Evaluating Ecological Tradeoffs
The primary strategy for restoring the Everglades ecosystem is the restoration, to the extent possible, of the hydrologic regime. However, the ultimate goal is restoration of the Greater Everglades Ecosystem while meeting society’s needs for flood control and water supply. Earlier in this report and in previous reports of this committee (e.g., NRC, 2003b), we have discussed the importance of evaluating the restoration effort; the NRC (2003b) also recommended that conflicts among ecological restoration targets should be identified and that system-wide indicators be developed. These issues have been a major focus of the Restoration Plan scientists as well. Major difficulties are associated with such evaluations. One is translating the general ecological and societal goals espoused in the Central and South Florida Restudy into realistic targets and practical performance measures (NRC, 2003b). Another difficulty is that restoration of one aspect of ecosystem functioning or of biological diversity might have to come at the expense of another. Furthermore, not all aspects of ecosystem structure and functioning are equally valued by all sectors of the public or even by all agencies in the region.
In addition, three major components of the Restoration Plan—aquifer storage and recovery (ASR), Lake Belt storage, and water reuse—have major uncertainties that still need to be resolved through pilot studies. For example, their efficiencies might be lower than assumed, their costs might be higher, or water-quality issues might be problematic.
All the above factors make it imperative that there be a quantitative framework for evaluating possible modifications to the plan as needed. Even if the general goals remain unchanged (and they might not), the restoration strategies and targets will need to be revisited and alternative management scenarios judged in the light of new information. Decision-makers and interest groups will probably prefer different alternatives based on how they view the tradeoffs among goals for subsets of the ecosystem and among desired endpoints. Experience suggests that a structured decision process that synthesizes information, reflects these tradeoffs, and accounts for different stakeholder preferences can promote constructive analysis and negotiation (e.g., Clemen, 1991; Ridgely and Rijsberman, 1992; NRC, 1996b; Prato, 2003; Brown et al., 2001) and thus help to operationalize adaptive management. To this end, the committee has considered multiattribute (or multicriterion) approaches to decision making. It proposes the use of a system
performance measure that could be used together with specific indicators and performance measures already in place to help to evaluate restoration progress and alternatives, including re-evaluation and refinement of restoration goals.
The proposed system performance indicator does not in itself lead to decisions; instead, it allows alternative scenarios or outcomes to be evaluated. To make decisions, it will be necessary to weight various outcomes and aspects of ecosystem structure and functioning. A recent NRC committee described these considerations in identifying options for protecting Atlantic salmon in Maine (NRC, 2004a). It described the need for “differences in perspectives [to be] taken into account so that the decision is informed by the views of all parties having legitimate interests in the outcome.” Like that earlier NRC committee, this committee cannot perform such weighting, because value judgments are involved, as well as scientific estimates. The best that can be attained is a clear description of a weighting algorithm so that policy makers and stakeholders in the Everglades restoration can do the hard work of agreeing on the weights to be assigned.
Characteristics of a System Performance Measure
In an ideal world, a system performance measure for evaluating alternative Everglades restoration plans would be a single measure of the degree to which a given plan meets the Restoration Plan objectives—water supply and flood control for the built environment, and ecosystem restoration. Such a system measure would need to quantify performance in a way that is consistent with societal preferences. This would mean specifying relative societal preferences both within and across the main categories of water supply, flood control, and ecosystem restoration. How does society value water supply for agriculture versus water supply for municipalities? Restoration of the ridge and slough landscape versus restoration of the marl prairie? Water supply versus flood control? Flood control versus ecosystem restoration?
Obviously, there are many conceptual and practical difficulties in developing an “ideal” system performance measure for Everglades restoration. Some of these difficulties can be avoided by excluding consideration of the built environment and measuring only the degree to which a given restoration plan meets the Restoration Plan ecosystem restoration objectives. To a large extent, Restoration Plan objectives pertaining to the built environment are legally mandated and cannot be compromised without changing the law. (Of course, laws can be changed; an overall ecological performance measure for the Restoration Plan could be used in an analysis to evaluate the degree to which current built environment mandates limit restoration success.)
The Everglades ecosystem consists of several identified, distinct components, such as marl prairie and ridge and slough terrain. Estimation of a system measure of the degree to which a particular restoration plan meets Restoration Plan objectives requires the ability to estimate how the value of a particular ecosystem component is affected by the restoration plan, as well as assignment of relative value to all identified ecosystem components. This clearly means that a system restoration performance measure must be based on restoration outcomes that can be both modeled and valued. Modeling of the Everglades is highly advanced, with respect to both hydrologic and ecosystem processes, although the latter are much more difficult to quantify. In the next section we develop a conceptual system performance measure that focuses on restoration of individual components of the Everglades ecosystem.
A Conceptual System Performance Measure
To help in its goal of restoring the hydrologic regime of the Everglades, the South Florida Water Management Model provides a quantitative tool for predicting how various restoration strategies would modify the hydrologic regime. For these reasons we base our conceptual system performance measure on hydrologic performance measures. The current set of hydrologic performance measures could be used for this purpose, although it might be desirable to modify this set. The restoration of each particular ecosystem component will depend on one or more hydrologic performance measures (i.e., system attributes). This dependence can be expressed mathematically as (in the lexicon of economics) a utility function that relates the numerical value of a performance measure to society’s degree of satisfaction with the restoration. A separate utility function is developed for each performance measure and an overall degree of utility is then calculated by weighting and then combining utility scores.
