Smith and Kostlan. Growth of Etelis csrbunculus from sagittal otolith radius 



469 



along the full length of the radial axis of all otoliths 

 sampled. For very small fish, increments near the focus 

 can be counted and integration is not necessary to get 

 an age estimate. The size of the sample would be deter- 

 mined by the variance of otolith radius as a function 

 of fork length, a characteristic which varies regional- 

 ly. Otolith radius, and thus estimated age, was highly 

 variable for a given fork length for all regions for which 

 a sufficiently large sample was available to make this 

 determination. This is evidence that a fairly large 

 sample is necessary in order to have a reliable growth 

 estimate. Sampling a representative number of fish 

 within a full range of sizes and obtaining increment 

 density measurements along the entire otolith radius 

 are important, and should not be done at random. 



Despite the utility of the technique of age estimation 

 developed, potential biases in size range and number 

 of otoliths sampled make it difficult to compare growth 

 estimates for Etelis carbunculus quantitatively. Results 

 presented imply that a more systematic sampling pro- 

 gram may document significant regional differences in 

 growth and population dynamics of this species. The 

 relationship between otolith growth rate and post- 

 rostral radius was similar for all regions. However, 

 there were regional differences in the width of otoliths 

 for a given fork length (Appendix 1), wider otoliths 

 being found in Hawaii and French Polynesia. Although 

 there were few fish sampled from Vanuatu and limited 

 overlap occurs with small organisms from the other 

 regions, fish from NMI and Vanuatu had smaller otolith 

 radius (lower estimated age) than fish of similar size 

 from the other two areas. This means that even if 

 growth rates of otoliths were essentially equal through- 

 out the Pacific, fish would be older at a given fork 

 length in Hawaii and French Polynesia. Thus, well 

 below the forced values of asymptotic length used in 

 regressions, regional differences in estimated age are 

 apparent. 



Fish apparently grow faster at Vanuatu and NMI 

 than in any of the other regions. However, at NMI they 

 reach less than half the size found in the region near 

 Vanuatu. This suggests there are differences in natural 

 mortality rates for these regions (asymptotic length 

 estimates were obtained from virgin stocks for NMI 

 and in the initial stage of the fishery at Vanuatu). The 

 growth curves for NMI yielded higher estimates of the 

 von Bertalanffy growth constant (K) and, as a result, 

 natural mortality estimates were higher. The differ- 

 ences in K hold true even for the NMI regression that 

 was constrained only in t , but as mentioned the 

 results from NMI should be interpreted with caution. 

 Hawaii and French Polynesia showed similar values of 

 K to Vanuatu, but fish from these regions had thicker 



otoliths and were apparently more slow growing. Pos- 

 sible explanations for such differences include genetic 

 differences in regional stocks, variation in the mean 

 annual temperature, and differences in feeding or food 

 availability. 



The estimates of the von Bertalanffy growth curve 

 were in the range of those obtained in previous studies. 

 Although Uchida et al. (1982) estimated K at 0.36 for 

 the Hawaiian Islands, Ralston and Kawamoto (1987) 

 obtained values of 0.15-0.17, which is more consistent 

 with the present results. Both Uchida et al. (1982) and 

 Ralston and Kawamoto (1987) based their estimates on 

 length-frequency data from exploited stocks. Ralston 

 and Williams (1988b) estimated K for virgin stocks at 

 NMI from 0.13 to 0.35, depending on the method used. 

 K estimates from commercial landings data at the onset 

 of the fishery at Vanuatu were in the range 0.07-0.19 

 (Brouard et al. 1983, Brouard and Grandperrin 1985). 

 Thus, the present K estimates seem reasonable. Esti- 

 mates of t were considerably greater than zero, ex- 

 cept in the case of NMI, which was almost certainly 

 a function of the limited number of small fish included 

 in the data. 



The similarity of otolith growth rate at any given 

 radial distance was pointed out by Ralston and Miya- 

 moto (1983), but these authors did not discuss the im- 

 plications with regard to otolith shape for fish with 

 different growth rates. The similarity of growth rate 

 of otoliths and regional variation in growth rate of fish 

 may explain some regional differences in otolith shape. 

 It also indicates that there are species-specific char- 

 acteristics of otolith growth from one region to another, 

 as would be expected. If the concept of narrower or 

 thinner otoliths corresponding to faster-growing organ- 

 isms within a given species can be generally applied 

 to fishes, then this is a tool that can be useful for a 

 number of purposes (for example, in evaluating pale- 

 ontological evidence of growth rates in sedimentary 

 strata). Although environmental and genetic compo- 

 nents of otolith shape must also be considered, differ- 

 ences in growth rate may explain many of the differ- 

 ences in shape which have commonly been used to 

 separate the otoliths of different stocks, or "races," 

 of the same species (Postuma and Zijlstra 1958, Par- 

 rish and Sharman 1959, Kotthaus 1961, Messieh 1972, 

 Bird et al. 1986). Variation in growth may also explain 

 sex-linked differences in the thickness of otoliths of sex- 

 ually dimorphic fishes (Gaemers and Crapon de Cra- 

 pona 1986). The present study is a contribution to the 

 development of methodology that will be useful in 

 evaluating these and other questions regarding the 

 growth rate and shape of otoliths. 



