Cichlasomatini is a well-supported clade among the seven Cichlinae tribes proposed by Ilves et al. (2018; viz. Cichlini, Retroculini, Astronotini, Chaetobranchini, Geophagini, Cichlasomatini, and Heroini). However, some of the relationships within this tribe are still unclear. One of the most important issues related to Cichlasomatini is the monophyly of Cichlasoma Swainson, 1839 and Aequidens Eigenmann & Bray, 1894, the two oldest generic names inside Cichlasomatini. Cichlasoma is nested within Aequidens (Musilová et al., 2008; Ilves et al., 2018) resulting in a paraphyletic group. However, Aequidens isdistinguished from Cichlasoma by morphological characteristics such as naked, rather than scaled dorsal and anal fins longer caudal peduncle with 2–3 vertebrae (2 or 0 in Cichlasoma), and 4–5 vertical bars from the midlateral spot to the caudal peduncle in Aequidens (6 in Cichlasoma) (Kullander, 1986). The uni or triserial predorsal squamation, the presence of a lateral band on the middle of the flank, and some meristic traits, such as a long peduncle with more vertebrae, are characteristics that differentiate Aequidens species from other Neotropical cichlid genera (Kullander, 1983, 1986; Kullander, Nijssen, 1989). The phenotypic character states used to distinguish species within Aequidens are mainly related to coloration, such as differences in the patterns of the longitudinal stripe and flank bars, cheek spot, and buccal stripes. The morphological characters traditionally used in species delimitation in Aequidens (and other Cichlasomatini as well) show a high degree of plasticity, such as meristic and morphometric. This makes the delimitation of the range of variation and, consequently the species delimitation, difficult and dependent on authoritative argumentation or evidence from other independent sources of data.

Aequidens comprises 17 valid species (Fricke et al., 2024) distributed in the rios Amazon-Orinoco-Guiana (AOG) region, and in the rio La Plata basin (Paraná-Paraguay) (Kullander, 2003; Hernández-Acevedo et al., 2015). The majority of Aequidens species were described from the Amazon-Orinoco-Guyana area (AOG region of van der Sleen, Albert, 2018). Aequidens plagiozonatus Kullander, 1984 is the only species known from the río La Plata basin (Paraná-Paraguay), occurring also in the rio Guaporé together with A. viridis (Heckel, 1840). The records of this species in other hydrographic basins are deemed non-native occurrences (Reis et al., 2020).

The rio Correntes is a tributary to the upper rio Paraguai basin draining the border regions of the Mato Grosso and Mato Grosso do Sul states, Central Brazil. In the surroundings of the city of Sonora, Mato Grosso do Sul, the river flows through a sinkhole (around 17°36’49”S 54°50’14”W) into a natural karst tunnel about 800 m long (SEMA, 2005). Along with the headwaters of the rio Araguaia basin upstream from Barra do Garças in Mato Grosso and the upper rio Taquari basin, also a tributary to the upper rio Paraguai basin, the upper rio Correntes basin (upstream the sinkhole) is part of one of the smallest endemic regions for Neotropical freshwater fishes (Dagosta et al., 2020). Currently, four species are known to be endemic to the upper rio Correntes basin: Melanorivulus dapazi (Costa, 2005), Eigenmannia correntes Campos-da-Paz & Queiroz, 2017, Characidium chicoi da Graça, Ota & Domingues, 2019, and Cyphocharax caboclo Melo, Tencatt & Oliveira, 2022.

During survey efforts in the rio Correntes basin (2018 to 2021), an undescribed species of Aequidens was captured exclusively in the upper portion of this basin (i.e., upstream from the sinkhole). In this paper, we combine morphological and molecular evidence to recognize and describe this new species of Aequidens, and discuss the occurrence of A. plagiozonatus in the rio Araguaia basin based on recent records in fish collections.

Morphological data. Measurements and counts were obtained under a stereomicroscope, in accordance with Ota et al. (2021) and Deprá et al. (2022). The following measurements are included: preanal distance, from the premaxillary symphysis to the base of the first anal-fin spine; distance from dorsal to caudal fin, from the base of the first dorsal-fin spine to the vertical through distal margin of the hypurals, between the dorsal and ventral lobes; distance from dorsal to anal fin, from the base of the first dorsal-fin spine to the base of the first anal-fin spine; distance from dorsal to pelvic fin, from the base of the first dorsal-fin spine to the base of the pelvic-fin spine; distance from dorsal to pectoral fin, from the base of the first dorsal-fin spine to the base of the first pectoral-fin ray; caudal peduncle length (straight), from the vertical line through the base of the last dorsal-fin ray, on the intersection with the lower lateral line, to distal margin of the hypurals; caudal peduncle length (oblique), from the base of the last anal-fin ray to distal margin of the hypurals; dorsal-fin base length (spine), from the base of the first dorsal-fin spine to the base of the last dorsal-fin spine; dorsal-fin base length (total), from the base of the first dorsal-fin spine to the base of the last dorsal-fin ray; anal-fin base length (spine), from the base of the first anal-fin spine to the base of the last anal-fin spine; anal-fin base length (total), from the base of the first anal-fin spine to the base of the last anal-fin ray. The series of scales between the dorsal-fin base and the upper lateral line were counted starting from: a) the scale reaching the base of the first spine (anterior series); and b) the scale reaching the base of the last spine (middle series). The scales on the lower lateral line are made up of the following: the number of scales to the distal end of hypurals plus the scales onto caudal fin (usually two scales). Counts from holotype are indicated by an asterisk. The counts are followed by their frequency in parentheses, unless when equal for all specimens. Measurements expressed as a percentage of head length.

Two specimens were cleared and stained (c&s) following the methodology of Taylor, Van Dyke (1985) for osteological examination. Vertebrae were counted as in Deprá et al. (2021). Fused PU1+U1 (pleural and ural vertebrae) was treated as a single bone. The lower pharyngeal tooth plate was examined as described by Barel et al. (1976).

