Evidence for Selection on Mitochondrial OXPHOS Genes in the Mediterranean Killifish Aphanius fasciatus Valenciennes, 1821

Simple Summary The old world killifish Aphanius fasciatus is a typical inhabitant of transition waters along the coasts of the central and eastern Mediterranean Sea, where it lives in environments (salt ponds, estuaries, and coastal lagoons) with large variations in salt, oxygen, and temperature. Therefore, A. fasciatus could be an interesting non-model species to be studied to evaluate the response of the mitochondrial oxidative phosphorylation (OXPHOS) genes to the variations and extremes of environmental factors. In this study, the sequences of three OXPHOS genes were analyzed in six populations of A. fasciatus to detect mutations and sites subject to selection. The results indicate that mutations were detected in two genes of the Greek population. Moreover, positively selected sites were also found. The information we obtained from the mitochondrial DNA sequences of A. fasciatus adds to the growing data on selective pressure acting on mitochondrial DNA. These results should be explored from the perspective of the local adaptation of a species highly tolerant to wide fluctuations of environmental parameters and should be supported using experimental evidence to better understand the interplay between historical climatic events and local adaptation and how each of them contributes to shaping the genetic structure of this species. Abstract Mitochondrial oxidative phosphorylation (OXPHOS) genes are a system subject to selection under determined environmental constraints despite a neutral evolution model that has long been hypothesized for the mitochondrial genome. In this study, the sequences of ND1, Cytb, and COI OXPHOS genes were analyzed in six populations of the eurythermal and euryhaline killifish A. fasciatus, to detect non-synonymous mutations leading to amino acid changes and to check whether selection acted on them using tests of recombination and selection. The results indicate a high COI and Cytb gene diversity and a high percentage of private haplotypes in all populations. In the Greek population, non-synonymous nucleotide substitutions were observed in the N-terminal region of COI and Cytb. Positively selected sites were also found. The information we obtained from the mitochondrial DNA sequences of A. fasciatus adds to the growing data on selective pressure acting on mitochondrial DNA in non-model species. These results should be explored from the perspective of the local adaptation of eurythermal and euryhaline species and supported using experimental evidence to better understand the interplay between historical climatic events and local adaptation and how each of them contributes to shaping the genetic structure of this species.


