Remote Sensing and Marine Animal Tags: How Much Data is Needed to Detect Environmental Selection in Marine Predators?

Understanding the selection of environmental conditions by marine predators requires knowledge of where they are, but also of where they are not. Tracking marine organisms is challenging because animals can only be detected when they are at the surface. Further, the accuracy of such detections depends on the tag used and on the time that the animal spends at the surface, with uncertainties ranging as large as several tens of kilometers for ARGOS tags. However, absence data are even harder to determine because they are hard to measure in an uncontrolled setting. To address this problem, "pseudo-absence" data are simulated. Pseudo-absences represent the movement of environmentally naive individuals, creating a set of locations that animals could have been, but likely were not if animals perform environmental selection.

Here, we present a method relying on presence data from tagged marine organisms and simulated pseudo-absence data to investigate environmental selection in marine organisms. Using a dataset of simulated tracks biased towards high sea surface temperatures, and a dataset of real marine organisms of unknown sea surface temperature selection, we show how we can test whether animals select for specific environmental conditions. Presence - pseudo-absence data is a valid approach to test for environmental selection in marine organisms, but results must be interpreted with caution - this method can suffer from high false positive rates, especially if the sample size is too limited, the track uncertainty too large, or the selection strength too weak.