CADDIS Volume 1: Stressor Identification
The Step-by-Step Guide Introduction
On This Page
- Before the causal analysis begins
- Decision-maker and stakeholder involvement
- As necessary: Acquire data and iterate process
- Step 1: Define the case
- Step 2: List the candidate causes
- Step 3: Evaluate data from the case
- Step 4: Evaluate data from elsewhere
- Step 5: Identify probable cause
- After the causal analysis is done
The Stressor Identification (SI) process, shown in the yellow box in the center of the Figure 1, follows five steps that conclude with the identification of a probable cause. The gray boxes around the Stressor Identification process show various interactions and the context for the analysis. You will see this figure throughout the Step-by-Step Guide, with different boxes highlighted in black to indicate where you are in the process. Each of the elements shown in Figure 1 are briefly reviewed below.
Before the causal analysis begins
Detect or suspect biological impairment
What is the impetus for a causal analysis? Usually, something or some observation triggers the need. It may be curiosity, pure and simple. But, more often the impetus is a problem that can't be fixed until the cause is identified. Many U.S. EPA water management programs may benefit or even require the identification of a cause of a biological impairment. One example is the requirement of the Clean Water Act to identify and remediate impaired bodies of water. The cause must be known in order to develop an effective management plan. The requirement for the U.S. EPA to identify, report and develop plans to improve impaired bodies of water appears in the 305b and 303d clauses of the Clean Water Act.
What are the characteristics of an impaired waterbody? This website assumes that you already know that your stream or river is biologically impaired. For more information on making this determination, refer to U.S. EPA guidance on evaluating the biological condition of bodies of water (Biocriteria and Biological Assessment). Types of biological impairments that may warrant a causal analysis include fish kills, algal blooms, loss of biological diversity or unusual composition of biological assemblages.
Causal analysis requires some data and knowledge about ecology. Ecology is an inclusive science that integrates understanding of biological, physical and chemical processes in complex, interactive landscapes. Individual scientists can independently perform causal analyses, but a team often does it better. Effective causal analysis teams include members with expertise in a wide variety of environmental fields, and strong analytical, social, and communication skills. Social skills are particularly necessary when communicating the needs of the team to stakeholders and decision-makers, and when enlisting their support after the cause has been identified and it is time to correct the problem.
Stakeholders and decision makers can be viewed as part of the extended team engaged in the investigation to find the cause of a biological impairment. They can be sources of information. They can suggest possible causes of impairment and often know about historical impacts that may no longer be present or which have become hidden by a changing landscape.
Involve your stakeholders and decision-makers often. When the cause has been identified and it's time to take action, you will need community support to make environmental improvement a reality.
Causal analysis requires data, which may come from any reliable source. Data are used to show causal relationships between candidate causes and specific effects. New data may become available at any step of the process or you may decide to obtain more information to fill identified data gaps. These are normal iterations of the Stressor Identification process and can occur at any time. In fact, a screening level causal analysis is very useful for identifying the types of measurements and locations for sampling that will clinch a case.
Defining the case involves gathering information that sets the stage for the causal analysis. Products of this step include:
- A description of the reason for the causal analysis, usually a description of the general biological impairment and any regulatory violation,
- Descriptions of the methods and data used to determine the biological impairment,
- Maps and descriptions of the land use, locations of sampling sites, and possible sources of pollutants,
- Descriptions of the specific biological impairments and criteria that were used to determine them.
Causal analysis identifies the most probable cause of an impairment from a larger set of candidate causes. The list of candidate causes arises from several \activities and sources including the information gathered in Step 1. There are several ways to develop a list of candidate causes. Additionally, you may choose to brainstorm a list with colleagues or stakeholders. If there are stakeholders that staunchly believe that a biological impairment is due to a particular cause, include that cause no matter how unlikely it may seem. They need to be shown the evidence for or against that candidate cause.
Here are several other suggestions:
- Consult lists of candidate causes that are commonly encountered.
- List sources and the stressors generated by the sources or land uses.
