FISHERY BULLETIN: VOL. 85, NO. 4 



major fleets span the years 1970 through 1980. 

 We processed the data into an "average year", 

 that is, the average (over the years 1970-80) of 

 catch by size, zone, month, and gear and the aver- 

 age of effort by zone, month, and gear (Kleiber 

 and Baker"^). The effort values were used directly 

 in the model and the catch and effort used to esti- 

 mate other input parameters. 



We estimated the 1970-80 average recruitment 

 and preliminary catchability values by size and 

 gear by use of a size-structured cohort analysis 

 (Jones 1981). The catch-at-size vector necessary 

 for this cohort analysis was obtained from the 

 average year by aggregating over zones, averag- 

 ing over months, and smoothing over size classes. 



To conduct the cohort analysis, we needed to 

 specify an average final cohort size, which was 

 unknown to us. We tried a series of values and 

 chose results for which the overall exploitation 

 rate (catch divided by recruitment estimate) was 

 close to the overall exploitation rate estimated 

 from tagging. Tagged albacore have been re- 

 leased in the U.S. fishery at an average size of 

 approximately 65 cm, and approximately 6% of 

 the tags have been recovered (Laurs"^). Nonre- 

 porting losses are small for the major fisheries 

 that recovered the tags (Laurs^). Assuming a 

 value of 10% for nonreporting and Type I and II 

 tag losses of 12% and 0.098 year'^ respectively 

 (Laurs et al. 1976), the exploitation rate of re- 

 cruits to 65 cm should be approximately twice the 

 raw recovery percentage. But the exploitation 

 rate in the cohort analysis and the simulation 

 model is based on recruits to 25 cm which should 

 be approximately twice as numerous as recruits 

 to 65 cm (based on growth and natural mortality 

 rates used in the model). Therefore, the exploita- 

 tion rate of recruits to 25 cm should be approxi- 

 mately equal to the raw tag recovery percentage 

 (6%). We chose a cohort analysis with an exploita- 

 tion rate of 6.3% as the basis for the results pre- 

 sented below except where we discuss sensitivity 

 to exploitation rate for which we repeated the 

 analysis several times starting at this point with 

 a series of cohort analyses at a series of higher 

 exploitation rates. 



3Kleiber, P., and B. Baker. 1986. Development of catch 

 and effort data base for the North Pacific albacore simulation 

 model. U.S. Natl. Mar. Fish. Serv., Southwest Fish. Cent., Ad- 

 min. Rep. LJ-86-26, 21 p. 



4Laurs, R. M. 1979. Results from North Pacific albacore 

 tagging studies. U.S. Natl. Mar. Fish. Serv., Southwest Fish. 

 Cent., Admin. Rep. LJ-79-17, 10 p. 



5R. M. Laurs, Southwest Fisheries Center La JoUa Labora- 

 tory, National Marine Fisheries Service, NOAA, P.O. Box 271, 

 La Jolla, CA 92038, pers. commun. March 1987. 



The cohort analysis yielded a Pacific-wide re- 

 cruitment estimate. We apportioned one third of 

 this recruitment to each of the three southern 

 zones, where albacore larvae are predominantly 

 found (Nishikawa et al. 1984). 



Size-specific overall fishing mortalities ob- 

 tained from the cohort analysis were apportioned 

 to gear type by the proportion of the total catch at 

 each size that is taken by each gear type in the 

 average year. We then converted the fishing mor- 

 talities into catchabilities by dividing by the over- 

 all average effort for each gear type. 



Size-specific growth coefficients, Gg, were the 

 growth rates in length (dl/dt) at the upper end of 

 each size class divided by the length of the size 

 classes (5 cm). We estimated the growth rates 

 from the derivative form of the von Bertalanffy 

 growth equation 



^J-^(Zoo-l) 



where U is 135.6 cm and k is 0.014 month" ^ 

 (Clemens 1961). We used these same values in 

 the size-structured cohort analysis, which re- 

 quired input of growth information. 



A value for natural mortality was also needed 

 both in the cohort analyses and in the model. We 

 used a value of 0.017 month"^ (0.2 year"^) (Suda 

 1966). 



The tag data would be a good source of informa- 

 tion to estimate migration coefficients except that 

 the tag recovery effort is not uniformly dis- 

 tributed and analytical techniques to deal with 

 that situation are not well developed. To get a 

 reasonable set of migration coefficients, we quan- 

 tified the experience of three experts, scientists 

 knowledgeable about the North Pacific albacore 

 fisheries and the available tag data. We first 

 asked the experts to identify the significant paths 

 (movement from one zone to an adjacent zone) for 

 each of a series of broad (10 cm) size classes. For 

 each path we then asked how the intensity of 

 migration via that path is distributed over the 

 months of the year and we evaluated the average 

 intensity by asking the experts the following 

 question: On average, during the season of this 

 migration, if 100 fish of the given size class are 

 now in the origin zone, how many of these (irre- 

 spective of mortality) would be expected to be in 

 the destination zone one month from now? We 

 calculated the average migration coefficient for 

 the particular path by 



706 



