Spatial variation in the stable isotopes of 13 C and 15 N and trophic position of Leporinus friderici ( Characiformes , Anostomidae ) in Corumbá Reservoir , Brazil

Stable isotopes of carbon (δ13C) and nitrogen (δ15N) were used to describe sources of energy and trophic position for adult Leporinus friderici in the area of the Corumbá Reservoir, Brazil. Samples were collected from April 1999 to March 2000. Spatial variations were not identified in the isotopic composition. The maximum and minimum contribution of C4 plants calculated integrating the variation of plants and fish were 47.7% and 2.4%, respectively. Among C3 plants, periphyton presented closer isotopic values to those observed for fishes, corresponding to an important carbon source. The proportion of ingested plant item is larger in rivers upstream from the reservoir (42.7%), which justifies the smaller trophic level among there. However, in the reservoir, the ingestion of fish was 81.4%, while ingested plants contributed with 18.6%. Downstream from the dam, participation of plant item was even smaller (14.4%). Although the trophic position calculated with diet data was proportional to the one calculated with δ15N values, the former elevated the trophic level of L. friderici in the food web, because estimated trophic positions were based on fish items belonging to the 2nd (a) and to the 3rd (b) trophic levels.


INTRODUCTION
The use of stable isotopes of carbon (δ 13 C) and nitrogen (δ 15 N) has been intensifi ed in the last years (Hobson and Wassenaar 1999).These isotopes are used to describe sources of energy and trophic relationships in food chains of terrestrial, marine and freshwater ecosystems (Peterson and Fry 1987).The δ 13 C usually identifi es pathways of carbon transference, starting from the primary producers, whereas the δ 15 N characterizes the trophic position of the organisms in food chains ( Van-der Zanden et al. 1997).The amount of δ 15 N in tissues of consumers is, usually, enriched in 3 • / •• in relation to their prey.However, δ 13 C is slightly enriched (1 • / •• ) with the increase in trophic levels (Jennings et al. 1997).
The application of such techniques has been useful in investigation of ecology (Martinelli et al. 1991, McArthur and Moorhead 1996, Keough et al. 1996, France 1997), as well as in analyzing effects of anthropogenic impacts (McClelland and Valiela 1998).In dammed areas, stable isotopes help to understand the processes determining dynamic changes imposed to the new environment and, consequently, supporting conservation and management decisions (Angradi 1994).
Transformations of carbon (or energy) began with CO 2 fi xation by plants.Plants C 3 and C 4 plants differ between themselves by their respective photosynthetic pathways, resulting in different values for carbon stable isotope (Farquhar et al. 1989).Due to selection for lighter isotope during fi xation of carbon, C 3 plants are signifi cantly more enriched in 12 C.These isotopic differences turn relatively easy to identify carbon of C 3 and C 4 plants (Forsberg et al. 1993).
Primary sources of energy in the area of influence of Corumbá Reservoir are C 4 grasses, C 3 plants (constituted by the riparian vegetation), phytoplankton and periphyton (Benedito-Cecilio et al. 2004).Aquatic macrophytes are scarce in the area of the Corumbá Reservoir (Luz- Agostinho et al. 2006).Studies have indicated that, in spite of the great quantity of biomass produced by C 4 plants, isotopic carbon signatures in fi sh are more related to algae based food web (Araújo-Lima et al. 1986, Forsberg et al. 1993).
Leporinus friderici is an abundant species in the Corumbá Reservoir (Agostinho et al. 1999).This species is economically important in other areas of the Paraná River basin, in spite of the environmental modifi cations imposed by impoundments (Agostinho et al. 1989(Agostinho et al. , 1994)).Ecological studies with Leporinus friderici were conducted in the Brazilian stretch of Paraná River without dams (Andrian et al. 1994, Vazzoler et al. 1997), in Itaipu Reservoir (Agostinho et al. 1992, Benedito-Cecilio et al. 1997) and in the dammed stretch of the basin (Lopes et al. 2000, Benedito-Cecilio et al. 2005).These studies generated valuable information to support management actions.
