Applied Ecosystem Services, LLC

The Environmental Issues Doctor

Photo of Nonpoint Source Pollution Regulation


Estimated reading time: 9 minutes

Nonpoint source water pollution adversely impacting a beneficial use is regulated by discharge allocations from a Total Maximum Daily Load. Whether a waterbody requires a TMDL depends on how its condition is assessed.

The Clean Water Act, as amended in 1987, specifies using environmental ambient conditions to assess waterbodies, but regulators focus on single chemical ions or compounds based on point source discharge regulations. The limitations of this approach and the benefits of a different approach are explained here.

Understanding the assessment process helps you to avoid, or resolve, concerns about your operations. Load allocations affect your permitted business operations if it discharges storm water over a diffuse area, not from pipe outfalls. For example, precipitation surface flows over crop lands, wood lots and forests, dairies, feedlots and other confined animal feeding operations (CAFOs), or mining operations with exposed soils.

Knowing how receiving waters are, and should be, assessed allows you to better understand the relationship of your operations to adjacent and nearby waterbodies.

Regulators often struggle with waterbody assessments and it’s common for the regulated public who are affected to be dissatisfied with the results.

You need to understand this process because assessments and resulting TMDLs focus on specific chemicals in the water rather than the CWA objective of restoring and maintaining the physical, chemical, and biological integrity of the nation’s waters.

The 1987 revision of the CWA directed the EPA to develop and implement regulatory programs for control of nonpoint sources of pollution based on ambient waterbody conditions. States and tribes must assess water quality status every two years to determine whether corrective TMDLs are needed.

A total maximum daily load, a TMDL, is the quantity of a chemical pollutant determined by the regulator to be the most that can be discharged into the waterbody ecosystem each day without adversely impacting a designated use.

The biannual water quality status assessments are the most important component of the process because everything else depends on their scientific robustness.

Assessments are specific to a designated use in a comparatively homogeneous area and are currently limited to a specific chemical pollutant rather than to ecosystem ambient conditions. More on this point later.

These issues make it difficult to determine a TMDL and effectively allocate discharge limits to point and nonpoint dischargers. Recommendations for increased effectiveness of the waterbody assessment process by incorporating appropriate environmental science and data analyses were developed in 2001 by the National Research Council. Not all states, nor the EPA, have implemented all their recommendations.

The focus of water quality laws was originally on human health, then gradually broadened to include aesthetic uses such as “fishable and swimmable.” The 1972 CWA amendment expanded the goal to “restore and maintain the chemical, physical, and biological integrity of the Nation’s waters.”

Despite the 1987 amendments’ re-focus on ambient conditions regulatory focus has remained on point sources, the discharge of pollutants into receiving waters at specific locations. However, there are pollutants entering waterbodies from diffuse sources and hydrologic alterations that create nonpoint source pollution. Applying point source effluent-based criteria to nonpoint source discharges is highly problematic and ineffective.

There are three major consequences from applying effluent-based, rather than ambient-based, standards to diffuse nonpoint source pollution discharges.

First, regulatory efforts to measure and communicate water quality status are described in terms of permit conditions rather than the receiving water condition. The two are not the same.

Second, effluent standards apply only to point sources. Pollutants from nonpoint sources such as agricultural, silvicultural, and construction activities largely escape oversight because they cannot be measured the same way.

Third, chemical pollutants came to dominate water quality policy while physical and biological determinants of the ambient condition were only infrequently considered.

In non-urban areas the condition of a waterbody depends on more than the loads of specified pollutants from sources discharging under a NPDES permit. For example, changes in the hydrologic regime associated with residential and industrial development activities can destabilize stream banks, increase loads of sediments and nutrients, eliminate key aquatic species, or otherwise change the aquatic ecosystem. Biological, hydrological, and physical changes to a waterbody are included in the CWA’s definition of pollution.

Compliance with CWA Section 303(d) requires states to identify waters not meeting ambient water quality standards in each assessment unit, define the pollutants and pollution factors and their sources, establish total maximum daily loads necessary to achieve those standards, and allocate daily loads to all sources within the assessment area. Degradation factors that do not fit the pollutant definition such as changes of habitat, flow alterations, channelization, and modification or loss of riparian habitat may be considered as reasons for not meeting standards and their effects on water chemistry must be considered in the TMDL development process.

The assessment process needs to be explained in terms of established biological and ecological theories as well as the appropriateness of statistical models applied to the available data.

The TMDL process depends on observations and measurements, not experiments. Therefore, it is greatly influenced by how water quality standards are determined. Currently, water quality standards are maximum concentrations of specific toxic metal ions or organic chemical compounds thought to be protective of fish, wildlife, livestock, and human uses.

For example, rather than the too-broad concept of fishable and swimmable a better description for a designated use of human contact recreation should protect humans from microbial pathogens while swimming, wading, or boating. A sufficiently detailed designated use might distinguish between beach use, primary water contact recreation, and secondary water contact recreation.

Other designated uses should be defined to protect humans, wildlife, and livestock from harmful substances in water and fish tissues. Specific aquatic life designated uses should be designed to protect fish and shellfish. It’s not only chemicals that need to be considered. Parasites and water temperature can be of greater importance to fish survival, growth, and reproduction than are the chemical constituent concentrations of the water.

