The Resilience 2011 conference was a unique opportunity to meet people and new ways of thinking about resilience. This post is dedicated to the sessions I enjoyed the most, and my research interests biased me towards sessions on regime shifts and coupled social-ecological system analysis.

As PhD student working with regime shifts, it was not surprisingly that the panel on research frontiers for anticipating regime shifts was on my top list. Marten Scheffer from Wageningen University introduced the theoretical basis of critical transitions on social-ecological systems. His talk was complemented by his PhD student Vasilis Dakos on early warnings. Their methods are based on the statistical properties of systems when approaching a bifurcation point. These are gradual increase in spatial and temporal auto-correlation, as well as variability. A perfect counterpoint to these theoretical approaches was offered by Peter Davies from University of Tasmania; who presented the case study of a river catchment in Tasmania. Davies and colleagues introduced Bayesian networks as a method to estimate regime shifts, their likelihood and possible thresholds. Victor Galaz from Stockholm Resilience Centre presented an updated version of his work with web crawlers, exploring how well informed Internet search can give early warnings on, for example, disease outbreaks. Galaz point out the role of local knowledge as fundamental component of the filtering mechanism for early warning systems.  Questions from the audience and organizers were focused on the intersections from theory and practical applications of early warnings.

While Dakos’ technique does not need deep understanding of the system under study, his time series analysis approach does require long time series. On the other hand, Bayesian networks require a deep understanding of the system and their feedbacks in order to make well-informed assumptions to design models. An alternative approach was proposed by Steve Lade from Max Planck Institute in a parallel session, who used generalized models to identify the model’s Jacobian. Although his approach does need a basic knowledge of the system, it is able to identify critical transitions with limited time series, typical of social-ecological datasets in developing countries.

Most of the work on regime shifts is based on state variables that reflect either ecological processes or social dynamics, but rarely both. Thus, I was also interesting in advances on operationalizing the concept of critical transitions to social-ecological systems in a broader sense. I looked for modeling examples where it is easier to track how researchers couple social and ecological dynamics. Here are some notes on the modeling sessions.

J.M. Anderies and M.A. Janssen from Arizona State University (ASU) presented their work on the impact of uncertainty on collective action. They used a multi-agent model based in irrigation experiments (games in the lab). Their work caught my attention because first they capture the role of asymmetries in common pool resources, which is often overlooked. In the case of irrigation systems, it is given by the relative positions of “head-enders” and “tail-enders” with different access to the resource.  Secondly, they used their model to explore how uncertainty both in water variability and shocks to infrastructure affects the evolution of cooperation.

Ram Bastakoti and colleagues (ASU) complemented the previous talk by bringing Anderies and Janssen insights to the field, particularly to cases in Thailand, Nepal and Pakistan. Batstakoti is studying the robustness of irrigation systems to different source of disturbances including policy changes, market pressure and the biophysical variability associated with resource dynamics. In the following talk, Rimjhim Aggarwal (ASU) presented the case of India, a highly populated country facing a food security challenge in the forthcoming decades; where groundwater levels are falling faster than expected. Aggarwal research explores the tradeoffs among development trajectories. His focus on technological lock-ins and debt traps as socially reinforced mechanism towards undesirable regimes makes his study case a potential regime shift example.

My colleagues from the Stockholm Resilience Centre at Stockholm University also presented interesting work on modeling social-ecological dynamics. Emilie Lindqvist uses a theoretical agent model to explore the role of learning and memory in natural resource management. Her main results point out that long-term learning and memory is essential for coping with abrupt decline or cyclic resource dynamics. On the other hand, Jon Norberg and Marty Anderies presented a theoretical agent model where social capital dynamics are coupled with a typical fishery model. Although their work is still prelimary, it was the only talk that I saw which actually coupled social and ecological dynamics.

Resilience 2011 gave me the opportunity to rethink and learn a lot about regime shifts. Although my main question: how to study regime shifts in coupled social-ecological system remains unsolved, the discussions in the panel sessions gave me some possible ways of tackling it.

The research agenda on regime shifts is strongly developing towards early warnings. Three competing methods arise:

1. look for signals in spatial and temporal data by examining the statistical properties of a system approaching a threshold: increase in variance and autocorrelation
2. acquire a deep knowledge of feedback dynamics and apply Bayesian networks to understand and predict potential interacting thresholds
3. use shallow knowledge of the system to estimate their Jacobian using short time series.

Social and ecological dynamics are hard to couple. It is not only because there are usually studied in different disciplines with different methods. My guess is that the rates of change of their main variables occur at very different rates. As consequence social scientists assume nature dynamics to be constant or as drivers, while natural scientists assume the “social stuff” to be constant as well.

Modelers have started breaking the ice by introducing noise to the external variables (e.g. rainfall variability, political instability, market pressure); or by looking at how memory or social capital at individual level scale up to resource dynamics. However, their main insights remain confined to study cases making difficult to generalize or study the coupling of society with global change trends.