228 



Fishery Bulletin 89(2), 1991 



MAM J J ASONDJ F MAM J J ASONDJFMAM 



MAM J J ASONDJ F MAM J J ASONDJFMAM 



100 

 80 

 60" 

 40 

 20 



(c) Vona Vona 

 S gracilis 



t 



MAMJJ ASONDJ FMAMJJ ASONDJ FMAM 



1987 1988 1989 



Time (months) 



Figure 7 



Monthly variation in proportion ( ± 95% confi- 

 dence limits) of female Archamia zosterophora 

 at Munda and Vona Vona (a,b), and female 

 Spratelloides gracilis at Vona Vona (c), March 

 1987-May 1989. 



out the year. At Vona Vona, however, there was a peak 

 in spawning activity in samples taken in October 1988. 

 The spawning data for S. gracilis are consistent with 

 data for other Spratelloides at Vona Vona, with a 

 decline in spawning activity during late 1988 and 1989 

 (Fig. 7). 



Eggs and larvae Encrasicholina eggs were present 

 in the zooplankton at each site most months of the 

 sampling period (Fig. 8). The abundance of eggs did 

 not vary greatly at each site during the period, al- 



though the samples from Vona Vona had fewer eggs 

 than the other sites. Overall, the mean Encrasicholina 

 egg density for the entire sampling period varied from 

 0.08 ± 0.03/m 3 at Vona Vona to 0.43 ± 0.10 at Munda. 

 The overall mean teleost egg density was also lower 

 at Vona Vona (1.34 ± 0.21/m 3 ) and highest at Munda 

 (4.98 ± /m 3 ). Encrasicholina egg density did not cor- 

 relate with total egg density at any site (P>0.3) or with 

 the proportion of spawning E. devisi or E. heterolobus 

 (P>0.5). No Spratelloides larvae were found, although 

 apogonid larvae were present but could not be iden- 

 tified to species. 



The density of Encrasicholina larvae followed a 

 similar pattern at all sites (Fig. 8): larvae were pres- 

 ent in most months, increasing every three or four 

 months. In contrast to egg densities, Encrasicholina 

 larvae reached higher densities at Vona Vona (0.05 ± 

 0.01/m 3 ) than at Munda (0.03 ± 0.006/m 3 ). Other fish 

 larvae were also in higher densities at Vona Vona 

 (1.15 ±0.27) than at the other sites (Munda 0.93 ± 

 0.26 and Tulagi 0.60 ± 0.22). The density of Encrasi- 

 cholina larvae was significantly correlated with the 

 density of other fish larvae at Munda (r 2 0.62, 

 P<0.05), but not at other sites. 



Proximate spawning stimuli 



To assess the independence of the 11 proximate stimuli, 

 all variables were correlated with one another (Table 

 3). At all sites, most measures of rainfall were signif- 

 icantly correlated (P<0.05). Total rainfall was nega- 

 tively correlated with moon phase at Vona Vona and 

 Tulagi, and with zooplankton biomass at Tulagi. Cloud 

 cover was correlated with total rain and days of rain 

 at Tulagi (P<0.05). Zooplankton biomass and density 

 were correlated at Munda and Tulagi. However, there 

 were no consistent correlations at all sites between non- 

 rainfall variables (Table 3). 



Principal component analysis was also performed 

 separately on the proximate variables from each site. 

 Rainfall variables at each site had loadings on the first 

 three factors that were similar in magnitude and direc- 

 tion. Zooplankton density and biomass also covaried at 

 each site. Temperature, moon phase, tide, and wind had 

 loadings that varied independently for the first three 

 factors at each site. These variables, zooplankton den- 

 sity, and the number of days of rain (which covaried 

 least with other rainfall variables) were used in a 

 separate stepwise regression analysis. 



Encrasicholina Analysis of the relationships between 

 all 11 environmental parameters and the proportion of 

 each sample spawning showed no consistent pattern 

 for either species of Encrasicholina (Table 4). At 

 Munda, spawning of E. devisi positively correlated 



