OWEN ET AL.: SPAWNING AND SURVIVAL OF LARVAL ANCHOVY 



plankton. Plankton community structure was 

 different between sites: most notably, predatory 

 copepods were more abundant at Site 1 and salps 

 were more abundant at Site 2. 



Larval Parameter Variations Within and 

 Between Drifters and Sites 



Larval Growth Rates: 



Ranges of size and age of larvae differed be- 

 tween sites. Site 2 samples included larvae that 

 were larger (and older) than those at Site 1. 

 The average age of larvae was 6.9 and 11.5 days 

 from hatching at Sites 1 and 2, respectively. 

 Larval length averaged 7.8 mm and 9.0 mm at 

 Sites 1 and 2, respectively. Thus the offshore 

 Site 2 was inhabited by older and slightly longer 

 larvae than Site 1. Offshore drift of larvae 

 hatched inshore into the region of Site 2 (Smith 

 1972) may account for this phenomenon, as may 

 higher predation on older larvae at Site 1. The 

 latter explanation is favored because the ob- 

 served flow patterns (Fig. 2) do not suggest 

 that larvae were transported offshore from 

 nearshore areas. 



To assess the growth of larvae in these two 

 stations, the Laird-Gompertz model was used to 

 fit the length-age data: 



L, = L.(Lo/LJ^>^Pl-«'l 



where t is time since first feeding at 5 days after 

 hatching, Lo is larval length at first feeding, and 

 L» is the asymptotic length. The maximum 

 and minimum fish lengths in the sample have 

 strong influence on the length parameter esti- 

 mates (Lo and L^) and on the growth coeffi- 

 cient (a). The resulting growth coefficients at 

 the two sites (Table 2) are 0.1 for the inshore 

 station and 0.05 for the offshore station. A stan- 

 dard growth coefficient of 0.05 is used by 

 CalCOFI (Methot and Hewitt 1980). 



We compared total growth rates of fish 

 younger than 18 days (after hatching) at the two 

 sites, assuming a linear growth rate to be 

 reasonable because of difficulties in comparing 

 differences between nonlinear curves. Larvae at 

 the offshore Site 2 had a larger asymptotic 

 length and a lower growth coefficient than at the 

 inshore site. The null hypothesis, that growth 

 was equal at the two sites, was tested using 

 analysis of covariance. When the test was per- 

 formed on lengths uncorrected for shrinkage or 

 on lengths corrected for the same shrinkage, the 



slopes at the two sites were not significantly 

 different but the mean lengths were. (A ti-ue 

 difference between mean lengths would indicate 

 that anchovy larvae had hatched at a larger size 

 at Site 1 than at Site 2, or that Site 2 larvae 

 underwent starvation at the first-feeding stage 

 and then recovered.) However, when larval 

 lengths from Site 2 were corrected for greater 

 handhng time, then neither the slopes nor the 

 adjusted lengths differed significantly between 

 the two sites (Table 3). In short, corrected data 

 indicate that no difference existed in larval 

 growth between the sites. 



Larval Production and Mortality Rates 



Larval production at age (larvae <20 days 

 from spawning) was used to model larval mortal- 

 ity. For mortahty analysis, age is defined as time 

 since spawning. Larval production was com- 

 puted as described above in Methods, and larval 

 age was derived from live size using a Gompertz 

 growth curve. Three growth curves were avail- 

 able for computing larval age: the standard 

 growth curve used for routine annual anchovy 

 larval assessment, and two site-specific growth 

 curves constructed from length-age data col- 

 lected at each of our sites. We elected to use the 

 standard growth curve to convert larval size to 

 age for both sites because no significant differ- 

 ence in total growth rates for larvae <20 days 

 since spawning was detected between two site- 

 specific growth curves, and because the growth 

 coefficients estimated from two sites were also 

 similar to that of the standard growth model (all 

 equal to 0.05) when the maximum and minimum 

 lengths of larvae were set to be 27 mm and 4.1 

 mm, corresponding to the standard growth 

 curve. Both nonlinear and log-linear regression 



Table 2. — Parameters of Laird-Gompertz growth curves 

 for northern anchovy larvae at Site 1 and Site 2. Lq is larval 

 length at first feeding; L. is the asymptotic length; and 

 a is the growth coefficient. Lengths are corrected for differ- 

 ential handling times. 



681 



