Individual Variation in Diet of Four Fishes in Shallow Tropical Estuarine Areas: Comparisons Between Seasons and Habitats

Abstract

For the development of conservation and management strategies, scientists often define a ‘typical’ individual from a population and act as if all members of that group are basically the same, ecologically speaking. However, studies have found that resource use can vary among individuals from the same population, often due to their life stage, and that this variation helps reduce intraspecific competition among them. In this study, the feeding habits and the degree of individual specialization in the use of food resources, determined by the individual specialization index, were analyzed for the populations of the Brazilian silverside (Atherinella brasiliensis), mojarra (Eucinostomus argenteus), and two anchovy species (Anchoa tricolor and Lycengraulis grossidens) in different estuarine habitats. A total of 3855 specimens were collected from vegetated and unvegetated areas during the rainy and dry seasons of 2014. Overall, the highest density and biomass were registered in unvegetated areas. Density differed significantly among sites for A. brasiliensis, while for anchovies it differed significantly among seasons. Species exhibited spatial segregation in feeding, with anchovies feeding primarily in the water column, while A. brasiliensis and E. argenteus were benthic feeders. Despite using similar items, benthic species showed little dietary overlap, coexisting through differential use of prey proportions. A. brasiliensis, E. argenteus and A. tricolor were generalists, with the first two showing high intrapopulational diet variation (low IS values), whereas L. grossidens was a specialist with narrower niche breadth and lower individual specialization. Corroborating the Niche Variation Hypothesis, generalist species had a higher degree of individual specialization, which may be related both to an attempt at resource partitioning and to reducing intraspecific competition.

1. Introduction

Diet variation is common in estuarine and marine fish populations and may occur in response to growth [1,2,3]), morphological shifts [2], and energy allocations by various changes in environmental factors, such as temperature, salinity and oxygen levels [4]. Traditionally, researchers have treated conspecific individuals as ecologically equivalent [5,6], but they have often found it difficult to explain why individuals of the same species end up using different resources [7]. Hence, if considering the individuals of the same population to be identical, this contradicts the basic principles of ecology, and thus, the proportions of individuals that may use different resources can vary among populations [8]. Furthermore, focusing only on population generalities can overlook those individuals that use other habitats or alternative food resources [6]. There is further strong evidence that the use of resources may differ substantially between the individual and population levels (“Hypothesis variation niche”) [9], linked to reducing intraspecific competition [10]. Most ecological studies, especially in relation to trophic niches, have attributed this variation to ontogenetic shifts or sexual dimorphism [11].

Intrapopulational diet variation cannot be associated only with ontogeny, sex or polymorphisms of species [7], since variations in resource utilization are also observed in individuals belonging to the same size class or in the same sex [12,13]. This individual-level variation is generally related to limitations in the ability of an individual to efficiently explore a wide range of resources; that is, there is greater intraspecific competition [12,14]. Individual specialization in resource use has been documented across several taxonomic groups and may have evolutionary, ecological, and conservation implications [7,9,14,15,16,17].

Among the theoretical models developed to explain species coexistence in an ecosystem, Niche Theory has been the most widely adopted [18]. According to Hutchinson (1957) [19], the niche is a multidimensional space defined by the range of biotic and abiotic conditions under which an individual or a species can survive indefinitely. The main idea behind this theory is that interspecific competition drives species to develop distinct niches [20,21], thereby facilitating species coexistence [22,23]. Following this argument, the Niche Variation Hypothesis proposes that populations with wider niches exhibit greater intrapopulation variation than those with narrower niches [9]. Thus, according to the same author, general populations would be composed of individuals with specialisms. For fishes, this was recorded in natural populations of perch Perca fluviatilis (L. 1758) in an oligotrophic lake in central Sweden [10] and demonstrated experimentally in populations of stickleback Gasterosteus aculeatus (L. 1758) in Canada [24].

Estuaries are among the most productive aquatic ecosystems on the planet [25,26,27], yet they are also considered the most threatened marine environments worldwide [28]. This is largely because estuaries sit close to human population centers, making them vulnerable to land-use change, habitat loss, pollution, dredging, wetland drainage, and destructive fishing practices [29]. According to Pereira et al. [30]), the main anthropogenic threats recorded in the Mamanguape River Estuary include pollution from solid waste deposits, domestic sewage, industries, and public service activities, as well as severe changes in water circulation, intensive shrimp aquaculture, sugarcane plantations, and deforestation and disorderly occupation [31]. Estuarine landscapes can be formed by multiple habitat types that support diverse flora and fauna [32,33,34,35]. Among them, seagrass beds, just like mangroves and salt marshes, are considered essential nursery areas for fishes [36,37,38]. This is due to their structural complexity, which provides numerous refuge areas [39,40,41,42], as well as resource availability that facilitates species coexistence and increases both richness and abundance of individuals [43,44,45]. However, we cannot underestimate the importance of unvegetated habitats [46], since they can also be highly productive [29,47] and function as nursery, feeding, and refuge grounds for a variety of organisms.

