Sampling and data collection

Larval Micropsectra specimens were obtained from the benthos of 5 streams in Minnesota, USA, during late winter/early spring 2010. All specimens were held in constant-temperature incubators at 5°C to approximate stream temperatures and allowed to develop to adulthood in mass rearings, resulting in a cumulative total of 99 adult Micropsectra with associated pupal exuviae. Rearing chambers were typically checked twice per day to ensure that emerging adults were associated with the appropriate exuviae. Upon emergence, adults and associated exuviae were stored in individual vials and preserved in 95% ethanol. Unfortunately in nearly all cases, larval exuviae could not be retrieved. The antennae, head, wings, legs, and hypopygium were dissected. The head and hypopygium were cleared in 10% KOH and slide mounted in Euparal® with the pupal exuviae, legs, wings, and antennae, for species identification. The thorax and abdomen were slide-mounted after DNA extraction (described below). On the few occasions where >1 fly of the same sex emerged between observations, the adults and exuviae were stored in the same vial until dissection. Adults were mounted on individual slides as described above, but both exuviae were mounted on a single, separate slide. All specimens were identified and separated into 4 morphotypes. A subset of 46 specimens collected from Minnesota streams was selected for molecular analyses (Table S1; available online from:  http://dx.doi.org/10.1899/12-020.1.s1 (10.1899_12-020.1.s1.doc)). Specimens were selected to include representatives from each Micropsectra morphotype. In addition, 1 Tanytarsus specimen was sequenced to include among the outgroup species.

The full COI data set consisted of 291 individuals, including 5 outgroup species and 44 Micropsectra species (31% of known Micropsectra species). Of these, 4 species (43 individuals) were collected in Minnesota: Micropsectra neoappendica n. sp., Micropsectra penicillata n. sp., Micropsectra subletteorum n. sp., and Micropsectra xantha (Roback, 1955) (Table S1). The remaining sequences were acquired from GenBank, and included 3 taxa of Tanytarsus and 2 taxa of Paratanytarsus, which served as outgroups (Table S1). A 2^nd reduced COI data set that conformed to the CAD data set (described below) also was generated and analyzed. The CAD data set consisted of 82 individuals from 37 species (33 Micropsectra and 4 outgroup taxa; 27% of known Micropsectra species). Twelve individuals were from the above-named Micropsectra taxa. The remaining sequences were acquired from GenBank, and included 2 Paratanytarsus and 2 Tanytarsus species that served as outgroups. Voucher and sequence information for all specimens collected are also accessible in BOLD project “Nearctic Micropsectra (NAMIC)” and the BOLD data set “Holarctic Micropsectra (DS-HOLMIC)” ( http://boldsystems.org). Remaining DNA extracts are stored at the Museum of Natural History and Archaeology, Norwegian University of Science and Technology (VM). A detailed record of all taxa, including collection localities and GenBank accession numbers, is available in Table S1.

DNA extraction, amplification, and sequencing

A GeneMole® instrument (Mole Genetics AS, Lysaker, Norway) was used to extract DNA from most specimens, following standard protocol for the MoleStrips DNA Tissue kit. DNA was extracted from remaining specimens using protocol for the Qiagen DNeasy tissue extraction kit (Qiagen, Oslo, Norway), with the exception that less elution buffer (150 µL) than recommended was used because of small specimen sizes. Polymerase chain reaction (PCR) amplification and sequencing was carried out on partial COI and CAD genes using the primer pairs and PCR programs described in table 1 of Ekrem et al. (2010b). We used HotStar Taq (Qiagen) when amplifying COI and TaKaRa HS ExTaq (Takara Bio Inc., Otsu, Shiga, Japan) for amplification of CAD. PCR products were purified using ExoSAP-IT® (Affymetrix Inc., Cleveland, Ohio) and sent to an external sequencing service (Eurofins MWG Operon, Ebersberg, Germany). All gene products were sequenced bidirectionally, and forward and reverse sequences were aligned and manually edited using Sequencher 4.8 (Gene Codes Corp., Ann Arbor, Michigan). Conflicting or ambiguous base calls were initially given the appropriate International Union of Biochemistry ambiguity symbol. Initial ambiguities were checked after alignment and corrected, if possible, with reference to trace files. Consensus sequences were assembled and initially aligned in BioEdit (version 7.0.5.3; Hall 1999) for inspection. Introns in CAD sequences were recognized by the guanine thymine–adenine guanine (GT–AG) rule (Rogers and Wall 1980) and eliminated from the alignment because these areas exhibited extreme variation among species. All CAD sequences had intron placement at position 725 and no indication was seen of paralogous gene copies. CAD exons were aligned by their amino acids in MEGA5 (Tamura et al. 2011; available from:  www.megasoftware.net) using the Clustal-W algorithm with the BLOSUM weight matrix and gap opening costs of 10 and gap extension costs of 0.1 and 0.2 in pairwise and multiple alignments, respectively. Alignment of COI sequences was trivial because no indels were observed within the sequences.

