Long-teim Tieiuls in Catchpci-iinit.of-effort 



The catch-per-hundi-ed-hooks is CPUE and 

 an index of apparent abundance for the albacore 

 on the North Pacific longline grounds. This 

 section provides indices that are used to esti- 

 mate trends in apparent abundance for the 

 longline fishery. 



In developing the indices we first considered 

 the use of the ratio of averages CPUE statistic 

 versus the average of ratios CPUE statistic. 

 (The ratio of averages CPUE statistic is com- 

 puted by dividing the total catch by the total 

 effort, whereas, the average of ratios CPUE 

 statistic is computed by averaging individual 

 CPUE ratios.) These two statistics are plotted 

 for a typical Marsden square in the longline 

 albacore region as a time series (fig. 3). Fig- 

 ure 3 shows that the ratio-of-averages statistic 

 approximates the average-of-ratios statistic 

 during the nonpeak fishing months, but during 

 the peak fishing months the ratio-of-averages 

 statistic generally exceeds the average-of- 

 ratios statistic. The latter situation (see Gul- 

 land, 1955) indicates that the fishermen are 

 concentrating their effort upon the fish (or the 

 fish are becoming more vulnerable in the most 

 heavily fished areas). We note, however, a 

 tendency for the average of ratios to exceed the 

 ratio of averages in months of relatively high 

 apparent abundance in some of the terminal 

 years of the time series. We do not know 

 whether this reversal in the relation between 

 the average-of-ratios statistic and the ratio- 

 of-averages statistic reflects a shift in fishing 

 emphasis from albacore to another species 

 such as bluefin tuna or to a response to a gen- 



eral decrease in the apparent abundance of 

 albacore. We decided to use the average-of- 

 ratios statistic in our analyses since (1) it 

 shows roughly the same declining trend as the 

 ratio-of-averages statistic, (2) it is unweighted 

 by the distribution of effort, and (3) by virtue 

 of the central limit theorem it is more amen- 

 able to the normality assumptions of regression 

 analysis (which will be used subsequently). 



The variability of the CPUE index and also 

 the effects of the varying relation between the 

 two ratio statistics were reduced by taking ad- 

 vantage of the rather well-defined (especially 

 when contrasted with trends in apparent abun- 

 dance of tunas in the low latitudes) seasonal 

 periodicity of albacore CPUE. This well- 

 defined seasonal periodicity reflects a fairly 

 consistent correlation among the monthly 

 CPUE's for any year. For example, the CPUE 

 in January is always relatively high whereas 

 the CPUE in July is always relatively low. 

 This consistent ranking suggests that the vari- 

 ability in a composite annual CPUE index can 

 be considerably reduced by not averaging the 

 monthly CPUE's for each year, but rather by 

 studying the trends among Januaries, among 

 Februaries, and the rest of the months. We 

 assume from examining our data that the er- 

 rors induced by any lack of synchrony (see fig. 

 3 for typical example) among the annual CPUE 

 cycles are relatively small. We also assume 

 that most of the information pertinent to long- 

 term changes in apparent abundance is con- 

 tained in the data for the midwinter months at 

 the peak of the fishery. 



s 

 o 



igure 2. --Albacore landings taken by § 

 various fisheries in the Pacific "* 

 Ocean, 1950-59 (from Yoshida and 

 Otsu, 1963: p. 317). 



1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 