Introduction
The mitochondrial oxidative phosphorylation (OXPHOS) genes have been studied in recent decades as a system that is subject to selection under certain environmental constraints [1][2][3] despite the fact that (i) a high level of preservation of its functions exists among eukaryotes and (ii) a neutral evolution model has long been hypothesized for the mitochondrial genome.This highly conserved system involved in the cellular respiration process for ATP production engages five main multimeric protein complexes (I-V) that together constitute the core of the OXPHOS system.The protein subunits that make up four (complexes I, III, IV, and V) of the five complexes of the electron transport chain are encoded by 13 mitochondrial genes together with several nuclear genes.Focusing on the function of these four complexes, the seven subunits of the NADH dehydrogenase complex I couple the oxidation of NADH to NAD+ with the translocation of 4 protons to the inner mitochondrial space.The cytochrome b (Cytb) subunit of the Cytochrome bc1 complex III is involved in the transfer of electrons from quinol to cytochrome c, which in turn transfers them to the inner mitochondrial membrane.The Cytochrome c Oxidase (COI) complex IV, including the COI, COII, and COIII subunits, generates water molecules from electrons carried by cytochrome c and molecular oxygen.Finally, the ATP synthase complex V, consisting of ATP6 and ATP8 subunits, synthesizes ATP via phosphorylation of ADP by exploiting the electrochemical energy generated using the proton gradient.
All these proteins are located in the inner mitochondrial membrane, and their function is highly conserved.However, it has been assessed that in order to meet the diverse metabolic needs of organisms via energy production, the OXPHOS genes are easily subjected to various selective pressures that promote adaptive changes.Indeed, the analysis of many of these genes has proven successful in identifying the traces of purifying (negative) and adaptive (positive) selection in various organisms associated with their adaptation to particular environmental factors.Evidence of positive selection on metabolism-related genes of mitochondrial DNA (mtDNA) has recently been detected for example in mammalian species of the order Cetartiodactyla [4], in species of the genus Lepus [5], in gentoo penguin Pygoscelis papua [3], in Polyplacophora Molluscs [6], in Orthoptera insects [7], in Drosophila melanogaster [1].In all cited cases, environmental and/or climatic changes appear to drive the species adaptation.
With regard to fishes, according to [8], changes in mtDNA evolution are generally driven by selective pressures to increase aerobic capacity or to adapt to changes in environmental temperature.In this context, the positive selection in the ND1, ND3, and ND4 genes detected in northern populations of Salmo salar has been hypothesized to be linked to the request for higher aerobic capacity at low temperatures [9].On the other hand, it has been demonstrated that thermal environmental stress deeply affects cellular metabolism via the shift of enzyme functions, reduction in OXPHOS efficiency, increased inner membrane proton leakage, and decreased ATP production, all of which adversely affect cellular and lastly, organismal survival [10].In addition, mitochondrial respiration is particularly dependent on habitat temperature in ectotherms, as they depend largely upon environmental radiation to thermoregulate themselves, and it has been demonstrated that environmental temperatures above the optimal temperature for each individual species could impair mitochondrial function [11].However, it is remarkable that fish and mollusk species living in coastal marine environments, which are naturally subject to rapid changes in environmental variables, are able to cope with a wide spectrum of temperature, oxygen, and salinity levels via a rapid modulation of the mitochondrial electron transport system [12,13].
The old world killifish Aphanius fasciatus is a typical inhabitant of transition waters along the coasts of the central and eastern Mediterranean Sea and shows peculiar biological traits.During the breeding season, the species lays large demersal eggs that remain attached to the vegetation; it lacks a larval stage and shows a rapid turnover due to the short generation time and the high reproductive rate.Adults are highly mobile but also show high site fidelity because they only move from one site to another adjacent site during heavy rain events that cause the coastal lake to overflow.Therefore, the gene flow between populations is very limited.Due to its strong euryhalinity and eurythermy, it is adapted to live in environments (salt ponds, estuaries, and coastal lagoons) with large salt, oxygen, Biology 2024, 13, 212 3 of 14 and temperature variations where it goes through its entire life cycle.In particular, this species occupies shallow aquatic environments that can be easily heated locally by solar insolation, where it experiences wide daily and seasonal fluctuations in water temperature along the Mediterranean coast.Therefore, A. fasciatus could be an interesting non-model species to be studied to evaluate the response of the OXPHOS genes to the variations, even extreme ones, in environmental factors.
Based on the above considerations, in this study, we analyze the sequences of ND1, Cytb, and COI genes in six populations of the killifish A. fasciatus sampled along the Italian, Albanian, and Greek coasts.The aims are: (i) to explore the nucleotide sequences of the three OXPHOS mitochondrial genes to detect the presence of mutations leading to nonsynonymous amino acid changes and (ii) to check whether selection acted on them using tests of recombination and selection based on different models of evolution.

Sampling Sites
Six populations of the killifish A. fasciatus were sampled along the coast of Italy (Grado, Tarquinia, and Casaraccio lagoons), Albania (Vain and Karavasta lagoons), and Greece (Korission lake) coasts (Figure 1, Table 1).Most of these coastal lagoons are not very deep, with a complex hydrographical network including tidal flats, tidal channels, and subtidal zones; the Korission lagoon is the deepest one, with a maximum depth of 2 m.
during heavy rain events that cause the coastal lake to overflow.Therefore, the gene between populations is very limited.Due to its strong euryhalinity and eurythermy adapted to live in environments (salt ponds, estuaries, and coastal lagoons) with larg oxygen, and temperature variations where it goes through its entire life cycle.In pa lar, this species occupies shallow aquatic environments that can be easily heated lo by solar insolation, where it experiences wide daily and seasonal fluctuations in temperature along the Mediterranean coast.Therefore, A. fasciatus could be an intere non-model species to be studied to evaluate the response of the OXPHOS genes t variations, even extreme ones, in environmental factors.
Based on the above considerations, in this study, we analyze the sequences of Cytb, and COI genes in six populations of the killifish A. fasciatus sampled along the It Albanian, and Greek coasts.The aims are: (i) to explore the nucleotide sequences three OXPHOS mitochondrial genes to detect the presence of mutations leading to synonymous amino acid changes and (ii) to check whether selection acted on them tests of recombination and selection based on different models of evolution.