In this study, the diets of four co-occurring species found in vegetated and unvegetated areas of the estuary of the Mamanguape River were studied to test the hypothesis of individual specialization: the Brazilian silverside Atherinella brasiliensis (Quoy & Gaimard 1825), the mojarra Eucinostomus argenteus Baird & Girard 1855, and the anchovies Anchoa tricolor (Spix & Agassiz 1829) and Lycengraulis grossidens (Spix & Agassiz 1829). These species were collected in greater abundance in this estuary [48]. They are common in estuaries and, except for A. brasiliensis, are economically important [49,50,51,52,53]. In this context, this study aimed to analyze the trophic ecology and niche breadth of these fish populations, assess how different vegetated and unvegetated habitats influence their diet, and investigate whether individual variation in the use of food resources occurs within these populations.

2. Materials and Methods

2.1. Study Area

The Mamanguape River Estuary (6°43′02″ S, 35°67′46″ W) lies on the northern coast of Paraíba, Brazil (Figure 1), within the Barra de Mamanguape Environmental Protection Area. This estuary stretches for about 24 km, with channel depths ranging from 1 to 5 m and a maximum width of roughly 2.5 km near its mouth [54,55]. Sandstone reefs partially close off the mouth, protecting the estuary from incoming wave energy. In the channel systems, intertidal and subtidal seagrass beds are found, formed by the species Halodule wrightii (Ascherson, 1868) [56], Halophila decipiens (Ostenfeld, 1902), and the rare and recently recorded Halophila baillonis (Ascherson ex Dickie, 1874) [57]. The estuary is also surrounded by 6000 hectares of mangrove forest, dominated by Rhizophora mangle, Avicennia spp., Laguncularia racemosa, and Conocarpus erectus [54]. The regional climate is tropical, with only two distinct seasons: rainy and dry [58]. During the collection months, average monthly rainfall ranged from 114.20 mm in the rainy season to 83.43 mm in the dry season, while temperatures varied between 26 and 44 °C [59].

Two areas were selected within the euhaline sector of the estuary. The first vegetated area (Vegetated Area 1) consisted of patches of H. wrightii seagrass. The second vegetated area (Vegetated Area 2) was a sandbank located in the estuary channel, covered by a mix of seagrass and marine macroalgae patches (Appendix A.1). Although seagrass density was a bit higher in Vegetated Area 2, both sites showed signs of disturbance from the methods local collectors use to catch shellfish. The unvegetated area consisted of mudflats in the lower estuary. Vegetated Areas 1 and 2 were about 2–3 km apart from each other, while the distance between the two unvegetated areas was roughly 0.8 km.

2.2. Sampling and Dietary Analysis

Sampling was performed during daylight hours in the rainy season (April, May, August) and dry season (September, November, December) of 2014. A beach seine net (10 m long × 1.5 m high) with mesh sizes of 12 mm in the wings and 8 mm in the central portion was used for collection. The net was hauled parallel to the shoreline over approximately 30 m at a maximum depth of 1.5 m during low tide. The sampling unit consisted of five standardized replicates spaced 50 m apart to reduce interference between successive hauls, aiming to capture individuals using the area for feeding. The design included 120 samples (4 areas × 5 replicates × 6 months). The effective fishing area per site totaled about 300 m². All fish collected were identified, counted, measured (total length in mm), and weighed (g).

The data for diet characterization was collected through stomach content analysis of fish caught within the vegetated and unvegetated areas. Diet composition was assessed by calculating two percentages for each food item, numerical (N) and volumetric (V), according to [61]. Volumes were obtained using the displacement method of [62]. The separated items were evenly flattened between two glass slides, and volume was recorded using a Petri dish divided into 100 grid points. Prey items and fragments that fell onto each grid point were counted, and their volume was subsequently recorded. All food items were identified to the lowest possible taxonomic level. Specimens containing empty stomachs were omitted from the analyses.

For studies on the availability of food resources, zooplankton were collected using a conical–cylindrical plankton net (total length 1.50 m; 60 cm of mouth opening and a mesh net size of 0.2 mm). At each sampling station, horizontal plankton hauls were performed during the day at each site at low tides. Each haul was standardized to a duration of 5 min at a boat speed of 1.5 knots, aiming to minimize fish escape. All samples were immediately stored and preserved in 4% formaldehyde diluted in seawater [63]. In the laboratory, 10 mL subsamples were examined using a stereomicroscope. Organisms were identified to the lowest possible taxonomic level [64] and subsequently counted. The abundance of each zooplankton organism was calculated using the following calculation:

$$A=\left({\displaystyle \frac{\frac{n \times va}{vs}}{V}}\right)\times 2,$$

where n is the number of organisms counted, va is the volume of sample (mL), vs is the volume of the subsample (in mL) and V is the volume of water filtered in the field (L).

