Evolutionary Tracing and Taxonomic Implications of the Mitochondrial Genome of Gephyrocharax atracaudatus (Meek and Hildebrand, 1912)

Simple Summary

Gephyrocharax atracaudatus originally belonged to “Characidae, Characiformes”. With the publication of more sequences, it seems that new insights have been gained regarding the classification of Characidae: the original Characidae should be divided into four new families (Spintheriobolidae, Stevardiidae, Characidae and Accentorhamphidae). This study jointly reveals the taxonomic status of G. atracaudatus through gene composition, codon usage characteristics, evolutionary pressure, and phylogenetic relationships. Concurrently, the evidence presented herein supports the categorisation of G. atracaudatus as part of the order Characiformes, specifically within the subfamily Stevardiinae and the family Stevardiidae (one of the four new families). This study establishes a robust model framework for understanding the evolution of G. atracaudatus, thereby providing fundamental data that supports the elucidation of the evolution mode of Characiformes.

Abstract

Gephyrocharax atracaudatus is a rare freshwater fish in Panama. This study provides a comprehensive analysis of the mitochondrial genome of G. atracaudatus, highlighting its gene composition, codon usage, evolutionary pressures, and phylogenetic relationships. The findings reveal unique evolutionary patterns and divergence times of G. atracaudatus that enhance the understanding of the genetic diversity within Characiformes. The increasing availability of genomic data has prompted taxonomic revisions for numerous Characiformes species, encompassing corrections to species names, subfamily, and family classifications. This study reconfirmed the classification of G. atracaudatus as “Gephyrocharax, Stevardiinae, Stevardiidae, Characiformes”. A cross-analysis model for species differentiation and tracking was established using divergence time comparison and phylogenetic analysis. The results show that G. atracaudatus, Hyphessobrycon roseus and Pristella maxillaris converge on one branch, indicating that the genetic relationship may be the most similar, and it occurred approximately 90.45 Mya in the Cretaceous. This study establishes a robust model framework for understanding the evolution of G. atracaudatus, and correctly determines the biological classification status of G. atracaudatus, providing basic data support for clarifying the evolution mode of Characiformes.

1. Introduction

The mitochondrial genome (mitogenome) of vertebrates is compact and maternally inherited [1]. These features make it an excellent molecular marker for reconstructing phylogenetic relationships and estimating divergence times. High-throughput sequencing has rapidly increased the number of available whole mitochondrial genome sequences in recent years [2,3]. This expansion provides high-resolution data for understanding vertebrate evolutionary history across various taxonomic levels. Maternal inheritance and the high mutation rate of mitochondrial DNA supply essential data for reconstructing vertebrate lineages, identifying species, and defining conservation units [4]. When combined with phenotypic and ecological data, mitochondrial genomes can also indicate potential adaptive differentiation [5]. A typical vertebrate mitochondrial genome contains 13 protein-coding genes, two ribosomal RNA genes (12S rRNA and 16S rRNA), and 22 transfer RNA genes [6]. Detailed analysis of base composition bias and codon usage patterns in these sequences reveals the selective constraints and evolutionary dynamics acting upon them [7]. This approach helps identify adaptive changes that may occur under specific ecological conditions [8,9]. It thus provides valuable evidence for understanding how vertebrates respond to environmental changes. At the same time, integrating mitochondrial genome variation data from multiple species supports the assessment of genetic diversity [5]. This integration lays the groundwork for future incorporation of nuclear genome information and the construction of a more comprehensive phylogenetic framework.

Gephyrocharax atracaudatus (Meek and Hildebrand, 1912) represents a species that is emblematic of the genus Gephyrocharax. This species is found in the central-eastern region of Panama, where it occupies small, independent watersheds along both the Pacific and Caribbean coasts [10]. Its limited geographic range makes it an ideal system for investigating the formation and ecological differentiation of local freshwater fish species. The species is relatively small in body size and exhibits strong habitat specificity for particular microhabitats. These characteristics render it an accessible model for studying ecological interactions and evolutionary relationships among populations. Within its distribution range, G. atracaudatus frequently co-occurs with closely related species in the same ecological domain. This pattern reflects the potential role of watershed dynamics as a driving force in species formation [11]. The topography of Panama is characterized by high spatial heterogeneity and extensive water system isolation. These features collectively provide a natural experimental setting for examining the roles of geographic barriers and environmental factors in genetic differentiation [11].

In previous studies, G. atracaudatus was typically defined as belonging to Characidae [10]. However, the study have indicated that species assignments within Characidae may contain taxonomic errors [12]. Recent research using ultraconserved elements (UCEs) has revealed that Gephyrocharax valenciae is more closely related to Corynopoma riisei than to other Gephyrocharax species, including Gephyrocharax venezuelae and Gephyrocharax machadoi [13]. These species do not form a monophyletic group. This finding suggests that the conventional classification of Gephyrocharax may be problematic. The same study further proposes dividing the traditional Characidae into four distinct families: Spintheriobolidae, Stevardiidae, Characidae, and Accentorhamphidae. Under this new framework, Gephyrocharax is classified within Stevardiidae [13]. Because G. atracaudatus belongs to Gephyrocharax, its systematic position requires re-examination. Consequently, mitochondrial genome data for this species are urgently needed to verify its precise taxonomic placement within this revised familial framework.

