89 



Abstract. - We examine some issues 

 concerning the accuracy of larval age 

 estimates for Sebastes jordani. To lo- 

 calize a starting point for daily incre- 

 ment counts, gestating and planktonic 

 larvae were examined to determine 

 whether an extrusion check forms at 

 parturition. We also assessed 1) the 

 resolution limitations of optical micros- 

 copy in ageing larvae, 2) the precision 

 of age estimates, and 3) the effect of 

 specimen age on the reliability of ages. 

 Preextrusion increments are com- 

 monly observed in gestating and plank- 

 tonic larvae, but a distinctive extrusion 

 check forms in the otolith at parturi- 

 tion. This feature occurs predictably at 

 a radius of 15-19 //m and provides an 

 unambiguous starting point for incre- 

 ment counts. Observations of larval 

 otoliths with a scanning electron micro- 

 scope revealed that the minimum size 

 of postextrusion increments is -0.6 /im, 

 which is large enough to be resolved 

 with a high-quality optical system. 

 Cross-checking of ages among three 

 readers showed 84</r agreement to 

 within ±1 d and no systematic differ- 

 ences among readers. The clarity of 

 otolith microstructure resulted in high 

 confidence scores for age estimates; 

 however, very young larvae (0-3 d) 

 were the most difficult to age. An expo- 

 nential growth model fitted to 2,203 

 aged larvae indicated that during the 

 first month of life, growth in length 

 proceeds at the rate of 2.75% d" 1 . 



Accuracy of age estimates for larval 

 Sebastes jordani 



Stephen Ralston 



Tiburon Laboratory 



Southwest Fisheries Science Center. NOAA 



3150 Paradise Drive. Tiburon, California 94920 



Edward B. Brothers 



EFS Consultants 



3 Sunset West. Ithaca, New York 1 4850 



Dale A. Roberts 

 Keith M. Sakuma 



Tiburon Laboratory 



Southwest Fisheries Science Center, NOAA 



3150 Paradise Drive. Tiburon, California 94920 



Manuscript accepted 15 September 1995. 

 Fishery Bulletin 94:89-97 (1996). 



Ichthyoplankton surveys are used 

 routinely to estimate spawning 

 stock biomass (Gunderson, 1993; 

 Hunter et al., 1993). Fundamen- 

 tally, this type of calculation re- 

 quires a determination of total daily 

 egg production (P [eggsd -1 ]), which 

 is estimated by regressing egg 

 abundance on egg age. Typically, 

 egg ages are obtain ?d indirectly by 



1 ) description of egg stage ontogeny, 



2) laboratory studies of the tem- 

 perature dependency of egg devel- 

 opmental rates, 3) measurement of 

 thermal conditions in the environ- 

 ment at the time of sampling, and 

 4) back calculation of egg age. Not 

 surprisingly, serious complications 

 can occur with this approach, espe- 

 cially when spawning occurs in deep 

 water (e.g. Loet al., 1993; Picquelle 

 andMegrey, 1993). 



Members of the rockfish genus 

 Sebastes (family Scorpaenidae) are 

 viviparous, bearing advanced yolk- 

 sac larvae at the time of parturition 

 ( Wourms, 1991; Bowers, 1992). Pre- 

 vious work has shown that Sebastes 

 larvae can be aged directly by enu- 

 meration of daily otolith increments 

 ( Penney and Evans, 1985; Laidig et 



al., 1991). Thus, rockfishes exhibit 

 a life history adaptation, i.e. vivi- 

 parity, which potentially lends itself 

 to accurate estimation of daily lar- 

 val production rates (P [larvae-d -1 ]). 

 The purpose of this study is to 

 examine the accuracy of larval 

 Sebastes age estimates obtained 

 through the study of otolith micro- 

 structure ( Stevenson and Campana, 

 1992). To be considered accurate, a 

 sample statistic must be close to the 

 true underlying "population" value 

 it estimates. Inaccuracy is due to 

 the combined effects of estimator 

 bias and imprecision. Although di- 

 rect ageing of rockfish larvae cir- 

 cumvents many of the problems as- 

 sociated with staging eggs (see 

 above), accurate larval age esti- 

 mates, characterized by low bias 

 and high precision, are still prereq- 

 uisite to an accurate larval produc- 

 tion estimate of spawning stock bio- 

 mass. Because larval mortality 

 rates in the marine environment 

 are high, commonly exceeding 

 0.15-d" 1 (Pepin, 1991; Houde, 1994), 

 systematic larval ageing errors as 

 small as one day can significantly 

 bias estimates of spawning biomass. 



