Introduction
Understanding the processes and mechanisms controlling a species’ trophic niche and potential consequences for coexistence of phylogenetically, morphologically or functionally similar species is a long-standing problem in ecology (Schoener, Reference Schoener1974; Layman & Winemiller, Reference Layman and Winemiller2005; Andrade et al., Reference Andrade, Fitzgerald, Winemiller, Barbosa and Giarrizzo2019). Niche overlap occurs when two or more species share, to some extent, the same realized niche (Hutchinson, Reference Hutchinson1957). If niche overlap is low or resources (e.g. food) are abundant, species will coexist without competition, but high niche overlap and limited resource availability will promote competition (Pianka, Reference Pianka1974; Giller, Reference Giller1984). Food resources used by a species (i.e. the trophic niche) represent one of the most studied dimensions of the niche (Bearhop et al., Reference Bearhop, Adams, Waldron, Fuller and Macleod2004). However, the myriad of factors affecting the trophic niche are complex due the spatiotemporal variations in food availability, prey–predator interactions and consumer trophic plasticity (Giller, Reference Giller1984; Park et al., Reference Park, Gaston and Williamson2016; Silva et al., Reference Silva, Gubiani, Neves and Delariva2017).
Food partitioning among species with inter-specific differences in morphological traits or through space and time is often considered an adaptation to minimize competition (e.g. for fishes: Alexandrou et al., Reference Alexandrou, Oliveira, Maillard, McGill, Newton, Creer and Taylor2011; Cardona, Reference Cardona, Crosetti and Blaber2015; Rohan & Buckley, Reference Rohan and Buckley2018). Several studies in distinct ecosystems revealed marked variation in food partitioning patterns among fishes coexisting in the same habitat (e.g. Correa & Winemiller, Reference Correa and Winemiller2014; Juncos et al., Reference Juncos, Milano, Macchi and Vigliano2015; Andrade et al., Reference Andrade, Fitzgerald, Winemiller, Barbosa and Giarrizzo2019; Malinowski et al., Reference Malinowski, Cavin, Chanton, Chasar, Coleman and Koenig2019). For instance, prior work on food partitioning in iliophagous mullet species (i.e. feeding mostly on microorganisms associated with fine sediment and debris; Dualiby, Reference Dualiby1988; Vieira, Reference Vieira1991; Cardona, Reference Cardona, Crosetti and Blaber2015) has shown trophic niche segregation in some estuaries (Le Loc'h et al., Reference Le Loc'h, Durand, Diop and Panfili2015; Garcia et al., Reference Garcia, Garcia, Vollrath, Schneck, Silva, Marchetti and Vieira2018), but trophic niche overlap in others (Cardona, Reference Cardona2001). Such discrepancies may be associated with differences in habitat characteristics, between-site variations in food availability, as well as technical differences employed to describe diets.
An interesting model to investigate use of food resources by sympatric species in coastal ecosystems is mullet species inhabiting estuaries. For example, the mullets Mugil curema Valenciennes, 1836 and Mugil liza Valenciennes, 1836 are highly euryhaline fishes that occur in marine, estuarine and freshwater ecosystems along tropical, subtropical and temperate systems (Crosetti & Blaber, Reference Crosetti and Blaber2015; Nelson et al., Reference Nelson, Grande and Wilson2016). These species are abundant in temperate and subtropical estuaries, comprising a major part of the fish assemblage in shallow areas (Garcia et al., Reference Garcia, Vieira, Winemiller and Grimm2004; Possamai et al., Reference Possamai, Vieira, Grimm and Garcia2018; Vieira et al., Reference Vieira, Román-Robles, Rodrigues, Ramos and Santos2019). They are marine estuarine-dependent (Elliott et al., Reference Elliott, Whitfield, Potter, Blaber, Cyrus, Nordlie and Harison2007) species that usually spawn in the sea whereas juveniles use shallow estuarine areas as nursery and feeding areas (Vieira, Reference Vieira1991; Lemos et al., Reference Lemos, Varela, Schwingel, Muelbert and Vieira2014; Garcia et al., Reference Garcia, Garcia, Vollrath, Schneck, Silva, Marchetti and Vieira2018; Mai et al., Reference Mai, Albuquerque, Lemos, Schwingel, Ceni, Saint'pierre and Vieira2019). Juveniles of mullets typically change feeding habits from planktonic to iliophagous (Blaber & Whitfield, Reference Blaber and Whitfield1977; Dualiby, Reference Dualiby1988), and this planktonic-benthic diet shift usually occurs between 20 and 30 mm total length (TL) depending on the species (Cardona, Reference Cardona, Crosetti and Blaber2015). In Patos Lagoon estuary in southern Brazil, juveniles of M. curema and M. liza consume microalgae, foraminifers and microcrustaceans and both recruit into Patos Lagoon estuary at less than 35 mm TL, suggesting they may shift feeding habits from planktivore in the marine adjacent surf-zone to iliophagous in the estuary benthic habitats (Vieira, Reference Vieira1991).
Studies of food partitioning among iliophagous species are relatively scarce, probably due to technical difficulties in describing detailed diet composition and food assimilation patterns (Cardona, Reference Cardona, Crosetti and Blaber2015). A study carried out at Tramandaí-Armazém estuary (South Brazil) compared the diet of two mullet species using stable isotopes analysis and stomach contents techniques and observed different niche overlaps depending on the habitat occupied by both species (Garcia et al., Reference Garcia, Garcia, Vollrath, Schneck, Silva, Marchetti and Vieira2018). However, although their work provided relevant evidence on the trophic niche of these iliophagous species, technical aspects hinder their conclusion. For example, the study had a low number of stomach content samples and lacked temporal replicates (i.e. it was only a diet snapshot based on a single field campaign). Their isotopic analysis was restricted to consumers’ variability and did not include food sources and isotope mixing model analysis, which are essential for a comprehensive investigation of the trophic niche (Phillips et al., Reference Phillips, Inger, Bearhop, Jackson, Moore, Parnell, Semmens and Ward2014). This last caveat was probably due to the difficulty to obtain pure samples of food items (microorganisms, especially microalgae) in an adequate amount to allow the determination of isotopic composition. Instead of analysing the isotopic composition of particular food items, prior studies had analysed particulate organic matter in suspension (POM) or in the sediment (SOM) as proxies for phytoplankton and phytobenthos, respectively, considering only single large (1.2–300 μm) pools of organic matter (Faye et al., Reference Faye, de Morais, Raffray, Sadio, Thiaw and Le Loc'h2011; Le Loc'h et al., Reference Le Loc'h, Durand, Diop and Panfili2015; Carassou et al., Reference Carassou, Whitfield, Moyo and Richoux2017; Cicala et al., Reference Cicala, Calizza, Careddu, Fiorentino, Caputi, Rossi and Costantini2019). This approach, however, does not allow differentiation among particle size classes such as picoplankton, nanoplankton and microplankton (Sieburth et al., Reference Sieburth, Smetacek and Lenz1978). The aggregation of a broad range of particle sizes into a single source pool prevents detection of possible resource partitioning based on particle sizes, which are potentially composed by distinct microalgae communities.