Studies using stable isotopes of carbon were fi rstly carried out in the Amazonian ecosystem in the 80's (Araújo-Lima et al. 1986, Martinelli et al. 1991, Forsberg et al. 1993).However, for the Paraná River, isotope ratios were not described so far for any biotic component.Concepts of energy that flows in food webs have only been based on diet analysis and stomach content of fi sh, which maybe limited due to diffi culties in identifying food items, or, when they can be identifi ed, it is not safe to affi rm that such items would be assimilated and, therefore, they will contribute to production (Jennings et al. 1997).In the present work, isotopic ratios of carbon and nitrogen of muscles of adult Leporinus friderici individuals are compared with available information in the literature, concerning the isotopic ratios of C 3 and C 4 plants and also the diet of the species.Our hypothesis is that the variations of δ 13 C and δ 15 N are specifi c for each area of the reservoir and, therefore, equivalent to the composition of the food ingested by the species.

MATERIALS AND METHODS
Leporinus friderici (Bloch 1794) was collected monthly from April 1999 to March 2000 in nine sites distributed in the lower Corumbá River basin and its tributaries localized predominantly scrubland in the Cerrado Biome.The Corumbá River dammed in September 1996, forming the Corumbá Hydroelectric Reservoir.Corumbá Reservoir presents a surface area of 65 km 2 , a total volume of 1500 × 106 m 3 , an average depth of 23 m and a hydraulic retention time of 30 days (Luz- Agostinho et al. 2006).Sample sites were grouped in three characteristic biotopes defi ned considering the influence of Corumbá Reservoir: (1) streams lotic and semi-lotic characteristics upstream from the reservoir (COPE, MOIT, AREI and PFOZ); (2) stations inside the reservoir (LISA, JACU, CPIR and PIRA); and (3) river downstream from the dam (JUSA) (Figure 1).
Gillnets with different mesh sizes were used to capture fi sh.For each fi sh, standard length (Ls) and total weight (Wt) were obtained.A sample of the muscle close to the insertion of the dorsal fi n was removed from each individual.Leaves of C 3 (riparian vegetation) and C 4 (grasses) plants were sampled on bank areas.No aquatic macrophytes were found.Periphyton samples were washed in distilled water, fi ltered and maintained in fi berglass fi lter.Filters (GF/C Whatman) were previously undergone combustion at 550 • C for 4 hours.Filtered samples were rinsed in 1N HCl solution to remove carbonates.Particulate Organic Carbon (POC) and zooplankton samples were collected respectively with 25µm-and 75µm -mesh nets.These samples were also conditioned in fi berglass fi lters.
To determine the δ 13 C of phytoplankton is problematic due to contamination by carbon from vascular plants.Considering the results presented in Fry and Sherr (1984) for food webs of aquatic communities, the isotopic composition of phytoplankton was established through zooplankton with 1 • / •• fractionation per trophic To determine the relative importance of C 4 plants as source of carbon for adults L. friderici, the following equation was used (Forsberg et al. 1993): where: %C 4 = C 4 plants contribution; δ 13 C fi sh = mean value of δ 13 C for L. friderici; δ 13 C C3 = mean value of δ 13 C for C 3 plants; δ 13 C C4 = mean value of δ 13 C for C 4 plants.
According to defi ned mean values of carbon for the groups of plants of the area of influence of Corumbá Reservoir (Benedito-Cecilio et al. 2004), the most negative group (phytoplankton = -29,4 • / •• ) was used to calculate the maximum contribution of C 4 plants, while the less negative group (periphyton = -21,6 • / •• ) was used to calculate the minimum contribution.The percentage of the carbon originated from C 3 plants, by defi nition, was %C 3 = (%C 4 ) -100.
Trophic position (TP) based on δ 15 N was calculated according to formula (Vander Zanden et al. 1997 The enrichment of δ 15 N was calculated in 3.4 • / •• for trophic level (Fry 1988, Vander Zanden et al. 1997).Diet of the species was described in Hahn et al. (2004).Diet-based mean trophic position (MTP) was estimated by the formula (Winemiller 1990, Vander Zanden andRasmussem 1996): where: Cn = percentage contribution of the n th food item; Tn = trophic position of n th food item.

