Feeding ecology of electric eel Electrophorus varii (Gymnotiformes: Gymnotidae) in the Curiaú River Basin, Eastern Amazon

In this study, the composition of the diet and the feeding activity of Electrophorus varii were evaluated. The influence of ontogeny and seasonality in these feeding parameters was also examined. Fish were collected in the Curiaú River Basin, Amazon, Brazil, from March 2005 to February 2006, during the rainy (January– June) and dry (July–December) seasons. Diet composition was characterized based on the analysis of stomach contents and feeding dynamics was assessed based on the Stomach Fullness Index (IR) calculated using stomach weight. Stomach content and RI data were grouped into four-cm size classes (40–80, 80–120, 120–160, and 160–200) and two seasonal periods (rainy and dry). The influence of ontogeny and seasonality in the diet was investigated through PERMANOVA, and in the food dynamics through ANOVA. The analysis of stomach contents revealed that fish were the most consumed preys by electric eels, especially Callichthyidae and Cichlidae. Diet composition and RI values of electric eels were not influenced by ontogeny and seasonality. Electric eels are fish predators, regardless of size class and seasonal period.


INTRODUCTION
Fish of the order Gymnotiformes are known for their specialized organs that generate electric discharges for electrolocation and electrocommunication (Moller, 1995). These electric fishes have a geographical distribution confined to freshwater ecosystems in the neotropics, from southern Mexico to northern Argentina (Albert, 2001), inhabiting lakes, wetlands, streams, and rivers (Crampton, 2011). The most notorious Gymnotiformes are electric eels, Electrophorus Gill, 1864, because of their strong electric discharges, large size (up to 2.5 meters in total length; Ellis, 1913;Campos-da-Paz, 2003), and air breathing through a modified oral organ (Johansen et al., 1968).
The information available on the diet of electric eels is speculative, as few stomach 3/17 ni.bio.br | scielo.br/ni contents have been analyzed (Saul, 1975;Soares et al., 1986;Planquette et al., 1996;, in addition to a lack of consensus about the feeding habits of Electrophorus. Some authors classify Electrophorus as specialized fish predators (Bullock et al., 1979;Soares et al., 1986;Westby, 1988;Mago-Leccia, 1994;Stoddard, 1999;Sá-Oliveira et al., 2014;Mendes-Júnior et al., 2016), while others have categorized them as generalist carnivores (Ellis, 1913;Sterba, 1959;Saul, 1975;Goulding et al., 1988;Planquette et al., 1996;Mérona, Rankin-de-Mérona, 2004;Crampton et al., 2013;Giora et al., 2014;Stoddard et al., 2019). Nakashima (1941) and Goulding (1980) have also reported the consumption by electric eels of fruits of açaí, Euterpe oleracea, a palm tree common in Amazonian floodplains. Electric eels are important components of the ichthyofauna in Amazonian floodplain systems (Crampton, 1996), where most fishes inhabit the main river channel and feed on a wide variety of preys (Junk et al., 1997). During the rainy season, the water level of rivers and their tributaries increases and floods marginal terrestrial habitats, influencing prey availability and quality (Junk et al., 1997). Fishes species that inhabit the main river channel and its tributaries migrate to the adjacent floodplains during the rainy season, where they feed and reproduce, with these areas serving as natural nurseries for juveniles of species with general feeding habits (Junk et al., 1997;Abelha et al., 2001) that then return to the main river channel in the beginning of the dry season. Some fish are trapped in puddles in the floodplain during the dry season, where they usually have carnivorous feeding habits (Junk et al., 1997). Individuals of E. varii are generally found in residual pools during the dry season that they share with other fish with adaptations to environments with low dissolved oxygen in the water, such as Hoplosternum littorale (Hancock, 1828) and Hoplerythrinus unitaeniatus (Spix & Agassiz, 1829) (Val et al., 1998. To understand the feeding ecology of fishes species in dynamic ecosystems such as floodplains, it is essential to study parameters other than diet, such as feeding dynamics. Temporal variation in stomach volume can be a useful indicator of patterns and degree of both daily and seasonal fish feeding (Eliassen, Jobling, 1985) and is among the most common methods for estimating prey consumption by fish in the wild (Elliot, Persson, 1978). The analysis of feeding dynamics allows the characterization of the period of feeding activity and the nutritional condition of fish in natural conditions (Barbieri, Barbieri, 1984), as well as the response of fish populations to changes in environmental conditions (Pereira et al., 2016). No information is available on the feeding dynamics of Electrophorus species in nature, and their period of feeding activity is still unknown.
