WEEDS

Evaluating Risk to Native Plant Communities from Selected Exotic Plant Species

Maria Mantas

The Nature Conservancy of Montana

Cohesive Strategy Team

Flathead National Forest

February 4, 2003

This document provides a brief summary of the content of this analysis and its purpose.  It is a precursor to a GTR publication which is in prep.  For a complete and detailed description of the analytical procedures used to spatially display risk, view the metadata file for any species listed in the table below, or refer to Appendix H.  Most of the Tables and Appendices that provide the raw data for this analysis can be viewed using the links listed below.

ABSTRACT

A three-tiered approach was developed to access risk to native plant communities in Montana and northern Idaho from exotic plant species.  This approach involved determining the susceptibility of areas to species’ establishment, assigning level of threat to susceptible areas, and factoring in the probability of exposure of each site to plant propagules by considering the various aspects that affect dispersal.  Native plant communities were represented by various biophysical settings that were modeled as 34 potential natural vegetation (PNV) groups. Susceptibility, threat, and probability of exposure were then combined to spatially model degree of risk across the study area from some of the most threatening extant exotic species in Montana and northern Idaho. 

PURPOSE:

Invasive species have been recognized as being second only to land development in the loss of biodiversity.  Some exotic species are so fast to colonize and convert native vegetation that little can be done in time to stave off the invasion.  A review of the timing of action taken to address the threats from these species has shown that action often comes too late.  Many species are not recognized and placed on noxious weed lists until they have already caused irreparable harm.  To remedy this problem, researchers and managers have recently moved towards developing more proactive approaches, such as analyzing the risk of exotic species to the environment.  Evaluating risk to native plant communities from invasion by the most imminent and threatening of exotic plant species is important in identifying opportunities for action.  The products generated by this assessment will allow managers across Montana and Northern Idaho to first, be alerted to which of the many listed and non-listed exotic plant species were recognized as being the most threatening to native plant communities, and second, to identify which areas are at greatest risk.  This product will then allow them to prioritize which areas, species, or combination of the two are the most important for weed management actions.

The purpose was not to focus on specific resource values at risk, rather it was to evaluate risk to the overall integrity of native plant communities.  This was accomplished by considering potential changes to plant community structure, ecosystem function, and native species composition caused by the introduction of selected exotic species.

Thomas (1999) states that a risk assessment should be two tiered, evaluating the risk of introduction of potential invaders, and prioritization regarding species already present in the area.  In this assessment the concern was for the imminent threat of rapidly spreading extant exotics, so focus is shifted away from the invasiveness of particular species to the vulnerability of native plant communities from extant species.  Therefore, this study only addresses half of the problem, according to Thomas (1999).  It is equally important to evaluate the risk from new invaders, and recognize the need for such in the analysis area.

It is important to realize that the products created by the models will only predict the level of risk to native vegetation under different scenarios (disturbed/undisturbed) if and when infestation occurs.  This model will not predict where infestations currently exist. In fact, some areas identified as high risk may already be severely infested while others are in a pristine condition.  An inventory of exotic species (i.e., something finer than at the county level) would help to portray the current status.  Unfortunately, a continuous inventory for exotic species at the same scale as this model does not exist for the analysis area, and it is unlikely that an inventory of this detail would be produced in the near future.  Therefore, the user must realize that this model has the greatest value in alerting managers where there may be problems in areas that are not already infested.  It is recommended that the user overlay whatever existing inventories are available from local sources with these products to identify areas that are potentially at risk from invasion.

The risk assessment will allow users to identify those species that pose the greatest threat to native communities.  Furthermore, we will be able to display which areas are at greatest risk.  By themselves, these products will help land managers to prioritize habitats most at risk, where they should conduct weed inventories, where they should conduct monitoring studies, where weed treatments are likely to be most successful, and where prevention should be applied in risky areas.  However, the real power of these data will become apparent when they are integrated with other data themes within an integrated status, risk, and opportunity assessment.  Managers will be able to evaluate tradeoffs among various kinds of treatments as well as the locations of treatments.  For example, using mechanical treatments to restore forest integrity may have a high risk of weed invasion in some areas, but not others.  Similarly, the same forest integrity objectives might be achieved using prescribed fire, but the risk of weed invasion may be much lower.

Determining Disturbance, Susceptibility, Threat, Exposure, and Risk

Disturbance

This coverage was developed to identify disturbed sites in the study area.  Disturbance data must be combined with potential natural vegetation (PNV) data to identify which areas are susceptible to each exotic plant species analyzed.  Various types of ground disturbance were considered:  animals;  burned areas;  physical disturbance;  and grazed areas.  Data concerns included:  the lack of sufficient disturbance data on non-Forest Service lands;  insufficient data to differentiate burned areas as to their degree of disturbance based on fire impacts;  lack of detailed grazing allotment records.  As a result, disturbance is underestimated on non-Forest Service lands, all areas within fire perimeters were considered disturbed, and all acres meeting the grazing criteria were considered disturbed.

Susceptibility

Susceptibility refers to the vulnerability of a native plant community to colonization and establishment of an exotic species.  Data, literature sources, and expert opinion were used to determine if a species could become established in each potential natural vegetation type (PNV).  Expert opinion came from a panel of botanists and ecologists who were convened to review the findings from data and literature, and provide further input where needed.  Susceptibility was rated using a categorical system where each combination of a species and PNV was coded with one of the following: 

U – Unknown:  Susceptibility of this PNV to the species is unknown

C – Closed:  The species generally does not occur within this PNV under any condition

I – Invasive:  The species is invasive in undisturbed conditions within this PNV.  If a species was rated as “I”, the assumption is that it would also invade with disturbance.

