﻿Molecular species delimitation and description of a new species of Phenacogaster (Teleostei, Characidae) from the southern Amazon basin

﻿Abstract Phenacogaster is the most species-rich genus of the subfamily Characinae with 23 valid species broadly distributed in riverine systems of South America. Despite the taxonomic diversity of the genus, little has been advanced about its molecular diversity. A recent molecular phylogeny indicated the presence of undescribed species within Phenacogaster that is formally described here. We sampled 73 specimens of Phenacogaster and sequenced the mitochondrial cytochrome c oxidase subunit I (COI) gene in order to undertake species delimitation analyses and evaluate their intra- and interspecific genetic diversity. The results show the presence of 14 species, 13 of which are valid and one undescribed. The new species is known from the tributaries of the Xingu basin, the Rio das Mortes of the Araguaia basin, and the Rio Teles Pires of the Tapajós basin. It is distinguished by the incomplete lateral line, position of the humeral blotch near the pseudotympanum, and shape of the caudal-peduncle blotch. Meristic data and genetic differentiation relative to other Phenacogaster species represent strong evidence for the recognition of the new species and highlight the occurrence of an additional lineage of P.franciscoensis.

Relative to other Characinae genera, Phenacogaster possesses two longitudinal series of elongate and imbricated scales producing a zigzag pattern in a flat preventral region, as well as the outer premaxillary tooth row divided into a medial and a lateral section separated by a diastema (Eigenmann 1917;Malabarba and Lucena 1995;Mattox and Toledo-Piza 2012). Lucena and Malabarba (2010) presented the most comprehensive taxonomic revision of the genus with descriptions of nine species of Phenacogaster, nearly doubling the species diversity, and an identification key for the species, with the exception of the so-called Phenacogaster pectinata complex with P. pectinata (Cope, 1870), P. microstictus Eigenmann, 1909, P. beni Eigenmann, 1911and P. suborbitalis (Ahl, 1936. Recently, three more species from the Brazilian Shield have been described: P. naevata Antonetti, Lucena & Lucena, 2018; P. eurytaenia Antonetti, Lucena & Lucena, 2018 from the Tocantins basin (Antonetti et al. 2018); and P. julliae Lucena & Lucena, 2019 from the Rio São Francisco (Lucena and Lucena 2019).
No study has been conducted to assess the interspecific genetic diversity of Phenacogaster, although species delimitation methods have been used for such purposes in other Characidae (Rossini et al. 2016;García-Melo et al. 2019;Brito et al. 2021;Malabarba et al. 2021;Mattox et al. 2023). A recent molecular phylogeny of Characinae revealed the presence of the two clades in Phenacogaster, the P. pectinata clade and the P. franciscoensis clade, as well as an undescribed species of Phenacogaster from the Xingu basin (Souza et al. 2022). To further investigate this question, we used mitochondrial data and species delimitation techniques to estimate intra-and interspecific genetic diversity within the genus. The results confirmed the presence of a new species in the upper Rio Xingu of the Amazonian Brazilian Shield, which is formally described in this paper.

Taxon sampling
The molecular analysis encompassed 74 taxa (Suppl. material 4), including 73 specimens of Phenacogaster and Tetragonopterus carvalhoi Melo, Benine, Mariguela & Oliveira, 2011 as an outgroup. Seventy sequences were generated, and four were retrieved from BOLD/GenBank: one Tetragonopterus carvalhoi, two P. wayana, and one P. calverti (Suppl. material 4). We used Phenacogaster specimens collected or received from ichthyological collections, which were identified morphologically using identification key (Lucena and Malabarba 2010). All fishes were collected in accordance with Brazilian law through SISBIO/MMA permit no. 3,245, and collection, maintenance, and analyses procedures were conducted in accordance with international guidelines for animal experiments via CEEAA IBB/UNESP protocol no. 304.

DNA amplification and sequencing
DNA was extracted from muscles or gills using the extraction method of Ivanova et al. (2006). The cytochrome c oxidase subunit I (COI) gene was amplified by polymerase chain reaction (PCR) using the FishF1/FishR1 and FishF2/FishR2 primers (Ward et al. 2005) or the L6252-Asn/H7271-COXI primers (Melo et al. 2011). PCR amplifications were performed in a total volume of 12.5 µl that included 1.25 µl of 10X buffer, 0.25 µl of MgCl 2 (50 mM), 0.2 µl dNTPs (2 mM), 0.5 µl of each primer (5 mM), 0.1 µl of PHT Taq DNA polymerase (Phoneutria), 1.0 µl of genomic DNA (200 ng) and 8.7 µl ddH 2 O. The PCR conditions included an initial denaturation (5 min at 94 °C), 30 cycles of chain denaturation (50 s at 94 °C), primer hybridization (45 s at 50-54 °C), and nucleotide extension (1 min at 68°C), and a final extension (10 min at 68 °C). All PCR products were checked on 1% agarose gels and then purified using ExoSap-IT (USB Corporation) according to the manufacturer's instructions. The purified PCR products were subjected to sequencing procedures with the BigDye Terminator v. 3.1 Cycle Sequencing Ready Reaction Kit (Applied Biosystems) and purified by ethanol precipitation. PCR products were loaded onto an ABI 3130 DNA Analyzer automatic sequencer (Applied Biosystems).

