- Tools have to be fast and easy to use if they are to be use- ogy alone (but see below for the ‘automatization’ issue). ful in routine studies. Clustering at preset thresholds is as The smallest interspecific distance was 16.5% (mean such a very fast and easy to use method. Though threshold 20.5%), values remarkably high for Coleoptera (Meier setting is arbitrary, we suggest that in our local approach et al. 2008); intraspecific distances were smaller, ranging it would have been an efficient diversity approximator. from 0 to 8.8% (average 1.2%). In six species, the intra- Agreement between morphological species count and specific distance was higher than 4%: sp. 8 (4.0–5.7%), sp. mtDNA clusters was perfect at 8–12% (Fig. 2). At 2%, we 12 (4.4–8.0%), sp. 39 (4.5%), sp. 48 (4.7–8.8%), sp. 30 found 59 clusters, which is a 15% overestimation of diver- (4.9%), sp. 11 (5.1–5.9%). No morphological differences sity, with 88% perfect clusters (or 12% error). At 16%, we could be detected within these genetically more divergent found only 31 clusters, which is a 39% underestimation of species, and such cases can only be addressed properly diversity (or error), with only 58% perfect clusters, i.e. based on more sophisticated sampling and an iterative only 30 clusters are congruent with some of the total of approach using different sources of information and more 51 recognized morphospecies. material, i.e. integrative taxonomy. Computation of statistical networks breaking down vari- Species should not be characterized solely based on ation on the basis of the information contained in the data DNA sequence data (Tautz et al. 2002), and our data set set itself revealed 62 networks; assuming these were shows that morphologically identical specimens from the species means 21% diversity overestimation, and 42 same locality can diverge as much as 8.8% cox1 p-distance networks perfectly corresponding to morphospecies, that (e.g. also Balke et al. 2008). Whether these are cryptic spe- is, an error of 18%. cies or not cannot and should not be decided with the information available now. Implications for taxonomy and diversity assessment In a case where morphology and mtDNA data reveal At present, only 46 species of Trigonopterus were formally monophyletic, congruent clusters, seeking to automatize described from the entire Papuan region, including the species-richness estimation to some degree is tempting. Moluccas, the Bismarck archipelago and the Solomon

ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters d Zoologica Scripta, 39, 1, January 2010, pp 63–74 69 Hyperdiversity of Trigonopterus weevils d A. Riedel et al.

