Review of available statistical approaches to help identify Marine Protected Areas for cetaceans and basking shark
The Marine (Scotland) Act 2010 makes provision for the designation of Nature Conservation Marine Protected Areas (hereafter MPAs). In response to this Marine Scotland established the Scottish MPA Project to develop the Scottish MPA network. Here we consider relevant habitat modelling methods and available survey data to help inform identification of MPAs for four charismatic megafaunal species: Risso’s dolphin (Grampus griseus), white-beaked dolphin (Lagenorhynchus albirostris), minke whale (Balaenoptera acutorostrata) and basking shark (Cetorhinus maximus). Our aims were to:
- review the appropriate habitat modelling techniques for the identification of marine protected areas,
- evaluate the quality, quantity and relevance of both the available dependent and explanatory data in evaluating Scottish MPAs (at different spatio-temporal scales),
- recommend appropriate modelling techniques for each species given the available data, and
- consider methods for delineating MPAs given the potential results.
- Preparing sightings data and explanatory covariate data for habitat modelling will take considerable time, even building upon efforts stemming from the Joint Cetacean Protocol (JCP) project. The cost in time and effort to organise these data should be considered along with benefits that might be derived from additional data.
- The following currently available dependent data should be considered:
- Risso’s dolphin: available data collated to inform the JCP project from Scottish territorial waters possibly augmented with JCP data from the Isle of Man. If only the west coast is of interest for this species then data should be restricted to this spatial extent.
- White-beaked dolphin: available data collated to inform the JCP project from Scottish territorial waters initially. If the influence of sandeel presence is negligible (i.e. sandeel presence is not chosen as a predictor), then Scottish shelf waters (i.e. to 200 m depth) should be considered. Sandeel data are not available for the entire shelf.
- Minke whale: available data collated to inform the JCP project from Scottish territorial waters but omitting winter data.
- Basking shark: available data provided for the JCP project (where basking shark were recorded) from Scottish territorial waters, augmented with the Speedie data, possibly additionally augmented with data from the Isle of Man but omitting winter data.
- In all cases a small buffer zone may be applied to the area from which input data are collated, to avoid edge effects in the predictions.
- Additional data from Cetacean Research and Rescue Unit (CRRU), Whale and Dolphin Conservation (WDC) and Hebridean Wildlife and Dolphin Trust (HWDT) may prove useful although some work will be required to integrate these data sets into the existing JCP data resource framework.
- GAMs should be used to create predicted relative density surfaces. It is likely that mixed model GAMs or GEE-GAMs will be used to manage the presumed spatial correlation in the data. It is possible for the data-sparse species (i.e. Risso’s dolphin) that a model cannot be fitted, in which case an empirical approach to the identification of regions of relatively higher density could be undertaken.
- Delineation of MPA proposals could be performed by drawing polygons using predicted relative animal densities for individual species. The resulting areas can then be considered by SNH alongside other contextual information (e.g. on behaviour) to inform their advice on areas to be considered for designation as Nature Conservation MPAs.