Introduction

Neotropic cormorants (Nannopterum brasilianum) are widely distributed, with breeding populations ranging from Argentina (Olrog 1975; Quintana et al. 2002; Casaux et al. 2012) to south of the United States (Kalmbach and Becker 2005). While a robust body of literature exists on their trophic ecology (Alarcón et al. 2012; Muñoz-Gil et al. 2013; Tette-Pomárico et al. 2020; Harding and Mesler 2022) and breeding colonies (Morrison et al. 1979; Azevedo 1995; Kalmbach and Becker 2005; Alves et al. 2011; Casaux et al. 2012; Quintana et al. 2002), research on these aspects in Brazil remains limited and outdated. On the past two decades, a few studies focused on cormorant diet (Barquete et al. 2008a; Branco et al. 2009; Oliveira et al. 2019) and population abundance of cormorants (Branco 2002; Barquete et al. 2008b; Silva et al. 2014). These abundant aquatic birds are distributed throughout Brazilian territory, with conspicuous populations in Amazonian basin (Silva et al.2014), Pantanal humid areas (Donatelli et al. 2014) and southern coasts of Brazil, such as Lagoa dos Patos (Barquete et al. 2008b) and Santa Catarina State (Branco 2002).

Given their feeding preference for fish and intensified foraging activity, cormorants can potentially impose significant predatory pressure on aquatic fauna and generate conflicts with fishing activities. Abundant cormorants from the northern hemisphere like the double-crested (Nannopterum auritum) and great cormorants (Phalacrocorax carbo) have shown considerable dietary overlap with fisheries, often leading to conflicts with the fishing industry (Carter et al. 1995; Adkins et al. 2014; Ovegård et al. 2017; Harding and Mesler 2022). In the Neotropics, studies have started to explore cormorants interactions with fisheries, although the focus has often been limited to local or isolated populations. Gil-Weir et al. (2011) explored N. brasilianum ecological functions (ecosystem services; bird watching, indicator of fish schools and nitrogen and phosphorus input) and economic values (i.e. cormorants as meat, eggs and guano as fertilizer) in Los Olivitos Estuary in Venezuela suggesting a minor impact on artisanal fisheries. Moreover, the growth of a Neotropic cormorant population in highland lakes in Ecuador has had tangible impacts on fish farmers, which migrating and non-breeding individuals have been observed preying on commercial fish ponds, potentially facilitating the species expansion into these highland regions (Guevara et al. 2011). In Patagonia, Argentina, a diet assessment of Neotropic cormorant identified not only a flexible feeding strategies but also an important regulatory role of two introduced exotic fish species, Oncorhyncus mykiss and Salmo trutta (Alarcón et al. 2012). While these studies emphasize some ecological roles and isolated Neotropic cormorants interactions, ecological conflicts arising from resource competition remain a prevalent issue that demands closer attention.

Such conflicts are typically fueled by the perception of resource competition, with cormorants high fish consumption rates often cited as a central concern (Kameda et al. 2003; Carss et al. 2009; Östman et al. 2013; Aguado-Giménez et al. 2018). Additionally, the conspicuousness of large cormorant flocks amplifies these perceptions, as their visible presence in key fishing areas is often misinterpreted as a direct indicator of population growth and increased pressure on fish stocks. This perceived abundance, coupled with the species high foraging activity, makes them a target of negative perceptions among fishermen, even in the absence of robust population data (Carss 2022; Dorr et al. 2022; Ludwig et al. 2023). Recently, in the southern Brazilian coast, a new cormorant colony near artisanal fisheries has raised concerns about resource competition and the depletion of fish stocks, highlighting the need for data to acknowledge the local management (Pimenta et al. 2024).

To address these concerns and provide a clearer understanding of cormorants diet and fishery interactions, it is essential to assess their dietary habits and quantify their predatory effects on fish stocks. While cormorants diet is relatively easy to assess, evaluating their predatory effects is more complex, as various factors influence fish distribution and assemblages (Ovegård et al. 2021). Thus, various methods may be employed to evaluate cormorants dietary habits, including direct observations, stomach content analysis, and the examination of regurgitated pellets (Barrett et al. 2007). Each one of these methods poses methodological challenges or advantages. Direct observations of cormorants submitting preys and stomach lavage are valid methods, but they are both qualitative data based approaches that represent only a snapshot of feeding process. Regurgitated pellets can generate both qualitative and quantitative data as different preys can be found on pellets (Barquete et al. 2008a; Tette-Pomárico et al. 2020), being non-invasive to the bird, relatively cheap and viable to collect large samples (Carss et al. 1997; Barrett et al. 2007). Despite that, pellets were long discussed to be a biased method to reconstruct size and mass of prey and demands extreme caution in considering prey sizes, as it can be overestimated by smaller prey omission (Carss et al. 1997; Votier et al. 2001).

