Diet of the freshwater stingray Potamotrygon motoro ( Chondrichthyes : Potamotrygonidae ) on Marajó Island ( Pará , Brazil )

The stomach contents of 137 examples of Potamotrygon motoro caught in 3 locations (Muaná, Afuá and Lake Arari) on Marajó Island were analysed. The values of the Index of Relative Importance (IRI) and its respective percentage (%IRI) were calculated. The level of repletion 1 (1⁄4 full) was the most representative for both sexes, as well as for immature and mature specimens. Most of the food items found were well-digested. The food items identification indicated the presence of 15 orders, including insects, mollusks, crustaceans, annelids and fish. Differences in diet were observed among the locations studied when comparing %IRI, crustaceans being the most preferred in Afuá, fish in Lake Arari and mollusks in Muaná.


Introduction
The Amazon Estuary involves the Amazon and Tocantins Rivers discharge area, as well as an archipelago of river-ocean islands where Marajó Island is located.Marajó Island occupies an area of approximately 50 thousand km 2 and is the largest river-ocean island in the world (Cruz, 1987).This region shelters both marine and freshwater ichthyofauna, including freshwater stingray species (Barthem, 1985;Sanyo Techno Marine, 1998).It presents heterogeneous environments, forming a mosaic of differentiated habitats among its vegetation types.These habitats, such as lakes, lagoons, beaches, streams and mangroves, are important for many fish species including elasmobranchs, since they have differentiated patterns of habitat use and preferences (Carrier et al., 2004).
The Potamotrygonidae family is considered unique among elasmobranchs since its members are completely adapted and restricted to freshwater environments (Thorson et al., 1983;Rosa, 1985).The feeding studies in continental stingrays have shown variations in composition and frequency of certain food items, as well as intra and inter-specific diet differences according to Some aspects of the feeding habits and diet of Potamotrygon motoro were previously studied by other authors using varying methodologies (Pântano-Neto and Souza, 2002;Lonardoni et al., 2006;Melo et al., 2007;Shibuya et al., 2007;Silva and Uieda, 2007).Results from these researchers demonstrated that this species has feeding plasticity and diet variability.Nevertheless, few of them were carried out in the Amazon region and none in the Amazon River mouth region.Taking into account that P. motoro is caught as an ornamental and food resource in this region, the understanding of its biology is essential for adequate management and conservation purposes.
The main objective of this study was to characterise the diet of Potamotrygon motoro taken from three sampling points on Marajó Island, evaluating possible differences between localities, sexes and sexual maturity stages (immature or juvenile and mature or adult).

