Elsevier

Environmental Modelling & Software

Volume 75, January 2016, Pages 318-332
Environmental Modelling & Software

Modelling systemic change in coupled socio-environmental systems

https://doi.org/10.1016/j.envsoft.2015.10.017Get rights and content

Highlights

  • We demonstrate and argue that systemic change pushes the model boundary by changing the language used to describe the system.

  • We briefly review the history of modelling systemic change in coupled socio-environmental systems.

  • A framework highlighting challenges in modelling systemic change is used to introduce papers in this thematic issue.

Abstract

Abrupt systemic changes in ecological and socio-economic systems are a regular occurrence. While there has been much attention to studying systemic changes primarily in ecology as well as in economics, the attempts to do so for coupled socio-environmental systems are rarer. This paper bridges the gap by reviewing how models can be instrumental in exploring significant, fundamental changes in such systems. The history of modelling systemic change in various disciplines contains a range of definitions and approaches. Even so, most of these efforts share some common challenges within the modelling context. We propose a framework drawing these challenges together, and use it to discuss the articles in this thematic issue on modelling systemic change in coupled social and environmental systems. The differing approaches used highlight that modelling systemic change is an area of endeavour that would benefit from greater synergies between the various disciplines concerned with systemic change.

Introduction

The collapse of ecosystems and the global financial crisis have much more in common than one may think at the first glance (Scheffer, 2009). Not only may these abrupt systemic changes be driven by internal and external processes of a similar nature, the system's reactions and early warning signals that indicate such changes may also share the same characteristics (Scheffer et al., 2009). While there has been much attention to studying regime shifts in ecological and economic systems independently, the attempts to do so for coupled socio-environmental systems (SES) are scarcer. (We deliberately use the term socio-environmental systems here, with a view to being as general as possible, though social-ecological systems are very much in view.) Understanding systemic change in coupled systems requires insights not only into the processes at macro and micro levels in both socio-economic and environmental subsystems but also into the role of feedbacks between them. Models can be instrumental here. This special issue aims to elicit and discuss the challenges of modelling systemic changes in coupled SES and point towards ways to address them by presenting recent examples of simulation models of systemic change. We begin our introduction to the issue using three challenges (terminology, structural change and subjectivity) as a basis for introducing systemic change in coupled socio-environmental systems.

One of the first obstacles in the study of systemic change in SES is terminology.1 Various disciplines in both the environmental and social sciences have engaged with relevant ideas – regime shift, structural change, non-marginal change, transition theory to name a few – and each claims ownership over their tokens. While ordinary mortals squabble over land and resources, in academia the territories are linguistic. For the purposes of introducing this thematic issue, we use the term systemic change, and include in Box 1 a brief glossary of terms. As modellers, we are interested in systems (though even this term is claimed), whether they are represented or analysed using equations, probability distributions, algorithms or any other formal approach. Systemic changes involve fundamental changes to the way in which a system is structured, covering such things as:

  • new classes of entity being formed, or new types of relationship between them;

  • the introduction of new processes and changes in feedback loops;

  • changes to the set of exogenous variables to which the system is sensitive;

  • other changes to the relevance of variables in or affecting the system;

  • the reorganisation of networks of interaction, possibly entailing different interaction topologies;

  • abrupt (step-wise) changes in functions or parameters describing the system.

All these may be needed to represent exogenous change, or endogenous evolution that comes as a result of the formation of new institutions, rules or norms governing behaviour.

If we conceive systemic change as going from one system α to another, β, then in comparison with models of system α exclusively, models simulating systemic change from α to β entail, to some degree or another, redefinition of system boundaries and pathways through which the social system interacts with its environment. Differences between the two kinds of model may also include appropriate temporal and spatial system resolutions and extent.