Aiming to advance the knowledge of iliophagous food habits and evaluate diet shifts associated with estuarine recruitment and trophic niche overlap between M. curema and M. liza, we used a technique to separate the particles size fractions of organic matter and estimated their isotopic values. We also compared the use of food resources between juveniles of the mullets M. curema and M. liza in marine and estuarine habitats of a subtropical coastal system. Based on the fact that mullet species can exhibit differences in gill rakers morphology (Konan et al., Reference Konan, Adepo-Gourene, Konan and Gourene2014; Cardona, Reference Cardona, Crosetti and Blaber2015) and these may lead to trophic niche partition by selection of food sizes (Rohan & Buckley, Reference Rohan and Buckley2018), we investigated which size fractions (1.2–20, 20–68 and 68–250 μm) of particulate organic matter were most assimilated in each habitat.
The marine surf-zone is dominated by dense blooms of microalgae (especially diatoms) comprising most of the in situ primary production (Odebrecht et al., Reference Odebrecht, Bergesch, Rörig and Abreu2010), whereas the estuary harbours more diverse phytoplanktonic (Haraguchi et al., Reference Haraguchi, Carstensen, Abreu and Odebrecht2015; Mendes et al., Reference Mendes, Odebrecht, Tavano and Abreu2016) and benthic microalgae assemblages (Coutinho & Seeliger, Reference Coutinho and Seeliger1984; da Silva et al., Reference da Silva, Torgan and Cardoso2010). In this sense, we expected that diet and food partitioning between juvenile mullet species would change across these coastal habitats with contrasting food availability. We hypothesized that with higher availability of diverse size food-resources (estuary), there will be no food-size partitioning between the species, while in the lower availability of diverse size food-resources (marine surf-zone) the species will partition the particle sizes. Moreover, this proposed methodology will allow us for the first time to evaluate isotopically the occurrence of feeding habit shifts from planktonic to iliophagous on mullets.
Materials and methods
Study area
This study was carried at Patos Lagoon estuary and the adjacent marine surf-zone in southern Brazil (Figure 1). Patos Lagoon constitutes the largest choked lagoon (10,360 km2) in the world (Kjerfve, Reference Kjerfve and Wolfe1986) and its estuarine area comprises about 10% of total lagoon area, connected with the ocean through a single inlet about 4 km long and 740 m wide at the mouth (Seeliger & Odebrecht, Reference Seeliger and Odebrecht2010) (Figure 1). Tidal range influence in the estuary is minimal (~0.47 m), and the hydrodynamic of the estuary is driven mainly for the winds, which are predominantly from NE to SW in the region (Möller et al., Reference Möller, Castaing, Salomon and Lazure2001). The shallow waters bottom (<1.5 m) is composed of sand and the channel substrate by sand, silt and clay (Calliari et al., Reference Calliari, Griep and Vieira1977; Ortega et al., Reference Ortega, Ibeiro, Rodrigues, Rodrigues and Dumont2020). Patos Lagoon estuary harbours more than 90 species of benthic macroalgae, including colonial and filamentous cyanobacteria, chlorophytes, phaeophytes, xanthophytes and rhodophytes. The main microalgae are diatoms, cryptophyta, cyanobacteria and dinoflagellates, and their temporal distributions may vary seasonally and also due to changes in abiotic factors such as salinity (Seeliger et al., Reference Seeliger, Odebrecht and Castello1997).
Patos Lagoon and its estuarine zone in southern Brazil. Black triangles in the inset figure denote sampling sites in the estuary (Estu1, Estu2) and black dots denote the adjacent marine surf-zone (Mar1, Mar2).
The adjacent marine surf-zone is characterized by an extensive sandy coastline of ~220 km with dissipative beaches directly exposed to waves with medium to high energy. The main autotrophs in this ecosystem are microalgae that are responsible for high in situ phytoplankton production and blooms dominated by frequent and dense diatoms, mainly Asterionellopsis guyunusae (Odebrecht et al., Reference Odebrecht, Bergesch, Rörig and Abreu2010, Reference Odebrecht, Preez, Abreu and Campbell2013; Franco et al., Reference Franco, They, Canani, Maggioni and Odebrecht2016). Centric diatoms and dinoflagellates are important components of the phytoplankton particularly during summer (Seeliger et al., Reference Seeliger, Odebrecht and Castello1997). Therefore, in contrast with the estuary, this marine surf-zone is characterized by the dominance of phytoplankton and the absence of other autotrophs such as aquatic macrophytes, macroalgae beds and seagrasses (Seeliger et al., Reference Seeliger, Odebrecht and Castello1997).
