Coupled and complex: Human–environment interaction in the Greater Yellowstone Ecosystem, USA

Available online at www.sciencedirect.com

Geoforum 39 (2008) 833–845 www.elsevier.com/locate/geoforum

Coupled and complex: Human–environment interaction in the Greater Yellowstone Ecosystem, USA David Bennett b

a,*

, David McGinnis

b

a 316 Jessup Hall, Department of Geography, The University of Iowa, Iowa City, IA 52242, USA 1500 University Avenue, Grants and Sponsored Programs, Montana State University-Billings, Billings, MT 59101, USA

Received 27 April 2006; received in revised form 13 April 2007

Abstract Complexity theory has received considerable attention over the past decade from a wide variety of disciplines. Some who write on this topic suggest that complexity theory will lead to a unifying understanding of complex phenomena; others dismiss it as a passing and disruptive fad. We suggest that for the analysis of coupled natural/human systems, the truth emerges from the middle ground. As an approach focused as much on the connections among system elements as the elements themselves, we argue that complexity theory provides a useful conceptual framework for the study of coupled natural/human systems. It is, if nothing else, a framework that leads us to ask interesting questions about, for example, sustainability, resilience, threshold events, and predictability. In this paper we attempt to demystify the ongoing discussions on complexity theory by linking its evocative and overloaded terminology to real-world processes. We illustrate how a shift in focus from system elements to connections among elements can lead to meaningful insight into human–environment interactions that might otherwise be overlooked. We ground our discussion in ongoing interdisciplinary research surrounding Yellowstone National Park’s northern elk winter range; a tightly coupled natural/human system that has been the center of debate, conflict, and compromise for more than 135 years. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Complexity theory; Coupled natural/human systems; Yellowstone National Park

1. Introduction What we know of the world around us is, in large measure, the product of reductionist science. The basic tenets of this approach tell us that truth can be found through an understanding of individual system components; a system is the sum of its parts. While this approach served us well through most of the previous century, many scientists now believe that a reductionist approach alone is insufficient for the study of natural and social systems. These scientists note that social and natural processes are often driven by subsystem interactions and feedback mechanisms and, thus, system-wide behavior cannot be understood by *

Corresponding author. E-mail addresses: [email protected] (D. Bennett), dmcginnis@ msubillings.edu (D. McGinnis). 0016-7185/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.geoforum.2007.05.009

analyzing system components in isolation; a system can be more than the sum of its parts. A focus on the interactions that exist among system components and the resulting feedback mechanisms is a hallmark of complexity scientists who, metaphorically speaking, study how complicated puzzles fit together and, importantly, attempt to understand how the coupling of seemingly unrelated pieces produce system-level patterns and behaviors. A science focused on interactions and feedbacks seems particularly appropriate for scientific inquiry into how humans are coupled to the natural environments in which they are situated—particularly when reductionist science has provided insight into how the individual pieces of these complex puzzles operate. Some, however, criticize complexity theory as being too immature and ill-defined to be of general utility in the social sciences (Johnson and Burton, 1994). Others find the dichotomy between the

834

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845

reductionism of traditional scientific method and the holism of complexity to be a false one. These writers note that you cannot study interaction without decomposing the whole into its constituent parts (a reductionist approach) and a consideration of interaction is not the sole domain of complexity theorists (Horgan, 1995). Yet scientists from a broad range of disciplines have found complexity theory compelling because it offers a unique epistemological perspective for scientific inquiry (O’Sullivan, 2004); a conceptual point of departure from which the connectedness of those components that define the larger whole can be explored. In this paper we explore the applicability of complexity theory to the study of coupled human and natural systems. We present a generalizable framework for landscape-scale research, discuss the importance of interdisciplinary perspectives, and suggest that the study of biocomplex couplings among natural and human systems represents a logical extension of Geography’s human–environment tradition. Note that it is not our intent to make grand contributions to the philosophical discussion of what ‘‘complexity theory’’ is or is not, or to debate its utility to the social sciences. We refer the reader to the literature for this ongoing discussion (Cilliers, 1998; Levin, 1998; Manson, 2001; Lansing, 2003; Reitsma, 2003; O’Sullivan, 2004; Manson and O’Sullivan, 2006; Portugali, 2006). Our goal here is more pragmatic: To render the generalities of complexity theory more concrete by placing them into a specific coupled natural/human system. We accept complexity theory as a useful conceptual framework that provides insight into complex and adaptive spatial systems (CASS) and use it to help develop theories and hypotheses about how humans and natural systems operate. To provide context to this discussion we present a case study situated in the Yellowstone National Park (YNP) USA northern elk winter range (NEWR) and its surrounding environs. The remainder of this paper is organized as follows. First, we describe the problem domain that we study; nature/human interactions in and around the NEWR. In Section 3 we define complexity in the context of this paper and then illustrate its applicability to the NEWR in Section 4. In Section 5 we discuss the utility of complexity science in theory and practice and provide concluding remarks Section 6.

encourage migratory behavior. The elk responded to reduced human predation and population levels climbed rapidly to a mid 1990s high of 19,000. Wolves were reintroduced to YNP in 1995–1996. Since 1996 the elk population has declined by approximately 50% (MFWP, 2005). Over the same time period (1996–2005) riparian vegetation began to grow to heights not seen for 100 years (NRC, 2002), beaver began to recolonize Yellowstone National Park, land use patterns began to change on the privately owned property within the region (Parmenter et al., 2003), and the number of elk moving out of the park during the winter migration season steadily increased (Lemke et al., 1998; Lemke, 2005; MFWP, 2005). Many residents and researchers quickly attributed changes in ecosystem structure to the reintroduction of the wolves. While this kind of top-down trophic control of ecosystem dynamics does seem to be operating (Ripple et al., 2001; Fortin et al., 2005), we suggest that this intricately coupled natural/human system cannot be fully understood by studying system components in isolation or in pairs. Furthermore, we argue that many of the basic concepts associated with complexity theory provide useful insight into the dynamics of this system. As illustrated in Fig. 1, the park’s northern boundary cuts across the NEWR and, thus, a patchwork of ownership patterns and management strategies has been overlaid onto what was once a highly integrated ecosystem. Controversy and conflict about the management of public and private land in this region has been a constant theme over the park’s long history (Pritchard, 1999). Yellowstone’s northern range elk population is often at the center of this conflict as different interest groups argue about what constitutes an appropriate and sustainable elk herd size. Elk have been viewed as self-regulating elements within the regional ecosystem, negative externalities to be minimized, common pool resources to be exploited, and amenities to be acquired. The elk population is impacted not only by the park’s emphasis on sustainable natural processes but also by land use decisions made by private and public deci-

2. Context Yellowstone’s NEWR has experienced significant and continual change since the park’s inception in 1872. Over the past 40 years this change has seemed particularly dramatic. In 1967 the northern range elk (Cervus) herd population was estimated to be 3172, a historic low. In 1968 Yellowstone National Park adopted a policy of ‘‘natural regulation’’ and, as a result, elk were no longer culled to maintain herd size at prescribed levels. At about the same time the Montana Department of Fish, Wildlife and Parks modified hunting regulations to reduce total harvest and

Fig. 1. YNP study region and the Northern Elk Winter Range (NEWR).

