Trichomycteridae belongs to the order Siluriformes and is a family geographically distributed throughout South and Central America (de Pinna, Wosiacki, 2003), comprising small fishes (about 10 cm standard length) known as “candirus”, “cambevas” or “guascas” with species that exhibit different habits, most being found among rocks in streams with stony bottoms or buried in the sand (Baskin, 1973; Wosiacki, de Pinna, 2008; Ferrer, Malabarba, 2011, 2013). This family forms a monophyletic lineage (Baskin, 1973; de Pinna, 1992a, 1998; Schmidt, 1993; Fernández, Schaefer, 2009; Henschel et al., 2017; Ochoa et al., 2017, 2020; Fernández etal., 2021), primarily supported by synapomorphies related to the modified opercular system (Baskin, 1973; de Pinna, 1992a, 2016).

Trichomycteridae has nine subfamilies, of which Trichomycterinae is the most species-rich (Fricke et al., 2023), with its monophyly corroborated by Datovo, Bockmann (2010) using miology, and Ochoa etal. (2017, 2020) and Fernández et al. (2021) with molecular data. Among the valid genera in this subfamily, Cambeva Katz, Barbosa, Mattos & Costa, 2018 is considered a monophyletic genus supported by molecular analyses (Katz et al., 2018; Ochoa et al., 2020). Nevertheless, other authors have studied the phylogenetic relationships of this group and found morphological synapomorphies between Cambeva and Scleronema Eigenmann, 1917 (see Katz et al., 2018; Ochoa et al., 2020).

Some species of Cambeva are identified through body color pattern, as it is considered a conservative characteristic (Bockmann et al., 2004). However, within Trichomycteridae, great variation in color pattern has been found, making identification difficult (Silva et al., 2010; Nascimento et al., 2017; Donin et al., 2022). Studies employing morphometric and osteological data (Ferrer, Malabarba, 2011), as well as those using molecular data into the analyses, proved to be efficient in distinguishing species (Reis et al., 2020a; Costa et al., 2023). These approaches have also highlighted the remarkable variation in color patterns exhibited by Cambeva species (Silva et al., 2010; Nascimento et al., 2017; Donin et al., 2022; Costa et al., 2023).

The upper rio Paraná is one of the main drainages in Brazil, occupying around 880,000 km², representing 10.3% of the Brazilian territory and extending across five Brazilian states (Agostinho et al., 2007). In the state of Paraná, the rio Piquiri and Ivaí basins are among the main tributaries of the left bank of the upper rio Paraná basin (Frota et al., 2016; Cavalli et al., 2018). The proportion of possible new species for the region is approximately 10% (Langeani et al., 2007; Frota et al., 2016; Cavalli et al., 2018; Reis et al., 2020a), this demonstrates how the description of species even in well-sampled regions such as the rio Piquiri and rio Ivaí, is still important for understanding local diversity, especially in headwaters, which are home to small species from the region. In this context, this study aims to describe a new species of Cambeva from the rio Piquiri and Ivaí basins, upper rio Paraná basin, Brazil, using both morphological and molecular data.

Morphological data. Morphometric data were taken point to point with a digital caliper (precision 0.1 mm) on the left side of specimens following Tchernavin (1944) for 1) maxillary, 2) nasal and 3) rictal barbels length; Costa (1992) for 4) mouth width and 5) supraorbital pores S6 distance; Donin et al. (2022) for 6) standard length, 7) head length, 8) head width, 9) predorsal length, 10) prepelvic length, 11) pre-anal length, 12) scapular-girdle width, 13) trunk length, 14) pectoral-fin length, 15) pelvic-fin length, 16) caudal-peduncle length, 17) caudal-peduncle depth, 18) body depth, 19) dorsal-fin base length, 20) anal-fin base length, 21) snout length, 22) interorbital distance and 23) eye diameter; and Nascimento et al. (2017) 24) body length, 25) anal-fin length, 26) dorsal-fin length and 27) Pelvic-anal length (hereafter Pelvic-anal fins distance). Number of morphometric measures do not correspond to the order shown in Tabs. 1-2. We conducted a Principal Component Analysis (PCA) in PAST v. 4.03 (Hammer et al., 2001) to check the overall morphometric variation among specimens and to determine if distinct groups were identified. The measurements indicated previously were treated with the Allometric Burnaby’s method to remove the allometric size-dependent shape variation from the multivariate data set, and therefore log-transforming the data (Hammer et al., 2001). Additionally, the following morphometric data were excluded for PCA analysis, as they may indicate an influence of size on the shape arrangement between the specimens (see Chuctaya etal., 2018): 1) standard length, 2) trunk length and 3) body length. The final PCA result was visualized using the ggplot2 package (Wickham, 2016) in R v. 4.3.2 (R Development Core Team, 2023). For osteological analysis, nine specimens were cleared and stained (c&s) according to procedures described by Taylor, Van Dyke (1985). Nomenclature of the osteological structures, laterosensory canals and associated pores follows Bockmann et al. (2004), except for the use of barbular, which followed de Pinna et al. (2020), and opercular and interopercular odontodophores, which followed de Pinna, Dagosta (2022). Vertebrae counts followed Ferrer, Malabarba (2013), excluding the Weberian complex and the compound caudal centrum (pu1+u1) was counted as a single element. Counts of odontodes from opercular and interopercular odontodophores were made only in c&s specimens. The counts of unsegmented rays (represented by lowercase Roman numerals) in c&s specimens are given before the number of unbranched and segmented rays. Unbranched rays (represented by uppercase Roman numerals) and branched rays (represented by Arabic numerals) were performed in 40 specimens. Holotype counts are marked with an asterisk and each meristic character is followed by the number of specimens examined in parentheses. Non-type specimens correspond to a cleared and stained individual, poorly ossified and with broken bones, and a poorly preserved specimens. Osteological illustrations were prepared in the digital software Photoshop version 2021 v. 22, based on photographs and direct observation of c&s specimens under a stereomicroscope. To verify possible morphotypes of the new species, we used traditional color pattern diagnosis of Cambeva species, and provided morphological and molecular analysis from possible morphotypes found. Morphological data of the specimens were based on the comparative material listed and the original descriptions and redescriptions of the congener species. We analyzed the photo and x-ray of the holotype of Pygidium paolence Eigenmann, 1917, available from https://collections-zoology.fieldmuseum.org/catalogue/637343. Institutional abbreviations follow Sabaj (2020).

