Zooming into small-scale fishing patterns : The use of vessel monitoring by satellite in fisheries science

Résumé

With the introduction of the European Vessel Monitoring by Satellite system (VMS), scientists could routinely access position data with an accuracy of around 100m. The introduction of VMS boosted the ability to explore the location of fishing activities (Murawski et al., 2005; Stelzenmuller et al., 2008; Fock, 2008), its relation to the habitat (Hiddink et al., 2006; Kaiser et al., 2006) and interaction with other fishing vessels (Poos and Rijnsdorp, 2007a) or users (Bastardie et al., 2015). My research aimed at expanding the use of VMS data to study the impact of fishing at small spatial scales (tens of meters) to be used in an international context taking account of the issues of confidentiality and transparency, allow for analyses to take place at small spatial and temporal scales, and gain a mechanistical understanding of how small scale fishing patterns arise. This to allow for predictions of fleet distribution to be made at small spatial scales.

The PhD-thesis started with the development of transparent, generic and efficient methods to process fisheries data, both VMS and logbook data. Provided that in most countries, but in the EU specific, VMS and logbook reports contain very similar fields of information, standardized data templates were designed at which a suite of analyses tools could be applied to gain understanding in fisheries behaviour and the impact of fishing. An R software package VMStools was developed. This contains a suite of standardized functions to clean VMS and logbook data from evident incorrect entries, link datasets together in time and space and allowed for activity tracking of fishing vessels. The package shows how data from multiple countries can be combined to provide a more complete overview of fishing intensity. Among the routines available in the VMStools package is a tool to interpolate in between successive VMS observations to artificially reduce the interval time of VMS data and hereby reconstruct a trawling track. Although there is uncertainty associated with reconstructing trawling tracks, the methodology performs better than the more simple straight line interpolations applied in earlier studies (Stelzenmuller et al., 2008). Making use of the tools designed in Chapter 2 and 3 we were able to reconstruct trawling tracks at meter scale for the Dutch beam trawl fleet and study the aggregation of fishing in time and space. Zooming in to meter scale was recommended by Ellis et al. ( 2014) to allow counting the number of times a very small area (i.e. grid cell) was trawled by the entire fleet segment over a year. The resulting frequency distribution of counts was approximated by a statistical distribution (Negative Binomial) and the degree of aggregation, one of the main outputs derived from this distribution, showed to be very stable over time at the ICES rectangle spatial scale. This suggest stable patterns in resource distribution and the attraction of fish resources to specific habitats that are routinely trawled by fishers. Since effort varied over the years, this shows that fishers visit in a predictable manner the small spatial scale of a grid cell. In addition to providing a measure for clustering (i.e. aggregation) on favourable fishing grounds, results also indicated the existence of untrawlable habitats and unfavourable grounds in a quantitative manner. This result provides context to areas that are hardly ever visited by fishers to whether these would constitute fishable areas or not, being relevant in the discussion on space available to fisheries in a crowded southern North Sea. First indications that untrawlable areas are associated with coarser habitat types was provided and let to the development of a habitat preference model for the Dutch beam trawl fleet in which the preference for fishers to fish in areas associated with certain environmental conditions was tested. Results of that study show that abiotic habitat characteristics describe well the fishing effort allocation made by beam trawl fishers (both traditional tickler chain fishers as the more recently developed fishery using electric stimuli: pulse fishing). Both types of fisheries prefer elevated landscapes with low gravel contents and often in-between sand ridges rather than on the top of larger scale sand dunes. Pulse fishers fished more in areas with higher gravel content, higher elevation differences and natural disturbance compared to tickler chain fishers which may be attributed to stronger focus on the southern North Sea. The results can be used to predict fishing effort distribution at high spatial scales under a range of scenarios such as closure of areal due to wind farm or Natura 2000 site development. The results also have a direct application in understanding the impact of fishing on ecosystem functioning as e.g. benthic community composition and function is often driven by habitat characteristics.