Introduction
Estuaries have physical and chemical gradients, especially salinity and trace elements, which can be used as natural markers of habitat for aquatic organisms (Albuquerque et al., Reference Albuquerque, Miekeley and Muelbert2010; Avigliano & Volpedo, Reference Avigliano and Volpedo2013; Mohan & Walther, Reference Mohan and Walther2015). The exposure of fish to different water masses through ontogenetic development directly influences the incorporation of trace elements into the otolith (Campana, Reference Campana, Cadrin, Kerr and Mariani2014). Otoliths are complex structures of calcium carbonate precipitated mostly as aragonite, and small amounts of other chemical elements immersed within an organic matrix (Campana et al., Reference Campana, Chouinard, Hanson, Fréchet and Brattey2000). The precipitation of calcium carbonate is an extracellular process regulated by fish physiology, environmental factors (such as salinity and temperature), and water composition. Nevertheless, it can also be dependent on genetics, food and even fishing pressure (Morales-Nin, Reference Morales-Nin2000; Elsdon & Gillanders, Reference Elsdon and Gillanders2003; Gillanders et al., Reference Gillanders, Clarke, Thorrold and Conover2011; Catalán et al., Reference Catalán, Alós, Díaz-Gil, Pérez-Mayol, Basterretxea, Morales-Nin and Palmer2018). The otolith is a chemically inert structure, since there is no resorption of the incorporated elements (Gauldie et al., Reference Gauldie, Thacker, West and Wang1998; Elsdon et al., Reference Elsdon, Wells, Campana, Gillanders, Jones, Limburg, Secor, Thorrold and Walther2008). The precipitation of divalent ions in the otolith is a competitive process, in which these ions substitute Ca2+ (Campana, Reference Campana1999; Brown & Severin, Reference Brown and Severin2009). For this reason, the natural markers in otoliths such as Ba, Mn, Li and Sr are always studied in relation to the concentration of Ca (element/Ca ratio). Specifically, for several euryhaline species the otolith Sr/Ca and Ba/Ca ratios have respectively positive and negative relationships with salinity (Campana, Reference Campana, Cadrin, Kerr and Mariani2014; Avigliano et al., Reference Avigliano, Miller and Volpedo2018); while ratios such as Mn/Ca (Limburg et al., Reference Limburg, Olson, Walther, Dale, Slomp and Hoie2011) or Li/Ca (Bouchard et al., Reference Bouchard, Thorrold and Fortier2015) may be associated with hypoxia, physiological or environmental factors. Then, the otolith has turned out to be a useful tool to track the life history of teleostean fishes in environments with physicochemical gradients (Shrimpton et al., Reference Shrimpton, Warren, Todd, McRae, Glova, Telmer and Clarke2014; Duponchelle et al., Reference Duponchelle, Pouilly, Pécheyran, Hauser, Renno, Panfili, Darnaude, García-Vasquez, Carvajal-Vallejos, García-Dávila, Doria, Bérail, Donard, Sondag, Santos, Nuñez, Point, Labonne and Baras2016; Avigliano et al., Reference Avigliano, Leisen, Romero, Carvalho, Velasco, Vianna, Barra and Volpedo2017a).
The species within the Ariidae family are of high commercial value in the South-western Atlantic (Freire & Pauly, Reference Freire and Pauly2005), Madmango sea catfish Cathorops spixii (Spix & Agassiz, 1829) being one of the most important. This is a benthic species widely distributed in estuaries and continental shelf between latitudes 13°N and 25°S (Dantas et al., Reference Dantas, Barletta, de Assis Almeida Ramos, Lima and da Costa2013; Denadai et al., Reference Denadai, Pombo, Santos, Bessa, Ferreira and Turra2013; Santos et al., Reference Santos, Cattani and Spach2016) and it has been characterized as a K-strategist fish with parental care and oral incubation of eggs and juveniles (Lima et al., Reference Lima, Barletta, Dantas, Possatto, Ramos and Costa2012). The artisanal fishery's mean annual catch of C. spixii in Brazil (1995–2000) was 2000 tons (Freire & Pauly, Reference Freire and Pauly2005), also being by-catch fauna of the shrimp fishery (Graça Lopes et al., Reference Graça Lopes, Tomás, Tutui, Severino Rodrigues and Puzzi2002; Cattani et al., Reference Cattani, Santos, Spach, Budel and Gondim Guanais2011). This species has an important ecological role because it is a frequent prey for marine ichthyophagous species such as the Atlantic cutlassfish Trichiurus lepturus (Linnaeus, 1758) and the kelp gull Larus dominicanus (Lichtenstein, 1823) (Bittar et al., Reference Bittar, Castello and Di Beneditto2008; Miotto et al., Reference Miotto, De Carvalho, Spach and Barbieri2017), and it is a good bioindicator of pollution due to its ability to accumulate trace elements in tissues (Azevedo et al., Reference Azevedo, Fernandez, Farias, Fávaro and Braga2009, Reference Azevedo, Braga, Silva de Assis and Oliveira Ribeiro2013). Currently, valuable information for designing fisheries management plans, such as variability in migration patterns, is still largely missing.
In this context, understanding the variation in chemical composition across estuarine salinity gradients could be useful for reconstructing environmental histories of C. spixii from calcified structure microchemistry. The present study aimed (1) to study the distribution of potential habitat markers (Sr/Ca, Ba/Ca, Mn/Ca and Li/Ca) in water from the Paranaguá Estuarine Complex (Brazil) and (2) to study habitat use patterns of C. spixii through ontogeny by using otolith microchemistry.
