Employing relative entropy techniques for assessing modifications in animal behavior

Area of Science:

• Animal behavior
• Information theory
• Ecological statistics

Background:

• Quantitative behavioral analysis requires distinguishing new observations from equilibrium behavior.
• Traumatic stress from instrumentation (e.g., telemetry transmitters) can bias baseline animal behavior.
• Information theory offers tools to quantify deviations from equilibrium behavior.

Purpose of the Study:

• To develop a framework using relative entropy for quantifying time-dependent differences between new behavioral observations and an animal's equilibrium behavior.
• To assess the impact of capture and release stress on Pacific bluefin tuna (Thunnus orientalis) behavior.
• To identify behavioral modifications and their duration following handling.

Main Methods:

• Applied information theory concepts, specifically relative entropy, to analyze behavioral data.
• Analyzed speed distributions of Pacific bluefin tuna over a 48-hour period post-capture and release.
• Modeled observed and equilibrium distributions as Gaussian to quantify behavioral shifts.

Main Results:

• Pacific bluefin tuna behavior was significantly altered for approximately 5 hours post-release (95% confidence level).
• Behavioral modification was primarily driven by changes in the central tendency of speed distributions.
• Identified a periodic speed fluctuation linked to daily light changes and diving patterns, occurring before sunrise.

Conclusions:

• Relative entropy provides a robust framework for quantifying behavioral stress responses in animals.
• Capture and release induce a significant, quantifiable stress response in Pacific bluefin tuna, lasting several hours.
• Tuna exhibit light-dependent behavioral patterns, including daily diving, which interact with stress-induced modifications.