We adopted the coloration-related terminology described by Říčan et al. (2005), which defined the homology and aspects of ontogeny of nine flank bars along the body. Specimens were deposited in the following Brazilian institutions: Coleção Ictiológica de Três Lagoas, Universidade Federal do Mato Grosso do Sul, Três Lagoas (CITL); Coleção Ictiológica do Núcleo de Pesquisas em Limnologia, Ictiologia e Aquicultura, Universidade Estadual de Maringá, Maringá (NUP); Coleção de Peixes da Universidade Federal do Mato Grosso, Cuiabá (CPUFMT); Museu de Ciências e Tecnologia, Pontifícia Universidade Católica do Rio Grande do Sul, Porto Alegre (MCP); Museu de Zoologia, Universidade Estadual de Londrina, Londrina (MZUEL); and Coleção do Laboratório de Biologia e Genética de Peixes da Universidade Estadual Paulista, Botucatu (LBP). To infer the IUCN criteria, the software GeoCAT (Geospacial Conservation Assessment Tool; http://geocat.kew.org) allowed the calculation of the Extent of Occurrence of the new species based on the currently known points of occurrence.

Molecular data. Sampling and sequencing. We obtained partial sequences of the cytochrome c oxidase subunit I (COI) gene of the new species, of Aequidensplagiozonatus from the Paraguai and rio Araguaia basins, and of A. gerciliae Kullander, 1995. The sequences of other Aequidens species were obtained from the GenBank database.

Voucher specimens were fixed in 10% formalin and then stored at 70º GL ethanol. Tissue samples were extracted from specimens prior to their fixation and preserved and stored at 95º GL ethanol. Total DNA was extracted following Ivanova et al. (2006). Subsequently, the DNA barcode region of the COI was amplified by polymerase chain reaction (PCR) using the primers FISH-F6 (5’-ACYAAYCACAAAGAYATTGGCA-3’) and FISH-R7 (5’-TARACTTCTGGRTGDCCRAAGAAYCA-3’) described by Jennings et al. (2019). The PCR was performed in a thermocycler with a final volume of 12.5 μl containing 7.85 μl distilled water (ddH2O), 0.30 μl deoxynucleotide triphosphate (dNTP) (2 mM), 1.25 μl PCR buffer (10×), 0.4 μl MgCl2 (50 mM), 0.25 μl each primer, 2 μL DNA (200 ng) and 0.20 μl Taq DNA polymerase PHT (Phoneutria). PCR was performed under the following conditions: an initial denaturation at 95 ºC for 5 min, followed by 30 cycles including denaturation at 95 ºC for 60 s, annealing (primer hybridization) at 52 ºC for 45 s and nucleotide extension at 68 ºC for 1 min, with a final extension at 68 ºC for 10 min. The PCR products were amplified and checked on a 1% agarose gel before being purified using ExoSAP-IT (USB Corporation, Cleveland, OH, USA) according to the manufacturer’s protocol.

To sequence both DNA strands, the purified products were used as templates. Further, the cycle sequencing reaction was accomplished using a BigDye Terminator v. 3.1 Cycle Sequencing Ready Reaction kit (Applied Biosystems, Austin, TX, USA) in a final volume of 7 μl containing 0.1 μl of purified PCR product, 0.35 μl primer (10 mM), 1.05 μl buffer 5×, 0.7 μl BigDye mix and 3.9 μl distilled water. The final mixture of the purified PCR product was then purified again using ethanol precipitation and sequenced using the 3500-Genetic Analyzer (Applied Biosystems) at IBTEC, at the Instituto de Biociências at Universidade Estadual Paulista “Júlio de Mesquita Filho”, UNESP, Botucatu.

Molecular analysis. Sequences of COI were obtained from nine specimens (two of the new species, six of A. plagiozonatus and one of A. gerciliae) and 16 additional sequences were obtained from GenBank, totalizing 25 sequences in the final matrix (23 of Aequidens and two of Pterophyllum Heckel, 1840). All sequences and GenBank accession numbers used in this study are described in Tab. S1. Species added to the analysis were: Aequidensdiadema (Heckel, 1840) (rio Amazonas basin in Iquitos), A. epae Kullander, 1995 (rio Tapajós basin), A. tetramerus (Heckel, 1840) (French Guiana and Suriname), A. pallidus (Heckel, 1840) (rio Negro and Jatapu basins), A. gerciliae (rio Tapajós basin), A. michaeli Kullander, 1995 (rio Xingu basin), Cichlasoma dimerus (Heckel, 1840) (Lower rio Paraná basin), C. orientale Kullander, 1983 (Atlantic mid-north-eastern Rivers), C. paranaense Kullander, 1983 (Upper rio Paraná basin) and C. sanctifranciscense Kullander, 1983 (Atlantic mid-north-eastern Rivers). Pterophyllum altum Pellegrin, 1903 was added as the outgroup. For the sequences obtained from GenBank we followed the identification from original studies. Sequences were aligned using Clustal W algorithm (Thompson et al., 1994) incorporated in MEGA X (Kumar et al., 2018). In MEGA X, sequences were translated into amino acids for the verification for stop codons along the sequences, which were absent.

For species delimitation, four approaches were used: i) Poisson Tree Process analysis (PTP; Zhang et al., 2013); ii) General Mixed Yule Coalescent Model analysis (GMYC; Pons et al., 2006; Fujisawa, Barraclough, 2013); iii) Automatic Barcode Gap Discovery analysis (ABGD; Puillandre et al., 2012); and iv) Assemble Species by Automatic Partitioning (ASAP; Puillandre et al., 2021).

