Panarchy Holling

M I N I R E V I E W S

Understanding the Complexityof Economic, Ecological, and

Social Systems

C. S. Holling

Department of Zoology, University of Florida, Gainesville, Florida 32611, USA

ABSTRACTHierarchies and adaptive cycles comprise the basisof ecosystems and social-ecological systems acrossscales. Together they form a panarchy. The panar-chy describes how a healthy system can invent andexperiment, benefiting from inventions that createopportunity while being kept safe from those thatdestabilize because of their nature or excessive ex-uberance. Each level is allowed to operate at its ownpace, protected from above by slower, larger levelsbut invigorated from below by faster, smaller cyclesof innovation. The whole panarchy is thereforeboth creative and conserving. The interactions be-tween cycles in a panarchy combine learning with

continuity. An analysis of this process helps to clar-ify the meaning of sustainable development. Sus-tainability is the capacity to create, test, and main-tain adaptive capability. Development is the processof creating, testing, and maintaining opportunity.The phrase that combines the two, sustainable de-velopment, thus refers to the goal of fosteringadaptive capabilities and creating opportunities. It istherefore not an oxymoron but a term that de-scribes a logical partnership.

Key words: hierarchy; adaptive cycles; multiplescales; resilience; sustainability.

INTRODUCTION

The ecological status of nations and regions is acurrent item for assessment and action on theagenda of several organizations. In the UnitedStates, the National Academy of Sciences and theHeinz Center have issued guidelines to identify sus-tainability indicators. Internationally, the SpeciesSurvival Commission of the World ConservationUnion (IUCN) has stated that sustainability, eitherin a region or of a species, depends on interactionsamong internal and external factors. The internalfactors may be social, political, ecological, or eco-nomic; the external factors include foreign debt,structural poverty, global environmental problems,

and social/political/economic conflicts. Indicators ofsustainability have been identified for all the inter-nal factors, while issues of concern have been sug-gested for the external ones. One unpublished re-port cited 76 specific sustainability indicators for theinternal factors and a more diffuse set of attributesfor the external factors.

All of these indicators and all of the attributesmake sense. The problem is not that they arewrong, or that they are not useful. They are, ifanything, incomplete. Rather, they suggest a com-plexity that can overwhelm understanding, evenwhen, in specific situations, only a subset of theseentities are relevant. There are two approaches tocomplexity.

One of them, which has been explored thor-oughly and incisively by Emory Roe (1998), viewscomplexity as anything we do not understand, be-cause there are apparently a large number of inter-

This paper has been adapted from Gunderson and Holling(2001), with permission of Island Press.Received 7 March 2001; accepted 16 March 2001.*e-mail: [email protected]

Ecosystems (2001) 4: 390405DOI: 10.1007/s10021-001-0101-5 ECOSYSTEMS

2001 Springer-Verlag

390

acting elements. The appropriate approach, accord-ing to Roe, is to embrace the complexity andresulting uncertainty and analyze different subsetsof interactions, each of which seem relevant from anumber of fundamentally different operational andphilosophical perspectives. A recent article in Con-servation Ecology offered a review of this thesis fromfour different disciplinary and policy perspectivesand a commentary on the reviews by the author(www.consecol.org/Journal/vol4/iss2/index.html).

An alternative view (Holling 2000; Gundersonand Holling 2001) suggests that the complexity ofliving systems of people and nature emerges notfrom a random association of a large number ofinteracting factors rather from a smaller number ofcontrolling processes. These systems are self-orga-nized, and a small set of critical processes create andmaintain this self-organization. (Self-organiza-tion is a term that characterizes the developmentof complex adaptive systems, in which multipleoutcomes typically are possible depending on acci-dents of history. Diversity and the individuality ofcomponents, localized interactions among compo-nents, and an autonomous process that uses theoutcomes of those local interactions to select a sub-set of those components for enhancement are char-acteristics of complex adaptive systems [Levin1999]). These processes establish a persistent tem-plate upon which a host of other variables exercisetheir influence. Such subsidiary variables or fac-tors can be interesting, relevant, and important, butthey exist at the whim of the critical controllingfactors or variables. If sustainability means any-thing, it has to do with the small set of criticalself-organized variables and the transformationsthat can occur in them during the evolutionaryprocess of societal development.

But these two views of complexity require alter-native perspectives and competing models and hy-potheses. The goal of each approach is to mobilizeevidence that can distinguish among competing ex-planations so that multiple lines of evidence beginto define what is known, what is uncertain, andwhat is unknown. We are always left with bestjudgments, not certainties.

The view presented here argues that there is arequisite level of simplicity behind the complexitythat, if identified, can lead to an understanding thatis rigorously developed but can be communicatedlucidly. It holds that if you cannot explain or de-scribe the issue of concern using at least a handfulof causes, then your understanding is too simple. Ifyou require many more than a handful of causes,then your understanding is unnecessarily complex.That level of understanding is built upon a founda-

tion of adequate integrative theory, rigorously de-veloped. This theory is rooted in empirical realityand communicated with metaphor and example.The first requirement is to begin to integrate theessence of ecological, economic, and social sciencetheory and to do so with the goal of being, inEinsteins words, as simple as possible but no sim-pler.

The purpose of this paper is to summarize a the-oretical framework and process for understandingcomplex systems. This concept has recently beendeveloped and expanded into a book-length thesis(Gunderson and Holling 2001). In its expanded ver-sion, it provides a means of assessing informationabout the internal factors and external influencesthat interact to determine systemic sustainability.To be useful, such a framework and process mustsatisfy the following criteria:

c Be as simple as possible but no simpler than isrequired for understanding and communication.

c Be dynamic and prescriptive, not static and de-scriptive. Monitoring of the present and past isstatic unless it connects to policies and actionsand to the evaluation of different futures.

c Embrace uncertainty and unpredictability. Sur-prise and structural change are inevitable in sys-tems of people and nature.

AN INTEGRATIVE THEORYBackground

The theory was developed under the auspices of theResilience Project, a 5-year collaboration amongan international group of ecologists, economists,social scientists, and mathematicians. The projectwas initiated to search for an integrative theory andintegrative examples of practice. Its goal was todevelop and test the elements of an integrativetheory that had the degree of simplicity necessaryfor understanding but also the complexity requiredto develop policy for sustainability. The results ofthat project are summarized in the final report tothe MacArthur Foundation found at http://www.resalliance.org/reports.

The heart of the work has now been amplified inPanarchy: Understanding Transformations in Humanand Natural Systems (Gunderson and Holling. 2001).This book expands the theory and explores its im-plications for ecological, political, institutional, andmanagement systems. It was intended to deepenour understanding of linked ecological/economic/decision systems through the use of a set of inter-active models, several analyses of institutions that

Understanding Complex Systems 391

link people and nature, and an extensive explora-tion of two prototypical systems, the savannas andgrasslands of Australia and the Everglades of Flor-ida. Table 1 summarizes the books contents.

Panarchy is the term we use to describe a con-cept that explains the evolving nature of complexadaptive systems. Panarchy is the hierarchicalstructure in which systems of nature (for example,forests, grasslands, lakes, rivers, and seas), and hu-mans (for example, structures of governance, set-tlements, and cultures), as well as combined hu-mannature systems (for example, agencies thatcontrol natural resource use) (Gunderson and oth-ers 1995) and social-ecological systems (for in-stance, co-evolved systems of management) (Folkeand others 1998), are interlinked in never-endingadaptive cycles of growth, accumulation, restruc-turing, and renewal. These transformational cyclestake place in nested sets at scales ranging from a leafto the biosphere over periods from days to geologicepochs, and from the scales of a family to a socio-political region over periods from years to centuries.If we can understand these cycles and their scales, itseems possible to evaluate their contribution to sus-tainability and to identify the points at which asystem is capable of accepting positive change and

the points where it is vulnerable. It then becomespossible to use those leverage points to foster resil-ience and sustainability within a system.