Fishes diet and feeding activity are influenced by several biotic and abiotic factors, including ontogeny (Griffiths et al., 2009;Ferriz, Iwaszkiw, 2014) and changes in hydrometric level (Junk et al., 1997). Generally, young fish of predatory species tend to consume a wider variety of prey than adults (Winemiller, 1989;Hahn et al., 1997;Novaes et al., 2004). These ontogenetic differences in diet are also observed in some species of predatory fishes in floodplains, with small individuals consuming microcrustaceans, medium size fish eating mostly aquatic insects, and larger individuals preying mainly on fish (Winemiller, 1989). Ontogenetic variations in the diet of fry and juveniles of E. varii were observed in the floodplain of Ilha do Marajó-PA, Brazil, with 7 cm fish exclusively feeding on conspecific eggs and developing embryos (cannibalism), 8 cm fish consuming mainly of eggs and embryos, with occasional ingestion of crustacean larvae, and 9 cm fish eating mostly crustacean larvae (Assunção, Schwasmann, 1995).
The diet of E. varii individuals larger than 9 cm is still unknown.
Some fishes diets and feeding dynamics change seasonally in floodplain ecosystems depending on the availability of food resources, with more autochthonous prey during the dry season and allochthonous ones during the rainy season, with predator fish displaying opportunistic behavior as their main feeding strategy in this type of habitat (Cardoso et al., 2019). Therefore, electric eels that inhabit floodplain systems might have higher trophic plasticity, due to greater availability of prey. However, electric eels might also maintain a more specialized feeding habit in floodplains, as electric discharges allow them to hunt more specific preys regardless of the time of year. The present study aims to characterize for the first time the diet of the electric eel E. varii in the wild, based on analysis of stomach contents and intestine size, and its feeding dynamics based on stomach weight variation (stomach repletion index -RI), in addition to determining the influence of ontogeny and seasonality in the stomach content and the feeding activities of this species in the Curiaú River basin, Amapá, Brazil.

MATERIAL AND METHODS
The river-floodplain system of the Curiaú River Environmental Protection Area (Curiaú River EPA) ( Fig. 1) is located in the urban expansion area of the Municipality of Macapá, state of Amapá, Brazilian Amazon (Chellappa et al., 2005). It covers a small area of 23,000 ha, which contrasts with its high diversity of ecosystems such as savannas, dry-land forests, forests, and floodplains. The Curiaú River and its floodplain form a floodplain-river system, with dry seasons occurring from July to December, and rainy seasons, from January to June (Chellappa et al., 2005).
Electrophorus varii specimens were sampled every two months, which included the dry (July to December 2005) and the rainy seasons (March 2005 to June 2005 and January and February 2006). Most individuals were collected by active search in the floodplain, using trawl nets, fishing nets, casting nets, and line and hooks at various points distributed in streams, permanent and temporary lakes of the Curiaú River Basin. Groups of gillnets (mesh of 3-10 cm between opposing knots) were set up overnight (12 hours of exposure). The difficulty in locating electric eels in the wild without the use of electric fish detectors, as described by Crampton et al. (2007), was offset by the assistance of quilombolas (current inhabitants of rural communities formed by descendants of enslaved Africans), not allowing the pre-establishment of fixed sampling units. Thus, the entire floodplain with its streams, temporary and permanent lakes was included the study area.
Ninety-five (45-178 cm in length) young (<85 cm, N=20) and adult (>85cm, N=75) electric eels were collected. Of the 95 specimens, 50 were collected in the dry (45-178 cm in length) and 45 in the rainy season (48-163 cm in length). Total weight (Wt) in grams (g) and total length (Lt) in centimeters (cm) were obtained. Fish were dissected, and intestine length (Li) in centimeters (cm) and stomach weight (Ws) in grams (g) were measured. Two intestines were damaged, and their length could not be measured. Voucher specimens were deposited in the biological collection of the Instituto Nacional de Pesquisas da Amazônia (INPA) under numbers INPA 41112 to 41122 and 41124. Stomachs were stored in 10% formalin solution. In order to determine the diet of electric eels, 47 stomachs were dissected with the aid of a stereoscopic microscope. Four were empty and of the 43 stomachs with preys, 15 had damaged identification tags and the information on size and period of collection could not be retrieved. The remaining 28 stomachs belonged to specimens with sizes ranging from 56 to 178 cm in length. Food items were identified at the lowest possible taxonomic level and grouped into the following taxonomic and ecological categories: fish (whole animals in varying degrees 6/17 ni.bio.br | scielo.br/ni of digestion, scales and bones), terrestrial insects (Hymenoptera, Coleoptera and insect remains), aquatic insects (Odonata larvae and Coleoptera), crustaceans (Palaemonidae shrimps and Trichodactylidae crabs), amphibians (frogs), and seeds. Plant remains were not considered as food items as they were probably accidentally ingested while eating other preys and are not commonly observed in stomach contents of other Neotropical electric fishes species (Giora et al., 2014).
