Ecotoxicology, 14, 163–180, 2005 2005 Springer Science+Business Media, Inc. Manufactured in The Netherlands.
Mercury in Freshwater Fish of Northeast North America – A Geographic Perspective Based on Fish Tissue Monitoring Databases NEIL C KAMMAN,1,* NEIL M. BURGESS,2 CHARLES T. DRISCOLL,3 HOWARD A. SIMONIN,4 WING GOODALE,5 JANICE LINEHAN,6 ROBERT ESTABROOK,7 MICHAEL HUTCHESON,8 ANDREW MAJOR,9 ANTON M. SCHEUHAMMER1,10 AND DAVID A. SCRUTON11 1
Vermont Department of Environmental Conservation – Water Quality Division, 103 S Main 10N, Waterbury, VT USA 05671-0408 2 Canadian Wildlife Service, Environment Canada-Atlantic Region, Mt. Pearl, NL Canada 3 Department of Civil and Environmental Engineering, Syracuse University, Syracuse, NY USA 4 New York State Department of Environmental Conservation, Rome Field Station, Rome, NY USA 5 Biodiversity Research Institute, Gorham, ME USA 6 Canadian Natural Resources Ltd., Calgary, AB Canada 7 New Hampshire Department of Environmental Conservation, Concord, NH USA 8 Massachusetts Department of Environmental Protection, Boston, MA USA 9 United States Fish and Wildlife Service, Concord, NH USA 10 Canadian Wildlife Service – National Wildlife Research Centre, Ottawa, ON Canada 11 Fisheries and Oceans Canada, St. John’s, NL Canada Accepted 4 December 2004
Abstract. As part of an initiative to assemble and synthesize mercury (Hg) data from environmental matrices across northeastern North America, we analyzed a large dataset comprised of 15,305 records of fish tissue Hg data from 24 studies from New York State to Newfoundland. These data were summarized to provide mean Hg concentrations for 40 fish species and associated families. Detailed analyses were carried out using data for 13 species. Hg in fishes varied by geographic area, waterbody type, and waterbody. The four species with the highest mean Hg concentrations were muskellunge (Esox masquinongy), walleye (Sander vitreus), white perch (Morone americana), and northern pike (Esox luscius). Several species displayed elevated Hg concentrations in reservoirs, relative to lakes and rivers. Normalized deviations from mean tissue levels for yellow perch (Perca flavescens) and brook trout (Salvelinus fontinalis) were mapped, illustrating how Hg concentrations in these species varied across northeastern North America. Certain geographic regions showed generally below or above-average Hg concentrations in fish, while significant heterogeneity was evident across the landscape. The proportion of waterbodies exhibiting exceedances of USEPA’s criterion for fish methylmercury ranged from 14% for standard-length brook trout fillets to 42% for standard-length yellow perch fillets. A preliminary correlation analysis showed that fish Hg concentrations were related to waterbody acidity and watershed size. Keywords: mercury; fish tissue; length; lake; reservoir; river; indicator; GIS *To whom correspondence should be addressed: Tel.: +1-802-241-3795; E-mail:
[email protected]
164 Kamman et al. Introduction The global consequences of atmospheric mercury (Hg) deposition resulting from anthropogenic emissions are becoming increasingly clear. Mercury is found in virtually all environmental matrices (USEPA, 1997; United Nations Environmental Programme, 2002), including air (Scherbatskoy et al., 1999), wetland soils (Norton et al., 1997), terrestrial vascular plants (Rea et al., 2002), and terrestrial biota (United Nations Environmental Programme 2002; Rimmer et al., 2005). Hg emissions from anthropogenic and natural sources in North America and other continents are well understood to enter the global Hg pool (Pacyna and Pacyna, 2002), depositing in areas worldwide (Lindberg et al., 2002; Biester et al., 2002). Aquatic ecosystems serve as sensitive receptors for deposited Hg. In aquatic ecosystems, biotic