How good are we at predicting the field host-range of fungal pathogens used for classical biological control of weeds?

After release, a classical biocontrol agent is expected to become a permanent part of its new environment and it is vital to ensure, as far as possible, that it will do more good than harm there. This is done through risk assessment, which invariably includes host-range testing. The use of pathogenic fungi as classical biological control agents for weeds began in 1971. To date, 26 species of fungi, originating from 15 different countries, have been used as classical biocontrol agents against over 26 species of weeds in seven countries. Information is presented on the performance of these pathogens in the field after release, compared with predictions made on the basis of prerelease host-range testing. The mechanisms and evolution of plant resistance are discussed in relation to risk assessment. No reports were uncovered of deliberately introduced fungi unexpectedly attacking nontarget plants after release. Indeed, host-range testing results have often proved conservative, with a number of examples of pathogens attacking nontarget plants in prerelease tests, but not being recovered from these species in the field. It is concluded that risk assessments based on rigorous host-range testing, combined with a good understanding of the taxonomy, biology, and ecology of the agent, the target weed, and nontarget species, can ensure that the introduction of exotic pathogens is a safe and environmentally benign method of weed control. (C) 2004 Elsevier Inc. All rights reserved.