Food items were analyzed using the occurrence frequency method (Hynes, 1950;Hyslop, 1980;Bowen, 1983) and the Volumetric Analysis Index (Lima-Junior, Goiten, 2001). The importance of each food item was determined by the Alimentary Index (Kawakami, Vazzoler, 1980) as follows: IAi = (Fi% x Vi%) / (∑Fi% x Vi%), where i = 1,2 , ... n food item, Fi% is the frequency of occurrence of food item i, and Vi% is the volume of food item i.
As a complementary analysis to characterize the diet of electric eels, the Intestinal quotient (IQ), which represents the ratio of intestine length (Li) to total length (Lt), was obtained according to Giora et al. (2005) as follows: IQ = Li / Lt, where IQ is the intestinal quotient, Li is the intestine length and Lt, total length.
To determine the feeding dynamics of electric eels, the stomach repletion index (RI), calculated as the ratio of stomach weight to total animal weight, was used and estimated according to the formula proposed by Santos (1978) and adapted as in Giora et al. (2005): RI = Ws x 100 / Wt, where Ws is the stomach weight and Wt is the total weight. Although this is the most commonly used method (Elliot, Persson, 1978), limitations in the use of stomach weight should be accounted for when evaluating fish feeding dynamics (Elliasen, Jobling, 1985;Bromley, 1994). Fish stomach weight can be overestimated when prey is eaten during capture (Bromley, 1994). Inversely, stomach weight can be underestimated due to prey regurgitation during capture (Bromley, 1994), progression of digestion and evacuation in passive fishing methods, and the influence of preservation methods (Elliasen, Jobling, 1985;Bromley, 1994). In the present study, sampling occurred in different seasonal periods, and included specimens in a wide range of size classes, which can mitigate the disadvantages of the use of IR values in the estimation of E. varii feeding dynamics.
In order to decrease the importance of individual differences when analyzing the influence of ontogeny and seasonality in diet composition and feeding dynamics (Elliassen, Jobling, 1985), stomach contents and RI data of E. varii were grouped into four size classes (40 to 80 cm, 80 to 120 cm, 120 to 160 cm, and 160 to 200 cm) and two seasonal periods (rainy and dry). To analyze whether diet composition differed among size classes of electric eels and between seasonal periods, volume (Vi) of groups of prey of 28 stomachs were transformed by division by sum. The transformed data were used in the construction of a Bray-Curtis dissimilarity matrix. The latter was the basis for a Permutational Multivariate Analysis of Variance (PERMANOVA), a non-parametric statistical test that allows the investigation of differences between predefined groups (Delariva et al., 2013). PERMANOVA was performed with 9,999 permutations in order to test the significance of the generated values of Pseudo-F.
Electric eel size and seasonality were also tested to examine if they affected feeding dynamics through ANOVA. This analysis was based on the Stomach Repletion Index (Ri) data from 95 specimens.
Since the objective is to assess the individual effect of ontogeny and seasonality 7/17 ni.bio.br | scielo.br/ni on diet and feeding dynamics, PERMANOVA and ANOVA included both factors simultaneously to eliminate any possible influence of the combined effect of the variables on the data. Probability values lower than 0.05 were considered significant. PERMANOVA and ANOVA were performed with the software R (R Core Team, 2020) using respectively the function adonis of the Vegan Package (Oksanen et al., 2019) and the function aov of Stats Package (R Core Team, 2020).

RESULTS
Fishes were the most important prey (IAi = 96.66%) in the E. varii diet (Tab. 1). Among the fishes that could be identified, the most frequent and large were specimens of Cichlidae and Callichthyidae. Occasional consumption of crustacea (IAi = 1.45%), aquatic insects (IAi = 0.72%) and terrestrial insects (IAi = 0.78%) was observed. Electric eels had a mean IQ of 0.40±0.06. No influence of ontogeny (PERMANOVA, pseudoF = 0.77299, p = 0.6241) and seasonality (PERMANOVA, pseudoF = 1.18292 p = 0.3024) was observed in the diet of E. varii. Fish were the main prey for electric eels in all size classes (Tab. 2). Fish was also the most consumed prey by E. varii both in the rainy and dry seasons in the Curiaú River EPA (Tab. 3).