and abiotic methlyation increases the bioavailability and toxicity of Hg (Regnell, 1994; DiPasquale et al., 2000), which then biomagnifies from algae and bacteria to zooplankton (Chen et al., 2000; Pickhardt et al., 2002), fishes (USEPA, 1997; Verdon and Tremblay, 1999), and ultimately to piscivorous wildlife (Evers et al., 2002) and humans (National Academy of Sciences, 2000). In response to this pattern, state, provincial and federal government agencies in North America, particularly in the northeast USA and eastern Canada expend significant resources controlling Hg discharges to the environment, by regulating Hg emissions from waste incineration and releases through the disposal of Hg-bearing products (e.g., Northeast States for Coordinated Air Use Management, 2003; Vermont Advisory Committee on Hg Pollution, 2004). Fish are probably the most well-studied ecosystem component with regards to Hg concentration and bioaccumulation. Indeed, the United States and Canada have adopted federal advisories warning against excessive consumption of certain Hg-tainted fishes (Health Canada, 2002; United States Food and Drug Administration, 2004). Hg accumulation in freshwater fish is influenced by several factors, including fish size (USEPA, 1997), waterbody pH, dissolved organic carbon, and watershed characteristics (e.g., Mierle and Ingram, 1991; Spry and Wiener, 1991; Driscoll et al., 1994; Haines et al., 1994), algal productivity (Pickhardt et al., 2002; Kamman et
al., 2004); and zooplankton community structure (Chen et al., 2000, 2005). While most Canadian provinces and US states routinely analyze fish for Hg, these data are often not published. Accordingly, while numerous publications describe fish Hg specific to individual study areas or species, the body of scientific literature regarding fish Hg concentrations remains underrepresented, as the large datasets used by governments for issuing consumption advisories have not been systematically combined and evaluated. The Northeast States Research Consortium (Evers and Clair, 2005) has sponsored the establishment of a northeast North American Hg workgroup to compile and analyze as large an assembly of Hg data as practical from a wide variety of environmental matrices, focused around freshwater ecosystems. Herein, we describe the fish tissue dataset assembled in conjunction with this initiative, spanning the entire NSRC study region (Evers and Clair, 2005). We analyze these data by geographic region, waterbody type, and waterbodyspecific factors. In so doing, we provide a geographically-based synthesis of Hg in fish across the NSRC study area, and discuss the suitability of certain freshwater fish for use as indicators of aquatic Hg contamination. While other large regional or national datasets are in development (e.g., USEPA, 2001a), to our knowledge, the analyses presented herein are the first to be published from such a large dataset (compiled from 24 different studies) describing fish Hg concentrations and variation at the subcontinental scale.
Methods Source datasets We assembled existing fish Hg databases from willing donors across all geographic regions of the NSRC study area, which is described by Evers and Clair (2005). The fish tissue database spans the geographic range from 39.5 to 54.7 N latitude, and 53.9 to 79.5 W longitude (Fig. 1). Monitoring programs that are carried out by provincial and state governments for the purpose of risk assessment provided the largest datasets. Datasets from random probability surveys conducted within the United States were also actively collected, as were
Fish Mercury in Northeast North America 165
Figure 1. Distribution of fish tissue sampling locations across northeastern North America. Symbols vary by individual project contributing data to the NSRC mercury project.
