CHANS and the Risks of Modelling

In their recent review of Coupled Human and Natural Systems (CHANS), Liu et al highlight several facets of the integrated study of these systems;

• Reciprocal Effects and Feedback Loops
  
• Nonlinearity and Thresholds
  
• Surprises
  
• Legacy Effects and Time Lags
  
• Resilience
  

Whilst the emphasis of the paper is on the emergence of complex patterns and processes not evident when human and natural systems are studied independently by social or natural scientists, for me the issue that should be highlighted is the importance of surprises and legacy effects when studying these systems. This goes back to what I have written before about the open, middle-numbered nature of these systems. In these systems history matters and events that occur outside the bounds of the system being studied can have an influence on system dynamics.

With this in mind, when I was recently asked where the risks lie in ecological-economic modelling (modelling that specifically considers the interactions of ecological and economic systems) I suggested we might consider three areas of risk:

1. The production of a integrated model that is not accepted or valued by those we hope it would (whether that be other scientists, decision-makers or members of the society we are modelling). For example, the nature of producing a model that lies somewhere between ecology and economics and/or between science and management has the potential to be accepted by neither party in these dichotomies (as it is not perceived by others to be ‘real ecology’ or ‘real science’ for example). However, this can be avoided by ensuring continued collaboration between economists and ecologists, and between scientists and managers, throughout the modelling process to ensure understanding or model structure.
   


2. The production of a model that is not fully integrated but is rather an ecological model used to examine various economic scenarios. In this case, the study remains integrated (examining the interactions between economic and ecological systems) but the model is not (as feedbacks back from the ecological systems into the economic system, for example in terms of prices and costs, are not fully accounted for). Alternatively, if the modelling process is understood to be iterative, then this initial reduced version of the model may simply be a single step in the complete ecological-economic modeling process.
   


3. Because of legacy effects, surprises etc, a misplaced confidence in what the model can accurately predict may arise. This is also related to the question of the limited capacity to validate models of complex ecological systems given limited empirical data. Again, this may be prevented by continued collaboration between scientist and manager to ensure the structure and limitations of a model are understood, and if a range of model results are predicted for different scenarios (in order to demonstrate the variability in potential outcomes).
   

The study of CHANS will become increasingly important in the future. But if political decisions are to be made based on the outcome of the knowledge gained, the risks present in the study (and specifically the modelling) of these systems must be minimized and accounted for.