Cytochrome c oxidase barcodes for aquatic oligochaete identification: development of a Swiss reference database

Sampling and morphological analysis

The sampling was performed between 2013 and 2017 in eleven streams/rivers of four Swiss cantons (Geneva, Vaud, Bern and Lucerne) and at ten sites in Lake Geneva (Table S1). Sediments from the upper 10 cm were sampled using a shovel, a Surber type net (0.2 mm mesh size) or an Ekman type grab sampler (deep zones of lake). Before sieving, the biological material was either fixed with absolute ethanol or 5% formalin. The fixation of oligochaetes with formalin does not prevent genetic analyses if the specimens are preserved in this medium for a short time (<1 week) (Vivien, Ferrari & Pawlowski, 2016). At the laboratory, sediment samples were sieved (using a sieve of 0.5 mm mesh size). The material retained in the sieve was transferred into a Tupperware box, then examined under a stereomicroscope and oligochaete specimens were extracted. Each specimen was cut in two. The anterior parts were fixed and preserved in 5% formalin for morphological analysis and the posterior parts were preserved in 100% ethanol for DNA analysis. Anterior parts were cleared in an acid lactic/glycerol solution and mounted between slide and coverslip in a coating solution composed of lactic acid, glycerol and polyvinylic alcohol (Mowiol 4–88). Oligochaete specimens were identified to the lowest level (species if possible). The identification keys of Sperber (1950) and Timm (2009) were mainly used. The anterior parts served as reference vouchers and will be deposited at the Museum of Natural History of the city of Geneva.

Genetic analyses

Total genomic DNA was extracted using guanidine thiocyanate as described by Tkach & Pawlowski (1999). A fragment of 658 base pairs of the COI gene was amplified from each DNA extract using LCO 1490 and HCO 2198 primers (Folmer et al., 1994). The ITS2 rRNA region was amplified from some DNA extracts using the primers described in Navajas et al. (1998). The PCRs were performed in a total volume of 20 μl containing 0.6 Unit of Taq polymerase (Roche, Basel, Switzerland), 2 μl of the 10× buffer (Roche) containing 20 mM of MgCl2, 0.5 μl of each primer (10 mM each), 0.4 μl of a mix containing 10 mM of each dNTP (Roche, Basel, Switzerland) and 0.8 μl of template DNA of undetermined concentration. The PCR comprised an initial denaturation step at 95 °C for 5 min, followed by 35 cycles of denaturation at 95 °C for 40 s, annealing at 44 °C for 45 s and elongation at 72 °C for 1 min, with a final elongation step at 72 °C for 8 min.

COI and ITS2 PCR products were then bi-directionally Sanger sequenced on an ABI 3031 automated sequencer (Applied Biosystems, Foster City, CA, USA) using the same primers and following the manufacturer’s protocol. The raw sequence editing and the generation of contiguous sequences were accomplished using CodonCode Aligner (CodonCode Corporation, Centerville, MA, USA). Multiple sequence alignments were automatically generated using Muscle (Edgar, 2004) as implemented in Seaview program (Gouy, Guindon & Gascuel, 2010).

Phylogenetic trees comprising the COI sequences obtained in the present work and in our previous work (Vivien et al., 2015) were constructed using maximum likelihood phylogeny (PhyML 3.0) as implemented in ATGC: PhyML (Guindon et al., 2010). An automatic model selection based on Akaike Information Criterion (AIC) was used for PhyML 3.0 yielding in a GTR substitution model being selected for the analysis. Additional trees were constructed using FastMe 2.0, a distance based phylogeny inference program as implemented in ATGC: FastMe (Lefort, Desper & Gascuel, 2015). Two substitution models (F84 and T93) were tested including tree refinement with Subtree Pruning and Regrafting (SPR). Bootstrap values are based on 100 replicates for all analyses.

Two trees combining sequences with a divergence <5% were constructed: one tree comprising all Tubificinae lineages and some lineages of the other families/subfamilies; a second tree comprising all Naidinae, Pristininae, Rhyacodrilinae, Lumbriculidae, Lumbricidae, Enchytraeidae, Haplotaxidae and two lineages of Tubificinae. Additionally, two trees combining sequences with a divergence <1% were constructed (provided as Figs. S1 and S2): one tree for Tubificinae and another one for the remaining families/subfamilies mentioned above. On the four illustrated trees, the bootstrap values (BV) higher than 70% are shown.

A 10% threshold of COI divergence was applied to distinguish between species (Vivien et al., 2015). The intra and inter-lineage distances were calculated using the K2P model in MEGA 5.1 (Tamura et al., 2011). For the genetic distance calculations, we took into account all the sequences obtained in the present work and one or two sequence(s) per lineage of our database (Vivien et al., 2015). The sequences of our database used for distance calculations are represented in the trees. The discrimination of lineages diverging by a distance situated between 10 and 13.5% were considered as doubtful if the specimens corresponding to these sequences showed no morphological difference. ITS2 of these specimens were sequenced to confirm or not the segregation of the lineages.