other datasets derived from large-scale research initiatives. Smaller datasets such as those derived from academic researchers were solicited. We collected geo-referenced datapoints from 24 research and monitoring projects across the study region (Table 1), comprising an aggregate 19,815 individual Hg datapoints. All fish Hg concentrations are reported on a total mercury wet weight (w.w.) basis. We attempted to retain the maximum amount of data possible, while ensuring that follow-on analyses were not unduly complicated by the large number of species present in the database. Accordingly, data were rejected from further analysis if they failed one or more of three criteria. First, only fish Hg measurements analyzed using cold-vapor atomic absorption or cold-vapor atomic fluorescence spectroscopy (USEPA 1994, method 245.1; USEPA 1996, method 1631; or equivalent) were retained. Second, only fish collected in 1980 or later were included in the final database. Fish Hg data from before 1980 may be elevated due to elevated emissions and deposition of Hg during that period (e.g., Engstrom and
Swain, 1996; Armheim and Geis, 2001, Kamman and Engstrom, 2002). Further, data from the Great Lakes and St. Lawrence River were excluded because these waterbodies were outside the focus of this NSRC assessment. Finally, only Hg concentrations derived from fish fillets or whole fish were retained, as these are commonly used for assessment of risk to humans and to wildlife. To ensure data quality, we subjected the dataset to a series of validation checks aimed at detecting outlier, mis-transcribed, or incorrect datapoints. These checks included examination of basic statistics, Tukey plots, and scatterplots of Hg by length and weight, by species and within project. Automated as database queries, the validity checks identified numerous datapoints with values that were either excessively high, presented in the wrong unit of measure, or mis-attributed to the wrong species. Datapoints so identified were either corrected from source datasets, or purged from the database. We also reviewed primary literature or ‘grey’ reports describing methodologies for each project. In summary, of the 19,178 original records
Intensive Intensive
Routine Intensive Routine Intensive
Routine Routine
Random Probability Intensive
AdxYP BRI
CTDEP CWS_ON DFO_LAB DFO_MART
DFO_NF EC_YLP
EMAP
Routine Random Probability Intensive
Routine Routine Intensive Routine Routine Routine Intensive
Routine Intensive
NB_LEPREAU
NB_SPF NHDES NHDES_SE NS_Power NS_SPF NY NY_Reservoir
QUE_SPF USGS_NE
MASS ME_REMAP
Kejimkujik
Purpose of project sampling
Project Name
Southwest New Brunswick New Brunswick New Hampshire Southeast NH Nova Scotia Nova Scotia New York New York City Reservoirs, NY Quebec Coastal rivers, Massachusetts, NH, Maine
Kejimkujik National Park, Nova Scotia Massachusetts Maine
Adirondacks, NY New York, Vermont, NH, Maine Connecticut Ontario Labrador Canadian Maritimes Newfoundland Canadian Maritimes New England
Project geographic area
)73.5828 )68.6657 )71. 7869 )76. 5000 )56. 6333 )63.2167 )53. 8333 )62.9000 )67.3006 )65.1778 )70.3900 )67.1750 )66. 5228 )65. 0833 )70. 8886 )71.0617 )61.5000 )60.5667 )73.3462 )73. 6357 )64. 7429 )69.6856
)75.2564 )72.2160 )73.5250 )79.4000 )67.4667 )69.3667 )59.2667 )67.5333 )78.8514 )65.4386 )73.1600 )71.0131 )66.7831 )68.3667 )72.4464 )71.5548 )65.9167 )66.0633 )79.3995 )74.4393 )79.5444 )71.8244
45.0250 41.9475
45.3667 42.7175 42.8885 43.9667 44.1667 40.6600 41.0894
45.1806
41.6300 43.2558
44.2831
39.4510
47.0833 44.1500
41.2512 44.4167 51.7000 43.8500
43.7133 43.1730
Max
Min
Min
53.2886 44.4736
47.9500 45.2986 43.1565 45.3000 46.1333 44.9607 42.3702
45.4494
42.8300 47.1197
44.4592
47.1998
51.4000 46.7333
42.0260 45.4667 54.9833 46.8500
44.5275 46.0133
Max
Geographic extent-west longitude
Geographic extent-north latitude
1983 1999
1994 1992 2001 1995 1994 1981 2001
1998
1986 1993
1996
1992
1981 1998
1995 1993 1982 1981
1991 1996
Yr begin
2003 2000
1994 2002 2001 1995 1994 1994 2002
1998
1994 1993
1997
1994
1981 1998
2001 1993 1982 1981
1991 2003
Yr end
Project duration
36 1
7 24 1 4 9 23 14
23
4 13
3
7
3 1
10 4 9 5
1 28
Species
272
18 63
40 142 15
15
37 79
26
113
74 33
53 24 81 98
15 49
Lakes
10
8 8
19
8
12 41
1 17
Reservoirs
199 26
6
8
8
3
15
Rivers
5408 26
126 1242 75 133 24 1609 391
177
254 353
332
132
863 97
682 200 1060 197
464 767
Hg datapoints
Number of database measurements by type
20 21
14 15 16 17 18 19 19
13
11 12
10
9
7 8
3 4 5 6
1 2
Reference
Table 1. Individual projects contributing data to the Northeast States Research Consortium analysis of mercury in fish tissues across northeastern North America