The advent of molecular sequencing technology has enabled routine acquisition of complete mitochondrial genome data. This development has substantially enhanced the robustness of phylogenetic analysis. It overcomes the limitations of traditional single-gene fragments regarding node support and topological consistency [3]. Compared with morphological classification, genomic information provides higher resolution for reconstructing evolutionary relationships [2,14,15]. Single-gene fragments, such as COX1 and CytB, have provided fundamental evidence for species identification and phylogenetic inference [16,17]. However, these markers contain limited information for analyzing deep temporal divergence or complex lineage relationships. At this point, multi-gene analyses covering the entire mitochondrial genome can provide more reliable phylogenetic signals. To address this gap, we conducted a second sequencing and systematic analysis of the complete mitochondrial genome of G. atracaudatus. We built upon the previous assembly [10]. For the first time, we estimated lineage divergence times based on this dataset. We also evaluated the systematic position of this species within Characidae. Compared with the costly and time-consuming process of de novo assembly of UCEs, mitochondrial sequencing offers a rapid and cost-effective approach for elucidating evolutionary relationships. This study provides new molecular evidence for reclassifying subfamilies within Characidae. It thus supplements the fragmented analyses of G. atracaudatus from previous studies. This work lays the foundation for subsequent biodiversity assessment and ecological research.

2. Materials and Methods

2.1. Sample Collection, DNA Extraction, PCR Amplification, and Sequencing

Specimens of G. atracaudatus were collected from Flamenco Island, Panama (9°2′7″ N, 79°28′43″ W) on 16 August 2017 (Figure 1). One specimen (Carcass length: 33.6 mm, Female) was deposited in the Laboratory of the Museum of Materials and Environmental Engineering (Zhaowen Liu, liuzhaowen92@163.com) at Chizhou University, under voucher number GR2102145.

The experiment on the specimen was approved by the Animal Ethics Committee of Chizhou University (CZU-AEC-2017-06-04), and the experimental details were recorded in accordance with the ARRIVE 2.0 guidelines [18]. Following sample fixation, 20 mg of muscle tissue was obtained and total DNA was extracted using a modified phenol chloroform method with a 260/280 ratio ≥ 1.8 [3]. The complete mitochondrial genome was obtained through Illumina NovaSeq 6000 platform 150 bp double-ended sequencing. The high variability areas of the gap and control region were validated by Sanger sequencing [19]. The primer design was based on the mitochondrial genome of G. atracaudatus (MH636341) on NCBI (Supplementary Table S1), and the PCR system and cycling parameters were strictly followed in accordance with the instructions provided by specifications of the reagent kit (Takara, Beijing, China: PrimeSTAR^® Max DNA Polymerase).

2.2. Sequence Analysis, Assembly, and Mitochondrial Genome Annotation

The mitochondrial genome sequence was assembled in CodonCode Align 5.1.5 (CodonCode Corporation, Dedham, MA, USA) to obtain a complete circular DNA molecule. Subsequently, the MITOS online (http://mitos2.bioinf.uni-leipzig.de, accessed on 29 April 2026) server was employed for automatic annotation, based on the characteristics of metazoan mitochondria. Thereafter, a combination of the results of tRNAscan SE and further manual review and correction of the tRNA secondary structure predicted by MITOS was conducted (Supplementary Figure S1) [20].

2.3. Amino Acid Composition and Nucleotide Substitution Saturation Index of PCGs

The codon usage count and the relative synonymous codon usage (RSCU) value are calculated in MEGA 11 [21]. The non-synonymous mutation rate (Ka), synonymous mutation rate (Ks), and the Ka/Ks ratio of each PCG were estimated using DnaSP 5 with the Nei-Gojobori method (Jukes-Cantor correction) [22]. The nucleotide substitution saturation of PCG was evaluated in DAMBE 7: initially, the genetic distance is estimated using the TN93 model, and subsequently, a scatter plot is generated of the conversion and crossover on the genetic distance [23].

2.4. Relative Evolutionary Rate Analysis

We downloaded nucleotide sequences of 13 mitochondrial protein-coding genes from 20 species in NCBI (

Table 1. Species attribution and accession number.

