RALSTON AND MIYAMOTO: OTOLITH INCREMENTS OF HAWAIIAN SNAPPER 



Table 2.— Parameter estimates from the 

 von Bertalanffy model fitted to 64 inte- 

 grated age estimates of Pristipomoides 

 filamentosus. 



which more than 2,500 opakapaka were measured, 

 many fish exceeded 70 cm FL. Consequently, the 64 

 data points were refitted to the model with L« con- 

 strained to a value of 78 cm FL, the largest specimen 

 we have observed (Table 2, dotted line in Figure 

 9). 



Analysis of Size-Frequency 

 Distributions 



The data presented in Figure 10 represent three 

 length-frequency distributions of opakapaka sam- 

 pled at French Frigate Shoals in the Hawaiian 

 Islands. All three samples were taken on different 

 dates but at the same position on the west side of the 

 atoll (lat. 23°47'N, long. 166°22' W). In each distribu- 

 tion a large mode of small fish is plainly visible. If one 

 accepts the premise that these small fish represent a 



single cohort of juveniles, these data afford the op- 

 portunity to estimate their size-specific growth rate 

 by examining the progression of modes in time (Flick- 

 er 1975; McNew and Summerfelt 1978). 



The simplified calculations presented in Table 3 

 show that these fish grew, on the average, at the rates 

 of 0.020 cm/d during the winter growth period 

 (October-May), 0.022 cm/d during the summer 

 growth phase (May-November), and an average 

 0.021 cm/d over the entire year (October 1979- 

 November 1980). If we compute the expected size- 

 specific (31 cm FL) growth rate of these fish, based 

 upon the fit of the integrated otolith age data to the 

 von Bertalanffy model (Table 2, Fig. 9), we predict 

 growth rates of 0.023 and 0.019 cm/d for the freely 

 fitted and the constrained versions of the model, re- 

 spectively. These results compare favorably with the 

 modal growth rates, further substantiating our age 

 estimates. 



TABLE 3.— Growth rate calculations for Pristipomoides filamentosus 

 based on size- frequency distributions. Samples from French Frigate 

 Shoals, Hawaiian Islands. 



October 1979 

 n = 166 



30 40 50 60 70 80 



Fork l«ngth - cm 



FIGURE 10.— Length-frequency distributions of Pristipomoides 

 filamentosus sampled at French Frigate Shoals, Hawaiian 

 Islands (lat 23°47'N, long. 166°22'W). 





DISCUSSION 



Elsewhere we have shown that the increment thick- 

 ness model presented here provides reasonably pre- 

 cise age estimates from a purely statistical stand- 

 point (Ralston and Miyamoto 1981). The frequency 

 distribution of coefficients of variation of T is cen- 

 tered on 6%, with 67 of the 81 values <10%. Thus, a 

 typical preparation provides an age estimate for 

 which the 95% confidence interval is about ±12% of 

 the estimate. Furthermore, the regression technique 

 is an effective means of accounting for variation in 

 otolith growth rates. The frequency distribution of r 2 

 values for the 81 regressions had a median value of 

 78% (Ralston and Miyamoto 1981). 



Several explicit and implicit assumptions, however, 

 underlie the aging method presented here. It is im- 

 portant to address these at this time and to review to 

 what extent they may or may not be justified. One im- 

 portant assumption is that it is appropriate to use the 

 interpolative power of the regression Equation (2) to 

 predict otolith growth rates in intermediate regions 

 where increment microstructure is unclear. This 



531 



