Integrative taxonomy of Metrichia Ross (Trichoptera: Hydroptilidae: Ochrotrichiinae) microcaddisflies from Brazil: descriptions of twenty new species

Metrichia is assigned to the Ochrotrichiinae, a group of almost exclusively Neotropical microcaddisflies. Metrichia comprises over 100 described species and, despite its diversity, only one species has been described from Brazil so far. In this paper, we provide descriptions for 20 new species from 8 Brazilian states: M. acuminata sp. nov., M. azul sp. nov., M. bonita sp. nov., M. bracui sp. nov., M. caraca sp. nov., M. circuliforme sp. nov., M. curta sp. nov., M. farofa sp. nov., M. forceps sp. nov., M. formosinha sp. nov., M. goiana sp. nov., M. itabaiana sp. nov., M. longissima sp. nov., M. peluda sp. nov., M. rafaeli sp. nov., M. simples sp. nov., M. talhada sp. nov., M. tere sp. nov., M. ubajara sp. nov., and M. vulgaris sp. nov. DNA barcode sequences (577 bp of the mitochondrial gene COI) were generated for 13 of the new species and two previously known species of Metrichia resulting in 64 sequences. In addition, COI sequences were obtained for other genera of Ochrotrichiinae (Angrisanoia, Nothotrichia, Ochrotrichia, Ragatrichia, and Rhyacopsyche). DNA sequences and morphological data were integrated to evaluate species delimitations. K2P pairwise distances were calculated to generate a neighbor-joining tree. COI sequences also were submitted to ABGD and GMYC methods to assess ‘potential species’ delimitation. Analyses showed a conspicuous barcoding gap among Metrichia sequences (highest intraspecific divergence: 4.8%; lowest interspecific divergence: 12.6%). Molecular analyses also allowed the association of larvae and adults of Metrichia bonita sp. nov. from Mato Grosso do Sul, representing the first record of microcaddisfly larvae occurring in calcareous tufa (or travertine). ABGD results agreed with the morphological delimitation of Metrichia species, while GMYC estimated a slightly higher number of species, suggesting the division of two morphological species, each one into two potential species. Because this could be due to unbalanced sampling and the lack of morphological diagnostic characters, we have maintained these two species as undivided.

and hence the new names contained in the electronic version are effectively published under that Code from the electronic edition alone. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix http://zoobank.org/. The LSID for this publication is: urn:lsid:zoobank.org:pub:D8D4049E-494B-4A30-92AC-F8F42D2B54B9. The online version of this work is archived and available from the following digital repositories: PeerJ, PubMed Central and CLOCKSS.

Sequence editing, alignment, and analyses
Forward and reverse sequences of each sample were assembled and manually edited in 3).
COI sequences were used to explore putative species limits with four different methodologies: (1) lineages recovered in neighbor-joining tree; (2) lineages recovered with Bayesian Inference; (3) ABGD; and (4) GMYC. The neighbor-joining tree was calculated in MEGA 6 using Kimura 2-Parameter (K2P) distances (Kimura, 1980), with partial deletion of missing information. Although the use of K2P distances in DNA barcoding is debated (Srivathsan & Meier, 2012), to allow comparison with previous works we also used this evolutionary model because it is frequently used in studies of species delimitation based on COI sequences. Branch support of neighbor-joining tree was assessed with 1,000 pseudoreplicates of non-parametric bootstrap (Felsenstein, 1985).
BI analysis was conducted with MrBayes v. 3.2.2 (Ronquist et al., 2012) with four independent runs, each one with four MCMC chains running for 50,000,000 generations, with sample frequency of 5,000. Convergence of sampled parameters was checked in Tracer v. 1.5 (Rambaut & Drummond, 2007) and the first 10% of sampled trees and parameters discarded as burnin. GTR+I+G was the best fit model selected by Akaike Information Criterion (AIC) with jModeltest v. 0.1.1 (Posada, 2008) and it was applied in BI analysis in MrBayes. Branch support was assessed by posterior probability (PP), presented on a 50% majority consensus tree.
ABGD analysis was run using the on-line version available in http://wwwabi.snv.jussieu.fr/public/abgd/, where the COI alignment was uploaded. The analysis was conducted with the following settings: Pmin=0.001; Pmax=0.1; steps=20; relative gap width=1.0, also based on K2P model. This method statistically infers the DNA barcode gap in a single locus alignment, partitioning the data based on this gap in putative species (Puillandre et al., 2012).
The GMYC analysis (Pons et al., 2006;Fujisawa & Barraclough, 2013) was performed in R (R Development Core Team 2010) using the SPLITS package (Ezard et al., 2009) with singlethreshold method. Basically, the method estimates branching patterns on an ultrametric tree, identifying the most likely transition point from coalescent to speciation branching. The ultrametric tree used here was obtained with BEAST v. 1.8 (Drummond et al., 2012) under a relaxed uncorrelated molecular clock (Drummond et al., 2006). The node including Ochrotrichia species was calibrated based on fossil evidence with a lognormal distribution offset at 20 mya and log(mean) = 2.8 to represent the possible range of 20-140 mya (Wells & Wichard, 1989); and the divergence of Ochrotrichiinae was calibrated based on Malm et al. (2013) with a normal distribution with mean 82.17 ± 12 mya. The BEAST analysis ran for 200,000,000 generations, sampled every 10,000 generations. Convergence was verified with Tracer and a maximum credibility tree was written using TreeAnotator, discarding the first 10% as burnin.   In all analyses performed using DNA barcode, Metrichia larvae collected in calcareous tufa were consistently associated with adult males of Metrichia bonita sp. nov. Therefore, in the following section, we describe these larvae within that species. Etymology. The species is named in allusion to the pointed apices of inferior appendages (from Latin, "acumin-" = "pointed").