166 Kamman et al.
6 6 43 40 19 1 1999 2000 1987 1999 )71. 8750 )70.9586 )73.4250 )73.2808 45.0083 44.8600 42.7417 42.8008 Vermont Vermont and NH Routine Random Probability VTDEC VTNH_REMAP
References: 1. Driscoll et al. (1994). 2. Evers et al. (2002). 3. Neumann et al.(1996); Hanten et al. (1998).4. Atchison (1994). 5. Scruton (1984). 6. Peterson et al. (1990). 7. Scruton (1983). 8. Rutherford et al. (1998). 9. USEPA (1995). 10. Drysdale et al. (2005). 11. Rose et al. (1999). 12. Stafford and Haines (1997); MEDEP (1995). 13. Barry and Curry (1998). 14. N.B. Dep’t Health Community Services (1994). 15. N.H. Department of Environmental Services (2003). 16. Major (2003). 17. Nova Scotia Power (1995). 18. Horwich (1994). 19. N.Y. Dept. Environ. Conserv. (2003). 20. Province of Quebec (2002). 21. Chalmers et al. (2002). 22. VTDEC (1995). 23. Kamman et al. (2004).
22 23 381 311
Hg datapoints Rivers Reservoirs Lakes Species Yr end Yr begin Max Min Min
Max
Geographic extent-west longitude Purpose of project sampling Project Name
Table 1. Continued
Project geographic area
Geographic extent-north latitude
Project duration
Number of database measurements by type
Reference
Fish Mercury in Northeast North America 167 submitted to the database, 15,305 records met screening criteria, passed validity checks, and were retained. Counts of Hg samples and of waterbody types within projects are shown in Table 1. Species and tissue types Data were classified by fish species, tissue type (fillet or whole body), and waterbody type (lake, reservoir, or river). While many of the species in the database were commonly sought gamefish, many others were non-game species or rare species. The final dataset provided Hg measurements for 64 freshwater fish species. This large roster of taxa under-represents certain species relative to others. For example, yellow perch (scientific names for fish species are listed in Appendix A) and brook trout are the most prevalent species, having 3821 datapoints across 21 projects, and 1456 datapoints across 14 projects, respectively. By contrast, several non-game or rare fish species are represented by only a few measurements, or were sampled only within a single project area. To further reduce the complexity of the analyses, we only analyzed data for those 13 species that either had 1000 or more Hg measurements per species, or were present in onethird or more (‡9) of the projects (Table 2). Fish Hg concentrations are known to vary with fish length, weight, and age (e.g., Driscoll et al., 1994; Kamman et al., 2004; Drysdale et al., 2005), and these factors were included in the dataset for records where available. Lengths were reported as either total length or fork length, depending on the project. We used fork length to total length ratios compiled from the literature by the online resource Fishbase (Froese and Pauly, 2003) to adjust all fork lengths to total lengths (Appendix A). For species where fork length to total length ratios were unavailable, we used relationships from species with similar morphology. Even while we constrained our detailed analyses to the 13 species listed in Table 2, we calculated basic statistics describing total Hg concentrations, for a given species or taxonomic family, for all species in the final dataset, as a baseline reference to support future inquiry (Appendix A). Species with more than 10 datapoints were treated separately (n=40 species), while species with 10 or fewer datapoints were grouped by family (n=6 families).
168 Kamman et al. Table 2. Analysis of covariance results for 13 freshwater fish species evaluating the influence of project area Tissue type
Fillet
Whole
Species
Brook trout Brown bullhead Brown trout Eastern chain pickerel Lake trout Landlocked salmon Largemouth bass Northern pike Smallmouth bass Walleye White perch White sucker Yellow perch Brook trout White sucker Yellow perch
Statistical model parameters Variance explained by project area (%)
Variance explained by waterbody within project (%)
Variance explained by length (%)
Overall model R2
7.4 NS 8.1 7.6 1.0 20.4 10.3 2.0 7.6 1.5 26.0 11.4 33.6 18.4 15.1 13.9
44.6 64.5 36.2 57.1 40.1 6.7 45.5 36.7 50.4 34.4 30.4 48.0 30.3 57.5 78.3 71.9
7.3 3.7 20.3 8.5 17.1 3.2 21.0 27.4 11.1 27.0 3.6 9.3 5.3 4.1