DA: Recommend restrictions to farmed salmon use?
Thic page contains the discussion in Science related to an article by Hites et al (2004).
Structure of the assessment
Focus
To study the recommendation about eating or not eating farmed salmon; and explore decision about actions on feed contaminants.
Scope
- Both health risks and benefits related to consumption of farmed salmon are included.
- Monetary issues are excluded.
- Health effects: Total mortality, cardiovascular mortality, total cancer incidence and mortality, breast and liver cancer incidence
- Population: Adults in Western Europe (European Economic Area)
- Pollutants: Toxaphene, dieldrin, dioxin, PCB
- Nutrients: Omega-3 fatty acids
- Decision: see focus
- Time: Current situation, no dynamics, no discounting
- Methods used: Existing literature
- Environmental effects: not included
Describe the value judgements.
Causal chain
- Driving force: Fish is a good food and farmed salmon is easily available with a reasonable price.
- Pressures: Fish is healthy, but on the other hand it contains pollutants such as pesticide residues (Toxaphene, dieldrin) and persistent organic pollutants (dioxin, PCB).
- States:
- Farmed salmon consumed: Data comes from EPIC study looking at fish consumption in 10 European countries by gender. We take the minimum, the unweighed average and the maximum of these values (7.5, 15.3, 31 g/d) in the distribution to represent uncertainty in population average fatty fish intake. All fatty fish is assumed to be salmon.
- Pollutant concentrations in farmed salmon (in µg/kg in fresh weight):
Dieldrin 3.299, Toxaphene 96.14, Dioxin 1.891e-003, PCB 36.25. Data according to Hites et al, 2004.
- Exposure: Pollutant exposures in general population (in µg/kg/d):
Dieldrin 7.417e-004, Toxaphene 0.02094, Dioxin 4.209e-007, PCB 8.105e-003 Data according to Tuomisto et al, 2004 model
- Effects:
- Dispute
- Dose-response function has a threshold or may even be U-shaped.
- Dose-response function is linear. In this assessment adopts this latter approach.
- Dispute
BAU | More limits on feed pollutants | |
BAU | -205.5 | -138.7 |
Recommend consumption restrictions | -149.3 | -101.1 |
BAU | More limits on feed pollutants | |
BAU | 31300 | 31680 |
Recommend consumption restrictions | 23050 | 23430 |
Data according to Tuomisto et al, 2004
- Actions:
BAU | More limits on feed pollutants | |
BAU | 16.13 | 16.3 |
Recommend consumption restrictions | 11.75 | 11.91 |
Data according to Tuomisto et al, 2004
Value judgements
- Values:
- Alternatives for farmed salmon are better, because they contain less pollutants: wild salmon, other fish, canola oil, flaxseed oil.
- Disgreement on the optimising strategy
- Net health benefit (public health) should be the main prioirity.
- Pollutant effect should be the main priority, and consumers themselves decide about the public health issues based on information provided.
- Conclusions:
- Pollutant risk is much smaller than the net health benefit of farmed salmon
- Some scientific and political uncertainties related to feed limits are important
- Scientific uncertainties related to recommendations are unimportant
- Questions:
- Should we change fish feed instead of giving fish consumption advisories?
- What has been done and what should be done to reduce pollutants in fish feed?
- Does omega 3 help other people than CHD patients?
Synthesis
The Health Benefits of Eating Salmon
THE RECENT REPORT BY R. A. HITES ET AL. about toxic contaminants in salmon (“Global assessment of organic contaminants in farmed salmon,” 9 Jan., p. 226) may have unintended negative consequences on human health. Yes, the source of toxic contaminants in farmed fish should be investigated and reduced as much as reasonably possible, but the proven health benefits of omega-3 fatty acids should not be overlooked.
The GISSI prevention trial (1) showed that consumption of 700 mg daily of omega-3 fatty acids from fish reduced total mortality by 20% in Italians with coronary artery disease (CAD). This corresponds to a number needed to treat (NNT) of 49 (95% confidence interval 30 to 175), indicating that one person with CAD avoids death when 49 people eat 21 g of omega-3 fatty acids monthly for 3.5 years. A monthly diet of 21 g of omega-3 fatty acids is present in ~1190 g (42 ounces) of salmon.
