Wildlife Technology Frontiers, All Rights Reserved 2023
More to come here when we conclude our analysis.
Mini-PATs do have pressure/depth sensors and temperature sensors. From this data, we can describe the vertical movement patterns of the sharks: how much time do they spend at different depths, do they use the entire water column from surface to sea floor, how quickly do they change depths, and how much does this vary by time of day, season or region, or by size and age of the shark?
From the temperature data, we can also describe the ambient temperature preferences of these animals. This may lead to important predictions of how their larger-scale, migratory movement patterns may be influenced for example by long-term increasing ocean temperatures, or short-term changes such as the warm water ‘blob’ anomaly.
For example, this image shows the vertical movement through the water column and associated water temperatures recorded from one sleeper shark, for several distinct periods:
Initially, we begin the analysis of horizontal movement information with only a few high-quality locations, usually, start (where we released a tagged shark) and endpoints (where the mini-PAT pops up at pre-programmed release time):
In between, we may have rare, opportunistic ‘hits’ from the ‘pinger’ being picked up by one of many listening stations in the region, occasional geolocation-by-light position estimates, and of course the two mrPAT locations at 4 and 8 months. However, we also have information about the water temperature-at-depth profiles shown in the previous section on Vertical Movement. We can compare these to known bathymetric temperature profiles. We also know that if a shark went to a maximum daily depth of, for example, 350m, it could not have been in an area that might be only 250m deep. We can use all this additional information to improve our estimates of where the shark may have been.
This approach uses a sophisticated computational model (for the geeks, it is called a Hidden Markov Movement Model). Originally developed for geolocation of Atlantic cod in the North Sea by Pedersen et al. 2008 and Thygesen et al. 2009, this approach was recently refined for spiny dogfish by our team member Julie Nielsen. In even more geek-speak, the Hidden Markov Model is a spatially and temporally discrete state-space model that provides tagged animal location estimates in the form of daily probability surfaces in a gridded study area. The beauty of the HMM approach lies in the use of the maximum amount of available information from different tag types and other data sources.
What can we do with any of the telemetry data we receive from pingers and PATs? We are of course interested in learning more about the movement patterns of the sharks, and their habitat use. The problem we previously explained, is that we only get limited information on the whereabouts of the sharks from the tags we deploy.
Connect with us