Species	Former Name	Subfamily	Family	Order	Accession No.
Gephyrocharax atracaudatus	\	Stevardiinae	Stevardiidae	Characiformes	NC_042882
Knodus borki	\	Diapominae			NC_070390
Psalidodon anisitsi	Hyphessobrycon anisitsi	Acestrorhamphinae	Acestrorhamphidae		NC_066994
Astyanax lacustris	\				NC_053756
Psalidodon rivularis	\				NC_053757
Grundulus bogotensis	\	Grundulinae			NC_026195
Nematobrycon palmeri	\	Rhoadsiinae			NC_051983
Hyphessobrycon amapaensis	\	Hyphessobryconinae			NC_066989
Hyphessobrycon herbertaxelrodi	\				NC_050876
Hyphessobrycon heterorhabdus	\				NC_080887
Hyphessobrycon pulchripinnis	\				MW 331227
Hyphessobrycon roseus	\				MW 315749
Inpaichthys kerri	\	Thayeriinae			NC_057167
Megalamphodus megalopterus	Hyphessobrycon megalopterus	Megalamphodinae			NC_053878
Megalamphodus socolofi	Hyphessobrycon socolofi				NC_066990
Paracheirodon axelrodi	\				AB 898197
Paracheirodon innesi	\				KT 783482
Pristella maxillaris	\	Pristellinae			NC_066992
Hemigrammus rodwayi	Hemigrammus armstrongi				NC_066991
Salminus brasiliensis	\	Salmininae	Bryconidae		NC_024941
Brycon nattereri	\	Bryconinae			NC_051927
Eurypharynx pelecanoides	\	Eurypharynx	Eurypharyngidae	Saccopharyngiformes	AB 046473
Saccopharynx lavenbergi	\	Saccopharynx	Saccopharyngidae		AB 047825

). We aligned each gene at the codon level using MUSCLE v3.8.31 (https://drive5.com/muscle/, accessed on 28 April 2026). We then concatenated the aligned sequences in the same species order using SequenceMatrix v1.7.8. This generated a 20 × 13 supermatrix. We implemented Tajima’s relative rate test using the TPCV module in the LINTRE (version 1) software package. We used the concatenated dataset for this analysis. We also evaluated significant rate heterogeneity in each branch [24].

2.5. Divergence Time Estimation

We estimated species divergence times in BEAST 2 using a strict molecular clock model. We implemented the following settings in the XML file. The nucleotide substitution model was GTR with four categories of discrete gamma distribution [25]. The tree prior was the Yule pure process (Yule Model). Log normal priors were set based on fossil records taken from the oldest credible fossil age of the corresponding node in the TimeTree database (http://www.timetree.org, accessed on 28 April 2026). The chain length was 10,000,000 generations, sampled once every 1000 generations, with the top 10% used as burn-in (MCMC). Convergence diagnosis was performed, ensuring that all parameters had an ESS greater than 200 with Tracer v1.7.2. Subsequently, the results were visualised using TVBOT to generate and plot a time-calibrated Maximum Credibility Tree (https://www.chiplot.online/tvbot.html, accessed on 28 April 2026) [26].

2.6. Phylogenetic Tree Construction

Twenty-one complete Characiformes mitochondrial genomes were downloaded from GenBank (https://www.ncbi.nlm.nih.gov/genbank/, accessed on 28 April 2026) for phylogenetic studies (Table 1). Two Saccopharyngiformes species, Eurypharynx pelecanoides and Saccopharynx lavenbergi, were used as outgroups. Excluding outgroups, we selected three families (Stevardiidae, Acestrorhamphidae and Bryconidae) of the order Characiformes for comparison (Table 1). The alignment of codon level nucleotide sequences of 13 mitochondrial protein coding genes (PCGs) was performed using MEGA 11 [21]. Subsequently, Gblocks 0.91b was utilised to eliminate segments exhibiting substandard alignment quality (allowing gap positions < 50% and minimum block length ≥ 10 bp) [27]. The reconstruction of phylogenetic relationships was conducted utilising Bayesian inference (BI) (Sample every 1000 generations and discard the first 25%, MrBayes 3.2.6) [28,29] and maximum likelihood (ML) (Node support is evaluated 1000 times) methodologies [30]. The resulting phylogenetic trees were visualized using FigTree v. 1.4.4 and its tools (https://itol.embl.de, accessed on 28 April 2026) [31].

3. Results

3.1. Characteristics, Structure and Overlapping of the Mitogenome

The complete mitochondrial genome of G. atracaudatus spans 17,049 bp, as per GenBank accession MH636341 (Figure 2). The circular mitochondrial genome contained 13 protein-coding genes (PCGs), 2 ribosomal RNA genes (12S rRNA and 16S rRNA), 22 transfer RNA genes, and a non-coding control region (D-loop). The L Strand hosted a smaller subset of genes, including 8 tRNAs: tRNA-Gln, tRNA-Ala, tRNA-Asn, tRNA-Cys, tRNA-Tyr, tRNA-Ser, tRNA-Glu, and tRNA-Pro. The remaining 14 tRNAs were positioned on the H strand. The tRNA genes ranged in length from 66 to 75 bp, reflecting their role in amino acid transfer during protein synthesis. Both the small (12S rRNA, 951 bp) and large (16S rRNA, 1683 bp) ribosomal subunit genes were present, with tRNA-Phe and tRNA-Leu positioned on the H strand, separated by tRNA-Val (

Table 2. Features of the mitochondrial genomes of G. atracaudatus.

