A flexible and efficient Bayesian implementation of point process models for spatial capture-recapture data.

Area of Science:

• Ecology and Wildlife Biology
• Statistical Modeling
• Computational Biology

Background:

• Spatial capture-recapture (SCR) is a standard method for wildlife population estimation.
• Current Bayesian SCR models lack accessible and efficient fitting tools for continuous spatial processes.
• Existing methods often rely on data augmentation, which can be computationally intensive.

Purpose of the Study:

• To develop and present custom Bayesian functions and distributions for fitting spatial capture-recapture models.
• To offer efficient model fitting with spatial covariates and utilize the semi-complete data likelihood (SCDL) approach.
• To provide a more accurate reflection of spatially continuous detection processes in area-search SCR studies.

Main Methods:

• Developed a Bayesian framework with custom functions and distributions for SCR models.
• Implemented the semi-complete data likelihood (SCDL) approach as an alternative to data augmentation.
• Tested the model formulation through simulations and quantified computational efficiency gains.

Main Results:

• The new Bayesian tools allow for more efficient SCR model fitting, especially with spatial covariates.
• The SCDL approach is computationally more efficient than data augmentation for simpler SCR models.
• The model effectively incorporates spatially continuous detection processes and spatial variation in density.

Conclusions:

• The developed Bayesian SCR framework provides accessible and efficient tools for wildlife population estimation.
• The SCDL approach offers computational advantages for specific SCR model complexities.
• The methodology was validated with simulations and a real-world application on wolverine (Gulo gulo) populations.