In this post Craig Allen and Hannah Birge discuss a paper from Christine Moore, John Grewar and Graeme S. Cumming ‘Quantifying network resilience: comparison before and after a major perturbation shows strengths and limitations of network metrics’
Humans are very good at creating mental models to simplify nature’s complexity, to make its many parts and interactions more understandable. Approaches that exemplify this include hierarchy theory, panarchy theory, and network theory. By identifying commonalities across disparate systems, these theoretical approaches help us understand repeating patterns across systems, enhance learning, and simplify management which would otherwise be idiosyncratic for every location and action taken.
In their paper “Quantifying network resilience: comparison before and after a major perturbation shows strengths and limitations of network metrics” Moore et al. apply network theory to understand the response of a complex social–ecological–economic system to an unexpected perturbation (or ‘surprise’). The network in this case was developed around the production of ostrich for meat and other products, in South Africa, and the surprise was an outbreak of avian influenza. The outbreak killed birds and led to a series of government and farmer responses that led to the immediate reduction in the number of nodes (farms) and connectivity (interactions among farms). The authors’ interest is in the response of the network to this perturbation, i.e. whether the network recovered to its previous state, adapted in some quantifiable way, or reorganized into an entirely different type of network. In other words, the authors are interested in whether the network displayed adaptation and learning following the perturbation
The network responded most obviously through a decrease in number of ostrich farms, or network nodes, following the outbreak of avian influenza. However, the number of nodes recovered in the years following outbreak to near pre-perturbation levels. Similarly, while movement of birds among nodes immediately and sharply reduced, it has since largely recovered. In fact, several years on, the South African ostrich production network has recovered to near pre-perturbation levels and is, surprisingly, even more highly connected than it was pre-perturbation.
Seemingly, and perhaps counter-intuitively, this production network was resilient to the perturbation that occurred. This is not entirely unexpected as systems should be resilient to recurring types of disturbances. Disease outbreaks are not rare in confined animal operations, and have previously occurred in South African ostrich farming. It is possible that the current observed network structure and rapid recovery from avian influenza is the result of past system adaptation to this very type of perturbation, and that it has evolved over time to its current structure partially to withstand recurrent disease outbreaks.
Moore et al.’s use of network theory revealed essential knowledge about this social-ecological system that might otherwise have been confounded by the sheer volume of information present. In an age where large quantities of data are readily available for complex systems, approaches that cut through the noise to reveal important patterns are increasingly valuable. Indeed, although the application of network analyses for complex social-ecological systems is still in its early stages, Moore et al.’s work provides an excellent example of its utility.