Honey Bee Overwintering Survival in Pennsylvania Predicted by Weather and Land Use

Martina Calovi from Pennsylvania State University used a three-year survey of Pennsylvania beekeepers to assess drivers of overwintering mortality in managed honey bee colonies. A Random Forest model predicted survival with 73.3% accuracy for colonies and 65.7% for apiaries, with growing degree days being the most important predictor. The study produced a composite map of long-term survival probability for Pennsylvania from 1981 to 2019.

Use Cases

• Predict colony survival probability based on weather variables like growing degree days.
• Model the impact of surrounding land use on honey bee mortality.
• Assess the relative importance of management factors versus environmental drivers on colony failure.
• Create spatial risk maps for beekeeping under current or projected climatic conditions.

Strengths

• Model achieved 73.3% accuracy for colony-level predictions via cross-validation.
• Analysis integrates three years of survey data across Pennsylvania.
• Long-term probability map covers a 39-year period (1981-2019).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Three years of data may not capture the full range of climatic conditions.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Pennsylvania State University

Collection Method

Three-year survey of Pennsylvania beekeepers.

Time Range

Survey covered three years; long-term map covers 1981-2019.

Geography

Pennsylvania, USA