This study aims to describe the feeding ecology of cormorants and evaluate their trophic interactions with fisheries in the lower portion of the Laguna Estuarine System (LES), Santa Catarina, Brazil. In order to do so, first, we estimated cormorant abundance through monthly counts at a major colony, considering abundances differences between seasons, breeding and non-breeding periods. Second, we identified prey taxonomic composition and sizes to estimate the total fish consumption by cormorants via regurgitated pellets samples. Then, we analyzed niche breadth, feeding strategies, and temporal diet variation to understand their predatory role. At last, we assessed resource overlap between cormorants and artisanal fisheries data to evaluate potential associations and support local future policy decisions.

Materials and methods

Study area

This work was conducted in a major cormorant roosting area located in the lower portion of LES. The roosting area is a semi-open lagoon, known as Noca Lagoon (48°45’27’’S, 28°30’23’’W), partially enclosed by SC-100 highway construction (Fig. 1). The lagoon is connected via water pipes to Santo Antônio Lagoon and the single channel that provides tidal influence from the sea (Barletta et al. 2017; Dantas et al. 2019). An adjacent urbanized area discharges waste directly into the lagoon, while seagrass vegetation dominates much of the lagoon’s area. The cormorant colony, covering an area of 54,000 m2, is situated on a hill predominantly covered by a secondary Atlantic rainforest.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Laguna Estuarine System location map, highlighting South of Brazil and Santa Catarina State. The figure emphasizes the three elliptical lagoons, Tubarão river discharge, the adjacent Camacho Lagoon, highway physical barrier (SC-100), Noca Lagoon and the cormorant roosting site area

Sampling design

Field sampling for both abundance counts and pellet collection was conducted concurrently from July 2021 to May 2023. Abundance data was assessed with two different methods. To determine the number of cormorants in non-breeding seasons (September to March), direct counts were made twice a month at dawn (around 6h00) using 8x42 binoculars and a mechanical counter. Cormorants tend to depart from the roosting site in small amounts (flocks of 10 to 30 birds). The observer was located on road SC-100, about 450 m west from the hill and started counting birds from 20 minutes before dawn until no bird departure occurred in a 30 minutes period. During breeding seasons (April – August), the cormorant colony had an intense traffic of birds (in and outwards) due breeding activities, such as nest construction and chick feeding, making count efforts not viable. So cormorant population was estimated by (1) counting the number of nests (1 pair of cormorants per nest) and (2) adding an estimated juvenile and non-breeding adults population. To estimate non-breeding adults, we conducted counts at perching sites along the estuary, away from the main colony. At these sites, we used plumage differences to calculate the proportion of juveniles and non-breeding adults to breeding adults. Reproductive adults show a blackish color with white filoplumes behind the beak and neck, while juveniles and non-reproductive adults maintain their natural olive-black colors (Sick et al. 1997). This ratio from the perching sites was then applied to the number of breeding adults (from the nest count) to estimate the total number of non-breeding birds, which was added for the final population estimate.

Diet data was assessed via regurgitated pellets, which can be assumed to be produced once per day, depending on physiological conditions and feeding success of cormorants, as an example of captive cormorants pellet production of 0.69 per day (McKay et al. 2003). Pellets were collected once per month during surveys on Noca roosting site, from July 2021 to May 2023. We targeted only fresh (covered in mucous) and undisturbed casks to assure pellets from the same day. During the breeding season, pellets were collected from the outer ring of the colony where non-reproductive adults and juveniles roosted. It was assumed that reproductive adults and non-reproductive individuals have the same food items type.

Pellets were stored frozen until process phase. First we defrosted in ambient water, washed the mucous in running water, while screening prey remnants on a 1 mm, 0.5mm and 63µm stacked sieves. Saggitae and lapillus (for ariidae – Genidens spp.) otoliths were used to count and identify food items. Other key hard structures such as chewing pads, pharyngeal jaws, eye lenses or vertebras were only used if otoliths were absence. We arranged otoliths by taxonomic groups, size and orientation (right and left). The total number of fish for each pellet was assumed to be the maximum number of otolith pairs counted. Fish prey was identified using the Atlas of marine bony fish of Southeastern-Southern Brazil Part I-VIII (Rossi-Wongtschowski et al. 2014; Brenha-Nunes et al. 2016; Siliprandi et al. 2016; Conversani et al. 2017; Giaretta et al. 2017; Santificetur et al. 2017a, b) and Atlas of marine fishes of southern Brazil (Haimovici et al. 2024). Crustaceans exoskeletons (especially penaeids) were not possible to be quantify due its soft and highly digestible nature, but were accounted as present or absent on samples.