Materials and methods
Study area
The study area includes the Paranaguá Estuarine Complex (PEC), an important subtropical ecosystem of the South Atlantic due to its continuous cover of the Atlantic rain forest and its consideration as a Natural Patrimony of Humanity (UNESCO, 1999). It is an extensive estuary with an area of ~612 km2, divided in two main axes, north–south and west–east (Figure 1, Lana et al., Reference Lana, Marone, Lopes and Machado2001).
Fig. 1. Study area map showing the sampling sites of Cathorops spixii (red circles) and water (red and black circles) in the Paranaguá Estuarine Complex (Brazil).
In the present study, we analysed the west–east axis, which is responsible for 82% of the fluvial discharge in the PEC (Rocha et al., Reference Rocha, Sá, Campos, Grassi, Combi and Machado2017). The largest contributors of fresh water are the Cachoeira and Nhundiaquara rivers. In more rainy periods the flow of the estuary is 28 × 106 m3 day−1 (Marone et al., Reference Marone, Machado, Lopes and and Silva2005). Despite the influence of some continental water bodies, the PEC is dominated by tidal waves, characterized as semi-diurnal tides with amplitude between 0.5 and 2 m (mean amplitude = 2 m) (Lana et al., Reference Lana, Marone, Lopes and Machado2001). Due to the influence of tides and the continental contribution, the salinity in the west–east varies between 0 and >31, presenting lower salinities at the surface than at the bottom (Dias et al., Reference Dias, Oliveira, Sanders, Carvalho, Sanders, Machado and Sá2016). The water temperature varies between 20–26°C according to the seasons (Mizerkowski et al., Reference Mizerkowski, Hesse, Ladwig, Machado, Rosa, Araújo and Koch2012).
Sample collection and preparation
In order to characterize the distribution of salinity and chemical markers in the PEC, four bi-monthly sampling campaigns (12.1.2015; 2.4.2016; 4.5.2016; 6.9.2016) were carried out to collect bottom water at eight sampling stations on the west–east axis of the PEC (Figure 1), with the exception of site 1 which was sampled only once (12.1.2015). Each field campaign was carried out in a period of less than 7 h to avoid temporal variations. These sampling stations are arranged between the innermost (site 1) and outermost area (site 8) of the PEC (Figure 1). The selection of the sampling points was based on the reported salinity distribution for PEC (Lana et al., Reference Lana, Marone, Lopes and Machado2001; dos Passos et al., Reference dos Passos, Contente, Abbatepaulo, Spach, Vilar, Joyeux, Cartagena and Fávaro2013; Dias et al., Reference Dias, Oliveira, Sanders, Carvalho, Sanders, Machado and Sá2016) in order to cover the widest possible range of salinity. Bottom water was collected because C. spixii has benthic habits (Dantas et al., Reference Dantas, Barletta, de Assis Almeida Ramos, Lima and da Costa2013). A Van Dorn bottle was used to obtain the bottom water samples for elemental analyses, while the characterization of bottom salinity was made using a Conductivity Temperature and Pressure (CTD) system (depth from 1 m at site 1 to 15 m at sites 7 and 8). Water samples were filtered under vacuum using 0.45 μm cellulose acetate filters, acidified to 0.2% (v/v) (pH < 2) with nitric acid and kept cooled until analysis (APHA, 2017).
On the other hand, fish were collected at different salinities to later relate the otolith edge composition with salinity, and to evaluate the variation in otolith chemical tags through ontogeny. Water salinity was confirmed in situ by using a CTD simultaneously with the capture of the fish. The specimens of C. spixii (N = 37) were collected using trawl nets and fishing rods at salinities of 20 (site 2, 9.28.2016 and 12.8.2016), 25 (site 5, 2.1.2017) and 33 (site 7, 3.1.2017) (Table 1 and Figure 1). Fish were collected at salinities higher than 20 because they are more abundant from that salinity (Barletta et al., Reference Barletta, Amaral, Corrêa, Guebert, Dantas, Lorenzi and Saint-Paul2008).
Table 1. Summary of Cathorops spixii descriptive statistics (mean ± SD and range) for each sampling site
N, sample size; TL, total length; TW, total weight; SD, standard deviation.
All the fish caught were adults (length at first maturity from 11.9–12.6 cm) (Melo & Teixeira, Reference Melo and Teixeira1992) and similar in size (mean ± SD, 18.2 ± 1.0 cm, 17.3 ± 2.3 cm and 19.0 ± 4.1 cm for sites 2, 5 and 7, respectively) between the collection sites. After recording total length (cm), the lapilli otoliths were extracted, washed with distilled water, dried and weighed. Afterwards, they were embedded in crystal epoxy resin and cut using a low speed metallographic saw (Buehler Isomet) to obtain 700 μm-thick sections. Otolith sections were fixed to glass slides using epoxy resin, polished using 10 μm-grit sandpaper and sonicated for 5 min in ultrapure water with a resistivity of 18.2 mOhm cm−1.
