Discover millions of ebooks, audiobooks, and so much more with a free trial

Only $11.99/month after trial. Cancel anytime.

New Models for Ecosystem Dynamics and Restoration
New Models for Ecosystem Dynamics and Restoration
New Models for Ecosystem Dynamics and Restoration
Ebook816 pages8 hours

New Models for Ecosystem Dynamics and Restoration

Rating: 0 out of 5 stars

()

Read preview

About this ebook

As scientific understanding about ecological processes has grown, the idea that ecosystem dynamics are complex, nonlinear, and often unpredictable has gained prominence. Of particular importance is the idea that rather than following an inevitable progression toward an ultimate endpoint, some ecosystems may occur in a number of states depending on past and present ecological conditions. The emerging idea of “restoration thresholds” also enables scientists to recognize when ecological systems are likely to recover on their own and when active restoration efforts are needed.

Conceptual models based on alternative stable states and restoration thresholds can help inform restoration efforts. New Models for Ecosystem Dynamics and Restoration brings together leading experts from around the world to explore how conceptual models of ecosystem dynamics can be applied to the recovery of degraded systems and how recent advances in our understanding of ecosystem and landscape dynamics can be translated into conceptual and practical frameworks for restoration.

In the first part of the book, background chapters present and discuss the basic concepts and models and explore the implications of new scientific research on restoration practice. The second part considers the dynamics and restoration of different ecosystems, ranging from arid lands to grasslands, woodlands, and savannahs, to forests and wetlands, to production landscapes. A summary chapter by the editors discusses the implications of theory and practice of the ideas described in preceding chapters.

New Models for Ecosystem Dynamics and Restoration aims to widen the scope and increase the application of threshold models by critiquing their application in a wide range of ecosystem types. It will also help scientists and restorationists correctly diagnose ecosystem damage, identify restoration thresholds, and develop corrective methodologies that can overcome such thresholds.
LanguageEnglish
PublisherIsland Press
Release dateMar 19, 2013
ISBN9781610911382
New Models for Ecosystem Dynamics and Restoration

Read more from Richard J. Hobbs

Related to New Models for Ecosystem Dynamics and Restoration

Titles in the series (29)

View More

Related ebooks

Science & Mathematics For You

View More

Related articles

Reviews for New Models for Ecosystem Dynamics and Restoration

Rating: 0 out of 5 stars
0 ratings

0 ratings0 reviews

What did you think?

Tap to rate

Review must be at least 10 words

    Book preview

    New Models for Ecosystem Dynamics and Restoration - Richard J. Hobbs

    Directors

    PREFACE

    The idea of putting together this book arose from the perception that the link between theory and practice in ecology and, in particular, restoration ecology is not as strong as it could be. There are interesting new ideas about how ecosystems work arising from both the theoretical and the practical side, and it seems essential that these ideas are examined critically and, where possible, tested in order to assess their applicability in restoration practice. Essentially, that is the aim of this book. We aimed to bring together the people that have been at the forefront of developing new conceptual models for ecosystem dynamics and people who have been using such models in an applied setting. We aimed to ask a number of questions surrounding this topic. What is the theoretical basis for the new models, and what evidence is there that such models can be applied to real ecosystems? How have such models been used to think about and manage a range of ecosystems? And is it possible to better mesh the theory and practice to improve both?

    We wish to thank all the people who have assisted us in bringing together this book. First, we thank the authors of the chapters for their contributions to the book, their gracious and timely responses to our requests, and their willingness to consider their work in a broader conceptual and/or practical context. Barbara Dean at Island Press is both an inspiration and a welcome and thoughtful guide through the process of putting a book together, and Barbara Youngblood is both patient and effective in the production process; without them and the rest of the Island crew, it is unlikely that the book would have been either started or finished.

    All chapters in the book were peer reviewed. In addition to internal review by other chapter authors, each chapter was also reviewed by someone not involved in the book. We gratefully acknowledge those who reviewed chapters and provided comment on the ideas presented within the book: Mitch Aide, Peter Bellingham, Margaret Brock, Raphael Didham, Susan Galatowitch, Karen Holl, Kathryn Kiel, Jamie Kirkpatrick, Anke Jentsch, David Lindenmayer, Sue McIntyre, Karel Prach, and Eric Seabloom. We are also grateful to the anonymous referees and the series editors, James Aronson and Don Falk, for their thoughtful comments on the initial book proposal.