Sampling Sites
Six populations of the killifish A. fasciatus were sampled along the coast of (Grado, Tarquinia, and Casaraccio lagoons), Albania (Vain and Karavasta lagoons Greece (Korission lake) coasts (Figure 1, Table 1).Most of these coastal lagoons ar very deep, with a complex hydrographical network including tidal flats, tidal cha and subtidal zones; the Korission lagoon is the deepest one, with a maximum dept m.Being shallow water bodies, their temperatures can be heavily influenced by air temperature: in any case, there is a significant temperature range between summer and winter.For example, in Grado, Karavasta, and Korission, temperature values can range from 6-7 • C to 26-30 • C [ [14][15][16], while in other cases, temperatures can be even higher up until 35-36 • C for Tarquinia [17].
Another important parameter concerns the salinity of the water, as it depends on the degree of freshwater inputs, which are mostly related to the rainfall amount or drainage channels.In some cases, there is a gradient existing within the water mass itself, for example, in Grado and Tarquinia [17,18], and this will regulate the distribution of fish fauna and macrozoobenthic species; while in other cases the salinity changes between seasons, being very much higher in summer, for example in Karavasta [19], Casaraccio [20] and Korission [16] so that they can be characterized by hypersaline water.
As for the trophic state, among the ones taken into consideration, the Casaraccio lagoon is the one with lower organic and mud contents because it has no significant anthropogenic impacts and it has been able to preserve the highest degree of naturalness [20] and also the Grado lagoon does not exhibit significant effects of eutrophication [14].Between the sites taken into account, the Vain lagoon is classified as eutrophic, because it presents green and murky water, with higher amounts of nutrients and algae; this was further demonstrated using the analysis of chlorophylls content which exhibited high values, showing that the high content of Cyanophyceae is linked to a higher eutrophication level [21].
Six populations of the killifish A. fasciatus were sampled along the Italian, Albanian, and Greek coasts for a total of 116 individuals (Figure 1).

Amplification and Sequencing of the OXPHOS Target Genes
Three mitochondrial genes, each encoding different subunits of the OXPHOS complex I (ND1 gene), complex III (Cytb gene), and complex IV (COI gene), were amplified using specific primer pairs reported in Table 2. PCR amplification mixes for each target gene were prepared following [23] and using different thermal profiles (Supplementary Material Figure S1).All obtained amplicons were sequenced at Eurofins Genomics (https://eurofinsgenomics.eu, accessed on 3 September 2023), and sequences were submitted to the GenBank database (Supplementary Material Tables S1-S3).
Table 2. List of primers used in this study.

Gene
ID Primer Sequence Reference

Recombination and Selection Tests
The Genetic Algorithms for Recombination Detection (GARD) (HyPhy package, accessed at www.datamonkey.org,accessed on 20 November 2023) was used to check for the presence of mitochondrial COI, Cytb, and ND1 recombinants.A preliminary one-tailed Z-test in MEGA X [29] was performed to test the presence of selection for each gene dataset.Furthermore, three codon models were used to estimate codons under positive or purifying selection: FEL (Fixed Effects Likelihood) [30], FUBAR (Fast, Unconstrained Bayesian Approximation for Inferring Selection) [31], and MEME (Mixed effects model of evolution) [32].Sites with p-values below 0.05 for FEL and MEME, as well as sites with posterior probability higher than 0.9 for FUBAR, were all considered as being under selection [33].

