- Email: [email protected]
Mercury bioacumulation in four tissues of Podocnemis erythrocephala (Podocnemididae: Testudines) as a function of water parameters
S CI EN CE OF T H E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m
w w w. e l s e v i e r. c o m / l o c a t e / s c i t o t e n v
Mercury bioacumulation in four tissues of Podocnemis erythrocephala (Podocnemididae: Testudines) as a function of water parameters Larissa Schneider a,⁎, Lauren Belger b , Joanna Burger c , Richard C. Vogt a a
Department of Freshwater Biology, Instituto Nacional de Pesquisas da Amazônia (INPA), Brazil Department of Ecology, Instituto Nacional de Pesquisas da Amazônia (INPA), Brazil c Department of Ecology, Evolution, and Natural Resources, and Environmental and Occupational Health Sciences Institute, Rutgers University, USA b
AR TIC LE D ATA
ABSTR ACT
Article history:
A number of environmental factors influence the dynamics of Hg in aquatic ecosystems, yet
Received 21 June 2008
few studies have examined these factors for turtles, especially from South America. Red-
Received in revised form
headed river turtle (Podocnemis erythrocephala) is easy to capture in the black waters of Rio
24 September 2008
Negro, making it the turtle species that is consumed most often by people of the region. In
Accepted 25 September 2008
this study, environmental factors and turtle size were investigated to determine their influence on the Hg concentration in blood, muscle, liver and carapace of the red-headed
Keywords:
river turtle. Factors investigated included turtle length, pH, dissolved organic carbon and
Podocnemis
availability of potential methylation sites (floodplain forests and hydromorphic soils). The
Amazon
study was conducted in the Rio Negro basin, where we collected water and turtle blood,
Mercury
muscle, liver and carapace samples from 12 tributaries for chemical analysis. Through radar
Methylation
imagery and existing soil maps with GIS, the percentage of alluvial floodplains and
Bioaccumulation
hydromorphic soils (potential methylation sites) was estimated for each drainage basin at
Rio Negro
sampling points. The mean Hg concentration in blood of P. erythrocephala was 1.64 ng g− 1
Ph
(SD = 1.36), muscle 33 ng g− 1 (SD = 11), liver 470 ng g− 1 (SD = 313) and carapace 68 ng g− 1
Physical factors
(SD = 32). Sex or length did not influence the Hg concentration in P. erythrocephala blood,
Size
muscle and liver, but Hg increased in carapace tissue when length size increased (ANCOVA
Turtles
p = 0.007). In the multiple regression analysis, none of the environmental factors studied had a significant relation with blood, muscle, liver and carapace. P. erythrocephala moves among habitats and in the open and interconnected aquatic systems of the Amazon basin, characterized by high levels of limnological variability, a good bioindicator of Hg concentration needs to be relatively sedentary to represent a specific habitat. However, the levels of Hg in liver were sufficient to pose a potential risk to humans that consume them, suggesting the usefulness of P. erythrocephala as a bioindicator. © 2008 Elsevier B.V. All rights reserved.
⁎ Corresponding author. Rua Hawai Quadra 33 casa 5, Cj Campos Elisios, Planalto, CEP 69045-400, Manaus, AM, Brazil. Tel.: +55 9182 3575; fax: +55 92 3643 3394. E-mail address: [email protected] (L. Schneider). 0048-9697/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2008.09.049
S CI EN C E OF TH E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
1.
Introduction
Mercury contamination is widespread in many environments (Haines et al., 1995) and is probably the most studied trace element in the world (Hylander et al., 2003). Mercury is a major pollutant in the river systems of the Amazon. Previously mercury contamination was attributed primarily to gold mining activities and resulted in research analyzing environmental factors and toxic effects of heavy metals in fish and humans (Malm et al., 1995; Roulet et al., 2000, 2001; Belger and Forsberg, 2006). However, in the 1990s, more extensive research found some of the highest levels of mercury in fish and humans in Amazon basin tributaries far removed from gold mines or any other known anthropogenic Hg sources (Forsberg et al., 1995). This mercury appears to be of natural origin as mercury levels in many Amazonian soils are naturally high (Roulet et al., 2001; Fadini and Jardim, 2001). These studies have found exceptionally high levels of mercury contamination in fish and human hair, representing a potential health risk for local fish-eating human populations (Forsberg et al., 1995; Belger and Forsberg, 2006). Fish are usually considered the main source of protein and Hg for riverine human populations. In parts of the Amazon, chelonians traditionally represent one of the most important sources of protein for humans and thus should be studied to analyze their potential health risk. In the Rio Negro basin, the capture of turtles is intense and Podocnemis erythrocephala is the most consumed and commercialized species for food, trade and special celebrations. P. erythrocephala is the smallest turtle in the Podocnemis genera and it is not preferred by the riverine people. But because populations of the larger species (P. expansa and P. unifilis) have been reduced due to the over-
1049
collecting of adults and eggs, P. erythrocephala became the only species of turtle that can be collected abundantly in the Rio Negro basin, and therefore is the most consumed (Vogt, 2001). Its preferred habitat is lentic water and it is primarily an opportunistic herbivore (Vogt, 2001). Living organisms are frequently the best indicators of environmental contaminants as they are bioaccumulated in body tissues. This is particularly true for mercury, which is eliminated very slowly from the body (Foo et al., 1993). Non migratory fish are commonly used as bioindicators in Amazon aquatic environments (Forsberg et al., 1995; Belger and Forsberg, 2006). In North America, chelonians have been shown to be good indicators of environmental contamination, and potential human risk (Golet and Haines, 2001; Burger, 2002; Storelli and Marcotrigiano, 2003). However, there have been no previous studies on mercury bioaccumulation in Amazonian turtles. The present paper adds new insight to the existing studies regarding Hg in the Amazon region. Environmental factors are potential determinants of mercury bioaccumulation, and in undisturbed environments, mercury levels in organisms usually reflect the average characteristics of the local environmental variability (SilvaForsberg et al., 1999; Belger and Forsberg, 2006). Mercury in water is positively correlated with Dissolved Organic Carbon (DOC) because Hg in the soil is adsorbed to organic particles being carried to the water (Bisinoti et al., 2007; Serudo et al., 2007). Also, organic matter reduces the ultraviolet light in water and pH, diminishing the transformation of Hg2+ into Hg0(Serudo et al., 2007), therefore increasing the Hg2+ available to be transformed into MeHg (Bisinoti et al., 2007). Organic matter can also be a substratum for methylator bacteria (Bodaly et al., 1984; Bisinoti et al., 2007). Hg in organisms correlates negatively with pH (Bisinoti et al., 2007; Serudo
Fig. 1 – Location of the 12 sample sites in the Rio Negro basin, with the Brazilian location on the right corner. Site descriptions are given in Table 1.
1050
S CI EN CE OF T H E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
et al., 2007). Acid waters promote Hg adsorption to the organic matter in soils, favoring the transformation of Hg to MeHg (Bisinoti et al., 2007). In addition, methylation occurs predominantly in anoxic environments (Bodaly et al., 1997; Bisinoti et al., 2007), such as wetlands (Roulet et al., 2001), floodplain forests, aquatic macrophyte beds (Guimarães et al., 2000) and hydromorphic soils (Branfireun et al., 1996), which are amply distributed throughout the Amazon Basin (Belger and Forsberg, 2006). The present study examines whether environmental factors influence the assimilation of Hg in a non-migratory turtle species, P. erythrocephala, the red-headed river turtle. We determined total mercury levels in four tissues of this species from the Rio Negro basin to investigate the influence of turtle size, water pH, dissolved organic carbon, and the relative abundance of floodable areas and hydromorphic soil on mercury bioaccumulation. Thus this paper focuses on the relationship of Hg levels in turtle to relevant environmental factors, and whether this species can be used as a bioindicator of local environmental conditions, and not on the correlations of Hg among tissues. P. erythrocephala, known locally as “Irapuca”, was used as a bioindicator of mercury contamination because it is widely distributed, relatively abundant and easy to collect, of reasonable size for tissue analysis, long lived and easily aged. To be a good bioindicator, a species should also accumulate pollutants above the detection limit without lethal effects, must exhibit a correlation between their pollutant content and the average pollutant concentration in the surrounding water, and this correlation must be comparable among individuals of the species, among all sites and under all conditions, and must be of interest to the public and governmental agencies (Burger, 2006a). We chose P. erythrocephala because it matches most of these criteria; it is the only species widely distributed along the Rio Negro Basin, and it can be collected in most of the tributaries. P. erythrocephala is frequently consumed by local human populations and does not migrate significantly, which makes them suitable indicators of both human interest and the influence of local environmental conditions on mercury contamination.