Molecular data analysis
Forward and reverse sequences were assembled using Geneious v. 7.1.9 (Kearse et al. 2012) and contigs aligned with MUSCLE (Edgar 2004) using the default parameters. Substitution saturation was determined using Xia et al. (2003)'s approach in DAMBE v5.3.38 (Xia 2013). Nucleotide variation, substitution patterns, and the best-fit model of nucleotide evolution were estimated in MEGA v. 10 (Kumar et al. 2018).
The maximum likelihood (ML) analysis was conducted using RAxML HPC-PTHREADS-SSE3 (Stamatakis 2014) with five random parsimony trees and the GTRGAMMA model on the Zungaro server at LBP/UNESP. The neighbor-joining (NJ) tree was estimated with the K2P+G model (Kimura 1980) and 1,000 bootstrap replicates in MEGA v10 (Kumar et al. 2018). Two species delimitation methods were performed: the Assemble Species by Automatic Partitioning (ASAP) analysis (Puillandre et al. 2021) via the webserver (https://bioinfo.mnhn.fr/abi/public/asap/asapweb. html) with Kimura (K80; 2.0); and the Poisson Tree Process (PTP; Zhang et al. 2013) using the ML tree as input, 100,000 generations, and other parameters at default in the PTP webserver (http://species.h-its.org/ptp/). MEGA v. 10 computed K2P+G distances across groups based on their morphological identification. The ASAP, PTP, and genetic distance analyses were conducted without the outgroup.

Morphological analysis
Morphometric and meristic data were collected on the left side of every specimen whenever possible, following Malabarba and Lucena (1995) and Lucena and Gama (2007). Point-to-point measurements were taken with a precision of 0.1 mm using a digital caliper. Counts of vertebrae, supraneurals, gill rakers, and teeth were obtained from cleared and stained (c&s) specimens prepared in accordance with Taylor and Van Dyke's (1985) methodology. Vertebral counts include the four centra of the Weberian apparatus as separate elements and the compound ural centrum as a single vertebra. Except for head subunits, which are reported as a percentage of head length (HL), other measurements are expressed as a percentage of standard length (SL). In the description, the frequency of each count is mentioned in parenthesis, and the holotype count is indicated with an asterisk. Institutional acronyms follow Sabaj (2020). Specimens from the Xingu basin were determined as types and specimens from Araguaia and Tapajós are listed as non-types. Examined material is organized by acronym and collection number, number of specimens, range of SL, locality, collection date, and collectors. Comparative material is classified according to the alphabetical order of species, and, within a species, it follows the same order as examined material.

