Instantaneous implementation of systematic conservation plans at regional scales is rare. More typically, planned actions are applied incrementally over periods of years or decades. This is certainly true in the central Philippines, where a network of 33 MPAs was gradually established between 1974 and 2008. 

In recent years there has been a rapidly growing literature on MPA network design, with increasingly sophisticated analytical approaches to considering connectivity between protected areas. Yet this literature generally fails to consider the important reality that protected area systems are usually implemented gradually, over time. In addition to identifying which locations should be prioritised for protection, this introduces a new factor to consider: how should we schedule protected area designation? Which sites should be established first? This might be especially important, given that whilst implementation is ongoing, threats to biodiversity will continue in unprotected sites and the character of the connected ecological system will change. 

A group of colleagues and I, led by Stuart Kininmonth, set out to understand the best strategy for scheduling management within a connected environment, using the central Philippines MPAs as a case study. This was a fun paper to be a part of because it brought together network modelling approaches, empirical data on larval connectivity in the Philippines, and conservation planning theory. 

We modelled the sequential application of no‐take MPAs within a metapopulation framework, using a range of network‐based decision rules including PageRank, maximum degree, closeness centrality, betweenness centrality, minimum degree, random, and a dynamic strategy called colonisation–extinction rate. We compared each strategy, and the real implementation schedule, in terms of its ability to ensure the persistence of a reef fish metapopulation over time. 

The PageRank schedule performed best, whilst the actual implementation from historical records ranked lowest. Our results indicate that protecting the sites ranked highest with regard to larval supply is likely to yield the highest benefit for fish abundance and fish metapopulation persistence.

Our analyses threw up a couple of surprises. First, the dynamic index we tested based on the evolving measure of colonisation–extinction rate ranked only third, after PageRank and betweenness centrality. The higher success of the PageRank and betweenness measures, over stochastic metapopulation dynamics, implies that each site can be ranked simply on the local neighbourhood structure in the first year of the schedule, which is good news for conservation planners. 

Second, the random allocation of MPAs also outperformed the actual implementation schedule. In general, a random selection process will distribute MPAs evenly throughout the network, whereas the actual implementation schedule was spatially biased towards certain areas, missing protection within the densely connected region of the network.  This might occur for a number of reasons, including opportunism, though representation-based conservation planning might have a similar effect. This highlights the importance of including network processes in conservation planning.

You can read the full paper here:

Kininmonth, S., Rebecca, W., Abesamis, R. A., Bernardo, L. P. C., Beger, M., Treml, E. A., et al. (2018). Strategies in scheduling marine protected area establishment in a network system. Ecological Applications, 1–9.

Co-author Lawrence Bernardo discussing early marvel connectivity modelling results with stakeholders in the Philippines to inform the placement of new MPAs. 

Leave a Reply