Step 2: List Candidate Causes

This image is a drawing of a caddisfly larva in its case. Caddisflies are aquatic insects that are used by biologists to monitor the environmental quality of streams.

Supplemental Pages

The list of candidate causes is developed by starting with an initial list, and then refining it by making a map; drawing on available information on sources, stressors, and exposures; constructing a conceptual model; and engaging stakeholders. Continue to assemble relevant references throughout the process (example reference worksheet).

Make an Initial List

To begin the process of identifying and listing candidate causes, it helps to make a relatively long list of all stressors thought to occur in a waterbody. Include stressors that stakeholders have good reason to believe may be important. Consult other ecologists for potential causes of the impairment. It may be helpful to review available lists, such as:

A list of stressors that can potentially cause biological impairment,
Lists of stressors from monitoring studies, and
Standard lists developed by the state or other entity (standard list of candidate causes used in Mississippi).

Standard lists can provide a useful and consistent starting point, but you should keep in mind that no list is complete. Some causes are unusual and therefore not included on standard lists. In addition, the listed causes are likely to be generic (e.g., metals or invasive non-native species) whereas candidate causes should be specific (e.g., zinc or zebra mussels). The initial list is refined by moving through the process of drawing a map, gathering information, constructing a conceptual model, and engaging stakeholders.

The Common Candidate Causes section of CADDIS provides some suggestions on what to look for when deciding whether to include a cause on your list. Information is currently available for:

Multiple stressors may act together to cause effects or may reflect different steps on the same causal pathway. Consider whether you should include specific combinations on your list or can combine some to simplify analysis in later steps.

Gather and Map Information on Potential Sources

Information about point and nonpoint sources near the waterbody can suggest potential candidate causes. Point sources, such as drainage pipes, outfalls, and ditches, are easily identified as sources. Constituents of the effluent or the effluent as a whole can be listed as candidate causes. Other sources may be located some distance from the resource; for example, motor vehicles and smoke stacks generate candidate causes such as acid rain or nitrogen enrichment. Particular land uses often generate a consistent suite of stressors. For example, siltation, nutrients and pesticides are commonly associated with agriculture.

Locations of sources and stressors should be mapped. Maps or other geographical representations that show the location and severity of impairments are essential for orienting participants, examining spatial relationships, and eliciting information from stakeholders. Maps can range from simple hand-drawn ones to computer-generated versions. Useful geographic information includes the location of the impairment and known point and nonpoint sources, cities, roads, dams, tributaries, and land uses (example map of sources).

You may want to examine your readily available data using exploratory analytical methods. Some of these include scatter plots, correlation analysis, box plots, and conditional probability analysis. Remember you are only exploring the data to see if it suggests any candidate causes. You are not yet undertaking the full fledged analysis that you will do in Steps 3 and 4.

Construct a Conceptual Model

Simplified conceptual model for sediment.

Click to Enlarge Image.

Figure 2-1. A conceptual model for increased water temperature as a candidate causal agent of biological impairment.

Conceptual models link the candidate causes with potential sources and effects. This part of the process documents a likely explanation of how the candidate cause could have produced the impairment. Using a pictorial conceptual model is useful to introduce the ecological relationships. Then a box and arrow diagram can be used to show details of the relationships among sources, stressors, receptors, and intermediate processes. The diagram should show the pathways and causes considered in the study. Separate diagrams for each stressor or pathway can keep the focus on the analysis steps that will follow. The figure to the right is an example of a box-and-arrow conceptual model that was developed for a case study in coastal Maine (another example conceptual model). For additional examples and advice on conceptual model development, consult Jorgensen (1994), Suter (1999), Cormier et al. (2000c), USEPA (1998a) (especially Appendix C), and the Conceptual Model Library.

Engage Stakeholders

Involving a broad range of stakeholders such as managers, local citizens, and scientists from other disciplines can be crucial for the success of a causal investigation. Involving the managers and other decision-makers ensures that the investigation will be relevant to decisions and actions. Involving a broad range of scientists ensures that important causal pathways are considered. Involving local citizens helps to ensure that the investigation is inclusive of local knowledge and concerns.