    RJH received financial support from the Australian Research Council as an ARC Australian professorial fellow during the preparation of this book. KNS acknowledges research support from the National Science Foundation and U.S. Department of Agriculture.

    Finally, we’d like to thank the colleagues with whom we’ve discussed and argued about the ideas in the book and, last but not least, our families, who have to put up with continual absences, permanently cluttered home workspaces, and long periods when we were tied to the computer.

    PART ONE

    Background: Concepts and Models

    Our objective in putting together this book was to collectively evaluate threshold modeling approaches as applied to ecological restoration. It was our aim not to develop a new suite of models but rather to examine when and where state transition, alternative state, and other threshold approaches are being used; how effective they are; and what types of evidence are being used to derive and apply the models. There are many ways to explore the nexus between the demand for strong inference and statistically robust approaches required by these models and the need to get something that works practically. In this emerging junction between theoretical and quantitative ecology and practical ecological restoration, it was our goal to explore both the potential and the pitfalls of this meshing of theory and practice as well as to develop innovative ways to maximize the potential and minimize the pitfalls. In this part, experts in the theoretical and quantitative aspects of these ecosystem models examine the possible ways of assessing ecosystem dynamics. These six chapters highlight the conceptual relevance of these models to restoration as well as the problems of rigorously testing the models in a restoration setting.

    Chapter 1

    Models of Ecosystem Dynamics as Frameworks for Restoration Ecology

    KATHARINE N. SUDING AND RICHARD J. HOBBS

    As rates of exotic species invasion, fragmentation, and climate change continue to accelerate, restoration faces increasingly greater challenges. Restoration must address the substantial, long-lasting reorganizations of ecosystems driven by these impacts. Among practitioners and scientists alike, there is increasing recognition that ecosystem dynamics can be complex, nonlinear, and often unpredictable (Wallington et al. 2005). Of particular importance is the recognition that some ecosystems may occur in a number of different states, which may be contingent on the history of disturbance, human intervention, or past restoration actions (Beisner et al. 2003; Suding et al. 2004). Complementary approaches using modifications of classical succession theory and the concept of assembly rules have also recently been investigated in the context of managing and restoring ecosystems (Young et al. 2001; Temperton et al. 2004; Hobbs et al. 2007). Adding further complexity, we better understand the importance of broad-scale processes and interactions between adjoining ecosystems; impacts in one place may be the result of events or management decisions elsewhere (Hobbs 2002). Taken together, these advances yield an exciting body of theory on which to rest restoration ecology (D’Antonio and Thomsen 2004; Hobbs and Norton 2004; Holl and Crone 2004; Young et al. 2005).

    This book addresses how recent advances in our understanding of ecosystem and landscape dynamics can be translated into the conceptual and practical frameworks for restoration, adding to a number of excellent recent books developing linkages between ecological theory and restoration (Whisenant 1999; Walker and del Moral 2003; Temperton et al. 2004; Falk et al. 2006; van Andel and Aronson 2006; Walker et al. 2007). We explore how ecosystem models, particularly those that encompass nonlinear and complex dynamics, can be applied to the recovery of degraded systems (Hobbs and Norton 1996; Prober et al. 2002; Lindig-Cisneros et al. 2003; Suding et al. 2004). In this introductory chapter, we trace the development of these new ecosystem models in restoration, describe the main restoration approaches that would be taken based on the different type of ecosystem dynamics, and provide a synopsis of suitable evidence and approaches that can be used to determine what models of ecosystem dynamics may be applicable for particular systems and restoration situations. Lastly, we delineate the limitations and important considerations of this evidence that will affect inference, starting a discussion that continues in the contributed works that follow.

    We also include a definitions of terms used throughout this book. Many terms have very precise definitions as they relate to ecosystem models. To avoid misperceptions and oversimplification, it is important to maintain very clear and unambiguous terminology as application of these models increase. We italicize our first use of terms throughout this chapter that we define in box 1.1.

    BOX 1.1 Definitions of Terms

    Sources for terminology include the glossary at the Resilience Alliance Web page (http://www.resal-liance.org), Gunderson (2000), and Carpenter et al. (2001).

    Alternative stable states. A regime shift involving alternate stable states occurs when a threshold level of a controlling variable in a system is passed, such that the nature and extent of feedbacks change, resulting in a change of direction (the trajectory) of the system. A shift occurs when internal processes of the system (rates of birth, mortality, growth, consumption, decomposition, leaching, and so on) have changed and the state of the system (as indicated by system state variables) begins to change in a different direction toward a different attractor or domain of attraction. The system changes direction when it crosses an unstable equilibrium or repellor into another basin of attraction. While the S-shaped curve (fig. 1.1c) is a common way of visualizing these dynamics, there are many types of relationships between system state variables and environmental conditions. Other terms that are synonymous with alternative stable states include alternative attractors, alternative basins of attraction, multiple stable states, and multiple stable equilibria.