2.
Materials and methods
The Rio Negro basin, which comprises about 10% of the Amazon basin with an area of 750 × 103 km2 (Goulding et al., 1988), does not have any recent history of gold-mining activities along its tributaries. The Rio Negro basin has a wide range of river chemistry characteristics and potential methylation sites; DOC concentration range from 3.73 to 23.40 mg/L between tributaries, while pH varies from 3.37 to 6.89 (Forsberg et al., 1995). The basin has well defined seasons of high and low waters and these factors vary temporally in response to seasonal floods (Castillo et al., 2004). Peak flooding usually occurs in June, near the end of the six-month rainy period (Junk, 1993). Turtle and water samples were collected from January 9 to February 2, 2007 during the falling water season. Samples were collected in 12 tributaries of the Rio Negro, located between the coordinates 00°25′27″S and 02°42′45″S latitude and 60°24′ 43″W and 66°23′55″W longitude (Fig. 1). In each tributary,
physical–chemical parameters including DOC and pH were measured at only one point. According to Tardy et al. (2005), variations of DOC and pH values are similar among tributaries of the Rio Negro over the course of the hydrological cycle. Tributaries were sampled 2–3 km upstream from the Rio Negro to avoid backwater effects. At each location, we recorded geographic coordinates with a Garmin model 50 GPS and an Orion model 205A field pH meter was used to measure pH in situ. We collected surface water for DOC analyses with a 60 mL syringe. Water was filtered trough a Whatman GFF glass fiber filter (pre-combusted for 1 h at 450°C) and stored in 25 mL glass scintillation vials (pre-combusted at 400 °C for 1 h) with Teflon lined caps. As a preservative, HgCl2 was added in a final concentration of 10 mg/L. DOC concentrations were determined with a Shimadzu TOC-5000A analyzer in CENA laboratory at the University of São Paulo. We collected two to five chelonians at most sampling sites using trammel nets, for a total of 34 specimens. Turtles were sexed visually by sexual dimorphism. Males have a distinctive bright red or reddish orange pattern on the head. Females are larger than males, and males have longer, thicker tails. Individuals with no sexual dimorphism were classified as juveniles. We used carapace length as an index of size since it is the most reliable index for growth of turtles and does not vary with breeding season, as does weight (Vogt, 2001). We measured the turtles using a 1 meter ruler with 1 mm precision. A skinless boneless muscle sample of approximately 2 cm3 was taken from the left hind leg, an equal quantity of liver was taken from each turtle, and a 1 cm2 piece of the 8th right marginal scale was also collected. Blood was removed by decapitation and subsequent extraction from the jugular vein, following the report of the American Veterinary Medical Association on Euthanasia. Muscle, carapace and liver samples were kept in plastic bags and maintained frozen until analysis for Hgtot (organic + inorganic Hg) at the Universidade Federal do Pará in Santarém. Digestion was done in tubes washed in 10% HNO solution and rinsed with deionized water. A 300 to 400 mg sample was added into 10:1 HNO3 and HC 6 mol L− 1and heating it to 120 °C during 4 h (Malm et al., 1989). Hgtot concentrations were determined by cold vapor atomic fluorescence spectroscopy (CVAFS). National Research Council of Canada (NRCC) certified reference materials TORT-2 were used to assess the accuracy of the method. Blood was frozen in eppendorf tubes and sent to the Environmental and Occupational Health Sciences Institute at Rutgers University, Piscataway, New Jersey, USA. All laboratory equipment and containers were washed in 10% HNO solution and rinsed with deionized water prior to each use. A 1.0 g (wet weight) sample of blood was digested in 4 mL Ultrex ultrapure nitric acid and 2 mL deionized water in a microwave (MDX 2000 CEM), using a digestion protocol of three stages of 10 min each under (3.5, 7 and 10.6 kg/cm2) at 70% of the maximum power. Digested samples were subsequently diluted to 10 mL with deionized water. Detection limit was 0.2 ng g− 1 for mercury. All specimens were analyzed in batches with known standards (NIST), calibration standards (25 ng g− 1), and spiked specimens. Recoveries ranged from 88% to 102%. Batches with recoveries of less than 85% were reanalyzed. The coefficient of variation on replicate, spiked samples ranged up to 10%.
S CI EN C E OF TH E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
Table 1 – Environmental data of the sample sites Sample site
Location
Environmental data
#
Name
Latitude
Longitude pH
1 2 3 4 5 6 7 8 9 10 11 12
Marié Téa Urubaxi Ayuanã Enuexi Mabahá Darahá Itú Quimicuri Jufari Caurés Cuieras
00°25′27″S 00°31′32″S 00°32′58″S 00°33′28″S 00°35′24″S 00°29′46″S 00°23′36″S 00°26′02″S 00°42′25″S 01°10′07″S 01°19′35″S 02°42′45″S
66°23′55″W 65°12′25″W 64°48′30″W 64°55′35″W 65°07′51″W 64°38′20″W 64°46′54″W 63°20′20″W 63°13′57″W 61°59′33″W 62°17′51″W 60°24′43″W
4.5 4.3 4.7 4.5 4.3 4.4 4.6 4.8 6.0 4.9 5.2 4.9
DOC FA (mg/L) (%) 12.6 22.4 10.3 14.0 19.9 7.0 15.7 10.1 6.9 4.2 9.3 9.0
10.6 11.6 16.8 12.8 12.0 25.6 7.0 53.7 56.4 50.2 12.5 4.7
HS (%) 64.6 73.8 67.3 47.2 50.0 26.7 64.1 59.6 49.5 43.4 47.1 10.4
DOC = Dissolved organic carbon; FA = flooded area; HS = hydromorphic soil.
We considered the seasonally flooded alluvial forests as potential Hg methylation sites in the Rio Negro Basin, known locally as “igapó”, and hydromorphic soils. The influence of these habitats was assumed to be proportional to the percent of the drainage basin, which turtles occupied upstream from the sampling site. We determined the geographical limits of the drainage basins upstream from each sampling point through the visual interpretation of Hydrosheds maps (Hydrological data and maps based on Shuttle Elevation Derivatives at Multiple Scales) from the World Wildlife Fund, derivative from the SRTM data (Shuttle Radar Topography Mission). All watercourses above the collecting point were considered part of the drainage basin of rivers and streams. In the case of lakes, all streams that terminated in the lake, as well as the upstream portion of the river to which it is connected, were considered part of the basin. Flooded forests within these drainage basins were identified using a dual-season mapping of wetland inundation and vegetation for the central Amazon basin acquired by mosaicked L-band synthetic aperture radar (SAR) (Hess et al., 2003). Flooded area was calculated using the ArcGIS 9.2. The percentage of hydromorphic soils in each drainage basin was obtained through digital integration soil maps obtained from the Instituto Brasileiro de Geografia e Estatística (IBGE). The soils were reclassified as hydromorphic or not hydromorphic, and calculated using the polygon select and histogram functions of ArcGIS 9.2. The normality of all variables was tested by Shapiro–Wilk. Correlation among independent variables (pH, DOC, percent of flooded area and percent of hydromorphic soils) was tested with a Pearson correlation and Bonferroni probability matrixes. Correlations between these variables and turtle size were also examined. The concentration of Hg in blood, muscle, liver and carapace of all individuals was compared using a correlation matrix. Further, ANCOVA was used to determine if sex (as a categorical variable) or size (continuous variable) influenced mercury bioaccumulation in blood, muscle, liver and carapace separately .The influence of environmental factors on Hg bioaccumulation in each tissue was tested with multiple
1051
regression analysis. If size or sex had an effect, it was considered as an independent variable together with pH, DOC, percentage of floodable area and percentage of hydromorphic soils against mercury concentration. To avoid spatial pseudo-replication, results for only one randomly chosen turtle from each site were used in this analysis.
3.
Results
Water pH varied from 4.3 to 6.0 (n = 12, µ = 4.7, SD = 0.5) and DOC varied from 4.2 to 22.3 mg/L (n = 12, µ = 11.8 and SD = 5.4). The percent of flooded forest in upstream drainage areas varied from 4.7 to 56.4% (n = 12, µ = 22.8%, SD = 5.4) while the percentage of hydromorphic soils varied from 10.4 to 73.8% (n = 12, µ = 50.3%, SD = 17.9). Environmental data are presented in Table 1. The pH was negatively correlated with DOC (p = 0.04, r = −0.59, Fig. 2a) and positively correlated with flooded areas (p = 0.04, r = 0.59, Fig. 2b). There was no correlation among the remaining environmental factors. In the multiple regression analysis, none of the environmental factors had significant effects on Hgtot concentration in turtle blood (p = 0.243), muscle (p = 0.596), liver (p = 0.877) and carapace (p = 0.284).