2.3. Size Structure

To examine the dietary shifts during ontogeny of the species, size classes were defined by Sturges’s Rule [65]:

$$K=1+3.3\times \log \left(n\right),$$

where K = the number of classes, and n = number of individuals. The total length of the captured individuals ranged from 10 to 132 mm in A. brasiliensis, from 23 to 96 mm in A. tricolor, from 10 to 124 mm in E. argenteus, and from 32 to 165 mm in L. grossidens. From Sturges’s Rule, individuals of these species were separated into approximately ten size classes. However, due to the small number of fish in certain size classes, some of the lower classes and some major classes were combined. Finally, the remaining size classes are described in

Table 1. Size classes of the four species studied of the estuary of the Mamanguape River, Northeastern Brazil, in 2014. TL in mm.

Species	TL1	TL2	TL3	TL4	TL5	TL6	TL7
Anchoa tricolor	≤36	37–43	44–50	51–57	58–64	≥65	-
Atherinella brasiliensis	≤42	43–53	54–64	65–75	76–86	87–97	≥98
Eucinostomus argenteus	≤31	32–42	43–53	54–64	65–75	≥76	-
Lycengraulis grossidens	≤45	46–59	60–73	74–87	88–101	≥102	-

.

2.4. Statistical Analysis

To test for significant differences in fish density (number per 1000 m²) and biomass (grams per 1000 m²) across different areas and sampling periods, we used a permutational analysis of variance (PERMANOVA) [66,67], followed by a posteriori pairwise comparisons tests [68]. We chose two fixed factors for the analysis: a temporal factor (with two levels: dry season and rainy season) and a spatial factor (also with two levels: vegetated and unvegetated areas). The data was transformed and arranged into a Euclidean distance matrix, which allowed us to test the effects of these factors. Statistical significance was assessed using 9999 permutations, following standard PERMANOVA procedures [66,69].

To visually explore the trophic structure of the fish community, we applied non-metric multidimensional scaling (nMDS) to the volumetric data of food items. Before running the ordination, we calculated the Bray–Curtis similarity coefficient following a fourth-root transformation of the data. Following the approach described by [70,71], a matrix of food items was built to reduce the number of samples and facilitate the detection of feeding patterns. Because the dietary data for a single individual contained large numbers of zero values, this led to instabilities in the calculation of similarities at the individual level, which greatly reduced the effectiveness of multivariate analyses of dietary data. This problem was efficiently minimized by averaging the dietary data (%V) and pooling stomach contents from 3 to 5 individuals to produce a new series of replicates for a given factor (i.e., species or season within species). One-way analysis of similarity (ANOSIM) was performed on the similarity matrix to test whether the composition of the diet differed significantly among the fish species. All multivariate analyses were carried out using the software package Primer v6 [68,72].

Feeding selectivity was quantified using Ivlev’s Electivity Index [73], which compares observed trophic habits with the availability of food resources in the environment. This index is commonly employed to determine the degree of predator preference for particular prey items [74,75] and is expressed as

$$E={\displaystyle \frac{r_i-p_i}{r_i+p_i}},$$

where r_i = relative abundance of prey i in the stomachs; and p_i = relative abundance of prey i in the environment. The index ranges from −1 to +1. Negative values indicate avoidance or inaccessibility of a given prey, zero denotes random feeding, and positive values reflect active selection [73,75,76].

The proportional similarity index (PSi) of Schoener [77]) was used to calculate the individual specialization, measuring the overlap between an individual diet and that of the population [5]. The average of the PSi values of all the individuals in a population is the average level of overlap in the diet of individuals and the population as a whole, i.e., the degree of individual specialization (IS) [5,78]. This index ranges from 0, when each individual consumes small subsets of food resources (specialists), to 1, when all individuals consume the full set of food resources used by the R population (generalist) [5]. The program used to make this analysis was the Studio software (version 4.2.2) [79] with the package “RInSp” [80]. The “RInSp” uses a non-parametric procedure of Monte Carlo (9999 replicates) for testing the value of IS against a null model [5]. The PSi values generated for each individual were used to compare the degree of individual specialization between the size classes of species in each study area and at each station through the Kruskal–Wallis test (p < 0.05), performed in R Studio software (version 4.2.2) [79].

Dietary niche breadth was estimated using the Shannon–Wiener index (H′). To evaluate niche overlap among species and across different size classes, we applied the simplified Morisita overlap index [81]. Overlap values exceeding 0.60 were considered biologically significant, following the criteria established by [82,83]. All calculations were conducted using the R package “spaa” [84]. The resulting Morisita index values were visualized as a heatmap generated with R Studio (version 4.2.2) [79].

