A flexible framework for spatial capture-recapture with unknown identities.

Area of Science:

• Ecology
• Wildlife Population Dynamics
• Bioacoustics

Background:

• Camera traps and acoustic recorders are standard tools for wildlife population sampling.
• Spatial capture-recapture (SCR) methods require individual animal identification, which is often labor-intensive and not always feasible.
• Existing methods struggle with datasets where individual identification is incomplete.

Purpose of the Study:

• To develop a generalized spatial capture-recapture (SCR) framework that accommodates imperfect individual identification.
• To integrate acoustic data into SCR models, enabling density estimation from vocalizations.
• To provide a unified approach for analyzing data from camera traps and acoustic recorders.

Main Methods:

• Formulated SCR as a marked Poisson process with a single counting process for all detections.
• Defined a flexible mark distribution to incorporate various observed characteristics (e.g., identity, location, sex, time).
• Applied the generalized SCR model to camera trap data (fisher) and acoustic data (Cape Peninsula moss frog), validated through simulation.

Main Results:

• The proposed method successfully estimates animal density from both camera trap and acoustic data.
• Latent identity SCR models incorporating additional marks (sex, time) proved reliable for density estimation.
• The framework integrates acoustic SCR and generalizes existing latent identity SCR models.

Conclusions:

• The generalized SCR framework offers a robust solution for wildlife density estimation with imperfect identification.
• This integrated approach enhances the utility of camera trap and acoustic monitoring data.
• The method provides a reliable tool for ecological research and wildlife management.