Spatial and temporal population genetic analysis of Semaprochilodus insignis (Prochilodontidae), an overexploited fish from the Amazon basin

Field sampling and laboratory protocols

In total, 241 specimens of the Semaprochilodus insignis were sampled from small artisanal fishing boats at 11 localities in the Amazonian basin (Fig. 1, Table 1), from Tabatinga to Santarém (west–east) along the main channel of the Amazon and in its northern and southern tributaries. These localities include black water rivers (Negro River), clear water rivers (Tapajos River) and white water rivers (localities in the mainstream of Solimões-Amazon Rivers and the tributaries Juruá/Purus). Samples of pectoral fins were collected and stored in 95% ethanol. Collections from different years were possible for four sites, Tabatinga (2007–2008), Purus (2006–2007), Santarém and Itaituba (2006–2008), enabling to test possible temporal variations in the genetic structure of S. insignis. We were also able to collect samples from geographically distant migrating aggregations, enabling us to test for divergence of migratory aggregations. All individuals were captured and sampled under license granted by the Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renováveis (IBAMA/SISBIO permit #11325-1). Sampling collection was undertaken in accordance with the ethical recommendations of the Conselho Federal de Biologia (CFBio; Federal Council of Biologists), Resolution 301 (December 8, 2012).

For the spatial analysis, 241 individuals were sequenced for mitochondrial DNA control region and 180 individuals were genotyped for eight microsatellite loci (Table 1). The same individuals that were genotyped were also sequenced for the control region. However, a few sequenced individuals could not be genotyped. A subset of this database, composed of samples collected in the same location in different years, was used to perform the temporal analyses, being composed of 114 individuals for the control region of mitochondrial DNA and 74 individuals for eight microsatellite loci.

From each sample, the total DNA was extracted following a standard proteinase K, phenol–chloroform extraction protocol (Sambrook, Fritsch & Maniatis, 1989). The mtDNA control region was amplified via the polymerase chain reaction (PCR) using the primers F-TTF (5′-GCCTAAGAGCATCGGTCTTGTAA-3′) and 12SR5 (5′-GGCGGATACTTGCATGT-3′) obtained from Sivasundar, Bermingham & Orti (2001) and Hrbek & Farias (2008), respectively. The PCR reactions were carried out in 15 µL total volume and consisted of 1 µL of total genomic DNA (50–100 ng/µL), 2 mM each primer, 1X PCR Buffer (2 mM Tris-KCL, pH 8.5), 1.5 mM MgCl2 (25 mM), 1.5 mM dNTPs, 1U/µL Taq DNA polymerase. The PCR condition consisted of 35 cycles of denaturation at 93 °C for 60 s, annealing at 50 °C for 40 s, extension at 72 °C for 90 s and final extension was performed at 72 °C for 5 min. Subsequent to PCR amplification, the PCR products were purified using exo-sap (Fermentas), and sequence reactions were carried out using the BigDye TM Terminator V3.1 Cycle Sequencing Kit (Life Technologies) following the manufacturer’s instructions. Sequences were resolved using the ABI 3130XL sequencer (Life Technologies) (Hall, 1999) and manually verified in the software BioEdit (Hall, 1999). Alignment was carried out in BioEdit using the software ClustalW (Thompson, Higgins & Gibson, 1996) under default conditions.

Eight microsatellite loci characterized by Passos et al. (2010) were used in this study. Seven of them were amplified in three multiplex PCRs, and the selection of multiplex systems was performed by combining allele sizes so that there was no overlap in the microsatellite ranges (Table S1). The total volume of PCR reaction was of 15 µL for a multiplex system consisting of three microsatellite loci, 14 µL for a multiplex system consisting of two microsatellite loci and 8.5 µL for an individually amplified microsatellite marker. The PCR reaction consisted of 1 µL of genomic DNA (50–100 ng/ µL), 1.5 mM MgCl2 (25 mM), 1.5 mM dNTPs, 1X PCR Buffer (2mM Tris-KCL, pH 8.5), 2 mM for tailed forward primer, 2 mM for reverse primer, 2 mM for primer M13 labeled primer(with FAM fluorescence) and 2.5 U Taq DNA polymerase. Reactions were submitted to the following cycling: hot start at 94 °C for 60 s, followed by 25 cycles of denaturing at 94 °C for 30 s, annealing at 55 °C for 30 s, and extension at 68 °C for 40 s; labeling step consisted of 20 cycles of denaturing at 94 °C for 20 s, annealing at 53 °C for 30 s, and extension at 72 °C for 60 s; final extension at 72 °C for 30 min. The PCR products were visualized in a 2% agarose gel, dyed with GelRed and photographed using a UV 302 nm transilluminator T26M (BioAgency).

PCR products generated with labeled primers were visualized on an ABI-3500 automatic sequencer (Life Technologies). Allele sizes were scored against a ROX size standard (DeWoody et al., 2004) and the results were analyzed in software GeneMapper^® version (Life Technologies).

