540 



Fishery Bulletin 101(3) 



Figure 4 



Frequencies of sea-surface water temperatures for young Pacific bluefin tuna with archival tags. 

 The number given is the sum (over all individuals) of days in the category based on temperature 

 at m depth from the summary files. Gray, thick bars indicate the sea-surface temperature range 

 that appeared to be accessible to the fish in each month. 



In the detail files obtained from wild fish, the following 

 three types of visceral temperature changes were identified 

 as indicators of feeding ( Fig. 7 1. Type A: a sharp decrease of 

 1-2°C followed by an increase above the initial temperature. 

 Type B: an increase above the initial temperature, preceded 

 by either no decrease or a slow decrease. Type C: a sharp 

 decrease of 1-2°C followed by a return to the initial temper- 

 ature. In addition, in only cases where the visceral tempera- 

 ture changes could not be explained by water temperature 

 changes were they counted as being caused by feeding. For 

 example, a slow decrease of visceral temperature when the 

 fish dived into cold water was not counted as a feeding event. 

 Cases where the water temperature changed frequently 

 were also excluded as too difficult to interpret. 



Feeding events were observed in 942 days out of 1494 to- 

 tal recorded days (63% ). Because it was found that fish did 

 not feed normally for approximately the first 30 days after 

 release (Itoh et al., 2003), data from the first 60 days after 

 release were excluded in the following description. After 

 exclusion of data for the first 60 days, feeding events were 

 observed in 726 out of the remaining 807 recorded days 

 (90%). The number of daily feeding events ranged from 

 zero to ten, with an average of 1.8 ±1.4. Feeding events oc- 

 curred most frequently in the daytime (69% of all obser\'ed 

 feeding events more than 60 days after release), followed 

 by dawn with 27% (Fig. 8). Feeding events rarely occurred 

 at dusk and at night, accounting for only 1% and 3%^ of the 

 total feeding events, respectively. For this analysis, dawn 