Mitogenome	Position		Length	Amino	Start/Stop	Intergenic Region (bp) *	Strand #
	From/To		(bp)	Acid	Codon		Heavy/Light
tRNA-Phe (F)	1	68	68			0	H
12S RNA	69	1019	951			0	H
tRNA-Val (V)	1020	1091	72			0	H
16S RNA	1092	2774	1683			0	H
tRNA-LeuUUA (L1)	2775	2849	75			0	H
ND1	2850	3821	972	324	ATG/TAA	0	H
tRNA-Ile (I)	3833	3904	72			11	H
tRNA-Gln (Q)	3973	3903	71			−2	L
tRNA-Met (M)	3984	4054	71			10	H
ND2	4056	5114	1059	353	ATG/TAG	1	H
tRNA-Trp (W)	5133	5205	73			18	H
tRNA-Ala (A)	5243	5168	76			34	L
tRNA-Asn (N)	5317	5246	72			2	L
tRNA-Cys (C)	5414	5349	66			32	L
tRNA-Tyr (Y)	5484	5414	71			−1	L
COX1	5486	7045	1560	520	GTG/AGG	1	H
tRNA-SerUCA (S1)	7104	7033	72			−13	L
tRNA-Asp (D)	7108	7179	72			3	H
COX2	7193	7880	688	229	ATG/T	13	H
tRNA-Lys (K)	7887	7953	67			6	H
ATP8	7955	8122	168	56	ATG/TAG	1	H
ATP6	8113	8795	683	227	ATG/TA	−10	H
COX3	8795	9578	784	261	ATG/T	−1	H
tRNA-Gly (G)	9579	9651	73			0	H
ND3	9652	10000	349	116	ATG/T	0	H
tRNA-Arg (R)	10001	10069	69			0	H
ND4L	10070	10366	297	99	ATG/TAA	0	H
ND4	10360	11740	1381	460	ATG/T	−7	H
tRNA-His (H)	11741	11809	69			0	H
tRNA-SerAGC (S2)	11810	11877	68			0	H
tRNA-LeuCUA (L2)	11879	11951	73			1	H
ND5	11952	13787	1836	617	ATG/TAA	0	H
ND6	14299	13784	516	171	ATG/TAG	−4	L
tRNA-Glu (E)	14367	14300	68			0	L
Cyt b	14371	15507	1137	379	ATG/TAA	3	H
tRNA-Thr (T)	15512	15585	74			4	H
tRNA-Pro (P)	15652	15582	71			−4	L
Dloop	15653	17049	1397			0	H

* Intergenic region: non-coding bases between the feature on the same line and the line below, with a negative number indicating an overlap. ^# H: heavy strand; L: light strand.

). Compared with genes in other coding regions, ND6 exhibits unique and typical arrangement features on the mitochondrial genome, located on the L chain (Table 2) (Figure 2).

3.2. Protein-Coding Genes and Codon Usage and Mitogenome Mutations

The coding region of the mitochondrial genome was 11,420 base pairs long, with 66.89% of it dedicated to PCGs (Table 2). GTG was the start codon of COX1, and other PCGs included ATG (ATG is a common start codon in the mitochondrial genome [5]). The Relative Synonymous Codon Usage (RSCU) analysis revealed notable variations in codon usage frequencies, highlighting evolutionary selection pressures on amino acids (Figure 3). Leu1, Thr, Ala, and Gly were more abundant, and the usage of four different codon types for Leu1, Val, Ser2, Pro, Thr, Ala, Arg, and Gly pointed to a buffering mechanism against genetic mutations (Figure 3). The TN93 (Tamura Nei, 1993) model is a commonly used nucleotide substitution model in molecular phylogenetics, which can simultaneously correct differences in nucleotide conversion rates and base frequency biases [32]. This study employed the TN93 model to analyze base substitution ratios and nucleotide frequencies across the three codon positions (Figure 4). Additionally, a network diagram was created to visualize correlations between gene fragments in the mitochondrial genome, including PCGs, rRNAs, and tRNAs (Supplementary Figure S1). Nodes represents individual gene fragments, while edges indicates significant correlations in sequence similarity or evolutionary rates.

3.3. Evolutionary Relationships in Stevardiidae Family

In this study, we selected twenty representative mitogenomes of Stevardiidae to assess the evolutionary selection pressure on G. atracaudatus. Additionally, we included Salminus brasiliensis (Salminus, Salmininae, Characiformes) and Brycon nattereri (Brycon, Bryconinae, Characiformes), species with close phylogenetic relationships to G. atracaudatus, as well as E. pelecanoides and S. lavenbergi, which are less related, to validate the accuracy of our results through comparative analysis. Based on the gene sequences of 13 PCGs from the mitogenome, we assessed the relative evolutionary pressures across different species. The results (Figure 5A) revealed distinct pressure patterns among species from different families and genera. Compared to the evolutionary pressure faced by G. atracaudatus, the pressure on Psalidodon anisitsi, Hyphessobrycon amapaensis, H. heterorhabdus, Megalamphodus megalopterus and P. rivularis seem to be smaller. Hyphessobrycon contains a greater number and richness of species, with over 160 species discovered [33]. In contrast, Grundulus bogotensis and B. nattereri were under greater relative evolutionary pressure, Species with comparable evolutionary pressure indices to that of G. atracaudatus include Hemigrammus rodwayi, H. herbertaxelrodi and Paracheirodon axelrodi. When analyzing synonymous mutations in amino acids, the Ka and Ks values for different species showed relatively similar Ka/Ks ratios (Figure 5B). In Figure 5B, the Ka values for most species were smaller than their respective Ks values, and these values varied proportionally. However, G. atracaudatus stood out with a higher Ka/Ks ratio, a pattern also observed in P. axelrodi and P. rivularis.

