Trophic Ecology and Resource Partitioning of Common Snook and Mexican Snook in the Usumacinta River Delta

The abundance and distribution of species and the resources they use in a community are defined by factors acting at different scales. In environments with severe or large-scale environmental changes, such as in a lotic system with strong fluctuations in water discharge, the communities are influenced by abiotic factors (Grossman et al. 1998). In contrast, in systems with a more restricted spatial scale where the environmental conditions are favorable, the occurrence of biological interactions, such as predation and competition, is more likely (Jackson et al. 2001). In heterogeneous environments, such as lagoon systems, the communities are affected by environmental variables and biological interactions, as well as the competition to maintain the structure of the community (Mouillot et al. 2007).

The ecological niche of a species is a concept consisting of three important axes: trophic interactions, habitat use, and their temporal variability (Chase and Leboid 2003). Species diet and trophic interactions are probably the most studied components of ecological niches (Gonzalez et al. 2019). Species that occupy a similar ecological niche and coexist in the same ecosystem often develop mechanisms of resource partitioning driven by physiological, morphological, and behavioral limitations (Grossman et al. 1998), interspecific competition (Schoener 1974), and habitat segregation (Grossman et al. 1998; Jackson et al. 2001). The mechanisms that allow the coexistence of species are related to the spatial and temporal partitioning of resources (Jepsen et al. 1997; Gonzalez et al. 2019), as well as to ontogenetic feeding changes (Feltrin-Contente and Stefanoni 2009; Montaña and Winemiller 2009). Studies on fish feeding are key to identifying and understanding these mechanisms (Kroetz et al. 2016; García et al. 2018; Malinowski et al. 2019).

Snook Centropomus spp. are catadromous fish that depend on estuaries during their life cycle (Orrell 2002; Miller 2009; Nelson et al. 2016). The Common Snook Centropomus undecimalis is distributed from North Carolina to Brazil, while the Mexican Snook Centropomus poeyi is endemic to the southeastern Gulf of Mexico (Chávez 1963; Castro-Aguirre et al. 1999; Orrell 2002; Miller 2009). These species are of great importance for commercial and recreational fishing. However, these activities have greatly reduced their populations (Lorán-Núñez et al. 2012; Perera-García et al. 2013; Chávez-Caballero et al. 2014), with negative effects on aquatic food web functioning and human food security (Adams et al. 2009; Perera-García et al. 2013).

The Common Snook is a carnivorous species that shows ontogenetic differences. Small individuals (<549 mm SL) feed on invertebrates such as plankton, shrimp, crabs, and small fish to a lesser extent (Aliaume et al. 2005; Blewett et al. 2006). Large individuals (>549 mm SL) are known to consume pelagic fish that inhabit the water column, such as Inshore Lizardfish Synodus foetens, Leatherjack Oligoplites saurus, and mullet Mugil spp. (Blewett et al. 2006; Malinowski et al. 2019). Hence, they have been classified as second-order consumers (Sepúlveda-Lozada et al. 2015; Gonzalez et al. 2019; Malinowski et al. 2019). The Common Snook presents spatial and temporal changes in the composition of its diet associated with the hydrological cycle (Stevens et al. 2010, 2018), the availability of food among habitats (Blewett et al. 2006; Souza-Lira et al. 2017), and anthropogenic alterations (Adams et al. 2009). Information on the Mexican Snook diet is more limited; it has been documented that fish are a primary food source, followed by crustaceans, insects, and plant remains (Chávez 1963; Wakida-Kusunoki and Toro-Ramírez 2016). The high proportion of plant remains in the gut content of Mexican Snook as compared with Common Snook suggests that it may feed in habitats with abundant vegetation (Blewett et al. 2006; Fertlin-Contente et al. 2009).

Traditionally, fish feeding studies have been based on stomach content analysis, which allows for the taxonomic identification and quantification of recently consumed prey (Hyslop 1980) and provides a snapshot of the diet of a certain consumer. Another powerful approach is stable isotope analysis, particularly employing carbon (δ¹³C) and nitrogen (δ¹⁵N) isotopes assimilated in the tissue of consumers to identify their basal sources of food and their trophic level on larger time scales (De Niro and Epstein 1978, 1981). The combination of these methods allows accurate interpretations of trophic resource partitioning among populations by characterizing trophic complexity and the dominant pathways through which energy flows (Krebs 1998; Jackson et al. 2011; Malinowski et al. 2019).