Prey size can be underestimated using otolith measurements, as they are susceptible to stomach digestion (McKay et al. 2003; Barrett et al. 2007). To mitigate this methodological bias when applying allometric equations (see Online Resource 1), we used a wear index with five degrees, considering the degree of wear on the otolith edges and the sulcus acusticus (Recchia and Read 1989). The levels were defined as: (01) Intact, otoliths with no wear, extracted directly from fish; (02) Minimal, otoliths with light wear, a visible sulcus acusticus, and well-defined edges; (03) Moderate, with a visible sulcus acusticus and slightly worn edges; (04) High, with a worn sulcus acusticus and completely worn edges; (05) Total, with excessively worn sulcus acusticus and edges. Fish total length (TL) and body mass (WG) were estimated using the mean lengths of right and left otoliths from levels 01, 02, and 03, which accounted for 53.3% of all screened and paired otoliths. Otoliths from Engraulidae species are relatively smaller and more susceptible to digestion, but remained mostly identifiable even at wear level 04. To avoid losing this group taxonomic information, we included level 04 otoliths and grouped these prey items at the family level. Prey total length, body mass and total biomass per pellet were then estimated through allometric regressions from the literature on bony fishes from the southern coast of Brazil (Naves 1999; Carvalho et al. 2014; Haimovici et al. 2024).

Diet analysis

Prey composition was measured via frequency of occurrence (%FO) and numeric frequency (%N):

$$\% FO = \frac{Ni}{N}*100$$
$$\% N = \frac{Ri}{R}*100$$

Where, Ni represents the number of samples which prey i occurred and N the total of samples. Ri, represents the total of prey i within all samples, and R the total number of prey. Prey was also classified according to water column distribution (demersal/benthic or pelagic).

Niche breadth was calculated as Levins (1968) and standardized to a 0 to 1 scale, following (Krebs and Davies 2010):

$${\text{B }} = \frac{1}{{\sum \left( {Ri} \right)^{2} }}$$
$${\text{Bs }} = \frac{{{\text{B }}{-}{ }1}}{{R{ } - { }1}}$$

We delve deeper into feeding strategies by using the graphical method of Amundsen et al. (1996), in which each prey specific-abundance (Pi) is plotted on a diagram with its respective frequencies of occurrence (%FO):

$$Pi = \frac{\sum Si}{{\sum Sti}}* 100$$

Si is the total number of prey i and Sti the total number of preys occurring on samples that prey i is present. The distribution of prey on the plot indicates their relative importance, rarity, and dominance within the diet. For instance, prey with high Pi and %FO is considered dominant, suggesting specialization, while prey with low values is more rare or incidental. The overall prey distribution provides insights into the predator’s feeding strategy, revealing whether it tends toward a generalist or specialist approach, and helps identify intra- and inter-phenotype differences in diet composition, as discussed by Amundsen et al. (1996).

To investigate the significance of monthly variation in the diet of N. brasilianum, we conducted a permutational multivariate analysis of variance (PERMANOVA) with 999 permutations (Anderson 2001). A Hellinger transformation was applied to balance prey abundance data, minimizing the dominance of abundant species and enhancing the representation of rare species, identified with Amundsen’s graphical analysis. As a post hoc analysis, we identified the most significant variations between groups (months) using pairwise p-values (Martinez Arbizu 2020). To further explore and identify food items contribution for specific months, we performed an Indicator Species Analysis (ISA) (De Cáceres 2013).

Fishing production data and diet overlap

Fishing production data were obtained from the Santa Catarina State Monitoring of Fishing Activities Project (PMAP-SC) for the period from July 2021 to June 2023 (PMAP-SC 2024). This project collects monthly sampling data on fishing production from various cities in Santa Catarina. This includes information on the type of fishing gear used, total effort (monthly), species captured, fishing production (in kilograms), and the number of discharges during the specified period. For this analysis, we focused on artisanal fishing types I and II, excluding industrial fishing. Type I involves fishing with or without vessels up to 10 meters, while Type II utilizes vessels larger than 10 meters. Both types were grouped to provide a comprehensive overview of artisanal fishing in the Laguna Estuarine System. We removed the fishing gears used in ocean waters to better explore those employed within the estuarine waters. Since cormorants forage mostly on estuarine waters, we removed ring nets, beach seines, and manual collection (which focus on mollusks as the target species). To compare artisanal fishing and cormorant biomass consumption, monthly and annual biomass consumption was estimated by multiplying cormorant abundance, days of a given month and calculated mean biomass per pellet (Barquete et al. 2008a). To estimate dietary overlap between Neotropic cormorants and artisanal fisheries we utilized Pianka’s Index (Pianka 1973). Seasonal and overall indexes were calculated with consumed biomass proportions, with index values ranging from 0 (no overlap) to 1 (total overlap):

$${\text{O}} = \frac{{\sum {\text{Q}}_{{{\text{ic}}}} {\text{Q}}_{{{\text{if}}}} }}{{\sqrt {\sum {\text{Q}}^{2}_{{{\text{ic}}}} {\text{Q}}^{2}_{{{\text{if}}}} } }}$$

where Qic represents the biomass proportion of prey species i consumed by Neotropic cormorants, and \(Q_{if}\) represents the biomass proportion of prey species i targeted by artisanal fisheries of LES.