3.4. Divergence Time and Phylogenetic Analysis

In this study, the PCGs of the mitochondrial genomes from twenty-two species were used to construct the evolutionary time tree (Figure 6). The relative differentiation times within the life evolution scale were calculated using the species pairs E. pelecanoides vs. S. lavenbergi and B. nattereri vs. S. brasiliensis (Supplementary Figure S2). B. nattereri and S. brasiliensis, as outgroups to Characiformes (Table 1), have a divergence time of 35.5–34.0 Mya. A divergence time of 35 Mya was selected as the optimal unit scale based on their species affinity (Supplementary Figure S2A). Similarly, E. pelecanoides and S. lavenbergi, as outgroups to Characiformes, have a divergence time ranging from 118.3–25.9 Mya. A divergence time of 39 Mya was chosen as the optimal unit scale for this pair, based on their species affinity (Supplementary Figure S2B). The results (Figure 6) (Table 1) suggested that the initial divergence between Saccopharyngiformes and Characiformes occurred approximately 361.80 Mya. In the divergence time topology, Characiformes species appeared to have split into two distinct clusters (Figure 6). In the first cluster, new species diverged approximately every 50–40 Mya, a pattern that seems consistent over time. Within the second clade, G. atracaudatus appeared to belong to an even older lineage. This species diverged from Hyphessobrycon roseus and Pristella maxillaris during an extensive differentiation period lasting approximately 90 million years. At the same time, an inter-species correlation network diagram was constructed based on statistical analysis of mitochondrial genome data from different species (

Table 3. The sequence characteristics and statistical analysis of 13 protein-coding genes (PCGs), 22 transfer RNAs (tRNAs) and 2 ribosomal RNAs (rRNAs) in the mitochondrial genome of G. atracaudatus were investigated. The analysis was performed using gene sequences from the mitochondrial genome, which were processed using DnaSP v6 software.