The new COI sequences for Switzerland were compared to Genbank (NCBI) sequences using BLAST (http://www.ncbi.nlm.nih.gov/BLAST/Blast.cgi) and to BOLD sequences (http://www.boldsystems.org/index.php/IDS_OpenIdEngine), until July 2017. Our sequences and Genbank’s or Bold’s sequences were considered as belonging to the same species if their genetic divergence were ≤10%.

The COI and ITS2 sequences (raw data) are provided as Files S1 and S2. The COI sequences are accessible in the European Nucleotide Archive (ENA) at LT598611–LT598612; LT598614, LT598615, LT598616, LT598617; LT598619, LT598620, LT598621; LT598625; LT598628, LT598629, LT598630, LT598631, LT598632, LT598633; LT899859, LT899860, LT899861, LT899862, LT899863, LT899864, LT899865, LT899866, LT899867; LT899869–LT899898; LT903797–LT903836; LT903777–LT903796; LT903837, LT903838, LT903839, LT903840, LT903841, LT903842, LT903843, LT903844, LT903845, LT903846; LT904767, LT904768, LT904769, LT904770, LT904771, LT904772, LT904773, LT904774, LT904775, LT904776; LT905357–LT905411; LT906396, LT906397, LT906398, LT906399, LT906400, LT906401; LT906407–LT906426. The ITS2 sequences are available at LT906402, LT906403, LT906404, LT906405, LT906406; LT906427, LT906428, LT906429, LT906430, LT906431, LT906432, LT906433.

New COI data

A total of 180 specimens originating from stream/river sediments and 36 from lake sediments were sequenced (Table S1). These specimens belonged to five families: 140 specimens to Naididae (22 Naidinae, 117 Tubificinae, one Pristininae), 31 to Lumbriculidae, two to Lumbricidae, 41 to Enchytraeidae and two to Haplotaxidae (Table 1).

All the lineages obtained were separated by more than 10% of COI variation, with the exception of four lineages within the genera Nais and Uncinais. The minimal interlineage variation of the species Nais christinae and Nais stolci/pardalis was slightly >10%, while it was between 8.1 and 9 for the species Nais alpina, Nais communis (lineage N10), Nais pseudobtusa and Uncinais uncinata. These lineages were clearly differentiated by the morphological analysis and so the threshold of genetic variation of COI to discriminate the species in these two genera was fixed at 8%. The sequences within the lineage of Limnodrilus udekemianus and the lineage of Globulidrilus riparius E11 presented a genetic variation situated between 10 and 13.5%. The variation within L. udekemianus was between 6.7 and 10.7%. The three specimens of this group, all in an immature state, presented no morphological differences. The ITS2 sequences of the specimens diverging in COI by 10.7% were identical. So these COI sequences could be considered as belonging to the same lineage as neither the morphological analysis nor ITS2 data allowed to differentiate them. In G. riparius E11, the sequences diverged by 10.6–13.3%. All the specimens of this group were in an immature form and presented no morphological differences. ITS2 sequences of these specimens were identical. These COI sequences were so grouped in one lineage.

A total of 53 lineages could be distinguished based on COI and ITS2 divergence and morphological analysis: 35 Naididae (23 Tubificinae, 11 Naidinae and 1 Pristininae), 13 Enchytraeidae, two Lumbriculidae, two Lumbricidae and one Haplotaxidae.

One hundred and twenty specimens were assigned to 19 already described lineages in Switzerland, while 96 corresponded to newly identified lineages for Switzerland. Thirty-four new lineages were added to the Swiss COI database. They included 23 species (Potamothrix hammoniensis, Potamothrix vejdovskyi, Potamothrix moldaviensis, Potamothrix heuscheri, Spirosperma ferox, Embolocephalus velutinus, Tasserkidrilus kessleri, Vejdovskyella intermedia, Henlea perpusilla, Nais alpina, Haplotaxis gordioides, Nais christinae, Nais stolci/pardalis, Nais pseudobtusa, Uncinais uncinata, Chaetogaster diastrophus, Pristina jenkinae, Tubifex tubifex, Marionina argentea, Enchytraeus buchholzi, Nais communis (two lineages), Cernosvitoviella minor, Globulidrilus riparius (three lineages), six lineages of Enchytraeidae (three Fridericia sp, two Achaeta sp and one Lumbricillus sp) and two lineages of Tubificinae (one Tubifex sp and one Tubificinae sp). COI sequences of the species T. tubifex, M. argentea, E. buchholzi and N. communis had already been reported (Vivien et al., 2015), so the new lineages of these species corresponded to cryptic species. A mature and identifiable specimen of the lineage T16, previously identified as “Tubificinae without hair setae” (Vivien et al., 2015), was found. This lineage corresponds to an additional cryptic species of Limnodrilus hoffmeisteri.