Systemic changes may arise through exogenous disturbances to a system or emerge endogenously either through the behaviour of the system itself, or through gradually accumulated responses of the system to relatively small exogenous perturbations (Walker and Meyers, 2004, Biggs et al., 2009, Carpenter et al., 2011). Systemic changes may be coupled with a collapse in existing (formal and informal) institutions, loss of key hubs in interaction networks, irrelevance of prior classification criteria, or entities no longer interrelating in a particular way. To consider a rather extreme example, the French revolution involved the collapse of the monarchy, the execution of much of the aristocracy, the irrelevance of feudal social stratification, and with that, at least in principle, an end to social interactions based on a presumption of inequality. Since such changes may themselves be seen as disturbances, a systemic change can also be understood as the propagation and amplification of a disturbance throughout the system, leading to a long-term change in the way the system is organised. All these issues pose challenges for modelling, not least because, in extreme cases, they may involve a fundamental shift in the vocabulary used to describe the system, which will be reflected in the model's ontology. For example, in equation-based models systemic change implies that not only parameters' or variables' values change but the entire functional forms used to relate them in the model transform, possibly with new variables and new processes being introduced and old ones being deleted.

More formally, a model of a system may be conceived as a triplet consisting of (L) a formal language describing the possible states it can have, (E) expressions in that language describing the specific state it currently has (such as, the existence of a particular entity, values the entity has for its variables, and the other entities it interacts with), and (P) algorithms for computing subsequent state(s) of the system given previous state(s). Systemic change as represented in the model is a change to combinations of L, P, and ‘major changes’ to E, each of which will be referred to as ΔL, ΔP and ΔE systemic changes respectively. In the case of ΔE, a systemic change occurs when a significant number of the entities in the system are replaced with new entities, but ones of the same types, interacting in the same way as before. In ΔP, the systemic change affects the way the system evolves. In ΔL, it is the whole vocabulary used to describe the system that changes. (See Box 2.) Notably, systemic changes are not necessarily associated with a ‘shock’ or disturbance – they can occur through the gradual evolution of the system, so are also relevant to those who do not believe in discontinuities in natural systems (the natura non facit saltus axiom). Gradual changes in a system's elements and micro-level processes may drive a system over a critical point when irreversible and significant macro-level structural changes occur.

Another major challenge for modelling systemic changes is that they – by definition – involve fundamental changes in system behaviour and structure that are often unknown beforehand. The promise of predictive modelling, however, is based on the assumption that the trend along which a system was developing in the past can be, with acceptable confidence, extrapolated into the future. We build our models based on what we know about the systems in the past. This is well recognized for statistical or data-based models. The advantage of process-based models, supposedly, is that if we are correctly describing processes then we can predict how systems will behave in the future. This is true if the systems do not change structurally. However in environmental sciences in most cases they do (Milly et al., 2008); especially if the human factor is a part of the system. A modeller needs to keep in mind that SES are complex adaptive systems, constantly changing and featuring nonlinear, self-organizing and out-of-equilibrium dynamics (Arthur et al., 1997, Folke, 2006). Thus, if our models rely only on the past behaviour of the system – which may not contain any period encompassing a systemic change – and do not account for how SES may evolve in the future, we cannot expect that our predictions to be true. This issue also closely relates to the notion of equilibrium. While it may be convenient to conceptualise the existence of a unique equilibrium that delivers Pareto efficient allocation, there may in fact be multiple equilibria or no equilibrium at all. While shifts between multiple equilibria or steady states, particularly in modelling of ecological regime shifts, have been subject of ecological research for several decades (Scheffer et al., 2001), traditional economic – and even environmental economic – models deal with marginal changes around an equilibrium (Stern, 2008). In other words, they are designed to explore marginal changes along a certain point or trend, in many cases reconstructed from historic data. Yet, when focussing on systemic changes in SES – such as social-ecological regime shifts – the emphasis is on exploration of the transient dynamics, which may be at times abrupt (Hughes et al., 2013).

The challenge of change in structure is covered extensively in the works of Beck, 2002, Beck, 2005, Beck, 2009. Beck's discussion of the issue is conceptualised around a classical mechanics representation, but an example using an agent-based model is shown in Box 2. Beck (2002) makes a distinction between ‘apparent’ and ‘true’ (evolutionary) structural change. To Beck, apparent structural change is largely a matter of ignorance: a model doesn't contain enough equations to capture all the dynamics in the real world, and its users are ‘surprised’ when the system behaves in a way the model cannot describe. True structural change (Beck, 2009) is the breakdown and restructuring of a system. The functions modelling the ‘before’ and ‘after’ systems have different domain and range, parameters, and links between the states. Using f0 to caricature the ‘naïve’ system of functions,2 Beck (2005) contrasts this with f1, a refined function including more phenomena. This is, to Beck, apparent structural change – when f0 was built, there wasn't enough information to build f1; the difference between f0 and f1 is represented in f1 alone by the model being in different subvolumes of state space. Further information and refinements create f2, f, and so on; at f we (never) reach the “truth of the matter” (Beck, 2005, p. 655). However, by induction, arguably all changes on the path to f are ‘apparent’, ruling out any ‘true’ structural change. Beck (2005) would be comfortable with this: in footnote 4 on p. 655, he effectively characterises ‘true’ systemic change as a change to the dimensionality of the state space over which f operates. Adding new variables to the state space, domain or range of f, however, just seems to be reflecting a different form of ignorance from that leading to adding new functions to f.

