Resource partitioning among stranded aquatic mammals from Amazon and Northeastern coast of Brazil revealed through Carbon and Nitrogen Stable Isotopes

Aquatic mammals play an important role in community structure. The present study applied stable isotope analysis (SIA) to evidence trophic relationships and resource partitioning among aquatic mammals inhabiting different environments in the Amazon estuarine complex and adjacent coastal zone (AE) and Northeastern coast (NC) of Brazil. In addition, isotopic niche partitioning among Sotalia guianensis, Inia spp. and Trichechus inunguis within the AE was also evaluated, and ecological S. guianensis stocks were characterized. Among marine delphinids, the carbon isotopic composition in offshore species mirrored that of nearshore species, contradicting the pattern of decreasing δ13C values characteristic of many areas around the world including areas in Southeastern and Southern Brazil. Isotopic niches were highly distinct, with no overlap among the assessed species inhabiting the AE. Inia spp. and T. inunguis occupied significantly larger isotopic niche spaces, suggesting high habitat plasticity. S. guianensis inhabited two coastal regions indicating an ecological distinction. Nitrogen values were similar between S. guianensis from the NC and AE, indicating comparable trophic positions. However, NC specimens presented more variable δ13C values compared to those from AE. SIA results also allowed for insights concerning habitat use and the trophic ecology of dolphin species inhabiting different oceanographic regions off Northern/Northeast Brazil. These findings provide novel data on the stable isotope composition for cetaceans and sirenians from this region, and aid in furthering knowledge on the trophic ecology and habitat use of the investigated species.


Material and methods
Study sites. The samples assessed herein were obtained from stranded carcasses recovered along three main sectors: (1) the Amazon lowland (AL); (2) the Amazon estuarine complex and adjacent coastal zones (AE) and, (3) the Northeastern coast of Brazil (NC) (Fig. 1). The Amazon lowland (AL) sector comprises the floodplain habitat named várzea, consisting of clearwater rivers 58 . Specimens from this region were collected in rivers around the Santarém municipality, Ayaya River, a small Amazon River tributary; at Oriximiná, Trombetas River; at Vitória do Xingu, in the lower Xingu River, and near Belém, Guajará Bay.
The Amazon estuarine complex and adjacent coastal zones (AE) covered in this study belong to the Northern and Eastern coast of Marajó Island (i.e., Marajó Bay), as well as the Eastern coast of the state of Pará. Marajó Bay is formed mainly by Pará discharges and the Araguaia-Tocantins River Basin can be considered as the main freshwater source to this bay 59 . This area receives a superficial saline intrusion during low river discharge 60 and undergoes a macrotidal regime, with maximum tides around 4 m on both sides of the bay 61 . The continental Amazon shelf is influenced by factors such as proximity to the Equator, strong tides (semidiurnal tides) and oceanic currents and winds (e.g., North Brazil Current, easterly trades), as well as the substantial discharge of the Amazon River (e.g., water, solutes and particulate materials). The region is considered part of the wet tropics, due to high precipitation rates and temperature 62 .
The Northeastern coast (NC) covers the coastline of the state of Piauí and the easternmost coastline of the state of Maranhão. The Parnaíba River discharge into the Atlantic Ocean forms a delta with five tidally influenced bays, Tutóia, Caju, Melancieiras, Canárias and Igaraçu, that form the Parnaíba Delta 63 . The region comprises a mesotidal coast, with tides ranging from 1.1 m to 3.3 m 64 and vast mangrove forest areas.  Most analyzed bone samples were obtained from the internal portion of the skull using pliers. In the absence of a skull, other available bones (n = 67) were used, following the priority order: 1st caudal vertebrae, 2nd chevrons, 3rd teeth and other bones (e.g., scapula, mandible, flipper). Tooth fragments were removed after demineralization and lipid extraction was performed for the whole tooth. Previous studies reported a difference of 0.2‰ in carbon and 0.3‰ in nitrogen between the teeth and bones of the same individual 65 . Therefore, the isotopic values of the sampled tissues were considered comparable.
Stable isotope analysis. Bone fragments were demineralized by repeated baths in hydrochloric acid (HCl, 0.5N for approximately 72 to 96 h 66 in order to isolate collagen. Each bone fragment (or tooth) was placed in a glass vial covered with acid and stored at 4°C overnight, with acid replacement every 24 h. Samples were subjected to successive baths with distilled water to achieve a neutral pH after a rubber-like flexibility was reached. After this process, lipid extraction was performed by manual rinsing of the samples three times in a 2:1 methanol:chloroform solution, discarding the old solution each time and replacing it with a new one. Samples were then washed in distilled water and dried for at least 24 h at 60°C 56 . Dried samples were then ground to a fine powder using a mortar and pestle. Bone/teeth collagen samples were finally weighed (0.5 to 0.6 mg) in tin capsules (Costech Analytical).
