Wide DNA Barcode Library As a New Tool for Arctic Research

Molecular Ecology Resources (2016) 16, 809–822 doi: 10.1111/1755-0998.12489

Establishing a community-wide DNA barcode library as a new tool for arctic research

H. WIRTA,1 G. VARKONYI,2 C. RASMUSSEN,3 R. KAARTINEN,4 N. M. SCHMIDT,5 P. D. N. HEBERT,6 M. BARTAK,7 G. BLAGOEV,6 H. DISNEY,8 S. ERTL,9 P. GJELSTRUP,10 D. J. GWIAZDOWICZ,11 12 13 14 15 € € 12 € € 1 L. HULDEN, J. ILMONEN, J. JAKOVLEV, M. JASCHHOF, J. KAHANPAA, T. KANKAANPAA, P. H. KROGH,10 R. LABBEE,6 C. LETTNER,9 V. MICHELSEN,16 S. A. NIELSEN,17 € T. R. NIELSEN,18 L. PAASIVIRTA,19 S. PEDERSEN,6 J. POHJOISMAKI,20 J. SALMELA,21 € P. VILKAMAA,12 H. VARE,22 M. VON TSCHIRNHAUS23 and T. ROSLIN 1,4 1Department of Agricultural Sciences, University of Helsinki, Latokartanonkaari 5, 00790 Helsinki, Finland, 2Finnish Environment Institute, Natural Environment Centre, Friendship Park Research Centre, Lentiirantie 342B, 88900 Kuhmo, Finland, 3Department of Bioscience, Aarhus University, Ny Munkegade 114, DK–8000 Aarhus, Denmark, 4Department of Ecology, Swedish University of Agricultural Sciences, Box 7044, 750 07 Uppsala, Sweden, 5Arctic Research Centre, Department of Bioscience, Aarhus University, Frederiksborgvej 399, DK-4000 Roskilde, Denmark, 6Biodiversity Institute of Ontario, University of Guelph, Guelph, ON N1G 2W1, Canada, 7Department of Zoology and Fisheries, Faculty of Agrobiology, Food and Natural Resources, Czech University of Life Sciences, 165 21 Praha 6 - Suchdol, Czech Republic, 8Department of Zoology, University of Cambridge, Cambridge CB2 3EJ, UK, 9Division of Conservation Biology, Vegetation Ecology and Landscape Ecology, Department of Botany and Biodiversity Research, University of Vienna, Rennweg 14, 1030 Vienna, Austria, 10Department of Bioscience, Aarhus University, Vejlsøvej 25, Silkeborg DK-8600, Denmark, 11Department of Forest Pathology, University of Life Sciences, Wojska Polskiego 71c, Poznan 60625, Poland, 12Finnish Museum of Natural History, Zoology Unit, University of Helsinki, Pohjoinen Rautatiekatu 13, 00100 Helsinki, Finland, 13Mets€ahallitus, Parks & Wildlife Finland, PO Box 94, 01301 Vantaa, Finland, 14Finnish Environment Institute, Mechelininkatu € 34A, 00250 Helsinki, Finland, 15Station Linne, Olands Skogsby 161, 38693 F€arjestaden, Sweden, 16Zoological Museum of the University of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen, Denmark, 17Department of Environmental, Social and Spatial Change, Roskilde University, Universitetsvej 1, PO Box 260, DK-4000 Roskilde, Denmark, 18Sandvedhagen 8, NO-4318, Sandnes, Norway, 19Ruuhikoskenkatu 17 B 5, 24240 Salo, Finland, 20Department of Biology, University of Eastern Finland, P.O. Box 11, 80101 Joensuu, Finland, 21Mets€ahallitus, Ounasjoentie 6, 96101 Rovaniemi, Finland, 22Finnish Museum of Natural History, Botany Unit, University of Helsinki, Unioninkatu 44, 00140 Helsinki, Finland, 23Fakulta¨t Biologie, Universita¨t Bielefeld, Universita¨tsstrasse 25, 33615 Bielefeld, Germany

Abstract DNA sequences offer powerful tools for describing the members and interactions of natural communities. In this study, we establish the to-date most comprehensive library of DNA barcodes for a terrestrial site, including all known macroscopic animals and vascular plants of an intensively studied area of the High Arctic, the Zackenberg Valley in Northeast Greenland. To demonstrate its utility, we apply the library to identify nearly 20 000 arthropod individuals from two Malaise traps, each operated for two summers. Drawing on this material, we estimate the coverage of previous morphology-based species inventories, derive a snapshot of faunal turnover in space and time and describe the abundance and phenology of species in the rapidly changing arctic environment. Overall, 403 terrestrial animal and 160 vascular plant species were recorded by morphology-based techniques. DNA barcodes (CO1) offered high resolution in discriminating among the local animal taxa, with 92% of morphologically distinguishable taxa assigned to unique Barcode Index Numbers (BINs) and 93% to monophyletic clusters. For vascular plants, resolution was lower, with 54% of species forming monophyletic clusters based on barcode regions rbcLa and ITS2. Malaise catches revealed 122 BINs not detected by previous sampling and DNA barcoding. The insect community was dominated by a few highly abundant taxa. Even closely related taxa differed in phenology, emphasizing the need for species-level resolution when describing ongoing shifts in arctic communities and ecosystems. The DNA barcode

Correspondence: Helena Wirta, Fax: +358-9-191-58582; E-mail: helena.wirta@helsinki.ﬁ

© 2015 John Wiley & Sons Ltd 810 H. WIRTA ET AL. library now established for Zackenberg offers new scope for such explorations, and for the detailed dissection of interspeciﬁc interactions throughout the community.

Keywords: arthropod, DNA barcode library, Greenland, high arctic, species diversity Received 24 June 2015; revision received 9 November 2015; accepted 17 November 2015

than previously thought (Hodkinson & Coulson 2004; Introduction Wirta et al. 2015b), but community-level descriptions of DNA sequences offer powerful tools for identifying the food webs including terrestrial arthropods are only members of natural communities (e.g. Kaartinen et al. beginning to emerge from the Arctic (Wirta et al. 2014, 2010) and for describing the interactions between them 2015b). Such descriptions rely on techniques for the iden- (Clare 2014; Wirta et al. 2014). Current techniques offer tification of species remains in gut contents and faecal scope for massive enumeration of taxa occurring at a material, as best resolved by molecular techniques (Clare given site at a given time (Hajibabaei et al. 2006) – but to 2014; Wirta et al. 2014). Overall, confronting each of these collate such information with taxa of known identity, challenges will call for a comprehensive resource allow- ecology and lifestyle, we need reference libraries of DNA ing the species-level resolution of all types of biological barcodes. While DNA barcode libraries are being stocked samples across arctic communities. at an impressive pace, they are often targeted at or To define the composition, phenology and interac- biased towards particular taxa. To offer scope for versa- tions in species of an arctic area, we here report the estab- tile assessment of full communities and food webs, the lishment of a barcode library for species from an establishment of comprehensive reference libraries cov- intensively studied site, the Zackenberg Valley in North- ering major parts of the flora and fauna remains a prior- east Greenland (Meltofte & Rasch 2008). To our knowl- ity for current research. edge, this is one of the first comprehensive DNA barcode Arctic ecosystems are particularly amenable to the libraries established for a terrestrial site, encompassing establishment of comprehensive DNA barcode libraries. more than 80% of the local diversity of animal and vascu- Here, species diversity is relatively low, and vegetation lar plant taxa as previously detected by morphological structure simple and accessible to sampling. Thus, where characters. Drawing on this novel resource, we test the the comprehensive assessment of the fauna of a single ability of DNA barcodes of standard regions (CO1 for tropical site may call for more than 65 person-years of animals, rbcLa and ITS2 for plants) to distinguish among work (Basset et al. 2012), the same can more easily be species within the local community. To demonstrate the achieved in a high-arctic context. utility of this tool, we apply it to community-level arthro- Knowing both the species and their ecology in the pod samples collected by Malaise traps. More specifi- Arctic is central for current research and part of arctic cally, the comprehensive DNA barcode library allows us monitoring efforts (CAFF Terrestrial Steering Group to answer a series of questions: How well do DNA bar- 2014). While the Arctic region is currently experiencing codes discriminate among morphologically defined taxa rapid climate warming (ACIA 2005), we still know rela- within the target community? What community proper- tively little about its animal and plant communities – ties are resolved by the Malaise trap material? How apart from the fact that they are changing (Settele et al. many new taxa are detected when identifying close to 2014). At least three components of Arctic transition call 20 000 individuals? What does the resulting individual- for accurate descriptions of communities. First, species’ level resolution reveal in terms of community structure? ranges are shifting, and this will likely increase the spe- Which taxa dominate the community, and when? cies richness of the Arctic. However, evaluating this Beyond the current objectives, the DNA barcode library hypothesis is challenging, as changes in species composi- has already allowed us to resolve trophic interactions tion are poorly documented at arctic sites (Killengreen within this high-arctic food web, with insights reported et al. 2007; Post et al. 2009; Wardle et al. 2011; Gilg et al. elsewhere (Wirta et al. 2014, 2015a,b). 2012). Second, drastic phenological changes are currently being observed in the Arctic (Høye et al. 2007, 2013). Materials and methods Nonetheless, current descriptions of such changes mostly rely on observations from higher taxa, with limited Study area understanding of species- and population-level change 0 0 (Høye et al. 2007; Miller-Rushing et al. 2010; Høye et al. Our study area, the Zackenberg Valley (74°28 N/20°34 W) 2013 but see e.g. Post & Forchhammer 2008; Gilg et al. is located within the Northeast Greenland National Park 2009; Mortensen et al. 2014). Third, arctic communities (Meltofte & Rasch 2008). The valley encompasses an area seem to be governed by more complex biotic interactions of approximately 100 km2, surrounded by mountains and