Element analysis
Water
Concentrations of 7Li, 43Ca, 55Mn, 88Sr and 138Ba isotopes in the water samples were determined with an Inductively Coupled Plasma Mass Spectrometer (ICP-QMS, Agilent 7500) equipped with a Micro Mist nebulizer (Glass Expansion, Waldbronn, Germany). For the external calibration, a quality control standard (QCS 21 atomic standard, Perkin Elmer® Pure, USA) was used. All concentrations were determined in triplicate with a relative standard deviation below 5%. Quality assurance and quality control were performed using the certified reference materials NIST 1640a for fresh water (trace element in fresh water, National Institute of Standards and Technology, USA) and SLEW-3 for estuarine water (trace element in estuarine water, National Research Council of Canada). Analysis of these reference materials showed acceptable accuracy recoveries ranging from 90.0 to 115%. The detection limits based on three times the standard deviation of the blank signal were 7.28 × 10−8 μmol l−1 for Ba and Li, 8.33 × 10−7 μmol l−1 for Mn, 3.42 × 10−7 μmol l−1 for Sr and 0.4 μmol l−1 for Ca. Elemental concentrations were expressed in μmol l−1 for Ba, Li, Mn and Sr, and mol l−1 for Ca, while element/Ca ratios were expressed in μmol mol−1.
Otolith
Determination of 7Li, 43Ca, 55Mn, 88Sr and 138Ba in otolith core-to-edge transects was performed by Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-ICP-MS) using a 193 nm ArF Excimer laser (Photon Machines Analyte G2, USA) coupled to an ICP-QMS Agilent 7700 (Japan). A repetition rate of 10 Hz and a laser fluence of 5 J cm−2 were used, with a circular aperture of 85 μm at a scan speed of 15 μm s−1. The ICP-QMS was operated at a power of 1600 W using Ar as plasma gas, while Helium (800 ml min−1) was employed as carrier gas in the laser ablation unit. The 238U/232Th (1.2) and 232Th16O/232Th (0.4%) ion ratios were monitored to control the oxide production rates and plasma robustness using the reference material NIST 612 (trace elements in silicate glass).
The standard reference materials NIST612, NIST610 (trace elements in silicate glass) and USGS MACS-3 (trace elements in synthetic calcium carbonate, Jochum et al., Reference Jochum, Scholz, Stoll, Weis, Wilson, Yang, Schwalb, Börner, Jacob and Andreae2012) were measured in triplicate at the beginning and at the end of the analytical sessions. Moreover, the three certified materials were analysed during the analysis batch every 12 otoliths to monitor drift. Calcium (38.3% weight, Yoshinaga et al., Reference Yoshinaga, Nakama, Morita and Edmonds2000) was used as internal standard and NIST612 as calibration standard to convert ion signal intensities to concentration. Moreover, NIST610 and MACS-3 USGS were used as secondary standards (Pearce et al., Reference Pearce, Perkins, Westgate, Gorton, Jackson, Neal and Chenery1997; Jochum et al., Reference Jochum, Weis, Stoll, Kuzmin, Yang, Raczek, Jacob, Stracke, Birbaum, Frick, Günther and Enzweiler2011). Recoveries based on secondary standards showed deviations within 0–12 and 0–20% of GeoREM (http://georem.mpch-mainz.gwdg.de) preferred values for NIST 610 and MACS-3, respectively. Otolith element/Ca ratios were expressed in mmol mol−1.
Data analysis
Linear regression analysis was performed to assess the relationships between water composition (elements and element/Ca ratios) and salinity.
Otolith edge composition (all ratios) was used to assess the ability of trace elements as habitat indicators (Avigliano et al., Reference Avigliano, Carvalho, Miller, Gironde, Tombari, Limburg and Volpedo2019). The otolith edge was defined as the last ablation point of each otolith edge-to-edge transect. Otolith edge element ratios were tested for assumptions of normality and homogeneity of variance using Shapiro–Wilk and Levene tests, respectively. Li/Ca and Sr/Ca met the parametric assumptions of normality and homogeneity (Shapiro–Wilk test: 0.86 < P < 0.82; Levene test: 0.71 < P < 0.21), while Ba/Ca and Mn/Ca met these after Log(x × 1000) transformation (Shapiro–Wilk test: 0.50 < P < 0.28; Levene test: 0.41 < P < 0.20). Because the incorporation of trace elements into the otolith edge can be ontogenetically influenced, the effect of total length, total weight and otolith weight on the elemental ratios was tested using analysis of covariance (ANCOVA) and Pearson correlation test. No significant covariation and correlation between elemental ratios and total length (ANCOVA: 0.02 < F < 1.35, 0.2 < P < 0.9, Pearson: −0.29 < r < −0.11, 0.09 < P < 0.52), total weight (ANCOVA: 0.04 < F < 1.0, 0.3 < P < 0.8, Pearson: −0.29 < r < −0.04, 0.09 < P < 0.81) and otolith weight (ANCOVA: 0.23 < F < 2.2, 0.13 < P < 0.61, Pearson: −0.30 < r < 17, 0.09 < P < 0.53) was observed.
Once the parametric assumptions and the absence of the potential effect of the covariates on the element/Ca ratios was verified, linear regression was used to explore the relationship between the otolith edge composition (all ratios) and salinity. Analysis of variance (ANOVA) with Bonferroni post-hoc t-test was also used to compare the otolith edge ratios between salinities. Otolith edge ratios were plotted using a box plot for each salinity separately (20, 25 and 33).
