McGURK: PACIFIC HERRING OTOLITH RINGS 



1.063. Alcohol-preserved lengths did not require 

 correction. 



Ring Counting 



After extraction from the skull the sagittae were 

 placed on a glass slide under immersion oil; their 

 diameters were measured with an ocular micrometer. 

 Sagittae are slightly flattened spheroids in young lar- 

 vae and tend to become more oval in shape as the fish 

 grows. The diameter measured was always the long- 

 est axis of the otolith. The sagittae were photo- 

 graphed at 400-1, 000X, the developed film was 

 projected on a screen, and the rings were counted. A 

 single ring consisted of a dark band and an adjacent 

 light band. All rings, no matter how faint, were coun- 

 ted in order to avoid observer bias towards a daily 

 ring pattern. Two classes of rings were observed: 1) A 

 group of 1-5 thin, faint rings clustered about the 

 nucleus surrounded by 2) wider, darker rings that 

 composed the majority of the rings in most larvae. In 

 some sagittae the second class of rings were 

 separated from the first by a distinct ring which may 

 have been a check deposited in response to the 

 exhaustion of the yolk. The two classes could not 

 always be clearly distinguished, particularly in slow- 

 growing fish. The first class corresponds to Geffen's 

 (1982) "yolk sac" rings and the second to her "nor- 

 mal" or "regular" rings. In this paper the first class 

 will be unnamed for two reasons: 1) Most of the rings 

 were found in the larvae that had completely 

 absorbed their yolk, so they were not exclusively 

 yolk-sac rings, and 2) it has not been established that 

 the two classes of rings are fundamentally different 

 from each other, so the introduction of new terminol- 

 ogy is premature. Geffen (1982) defined a "first 

 heavy ring" that was found between the outer margin 

 of the nucleus and the first normal ring. This term has 

 not been used because the first normal ring was not 

 always distinguishable from subsequent normal 

 rings on the basis of width or darkness. 



Each sagitta was counted three times, and the mean 

 of the three counts was taken as the final count of that 

 sagitta. The ring count of a fish was the mean of the 

 final counts of its two sagittae. The mean (± 1 SD) dif- 

 ference in final counts between sagittae from the 

 same fish was 1.3 ± 1.4 which was not significantly 

 different from zero (t = 0.9028, df = 148, 0.4 > P > 

 0.2). The sagittae of 21 large larvae (live length range 

 = 14-29 mm, age range = 20-54 d posthatch) se- 

 lected at random from several groups were photo- 

 graphed and then fixed to a glass slide with 

 cyanoacrylate glue and ground to the midplane with 

 metallic lapping paper. They were rephotographed 



and recounted. The mean (± 1 SD) difference was 1.1 

 ± 2.0 which was not significantly different from zero 

 (t = 0.5273, df = 20, 0.5 > P > 0.9). Inspection of the 

 data revealed no trend of the difference with age or 

 with the ring count of the nonground sagittae. 



Data Analysis 



The average rates of ring deposition and of growth 

 in length were calculated as the slopes of linear pre- 

 dictive regressions of mean ring number and mean 

 length on age posthatch. The homogeneity of the 

 variances of the means of a group was tested with 

 Bartlett's test (Sokal and Rohlf 1969), and, if they 

 were found to be heterogenous, each mean was 

 weighted with its sample size divided by its variance. 

 T-tests were used to test the significance of differ- 

 ences between the slope of a regression of mean ring 

 number on age and 1 ring/d and ring/d. F-tests were 

 used in covariance analyses to test for significant dif- 

 ferences between two slopes. 



RESULTS 



Growth in live standard length was positive in all 

 groups except 1980C and 1980B, in which the starv- 

 ing larvae shrank (Fig. 1). There are indications that 

 growth was curvilinear, especially in 1980A and 

 198 IB where the growth rates between the two last 

 sampling dates in each group were much less than the 

 previous growth rates. However, linear growth was 

 assumed for the purpose of obtaining average growth 

 rates to compare with the average ring deposition 

 rates (Table 2). Growth rate was highest in the 2,000 1 

 culture chamber and lowest in the 25 1 aquarium, and 

 there was a positive but nonsignificant correlation 

 between growth rate and container size in the four fed 

 groups (n = 4, r = 0.90, 0.05 > P > 0.10). 



Thin, faint rings of the first class were found in the 

 otoliths of most of the 1980 fish that were < 14 mm 

 long, but were not found in the otoliths of any 1981 

 and 1982 fish (Fig. 2). These rings may have been 

 deposited at any time between the late embryo and 

 the postyolk-sac stage. The only sample of otoliths 



TABLE 2. — Linear regressions of mean standard length on age in ' 

 groups of Pacific herring larvae. 



115 