Field collections and sample processing
Juvenile mullets (M. curema and M. liza) were sampled monthly during the austral summer (January to March) of 2018 in four sampling stations with depth <1.5 m: two inside the estuary (Estu1, Estu2) and two in the adjacent marine surf-zone (Mar1, Mar2) (Figure 1). This season was chosen based on prior work showing that both juvenile mullet species co-occur in this estuary and its adjacent marine shallow area only during summer (Vieira, Reference Vieira1991; Rodrigues et al., Reference Rodrigues, Cabral and Vieira2015). Mai et al. (Reference Mai, Santos, Lemos and Vieira2018) demonstrated through chemical analysis of otoliths that both species spawn in marine waters. However, some specimens of M. curema can spend their entire life cycle in seawater, others have sporadic entries in brackish or fresh water and finally there are some individuals who remain in brackish water for a longer period of time (Mai et al., Reference Mai, Santos, Lemos and Vieira2018). The species M. liza, on the other hand, seems to be more dependent on the estuary, as they spend most of their life cycle in fresh and/or brackish water (Mai et al., Reference Mai, Santos, Lemos and Vieira2018). The mullet species reach the first maturity for both sexes with an average size of 408.3 mm for Mugil liza and 248.6 mm for Mugil curema (Fernandez & Dias, Reference Fernandez and Dias2013; Lemos et al., Reference Lemos, Varela, Schwingel, Muelbert and Vieira2014). Fishes were sampled using a 9 m beach seine (13 mm bar mesh in the wings and 0.5 mm mesh in the 3 m centre section) that was pulled to cover an area of about 60 m2 during each haul (Garcia et al., Reference Garcia, Vieira and Winemiller2001).
Stable isotopic analysis
Ten individuals measuring between 23 and 45 mm TL of each species (M. curema: 33.9 ± 3.0 mm and M. liza: 30.4 ± 5.8 mm TL) at each sampling station were collected in March 2018. The exemplars were euthanized with eugenol (0.4 ml/l of eugenol solution 1:10 in ethanol 70%) for subsequent collection of muscle tissue for analysis of carbon and nitrogen stable isotope ratios (δ13C and δ15N, respectively). This anaesthetic was used because it does not significantly alter stable isotopic composition (Nahon et al., Reference Nahon, Séité, Kolasinski, Aguirre and Geurden2017). Only the individuals collected in March were used in the SIA because their isotopic turnover reflects food assimilation in the prior two months of sampling (Oliveira et al., Reference Oliveira, Mont'Alverne, Sampaio, Tesser, Ramos and Garcia2017) and thus the time that both species co-occur in the study system.
In order to quantify potential diet shifts, different size classes of particulate organic matter in suspension (POM) and in the sediment (SOM) were sampled and used as proxies for basal food sources (planktonic and benthonic microalgae) consumed by juvenile mullets. One sub-sample of each SOM size fraction was analysed under the microscope to ensure that the samples contained microphytobenthos. Three samples of each size fraction of POM and SOM were obtained monthly between January and March 2018 at each sampling station. The POM fractions were obtained by filtering water through a sequence of filters with decreasing porosity. Initially, water was passed through a 250 μm pore sieve to retain and subsequently discard larger zooplankton and coarse detritus (e.g. sediment, leaves). The remaining water was filtered in sequence into 68 μm and 20 μm pore sieves and finally a pre-combusted (450 °C for 4 h) 1.2 μm Whatman glass-fibre filter. The material obtained in the 68 μm and 20 μm pore sieves and the 1.2 μm filter was placed in Petri dishes, sealed and stored on ice until their transport to the laboratory. These procedures resulted in three fractions of POM with different size classes: 1.2–20 μm, 20–68 μm and 68–250 μm. A similar procedure was applied to obtain size fractions of SOM (i.e. microphytobenthos). Initially, the upper 2 cm from the sediment was collected using a plastic core (10 cm diameter) at each sampling station and stored on ice until processing. In the laboratory, the sediment was moved to a clean plastic bottle, partially filled with distilled water and manually mixed. After sedimentation of sand grains, the supernatant was collected and filtered into the sieves and glass-fibre filter following the same procedures described for POM. The samples were not acidified because the sediments of Patos Lagoon estuary are poor in carbonates and previous studies did not find differences in the carbon isotopic composition of acidified samples compared with non-acidified (Claudino et al., Reference Claudino, Abreu and Garcia2013).
All samples for SIA were processed in the laboratory following standard procedures (Garcia et al., Reference Garcia, Hoeinghaus, Vieira and Winemiller2007; Hoeinghaus et al., Reference Hoeinghaus, Costa, Garcia, Bemvenuti, Vieira and Winemiller2011). Briefly, muscle tissue was dissected from the anterior-dorsal region of each individual fish, rinsed with distilled water, and dried in sterile Petri dishes in an oven at 60 °C to constant weight (minimum of 48 h). Similarly, POM and SOM size fractions, including the filter for the smallest size fraction, were dried in sterile Petri dishes to constant weight. Dried samples with the exception of filters were ground to a fine powder with a mortar and pestle and stored in clean Eppendorf tubes. Subsamples were weighed (~1 mg for animal tissues and 25–30 mg for SOM and POM) and pressed into ultra-pure tin capsules (Costech Analytical Technologies, Valencia, CA, USA). Isotopic analyses were conducted at the Stable Isotope Ecology Laboratory at the University of North Texas using continuous flow elemental analyser isotope ratio mass spectrometry (EA-IRMS) in a system comprised by a Thermo Flash 2000 EA, ConFlo IV interface and Delta V Advantage IRMS. Modified single-point normalization (equivalent to two-point and multi-point normalization methods; Paul et al., Reference Paul, Skrzypek and Fórizs2007; Carter & Barwick, Reference Carter and Barwick2011) was used to normalize preliminary data, based on quantified and known values of two certified reference materials (USGS 40 and USGS 62) analysed with each sequence. Reference materials were chosen such that their known values approximately bracket the estimated ranges of C and N isotope values of samples being analysed. Stable isotope values are reported as parts per thousand (‰) differences from corresponding international standards VPDB (Vienna PeeDee Belemnite) and air for carbon and nitrogen, respectively: δX = [(R sample/R standard) − 1] × 103, where R = 13C/12C or 15N/14N. Standard deviations for δ13C and δ15N of replicate analyses of lab standards analysed with each sequence were 0.13‰ and 0.07‰ for animal, and 0.08‰ and 0.24‰ for plant, respectively.