Migratory pattern analyses were based on Sr/Ca variation through ontogeny. The effectiveness of otolith Sr/Ca as a salinity proxy was first tested with linear regression and ANOVA. The other ratios were not used as salinity proxies because it is known that in ariids, they can be strongly susceptible to other environmental and physiological factors (Avigliano et al., Reference Avigliano, Leisen, Romero, Carvalho, Velasco, Vianna, Barra and Volpedo2017a, Reference Avigliano, Carvalho, Miller, Gironde, Tombari, Limburg and Volpedo2019; Maciel et al., Reference Maciel, Avigliano, Maichak de Carvalhoc, Miller and Viannaa2020). Several authors have estimated Sr/Ca thresholds to infer movements between environments with different salinity using the chemical signature of the otolith edge (Bradbury et al., Reference Bradbury, Campana and Bentzen2008; Panfili et al., Reference Panfili, Darnaude, Lin, Chevalley, Iizuka, Tzeng and Crivelli2012; Avigliano et al., Reference Avigliano, Miller and Volpedo2018). However, estimating thresholds requires laboratory tests or chemical signatures of fish caught in extreme environments (e.g. freshwater or marine environments) (Bradbury et al., Reference Bradbury, Campana and Bentzen2008; Panfili et al., Reference Panfili, Darnaude, Lin, Chevalley, Iizuka, Tzeng and Crivelli2012; Avigliano et al., Reference Avigliano, Miller and Volpedo2018). In addition, salinity varied over relatively short distances in the present study, especially between sites 1 and 4, which could introduce errors in the threshold calculation. For these reasons, no thresholds were estimated in the present study.
To assess if there were significant shifts in element/Ca signatures through ontogeny, Change-point analyses (CPA) were performed with Change-Point Analyzer 2.3 software package using a bootstrapping mode (Avigliano et al., Reference Avigliano, Leisen, Romero, Carvalho, Velasco, Vianna, Barra and Volpedo2017a, Reference Avigliano, Miller and Volpedo2018). This analysis uses a combination of cumulative sum charts and bootstrapping to identify stable signatures and changes through time series. The analyses were performed using 95% confidence levels. Only the Sr/Ca stable signatures met the parametric assumptions of normality and homogeneity (Shapiro–Wilk test: P = 0.82; Levene test: P = 0.12), while Ba/Ca, Li/Ca and Mn/Ca did not meet normality, even after Log(x) transformation (Shapiro–Wilk test: 0.02 < P < 0.04). ANCOVA was performed to test the relationship between fish size (covariable) and the number of stable signatures (variables). In addition, univariate tests (ANOVA for Sr/Ca and Kruskal–Wallis for Ba/Ca, Li/Ca and Mn/Ca) were performed among sampling sites to tests if there were differences in the number of stable signatures between salinities.
Finally, the otolith sections were photographed after chemical analysis immersed in water to observe the ablation line. The position of the annuli (age not validated) in relation to the ablation line was determined on the images using Image-Pro Plus 4.5 software.
Statistical analyses were performed using Infostat software (Di Rienzo et al., Reference Di Rienzo, Casanoves, Balzarini, Gonzalez, Tablada and Robledo2011).
Results
Water
The salinity of the sampling sites ranged from 4.54–31.4. An increase in salinity from the internal to the outermost sites was observed (Figure 2A).
Fig. 2. Salinity and element concentration in water samples from of the Paranaguá Estuarine Complex (Brazil). (A) Boxplot of salinity for each sampling site; (B–F) element concentration in water along the salinity gradient.
The Sr, Ba, Li and Mn (in μmol l−1) varied from 0.011 to 0.083, 0.0002 and 0.0221, 0.007 and 0.025, 0.00240 and 0.013, respectively (Figure 2). The Sr/Ca, Ba/Ca, Li/Ca and Mn/Ca ratios (in μmol mol−1) ranged from 3.49–15.2, 0.028–1.33, 1.38–3.74, and 0.007–0.63, respectively (Figure 3).
Fig. 3. Relationship between the element ratios in water and salinity along the salinity gradient of the Paranaguá Estuarine Complex (Brazil).
Significant and linear relationships were found between all elements and calcium ratios with salinity (except for Li/Ca: r 2 = 0.19, P = 0.28; Figure 3B), these being positive for Sr (r 2 = 0.98, P < 0.0001), Li (r 2 = 0.97, P < 0.0001), Ca (r 2 = 0.93, P < 0.0001) and Sr/Ca (r 2 = 0.82, P = 0.002), and negative for Ba (r 2 = 0.82, P < 0.0001), Mn (r 2 = 0.97, P < 0.0001), Ba/Ca (r 2 = 0.74, P < 0.006) and Mn/Ca (r 2 = 0.90, P = 0.0003) (Figures 2 and 3).
Otolith
Otolith chemistry as a habitat proxy
Considering all datasets (N = 37), a significant positive relationship was found between otolith edge Sr/Ca (r 2 = 0.40; P < 0.0001) and Li/Ca (r 2 = 0.23; P = 0.002) and salinity. No significant relationship was found between otolith edge Ba/Ca (r 2 = 0.02; P = 0.34) and Mn/Ca (r 2 < 0.0001; P = 0.97) and salinity.
The highest otolith edge Sr/Ca (F = 15.2, P < 0.0001) and Li/Ca (F = 5.5, P < 0.008) ratios were found at salinity 33, and the lowest values at salinity 20 (Figure 4). The Ba/Ca ratio was significantly higher at salinity 25 and lower at salinity 31 (F = 4.9, P < 0.001, Figure 4). No significant differences were found between salinities for Mn/Ca (F = 1.7, P = 0.2, Figure 4).