RESULTS AND DISCUSSION
For adult L. friderici (Ls above 17.5 cm), the δ 13 C mean value and standard deviation was -21.4 • / •• ± 1.7 (Table I).In Central Amazon, isotopic values superior to that were verifi ed for Schizodon fasciatus (-18.8 • / •• ), and an average of -28.8 • / •• for the entire fi sh assemblage (Forsberg et al. 1993).The low value could be related to the formation of Corumbá Reservoir that influenced the access to the sources of energy for the species.The analysis of stomach content, in river phase and immediately after Corumbá Reservoir fi lling (Ferreira et al. 2002, Luz-Agostinho et al. 2006), demonstrated variations in diet of species.In the river phase, the item fi sh was predominant in those the diet, whereas in the reservoir phase L. friderici ingested, basically, plants and a small proportion of fi sh and insects.These fi ndings are reinforced by Andrian et al. (1994) for the Paraná River floodplain, which classifi ed the species as opportunist.Isotopic variations of δ 13 C for primary producers analyzed by Benedito-Cecilio et al. (2004) in the same area are presented in the Figure 2. C 4 plants were strongly enriched in δ 13 C (-2.7 • / •• ± 0.7), but phytoplankton was the more negative group (-29.3 • / •• ± 1.6).Phytoplankton carbon is usually lighter than vascular plants carbon (Hamilton andLewis 1992, Victoria et al. 1992).However, in the area studies C 3 plants (riparian vegetation, C 3 grasses, periphyton and phytoplankton) presented signifi cant different isotopic ratios (F 3,22 = 22.59; p < 0.001).Periphyton presented positive values of δ 13 C (-21.6 • / •• ; ± 3.4).Nevertheless, planktonic and periphytic algae presented more positive average values if compared with those registered for the Amazonian basin, where mean values were -33.3 and -26.2 • / •• , respectively (Araújo-Lima et al. 1986).Spatial variations were not identifi ed in the composition of δ 13 C for adult L. friderici (Table I).Large variance was detected for isotopic values of carbon in lotic environments.Spatial differences in isotopic ratio for adults Colossoma macropomum and Prochilodus nigricans were verifi ed by Benedito-Cecilio et al. (2000) in Central Amazon, where the authors observed depletion of δ 13 C from downstream to upstream.Similarly, Thomas and Cahoon (1993) demonstrated significant differences in the ratio of δ 13 C and 15N for fi sh in coral reefs.For L. friderici, although signifi cant spatial differences in δ 13 N were not been identifi ed (F 2,12 = 3.18; p > 0.05), values were greater downstream from the dam (Table I).In spite of differences were not signifi cant, the species may be adopting specifi c trophic strategy for each environment.This is confi rmed if we consider that fi sh sampled downstream consumed more (Luz- Agostinho et al. 2006), denoting the use of protein originated from superior trophic levels.
Variations in the isotopic composition of L. friderici can also be due to the spatial variability in the isotopic ratio of the same food item.The spatial analysis, relative to the distance of to the dam, of the isotopic variations of δ 13 C for primary producers is presented by Benedito-Cecilio et al. ( 2004) (Figure 3).Although the studied area was relatively short (100 km), spatial correlations were detected for phytoplankton (r = 0.97; p < 0.05) and POC (r = 0.65; p < 0.05).In the system Solimões-Amazonas (between Tefé and Santarém), spatial differences were also verifi ed in δ 13 C of C 4 macrophytes (Benedito-Cecilio et al. 2000).Gradients of carbon stable isotopes can exist in ecosystems and this may have influenced the isotopic ratios of plants.The upstream stretches, not impacted by the reservoir, are 1.5 to 2 times more saturated in CO 2 and present higher values of δ 13 C than downstream (Lajtha and Marshall 1994).POC, which is composed by organic carbon originated from parts of plants and animals, can represent that reduction in the downstream values of δ 13 C.
The inverse tendency verifi ed for phytoplankton seems to be associated to diel variations in 13 CO 2 concentration (Martinelli et al. 1991).Jackson and Harkness (1987) found spatial variation in δ 13 C for plants.Such variations may happen due to environmental alterations induced in plant physiology, which means that, δ 13 C values could be related to environmental conditions (tem- perature, salinity, seasonality) and to geographical and temporal variations.All these have potential to induce alterations in plant metabolism.