Protein Structure Analysis
To show the spatial position of sites under positive selection in a 3-dimensional space, we used the SWISS-MODEL server (https://swissmodel.expasy.org/,accessed on 10 December 2023) with default parameters [34].In order to visualize the structural position of positively selected sites, the secondary structures of the three protein models were predicted using the PSIPRED server (http://bioinf.cs.uci.ac.uk/psipred/, accessed on 13 December 2023) [35], and the positions of transmembrane helices were predicted using the MEMSAT-SVM server (http://bioinf.cs.uci.ac.uk/psipred, accessed on 13 December 2023).

Gene Sequences Analysis and Distribution of Genetic Variation
The 116 individuals of killifish A. fasciatus were processed to amplify and sequence the three targeted genes.Sizes of the fragments utilized varied across genes, specifically COI (669 bp), Cytb (1125 bp), and ND1 (909 bp).All sequences produced 55 haplotypes for the COI gene (GenBank Acc.N. OR064287-OR064341), 56 for Cytb (GenBank Acc.N. OR134774-OR134829), and 30 for ND1 (GenBank Acc.N. OR097693-OR097722) (Supplementary Material Tables S1-S3).As expected, genetic diversity indexes varied among analyzed genes (Figure 2).).As expected, genetic diversity indexes varied among analyzed genes (Figure 2).Low values of haplotype diversity were found in the ND1 gene for all populations with the exception of Korission, where h was 0.900 (Figure 2); h values ranging from 0.629 to 0.932 were found in COI e Cytb genes for all populations.In addition, low values of nucleotide diversity were observed for the ND1 gene with the exception of the Grado pop- Low values of haplotype diversity were found in the ND1 gene for all populations with the exception of Korission, where h was 0.900 (Figure 2); h values ranging from 0.629 to 0.932 were found in COI e Cytb genes for all populations.In addition, low values of nucleotide diversity were observed for the ND1 gene with the exception of the Grado population, while the highest π value was found in the COI gene of the Tarquinia population.
The haplotype networks (Figure 3) show only a few shared haplotypes for the COI and ND1 genes (1 shared haplotype between Karavasta and Vain and three shared haplotypes) and none for the Cytb gene.Accordingly to the COI and Cytb haplotype networks, there are many private haplotypes for each population represented by a few individuals, whereas, for the ND1 gene, the number of identified haplotypes is lower with the presence of more shared haplotypes, namely H8, H12, and H15.The ML trees constructed using COI, Cytb, and ND1 sequences were reported in Figure 4.In line with the network, the COI ML tree shows that all populations have private haplotypes with the exception of Karavasta and Vain populations (Albania), which shared In line with the network, the COI ML tree shows that all populations have private haplotypes with the exception of Karavasta and Vain populations (Albania), which shared the HAP29, while no shared haplotypes were found in the Cytb ML tree.The ND1 ML tree shows that there are three shared haplotypes: HAP8, present in Italian populations of Grado and Tarquinia; HAP12, present in the Italian populations of Grado and Casaraccio; and HAP15, found in the Albanian populations of Vain and Karavasta (Figure 4).

Evidence of Selection Signals in Mitochondrial Genes
No evidence for recombination was found in COI and ND1 datasets with GARD.For the Cytb dataset, GARD showed a recombination break, so the two partitions were analyzed separately.The null hypothesis of strict neutrality was rejected by the Z test in favor of the alternative hypothesis of positive selection in the three analyzed genes.
For the COI gene, eight codons were under purifying selection, and two codons were under positive selection (codons 9 and 111).In particular, the MEME method detected episodic positive selection for codon 9, while FUBAR revealed pervasive positive selection for codon 111.For the Cytb gene, five codons were under purifying selection, and four codons were under positive selection.Codons 318 and 375 were under episodic positive selection with the MEME method, while FUBAR detected pervasive positive selection for codons 9 and 320.For the ND1 gene, eight codons were under purifying selection, and only codon 300 resulted in positive selection with the FUBAR method (Table 3).

