Abstract As the first step in investigating the Pacific bonito (Sarda chiliensis) as a model system for studying the acquisition of swimming behavior and underlying neural activity in teleost fish, we studied the changes in the swimming behavior, anatomy of motoneurons, and muscle activation throughout the larval developmental period. The larval period lasts just 12 days, during which there is rapid development of locomotion. At 12 days post fertilization (dpf), bonito undergo a rapid metamorphosis to the juvenile stage. The short, 12-day developmental period suggests rapid development and organization of the motoneuron and motor system. Swimming behavior was videotaped for five fish in each day, and analyzed by determining a time budget for active time and determining maximum body flexion exhibited by each fish. Motoneurons were fluorescently labeled with Dil and muscle activity was recorded using electromyography. Locomotion is initially produced using the trunk and develops to include use of the full tail and tip of the tail. Early in development, larval bonito spend the majority of their time inactive but sustained swimming periods become more common as the fish develops. Maximum body flexion did not appear to change with development. The number of motoneurons increased from less than 10 to over 30 cell bodies per segment. We also observed two types of motoneurons which differed in morphology. Cell bodies also appeared to develop in tiers until they were of the same size. Motoneuron development first took place in more rostral segments of the fish with the number and complexity of the motoneurons decreasing toward caudal segments. Signals obtained from EMG recordings appeared to correspond to the basic swimming modes of rapid bursts and longer duration swimming using tail beats. We cannot draw a direct correlation between swimming behavior, motoneuron anatomy, and muscle activity based on the data obtained thus far, however, we have shown that development of bonito swimming behavior is associated with rapid development of the motoneuron system and muscle activity which correlates to swimming mode. Introduction The neural circuitry underlying swimming movements has been well studied in cyclostomes and embryonic and larval amphibians, but little is known about the neural basis of the swimming motor pattern generator in teleost fish (Yoshida et al., 1996). The zebrafish (Danio rerio) has been used as a model system for swimming behavior and nervous system development due to its rapid development and transparent larvae, making it easy to visualize motoneurons. Preliminary observations of the Pacific bonito (Sarda chiliensis) suggested that bonito would serve as an ideal model system in which to study the anatomical, mechanical, and behavioral processes underlying the acquisition of swimming behavior in teleosts. Bonito larvae, like zebrafish, are transparent. They hatch within 2 days after fertilization at 22-23°C and undergo rapid growth throughout a larval life lasting only 11 or 12 days post fertilization (dpf) before undergoing a rapid metamorphosis to the juvenile stage. The short larval period suggests rapid development and organization of the spinal cord motoneurons and axial motor systems. The purpose of this study was to begin the investigation of the bonito as a model system for studying the acquisition of swimming behavior and underlying neural activity in teleost fish. In order to produce swimming behavior an optical stimulus, such as a rotating striped drum, is often used (Harden Jones, 1963). This type of optical stimulus has also been used to study the development of vision in zebrafish, where the rotation of a striped drum evokes an optokinetic response (OKR) (Easter and Nicola, 1996). Behavior of a fish can be quantified using a time budget, which accounts for all activity of the fish in a given time period. The most basic time budget divides behavior into two categories; active and inactive. The maximum flexion a fish can perform may also be important because in order to produce undulatory movements, the fish must be flexible in lateral directions (Videler, 1993). In order to investigate the neural circuitry responsible for producing swimming behavior, we studied the motoneuron development during the larval period and electromyography (EMG) recordings. We began our investigation of the changes associated with the development of swimming activity in larval bonito by examining swimming behavior, labeling motoneurons using the fluorescent dye Dil, and using EMG analysis to study muscle activity. We show that development of bonito swimming is associated with rapid development of the motoneuron system and muscle activity which correlates to the swimming mode exhibited. Materials and Methods Animals Larval bonito were obtained courtesy of the Monterey Bay Aquarium. Day 2 through 6 fish used in the behavior and motoneuron