FISHERY BULLETIN: VOL. 80, NO. 3 



categories: 1) those that made more compensa- 

 tory movements than one would expect by 

 chance, and 2) those that did not. 



In Experiment II, each herring was exposed to 

 a random sequence of 23 changes in pressure. If 

 we regard nonresponse as noncompensatory, 

 then the binomial distribution provides a basis 

 for classifying the larvae into two groups, those 

 that moved in the compensatory direction more 

 frequently than one would expect by chance, and 

 those that did not. Under the null hypothesis the 

 probability that a herring would make 16 or 

 more compensatory shifts in vertical position is 

 <0.05. Using this criterion, we classified 5 of the 

 stage-I larvae, 4 of the stage-II larvae, and 10 of 

 the stage-Ill larvae as having made more com- 

 pensatory vertical movements than one would 

 expect by chance. A chi-square test showed that 

 the stage-Ill larvae more frequently compen- 

 sated for the imposed pressure change than the 

 two earlier stages (chi-square = 7.69, df = 2, 

 P<0.03). This implies that the bulla system con- 

 tributes to the larval herring's hydrostatic pres- 

 sure perception only after it contains gas. 



Because the average position of a larval her- 

 ring was determined for each pressure level, we 

 could calculate the average vertical distance it 

 moved for each change in pressure. The average 

 distances moved for the 19 successful fish were 

 regressed against the corresponding change in 

 pressure (Fig. 5). The lines for stage-I and stage- 

 II larvae nearly coincide and their slopes are 

 about half that of the regression for stage-Ill 

 herring. The greatest departure from these 

 lines, fitted through the origin, is for the stage- 

 Ill herring at the -66 cm H 2 OAF. They failed 

 to move downward in the column as much as 

 their performance at other pressure changes 

 would predict. We note that even the stage-Ill 

 herring moved only about 17% of the distance re- 

 quired to compensate fully for an imposed pres- 

 sure change. 



Discussion 



The average proportion of tests in which a lar- 

 val herring moved vertically to compensate was 

 higher in the second experiment than in the first. 

 This was probably due to the shorter duration of 

 the trials and the test series which should have 

 reduced any effects of habituation or fatigue. 

 However, the random nature of the pressure 

 changes in the second experiment may also have 

 contributed to the enhancement of the response. 



E 

 o 



Q 



> 



o 



LU 

 O 



CO 



O 



—I 

 < 



o 

 a. 



LU 



> 



LU 

 O 

 < 



cr 



LU 



> 

 < 



-66 -39 -13 +13 +39 +66 



CHANGE IN PRESSURE (cm H 2 0) 



Figure 5.— Average vertical distances moved by herring lar- 

 vae to compensate for changes in pressure. 



Both of the experimental designs employed in 

 this investigation yielded information. The first 

 design revealed that when two herring are tested 

 simultaneously, the response of each is influ- 

 enced by the presence of the other. It also re- 

 vealed that the response of a herring to a repeated 

 pressure change tends to increase and then de- 

 crease over a 20-min period. The second design 

 provided a more satisfactory test of the null 

 hypothesis that a herring's response to pressure 

 change is independent of the developmental 

 stage of the bulla system and also confirmed an 

 implication from the first experiment: namely, 

 that even before the full development of the bulla 

 system, herring are capable of detecting changes 

 in pressure of the magnitude used in this investi- 

 gation. Finally, the second experiment demon- 

 strated that herring possessing a gas-filled bulla 

 system will exhibit a markedly improved per- 

 formance when compared with less mature lar- 

 vae. 



In very few instances did the larvae move a 

 sufficient vertical distance to fully compensate 

 for the imposed pressure change— a similar find- 

 ing to that of Blaxter and Denton (1976). Even 

 the stage-Ill larvae, on the average, only moved 

 17% of the distance to compensate fully. This is 

 partly a statistical artifact of the manner in 

 which we measured a larva's response. We re- 



572 



