Ichthyofauna on the move: fish colonization and spread through the São Francisco Interbasin Water Transfer Project

Augusto Luís Bentinho Silva¹ , Giancarlo Arraes Galvão¹, Aline Alves Ferreira da Rocha¹, Silvia Maria Millan Gutierre¹, Geiza Rodrigues dos Santos¹, Bruno Dourado Fernandes da Costa¹, Luiz Cezar Machado Pereira² and Patricia Avello Nicola^2,3

PDF: EN    XML: EN | Supplementary: S1 S2 S3 | Cite this article

Abstract​

---

Introduction​

---

River basin integration projects have been implemented worldwide to complement water supply for human and animal use, irrigation projects, and/or industrial activities (Davies et al.,1992; Das, 2006; Qin et al.,2019). In general, the goal is to redistribute water in order to alleviate the imbalance between supply and demand for water resources, especially in arid and semi-arid regions (Grant et al.,2012; Zhuang, 2016). Regardless of the purpose for which the project was designed, there are two common universal characteristics: high structural and functional complexity, and generate questions about environmental sustainability (Das, 2006).

According to Lévêque et al. (2008), the Brazilian Northeast region is recognized for harboring one of the biggest gaps in terms of ichthyofauna knowledge and also faces strong environmental pressures that can lead to a decline in populations and communities (Collen et al., 2013). The pressures involve, for example, rainfall scarcity, habitat loss, the introduction of non-native species (Langeani et al., 2009), the construction of artificial reservoirs (Rebouças, 1997), and more recently, the construction of the São Francisco Interbasin Water Transfer (SF-IWT) system that transfers part of the waters from the São Francisco River to different receiving basins in the semiarid region.

During inter-basin water transfer projects, one of the major concerns is to continuously assess the environmental impacts, especially the interchange of aquatic organisms between the donor and the receiving basins (Meador, 1992; Rahel, 2007). Species introduction and the integration of watersheds are among the factors that most threaten the conservation of ichthyofauna in the world (Dudgeon et al.,2006; Pittock et al., 2009; Pelicice et al.,2017; Dudgeon, 2019; Albert et al.,2020) and, more specifically, in the Brazilian Northeast semi-arid region (Albuquerque et al., 2012). The ecological consequences of introducing non-native species into a river basin are, among others, the loss of native biodiversity and fishery resources, the spread of pathogens, trophic alterations, and biotic homogenization (Davies et al.,1992; Arismendi et al., 2009; Gozlan et al., 2010; Grant et al.,2012; Vitule, Prodocimo, 2012; Simberloff, Vitule, 2014; Vitule et al., 2019; Geller et al., 2021).

According to Berbel-Filho et al. (2016), the biotic homogenization of the ichthyofauna is one of the probable environmental impacts of the São Francisco River Integration Project and may also pose a danger to rare and/or threatened species in the receiving basins. Brito et al. (2020) discuss that the homogenization of ichthyofauna, sometimes caused by the introduction of non-native species, has been an obstacle to the maintenance of native fish species, especially in semiarid regions. Currently, the São Francisco and Paraíba do Norte River basins share less than 25% of native species among themselves (Silva et al.,2020), and the interchange of non-native cosmopolitan species may cause additional pressure on native and endangered species (Misra et al., 2007; Gallardo, Aldridge, 2018). Therefore, it is essential to conduct ecological studies to assess possible impacts on native fishes in water transfer projects, in addition to measuring homogenization, invasion capacity (Blackburn et al., 2011; Hui et al.,2016; Pereyra, 2016), and environmental invasibility (Ricciardi, Cohen, 2007).

