Sampling Error for Hatchling Turtle Measurements: Probing a Rule-of-Thumb

Despite the importance of understanding error associated with sampling, there has been a paucity of discussion in many analyses within sea turtle biology. We investigated sampling and non-sampling errors for size measurements (straight carapace length, straight carapace width, and mass) of hatchling Flatback Turtles, Natator depressus, from two study sites in tropical Australia using a priori and post hoc methods. A priori and post hoc methods produced similar error estimates (max difference = 0.19 g and 0.24 mm), highlighting methodologies that may be transferred for other species without requiring a census of every hatchling in a clutch. The standard practice of sampling ten hatchlings per nest resulted in reasonably low sampling error (upper 95% CI = 0.9 g for mass, 0.7 mm for straight carapace length, and 1.0 mm for straight carapace width), but sample size may be reduced if less precise size measurements are required. Effort analysis showed that sampling error from a nesting population was minimized when sampling effort was spread over the maximum number of clutches rather than concentrating effort within fewer clutches.

Fig. 1

Sampling error for (A) mass, (B) straight carapace length to notch (SCLmin), (C) straight carapace length to tip (SCLn-t), and (D) straight carapace width (SCW) measurements of hatchling Flatback Turtles calculated using a sampling theory formula with a mean clutch size correction factor (dashed lines) and resampling from the true data (solid line).

Fig. 2

Difference in sampling error between methods using a priori sampling theory and resampling real data for (A) mass, (B) straight carapace length to notch (SCLmin), (C) straight carapace length to tip (SCLn-t), and (D) straight carapace width (SCW) of hatchling Flatback Turtles. A priori sampling theory with a finite population correction factor equal to clutch size for each nest was used as it had the closest approximation to sampling from real data.

Fig. 3

Boxpot showing within clutch variability in measurements for straight carapace length to notch (SCLmin), straight carapace length to tip (SCLn-t), straight carapace width (SCW), and mass. Plot shows grouping by study site, year, and collection method and presents the minimum, first quartile, median, third quartile, and maximum values for each measurement.

Fig. 4

Sampling error within nests and within the nesting population for sample sizes from 1–49 hatchlings and 1–31 clutches, respectively for (A) mass, (B) straight carapace length to notch (SCLmin), (C) straight carapace length to tip (SCLn-t), and (D) straight carapace width (SCW) of hatchling Flatback Turtles.

Fig. 5

Combined sampling error (within nest and within population) for mass of hatchling Flatback Turtles for (A) the number of hatchlings sampled per clutch and (B) the number of clutches sampled per population given a set sampling effort of 100, 200, 300, and 500 hatchlings. Within population sampling error for 30 clutches was used for sample sizes of greater than 30 clutches due to the limit of 31 clutches sampled in this study.

Contributor Notes

¹ Faculty of Education, Health and Science, Charles Darwin University, Northern Territory 0909, Australia; E-mail: andrea.whiting@cdu.edu.au; and au.whiting@gmail.com. Send reprint requests to this address.

² Ecological Modelling Services P/L, P.O. Box 6150, University of Queensland, St. Lucia, Queensland, 4067, Australia.

³ Environmental Protection Agency, P.O. Box 15155, City East, Brisbane, Queensland 4002, Australia.

Received: 27 Aug 2007

Accepted: 21 May 2008