Explaining detection heterogeneity with finite mixture and non-Euclidean movement in spatially explicit...

Area of Science:

• Ecology
• Wildlife Biology
• Conservation Science

Background:

• Landscape structure influences animal movement patterns and home range sizes.
• Heterogeneity in movement can bias population size and density estimations.
• Traditional methods for accounting for heterogeneity, like finite mixture models, are data-intensive and can yield unreliable estimates.

Purpose of the Study:

• To evaluate the effectiveness of novel non-Euclidean spatially explicit capture-recapture (SCR) models in improving black bear (Ursus americanus) density inference.
• To compare the performance of non-Euclidean SCR models against standard SCR and finite mixture models.

Main Methods:

• Fit standard and non-Euclidean SCR models to black bear capture-recapture data from 51 study areas.
• Incorporated landscape connectivity using non-Euclidean least-cost paths in advanced SCR models.
• Assessed model support and sensitivity to spatial scale (pixel resolution).

Main Results:

• Non-Euclidean models were supported in over half of the study areas, particularly in structured landscapes.
• Density estimates from non-Euclidean models were higher and less precise than simple models.
• Finite mixture models, despite large sample sizes, yielded potentially unreliable abundance estimates.

Conclusions:

• Ignoring landscape heterogeneity can lead to severe negative bias in density estimates.
• Non-Euclidean SCR models offer improved inference in structured landscapes but require careful application.
• Finite mixture models should be used cautiously, especially with limited data, due to potential unreliability.