The Study of Environmental Impacts (SEI – Brasil, 2004), a mandatory document presented before the implementation of mega infrastructure projects in Brazil, predicted the mixing of fish communities between the donor (São Francisco) and the receiving (Paraíba do Norte) basins as a consequence of the SF-IWT implementation. Moreover, in the SEI it was reported the possible depletion of native fish populations in receiving hydrographic basins (Andrade et al., 2011). Studies that could fill up the gaps in species diversity, ecology, and niche occupation are important and necessary in the Brazilian Northeast, not only to supplement the ichthyofauna community data but also to assess the possible impacts of the ecological changes during the SF-IWT project. Specifically, long-term studies that evaluate the changes in fish assemblages on the integrated basins, the species interchange, and describe the temporal events occurring in a water transfer system. Those studies are yet scarce, even though they are fundamental tools to improve management, conservation, and impact mitigation on ichthyofauna.

To detect the possible changes in the fish communities caused by the integration of the São Francisco and Paraíba do Norte River basins, this study presents the results of a decade of ichthyofauna monitoring in the SF-IWT East Axis. We present data to supplement the species list of the São Francisco and Paraíba do Norte Rivers, report the fish colonization in the artificial SF-IWT canals, investigate the fish dispersion from the donor to the receiving basin through the canals, and the occurrence of São Francisco basin endemic species in the Paraíba do Norte basin. The status of the species translocated by SF-IWT is analyzed, as well as the risk of introducing other fish species from the Sao Francisco River into the receiving basin. Finally, the feeding and reproduction behavior of two translocated species are described to characterize the species’ invasiveness.

Material and methods

---

Study area. The São Francisco Interbasin Water Transfer (SF-IWT) to the Northeastern Hydrographic Basins Project is the largest water infrastructure project in Brazil. The enterprise is divided into two main axes (North and East Axis) that aim to guarantee the water security of 12 million people in 390 municipalities in the Caatinga biome, covering the states of Pernambuco, Ceará, Rio Grande do Norte, and Paraíba (Brasil, 2004). The operationalization of the system is intended to result in the improvement of the Jaguaribe (Ceará), Apodi-Mossoró (Rio Grande do Norte), Piranhas-Açu (Paraíba and Rio Grande do Norte), Pajeú, Moxotó, Brígida, and Terra Nova (Pernambuco) River basins (Brasil, 2004).

The Caatinga region of Northeastern Brazil, where SF-IWT is located, has semiarid climate, with average temperatures ranging from 25° to 30°C, and reaching higher temperatures during the dry season (Brasil, 2004; Albuquerque et al., 2012). The short rain periods are concentrated from January to May (rainy season). The average rainfall of 600 mm annually in this region is very low compared to 1,900 mm in the Southeastern Brazil, for example. As a result, rivers of the Caatinga biome are, for the most part, intermittent, as they can be completely dry for several months or even years (Brasil, 2004). As geographical reference, we use the freshwater ecoregions proposed by Abell et al. (2008).

The fish specimens were caught from 19 sampling sites between August 2012 and December 2021, during the SF-IWT East Axis Installation (between 2012 and 2017) and Operation (2018-currently; Brasil, 2018) phases, within the limits of the São Francisco (SF) and Paraíba do Norte (PB) basins (Fig. 1; Tab. S1). The monitoring campaigns were conducted every six months (once in the dry and once in the rainy period) in two locations of the Itaparica Reservoir (donor basin – DB – São Francisco River; Fig. S2 A–B) and five locations in the Paraíba do Norte River basin (receiving basin – RB; Fig. S2 G–J). In addition to these, all the 12 artificial reservoirs along the East Axis (EAR) were monitored after their filling (Fig. S2 C–F). The first reservoir in the East Axis (Areias Reservoir – EAR 1; Fig. S2 C) was sampled for the first time in March 2015, while the last reservoir (Barro Branco Reservoir – EAR 12; Fig. S2 F) was sampled for the first time in December 2017.

FIGURE 1| Schematic representation containing the monitored sites along the East Axis of the São Francisco River Integration Project (SF-IWT) and the exact location of the new records of Anchoviella vaillanti in the Poções and Epitácio Pessoa Reservoirs. DB = Donor basin, EAR = East Axis Reservoirs, RB = Receiving basin. Detailed location list of the sampling sites in Tab. S1.

