Genetic diversity and population structure of Brycon nattereri (Characiformes: Bryconidae): a Neotropical fish under threat of extinction

1Laboratório de Genética e Conservação (GECON), Universidade Estadual do Norte Paraná, Rua Portugal, 340, 86300-000 Cornélio Procópio, PR, Brazil. (BAG) bruno@uenp.edu.br, https://orcid.org/0000-0003-0517-1004, (DGF) dhiegouenp@gmail.com, https://orcid. org/0000-0003-4375-6556 (corresponding author) 2Laboratório de Genética e Ecologia Animal (LAGEA), Departamento de Biologia Geral, Universidade Estadual de Londrina, Rod. Celso Garcia Cid, km 380, 86051-970 Londrina, PR, Brazil. (CAS) carolapolinario07@gmail.com https://orcid.org/0000-0001-5810234X, (MCT) marianauenp@gmail.com, https://orcid.org/0000-0002-2907-405X, (NGOA) aprigio.bio@hotmail.com, https://orcid. org/0000-0002-5142-5119, (FSA) fernandasa@uel.br, https://orcid.org/0000-0002-0130-5047, (SHS) shsofiabelh@gmail.com, https:// orcid.org/0000-0002-3443-0696 3Programa de Pós-Graduação em Ecologia de Ambientes Aquáticos Continentais, Universidade Estadual de Maringá. Av. Colombo, 5790, 87020-900 Maringá, PR, Brazil. (RRO) renatarubia.ota@gmail.com, https://orcid.org/0000-0002-5743-2758 4Laboratório de Ictiologia da Amazônia Ocidental (LIAO), Departamento de Engenharia de Pesca, Universidade Federal de Rondônia, Rua da Paz, 4376, Lino Alves Teixeira, 76916-000 Presidente Medici, RO, Brazil. (WMO) ohara@unir.br, https://orcid.org/0000-0003-1684-4198 Introduction


Introduction
It is widely acknowledged that comprehensive studies on the genetic diversity and structure of endangered species are both necessary and urgent if we are to undertake effective environmental conservation and management action (Wu et al., 2015).As one of three levels of biodiversity recommended for the conservation of the planet, genetic diversity is 2 e180071 [2] thought to be essential if populations are to evolve in response to environmental changes (Reed et al., 2002), such as those resulting from anthropogenic disturbances.It is a well-known fact that small and isolated populations are more likely to suffer impaired genetic diversity, mainly due to the effects of genetic drift and inbreeding (Frankham et al., 2010).Anthropogenic disturbances can lead to changes in the environment that can result in the isolation and decline of natural populations and are now a major threat to numerous species throughout the world (Frankham et al., 2010;Allendorf et al., 2012).
In the upper Paraná River basin, a tributary system heavily impacted by anthropogenic activities in the Neotropical region (Agostinho et al., 2005), B. nattereri has recently been reported in the Corumbá (Pavanelli et al., 2007), Tibagi (Shibatta et al., 2007), Grande (Pompeu et al., 2009;Belei, Sampaio, 2012;Casarim et al., 2012) and Ivaí rivers (Viana et al., 2013;Frota et al., 2016).Although some studies report a number of populations of B. nattereri (Belei, Sampaio, 2012;Vitorino Júnior et al., 2014), this species is very rare and has low abundance in tributaries of the upper Paraná River (Casarim et al., 2012;Viana et al., 2013;Frota et al., 2016), including the Paranapanema River subbasin (Shibatta et al., 2007).Oyakawa et al. (2009) drew attention to the fact that, although B. nattereri used to be an important fishing resource in the Brazilian southeast during the last century, natural populations seem to have become extinct in some of these basins.