Tree-based species delimitation and phylogenetic analyses

Phylogenetic analysis of the initial full COI data set was inferred with neighbor joining (NJ) and maximum likelihood (ML) methods in MEGA5. The Kimura 2-parameter (K2P) substitution model was used for the NJ analysis to infer evolutionary distances because it has been a standard substitution model in other barcode studies, thus allowing comparisons among studies. Although this method is commonplace in barcoding studies, recent authors have questioned the strict use of the K2P model (Srivathsan and Meier 2012). This concern prompted us to include ML analyses with 500 bootstrap replicates using the General Time Reversible model with a discrete Gamma distribution and invariable sites (GTR+G+I), which was the most appropriate nucleotide substitution model for our data set based on the Bayesian (BIC) and corrected Akaike Information Criterion (AIC_c) in the ML model selection feature of MEGA5.

The phylogenetic relationships among species in Micropsectra were further explored using CAD because of its greater ability to reconstruct lower-level phylogenies in Diptera (Winterton et al. 2007, Winkler et al. 2009, Ekrem et al. 2010b). The best-fit model of substitution was identified using a ML criterion in MEGA5. BIC indicated that the Tamura 3-parameter model with a discrete Gamma distribution and invariant sites (T92+G+I) was the most appropriate model, whereas AICc indicated that the GTR+G+I model best fit our CAD data set. Therefore, we used the GTR+G+I model for all ML and Bayesian analyses because the T92 model is not incorporated in RAxML (version 7.4.2; Stamatakis 2006) or MrBayes (version 3.1.2; Huelsenbeck and Ronquist 2001;  http://mrbayes.sourceforge.net/). The best-fit partition scheme was found using PartitionFinder (version 1.0.1; Lanfear et al. 2012) and contained 2 partitions in the nucleotide sequences: 3^rd position and 1^st+2^nd position. Multiple tree-building methods were used to assess whether phylogenetic results were robust to different analytical assumptions. ML analysis with 500 bootstrap replicates was run on the partitioned data set in RAxML using the software raxmlgui (Silvestro 2012). Bayesian analysis (BA) was done on the partitioned data set with MrBayes. For BA, we specified flat prior probabilities and ran 2 parallel analyses, each with 4 chains, for 5 million generations, sampling every 1000^th generation and using a 10% burn-in. We examined the trace files with Tracer 1.5 (Rambaut and Drummond 2008;  http://tree.bio.ed.ac.uk/software/tracer/) to verify that Markov Chain Monte Carlo (MCMC) convergence had occurred and that the estimated sample size (ESS) had reached an acceptable level before summarizing the data from the sampled trees in a consensus. Trees created by BA were imported and modified for presentation in MEGA5. A reduced COI data set, corresponding to the same individuals analyzed for CAD, also was analyzed phylogenetically with ML using the GTR+G+I substitution model and 500 bootstrap replicates in MEGA5. Last, a concatenated data set containing COI and CAD sequences was constructed and analyzed with both RAxML and MrBayes with the following 5 partitions identified by PartitionFinder: COI 1^st position, COI 2^nd position, COI 3^rd position, CAD 1^st+2^nd position, CAD 3^rd position.

Species descriptions and specimen deposition

Most of the morphological terminology and abbreviations used here follow Sæther (1980), with the following exceptions: taeniae, introduced by Langton (1994), is used to describe the flattened setae found on the pupal exuviae and the term setiger (Spies 1998) describes the setae-bearing area of the superior volsella on the male hypopygium. The following abbreviations also are used: P♂  =  associated pupal exuviae and adult male; P♀  =  associated pupal exuviae and adult female; L_hcP♂  =  associated larval head capsule, pupal exuviae, and adult male. Measurement techniques follow those outlined by Soponis (1977), with the exception of the gonocoxite length, which was measured along the limb's longest access, as described by Stur and Spies (2011). Lengths of the genital volsellae were measured along their median margin, and the anal point was measured from the anterior ends of the anal crests to the tip of the anal point. Mensural data are presented as ranges, followed by the mean and number of observed specimens.

Holotypes and most paratypes are deposited in the University of Minnesota Insect Collection in St Paul, Minnesota, USA (UMSP). Additional paratypes are deposited in VM. We also examined type material and other reference material of closely related taxa from the following collections (abbreviations correspond to those used in text): The Natural History Museum, London, UK (BMNH); Canadian National Collection of Insects, Arachnids and Nematodes, Ottawa, Ontario, Canada (CNC); private collection of Peter H. Langton, Londonderry, Northern Ireland (PHL); UMSP; VM; Natural History Collections, Bergen Museum, University of Bergen, Norway (ZMBN); and Zoologische Staatssammlung München, Munich, Germany (ZSM).

Tree-based species delimitation

All Nearctic specimens collected for our study formed distinct monophyletic groups at the species level in analyses of the COI, CAD, and the concatenated COI/CAD data, and distinctly separated the 4 Nearctic Micropsectra from morphologically similar Palearctic counterparts (Figs 1, 2, Figs S2, S3; available online from:  http://dx.doi.org/10.1899/12-020.1.s2 (10.1899_12-020.1.s2.pdf) and  http://dx.doi.org/10.1899/12-020.1.s3 (10.1899_12-020.1.s3.pdf)). This result is particularly notable for 2 of the Nearctic species, M. neoappendica n. sp. and M. subletteorum n. sp., which are very similar morphologically to species known from the Palearctic. Micropsectra subletteorum n. sp. is closely allied morphologically with many species within the notescens species group. Males of M. neoappendica n. sp. fit the species description of the Palearctic Micropsectra appendica Stur and Ekrem, 2006 almost perfectly. Only after divergence patterns were discovered with molecular analyses of COI and CAD markers, did subsequent examination reveal subtle but corroborating morphological characters that can be used to differentiate M. neoappendica and M. appendica (see Species Description below).

