Ecotoxicology (2007) 16:341–346 DOI 10.1007/s10646-007-0138-0
Annual flooding and fish-mercury bioaccumulation in the environmentally impacted Rio Madeira (Amazon) Wanderley R. Bastos Æ Ronaldo de Almeida Æ Jose´ G. Do´rea Æ Antonio C. Barbosa
Accepted: 13 February 2007 / Published online: 14 March 2007 Springer Science+Business Media, LLC 2007
Abstract Regular annual flooding of the Amazonian rivers changes the aquatic environment affecting fish feeding strategies. The Rio Madeira has been greatly impacted by deforestation for agricultural projects, damming for a hydroelectric power plant, and alluvial gold extraction. We studied fish-Hg concentrations within defined weight ranges of representative species at the top of the food web, comparing high and low water seasons. Selected piscivorous species (Cichla spp, Hoplias malabaricus, Pinirampus pirinampu, Serrasalmus spp) showed a large variation of Hg concentrations but only ‘‘traı´ra’’ (Hoplias malabaricus) showed a statistically significant difference between seasons. However, the bioaccumulation trends during high and low waters were similar for ‘‘tucunare´’’ (Cichla spp) and ‘‘traı´ra’’ (Hoplias malabaricus) but different for ‘‘piranhas’’ (Serrasalmus spp), ‘‘barba chata’’ (Pinirampus pirinampu) and the detritivorous Prochilodus nigricans. Fish-Hg bioaccumulation is species specific; changes in feeding strategies brought by flooding seasons can change the bioaccumulation pattern without systematically affecting the overall accrual of methylmercury in tertiary consumer species. It appears that naturally occurring Hg and the high sediment load of the Rio Madeira are
W. R. Bastos R. de Almeida Laborato´rio de Biogeoquı´mica Ambiental, Universidade Federal de Rondoˆnia, Porto Velho, RO, Brazil J. G. Do´rea (&) Faculdade de Cieˆncias da Sau´de, Universidade de Brası´lia, C.P. 04322, Brasilia, DF 70919-970, Brazil e-mail: [email protected]
A. C. Barbosa Instituto de Quı´mica, Universidade de Brası´lia, Brasilia DF, Brazil
secondary factors in the Hg bioaccumulation pattern of fish species at the top of the food chain. Keywords Inundation Floodplain Fish Hg Gold mining
Introduction The Rio Madeira is the largest tributary of the Rio Amazonas basin (Fig. 1). Its catchment spreads over an area of 1,390 · 103 km2, representing 14% of the total area of the Brazilian Amazon forest and 15% of the water that the Rio Amazonas delivers to the Atlantic Ocean (Latrubesse et al. 2005). Well-defined seasonal inundation causes waters to raise as much as 20 m (March–April, high-water season) above the low-water level in August–September (Fig. 2). This fairly predictable seasonal fluctuation in water level is attributed to a rainfall of 2,000–2,200 mm/ year and melting of Andean snow-caps. Its basin runs 25% in the Andes, 27% on the Brazilian shield and 48% on the plain (Maurice-Bourgoin et al. 2000). The Rio Madeira tributaries, draining the Andes, are characterized by highsuspended load and high sediment yields. Latrubesse et al. (2005) summarized data for the Rio Madeira basin showing that it contributes around 50% of the total suspended load transported by the Rio Amazonas (estimated values ranging between 248 and 600 million tons/year). In the last 30 years the Rio Madeira basin has been impacted by anthropogenic activities, including deforestation for agriculture projects, damming for a hydroelectric power plant, and alluvial gold extraction, directly affecting Hg availability to aquatic life. Remobilization of Hg from bottom sediments plus re-emission from soils due to landuse changes are thought to keep Hg concentrations high in
