The decision to release a potential weed biological control agent into the field is based on judgements about risk of damage to non-target plants of economic or conservation value. These risks should be balanced against the possible benefits of biological control of the target weed.

The risk of damage to non-target plants is assessed from literature searches, and field observations in the country of origin or other regions in which the agent has been released, and from host specificity tests. The predicted host range of potential biological control agents varies from high specificity, so that only the target weed will be damaged in the field, to broad specificity, so that several non-target plants of economic or conservation value are likely to be seriously affected if the agent was released in the field. Decision making at these extremes is straightforward. If there is no risk to non-target plants then the agent should be released. On the other hand, if several valuable non-target plants are likely to be seriously damaged, the potential agent should not be released.

The larva of Neurostrota gunniella a control
agent for Mimosa pigra.

Between the extremes of possible outcomes of host specificity testing are cases of potential agents which may cause some damage in the field to one or a limited range of valuable non-target plants. In these cases a risk analysis approach balancing the costs (i.e. risk of damage to non-target plants) and benefits (i.e. possibility of control of the target weed) of release of the potential agent is necessary in making the decision to release the agent in the field. Political and social costs need to be considered, and may sometimes outweigh the economic costs.

In this section three case studies from biological control programs in Australia are presented to illustrate this approach:

This is followed by a brief discussion of the decision making used in these cases and what can be learnt from it.