The idea of panarchy combines the concept ofspace/time hierarchies with a concept of adaptivecycles. I will deal with each in turn and then showthe consequence of combining them in a synthesis.

Hierarchies

Simon (1974) was one of the first to describe theadaptive significance of hierarchical structures. Hecalled them hierarchies, but not in the sense of atop-down sequence of authoritative control.Rather, semi-autonomous levels are formed fromthe interactions among a set of variables that sharesimilar speeds (and, we would add, geometric/spa-tial attributes). Each level communicates a small setof information or quantity of material to the nexthigher (slower and coarser) level. Figure 1 showsan example for a forested landscape, Figure 2 showsa wetland system, and Figure 3 shows a social sys-tem.

As long as the transfer from one level to the otheris maintained, the interactions within the levelsthemselves can be transformed, or the variableschanged, without the whole system losing its integ-

Table 1. Table of Contents for Panarchy: Understanding Transformations in Human and Natural Systems

Part I. IntroductionChapter 1. In Quest of a Theory of Adaptive Change. C.S. Holling, L.H. Gunderson, and D. Ludwig

Part II. Theories of ChangeChapter 2. Resilience and Adaptive Cycles. C.S. Holling and L.H. GundersonChapter 3. Sustainability and Panarchies. C.S. Holling, L.H. Gunderson, and G.D. PetersonChapter 4. Why Are Systems of People and Nature not just Ecological or Social Systems? F. Westley, S.R. Carpenter,

W.A. Brock, C.S. Holling, and L.H. GundersonChapter 5. Back to the Future: Ecosystem Dynamics and Local Knowledge. F. Berkes and C. FolkeChapter 6. The Dynamics of Political Discourse in Seeking Sustainability. L. Pritchard Jr. and S.E. Sanderson

Part III. Myths, Models, and MetaphorsChapter 7. Collapse, Learning, and Renewal. S.R. Carpenter, W.A. Brock, and D. LudwigChapter 8. Dynamic Interaction of Societies and Ecosystems: Linking Theories from Ecology, Economy, and

Sociology. M. Scheffer, F. Westley, W.A. Brock, and M. HolmgrenChapter 9. A Future of Surprises. M. JanssenChapter 10. Resilience and Sustainability: The Economic Analysis of Non-Linear Dynamic Systems. W.A. Brock, K.G.

Maler, and C. PerringsPart IV. Linking Theory to Practice

Chapter 11. Resilient Rangelands Adaptation in Complex Systems. B. Walker and N. AbelChapter 12. Surprises and Sustainability Cycles of Renewal in the Everglades. L.H. Gunderson, C.S. Holling, and G.D.

PetersonChapter 13. The Devil in the Dynamics: Adaptive Management on the Front Lines. F. WestleyChapter 14. Planning for Resilience: Scenarios, Surprises, and Branch Points. G. C. Gallopin

Part V. Summary and SynthesisChapter 15. Discoveries for Sustainable Futures. C. S. Holling, S. R. Carpenter, W. A. Brock, and L. H. GundersonChapter 16. Towards an Integrative Synthesis. R. Yorque, B. Walker, C. S. Holling, L. H. Gunderson, C. Folke, S. R.

Carpenter, and W. A. Brock

392 C. S. Holling

rity. As a consequence, this structure allows widelatitude for experimentation within levels, therebygreatly increasing the speed of evolution.

Ecologists were inspired by Simons seminal arti-cle to apply the term hierarchy to ecological sys-tems and develop its significance for a variety ofecological relationships and structures. In particu-lar, Allen and Starr (1982) and ONeill and others(1986) stimulated a major expansion of theoretical

understanding by shifting attention from the small-scale view that characterized much of biologicalecology to a multiscale and landscape view thatrecognized that biotic and abiotic processes coulddevelop, mutually re-enforcing relationships overdistinct ranges of scale. More recently, Levin (1999)has expanded that representation of cross-scale dy-namics in a way that greatly deepens our under-standing of the self-organized features of terrestrialecosystems.

Simons key arguments are that each of the levelsof a dynamic hierarchy serves two functions. One isto conserve and stabilize conditions for the fasterand smaller levels; the other is to generate and testinnovations by experiments occurring within alevel. It is this latter, dynamic function we call anadaptive cycle (Holling 1986). It is a heuristicmodel, a fundamental unit that contributes to theunderstanding of the dynamics of complex systemsfrom cells, to ecosystems, to societies, to cultures.

The Adaptive Cycle

There are three properties that shape the adaptivecycle and the future state of a system:

c The inherent potential of a system that is avail-able for change, since that potential determines

Figure 1. Time and space scales of the boreal forest(Holling 1986) and the atmosphere (Clark 1985) andtheir relationship to some of the processes that structurethe forest. Contagious meso-scale processes, such as in-sect outbreaks and fire, mediate the interaction betweenfaster atmospheric processes and slower vegetation pro-cesses. (Reprinted from Gunderson and Holling 2001with permission of Island Press)

Figure 2. Time and space scales of levels of a hierarchy inthe Everglades. (Reprinted from Gunderson and Holling2001 with permission of Island Press)

Figure 3. Institutional hierarchy of rule sets. In contrastto ecological hierarchies, this hierarchy is structuredalong dimensions of the number of people involved inrule sets and approximate turnover times (Gundersonand others 1995; Westley and others 2001). (Reprintedfrom Gunderson and Holling 2001 with permission ofIsland Press)


the range of future options possible. This prop-erty can be thought of, loosely, as the wealthof a system.

c The internal controllability of a system; that is,the degree of connectedness between internalcontrolling variables and processes, a measurethat reflects the degree of flexibility or rigidity ofsuch controls, such as their sensitivity or not toperturbation.

c The adaptive capacity; that is, the resilience ofthe system, a measure of its vulnerability tounexpected or unpredictable shocks. This prop-erty can be thought of as the opposite of thevulnerability of the system.

These three propertieswealth, controllability,and adaptive capacityare general ones, whetherat the scale of the cell or the biosphere, the individ-ual or the culture. In case examples of regionaldevelopment and ecosystem management (Gun-derson and others 1995), they are the propertiesthat shape the responses of ecosystems, agencies,and people to crisis.

Potential, or wealth, sets limits for what is possi-bleit determines the number of alternative op-tions for the future. Connectedness, or controllabil-ity, determines the degree to which a system cancontrol its own destiny, as distinct from beingcaught by the whims of external variability. Resil-ience, as achieved by adaptive capacity, determineshow vulnerable the system is to unexpected distur-bances and surprises that can exceed or break thatcontrol.

A stylized representation of an adaptive cycle isshown in Figure 4 for two of these propertiespotential and connectedness. The trajectory alter-nates between long periods of slow accumulationand transformation of resources (from exploitationto conservation, or r to K), with shorter periods thatcreate opportunities for innovation (from release toreorganization, or V to a). That potential includesaccumulated ecological, economic, social, and cul-tural capital as well as unexpressed chance muta-tions and inventions. During the slow sequencefrom exploitation to conservation, connectednessand stability increase and capital is accumulated.Ecosystem capital, for example, includes nutrients,biomass, and physical structure. Although this ac-cumulated capital is sequestered for the growing,maturing ecosystem, it also represents a gradualincrease in the potential for other kinds of ecosys-tems and futures. For an economic or social system,the accumulating potential could as well derivefrom the skills, networks of human relationships,and mutual trust that are developed incrementally

and integrated during the progression from r to K.They also represent a potential that was developedand used in one setting but could be available intransformed ones.

