Fig. 2). We measured human population density as the
total number of inhabitants within the boundary of each
site per coral reef area. We estimated coral reef area at
each site with Google Earth Pro and defined the area as
the total visible coral reef area within site boundaries, as
derived from the expert-elicited site boundary maps (Fig.
1b) (Brewer et al. 2009).We measured market proximity
with ArcGIS (ESRI, Redlands, California) as the shortest
distance from the center of each ecological sample location
to the center of the nearest local fish market, provincial
capital, and national capital (all of which have fish
markets). Rather than Euclidean distance, we measured
market proximity as the shortest distance by roads and
sea from ecological sampling sites to markets (Table 1).
We developed a single measure of market proximity for
each site by summing the unweighted distances from
the ecological sampling location to the nearest local fish
market, provincial capital, and the national capital (Table
1). We measured modernization as the sum of a set
of unweighted infrastructures and amenities within site
boundaries (Table 1). The indicators of modernization
we used are similar to those used in previous studies
(Cinner et al. 2009a; Pollnac et al. 2010). Our method
does not provide a complete treatment of the theory
of modernization, but it does capture key dimensions of
modernization, as understood at the local scale, including
affluence and urbanization. To control for infrastructure
and amenity accessibility, we divided the aggregate score
for each site by the number of villages at each respective
site (Cinner & McClanahan 2006).