A maximum likelihood (ML) analysis was performed in MEGA-X v. 10.2.1 (Kumar et al., 2018) using the best nucleotide substitution model (K2+G, lowest BIC score: 4685,00 with 91 parameters), five random searches and 1,000 bootstrap replicates and other parameters at default, showing the best tree. The resulting ML tree was used as an input tree for PTP analysis performed on the PTP web server (species.h–its.org/server) using 500,000 MCMC generations with 0.1 burn-in rate.

For GMYC analysis, a Bayesian Inference analysis was performed to construct a phylogenetic tree, inserting the fasta file in BEAUTi (Drummond et al., 2012), selecting the HKY+G+I nucleotide evolutionary substitution model, the uncorrelated relaxed clock and the speciation birth-death model (Yule process), on an arbitrary timescale. A random tree was used as a starting tree for the MCMC searches with a run of 10 million generations, and a tree sampled every 1,000 generations and posteriorly running the analysis in BEAST v. 1.8.4 (Drummond et al., 2012). Following that, Tracer v. 1.7.2 (Rambaut et al., 2018) was used to investigate the distribution of log-likelihood scores and the identification of the stationary phase for each search (to determine whether more runs were necessary to achieve convergence). In the burn-in procedure, the sampled topologies below the asymptote were removed (10%), and the remaining trees were used to construct a 50% majority-rule consensus tree in TreeAnnotator v. 1.8.4. (included in BEAST). The resulting ultrametric tree was displayed in FigTree v. 1.4.3 (Rambaut, 2019) and a newick archive exported was employed as an input file for the GMYC analysis done using a single threshold method at the GMYC webserver (species.h–its.org/gmyc/R).

The ABGD analysis (Puillandre et al., 2012) was performed on the ABGD web server (bioinfo.mnhn.fr/abi/public/abgd/abgdweb.html), inserting the fasta file with the aligned sequences, using Kimura (K2P; 2.0) distance model and other parameters at default (Pmin = 0.001; Pmax = 0.1). In addition, the ASAP analysis (Puillandre et al., 2021) was performed on the ASAP web server (https://bioinfo.mnhn.fr/abi/public/asap/asapweb.html), inserting the fasta file with the aligned sequences, using Kimura (K80; 2.0) distance model.

Aequidens pirilampo,new species

urn:lsid:zoobank.org:act:6F182F43-5BFE-46B4-B7DE-6580FE19C11F

(Figs. 1–5; Tab. 1)

Aequidens. —Melo et al., 2022:332 (Brazil, Mato Grosso, Ribeirão Comprido; first record).

Aequidens sp. —Gimênes, Rech, 2022:595 (Brazil, Mato Grosso, photo).

Holotype. NUP 23543, 94.7 mm SL, Brazil, Mato Grosso State, Itiquira Municipality, ribeirão Comprido, tributary of the rio Correntes, rio Paraguai basin, 17º32’04”S 54º25’36”W, 27 Jan 2021, L. F. C. Tencatt & M. N. Souza.

Paratypes. All from Brazil, rio Paraguai basin. Mato Grosso State: Itiquira Municipality: LBP 30742, 1, 42.1 mm SL, 18 Sep 2021, L. F. C. Tencatt, M. N. Souza & M. A. Alves; LBP 30743, 1, 48.3 mm SL, 5 Apr 2018, L. F. C. Tencatt, M. N. Souza & M. A. Alves; LBP 31524, 4, 41.8–94.0 mm SL, 18–21 Sep 2021, L. F. C. Tencatt, M. N. Souza & M. A. Alves, same data as holotype. MCP 54880, 4, 42.6–67.0 mm SL, MZUEL 22500, 4, 43.0–76.6 mm SL, NUP 23546, 5, 39.8–66.2 mm SL, NUP 23547, 2 c&s, 59.2–59.6 mm SL, collected with LBP 31524. NUP 21644, 9, 28.7–68.5 mm SL, córrego de Cima, tributary of the rio Correntes, 17°39’52”S 54°14’46”W, 31 Aug 2018, Nupélia staff. NUP 23544, 7, 14.9–87.2 mm SL, collected with holotype. Mato Grosso do SulState: Sonora Municipality: CITL 392, 10, 24.5–74.5 mm SL, córrego de Baixo, tributary of the rio Correntes, 17º42’33”S 54º21’32”W, 7 Aug 2021, L. F. C. Tencatt, M. N. Souza, V. Carvalho & M. A. Alves. CITL 393, 4, 17.3–39.3 mm SL, unnamed stream, tributary of the rio Correntes, 17º35’42”S 53º53’33”W, 29 Dec 2020, L. F. C. Tencatt & M. N. Souza. CPUFMT 7763, 2, 64.8–66.1 mm SL, same data as CITL 392, 2 Oct 2021, L. F. C. Tencatt, M. N. Souza & M. A. Alves.

