The holotype of the new species was collected by the senior author by sweeping in Muránska planina National Park in central Slovakia. This national park belongs to the most valuable protected areas in Slovakia with regard to biodiversity, with many rare and thermophilous species, often reaching there the northernmost limit of their distribution (cf. Ševčík and Kurina 2011a, b). The habitats include mainly karst valleys and limestone rocky slopes, covered mostly by beech and spruce forests.

The material used for DNA extraction was collected with Malaise traps or sweep net at different localities of the Czech and Slovak Republics in the years 2012–2015. The voucher specimens are deposited in the collection of Jan Ševčík (JSEVC) or in the Silesian Museum, Opava, Czech Republic (SMOC).

All the specimens were collected in 70% ethanol. The holotype is stored in the glycerine medium in a plastic pinned microvial. Before placing in glycerol the specimen was incubated in proteinase K to extract DNA.

Genomic DNA was extracted using NucleoSpin Tissue Kit (Macherey-Nagel, Düren, Germany) following the included protocol. The specimens were cleaned with PBS (Phosphate Buffered Saline) and left in lysis buffer with proteinase K overnight at 56 °C. The primers used for the PCR amplifications were as follows: LCO1490 (5′-GGTCAACAAATCATAAAGATATTGG-3′) and HCO2198 (5′-TAAACTTCAGGGTGACCAAAAAATCA-3′) (Folmer et al. 1994) for COI; CYTB-F (5’-TATGTTTTATGAGGACAAATATC-3’) and CYTB-R (5’-AAATTCTATCTTATGTTTCAAAAC-3’) (Su et al. 2008) or mCYTB-R (5’-ATTACTCCCCCTAATTTATTAGGAAT-3’) (Ševčík lab) for cytochrome B; ITS2a (5’-TGTGAACTGCAGGACACAT-3’) and ITS2b (5’-TATGCTTAAATTCAGGGGGT-3’) (Beebe & Saul, 1995) for ITS2; and mCOII-F (5’-CAAGATAGAGCTTCTCCTCTTATAG-3’) and mCOII-R (5’-GGCATAAATCTATGATTAGCCCCAC-3’) (Ševčík lab) for COII. The amplification programme for the gene fragments was 94 °C for 3 min, followed by 35 cycles of 94 °C for 1 min, 49–50 °C for 1 min and 72 °C for 1:30 min and a final extension step at 72 °C for 7 min. The obtained PCR products were purified using Gel/PCR DNA Fragments Extraction Kit (Geneaid, New Taipei City, Taiwan) following manufacturer’s protocol and sequenced by Macrogen Europe (Netherlands).

German material discussed here was obtained within the German Barcode of Life Project (GBOL). Specimens used for DNA extraction were collected with Malaise traps at two different localities in Germany in the years 2013 and 2015. All the specimens were collected in pure 96% ethanol.

Genomic DNA was extracted from legs of the specimens using the BioSprint96 magnetic bead extractor by Qiagen (Hilden, Germany). Polymerase chain reaction (PCR) was carried out in total reaction mixes of 20 μl, including 2 μl of undiluted DNA template, 0,8 μl of each primer (10 pmol/μl), 2 μl of ‘Q-Solution’ and 10 μl of ‘Multiplex PCR Master Mix’, containing hot start Taq DNA polymerase and buffers. The latter components are available in the Multiplex PCR kit from Qiagen (Hilden, Germany). PCR reactions were run individually and not multiplexed.

Thermal cycling was performed on GeneAmp PCR System 2700 (Applied Biosystems, Foster City, CA, USA) as follows: hot start Taq activation: 15 min at 95 °C; first cycle set (15 repeats): 35-s denaturation at 94 °C, 90-s annealing at 55 °C (−1 °C/cycle) and 90-s extension at 72 °C. Second cycle set (25 repeats): 35-s denaturation at 94 °C, 90-s annealing at 40 °C and 90-s extension at 72 °C; final elongation 10 min at 72 °C using the primers LCO1490-JJ: 5´-CHACWAAYCATAAAGATATYGG- 3´ with HCO2198-JJ: 5´-AWACTTCVGGRTGVCCAAARAATCA- 3´ respectively (Astrin and Stüben 2008).