A three-days-sampling effort was conducted on all analyzed sites. For sites located in the donor basin, there was a total of 16 three-days-sampling. On the receiving basin, 12 three-days-sampling were conducted in each of the five sites monitored, five samplings occurred before the water input from the SF-IWT and seven after. For all 12 artificial reservoirs, there were different sampling sizes because the filling date of each reservoir varied, meaning that the SF-IWT waters reached those reservoirs at different dates. The sampling effort employed in the 12 artificial reservoirs ranged from five to 12 three-days-sampling, depending on the filling date. Eight reservoirs had five samplings (EARs 2, 3, 4, 6, 7, 9, 10, 11), three reservoirs had eight samplings (EARs 5, 8, 12), and 12 samplings were conducted at oldest reservoir, Areias (EAR 1).

Capture, processing, and preservation of biological material. Six fishing methods were applied, including five actives: trawl (10 m long, 5 mm mesh), sieve (60 cm diameter, 5 mm mesh), cast nets (mesh sizes of 15 and 30 mm between adjacents knots), hand net (40 mm/side rectangular base, 5 mm mesh), and ichthyoplankton conical net (300 µm mesh); and one passive method: gill nets (10 or 50 m long with mesh sizes of 20, 30, 40, 50, 60, and 80 mm between adjacents knots). For each site, it was established a minimum of active capture attempts for the four active methods, except ichthyoplankton: three times per day, totalizing at least nine times per site (three days per site). Additional attempts were added if different species kept being caught. For ichthyoplankton, sampling occurred once during the daytime (around 8 am) and once at night-time (around 6 pm), for 10 min per period in each site. The net was positioned in areas of higher river flow or dragged by the boat at low speed (in reservoirs), and kept both in the surface water and around 3 m deep for 10 minutes each. Total ichthyoplankton effort was 40 min per site. Gill nets were kept overnight (12–14 h). Photographic records of specimens collected by local fishermen were also taken into consideration (Tab. 1). Collected specimens were euthanized by overexposure to 1 g/mL clove oil (based on MCTI – CONCEA, 2018), fixed in a 10% formaldehyde solution and preserved in 70° GL alcohol. Vouchers were deposited in the Ichthyological Collection of the Museu de Fauna da Caatinga (MFCI), Universidade Federal do Vale do São Francisco (UNIVASF).

Reproductive and dietary analysis of non-native species. Supplementary analyzes were conducted for the SF translocated species Anchoviella vaillanti (Steindachner, 1908) and Moenkhausia costae (Steindachner, 1907), both captured in the receiving basin, to assess the ecological niche. Diet data were analyzed using the Alimentary Index – IAi (Kawakami, Vazzoler, 1980), while reproduction data was evaluated through macroscopic visualization of the gonads and Gonadosomatic Index – GSI (Vazzoler, 1996). Sixty individuals of M. costae were analyzed in RB 1 (15 in the dry season and 45 in the rainy season) and 60 individuals of A. vaillanti in RB 1 (15 in the dry season) and RB 2 (45 in the dry and rainy seasons). There were not enough specimens of A. vaillanti to be analyzed in RB 1 during the rainy seasons. All dissected specimens were used in both reproductive and dietary analyses.

Terminology, taxonomic classification, and conservation status. The species were identified according to Britski et al. (1988) and Ramos et al. (2018), complemented by reviews of some taxonomic groups. Larvae specimens were identified according to Nakatani et al. (2001) and Silva et al. (2010). The nomenclature and systematic classification of species were based on Betancur-R et al. (2017) and Fricke et al. (2022). The definition of endemic species was based on Reis et al. (2003), Rosa et al. (2003), Barbosa et al. (2017), Lima et al. (2017), and Silva et al. (2020).

The geographical distribution data and historical records of the species were obtained from SpeciesLink (https://specieslink.net/), SiBBr (Sistema de Informação sobre a Biodiversidade Brasileira, https://www.sibbr.gov.br), Portal da Biodiversidade (https://portaldabiodiversidade.icmbio.gov.br/), and GBIF (Global Biodiversity Information Facility, https://www.gbif.org). The endangered species were assessed using the Livro Vermelho da Fauna Brasileira Ameaçada de Extinção updated list (MMA, 2022).

