Introduction
Coastal lagoons are common ecosystems of global occurrence, occupying ~10% of the world's coastline (Moreno et al., Reference Moreno, Ávila and Losada2010). These ecosystems are transition zones between land and sea, known for high productivities (on average 300 g C m−2 y−1) supported by several factors, such as a high number of primary producers, high input of nutrients and efficient recycling of matter (Mouillot et al., Reference Mouillot, Gaillard, Aliaume, Verlaque, Belsher, Troussellier and Do Chi2005; Vizzini & Mazzola, Reference Vizzini and Mazzola2008). These ecosystems are exposed to wide variations in temperature and salinity (Bintz et al., Reference Bintz, Nixon, Buckley and Granger2003), which in turn may have consequences for the local food web by changing its trophic structure (Bruno et al., Reference Bruno, Barbini, Astarloa and Martos2013; Rakhesh et al., Reference Rakhesh, Madhavirani, Charan-Kumar, Raman, Kalavati, Prabhakara Rao, Rosamma, Ranga Rao, Gupta and Subramanian2015).
Fish are present from lower to top levels of coastal lagoon food webs, and previous studies have shown that fish respond to spatial and/or temporal variations of abiotic factors and the biotic community in a number of different ways, from changes regarding habitat use to ontogeny and feeding habits (Davias et al., Reference Davias, Kornis and Breitburg2013; Mont'Alverne et al., Reference Mont'Alverne, Pereyra and Garcia2016; Sánchez-Hernández et al., Reference Sánchez-Hernández, Elloranta, Finstad and Amundsen2017; Muro-Torres et al., Reference Muro-Torres, Soto-Jimenéz, Green, Quintero and Amezcua2019).
To help improve understanding of how changes in fish assemblages take place, the use of trophic guilds, defined by Root (Reference Root1967) as ‘a group of species that exploit the same resources in a similar way’, has become popular. The benefits of this approach in trophic structure studies are two-fold. First, trophic guilds are considered basic building blocks of a community, meaning that by applying the concept it is possible to group species based on ecological roles that will ultimately represent the different compartments of a community. Second, it allows for easy comparison among studies, since trophic guilds of different communities are very similar even when species compositions differ (Simberloff & Dayan, Reference Simberloff and Dayan1991).
Stable isotope analysis (SIA) can be used in trophic studies to infer resource assimilation by consumers since, as isotopic ratios of carbon (13C/12C) and nitrogen (15N/14N) from producers pass from one trophic level to the next, their heavy isotope content tends to increase due to fractionation during metabolism (Peterson & Fry, Reference Peterson and Fry1987). While δ13C signatures of primary producers can be distinct enough for this element to serve as a tracer for different sources of carbon, δ15N values go through a constant enrichment from the lower to the upper levels in a chain or food web, therefore providing an indication of trophic level (Peterson & Fry, Reference Peterson and Fry1987; Post, Reference Post2002; Bearhop et al., Reference Bearhop, Adams, Waldron, Fuller and Macleod2004).
The isotopic niche, useful as an indicator of δ13C and δ15N dispersal and resource use (Newsome et al., Reference Newsome, Rio, Bearhop and Phillips2007; Jackson et al., Reference Jackson, Inger, Parnell and Bearhop2011), will change as a response to variations in environmental factors, trophic structure and resources exploitation over time and space. For example, Fry (Reference Fry2002) and Harrod et al. (Reference Harrod, Grey, McCarthy and Morrissey2005) suggested a positive correlation between salinity and δ13C values of fish, while Davias et al. (Reference Davias, Kornis and Breitburg2013) observed the influence of environmental factors over the δ13C and δ15N signatures of three fish species in Chesapeake Bay, USA.