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845

sion-makers outside of the park who manage the landscape to achieve multiple and sometimes very different objectives (e.g., USFWS, USFS, Montana Fish, Wildlife and Parks, ranchers, and amenity buyers); objectives that vary by agency and individual, and across space and time. Collectively these land use decisions translate into a continually changing landscape comprised of opportunities (e.g., accessible forage) and threats (e.g., predation) through which elk must navigate. Superimposed onto these local decisions about land use and land cover are the effect of national and international-scale social, economic, and biophysical processes, including: (1) agricultural economics; (2) the demand for ‘‘amenity property’’; (3) the international significance of Yellowstone National Park; and (4) global climate change. The system is driven by processes occurring at multiple, interacting scales. 2.1. Case study goals and objectives In the fall of 2002 an interdisciplinary team was formed to broaden the debate on elk ecology in the NEWR by more explicitly considering the consequences (intended and unintended) of human decision-making. This team was comprised of individuals with expertise in human geography, environmental history, economics, riparian and elk ecology, climatology, environmental simulation, and geographic information science. The conceptual framework produced by this group is presented in Fig. 2. It is assumed that land managers make decisions designed to produce a particular set of ecosystem services (e.g., the production/maintenance of cattle, elk, clean water, or scenic vistas) within the context of cultural, economic, biophysical, and regulatory constraints. Land use decisions change land cover that, in turn, create

Endogenous (e.g. LU policy) and exogenous (e.g. climate) drivers affect LULC decisions a

Human subsystems affect the biophysical sub-systems a LULC decisions affect the productions of ecosystem services

835

changes in the set of ecosystem services produced. Those who are sufficiently dissatisfied with the status quo will attempt to effect changes in the set of constraints that led to the production of those services (e.g., change in public policy through political processes). We focus our analysis on the impact of human decision-making on the number and spatial distribution of elk because of the important but changing role that elk play in the region and their connection to both human and natural processes, but the framework is generally applicable to the production of ecosystem services broadly defined. To better understand the processes that affect the regional elk population we set forth the following research objectives: 1. Assess the knowledge/belief systems of NEWR stakeholders with respect to environmental change and ecosystem service and relate these systems to socioeconomic characteristics and stakeholder identity. 2. Improve and expand the empirical record associated with land use/land cover change and use this record to model the impact of residential development and other human activities on the NEWR. 3. Improve and extend individual-based models that represent large-mammal behavior. 4. Develop plausible climate scenarios that illustrate how the NEWR may be impacted by global climate change/variability. 5. Use knowledge gleaned from 1 to 4 to model decisionmaking using quantitative (intelligent agents) and qualitative (scenario analyses) methods and to merge biophysical and decision-making models into an ecosystem model that allows us to explore alternative future scenarios for the NEWR.

Desire for change affects policy through political processes

Policy makers State changes within human subsystems Land Managers Landscape Biological State changes within biophysical systems

Physical

a

Biophysical subsystems affect the human subsystems a Produced ecosystem services affects satisfaction and desire for change.

Fig. 2. Conceptual framework for interdisciplinary study of coupled natural/human systems in and around Yellowstone’s NEWR.

836

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845

As indicated by Fig. 2, inter- and intra- system interactions and feedbacks are viewed as key drivers of landscape dynamics and, thus, complexity theory provides a useful organizational framework within which to integrate the seemingly disparate elements of this study.

3. Complexity defined There is considerable debate about what is meant by ‘‘complexity theory’’ (see, e.g., Horgan, 1995; Manson, 2003; Reitsma, 2003). Perhaps this is to be expected from a term attached to such divergent problems as computational intractability (e.g., solving np-hard problems, Flum et al., 2006), the development of soil morphology (Phillips, 1999), fisheries management (Carpenter and Brock, 2004), and international relations (Axelrod, 1997). Connecting all related concepts into a grand unifying theory of complexity that has meaningful explanatory or predictive power is probably not a realistic expectation (Cilliers, 1998; Manson and O’Sullivan, 2006). Manson (2001) attempts to provide structure to the various concepts associated with complexity theory by placing related research into one of three classes: algorithmic, deterministic, and aggregate. Studies of social and ecosystem dynamics that claim a theoretic link to the complexity sciences (and some that do not) are typically studies of what Manson would class as aggregate complexity (we refer the reader to Manson (2003) for a discussion of algorithmic and deterministic complexity). There is considerable overlap between Manson’s aggregate complexity and Holland (1995) complex adaptive systems (CAS). Both authors speak to the importance of connectivity among system components, feedback processes that produce non-linear behavior, self-organization, and emergent properties. They do not, however, use consistent ter-

minology when describing these properties and processes. The lack of a common lexicon for use by those who characterize their work as complexity science causes confusion and impairs communication (Phillips, 1999; Reitsma, 2003; O’Sullivan, 2004). Different terms have been used in the literature to describe the same process, and different processes have been described using the same term (Phillips, 1999). Furthermore, as O’Sullivan (2004) notes, the words used to describe complexity theory are often evocative, overloaded terms that carry considerable common use baggage. The problem with such words, of course, is that they do not necessarily evoke the same image in everyone’s mind. Until the science matures and, dare we say, self organizes into more distinct lines of inquiry it will be important for authors to clearly state what they mean when they use the terminology associated with complexity theory. In Table 1 we present a glossary to define what we mean in the context of this study when we refer to properties commonly used to define complex systems and the theories that explain them. For a broader discussion of complexity theory we refer the reader to the many excellent review articles available in the literature (e.g., Manson, 2001; O’Sullivan, 2004). The properties of non-linearity, feedback mechanisms, path dependence, emergent behavior, and cross-scale connections can have significant ramifications for science and management. If, for example, we accept an explanation based on complexity then we must also accept that an existing system state is just one realization of many possible alternative states. A realization produced, perhaps, by a seemingly insignificant event that starts a system down a new and unexpected path or pushes a system past some critical threshold to a new metastable state. Complexity suggests, therefore, that the same underlying processes can produce many different patterns (what Brown et al., 2006 refer to as the predictability problem).