Non-Synonymous Changes Analysis
Nucleotide substitutions leading to non-synonymous aminoacidic changes were observed in the Korission population in the COI (223 aa sequence long) and Cytb (375 aa sequence long) genes.For the COI gene, a nucleotide substitution in position 64 from G to A led to the non-synonymous aminoacidic change from alanine (A) to threonine (T) in codon 22.
For Cytb gene: (i) a nucleotide substitution in position 8 from A to C resulted in a non-synonymous aminoacidic change from asparagine (N) to threonine (T) in codon 3; (ii) nucleotide substitution in position 11 from C to A led a non-synonymous aminoacidic Biology 2024, 13, 212 9 of 14 change from proline (P) to histidine (H) in codon 4; (iii) a nucleotide substitution in position 17 from C to T leading a non-synonymous aminoacidic change from proline (P) to leucine (L) in codon 6 and (iv) a nucleotide substitution in position 21 from T to A resulting in an aminoacidic change from phenilalanine (F) to Leucine (L) in codon 7.For the ND1 gene (303 aa sequence long), only synonymous changes were observed (Table 4).Mapping of the non-synonymous aminoacidic changes on the 3-dimensional structure of the obtained models for the examined mitochondrial proteins is shown in Figure 5.

3D Models of COI, Cytb, and ND1 Fragment Proteins
In order to understand the position of the sites under positive selection, three protein models for each mitochondrial subunit were generated (Figure 5).
In the COI barcode region, the non-synonymous mutation Ala22Thr is located in the N terminal tail, where we also found codon 9 under episodic positive selection, while codon 111 under pervasive positive selection is placed in the coil of the protein facing towards the intermembrane space (IMS) (Figure 5).
In the Cytb segment, four non-synonymous aminoacidic mutations (codons: 3, 4, 6, 7) are sited in the N-terminal tail, while the codon change in 111 positions is located in the third transmembrane α-helix.We found one positively selected codon (9) in the N-terminal tail, while codons 318 and 320 under positive selection are in the seventh α-helix and codon 375 in the C-terminal tail (Figure 5).
In the ND1 region, the only codon (300) found under positive selection is located in the C-terminal tail (Figure 5).