W. R. Bastos et al.
Fig. 1 Map of the Rio Amazon showing the Rio Madeira basin
Water level (m)
14 12 10 8 6 4 2 1
Fig. 2 Annual average of water levels of the Rio Madeira at Porto Velho (adapted from CPRM, 2002)
the Rio Madeira fish (Bastos et al. 2006). At present, alluvial gold extraction is scarce and declining (Bastos et al. 2006). Despite sediment removal being recommended as a deterrent to methylation potential of aquatic environments (Mailman et al. 2006), metallic Hg used for gold amalgamation was originally thought to raise fish-Hg levels in Amazon rivers (Dorea 2003). During the gold rush, the amount of sediment removed was substantial, but it did not result in lower fish-Hg concentrations. Current fish-Hg concentrations are similar to those measured during the gold rush (Bastos et al. 2006). Natural Hg in soil is the principal source of Hg present in the Rio Madeira ecosystem, but human activities are associated with soil-Hg release. Maurice-Bourgoin et al. (2000) reported a gradient of water-Hg concentrations that
increased from the glacial waters of head rivers (Rio Beni, Bolivia) to the lower Rio Madeira (Porto Velho, Brazil). More recently, in the Rio Madeira basin, Almeida et al. (2005) reported that soil-Hg loss was directly related to Hg concentrations and that estimated Hg losses were only significant between pasture and forest soils in the topsoil. Indeed, it has been shown that in the Amazonian soils only Hg+2 occurs and organically bound Hg predominates (Valle et al. 2005). During seasonal variations in water levels the net monomethyl (MMHg) production is influenced by variation in depth and location of sampling sites. The methylation potential was one order of magnitude higher in C-rich macrophyte and flooded forest (Guimara˜es et al. 2000). In Amazonian fish, MMHg is transferred more through the aquatic food chain than in the form of water-borne Hg (Barbosa et al. 2003). Predatory fish end up with large amounts of MMHg that are bioaccumulated as a function of fish size (Barbosa et al. 2003). The large fish biomass of the Amazonian rivers does not depend upon primary production for food. Annual inundations facilitate the allochthonous input from the terrestrial ecosystem (Gragson 1992). Based on stomach contents, Gragson (1992) estimated that 47% of fish food items are terrestrial in origin. Fish feeding behavior adapts to changes in habitat due to seasonal inundation of the rain forest. Therefore, fish feeding strategies depend on terrestrial (plants and invertebrates) and aquatic (plant, invertebrates, smaller fish/ crustacean) life (Gragson 1992). As a consequence, fishMMHg bioaccumulation varies considerably with foodchain structure and length. Flooded areas increase forage ranges and interspecies interactions, which in turn influence
Annual flooding and fish-Hg bioaccumulation
nutrient availability and growth rates of some Amazonian fish. It was shown that, in some species, growth-rates are 60% faster during the inundation season (Junk 1985). Fish-Hg bioaccumulation studies in the Amazon tropical forest lack a solid knowledge base. The ecology of the Amazon aquatic environment is complex; despite the environmental changes brought by annual flooding, the knowledge of Hg transfer has been built on a poor or non-existent conceptual framework of the functioning of this complex ecosystem. MMHg availability at the base of the aquatic food chain modulates the MMHg levels in fish at the top of the chain. This is not so predictable in the special situation found in the Amazon, where soils are annually flooded. Seasonal inundation of this kind can affect methylation rates, increase the forage ground and the interaction between terrestrial and aquatic life; this in turn may lead to changes in Hg transfer up through the food chain (Dorea et al. 2006). Decomposition of organic matter and Hg leaching from soil in the impacted basin of the Rio Madeira can change during flooding and low-water seasons, and the uncertainty surrounding these shifts has led to equivocal assumptions when managing fish-Hg information crucial for public health. Little is known about Hg bioaccumulation as a function of hydrological cycles in the complex Amazonian ecosystem of regular flooding (Dorea et al. 2006). Because food is the dominant pathway of Hg uptake by fish, small changes in the environmental MMHg availability can be amplified as a function of the food chain length. Therefore we studied predatory species of the impacted Rio Madeira, with distinct feeding strategies, to evaluate the influence of seasonal inundation on fish MMHg bioaccumulation.