As the progression to the K phase proceeds in anecosystem, for example, the accumulating nutrientand biomass resources become more and moretightly bound within existing vegetation, prevent-ing other competitors from utilizing them. The po-tential for other use is high, but it is expropriatedand controlled by the specific biota and processes ofthe ecosystem in place. That is, the systems con-nectedness increases, eventually becoming over-connected and increasingly rigid in its control. Itbecomes an accident waiting to happen.

The actual change is triggered by agents of dis-turbance, such as wind, fire, disease, insect out-break, and drought. The resources accumulated andsequestered in vegetation and soil are then sud-denly released and the tight organization is lost.Human enterprises can exhibit similar behavior, as,for example, when corporations such as IBM,AT&T, or General Motors accumulate rigidities tothe point of crisis and then attempt to restructure(Hurst and Zimmerman 1994; Hurst 1995; Holling

Figure 4. A stylized representation of the four ecosystemfunctions (r, K, V, a) and the flow of events among them.The arrows show the speed of the flow in the cycle. Short,closely spaced arrows indicate a slowly changing situa-tion; long arrows indicate a rapidly changing situation.The cycle reflects changes in two properties: the y axis(the potential that is inherent in the accumulated re-sources of biomass and nutrients) and the x axis (thedegree of connectedness among controlling variable). Theexit from the cycle indicated at the left of the figuresuggests, in a stylized way, the stage where the potentialcan leak away and where a flip into a less productive andless organized system is most likely (Holling 1986). (Re-printed from Gunderson and Holling 2001 with permis-sion of Island Press)

394 C. S. Holling

and others 2001). The Soviet Union is a societalexample of accumulated rigidities that precipitate asudden collapse. The proximate agents of distur-bance in these cases can be stakeholder revolts,public-interest attacks through the legal system, ormore extreme societal revolts.

The phase from V to a is a period of rapid reor-ganization during which novel recombinations canunexpectedly seed experiments that lead to inno-vations in the next cycle. The economist J. A.Schumpeter (1950) appropriately called this phasecreative destruction. Initially, the front loop ofthe trajectory, from r to K, becomes progressivelymore predictable as it develops. In contrast, theback loop of the adaptive cycle, from V to a, isinherently unpredictable and highly uncertain. Atthat stage, the previously accumulated mutations,inventions, external invaders, and capital can be-come reassorted into novel combinations, some ofwhich nucleate new opportunity.

It is as if two separate objectives are functioning,but in sequence. The first maximizes productionand accumulation; the second maximizes inventionand reassortment. The two objectives cannot bemaximized simultaneously but only occur sequen-tially. And the success in achieving one inexorablysets the stage for its opposite. The adaptive cycletherefore embraces two opposites: growth and sta-bility on the one hand, change and variety on theother.

Figure 5 adds the third dimension, resilience, tothe adaptive cycle. The appearance of a figure 8 inthe path of the adaptive cycle, as in Figure 4, is theconsequence of the projection of a three-dimen-sional object onto a two-dimensional plane. We canview that three-dimensional object from differentperspectives, emphasizing one property or another.Figure 5 rotates the object to expose the resilienceaxis.

This orientation of the figure shows that as thephases of the adaptive cycle proceed, a systemsecological resilience expands and contracts. Theconditions that occasionally foster novelty and ex-periment occur during periods in the back loop ofthe cycle, when connectedness, or controllability, islow and resilience is high (that is, during the aphase). The low connectedness, or weak control,permits novel reassortments of elements that werepreviously tightly connected to others in isolatedsets of interactions. The high resilience allows testsof those novel combinations because the system-wide costs of failure are low. The result is the con-dition needed for creative experimentation. Thisrecognition of resilience varying within a cycle addsan element that can reconcile the delicious para-

doxes of conservative nature vs creative nature;sustainability vs creative change.

The a phase is the stage that is least examinedand the least known. It is the beginning of a processof reorganization that provides the potential forsubsequent growth, resource accumulation, andstorage. At this stage, ecological resilience is high, asis potential. But connectedness is low and internalregulation is weak. There is a wide stability region,with weak regulation around equilibria, low con-nectivity among variables, and a substantialamount of potential available for future options.Because of those features, it is a fertile environmentfor experiments, for the appearance and initial es-tablishment of entities that would otherwise be out-competed. As in good experiments, many will fail,but in the process, the survivors will accumulate thefruits of change. It is a time of both crisis andopportunity.

In summary, there are four key features thatcharacterize an adaptive cycle, with its properties ofgrowth and accumulation on the one hand and of

Figure 5. Resilience is another dimension of the adaptivecycle. A third dimension, resilience, is added to the two-dimensional box of Figure 4 to show how resilience ex-pands and contracts throughout the cycle. Resilienceshrinks as the cycle moves towards K, where the systembecomes more brittle. It expands as the cycle shifts rapidlyinto a back loop to reorganize accumulated resources fora new initiation of the cycle. The appearance of a figure 8in Figure 4 is the consequence of viewing a three-dimen-sional object in a two-dimensional plane. (Reprintedfrom Gunderson and Holling 2001 with permission ofIsland Press)


novelty and renewal on the other. All of them aremeasurable in specific situations:

1. Potential (that is, wealth as expressed in eco-system structure, productivity, human relation-ships, mutations, and inventions) increases in-crementally in conjunction with increasedefficiency but also in conjunction with in-creased rigidity. This is the phase from r to K inFigure 4.

2. As potential increases, slow changes graduallyexpose an increasing vulnerability (decreasedresilience) to such threats as fire, insect out-break, competitors, or opposition groups. Thesystem becomes an accident waiting to happen.A break can trigger the release of accumulatedpotential in what the economist Schumpetercalled creative destruction (1950). The trajec-tory then moves abruptly into a back loop fromK to V.

3. Innovation occurs in pulses or surges of inno-vation when uncertainty is great, potential ishigh, and controls are weak, so that novel re-combinations can form. This is the phase ofreorganization represented in a (Figure 4)where low connectedness allows unexpectedcombinations of previously isolated or con-strained innovations that can nucleate new op-portunity.

4. Those innovations are then tested. Some fail,but others survive and adapt in a succeedingphase of growth from r to K.

Not All Adaptive Cycles Are the Same

Efforts to find exceptions that might invalidate thepreceding representation have identified differentclasses of systems that represent distinct variants of,or departures from, that cycle. Examples of theseexceptions include:

c Physical systems where the lack of inventionand mutation limits the potential for evolution-ary change. Examples: tectonic plate dynamics,and Per Baks (1996) sand pile experimentsdemonstrating organized criticality from K toV).

c Ecosystems and communities of plants and ani-mals that are strongly influenced by uncontrol-lable or unpredictable episodic external inputsand have little internal regulation and highlyadaptive responses to opportunity. Examples:exploited arid rangelands, pelagic biotic commu-nities. These systems tend to remain largely inthe lower left quadrant of the cycle, oscillating in

the a and r phases, dominated by trophic dy-namics (Walker and Abel 2001).

c Ecosystems and human organizations with pre-dictable but variable inputs and some significantinternal regulation of external variability overcertain scale ranges. For example, productivetemperate forests and grasslands, large bureau-cracies. These systems represent the full cycle ofboom-and-bust dynamics shown in Figure 4(Holling and Gunderson 2001).

c Biological entities with strong and effective ho-meostatic internal regulation of external vari-ability. Examples: cells and ionic regulation,warm-blooded organisms with endothermiccontrol of temperature. System variables remainnear an equilibrium and the individual is freedto exploit a wider range of opportunities withina community or ecosystem. This is an exampleof local control that can release external oppor-tunity and variability at a different scaleatransfer of the full adaptive cycle to the largerarena of a higher level in the hierarchy.

c Human systems with foresight and active adap-tive methods that stabilize variability and exploitopportunity. Examples: entrepreneurial busi-nesses, futures markets and resource scarcity,some traditional cultures. The high variability ofthe adaptive cycle can be transferred from thesociety to an individual entrepreneur or, in atraditional culture, to a wise person (Westleyand others 2001; Berkes and Folke 2001).