Remarks.
This new species belongs to the penicillata group based on: (1) internal pouches between abdominal segments IV and V; (2) setal brushes on segments V, VI, and VII; and (3) phallus with two subapical spines. The male genitalia and complex abdominal modifications resemble M. penicillata (Flint, 1972) and M. trigonella (Flint, 1972). These three species have inferior appendages with acute apices in lateral view; phallus with two subapical spines; and abdominal terga with brushes of very long and stout setae. The new species can be distinguished We were not able to obtain COI sequences for individuals from Ceará State, so the five sequences analyzed belong to specimens from Alagoas State, which shared the same haplotype.

Remarks.
This new species is another member of the penicillata group based on internal pouches between segment V and VI and the long subapical spines of the phallus. The new species shares similarities of the male genitalia with M. biungulata (Flint, 1972)

Manuscript to be reviewed
Holzenthal, 2003 particularly the rounded aspect of inferior appendages, but can be easily distinguished from those species by the absence of tooth-like processes on inferior appendages; more elongate preanal appendages; and dorsal hook only slightly downturned in lateral view. Labrum with pair of stout setae (Fig. 7F). Mandibles with inner margin sinuous and darkened.
Thoracic legs brown, short and stout, almost the same size of each other. Foreleg with stout setae; tibia with a posteroventral lobe with a spine-like seta ( Paratypes. Same data as holotype, 3 males (DZRJ), 1 male (MNRJ).
Etymology. This species is named in reference to the type locality ( Fig. 8), the municipality of Bonito in the state of Mato Grosso do Sul. In Portuguese, the word "bonita" (the feminine form) means "beautiful". Manuscript to be reviewed  Manuscript to be reviewed Paprocki et al. (2003) discussed the role of Smicridea travertinera Paprocki, Holzenthal & Cressa, 2003 in calcareous tufa formation (travertine). According to these authors, larvae of that species interfere in the deposition and erosion of the calcareous substrate by their netbuilding activities (Paprocki et al., 2003). Cyanobacteria and diatoms are known to participate in travertine formation, but the role played by macroinvertebrates is poorly understood (Drysdale, 1998;1999). It is possible that cases of Metrichia bonita sp. nov. are impregnated passively with calcareous particles, but as commented by Drysdale (1999) for other aquatic insects, they could be important in travertine biogenesis by producing new nucleation sites or eroding other ones.

