A Poorly Known Catfish Clade in an Endangered Neotropical Biodiversity Hotspot: Relationships and Distribution Patterns of the Cambeva variegata Group (Siluriformes: Trichomycteridae)

Abstract

The Cambeva variegata group (CVG) is endemic to a region situated in the intersection of two endangered biodiversity hotspots, Cerrado and Atlantic Forest, and drained by two important South American river basins, the upper Rio Paraná and upper Rio São Francisco basins. Presently, CVG comprises two nominal species, besides some still undescribed. We first performed a molecular phylogenetic analysis (total of 3368 bp) for five species of the CVG and 30 outgroups, which supported the monophyly of the CVG and its inclusion in Cambeva. Most morphological character states distinguishing the CVG from congeners are also present in Scleronema, possibly consisting of plesiomorphic features. We also performed the first time-calibrated phylogeny of the group, which supported possible relationships between present geographical distribution patterns and palaeogeographical events. The estimated time of origin of CVG in the Middle Miocene is nearly contemporaneous to a past hydrographical configuration when part of the upper Rio Paraná basin was connected to the Rio São Francisco basin. The first CVG lineage split occurring in the Miocene end corresponds to a major break in that palaeo basin. Species diversification between the Pliocene and early Pleistocene is compatible with final drainage rearrangement. This study highlights the urgent need for more detailed studies on the diversity and phylogenetic relationships of still poorly known organisms in this highly diverse and threatened region.

1. Introduction

Fast-flowing rivers and streams draining the mountain ranges of southeastern Brazil in the region comprising the upper Rio Paraná basin and the upper Rio São Francisco basin (hereafter Upper Paraná–São Francisco Region, UPSFR), such as Serra da Canastra, Serra do Espinhaço, and Serra da Mantiqueira, among others, shelter a huge diversity of trichomycterine mountain catfishes, mostly revealed in recent years (e.g., [1] and included references). The great majority of trichomycterines endemic to this region belong to the genus Trichomycterus Valenciennes, 1832, whereas species of Cambeva Katz, Barbosa, Mattos, and Costa, 2018, are rare. Cambeva presently includes about 50 species occurring between the tropical area drained by the Rio São Francisco basin, about 16°10′ S, and the subtropical area drained by the rivers connected to the Lagoa dos Patos system, about 29°30′ S [2]. However, only two species of Cambeva occur and are endemic to UPSFR, Cambeva concolor Costa, 1992, and Cambeva variegata Costa, 1992, both endemic to the upper Rio São Francisco basin [3]. These species are easily distinguished from all other species of Cambeva by the presence of a prominent skin crest on the dorsal margin of the caudal peduncle [3], a condition similar to that occurring in Scleronema Eigenmann, 1917 [4,5,6], and to the adipose fin of the Copionodontinae [7]. Our field studies have shown that this group of Cambeva, hereafter the Cambeva variegata group (CVG), also includes undescribed species, some of which are endemic to the Rio Grande drainage, upper Rio Paraná basin, consisting of the first records of CVG for UPSFR.

Whereas the great species diversity of Trichomycterus from UPSFR has been reported in frequent publications (e.g., [2,8] and included references), little has been published about species of the CVG from the same region. Mentions of this group are restricted to a single taxonomic paper published 32 years ago [3] and a few phylogenetic studies inferring the positioning of C. variegata among other congeners (e.g., [2,9,10]). The scarcity of studies on the CVG is probably due to two factors. Firstly, most species of this group are extremely similar when compared only by external morphology, making the identification and recognition of new species inaccessible to biologists not trained in osteological examination or molecular biology. Secondly, there is great difficulty in field collecting, since species in this group have shown to be restricted to small areas and are often rare in their environments.