Nitrogen and carbon isotope ratios were measured by Elemental Analyzer Continuous Flow Isotope Ratio Mass Spectrometry in the Center for Stable Isotopes, University of New Mexico using a Costech ECS 4010 Elemental Analyzer coupled to a ThermoFisher Scientific Delta V Advantage mass spectrometer via a CONFLO IV interface. Isotope ratios are reported using the standard delta (δ) notation relative to V-AIR and to Vienna Pee Dee Belemnite (V-PDB), respectively. The three internal laboratory standards are: UNM-CSI Protein std#1, casein purchased from Sigma Aldrich with δ 15 N and δ 13  Data analysis. For T. inunguis (n = 11) and Inia spp. (n = 10) we obtained samples from museum specimens of distinct years (1910′s to 2010), therefore, before pooling samples from different decades, possible temporal trends were tested. An ordinary least squares linear trend performed for both taxa didn't detect significant differences in historical trends for Inia spp. (δ 13 C: r 2 = 0.25, p = 0.13; δ 15 N: r 2 = 0.25, p = 0.13) and for T. inunguis (δ 13 C: r 2 = 0.06, p = 0.45; δ 15 N: r 2 = 0.24, p = 0.12). Before pooling different bones from the same species (i.e., Sotalia guianensis, the only species with enough sampling to perform the test: n = 220), possible differences in stable isotope signatures were assessed using a permutational multivariate analysis of variance test and no differences were detected (pseudo-F = 1.60, p = 0.11). So, we assume that for our data set distinct bone tissue of samples didn't affect results. Due to the small sample size for some species and lack of essential data (i.e., age, standard length, sex), neither sex or maturity stage were considered for the analyses. δ 13 C and δ 15 N data normality and homogeneity of variance were tested by the Shapiro-Wilk and Levene tests, respectively. The Student's t-test was used to compare isotopic values between S. guianensis populations and between marine and freshwater species. A significance level of 0.05 was assumed for all tests.
To investigate isotopic niche variations, species were grouped into Delphinids (all delphinids except S. guianensis), Inia spp. (I. geoffrensis and I. araguaiaensis) and T. inunguis. S. guianensis specimens were grouped according to their sampling sector: Sgui_AE for individuals from AE and Sgui_NC, for specimens from NC. Isotopic niche areas were calculated through standard ellipse areas corrected for small sample sizes (SEAc) using Stable Isotope Bayesian Ellipses (SIBER routine in SIAR package in R 67 ). Overlaps among ellipses were also calculated, in order to quantify the trophic overlap among groups. Probabilities were estimated by Bayesian inference indicating uncertainty and central tendency measures based on permutations presenting 50%, 75% e 95% credibility intervals.
isotopic niche variation. The isotopic niches were highly distinct among the aquatic mammals that inhabit the AE region (T. inunguis, S. guianensis, I. geoffrensis and I. araguaiaensis) with no overlaps (standard ellipse areas, SEAc). In addition, no overlap among the most representative species of the study area (Sotalia guianensis, Inia spp. and Trichechus inunguis) was observed, while a significant overlap among Delphinids and S. guianensis populations was noted, expressing isotope niche partitioning.
The smallest isotopic niche area was occupied by Sgui_NC, followed by Sgui_AE. S. guianensis populations presented a niche area overlap of 68.98% (Fig. 6).