Data analysis. A Venn diagram was generated to illustrate the species’ data logical relationships between the three regions (SF, EAR, PB) (https://bioinformatics.psb.ugent.be/webtools/Venn/). The seriated ordination of species based on appearance events consists of a presence/absence matrix, with sampling sites in columns and taxa in rows. This analysis was performed using PAST 3 software (Hammer et al., 2001). To indicate the groups affected by the SF-IWT project, the conservation status, origin, adaptability, habitat usage, and trophic ecology data were considered.

Changes in the fish community across studied regions (DB-SF, RB-PB, and EAR) were evaluated using the Bray-Curtis similarity index, observing matrices of dissimilarity. Abundance data was log-transformed to mitigate high-abundance species bias. Analysis of Similarity (ANOSIM) was used to compare the dissimilarity matrices, evaluating whether there were differences in fish communities between regions for richness and abundance. The ANOSIM analyses the variance and multivariate differences in groups through permutations (Clarke, Gorley, 2006).

Spatial beta diversity was analyzed by partitioning diversity into LCBD (Local Contribution to Beta Diversity) and Species Contribution to Beta Diversity (SCBD) components of species richness difference and species replacement (Legendre, 2014), using Podani’s family indices. This analysis allows us to assess species and sites that contribute to beta diversity (Legendre, Caceres, 2013). Presence-absence data determined the composition of species in the basin, while abundance data provided insights on the degree of occupation of each location. The data was previously Hellinger transformed to limit the relevance of rare species. High LCBD values indicate strong differences in species composition from the mean sites. LCBD is partitioned into components of Replacement and Abundance/Richness Difference. Meanwhile, SCBD indicates important species for overall local diversity.

Richness, abundance, and Shannon diversity were included as variables to find the best explanatory model for fish assemblage. Modeling was conducted via stepwise selection by the Akaike information criterion (AIC) until the minimum adequate model was obtained (Crawley, 2013). The explanatory variables ‘abundance’ and ‘Shannon diversity’ were excluded from analysis due to collinearity issues, leaving ‘richness’ as the only explanatory variable in the final model. Data was analyzed using a Canonical Analysis of Principal coordinates (CAP). The CAP allows the detection of linear relationships on (dis)similarities matrices, highlighting the relative contribution of predictor variables to the fish assemblages (Legendre, Anderson, 1999). The dissimilarity based on the matrix obtained by Bray-Curtis distance was tested using PERMANOVA for explanatory variables, with 999 permutations.

To verify the stabilization trends of non-native species populations in the receiving basin, the G test was used to assess whether there is a significant difference (α = 0.05) between the sampling abundances of the species transposed to stretches of the Paraíba do Norte River.

The main statistical analysis were performed in R software 4.1.2 (R Development Core Team, 2021) using functions available in the package Vegan (Oksanen et al., 2013). For beta diversity analysis, the Adespatial package (Dray et al., 2021) was used. Models were run using the MASS package (Ripley et al., 2013). The plots and map (Fig. 5) came from package ggplot2 (Wickham, 2006).

To compare the diet of the non-native species on PB (Anchoviella vaillanti and Moenkhausia costae) over the seasons (rainy and dry), fish stomach volume was analyzed. Data were log-transformed prior to the analysis. Species diet and season groups were sorted by Non-metrical Multidimensional Scaling analysis (NMDS) using the Bray-Curtis similarity index and compared by the similarity analysis test (ANOSIM). The similarity percentage analysis (SIMPER) was used to evaluate which food item contributed most to the differentiation of the groups. The NMDS, ANOSIM, and SIMPER tests were performed on Primer 6 & Permanova software (Clarke, Gorley, 2006). The average GSI for each gonadal stage was compared using analysis of variance (ANOVA; Tukey’s post hoc), with a significance level of α<0.05.