Table 1 Glossary of terms Interaction

Elements that comprise a system affect the state of other elements (or themselves)

Feedback

Interaction produces feedback where the current state of a system directly or indirectly affects its own future state (e.g., population growth, predator/prey relationships, social organization). Feedback mechanisms lead to non-linear processes where the rate of change associated with some state variable(s) is not constant through time (e.g., populations unconstrained by resource limitations, the spread of infectious diseases, the rise and subsequent collapse of a culture). Interaction and feedback processes lead to adaptive changes in structure or behavior (e.g., elk adapt their spatial and temporal behavior in response to interaction with predators). Interaction, adaptation and feedback cause elements within a complex adaptive system (CAS) to form organized structure or behavior, the system becomes more ordered as patterns emerge. Organization comes at a cost. Maintaining a system in a highly ordered, low entropy state requires an input of energy. The theory of self-organized criticality (SOC) suggests that a system will accumulate only so much energy before intra-system stress causes its organizational structure to collapse; an energy dissipating event occurs (Bak, 1996; Jensen, 1998). Systems at these threshold states between order and disorder, low entropy and high entropy, are said to be ‘‘far from equilibrium’’ and ‘‘at the edge of chaos’’. The current state of the system is dependent on specific events that occurred in the past. Path dependency means that today’s decisions limit future opportunities (history matters)—outcomes predicated on the ‘‘paths not taken’’ may become very expensive to produce. Interaction, feedbacks, adaptation, self-organization, and path dependency lead to the development of system-level patterns and behaviors that could not have been predicted from the analysis of individual elements in isolation.

Non-linearity

Adaptation Self-organization Self-organized criticality

Path dependency

Emergence

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845

Models or theories that fail to reproduce existing states are not, therefore, necessarily in error. Similarly, the concept of equifinality suggests that many different processes can produce the same pattern. This implies that models or theories that mimic real-world patterns are not necessarily valid. The difficulties associated with predicting the future state of complex systems have led some to refer to this as the science of surprise (Casti, 1996) and attribute almost mystical qualities to the phenomenon of emergence. Perhaps a more productive response is to recognize that the dynamics of CASS cannot be explained by deterministic representations of process. Rather, a population of alterative system states, each element of which is a plausibly outcome given the processes of adaptation, bifurcation and path dependence, and self-organization, must be explored. A failure to consider such states may well produce the kinds of unexpected and unintended consequences attributed to emergence (Robbins, 2001). In the following section we illustrate how these commonly accepted properties of complex systems do, or do not, provide insight into processes that drive the study area. 4. Complexity applied We now use the NEWR area as a case study to illustrate how the characteristics that define complexity theory apply to real-world systems. We link specific characteristics with specific processes but, just like complexity itself, these characteristics and processes are interrelated. Emergent behaviors, for example, can reinforce path dependency, and interaction can lead to criticalities. The conceptual diagram represented in Fig. 2 will be used as an organizational device to facilitate this discussion; beginning in the upper left-hand corner and working counterclockwise. Please note that our intent is not to provide a detailed discussion on the various research elements that comprise the larger project; we leave such discussions to the individual members of the research team in separate publications. 4.1. Endogenous and exogenous drivers of land use/land cover change: feedbacks and non-linearities This research is based on the premise that decisionmakers in rural landscapes are driven by a desire to produce particular kinds of ecosystem services given an existing cultural, political, economic, and biophysical context. Furthermore, they attempt to produce these services within a decision-framework (e.g., knowledge, analytical tools, and data) that is incomplete and imperfect. Stated differently, boundedly rational decisions (Simon, 1957) are made in an attempt to ‘‘optimize’’ a multi-objective function subject to numerous and significant social, economic, and biophysical constraints. We place optimize in quotes because it should be understood that not all objectives or constraints can be fully defined, quantified and incorporated into traditional optimization algorithms. Furthermore, it should be noted that objectives often con-

837

flict and stakeholders often disagree on the relative importance of alternative objectives (i.e., an optimum optimorum may not exist). The production of desired ecosystem services through collective and individual actions is a daunting task indeed, but does this necessarily lead to complexity? To address this question we consider a set of ongoing processes in and around the NEWR that push decision-makers away from the production of cattle as an ecosystem service and toward the production of other environmental amenities. Let us begin by considering the set of objectives that impact the decision-making processes of ranchers. Maximizing economic return is certainly important, but interviews conducted by team members indicate that it is not the only objective that ranchers are concerned about (Haggerty, 2004). Perpetuating a way of life, preserving options for the next generation, staying on the land, and maintaining social standing are also important objectives. Efforts to meet these objectives produce a negative feedback mechanism that helps to maintain the landscape in a familiar state. Exogenous forces, however, also push toward change and the production of alternative ecosystem services. The ranch economy often struggles, which lowers the ecosystem’s value as a producer of food resources. At the same time, outside interests are placing high value on the ecosystem’s ability to produce a desirable rural lifestyle. The price these ‘‘amenity buyers’’ are willing to pay for this service exceeds the price ranchers can pay to produce beef. Furthermore, if a ranch is passed on to heirs upon the death of a landowner, the land is valued at its highest value potential rather than its traditional or intended use. As a result, many heirs cannot afford to pay the inheritance tax because high amenity locations are valued for their real estate development potential rather than ranching. Obviously, strong economic pressure exists for landscape transformation when the economic value associated with the land’s ability to produce an aesthetic quality exceeds that of its ability produce cattle. Ranches that become an amenity landholding often change the biophysical and cultural landscape in a way that encourages further transformation; a positive feedback mechanism is established (Haggerty, 2004) that leads to non-linear growth in amenity ownership. This positive feedback mechanism is driven by four distinct processes. First, as stated above, many of the objectives that push the systems toward the production of cattle are social and cultural. With each new amenity acquisition, the social network that supports the ranching community is weakened and, thus, the social pressure to maintain ranching as a way of life declines. Couple this trend in social structure with an aging ranching community and the tendency for younger generations to choose alternative professions, and the social costs associated with selling a ranch are reduced. Second, amenity buyers often have distinctly different land management objectives. For example, many new land managers view elk as an amenity to be cultivated rather than a negative externality from the park or an economic resource to be

838

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845

harvested (Haggerty, 2004). This different environmental perspective is made manifest by changes in land use (e.g., amenity owners place land into conservation easements and prohibit hunting) and land cover (e.g., fields are managed to attract elk). Whether the threat is real or perceived, increased competition from ungulates is viewed as an additional threat by ranchers (Haggerty, 2004). Changes in land use/land cover practices that conflict with traditional uses can, therefore, be used as a justification to sell even when there is social pressure to maintain the status quo. Third, the declining number of ranches can have a concomitant affect on the supporting infrastructure. Ranch supply stores, banks that lend to ranchers, and available labor pools all feel the pressure of declining ranching activity. Stores will close, banks will refocus on the housing market and labor will shift to meet emerging demands (e.g., construction). These changes in the supporting infrastructure will make ranching less attractive and, potentially, less economically viable. Finally, given a declining supply of ranchland and stable or increasing demand for amenity properties, land value, and thus the pressure to transform land out of agricultural production, will continue to climb. The landscape is driven by non-linear processes, feedback mechanism, and path dependency, and, thus, meets our definition of a complex adaptive system. Furthermore, knowledge can be gained by considering the interactions that occur among elemental entities. An argument can be made, therefore, that complexity theory provides insight into this system. 4.2. Land use/land cover affects ecosystem services: selforganization and adaptive processes Ecosystems are comprised of a large number of intricately linked components. A change in one component ripples through the system affecting the state of components that are potentially several links away. An increase in plant biomass (i.e., primary productivity), for example, has a positive impact on herbivores that, in turn, precipitates an increase in predator populations. This bottom-up trophic control of ecosystem dynamics is, of course, constrained— negative feedback mechanisms operate to keep components in relative balance (ecosystem components adapt and self-organize in response to system dynamics). Early ecological theory was often based on the assumption that ecosystems were regulated from the bottom up. By 1994, however, studies on Isle Royale, USA, began to suggest that trophic control can also be top-down (McLaren and Peterson, 1994). In this situation, wolves controlled the moose population and the moose population, in turn, impacted the growth of alder. Landscape changes that have occurred since the reintroduction of wolves into YNP suggest that similar top-down trophic cascade mechanisms exist in the NEWR (Ripple et al., 2001; Fortin et al., 2005). Photographs taken in YNP during the early part of the 20th century depict riparian corridors with tall stands of willow. By the late twenti-