islands (Gressitt 1982), including 40 species from New 15 specimens can be dissected and processed per day. Guinea. Only four species were recorded from mainland The speed of DNA barcoding depends much on the West New Guinea (Setliff 2007). Thus, the number of equipment and the workflow, but under average condi- species recorded within a 2-week survey in parts of only tions, the latter should be about two to three times fas- one small mountain range in the region (51 species) ter. Moreover, it is easier to automatize. However, it exceeds the number of species known from the entire must be noted that once the fauna of a given area is region. A comparison with historic type material revealed well-known and documented, it may again be possible that at most three out of our 51 species belong to to identify a great proportion of specimens faster using described species, thus at least 94% or 48 of the species traditional techniques, for example, by combining mor- found are undescribed. A simple (and obviously oversim- phological data with ecological data (e.g., the elevation plified) arithmetic extrapolation would lead to an expected of a given sample ⁄ food-plants, etc.). DNA barcoding number of 1500 species worldwide and to 767 species could then be restricted to the ‘doubtful’ specimens from the Papuan region. Based on personal field observa- that are hard or impossible to identify by morphologi- tions, these numbers appear quite conservative. Thus, cal means. This has proved to be an effective strategy although the number of described Trigonopterus species before, especially under sympatric conditions (Petersen does not appear very impressive, the genus is in fact et al. 2007). hyperdiverse, that is, very species rich as well as abundant. 4 Five species (sp. 17, 20, 26, 30, 34; Figs S1–S51) were New technologies to speed up the process of making this represented by females only. Morphological differences plethora of mostly similar species identifiable would be between males and females are frequently bigger than highly welcome. that between specimens of the same sex of closely In recent years, there has been an intense debate about related species (Fig. 4, species 10), and the morpholog- the benefits of DNA taxonomy or DNA barcoding (e.g. ical characters critical for species identification are Hebert & Gregory 2005; Will et al. 2005 exemplifying found only in males (Fig. 5). Therefore, it is difficult opposing views). What is the practical experience from to decide about the species-status of these specimens our study? based on morphological data alone. 1 Technically, the method applied works well for Trigon- 5 Based on the molecular data, intraspecific morpho- opterus. The strongly sclerotized specimens endure the logical variation in the sexes could be clearly sepa- extraction process well and the morphological charac- rated from specific differences, e.g. in species 18 ters can be studied even more easily after an extraction (Fig. 4). than before. 6 Species were clearly delineated in this study. Molecular 2 cox1 data helped to sort very similar specimens rapidly and morphological data were perfectly congruent. This into well-defined entities or clusters which we, based means that the same (or a very similar) estimation of on additional morphological evidence, suggest to rep- species diversity based on inspection of tree topology resent species. Our starting hypothesis based on exter- and divergence between clusters would have been made nal morphological characters suggested the presence based on molecular data alone. This may change to of 40 species, which means that despite expert taxo- some degree when material from allopatric populations nomic sorting, 11 species or 22% had been over- is included. However, having sampled some variety of looked. There was an especially severe habitats over multiple sampling points and significant underestimation of diversity among the numerous altitudinal gradient, we do not find incongruence of similar foliage-frequenting species (29%). It can be clusters because of introgression and ⁄ or lineage sorting stated with confidence that the percentage of over- phenomena, as seen in other, similarly recent Melane- looked species would have been much higher if the sian radiations (Monaghan et al. 2006). Paraphyly of process had been performed by a nonexpert in weevil one species appeared in a revision of Malagasy ants taxonomy (see Krell 2004). This error rate is higher based on both morphological and molecular cox1 data than relying merely on cox1 in conjunction with com- (Fisher & Smith 2008). putation of statistical networks; it is also higher than From this study, DNA sequencing clearly emerges as an the results of clustering with a preset threshold efficient tool for the rapid and sustainable surveying of an between 2% and 15%. individual area. With this technique, the number of spe- 3 The preparation of genitalia is critical for a reliable cies in samples from a given study site could be easily species assignment based on morphology, but it is a determined, even by a nonexpert. Most importantly, data time-consuming task. If this is carried out with a qual- thus gained can readily be compared with data from any ity as that required for subsequent taxonomic work, c. other area in the world because DNA sequence data are

70 Zoologica Scripta, 39, 1, January 2010, pp 63–74 d ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters A. Riedel et al. d Hyperdiversity of Trigonopterus weevils

Fig. 4 Maximum likelihood tree of 216 specimens of the genus Trigonopterus plus 3 outgroup representatives from the Cyclops Mountains. Clusters correlate with the assignment of 51 morphospecies given to the right.

ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters d Zoologica Scripta, 39, 1, January 2010, pp 63–74 71 Hyperdiversity of Trigonopterus weevils d A. Riedel et al.