On the Brazilian coastline, lagoons are more frequent in the south-eastern and southern regions. Most of them are choked lagoons, i.e. connected to the sea by a single, narrow and shallow tidal channel, with restricted water exchange (Kjerfve, Reference Kjerfve and Wolfe1986). The Araruama Lagoon (AL) occupies an area of 220 km2, which makes it the largest permanent hypersaline lagoon in South America. It is surrounded by five towns with a permanent population of ~470,000 (IBGE, 2018), with the population increasing five-fold during the summer holidays (PROLAGOS, 2016). Domestic waste discharged into the lagoon, although treated by five sewage plants (two of them tertiary treatment), contains large amounts of nitrogen and phosphorus (Braga et al., Reference Braga, Vianna and Kjerfve2003), and blooms of macroalgae and cyanobacteria have been recorded (Clementino et al., Reference Clementino, Vieira, Cardoso, Nascimento, Silveira, Riva, Gonzalez, Paranhos, Albano, Ventosa and Martins2008). Despite these increased environmental troubles for the AL, few comprehensive studies are available. As a first attempt, using stable isotopes, to understand how the trophic fish guilds interact with each other and how salinity variation influences the fish population of the AL, the present study will provide valuable information for future attempts of ecosystem management.
Fish trophic guilds, with focus on ontogeny of the most abundant species and isotopic niche overlap were investigated using stable isotopes in order to evaluate the spatial changes of the fish trophic guilds from the AL. Specifically, this study aims to answer the following questions: (1) How do possible variations in fish assemblages, size of the focal species and environmental factors influence the isotopic niche of the trophic guilds? (2) Do trophic guilds show spatial changes regarding their isotopic niche? (3) Do these trophic guilds exhibit isotopic niche overlap and, if they do, what is the degree of the overlap?
Materials and methods
Study site and sampling
Sampling was conducted in the AL, located on the south-east coast of Brazil (22o49′ and 22o57′S 42oand 42o23′W). The lagoon has an elliptic shape that could be divided into seven inlets limited by internal sand spits. The maximum width and length are 14 and 33 km, respectively. The lagoon has a microtidal regime and the mean depth is 2.5 m. The local climate is semi-arid with low rainfall volumes during the year (900 mm annual mean) and high evaporation rates (1400 mm) promoted by frequent and intense north-east winds (Barbieri, Reference Barbieri1975). The only connection to the sea is by the 5.5 km long Itajuru Channel, located to the east of the lagoon, which associated with the climate, balance of rainfall/evaporation, low input of continental discharges (the local watershed has only two permanent low-volume discharge rivers), and long renewal time of its water (84 days) are responsible for the hypersaline (historical mean salinity of 52) regime waters (Kjerfve et al., Reference Kjerfve, Schettini, Knoppers, Lessa and Ferreira1996; Moreira-Turcq, Reference Moreira-Turcq2000; Souza et al., Reference Souza, Kjerfve, Knoppers, Landim and Damasceno2003). The mean temperatures of the surface and bottom layer of the water column are 28.4°C and 24.4°C, respectively (André et al., Reference André, Oliveira, Okuda, Horta, Soldan, Moreira, Rollemberg and Heinzen1981).
Fish were collected at sampling stations positioned in two sectors of the lagoon (Figure 1), following the model proposed by Slack-Smith et al. (Reference Slack-Smith, Faria, Jablonski and Rodrigues1977) that identified environmentally distinct sectors in the lagoon based on depth, sedimentary features and salinity. Sampling stations were located in the outer (site 1) and inner zone (site 2). In order to obtain the best possible representation regarding length variability of the species, fish were sampled using a set of gill nets (15, 30 and 45 mm between opposing knots) in February, May, July and October 2011. The nets were placed at eight sampling stations and retrieved after 24 h. The fish sampled were frozen and taken to the laboratory for identification, measurement of total length (the distance between the tip of the snout to the tip of the longer lobe of the caudal fin) and subsequent muscle extraction.
Fig. 1. Map showing the location of Araruama Lagoon in Brazil (above on the left), its location in Rio de Janeiro State (below on the left) and the study area showing sampling stations at site 1 (outer zone) at site 2 (inner zone).
At the same sites where fish sampling occurred, dissolved oxygen (mg l−1), temperature (°C), salinity and pH of the water were measured using a multi-parameter gauge (HANNA model HI9828). Due to a low catch during summer, a temporal analysis was not possible and only a spatial data analysis was performed. Fish species were grouped in trophic guilds following previous work developed in the same area (Saad, Reference Saad2003, Almeida-Silva et al., Reference Almeida-Silva, Tubino, Zambrano, Hunder, Garritano and Monteiro-Neto2015, Cruz et al., Reference Cruz, Santos and Santos2018). For logistical reasons, it was not possible to collect other components of the food web.
