Fig 1.
Schematic flow-through diagram describing each stage of the method used to calculate Predicted Fishing Effort (PFE).
Method runs clockwise from (1) thru (5).
Fig 2.
Map of the case-study area, the Gulf of California, used to design the model of Predicted Fishing Effort (PFE).
Original locations of raw data used: a) locations at which human population census data were used (n = 153 cells), b) locations in which images of boats were counted (n = 163 cells), c) location of fishing offices (n = 31 cells).
Fig 3.
Relationship between human coastal population and number of small scale fishing boats pre (A) and post (B) kernel density estimation. Points represent the raw data, solid lines the fitted model and dashed lines the 95% confidence intervals (see also S1A and S1B Fig for mapped post KDE data).
Fig 4.
The Predicted Fishing Effort (PFE) presented in 500 km2 grid cells (n = 565).
The unit of PFE is the number of boats adjusted by the number of people in local coastal populations.
Fig 5.
Relationship between Predicted Fishing Effort (# of boats per 500 km2) and mean total annual catch.
In logarithmic form (A) and arithmetic form (B). Points represent raw data (one value per 500 km2 grid cell), solid lines are the fitted non-linear models, the dashed lines are the 95% prediction intervals of the fit, and the outer dotted lines show one standard deviation for the regression residuals. Note the funnel-shaped errors: as PFE increases so does dispersion in the data.
Fig 6.
Map of the upper Gulf of California showing fishing track lines of small-scale fishing vessels (A) and the relationship between PFE per cell and fishing frequency per cell (B). Asterisks denote outliers, solid line represents the fitted model and broken lines represent 95% confidence intervals.