Similar RI values were obtained for the size classes of E. varii ( Fig. 2A), as well as   between the rainy and dry seasons (Fig. 2B). Ontogeny (ANOVA, F = 1.727, p = 0.167) and seasonality (ANOVA, F = 2.524, p = 0.116) did not influence the feeding dynamics of electric eels in the Curiaú River EPA.

DISCUSSION
The results of the present study indicate that E. varii is a piscivorous predator, in disagreement with the hypothesis proposed by other authors that electric eels are generalist carnivores (Ellis, 1913;Sterba, 1959;Saul, 1975;Goulding et al., 1988;Planquette et al., 1996;Mérona, Rankin-de-Mérona, 2004;Crampton et al., 2013;Giora et al., 2014;Stoddard et al., 2019). The presence of açaí seeds in the stomach contents of E. varii was considered accidental, as it was observed in few stomachs, with only one seed per stomach, and in general seeds were accompanied by animal prey in the stomach, contrary to the hypothesis of herbivory in electric eels (Nakashima, 1941;Goulding, 1980). The lack in information on the feeding habits of these fish is probably due to the limited sample size in the few studies available (Saul, 1975;Goulding et al., 1988;Mérona, Rankin-de-Mérona, 2004;Crampton et al., 2013;Dary et al., 2017). Cichlidae and Callichthydae were the most consumed groups of fishes by E. varii, which may be explained by their greater abundance in floodplains (Arrington,  Barthem, 2008), particularly in the Curiaú River EPA (Chellappa et al., 2005). Ingestion of prey with anti-predation morphological adaptations, such as fin spines of Cichlidae and lateral dermal plates of Callichthyidae (Kirchheim, Goulart, 2010), reflects the efficacy of predatory tactics of electric eels, such as "Remote Control" described by Catania (2014). In addition to strong electrical discharges, electric eels breathe air (Johansen et al., 1968), allowing them to access shallow and anoxic areas commonly used as a refuge by small prey (Anjos et al., 2008), and consumption of preys with morpho-physiological adaptations to water with low dissolved oxygen concentrations such as Callichthyidae fish (Brauner et al., 1995). Some electric fish were found in the stomach content of E. varii in the Curiaú River EPA, which was also observed in nature and experiments in the laboratory (Bullock et al., 1979;Westby, 1988;Stoddard, 1999). However, considering its low importance in the E. varii diet, as suggested by Stoddard et al. (2019), electric eels cannot be considered as predators specialized in electric fish (Stoddard, 1999;Moyle, Cech Jr, 2004). Fishes were the main preys of electric eels regardless of size class or seasonal period, reinforcing that piscivory is the main feeding habit of E. varii in the Curiaú River EPA. However, there was very occasional consumption of arthropods in most size classes and in the rainy and dry seasons. The sporadic consumption of arthropods by piscivorous predators, such as electric eels, is not surprising (Luz-Agostinho et al., 2008), since plasticity and trophic opportunism in the consumption of prey are common characteristics of freshwater fishes (Abelha et al., 2001), especially in floodplains with seasonal changes in food availability and quality (Junk et al., 1997;Mortillaro et al., 2015). Predation of amphibians and small mammals has also been reported in E. electricus in French Guiana (Planquette et al., 1996) and in a sample of E. voltai in the Jari River, Brazil , but the presence of a frog in a single stomach in the present study suggests that predation of small vertebrates by E. varii in the wild is unusual.
Electric eels with sizes between 40 and 178 cm have a fish-eating habit. Because the mean size of the first gonadal maturation of E. varii is 85 cm (Mendes-Júnior et al., 2016), young individuals larger than 40 cm and adults of E. varii can be considered fish predators. However, the influence of ontogeny in the diet of electric eels can not be ruled out, given that in Ilha do Marajó, E. varii with total length up to 8 cm consumed mainly conspecific eggs and embryos (cannibalism), followed by the replacement of these preys with crustacean larvae by 9 cm electric eels (Assunção, Schwassmann, 1995). This suggests that ontogenetic changes in the diet of E. varii of sizes ranging between 10 to 40 cm can occur.
