On the systematics of the fungus gnat subfamily Mycetophilinae (Diptera): a combined morphological and molecular approach

Sampling

The specimens included in the molecular study were collected at 10 localities in Norway and Sweden and at one locality in Korea (Table 1) using sweep nets and Malaise traps with 80% ethanol as fixative.

We attempted to include representatives of all currently recognized Mycetophilinae genera in the study, but this could not be achieved. Unfortunately, available collection material turned out not suitable for molecular analyses, i.e. the DNA extractions did not yield genetic material of reasonable quality to serve as appropriate PCR template. The outgroup taxa were the same genera as used by Rindal and Søli (2006), i.e. Boletina, Leia and Docosia belonging to Gnoristinae and Leiinae (Mycetophilidae).

Morphological data

The morphology-based taxonomy and nomenclature of Mycetophilinae species follows Søli (1997), and the morphological characters and the respective data matrix have been published by Rindal and Søli (2006). The present study is based on these morphological data. The data set of Rindal and Søli (2006) was trimmed to match the taxa included in the molecular analyses presented here, and consist of 65 characters for 24 genera.

Molecular data

DNA was extracted from the abdomen of the specimens following the instructions of the Puregene kit (Gentra Systems, Minneapolis, MN, USA). The genitalia were stored in glycerol in micro vials as vouchers and deposited in the entomological collection of the Natural History Museum, Oslo.

Details on the molecular methods for amplification and sequencing of the nuclear 18S rDNA, and the mitochondrial 16S rDNA and coxI genes are described in Rindal et al. (2007). The amplification programme for the 28S gene was 94°C for 3 min; 35 cycles of 94°C for 30 s, 50°C for 30 s and 72°C 1 min 50 s; and a final extension step at 72°C for 7 min. All PCR amplifications were performed using the recombinant Taq polymerase of Roche (Basel, Switzerland). The primers used for PCR amplifications and sequencing for 28S are Forward_C1: ACC CGC TGA ATT TAA GCA T and Reverse_C1: TGA ACT CTC TCT TCA AAG TTC TTT TC. All sequences have been deposited in Genbank and their accession numbers are listed in Table 1.

Proofreading of the obtained nucleotide sequences and subsequent alignment was straightforward and initially performed using genetools 2.0 (Wishart and Fortin 2001) and the alignment was subsequently optimized by eye. Variable regions in the 18S and 28S sequence alignment that were considered arbitrary because of the occurrence of indels (up to 51 bp per sequence), were omitted from the subsequent analyses.

Phylogenetic reconstruction

The phylogenetic analyses were performed on two different data sets. First the nucleotide sequence data were analysed, using both a maximum parsimony (MP) and a Bayesian approach. Subsequently, a combined data set consisting of both morphological and molecular data was subjected to a MP analysis.

Bayesian analyses of the molecular data set were conducted with an online version of MrBayes (Huelsenbeck and Ronquist 2001) implemented at the Bioportal at the University of Oslo (http://www.bioportal.uio.no). Modeltest 3.06 (Posada and Crandall 1998) was used to estimate the best-fitting substitution model for the analyses. Using the Akaike information criterion (AIC), the best model of nucleotide substitution for the 18S, 16S and coxI was the general time reversible model with gamma distributed rate heterogeneity and a significant proportion of invariable sites (GTR + I + G), for the 28S data set it was the GTR + I model. Bayesian inference analyses were performed under 4 000 000 generations and four Metropolis-coupled Markow chains, taking samples every 100 generations, with the first 4000 samples discarded as burn-in. From the resulting trees a posteriori probabilities for individual clades were assessed based on their observed frequencies.

Due to the substitution saturation of cox1 (see Results), additional runs were conducted: (1) without coxI, (2) with 3rd position of coxI excluded and (3) using only the 2nd position of coxI.

The settings for the parsimony analyses were the same for both the molecular and the combined approach. paup* 4 beta 10 win (Swofford 2003) as implemented at the Bioportal at the University of Oslo (http://www.bioportal.uio.no) was used to construct the most parsimonious (MP) cladograms. The parsimony analysis utilized a heuristic search with 1 million replicates and treating gaps as a fifth character state. Gaps can be treated in different ways. If coded as missing data they will not be informative, and thus not contribute to the phylogenetic reconstruction. Alternatively, treating gaps as a fifth state allows for retaining the evolutionary information associated to an assumed indel. Bootstrap analyses were performed with 1000 replicates and 100 searches within each bootstrap replicate.

Pair-wise partition homogeneity tests as implemented in paup* 4 beta 10 win (Swofford 2003) with 200 replicates and 10 searches within each replicate were conducted for the 18S, 28S, 16S and coxI data sets.

Saturation plot

Saturation plots (Fig. 1a–g) were made using p-distances plotted against the distances based on the model chosen by the Modeltest analyses, i.e. GTR + I + G distances for 18S, 16S and coxI and GTR + I distances for 28S, in accordance with Sullivan and Joyce (2005). Individual plots were made for the 18S, 28S, 16S and coxI respectively. The different codon positions of the coxI gene were also plotted separately.