THE PANARCHY: A SYNTHESIS

Because the word hierarchy is so burdened by therigid, top-down nature of its common meaning, wedecided to look for another term that would capturethe adaptive and evolutionary nature of adaptivecycles that are nested one within each other acrossspace and time scales. Our goal was to rationalizethe interplay between change and persistence, be-tween the predictable and the unpredictable. Wetherefore melded the image of the Greek god Pan asthe epitoma of unpredictable change with the no-tion of hierarchies across scales to invent a newterm that could represent structures that sustainexperiment, test its results, and allow adaptive evo-lution. Hence, panarchy.

A panarchy is a representation of a hierarchy as anested set of adaptive cycles. The functioning ofthose cycles and the communication between themdetermines the sustainability of a system. That syn-thesis will be explored in this section.

The adaptive cycle, as shown in Figures 4 and 5,

396 C. S. Holling

transforms hierarchies from fixed static structuresto dynamic, adaptive entities whose levels are sen-sitive to small disturbances at the transition fromgrowth to collapse (the V phase) and the transitionfrom reorganization to rapid growth (the a phase).At other times, the processes are stable and robust,constraining the lower levels and immune to thebuzz of noise from small and faster processes. It is atthe two-phase transitions between gradual andrapid change and vice versa that the large and slowentities become sensitive to change from the smalland fast ones.

However, the structural, top-down aspect of hi-erarchies has tended to dominate theory and appli-cation, reinforced by the standard dictionary defi-nition of hierarchy as a system of vertical authorityand control. Therefore, the dynamic and adaptivenature of such nested structures has tended to belost.

It is certainly true that slower and larger levels setthe conditions within which faster and smaller onesfunction. Thus, a forest stand moderates the climatewithin the stand to narrow the range of tempera-tures experienced by its individuals constituents.Similarly, cultures of different people establishnorms that guide the actions of human individuals.But this representation has no way of accountingfor the dynamics of each level as symbolized in thefour-phase cycle of birth, growth and maturation,death, and renewal.

This adaptive cycle captures in a heuristic fashionthe engine that periodically generates the variabilityand novelty upon which experimentation depends.As a consequence of the periodic, but transient,phases of creative destruction (V stage) and re-newal (a stage), each level of a systems structureand processes can be reorganized. This reshufflingin the back loop of the cycle allows the possibility ofnew system configurations and opportunities utiliz-ing the exotic and entirely novel entrants that hadaccumulated in earlier phases. The adaptive cycleopens transient windows of opportunity so thatnovel assortments can be generated.

For organisms, those novel entrants are mutatedgenes or, for some bacteria, exotic genes that aretransferred occasionally between species. For eco-systems, the novel entrants are exotic, potentiallyinvasive species or species in the wings waitingfor more appropriate conditions. For economic sys-tems, these novel entrants are inventions, creativeideas, and innovative people. The adaptive cycleexplicitly initiates a slow period of growth duringwhich mutations, invasions, and inventions can ac-cumulate, followed by a briefer period when theyundergo rearrangements. This process can occur

periodically within each hierarchical level, in a waythat partially isolates the resulting experiments, re-ducing the risk to the integrity of the whole struc-ture.

The organization and functions that form biolog-ical, ecological, and human systems can thereforebe viewed as a nested set of four-phase adaptivecycles. Within these cycles, there are opportunitiesfor periodic reshuffling within levels, which main-tain adaptive opportunity, while simple interactionsacross levels maintain integrity. One major differ-ence among biological, ecological, and human sys-tems is the way that inventions are accumulatedand transferred over time. But more on that later.

There are two features that distinguish the pan-archical representation from traditional hierarchicalones. The first, as discussed earlier, is the impor-tance of the adaptive cycle and, in particular, the aphase as the engine of variety and the generator ofnew experiments within each level. The variouslevels of the panarchy can be seen as a nested set ofadaptive cycles (Figure 6).

The second feature is the connections betweenlevels. There are potentially multiple connectionsbetween phases at one level and phases at anotherlevel. But two of these connections are particularlysignificant to our search for the meaning of sustain-ability. They are labeled as revolt and remem-ber in Figure 7, where three levels of a panarchyare represented. The revolt and remember connec-tions become important at times of change in theadaptive cycles.

Figure 6. A stylized panarchy. A panarchy is a cross-scale, nested set of adaptive cycles that indicates thedynamic nature of structures depicted in the previousplots. (Reprinted from Gunderson and Holling 2001 withpermission of Island Press)


When a level in the panarchy enters its V phaseof creative destruction, the collapse can cascade tothe next larger and slower level by triggering acrisis. Such an event is most likely if the slower levelis at its K phase, because at this point the resilienceis low and the level is particularly vulnerable. Therevolt arrow in Figure 7 suggests this effect, onewhere fast and small events overwhelm slow andlarge ones. Once triggered, the effect can cascade tostill higher, slower levels, particularly if those levelshave also accumulated vulnerabilities and rigidities.

An ecological version of this situation occurswhen conditions in a forest allow a local ignition tocreate a small ground fire that spreads first to the

crown of a tree, then to a patch in the forest, andthen to a whole stand of trees. Each step in thatcascade moves the transformation to a larger andslower level. A societal version occurs when localactivists succeed in their efforts to transform re-gional organizations and institutions, because thelatter have become broadly vulnerable. Such achange occurred in New Brunswick, Canada whena few small groups opposed to spraying insecticideover the forest were able to transform this regionsvulnerable forest management policies and prac-tices (Baskerville 1995).

The arrow labeled remember in Figure 7 indi-cates a second type of cross-scale interaction that isimportant at times of change and renewal. Once acatastrophe is triggered at one level, the opportuni-ties for, or constraints against, the renewal of thecycle are strongly influenced by the K phase of thenext slower and larger level. After a forest fire, forexample, the processes and resources that have ac-cumulated at a larger level slow the leakage ofnutrients that have been mobilized and releasedinto the soil. At the same time, the options forrenewal include the seed bank, physical structures,and surviving species, which comprise biotic lega-cies (Franklin and MacMahon 2000) that have ac-cumulated in the course of the forests growth.Similarily, for its reorganization and renewal, acoral reef hit by a storm draws on its own legaciesand the memory of the seascape of which it is a part(Nystrom and Folke 2001). It is as if this connectiondraws on the accumulated wisdom and experiencesof maturity; hence, the word remember.

In a similar vein, Stewart Brand, in his marvelousmeditation on buildings (1994), described them asadaptive, hierarchical entities. Buildings of endur-ing character are a reflection of seasoned maturi-tythe culmination of a series of idiosyncratic,wise, and thought-provoking experiments in theform and content of a mature, evolved structure. InThe Clock of the Long Now, Brand (1999) extendsthese ideas and generalizes the concept of fast andslow processes to society as a whole. His work res-onates with features reminiscent of panarchy the-ory. Similarly, Levins Fragile Dominion (1999) is anaccessible and effective disquisition on self-organi-zation as it characterizes adaptive, complex ecolog-ical systems.

