A Spatio-Temporally Explicit Random Encounter Model for Large-Scale Population Surveys

Area of Science:

• Ecology
• Wildlife Biology
• Statistical Modeling

Background:

• Random encounter models estimate population abundance using non-invasive sampling like track counts.
• The Formozov-Malyshev-Pereleshin (FMP) estimator provides population density from track counts but requires daily movement data.
• Existing methods may not fully capture spatio-temporal variations in animal populations.

Purpose of the Study:

• To extend the FMP estimator using a flexible Bayesian modeling approach.
• To estimate not only total population size but also spatio-temporal variation in population density.
• To introduce a weighting scheme for estimating density in unsurveyed habitats.

Main Methods:

• Utilized generalized linear models with spatio-temporal error structures.
• Developed a Bayesian modeling framework extending the FMP estimator.
• Incorporated a weighting scheme for habitat density estimation.
• Conducted a simulation study mimicking the Finnish winter track count survey.

Main Results:

• The spatio-temporal modeling approach effectively borrows information from neighboring locations and times.
• Spatio-temporal and temporal smoothing models yielded improved estimates of total population size compared to the FMP method.
• The new approach demonstrated enhanced accuracy in estimating population density and its variations.

Conclusions:

• The proposed Bayesian spatio-temporal models offer a more robust and accurate method for estimating wildlife population abundance and density from track count data.
• This approach enhances ecological inference by accounting for spatial and temporal dynamics.
• The methodology provides a valuable tool for wildlife management and conservation efforts.