Gene	G + C Content	Total Number of Mutations	Nucleotide Diversity (ND)	ND’s Standard Deviation	Average Number of Nucleotide Differences	Number of Haplotypes	Haplotype Diversity	Standard Deviation of Haplotype Diversity	Fu’s Fs Statistic	Tajima’s D	Number of Segregating Sites Analyzed	Fu and Li’s D-Star Test Statistic	Fu and Li’s F-Star Test Statistic	Achaz Y-Star Test Statistic
tRNA-Phe (F)	0.41	98	0.31	0.051	21.4	19	0.99	0.018	−3.6	−0.874	22	−0.145	−0.06	0.371
tRNA-Val (V)	0.5	28	0.08	0.008	5.7	17	0.98	0.02	−8.47	−1.03	20	−0.316	−0.521	−1.092
tRNA-LeuUUA (L1)	0.51	85	0.26	0.038	19.6	19	0.99	0.018	−4.01	−0.696	33	−0.413	−0.424	−0.156
tRNA-Ile (I)	0.51	68	0.25	0.052	18	20	0.99	0.016	−6.22	−0.188	41	1.567	1.275	−0.701
tRNA-Gln (Q)	0.39	67	0.22	0.031	15.9	17	0.98	0.02	−2.42	−0.578	39	−1.14	−0.983	0.608
tRNA-Met (M)	0.4	115	0.36	0.05	25.2	17	0.99	0.021	−2.06	−0.979	28	−0.105	−0.2	−0.491
tRNA-Trp (W)	0.4	133	0.45	0.033	31	20	0.99	0.016	−3.56	−0.664	13	0.311	0.177	−0.529
tRNA-Ala (A)	0.38	36	0.1	0.014	7	18	0.99	0.019	−8.79	−1.178	22	−1.146	−1.263	−1.067
tRNA-Asn (N)	0.48	38	0.09	0.018	6.7	17	0.98	0.02	−7.28	−1.448	27	−2.323	−2.267	−0.547
tRNA-Cys (C)	0.49	127	0.46	0.038	30.1	20	0.99	0.016	−3.68	−0.601	12	0.627	0.615	0.053
tRNA-Tyr (Y)	0.49	67	0.21	0.045	14.8	19	0.99	0.018	−5.47	−0.822	33	0.284	0.039	−0.978
tRNA-SerUCA (S1)	0.48	60	0.14	0.059	10.1	10	0.89	0.039	1.77	−1.58	36	0.726	0.119	−2.195
tRNA-Asp (D)	0.36	116	0.44	0.042	31.3	18	0.98	0.019	−1.04	−0.116	19	1.008	1.113	0.559
tRNA-Lys (K)	0.5	107	0.25	0.067	18.2	17	0.98	0.02	−1.9	−1.577	26	0.127	−0.138	−1.143
tRNA-Gly (G)	0.3	101	0.4	0.037	28	17	0.98	0.02	−0.47	−0.007	24	0.465	0.522	0.324
tRNA-Arg (R)	0.41	74	0.28	0.043	19.4	19	0.99	0.018	−4.04	−0.22	36	0.565	0.601	0.251
tRNA-His (H)	0.31	75	0.19	0.046	13.4	17	0.98	0.02	−3.16	−1.438	32	0.061	−0.317	−1.667
tRNA-SerAGC (S2)	0.48	97	0.23	0.058	15.6	18	0.98	0.019	−3.69	−1.713	30	−1.947	−2.055	−1.684
tRNA-LeuCUA (L2)	0.42	50	0.09	0.042	6.7	12	0.89	0.059	1.28	−2.067	36	−3.139	−3.042	−0.961
tRNA-Glu (E)	0.42	65	0.22	0.05	15.2	17	0.97	0.026	−2.61	−0.631	34	0.835	0.64	−0.582
tRNA-Thr (T)	0.49	91	0.35	0.031	25.4	20	0.99	0.0003	−4.41	0.02	36	−0.563	−0.405	0.673
tRNA-Pro (P)	0.36	86	0.26	0.053	18.4	19	0.99	0.018	−4.31	−0.934	36	0.565	0.367	−0.673
ND1	0.42	1825	0.5	0.048	482	20	0.99	0.016	2.65	−0.207	296	1.495	1.584	0.563
ND2	0.41	2565	0.54	0.042	563.5	20	0.99	0.016	2.94	−0.871	59	1.141	1.08	−0.005
ND3	0.43	484	0.31	0.047	106.6	20	0.99	0.016	−0.14	−0.859	175	0.999	0.483	−1.652
ND4L	0.46	284	0.24	0.011	71.9	20	0.99	0.016	−0.99	−0.367	102	0.107	0.175	0.318
ND4	0.42	1850	0.28	0.033	382.4	20	0.99	0.016	2.23	−1.065	619	−2.405	−2.261	0.089
ND5	0.4	3525	0.35	0.061	642.6	20	0.99	0.016	3.18	−1.431	467	−0.456	−0.739	−1.493
ND6	0.42	939	0.35	0.042	180.4	20	0.99	0.016	0.87	−1.282	149	−1.564	−1.576	−0.655
COX1	0.44	1025	0.18	0.006	274.5	20	0.99	0.016	1.63	−0.152	350	0.992	1.116	0.637
COX2	0.42	1122	0.28	0.059	192.2	19	0.99	0.018	2.47	−1.593	270	−1.865	−1.94	−1.335
COX3	0.45	470	0.16	0.006	125.2	19	0.99	0.018	1.46	−0.173	190	0.573	0.632	0.347
ATP8	0.35	249	0.33	0.029	54.8	20	0.99	0.016	−1.68	−0.857	59	−0.079	−0.176	−0.455
ATP6	0.41	1306	0.39	0.051	264.4	20	0.99	0.016	1.57	−1.128	180	−0.12	−0.363	−1.119
Cyt b	0.42	1345	0.24	0.037	267.8	20	0.99	0.016	1.59	−1.178	559	−2.554	−2.394	0.175
12S RNA	0.46	2535	0.63	0.018	602.4	20	0.99	0.016	3.06	−0.603	50	−0.761	−0.798	−0.411
16S RNA	0.43	4446	0.64	0.019	1060.9	20	0.99	0.016	4.12	−0.588	55	0.766	0.639	−0.332

) (Figure 7). As shown in Figure 7, nodes representing species within the same genus exhibited closer correlations, supporting the phylogenetic relationship and the relationship between evolutionary scales within the Characiformes population, a pattern that aligns with the results in Figure 6. All species, except for those in the cross-cluster (Inpaichthys kerri and G. bogotensis), showed consistency in both the time of divergence and in the construction of the sequence tree (Figure 6 and Figure 8). The tree structures from the two analyses displayed nearly identical clustering topologies. Notably, species within the same Subfamily often clustered with those from other Subfamily. Most species of Characiformes formed natural clusters, with H. roseus being a notable exception. The topology of the divergence times, a pattern also observed in the results of the systematic evolution, indicated that most of the new time points appeared during the Mesozoic period (Figure 8).

4. Discussion

Mitochondrial genomes contain abundant genetic information and serve as powerful molecular markers for establishing phylogenetic relationships and estimating divergence times [3,4]. In this study, G. atracaudatus showed a highly conserved gene arrangement (Figure 2). The use of GTG rather than the traditional ATG as the start codon for COX1 indicates evolutionary flexibility. This variation may reflect adaptations that influence protein synthesis efficiency or mitochondrial function stability [5]. Incomplete stop codons are common in vertebrate mitochondrial genomes. These truncated codons are converted into complete termination signals through mRNA polyadenylation after transcription. Their widespread occurrence reflects strong selection for maintaining a compact genome size and replication efficiency [34]. Complete stop codons such as TAG and TAA also occur and serve as canonical termination signals without post-transcriptional modification. In many fish mitochondrial genomes, both types coexist, reflecting the dual influence of structural constraints and translational efficiency [3].