dye injections were fertilized on 25 April and raised at 26.5°C +-0.5°. The fish used for days 7 through 13 in the behavior and in the motoneuron dye injections were obtained from a separate hatching. These fish were fertilized on 30 April and raised at 22.5°C +-0.5°. The fish used in the EMG recordings were fertilized on 19 May and raised at 22.5°C +0.5°. On day 7 of development for this population, the population died out and we were unable to obtain fish to finish the time course of development for EMG recordings. The motoneuron dye injections for the day 4 fish was also obtained from this last population, since earlier attempts for motoneuron labeling had been unsuccessful. Error may have occurred in experiments performed on days 2 through 6 fish, which were raised at a higher temperature the other stocks of bonito larvae observed (26°C versus 22°C). The effects of temperature on bonito larval development have not been fully investigated, though preliminary studies indicate that under optimal feeding conditions increased temperature may lead to more rapid behavioral development. Swimming Behavior A rotating drum was constructed using a plastic bucket with a 29 cm diameter. The interior was painted white, and the exterior wrapped in black gaffing tape. Black stripes were placed on the interior of the drum using black electrical tape. Each black stripe was 1.9 cm wide, separated by a white stripe 2.6 cm in width, resulting in a total of 20 stripes 18° apart (Fig. 1). The drum was attached to a rotating base, controlled by a DC motor. The speed and direction of rotation could be varied. The bottom of the drum was painted white, with a hole in the center in which an 9 cm diameter black-bottomed petri-dish containing the fish were placed. The fish were filmed using a Sony CCD camera mounted above the dish. Recordings were made on a Panasonic Time Lapse Video Cassette Recorder (6740). Five fish were placed in the petri-dish for each swimming experiment. For days 2 through 6, each experiment lasted 2 hours. This was broken down into 12 ten-minute time periods. The observation began with a period in which the drum was stationary, followed by counterclockwise and clockwise rotations of the drum. The counterclockwise and clockwise rotation periods were separated by a period in which the drum did not rotate. The drum was rotated at three speeds in both the counterclockwise and clockwise directions: 2 to 3°/8, 4 to 5%/, and 6 to 7°/s. Beginning with the day 7 fish, we reduced the duration of the experiment because we noticed that the fish were becoming tired by the end of the 2 hour period. The revised protocol lasted a total of 25 minutes and was repeated for a total length of 50 minutes. The 25 minute period was broken down into 5 five-minute time periods, beginning with a the stationary drum and alternated with counterclockwise and clockwise rotations at 6 to 7%/s periods, and ending with the a period in which the drum did not rotate. Behavior was observed qualitatively by viewing the tapes. We also looked for optokinetic response by counting the number of counterclockwise and clockwise turns during each time period in the day 7 and day 8 fish. A time budget for the fraction of time spent active for a 2 minute period was determined, as well as the maximum body flexion that each fish exhibited during a 2 minute time period. A time budget of activity was determined for a 2 minute time period while the drum was stationary. A fish was considered to be active if any part of its trunk or tail was moving. A fish undergoing no translation but snapping its jaw was considered inactive while a fish undergoing no translation but flicking its tail was considered active. Each fish was followed for the 2 minute period and the amount of time spent active was determined using a stop watch. Maximum body flexion was measured by observing each fish individually during a 2 minute time period when the drum was stationary. When a turn which included a large body flexion was observed, the behavior was retraced frame by frame on the VCR and traced onto acetate paper. Body flexion angle was measured by determining the angle between lines drawn between the head of the fish and the tip of the tail. Fig. 2 shows examples of tracings of fish and how the angle of flexion was determined. The smaller the angle, the greater the flexion of the fish. Motoneuron Dye Injections The long-chain dialklycarbocyanine lipophilic tracer, Dil, was used at a concentration of 2.5 mg/mL in dimethylforamamide (DMF) to visualize motoneurons. Fish were anesthetized in ice-water before being fixed in 4% formaldehyde for 2 hours. Fixed fish were injected with Dil using microelectrodes approximately 1 um in diameter made from 1.5 mm capillary tubes. Fish were injected 2 to 5 times in the ventral musculature and stored overnight in a 4°C fridge. The days 2 through 5 fish were stored in