Quantifying site fidelity in marine turtles is often hampered by study-specific metrics and untested sampling representativeness. We conducted snorkel-based photo-identification surveys of foraging green turtles at two adjacent reef sites on Liuqiu Island, Taiwan, across nine quarterly sampling periods (October 2020–November 2022). Of 4,191 sightings, 3,845 (92%) were valid and identified 398 unique turtle individuals. Using iNEXT sample coverage, we assessed four survey designs: coverage increased from moderate (0.65–0.70) under single-event surveys to sufficiently high (0.93–0.95) under triple-event surveys, with modest gains under > triple-event surveys (0.96–0.98) and higher resighting probability (1.1-fold vs. triple-event; 1.7-fold vs. single-event). Tide cycles influenced detections: flood and high tides yielded the most sightings, whereas flood and ebb tides yielded more unique individuals, indicating detectability bias if tide structure is ignored. Using the representative mark-resight dataset, 82% (n = 327) of turtles were resighted, and 55% (218) stayed for over 1 year. The Standardized Site Fidelity Index (SSFI) and one-dimensional hierarchical clustering identified two fidelity groups: Lower (49%; 0–0.021) and Higher (51%; 0.0212–0.0827), with adult-sized turtles had significantly lower SSFI than juveniles and subadults. Interstudy comparisons show that our surveys produced a more evenly distributed SSFI profile and moderate-to-high SSFI values. Because SSFI max depends on sampling frequency and study duration, comparisons should focus on distributional patterns and explicitly report sampling effort. Together, coverage diagnostics, standardized fidelity metrics, and open data provide a transferable framework for comparable site-fidelity inference and evidence-based management of foraging turtle hotspots.
Fong et al. (Wed,) studied this question.