D – Disturbance:  The species occurs in this PNV where there has been evidence of recent disturbance.

Susceptibility codes used for each species/PNV combination can be found in the risk assessment data matrices.  There are different matrices for east and west of the Continental Divide due to differences in reported vulnerability of PNVs in the two areas, and because some species were only analyzed on one side of the Continental Divide.  Data concerns are a result of the paucity of field data and research for assessing susceptible PNVs.  Much of the information used to assess susceptibility was derived from expert panel opinion.  There are gaps in the coverage where there was no information for particular species/PNV combinations.

Threat

Threat refers to the degree of change to the structure, composition, or function of a native community from an exotic species.  Threat is displayed using a qualitative ranking of three classes: low, high, and none.  Factors taken into consideration for classing threat were as follows:

L – Low Threat:  Species can become established; however, they cannot compete well with native vegetation, even in disturbed settings.  Species with low threat never increase substantially in cover without the aid of severe site disturbance.  Even in cases of moderate to mild disturbance events (e.g. low intensity fires, moderate grazing) native plants still are able to compete successfully. 

H – High Threat:  Species were rated as having high threat if once established they can compete successfully with native vegetation.  These changes would have to be significant enough to where the function of the plant community is substantially altered.  These changes would include alteration in natural pathways of succession, a change in the natural fire regime, and/or significant changes to the composition and canopy cover of native plant species.

N – No Threat:  A species can only be assigned no threat to a PNV if it is closed (C) to that PNV.

The data source for determining threat can be found in the threats data matrices.  As with susceptibility, threat was sometimes classed differently from east to west of the Continental Divide, therefore two separate threats matrices were developed.

U – Threat Unknown

Probability of Exposure

Probability of exposure (POE) was calculated by combining various factors that were considered to be influential on the probability that the particular pixel would be exposed to seeds of the species being evaluated.  Therefore, a POE grid was constructed for each species analyzed.  Factors used to construct POE Grid:

Distance to nearest county with known species occurrence.

Moving window road density classes.

Within specified distance to primary/secondary federal or state highway.

Within a recently active grazing allotment.

The four factors were added together, and based upon the final value, a probability of exposure rating of low, moderate, or high was assigned.

Risk

Data from the three input sources (susceptibility, threat, and exposure) were used in the following rule set to determine the level of risk to a site from each species. 

General rules were used to help determine breaks for level of risk for the various combinations.  These were:  1)  If a pixel is not susceptible to a species, then the risk is always none;  2)  If the threat is low for a species in a susceptible pixel, then the risk will always be low;  and 3)  If the threat is high for a species in a susceptible pixel, then the risk can never be low.  The final risk rating results pointed out that there weren’t enough data or knowledge among the expert panel members to add another threat level (i.e. moderate).  The fact that there was no moderate threat class resulted in very little of the final risk being coded as moderate.  Rather than force the analysis to evenly distribute risk among the four classes (none, low, moderate, high), we chose to use the general rules stated above to assure that results were more meaningful with regards to actual risk, than a distribution of risk based mathematically.

TWO SCENARIOS ANALYZED:

The maps and tables generated by using the previously described methods display the risk to the condition of the study area at the time of this analysis.  It was also useful to display the level of risk to PNVs in a “worst-case” scenario that is considering every site, or 100% of the study area, as disturbed. This scenario will be useful in evaluating the risk of future vegetation treatments or other disturbances.  Each species has two final grids:  1) one that evaluates risk at the time of this analysis;  and 2) one that evaluates risk in a “worst-case” scenario.

GENERAL LIMITATIONS:

These data were designed to characterize broad-scale patterns for regional and subregional assessments.  Any decisions based on these data should be supported with field verification, especially at scales finer than 1:100,000.  Although the resolution of the layer is at 90 meter cell, the expected accuracy does not warrant their use for analyses of areas smaller than about 10,000 acres (for example, assessments that typically require 1:24,000 data).

List of Tables

Table 1. Exotic Species Analyzed

Table 2. Pasture grasses that have aggressive or competitive tendencies in native plant communities

Table 3. Susceptibility Code Definitions

Table 4. Possible risk rating outcomes for each site

List of Appendices

Appendix A: Expert Panelists and Project Reviewers

Appendix B. Species Considered in this Analysis

Appendix C. Species Considered but not Analyzed

Appendix D. Potential Natural Vegetation Types

Appendix E. East Side Susceptibility/Threat Matrices

Appendix F. West Side Susceptibility/Threat Matrices

Appendix H. Procedural outline for GIS analysis to determine risk to PNVs

Appendix I. Literature Sources for Species

Choose from species code or common name:

Powerpoint Presentation of Weed Risk Assessment Process

* if you do not have Powerpoint use this html version

For copies of the GIS coverages (grid data) contact:

Maria Mantas
The Nature Conservancy of Montana
mmantas@tnc.org
406-862-6494

Weeds
Description (click on name for lit_abstracts)	Image
Canada thistle
cheatgrass
common crupina
common speedwell, Paul's betony
common St. John's-wort
common tansy
common toadflax, yellow toadflax
Dalmatian toadflax
diffuse knapweed
dyer's woad
field scabiosa
hoary cress
hound's-tounge
Japanese brome
Japanese knotweed
leafy spurge
meadow hawkweed
orange hawkweed
quackgrass
rush skeletonweed
Russian knapweed
Russian olive, oleaster
soft brome
spotted knapweed
sulphur cinquefoil
white bryony
yellow starthistle
yellow sweetclover