Conservation status. We calculated the extent of occurrence (EOO) and area of occupancy (AOO) using GeoCAT webserver (http://geocat.kew.org/). The conservation status followed the guidelines of the International Union for Conservation of Nature for the red list of threatened species (IUCN Standards and Petitions Committee, 2024).

Molecular data and analyses. DNA extractions from previously ethanol-preserved tissue samples were carried out using the Wizard Genomic DNA Purification kit (Promega), following the manufacturer’s protocol. For the amplification of the partial fragment of the cytochrome c oxidase subunit I (COI) gene we used the primers FR1d (Ivanova et al., 2007) and FishF1 (Ward et al., 2005) with the following conditions: initial denaturation at 95 °C for 5 min, followed by 35 cycles at 94 °C for 30 s, 52 °C for 40 s, and 72 ºC for 1 min, with a final extension at 72 °C for 10 min (adapted from Ivanova et al., 2007). All samples were then quantified using NanoDrop™ Lite Spectrophotometer, purified with polyethylene glycol 8000 (Rosenthal et al., 1993), and sequenced at ACTGene Análises Moleculares LTDA, using an ABI 3500 Applied Biosystems automated sequencer.

Sequences were edited and aligned using BioEdit (Hall, 1999) and ClustalW (Thompson et al., 1994) algorithm in MEGA7 (Kumar et al., 2016), which was also used to calculate genetic distances using the Kimura-2-Parameter (K2P) model. A maximum likelihood (ML) tree and its substitution model were calculated using PhyML (Guindon et al., 2010), with branch support tested using 1,000 replicates.

The Bayesian tree with all specimens (Fig. S1) was created with a relaxed clock with speciation birth-death model, on an arbitrary timescale, using BEAST v. 1.8.4 (Drummond et al., 2012). The best-fitting model of molecular substitution was found using the Bayesian Information Criterion (BIC) implemented in the web server of IQ-TREE (http://iqtree.cibiv.univie.ac.at/, see Trifinopoulos et al., 2016). A random tree was used as a starting tree for MCMC searches with two independent runs of 40,000,000 generations, and trees were sampled at every 4,000th generation. Chain convergence was analyzed by Tracer 1.6 to determine the stationary phase and an effective sample size > 200 (Rambaut et al., 2018). Ten percent of the chain was discarded as a burn-in procedure in Tree Annotator v. 1.8.4 to find the Maximum Clade Credibility Tree (MCC) (Drummond et al., 2012). The final tree was edited using Interactive Tree of Life (iTOL) (Letunic, Bork, 2021).

Trichomycterus nigricans Valenciennes, 1832 (MN385796) was used as an outgroup and all the sequences obtained in this study were deposited in GenBank (OQ756220, OQ756221 and PP281332). Access to the genetic heritage of the species was authorized by the National System for Management of Genetic Heritage and Associated Traditional Knowledge (SisGen) under the registration number AB3B0CA. All molecular data used and taxonomic identifications provided by the authors are listed in Tab. S2.

Species delimitation methods. In order to achieve optimal performance in the delimitation analyses and to obtain more accurate results from the delimitation methods, we used only closer sister groups of the new species in these analyses. Cambeva horacioi Reis, Frota, Fabrin & Graça 2019 was used as an outgroup. For this study, four different species delimitation methods were used to validate the results: Assemble Species by Automatic Partitioning (ASAP; Puillandre et al., 2021), the Poisson tree process (PTP) and its Bayesian implementation (bPTP; Zhang et al., 2013), and the General Mixed Yule Coalescent approach (GMYC; Fujisawa, Barraclough, 2013). For the first method, the analysis was performed in the web server (https://bioinfo.mnhn.fr/abi/public/asap/), using the alignment file. The PTP method used a new ML tree from PhyML as input in the web server (https://species.h-its.org/), and its Bayesian implementation utilized 500,000 MCMC generations with 0.2 burn-in, along with all other default parameters. For the latter method, GMYC, a new Bayesian inference of the gene tree, using only sister groups of the new species, and with unique haplotypes was estimated. We used the same settings as the bayesian tree created for all specimens, except for the Markov chain generations, being performed searches with two independent runs of 50,000,000 generations, and trees were sampled at every 5,000th generation. The Maximum Clade Credibility Tree (MCC) was checked in FigTree v. 1.4.4 (Rambaut, 2018) and used as an input file (Newick format) for the GMYC analyses performed in the web server (https://species.h-its.org/gmyc/), using a single threshold method. The final tree was edited using Interactive Tree of Life (iTOL) (Letunic, Bork, 2021).

Cambeva perobana,new species

urn:lsid:zoobank.org:act:C0BF9D84-4F67-4B57-8BCA-267A8F608AC9

(Figs. 1–8; Tab. 1)

Trichomycterus sp. 1 —Delariva, Silva, 2013:552 (Checklist of fishes from Perobas Biological Reserve).

Trichomycterus sp. 2 —Delariva, Silva, 2013:552 (Checklist of fishes from Perobas Biological Reserve).

