Unruly Complexity: Ecology, Interpretation, Engagement

by Peter J. Taylor

Overview

Ambitiously identifying fresh issues in the study of complex systems, Peter J. Taylor, in a model of interdisciplinary exploration, makes these concerns accessible to scholars in the fields of ecology, environmental science, and science studies. Unruly Complexity explores concepts used to deal with complexity in three realms: ecology and socio-environmental change; the collective constitution of knowledge; and the interpretations of science as they influence subsequent research.

For each realm Taylor shows that unruly complexity-situations that lack definite boundaries, where what goes on "outside" continually restructures what is "inside," and where diverse processes come together to produce change-should not be suppressed by partitioning complexity into well-bounded systems that can be studied or managed from an outside vantage point. Using case studies from Australia, North America, and Africa, he encourages readers to be troubled by conventional boundaries-especially between science and the interpretation of science-and to reflect more self-consciously on the conceptual and practical choices researchers make.

Product Details

ISBN-13:	9780226790350
Publisher:	University of Chicago Press
Publication date:	09/15/2005
Pages:	232
Product dimensions:	6.00(w) x 9.00(h) x 0.90(d)

About the Author

Peter J. Taylor is associate professor and director of the Programs in Science, Technology & Values and Critical & Creative Thinking at the University of Massachusetts, Boston, where he teaches environmental studies, science studies, and reflective practice.

Read an Excerpt

Unruly Complexity
Ecology, Interpretation, Engagement

By Peter J. Taylor The University of Chicago Press
Copyright © 2005
The University of Chicago
All right reserved.

ISBN: 978-0-226-79036-7


Chapter One Problems of Boundedness in Modeling Ecological Systems

A. The Construction of Complexity

What have ecological theorists learned about the relationship between complexity and stability?

MacArthur (1955) proposed that "the amount of choice which the energy has in following the paths up through the food web is a measure of the stability of the community" (fig. 1.1). This formulation gave a mathematical touch to an old intuition that the diversity or complexity of interrelations among organisms ensures harmony or stability in the order of nature (Egerton 1973). By 1975, however, Goodman, after reviewing the voluminous literature and sorting through diverse definitions of complexity and stability, had concluded that the few well-defined empirical and theoretical studies tended to contradict the expectation that complexity promotes stability (Goodman 1975; see also INTECOL 1974). Ecologists continued to tease apart the different meanings of the terms and examine a variety of complexity-stability relationships (Pimm 1984, 1991), but by the late 1980s, most had become skeptical about any robust relationship ever emerging. For example, in a poll of the members of the British Ecological Society for its 75th anniversary symposia, "the diversity-stability hypothesis" ranked only 35th in a list of the 50 "most significant" ecological concepts. These ecologists deemed "ecosystem fragility" (ranked 10th) a much more "significant" concept (Cherrett 1989).

What exactly could ecologists claim to have learned in turning away from the complexity-stability issue by the end of the 1980s?6 Was the lesson that change, rather than constancy and stability, is the normal state of ecology (Botkin 1990)? Hardly-change, process, and dynamics were central concepts for the founders of the discipline (Kingsland 1995, 9-24). Even Frederic Clements (1874-1945), whose name is associated with the idea of stable, self-sustaining climax communities of plants, was primarily concerned with explaining ecological succession-the changes at some site over time that led predictably, he thought, to some such climax (Clements 1916). And, in the writings of Clements's leading critic, Henry Gleason (1882-1975), nothing was static. Ecologists, Gleason proposed, should analyze the shifting associations of individual species as they respond to variable environmental conditions and view succession as "an extraordinarily mobile phenomenon" (Gleason 1927).

No, the ubiquity of change was not a new insight for ecology. Admittedly, given that environmentalists still invoke the age-old idea of a balance of nature, there was room still for scientific debunking of this concept (Botkin 1990). The complexity-stability question, however, need not be disposed of in a package with the balance-of-nature myth. Nor need it be reduced to questions of productivity and constancy of numbers in mixtures of plants (Tilman 1999). Instead, I believe, attempts to find a stability-complexity connection can be viewed in relation to the long-standing problem of accounting for structure and organization as well as the possibility of change-in short, for ongoing restructuring. In this light, let me reconsider the search during the 1970s and 1980s for a significant relationship between complexity and stability, in which the definition of both terms varies among authors. The lesson will not be that ecological interactions are without any structure or constraints on change. A myth of "anything goes"-that all disruptions are qualitatively alike-would offer no better guidance for human intervention in (or within) nature than the myth of harmony and balance.