Sequencing of the unpurified PCR products in both directions was conducted at Beijing Genomics Institute (Hongkong, CN) by using the amplification primers. Barcode sequence analysis was done using the Geneious® software version 7.1.7 (http://www.geneious.com).

Vouchers were deposited in the collection of the Zoologisches Forschungsmuseum Alexander Koenig, Bonn, Germany.

The sequences were aligned using MAFFT version 7 (Katoh and Standley 2013) on the MAFFT server (http://mafft.cbrc.jp/alignment/server/) with default settings and then manually edited. The protein-coding gene COI, COII and CytB sequences were checked based on amino-acid translations and yielded indel-free nucleotide alignments. All unreliably aligned regions of ITS2 fragment were removed in GBLOCKS 0.91b program (Castresana 2000) on the Gblocks server (http://molevol.cmima.csic.es/castresana/Gblocks_server.html). We created four alignments, one for each gene, and one concatenated alignment for all gene fragments with 15 taxa, including additional sequences from GenBank (KT316839, KC435639, KC435683 and KC435708). The final molecular dataset consists of 2039 characters: COI–658, COII–546, CytB–433, ITS2–402 bp. The datasets were analysed using maximum likelihood analyses conducted on the CIPRES computer cluster using RAxML-HPC BlackBox 7.6.3 (Stamatakis 2006) employing automatic bootstrapping on the partitioned dataset. All the sequences were deposited into GenBank and BOLD (https://doi.org/10.5883/DS-DOCODENT), where detailed metadata is available (see Table 1). As outgroup taxa we selected two representatives of the subfamily Gnoristinae, in concordance with previous molecular studies (Rindal et al. 2009, Ševčík et al. 2013).

Genetic distances were calculated in MEGA6 (Tamura et al. 2013) using Kimura 2-parameter model (K2P) with pairwise deletion for the treatment and they are demonstrated in Table 2 and Fig. 5.

The genetic distances among the 13 species of Docosia and two outgroup species for particular gene markers are presented in Tab. 2. The interspecific differences among Docosia species ranged from 0.3% (in ITS2 between D. dentata and D. muranica) to 26.0% (in CytB between D. fumosa and D. lastovkai), with the mean interspecific distances for particular genes: 12.8% (COI), 10.3% (COII), 14.9% (CytB) and 4.5% (ITS2). The genetic distances for the nuclear ITS2 marker were remarkably lower than for the mitochondrial markers. Fig. 5 shows that the distribution of K2P distances for cytB is rather broad (6.6–28%), while the distributions for COI, COII and ITS2 are relatively narrow.

The phylogenetic tree for the concatenated dataset is presented in Fig. 4. The genus Docosia was found to be monophyletic with maximum bootstrap support value (BV = 100). Surprisingly, D. fumosa branched basally as a sister group to all the other Docosia species included in this analysis, followed by D. gilvipes. All the other Docosia species grouped together as a monophyletic group with maximum support (BV = 100). Within this group, only D. diutina and D. landrocki were found to be closely related (BV = 100) while the other relationships between any two species are less supported. Also the monophyly of the group comprising eight terminal species (see Fig. 4) is highly supported (BV = 100). A sister taxon to this group is D. fuscipes, while D. muranica is sister taxon to the entire latter group. Finally, D. lastovkai is sister taxon to the all Docosia species in this dataset, except D. fumosa and D. gilvipes.

Figure 4. 

Maximum likelihood hypothesis for relationships among selected species of Docosia Winnertz based on DNA sequence data (COI, COII, CytB and ITS2), 2039 characters. Above node number = bootstrap support for ML.

Figure 5. 