    Controlling variable. The factor (or combination of factors) that drives the change in responding state variables (e.g., grazing intensity, fire frequency, pollution, or nutrient loading). They are sometimes called slow variables. For alternative stable states, controlling variables are assumed to be external to the system (i.e., that changes to the state variables will not change the controlling variables), and the term cross-scale interactions is used when external factors drive transitions.

    A variation of these ideas, termed slow–fast cycles (also called relaxation oscillations or oscillatory behavior) may occur if the state variable (the fast variable) is internally coupled to the controlling variable (the slow variable) and responds discontinuously to it. Such behavior can describe dynamics like the regular spruce-budworm outbreak in boreal forests and differs from alternative stable state behavior, which assumes that the fast state variable does not affect the slow controlling variable.

    Disturbances are changes in the system state variable or in the controlling variable (either of which could be affected by management). It is important to clarify what is being altered as part of a disturbance. For instance, disturbances to the system state variable are sometimes termed perturbations and distinguished from disturbances to the controlling environmental variable, which are sometimes called external shocks.

    Gradual continuum models describe system dynamics without thresholds. In this case, change in the controlling variable is proportionate to change in the state variable. A state variable would not be paired in this case (e.g., no dichotomy of grass versus shrub domination) but rather a continuum of more or less grasses and shrubs in the system. These models imply unassisted recovery following the removal of adverse disturbances (or given the natural disturbance regime) to a desired set point.

    Hysteresis refers to how a system responds or, more specifically, the return path taken following some disturbance or change due to cumulative effects. When the system follows a different path on return to its former state, this is called a hysteresis effect. Hysteresis would occur in alternative stable states and relaxation oscillations but not other types of regime shifts or nonregime behavior (gradual continuous and stochastic models). For instance, a system exhibiting threshold behavior but not alternative stable states should respond discontinuously but similarly to disturbance regardless of whether it is moving forward or backward.

    Persistence. One criterion for stability is that a state is persistent: maintained beyond one complete turnover of individuals. As the time frame of vegetation dynamics generally exceeds the time scales of human observation or management goals, restoration decisions may often need to be made without the certain documentation of persistence. States without strong evidence of persistence are sometimes termed transient states.

    Positive feedbacks. One important consideration about whether a system has crossed a threshold is whether feedbacks affecting internal processes (rates of birth, mortality, growth, consumption, decomposition, leaching, and so on) have changed. When a threshold is crossed, positive feedback loops (the output of a process influences the input of the same process in a way that amplifies the process) result in disproportionate changes relative to the controlling variable. Near a regime or attractor, negative feedbacks increase in importance to stabilize the system.

    Resilience is the capacity of a system to absorb disturbance and reorganize so to retain essentially the same function, structure, identity, and feedbacks (i.e., remain in the same regime). Resilience is also defined as the width of the basin of attraction. The resilience of a system can change; it is not static or defined solely by controlling variables. For instance, the controlling variable for grass- and shrub-dominated rangeland could be grazing intensity, but fire frequency could be a factor that influences how resilient the system is to changes in grazing (i.e., how much disturbance it takes to shift a grassland system to a shrub-dominated system).

    State variables. Characteristics such as abundance, composition, or some ecosystem function are state variables that describe responses of the system. For example, crossing a threshold from clear to turbid water in lake systems brings about a sudden, large, and dramatic change in the responding state variables: the species shift from macrophytes to algae. Another example of system state variables is grass- or shrub-dominated rangeland. State variables are also sometimes called fast variables because they respond to changes in a controlling variable.

    Stochastic dynamics are nonequilibrium with few relationships between environment and system condition (i.e., there is no well-defined controlling variable). Often these systems have been found to be controlled mainly by random processes, chance events, or climate variation.

    Threshold. A breakpoint between two states of a system. A characteristic feature of a threshold is a change in system feedbacks. As resilience declines, the amount of disturbance needed to cross the threshold declines. Restoration thresholds indicate breakpoints that need to be addressed by restoration efforts for recovery to occur, and degradation thresholds indicate the point where environmental change precludes recovery to the same state without management or restoration actions.