3. Results

3.1. Spatial and Temporal Distribution

A total of 3855 specimens were collected (1741 to Atherinella brasiliensis, 1068 to Anchoa tricolor, 583 to Eucinostomus argenteus and 463 to Lycengraulis grossidens). Density differed significantly among sites for A. brasiliensis (Pseudo-F_1.119 = 4.4515; p < 0.05), with the highest values being recorded in unvegetated areas both in the rainy (0.31 ± 0.08) and in the dry season (0.17 ± 0.10). The species L. grossidens (Pseudo-F_1.119 = 7.85; p < 0.001) and E. argenteus (Pseudo-F_1.119 = 4.96; p < 0.001) showed differences in relation to the temporal factor (Figure 2). The anchovy L. grossidens reached the highest values during the dry period, while the mojarra E. argenteus showed higher densities in the rainy period (Figure 2).

Regarding biomass, it was observed that only E. argenteus showed significant differences in relation to the temporal factor (Pseudo-F_1.119 = 4.93; p < 0.05), with the highest values being recorded during the rainy season (Figure 2). No significant difference was observed in the other species. However, the anchovies also showed higher biomass during the rainy season (Figure 2), while for A. brasiliensis, the highest values of biomass were found in the vegetated areas during the rainy and in the unvegetated areas during dry period (Figure 2).

3.2. Diet Composition

For the dietary study, we analyzed the stomach contents of 768 individuals of A. brasiliensis, 284 of E. argenteus, 365 of A. tricolor and 382 of L. grossidens. Of these, approximately 2% of the individuals of each species had no food in their stomachs. In general, the Brazilian silversides A. brasiliensis ingested a greater variety of prey (56 items), while the mojarras E. argenteus ingested only 40 different food items (Appendix A.2, Appendix A.3, Appendix A.4, Appendix A.5, Appendix A.6, Appendix A.7, Appendix A.8 and Appendix A.9).

The analysis of the Volumetric Frequency of the dietary items showed an ontogenetic variation in all species in both habitats. The diets of the smallest length classes comprised mainly copepods Cyclopoida and Calanoida, while larger individuals showed preference for most volumetrically important prey, such as Algae, Polychaeta, Caprella and fish (Appendix A.2, Appendix A.3, Appendix A.4, Appendix A.5, Appendix A.6, Appendix A.7, Appendix A.8 and Appendix A.9).

In general, the Brazilian silversides A. brasiliensis ingested primarily copepods (Cyclopoida and Calanoida) in both areas in all seasons. However, these species also ingested Diatoms and Algae in higher abundance in vegetated areas during the rainy period and Diatoms and Cumacea in the dry period. In unvegetated areas, they also consumed a great volume of Algae and Hymenoptera in both the rainy and the dry periods. The diet of A. tricolor during the rainy season was formed primarily by Calanoida, Brachyura Zoea and fish in vegetated areas and Gastropoda Larvae additionally in unvegetated areas. During the dry period, they consumed Calanoida and Cumacea in vegetated areas and Trematoda, Calanoida, Harpacticoida, Brachyura Larvae and fish in unvegetated areas. The mojarra E. argenteus consumed mainly Cyclopoida, Nematoda, sedentary Polychaete and Caprellidae in vegetated areas during the rainy period and Harpacticioda, Caprellidae and sedentary Polychaete during the dry period. In unvegetated areas, the mojarra consumed primarily Cyclopoida, Calanoida, plant material, and sedentary Polychaete during the rainy period and Cyclopoida, Calanoida, plant material, Nematoda, sedentary Polychaete, and tube Polychaete in the dry period. The main items consumed by L. grossidens in the vegetated areas during the rainy season were Trematoda, fish and Brachyura Zoea. In the dry season, they consumed mainly Calanoida, Cumacea and fish. In the unvegetated areas, the anchovy L. grossidens ingested mainly fish in the rainy and Calanoida, Harpacticoida, Brachyura Zoea, fish, and plant material during dry period (Appendix A.2, Appendix A.3, Appendix A.4, Appendix A.5, Appendix A.6, Appendix A.7, Appendix A.8 and Appendix A.9).

4. Ontogenetic Shifts

In vegetated areas, the diet of the first size classes (TL1 to TL3) in all species consisted predominantly of zooplankton (Cyclopoida, Calanoida and Brachyura zoea) during both seasons (Figure 3). The contribution made by Caprellidae to the diet of E. argenteus remained at TL6, whereas that of Brachyura Zoe was to the same length class of A. tricolor. In the larger classes (TL4 to TL6), the contribution of Polychaeta and fish rose alongside the corresponding increase in size in E. argenteus and L. grossidens, respectively; increases in these items were observed during two seasons (Figure 3).