Stable isotope analysis
Isotopic composition (δ13C and δ15N) was determined from a fragment of a dorsal muscle of each fish specimen. For each species, a subsample comprised of several specimens were chosen that covered the size variation observed to be representative for that species. Samples were freeze-dried, homogenized and ~0.5 mg was weighed and analysed with a continuous-flow mass spectrometer (Thermo Finningan Delta V Plus) coupled to an elemental analyser and a Thermo Conflo III interface. Results are expressed as delta (δ), in parts per thousand (‰), relative to Pee Dee Belemnite for δ13C and atmospheric N2 for δ15N, according to equation (1):
$${\rm \delta } = \left({\displaystyle{{R_{{\rm sample}}} \over {R_{{\rm standard}}}}- 1} \right)\times 10^3$$where R sample and R standard are the corresponding ratios of rare to common isotopes (13C/12C and 15N/14N) in the samples and international standards, respectively (Peterson & Fry, Reference Peterson and Fry1987). The analytical precision was ± 0.3‰ for δ15N and 0.2‰ for δ13C.
Lipids are depleted in 13C and high levels of this compound in fish samples, evidenced by a C:N ratio ≥3.5, may compromise δ13C results and interpretation (Kilujen et al., Reference Kilujen, Grey, Sinisalo, Harrod, Immonen and Jones2006). However, elemental composition of the samples analysed indicated a low C:N ratio (≤3.5) and therefore a lipid extraction was not necessary.
Statistical analyses
Principal component analysis (PCA) was applied on the environmental data matrix in order to identify spatial patterns of the water variables measured.
Differences in the isotopic values (δ13C and δ15N) among trophic guilds were tested separately with non-parametric analysis of variance (Kruskal–Wallis), followed by a posteriori Dunn's test. For each abundant species (N > 30) with at least 10 specimens sampled on each site, a linear regression between δ15N and δ13C values was performed to check for a possible relationship between the variables. In addition, a linear regression between δ15N and total length was also performed to check for the possibility of changes in trophic level during the life cycle of the species, suggesting ontogeny. Differences in δ13C and δ15N values of the same trophic guild between sites were tested and a non-parametric test was performed when the homoscedasticity of variances was not observed.
To investigate the trophic structure of the fish assemblage in both sites, six community-wide metrics were applied and are described by Layman et al. (Reference Layman, Arrington, Montaña and Post2007) as follows: (1) δ15N Range (NR): Distance (i.e. maximum δ15N − minimum δ15N) between the two species with the most enriched and the most depleted δ15N values. NR represents the vertical structure (trophic position) within a food web. (2) δ13C Range (CR): Distance (i.e. maximum δ13C − minimum δ13C) between the two species with the most enriched and the most depleted δ13C values. CR values would increase when there are multiple basal resources with varying δ13C values. (3) Total Area (TA): Convex hull area encompassed by all species in a δ13C – δ15N bi-plot space representing a measure of the total niche space occupied and an indicator for the total extent of trophic diversity in a food web. (4) Mean Distance to Centroid (CD): Average Euclidean distance of each species to the δ13C – δ15N centroid, described as the mean δ13C and δ15N values for all species in the food web. It provides a measure of the average degree of trophic diversity within a food web and it may be more reliable than TA when there are outlier species differentially affecting the latter. (5) Mean Nearest Neighbour Distance (MNND): Mean of the Euclidean distances to each species' nearest neighbour in a δ13C – δ15N bi-plot space. Food webs with proportionally more species characterized by similar trophic modes will exhibit an increased trophic redundancy and a smaller MNND. (6) Standard Deviation of Nearest Neighbour Distance (SDNND): A measure of the evenness of species packing in a δ13C – δ15N bi-plot space. Low SDNND values indicate a more even distribution of trophic niches. Differences in the CD and MNND values between communities at sites 1 and 2 were tested separately with a T-test.