Stomach content analysis
For the stomach content analyses (SCA), five individuals ranging from 25–50 mm of total length of each species (M. curema: 41.4 ± 5.7 mm and M. liza: 32.8 ± 8.5 mm TL) at each sampling station in January and February 2018 were randomly selected from the individuals collected. In the laboratory, stomachs were removed and fixed in 10% formaldehyde for four days then stored in 70% alcohol until analysis. Determination of stomach contents was based on subsamples. For each individual, the contents of the stomach were removed, mixed with ethanol and stirred. Following procedures in Cardona (Reference Cardona, Crosetti and Blaber2015) and Garcia et al. (Reference Garcia, Garcia, Vollrath, Schneck, Silva, Marchetti and Vieira2018), one aliquot was sampled with a micropipette and put into a Fuchs Rosenthal chamber (0.2 mm deep with a grid of 16 quadrats of 1.0 mm2 each) where microalgae and zooplankton were identified and counted at 400× magnification using a light microscope. Items were counted until reaching at least 200 individuals (cells, colonies or filaments) in each sample. In case of not reaching 200 individuals in the first aliquot, an additional aliquot was analysed up to the maximum of three aliquots or until reaching the established limit of counted individuals. Additional subsamples were acid cleaned and mounted on glass slides using Naphrax™ (Brunel Microscopes Ltd, Chippenham, UK) in toluene and examined at 1000× under a light microscope (Biggs & Kilroy, Reference Biggs and Kilroy2000) to identify diatoms. Microalgae and zooplankton were identified to the lowest practical taxonomic level based on specialized literature.
Data analyses
SIA – sources contribution to the mullets
Initially, a two-way ANOVA was used to test for differences in average δ13C and δ15N among species (M. curema and M. liza) and habitats (Mar1, Mar2, Estu1, Estu2), followed by a Tukey (HSD) test for post-hoc comparisons. Biological data were tested for normality (Shapiro–Wilk test) and homogeneity of variances (Cochran test). This analysis did not reveal differences in δ13C or δ15N for either species between the two sampling stations within the marine surf-zone or within the estuary (P-value >0.05). Therefore, subsequent SIA were carried out considering only marine vs estuary habitats without specific locations within each.
Biplots of δ13C and δ15N were used to first examine between-habitat patterns in isotopic composition of consumers (M. curema and M. liza) and basal food sources (Peterson & Fry, Reference Peterson and Fry1987). A Bayesian isotope mixing model (SIMMR; Stable Isotope Mixing Models in R; Parnell, Reference Parnell2016) was used to estimate relative contributions of different size fractions of POM and SOM (1.2–20 μm, 20–68 μm and 68–250 μm) assimilated by each species in each habitat. This method employs Gaussian likelihood and fits the model to the data via Markov chain Monte Carlo (MCMC). For each mixing model, 500,000 iterations were run, followed by a burn-in and thinning of 50,000 and 5000, respectively. Models were run with uninformative priors. The trophic enrichment factor (TEF) values used were 2.1 ± 0.7 for carbon and 3.8 ± 0.4 for nitrogen, estimated specifically for juvenile mullet M. liza through controlled diet experiments (Oliveira et al., Reference Oliveira, Mont'Alverne, Sampaio, Tesser, Ramos and Garcia2017). In the absence of similar experimental data for its congeneric species, the same TEF values for M. curema were applied in the Bayesian mixing models. The nitrogen content in the POM and SOM samples was relatively low and close to the detection limit of the Isotope Ratio Mass Spectrometry (IRMS), therefore one additional standard deviation was added to the instrumental precision of δ15N values obtained for each fraction of POM and SOM to cope with this potential analytical variability.
A fundamental assumption of isotope mixing models is that the isotopic variability of consumers, after accounting for fractionation corrections, is contained within the variability of assimilated food sources (Phillips et al., Reference Phillips, Inger, Bearhop, Jackson, Moore, Parnell, Semmens and Ward2014). This assumption was evaluated using isotope mixing polygon simulations, which quantitatively establish boundaries of possible source values in the δ13C-δ15N biplot space that can explain the isotopic variability of consumers (Smith et al., Reference Smith, Mazumder, Suthers and Taylor2013). These simulations were run with the packages sp (Pebesma & Bivand, Reference Pebesma and Bivand2005; Bivand et al., Reference Bivand, Pebesma and Gomez-Rubio2013) and splancs (Rowlingson & Diggle, Reference Rowlingson and Diggle2017) in R (R Core Team, 2019). Simulations were conducted for each habitat separately and those individual consumers located outside the 95% mixing polygon region (Supplementary Figure S1), which indicate they could not be confidently explained by the food sources (Phillips et al., Reference Phillips, Inger, Bearhop, Jackson, Moore, Parnell, Semmens and Ward2014), were omitted from the subsequent Bayesian isotope mixing model analyses. For the surf-zone, the majority of individuals could not be explained by local basal food sources in the polygon simulation. In this sense, we added the POM offshore as an additional source in the Bayesian Mixing Model once it meets the marine surf-zone polygons. These samples (N = 6) were obtained during spring 2014 between 5 and 150 nautical miles off the coast (E. Secchi & S. Botta, unpublished data; Possamai et al., Reference Possamai, Hoeinghaus, Odebrecht, Abreu, Moraes, Santos and Garcia2020). Although offshore food sources were not collected in the same season and year of the fish samplings, the use of these data is justified by the lower variation in isotopic values of marine POM compared with marine surf-zone POM (Bouillon et al., Reference Bouillon, Connolly and Gillikin2011; Rosli et al., Reference Rosli, Muhammad and Fadhullah2017; Garcia et al., Reference Garcia, Oliveira, Odebrecht, Colling, Vieira, Rodrigues and Bastos2019).