Results​

---

A total of 89,372 specimens, distributed in 69 species, 24 families, and eight orders were recorded. At the two sites on the São Francisco River, 50 species were recorded, 25 in the artificial reservoirs along the East Axis, and 35 in the Paraíba do Norte River basin (Tab. 1). Characidae was the most representative family, with 16 species, followed by Cichlidae with seven species, and Loricariidae represented by six species. The majority of small-sized fish were captured with trawls, sieves, and cast nets, while medium- and large-sized individuals were caught with gill nets.

TABLE 1 | List and respective occurrences and abundances of species recorded during the SF-IWT Ichthyofauna Monitoring. The Origin column indicates native (N), endemic (E) and non-native species (NN) considering the hydrographic ecoregions of São Francisco (SF) and Northeastern Caatinga and Coastal Drainages (NCCD). SL = standard length (mm).

Species composition and distribution. Among the 50 species recorded in the SF basin, 24% (n = 12) are endemic to this ecoregion, four of those were registered at least in one of the EARs, and one was recorded at two sites in the PB basin (Anchoviella vaillanti) (Tab. 1). Of the 25 species recorded in the East Axis artificial reservoirs, only one was not recorded at the locations on the SF basin: Myleus micans (Lütken, 1875). For the Paraíba do Norte River basin, only two of the 35 captured species are considered endemic to the Northeastern Caatinga and Coastal Drainage ecoregion: Parotocinclus spilosoma (Fowler, 1941) and Apareiodon davisi Fowler, 1941. The only threatened species recorded was A. davisi (Endangered species (EN) according to MMA (2022)). This species was registered at two out of the five sites in the PB basin (RB 3 – Acauã Reservoir and RB 5 – Gurinhém River). The seriated ordination analysis graph (Fig. 2) revealed species with restricted distribution its extremes or species with shared occurrence among the two basins at its center. The only species recorded in all 19 locations sampled was the exotic Oreochromis niloticus (Linnaeus, 1758).

FIGURE 2| Seriated ordination of the species presence/absence at the sampling sites of the East Axis of the São Francisco River Integration Project (SF-IWT). Black cells represent the presence of the species at a particular location. The species written in bold represents the new occurrence record in the receiving basin. In the upper right corner, Venn diagram showing species richness interactions between groups of sites. SF = São Francisco River basin, PB = Paraíba do Norte River basin, EAR = East Axis Reservoirs.

No fish eggs were found in our samplings. Meanwhile, the great majority of larvae(n = 2,045) was from Anchoviella vaillanti. Moreover, there were three larvae of Characidium bimaculatum Fowler, 1941 and three Oreochromis niloticus. The larvae not possible to identify at species level were: Hypostomus sp. (n = 1), Sciaenidae (n = 1), and not identified (n = 15). Most of the larvae were in pre-flexion stage (45%), followed by yolk sac (27%), flexion (15%), and post-flexion (12%).

Species diversity and community patterns. The total beta diversity index for the study was 0.39 out of the maximum possible value of 1 (when all sites contain different species). At DB-SF, beta diversity was 0.34, at RB-PB was 0.36 and at EAR was 0.29. The contribution of individual samples to beta diversity (LCBD), ranged from 0.024 to 0.095, with EAR sites having the lowest values (<0.044), and PB and SF the highest (>0.075). The p-values for the significantly higher beta diversity samples ranged from 0.001 to 0.039. These ecologically unique samples were at SF and PB sites.

The ANOSIM analysis showed that the sites and species within a region (SF, EAR, or PB) are more similar to each other and dissimilar to the sites and species from a different region (r = 0.9773). Likewise, p<0.05 indicated a significant difference between regions (Fig. 3). The effect of richness was highly significant (PERMANOVA marginal significance test, p<0.001) in structuring fish communities across the basins. The first horizontal axis (CAP1, p<0.001), represented by richness difference across regions, accounted for 37% of the model explanation. Sites on the right side of the horizontal axis, represented by EAR, had a minimal influence of richness when compared to sites on the left represented by SF and PB basins. Meanwhile, axis 2 (MDS1) of unconstrained data reflecting the regions’ separation, had SF more closely related to EAR, and PB as an isolated group.