Material and Methods
Seventeen sampling periods of approximately seven days each were carried out and distributed over the dry, rainy and transitional seasons, between 2005 and 2007.The specimens were caught with bottom longlines and other fishing gears (seine nets, gillnets, spears, and handlines) mentioned in the literature as being efficient methods for catching freshwater stingrays (Barthem, habitat (Charvet-Almeida, 2001;2006;Rincon, 2006).These studies indicate that potamotrygonids could be generically classified as carnivorous, preying on fish, crabs, shrimps, insects and mollusks.
Potamotrygon motoro (Natterer in Müller and Henle, 1841) (Ocellate River Stingray) is a widely distributed species, present from the Orinoco River (Venezuela) down to La Plata River, but is absent from the Atrato-Magdalena (Maracaibo) basin and Paranaíba River (Rosa, 1985).This species is popular in the ornamental fish trade (Araújo, 1998), is polychromatic (Rosa, 1985), is highly fecund in relation to other potamotrygonids (Charvet-Almeida et al., 2005) and poorly known, with sparse life history and population data available at the present moment (Drioli and Chiaramonte, 2005).
Research on fish feeding habits (diet) and feeding strategies has grown in importance in ichthyologic studies.Such studies provide information for better understanding of how ecosystems function, and consequently on how to manage them correctly (Zavala-Camin, 1996).
Elasmobranchs perform an important role in energy transfer among the higher trophic levels (Wetherbee and Cortés, 2004).Precise descriptions of diets in this group are hampered by the plasticity of their feeding habits that often result in ontogenetic and spatiotemporal shifts (Achenbach and Achenbach, 1976;Wetherbee and Cortés, 2004).1981), where the percentage by volume originally used was substituted for a percentage by weight.This modification was needed owing to the difficulty in calculating the percentage by volume for very small food items.The IRI modified method was based on the following equation: IRI = % F. O. × (% P + % N).
The IRI was also transformed into a percentage (% IRI) for a better interpretation of the data in accordance with the recommendations in the literature (Cortés, 1997).A three-dimensional graphic representation of the stomach content data was prepared with Statistica 7.0 (Statsoft, Inc.).
The stages of sexual maturation were obtained from another study in progress on the reproductive characteristics of this species in the same region (personal observation of M. P. Almeida).In the present study, the sexual maturity stages were divided into immature (juvenile) and mature (adult).
The levels of repletion were observed and the following values were attributed: 0 = empty; 1 = ¼ full; 2 = up to half full; 3 = ¾ full and 4 = full.The degree of digestion of the items followed the scale suggested by Zavalla-Camin (1996), who used the following values: 1 = undigested food; 2 = external parts partially digested; 3 = external parts and muscular mass partially digested; 4 = only the axial skeleton and part of the muscular mass present; and 5 = only fragments are found.
Direct comparisons were made with the calculated %IRI values regarding the diet between sampling points, sexes, and immature and mature specimens.The degrees of repletion between immature and mature specimens and between sexes were compared by using the Chi-Square Test (χ 2 , degree of freedom DF = 1; level of significance P < 0.05) (Zar, 1999;Ayres et al., 2003).
Empty stomachs or those containing only amorphous substance were not considered for analysis.The spiral valves were also not taken into account for analysis since they presented contents at a very advanced stage of digestion.
In this study, plant fragments were not considered a food item, since this material is probably incidentally ingested by suction and is unlikely part of the diet of freshwater stingrays (Bragança, 2002;Charvet-Almeida, 2001;2006).
Detailed analysis of the diet verified that about 15 major prey orders were identified among this species' stomach contents.Decapoda was the most representative food item order considering FO, N, W and IRI percentages.The second most important order in terms of %IRI 1985; Lasso et al., 1996;Charvet-Almeida, 2001;2006;Rincon, 2006).Longlines were the main fishery gear used.They were usually set at dusk and recovered at dawn, remaining in the water for a period varying from two to twelve hours.Fishing during daytime was impossible due to the presence of bait competitors (Characidae, mainly Serrasalmus spp.).Nets were used only as complementary gears and in a non-standardised way.