UPSFR is situated in the intersection of two phytogeographical provinces considered among the main biodiversity hotspots of the world, the savannah-like Cerrado and the Atlantic Forest [11]. On the other hand, rivers and streams of this area are highly impacted by anthropogenic factors, such as large dams for the generation of electricity, destruction of the hydrographic structure for mineral extraction, water pollution by domestic and industrial sewage and by pesticides used in agriculture, and the introduction of exotic species and deforestation of marginal areas (e.g., [1]). The growing process of environmental degradation requires urgency in studies focusing on the diversity of this rare fish group. The objectives of the present study are to present the first phylogenetic analysis involving a large species sample of the CVG, testing the monophyly of the group; to provide a comparative osteological analysis allowing morphological diagnoses of clades supported in molecular analyses; and to present the first time-calibrated phylogeny of the group in order to find evidence of possible relationships between the present geographical distribution patterns and palaeogeographical events.

2. Materials and Methods

2.1. Specimens

Fish collections were performed during daylight with dip nets. Collecting permits were provided by ICMBio (Instituto Chico Mendes de Conservação da Biodiversidade; permit numbers: 38553-13 and 64415-5). Methods for fish collection, euthanasia using a buffered solution of tricaine methane sulphonate (MS-222) at a concentration of 250 mg/L (e.g., AVMA Guidelines for the Euthanasia of Animals) [12], and fixation were approved by CEUA-CCS-UFRJ (Ethics Committee for Animal Use of Federal University of Rio de Janeiro; permit numbers: 065/18 and 084/23). For fixation, specimens were kept in formalin for two weeks, or fixed in absolute ethanol in cases of specimens used in the molecular analysis. Formalin-fixed specimens were preserved in 70% ethanol. Among these specimens, between three and five were prepared for osteological examination, using clearing and staining techniques described by Taylor and Van Dyke [13], and later preserved in glycerine. In the list of specimens below, C&S means specimens cleared and stained for osteological examination, and DNA means specimens directly fixed in absolute ethanol for molecular analysis. Specimens were deposited in the ichthyological collections of Instituto de Biologia, Universidade Federal do Rio de Janeiro (UFRJ), and Centro de Ciências Agrárias e Ambientais, Universidade Federal do Maranhão (CICCAA).

Specimens examined: Cambeva variegata (UFRJ 8355, 7 ex.; UFRJ 8314, 1 ex. (DNA); UFRJ 8318, 1 ex. (DNA); UFRJ 8319, 1 ex. (DNA); UFRJ 9346, 2 ex. (C&S): topotypes: 20°15′09″ S 46°24′23″ W; UFRJ 585, 2 paratypes, (C&S); UFRJ 12857, 3 ex.; UFRJ 12833, 1 ex. (DNA): 20°15′50″ S 46°20′56″ W; UFRJ 14072, 9 ex.; UFRJ 12928, 3 ex. (DNA): 20°15′50″ S 46°20′56″ W); Cambeva sp. 1 (UFRJ 14061, 12 ex.; UFRJ 14062, 5 ex., C&S; UFRJ 14063, 6 ex., DNA: 22°28′02″ S 45°21′38″ W); Cambeva sp. 2 (UFRJ 14059, 5 ex.; UFRJ 14060, 2 ex., C&S: 20°34′47″ S 46°20′56″ W); Cambeva sp. 3 (UFRJ 14050, 1 ex.; UFRJ 14051, 11 ex; UFRJ 14052, 4 ex. (C&S); UFRJ 14042, 2 ex. (DNA); CICCAA 08193, 4 ex.: 20°36′27″ S 46°26′09″ W); Cambeva sp. 4 (UFRJ 13637, 1 ex.; UFRJ 14054, 1 ex.; UFRJ 13638, 1 ex. (DNA); UFRJ 14055, 9 ex.; UFRJ 14056, 4 ex. (DNA); UFRJ 14057, 5 ex. (C&S): 20°35′19″ S 46°13′41″ W); Cambeva sp. 5: (UFRJ 13653, 1 ex.; UFRJ 12967, 6 ex.; UFRJ 12968, 22 ex.; UFRJ 12983, 9 ex.; UFRJ 13654, 4 ex. (C&S); CICCAA 07950, 10 ex.: 19°58′29″ S 43°51′39″ W; UFRJ 12965, 7 ex.; UFRJ 12966, 15 ex.; UFRJ 12982, 3 ex.: 19°58′26″ S 43°47′47″ W). In addition, we accessed data taken from the type material of C. concolor and C. variegata, deposited in Museu de Zoologia, Universidade de São Paulo, at the time of the original description by one of us (WJEMC). Comparative material appears in Costa et al. [11] and included references.

