Movement modeling and patterns of within- and among-individual behavioral variation across time scales in neonate garter snakes (Thamnophis elegans)

Abstract

Animal movement modeling and animal personality frameworks are both useful approaches to identify patterns of behavioral variation. Yet these approaches are seldom applied jointly, especially in the context of a controlled laboratory experiment. In this study, we combine these approaches to identify patterns in the behavior of neonate snakes at different time scales (within several minutes of a trial and across the days of the experiment). Specifically, we utilize Bayesian hierarchical hidden Markov models on fine-scale movement data and Bayesian mixed linear models on additional behavioral data (average movement speed, tongue-flick rate, latency to emerge, and area explored) measured repeatedly in captive-born western terrestrial garter snakes (Thamnophis elegans). This novel approach identified distinct behavioral states that describe movement patterns and predictable patterns of transitions in behavioral states at different time scales. Snakes exhibited consistent among-individual variation for all traits, and behaviors were correlated across contexts. Patterns of behavioral changes over time demonstrate responsiveness consistent with habituation, a basic form of learning, but snakes did not learn to associate a novel object with a simulated attack. Our unique approach provides a useful case study in how the joint application of statistical methodologies often relegated to different fields can be leveraged to describe personality axes and to characterize shifts in behavior at short (minutes) and long (across days) time scales in an experimental design context.

Significance statement

Using a combination of traditional statistical approaches and novel movement modeling, we identified patterns in the behavior of baby snakes at different time scales (within several minutes of a trial and across the days of the experiment). Snakes were consistent in their behaviors and exhibited individual differences in these behaviors. Further, behaviors were correlated across contexts: snakes that were more active in an open field test were also more likely to emerge from a hide and explore. Over time, snakes decreased their responsiveness to repeated stimuli, which suggests a basic form of learning. However, we found no evidence that snakes could learn to associate a novel object with a simulated attack within the timeframe of our experiment.