In unvegetated areas, zooplankton organisms were the most important food items for the diet of the smallest classes for all species (TL1 to TL3) in both seasons (Figure 4). For A. brasiliensis, Hymnoptera and Algae also made the greatest contributions to the volume of this species, whereas the proportions of fish, in the case of L. grossidens, remained high in TL2 and TL3 (Figure 4). For the larger individuals (TL4 to TL6), the contributions made by zooplankton items to the diet of A. brasiliensis, E. argenteus and L. grossidens declined markedly, whereas the proportions of Hymnoptera, Polychaeta and fish increased, respectively (Figure 4). In contrast, A. tricolor consistently ingested relatively large proportions of zooplankton, principally Brachyura Zoe and Gastropoda Larvae.

When the means of the volumetric contributions of the different dietary categories in each dietary sample of each species are subjected to ordination, the nMDS plot shows separation among the species. While A. brasiliensis and E. argenteus are grouped in the upper part of the diagram due to their similar diets, samples A. tricolor and L. grossidens, which showed preference for zooplankton items, lie on the lower side of the diagram (Figure 5). The global test of ANOSIM indicated a significant difference between species (R = 0.449, p < 0.01).

4.1. Feeding Selectivity

Based on the Ivlev Index, in vegetated areas, A. brasiliensis and E. argenteus positively selected Caprella and Cumacea (Figure 6), whereas A. tricolor and L. grossidens positively selected y Peneidae Zoe, Brachyura Zoe and Cyprid during both seasons (Figure 6).

In the unvegetated areas in both seasons, the zooplankton items showed positive selection by A. tricolor and L. grossidens, with Brachyura zoea and Cumacea exhibiting the highest positive electivities (Figure 7). In the case of A. brasiliensis and E. argenteus, similar results were seen for positive selection in both vegetated and unvegetated areas, with prevalence in the selection of epiphytic habit organisms, such as Caprellidae (Figure 6).

Figure 6. Food electivity (Ivlev’s Index) of Anchoa tricolor, Atherinella brasiliensis, Eucinostomus argenteus and Lycengraulis grossidens for different types of prey (see Table 2) in the vegetated areas of the estuary of the Mamanguape River, Northeastern Brazil, during the rainy and dry periods of 2014. The dashed line represents the X-axis.

Figure 6. Food electivity (Ivlev’s Index) of Anchoa tricolor, Atherinella brasiliensis, Eucinostomus argenteus and Lycengraulis grossidens for different types of prey (see Table 2) in the vegetated areas of the estuary of the Mamanguape River, Northeastern Brazil, during the rainy and dry periods of 2014. The dashed line represents the X-axis.

Figure 7. Food electivity (Ivlev’s Index) of Anchoa tricolor, Atherinella brasiliensis, Eucinostomus argenteus and Lycengraulis grossidens for different types of prey (see Table 2) in the unvegetated areas of the estuary of the Mamanguape River, Northeastern Brazil, during the rainy and dry periods of 2014. The dashed line represents the X-axis.

Figure 7. Food electivity (Ivlev’s Index) of Anchoa tricolor, Atherinella brasiliensis, Eucinostomus argenteus and Lycengraulis grossidens for different types of prey (see Table 2) in the unvegetated areas of the estuary of the Mamanguape River, Northeastern Brazil, during the rainy and dry periods of 2014. The dashed line represents the X-axis.

Table 2. Relative abundance of planktonic organisms in vegetated and unvegetated areas of the estuary of the Mamanguape River, Northeastern Brazil, during the rainy and dry seasons of 2014.

	Vegetated Areas		Unvegetated Areas
Prey (Abbreviation)	Rainy	Dry	Rainy	Dry
Foraminifera (FRM)	0.24	0.77	0.04	0.02
Diatomaceous (DIA)	5.93	1.34	5.52	9.80
Centric diatomaceous (DCE)	2.34	0.71	1.63	0.79
Pennate diatomaceous (DPE)	-	-	-	0.30
Invertebrate egg (EIN)	0.03	0.17	0.14	-
Nematoda (NEM)	0.13	-	-	0.03
Polychaeta Larvae (LPO)	0.68	0.37	0.02	0.02
Polychaeta (POL)	0.05	0.01	0.01	0.01
Cyclopoida (CYC)	10.31	9.96	15.22	17.89
Calanoida (CAL)	42.56	67.25	45.09	43.18
Harpacticoida (HAR)	15.87	6.48	14.17	11.00
Cyprid (CYP)	0.26	0.26	0.15	0.06
Nauplius (NAU)	5.71	5.61	2.21	7.81
Penaeidae zoea (ZPE)	0.21	0.03	0.34	0.06
Penaeidae (PEN)	-	-	-	0.01
Brachyuran zoea (ZBR)	11.49	3.43	12.58	3.34
Brachyuran antizoea (ABR)	-	-	-	0.02
Brachyuran megalopa (MBR)	-	-	-	0.01
Porcellanidae zoea (ZPO)	-	-	-	0.05
Mysis (MYS)	0.09	0.03	0.05	0.07
Mysida (MYA)	0.07	0.28	0.26	0.29
Cumacea (CUM)	0.38	0.07	0.06	0.03
Isopoda (ISO)	0.07	-	0.01	-
Gammaridae (GAM)	0.05	0.08	0.01	-
Caprellidae (CAP)	0.10	0.06	0.15	0.02
Amphipoda (AMP)	0.30	0.62	0.09	0.02
Ostracoda (OST)	0.21	0.51	0.07	0.04
Ceratopogonidae larvae (LCE)	0.02	-	-	0.01
Ceratopogonidae pupa (PCE)	0.02	-	-	-
Diptera (DIP)	-	-	-	0.02
Hymenoptera (HYM)	0.02	-	-	0.01
Gastropoda larvae (LGA)	0.14	0.31	0.79	4.60
Bivalvia larvae (LBI)	-	-	0.19	0.02
Lophophorata (LOF)	0.21	0.03	0.08	-
Chaetognatha (CHA)	0.44	0.27	0.72	0.08
Fish eggs (EFI)	1.11	0.27	0.07	0.30

