Mann et al  Reproduction and recruitment oi Mugil curema 



813 



Figure 4 



Number of growth increments (counted on seven independent occasions) on otoliths 

 from 1 1 juvenile white mullet of varying size. Means, standard errors, and standard 

 deviations are illustrated. 



Ul=- 



100 



where f = Coriolis parameter; 



Tjj. = surface wind stress; and 



p„, = average density of the water (1025 kg/m'^). 



The term /'was calculated as 



/■=2ojsin(/;), 



where co = angular velocity of rotation of the earth 

 (7.29xl0-5s [seconds]): and 

 /j = latitudinal position at the place i. 



The term t^^ represents surface wind stress measured in 

 the .r-axis perpendicular to the coast (Bowden, 1983), often 

 considered in terms of the empirical equation 



Tj^ = ^ X p_j X W^, 



where k = empirical drag coefficient (1.11 to 3.25, as a 

 function of wind velocity; Bowden 1983); 

 p^ = mean density of the air ( 1.25 kg/m^); and 

 W = wind velocity. 



pling period during each of the 18 months of the study by 

 applying the frequency distribution of birth dates of aged 

 juveniles to the total catch for that date. A total of 398 

 juveniles were aged by otolith analysis. If x% of aged fish 

 captured on a given date were hatched on Julian Day y, 

 this percentage was applied to the total catch of juveniles 

 for that sampling date. Secondly, all fish hatched on a given 

 day were summed across the 18 monthly sampling dates. 

 This frequency distribution was then correlated with the 

 distribution of UI estimates over the same period of time 

 as that of the birth dates. 



Before proceeding with correlation, trends in birth date 

 and UI data series were described by using a smooth- 

 ing spline. The spline fit uses a set of smoothly spliced 

 3'''' degree polynomial segments (Simonoff, 1996; JMP® 

 software, version 3.2.1, SAS Institute, Gary, North Caro- 

 lina). Predicted values were correlated with the raw data 

 points in order to optimize the value of lambda used to 

 fit the smoothing spline. Increasing the value of lambda 

 increases the degree of smoothing but weakens the correla- 

 tion between predicted and raw data. Pearson correlation 

 and cross-correlation functions were used to describe the 

 temporal relationship between upwelling and the date of 

 hatching offish recruited to the coastal lagoon. 



The drag coefficient, k, changes as a function of wind 

 velocity and gives values equivalent to those of Bakun et 

 al.(1974). 



The relationship between upwelling and the birth dates 

 of successful recruits (captured in La Restinga Lagoon) 

 was determined in two steps. First, we calculated the 

 birth dates of juveniles captured during one monthly sam- 



Results 



The white mullet surveyed in the commercial catches 

 measured from 4 to 36 cm SL. The largest fish were from 

 the Chacopata zone where the most abundant sizes classes 

 were those from 22 to 30 cm. The most abundant sizes in 

 the Cariaco Gulf and Margarita Island zones were 20 to 