eth century, however, these same areas sported only short stubby willows; elk browsed them (and aspen shoots too) down to the snow depth line each winter when forage became unavailable (NRC, 2002). Wolves were reintroduced, elk populations declined, and some willow patches began to grow into substantial stands again (Ripple et al., 2001; Fortin et al., 2005); suggesting a top-down trophic cascade. But this might not be the entire story. Why, for example, did elk remain in the park to browse on willow and aspen and not travel to lower elevations where more nutritious forage was available? What evidence do we have to suggest that the 1994 NEWR elk population was sustainable? There are, in fact, several competing hypotheses documented in the literature about the tropic relationships of the NEWR that need to be considered. These hypotheses are based on alternative assumptions about the changing spatial pattern of resources and risks and can be used to illustrate different perspectives on the dynamics of complexity systems. 4.2.1. Hypothesis 1. Ungulate population cycles – bottom uptrophic control Ungulate population levels are known to be cyclical and occasional large-scale die-offs are possible (Young, 1994; Erb and Boyce, 1999). A common objective of wildlife managers is to regulate this variability by keeping wildlife population at or below the presumed carrying capacity of the landscape. In 1930 the capacity of the northern range was estimated to be 12,000 to 14,000 elk (Barmore, 2003). This estimate was reduced to 7000 in the late 1930s and to 5000 between 1945 and 1968. The elk population was actively managed throughout this period to produce these population levels. The elk herd was culled by park officials within the park boundary and hunters harvested prescribed numbers of elk outside of the park. The effect of these management strategies is apparent in Fig. 3 where we see a downward sloping trend in elk population from 1929 to 1968. In 1968, natural regulation was adopted and the culling of elk by YNP terminated. The assumption was that bottom-up tropic control would regulate the elk population. At the same time a moratorium on late season hunting was established (a January hunt used to further reduce the elk population). This moratorium ended in 1975, but the reopened hunt was managed to help facilitate migratory behavior and the use of traditional winter range. Given plentiful food and significantly reduced predation rates (by humans) the elk population quickly rebounded, reaching a peak in 1994. The population then began a relatively steady decline to its current level. If the elk population exceeded the capacity of NEWR, then the decline in the elk population can be explained by the over-consumption of forage and bottom-up tropic control. The NEWR ecosystem would, therefore, be driven by negative feedback mechanisms that protect the system from extreme oscillations. Research in and around Yellowstone has shown a negative correlation between population growth and population size (Houston, 1982; Merrill and Boyce, 1991) and

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845

839

Fig. 3. Elk population trends: (a) Total estimated population level and (b) Percent of the elk population counted north of the Dome Mountain.

this suggests density dependent, bottom-up control (NRC, 2002; Taper and Gogan, 2002). Processes associated with this hypothesis are driven by feedback mechanisms and exhibit non-linear behavior. The system exhibits a self organized response as elk respond to available forage, and vice versa. We can say, therefore, that the system is operating as a complex adaptive system. However, because ecosystem changes would be largely predictable from population level dynamics and equilibrium theory, the insights gained by analyzing this hypothesis through the lens of complexity theory are not significant. 4.2.2. Hypothesis 2. The impact of predation and top-down trophic cascade An alternative hypothesis suggests that the introduction of new predators into the park lowered elk population levels and fewer elk means that less vegetation matter is removed by browsing—a classic top-down trophic cascade (Ripple et al., 2001; Fortin et al., 2005). White and Garrott, 2005 argue that predation by wolves and humans will continue to reduce the NEWR elk herd population and this

poses a serious threat to the long term sustainability of the herd. Varley and Boyce (2006), on the other hand, suggest that wolves will have a stabilizing effect on an elk population that can sustain moderate hunting pressure. This difference in opinion is derived from different assumptions about the reproductive potential of predated animals and the affect of prey density on predator success (i.e., the efficacy of negative feedback mechanisms). Which set of assumptions most accurately reflect reality has important implications for elk management. Under this hypothesis the linkages among system components that drive organization and adaptation occur at the population level. Like bottom-up trophic control, the underlying assumption is one of equilibrium between prey and predator and, thus, negative feedbacks act to protect the system from extreme oscillations. Again, the patterns that emerge are predicable from system-level dynamics and, while it is likely that top-down trophic processes are contributing to ecosystem dynamics in Yellowstone, little additional understanding is gained by evoking complexity theory in the analysis of this hypothesis.

840

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845

4.2.3. Hypothesis 3. The impact of wolves on the spatial pattern of elk; adaptation and self-organization Wolves could also effect change in ecosystem structure by impacting the spatial behavior of elk and this change in behavior could occur at multiple spatial scales—a response not considered by hypotheses 1 or 2. At a broader landscape-scale we can postulate the following: Given a spatially constrained predator (wolf due to home ranges) dependent on a spatially unconstrained prey that modifies the landscape (elk who move to avoid predation and consume forage and browse), a shifting and cyclical landscape pattern emerges. Wolf predation affects the spatial pattern of elk, concentrating them into areas of low predation pressure (Mech, 1977; Edwards, 1983; Ferguson et al., 1988; White et al., 1998; Hepplewhite et al., 2005; Van de Koppel et al., 2005). Within these areas willow and aspen become over browsed (Ripple et al., 2001; McLaren and Peterson, 1994), which in turn has a negative impact on riparian fauna. The wolf population responds to the new spatial pattern of elk as different packs prosper and new spatial patterns of predation emerge. Elk adapt to the changing pattern of risk, thus reducing the pressure on plant resources in one area and increasing it in another. This creates a shifting pattern of resource use where some locations temporarily experience over-exploitation, but then rebound when released from intense herbivory. The ecosystem is sustainable at a landscape-scale. Long term studies are needed to test this hypothesis empirically. Changes in the pattern of resource use at a local level, however, may also be occurring and these are more readily identified. This hypothesis was put forth and evaluated by Creel and Winnie (2005) whose work shows that elk tend to move away from open areas and toward forested areas in the presence of wolves. Less time spent in the broad flat riparian zones results in reduced browsing. The spatio-temporal interactions that drive the system dynamics hypothesized here represent a self-organized and adaptive response to inter and intra specie interactions. The system is acting as a complex adaptive system. Furthermore, the patterns hypothesized emerge from the spatio-temporal behavior of individuals and rely on processes that occur at multiple spatio-temporal scales. Patterns are produced, therefore, that may not be predicted from hypotheses informed by population-level equilibria alone. Complexity theory adds to our understanding of system behavior by focusing attention on how individual-level interactions produce self-organized and adaptive behavior at multiple spatio-temporal scales. 4.2.4. Hypothesis 4. The impact of climate change; crossscale relationships Over the past several years the Greater Yellowstone Ecosystem (GYE) has experienced mild winters and low snow fall. Lower snow depth levels provide: (1) greater access to grasses and, thus, the dependence of elk on willow during the winter season is reduced; and (2) less pressure to

