National Water-Quality Assessment (NAWQA) Project
Chalmers, A.T., Argue, D.M., Gay, D.A., Brigham, M.E., Schmitt, C.J., and Lorenz, D.L., 2010, Mercury trends in fish from rivers and lakes in the United States, 1969–2005. Environmental Monitoring and Assessment. Published online June 10, 2010. (available for free download from Environmental Monitoring and Assessment)
Four separate spreadsheets within the file Chalmers2010_TrendsDatasets.xls (Microsoft Excel file) contain the data sets used for analysis.
The source of data is described in the Introduction section of the paper. Trends were analyzed on fish Hg concentrations in U.S. rivers and lakes aggregated by site and by state. Specific subsets of data used for trend analysis are described in the Methods section of the paper, and are summarized here to provide a key for the data sets.
Data was retained for site trends analysis if fish Hg concentration, fish lengths, fish species, tissue type, and coordinate information was available. Each site trend analysis was limited to one fish species and tissue type. A restricted range of fish lengths were used to normalize data for differences in fish age and size. Data within 10 percent of the median length for each species and tissue type for a site were retained for trend analyses. Sites with all fish Hg concentrations below detection were dropped from the site trend analysis. If the difference between the highest and lowest fish Hg concentration was less than or equal to the detection level, the trend was not considered significant, even if the p-value was less than 0.05. Site trends were analyzed over two periods, 1969 – 2005, and 1969 – 1987.
Sites were selected for trend evaluation from 1969 to 2005 if they had at least seven fish samples within the restricted range of fish lengths that were of the same species and tissue type collected for at least different five years over more than a 10-year period between 1969 and 2005. Ninety of the original 7,759 river and lake sites met the criteria for trend analyses between 1969 and 2005.
Sites were selected for trend evaluation from 1969 to 1987 if they had at least seven fish samples within the restricted range of fish lengths that were of the same species and tissue type collected for at least different five years over more than a 10-year period between 1969 and 1987. Fifty trends sites met these criteria.
Data was retained for statewide trends analysis if information was included about fish Hg concentration, fish lengths, fish species, tissue type, and the state the sample was collected in. Each statewide trend analysis was limited to one fish species and tissue type. A restricted range of fish lengths were used to normalize data for differences in fish age and size. Data within 10 percent of the median length for each species and tissue type in a state were retained for trend analyses. Statewide trends were analyzed over 2 periods, 1988 – 2005, and 1996 – 2005.
Statewide trends were analyzed on an aggregate of all samples of a single fish species and tissue type from a state that met the following criteria: during the period 1988–2005, fish samples of one species and tissue type within the restricted length were collected for at least 7 different years for over at least a 10-year period. In addition, for each fish species and tissue type evaluated in the statewide trend, at least 30 sites had to have been sampled in the state over the 10-year period, with at least three sites having been sampled in both the first and last years of sample collection. Statewide trends were not analyzed prior to 1988 because of insufficient data.
Statewide trends were also analyzed from 1996 to 2005 using the same statewide criteria except that the period of record was reduced to at least 5 different years over a 6-year period from 1996 to 2005. The results of the fish Hg trend analyses from 1996–2005 provide a direct comparison with Hg wet deposition data that the National Atmospheric Deposition Program Mercury Deposition Network (NADP MDN) began collecting in 1995.
Statistical analysis was performed using Version 9.1.3 of the Statistical Analysis System (SAS Institute 2002). Trends were determined through the use of the SAS LIFEREG (linear regression) procedure, which yields estimates of parametric regression models for censored data using the method of maximum likelihood (Allison 1995). Detection levels that changed consistently over time (ie. typically decreasing over time with improve methods) in a given site or state were set to the highest detection level to reduce the potential for false trends. Hg concentrations in fish were natural log transformed to remove much of the heterogeneity of residuals. Linear changes in log Hg concentrations were considered significant if the p-value was less than 0.05. Three diagnostic statistics, Cook's D, leverage, and standardized residual statistics were used to identify data values that substantially influenced the fit of the regression line (Helsel and Hirsch 1992). Data values that diagnostic statistics identified as outliers were removed resulting in changes in trends at two sites. The annual percent change in fish Hg concentrations was calculated using the equation, (eB – 1) x 100, where B is the slope of the regression line (Hirsch et al. 1991).
Allison, P. D. (1995). Survival analysis using SAS: A practical guide. Cary, NC: SAS Institute Inc.
Hirsch, R. M., Alexander, R. B., & Smith, R.A. (1991). Selection of methods for the detection and estimation of trends in water quality. Water Resources Research, 27(5), 803-813.
Ann T. Chalmers
U.S. Geological Survey
87 State Street, PO Box 628
Montpelier, VT 06501, USA
Denise M. Argue
U.S. Geological Survey
331 Commerce Way
Pembroke, NH 03275-3718, USA
Mark E. Brigham
U.S. Geological Survey
2280 Woodale Drive
Mounds View, MN 55112, USA