Monitoring partially marked populations using camera and telemetry data

Area of Science:

• Wildlife ecology
• Conservation biology
• Statistical modeling

Background:

• Long-term wildlife monitoring is crucial for conservation but often hindered by costly and complex density estimation methods.
• Spatial mark-resight (SMR) models offer a more accessible approach by utilizing data from both marked and unmarked individuals.
• Existing SMR models face challenges with large datasets and long-term monitoring initiatives.

Purpose of the Study:

• To develop a generalized, two-stage spatial mark-resight (SMR) model adaptable for long-term camera data and auxiliary telemetry data.
• To improve spatiotemporal inference for wildlife monitoring by reducing computational demands.
• To provide a flexible framework for density estimation in long-term monitoring where auxiliary data may be limited.

Main Methods:

• A two-stage generalized spatial mark-resight (SMR) model was developed.
• Stage 1: Detection parameters estimated using telemetry data and camera detections of instrumented individuals.
• Stage 2: Density estimated using camera data, treating all individuals as unmarked, with time-series models accounting for serial correlation.

Main Results:

• The model was applied to 3 years of white-tailed deer data in Florida, integrating telemetry and camera trap data from 59 marked females.
• Seasonal fluctuations significantly influenced temporal density variation, with one area showing a slight population decline.
• The two-stage approach demonstrated lower computational demands compared to joint SMR models, facilitating practical application.

Conclusions:

• The developed two-stage SMR framework offers a computationally efficient and practical solution for large-scale, long-term wildlife density estimation.
• This model enhances the utility of modern technologies like camera traps for conservation by addressing analytical challenges posed by massive datasets.
• The model's flexibility allows for density estimation even without synchronous auxiliary data on marked individuals, a common scenario in long-term monitoring.