Molecular species delimitation
Partial COI gene sequences were obtained from 68 specimens representing 13 of the 23 valid species of Phenacogaster (56.2%), and for five specimens that represent the species described in this study. The matrix consisted of 678 bp (153 variable sites) and had a nucleotide composition of 24.6% adenine, 27.5% cytosine, 18% guanine, and 30% thymine. In both asymmetrical (Iss.cAsym) and symmetrical (Iss.cSym) topologies, neither transitions nor transversions were found to be saturated by DAMBE. Both ML and NJ trees recovered similar topologies and supported the recognition of P. lucenae as a new species (Fig. 1, Suppl. material 1). The best partition provided by ASAP identified 14 species (1.00 asap-score) (Fig. 1, Suppl. material 2) and supported P. lucenae as new species. The PTP analysis defined 15 species and recognized the new species as a distinct lineage (Fig. 1, Suppl. material 3). Both methods recovered the same species limits, except for P. franciscoensis which was split in two by the PTP method. The overall mean K2P genetic distance was 0.077 ± 0.007. Interspecific genetic distances were between 0.026 ± 0.007 and 0.143 ± 0.020, and intraspecific genetic distances ranged between 0.000 ± 0.000 and 0.010 ± 0.003 (Table 1). Table 1. Pairwise K2P genetic distances and intraspecific genetic variation of Phenacogaster species included in this study. Numbers below the diagonal represent the interspecific distance, while the numbers above the diagonal represent the relative standard deviation.  Diagnosis. Phenacogaster lucenae is distinguished from all congeners except P. tegata (Eigenmann, 1911), P. carteri (Norman, 1934), P. napoatilis Malabarba, 2010, andP. capitulata Lucena &Malabarba, 2010 by having an incomplete lateral line (vs. complete lateral line). It differs from P. tegata by the presence of a round or slightly longitudinal oval humeral blotch near the pseudotympanum and distant from the vertical through dorsal-fin origin (vs. humeral blotch longitudinally elongated distant from pseudotympanum, closer to vertical through dorsal-fin origin). The new species differs from P. carteri by having a humeral blotch in males and females (vs. absence of humeral blotch in both sexes) and from P. napoatilis and P. capitulata by having a humeral blotch in both sexes (vs. absence of humeral blotch in males). In addition to the incomplete lateral line (vs. complete), P. lucenae differs from P. retropinna Lucena & Malabarba, 2010 by the anal-fin origin at vertical through base of first or second dorsal-fin branched ray (vs. anal-fin origin located posteriorly to that point), and from P. ojitata Lucena & Malabarba, 2010 by the round caudal peduncle blotch slightly reaching over the middle caudal-fin rays (vs. a diamond-shaped caudal peduncle blotch and further extending over the middle caudal-fin rays).  Table 2. Body compressed. Dorsal profile convex from anterior tip of upper jaw to origin of dorsal fin with a slight concavity in occipital region; slightly straight from dorsal-fin base to origin of adipose fin and slightly concave from that point to base of dorsal procurrent caudal-fin rays. Ventral profile of body convex from tip of lower jaw to anal-fin origin, straight along anal-fin base, straight to slightly concave from that point to ventral procurrent caudal-fin rays. Preventral area flattened with two longitudinal series of elongate scales overlapping; scales different in shape from remaining body scales and forming zigzag pattern in ventral view. Pseudotympanum triangular extending from region of rib of fifth vertebra to anterior border of rib of sixth vertebra.

Color in alcohol.
Overall ground coloration pale yellow ( Fig. 2A). Dorsolateral region of body with melanophores along margins of scales. Ventrolateral region less pigmented. Thin lines of melanophores accompanying myosepta along flanks, more evident in the hypaxial musculature. Females and males with rounded or slightly longitudinally oval humeral blotch immediately posterior to pseudotympanum, covering roughly three to five scale rows vertically and two to five scales longitudinally. Caudal peduncle with circular patch of melanophores covering whole caudal peduncle depth and reaching base of caudal-fin middle rays. Thin line of melanophores extending along horizontal septum between humeral and caudal peduncle blotches. Anal, pelvic, pectoral, and dorsal fins scattered by small melanophores. Adipose fin hyaline (Fig. 2).
Color in life. Overall ground coloration yellowish to golden on slightly translucent background (Fig. 2B, C). Dorsolateral body region with melanophores along margins of scales. Ventrolateral area less pigmented. Humeral blotch rounded or oval with anterior portion black and posterior edge iridescent yellow to orange. Round black blotch on middle portion of caudal peduncle extending vertically over entire caudal peduncle depth and extending posteriorly to proximal portion of caudal-fin middle rays. Some specimens with bright golden or white patches on posterior portion of caudal peduncle blotch, covering base of caudal-fin rays in upper and lower lobes. Thin line of melanophores between humeral and caudal peduncle blotches. Abdominal cavity, opercular series and portion of infraorbitals covered with guanine. All fins orange to yellowish coloration, with anterior halves of caudal-fin lobes more intensely colored. Posterior tip of caudal and dorsal fins scattered by small dark chromatophores (Fig. 2B, C).
Sexual dimorphism. Our samples consist of two adult males (MZUSP 97621, 30.4-34.4 mm SL) with hooks on pelvic-and anal-fin rays (Fig. 4). Four to six lateralmost branched pelvic-fin rays with five to nine curved hooks on medial edge of rays, one hook per segment towards the tips and more hooks per segment toward the base of the rays. Hooks more developed and frequent on medial regions of branched rays (Fig. 4A). Anal-fin rays with four to nine curved hooks on the posterior edges of the last unbranched and the first to eleventh branched fin rays. Fin hooks more developed and abundant on anterior branched rays. In most cases, one pair of small hooks per segment, but occasionally two pairs per segment. Hooks in some cases incipient in the form of bumps. A few rays with additional single hook on the anterior edge of distal portion (Fig. 4B).
Etymology. Phenacogaster lucenae is named in honor of Dr. Zilda Margarete Seixas de Lucena, an eminent ichthyologist who has significantly contributed to our knowledge of Phenacogaster taxonomy. A noun in genitive case.

