Skip to main content
U.S. flag

An official website of the United States government

The statistical power to detect regional temporal trends in riverine contaminants in the Chesapeake Bay Watershed Completed

By Chesapeake Bay Activities January 20, 2022

Issue: Chemical contamination of riverine ecosystems is a global concern, with potentially negative effects for human and ecological health. Land management activities (e.g., Best Management Practices; BMPs) are an important tool that can be used to reduce point and non-point sources of pollution. The Chesapeake Bay Program (CBP) has goals for (1) reducing toxic contaminants and (2) reducing nutrients and sediment to improve water-quality conditions in the Bay and across the Chesapeake Bay Watershed. The BMPs to reduce nitrogen, phosphorus, and sediment also have a potential co-benefit of reducing toxic contaminants. However, the ability to confidently make predictions about the effects of land-management activities on reducing in-stream chemical concentrations is poorly understood. The CBP stakeholders need improved approaches to detect temporal changes (i.e., statistical power) in riverine chemical concentrations to help inform the design of monitoring programs; and to develop expectations regarding the amount of time needed to detect water-quality changes after BMP implementation (see figure 1).

Graphical abstract showing watershed sampling sites and BMPs
Sources/Usage: Public Domain. View Media Details
Figure 1. Graphical abstract. (from Wagner and others, 2021)

USGS study

The U.S. Geological Survey (USGS) conducted a study to provide new insights into the number of monitoring stations and amount of time needed to detect changes in river contaminants. Using data from USGS National Water Quality Network, the National Water Quality Monitoring Council Water Quality Portal, and the Pennsylvania Department of Environmental Protection, USGS scientists had two objectives: 

  • Quantify existing regional (watershed -wide) temporal trends in concentrations of select contaminants over the past several decades and estimate components of spatio-temporal variation, and 

  • Using estimated parameters from objective 1, perform simulations to determine the statistical power needed to detect regional temporal trends under different monitoring scenarios. The scenarios included varying the magnitude of the annual decline, the number of river sites sampled in the Chesapeake Bay Watershed each year, the number of years sampled, and annual sampling frequency. The selected contaminants included water concentrations of atrazine, metolachlor (herbicides), sediment concentrations of total polychlorinated biphenyls (PCBs), and total estrogenicity, which is an indicator of estrogenic activity in the water column. 

 

Major findings

The analysis of existing contaminant data in the watershed found:

  • For herbicides atrazine and metolachlor, that have had 30 years of regular sampling across the Chesapeake watershed, there were significant declining regional temporal trends in river concentrations of approximately 4% per year. 

  • Temporal trends for total estrogenicity and total PCBs were also negative, but not statistically significant.

 

Findings from the power analysis revealed (see figure 2)
  • Monitoring programs aimed at detecting small annual declines (< 5 – 7 % declines per year) are underpowered and unlikely to detect these small rates of decreasing contaminant concentrations, unless sampling has occurred at roughly 100 or more sites for at least 20 years. 
  • Monitoring for short-time periods (e.g., 5 years) is inadequate for detecting regional temporal trends, regardless of the number of sites sampled or the magnitude of the annual declines. 
  • Sampling sites across the watershed once a year did not increase the ability to detect regional trends for any monitoring scenario compared to sampling sites every other year. This suggests that sampling all sites every year is not necessary to detect trends.
  • In addition to these general patterns that emerged from the power analysis, there was variability among chemicals, suggesting that management expectations for detecting changes in concentrations will vary depending on the chemical of interest, and that some chemicals may be better ecological indicators than others. 
  • Overall, the ability to detect temporal trends was greatest for total estrogenicity, suggesting that this aggregate measure of estrogenic activity may be a useful indicator.
Power curves for detecting regional temporal trends in total estrogenicity, atrazine, metolachlor, and sediment total polychlorinated biphenyl (PCB) concentrations in rivers in the Chesapeake Bay Watershed
Sources/Usage: Public Domain. View Media Details
Figure 2. Power curves for detecting regional temporal trends in total estrogenicity, atrazine, metolachlor, and sediment total polychlorinated biphenyl (PCB) concentrations in rivers in the Chesapeake Bay Watershed, USA. Power is a function of the annual percent decline (-5, -7, -10, -15, -20%/year) and the number of sites sampled each year (10, 20, 30, 50, 80, 100 sites). The horizontal dashed line at a power of 0.8 was added to aid interpretation. (from Wagner and others, 2021)

Management implications

Several groups of toxic contaminants, including PCBs and some pesticides, are widespread across the watershed (Chesapeake Bay Program, 2012).  Ecological monitoring is critical for evaluating the success of management efforts to reduce toxic contaminants within the Chesapeake Bay Watershed and implications from the paper include:

  • The ability to detect temporal changes (i.e., statistical power) in these chemical concentrations is important to help inform the design of monitoring programs, and to develop expectations regarding the amount of time needed to detect temporal changes after BMP implementation.   
  • There was variability among chemicals, suggesting that management expectations for detecting changes in concentrations will vary depending on the chemical of interest, and that some chemicals may be better ecological indicators than others. 
  • Expectations should be set to detect regional trends over decadal time periods for determining the response of contaminant concentrations to BMPs. For example, the results suggest monitoring will need to be conducted for 20 years to detect the response of some toxic contaminants in rivers to BMP implementation.

 

For more information

Wagner, T., McLaughlin, P., Smalling, K., Breitmeyer, S., Gordon, S., and Noe, G. 2022. The statistical power to detect regional temporal trends in riverine contaminants in the Chesapeake Bay Watershed, USA: Science of the Total Environment, vol. 812, article 152435, https://doi.org/10.1016/j.scitotenv.2021.152435.

