ble for several of the programs, but there are other parameters that are equally valid measures of the success of biocontrol programs. For example, knapweed increases surface runoff and stream sedimentation ( Lacey et al. 1989), and there is a suggestion that it reduces ponderosa pine regeneration. Both of these could have been used as project objectives. A concern with leafy spurge is that it is displacing the western prairie fringed orchid (Plantanthera praeclara), of which there are about 2,000 plants in North Dakota ( Federal Register 1989). Hence the objective of the U.S. Forest Service group responsible for managing the main orchid populations should be to maintain or increase it. The Canadian Department of National Defense is concerned with spurge at Shilo, Manitoba, because it does not bind the sandy soil, so the area is subject to wind erosion after tank exercises. Biocontrol success would be an increase in the root mat near the surface as grass replaces spurge. Military authorities are also concerned with the maintenance of native plant communities and being a good neighbor to surrounding farmers, both of which have to be measured in other terms. One of the government of Saskatchewan's goals for the leafy spurge biocontrol program is the elimination of an annual $150,000 subsidy for its chemical control; the results achieved are sufficiently encouraging that this subsidy has already been withdrawn, so this goal has been met. The point is that each of these objectives needs to be measured in terms other than weed reduction, and many weed biocontrol projects are not taking this final step.
The impact of biocontrol agents--such as degree of defoliation or reduction in weed population vigor--on their target weed is not linearly related to the benefits of weed biocontrol, such as forage yield. Indeed, the benefits are usually not studied because they are not established because project goals.
Monitoring weed biocontrol projects is difficult because many parameters must be followed and the effects do not necessarily occur where and when they are expected. The system suggested is to divide the project into the following steps, each with its own goal, to be investigated in sequence: agent establishment, intensity of agent attack, impact of the agent on the weed, and project benefits. Many projects are terminated with an assessment of the agent impact on the weed, which is done by clipping or sorting vegetation inside and outside the agent area. However, satisfactory results can usually be obtained with much less effort with Daubenmire's approach for assessing plant species cover. This is not the end of a project, since the benefits of reducing weed cover have to be measured in other terms, such as forage yield, reduced erosion, reduced herbicide use, and the protection of rare species. These are the benefits that justify starting a biocontrol project, and they vary with each weed species.
I gratefully acknowledge the assistance of Alan Sturko of the British Columbia Forest Service for collecting samples and providing data from Daubenmire transects,