Remarks. Metrichia bonita
Metrichia bonita sp. nov. is the only microcaddisfly known to inhabit ( elongate and bearing very long setae (Fig. 9B). Subgenital plate apparently absent. Tergum X membranous and with shallow U-shaped incision (Fig. 9B). Phallus tubular, elongate and slender, slightly constricted mesally; without spines, but with a sclerotized process arising from a subapical constriction; apex rounded and folded; ejaculatory duct sclerotized, straight and not protruding apically (Fig. 9D). Etymology. The species is named in allusion to the river where the holotype was collected.

Holotype
Remarks. This new species can be assigned to the patagonica group because of the absence of curved spines of the phallus. The general aspect of the male genitalia resembles M. patagonica (Flint, 1983), M. pernambucana, and M. pseudopatagonica Bueno-Soria & Holzenthal, 2003.
Metrichia bracui sp. nov. differs from these species and others in the group specially by the phallus bearing a sclerotized process on a constricted region.
The four COI sequences generated for M. bracui sp. nov. were from specimens collected in a locality at Parque Nacional do Itatiaia, Rio de Janeiro, Brazil. The highest pairwise intraspecific divergence between sequences was 0.4%, and the lowest interspecific divergence was 21.4% between M. bracui sp. nov. and Angrisanoia cebollati (Angrisano, 1995). Until now, there have been no formal studies on the relationships among species or genera included in Ochrotrichiinae. However, to infer phylogenetic hypothesis for the entire subfamily is beyond the scope of this work, and also much more data and taxa sampling are necessary to generate robust hypotheses. Manuscript to be reviewed palpus 3-articulated. Mesoscutellum with transverse suture. Metascutellum subtriangular.
Anterior femur without processes.  10A); in lateral view narrower anteriorly than posteriorly (Fig. 10C). Inferior appendage covered by long setae, subtrapezoidal in ventral view (Fig. 10A), apex oblique and projected mesad into a large process bearing a stout spine-like setae. Dorsal hook short and straight; in lateral view, truncate and broader apically (Fig. 10C). Preanal appendage short, truncate and bearing very long setae (Fig. 10B). Subgenital plate apparently absent. Tergum X membranous and truncate (Fig. 10B). Phallus tubular, elongate and slender, slightly constricted mesally; with two curved subapical spines, one short and another long; apex emarginate; ejaculatory duct sclerotized and not protruding apically (Fig. 10D). Etymology. This species is named in reference to the stream where type specimens were collected.

Remarks.
This new species has very distinctive male genitalia. Based on the dorsoapically produced inferior appendages, it resembles M. lenophora (Flint, 1991 Manuscript to be reviewed Anterior femur without processes.  11A); in lateral view narrower anteriorly than posteriorly (Fig. 11C).
Phallus tubular, elongate and slender, slightly constricted mesally and with a median process; with two short subapical spines; apex rounded and sclerotized; ejaculatory duct sclerotized, straight and not protruding apically (Fig. 11D). Etymology. The new species name is an allusion to the rounded and simple inferior appendages, derived from the Latin, "circuli-" and "form" meaning "rounded shape".

Holotype. BRAZIL: Rio de
Remarks. This species has simple male genitalia and abdomen with only brushes of long setae on segments VI and VII. General aspect of the male genitalia is similar to M. riva (Bueno-Soria, 1983) and M. quadrata (Flint, 1972), particularly their inferior appendages short and subrectangular and phallus with two subapical spines. However, M. circuliforme sp. nov. can be easily distinguished from M. quadrata by the absence of internal sacs in the abdomen. It can be distinguished from M. riva by the elongate preanal appendages and phallus with subequal hook spines subapically.
We obtained seven COI sequences for M. circuliforme sp. nov. and although all of them came from specimens collected at the same locality, intraspecific divergences were relatively high, reaching 3.5%. Besides that, GMYC estimated two species for these sequences instead of one. Re-analysis of the morphology of these specimens did not reveal any conspicuous variation that could justify splitting this species into two taxonomic groups. Compared to other studies using DNA barcodes of caddisflies, this genetic distance is still low, for example Pauls et al. GMYC is known to be more sensitive to geographic range coverage and/or other sampling schemes, resulting in oversplitting (Lohse, 2009;Talavera et al., 2013). Therefore, we consider Manuscript to be reviewed nov. were 18.4%, and again, these two species are very distinct based on morphological features and apparently are not even closely related to each other (Fig. 2). Manuscript to be reviewed  12A); in lateral view, narrower anteriorly than posteriorly (Fig. 12C).