Trichomycterus sp. 3 —Delariva, Silva, 2013:552 (Checklist of fishes from Perobas Biological Reserve; partim: NUP 11703).

Trichomycterus sp. —Delariva, Silva, 2014:61 (Checklist of fishes from Perobas Biological Reserve; fig. in p. 61).

Holotype. NUP 23907, 75.9 mm SL, Brazil, Paraná State, municipality of Tuneiras do Oeste, rio Mouro, tributary of rio Goioerê, rio Piquiri basin, upper rio Paraná basin, 23°52’54”S 52°49’46”W, 15 Jul 2011, A. G. Bifi & G. C. Deprá.

Paratypes. All from Brazil, Paraná State, upper rio Paraná basin. Rio Ivaí basin: NUP 11020, 2, 56.1–70.0 mm SL, municipality of Araruna, rio Ligeiro, tributary of rio Ivaí, 23°50’52”S 52°33’47”W, 25 Oct 2010, C. H. Zawadzki, L. Raisi & G. C. Zawadzki. NUP 11726, 2, 57.8–71.9 mm SL, municipality of Cianorte, rio dos Índios, 23°51’11”S 52°42’03”W, 6 Dec 2010, R. Delariva. NUP 24163, 1, 56.3 mm SL, municipality of Cianorte, rio dos Índios, tributary of rio Ivaí, 23°51’11”S 52°42’03”W, 6 Dec 2010, R. Delariva. NUP 24274, 1 c&s, 50.0 mm SL, municipality of Cianorte, rio dos Índios, 23°51’11”S 52°42’03”W, 6 Dec 2010, R. Delariva. Rio Piquiri basin: MCP 54936, 3, 67.6–79.5 mm SL, municipality of Tuneiras do Oeste, rio Mouro, tributary of rio Goioerê, 23°53’10”S 52°49’19”W, 1 May 2017, R. Delariva & C. Larentis.NUP 11703, 5, 38.4–67.5 mm SL, municipality of Tuneiras do Oeste, rio Concórdia, tributary of rio Mouro, 23°52’51”S 52°49’56”W, 4 Dec 2010, R. Delariva. NUP 11707, 2, 23.3–58.8 mm SL, municipality of Tuneiras do Oeste, rio Concórdia, tributary of rio Piquiri, 23°52’51”S 52°49’56”W, 4 Dec 2010, R. Delariva. NUP 11712, 4, 27.3–68.8 mm SL, municipality of Tuneiros do Oeste, rio Mouro, tributary of rio Piquiri, 23°52’07”S 52°48’56”W, 6 Dec 2010, R. Delariva. NUP 14648, 5, 22.0–63.8 mm SL, municipality of Tuneiras do Oeste, rio Saquarema, tributary of rio Piquiri, 23°52’02”S 52°46’30”W, 28 Aug 2011, R. Delariva. NUP 16051, 2, 31.0–58.3 mm SL, municipality of Janiópolis, NN stream, tributary of rio Barreiro, 24°12’44”S 52°47’50”W, 25 Mar 2014, W. J. da Graça, W. M. Domingues, F. A. Teixeira & R. J. da Graça. NUP 16086, 1, 52.5 mm SL, municipality of Tuneiras do Oeste, rio Água Cinquenta e Cinco, tributary of rio Goioerê, 23°56’02”S 52°45’52”W, 29 Jan 2014, W. J. da Graça, W. M. Domingues, F. A. Teixeira & R. J. da Graça. NUP 17219, 2, 34.5–68.6 mm SL, municipality of Farol, NN stream, tributary of rio Farol, 24°22’45”S 52°40’53”W, 12 Sep 2014, C. H. Zawadzki. NUP 17223, 1, 35.0 mm SL, municipality of Farol, rio Farol, tributary of rio Goioerê, 24°22’45”S 52°40’ 54”W, 12 Sep 2014, C. H. Zawadzki. NUP 17228, 1, 45.7 mm SL, municipality of Farol, NN stream, tributary of córrego Água da Granada, 24°22’53”S 52°35’58”W, 12 Sep 2014, C. H. Zawadzki. NUP 17232, 3, 27.6–43.1 mm SL, municipality of Farol, Córrego Água da Granada, tributary of rio Goioerê, 24°16’55”S 52°41’31”W, 13 Sep 2014, C. H. Zawadzki. NUP 23908, 2, 29.0–68.3 mm SL, municipality of Tuneiras do Oeste, NN stream, tributary of rio Mouro, 23°52’54”S 52°49’46”W, 23 Mar 2022, A. G. Bifi & G. C. Deprá. NUP 24159, 2, 60.9–67.5 mm SL, same data as holotype. NUP 24160, 1, 41.2 mm SL, municipality of Farol, NN stream, tributary of córrego Água da Granada, 24°22’53”S 52°35’58”W,12 Sep 2014, C. H. Zawadzki. NUP 24161, 1, 49.9 mm SL, municipality of Farol, NN stream, tributary of rio Farol, 24°22’45”S 52°40’53”W, 12 Sep 2014, C. H. Zawadzki. NUP 24162, 1, 74.7 mm SL, municipality of Tuneiras do Oeste, rio Água Cinquenta e Cinco, tributary of rio Goioerê, 23°56’02”S 52°45’52”W, 29 Jan 2014, W. J. da Graça, W. M. Domingues, F. A. Teixeira & R. J. da Graça. NUP 24164, 2 c&s, 45.7–50.0 mm SL, municipality of Janiópolis, NN stream, tributary of rio Barreiro, 24°12’44”S 52°47’50”W, 25 Mar 2014, W. J. da Graça, W. M. Domingues, F. A. Teixeira & R. J. da Graça. NUP 24165, 1, 68.8 mm SL, rio Concórdia, tributary of rio Mouro, 23°52’54”S 52°49’46”W, 15 Jul 2011, A. G. Bifi & G. C. Deprá. NUP 24166, 2, 46.0–63.2 mm SL, rio Mouro, tributary of rio Goioerê, 23°52’54”S 52°49’46”W, 9 Sep 2022, I. C. Martins, Renan B. Reis, B. H. M. Stabile & M. Z. Roloff. NUP 24167, 6, 25.9–54.6 mm SL, rio Concórdia, tributary of rio Mouro, 23°53’0.31”S 52°49’52.00”W, 9 Sep 2022, I. C. Martins, R. B. Reis, B. H. M. Stabile & M. Z. Roloff. NUP 24271, 1 c&s, 48.1 mm SL, municipality of Tuneiras do Oeste, rio Água Cinquenta e Cinco, tributary of rio Goioerê, 23°56’02”S 52°45’52”W, 29 Jan 2014, W. J. da Graça, W. M. Domingues, F. A. Teixeira & R. J. da Graça. NUP 24272, 1 c&s, 50.5 mm SL, municipality of Tuneiras do Oeste, rio Água Cinquenta e Cinco, tributary of rio Goioerê, 23°56’02”S 52°45’52”W, 29 Jan 2014, W. J. da Graça, W. M. Domingues, F. A. Teixeira & R. J. da Graça. NUP 25070, 7, 25.7–57.8 mm SL, municipality of Tuneiras do Oeste, rio Concórdia, tributary of rio Mouro, 23°53’03”S 52°49’52”W, 9 Sep 2022, I. C. Martins, R. B. Reis, B. H. M. Stabile & M. Z. Roloff. NUP 25181, 3 c&s, 52.4–75.1 mm SL, municipality of Tuneiros do Oeste, rio Mouro, tributary of rio Piquiri, 23°52’07”S 52°48’56”W, 6 Dec 2010, R. Delariva.