The corrected standard ellipse area (SEAc) was calculated to represent the isotopic niche width of the community of each site and of the different trophic guilds. The SEAc is insensitive to bias associated to small sample size and the ellipses were created with a 40% probability of containing data samples obtained posteriorly (Jackson et al., Reference Jackson, Inger, Parnell and Bearhop2011). An ellipses' overlap analysis via Bayesian model based on maximum likelihood estimation was also performed among trophic guilds for each site. All community metrics, SEAc and overlap percentages were calculated with the SIBER package (Stable Isotope Bayesian Ellipses tool in R, Jackson et al., Reference Jackson, Inger, Parnell and Bearhop2011) available in the software R (R Core Team, 2017).
Results were considered significant at P < 0.05. All statistical analyses were conducted using Statistica 8.0, Graphpad 3.1 and the supplement ActionStat for Microsoft Excel.
Results
Abiotic factors
Mean values of dissolved oxygen and temperature did not vary greatly between sites (Table 1), and a significant difference was not observed (t-test: t = 0.58, df = 30, P > 0.05 and t = −1.20, df = 30, P > 0.05 for DO and temperature, respectively). Mean salinity was above 40 at both sites and significantly higher at site 2 (Mann–Whitney U test, U = 186.5, Z = −4.37, P < 0.001). Mean water pH was very similar between the two sampling areas. Still, a low within-site variation contributed to a significantly higher value at site 1 (Mann–Whitney U test, U = 354.0, Z = 2.12, P < 0.05).
Table 1. Mean values and standard deviation of abiotic factors for both areas sampled
Superscript letters show significant differences among sites (P < 0.05). Data were pooled from all sample periods.
Eigenvalues greater than 1.0 were observed for two principal components (PC-1 and PC-2) that together explained 68.9% of the data variability (Table 2). Salinity was positively correlated to the first axis, whereas water pH was negatively correlated. PC1 showed a significant difference between the two areas (t = −2.26, df = 30, P < 0.05). Dissolved oxygen and temperature were positively correlated to PC2, but its scores did not differ significantly between areas (t = 0.39, df = 30, P > 0.05).
Table 2. Results of the principal component axis (PCA) with correlations for each abiotic factor, eigenvalues (λ) and percentage of explanation for each axis (correlation values above 0.4 are in bold according to Hair et al., Reference Hair, Anderson, Tatham and Grablowski1984).
Fish species composition, δ13C × δ15N and TL × δ15N of the focal species
Overall, 440 muscle samples were analysed, representing 17 fish species that were grouped into five trophic guilds, as follows: three planktivorous, three invertivorous, one omnivorous, five piscivorous and five detritivorous (Table 3). The highest δ13C mean value was observed for an omnivorous species, Diapterus rhombeus (Cuvier, 1829), whereas the lowest carbon signatures were found for the three fish species belonging to the planktivorous guild, as follows: Brevoortia pictinata (Jenyns, 1842), Opisthonema oglinum (Lesueur, 1818) and Sardinella janeiro (Eigenmann, 1894). The species O. oglinum together with the invertivorous Achirus lineatus (Linnaeus, 1758) also showed the lowest δ15N value, while the highest δ15N signature was observed for the piscivorous fish Menticirrhus americanus (Linnaeus, 1758) (Table 3).
Table 3. Guild and species from a hypersaline lagoon in south-east Brazil, including number of specimens caught, mean, standard deviation and range of total length (mm), δ13C (‰) and δ15N (‰)
A δ13C × δ15N bi-plot and a possible correlation between total length and δ15N was assessed for each abundant species (N > 30) with at least 10 specimens sampled on each site (Figure 2). With the exception of B. pectinata, highest δ13C values observed for the species were found at site 1, as well as the lowest δ15N for all species. Specimens of B. pictinata and Micropogonias furnieri (Desmarest, 1823) showed a significantly higher TL and δ15N at site 2 (TL = 218.5 ± 11.0 mm, δ15N = 12.7 ± 0.4 and TL = 198.7 ± 68.1 mm, δ15N = 13.5 ± 0.6, respectively), whereas mean 13C signatures observed for E. argenteus, M. furnieri and E. gula were significantly more enriched at site 1 (δ13C = −11.5 ± 1.5, δ13C = −12.7 ± 0.8, δ13C = −12.0 ± 1.3, respectively) (Table 4). A positive significant correlation between total length and δ15N was observed for B. pictinata, Eucinostomus argenteus Baird & Girard, 1855 and M. furnieri (Figure 2).