Stability of Simple Multipopulation Models

Let me conduct this review in the spirit of MacArthur; namely, by exploring the qualitative behavior of simple mathematical models in the hope of finding some theoretical unity among the disparate facts of ecology. The terms of the models will be the numbers of organisms in each population and the interactions-competition, predation, parasitism, and so on-that influence the growth or decline of those numbers. Contingent circumstances that could obscure any regularities will not be considered (Kingsland 1995, 176-205). These simple models, which have few explicit biological assumptions, are intended-using the analogy of MacArthur's physicist-turned-theoretical-ecologist successor, Robert May-to be like the perfect crystals of solid-state theory in physics. They are gross simplifications and obtain nowhere, but investigation of the deviations of reality from the ideal will suggest new features to add to the initial models-or, as often will be the case, to new formulations of the question at hand. The justification for this approach is that if we were to write down at the outset a model with many specific biological features, it would be difficult to establish whether its behavior depended on the specific features or simply resulted from the basic dynamics common to many other models (May 1973, 10-12). (In chapter 2 this "light-on-data" strategy of modeling is compared with other strategies of modeling.)

At first sight, the impression that nature is complex and nature is stable admits of two broad interpretations:

1. Nature is stable because complexity in general promotes stability. This interpretation is consistent, for example, with the observation that monocultures are much more vulnerable to pest outbreaks than are diverse, natural floras and faunas. 2. Nature is stable because of its particular complexity. This interpretation is consistent with the idea that human disturbances can threaten nature's finely tuned checks and balances.

Let me start with the first view. Could stability as a consequence of complexity be a general property of complex systems? Beginning around 1970, this possibility was examined through computer-aided investigations of mathematical systems governing the rate of change of many interacting components (Gardner and Ashby 1970). For the purposes of ecological theory, the systems could be thought of as ecological communities and their components as populations of interacting species (May 1972). May's approach was to compare many samples, each made up of many model communities. Within any sample, each community would have the same complexity, in the sense of the number of populations, strength of interactions among populations, and interconnectedness (the proportion of pairs of populations actually interacting). The communities would vary only in the arrangement of positive, negative, and zero interactions. The equations relating rates of change of populations to their interactions were bare of biological detail (such as spatial location or individual variability) and were not constrained to fit any recognizable "trophic" structure (such as food chains from plants to carnivores). The equations were simple enough to calculate the equilibrium or steady state population sizes of each community and to determine the community's stability in the sense of tendency to return, when perturbed slightly, to that equilibrium.

With the help of their computers, modelers such as May discovered that stable communities become more rare in the samples with higher complexity. Moreover, for the stable communities to be of biological relevance, they must also be "feasible"-that is, all populations must be positive-and the chance of feasibility drops rapidly with the size of the community (Roberts 1974). In short, complexity almost ensured instability (fig. 1.2).

Responses to the "complexity-instability" results were varied, and each response reformulated the issue in some way: In what sense and to what extent are real ecological communities stable? Do real communities keep below the limiting levels of complexity and strength of interactions suggested by the models? Might complexity promote stability in models whose equations captured more biological detail? If complexity in general does not promote stability, what are the "devious strategies which make for stability in enduring natural systems" (May 1973, 174)?

Goodman's (1975) review concluded that experiments and observations complement the results from modeling. Complexity, when measured by various indices of diversity (Pielou 1975), bore no consistent, let alone positive, relationship to stability, as assessed by the degree of fluctuations of populations or simply by their persistence. Subsequently McNaughton (1977) reported a positive relationship for grasslands. But it was ecological modeling, more than new empirical evidence, that kept the issue alive into the 1980s. Or, one might say, it was the different formulations of ecological modelers that gave the complexity-stability issue new lives.

