B|orndal et al.: Estimates of survival for Caretta caretta In the North Atlantic 



733 



variety of vessels. Loggerhead sea turtles captured 

 on longline hooks were excluded from this sample 

 to meet the requirement of equal probability of cap- 

 ture across the age range ( see "Results" section ). The 

 turtles were measured, tagged, and released soon 

 after capture. The juvenile loggerhead sea turtles 

 ranged in size from 8.5 to 82.0 cm curved carapace 

 length (mean=33.1 cm, SD=11.6) measured from 

 the anterior point at midline to the posterior notch 

 at midhne between the supracaudals (Bolten, 1999). 

 For 248 turtles, straight-line carapace measure- 

 ments were converted to curved carapace length, 

 as described in Bjorndal et al. (2000). 



The second age distribution ("tuna sample") was 

 a subset of the total sample and comprised 733 log- 

 gerhead sea turtles captured by crews of commercial 

 tuna vessels in the Azores between 1990 and 1992. 

 We analyzed the tuna sample in addition to the to- 

 tal sample because the tuna sample was collected 

 over a shorter interval ( 1990-92) than was the total 

 sample (1984-95). In addition, sizes of vessels from 

 which turtles were captured were more consistent 

 for the tuna sample. This collaborative project with 

 the tuna fleet is described in Bolten et al. (1993); 

 sea turtles are not bycatch in the tuna fishery. Sea 

 turtles were sighted at the surface while the crews 

 were scanning for indications of tuna feeding ac- 

 tivity. The turtles were then captured in dipnets, 

 tagged and measured by a crew member, and re- 

 leased at sea. The juvenile loggerhead sea turtles 

 ranged from 11.0 to 82.0 cm curved carapace length 

 ( mean=33.5, SD= 1 1.2 ). No conversion from straight 

 to curved measurements was required. 



The size distributions were converted to age dis- 

 tributions by using a size-at-age function developed 

 for this population based on a skeletochronological 

 study (Table 1; Bjorndal et al., 2003). Catch curves 

 were generated for each age distribution by plot- 

 ting the natural log of A^^ against x, where N^ is 

 the number of turtles of age x. The catch curves 

 were truncated by excluding age classes with 

 fewer than five individuals, as recommended by Seber 

 (1982). The age at which the population fully recruited 

 to the capture method (threshold age) was identified as 

 the age with the highest ln(A'',). Linear regression analy- 

 ses of the values on the right-hand or declining slope of 

 the distribution were used to generate estimates of total 

 instantaneous mortality rate (Z), which is expressed on 

 an annual basis and is the absolute value of the slope of 

 the regression line. Annual apparent survival probability 

 (<P) was estimated as e^^. The statistical software S-PLUS 

 (Guide to statistics, vol. 1, MathSoft, Seattle, WA) was used 

 for regression analyses. 



Results 



The age distributions for the two samples are shown in 

 Figure 1 and the catch curves in Figure 2. For catch-curve 

 analyses, the age distributions were truncated to the 12- 



12 14 16 18 



120 



Age (years) 



Figure 1 



Age distributions of (A) total sample (/! = 1600) and (B) tuna sample 



(n=733). 



year age class for the total sample and to the 11-year age 

 class for the tuna sample because older age classes con- 

 tained fewer than 5 individuals (Seber, 1982). The selection 

 of the appropriate threshold age — the first fully recruited 

 age class to the capture method — can be difficult but is 

 critical for the analyses (Seber, 1982; Isaac, 1990). We des- 

 ignated 2 years as the threshold age for the total sample 

 and 4 years for the tuna sample; in the tuna sample, the 

 4-year age class was slightly larger than the 2-year age 

 class (120 vs. 118 turtles. Table 1). 



We believe that turtles between ages 2 and 12 in the total 

 sample and between 4 and 11 in the tuna sample meet the 

 assumption of equal capture probability. With increase in 

 turtle size, capture probability is a compromise; greater vis- 

 ibility of larger turtles is countered by greater difficulty in 

 capturing larger turtles. Loggerhead sea turtles captured 

 incidentally by longline vessels were excluded from our 

 sample because there is a capture bias toward the largest 

 size classes. The mean size of the turtles in the total sample 



