McGOWAN: SPAWNING OF NORTHERN ANCHOVY 



Table 5.— Bivariate correlations between northern anchovy eggs, larvae, and other 

 variables. EGGS: log (eggs-m -3 ); LARV: log (larvae -m" 3 ); ZOOP: log (zooplank- 

 tersm -3 ); MICR: microzooplankton; TEMP: surface water temperature; SALI: sur- 

 face water salinity; TSTR: temperature stratification; SSTR: salinity stratification; SECC: 

 Secchi disk depth. 



where E = log (eggs/1,000 m 3 + 1) 

 T = surface temperature (°C) 

 M = log (microzooplankton/^ + 1), (Table 6). 



No single variable explained the majority of the 

 variability in larval density (Table 7). Secchi depth 

 was the single best predictor, accounting for 11% 

 of the variance of larval density (r 2 = 0.113). The 

 combination of surface temperature with Secchi 

 depth increased the coefficient of determination to 

 0.306. All of the variables combined explained just 

 50% of the variability of larval density (r 2 = 

 0.498). Five variables improved the prediction of the 

 set of independent variables by more than 1% when 

 added to the model. The predictive equation for lar- 

 val density based on using these five is 



L = -0.842 - 0.591X + 0.1267/ + 0.515Z 



Table 6. — Stepwise multiple regression: northern anchovy 

 egg density vs. biological and environmental variables. 



- 0.57LW + 0.029S 



where L 

 X 

 T 

 Z 



M 

 S 



+ 1) 



log (larvae/1,000 m 3 

 Secchi depth (m) 

 surface temperature (°C) 

 log (zooplankton/1,000 m 3 

 log (microzooplankton/^ + 

 surface salinity (%>o). 



+ 1) 

 1) 



The results of the multiple regressions show that 

 northern anchovy egg density could be predicted 

 largely by surface water temperature. Larval den- 

 sity could not be predicted well by a single variable 

 or by the five variables which, when combined, ac- 

 counted for only 49% of the variability. 



Spawning Stock Estimates 



Based on estimates of egg production, the spawn- 



Table 7.— Stepwise multiple regression: northern anchovy 

 larval density vs. biological and environmental variables. 



ing stock biomass of northern anchovies in the part 

 of San Francisco Bay sampled in this study ranged 

 from undetectable in December 1978 and January 

 1979 (no eggs collected) to 696 t (metric tons) (767 

 short tons) in July 1978. If the area of the Bay which 

 is <2 m deep were included, the estimate of July 

 biomass would have been 2,030 1 (2,240 short tons). 



Length Frequencies of Larvae 



Monthly samples could contain larvae from the 

 current month and 2 previous ones because meta- 

 morphosis is not complete until 35 mm, age 74 days 

 at 16°C (Hunter 1976). However, larvae longer than 

 15 mm were not taken at the standard stations from 

 August through October, although eggs and smaller 



887 



