Selecting sampling frequency for measuring diving behaviour

The use of time-depth recorders has revolutionised knowledge of diving activity of marine animals to the point where very detailed studies of variation in diving behaviour and performance between individuals, seasons and years are possible. The degree to which the sampling interval selected effects detection of dives and statistics of diving behaviour is investigated, using data from Antarctic fur seals and southern elephant seals representative of the extremes of diving in pinnipeds.
For both species the proportion of surface intervals recognised incorrectly (i.e. real dives artificially concatenated) increased monotonically as sampling interval increased. Effects were especially marked for fur seals: an increase in interval from 5s to 15s resulted in 20% of dives being unrecognised, a 38% increase in mean maximum dive depth, a 29% increase in mean dive duration and a 12% increase in duration of surface interval. In elephant seals an increase in interval from l0s to l00s produced changes of 10%. 5%, 13% and 18% respectively. Choice of sampling interval can therefore create significant biases, especially for species with diving characteristics similar to fur seals; critical comparisons should be confined to data collected using similar sampling intervals.