Abstract Developmental changes in the swimming behavior and the underlying muscular function and motoneural anatomy in the Pacific Bonito, Sarda chiliensis, were investigated in the course of the first twelve days of post-hatch development of the fish. Quantitative observations of behavior revealed peaks in activity at days 8 and 12 of development Analysis of maximum body angle while turning showed no trend towards decreasing or increasing flexibility of the fish (f-value = 18.4, r’ =.26), even though Bonito use different amounts of flexion while swimming in different stages of development. Motoneurons in the spinal cord from days 2 through 7 are organized into tiers: one tier was seen on days 2 and 3, two tiers on days 5 and 6, and three tiers on day 7. Tier distinction decreases as the fish grow older, and tiers are no longer distinguishable on day 11. Spinal cord cells are also divided into two categories-cells in pattern A tend to be larger and extend their axons ventrally to a fiber tract which travels parallel to the spinal cord and then veers to form the ventral root, while pattern B cells are smaller and extend their axons directly to their ventral root. Finally, neuritic connections between cells were seen in early days and multiplied with age. In later fish, extensive swellings were seen in the ventral roots and neurites of the spinal cord. Electrophysiology studies revealed that on day 7, the motor units of the fish began to fire in sustained low frequency series' reminiscent of the tail beat in continuous swimming. Also on day 7, a variable pattern of firing was observed for the first time. These data correlate acquisition of swimming behavior with increased complexity of motoneurons and greater sophistication of muscle activity. Understanding Bonito's striking development is a step towards developing a well-characterized model system for the study of the acquisition of swimming in teleosts. Introduction Developmental studies in fishes are focused mainly on swimming-one of the simpler forms of vertebrate locomotion. In swimming, muscle movements are reduced to lateral sets of myotomes which contract alternatively to propel the fish forward. These myotomes overlay the spinal cord and demarcate its segments. This simple organization facillitates ease of both neural control in the fish and muscular and neural studies (Sillar et al, 1993). Correlational studies of the development of swimming behavior, muscle activation, and motoneural anatomy have been performed in several swimming organisms, including cyclostomes and amphibians, but not extensively in teleost fishes. Larval zebrafish (Brachydanio rerio) motoneural anatomy has been studied extensively, and swimming acquisition and underlying neural patterns have been analyzed in larval Angelfish (Pterophyllum scalare), but developmental biology still lacks a well¬ characterized model teleost species for neural control and muscle activation during swimming (Yoshida et al, 1996) This paper introduces another candidate teleost model, the Pacific Bonito (Sarda chiliensis). Bonito offer an auspicious and interesting model for the study of teleosts for three reasons. First, it is possible to detach the intact spinal cord from the fish with minimal effort, allowing for neuroanatomical and physiological studies. Second, these fish are highly active from a very early age and exhibit a range of behaviors including turning, burst swimming, listing, and, later in development, hovering, continuous swimming, and predation. Bonito development is accelerated compared to that of other teleosts. For example, larval Angelfish, Pterophyllum scalare, do not show locomotion until S days after hatching (Yoshida et al, 1996), while Bonito can swim immediately post-hatch. Finally, healthy Bonito undergo a rapid metamorphosis at the end of the larval stage after which they immediately act as predators. We feel that this metamorphosis, in which a Bonito triples in size in approximately 48 hours, must be preceded by substantial progress in motoneural organization and muscle coordination so that the fish are advanced swimmers upon reaching full size. As of yet, there are no theories or data regarding how they might accomplish this preparation. We seek to quantify Bonito's accelerated behavioral development prior to metamorphosis and address how it is controlled by examining motoneural anatomy and analyzing electromyograms of muscle activation during normal swimming. The hypothesis is that all three levels undergo detectable development within the short period chosen- development that can be correlated as a function of culture time. The experimenters gained access to colonies of Bonito bred at the Monterey Bay Aquarium. For each of the first