Complementary studies between stomach content analysis and stable isotopes have shown differences in the use of habitat and in the behavior of piscivores to reduce competition for food. For instance, in Florida estuaries it has been found that the temporal partitioning of the trophic niche between Common Snook, Florida Gar Lepisosteus platyrhincus, and Largemouth Bass Micropterus salmoides, as well as between Common Snook and Smallscale Fat Snook C. parallelus, is related to the different patterns of movement of each species, such as seasonal migration to different sites for reproduction and the physiological tolerance of each species to environmental variables such as temperature (Blewett et al. 2013; Stevens et al. 2020). Recent studies have shown a substantial trophic niche overlap between Common Snook and Smallscale Fat Snook, as well as between Common Snook and Red Drum Sciaenops ocellatus (Souza-Lira et al. 2017; Gonzalez et al. 2019; Malinowski et al. 2019). However, in both cases the species developed strategies for spatial partitioning of trophic resources; while Common Snook consumed pelagic prey, Smallscale Fat Snook and Red Drum fed on prey that inhabit shallower sites with greater vegetation and woody debris located in the benthic zone of the estuary (Malinowski et al. 2019; Stevens et al. 2020). Studies in estuaries in Brazil indicate that Common Snook and Smallscale Fat Snook trophic niches present spatial variations in the degree of breadth and overlap, related to the availability and diversity of trophic resources in these systems (Souza-Lira et al. 2017; Gonzalez et al. 2019).

In this study, we analyzed the feeding of Common Snook and Mexican Snook, which coexist in the Usumacinta River delta, in the southern Gulf of Mexico, with the aim of identifying patterns of trophic resource partitioning between the species. The objectives were to estimate (1) variability in diet among species, size-classes, season (dry, rainy, and “nortes”), and site (Pom−Atasta and Palizada River systems), (2) differences in trophic level related to the ontogenetic changes of the species, and (3) trophic niche breadth and the degree of trophic overlap between species. To do this we analyzed stomach content and stable isotope ratios of carbon and nitrogen of the two fish species. Our hypotheses were that the diet of both snook species was made up of fish, crustaceans, and mollusks, with fish being the main component. We expected that ontogenetic changes would be observed in the diet; the specimens of adult sizes would feed mainly on fish, while the small sizes would have greater consumption of invertebrates, showing changes in trophic levels during their growth. We expected trophic overlap between snook species, although they would present strategies of spatial and temporal partitioning of resources.

METHODS

Study area

The Usumacinta River delta, in the southern Gulf of Mexico, is characterized by coastal ecosystems with high freshwater inputs and widespread wetlands that host a great diversity of birds, fish, and invertebrates. The Centla Biosphere Reserve was established to protect this area, which also includes the mouth of the San Pedro–San Pablo River, while the Palizada River mouth is located in the Terminos Lagoon Protected Area (Yáñez-Arancibia and Day 2009; Soria-Barreto et al. 2018); both areas are Ramsar sites. These rivers feed the Pom−Atasta and Palizada estuarine lagoon systems, located at the eastern end of the Usumacinta River delta and west of Terminos Lagoon in the Gulf of Mexico (Figure 1). This area is characterized by three hydrological seasons: a dry season from March to May, a rainy season from June to October, when maximum rainfall occurs, and a “nortes” season from November to February, characterized by strong winds and intermediate rainfalls (Yáñez-Arancibia et al. 2009).

The Pom−Atasta (PA) is a system of two lagoons joined by a narrow channel (Figure 1). The Pom Lagoon is fed by the San Pedro–San Pablo River, an effluent from the Usumacinta River, through natural and artificial channels that generate low-salinity conditions. By contrast, conditions of higher salinity predominate in Atasta Lagoon due to the entry of water from Terminos Lagoon through a complex system of smaller lagoons, forming a salinity gradient from west to east of 0‰ to 28‰ (Muciño-Márquez et al. 2017). The vegetation is dominated by the red mangrove Rhizophora mangle and the black mangrove Avicennia germinans.

The Palizada estuarine lagoon system (PR) comprises the mouth of the Palizada River, the main effluent of the Usumacinta River, in Terminos Lagoon (Figure 1). This system has two open deltaic lagoons, El Vapor and Del Este, where four smaller permanent rivers drain. The Palizada River flows into Terminos Lagoon through two mouths, the main one being Boca Chica (Fuentes-Yaco et al. 2001). The system has a substantial influence of fresh water from the Palizada River, as well as brackish water due to its direct connection with Terminos Lagoon. There is a salinity gradient of 0‰ to 10‰ from south to north, and vegetation dominance shifts from the reed Phragmites australis to mangroves along the same orientation (Fuentes-Yaco et al. 2001; Muciño-Márquez et al. 2017).

Field surveys

Surveys were conducted during the rainy season (June and September 2017), nortes season (January 2017 and 2018), and dry season (March 2017 and May 2018). The fish were captured with multimesh gill nets that were 55 m long and 2.5 m high, with a mesh size of 2.5 to 9 cm. Longlines 50 m long and cast nets 2 m high (1-cm mesh size) were also employed. The gill nets and longlines were placed on the margins of the lagoons during the mornings for an approximate period of 4 h. With the cast net, 15 sets were made at each sampling site during the mornings. Commercial fishermen further helped collecting snook stomachs at the study sites. The fish were caught using gill nets (a mesh size of 10 to 11.5 cm) and harpoon.

In the field, we measured weight (g) and standard length (cm) of each fish; the fish and their stomachs were then fixed with a 10% formalin solution. In the laboratory, the specimens and stomachs were rinsed with tap water and preserved in 70% alcohol. Later they were identified to the species level with the keys of Castro-Aguirre et al. (1999) and Orrell (2002).

