Sascha K. Hooker, Susan G. Heaslip, Jason Matthiopoulos, Oliver Cox, Ian L. Boyd. 2008.
Marine Technology Society Journal 42(2): 65-75.
*This paper was originally submitted for inclusion in the MTSJ special issue on "Advances
in Animal-Borne Imaging," Winter 2007/2008, Volume 41, Number 4, guest-edited by Greg
Marshall et al., National Geographic Society Remote Imaging Department.
Although the use of animal-borne video cameras has brought great insight into the lives
of animals, there are still several limitations in using them to record observational data. The
ethical demand for reduced camera size limits battery capacity, which thus restricts recording
duration. In addition to design considerations, there are several sampling options available
for extending sampling duration following tag deployment. These options include the use of
duty cycling (for sampling at preset times), a depth trigger and varying the sampling interval
between images. We investigate the consequences of these options based on foraging data
collected from Antarctic fur seals using the Venus camera system (Wild Insight Ltd). We show
the effect of differing sampling protocols on the resulting data. By sub-sampling our dataset,
we show how changes in sampling protocol affect the number of prey images collected and the
measurement of simple foraging variables such as the time spent feeding. In addition, we ran
simulations based on altering the foraging parameters: prey-encounter rate (waiting time) and
handling time (feeding event duration), to more broadly investigate how changes in sampling
frequency would impact data collection for species with varying foraging characteristics. We
show that, irrespective of sampling frequency, the proportion of prey images recorded remains
constant, suggesting that coverage of a foraging trip could be extended for investigation of
time-invariant processes by using an increased sampling interval. Measurement of rates
or durations of behavioral events (such as prey-encounter rate or handling time), however,
requires higher sampling rates to increase the precision of parameter values.