Fig. 2. Stable isotope bi-plot (δ13C and δ15N) and correlation between total length (mm) and δ15N for B. pictinata, O. oglinum, M. furnieri, E. argenteus and E. gula. P < 0.05 represents significant correlation (black circles – site 1, grey circles – site 2).
Table 4. Main species from a hypersaline lagoon in south-east Brazil, including number of specimens caught, mean and standard deviation of total length (mm), δ13C (‰) and δ15N (‰) for each site together with ANOVA results. Lower-case letters show significant difference of a species between two sites
Trophic guilds and niche overlap
Detritivorous fish were the most abundant at both sampling sites (Table 5), and represented about 60% (N = 236) of the total individuals caught. Samples of omnivorous fish (both sites), and piscivorous fish (site 1), comprised less than 10 individuals. For both sampling sites, piscivorous fishes showed the highest δ15N (δ15N = 13.5 ± 1.1, δ15N = 14.4 ± 1.4 for site 1 and 2, respectively), which differed significantly from the remaining trophic guilds at site 2 (Kruskal–Wallis, MS = 0.93, df = 128, P < 0.001). Apart from the omnivorous group, the lowest mean δ15N values of both sites were observed for planktivores, followed by the invertivorous guild (Table 5). Mean carbon values of the planktivorous guild were significantly lower than the remaining groups at site 2 (δ13C = −16.1 ± 1.9, Kruskal–Wallis, MS = 3.23, df = 128, P < 0.001). At site 1, the carbon signature of the planktivorous was only similar to the piscivorous fishes, probably due to a high standard deviation observed for both trophic groups (Table 5). Significant intra-guild differences in δ13C and δ15N values between sites were observed for the invertivorous guild (t-test, t = 6.78, df = 101, P < 0.0001 and t-test, t = −2.70, df = 101, P < 0.01, respectively) and between δ13C values for the detritivorous guild (t-test, t = 7.21, df = 234, P < 0.0001).
Table 5. Trophic guilds from a hypersaline lagoon in south-east Brazil, including site of occurrence, number of specimens, and mean and standard deviation of δ13C (‰) and δ15N (‰)
Superscript letters show a significant difference among guilds in each site. § indicates significant difference in δ13C (‰) and δ15N (‰) for the same guild between sites.
The spatial difference analysis of the trophic niche (Figure 3) showed that the detritivores showed the highest SEAc value (9.7‰2) at site 1, followed by the piscivorous (6.7‰2) and the planktivorous guild (6.0‰2). At site 2, the invertivores showed the largest trophic niche (7.7‰2), followed by the piscivorous (6.8‰2) and detritivorous guilds (3.7‰2). While the smallest SEAc at site 1 was observed for the omnivorous guild, it was not possible to calculate its trophic niche at site 2 due to a low number of specimens (N < 3).
Fig. 3. Stable isotope bi-plots representing the isotopic niche areas of the trophic guilds at sites 1 and 2. Values in ‰2 indicate the corrected standard ellipse areas (SEAc).
The isotopic niche overlap measures for site 1 (Table 6) showed a complete segregation between the omnivorous guild and all others, except the detritivorous guild. When compared with site 1, the detritivorous guild presented a smaller SEAc at site 2, but also a higher percentage overlap with the piscivorous (58.3%) ellipses. The highest overlap at site 2 was observed between the invertivorous and planktivorous guilds (59.0%), but only 24.5% of the ellipse of the latter overlapped with that of the former. As observed at site 1, the ellipses of the planktivorous and detritivorous guilds presented a small overlap at site 2 (3.4%). A complete segregation at site 2 was only observed between the piscivorous and the planktivorous guilds.