twelve post-hatch days of development (days 2-13), sample Bonito fish were observed and then injected with a fluorescent dye to fill their motoneurons. Electromyograms (EMGs) were recorded from individual fish from days 2 through 7. The observations were quantified into behavior statistics which showed that Bonito acquire swimming behavior quickly. The progression of motoneuron development and trends in muscle activity led us to a general theory of swimming acquistion in larval Bonito. Based on all of the data, we believe that Bonito acquire the capability to execute complex swimming behaviors such as hovering, continuous swimming, and predation as the number and complexity of motoneurons increases and the ability to coordinate motor units into sustained firing in addition to variable firing develops. Materials and Methods Fish Bonito were cultured at the Monterey Bay Aquarium and collected live each day of experimentation. In our numbering system, eggs are collected from a tank of adult bonito and set in the culture tank on day 0. The eggs usually hatch on day 2. With the exception of electromyographic (EMG) recordings, the experiments included 12 days of the larval Bonito's maturation period- days 2 through 13. EMG experiments comprised recordings from days 2 through 7. All fish used in electrophysiology studies were raised at 22°C. In the course of the experiments, we sampled fish from three broods, raised under the same conditions with the exception of temperature. Day 2, 3, 5, and 6 fish used for motoneuronal staining were cultured at 26.5°C +.5°C. All other fish were cultured at 22.5°C+5°C Behavior In order to study swimming behavior, we constructed an apparatus to elicit propulsion and direction changes, modeled after Jones (1963), in which a similar apparatus is used to measure the responses of fish to moving backgrounds. In the general setup, fish swim freely in a container inside a rotating black and white striped cylindrical drum, intended to simulate a water current acting on the fish. Many species of fish have been shown to swim with the rotating stripes, against the artificial current (Jones, 1963). In our apparatus, illustrated in Fig. 1, the larval Bonito swam in a raised petri dish of diameter 8.8 cm with an opaque bottom. The petri dish was placed in a cut opening in the bottom of a plastic bucket which had been modified for the experiment. The 28.6 cm diameter (at base), 30.3 cm high bucket was positioned open end up on a revolving tray which was powered by a motor. The motor was modulated by a 1 kQ potentiometer to allow for variable rotational velocity and by a three way switch to facilitate both directions of rotation. To approximate conditions in Jones (1963), each vertical black and white pair of stripes covered 18 degrees of the drum's inside area; each black stripe was 1.9 in width and each white stripe was 2.6 cm in width. The fish were filmed at 1:1.6 magnification by a Sony CCD camera suspended 30.3 cm above the petri dish, and the footage recorded by a Panasonic Time Lapse VCR 6740. To facilitate the experimenter's visibility of the fish, the base of the petri dish was painted black for 2-12 day old fish which are largely unpigmented, and white for 13 day old fish which are dark in color. Each day, five fish were placed in the apparatus at approximately lOam and the drum cycled through its rotating capabilities—clockwise and counterclockwise at each possible speed. To gain insight into the mechanics of Bonito swimming, we quantified percent time active versus time at rest and maximum angle of body flexion across all 12 days of culture time. First, we inspected the video tapes and were unable to find any quantifiable response to the rotating drum, even at very high speeds- the fish did not turn or swim a different amount under the varying conditions. However, the fish were sufficiently active when the drum was not spinning to use this time segment of each tape for scoring. Activity was quantified with a stopwatch. Each two minute sample, one sample per day, began after the first minute in the rotating drum apparatus to allow for attenuation. Because the purpose of this study is to analyze swimming behavior, a fish was considered active if any part of its trunk or tail was visibly moving. Thus, a fish holding still but snapping its jaw was considered inactive, while a fish hovering in place with its tail beating was scored as active (we chose to include hovering in swimming behavior). Initially, both active time and inactive time were scored and summed to determine accuracy of scoring. As I improved my timing technique, and the sums became consistently accurate, this precaution was abandoned. All five fish were surveyed and the mean percent time