SOURCES OF VARIATION IN CATCH PER UNIT EFFORT 



OF YELLOW! AIL FLOUNDER, LIMANDA FERRUGINEA (STORER), 



HARVESTED OFF THE COAST OF NEW ENGLAND 



LORETTA O BrIEN AND RALPH K. MaYO^ 



ABSTRACT 



Factors affecting variability in commercial catch per unit effort (CPUE) of yellowtail flounder were 

 examined in order to establish a basis for standardizing fishing effort. Analysis of variance ( ANOVA) 

 procedures were employed to test for differences in CPUE among vessel tonnage class, fishing area, 

 and depth zone and the interactions between tonnage class and area, and tonnage class and depth. 

 Vessel tonnage class and fishing area accounted for highly significant IP < 0.01) sources of variation 

 in CPUE whereas depth was not significant (P > 0.05) in most cases. Interactions between tonnage 

 class and stock area were also highly significant in all cases. A series of annual fishing power 

 coefficients was computed for each tonnage class relative to a standard for each stock based on 

 parameter estimates obtained by fitting the CPUE observations to a linear model with tonnage class 

 as the independent variable. Deviations of annual fishing power coefficients from the 20-year mean 

 were found to exhibit significant first order autocorrelations. Consequently, annual coefficients were 

 computed over the entire 1964-83 period by incorporating tonnage class, annual and seasonal effects 

 as independent variables in a three-way linear model. Although the standardized CPUE estimates 

 obtained from this procedure are similar to those obtained by previous methods, the revised proce- 

 dures described in this paper insure adequate representation of all vessel classes engaged in the 

 yellovfcrtail fishery in the CPUE calculations. 



Fishing effort and resulting catch per unit effort 

 (CPUE) indices are routinely used in assessing 

 the impact of commercial fishing operations on 

 stock abundance. The traditional concept that ag- 

 gregate CPUE indices may be used to measure 

 annual changes in relative stock abundance is 

 based on the principal assumption that the catch- 

 ability coefficient (q) either remains constant 

 over all fleet components, or that nominal effort is 

 adjusted to account for differences in relative effi- 

 ciencies (Pope and Parrish 1964; Kimura 1981). 

 Variation in q may be due to persistent differ- 

 ences in fishing power of various types of gear or 

 to technological innovations which may be gradu- 

 ally introduced over time (Gulland 1964; Sis- 

 senwine 1978). Biological interactions such as 

 changes in availability of a species due to sea- 

 sonal distribution patterns or to annual changes 

 in abundance may also affect the overall catcha- 

 bility of demersal species (Garrod 1964; Pope and 

 Garrod 1975). Variability in catchability coeffi- 

 cients may be taken into account by relating nom- 

 inal fishing effort of each fleet component to some 

 chosen standard category. 



'Northeast Fisheries Center Woods Hole Laboratory, Na- 

 tional Marine Fisheries Service. NOAA, Woods Hole, MA 

 02543. 



Manuscript accepted October 1987. 

 FISHERY BULLETIN; VOL. 86. NO. 1, 1988. 



Numerous authors have described the basic 

 procedures for calculating relative fishing power 

 of various fleet components. Beverton and Holt 

 (1957) provided evidence to suggest that the dis- 

 tribution of logarithms of fishing power factor/ 

 vessel tonnage ratios could be described by a nor- 

 mal curve while Gulland (1956) employed an 

 analysis of variance (ANOVA) model of log 

 CPUE. The properties of the ANOVA model were 

 further examined by Robson (1966) who extended 

 the techniques developed by Gulland (1956) and 

 formally specified the analysis of Beverton and 

 Holt (1957) as a two-way multiplicative ANOVA 

 model. Stern and Hennemuth (1975) employed 

 the method of Robson (1966) in their analysis of 

 fishing effort in the U.S. Georges Bank haddock 

 fishery using depth fished and vessel tonnage as 

 classification variables. In a previous study, 

 Rounsefell (1957) computed standardized log 

 CPUE indices to determine relative abundance of 

 several co-occurring species on Georges Bank ac- 

 cording to depth. More recently, Gavaris (1980) 

 and Kimura (1981) have developed modifications 

 of the ANOVA model to estimate annual stan- 

 dardized CPUE indices from time series of catch 

 and effort data by incorporating a year effect in 

 the model. 



Standardized annual CPUE indices based on 



91 