Table 6. Overlapping SEAc (%) between trophic guilds in both sites (the values indicate the percentage the ellipses from the guilds in the columns overlap with the ones from the guild in the row)
A SIBER analysis considering all the trophic guilds of each site as one community was performed, and the results showed higher values of SEAc, total area (TA), δ13C range (CR), and MNND at site 1 (Figure 4 and Table 7). In theory, these results indicated that the community at site 1 had a larger feeding plasticity (TA), associated with a higher variability of food sources (δ13C range). On the other hand, a wider δ15N range (NR) observed at site 2 suggested a higher range of trophic level. Significantly smaller MNND values were found at site 2 (t = −6.92, df = 439, P < 0.0001) compared with site 1, indicating a higher trophic redundancy. A significant difference in CD values between sites was not observed (t = 0.19, df = 439, P > 0.05).
Fig. 4. Stable isotope bi-plots representing the isotopic niche areas of the fish communities of both sampling sites in Araruama Lagoon. Values in ‰2 indicate the corrected standard ellipse areas (SEAc).
Table 7. Layman's metrics obtained for the fish assemblages from both sampling areas in Araruama Lagoon, including number of specimens, δ13C‰ (CR range), δ15N‰ (NR range), total area (TA), mean distance to centroid (CD), mean nearest neighbour distance (MNND) and standard deviation of the mean nearest neighbour distance (SDNND)
Discussion
Stomach contents and biomass studies have been made at AL in order to understand how salinity would affect fish composition and trophic structure. While Almeida-Silva et al. (Reference Almeida-Silva, Tubino, Zambrano, Hunder, Garritano and Monteiro-Neto2015) concluded that hypersalinity was not a predominant factor influencing the trophic ecology of fish in this lagoon, Saad (Reference Saad2003) and Cruz et al. (Reference Cruz, Santos and Santos2018) suggested that species composition, biomass and, consequently, food web structure across the ecosystem were under influence of a gradient that separated species according to their tolerances to salinity variation. In hypersaline lagoons, salinity is considered to be the main abiotic factor known to influence fish composition (Vegas-Sandejas & Santillana, Reference Vegas-Sandejas and Santillana2004; Deegan et al., Reference Deegan, Lamontagne, Aldridge and Brookes2010), and changes in species and guild composition will influence the food web structure, leading to feeding (stomach contents) and isotopic variation (O'Farrell et al., Reference O'Farrell, Bearhop, McGill, Dahlgren, Brumbaugh and Mumby2014; Houssain et al., Reference Houssain, Ye, Leterme and Qin2016; Rosa et al., Reference Rosa, Alberto, Ribas, Neves and Fernandes2016). The present case study was the first attempt to identify possible fish trophic structure changes in AL using stable isotopes, and conclusions should be drawn with care.
A lower salinity year-round had been observed at site 1 (closer to the sea), with more constant, higher values at site 2 (Kjerfve et al., Reference Kjerfve, Schettini, Knoppers, Lessa and Ferreira1996, Cruz et al., Reference Cruz, Santos and Santos2018). These spatial trends regarding salinity were also observed herein, leading to a significant difference between sites (41.1 ± 3.3, 45.0 ± 1.7, Mann–Whitney U test, U = 186.5, Z = −4.37, P < 0.001, Table 1). Salinity variability can be relevant for the isotope values of the fish assemblage, since it may influence the carbon signature of the CO2. The CO2, responsible for the majority of the dissolved inorganic carbon (DIC) in the oceans, usually reflects the value of calcium carbonate (0–2‰). However, this is observed when the salinity is around 35 (Martinelli et al., Reference Martinelli, Ometto, Ferraz, Victoria, Camargo and Moreira2009). According to Fry (Reference Fry2002) and Gilikin et al. (Reference Gilikin, Lorrain, Bouillon, Willenz and Dehairs2006), in regions with lower salinity values, and consequently less carbonate, DIC δ13C signatures tend to be more negative, while a higher salinity leads to higher carbon isotope values. Consequently, salinity fluctuations will promote changes in primary carbon sources, and will eventually influence organisms on higher trophic levels. Still, the opposite was observed, since all three most abundant species (E. gula, E. argenteus and M. furnieri), with significant δ13C spatial difference (Table 4) showed lower values (δ13C = −13.6 ± 1.3, δ13C = −12.5 ± 1.6 and δ13C = −15.2 ± 2.2 for E. gula, E. argenteus and M. furnieri, respectively) in the more saline site 2 compared with site 1; this may be a consequence of a greater contribution of a δ13C depleted pelagic source (such as phytoplankton), a lower participation of enriched carbon sources (such as C4 plants) to the diet of the fish species or even a combination of the two (Bouillon et al., Reference Bouillon, Connolly, Gillikin, Wolanski and McLusky2011). Also, a spatial shift in diets supported by δ13C significant different baselines was proposed by Litvin & Weinstein (Reference Litvin and Weinstein2004) and Harrod et al. (Reference Harrod, Grey, McCarthy and Morrissey2005) as a possibility for distinct δ13C signatures found for fish species along a salinity gradient. The latter, associated with different contributions of carbon sources between sites, would explain a significant spatial difference in δ13C values of E. argenteus and E. gula.