2.2. Morphological Data

Methods for taking morphological data and terminology followed the methods of our most recent studies on the systematics of eastern South American trichomycterines (e.g., [1]), which were based on Costa [14] and Kubicek [15] for bone nomenclature, Arratia and Huaquin [16] and Bockmann and Sazima [17] for the terminology of pores of the cephalic latero-sensory system, and Bockmann and Sazima [17] for fin-ray formulae. Morphological comparisons were made in the two nominal species, C. concolor and C. variegata, and in five undescribed species, Cambeva sp. 1–5 (see list of specimens above).

2.3. DNA Extraction, Amplification, and Sequencing

Methods for DNA extraction are those described in our recent studies on trichomycterines [1]. Polymerase chain reaction (PCR) was performed with the following primers for mitochondrial encoded genes: Cytb Siluri F and Cytb Siluri R [18] for cytochrome b (CYTB); FISH-F1 and FISH-R1 [19] for cytochrome c oxidase I (COX1); L11935 and H12857 [20] for NADH: ubiquinone oxidoreductase core subunit 4 (ND4), along with t-RNA-His, Ser, and Leu; and for the nuclear gene recombination activating 2 (RAG2), RAG2 TRICHO F and RAG2 TRICHO R [21]. PCR was performed in 45 μL with the following reagent concentrations: 5× GreenGoTaq Reaction Buffer (Promega), 2.0 mM MgCl₂, 1 μM of each oligo, 0.2 mM of each dNTP, 1 μL of Promega GoTaq Hot Start polymerase, and 50 ng of total genomic DNA. Negative controls were used to check for contaminants. The thermal profile consisted of initial denaturation at 95 °C for 5 min; followed by 35 cycles of denaturation at 94 °C for 1 min, annealing at 45–50 °C for 1 min, and extension at 73 °C for 1–1.5 min; with a final extension at 73 °C for 7 min. PCR products were purified using the Wizard SV Gel and PCR Clean-Up System (Promega). Sequencing reactions were performed in 20 μL reaction volumes containing 4 μL BigDye, 2 μL sequencing buffer 5× (Applied Biosystems), 2 μL of the PCR products (30–40 ng), 2 μL primer, and 10 μL ultrapure water, and the thermal profile was 35 cycles of 30 s at 95 °C, 30 s at 55 °C, and 1.5 min at 72 °C. MEGA 11 [22] was used for analysing sequencing chromatograms and sequence annotation, and for translating DNA sequences into amino acid residues to confirm the absence of premature stop codons or indels. GenBank accessions are provided in Appendix A.

2.4. Phylogenetic Analysis

Terminal ingroup taxa comprised a total of fifteen species, including five species of the CVG and ten species representing other lineages of Cambeva. The species of the CVG analysed were C. variegata and four undescribed species. Cambeva concolor and Cambeva sp. 2, without available material for DNA sequencing, were not included in the phylogenetic analysis, with their possible relationships inferred from an examination of morphological traits (see list of specimens examined above). Outgroups comprised five species of Scleronema, the sister group of Cambeva, and five species of Trichomycterus s.s., the sister group of Cambeva plus Scleronema [2], two trichomycterines representing another subfamilial lineage, three species representing other trichomycterid subfamilies, four species representing other catfish families, and one representative of another Ostariophysi lineage, order Characiformes. The gene datasets were aligned using the Clustal W algorithm [23] implemented in MEGA 11, not finding gaps or stop codons. The 3368 bp complete dataset (COX1 732 bp, ND4 693 pb, tRNA His Ser Leu 162 pb, CYTB 993 bp, RAG2 788 bp) was analysed using PartitionFinder2.1.1 [24] for optimal partitioning and evolutionary models, using the Corrected Akaike Information (AICc) selection criteria. Partitions and respective models appear in Appendix B.