Fig. 2 – pH measurements taken from different tributaries of the Rio Negro in relation to (a) DOC and (b) percent of flooded area.
1052
S CI EN CE OF T H E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
There was no significant correlation among Hg concentration in tissues. Liver showed the highest Hg level, followed by carapace, muscle and blood. The mean Hg concentration in P. erythrocephala blood was 1.64 ng g− 1 (n = 34, SD = 1.36), ranging from 0.1 to 6.0. Muscle was 33 ng g− 1 (n = 34, SD = 11), ranging from 10 to 106 ng g− 1. The mean Hg concentration in liver was 470 ng g− 1 (n = 19, SD = 313), ranging from 79 to 1358 ng g− 1. The carapace Hg mean concentration was 68 ng g− 1 (n = 21, SD = 32), ranging from 9 to 155 ng g− 1. Individual's length varied from 6.6 cm to 30.0 cm (n = 34, µ = 21.6, SD = 4.9) and weighted from 90 g to 2140 g (n = 34, µ = 1033, SD = 523). The ANCOVA results showed that P. erythrocephala Hg bioacomulation was not influenced by sex or size for blood (p = 0.493; p = 0.224, respectively), liver (p = 0.195; p = 0.388, respectively) and muscle (p = 0.115; p = 0.195, respectively), but it was influenced by the carapace length (p = 0.1255; p = 0.007, respectively) (Fig. 3).
4.
Discussion
2001). The low variation in mercury levels was insufficient to track differences in the physical characteristics of the local environment. Similarly, Hg in herbivorous fish in the Rio Negro basin also had no systematic correlation with sitedependent geochemistry (Barbosa et al., 2003). We had expected a positive correlation between pH and flooded areas because the anoxic condition that occur in flooded areas favor the accumulation and release of organic materials into the system, including fulvic and humic acids (Junk, 1997). Also, a correlation observed between DOC and pH indicates a mechanistic relation between these parameters. DOC in the Rio Negro system consists predominantly of humic and fulvic acids (Hedges et al., 1986) and since most tributaries in the system are extremely poor in carbonate buffering capacity (Belger and Forsberg, 2006), pH is controlled predominantly by the concentration of these organic acids. The pH and DOC values were similar to the ones found by Belger and Forsberg (2006), varying from an acid pH of 4.0 to 6.0 and a low DOC of 7 to 25 mg/L.
4.1.
Hg levels and environmental correlates
4.2.
The results of this study showed that environmental factors did not explain the variations in the concentration of mercury in the blood, muscle, liver and carapace of P. erythrocephala in the Rio Negro Basin. This is likely a reflection of the fact that P. erythrocephalata, although it is not a migratory species, moves extensively from one area to another during the year. Belger and Forsberg (2006) studied two species of fish (Cichla spp. and Hoplias malabaricus) and found that mercury levels in the relatively sedentary H. malabaricus were more correlated with environmental factors than in the more erratic Cichla spp. Thus, even though P. erythrocephala is a non-migratory species, the fact that it moves extensively in different tributaries suggests that the species is not a good bioindicator of local environmental conditions. In an open and interconnected aquatic systems, such as the Amazon basin that is characterized by high levels of limnological variability, a more locallydistributed species might be a better indicator of environmental characteristics. In addition, the mercury levels in P. erythrocephala were relatively low due to its exclusively herbivorous diet (Vogt,
Fig. 3 – Relation between Hgtot in carapace and turtle size, using the carapace length as an index of size.
Hg levels and biological correlates
Although Hg levels in P. erythrocephala did not reflect local environmental conditions, they could reflect internal Hg levels. However, the lack of correlation between mercury in muscle, liver and carapace indicates that the non lethal collection of a small piece of carapace is not effective as a bioindicator because the carapace does not predict Hg levels in the muscle or liver. For snapping turtles (Chelydra serpentina), the scute was a suitable material for mercury monitoring because there was a high correlation between scute concentrations and muscle, which also occurs with hair in mammals and feathers of birds (Golet and Haines, 2001; Burger and Gochfeld, 2001). However, this did not occur for the redheaded river turtle in the Rio Negro Basin. The Hg concentration in the herbivorous P. erythrocephala muscle in this study (10 to 106 ng g− 1) was lower than previous studies with the omnivore Chelydra serpentina: 50 to 500 ng g− 1 in 5 lakes in Connecticut (Golet and Haines, 2000), 50 to 300 ng g− 1 in Minnesota (Helwig and Hora, 1983) and 100 to 170 ng g− 1 in Tennessee (Meyers-Schöne et al., 1993). P. erythrocephala have similar Hg levels, around 100 ng g− 1, as herbivorous fish from the Rio Negro basin (Barbosa et al., 2003) and from the Rio Tapajós Basin (Malm et al., 1995). The significant accumulation of Hg in Amazonian turtles and fish reflects environmental Hg bioavailability (Roulet et al., 2000). Mercury bioaccumulation in aquatic systems varies considerably with food-chain structure and length. Mercury levels were relatively low in P. erythrocephala in the present study, perhaps because it is exclusively herbivorous and the aquatic plants and fruits it eats are typically low in mercury (Vogt, 2001). A lack of correlation between body size and mercury levels was also reported by Helwig and Hora (1983) and Golet and Haines (2000) who studied the omnivorous Chelydra serpentina. They suggest that chelonians may be eliminating mercury from the body tissues at higher rates than fish, and can reach equilibrium with ingested mercury. This hypothesis is supported by the lack of correlation found in this study between size and blood, liver and muscle of the Red-headed river turtle; there was, however, a
S CI EN C E OF TH E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
strong correlation between carapace and size. Blood, muscle and liver are likely to be more variable with daily intake and elimination of mercury, whereas carapace integrates mercury intake over longer periods of time. The shell of chelonians is composed of a deep, bony dermal-endoskeleton and an external corneous epidermis subdivided into rows of scutes (Zangerl, 1969). These scutes are the hard keratinized plates that armor the carapace and they originate from growing regions at the basal infoldings of the epidermis. The scutes form a series of nonliving keratin layers deposited one on the top of the other (Solomon et al., 1986). Hg is deposited in a metabolically inactive keratin matrix or is bound in relatively immobile inorganic Hg complexes. Consequently, since Hg is very stable in non-living proteinaceous tissue (Burger and Gochfeld, 2001; Day et al., 2007), a correlation between Hg and turtle size can be obtained by analyzing the scutes (as was done for P. erythrocephala in the present study).
4.3.
Hg levels and human health
Because they are long-lived, widely-distributed, and seem to have similar amounts of Hg regardless of size, they may be a useful species for biological monitoring of environmental mercury contamination of specific interest for humans consuming them. The levels of total mercury we found in P. erythrocephala in muscle are under the maximum level of 500 ng g− 1 recommended by the World Health Organization and do not constitute a serious health risk for fish-eating human populations in the Rio Negro Basin. However, we found mercury levels in the liver higher than the maximum level recommended by the World Health Organization, and special attention should be given to this part of the body, especially since it is eaten by local people. Recommendations should be given to the local people to avoid eating the liver of turtles. Also, we recommend that carnivorous species of turtles in the Amazon areas with potentially high Hg levels, such as Chelus fimbriatus, be studied. This is a species traditionally consumed from the Rio Negro, and can be an important source of protein for the human population. Carnivorous species in the Amazon basin that have Hg levels higher than 500 ng g− 1, the maximum level recommended by the World Health Organization, should be given special attention; the muscle of C. fimbriatus might represent a health risk to the riverine people. Such a higher trophic level species could be a better bioindicator of exposure and effects (Burger 2006b).
5.
Conclusions
Environmental factors did not explain the variation in the concentration of mercury in the muscle of P. erythrocephala in the Rio Negro basin, which may occur because this species moves among local habitats and tributaries. There was no correlation between body size and mercury levels in P. erythrocephala, perhaps because it is exclusively herbivorous (and thus has low mercury levels overall), and may also be able to eliminate mercury from the muscle and liver. However, this Amazonian turtle accumulated significant amounts of Hg,
1053
comparable to those of herbivorous fish in the Rio Negro and other Amazonian Rivers, reflecting Hg bioavailability. Because they are long-lived, widely-distributed and seem to have similar amounts of Hg regardless of size, they may be a useful species for biological monitoring of environmental mercury contamination that can affect consumers of these turtles, including humans.
