A high correlation exists between atmospheric deposition rate and concentrations of mercury (Hg) in fish (Fjeld and Rognerud 1993). Even the biota from lakes far away from point sources can have high levels of Hg contamination (Chen et al. 2005) which indicates its high bioaccumulation and biomagnification properties. All the biota accumulates Hg mainly in the form of methylmercury (MeHg) and it has been shown that MeHg contributes up to 98 % of Hg in fishes (Carrasco et al. 2011). MeHg is highly neurotoxic (Bloom 1992) with high bioaccumulation and biomagnification properties (Wang et al. 2010); and fish is the main route of MeHg poisoning to humans.

Despite the fact that mercury pollution in fish has attracted considerable attention worldwide, studies related to accumulation of contaminants in biota and their trophic transfer are still lacking in Nepal where subsistence as well as commercial fisheries is gaining popularity. Therefore, the present work was carried out to investigate the concentration of Hg in Lake Phewa, Nepal focusing mainly on the fish community. The present findings not only provide baseline information of Hg contamination, but also shed light on the bioaccumulation properties of this pollutant in commercially important fish species of Lake Phewa, Nepal.

Materials and Methods

Lake Phewa is situated at an elevation of 782 m approximately at the center of the country map, Nepal. The lake has a surface area of 4.35 km2 and an average depth of 8.6 m (Gurung et al. 2010) with a maximum depth of 22.5 m. It is a small and warm monomictic lake near a sub-metropolitan city (Pokhara), Nepal.

A total of 122 samples belonging to four species of fish: Tilapia Oreochromis niloticus, Spiny Eel Mastacembelus armatus, African catfish Clarias gariepinus and Sahar Tor putitora were identified in situ and collected. Following the measurement of total length (TL, cm) and total weight (TW, g), fillets just behind the dorsal fin (axial muscles) was collected after removal of the skin for the analysis of Hg (both THg and MeHg species), δ15N and δ13C. Stomach contents were collected and preserved in 70 % ethanol to characterize the dominant diets. The muscle samples were kept in a freezer until they were brought to the laboratory. Samples were analyzed in the laboratory of the Department of Biology, Hong Kong University of Science and Technology (HKUST) for the analysis of THg and MeHg. THg were also analyzed at Environmental Chemistry Section of the Department of Plant and Environmental Sciences, Norwegian University of Life Sciences (UMB).

At HKUST, fish muscle samples were freeze-dried and ground into fine powders. For the analysis of THg, all containers were vigorously cleaned with 4 N HCl. Approximately 0.2 g dried tissues were weighed and digested at 190°C with ‘aqua regia’ (2 mL HNO3:6 mL HCl) in a microwave digestion system. An aquilot of digested samples was taken and diluted as appropriate. Bromine monochloride (0.5 % v/v) was added to the diluted sample until stable yellow color was obtained, after which the samples were pre-reduced by addition of NH2OH·HCl. THg was quantified using the single gold trap amalgamation technique by Cold Vapour Atomic Fluorescence Spectroscopy (CVAFS, QuickTrace® 8000, USA). At UMB, THg has been analyzed using Flow Injection Mercury System (Perkin-Elmer, model FIMS 400) (details in Sharma et al. 2008).

For analysis of MeHg, approximately 40 mg of tissues were digested with 25 % KOH in methanol at 60°C for 3 h. MeHg in the extract was measured with an automated MeHg analytical system (MERX, Brooks Rand). Briefly, 20–50 μL of extract was buffered with sodium acetate at pH 4.9, and ethylated by sodium tetraethyl borate in a 40 mL Teflon line borate glass bottle. The quantification of MeHg was automatically carried out by the MeHg analyzer with gas chromatographic separation and pyrolysis following atomic fluorescence detection. The recovery of certified reference material (IAEA-436, tuna fish) when used in conjunction with sample batches was between 87.3  % ± 4.2 % for MeHg and 90.5 % ± 3.6 % for THg.

For the analysis of δ15N, samples were burned under high temperature to convert into N2 and measured for 15N and 14N using elemental analyzer in conjunction with Isotope-ratio mass spectrometry (IR-MS). Similarly, for δ13C, samples were burned under high temperature to convert into CO2 and measured for 13C and 12C. The δ15N was reported relative to atmospheric N2 and δ13C using global standard reference material (Pee Dee Belemnite or PDB). The accuracies of δ15N and δ13C analyses were ±<0.2 ‰ and ±<0.1 ‰, respectively. Details for the sample analysis at UMB for the analysis of δ15N and δ13C are described in Sharma et al. (2008).

