Fig. 2
From: Disturbance modifies payoffs in the explore-exploit trade-off
Quantifying exploration in vessel movement tracks. a The Vessel Monitoring System records hourly positions (pings) of commercial fishing vessels in the Gulf of Mexico. In the simulated vessel track shown, markers represent pings that have been classified into one of three activities: fishing (red circles), transiting (●) or in port (◆) using a supervised learning algorithm trained using data from trips when on-board fisheries observers were present. To discretize fishing grounds, a grid is overlaid on the pings and then cells containing fishing activity are identified, allowing a portfolio of fishing locations to be created for each vessel. The location choices of less-exploratory vessels who repeatedly revisit the same locations are easier to predict (lower choice entropy) than those of more-exploratory vessels who regularly sample new locations (higher choice entropy). b Calculated using 2 years of mobility data from before the disturbance, individual choice entropy trajectories approach plateaus for the 106 vessels in our analysis, showing consistent and substantial separation among vessels in exploratory behavior. The first half of each vessel’s trajectory is used as a burn-in period, then the mean of the second half is calculated to quantify the vessel’s explore-exploit trade-off (EETO) strategy. Vessels with higher choice entropy scores have more-exploratory strategies. c There is a weak negative correlation (−0.242; solid line) between choice entropy and patch residence time. Markers show the centred scores for each vessel expressed in standard deviations around zero means. Darker shading indicates higher pre-disturbance performance