Fig. 4. Box plot of Cathorops spixii otolith edge element/Ca ratios (mmol mol−1, sampling size = 37) for each salinity. Different letters show significant differences between salinities (ANOVA, P < 0.05).
Sr/Ca through life history
Considering all core-to-edge transects, Sr/Ca ratio ranged from 1.89–9.73 mmol mol−1. According to CPA, around 46% (N = 17) showed relatively low stable chemical signatures (<2.7 mmol mol−1, e.g. Figures 5B and 6C–E) at the beginning of the transects, while the rest (54%, N = 20) showed relatively high initial stable values (Sr/Ca >2.7 mmol mol−1, Figures 5A, C–F and 6A, B and F). Most of the fish (95%, N = 35) showed a tendency to increase Sr/Ca in the area within the first annulus, where a peak was observed (Figure 5C). This peak was absent in only two specimens (5%) (Figure 5C).
Fig. 5. Representative otolith Sr/Ca (mmol/mol) profiles of Cathorops spixii from core to edge. Solid horizontal lines illustrate stable signatures identified using change-point analysis. Arrow indicates the annuli positions, the red arrow being the first one. TL, total length (cm). The salinity of the capture site is shown in green.
Fig. 6. Representative otolith microchemical profiles of Cathorops spixii from core to edge. Fish were caught at salinities of 20 (A,B), 25 (C,D) and 33 (E,F). Solid horizontal lines illustrate stable signatures identified using change-point analysis for the Sr/Ca ratio. Arrow indicates the annuli positions, the red arrow being the first one. TL, total length (cm). Ratios were expressed in different units for better visualization (Sr:Ca in mmol/mol, Mn/Ca in mmol/mol × 100, and Ba:Ca and Li/Ca in μmol mol−1). The salinity of the capture site is shown in green.
According to the CPA, no specimen showed a constant stable signature (absence of significant changes) through ontogeny. Fish showed between 3 and 9 stable Sr/Ca signatures throughout the history of life (global mean: 6.1 ± 1.9, Table 2), ranging from 2.50 (Figure 5F) to 6.85 mmol mol−1 (Figure 5B). ANCOVA revealed that the number of stable signatures did not covary with the fish size (F = 2.96, P = 0.1). No significant differences (ANOVA, F = 1.47, P = 0.2) were found in the number of stable signatures between salinities.
Table 2. Number of otolith stable chemical signature or shift (mean ± SD) through ontogeny according to Change Point Analysis
No significant differences (P > 0.05) were found between salinities/sampling sites (ANOVA for Sr/Ca and Kruskal–Wallis for Ba/Ca, Li/Ca and Mn/Ca)
Mn/Ca, Ba/Ca and Li/Ca through life history
The Mn/Ca, Ba/Ca and Li/Ca ratios (in mmol mol−1) of the otolith core (within the area of the first annulus) varied between 0.00006–0.088, 0.00007–0.017 and 0.0012–0.016, respectively. All fish showed Mn/Ca, Ba/Ca and Li/Ca peaks in the primordium or at the edge of the core (Figure 6). In general, the peak otolith core Sr/Ca coincides with Ba/Ca and Mn/Ca peaks, all of them significant jumps according to the CPA. All fish showed the highest Li/Ca ratio at the beginning of life, decreasing after that and subsequently showed a peak, which overlapped with the Ba/Ca and Mn/Ca peaks (Figure 6).
After the first annulus, the otolith Mn/Ca, Ba/Ca and Li/Ca (in mmol mol−1) varied from 0.00033–0.088, 0.0004–0.013 and 0.00043–0.015, respectively.
The Mn/Ca and Ba/Ca ratios showed a global tendency to decrease through ontogeny after the first annulus (Figure 6). For Mn/Ca, only three fish showed other significant peaks after the first mark (Figure 6D), the rest showed a clear decreasing trend (Figure 6A–C, E–F). Within the declining global trend, Ba/Ca showed variations or changes throughout ontogeny, there being no obvious association with the annuli (Figure 6). When Mn/Ca (Figure 6D) or Ba/Ca (Figure 6C) peaks were observed, these were lower than those observed at the core.
After the first annulus, the Li/Ca ratio had a global tendency to decrease (Figure 6A, C, and E) through ontogeny in 84% (N = 31) of the specimens. Nevertheless, according to the CPA, the rest of the samples showed significant fluctuations with no decreasing trend (Figure 6D, B, and F).
All fish showed changes in the stable signatures of Mn/Ca (6.9 ± 1.5), Ba/Ca (7.8 ± 2.0) and Li/Ca (5.0 ± 1.8) throughout ontogeny (Table 2), however, no significant differences (Kruskal–Wallis, 0.25 < H < 2.3, 0.2 < P < 0.9) were found between salinities. ANCOVA revealed that the number of stable signatures of Mn/Ca, Ba/Ca and Li/Ca through life history did not covary with the fish size (0.02 < F < 4.0, 0.09 < P < 0.9).
