Scan-sampling interval shapes the reliability of immersion-activity estimates in captive penguins
Abstract
Instantaneous scan sampling is widely used across behavioural observation settings, but sampling intervals are often chosen for practical reasons despite their influence on estimation accuracy. We evaluated how sampling interval affects estimates of immersion activity in captive Magellanic penguins ( Spheniscus magellanicus ) and whether errors differ between group- and individual-level analyses. Using biologging-derived immersion records, we reconstructed immersion-state sequences and simulated scan sampling at intervals of 30 s and 1, 2, 3, 4, 5, and 30 min, accounting for scan-start phase. Estimation error increased nonlinearly with sampling interval but was smaller and less variable at the group level. Median group-level error remained below 0.7% at 5-min intervals and reached 3.6% at 30 min. In contrast, individual-level error increased sharply beyond 4–5 min and reached 8.5% at 30 min. Sampling interval dominated estimation error, with smaller detectable seasonal effects. Frequent land–water transitions further amplified errors at longer intervals. These results indicate that group-level aggregation buffers error but may also conceal individual-level heterogeneity. For group-level summaries, intervals up to 5 min may provide stable estimates, whereas individual-level monitoring requires finer resolution, approximately 2–3 min. More broadly, our findings highlight that sampling interval, scan phase, state switching, and aggregation level jointly determine the reliability of scan-sampling estimates.
Related
- Reliability-Centered Evaluation of Sparse Longitudinal CT Lesion-Size Forecasting with Conformal Interval Calibration and Gompertz-Inspired Regularization
- Available time-series of terrestrial animals provide low reliability in population trend estimates
- Geometric Reliability of Neural Population Codes: Sampling Calibration and Within-Session Nonstationarity
- Immersion 360, a european patient-partnered research project aimed at developing communication skills in paediatrics and raising awareness of interprofessional collaboration practice in partnership.
- Inducing and measuring identity levels in professional rugby players during immersion in an interactive virtual environment
BibTeX
@article{yamamoto2026scansampling,
title = {{Scan-sampling interval shapes the reliability of immersion-activity estimates in captive penguins}},
author = {Takashi Yamamoto and Tomohiro Kakizaki and Yukari Takashima and Hiroe Tsunesumi and Ken Yoda},
journal = {Scientific Reports},
year = {2026},
month = oct,
doi = {10.1038/s41598-026-77369-w},
url = {https://www.nature.com/articles/s41598-026-77369-w},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Scan-sampling interval shapes the reliability of immersion-activity estimates in captive penguins
AU - Takashi Yamamoto
AU - Tomohiro Kakizaki
AU - Yukari Takashima
AU - Hiroe Tsunesumi
AU - Ken Yoda
PY - 2026
DA - 2026/10/28
JO - Scientific Reports
DO - 10.1038/s41598-026-77369-w
UR - https://www.nature.com/articles/s41598-026-77369-w
AB - Instantaneous scan sampling is widely used across behavioural observation settings, but sampling intervals are often chosen for practical reasons despite their influence on estimation accuracy. We evaluated how sampling interval affects estimates of immersion activity in captive Magellanic penguins ( Spheniscus magellanicus ) and whether errors differ between group- and individual-level analyses. Using biologging-derived immersion records, we reconstructed immersion-state sequences and simulated scan sampling at intervals of 30 s and 1, 2, 3, 4, 5, and 30 min, accounting for scan-start phase. Estimation error increased nonlinearly with sampling interval but was smaller and less variable at the group level. Median group-level error remained below 0.7% at 5-min intervals and reached 3.6% at 30 min. In contrast, individual-level error increased sharply beyond 4–5 min and reached 8.5% at 30 min. Sampling interval dominated estimation error, with smaller detectable seasonal effects. Frequent land–water transitions further amplified errors at longer intervals. These results indicate that group-level aggregation buffers error but may also conceal individual-level heterogeneity. For group-level summaries, intervals up to 5 min may provide stable estimates, whereas individual-level monitoring requires finer resolution, approximately 2–3 min. More broadly, our findings highlight that sampling interval, scan phase, state switching, and aggregation level jointly determine the reliability of scan-sampling estimates.
SN - 2045-2322
ER -