(a) (b) Fig. 1 (a) Map of Greenland with Zacken- berg indicated by a square. (b) Sampling points in the Zackenberg valley shown by black dots, with research zones (1, 1a, 1b, 1 1c; cf. Schmidt et al. 2012) indicated. The two Malaise traps of which full catches were barcoded for 2 years are indicated by triangles. Malaise trap A was operated 1b in years 2012 and 2014, whereas trap B was operated in 2013 and 2014. 1a

N Malaise trap A Malaise trap B 5mk

the ocean (Fig. 1). As the valley is environmentally we operated two Malaise traps, each during two summer diverse, most species known from Northeast Greenland seasons (one trap during summers 2012 and 2014, and one have been detected in the area (Meltofte & Rasch 2008). during 2013 and 2014; Fig. 1). The seasons 2012 and 2014 The area with its flora and fauna has been intensively were both late with respect to snowmelt, while 2013 was studied since the establishment of the Zackenberg unusually early (Jensen et al. 2014; J. Hansen 2015 per- Research Station in 1996 (cf. http://zackenberg.dk/publi- sonal communication). The traps were emptied weekly cations/). The station sustains an intensive monitoring and the specimens processed following the Global Malaise programme aimed at simultaneously recording concur- Trap Program (Appendix S4, Supporting information). rent changes in the physical, biological and feedback Vascular plants were sampled in 2012 by R. Kaarti- properties of a single ecosystem. In short, this area offers nen, G. Varkonyi and T. Roslin using visual search, an excellent context for studying an arctic community and guided by prior locality information. Further vascular changes in it (Forchhammer et al. 2008). plant samples from the region were provided by S. Ertl and C. Lettner. Description of the species composition of the region DNA barcoding To describe the terrestrial fauna and flora of the region, a preliminary list of macroscopic species was established To examine possible intraspecific variation, five individu- based on previous studies of animals (BioBasis Zacken- als of each collected arthropod species and three individ- berg, N. M. Schmidt 2012 personal communication) and uals of each vascular plant species (when available) were vascular plants (Fredskild & Mogensen 1997; Bay 1998; DNA barcoded. For animals, we used the well-estab- BioBasis Zackenberg, N. M. Schmidt 2012 personal com- lished DNA barcode region of cytochrome c oxidase 1 munication). To complement the list, arthropods were (CO1; Hebert et al. 2003). For plants, we used two regions caught with a multitude of trapping methods as described previously shown to amplify, sequence and distinguish by Varkonyi & Roslin (2013) for 2009–2012. In addition, species well: the rbcLa region and the internal transcribed soil arthropods were sampled by five soil cores (5 * 5cm, spacer 2 (ITS2; Chen et al. 2010; Hollingsworth 2011; Hol- 10 cm deep) in 2012 and extracted by the Tullgren funnel lingsworth et al. 2011; Kuzmina et al. 2012). For further method (Tullgren 1918). Sampling covered all habitat details, see Appendix S1 (Supporting information). types in the study area. For a list of taxonomists identify- Tissue sampling, DNA extraction, amplification and ing the samples, see Table 1 and Table S1, Supporting sequencing of both arthropods and vascular plants were information; for other details of sampling and sorting, see implemented in accordance with the standard protocols Appendix S1 (Supporting information). The specimen are of the Canadian Centre for DNA Barcoding (CCDB; all uniquely coded and stored in the University of Hel- Appendix S1, Supporting information). sinki, University of Copenhagen and in private collections Separately, each individual from two Malaise traps (Appendix S1, Supporting information; for specimen data for two seasons was DNA barcoded following the see Data accessibility). standard protocol of the Global Malaise Trap Program To derive ecological samples of local communities (http://globalmalaise.org/; for details see Appendix S4, present at a given site at a given time within the valley, Supporting information).

Table 1 Animal orders and species richness encountered at Zackenberg

Phylum Class Order No. species Identiﬁed by

Arthropoda Arachnida Araneae 10 G. Blagoev, H. Wirta Actinedida 11 P. Gjelstrup Gamasida 14 D. Gwiazdowicz Oribatida 9 P. Gjelstrup Collembola Entomobryomorpha 8 Z. Gavor & E. Jørgensen Neelipleona 1 Z. Gavor & E. Jørgensen Poduromorpha 11 Z. Gavor & E. Jørgensen Symphypleona 3 Z. Gavor & E. Jørgensen Insecta Coleoptera 4 G. Varkonyi Diptera 168 M. Bartak, H. Disney, L. Hulden, J. Ilmonen, J. Jakovlev, M. Jaschhof, J. Kahanpa€a,€ V. Michelsen, S. A. Nielsen, T. Nielsen, L. Paasivirta, J. Pohjoismaki,€ J. Salmela, M. von Tschirnhaus, P. Vilkamaa Hemiptera 8 C. Rasmussen, E. Maw Hymenoptera 64 G. Varkonyi, H. Wirta Lepidoptera 21 T. Roslin, C. Rasmussen, J. Kullberg, Psocodea 2 J. Kanervo Thysanoptera 1 J. Kettunen Trichoptera 1 J. Salokannel Chordata Aves Anseriformes 11 BioBasis Zackenberg; N. M. Schmidt Charadriiformes 34 BioBasis Zackenberg; N. M. Schmidt Falconiformes 2 BioBasis Zackenberg; N. M. Schmidt Galliformes 1 BioBasis Zackenberg; N. M. Schmidt Gaviiformes 2 BioBasis Zackenberg; N. M. Schmidt Passeriformes 8 BioBasis Zackenberg; N. M. Schmidt Procellariiformes 1 BioBasis Zackenberg; N. M. Schmidt Strigiformes 1 BioBasis Zackenberg; N. M. Schmidt Mammalia Artiodactyla 1 BioBasis Zackenberg; N. M. Schmidt Carnivora 4 BioBasis Zackenberg; N. M. Schmidt Lagomorpha 1 BioBasis Zackenberg; N. M. Schmidt Rodentia 1 BioBasis Zackenberg; N. M. Schmidt

Column ‘Identiﬁed by’ offers the taxonomists primarily responsible for identifying each taxon. For a full species list, see Table S1 in Supporting information.