migrate out of the park during the winter. Over the same time period, the region has experienced a prolonged drought with a concomitant reduction in primary productivity (Vucetich et al., 2005). Reduced primary productivity could have contributed to reduced herd size and pushed elk north of the park in search of forage. The work of Vucetich et al. (2005) suggests that the combined effect of drought and human predation is sufficient to explain the post1994 elk population decline and these researchers go on to suggest that predation by wolves is almost entirely compensatory. Again, the conclusions reached by Vucetich et al. (2005) differ from those of reached White and Garrott, 2005 and Varley and Boyce (2006) because of the underlying assumptions about elk reproductive potential, prey selection, and predation rates used in their analysis. It is plausible that these recent anomalies in NEWR weather patterns are a localized response to global climate change and, thus, local ecosystem dynamics are adapting to biophysical and human processes that are occurring at a larger scale. The system, therefore, is acting in a complex manner. 4.2.5. Hypothesis 5. The impact of human management strategies; adaptation, self-organization, and path dependence All of the above hypotheses lack the ability to explain why elk remained in the park until the late 1980’s, even during deep snow conditions, to browse on willow and aspen rather then traveling to lower elevations where forage was more plentiful. Given the unique spatial location of the NEWR it is reasonable to question the role that humans might play in this coupled system (Houston, 1982; Boyce, 1989) and, in fact, there is at least anecdotal evidence that human activity might, in fact, be an important piece of the puzzle. Humans differ from natural predators in two important ways. First, human population levels are not a function of the prey population and, second, the spatial–temporal pattern of human predation is patchy—regulations and access restrict hunting to specific locations and time periods. When predator population levels are not controlled by prey population (e.g., the human population in the western US is not controlled by the elk population), prey can become overexploited (Van de Koppel et al., 2005). Furthermore, when the threat of predation is patchy, prey seek areas of lower threat (e.g., YNP before wolves) and, potentially overexploit food sources (e.g., willow) at these locations (Mech, 1977; White et al., 1996; Tschirhart, 2003; Van de Koppel et al., 2005). Topography and deep snow funnel migrating elk down the Lamar and Yellowstone River valleys and out of the park near Gardiner, MT where they traditionally would have left park boundaries in search of winter forage (see Fig. 1). From the 1940s to the late 1960s the NEWR herd was subjected to considerable hunting pressure at this critical location. It is commonly acknowledged that elk learned about and adapted to this boundary (the ‘‘firing line’’).

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845

Lemke (1995), a long time Montana Fish, Wildlife and Parks wildlife manager, notes ‘‘. . .the elk lost their migratory behavior, never moving far from the national park boundary.’’ Lemke’s viewpoint echoes that of an old-time game warden, Joe Gaab, who stated that (reported as a personal communication in Lemke, 1995) ‘‘We needed to let the lead elk through and keep the memory bank alive...’’ Why did migration patterns change? Lemke, Gaab and others suggest that it was hunting pressure. An alternative hypothesis about trophic relationships in and around YNP, therefore, is that the significant hunting pressure outside of the park and the lack of natural predators inside the park produced a significant spatial imbalance in the risk surface to which elk were subjected (Boyce, 1989). Elk perceived it to be in their best interest to remain in the park and browse on willow and aspen (with its cascading impacts on riparian ecosystems) rather than risk predation by humans. Human activity turned YNP into an island refuge for elk, thereby truncating in space and time the natural processes (described in Section 4.2.3) that were hypothesized to produce landscape-level sustainability and resiliency. Interestingly, two of the best examples of top-down trophic cascades associated with large animals are associated with island ecosystems (McLaren and Peterson, 1994; Terborgh et al., 2001). This risk surface for elk changed with the 1968 hunting regulations. Elk, however, did not significantly adapt their spatial behavior until 1989, when an increase in the number of elk migrating to locations north of the park occurred (Fig. 3). One explanation for the delayed response in elk behavior lies in the lost spatial knowledge referred to by Lemke and Gaab. In 1988 a major fire occurred in YNP, followed by a particularly harsh winter. The lack of food and the deep snow (i.e., a change in the risk surface) forced elk to leave the park in record high numbers and, thereby, rediscover forgotten migratory paths. The percentage of elk leaving the park has remained relatively high since this event. The 1995 reintroduction of wolves modified the risk surface over which elk navigate once again. However, teasing out the relative impact of forage, snow, and predation (human and wolf) is challenging. The point we wish to make here is that the processes of adaption, interaction, perturbation and the emergence of new structure, and path dependency all seem to be operating in this system. The system as described here is decidedly complex. 4.2.6. Hypothesis 6. Some combination of all of the above Each of the general hypotheses describe above have been supported by at least one study and, conversely, studies exist that failed to find support for many of these very same ideas. While some of this controversy stems from disagreements about underlying assumptions (White and Garrott, 2005), it is probable that more than one of the processes described above are affecting the NEWR ecosystem. The NEWR is acting as a complex, open system driven by biophysical and socio-economic processes that operate at multiple scales. Unfortunately, traditional scien-