Data analysis

Genetic diversity

The control region data were composed of 241 nucleotide sequences with a total of 1,145 base pairs (bp) of which 1,031 sites were monomorphic and 114 variables. The sequences are available at GenBank: ON972834–ON973074. In the 241 individuals collected along the main channel and tributaries, 116 haplotypes were found. The lowest number of haplotypes was observed in the locality of Careiro (seven) and the highest in Purus and Santarém (27) (Table S2). The H4 haplotype had the highest frequency (45 individuals) and geographical distribution, being present in all sampled locations in the Amazon basin. Of the 116 haplotypes, 89 haplotypes were singletons.

The results of the genetic diversity parameters show that gene diversity was 1 in all locations (Table 2). The lowest haplotype diversity was observed in Tabatinga and Novo Airão (0.009) and the highest values in Coari, Piagaçu-Purus, Manaus and Careiro (0.012) (Table 2). The results of the haplotype network show extensive haplotype sharing and several unique haplotypes indicating an absence of genetic structure for the sampled locations in the Brazilian Amazon (Fig. 2).

The analysis based on microsatellite loci of the 180 individuals of S. insignis showed 191 alleles in the eight analyzed loci. Matrix of genotypes is available at https://github.com/legalLab/datasets and https://doi.org/10.5281/zenodo.7641680. The number of alleles per locus ranged from 12 (SinC10 locus) to 31 (SinF05 and SinE08 loci) (mean = 23.88 alleles per locus). The average expected heterozygosity(HE) ranged from 0.79 (locus SinB04) to 0.94 (locus SinE08), with an average of 0.88 per locus (Table S3). After Bonferroni correction, no loci deviated significantly from HWE and there was no evidence of linkage disequilibrium. Within locations the average number of alleles ranged from 7.4 in Coari to 15.0 in Jacareacanga. The average values of gene diversity over loci were very similar in all locations with the minimum value in Coari (0.85 ± 0.48) and the maximum value in Eirunepé and Jacareacanga (0.91 ± 0.49). The average allelic richness was also quite homogeneous among locations, ranging from 5.42 in Tefé to 6.09 in Jacareacanga. Observed heterozygosity ranged from 0.74 in Itaituba to 0.85 in Tabatinga, Eirunepé and Tefé, while the expected heterozygosity ranged from 0.87 in Tefé, Coari, Piaguaçu-Purus, Santarém and Itaituba to 0.91 in Jacareacanga (Table 2). The inbreeding coefficient (F_IS) ranged from 0.021 in Tefé to 0.234 in Manaus, showed statistical significance for eight out of 11 sampled locations (Eirunepé, Coari, Manaus, Novo Airão, Careiro, Santarém, Itaituba and Jacareacanga); it also was significant when the analysis was conducted considering a single and large population. Proportion of private alleles ranged from 0.40 in Tefé and Coari to 0.85 in Jacareacanga. Private alleles values above 0.5 were observed in the locations with the greatest geographical distance (Tabatinga, Eirunepé, Itaituba, and Jacareacanga) (Table 2).

Temporal and spatial population structure

For temporal analysis based on control region data, the Φ_ST values of AMOVA results (Table 3) ranged from −0.01246 for comparisons between different years in Piagaçu-Purus (2006 × 2007) to 0.22195 for comparisons between different years in Tabatinga (2006 × 2008) with statistical significance for Tabatinga (p < 0.010). For this locality 22.19% of variance is found between samples from different years, indicatating temporal structure. The values of F_ST of AMOVA for the temporal analysis based in microsatellite data ranged from −0.00183 for Santarém (2006 × 2008) to 0.01429 for Itaituba (2006 × 2008), however, none of the values were significant (Table 3). Likewise, pairwise comparisons for both markers showed low and non significant F_STvalues with the aforementioned mtDNA Tabatinga (2006 × 2008) comparison (Table 3).

The AMOVA based on the spatial analysis revealed that the higher genetic variation was observed within sampled populations (99.65% for control region and 99.19% for microsatellites), and no genetic differentiation was observed among populations (Φ_ST = 0.00352, p = 0.335, and F_ST = 0.00813, p = 0.937, for control region and microsatellites, respectively). For the control region the values of pairwise comparisons of Φ_ST for spatial analysis ranged from −0.008 for Novo Airão and Manaus to 0.148 for Tabatinga and Careiro, however, no value was statistically significant after Bonferroni correction (p < 0.0009) (Table 4). For the microsatellite data the values of pairwise comparisons of F_ST ranged from −0.0003 for Eirunepé and Jacareacanga to 0.037 for Coari and Novo Airão. After Bonferroni correction only the Tabatinga and Careiro (F_ST = 0.032, p < 0.0001) and Tefé and Jacareacanga (F_ST = 0.028, p < 0.0001) comparisons were significant (Table 5). All Nm values were above thousands of effective migrants per generation. When testing population groups as function of different types of water, AMOVA did not show any genetic differentiation (Φ_ST = 0.00513, p = 0.79, and F_ST = 0.00039, p = 1.00, for control region and microsatellite, respectively).

The STRUCTURE v.2.3.4 delimited only one biological population among all individuals sampled from the 11 localities, with the highest posterior probability of −7259.2 (Fig. S1, Table S4). These results were also in agreement with the DAPC analysis which also identified only one group present within S. insignis (Fig. 3).