active calculated. Maximum body flexion was scored similarly, using two minute segments and data from all five fish. When a large body flexion was observed, the behavior was traced ster by step onto acetate paper from a frame by frame playback on the Panasonic Time Lapse VCR. The flexion of the fish's body was measured by drawing lines through the axis of the head and tail and finding the angle with a protractor-the smaller the angle, the greater the flexion. The smallest angle for each fish was chosen and the mean of these maximum flexions calculated for each day. Anatomy of Motoneurons We labeled spinal motoneurons by injecting ventral muscle with the retrograde neuronal tracer Dil (Molecular Probes*) at 2.5 mg/mL. Dil is a long-chain dialkylcarbocyanine which diffuses laterally in the plasma cell membranes of axons in muscle tissue to fill the cell bodies in the spinal cord. Fish from days 2 through 13 were anesthetized by placing in a beaker on ice and then fixed for two hours in 3.7% formaldehyde. The fixed fish were then injected on a microscope stage using injection micropipettes directed by a Narishige micromanipulator. The micropipettes were pulled from 1.5mm diameter, 4 inch long capillary tubes on a Narishige vertical puller. Dil was drawn into the tips of the microelectrodes by capillary action. A sample injection is shown in Fig 2 (rostral: down, ventral: left), though electrode tips were usually drawn out of the fish without breaking. The injected fish were stored in either distilled water, formaldehyde solution, or Phosphate Buffered Saline (PBS) overnight at 4 degrees Celsius. They were then visualized at 4, 10, 20, and 40X on an Olympus fluorescence microscope and videotaped to a Sony VCR SLV-373UC. Images were captured from the footage using the Megavision 1024XM System v.4.3. and processed on Adobe Photoshop. Motoneuronal anatomy was analyzed qualitatively directly from the footage and from captured, processed images. Electrophysiology Extracellular recordings were read from a single tungsten wire inside a capillary tube pulled on the Narishige vertical puller to make an electrode. The fish was stabilized in 10% Agarose and supplied with water in a dish. The electrode was positioned in the fish's muscle using a Narishige micromanipulator. During readings, careful attempts were made to observe any movement which might be matched with action potentials. The stage was lighted with a D.C. battery, rather than with the AC light source, to avoid creating excess noise in the signal. The experimenter noted any apparent artifacts due to movement near the specimen. The signal was recorded on a Sony DAT 670 and selected time periods processed with PAT—Single Channel Current Analysis Program v6 6d. EMG's were then converted to Microsoft Excel and further processed. Results Behavior Qualitative observations were made of Bonito behavior. All observations were made at 22°C. Starting at day 2, Bonito can swim either upright or on their sides, and alternate easily between the two. Fish of all ages sampled can execute a C-type fast start as described in literature, but do not appear to use the behavior as an escape response or a predatory attack response (Williams et al, 1996). Three swimming strategies are seen over the course of development. In days 2-3, the fish swim by undulating their entire bodies. In days 4-6, they begin to swim using tips of their tails. Finally, in days 7-13, the fish are capable of tail tip locomotion, as before, and full tail locomotion. In Days 7 through 9, the fish spend much of their time either hovering-beating their tails without propelling forward-or swimming continuously around the dish. Fish are seen to attack each other beginning at day 8. As is illustrated in Fig. 3, no trend was found in the maximum body flexion for the time period sampled (f-value = 18.4, r=26). The means of percent time that fish of each age were active are higher in days 5-13 than in days 2-4 (Fig. 4). Figure 5 is an alternative plot of this data, in which all data points outside one standard deviation of the original mean are eliminated. In figure 5, the highest mean is at day 8 and the lowest is at day 2, and there is a drop from day 12 to day 13. There is a decrease in activity from days 8-10 and an increase from days 10 -12. Anatomy of Motoneurons We found two patterns of motoneurons in segments of the spinal cord. Cells in pattern A (Fig. 6), which is more common, tend to have larger cell bodies which extend their axons straight down to a long fiber tract parallel to the spinal cord. This fiber tract veers ventrally to form the root. It appears that these cells extend their axons to the ventral root just caudal to the cell. Cells in pattern B (Fig. 7) are