Size structure

The size-classes were assigned based on Stevens et al. (2007), Chávez-Caballero (2011), and Lorán-Nuñez et al. (2012), which report the size for juveniles (350 mm SL) and size at first sexual maturity (>500 mm SL). Three size-classes were defined accordingly: C1 (<350 mm SL), C2 (351–499 mm SL), and C3 (>500 mm SL).

Diet analysis

We measured the total length of the digestive tract and recorded visually the degree of fullness, according to the following categories: full (75–100%), half full (50–75%), almost empty (25–50%), and empty (0–25%) (Bowen 1996). The contents were removed from the anterior portion of the tract. The diet was analyzed with the volumetric method proposed by Hyslop (1980), which consists of recording the water displacement produced by each food component in a graduated cylinder. Small components were visually compared to a drop of water of a specified volume (Bowen 1996). Empty or highly digested stomachs were not included in the analysis. The food components were taken from the anterior third of the intestine and were identified to the lowest possible taxon. Fish were identified with the taxonomic keys of Schultz and Miller (1971), Castro-Aguirre et al. (1999), Orrell (2002), Armbruster and Lawrence (2006), Betancur and Willink (2007), Marceniuk and Betancur (2008), and Miller (2009), while crustaceans and mollusks were identified with keys from Tavares (2002a, 2002b) and Leal (2002a, 2002b).

Trophic diversity accumulation curves were performed for each species, size-class, season, and system using the EstimateS 9 software (Colwell 2013) to determine if sample size was adequate to describe the diet variation. The accumulation curves reached or showed a tendency to asymptotes, which indicates that the number of stomachs was adequate to represent the variation in the diet of both species (Figures S1–S4 in the Supplement provided in the online version of this article).

For statistical analysis, prey were grouped into 18 categories based on their taxonomic affinities: (1) fish remains (bones, scales, muscle, eyes, and eggs of unidentifiable species), (2) Cyprinodontiformes (Cyprinidon sp., Pike Killifish Belonesox belizanus, Gambusia sp., Yucatan Gambusia G. yucatana, Spottail Killifish Heterandria bimaculata, Poecilia sp., Péten Molly P. kykesis, Shortfin Molly P. mexicana, Southern Platyfish Xiphophorus maculatus), (3) Cichliformes (Paracrhomis sp., Jack Dempsey Rocio octofasciata, Yellowbelly Cichlid Trichromis salvini, Quetzal Cichlid Vieja melanurus, species in the family Cichlidae), (4) Clupeiformes (species in the family Engraulidae, Atlantic Anchoveta Cetengraulis edentulus, Longfin Gizzard Shad Dorosoma anale, Threadfin Shad D. petenense), (5) Pterygoplichthys (Amazon Sailfin Catfish Pterygoplichthys pardalis, Vermiculated Sailfin Catfish P. disjunctivus, sailfin catfish Pterygoplichthys sp.), (6) Ariidae (Cathorops sp., Hardhead Catfish Ariopsis felis, species in the family Ariidae), (7) Thorichthys (Firemouth Cichlid T. meeki, Yellow Cichlid T. helleri, Thorichthys sp.), (8) Perciformes (Burro Grunt Pomadasys crocro, Freshwater Drum Aplodinotus grunniens), (9) Gobiidae (Gobiidae species 1 and 2, Fat Sleeper Dormitator maculatus), (10) Ladyfish Elops saurus, (11) tetra Astyanax finitimus, (12) mullet Mugil sp., (13) mojarra Diapterus sp., (14) other decapods (cock shrimp Exhippolysmata oplophoroides, Hippolytidae, crayfish Procambarus llamasi, shrimp Litopenaeus sp., cinnamon river shrimp Macrobrachium acanthurus, crustacean remains), (15) invertebrate remains (Diptera: Chironomidae, remains of other insects, remains of polychaetes), (16) mollusks (Gastropoda 1 and 2, remains of bivalves), (17) crab Callinectes sp., and (18) plant remains.

For each specimen, the volumes of each category were recorded in proportions from 0 to 1 and were transformed with the arcsine square root to reduce the influence of the prey with greater volume. Subsequently, the normality and homoscedasticity of the data were analyzed with the Kolmogorov–Smirnov and Barlett tests, respectively.

To analyze the ontogenetic, temporal, and spatial effects, the diet was compared among species (Common Snook and Mexican Snook), seasons (rainy, nortes, and dry season), sites (PR, PA), and size-classes (C1, C2, and C3). We further analyzed the joint effect of these factors with a four-way permutational multivariate analysis of variance (PERMANOVA) (9,999 permutations) with a Bray–Curtis matrix. The analyses were carried out with the vegan package (Oksanen et al. 2019) in R (R Core Team 2018). Significant interactions were analyzed in pairs with a Tukey’s honestly significant difference post hoc test using the "TukeyHSD" function.