Non-types. CIG 869, 2, 51.0–67.0 mm SL, municipality of Campo Mourão, rio Mourão, 24°11’18”S 52°22’43”W, 1 Jul 2010, V. A. Frana. NUP 24273, 1 c&s, 54.9 mm SL, municipality of Cianorte, rio dos Índios, 23°51’11”S 52°42’03”W, 6 Dec 2010, R. Delariva.

Diagnosis.Cambeva perobana can be distinguished from most congeners by having six branched pectoral-fin rays (vs. four in C. alphabelardense Costa, Feltrin & Katz, 2022; C. betabelardense Costa, Feltrin & Katz, 2022, C. pascuali (Ochoa, Silva, Costa e Silva, Oliveira & Datovo, 2017); five in C. brachykechenos (Ferrer & Malabarba, 2013), C. flavopicta Costa, Feltrin & Katz, 2020, C. grisea Costa, Feltrin & Katz, 2021, C. mboycy (Wosiacki & Garavello, 2004), C. naipi (Wosiacki & Garavello, 2004), C. podostemophila Costa, Feltrin & Katz, 2023, C. poikilos (Ferrer & Malabarba, 2013), C. taroba (Wosiacki & Garavello, 2004) and C. tourensis Costa, Feltrin & Katz, 2023; or seven in C. barbosae Costa, Feltrin & Katz, 2021, C. castroi (de Pinna, 1992), C. concolor (Costa, 1992), C. crassicaudata (Wosiacki & de Pinna, 2008), C. diabola (Bockmann, Casatti & de Pinna, 2004), C. difficilis Costa, Feltrin & Katz, 2024, C. guaraquessaba (Wosiacki, 2005), C. igobi (Wosiacki & de Pinna, 2008), C. iheringi (Eigenmann, 1917), C. melanoptera Costa, Abilhoa, Dalcin & Katz, 2022, C. tupinamba (Wosiacki & Oyakawa, 2005), C. variegata (Costa, 1992),and C. ytororo (Terán, Ferrer, Benitez, Alonso, Aguilera & Mirande, 2017)). Cambevaperobana can be distinguished from C. chrysornata Costa, Feltrin, Mattos, Dalcin, Abilhoa & Katz, 2023, C. davisi (Haseman, 1911), C. orbitofrontalis Costa, Feltrin & Katz, 2021, and C. stawiarski (Miranda Ribeiro, 1968)by the lower number of interopercular odontodes (28–29 vs. 30–38), or from C. balios (Ferrer & Malabarba, 2013), C. biseriata Costa, Feltrin, Mattos, Dalcin, Abilhoa & Katz, 2023, C. diatropoporos (Ferrer & Malabarba, 2013), C. diffusa Costa, Feltrin & Katz, 2021, C. duplimaculata Costa, Feltrin & Katz, 2021, C. horacioi, C. imaruhy Costa, Feltrin & Katz, 2021, C. longipalata Costa, Feltrin & Katz, 2021, C. notabilis Costa, Feltrin & Katz, 2021, C. panthera Costa, Feltrin & Katz, 2021, C. pericoh Costa, Feltrin & Katz, 2021, C. perkos (Datovo, Carvalho & Ferrer, 2012), C. plumbea (Wosiacki & Garavello, 2004), C. tropeiro (Ferrer & Malabarba, 2011), C. urubici Costa, Feltrin & Katz, 2021, and C. ventropapillata Costa, Feltrin, Mattos, Dalcin, Abilhoa & Katz, 2023 by the greater number of interopercular odontodess (28–29 vs. 11–26); from C. balios, C. botuvera Costa, Feltrin & Katz, 2021, C. chrysornata, C. diffusa, C. duplimaculata, C. gamabelardense Costa, Feltrin & Katz, 2022, C. guaratuba Costa, Feltrin, Mattos, Dalcin, Abilhoa & Katz, 2023, C. horacioi, C. imaruhy, C. longipalata, C. notabilis, C. orbitofrontalis, C. paolence (Eigenmann, 1917), C. pericoh, C. perkos, C. piraquara Reis, Wosiacki, Ferrer, Donin & Graça, 2023, C. stawiarski, and C. urubici differs by the number of vertebrae (34–37 vs. more than 37); from C. biseriata, C. botuvera, C. chrysornata, C. davisi, C. imaruhy, C. notabilis, C. papillifera (Wosiacki & Garavello, 2004), C. plumbea, C. stawiarski, and C. urubici differsby the lower number of opercular odontodes (13 vs. 14–28), or from C. cubataonis (Bizerril, 1994), C. longipalata, and C. panthera differs by the lower number of opercular odontodes (13 vs. 7–11); from C. cubataonis, C. diffusa, C. guaratuba, C. guareiensis, C. imaruhy, C. longipalata, C. notabilis, C. orbitofrontalis, C. panthera, C. pericoh, and C. zonata (Eigenmann, 1918) differs by the number of branchiostegal rays (9 vs. 7–8); from C. papillifera,and C. ventropapillata differs by the absent of conspicuous papillae on ventral region of head (vs. present);and from C. tropeiro is distinguished by the presence of pelvic girdle and fin (vs. absent).