Bayesian Inference (BI) and Maximum Likelihood (ML) analyses were utilised as independent approaches for phylogenetic reconstruction, aiming to mitigate methodological biases. BI was conducted using Beast 1.10.4 [25] with the following parameters: Birth–Death process as the tree prior [26], and two independent Markov chain Monte Carlo (MCMC) runs with 10⁷ generations with a sampling frequency of 1000 each. The convergence of the MCMC chains and the proper burn-in value were determined by evaluating the achievement of the stationary phase and the effective sample size for all the analysis parameters in both runs using Tracer 1.7.2 [27]. LogCombiner v.1.10.4 and the Tree Annotator version 1.10.4 [25] were employed to combine and calculate the consensus tree, apply the burn-in, and annotate the Bayesian posterior probabilities. The ML analysis utilised IQTREE 2.2.2.6 [28]. Node support was assessed by ultrafast bootstrap (UFBoot) [29] and the Shimodaira–Hasegawa-like approximate likelihood ratio test (SH-aLRT), both with 1000 replicates [30].

2.5. Divergence Time Estimation

The divergence time analysis was conducted in Beast 1.10.4 using the same dataset, partitions, evolution models, tree priors, and parameters as described above. Additionally, the analysis incorporated a lognormal uncorrelated relaxed clock model. Calibration points were established as follows: the origin of Siluriformes Order with a normal prior distribution (mean = 140 MA, SD = 7) following Lundberg et al. [31], the origin of the Trichomycteridae with a normal prior distribution (mean = 106 MA, SD = 5.0) following the estimative of Betancur-R et al. [32]; and the origin of the genus Corydoras Lacépède, 1803, with a lognormal prior distribution (mean = 55 MA, SD = 2.5), based on the dating of the oldest known Corydoras fossil species, Corydoras revelatus Cockerell, 1925. MCMC chains were assessed to verify convergence by evaluating the effective sample size of the runs in Tracer 1.7.1. The time-scaled tree was obtained using Tree Annotator version 1.10.4 to generate the consensus tree.

3. Results

3.1. Phylogenetic Relationships and Comparative Morphology

The CVG was supported as monophyletic with maximum values, whereas relationships among the more inclusive lineages of Cambeva were weakly supported (Figure 1). Individual tree loci generated compatible results (Supplementary File S1), with the CVG recovered as monophyletic in both trees, but without any internal resolution in the RAG2 tree. In addition to the presence of an adipose crest on the caudal peduncle (Figure 1A,B), CVG species differ from other congeners by the presence of an interrupted supraorbital canal, with an s4 pore (Figure 2; vs. continuous, s4 absent), a pronounced narrowing at the lateral end of the premaxilla (Figure 3A–E; vs. expansion absent), and a ventral expansion in the preopercle (Figure 4; vs. expansion absent). In addition, only in species of the CVG, the foramen of the parurohyal is rudimentary or absent (Figure 5B–E), but a broad foramen is present in a new species of this clade described below (Figure 5A), possibly as a result of a reversal due to its apical position in the phylogenetic tree (Figure 1). The CVG group comprises two main lineages, Cambeva sp. 1 from tributaries of the Rio Grande drainage at the Serra da Mantiqueira and the Cambeva variegata complex, a group of similar species occurring in a broad geographical region encompassing the upper Rio São Francisco basin and an adjacent area of the Rio Grande basin at the Serra da Canastra (Figure 6). A taxonomical revision of the Cambeva variegata complex is in progress by the authors and new species here cited will be described elsewhere.

