Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
10 A Scientific Framework For Environmental Problem-Solving To manage the effects of environmental manipulations, we must be able to predict them. However, knowledge is seldom sufficient to allow accurate prediction, so studies are necessary to provide the information needed to make decisions. Such studies must be carefully planned, because they are expensive in time, money, and effort. This chapter presents a general framework for identifying, scoping, and planning studies of en- vironmental problems. The framework is, in essence, an admonition to think before acting and to use established scientific principles. Table 1 makes it clear that deficiencies in environmental impact assessment are due not only to scientific difficulties the ones with which this chapter is primarily concerned but also to political, administrative, and eco- nomic difficulties. Despite their bewildering variety, environmental problems share some basic features, including actions that result in environmental changes, public and scientific concern about those changes, a need for methods for predicting environmental responses to human actions, and limited re- sources for the acquisition and analysis of relevant ecological information. We draw heavily on a number of recent efforts to make environmental assessment and management scientifically more credible (Andrews et al. 1977; Anonymous, 1980; Council on Environmental Quality, 1978; Fritz et al., 1980; Holling, 1978; Larkin, 1984; Rosenberg et al., 1981; Sanders et al., 1980; Sharma et al., 1976; States et al., 1978; Walters, in press; Ward, 1978) and in particular on a recent Canadian review (Beanlands and Duinker, 19831. 104
A SClE=IFIC FRAMEWORK FOR E^IRONME=^ PROBLEM-SOLVING 105 TABLE 1 Some Recent Criticisms of Ecological Impact Assessment, Based Primarily on Beanlands and Duinker (1983), Carpenter (1976), Rosenberg et al. (1981), and Skutch and Flowerdeu (1976) Guidelines too elaborate and requirements too diverse Time and money constraints not recognized Unreasonable expectations of decision-makers Tendency to start gathering baseline data immediately, at the expense of careful planning Failure to formulate objectives clearly and to develop a study strategy Unwarranted belief that ecological principles used in managed systems are as appropriate to unmanaged systems Failure to recognize the value of early input from those who might later be involved in re view, leading to an adversarial process Failure to define project boundaries Failure to consider cumulative effects Failure to state the bases of value judgments Lack of scientific standards for impact assessment Lack of respect in academe for impact assessment Vague and unverifiable predictions Lack of a rigorous, quantitative approach, especially in monitoring Lack of continuity in studies conducted during planning, developmental, and operational phases of a project Failure to follow actions with adequate monitoring studies Use of impact assessment for disclosure, rather than for learning Failure to recognize the scientific value of experimentation and monitoring Failure to consider the recovery potential of species and ecosystems Poorly written reports in which major points are buried in enormous amounts of information Inordinate expenditure of effort on descriptive studies with little potential for predictive value Inaccessibility of reports and results of studies, making them difficult to evaluate and learn from DEFINING ENVIRONMENTAL GOALS AND SCIENTIFIC QUESTIONS In spite of the difficulties and controversies associated with identifying environmental goals, a clear statement of goals early on can help to focus research and can increase the chance of protecting components of the environment likely to be identified as valuable to society. The first step in defining such goals is to identify the components of the environment perceived as valuable, such as salmon in rivers of the northwestern and northeastern United States, a "natural-looking" community of plants on reclaimed land (Chapter 18), clean water (Chapter 20), clean air, forest productivity (Chapter 19), and fishery productivity (Chapter 12~. The second step is to determine the desired degree of protection, ex- ploitation, or control. This decision usually involves choosing a state in which to maintain the ecological system in question and a length of time