Table 2. Relative abundance of planktonic organisms in vegetated and unvegetated areas of the estuary of the Mamanguape River, Northeastern Brazil, during the rainy and dry seasons of 2014.

	Vegetated Areas		Unvegetated Areas
Prey (Abbreviation)	Rainy	Dry	Rainy	Dry
Foraminifera (FRM)	0.24	0.77	0.04	0.02
Diatomaceous (DIA)	5.93	1.34	5.52	9.80
Centric diatomaceous (DCE)	2.34	0.71	1.63	0.79
Pennate diatomaceous (DPE)	-	-	-	0.30
Invertebrate egg (EIN)	0.03	0.17	0.14	-
Nematoda (NEM)	0.13	-	-	0.03
Polychaeta Larvae (LPO)	0.68	0.37	0.02	0.02
Polychaeta (POL)	0.05	0.01	0.01	0.01
Cyclopoida (CYC)	10.31	9.96	15.22	17.89
Calanoida (CAL)	42.56	67.25	45.09	43.18
Harpacticoida (HAR)	15.87	6.48	14.17	11.00
Cyprid (CYP)	0.26	0.26	0.15	0.06
Nauplius (NAU)	5.71	5.61	2.21	7.81
Penaeidae zoea (ZPE)	0.21	0.03	0.34	0.06
Penaeidae (PEN)	-	-	-	0.01
Brachyuran zoea (ZBR)	11.49	3.43	12.58	3.34
Brachyuran antizoea (ABR)	-	-	-	0.02
Brachyuran megalopa (MBR)	-	-	-	0.01
Porcellanidae zoea (ZPO)	-	-	-	0.05
Mysis (MYS)	0.09	0.03	0.05	0.07
Mysida (MYA)	0.07	0.28	0.26	0.29
Cumacea (CUM)	0.38	0.07	0.06	0.03
Isopoda (ISO)	0.07	-	0.01	-
Gammaridae (GAM)	0.05	0.08	0.01	-
Caprellidae (CAP)	0.10	0.06	0.15	0.02
Amphipoda (AMP)	0.30	0.62	0.09	0.02
Ostracoda (OST)	0.21	0.51	0.07	0.04
Ceratopogonidae larvae (LCE)	0.02	-	-	0.01
Ceratopogonidae pupa (PCE)	0.02	-	-	-
Diptera (DIP)	-	-	-	0.02
Hymenoptera (HYM)	0.02	-	-	0.01
Gastropoda larvae (LGA)	0.14	0.31	0.79	4.60
Bivalvia larvae (LBI)	-	-	0.19	0.02
Lophophorata (LOF)	0.21	0.03	0.08	-
Chaetognatha (CHA)	0.44	0.27	0.72	0.08
Fish eggs (EFI)	1.11	0.27	0.07	0.30

4.2. Individual Specialization

There was evidence of significant individual specializations in relation to the taxa consumed for all species in both studied areas in both seasons (

Table 3. Individual specialization index (IS) for Atherinella brasiliensis, Eucinostomus argenteus, Anchoa tricolor and Lycengraulis grossidens in vegetated and unvegetated areas of the estuary of the Mamanguape River, Northeastern Brazil, during the rainy and dry seasons of 2014. * p < 0.0001 (Monte Carlo; 9999 simulations).