between morphologically defined species and Barcode DNA barcode libraries Index Numbers (BINs; Ratnasingham & Hebert 2013). To compile a DNA barcode library of morphologically For this purpose, we used the data set ZackAnim (Zack- defined taxa, all species for which specimens were avail- enberg animals, also public DNA barcodes0; 64 verte- able from Zackenberg were sampled, identified and pro- brate and 297 arthropod species). Here we only cessed to obtain DNA barcodes. Further DNA barcodes considered specimens assigned to both species and BINs, emanating from a study by Rasmussen et al. (2013), as thus omitting specimens assigned to genera (i.e. Hyme- also conducted at Zackenberg, were likewise included. noptera: Atractodes, Syrphoctonus and Cotesia, each For details on the contents of libraries created and here encompassing multiple species) and species for which published, see Appendix S2 (Supporting information). only a short DNA barcode was available (which were For purposes of completeness, we note that even speci- not assigned to a BIN). A BIN (or multiple BINs) mens identified to only family or genus, as well as taxa detected uniquely within a single morphologically lacking any DNA barcodes, are also included in the data defined taxon allows each individual assigned to this set. BIN to be reliably assigned to the morphologically defined taxon in question. In addition, in some cases where the same BIN was shared among two or more Discrimination of species by DNA barcodes morphologically defined taxa, analysis revealed species- To verify the ability of DNA barcodes to distinguish specific monophyletic clusters within this BIN – thus between animal species, we examined the match allowing successful discrimination of species (see

Appendix S3, Supporting information for details). For and years. For the analysis of phenology, we chose the the latter purpose, we constructed a neighbour joining year with the longest trapping season, 2012, to best rep- tree in BOLD (using the BOLD aligner and the Kimura resent the full phenology of the species. 2-parameter distance model while including sequences >50 bp and excluding potentiallly contaminated or Results misidentified specimens; Ratnasingham & Hebert 2007). As the BIN system is not applicable to plants, we used DNA barcode coverage a separate approach to estimate how well the two plant barcode regions distinguished among plant species. We Overall, 403 terrestrial animal species are now known constructed a neighbour joining tree in BOLD (with the from Zackenberg (Table S1, Supporting information), same settings as above but using MUSCLE to align the including 67 vertebrates (60 birds and 7 mammals) and sequences; Ratnasingham & Hebert 2007) for the two 336 arthropods. Of the latter, 269 are insects (Insecta), 23 regions ITS2 and rbcLa separately. When morphologi- springtails (Collembola), ten spiders (Arachnida: Ara- cally defined species or subspecies formed monophyletic neae) and 34 mites (Arachnida: Acari; Table 1 and clusters, we judged that the barcode region could cor- Table S1, Supporting information). The number of mite rectly assign the specimens to the taxon. For these analy- species is likely to increase based on preliminary data ses, we used the data set ZackPlan (Zackenberg vascular (BioBasis Zackenberg, N. M. Schmidt 2015 personal com- plants0; 137 species and subspecies). munication). To examine how well the two plant barcoding regions Sequencing success for animals was relatively high, separate between plant species, we first concatenated the with 81% of 313 species (84% of 2376 samples) sampled alignments of ITS2 and rbcLa with SequenceMatrix (Vai- at Zackenberg yielding a DNA barcode. The resultant dya et al. 2011). For this analysis, all specimens with an DNA barcode library (supplemented with publicly avail- rbcLa barcode were used (total n = 369 specimens), with able sequences; cf. Table S1, Appendix S2, Supporting a few specimens lacking an ITS2 barcode (Appendix S2, information) covers 86% of all animal species known Supporting information, Data set ZackPlan). Again, we from Zackenberg (although only short sequences are constructed a neighbour joining tree (in MEGA 6, using available for six species: four with 400–600 bp, one with the Kimura 2-parameter model and bootstrapping the 180 bp and one with 120 bp). For Collembola and Arach- phylogeny 1000 times, with all other settings kept at their nida: Acari, barcode coverage is poor (48% and 41% of default values; Tamura et al. 2013), and assessed whether taxa, respectively), while for insects, coverage is high specimens of each morphologically defined taxon (91%; Table S1, Supporting information). formed a monophyletic clade. In terms of vascular plants, 160 species and subspecies representing three phyla and 25 families have been detected at Zackenberg (Table 2 and Table S2, Sup- Applying barcodes to community samples from Malaise porting information). For the species with specimens traps available for the current study, sequencing success was Catches from the two Malaise traps were identified by very high for both markers, with 99% of samples and the identification engine of BOLD (Ratnasingham & 100% of taxa yielding an rbcLa barcode, and 89% of sam- Hebert 2007; details of the community samples from ples and 96% of taxa yielding ITS2 barcodes. As a result, Malaise catches in Appendix S4, Supporting informa- rbcLa barcodes are now available for 137 (86% of 160) tion). To estimate the turnover in community composi- taxa, and ITS2 barcodes for 132 (82% of 160) taxa. tion in space and time, we compared species richness and composition between years within sites and between Species discrimination by DNA barcodes sites within years, describing differences in both species composition and in the relative abundance of species Most animal species from Zackenberg are readily distin- (Appendix S5, Supporting information). We also guishable by DNA barcodes: 92% of the species pos- described the phenology of each BIN with ten or more sessed unique BINs and 93% formed monophyletic representatives (considering a maximum of ten BINs per clusters (Fig. 2). We found some evidence for cryptic family; Appendix S5, Supporting information). diversity within morphologically identified taxa, with 18 For these analyses, we focused on insects alone, cases of more than one BIN being detected within a mor- excluding arachnids (as less comprehensively sampled phologically defined species (cf. Table S1, Appendix S3, by Malaise traps). To avoid any redundant species turn- Fig. S1, Supporting information). Higher taxonomic units over due to different time periods being sampled in dif- were well discriminated in all classes, with 200 (90%) of ferent years, we used pooled catches from the same 223 genera forming monophyletic clades (Appendix S2, 6 weeks from each year for the analyses comparing sites Fig. S1, Supporting information).

Table 2 Vascular plant orders and species richness encountered at Zackenberg

Phylum Class Order Species Identiﬁers

Lycopodiophyta Lycopodiopsida Lycopodiales 1 G. Varkonyi, H. Vare€ Magnoliophyta Liliopsida Alismatales 2 R. Kaartinen, G. Varkonyi, H. Vare€ Poales 59 C. Lettner, S. Ertl, R. Kaartinen, G. Varkonyi, H. Vare,€ BioBasis Zackenberg; N. M. Schmidt Magnoliopsida Asterales 8 R. Kaartinen, G. Varkonyi, H. Vare€ Boraginales 1 BioBasis Zackenberg; N. M. Schmidt Brassicales 17 S. Ertl, R. Kaartinen, G. Varkonyi, H. Vare,€ BioBasis Zackenberg; N. M. Schmidt Caryophyllales 22 S. Ertl, R. Kaartinen, G. Varkonyi, H. Vare,€ BioBasis Zackenberg; N. M. Schmidt Ericales 7 R. Kaartinen, G. Varkonyi, H. Vare€ Fagales 1 R. Kaartinen, H. Vare€ Lamiales 5 R. Kaartinen, H. Vare,€ BioBasis Zackenberg; N. M. Schmidt Malpighiales 2 G. Varkonyi, H. Vare€ Myrtales 2 R. Kaartinen, G. Varkonyi, H. Vare€ Ranunculales 7 R. Kaartinen, G. Varkonyi, H. Vare€ Rosales 9 H. Vare,€ BioBasis Zackenberg; N. M. Schmidt Saxifragales 13 S. Ertl, G. Varkonyi, H. Vare,€ BioBasis Zackenberg; N. M. Schmidt Pteridophyta Pteridopsida Equisetales 2 G. Varkonyi, H. Vare€ Polypodiales 2 G. Varkonyi, H. Vare€

Column ‘Identiﬁers’ offers the taxonomists primarily responsible for identifying each taxon. For a full species list, see Table S2 in Supporting information.

