516 



Fishery Bulletin 101(3) 



second always contained the latest data. We divided the file 

 into two 40-day sections for releases in 1995 and 1996, and 

 into 20- and 60-day sections for releases in 1997. 



Reliability and calibration of archival tags in air 



To examine the reliability of location estimates made 

 by archival tags, 117 archival tags were left outdoors 

 (34°59'N; 138°59'E) where they were not affected by artifi- 

 cial light during July-September 1996 (55 days, five tags), 

 May-August 1997 (86 days, 14 tags), and October 1997 

 (five days, 100 tags). Two of the tags were used in two of 

 the experiments. 



Calibration tests of internal and external temperature 

 sensors were conducted for all tags before being implanted 

 in fish that were released, and the sensors were recalibrat- 

 ed for nine tags after they were recovered. Temperature 

 calibration was done by immersing tags into a series of 

 water tanks that were set to temperatures ranging from 

 5.0° to 30.0°C by 5°C intervals. Calibration tests of pres- 

 sure sensors were also conducted for all tags before release 

 and on 27 tags after being recovered. Tags were placed in 

 a pressure chamber with a resolution of 0.1 bar and exam- 

 ined up to 20 bar. The tags were left at least five minutes 

 at each temperature or pressure to obtain at least two 

 measurements at the 128-second recording interval. 



Experiment with pen-held fish 



Archival tags were attached to three pen-held young Pacific 

 bluefin tuna of 93-97 cm fork length (FL) at Kasasa in 

 Kagoshima Prefecture (31°25'N; 130°11'E) in November 

 1994. The fish had been reared in a net pen (40 m x 25 m 

 with 12 m depth) for more than two years and were accli- 

 mated to the environment at the time of the experiment. 

 Archival tags were inserted into the abdominal cavities 

 of two of the three fish by the following method. A fish 

 caught by hook and line was put into a styrofoam box, and 

 its eyes were covered with a black polyethylene bag. The 

 belly of the fish was cut with a scalpel about 4 cm anterior 

 to the anus, 3—4 mL of antibiotic (artificial penicillin, Doil, 

 Tanabe Seiyaku Co., Ltd., Osaka, Japan) was injected 

 into abdominal cavity of the fish, and an archival tag was 

 inserted there with the stalk extending through the inci- 

 sion. A stitch was made in the middle of the incision with 

 an absorbable suture (Coated Vicryl, type J583G, Ethicon 

 Inc., Cornelia, GA ), and the fish was released back into the 

 pen. All tools and tags were disinfected with 100% ethanol. 

 No anesthetic was used because with their eyes covered, 

 the fish remained quiet during the surgery. This simplified 

 procedure (from making the incision to releasing the fish) 

 could be completed in less than 90 seconds, thus minimiz- 

 ing total stress on the animal and, in later experiments 

 on wild fish, providing the best chance for the animal to 

 rejoin its original school. In this pen study, the third fish 

 was tagged externally instead of internally, the tag being 

 connected by a thin wire rope to a small metal arrowhead 

 inserted in a muscle near the second dorsal fin base. 



During the pen-held fish experiment, none of the fish 

 were observed to die as a result of tagging. The tag that 



had been attached externally came loose from the fish 

 and was retrieved from the bottom of the pen four days 

 after tagging. One tagged fish escaped when the pen was 

 broken. The remaining tag was recovered 453 days later 

 when the fish was caught from the pen as part of a com- 

 mercial catch. 



Experiments with wild fish 



Tag and release experiments on wild young Pacific bluefin 

 tuna were conducted near Tsushima, at the northeastern 

 end of the East China Sea, by using chartered commercial 

 trolling vessels, every November and December from 1995 

 to 1997. A total of 166 fish, ranging from 43 to 78 cm FL 

 (age or 1), were internally tagged as described above and 

 released immediately. Two dart-type conventional tags 

 were also attached to the second dorsal-fin base of each fish 

 in the 1997 experiment as visual markers in an attempt to 

 improve the recovery rate. 



Thirty archival tags (18.1%) were recovered. The dura- 

 tions at sea were 50 days or less for 13 fish, 96-211 days 

 for 13 fish, and 359-375 days for three fish, all recaptured 

 around Japan. One additional fish was recaptured off the 

 west coast of Mexico, on the east side of the Pacific Ocean, 

 at 610 days after release. Data could not be downloaded 

 from one archival tag released in 1995 and recovered 30 

 days after release; all other tags returned data. 



Results 



Reliability of location estimates 



The tag recovered from a fish penned in a known location 

 for 453 days yielded a record of positions automatically 

 estimated during that time. Figure 1 plots the errors in 

 those estimates and the date when each was made. This 

 tag provided the only position sequence of long duration 

 obtained from a captive fish. Unfortunately it was discov- 

 ered later, after the experiment was completed and after 

 this particular tag was no longer available for further test- 

 ing, that the light sensitivity of this tag, as well as that of 

 the tag that yielded data for four days in the captive fish 

 experiment, was at least a factor of ten lower than that of 

 other tags. This discrepancy in light sensitivity could be 

 seen in the daily noon-light intensity data in the summary 

 file, both during the in-water experiment (when compared 

 with typical values for tags in wild fish) and when tested 

 in air (compared with other tags of the group tested in air). 

 On dark days there was an unusual pattern of early sunset 

 times and late sunrise times that the tag manufacturer 

 interpreted as being associated with the low light sensi- 

 tivity. Thus, although the general trends of error size with 

 season can be expected to be representative, the absolute 

 size of the errors was likely inflated in this, the only long- 

 term record obtained from a captive fish. 



Longitude error showed no change with season, but 

 latitude error increased dramatically near the equinoxes 

 as expected because day length does not vary significantly 

 with latitude at that time, and therefore carries little in- 