Materials and methods As part of an ongoing project for monitoring Hg contamination of the Rio Madeira basin we sampled fish, soil, and sediment during study campaigns from 1997 to 2004. Previous publication described the protocol for fish sampling and determination of Hg concentrations (Bastos et al. 2006); fish were caught in the Rio Madeira between Porto Velho and where it joins the Rio Amazon, at the localities shown in the map (Fig. 1). Fish was caught by local professional fishermen and species were identified according to Santos et al. (1991); after that they were weighed and length measured, then approximately 20 g of muscle samples were immediately cut, frozen and transported to the laboratory at the University of Rondoˆnia for analysis. Mercury determination was done in the samples after treatment with H2SO4:HNO3 (1:1) solution and KMnO4 (5%) for oxidation. For 500 mg of sample, 5.0 mL of acid
mixture was added and digested in a digestion block for 60 min (Tecnal—Mod.007A, Piracicaba, Sa˜o Paulo, Brazil). After digestion, 4.0 ml of a KMnO4 solution (5%) was added to the sample, leaving it for more 30 min in the digestion block. After cooling until the temperature reached 25C (room temperature), drops of hydroxylamine solution at 12% were added and the sample was transferred to a volumetric flask of 10.0 ml with ultra-pure H2O (Milli-Q Plus, Millipore, Bedford, MA, USA). All glassware was washed clean in 10% HNO3 and rinsed with ultra-pure H2O. Total Hg measurements were carried out by cold vapor atomic absorption spectrophotometry (Flow Injection Mercury System-FIMS-400-Perkin Elmer, Ueberlingen, Germany) (Malm et al. 1989; Bastos et al. 1998). Precision and accuracy of Hg determinations were assured by the use of internal standards prepared in our laboratory against certified reference materials (Dogfish Muscle-DORM-2-National Research Council of Canada, Ottawa) and used in intercalibration exercises among Brazilian laboratories. It is essential to compare fish-Hg concentrations within same species and between groups with similar body weight. For each species, we selected individuals within a similar body weight range to compare Hg concentrations in the muscle from fish collected during the high and the low water seasons. In order to account for differences related to fish weight between species and Hg concentrations, we estimated the mass increase factor (MIF) between species. Because possible bias caused by relationship of fish size and Hg bioaccumulation, we compared size differences between species as mass increase factor (MIF). We calculate the mass increase factor as MIF = Maximum fish mass/Minimum fish mass. Therefore, differences in variability of achieved growth and biodilution/bioconcentration of Hg can be compared within and between species. To compare the effects of seasonal flooding on Hg concentrations of different fish species, we performed the nonparametric Wilcoxon test. Statistical analysis and data summarization (mean, SD) were performed using an SAS (SAS Institute, Cary, NC, USA). We used Pearson correlation between variables of fish weight and Hg concentrations. For the statistical test, p < 0.05 was considered significant.
Results Scatter plots of fish weight and Hg concentrations for piscivorous species caught in the Rio Madeira during flooding and low waters are illustrated in Fig. 3, while fish mass and mean Hg concentrations are summarized in Table 1. In order to contemplate different feeding strategies a detritivorous species (Prochilodus nigricans) was shown along with the piscivorous species (Table 1 and
