Barcoding the Collembola of Churchill: a Molecular Taxonomic Reassessment of Species Diversity in a Sub-Arctic Area
Total Page：16
File Type：pdf, Size：1020Kb
Molecular Ecology Resources (2013) doi: 10.1111/1755-0998.12172 Barcoding the Collembola of Churchill: a molecular taxonomic reassessment of species diversity in a sub-Arctic area € DAVID PORCO,* DARIUSZ SKARZ˙ YN´ SKI,† THIBAUD DECAENS,* PAUL D. N. HEBERT‡ and LOUIS DEHARVENG§ *EA 1293 ECODIV, Universite de Rouen, UFR Sciences et Techniques, SFR SCALE, B^atiment IRESE A, Mont Saint Aignan Cedex 76821, France, †Department of Evolutionary Biology and Ecology, Wrocław University, Przybyszewskiego 63/77, Wrocław 51-148, Poland, ‡Biodiversity Institute of Ontario, University of Guelph, 50 Stone Road East, Guelph, ON N1G 2W1, Canada, §Museum National d’Histoire Naturelle, UMR7205, «Origine, Structure et Evolution de la Biodiversite», 45 rue Buffon, CP50, Paris 75005, France Abstract Although their functional importance in ecosystems is increasingly recognized, soil-dwelling micro-arthropods are usually poorly known in comparison with their above-ground counterparts. Collembola constitute a significant and species-rich component of the soil biodiversity, but it remains a woefully understudied group because of the taxo- nomic impediment. The ever-increasing use of molecular taxonomic tools, such as DNA barcoding, provides a possi- ble solution. Here, we test the use of this approach through a diversity survey of Collembola from the vicinity of Churchill, Manitoba, Canada, and compare the results with previous surveys in the same area and in other sub-Arctic regions. The systematic barcoding campaign at Churchill revealed a diverse collembolan fauna consisting of 97 spe- cies-level MOTUs in six types of habitats. If all these MOTUs are confirmed as species, this richness would be far higher than prior records for Arctic Canada and could lead to reconsider the actual diversity of the group in Arctic environments. Keywords: Arctic, Collembola, cryptic diversity, diversity, DNA barcoding, species proxy Received 24 June 2013; revision received 9 September 2013; accepted 13 September 2013 Applying the ratio found by Mora et al. (2011) between Introduction described and unknown species for the whole animal The class Collembola is one of the most abundant groups kingdom, we estimate that the global diversity of Col- of decomposers in soil. With a global diversity of nearly lembola could reach about 65 000 species. The key reason 8000 described species, it occurs in a broad range of habi- for the current underestimation of collembolan diversity, tats, including the most extreme environments (Hopkin as in many other soil invertebrates, is the taxonomic 1997). Even in well-explored areas, such as Western impediment (Deca€ens 2010), that is, difficult taxonomy Europe, species new to science are still being discovered combined with a paucity of specialists. DNA barcoding, (Deharveng 2004), and many more are expected in tropi- a molecular taxonomic tool, using a portion of the 5′ end cal areas usually accepted as a rich and underinvestigat- of the mitochondrial gene COI as a species tag (Hebert ed source of biodiversity in many groups (May 2010). et al. 2003) provides a possible solution. The systematic The Neotropical zone currently includes 1557 described use of this tool for the development of extensive refer- species of Collembola (Mari Mutt & Bellinger 1990, ence libraries in Collembola has shown that even in tem- 1996). By contrast, Arctic environments have generally perate-zone environments and for common species, been found less rich in species: only 420 species of Col- considered well known, a significant part of the diversity lembola are known from the entire Arctic (Babenko & has been overlooked (Cicconardi et al. 2010; Porco et al. Fjellberg 2006). But the species richness of the different 2012a,b). These results suggest that the global diversity biomes are currently underestimated (Deharveng 2004). of Collembola is higher than expected, with many spe- cies new to science awaiting description in regions that Correspondence: David Porco, Fax: +33 2 35 14 66 55; were considered to be well explored, such as Western E-mail: [email protected] Europe. As a consequence, the current species count for © 2013 John Wiley & Sons Ltd 2 D. PORCO ET AL. Arctic regions, which have been less sampled, could Bay near the estuary of the Churchill River. This region likely be a great underestimate as well. Moreover, in the sits on the interface between boreal forest and tundra. context of global warming, the accurate estimation of the Habitats with a maritime influence such as coastal tun- species communities present in Arctic ecosystems is par- dra, bluff and beach are also present. ticularly critical as it will impact the cryophilic native fauna, favour introduced