Clusters Perfect

Fig. 5 Clustering the cox1 data at different 20 preset thresholds with TaxonDNA software, from 2% to 16% uncorrected p-distance. Left bar = number of clusters 10 found, right bar = perfect clusters, containing all members and only those of a morphospecies delineated based on 0 external and genital characters. Both 2 34567891011 12 13 14 15 16 categories agree setting the threshold at (%) 8–12% in this case. universal, unlike morphospecies descriptions which leave general identification tool, even among allopatric popula- room between different workers (Krell 2004). tions. In the light of this study, one should be prepared for DNA barcoding showed its potential to speed up taxo- a large number of species involved in this region. nomic work of a taxonomically demanding group. Based More fieldwork in the Cyclops Mountains would be on our previous experience with weevil taxonomy (Riedel rewarding, too. So far, the examined material was obtained 2001, 2002, 2006), we suggest that the final species count only from one transect on the Southern slopes. It would could have also been achieved with morphological data be interesting to check if a transect on the Northern alone, but it would have been more time-consuming. This slopes shows a different species composition. Relatively lit- resorts especially to the classification of unique specimens. tle time was spent in the higher elevations above 1400 m. Additional field work would have been necessary to obtain The area between 1520 m and the peak was not sampled enough specimens for a decision of their status. Often this at all. However, a high number of species were exclusively is impossible, especially in the tropics. The same applies found in the montane forests above 1000 m, such as species to intraspecific variation. If long series are available, pref- 4, 5, 6, 16, 17, 20, 24, 29, 30, 38, 39, 40, 44. If information erably in combination with data on the ecology of the spe- on endemic species is an argument for conserving habitats cies, then a decision upon the status of questionable of a given area, then Trigonopterus should be a taxon that specimens can be made equally well with traditional meth- gives strong support for protection of the Cyclops. ods. But the molecular approach proves to be an elegant shortcut. Acknowledgements In future studies, we will include the Trigonopterusof We thank J. Astrin (Bonn) for providing us Cryptorhyn- neighbouring localities, such as the Bewani, Torricelli or chinae-specific primer sequences. Rene Ta¨nzler (Mu¨nchen) Foja Mountains. This could lead to the discovery of closely helped with some of the laboratory work. related sister species. Despite their superficial similarity, it is possible that the species of the Cyclops Mountains are References phylogenetically rather distant – a suspicion nourished by Alonso-Zarazaga, M. A. & Lyal, C. H. C. (1999). A World their long branches. Possibly, more closely related sister Catalogue of Families and Genera of Curculionoidea (Excepting species might occur allopatrically and they would break Scolytidae and Platypodidae). Barcelona: Entomopraxis, 316 pp. down the long species branches. Determining this will be Astrin, J. J. & Stu¨ben, P. E. (2008). Phylogeny in cryptic weevils: molecules, morphology and new genera of western Palaearctic the acid test for a general usage of DNA barcoding as a

72 Zoologica Scripta, 39, 1, January 2010, pp 63–74 d ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters A. Riedel et al. d Hyperdiversity of Trigonopterus weevils