Fig.1. 

Full mitochondrial cytochrome c oxidase subunit I (COI) maximum likelihood (ML) phylogeny of 291 individuals from 49 species. Taxa denoted with ▪ were collected in Minnesota and are described/redescribed in our study. Numbers on branches indicate ML bootstrap support values. Values within highly supported (>90%) species clades were removed for ease of interpretation. Low bootstrap support values are included for the higher-level relationships to indicate the inability of COI sequences to resolve relationships between species and species groups clearly, particularly when compared to nuclear carbamylphosphate synthetase (CAD) sequences (Fig. 2). The scale bar indicates branch lengths as number of substitutions per site corrected by the General Time Reversible model with a discrete Gamma distribution and invariable sites model (GTR+G+I).

Fig.2. 

Carbamylphosphate synthetase (CAD) phylogeny of 82 individuals from 37 species, based on Bayesian (BA) and maximum likelihood (ML) analysis. Taxa denoted with ▪ were collected in Minnesota and are described/redescribed in our study. Shaded boxes enclose current or proposed species groups. Values on branches indicate BA posterior probabilities and ML bootstrap support values, respectively. The scale bar indicates branch lengths as number of substitutions per site corrected by the General Time Reversible model with a discrete Gamma distribution and invariable sites model. ** indicates discrepancies between the BA and ML analysis (e.g., Micropsectra roseiventris was sister to remaining in-group taxa in BA, but sister to Paratanytarsus taxa in ML).

NJ and ML analysis yielded very similar results for the analysis of our COI data. Other than minor differences in bootstrap support values, no notable differences were found, and all species clusters were well resolved in both analyses. However, because of recent contention regarding the use of K2P and strictly NJ approaches (Srivathsan and Meier 2012), the results based on the ML analysis are the focus of our discussion. COI ML analyses of the full and the reduced data sets formed strongly supported species clusters. All bootstrap support values for species clusters in the analysis of the full data set were >90% (Fig. 1). A similar trend was seen in the reduced COI ML and NJ/K2P analyses, with all species clusters with bootstrap support > 90%, except Micropsectra sofiae Stur and Ekrem, 2006 (bootstrap support  =  65%).

Phylogenetic analyses

CAD phylogenies from ML and BA yielded similar species-group patterns, with no notable differences other than support values for certain clades (Fig. 2). One such difference was found in the support for the Micropsectra kurobemaculata–Micropsectra lindrothi clade. BA indicated strong support for this clade and placed these 2 species together (posterior probability [pp]  =  95%) and as sister to the remaining species within the notescens group (pp  =  99%), whereas ML analysis indicated much weaker support for these placements (bootstrap support values <50%). We choose to leave these 2 species as members of the notescens group at this point, but recognize that as more species are included in analyses, a more appropriate placement may exist for M. kurobemaculata (Sasa and Okazawa, 1992) and M. lindrothi Goetghebuer, 1931. Another difference between BA and ML analysis was in the placement of Micropsectra roseiventris (Kieffer, 1909). BA placed this species as sister to the remaining in-group taxa (Fig. 2), with a pp  =  0.66. ML analysis, on the other hand, weakly indicated that this taxon is sister to Paratanytarsus (bootstrap support  =  0.44). Further analysis of the relationship between Paratanytarsus and Micropsectra is beyond the scope of our study, but warrants future research.

Analysis of the concatenated COI/CAD data yielded results that were similar to the CAD phylogeny with a few discrepancies (Fig. S2). Most notable is the placement of the lacustris group within the phylogeny. The CAD phylogeny depicts this group as sister to the atrofasciata group, whereas the combined COI/CAD phylogeny indicates this group is sister to the atrofasciata, notescens, and recurvata groups.

Analyses yielded strongly supported atrofasciata and notescens species groups and a paraphyletic recurvata group, results that correspond well with conclusions of Ekrem et al. (2010b) (Fig. 2). Micropsectra lacustris Säwedal, 1975, here designated as a member of the newly re-erected lacustris group was previously considered a member of the recurvata group. We could not confirm the finding of a paraphyletic attenuata group found by Ekrem et al. (2010b) because CAD sequences are not currently available for Micropsectra pharetrophora Fittkau and Reiss, 1999, the specimen accounting for paraphyly in this group. Results for the species described in our study show M. neoappendica n. sp. as sister to M. appendica in the atrofasciata group, M. penicillata n. sp. as sister to M. lacustris in the lacustris group, M. subletteorum n. sp. as sister to M. insignilobus Kieffer, 1924 in the notescens group, and M. xantha as sister to M. recurvata Goetghebuer, 1928 in the recurvata group (Fig. 2).