The δ15N and δ13C are used to study the feeding relationships of organisms (McCutchan et al. 2003) where δ15N indicates the trophic position and δ13C indicates the carbon source. Stomach contents were analyzed in the Aquatic Ecology Centre at Kathmandu University under dissecting microscope. Mean volume percentages of prey items were measured in all fish species. 52 out of 105 analyzed fish had empty stomachs. A regression of log-transformed THg (logTHg) concentrations against δ15N values across all fish species was carried out to see if this gives an idea of the quantitative measure of biomagnification rate (Kidd 1998).

Comparisons of Hg concentrations between different fish species were performed by ANCOVA (statistic F) using length as covariate. Spearman Rank Order test was performed to evaluate the correlations. All the analysis were considered statistically significant at p < 0.05 unless otherwise stated.

Results and Discussion

Mercury concentrations in all fish samples were <0.31 mg kg−1 in the present study (Table 1). Most African freshwater systems sharing at least some fish species with Lake Phewa (e.g., C. gariepinus, O. niloticus) also showed low Hg concentrations (Desta et al. 2007; Tadiso et al. 2011). The concentrations of Hg in C. gariepinus and O. niloticus in Lake Phewa were comparable to the respective species (same length classes) in Lake Awassa (Desta et al. 2007) and Lake Ziway (Tadiso et al. 2011) in Africa. The reason for such a low concentrations of Hg in Lake Phewa could be the absence of local sources of Hg to the lake as also described by Tadiso et al. (2011) for Lake Ziway.

Table 1 Mean (± SD) and range of length (cm), weight (g), T-Hg (mg kg−1, ww), Me-Hg (mg kg−1, ww), δ15N (‰), and δ13C (‰) of four commercial fish species from Lake Phewa
Full size table

Regarding Hg speciation, earlier studies (e.g., Bloom 1992) indicated that most of the Hg in fish muscles from higher trophic levels is in the form of MeHg. The concentrations of MeHg followed the same trend as THg (Fig. 1a) because there was a significant positive correlation between THg and MeHg in fish muscles (r2 = 0.92, p < 0.001; Fig. 1b). In the present study, an average of 82 % of THg was found in the form of MeHg in fish fillets. Some recent studies indicated varied ranges (50 %–98 %) of MeHg in different fish species (e.g., Carrasco et al. 2011).

Fig. 1
figure 1

a Comparison of Hg concentrations (both THg and MeHg; mg kg−1, ww) in four species of fish (both in O. niloticus, M. armatus, C. gariepinus and only THg in T. putitora) from Lake Phewa; b regression between THg and MeHg in the fish community shows their relationships

Full size image

Mercury concentrations and body length showed significant positive correlations in C. gariepinus (r2 = 0.74; p < 0.001) and M. armatus (r2 = 0.64; p < 0.001) but not in other two species (Table 2). In general, there are significant positive relations between fish size and Hg concentrations in their muscle tissues (Gewurtz et al. 2011). Increasing trend (some significant and some non-significant) of Hg concentrations with body size in C. gariepinus have been reported in African lakes (Desta et al. 2007; Tadiso et al. 2011). In their studies, the normal diet of large size-classes of C. gariepinus included organisms from higher trophic levels, e.g. fish. The Hg concentrations in C. gariepinus in the present study showed a significantly increasing trend (Table 2) with the body size (TL). The diet composition of the larger individuals of this species was also dominated by the organisms from higher trophic positions (Fig. 2a). It has been recorded that the proportions of prey-fish in the diets of C. gariepinus increased as they grew in size (Fig. 2a, b) as also indicated by Desta et al. (2007) in African lakes. This phenomenon further justifies that the diet plays an important role in the increment of the mercury in fish flesh.