841

tific method is ill-suited to the direct study of such systems. Practical considerations of available time and resources, along with tightly held epistemological ideals, like replicability, experimentation, control, and hypothesis testing, tend to focus scientific effort on specific processes (e.g., elk bioenergetics) or process pair (e.g., wolf/elk interaction). This may, in part, explain the continuing controversy about what drives ecosystem dynamics in the NEWR—the proverbial problem of describing an elephant through reduction. While studying each process (e.g., elk bioenergetics) or process pair (e.g., wolf/elk interaction) in isolation produces important new knowledge, it may not be sufficient to gain a complete picture of ecosystem dynamics. Stated more positively, the NEWR is a complex adaptive system and the concepts associated with complexity theory prove useful when thinking about how its individual components interact to produce system-level behavior. 4.3. Dissatisfaction with the produced ecosystem services can affect policy: Self-organization, self-organized criticality, and emergence Consider Bak’s (1996) often discussed experiment in self-organized criticality (see Table 1). Grains of sand are slowly added to a growing pile of sand. Each additional grain causes a disturbance (the kinetic energy of the falling sand grain is transferred to surrounding grains causing a sand slide). Individual grains, however, quickly lock together, stopping the sand slide and storing some of this energy in the system. As this process continues the sides of the pile become steeper and the sand pile system moves farther from its low energy state. Energy accumulates and reaches a critical threshold, far from the system’s equilibrium, and a large-scale energy dissipating event occurs (an avalanche occurs when the angle of repose is exceeded). It turns out that the magnitude of these energy dissipating events follows a power law distribution. The system, therefore, self-organizes at two distinct scales. Individual elements self-organize to form metastable states and the system self-organizes to form a power law distribution. While we cannot directly apply the second law of thermodynamics to social systems, metaphorically the concepts associated with Bak’s sand pile experiment also apply to social processes. History is replete with examples of civilizations that existed for centuries in far-from-equilibrium states, only to fall catastrophically from dominance due to processes that began at local scales (Diamond, 2005). Stress builds in social systems until a threshold is reached, at which point system components adapt, self-organize into new states, and stored energy is dissipated; a phenomenon sometimes referred to as a tipping point in the social science literature. It is at these tipping points that some suggest ‘‘new structural arrangements and morphogenic changes are most likely to occur’’ (Mathews et al., 1999). Similar processes might apply to more local social dynamics. Land use policies and regulations are designed, in part, to produce a particular suite of ecosystem services

842

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845

that reflects the relative importance that decisions-makers place on alternative services. The system can be considered to be in equilibrium (a low energy state) when the production of ecosystem services equals the demand for those services. What if one new individual is added to the system that possesses alternative views on the relative importance of ecosystem services (e.g., the first amenity buyer, a metaphorical grain of sand)? The system moves slightly out of equilibrium, but this grain of sand has only a very minimal and localized impact on policy and the production of ecosystem services. The demographics of the region continue to change slowly; amenity buyers trickle in and increasingly demand a reassessment of the policies and regulations that guide land use. Entrenched leadership, social networks, and ideas that have become embedded in the belief systems of decision-makers can all operate to maintain the status quo. However, maintaining the system in such a non-equilibrium state requires an investment of energy. Stress continues to build as the disparity between what is desired and what is produced grows. New social organizations emerge to advocate for change. As the system moves to a farfrom-equilibrium point, a threshold will be reached and a large scale system changing and energy dissipating event (a tipping point) will occur. New officials are elected and new policies enacted. Changes in policy and regulations change the decision-making environment of land–owners, which changes the mix of ecosystems services produced (and thus cycle begins again, see Section 4.1). Social, economic, and political processes are complex. 5. Why analyze given a science of surprise? Interaction, feedback mechanisms, path dependency, self-organization, trigger points, and emergent behavior all limit our ability to explain or predict outcomes—and the farther into the future we try to apply our science, the more futile the exercise becomes. So why analyze given a science of surprise? This question has both a theoretical and a practical answer. 5.1. Theory In science we strive to describe both the state of a system and the driving forces that transform a system from one state to another. We like to assume deterministic connections between form and process so that, theoretically, direct and indirect links among causes and effects can be understood. Complexity theory cautions us against making such assumptions at the spatio-temporal scales at which geographers work and this pushes us to adopt alternative approaches to research because: (1) each cause can produce many effects; (2) each effect can be produced by many causes; and (3) causes can interact in complex ways that produce unexpected effects (Brown et al., 2006). Consider, for example, the discussion of elk dynamics in Section 4.2. The spatial behavior of elk and their effect on vegetation may not be adequately understood without a consideration

the complex interactions of humans, wolves, fire, climate, forage, and, perhaps, even the way in which individual elk encode and use spatial knowledge (Bennett and Tang, 2006). From the perspective of complexity theory, therefore, the interesting questions relate to how social and environmental dynamics lead to: (1) systems that are near equilibrium; (2) systems that are far from equilibrium; (3) emergent behavior; (4) thresholds or tipping points; and (5) rapid change and adaptation. Insight into these questions might be best found by shifting our focus from the boxes that form the typical system diagram (i.e., the entities that comprise a system) to the linkages that exist among them. While part of the answers to such questions are likely to be contextual, other elements of coupled natural/human systems are likely to be highly generalizable (e.g., the ways in which power is maintained, inertia, response to stimuli, fear of the unknown). 5.2. Practice The fact that similar spatial structures emerge in many different places is reassuring; geography is, to some extent, predictable and generalizable. Biophysical, cultural, political and economic processes provide bounds on what is possible and produce a form of self organization that can be used in decision-making (the NEWR looks different than Silicon Valley for a reason). Metastable systems emerge and are maintained by the set of dynamic processes that they produce (e.g., flows of energy, materials, people, capital, and ideas). Landscape level change is, of course, inevitable but the basic way in which we manage the landscape and our expectation of some delivered set of ecosystem services remains relatively constant, even if the relative value placed on different services changes through time. Complexity theory, therefore, provides insight into the practice of land use/land cover management in two important ways. First, assume that Fig. 4 represents a simplified decision space. A decision-maker has a binary decision to make given information on decision variables 1 and 2. Assuming that these decision variable estimates are relatively certain, the decision-maker will not regret his or her choice when decision-variables fall in regions A or B; decisions are robust in these regions (Lembert et al., 2003). Region C represents a region of uncertainty where a clearly superior solution cannot be ascertained and D represents a threshold condition where the preferred action changes abruptly with small changes in the value of decision-variables. From this diagram we are reminded that the set of plausible future states is large (probably larger than typically envisioned) and this realization pushes us to explore more fully the breadth of possible outcomes produced by our decisions. Through an exploration of this many-to-many relationship among decisions and outcomes, and a careful evaluation of the range of values decision-variables can assume, decision-makers are better able to balance risk and uncertainty against desired results. We

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845

Fig. 4. The concepts of complexity theory (e.g., path dependency, threshold points, and emergence) guide us to explore uncertainty and the existence of critical thresholds in decision and solution spaces.

are further reminded that the set of decision options is often relatively small. This suggests that decision-makers are likely to reach the same conclusion given many different current and future scenarios. Complexity theory warns us to pay particularly close attention to decisions made near thresholds (decision points at the edge of chaos) and regions of uncertainty because it is in these regions that the impacts of initial conditions, path dependences, and unexplored interactions are most likely to produce regrettable outcomes. Second, note that each point in a decision space can be mapped to a point in a solution space (i.e., decisions have real-world outcomes). By viewing the surface presented in Fig. 4 from this perspective, complexity theory leads us to ask interesting questions about sustainability, resiliency, and threshold events given alternative assumptions about, for example, initial conditions, adaptation, metaequilibria, and linkages among system elements. 6. Conclusions It seems obvious that coupled natural/human systems are complex. They are, after all, comprised of heterogeneous components whose actions combine to produce emergent behavior that creates results that are often unexpected. These systems are driven in large measure by the linkages that exist among system elements and characterized by feedback mechanisms, non-linearities, path dependence, cross-scale relationships, and self-organization (Holland, 1995; Auyang, 1998; McKelvey, 2001; Manson, 2003; O’Sullivan, 2004). While we remain skeptical of some of the more grandiose claims made about complexity theory (Kauffman, 1995; Bak, 1996; Wolfram, 2002), we believe that it does provides a useful conceptual framework for the study of such systems and focuses attention on system interactions that are often overlooked when traditional approaches are applied.