Additionally, C. perobana can be distinguished from all congeners from the upper rio Paraná and rio Iguaçu basin by the presence of pelvic girdle (vs. absence of pelvic girdle in C. pascuali); by the long rictal barbel, 40.0–73.0% of HL (vs. short, 12.0–30.0% in C. papillifera, 33.0–40.0% in C. castroi); by the pelvic fins not reaching the urogenital opening (vs. reaching to anal-fin origin in C. crassicaudata and C. paolence, reaching urogenital opening in C. cauim Reis, Ferrer & Graça, 2021, C. davisi, C. guareiensis Katz & Costa, 2020, and C. taroba); by the caudal fin truncate to rounded (vs. caudal fin forked in C. crassicaudata; distal margin strongly concave in C. cauim, or slightly concave in C. stawiarski); by the first pectoral-fin ray not prolonged as a filament (vs. prolonged as a rudimentary filament in C. pascuali, and C. piraquara; short filament in C. paolence, and some specimens of C. piraquara; and long filament in C. taroba); by the color pattern of the caudal fin, presenting blotches, spots, or a homogeneous dark-brown or gray pattern in the proximal region (vs. presenting a pale-yellow to unpigmented stripe in the proximal region in C. castroi, C. diabola, C. difficilis, and C. melanoptera); by the presence of a narrow mid-lateral stripe in some specimens (vs. presence of four narrow stripes on the body: mid-sagital, mid-dorsal, mid-lateral, and ventro-lateral stripes in C. naipi); by the shorter head, 17.2–20.6% of SL (vs. longer, 23.8–26.8% of SL in C. igobi); by the longer pelvic and pectoral fins, 10.3–16.0% of SL and 7.5–9.8%, respectively (vs. 7.7–9.1%, and 5.6–7.3%, respectively, in C. mboycy); by the color pattern of some specimens, consisting in a dark-gray to dark-brown, becoming lighter towards ventral portion of the body and head, and absence of chromatophores, or some specimens with irregular dark-brown blotches on the inner skin layer, larger than two or three times the diameter of the orbit, sometimes coalescing and forming larger blotches, and an interrupted stripe in the mid-lateral region of the flank, over a plain yellowish to brown background (vs. color pattern consisting in dorsal and lateral surface of body with scattered circular well-defined dark-brown blotches, variable in size in inner skin layer and small black spots on the outer skin layer in C. horacioi; or lateral and dorsal surface of body with numerous small spots, irregularly distributed, and well-defined dark-brown blotches, larger than orbit diameter, along dorsal and mid-lateral surface of body in C. iheringi); and by the number of branchiostegal rays (9 vs. 7–8 in C. iheringi).

Description. Summarized morphometric data of two morphotypes in Tab 1. Body elongate, trunk roughly cylindrical close to head and gradually becoming laterally compressed towards caudal fin. Dorsal profile of trunk slightly convex along anterior half of body to insertion of dorsal fin. Ventral profile of trunk slightly convex. Dorsal and ventral profiles of caudal peduncle slightly convex.

Head depressed, trapezoidal in dorsal view, wider posteriorly and anterior portion slightly rounded. Dorsal and ventral profiles of head straight to slightly convex in lateral view. Eyes located dorsolaterally on anterior region of head, at same longitudinal line of nasal barbel. Eyes with a round to elliptical shape anteroposteriorly, covered by thin and translucent skin. Orbital rim not free. Eyes visible from lateral view.

Anterior nostril slightly smaller than diameter of eye, surrounded by flap of integument posterolaterally continuous with base of nasal barbel. Posterior nostril slightly smaller than diameter of eye, surrounded anterolaterally by thin flap of integument. Gill openings not constricted, forming free fold reaching pectoral-fin insertion. Mouth subterminal and slightly curved with corners posteriorly oriented in ventral view. Upper lip thicker laterally. Lower lip with conspicuous fleshy lobes at corners of mouth, continuous with base of rictal barbels. Lips with small rounded papillae of approximately same size.

Barbels with broad bases, tapering gradually towards tips. Nasal barbel emerging from posterolateral region of anterior nostril with tip surpassing infraorbital pores i10 when adpressed to head. Maxillary barbel emerging from corner of mouth with tip reaching to anterior region of interopercular odontodophore when adpressed to head. Rictal barbel emerging from corner of mouth, shorter than maxillary barbel.