Discussion
The sequence analysis of three OXPHOS genes, ND1, Cytb, and COI, was carried out on six populations of the Mediterranean killifish A. fasciatus, sampled in coastal ponds along a North-South thermal gradient: the Italian population of Grado was the northernmost, and the Greek one of Korission was the southernmost.The results indicate that a high haplotype and nucleotide diversity characterizes the COI and Cytb genes and that all three genes show a high percentage of private haplotypes.Very peculiar in this regard seems to be the Greek population of Korission, which in the ML trees does not share any haplotype with the other populations for each gene examined.Overall, these results confirm those of previous studies on the genetic structure of A. fasciatus using various mitochondrial molecular markers [35][36][37][38][39][40][41][42].They can be explained by considering the peculiar biological traits of this species that trigger the divergence among populations.In particular, the absence of larval stages and the low dispersal potential of adults promote high fidelity to the site.In addition, the adaptation to naturally fragmented environments with large variations in temperature (from 4 to 40 • C) and salinity (from freshwater to 40‰ hypersaline water bodies) promotes the selection of population-specific gene pools.For these reasons, A. fasciatus is an ideal species to study microevolutionary processes driven by adaptation to the environment.In this context, although the assumption of neutrality of mitochondrial DNA has long been advocated, it is now demonstrated that mitochondrial genes coding for OXPHOS enzymes are easily subjected to selective pressure in variable environmental contexts since they are responsible for the metabolic performance of organisms.Therefore, several investigations have explored the nucleotide sequences of the OXPHOS mitochondrial genes to detect the presence of non-synonymous mutations leading to amino acid changes, eventually affecting the function of the encoded proteins under environmental constraints [5].In our case, non-synonymous nucleotide substitutions leading to amino acid changes were observed in the Korission population of A. fasciatus in codon 22 of the COI gene and in codons 3,4,6 and 7 of the Cytb gene.All these mutations are located in the N-terminal tail of both proteins, and while the main part of them entailed the substitution of amino acids with the same polarity, only the amino acid substitution at codon 22 of COI protein and that at codon 4 of the Cytb protein resulted in a change of the physiochemical properties of the amino acids.In particular, the non-polar alanine (A) was substituted by the polar uncharged threonine (T) in codon 22 of the COI protein.Alanine and threonine are both amino acids small in size, but while alanine has a side chain non-reactive, threonine has a fairly reactive hydroxyl group.Accordingly, while A is rarely involved in protein function, T is common in protein functional centers [43].Looking at the nucleotide sequence of the COI barcode region examined by us, it is known that its discriminative power for species identification is based on the high variability of the third codon position of amino acids, but taking into account the amino acids deriving from changes in this position, it should be noted that about 72% of them are synonymous and do not lead to amino acid changes, while only 5% of changes in the first position are synonymous, proving that the amino acid sequence of COI is basically conserved as well as its function [44].In our case, the nucleotide triplets encoding alanine (four codons) and threonine (four codons) amino acids differ only in the first position, which is G in alanine and A in threonine.Furthermore, our COI sequences perfectly overlap with those of 323 Actinopterygii analyzed by [44], and based on the degree of conservation of each amino acid position, alanine 22 should be highly conserved contrary to our observations.Looking at Cytb protein, the non-polar proline (P) was substituted by the polar basic histidine (H) in codon 4. The proline side chain is very non-reactive, and it is rarely involved in protein-active or binding sites, while histidine is the most common amino acid in protein-active or binding sites [43].It is noteworthy that the N-terminal tail of the Cytb protein belongs to the matrix domain, which is involved in a few functional activities.However, it has been highlighted that a few amino acid residues close to the N-terminal tail are implicated in the creation of the proton gradient [45,46].The functional importance of amino-acid substitution and its effects on the physicochemical properties of the N-terminal region of the Cytb protein was also suggested by the results of the investigation by [47], which demonstrated that the substitution that introduces a threonine (polar) instead of isoleucine (non-polar) in the position 7 of the human Cytb, led the region surrounding site 7 to become more hydrophilic and more open and free to interact.Accordingly, we hypothesize that the substitution of amino acids we detected could have important functional implications, such as making the N-terminal region of the Cytb protein more hydrophilic.
The second aim of our investigation was to check whether selection acted on the three OXPHOS genes under consideration using recombination and selection tests based on different models of evolution.Notably, the main part of codons was found under purifying or negative selection, as expected for genes encoding proteins involved in important mitochondrial functions as they are OXPHOS enzymes.However, the search for sites that evolve under positive selective pressure yielded interesting results, especially for COI and Cytb proteins.In particular, codon 9 of the COI, again located in the N-terminal tail facing the mitochondrial inner membrane (MIM), was found to be under episodic positive selection detected using MEME, and codon 111, located in the coil of the protein facing towards the intermembrane space, under pervasive positive selection detected using FUBAR.For Cytb, codons 9 and 320 were found under pervasive positive selection, and codons 318 and 375 were under episodic positive selection.Apart from codon 375, located at the C-terminal tail within the matrix, all the remaining codons are located in the inner mitochondrial membrane.Finally, for ND1, codon 300, once again located in the N-terminal region, was found under pervasive positive selection.However, none of the COI and Cytb codons under positive selection were affected by non-synonymous substitutions.
Signatures of positive selection on OXPHOS genes have been found in several teleost species in response to the need to increase aerobic capacity or adapt to changes in environmental temperature.Positively selected sites occurring in the COI gene of Glyptosternoid fishes from the southeastern Tibetan Plateau and other OXPHOS genes were found to drive adaptation to the high-elevation environment.It has also been hypothesized that the interaction between positively selected sites of COX subunits could affect enzyme regulation and OXPHOS efficiency in Istiophoridae, a family of highly aerobic fish [48].A signature of positive or directional selection was found in three sites of Cytb in Sardina pilchardus populations, and a correlation between the most frequent Cytb haplotype and the minimum sea surface temperature was detected [49].Interestingly, the mitochondrial performance of the heart and brain, following thermal acclimation to extreme temperatures, in locally adapted subspecies of the Atlantic killifish Fundulus heteroclitus, highlighted a complex mitochondrial regulation in this deeply eurythermal species, with OXPHOS efficiency and reactive oxygen dynamics probably playing a pivotal role in thermal acclimation and local adaptation [50].In addition, the electron transport system and OHPHOS enzymes have been indicated as promising targets to study the high capacity of mitochondrial regulation in intertidal organisms (mollusks, crustaceans, and fish) facing rapid changes in environmental parameters [13].In our case, it is evident that the N-terminal region of COI and Cytb, which host both non-synonymous mutations leading to amino acid changes and codons under positive selection, could play an important functional role within the OXPHOS system.However, we did not explore the relationship between the sequence mutation and/or the positively selected sites detected and environmental parameters.In this respect, it should be taken into account that each A. fasciatus population may experience the same extreme environmental parameters in the transitional habitats in which they live, characterized by wide daily and seasonal fluctuations in temperature, dissolved oxygen, salinity, and other physicochemical parameters.Furthermore, we cannot overlook the peculiar biological characteristics of this Mediterranean killifish, which, on the one hand, favor the genetic isolation of populations and, on the other, make it highly resistant to the stressful conditions typical of the environments in which it lives.
It is noteworthy, however, that non-synonymous mutations are exclusively found in the Greek population of Korission, which is the southernmost of those we tested.In contrast, the sites subjected to selection remain as such even when the Greek population is removed from the analysis.Consequently, in order to understand which factors may individually or synergistically bring about functional changes in the proteins of the OXPHOS genes, experimental evidence must be sought to test the response of these mitochondrial genes to the variation of environmental parameters.