Isotope analysis

Dorsal muscle samples (1 cm³) were taken from the freshly caught snook during the rainy, nortes, and dry seasons. Samples of the main basal resources were collected to determine the isotopic baseline of the aquatic food web. The sources included phytoplankton, leaves of dominant riparian vegetation (i.e., reed, red and black mangrove), epiphytic algae attached to mangroves and wood remains, leaf litter, seston, and sediment.

Phytoplankton samples were collected with a cascade filtering device with three stacked sieves (15, 30, and 50 μm), through which large volumes of water were filtered manually using a 20-L container until the 15-μm sieve was saturated (Sepúlveda-Lozada et al. 2015). The retained fraction was washed with distilled water and filtered through precombusted Whatman GF/F filters of 0.7-µm pore size. The seston was collected by filtering 60 mL of surface water through precombusted Whatman GF/F filters. The filters were wrapped in aluminum foil and stored on ice for transportation and later analysis. The leaves of the riparian vegetation were cut directly from the plants. The algae, leaf litter, and sediment were collected manually. Snook tissue samples and basal resources were placed in salt (Arrington and Winemiller 2002). All samples were stored on ice for conservation and transportation.

In the laboratory, the samples were washed with distilled water to remove excess salt and dried at 60°C for 48 h. All samples were pulverized with a mortar and pestle (Arrington and Winemiller 2002). Subsamples of 1–2.5 mg were placed in tin capsules and sent to the University of California–Davis Stable Isotope Facility, California, to determine the content of δ¹³C and δ¹⁵N. The carbon and nitrogen results were expressed in parts per thousand (‰) as δ¹³C and δ¹⁵N values, respectively, derived from the relationship between the isotopic value of the sample and a known standard with the formula

$${\mathrm{\delta }}^{13}\text{C or}{\mathrm{\delta }}^{15}\mathrm{N}=\left[\left(R_{\text{sample}}/R_{\text{standard}}\right)-1\right]\times {10}^3,$$

where R corresponds to the ¹³C/¹²C rate or the ¹⁵N/¹⁴N rate (Fry 2006). The standards used for δ¹³C and δ¹⁵N were Vienna Pee Dee Belemnite limestone and atmospheric nitrogen, respectively.

Comparisons of isotope values between snook species and between systems were made with a Mann–Whitney U-test (Zar 1996). Comparisons of isotopic values among size-classes and seasons were made with a Kruskall–Wallis test using the "kruskal.test" function in R.

Trophic level

The trophic level of snook was compared between species and size-classes, using diet and isotope data. The determination of trophic position from the volumetric data (TP_SCA) was carried out with the formula proposed by Winemiller (1990):

$${\text{TP}}_{\text{SCA}}=1+\sum T_j\left(V_{\mathit{ij}}\right),$$

where V_ij and T_j are the volumetric contribution of species i consisting of prey species j and the trophic level of prey j, respectively. Prey trophic levels were obtained from Adams et al. (1983).

The determination of the trophic level from the δ¹⁵N values (TP_SIA) was carried out with the formula proposed by Post (2002):

$${\text{TP}}_{\text{SIA}}=\left({\mathrm{\delta }}^{15}{\mathrm{N}}_{\text{consumer}}-{\mathrm{\delta }}^{15}{\mathrm{N}}_{\text{baseline}}\right)/\text{TDF}+{\text{TP}}_{\text{baseline}},$$

where δ¹⁵N_consumer and δ¹⁵N_baseline are the δ¹⁵N values of the consumers and the baseline, TDF is the trophic discrimination factor (equal to 2.4‰ according to Vanderklift and Ponsard [2003] for marine fish), and TP_baseline (equal to 1) is the trophic position of the isotopic baseline. The average δ¹⁵N of phytoplankton, seston, epiphytic algae, and leaf litter was assumed as δ¹⁵N_baseline. From the analysis of stable isotope biplots (Figure 2), it was determined that these basal sources were the ones that contributed the most to the snook biomass. The relative position of the consumers on the δ¹³C axis reflects the potential contribution of various carbon sources to its biomass (Soria-Barreto et al. 2021).

The trophic level was estimated with the tRophicPosition package (Bayesian Trophic Position Calculation with Stable Isotopes) in R (Quezada-Romegialli et al. 2017). The TP_SCA and TP_SIA results were compared between size-classes for each species with a Kruskal–Wallis test.

Niche width and niche overlap

The breadth of the diet (B_A) was determined with the standardized index of Levin (1968):

$$B_A=\left[\left(1/\mathrm{\Sigma }{p}_{\mathit{ij}}-1\right)\right]/\left(n-1\right),$$

where p_ij is the proportion of prey i in the diet of species j, and n is the total number of prey considered. If the values are close to 0, the consumer is considered to be a specialist, if the values are close to 1, the consumer is considered to be a generalist (Krebs 1998).