Acknowledgments The study was conducted under the auspices of IBAMA permit 147/2006 and financed by FINEP. Larissa Schneider was financed by CNPq. Blood analysis was funded by P30ES005022 and DEFC01-06EW07053. We thank Dr. Reinaldo Peleja and crew for helping in the lab work, Dr. Bruce Forsberg, Chris Jeitner and Dr. Michael Gochfeld for the comments and lab facilities, and Camila Ferrara and Ladislau Santos-Junior for helping to collect turtle samples.
REFERENCES Barbosa AC, Souza J, Dórea JG, Jardim WF, Fadini PS. Mercury biomagnification in a tropical black water, Rio Negro, Brazil. Arch Environ Contam Toxicol 2003;45:235–46. Belger L, Forsberg BR. Factors controlling Hg levels in two predatory fish species in the Negro River Basin, Brazilian Amazon. Sci Total Environ 2006;367:451–9. Bisinoti MC, Sargentini-Junior E, Jardim WF. Seasonal behavior of mercury species in waters and sediments from the Negro River Basin, Amazon, Brazil. J Braz Chem Soc 2007; 18(3):544–53. Bodaly RA, Hecky RE, Fudge RJ. Increases in fish Hg levels in lakes flooded by the Churchill river diversion, Northern Manitoba. Can J Fish Aquat Sci 1984;41:682–91. Bodaly RA, St. Louis VL, Paterson MJ, Fudge RJP, Hall BD, Rosenberg DM, et al. Metal ions in biological systems. In: Siegel A, Siegel H, editors. Hg and its effects on environmental and biology. New York: Marcel Dekker; 1997. p. 259–87. Branfireun BA, Heus A, Roulet NT. The hydrology and methylHg dynamics of Precambrian shield headwater peatland. Water Resour Res 1996;32:1785–94. Burger J. Metals in diamondback terrapins from New Jersey. Environ Monit Assess 2002;77:255–63. Burger J. Bioindicators: types, development, and use in ecological assessment and research. Environ Bioindicat 2006a;1:22–39. Burger J. Neurotoxicology and behavioral effects in reptiles. In: Gardner SC, Oberdorster E, editors. Toxicology of reptiles. Boca Raton: CRC Press; 2006b. p. 173–98. Burger J, Gochfeld M. Effects of chemicals and pollution on seabirds. In: Schreiber EA, Burger J, editors. Biology of marine birds. Boca Raton: CRC Press; 2001. p. 525. Castillo MM, Allan JD, Sinsabaugh RL, Kling GW. Seasonal and interannual variation of bacterial production in lowland rivers of the Orinoco basin. Freshw Biol 2004;49:1400–14. Day RD, Segars AL, Arendt MD, Peden-Adams MM. Relationship of blood mercury levels to health parameters in the loggerhead sea turtle (Caretta caretta). Environ Health Perspec 2007; 115(10):1421–7. Fadini PS, Jardim WF. Is the Negro River Basin (Amazon) impacted by naturally occurring Hg? Sci Total Environ 2001;275(1):71–82. Foo SC, Ngim CH, Salleh I, Jeyaratnan J, Boey KW. Neurobehavioral effects in occupational chemical exposure. Environ Res 1993;60:267–73.
1054
S CI EN CE OF T H E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
Forsberg BR, Silva-Forsberg MC, Padovani CR, Sargentini E, Malm O. High level of Hg in fish from the Rio Negro Basin (Brazilian, Amazon): natural background or anthropogenic contamination? In: Kato H, Pfeifer WC, editors. International workshop on environmental hg pollution and its health effects in the Amazon River basin, National Institute for Minamata Disease. Rio de Janeiro: Universidade Federal do Rio de Janeiro; 1995. p. 33–40. Golet WJ, Haines TA. Snapping turtles (Chelydra serpentina) as monitors for mercury contamination of aquatic environments. Environ Monit Assess 2001;71:211–20. Goulding M, Carvalho ML, Ferreira EG. Rio Negro: rich life in poor water: Amazonian diversity and foodplain ecology as seen through fish communities. The Hague: SPB Academic Publishing; 1988. p. 200. Guimarães JRD, Meili M, Hylander LD, Silva EDE, Roulet M, Mauro JBN, et al. Mercury net methylation in five tropical flood plain regions of Brazil: high in the root zone of floating macrophyte mats but low in surface sediments and flooded soils. Sci Total Environ 2000;261:99–107. Haines TA, Komov V, Matey V, Jagoe C. Perch mercury content is related to acidity and color of 26 Russian lakes. Water Air Soil Pollut 1995;85:823–8. Hedges JI, Ertel JR, Quay PD, Grootes PM, Richey JE, Devol AH, et al. Organic carbon-14 in the Amazon River System. Science 1986;231:1129–31. Helwig D, Hora M. Polychlorinated biphenyl, mercury and cadmium concentrations in Minnesota snapping turtles. Bull Environ Contam Toxicol 1983;30:186–91. Hess L, Melack J, Novo EMLM, Barbosa CCF, Gatil M. Dual-season mapping of wetland inundation and vegetation for the central Amazon basin. Remote Sens Environ 2003;87:404–28. Hylander LD, Sollenberg H, Westas H. A three-stage system to remove mercury and dioxins in flue gases. Sci Total Environ 2003;304(1–3):137–44. Junk WJ. Wetlands of tropical South America. Wetlands of the World 1993;11:679–739. Junk WJ. The central Amazon floodplain. Ecology of a pulsing system. Berlim: Springer; 1997. p. 100. Malm O, Branches FJP, Akagi H, Castro MB, Pfeiffer WC, Arada M, et al. Mercury and methylmercury in fish and human hair from Tapajós river basin, Brazil. Sci Total Environ 1995;175 (2):141–50.
Malm O, Pfeifer WC, Bastos WR, Souza CMM. Utilização de acessório de geração de vapor frio para análise de mercúrio em investigações ambientais por espectometria de absorção atômica. Ciência e Cultura 1989;41:88–92. Meyers-Schöne L, Shugart L, Beauchamp J, Walton B. Comparison of two freshwater turtle species as monitors of radio nuclide and chemical contamination: DNA damage and residue analyses. Environ Toxicol Chem 1993;12:1487–96. Roulet M, Guimarães JRD, Lucotte M. Methylmercury production and accumulation in sediments and soils of an Amazonian floodplain: effect of seasonal inundation. Water Air Soil Pollut 2001;128:41–61. Roulet M, Lucotte M, Rheault I, Guimarães JRD. Methylmercury in the water, seston and epiphyton of an Amazonian River and its floodplain, Tapajós River, Brazil. Sci Total Environ 2000;261:43–59. Serudo RL, Oliveira LC, Rocha JC, Paterlini WC, Rosa AH, Silva HC, et al. Reduction capability of soil humic substances from the Rio Negro basin, Brazil, towards Hg(II)studied by a multimethod approach and principal component analysis (PCA). Geoderma 2007;138:229–36. Silva-Forsberg MC, Forsberg BR, Zeidemann ZK. Mercury contamination in humans linked to river chemistry in the Amazon Basin. Ambio 1999;28:519–21. Solomon SE, Hendrickson JR, Hendrickson LPJ. The structure of the carapace and plastron of juveniles turtles, Chelonia mydas (the green turtle) and Caretta caretta (the loggerhead turtle). J Anat 1986;145:123–31. Storelli MM, Marcotrigiano GO. Heavy metal residues in tissues of marine turtles. Mar Pollut Bull 2003;46:397–400. Tardy Y, Bustillo V, Roquin C, Mortatti J, Victoria R. The Amazon. Bio-geochemistry applied to river basin management Part I. Hydro-climatology, hydrograph separation, mass transfer balances, stable isotopes, and modeling. Appl Geochem 2005;20:1746–829. Vogt RC. Turtles of the Rio Negro. In: Chao NL, Petry P, Prang G, Sonneschien L, Tlusty M, editors. Conservation and management of ornamental fish resources of the Rio Negro Basin, Amazonia, Brazil. Manaus: Editora Universidade do amazonas; 2001. p. 245–62. Zangerl R. The turtle shell. In: Gans C, editor. Volume 1: morphology. London: Academic Press; 1969. p. 311–39.