106 KINDS OF ECOLOGICAL KNOWLEDGE AND THEIR APPLICATIONS for which to maintain it. For example, we might wish to reduce the population of a pest or the amount of damage it causes, increase the yield of a harvested species, or maintain the species composition of a valued habitat. The period of management might be months, decades, or cen- turies. The third step is to learn the environmental and financial costs of managing the system. Maintaining an ecological system in other than the "natural" condition usually requires some expenditure and might produce unwanted side effects. Harvesting a population for maximal yield can increase the variability of the yield and the likelihood of overharvesting (Chapter 11. Increasing production of an agricultural crop or forest often involves the use of hazardous pesticides (Chapter 24), which can have several deleterious cumulative side effects (Chap- ters 1, 3, and 41. Identifying environmental goals is complex and requires input from the public and from scientists. The public is concerned primarily with the choice of environmental goals. Scientists can help to identify non- obvious goals and can indicate the environmental and economic costs involved. Scientists are also needed to translate environmental goals into scientific objectives, which show what information is needed to answer the major questions and hence help in the planning of studies. As in any research plan, scientific objectives are based not only on the need for particular information, but also on how easily that information can be obtained. The issues on which environmental goals are based are specified in part by law and in part by public concern (see Table 24. The National Envi- ronmental Policy Act requires early public and professional input in iden- tifying those issues (Council on Environmental Quality, 1978~. The goal of protecting the Southern Indian Lake whitefish fishery was economically motivated (Chapter 211. Provincial wildlife biologists recognized caribou migration as a major public concern in the Newfoundland hydroelectric development case (Chapter 161. In the case of Lake Washington (Chapter 20), interested scientists and the public cooperated to define goals and develop an appropriate response; the DDT case (Chapter 24) shows how such interactions can lead to new understanding and to legislation. Attempts to achieve a goal sometimes have unexpected results. The New Brunswick forest case study shows how attempting to maximize forest timber production on the basis of individual stands might not only fail to maximize yield over the whole forest, but fail to provide consistency in yields over a long period. In fisheries, managing for maximal sustainable yield often produces large variations in both yields and stock abundance (May 1980), making overexploitation and population collapse more likely
A SCIENTIFIC FRAMEWORK FOR ENVIRONMENTAL PROBLEM-SOLVING 107 TABLE 2 Some Common Cntena for Identifying Important Issues and Valued Ecosystem Components in Impact Assessment Legal requirements Air and water quality standards Public health Rare, threatened, and endangered species Protected areas or habitats Aesthetic values Landscape appeal Attractive communities Appealing species (e,g., large ungulates, colorful birds, cacti) Species at higher trophic levels (e.g., eagles and tigers) Clear air and water . Economic concerns Species or habitats of recreational or commercial interest Ecosystem components Environmental values and concerns Ecosystem rarity or uniqueness Sensitivity of species or ecosystems to stress Ecosystem " naturalness " Genetic resources Ecosystem services Recovery potential of ecosystems "Keystone" species than more conservative management would (e.g., Murphy, 19771. Har- vesting or managing populations over long periods can also produce un- desired cumulative genetic changes (Chapter 11. SCOPING THE PROBLEM Scoping involves bringing together all interested parties public, busi- ness, government, and scientific so that they can interact and express their views before major actions or studies are initiated (Council on En- vironmental Quality, 19781. Early scoping can help to identify the im- portant issues and potential environmental effects associated with planned actions. It can help to define scientific objectives and guide the design of ecological studies. Scoping can also be useful during a program or project to ensure that the most important issues are being addressed, that studies are producing useful results, and that important new issues are noted (Fritz et al., 1980; Sanders et al., 19801. Once the valued ecosystem components, significant issues, and major potential effects have been identified, ecologists can establish scientific objectives. When prediction of environmental effects is a major purpose