SIA – isotopic niche overlap
Isotopic niche overlap was evaluated by the area and overlap among standardized isotopic ellipses (SEAc) in the bivariate δ13C–δ15N space for mullet species at each habitat using the SIBER (Stable Isotope Bayesian Ellipses, Jackson et al., Reference Jackson, Inger, Parnell and Bearhop2011) package in R. Overlap in SEAC between juvenile mullet species in each habitat was expressed as a proportion of the sum of the non-overlapping areas of the ellipses, where values range from zero (ellipses are completely distinct) to 1 (ellipses are identical) (Jackson, Reference Jackson2019). Although not directly equivalent to trophic niche metrics based on stomach contents analyses (Bearhop et al., Reference Bearhop, Adams, Waldron, Fuller and Macleod2004; Newsome et al., Reference Newsome, Martinez Del Rio and Phillips2007; Hoeinghaus & Zeug, Reference Hoeinghaus and Zeug2008; Hette-Tronquart, Reference Hette-Tronquart2019), isotopic ellipses can be useful as a proxy to compare amplitude and overlap among species’ trophic niches (Jackson et al., Reference Jackson, Inger, Parnell and Bearhop2011).
SCA – mullets’ diet composition and trophic niche overlap
The comparison of species’ diet was performed by a Permutational Analysis of Variance (PERMANOVA) for each habitat separately, using the vegan package in R software. The abundance matrix of diet items was used and the α = 0.05. In order to express the main food items consumed by each species, the Numeric Frequency was calculated by FN% = (Ni/Nt) × 100, where Ni is the total abundance of item i and Nt is the total abundance of items.
Trophic niche overlap between juveniles of the two mullet species was based on stomach contents data for each habitat and was analysed using the Morisita–Horn index:
$$\hat{{\rm C}}_{\rm H} = \displaystyle{{2\mathop \sum \nolimits_i^n p_{ij}p_{ik}} \over {\mathop \sum \nolimits_i^n p_{ij}^2 + \mathop \sum \nolimits_i^n p_{ik}^2 }}, \;$$where ĈH is the diet niche overlap between species j and k, pij is the proportion of resource i from the total resources used by species j, pik is the proportion of resource i from the total resources used by species k, n is the total number of resources. This index varies from 0 to 1, with higher values showing higher dietary niche overlap (Krebs, Reference Krebs1998). The Morisita–Horn index was calculated using the divo package in R software, calculating 500 bootstrap values and confidence interval of 95% (Erazmus et al., Reference Erazmus, Figueras, Luiselli and Burke2018).
The trophic niche breadth of each juvenile mullet at each habitat was estimated using Levins’ standardized niche breadth index (Hurlbert, Reference Hurlbert1978) following the formula:
$${\rm Ba} = \displaystyle{{\left({1/\sum p_j^2 } \right)-1} \over {n-1}}$$where Ba is the trophic niche breadth, pj is the proportion of the food item j in the diet of species i and n is the number of food items. This index ranges from 0 to 1, where 0 is an extreme specialist and 1 an extreme generalist. The trophic niche breadth was calculated with 95% confidence intervals using the Ecological Methodology program (Krebs, Reference Krebs1998).
Results
Food assimilation and isotopic niche inferred by SIA
The range of δ13C values in the estuary was from −16.9 to −10.9‰ for M. curema and from −17.0 to −11.7‰ for M. liza, and in the marine surf-zone was −19.2 to −15.9‰ and −22.9 to −19.3‰, respectively (Figure 2, Table 1). For δ15N values, the ranges in the estuary were 9.2–11.7‰ for M. curema and 9.4–12.1‰ for M. liza, and in the marine surf-zone was 8.6–13.2‰ and 7.0–12.3‰, respectively. In relation to size fractions of POM and SOM, wider range and lower overlap among average δ13C and δ15N values were observed in the marine surf-zone than in the estuary (Figure 2). Average δ13C values were significantly different across habitats (F = 45.359, df = 3, P <0.001) and species (F = 9.347, df = 1, P = 0.004), but the interaction was not significant (F = 2.239, df = 3, P = 0.102) (Figure 3). In contrast, average δ15N did not vary across habitats (F = 0.698, df = 3, P = 0.560) or species (F = 1.263, df = 1, P = 0.269) (Figure 3).
Carbon (δ13C) and nitrogen (δ15N) isotopic values of the mullets Mugil curema (open/blue circles) and M. liza (grey/red circles) and average (± SD) values of the different size fractions (1.2–20, 20–68, 68–250 μm) of particulate organic matter in suspension (POM: P) and in the sediment (SOM: S) in the marine surf-zone (upper panel) and in the estuary (lower panel).
Comparison between carbon (δ13C) (upper panel) and nitrogen (δ15N) (lower panel) isotope values of the mullets M. curema and M. liza in the marine surf-zone (Mar1, Mar2) and in the estuary (Estu1, Estu2). Different letters (a, b) denote statistically significant differences (Tukey post-hoc test, P >0.05) between average values.
Number of samples (n) and average values (±SD) of carbon (δ13C) and nitrogen (δ15N) stable isotope values of consumers (Mugil curema and M. liza) and the different size fractions (1.2–20, 20–68, 68–250 μm) of particulate organic matter in suspension (POM) and in the sediment (SOM) in the marine surf-zone and in Patos Lagoon estuary
Bayesian mixing models revealed that offshore POM was the most assimilated basal food source at the marine surf-zone by both M. curema (Median = 60%; 95% Credibility Interval = 7–77%) and M. liza (88%; 74–95%) (Figure 4). In contrast, the largest size fraction (68–250 μm) of SOM was the most assimilated in the estuary by M. curema (48%; 3–80%) and M. liza (43%; 2–77%) (Figure 4).
Relative assimilation (Median: 50%; Credibility interval: 95%) of the different size fractions (1.2–20, 20–68, 68–250 μm) of particulate organic matter in suspension (POM: P) and in the sediment (SOM: S) by juvenile mullets Mugil curema and M. liza of the marine surf-zone (upper panel) and the estuary (lower panel). An additional food source (POM offshore) was included in the mixing models in the marine surf-zone (see Materials and methods for details).
The size and overlap of isotopic ellipses (SEAc) of both species differed across habitats (Table 2). In the marine surf-zone, M. curema presented lower SEAc than M. liza (4.11 and 7.62, respectively), whereas in the estuary both species showed similar values (4.73 and 4.85, respectively). The proportional overlap between isotopic ellipses of both species was higher in the estuary (64.7%) compared with the marine surf-zone (0.7%) (Table 2).