Type material

Description

Adult male.—

Measurements and ratios in Table 1.

Table 1. 

Measurements and ratios for adult males of Micropsectra species. Lengths are in µm unless indicated. Data are presented as ranges, followed by the mean and the number of specimens observed in parentheses. VR  =  venarum ratio, AR  =  antennal ratio, LR  =  leg ratio, BV  =  “Beinverhältnisse”, BR  =  bristle ratio, SV  =  “Schenkel-Schiene-Verhältnis”, HR  =  hypopygium ratio. Morphological terminology and abbreviations follow Sæther (1980).

Pupa.—

Measurements given in Table 2.

Table 2. 

Measurements of pupal exuviae and pupal structures of Micropsectra species. Lengths in µm unless indicated. Data are presented as ranges, followed by the mean with the number of specimens observed in parenthesis. DC  =  dorsocentral setae. Morphological terminology and abbreviations used here follow Sæther (1980), except taeniae (Langton 1994), is used to describe the flattened setae found on the pupal exuviae.

Taxonomic remarks

Adult males of M. neoappendica are quite similar morphologically to M. appendica, and thus, fit well in the atrofasciata group. Two characters that help to differentiate these 2 species include the comparatively shorter median volsella of M. neoappendica (120–125, 122.5 µm than the 130–144, 137 µm reported in Stur and Ekrem 2006) and the shape of the elevated hump anterior to the anal point base (Fig. 3). In the specimens we have examined, this hump appears less curved in M. neoappendica and typically exhibits a small rounded point or hump at the tip of the elevated region. Setae extending from the gonocoxite and the seta on the stem of the superior volsella appear to be somewhat stronger than those on M. appendica (Fig. 3). The 2 species are best differentiated in the pupal stage. Pupal exuviae of M. neoappendica can be separated by the placement of the dorsocentrals, where the distance between anterior and posterior pairs is approximately equal, as opposed to having a much greater separation of the anterior pair (∼4× greater) than the posterior pair as seen in M. appendica. In addition, the thickness of DC₃ and DC₄ is similar for M. neoappendica, whereas 1 thick and 1 thin setae are noted for M. appendica. The nose of the wing sheath of M. neoappendica is typically quite strong, whereas it is generally weak in M. appendica. Spine patches of TIV and TV further differentiate the species, with long longitudinal spines on TIV of M. neoappendica extending nearly to the posterior end of the tergite and spinule/point patches of TV distinctly shorter than spine patch of TIV, with patch tapering off ∼⅔ the length of the tergite (Fig. 4A). Conversely, longitudinal spines of TIV of M. appendica typically taper off approximately at the mid-point of the tergite, with much shorter spines and spinules extending to the posterior end of the tergite; the spine/spinule patch of TV is subequal in size and pattern to TIV patches (Fig. 4B). Last, pupae of M. neoappendica seem to have fewer anal lobe taeniae than M. appendica (27–33, 30 as compared to 35–42, 38 as described in Stur and Ekrem 2006). Males of M. neoappendica are also quite similar to Micropsectra tusimalemea Sasa and Suzuki, 1999, but can be separated by the differentiating characters for M. appendica described in Stur and Ekrem (2006).

Ecology and distribution.—

Micropsectra neoappendica is known from a shallow stretch of a small coldwater trout stream in northeastern Minnesota, USA.

Micropsectra penicillata Anderson, Stur & Ekrem, new species

(Fig. 5A–H)

Fig.5. 

Micropsectra penicillata n. sp., male. A.—Hypopygium. B.—Pupal abdominal segments II–VI, dorsal view. C.—Anal point, lateral view. D.—Pupal frontal apotome. E.—Posterolateral combs of pupal segment VIII. F.—Pupal thorax. G.—Pupal thoracic horn. H.—Pupal abdominal segments VIII–IX.

Type material

Description

Taxonomic remarks

Micropsectra penicillata exhibits morphological characteristics that tentatively place this species as a member of the Palearctic recurvata group (sensu Säwedal 1981), with the exception of the lack of a digitus, an atypical character for the current recurvata group. Specimens from this species have been collected from various localities throughout North America but have never been formally described. The Sublette collection at UMSP contains adult male specimens of this species from various localities including 1 adult male from Colorado, USA, that was previously identified as M. recurvata because of morphological similarities. Several others from the collection were left undetermined. Adult males can be differentiated from M. recurvata by the absence of the digitus in M. penicillata, and appearance of the median volsella lamellae, which lack the spoon-shaped appearance of many Micropsectra species, including M. recurvata, and instead appear individually as thin leaves that are often held tightly together in a cluster, giving an appearance similar to a thin, pointed, round paintbrush or pointed tail (Fig. 5A), showing most morphological similarity to M. lacustris (see discussion below). Pupae of M. penicillata differ from M. recurvata by the presence of long longitudinal spines on tergite IV (Fig. 5B) as opposed to a short spine patch in M. recurvata. Males of M. penicillata also share morphological similarities with M. lacustris, and phylogenetic analyses based on CAD nuclear gene sequences indicate a sister relationship between these 2 species (Fig. 2). Micropsectra penicillata can be differentiated from M. lacustris based on a distally tapered and laterally curved inferior volsella, the absence of a digitus, and a comparatively longer median volsella (102–137, 122 µm compared to 74–79 µm reported by Säwedal (1975) for M. lacustris). The median volsella lamellar setae of M. penicillata are often held together more tightly than those of M. lacustris, which tend to appear more fanned. Pupae of the 2 species are differentiated based on the pattern of tergite armament; specifically, M. penicillata does not exhibit long longitudinal spines on TV (Fig. 5B) as seen in M. lacustris (see fig. 6 by Säwedal 1975). Further, Säwedal indicates that M. lacustris has 2 L and 2 LS setae on TIV, whereas M. penicillata has 1 L and 2 LS setae.

