670 



Fishery Bulletin 93(4), 1995 



200 



150 



100 



• •••• •••• 



1990-91 1991-92 1992-93 1993-94 



n = 316 n=467 n= 1,421 n = 87S 



• • • • 



1990-91 1991-92 



• • • • • 



1992-93 1993-94 



Figure 3 



The total hours per night when the moon was 

 below the horizon and the tide was flooding 

 (hours of dark flood) throughout the four winter 

 study periods (top graph). Dotted vertical lines 

 indicate where years are joined by lunar date. 

 Circles on abscissae indicate new moons. Total 

 abundance of bonefish, Albula vulpes, larvae col- 

 lected by channel nets during the four consecu- 

 tive winters (bottom graph). Values above each 

 year indicate total catch. 



Few significant correlations between recruitment 

 of bonefish and oceanographic or meteorological con- 

 ditions were detected (Table 1). Recruitment was 

 positively correlated with along-shore winds to the 

 northwest with a lag of zero days for the 1991-92 

 season, whereas recruitment peaks the following 

 winter lagged winds to the southeast by three days. 

 No significant relationships between along-shore 

 winds and recruitment were found during the other 

 two winters, nor were relationships detected between 

 recruitment and cross-shelf winds or either cross- 

 shelf or along-shore currents. 



Temporal patterns of recruitment during 

 summer 



A total of 1.112A. vulpes leptocephali were collected 

 during summer 1992. Over 76% of these fishes were 



O 



O 



o 



30 9.5 5.6 4 3.1 2.5 2.1 



Cycling period (days) 



-0.2 



5 10 15 20 25 30 35 40 45 



Lag (days) 



o 



o 



"°' 6 -20 -15 -10 -5 5 10 15 20 



Lag (days) 



Figure 4 



Periodogram plot of logged larval concentrations 

 of bonefish, Albula vulpes (top graph), auto- 

 correlation function of log (rc + 1) larval abundance 

 l middle graph) and cross-correlation function of 

 log (n + 1) larval abundance with hours of dark 

 flood tide (bottom graph). All analyses were con- 

 ducted on the 277-day period which resulted from 

 joining the four years of recruitment (and hours 

 of dark flood) by lunar date. Dashed lines in 

 autocorrelation and cross-correlation plots indi- 

 cate approximate 95% confidence intervals. The 

 cross-correlation plot coincides with lunar days 

 starting with the new moon (lag of zero). 



taken during the first 12 days of the 72-day sam- 

 pling season, and a peak of 160 leptocephali were 

 taken during a single night (Fig. 5). Although this 

 sampling period is too short for time-series analysis, 

 recruitment appears to be limited to periods with 

 relatively high amounts of flood tide occurring un- 



