Assessing Probabilistic Risk Assessment Approaches for Insect Biological Control Introductions

Area of Science:

• Ecology
• Entomology
• Biological Control

Background:

• Introduction of biological control agents necessitates host specificity testing to assess non-target impacts.
• Current methods rely on conservative physiological host ranges under captive conditions, ignoring ecological factors influencing realized host range.

Purpose of the Study:

• To validate a probabilistic risk assessment (PRA) procedure for non-target impacts using historical and current field data.
• To assess the predictive power of PRA by comparing predictions with actual non-target parasitism levels of introduced parasitoids in Hawaii.

Main Methods:

• Utilized historical and current field data from introduced parasitoids targeting an endemic Lepidoptera species in Hawaii.
• Employed data on the known host range and habitat use in the parasitoids' native regions.
• Compared predictions from PRA using different data types (apparent mortality vs. marginal attack rates) and incorporation of ecological data.

Main Results:

• Comprehensive data from the agents' native regions enabled reasonable predictions of potential non-target impacts.
• Scant data from native regions led to poor predictive accuracy.
• Using apparent mortality data in PRA overestimated non-target impacts compared to marginal attack rate estimates.
• Incorporating ecological data into PRA models significantly improved predictive power.

Conclusions:

• PRA can be a valuable tool for predicting non-target impacts of biological control agents, provided comprehensive data are available.
• Ecological data integration is crucial for enhancing the accuracy and reliability of risk assessments.
• Future risk assessments should prioritize collecting detailed ecological information from the agents' native habitats.