Researchers have demonstrated that analyzing accelerometer data from GPS collars can significantly refine the estimated timing of predation events in white-tailed deer fawns. The study, published in Animal Biotelemetry, addresses a common limitation in wildlife survival studies where collar mortality alerts are frequently delayed because predators move carcasses after the initial kill. By applying piecewise regression to activity data, the team identified abrupt shifts in movement that likely correspond to the actual moment of predation, providing a more accurate timestamp than traditional collar alerts alone.
The research was conducted by Maya Silver and Tyler R. Obermoller, who aimed to improve the accuracy of identifying mortality timing in neonate ungulates. Standard GPS collars equipped with activity sensors are widely used to detect mortality events in large mammals. However, when a predator such as a coyote kills a collared animal, it often drags or moves the carcass before the collar’s sensor triggers a mortality alert. This physical displacement creates a temporal gap between the actual predation event and the recorded death time, which can distort survival estimates in ecological studies.
To test a method for correcting this delay, the researchers utilized data from 171 white-tailed deer fawns that died by coyote predation. Each fawn was fitted with a GPS collar that included an accelerometer, providing fine-scale activity data in addition to location information. The study focused specifically on neonate ungulates, a population where precise timing of mortality is critical for understanding survival dynamics during biologically relevant periods. The use of accelerometer data allowed the team to examine the specific movement patterns surrounding the death of each animal.
The analytical approach involved using piecewise regression to detect activity patterns that might indicate a flee response followed by a predation event. This statistical technique identifies changepoints in the data, which are specific moments where the trend in activity shifts abruptly. The researchers used these model-selected changepoints to estimate the true time of mortality, comparing these estimates against the standard collar-detected mortality alerts. The method was designed to isolate the short burst of activity likely associated with the predator’s attack from the subsequent period of inactivity or carcass movement.
The results showed that the piecewise regression models were successful at identifying these abrupt shifts in activity. The analysis selected a three-segment model for 17.5% of the individuals, which corresponded to 30 fawns. For the remaining 82.5% of the individuals, which was 141 fawns, the two-segment model was selected. These models effectively distinguished the pre-predation activity from the post-predation state, allowing for a more precise estimation of when the kill actually occurred.
On average, the models detected a changepoint in the activity data 13.5 hours earlier than the collar-detected mortality alert. The standard deviation for this difference was 10.1 hours, indicating some variability in the delay. The range of differences observed across the 171 individuals extended from 0.0 to 44.5 hours. This finding highlights that relying solely on collar alerts can lead to a substantial overestimation of the time of death. In many cases, the actual predation event occurred well before the collar registered the mortality, with some delays reaching nearly two days.
The study concludes that many neonatal ungulate survival studies may be overestimating the time of death when determined from GPS-collar mortality alerts alone. By applying simple piecewise regression to activity data surrounding a mortality event, researchers can identify abrupt changes in activity that likely correspond with predation events. This approach provides refined estimates of mortality timing, which is crucial for accurate survival analysis. The method helps account for the temporal error associated with collar alerts, thereby improving the precision of survival estimates during critical life stages.
Furthermore, the refined timing can provide additional evidence when assigning cause-specific mortality.
Subject of Research: Refining predation timing in white-tailed deer fawns using accelerometer data from GPS collars
Article Title: Using accelerometer data to refine timing of predation events of white-tailed deer fawns (Odocoileus virginianus)
Article References: Silver, M., & Obermoller, T. R. (2026). Using accelerometer data to refine timing of predation events of white-tailed deer fawns (Odocoileus virginianus). Animal Biotelemetry. https://doi.org/10.1186/s40317-026-00508-7
Image Credits: AI Generated
DOI: 10.1186/s40317-026-00508-7
Keywords: white-tailed deer, accelerometer data, GPS collars, predation timing, wildlife telemetry, survival analysis, coyote predation, piecewise regression, animal biotelemetry, neonate ungulates, accelerometer, data
News Source: Drew Townsend. (November 10, 2026). Accelerometer Data Refines Timing of Deer Fawn Predation. Scienmag.
