Resilience of Large-Scale Resource Systems

Lance H. Gunderson, C. S. Holling, Lowell Pritchard Jr., and Garry D. Peterson

Regional-scale systems of people and nature provide some of the most vexing challenges for attaining social goals of sustainability, biological conservation, or economic development. There are many more examples of failures than successes, as measured by numerous resource systems that exist in a constant or recurring state of crisis (Ludwig et al. 1993). In the Florida Everglades, agricultural interests, environmentalists, and urban residents contest with one another for control over clean water (Light et al. 1995). In the Pacific Northwest region of the United States, various advocates of salmon argue over the appropriate use of the Columbia River with those who prefer cheap hydroelectric power (Lee 1993; Volkman and McConnaha 1993). The nations surrounding the Baltic Sea struggle with issues of governance as the fish populations and water quality of the sea declines (Jansson and Velner 1995). Within Zimbabwe, large-scale land use conversions are testing stabilities of both ecological and political structures. In these cases resource management has taken a pathological form in which the complexity of the issues, institutional inertia, and uncertainty lead to a state of institutional gridlock, when inaction causes ecological issues to be ignored and existing policies and relationships to be continued.

Paradoxically, this failure often arises from the success of initial management actions. Managers of natural resource systems are often successful at rapidly achieving a set of narrowly defined goals. Unfortunately, this success encourages people to build up a dependence upon its continuation while simultaneously eroding away the ecological support that it requires. This leads to a state in which ecological change is increasingly undesirable to the people dependent upon the natural resource and simultaneously more difficult to avoid. This management pathology leads to unwanted changes in nature, a loss of ecological resilience, conservative management policies, and loss of trust in management agencies.

Recent work reveals a way out of this pathology in large, regional-scale systems. These systems move through periods of surprise, crisis, and reformation (Gunderson et al. 1995). Managers are surprised when the inadequacies of many, if not most, management policies are revealed by ecosystem dynamics. A crisis occurs when it is becomes unambiguously clear that existing policies caused this surprise. The crisis is followed by periods of denial, resistance, and often, finally, by a period of reformation during which new policies are developed and implemented. It is during these periods of crisis that institutions and the connections between them are most open to dramatic transformation. This ability to transform and survive requires that the resource system have sufficient resilience to permit the experimental development of new management policies.

What Is Resilience?

Resilience has been defined in two different ways in the ecological literature, each reflecting different aspects of stability. One definition focuses on efficiency and depends on constancy and predictability—all attributes of engineers' desire for fail-safe design. The other focuses on persistence, despite change and unpredictability—all attributes embraced and celebrated by evolutionary biologists and by resource managers who search for safe-fail designs. Holling (1973) first emphasized these contrasting aspects of stability to draw attention to the tensions between efficiency and persistence, between constancy and change, and between predictability and unpredictability.

The more common definition, which we term engineering resilience (Holling 1996), conceives ecological systems to exist close to a stable steady state. Engineering resilience, then, is the speed of return to the steady state following a perturbation (Pimm 1984; O'Neill et al. 1986; Tilman and Downing 1994). This idea of disturbance away from and return to a stable state is also at the center of twentieth-century economic theory (Varian 1992; Kamien and Schwartz 1991).

The second definition, which we term ecological resilience (Walker et al. 1981; Holling 1996), emphasizes conditions far from any stable steady state, where instabilities can shift or flip a system into another regime of behavior—in other words, to another stability domain (Holling 1973). In this case, resilience is measured by the magnitude of disturbance that can be absorbed before the system is restructured with different controlling variables and processes.

The differences between these two aspects of stability—essentially between a focus on maintaining efficiency of function (engineering resilience) and a focus on maintaining existence of function (ecological resilience)—are so fundamental that they can become alternative paradigms in which subscribers dwell on received wisdom rather than the reality of nature. Those using the concept of engineering resilience tend to explore system behavior near a known stable state, while those examining ecological resilience tend to search for alternative stable states and the properties of the boundaries between states.