smaller and less densely crowded into a segment than cells in pattern A. Their axons extend directly to the ventral root and tend to intersect each other. In general, segments exhibit exclusively pattern A or patter B, but in some segments of a Day 7 fish, a set of pattern B cells was observed ventral to two sets of pattern A cells. The number of cell bodies per segment is highest in pattern A structures in the last two days, at greater than 25 cells per segment. A general increasing trend in number of cells per segment was apparent. Many cells, even at day 2, have extensive dendritic arbors, which increase in number of dendrites and extent of arborization with age (data not shown). Fish older than day 8 also have many swellings in the ventral roots. In fish from days five through 7, cells in a segment are divided into tiers which are distinguishable by margins which separate them and, in the case of day 7 fish, by the morphology of the cells in each tier. Cells in the more ventral tiers are usually smaller than the cells above them. Two day 2 Bonito were successfully injected and visualized. The stained portions of both fish extend rostrally from the midsection of the fish. Fig. 8 shows the most rostral neurons stained. Only four patter A cells are visible, which is possibly due to poor staining. More neurons are seen in the most caudal segment stained (Fig. 6). This segment contains more than 15 cells, in pattern A. In both segments, the cells extend their axons caudally into the ventral root behind them. A more complete picture of early larval spinal cord anatomy is shown in Fig. 9, a reconstruction of a day 3 Bonito spinal cord. This spinal cord has five sections which represent different morphologies. In the most caudal section, between the tail tip and first arrow, each segment contains approximately three cell bodies which are stacked in pyramid form (20X, Fig. 10). The next section, between arrows 1 and 2, is either quite simple compared to the previous segment or not well stained, but it does shown one repeating well stained dorsal cell body. Sections 3 and 5 show complicated segments of pattern A cells, while section 4 is either simple or unstained. A 20X image of more complex cell morphology, taken from the most rostral segment of the fish's body, is shown in Fig. 11. It appears that the cells shown in Fig. 11 extend their axons to ventral roots both caudal and rostral to themselves, however this is difficult to discern. The ventral roots are more likely close enough together that cell bodies from one appear to extend processes to the rostrally adjacent ventral root. A partial reconstruction of the day 5 spinal cord is compared with the day 3 spinal cord and a partial day 7 spinal cord in Fig. 12. Only the rostral portions of the day 5 and 7 fish were successfully stained. Days 3 and 7 show increasing complexity and number of cells moving from caudal to rostral, but this trend is not clear in the day 5 reconstruction. Day 5 segments appear to have two loosely defined tiers of cells, both of which are arranged in pattern A. The day 7 spinal cord contained the first visible third tier of cell bodies, in the fourth through eleventh segments. Fig. 13 shows one segment of spinal cord in two different planes of focus-one showing the more dorsal, medial upper tiers, in pattern A, and the other showing the ventral, lateral, third tier in pattern B. A third tier has approximately 8 cells; an entire segment has more than 20 cells. Video footage suggested a two tier structure in the most rostral day 7 ventral roots, the more ventral tier containing smaller cell bodies. In day 10, there is extensive connectivity-fine processes are visible and swellings on the fibers in the ventral roots are abundant. The processes projecting from the ventral root in Fig. 14 appear to be innervating muscle. The most ventral of these processes appears to end in a neuromuscular junction. The clearest example of pattern B, shown above, was from a day 10 fish (Fig. 7). In fact, all neurons seen in this day 10 fish were in pattern B. Images from day 11 and day 13 fish show a marked increase in cell number per segment—each segment shown in Fig. 15 has at least 30 cell bodies. These are pattern A cells, and they have lost all organization which would suggest tiers. The best defined image from day 13 shows a more complex morphological structure than any seen in earlier fish (Fig. 16), which under better resolution is interpreted as a pattern B cluster caudal to a pattern A cluster. The pattern A ventral root connects with more than 25 cells; the pattern B ventral root has fewer. Electrophysiology EMGs were recorded from days 2, 3, 5, and 7. A record from a day 3 fish is shown in Fig. 17. The wholescale