distilled water until we noticed that the osmolarity of the distilled water differed from that of the fish and was reducing the number of successful injections. The remaining fish were generally stored in phosphate buffer solution (PBS), which was used to dilute the fixative and would not disrupt osmotic balance, and occasionally kept in the 4% formaldehyde. The injections were visualized using an Olympus fluorescence microscope at 4x, 1Ox, 20x, and 40x magnification. Successful stains were filmed using a camera attached to the microscope and a Sony VCR (SLV-373UC). Images of labeled motoneurons were digitized using the Mega Vision 1024XM System v.4.3 and Adobe Photoshop 2.5. EMG Recordings EMG recordings were obtained using a tungsten microelectrode approximately 1 um in diameter. The signal was fed through an amplifier with a gain of 10000 and a 1 kHz filter and then digitized using a Sony DAT Recorder (670). Analysis of the EMGs was performed using Dempster software. The fish were placed in a drop of agarose in a small petri-dish of distilled water and then penetrated with the electrode. They were also viewed through a light microscope for movements which could be correlated with EMG recordings. Results Staging Bonito development was broken down into seven stages. In Stage 1, the Yolk-Sac stage, the fish were characterized by lack of eye pigments, a transparent head, and an internalized oil droplet in a large yolk-sac (Fig. 3A). Stage 1 encompassed day O (day of fertilization) to 2 dpf. In Stage 2, the Eye Pigment stage, the fish developed retinal epithelium pigments by 3 dpf, proto-pectoral fins, and by 4 dpf, a functioning jaw (Fig. 3B). Stage 2 ranged from 3 dpf to 5 dpf. During Stage 3, the Predator stage, the bonito began feeding on artemia, giving them an orange gut by late 5 dpf. Yellow and black spots appeared on the skin, and teeth appeared at 6 dpf. Stage 3 included 5 dpf to 9 dpf (Fig. 30). Morphological changes during Stage 4, the Constant Swimmer stage, included increased definition of trunk structure, spots on the skin covering the head and gut, development of cartilage or bone in the head, greater definition in the nasal area, and clear myotome definition by 11 dpf (Fig. 3D). Stage 4 ranged from 9 dpf to 11 dpf. Stage 5, the Metamorphosis stage, was characterized by a 2-fold increase in total length, increased coloration on the gut and snout, reflective eyes and gut, the appearance of a bony dorsal fin, development of bone in the trunk, and definition of the caudal fin, muscle structure, and nasal area (Fig. 4). Stage 5 consisted of 12 dpf. Stage 6, the Pre-Juvenile stage, was marked by increased trunk coloration and steady growth. Stage 6 consisted of 13 dpf to 14 dpf. At Stage 7, the fish were completely opaque and approximately 2 cm in length. Stage 7 generally began at 15 dpf. Swimming Behavior Stage 1 fish spent the majority of their time inactive, interrupted by short bursts of activity. These short periods of activity typically consisted of short, rapid forward swimming propulsions or quick tight circling, both produced primarily by movements of the trunk. By Stage 2, the fish were swimming more consistently and began to perform graded turns, where the head moved first followed by the rest of the body. Locomotive behavior appeared to involve use of the full tail as well as the trunk. Sustained swimming periods were more common by Stage 3. These periods of steady swimming were often interrupted by short, rapid forward swimming propulsions. Turns were consistently performed by movement of the head first followed by the tail. In Stage 3, a new behavior termed "hovering" was exhibited, where a fish displayed repeated tail flicks but did not translate in any direction. Locomotion appeared to be controlled both by use of the entire trunk and the tip of the tail alone. Stage 4 fish swam nearly constantly, with short periods of inactivity. Hovering was performed using the tip of the tail, and accelerated swimming was achieved using the full tail and trunk. Optokinetic Response Attempts to quantify an optokinetic response in the day 7 and 8 fish were unsuccessful. It did appear that older fish which were stationary oriented towards the rotating drum by pivoting in place, although they did not respond to the drum in a stereotyped manner if they were actively swimming. Activity A trend was seen for increase in activity with development (Fig. 5). Peaks of activity were seen on days 3, 8, and 12, but the overall time spent active increased over the 12-day period. Maximum Body Flexion No significant trend was seen in the change of the maximum body flexion for the bonito throughout the 12 day developmental period studied (Fig. 6). The data was quantified using regression analysis (n-5, r-0.26, F-18.4), which indicated that the difference of the degree of flexion throughout the larval developmental period was not statistically