Cryptorrhynchinae (Coleoptera: Curculionidae). Invertebrate ‘barcoding gap’ and leads to misidentification. Systematic Biology, Systematics, 22, 503–522. 57, 809–813. Balke, M., Alarie, Y., Ribera, I. & Wewalka, G. (2007). Monaghan, M. T., Balke, M., Pons, J. & Vogler, A. P. (2006). Molecular Phylogeny of Pacific Island Colymbetini: radiation Beyond barcodes. Complex DNA taxonomy of a South Pacific of New Caledonian and Fijian species. Zoologica Scripta, 36, Island radiation. Proceedings of the Royal Society of London. Series 173–200. B: Biological Sciences, 273, 887–893. Balke, M., Go´mez-Zurita, J., Ribera, I., Viloria, A., Zillikens, A., Monnier, C., Girardeau, J., Pubellier, M., Polve´, M., Permana, H. Steiner, J., Garcıá, M. & Vogler, A. P. (2008). Ancient & Bellon, H. (1999). Petrology and geochemistry of the associations of aquatic beetles and tank bromeliads in the Cyclops ophiolites (Irian Jaya, East Indonesia): consequences Neotropical forest canopy. Proceedings of the National Academy of for the Cenozoic evolution of the north Australian margin. Sciences of the USA (PNAS), 105, 6356–6361. Mineralogy and Petrology, 65, 1–28. Borisenko, A. V., Burton, K. L., Ivanova, N. V., Hanner, R. H. & Morimoto, K. (1978). On the genera of Oriental Cryptorhynchinae Hebert, P. D. N. (2008). DNA barcoding in surveys of small (Coleoptera: Curculionidae). Esakia, 11, 121–143. mammal communities: a field study in Suriname. Molecular Oberprieler, R. G., Marvaldi, A. E. & Anderson, R. S. (2007). Ecology Resources, 8, 471–479, doi: 10.1111/j.1471-8286.2007. Weevils, weevils, weevils everywhere. In Z.-Q. Zhang & W. A. 01998.x. Shear (Eds) Linnaeus Tercentenary: Progress in Invertebrate Clement, M., Posada, D. & Crandall, K. A. (2000). TCS: a Taxonomy. Zootaxa, 1668, 491–520. computer program to estimate gene genealogies. Molecular Pascoe, F. P. (1885). List of the Curculionidae of the Malay Ecology, 9, 1657–1659, doi: 10.1046/j.1365-294x.2000.01020.x. Archipelago collected by Dr. Odoardo Beccari, L. M. Faust, J. (1898). Curculioniden aus dem Malayischen und D’Albertis, and others. Annali del Museo Civico di Storia Naturale Polynesischen Inselgebiet. II. Stettiner Entomologische Zeitung, di Genova, 22, 201–332 + pl. I–III. 59, 140–213. Petersen, F. T., Damgaard, J. & Meier, R. (2007). DNA Fisher, B. L. & Smith, M. A. (2008). A revision of Malagasy taxonomy: how many DNA sequences are needed for solving a species of Anochetus Mayr and Odontomachus Latreille taxonomic problem? The case of two parapatric species of louse (Hymenoptera: Formicidae). PLoS ONE, 3, e1787, 1–23. Flies (Diptera: Hippoboscidae: Ornithomya (Latreille, 1802). Flannery, T. F. & Groves, C. P. (1998). A revision of the genus Arthropod Systematics & Phylogeny, 65, 111–117. Zaglossus (Monotremata, Tachyglossidae), with description of Polhemus, D. A. (2007). Tectonic geology of Papua. In A. J. new species and subspecies. Mammalia, 62, 367–396. Marshall & B. M. Beehler (Eds) The Ecology of Papua, Part I. Folmer, O., Black, M., Hoeh, W., Lutz, R. & Vrijenhoek, R. (1994). The Ecology of Indonesia Series Volume VI (pp. 137–164). NA primers for amplification of mitochondrial cytochrome c Singapore: Periplus. oxidase subunit I from diverse metazoan invertebrates. Molecular Polhemus, D. A. & Richards, S. (2002). Geographic overview Marine Biology and Biotechnology, 3, 294–299. of the Cyclops Mountains and the Mamberamo Basin. In S. Gressitt, J. L. (1982). General introduction. In J. L. Gressitt (Ed) J. Richards & S. Suryadi (Eds) A Biodiversity Assessment of Monographiae Biologicae 42, Biogeography and ecology of New Yongsu – Cyclops Mountains and the Southern Mamberamo Basin, Guinea (pp. 3–13). The Hague: Dr W. Junk Publishers. Papua, Indonesia. RAP Bulletin of Biological Assessment 25 Hebert, P. D. N. & Gregory, T. R. (2005). The promise of DNA (pp. 32–37). Washington, DC, USA: Conservation barcoding for taxonomy. Systematic Biology, 54, 852–859. International. Hebert, P. D. N., Cywinska, A., Ball, S. L. & deWaard, J. R. Ratcliffe, J. B. (1984). Cagar Alam Pegunungan Cyclops Irian Jaya: (2003). Biological identifications through DNA barcodes. Management Plan 1985–1989. Jayapura: World Wildlife Fund. Proceedings of the Royal Society of London. Series B: Biological Richards, S. J. & Suryadi, S. (2002). A Biodiversity Assessment of Sciences, 270, 313–321. Yongsu – Cyclops Mountains and the Southern Mamberamo Basin, Heller, K. M. (1916). Die Ka¨fer von Neu-Caledonien und den Papua, Indonesia (180 pp.) RAP Bulletin of Biological benachbarten Inselgruppen. In F. Sarasin & J. Roux (Eds) Nova Assessment 25. Washington, DC, USA: Conservation Caledonia. A. Zoologie 2 (3): 229-356 + 1 [unn.] + pl. X–XI. International. Hoagland, K. E. (1996). The taxonomic impediment and the Riedel, A. (2001). The pygmaeus-group of Euops Schoenherr convention of biodiversity. Association of Systematics Collections (Coleoptera, Curculionoidea, Attelabidae), weevils associated Newsletter, 24, 61–62, 66–67. with Nothofagus in New Guinea. Journal of Natural History, 35, Krell, F.-T. (2004). Parataxonomy vs. taxonomy in biodiversity 1173–1237. studies – pitfalls and applicability of ‘morphospecies’ sorting. Riedel, A. (2002). Revision of the subgenus Neosynaptops Voss of Biodiversity and Conservation, 13, 795–812. Euops Schoenherr (Coleoptera, Curculionoidea, Attelabidae) Maddison, W. P. & Maddison, D. R. (2000). MacClade – Analysis from the Papuan region. Zoologica Scripta, 31, 135–165. of Phylogeny and Character Evolution, Version 4.07 [Computer Riedel, A. (2006). Revision of the subgenus Metaeuops Legalov of software and manual]. Sunderland, MA: Sinauer. Euops Schoenherr (Coleoptera, Curculionoidea, Attelabidae) Meier, R., Kwong, S., Vaidya, G. & Ng, P. K. L. (2006). DNA from the Papuan region. Zootaxa, 1181, 1–102. Barcoding and taxonomy in Diptera: a tale of high intraspecific van Royen, P. (1965). Sertulum Papuanum 14. An outline of the variability and low identification success. Systematic Biology, 55, flora and vegetation of the Cyclops Mountains. Nova Guinea 715–728. N.S. Botany, 21, 451–469. Meier, R., Zhang, G. & Ali, F. (2008). The use of mean instead Schindel, D. E. & Miller, S. E. (2005). DNA barcoding a useful of smallest interspecific distances exaggerates the size of the tool for taxonomists. Nature, 435, 17.

ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters d Zoologica Scripta, 39, 1, January 2010, pp 63–74 73 Hyperdiversity of Trigonopterus weevils d A. Riedel et al.

Setliff, G. P. (2007). Annotated checklist of weevils from the Austin. Available via GARLI: http://www.bio.utexas.edu/faculty/ Papuan region (Coleoptera, Curculionoidea). Zootaxa, 1536,1– antisense/garli/Garli.html. 296. Smith, M. A., Fisher, B. L. & Hebert, P. D. N. (2005). DNA Supporting Information barcoding for effective biodiversity assessment of a hyperdiverse Additional Supporting Information may be found in the arthropod group: the ants of Madagascar. Philosophical online version of this article: Transactions of the Royal Society of London. Series B: Biological Sciences, 360, 1828–1834. Figs S1–S27 Habitus of different foliage-frequenting Stoeckle, M. Y. & Hebert, P. D. N. (2008). Barcode of life. Trigonopterus species from the Cyclops Mountains in Scientiﬁc American, 299, 66–71. defense position. Illustration number equals number of Swofford, D. L. (2002). PAUP*. Phylogenetic Analysis using species as used throughout the text. Scale line 1 mm. Parsimony (*and other methods), Version 4.0b10 [Computer software and manual]. Sunderland, MA: Sinauer Associates. Figs S28–S51 Habitus of different Trigonopterus species Tautz, D., Arctander, P., Minelli, A., Thomas, R. H. & Vogler, from the Cyclops Mountains in defense position. Illustra- A. P. (2002). DNA points the way ahead in taxonomy. Nature, tion number equals number of species as used throughout 418, 479. Wheeler, Q. D., Raven, P. H. & Wilson, E. O. (2004). the text. (28–31) Foliage-frequenting species; (32–51) Taxonomy: impediment or expedient? Science, 303, 285. edaphic species. Scale line 1 mm. Will, K. W., Mishler, B. D. & Wheeler, Q. D. (2005). The perils Please note: Wiley-Blackwell is not responsible for the of DNA barcoding and the need for integrative taxonomy. content or functionality of any supporting materials sup- Systematic Biology, 54, 844–851. Zwickl, D. J. (2006). Genetic algorithm approaches for the phylogenetic plied by the authors. Any queries (other than missing analysis of large biological sequence datasets under the maximum material) should be directed to the corresponding author likelihood criterion. PhD dissertation, The University of Texas at for the article.

74 Zoologica Scripta, 39, 1, January 2010, pp 63–74 d ª 2009 The Authors. Journal compilation ª 2009 The Norwegian Academy of Science and Letters Copyright of Zoologica Scripta is the property of Blackwell Publishing Limited and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.