Saunders and Bazin (1975), for example, realized that if there is any biological basis to the idea that complexity and stability are related, it must have been derived from actual ecological communities. For these communities to have been observed, they could not have been ephemeral. Saunders and Bazin, therefore, considered only model communities having a stable equilibrium and asked whether such communities were more resistant on average to perturbations when more complex (fig. 1.3). (Their question, it should be noted, falls under the second broad interpretation; namely, that nature is stable because of its particular complexity.) They discovered, however, that the speed of recovery from perturbation decreased with complexity. Pimm (1979) extended this line of inquiry by investigating stability after a much larger perturbation; namely, deletion of one population from a stable model community. He discovered that the greater the interconnectedness of populations, the less likely the community was to be stable after the deletion of a population. In short, the reformulation from the proportion of stable systems to the degree of resistance of stable systems to perturbation did not seem to illuminate why complex ecological communities would be stable.

Suppose that the stable complex communities examined by modelers had turned out to be more stable, in some sense, than stable simple communities. Theory would still have had to explain how, given the rarity of such complex communities in the samples of May and others, any of them could come to exist. In the spirit of May's call to expose the "devious strategies" promoting stability, various modelers sought to elucidate the special conditions or structures that ensure that complexity promotes stability, or, at least, ensure that complexity does not make stability very unlikely.

McMurtrie (1975), for example, showed that complex communities are more likely to be stable if nonzero interactions between populations form long cycles from one population through others, then back to the first population. Siljak (1975), following the lead of Simon (1969), suggested that stable communities can be "decomposed" or partitioned into blocks having strong intra-block interactions and weak inter-block interactions. This would mean that in stable communities, intraspecific interactions overshadow interspecific interactions. Features such as low efficiency of assimilation of consumed food, absence of loops of the kind "X eats Y eats Z eats X," and low predation efficiency at low densities were also shown to enhance the complexity-stability relationship (DeAngelis 1975; Lawlor 1978; Nunney 1980; but see Abrams and Taber 1982). Other authors examined the consequences of particular features, such as time lags-between a population change and its effect on an interacting population-and fluctuating environments (May 1973), more complex forms of the equations (Pomerantz, Thomas, and Gilpin 1980), and inclusion of nutrient uptake and cycling (DeAngelis 1992).

Cycles, partitioning, and so on may enhance the possibility of stability in model communities. Yet for each addition that ecologists make to the list of stability-enhancing mechanisms, nature becomes less obliged to employ any particular one of them. Furthermore, even if ecological data display some of the features found in models (Pimm 1984, 1991), it does not follow that they exist in nature because of their stabilizing effect. A stronger case for any mechanism is made if the modeler can account for how nature arrives at the special structures, or how populations violating the special conditions are eliminated. That is, modelers need the dynamic evolution of their model communities to produce the feature in question. This means that modelers have to resist pressure for their models to look realistic to other ecologists at the outset. When modelers impose some biologically realistic constraint on their model communities in advance, they lose the chance of explaining it dynamically.

If dynamic evolution is required, then the question of how any stable complex community exists is modified: In what ways does nature arrive at the arrangements that persist longer than those that preceded them? Whereas May's complexity-promotes-instability result assumes that complex ecological communities are outcomes of a sampling process, most actual communities are the result of some process of development over time. Many authors have suggested that natural selection might operate in a way that modifies interactions among populations in a direction that enhances the stability of communities (Lawlor and Maynard-Smith 1976; Roughgarden 1977; Saunders 1978; Lawlor 1980). However, early theoretical investigations of a very simple community (one predator-one prey) suggested that this is not necessarily true (Rosenzweig 1973), and subsequent modeling of more complex communities of competitors indicated that interactions become progressively weaker until only a few interacting populations dominate (Ginzburg, Akcakaya, and Kim 1988).

