C = n r * n s * n t * C r + n t * C t (8) 



where C = yearly cost of seine program in personnel hours (362h) 

 C r = cost of processing one replicate (2h) 

 C t = cost of one 2-day sampling trip (9.25h) 

 n = number of replicates 

 n = number of stations sampled 

 n t = number of 2-day sampling trips 



while an appropriate description of the variance components was given 



by: 



s 2 s 2 s 2 



s 2 » _1 + _L + L_ (9) 



n t n g n r n s n t 



where s 2 = variance of mean log^Q(CPUE +1) 



s 2 = estimated variance component due to time 

 s| = estimated variance component due to station 

 s| = estimated residual variance 



To obtain the solution that minimizes the variance for a fixed 

 cost, Lagrange multiplier techniques were followed (see TASC 1979). 

 This gave: 



s 2 C + s 2 C /n 2 

 s e t r 



n r C r s t 



(10) 



n t = C/(n r n s 'C r + C t ) (11) 



as the number of stations and times per year to sample for a given level 

 of replication (1, 2 or 3) . 



The log.j _ transformed catch data were analysed as time series. The 

 autocorrelation structure of the data was investigated by visually 

 evaluating the autocorrelation function generated by the PROC ARIMA 



procedure of SAS. The general hypotheses being considered were that the 



44 