	Vegetated Areas				Unvegetated Areas
	Rainy		Dry		Rainy		Dry
Species	IS	n	IS	n	IS	n	IS	n
Atherinella brasiliensis	0.190 *	221	0.285 *	110	0.206 *	191	0.244 *	184
Eucinostomus argenteus	0.281 *	173	0.307 *	14	0.272 *	71	0.225 *	16
Anchoa tricolor	0.302 *	81	0.265 *	93	0.368 *	138	0.282 *	32
Lycengraulis grossidens	0.399 *	11	0.252 *	142	0.298 *	64	0.222 *	144

). The individuals of A. brasiliensis reached a higher degree of specialization (lower IS value) during the rainy season, while anchovies A. tricolor and L. grossidens were more specialized in the dry season (Table 3). The mojarra E. argenteus presented lower SI values in the unvegetated areas in both climatic seasons (Table 3). In relation to ontogenetic variation, it can be observed that most individuals presented a high degree of individual specialization, since most of the values of the proportional similarity index (PSi) are below 0.5 (Figure 8).

4.3. Niche Breadth and Dietary Overlap

The niche breadth values (H′) of species varied from 0.33 to 0.98 (

Table 4. Niche breadth values (H′) of the species analyzed in vegetated and unvegetated areas of the estuary of the Mamanguape River, Northeastern Brazil, during rainy and dry seasons of 2014.

	Vegetated Areas		Unvegetated Areas
	Rainy	Dry	Rainy	Dry
Species	H′	H′	H′	H′
Atherinella brasiliensis	0.63	0.64	0.49	0.43
Eucinostomus argenteus	0.49	0.98	0.67	0.72
Anchoa tricolor	0.76	0.55	0.73	0.89
Lycengraulis grossidens	0.33	0.48	0.34	0.55

). The Brazilian silversides A. brasiliensis and the mojarra E. argenteus presented the highest values in the vegetated areas, while the anchovy A. tricolor obtained a larger niche in the unvegetated areas (Table 4). The anchovy L. grossidens had the lowest values of niche breadth in relation to the other species in both habitats (Table 4). Significant statistical difference was found in niche breadth among areas for A. tricolor (Pseudo-F_1.354 = 8.525; p = 0.004) and A. brasiliensis (Pseudo-F_1.746 = 28.005; p = 0.0001) and seasons for E. argenteus (Pseudo-F_1.279 = 9.4049; p = 0.002) and L. grossidens (Pseudo-F_1.373 = 7.3684; p = 0.0061).

The analyses of dietary overlap based on Morisita’s index found relatively low values for most pairs of species in two habitats. The highest values of diet overlap occurred between A. brasiliensis and A. tricolor in both vegetated and unvegetated areas, in both rainy and dry periods (Figure 9). The anchovy L. grossidens overlapped its diet with A. tricolor and A. brasiliensis diets in unvegetated areas, and only in the dry season in vegetated areas (Figure 9). The mojarra E. argenteus, most of the time, only overlapped between their own size classes (Figure 9). All overlap values are available in Appendix A.10 and Appendix A.11.

5. Discussion

With the exception of Brazilian silverside A. brasiliensis, our results evidenced a clear ontogenetic change in the diet of the species studied. While the smaller individuals fed mainly zooplanktonic copepods, the larger ones generally showed preference for more benthic items. In general, this shift from planktonic to benthic feeding behavior is related to the search for nutrients that are more energetically profitable to the predator, and to the morphological alterations during its growth (mouth and body size) [85,86]. The absence of an ontogenetic diet shift in A. brasiliensis may be associated with differences in foraging abilities, reflecting increased efficiency in handling benthic prey (such as Polychaeta, Brachyura, and Gastropoda), combined with changes in swimming ability for prey capture [87]. Furthermore, the body form of Brazilian silversides, particularly the forked caudal fin, enhances maneuverability and mobility, which likely influences feeding performance and may favor a more specialized benthic foraging strategy throughout ontogeny [51]. In the case of E. argenteus, the change in jaw protrability is one of the most important events for success in the ability to use other food resources on substrate, such as polychaetes and molluscs [88]. The study by [2], for example, showed that there is a positive correlation between the total length of the individuals and the gradual increase in the mouth size, thus suggesting that their prey selection is related to the mouth area.

In relation to niche breadth, it was observed that A. brasiliensis, E. argenteus and A. tricolor presented as generalists and only L. grossidens presented as a specialist. Although they presented as generalist species, A. brasiliensis and E. argenteus obtained the lowest values in the individual specialization index (IS), indicating a high degree of intrapopulational variation in their diets. Thus, individuals used only subsets of the total food items recorded for the diet of these species. These co-occurring individuals may be consuming different items due to phenotypic variations that produce interindividual differences in the ability to use alternative prey, due to different physiological requirements or differences in their intraspecific competition abilities [13,89]. However, in the case of A. brasiliensis, the greater availability of food items in unvegetated areas decreased the degree of individual specialization, despite the higher density of this species in these areas. This corroborates with [13]. According to these authors, intraspecific competition can either increase or decrease individual specialization, depending on the pattern of variation in preferred items predicted by the foraging theory.

