Integrative taxonomy supports two new species of Chimarra Stephens, 1829 from Brazil (Trichoptera: Philopotamidae)

Chimarra Stephens, 1829 is the largest genus of the Philopotamidae with about 930 species and cosmopolitan distribution. Recent taxonomic revisions have subdivided the genus into four subgenera: Chimarra, Curgia Walker, 1860, Chimarrita Blahnik, 1997, and Otarrha Blahnik, 2002, the last three restricted to the New World. In this paper, we describe and illustrate two new species of Chimarra from Brazil, C. (Otarrha) paraodonta sp. nov. from Rio de Janeiro State and C. (Chimarrita) truncata sp. nov. from Amazonas and Pará states. Partial sequences of cytochrome oxidase subunit I (COI, DNA barcodes) were generated and integrated with morphological evidence to delimit the new species and evaluate their phylogenetic relationships within the genus. A maximum likelihood analysis of 48 COI sequences representing 19 species of Chimarra corroborated their subgeneric assignment based on morphology and highlighted their putative sister species. Both new species showed high K2P divergences when compared to their sister species: Chimarra (O.) paraodonta sp. nov. and C. (O.) odonta (17.4–21.3%) and Chimarra (C.) truncata sp. nov. and C. (C.) simpliciforma (20.0–21.3%). These distances are comparable to the range of interspecific distances calculated for the whole genus (13.6–22.7%), adding support to their description as new species. This analysis was especially important because of the high morphological similarity of C. paraodonta sp. nov. and C. odonta. Finally, analysis of the sequences of Chimarra odonta suggests that the nominal species may represent a complex of cryptic species with high intraspecific divergences (up to 18.1%), with at least two of those lineages co-occurring with C. paraodonta sp. nov. at Parque Nacional do Itatiaia.


Introduction
Philopotamidae comprise approximately 1,500 species in 26 extant genera distributed worldwide, but most of the diversity is found in the tropics ; Morse et al. 2019). Currently, the family is divided into three subfamilies: Rossodinae, with 16 species solely in one genus, endemic to Madagascar; the cosmopolitan Chimarrinae, with about 950 extant species distributed into three genera; and the Philopotaminae, with more than 400 species in 18 genera Morse 2021;Cartwright 2020). Chimarra Stephens, 1829 is one of the most diverse caddisfly genera with about 930 species (Kjer et al. 2014;Cartwright 2020); of these about 260 species occur in the Neotropical Region and 54 species in Brazil (Vilarino and Calor 2015;Holzenthal and Calor 2017;Dumas and Santos 2021). Chimarra adults range from 3−8 mm in size and are generally black or dark-brown colored, being commonly found near a wide variety of flowing waters, where larvae usually construct tubular, silken retreats attached beneath or between large substrates, like logs or rocks (Blahnik 1997;Holzenthal and Calor 2017). They can be characterized by having tibial spur formula 1-4-4 and presence of an anal loop on hind wing, in which the 2A vein is looped to join 1A vein (Blahnik 1998). Currently, Chimarra is subdivided into four subgenera and the New World species have been the subject of relatively recent taxonomic revisions (Blahnik 1997(Blahnik , 1998(Blahnik , 2002Flint 1998). The nominotypical subgenus Chimarra is the largest with about 550 species known from all zoogeographical regions, being more diverse in the Oriental Region with 321 species. In the New World approximately 100 species are recorded, being placed in 19 species groups, about half of them including only one or two species (Blahnik 1998;Kjer et al. 2014;Holzenthal and Calor 2017). The subgenera Curgia Walker, 1860, Chimarrita Blahnik, 1997, and Otarrha Blahnik, 2002 are confined to the New World, especially in the Neotropics (Blahnik and Holzenthal 2012). Curgia has 93 species distributed in 16 species groups ranging from the southwestern portion of the United States through Central and South America, extending into the Lesser and Greater Antilles (Flint 1998;Santos and Nessimian 2009). Chimarrita (21 spp.) and Otarrha (33 spp.) are endemic to the Neotropics, being widespread in Central and South America, including the Antilles (Blahnik 1997(Blahnik , 2002Blahnik and Holzenthal 2012;Vilarino and Calor 2015;Desidério et al. 2018;Camargos 2016).
