Benson et al.: Abundance, distribution, and habitat of Dermochelys corioceo off California 



341 



minant of the level of survey coverage achieved. 

 Leatherback turtle encounter rates were identi- 

 cal for Beaufort sea states 0-1, 2, and 3 (0.003 

 turtles/km). Cloud cover and glare categories 

 did not have a significant effect on encoun- 

 ter rates (P=0.08 and P=0.23, respectively). 

 The number of leatherback turtles seen per 

 year ranged from 2 to 28, and totaled 100 in- 

 dividuals for all years (Table 1). The majority 

 of turtles were subjectively estimated to be 5-7 

 ft (1.5-2.1 m) in total length, but only three 

 smaller individuals (3.5-4.5 ft; 1.1-1.4 m) and 

 one very large individual, estimated to be about 

 7.5 ft (2.3 m), were also recorded. Whenever 

 possible, the presence of a long tail (indicating 

 an adult male) was noted; however, this feature 

 was often difficult to determine from the air- 

 craft. In particular, males with tails of interme- 

 diate length may have had a greater likelihood 

 of being recorded as "tail length undetermined." 

 The proportion of identified males, 6 of 44 (149c), 

 therefore, is a minimum proportion of males in 

 the study area. Greatest concentrations of leath- 

 erback turtles were observed in the Gulf of the 

 Farallones stratum, but turtles were observed 

 in all geographic strata (Fig. 1). 



Estimation of line-transect parameters 



All three detection function models yielded simi- 

 lar estimates of /"(O), and AIC values were within one 

 point. The Hazard rate model (Buckland et al., 2001) 

 was selected because it provided the best fit, especially 

 near the transect line (Fig. 2), yielding an estimated 

 /■(0)=4.465 (coefficient of variation, CV=0.136l. 



During the visibility calibration experiment, only 

 the Secchi disk at 1 m depth was visible to the aerial 

 observers; therefore, the depth at which leatherback 

 turtles were detected was estimated to be about 1 m. 

 Time-depth recorders (TDRs) were attached to three 

 turtles (1 male, 2 females) on 29 September (for 153 

 minutes), 30 September (167 minutes), and 13 Octo- 

 ber, 2005 (229 minutes). There was no visible reaction 

 by the turtles to the application of the tag, and the 

 proportion of time spent within 1 m of the surface did 

 not differ between the first 30 minutes and subsequent 

 30-minute periods of tag deployment (P=0.08). The 

 parameter g(0) was, therefore, estimated from the com- 

 plete TDR dive record as the average proportion of time 

 leatherbacks spent at or above 1 m depth. The three 

 individuals exhibited a remarkably similar proportion 

 of time spent within the upper meter of the sea surface 

 (Table 2), andg(O) was estimated as 0.471 (CV=0.029). 

 Corrected estimates of abundance thus are about twice 

 the uncorrected values (Fig. 3). 



Abundance and density 



Estimated leatherback turtle abundance was vari- 

 able among years (Fig. 3; Table 3), ranging from 12 



(CV=0.74) during 1995 to 379 (CV=0.23) during 1990. 

 The greatest proportion of turtles was encountered 

 within the two central California strata (Monterey Bay 

 and Gulf of the Farallones), accounting for an average of 

 72% (range 31-97%, CV=0.37) of the total abundance. 

 In partial survey years, when only these two strata were 

 surveyed, total abundance within the study area was 

 estimated as the central California abundance divided 

 by the mean percentage (72%). For all years combined, 

 estimated leatherback turtle abundance averaged 140 

 (CV=0.17) within the central California strata and 178 

 (CV=0.15) for the entire study area (Table 3). Although 

 the Gulf of the Farallones stratum contributed the most 

 to overall abundance because of its larger size, turtle 

 densities were only slightly less for the Monterey Bay 

 and Pt. Arena strata (Table 3). The South Central 

 California and North Coast strata had the lowest den- 

 sities. Monthly encounter rates of leatherback turtles 

 by stratum (Fig. 4) were consistent with past reports 

 of frequent sightings in Monterey Bay during August 

 (Starbird et al., 1993); however, in our study, encounter 

 rates were also high during September in the Monterey 

 Bay and Gulf of the Farallones strata, and during Octo- 

 ber within the Gulf of the Farallones. Encounter rates 

 decreased markedly throughout the study area during 

 November. Interannual variability was least during 

 September, and regionally within the Gulf of the Faral- 

 lones stratum. 



The estimates of abundance of leatherback turtles off 

 California (Fig. 3) did not exhibit a trend between 1990 

 and 2003 (P=0.19 when data for all ten survey years 

 were used, P=0.41 including only the seven coastwide 

 survey years) but appeared to be related to the average 

 annual NOI (Schwing et al., 2002), i.e, there were posi- 

 tive index values associated with greater leatherback 

 turtle abundance and vice versa (Fig. 5; P=0.03 when 