As defined earlier, systemic change can arise from change in structure (ΔL and ΔP) as well as ‘significant’ change in state (ΔE). Beck's (2002) ‘apparent’ structural change corresponds closely with ΔP (adding new links in the network of possible state transition functions). ‘True’ structural change, in Beck's terminology, is more related to what we have called ΔL systemic change. Change in state ΔE would not be structural change under Beck's formalism, but note that it is reflected in f1's trajectory operating in a superset of the state space that f0 can, which may explain his choice of the term ‘apparent’ to reflect the ΔP systemic change from f0 to f1. Whilst we agree that ΔL systemic change is more significant, for our purposes, since systemic change is about change in the language used to describe systems, ΔE and ΔP systemic change are no less ‘true’. Indeed, the example in Box 2 shows how the presence of ΔE systemic change in a model can suggest a need for ΔP and ΔL.

As the arguments above suggest, there are unavoidable subjectivities in how systemic changes are understood and explained that will be uncomfortable to traditional scientific thinking. Clearly the language describing a system and the algorithms describing its dynamics are themselves sources of subjectivity. With more knowledge and data, a different language or set of algorithms might have been used that would have accommodated a systemic change. This subjectivity is also closely related to the system boundaries and scales: what may be considered a systemic change in the short term, may revert to the previous dynamics in the long term; what may be seen as dramatic and systemic by one, may be seen as normal and within bounds by another. The temporal perspective on systemic change is considered in more detail by Beck (2005), since one source of the information that leads from f0 to fn is deeper knowledge of the states the system has had in the past. Subjectivity also arises from the uses to which a model would be put, and given a requirement to fit the recent past and predict to the near future, many modellers would prefer f0 for reasons of parsimony (higher orders of f would be ‘overfitting’).

Subjectivity with respect to systemic change also has an emotional dimension in that negative connotations may be associated with the immediate aftermath of a shock and those who have suffered from it, but systemic change can also be part of positive transformations (Folke et al., 2010). Examples might include the transition to democratic modes of governance or a shift in consumer preferences to environmentally friendly goods or low-carbon energy sources, though even in these cases, not all will agree they are necessarily changes for the better. The contrast between different perspectives on the same event can be seen in an ecological case: the local extinction of one species presents an opportunity for others. This means that given our inherent loss-aversion (Kahneman, 2011) we will be more likely to pay more attention to modelling the changes that we may see as negative rather than those that we deem as positive, potentially limiting the range of outcomes that could be considered.

Formal models, and especially simulation models, play an essential role in understanding systemic changes in coupled SES. Theoretical and conceptual models (those not necessarily fitted to empirical data) have been fundamental in studying ecological (Carpenter et al., 1999, Gunderson and Holling, 2002) and social-ecological regime shifts3 (Lade et al., 2013), and in developing early warning signals for approaching regime shifts (Scheffer et al., 2009, Dakos et al., 2015). Theoretical models are instrumental in exploring potential drivers and interventions to achieve desirable future SES states (Biggs et al., 2009, Polasky et al., 2011, Levin et al., 2013). Such exploration is especially vital in the modern interconnected world where regime shifts in SES are often not just a problem for a single region or ecosystem but increasingly become a global common good problem. Advances in data-driven computational models are also being made in studying the emergence of nonlinearities in SES (Vespignani, 2012), but may experience limitations when there are no data on anticipated future states of the system. Given the acknowledged challenges in integrated modelling of coupled SES (Voinov and Shugart, 2013), modelling fundamental changes in their structure may need to push current SES modelling practices beyond the state-of-the-art. This could possibly be achieved by learning from modelling of systemic changes in SES experienced by ancient civilizations (Axtell et al., 2002, Heckbert, 2013), although behavioural sub-models may still need to rely on stylized theory grounded in ethnographic records (Janssen, 2009).

