Total mercury concentrations in fish from Urrá reservoir (Sinú river, Colombia). Six years of monitoring

Objective. The aim of this study was to monitor the total mercury (T-Hg) concentrations in fish from the Urrá reservoir, after impoundment. Materials and methods. Five fish species at different trophic levels were sampled from 2004 to 2009 and analyzed by cold-vapor atomic absorption spectroscopy for T-Hg concentrations in muscle tissue. Water quality parameters were evaluated. Results. The highest (1.39±0.69 μg/g ww) and lowest (0.15±0.02 μg/g ww) T-Hg concentrations were detected in Hoplias malabaricus (piscivorous) and Cyphocharax magdalenae (iliophagous/detritivorous) respectively, whereas Leporinus muyscorum (omnivorous) had an intermediate level (0.40±0.11 μg/g ww). The organic matter content in the water increased with time and depth, whereas dissolved oxygen and pH decreased. A covariance analysis (with fish length as a covariate) shows a steady increase of T-Hg levels in all the studied species after impoundment. Conclusions. The T-Hg concentrations in the evaluated fish species, increased after impoundment. The water quality variables showed conditions favoring Hg methylation and its biomagnification, this last was evident in the fish food chain of the reservoir.


INTRODUCTION
Among the metals of environmental concern, mercury (Hg) is considered a priority pollutant due to its ubiquity, persistency, accumulation, and toxicity.Monomethylmercury (MeHg) is an important form of Hg because it is a potent neurotoxin with lipophilic characteristics and is able to bind proteins, causing severe damage to vertebrates (1).Hg uptake by humans occurs through the contaminated food ingestion and is the main route of uptake for populations that depend on fishing because Hg transferred to the biotic compartment is bioaccumulated and biomagnified by fish (2).
Reservoirs are created for various purposes, including hydropower generation, irrigation, flood control, water supply, fishery production, navigation and recreation.Although reservoirs provide significant benefits, they also have potential socioeconomic and environmental impacts (3).It has been demonstrated that the transfer of Hg to the aquatic food chain, increases in freshwater impoundments (4).Indeed, increased Hg levels in fish from hydroelectric reservoirs have been frequently reported, and these findings also show that fish species at high trophic levels have the highest Hg concentrations (5,6).
The main source of Hg in reservoirs is the soil cover, which retains Hg bonded to organic matter.Other sources of Hg, such as dry deposition and watershed runoff, may be important.The timing and magnitude of the response in fish to Hg concentrations will vary depending on ecosystem-specific variables such as sulfur and iron levels, pH, the amount of flooded terrestrial carbon and Hg speciation (7).Given that the creation of hydroelectric reservoirs in the tropics generally involves the flooding of large areas of dense forest and the occurrence of high water temperatures, an increase in the activity of sulfate-reducing bacteria and plankton microorganisms in anoxic conditions, favored by the decomposition of submerged organic matter, also enhances MeHg production, primarily from sediments but also in the water column (6,8).
In Colombia, there is evidence of Hg contamination in fish from sites traditionally dedicated to and/ or influenced by gold mining (9)(10)(11).However, although there are currently 24 active reservoirs in Colombia (12), with 10 more planned over the years 2010 -2024 (13), only there are published studies about the potential impact of dams on Hg concentrations in fish and the potential risk for human health resulting from continuing consumption of fish, for one reservoir (14,15).In

MATERIALS AND METHODS
Study site.The Urrá hydroelectric power station is located on the Sinú River, in northwestern Colombia (N 7°49′51″ -8°00′56″, W 76°12′45″ -76°18′55″) (Figure 1).The station is situated 30 km south of the municipality of Tierralta, Department of Córdoba, Colombia.The reservoir is fed by the Sinú River (average inflow 350 m 3 /s) and its tributaries.Several of these tributaries are located in the Paramillo Nature Reserve catchment area, part of which was flooded in the year 2000 with the filling the reservoir.The flood area covers 7400 ha of tropical rain forest, with a monthly rainfall range from 20 to 350 mm, with an annual average of 2212 mm.A maximum temperature of 37.5, a mean of 28.2 and a minimum of 18.7°C have been recorded (16,17).
Esas especies incluyeron los carnívorospiscívoros Hoplias malabaricus (Moncholo) y Caquetaia kraussii (Mojarra amarilla), el six yearly sampling periods in January from 2004 to 2009, that is between four and nine years after filling the reservoir.The samples were collected with the help of local fishermen with a gill net.For each species, 3 -6 samples were collected annually within the reservoir, 3 samples approximately 3 km downstream from the reservoir and 3 samples at a comparison site located approximately 60 km downstream from the reservoir.The fishes were individually packed, stored in iceboxes at 4°C and shipped to the laboratory, where they were immediately measured (standard length, from the nose to the caudal fin base: (±0.5 cm)).One piece of white muscle was removed from below the dorsal fin for the determination of T-Hg.The muscle samples were kept frozen at -20°C until analysis.
From 2004 to 2008, water samples were collected at different depths (0, 5 and 32 m) using a Van Dorn bottle from a boat at one of the sites of maximum water depth of the reservoir (W station, Figure 1).The samples were used for the determination of water pH, dissolved oxygen (DO) and total organic matter (OM).In 2009, water and sediment samples were collected at different stations on the reservoir (S stations, Figure 1) for the analysis of T-Hg.The water samples (2 L per station) were collected with a polycarbonate Van Dorn bottle to 1 m below the surface of the water and then poured into clean, acid-washed polyethylene (HDPE) containers.
The samples were acidified with HNO 3 to pH<2 and were kept refrigerated until analysis within 1 week after collection.Sediment samples were obtained by lowering a Van Veen grab from a boat.At each station, four sediment subsamples were collected at all cardinal points within a 2 m radius from the reference point.The samples were placed in plastic bags, labeled and packed in ice, transported to the laboratory, and dried at 40ºC in a drying oven.