843

Complexity theory leads us toward questions that, when answered, provide important insight into the resolution of difficult problems (e.g., questions about sustainability, resiliency, and threshold events) that can lead to significant and sudden changes in system state. Furthermore, we believe that complexity theory provides a catalyst for interdisciplinary work. Just as the need to store data using a geographic information system framework often provides the impetus for researchers from various disciplines to resolve differences in representation and methodological approach, complexity theory, by emphasizing linkages among system elements, encourages researchers from different disciplines to resolve differences in conceptual and theoretical perspective. Complexity theory has, therefore, the potential to promote tightly integrated, interdisciplinary research into important problems facing human-kind today. We are fortunate to have a research team who embraced, argued, and discussed how to cross disciplinary boundaries to explore new ways to understand and represent the variables we studied. The study of complex and adaptive geographic systems, however, does present unique methodological and theoretical challenges. Path dependence, self-organization, and emergent properties, for example, severely limit the utility of traditional methods of hypothesis testing and model development. New forms of computer modeling are beginning to evolve derived from agent-based technologies that are designed to help study CAS. These tools hold significant promise (e.g., see other papers in this issue, Janssen et al., 2000; Parker et al., 2003; Bennett and Tang, 2006). Before these tools reach their full potential, however, researchers must overcome representational issues (e.g., the digital representation of spatial knowledge, social capital, or collaborative action) and issues associated with model validation (e.g., how to validate given path dependency and the lack of deterministic processes, Pontius, 2002; Brown et al., 2005). While the challenges of coupled natural/human systems research are significant, they are reflective of real-world processes that must be considered if we are to understand many complex problems associated with modern geographic inquiry. Solutions to these problems will benefit from basic research into what motivates the social, cultural, and spatial behavior of humans and other organisms, better methods for the formalization of spatial decision-making, and a deeper understanding of natural processes and the affects of humans on these processes. Complexity theory, while immature and struggling to gain a well defined identity, holds promise as an intellectual jumping-off point for the study of such problems. Acknowledgements The authors would like to thank the National Science Foundation (BE-CNH Award #0216588, ‘‘Complexity Across Boundaries: Coupled Human and Natural Systems in the Yellowstone Northern Elk Winter Range’’) and The University of Iowa Center for Global and Regional

844

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845

Environmental Research for their support of this project. The authors also thank our entire research team for their support in research related to this project. References Auyang, S.Y., 1998. Foundations of Complex-system Theories: In Economics, Evolutionary Biology, and Statistical Physics. Cambridge University Press, Cambridge, UK. Axelrod, R., 1997. The Complexity of Cooperation: Agent-based Models of Competition and Collaboration. Princeton University Press, Princeton, NJ. Bak, P., 1996. How Nature Works: The Science of Self-organize Criticality. Springer-Verlag, New York. Barmore, W.J., 2003. Ecology of Ungulates and Their Winter Range in Northern Yellowstone National Park: Research and Synthesis 1962– 1970 (Mamoth Hot Springs: National Park Service). Bennett, D.A., Tang, W., 2006. Modeling Yellowstone’s northern range elk herd as adaptive, spatially aware, and mobile agents. International Journal of Geographical Information Science 20 (9), 1039–1066. Boyce, M.S., 1989. The Jackson Elk Herd. Cambridge University Press, Cambridge, UK. Brown, D.G., Aspinall, R., Bennett, D.A., 2006. Landscape models and explanation in landscape ecology—a space for generative landscape science? The Professional Geographer 58 (4), 369–382. Brown, D.G., Page, S., Riolo, R., Zellner, M., Rand, W., 2005. Path dependence and the validation of agent-based spatial models of land use. International Journal of Geographical Information Science 19 (2), 153–174. Carpenter, S.R., and Brock, W.A. 2004. Spatial complexity, resilience, and policy diversity: Fishing on lake-rich landscape. Ecology and Society 9(1), 8. URL: http://www.ecologyandsociety.org/vol9/iss1/art8. Casti, J.L., 1996. Would-Be Worlds: How Simulation Is Changing the Frontiers of Science. John Wiley and Sons, New York, NY. Cilliers, P., 1998. Complexity and Postmodernism: Understanding Complex Systems. Routledge, London. Diamond, J., 2005. Collapse: How Societies Choose to Fail or Succeed. Penguin Group, New York. Edwards, J., 1983. Diet shifts in moose due to predator avoidance. Oecologia 60, 185–189. Creel, S., Winnie, J.A., 2005. Responses of elk herd size to fine-scale spatial and temporal variation in the risk of predation by wolves. Animal Behaviour 69, 1181–1189. Erb, J.D., Boyce, M.S., 1999. Distribution of population declines in large mammals. Conservation Biology 13, 199–201. Flum, J., Grohe, M., Weyer, M., 2006. Bounded fixed-parameter tractability and log2n nondeterministic bits. Journal of Computer and System Sciences 72, 34–71. Fortin, D., Beyer, H.L., Boyce, M.S., Smith, D.W., Duchesne, T., Mao, J.S., 2005. Wolves influence elk movements: behavior shapes a trophic cascade in Yellowstone National Park. Ecology 86, 1320–1330. Ferguson, J.W., Bergerud, A.T., Ferguson, R., 1988. Predator risk and habitat selection in the persistence of a remnant caribou population. Oecologic 76, 236–245. Haggerty, J. 2004. A ranchland genealogy: land, livestock and community in the Upper Yellowstone Valley, 1866–2004. University of Colorado, Dissertation thesis. Hepplewhite, M., White, C.A., Nietvelt, C.G., McKenzie, J.A., Hurd, T.E., Fryxell, J.M., Bayley, S.E., Paquet, P.C., 2005. Human activities mediate a trophic cascade caused by wolves. Ecology 86 (8), 2135– 2144. Holland, J.H., 1995. Hidden Order: How Adaptation Builds Complexity. Perseus Books, Cambridge, MA. Horgan, J., 1995. From complexity to perplexity: can science achieve a unified theory of complex systems? Scientific American 284, 104–109. Houston, D.B., 1982. The Northern Yellowstone Elk: Ecology and Management. Macmillan, New York.