The absence of effect of seasonality on the diet of electric eels may be due to the predominance of fish consumption in both rainy and dry seasons, as electric eels are predatory piscivorous regardless of the time of year. Fish in river-floodplain systems are generally more easily caught during the dry season, when water bodies retract and preys are confined in the river channel and in marginal residual pools. Comparatively, in the rainy season, the flooding of extensive land areas due to rains promotes the dispersion of fish in the floodplain (Junk et al., 1997), forcing piscivorous predators to modify their diets according to the seasonal period (Lowe-McConnell, 1999;Luz-Agostinho et al., 2008). The varied hunting tactics of electric eels (Catania, 2019) allow them to locate fish, their main prey, dispersed in the floodplain in the rainy season as well as confined to residual pools during the drought, which would explain the absence of effect of seasonality on the diet of E. varii. 11/17 ni.bio.br | scielo.br/ni Considering that the ratio between intestine size and fish size reflects their eating habits, ranging from short intestines of carnivores to long-intestines of herbivores, and intermediate-sized intestines of omnivores (Fryer, Iles, 1972;Barbieri et al., 1994;Gerking, 1994;Ward-Campbell et al., 2005), the short intestine of electric eels reflects their carnivorous eating habits. The average IQ of E. varii is short when compared to that of the piscivore Hoplias malabaricus (Bloch, 1794), with IQ values above 0.53 (Mazzoni, Costa, 2007;Peretti, Andrian, 2008), but the small visceral cavity in relation to the body size of Gymnotiformes do not allow a comparison of IQ values of this group with those of other Teleostei fish. The mean IQ of E. varii is lower than the IQ of the generalist carnivore  (Peretti, Adrian, 1999;Giora et al., 2012;Giora et al., 2014), which contradicts the idea that the IQ of fish-eating predators tends to be lower than that of invertivores. The values observed in E. varii might be higher due to its larger body cavity (covering more than 30 vertebrae) compared to the 16 to 19 vertebrae in Brachyhypopomus Mago-Leccia, 1994and Eigenmannia Jordan & Evermann, 1896(Peixoto et al., 2015Crampton et al., 2016). If the intestine is a good indicator of the diet of electric fish, and since Gymnotus, the putative sister group of Electrophorus (e.g. Alda et al., 2019), consists of voracious predators of fish and insects (Campos-da-Paz, 2003), their intestine is likely intermediate in length between the ones observed in Electrophorus (IQ = 0.40) and those documented in Eigenmannia and Brachyhypopomus species (IQ = 0.25).
Similar to the observed for diet composition, the feeding dynamics of E. varii did not differ among size classes and between seasonal periods in the Curiaú River EPA, which may reflect the high capacity of electric eels to locate and subdue prey using weak and strong electrical discharges, respectively. The electrical organ of E. varii individuals with a length of at least 40 cm can produce discharges greater than 300 volts (from de Santana et al., 2019), thus, electric eels of all size classes analyzed in the present study were equally lethal to prey fish. In addition, the hunting tactics of electric eels described by Catania (2019) allow them to find and control mobile or sedentary prey, regardless of the time of year. Visually oriented piscivorous predators, such as species of Hydrolycus Müller & Troschel, 1844, have different feeding dynamics than those observed in electric eels. They modify their feeding activity according to the seasonal period (Barbosa et al., 2018), since the flooding of river banks in the rainy season provides new refuges for prey, making it difficult for predators to hunt (Lowe-McConnell, 1999), which probably does not occur with E. varii. The present study reports the initial findings of the feeding dynamics of electric eels, due to limitations in the use of stomach weight to measure feeding activity in fishes species (Elliasen, Jobling, 1985;Bromley, 1994).
In conclusion, our findings reveal that electric eels are piscivorous predators and that fishes are the main prey of young individuals (Lt 45-85 cm) and adults (Lt > 85 cm), regardless of the time of year. Future analyzes of the stomach contents of 10-40 cm individuals may indicate in what size class there is a change from a diet rich in crustacean larvae (Assunção, Schwassmann, 1995) to a piscivorous diet (present study). Our initial findings also indicate that the feeding dynamics of electric eels are not influenced by 12/17 ni.bio.br | scielo.br/ni ontogeny and seasonality. However, future studies with more robust methods than the analysis of stomach weight are needed to more effectively measure the feeding dynamics of electric eels. Protection and management strategies and actions are essential to ensure that future generations can experience the celebrate electric eel not only in zoos, public aquariums, wildlife books and documentaries, but in their natural habitat.