Discussion
Water
An increase in salinity was observed from the internal sector of the estuary towards the ocean, which was more pronounced between sampling points 1 and 4 (Figure 2A). This spatial variation in salinity was previously reported for PEC (Lana et al., Reference Lana, Marone, Lopes and Machado2001; dos Passos et al., Reference dos Passos, Contente, Abbatepaulo, Spach, Vilar, Joyeux, Cartagena and Fávaro2013; Dias et al., Reference Dias, Oliveira, Sanders, Carvalho, Sanders, Machado and Sá2016). Mizerkowski et al. (Reference Mizerkowski, Hesse, Ladwig, Machado, Rosa, Araújo and Koch2012) have shown that the contribution of inland waters does not produce significant changes in the salinity configuration in the inner sector of the PEC, regardless of the season. On the other hand, the intermediate and external sectors are influenced by the sea (Mizerkowski et al., Reference Mizerkowski, Hesse, Ladwig, Machado, Rosa, Araújo and Koch2012), presenting higher salinities in relation to the internal sector, as observed in the present study (sites 5–8).
Like salinity, water Ca, Li, Sr and Sr/Ca levels were higher near the mouth of the PEC, this being explained by the abundance of these elements in the seawater (Mohan & Walther, Reference Mohan and Walther2015; Walther & Nims, Reference Walther and Nims2015). Strontium is more abundant in the marine environments, mainly related to salts of Sr and it can be found in different forms such as SrCl2, SrSO4 and others (Martin et al., Reference Martin, Allen, Cooper, Johns, Lampitt, Sanders and Teagle2016). Burton & Vigier (Reference Burton and Vigier2011) have shown that Li is an element available to the marine environment through the weathering of rocks. In this sense, some aquatic systems show a positive relationship between the water Li level and salinity (Bouchard et al., Reference Bouchard, Thorrold and Fortier2015). The patterns found in this study are consistent with the available literature. For example, Daros et al. (Reference Daros, Spach and Correia2016) have determined Sr in water (1.5 ± 0.3 and 2.5 ± 0.9 mg l−1) at two extreme sampling sites from the PEC, suggesting a potential relationship with salinity. The same pattern has also been observed in other estuaries for both Li (Burton & Vigier, Reference Burton and Vigier2011) and Sr (Tabouret et al., Reference Tabouret, Bareille, Claverie, Pécheyran, Prouzet and Donard2010; Albuquerque et al., Reference Albuquerque, Miekeley, Muelbert, Walther and Jaureguizar2012; Avigliano & Volpedo, Reference Avigliano and Volpedo2013; Mohan & Walther, Reference Mohan and Walther2015; Walther & Nims, Reference Walther and Nims2015).
In this study, Ba, Mn, Ba/Ca and Mn/Ca decreased with increasing salinity. Barium is related to inputs of inland waters (freshwater sediments) (Li & Chan, Reference Li and Chan1979; Coffey et al., Reference Coffey, Dehairs, Collette, Luther, Church and Jickells1997; Tabouret et al., Reference Tabouret, Bareille, Claverie, Pécheyran, Prouzet and Donard2010), which could explain the relatively high concentrations found in the most internal sites of the PEC. Moreover, Ba in fresh water can be free or complex ions (e.g. BaCO3, BaCl or BaSO4) and its concentration can be associated with run-off. The Ba/Ca pattern found in this study has also been reported in other estuaries such as Chesapeake Bay (Dorval & Jones, Reference Dorval and Jones2005), French West Indies in the Caribbean Sea (Tabouret et al., Reference Tabouret, Bareille, Claverie, Pécheyran, Prouzet and Donard2010) and La Plata River Estuary (Avigliano & Volpedo, Reference Avigliano and Volpedo2013; Avigliano et al., Reference Avigliano, Carvalho, Miller, Gironde, Tombari, Limburg and Volpedo2019). The negative relationship between Mn and salinity was previously reported for the PEC (Rocha et al., Reference Rocha, Sá, Campos, Grassi, Combi and Machado2017). The oxy-redox sensitivity of Mn can cause spatio-temporal variations in estuaries (Walther & Nims, Reference Walther and Nims2015), while precipitation can be favoured in well-oxygenated environments (Du Laing et al., Reference Du Laing, Rinklebe, Vandecasteele, Meers and Tack2009). However, this Mn trend is not a general rule for all estuaries. The opposite pattern has been observed in some estuaries from the Gulf of Mexico (Mohan & Walther, Reference Mohan and Walther2015), while no relationship was found between Mn and salinity in others (Walther & Nims, Reference Walther and Nims2015).
Although both Li and Ca varied positively with salinity, Li/Ca was the only ratio that did not show a significant relationship with salinity. A possible explanation could be the seasonal variation in the water composition of the PEC, which was previously reported by Dos Anjos et al. (Reference Dos Anjos, da Machado and Grassi2012). They reported temporal variations in physicochemical parameters and trace elements (mainly arsenic species), probably due to different biogeochemical and hydrological scenarios. In this study, the relatively high dispersion observed in the water chemical composition (Figures 2 and 3) suggests that there is temporal variation, which is more noticeable in the elements that were in lower concentrations, such as Li, Ba and Mn. Nevertheless, despite the temporal variation in the trace element concentration, the spatial variation was strong enough to show significant relationships with salinity for the rest of the elements (Figures 2 and 3).
On the other hand, in the area near the mouth the PEC shows a vertical salinity gradient in addition to its horizontal salinity-dominated ecocline (Barletta et al., Reference Barletta, Lima, Costa, Dantas, Finkl and Makowski2017). Local changes in environmental conditions such as wind or tides can affect the stratification of the water column, modifying the overall relationships between the parameters studied.