Ecology and distribution.—

Micropsectra penicillata is currently known from stream localities from Colorado eastward in the USA and from Manitoba and Ontario, Canada.

Micropsectra subletteorum Anderson, Stur & Ekrem, new species

(Fig. 6A–G)

Fig.6. 

Micropsectra subletteorum n. sp., male. A.—Hypopygium. B.—Pupal abdominal segments II–VI, dorsal view. C.—Posterolateral combs of pupal segment VIII. D.—Pupal frontal apotome. E.—Pupal thorax. F.—Pupal thoracic horn. G.—Pupal abdominal segments VIII–IX.

Type material

Description

Taxonomic remarks

Micropsectra subletteorum is closely allied with many species in the notescens group, particularly M. apposita (Walker, 1856), Micropsectra brundini Säwedal, 1979, Micropsectra contracta Reiss, 1965, Micropsectra insignilobus, Micropsectra junci (Meigen, 1818), Micropsectra lindebergi Säwedal, 1976, Micropsectra nigripila (Johannsen, 1905), and Micropsectra notescens (Walker, 1856). The primary characteristic distinguishing adult males of M. subletteorum from the aforementioned species is that M. subletteorum lacks a field of microtrichia on the basal part of the superior volsella and lateral teeth on the anal tergite (Fig. 6A). The following characteristics also are useful for distinguishing males of M. subletteorum from M. brundini, M. insignilobus, and M. lindebergi: a comparatively higher LR_1, and stronger anal point crests (see tables 1–3 in Säwedal 1979). The gonostylus of M. subletteorum also is gradually narrowing towards the distal end (Fig. 6A) and the AR is lower than seen in the 3 aforementioned species. Säwedal (1979) indicates that the low number of sensilla chaetica of M. brundini (0–2) also may help differentiate M. brundini from other notescens group species. However, we have observed specimens of M. brundini from the PHL collection that have up to 4 sensilla chaetica, indicating this character may not be reliable. Micropsectra subletteorum can be further differentiated from M. insignilobus by having a comparatively longer median volsella (68–85 µm compared to 40–59 µm). Micropsectra subletteorum differs from M. junci and M. nigripila by having a comparatively longer digitus that often extends to or past the margin of the superior volsella (Fig. 6A). The digitus of M. subletteorum also exhibits more curvature than that of M. notescens, M. contracta, and M. apposita. Micropsectra subletteorum is further differentiated from M. notescens by lacking the ball-shaped swelling at the tip of the inferior volsella (see fig. 10 by Säwedal 1976) and having a longer median volsella, 67–85, 74 compared to 38–64 µm (Säwedal 1976). The short, blunt-tipped anal point of M. nigripila also differentiates males of the 2 species, and the rounded tip of the superior volsella (Fig. 6A) sets M. subletteorum apart from the more triangular superior volsella of M. contracta and M. apposita.

Pupae of M. subletteorum can be easily differentiated from M. contracta and M. apposita by the absence of semi-long longitudinal spines posterior to the oval spinule patch on TIV (Fig. 6B). The species differs from M. insignilobus and M. lindebergi by exhibiting 2 LS and 1 L setae on TIV rather than 3 LS setae. The thoracic horn of M. subletteorum also is shorter than in these 2 species, M. nigripila, and the parthenogenetic Micropsectra silvesterae Langton, 1998, (∼450 µm as compared to 680 µm for M. insignilobus and M. lindebergi; 624–947, 825 µm for M. nigripila (Oliver and Dillon 1994); and 440–835, 672 µm for M. silvesterae; Säwedal 1976, Oliver and Dillon 1994, Langton 1998). Further, thoracic horn setae start basally for both outer and inner lateral margins, and filament length is ⅓–½ the length of the thoracic horn as compared to ⅕–⅙ for M. insignilobus and M. lindebergi (see Säwedal 1976); the particularly long thoracic horn with short chaetae of both M. nigripila and M. silvesterae also serves to differentiate M. subletteorum from these species (cf. fig. 10 by Oliver and Dillon 1994, fig. 2C by Langton 1998). Micropsectra subletteorum also can be distinguished from M. silvesterae by the presence of only 1 median antepronotal seta as compared to 2 in M. silvesterae (Langton 1998). Micropsectra subletteorum differs from pupae of M. junci by lacking the finger-like prealar tubercle in front of the wing sheath. Micropsectra subletteorum can be further differentiated from M. insignilobus and M. junci in that it has 40–47, 44 anal lobe filaments, as compared to 53–60 in M. insignilobus and 62–73 in M. junci. Pupae of M. subletteorum are perhaps most similar to M. notescens, and few reliable characters have been found to separate the 2. The thoracic horn length and setae arrangement is perhaps the best distinguishing character at present, with the thoracic horn of M. subletteorum typically shorter than M. notescens. Säwedal (1976) reported the average length of the thoracic horn of M. notescens as 660 µm, but we have seen specimens with thoracic horn length as low as 525 µm, which falls within the range we report for M. subletteorum. Thoracic horn setae appear to begin slightly lower on the inner lateral margin for M. subletteorum, covering to as compared to ⅔ for M. notescens.