Conservation status.
Phenacogaster lucenae is found in the upper Xingu, Araguaia, and Tapajós basins, where specimens were collected during focused expeditions. Although deforestation and hydroelectric plants have affected the region, 18 specimens of P. lucenae have been collected in recent years (2021)(2022), demonstrating a likely high resilience to anthropogenic impacts. Therefore, we suggest the categorization Least Concern (LC) according to the International Union for Conservation of Nature criteria (IUCN 2014, Standards and Petitions Subcommittee).

Discussion
This is the first molecular delimitation using barcode sequences of the genus Phenacogaster spanning more than half of the known species diversity and supplements the phylogenetic study of the Characinae recently published including 16 species of Phenacogaster (Souza et al. 2022). Based on the application of the species delimitation methods, results identified 14 or 15 species (ASAP and PTP) and both supported P. lucenae as a new species (Fig. 1). The minor difference of ASAP and PTP results may be attributable to the range of algorithms and implementations, population size, species diversity, and speciation rates (Ahrens et al. 2016;Puillandre et al. 2021). The species delimitation methods are useful tools that, when combined with other types of information such as morphological data, may constitute solid evidence for species delimitation Mateussi et al. 2020;Lozano et al. 2022).
The Phenacogaster pectinata complex (P. pectinata, P. microstictus, P. suborbitalis and P. beni) was proposed for widely distributed species characterized by humeral blotch present only in females, humeral blotch absent or restricted to a few chromatophores in males, complete lateral line, and 32-42 branched anal-fin rays (Lucena and Malabarba 2010). Phylogeny based on genomic evidence supports the group (P. pectinata clade) and adds P. capitulata, P. megalostictus, P. prolata, P. suborbitalis, and P. tegata (Souza et al. 2022). Both topologies of our study concur with the molecular phylogeny (Fig. 1, Suppl. material 1) and adds P. microstictus from the Rupununi River as another member of the clade closer to P. prolata from the Negro basin (Fig. 1).
Phenacogaster lucenae belongs to the P. franciscoensis clade (Souza et al. 2022). In fact, these authors sequenced ultraconserved elements for P. lucenae (identified there as Phenacogaster sp. Xingu) and discovered that it is the sister species of P. retropinna (Tapajós and Xingu) (Souza et al. 2022). Lucena and Malabarba (2010) described P. retropinna from the Amazonian rivers Negro, Madeira, Xingu, and Araguaia. Here, both molecular and morphological evidence support the distinction between P. lucenae and P. retropinna. The mitochondrial data analysis revealed a reasonably high genetic divergence (0.038±0.008) between these species (Table 1, Fig. 1). Lucena and Malabarba (2010) described the endemic Phenacogaster ojitata from the Rio Curuá, a left tributary of the Xingu. Unfortunately, there are currently no tissues of P. ojitata for molecular analyses. Morphologically, P. lucenae can be recognized from P. ojitata by the round caudal peduncle blotch slightly reaching over the middle caudal-fin rays and a larger orbital diameter (34-42.9% of HL; see diagnosis). In addition, P. lucenae can be distinguished from other Phenacogaster with incomplete lateral line by the presence of humeral blotch (vs. absence of humeral blotch in P. carteri), presence of humeral blotch in males and females (vs. absence of humeral blotch in males of P. napoatilis and P. capitulata); humeral blotch near pseudotympanum and distant from vertical line through dorsal-fin origin in males and females (vs. humeral blotch distant from pseudotympanum and near dorsal-fin origin in males and females of P. tegata).
Reduction or lack of pores in the laterosensory system is a classic reductive trait in fishes (Myers 1958) and most likely results from the loss of terminal developmental stages as consequence of the body size reduction (Marinho et al. 2021). As stated previously, the incomplete lateral line is present in four of the currently 23 valid species of Phenacogaster (Lucena and Malabarba 2010) in addition to P. lucenae described here. Although we did not have access to tissue samples from all these species, our results indicate that reduction of the lateral line independently evolved three times in the phylogeny (Souza et al. 2022). We detected incomplete lateral lines in both juveniles and adults of P. lucenae, with only six of 32 specimens exhibiting scale interruptions along the lateral line (i.e., incomplete pored series with two or three pored scales towards the end of the scale series, and a long gap of non-pored scales between them). The evolutionary significance for this modification still needs additional research as well as the investigation of sympatric occurrence of species with completely and incompletely developed laterosensory system.
Additional research on Phenacogaster can concentrate on taxa that have not been sampled and additional gene sampling. The presence of two distinct lineages of P. franciscoensis, one in the Rio São Francisco and another in the Rio Parnaíba is under investigation. Additional undescribed species are also expected for the genus as we increase taxon sampling in research projects. Finally, further research is required to understand the historical biogeographic processes that contributed to the disjunct distribution of the Phenacogaster species across the Brazilian Shield.