Figure 1 .
Figure 1.Heatmap of annual mean sea surface temperature and sampling locations (black d the killifish Aphanius fasciatus.

Figure 1 .
Figure 1.Heatmap of annual mean sea surface temperature and sampling locations (black dots) of the killifish Aphanius fasciatus.

Figure 2 .
Figure 2. Plots of haplotype (h) and nucleotide (π) diversity indexes of the killifish Aphanius fasciatus OXPHOS target genes.Error bar indicates standard deviation.

Figure 2 .
Figure 2. Plots of haplotype (h) and nucleotide (π) diversity indexes of the killifish Aphanius fasciatus OXPHOS target genes.Error bar indicates standard deviation.

Biology 2024 , 17 Figure 3 .
Figure 3. Median-joining network for COI, Cytb, and ND1 sequences of the killifish Aphanius fasciatus.The networks mostly illustrate haplotype diversity and relationships since circle sizes are representative of sample size.Crossbars indicate single substitutions.The ML trees constructed using COI, Cytb, and ND1 sequences were reported in Figure 4.

Figure 3 .
Figure 3. Median-joining network for COI, Cytb, and ND1 sequences of the killifish Aphanius fasciatus.The networks mostly illustrate haplotype diversity and relationships since circle sizes are representative of sample size.Crossbars indicate single substitutions.

Figure 4 .
Figure 4. Maximum likelihood trees showing phylogenetic relationships among killifish Aphanius fasciatus sequences based on COI, Cytb, and ND1 genes.Values at nodes indicate bootstrap support.Only values higher than 70% are shown.The topology is rooted in the Spanish toothcarp Apricaphanius iberus.

Figure 4 .
Figure 4. Maximum likelihood trees showing phylogenetic relationships among killifish Aphanius fasciatus sequences based on COI, Cytb, and ND1 genes.Values at nodes indicate bootstrap support.Only values higher than 70% are shown.The topology is rooted in the Spanish toothcarp Apricaphanius iberus.

Table 1 .
Sampling sites of the killifish A. fasciatus populations.Numbers of specimens examined, N and number of private haplotypes (h) detected for each gene.