significant. Motoneuron Dye Injections Over the larval developmental period, the number of cell bodies per segment increased from less than 10 to over 30. We noted that motoneurons could be classified into two different types, dependent on morphology. Type A, which was seen more consistently, was characterized by uniform branching of cell bodies off the main ventral root, while Type B motoneurons were characterized by more erratic branching off the main ventral root and greater secondary branching (Fig. 7). We also noted that motoneurons appeared to occur in tiers, with only one tier occurring in the Stage 1 fish (Fig. 8A). Beginning with Stage 3, two tiers of the Type A motoneuron cells were seen, with the second tier of cells being smaller in size (Fig. 8B). In a mid Stage 3 fish, three tiers of cells were observed. The third tier of cells occurred just caudal to the gut and was composed of small Type B cells (Fig 8C, Fig. 9). We were not able to identify a third tier of cells in later stage fish. By the end of Stage 4, the tiers of cell bodies were no longer distinctly layered, nor was there a distinct difference in the size of the cell bodies (Fig. 8D). Motoneuron development first took place in the more rostral segments of the fish, with the number and complexity of the motoneurons decreasing towards the caudal segments. At later stages, the number and complexity of the caudal motoneurons increased (Fig. 10). EMG Recordings Late Stage 1 fish displayed a quick burst of electrical activity. Each spike lasted about 7 ms and occurred at a frequency of 142 Hz. The total duration of this activity reached about 360 ms, preceded by a movement artifact of approximately 170 ms. The amplitude of the spikes within the burst was fairly constant (Fig. 11). Electrical activity generated by early Stage 2 fish consisted of two signals followed by a movement artifact. The first signal was more complicated in nature and consisted of at least three main spikes, all of which may contain overlapping signals. The overall signal lasted approximately 30 ms with a frequency of 33 Hz. Individual spikes within the signal lasted approximately 10 ms with a frequency of 100 Hz. Approximately 50 ms later, this signal was followed by a second spike, which was simpler, consisting of a single spike which lasted about 5 ms with a frequency of 200 Hz and had an amplitude about half that of the first signal. The movement artifact which followed the second signal lasted about 30 ms (Fig. 12). A more complicated swim burst pattern was seen with early Stage 3 fish. Each spike was more complicated in structure, appearing to consist of a primary and secondary spike which overlapped. The individual signals lasted approximately 10 ms, with a frequency of 100 Hz. The entire duration of the pattern lasted just over 100 ms, with the amplitudes of the spikes in a ration of 3:4:5:6. These swim burst patterns were correlated with head movements (Fig. 13, Fig. 14). Swim bursts recorded from middle Stage 3 fish were variable in duration, lasting 300 to 650 ms. The first swim burst pattern was also more complicated, consisting of two consecutive spikes which were repeated approximately 50 ms after the previous spikes ended. The first of the two spikes was larger and relatively unitary in appearance. The second spike was only about half the amplitude of the first spike, but more complicated, appearing to consist of two overlapping signals. Each event lasted approximately 10 ms, with a frequency of 100 Hz. The entire duration of the burst was approximately 650 ms (Fig. 15). The second swim burst recorded was simpler, consisting mostly of unitary spikes, although some spikes were complicated by underlying signals. The spikes lasted about 5 ms, had a high frequency of 200 Hz, and relative amplitudes in a 2:3:5 ratio. The entire duration of the pattern was approximately 300 ms (Fig. 16). Series of action potentials were also recorded, with each individual action potential lasting less than 10 ms. à frequency of 100 Hz. The first two spikes were separated by only about 15 ms, with a longer separation time of about 50 ms between the remaining spikes. The signals had a ration of amplitudes of 1:3, with the middle signal having the largest voltage (Fig. 17). Discussion Swimming Behavior Qualitative observations of swimming behavior indicated that active locomotive behavior increased with time. The gradual change to more complete use of the trunk and tail for swimming seems to indicate a developmental period in which larval bonito gains greater control over their muscle groups to produce more adult-type swimming. The time budget of activity clearly showed a trend of increased activity with age (Fig. 3). This falls into place with our general observation of increased activity with age, and implies that the larval fish are developing swimming behavior over time to reach adult swimming