FISHERY BULLETIN: VOL. 79, NO. 2 



bait fishing vessels on charter to AFRF. Approx- 

 imately lO^c of the tagging was done by commer- 

 cial fishermen trained in tagging procedures, the 

 rest by NMFS technicians. Single Floy spaghetti- 

 dart tags were inserted on the left side below the 

 second dorsal fin with the aid of a beveled stainless 

 steel tube so that the tag barb was lodged in the 

 pterygiophores of the fin. Only fish judged to be in 

 very good condition were tagged; fish hooked in 

 the roof of the mouth or showing signs of extreme 

 exhaustion or severe bleeding were rejected. For 

 each tagged fish a record was kept on 1) tag 

 number, 2) date and time of release, 3) fork 

 length at time of release, 4) condition at tagging, 

 and 5) longitude and latitude of release. Addi- 

 tional tagging details are given in Laurs et al. 

 (1976). 



Recovery Procedures 



Recoveries were made by sport and commercial 

 fishermen, unloaders, and cannery workers. In- 

 formation was obtained on 1) tag number, 2) date 

 of recapture, 3) fork length at time of recovery, and 

 4) longitude and latitude of recapture. Most re- 

 capture locations were given as loran coordinates, 

 which were converted to longitude and latitude, 

 but the recapture locations for tags recovered by 

 unloaders and cannery workers were often re- 

 ported inexactly, e.g., as "off central California." 

 Direct measurements of fork length were avail- 

 able for about one-half the fish recovered. For 

 most of the remainder only the weight at recovery 

 was given, and fork length was estimated using 

 Clemens' (1961) weight-length relation. Observed 

 fork lengths were measured to the nearest centi- 



•* Mention of a commercial company or product does not 

 con.stitute an endorsement by the National Marine Fisheries 

 Service, NOAA. 



meter. Bias in estimating fork length from ob- 

 served weight using the inverted weight-length 

 relationship was judged to be <0.5 cm in absolute 

 value over the length range represented in the 

 recapture sample, so subsequent growth analyses 

 were based on the combined sets of observed and 

 estimated lengths. 



Data Screening 



The tag return data were screened to exclude 

 cases where information was incomplete, unreli- 

 able, or clearly inaccurate. Out of 741 tag returns 

 made from 1971 through 1978, 305 were rejected 

 (Table 1). In 15 cases length at release was not 

 measured, in 116 cases the recapture date was 

 unknown, and in 68 cases neither length nor 

 weight was measured at recovery. In 79 other 

 rejected cases the length at recovery was not 

 measured and the weight only guessed without 

 the use of scales. Additionally, in 27 instances a 

 gross error was apparent in the measurement of 

 fork length either at release or recovery. 



The final accepted data set of 436 cases includes 

 observations on 28 albacore showing negative 

 estimated growth. We assume these are a result 

 solely of measurement error or error in estimation 

 in cases where the recovery weight was converted 

 to length, and we assume such error occurs 

 throughout the data set independently of size or 

 time between release and recapture (time out). 



One of the common steps in screening tag 

 recovery data for growth studies is to partition the 

 data according to length at release, compute 

 linear regressions of growth increment on time 

 out within each subset, and then reject rare 

 observations, usually those departing from expec- 

 tations by more than two standard deviations 

 (Schaefer et al. 1961; Joseph and Calkins 1969). 

 We abandoned this step because the number of 



Table l. — Summary of number of tagged fish released, recovered, rejected, and accepted for analysis of albacore growth. 



Rejected tag recoveries (no.) 



294 