    Threshold dynamics. The term alternate states is commonly used to describe the phenomenon whereby systems can exhibit a big change from one kind of regime to another. However, this terminology can be confusing because this would include what truly are alternate stable states (i.e., two or more stable point attractors separated by unstable thresholds) and the various other kinds of big changes that systems may exhibit. To avoid confusion over terminology, we suggest using threshold dynamics to include all the various kinds of thresholds and multiple system regimes that occur, including alternate stable states. Thus, these models exhibit threshold behavior and can (but do not need to) indicate alternative stable states. For convention in light of state-and-transition models, we still use states to specify the regimes indicated by any of this class of models (not solely restricted to alternative stable states). The terms regime shift and regime (Scheffer and Carpenter 2003) also describe these general threshold dynamics.

    Application of Ecosystem Dynamics to Restoration

    Over the past 100 years, extensive work has documented how communities and ecosystems change in response to disturbance, external changes, or other types of perturbations. Despite the extensive documentation of patterns, there is no one agreed-on general conceptual framework concerning the controls on species turnover and ecosystem development, and new frameworks are still being proposed. Over the past decades, the focus of these models has expanded from assuming gradual continuum dynamics to models that incorporate alternative trajectories, thresholds, and stochasticity (fig. 1.1). This expansion has generated much interest in the synergistic potential among the models and restoration, as evidenced by the many recent reviews on the topic (Chapin et al. 2004; Mayer and Rietkerk 2004; Suding et al. 2004; Bestelmeyer 2006; Briske et al. 2006; Groffman et al. 2006; King and Hobbs 2006).

    An early classic view that has been used to guide restoration was one of gradual linear change along a continuum (Clements 1916; Odum 1969; Pickett and McDonnell 1989), similar to succession toward a single climax state. One of the first conceptual models in restoration (Bradshaw 1984) depicted restoration as following this continuum model, hitting a single target restoration goal along a linear pathway (fig. 1.2). By understanding succession, this model implies that it is possible to predict, control, and perhaps accelerate community recovery after disturbances (Bradshaw 1987; Luken 1990; Dobson et al. 1997). It also assumes that there is an ultimate climax state that can be identified as the end point of the restoration effort. While a continuum approach has widespread appeal because it implies that we can predict and guide change in a system, it may presume that restoration efforts follow overly simple or unrealistic pathways in some cases (Lindig-Cisneros et al. 2003; Suding et al. 2004; Young et al. 2005; King and Hobbs 2006).

    The idea that communities can develop into alternative stable states rather than into a single climax state was first proposed by Lewontin (1969). These models predict threshold dynamics, with a small change in environmental conditions causing an abrupt change in ecosystem function and/or community structure (fig. 1.1c). Because multiple states can exist given similar environmental conditions, feedbacks are important to maintain a system in particular state. Because of these feedbacks, the trajectory of shift from one state to another will differ from the trajectory required to return to the original state, leading to the possibility of an irreversible collapse. While well documented theoretically, the existence of alternative stable states in ecological systems has met with much debate (Sutherland 1974; Connell and Sousa 1983; Grover and Lawton 1994). It has proven difficult to test for the existence of alternative equilibria empirically (Sutherland 1974; Connell and Sousa 1983; Grover and Lawton 1994; Petraitis and Latham 1999) because of criteria (e.g., stability, population turnover) that are often hard if not impossible to meet in natural systems. While there have been some recent successful demonstrations (see Schröder et al. 2005), rigorous tests of whether a degraded system truly represents an alternative and stable equilibrium are difficult and beyond the scope of most restoration efforts.

    e9781610911382_i0003.jpg

    FIGURE 1.1. Three conceptual models depicting the shift between a native state (or otherwise restoration goal) to a degraded state might occur along an environmental gradient (adapted from Scheffer and Carpenter 2003): (a) gradual continuum model; (b) one that predicts thresholds; and (c) a more specific type of threshold model, alternative stable states. In (c), the dotted line denotes an unstable equilibrium where positive feedbacks push the system to one or the other state, and the arrows denote different pathways of degradation and recovery. We do not show another model, stochastic dynamics, as it predicts highly variable dynamics without equilibria.