108 KINDS OF ECOLOGICAL KNOWLEDGE AND THEIR APPLICATIONS of studies, four questions are usefully posed (Beanlands and Duinker, 19831: · Are valued ecosystem components expected to be affected either directly or indirectly by the project or action? · Can the valued ecosystem components be studied directly? (In the absence of adequate guidance from experience or the literature, pilot investigations might be needed to indicate the feasibility of particular studies.) · Is it possible to study valued ecosystem components indirectly? For example, because large carnivores are often difficult to study directly, the effects of an action on their prey base or their habitat could be studied instead. Studies of indirect effects are most appropriate when they are reliably associated with effects on the valued ecosystem components. · Would the use of indicators of impact be helpful? (See Chapter 7.) Formulating a conceptual model of the relationships between the pro- posed action and the receiving environment can help to identify pertinent questions and potential environmental effects. The purpose of such models is to identify the physical and biological pathways by which an action can produce ecological effects. By focusing on relationships important to the manifestation of effects, they help to develop specific, testable hypotheses to explain why particular effects should or should not occur. Conceptual models can also help to identify logical errors, to highlight factors that require special study, to synthesize ideas and knowledge, and to com- municate information (Beanlands and Duinker, 19831; guidance for de- veloping conceptual models can be found in Holling (1978), Ward (1978), Fritz et al. (1980), and Beanlands and Duinker (19831. Multidisciplinary workshops can be used to articulate a problem and plan studies (Holling, 1978) and have been used to advantage in this way (ESSA, 19821. Given adequate time and resources, sophisticated modeling should be considered (Munn, 1979~. Basic guidance in development and use of such models can be found in Frenkiel and Goodall (1978), Holling (1978), and Ward (1978~. Quantitative modeling can help by forcing assumptions to be made explicit, by making their consequences clear, and by revealing the sensitivity of outcomes to details of various assumptions. Simulation models are necessarily based on numerous unverified assumptions and cannot predict quantitative changes very accurately (Hilborn, 1979; Wal- ters, 19751. But they can be useful in identifying potential qualitative effects and exploring the consequences of alternative management plans. Sensitivity analysis allows an exploration of the consequences of altering
A SCIE=IFIC REWORK FOR EWIRONMEV~ PROBLEM-SOLVING 109 the assumptions of a model. Simulation models are often used in con- nection with freshwater systems (Chapter 21), in which the driving phys- ical and chemical processes are fairly well understood; fishery and wildlife management (Chapter 121; epidemiology (Chapter 151; and forest man- agement (Chapter 191. ESTABLISHING STUDY BOUNDARIES One of the first and most important tasks in the design of research on environmental effects is to establish a set of boundaries-temporal, spa- tial, and ecological. When might effects appear, and how long might they last? How long must studies last to allow reasonable predictions and reliable diagnosis of effects? Over what area will effects occur? Are there any natural barriers to the transmission of effects? Are any physical or biological processes likely to spread effects to other areas? What ecosystem components will be affected? At what levels of biological organization will effects appear? What species or ecosystem processes need to be studied and over what area? Boundaries in open systems, such as the ocean or atmosphere, are the most difficult to define. Variations in eco- system components of interest can strongly influence the time required for biological effects to appear (Holling, 1973~. Making judgments about boundaries is difficult, and many surprises have occurred. For example, large water impoundments can influence local climate or induce earthquakes (Baxter and Glaude, 19801. When DDT was first used as a pesticide, no one expected it to appear in animals in the ocean (Chapter 241. The DDT story and similar cases (e.g., that of acid rain) have shown that environmental effects can spread by subtle pathways. Assumptions implicit in management decisions might result in setting boundaries that omit critical processes. Attempts to increase stocks of anadromous fish by increasing reproduction in rivers might fail if survival in the ocean is already limited by food supply (Peterman, 19844. The cumulative effects of multiple actions have taught us that specific projects and actions must be viewed in the context of related actions (Odum, 1982~. Recent efforts to protect and conserve species have shown how management of a population requires consideration of its relationship with other populations (Franker and Soule, 1981; Schonewald-Cox, 1983; Soule and Wilcox, 19801. And only recently has it been recognized that systematic management procedures e.g., sex-biased or size-biased har- vesting-can lead to undesirable genetic changes over remarkably short periods (Chapter 11.
