Background: Biologging and tracking instruments provide valuable, remote surveillance on otherwise unobservable marine animals. Instruments can be consumed (ingested) by predators while collecting data, and if not identified, the retrieved dataset could be assigned to the incorrect individual and/or species. Consumption events of instruments, such as pop-up satellite archival tags and data loggers that record ambient light, are typically identified by negligible light levels and visual assessment of data records. However, when light-level data are not available (e.g., environments below the euphotic zone, instrument model), instrument consumption is not easily discernible. Instruments that record concurrent, time-series temperature and depth data provide detailed information on the ambient temperature in the water column. However, if the instrument is consumed, the temperature profile may dissociate from the depth profile, providing evidence and a means for detecting consumption.
Results:To quantify the dissociation between time-series depth and temperature profiles, we applied the crosscorrelation function to evaluate the time delay and uncoupling between time-series depth and temperature profiles, suggestive of instrument consumption. Given that instruments may be consumed midway through the deployment duration, we extended the cross-correlation function to systematically slide across time-series profiles, sequentially considering subsets of data, to infer time of consumption. This method was applied to datasets from both deep-water (disphotic and aphotic) and epipelagic (euphotic) environments to evaluate instrument consumption. Results were dependent on ambient environment, data sampling rate, predator physiology, and function parameters.Conclusions: Utilization of the cross-correlation function objectively indicates potential instrument consumption events without the bias induced by subjective methods such as visual assessment of tag-recorded data, and does not require the simultaneous collection of light-level data. This methodology aids in the appropriate biological interpretation of tag-recorded data, ensures that data are not attributed to the incorrect species, and can be used to authenticate data during the validation process. Additionally, it is particularly useful for contrasting datasets from comparable studies (i.e., same location and species) and is applicable across taxa and electronic biologging instrument variations.