Ecology and distribution.—

This species is common in small groundwater-dominated trout streams throughout Minnesota and has been found in other lotic and lentic localities in the eastern half of the USA and Canada.

Micropsectra xantha (Roback, 1955)

(Figs 7A–G, 8A, B)

Fig.7. 

Micropsectra xantha, male. A.—Hypopygium. B.—Pupal abdominal segments II–VI, dorsal view. C.—Superior, median, and inferior volsellae. D.—Pupal frontal apotome. E.—Pupal thorax. F.—Pupal thoracic horn. G.—Pupal abdominal segments VIII–IX.

Fig.8. 

A.—Superior volsella, Micropsectra xantha. B.—Anal point, M. xantha. C.—Superior volsella, Micropsectra connexa. D.—Anal point, M. connexa.

Calopsectra (Micropsectra) xantha Roback, 1955: 4

Type material

Description

New Nearctic Micropsectra and cryptic diversity

This small survey of only 5 Minnesota streams, each visited on only 1 sampling occasion, yielded 3 new species of a single chironomid genus, a clear indication that many Nearctic chironomid species are left to be discovered and described. These new species more than double the number of Micropsectra previously known from Minnesota with 24 species now known from the Nearctic. Without the aid of molecular techniques or examination of multiple life stages, 2 of the species described in this work, M. neoappendica and M. subletteorum, probably would have gone unnoticed if identified using existing Palearctic keys and species descriptions, thereby satisfying the definition of cryptic species by Bickford et al. (2007). This fact is particularly important for M. neoappendica, which, to date, has been found only in 2 neighboring, pristine streams in northeastern Minnesota. The 3 other species described/redescribed in our work have been collected by the authors at multiple Nearctic localities or were found in the large set of Micropsectra material collected and curated by Jim Sublette, now at UMSP. To contrast—and to point out potential habitat differences between the appendica/neoappendica cryptic species complex (the Palearctic counterpart of M. neoappendica)—M. appendica is described as having a wide geographical distribution, with collections made from various types of ponds, rivers, brooks, streams, and springs (Stur and Ekrem 2006). This broad distribution suggests different habitat preferences between the 2 species and potential differences in pollution sensitivity. Habitat partitioning by cryptic species, such as suggested here, is not uncommon and is known for various cryptic invertebrates (e.g., Tarjuelo et al. 2001, Folino-Rorem et al. 2009) and vertebrates (e.g., Davidson-Watts et al. 2006). Therefore, considering that cryptic species may exhibit different responses to environmental conditions, use of a single species, rather than a cryptic species-complex, will allow much more accurate biological monitoring and is important for appropriate and accurate conservation and management programs (Bickford et al. 2007).

Molecular species delimitation and phylogeny

As pointed out by Stur and Spies (2011), including species-specific DNA sequences as a standard component of new species descriptions has clear advantages that extend beyond the ability of recognizing cryptic diversity. These advantages include improved consistency in appropriate use of species names (this paper, Stur and Spies 2011), association of multiple or unknown life stages (Zhou et al. 2007, Ekrem et al. 2010a, Pauls et al. 2010, Stur and Ekrem 2011), and an additional set of characters that can be used to identify and establish relationships among species.

Our results emphasize that COI is an effective tool for species delimitation and recognition of cryptic species, but we choose not to discuss relationships among species based on this gene. COI has a weak phylogenetic signal and is unsuitable for inference of species relationships in the Chironomidae (Ekrem et al. 2007, 2010b). The COI trees generated in our analyses did not correspond well with our knowledge of Micropsectra phylogeny and many of the relationships between species had low (<50%) bootstrap support (Figs 1, S3). Furthermore, in the COI phylogenies of both the full (Fig. 1) and reduced (Fig. S3) data sets, the 2 outgroup Paratanytarsus species were nested within the Micropsectra, rendering this genus paraphyletic. Thus, we recognize the power of COI for species delimitation and that it can be quite effective at identifying higher relationships in certain taxa, but we have chosen not to focus on interpreting results from our COI or combined COI/CAD data set to avoid misinterpretations of the data with regard to phylogenetic relationships. Based on results of our analyses and those of others (Winterton et al. 2007, Winkler et al. 2009, Ekrem et al. 2010b), we think that CAD is most reliable for reconstruction and interpretation of phylogenetic relationships for the Chironomidae and some other dipteran families. Results based on our combined COI/CAD data did largely conform to our CAD-only results (Figs 2, S2). The few discrepancies that exist between the 2 methods of analysis probably are the result of conflicting phylogenetic signals from COI that partially mask more reliable species-group hypotheses based on the CAD marker alone. Consequently, all discussion of phylogenetic relationships is based on the nuclear CAD gene, shown by others to have a strong phylogenetic signal for estimating genus-level relationships within Chironomidae (Ekrem et al. 2010b).