110 KINDS OF ECOLOGICAL KNOWLEDGE AND THEIR APPLICATIONS The establishment of boundaries is constrained by administrative, proj- ect-related, technical, and ecological factors (Beanlands and Duinker, 19831. Administrative constraints include jurisdictional limits, insufficient time or funding, and political factors. Spatial boundaries are often obvious, unless long-range transport phenomena are involved. Temporal boundaries might be less well defined, because of political and other uncertainties; as we go from short-term, local effects at the population level to long- term, regional effects at the ecosystem level, we are less able to predict them (Christensen et al., 19761. Other technical constraints are imposed by environmental variability, project location, and logistical problems. Setting appropriate temporal and spatial boundaries is important in the management of species populations, whether for protection, control, or harvest. When populations become small, patterns and rates of interchange of individuals and genes between populations become critical. The sizes of populations needed for the long-term maintenance of the spotted owl in the Pacific Northwest depend on whether the Columbia River is a dispersal barrier (Chapter 171. When timber is managed on a forest-wide basis, rather than by stands, yields are higher and more consistent (Chapter 191. Physical and chemical processes can be critical in defining boundaries in aquatic systems, particularly when the spread and accumulation of pollutants are involved. The control of eutrophication in Lake Washington depended on knowledge of flow rates through the lake and the low turnover of phosphate in lake sediments (Chapter 201. DEVELOPING AND TESTING HYPOTHESES Statements about relationships between proposed actions and ecosystem components or processes are, in effect, hypotheses that can be tested. Studies designed to test them can increase our ability to predict environ- mental effects. In addition, the explicit statement of hypotheses helps us to identify important assumptions and formulate specific objectives for ecological studies. However, despite the acknowledged value of testing hypotheses in solving environmental problems, many studies are not de- signed and conducted to do so. Many studies in wildlife management, for example, involve elaborate collection of field data with only after-the-fact attempts at explanation (Romesburg, 19811. What happened as a result of a project is rarely studied (Beanlands and Duinker, 1983; Larkin, 1984~. In practice, most general hypotheses are evaluated by testing specific predictions that arose from them. In environmental impact assessment, the predictions themselves are a major product of preproject research. It is often helpful to develop several hypotheses about possible effects and their causes, so that studies can be designed to distinguish among them.
A SCIENTIFIC FRAMEWORK FOR ENVIRONMENTAL PROBLEM-SOLVING 111 A hypothesis can be tested by studies before a project and by treating the project itself as a test. Methods for testing ecological hypotheses in preproject studies include the use of microcosms (Crow and Taub, 1979; Heath, 1979), field and laboratory experiments (Giddings, 1980; Suter, 1982; Ward, 1978), and computer simulations (Frenkiel and Goodall, 1978~. Despite the difficulty of assigning causality in field experiments (Sharp et al., 1979) and of extrapolating from small studies to large problems (Hilborn and Walters, 1981), pilot-scale perturbation studies could be the most productive research tool for impact assessment, although underused (Beanlands and Duinker, 1983; Ward, 19781. If projects are to be treated as large-scale experiments, baseline data must be collected before the project begins (Beanlands and Duinker, 1983; Hilborn and Walters, 1981; Larkin, 1984~. The baseline can best be viewed as a description of the mean values and natural variability in the system (Hirsch, 19801. Judgments of how much information is needed are often difficult, because of periodic cycles, random events, and spatial hetero- geneity and because many variables can change systematically during the baseline study period (e.g., owing to succession). Statistical guidance is available for designing baseline and monitoring programs once the variables of interest have been identified (Cowell, 1978; Eberhardt, 1976, 1978; Green, 1979; Kumar, 1980; Lucas, 1976; Sharp et al., 1979; Ward, 1978; Zar, 1976~. Two common problems that make it difficult to design projects as experiments properly are the lack of adequate controls (Cowell, 1978) and the lack of true replicates (Eberhardt, 19761. Eberhardt (1976) suggested a "pseudodesign" with baseline data on a control area and the project site. They can be compared with data collected when the project is complete, with replicates in time substituting for spatial replicates. Baseline