(a)300 (b) 100

50 80 200 60 150

40 100 Number of taxa

50 20

0 0 Arachnida: Collembola Insecta Aves Mammalia Lycopodiopsida Magnoliopsida Acari Araneae Liliopsida Pteridopsida

Fig. 2 Coverage and discriminatory power of DNA barcodes for the terrestrial (a) fauna and (b) vascular plant flora of Zackenberg, with bars representing classes. For animals, we show the number of taxa for which each molecularly defined BIN can be attributed to a single morphologically defined taxon (black fraction); for which the same BIN is shared among multiple taxa (grey fraction), and for which a DNA barcode is missing (white fraction). For vascular plants, we combine information from barcoding regions rbcLa and ITS2, showing the number of taxa for which combined information will yield monophyletic clades attributable to a single morphologically defined taxon (black fraction); for which each such molecularly defined clade splits into several morphologically defined taxa (grey fraction), and for which the DNA barcodes are missing (white fraction). For animals, 272 species have DNA barcodes for specimens collected in Zackenberg, 2 for specimens collected from other parts of Greenland, 4 for specimens from the full Malaise trap barcoding and 78 species with barcodes for specimens sampled in northern regions and extracted from BOLD (with IDs for the latter offered in Table S1, Supporting information), while the DNA barcodes for vascular plants are all from specimens collected in Zackenberg.

Regarding vascular plants, the DNA barcodes offered 132 taxa with ITS2 sequences (Fig. 2, Appendix S3, Figs weaker resolution among species. Of the 137 subspecies S2 and S3, Supporting information). At the level of gen- and species with rbcLa sequence, 61 taxa (45%) formed era, resolution was slightly higher, with 49 (78%) of 63 monophyletic clusters, as compared to 69 (52%) for the genera examined forming monophyletic clades based on

© 2015 John Wiley & Sons Ltd A HIGH-ARCTIC COMMUNITY DNA BARCODED 815 rbcLa, and 42 (69%) of 61 taxa assessed showing mono- catches, but the other half was only detected by one phyly of ITS2 clades. method – with the specific fraction varying with the class Combining information from both gene regions (Fig. 3). As expected, flying insects formed the most abun- added some resolution, with 74 (54%) of 137 species and dant class among the specimens found in the Malaise subspecies then forming monophyletic clusters. At the traps, with Diptera being the most abundant order, and level of genera, no further resolution was added beyond Chironomidae the most abundant family (Appendix S4; that offered by rbcLa, with 49 (78%) of 63 genera assessed Fig. S5, Supporting information). Overall, the arctic insect forming monophyletic clades for the regions combined community was strongly dominated by a very few hyper- (Appendix S3, Fig. S4, Supporting information). abundant taxa (Fig. 4; Appendix S4; Fig. S5, Supporting In terms of sequence divergence within vs. between information). Only five BINs were represented by more species, the CO1 region used for animals revealed sub- than ten individuals per trap day during at least one sea- stantially higher sequence divergence between than son – but of these, the two most abundant ones accounted within species, analysed through Barcode Gap Analysis for 5230 and 2256 individuals (Appendix S4; Table S7, in BOLD (using the BOLD aligner (ZackAnim)/MUSCLE Supporting information). Of abundant BINs (as repre- (ZackPlan) and the Kimura 2-parameter distance model sented by over 100 individuals in the Malaise traps over- while including sequences >50 bp and excluding poten- all), 22 of 27 were also detected by traditional sampling tially contaminated or misidentified specimens; Ratnas- (Appendix S4; Table S7, Supporting information). Sixteen ingham & Hebert 2007). In 93.1% (257 of the 276 species abundant BINs were midges in the family Chironomidae, with multiple specimens with a CO1 DNA barcode) of showing their dominance in the arctic fauna. the animal species the maximum intraspecific divergence Using the individually sequenced specimens from was smaller than the distance to the nearest neighbour, Malaise traps, we found substantial species turnover in showing a barcode gap (Hebert et al. 2004b; Meyer & both space and time. Focusing on the 1 year (2014) in Paulay 2005). For the majority of vascular plant species, which both Trap A and B were operated, we detected a the difference in maximum sequence divergence within difference of more than half of local species over a dis- species vs. the distance to nearest neighbours remained tance of <1 km (Fig. 1; Appendix S5, Supporting infor- vague in both rbcLa and ITS2. For rbcLa, a barcode gap mation). Examination of patterns in time similarly was found in 51.2% (62 of the 121 species with multiple revealed large turnover. Year 2013, which was character- specimens with an rbcLa DNA barcode) of nearest neigh- ized by an early spring and unusually little snow, was bour pairs and in 48.7% (56 of the 115 species with multi- different from the years of 2012 and 2014 which shared a ple specimens with ITS2 DNA barcode) for ITS2. late snow melt (Appendix S5, Supporting information). In terms of species’ phenology in 2012, species within families showed extreme variation, with, for example, Applying DNA barcodes to resolving the structure of an some midges showing a median flight period almost a arctic community in space and time month before others – within an Arctic summer span- A total of 21 512 arthropods were individually sorted ning <2 months (Fig. 5). and processed from the two Malaise traps combined (with 8810 individuals from Trap A in 2012 and 4769 in Discussion 2014 vs. 3283 individuals from Trap B in 2013 and 4 650 in 2014). Of these individuals, most (20 117 individuals, This study reports substantial progress towards the 94%) were successfully sequenced and assigned to a BIN assembly of a comprehensive library of DNA barcodes (19 896 individuals, i.e. 92% of all specimens and 99% of for an intensively studied region in the High Arctic. To those yielding a sequence). In total, we detected 303 our knowledge, it offers the highest coverage of the local BINs, of which 122 had not been previously detected by species complement achieved for any study site to date. other sampling techniques (Fig. 3). Targeting terrestrial animals and vascular plants, we Combining the diversity revealed by the Malaise traps show that the standard barcoding markers deliver good with that detected by previous sampling yielded a total of species-level resolution for animals, but less so for 401 arthropod BINs at Zackenberg. Of these, 30 were plants. Applying this tool to community-wide samples Arachnida:Acari, 9 Arachnida:Araneae, 16 Collembola of arthropods from Malaise traps, we found substantial and 346 Insecta, confirming the preponderance of insects diversity beyond that revealed by traditional sampling in the high arctic fauna. That so many new taxa (122 BINs) approaches and morphological identification. Identifying were detected by Malaise sampling reveals the difficulty full trap catches with the aid of DNA barcodes also in comprehensive sampling the fauna in even a low- revealed new patterns in the ecology of the lesser known diversity area. Approximately half of all BINs were parts of arctic biodiversity. In the course of the Discus- detected by both previous sampling and in the Malaise sion, we address each of these points in more detail.

(a) (b) 350 120

300 100

250 80

200

150 Number of taxa

40 100

20 50

0 0 1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 22 23 25 27 29 31 33 34 36 38 40 42 44 45 46 11 13 15 17 19 21 24 26 28 30 32 35 37 39 41 43 Acari Coleoptera Diptera Insecta Hemiptera Hymenoptera Lepidoptera Thysanoptera Araneae Trichoptera Arachnida: Arachnida: Collembola

Fig. 3 The relative dominance of different arctic arthropod taxa, and the coverage of our faunal inventory, as revealed by overlap in species detected by different sampling techniques. Shown is the number of taxa (here represented by BINs) for (a) arthropod classes, and (b) insect families only detected by traditional sampling and morphological identiﬁcation (white fraction), by both methods (grey fraction) and only by individually sequenced specimens from Malaise traps (black fraction). Note that only individuals successfully sequenced and assigned to BINs could be included, thus omitting a small fraction of the material. The insect families in B) are 1. Coc- cinellidae, 2. Agromyzidae, 3. Anthomyiidae, 4. Calliphoridae, 5. Cecidomyiidae, 6. Ceratopogonidae, 7. Chironomidae, 8. Culicidae, 9. Empididae, 10. Heleomyzidae, 11. Limoniidae, 12. Muscidae, 13. Mycetophilidae, 14. Pallopteridae, 15. Phoridae, 16. Piophilidae, 17. Scathophagidae, 18. Sciaridae, 19. Syrphidae, 20. Tachinidae, 21. Tipulidae, 22. Trichoceridae, 23. Aphididae, 24. Lygaeidae, 25. Apidae, 26. Braconidae, 27. Encyrtidae, 28. Eulophidae, 29. Figitidae, 30. Ichneumonidae, 31. Megaspilidae, 32. Pteromalidae, 33. Tenthredinidae, 34. Crambidae, 35. Erebidae, 36. Geometridae, 37. Lycaenidae, 38. Noctuidae, 39. Nymphalidae, 40. Pieridae, 41. Plutellidae, 42. Pterophoridae, 43. Pyralidae, 44. Tortricidae, 45. Thripidae and 46. Apataniidae.