W. R. Bastos et al.
Total Hg, µg/g
0.3 0.2 0.1 0.0 500
3 High water
Total Hg, µg/g
Total Hg, µg/g
High water Low water
1500 High water
Total Hg, µg/g
Discussion High water
Total Hg, µg/g
Low water 1.5
Fish weight, g
Fig. 3 Fish Hg concentrations as a function of weight during high and low waters of the Rio Madeira
Fig. 3). In the studied species, within the controlled weight range, a significant correlation between fish weight and Hg concentration was observed during low water for Serrasalmus spp (p < 0.0001) and during high water for Pinirampus pirinampu (p = 0.049). Coincidentally, in these species, the trend in Hg bioaccumulation (as a function of fish weight) followed different patterns between seasons (Fig. 3); thus denoting difference in habitats and attendant feeding strategies. Comparing flooding and low-water seasons, the trends of fish-Hg bioaccumulation were similar only for ‘‘tucunare´’’ (Cichla spp) and ‘‘traı´ra’’ (H. malabaricus), albeit with no significant correlation. Because we could not assess fish growth rate we controlled for fish weight differences by choosing fish with the same range of body weight; within each species we estimated the mass increase factor (MIF). The fish MIF were respectively 19 (weight range: 100–1,900), 10 (weight range: 200–2,000), 5 (weight range: 192–1, 450), 3 (weight range: 426–1,298) and 2.7 (weight range: 390–1,051) for Serrasalmus spp, Cichla spp, P. nigricans, P. pirinampu and H. malabaricus (Fig. 3). Only H. malabaricus showed significant differences between seasons for both fish weight and fish Hg concentrations (Table 1). Although the fish weight range was similar between seasons the MIFs illustrate the large variation between species. While Serrasalmus and H. malabaricus had similar mean weights their MIF were quite different (19 and 2.7 respectively); despite a larger weight interval, Serrasalmus specimens showed consistently higher Hg concentration values. Collectively, these results suggest that variation in habitats due to regular flooding modulates bioaccumulation of Hg; Amazonian fish at the top of the food chain adapting to changes in food resources may explain changeable Hg biomagnification patterns.
Despite the Rio Madeira high sediment load and anthropogenic impacts (deforestation, agriculture, hydroelectric reservoir, and alluvial gold mining) on natural Hg release, mean fish-Hg concentration in tertiary-consumer species was not systematically different between the two seasonal habitats (flooding and low waters). As shown in Table 1, the trophic level of the species is the modulator of levels of fish-Hg concentrations. Mean Hg concentrations of predatory species were much higher than the Hg concentration of the detritivorous species (P. nigricans). Environmental factors related to natural-Hg release from soil and physicochemical characteristics of high and low waters of the Rio
Annual flooding and fish-Hg bioaccumulation Table 1 Comparison of fishHg concentrations in selected species (with similar weight range) during high and low waters of the Rio Madeira
Serrasalmus spp Weight, g
[Hg], lg/g Pinirampus pirinampu
Cichla spp Level of statistical significance by the nonparametric Wilcoxon test a
p = 0.0510
p = 0.0129
Madeira are secondary to food-Hg factors related to the length of the aquatic food chain. Adaptation to changing food resources can be monitored as fish-Hg biomagnification trends in tertiary consumers. This conforms to studies done in the aquatic environment of the Rio Negro (Dorea et al. 2006); Sampaio da Silva et al. (2005) used C and N stable-isotope ratios and reached similar conclusions in fish of lotic waters of Amazon lakes (Rio Tapajo´s). In Amazonian rivers, fish-Hg concentration is primarily influenced by MMHg in the aquatic food chain. Diet modulates fish growth rates, which in turn can also influence fish-Hg concentrations (Mailman et al. 2006). We controlled for this confounding variable by comparing fish of equivalent size interval within the same species; we also calculated a factor related to differences in fish weight (MIF) to partly compensate for the lack of information on fish growth rate. Thus, we can compare Hg bioaccumulation between species at different stages of development. Despite a higher MIF for tucunare´ (10) than for traı´ra (3), these species showed comparable mean