The subfamily Mycetophilinae

The present study, based on a combined data set including nucleotide sequences of nuclear (18S and 28S rDNA), mitochondrial (16S rDNA and coxI) and morphological characters, confirms the monophyly of the subfamily Mycetophilinae with high statistical support. The current results are congruent to those obtained by Rindal and Søli (2006) and Rindal et al. (2007). Rindal and Søli (2006) analysed exclusively morphological characters, whereas Rindal et al. (2007) based their conclusions exclusively on nucleotide sequence data with the tribe Mycetophilini only represented by two genera.

The tribe Mycetophilini

The combined analyses of the morphological and molecular data also provided strong support for the monophyly of the tribe Mycetophilini including the genus Dynatosoma. As with the subfamily Mycetophilinae, this is in agreement with previous results by Rindal and Søli (2006) and Rindal et al. (2007). Hence, the notion of Tuomikoski (1966), who suggested that Mycetophilini might be paraphyletic and instead considered the Mycetophila- and the Phronia-groups tribes in their own right, is not supported, but cannot formally be rejected on the basis of the taxa included in this study.

The systematic position of Dynatosoma is particularly interesting. Dynatosoma was considered by Tuomikoski (1966) to be a member of the Phronia-group, without specifying synapomorphic characters in more detail. Later, Rindal and Søli (2006) found Dynatosoma in a trichotomy with Epicypta and Plarurocypta, thus considering the genus as part of the Mycetophila-group. In the parsimony analysis of the complete molecular data set, Dynatosoma is found in a trichotomy with the Mycetophila- and Phronia-group, whereas in the combined data set it is the sister-taxa to all other Mycetophilini taxa. On the other hand, the Bayesian analysis places the genus basal to the Phronia-group, though with low posterior probability. Nevertheless, the clade consisting of Phronia, Trichonta and Macrobrachius is recovered with high posterior probability in the Bayesian analysis.

The Mycetophila- and the Phronia-groups are also well resolved within the Mycetophilini. A diagnostic morphological character for separating the Phronia-group from the Mycetophila-group is the presence of a distinct, small, ovate plate above the antennal socket (for further details see Fig. 4 published in Rindal and Søli 2006). In Dynatosoma this area is well sclerotized and furnished with setae, but connected to the frons. At first glance, this condition is more similar to that found in the species of the Phronia-group than it is to the outline of a bare, weakly sclerotized membrane that is characteristic for the species in the Mycetophila-group. This may be taken as support for a basal position of Dynatosoma in the Phronia-group as recovered by the Bayesian analyses presented here. However, it needs to be stressed that the affinity of Dynatosoma and the Phronia-group in the Bayesian hypothesis is not supported by high posterior probability. Moreover, Dynatosoma does not share similarities with members of the Phronia-group in general appearance or in genital structures.

Within the Mycetophila-group, Sceptonia and Zygomyia are recovered as sister-groups in the combined analysis. This may indicates that the loss of M₄ in wing venation, following the interpretation of Amorim and Rindal (2007) is a suitable diagnostic synapomorphy for this group. However, the Bayesian and parsimony analysis based on molecular data contradicts this and places Sceptonia together with Epicypta. The phylogenetic position of Mycetophila as the sister-group to the other genera within the Mycetophila-group is found in all trees. The monophyly of the highly diversified genus Mycetophila has never been properly demonstrated, and, currently, it cannot be excluded that the genus is paraphyletic. However, an adequate test of the monophyly of Mycetophila is beyond the scope of this study.

The tribe Exechiini

The combined analysis of morphological and molecular data did not recover any intergeneric relationships within the tribe Exechiini with high statistical support. In the parsimony analyses (2, 3) Exechiini is divided into three clades, though the composition differ between the trees and little statistical support is found for any of the arrangements. The only noteworthy grouping relates to the genera Cordyla and Brachypeza that form a common clade in all trees, though with low statistical support. It is interesting to note that these two genera also share some morphological traits, in particular between Cordyla and Paracordyla, a subgenus of Brachypeza. Their resemblance was also mentioned by Tuomikoski (1966) in his description of the genus. Tuomikoski, however, suggested that this might be a result of convergence.

Concluding remarks

The present study provides support for the monophyly of the subfamily Mycetophilinae and its two tribes, Mycetophilini and Exechiini, as well as the monophyly of the Phronia-group within Mycetophilini. The internal phylogeny of Exechiini, however, remains largely unresolved. Combining data set might give an increased resolution as compared to data sets including only morphological or molecular characters (reviewed in Wortley and Scotland 2006), and sometimes also a better statistical support (Wahlberg et al. 2005). Our study yielded a better resolution, but did not show a significantly better support than found in Rindal and Søli (2006) and Rindal et al. (2007). Thus, future studies need to address the phylogenetic relationships within Exechiini through new approaches. Better phylogenetic resolution may be achieved by including more species for each genus. Geographical variation and biogeography may offer additional useful criteria for the selection of taxa.