behavior. The peak of activity seen on day 8 may have been due to the hoverins behavior. Hovering was first exhibited on day 8, and the fish seemed to spend more time hovering on the first day of demonstrating the behavior than on consecutive days. The drop in activity seen on day 13 may have been a result of metamorphosis, which generally occurs on day 12. Fish that did not undergo metamorphosis on day 12 may have begun to die. Although there did appear to be a slight trend for increased flexion over de velopment, further data collection would be necessary to quantify the significance of this trend since our data should that the difference in flexion was not significant. Past studies have hypothesized that fish may use bending stiffness of their bodies to modulate their swimming behavior (Long and Nipper, 1996), which implies that as a fish gains greater control over its locomotion its flexibility may change. Our data did not show a significant change in maximum body flexion. However, we did not account for muscle development. which is occurring concurrently. Greater muscle development may allow fish to modulate their flexion. Motoneuron Dye Injections While we were able to identify two different types of motoneurons occurring in the bonito (Fig. 7), we were not able to determine the exact areas innervated by individual motoneurons. It is possible that the Type A and Type B motoneurons correspond to the primary and secondary motoneurons characterized in the zebrafish (Myers et al., 1986). It was shown in zebrafish that the three primary motoneurons precisely innervate cell-specific subsets of contiguous muscle fibers in mutually exclusive regions (Westerfield et al. 1986), which could also occur with the two types of bonito motoneurons. However, to determine if either of these possibilities are plausible, more detailed experiments would be needed to determine if this is the case. If Type A and Type B motoneurons corresponded to the primary and secondary motoneurons identified in zebrafish, they could be activated differently dependent on the swimming mode as was demonstrated in the zebrafish for fast and slow swimming (Liu and Westerfield, 1987) or act synchronously as in the zebrafish escape response (Fetcho and O'Malley, 1995). The number of motoneurons we visualized throughout the larval period was greater than the 3 primary motoneurons identified in the zebrafish (Myers et al., 1986; Westerfield et al., 1986). All of the motoneurons that were seen may not have been primary motoneurons, nor were we able to identify specific motoneurons and track their development, so a direct comparison cannot be drawn to the zebrafish. It does seem, however, that motoneuron development is occurring more rapidly in the bonito than in the zebrafish. The development of cell body tiers (Fig. 8) throughout Stages 1 through 4 seemed to begin with a single tier in Stage 1 which expanded to two tiers seen in early Stage 3. The third tier of cells in mid Stage 3, which consisted of Type B cells, may correspond to innervation of motoneurons into different muscle groups. This could result in the finer locomotive behavior seen in Stage 3, where the tail is used in addition to the trunk and turns are initiated by the head followed by the tail. By late Stage 4, the obvious tiering of cell bodies disappeared, with the cell bodies becoming more evenly layered and equal in size. If the various tiers of cells had innervated different muscle groups, more uniform distribution and size could indicate equal control over all muscle groups producing locomotive behavior. This could correspond to our observations that early in development, locomotive behavior is dependent on movements of the trunk and evolves to consist of use of the trunk and tip of the tail. Again, we could not specifically determine innervation of motoneurons into specific muscle groups. However, development of the more complex tiering of cells through Stages 1 to 4 does correlate with the notion that the nervous system of the larval bonito is undergoing rapid morphological and organizational changes before the onset of metamorphosis, which occurs in Stage 5. Our observation that motoneuron development first takes place in more rostral segments and then in the more caudal segments in later stages (Fig. 10) seems to indicate that larval bonito have less control over their caudal segments earlier in development, consistent with observations that Stage 1 fish use their trunk, or more rostral segments, for locomotion. Later in development, the number and complexity of the caudal motoneurons increased, indicating that the fish would have greater control over their tails for use in locomotive behavior. This is also consistent with the observation that later stage bonito use their tail in locomotive behavior. EMG Recordings The high frequency burst of electrical activity seen in Stage 1 fish (Fig. 11) may