Stevens et al.: Radiometric validation of age, growth, and longevity of Sebastes melanostomus 



713 



otolith growth (i.e., the core, representing the first few 

 years of life). Radium-226 is present at such low activity 

 levels, however, that many otolith cores from fish of a 

 similar age and same sex must be pooled to acquire the 

 mass of material needed for detection (-0.5 to 1 gram; 

 Andrews et al., 1999a, 1999b). Because we possessed 

 a limited number of blackgill rockfish otoliths (-1200), 

 an age prediction model was created to conserve otolith 

 material for radiometric analysis. It was appropriate to 

 assume from the results of Francis (2003) that within- 

 sample heterogeneity with respect to otolith age and 

 mass growth rate was negligible in the core material. 



To determine age groups for radiometric analyses, 

 final ages for fish whose otoliths were sectioned, along 

 with their corresponding average otolith weight (left 

 and right, «=2), were used to predict age for the re- 

 maining fish in the collection. Several parameters were 

 regressed to determine a predictive relationship be- 

 tween average otolith weight (henceforth termed "oto- 

 lith weight") and estimated age (i.e., section age). The 

 following regressions were compared to estimated age 

 by using Kruskal-Wallis (nonnormal) ANOVA: 1) oto- 

 lith weight (to the nearest 0.001 g), 2) otolith weight 

 and fish length (to the nearest 1 mm), and 3) otolith 

 weight plus otolith length (to the nearest 0.001 mm) 

 multiplied by otolith weight (as an interaction term). A 

 power function was also investigated but did not result 

 in a better fit than that provided by a simple linear 

 regression (either log-transformed or normal). A paired 

 sample (-test and student's (-test for slopes were used 

 to determine if a significant difference existed between 

 male and female otolith weight, and between male and 

 female otolith weight-to-age regressions, respectfully. 

 The final regression equations were applied to the aver- 

 age otolith weight for all individual remaining fish to 

 obtain a predicted age. Age groups were created if there 

 was sufficient otolith material from fish of the same sex 

 and of a similar predicted age. 



The predicted age range for each group was kept as 

 narrow as possible while permitting enough material for 

 analysis; approximately 25 to 50 otoliths were needed at 

 a target core weight of 0.02 g. Fish that had both oto- 

 liths intact (not sectioned or broken) were preferred to 

 reduce the number of fish for each radiometric sample. 

 To better insure sample conformity, 90% confidence 

 intervals with respect to fish length and otolith weight 

 were used to eliminate from each group dissimilar fish 

 that may have varied significantly from predicted age. 

 In addition to this discriminating technique, groups 

 were further confined by capture year and location. 

 Only samples caught in the same year and similar geo- 

 graphic location (based on the majority of port locations 

 within 300 miles) were included in the same group. 



Core size was determined by viewing several whole 

 juvenile blackgill rockfish otoliths with estimated ages 

 between 1 and 7 years. The first annulus was deter- 

 mined to be approximately 2 mm wide, and a 3-year-old 

 otolith was measured at 3 mm wide, 4 mm long, and 1 

 mm thick, and having a weight of 0.02 g. These dimen- 

 sions were chosen as the target core size because a core 



of this size could be easily extracted, yet was young 

 enough to minimize the possible error associated with 

 variable 226 Ra uptake in the first few years of growth. 

 Otoliths from adult fish were ground down to the tar- 

 get core size with a lapping wheel and 80- to 120-grit 

 silicon-carbide paper. Otoliths from selected juveniles, 

 if older than age 3 (core size), were also ground to the 

 target core size. 



Radiometric analysis 



The radiometric analysis was conducted as described in 

 Andrews et al. (1999a, 1999b). Because previous studies 

 have revealed extremely low levels of 210 Pb and 226 Ra in 

 otolith samples, trace metal precautions were employed 

 throughout sample cleaning and processing (Bennett et 

 al., 1982; Campana et al., 1990; Andrews et al. ,1999a). 

 Acids were double distilled (GFS Chemicals", Powell, 

 OH) and all dilutions were made using Millipore- filtered 

 Milli-Q water (18 MQ/cm). Samples were thoroughly 

 cleaned, dried, and weighed to the nearest 0.0001 g 

 prior to dissolution. Whole juvenile otoliths groups were 

 analyzed first to determine if exogenous 210 Pb was a 

 significant factor, and to determine baseline levels of 

 226 Ra activity. 



Because of the low-level detection problems associ- 

 ated with (beta) /3-decay of 2ln Pb, the activity 210 Pb 

 was quantified through the autodeposition and (alpha) 

 a-spectrometric determination of its daughter proxy, 

 polonium-210 ( 210 Po, half-life=138 days; Flynn, 1968). In 

 preparation for 210 Po analysis, samples were dissolved 

 in acid and spiked with a calibrated yield tracer, 208 Po, 

 estimated to be 5 times the activity of 210 Po in the oto- 

 lith sample. Polonium isotopes from the sample were 

 autodeposited onto a purified silver planchet (A.F Mur- 

 phy Die and Machine Co., North Quincy, MA) held in 

 a rotating Teflon™ holder over a 4-hour period (Flynn, 

 1968). The activity of 208 Po and 210 Po on the planchets 

 was measured with ion-implant detectors in a Tennelec 

 (Oak Ridge, TN) TC256 or-spectrometer interfaced with 

 a multichannel analyzer and an eight channel digital 

 multiplexer. Counts were recorded with Nucleus'-' soft- 

 ware (Nucleus Personal Computer Analyzer II, The Nu- 

 cleus Inc., Oak Ridge, TN) on an IBM computer. Counts 

 measured over periods that ranged from 28 to 50 days 

 accumulated from 160 to 919 total counts. Lead-210 ac- 

 tivity, along with uncertainty, was calculated in a series 

 of equations that corrected for background and reagent 

 counts, as well as error associated with count statistics 

 and procedure (pipetting error, yield-tracer uncertainty, 

 etc; Andrews et al., 1999a). The remaining sample was 

 dried and conserved for 226 Ra analysis. 



Determination of 226 Ra employed an elemental sepa- 

 ration procedure followed by isotope-dilution thermal 

 ionization mass spectrometry (TIMS) as described in 

 Andrews et al. (1999a, 1999b). The sample was spiked 

 with a known amount of 22s Ra yield tracer estimated 

 to produce a 226 Ra: 228 Ra atom ratio close to one. The 

 samples were dissolved in strong acid and dried re- 

 peatedly (~90-100°C) until the sample color was bright 