Out of the 34 newly found lineages in Switzerland, the sequences of Potamothrix hammoniensis, Potamothrix vejdovskyi, Potamothrix moldaviensis, Spirosperma ferox, Henlea perpusilla, Cernosvitoviella minor, Nais alpina, Nais christinae, Nais communis (two lineages), Nais stolci/pardalis, Tubifex tubifex, Tasserkidrilus kessleri, Uncinais uncinata, Enchytraeus buchholzi, Vejdovskyella intermedia and one of the three lineages of Gobulidrilus riparius (E9) were present in the Genbank database. However, in Genbank, the sequences of Tasserkidrilus kessleri and Enchytraeus buchholzi were identified at the family level, the sequences of Nais pseudobtusa and Vejdovskyella intermedia were identified at the genus level and the sequence of Uncinais uncinata was falsely identified (Nais sp). Sequences corresponding to Potamothrix heuscheri and Chaetogaster diastrophus were present in Genbank but differed from our sequences of these species (COI divergence >13%). Our specimens of Henlea perpusilla and Cernosvitoviella minor were immature and so were identified only based on Genbank data. The lineage N12 could correspond to Nais stolci or Nais pardalis. According to the Genbank database, they belonged to N. stolci, but our two specimens of this lineage did not present clearly the specific features of this species, namely strongly enlarged ventral crotchets from segment VI. So they seemed closer to N. pardalis than to N. stolci. The differentiation of these two species is difficult on the basis of chaetal morphology as N. pardalis can also possess strongly enlarged ventral chaetae (Timm, 2009). N. pardalis also differs from N. stolci in revealing abrupt stomach dilatation and presenting equally long teeth of posterior ventral crotchets. The stomach dilatation was not visible in our preparations and the character of the length of teeth of posterior crotchets could not be considered as the posterior parts of the specimens had been used for genetic analysis.

Inventory of COI lineages in Switzerland

In total, 75 lineages have been found in Switzerland (Table S2): 33 Tubificinae, 15 Naidinae, one Pristininae, one Rhyacodrilinae, three Lumbricidae, 17 Enchytraeidae, four Lumbricidae and one Haplotaxidae. They corresponded to 44 morphospecies and 15 unidentified species and to cryptic species. The total number of cryptic species was 23: five in Tubifex tubifex, six in Limnodrilus hoffmeisteri, two in Enchytraeus buchholzi, two in Eiseniella tetraedra, two in Marionina argentea, three in Globulidrilus riparius and three in Nais communis. The sequence E3 (Enchytraeidae), identified in our previous work (Vivien et al., 2015) as Lumbricillus rivalis Levinsen 1884, was attributed by Klinth, Rota & Erséus (2017) to Lumbricillus rutilus as part of a systematics study of the genus Lumbricillus. Therefore, we assigned the species L. rutilus to the lineage E3.

Phylogenetic analysis of all COI lineages found in Switzerland

The phylogenetic trees (Figs. 1 and 2, Figs. S1 and S2) represent all the lineages found so far in Switzerland. The two different substitution models (F84, T93) tested for FastMe 2.0 yielded congruent results.

The subfamily Tubificinae was monophyletic, sustained by a BV of 86% (Fig. 1). The Tubificinae with and without hair setae were globally well separated. The Tubificinae with hair setae correspond to the lineages T1–T13, T24–T29 and T31–T33 and the Tubificinae without hair setae to the lineages T14–T23 and T30. These two groups did not branch together, with the exception of the species Branchiura sowerbyi (Tubificinae with hair setae) and Potamothrix moldaviensis (Tubificinae without hair setae). This last species logically branched with the other Potamothrix species that all possess hair setae. The genera Tubifex and Limnodrilus appeared as polyphyletic. The tree combining sequences with a divergence <1% (Fig. S1) was almost identical to the one represented in Fig. 1. Differences occurred in the branching position of lineages T24 and T25 that clustered at the base of Tubificinae in Fig. 1 but branched with lineage T4 in Fig. S1. Lineages T13 and T33 branched at the base of Tubificinae next to lineage T8 in Fig. 1, while they branched with lineages T4, T24 and T25 in Fig. S1.

The families Enchytraeidae, Lumbriculidae and Lumbricidae were monophyletic but none of them showed any support (Fig. 2). Within the family Naididae, the subfamilies Naidinae, Tubificinae and Rhyacodrilinae were monophyletic and well supported (BV respectively 93%, 99% and 100%). The subfamily Pristininae branched at the base of the subfamily Naidinae but the support was moderate (BV 72%). The family Haplotaxidae branched at the base of the family Enchytraeidae but the branching was not supported. The tree combining sequences with a divergence <1% (Fig. S2) differed from the one represented in Fig. 2 in the branching position of Haplotaxidae, clustering within Enchytraeidae, but the branching was not supported.