Recent phylogenetic analyses inferred from molecular data (Kjer et al. 2014;Wahlberg and Johanson 2014) support the monophyly of Chimarra, as well as that of its subgenera, but minor incongruences among some lineage relationships were found when compared with morphological data proposed in subgeneric revisions (Flint 1998;Blahnik 1997Blahnik , 1998Blahnik , 2002. Wahlberg and Johanson (2014) performed a biogeographical analysis pointing to an early Cretaceous origin, circa 138 million years ago, in the Neotropical region, followed by subsequent radiations into the Oriental, Palearctic, and Australasian regions, with several independent colonization events into the Afrotropical region. Both works are of great value as they provide unprecedented amounts of nuclear and mitochondrial sequence data for a large and worldwide distributed caddisfly genus.
In the past few decades, molecular techniques have been used widely for species separation and identification (Vogler and Monaghan 2006) and have become increasingly popular for taxonomic studies (Doyle 1992;Soltis et al. 2000;Moore et al. 2010). The DNA barcode, a short standard DNA region of the mitochondrial cytochrome oxidase I gene (COI), is the most commonly used marker for animal species identification, including several freshwater insect groups (e.g., Ephemeroptera -Ball et al. 2005;Trichoptera -Zhou et al. 2016;Simuliidae -Rivera and Currie 2009;Ephemeroptera, Plecoptera and Trichoptera -Morinière et al. 2017). DNA barcoding is not used only for species identification, but also for evolutionary, ecological, and conservation research (Hebert et al. 2003;Valentini et al. 2009;Leese et al. 2018). In addition, in recent years the use of DNA barcoding has increased the discovery of cryptic species in different taxa, habitats, and regions (e.g., Pfenninger and Schwenk 2007;Zakšek et al. 2009;Pauls et al. 2010;Jackson et al. 2014;Weiss et al. 2014).
Although the use of COI sequences has become common in taxonomic studies at the species level, their use is still rare for Neotropical caddisflies. The first comprehensive work using this tool in the Neotropics was made by Pauls et al. (2010), where authors corroborated two new species and revealed the existence of cryptic diversity of Smicridea (Smicridea) McLachlan, 1871 in Chile. After that, only few works using DNA barcodes were carried out in order to associate different life-stages (e.g., Santos et al. 2016;Barcelos-Silva et al. 2018) and to evaluate species delimitation in integrative taxonomy efforts (e.g., Santos et al. 2016;Vilarino et al. 2019).
In this study two new species of Chimarra are described and illustrated in the subgenera Chimarrita and Otarrha through an integrative taxonomic approach based on adults from the Amazon and Atlantic Forest biomes of Brazil. In addition, DNA barcodes were gen-erated and integrated in order to evaluate species delimitation and relationships of the new taxa within the genus Chimarra.

Material examined
Specimens of the new species were collected with Malaise traps (Gressit and Gressit 1962) and Pennsylvania light traps (Frost 1957) in two Brazilian biomes: Amazon Forest in Amazonas and Pará states and Atlantic Forest in Rio de Janeiro State (Fig. 1). Collected specimens were preserved in 96% ethanol. Specimens were identified based on male genitalic morphology using Blahnik (1997Blahnik ( , 2002, Holzenthal (2012), Vilarino et al. (2015), Camargos (2016), and Desidério et al. (2018). In order to observe genital structures, the abdomen of each specimen was removed and cleared using heated 10% KOH or, alternatively, hot lactic acid for a few minutes (Blahnik et al. 2007), followed by a rinse in distilled water. After clearing, the abdomen was mounted on a temporary slide using glycerin or glycerin jelly, and it was examined under a Carl Zeiss Axiolab compound microscope equipped with a camera lucida. Finally, the abdomen was stored in a microvial with 96% ethanol together with the respective specimen. Pencil sketches were produced and then used as templates for vectorization in Adobe Illustrator CS6 (v. 16.0.0, Adobe Systems, Inc.) to create illustrations.
