Movement Ecology of a Cryptic Ambush Predator: Integrating Radio Telemetry and Tri-Axial Accelerometry to Evaluate Spatial Strategies and Activity Patterns by Western Diamond-Backed Rattlesnakes (Crotalus atrox)

Dominic Louis DeSantis, University of Texas at El Paso

Abstract

An animal’s decision to move from one location to another within its environment is determined by a complex blend of internal and external factors. Teasing apart the relative roles of specific variables in this web of interacting mechanisms has been a long-standing challenge in animal movement ecology. Historically, this problem was viewed as a sort of black box for which a myriad of methodological limitations precluded rigorous study. Recently, a diversity of animal-borne transmitters and dataloggers (i.e., bio-loggers) have circumvented many of these traditional limitations and transformed field studies of animal movement, behavior, and physiology – in some cases, allowing for testing of entirely new theories. Among these technologies, tri-axial accelerometers (ACTs), which enable remote and continuous recording of animal activity, are becoming increasingly commonplace in longitudinal field studies. One of the many strengths offered by ACTs is the ability to be paired with other sensors to provide multiple, complementary data types. In the recent explosion of such integrative bio-logging applications, a distinct taxonomic bias is evident, with smaller-bodied terrestrial taxa often being overlooked because of greater difficulty in device attachment or implantation. Significant computational challenges also remain with these “big data” that are often exacerbated in pilot studies with novel study species. Herein, an integrative framework coupling radio telemetry and accelerometry (RT-ACT) is developed and validated through a case study on Western Diamond-backed Rattlesnakes (Crotalus atrox), representing the first example of snakes as a focal organism in bio-logging research. Telemetry proved to be critical in ACT validation procedures, enabling periodic field observations of rattlesnake behavior that were used to train and test supervised machine learning models for behavioral classification. Following model training, Random Forest and Generalized Linear-NET algorithms distinguished between periods of “activity” and “inactivity” at very high accuracies (99.0% and 97.0%, receptively), allowing automated classification of activity in extensive ACT field datasets (94 ± 99 days, range = 6–289 days). These classifications enabled the construction of continuous activity budgets for evaluation of the timing and duration of activity at multiple temporal scales. In general, activity patterns were found to be highly variable within and between individuals, as the proportion of time spent “active” per individual dataset ranged from 1.6% to 37.1%. The same general daily activity pattern was conserved across all active seasons (spring, summer, fall), with the majority of activity occurring during the evening or nocturnal diel periods. There was seasonal variation in activity duration within diel periods, as activity increased during the summer-mating season, possibly reflective of characteristic mate-searching efforts by male rattlesnakes. Moving forward, long-term and low-frequency ACT field-monitoring could play an important role in improving our understanding of organismal responses to shifting environmental conditions, particularly in small, secretive terrestrial taxa for which other bio-logging technologies are not applicable. Independent captive observations might also allow classification of additional cryptic behaviors (even at very low ACT recording frequencies (1-Hz)) not often observed in the field, ultimately enabling real-time tracking of individual behavior and performance that can be linked to population dynamics. In addition to its role in validating the RT-ACT framework, radio telemetry was also used to explore the effects of sex, behavioral season, and critical resource distribution on the spatial strategies of C. atrox. Specifically, seasonal movement and space use patterns were used to test whether Native Habitats (NH) and human-made Resource Hotspots (RH) on the Indio Mountains Research Station facilitate divergent search strategies in response to critical resources, including potential mating partners, being widely dispersed in NH and clustered in RH. Independent of habitat category, seasonal patterns largely reflected those expected in a male-search based mating system. However, accounting for individuals using primarily NH and those using RH revealed divergent strategies. NH males used more space than RH males within both behavioral seasons, and NH males increased movement distances and space use during the mating season while RH males displayed no significant seasonal shifts. NH females elevated movement distances during the mating season while RH displayed no seasonal shifts in movement or space use. Collectively, seasonal spatial patterns and observations of reproductive behavior uncovered contrasting patterns by NH and RH individuals that might represent alternative optimal strategies in this unique system, highlighting the potential for multiple interacting mechanisms (sexual selection, behavioral plasticity, and habitat heterogeneity) to drive disparate tactics within populations. In summary, this dissertation illustrates: 1) the transformative potential of integrative bio-logging approaches in field studies of movement behavior through the development and validation of the novel RT-ACT framework, and 2) the retained value of traditional data collection techniques (i.e., radio telemetry) in specific contexts.

Subject Area

Ecology|Zoology|Biology

Recommended Citation

DeSantis, Dominic Louis, "Movement Ecology of a Cryptic Ambush Predator: Integrating Radio Telemetry and Tri-Axial Accelerometry to Evaluate Spatial Strategies and Activity Patterns by Western Diamond-Backed Rattlesnakes (Crotalus atrox)" (2019). ETD Collection for University of Texas, El Paso. AAI22616442.
https://scholarworks.utep.edu/dissertations/AAI22616442

Share

COinS