a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m
w w w. e l s e v i e r. c o m / l o c a t e / s c i t o t e n v
Mercury bioacumulation in four tissues of Podocnemis erythrocephala (Podocnemididae: Testudines) as a function of water parameters Larissa Schneider a,⁎, Lauren Belger b , Joanna Burger c , Richard C. Vogt a a
Department of Freshwater Biology, Instituto Nacional de Pesquisas da Amazônia (INPA), Brazil Department of Ecology, Instituto Nacional de Pesquisas da Amazônia (INPA), Brazil c Department of Ecology, Evolution, and Natural Resources, and Environmental and Occupational Health Sciences Institute, Rutgers University, USA b
AR TIC LE D ATA
ABSTR ACT
Article history:
A number of environmental factors influence the dynamics of Hg in aquatic ecosystems, yet
Received 21 June 2008
few studies have examined these factors for turtles, especially from South America. Red-
Received in revised form
headed river turtle (Podocnemis erythrocephala) is easy to capture in the black waters of Rio
24 September 2008
Negro, making it the turtle species that is consumed most often by people of the region. In
Accepted 25 September 2008
this study, environmental factors and turtle size were investigated to determine their influence on the Hg concentration in blood, muscle, liver and carapace of the red-headed
Keywords:
river turtle. Factors investigated included turtle length, pH, dissolved organic carbon and
Podocnemis
availability of potential methylation sites (floodplain forests and hydromorphic soils). The
Amazon
study was conducted in the Rio Negro basin, where we collected water and turtle blood,
Mercury
muscle, liver and carapace samples from 12 tributaries for chemical analysis. Through radar
Methylation
imagery and existing soil maps with GIS, the percentage of alluvial floodplains and
Bioaccumulation
hydromorphic soils (potential methylation sites) was estimated for each drainage basin at
Rio Negro
sampling points. The mean Hg concentration in blood of P. erythrocephala was 1.64 ng g− 1
Ph
(SD = 1.36), muscle 33 ng g− 1 (SD = 11), liver 470 ng g− 1 (SD = 313) and carapace 68 ng g− 1
Physical factors
(SD = 32). Sex or length did not influence the Hg concentration in P. erythrocephala blood,
Size
muscle and liver, but Hg increased in carapace tissue when length size increased (ANCOVA
Turtles
p = 0.007). In the multiple regression analysis, none of the environmental factors studied had a significant relation with blood, muscle, liver and carapace. P. erythrocephala moves among habitats and in the open and interconnected aquatic systems of the Amazon basin, characterized by high levels of limnological variability, a good bioindicator of Hg concentration needs to be relatively sedentary to represent a specific habitat. However, the levels of Hg in liver were sufficient to pose a potential risk to humans that consume them, suggesting the usefulness of P. erythrocephala as a bioindicator. © 2008 Elsevier B.V. All rights reserved.
⁎ Corresponding author. Rua Hawai Quadra 33 casa 5, Cj Campos Elisios, Planalto, CEP 69045-400, Manaus, AM, Brazil. Tel.: +55 9182 3575; fax: +55 92 3643 3394. E-mail address: [email protected] (L. Schneider). 0048-9697/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2008.09.049
S CI EN C E OF TH E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
1.
Introduction
Mercury contamination is widespread in many environments (Haines et al., 1995) and is probably the most studied trace element in the world (Hylander et al., 2003). Mercury is a major pollutant in the river systems of the Amazon. Previously mercury contamination was attributed primarily to gold mining activities and resulted in research analyzing environmental factors and toxic effects of heavy metals in fish and humans (Malm et al., 1995; Roulet et al., 2000, 2001; Belger and Forsberg, 2006). However, in the 1990s, more extensive research found some of the highest levels of mercury in fish and humans in Amazon basin tributaries far removed from gold mines or any other known anthropogenic Hg sources (Forsberg et al., 1995). This mercury appears to be of natural origin as mercury levels in many Amazonian soils are naturally high (Roulet et al., 2001; Fadini and Jardim, 2001). These studies have found exceptionally high levels of mercury contamination in fish and human hair, representing a potential health risk for local fish-eating human populations (Forsberg et al., 1995; Belger and Forsberg, 2006). Fish are usually considered the main source of protein and Hg for riverine human populations. In parts of the Amazon, chelonians traditionally represent one of the most important sources of protein for humans and thus should be studied to analyze their potential health risk. In the Rio Negro basin, the capture of turtles is intense and Podocnemis erythrocephala is the most consumed and commercialized species for food, trade and special celebrations. P. erythrocephala is the smallest turtle in the Podocnemis genera and it is not preferred by the riverine people. But because populations of the larger species (P. expansa and P. unifilis) have been reduced due to the over-
1049
collecting of adults and eggs, P. erythrocephala became the only species of turtle that can be collected abundantly in the Rio Negro basin, and therefore is the most consumed (Vogt, 2001). Its preferred habitat is lentic water and it is primarily an opportunistic herbivore (Vogt, 2001). Living organisms are frequently the best indicators of environmental contaminants as they are bioaccumulated in body tissues. This is particularly true for mercury, which is eliminated very slowly from the body (Foo et al., 1993). Non migratory fish are commonly used as bioindicators in Amazon aquatic environments (Forsberg et al., 1995; Belger and Forsberg, 2006). In North America, chelonians have been shown to be good indicators of environmental contamination, and potential human risk (Golet and Haines, 2001; Burger, 2002; Storelli and Marcotrigiano, 2003). However, there have been no previous studies on mercury bioaccumulation in Amazonian turtles. The present paper adds new insight to the existing studies regarding Hg in the Amazon region. Environmental factors are potential determinants of mercury bioaccumulation, and in undisturbed environments, mercury levels in organisms usually reflect the average characteristics of the local environmental variability (SilvaForsberg et al., 1999; Belger and Forsberg, 2006). Mercury in water is positively correlated with Dissolved Organic Carbon (DOC) because Hg in the soil is adsorbed to organic particles being carried to the water (Bisinoti et al., 2007; Serudo et al., 2007). Also, organic matter reduces the ultraviolet light in water and pH, diminishing the transformation of Hg2+ into Hg0(Serudo et al., 2007), therefore increasing the Hg2+ available to be transformed into MeHg (Bisinoti et al., 2007). Organic matter can also be a substratum for methylator bacteria (Bodaly et al., 1984; Bisinoti et al., 2007). Hg in organisms correlates negatively with pH (Bisinoti et al., 2007; Serudo
Fig. 1 – Location of the 12 sample sites in the Rio Negro basin, with the Brazilian location on the right corner. Site descriptions are given in Table 1.
1050
S CI EN CE OF T H E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
et al., 2007). Acid waters promote Hg adsorption to the organic matter in soils, favoring the transformation of Hg to MeHg (Bisinoti et al., 2007). In addition, methylation occurs predominantly in anoxic environments (Bodaly et al., 1997; Bisinoti et al., 2007), such as wetlands (Roulet et al., 2001), floodplain forests, aquatic macrophyte beds (Guimarães et al., 2000) and hydromorphic soils (Branfireun et al., 1996), which are amply distributed throughout the Amazon Basin (Belger and Forsberg, 2006). The present study examines whether environmental factors influence the assimilation of Hg in a non-migratory turtle species, P. erythrocephala, the red-headed river turtle. We determined total mercury levels in four tissues of this species from the Rio Negro basin to investigate the influence of turtle size, water pH, dissolved organic carbon, and the relative abundance of floodable areas and hydromorphic soil on mercury bioaccumulation. Thus this paper focuses on the relationship of Hg levels in turtle to relevant environmental factors, and whether this species can be used as a bioindicator of local environmental conditions, and not on the correlations of Hg among tissues. P. erythrocephala, known locally as “Irapuca”, was used as a bioindicator of mercury contamination because it is widely distributed, relatively abundant and easy to collect, of reasonable size for tissue analysis, long lived and easily aged. To be a good bioindicator, a species should also accumulate pollutants above the detection limit without lethal effects, must exhibit a correlation between their pollutant content and the average pollutant concentration in the surrounding water, and this correlation must be comparable among individuals of the species, among all sites and under all conditions, and must be of interest to the public and governmental agencies (Burger, 2006a). We chose P. erythrocephala because it matches most of these criteria; it is the only species widely distributed along the Rio Negro Basin, and it can be collected in most of the tributaries. P. erythrocephala is frequently consumed by local human populations and does not migrate significantly, which makes them suitable indicators of both human interest and the influence of local environmental conditions on mercury contamination.