Afuá is located on the northwestern portion of Marajó Island.There are many river channels (depths ranging mostly from 0.5 to 6 m) and islets in this area, which is highly influenced by the Amazon River discharge and by tidal variations.Muana is on the southeastern part of the island, north of the Marajó Bay.There are several river channels (similar depth ranges as in Afuá) and flooded areas, influenced mainly by the Tocantins River discharge.Tidal change variations are intense.Arari is the largest lake on Marajó Island, located on the central eastern part of the island.Around this lake there are flooded grass fields and many streams that drain towards it.Water level in this environment seasonally varies with rainfall but limited tidal influence was noted in this area.Muddy substrates were predominant in all sampling points.
The freshwater stingrays were anesthetised, killed and eviscerated through a ventral semi-circular incision performed below the scapulocoracoid cartilage.Stomachs had their extremities tied to avoid content loss, were removed, labeled, fixed in formaldehyde buffered (sodium tetraborate) solution and conserved in 70% ethanol.
During laboratory analysis, stomach contents were emptied onto a Petri dish and examined under a stereoscopic magnifier (Nikon ® SMZ-10A).The food items were weighed on a precision scale (0.001 g -Sartorius LC621S) but excess liquid was removed with soft paper towels previously to weighing.
The stomach content analysis was carried out using the most specific taxonomic level.Afterwards, the previously identified food items were sent to specialists in each of the groups found to confirm their identification.
The values for each food item were calculated as follows: a) Percentage frequency of occurrence (% F.O. = 100 × Fi / n, where: Fi = number of times the food item occurred; and n = total number of stomachs with food items).b) Percentage by number (% N = 100 × Ni / n, where: Ni = number of items of each taxon; and n = total number of items of all taxa).c) Percentage by weight (% W = 100 × Wi / n, where: Wi = total weight of one item (taxon); and n = sum of all the weights of all stomach contents).The Index of Relative Importance (IRI) (Pinkas et al., 1971) was calculated in a modified way (Hacunda, diet of P. motoro at Marajó Island were fish and crustaceans (Figure 2).
The IRI ranked five groups of food items considering the location and sex (Table 2).The dominant items were different for the different sampling points, crustaceans were predominant in Afuá (IRI = 97%), fish in Arari Lake (64%) and mollusks in Muaná (58%).The main food item was the same for both sexes within each sampling point.Nevertheless, when considering the sampling points altogether, IRI values for fish and crustaceans were dominant in the diet of Potamotrygon motoro.
In Afuá and Lake Arari there were no diet variations when comparing immature and mature specimens.However, in Muaná the dominant food item for immature specimens was crustaceans, whilst the sexually mature potamotrygonids preferred mollusks.On Marajó Island (locations altogether) the dominant food item for immature stingrays was crustaceans and fish for mature specimens (Table 3).
Most of the food items were well-digested, pertaining to category 4 (31%) and 5 (37%).This characteristic posed a problem to precisely identify the food items.A digestion was Siluriformes (Osteichthyes).The remaining orders represented less than 5% (Table 1).
The families that could be identified in the samples were: Hyriidae, Corbiculidae and Ampullaridae (Gastropoda); Grapsidae, Among these, the ones representing the greatest %IRI were: Pomacea sp.(gastropods); Dilocarcinus pagei (Malacostraca); Polymitarcyidae (Insecta) and Callichthyidae (Osteichthyes).In Oligochaeta a more specific identification other than class was impossible to be achieved.
According to the three-dimensional graphic representation of the %IRI, the dominant food items in the Table 1.Results of the stomach content analysis for P. motoro (n = 137) indicating the percentage frequency of occurrence (%FO), percentage by number (%N), percentage by weight (%W), Index of Relative Importance (IRI) and its respective percentage (%IRI) for each taxonomic order identified.