The panarchy is a representation of the ways inwhich a healthy social-ecological system can inventand experiment, benefiting from inventions thatcreate opportunity while it is kept safe from thosethat destabilize the system because of their natureor excessive exuberance. Each level is allowed tooperate at its own pace, protected from above by

Figure 7. Panarchical connections. Three selected levelsof a panarchy are illustrated to show the two connectionsthat are critical in creating and sustaining adaptive capa-bility. One is the revolt connection, which can cause acritical change in one cycle to cascade up to a vulnerablestage in a larger and slower one. The other is the re-member connection which facilitates renewal by draw-ing on the potential that has been accumulated andstored in a larger, slower cycle. An example of the se-quence from small and fast through larger and slower andthence to largest and slowest for a boreal forest ecosystemincludes needles, tree crowns, and patches. For institu-tions, those three speeds might be operational rules, col-lective choice rules, and constitutional rules (Ostrom1992); for economies, they might be individual prefer-ences, markets, and social institutions (Whitaker 1987);for developing nations, they might be markets, infrastruc-ture, and governance (Barro 1997); for societies, theymight be allocation mechanisms, norms, and myths(Westley 1995); for knowledge systems, they might belocal knowledge, management practice, and world view(Gadgil and others 1993; Berkes 1999; Holling and others2001). (Reprinted from Gunderson and Holling 2001with permission of Island Press)

398 C. S. Holling

slower, larger levels but invigorated from below byfaster, smaller cycles of innovation. The whole pan-archy is therefore both creative and conserving. Theinteractions between cycles in a panarchy combinelearning with continuity.

This process can serve to clarify the meaning ofsustainable development. Sustainability is the ca-pacity to create, test, and maintain adaptive capa-bility. Development is the process of creating, test-ing, and maintaining opportunity. The phrase thatcombines the two, sustainable development,therefore refers to the goal of fostering adaptivecapabilities while simultaneously creating opportu-nities. It is therefore not an oxymoron but a termthat describes a logical partnership.

Collapsing Panarchies

Stochastic events external to a cycle can triggerspasmodic collapses, particularly if they encountervulnerabilities within an adaptive cycle. Extremelylarge events can overwhelm the sustaining proper-ties of panarchies, destroying levels, and triggeringdestructive cascades down the successive levels of apanarchy. The cataclysmic loss of biological diver-sity that occurred some 65 million years ago, de-stroying about 70% of Earths species; Jablonski1995), for example, is likely to have been caused bythe impact of an asteroid (Alvarez and others 1980).That event, which may also be associated with mas-sive volcanic eruptions that occurred around thesame time, unraveled the web of interactionswithin and between panarchical levels over scalesfrom biomes to species.

Since recovery from these events is so delayed, itis likely that mass extinction events eliminate notonly species but also ecological niches. For theircontinued existence, species depend on an environ-ment that is created by life. Because they destroymost species, mass extinction events concomitantlyeliminate many ecological niches. The recovery ofbiodiversity from such cataclysmic events requiresthe reconstruction of these niches, as new speciesevolve to fill them.

Notably, different families, orders, and speciesdominated the new assemblages after recovery;novel inventions and new ways of living emerged.The dinosaurs became extinct during the collapsethat occurred 65 million years ago; the mammals,inconspicuous before that time, exploded in a di-versification that created new opportunity. Theconservative nature of established panarchies cer-tainly slows change, while at the same time accu-mulating potential that can be released periodicallyif the decks are cleared of constraining influences bylarge, extreme events.

Similarly, a long view of human history revealsnot regular change but spasmodic, catastrophic dis-ruptions followed by long periods of reinventionand development. In contrast to the sudden col-lapses of biological panarchies, there are long peri-ods of ruinous reversal, followed by slow recoveryand the restoration of lost potential. Robert Ad-amss magnificent reconstruction of Mesopotamiansocieties (1966, 1978) and a later review of otherarchaeological sequences at regional or larger scales(R. M. Adams unpublished) led him to identify twotrends in human society since the Pleistocene. Thefirst is an overall increase in the hierarchical differ-entiation and complexity of societies. That is, levelsin the panarchy are added over time. If enoughpotential accumulates at one level, it can pass athreshold and establish another, slower and largerlevel. The second trend is defined by the occurrenceof rapid discontinuous shifts, interspersed by muchlonger periods of relative stability. A number ofscholars have focused on the study of such societaldynamics in more recent history. For example,Goldstone (1991) examined the wave of revolu-tions that occurred in Eurasia after a period of calmin the 17th century. He hypothesized that politicalbreakdown occurs when there are simultaneouscrises at several different organizational levels insociety. In other words, adaptive cycles at differentlevels in a panarchy become aligned at the samephase of vulnerability. Thus, he explicitly posits acascading, panarchical collapse.

In The Great Wave, David Fischer (1996) presentsa somewhat similar model of political breakdownthat focuses less on social stratification and revolu-tionary dynamics than on empirical price data andinflation. According to Fischer, at least three wavesof social unrest swept Eurasia, first in the 14thcentury and later in the 17th and late 18th centu-ries. He argues that currency mismanagement andthe outbreak of diseases aggravated the destabiliz-ing effects of an inflation that in turn was driven bypopulation growth.

In effect, both of these models of societal changepropose that slow dynamics inform social organiza-tion. Periods of success carry the seeds of subse-quent downfall, because they allow stresses andrigidities to accumulate. Organizations and institu-tions often fail to cope with these slow changeseither because the changes are invisible to them, orthey are so complex and highly contested that noaction can be agreed upon.

Modern democratic societies are clearly vulnera-ble to the same process, but they have inventedways to diffuse large episodes of creative destruc-tion by creating smaller cycles of renewal and


change through periodic political elections. So longas there is a literate and attentive citizenry, thepainful lessons learned from the episodic collapsesof whole societal panarchies can be transferred tofaster learning at smaller scales. Various designs inbusiness, from the creation of skunk works to theintroduction of total quality management, serve thesame purpose.

Poverty Traps and Rigidity Traps

Collapsing panarchies begin to decline within spe-cific adaptive cycles that have become maladaptive.Earlier, I described the path of an adaptive cycle asoscillating between conditions of low connected-ness, low potential, and high resilience to theiropposites. Could there be systems with other com-binations of those three attributes in which variabil-ity is sharply constrained and opportunity is lim-ited? We suggest two such possibilities in Figure 8.If an adaptive cycle collapses because the potentialand diversity have been eradicated through misuse

or due to an external force, an impoverished statecan result, with low connectedness, low potential,and low resilience, thus creating a poverty trap.

This condition can then propagate downwardthrough levels of the panarchy, collapsing levels asit goes. An ecological example is the productivesavanna that, through human overuse and misuse,flips into an irreversible, eroding state, beginningwith sparse vegetation. Thereafter, subsequentdrought precipitates further erosion, and economicdisincentives maintain sheep production. The samepersistent collapse might also occur in a societytraumatized by social disruption or conflict, so thatits cultural cohesion and adaptive abilities are lost.In such a situation, the individual members of thesociety would be able to depend only on themselvesand perhaps their immediate family members.

Some such societies might continue to exist inthis degraded state of bare subsistence, barely ableto persist as a group, but unable to accumulateenough potential to form the larger structures andsustaining properties of a complete panarchy. Oth-ers might simply collapse into anarchy. Berkes(1999) and Folke and others (1998) tried to deter-mine how far such erosion must progress beforerecovery becomes impossible. When recovery ispossible, it would be useful to know what criticalattributes need to be reinvented and reestablishedfrom the residual memory stored in slowly fadingtraditions and myths to recreate a new, sustainingpanarchy.