Discussion
The isotopic composition in bone/teeth that integrate a long-term diet were evaluated for the first time in aquatic mammals from the Amazon estuarine complex and Northeastern Brazil. No isotopic overlap was observed among coexisting species within the Amazon estuarine complex and adjacent coastal areas (AE). Indeed, isotopic niches were highly distinct among S. guianensis, Inia spp. and Trichechus inunguis, indicating strong niche segregation among these sympatric species, at least in Marajó Bay. Among marine delphinids, the carbon isotopic composition in the bone/tooth collagen in offshore species (i.e. Grampus griseus, Globicephala macrorhynchus) mirrored that of nearshore species (i.e. Sotalia guianensis), contradicting the pattern of decreasing δ 13 C values characteristic of many areas around the world 56 including areas in Southeastern 68 and Southern Brazil 69 . This pattern can be explained by the huge freshwater discharge into the adjacent oceanic areas that contributes with 13 C-depleted organic material derived from terrestrial and mangrove sources to their dissolved and particulate organic pool 31,70,71 . Therefore, carbon isotopic values in offshore pelagic areas under the influence of the Amazon plume exhibit low δ 13 C values, while higher values are reported for high salinity oceanic waters 72 . The similar carbon isotopic values found in the collagen of estuarine and oceanic species, suggest that the low Table 1. Mean and standard deviation (SD) of δ 13 C and δ 15 N isotope values of (in‰) in aquatic mammal bone samples. Specimens were collected from Northern, Amazon lowland (AL), Amazon Estuary and adjacent coastal zones (AE), and Northeastern Brazil (Maranhão and Piauí, NC).     39,48 , mostly associated with negative interactions with artisanal fisheries. Indeed, stranding events are considered an important source to access ecology data of marine mammals. The study area is influenced by the water and wind regime of the North Brazil Current (NBC), defined as a major low latitude western boundary current in the Atlantic that transports upper-ocean waters northward across the Equator. NBC does not show significant seasonal variation in the lower-frequency fluctuations in the thermocline layer throughout the year 78 . As such, carcasses are expected to be transported westwards on a regular basis, following  www.nature.com/scientificreports/ trade winds, drifting in a water mass of warm (> 24°C) temperatures. Teutophagous species (e.g., Globicephala macrorhynchus, Peponocephala electra and Grampus griseus) and species that feed on small pelagic fish and squid (e.g., Lagenodelphis hosei and Stenella attenuata) displayed similar δ 13 C and δ 15 N values, coinciding with their low trophic level predation in offshore areas 79 . So far, no previous isotopic data for G. macrorhynchus, P. electra and Physeter macrocephalus have been reported for Brazilian waters. P. macrocephalus samples were the most 15 N-enriched and 13 C-depleted compared to the other assessed taxa. Due to the long-time interval integrated in the bone collagen of this long-lived cetacean, these values are possibly averaging their movement between isotopically distinct foodwebs throughout their lives, thus limiting the interpretation of their trophic position within the studied cetacean community. Nevertheless, it is expected that the species occupy high trophic levels within offshore foraging grounds 80,81 which is supported by the isotopic values in the two mature females and the immature specimens analyzed in this study. Higher nitrogen isotopic values were expected for the coastal Sotalia guianensis, regardless of its trophic position, as higher basal nitrogen values were observed in the estuarine waters 82 . Basal nitrogen isotopic values seem to track the main nitrogen sources used by the producers 83 . Indeed, higher δ 15 N values are found within estuarine and nearshore areas dominated by diatoms while lower  www.nature.com/scientificreports/ δ 15 N values, typical of diatom-diazotroph associations and oceanic diazotroph producers, which are reported within the Amazon plume and the adjacent oceanic areas, respectively 32,82,83 . In this context, a higher relative importance of low trophic level prey such as shrimps and squids in the diet of the species 84 may account for the low nitrogen isotopic values found in this species.
niche partitioning. The largest isotopic niche areas were found for Inia spp. and T. inunguis, mainly resulting from a wide range of δ 13 C values, reflecting higher habitat plasticity, i.e., foraging along a gradient encompassing lowland freshwaters and AE estuarine habitats. Variable carbon isotope values in these aquatic mammals may be the result of a diversity of basal sources found mainly in the Amazon estuarine complex (AE), characterized by the presence of both 13 C-enriched C 4 plants (e.g., seagrasses) and low δ 13 C plants, such as mangrove leaves (δ 13 C = − 28.4 ± 0.5‰ 85 ). Ontogenetic variation were found in manatee diets at the Tapajós and Negro rivers, in the Amazon basin, with a proportional consumption of C4 and C3 plants for lactating females and other adults, respectively 86 . Some potential diet items were identified for both manatees, Amazonian and Antillean, in estuarine Marajó Bay areas (i.e., Blutaparon portulacoides, Eleocharis geniculata, Crenea maritima), including both C3 and C4 plants 87 . Vegetation composition and rainfall season could influence the biomass and exposure of seagrass and macroalgae and, consequently, manatee foraging habits 88 . In the Amazon estuarine complex, the local tidal regime probably causes changes in vegetation availability, such as substrate and flood level point differences in size, shape and plant appearance 87 . Manatee movements inside this dynamic habitat can be influenced by food availability, which should vary according to the dry and rainy seasons. The rainy season has a strong influence on the presence of manatees in Marajó Bay 89 .