USGS Pennsylvania Cooperative Fish and Wildlife Research Unit

 

Posted January 20, 2021

Issue: Chemical contamination of riverine ecosystems is a global concern, with potentially negative effects for human and ecological health. Land management activities (e.g., Best Management Practices; BMPs) are an important tool that can be used to reduce point and non-point sources of pollution. The Chesapeake Bay Program (CBP) has goals for (1) reducing toxic contaminants and (2) reducing nutrients and sediment to improve water-quality conditions in the Bay and across the Chesapeake Bay Watershed. The BMPs to reduce nitrogen, phosphorus, and sediment also have a potential co-benefit of reducing toxic contaminants. However, the ability to confidently make predictions about the effects of land-management activities on reducing in-stream chemical concentrations is poorly understood. The CBP stakeholders need improved approaches to detect temporal changes (i.e., statistical power) in riverine chemical concentrations to help inform the design of monitoring programs; and to develop expectations regarding the amount of time needed to detect water-quality changes after BMP implementation (see figure 1).

Graphical abstract showing watershed sampling sites and BMPs
Sources/Usage: Public Domain. View Media Details
Figure 1. Graphical abstract. (from Wagner and others, 2021)

USGS study

The U.S. Geological Survey (USGS) conducted a study to provide new insights into the number of monitoring stations and amount of time needed to detect changes in river contaminants. Using data from USGS National Water Quality Network, the National Water Quality Monitoring Council Water Quality Portal, and the Pennsylvania Department of Environmental Protection, USGS scientists had two objectives: 

  • Quantify existing regional (watershed -wide) temporal trends in concentrations of select contaminants over the past several decades and estimate components of spatio-temporal variation, and 

  • Using estimated parameters from objective 1, perform simulations to determine the statistical power needed to detect regional temporal trends under different monitoring scenarios. The scenarios included varying the magnitude of the annual decline, the number of river sites sampled in the Chesapeake Bay Watershed each year, the number of years sampled, and annual sampling frequency. The selected contaminants included water concentrations of atrazine, metolachlor (herbicides), sediment concentrations of total polychlorinated biphenyls (PCBs), and total estrogenicity, which is an indicator of estrogenic activity in the water column. 

 

Major findings

The analysis of existing contaminant data in the watershed found:

  • For herbicides atrazine and metolachlor, that have had 30 years of regular sampling across the Chesapeake watershed, there were significant declining regional temporal trends in river concentrations of approximately 4% per year. 

  • Temporal trends for total estrogenicity and total PCBs were also negative, but not statistically significant.

 

Findings from the power analysis revealed (see figure 2)
  • Monitoring programs aimed at detecting small annual declines (< 5 – 7 % declines per year) are underpowered and unlikely to detect these small rates of decreasing contaminant concentrations, unless sampling has occurred at roughly 100 or more sites for at least 20 years. 
  • Monitoring for short-time periods (e.g., 5 years) is inadequate for detecting regional temporal trends, regardless of the number of sites sampled or the magnitude of the annual declines. 
  • Sampling sites across the watershed once a year did not increase the ability to detect regional trends for any monitoring scenario compared to sampling sites every other year. This suggests that sampling all sites every year is not necessary to detect trends.
  • In addition to these general patterns that emerged from the power analysis, there was variability among chemicals, suggesting that management expectations for detecting changes in concentrations will vary depending on the chemical of interest, and that some chemicals may be better ecological indicators than others. 
  • Overall, the ability to detect temporal trends was greatest for total estrogenicity, suggesting that this aggregate measure of estrogenic activity may be a useful indicator.
Power curves for detecting regional temporal trends in total estrogenicity, atrazine, metolachlor, and sediment total polychlorinated biphenyl (PCB) concentrations in rivers in the Chesapeake Bay Watershed
Sources/Usage: Public Domain. View Media Details
Figure 2. Power curves for detecting regional temporal trends in total estrogenicity, atrazine, metolachlor, and sediment total polychlorinated biphenyl (PCB) concentrations in rivers in the Chesapeake Bay Watershed, USA. Power is a function of the annual percent decline (-5, -7, -10, -15, -20%/year) and the number of sites sampled each year (10, 20, 30, 50, 80, 100 sites). The horizontal dashed line at a power of 0.8 was added to aid interpretation. (from Wagner and others, 2021)

Management implications

Several groups of toxic contaminants, including PCBs and some pesticides, are widespread across the watershed (Chesapeake Bay Program, 2012).  Ecological monitoring is critical for evaluating the success of management efforts to reduce toxic contaminants within the Chesapeake Bay Watershed and implications from the paper include:

  • The ability to detect temporal changes (i.e., statistical power) in these chemical concentrations is important to help inform the design of monitoring programs, and to develop expectations regarding the amount of time needed to detect temporal changes after BMP implementation.   
  • There was variability among chemicals, suggesting that management expectations for detecting changes in concentrations will vary depending on the chemical of interest, and that some chemicals may be better ecological indicators than others. 
  • Expectations should be set to detect regional trends over decadal time periods for determining the response of contaminant concentrations to BMPs. For example, the results suggest monitoring will need to be conducted for 20 years to detect the response of some toxic contaminants in rivers to BMP implementation.

 

For more information

Wagner, T., McLaughlin, P., Smalling, K., Breitmeyer, S., Gordon, S., and Noe, G. 2022. The statistical power to detect regional temporal trends in riverine contaminants in the Chesapeake Bay Watershed, USA: Science of the Total Environment, vol. 812, article 152435, https://doi.org/10.1016/j.scitotenv.2021.152435.

USGS Pennsylvania Cooperative Fish and Wildlife Research Unit

 

Posted January 20, 2021