Diet overlap (O_ik) was compared between species and their size-classes using the volumetric data of the diet with the Pianka index (1976):

$$O_{\mathit{ik}}=\mathrm{\Sigma }\left(p_{\mathit{ij}}\times p_{\mathit{ik}}\right)/\left[\sqrt{\left(\mathrm{\Sigma }{p}_{\mathit{ij}}^2\times p_{\mathit{ik}}^2\right)}\right],$$

where p_ij is the proportion of resource i in relation to the total resources of species j and p_ik is the proportion of resource i in relation to the total resources of species k. This measure of overlap ranges from 0 (no similarity in the use of resources) to 1 (complete overlap) (Krebs 1998).

The isotopic niche breadth was estimated using the standard ellipse area (SEA), which represents a bivariate measure of the distribution of individuals in a two-dimensional (i.e., carbon and nitrogen) isotope space (Jackson et al. 2011). The SEAs were calculated based on Bayesian inference (SEA_B), which produces a range of confidence intervals (95, 75, and 50%) for the calculated ellipses and is expressed in parts per thousand squared (‰²). We further calculated SEAs corrected for small sample size (SEA_C), which contain approximately 40% of the total isotope data and are used for spatial and temporal comparison of isotopic niches.

A SEA overlap index was used to estimate the overlap in isotopic space between species in each system. The SEA overlap index was calculated as the ratio between the overlapping and nonoverlapping area of the ellipses based on Montaña et al. (2020). An SEA overlap index = 1 indicates a complete overlap between the ellipses, meaning that the species share the same trophic niche; a value <1 indicates that the overlapping area is less than the nonoverlapping area, and thereby points to a limited overlap of trophic niches; a value >1 implies that the overlapping area is greater than the nonoverlapping area, hinting to a greater overlap between trophic niches. The SEA and the SEA overlap index were applied for the different species and size-classes. The estimation of SEAs and their overlap was carried out with the SIBER package (Stable Isotope Beyesian Elipses in R) in R (Jackson et al. 2011).

RESULTS

Stable Isotope Analysis

A total of 213 dorsal muscle samples were obtained, 172 from Common Snook and 41 from Mexican Snook. Common Snook δ¹³C values ranged from −32.57 to −20.31‰ (mean ± SD = −26.70 ± 2.20), and δ¹⁵N ranged from 8.17 to 14.18‰ (11.52 ± 1.16) (Figure 2). For the Mexican Snook, the δ¹³C values varied from −31.63 to −23.41‰ (−27.80 ± 1.70) and the δ¹⁵N from 8.14 to 14.08‰ (11.14 ± 1.49) (Figure 2). Statistically significant differences between snook species were observed for the δ¹³C values (P < 0.001); Common Snook presented higher values (Table 3). There were no significant differences between species for δ¹⁵N values (P = 0.192).

TABLE 3. Range (mean ± SD in parentheses) of carbon (δ¹³C) and nitrogen (δ¹⁵N) isotopes (measured in ‰) for three size-classes (C1, C2, C3) of the two snook species. Analysis was done using a Kruskal–Wallis test (P = 0.05). Significant values are shown in bold italics.

Species	Species in		C1		C2		C3
	δ13C	δ15N	δ13C	δ15N	δ13C	δ15N	δ13C	δ15N
Common Snook	−32.57 to −20.31 (−26.61 ± 2.20)	8.71 to 14.18 (11.51 ± 1.16)	−32.57 to −21.09 (−26.78 ± 1.89)	9.29 to 13.50 (11.35 ± 1.15)	−31.47 to −20.37 (−26.54 ± 2.60)	8.71 to 13.67 (11.70 ± 1.07)	−30.26 to −20.31 (−26.01 ± 2.7)	10.09 to 14.18 (11.94 ± 1.24)
Mexican Snook	−31.63 to −23.47 (−27.80 ± 1.70)	8.14 to 14.08 (11.14 ± 1.49)	−31.63 to −26.20 (−27.83 ± 1.71)	8.14 to 11.98 (9.92 ± 1.11)	−30.06 to −25.04 (−27.49 ± 2.11)	8.95 to 11.18 (9.57 ± 0.91)	−30.25 to −23.47 (−27.85 ± 1.69)	9.98 to 14.08 (11.92 ± 1.12)

The Common Snook showed greater variability in isotopic values in the three size-classes when compared to the Mexican Snook (Table 3). There were no statistical differences among the Common Snook size-classes either for δ¹³C (P = 0.721) or δ¹⁵N values (P = 0.06). Among the size-classes of Mexican Snook, there was no statistical differences for the δ¹³C values (P = 0.698), while the δ¹⁵N values were significantly higher (P < 0.05) in the C3 class. The spatial and temporal variability in δ¹³C and δ¹⁵N of the size-classes of Common and Mexican snooks are presented in Figures 3 and 4. Statistical comparisons of the breadth of the isotopic space between size-classes were not performed due to the limited number of samples.

In the PA system, there were no statistically significant differences between species for the δ¹³C values (P = 0.349). Nonetheless, the Common Snook showed significantly higher (P = 0.001) δ¹⁵N values compared with the Mexican Snook. In the PR system, the Common Snook presented significantly higher (P < 0.001) δ¹³C values, while no statistically significant variation was observed for δ¹⁵N (P = 0.361) among species. The δ¹⁵N values of leaf litter and epiphytic algae were significantly higher (P = 0.0001) in the PR system than in the PA system, while phytoplankton and seston did not differ significantly (P > 0.05) among systems.