Until recently, the Amazon River dolphin, Inia geoffrensis, was considered endemic to the Amazon and Orinoco basins 90 while the newly described I. araguaiaensis was restricted to the Tocantins-Araguaia River basin 52 . However, Costa et al. 91 described the presence of Inia spp. in estuarine areas of Marajó Island and the eastern Pará coast. Further molecular analyses confirmed the occurrence of both Inia species around Marajó Is. and the Curuçá Estuary, extending I. araguaiaensis distribution area in nearly 500 km 42 . The small sample size in the present study did not allow for comparisons between the isotopic niche of these two species. Data on one I. araguaiaensis collected at Curuçá Estuary was 13 C-depleted in δ 13 C values, similar to samples from the AL sector, suggesting that it probably fed in a 13 C-depleted food web (e.g., mangrove) 85 and moved to estuarine areas. Prey availability could determine this species movements and might affect species distribution 92,93 . Although I. geoffrensis and I. araguaiaensis are considered exclusively freshwater species 52,90 , the findings reported herein reinforce that the use of estuarine areas is not circumstantial but rather reflects their occupancy in these environments for active foraging.
Although S. guianensis and Inia spp. are considered piscivorous species 90,94,95 no isotopic niche overlap was observed, indicating spatial and trophic partitioning, at least in Marajó Bay. Studies on the trophic ecology of I. geoffrensis are scarce, although the analysis of stomach contents revealed at least forty-three fish species 90 . However, no studies describing the species' diet in flooded areas of the eastern Amazon or in the Amazon estuarine complex are available to date.
Isotopic values indicate that Inia spp. forage at higher trophic levels than S. guianensis. Morphological Inia characters (e.g., lateral mobility) probably allow individuals to explore areas with dense vegetation in floodplains (i.e., várzea) and flooded forests (i.e., igapó) 96 . In the Central Amazon I. geoffrensis and S. fluviatilis have been observed foraging in extensive floodplain areas around main rivers (Solimões and Amazonas). In this region, the flood cycle impact can determine both habitat and prey availability 92 . Differences in habitat use, foraging strategies and prey availability, as well as consumption of prey belonging to different trophic guilds, could reduce inter-specific competition between Inia and Sotalia in sympatric areas.
ecological Sotalia guianensis stocks. At least six management units for S. guianensis have been suggested alongshore Brazil (Pará, Ceará, Rio Grande do Norte, Bahia, Espírito Santo and South-Southeastern area) through the use of molecular markers, evidencing a strong population structure 51 . Other ecological parameters have been used to differentiate ecological stocks. Different geographic areas (Espírito Santo, north and south Rio de Janeiro and São Paulo) have been recognized mainly for cranial and feeding apparatuses variables 97 . Cranial morphometry exhibited a latitudinal growth pattern from North to South, i.e., specimens from São Paulo had smaller skulls. Skull analyses from the north, northeast and southeast have been demonstrated a complete separation between Pará and Rio de Janeiro populations, and a partial separation between Pará, Maranhão and Piauí 98 . Analyses of the periotic-tympanic bone complex of three Brazilian regions also reported geographic variations 99 .
In order to evaluate ecological S. guianensis stocks, Botta 100 analyzed stable isotopes in Guiana dolphin teeth from the Amazon River estuary, Ceará, Espírito Santo, northern Rio de Janeiro, southern São Paulo and northern Paraná and northern Santa Catarina. Four groups were recognized, indicating that SIA is a powerful tool to confirm ecological stock differences.
Significant differences in isotopic carbon composition were found for S. guianensis from AE and NC in the present study, although with similar nitrogen isotope values. Sgui_AE bone collagen samples were more depleted in 13 C than those from Sgui_NC. Dolphins from AE occupy a broader isotopic niche area, indicating higher trophic plasticity. Groups in this sector probably have a larger availability of more diverse prey associated to freshwater habitats 101,102 and carbon source variability due to the huge influence of the Amazon River and Pará River estuary (i.e. Marajó Bay). Individuals from Sgui_NC occupy a narrower isotopic niche area, probably resulting from the association of a greater consumption of marine fishes, as identified in stomach contents of individuals from that population 102  www.nature.com/scientificreports/ In summary, this study provided a first attempt to evidence trophic relationships and resource partitioning among aquatic mammals from the Amazon estuarine complex and its adjacent coastal zones. The obtained SIA results allowed for insights on habitat use and the trophic ecology of marine mammals inhabiting different oceanographic regions off Northern/Northeastern Brazil. Furthermore, δ 13 C and δ 15 N values in S. guianensis inhabiting two coastal regions (AE and NC) indicated an ecological distinction between the populations. These findings provide novel data on the stable isotope composition for these threatened cetaceans and sirenian species in this large section of Brazilian coast.