Revision and expansion of Micropsectra species groups

Palearctic species of Micropsectra have traditionally been subdivided into species groups to aid in identification, species revisions, and classification. Recent authors have questioned the appropriateness and accuracy of these groups (Giłka 2001, Giłka and Paasivirta 2008, Ekrem et al. 2010b). Present molecular and morphological evidence shows a well supported atrofasciata group (e.g., Ekrem et al. 2010b, Stur and Ekrem 2006), but lines of evidence supporting monophyletic attenuata, recurvata, and notescens groups are less distinct (e.g., Giłka 2001, Giłka and Abramczuk 2006, Giłka and Paasivirta 2008, Ekrem et al. 2010b), particularly considering the recent synonymy of Krenopsectra and Parapsectra with the Micropsectra (Ekrem et al. 2010b). Additional problems with the groups become apparent when attempting to place Nearctic species in the Palearctic-derived classification system.

Most Nearctic Micropsectra have not yet been described (Reiss 1995, our study). Many species awaiting description and some that have been described do not conform to the current species-group classification, which is based primarily on the adult male (e.g., Micropsectra spinigera Reiss, 1995, Micropsectra borealis (Kieffer, 1922), Micropsectra radialis Goetghebuer, 1939, and the parthenogenetic species Micropsectra sedna Oliver, 1976, and M. silvesterae). This situation highlights the need for a major revision of the genus and expanded species-group system.

In our study and in the study by Ekrem et al. (2010b), molecular evidence indicated a paraphyletic recurvata group, with former recurvata group members M. recurvata and M. lacustris falling into genetically distinct clades (Figs 2, S2; figs 3, 4 by Ekrem et al. 2010b). If chironomid taxonomists choose to retain current species groups, perhaps Säwedal's original classification of a separate lacustris group (Säwedal 1975) was more appropriate than his later placement of M. lacustris in the recurvata group (Säwedal 1981). Molecular evidence from our CAD data set suggests placement of M. penicillata as sister species to M. lacustris. Micropsectra penicillata shares similarities with the recurvata group, as pointed out by preliminary classification of the undescribed species by J. Sublette in his personal collection (now part of UMSP), but distinct differences, which we pointed out in the taxonomic remarks for the species (see above) also exist. These morphological differences, combined with molecular evidence (Figs 2, S2), indicate an alternative grouping for the species.

A further drawback of the existing species-group classification is that it is based primarily on the morphological characteristics of the adult male, and it was founded before molecular data were commonly used in systematics. The current species-group classification does not extend to immature stages (Pinder and Reiss 1986). Most chironomid descriptions historically have been based on the adult male, but descriptions of immature life stages are becoming increasingly important, particularly for purposes of biological monitoring and studies of species diversity (e.g., Ekrem et al. 2010a, Zhou et al. 2010).

Chironomid pupal exuviae are a simple and effective method for assessing species diversity (e.g., Ferrington et al. 1991, Calle-Martinez and Casas 2006, Raunio et al. 2007, Ruse 2011), particularly since descriptions of pupal exuviae have become more common and readily available. We think that this life stage may provide important features that, combined with characters of the adult male and molecular data, may help resolve future species-group concepts. For example, when examining Fig. 2 (our study) and fig. 3 by Ekrem et al. (2010b), the exuviae of all members of the atrofasciata group that have described exuviae have long longitudinal spine patches on abdominal tergite IV. Conversely, exuviae known for the notescens group, with the exception of M. contracta and M. apposita, have short anterior patches consisting of spinules and points. Micropsectra contracta and M. apposita have semilong spines (none long enough to be synonymous with spine-type from the atrofasciata group) extending partway down tergite IV. Members of a reduced recurvata group, including M. recurvata and M. xantha, also fit the exuviae generalization of the notescens group, forming a recurvata–notescens clade. Adult male morphology clearly separates recurvata group members from the notescens group members, but we have not yet identified pupal characters that can be used to separate them. Furthermore, if the lacustris group is re-established with M. penicillata as a member, this group fits better with a combined lacustris–atrofasciata clade (Fig. 2). As with the recurvata–notescens clade, members of the lacustris–atrofasciata clade have similar spine patch patterns on the exuviae, with known species in this clade exhibiting long longitudinal spine patches on tergite IV. Adult males separate the lacustris group from the atrofasciata group, but no definite pupal characters distinguish the groups at this point. The results of our analysis of the concatenated data placed the lacustris group as sister to the atrofasciata, notescens, and recurvata groups (Fig. S2) rather than to the atrofasciata group. Placement of this group should become clearer as more species and markers are added to the data set. However, most importantly, all species groups we propose are strongly supported in both CAD and combined COI/CAD analyses.