Statistical analysis

All statistical analyses were performed using R version 4.3.2 (R Core Team 2024). The abundance data were tested for normality, heteroscedasticity, and dispersion. Given the temporal nature of the count data, we tested for autocorrelation using the acf() function from the ‘tseries’ package (Trapletti et al. 2015). We identified temporal autocorrelation and accounted the discrete structure of the data by employing a Generalized Linear Mixed Model (GLMM) with a negative binomial distribution, using the glmmTMB() function from the ‘glmmTMB’ package (Brooks et al. 2023). Reproductive periods were included on the model as a fixed effect to test for differences on bird counts between breeding and non-breeding periods. We treated the monthly autocorrelation implementing a first order structure nested by year, ‘ar1(month_num + 0 | year)’. Model residuals were then evaluated using the DHARMa package, which confirmed that the temporal autocorrelation was effectively corrected and that the model appropriately accounted for the discrete structure of the data (see Online Resource 2).

To test for temporal diet composition variation, a PERMANOVA was performed using the function adonis2() from ‘Vegan’ package (Oksanen et al. 2022), while the post hoc test was performed with the function pairwise.adonis() from ‘pairwiseAdonis’ package (Martinez Arbizu 2020). The function multipatt() from package ‘Indicspecies” was used to execute the Indicator Species Analysis (De Cáceres 2013). Dietary overlap between cormorants and artisanal fisheries was calculated using the function pianka() from package ‘EcoSimR’ (Gotelli et al. 2013). The significance of the dietary overlap was assessed by comparing the observed values to those generated through randomization, with 1000 iterations performed on the function niche_null_model() (method “pianka” and algorithm “r3”). The observed overlap index was compared to the distribution of simulated indices, and a standardized effect size (SES) was calculated for future comparisons.

Results

Cormorant abundance

Cormorant abundance at Laguna Estuarine System varied throughout months of all sampled years. We identified seasonal tendencies of higher peaks of abundance during autumn and winter (March - August), while a decreased numbers during September and February. We accounted for a monthly mean abundance of 4390 ± 1626 cormorants. Lowest abundance occurred in September 2021 with 1345 cormorant and highest numbers occurred during the breeding period July 2023, an estimate of 5040 reproductive adults and 2308 juveniles and non-breeding adults (not accounting for born chicks) (Fig 2).

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

Observed temporal abundances of Nannopterum brasilianum at Noca Lagoon dormitory, Laguna, Santa Catarina, Brazil. Horizontal arrows represent non-breeding (blue) and breeding periods (red)

The negative binomial GLMM with an AR(1) correlation structure indicated a significant effect of the reproductive season on abundance, with higher counts during the reproductive period (Estimate = 0.44, z = 4.48, p < 0.001; Fig. 3a). Furthermore, the density distribution of counts across seasons (Fig. 3b) highlighted greater abundance during Autumn and Winter. The AR(1) structure effectively addressed temporal autocorrelation, as confirmed by the residual analysis (Fig. 3c). The random effect of year accounted for interannual variability (Variance = 0.043, SD = 0.208).

Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

Overview of the effect of season on abundance a GLMM showing the relationship between seasons and abundance, highlighting higher values during the breeding season (Autumn and Winter). b Density distribution of counts across seasons, illustrating seasonal variations. c Diagnostic plot of Pearson residuals by month for the negative binomial GLMM with an AR(1) correlation structure

Diet composition, prey sizes and feeding strategies

We collected a total of 201 pellets over 19 months, with a mean of 10.15 ± 3.2 samples per month. These samples contained 31 different food items distributed across 17 families. Teleost fish were the dominant prey, occurring in 90% of samples (n = 181), while crustaceans were present in 29.1% (n = 59). In total, 4494 individual prey items were screened, with an average of 16.4 ± 18.6 prey per sample and a maximum of 188 prey in a single pellet.

From the total, 2397 prey items were identified to the lowest possible taxonomic level using a wear index to account for digestive erosion. Demersal and pelagic prey were equally represented in terms of numerical abundance (%N), accounting for 54.9 and 45.1% of the diet, respectively. A notable example of the identification challenges and solutions was the Engraulidae family: from 376 engraulid prey, we identified 188 as Anchovia clupeoides, 31 to the genus Anchoa sp., and four as Anchoviella lepidentostole. The remaining 153 engraulids could not be identified to a lower taxonomic level due to high wear (Table 1).

Table 1 Diet composition of N. brasilianum arranged in numerical (N) order; Frequency of occurrence (%FO), Specific abundance of prey i (Pi), Seasonal abundances of each prey (Win – Winter, Spr – Spring, Sum – Summer, Aut – Autumn); Mean prey total length in mm (TL), Minimum length for fishing capture (Lmin fisheries), Prey feeding habit (demersal or pelagic), Habitat use (E/M – estuarine marine, E – estuarine, M – marine)
Full size table

For 1,554 prey with suitable otoliths, we retrocalculated total length and mass using allometric equations. The estimated average prey size was 117.1 ± 67.7 mm (ranging from 26.6 to 326.9 mm). The mean consumed biomass per pellet was 382.8 ± 555.8 g, ranging from 1.2 g to 979.8 g.