Fruitfully engaging stakeholders in not a trivial task. Advice on engaging stakeholders can be found in Watershed Academy's Getting In Step Guide. A fairly comprehensive list of Public Involvement Techniques can be found on the EPA sponsored website SMARTe .

The draft map and conceptual model can provide a useful focus for stakeholder discussions. Questions may include the following:

What have we missed? Is there additional information that should be considered in the investigation?
Are there potential causes that should be added to the list for us to analyze?

Finalize the List

After a list of candidate causes is developed, it should be winnowed to the list of candidate causes that will be evaluated further (example list of candidate causes).

Listing candidate causes requires balancing two issues. If you include every stressor, the causal analysis will be burdensome, but being too selective in this step may eliminate the true cause.

There are two logical strategies that you can use to finalize the list of candidate causes:

Use the initial list as a check list by selecting candidate causes from it, or
Use the initial list as a scratch list by eliminating or deferring listed potential causes from further consideration.

We recommend that you document the rationale for omitting potential candidate causes from further consideration. The list of omitted causes may best be thought of as a list of deferred or second tier candidate causes. That is, they may be revisited if a probable cause is not identified and an iteration of the process is needed.

Generalizations about when to exclude or defer a cause from further consideration is difficult, but it is usually based on historical knowledge of the water body and its watershed, or knowledge of biology and mechanism. Four general rationales are used, listed below in rough order of decreasing level of confidence for either excluding or deferring analysis of a potential cause.

Evidence that the cause is absent based on high quality stressor measurements — In most cases, this evaluation is equivalent to analysis and evaluation of spatial and temporal co-occurrence. That is, although the cause could occur, data indicate that it has not occurred at the site. For example, if continuous data loggers indicate that DO concentrations at the impaired location are not lower than those in a reference location during the preceding year, low DO may not be listed as a candidate cause. The degree of quality needed to defend omission from the analysis will vary with the stressor and the level of confidence desired. In general, long term (e.g., a year or more) temporally continuous and spatially extensive data is needed for a high level of confidence in omitting a potential cause. Since the data are available and must be analyzed in order to omit the potential cause, we suggest that you include these less likely candidate causes in the strength-of-evidence analysis (Step 3). By including rather than omitting, these candidate causes, they can be compared to the other candidate causes; the assessment is more complete and transparent; and there is less likelihood of overlooking a cause.

Evidence of an impossible cause or mechanism — Indisputable evidence that a cause would not occur at the site (e.g. low dissolved oxygen in a cataract or very turbulent stream) or that the biological effect is never caused by the agent (e.g., over-harvesting causing liver cancer in fish). This type of information is sufficient to omit a potential cause from the list of candidate causes. When available, evidence that demonstrates a candidate cause is impossible is discussed in the Common Candidate Causes section.

No evidence that the cause is present — A lack of the observations that usually accompany a biological effect due to a particular cause. We provide check lists of factors that suggest that a potential cause be listed (see Common Candidate Causes). If these sources, site evidence or biological effects are absent, then the cause is either unlikely or you are lacking information. For example, if the watershed of an impaired stream is undisturbed forest, if sediment accumulation is not observed at the impaired site and if known sediment-sensitive species are present, then sediment may be omitted from the list of candidate causes. Using “absence of evidence” as a reason to omit a potential cause from the analysis is risky. Use it with great caution, because it could be that the relevant evidence is present but just has not yet been observed. For example, you may have observations from the autumn, but the exposures may occur in the winter.

Insufficient data and evidence — No data are available or available data are untrustworthy for evaluating the potential cause. In these situations, you may decide to defer analysis until data become available or until analysis is not needed because a probable cause was identified, remediated, and the biological condition improved. By deferring a potential candidate cause rather than excluding it, you demonstrate scientific awareness and lack of bias.

Additional tips for listing candidate causes are provided that may help you during this process.

Previous: Overview | Next: Results and Next Steps | Continue to Step 3 | Step-by-Step Guide Home Page