Niche width and niche overlap of M. curema and M. liza in Patos Lagoon estuary and adjacent marine surf-zone, South Brazil, estimated by both stable isotopes and stomach contents techniques
Isotopic niche width (SEAc: ‰2), trophic niche breadth (Ba), isotopic niche overlap (INO) and trophic niche overlap (CH).
Diet composition and trophic niche inferred by SCA
Stomach content analyses (SCA) of both mullet species revealed diets composed of microalgae such as Bacillariophyceae, Coscinodiscophyceae, Mediophyceae, Chlorophyceae, Cyanophyceae, Euglenophyceae, Zygnematophyceae, and also fragments of zooplankton (Supplementary Table S1).
The comparison of the diet between M. curema and M. liza showed different results depending on the habitat. In the marine area, the species presented distinct diets (F = 4.829, df = 1, P < 0.001); however, in the estuary, the diet was similar (F = 1.308, df = 1, P = 0.226). In the marine surf-zone, M. curema feed mainly on Asterionellopsis guyunusae (Bacillariophyceae), while M. liza consumed mainly Coscinodiscophyceae (fragments) (Supplementary Figure S2). Nevertheless, although in the estuary both species fed on a greater diversity of algae, the four more consumed algae were the same for these two species: Bacillariophyceae spp., Cylindrotheca closterium, Nitzschia sp. 4 and Skeletonema spp., together comprising 40.8% of the M. curema diet and 39.9% of the M. liza diet (Supplementary Figure S2). Moreover, these four algae species were also consumed in similar proportions.
Trophic niche overlap (CH) of M. curema and M. liza showed similar results between marine surf-zone (mean = 0.86; 95% Credibility Interval = 0.69–0.97) and estuary (mean = 0.83; 95% Credibility Interval = 0.76–0.90) (Table 2). Trophic niche breadth (Ba) did not vary between species within the same habitat, but lower niche breadth was observed for both species in the marine surf-zone (M. curema = 0.03 and M. liza = 0.06) compared with the estuarine zone (M. curema = 0.14 and M. liza = 0.16) (Table 2).
Discussion
Assimilated diet and isotopic niche overlap inferred by SIA
Stable isotope analyses based on different size fractions (1.2–20, 20–68 and 68–250 μm) of organic matter revealed the preference of both species for larger particles in the estuary. Compared with previous studies (e.g. Faye et al., Reference Faye, de Morais, Raffray, Sadio, Thiaw and Le Loc'h2011; Le Loc'h et al., Reference Le Loc'h, Durand, Diop and Panfili2015; Carassou et al., Reference Carassou, Whitfield, Moyo and Richoux2017; Cicala et al., Reference Cicala, Calizza, Careddu, Fiorentino, Caputi, Rossi and Costantini2019) that analysed bulk organic matter that included all size fractions, our analyses provide new isotopic evidence supporting the hypothesis that juvenile mullets exhibit particle preferences during feeding. Higher assimilation of SOM with size particles between 68 and 250 μm suggests a feeding preference for microphytobenthos (20–200 μm), compared with smaller cell sizes such as nanophytobenthos (2–20 μm) and picophytobenthos (0.2–2 μm).
SIA also revealed a diet shift between juvenile mullets caught in the marine surf-zone and inside the estuary. Particulate organic matter in suspension (POM) was the most assimilated dietary item by juvenile mullets in the marine surf-zone, suggesting planktonic feeding in the water column. In contrast, the most assimilated item inside the estuary was SOM, which suggests iliophagous feeding behaviour (i.e. consuming microorganisms associated with fine sediment and detritus). This diet shift has already been described for other Mugilidae species (Cardona, Reference Cardona, Crosetti and Blaber2015) and has also been suggested for M. liza based on inspection of stomach contents (Vieira, Reference Vieira1991). The present work is the first to provide isotopic evidence of the planktonic–benthic diet shift for M. curema and M. liza during their recruitment from the marine surf-zone into the estuary. Finally, it is worth noting that comparison of food niche partitioning between congeneric species in different types of habitat could be partially confounded by potential differences in feeding preference simply related to ontogeny. In order to minimize this potential effect, we analysed only juvenile forms with similar body size ranges (25–50 TL mm) in both habitats. Nevertheless, future experimental studies could investigate if diet changes in juvenile mullets may occur between early developmental stages compared with later ones regardless of the resource diversity available.
Analysis of isotopic niche overlap of mullet species in the marine (0.07%) and estuarine (64.7%) habitats support our initial hypothesis of higher resource partitioning in the habitat with lower resource diversity. Interestingly, a preliminary study carried out in another subtropical system (Tramandaí-Armazém estuary) comparing the same species, found an opposite pattern, with isotopic niche overlap in the marine habitat, and resource partitioning in the estuary (Garcia et al., Reference Garcia, Garcia, Vollrath, Schneck, Silva, Marchetti and Vieira2018). This discrepancy could be related to differences between estuaries in food composition and availability and time of arrival of the juvenile mullets in the marine habitat. For example, individuals recently arrived in the marine habitat had less time to achieve isotopic equilibrium with local food sources in the current work compared with Garcia et al. (Reference Garcia, Garcia, Vollrath, Schneck, Silva, Marchetti and Vieira2018), which may have influenced the isotopic niche overlap. In future studies, it would be interesting to perform isotopic analysis of tissues with a faster turnover rate than muscle tissue (e.g. blood, plasma, liver), because they reflect what was consumed in a shorter period of time (Thomas & Crowther, Reference Thomas and Crowther2015; Barton et al., Reference Barton, Litvin, Vollenweider, Heintz, Norcross and Boswell2019). This would increase the likelihood of juvenile mullets reflecting food items eaten in the habitat in which they were captured. However, the comparison with this prior work is limited because they did not analyse isotopic composition of food sources. In contrast, the differentiation of the organic matter size fractions used in the present study revealed differences in food habits between M. curema and M. liza, which were not possible in other works that analysed the organic matter pool as a single larger (1.2–300 μm) fraction (Faye et al., Reference Faye, de Morais, Raffray, Sadio, Thiaw and Le Loc'h2011; Le Loc'h et al., Reference Le Loc'h, Durand, Diop and Panfili2015; Carassou et al., Reference Carassou, Whitfield, Moyo and Richoux2017; Cicala et al., Reference Cicala, Calizza, Careddu, Fiorentino, Caputi, Rossi and Costantini2019).