FIGURE 3| Canonical analysis of principal coordinates (CAP) for the species and sites that contributed to the differences between basins and reservoirs (SF = São Francisco River basin, PB = Paraíba do Norte River basin, EAR = East Axis Reservoirs). Fitted site scores are colored and shaped according to the basin or EAR designation. A subset of species that explain at least 70% of the variation among sites is represented by species name abbreviation (two first letters of genus and the two first of the epithet). The only predictor significant for the linear model was richness.

Species that explained at least 70% of the variation among sites were selected to be shown in Fig. 3. Of those species, ten were also pointed by SCBD as indicator species (varied the most): Anchoviella vaillanti, Astyanax lacustris (Lütken, 1875), Bryconops cf. affinis, Hemigrammus marginatus Ellis, 1911, Hyphessobrycon cf. parvellus, Moenkhausia costae, Poecilia hollandi (Henn, 1916), Psalidodon fasciatus (Cuvier, 1819), Serrapinnus heterodon (Eigenmann, 1915),and Serrasalmus brandtii Lütken, 1875 (Fig. S3H).

Non-native species. Eight non-native species (11.6% of the total number of species) were documented (Tab. 1). Three of those species, Hoplosternum littorale (Hancock, 1828), Metynnis lippincottianus (Cope, 1870), and Piaractus mesopotamicus (Holmberg, 1887) (Fig. S3E), were sampled exclusively in the SF basin. The five others were registered in both sampled basins (Tab. 1).

Ten of the 25 species (40%) that inhabit the EAR have no recorded occurrence in the PB basin (Tab. 1) and may become introduced species in this receiving basin: Acestrorhynchus lacustris (Lütken, 1875), Astyanax lacustris, Bryconops cf. affinis, Cichlasoma sanctifranciscense Kullander, 1983, Hemigrammus brevis Ellis, 1911, Myleus micans, Poecilia hollandi, Roeboides xenodon (Reinhardt, 1851), Serrasalmus brandtii,and Triportheus guentheri (Garman, 1890). Two species (Anchoviella vaillanti and Moenkhausia costae) were proven to be translocated from SF to the PB basin.

Non-native species dispersed to the receiving basin. Moenkhausia costae showed variations in abundance in RB 1 with a gradual decrease from its first occurrence, in August 2018, until the last sampling, performed in December 2021 (Fig. 4). Statistical analysis revealed that variations in abundance between the campaigns in which M. costae was captured in RB 1 are still significant (G test = 276.7; df = 7; p <0.001). In contrast to M. costae, the abundance of A. vaillanti has been increasing since the first records (Fig. 4). The species presented the second highest abundance in site RB 2 (n = 124), only after H. marginatus (n = 165). A significant variation was also detected between its abundances in RB 1 (G test = 232.2; df = 7; p <0.001) and RB 2 (G test = 681.9; df = 7; p <0.001).

FIGURE 4| Variation in the abundance of non-native species Anchoviella vaillanti and Moenkhausia costae in the Poções Reservoir (RB 1) in the campaigns conducted after the arrival of the SF-IWT waters.

New distribution record. An endemic species from the donor basin (São Francisco) – Anchoviella vaillanti (Clupeiformes, Engraulidae; Fig. S3 A) – was recorded in the receiving basin for the first time after the beginning of the operation of the East Axis. The graphic variation in A. vaillanti abundance at all sampling sites is represented in Fig. 5. All records are from Brazil, Paraíba State, Paraíba do Norte River basin: MFCI 7633, 5, 41–56 mm SL, Poções Reservoir (RB 1), Monteiro, 07°53’19.67”S 36°59’56.96”W, 28 Aug 2018, A. L. B. Silva. MFCI 8280, 19, 32–52 mm SL, Epitácio Pessoa Reservoir (RB 2), Boqueirão, 07°33’26.3”S 36°16’29.2”W, 14 Mar 2019, A. L. B. Silva & G. R. dos Santos. Another 326 individuals were caught at larval stages (32 in RB 1 and 294 in RB 2) with the ichthyoplankton net.