Diagnosis.Aequidens pirilampo is distinguished from all congeners, except A. plagiozonatus, by having anteriorly oblique dark brown flank bars (vs. vertical). The new species differs from A. plagiozonatus by having the dorsal head contour, from the tip of the snout to the vertical through the posterior margin of the eye, almost straight, except for a conspicuous concavity at the interorbital region (Fig. 2) (vs. dorsal head contour convex, with a subtle concavity at the interorbital region in occasional specimens), by having longer lower jaw (40.2–46.9% HL and 16.7–18.5% SL vs. 35.2–39.3% HL and 13.2–15.2% SL in A. plagiozonatus), and longer upper jaw, (12.2–15.1% SL vs. 9.3–12.1% SL in A. plagiozonatus). Additionally, A. pirilampo is distinguished from its congeners, except A. chimantanus Inger, 1956, A. diadema, A. epae, A. mauesanus Kullander, 1997, A. michaeli, A. patricki Kullander, 1984, A. plagiozonatus, A. potaroensis Eigenmann, 1912, and A. tubicen Kullander & Ferreira, 1991, by having a discontinuous lateral band in fixed specimens (vs. continuous). Aequidens pirilampo differs from A. diadema and A. epae by having the lateral band divided into blotches more conspicuous at the encounter with flank bars (vs. lateral band not divided into blotches), and by the absence of unpigmented areas anteriorly and posteriorly to the midlateral spot, not forming a whitish circle above the longitudinal stripe (vs. presence). Aequidens pirilampo, also differs from A. epae by having the dorsal portion of the flank bar 5 anteriorly inclined and less conspicuous than the midlateral spot (vs. posteriorly inclined and as dark as the midlateral spot). Aequidens pirilampo also differs from A. rondoni (Miranda Ribeiro, 1918)by having eight flank bars (1a, 1p, 2, 3, 4, 5, 6, and 7; Fig. 4) instead of nine (1a, 1p, 2 bars, 3–4 expressed as three bars, 5, 6, and 7). Aequidens pirilampo differs from A. mauseanus, A. michaeli and A. potaroensis by having the lateral band equally conspicuous along its entire length (vs. lateral band more inconspicuous anterior to midlateral spot in A. mauseanus and A. potaroensis, and posterior to midlateral spot in A. michaeli); from A. mauseanus by the absence of a dark-brown blotch dorsal to midlateral spot (vs. presence); from A. michaeli by having a cheek spot (vs. absence), and by the absence of iridescent vermiculations on the cheek, opercle and pectoral girdle (vs. more ornamented pattern with stripes present in the cheek, opercle and pectoral girdle); from A. paloemeuensis Kullander & Nijssen, 1989 and A. potaroensis by the suborbital stripe not retained in adult specimens (vs. retained). Aequidens pirilampo differs from A. patricki by the absence of large, dark-brown blotches on the cheek and opercle (vs. presence), and by the brown lachrymal region (vs. unpigmented). Aequidens pirilampo differs from A. tubicen by the absence of a preopercular spot (vs. presence). Additionally, Aequidens pirilampo differs from A. gerciliae by the round midlateral spot (vs. longitudinally elongated); from A. metae Eigenmann, 1922 by the cheek spot not spanning the entire anterior margin of the vertical arm of the preopercle (vs. spanning the entire anterior margin of the vertical arm of the preopercle); from A. superomaculatum Hernández-Acevedo, Machado-Allison & Lasso, 2015 by the absence of a posterodorsal blotch on the flank (vs. presence).

Description. Based on holotype and paratypes. Measurements in Tab. 1. See also Figs. 2–4 for details of shape and color pattern. Body laterally compressed. Predorsal contour ascending straight from tip of snout to vertical through posterior margin of orbit. Moderately convex from this point to end of dorsal fin; straight along caudal peduncle. Prepelvic contour descending straight or slightly convex from tip of snout to vertical through posterior margin of preopercle, at same angle of predorsal contour (young specimens have more obtusely angled predorsal contour than prepelvic). Abdominal contour straight or slightly convex and horizontal. Anal-fin base contours straight and oblique. Caudal peduncle ventral contour straight or slightly concave and horizontal.

Head elongate, triangular in lateral view, with dorsal margin, from tip of snout to end of supraoccipital, ascending straight and oblique, ventral contour less steep than dorsal contour. Groove between eyes. Snout long, with frontal contour elongated and continuous with dorsal and ventral contour of head. Lips thick and of “American type”, i.e., lower lip fold covers upper lip. Tip of maxilla almost reaching vertical through anterior margin of eye. Nostril dorsolaterally situated, below horizontal through lower margin of orbit, halfway between tip of snout and orbit (in young specimens, closer to eye, horizontal through middle of eye). Orbit large, situated on dorsal half of head, pupil ventral to level of upper lateral line. Posterior margin of preopercle, opercle, subopercle, interopercle and suprachleitrum smooth, without serrations.

E1 scales 24(18), 25*(14) or 26(1). Scales between upper lateral line and dorsal-fin 3* (24), 3½ (9) at base of first dorsal-fin spine, 1½(27), 2*(5), 2½(2) at base of last dorsal-fin spine. Scale rows between lateral line 2*(27). Scales on lateral line 13/8(1), 14/8(1), 15/7(2), 15/8(2), 16/6*(2), 16/7(5), 16/8(5), 17/6(1), 17/7(2), 17/8(5); additionally 1(4), 2*(23) or rarely 3(1) scales of lower lateral line onto caudal fin. Cheek scales in 3*(9), 3½(10), or 4(10) rows, ctenoid. Opercle scales 4(1), 5(2) or 6*(23), large and cycloid, stochastically arranged. Subopercle covered with 3*(32) cycloid scales. Interopercle with 2(15), 3*(14) or 4(2) scales embedded in skin. Scales absent from preopercle. Infraorbital scales 5(3), 6*(25) or 7(4). Circumpeduncular scale rows 16*(35), including lateral line scales. Predorsal scales uniserial(2), biserisal(2), triserial*(24) or stochastic(1), cycloid, slightly smaller than flank scales. Flank scales ctenoid. Prepelvic scales 6(1), 7(3), 8(4), 9(1), 10*(12), 11(7), 13(1), ctenoid, same size towards gular region. Abdominal scales ctenoid, slightly smaller than flank scales. Pectoral, pelvic, dorsal, and anal fins without scales. Caudal-fin base covered with stochastically distributed transition scales, intermediate in size between peduncular and inter-radial scales; caudal fin with cycloid inter-radial scales from base of rays to ⅓ of its length; series increasing ontogenetically, 2–5 scales in specimens up to 40 mm covering from basal ⅛ to basal ⅙, 4–14 scales in specimens over than 40 mm SL, covering from basal ¼ to basal ⅓, without secondary series; one specimen (NUP 21644, 68.5 mm SL) with series of scales with pores and canals between rays D3–D4 (8 scales) and V3–V4 (8 scales).