In Corumbá Reservoir, the maximum and minimum contribution of C4 plants, for adults of Leporinus friderici, calculated integrating the variation of plants and fi sh were 47.7% and 2.4%, respectively.This is an expressive contribution of carbon from C4 plants, if compared to the fi sh assemblages studied in Central Amazon.In that ecosystem, only four species presented maximum contribution of C 4 plants superior to 38%.The largest proportion of C 3 carbon in adult fi sh could be, however, due to the preferential consumption of C 3 plants (Forsberg et al. 1993).
The low digestibility and the diminished nutritional value of C 4 plants for herbivores were demonstrated by Caswell et al. (1973).On the other hand, algal protein is highly nutritive and easily assimilated by most animals (Waslien 1979).Among C 3 plants, periphyton presented closer isotopic values to those observed for fi shes, corresponding to an important carbon source to L. friderici.
The intra specifi c variability in trophic position for the species, calculated from the obtained values of δ 15 N (Vander Zanden et al. 1997) and diet data (Luz- Agostinho et al. 2006), are presented in Table II.The proportion of ingested plant item is larger in rivers upstream from the reservoir (42.7%), which justifi es the smaller trophic level among there.However, in the reservoir, the ingestion of fi sh was 81.4%, while ingested plants contributed with 18.6%.Downstream from the dam, participation of plant item was even smaller (14.4%).Trophic position indicates how many times the biomass consumed by an organism have been metabolized along the food chain (Vander Zanden et al. 1997).In this case, the omnivorous behavior of the species, a characteristic of tropical ecosystems complexity, makes difficult the understanding of energy flow and mass transfer in aquatic ecosystems.The trophic position variability of the species can be attributed to the following factors or even to the combination of both: i) high flexibility in feeding species, already justifi ed by Andrian et al. (1994) for the Paraná River floodplain, and ii) variation in the trophic position of preys.In this last case, the diffi culty to identify prey is due to a characteristic of the species that removes pieces of fi shes when feeding.This impedes the determination of the prey trophic level (Luz- Agostinho et al. 2006).For an appropriate correction of this variation, experimentations to quantify the degree of trophic flexibility and to determine preferential prey are fundamental.Although the trophic position calculated with diet data was proportional to the one calculated with 15N values, the former elevated the trophic level of L. friderici in the food web, because estimated trophic positions were based on fi sh items belonging to the 2 nd (a) and to the 3 rd (b) trophic levels (Table II).
Determination of trophic position based on diet, compared to the use of δ 15 N, involves distinctions in the way as these methods integrate variations in trophic positions (Vander Zanden et al. 1997).The δ 15 N presents, in a more robust way, the integration in longer time, through the food web, the energy assimilated by lower trophic levels.However, a better estimate of the results obtained with δ 15 N is only possible based on diet composition.
Results obtained with the use of isotopes made possible a better understanding of the role of L. friderici in the flow of energy in the food web of the area of Corumbá Reservoir influence.The primary sources of carbon for the species, after the fi rst year reservoir fi lling, were constituted by periphyton and C 4 grasses.Although studies have not been conducted in the river phase, such sources might not have been the same during the two phases (before and after the formation of the reservoir), once the diet of the species presented similar item, but in different relative importance in the reservoir phase.On the other hand, trophic position of the species, in the adult phase, based on diet data and δ 15 N, ranks it above the second trophic level.However, the pattern of carbon flow and trophic dynamics in juveniles of this species may be distinct to those presented in this work.

ACKNOWLEDGMENTS
We are grateful to M.F.A and L.C. Gomes for valuable comments, to Núcleo de Pesquisas em Limnologia, Ictiologia e Aquicultura (NUPELIA) and FURNAS Centrais Elétricas S.A. for infrastructural support and Conselho Nacional de Desenvolvimento Científi co e Tecnológico (CNPq), for the scientifi c iniciation scholarship to ALP.
fi sh = mean value of δ 15 N for L. friderici; 5.7 = average δ 15 N for vascular plants; 3.4 = increase of trophic level for δ 15 N.