and monitoring studies are most effective if they are statistically designed to detect changes of the magnitude expected (Zar, 19761. This expectation in turn determines the extent of sampling required (Hartzbank and McCusker, 19791. In highly variable systems, adequate sampling might be too expensive, and resources might be better used in carrying out less direct studies. Baseline information can sometimes be derived after impacts have already occurred (e.g., Cowell and Syratt, 19791. The Lake Washington case (Chapter 20) is an excellent example of testing hypotheses concerning the effect of lake fertilization changes on the makeup of plankton communities. A great deal was learned from this case, because monitoring continued throughout the development and treat- ment of the problem. Similarly, scientists at Southern Indian Lake (Chapter 21) were able to test hypotheses derived from the results of other lake studies and current limnological models. Carefully designed studies before
112 KINDS OF ECOLOGICAL KNOWLEDGE AND THEIR APPLICATIONS and after project development showed that some of the predictions were wrong because the models used were not based on knowledge of lakes in a permafrost zone. Field and laboratory experiments can be used to test hypotheses. The Garki malaria project (Chapter 15) was itself a large-scale experiment to investigate a model for controlling malaria through a combination of drugs and mosquito control. Careful monitoring studies before, during, and after applications allowed the model to be tested, and an important phenomenon exophily was discovered. SPECIFYING PREDICTIONS AND DETERMINING THE SIGNIFICANCE OF EFFECTS A major purpose of developing general ecological hypotheses is to generate a set of specific predictions of ecological change that can be used in decision-making. The predictions should be as clearly and precisely stated as possible. The period over which a change is expected to occur, the bases of the prediction, and the degree of uncertainty should be spec- ified. Determining the significance of predicted or observed ecological changes is often very difficult, because ecological systems are not fully understood. A clear distinction, if it can be made, between the magnitude of a change and its biological importance is useful. The rates of change and recovery are often important components of ecological effects (e.g., Cairns and Dickson, 19801. The overall significance of an effect is tied closely to the definition of environmental goals. The best course for scientists is to predict or describe changes precisely. Whether or not a change is " significant" is a judgment that transcends science and is best made by all interested parties. Several of the cases studied were organized around tests of specific predictions. In the derelict lands restoration case (Chapter 18), predictions were derived from empirical results of other restoration efforts and basic plant ecological theory. The bases of these predictions were clearly stated, and tests of them produced results of value to other restoration projects. Predictions in the Atomic Energy Commission radiation studies were de- rived from knowledge of food-chain dynamics and laboratory studies, and hypotheses were continually revised as predictions were tested experi- mentally (Chapter 221. In the Lake Washington case, scientists predicted not only specific changes in water quality, but also the periods over which deterioration and recovery would occur (Chapter 201. In the Southern Indian Lake studies (Chapter 21), predictions were based on analogs and limnological
A SCIENTIFIC FRAMEWORK FOR ENVIRONMENTAL PROBLEM-SOLVING 1 13 principles, but were mostly qualitative. In both cases, careful monitoring was incorporated into comprehensive postproject analyses to improve un- derstanding. MONITORING Biological monitoring is used in ecological studies in two basic ways. Studies conducted during or after an action or project are designed to learn what ecological changes resulted. Anticipatory monitoring is designed to track the effects of activities that might be cumulative or pose hazards to human health (Baker, 19761. Properly done, monitoring provides contin- uous indexes of environmental quality that can signal environmental deg- radation or improvement (Chapter 7~. In the event of unexpected environmental changes, monitoring can facilitate adaptive changes in management and in the design of ecological studies (Hilborn et al., 1980; Holling, 1978; Walters and Hilborn, 19761. From a broader perspective, followup monitoring and retrospective anal- ysis are ways to learn from experience and improve the prediction of ecological effects. Monitoring is most effective when it is designed to test ecological hypotheses and when preproject studies have provided baseline information (see Beanlands and Duinker, 19831. Postproject studies of the