¹Laboratory of Marine Ecosystem Change Analysis

Field Science Center for the Northern Biosphere, Hokkaido University

Hakodate, JAPAN

kawabe@fish.hokudai.ac.jp

　

²National Institute of Polar Research, Tokyo, JAPAN

³Hokkaido University, Hakodate, JAPAN

　

　

The tail beat and swimming behavior of four captive Japanese flounder (Paralichthys olivaceus) were monitored while swimming in an aquarium using acceleration data-loggers. Depth, swimming speeds and two-axis acceleration data were collected continuously for ca. 20 hours per fish. Simultaneously, the swimming behaviors of the fish were filmed at different angles. Using the specific characteristic of the acceleration profiles, in tandem with other types of data (e.g., speed and depth), four behavioral patterns could be recognized; (1) 'active' swimming, and (2) burying patterns, (3) 'inactive' gliding, and (4) lying on the bottom. Tail beat frequency ranged from 1.65±0.47 to 2.04±0.25 Hz (mean±S.D., N=4). Using the relationship between tail beat frequency and swimming speed, their 'preferred' swimming speed was estimated to be between 0.6 and 1.2 Body length (BL)/sec. Additionally, fish rarely swam faster than 1.2 BL/sec. This study shows that the acceleration data-logger is a useful and reliable system for accurate recording of the tail beat of free-ranging fish and estimating flatfish behavior.

　

　

In order to understand fish ecology, further more the underwater behavior of free-ranging fish must be monitored precisely. However, since the difficulty of direct assessments of swimming speed and the proportion of time spent swimming by free-ranging fish (Hinch and Collins, 1991) a reliable methodology that would provide accurate information on the behavior and energetics of free-ranging fish have been required. Behavioral studies using water speed-sensing transmitters (Block et al., 1992) have allowed researchers to investigate the actual underwater behavior of free-ranging fish. Unfortunately, speed sensors are known to overestimate swimming speed. Moreover, water-current speed and direction may influence the energy expenditures of gliding fish (Carey and Scharold, 1990; Holland et al., 1990). Therefore fish swimming energy expenditures cannot be estimated using speed transmitters. Facilitated by recent advances in micro device technology, data-loggers that record body movements through two accelerometer signals were developed and deployed on a variety of free-ranging aquatic animals. This enabled researchers to monitor various activities, such as the porpoising behavior and postures of penguins (Yoda et al., 1999, 2001) or the fin-beating activity of free-ranging migrating chum salmon, in tandem with records of swimming speed, diving angle and diving depth (Tanaka et al., 2001).

　

So far, few behavioral data are available for demersal fish such as flatfish. Flatfish do not have an gas bladder and are therefore negatively buoyant in their medium. This forces them to lie on one side on the seabed, which may be advantageous for camouflage and cryptic behavior (Norman, 1969). To measure the energetic costs of an active behavior, it is important to observe all of the activities of free-ranging flatfish for a long time and establish behavioral time budgets. However, time allocation and/or precise energy budgets cannot be estimated using only visual observation of captive fish. Only the data-logger is considered to be possible to measure the energy budgets of free-ranging flatfish.

　

In this regard, in seawater aquarium, acceleration data-loggers were attached to freely swimming Japanese flounders, (Paralichthys olivaceus - a key species in stock enhancement and marine ranching activities in Japan) and simultaneous measurement of their tail beat activity and swimming speed was conducted. The behavior of instrumented fish was also filmed by video cameras of multiple angles.

　

　

Four Japanese flounders (body length 52.0-54.0 cm, body weight 2.7-3.1 kg; Fish 1, 2, 3 and 4) were captured by commercial fishermen off the coast of Kashima, Ibaraki prefecture, Japan and transported to the Fish Behavior Laboratory of the National Research Institute of Fisheries Engineering (NRIFE). Fishes were kept in a circular tank of 2.5 m in diameter and 0.9 m deep (5,000 L) with a constant flow-through of seawater. The water temperature in the tank was maintained at ambient ocean temperatures (approximately 14℃) and natural light cycles were also simulated.

　

The behaviors of the flatfish were monitored with a 12-bit resolution, 16 MB memory, four channel ('acceleration') data-loggers with two acceleration sensors (UWE-200PD2G, Little Leonardo, Tokyo). These cylindrical-shaped, 20 mm diameter 120 mm long loggers weigh 64.0 g in air and 22.0 g in seawater. The devices include a depth recorder, speed meter and two piezoresistive accelerometers (Model 3031, IC Sensors). These accelerometers record the 'surging acceleration' in the direction of the main axis of the flatfish (forward and backward) and the 'heaving acceleration' along the axis crossing the fish's body from the eyed (upward facing) side to the blind (downward facing) side (Fig.1).

　

　

The measuring ranges of both accelerometers were between -39.2 and 39.2 m/sec² (-4 and 4G, parallel and orthogonal to the main axis of the data-logger, respectively). The amplitude of acceleration was sampled at 16 Hz and filtered using an analogue sensor signal in the band pass filter between 0.5 and 8.0 Hz. Depth and swimming speed were sampled every second, with the resolution of 5 cm and 5 cm/sec, respectively. Swimming speed was measured by counting the number of revolutions per second (RPS) of an anterior-mounted propeller. The stall speed of the speed sensor was determined experimentally to be 25 cm/sec. Speeds below these values were considered to be indistinguishable from zero. A regression line was used to relate RPS to swimming speed. To calibrate the speed sensor, we examined the relationship between RPS and flow velocity (cm/sec) in NRIFE. The relationship was linear from 25.0 to 120.0 cm/sec and the regression coefficient was greater than 0.98.

　

Fish were transferred from the holding tank into individuals, where they were anesthetized by briefly submerging into well-oxygenated seawater containing 0.125 g/l of MS-222 (Ethyl m-Aminobenzoate Methanesulfonate; NACALAI TESQUE, INC., USA). The acceleration data-logger was attached to each fish with two nylon straps, inserted through the dorsal musculature and aliened along the body axis. The fish were then allowed to recover for at least 12 hours before being released into an experimental rectangular seawater aquarium (13.0 m x 7.5 m, 2.5-2.3 m deep). The water temperature in the aquarium was maintained between 15.5 and 18.5℃. The fish were also observed from three different angles using video cameras. One video camera was hand-held and two others were placed on the bottom of aquarium. Once the experiments ended, the fish were removed from the experimental aquarium and the acceleration data-loggers were retrieved.

　

After retrieval of the acceleration data-logger, data were downloaded into a laptop computer and analyzed with Igor Pro 3.1.4 software (WaveMetrics, USA). Statistical analysis was conducted significance using Stat View 4.5 software (SAS institute, USA) and mean ± standard deviation (S.D) were provided. Swimming movements off the bottom to depths of at least twice the depth resolution (±0.05 m) of the acceleration data-loggers were considered to be vertical and horizontal movements and were analyzed for swimming duration (sec) and swimming speed (cm/sec and BL/sec). 'BL' stands for 'Body length', BL/sec being a common unit of measure for swimming speed in fish. The acceleration profiles were compared frame by frame with a visual analysis of the videotapes (30 frames/sec), and synchronized using a dubbed time interval (0.033sec) of the data-logger. Dynamic behaviors (i.e. swimming and burying) were categorized by usually examining the acceleration profiles. The periodic properties of the acceleration signals obtained from the dynamic behavior of the flatfish allowed us to apply an auto-correlation and fast Fourier Transform (FFT) analysis to determine the frequency of swimming and burying.