    The importance of stochastic or nonequilibrium dynamics gained widespread acceptance throughout many branches of ecology in the 1970s and 1980s (Pickett et al. 1987; Luken 1990). Empirical evidence from a variety of systems has shown that disturbance type, biological legacies, and chance can create multiple trajectories and influence rates of change (Drury and Nisbet 1973; Coffin et al. 1996; Pickett et al. 2001). This perspective acknowledges that succession can be unpredictable with no tendencies toward any one permanent state (Zedler and Callaway 1999; Bartha et al. 2003; Benincà et al. 2008). It also acknowledges that stochastic dynamics influence the assembly of most species and that the degree to which stochastic forces influence assembly might be predictable based on environmental or regional processes (Chase 2003, 2007).

    e9781610911382_i0004.jpg

    FIGURE 1.2. (a) Example output from a gap model of succession (reprinted with permission from Shugart et al. 1988), illustrating the continuum linear view of succession. (b) Similar model in restoration (adapted from Bradshaw 1984; Dobson et al. 1997) depicting restoration with ecosystem structure and function developing along a linear pathway.

    In addition, the early 2000s brought increased attention to regime shifts or dynamic regime models (Scheffer et al. 2001; Foley et al. 2003; Scheffer and Carpenter 2003; Collie et al. 2004; Mayer and Rietkerk 2004). Based on seminal work from Holling (1973), these models describe complex threshold dynamics with or without alternative states, avoiding much of the empirical difficulty associated with specific predictions of alternative stable state models and still describing many of the dynamics applicable to restoration, such as resilience (Gunderson 2000; Carpenter et al. 2001), adaptive capacity (Elmqvist et al. 2003), and feedbacks (Mayer and Rietkerk 2004) at multiple scales.

    While the idea of alternative states was discounted for several decades in much of ecology, rangeland ecology adopted many of these ideas in the late 1980s in the form of state-and-transition models (fig. 1.3) (Westoby et al. 1989; Friedel 1991). A result of dissatisfaction with classic successional approaches to range condition assessment, state-and-transition models of rangeland vegetation dynamics split changes in rangeland systems into discrete states and describe processes that cause transitions between states (Bestelmeyer et al. 2003; Briske et al. 2003). For example, overgrazing enhances the survival of woody vegetation. Reduction of grazing intensity is not sufficient to restore the system to a healthy rangeland once this transition occurs (Friedel 1991); burning is needed to remove woody plants (Westoby et al. 1989). State-and-transition models have altered the general idea of rangeland management, refuting the general dogma that removing grazing from overgrazed rangeland is sufficient for recovery. Most state-and-transition models describe states, transitions, and thresholds largely qualitatively, although quantitative approaches are also possible (Allen-Diaz and Bartolome 1998; Jackson and Bartolome 2002). Rangeland ecology remains a leader in developing these ideas, with recent newer models proposed such as the spiral of degradation (fig. 1.3b) as well as the increased recognition of the importance of stochastic dynamics (Fynn and O’Connor 2000; Jackson and Bartolome 2002).

    Shallow lakes are another system where these ideas of alternative states and thresholds gained credence in the late 1990s (Scheffer et al. 1997; Bachmann et al. 1999; van Nes et al. 2002). Lakes can exist in a state either where the water is clear and rooted plants are abundant or where the water is turbid and phytoplankton are abundant. In the clear lakes, rooted plants stabilize the sediment, reducing turbidity, and provide refuges for fish that eat phytoplankton. However, if the plants are removed or if fishing pressure is high, turbidity blocks light and resuspends sediment for phytoplankton, causing a rapid and dramatic shift (Moss et al. 1996; Carpenter et al. 1999). A turbid lake can be restored by manipulating the feedbacks that maintain the system in the turbid state: increasing the population of fish that consume phytoplankton, decreasing the number of predators that eat the phytoplankton-consuming fish, reducing nutrient loading, and installing wave barriers to create refuges for plants (Bachmann et al. 1999; Dent et al. 2002).

    e9781610911382_i0005.jpg

    FIGURE 1.3. (a) An example of a state-and-transition model that groups communities into discrete states and describes processes that cause transitions between states (modified from Westoby et al. 1989). (b) The spiral of degradation conceptual model proposed by Whisenant (1999) and expanded by King and Hobbs (2006) that describes the interactive dynamics involved in degradation, characterizing them as a stepwise process with feedback. As applied to restoration, the goal is to generate feedback loops that will initiate and promote the continued improvement of ecosystem components, reversing the direction of the arrows of degradation through autogenic recovery.