Otolith
Otolith chemistry as a habitat proxy
Otolith edge Sr/Ca and Li/Ca showed a significant relationship with salinity. However, the Li/Ca decreasing trend found in most specimens (Figure 6) suggests that this ratio has a strong ontogenetic rather than environmental explanation. Furthermore, there was no clear trend between water Li/Ca ratio and salinity, while Sr/Ca showed a better linear fit (Figure 3), suggesting that this ratio was the best salinity proxy. A positive relationship between otolith Sr/Ca and salinity was reported for other sympatric species of the same family such as Genidens barbus (Lacépède, 1803) (Avigliano et al., Reference Avigliano, Leisen, Romero, Carvalho, Velasco, Vianna, Barra and Volpedo2017a, Reference Avigliano, Carvalho, Miller, Gironde, Tombari, Limburg and Volpedo2019). Strontium efficiency as a tool to infer life history is relatively low at high salinity (e.g. marine environments), where the relationship between water Sr/Ca and salinity tends to be asymptotic (Brown & Severin, Reference Brown and Severin2009; Avigliano & Volpedo, Reference Avigliano and Volpedo2013). In relation to our data, it was observed for both water (Figure 3A) and otoliths (Figures 3 and 4), at high salinities (e.g. >27). In summary, this natural marker is more efficient in areas with a strong saline gradient (when the curve between Sr/Ca and salinity has a high slope).
The high variability found in the otolith edge signatures (Figure 4) suggests that the last ablation point could not represent exactly salinity at the time of capture, but the average of salinities experienced in the last period of life (perhaps some days). This suggests that the edge chemical signatures obtained for the different collection salinities are not valid to estimate migration thresholds between environments because there could be overlapping values. At this point, the need to estimate reference values in future laboratory tests or field studies is highlighted. Another source of error could be the vertical variation of salinity that exists in the PEC. If catfishes move vertically, they could register salinity that does not correspond to the calculated salinity (bottom salinity). However, because it is a species with benthic habitats, this effect could be minimal. Additionally, we do not know the minimum exposure time necessary to influence the chemical composition of the otolith.
After Sr/Ca ratio, Ba/Ca is the most widely used indicator to study migrations of euryhaline fish (Elsdon & Gillanders, Reference Elsdon and Gillanders2005; Avigliano et al., Reference Avigliano, Leisen, Romero, Carvalho, Velasco, Vianna, Barra and Volpedo2017a). Unlike this study, a negative relationship between the otolith Ba/Ca and salinity has been observed in many migratory species (e.g. Tabouret et al., Reference Tabouret, Bareille, Claverie, Pécheyran, Prouzet and Donard2010; Mohan et al., Reference Mohan, Halden and Rulifson2015). In the PEC, the water Ba/Ca ratio is relatively constant at salinity greater than ~7 (Figure 3C), therefore, fish that move to higher salinities than this are exposed to relatively constant Ba values. Furthermore, due to the relatively low concentrations, Ba could be very sensitive to environmental factors such as run-off and precipitation, which could affect the spatio-temporal distribution in the PEC. In tune with our results, the absence of a relationship between otolith Ba/Ca and salinity was reported in other neotropical migratory fishes such as Odontesthes bonariensis (Cuvier & Valenciennes, 1835) (Avigliano et al., Reference Avigliano, Miller and Volpedo2018) and Prochilodus lineatus (Valenciennes, 1847) (Avigliano et al., Reference Avigliano, Pisonero, Dománico, Sánchez and Volpedo2017b).
Inferring the habitat use
The CPA based on Sr/Ca stable signature (indicator of displacement between different salinities) (Figures 5 and 6) suggested that C. spixii migrates between environments with different salinity. The number of Sr/Ca stable signatures was independent of the fish size, which suggests that there is a high plasticity in the habitat use. In other words, there are smaller fish that would make more movements than other larger ones. Moreover, no significant differences in the number of Sr/Ca stable signatures between collection sites were found (Table 2), suggesting that there is a similar migration pattern.
According to the edge Sr/Ca ratio (Figure 4), the core-to-edge Sr/Ca values (1.89–9.73 mmol mol−1) correspond to a wide range of salinities, which suggests that displacements could occur, at least, between salinities of ≤20 to 33. These displacements could be both horizontal, through the salt wedge, and vertically (least likely). The maximum Sr/Ca values found in this study were comparable with those reported for other ariids that coexist with C. spixii such as G. barbus (range: 0.90–9.83 mmol mol−1, Avigliano et al., Reference Avigliano, Leisen, Romero, Carvalho, Velasco, Vianna, Barra and Volpedo2017a) and G. genidens (range: 2.4–9.0 mmol mol−1, Maciel et al., Reference Maciel, Avigliano, Maichak de Carvalhoc, Miller and Viannaa2020), which make migrations between fresh water or estuary and sea. Particularly, G. barbus use freshwater environments, which could explain the relatively lower minimum Sr/Ca ratio (0.90 mmol mol−1) compared with that found in C. spixii (1.89 mmol mol−1).