Frequency distributions of genetic distances (K2P) for particular gene markers.

According to the key by Laštovka and Ševčík (2006), the new species runs to couplet 10 (D. montana Laštovka & Ševčík, 2006), because of bare laterotergites, yellow coxae and basal flagellomeres relatively long. The structure of the male terminalia is, however, quite different from this species and also from two other species described later (D. chandleri Ševčík & Laštovka, 2008 and D. matilei Ševčík & Laštovka, 2008), which would run to couplet 10 too. It is thus difficult to find the closest relative of D. dentata according to morphological characters.

Considering the genetic distance, the closest relative of Docosia dentata is D. setosa, based on COI (6.0 %), COII (5.1%) and CytB (6.6%) gene markers. In the phylogenetic tree based on all the four gene markers (Fig. 4), D. dentata forms a sister branch to a group of the following four species: D. flavicoxa, D. sciarina, D. montana, and D. setosa.

The comparison of the four variable gene regions revealed that the genetic distances between Docosia species are quite high for CytB and COI, with the average values being 14.9% and 12.8%, respectively. From this point of view, CytB performs as the best barcoding marker for Docosia species, followed by the traditional animal barcoding region (COI), COII and ITS2. The ITS2 sequences show remarkably high similarity in the genus Docosia and do not appear as a suitable barcode marker in this case. The high uniformity of ITS2 has been reported in several studies (e.g. Lv et al. 2014 or Navajas et al. 1988). At least within Mycetophilidae, the ITS2 marker may thus possibly be more suitable for intergeneric comparisons and higher phylogeny, similarly as the neighbouring ribosomal 28S region. In mycetophilids, ITS2 has already been successfully used for inferring phylogeny at generic or subgeneric level by Ševčík et al. (2013, 2014). This issue definitely deserves further study.

There are many studies comparing the utility of various gene markers for DNA barcoding and the identification of species (e.g. Bourke et al. 2013, Lv et al. 2014, Schwarzfeld and Sperling 2014). Concerning fungus gnats, comparisons between COI and ITS2 were recently provided by Jürgenstein et al. (2015) and Kurina et al. (2015). They both came to the conclusion that COI performed better.

The CytB region has mostly been used in the studies devoted to vertebrates but it was recently used also for barcoding of Calliphoridae (GilArriortua et al. 2013), tse-tse flies (Orji et al. 2015) or aphids (Chen et al. 2012).

There has been no previous phylogenetical hypothesis for Docosia species published to date but if we take into account the characters used in most available keys, there is only little congruence between those morphological characters and the molecular results presented here. For example, Docosia fumosa does not appear as particularly separated from the other species by morphological characters. It belongs to the group of species with pubescent lateroterites, together with D. flavicoxa, D. gilvipes, D. moravica, D. sciarina, and several other species, not represented in our molecular dataset. These latter species do not constitute a monophyletic group in the molecular tree (Fig. 4) nor do the species with bare laterotergites. This means that this widely used character (pubescence of laterotergites) has most probably no or little value in reconstructing relationships within the genus.

The second widely used character in the keys is the coloration of legs, mainly the coxae, which also does not help much in defining any group congruent with the clades in the molecular tree presented here.

Further, it is difficult to find clear morphological synapomorphies for the highly supported clades in the tree (Fig. 4), e.g. for the sister relationship of D. diutina + D. landrocki (except for the overall resemblance of their male terminalia).

Interestingly, in two of the four trees based on one gene region only (in COI and ITS2, data not shown), D. gilvipes (not D. fumosa) branched basally as a sister group to the rest of Docosia, a result which would be expected on the basis of morphological and ecological data (D. gilvipes is the only species in the dataset with Sc setose and ending free, different structure of male terminalia, endomycophagous larvae etc.).

Nevertheless, the number of species included in this analysis is rather limited to reach any final conclusion. A more comprehensive phylogeny of the genus, as well as of other genera of fungus gnats, is thus needed.