Determinación de mercurio.
El Hg-T en muestras sin filtrar de agua fue medido utilizando espectrometría de absorción atómica de vapor frío (CVAAS) luego de la digestión con una solución diluida de KMnO 4 -K 2 S 2 O 8 a 95°C (18).El límite de detección (tres veces la desviación estándar de 10 muestras testigo) fue 0.1 μg/L.pH, DO and OM determination.DO and pH were measured in situ using a portable multimeter (Thermo Scientific, Orion 3STAR and Orion 4STAR models, respectively).The instrument for measuring dissolved oxygen was calibrated at the laboratory, and the pH meter was calibrated at each sampling site with two buffer solutions.The organic matter content was estimated based on standard methods of APHA (20).

Statistical treatment of data.
To study the effects of site, species and year on T-Hg concentrations, an analysis of variance (ANOVA) and an analysis of covariance (ANCOVA) were performed after checking the normality and homoscedasticity (Shapiro-Wilk and Bartlett tests) of the error term.The length of the fish was used as a covariate in the ANCOVA.
If the null hypothesis (no factor effect) was rejected, a Duncan test was used to evaluate the differences between pairs of treatment means.For all statistical tests, p<0.05 was considered significant.The slopes of the changes in T-Hg with length for each species and year were determined by linear regression.The computations were performed with Statistica V 7.0 software.

RESULTS
T-Hg concentrations in fish.For all species, the T-Hg levels in the fish collected in the reservoir were higher than were those at the comparison site during all sampling years (Table 1 Las concentraciones de Hg-T en los sedimentos se reportaron con base en su peso seco, mientras que en los peces se expresan con base en su peso freso o húmedo.Todas las determinaciones de Hg se realizaron con un Espectrómetro de Absorción Atómica S-Serie 4 (Thermo Electron Corporation, Reino Unido).
Las pendientes del cambio del Hg-T con longitud para cada especie y años se determinaron por regresión linear.Los cómputos se realizaron con el software Statistica V 7.0 T-Hg concentrations of fish from Urrá reservoir were normalized with the length for evaluate their evolution during the study period (Figure 2).The concentrations increased from 2004 to 2008 in the carnivorous fish, and from 2004 to 2007 in the omnivorous and detritivoruos fish.
Subsequently, the T-Hg levels showed a tendency to decrease.

Water parameters and T-Hg in fish.
The changes in the water parameters from 2004 to 2008 showed an increase in the OM content    contaminación incrementó de especies detritívora a piscívora, con los peces omnívoros mostrando concentraciones intermedias de Hg-T.Las concentraciones de Hg-T de los peces del ensamble Urrá fueron normalizadas con la longitud para evaluar su evolución durante el periodo de estudio (Figura 2).Las concentraciones incrementaron entre 2004 y 2008 en los peces carnívoros, y entre 2004 y 2007 en los peces omnívoros y detritívoros.Subsiguientemente, los niveles de Hg-T mostrar una tendencia a disminuir.
Los resultados del ANCOVA (modelo de pendientes separadas, con la longitud como covariable) se presentan en la tabla 4. Los resultados confirman en general la dependencia de las concentraciones de Hg-T, de los hábitos dietarios de los peces y el año.Se encontraron excepciones a este patrón, luego de remover la influencia de la longitud, T-Hg and fish length.A fitted linear model   of the T-Hg concentration as a function of fish length (Table 3) showed a significant correlation (p<0.05) between these variables for the piscivorous fish in all years analyzed and at least one year for omnivorous and detritivorous species.The piscivorous fish showed the steepest slopes, followed by the omnivorous fish and detritivorous fish.The slopes increased during the study period for all species. The

