Rattlesnake feeding ecology: using high frequency accelerometry to capture feeding events across Crotalus
Hanscom, Ryan J.
rhanscom6545@sdsu.edu
Hill, Jessica
Clark, Rulon W.
Sukumaran, Jeet
Marbach, Tyler
Department of Biology
San Diego State University
San Diego, California USA
DeSantis, Dominic L.
Tipton, Anna F.
Thompson, Morgan
Department of Biological and Environmental Sciences
Georgia College and State University
Milledgeville, Georgia USA
Higham, Tim
Department of Ecology, Evolution, and Organismal Biology
University of California, Riverside
Riverside, California USA
Rattlesnakes are cryptic animals that are rarely encountered and difficult to directly observe because they spend large portions of their lifetime hidden and are sensitive to the nearby presence of humans. Some methodological approaches (e.g., field-based surveillance while a rattlesnake remains in ambush) have been developed to assemble large datasets of different aspects involved in the feeding ecology of rattlesnakes (e.g., encounter rates, strike success rates, prey species encountered, etc.). However these methods are limited and labor intensive. Over the past decade, animal-borne accelerometers have been used by a variety of ecologists to quantify activity and moment-to-moment behavior of free ranging animals. Accelerometry can provide new insight into the cryptic lives of rattlesnakes, and here, we propose a new method to quantify feeding events, and in turn the foraging rates, of rattlesnakes. Accelerometers were externally attached to individual rattlesnakes (3 species; C. horridus, C. oreganus, and C. viridis) and logged acceleration data at rates of 25 Hz. We used direct observations in the lab to validate our classification model to detect the feeding event of a rattlesnake via acceleration patterns. The following behavioral states during a rattlesnake feeding event were classified: strike, strike-induced chemosensory searching, carcass investigation, carcass dragging, ingesting, and swallowing. By validating this method and training supervised machine learning models to automate classification of a feeding event in rattlesnakes by using this lab observation approach, we believe we will be able to accurately determine the foraging rate of free ranging rattlesnakes.