Paratypes. Same data as holotype, 3 males (DZRJ), 3 males (MNRJ).
Etymology. The specific name is a reference to the very short inferior appendage; in Portuguese "curta" means "short".  Manuscript to be reviewed Anterior femur without processes. Tibial spur formula 1-3-4. Wing venation reduced in both wings. Abdomen without modifications; segment VII bearing specialized setae dorsally.
Tergum X membranous and rounded (Fig. 13B). Phallus tubular, elongate and slender, slightly constricted mesally; with a stout subapical spine; apex rounded and sclerotized; ejaculatory duct sclerotized, sinuous, and protruding apically (Fig. 13D).   Manuscript to be reviewed Although more than 10 specimens of M. farofa sp. nov. were submitted to DNA extraction and many attempts of COI amplification via PCR were conducted, but we were not able to obtain sequences of this species, even using recent material collected after 2013. Anterior femur with small acute apical process. Tibial spur formula 1-3-4. Wing venation reduced in both wings. Abdominal segment VII bearing internal pouches in anterior area. Ventromesal process on segment VII absent. Segment VIII shorter ventrally than dorsally. Male genitalia.

Metrichia forceps
Segment IX reduced dorsally; sternum subpentagonal (Fig. 14A); in lateral view, narrower anteriorly than posteriorly (Fig. 14C). Inferior appendage covered by long setae, subrectangular in ventral view (Fig. 14A); in lateral view, with posterior margin excavated and with two acute and sclerotized process (Fig. 14C). Dorsal hook long, almost reaching the inferior appendage campana (Flint, 1968), M. similis (Flint, 1968), and M. continentalis (Flint, 1972), particularly by their inferior appendages with excavate posterior margins, forming two pointed processes, one ventral and another dorsal. The new species can be distinguished from the others by its deeply notched tergum X; dorsal hook of inferior appendages elongate and downturned, and phallus apex bearing two small spines and a sclerotized flap surrounding the protruding ejaculatory duct. Anterior femur without processes. Tibial spur formula 1-3-4. Wing venation reduced in both wings. Abdominal segment V ventrally with a mesal brush of long setae; segment VI with tergum as a sclerotized triangular plate surrounded by specialized setae (Fig. 26B), internally with pair of internal pouches; segment VII bearing specialized setae ventrally and dorsally (Fig. 26B).

Manuscript to be reviewed
Etymology. This species is named in reference to Serra de Itabaiana, Sergipe, where the holotype was collected.

Remarks.
This new species appears to be a member of the campana group because of internal pouches between abdominal segments VI and VII and pair of long subapical spines on phallus, but it lacks the acute process on the mesal area of phallus. Anterior femur with small acute apical process. Tibial spur formula 1-3-4. Wing venation reduced in both wings. Abdominal segment IV with dorsal area expanded posteriorly bearing stout setae; segment VI with stout and striate setae (Fig. 26C); segment VII with stout and striate setae (Fig.   26C). Ventromesal process on segment VII absent. Segment VIII shorter ventrally than dorsally. Inferior appendage covered by long setae, short and rounded, with apex slightly excavated (Fig.   20A); in lateral view, rounded (Fig. 20C). Dorsal hook long, more than half length of inferior appendage; in lateral view, downturned (Fig. 20C). Preanal appendage elongate, but shorter than half length of inferior appendage, and bearing stout and striate setae (Fig. 20B). Subgenital plate apparently absent. Tergum X membranous and truncate (Fig. 20B). Phallus tubular, elongate and slender, slightly constricted mesally, with a median process; with two long, curved, subapical spines, and a membranous lobe; apex rounded and sclerotized; ejaculatory duct sclerotized, sinuous, and not protruding apically (Fig. 20D).