    Recently, there has been an expansion of threshold and alternative state ideas to the general challenge of restoration of degraded land beyond rangeland and lake ecosystems. One conceptual model that has resonated widely is a two-threshold model where the first threshold denotes changed biotic interactions and the second, further down the degradation pathway, denotes changed abiotic limitations (fig. 1.4a; Hobbs and Harris 2001). Cramer and Hobbs (2005) modified this model by including multiple processes that interact to affect resilience and lead to degradation (fig. 1.4b). These and many other related conceptual models have been used as heuristic devices to guide restoration efforts and often prove to be consistent with land managers’ perception of the restoration process (Wallington et al. 2005). We focus on this expansion of models in this book.

    Ecosystem Models and Restoration Approaches

    Ecosystem models can be crucial decision-making tools in restoration and land management. For instance, in restoration, it is important to recognize when ecological systems are likely to recover unaided and when they require active restoration efforts. This assessment involves the understanding (or estimation) of the recovery trajectory and identification, if any, of restoration thresholds that serve as barriers to prevent the recovery of degraded systems. These barriers can result from biotic factors (e.g., weed invasion, herbivory, lack of pollination) or abiotic factors (e.g., changes in hydrology or soil structure and processes) (fig. 1.4). Conceptual models of ecosystem dynamics can aid in this understanding and may reduce the risk of unpredicted or undesired change in restoration projects, suggesting ways to correctly diagnose ecosystem damage identify restoration constraints and develop corrective methodologies that aim to overcome such constraints.

    e9781610911382_i0006.jpg

    FIGURE 1.4. (a) An influential view about how the state-and-transition framework can be applied to restoration ecology (modified from Hobbs and Harris 2001; Whisenant 1999). Primary processes are fully functional in intact ecosystem state (far left). When the system crosses the transition threshold controlled by biotic interactions, most primary processes are still functional, and recovery requires vegetation manipulation (e.g., invasive species removal). However, when the system crosses the threshold primary controlled by abiotic limitations, primary processes are nonfunctional, and recovery requires modification of the physical environment. (b) Another state-and-transition model of ecosystem dynamics where the balls represent vegetation states and the arrows transition between states (based on Cramer and Hobbs 2005). This representation incorporates two process dimensions in the degradation sequence (in this case of salinity, salt accumulation, and waterlogging) and the degree of reversibility of the transition (the depth of the depression in which the ball sits).

    While there are many types of ecosystem models and many ways to distinguish among the different types, three are particularly applicable to restoration ecology (table 1.1). First, continuum models describe dynamics without thresholds, where a change in the environmental controlling variable is more proportional to the system response. Second, stochastic models describe highly variable nonequilibrium relationships between system response and environment. These two types of models stand as alternatives to the third group, which constitutes the new models reference in the book title, threshold or regime shift models. These describe abrupt changes with small changes in environmental conditions. In the previous section, we described some of the historical developments related to these models; in this section, we describe the main restoration approaches that would be taken based on the different type of ecosystem dynamics.

    Gradual Continuum Models

    These models assume that systems respond in a continuous manner to environmental change and return to their predisturbance state or trajectory following disturbance (table 1.1). These models predict a classical successional trajectory: steady, directional change in composition to a single equilibrium point (Clements 1916; Odum 1969) with perturbations causing shifts along a common trajectory. Recovery is seen as a predictable consequence of interactions among species with different life histories and the development of ecosystem functions. Strong internal regulation occurs through negative feedback mechanisms, including competition and herbivore–predator interactions, as well as climate–ecosystem couplings and life history trade-offs.

    In cases where these models apply to restoration, efforts can be designed from a perspective of initiating or assisting succession (Prach 2003; Sheley and Krueger-Mangold 2003). In some cases, community development can proceed spontaneously, relatively unassisted, to reach desirable target states (Prach et al. 2001; Khater et al. 2003; Novak and Prach 2003). In other cases, restoration can take this approach to spur recovery along a successional trajectory through the use of an understanding of positive and negative species interactions to either accelerate rates of change (Luken 1990; Choi and Wali 1995) or identify times when change is slowed and intervention is needed (Mullineaux et al. 2003). Restoration effort can then accelerate natural succession so that the ecosystem develops along the same trajectory as it would in the absence of intervention but reaches the goal end point sooner. For instance, prescribed burning of degraded grasslands can promote restoration, particularly if applied according to historical patterns (Baer et al. 2002; Copeland et al. 2002), and reinstating the original flow regime of a severely degraded river can spur recovery of the surrounding plant communities (Lytle and Poff 2004).