The overall diet niche breadth for N. brasilianum was 0.287. Amundsen’s graphical analysis revealed the prey with the highest prey-specific abundance: Trachinotus spp., Menticirrhus americanus, Urophycis spp., Genidens spp., and Eucinostomus spp. (Fig. 4). These species were distributed across different quadrants of the plot according to their respective frequencies of occurrence.

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

Amundsen’s graphical analysis of feeding strategies for Nannopterum brasilianum at southern coast of Santa Catarina, Brazil. Dots represent the relative position of each prey, classified by numbers. 1- Genidens spp., 2- Engraulidae, 3- Micropogonias furnieri, 4- Eucinostomus spp., 5- A. brasiliensis, 6- M. curema, 7- M. liza, 8 – D. rhombeus, 9- Sardinella brasiliensis, 10- C. spilopterus, 11- Ctenogobius spp., 12- Centropomus spp., 13- P. isoceles, 14- S. rastrifer, 15- M. litorallis, 16- Saurida brasiliensis, 17- Urophycis spp., 18 – e.n.i., 19- Trachinotus spp., 20- M. americanus, 21- P. brasiliensis, 22- Cynoscion spp., 23- Ctenosciane spp., 24- Strongylura spp., 25- Achirus lineatus

Micropogonias furnieri exhibited low prey-specific abundance but high frequency of occurrence, confirming its status as a dominant and constant resource in the diet, alongside Genidens spp. In contrast, Atherinella brasiliensis and Mugil curema showed relatively low values for both metrics, indicating lower overall importance. The remaining prey, located in the bottom left quadrant of the plot, are considered rare or incidental food items.

Temporal diet variation

The PERMANOVA results, based on Hellinger dissimilarity, indicated significant temporal variation in the diet of cormorants (F = 3.305, P < 0.001, perm = 999) (Table 2). Pairwise comparisons revealed 23 significant pairwise dissimilarities among months. The winter months of July and August 2021 were the primary drivers of this variation, collectively involved in 12 of the 23 significant comparisons (July 2021: 7 comparisons; August 2021: 5 comparisons). This indicates that the diet composition during the winter of 2021 was distinctly different from most other periods. Furthermore, September and October of 2022 also showed highly distinct diets, with each being significantly dissimilar to a wide range of other months (see Online resource 4). Indicator Species Analysis (ISA) revealed significant contributions of Genidens spp., M. furnieri and M. curema to temporal variations in cormorant diet thought the whole sampling period. A few other preys contributed significantly on specific months, such as M. liza, engrualids, A. brasiliensis and gerreids (Table 3). Mugil liza was more prominent in warmer months, such as January and February, while Gerreids and A. brasiliensis were associated with a broader range of months but exhibited peaks in cooler months, such as May, June, and July. Additionally, engraulids emerged as significant contributors, particularly during cooler months (July 2022 and May–June 2023), reflecting their availability and potential aggregation during these periods. Interestingly, Saurida brasiliensis showed a significant association with a limited number of sampling groups (e.g., August 2022 and June 2023). This last result likely arises from its occurrence in just two unique and distinct diet samples, where it was the only prey item recorded, thereby inflating its indicator value.

Table 2 Results of PERMANOVA analysis applied to diet composition variation of cormorants in Laguna Estuarine System
Full size table
Table 3 Species Indicator Analysis (ISA). Association between prey abundance and sampling group combination (months)
Full size table

Fishing production and dietary overlap

LES artisanal fisheries targeted 39 different species (excluding Elasmobranchii). The most frequently captured species were Callinectes spp. (36.2%), mullets (24.2%), penaeids (12.9%), Whitemouth croakers (6.8%) and clupeiformes (6.5%) (see Online Resource 3). Over the sampled period, artisanal fisheries extracted more than 13,625 tons of aquatic resources, while cormorants were estimated to consume 1,665 tons. A dietary overlap index of 0.78 was estimated between cormorants and artisanal fisheries, with a mean simulated value of 0.52 (nReps = 1000, P < 0.001). Seasonal indexes ranged from 0.63 on summer 2021/2022 to 0.87 on winter 2023, with a mean value of 0.80 ± 0.08.

Discussion

Our multifaceted results reveal that the Neotropic Cormorant population in the Laguna Estuarine System is both abundant and exhibits an opportunistic feeding strategy, which differs from patterns reported in other studies. This detailed assessment of its diet provides crucial insights into its feeding ecology and resource use, particularly by highlighting the potential for competition with artisanal fisheries. Given that this species remains underexplored in Brazilian waters, our findings help to fill knowledge gaps and establish a baseline for understanding the species ecological roles and mitigating potential new conflicts, as have already occurred in the LES (Pimenta et al. 2024).