barcodes to shorter but still-diagnostic regions for speci- How well do DNA barcodes resolve arctic species? fic applications, for example for species-level identifica- The low species diversity of high-latitude ecosystems tion of degraded gut contents (cf. Wirta et al. 2014, 2015a, offers a rare opportunity to assemble a comprehensive b). barcode library for an entire community with realistic Some morphologically distinct bird, fly and wasp spe- effort. Among the animal species detected, DNA bar- cies share BINs (Table S1, Fig. S1, Supporting informa- codes provided strong resolution in separating morpho- tion). Most of these are likely valid species, but logically distinguishable taxa. Sequence divergence apparently closely related and oftentimes sharing a within taxa was generally lower than that among species monophyletic BIN. In addition, some morphologically (Appendix S3, Supporting information). Approximately similar mite, collembolan, wasp and fly taxa splitting 93% of the species in this high-arctic community could into more than one BIN will likely be complexes of cryp- be assigned to a particular species based on barcode data tic species, and are now being taxonomically revised (P. – thereby matching previous findings from barcode Gjelstrup, G. Varkonyi, S. A. Nielsen and L. Paasivirta). libraries of specific taxa (e.g. Hausmann et al. 2011; Compared to the resolution provided by CO1, the Renaud et al. 2012a; Webb et al. 2012; Moriniere et al. two gene regions chosen for vascular plants, ITS2 and 2014; Raupach et al. 2014). Importantly, the low species rbcLa, delivered less resolution both when applied indi- richness of the area and the limited diversity of, for vidually and in combination. In both cases, these regions example, congeneric taxa, will allow the trimming of full distinguished about half of the species, and two-thirds of

30 2012; Stahlhut et al. 2013). This general composition of communities suggests that any profound representation 25 of arctic change will call for new attention to its smallest 20 inhabitants – a focus which certainly has not been real- ized in recent research (cf. Post et al. 2009). 15 The species list associated with the current, compre-

10 hensive DNA barcode library provides a strong baseline –

Individuals/trap day for gauging future changes in species composition that 5 is for monitoring a key dimension of community change Thysanoptera Coleoptera (Parmesan & Yohe 2003; Wardle et al. 2011; Gilg et al. 0 2012). Importantly, DNA barcodes will allow the detec- Fig. 4 Rank-abundance plot of the insect community sampled tion of new invaders in whatever form they come (cf. by Malaise traps. Shown is the rank abundance of individual Gibson et al. 2014). While a new vertebrate species enter- BINs per order in the pooled material from both traps (cf. Fig. 1) ing a community may be easily recognizable, most inver- over the 3 years. BINs from the order Hymenoptera are shown tebrates are much less so – and/or may first turn up in in black, Hemiptera in dark grey, Diptera in grey, Lepidoptera the form of a prey already ingested by a predator, or as a in light grey, Coleoptera in very light grey and Thysanoptera in larva still within the tissues of its host. For example, in white. Note that to make the colours representing orders more visible, the x-axis crosses at y = 1. Coleoptera and Thysanop- previous work, we detected parasitoids new to the area tera have been separately marked for clarity. from sequences of parasitoid larvae within their hosts (Wirta et al. 2014). In the current study, we detected a migratory moth new to the region through DNA barcod- the plant genera. This low discriminatory power corre- ing of Malaise catches from 2014 (the diamond-back sponds to that reported for another locality-wide library moth Plutella xylostella, BOLD: AAA1513; Appendix S4, of arctic ITS2 and rbcLa barcodes (Kuzmina et al. 2012), Supporting information; Data set ZackGM). In terms of but is lower than that found in studies using matK (Le guild-wide change, a study by Fernandez-Triana et al. Clerc-Blain et al. 2010; Burgess et al. 2011; de Vere et al. (2011) reported substantial turnover of parasitoid wasp 2012). Overall, the low resolution provided by ITS2 and species at a subarctic site repeatedly sampled through rbcLa conflicts with other studies identifying ITS2 as a time, whereas two other studies detected no clear promising region for species-level identification (Chen changes in guilds of either parasitoid wasps or Diptera et al. 2010; Hollingsworth 2011; Hollingsworth et al. (Renaud et al. 2012b; Timms et al. 2013). Here, new tools 2011; Pang et al. 2011, 2013). Our study exemplifies the like the current DNA barcode library offer the potential trade-off between high sequencing success (for ITS2 and to demonstrate how quick and widespread rates of rbcLa) vs. relatively low discriminatory power (Hollings- change may be when assessed not for single guilds or worth et al. 2011). However, our results do document taxa, but across entire communities. just what taxa ITS2 and rbcLa can distinguish in our tar- It is evident that the current species list for Zacken- get flora (Table S2, Supporting information) – thereby berg is still not exhaustive. As total species richness is offering researchers the a priori information needed to based on both common and rare species, no sampling judge which questions they may address with DNA bar- effort is ever guaranteed to detect all species (e.g. Chao code analysis. 2005). Thus, our sequencing of some 20 000 individuals from two Malaise traps revealed many species over- looked by previous sampling, and added substantially to How diverse is the target community? the coverage of our resultant barcode library. The total number of animal and vascular plant species is Exactly how many of the additional BINs detected by certainly low in our high-arctic target community. With molecular methods are ‘biologically valid species’ is per- about 400 terrestrial BINs, the species richness at Zacken- haps a moot point. Molecularly based biodiversity inven- berg Valley approximates 1% of the estimated species tories will usually suggest higher species and/or BIN richness of a northern temperate country like Finland diversity than counts based on morphological identifica- (with 45 000 estimated species; Rassi et al. 2010). Within tion. Such discrepancies may arise both due to the reso- morphologically identified taxa, DNA barcodes revealed lution of true but cryptic species, and to the successful few indications of cryptic diversity at Zackenberg. Sum- identification of specimens lacking useful morphological ming across taxa detected, the species richness of arthro- characters (e.g. of females in taxa where all characters pods was far greater than that of all other animal groups, relate to male genitals, of juveniles in the many taxa for supporting patterns from other latitudes (Pimentel & which all characters relate to adult morphology and of Andow 1984; Hodkinson & Coulson 2004; Basset et al. specimens in generally poor condition; Smith et al. 2009;

(a) 228

218

208

198

Day of year 188

178

168 1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 21 23 25 27 28 29 30 32 34 35 11 13 15 17 19 22 24 26 31 33 Agromyzidae Ceratopogonidae Culicidae Muscidae Phoridae Sciaridae Tipulidae Anthomyidae Chironomidae Empididae Mycetophilidae Tachinidae Scatophagidae