According to various authors, urgent action to conserve this species should include the identification and conservation of areas with healthy populations in upper stretches of tributaries with riparian vegetation (Mikich, Bérnils, 2004;Lima et al., 2008;Rosa, Lima, 2008).Although B. nattereri is recognized as an endangered species (Lima et al., 2008;Rosa, Lima, 2008), its population genetics is still largely unknown.Population analysis data produced using nuclear and mitochondrial markers are present in the literature, but only for other species of the genus Brycon (Hilsdorf et al., 2002;Barroso et al., 2005;Sanches, Galetti-Jr, 2012;Ashikaga et al., 2015).Thus, in this context, understanding the genetic diversity and population structure of the few populations of B. nattereri known to exist in nature would be of great relevance for its preservation.Since this species is still assigned "vulnerable" status in the Brazilian list of endangered species (ICMBio, 2018), genetic data could support management actions that prevent higher levels of threat of extinction.In view of this, there is hope that management initiatives intended to protect its populations can still be implemented.
In discussing the urgent actions required to conserve B. nattereri and possible threats to its populations, this study examines the diversity and genetic structure of B. nattereri along a recently identified area of occurrence in one of the tributaries of the Paranapanema River, in the upper Paraná Basin.Since this species possibly performs short migratory movements (Vieira et al., 2005;Vitorino Júnior et al., 2014) and its occurrence in the studied river seems restricted to an upper stretch (without apparent obstacles), a single population would be expected to occur in this area, possibly still exhibiting some genetic diversity among individuals.The present study tested and discussed these hypotheses, aiming to provide information that could improve our knowledge of the genetics of this species and help in developing future management and conservation initiatives.

Material and Methods
Study area.Samples were collected along a stretch (approx.80 km) of the upper Laranjinha River (Fig. 1).The source of this river is in the municipality of Ventania, state of Paraná, southern Brazil, and it flows into the Cinzas River, a tributary on the left of the Paranapanema River, upstream of the Paraná River.
Since the Paranapanema River is currently fragmented by eleven hydroelectric dams (Britto et al., 2003), some studies have suggested that its dam-free tributaries, such as the Laranjinha and Cinzas rivers, could have great importance for the maintenance of ichthyofauna that still exists in the basin (Hoffmann et al., 2005;Lopes et al., 2007;Vianna, Nogueira, 2008).Based on this, a previous study, conducted by Galindo (2014), sought to understand the composition and distribution of the ichthyofauna in the Laranjinha River.In that study, seven sampling sites were chosen (B, D-J in Fig. 1) from the source to the mouth of the river (spaced about 50 km), which were sampled seasonally (six times at each site) from October 2010 to April 2012, using gill nets (different mesh-sizes) and other methodologies (cast nets and sieves) in stretches of approximately 500 m, including 48h efforts.In addition to registering more than 100 species along the Laranjinha River, Galindo (2014) also identified an occurrence area not yet reported for B. nattereri.Interestingly, even after several collections, this species was found only at sites B (40 individuals from six collections) and D (five individuals, obtained in a single collection of six performed).
Specimen sampling and marking.In order to understand whether there are one or more populations of B. nattereri in this area as well as their current levels of genetic diversity, the present study carried out new samplings of B. nattereri at three sites with in the area of occurrence, including site B (site where the species was first found), site A (upstream of B and downstream of site F, where the species was not found), and site C (downstream of B and upstream of D, where the occurrence of the species was accidental) (Fig. 1).
Considering that there are no apparent barriers to fish migration within the study area, the locations of sites A and C were chosen aiming to perform samplings at equidistant sites (spaced about 20 km), studying the stretch of river that still exhibits some characteristics commonly required by B. nattereri, such as good quality water and preserved riparian vegetation (Lima et al., 2008;Pompeu et al., 2009).In fact, although this stretch is fenced in for the cultivation of exotic trees, it still has some riparian vegetation (composed of native plants), which looks better than in other parts of the river.In addition, sites A and C were those with the best access conditions to the river, since the relief (Valleys) also presented limitations to sampling in other sections.
New samplings were performed seasonally (four times at each site) at sites A, B, and C, from July 2014 to March 2015, using specific gill nets for B. nattereri (15 to 50 mm between knots), put in place for 24 h and checked every two hours.After obtaining a small sample of pelvic fin, the individuals captured were microchipped to prevent resampling and rapidly released into the environment.All samples were collected under Environmental Authorization Number 23315-1, issued by SISBIO/MMA.Dead specimens were deposited in the fish collection kept by the Museu de Zoologia, Universidade Estadual de Londrina under catalog number: MZUEL 9445-9446.