Cormorant abundances and fluctuations

Neotropic cormorant population of LES exhibited fluctuations patterns of abundance with higher peaks of individuals during autumn and winter, while a decrease in numbers during summer and spring. This seasonal pattern coincides with the establishment of a reproductive colony in LES from March to August as cormorants distributions can be shifted within breeding and feeding grounds (Barquete et al. 2008a,b; Branco et al. 2009). The establishment of cormorant breeding colonies is perceived to increase bird numbers, enhances their feeding activities (Quintana et al. 2002), raise fish predation and generate ecological conflicts among fisheries activities of Laguna region (Pimenta et al. 2024). Last publication focused on population assessment of N. brasilianum in south of Brazil occurred over 20 years ago in Lagoa dos Patos (Barquete et al. 2008b) and Saco da Fazenda Estuary (Branco 2002), accounting a max population of approximately 1390 and 650 respectively. LES cormorant population fluctuated from approximate 1300 individuals at spring of 2021 to 7800 during 2023 breeding period. Our study not only brings new comparable population numbers, but also the information of an established and conspicuous reproductive colony on southern Brazil, which may contribute significantly to cormorants distribution on Brazilian Southern coasts, Pantanal and Argentina. Newly established colonies and increased cormorant numbers could be an indirect response to climate change, as changes in aquatic habitats influence systems from the individual to the community level and also decrease biodiversity (Wrona et al. 2006; Harley 2011; Rodrigues-Filho et al. 2023), thereby favoring generalist predators such as cormorants. Furthermore, extreme climate change events may also be driving these changes, for instance, by causing higher chick mortality during heat waves (Quintana et al. 2022) or depletion of food availability through severe droughts, since this birds forage on waterbodies (Braz-Mota and Val 2024).

Diet composition and feeding strategies

Neotropic cormorant at LES showed opportunistic behaviors with a diverse diet and fish as main food item. We identified 31 different prey types within 17 taxonomic families. The majority of fish consumption by N. brasilianum is observed in other regions of the species distribution range, including Brazil (Barquete et al. 2008a; Branco et al. 2009; Oliveira et al. 2019), Venezuela (Gil-Weir et al. 2011), Argentina (Alarcón et al. 2012; Casaux et al. 2009, 2012), Colombia (Tette-Pomárico et al. 2020) and United States (Harding and Mesler 2022). Within all identified teleost, Catfishes (Genidens spp.), Engraulids, Whitemouth croakers (M. furnieri), Gerreids (Eucinostomus spp. and D. rhombeus), flounders (C. spilopterus), Brazilian silverside (A. brasiliensis) and mullets (M. liza and M. curema) were previously observed to be abundant species in LES (Sunye et al. 2014; Barletta et al. 2017; Dantas et al. 2019; Frischknecht et al. 2023) and major preys of LES cormorants population.

Neotropic Cormorants are considered a generalist predator due to its wide distribution and plasticity in food resource use. Our niche breadth value of 0.287 for the LES population is substantially higher than values reported for other populations, such as 0.056 in Lagoa dos Patos (Barquete et al. 2008a) and 0.02 in northern Colombia (Tette-Pomaricó et al. 2020). This key difference is not a contradiction but rather a clear reflection of contrasting prey availability across these ecosystems. The extremely low value in Colombia (0.02) indicates a forced specialization, where the diet was dominated by a single family (Ariidae) due to a documented decline in historically important Engraulidae prey populations caused by overfishing (Tette-Pomaricó et al. 2020). The low value in Lagoa dos Patos (0.056) also indicates a specialized diet on a few abundant demersal fish, which together comprised over 94% of local fish catches (Barquete et al. 2008a; Vieira, 1991). In contrast, the Laguna Estuarine System supports a dynamic fish assemblage (Freischneck et al. 2023), providing a wider spectrum of abundant prey throughout the year. This greater resource diversity allows the local cormorant population to exploit a broader niche, resulting in the higher Levin’s index. This pattern strongly supports the concept that the niche breadth in Neotropic cormorants is a plastic indicator of local resource availability rather than a fixed taxonomic trait. As noted by Barquete et al. (2008a), cormorants can prey heavily on abundant resources (a specialist tactic) within an overall generalist and opportunistic framework.

Using the feeding strategy graphical analysis as a qualitative approach (Amundsen et al. 1996), we identified several abundant fish species, such as Catfishes, Whitemouth croakers, Engraulids, Gerreids and Brazilian silverside, consistent with an opportunistic feeding strategy. We also identified a few prey types with high specific abundances but low frequency of occurrence, meaning they were consumed by only a few individuals (Amundsen, 1996) and appeared in a limited number of samples, often as the sole prey type. These preys in Amundsen’s plot are Menticirrhus americanus, Trachinotus sp. and Urophycis sp., being all marine only species. The presence of exclusively marine species may be attributed to intra-specific variations within the cormorant population, as individual birds may exploit different resources based on factors such as prey spatial distribution, individual experience, sex, and age (Ceia and Ramos 2015). Also, cormorants are observed to forage on flocks and individually (Quintana et al. 2004. Oliveira et al. 2019), on shallow ocean waters (Quintana et al. 2004) and estuarine waters (Barquete et al. 2008a; Branco et al. 2009). Although our results indicate a preference for estuarine waters in the LES, cormorants are observed to forage in shallow oceanic waters less frequently (not published).