The distribution map was prepared using QGIS Las Palmas 2.18.10 software (QGIS Developed Team 2016). Morphological terminology for male genitalia follows that of Blahnik (1997Blahnik ( , 2002
PCR products were stained with GelRed™ (Biotium) and underwent agarose gel electrophoresis in 1.0% TBE and visualized under UV light. Amplicons were purified using ExoSAP-IT® (USB Affymetrix). DNA sequencing was performed in both forward and reverse directions using the same PCR primers by Macrogen (Seoul, South Korea). All sequences generated as part of this study were deposited in GenBank under accessions OM964809-OM964828.
Consensus sequences were constructed based on electropherogram assemblies in Geneious® v9.1.2 (Kearse et al. 2012) and used in a comparative similarity search at GenBank using the BLAST program (Altschul et al. 1990). Consensus sequences were aligned in Geneious using the ClustalW algorithm (Thompson et al. 1994) with a gap opening cost of 15 and a gap extension of 6.66. Amino acid translations were conducted to check for the absence of stop codons in the alignment.

Taxon sampling
In addition to sequences generated herein of the new species and two other Chimarra species, COI sequences publicly available from GenBank were included in the alignment to a total of 52 sequences (Table 1). As outgroups, four species of Philopotamidae were included: Wormaldia planae Ross & King (KX292642), Philopotamus montanus (Donovan, 1813) (MZ046700), Sortosa chilensis (Navás, 1918) (KM225345), and Chimarrhodella peruviana (Ross, 1956) (KX107274), the latter used for rooting according to Wahlberg and Johanson (2014). Sampling of Chimarra included the incertae sedis C. usitatissima Flint, 1971 and species representatives of the subgenera Curgia (2 spp.), Chimarra (2 spp.), Chimarrita (5 spp.), and Otarrha (6 spp.). In the latter, a more extensive sampling of individuals belonging to the Chimarra odonta complex (16 sequences) were also included. Although more species of Chimarra have COI sequences available in GenBank, we have selected species that were more closely related to the new species according to their morphology.

Phylogenetic placement and COI divergences
The with COI sequences were recovered as a monophyletic lineage with maximum bootstrap support (Fig. 2, in  green). Surprisingly, the maximum intraspecific K2P divergence for this species was low (0.5%), given that specimens were collected in two localities very distant geographically (approximately 660 km). Unfortunately, only a single individual of the other new species proposed, C.  (Otarrha) paraodonta sp. nov., had the COI successfully sequenced (Fig. 2, in blue).
It is important to note that, although we are treating C. (Otarrha) odonta (Fig. 2, in red) as a single nominal species (terminals labeled as C. odonta, C. nr. odonta, or C. odonta complex sp.), it is most likely a complex of cryptic species based on hitherto inconclusive morphological and molecular evidence, such as showing high intraspecific K2P COI divergences (up to 18.1%, see Discussion below). Diagnosis. Chimarra (Chimarrita) truncata sp. nov. is most similar to C. (Chimarrita) xingu Blahnik, 1997 by the short, fully divided tergum X with dorsal ridges and foldlike areas on outer margin of each lobe and the simple, almost uniform in width inferior appendages. However, the new species has an apically subtruncated ventral process of segment IX, while in C. (Chimarrita) xingu this process is pointed. Furthermore, C. (Chimarrita) truncata sp. nov. can be recognized by inferior appendages with truncated apices both in lateral and ventral view (rounded in C. xingu).

Diagnosis. The new species is closely similar to C.
(Otarrha) odonta Blahnik, 2002 by some shared primitive characters of the subgenus, like hindwing venation pattern with Rs 4-branched and the undivided anterior head setal warts. Both species also have a simple, subtriangular, and completely divided tergum X and an inner process on each inferior appendage. However, the new species has the Otarrha synapomorphic hindwing venation with Sc+R1 fused, narrower and more uniform lobes of tergum X, and inferior appendage rhomboidal (in lateral view) and more elongated and spatulated (in ventral view). Additionally, the dorsomesal process of the inferior appendage in C. (Otarrha) paraodonta sp. nov. is thornlike, more robust, and positioned subapically; while in C. odonta, this process is tooth-like, blunt, and positioned more apically. Furthermore, C. (Otarrha) paraodonta sp. nov. can be recognized by its differently shaped tergum IX as viewed dorsally, the more robust ventral process, and simple phallotremal sclerite.
