462 



Fishery Bulletin 89(3). 1991 



Commonwealth of the Northern Marianas Islands 

 (NMI). In the single case where the estimated von Ber- 

 talanffy growth curve did reach an asymptote (379 mm 

 for NMI), it did so considerably below the maximum 

 length commonly found in that region (580 mm) and 

 well below the maximum length included in the data. 

 The present study reevaluated the method of age and 

 growth estimation using the same data, in an attempt 

 to see if growth estimates could be improved. In 

 response to some of the problems encountered in 

 reprocessing the data, the methodology of integrating 

 and fitting was modified in a manner that may prove 

 useful to other researchers. In addition, some impor- 

 tant guidelines for collecting data for growth estima- 

 tion from integrated microincrement densities were 

 developed. 



Methods 



Ralston and Williams' otolith growth data (unpubl.) 

 from Hawaii, Vanuatu, French Polynesia, and NMI 

 were reevaluated, using a new method that employs 

 both analytical and numerical integration of the rate 

 curve. Fork lengths were recorded and sagittal otoliths 

 removed from fish caught with bottom handlines at 

 depths of 80-200 m in the four regions. The data were 

 collected and compiled by Ralston and Williams from 

 1982 to 1988. They collected a wide size range of fish, 

 but made no systematic sample of a representative 

 number of organisms of each size. Small fish, not ordi- 

 narily captured with the large hooks used by commer- 

 cial fishers, were underrepresented in all regional 

 samples. Collections were essentially random, within 

 the size range selected by this fishing gear. Since the 

 principal objective was to find an appropriate method of 

 fitting and integration, this limitation was acceptable. 

 Otoliths were prepared and microincrement densities 

 on the postrostral axis recorded from video-relayed 

 images using techniques described by Ralston and 

 Williams (1988a). The data available for each fish in- 

 cluded fork length, total otolith postrostral radius, and 

 a series of estimates of microincrement density at 

 various points along the otolith radius. Microincrement 

 densities were registered in microns per increment, 

 calculated by dividing the length of the radial segment 

 by the number of increments counted within it. Follow- 

 ing the assumption of Ralston and Williams (1988a) and 

 Brouard et al. (1983) that microincrements are de- 

 posited on a daily basis for Etelis carbunculus, incre- 

 ment densities represent otolith growth rate (microns/ 

 day) as a function of distance from the origin of growth, 

 or focus. Validation of daily increment deposition for 

 most deep slope snappers has not been obtained be- 

 cause of low survival when fish are brought to the sur- 



face. However, validation was obtained for Pristipo- 

 moides filamentosus (Ralston 1981), another eteline 

 snapper found in similar habitat and locations. 



Ralston's method of age integration consists of esti- 

 mating the time transpired within consecutive 500-^m 

 intervals of otolith radius, by multiplying the length 

 of each interval by its mean growth rate (microns/day). 

 The estimated number of days required to grow 

 through each 500-^m radial interval are then summed 

 numerically, producing a series of estimated ages for 

 fish with otoliths measuring 500, 1000, 1500^m, etc. 

 The relationship between the natural logarithms of fork 

 length and otolith radius, determined by regression 

 from each regional sample, is then used to estimate fish 

 length at the midpoint of each 500-fim interval of otolith 

 radius. The resultant age-at-length data are fit to a von 

 Bertalanffy growth curve by nonlinear regression. In 

 this way, estimated fork lengths for a hypothetical 

 group of fish with otoliths measuring 250-9250 nm (by 

 intervals of 500) are fit to age estimates for otoliths 

 at 500-9500 ^m (by 500-^m intervals). Actual fork 

 lengths and otolith measurements are used only to 

 estimate the relationship between fork length and 

 otolith radius. In addition to the slight difference in 

 otolith radius (250 ^m) between age and length esti- 

 mates, the validity of the assumption that the length 

 of fish as a function of otolith radius can be accurately 

 predicted from a regression curve obtained from the 

 original sample depends on having had a representative 

 size range and number of individuals from each popula- 

 tion. Rather than incorporate this sampling error, it 

 was decided to concentrate on using the otolith growth- 

 rate data to obtain integrated age estimates for the fish 

 and otoliths sampled. 



Early attempts at integrating the otolith growth 

 function were aimed at reducing the size of the radial 

 interval used to approximate the growth rate. This was 

 found to be impractical for various reasons. The aver- 

 age number of microincrement density estimates 

 ("readings") per fish ranged from 4.1 to 61.5 regional- 

 ly. Otoliths from Hawaii and French Polynesia were 

 measured more extensively; 1681 and 3877 readings 

 were made from 37 and 63 otoliths, respectively. For 

 NMI and Vanuatu, fewer measurements were made 

 (252 and 141 from 62 and 13 otoliths, respectively). 

 Because of these differences in the number and distri- 

 bution of microincrement density readings along the 

 otolith radius, the reliability of mean otolith growth 

 rates and resultant age estimates varied widely as a 

 function of the length of the radial interval chosen. The 

 asymmetry of the otolith growth-rate function com- 

 pounded this problem. Therefore, an alternative 

 method of integration that did not rely on mean 

 microincrement densities as a function of radial inter- 

 vals was chosen. 