correspond to the quick burst of swimming activity observed during behavior. The signals produced by the early Stage 2 fish (Fig. 12) differ from the activity seen in the Stage 1 fish, consisting of less spikes overall which were about equal to or lower in frequency. This signal probably constitutes a different swimming pattern than that in the Stage 1 fish, possibly corresponding to the slower tail beat used in locomotion rather than a quick burst of rapid swimming. The bursts of electrical activity seen in early Stage 3 fish (Fig. 13, Fig. 14) were observed immediately after head jerks. These head jerks would probably have resulted in full body movements if the fish had not been embedded in agarose, indicating that this swimming behavior is similar to the burst of swimming seen in the Stage 1 fish rather than the simpler tail beat. The signals obtained from the early Stage 3 fish were more complicated and lasted longer than the Stage 1 fish, but lower in frequency, indicating a more sustained period swimming with more motoneurons firing. The decrease in frequency could be due to the fact that while the Stage 1 spikes were more unitary and easily distinguished from each other, the signal in question was more complicated and probably consisted of multiple spikes whose individual frequencies could not be determined. Both the bursting pattern (Fig. 15, 16) and the slower tail beat (Fig. 17) were seen in the mid Stage 3 fish. Again, the burst pattern consisted of more complicated signals than the quick burst patterns seen in younger fish, but were comparable in frequency to the signal seen in the early Stage 3 fish. The duration of the pattern varied from approximately 300 ms in one instance to 650 ms in another, possibly indicating that the fish has the ability to regulate the length of the sustained swimming period. Variation in the duration of the pattern is similar to studies performed on the larval angelfish (Pterophyllum scalare), where it was shown that the duration of bursts in older fish were more variable suggesting that the force of the tail beat could be regulated flexibly as the fish developed (Yoshida et al., 1996). The series of action potentials in Fig. 17 most likely correspond to the slower tail beat patter, similar to the one seen in the early Stage 2 fish. The greater number of action potentials indicate more tail beats over a longer duration than was seen in the early Stage 2 fish. Increase in the complexity, length, and number of signal pulses with age has been seen in both larval angelfish (Yoshida et al., 1996) and in Xenopus (Sillar et al, 1993). The electrical signals recorded from the Stage 1 through 3 fish could coincide with the behavior observations of the bonito. Early in development, the fish display mostly short periods of quick forward propulsive activity, which could correspond to the high frequency, short burst of electrical activity. Later in development, the fish are able to maintain longer periods of swimming, along with longer bursts of rapid activity. The slower more sustained tail beats seen in steady swimming could possibly correspond to the singular action potentials and the bursts of rapid forward propulsion could correspond to the more complicated burst patterns. Our EMG recordings cannot be correlated directly with specific behavior patterns or firing of specific motoneurons because typically, a directly correlatable behavior was not observed and insertion techniques were not optimal. Use of the single-insertion electrode makes it impossible to determine the source of the electrical signal. Furthermore, because the fish was much smaller in size relative to the electrode, it was difficult to determine where exactly in the fish the electrode was inserted and stereotyped activity produced by each motor unit which could have been recorded is unknown. Conclusion We cannot draw a direct correlation between swimming behavior, motoneuron anatomy, and muscle activity based on the data obtained thus far. However, it does appear that rapid development of locomotion is accompanied by rapid development of motoneurons, in terms of number, complexity, and organization. EMG recordings seem to correspond to the basic swimming modes of rapid bursts and longer duration swimming using tail beats. Our study only begins to scratch the surface of the full-scale investigation of bonito development. Further studies are needed to continue the development of the bonito model system. Acknowledgments Iwould like to thank my partner in carrying out this project, Linda Nevin, along with Stuart Thompson, Matt McFarlane and Christian Reilly, for their assistance in the project. Greatly appreciated technical assistance was provided by John Lee. Larval bonito were husbanded by David Cripes and Timothy Cooke, who generously provided them to us for use in experimentation. Literature Cited Easter, S.S, Jr., and Nicola, G.N. 1996. The Development of Vision in the Zebrafish (Danio rerio). Dev. Biol. 180: 646-663. Fetcho, J.R., and O'Malley, D.M. 1995. Visualization of Active Neural Circuitry in the Spinal Cord of Intact Zebrafish. J. Nuerophsyiol. 