2.
Materials and methods
The Rio Negro basin, which comprises about 10% of the Amazon basin with an area of 750 × 103 km2 (Goulding et al., 1988), does not have any recent history of gold-mining activities along its tributaries. The Rio Negro basin has a wide range of river chemistry characteristics and potential methylation sites; DOC concentration range from 3.73 to 23.40 mg/L between tributaries, while pH varies from 3.37 to 6.89 (Forsberg et al., 1995). The basin has well defined seasons of high and low waters and these factors vary temporally in response to seasonal floods (Castillo et al., 2004). Peak flooding usually occurs in June, near the end of the six-month rainy period (Junk, 1993). Turtle and water samples were collected from January 9 to February 2, 2007 during the falling water season. Samples were collected in 12 tributaries of the Rio Negro, located between the coordinates 00°25′27″S and 02°42′45″S latitude and 60°24′ 43″W and 66°23′55″W longitude (Fig. 1). In each tributary,
physical–chemical parameters including DOC and pH were measured at only one point. According to Tardy et al. (2005), variations of DOC and pH values are similar among tributaries of the Rio Negro over the course of the hydrological cycle. Tributaries were sampled 2–3 km upstream from the Rio Negro to avoid backwater effects. At each location, we recorded geographic coordinates with a Garmin model 50 GPS and an Orion model 205A field pH meter was used to measure pH in situ. We collected surface water for DOC analyses with a 60 mL syringe. Water was filtered trough a Whatman GFF glass fiber filter (pre-combusted for 1 h at 450°C) and stored in 25 mL glass scintillation vials (pre-combusted at 400 °C for 1 h) with Teflon lined caps. As a preservative, HgCl2 was added in a final concentration of 10 mg/L. DOC concentrations were determined with a Shimadzu TOC-5000A analyzer in CENA laboratory at the University of São Paulo. We collected two to five chelonians at most sampling sites using trammel nets, for a total of 34 specimens. Turtles were sexed visually by sexual dimorphism. Males have a distinctive bright red or reddish orange pattern on the head. Females are larger than males, and males have longer, thicker tails. Individuals with no sexual dimorphism were classified as juveniles. We used carapace length as an index of size since it is the most reliable index for growth of turtles and does not vary with breeding season, as does weight (Vogt, 2001). We measured the turtles using a 1 meter ruler with 1 mm precision. A skinless boneless muscle sample of approximately 2 cm3 was taken from the left hind leg, an equal quantity of liver was taken from each turtle, and a 1 cm2 piece of the 8th right marginal scale was also collected. Blood was removed by decapitation and subsequent extraction from the jugular vein, following the report of the American Veterinary Medical Association on Euthanasia. Muscle, carapace and liver samples were kept in plastic bags and maintained frozen until analysis for Hgtot (organic + inorganic Hg) at the Universidade Federal do Pará in Santarém. Digestion was done in tubes washed in 10% HNO solution and rinsed with deionized water. A 300 to 400 mg sample was added into 10:1 HNO3 and HC 6 mol L− 1and heating it to 120 °C during 4 h (Malm et al., 1989). Hgtot concentrations were determined by cold vapor atomic fluorescence spectroscopy (CVAFS). National Research Council of Canada (NRCC) certified reference materials TORT-2 were used to assess the accuracy of the method. Blood was frozen in eppendorf tubes and sent to the Environmental and Occupational Health Sciences Institute at Rutgers University, Piscataway, New Jersey, USA. All laboratory equipment and containers were washed in 10% HNO solution and rinsed with deionized water prior to each use. A 1.0 g (wet weight) sample of blood was digested in 4 mL Ultrex ultrapure nitric acid and 2 mL deionized water in a microwave (MDX 2000 CEM), using a digestion protocol of three stages of 10 min each under (3.5, 7 and 10.6 kg/cm2) at 70% of the maximum power. Digested samples were subsequently diluted to 10 mL with deionized water. Detection limit was 0.2 ng g− 1 for mercury. All specimens were analyzed in batches with known standards (NIST), calibration standards (25 ng g− 1), and spiked specimens. Recoveries ranged from 88% to 102%. Batches with recoveries of less than 85% were reanalyzed. The coefficient of variation on replicate, spiked samples ranged up to 10%.
S CI EN C E OF TH E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
Table 1 – Environmental data of the sample sites Sample site
Location
Environmental data
#
Name
Latitude
Longitude pH
1 2 3 4 5 6 7 8 9 10 11 12
Marié Téa Urubaxi Ayuanã Enuexi Mabahá Darahá Itú Quimicuri Jufari Caurés Cuieras
00°25′27″S 00°31′32″S 00°32′58″S 00°33′28″S 00°35′24″S 00°29′46″S 00°23′36″S 00°26′02″S 00°42′25″S 01°10′07″S 01°19′35″S 02°42′45″S
66°23′55″W 65°12′25″W 64°48′30″W 64°55′35″W 65°07′51″W 64°38′20″W 64°46′54″W 63°20′20″W 63°13′57″W 61°59′33″W 62°17′51″W 60°24′43″W
4.5 4.3 4.7 4.5 4.3 4.4 4.6 4.8 6.0 4.9 5.2 4.9
DOC FA (mg/L) (%) 12.6 22.4 10.3 14.0 19.9 7.0 15.7 10.1 6.9 4.2 9.3 9.0
10.6 11.6 16.8 12.8 12.0 25.6 7.0 53.7 56.4 50.2 12.5 4.7
HS (%) 64.6 73.8 67.3 47.2 50.0 26.7 64.1 59.6 49.5 43.4 47.1 10.4
DOC = Dissolved organic carbon; FA = flooded area; HS = hydromorphic soil.
We considered the seasonally flooded alluvial forests as potential Hg methylation sites in the Rio Negro Basin, known locally as “igapó”, and hydromorphic soils. The influence of these habitats was assumed to be proportional to the percent of the drainage basin, which turtles occupied upstream from the sampling site. We determined the geographical limits of the drainage basins upstream from each sampling point through the visual interpretation of Hydrosheds maps (Hydrological data and maps based on Shuttle Elevation Derivatives at Multiple Scales) from the World Wildlife Fund, derivative from the SRTM data (Shuttle Radar Topography Mission). All watercourses above the collecting point were considered part of the drainage basin of rivers and streams. In the case of lakes, all streams that terminated in the lake, as well as the upstream portion of the river to which it is connected, were considered part of the basin. Flooded forests within these drainage basins were identified using a dual-season mapping of wetland inundation and vegetation for the central Amazon basin acquired by mosaicked L-band synthetic aperture radar (SAR) (Hess et al., 2003). Flooded area was calculated using the ArcGIS 9.2. The percentage of hydromorphic soils in each drainage basin was obtained through digital integration soil maps obtained from the Instituto Brasileiro de Geografia e Estatística (IBGE). The soils were reclassified as hydromorphic or not hydromorphic, and calculated using the polygon select and histogram functions of ArcGIS 9.2. The normality of all variables was tested by Shapiro–Wilk. Correlation among independent variables (pH, DOC, percent of flooded area and percent of hydromorphic soils) was tested with a Pearson correlation and Bonferroni probability matrixes. Correlations between these variables and turtle size were also examined. The concentration of Hg in blood, muscle, liver and carapace of all individuals was compared using a correlation matrix. Further, ANCOVA was used to determine if sex (as a categorical variable) or size (continuous variable) influenced mercury bioaccumulation in blood, muscle, liver and carapace separately .The influence of environmental factors on Hg bioaccumulation in each tissue was tested with multiple
1051
regression analysis. If size or sex had an effect, it was considered as an independent variable together with pH, DOC, percentage of floodable area and percentage of hydromorphic soils against mercury concentration. To avoid spatial pseudo-replication, results for only one randomly chosen turtle from each site were used in this analysis.
3.
Results
Water pH varied from 4.3 to 6.0 (n = 12, µ = 4.7, SD = 0.5) and DOC varied from 4.2 to 22.3 mg/L (n = 12, µ = 11.8 and SD = 5.4). The percent of flooded forest in upstream drainage areas varied from 4.7 to 56.4% (n = 12, µ = 22.8%, SD = 5.4) while the percentage of hydromorphic soils varied from 10.4 to 73.8% (n = 12, µ = 50.3%, SD = 17.9). Environmental data are presented in Table 1. The pH was negatively correlated with DOC (p = 0.04, r = −0.59, Fig. 2a) and positively correlated with flooded areas (p = 0.04, r = 0.59, Fig. 2b). There was no correlation among the remaining environmental factors. In the multiple regression analysis, none of the environmental factors had significant effects on Hgtot concentration in turtle blood (p = 0.243), muscle (p = 0.596), liver (p = 0.877) and carapace (p = 0.284).
