Research Review: Disentangling intangible social–ecological systems, by Orjan Bodin and Maria Tengo
Today for lab meeting the CEPB crew is reading Orjan Bodin and Maria Tengo's excellent new paper "Disentangling Intangible Social-Ecological Systems". These two scholars have been international leaders in developing theoretical and empirical approaches to studying social-ecological systems (SES). In my opinion, the concept of SES is one of the most important ideas for furthering our understanding of environmental governance and policy, and how social decisions link to environmental outcomes.
In this paper, Bodin and Tengo use network analysis methods to model SES as a multiplex network, with connections within and between two types of network nodes: social actors (e.g.; a fishing group) and ecological features (e.g.; different fishing grounds in a coastal area). Based on data from a forest ecosystem in Madagascar, the most important finding of the paper is that in comparison to networks generated by a random process, there is a high frequency of network motifs where ecological connectivity is associated with social connectivity. According to the authors, this "match" between social and ecological connectivity is a fingerprint of more successful governance.
There is no doubt that this paper is a step forward in this area, but it raises a number of points worth debating. First and foremost, the central theoretical argument is that functioanl governance is associated with a "match" between ecological connectivity and social connectivity. For example, if two forest patches are connected by a corridor, then the different communities who use those patches should also be communicating. In the paper, the social links are measured in terms of various kinship ties.
The "match" hypothesis fits within the larger literature on social networks and governance, where people are looking for different network "motifs" that are expected to perform different functions. For example, the Berardo and Scholz "risk hypothesis" expects to find "closed" network structures when the actors are facing collective action problems. I suppose the "match" hypothesis has some face validity, for example it seems likely that shared resources would be hard to manage well if the actors sharing the resources are not communicating.
But I'm not sure the existence of a link between the actors sharing different resources is enough to assume that cooperation and successful resource management is happening. In the specific case studied in the paper, the kinship networks could exist and the groups could easily still be in conflict over any number of issues including resource governance. More generally, any type of social link might be more likely to occur when actors are sharing resources, but without some more specific knowledge about the nature of that link, it is hard to judge whether it would improve resource governance. For example, it could be that if you compare two forest regions in Madagascar, one that is experiencing severe deforestation and another with more successful governance, you would see the same types of network patterns seen from the one case study in the paper. This points to the continued need to link measurable ecological outcomes to network structure, and compare many cases. To be fair, the authors do acknowledge these concerns.
Another interesting point is limiting their analysis of "motifs" to two actors and two ecological nodes. Even with such a limited number of nodes, they come up with a large number of unique combinations. Analytic tractability is the main advantage of focusing on this small family of motifs. But one can imagine many other motifs, for example what if there are three interconnected ecological nodes? However, this criticism is somewhat lame because the authors are just experiencing the same problem being encountered by so much network research: what structural motifs out of the nearly infinite possibilities does the theory tell us we should focus on? Still, it leaves me wondering if there is a way to design a "matching" measure where there is a count of the number of "matching" configurations across many different numbers of social and ecological nodes? This might require some type of judgement about the "degree" of match, rather than just saying match, yes or no.
Finally, they argue that applying their framework requires the theoretical possibility of finding connectivity between any pair of ecological nodes. But this is really unlikely in the context of most ecological systems, where spatial proximity has a huge effect on potential connectivity. For example, there is a distinct upstream-downstream pattern to hydrological connectivity in watersheds. Hence the analysis must somehow be constrained to recognize how space and ecological processes limit potential connections. This might be possible using expotential random graph models that explicity forbid some ecological links, or use attributes of nodes to predict links.
Despite these criticisms, this is a very nice paper. It moves the ball forward, and provides good food for thought about how to analyses SES with network methods. We need more of this to continue improving our knowledge about how SES function.