FISHERY BULLETIN: VOL. 70, NO. 3 



saturated with oxygen and in ones containing 

 water below saturation at 17°C. In the tests 

 below saturation the initial level was 30 9f of 

 saturation, but it increased to 65 ^.r by the end 

 of the observation period 7 hr later. 



At oxygen concentrations below saturation, 

 larvae of both ages swam more frequently than 

 did the controls of the same age (P < 0.008 

 Mann Whitney U Test, Siegel, 1956). Age 

 day larvae spent 5.36 ± 1.41% (±2 X SE, A^ 

 = 15) of the time swimming in water below 

 saturation whereas they spent 1.97 ± 0.64% 

 (N = 10) of the time swimming at saturation. 

 Age 1 day larvae spent 12.48 db 4.50% (A^ = 5) 

 of the time swimming in water below saturation 

 whereas the controls spent 3.30 ± 2.31% (N = 

 5). The proportion of time spent swimming by 

 the controls in both tests did not differ from that 

 given in Table 1 for larvae of the same age. 



This experiment suggests that the regular 

 bursts of swimming of yolk-sac anchovy larvae 

 have a respiratory function. On the other hand, 

 the increase in swimming could have been the re- 

 sult of stress induced by low oxygen concentra- 

 tions and bear no relationship to behavior under 

 normal conditions. I am not inclined to accept 

 this explanation because except for the increase 

 in the duration and frequency of swimming, the 

 behavior of the larvae was normal. A different 

 motor pattern, vigorous shaking of the head, ap- 

 pears at lethal or near lethal levels of oxygen, 

 about 12% of saturation at 17 °C. 



STRUCTURE OF CONTINUOUS AND 

 INTERMITTENT SWIMMING 



To estimate the relationship between larval 

 anchovy tail movement, size, and speed during 

 continuous and intermittent swimming, 53 film 

 sequences were analyzed frame by frame. They 

 included sequences of artificially stimulated and 

 spontaneous bursts of continuous swimming and 

 of bouts of intermittent swimming. In each 

 swimming sequence the mean tail beat ampli- 

 tude, swimming speed, and tail beat frequency 

 were measured by use of a coordinate reader 

 and digitizer (Hunter and Zweifel, 1971). I 

 assumed that the net course swam was equiva- 

 lent to a path formed from the midpoints of the 

 tail beat. If the course was straight, this esti- 



mate was the same as a regression of the X and 

 Y coordinates for the positions occupied by the 

 head or about the same as a straight line fit by 

 eye through frame by frame tracings of the 

 larva. If the course was curved, the path formed 

 by the midpoints provided a reasonable estimate 

 of the net curvilinear path followed by the 

 larva. 



Typically intermittent swimming could be sep- 

 arated from continuous swimming at a glance, 

 but when the tail beat frequency approached that 

 of continuous swimming it was difficult to dis- 

 tinguish between the two types of swimming. 

 Thus, to separate all data into one of the two 

 classes of swimming it was necessary to deter- 

 mine the beat frequency at which larvae changed 

 from intermittent to continuous swimming. This 

 was accomplished by measuring the elapsed time 

 between beats in the slower swimming sequen- 

 ces. 



When the tail was beat at a frequency of 4.7 

 beats/sec or higher, the movement of the tail 

 was continuous, that is, the interval of rest be- 

 tween beats was equal to or less than 0.0078 sec 

 (1 frame at 128 fps). At tail beat frequencies 

 at or below 4.4 beats/sec the movement of the 

 tail was not continuous but rather pauses of 

 0.086 to 0.811 sec existed between beats. The 

 mean of the duration of rest between beats was 

 0.30 ± 0.22 sec (±2 x SE) whereas the dura- 

 tion of a single beat was 0.13 ± 0.05 sec. The 

 duration of the pause or glide between beats was 

 independent of larval size or swimming charac- 

 teristics and was quite variable. The speed of 

 tail movement was also independent of size but 

 was about the same in all larvae. Thus, at beat 

 frequencies below 4 beats/sec, larvae decreased 

 speed by increasing the interval between beats 

 but maintained about the same speed of tail 

 movement. 



Continuous and intermittent swimming data 

 (Table 2) were analyzed separately to determine 

 the relationship between tail beat amplitude, 

 speed, length, and tail beat frequency. The gen- 

 eral equation V/A = a + bF, where A is ampli- 

 tude in cm, F is tail beat frequency, V is speed 

 in cm/sec, and Si is the standard deviation about 

 the line, provided the best fit to continuous and 

 intermittent data sets. The intercepts and slopes 



824 