On the basis of U.S. Environmental Protection Agency (EPA) estimates, Hites et al. suggest that consumption of 55 g monthly of the most contaminated salmon would increase theoretical cancer risk by 1 in 100,000. If cancer risk were linear, 1190 g of salmon monthly would increase cancer risk by 22 cases per 100,000, corresponding to a number needed to harm (NNH) of 4500. This suggests that one person would develop cancer for every 4500 people who eat 1190 g of the most contaminated salmon (21 g of omega-3 fatty acids) monthly.
Therefore, the ratio of the NNH for cancer to the NNT for total mortality in people with CAD is 4500/49 = 92. This suggests that 92 Italians with CAD who eat 1190 g of salmon monthly would avoid death for every Italian that develops cancer. This analysis suggests that eating even the most contaminated salmon has clear health benefits. Clearly, salmon with fewer contaminants or other clean sources of omega-3 fatty acids would be better, but avoidance of salmon to avoid these contaminants without replacing the omega- 3 fatty acids from other sources would have adverse health implications.
The quality of the evidence in favor of the benefits of omega-3 fatty acids in people with CAD is high: The GISSI prevention trial was a randomized clinical trial. The GISSI trial estimates were only for 3.5 years of treatment. The benefit is likely greater for more prolonged treatment. In contrast, the EPA guidelines are estimates based on nonhuman toxicity or observational studies: There is no clinical trial showing that these toxins, when given to humans, cause cancer. Overall, this analysis suggests that ingestion of salmon should not be limited, especially in people with CAD.
CHRISTOPHER M. REMBOLD
University of Virginia, Box 801395, Charlottesville, VA 22908, USA.
Reference 1. F.Valagussa et al., Lancet 354, 447 (1999).
Response
REMBOLD RECOMMENDS THAT THE “INGESTION of salmon should not be limited, especially in people with [coronary artery disease].” As we acknowledge in our Report, there are health benefits associated with consumption of omega-3 fatty acids from fish, and calculation of the overall riskbenefit of contaminants in fish that are high in heart-healthy fats is challenging. To date, there has been no comprehensive assessment of the risk/benefit trade-offs associated with the consumption of such fish. In fact, the most recent scientific statement from the American Heart Association, although recommending fish as a heart-healthy food, states that “the fish recommendations must be balanced with concerns about environmental pollutants, in particular PCB and methyl mercury…” (1). In addition, omega-3 fatty acids do not protect against cancer (2).
To complicate matters, the contaminants we report in salmon are also associated with a variety of noncancer effects. Of particular note are the anthropometric and neurobehavioral effects of PCBs in children exposed in utero and during early childhood (3).
A comprehensive risk/benefit analysis for contaminated, farmed Atlantic salmon must account for all health risks associated with exposure to contaminants in these fish. Nevertheless, such an assessment, although helpful, is not essential to protect public health because there are many alternative sources of omega-3 fatty acids that have considerably lower contaminant concentrations, including wild Pacific salmon. These sources include other seafood that is not contaminated and nonfish foods, including flaxseed, walnut and canola oils, nuts, and legumes. One does not need to eat contaminated salmon to consume omega-3 fatty acids.
RONALD A. HITES,1* JEFFERY A. FORAN,2 DAVID O. CARPENTER,3 M. COREEN HAMILTON,4 BARBARA A. KNUTH,5 STEVEN J. SCHWAGER6
1School of Public and Environmental Affairs, Indiana University, Bloomington, IN 47405, USA. 2Midwest Center for Environmental Science and Public Policy, Milwaukee, WI 53202, USA. 3Institute for Health and the Environment, University at Albany, Rensselaer, NY 12144, USA. 4AXYS Analytical Services, Post Office Box 2219, 2045 Mills Road, Sidney, BC, Canada V8L 3S8. 5Department of Natural Resources, Cornell University, Ithaca, NY 14853, USA. 6Department of Biological Statistics and Computational Biology, Cornell University, Ithaca, NY 14853, USA. E-mail: HitesR@Indiana.edu
References
1. P. M. Kris-Etherton,W. S. Harris, L. J. Appel, Circulation 106, 2747 (2002). 2. P. D. Terry, T. E. Rohan, A.Wolk, Am. J. Clin. Nutr. 77, 532 (2003). 3. S. I. Schantz, J. J.Widhom, D. C. Roce, Environ. Health Perspect. 111, 357 (2003).