Cambeva sp. 1, previously identified as Trichomycteridae sp. by Thereza and Langeani [8], was strongly supported as a member of the CVG here. Externally, it differs from all other congeners by having a long dorsal fin with twelve or thirteen principal rays (vs. nine or ten) (Figure 1A). Seven osteological features distinguish this species from all other congeners: postero-lateral process of the autopalatine posteriorly directed (Figure 3E; vs. postero-laterally directed, Figure 3A–D); posterior extremity of the lateral process of the vomer sinuous (Figure 3G; vs. convex, Figure 3F); six branchiostegal rays (vs. seven or eight); epibranchial 2 elongated, lacking the postero-distal process (Figure 7B; vs. not elongated, with a postero-distal process, Figure 7A); pharyngobranchial 4 relatively large, occupying more than half the surface of the adjacent dentigerous plate (Figure 7B; vs. one-third or less, Figure 7A); dorsal and ventral hypural plates in close proximity, separated by a short posterior interspace (Figure 8B; vs. completely separated, Figure 8A); and pelvic bone relatively broad, with short anterior processes (Figure 8D; vs. narrower, with longer anterior processes, Figure 8C). In addition, the entire anterior portion of the neurocranium is proportionally slenderer (Figure 3E) than in other congeners (Figure 3A–D). This taxon also differs from all other congeners of the CVG by the absence of a pectoral fin filament and the presence of a lateral expansion of the mesethmoid (Figure 3E; vs. absence); the mesethmoid cornua having a relatively short and sharply pointed cornu (Figure 3E; vs. longer, not sharply pointed, Figure 3A–D); and pharyngobranchial 3 with distinctive lateral constriction (Figure 7B; vs. without lateral constriction, Figure 7A). This species is being formally described by F. Langeani and collaborators (Langeani personal communication to WJEMC, 2 January 2024).

The C. variegata complex, comprising Cambeva variegata, Cambeva sp. 3, Cambeva sp. 4, and Cambeva sp. 5, was corroborated in the phylogenetic analysis with high support values (Figure 1). In addition to the unique character states described for Cambeva sp. 1 that are not present in the C. variegata complex (see above), species of this complex are distinguished from all congeners by a relatively short posterior process of the vomer, which is about 1.5 times longer than the vomer length excluding the posterior process (Figure 3F; vs. about two times longer or more, Figure 3G). Relationships within the C. variegata group were weakly supported, except for the sister group relationships between Cambeva sp. 3 and Cambeva sp. 4, two species occurring in close localities of the upper Rio Paraná basin at Serra da Canastra (Figure 6).

3.2. Time-Calibrated Phylogeny

According to the time-calibrated analysis, the origin of Cambeva would have occurred in the late Oligocene, whereas the origin of CVG occurred in the middle Mioceno, which was followed by the split between Cambeva sp. 1 and the C. variegata complex in the late Miocene, with the diversification within the C. variegata complex occurring between the Pliocene and Pleistocene (Figure 9).

4. Discussion

4.1. Phylogenetic Relationships

Species of the CVG share four morphological character states that are unique among congeners: 1—the presence of an adipose crest on the caudal peduncle; 2—an interrupted supraorbital canal, with an s4 pore (Figure 2); 3—a pronounced narrowing at the lateral end of the premaxilla (Figure 3); 4—a ventral expansion in the preopercle (Figure 4). However, the first three character states are also present in species of the genus Scleronema [4,5,6], which is sister to Cambeva [2,33], but not in Trichomycterus s.s. [14], sister to the clade comprising Cambeva and Scleronema [2].

Considering only the distribution of these morphological character states, the most plausible hypothesis would be to interpret Cambeva as paraphyletic, with the CVG being more related to Scleronema than to the other lineages of Cambeva. However, all molecular phylogenetic analyses available to date point to the monophyly of Cambeva when including the species of the CVG (e.g., [2,9,10]), although the position of this group within Cambeva was different in previous studies (i.e., sister to Cambeva iheringi instead of to the Cambeva gama-clade as in the present study). It is important to note that the topology found in previous studies that used a smaller sample of genetic markers, where the CVG is sister to C. iheringi, is congruent with the unique possession of a long supraorbital sesamoid in these two groups (Figure 3), which does not occur in other species of Cambeva and Scleronema. The possibility of the three conditions shared by species of the C. variegata group and Scleronema consisting of primitive conditions present in the most recent common ancestor of the clade comprising Cambeva plus Scleronema and subsequently lost in other lineages of Cambeva cannot be ruled out. An accurate optimisation of these character states depends on a phylogeny where all Cambeva lineages are positioned with high support values, which is still not available.