Pectoral fin with distal margin rounded, I,6*(40), first ray unbranched, not prolonged as filament. Pelvic fin with distal margin rounded, not covering anterior margin of urogenital papilla; with I,4* rays (40). Pelvic-fin insertion anterior to dorsal-fin origin. Inner margins of pelvic fins close basally. Urogenital papilla closer to distal margin of pelvic fin than to origin of anal fin. Dorsal fin with distal margin rounded, ii(5), II,7* rays (40). Origin of dorsal fin located at vertical through last third of pelvic fin. Anal fin elongated with distal margin rounded and slightly smaller than dorsal fin, ii(5), II,5*(29) or II,6(11) rays. Origin of anal fin located at vertical through half or last third of dorsal-fin base. Caudal fin with distal margin rounded or truncate in specimens smaller than 35.0 mm SL; upper caudal plate with I,5* rays (40), lower caudal plate with I,6* rays (40).

Osteology. Mesethmoid with anterior margin straight to slightly concave and cornua short, with tapering distal ends. Anterior cranial fontanel restricted to small, rounded opening situated between frontals and epiphyseal bar. Posterior cranial fontanel long and wide extending from posterior portion of frontals to parieto-supraoccipital. Epiphyseal bar longer than wide. Antorbital anteriorly expanded and posteriorly elongated, extending over anterior third of autopalatine. Barbular bone elongate, with medial process in anterior third. Sphenotic, prootic, and pterosphenoid fused, anterior portion anterolaterally directed in dorsal view (Fig. 2). Vomer arrow-shaped with long posterior process extending to parasphenoid. Parasphenoid with long and pointed posterior process extending to basioccipital. Weberian capsule with lateral openings and anterior margin fused to the basioccipital.

Premaxilla rectangular with 38 or 40 (2) spatulate to conical teeth similar in size and roughly distributed in four irregular rows. Dentary teeth extending from coronoid process base to near dentary symphysis (Figs. 3A–B). Maxilla boomerang-shaped, shorter than premaxilla. Autopalatine with lateral margin concave; anterior margin slightly convex; medial margin slightly concave and long posterior process extending over posterior portion of metapterygoid. Metapterygoid large and laminar, connected to quadrate through cartilage; posterior portion with notch (arrow in Figs. 3C–D). Quadrate L-shaped with concavity in anterior portion. Hyomandibula well developed, dorsal margin with concavity (Figs. 3C–D). Opercle longer than interopercle. Opercular odotodophores ovoid to rounded with 13 (2) conical odontodes, gradually curving medially and increasing in size posteriorly, arranged in five irregular transverse rows. Interopercular patch of odontodes elongate with 28 or 29 (2) conical odontodes, arranged in two transverse rows.

Ventral hypohyal trapezoid-shaped. Anterior ceratohyal elongate and wider at anterior and posterior ends. Posterior ceratohyal short, triangular, with rounded tips. Nine branchiostegal rays (2): six in contact with anterior ceratohyal, one with interceratohyal cartilage, and two with posterior ceratohyal. Four posteriormost branchiostegal rays wider distally (Figs. 4A–B). Urohyal with expanded anterior head, two elongate lateral processes with wide bases and decreasing in width distally with rounded tips, and sharp and elongated posterior process. Posterior process of urohyal shorter than lateral processes.

Basibranchials 2 and 3 elongated, connected by cartilage; basibranchial 2 slightly wider than basibranchial 3. Basibranchial 4 hexagonal and entirely cartilaginous. Hypobranchial 1 elongated, with cartilaginous tips, approximately of same size than basibranchial 2. Hypobranchials 2 and 3 with narrow anterolateral ossified processes with large area of cartilage posteriorly; hypobranchials 2 and 3 equal in size. Five elongate ceratobranchials with cartilaginous tips. Ceratobranchial 3 with prominent concavity on posterior margin. Ceratobranchial 5 with 16 or 20 (2) conical, elongated, and pointed teeth, arranged in three irregular rows. Four epibranchials; anteriormost three elongated and narrow. Epibranchials 1 L shaped, with anterior pointed process; epibranchial 2 with mesial-anterior and distal-posterior process; epibranchial 3 with wider process on distal-posterior margin, slightly curved mesially. Epibranchial 4 rectangular. Pharyngobranchial 3 straight and elongated, shorter than hypobranchial 1. Pharyngobranchial 4 ossified and connected to curved plate with 23 or 28 (2) conical, elongated, and pointed teeth, arranged in up to three irregular rows; teeth increasing in size posteriorly (Figs. 4C–D).

Dorsal fin with eight pterygiophores (4) or seven (2), first inserted anterior to neural spine of 18th or 19th post-weberian vertebrae. Anal fin with six pterygiophores (6), first inserted anterior to hemal spine of 22nd post-weberian vertebrae. Procurrent caudal-fin rays 17(1), 18(1), or 19(4) dorsally and 12(1), 13(2), or 14(22) ventrally. Upper caudal plate with one unbranched ray and five branched rays; hypural 3 free and hypurals 4 and 5 fused to each other. Single lower caudal plate with one unbranched ray and six branched rays. Parhypural and hypurals 1 and 2 co-ossified and fused to compound caudal centrum. Post weberian vertebrae 34(1), 35(1), 36(3), 37(3), ribs 13(4), or 15(1).

Laterosensory system. Laterosensory canals with simple (non-dendritic) branches ending in single pores. Nasal and frontal branches of supraorbital canal continuous, with three paired pores s1, s3, and s6. Supraorbital pore s1 located at posterior portion of anterior nostrils, pore s3 at same longitudinal line of pore s1, posteriorly to posterior nostrils, and pore s6 aligned with posterior margin of eyes. Antorbital segment of infraorbital canal absent. Sphenotic canal present with two pores, i10 located behind eyes, and i11 located laterally to posterior margin of eye. Otic and postotic canals present with two pores associated: po1 located anterolaterally to opercular odontodophore and po2 located laterally to half-length of opercular odontodophore. Lateral line canal short with two* (36) pores located above pectoral-fin insertion and posterior to gill opening.