In unvegetated areas, despite the higher density of A. brasiliensis, we observed a lower degree of individual specialization (IS). This finding may seem counterintuitive, since higher conspecific density typically increases intraspecific competition, which in turn is predicted to increase individual specialization. When resources are abundant and diverse, individuals can opportunistically feed on a wide range of prey without needing to specialize, even under high density [13]. In contrast, competitive pressure would only lead to increased specialization if preferred resources were limited. Thus, our results suggest that in unvegetated areas, food availability is sufficiently high to buffer the intraspecific competitive effects of increased density, allowing individuals to maintain more generalized diets. This interpretation is consistent with the Niche Variation Hypothesis, which predicts that generalist populations exhibit higher individual specialization only when resource diversity is limited. Several studies have demonstrated this positive relationship between the degree of individual specialization and intraspecific competition [10] or the low abundance of resources [90].

In contrast, the anchovy L. grossidens showed a smaller niche breadth; i.e., individuals used a limited number of food resources, reducing the intrapopulational specialization of this species. These results corroborate the hypothesis of the niche variation described by [9], which states that there are more individual variations in populations with higher niche ranges than in populations with narrower niches [91]. This was also observed in populations of Perca fluviatilis (Percidae) in lakes in the south–central part of Sweden [92] and among four populations of frogs (Leptodactylidae) in the Brazilian Cerrado [93], which indicates this may be a general pattern in natural populations. In addition, it was possible to verify that for L. grossidens there was a reduction in individual specialization (increase in PSi values) with the increase in body size, since as they grow, individuals use a smaller variety of food items, becoming more specialized.

Differences in the location of individuals of each species in nMDS also emphasized the spatial variation in the diet of the species. A separation was observed between those fed primarily on the water column (A. tricolor and L. grossidens) and those that had a preference for bottom items (A. brasiliensis and E. argenteus). Although A. brasiliensis and E. argenteus presented as benthic feeders and used a large amount of similar food items, there was little or no significant food overlap among these species. These species, in addition to using different proportions of the same food resource (copepod), selected different items to complement their diet. While one used seaweed, pennate diatomaceous and Cumacea, the other showed great contribution of sedentary/tube polychaete and Caprella in their diet. Thus, the Brazilian silverside A. brasiliensis and the mojarra E. argenteus were able to coexist and feed themselves in the same distribution of the environment through strategies of niche complementarity and partition of resources. These strategies are based on the fact that species that occupy similar positions in a given niche dimension tend to use alternative resources in a way that complements and differentiates their niches [94] and that through the food partition these species coexist using different proportions of the same item, reducing interspecific competition and food overlap between them [95,96].

Similarly, the species that fed on the water column, A. tricolor and L. grossidens, managed to coexist through the differentiation of complementary items and ontogenetic variation in their diets. All size classes of A. tricolor were presented as zooplanktivores, using items from different strata of the water column. However, the anchovy L. grossidens reduces the contribution of zooplanktonic elements present in the diet of the smallest individuals and uses a higher proportion of fish in the diet of the larger individuals [53](. Small morphological differences may be related to the variation in the trade-offs of these species, causing this variation in diet and facilitating coexistence. The importance of the morphological variations in dentition and gill traces has already been observed and related to the diet of engraulids in Venezuela: the piscivorous ones are characterized by great canine teeth and a reduced amount of gill traces, whereas reduced dentition and numerous traces are characteristic of zooplankivorous engraulids [97]. Our results demonstrated decreased food overlap between L. grossidens and A. tricolor during ontogeny. While not directly tested here, we propose that the increasing dentition and gill-raker spacing in L. grossidens reduce its zooplankton feeding capacity, a hypothesis warranting future investigation.

In addition, it can be observed that zooplankton was the most important food resource in feeding species in unvegetated areas. This result had already been observed for unvegetated tidal flats in Japan [98] and on one of the tidal flats studied, where the zooplankton trophic guild was the species richest in both periods of the hydrological regime [31]. In the vegetated areas there was a greater use of microcrustaceans, such as Caprella and Gammaridea, which live on or between the leaves of the seagrass, corroborating the results of [99] that suggested that the biomass of seagrass in the Swartvlei estuary (South Africa) was ineffective at preventing the predation of amphipods by resident fish.

In summary, the greater degree of individual specialization in populations with a larger niche (generalist) than in specialist populations seems to be a standard even for natural populations. However, this specialization did not vary between vegetated and unvegetated areas, due to the fact that the seagrass beds used in this study are formed by small patches (sparse seagrass meadows span approximately 400 m in length and 30 m in width) [100], which reduces the complexity of this mesohabitat when compared to more extensive and/or formed banks of larger phanerogam species. Thus, interindividual variations in populations appear to have been influenced by differences in food availability and the magnitude of intraspecific competition, rather than by the dynamics between vegetated and unvegetated areas of this estuary.