McGOWAN: SPAWNING OF NORTHERN ANCHOVY 



collected in February and very few were collected 

 in November. Larvae were present every month and 

 at every station each month with four exceptions: 

 during June, no larvae were collected at station 1, 

 the southernmost station; during July and August 

 no larvae were collected at station 6, the Golden 

 Gate Bridge station; during March no larvae were 

 collected at station 3 in South Bay. Eggs were pres- 

 ent on each of the occasions when larvae were 

 absent from the samples. 



Egg density varied from to 55,000 per 1,000 m 3 

 (mean = 3,000). The greatest number of eggs in a 

 single sample was 14,640 at station 2 in July. Occur- 

 rence of eggs was seasonal: they were abundant in 

 summer and absent in winter (Fig. 2). 



Larvae varied from to 4,400 per 1,000 m 3 

 (mean = 259). The greatest number of larvae in a 

 single sample was 1,420 in September at station 2. 

 Larval abundance was also seasonal with peak den- 

 sity in late summer and fall (Fig. 2). 



Two-way ANOVA of log-transformed standard- 

 ized densities of eggs and larvae were performed 

 with month and station as fixed factors in separate 

 analyses. The interaction mean square (not signifi- 

 cant) was used as the denominator in the F-tests 

 because there was just one observation per cell of 

 the design (Montgomery 1976:156). Densities of 

 eggs differed significantly among months (P < 

 0.001, Table 1) but not among stations (P = 0.104). 

 Densities of larvae were significantly different 

 among months (P = 0.010) and among stations (P 

 = 0.014) (Table 2). 



Seasonal patterns of abundance of eggs and lar- 



Table 1 . — Analysis of variance of northern anchovy eggs: 

 month by station. 



Table 2.— Analysis of variance of northern anchovy larvae: 

 month by station. 



vae were unmistakeable, but differences among sta- 

 tions were not as clear so three hypotheses were 

 tested: 1) stations 1, 2, and 3, South Bay stations, 

 differed from stations 4, 5, and 6; 2) stations 4 and 

 6, Golden Gate and Central Bay stations, differed 

 from stations 1, 2, 3, and 5, South Bay stations plus 

 the station at the outflow of San Pablo Bay; 3) sta- 

 tions 3, 4, and 6, the stations most influenced by 

 ocean water, differed from stations 1, 2, and 5, the 

 Bay stations. These hypotheses were tested using 

 linear contrasts (Nie et al. 1975:425), a procedure 

 that compared the geometric means of the groups 

 of stations. 



None of the three contrasts was significant for 

 eggs but all three were significant (P < 0.05) for lar- 

 vae. The difference between the mean of stations 

 4 and 6 and the mean of stations 1, 2, 3, and 5 was 

 highly significant (P = 0.001). 



Further comparisons of mean densities of larvae 

 were done using Duncan's Multiple Range test. This 

 a posteriori procedure identified groups of means 

 which did not differ significantly from each other 

 at a specified level (Nie et al. 1975:427). The rank 

 order of the stations in increasing mean density of 

 larvae was 4,6, 1,3,5,2. Three groupings were pro- 

 duced by the Duncan procedure at the 0.05 level. 

 The mean of stations 4 and 6 was smaller than the 

 mean of the other four. The mean of stations 5 and 

 2 was greater than that of the other four. Station 

 4 was significantly lower and station 2 significant- 

 ly higher than the mean of the other four stations. 



A summary of the analyses of variance follows. 

 Eggs and larvae were seasonal in abundance, eggs 

 more strongly than larvae. Numbers of eggs, which 

 would be subject to passive drift and dispersal, were 

 not significantly different among locations in the 

 Bay. Larvae did differ in abundance among the six 

 stations. Based on a priori and a posteriori tests, 

 station 4 and station 6, the stations most influenced 

 by oceanic water, had low densities of larvae while 

 the other stations within the Bay had high mean den- 

 sities of larvae. This pattern was true for station 

 5, near the Richmond-San Rafael Bridge, as well as 

 for stations 1, 2, and 3 in the South Bay. Among 

 the within-bay stations, station 1, the southernmost, 

 ranked lowest in both egg density and larval den- 

 sity although it was not statistically different from 

 the other inner stations— 2, 3, and 5. 



The stations also differed in the proportion of eggs 

 to larvae. While the ratio of eggs to larvae was 

 generally greater than 10:1, at station 3 the ratio 

 of the mean number of eggs to mean number of lar- 

 vae was <10:1 (Fig. 3). The proportions were sta- 

 tistically different among stations (Chi-square P < 



883 