In the current study, the isotopic composition of juvenile mullets did not isotopically match the autochthonous organic matter pools in the marine surf-zone. Rather, Bayesian mixing models indicated that the juvenile mullets sampled in the marine surf-zone reflected assimilation of ocean-derived organic matter. This trophic connection between the marine surf-zone and the offshore ocean is plausible considering the life cycle of these mullet species. These marine estuarine-dependent species spawn in offshore ocean waters and their larvae and recruits migrate towards shallow coastal zones, such as marine beaches and estuaries (Vieira, Reference Vieira1991; Lemos et al., Reference Lemos, Varela, Schwingel, Muelbert and Vieira2014). Therefore, it is reasonable to suggest that the juvenile mullets found in the surf-zone with isotopic composition distinct from local autotrophic sources are oceanic-derived individuals who recently recruited into the surf-zone. As newly arrived individuals, they would not have enough time (~2 months according to Oliveira et al., Reference Oliveira, Mont'Alverne, Sampaio, Tesser, Ramos and Garcia2017) to achieve isotopic equilibrium with local food sources, and still reflect isotopically their original oceanic habitat.
The high isotopic niche overlap found in the estuary for the juvenile mullets does not necessarily imply competition. High biomass and diversity of microalgae in the studied estuary suggest a lack of food limitation and, therefore, absence of intra- and inter-specific competition for food resources in juvenile mullets. Other factors that minimize strength of competition and promote coexistence between these mullet species include high spatiotemporal variability of microalgae assemblages in this estuary (Haraguchi et al., Reference Haraguchi, Carstensen, Abreu and Odebrecht2015; Mendes et al., Reference Mendes, Odebrecht, Tavano and Abreu2016) and trophic plasticity frequently observed in fishes, including mullets (Gerking, Reference Gerking1994; Cardona, Reference Cardona, Crosetti and Blaber2015; Garcia et al., Reference Garcia, Garcia, Vollrath, Schneck, Silva, Marchetti and Vieira2018). Prior research suggested that these factors can suppress competition even in conditions of resource limitation (Cardona, Reference Cardona2001; Park et al., Reference Park, Gaston and Williamson2016; Silva et al., Reference Silva, Gubiani, Neves and Delariva2017). For instance, Garcia et al. (Reference Garcia, Garcia, Vollrath, Schneck, Silva, Marchetti and Vieira2018) attributed the trophic niche segregation between juveniles of M. curema and M. liza in the Tramandaí-Armazém estuary (southern Brazil) not to resource limitation and competition, but rather, to exploitation of microhabitats with different depths and/or particle size profiles harbouring distinct microalgae assemblages. Several studies of feeding apparatus anatomy in mullets have shown differences among species in the gap size between gill rakers that could lead to differential retention of particles (e.g. microalgae cells) (Guinea & Fernandez, Reference Guinea and Fernandez1992; Konan et al., Reference Konan, Adepo-Gourene, Konan and Gourene2014; Cardona, Reference Cardona, Crosetti and Blaber2015; Menezes et al., Reference Menezes, Nirchio, Oliveira and Siccharamirez2015).
Diet composition and niche overlap inferred by SCA
Diet composition of juveniles of both mullet species was dominated by microalgae and showed higher diversity in the estuary than in the marine surf-zone, coinciding with the high diversity and spatiotemporal variability in the estuarine microalgae assemblage (Odebrecht et al., Reference Odebrecht, Bergesch, Rörig and Abreu2010; Haraguchi et al., Reference Haraguchi, Carstensen, Abreu and Odebrecht2015; Mendes et al., Reference Mendes, Odebrecht, Tavano and Abreu2016). According to the Morisita–Horn index, the diet of both species showed a high overlap in both habitats. However, the PERMANOVA model demonstrated that the diet of both species was different in the marine surf zone, but it was not in the estuary, agreeing with our initial hypothesis that partitioning may occur in less resource-diverse environments.
Diatoms are the predominant microalgae throughout the year in both the marine surf-zone and estuary (Haraguchi et al., Reference Haraguchi, Carstensen, Abreu and Odebrecht2015; Mendes et al., Reference Mendes, Odebrecht, Tavano and Abreu2016) and distinct diatom groups dominated the stomach contents of both species. Although diatoms are ubiquitous in both habitats, the marine area is largely dominated by dense (109 cells l−1) blooms of the surf-zone diatom Asterionellopsis guyunusae; it is a very difficult environment due to the waves' energy, and few species are adapted to these conditions (Odebrecht et al., Reference Odebrecht, Bergesch, Rörig and Abreu2010; Franco et al., Reference Franco, They, Canani, Maggioni and Odebrecht2016; Mendes et al., Reference Mendes, Odebrecht, Tavano and Abreu2016). In contrast, microalgal diversity is much higher in the estuary; in addition to diatoms, cyanobacteria, green algae and cryptophytes are important components of planktonic and benthic habitats (Mendes et al., Reference Mendes, Odebrecht, Tavano and Abreu2016). Estuarine microalgal assemblages are characterized by marked variation in species composition and abundance at both short (hours, weeks) and long (seasons, years) time scales (Haraguchi et al., Reference Haraguchi, Carstensen, Abreu and Odebrecht2015). That variation is largely attributed to environmental variability in salinity, temperature, nutrients, freshwater inflow and water transparency (Haraguchi et al., Reference Haraguchi, Carstensen, Abreu and Odebrecht2015; Mendes et al., Reference Mendes, Odebrecht, Tavano and Abreu2016). This high variability in species composition and abundance of microalgal assemblages is likely the main driver of higher trophic niche width of M. curema and M. liza juveniles in the estuary (0.14 and 0.16, respectively) compared with the adjacent marine habitat (0.03 and 0.06, respectively).