This thematic issue4 of Environmental Modelling and Software focuses on modelling systemic change in coupled SES. The goal of the thematic issue is to provide a systematic overview on how model design and analysis is (or should be) aimed at representing the processes and consequences of systemic changes in SES, supported by modelling examples. The papers contributing to this thematic issue address questions of model design, and how this needs to be approached differently to cater for systemic change in coupled SES. We attempted to solicit contributions that cover a variety of modelling techniques, and include case studies from diverse geographical regions. It is therefore gratifying that the articles in this thematic issue feature case studies in the “West” as well as the Global South, and in temperate and tropical zones. There is also a variety of modelling approaches, including Bayesian and System Dynamics methods, as well as Agent-Based Modelling.

In this article, we briefly consider the history of modelling systemic change. We then propose a framework summarising what we see as the key points to discuss in modelling systemic change, and use it to introduce the papers in this thematic issue, before concluding with a discussion of our findings.

Section snippets

Modelling systemic change: A brief history

Modelling systemic or structural change in socio-environmental systems is not new. In ecology the dominant view that ecosystems are in or developing towards a single global equilibrium was challenged with the discovery of multiple stable states in ecosystems (e.g. Holling, 1973). This has led to a growing body of research investigating the transitions between alternative states (Scheffer et al., 2001). Efforts in analysing systemic change in socio-economic, social and socio-environmental

The contributions to this thematic issue

In this section, we develop a framework to discuss the modelling contributions to this thematic issue, drawing on the review of Filatova et al. (2015). After a brief summary of the contributions to this special issue we then apply it to analyse how the different models address systemic change.

Discussion and concluding remarks

Modelling the co-evolving dynamics of complex coupled SES to anticipate the potential outcomes of exogenous shocks, or to study the emergence of endogenous systemic change therein, is ambitious. The most serious challenge in modelling systemic changes in SES concerns the restructuring of a model that may be required if a system it represents experiences a systemic change (Beck, 2005). Contemporary models based on systems of differential or difference equations and dynamical systems theory

Acknowledgements

The authors are grateful to the financial support of the Scottish Government Rural Affairs and the Environment Portfolio Strategic Research Theme 1 ‘Ecosystem Approach’ (JGP), the Netherlands Organisation for Scientific Research (NWO) VENI grant no. 451-11-033 (TF), the European Research Council under the European Union's Seventh Framework Programme (FP/2007-2013)/ERC grant agreement no. 283950 SES-LINK and a core grant to the Stockholm Resilience Centre by Mistra (MS), and the European Union's

References (98)

  • S.E. Jorgensen

    Chapter 2 – structurally dynamic models of lakes

    Dev. Environ. Model.

    (2014)
  • D. MacDonald et al.

    Agricultural abandonment in mountain areas of Europe: environmental consequences and policy response

    J. Environ. Manag.

    (2000)
  • P. Mulder et al.

    Structural change and convergence of energy intensity across OECD countries, 1970-2005

    Energy Econ.

    (2012)
  • D.C. Parker et al.

    Complexity, land-use modeling, and the human dimension: fundamental challenges for mapping unknown outcome spaces

    Geoforum

    (2008)
  • S. Polasky et al.

    Optimal management with potential regime shifts

    J. Environ. Econ. Manag.

    (2011)
  • J. Robinson

    Future subjunctive: backcasting as social learning

    Futures

    (2003)
  • M. Scheffer et al.

    Catastrophic regime shifts in ecosystems: linking theory to observation

    Trends Ecol. Evol.

    (2003)
  • A. Smajgl et al.

    Behaviour and space in agent-based modelling: poverty patterns in East Kalimantan, Indonesia

    Environ. Model. Softw.

    (2013)
  • A. Tavoni et al.

    The survival of the conformist: social pressure and renewable resource management

    J. Theor. Biol.

    (2012)
  • E. Vakhtina et al.

    Capital market based warning indicators of bank runs

    Phys. Stat. Mech. Appl.