The presence of pelagic and demersal preys on cormorant diet also indicates an opportunistic behavior which may be explained by the capacity on changing foraging strategies (Quintana et al. 2004), even under high turbidity scores (Grémillet et al. 2012). For instance, cormorants can switch targets during diving sessions (Cosolo et al. 2010), transitioning from predictable, sediment-associated demersal prey (Cook et al. 2006) to dense and conspicuous schools of pelagic fish. Overall, the diverse and adaptable diet of Neotropic cormorants at LES highlights their opportunistic feeding strategy, shaped by prey availability and environmental conditions, reinforcing their role as a key predator in the estuarine ecosystem.

Prey size and daily food intake

Our data estimates a mean prey size of 117.1mm on the diet of N. brasilianum. Similar sizes of 113.4mm and 120 mm were also estimated by Barquete et al. (2008a) and Harding and Mesler (2022), respectively in Lagoa dos Patos, southern Brazil and Arizona, United States. With a prey size ranging from 26 to 326 mm, Neotropic cormorants feeds not only juveniles, but also adult sizes, although cormorant morphology may limit the ingestion of larger fishes (~300mm), as discussed by Barquete et al. (2008a). Additionally, the growth patterns of fish, particularly the time required to reach a size detectable and capturable by piscivorous birds, as well as the period they remain within a consumable size range (i.e. >300mm), are secondary, but important factors that influence their presence on cormorant diet (Riedel et al. 2007). In addition, we also suggest the possibility of cormorants feeding upon discarded heads by fisherman cleaning fish along the estuary, as artisanal fisheries are an intensive activity at Southern State of Santa Catarina (Barletta et al. 2017). By consuming a broad range of sizes, cormorants may regulate fish population density and minimizes competition among fish of various size classes, which could favor growth and surviving of key species by reducing competition (Casaux et al. 2012).

If assumed that one pellet is produced per day, we evaluated a daily food intake of 382 g of fish, a lower number compared to the 425 g per day of Neotropic cormorants at Lagoa dos Patos (Barquete et al. 2008a), 828 g of P. carbo (Grémillet et al. 1996) and many other species around the world (Ridgway 2010). We aimed to reduce size estimation bias by using a worn index, but most of our smaller preys were removed from the analysis, possibly influencing towards a higher mean biomass per pellet (M°Kay et al. 2003; Barret et al. 2007; Ridgway 2010; Tette-Pomárico et al. 2020).

Temporal variation on prey preferences

Estuarine fish assembly is spatially and temporally diverse (species composition and life stages) as abiotic conditions (i.e. tide regimes, salinity, precipitation and water depth) creates different habitats for fish during their life (Henriques et al. 2017; Rodrigues-Filho et al. 2023). For instance, absence of fresh water prey in this study may be explained due cormorants night roosting site, Noca Lagoon, be closer to ocean waters and suffer influences from coastal waters and salinity, key factor that shapes distributions of fishes from Ariidae family (Dantas et al. 2012). For instance, Genidens genidens is an estuary-dependent species and exhibits different distributions and behaviors throughout its life stages (Silva Junior et al. 2013), with juveniles using shallow waters and adults occupying deeper waters (Dantas et al. 2019). The presence of juvenile catfishes (<45mm) in cormorant diet may be associated to fish inexperience (Van Eerden and Carss 2012) and a limited moving capability (Barbieri et al. 1992). Thus, higher population densities of G. genidens occurs after spawning periods, especially in warmer months (spring-summer) (Silva Junior et al. 2013), increasing the species abundance and raising food availability for cormorants, which could explain higher occurrences of catfishes on cormorant diet.

Another important prey to cormorants, gerreids, is a group strongly associated with salinity gradients and abundant in LES. For example, Eucinostomus melanopterus is more common during higher salinity conditions in spring, while E. argenteus is associated with lower salinity scores in autumn (Frischknecht et al. 2023). Our cormorant prey abundance data indicates higher abundances of gerreids during both periods, suggesting that prey availability may influence cormorant diet in specific months.

