Itoh et al,: Migration patterns of Thunnus orienta/is determined with archival tags 



515 



et al., 1997). These methods have not yielded detailed 

 information regarding migration, behavior, and their rela- 

 tion to environmental factors for Pacific bluefin tuna over 

 a long period. 



An archival tag is an electronic device that measures 

 environmental variables and records data in its memory. 

 When attached to an animal, it allows direct examination 

 of the relationship between an animal's behavior and 

 physiological condition, or the ambient environment. One 

 type of "archival" tag merely stores data; however, another 

 type not only stores data but also provides daily geographi- 

 cal locations of the fish by processing the measured envi- 

 ronmental data. This type of archival tag was anticipated 

 since the 1980s as a tool that could collect detailed infor- 

 mation on individual fish behavior (Hunter et al., 1986; 

 Anonymous, 1994). Metcalfe and Arnold (1997) estimated 

 the geographical locations and tracks of plaice, a demersal 

 species, by comparing tidal depth variations with the time 

 series depth data recorded by archival tags attached to the 

 fish. However, this method is not suitable for pelagic fish, 

 which change swimming depth freely. A type of archival tag 

 that can estimate geographical locations based on change 

 of light levels during a day — a method more suitable for 

 pelagic species — has been commercially available since the 

 early 1990s. So far, archival tags of this type have been used 

 in several tagging projects (Arnold and Dewar, 2001). The 

 results published in a few reports on southern bluefin tuna 

 (T. maccoyii) (Gunn and Block, 2001), and Atlantic bluefin 

 tuna (T! thynnus) (Block et al., 2001), show the remarkable 

 value of archival tag data. 



As archival tags have come into wide use, results of sev- 

 eral experiments conducted to evaluate the reliability of 

 its geolocation estimates have been published (Welch and 

 Eveson, 1999; Musyl et al., 2001; Gunn et al.^). However, 

 several points remain to be tested: tag reliability when a 

 number of tags are deployed for long duration, reliability 

 of sensors for variables other than light, and the effects of 

 attaching the tag to fish. 



After two preliminary experiments with tags in 1994, 

 the first with tags placed at a known outdoor location on 

 land and in air, and the second with tags attached to young 

 Pacific bluefin tuna held in pens, we applied archival tags 

 to wild young Pacific bluefin tuna to investigate their be- 

 havior and migration. In the present study we report the 

 characteristics of migration for this species based on data 

 on daily geographical location, as well as the reliability 

 of archival tag data and the effect of attachment of the 

 tag to fish. Analyses for swimming depth, ambient water 

 temperature, and feeding frequency of the species are 

 undertaken in other papers (Kitagawa et al., 2000; Itoh 

 et al.,2003). 



2 Gunn, J., T. Polacheck, T. Davis, M. Sherlock and A. Betlehem. 

 1994. The development and use of archival tags for studying 

 the migration, behavior and physiology of southern bluefin tuna 

 with an assessment of the potential for transfer of the technol- 

 ogy to groundfish research. Proc. ICES mini-symposium on 

 migration, St. Johns, Newfoundland. ICES CM. Mini:2.1, 23 p. 

 International Council for the Exploration of the Sea, Palaegade 

 2-4, DK-1261 Copenhagen K, Denmark. 



Materials and methods 



Outline of the archival tag used in this study 



The archival tag used in this study (Northwest Marine 

 Technology, Inc. Shaw Island, WA ) had a cylindrical stain- 

 less-steel body ( 16 mm in diameter and 100 mm long, and 

 weighing 52 g) that was implanted in the animal. A flexible 

 sensor stalk 2.2 mm in diameter and 150 mm long extended 

 from the tag through the skin of the animal into the water. 

 The end of the stalk housed an external temperature sensor 

 and a light capture region. Light was led from the capture 

 region by optical fiber to a photodiode sensor in the body 

 of the tag, which also housed sensors for pressure, internal 

 temperature, and light. Response times for the tempera- 

 ture sensor were three seconds for the external sensor and 

 20 seconds for the internal sensor, and temperature resolu- 

 tion was 0.2°C for both sensors. Resolution of the pressure 

 sensor record was 1 m at shallow depths up to 126 m, then 

 changed to 3 m from that depth to the scale limit of 510 m. 

 Clock drift was less than 30 seconds per year. The tag had 

 a data measurement interval of 128 seconds, a 256-kByte 

 data memory, and an operating life exceeding seven years. 

 Data were downloaded from recovered tags by using a per- 

 sonal computer and a fiber-optic connector. 



Two types of data files were created within the tag 

 memory. One data file stored daily records containing date, 

 estimated times of sunrise and sunset, water temperatures 

 at m plus two other selectable depths (we selected 60 m 

 and 120 m), and other information required for, or produced 

 in, the course of location estimates for each day. This file 

 is referred as the "summary file" in the "Results" section, 

 and it stored data for all days after the memory was last 

 cleared. The times of sunrise and sunset were estimated 

 within the tag from sea-surface light intensities, which 

 were inferred from measured depth and measured light 

 intensity at depth and a water opacity factor determined 

 from the measured data each Universal Time (UT) day. The 

 time of midday was determined as the midpoint between 

 sunrise and sunset times, and longitude was calculated 

 from the difference between the midday time and 1200h 

 UT, at a rate of 15 degrees longitude per hour, corrected 

 for astronomical effects. Latitude was estimated from the 

 duration of daylight (Hill, 1994). 



The second data file contained unprocessed time series 

 data records taken at 128-second intervals. The tag could 

 record at any integer multiple of its 128-second measure- 

 ment interval and a multiple of one was chosen. Each 

 record consisted of external temperature, internal tem- 

 perature, pressure, and light intensity, and corresponded 

 to a known time. This is referred to as the "detail file" in 

 the "Results" section. It could hold about 54,000 records, 

 or about 80 days of steady recording at the high data rate 

 chosen — a small fraction of the tag's overall lifetime. The 

 time-series memory was divided into two sections, and the 

 size allocations for the two sections were determined by the 

 user The first section filled first and did not change there- 

 after. The second section filled next, but once full, it was 

 continually overwritten by new data. Thus the first section 

 always contained the earliest data retrieved from a tag; the 