The Mantel test results suggest no significant correlation between genetic divergence and geographical distance of sampled localities neither for control region (r =  − 0.014751, p = 0.505) nor for microsatellites (r =  − 0.267629, p = 0.904).

Fishing effect and implications for conservation and management

In the central Amazon basin region, the low cost of Semaprochilodus insignis is offset by fishery volume (Faria Jr & Batista, 2019), making commercial fishery of the S. insignis lucrative. The contemporary populations of S. insignis show evidence of bottleneck in all analyzed metrics (Table 6), which means that S. insignis are indeed suffering from overexploitation. Considering that the alleles are lost more quickly than the heterozygosity, the results based on the analyzes of the Garza-Williamson index (allele loss) and the Bottleneck software (heterozygosity) showed that the reduction of the effective population size has been ongoing for some time (Table 6). The same pattern of the effects of intense fishing was observed in the pirarucu (Arapaima gigas), a sedentary species and of great economic value for the Amazon region, and also considered overfished in the rivers of the Amazon basin (Farias et al., 2019). The genetic diversity of S. insignis is large (Table 2), and thus may lead some to interpret this as lack of effect of overfishing, or that “genetic health” of S. insignis has not been affected by overfishing. However, this interpretation is not correct, since genetic diversity per se is not relevant, but rather temporal trends in changes in generic diversity are. While the genetic diversity of S. insignis is large, as in other migratory Amazonian fishes with large geographic distributions, we have no baseline, pre-overfishing estimate of genetic diversity, and thus no information on temporal changes in genetic diversity. Since the Garza-Williamson index is based on allele loss and it is significant, S. insignis has suffered from loss of diversity despite still having relatively high levels of diversity.

Based on the analyses of 1996 to 2000 fisheries data from the central Amazon, Vieira (2003) concluded that all size classes were exploited by the commercial fishery. However, the majority of the catch was concentrated on size classes smaller than 24 cm, a size at which 50% of the individuals reached reproductive maturity. Overfishing and fishing of sub-reproductive individuals consequently has led to selection for reproductive maturity at smaller size. These observations are consistent with studies of Batista (1999) who analyzed fish landing data from 1994 to 1996, and also with Ribeiro (1983) who analyzed fisheries data from the late 1970s. All three studies clearly demonstrate that S. insignis is overfished.

Given that S. insignis is suffering from overexploitation evidenced by (1) decrease of census population size indicated by lower fishing efficiency, (2) its adaptive response of reduction in size at first reproduction (Batista, 1998; Vieira, 2003; Batista et al., 2012), and (3) that it is experiencing a genetic bottleneck, the preservation of the genetic diversity of S. insignis must be a high conservation priority. Loss of genetic diversity (effective population size) lags behind the loss of census size, and thus provides a window of opportunity for implementing conservation and management strategies that are able capitalize on the still available genetic diversity.

Implications for conservation and monitoring

In 2007, IBAMA established a closed fishery period to protect selected overfished species (Ordinance No. 48 of November 5, 2007). The closed period aims to close fishing during the reproductive season, in order to guarantee the reproduction of the selected species and help in the maintenance of fish stocks. However, despite clear evidence of overexplitation detected as early as 40 years ago (Ribeiro, 1983; Batista, 1999; Vieira, 2003), S. insignis fishing regulations have been implemented in a very irregular way, if at all, and independently by each region of the state of Amazonas (http://www.ipaam.am.gov.br/defeso/, accessed on January 3, 2023). Rather, the entire fishery focuses on harvesting reproductive migrations. Considering the results presented here, S. insignis is clearly showing signs of rapid demographic population collapse and loss of genetic diversity. Compounded on the overfishing, loss of genetic diversity and lack of fishing regulations, a study by Ramos (2016) demonstrated that deforestation in the varzea and igapo forests will negatively affect the reproductive success of S. insignis. The varzea and igapo forests are floodplain forests and integral part of the riverine ecosystem, but also are the main area of human settlement and agricultural conversion and unsustainable resource extraction.

The situation of S. insignis appears to be analogous to that of the north Atlantic cod (Gadus morhua). The north Atlantic cod, a historically abundant r-strategist, has been subject to sustained but sustainable fishery since the Viking period (Sodeland et al., 2022). After WWII, as fishing became industrialized, catches peaked despite ever increasing effort to yield ratio, resulting in a partial (60–70%) collapse in the early 1970’s. Despite repeated warnings from the scientific community, the fisheries were unsustainably managed, leading to a sudden collapse in 1992, when the north Atlantic cod population dropped to 1% of its historic levels (Hamilton & Butler, 2001). Thirty years later, the north Atlantic cod has yet to recover.

The scenario of S. insignis appears not to be different. Diverse indicators, whether biological or fisheries, indicate a stressed, overfished and unustainably harvested species on the brink of a demographic collapse. Therefore it is essential that S. insignis is placed on the list of species for which fishing is closed during the reproductive season (November 15 to March 31). This will not only decelerate observed negative population trends, but also provide time during which managers and scientists can devise and implement an effective conservation and management plan.