According to previous studies, C. spixii adults perform reproductive migrations from high salinity waters towards relatively low salinity waters, using areas of relatively low dissolved oxygen (Dantas et al., Reference Dantas, Barletta, Lima, de Assis Almeida Ramos, da Costa and Saint-Paul2012). After the reproductive event, the juveniles will be transported by their parents to waters of higher salinity (Dantas et al., Reference Dantas, Barletta, Lima, de Assis Almeida Ramos, da Costa and Saint-Paul2012). The highest occurrence of mouthbrooder males eggs, free embryos and young juveniles are typically observed in the inner and middle estuary (Dantas et al., Reference Dantas, Barletta, Lima, de Assis Almeida Ramos, da Costa and Saint-Paul2012). In this sense, the Sr/Ca values at the beginning of the core-to-edge time series supports the hypothesis of the use of environments with relatively low salinity in the early stages (before the Sr/Ca peak).
The Ba/Ca and Mn/Ca peaks could be explained by the use of low salinity environments and low levels of O2 in the early stages. This study showed that the inner estuary is characterized by high Ba/Ca and Mn/Ca values, unlike the external estuary. In this case, high Ba/Ca and Mn/Ca ratios in the water could be reflected in the otolith. Furthermore, it has been reported that low levels of O2 (hypoxia) can be associated with Mn/Ca peaks in otoliths (Limburg et al., Reference Limburg, Olson, Walther, Dale, Slomp and Hoie2011, Reference Limburg, Walther, Lu, Jackman, Mohan, Walther, Nissling, Weber and Schmitt2015). Then, the combination of the Mn/Ca and Ba/Ca peaks support reproduction in the inner estuary, where they experience low salinity (<3) and low O2 (<3.4 g l−1) (Dantas et al., Reference Dantas, Barletta, Lima, de Assis Almeida Ramos, da Costa and Saint-Paul2012).
However, the environment is not the only factor that can influence the incorporation of trace elements. If this were the case, Mn/Ca and Ba/Ca peaks should also be observed throughout the ontogeny, during adult reproductive incursions into areas of low salinity and low O2. Here, only three fish showed a second peak of Mn/Ca, and were not clearly associated with low salinity water (low levels of Sr/Ca, Figure 5B). The maternal effect and prolonged yolk-feeding could also have an influence on the incorporation of trace elements (Kalish, Reference Kalish1990; Liberoff et al., Reference Liberoff, Miller, Riva-Rossi, Hidalgo, Fogel, Pascual and Tierney2014). The Ariidae family is characterized by large eggs, long paternal incubation periods and a prolonged yolk-sac period (Reis, Reference Reis1986; Lima et al., Reference Lima, Barletta, Dantas, Possatto, Ramos and Costa2012). In C. spixii, the beginning of autonomous feeding begins after total consumption of the yolk sac, around the 25th week of life (up to 8.2 cm TL) (Lima et al., Reference Lima, Barletta, Dantas, Possatto, Ramos and Costa2012). In addition, physiological changes could occur in the transition to free-embryo or juvenile life (Lima et al., Reference Lima, Barletta, Dantas, Possatto, Ramos and Costa2012), which could cause changes in the rates of incorporation of Mn/Ca, Ba/Ca and Li/Ca. Similar to the present study, Mn/Ca (Rogers et al., Reference Rogers, Fowler, Steer and Gillanders2019) and Ba/Ca (Tabouret et al., Reference Tabouret, Lord, Bareille, Pécheyran, Monti and Keith2011) peaks were associated with the early stages (primordium) of several species such as Sicydium punctatum Perugia, 1896 and Sillaginodes punctatus (Cuvier, 1829). In C. spixii, the peaks were found on the margin of the primordium; there could be a potential physiological effect caused by metabolic changes (perhaps associated with ontogenetic changes) or a strong maternal/yolk influence which could mask the effect of other variables, such as environmental or physiological. In this sense, we recommend being cautious when using otolith chemistry in the core area and emphasize the need to study in greater depth the factors that influence the incorporation of trace elements into C. spixii otolith. The CPA suggests that there could be a common source of variation for all the sampled sites, because no significant differences in the number of stable signatures were found between catch sites for Ba/Ca, Li/Ca and Mn/Ca (Table 2).
Finally, in the time interval studied, spatial variations have prevailed over temporal ones. Nevertheless, when chemical markers are used, large temporal variations in the hydrodynamics of the estuary could confuse interpretations of fish migrations. Therefore, a better compression of temporal variation (seasonal and inter-annual) is required to make inferences based on otolith chemistry.
Final remarks
This study reports for the first time significant relationships between trace elements in water and salinity in the PEC. These relationships can have important implications for reconstructing life history of several migratory fish species from calcified structures chemistry. Otolith Sr/Ca ratio would appear to be a good salinity proxy for C. spixii, while otolih Ba/Ca, Li/Ca and Mn/Ca would appear to be physiologically influenced. Otolith Sr/Ca time series suggested that C. spixii migrate between environments with different salinity with an important plasticity in the migratory patterns. Finally, it is recommended deepening the research on the spatio-temporal variation in the water chemical composition on both the horizontal and vertical axes, to improve the interpretations based on otolith chemical markers.
Acknowledgements
This study was supported by the CAFP-BA/SPU and UFPR (Fundação Araucária, grant number 292/14), Becar program (Ministerio de Educación, Argentina), Government of Principality of Asturias (grant number IDI/2018/000186), ANPCyT (grant number PICT-2015-1823), UBA (grant number UBACyT-20020150100052BA), CNPQ (B.M.C.'s scholarship, grant number 141267/2015-1) and PDJ – CNPQ (grant number 153090/2019-7).
