Insect and Pathogen Risk and Hazard Rating Systems for Use in Forest Threat Assessments
Eric L. Smith and Andrew J. McMahan
The basis of assessing and predicting threats to forest health lies in being able to relate site and forest conditions to the likelihood and intensity of disruption by organisms and events. One major class of disruption agents are insects and pathogens: “pests” when and where the disruption decreases socially desired forest resource benefits. Since site and stand conditions have long been observed to influence the likelihood and intensity of impacts of a wide range of pests, a number of rating systems, simulation models, and related studied have been developed to assist managers in evaluating conditions, prioritizing treatment areas, and selecting treatments to be applied.
These models, hazard and risk rating systems and others, represent a quantification of a significant portion of the academic and applied knowledge of the nation’s forest pests, and pest risk response to management. These products are potentially useful tools in broad scale threat assessments. They are based on empirical data, knowledge of biological relationships, or constructed by experts with significant experience and insight. Many have been improved through testing or use; and many have some level of acceptance by natural resource professionals. The knowledge represented by these studies and tools represent an investment in time and resources, so if existing studies are not used it is not likely results from new studies would be available soon.
On the other hand, there are statistical, decision analytic, and other quantitative issues to consider when considering the use of these tools for broad scale assessments. These systems have usually been developed and calibrated for a limited geographical region or range of forest conditions. The systems may have been constructed as decision tools with imbedded management assumptions, not appropriate to current conditions. The analysis used in their construction may have been flawed or inappropriate; at least the goodness of fit of statistically-based models needs to be considered. In any case, the original data and analytical details of their construction may no longer be available to verify or modify the analysis. To apply the systems in broad scale assessments, the data for the model variables need to be available and of sufficient quality. Issues regarding the spatial scale of the original analysis relative to that of broad scale assessments should also be considered.
We have identified almost 200 published North American forest pest hazard and risk systems, and related studies and models, which could be considered for use in threat assessments. In this paper, we classify the relevant features of these systems which are needed to evaluate their potential usefulness in broad scale assessments. These features include scope of the original system (pest and host species, geographic or ecological range of application), original purpose (descriptive statistics of a sample, treatment priority ranking, marking or thinning guide, and others); method of analysis or model construction (regression-type analyses, scoring systems, expert opinion or multi-criterion approaches, complex computer simulation models, others); and other relevant factors. The systems and their features will be catalogued in a relational database and summary tables will be presented in this paper. The scope and applicability of existing published systems for specific pests will be compared to the recent and projected activity of these pests.
Additional issues concerning the application of single pest rating systems in ecosystem assessments will be addressed. Many systems provide an ordinal classification of stand hazard (high, medium, low, for example), or an index system which provides an ordinal scoring of conditions. Such systems for individual pests are not easily integrated into an assessment of multiple pests. Where systems produce impact outputs in absolute terms (BA mortality per area), a difficulty arises in representing positive interactions between multiple pest organisms, and consideration of the additivity of the impacts, so that the same tree is not projected as being killed more than once. Although many would consider empirical probability based models (for tree mortality, for example) to be superior to ordinal classification systems, it may be inappropriate to apply a probability model developed at one time and place to a different time and place.
Monitoring Methods Session - Tuesday Afternoon
corresponding author:
Eric L. Smith
USDA Forest Service
Forest Health Technology Enterprise Team
2150A Centre Avenue
Fort Collins, CO 80526-8121
970-295-5841
elsmith@fs.fed.us
Encyclopedia ID: p127