Coloration in alcohol. We differentiated two morphotypes for Cambeva perobana.

Morphotype I. Background of body and head consisting in a dark-gray to dark-brown coloration, becoming lighter towards ventral portion of body and head. Chromatophores slightly concentrated on dorsum, humeral, occipital, and opercular and interopercular regions. Some specimens with inconspicuous narrow dark-brown mid-lateral stripe, sometimes interrupted, extending from opercular odontodophores to base of caudal-fin rays (Figs. 1, 5A–C). Pectoral, anal, dorsal, and caudal fins with dark brown pigmentation, concentrated in proximal region in specimens larger than 40.0 mm SL. Pelvic fin unpigmented. Specimens smaller than 40.0 mm SL without pigmentation in all fins. Barbels with dark-brown pigmentation concentrated on dorsal surface.

Morphotype II. Background of body and head consisting in a yellowish to dark-brown coloration, becoming lighter towards ventral portion of body and head. Lateral and dorsal surface of body and head with irregular dark-brown blotches larger than two or three times orbit diameter, on inner skin layer, sometimes coalescing and forming larger irregular blotches. Some specimens with inconspicuous narrow dark-brown mid-lateral stripe, sometimes interrupted, extending from opercular odontodophore to base of caudal-fin rays (Figs. 5D, E). Pectoral, anal, dorsal, and caudal fins with dark brown spots in specimens larger than 40.0 mm SL. Pelvic fin unpigmented. Barbels with dark-brown pigmentation concentrated on dorsal surface.

Coloration in life. Similar to coloration in alcohol. In morphotype I, dark-gray to brown pigmentation more intense on flank and base of dorsal, anal, and pectoral fins (Fig. 6).

Molecular data. A total of 82 COI gene sequences (573 bp) were used in this study, including three of the newly described species (632 bp), corresponding to two sequences from morphotype I (NUP 24166, NUP 24167), and one sequence from morphotype II (NUP 25070). The final alignment revealed 142 polymorphic sites, of which 109 were informative. Genetic distances (K2P) between groups showed that Cambeva perobana had a distance of 1.2–1.4% from C. davisi and 1.6–1.8% from C. stawiarski. Full genetic distance results are listed in Tab. S3.

The best substitution models according to BIC were GTR+R for the ML tree and the TN+F+I+G4 for the Bayesian tree, both showed that Cambeva perobana formed a monophyletic group with C. davisi (from rio Iguaçu basin), sequences identified as C. barbosae (from rio Cubatão do Sul basin), and C. diabola (from rio Paranapanema basin) as sister groups. The last two species formed a cluster with variably internested sequences. Three out of the four delimitation methods tested supported the assignment of C. perobana as a distinct species. Only GMYC lumped it together with C. davisi, C. barbosae, and C. diabola (Fig. 7).

Geographical distribution.Cambevaperobana is known from the upper section of Paraná System, Paraná State, Brazil, in the rio Piquiri basin: rio Mouro (type-locality), rio Água Cinquenta e Cinco, rio Concórdia, rio Barreiro, rio Farol, rio Saquarema, Córrego Água da Granada, and stream tributaries of rio Goioerê, and in the rio Ivaí basin: rio dos Índios and rio Ligeiro (Fig. 8).

Ecological notes. The type-locality of Cambevaperobana is located at an elevation of 450 m, and the other localities are at 410 to 645 m above sea level. Substrate was composed of pebbles and rocks of 1 to 4 cm, and sand. The marginal vegetation was composed of predominant bushes and grassy banks (Fig. 9), where the specimens were found. The streams are surrounded by riparian vegetation, but only part of the rivers are inside the Perobas Biological Reserve. The new species was found in sympatry at the type-locality with Ancistrus sp., Cetopsorhamdia iheringi Schubart & Gomes, 1959, Characidium gomesi Travassos, 1956, Cichlasoma paranaense Kullander, 1983, Corydoras aeneus (Gill, 1858), Gymnotus inaequilabiatus (Valenciennes, 1839), Heptapterus sp., Hisonotus pachysarkos Zawadzki, Roxo & da Graça, 2016, Hypostomus ancistroides (Ihering, 1911), Neoplecostomus sp., and Rhamdia aff. quelen (Quoy & Gaimard, 1824).

Etymology. The specific epithet “perobana” is in allusion to the “Reserva Biológica das Perobas”, a conservation unit in the Paraná State, Brazil, and the area of the type-locality of the new species. The feminine Latin suffix “-ana”, which mens pertaining to, is added to the singular noun “Peroba”. An adjective.

Conservation status. Cambeva perobana is found in streams of the Reserva Biológica das Perobas, which is a Federal Conservation Unit and the major reminiscent forest in northwest Paraná, Brazil (Delariva, Silva, 2013), although there are pastures and crops, which can result in contamination of its water bodies by agricultural pollutants. Furthermore, some streams located in the vicinity of the reserve show apparent degradation, due to pollution from solid waste, however, we were not able to measure these impacts on C.perobana populations. The new species has an EOO of 1,863 km² (< 20,000 km² in criteria B1), and an AOO of 44 km² (< 2,000 km² in criteria B2). Although, C. perobana does not meet any other condition of this criteria, the new species can be classified as Least Concern (LC), according to the IUCN criteria and categories (IUCN Standards and Petitions Committee, 2024).