Risk-Benefit Analysis of Eating Farmed Salmon
IN THEIR REPORT “GLOBAL ASSESSMENT OF organic contaminants in farmed salmon,” R. A. Hites and co-workers analyzed wild and farmed salmon samples from North and South America and Europe for organic pollutants (9 Jan., p. 226). The authors conclude that, because of chemical contaminants, farmed salmon should not be eaten more often than 0.25 to 1 times per month. However, the model used does not take into account any beneficial effects of eating fish.
We analyzed the risks and benefits of the recommendation to reduce the intake of farmed salmon to 1 meal (227 g) per month. The authors estimated cancer risk of polychlorinated biphenyls (PCBs), toxaphene, and dieldrin using a U.S. Environmental Protection Agency model, which maximizes the estimated risk by assuming a linear correlation between cancer and exposure and by using upper confidence interval estimates (1). We included cardiovascular benefits of omega-3 fatty acids (2, 3), but no other positive effects associated with fish. We calculated the benefits as the best available estimates, being careful not to exaggerate benefits, if data were sparse.
We also performed a value-of-information (VOI) analysis (4, 5) for the decision. There is always uncertainty about the true values of variables affecting the decision. It often prevents the decision-maker from knowing the optimal alternative. VOI is defined as the expected benefit that occurs when an uncertainty is resolved, and the decision can be based on more solid evaluation. In our analysis, political questions (other than the decision under analysis) and scientific uncertainties were both treated in the same manner as uncertain variables.
The effects were estimated for the European Economic Area countries (population 387 million). Excess cancer mortality due to pollutants in farmed salmon was estimated at 210 cases per year [90% confidence interval (CI) 110 to 340], supporting restrictive recommendations. The number of cancer deaths that could be prevented by the restrictive recommendation on farmed salmon use was estimated at 40 deaths per year (90% CI 2 to 110). However, the recommendation would worsen the net health effect (cancer and cardiac deaths combined) by 5200 deaths per year (90% CI 34 to 19,000). It is therefore clear that if the main concern is the net health benefit, the decision-maker will not recommend restrictions. None of the scientific uncertainties considered, e.g., levels of pollutants in farmed versus wild salmon (4), changed this conclusion. The cost of not knowing (i.e., the VOI) whether pollutant effect or net health effect should be considered was estimated at 20 avoidable deaths per year. As shown above, scientific uncertainties have little relevance for the decision about recommending reduction in the intake of farmed salmon.
However, the importance of scientific uncertainties depends on the decision under analysis. This is clearly seen when considering an alternative way to reduce the cancer risk due to pollutants in farmed salmon. We analyzed a decision to lower the amount of pollutants in fish feed. This lowering was estimated to save 360 deaths per year (90% CI 3200 to +4100), mainly because of possible increase in consumption of salmon. In this case, several scientific and political uncertainties influenced which decision alternative appeared to be the optimal one (see figure). The variable with the largest VOI was how consumers would change their consumption of salmon after being informed that further regulations are needed for fish feed.
In conclusion, the question about restricting consumption of farmed salmon appears to be nonscientific, because the outcome of the analysis was totally driven by a political variable, whether to ignore the health benefits of fish. The question about fish feed regulation was partly scientific and would benefit from further research.
JOUNI T. TUOMISTO,1 JOUKO TUOMISTO,1,2 MARKO TAINIO,1 MARJO NIITTYNEN,1 PIA VERKASALO,1 TERTTU VARTIAINEN,1,3 HANNU KIVIRANTA,1 JUHA PEKKANEN1
1Centre of Excellence of Environmental Health Risk Analysis, Department of Environmental Health, National Public Health Institute, Post Office Box 95, FIN-70701 Kuopio, Finland. 2Department of Public Health and General Practice, University of Kuopio 70210, Finland. 3Department of Environmental Sciences, University of Kuopio, Kuopio 70210, Finland.
References and Notes
1. U.S. EPA, Guidance for Assessing Chemical Contaminant Data for Use in Fish Advisory, vol. 2, Risk Assessment and Fish Consumption Limits. (U.S. EPA, Washington, DC, ed. 3, 2000).
2. J. N. Din, D. E. Newby, A. D. Flapan, Br. Med. J. 328, 30 (2004).
3. C. R. Harper, T. A. Jacobson, Arch. Int. Med. 161, 2185 (2001).
4. M. G. Morgan, M. Henrion, Uncertainty:A Guide to Dealing with Uncertainty in Quantitative Risk and Policy Analysis (Cambridge Univ. Press, Cambridge, 1990).