    TABLE 1.1 Comparisons of four types of ecosystem models, with patterns and tests consistent with each type. Alternative stable states are specific types of threshold models. See glossary for definition of terminology.

    e9781610911382_i0007.jpg

    Stochastic Dynamics

    Nonequilibrium theory assumes that external factors (e.g., climate, pollution) play a larger role in the behavior of ecosystems than do internal processes, such as competition and predation, and predicts divergent, cyclic, or arrested trajectories that never arrive at a common state. There are many examples that support this view. For instance, following the eruption of Mount St. Helens, in Washington, USA, there were variable rates of recovery along several distinct pathways (Franklin and MacMahon 2000) where chance colonization determined changes over time (del Moral 1998, 1999; Walker and del Moral 2003). In arid rangelands, large fluctuations in precipitation prevent herbivores from regulating primary production, thereby minimizing negative feedbacks that would cause equilibrium behavior (Ellis and Swift 1988; Jackson and Bartolome 2002). Stochastic effects resulting from isolation and dispersal limitations have also been shown to override the deterministic effects of competition (Underwood and Fairweather 1989; del Moral 1998; Foster et al. 1998). As seed limitation may be a very common barrier in degraded systems, stochasticity associated with dispersal may be important to consider in restoration projects (Young et al. 2005).

    Threshold Models

    There are several types of models that predict threshold dynamics. Two that are particularly applicable to restoration are dynamic regime or threshold models and a particular subset of these, alternative stable state models (table 1.1). While we expect that it may be impossible to identify any more specific behavior than whether threshold dynamics may occur in much restoration work, the distinction between these models is important heuristically. Alternative stable states refer to a specific case in which there are two or more stable point attractors separated by unstable equilibria. In this case, the pathway of degradation differs from recovery (termed hysteresis), which may cause transitions that are particularly hard to reverse.

    As a group, threshold models provide important insights into system dynamics that can be applied to the recovery of degraded systems (Hobbs and Norton 1996; Prober et al. 2002; Lindig-Cisneros et al. 2003; Suding et al. 2004). In particular, restoration needs to consider positive feedbacks that can make the degraded state resilient to restoration efforts. Restoration is often dealing with the challenge of nonnative species, which can have distinctive traits that change ecosystem processes (Ehrenfeld 2003). If these effects are beneficial to the modifying species, they can produce positive feedbacks that make it resilient to restoration efforts. For example, introduced grasses in woodlands in Hawaii alter nitrogen cycling and promote fire, which further benefits introduced grasses at the expense of native shrub species, creating an internally reinforced state that has proven very difficult to change (Mack et al. 2001).

    Instead of a focus on species differences in competitive ability and abiotic tolerance forming predictable species turnover, threshold models often assume other, less predictable types of species turnover (Bakker and Berendse 1999; Gunderson 2000). These include priority effects, asymmetric competitive interactions where one resident species is able to exclude an invading species and vice versa (D’Antonio et al. 2001; Seabloom et al. 2003; Lulow 2006; Prober et al., chapter 11, this volume), trophic triangles (Carpenter and Scheffer, this volume), sequential species loss effects (Eriksson and Eriksson 1998), and shifts in competitive rankings due to alteration of disturbance regimes (Dudgeon and Petraitis 2001; Suding 2001).

    Changes in environmental conditions such as climate or pollution, even when their rates of change are relatively gradual, can push a system past a threshold (van Auken 2000; Holmgren and Scheffer 2001; Cione et al. 2002). These changed external factors do not respond quickly to changes in local population abundance or community structure and consequently are often harder to manage within any given reserve or project area. Thus, the possibility of a threshold event is important to consider even if the system does not appear to be tracking the external changes (Carpenter et al. 1999).

    Determining System Dynamics

    It is a challenge to predict ecosystem dynamics, particularly in a degraded system and in response to management actions. Our goal in this section is to lay out possible ways to look for different types of ecosystem dynamics, with the understanding that some ways (or combinations of ways) are better than others. Quantitative-based analysis of dynamics may be neither possible nor appropriate in all cases, particularly in actual restoration situations. Thus, we expect that few practical restoration projects will have the capacity to test for these dynamics following stringent criteria. Indeed, it may be that threshold models are used predominantly as heuristic devices, both for hypothesis testing and for management decision making. While empirical tests certainly have high value in understanding system dynamics, other approaches are also very valuable. Dynamics can be inferred through use of indicators, gathering and utilizing expert knowledge, and consideration of cultural drivers.

