By Erica Goldman

Uncertainty — the disparity between what is known and what actually is or will be — will inevitably color the high profile decision to introduce or not introduce a non-native oyster to the Bay. Scientists can predict the abundance and distribution of oyster populations under different environmental conditions. They can model the potential for oysters to improve water quality in the Bay and evaluate the risk that a new disease or habitat change might cause to the ecosystem. Economists can predict the potential benefit of a restored oyster population for the fishery. Anthropologists can assess the social dimension of an introduction. But in the end, the Chesapeake Bay cannot simply fast-forward to 2015 to reveal what will happen under each proposed restoration scenario.

We live in an uncertain world.

"The tools we have to attach certainty to our understanding of complex systems are still evolving," explains Ann Kinzig, a biologist at Arizona State University in Phoenix who has worked extensively in the national policy arena, including a recent fellowship in the Office of Science and Technology Policy in the Office of the President. Many experiments that we undertake with ecosystems, such as emitting gases into the atmosphere, are a one shot deal, she says. Many of the statistical tools used by repeatable manipulative experiments simply do not apply.

The introduction of a non-native oyster would be a clear case of Kinzig's "one-time experiment." Once reproducing populations of the non-native oyster enter the Bay, the decision becomes irreversible, with consequences that could extend far beyond the Chesapeake region. So when it comes to the great oyster controversy, how should policymakers approach scientific uncertainty and what tools do they have at their disposal?

Uncertainty and the Oyster

To make the final decision on the oyster Environmental Impact Statement (EIS), policy makers must weigh multiple sources of uncertainty. An "uncertainty analysis" of predictions from the oyster population model forms one key part of that total evaluation, explains Jon Vølstad, from the consulting company Versar.

Scientists turn to statistical methods to quantify uncertainty in model predictions. Vølstad and Mary Christman, with input from collaborators Jodi Dew at Versar and Danny Lewis at the University of Maryland, will run the juvenile/adult demographic model thousands and thousands of times. This repetition allows them to evaluate the effect of natural variation in the system — the fact that not every oyster grows at the same rate, for example, or the fact that oysters might experience higher or lower disease-related mortality as salinity changes in wet and dry years.

Modelers can also assess the effect of uncertainty in their choice of parameters. Since data for the non-native oyster rely predominantly on lab-based studies — the species does not live in the Bay — estimates for parameters like growth rate will carry a higher degree of uncertainty than for the native oyster. To ensure that they have the best possible information to plug into the model, the researchers work collaboratively with different advisory groups, including a special "growth rate advisory committee," explains Christman.

To address sources of uncertainty in the model, Christman, a statistician at the University of Florida, Gainesville will also conduct what is known as a "sensitivity analysis." This will help deal with environmental situations that may factor significantly in the model's predictions, but occur intermittently and remain hard to predict. For example, if she finds oysters in the model sensitive to short-term patches of freshwater, Christman will ask other scientists to determine the probability that a patch of freshwater (freshet) will occur in a given area. She can incorporate this probability into the model.

"From my perspective," says Christman, "If you tell me you are uncertain, I can run the model under different conditions. But understanding that uncertainty is one thing, interpreting what to do with it is another."

The interpretation of uncertainty will occur through a formal risk assessment process, explains Vølstad. The risk assessment will provide synthesis of the total body of knowledge available and will encompass the results of all of the different components of the Environmental Impact Statement (EIS) — including modeling efforts, a literature review, and results of the ecological, economic and cultural assessments (see Assessing a Potential Introduction). Scientists and managers will evaluate the quality of that information and associated risks, and make recommendations for action.

To sort and evaluate various streams of information from different sources, the Maryland Department of Natural Resources has developed a matrix with the different parts of the Environmental Impact Assessment spelled out — a decision-making worksheet of sorts. This worksheet concisely distills years' worth of research and analysis by scientists, economists, and anthropologists into a set of "decision factors."

To make this worksheet useful to decision makers, an Ecological Risk Assessment Advisory Team developed a set of objective criteria to evaluate the risk and uncertainty associated with each entry. These criteria assign each entry in the matrix with an estimated level of risk: high, medium, or low, and an uncertainty code: very certain (as certain as we are going to get); reasonably certain; moderately certain (more certain than not), reasonably uncertain, and very uncertain (a guess).

The Team will apply risk and uncertainty codes to each decision factor in the matrix for each scenario in the Environmental Impact Statement. This approach to risk assessment emulates the U.S. Geological Survey's protocol, developed when Maryland faced the first unintentional introduction of the northern snakehead fish in 2002, explains Vølstad, who works closely with the Team.

