944 



Fishery Bulletin 101(4) 



ational fishery off Bermuda. The four blue marlin tagged 

 off Florida during 2000 moved northeast to east in direction, 

 which is inconsistent with the generally northward flow of 

 the Gulf Stream as it exits the Florida Straits. Both the 

 direction and distance moved by these five fish are evidence 

 of their survival; had these fish been dead and floating, the 

 expected path of movement would have been northward 

 at a velocity of 1.5-3 knots in the Gulf Stream. Of the 

 two fish tagged in 2001, one (30D-1) moved 1126.4 km 

 (608.3 nmi) northeast toward the central north Atlantic, 

 and the other (30D-2) moved 2249.4 km ( 1214.7 nmi) north 

 toward the Grand Banks. 



According to the minimum straight-line distances, blue 

 marlin in this study moved an average of 0.95 nmi/hour 

 (22.9 nmi/day average; range: 15.1-39.2 nmi), which is 

 slightly faster than the average speed of 0.73 nmi/hour 

 (17.6 nmi/day average; range: 10.9-26.4 nmi) reported by 

 Graves et al. (2002). The results of both PSAT studies are 

 consistent with the swimming velocities of 1-2 nmi/hour 

 reported in the acoustic tracking of Pacific blue marlin by 

 Holland et al. ( 1990), but at the lower end of the range of 

 directly measured swimming speeds of 0.29-4.37 nmi/hour 

 for blue marlin with an acoustic tag reported by Block et 

 al. (1992). However, both measurements from acoustic 

 studies were of short duration deployments. As noted in 

 Block et al. (1992), high initial movement rates would be 

 consistent with actions likely taken by ram-ventilating fish 

 under oxygen debt; slower average speeds over the dura- 

 tion of the deployment could result from the averaging of 

 high initial speed with later slower movements upon the 

 return to normal behavioral patterns. 



Forward movement and inclinometer data 



The inclinometer values for each of the seven reporting 

 5-day tags indicated that the PSAT was depressed below 

 30° above horizontal for an average of 47.25% (range: 

 46.77-47.74%) of the measurements. These values indi- 

 cate continual forward movement through the end of the 

 5-day sampling period and are similar to the observations 

 of Graves et al. (2002), who reported substantial forward 

 movement for more than 40% of their inclinometer mea- 

 surements. They are also consistent with the observed net 

 displacements of the blue marlin in the present study. Incli- 

 nometer values following release for all PSATs indicated 

 that the tags were floating in a vertical position. 



Depth and temperature 



The temperature data of the 5-day tags indicated numer- 

 ous vertical movements into cooler water for each fish (Fig. 

 2, A-E). The maximum temperature recorded by each tag 

 was equivalent to, or slightly greater than, the sea surface 

 temperature (SST) recorded by available SeaWIFS satel- 

 lite data for that area and date. The slightly higher tem- 

 perature is possibly an artifact of the 5-day tag; its black 

 coloration may have absorbed heat from direct sunlight 

 while at the surface. Each fish demonstrated movement 

 into slightly cooler waters immediately following release, 

 a behavior also noted in acoustic tracking studies (e.g. Hol- 



land et al, 1990; Holts and Bedford, 1993), although it is 

 likely that this movement to cooler waters was still within 

 the upper strata of the water column. 



Differences in vertical behavior were noted among indi- 

 viduals. The vast majority (98.6%) of values reported by the 

 5-day tag on the Bermuda-released fish fell within a range 

 of 2°C (28.6°-30.6°C) (Fig. 2A). This range is smaller than 

 those reported by Graves et al. (2002) for blue marlin in the 

 same general area (range: 26-31°C) but of a much smaller 

 sample size (n=l vs. n=8 recovered tag datasets). The four 

 blue marlin tagged with 5-day tags off Florida exhibited 

 greater variation in thermal habitat, but their tempera- 

 ture values remained within a range of 6.5°C (Fig. 2, B-E). 

 These four fish also displayed greater vertical movements 

 during the morning hours, especially 5D-3 and 5D-7 (Fig. 

 2, B and E). For the four blue marlin tagged with 5-day 

 PSATs off Florida in 2000, there was a significant differ- 

 ence between the average temperature during the night, 

 day, and a composite dawn and dusk period (one-way 

 ANOVA; P=0.0003, 2 df). 



The 30-day PSATs provided much more detailed tem- 

 perature information than the 5-day tags, as well as depth 

 data, allowing for a higher resolution of blue marlin verti- 

 cal movements over the course of a day. The two fish with 

 the 30-day tags spent the vast majority of their total time 

 within the upper five meters of the water column (65.4%- 

 for tag 30D-1 and 81.5% for 30D-2, Fig. 3). The higher 

 temperatures recorded by tag 30D-1 are likely due to 

 the warmer surface waters encountered by that fish as it 

 moved toward the central Atlantic, whereas the blue mar- 

 lin with tag 30D-2 moved northeast into cooler waters near 

 the Grand Banks (Fig. 1). Examination of the temperature 

 histograms generated the by tags for each hour-long period 

 also indicates that these two fish spent all of the first two 

 (30D-1) and six (30D-2) days following release at or near 

 the surface (depth <5 m). Only after this initial period did 

 the fish resume the repetitive deep diving behavior seen 

 later during the deployments. 



The two 30-day tags deployed in 2001 recorded a broader 

 range of temperatures (30D-1: 29.6-17.8°C and 30D-2: 

 30.6-16.6°C) than the 5-day tags attached to the blue 

 marlin in 2000. This apparent difference may relate to the 

 measurement protocol of each type of tag. The 5-day tags 

 recorded temperature once an hour and stored the average 

 of two hourly values. In contrast, the 30-day tags recorded 

 temperature values every minute and stored them in the 

 form of an hourly histogram. Inspection of the 30-day tag 

 data revealed that excursions into cooler water were typi- 

 cally of short duration. It is therefore likely that such an 

 excursion could have been missed with a once-an-hour 

 temperature measurement by the 5-day tag. Alternatively, 

 if a lower temperature were encountered at one measure- 

 ment, it would probably be increased by being averaged 

 with another (likely warmer) measurement from the pre- 

 vious or next hour before being archived. It can be shown 

 mathematically that if the two observations in each pair 

 are independent (which would occur if they are sufficiently 

 separated in time), then the effect of averaging them would 

 be to reduce the variance by one-half However, cases can be 

 constructed where the observations in each pair would not 



