Fig 1.
Schematic diagram of a typical FAD deployed from the field site.
Designs vary around the island. The head consists of series of floats at the surface (~20cm–40cm diameter) connected by rope or other material. The head can be 20m long. Referred to as attraction in Dominica, tarps and nets are attached near the surface in order to attract fish. Attraction is sometimes attached to the head as well to the top section of mainline as shown here. Metal swivels are placed at key points to allow the device to better withstand strong current and to act as weight to keep the rope from floating to the surface. A length of polypropylene rope is attached to an anchor made from half a 55 gallon drum filled with concrete. FADs are commonly placed at depths of 1000m–2500m deep at distances of 15–50km from shore. The schematic is not to scale.
Fig 2.
Four example GPS tracks of foraging Dominican FAD fishers.
The east coast of Dominica is indicated on the left of each plot. Trips originate at Desa Ikan and travel eastward toward the FADs. Track points are colored to indicate speed. (A) Trip to one FAD. (B) Trip to two FADs (C) Trip to a FAD and a near shore shallow water FAD. (D) Trip to two FADs and a wood. Each track is identified by a Trip ID number (TID). Arrows indicate direction of travel. Transparent blue circles indicate areas that were identified as patches by the observer and subsequently identified as ARS. See Fig. 5 for a higher resolution plot of tracks at patches.
Table 1.
Parameters for individual foraging trips by Dominican FAD fishers.
Table 2.
Results of k-means analysis of travel speed.
Fig 3.
Plot of speed versus time for FAD fishing trips.
Plots show the four trips in Fig. 2. Points are color coded in the same way. Periods of travel are fairly homogenous, while behaviors at patches show heterogeneity.
Fig 4.
Frequency distribution of speeds.
Plots are for the trips in Fig. 2. Histograms show the tri-model pattern characteristic of all the trips in the sample. The inset scree plots show an inflection point at three groups indicating that fewer clusters have a sharply increased total within sum of squares.
Fig 5.
Higher resolution plots of the track segments identified as patches in the tracks of Fig. 2.
These segments were identified as patches by the observer and subsequently identified as ARS in the analysis. Tracks are color coded to indicate speed according to the legends in Fig. 2. Within patch movements show speed heterogeneity and patch sinuosity characteristic of ARS.
Fig 6.
Cumulative sum time series plots generated from Equation 1.
Plots are for the trips in Fig. 2. Local maxima and minima are indicated by the red circles, corresponding to points where the track changes between travel and ARS. Segments with positive slope are all travel. Segments with negative slope are ARS and are annotated to indicate patch type identified by observers during trips. FVID indicates a FAD visit identification number.
Table 3.
Patches and behaviors associated with segments.
Fig 7.
Travel segments in the harbor as trips were getting underway or coming to an end.
These segments were misidentified as ARS in the MMA analysis and describe slow movement near the landing site.
Table 4.
FADs visited.
Fig 8.
Mean locations of ARS at FADs.
The FAD patches were moving synchronously across the seascape, presumably with currents. Points are color coded to indicate the week. The single visit to a third FAD is not shown.