Assessing Whether Habitat Suitability Models Can Predict Abundance of an Insect Pest

Area of Science:

• Ecology
• Pest Management
• Ecological Modeling

Background:

• Predicting pest abundance across landscapes is crucial for effective pest management.
• Habitat suitability models (HSMs) are used for forecasting pest abundance but struggle with heteroscedastic distributions.
• Understanding how HSMs handle wedge-shaped relationships between pest abundance and habitat suitability is key.

Purpose of the Study:

• To evaluate the effectiveness of six habitat suitability models in predicting the abundance of the blueberry maggot (Rhagoletis mendax).
• To assess model performance in capturing heteroscedastic pest distributions and wedge-shaped abundance patterns.
• To compare model interpolation and transferability for pest management applications.

Main Methods:

• Utilized satellite-derived environmental data to construct six distinct habitat suitability models.
• Collected 4-year field survey data (2009-2012) on Rhagoletis mendax abundance in New Jersey, USA.
• Employed quantile regressions to model habitat suitability's prediction of various abundance quantiles, not just the mean.

Main Results:

• Only BioClim and random forest models identified a positive wedge-shaped relationship between pest abundance and habitat suitability.
• Four models showed a negative relationship, indicating potential limitations in capturing pest distribution.
• Random forest excelled in quantile regressions but lacked transferability; BioClim showed poor interpolation but good transferability.

Conclusions:

• Habitat suitability models must account for pest abundance gradients within suitable habitats to accurately predict pest density.
• Different modeling approaches may be required for effective spatial interpolation versus transferability.
• The choice of HSM impacts the ability to predict heteroscedastic pest distributions and inform management strategies.