Monophyly of the C. variegata group is tentatively supported by the presence of a ventral expansion in the preopercle (character state 4 above), a condition variable within the CST clade. A rudimentary or absent middle foramen of the parurohyal (Figure 5B–E) may be considered further evidence of monophyly if admitting a reversion in Cambeva sp. 3, which is the only species of the group with a broad foramen (Figure 5A) and appearing in an apical position within the CVG topology (Figure 1).

4.2. Distribution Patterns

The geographical distribution of the CVG comprises a wide region of southeastern Brazil encompassing the upper Rio São Francisco basin and the Rio Grande drainage, which is a part of the Rio Paraná basin (Figure 6). As in most groups of Trichomycterinae, the distribution of the CVG is concentrated in mountain ranges, such as the Serra da Canastra, Serra do Espinhaço, and Serra da Mantiqueira. Geological data indicate that the Serra da Mantiqueira region, the area of endemism of Cambeva sp. 1, which is strongly supported as the sister group to all other CVG species, already acted as a watershed between the basins located south of this mountain range and the Rio Grande drainage at least since the Eocene, which happened synchronously with the installation of the Continental Rift of southeastern Brazil [34]. Evidence indicates that the natural flow path of the Rio Grande drainage course headed towards the São Francisco River basin, with which it was connected in the past [35]. Thus, at least during the Paleogene and part of the Neogene, the Rio Grande drainage acted as an upper tributary of the Rio São Francisco basin, not as a main tributary of the Rio Paraná River as presently. Therefore, the origin of the CVG here estimated to have occurred in the Middle Miocene (Figure 9) portrays a palaeogeographic scenario in which the Rio Grande drainage and the São Francisco basin formed a single hydrographic basin.

Geological data indicate that the disruption of the ancient connections between the current Rio Grande drainage and the São Francisco River basin, and the consequent capture of the Rio Grande drainage by the Rio Paraná basin, began in the Middle Miocene, following a process of widespread uplift in the region [35]. This historical process is compatible with the divergence found between Cambeva sp. 1, endemic to the Rio Grande drainage, and the C. variegata complex, endemic to the upper Rio São Francisco basin and a small area of the Rio Grande drainage on the periphery of the upper Rio São Francisco, which would have occurred at the end of the Miocene according to our estimates (Figure 9). Therefore, a major rupture in the Grande-São Francisco palaeo basin would be responsible for a vicariance event involving Cambeva sp. 1 and the Cambeva variegata complex. A similar vicariance event was hypothesised by Vilardo et al. [36] for lineages of the subgenus Paracambeva Costa, 2021, of the genus Trichomycterus inhabiting the same area.

Additionally, geological data suggest that the drainage rearrangement was a gradual process, reaching the current hydrographic configuration only in the Pliocene or early Pleistocene [35]. Precisely, the area of the Serra da Canastra region that separates the distribution of C. variegata, endemic to the São Francisco River basin, from the area inhabited by the clade comprising Cambeva sp. 3 and Cambeva sp. 4 (Figure 6) is one of the points of altimetric anomalies, characterising an area of probable past connection between the drainage of the Rio Grande and the Rio São Francisco [35]. Therefore, despite the low support values found at internal nodes of the C. variegata complex, making the relationship hypothesis weak, the estimated split between those two lineages in the Pliocene may indicate a past rupture event in this particular area, artificially restored with transposition works during construction of the Furnas hydroelectric power dam carried out in the 1960s [37].

5. Conclusions

This study reports the occurrence of the genus Cambeva in a vast area of southeastern Brazil, including the first records of the genus in two mountain ranges considered among the most important centres of biodiversity in the world, with high rates of species endemism: the Serra do Espinhaço and Serra da Mantiqueira [38,39]. Over 40 trichomycterines of the genus Trichomycterus s.s. have been continuously reported from these two mountain ranges since the nineteenth century [40,41] and early twentieth century [42,43] to the present (e.g., [2] and included references), where they are usually common in every river drainage, contrasting with only two species of the CVG reported here for the first time. Therefore, this study demonstrates the relative rarity of species of the CVG and its important role in recovering ancient biogeographic patterns, highlighting the urgent need for more detailed studies on the diversity and phylogenetic relationships of still poorly known organisms in this highly diverse and threatened region.