This continuing series of negative or not-so-compelling complexity-stability relationships might lead one to conclude that there is, as Goodman (1975) intimated, nothing there to be discovered. This is the position adopted by DeAngelis and Waterhouse in their important review (1987). To these authors, May's complexity-instability result indicates that a "potential biotic feedback instability [is] inherent in complex natural systems" (DeAngelis and Waterhouse 1987, 5). (By the 1980s, this result had become conventional wisdom, a status that persists, even among the plant ecologists who have revived interest in diversity: see note 6; but see May 2000.) Noting the limited success of different attempts to counteract this instability, DeAngelis and Waterhouse concluded that one should not "base any theory of ecological communities solely on the notion of inherent ecological stabilizing mechanisms" (1987, 9).

DeAngelis and Waterhouse's review of ecological modeling mapped even more paths than I have so far (fig. 1.4). They started, as I did here, from an "equilibrium" view, in which communities move toward or away from a steady state. Attention to disruption from internal feedbacks led ecologists to emphasize "biotic instability" (path a in the figure). Similarly, environmental disturbances that are irregular (or "stochastic") led to "stochastic domination" (path b). To account for the persistence of communities despite these disruptions, several paths have been taken: exploration of potential stabilizing mechanisms (as I have discussed; path 1), studies of disturbance patterns that might interrupt any process that eliminates an interacting population (path 2), and studies of biological mechanisms that might compensate when populations have been driven to low numbers (path 3). But to DeAngelis and Waterhouse, the most promising approach entails a "landscape" view, in which a community may persist in a landscape of interconnected patches even though the community is, because of instability and environmental disruption, transient in each of the patches (paths 4 and 5). Although I endorse most of DeAngelis and Waterhouse's interpretations, I want to draw attention to another path that leads to the landscape view, which they, and most other modelers, have overlooked or discounted (path c and 6). To introduce this path, I need to return to the question of how nature arrives at the arrangements that persist.

A Constructionist View

Real ecological communities develop over time through succession-that is, the addition, growth, decline, and elimination of populations. Several modeling studies have shown that although stable, feasible model communities may be extremely rare as a fraction of the communities sampled, they can be readily constructed over time by the addition of diverse populations from a pool of populations and by the elimination of populations from communities not at a steady state (fig. 1.5). (Tregonning and Roberts [1979] made important initial contributions, which I followed; see Taylor 1988b and references therein.) I call this insight about complexity-stability theory a constructionist view.

(Continues...)

Excerpted from Unruly Complexity by Peter J. Taylor Copyright © 2005 by The University of Chicago. Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.

Table of Contents

Acknowledgments
Prologue
Part I - Modeling Ecological Complexity
1. Problems of Boundedness in Modeling Ecological Systems
A. The Construction of Complexity
B. The Hidden Complexity of Simple Models
2. Open Sites in Model Building
Part II - Interpreting Ecological Modelers in Their Complex Social Context
3. Metaphors and Allegory in the Origins of Systems Ecology
A. Social-Personal-Scientific Correlations in the Work of H. T. Odum: A First Reading
B. Another Look: Diagrams and Physical Analogies
4. Reconstructing Heterogeneous Webs in Socio-Environmental Research
A. The Simulated Future of a Salt-Affected Agricultural Region
B. An Intersection of Domains of Action that Include MIT, USAID, System Dynamics Modelers, and Nomadic Pastoralists
Part III - Engaging Reflexively Within Ecological, Scientific, and Social Complexity
5. Reflecting on Researchers' Diverse Resources
A. Further Intersections that Affect Researchers and Interpreters Extending Their Webs
B. Workshops in which Ecologists Map Their Webs of Knowledge-Making
C. Two Terms that Help Researchers Conceptualize More Complexity
6. Reasoned Understandings and Social Change in Research on Common Resources: Introducing a Framework to Keep Tensions Active, Productive, and Ever-Present
A. Researchers Conduct a Dialogue, Involving Concepts and Evidence, with the Situations Studied
B. Socially Situated Researchers Interact with Other Social Agents to Establish What Counts as Knowledge
C. Researchers Pursue Social Change by Addressing Self-Consciously the Complexities of the Situations They Study and Their Own Social Situadness
Epilogue: Three Stories
A. Participation
B. Flexible Engagement
C. Open Questions
Summary of Themes and Questions Opened Up
Glossary
Notes
References
Index