Improving emergence location estimates for Argos pop-up transmitters

Abstract

Background: Recent advances in satellite tagging technologies for marine animals have provided opportunities to investigate the spatial ecology of pelagic species including at-sea behavior and predator-prey interactions. Implantable Life History Transmitters (LHX tags) provide postmortem data on location and causes of mortalities from tagged individuals. Following a mortality event and extrusion of the tag from within an animal, varying amounts of time may elapse between the onset of satellite transmissions, the first successful uplinks to the Argos satellite system and the first location estimate obtained from multiple uplinks. Externally attached pop-up archival transmitters (PAT tags) also only commence transmissions following pre-programmed detachment from a host. The amount of delay for both types of tags may vary with programming, sea state (e.g., wind, waves, currents and tides) tag exposure and satellite coverage. Thus, actual emergence locations for LHX and other pop-up satellite transmitters are hard to accurately determine. Larger errors (~ 10-50 km) may be associated with emergence locations than the errors inherent in the Argos system. Here we present a new approach based on a time-reversed state-space model to improving emergence location estimates and quantifying their uncertainty for pop-up satellite transmitters, using data from 24 LHX tags deployed at known locations in the Gulf of Alaska. Results: Between May and June 2017, we deployed 12 LHX tags in two locations in Resurrection Bay and 12 tags in two locations in Prince William Sound, Alaska. When tracking models included all successful uplinks that resulted in Argos locations immediately after deployments, the emergence location could be predicted to within 3 km, on average. However, increasing transmission delays up to 16 h progressively reduced the accuracy to 6-12 km, and for delays of 24 h or longer, the actual emergence locations were outside of the 95% isopleth of estimates. Conclusions: Emergence locations for pop-up satellite transmitters can be estimated by a time-reversed state-space model. The area confined by 95% isopleth of model output is an effective way to characterize an emergence location with an associated uncertainty. Our findings illustrate the importance of programming tags to enhance satellite uplinks to provide immediate and high-quality locations.