Biotic elements such as presence of vegetation (e.g., seagrass and mangroves), reproductive patterns, spawning periods, life stages, and migration also play a crucial role in determining fish assembly. Spawning may enhance abundances of Atherinella brasiliensis at post reproductive periods, presenting higher abundances during Summer and Autumn (Cattani et al. 2019). Although this species abundance and resilient presence during the whole year may be explained due higher niche breadth and capability to occur on different habitats (e.i seagrass and mangroves) (Rodrigues-Filho et al. 2024). Also, White mullets (Mugil curema) are opportunistic estuarine resident species in LES and abundant on the region, which are favored by temperatures above of 20 °C and higher salinity scores (Vieira and Vieira 1991; Mai et al. 2018). Another important mullet species in the LES is Mugil liza. This marine migratory species is highly dependent on the estuarine environment during its juvenile phase, with individuals occurring in the system year-round (Garbin et al. 2014; Herbst and Hanazaki 2014; Lemos et al. 2014). However, the fishery primarily targets large, migratory adults that enter the estuary in massive schools during the winter reproductive migration (Daura-Jorge et al. 2013). Both mullets have different biological demands but similar morphological characteristics, which their different temporal variation appearances in cormorant diet may be explained by fluctuations in environmental conditions and their different life history strategies (Meurer and Netto 2007; Frischknecht et al. 2023). Another key factor that influences fish and other aquatic organisms is precipitation, as rain regimes introduce fresh water to the estuarine system shifting the salinity gradient and turbidity (Barletta et al. 2017). The latter factor is essential for prey accessibility as cormorants use visual cues of prey during flights and dives, but in contrast, also influence prey awareness (Strod et al. 2008). For future diet assessments, we suggest subsequent daily sampling efforts to test daily diet composition and associations between precipitation scores and prey diversity in cormorant diet, associated to salinity and fresh water species.

Cormorants interactions with LES artisanal fisheries

The high diet overlap index (0.78) between Neotropic cormorant and artisanal fisheries in the LES suggests strong potential for resources competition. However, a detailed analysis of species composition and temporal dynamics reveals mechanisms that may mitigate this direct competition.

The high overlap is primarily driven by shared key species such as Mugil Liza and Micropogonias furnieri, which account for significant fishery landings of 24.2 and 6.8% of estuarine biomass during the sampling period, respectively (Sunye et al. 2014; Piazza et al. 2021; PMAP-SC 2024). Despite the high numerical overlap, resource partitioning is evidenced by cormorants predominantly preying on M. furnieri juvenils, while fisheries target adults of commercially valuable sizes (size partitioning). For M. liza, niche partitioning occurs on a temporal scale and by fish size. Cormorant predation of juveniles peaks in summer, revealed by Indicator Species Analysis (Table 3), whereas the fishery commercial sizes peaks occurs during the winter which targets a massive migratory adults (>300mm) (Daura-Jorge et al. 2013; Schroeder et al. 2023). In addition, the seasonal shift in cormorant feeding preferences, focusing on other abundant prey like engraulids and gerreids during winter, effectively reduces competition for the key shared resources (M. liza and M. furnieri). The constant presence of Genidens spp. in the cormorant diet across all seasons also contributes to the baseline overlap.

The broader context of the artisanal fishery, which exploits over 39 species but concentrates effort on five key resources (crabs, mullets, penaeids, whitemouth croakers, and clupeiforms, comprising 87% of landed biomass) (Barletta et al. 2017; Sunye et al. 2014; Piazza et al. 2021; PMAP-SC 2024), further explains the mitigated overlap dynamic. Additionally, non-selective fishing methods, such as fyke nets, generate significant bycatch and discards (Garthe et al. 1996; Furness et al. 2007), which likely subsidize the cormorant diet (e.g., crabs) and influences the distribution of many other aquatic birds (Branco 2001; Barbieri 2010; Carniel and Krul 2012). Crucially, the majority of the cormorant’s diet consists of non-target or low-value species (e.g., Genidens spp. at ~0.4% of fishery biomass).

Our data suggests a minimal economical impact of cormorant predation on artisanal fisheries. For that, we considered (1) the size and temporal partitioning of key overlapping species, (2) cormorant predation mostly on non-target species, (3) the natural density-dependent juvenile mortality (Ovegård et al. 2021) and (4) an estimative of cormorant consumption (1,665 tons) representing a 12.2% of fishery extraction (13,625 tons). Cormorants likely play a vital ecological role by preying on abundant juveniles and utilizing discards, thereby contributing to local diversity maintenance and potentially reducing competitive pressures among fish species (Rodrigues-Filho et al. 2024).

Conclusions

This study provides new insights into the trophic ecology of the Neotropic Cormorant and its relationship with artisanal fisheries in southern Brazil. Here we documented seasonal abundance patterns, with the highest population peaks linked to the establishment of a reproductive colony during the austral autumn and winter. Diet assessments revealed a varied use of food resources, dominated by teleost fish and complemented by penaeids, while significant temporal variation underscored the species opportunistic foraging strategy. By delving deep into this multifaceted diet analysis, temporal variation on prey availability and prey sizes are determinants for niche partitioning between cormorants and artisanal fisheries of LES, mitigating direct competition for resources. The data presented here provide a critical base for stakeholders to develop targeted management measures for the Laguna Estuarine System and other similar socio-ecological systems where such predator-fishery dynamics occur.