Simplified large African carnivore density estimators from track indices

Area of Science:

• Ecology
• Conservation Biology
• Statistical Modeling

Background:

• Assessing large carnivore populations is crucial for global conservation.
• Track-based surveys on suitable substrates can estimate carnivore population size and trends.
• Conventional linear models may not accurately represent zero-origin relationships in ecological data.

Purpose of the Study:

• To evaluate the appropriateness of linear regression through the origin versus linear regression with intercept for modeling large African carnivore densities and track indices.
• To determine if a general model can estimate carnivore densities across species and study areas.

Main Methods:

• Simple linear regression with intercept and simple linear regression through the origin were performed.
• Model fit was evaluated using confidence intervals for the regression coefficient (ß), Standard Error of Estimate, Mean Squares Residual, and Akaike Information Criteria.

Main Results:

• Linear regression through the origin provided a better fit for all tested models, indicated by lower Standard Error of Estimate and Mean Square Residuals.
• Akaike Information Criteria supported linear models through the origin, with no substantial support for models including an intercept.
• Some models with intercept were not significant, and their confidence intervals included zero.

Conclusions:

• Linear regression through the origin is justified for modeling large African carnivore densities from track data.
• A general formula, observed track density = 3.26 × carnivore density, can estimate densities on sandy substrates for carnivore densities ≥ 0.27/100 km².
• Further validation with independent data is needed to refine models, especially for low-density non-linear relationships.