Codon usage patterns are key to understanding molecular evolution [35]. Amino acids with longer side chains tend to have more codon variants, reflecting their complex structural requirements (Figure 3) [36]. The abundance of Leu1, Thr, Ala, and Gly in this study suggests their critical role in protein synthesis for G. atracaudatus. The use of multiple codons for these amino acids points to a buffering mechanism against genetic mutations (Figure 3). This redundancy allows the genetic code to tolerate base changes without affecting protein function, contributing to genetic stability [37]. Conversely, codons with lower frequency for certain amino acids indicate vulnerability to mutations or drift, potentially leading to changes in codon preferences over time [36].

The TN93 model is commonly used in molecular phylogenetics. It corrects differences in nucleotide substitution rates and base frequency biases [32]. In this study, mutations at different codon positions affected amino acid evolution differently (Figure 4). Non-synonymous mutations at the second codon position significantly impact protein structure, while synonymous mutations at the third position have more subtle effects on genetic variation without altering amino acid sequences (Figure 4B,C) [38,39]. This distinction is crucial for understanding how genetic variations influence evolution. The larger impact of second-position mutations aligns with their critical role in determining amino acid properties [40]. The conservative nature of third-position mutations reflects evolutionary mechanisms that buffer against detrimental changes. The similarity between first and third codon frequency distributions indicates evolutionary conservatism in G. atracaudatus, where most mutations tend to be neutral. This aligns with the theory of purifying selection, where natural selection eliminates deleterious mutations [41].

A network diagram visualized correlations between mitochondrial gene fragments, including PCGs, rRNAs, and tRNAs (Supplementary Figure S1). Nodes represent individual gene fragments, while edges indicate significant correlations in sequence similarity or evolutionary rates. Strong correlations suggest that mitochondrial genes are conserved and co-evolve, likely driven by shared selective pressures and functional interactions [42]. Highly connected nodes indicate that these genes face shared selective pressures, possibly driven by the need for coordinated functionality in energy production and metabolic processes [43]. Stronger correlations may indicate shared selection pressures and evolutionary rates, suggesting that co-evolution is driven by functional interactions [44].

In this study, we selected 20 representative mitogenomes of Stevardiidae to assess the evolutionary selection pressure on G. atracaudatus. We evaluated the relative evolutionary pressure index based on the gene sequences of 13 PCGs from the mitochondrial genome. We compared pattern differences among species (Figure 5A). These species inhabit diverse environments. They may experience relatively lower mutation pressures, potentially due to smaller fluctuations in temperature or light. Compared with Hyphessobrycon, G. atracaudatus faced higher relative evolutionary pressure. This difference could be attributed to its specific habitat. G. atracaudatus is native to Flamenco Island, Panama (9°2′7″ N, 79°28′43″ E). It may be significantly influenced by interactions with other species or by landform evolution. These factors could contribute to greater evolutionary pressures on this species. In contrast, Grundulus bogotensis and Brycon nattereri were under greater relative evolutionary pressure. The specific factors influencing these pressures require further investigation, particularly regarding potential changes in their habitats. Species with comparable evolutionary pressure indices to that of G. atracaudatus include Hemigrammus rodwayi, Hyphessobrycon herbertaxelrodi, and Paracheirodon axelrodi. This similarity may be linked to the genetic stability of their mitochondrial genomes [45]. The consistent mutation pressure index in the mitochondrial genome confirms strict maternal inheritance characteristics. Combined with recorded coastal environmental changes, this consistency indicates that a common selection mechanism serves as an effective driving force for species distribution [46]. In amino acid synonymous mutation analysis (Figure 5B), the small variation in Ka values among different Characiformes species suggests that the frequency of neutral evolution was comparable among species. The accumulation of neutral mutations potentially contributed to the lack of environmental selectivity in their mitogenomes [47]. Compared with other species, G. atracaudatus showed a higher Ka/Ks ratio. This pattern may reflect the unique evolutionary dynamics of the species. The evolutionary selection pressure in P. axelrodi was similar to that in G. atracaudatus regarding the Ka/Ks ratio (Figure 5A). This similarity may be due to comparable habitat conditions. The habitat environments of G. atracaudatus (9°2′7″ N, 79°28′43″ E) and P. axelrodi (3°8′7″ N, 65°5′9″ E) are very similar. These localities are recorded in the Global Biodiversity Information Facility (GBIF) (https://www.gbif.org/species/2353911, accessed on 28 April 2026). This similarity suggests that the external environments and natural pressures faced by these species may follow comparable trends. For other Characiformes species, both constrained and divergent evolution appeared to be closely linked to gene mutations (Ka/Ks). Generally, evolutionary selection of mitogenomes tends to eliminate harmful mutations. This process preserves the stability of amino acid sequences over time [48].