accuracy of predictions are useful, but are not as useful as followup monitoring that coordinates preproject and postproject sampling and that tests relevant hypotheses. Periodic analysis of results can help to detect unexpected changes and evaluate sampling programs, allowing them to be changed in a timely way. Thus, an iterative approach to monitoring with results fed into study design is often effective, particularly when methods have not been well tested and when effects are uncertain. Any changes in sampling, however, must be made carefully, to ensure that new data are statistically comparable with those already collected. Baseline monitoring of characteristics with substantial variation has a low probability of helping to detect changes due to a project. Measure- ments of baseline variability can help to identify the characteristics that it will be useful to measure in followup studies (Green, 1979) and can be used, with estimates of the duration of effects, to determine how long followup should continue. Even if all the above criteria are met, followup studies of ecological effects can help in planning only if they are made available in an easily digestible form, ideally as published summaries and as complete postproj- ect analyses (Hilborn and Walters, 19811. In several case studies, followup monitoring was shown to be part of
114 KINDS OF ECOLOGICAL KNOWLEDGE AND THEIR APPLICATIONS an overall study design to test hypotheses (Chapters 15, 22, and 23~. The hydroelectric development in Southern Indian Lake (Chapter 21) was treated as a large-scale experiment, and monitoring provided increased understanding of artificial lakes. Because the results were analyzed and published, they can be applied to other cases. In the caribou case (Chapter 16), monitoring of caribou movements and herd productivity began before development and continued during con- struction. Daily information on caribou movement was incorporated into constraints on construction activities. Annual monitoring of catch and fishing effort is used by the International Pacific Halibut Commission to set fishing quotas. Monitoring of conditions in Lake Washington allowed scientists to detect changes, predict trends in eutrophication, and predict and document recovery of the lake after action was taken; because the work was published, its lessons are readily available to managers of similar projects. Monitoring for DDT in the environment first identified the spread of another important group of toxic chemicals, PCBs (Chapter 241. SUMMARY: DEVELOPING A STUDY STRATEGY A study strategy is a plan for conducting ecological studies to help to predict and manage ecological effects. It is motivated by the environmental goals identified in scoping and is organized around the scientific objectives defined on the basis of the goals. Scoping identifies what information is required, and the study strategy specifies how to acquire it. A problem must be carefully thought through before studies aimed at solving it begin (Beanlands and Duinker, 19831. Potential study objectives should be evaluated, so that efforts can be devoted to studies with some chance of producing useful results. What information is needed? Why is it needed? Is it possible to acquire adequate information? How will the information be used to satisfy the ecological goals? How will it be used in decision-making? Highly accurate charac- terization of a variable is of little use if decisions are made on the basis of other considerations. Decision analysis helps to ensure that modeling and research remain focused on the objectives. A basic first step in designing studies is a review of what is already known about the problem. Larkin (1984) believes that literature review can provide more than 50% of the information needed in most initial impact assessments, and as much as 75% when coupled with brief re- connaissance surveys. To summarize, a broad ecological study strategy:
A SCIENTIFIC FRAMEWORK FOR ENVIRONMENTAL PROBLEM-SOLVING 1 15 · Is based on thought-out environmental goals. · Is organized around clearly defined scientific objectives designed to satisfy the environmental goals. · Includes a description of the boundaries established for the problem, with demonstration of their appropriateness. · Is designed to evaluate hypotheses about how the ecological system functions and will be affected by perturbations. · Specifies predictions that will be tested, with the basis of the pre- dictions and a statement of confidence in their accuracy. · Defines the basis for choosing environmental goals and evaluating their significance. · Explains clearly how each part of the study fits into the overall design. · Provides for baseline and followup monitoring to determine the ef- fects of the project or perturbation. · Allows the results of the study to be used to evaluate the plan and to modify it if necessary.