After the additional samplings, a total of 92 individuals was analyzed in the genetic study, obtained in the new samplings and by Galindo (2014), including: five from "site A" (all from new samplings), 50 from "site B" (40 from previous study and 10 from new samplings), 32 from "site C" (all from new samplings), and five from "D" (all from previous study).
DNA extraction and quantification.Genomic DNA was obtained from pelvic fin clips using phenol/chloroform extractions (Almeida et al., 2001).DNA concentrations were determined using a Qubit fluorometer (Invitrogen) and samples were diluted to 5 ng/µL.
Population structure analyses.Since the null hypothesis would be the presence of a single population along the area of occurrence, the first analyses tested the genetic structuring between the samples, allowing understanding as to whether there are one or more populations.In a first step, microsatellite data were evaluated on a Bayesian cluster analysis in Structure v.2.3.3 (Pritchard et al., 2000), estimating the posterior probabilities for K clusters, assuming HWE conditions within clusters while minimizing linkage disequilibrium.This analysis allows inference of which population an individual belongs to, regardless of their collection site.It includes an "attribution method" which calculates the probability of the different genotypes being observed in each population (without any a priori inference) and assigns the individuals to the populations, according to the possibilities of the genotypes belonging to them (Beaumont, Rannala, 2004).
In Structure v.2.3.3, twenty replicates were run for each K value (number of clusters), which ranged from one to seven (K = 1-7, number of sampling sites plus three), according to Evanno et al. (2005).Each replicate was implemented by running 100,000 Markov Chain Monte Carlo (MCMC) iterations discarded as burn-in, followed by 1,000,000 MCMC iterations.Convergent values of key summary statistics were used to obtain an accurate estimate of K, according to Pritchard et al. (2000).Estimates of posterior sample distribution were obtained using the admixture model and correlated allele frequencies among populations.Log-likelihood values for data conditional on K, lnPr(X/K) (Pritchard et al., 2000), and ΔK ad hoc statistics (Evanno et al., 2005) were used to choose the most likely number of clusters (K) using Structure Harvester v.0.6.7 (Earl, von Holdt, 2012).Results of independent Structure runs were summarized for the best K using the "greedy" algorithm in Clumpp 1.1.2(Jakobsson, Rosenberg, 2007).Graphs representing the membership coefficient of each sampled individual were plotted using Distruct 1.1 (Rosenberg, 2004).
Additionally, absence of genetic differentiation was also tested from the Analysis of Molecular Variance (AMOVA).This analysis was run on both molecular markers in the Arlequin v.3.5.1.3program (Excoffier, Lischer, 2010), estimating the partitioning of genetic variation within and between samples.AMOVA also provided the general Φ ST values (a genetic differentiation index).Estimates of significance were conducted based on 10,000 permutations.

Genetic diversity analyses.
After defining the number of existing populations in the study area, analyses were carried out to estimate the genetic diversity levels in the sample set.The total allele number (A), average alleles per locus (N A ), average effective alleles (N E ), private allele number (N p ), observed heterozygosity (H O ) and expected heterozygosity (H E ) were estimated using the Popgen v.1.31program (Yeh et al., 1999).The Fstat v.2.9.3 program (Goudet, 2001) was used to estimate the rate of inbreeding (F IS ).Testing for deviations from the Hardy-Weinberg Equilibrium (HWE) and linkage disequilibrium was performed using the Genepop v. 1.2 program (Raymond, Rousset, 1995), and alpha values adjusted using sequential Bonferroni corrections (Rice, 1989).Micro-Checker v.2.2.1 program (Van Oosterhout et al., 2004) was employed to evaluate null alleles or genotyping errors, such as allelic dropout and stutter peaks.