Although the marine surf-zone is less diverse and dominated by one microalgae species (Odebrecht et al., Reference Odebrecht, Bergesch, Rörig and Abreu2010), the two mullet species did not prey mainly on the Asterionellopsis guyunusae microalgae; instead, Mugil liza fed mainly on Coscinodiscophyceae (fragments). Asterionellopsis guyunusae was also an important item in the M. liza diet, however, it was the second most consumed by this species, while for M. curema it was the dominant consumed algae. This difference in the predominance of each algae in the stomach contents can show a degree of the partitioning of the resources, in that each mullet feeds on a specific prey, avoiding trophic overlapping in this less resource-diverse habitat. However, the Morisita–Horn index demonstrated a high similarity in the diet of both species in the marine surf zone, this may have occurred due to the high sensitivity to more abundant species (Chao et al., Reference Chao, Chazdon, Colwell and Shen2006). Therefore, as seen above, the mullet species consume the microalgae Asterionellopsis guyunusae and Coscinodiscophyceae (fragments), and they represent about 80% of their diet, but each mullet has a preference for one of these food resources. And the Morisita–Horn index, in addition to being insensitive to rare species (Chao et al., Reference Chao, Chazdon, Colwell and Shen2006), also did not detect this preference for the abundant food items present in the diet of both mullets.
Concerning the diet of the mullets in the estuary, a factor that contributes to higher diet diversity and niche width in this environment is greater microhabitat heterogeneity in the estuary compared with the marine surf-zone habitat (Seeliger et al., Reference Seeliger, Odebrecht and Castello1997). Due to this heterogeneity of habitats, and the nutrients input provided by the continent, the abundance and diversity of the phytoplankton in the estuary is higher than the marine surf-zone. With a high offer of food resources, the consumers did not need to partition resources, and this may be what is happening with these mullets in the estuary, since their diets were very diverse, but did not differ.
Integrating SCA, SIA and other diet tracer approaches
The results from both SIA and SCA showed that in the less resource-diverse environment (marine zone) the trophic niche overlap was lower than in the most resource-diverse environment (estuary). The diet of both species was different in the marine zone, but was not in the estuary, agreeing with our initial hypothesis that partitioning may occur in a less resource-diverse environment compared with a more diverse one. This difference was possible to observe due to the SCA, as opposed to the SIA that did not show this pattern. Using SIA we were able to describe the shift in the mullets' diet when they changed habitats, but as this analysis doesn't have the refinement of identifying each species used as a resource, we could not distinguish the differences in the resources used. However, when we analysed the trophic niche overlap, the SIA based on isotopic ellipses was more informative that the trophic niche calculated by SCA. So both analyses showed the same pattern, although they showed a different time-lag of resources use.
The SCA is a higher resolution snapshot of what was recently (hours to days) consumed, while the SIA provides time-integrated (weeks to months; Oliveira et al., Reference Oliveira, Mont'Alverne, Sampaio, Tesser, Ramos and Garcia2017) data reflecting dietary items assimilated in consumer tissues according to the isotopic turnover time of each particular tissue type and metabolic pathway (Hesslein et al., Reference Hesslein, Hallard and Ramlal1993; Mont'Alverne et al., Reference Mont'Alverne, Jardine, Pereyra, Oliveira, Medeiros, Sampaio and Garcia2016). The isotopic approach has an advantage over stomach content analysis in estimating the trophic niche overlap because it reflects assimilated food sources over a wider time window of feeding activity, which is particularly relevant considering the marked spatiotemporal variability in the availability of microalgae in the estuary (Haraguchi et al., Reference Haraguchi, Carstensen, Abreu and Odebrecht2015; Mendes et al., Reference Mendes, Odebrecht, Tavano and Abreu2016). Stomach contents analysis is complementary in that it provides a detailed taxonomic description of food items consumed, which is not feasible using SIA due to the lack of substantial isotopic differences among sources and technical difficulties in obtaining samples of microalgae at the species level. Moreover, SIA requires an isotopic equilibrium between sources and consumers, which is not easy to address in migratory organisms, as the mullets are in the present work. Therefore, the use of both approaches together is recommended to better evaluate the intra-specific interactions and clarify the use of resources by the species.
Research on the trophic ecology of iliophagous species, such as mullets, comes with several technical challenges, which could be addressed in part by integrating multiple approaches such as analyses of compound-specific stable isotopes, fatty acid biomarkers and DNA tracers (Majdi et al., Reference Majdi, Hette-Tronquart, Auclair, Bec, Chouvelon, Cognie, Danger, Decottignies, Dessier, Desvilettes, Dubois, Dupuy, Fritsch, Gaucherel, Hedde, Jabot, Lefebvre, Marzloff, Pey, Peyrard, Powolny, Sabbadin, Thebault and Perga2018). Given the costs of those approaches and their own assumptions and limitations, we suggest that combined SIA and SCA approaches, as in this study, are important to provide specific hypotheses for more narrowly targeted research using those approaches. Such further studies to advance our understanding of the trophic ecology of mullets are needed, especially considering their commercial value for fisheries, increasing anthropogenic pressures on their habitats, and their important (yet not fully understood) role in estuarine food web structure and dynamics (Reis & D'incao, Reference Reis and D'incao2000; Oliveira et al., Reference Oliveira, Bastos, Claudino, Assumpção and Garcia2014; Haimovici & Cardoso, Reference Haimovici and Cardoso2017; Santana et al., Reference Santana, Kinas, Miranda, Schwingel, Castello and Vieira2017; Possamai et al., Reference Possamai, Hoeinghaus, Odebrecht, Abreu, Moraes, Santos and Garcia2020).
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0025315421000242
Acknowledgements
We thank the ‘Laboratório de Ecologia e Conservação de Tartarugas e Mamíferos Marinhos’ for providing marine data.
Financial support
This work was funded by the ‘Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)’ through the Long Term Ecological Research Program (PELD – Pesquisas Ecológicas de Longa Duração site 8 – Lagoa dos Patos e região marinha adjacente) (Grant: 441492/2016-9). SV was supported by ‘Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)’ through the Graduate Student Fellowship (code 001). AMG was supported by CNPq through the Research Fellowship (Grant: 309208/2018-1).
