Program

I.7 -- Invited, IWTCP-3

Date: Tuesday, 28 July 2015

Time: 09h05 - 09h40

Defining behavior states of indicator animals under stressful or conflict conditions in a confined area

Tae-Soo Chon (1,2)

(1) Department of Biological Sciences (Prof. Emeritus), Pusan National University, Busan (Pusan) 609-735 Republic of Korea; (2) Ecology and Future Research Association, Busan, 609-802 Republic of Korea

Behavior monitoring would be feasible in on-line, early warning system for ecological risk assessment since stress affectedness could be sensitively and continuously reflected on overall motion of animals by reflecting total physiological network responses to stressors. However, behavior data are complex and difficult to analyze, especially under stressful conditions (e.g., toxic chemical treatment). In this study, recent computational approaches are overviewed in defining behavior states of indicator animals across different time scales in a confined observation arena. Parameters regarding position distribution (e.g., skewness) are used for expressing instantaneous behaviors of animals to detect affectedness by stimuli under conflict conditions. Internal behavior states and transition probability matrix are defined by Hidden Markov Model in a short time sequence (i.e., seconds – minutes) based on external observation data during the course of response to stressors. In an intermediate time scale (i.e., minutes – hours), the whole data sets were transformed including filtering by integration to address test organisms’ response and adaptation to stressors (e.g., intoxicating and recovering processes). Intermittency is additionally used for analyzing overall data structure in long time sequences (i.e., hours – days) to define behavior states in specific areas in the observation arena in individuals and groups.

Presenter: Tae-Soo Chon