- Search the literature for candidate causes associated with the specific effects that have been observed.
- Construct or consult conceptual models that show sources, processes, and candidate causes that are known to give rise to specific effects. For example, Figure 2 is a very simple example of a conceptual model linking an effect (fish declines) to a candidate cause (low dissolved oxygen), and back to two ecological processes (nutrients and channelization) that might give rise to the candidate cause (and, by extension, the observed effect).
The output of Step 2 includes:
- A list of candidate causes,
- A conceptual model and description of sources, processes and candidate causes that are known to give rise to the specific effects observed in the waterbody being investigated.
The most powerful evidence is based on cause-effect relationships developed from data obtained from the case itself. This evidence is developed in Step 3. Step 3 has two goals. The first is to use data from the case to develop evidence that allows you to confidently eliminate very improbable causes or use symptoms to refute or diagnose a cause.
The second goal of Step 3 is to begin building the body of evidence for those candidate causes that cannot be refuted or diagnosed, which will be used in Step 5 to identify the most probable causes. The more types of evidence and the more characteristics that support a candidate cause the more likely that it IS the true cause. The more types of evidence that weaken the case for a candidate cause the more likely that it IS NOT the true cause. Logical and statistical approaches that are used to develop evidence are described in more detail in Fundamentals of Causal Analysis and in Step 3 of the website.
The products of Step 3 describe the types of evidence used, the conclusions reached for each type of evidence, and the supporting documentation. The conclusions reached for each type of evidence are supported by:
- Describing the data,
- Describing the analysis,
- Scoring the evidence,
- Discussing the scores, including an evaluation of the quantity and quality of the data, and the reasons for applying the scores
- Documenting the candidate causes refuted or confirmed by diagnosis.
Refuted causes require no further analysis, but are revisited when summarizing the overall conclusions. Confirmed diagnoses by definition are probable causes and require no further analysis, but are also revisited when describing probable causes in the final summary. All other candidate causes are further evaluated in Step 4.
Criteria for eliminating a candidate cause: If data from the cases shows that any characteristic of a causal relationship is very improbable, you have a very strong case for eliminating a cause. For example, in Figure 3, the red "x"’s on the arrows are drawn because riffles were observed at the site, algal growth was minimal, and continuously monitored oxygen levels were always near saturation. Therefore the case for low dissolved oxygen can be confidently refuted.
Criteria for diagnosis: Diagnosis is possible when the effect is described by certain symptoms that are always present and those symptoms only have one cause. When the symptoms are absent, the candidate cause is refuted and eliminated from further consideration. When the symptoms are present, the candidate cause is diagnosed and is confirmed as the probable cause.
It sounds deceptively easy, but is actually rather difficult to do. It becomes more difficult because many uncertainties usually arise. For instance, measurements of the impairment and the agent may have been made on different days. The more detailed guidance suggests what to do in some of these situations, but the point here is that uncertainty is very common. What is reasonable?
- Use the data you have. It may not give you a confident answer, but it usually tells you something. At a minimum, you can use it to determine what additional information you would like to collect.
- Document the assumptions and choices you make. For example, you may have decided to use maximum values to represent a worst-case scenario. Or you may have chosen to use average values because the impairment is chronic rather than acute or episodic in nature. In either case, document the decisions!
Most candidate causes cannot be diagnosed or eliminated. Further evaluation is required that uses knowledge gained from laboratory studies and from past experiences and observations in other waterbodies. For example, one of the most useful types of evidence from elsewhere uses the stressor-response relationships developed from laboratory studies. Familiar examples are single chemical, single species toxicity tests. The primary goal in Step 4 is to develop evidence that uses data from elsewhere. Logical and statistical approaches that are used to develop evidence are described in more detail in Fundamentals of Causal Analysis and in Step 4 of the website.
The products of Step 4 describe the types of evidence used, the conclusions reached for each type of evidence, and supporting documentation. The conclusions reached for each type of evidence are supported by:
- Describing the data,
- Describing the analysis,
- Scoring the evidence,
- Discussing the final scores, including an evaluation of the quantity and quality of the data, and any extrapolations needed to apply the evidence to the case at hand.