    (2015)
  • A. Voinov et al.

    'Integronsters', integral and integrated modeling

    Environ. Model. Softw.

    (2013)
  • A.A. Voinov et al.

    Qualitative model of eutrophication in macrophyte lakes

    Ecol. Model.

    (1987)
  • R. Winkler

    Structural change with joint production of consumption and environmental pollution: a neo-Austrian approach

    Struct. Change Econ. Dyn.

    (2005)
  • K. Anand et al.

    Epidemics of rules, rational negligence and market crashes

    Eur. J. Finance

    (2013)
  • J.M. Anderies et al.

    Grazing management, resilience, and the dynamics of a fire-driven rangeland system

    Ecosystems

    (2002)
  • P.W. Anderson et al.

    The Economy as an Evolving Complex System

    (1988)
  • O. Arce et al.

    Housing bubbles

    Am. Econ. J. Macroecon.

    (2011)
  • W.B. Arthur et al.
    (1997)
  • R.L. Axtell et al.

    Population growth and collapse in a multiagent model of the Kayenta Anasazi in Long House Valley

    Proc. Natl. Acad. Sci. U. S. A.

    (2002)
  • J. Bai et al.

    Computation and analysis of multiple structural change models

    J. Appl. Econ.

    (2003)
  • W.J. Baumol et al.

    Chaos: significance, mechanism, and economic applications

    J. Econ. Perspect.

    (1989)
  • M.B. Beck

    Grand Challenges of the Future for Environmental Modeling. In the Setting of NSF's Environmental Observatories Initiatives

    (2009)
  • F. Berkes et al.

    Linking Social and Ecological Systems

    (1998)
  • R. Biggs et al.
  • R. Biggs et al.

    Turning back from the brink: detecting an impending regime shift in time to avert it

    Proc. Natl. Acad. Sci. U. S. A.

    (2009)
  • R. Biggs et al.

    Principles for Building Resilience: Sustaining Ecosystem Services in Social-ecological Systems

    (2015)
  • W.A. Brock
  • W.A. Brock et al.

    Variance as a leading indicator of regime shift in ecosystem services

    Ecol. Soc.

    (2006)
  • W.A. Brock et al.

    Discrete choice with social interactions

    Rev. Econ. Stud.

    (2001)
  • W.A. Brock et al.

    Nonlinear Dynamics, Chaos, and Instability: Statistical Theory and Economic Evidence

    (1991)
  • W.A. Brock et al.

    Managing systems with non-convex positive feedback

    Environ. Resour. Econ.

    (2003)
  • S. Carpenter et al.

    Early warnings of regime shifts: a whole-ecosystem experiment

    Science

    (2011)
  • S.R. Carpenter

    Regime Shifts in Lake Ecosystems: Pattern and Variation

    (2003)
  • S.R. Carpenter et al.

    Management of eutrophication for lakes subject to potentially irreversible change

    Ecol. Appl.

    (1999)
  • A.S. Crepin

    Using fast and slow processes to manage resources with thresholds

    Environ. Resour. Econ.

    (2007)
  • V. Dakos et al.

    Resilience indicators: prospects and limitations for early warnings of regime shifts

    Philos. Trans. R. Soc. B Biol. Sci.

    (2015)
  • V. Dakos et al.

    Spatial correlation as leading indicator of catastrophic shifts

    Theor. Ecol.

    (2010)
  • F. Eboli et al.

    Climate-change feedback on economic growth: explorations with a dynamic general equilibrium model

    Environ. Dev. Econ.

    (2010)
  • B. Edmonds

    The revealed poverty of the gene-meme analogy – why memetics per se has failed to produce substantive results

    J. Memet. Evol. Models Inf. Transm.

    (2005)
  • Cited by (41)

    • Taking a whole-of-system approach to food packaging reduction

      2022, Journal of Cleaner Production
      Citation Excerpt :

      From a shifting the burden archetype perspective, economic degrowth could be framed as a fundamental solution (Fig. 5). Systemic change is defined as a fundamental change in the way the system is structured (Polhill et al., 2016). Degrowth focuses on voluntary and fair processes of constraining production and consumption to reduce the ecological footprint of the global economy (Khmara and Kronenberg, 2018).

    View all citing articles on Scopus
    View full text