5. See supplementary online material at Science Online at www.sciencemag.org/cgi/content/full/305/5683/ 476/DC1.
6. This study was funded by the Academy of Finland, grant 53307, the National Technology Agency of Finland (Tekes), grant 40715/01, and European Commission, contract QLK4-CT-1999-01446.
Response
TUOMISTO ET AL. PROVIDE AN INTERESTING analysis of the risk/benefit trade-offs associated with consuming farmed salmon with elevated contaminant concentrations. We agree with their point that any decision regarding the consumption of a contaminated food must balance risks and benefits. As we state in our Report, the presence of elevated levels of contaminants in farmed salmon complicates the risk/benefit equation. Without contaminants, farmed salmon would indeed be an ideal source of protein, rich in heart-healthy omega-3 fatty acids.
Cancer risk estimates allow comparison of the health risks associated with consumption of farmed Atlantic versus wild Pacific salmon and demonstrate the importance of considering alternative sources of salmon that provide the benefits of high omega-3 concentrations with considerably lower contaminant concentrations. Regardless of the methodological issues, the risk/benefit equation is clearly tipped in the direction of net benefit for fish low in contaminants and high in hearthealthy fats such as wild Pacific salmon. Other foods can also provide these same benefits, without commensurate contaminant- associated risk.
Tuomisto et al. conclude that reducing contaminants in fish feed is the most effective way to reduce risk and preserve benefits of consuming farmed Atlantic salmon. We agree with this conclusion, although further research is required to fully understand the role of feed in contributing to tissue concentrations of the chlorinated organic contaminants in farmed Atlantic salmon. We understand that some feed producers have been taking steps to reconstitute their feed, in part by looking to substitutes for fish meal and oil (1). We would expect a reduction in contaminants in feed as a result, although the process must be accompanied by a comprehensive monitoring program to ensure that the intended benefits are realized. Until then, consumers need to be aware of the risks, as well as benefits, of consuming contaminated salmon so they can make informed choices based on their own health concerns.
RONALD A. HITES,1* JEFFERY A. FORAN,2 DAVID O. CARPENTER,3 M. COREEN HAMILTON,4 BARBARA A. KNUTH,5 STEVEN J. SCHWAGER6
1School of Public and Environmental Affairs, Indiana University, Bloomington, IN 47405, USA. 2Midwest Center for Environmental Science and Public Policy, Milwaukee,WI 53202, USA. 3Institute for Health and the Environment, University at Albany, Rensselaer, NY 12144, USA. 4AXYS Analytical Services, Post Office Box 2219, 2045 Mills Road, Sidney, BC, Canada V8L 3S8. 5Department of Natural Resources, Cornell University, Ithaca, NY 14853, USA. 6Department of Biological Statistics and Computational Biology, Cornell University, Ithaca, NY 14853, USA. E-mail: HitesR@Indiana.edu
Reference
1. D. Rideout, Canadian Aquaculture Industry Alliance, personal communication.
Cancer Risk and Salmon Intake
IN THEIR REPORT “GLOBAL ASSESSMENT OF organic contaminants in farmed salmon,” R. A. Hites et al. give firm recommendations for safe monthly intake of farmed Atlantic salmon on the basis of concentrations of organic pollutants in raw fish (9 Jan., p. 226). In Norway, the advice not to eat more than half a portion of farmed salmon per month (approximately 110 g of raw fish) received extensive attention because the estimated per capita intake is about 200 g of farmed salmon per month. In the “Norwegian Women and Cancer study,” there are data that may clarify the implied cancer risk. In this study (1–3), the consumption of fish was assessed by postal questionnaires, with questions on species (salmon and trout), seasonality, frequency, and amount eaten. Because wild salmon is not generally available in Norway, the consumption information reported here corresponds mostly to farmed fish. Follow-up was based on linkage to the Cancer Registry of Norway, registers of death, and emigration data for 64,674 women with complete information (mean age = 51.1 years).
As shown in the table, the relative risk for cancer was not increased among women having a higher self-reported consumption than among those with less than the recommended intake. We found no indication of a dose response.
Dioxin concentrations in Norwegian farmed Atlantic salmon, based on a larger sample size, are about 20% lower than those reported by Hites et al. (4). Our findings suggest that the cancer risk assigned to PCBs, dieldrin, and toxaphene is overestimated, presumably reflecting the inherent uncertainty in the EPA cancer slope factors employed (5).