FIGURE 5| Spatial distribution of Anchoviella vaillanti. The size of the circles represents juveniles/adults’ abundance at each sampling site. The species went from the Sao Francisco donor basin through the SF-IWT East Axis artificial canals and reservoirs, reaching the receiving Paraíba do Norte basin sites, in the states of Pernambuco (PE) and Paraíba (PB), Brazil. DB = Donor basin, EAR = East Axis Reservoirs, RB = Receiving basin. Detailed location list of the sampling sites in Tab. S1.

Feeding and reproductive analysis of dispersed non-native species. The Moenkhausia costae diet was based on six food items (Tab. 2). The most important items were Ostracoda and Zooplankton. Only three of the 60 analyzed stomachs of M. costae were empty. The Anchoviella vaillanti diet was composed of eleven food items. Copepoda and Organic Matter represented the majority of items ingested (Tab. 2). All analyzed A. vaillanti stomachs contained at least one item. Both species presented mainly zooplanktivorous feeding habits in the receiving basin, although A. vaillanti had also an opportunistic insectivorous feedind (due to the high Chironomidae abundance).

TABLE 2 | Alimentary Index (IAi) of the items consumed by non-native species of the Paraíba do Norte River basin (RB), Moenkhausia costae and Anchoviella vaillanti. N = number of non-empty stomachs analyzed; NI = not identified.

The NMDS followed by ANOSIM indicated no differences between species diet (R = 0.44; p>0.05), nor between rainy and dry seasons (R = 0.12; p>0.05). The items Organic matter, Ostracoda, and Zooplankton (NI) accounted for 43% of the season dissimilarity (81% total), while Organic matter, Zooplankton (NI), and Copepoda accounted for 45% of the dissimilarity between species (77% total).

The standard length of Moenkhausia costae ranged from 20 to 56 mm in RB 1, with the lowest values obtained in the dry season (ANOVA; F = 91.25; df = 1; p<0.0001). The smallest breeding female was 38 mm, and the male was 32 mm. In the dry season, M. costae had three gonadal stages: immature (53.3%), maturing (40% of the individuals analyzed), and mature (6.7%). The mean GSI was 0.98 ± 0.74 for males and 1.42 ± 1.26 for females. In the rainy season, three gonadal stages were observed: mature (64.45%), maturing (33.33%), and partially emptied (2.22%). The mean GSI (GSI_m) obtained was 6.34 ± 5.47 for females and 2.28 ± 0.64 for males. In the rainy season, mature females had the highest GSI_m values (ANOVA; F = 70.03; df = 1; p<0.0001). Juvenile specimens were observed only in the dry season (Tab. 3). For Anchoviella vaillanti, the standard length varied between 32 and 56 mm, with the smallest lengths recorded in the rainy season of RB 2 and the largest in the dry season of RB 1. Considering the specimens from both seasons, it was observed that fish in RB 1 are larger than those of RB 2 (ANOVA; F = 49.42; df = 1; p<0.001). The smallest breeding female was 44 mm, and the male was 32 mm. For the A. vaillanti three gonadal stages were observed: maturing (66.67%), mature (30%), and partially emptied (3.33%). The average GSI obtained was 5.05 ± 1.08 for females and 2.76 ± 0.93 for males. When comparing the GSI_m values between the gonadal stages, it was detected that the male indices in stage III were significantly higher than in stage II (ANOVA; F = 14.57; df = 1, p<0.05). In RB 1, 14 of 15 analyzed specimens were maturing females (Tab. 3).

TABLE 3 | Abundance and mean values of the gonadosomatic index (GSI_m) by gonadal stages obtained for Moenkhausia costae (RB 1) and Anchoviella vaillanti (RB 1 and RB 2). RB = Paraíba do Norte River basin; SL = standard length; N = abundance; M = males; F = females; I = indeterminate. Different bold letters (a, b, c) in the columns indicate statistical differences (ANOVA; p<0.05) between gonadal stages of each sample.