Review of analytical approaches for identifying usage and foraging areas at sea for harbour seals

The purpose of this report is to build upon the outputs of work undertaken by the Sea Mammal Research Unit (SMRU, University of St Andrews) funded by the Department of Energy and Climatic Change (DECC) and the Scottish Government, in order to draw these strands of work together with the aim of identifying whether discrete and persistent foraging areas for harbour seals (Phoca vitulina) can be identified in the UK marine area.
A usage map of harbour seal at-sea distribution is presented in order to identify important at-sea areas for the species. Usage is the estimate of the instantaneous density of seals at sea. This is based upon movement data derived from 277 telemetry tags deployed between 2003 and 2013, combined with terrestrial haul-out count data from 1996 to 2013. These data were collected by two tag types: Satellite Relay Data Loggers (SRDL) that use the Argos satellite system for location estimation and data transmission, and GPS phone tags that use the GSM mobile phone network for data transmission. The usage map extends to the limit of UK harbour seal movements and has a spatial resolution of 5km x 5km. Seal usage within the 5km x 5km cells is estimated (using current and historical data where appropriate) for the year 2013. The data do not support breaking down this usage map by year, season, or by the intrinsic properties of seal age or sex. Uncertainty is incorporated into the usage map estimation so that 95% confidence intervals of individual cell usage are available.
Regional significance is demonstrated by dividing the usage map into five Harbour Seal Areas (HSAs). Within each HSA, grid cells are ranked in descending order based on the estimated usage in each cell. Grid cells are selected, beginning with the most intensively used cells, until 10% of the total usage of each HSA is included. This is repeated in 10% increments (up to 90%) of individual HSA usage and the resulting maps are presented.
Usage maps could be made more accurate with strategically-located future deployments of tags. This would also help indicate whether high usage areas are persistent in the long term. The usage maps are weighted by the terrestrial distribution of harbour seals surveyed during the moult (August) period. However, there may be redistribution over the rest of the year so synoptic haul-out counts outside the moult period are needed to test whether this is a significant issue. Usage maps present a ‘snapshot’ of seal usage: they show the estimated number of seals per grid square at any instant in time, in this case the year is 2013.
Estimating activity-unclassified usage is informative for management, since it integrates all activities (all of which are considered to be essential for harbour seals) into one simple index. There may be added value in classifying activity because it would allow activity-specific management. For instance, changes in prey abundance and distribution may be important in areas where foraging dominates, but less important in areas used primarily for travelling. By contrast, in some cases disturbance may have more serious implications in travelling areas; if only one travelling route connects a haul-out and offshore foraging area then disturbance on such routes, for example during wind farm construction, may cause barriers to movement.
A State Space Modelling (SSM) framework was developed that uses track speed and tortuosity, and diving behaviour to disaggregate foraging, travelling and resting activities. Using this framework, it was demonstrated that data collected using SRDL tags are not suitable for estimation of activity states in harbour seals (e.g. foraging); only data from GPS phone tags are of the high temporal resolution required to define activities. There are no usage maps of these activities currently available. However, activity-specific locations of tagged individuals from a DECC-funded study at The Wash were overlaid onto local population usage. This simple comparison indicates that offshore high usage areas, in this region at least, are typically associated with a relatively high density of foraging locations. Such a comparison would not be appropriate elsewhere because other regions for which
GPS data are available are dominated by complex coastlines encompassing multiple haul-out sites. In such regions usage in a given at-sea grid cell results from tracks emanating from multiple haul-out site cells; the usage resulting from each haul-out site cell is then weighted by the population size of that haul-out site cell. Therefore a comparison of the locations of tagged individuals with population level usage would not be appropriate. Such a comparison is possible for The Wash because it comprises a small group of isolated haul-out site cells meaning that both population level usage and foraging locations emanate from the same sites.
Extensive work would be required to use the activity data to quantify key foraging areas. Even though the activity-unclassified usage maps included data from SRDL tags, a lack of telemetry data associated with haul-out site cells meant that 48% of the total harbour seal usage had to be predicted based on the relationship between usage and distance to haul-out sites for those haul-out site cells for which telemetry data were available. This percentage would increase with the exclusion of SRDL data, and predictions based on distance to haul-out sites would be inappropriate for predicting foraging areas. Thus, habitat preference analyses, for which environmental covariates are linked to foraging, is likely to be the most suitable way to in predict key foraging areas.