Oily fish is an important source of essential fatty acids and lipid soluble vitamins. The positive health effects are well documented for cardiovascular diseases (6). Unilateral promotion of very limited health hazards could affect the overall health of populations by urging people to give up a healthy diet, causing them to substitute fish with less healthy and perhaps less safe foods.
EILIV LUND,1 DAGRUN ENGESET,1 ELIN ALSAKER,1 GUN SKEIE,1 ANETTE HJÅRTAKER,2 ANNE-KATRINE LUNDEBYE,3 EVERT NIEBOR4
1Institute of Community Medicine, University of Tromsø, 9037 Tromsø, Norway. 2Department of Statistics, Institute of Basic Medical Science, University of Oslo, Box 1122 Blindern, 0317 Oslo, Norway. 3National Institute of Nutrition and Seafood Research (NIFES), Box 176 Sentrum, 5804 Bergen, Norway. 4Department of Biochemistry, McMaster University, 1200 Main Street West, Hamilton, ON, Canada L8N 3Z5.
References
1. E. Lund et al., Cancer Causes Controls 14, 1001 (2003).
2. E. Riboli et al., Public Health Nutr. 5 (6B), 1113 (2002).
3. Supporting material available at www.ism.uit.no/kk/e
4. A.-K. Lundebye et al., Aquacul. Nutr., in press.
5. U.S. EPA, Guidance for Assessing Chemical Contaminant Data for Use in Fish Advisories, vol. 2, Risk Assessment and Fish Consumption Limits (U.S. EPA, Washington, DC, ed. 3, 2000).
6. P. M. Kris-Etherton et al., Circulation 106, 2747 (2002).
Response
LUND ET AL.ADDRESS THE RISK/BENEFIT ISSUE with a bit of epidemiological data. The risk assessment used by EPA is to protect against one extra case of cancer in 100,000 (not the more conservative and usual 1 × 10 6). Lund et al. report on only 1517 cases of cancer in a total population of about 61,000. This sample is much too small to detect elevations at this level of risk. In addition, the power of the study discussed by Lund et al. to detect excess cancer risk associated with consuming contaminated fish is quite low. Indeed, using a cancer risk of 1 in 100,000, as we did, gives <1 additional case of cancer in the 64,000 people they studied. Furthermore, the “higher exposure group” in this study (individuals consuming more than 110 g/month of fish, or about one-half of a meal per month) is likely still at the very low end of the exposure range included in our study. The most restrictive, risk-based advice for the most contaminated fish was in this same range, and the “safe consumption range” of farmed Atlantic salmon from Norway (consumption that reduces cancer risk to 1 × 10 5) is approximately one-half of a meal per month.
Although omega-3 fatty acids do protect against sudden cardiac death after a heart attack, young people are not at risk of heart attacks, but their risk of cancer at older ages is increased by exposure to these compounds. In addition, the noncancer effects of exposures, such as reduced cognitive function of children exposed before birth, could overshadow a protective effect against sudden cardiac death in young people and were not factored into the cancer-based risk analysis in our Report.
RONALD A. HITES,1* JEFFERY A. FORAN,2 DAVID O. CARPENTER,3 M. COREEN HAMILTON,4 BARBARA A. KNUTH,5 STEVEN J. SCHWAGER6
1School of Public and Environmental Affairs, Indiana University, Bloomington, IN 47405, USA. 2Midwest Center for Environmental Science and Public Policy, Milwaukee, WI 53202, USA. 3Institute for Health and the Environment, University at Albany, Rensselaer, NY 12144, USA. 4AXYS Analytical Services, Post Office Box 2219, 2045 Mills Road, Sidney, BC, Canada V8L 3S8. 5Department of Natural Resources, Cornell University, Ithaca, NY 14853, USA. 6Department of Biological Statistics and Computational Biology, Cornell University, Ithaca, NY 14853, USA. E-mail: HitesR@Indiana.edu
Contaminant Levels in Farmed Salmon
THE REPORT “GLOBAL ASSESSMENT OF organic contaminants in farmed salmon” by R. A. Hites et al. (9 Jan., p. 226) reaches a conclusion that the contaminant levels in farmed salmon are significantly different than those in “wild” salmon by making comparisons between different species of salmon from different water sources. Yet there have been many news stories (1) that provide data indicating that the levels in “wild” salmon near areas such as Puget Sound and the Columbia River have contamination levels comparable to the highest contamination levels in farmed salmon in Scotland.