Dorsal-fin rays XIV.10(1), XIV.11(3), XV.9(2), XV.10(19), XV.11*(7); dorsal spines increasing in size up to 6th, first spine about one-fourth length of last. Dorsal-fin rays 4–5 forming filament reaching over ⅔ of caudal fin length in some specimens; lappets rounded to pointed, with posterior margin free, slightly surpassing tip of spines. Anal-fin rays III.7.i(3), III.8(4), III.8.i(19), III.8.ii(2), III.9*(4) or IV.9(1); middle rays longest, pointed, in some specimens forming filament reaching up to half of caudal-fin length. Caudal-fin rounded, with 14*(32) principal rays. Total pectoral-fin rays 12(2), 13(13), 14*(18) or 15(1). Pectoral fin ventrally rounded, dorsally pointed, fourth ray longest, in some specimens passing posterior margin of midlateral spot. Pelvic-fin rays I.5*(34); second ray longest, with filamentous extension, reaching or passing anal-fin origin.

Gill rakers externally on first epibranchial 1(10) or 2*(24); 0(2) or 1*(32) on angle; 4(8), 5*(20), 6(3) or 7(1) on ceratobranchial 1.

Teethunicuspid, decreasing gradually from symphysis. Symphysis of both jaws lacking teeth. Upper jaw series with 3–4 rows; lower jaw series with 2–4 rows. External hemiseries of upper jaw right/left sides with 9–13/11–12 in specimens up to 40 mm; 10–22/6–23 in specimens over 40 mm. External hemiseries of lower jaw right/left sides with 16–21/17–21 in specimens up to 40 mm; 17–25/11–26 in specimens over 40 mm.

Suture between contralateral ceratobranchial 5 not including interdigitations ventrally; posteromedial teeth large, cylindrical, with large, blunt, dorsally oriented cusp and 1–2 very small, anteriorly oriented cusps; anterolaterally, teeth gradually diminishing in size; outer teeth compressed transversally, with large cusp and 1–2 very small, upward cusps. Lower pharyngeal jaw tooth-plate (ceratobranchial 5) (Fig. 5) length including posterolateral processes 89.9% of width; length of dentigerous area 59.8% of width; 13 teeth along posterior margin each side;seven teeth along symphyseal margin; 22 teeth along outer margin. Pharyngobranchial 2 with seven teeth arranged in two rows, posteriormost ones larger, turned anteriad. Pharyngobranchial 3 with 34 teeth arranged in 6 rows turned backwards. Tooth plate 4 with 51 teeth arranged in 7 rows turned posteriad; three concavities in the frayed zone at the posterior margin. Ceratobranchial 4 with 2 tooth plates, posterior with 5–6 teeth, anterior with 6–9 teeth. Two supraneurals, anterior to first neural spine. Twenty-seven total vertebrae, of which 13(2) abdominal (4 and 5 type A’, others type A) and 13(2) caudal (first 12 type and PU1+U1). Three posteriormost vertebrae (CV3, CV2 and PU1+U1) completely included in the caudal peduncle. Ribs 10, abdominal. Vertebrae bearing ribs, 3rd–12th. Epineurals absent. Twenty-four dorsal-fin pterygiophores (one spine or ray for each pterygiophore), surrounded by vertebrae 1–20. Ten anal-fin proximal pterygiophores (first one bearing first two spines; last one bearing last two rays), surrounded by vertebrae 13–20 (anteriormost pterygiophore touches the anterior margin of haemal spine of 14th vertebra). Two epurals. One uroneural. Five branchiostegal rays. First branchial arch with 7 outer rakers (one on epibranchial, one on angle, and five on ceratobranchial) and 11 inner rakers (2 on epibranchial, one on angle, and 8 on ceratobranchial). Second arch with 12 external rakers (two on epibranchial, one between epibranchial and ceratobranchial, eight on ceratobranchial, and one between ceratobranchial and hypobranchial) and 8 inner rakers (on ceratobranchial). Third arch with 11 external rakers (one on epibranchial, one between epibranchial and ceratobranchial, and nine on ceratobranchial) and 11 inner rakers (one on epibranchial, two between epibranchial and ceratobranchial, and eight on ceratobranchial). Fourth arch with 11 external rakers (one on epibranchial one between epibranchial and ceratobranchial, and nine on ceratobranchial) and 10 inner rakers (epibranchial lacking rakers, all on ceratobranchial). Two procurrent caudal-fin rays dorsally and three ventrally. Microbranchiospines on all four branchial arches.

Coloration in alcohol.Background light beige to yellowish brown; ventral region yellowish white; dorsal region dark brown. Posterior margin of flank scales with diffuse brown pigmentation. Head brownish on neurocranial region and nape; yellowish-brown on cheek, preopercle and opercle region; and yellowish white ventrally. Cheek spot dark brown. Three oblique, beige stripes continuous across dorsal midline of head. First along anterodorsal margin of lachrymal, from tip of snout to anterior margin of orbit, through nostril. Second from near anteroventral margin of lachrymal to antero-ventral margin of orbit. Third along posteroventral margin of lachrymal and infraorbitals. Stripes absent from cheek, gill cover and pectoral girdle. Nine bars along body being eight bars on flank and one (bar 9) on head (Figs. 4A, B): bar 1p, at caudal-fin base, forming blotch; bar 1a, on distal portion of caudal peduncle; bar 2, at vertical through last soft dorsal- and anal-fin rays; bar 3, at vertical through first soft dorsal- and anal-fin rays; bar 4, at vertical through last dorsal-fin spine and anal-fin spines; bar 5, usually at vertical through 8th–11th dorsal-fin spines and vent; bar 6, usually at vertical through 4th–8th dorsal-fin spines and posterior to pelvic-fin base; bar 7, usually at vertical through base of first dorsal-fin spines to pectoral-fin origin; bar 8, on nape and, more diffusely, on opercle; bar 9, formed by supraorbital and infraorbital bars, the latter present only in young specimens (Figs. 4C–F). Lateral band dark brown, divided into blotches at intersections with flank bars, mainly concentrated along E1 scale series. Midlateral spot on bar 5, covering E1 and E2 series. Beige to brown fins. Dorsal fin lighter on distal margin; small, rounded white blotches on soft portion, in occasional specimens forming oblique stripes. Anal fin with same pattern as dorsal fin. Caudal fin lighter on distal margin; small white blotches more concentrated on anterior two thirds, forming dotted or striped pattern (better viewed in fixed and smaller specimens) (Fig. 4); one black blotch, usually ocellated, corresponding to bar 1p, at base of all rays of dorsal lobe.