Those who explore engineering resilience and the near-equilibrium behavior of ecosystems operate in the primarily deductive tradition of mathematical theory (e.g., Pimm 1984) that imagines simplified, untouched ecological systems; or they draw upon the traditions of engineering, which are motivated by the need to design systems with a single operating objective (Waide and Webster 1976; DeAngelis 1980; O'Neill et al. 1986). These approaches simplify the mathematics and accommodate the engineer's drive to develop optimal designs. However, there is an implicit assumption that ecosystems exhibit only one equilibrium steady state or, if other operating states exist, that those states should be avoided (figure 1.1).

On the other hand, those who emphasize ecological resilience come from traditions of applied mathematics and applied resource ecology at the scale of ecosystems—for example, of the dynamics and management of freshwater systems (Fiering 1982), of forests (Holling et al. 1977), of fisheries (Walters 1986), of semi-arid grasslands (Walker et al. 1981) and of interacting populations in nature (Sinclair et al. 1990; Dublin et al. 1990). Because these researchers are rooted in inductive rather than deductive theory formation, and because they have experience with the impacts of large-scale management actions, they believe that it is the variability of critical variables that forms and maintains the stability landscape. When this variability is reduced, an ecosystem can flip from one organization to another (figure 1.1).

In economics, there has also been a focus on single stable state. The history of economics has been to rapidly move from establishing the existence of a general equilibrium to examining issues of equilibrium uniqueness, stability, and comparative statics. If multiple equilibria are shown to theoretically exist, then the challenge is to theoretically reduce the salience of alternate stable states by proposing that expectations, norms, and social institutions make some equilibria unlikely. This approach does not examine or explain the conditions that can cause a system to move from one stability domain to another. Recently, however, the identification of multi-stable states due to path dependence (Arthur et al. 1987), chreodic development (Clark and Juma 1987), and non-convexities such as increasing returns to scale (David 1985) has reintroduced multiple stable states to economics.

The existence, or at least the importance, of multiple or single stable states determines the appropriateness of an engineering or ecological approach to resilience. If it is assumed that only one stable state exists or can be designed to exist, then the only possible definition and measures for resilience are near-equilibrium ones—such as characteristic return time. And that is certainly consistent with the engineer's desire to make things work—and not to intentionally make things that break down or suddenly shift their behavior. But nature and human society are different.

Why Study Resilience?

Complex resource systems are organized from the interactions of a set of ecological, social, and economic systems across a range of scales. Resilience is central to understanding the dynamics of these systems and their vulnerability to various shocks and disruptions. Resilience measures the strength of mutual reinforcement between processes, incorporating both the ability of a system to persist despite disruptions and the ability to regenerate and maintain existing organization. Resilience allows a system to withstand the failure of management actions. Management is necessarily based upon incomplete understanding, and therefore ecological resilience allows people in resource systems the opportunity to learn and change.

The importance of the role of resilience in ecosystems, flexibility of institutions, and incentives in economies emerged in a sequence of meetings held on the island of Askö in the Swedish archipelago. Sponsored by the Beijer International Institute for Ecological Economics, these meetings brought together economists and natural scientists to explore similarities and differences in views and experiences of change. Their conclusions were that economic growth is not inherently good, nor inherently bad, but that economic growth cannot in the long term compensate for declines in environmental quality. They also concluded that the growing scale of human activities is encountering the limits of nature to sustain that expansion (Folke and Berkes 1998; Arrow et al. 1995).

The familiar responses to these issues are often flawed, because the theories of change underlying them are inadequate. The stereotypical economist might say "get the prices right" (i.e., ensure that prices internalize significant environmental externalities) without recognizing that price systems require a stable context where social and ecosystem processes behave "nicely" in a mathematical sense (i.e., are continuous and convex). The stereotypical social scientist might say "get the institutions right" without comprehending the degree to which those institutions submerge ecological uncertainties and economic and political interests. The stereotypical ecologist might say "get the indicators right" without recognizing the surprises that nature and people inexorably and continuously generate. And the stereotypical engineer might say "get the technological control right and we can eliminate those surprises" without recognizing the limits to knowledge and control imposed by the inherent uncertainty and unpredictability of the ever-evolving interaction of people and nature.