species, possibly affecting both Sampling ecosystem functioning and services delivery. This study presents information on the diversity of col- Two sampling trips focused on the springtail fauna (July lembolan species present near Churchill (Manitoba, Can- 2008 and July 2009), but extra samples were collected ada). The sampling activity and the sequencing of the during previous generalist expeditions (September 2000, specimens were carried out in the context of a large-scale August 2006, June 2007) (Table S1). The collembolan DNA barcoding study of polar life (PolarBOL – http:// fauna from soil, moss, lichen and rotten wood was www.polarbarcoding.org). This survey of the collembo- extracted using Berlese funnels. In addition, sweep net, lan fauna combined morphological analysis with DNA hand collecting, pitfall traps and bush beating were used barcodes to provide a comprehensive appraisal of the to complete the sampling for epiedaphic species. Six local diversity of this group in an Arctic environment. types of habitats (coastal tundra, tundra, beach, bluff, forest and fen) were investigated in the Churchill area and its vicinity, for a total of 48 sampling localities (Table Materials and methods S1, Fig. 1). From 3 to 20 replicates were sampled for each habitat type. Except for the fen, the sampling effort was Study site proportional to the representation of those habitats in The Churchill area is located in northern Manitoba, in the surrounding landscape (Table 1). All samples were the Canadian sub-Arctic region, on the shores of Hudson preserved in 95% ethanol. The specimens were sorted Fig. 1 Map of sampling sites. © 2013 John Wiley & Sons Ltd BARCODING THE COLLEMBOLA OF CHURCHILL 3 Table 1 Number of replicates and specimens collected per (http:dx.doi.org/10.5883/DATASET-CHU-COL1) and on habitat GenBank (GU657078–GU6573 35, HM390591–HM390627, HM390653–HM390662, HM 424138–HM424148, Habitats No. of replicates N HM431644–HM431645, HM893770–HM893811, HQ992117, – – Forest 3 125 JF884220 JF884222, JF884258, JN 269617 JN269618, Fen 4 26 JN306349–JN306356, JX261789, JX26 1792, JX261796, Bluff 5 123 JX261798, JX261801, JX261804, JX261808, JX261809, Beach 7 109 JX261814, JX261819, JX261833, JX261837, JX 261843, Coastal tundra 14 260 JX261844, JX261848, JX261857, JX261876, JX26 1884, Tundra 20 423 JX261886, KF641929–KF642599). Data analyses into morphospecies, and several individuals of each were sequenced from each sampling site. A total of 1066 Molecular distance analysis and MOTUs delineation. Dis- specimens were sequenced for the 48 sampling sites tance analyses were performed with MEGA5 (Tamura (Table S1). Identified specimens were deposited in the et al. 2011), utilizing a neighbour-joining (Saitou & Nei Museum of Natural History in Paris. Except for unique 1987) algorithm with the Kimura-2 parameter model specimens, duplicates of the vouchers were deposited at (Kimura 1980) to estimate genetic distances. The robust- the Biodiversity Institute of Ontario. ness of nodes was evaluated through bootstrap re-analy- sis of 1000 pseudoreplicates. The tree was replotted using the online utility iTOL (Letunic & Bork 2007). Molecular analysis MOTUs were defined with ‘mothur’ using Hcluster com- DNA was extracted from entire specimens in 30 lLof mand with the option ‘Furthest neighbour’ (Schloss et al. lysis buffer (http://www.ccdb.ca/docs/CCDB_DNA_ 2009). Extraction.pdf), and proteinase K incubated at 56°C over- night. DNA extraction followed a standard automated Rarefaction curves and diversity estimators. The data were protocol using 96-well glass fibre plates (Ivanova et al. analysed using rarefaction procedures that are specifi- 2006). Specimens were recovered after DNA extraction cally designed to avoid the potential bias generated by using a specially designed work flow allowing their mor- uneven sampling. Rarefaction curves for specific diver- phological examination (Porco et al. 2010a). Species were sity were generated with EcoSim 7.71 (Gotelli & identified using standard identification keys (mainly Pot- Entsminger 2006) with a 95% confidence level and plot- apov 2001; Fjellberg 1980; Christiansen & Bellinger 1998; ted with R 2.15.0 (R Development Core Team 2012) using Bretfeld 1999). The 5′ region of COI used as a standard the package ‘Plotrix’ (Lemon 2006). Rarefaction curves DNA barcode was amplified using M13-tailed primers were calculated separately for MOTUs and species accu- LCO1490 and HCO2198 (Folmer et al. 1994). Samples mulation as a function of sampling effort (number of col- that failed to generate an amplicon were subsequently lected individuals) in the four orders of Collembola, and amplified with a pair of internal primers combined with for MOTUs accumulation in the six different types of full length ones (LepF1-MLepR1 and MLepF1-LepR1) habitats sampled (coastal tundra, tundra, beach,