440 



Fishery Bulletin 104(3) 



these nontransmitting tags were included as evidence 

 of mortalities. 



Nine white marlin were hooked in or near the eye. 

 Seven fish were hooked on either circle or J-style hooks 

 through the eye socket (with no visible damage to the 

 eyeball) and all survived for the 5- or 10-day PSAT 

 deployments. Two PSATs were attached to animals 

 that had been hooked with a circle hook through the 

 eye itself. One PSAT transmitted data consistent with 

 survival, and the other tag did not transmit data. Only 

 one white marlin tagged in this study was foul-hooked, 

 caught in the ventral musculature by a size 18/0 circle 

 hook. The PSAT attached to this fish separated from 

 the fish prematurely. 



Discussion 



The amount of data archived and transmitted varied 

 greatly among the three models of satellite tags, as 

 well as among the 16 transmitting PTT-100 HRs. The 

 early model PTT-100 tags archived only 63 data points, 

 but 100% of the archived information was transmitted, 

 providing sufficient information to infer survival (Graves 

 et al., 2002; Kerstetter et al., 2003). In contrast, the 

 newer PTT-100 HR tags archived either 4500 or 9145 

 data points, but not all archived data were transmitted. 

 In this study, most of these tags transmitted a rela- 

 tively large percentage of the archived data, facilitating 

 determination of the fate of the released white marlin. 

 However, one tag (MA-04-08) had an unusually low data 

 transmitting rate of 4.4% , representing 315 data points 

 over the ten-day tag deployment. Because these data 

 points were transmitted in 11-minute blocks (approxi- 

 mately 9 data points each), they often included complete 

 short-duration movements of a fish from the surface to 

 depth. As the transmitted blocks of data were distributed 

 haphazardly over the entire ten-day tagging period, it 

 remained possible to determine postrelease survival from 

 a high-resolution tag with a low data recovery rate. 



Prior studies of postrelease survival have used differ- 

 ent lengths of time to ascertain the effects of capture. 

 These have included studies focused on postrelease 

 survival as well as others addressing long-term be- 

 havior, movements, and habitat preferences. Graves 

 et al. (2002) justified a five-day deployment period for 

 blue marlin by citing reports of blue marlin recaptured 

 within five days after being released with conventional 

 tags from the recreational fishery, thus demonstrating 

 a return to feeding. Kerstetter et al. (2003) adopted a 

 similar position, although their study on blue marlin 

 also included the deployments of two PSATs for 30 days 

 to evaluate the possibility of delayed mortality. Do- 

 meier et al. (2003) used a variety of deployment periods 

 (1-12 month durations) to assess postrelease survival in 

 striped marlin. However, the longer the PSAT deploy- 

 ment period, the more susceptible the animal becomes 

 to both fishing (i.e., recapture) and natural mortality, 

 such as predation, biasing upwards the estimate of 

 postrelease mortality (Goodyear, 2002b). 



In our study, we primarily used tags with a ten-day 

 deployment period and believe that this period is suf- 

 ficiently long to document short-term mortality. Five of 

 seven white marlin mortalities reported in Horodysky 

 and Graves (2005) occurred within the first six hours 

 of release, and the other two died less than three days 

 later. All of the mortalities inferred for the closely re- 

 lated striped marlin by Domeier et al. (2003) occurred 

 within six days of release, and 75% of these mortalities 

 happened in less than two days. The two documented 

 mortalities in the present study (GB-02-01 and MA-03- 

 04) occurred within 24 hours of release. 



Direct comparisons of estimates of postrelease sur- 

 vival of billfishes among previous acoustic and PSAT 

 studies are problematic. Many acoustic tracking studies 

 had relatively short observation periods and low sample 

 sizes, and often fish in marginal physical condition were 

 not tagged (reviewed in Domeier et al., 2003). Even 

 among PSAT tagging studies, nontransmitting tags 

 have been addressed with different protocols by various 

 authors. Neither Graves et al. (2002) nor Kerstetter et 

 al. (2003) directly observed mortalities of PSAT-tagged 

 blue marlin. However, in both studies a conservative 

 approach was adopted to estimate postrelease survival 

 by considering nontransmitting tags as representing 

 mortalities; this approach was adopted in part because 

 of a lack of emergency release software or mechanisms 

 on the tags themselves that would release the PSAT 

 prior to its sinking with a dead fish below the depth at 

 which the tag would be crushed. Some new models of 

 satellite tags possess such emergency release software 

 or physical mechanisms, such as glass implosion devices 

 (Domeier et al., 2003) or the RD-1500 metal guillotine 

 from Wildlife Computers (Redmond, WA) that sever 

 the tether of the tag prior to reaching the depth limit 

 of the tag. New generations of tags are also rated to 

 greater crush depths (ca. 2000 m) than earlier models. 

 The PSATs used in our study, with the exception of the 

 one PAT tag, did not possess emergency release soft- 

 ware or physical mechanisms. Because all the animals 

 in this study were tagged over or near waters deeper 

 than the crush depths of the tags, any deaths of tagged 

 white marlin could have resulted in the PSATs being 

 destroyed prior to transmitting data while the tag re- 

 mained attached to the sinking, moribund fish. 



There are several reasons why PSATs may not re- 

 port even with emergency releases, including recovery 

 of the tag by a noncooperative fishing vessel, internal 

 malfunction, or biological activities. Kerstetter et al. 

 (2004) reported on three PSAT tags that were presum- 

 ably ingested by sharks during predation or scavenging 

 and suggested that a number of nontransmitting tags 

 in all PSAT studies could result from biological activity. 

 Goodyear (2002b) noted that including nontransmitting 

 tags as mortalities would bias mortality estimates up- 

 wards if the failure to transmit data was due to causes 

 other than mortality. 



The combination of physically more robust tags, emer- 

 gency release capabilities, and demonstrated mortalities 

 has led authors (e.g., Domeier et al., 2003) to specifi- 