DISCUSSION
The higher T-Hg concentrations of fish from the Urrá reservoir than in the comparison site (Table 1) show the Hg accumulation in reservoir, and their increase with the length of fish (Table 3) suggest an Hg biomagnification process within the aquatic food web.These results are remarkable because the levels of T-Hg detected in the fish samples of Urrá reservoir, are generally higher for the same trophic level than those reported in other regions of Colombia and Brazil in the presence of strong mining activity (Table 5).Note that the carnivorous and piscivorous fish investigated in Lago Manso, Brazil, had on average, twice the length of those evaluated in the present study.16) and the first and second place in the fish consumption into diet of children and women of childbearing age of the riverside population from Urrá reservoir in 2010 (15), suggest an increased risk for mercury exposure, which must be studied promptly.
Although there are no punctual sources of Hg in the study area, a diffuse source exists in a nearby watershed (San Jorge River basin), where ferronickel, coal and gold mining occur.This activity has expanded
En conclusión, las concentraciones de Hg-T en las especies de peces evaluadas en el embalse Urrá incrementaron después del represamiento.Las variables de calidad de agua mostraron condiciones que favorecen la metilación del Hg y su biomagnificación, esta última fue evidente en la cadena alimenticia de los peces del embalse, lo in recent years due to an increase in the ferronickel production, the building of the new coal-based thermoelectric plant (23) and the opening of many artisanal fronts of gold mining.Recent research in this area has shown high atmospheric Hg concentrations that are transported to other areas outside the mining areas (24).Thus, the high T-Hg levels in fish could reflect atmospheric transport from mining areas near the reservoir, or could simply result from a higher mobility of Hg in the area investigated.
In this regard, the data about to water parameters collected from the Urrá reservoir, showed an increase in the anoxic conditions, acidification and productivity based on OM concentration (Figure 3); conditions that enhance the rates of Hg methylation (8) and subsequent biomagnification in food chain.Indeed, it has been reported that acidic pH values increase mercury uptake by freshwater unicellular algae, primary producers in aquatic food chains (25).In the same direction, are highlighted the significant (p<0.05)correlations recorded in this study between T-Hg concentrations of C. kraussii with OM, pH and DO values of water (Table 2).
On the other hand, the T-Hg concentration in water (0.14 ± 0.04 µg/L) measured in 2009 in the Urrá reservoir was higher than the background values (0.003 µg/L) reported for Lago Manso reservoir in Brazil (6).Most likely, this difference results from local sources of Hg contamination.This result could reflect atmospheric transport from gold mining areas not far from the reservoir, or it could simply result from higher mobility of Hg in the areas investigated as was describe above.However, the results obtained were lower than those reported in marshes or rivers of heavily contaminated areas, such as the Grande marsh in Colombia (average 0.33 ± 0.03 µg/L) (26) and rivers from Indonesia (up to 0.25 µg/L) (27), which are located around gold mining areas.
The Hg concentrations in the sediment samples collected in this study (173.2 ± 14.1 µg/kg dw) were lower than the values of Hg in sediment for other contaminated areas in Colombia, such as Ayapel marsh (243 ± 62 µg/kg dw) (28) and marshes of Mina Santa Cruz (up to 355 µg/kg dw) (29), but greater than those reported in areas such as the marshes from Caimito village in the San Jorge River basin, Colombia (155 ± 16 ug/kg dw) (11), which are also impacted by gold mining.
According to the above description, The Hg concentrations in the waters and favorable physicochemical characteristics present in the Urrá reservoir, may produce a greater efficiency of metal uptake by the fish and cause higher concentrations compared with other ecosystems.
In conclusions, the T-Hg concentrations in the evaluated fish species of the Urrá reservoir increased after impoundment.The water quality variables showed conditions favoring the Hg methylation and its biomagnification, this last was evident in the fish food chain of the reservoir, which poses a risk for riverside population that consume these fish.This study illustrates one of the potentially negative consequences of the construction of hydroelectric reservoirs and shows the need to include these potential environmental costs in the assessment of the impact of future projects of this type.

a
Food habit: C = Carnivorous, O = Omnivorous, DI = Detritivore iliophagous b Ratio T-Hg in Urrá/T-Hg at comparison site; ns = no specimens caughtRESULTADOSConcentraciones de Hg-T en peces.Para todas las especies, los niveles de Hg-T de los peces recolectados dentro del embalse fueron mayores que los del sitio de comparación durante todos los años de muestreo (Tabla 1).En orden decreciente de niveles de Hg-T (y con diferencias respecto al sitio de comparación), las especies pueden organizarse en el siguiente orden H. malabaricus > C. kraussii > L. muyscorum > P. magdalenae ≥ C. magdalenae.El grado de and decrease in DO and pH with time and depth (Figure3).All variables were strongly and significantly correlated (R ≥ 0.991).The T-Hg concentrations of samples collected in 2009 were 173.2 ± 14.1 µg/kg dw in sediment and 0.14 ± 0.04 µg/L in water.The analysis of correlation between T-Hg concentration of fish with OM, pH

Figure 2 .
Figure 2. Values of T-Hg concentrations in fish from the Urrá reservoir normalized to unit length (cm) vs year.

Table 2 .
Correlation coefficients of T-Hg concentrations of fish with organic matter (OM), dissolved oxygen (DO) and pH of water.

Table 3 .
Results of linear fitting of T-Hg concentrations vs length of fish.
a Food habit: C = Carnivorous, O = Omnivorous, DI = Detritivore iliophagous Note: Significant (p<0.05)sum of squares of whole model/sum of squares of residuals are indicated in bold.ns = no specimens caught

Table 5 .
Comparison of T-Hg levels found in fish in this study (in bold) with values from other studies in Colombia (Col) and Brazil (Br).