Hg concentrations (Table 1). In the high Hg environment characterized by high methylating potential of the Amazon soil and decaying biomass, species interactions occur during high and low water seasons but do not change the overall steady state of fish-Hg bioaccumulation. The complex interaction of biological factors (fish weight and diet) is species specific and is reflected in the variability of fish-Hg bioaccumulation patterns of the high or low water habitats. The nature of the interactions between plant and animal used as food by fish (nearly 5,000 species) is even more complex in the dynamic environmental changes brought about by hydrological cycles of the Rio Amazon basin
(Junk 1985). It appears that Amazonian fish are facultative feeders on a seasonal basis; stomach contents showed that fish classified as frugivorous can consume arthropods and that some seed-eating piranhas display specialized behavior for preying on fins and scales from other fish (Goulding et al. 1988). Thus, variability in prey-Hg levels caused by modification of fish habitats explains the variability in fishHg bioaccumulation (Fig. 3). Goulding et al. (1988) give a detailed description of the feeding strategy for fish feeding on detritus. Some species are selective and lay down a large fat mass during the high water period. In this study, the detritivorous species (barba chata) showed no significant difference in mean Hg concentration but apparently had a distinct pattern of Hg bioaccumulation associated with high or low water season. In the Rio Amazon basin, the association of fish size and Hg-concentration has been reported in a seemingly inconsistent way. Such inconsistencies have been partly explained as shortcomings in study protocols (Barbosa et al. 2003). In the Rio Madeira, consistent with the present study, Malm et al. (1997) showed no significant correlation between tucunare´ size and Hg concentrations. Indeed, tertiary consumers can be facultative feeders adapting their feeding strategies to changes in food (and attendant MMHg) availability. In another study, consistent with the present one, Dorea et al. (2006) showed no significant correlation between fish size and Hg concentration for tucunare´, traı´ra and piranha during high and low waters of the Rio Negro. Indeed, fish adaptation to food resources brought by the seasonal flooding could partly explain inconsistencies in the Hg concentrations of tucunare´, traı´ra and piranha (Dorea et al. 2006).
In the studied species, linear correlation between fish weight and Hg concentration did not occur in a systematic pattern. Indeed, for some species (Serrasalmus spp, P. pirinampu) there was an inverse trend in Hg bioaccumulation between the seasonal habitats (Fig. 3). Mean fishHg concentration by a secondary consumer (P. nigricans), or in the tertiary consumers (predatory fish), are not directly influenced by annual inundation. Earlier studies in the Rio Madeira led researchers to conclude that observed fish-Hg variability represented contamination caused by the direct release of Hg during gold extraction. However, Lechler et al. (2000) surveyed 900 km of the Rio Madeira and concluded that elevated water-Hg levels are due to natural sources and biogeochemical processes. MMHg acquisition and retention by fish, however, depends on fish feeding strategy. We have showed that neither gradient (3-fold) of Hg in the water column nor DOC (8-fold) was proportional to fish-Hg in non-predatory species (Barbosa et al. 2003). However, a 10-fold increase in fish-Hg concentrations from herbivorous vs. predatory species was present at all levels of physico-chemical parameters of the Rio Negro waters (Barbosa et al. 2003), which is relatively less impacted than the Rio Madeira. Deforestation and agricultural practices associated with higher sediment loads contribute to natural Hg load in the Rio Madeira ecosystem (Almeida et al. 2005; Maurice-Bourgoin et al. 2000). Despite that, there are comparable patterns of seasonal fish-Hg bioaccumulation between the two rivers (Dorea and Barbosa 2007).
Conclusion Fish of the Rio Madeira adapt to water habitat (flooding or low waters) by showing variations in patterns of Hg bioaccumulation. Acknowledgements This work was partly supported by the National Research Council of Brazil-CNPq (PPG7, project-556985/ 2005-2).