Holotype. BRAZIL
Etymology. This species is named in reference to the simple aspect of the male genitalia and abdomen, without modifications and processes seen in other Metrichia species.   23A); in lateral view, narrower anteriorly than posteriorly (Fig. 23C). Inferior appendage elongate, apex with an acute projection; with a deep C-shaped notch in ventral view (Fig. 23A); in lateral view, with an acute projection (Fig. 23C). Dorsal hook short, less than half length of inferior appendage; in lateral view, downturned (Fig. 23C). Preanal appendage short, rounded and bearing very long setae (Fig. 23B). Subgenital plate apparently absent. Tergum X membranous and truncate (Fig. 23B). Phallus tubular, elongate and slender, slightly constricted mesally, with a median process; with two curved subapical spines, one short and another long; apex rounded and sclerotized; ejaculatory duct sclerotized, straight and protruding apically (Fig. 23D). in lateral view, narrower anteriorly than posteriorly (Fig. 24C). Inferior appendage covered by long setae, apex rounded; elongate and narrow in ventral view (Fig. 24A); in lateral view, rounded (Fig. 24C). Dorsal hook short, less than half length of inferior appendage; apex downturned; basally with a wide and sclerotized projection; in lateral view, C-shaped (Fig. 24C).

Holotype. BRAZIL
Preanal appendage elongate, but shorter than half length of inferior appendage, and bearing stout and striate setae (Fig. 24B). Subgenital plate apparently absent. Tergum X membranous and truncate (Fig. 24B). Phallus tubular, elongate and slender, slightly constricted mesally; with two long, curved, subapical spines; apex rounded and folded; ejaculatory duct sclerotized, sinuous, and protruding apically (Fig. 24D). different populations as the same species, with K2P intraspecific divergences up to 4.8%. This is the highest intraspecific divergence found in our work, but this value is still lower than that observed among species in other caddisfly groups . GMYC analyses grouped sequences into two 'species', but these groups were not related to their geographic distribution: one group included samples from Rio de Janeiro and Minas Gerais and the other group, samples from Rio de Janeiro, Minas Gerais, and Goiás. The broader geographical and specimen sampling of this species, which is directly associated with its higher intraspecific divergences, may explain the oversplitting by the GMYC method, as discussed by Talavera et al. (2013).
Although such wide distribution is not common for Metrichia species, other microcaddisflies can show continental distributions (e.g., Oxyethira tica). Because the knowledge about Neotropical microcaddisflies is very poor, this pattern may be more common than currently thought.

DISCUSSION
Although GMYC analysis overestimated the number of Metrichia species in our study (suggesting the split of M. circuliforme sp. nov. and M. vulgaris sp. nov. each into two 'species'), COI sequences strongly corroborated species limits previously defined based on morphological features. In general, COI sequences of caddisflies show a robust 'barcoding gap' , making this molecular marker appropriate as a source of additional information to corroborate species delimitation or associations of different life stages.
Microcaddisflies are extremely diverse and poorly known, and when associated with the morphological data, the use of molecular information can result in a more robust taxonomy for this group. Although methods such ABGD and GMYC should not be used alone to determine 'species', they are useful tools to identifying 'potential species' (Puillandre et al., 2012;Talavera et al., 2013), especially in very diverse groups and/or with dubious morphology-based identification. The wide distribution of Metrichia vulgaris sp. nov. could indicate the existence of different cryptic species, however morphology and barcode data agreed to define this group as a single species. Although GMYC overestimated the number of Metrichia species in our analysis, we consider this method an important tool for preliminary distinction when taxonomic information is poor.
Based on the presence of abdominal modifications, such as internal pouches, external sclerotized plates, and brushes of long setae, and features of male genitalia, six species groups have been proposed for Metrichia (Flint, 1972;Bueno-Soria & Holzenthal, 2003). These abdominal modifications usually arise from segments V, VI, and VII, and, in general aspect, are very distinctive from each other (Fig. 26), possibly representing non-homologous structures.
Nevertheless, most of the species described herein fit in these previously defined groups. Present analyses of COI sequences ( Fig. 2 and Supp. 4) recovered neither the monophyly of Metrichia nor any of the species groups tested (aberrans, campana, and nigritta groups). However, these relationships need to be analyzed further with better taxon and molecular marker sampling.

Bayesian inference estimated tree
Consensus phylogram (50% majority-rule) from BI analyses of COI sequences (mean lnL= -5464.29) of Metrichia and related microcaddisflies. Values displayed near branches are posterior probabilities.

K2P pairwise divergences
Pairwise K2P divergences of COI sequences of Metrichia and related microcaddisflies.