A greater salinity standard deviation at site 1 (Table 1) was expected, since it is considered a transitional zone between the sea and the inner lagoon (Bidegain & Bizerril, Reference Bidegain and Bizerril2002). A salinity variation would mainly influence carbon primary sources and organisms in lower trophic levels (Doi et al., Reference Doi, Zuykova, Shikano, Kikuchi, Ota, Yurlova and Yandrenkina2013, Davias et al., Reference Davias, Kornis and Breitburg2013), and this seems to be an explanation for the two highest δ13C standard deviations observed for the planktivores B. pictinata and O. oglinum collected at site 1 (Table 4). Considering this is the site closest to the sea, the greater δ13C standard deviation of these two planktivorous species could also be a consequence of signatures of specimens migrating from the sea, which would have associated with a lower constant salinity (~35) compared with site 2.
The effect of a salinity variation appeared to be less clear in species of other guilds, and maybe different factors should be considered. For example, size (total length) seemed to be relevant to variations found in δ13C and δ15N of the invertivore M. furnieri. For this species, significantly smaller specimens (TL = 136.9 ± 51.5) were observed at site 1, followed by a significant spatial difference in δ15N (Table 4). These results suggest that the larger, adult individuals, mainly observed at site 2, may occupy a higher trophic level, and a positive correlation between total length and δ15N (Figure 2) corroborates this assumption. Also, a study conducted at AL by Almeida-Silva et al. (Reference Almeida-Silva, Tubino, Zambrano, Hunder, Garritano and Monteiro-Neto2015) found a higher trophic plasticity in adults that could lead to the larger variation in δ13C values observed in site 2, if M. furnieri would begin feeding on prey linked to different carbon sources. In fact, the spatial differences observed for M. furnieri can be related to size, different prey being assimilated at both sites or a combination of these two, and this is yet to be investigated.
The SEAc analysis detected spatial differences and varying degrees of isotopic niche overlap among trophic guilds (Figure 3, Table 6), and this seems mainly promoted by a narrower standard deviation and spatial changes in δ13C values of the trophic guilds (Table 5). This was observed for the carbon signatures of the planktivorous species, where a wider standard deviation at site 1 led to a larger SEAc for the guild and a larger δ13C range (CR) at a community-level when compared with site 2 (Figure 4, Table 7). The invertivorous guild (composed mostly by M. furnieri) at site 2 showed a SEAc more than two times the area observed for site 1, probably due to the trophic plasticity of adult M. furnieri (Almeida-Silva et al., Reference Almeida-Silva, Tubino, Zambrano, Hunder, Garritano and Monteiro-Neto2015) reflected in the larger δ13C standard deviation of the guild in site 2. As a consequence, the degree of overlap between the invertivorous and the other guilds was higher (Figure 3, Table 6), as well as the trophic redundancy (MNND values, Table 7) of the community in that site. Compared with site 1, closer δ13C values and smaller SEAc for the detritivorous and the planktivorous guild in site 2 also seems to be the reason for a larger degree of isotopic niche overlap between them.