The authors state that “[u]ptake of organic contaminants from water to fish is a minor accumulation pathway.” However, the reference that they cite (2) states that the relative significance of water for both uptake and elimination is a function of the log Kow (octanol/water partition coefficients) of the specific chemical. The range of log Kow’s (3) for the chemicals in question is similar to those of compounds for which both feed and water can be a source and contaminant-free water can depurate (be removed from the organism) (4) on a reasonable time scale (5).
Without data on the specific contaminants in the feed and fish, it is not possible to make an accurate transport model (6). However, a rough model (7) indicates that about 90% of the contamination in the farmed salmon could be from the water, not the feed. Thus, changing fish farm locations or installing depuration systems in a cleaner water area may be a preferable solution to a contamination problem than changing fish feed formulations; hence, it is critical to know the real source of the contaminants.
DALLAS E.WEAVER
Scientific Hatcheries, 5542 Engineer Drive, Huntington Beach, CA 92649, USA. E-mail: Deweaver@surfcity.net
References and Notes
1. For example, see R. McClure, L. Stiffler, “Sound’s salmon carry high PCB levels,” Seattle Post- Intelligencer, 15 Jan. 2004 (available at http://seattlepi. nwsource.com/local/156714_warning15.html), using data from Washington Department of Fish and Wildlife PSAM program; see wdfw.wa.gov/fish/ psamp/findings.htm
2. R.W. Russell, F. A. P. C. Gobas, G. D. Haffner, Environ. Toxicol. Chem. 18, 1250 (1999).
3. A. Sabljic, H. Giisten, J. Hermens, A. Opperhulred, Environ. Sci. Technol. 27, 1394 (1993).
4. P. N. Fitzsimmons, J. D. Fernandez, A. D. Hoffman, B. C. Butterworth, J. W. Nichols, Aquatic Toxicol. 55, 23 (2001).
5. F. Verweij, K. Booij, K. Satumalay, N. van der Molen, R. van der Oost, Chemosphere 54, 1675 (2004).
6. M. C. Barber, Environ. Toxicol. Chem. 22, 1963 (2003).
7. Assuming mass-transport kinetics time constants on the order of 40 days, food conversion ratios of 1:1 (dry feed/animal wet weight), standard feed tables (feed rate as a function of size and temperature), and normal life cycle for Atlantic Salmon.
Response
INDEED,THERE ARE ELEVATED CONCENTRATIONS of contaminants in local populations of wild Pacific salmon, such as those cited by Weaver, and in Pacific salmon from the Great Lakes (which are not sold commercially). There are likely other examples of local contaminant hot spots in wild Pacific salmon as well. However, the wild Pacific salmon from Alaska and Canada that we collected make up a substantial portion of wild fish sold in commercial markets globally. Since consumers purchase these wild salmon for consumption, these samples were the most appropriate for comparison with farmed Atlantic salmon.
We state in our Report that the most likely source of contaminants in farmed Atlantic salmon is the feed. We disagree with Weaver’s contention that the main source is the water. Farmed Atlantic salmon are fed over extended periods of time a diet of fish feed with elevated contaminant concentrations. Depuration would only be pertinent if the contaminant exposure occurred only at a limited number of feedings. This is not the case. Furthermore, the consistency of contaminant concentrations between farms within regions, which are likely to use feed composed of fish from within their region, suggests that accumulation from feed is the likely pathway. In addition, it is unlikely that contaminants from water contribute significantly because these substances are not very water soluble. However, further studies of contaminant sources for farmed species such as Atlantic salmon are needed.
RONALD A. HITES,1* JEFFERY A. FORAN,2 DAVID O. CARPENTER,3 M. COREEN HAMILTON,4 BARBARA A. KNUTH,5 STEVEN J. SCHWAGER6
1School of Public and Environmental Affairs, Indiana University, Bloomington, IN 47405, USA. 2Midwest Center for Environmental Science and Public Policy, Milwaukee, WI 53202, USA. 3Institute for Health and the Environment, University at Albany, Rensselaer, NY 12144, USA. 4AXYS Analytical Services, Post Office Box 2219, 2045 Mills Road, Sidney, BC, Canada V8L 3S8. 5Department of Natural Resources, Cornell University, Ithaca, NY 14853, USA. 6Department of Biological Statistics and Computational Biology, Cornell University, Ithaca, NY 14853, USA. E-mail: HitesR@Indiana.edu