While one set of chemical concentration criteria could include acceptable constituent measurements of inorganic and organic molecules, one constituent alone is insufficient as animals living in the water (as well as human fullbody contact) are affected by the entire composition of constituents rather than a single one, or a small subset.

Biological criteria should be used along with physical and chemical criteria to determine whether a assessment unit is meeting its designated use. The most weight should be placed on biological criteria because they are most closely related to the long term and fluctuating ambient conditions than are chemical measurements.

Measurements of pollutants in surface and ground waters, sediments, and soils are highly variable. Analyzing these observational data requires different statistical models than do experimental, financial, and public data. Statistical models in spreadsheets and basic stat courses taught in science and business programs cannot produce valid results because they were designed for non-environmental data. This means that the common frequentist hypothesis-testing models can be used in only very limited situations with environmental data.

With this background it’s time to look at how waterbodies are assessed for compliance with the CWA goal.

One state’s regulator says they use a statistical hypothesis testing approach, a binomial test, to derive a critical number of sample excursions that scales with the number of representative samples to evaluate beneficial use attainment status of waterbodies.

There is much wrong with this sentence.

The frequentist hypothesis testing approach was developed for use in analyzing biological experimental results, comparing the frequency of a desired result in a treatment group to that in a control group. It cannot apply to environmental chemical data because neither one nor a set of constituent concentrations can be assigned to a null or an alternative hypotheses. In addition, the hypothesis testing approach cannot be applied to diffuse pollution elements, such as habitat loss, channel alteration, or biotic community structure and function in the assessment unit.

The binomial test is not a frequentist hypothesis test and cannot provide correct water quality results.

“Binomial” literally means two numbers. A binomial distribution shows the number of successes (heads or tails) in a set of experiments such as flipping a coin. All binomial tests of data ask a simple question, such as success or failure, yes or no, true or false, present or absent.

Chemical concentrations are continuous values, not integer counts. Neither a concentration nor a set of concentrations can be characterized into yes/no, success/failure, or true/false dichotomies.

In addition to the inappropriate application of frequentist hypothesis tests and the binomial distribution the state uses other statistical models that have assumptions not met by water quality data.

Telling you how a regulatory process produces incorrect results is only half of my job. I need also to provide an alternative process because the assessment results can put waterbodies on the TMDL implementation list when they are not impaired based on complete ambient conditions.

Local biota reflect ambient water quality much better than do fluctuating chemical constituent concentrations because the biota are there as water chemistry varies within its natural range.

Fish are not found in every assessment unit, in all reaches of those they do inhabit, and they move around a lot so they’re not suitable indicators of ambient water quality. But, benthic macroinvertebrates, the animals living on the waterbody’s substrate, are found almost everywhere along the river network from the highest first-order streams to the mouth of the river, and in almost all ponds, lakes, and reservoirs. Benthic macroinvertebrates include the juvenile stages of insects, arachnids such as mites, snails, fresh water mussels, worms, and other small animals. Their community composition remains stable because their food resources are stable and characteristic of the reach’s position within the river continuum.

Benthic macroinvertebrates are easily associated one of five principal feeding strategies reflecting the primary type of food resource: shredders, filterers, gatherers, grazers, and predators.

Because the primary source of food varies along the river continuum the relative proportions of the five feeding strategies also varies. But, within a limited assessment area the relative proportions remain quite constant throughout the year.

Before determining whether a permitted activity has adverse impacts on a specific designated use it’s necessary to quantify inherent natural variability of functional feeding group compositions. By common acceptance, all values within 95% of that variability range is accepted as normal.

When an assessment based on ambient conditions indicates the need for a remedial TMDL a location-specific approach should be developed that is based on established ecological, hydrological, and geomorphic theories.

What I want you to remember are:

  1. Ambient-based water quality standards include pollution as well as chemical pollutants.

  2. For TMDL assessment units biota provide more technically sound and legally defensible decisions than do chemical data, especially effluent-based standards.

  3. The assessment methods used, and how load allocations are assigned, must be related to the CWA’s objective of ecosystem integrity.

This work was originally published on the Applied Ecosystem Services, LLC web site at

It is offered under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International license. In short, you may copy and redistribute the material in any medium or format as long as you credit Dr. Richard Shepard as the author. You may not use the material for commercial purposes, and you may not distribute modified versions.

Keep reading

  1. Photo of Endangered Species Act: Critical Habitats

    ESA: Critical Habitats

    Estimated reading time: 4 minutes

    Natural resource operators are directly affected by habitat preservation requirements for species listed under the ESA and state equivalents. One possible explanation is that environmental decision-makers do not have sufficient information, ecological training, or appropriate analytical tools so they fall back on the precautionary principle and declare that all actual and potential habitat for the species be left untouched for population sustainability. This is both unnecessary and wasteful as there are robust statistical and spatio-temporal models that can inform technically sound and legally defensible decisions, even with limited data.
  2. Photo of The Fate of Biological Data: Too Little Information

    The Fate of Biological Data


    Estimated reading time: 3 minutes

    It is widely accepted that raw data need to be converted to information (commonly by statistical analyses) and the results interpreted to form knowledge before informed decisions can be made. With biological data this process is not followed as frequently as it should. Modern spatial analyses and statistical models can provide valuable and useful information that is otherwise lost. Biological data are counts, presence/absence, proportions, and frequencies. They are not continuous variables with a true zero so the familiar parametric statistics cannot be used.

For more information contact me.