Table 2 The relationship between mercury concentrations (mg kg−1, ww) and stable isotope ratio of nitrogen (δ15N) for all fishes from Lake Phewa was calculated
Full size table
Fig. 2
figure 2

a Mean volume (%) of diet composition (stomach contents) of large-size class of C. gariepinus, M. armatus, O. niloticus, and T. putitora. b Mean volume (%) of diet composition (stomach contents) of small size-class of C. gariepinus, M. armatus, and O. niloticus

Full size image

There was a significantly lower concentration of THg detected in O. niloticus compared to T. tor (F = 9.974; df = 1,117; p < 0.001), whereas no statistically significant difference was detected between C. gariepinus and M. armatus (Fig. 1a). The predator fish species in Lake Phewa, e.g. T. putitora, M. armatus and C. gariepinus had higher concentrations of Hg in only one individual from each species. The individuals of M. armatus showed significant correlations between Hg concentrations and body size (r2 = 0.34; p < 0.001; Table 2). Both small and large size classes of this species had prey-fish in their stomach contents (Fig. 2a, b). It has been demonstrated that the dominant pathway of overall mercury accumulation in O. niloticus is its dietary exposure (Wang et al. 2010). Major proportion of diets from the lower trophic levels, such as plant parts, is one of the major factors for such a low concentrations of mercury in C. gariepinus and O. niloticus (Tadiso et al. 2011). The diet of T. putitora contained the highest percentage of fish followed by M. armatus (Fig. 2a, b). The aquatic plants were the only diet of large O. niloticus, whereas small individuals had a large proportion of insect larvae. C. gariepinus mainly fed on insect larvae irrespective of the size classes (Fig. 2a, b).

The relative trophic position, based on δ15N, indicated that M. armatus occupied the higher position whereas O. niloticus the lower (Fig. 3). The lowest δ15N was found for O. niloticus (1.35) and the highest in M. armatus (11.93). This indicates a food chain of about 3–4 trophic levels. All the individuals from the lowest trophic position (as indicated by the stomach contents as well as the δ15N signatures), O. niloticus, had plants as their dominant diet. The stomach contents of M. armatus, the species at the highest trophic position based on δ15N signatures, were dominated by animal diets (e.g., insects and prey-fish). In addition, δ13C signature of M. armatus did not indicate a shift in food source as they grow in size (Table 2). T. putitora, having the highest Hg concentrations, occupy the intermediate trophic position among the four species. This could be mainly due to the fact that only large individuals were represented in the present investigation. Further study is necessary for this species to find out the dynamics of Hg in all size-classes.

Fig. 3
figure 3

Simple food web structure based on the relationships between δ15N and δ13C in four fish species (O. niloticus, M. armatus, C. gariepinus and T. putitora) from Lake Phewa. Ranges of error bars indicate 95 % confidence interval from the mean; vertical bars for δ15N and horizontal bars for δ13C values

Full size image

The Hg levels in O. niloticus were found to decline with the increase in size in some African lakes (Desta et al. 2007; Tadiso et al. 2011) possibly due to shift in their diet; younger individuals feeding on zooplankton and switching to plant materials when they grew larger (Desta et al. 2007). The present investigation also found that there was a shift in diets in this species from 〈mixture of insect larvae and plants〉 to 〈plants only〉 when they grew in size (Fig. 2a ,b). This species did not show an increasing tendency of Hg concentrations with body size in the present investigation. This could mainly be due to the fact that O. niloticus fed mainly on plant materials as also indicated in the study conducted by Zengeya et al. (2011).

An effort to regress the logTHg versus δ15N produced an equation with slope 0.041 per  ‰ δ15N (p < 0.001; Table 2), however the food chain seems to have no direct feeding relationship among the fish community (Fig. 3). The resource partitioning was observed in this fish community with depletion of carbon in C. gariepinus and T. putitora compared to O. niloticus and M. armatus (Fig. 3). Our study also showed that T. putitora, although had higher concentration of Hg and fish contents in the diet, occupied the lower position in the food chain. This warrants the investigation of more fish that can act as a bridge between the investigated fish community. Hence, the slope of this regression should not be interpreted as the bio magnification rate of the fish community in this case.

$${\text{logTHg}}({\text{mg}}\,{\text{kg}}^{{-1}} ) = - 1.702 + 0.041\updelta^{15} {\text{N}}(\permille).$$

Fishes of Lake Phewa have shown low concentrations of Hg although water and sediments contain comparable Hg concentrations (own unpublished data) to other uncontaminated freshwaters around the globe (Ullrich et al. 2001). Here, diets of these individual could explain the Hg burden in muscle tissues (Ullrich et al. 2001), and warrants a detailed study including the entire food web.