Divergence time estimation has gained significant attention in evolutionary biology in recent years [49]. Advances in methodology and empirical research now allow for more accurate time tree estimations than ever before [29]. The molecular clock theory represents a key method for assessing genetic differences between species or populations [50,51]. It assumes that gene mutations accumulate at a relatively constant rate over time within a given lineage. This assumption allows scientists to estimate divergence times between species based on genetic variations [52]. This principle extends to biomolecules such as DNA and proteins, where mutations are presumed to occur at a predictable rate throughout evolution [53]. If mutation accumulation on an evolutionary branch is proportional to the duration of its independent evolutionary history, the substitution rate is expected to remain roughly constant [54,55]. This suggests that the mutation rate tends to stabilise over evolutionary time [51]. Molecular clocks often rely on specific genetic markers, such as mitochondrial DNA, ribosomal RNA genes, or certain protein-coding genes [56]. These markers are considered neutral because they are less influenced by selection pressures. They provide a more stable rate of evolution compared with other genomic regions [57]. By tracking the frequency of these mutations, we can estimate the time since two species or populations shared a common ancestor [53,57]. However, mutation rates are not always constant in practice. Factors such as selection pressure, environmental influences, and genetic drift can cause significant variation in mutation rates. This variation leads to deviations from the expected molecular clock [58,59]. Despite these challenges, molecular clocks remain a valuable tool in evolutionary biology. They offer important insights into the timing of evolutionary events.

In this study, the temporal divergence tree showed that the initial split between Saccopharyngiformes (outgroup) and Characiformes occurred approximately 361.8 million years ago. The families Acestrorhamphidae and Stevardiidae appeared approximately 316.5 million years ago, according to mitochondrial genome analysis (Figure 6) (Table 1). In the divergence time topology, Characiformes species split into two distinct clusters (Figure 6). In the first cluster, new species diverged approximately every 40 to 50 million years. This pattern appears consistent over time and may reflect the natural dynamics of species evolution. Environmental factors such as latitude, light, and temperature likely influenced convergent evolution and drove divergent evolutionary processes between regions [45]. However, comparison of mitochondrial genomes alone is insufficient to fully explain this phenomenon. A comprehensive analysis of the distribution ranges of all species worldwide is essential. Within the second clade, G. atracaudatus belonged to an older lineage. This species diverged from Hyphessobrycon roseus and Pristella maxillaris during an extensive differentiation period lasting approximately 90 million years. This result and the new classification are mutually corroborative. They support the recognition of Stevardiidae as sister to Acestrorhynchidae [12,13]. However, the temporal divergence tree constructed from the mitochondrial genome suggests that Stevardiidae evolved from Acestrorhamphidae. Therefore, more evidence is needed to validate Stevardiidae as an independent family from Acestrorhamphidae.

In the interspecies network diagram (Table 3) (Figure 7), different groups or isolated nodes may represent distinct evolutionary paths. These paths may be shaped by different selection pressures, mutation hotspots, or environmental factors. The nodes of species within the same genus show closer correlation (Figure 7). This pattern supports the relationship between phylogenetic relationships and evolutionary scales in Characiformes (Figure 6). These insights provide a foundation for further research into mitochondrial dysfunctions, evolutionary biology, and species-specific adaptations [6,26,54]. Combining phylogenetic or structural analyses with network data could provide a deeper understanding of mitochondrial genome evolution.

The phylogenetic tree results (Figure 8) were basically consistent with the temporal divergence topology (Figure 6). Both analyses showed the same clustering pattern. Generally, species of the same family and genus cluster together. However, species within the same subfamily often cluster with other subfamilies. This pattern seems related to the limited information content of mitochondrial genomes [19]. It is insufficient to construct trees using only part of the genome. Real divergence between species requires consideration of whether habitat, environmental temperature, or light have changed. The topology of divergence times indicated that most new divergence events appeared during the Mesozoic period (Figure 8). This pattern was also observed in the systematic evolution results. Glacial and crustal shifts likely altered original habitats. These changes may have led to geographical isolation within the same water systems. However, the ancient nature of mitochondrial genomes can result in discrepancies between gene trees and species trees. A more comprehensive analysis incorporating coastal geological changes would provide a clearer understanding of these patterns.

5. Conclusions

This study provides a comprehensive mitochondrial genome analysis of G. atracaudatus. It offers new insights into its molecular characteristics, evolutionary pressures, and phylogenetic position within Characiformes. The mitochondrial genome of G. atracaudatus exhibits a highly conserved gene arrangement. This pattern reinforces its suitability as a reliable marker for evolutionary inference. As more genomes have become publicly available, the species names, former names, and subfamily classifications of many Characiformes species have been corrected. Even family-level classifications have been revised [12,13]. Despite the publication of the complete mitochondrial genome of G. atracaudatus, further analysis is required to clarify its species differentiation and evolution. This study reconfirmed the classification of G. atracaudatus as Gephyrocharax, Stevardiinae, Stevardiidae, Characiformes. It also provided a solid theoretical basis for this classification. Nevertheless, this study has limitations. All phylogenetic and temporal inferences rely solely on mitochondrial genomes. These genomes represent a single, maternally inherited genetic system. This constraint may lead to discrepancies between gene trees and true species histories, particularly in ancient lineages. In addition, environmental interpretations remain indirect. Detailed ecological and geological data were not explicitly integrated. Consequently, while this work establishes a robust mitochondrial framework for understanding G. atracaudatus evolution, future studies are essential. These studies should incorporate nuclear genomes, broader taxon sampling, and environmental context to fully resolve its evolutionary history.