Changes in the isotopic niches based on carbon values were expected due to different ranges observed for δ13C and δ15N. Considering the δ13C values published by Bouillon et al. (Reference Bouillon, Connolly, Gillikin, Wolanski and McLusky2011), a range between −7.5 and −18.9‰ shows that fish species from different trophic guilds prey spatially on carbon sources with distinct δ13C signatures. While for δ15N, a range between 9.4 and 15.8‰ suggests that fish species from AL have a limited variability regarding trophic position (Post, Reference Post2002). As an example, there were no significant differences in the δ15N mean values of the planktivorous, invertivorous, detritivorous and piscivorous guilds at site 1, contributing to a higher isotopic niche overlap among them (Shaw et al., Reference Shaw, Frazier, Kucklick and Sancho2016).
Despite the narrow δ15N range observed and with the exception of a high degree of overlap (~92%) between the omnivorous and the detritivorous guild at site 1 (Figure 3, Table 6), the overlap percentages (<60%) revealed by the SEA analysis suggest that, to some extent, the guilds have unique isotopic niche spaces. These results are in agreement with analyses of stomach content data from previous studies from AL that showed the same prey items with varying relative importance among the most abundant species. For example, while crustaceans and molluscs were the main prey categories observed for M. furnieri (Cruz et al., Reference Cruz, Santos and Santos2018), they were less important but still present in the stomachs of the detritivorous Eucinostomus spp. and the planktivorous O. glinum. The same was observed regarding polychaetes, suggested as the main food item for E. gula, and rare but still present for M. furnieri and O. glinum (Almeida-Silva et al., Reference Almeida-Silva, Tubino, Zambrano, Hunder, Garritano and Monteiro-Neto2015; Cruz et al., Reference Cruz, Santos and Santos2018). Further analyses regarding prey assimilation by the fish species are necessary in order to properly correlate stomach contents and isotopic signatures.
The present case study was the first attempt to evaluate the fish trophic structure from the AL using stable isotopes, and there was evidence of spatial differences. Salinity variation seemed to affect δ13C values of the two most abundant planktivorous species, and this influence can be seen at a guild and at a community-level. Size seemed to play an important role for M. furnieri, leading to a spatial difference in the isotopic niche of the invertivore guild. Spatial differences regarding isotopic niche overlap between guilds are most affected by δ13C values, while a narrow δ15N standard deviation detected suggests the presence of fish species from different trophic guilds in the same trophic level.
The results presented herein are the first evidence of how salinity influences fish composition and trophic structure at an isotopic level. Coupled with stomach content data from the literature, this study provides new data that may be used in the future to help decision making regarding ecosystem management. This is especially true for ecosystems under high anthropogenic pressure, such is the case with the AL. Nevertheless, further studies are necessary in order to understand how other factors may affect fish composition, feeding behaviour and stable isotope signature. Variation in isotopic signatures at different levels (species, trophic guild and community) may also be related to migration patterns, changes in prey availability, changes in contribution of primary sources, changes in baseline isotopic signatures or even any combination of the above. In the present study, data were not available for these assessments that need to be addressed in future work in order to elucidate the role of other factors in fish trophic structure of hypersaline lagoons.
Acknowledgements
This work was funded by FAPERJ, Fundação Carlos Chagas Filho de Amparo a Pequisa do Estado do Rio de Janeiro, Brazil (code 142572/2004-7 and code 210199/2006-7), and PROPPI, Pro Reitoria de Pesquisa, Pos-Graduação e Inovação – Fluminense Federal University – UFF. The authors are also grateful to the Laboratorio de Ciências Ambientais of the Centro de Biociências e Biotecnologia at the Universidade Estadual do Norte Fluminense for the use of its infrastructure. Thanks are also extended to INCT-TMCOcean on the Continent-Ocean Materials Transfer (CNPq: 573.601/08-9). Marcos A.L. Franco received postdoctorate financial support from CAPES. C.E. Rezende received financial support from CNPq (506.750/2013-2) and FAPERJ (E-26/110.032/2011).