shift in voltage (232-396 msec) is preceded by 100 msec by a small spike of amplitude 1.8 uV. Record portions such as this were correlated with movement of the fish. The next section of the record is a series of short duration spikes. The high amplitude peaks in this series (greater than 1.1 microV; average 1.75 microV) have a firing frequency of 117 Hz and a constant interpeak interval. These spikes are at such a high frequency that they cannot be correlated with action potentials in muscle, and may in fact be neurons firing. Fig. 18 shows a record from a different day 3 fish, with two consecutive events consisting of two spikes followed by movement, with a 28 msec delay between events. Small voltage changes appear to sum into a small mound immediately prior to both double spikes, which are 667 Hz and 286 Hz respectively. This simple activity demonstrates the ability of a day 3 fish to fire high frequency potentials which are followed by movement without observable delay. A recording from a day 5 fish (Fig. 19) includes a 95 msec action potential series of lower frequency (80 Hz) than the record from day 3 (Fig. 17). This series is more likely but not definitively from muscle. The interpeak interval of spikes in this record is again approximately constant. Series of similar and lower frequencies were consistently found in day 5 and older fish. Two successful records from the final day of recording, day 7, show significantly different electrical patterns, both of which appear to be muscle activity. Fig. 20 shows a long (627 msec) duration swim burst of spike clusters with maximal amplitude of about 4 uV and a relatively low 17 Hz frequency. Fig. 21 shows a non-periodic record with peaks of two recurring amplitudes (about 8 uV and about 3.5 uV). Unlike previous events, this pattern shows a variable interpeak interval, which is interpreted as a more sophisticated pattern. Discussion We believe we have made progress towards our overall goal of correlating the behavior, motoneural anatomy, and electrophysiology of Bonito’s pre-metamorphosis development Our general conclusions are as follows. First, the fish become increasingly active over the period studied, and enhance their repertoire of swimming behaviors. Second, the anatomy of motoneurons increases in complexity. And third, the muscle activity becomes more sophisticated A source of error exists in the husbandry of the fish used in behavioral and neuroanatomical experiments. Fish from days 2 -7 were cultured at 26.5°C while fish from days 8 -13 were taken from a different brood cultured at 22.5°C, the preferred culture temperature. It has been suggested that the growth cycle is accelerated when the temperature is increased (MacFarlane, pers. comm.). Since in our experiments the earlier fish might have had accelerated growth, the results would underestimate the differences between younger and older fish. The greatest complication is in correlating EMG data with other data; fish used in EMG's were raised at 22°C, while most of their counterparts (days 2-6) in other experiments were raised at 26°C. The general increase in activity over culture time is likely to result from all or part of the following explanation. Natural selection may prefer fish that swim more at early ages because the exercise strengthens the pattern generator which controls swimming as well as the muscles which execute it. However, the younger fish may lack the ability to coordinate their muscles into sustained swimming activity. This second statement is supported by the lack of observation of any fish before day 5 swimming continuously, the simplicity of motoneural anatomy until day 5, and day 3 EMGs, which show no record of low frequency sustained muscle group firing. The peak in swimming activity at day 8 corresponds to the hovering behavior, which is seen constantly at this age. Older fish return to floating when they are resting, while day 8 fish hover while resting. Therefore, although fish from days 9 through 12 swim more continuously than day 8 fish, the fish have highest overall activity on day 8. The drop in swimming activity from day 12 to day 13 is difficult to explain. It is possible that the fish used in the day 13 experiment were not healthy. Äfter the experiment, they were examined and found not to have undergone metamorphosis. Possibly fish which have not metamorphosed by day 13 begin to decline. However, they were no other signs of decline in these fish. The flexion experiments lent little insight into the changing swimming technique of these fish. The fish swim with increasing stiffness as they age, but their bodies do not loose any flexibility. One alternative explanation to changing flexibility is acquisition of the capability to use