73: 399-406. Harden Jones, F.R. 1963. The Reaction of Fish to Moving Backgrounds. J. Exp. Biol. 40: 437-446. Liu, D.W., and Westerfield. 1988. Function of Identified Motoneurones and Co-ordination of Primary and Secondary Motor Systems During Zebra Fish Swimming. J. Physiol. 403: 73-89. Long, J.H., Jr., and Nipper, K.S. 1996. The Importance of Body Stiffness in Undulatory Propulsion. Amer. Zool. 36: 678-694. Myers, P.Z., Eisen, J.S., and Westerfield, M. 1986. Development and Axonal Outgrowth of Identified Motoneurons in the Zebrafish. J. Neurosci.. 6(8): 2278-2289. Sillar, K.T., and Wedderburn, J.F.S. 1993. Control of Locomotor Movements During Vertebrate Development. Int. Union Physiol. Sci./Am. Physiol. Soc. 8: 107-111. Videler, J.J. 1993. The structure of the swimming apparatus: body axis and fins. pp. 41-70. in Fish Swimming, Chapman and Hall, New York. Westerfield, M., McMurray J.V., and Eisen, J.S. 1986. Identified Motoneurons and Their Innervation of Axial Muscles in the Zebrafish. J. Neurosci. 6(8): 2267-2277. Figure Legend Fig. 1 Rotating striped drum used in the swimming behavior experiments. Fig. 2 (A) Method used for determining maximum body flexion. Each fish represents is a single field at 60 Hz. Translation is not represented, only change in body shape of the fish. One fish from each stage is represented. (B) Representation of the angle of flexion, measured between the head and the tip of the tail. Fig Photos of fish from each stage. Scale bars are 1 mm. (A) Stage 1. (B) Stage 2. (C) Stage 3. (D) Stage 4. Fig. 4 Photos of 12 day old bonito. Scale bars are 1 mm. (A) Before metamorphosis. (B) After metamorphosis. Fig. 5 Scatter plot of fraction of time spent active for each fish in a two-minute period. Boxes are means for each day with standard deviations. Fig. 6 Scatter plot of maximum body flexion observed in a two-minute period for each fish. Greater flexion translates into a smaller angle. Fig. 7 Comparison of Type A and Type B motoneurons for Stages 1 and 4. (A) Stage 1, Type A motoneuron. (B) Stage 4, Type A motoneuron. (C Stage 1, Type B motoneuron. (D) Stage 1, Type B motoneuron. Fig. 8 Comparison of tier development for Stages 1 through 4. (A) Stage 1. Only one distinguishable tier. (B) early Stage 3. Two tiers of motoneuron cells are visible, both Type A. (C mid-Stage 3. Three tiers of motoneuron cells are visible. The third tier consists of small Type B cells. (D) Stage 4. Tiers are no longer distinguishable from each other. 9 40x view of the third tier seen in mid-Stage 3. (A) Focus on the upper two tiers Fig. of cell bodies. (B) Focus on the third tier of cell bodies. Fig. 10 Spinal cord comparison of Stage 2 and 3 fish. Rostral is to the right and ventral is down. (A) Stage 2. The majority of motoneuron development is in the rostral segments, with fewer motoneurons caudally. (B) early Stage 3. Caudal motoneurons have increased in number and complexity. (C) mid Stage 3. The number and complexity of all motoneurons has increased. Fig. 11 Burst pattern from Stage 1. (A) Overview of the entire signal, with baseline shown at the beginning and end. (B) Close-up of the spikes seen in the bursting pattern. Fig. 12 Action potentials from Stage 2. Fig. 13 Burst pattern in an early Stage 3 fish. (A) Overview of the entire signal, with baseline shown at the beginning and end. (B) Close-up of the spikes seen in the bursting pattern. Fig. 14 Burst pattern from early Stage 3 fish. Similar to the one shown in Fig. 13. Fig. 15 Burst pattern from mid Stage 3 fish. (A) Overview of the entire signal, with baseline shown at the beginning and end. (B) Close-up of the spikes seen in the bursting pattern. Fig. 16 Simpler burst pattern from mid Stage 3 fish. (A) Overview of the entire signal, with baseline shown at the beginning and end. (B) Close-up of the spikes seen in the bursting pattern. Fig. 17 Action potential series in a mid Stage 3 fish. Fig. 1 18 1925 e 8.8 cm 28.6 cm 30.3 cm — video 20 I V 2) U V 0 0 a I c 00 I Heh + —. •54 ° • + + •E SAljoe Suijj uojjoelj 8 -00 - 8 8 8 (seaibap) uoixojq 1o jsuy Fig. 7 65 5 Fig. 11 2.50E400 2.00E400 1.5OE-O 1005.00 5060. O.OOEO -5.00E-01 -1.O0E-00 aaaaaaaaa- aaaaa- Laaaaa. Time (ms . Time (ms. aa. (An) 0 v92 822 Fig. 13 .. aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa- Time (ms. aaa- aa . Time (ms a 82 112 112 v22 822 S12 802 202 561 681 281 921 O91 Evl 2 o81- 121. 211 - 111 0 (An) Fig. 15 45. aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa- Time (ms M. aataaaaaaaaaata- aaaai Time (ms Fig. 16 E saataaataaaaata Time (ms. rk 2 eguasguangnbgntg ne S8 EE STRRRRESRARESESS Time (ms aata- (An) 60A v92 12 v82 281 921 691 891 991 091 Evl 181 01 121 A 111