Trophic Niche Size and Overlap

The estimated diet breadth (B_A) of the Common Snook was B_A = 0.19, and similar values were found between systems (B_A = 0.14 in both the PA and PR systems). For the Mexican Snook, B_A = 0.22; larger values were found in the PA system (B_A = 0.23) than in the PR system (B_A = 0.11). The estimated B_A for C1 and C2 classes of the Common Snook was 0.09, while for C3 it increased considerably (0.23). For the Mexican Snook, B_A values increased according to the size-classes (C1 = 0.10, C2 = 0.19, C3 = 0.23). Overall diet overlap between Common and Mexican snooks was relatively high (O_ik = 0.80). The overlap between the two species was greater in the PA system (O_ik = 0.82) than in the PR system (O_ik = 0.68). With regard to size-classes, diet overlap among the two species was greater for classes C1 (O_ik = 0.72) and C2 (O_ik = 0.74) compared with C3 (O_ik = 0.64).

Isotopic niche breadth for snook varied among seasons; Common Snook presented greater isotopic niche space compared with Mexican Snook. For Common Snook, the higher values presented in the nortes season (SEA_C = 10.33‰²) compared with the rainy (SEA_C = 5.31‰²) and dry (SEA_C = 2.98‰²) seasons. Mexican Snook had higher values in the rainy season (SEA_C = 5.31‰²) than in the dry (SEA_C = 4.74‰²) and nortes (SEA_C = 3.73‰²) seasons (Figure 5). The credibility interval values are presented in Table 5. Isotopic niche overlap between Common and Mexican snooks was greater in the nortes season (SEA overlap index = 2.39) than in the rainy (SEA overlap index = 0.74) and dry (SEA overlap index = 0.008) seasons. These results suggest that in the nortes season the species presented share the isotopic niche space, compared with rainy and dry seasons.

TABLE 5. Bayesian credibility intervals of SEA_B from Common Snook (Cu) and Mexican Snook (Cp) in the rainy (R), nortes (N), and dry (D) seasons.

Season and species	Percent (%)	Interval
		Lower	Upper
R-Cu	99	3.494865	7.487365
	95	3.942894	6.701504
	50	4.741515	5.661386
R-Cp	99	1.515536	13.795968
	95	2.146072	10.168143
	50	3.599609	5.940807
N-Cu	99	7.102475	13.86077
	95	7.871904	12.79574
	50	9.363292	11.01242
N-Cp	99	2.073574	6.030115
	95	2.377411	5.239293
	50	3.05427	4.001837
D-Cu	99	1.960192	4.201437
	95	2.164851	3.862028
	50	2.621971	3.200525
D-Cp	99	0.9096364	11.176687
	95	1.503498	8.246602
	50	2.8109329	4.772576

Isotopic niche breadth of snook varied among systems. Common Snook presented lower values in the PA system (SEA_C = 5.07‰²) compared with the PR system (SEA_C = 7.01‰²). In contrast, the isotopic niche space of Mexican Snook was greater in the PA system (SEA_C = 7.19‰²) than in the PR system (SEA_C = 2.93‰²) (Figure 6). The credibility interval values are presented in Table 6. Isotopic niche overlap between Common Snook and Mexican Snook was greater in the PA system (SEA overlap index = 2.09) than in the PR system (SEA overlap index = 0.89). This indicates that in the PA system the two snook species present a similar trophic niche, while in the PR system lower overlap between species points to more diverse trophic niches (Figure 7).

TABLE 6. Bayesian credibility intervals of SEA_B from Common Snook (Cu) and Mexican Snook (Cp) in the Pom−Atasta (PA) and Palizada River (PR) systems.

Season and species	Percent (%)	Interval
		Lower	Upper
PA-Cu	99	3.71146	6.63481
	95	4.001598	6.189696
	50	4.620774	5.361452
PA-Cp	99	5.217565	9.041004
	95	5.553517	8.533842
	50	6.39272	7.378774
PR-Cu	99	3.247643	13.396048
	95	3.798425	11.083931
	50	5.467241	7.661113
PR-Cp	99	1.521134	4.800451
	95	1.827951	4.154412
	50	2.43728	3.214374

DISCUSSION

Diet analysis indicated that both Common Snook and Mexican Snook are piscivorous fish. In agreement with the hypothesis about resource partitioning between snook species, our data suggest that variability in their diet composition is related to the exploitation of different types of prey and the differential use of space for the acquisition of food. The Common Snook consumed primarily fish from the families Engraulidae, Clupeidae, and Elopidae, highly mobile species that inhabit the water column (Luczkovich et al. 1995; Blewett et al. 2006). In contrast, the Mexican Snook consumed more fish of the Cichlidae, Poeciliidae, and Loricariidae families, which are associated with aquatic vegetation located in the shallow areas of the estuary (Miller 2009; Wakida-Kusunoki and Toro-Ramírez 2016). Accordingly, gut content analysis revealed a higher proportion of plant remains in Mexican Snook versus Common Snook. Moreover, significant differences in carbon isotope values of the two species suggest the use of distinct basal resources, habitats, or both. Hence, our findings potentially indicate a process of spatial partitioning of resources. While the Common Snook captures prey in the pelagic zone of the estuary, the Mexican Snook feeds in areas of abundant vegetation (e.g., on the edges of rivers or lagoons of the estuary). These results are consistent with previous research from estuaries of Florida and Brazil, which highlighted spatial segregation among snook species (Souza-Lira et al. 2017; Gonzalez et al. 2019; Stevens et al. 2020).