the muscles to modulate body stiffness. It has been shown that the largemouth bass can increase body stiffness by using their muscles to generate negative mechanical work (Long and Nipper, 1996). Perhaps as the Bonito gain sophistication of muscle firing pattern they are able to stiffen as they swim. The first aspect of anatomy of motoneurons, the existence of pattern A and pattern B cells, remains developmentally and functionally enigmatic. One suggestion based on the data is that cells are born in pattern B and grow into pattern A. This contention requires further investigation to be convincing. Though pattern A cells are larger than those of pattern B, there is scant evidence that pattern A cells arise first. When we examined the development of the tier structure across culture time, it appeared that cells are born in groups, which we saw as tiers, and then migrate dorsally and medially to join the other cells. If this is the case, the first tier of cells is born pre-hatch and migrates to the dorsal third of the spinal cord by day 2. Embryonic studies are necessary to examine the pattern of the first tier of cells at their birth. The second tier may then appear at day four and still be apparent at day five, when we observed it first. Further staining and visualization of day four fish is necessary to show the birth of the second tier if it does occur at that point. If this ontogeny is correct, the cells which appear ventrally and laterally on day 7 are nascent cells-the only nascent cells that we saw, which happened to be in pattern B. However, we did see larger cells in pattern B in days 10 and 13, when no tier structure was apparent. Therefore, a developmental trend in appearance of patterns A and B has not yet been conclusively observed. The functions of cells in these patterns may be partially related to external structure. The most striking difference between the two patterns was that in pattern A, cells project their axons to a fiber tract running parallel to the spinal cord which then veers ventrally to form the root. A study of this fiber tract-its structure, and any synaptic connections in it, could lend insight into how the input of cells of type A is modulated on its path to the muscle. Any function of these fiber tracts with regard to electrical activity could create a difference in type A and type B electrical outputs. The succession of tiers in the spinal cord correlates with the acquisition of swimming behavior. We observed one tier of cells on days 2 and 3, two tiers on days 5 and 6, and a third tier on day 7. Beginning day 8, the distinction between tiers of cells disappears and by day 11 is completely nonexistent. It is interesting to note first that activity reaches a maximum at day eight, after all tiers are present. Further, the more complex behaviors which we seek to understand—hovering, continuous swimming, and predation-are each established by day 8. The fish increase their activity level and repertoire of swimming behaviors as the number of tiers increase, and reach full capability when all tiers are present. Spinal motoneurons in vertebrates are multipolar, and the number of dendrites corresponds to the amount of synaptic input the cell receives (Kandel et al, 27). Large dendritic arbors and many synaptic connections within the spinal cord and between ventral roots and muscle are affiliated with finer motor control and more complex behaviors. The enhanced dendritic arborization and complex topology of ventral roots seen as the fish ages is likely responsible for the increase in coordinated behaviors such as hovering. The putative neuromuscular junction seen in the day 10 fish (Fig. 14) may be a direct example of increased motor control. A single electrode inserted into muscle tissue detects signal from asynchronous motor units in the vicinity. The signals tend to interfere with each other, constructively or destructively, so the resulting EMG is often replete with random and complicated signals (Loeb and Gans, 50). For this reason, we examined extracellular EMGs not for precise analysis of swimming mechanism but for comparison across ages of larval fish. The most important information gleaned from EMGs comes from the two day 7 recordings. For the first time in day 7, we saw sustained low frequency firing which almost certainly comes from motor units. The 17 Hz frequency of the series in Fig. 20 corresponds to the frequency of tail beat of a day 7 fish swimming continuously around the dish. The clusters of spikes in this EMG suggest that motor units have been able to synchronize. Further, we see in Fig. 21 that a day 7 fish can create a variable pattern of firing. This sophisticated pattern suggests complex behavior, such as coordinating fins