(b) 228

218

208

198

Day of year 188

178

168 36 38 40 42 43 37 39 41 Braconidae Ichneumonidae Pteromalidae

Fig. 5 Phenology of individual taxa within (a) Diptera and (b) Hymenoptera as detected in Malaise trap A (see Fig. 1). Shown is the median (horizontal line), quartiles (box) and range (vertical lines) of capture days for BINs represented by ten or more individuals in 2012. Within families, BINs are ordered from early to late taxa. Include are the most abundant species, each of which was represented by at least ten specimens in the pooled trap catch. Taxa in A) are 1. BOLD:ABW5539 Phytomyza aquilonia, 2.BOLD:ACA8845 Chromato- myia puccinelliae, 3. BOLD:AAG1723 Zaphne frontata, 4. BOLD:ACP6173 Pegomya icterica, 5. BOLD:ABW3845 Brachypogon sp., 6. BOLD: AAG6532 Brachypogon sp., 7. BOLD:ACR7182 Forcipomyia sp., 8. BOLD:AAM6201 Culicoides sp., 9. BOLD:AAE8704 Smittia extrema, 10. BOLD:ABA7011 Smittia sp., 11. BOLD:AAC4201 Paraphaenocladius impensus, 12. BOLD:AAL9235 Limnophyes pumilio, 13. BOLD: AAU3704 Limnophyes ninae, 14. BOLD:ACM4349 Limnophyes brachytomus, 15. BOLD:AAB7912 Limnophyes brachytomus, 16. BOLD: AAI3491 Orthocladius gelidorum, 17. BOLD:AAB1171 Orthocladius rivicola, 18. BOLD:AAL1593 Allocladius nanseni, 19. BOLD:ACA4801 Orthocladius roussellae, 20. BOLD:AAA3750 Aedes nigripes, 21. BOLD:AAW0121 Rhamphomyia filicauda, 22. BOLD:AAF9804 Rhamphomyia nigrita, 23. BOLD:AAW1212 Phaonia bidentata, 24. BOLD:AAL9801 Drymeia groenlandica, 25. BOLD:AAM9109 Spilogona sanctipauli, 26. BOLD:ACA4207 Spilogona monacantha, 27. BOLD:AAM8957 Brevicornu fuscipenne, 28. BOLD: AAZ6184 Megaselia cirriventris, 29. BOLD: AAM7340 Gonarcticus arcticus, 30. BOLD:AAZ6073 Lycoriella riparia, 31. BOLD:AAV1299 Camptochaeta aff. flagellifera, 32. BOLD:AAP8779 Sciaridae sp., 33. BOLD:AAL7869 Lycoriella flavipeda, 34. BOLD:AAZ5252 Peleteria aenea and 35. BOLD:AAM7267 Tipula arctica. Taxa in B) are 36. BOLD:AAE7186 Praon sp., 37. BOLD:AAH7424 Praon brevistigma, 38. BOLD:AAY9781 Cryptus arcticus, 39. BOLD:AAH1503 Pimpla sodalis, 40. BOLD:AAN7603 Acrolyta glacialis, 41. BOLD:ABA0 403 Campoletis horstmanni, 42. BOLD:ABY5384 Neurateles sp. 1ZERO and 43. BOLD:ACJ0792 Pachyneuron sp.

Valentini et al. 2009; Stahlhut et al. 2013). In terms of the Malaise samples relies on first sequencing the individual, current study, the identification of specimens lacking then attributing it to a species see Appendix S4, Support- useful morphological characters seems a likely explana- ing information). In addition, the higher number of BINs tion for many of the new taxa from Malaise traps. We than morphologically defined species may be due to bio- expect that many of the ‘new’ BINs in Malaise catches logical species including more than one BIN. Overall, for are present in previous samples (Appendix S1, Support- our low-diversity target site, the actual fraction of species ing information), but were not assigned to a species by ‘exposed’ by individually sequencing Malaise catches taxonomists and were hence not DNA barcoded (note will most likely not be as high as suggested by Fig. 3 – that the barcoding of voucher specimens relies on first and will differ from reports from more-diverse faunas obtaining an identified individual, then subjecting it to where DNA barcoding has revealed a wealth of cryptic sequencing). By contrast, the protocol for processing species (e.g. Hebert et al. 2004a; Lees et al. 2014).

For one dipteran family, the pattern revealed did several expert taxonomists in identifying massive num- deviate from the rest. Nearly two-thirds of the BINs that bers of specimens of their target taxa over one or several were only detected in Malaise catches were species of seasons. Using our barcode library, we can now identify Chironomidae. As this family accounts for most of the taxa and assign them to ‘early’ vs. ‘late’ groups, allowing arthropod species (Fig. 3), and as chironomids are also us to test topical theory (Høye et al. 2013) on how species the most abundant insects of the High Arctic (this study; with different phenologies respond to climate change. Høye et al. 2013; Jensen et al. 2013), the pattern exposes Second, the application of our library provides one of the the need for vigorous research on chironomids. While first well-resolved descriptions of community structure midges may be less charismatic than arctic mammals, of insects in the Arctic. Compared to communities at midge diversity apparently outnumbers the latter by a lower latitudes, our target assemblage proves not only factor of at least 10:1 (Fig. 2) and more likely 20:1 (Fig. 3), species-poor, but also strongly dominated by very few thereby even exceeding tropical Diptera:Mammals ratios hyperdiverse taxa. This suggests that a large proportion (Basset et al. 2012). Just how midge BINs map onto mor- of total arctic biomass depends on a few populations, phologically defined species is still being studied. In gen- making them the key species for both feeding other taxa eral, DNA barcodes have been shown to be effective and perhaps for sustaining key processes like parasitism tools in identifying and detecting chironomid species and pollination (Høye et al. 2013). Third, the comparison (Pfenninger et al. 2007; Ekrem et al. 2010), although of community structure in space and time suggests that divergence in CO1 within and between species may the arctic tundra offers a more heterogeneous environ- overlap in some genera (Ekrem et al. 2007). ment than oftentimes thought. Species turnover proved What will allow us to dig deeper into the match large among sites with similar habitats separated by dis- between molecularly guided BINs and morphologically tances of less than a kilometre, with more than half of guided species is the ongoing, fruitful collaboration the species present at one site lacking at the other. Turn- between ecologists and taxonomists in generating bar- over in community structure was larger among a late code libraries. Vouchers of new taxa revealed by the and an early summer only 1 year apart than among two Malaise trap material have been and are still being similarly late years separated by 2 years. Although the returned to taxonomists for morphology-based species sampling is unreplicated, both patterns suggest that arc- identification, thus linking molecularly defined taxo- tic community structure is sensitive to local abiotic con- nomic units, here BINs, to previously described species ditions. Finally, having both names (be they BINs or with more or less known biology. Indeed, this continu- scientific binomials) and the tools to assign samples to ing feedback between taxonomists and ecologists is crit- them is really what counts in describing and studying a ical for making barcode libraries informative and community. DNA barcoding provides a good tool to col- useful. late new ecological information to known biological enti- ties, thereby developing the joint ‘Wikipedia’ of biodiversity. It is to this global initiative that we want to What’s new from barcodes? Advancing arctic research contribute with our barcode library from the High Arctic. by resolving patterns at the population level Until now, it has been difficult to address long-term Acknowledgements changes in abundance for the vast majority of arctic species, because long-term species abundance data have not We thank Juhani Kanervo, Jaakko Kullberg, Jukka Kettunen, been available. This is particularly true for mega-diverse Eric Maw and Juha Salokannel for their contribution to identi- groups like terrestrial arthropods, where seminal fication of specimens from Zackenberg, Lars Vilhelmsen for offering us the Greenland collection at the Zoological Museum research has – per necessity – been based on rough taxo- of the University of Copenhagen, and Bess Hardwick and the nomic groups (cf. Høye et al. 2007; Post et al. 2009; Høye Zackenberg station for excellent logistics. Funding by INTER- et al. 2013; but see Høye et al. 2014; Bowden et al. 2015). ACT (projects QUANTIC and INTERPRED) under the Euro- Here, DNA barcodes offer good tools for studying these pean Community’s Seventh Framework Programme (to H. W. groups with high resolution. Overall, our successful and T. R.), by the University of Helsinki (grant number 788/ identification of two Malaise trap catches over two sum- 51/2010 to T. R.), by the Academy of Finland (grant number mers reveals some of the horizons opened up by this 1276909 to T. R.), by the Kone foundation (to H. W.), by Carls- bergfondet (to C. R.) and by Aage V. Jensen Charity Founda- approach. First, it allows us to describe the phenology of tion (to N. M. S.) is gratefully acknowledged. A grant from the species. While previous work has suggested that species government of Canada through Genome Canada and the with a late vs. early occurrence will respond differently Ontario Genomics Institute supported the sequence analyses. to climate change (Høye et al. 2007, 2014), this hypothesis The studies at Zackenberg are one component of the Global has never been well tested – as establishing the phenol- Malaise Trap Program within the International Barcode of Life ogy of diverse taxa requires the concerted efforts of Project.