Morphometric analysis. The principal component analysis (PCA) returned the axis of component 1 with 26.6% of variation, component 2 with 16.1%, and component 3 with 11.9% of variation, totaling 54.6% of variation. There is a large overlap between Morphotypes I and II and no tendency of discrimination between them (Fig. 10). Measurements with higher positive loadings for PC1 are supraorbital pores s6 distance and pelvic-fin length, for PC2 are nasal-barbel length and supraorbital pores s6 distance, and for PC3 are pelvic-anal distance and anal-fin length (Tab. 2).

The color pattern of Cambeva species has been used to diagnose several species and has been shown to be diagnostic for species as C. castroi, C. diabola, C. horacioi, C. melanoptera, and C. piraquara (see de Pinna, 1992b; Bockmann et al., 2004; Reis et al., 2020b, 2023; Costa et al., 2022). Although color pattern variation initially suggests a greater number of species in Cambeva, it is important to be cautious when using it as a diagnostic to not overestimate species number (Silva et al., 2010; Nascimento et al., 2017; Donin et al., 2022). The same situation applies to C. perobana, as it exhibitsa high intraspecific variation in the body color pattern and is therefore divided into Morphotype I and Morphotype II (see remarks and coloration in alcohol for more information and description of the morphotypes). On the other hand, the similarity in the color pattern between C. perobana Morphotype II and C. davisi (see type-series in Nascimento et al., 2017) could hide the diversity of the genus, however the use of osteological data allows separation between species that are similar in external morphology.

Variation in color pattern of C. perobana does not seem to be due to ontogeny, since juvenile and adult specimens have both morphotypes (see Tab. 1 for SL range in both morphotypes). According to our results in both meristic, morphometric (PCA), molecular (species delimitation), and osteology, the morphotypes can be associated with a unique species, described here. Therefore, our results demonstrate how integrating the tools for delimiting Cambeva species helps to avoid overestimating the number of species in the group (Silva et al., 2010; Ferrer, Malabarba, 2013; Nascimento et al., 2017; Reis et al., 2020b; Donin et al., 2022).

Cambevaperobana specimens formed a monophyletic group in both ML and Bayesian trees and had C. davisi (from rio Iguaçu basin) as the genetically closer species with 1.3% of mean genetic divergence between the species. These findings are consistent with Pereira etal. (2013) observations of interspecific values <2% for Trichomycterinae. Despite the low genetic distance, the majority of species delimitation methods indicated that C. perobana is a distinct and unique species, including the two morphotypes found in this study. Although the GMYC method combined the new species with C. davisi, C. diabola, and C. barbosae, its results were also more conservative with other species already defined based on morphology and molecular methods (e.g., C. stawiarski, C. cauim, C. chrysornata, C. cubataonis, and C. guaratuba). The disparity between delimitation methods with different approaches is expected in species with high speciation rates, and the consensus should be accepted as the more parsimonious result (Dellicour, Flot, 2018).

The distribution of Cambevaperobana shows a similar pattern to other species occurring in both Piquiri and Ivaí basins, such as Apareiodonvladii Pavanelli, 2006, Planaltina kaingang Deprá, da Graça, Pavanelli, Avelino & Oliveira, 2018 (Reis et al., 2020a), which can be explained by the exchange of the ichthyofauna through headwater capture in such basins (Morais-Silva et al., 2018). The discovery of the new species in the rio Ivaí basin, located near the headwaters of rio Piquiri (ca. 5 km), further supports this hypothesis.

In addition to Cambeva perobana, six congeneric species, C. diabola, C. pascuali, C. horacioi, C. iheringi, C. guareiensis and C. paolence were described from the upper rio Paraná basin (Fricke et al., 2023), and only C. horacioi is sympatric with C. perobana (see Eigenmann, 1917; Bockmann et al., 2004; Ochoa et al., 2017; Katz, Costa, 2020; Reis et al., 2020b). Some studies suggest that part of the biodiversity found in the Atlantic Forest biome is present in the upper rio Paraná basin (Agostinho et al., 2007; Delariva, Silva, 2013; Delariva et al., 2014), which is one of the most extensively studied areas compared to other Brazilian basins, although the number of species that have not been formally described still account for approximately 10% (Agostinho et al., 2007; Reis et al., 2020a). Therefore, knowing and describing these fishes species from the region is an important factor for conservation of the upper rio Paraná basin, which is situated in one of the largest industrial and urban regions in South America and has been suffering from anthropogenic impacts (Agostinho et al., 2008). Considering the description of C. perobana, which is the first species of the genus in the rio Piquiri and the second in the rio Ivaí, it is essential to emphasize the importance of these rivers, which in their headwater regions probably still harbor species to be described (see possible new species in Delariva et al., 2013; Frota et al., 2016; Cavalli et al., 2018; Reis et al., 2020a).

Comparative material examined. All from Brazil.Cambeva cauim: rio Iguaçu basin: MPEG 39109, 4 paratypes, 29.2–82.3 mm SL. NUP 2416, 19, 20.0–97.4 mm SL. NUP 22756, holotype, 89.6 mm SL. Cambeva crassicaudata: rio Iguaçu basin: NUP 3783, 3, 78.6–134.5 mm SL. Cambeva davisi:rio Iguaçu basin: NUP 15994, 47, 19.6–80.1 mm SL. NUP 15914, 67.2–78.2 mm SL. Rio Ribeira de Iguape basin: NUP 17409, 8 (2 c&s), 31.7–63.1 mm SL. Cambeva paolence: rio Tietê basin: LBP 7684, 1, 54.1 mm SL. Cambeva papillifera:rio Iguaçu basin: NUP 1615, 1 paratype, 94.4 mm SL. Cambeva plumbea:rio Iguaçu basin: NUP 1614, 3 paratypes, 72.5–78.5 mm SL. Cambeva taroba:rio Iguaçu basin: NUP 1616, 3 paratypes, 47.4–53.8 mm SL.