Discussion
Feeding habits and diet studies have been through various revisions and proposals of new methodologies to analyse data in different animal groups in the recent degree 5 was more frequent when females (34%) and males (46%) were considered separately.The most frequent results were degree 4 (40%) for the immature specimens and degree 5 for the sexually mature specimens (39%).

Food item / sexual maturity stage
Afuá (1) Arari Lake ( 2  The numbers in bold correspond to the highest values. River (São Paulo and Mato Grosso).Results from these authors demonstrate that this species has feeding plasticity and diet variability.Certainly the feeding plasticity of P. motoro has contributed to its wide geographical distribution and favours its expansion into new areas/habitats.P. motoro is considered an invading species in the Paraná River (Agostinho et al., 1997;Lonardoni et al., 2006;Garrone-Neto et al., 2007) and probably it is capable of adapting to the food items available in its new distribution range.
Differences in the diet between sexes were expected, as observed in Potamotrygon leopoldi in the Xingu River region (Charvet-Almeida, 2006).In this group of stingrays the reproductive strategy (matrotrophic viviparity with trofodermata -Charvet-Almeida et al., 2005) depends on lipid reserves accumulated in the female liver (Baldridge, 1970;Craik, 1978) and therefore is directly related to the feeding process.However, the dominant item in the diet of P. motoro was the same for both sexes on Marajó Island.
Several elasmobranch species have diet changes associated to ontogeny and sexual maturity (Lowe et al., 1996;Muto et al., 2001).Some diet differences were noted between immature and mature specimens.These were observed in Muaná and in the overall results.According to the overall data, immature stingrays seemed to prey mainly on crustaceans, while mature specimens on fish.However, this predominance varied among sampling points and is possibly associated to prey availability in each area.In the Paranã River (Tocantins River drainage) an ontogenetic diet shift was registered for Potamotrygon orbignyi (Rincon, 2006).Rincon (2006) pointed out that a higher diversity of food items (mainly insects) was consumed by larger stingrays and that this shift was not associated to sexual maturity.
The presence of plant fragments in potamotrygonid species stomach content analysis (Charvet-Almeida, 2001;2006;Rincon, 2006;Lonardoni et al., 2006) contribute to the idea that these stingrays present speculative habits, as indicated by Zuanon (1999).The presence of small quantities of these fragments in the stomachs also suggests that these particles are not actually part of these stingray's diet, but are probably ingested accidentally by suction during the feeding process (Bragança, 2002).past.Cortés (1997) presented the first critical revision on the study methods used in fish feeding based on a stomach content analysis with a specific approach for elasmobranchs.It recommended the use of the Index of Relative Importance (IRI) (Pinkas et al., 1971), its percentage and also indicated the use of a three-dimensional graphic representation, as being the most appropriate (Cortés, 1997).
When considering the relative environment similarity and the wide distribution range of P. motoro in the Neotropical region (Rosa, 1985), little or no diet variation was expected.The greatest variety in food items found for P. motoro on Marajó Island occurred with the bony fish and insect categories, both of them with at least 5 different orders.Despite this large variety, insects always showed low %IRI values, and never represented more than 3% of the samples taken.
Until now P. motoro seems to be the freshwater stingray species that displays the most varied feeding habits.This affirmation is based on the results available from this and other studies.Pântano-Neto and Souza (2002) in the Cristalino River (Goiás) region identified a diet based on insects, mainly Chironomidae (Diptera); Melo et al. (2007) on the Juruá River Basin (Acre) indicated this specie to be insectivorous; Shibuya et al. (2007) showed that the preferential food item were crustaceans in the Negro River Basin (Amazonas); Lonardoni et al. (2006) showed that P. motoro preferred mollusks in the flood season and insects in the dry season in the Paraná River (Paraná) and Silva and Uieda (2007) demonstrated that this species fed on both insects and fish in the Paraná  Despite being highly regarded, the IRI (Pinkas et al., 1971) holds the premise that the analysis should be performed with all items at the same identification level in order to avoid results distortions.Since food items found in elasmobranchs are usually highly digested, the results almost always require higher taxa groupings (e.g.insects, crustaceans, etc.).
In the present study, it was observed that the food items examined were well-digested, and therefore it was difficult to identify the samples at more specific taxonomic levels.Given the condition of items, at higher taxa grouping, P. motoro ingested predominantly fish and crustaceans.However, when the food item analysis was performed at an order level, the crustacean group became the most representative, followed by fish, producing, therefore, an inversion in importance of the main food items observed.Crustaceans were represented by a single order (Decapoda), while fish comprised of at least five orders.If fish were grouped under a single taxonomic category (e.g.teleosts), the %FO and %P and consequently IRI values would be higher than the ones calculated for crustaceans.Cortés (1997) considered IRI efficient to describe species diets; however, he also recognised that that IRI values depend a lot on the taxonomic resolution levels attained in the analysis.
The occurrence of high percentages of empty stomachs, or stomachs with low levels of repletion, was evident in the sample studied and is a common characteristic observed among marine stingrays (Wetherbee et al., 1990), as well as in other freshwater stingray species (Charvet-Almeida, 2001;Bragança, 2002;Bragança et al., 2004;Rincon, 2006;Charvet-Almeida, 2006).These results suggest that potamotrygonids possibly feed on small quantities, perhaps several times a day.An hourly sampling standardisation throughout a 24-hour period to check precisely the feeding time could not be done since daytime fisheries were impracticable due to bait competitors.This way, results obtained in the present study corresponded mainly to nighttime sampling.The same problems with daytime fisheries and similar methodologies were adopted by other freshwater stingray diet researchers in the Amazon region (Bragança, 2002;Bragança et al., 2004;Charvet-Almeida, 2006;Rincon, 2006).
Results demonstrated that in the different environments studied P. motoro varied its feeding habits and always displayed differentiated diets with main food items %IRI values higher than 50%.This variability in feeding can be related to the abundance/availability of specific food items in each sampling point or to a specific feeding behaviour expressed in different locations.

Figure 2 .
Figure 2. Three-dimensional graphic representation of the Percentage of the Index of Relative Importance (%IRI) obtained for the food items of P. motoro on Marajó Island, where: D = dominant food category and R = rare food category (referring to the prey), and G = generalised diet and S = specialised diet (referring to the predator).

Table 2 .
Percentage Index of Relative Importance (%IRI) calculated for the food item groups of P. motoro by location (1, 2 and 3 and overall total) and sex (M and F).

Table 4 .
Chi-Square Test (χ 2 ) comparing stomach repletion levels to sex and sexual maturity.

Table 3 .
Percentage Index of Relative Importance (%IRI) calculated for the food item groups of Potamotrygon motoro by location (1, 2 and 3 and overall total) and stage of sexual maturity.