Janssen, M.A., Walker, B.H., Langridge, J., Abel, N., 2000. An adaptive agent model for analysing co-evolution of management and policies in a complex rangeland system. Ecological Modelling 131, 249– 268. Jensen, H.J., 1998. Self-Organized Criticality: Emergent Complex Behavior in Physical and Biological Systems. Cambridge University Press, New York. Johnson, J.L., Burton, B.K., 1994. Chaos and complexity theory for management: Caveat emptor. Journal of Management Inquiry 3 (4), 320–328. Kauffman, S.A., 1995. At Home in the Universe: The Search for Laws of Complexity. Penguin, London. Lansing, J.S., 2003. Complex adaptive systems. Annual Review of Anthropology 32 (1), 183–204. Lemke, T.O., 1995. A Montana Tradition. Montana Outdoors March/ April: 4–9. Lemke, T.O., 2005. Personal communication. Lemke, T.O., Mack, J.A., Houston, D.B., 1998. Winter range expansion by the northern Yellowstone elk herd. Intermountain Journal of Sciences 4 (1/2), 1–9. Lembert, R.J., Popper, S.W., Bankes, S.C., 2003. Shaping the Next One Hundred Years: New Methods for Quantitative, Long-Term Policy Analysis, Santa Monica, Rand. Levin, S.A., 1998. Ecosystems and the biosphere as complex adaptive systems. Ecosystems 1, 431–436. Manson, S.M., 2001. Simplifying complexity: a review of complexity theory. Geoforum 32 (3), 405–414. Manson, S.M., 2003. Epistemological possibilities and imperatives of complexity research: a reply to Reitsma. Geoforum 34 (1), 17–20. Manson, S., O’Sullivan, D., 2006. Complexity theory in the study of space and place. Environment and Planning A 38, 677–692. Mathews, K.M., White, M.C., Long, R.G., 1999. Why Study the complexity sciences in the social sciences. Human Relations 52 (4), 439–462. McKelvey, B., 2001. Foundations of ‘‘New’’ Social Science: Institutional Legitimacy from Philosophy, Complexity Science, Postmodernism, and Agent-based Modeling. Presented at the Sackler Colloquium on Agents, Intelligence and Emergent Human Organization: Capturing Complexity Through Agent-Based Modeling, University of California, Irvine, October 5th. McLaren, B.E., Peterson, R.O., 1994. Wolves, moose, and tree rings on Isle Royale. Science 266, 1555–1558. Mech, L.D., 1977. Wolf-pack buffer zones as prey reservoirs. Science 198, 320–321. Merrill, E.H., Boyce, M.S., 1991. Summer range and elk population dynamics in Yellowstone National Park. In: Keiter, R.B., Boyce, M.S. (Eds.), The Greater Yellowstone Ecosystem: Redefining America’s Wilderness Heritage. Yale University Press, New Haven, Connecticut, pp. 263–274. MFWP, 2005. Elk count data provided by T. Lemke, Montana Fish, Wildlife, and Parks. NRC, 2002. Ecological Dynamics on Yellowstone’s Northern Range. National Academy Press, Washington, DC. O’Sullivan, D., 2004. Complexity science and human geography. Transactions of the Institute of British Geographers 29 (3), 282–295. Parker, D.C., Manson, S.M., Janssen, M., Hoffmann, M.J., Deadman, P.J., 2003. Multi-agent systems for the simulation of land use and land cover change: a review. Annals of the Association of American Geographers 93 (2), 314–337. Parmenter, A.W., Hansen, A., Kennedy, R., Cohen, W., Langner, U., Lawrence, R., Aspinall, R., Maxwell, B., Gallant, A., 2003. Land Use and Land Cover Change in the Greater Yellowstone Ecosystem 1975– 95. Ecological Applications 13, 687–703. Phillips, J.D., 1999. Divergence, convergence, and self-organization in landscapes. Annals of the Association of American Geographers 89, 466–488. Pontius, R.G., 2002. Statistical methods to partition effects of quantity and location during comparison of categorical maps at multiple

D. Bennett, D. McGinnis / Geoforum 39 (2008) 833–845 resolutions. Photogrammetric Engineering & Remote Sensing 68, 1041–1049. Portugali, J., 2006. Complexity theory as a link between space and place. Environment and Planning A 38, 647–664. Pritchard, James A. 1999. Preserving Yellowstone’s Natural Conditions: Science and the Perception of Nature. University of Nebraska Press, Lincoln. Reitsma, F., 2003. A Response to Simplifying Complexity. Geoforum 34 (1), 13–16. Ripple, W.J., Larsen, E.J., Renkin, R.A., Smith, D.W., 2001. Trophic cascades among wolves, elk and aspen on Yellowstone National Park’s northern range. Biological Conservation 102 (3), 227–234. Robbins, P., 2001. Tracking invasive land covers in India or why our landscapes have never been modern. Annals of the Association of American Geographers 91 (4), 637–654. Simon, H., 1957. Models of Man. Wiley, New York. Taper, M.L., Gogan, P.J.P., 2002. The Northern Yellowstone elk: density dependence and climatic conditions. Journal of Wildlife Management 66 (1), 106–122. Terborgh, J., Lopez, L., Nunez, V., Rao, M., Shahabuddin, G., Orihuela, G., Riveros, M., Ascanio, R., Adler, G.H., Lambert, T.D., Balbas, L., 2001. Ecological meltdown in predator-free forest fragments. Science 294, 1923–1926.

845

Tschirhart, J., 2003. A new adaptive system approach to predator–prey modeling. Ecological Modelling 176 (2004), 255–276. Van de Koppel, J., Bardgett, R.D., Bengtsson, J., Rodriguez-Barrueco, C., Rietkerk, M., Wassen, M.J., Volkmar, W., 2005. The effects of spatial scale on trophic interactions. Ecosystems 8, 801–807. Varley, N., Boyce, M.S., 2006. Adaptive management for reintroductions: updating a wolf recovery model for Yellowstone National Park. Ecological Modeling 193, 315–339. Vucetich, J.A., Smith, D.W., Stahler, D.R., 2005. Influence of harvest, climate, and wolf predation on Yellowstone elk, 1961–2004. Oikos 111, 259–270. Wolfram, S., 2002. A New Kind of Science. Wolfram Media, Champaign. White, C.A., Olmsted, C.E., Kay, C.E., 1998. Aspen, elk, and fire in the Rocky Mountain national parks of North America. Wildlife Society Bulletin 26, 449–462. White, K.A.J., Murray, J.D., Lewis, M.A., 1996. Wolf–deer interactions: a mathematical model. Proceeding of the Royal Society of London Series B – Biological Sciences 263 (1368), 299–305. White, P.J., Garrott, R.A., 2005. Yellowstone’s ungulates after wolves— expectations realizations, and predictions. Biological Conservation 125, 141–152. Young, T.P., 1994. Natural die-offs of large mammals: implications for conservation. Conservation Biology 8 (2), 410–418.