Coloration in life.Based on Fig. 3 and field observations by LFCT (2018–2021). Background pale to light beige; ventral region whitish. Head brownish in dorsal region, with silvery scales on nape; lachrymal and cheek silvery, border of the scales yellow to brownish, with black cheek spot posteroventral to orbit, between third oblique stripe below lacrimal and preopercle upper portion; preopercle and interopercle silvery, subopercle light beige; opercle light brownish, border of the scales yellow to silvery brownish; middorsal portion typically with brownish orange patch; ventral region light beige. Body covered by green, yellow, and blue iridescent coloration. Scales on flank silvery, forming a more evident reticulated pattern, with brownish borders at posterior margin. Pattern of dark blotches, stripes, and bars less evident than in preserved specimens, but similar in relation to number and position of color-pattern elements. Fins yellowish. Dorsal fin with white blotches at membranous region between spines/rays, with distal portion yellow. Anal fin with distal portion yellow, blotches as same pattern as preserved specimens. Pelvic fin yellowish at first spine and rays region, proximal region light beige, distal region yellowish. Caudal fin yellowish, base with blue iridescent spots/stripes.

Geographical distribution.Aequidens pirilampo is currently only known from tributaries of the upper rio Correntes, a tributary of the rio Paraguai basin, in the border region between Mato Grosso and Mato Grosso do Sul states, Brazil (Fig. 6).

Ecological notes. General view on the collecting sites of Aequidens pirilampo in Fig. 7.The new species lives in lentic mesohabitats with depths ranging from around 10 cm (particularly hatchlings) to slightly over 1 m with a lot of submerged vegetation (macrophytes in general, including those floating in some points). Unlike other Aequidens species, which are usually found in brownish water (tea-colored water) environments, the upper rio Correntes and its tributaries present extremely crystalline water except for environments degraded by anthropic impacts, allowing the observation of solitary fish foraging close to submerged vegetation during the day. The species is found in syntopy with Astyanax sp., Characidium chicoi, Characidium aff. zebra Eigenmann, 1909, Cnesterodon cf. septentrionalis Rosa & Costa, 1993, Cyphocharax caboclo, Eigenmannia correntes, Hyphessobrycon sp., Hypostomus sp., Melanorivulus cf. dapazi (Costa, 2005), and Moenkhausia lopesi Britski & Silimon, 2001. The subterranean stretch (or the sinkhole itself) of the rio Correntes is hypothesized to be a barrier to the fish fauna of the basin. However, Aequidens pirilampo,as well as the other four endemic species described from the upper rio Correntes basin, were not found in all sampling sites upstream the sinkhole during the four years of survey efforts in the region. In the Ponte de Pedra hydroelectric power station Reservoir (just upstream of the sinkhole), none of these endemic species was captured. At this site, only A. plagiozonatus was found (Fig. 8A), along with several other species common to most tributaries of the upper rio Paraguai basin (e.g., Saxatilia lepidota (Heckel, 1840), Hyphessobrycon herbertaxelrodi Géry, 1961, Megalechis thoracata (Valenciennes, 1840), Metynnis cf. maculatus (Kner, 1858), and Satanoperca pappaterra (Heckel, 1840), including some introduced species like Cichla spp. and Colossoma sp. (the last one not captured, with occurrence confirmed by several locals). Considering the currently available data, it seems that a large waterfall (estimated to be at least 50 m high; 17°32’06”S 54°26’01”W) in the area of a small hydroelectric power plant represents the downstream limit of occurrence of the new species. Aequidens pirilampo is the only cichlid known to occur upstream at this point. None of the aforementioned species, except for the new species, was captured upstream of this waterfall.

Etymology. The specific epithet “pirilampo” means firefly in the popular Portuguese naming in the region the new species occurs. It is a bioluminescent Coleoptera very common in this region. These insects emit an intense green light, which alludes to the color pattern in life displayed by the new species. A noun in apposition.

Conservation status. The new species is currently known from two tributaries of the upper rio Correntes basin, which is severely impacted due to intense agricultural activity (see Melo et al., 2022) causing a decrease in riparian plant cover and, as a result, a silting process. Additionally, the region has been impacted by the installation of hydroelectric power plants, and further hydroelectric projects are already planned (see ANA, 2019). The Extent of Occurrence (EOO) of Aequidens pirilampo was estimated at about 200 km2. Therefore, considering the relatively restricted geographic distribution of the new species and the severe anthropic impacts in the region, and according to the International Union for Conservation of Nature (IUCN) categories and criteria (IUCN, 2022), Aequidens pirilampo is herein suggested to be classified as Near Threatened (NT), approximating the Endangered (EN) category by criterion B1b(iii).

Common name. Aequidenspirilampo is popularly known as ‘cará’.

Molecular analysis. The final matrix of Aequidens species comprises 25 terminal taxa (including 23 species of Aequidens) and 402 characters, being 303 conserved and 99 variable sites (of which 96 were parsimony-informative), with 22.8% adenine, 28.7% cytosine, 31.7% thymine and 16.8% guanine (% in relation to the entire alignment).