References National Academy of Sciences of the United States of America, 101, 14812– 14817. ACIA (2005) Arctic Climate Impact Assessment. Cambridge University Hebert PDN, Stoeckle MY, Zemlak TS, Francis CM (2004b) Identification Press, Cambridge, UK. of birds through DNA barcodes. PLoS Biology, 2, 1657–1663. Basset Y, Cizek L, Cuenoud P et al. (2012) Arthropod diversity in a tropi- Hodkinson ID, Coulson SJ (2004) Are high Arctic terrestrial food chains cal forest. Science, 338, 1481–1484. really that simple? The Bear Island food web revisited. Oikos, 106, 427– Bay C (1998) Vegetation Mapping of Zackenberg Valley, Northeast Greenland. 431. Danish Polar Center & Botanical Museum, University of Copenhagen, Hollingsworth PM (2011) Refining the DNA barcode for land plants. Pro- Copenhagen. ceedings of the National Academy of Sciences of the United States of America, Bowden JJ, Hansen RR, Olsen K, Hoye TT (2015) Habitat-specific effects 108, 19451–19452. of climate change on a low-mobility Arctic spider species. Polar Biol- Hollingsworth PM, Graham SW, Little DP (2011) Choosing and using a ogy, 38, 559–568. plant DNA barcode. PLoS ONE, 6, e19254. Burgess KS, Fazekas AJ, Kesanakurti PR et al. (2011) Discriminating plant Høye TT, Post E, Meltofte H, Schmidt NM, Forchhammer MC (2007) species in a local temperate flora using the rbcL plus matK DNA bar- Rapid advancement of spring in the High Arctic. Current Biology, 17, code. Methods in Ecology and Evolution, 2, 333–340. R449–R451. Chao A (2005) Species estimation and applications. In: Encyclopedia of Sta- Høye TT, Post E, Schmidt NM, Trojelsgaard K, Forchhammer MC (2013) tistical Sciences (eds Kotz S, Balakrishnan N, Read CB & Vidakovic B), Shorter flowering seasons and declining abundance of flower visitors pp. 7907–7916. Wiley, New York. in a warmer Arctic. Nature Climate Change, 3, 759–763. Chen SL, Yao H, Han JP et al. (2010) Validation of the ITS2 region as a Høye TT, Eskildsen A, Hansen RR et al. (2014) Phenology of high-arctic novel DNA barcode for identifying medicinal plant species. PLoS butterflies and their floral resources: species-specific responses to cli- ONE, 5, e8613. mate change. Current Zoology, 60, 243–251. Clare EL (2014) Molecular detection of trophic interactions: emerging Jensen LM, Rasch M, Schmidt NM (2013) Zackenberg Ecological trends, distinct advantages, significant considerations and conserva- Research Operations, 18th Annual Report 2012. Zackenberg Ecological tion applications. Evolutionary Applications, 7, 1144–1157. Research Operations, p. 122. Aarhus University, DCE – Danish Centre Ekrem T, Willassen E, Stur E (2007) A comprehensive DNA sequence for Environment and Energy, Aarhus University, Denmark. library is essential for identification with DNA barcodes. Molecular Jensen LM, Christensen TR, Schmidt NM (2014) Zackenberg Ecological Phylogenetics and Evolution, 43, 530–542. Research Operations, 19th Annual Report 2013. DCE - Danish Centre for Ekrem T, Stur E, Hebert PDN (2010) Females do count: documenting Chi- Environment and Energy, Aarhus University, Roskilde, Denmark. ronomidae (Diptera) species diversity using DNA barcoding. Organ- Kaartinen R, Stone GN, Hearn J, Lohse K, Roslin T (2010) Revealing isms Diversity & Evolution, 10, 397–408. secret liaisons: DNA barcoding changes our understanding of food Fernandez-Triana J, Smith MA, Boudreault C et al. (2011) A poorly webs. Ecological Entomology, 35, 623–638. known high-latitude parasitoid wasp community: unexpected diver- Killengreen ST, Ims RA, Yoccoz NG et al. (2007) Structural characteristics sity and dramatic changes through time. PLoS ONE, 6, e23719. of a low Arctic tundra ecosystem and the retreat of the Arctic fox. Bio- Forchhammer MC, Christensen TR, Hansen BU et al. (2008) Zacken- logical Conservation, 135, 459–472. berg in a circumpolar context. Advances in Ecological Research, 40, Kuzmina ML, Johnson KL, Barron HR, Hebert PDN (2012) Identification 499–544. of the vascular plants of Churchill, Manitoba, using a DNA barcode Fredskild B, Mogensen GS (1997) ZERO Line. Final Report 1997, p. 36. library. BMC Ecology, 12, 25. Greenland Botanical Survey and Botanical Museum, University of Le Clerc-Blain J, Starr JR, Bull RD, Saarela JM (2010) A regional approach Copenhagen., Copenhagen, Denmark. to plant DNA barcoding provides high species resolution of sedges Gibson J, Shokralla S, Porter TM et al. (2014) Simultaneous assessment of (Carex and Kobresia, Cyperaceae) in the Canadian Arctic Archipelago. the macrobiome and microbiome in a bulk sample of tropical arthro- Molecular Ecology Resources, 10,69–91. pods through DNA metasystematics. Proceedings of the National Acad- Lees DC, Kawahara AY, Rougerie R et al. (2014) DNA barcoding reveals emy of Sciences of the United States of America, 111, 8007–8012. a largely unknown fauna of Gracillariidae leaf-mining moths in the Gilg O, Sittler B, Hanski I (2009) Climate change and cyclic predator–prey Neotropics. Molecular Ecology Resources, 14, 286–296. population dynamics in the high Arctic. Global Change Biology, 15, Meltofte H, Rasch M (2008) The study area at Zackenberg: high Arctic 2634–2652. ecosystem dynamics in a changing climate: ten years of monitoring Gilg O, Kovacs KM, Aars J et al. (2012) Climate change and the ecology and research at Zackenberg research station, Northeast Greenland. and evolution of Arctic vertebrates. Year in Ecology and Conservation Advances in Ecological Research, 40, 101–110. Biology, 1249, 166–190. Meyer CP, Paulay G (2005) DNA barcoding: error rates based on compre- Group TS (2014) Arctic Terrestrial Biodiversity Monitoring Plan: Imple- hensive sampling. PLoS Biology, 3, 2229–2238. mentation and work plan Akureyri, Iceland, February 25-27, 2014. Miller-Rushing AJ, Hoye TT, Inouye DW, Post E (2010) The effects of CAFF Monitoring Series Report, CAFF International Secretariat, Akur- phenological mismatches on demography. Philosophical Transactions of eyri, Iceland. the Royal Society B-Biological Sciences, 365, 3177–3186. Hajibabaei M, Janzen DH, Burns JM, Hallwachs W, Hebert PDN (2006) Moriniere J, Hendrich L, Hausmann A et al. (2014) Barcoding fauna DNA barcodes distinguish species of tropical Lepidoptera. Proceedings Bavarica: 78% of the Neuropterida fauna barcoded! PLoS ONE, 9, of the National Academy of Sciences of the United States of America, 103, e109719. 968–971. Mortensen LO, Jeppesen E, Schmidt NM et al. (2014) Temporal trends Hausmann A, Haszprunar G, Hebert PDN (2011) DNA barcoding the and variability in a high-arctic ecosystem in Greenland: multidimen- geometrid fauna of Bavaria (Lepidoptera): successes, surprises, and sional analyses of limnic and terrestrial ecosystems. Polar Biology, 37, questions. PLoS ONE, 6, e17134. 1073–1082. Hebert PDN, Ratnasingham S, deWaard JR (2003) Barcoding animal life: Pang X, Song J, Zhu Y et al. (2011) Applying plant DNA barcodes for cytochrome c oxidase subunit 1 divergences among closely related spe- Rosaceae species identification. Cladistics, 27, 165–170. cies. Proceedings of the Royal Society B-Biological Sciences, 270(Suppl. 1), Pang XH, Shi LC, Song JY, Chen XC, Chen SL (2013) Use of the potential S96–S99. DNA barcode ITS2 to identify herbal materials. Journal of Natural Hebert PDN, Penton EH, Burns JM, Janzen DH, Hallwachs W (2004a) Medicines, 67, 571–575. Ten species in one: DNA barcoding reveals cryptic species in the Parmesan C, Yohe G (2003) A globally coherent fingerprint of climate neotropical skipper butterfly Astraptes fulgerator. Proceedings of the change impacts across natural systems. Nature, 421,37–42.