A review of the extensive literature on methods for deriving weights and combining utility scores is beyond our scope here. Multiattribute decision-making frameworks, however, are increasingly used in ecosystem management and are appealing for their simplicity, abiltity to engage stakeholders, and their flexibility in handling nonmarket values that have challenged more traditional cost-benefit approaches (e.g., Prato, 1999). One measure of overall restoration utility is the weighted sum of such individual utility functions, a widely used formulation (Keeney, 1982; Poyhonen and Hamalainen, 2001). The functions account for the degree to which each ecosystem component is restored as a result of a hydrologic regime corresponding to the hydrologic performance measures. The weights account for the value given to each restoration component relative to other components. Hence the overall restoration hydrologic performance measure estimates the relative value of the restoration associated with alternative restoration plans. This concept is presented more formally below.
Measuring System Restoration Performance
Assume there are “n” ecosystem components and a set of hydrologic performance measures. Let X0 be the vector of all the hydrologic performance measures under pre-restoration conditions. Let XT be the vector of the values of all the target hydrologic performance measures. For the restoration alternative j, let Xj be the corresponding vector of values of all the hydrologic performance measures. The vectors X0, XT, and Xj, can be obtained from simulations of the South Florida Water Management Model. Let fi(Xj) be a multivariate function representing the expected fractional restoration of the ecosystem i under restoration alternative j. (Note that fi will be insensitive to hydrologic performance measures that do not relate to ecosystem i.) Also, for all i, fi(X0) = 0 and fi(XT) = 1. In general, fi(Xj) would be expected to vary between zero and one, although it would be possible to evaluate restoration plans that would degrade one or more ecosystem components relative to pre-restoration conditions (in which case fi(Xj) would be less than one), or that would restore one or more components to conditions that exceed the target conditions (fi(Xj) > 1). Also, the value represented by 1 might change with the advent of new knowledge.
Let wi be a number between zero and one representing the imputed value of ecosystem i relative to the imputed value of the most-highly valued ecosystem, where the sum of all wi equals one. Then the system performance measure (SPM) associated with a particular restoration configuration on a set of independent hydrologic performance measures is given by
We recommend that in the initial uses of the proposed indicator, all of the weights be equal, in which case wi = 1/n for all i. If all weights are chosen equal,
Clearly, specification of the fi's would present the greatest challenge to the implementation of this measure. The existing ecological models would provide a scientific basis defining these functions, but some degree of subjective judgment also would be required. The Habitat Suitability Indices recently developed by the South Florida Water Management District would provide an excellent starting point. Another concern is whether the different performance measures are fully substitutable for each other, as is implied by additive weighting.
Application of Proposed System Performance Measure
A system restoration performance measure like the one described above is intended to complement rather than substitute for other evaluation tools already in place (e.g., ecological assessment models like ATLSS). Its main value would be in the context of an inclusive and explicit group-process for evaluating policy and management tradeoffs and alternatives (e.g., Ridgley and Rijsberman, 1992; Prato, 2003). As such the system performance measure could be used as a tool for re-examining restoration targets, for comparing the relative contributions of different components of the Restoration Plan to overall restoration progress, and for examining the sensitivity of measured restoration progress to differences in weights assigned by different stakeholders.
For example, during the screening studies preceding the Restoration Plan it was demonstrated that manipulations of water level in a partitioned or whole Lake Okeechobee would provide a cost-effective alternative to ASR. However, this strategy was not considered because it prevented restoration of the Lake Okeechobee littoral zone. An overall restoration performance measure would allow for a more formal examination of the balance between restoration benefits in the Lake Okeechobee ecosystem against other restoration benefits.
Simple Hypothetical Example of the Use of a System Performance Indicator
Consider two ecosystems that must share water provided by a restoration project. Assume that there is only one hydrologic performance measure for each ecosystem, the average annual flow to that ecosystem. Let x1 and x2 be the respective flows and let x1T and x2T be the respective target flows required for full restoration. Assume that x1T = x2T = Q*/2, where Q* is the total flow required for full restoration. Assume that the two ecosystems are equally valued, and hence w1 = w2 = 0.5.
Let f1(x1/x1T) and f2(x1/x1T) be functions quantifying fractional restoration as a function of the ratio of average annual flow to the target flow. Assume that these functions are given in Figure 5-1 above.
If the targets are met for each ecosystem, the value of the system performance measure is given by
Consider the case for which the available water supply, Q, is less than Q*. If water is allocated to optimize the SPM, the values of f1(x1/x1T) and f2(x1/x1T) must be equal, given that the ecosystems are assumed to have equal value (w1 = w2). From Figure 5-1 it is clear that this will require that ecosystem 1 be allocated much less water, as it is much less sensitive to the supply of water. Figure 5-2 illustrates the amount of water that must be allocated to each ecosystem to maximize the SPM, for various values of Q/Q*.
As an example, consider Q/Q* = 0.8.
Based on Figure 5-2, the optimal SPM is attained for x1/x1T = 0.61 and x2/x2T = 0.99. From Figure 5-1 it can be seen that the fractional restorations at these ratios are each 0.96. Hence the optimal SPM is 0.96.
Assume that for the case of Q/Q* = 0.8, water is instead allocated equally to ecosystems 1 and 2, that is, x1/x1T = x2/x2T = 0.8. From Figure 5-1 we see that
The resulting SPM is
This SPM is much lower than the optimal value of 0.96, indicating a very inefficient allocation of water. However, this allocation is optimal if w1 = 0.955 and w1 = 0.45, since for this set of weights
For every allocation of water there is a set of weights that makes that allocation optimal, and the weights are equal for only one of these allocations. In general, a restoration involving more than one ecosystem will likely involve unequal preferences unless an attempt is made to equalize preferences.