to execute hovering. It is interesting to note that the fish acquire sustained low frequency firing, sophisticated pattern firing, a third tier of motoneurons, and the ability to hover on day 7. Clearly, the acquisition of swimming behavior is correlated with changes in motoneural anatomy and electrophysiology of the fish. That these correlations were made using accessible techniques in a relatively short period of time attests to the utility of the Bonito in studies of teleosts. Acknowledgements Stuart Thompson Matthew MacFarlane Christian Reilly project conducted in conjunction with Elizabeth Hsu Literature Cited Jones, F. R. Harden. (1963). The reaction of fish to moving backgrounds. Journal of Experimental Biology 40: 437-446. Kandell, Eric R., James H. Schwartz, and Thomas M. Jessell. Essentials of Neural Science and Behavior. Appleton & Lange. Stamford, Connecticut. Q 1995. Loeb, Gerald E. and Carl Gans. Electromyography for Experimentalists. The University of Chicago Press. Chicago, Illinois. Q 1986. Long, John H. Jr. And Karen S. Nipper. (1996). The importance of body stiffness in undulatory propulsion. American Zoology 36: 678-694. Liu, Dennis W. and Monte Westerfield. (1988). Function of identified motoneurones and co-ordination of primary and secondary motor systems during Zebrafish swimming. Journal of Physiology 403: 73-89. Sillar, Keith T., John F.S. Wedderburn, Anne-Marie Woolston, and A. John Simmers. (1993). Control of locomotor movements during vertebrate development. News in Physiological Sciences 8: 107-111. Weis, Judith Shulman. (1968). Analysis of the development of the nervous system of the zebrafish, Brachydanio rerio. Journal of EmbryologicalExperiments in Morphology 19(2): 109-119. Williams, P.J., J.A. Brown, V. Gotceitas, and P. Pepin. (1996). Developmental changes in escape response performance of five species of marine larval fish. Canadian Journal of Fisheries and Aquatic Science 52: 1246-1253. Yoshida, Masayuki, Kentaro Matsuura, and Kazumas Uematsu. (1996). Developmental Changes in the Swimming Behavior and Underlying Motoneuron Activity in the Larval Angelfish, Pterophyllum scalare. Zoological Science 13: 229-234. Figure Legends Fig. 1 Behavior experimental apparatus. Fig. 2 Injection site in ventral muscle. Rostral is down; ventral is left. Fig. 3 Change in fish flexibility over culture time. Smaller angle of flexion equals greater flexibility. Fig. 4 Change in activity levels of fish over culture time. All data shown, means are outlined squares, error bars are standard deviations. Fig. 5 Change in activity levels of fish over culture time. All data shown, means and standard deviations calculated with all data points greater than one standard deviation outside the mean removed. Fig. 6 Day 2 segment in pattern A. Rostral is right, ventral is down. Fig. 7 Day 10 segment in pattern B. Rostral is right, ventral is down. Fig. 8 Day 2 rostral segment in pattern A. Fig. 9 Day 3 spinal cord reconstruction displayed to scale with day 3 fish. Rostral is down, ventral is left. Fig. 10 Day 3 caudal motoneurons. Rostral is right, ventral is down. Fig. 11 Day 3 rostral motoneurons. Rostral is right, ventral is down. Fig. 12 Spinal cord reconstructions. Rostral is down, ventral is left. Top: day 3. Center: day 5. Bottom: day 7 Fig. 13 Two planes of focus of one day 7 spinal cord segment. Rostral is right, ventral is down. Fig. 14 Ventral roots of day 10 fish. Rostral is right, ventral is down. Fig. 15 Day 11 spinal cord segments. Rostral is right, ventral is down. Top: 20X. Bottom: 40X Fig. 16 Day 13 spinal cord segment. Rostral is right, ventral is down. Fig. 17 EMG from day 3 fish. Fig. 18 EMG from day 3 fish. Fig. 19 EMG from day 5 fish. g. 20 EN om day 7 fish. Fig. 21 EMG from day 7 fish. Fig. 1 18 1.92.6 1:1.6 8.8 cm 28.6 cm — video 30.3 cm Fig. 2 8 (soaibap) uoikj jo juy 1.0 0.8 0.6 5 0.4 0.2 0.0- 7 1 2 3 4 5 6 7 8 9 10 11 12 13 1 age (days) Fig 4 1.2 1.0 0.8 0.4- 6 0.2 0.0 1 6 * 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 age (days) 9. 5 E Fig. 6 Fig. 7 Fig. 9 Fig. 10 Fig. 11 ta ta ta- Fig. 15 Fig. 16 D (sijoou) eo 8811 9911 ze 680 9901 8201 066 196 176 168 898 978 261 691 921 869 099 189 169 199 829 967 29t 671 968 898 0e8 167 192 182 861 991 281 (Sou) o 797 912 882 182 122 112 012 802 961 681 281 911 891 v91 Ivi Ovl e1 8 921 611 211 901 0. (Sijoou) o vO9 06v 91v 291 8vt vet 02v 90v 268 818 v98 O98 988 28 808 162 082 992 297 882 v2 012 961 281 891 v91 Ovl 921 211 O u) o . 9 166 696 9Ov6 216 9888 998 9928 861 9691 1v1 9711 189 9999 129 9 869 019 g1v9 C19 9v8t 991 912v 668 9018 2ve 9818 987 9992 827 97661 111 9Zv1 vI1 998 19 982 u 809 86t 611 v9t OSV 981 12v 901 268 L8 898 818 618 908 062 912 192 1v7 812 202 681 v11 091 Sv1 181 911 zo1 67