Figure 8 also suggests that it is possible to have asustainable but maladaptive system. Imagine a sit-uation of great wealth and control, where potentialis high, connectedness great andin contrast to thephase where those conditions exist in an adaptivecycleresilience is high; that is, a wealthy, tightlyregulated, and resilient system. The high resiliencewould mean that the system had a great ability toresist external disturbances and persist, even be-yond the point where it is adaptive and creative. Itwould have a kind of perverse resilience, preservinga maladaptive system. The high potential would bemeasured in accumulated wealth or abundant nat-ural capital. The high connectedness would be cre-ated by efficient methods of social control, in whichany novelty is either smothered or its inventorejected. It would represent a rigidity trap.

We see signs of such sustained but maladaptiveconditions in great hierocracies, such as societiesthat operate under rigid and apparently immutablecaste systems. Other examples occur in regions ofthe developing world that have abundant naturalresources but are subject to the rigid control ofcorrupt political regimes. But all such systems are

Figure 8. Maladaptive systems. A poverty trap and arigidity trap are illustrated as departures from an adaptivecycle. If an adaptive cycle collapses because the potentialand diversity have been eradicated due to misuse or anexternal force, an impoverished state can result, with lowconnectedness, low potential, and low resilience, thuscreating a poverty trap. A system with high potential,connectedness, and resilience is represented by the rigid-ity trap. It is suggestive of the maladaptive conditionspresent in hierocracies, such as large bureaucracies(Holling and others 2001). (Reprinted from Gundersonand Holling 2001 with permission of Island Press)

400 C. S. Holling

likely to have the seeds of their own destructionbuilt in, as was the case with the totalitarian bu-reaucracy of the now defunct Soviet Union (Levinand others 1998).

What Distinguishes Human Systems?

Human systems exhibit at least three features thatare uniquefeatures that change the character andlocation of variability within the panarchy and thatcan dramatically enhance the potential of the pan-archies themselves. Those three features are fore-sight, communication, and technology.

Foresight and intentionality. Human foresight andintentionality can dramatically reduce or eveneliminate the boom and bust character of somecycles. Predictions of looming economic crises andcollapses caused by resource scarcity, for example,are an important issue in debates about sustainabil-ity. The economist R. Solow (1973) provided awithering critique of such doomsday scenarios,pointing out that they ignore the forward-lookingbehaviors of people. These behaviors play a role intransmitting future scarcities into current prices,thereby inducing conservation behaviors in the realeconomic world. This forward-looking processfunctions through futures markets and the strategicpurchase and holding of commodities. They providevery large incentives for some people to forecast thecoming scarcity better than the rest of the marketand to take a position to profit from it. But whatone market participant can do, all can do; thus, thisprocess transmits information to the market as awhole.

But there limits to this process, as described byCarpenter and others (1999, 2001). These limits areillustrated in specific examples of models that com-bine ecosystem simulations with economic optimi-zation and decision processes. These models suggestthat even when knowledge is total, a minimallycomplex ecosystem model, together with stochasticevents, can thwart the forward-looking economicand decision-making capacity to eliminate boomsand busts. These minimal requirements for the sys-tem are the same ones that characterize the ecosys-tem panarchythat is, at least three speeds of vari-ables, separation among those speeds, andnonlinear, multistable behavior.

That analysis is the source of our conclusion thatecosystems have a minimal complexity we call theRule of Hand whose features make linear policiesmore likely to produce temporary solutions and agreater number of escalating problems. Only anactively adaptive approach can minimize the con-sequences.

Finally, how can we explain the common ten-

dency for large organizations to develop rigidities,thus precipitating major crises that initiate restruc-turing in a larger social, ecological, economic set-ting? Or, the many examples of long-term, ruinousreversals in the development of societies? Thesecollapses seem to be more extreme and requiremuch longer recovery than the internally generatedcycles of ecosystem panarchies.

Certainly, in management agencies, the exerciseof foresight and intentionality is often brilliantlydirected to protect the positions of individualsrather than to further larger societal goals. The fore-sight that maintains creativity and change whenconnected to an appropriate economic market canlead to rigid organizations that are maintained evenwhen that particular market no longer exists. Themarket in these cases is a market for political powerof the few, not a free market for the many (Prit-chard and Sanderson 2001). Foresight and inten-tionality can therefore precipitate ruinous reversalsif they are not connected to a market with essentialliberal and equitable properties.

Communication. Organisms transfer, test, andstore experience in a changing world genetically.Ecosystems transfer, test, and store experience byforming self-organized patterns that repeat them-selves. These patterns are formed and refined by aset of interacting variables that function over spe-cific scale ranges and form a mutually reinforcingcore of relationships. In fact, an ecosystem is devel-oped out of a few such sets that establish a repro-ducing, discontinuous template to provide nichesfor species diversification and the adaptation of in-dividual organisms.

In human systems, the same self-organized pat-terns are strongly developed, but humans uniquelyadd the ability to communicate ideas and experi-ence. As they are tested, these ideas can becomeincorporated into slower parts of the panarchy,such as cultural myths, legal constitutions, andlaws. Many sources of information, including tele-vision, movies, and the Internet, are global in theirconnectedness and influence. These media are con-tributing to a transformation of culture, beliefs, andpolitics at global scales.

Technology. The scale of the influence exertedby every animal other than humans is highly re-stricted. But technology amplifies the actions ofhumans so that they affect an astonishing range ofscales from the submicroscopic to global andhowever modestly at the momenteven extendbeyond Earth itself.

As human technology has evolved over the lasthundred thousand years, it has progressively accel-erated, changing the rules and context of the pan-


archies in the process. The specialized tools, habita-tion, and weapons of hunter-gatherers, forexample, together with the domestication of ca-nines for use as hunting companions, created op-portunities over wide scales. The use of fire by earlyhumans made them part of the ecological structur-ing process. In temperate North America and Aus-tralia, for example, they became capable of trans-forming mosaics of grasslands and woods intoextensive regions of contiguous grasslands or for-ests (Flannery 1994).

Progressively, the horse, train, automobile, andaircraft have extended the ambit for human choicesfrom local to regional and thence to planetaryscales, but the time allotted for each of these choiceshas changed little, or even decreased. Trips betweenhome and work, for example, have always beenlargely limited to less than an hour or so, althoughthe spatial scale has expanded from a maximum ofa few kilometers by foot to potentially a few hun-dred kilometers by commuter aircraft. The slope ofthe decision panarchy for humans, if plotted in thesame space as in Figures 13, now angles sharplyupward, intersecting and dominating other panar-chies of nature.

Assessing Sustainability

The current state of our understanding of panar-chies is summarized in Table 2. The theory is suffi-ciently new that its practical application to regionalquestions or the analysis of specific problems hasjust begun. Panarchy theory focuses on the criticalfeatures that affect or trigger reorganization andtransformation in a system. First, the back-loop ofthe cycles is the phase where resilience and oppor-tunity is maintained or created, via release andreorganization (Figures 4 and 5). Second, theconnections between levels of the panarchy arewhere persistence (via remembrance) and evolv-ability (via revolt) (Figure 7) are maintained.

These four phases or processes make up the fourRs of sustainability and development: release, re-organization, remembrance, and revolt. They pro-vide new categories that can be used to organize themore specific indicators and attributes discussed indocuments aimed at finding ways to evaluate sus-tainability and development.

To summarize: The panarchy describes how ahealthy socioecological system can invent and ex-periment, benefiting from inventions that createopportunity while it is kept safe from those thatdestabilize the system due to their nature or exces-sive exuberance. Each level is allowed to operate atits own pace, protected from above by slower, largerlevels but invigorated from below by faster, smaller

cycles of innovation. The whole panarchy is there-fore both creative and conserving. The interactionsbetween cycles in a panarchy combines learningwith continuity.