Although based on bad or insufficient theory, such simple prescriptions are attractive because they seem to replace inherent uncertainty with the spurious certitude of ideology, of precise numbers or of action. The theories implicit in these examples ignore multi-stable states. They ignore the possibility that the slow erosion of key controlling processes can cause an ecosystem or economy to abruptly flip into a different state that might effectively be irreversible. In an ecosystem, this might be caused by the gradual loss of a species in a keystone set that together determine structure and behavior over specific ranges of scale. In a resource-based economy, it might be implementation of maximum sustained yield policies that reduce spatial diversity, evolve ever-narrower economic dependencies, and develop more rigid organizations. In an economy, it might be caused by the channeling of loans through personal networks, allowing bad loans to accumulate to such a point that they cause an entire banking and finance system to collapse—such as the Asian financial crisis in the late 1990s.

It increasingly appears that effective and sustainable development of technology, institutions, economies, and ecosystems requires ways to deal not only with near equilibrium efficiency but also with the reality of more than one possible equilibrium. If there are multiple equilibria, in which direction should the finger on the invisible hand of Adam Smith point? If there is more than one objective function, where does the engineer search for optimal designs? In such a context, a near-equilibrium approach is myopic. Attention should shift to determining the constructive role of instability in maintaining diversity and persistence and to management designs that maintain ecosystem function despite unexpected disturbances. Such designs maintain or expand the ecological resilience of those ecological "services" that invisibly provide the foundations for sustaining economic activity and human society.

The goal of this volume is to begin to understand how the properties of ecological resilience and human adaptability interact in complex, large systems (regional scale). To lay a foundation for this volume, we initially review other key properties of complex adaptive systems that contribute to resilience.

Properties of Complex Adaptive Systems

We propose that the behavior of complex adaptive systems depends upon four key properties: ecological resilience, complexity, self-organization, and order. As discussed above, resilience is the extent to which a system can withstand disruption before shifting into another state. Complexity is the variety of structures and processes that occur within a system. Self-organization is the ability of these structures and processes to mutually interact to reinforce and sustain each other. The process of self-organization produces order from disorder, but the interaction of processes across scales also destroys, and reconfigures, ecological organization, producing complex ecological dynamics. The next three sections elaborate upon the role these properties play in complex systems, and how these other properties contribute and interact with resilience.

Diversity and Stability

The relationship between biological diversity and ecological stability has been an ongoing debate in ecology since the time of Darwin (1860; also Elton 1958; May 1973; Tilman and Downing 1994, 1996). The question is whether an ecosystem that includes more species is more stable than one that includes fewer species?

Tilman and Downing (1994) and Tilman (1996) demonstrated that an increase in species number increases the efficiency and stability of some ecosystem functions but decreases the stability of the populations of the species, at least over ecologically brief periods. Although this work is important and interesting, it focuses only on the behavior of ecosystems near some steady state. But, as we've discussed above, we feel it is important to discover the role of ecological diversity over a much broader range of variations. This is where the relationship between diversity and resilience has been poorly developed.

When grappling with this broader relationship between diversity and resilience, most turn to two commonly discussed hypotheses: Ehrlich's (1991) rivet hypothesis and Walker's (1992) driver and passengers hypothesis. The rivet hypothesis proposes that there is little change in ecosystem function as species are added or lost, until a threshold is reached. At that threshold the addition or removal of a single species leads to system reorganization (just as popping rivets from a seam causes little change at first, but at some point sudden, disastrous change will occur). The rivet hypothesis assumes that species have overlapping roles and that as species are lost the ecological resilience of the system is decreased, and then overcome entirely. Walker proposes that species can be divided into functional groups, or guilds, which are groups of species that act in an ecologically similar way. Walker proposes that these groups can be divided into "drivers" and "passengers." Drivers are "keystone" species that control the future of an ecosystem, while the passengers live in but do not significantly alter their ecosystem. However, as conditions change, endogenously or exogenously, species shift roles. Removing passengers has little effect, while removing drivers can have a large impact. Ecological resilience resides both in the diversity of the drivers and in the number of passengers who are potential drivers. These two hypotheses provide a start, but richer models of ecological complexity are needed that better incorporate ecological processes, dynamics, and scale.

Ecosystems are resilient when ecological interactions reinforce one another and dampen disruptions. Such situations may arise due to compensation when a species with an ecological function similar to another species increases in abundance as the other declines (Holling 1996) or as one species reduces the impact of a disruption on other species.