Gut content analysis showed variability in the dietary composition of snook in relation to their size. Smaller (i.e., C1) Common Snook consumed more crustaceans compared with larger (i.e., C2 and C3) individuals. The higher proportion of invertebrates is due to the fact that these are prey with low mobility that are relatively easy to capture and they fit in a limited gape size, which limits the larger prey consumed by small-sized snook (Luczkovich et al. 1995; Aliaume et al. 2005; Blewett et al. 2006; Fertlin-Contente et al. 2009; Dukta-Gianelli 2014). As expected, more elusive and larger prey were found in the stomachs of larger snook (i.e., C2 and C3). This result is consistent with previous evidence that prey selection by snook is related to their size, skills, and metabolic requirements at different stages of development (Luczkovich et al. 1995; Fertlin-Contente et al. 2009), establishing a positive relationship between predator and prey size (Blewett et al. 2006; Dukta-Gianelli 2014). Comparison of isotopic data among the different Common Snook size-classes did not show significant variation, indicating the consumption of prey with similar carbon and nitrogen isotope values (Arrington and Winemiller 2002; Post 2002). The discrepancy between volumetric and isotopic data can be explained by the similar trophic levels occupied by several crustaceans and fishes; these groups often share habitats and rely on similar basal resources (Sepúlveda-Lozada et al. 2015).

The positive relationship between predator and prey size has been observed in different snook species (Blewett et al. 2006; Fertlin-Contente et al. 2009; Dukta-Gianelli 2014; Gonzalez et al. 2019). The opportunistic nature of snook has also been highlighted, considering the consumption of small prey by large snook individuals when the former is abundant (Blewett et al. 2006; Dukta-Gianelli 2014). In our study, the high proportion of Poeciliidae in the diet of the Mexican Snook in C2 can be explained by their high relative abundance (Aragón-Flores et al. 2021) and availability for the predators. For instance, one specimen of Mexican Snook from C2 contained 20 poecilids in its stomach (Petén Molly, Shortfin Molly, Spottail Killifish, and Southern Platyfish). Consumption of small prey by large snook has been previously observed in estuaries of southwestern Florida (Blewett et al. 2006).

The trophic level of Common and Mexican snooks determined with volumetric data and nitrogen stable isotopes was estimated to range between 3 and 4, placing them as second or third order consumers within their food webs (Sepúlveda-Lozada et al. 2015). These results agree with the piscivorous habits previously reported for Common Snook of different sizes in estuaries of Brazil and Florida (Gonzalez et al. 2019; Malinowski et al. 2019). The significant increase in the trophic level of Mexican Snook in C3 determined by nitrogen stable isotope analysis may well be related to the higher consumption of prey such as the Shortfin Molly, Yellowbelly Cichlid, and Hardhead Catfish, which present relatively high δ¹⁵N values due to their planktivorous, zoobenthivorous, and detritivorous habits, respectively (Sepúlveda-Lozada et al. 2015, 2017).

Temporal changes in the Common Snook diet may be associated with the seasonal hydrological cycle as the latter affects the composition of fish species in the studied systems and influences the temporal availability of prey (Ferreira et al. 2019; Aragón-Flores et al. 2021). During the rainy season, the low-salinity conditions caused by high riverine discharge and runoff (Barriero-Güemes and Aguirre-León 1999; Fuentes-Yaco et al. 2001) favor the migration of species such as Fat Sleeper and Atlantic Anchoveta to the PA and PR systems, where the species gather in large numbers for reproduction (Amezcua-Linares and Yáñez-Arancibia 1980; Sánchez-Velazco et al. 1996). In contrast, during the dry season the salinity increases and species such as Threadfin Shad and penaeid shrimps become more abundant (Tavares 2002a; Aragón-Flores et al. 2021), thus increasing their availability and importance in the Common Snook diet.

