Video analysis of fish schooling behavior in finite space using a mathematical model

Abstract

The schooling behavior of the fish Parapristipoma trilineatum (chicken grunt) was examined for a range of school sizes in a water tank containing a central columnar structure using a mathematical model of fish schooling behavior that was modified from that of Sannomiya and Matuda. The two-dimensional motion of individuals was captured on digital video over about 10 minutes, and these time-series data were used to tune the model to the experimental conditions. The model parameters were then used to evaluate schooling behavior quantitatively. Comparing force magnitudes between experimental conditions revealed that propulsive force was less in a tank without the central structure than in a tank with the central structure, under all conditions tested.These results indicate that the school was less active under unobstructed conditions.Propulsive force also decreased with increasing school size. Attractive and repulsive forces of the walls were dominant for smaller school sizes; the attractive and repulsive forces of the structure were significant only for the largest school tested (25 individuals).Thus, changes in behavioral patterns caused by the number of individuals or the presence of structures can be expressed using the mathematical model.

Comparing the EthoVision 2.3 system and a new computerized multitracking prototype system to measure the swimming behavior in fry fish

“…If the fish are marked (see, e.g., Ylieff, 2002;Ylieff et al, 2000), the number of tracked fish is very limited by the EthoVision Color-Pro 2.3 system. In contrast, our multitracking system permits the tracking of a higher number of unmarked fish (up to 100), more than five times the number tracked by Buma and colleagues (Buma et al, 1998;Buma et al, 1996) and Suzuki et al (2003). However, the duration of tracking with our prototype is limited to a few minutes, due to memory capacity-mostly a consequence of the treatment of images.…”

“…These data can permit the determination of some structure parameters, notably cohesion and order parameters, thanks to knowledge of the positions, angles, and speed. These parameters permit direct comparisons with the computer and mathematical models elaborated by physicists (Grégoire & Chaté, 2004;Vicsek et al, 1995;Vicsek et al, 1999) and biologists (Parrish et al, 2002;Suzuki et al, 2003;Viscido et al, 2004).…”

“…The second approach, consisting of the development of mathematical models by computer scientists (Reynolds, 1987), by biologists (Parrish, Viscido, & Grünbaum, 2002;Suzuki, Tsumonu, & Hiraishi, 2003;Viscido, Parrish, & Grünbaum, 2004), or by physicists (Grégoire & Chaté, 2004;Vicsek, Czirók, Ben-Jacob, Cohen, & Shochet, 1995;Vicsek, Czirók, Farkas, & Helbing, 1999) is artificial. It requires a comparison with empirical results to validate and determine the degree of validity for each model.…”

“…In EthoVision Multi-Pro, when the trajectories of two fish cross each other and their images become one, the program is unable to attribute the correct identity to each fish after they uncross (Buma, Moskal, & Liang, 1998;Buma, Moskal, Thomas, & Jongbloed, 1996). Suzuki et al (2003) used another system, but without explicating its limits. Buma's team tracked up to 16 fish, and Suzuki's team tracked 25.…”

Posture tracking of multiple individual fish for behavioral monitoring with visual sensors

“…With the vast improvements in the computational capability of personal computers and the implementation of reliable image processing techniques, the visual monitoring of fish behavior has becomes a common tool for biological and environmental studies, e.g., exploiting the fish behavior for environmental risk assessments and their responses under specific stimulation, such as chemical stress analysis (Barry, 2012;Kane et al, 2004;Ma et al, 2010). A number of video surveillance systems have been developed for automatic observations and measurements of fish behavior since the 1990s (Delcourt et al, 2013;Kato et al, 1996;Kuklina et al, 2013;Papadakis et al, 2012;Suzuki et al, 2003). One of the major advantages of an automatic behavioral monitoring system is that no human interference is required during the observation process.…”