Pfenninger M, Nowak C, Kley C, Steinke D, Streit B (2007) Utility of de Vere N, Rich TCG, Ford CR et al. (2012) DNA barcoding the native DNA taxonomy and barcoding for the inference of larval community flowering plants and conifers of Wales. PLoS ONE, 7, e37945. structure in morphologically cryptic Chironomus (Diptera) species. Wardle DA, Bardgett RD, Callaway RM, Van der Putten WH (2011) Ter- Molecular Ecology, 16, 1957–1968. restrial ecosystem responses to species gains and losses. Science, 332, Pimentel D, Andow DA (1984) Pest-management and pesticide impacts. 1273–1277. Insect Science and Its Application, 5, 141–149. Webb JM, Jacobus LM, Funk DH et al. (2012) A DNA barcode library for Post E, Forchhammer MC (2008) Climate change reduces reproductive North American Ephemeroptera: progress and prospects. PLoS ONE, success of an Arctic herbivore through trophic mismatch. Philosophi- 7, e38063. cal Transactions of the Royal Society B-Biological Sciences, 363, 2369– Wirta H, Hebert PDN, Kaartinen R et al. (2014) Complementary molecu- 2375. lar information changes our perception of food web structure. Proceed- Post E, Forchhammer MC, Bret-Harte MS et al. (2009) Ecological dynam- ings of the National Academy of Sciences of the United States of America, ics across the Arctic associated with recent climate change. Science, 111, 1885–1890. 325, 1355–1358. Wirta H, Weingartner E, Hamback€ P, Roslin T (2015a) Extensive niche Rasmussen C, Dupont YL, Mosbacher JB, Trøjelsgaard K, Olesen JM overlap among the dominant arthropod predators of the High Arctic. (2013) Strong impact of temporal resolution on the structure of an eco- Basic and Applied Ecology, 16,86–92. logical network. PLoS ONE, 8, e81694. Wirta H, Vesterinen E, Hamback€ PA et al. (2015b) Exposing the structure Rassi P, Hyvarinen€ E, Juslen A, Mannerkoski I (2010) The 2010 Red List of of an Arctic food web. Ecology and Evolution, 5, 3842–3856. Finnish Species. Ymparist€ oministeri€ o€ & Suomen ymparist€ okeskus,€ Hel- sinki, Finland, pp. 685. Ratnasingham S, Hebert PDN (2007) BOLD: the barcode of life data sys- H.W. and T.R. designed the research. H.W., G.V., C.R., – tem (www.barcodinglife.org). Molecular Ecology Notes, 7, 355 364. R.K., N.M.S., P.D. N.H., M.B., G.B., H.D., S.E., P.G., Ratnasingham S, Hebert PDN (2013) A DNA-based registry for all animal species: the Barcode Index Number (BIN) system. PLoS ONE, 8, D.J.G., L.H., J.I., J.J., M.J., J.K., T.K., P.H.K., R.L., C.L., e66213. V.M., S.A.N., T.R.N., L.P., S.P., J.P., J.S., P.V., H.V., Raupach MJ, Hendrich L, Kuechler SM et al. (2014) Building-up of a M.v.T. and T.R. performed the research. H.W. and T.R. DNA barcode library for true bugs (Insecta: Hemiptera: Heteroptera) wrote the study with contributions by all authors. of Germany reveals taxonomic uncertainties and surprises. PLoS ONE, 9, e106940. Renaud AK, Savage J, Adamowicz SJ (2012a) DNA barcoding of North- ern Nearctic Muscidae (Diptera) reveals high correspondence between morphological and molecular species limits. BMC Ecology, Data Accessibility 12, 24. Renaud AK, Savage J, Roughley RE (2012b) Muscidae (Diptera) diversity All specimen data and DNA barcodes are available in in Churchill, Canada, between two time periods: evidence for limited BOLD (Ratnasingham & Hebert 2007) as data sets DS- changes since the Canadian Northern Insect Survey. The Canadian ZackAnim (Zackenberg animals; http://dx.doi.org/ – Entomologist, 144,29 51. 10.5883/DS-ZACKANIM), DS-ZackPlan (Zackenberg Schmidt NM, Hansen LH, Hansen J, Berg EB, Meltofte H (2012) BioBasis Manual Conceptual Design and Sampling Procedures of the Biological Moni- vascular plants; http://dx.doi.org/10.5883/DS-ZACK- toring Programme within Zackenberg Basic. Aarhus University, Roskilde, PLAN), DS-GreeAnim (Greenlandic animals outside Denmark, pp. 110. Zackenberg; http://dx.doi.org/10.5883/DS-GREEANIM), Settele J, Scholes R, Betts R et al. (2014) Terrestrial and inland water sys- DS-GreePlan (Greenlandic vascular plants outside tems. In: Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working Group II (eds Zackenberg; http://dx.doi.org/10.5883/DS-GREEPLAN; Field CB, Barros VR & Dokken DJ, et al.), pp. 271–359. Cambridge Appendix S2, Supporting information) and DS-ZackGM University Press, Cambridge, United Kingdom and New York, NY, (Zackenberg Global Malaise Trap catches; http://dx.- USA. Smith MA, Fernandez-Triana J, Roughley RE, Hebert PDN (2009) DNA doi.org/10.5883/DS-ZACKGM; Appendix S4, Supporting barcode accumulation curves for understudied taxa and areas. Molecu- information). The alignments of the data sets are available lar Ecology Resources, 9, 208–216. in BOLD, through the permanent data sets and the Align- Stahlhut JK, Fernandez-Triana J, Adamowicz SJ et al. (2013) DNA barcod- ment Browser. The DNA barcode sequences created for ing reveals diversity of Hymenoptera and the dominance of parasitoids in a sub-arctic environment. BMC Ecology, 13,2. morphologically identified specimens are also available in Tamura K, Stecher G, Peterson D, Filipski A, Kumar S (2013) MEGA6: GenBank (Accession numbers KT959567-KT960035, molecular evolutionary genetics analysis version 6.10. Molecular Biol- KT960048-KT960368, KT960378-KT960746, KU373125-KU – ogy and Evolution, 30, 2725 2729. 374802). Raw data of intraspecific variation and distance Timms LL, Bennett AMR, Buddle CM, Wheeler TA (2013) Assessing five decades of change in a high Arctic parasitoid community. Ecography, to nearest neighbour from data sets DS-ZackAnim and 36, 001–009. DS-ZackPlan are available in Dryad (doi: 10.5061/ Tullgren A (1918) Ein sehr einfacher Ausleseapparat fur€ terricole Tier- dryad.sg5s0). Specimen data is also available in GBIF € – faunen. Zeitschrift fur angewandte Entomologie, 4, 149 150. (http://www.gbif.org/dataset/04838789-732b-4cbf-8370- Vaidya G, Lohman DJ, Meier R (2011) SequenceMatrix: concatenation software for the fast assembly of multigene datasets with character set 0187e49b5d9c, http://www.390d8ae9-6ada-4485-8d32-397 and codon information. Cladistics, 27, 171–180. 84f66fe32, http://www.01479f71-9e66-4aa9-9481-dabeb5a Valentini A, Pompanon F, Taberlet P (2009) DNA barcoding for ecolo- ceaf4, http://www.0041409c-519a-4738-9a25-a38e641e16ef gists. Trends in Ecology & Evolution, 24, 110–117. and http://www.5ff795ba-b051-4c89-8489-43e053bef0d0) Varkonyi G, Roslin T (2013) Freezing cold yet diverse: dissecting a high- Arctic parasitoid community associated with Lepidoptera hosts. The [Correction added on 26 February 2016, after first online Canadian Entomologist, 145, 193–218. publication: accession number has been added].

Supporting Information Table S1 List of terrestrial animal species recorded in Zacken- berg. Additional Supporting Information may be found in the online version of this article: Table S2 List of terrestrial vascular plant species recorded in Zackenberg. Appendix S1 Methods of sampling and sorting. Fig. S1 Neighbour joining tree of animals in Zackenberg based Appendix S2 DNA barcode libraries for terrestrial animals and on CO1. vascular plants from Zackenberg and other parts of Greenland. Fig. S2 Neighbour joining tree of vascular plants in Zackenberg Appendix S3 The ability of DNA barcodes to discriminate based on rbcLa. among species in a local library. Fig. S3 Neighbour joining tree of vascular plants in Zackenberg Appendix S4 Details of applying DNA barcodes to community based on ITS2. samples from Malaise traps. Fig. S4 Neighbour joining tree of vascular plants in Zackenberg Appendix S5 Comparison of Malaise trap catches among sites based on rbcLa and ITS2. and years.