526 



Fishery Bulletin 101 (3) 



125°E 



130°E 



35° N 



30° N • 



rr«.».-^ 



35° N 



30° N ■' 



Figure 9 



Rapid movements of a fish with an archival tag in the residency 

 phase in the East China Sea. The fish moved rapidly south (upper 

 panel; open circles are in December 1995 and solid circles are in 

 January 1996) and north (lower panel; open circles are in April 

 1996 and solid circles are in May 1996). 



Discussion 



Reliability of archival tag data 



The reliability of archival tag data for geolocation estimates 

 based on measured light inten.sity ha.s been examined by 

 implanting the tags in pen-held fish or attaching the tags to 

 a stationary subsurface mooring (Welch and Eveson, 1999; 

 Musyl et al., 2001; Gunn et al.-». About one degree of reli- 

 ability for both longitude and latitude were the results. Our 

 study included further tests: for a large number of tags; for 

 sensors other than light sensors; for reliability of tags over 

 time; for tags manufactured by Northwest Marine Technol- 

 ogy that were applied to wild young blucfin tuna and not 

 fully examined in previous studies; and finally for the effect 

 of tag attachment on Pacific bluefin tuna. 



The benchmark test in this study showed that longitude 

 estimated by archival tags had an error (mean ±standard 

 deviation) of -0.54° ±0.75°, which differed by only -0.1° 

 ±0.8° from a comparison of in-water tag position results 

 with actual recapture locations. In the on-land benchmark 

 test, the mean error did not change with tag age, although 

 the standard deviation slightly increased. Ninety-six 

 percent of all tags tested were considered to have suffi- 

 cient reliability in longitudinal estimation. We concluded, 

 therefore, that the archival tag is a reliable tool to estimate 

 longitude on a scale of about one degree. 



Latitudes estimated automatically from day length car- 

 ried larger errors than estimations of longitude, and the 

 accuracy of estimation changed with season as well as with 

 the latitude itself (Hill, 1994; Hill and Braun, 2001). Smith 

 and Goodman (1986) recommended estimating latitude by 