Etymology. The specific epithet is a reference to the close similarity of the new species to Chimarra (Otarrha) odonta. Derived from the Greek, "para" = beside or near.

Discussion
Currently, four subgenera are recognized in Chimarra: Chimarra, Chimarrita, Curgia, and Otarrha, all of them recently reviewed (Blahnik 1997(Blahnik , 1998(Blahnik , 2002Flint 1998). Curgia was originally established for C. braconoides by Walker (1860) but was treated by Ulmer (1905) under Chimarra, and the genus was officially placed as a subgenus within Chimarra by Milne (1936). Until the end of the 1990s, many species were described and generally assigned to the subgenera Curgia and Chimarra, but without clearly defining them. Even after revisional works for Chimarra (Blahnik 1998) and Curgia (Flint 1998) proper diagnostic characters for them remain ambiguous, although subgeneric limits were better established. The nominotypical Chimarra is defined by the stem of forewing Rs vein with conspicuous curvature before discoidal cell, but it is less conspicuous in members of aterrima and obscurum Groups from East North America and is not true for several Afrotropical, Australasian, and Oriental species (Blahnik 1998;Wahlberg and Johanson 2014). Besides that, the male tergum X completely divided and widely separated is also pointed out as diagnostic for the subgenus, although it is not exclusive (Blahnik 1998). Curgia can be diagnosed by an anal vein of forewings with 2A looped to 1A, without forks, and by a combination of more or less inconclusive male genital characters, such as tergum X often entire or knoblike, or when it is apically separated, lobes are not totally divided mesally, and the short and linear inferior appendages (Blahnik 1998;Flint 1998). Blahnik (1997Blahnik ( , 2002 erected the subgenera Chimarrita and Otarrha. Chimarrita was created in 1997 to include several species from South America and one from the Greater Antilles besides three described species formerly placed in subgenus Chimarra: C. simpliciforma Flint, 1971, C. rosalesi Flint, 1981, and C. maldonadoi Flint, 1964. Phylogenetic analysis based on morphological data (Blahnik 1997) support the monophyly of the subgenus mainly by male abdominal and genital characters: ventral process of segment IX elongate, narrow, acute, and projecting; anteroventral margin of segment IX distinctly projecting and narrowed, acute mesally; preanal appendages very short and fused basally; and phallic spines with at least a slight helical twist. Chimarra  (Chimarrita) truncata sp. nov. has all of these characters, being readily placed within the subgenus.
Chimarrita was originally divided into three species groups by Blahnik (1997): maldonadoi, rosalesi, and simpliciforma Groups. However, Kjer et al. (2014) did not recover a monophyletic Chimarrita because of the position of C. maldonadoi as sister to Otarrha. It corroborates the speculative position of the maldonadoi Group in the original subgenus description based primarily on the structure of the female genitalia, which is distinctly elongate, although less so than in the other members of Chimarrita (Blahnik 1997). Therefore, based on this phylogenetic placement, this group, comprising two Antillean species, was considered incertae sedis within the genus, leaving Chimarrita with only two species groups (Kjer et al. 2014). The simpliciforma Group is easily recognized by the single, elongate spine emerging basodorsally from the phallotheca of the phallic apparatus. Interestingly, four out of the five Chimarrita species described recently belong to the simpliciforma Group (Blahnik and Holzenthal 2012;Vilarino and Calor 2015;Desidério et al. 2018), suggesting that, although this species group is the most diverse for the Neotropical Region, there are most likely many more species to describe. Chimarra (Chimarrita) truncata sp. nov. is clearly a member of the simpliciforma Group based on the characteristic of the group mentioned above. In the Kjer et al. (2014) phylogenetic analysis based on molecular data, they did recover the simpliciforma Group as monophyletic, but only sampled two species. In our analysis, C. (Chimarrita) truncata sp. nov. was recovered as sister to Chimarra simpliciforma, corroborating its position in the simpliciforma Group, although morphological analysis suggests that the new species is most similar (and likely more related to) C. xingu, which unfortunately was not sampled herein. However, the simpliciforma Group represented by five species was not recovered as monophyletic, but this result should be taken lightly as taxon sampling was low and internal branches had no significant support in the present analysis.
