A strategy for establishing the specificity and safety of an organism as a biological weed control agent is described. A critical first step is to expose to its attack a small group of plants very closely related and exhibiting morphological and biochemical similarities to the weed. To prevent an erroneous negative result tests are also made on selected cultivated plants, including those closely related to the weed, those of which the associated insects and fungi are little known, those that have evolved apart from or been little exposed to the agent, those attacked by closely related organisms and those already recorded as hosts.
The circumstances under which the strategy might fail to indicate safety are discussed, i.e. polyphagous organisms attacking plants irregularly distributed throughout many families, organisms highly specific to two alternate hosts, and those attacking two or three phylogenetically widely separated plant groups. The additional crop plant testing, included in the overall strategy to deal with such possible failures, is discussed.
It is shown that the strategy would have included Sesamum tndicum in the list of plants challenged by the bug Teleonemia scrupulosa in biological testing for control of Lantana camara, thereby forewarning of the attack that was subsequently observed in Africa.