The estimated index of substitution saturation (Iss) performed in DAMBE 5.2.31 (Xia, Xie, 2001) showed that the data was not saturated (i.e., Iss.c value greater than Iss). Genetic distances (Kimura, 1980) of the COI gene between Aequidens pirilampo and other Aequidens and Cichlasoma species are presented in Tab. 2. All species delimitation analyses corroborate the separation of A. pirilampo from the other Aequidens and Cichlasoma species included in this study (Fig. 9).

The ML tree inferred through MEGA-X recovered the two groups consistent with the genetic distance analysis and the previous morphological identification of species and exhibited strong node support for each species, i.e., 100% for the A. pirilampo clade and 99% for A. plagiozonatus clade (Fig. 9) (see Figs. S2, S3, S4, S5) for details in the delimitation analyses). The ABGD analysis resulted in eight partitions that ranged from 96 (P = 0.036) to 169 (P = 0.00113) lineages, with three partitions (P = 0.0031–0.00821) resulting in 15 lineages, and the ASAP analysis resulted in 10 partitions that ranged from 7 (score = 5.5) to 21 (10.5) lineages, with one partition with 16 lineages (lowest score = 3.0) (Fig. 9). All species delimitation analyses support the presence of two species of Aequidens occurring in the rio Paraguai basin, the first, A. pirilampo, described herein, and the second, A. plagiozonatus occurring in both the rio Paraguai and Araguaia basins, however, PTP analysis recovered several lineages within A. plagiozonatus.

Material comparative examined.Brazil, rio Amazonas basin: Aequidens gerciliae: NUP 7200, 3, 52.7–59.1 mm SL. NUP 7483, 1, 46.7 mm SL. NUP 18442, 1, 82.4 mm SL.Aequidens pallidus: NUP 17281, 83.0 mm SL. Aequidens tetramerus: INPA 972, 84.7 mm SL. NUP 17983, 6, 34.9–140.9 mm SL. NUP 19382, 1, 76.9 mm SL. Rio Araguaia basin: Aequidens plagiozonatus: LBP 30741, 1, 39.2 mm SL. Rio Paraguai basin: Aequidens plagiozonatus: LBP 1912, 5, 43.9–61.8 mm SL. LBP, 1925, 3, 40.7–54.0 mm SL. LBP 10767, 3, 41.1–56.7 mm SL. NUP 194, 3, 77.0–88.8 mm SL. NUP 13372, 3 (2 c&s), 57.1–76.6 mm SL. NUP 21604, 3, 37.4–52.0 mm SL. NUP 21621, 3, 25.4–48.2 mm SL. NUP 21368, 1, 81.4 mm SL. Rio Tapajós basin: Aequidens rondoni: MNRJ 1616, photograph and x-ray of the holotype of Acaropsis rondoni. Guyana, rio Potaro: Aequidens potaroensis: FMNH 53892, photograph of the holotype, 140 mm SL. Venezuela, rio Orinoco basin: Aequidens chimantanus: FMNH 45702, photograph of the holotype.

The generic allocation of Aequidens pirilampo in Aequidens as opposed to Cichlasoma is supported by the scaleless dorsal and anal fins (scaled in Cichlasoma), a longer caudal peduncle with 2–3 vertebrae (2 or 0 in Cichlasoma), and 4–5 vertical bars from the midlateral spot to the caudal peduncle in Aequidens (6 in Cichlasoma)(Kullander, 1986). Our results from the molecular analysis support previous hypotheses in which Cichlasoma is nested together with Aequidens (Musilová et al., 2008; Ilves et al., 2018).

Herein, morphological and molecular analyses support the species delimitation of Aequidens pirilampo. The only other valid species of Aequidens described from the rio Paraguai basin is A. plagiozonatus (Fig. 8), which shares with A. pirilampo, the obliquely oriented flank bars, but in the new species this pattern is less conspicuous. Unlike A. plagiozonatus, the caudal fin of some individuals of A. pirilampo, has a dotted or striped pattern, more conspicuous in fixed and smaller to medium specimens. The only other species with anteriorly inclined flank bars is A. viridis (see Kullander, 1984). However, in A. viridis only bars 6 and 7 are inclined, while in A. pirilampo, bars 3 through 7 are inclined. Furthermore, the head profile in lateral view is very triangular in A. pirilampo, instead as rounded as in A. viridis and A. plagiozonatus.

Characters previously used to group species of Aequidens are not well-defined. For example, buccal stripes and counts in vertebrae were used to classify species in unnamed artificial groupings (Kullander, 1995). Here, A. pirilampo has 13+13 vertebrae count, with only a distinction in the fourth and fifth abdominal vertebrae presenting short, paired hypapophyses, but co-ossified (see character 77 in Kullander, 1998). However, as Kullander (1995) identified the usual 13+13 vertebrae count in A. epae, A. gerciliae, and A. michaeli (the latter occasionally with 25, 26 or 27), being traits not exclusive. The new species is more closely related to A. plagiozonatus in the ML approach, and does not fit in any previously defined artificial species group of Aequidens.

Říčan et al. (2005) investigated the evolution and development of the melanophoric coloration in Neotropical cichlids and observed that the homology between individual marks is better understood when ontogenetic series are available for interspecific comparisons. They concluded that, in Cichlasoma, flank bar 4 splits during development, giving place to bar 4a and bar 4p (Říčan et al., 2005, figs. 33–36). By contrast, in species of Aequidens the flank bar 4 usually does not split during ontogeny (Říčan et al., 2005, figs. 26, 28), albeit some specimens of Aequidens may present a split flank bar 4 (see Kullander, 1995, fig. 17, for an example of Aequidens michaeli). Furthermore, in some species of Aquidens, such as A. diadema, A. mauseanus, A. metae, A. pallidus, A. superomaculatum, the flank bars 2 and 3 appear to be fused, similarly to the pattern in Satanoperca setepele Ota, Deprá, Kullander, Graça & Pavanelli, 2022. Therefore, the increased number of bars in Cichlasoma represents a division of the same bar by melanophore migration through ontogeny.