Fig. 2 – pH measurements taken from different tributaries of the Rio Negro in relation to (a) DOC and (b) percent of flooded area.
1052
S CI EN CE OF T H E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
There was no significant correlation among Hg concentration in tissues. Liver showed the highest Hg level, followed by carapace, muscle and blood. The mean Hg concentration in P. erythrocephala blood was 1.64 ng g− 1 (n = 34, SD = 1.36), ranging from 0.1 to 6.0. Muscle was 33 ng g− 1 (n = 34, SD = 11), ranging from 10 to 106 ng g− 1. The mean Hg concentration in liver was 470 ng g− 1 (n = 19, SD = 313), ranging from 79 to 1358 ng g− 1. The carapace Hg mean concentration was 68 ng g− 1 (n = 21, SD = 32), ranging from 9 to 155 ng g− 1. Individual's length varied from 6.6 cm to 30.0 cm (n = 34, µ = 21.6, SD = 4.9) and weighted from 90 g to 2140 g (n = 34, µ = 1033, SD = 523). The ANCOVA results showed that P. erythrocephala Hg bioacomulation was not influenced by sex or size for blood (p = 0.493; p = 0.224, respectively), liver (p = 0.195; p = 0.388, respectively) and muscle (p = 0.115; p = 0.195, respectively), but it was influenced by the carapace length (p = 0.1255; p = 0.007, respectively) (Fig. 3).
4.
Discussion
2001). The low variation in mercury levels was insufficient to track differences in the physical characteristics of the local environment. Similarly, Hg in herbivorous fish in the Rio Negro basin also had no systematic correlation with sitedependent geochemistry (Barbosa et al., 2003). We had expected a positive correlation between pH and flooded areas because the anoxic condition that occur in flooded areas favor the accumulation and release of organic materials into the system, including fulvic and humic acids (Junk, 1997). Also, a correlation observed between DOC and pH indicates a mechanistic relation between these parameters. DOC in the Rio Negro system consists predominantly of humic and fulvic acids (Hedges et al., 1986) and since most tributaries in the system are extremely poor in carbonate buffering capacity (Belger and Forsberg, 2006), pH is controlled predominantly by the concentration of these organic acids. The pH and DOC values were similar to the ones found by Belger and Forsberg (2006), varying from an acid pH of 4.0 to 6.0 and a low DOC of 7 to 25 mg/L.
4.1.
Hg levels and environmental correlates
4.2.
The results of this study showed that environmental factors did not explain the variations in the concentration of mercury in the blood, muscle, liver and carapace of P. erythrocephala in the Rio Negro Basin. This is likely a reflection of the fact that P. erythrocephalata, although it is not a migratory species, moves extensively from one area to another during the year. Belger and Forsberg (2006) studied two species of fish (Cichla spp. and Hoplias malabaricus) and found that mercury levels in the relatively sedentary H. malabaricus were more correlated with environmental factors than in the more erratic Cichla spp. Thus, even though P. erythrocephala is a non-migratory species, the fact that it moves extensively in different tributaries suggests that the species is not a good bioindicator of local environmental conditions. In an open and interconnected aquatic systems, such as the Amazon basin that is characterized by high levels of limnological variability, a more locallydistributed species might be a better indicator of environmental characteristics. In addition, the mercury levels in P. erythrocephala were relatively low due to its exclusively herbivorous diet (Vogt,
Fig. 3 – Relation between Hgtot in carapace and turtle size, using the carapace length as an index of size.
Hg levels and biological correlates
Although Hg levels in P. erythrocephala did not reflect local environmental conditions, they could reflect internal Hg levels. However, the lack of correlation between mercury in muscle, liver and carapace indicates that the non lethal collection of a small piece of carapace is not effective as a bioindicator because the carapace does not predict Hg levels in the muscle or liver. For snapping turtles (Chelydra serpentina), the scute was a suitable material for mercury monitoring because there was a high correlation between scute concentrations and muscle, which also occurs with hair in mammals and feathers of birds (Golet and Haines, 2001; Burger and Gochfeld, 2001). However, this did not occur for the redheaded river turtle in the Rio Negro Basin. The Hg concentration in the herbivorous P. erythrocephala muscle in this study (10 to 106 ng g− 1) was lower than previous studies with the omnivore Chelydra serpentina: 50 to 500 ng g− 1 in 5 lakes in Connecticut (Golet and Haines, 2000), 50 to 300 ng g− 1 in Minnesota (Helwig and Hora, 1983) and 100 to 170 ng g− 1 in Tennessee (Meyers-Schöne et al., 1993). P. erythrocephala have similar Hg levels, around 100 ng g− 1, as herbivorous fish from the Rio Negro basin (Barbosa et al., 2003) and from the Rio Tapajós Basin (Malm et al., 1995). The significant accumulation of Hg in Amazonian turtles and fish reflects environmental Hg bioavailability (Roulet et al., 2000). Mercury bioaccumulation in aquatic systems varies considerably with food-chain structure and length. Mercury levels were relatively low in P. erythrocephala in the present study, perhaps because it is exclusively herbivorous and the aquatic plants and fruits it eats are typically low in mercury (Vogt, 2001). A lack of correlation between body size and mercury levels was also reported by Helwig and Hora (1983) and Golet and Haines (2000) who studied the omnivorous Chelydra serpentina. They suggest that chelonians may be eliminating mercury from the body tissues at higher rates than fish, and can reach equilibrium with ingested mercury. This hypothesis is supported by the lack of correlation found in this study between size and blood, liver and muscle of the Red-headed river turtle; there was, however, a
S CI EN C E OF TH E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
strong correlation between carapace and size. Blood, muscle and liver are likely to be more variable with daily intake and elimination of mercury, whereas carapace integrates mercury intake over longer periods of time. The shell of chelonians is composed of a deep, bony dermal-endoskeleton and an external corneous epidermis subdivided into rows of scutes (Zangerl, 1969). These scutes are the hard keratinized plates that armor the carapace and they originate from growing regions at the basal infoldings of the epidermis. The scutes form a series of nonliving keratin layers deposited one on the top of the other (Solomon et al., 1986). Hg is deposited in a metabolically inactive keratin matrix or is bound in relatively immobile inorganic Hg complexes. Consequently, since Hg is very stable in non-living proteinaceous tissue (Burger and Gochfeld, 2001; Day et al., 2007), a correlation between Hg and turtle size can be obtained by analyzing the scutes (as was done for P. erythrocephala in the present study).
4.3.
Hg levels and human health
Because they are long-lived, widely-distributed, and seem to have similar amounts of Hg regardless of size, they may be a useful species for biological monitoring of environmental mercury contamination of specific interest for humans consuming them. The levels of total mercury we found in P. erythrocephala in muscle are under the maximum level of 500 ng g− 1 recommended by the World Health Organization and do not constitute a serious health risk for fish-eating human populations in the Rio Negro Basin. However, we found mercury levels in the liver higher than the maximum level recommended by the World Health Organization, and special attention should be given to this part of the body, especially since it is eaten by local people. Recommendations should be given to the local people to avoid eating the liver of turtles. Also, we recommend that carnivorous species of turtles in the Amazon areas with potentially high Hg levels, such as Chelus fimbriatus, be studied. This is a species traditionally consumed from the Rio Negro, and can be an important source of protein for the human population. Carnivorous species in the Amazon basin that have Hg levels higher than 500 ng g− 1, the maximum level recommended by the World Health Organization, should be given special attention; the muscle of C. fimbriatus might represent a health risk to the riverine people. Such a higher trophic level species could be a better bioindicator of exposure and effects (Burger 2006b).
5.
Conclusions
Environmental factors did not explain the variation in the concentration of mercury in the muscle of P. erythrocephala in the Rio Negro basin, which may occur because this species moves among local habitats and tributaries. There was no correlation between body size and mercury levels in P. erythrocephala, perhaps because it is exclusively herbivorous (and thus has low mercury levels overall), and may also be able to eliminate mercury from the muscle and liver. However, this Amazonian turtle accumulated significant amounts of Hg,
1053
comparable to those of herbivorous fish in the Rio Negro and other Amazonian Rivers, reflecting Hg bioavailability. Because they are long-lived, widely-distributed and seem to have similar amounts of Hg regardless of size, they may be a useful species for biological monitoring of environmental mercury contamination that can affect consumers of these turtles, including humans.