    There are many ways to discern whether a system’s dynamics are consistent with threshold models (Scheffer and Carpenter 2003; Schröder et al. 2005; Schröder, this volume), several of which are workable to people collecting and interpreting results from restoration projects. We acknowledge that a fine balance needs to be met in testing theory in restoration: While feasibility in restoration may make some suggestions weak, in that they might not completely meet stringent criteria of the theoretical models, it is crucial to not contribute to the misperceptions and oversimplifications that may accompany increased application of these models.

    Observationally, threshold patterns can be discerned in time-series analyses, indicated by abrupt transitions over time, or spatial analyses, indicated by sharp spatial boundaries in the absence of associated sharp discontinuities in environmental parameters. For example, Dodd and coworkers (2006) monitored squirrel density on nine sites that varied in proportion of unlogged high-quality forested habitat and found evidence for thresholds in squirrel density when high-quality habitat was reduced past 42%. Spatial clumping patterns can also be assessed in a multivariate approach, often used in state-and-transition models (Allen-Diaz and Bartolome 1998; Jackson and Bartolome 2002; Bartolome et al., this volume). Jackson and Bartolome (2002) used a multivariate classification approach to produce a state-and-transition model of vegetation across years, sites, and grazing manipulation. Their model indicated that seasonal weather patterns are the primary determinant of community change rather than grazing management, suggesting that management by manipulating grazing intensity may not be effective. Lastly, recent work suggests that a system will likely exhibit more variance (Carpenter and Brock 2006) and a slower rate of recovery to a perturbation (van Nes and Scheffer 2007) as it approaches a threshold. These approaches may prove to be useful to predict the risk of crossing a threshold before it actually occurs.

    Manipulative experiments offer direct evidence for the presence or absence of predicted dynamics in a system, but they are rarely conducted and sometimes are ambiguous in design and interpretations (Petraitis and Dudgeon 2004; Schröder et al. 2005). Schröder et al. (2005) detail tests to determine whether systems follow alternative state dynamics; several of these tests are applicable in restoration research. Priority effects can be tested when all interacting populations are started at different density ratios (Chase 2003; Scheffer et al. 2003) or added in different sequences (Chase and Leibold 2003). These experiments may consist of assembling target and degraded communities and testing whether species of each can invade each other (if they can, it would not be evidence of alternative stable states). Another test of alternative states would be whether replicate plots, such as restoration monitoring plots, consistently diverge on different trajectories over time. Divergence could also be due to random stochasticity, so care needs to be taken to demonstrate a consistent pattern of divergence. Lastly, tests of positive feedbacks, the mechanism by which systems are driven to different states given similar initial conditions, may help identify system dynamics. Plant–soil feedbacks may be particularly important in restoration work; a soil-culturing experiment can test whether species effects on soil are advantageous to the modifying species (Bever 2003).

    Although management decisions are generally conducted at much larger scales, many of the previously mentioned tests requiring experimentation are often conducted at smaller spatial scales (Carpenter and Scheffer, this volume). Small-scale experimentation can have substantial value in restoration ecology because it can inform managers or restoration ecologists about potential system behavior, knowledge that then can be incorporated into adaptive learning processes to broaden both the set of models that managers use and ideas about the large-scale, long-term behavior of the system. However, there is also risk involved in extrapolating results from small-scale experimentation to large-scale management. It will be important to use large-scale evidence to confirm small-scale experimentation whenever possible. Additionally, multiple overlapping processes influence ecological systems that are not entirely separable. It is important to note that there are many lines of evidence supportive of model dynamics, but there is no silver bullet, and it is wise to make decisions based on several types of evidence (Holling and Allen 2002).

    Relevance to Restoration: Can Restoration Projects Benefit from These Models?

    There is great potential to link models of ecosystem dynamics to more applied problems in restoration. However, theoretical models do not provide simple or universal answers for the challenges that confront restoration. Knowledge that systems may express continuum, stochastic, and threshold dynamics can inform restoration, but the knowledge needs to be supplemented with consideration of socioeconomic constraints and historical understanding specific to the system in question (Zedler 2000; Sutherland 2002). While experimentation and monitoring of the responses to management can provide evidence as to possible dynamics, utilizing inference and expert knowledge is also essential.

    While evidence is increasing that thresholds are important in some systems, many others exhibit nonthreshold dynamics. Currently, no clear generally applicable method exists for recognizing where thresholds are likely to be important. This results in two undesirable scenarios in which either (1) there is a hands-off approach to restoration or species recovery and nothing happens because there are unidentified thresholds or (2) active intervention is carried out when, in fact, the ecosystem or species could have recovered adequately

    Enjoying the preview?
    Page 1 of 1