Kerstetter and Graves: Survival of Tetrapturus albidus after release from longline gear 



437 



Data analysis 



Survival of tagged animals was inferred from three 

 types of environmental data provided by the tag: water 

 temperature changes, depth changes, and ambient light 

 intensity. Frequent short-scale t<l hour) variations in 

 both depth and temperature were used as indicators 

 of a live white marlin. The survival of individual fish 

 was also supported by the net displacement, calculated 

 as the distance from the location of the vessel at the 

 time the white marlin was released to that of the first 

 good transmission from the free-floating PSAT to the 

 ARGOS satellite system. The precision of reported loca- 

 tion estimates was based on the attitude of the receiving 

 satellite, and transmissions were generated through the 

 ARGOS system (Service Argos, Inc., Largo, MD) and 

 categorized into seven location accuracy codes. Loca- 

 tions were considered "good" for our study if the ARGOS 

 system reported an accuracy code that corresponded to 

 a distance of less than 1000 meters. If a good position 

 was not obtained directly from ARGOS, an average of 

 all location code "0" readings from the first 24-hour 

 period of transmission was used as a proxy location. All 

 distances were calculated with PROGRAM INVERSE 

 (NGS3). 



Estimates of white marlin postrelease survival were 

 calculated both by including nontransmitting tags as 

 evidence of mortalities and by excluding nontransmit- 

 ting tags. The 95% confidence intervals associated with 

 these estimates were calculated by using the RELEASE 

 MORTALITY version 1.1.0 software developed by Good- 

 year (2002b). These confidence intervals were based 

 on 10,000 simulations where underlying postrelease 



3 NGS (National Geological Survey). 1975. Version 2.0, modi- 

 fied by M. Ortiz, NOAA/NMFS Southeast Fisheries Science 

 Center, 75 Virginia Beach Drive, Miami, FL 33149. 



mortality rates derived from the transmitted data were 

 assumed to have no error sources (e.g., no premature 

 releases or tag-induced mortality). For the purpose of 

 these simulations, natural mortality was also assumed 

 to be zero because of the relatively short duration of the 

 tagging deployment period. Unless otherwise noted, all 

 statistical analyses for this study were conducted with 

 SAS version 8.3 (SAS Institute, Gary, NC). 



Results 



Eight trips (n=112 sets) were taken between June 2002 

 and August 2004 on the FV Carol Ann, a U.S. -registered 

 commercial pelagic longline vessel that operated during 

 the winter and spring in the Caribbean Sea targeting 

 swordfish and during the summer and fall in the mid- 

 Atlantic and Georges Bank region targeting both tuna 

 and swordfish. A summary of these trips and sets is 

 provided in Table 1. Sets were typically made overnight, 

 and gear was deployed at dusk and retrieved at dawn. 

 Catch rates (catch per 1000 hooks) for target and 

 bycatch species varied by season and location. Sword- 

 fish catch rates for retained animals ranged from 1.6 

 (mid-Atlantic, summer 2005) to 23.9 (Caribbean and 

 Gulf of Mexico, spring 2004). Retained tuna (yellowfin 

 [Thunnus albacares]; bigeye, T. obesus; and albacore 

 [T. alalunga]) catch rates ranged from 0.8 (Caribbean 

 and Gulf of Mexico, spring 2004) to 44.2 (mid-Atlan- 

 tic, summer 2004). Istiophorid billfishes (blue marlin, 

 white marlin, longbill spearfish [Tetrapturus pfluegeri], 

 and sailfish [Istiophorus platypterus}) represented ap- 

 proximately 3% of the catch by number, and the overall 

 mean catch rate of white marlin was 1.87 per 1000 

 hooks. Proportions of white marlin dead at the time of 

 haulback varied among sets, trips, seasons, and loca- 

 tions. The lowest observed proportion dead was 34.4% 

 (mid- Atlantic, summer 2005) and the highest was 50% 