Acknowledgments The study was conducted under the auspices of IBAMA permit 147/2006 and financed by FINEP. Larissa Schneider was financed by CNPq. Blood analysis was funded by P30ES005022 and DEFC01-06EW07053. We thank Dr. Reinaldo Peleja and crew for helping in the lab work, Dr. Bruce Forsberg, Chris Jeitner and Dr. Michael Gochfeld for the comments and lab facilities, and Camila Ferrara and Ladislau Santos-Junior for helping to collect turtle samples.
REFERENCES Barbosa AC, Souza J, Dórea JG, Jardim WF, Fadini PS. Mercury biomagnification in a tropical black water, Rio Negro, Brazil. Arch Environ Contam Toxicol 2003;45:235–46. Belger L, Forsberg BR. Factors controlling Hg levels in two predatory fish species in the Negro River Basin, Brazilian Amazon. Sci Total Environ 2006;367:451–9. Bisinoti MC, Sargentini-Junior E, Jardim WF. Seasonal behavior of mercury species in waters and sediments from the Negro River Basin, Amazon, Brazil. J Braz Chem Soc 2007; 18(3):544–53. Bodaly RA, Hecky RE, Fudge RJ. Increases in fish Hg levels in lakes flooded by the Churchill river diversion, Northern Manitoba. Can J Fish Aquat Sci 1984;41:682–91. Bodaly RA, St. Louis VL, Paterson MJ, Fudge RJP, Hall BD, Rosenberg DM, et al. Metal ions in biological systems. In: Siegel A, Siegel H, editors. Hg and its effects on environmental and biology. New York: Marcel Dekker; 1997. p. 259–87. Branfireun BA, Heus A, Roulet NT. The hydrology and methylHg dynamics of Precambrian shield headwater peatland. Water Resour Res 1996;32:1785–94. Burger J. Metals in diamondback terrapins from New Jersey. Environ Monit Assess 2002;77:255–63. Burger J. Bioindicators: types, development, and use in ecological assessment and research. Environ Bioindicat 2006a;1:22–39. Burger J. Neurotoxicology and behavioral effects in reptiles. In: Gardner SC, Oberdorster E, editors. Toxicology of reptiles. Boca Raton: CRC Press; 2006b. p. 173–98. Burger J, Gochfeld M. Effects of chemicals and pollution on seabirds. In: Schreiber EA, Burger J, editors. Biology of marine birds. Boca Raton: CRC Press; 2001. p. 525. Castillo MM, Allan JD, Sinsabaugh RL, Kling GW. Seasonal and interannual variation of bacterial production in lowland rivers of the Orinoco basin. Freshw Biol 2004;49:1400–14. Day RD, Segars AL, Arendt MD, Peden-Adams MM. Relationship of blood mercury levels to health parameters in the loggerhead sea turtle (Caretta caretta). Environ Health Perspec 2007; 115(10):1421–7. Fadini PS, Jardim WF. Is the Negro River Basin (Amazon) impacted by naturally occurring Hg? Sci Total Environ 2001;275(1):71–82. Foo SC, Ngim CH, Salleh I, Jeyaratnan J, Boey KW. Neurobehavioral effects in occupational chemical exposure. Environ Res 1993;60:267–73.
1054
S CI EN CE OF T H E T OTAL EN V I RO N M EN T 4 0 7 ( 2 0 09 ) 10 4 8–1 05 4
Forsberg BR, Silva-Forsberg MC, Padovani CR, Sargentini E, Malm O. High level of Hg in fish from the Rio Negro Basin (Brazilian, Amazon): natural background or anthropogenic contamination? In: Kato H, Pfeifer WC, editors. International workshop on environmental hg pollution and its health effects in the Amazon River basin, National Institute for Minamata Disease. Rio de Janeiro: Universidade Federal do Rio de Janeiro; 1995. p. 33–40. Golet WJ, Haines TA. Snapping turtles (Chelydra serpentina) as monitors for mercury contamination of aquatic environments. Environ Monit Assess 2001;71:211–20. Goulding M, Carvalho ML, Ferreira EG. Rio Negro: rich life in poor water: Amazonian diversity and foodplain ecology as seen through fish communities. The Hague: SPB Academic Publishing; 1988. p. 200. Guimarães JRD, Meili M, Hylander LD, Silva EDE, Roulet M, Mauro JBN, et al. Mercury net methylation in five tropical flood plain regions of Brazil: high in the root zone of floating macrophyte mats but low in surface sediments and flooded soils. Sci Total Environ 2000;261:99–107. Haines TA, Komov V, Matey V, Jagoe C. Perch mercury content is related to acidity and color of 26 Russian lakes. Water Air Soil Pollut 1995;85:823–8. Hedges JI, Ertel JR, Quay PD, Grootes PM, Richey JE, Devol AH, et al. Organic carbon-14 in the Amazon River System. Science 1986;231:1129–31. Helwig D, Hora M. Polychlorinated biphenyl, mercury and cadmium concentrations in Minnesota snapping turtles. Bull Environ Contam Toxicol 1983;30:186–91. Hess L, Melack J, Novo EMLM, Barbosa CCF, Gatil M. Dual-season mapping of wetland inundation and vegetation for the central Amazon basin. Remote Sens Environ 2003;87:404–28. Hylander LD, Sollenberg H, Westas H. A three-stage system to remove mercury and dioxins in flue gases. Sci Total Environ 2003;304(1–3):137–44. Junk WJ. Wetlands of tropical South America. Wetlands of the World 1993;11:679–739. Junk WJ. The central Amazon floodplain. Ecology of a pulsing system. Berlim: Springer; 1997. p. 100. Malm O, Branches FJP, Akagi H, Castro MB, Pfeiffer WC, Arada M, et al. Mercury and methylmercury in fish and human hair from Tapajós river basin, Brazil. Sci Total Environ 1995;175 (2):141–50.
Malm O, Pfeifer WC, Bastos WR, Souza CMM. Utilização de acessório de geração de vapor frio para análise de mercúrio em investigações ambientais por espectometria de absorção atômica. Ciência e Cultura 1989;41:88–92. Meyers-Schöne L, Shugart L, Beauchamp J, Walton B. Comparison of two freshwater turtle species as monitors of radio nuclide and chemical contamination: DNA damage and residue analyses. Environ Toxicol Chem 1993;12:1487–96. Roulet M, Guimarães JRD, Lucotte M. Methylmercury production and accumulation in sediments and soils of an Amazonian floodplain: effect of seasonal inundation. Water Air Soil Pollut 2001;128:41–61. Roulet M, Lucotte M, Rheault I, Guimarães JRD. Methylmercury in the water, seston and epiphyton of an Amazonian River and its floodplain, Tapajós River, Brazil. Sci Total Environ 2000;261:43–59. Serudo RL, Oliveira LC, Rocha JC, Paterlini WC, Rosa AH, Silva HC, et al. Reduction capability of soil humic substances from the Rio Negro basin, Brazil, towards Hg(II)studied by a multimethod approach and principal component analysis (PCA). Geoderma 2007;138:229–36. Silva-Forsberg MC, Forsberg BR, Zeidemann ZK. Mercury contamination in humans linked to river chemistry in the Amazon Basin. Ambio 1999;28:519–21. Solomon SE, Hendrickson JR, Hendrickson LPJ. The structure of the carapace and plastron of juveniles turtles, Chelonia mydas (the green turtle) and Caretta caretta (the loggerhead turtle). J Anat 1986;145:123–31. Storelli MM, Marcotrigiano GO. Heavy metal residues in tissues of marine turtles. Mar Pollut Bull 2003;46:397–400. Tardy Y, Bustillo V, Roquin C, Mortatti J, Victoria R. The Amazon. Bio-geochemistry applied to river basin management Part I. Hydro-climatology, hydrograph separation, mass transfer balances, stable isotopes, and modeling. Appl Geochem 2005;20:1746–829. Vogt RC. Turtles of the Rio Negro. In: Chao NL, Petry P, Prang G, Sonneschien L, Tlusty M, editors. Conservation and management of ornamental fish resources of the Rio Negro Basin, Amazonia, Brazil. Manaus: Editora Universidade do amazonas; 2001. p. 245–62. Zangerl R. The turtle shell. In: Gans C, editor. Volume 1: morphology. London: Academic Press; 1969. p. 311–39.