Predicting fishing footprint of trawlers from environmental and fleet data: an application of artificial neural networks

The increasing use of tracking devices, such as the Vessel Monitoring System (VMS) and the Automatic Identification System (AIS), have allowed, in the last decade, detailed spatial and temporal analyses of fishing footprints and of their effects on environments and resources. Nevertheless, tracking devices usually allow monitoring of the largest length classes composing different fleets, whereas fishing vessels below a regulatory threshold (i.e., 15 m in length-over-all) are not mandatorily equipped with these tools. This issue is critical, since 36% of the vessels in the European Union (EU) fleets belong to these "hidden" length classes. In this study, a model [namely, a cascaded multilayer perceptron network (CMPN)] is devised to predict the annual fishing footprints of vessels without tracking devices. This model uses information about fleet structures, environmental characteristics, human activities, and fishing effort patterns of vessels equipped with tracking devices. Furthermore, the model is able to take into account the interactions between different components of the fleets (e.g., fleet segments), which are characterized by different operating ranges and compete for the same marine space. The model shows good predictive performance and allows the extension of spatial analyses of fishing footprints to the relevant, although still unexplored, fleet segments.