The decision matrix provides a scheme to quantify scientists' confidence in the body of knowledge on the non-native oyster in a manner that policy makers can easily interpret. But when the time comes for the final decision on whether to introduce the non-native oyster to the Chesapeake, data and decision matrices will only go so far. Different stakeholders will have different perspectives on how much risk they can tolerate. Societal values will play a key part in the final decision.

Oyster Advisory Panel chair Brian Rothschild, from the University of Massachusetts, Dartmouth, sketches the following scene: Picture a hungry man standing on a street corner. On the opposite corner, a restaurant beckons but cars zoom through the intersection. If the man could be described as normal with respect to risk tolerance, he would look both ways, cross the street, and go to the restaurant. A risk-prone man would dash into the street without looking, while a risk-averse man would never cross the street and never make it to the restaurant. Part of the challenge with the oyster decision, says Rothschild, stems from the fact that we have each of these three types of street-crossers in the Bay.

At Scientific and Political Crossroads

Finding common ground between the spheres of science and policy when it comes to interpreting risk and uncertainty presents no small challenge. Uncertainty often complicates policy decisions related to the environment, especially when the stakes are high, according to Daniel Sarewitz, Director of the Consortium for Science, Policy and Outcomes, a think tank at Arizona State University. Scientific research can help reduce uncertainty to an extent, he argues in a 2004 paper published in the journal Environmental Science & Policy, but it will never eliminate it. And at the end of the day, policy decisions related to ecological problems — such as whether to introduce the non-native oyster to the Chesapeake Bay — must be made despite scientific uncertainty.

Reconciling scientific uncertainty with the political process requires balancing the fundamentally different goals of science and policy, based on significantly different standards of evidence, asserts Kinzig and her colleagues in a paper entitled "Coping with Uncertainty: A Call for a New Science-Policy Forum." Published in the journal Ambio, the article resulted from a meeting of ecologists and economists sponsored by the Royal Swedish Society in 2002. "Science doesn't tell you what you should do under a given scenario," Kinzig says.

Scientific studies must reach either a 95 percent or often a 99 percent statistical level of confidence to be considered conclusive, she explains. In contrast, the standard of evidence for many political decisions can vary, becoming more or less stringent depending on whether the perceived cost of being wrong is low or high. If a physician is certain that a patient is going to die shortly, for example, there is little hazard in prescribing a drug whose efficacy is largely unknown, but that could offer some hope of life extension, Kinzig's paper argues.

Kinzig and her colleagues identify four factors related to the difference in "evidentiary standards" between science and policy that can introduce difficulties to environmental decision making:

• A failure to communicate about the nature of the difference in standards between science and policy may cause fundamental misunderstandings.
• The need for a scientific conclusion to reach 95 percent confidence can slow the introduction of important information to policymakers, especially in studies that involve complex systems.
• The probabilities associated with future environmental scenarios can be too intractable for scientists to quantify.
• Scientific information cannot answer a value-based question about how to act, only help illuminate future outcomes and potential trade-offs.

So with all of these differences in how the scientific and political realms deal with uncertainty, do any unifying themes emerge to guide decision makers in their decision on non-native oysters in the Chesapeake Bay?

In a crowded room at the headquarters of Maryland Department of Natural Resources in Annapolis, resource economist Doug Lipton offered one answer. Lipton, an associate professor at the University of Maryland, College Park and program leader of the Maryland Sea Grant Extension Program, spoke at the third in the series of public outreach meetings for the oyster Environmental Impact Statement. He concluded his presentation on economic projections for oyster restoration with a slide that read: "Decision making under large uncertainty calls for a precautionary approach. But what is precautionary is in the eye of the beholder."

Risk may mean something different to different stakeholders, Lipton explains. Faced with near economic extinction, the oyster industry may perceive not introducing the non-native oyster as the riskier option. For other stakeholders, potential risks associated with introducing a new species to the Bay, such as the possibility of introducing a new disease, habitat destruction, or extinction of the native oyster, may far outweigh the risk of doing nothing.

Whether action or inaction would constitute a precautionary approach depends on the future outcome desired — again often a question of values and societal preferences. Uncertainty does not make a possible outcome less harmful, says Kinzig, nor is it an excuse for inaction. This is especially true in cases with clear global impact, such as climate change, she says. On the other hand, "when we first exploded the atom bomb, we didn't know that it would not ignite the atmosphere," Kinzig says. "In this case it might have been good to wait."

This page was last modified December 20, 2005

h/V04N3/side3