The four Rs, then, represent the critical processesthat manage the balance and tension betweenchange and sustainability.

It is often useful to begin the analysis of a specificproblem with a historical reconstruction of theevents that have occurred, focusing on the surprisesand crises that have arisen as a result of both ex-ternal influences and internal instabilities. In es-sence, a sequence of adaptive cycles can be de-scribed, for the so-called natural system, theeconomy, management agencies, users, and poli-tics. We think it is necessary to consider three scaleranges for each system, although the particularscales might be different for different subsystems.One of the principal aims is to define where in theirrespective adaptive cycles each of the subsystems isnow. Actions that would be appropriate at onephase of the cycle might not be appropriate at otherphases. Knowing where you are helps you to definewhat action needs to be taken.

In many instances, the motive for an assessmentis a crisis or transformation that has already oc-curred or is anticipated. In these situations, theconditions of the back loop of the adaptive cycle(Figure 4) dominate. However, it is these times ofgreatest threat that offer the greatest opportunity,because many constraints have been removed. Inan insightful analysis of local communities as seenfrom this perspective, Berkes and Folke (2001)showed that local societies often develop reservesthat are necessary during back-loop restructuring.In the same book, Westley (2001) presented anequally incisive analysis of a sequence of decisionsand actions taken in specific examples of problemsolving by a resource manager. Figure 9 provides anexample of the kind of analysis that is possible.

Such transformations across scales are qualita-tively different from the incremental changes thatoccur during the growth phase of the adaptive cy-cle. They are also qualitatively different from thepotentially more extreme changes and frozen acci-dents that can occur during the more revolutionaryshift from creative destruction (V) to renewal (a).These transformations cascade and transform thewhole panarchy along with its constituent adaptivecycles.

Because a unique combination of separate devel-opments has to conspire to occur simultaneously,extreme events are rare. Some developmentsemerge within adaptive cycles during the back loopof the cycle, when recombinations and external

402 C. S. Holling

Table 2. Summary Findings from the Assessment of Resilience in Ecosystems, Economies, andInstitutions

Statement Brief Explanation

Multistable states are common in many systems. Abrupt shifts among a multiplicity of very different stabledomains are plausible in regional ecosystems, someeconomic systems, and some political systems.

The adaptive cycle is a fundamental unit of dynamicchange.

An adaptive cycle that aggregates resources and thatperiodically restructures to create opportunities forinnovation is a fundamental unit for understanding complexsystems, from cells to ecosystems to societies to cultures.

Not all adaptive cycles are the same and some aremaladaptive.

Variants to the adaptive cycle are present in different systems.These include physical systems (because of the absence ofmutations of elements), ecosystems strongly influenced byexternal pulses, and human systems with foresight andadaptive methods to stabilize variability. Some systems aremaladaptive and trigger poverty and rigidity traps.

Sustainability requires both change and persistence. We propose that sustainability is maintained by relationshipsthat can be interpreted as a nested set of adaptive cyclesarranged as a dynamic hierarchy in space and timethepanarchy.

Self-organization shapes long-term change. Self-organization of ecological systems establishes the arenafor evolutionary change. Self-organization of humaninstitutional patterns establishes the arena for futuresustainable opportunity.

There are three types of learning. Panarchies identify three types of change, each of which cangenerate a different kind of learning: (a) incremental (r toK, Figure 4), (b) lurching, (V to a, Figure 4), and (c)transforming.

The world is lumpy. Attributes of biological and human entities form clumpedpatterns that reflect panarchical organization, creatediversity, and contribute to resilience and sustainability.

Functional diversity builds resilience. Functional groups across size classes of organisms maintainecosystem resilience.

Tractability comes from a Rule of Hand. The minimal complexity needed to understand a panarchyand its adaptive cycles requires at least three to five keyinteracting components, three qualitatively different speeds,nonlinear causation. Vulnerability and resilience changewith the slow variables; spatial contagion and biotic legaciesgenerate self-organized patterns over scales in space andtime.

Emergent behavior emerges from integrated systems. Linked ecological, economic, and social systems can behavedifferently from their parts. Integrated systems exhibitemergent behavior if they have strong connectivity betweenthe human and ecological components and if they have keycharacteristics of nonlinearity and complexity as suggestedin the Rule of Hand.

Management must take surprise and unpredictabilityinto consideration.

Managing complex systems requires confronting multipleuncertainties. These can arise from technical considerations,such as models or analytic frameworks. The examplessuggest that as much complexity exists in the socialdimensions as in the ecological ones and that managersmust juggle shifting objectives.

Is adaptive management an answer? For linked ecological/social/economic systems, slow variables,multistable behaviors, and stochasticity cause active adaptivemanagement to outperform optimization approaches thatseek stable targets.

Reprinted from Gunderson and Holling 2001 with permission of Island Press


influences can generate unexpected new seeds ofopportunity that can nucleate and modify the sub-sequent phase of growth. So long as connectionsare maintained with other levels, those innovationsare contained and do not propagate to other levels.

But if these recombinations and inventions accu-mulate independently in a number of adjacent lev-els, a time will come when the phases of severalneighboring cycles become coincident, and eachbecomes poised as an accident waiting to happen ina shift from V to a. Windows open that can thenallow those independent inventions and adapta-tions to interact, producing a cascade of novel self-organized patterns across a panarchy and creatingfundamental new opportunity. There is an align-ment of the stars. Such a coincidence in phases ofvulnerability at multiple scales is quite rare. That is,true revolutionary transformations are rare,whether in systems of people or systems in nature.

Under conditions of crisis in a region, the ele-ments of a prescription for facilitating constructivechange are as follows:

c Identify and reduce destructive constraints and in-hibitions on change, such as perverse subsidies.

c Protect and preserve the accumulated experi-ence on which change will be based.

c Stimulate innovation and communicate the re-sults in a variety of fail-safe experiments de-signed to probe possible directions in a way thatis low in costs in terms of human careers andorganizational budgets.

c Encourage new foundations for renewal thatbuild and sustain the capacity of people, econo-mies, and nature to deal with change.

c Encourage programs to expand an understand-ing of change and communicate it to citizens,businesses, and people at different levels of ad-ministration and governance, engaging them inthe process of change.

A principal conclusion from the Resilience Projectis that the era of ecosystem management via incre-mental increases in efficiency is over. We are nowin an era of transformation, in which ecosystemmanagement must build and maintain ecologicalresilience as well as the social flexibility needed tocope, innovate, and adapt.

ACKNOWLEDGMENTS

This paper draws extensively on Gunderson andHolling, (2001), particularly the three chapters I wrotewith my colleagues Buz Brock, Steve Carpenter,Lance Gunderson, and Garry Peterson. Along withCarl Folke and Brian Walker, they served as the co-organizers of the Resilience Project, which providedthe crucible for these ideas, models, and examples. Ithank these marvelous friends and collaborators fortheir contributions, their inspiration, and their imag-ination. But all of the authors who contributed to thebook were equally important to the development ofthese ideas. They comprised an international group ofecologists, economists, social scientists, and mathema-ticians whose depth of knowledge in their respectiveareas helped to produce a very real synthesis. In ad-dition to those just mentioned above, they includeNick Abel, Fikret Berkes, Pille Bunnell, Gilberto Gal-lopin, Milena Holmgren, Marco Janssen, Don Lud-wig, Karl-Goran Maler, Charles Perrings, Rusty Prit-chard, Steve Sanderson, Marten Scheffer, FrancesWestley, and Ralf Yorque. Over the 5 years of theproject, we became the best of friends and collabora-tors. Finally, I thank the MacArthur Foundation forthe support of a grant and Dan Martin of theMacArthur Foundation for his sustained advice andhelp.