Demographic analyses.Since bottleneck tests are important to understand whether current levels of genetic diversity in a population were influenced by some decline in effective population size (Luikart, Cornuet, 1998), these tests were run on both molecular markers.Recent population bottleneck signs were evaluated on microsatellite data using Bottleneck v.1.2.02 program (Piry et al., 1999), considering deviations from the mutation-drift equilibrium.Three tests were used, including two tests to indicate bottlenecks in the presence of significant excess heterozygosity: "Sign test" (Cornuet, Luikart, 1996) and the "Wilcoxon sign-rank test" (Luikart, Cornuet, 1998), both based on the Infinite Alleles Model (IAM), Stepwise Mutation Model (SMM) and Two--Phase Model (TPM -with 90% SMM and 10% IAM), with a P-value < 0.05.The third test was the "Mode-shift test" that indicates bottlenecks resulting from alterations in allele frequency distributions (Luikart et al., 1998).

Results
Population structure.All results corroborate the existence of a single population in the study area.Bayesian clustering analysis (Structure) used on the microsatellite data showed that the most probable K (K = cluster number) was K = 1 from lnPr(X/K) and K = 2 from ΔK (Figs. 2a, b).Since the ∆K cannot find the best K if K = 1 (Evanno et al., 2005), the graphic representation (Fig. 2c) was presented using K = 2 (the lowest possible K from ΔK) aiming to show that although it displays individuals with contrasting ancestry values (some with q < 0.8), there was no clearly defined clustering and ancestry values were evenly distributed over the samples.
Additionally, AMOVA analysis showed that, in terms of both the nuclear DNA and the mtDNA, most of the variation occurred within populations (99.74% and 99.04%, respectively) and a small percentage between samples (0.26% and 0.95%, respectively).In both cases, Φ ST values were not significant (nuclear DNA: Φ ST =0.0026, P-value >0.05, and mtDNA: Φ ST =0.0094, P-value >0.05), corroborating the absence of genetic structuring among the samples.nd no null alleles among the seven loci used.
A 555 bp fragment, including 497 bp from the D-loop mitochondrial region and 58 bp from the tRNA-Pro gene, was amplified and sequenced for the 92 individuals.A total of four polymorphic sites were found, revealing five different haplotypes (MF377398-MF377402).The only singleton was haplotype H5.The haplotype (h) and nucleotide (π) diversity values were 0.616 and 0.0014, respectively.
The haplotype network showed few mutational steps among the haplotypes found.H1 was the most frequent haplotype and was found in all samples (Fig. 3).Genetic diversity.After genetic structure results, all analyzes were conducted considering the existence of a single population.In all, 48 microsatellite alleles were obtained, average 6.857 alleles per locus.The average effective alleles (N E ) was 3.036, expected heterozygosity (H E ) 0.574, and observed (H O ) heterozygosity 0.564.The inbreeding coefficient was not significant, F IS = 0.024, P-value > 0.05.After applying Bonferroni sequential correction, there were no significant deviations (P < 0.05) in the Hardy-Weinberg equilibrium (HWE) of any of the microsatellite loci.This correction was also applied to the linkage disequilibrium (LD) tests, and significant values were found only between loci BoM05 and BoM06.The Micro-Checker program fou-  The signed-rank test did not produce significant values in any of the mutational models (IAM, SSM or TPM) and the mode-shift test showed typical L-shaped distribution (no bottleneck) in allele frequency (Tab.1).The values in the Fu (Fs= -0.194) and Tajima (D= 0.059) tests were not signi-6 e180071[6] ficant (P-values > 0.05).Furthermore, the mismatch distribution graph (Fig. 4) showed unimodal distribution for all haplotype frequencies.
Brycon.This has been shown to be the case for other endangered species, such as B. opalinus, B. insignis and B. orbignyanus (H E = 0.486 to 0.892 and h = 0.405 to 1.000) (Hilsdorf et al., 2002;Barroso et al., 2003Barroso et al., , 2005;;Matsumoto, Hilsdorf, 2009;Ashikaga et al., 2015), and also for species not yet classified as under threat, such as B. hilarii and B. sinuensis (H E = 0.660 to 0.740) (López, 2006;Sanches, Galetti-Jr, 2007, 2012).Thus, although the population of B. nattereri studied herein exhibited a low π (0.0014) and only five haplotypes, estimates of H E and h seem satisfactory.According to Freeland (2005), h is considered the haploid equivalent of H E in data on diploids.Therefore, the similarity among these estimates suggests that the current variations in nuclear and mitochondrial DNA are evenly distributed throughout the population.
