COLBY ET AL: PRESSURE SENSITIVITY OF ATLANTIC HERRING 



66 

 39 



13 

 ATM 

 -13 



CM 



I 



E 

 o 



^^ 



LU 

 DO 



CO 

 CO 



UJ 



cc 

 a. 



-39 



-66 



Mnj w 



20 40 60 80 100 120 



TIME (minutes) 



Figure 1.— The upper figure depicts the fixed pattern of pres- 

 sure changes to which pairs of larval herring were exposed in 

 Experiment I. The lower figure shows an example of one of the 

 randomized sequences of pressure changes to which a larval 

 herring was exposed in Experiment II. An independently ran- 

 domized sequence was drawn for each larval herring tested. 



of 5 atm (5,000 cm H2O) before the onset of the 

 regular pressure series to determine if this 

 abrupt AP would impair subsequent pressure 

 sensitivity differently in the different develop- 

 mental stages. These are referred to as treated 

 larvae, whereas those not preexposed are re- 

 ferred to as control larvae. 



Results 



vertical direction to compensate for the change 

 in pressure; -, if it moved in the opposite direc- 

 tion; and nr, if it showed no net change in its ver- 

 tical position. We have used these assigned scores 

 in the analyses to follow. Larvae not responding 

 were treated as if they had moved in the non- 

 compensatory direction, so the analyses are con- 

 servative. The scoring technique allows one to 

 evaluate the frequency with which a fish, con- 

 tending with a dynamic pressure regime, moved 

 correctly, i.e., moved in the appropriate vertical 

 direction to compensate for the imposed pressure 

 change. 



A separate analysis of variance of the number 

 of compensatory responses was calculated for 

 each treatment group, using only the data for 

 members of homogeneous pairs to determine 

 whether the paired larvae tend to respond to- 

 gether. The intraclass correlation coefficients 

 ranged from 0.91 to 0.92 for the two stage-I 

 groups to 0.20 and 0.32 for the two stage-Ill 

 groups, respectively (Table 1 ). Thus for the least, 

 well-developed fish (stage I) the variation among 

 members of a pair was only one-ninth as great as 

 the variation between average values for pairs of 

 fish, i.e., the two members of a pair tended to re- 

 spond together as a unit. The stage-II control 

 group was an exception to this general pattern. 

 For that group the variation among members of 

 a pair was greater than the variation among 

 pairs, suggesting that the members of a pair 

 tended to move away from one another as pres- 

 sure within the cylinder was changed. 



The lack of independence in the responses of 

 fish tested simultaneously invalidates use of the 

 data for mixed pairs (i.e. .pairs of fish of different 

 developmental stages) and requires that we con- 

 sider pairs of larvae as the experimental unit in 

 testing hypotheses about the average perfor- 

 mances. An analysis of variance of data for homo- 



Although pairs of larvae selected for simultan- 

 eous testing were judged visually to be of equal 

 length, the developmental status of their bulla 

 systems was evaluated only after they were sub- 

 jected to the pressure tests and was sometimes 

 found to differ (Table 1). 



A total of 480 locations within the cylinder 

 were recorded for each herring tested. We aver- 

 aged 10 locations of a larva during a 2.5-min test 

 at a given pressure to obtain a more concise sum- 

 mary of the response pattern. We then assigned a 

 score to the larva for each 2.5-min series: +, if its 

 average position indicated it had moved in the 



Table 1.— Numbers of herring larvae, numbers of pairs of lar- 

 vae of the same developmental stage, and intraclass correlation 

 coefficients for fish tested in Experiment I. 



569 



