380 



Fishery Bulletin 89(3), 1991 



baitboat fleet, operating in the western North Pacific, 

 and wider-ranging Asian longline and gillnet fleets. 



Except for the gillnet fleet, catch and effort in the 

 surface fisheries has declined in the past 15 to 20 years. 

 The U.S. commercial catch of albacore dropped from 

 approximately 20,000 metric tons per year in the early 

 1970s to less than 5000 metric tons in the late 1980s, 

 while effort dropped over the same period from 40,000 

 boat days to less than 5000 boat days. 



The albacore are very patchily distributed, but evolv- 

 ing satellite technology has helped U.S. fishermen to 

 locate areas of high concentration. Albacore tend to 

 migrate along oceanic thermal fronts, and to form tran- 

 sient aggregations in areas where the frontal struc- 

 tures favor local enrichment (Laurs 1983, Laurs and 

 Lynn 1977). These conditions are detectable by satellite 

 (Laurs et al. 1984, Svejkovsky 1988). Over the past 

 decade, increasingly sophisticated fishing advisories 

 have been provided to fishermen. The advisories in- 

 dicate, from satellite data, the locations of oceano- 

 graphic conditions conducive to albacore aggregation, 

 and fishermen have been taking increasing advantage 

 of these advisories (Laurs 1989). 



The ranges of the population and of the fishing 

 grounds are variable both within the fishing season and 

 from year to year. Albacore in the size range vulnerable 

 to the U.S. fishery are entrained in an annual east-west 

 migration pattern. The U.S. fishery peaks during the 

 summer and autumn months when the albacore are 

 closest to the North American coast. Albacore appear 

 to be separated into northern and southern subgroups 

 divided approximately by the 40°N latitude line (Laurs 

 and Lynn 1977). The timing and extent of albacore 

 migration are variable and without synchrony between 

 the two subgroups. The location of oceanic fronts is also 

 variable. As a result, the boundaries of the fishing 

 grounds are extremely ill-defined and fluid, as is the 

 extent to which the fishing ground overlaps the range 

 of the albacore population. The traditional U.S. fishery 

 is primarily nearshore. However, in the past 10 years, 

 jig boats have been venturing farther offshore, some 

 as far west as the dateline, earlier in the season in an 

 attempt to meet the migrating albacore on their way 

 toward the North American coast. 



Theory and methods 



The routine procedure for estimating fishing power in 

 the U.S. albacore fleet makes use of a computer pro- 

 gram, FPOW, coded by the California Department of 

 Fish and Game and described by Fox (1971) and Berude 

 and Abramson (1972). The basic theory behind this 

 program is described by Robson,(1966). The heart of 

 the method is an analysis of variance which seeks to 



account for variation in the logarithm of CPE due to 

 vessel length-classes, as well as to other factors such 

 as time-area strata. Unlike usual analysis of variance, 

 FPOW concentrates on reporting estimates of the coef- 

 ficients in the statistical model. These coefficients are 

 the logs of the fishing power of vessel classes relative 

 to that of a reference class (45-foot vessels in this case). 

 FPOW does not list a table of residual variances and 

 F statistics, which would indicate the degree of statis- 

 tical significance of the various factors. Accordingly, 

 for the most recent year for which we had data (1988), 

 we used a different analysis of variance program 

 (BMDP) to better reveal the significance of variation 

 due to vessel length in relation to other sources of 

 variation. 



As we looked into the routine procedure for aggre- 

 gating catch and effort over all the time-area strata 

 in a year, we found that effort (standardized for vessel 

 length) and catch in the individual strata have been 

 summed over strata, and a pooled CPE given by 



CPE 



pooled 



(1) 



has been calculated, where Cj and ej are the catch and 

 effort in the t-th stratum. In the case where popula- 

 tion density varies between strata, it is well known that 

 such a pooled CPE is not a good population index 

 because even though CPE might be proportional to 

 population density in individual strata, that propor- 

 tionality is destroyed with a pooled CPE. However, 

 that proportionality can be maintained with a stratified 

 CPE given by 



0PE strat - y 



N 



(2) 



where N is the number of strata (Beverton and Holt 

 1957:148-151). This is simply the average of CPE-s in 

 individual strata. An effort aggregation scheme that 

 in effect does the same thing has been used for the 

 Japanese longline fishery for albacore (Honma 1974). 

 Strictly speaking, Equation 2 presumes that there 

 is at least some effort in all strata. When that is not 

 true, estimates should be provided of what CPE would 

 have been in each of the missing strata had the fishery 

 visited them. In the case of the U.S. North Pacific 

 albacore fishery, the fluid nature of the fishing grounds 

 described above makes it difficult to say whether a 

 given stratum should be considered missing or not 

 present in the fishing ground for a particular year. For 

 the purpose of our new CPE series, we ignored the 

 problem by ignoring unvisited strata. However, if the 



