Estimating animal abundance at multiple scales by spatially explicit capture-recapture

Area of Science:

• Wildlife ecology and population estimation
• Spatial statistics and survey design

Background:

• Accurate animal abundance data is crucial for wildlife management but traditional surveys are costly and time-consuming.
• Existing statistical methods for spatially replicated surveys often rely on assumptions of uniform detectability and density, which can be violated in real-world scenarios.

Purpose of the Study:

• To quantify the bias introduced by unmodeled spatial heterogeneity in detectability and density.
• To evaluate novel, design-based estimators for average animal density across replicate study areas.
• To assess the performance of different variance estimators for average density.

Main Methods:

• Utilized simulation studies and empirical data from over 3500 black bears across 73 study areas in Ontario, Canada.
• Employed spatially replicated capture-recapture (SECR) surveys, treating each study area as an independent replicate.
• Compared a design-based estimator with traditional methods assuming homogeneous spatial processes.

Main Results:

• Assuming spatially constant detectability led to negative bias (20-30%) in density estimates.
• The design-based estimator, treating study areas as independent replicates, yielded unbiased density estimates at both local and landscape scales.
• This independent replicate approach maximized precision (7-18% relative SE) and computational efficiency for black bear density estimation.

Conclusions:

• Treating spatially replicated SECR surveys as independent replicates effectively avoids biases from spatial heterogeneity in detectability and density.
• This methodology provides efficient and precise estimates of animal density at multiple scales, applicable to various ecological questions and data types.
• The findings support flexible survey designs that accommodate spatial variation, enhancing the reliability of wildlife management decisions.