The ventral portion of bar 9 forms the suborbital stripe and is present in the young of all Aequidens species. This pattern is retained in adults only in A. paloemeuensis and A. potaroensis according to original descriptions and Kullander (2012), but lost in the other species. These two species were found to belong to the genus Krobia (K. potaroensis, K. paloemeuensis) by Musilová et al. (2009). However, according to morphological characters, both species pertain to Aequidens (see discussion section in Kullander, 2012). Due to a lack of larval material, we were unable to identify the development of the bars via the union of melanophore patches as described by Ota et al. (2021) and Deprá et al. (2022). A cheek spot is located between the posteroventral margin of the eye and the vertical arm of the preopercle, and is more evident in adults than in the young.

Integrative analyses have been performed as a useful tool to support species delimitations and to confirm species assignments (Padial et al., 2010; Gomes et al., 2015; Dellicour, Flot, 2018; Garavello et al., 2021), as well as to access phylogenetic relationships among the different species.

Herein, the validity of Aequidens pirilampo is investigated through phenotypical exam and several analyses using molecular data. The ML tree based on mt-co1 sequences recovers A. pirilampo and A. plagiozonatus as sister species. The analysis of barcoding indicates interspecific genetic divergence more than 2%, which is considered a suitable threshold for species delimitation (Carvalho et al., 2011; Pereira et al., 2011a,b, 2013) including for some cichlid species (Souza et al., 2018). Depending on the group studied, the interspecific genetic divergence can be found lower than 2% and this cutoff value is only a start point based on the distribution of K2P values of species barcoded available in BOLD system showing that evolutionary history should be considered (http://www.boldsystems.org, and see discussion in Pereira et al., 2013). But an operational taxonomic unit (OTU) has most likely been retrieved when more than 50% of the species delimitation techniques agree (Ramirez et al., 2023).

The Central Brazil, constituted by the rio Amazon-Paraguai-Araguaia basins, is known to have fish species interchange, as evidenced by some events as headwater-capture (i.e., adjacent basins capturing ichthyofauna from each other) (Aquino, Colli, 2017; Dagosta, de Pinna, 2017; Machado et al., 2018; specifically for the studied area Melo et al., 2022). Aequidensplagiozonatus is also present in tributaries of the rio Taquari basin, which is biogeographically hybrid, with species shared with rio Paraguai and Amazon basins (Dagosta, de Pinna, 2021). Headwater-capture occurrences were also noted by Melo et al. (2022) as a probable cause of this distribution pattern of Cyphocharax species from the same region. Because of their correlated geological history, there may be faunal interaction between the Taquari basin and the Araguaia headwaters (Dagosta, de Pinna, 2021). Such events may also enhance Aequidens dispersion in the Paraguai and rio Araguaia basins, highlighting the role of hydrogeological theory in this region (Hubert, Henno, 2006; Machado et al., 2018). The new species and the new record of A. plagiozonatus in the rio Araguaia basin comes from an endemic zone for Characidae, Callichthyidae, Rivulidae, Aspredinidae, Trichomycteridae, and Curimatidae species (Lima, Moreira, 2003; Dagosta et al., 2020; Dagosta, de Pinna, 2021; Melo et al., 2022).

Regarding other Aequidens species, morphological and ABGD analyses revealed a monophyletic lineage of A. pallidus from the rio Jatapu and rio Negro, however GMYC, ABGD and ASAP analyses separated these populations in two species. Using the same delimitation methods, Carvalho et al. (2019) identified A. pallidus as a non-monophyletic group. In the case of A. tetramerus, morphological and PTP analyses suggest that the specimens studied are monophyletic, but the opposite occurs in GMYC, ABGD, and ASAP analyses. The unique specimen of A. tetramerus from French Guiana (rio Maroni basin) is distinct from the others from Suriname (rio Paloemeu basin), and they form a polytomy with the A. epae clade from Brazil (rio Tapajós basin). This indicates the need for further research into the Aequidens specimens found in the vast and interconnected AOG region.

Among the techniques employed herein, distinct scenarios of lineages have been recovered. For instance, for Cichlasoma paranaense Kullander, 1983 PTP and ABGD returned one species, while GMYC and ASAP separated into two species. PTP analysis oversplitted some Aequidens species, such as A. plagiozonatus, A. epae, A. pallidus and A. diadema. Dellicour, Flot (2018) brings out that the size of population data sets influences the delimitation tests, the larger the data set, the lower the chance of correctly delimiting the studied group; on the other hand, when more than 50% of the tests are in agreement, the delimitation is probably correct (Ramirez et al., 2023). About distinct delimitation results, C. paranaense presented one specimen with a few differences in the sequenced nucleotides, from the upper portion of the rio Paraná basin (rio Grande sub-basin), however, the genetic distance reveals maximum intraspecific values lower than minimum interspecific values, showing the species present the barcoding gap. Same is possible to say about the Aequidens species. The perfect congruence among all delimitation methods is not always observed using single-locus approaches (Miralles, Vences, 2013; Dellicour, Flot, 2018). Therefore, the multilocus species approach can be used to support the delimitation study and may be able to handle high rates of speciation and huge population numbers (Dellicour, Flot, 2018). Alternatively, depending on the group studied, phylogenomics might be applied to resolve recent, rapid radiation, revealing even the hybridization cases (Giarla, Esselstyn, 2015; Irisarri et al., 2018; Huang et al., 2019).

The new species is found in Brazil’s Cerrado biome, a biodiversity hotspot that is home to a wide variety of threatened fish species (ICMBio, 2018). Since human activities have had an impact on this area, conservation efforts should focus on strengthening sampling and investigation. This, together with the increased new species descriptions for this region, reinforce the importance of the Cerrado and rio Paraguai basin in the conservation of endemic species.