In Step 5, all of the evidence is considered to reach final conclusions about the probable cause. The products of Step 5 include:
- Scores for all the types of evidence used displayed in a table,
- Evaluation of the consistency and credibility of the case based on the scores,
- Identification of the most compelling lines of evidence,
- Determination of the candidate cause as probable, unlikely, refuted or diagnosed,
- Discussion of the reasons for that determination,
- A report documenting the causal assessment.
Scores for all types of evidence are displayed in a comprehensive table. The score for the evidence from the case and from elsewhere is influenced by the quantity and quality of the evidence. The magnitude of the score is based on the likelihood of observing the effect due to the true cause rather than from a chance outcome. Therefore, the highest scores are given to the types of evidence that:
- Use data from the case,
- Are based on more than one piece of evidence,
- Effectively link the causal agent with the effect.
All of the evidence is evaluated for consistency by looking at the overall pattern of scores. Confidence in the argument for or against a candidate cause is increased when it is supported by many types of evidence. If there are inconsistencies, now is the time to determine whether they are important or can be explained.
Now, the hard part! The overall body of evidence for each cause is weighed, the most compelling lines of evidence are reviewed, and a determination is made. The criteria for determining probable and unlikely causes are described in Step 5 of this guide. But be forewarned, there is no mathematical formula for making the overall determination. Statistical hypothesis testing is inappropriate. Instead, the determination is based on the scientific evidence that you have. Does the evidence show that the candidate cause and the effect exhibit more than one characteristic of causal relationships? Are the characteristics of causal relationships evident and supported by several types of evidence? Is the evidence compelling, consistent and credible?
Each of the candidate causes is evaluated this way. Then, the cases for all of the candidate causes are compared against each other to reach a final conclusion. It is not uncommon to identify more than one candidate cause, or to conclude that multiple causes are operating. The Stressor Identification process should help clarify the possibilities and provide insights into management options. For example, it may be possible to remediate more than one candidate cause at a time. Other typical situations in which the findings are not clear cut are described in Step 5 of the guide. Remember that you are not trying to prove or disprove a grand scientific theory. Rather, you are comparing the relative merits of the candidate causes in order to support decision-making.
We recommend that you document your causal analysis in a report. The report typically describes all steps in the Stressor Identification process including:
- The reason for the causal analysis,
- A list of the candidate causes and the ecological theory supporting them,
- The sources of the data used in the analysis,
- Tables of the evidence and scores derived from the data,
- Conceptual models of the causal pathways, and
- The key evidence used to distinguish the probable from the unlikely causes.
The more costly or controversial the remediation and the more skeptical the decision-maker, the more complete the documentation must be. The bottom line is a statement of your reasons for identifying the probable cause and for determining that other candidate causes are unlikely. Some people find summary tables and narratives helpful; others prefer to annotate the conceptual models with evidence. Above all else, use what works for you.
Identify and apportion sources, management action, biological condition restored or protected
Once the probable cause has been identified, the task of managing the factors responsible for the impairment problem still remains. If the process has been set up properly, there should be stakeholders and managers ready to receive the information. The probable cause may have more than one source or may require source identification. There are several U.S. EPA websites that contain information on best management practices (BMPs) for reducing causal agents in a variety of situations, including stormwater BMPs, agricultural BMPs, and forestry BMPs. Research underway at the U.S. EPA and in other federal, private and academic research programs will continue to provide insights for selecting and implementing effective management options. So, don't stop with just our suggestions.
If a management action occurs at your site, monitoring the reduction in the causal agent and changes in the biological impairment can demonstrate progress and confirm or refine the findings of the causal analysis. That is, changes stemming from the management action can allow you to determine whether the "real" cause was targeted for action or not. If biological improvement is marginal even with significant reductions in the causal agent, there may be other unidentified causes. If the stream improves dramatically, call a press conference and have a celebration with your stakeholders. Let us know, too—we like to celebrate as well!