Additionally, microsatellite date analysis showed very weak evidence of recent bottlenecks in the population studied (only in the IAM model of Wilcoxon sign-rank test).However, despite the positive current scenario of this population (absence of recent bottlenecks and satisfactory H E and h), some results, mainly from mtDNA, suggest that an ancient bottleneck could have affected its genetic composition in the past (Slatkin, Hudson, 1991;Rogers, Harpending, 1992;Grant, Bowen, 1998).Examination of 497 bp in the D-Loop region, one of most variable regions of mtDNA (Frankham et al., 2010;Santos et al., 2007), revealed only five haplotypes and 0.14% π in over ninety individuals.Patterns containing low π values (< 0.5%) combined with high h values (> 0.5) usually suggest the occurrence of a bottleneck followed by rapid population growth and an accumulation of mutations (Grant, Bowen, 1998).In this scenario, the current levels of genetic diversity of B. nattereri seem to be in a process of recovery, which has been happening faster in microsatellite loci than in mtDNA.
Mutation rates in microsatellite loci are acknowledged to be higher than those in mtDNA (Estoup, Angers, 1998;Santos et al., 2007).Furthermore, in contrast to mtDNA, nuclear loci exhibit biparental inheritance and allow different alleles to be distributed and combined at higher rates (Freeland, 2005;Allendorf et al., 2012).Thus, signs of ancient bottlenecks can be less evident in microsatellite loci, since they tend to recover from variation more rapidly than mitochondrial sequences.At the same time, the recovery of π after a genetic bottleneck is slower than h in mtDNA (McCusker, Bentzen, 2010).Therefore, although h in B. nattereri seems to be recovering, π still seems to show signs of a dramatic drop in population that occurred in the past.Although this study does not reveal significant values for Tajima's D test and Fu's Fs test, the fact that a star-shaped haplotype network was obtained suggests population expansion after a bottleneck (Slatkin, Hudson, 1991;Grant, Bowen, 1998), indicating that all the current haplotypes are closely related (three mutations at most) and derived from a single main haplotype (H1).In addition, the pattern of mismatch distribution is unimodal, which is usually ascribed to recent population expansion processes (Rogers, Harpending, 1992),

Discussion
The genetic information obtained in this study indicates the existence of a single population of B. nattereri in the study area, showing satisfactory levels of genetic diversity and the absence of significant genetic structuring.In the case in point, the absence of natural or man-made barriers, as well as the presence of rapids, both acknowledged as important characteristics for the occurrence of B. nattereri (Lima et al., 2008;Vitorino Júnior et al., 2014), together with the restricted size of the area of occurrence, are possible factors that may account for the presence of a single population.
Despite the lack of information on the genetic diversity and population structure of B. nattereri, the majority of genetic studies conducted on other Brycon species show higher levels of genetic structure, influenced by reproductive behaviors (Sanches, Galetti-Jr, 2007, 2012;Sanches et al., 2012), geographic distance, natural barriers or anthropogenic interference (Hilsdorf et al., 2002;Barroso et al., 2005;Matsumoto, Hilsdorf, 2009;Ashikaga et al., 2015).For B. nattereri, non-genetic studies have revealed that this species exhibits seasonal patterns of reproduction and migratory movements restricted to short river stretches, mainly in source regions (Vieira et al., 2005;Vitorino Júnior et al., 2014).Interestingly, despite the fact that B. nattereri is an endangered species (Lima et al., 2008;Rosa, Lima, 2008), the data in our study, and especially the levels of genetic diversity (A = 48, N A = 6.857 and H E = 0.574), the absence of inbreeding, non-significant sub structuring and the lack of HW disequilibrium, all suggest that the sampled population has been able to maintain satisfactory population size over a limited stretch (approx.80 km) of the river.
The genetic diversity indices derived from both molecular markers indicate that the population of B. nattereri in the Laranjinha River exhibits estimated values (such as H E and h) within the limits of variation reported for the genus 7 e180071 [7] including the accumulation of mutations with minimum loss of lineage (Cassone, Boulding, 2006).