To expand our ideas further and to express the need for a revision and expansion of current species groups, M. radialis and M. borealis form a separate group in our phylogeny (Fig. 2) and that of Ekrem et al. (2010b). Moreover, the morphologies of the adult male and pupa are markedly different from those of other Micropsectra species mentioned here. Thus, these species do not to conform to any of the current species groups or the proposed system described above. Säwedal (1981) mentioned constructing a borealis group, based on redescription of M. borealis (Säwedal and Willassen 1980), but apparently, no subsequent treatment of the group was completed. Säwedal and Willassen (1980) suggested that M. borealis and M. radialis form a monophyletic clade with the atrofasciata group based largely on similarities of the superior volsella. We dispute placement of these species in the atrofasciata group. Instead, we propose erecting the borealis group, which currently includes M. borealis and M. radialis as members based on genetic similarity, morphological similarities of the male genitalia, and reasons outlined by Stur and Ekrem (2006).

The former Parapsectra and Krenopsectra taxa were recently synonymized with Micropsectra (Ekrem et al. 2010b). When considering the former Parapsectra, all taxa less one form a monophyletic group. Only M. mendli (Reiss, 1983) falls outside of this group and forms its own clade (Fig. 2). Therefore, we suggest formation of a nana group that currently includes Micropsectra nana (Meigen, 1818), Micropsectra bumasta (Giłka & Jażdżewska, 2010), Micropsectra chionophila (Edwards, 1933), and Micropsectra uliginosa (Reiss, 1969).

We are hesitant to place Micropsectra acuta Goetghebuer, 1934 firmly in the attenuata group (Fig. 2). The adult male of M. acuta shows similarities to other species in the attenuata group, and molecular results clearly indicate a linkage, but pupal morphology is somewhat different in that M. acuta lacks a laterally curved patch of longitudinal spines on tergite III as seen in the other attenuata-group species. In addition, M. acuta has no lateral taeniae on any tergite and has only lateral setae (a trait unusual for Micropsectra), and tergite III has only 4 setae rather than the usual 5. These differences lead us to think that as more Micropsectra species are described and sequenced for appropriate markers, M. acuta could form a new species group in Micropsectra, probably with Micropsectra fallax Reiss, 1969 and Micropsectra nohedensis Moubayed & Langton, 1896, both previously placed in Krenopsectra.

Last, our analyses suggest that the placement of M. roseiventris is unique with respect to the rest of the Micropsectra, perhaps rendering the genus paraphyletic as currently classified (Figs 2, S2). Further research at the molecular and morphological levels is needed to confirm the placement and stance of this species in relation to the rest of the Micropsectra.

Our results clearly suggest a change in the current species-group organization in Micropsectra. They indicate that a combination of morphological characters from adult male and pupal exuviae and molecular data from protein-coding markers is an appropriate approach to divide large Chironomidae genera into taxonomically useful groups that reflect likely genealogical relationships.

Conclusions

Our research demonstrates strong concordance between morphological species concepts, DNA barcodes, and CAD sequence data. We have used partial CAD sequences to confirm the species-divergence patterns detected with COI DNA barcodes. COI barcodes were not robust enough to relay information about relationships between species of Micropsectra, but concordance between CAD results and adult and pupal morphology was very strong. This strong concordance of CAD and morphological data is especially important considering that morphological phylogenetic analyses are of limited value for assessing species-level phylogenies of Chironomidae because of a high level of homoplasy in morphological data sets (Ekrem 2003, Stur and Ekrem 2006). As Micropsectra species continue to be described, revealing new morphological character combinations, and as the DNA barcode database for the genus is expanded, particularly with regard to Nearctic species and those in other less-studied localities, the taxonomic framework for Micropsectra will have to be reconsidered. This need for reconsideration probably will be a trend throughout other major genera of Chironomidae and, perhaps, other large families of aquatic insects. We anticipate further shifts and alterations in the current species-group classification system and a continued need for chironomid taxonomists to revisit the appropriateness of these groups. With recent and forthcoming advances in molecular taxonomy, taxonomic work that combines forces of molecular and morphological methods will rapidly advance our understanding of phylogenetic relationships in species-rich genera, such as Micropsectra, emphasizing the need for taxonomists to use all data available and to be well versed in both molecular and morphological techniques.

Taxonomic note: Micropsectra brundini was described from oligotrophic lakes in Greenland (Säwedal 1979) and is morphologically similar to M. lindebergi and M. insignilobus in both the adult male and pupal stages; females of this species are not yet known. The type material of this species could not be located at ZSBS (M. Spies, ZSBS, personal communication). We attempted unsuccessfully to locate type material of this specimen at various other museums. However, we have examined material identified as M. brundini in the collections of Peter Langton, Claus Lindegaard, and the Sublette material at UMSP and confirm that this species is morphologically very similar to M. lindebergi and M. insignilobus. Values for the AR seem to be the most consistent and obvious differences used to separate M. brundini from M. lindebergi (1.75–2.14 and 1.31–1.58, respectively), and a longer median volsella aids in differentiation from M. insignilobus (63–78 vs 40–59 µm). Säwedal (1979) lists other characters that may aid in differentiation, but overlap between these species, particularly M. brundini and M. insignilobus, is considerable. Resolution of these taxonomic issues is beyond the scope of our paper, but we think sampling of localities known to produce M. brundini and subsequent molecular analyses of fresh specimens could help assess relationships among these species.