Otarrha was established in 2002 to accommodate eighteen species formerly placed either in the subgenus Chimarra or unplaced to subgenus (informally referred to as patosa Group) and thirteen new species. The subgenus is particularly well represented in the Lesser and Greater Antilles, and northern South America, with a few species recorded from Central or southern South America (Blahnik 2002). As in Chimarrita, morphological phylogenetic analyses were performed (Blahnik 2002), revealing a monophyletic group. However, universal characters applicable to all species inserted into the subgenus are hard to define and some combinations of head warts, wing venation, and male genital structure are necessary. Most Otarrha species can be diagnosed by the anterior setal warts divided and reduced venation of hindwing, with R1 fused to Sc, Rs three-branched, and M two-branched. In addition, male genital apparatus of most species have earlike preanal appendages and ventral process relatively elongate and wider subapically. However, C. (Otarrha) paraodonta sp. nov. and the very closely related C. (O.) odonta are the only species of the subgenus that share the absence of some subgeneric diagnostic characters in head anterior setal warts, which are not divided, and in the reduced venation of hindwings, which bear fork I (Rs fourbranched). Nevertheless, both species fit all male genital characters typical for the subgenus and are found herein nested within Otarrha in the phylogenetic analysis.
Although there is weak morphological support for most subgenera, recent species-level molecular data phylogenies, using a large subset of taxa and multiple molecular markers (Kjer et al. 2014;Wahlberg and Johanson 2014), corroborate the monophyly of Chimarra and all of its subgenera, except Chimarrita, which was recovered as monophyletic only with the exclusion of C. maldonadoi (Kjer et al. 2014), which appears closer to Otarrha. Furthermore, in Kjer et al. (2014), Chimarra usitatissima was unplaced to subgenus, corroborating the incertae sedis position established by Blahnik (2002) for it by its unusual morphology. In our present analysis, Chimarrita was recovered as monophyletic only with the inclusion of C. usitatissima, although internal branches did not have significant support (probably an artefact of low gene and taxon sampling), thus our results are inconclusive. Phylogenetic relationships among Chimarra subgenera are mostly consensual in that the subgenus Chimarra is sister to a clade containing all other Chimarra species, but relationships within this other clade varies among published analyses. In the maximum likelihood tree based on COI and 28S, Kjer et al. (2014) recovered Curgia and Otarrha in a clade together with the unplaced maldonadoi Group and C. usitatissima, this clade being sister to Chimarrita. Wahlberg and Johanson (2014) presented two different analyses based on COI, CAD, and POL-II: in the parsimony analysis of the data, Chimarrita was recovered as sister to all Chimarra, which was divided in two clades: the subgenus Chimarra and a clade with Otarrha+Curgia; while in the Bayesian analyses Otarrha was recovered as sister to Chimarrita+Curgia. However, it is important to consider that some relationships of the mentioned analyses between subgenera had low branch support and Chimarrita was represented by a single species in Wahlberg and Johanson (2014).
Of a total of 33 species, only three species of Otarrha are recorded from Brazil: C. diakis Flint, 1971 (Amazon), C. odonta (Atlantic Forest), and C. machadoi Camargos, 2016 (Cerrado) (Blahnik 2002;Blahnik and Holzenthal 2012;Camargos 2016). The finding of a new species of Otarrha from Parque Nacional do Itatiaia is actually surprising, considering that it is one of the most well sampled national parks in Brazil for caddisflies (Dumas and Nessimian 2012), exposing that taxonomic gaps may still exist in Southeastern Brazil. Although no formal quantitative study was performed, the new species does appear to be rare when compared to the sympatric and morphologically similar C. odonta. While sorting Malaise traps from where only the three type specimens of C. paraodonta sp. nov. were found, over 200 specimens of C. odonta were identified.