The composition of Common Snook and Mexican Snook diets presented spatial variability that may be associated with the differences in the freshwater and saltwater inputs received by the PR and PA systems. The PR system presents greater salinity fluctuations due to the constant inflow of freshwater from the Palizada River and its direct connection with the brackish waters of Terminos Lagoon (Fuentes-Yaco et al. 2001). These conditions favor the establishment of species that may tolerate strong variations in salinity, such as Atlantic Anchoveta and species of the genus Pterygoplichthys (Castro-Aguirre et al. 1999; Orrell 2002; Capps et al. 2011), which were abundant in the diet of both snook species in this system. In the PA system, lower brackish water inputs from Terminos Lagoon and the establishment of a more lentic environment favored by its semicircular shape (Barreiro-Güemes and Aguirre-León 1999; Ramos-Miranda et al. 2006) allow the presence of species such as the Burro Grunt and the Fat Sleeper (Sánchez-Velazco et al. 1996; Miller 2009), as well as species from the Cichlidae and Poeciliidae families (Myers 1949; Miller 2009), which were important in the Common Snook and Mexican Snook diets in this system.

The Levin’s index results for Common and Mexican snooks were mainly related to the high proportion of fish remains in their diet (>35%). The Levin’s standardized index weighs the proportions of prey in consumers. When the proportion of a single prey is high, the predator is considered a specialist (Krebs 1998). Our results are consistent with findings from estuaries of Brazil and the Pacific of Mexico, where the Smallscale Fat Snook, Common Snook, and Yellowfin Snook Centropomus robalito presented similar values of the Levin’s index, which were attributed to the high reliance on fish and crustaceans (Fertlin-Contente et al. 2009; Moreno-Sánchez et al. 2015; Souza-Lira et al. 2017). We acknowledge that treating unidentified fish remains as a food item may pose a bias in the interpretation of dietary results (Baker et al. 2014). Prey identification in piscivorous species is often challenging, mainly due to their rapid digestion (Beukers-Stewart and Jones 2004; Souza-Lira et al. 2017). We deemed it opportune to include this category of diet considering the high proportions in the stomach and the important nutritional contribution to piscivorous species (Souza-Lira et al. 2017). A high proportion of fish remains in snook stomachs have also been reported in previous studies (Aliaume et al. 2005; Blewett et al. 2006; Moreno-Sánchez et al. 2015; Toro-Ramírez et al. 2014). The diet breadth increased in relation to body size of Mexican Snook; this was associated with a greater number of prey consumed by C3 (24 items) than by C1 (9 items) fish, which increased the Levin’s index values (Kroetz et al. 2016; Souza-Lira et al. 2017; García et al. 2018).

The trophic overlap among Common and Mexican snooks varied between systems, presenting higher values in the PA system than in the PR system, as indicated by both isotopic and volumetric data. This result may be associated with the availability of trophic resources in the study systems (Souza-Lira et al. 2017; García et al. 2018; Gonzalez et al. 2019). In the PA system, we observed no significant differences in δ¹³C values between Common and Mexican snooks and 50% of the gut content in both species was composed by fish remains and Fat Sleeper. By contrast, in the PR system the δ¹³C values of Common and Mexican snooks differed significantly, pointing to differential use of carbon sources. Volumetric data also showed that 45% of their diet was different (i.e., with predominance of Atlantic Anchoveta in the diet of Common Snook and unidentified fish remains, which may have included the Atlantic Anchoveta, in Mexican Snook). These results indicate that, in the PA system, the two snook species may feed on similar trophic resources, encompassing a wide prey spectrum (García et al. 2018; Gonzalez et al. 2019). In contrast, in the PR system, the two snook species may rely on different trophic resources. Partitioning processes may be associated with lesser prey availability in this system (Losos 2000; Gonzalez et al. 2019) or reliance on different feeding habitats (Malinowski et al. 2019; Stevens et al. 2020).

This study highlights the occurrence of spatial partitioning of trophic resources between Common and Mexican snooks, which allows their coexistence in estuaries of the Usumacinta River delta. The Common Snook consumed more elusive prey in the pelagic zone of the estuary, while the Mexican Snook fed on prey associated with vegetation and littoral areas. However, a substantial overlap in their trophic niches reflects that these species are not segregated and consume similar prey in estuaries. This result may be related not only to interspecific competition, but other biological interactions and environmental variables may also be affecting the exchange of resources (Jackson et al. 2001; Mouillot et al. 2007). Based on their specialist habits and trophic level, both snook species appear to be top predators in the studied systems (Adams et al. 2009; Blewett et al. 2013; Ferreira et al. 2019).

Overfishing represents a serious threat to these species’ populations in the southern Gulf of Mexico (Perera-García et al. 2013), mainly in the states of Tabasco and Campeche, where there are no official regulations that regulate the capture of Common Snook and Mexican Snook, which allows large catches to be made during the months of their reproduction and catches of organisms of small size that have not reached maturity (Chávez-Caballero et al. 2014). Understanding spatiotemporal variability in the diet of these snook species is fundamental in order to establish effective conservation and management strategies, such as preserving their feeding and breeding grounds, establishing closed fishing seasons, and promoting the conservation of important prey (i.e., shrimp, Atlantic Anchoveta, and Fat Sleeper) in the Usumacinta River delta. This is particularly valid for the Mexican Snook because the availability of its prey depends on riparian vegetation, it is a species native to the Mexican coasts, and it is less abundant and has had fewer ecological studies compared with the Common Snook (Chávez-Caballero et al. 2014).