REFERENCES

Adams RM. 1966. The evolution of urban society: early Meso-potamia and Prehispanic Mexico. Chicago: Aldine.

Adams RM. 1978. Strategies of maximization, stability and re-silience in Mesopotamian society, agriculture and settlement.Proc Am Philos Soc 122:32935.

Allen TFH, Starr TB. 1982. Hierarchy: perspectives for ecologicalcomplexity. Chicago: University of Chicago Press.

Alvarez LW, Alvarez W, Asaro F, Michel HV. 1980. Extraterres-trial cause for the Cretaceous-Tertiary extinction. Science 208:1095108.

Figure 9. Separate adaptive cycles are used to depictphases of issues as interpreted in four systemspolitical,organizational, interorganizational, and individual. Man-agers actions and solutions must account for these dy-namics of these systems (Westley 2001). (Reprinted fromGunderson and Holling 2001 with permission of IslandPress)

404 C. S. Holling

Bak P. 1996. How nature works: the science of self-organizedcriticality. New York: Copernicus Press, Springer-Verlag.

Barro R. 1997. Determinants of economic growth: a cross coun-try empirical comparison. Cambridge (MA): MIT Press.

Baskerville G. 1995. The forestry problem: adaptive lurches ofrenewal. In: Gunderson LH, Holling CS, Light SS, editors.Barriers and bridges to the renewal of ecosystems and institu-tions. New York: Columbia University Press.

Berkes F. 1999. Sacred ecology: traditional ecological knowledgeand management systems. Philadelphia and London: Taylor &Frances.

Berkes F, Folke C. 2001. Back to the future: ecosystem dynamicsand local knowledge. In: Gunderson L, Holling CS, editors.Panarchy: understanding transformations in human and nat-ural systems. Washington (DC): Island Press. in press

Brand S. 1999. The clock of the long now. New York: Basic Books.

Brand S. 1994. How buildings learn. New York: Viking Penguin.

Brock W. 1972. On models of expectations that arise from max-imizing behavior of economic agents over time. J Eco Theory5:34876.

Carpenter SR, Brock WA, Hanson P. 1999. Ecological and socialdynamics in simple models of ecosystem management. Con-serv Ecol 3:4. [online] URL: http://www.consecol.org/vol3/iss2/art4

Carpenter SR, Brock WA, Ludwig D. 2001. Collapse, learningand renewal. In: Gunderson L, Holling CS, editors. Panarchy:understanding transformations in human and natural systems.Washington (DC): Island Press. in press

Clark WC. 1985. Scales of climate impacts. Clim Change 7:527.

Fischer DH. 1996. Great wave: price revolutions and the rhythmof history. New York: Oxford University Press.

Flannery T. 1994. The future eaters: an ecological history of theAustralasian lands and people. New York: Braziller.

Folke C, Berkes F, Colding J. 1998. Ecological practices and socialmechanisms for building resilience and sustainability. In:Berkes F, Folke C, editors. Cambridge (UK): Cambridge Uni-versity Press. in press

Franklin JF, MacMahon JA. 2000. Ecology: messages from amountain. Science 288:11835.

Gadgil M, Berkes F, Folke C. 1993. Indigenous knowledge forbiodiversity conservation. Ambio 22:1516.

Goldstone J. 1991. Revolution and rebellion in the early modernworld. Berkeley (CA): University of California Press.

Gunderson L, Holling CS, editors. 2001. Panarchy: understand-ing transformations in human and natural systems. Washing-ton (DC): Island Press.

Gunderson LH, Holling CS, Light SS. 1995. Barriers and bridgesto the renewal of ecosystems and institutions. New York:Columbia University Press.

Holling CS. 1986. Resilience of ecosystems; local surprise andglobal change. In: Clark WC, Munn RE, editors. Sustainabledevelopment of the biosphere. Cambridge (UK): CambridgeUniversity Press. 292317. Reprinted in: Cleveland C,Costanza R, Perrings C, editors. 1997. The development ofecological economics. Brookfields: Edward Elgar Publishing.

Holling CS. 2000. Theories for sustainable futures. Conserv Ecol4:7. [online] URL:http://www.consecol.org/vol4/iss2/art7

Holling CS, Gunderson LH. 2001. Resilience and adaptive cycles.In: Gunderson L, Holling CS, editors. Panarchy: understanding

transformations in human and natural systems. Washington(DC): Island Press. in press.

Holling CS, Gunderson LH, Peterson GD. 2001. Sustainabilityand panarchies. In: Gunderson L, Holling CS, editors. Panar-chy: understanding transformations in human and naturalsystems. Washington (DC): Island Press. in press

Hurst DK. 1995. Crisis and renewal. Boston (MA): HarvardBusiness School Press.

Hurst DK, Zimmerman BJ. 1994. From life cycle to ecocycle.Manage Inquiry 3:33954.

Jablonski D. 1995. Extinction in the fossil record. In: Lawton JH,May RM, editors. Extinction rates. Oxford (UK): Oxford Uni-versity Press. pp. 2544.

Levin S. 1999. Fragile dominion. Reading (MA): Perseus Books.

Levin SA, Barrett S, Aniyar S, Baumol W, Bliss C, Bolin B,Dasgupta P, Ehrlich P, Folke C, Gren IM, and others. 1998.Resilience in natural and socioeconomic systems. Environ DevEcon 3:22235.

Nystrom M, Folke C. 2001. Spatial resilience of coral reefs.Ecosystems. 4:406417.

ONeill RV, DeAngelis DL, Waide JB, Allen TFH. 1986. A hier-archical concept of ecosystems. Princeton (NJ): Princeton Uni-versity Press.

Ostrom E. 1992. Crafting institutions for self-governing irrigationsystems. San Francisco: Institute for Contemporary Studies Press.

Pritchard L Jr, Sanderson SE. 2001. The dynamics of politicaldiscourse in seeking sustainability. In: Gunderson L, HollingCS, editors. Panarchy: understanding transformations in hu-man and natural systems. Washington (DC): Island Press.

Roe E. 1998. Taking complexity seriously: policy analysis, trian-gulation and sustainable development. Boston (MA): KluwerAcademic Publishers.

Schumpeter JA. 1950. Capitalism, socialism and democracy.New York: Harper & Row.

Simon HA. 1974. The organization of complex systems. In: HHPattee, editor. Hierarchy theory: the challenge of complexsystems. New York: Braziller. pp. 327.

Solow R. 1973. Is the end of the world at hand? ChallengeMarch/April:3950.

Walker B, Abel N. 2001. Resilient rangelands: adaptation incomplex systems. In: Gunderson L, Holling CS, editors. Pan-archy: understanding transformations in human and naturalsystems. Washington (DC): Island Press. in press

Westley F. 2001. The devil in the dynamics: adaptive manage-ment on the front lines. In: Gunderson L, Holling CS, editors.Panarchy: understanding transformations in human and nat-ural systems. Washington (DC): Island Press. in press

Westley F. 1995. Governing design: the management of socialsystems and ecosystems management. In: Gunderson LH,Holling CS, Light SS, editors. Barriers and bridges to the re-newal of ecosystems and institutions. New York: ColumbiaUniversity Press. in press

Westley F, Carpenter SR, Brock WA, Holling CS, Gunderson LH.2001. Why are systems of people and nature not just ecolog-ical or social systems? In: Gunderson L, Holling CS, editors.Panarchy: understanding transformations in human and nat-ural systems. Washington (DC): Island Press. in press

Whitaker JK. 1987. Alfred Marshall. In: Eatwell J, Milgate M,Newman P, editors. The new Palgrave: a dictionary of eco-nomics. London: MacMillan Press.


Documents

Panarchy Holling