In fact, B. nattereri was not the first fish species to show signs of an ancient bottleneck in the study area.Ferreira et al. (2015) studied the pearl cichlid Geophagus brasiliensis (Quoy, Gaimard, 1824) at numerous points along the Laranjinha River and found a considerable drop in the genetic diversity of a sample obtained from point "B".In addition, as in our study, the ancient reduction in effective population size was less evident in the microsatellite data than in the mtDNA data.In another study, Galindo (2014) examined samples of the armored catfish Hypostomus ancistroides (Ihering, 1911) at numerous points along the same river, and its estimated genetic diversity was also lower at point "B".Despite the similarity of results obtained in the study area, we are still not in a position to discuss the factors that led to an ancient bottleneck.
It is a matter of some urgency to identify and conserve areas with healthy populations of B. nattereri (Mikich, Bérnils, 2004;Lima et al., 2008;Rosa, Lima, 2008).This study provides information on a recently identified area of occurrence of B. nattereri and shows that the population in this location still retains satisfactory levels of genetic diversity.Although further study is required on this population, including the analysis of ecological and reproductive aspects, the data obtained herein constitute essential information for drafting future management and conservation initiatives.In addition to being an endangered species, B. nattereriis very demanding and requires well-conserved areas with good quality water and preserved riparian vegetation (Lima et al., 2008;Pompeu et al., 2009).Therefore, action taken to research, elucidate and improve the environmental quality of the Laranjinha River basin could make a significant contribution to conserving B. nattereri population and its genetic diversity levels.
As the Laranjinha River basin stands at present, the main threats to the conservation of B. nattereri seem to be related to recent or future anthropogenic impacts.Therefore, the main actions for conserving this species should include: a) improving and maintaining water quality.During our sampling operations, activities that routinely influence water quality, such as the use of agrochemicals and discharge of domestic and industrial waste, were observed both upstream and downstream of the area of occurrence; b) preservation and recovery of riparian vegetation.Despite the presence of some riparian vegetation along the sampled stretch, this has been fenced in for the cultivation of exotic trees.Furthermore, the recovery and expansion of riparian vegetation could also help expand the area of occurrence of B. nattereri, as has already been suggested for other species of the genus Brycon (Gomiero, Braga, 2007;Gomiero et al., 2008;Matsumoto, Hilsdorf, 2009;Vitorino Júnior et al., 2014); c) prohibition of hydroelectric dams.Although not present in the study area, a number of projects for the construction of small hydropower stations may pose a threat to the Laranjinha River (Agência Nacional de Energia Elétrica, 2015).
To conclude, the data herein indicate that the population of B. nattereri identified in the Laranjinha River (upper Paraná River) still retains satisfactory levels of genetic diversity and that this genetic diversity is evenly distributed in the area of occurrence.Measures including the aforementioned actions could help manage and conserve this population.At present, additional activities such as repopulation do not seem necessary.However, long-term monitoring of genetic diversity and inbreeding could help conserve this population and provide a basis for future decision-making.

Fig. 1 .
Fig. 1.Brycon nattereri sampling sites along the Laranjinha River (A, B, C and D).Also shown are the sampling sites used in a previous study (D, E, F, G, H, I and J) and the number of individuals of B. nattereri collected at each site (in parentheses).On the South America map, numbers and arrows indicate the Paraná, Tocantins, and São Francisco basins and the asterisks indicate the areas with records of Brycon nattereri(Rosa, Lima, 2008;Viana et al., 2013;Frota et al., 2016).

Fig. 2 .
Fig. 2. Bayesian analysis results (Structure).a. Values of K obtained based on ΔK. b.Values of K obtained based on mean likelihood Ln (K).c.Graphic representation of K = 2.Each column represents a different individual and the colors denote the probable ancestry coefficient of the individual and each genetic cluster.

Fig. 3 .
Fig. 3. Haplotype network based on partial sequencing of the D-loop region (mtDNA) of 92 individuals of Brycon nattereri from the Laranjinha River.Circle sizes are proportional to haplotype frequency.