References Almeida MD, Lacerda LD, Bastos WR, Herrmann JC (2005) Mercury loss from soils following conversion from forest to pasture in Rondonia, Western Amazon, Brazil. Environ Pollut 137:179– 186 Barbosa AC, Souza JR, Do´rea JG, Jardim W, Fadini P (2003) Mercury biomagnification in tropical black water, the Rio Negro, Brazil. Arch Environ Contam Toxicol 45:235–246 Bastos WR, Malm Olaf, Pfeiffer WC, Cleary D (1998) Establishment and analytical quality control of laboratories for Hg determination in biological and geological samples in the Amazon, Brasil. Cieˆnc Cult 50:255–260 Bastos WR, Gomes JP, Oliveira RC, Almeida R, Nascimento EL, Bernardi JVE, Lacerda LD, Silveira EG, Pfeiffer WC (2006)
W. R. Bastos et al. Mercury in the environment and riverside population in the Madeira River Basin, Amazon, Brazil. Sci Total Environ 368:344–351 CPRM (2002) Companhia de Pesquisa de Recursos Minerais. Servic¸o Geolo´gico do Brasil—Perspectivas do Meio Ambiente do Brasil; Rio de Janeiro, Brasil, pp 27 Dorea JG (2003) Fish are central in the diet of Amazonian riparians: should we worry about their mercury concentrations? Environ Res 92:232–244 Dorea JG, Barbosa AC, Silva GS (2006) Fish-mercury bioaccumulation as a function of feeding behavior and hydrological cycles of the Rio Negro, Amazon. Comp Biochem Physiol 142:275–283 Dorea JG, Barbosa AC (2007) Anthropogenic impact of mercury accumulation in fish from the Rio Madeira and Rio Negro Rivers (Amazonia). Biol Tr El Res (in press) Goulding M, Carvalho ML, Ferreira EG (1988) Rio Negro: rich life in poor water: Amazonian diversity and floodplain ecology as seen through fish communities. SPB Academic Publishing The Hague, p 200 Gragson TL (1992) Fishing the waters of Amazoˆnia: Native subsistence economies in a tropical rain forest. Am Anthrop 94:428–440 Guimaraes JR, Roulet M, Lucotte M, Mergler D (2000) Mercury methylation along a lake-forest transect in the Tapajo´s river floodplain, Brazilian Amazon: seasonal and vertical variations. Sci Total Environ 261:91–98 Junk WJ (1985) Temporary fat storage, an adaptation of some fish species to the water level fluctuations and related environmental changes of the Amazon River. Amazoniana 9:315–351 Latrubesse EM, Stevaux JC, Sinha R (2005) Tropical rivers. Geomorphology 70:187–206 Lechler PJ, Miller JR, Lacerda LD, Vinson D, Bonzongo JC, Lyons WB, Warwick JJ (2000) Elevated mercury concentrations in soils, sediments, water, and fish of the Madeira River Basin, Brazilian Amazon: a function of natural enrichment? Sci Total Environ 260:87–96 Mailman M, Stepnuk L, Cicek N, Bodaly RA (2006) Strategies to lower methyl mercury concentrations in hydroelectric reservoirs and lakes: A review. Sci Total Environ 368:224–235 Malm O, Pfeiffer WC, Bastos WR, Souza CMM (1989) Utilizac¸a˜o do Acesso´rio de Gerac¸a˜o de Vapor Frio para Ana´lise de Mercu´rio em Investigac¸ o˜es Ambientais por Espectrofotometria de Absorc¸a˜o Atoˆmica. Cieˆnc Cult 41:88–92 Malm O, Guimara˜es JRD, Castro MB, Bastos WR, Viana JP, Branches FJP, Silveira EG, Pfeiffer WC (1997) Follow-up of mercury levels in fish, human hair and urine in the Madeira and Tapajo´s basins, Amazon, Brazil. Water Air Soil Pollut 97:45–51 Maurice-Bourgoin L, Quiroga I, Chincheros J, Courau P (2000) Mercury distribution in waters and fishes of the upper Madeira rivers and mercury exposure in riparian Amazonian populations. Sci Total Environ 260:73–86 Sampaio da Silva D, Lucotte M, Roulet M, Poirier H, Mergler D, Oliveira Santos E, Crossa M (2005) Trophic structure and bioaccumulation of mercury in fish of three natural lakes of the Brazilian Amazon. Water Air Soil Pollut 165:77–94 Santos GM, Ferreira EJG, Jegu M (1991) Cata´logo dos Peixes do Rio Jamari, Rondoˆnia. Instituto Nacional de Pesquisas da Amazoˆnia/ INPA. Manaus, Amazonas, 123 Valle CM, Santana GP, Augusti R, Egreja Filho FB, Windmoller CC (2005) Speciation and quantification of mercury in Oxisol, Ultisol, and Spodosol from Amazon (Manaus, Brazil). Chemosphere 58:779–792