Considering the high morphological resemblance of C. paraodonta sp. nov. to C. odonta, we decided to include in the present study all C. odonta barcode sequences available (from Zhou et al. 2016) and generated a few more. Based on the present molecular phylogenetic analysis and K2P divergences, it seems evident that C. odonta is likely a species complex, comprised of at least four distinct genetic species distributed from Rio de Janeiro to Santa Catarina states in the southeastern and southern Brazil, respectively. Two of these species apparently co-occur in Parque Nacional do Itatiaia, together with C. paraodonta sp. nov. These genetic species may be supported by fine morphological features of the phallus, as variation seems to be found in this complex (R Blahnik, in litt. 2021). However, in order to robustly delimit this species complex, an integrative analysis is needed comprising a detailed morphological study, especially embracing its full distributional range (Vilarino and Calor 2015), and sequencing of additional molecular markers.
Our study shows that the combination of detailed morphological observation and molecular sequence data is constructive in discovering and describing new species and provides another example of the effectiveness of DNA barcodes as a tool for species delimitation. Due to efforts of the global initiative Trichoptera Barcode of Life (TBOL) that started in 2007, a comprehensive COI barcode reference library is available for about one-third of the described caddisfly species (Zhou et al. 2016). So, this work helps to expand this database, and thus facilitate the knowledge of caddisfly diversity and provide tools for fast and reliable identification. Most barcode studies on Trichoptera show clearly distinguishable intraspecific and interspecific variability, clearly seen in this study, which is known as a barcoding gap (e.g., Graf et al. 2005;Waringer et al. 2008;Zhou et al. 2010;Pauls et al. 2010;Previšić et al. 2014;Santos et al. 2016), but in some taxa levels of these variations are not conclusive (e.g., Waringer et al. 2007;Pauls et al. 2010). Potential limitations of using mtDNA to infer species boundaries include retention of ancestral polymorphism, male-biased gene flow, selection on any mtDNA nucleotide (as the whole mitogenome is one linkage group), introgression following hybridization, and paralogy resulting from transfer of mtDNA gene copies to the nucleus (Moritz and Cicero 2004). As a result, solely using mtDNA divergence or phylogenetic signal may lead us to biased species delimitations. Therefore, using multilocus sequence data can increase delimitation success (Dupuis et al. 2012) and provide support for species delimitations under different theoretical models that combine species phylogenies and gene genealogies via ancestral coalescent processes (Yang and Rannala 2010). Thus, integrative taxonomy approaches should ideally combine morphology, ecology and/or behavior with multi-locus nu/mtDNA data.
Caddisflies are excellent freshwater biological indicators due to their ecological diversity and intolerance of most species to pollution and disturbances (Resh 1993;Houghton 2004). However, the practical use of caddisfly larvae in bioassessment monitoring programs are limited in many areas by the lack of lower-level taxonomic resolution. Higher levels of identification, such as family or even genus, can often mask the variability of environment and species interaction, causing ecological information loss or redundancy (Resh and Unzincker 1975;Ruiter et al. 2013). In the Neotropical Region only about 9% of caddisfly species have their immature stages described . COI sequences have proven successful in associating larval and adult stages in Neotropical Trichoptera species (e.g., Santos et al. 2016;Barcelos-Silva et al. 2018), and its use can considerably reduce this gap of knowledge. Although life stage association was not the purpose of this work, the barcode data generated here will provide a valuable tool for future Chimarra larvae association, also contributing to biodiversity and conservation assessments.

Authors' contributions
All authors contributed to the study conceptualization and design. PDM, LLD, and GRD identified specimens and produced illustrations and maps. PDM and MPR carried out the molecular laboratory work under supervision of DMT. All authors contributed to the draft of the manuscript. All authors read and approved the final manuscript.