Back to Index

An Ecologically Based Decision Support System for Managing Leafy Spurge (Euphorbia esula) Infested Rangeland

Roger Sheley

Question 1: What major problem or issue is being resolved and how are you resolving it?

Although integrated approaches can have a synergistic effect on weeds, most management tactics are not effective when used in combination (Lym 1992). Rangeland managers need useful models for evaluating alternative integrated strategies (Archer 1989, Schatterer 1989, Laylock 1991). It is imperative that these models are capable of predicting the response of plant communities to management tactics such as herbicides, biological control, grazing, revegetation and their integration. These models should help land managers determine if proposed management strategies are ecologically sound and cost-effective. Management decisions must be based on a predicted outcome, and these predictions must be based on our scientific understanding of the biology and ecology of the plants in the community (Luken 1990, Sheley et al. 1996).

Life-history models have been used to predict the dynamics of weed populations. These models consist of a series of states (growth stages) and a series of transitions that represents the flow of individuals from one growth stage to the next. Each state represents the number of individuals of a particular growth stage that are present per unit area. The transitions have values that correspond to the proportion of individuals that advance from one state to the next within a generation. The state value is multiplied by the transition value to yield the number of individuals that advance to the next state. A single species life-history model was used to demonstrate the response of leafy spurge populations to picloram (Bowes and Thomas 1978). A second single species life-history model was used to simulate development of leafy spurge populations from the introduction of propagules and the response of leafy spurge populations to the reduction of particular transition values (Watson 1985). The transition values were constant in these earlier models. Density dependent factors were not considered. A later model incorporated intra-specific density dependent transition functions (Maxwell et al. 1988).

Sensitivity analysis was run on this model to identify vulnerable areas in the life cycle of leafy spurge. More recently, Sheley and Larson (1994) developed a two species life-history model that was used to compare the population dynamics of two winter annuals, cheatgrass (Bromus tectorum) and yellow starthistle (Centaurea solstitialis). This information was developed into a preliminary computerized tool to assess integrated weed management.
These models have centered on controlling the weed with minimal regard for the original and resulting plant community (Sheley et al. 1996). Management practices based on these models have resulted in re-invasion by the same or different weed species because weeds rapidly occupy the niches opened by control procedures. Integrated approaches that favor desired species over weeds are necessary for the successful management of rangeland weeds, therefore multi-species models must be developed (Andrascik 1994). Although existing models have provided a method for understanding and predicting single-species population dynamics, they have not been useful in developing decision support systems. On rangeland, a useful decision support system must be capable of predicting the responses of interactions among desired species and weeds (Sheley et al. 1996). The life-history parameters of multi-species systems are dependent upon intra- and inter-specific interactions (Pyke 1991). Once multi-species models are developed, the effects of management strategies on the weed and desired species can be predicted.
Our overall objective is to develop an ecologically based decision support system for leafy spurge infested rangeland. The decision aid will be built using a multi-species life history model. The decision support system will be user friendly, and it will allow land managers to base management strategies on predicted outcomes.

Question 2: How serious is the problem? Why does it matter?

Various rangeland weed control techniques have been developed and used with unpredictable outcomes (Hansen 1993). Herbicidal control of most weeds requires continuous and periodic applications. For example, the most effective herbicide for leafy spurge control, picloram (4-amino-3, 5, 6-trichloropicolinic acid) provides only temporary reductions in plant density (Lym and Messersmith 1985, Lym and Messersmith 1994). Performance of biological control organisms is highly dependent upon the specific weed, quantities of agents released and site conditions (Rees and Spencer 1991). Biological agents may significantly impact populations during some growing seasons, while having little or no effect in other years. Research suggests that sheep and goat grazing can be used effectively to control some weeds, but outcomes vary depending on specific grazing management strategies (Lym et al. 1994). More recently, revegetation of infested rangeland has shown some potential for successful weed management, but environmental and economic considerations associated with this intensive system require its thoughtful use (Lym and Tober 1997, Ferrel et al. 1998, Sheley et al. 1999). Although single weed control methods are rarely effective, developing integrated weed management strategies may provide effective, long-term solutions to noxious weed problems (Andrascik 1994).

Question 3: How does it relate to the National Program(s) and National Program Component(s) to which it has been assigned?

The TEAM Leafy Spurge project is a part of the USDA/ARS Area-Wide Management Program. It is a component of Crop and Commodity Pest Biology, Control and Quarantine (304). TEAM Leafy Spurge complements efforts to develop new and improved pest control technologies and assess component technologies for integrated pest management (IPM) systems.

Question 4: What were the most significant accomplishments this past year?

Our most significant accomplishment was the establishment of research plots that are explained in the next section. Information from these plots will be used to accurately parameterize the decision support system.

Question 5: Describe the major accomplishments over the life of the project, including their predicted or actual impact.

A conceptual three-species life-history model has been developed that will serve as a generalized system for understanding multi-species relationships on rangeland. The three species in the life-history model are Kentucky bluegrass (Poa pratensis), western wheatgrass (Agropyron smithii) and leafy spurge. These grasses are often found growing in association with leafy spurge and offer an excellent opportunity to study multi-species interactions (Nowierski and Harvey 1988). The leafy spurge states in the model include: number of seeds in the seed bank, number of seedlings, number of vegetative and crown shoots and number of flowering shoots. Due to the difficulty associated with monitoring individual grass shoots, only the volume and forage production of the grasses is included in the life-history model. The grass states include initial volume, volume at peak biomass and forage production within a generation. Estimates of grass volume can be obtained by measuring height and canopy area. Volume was previously used to quantify interference among weeds and corn (Zea mays) (Bussler et al. 1995). Life-history models have been incorporated into computer programs that are capable of conducting the mathematical computations necessary for tracking populations and plant communities through time (Watson 1985, Maxwell and Sheley 1997). Our life-history model has been incorporated into a preliminary computer program written in QuickBASIC version 4.5. The program contains a series of formulas that predict densities of leafy spurge shoot types as well as Kentucky bluegrass and western wheatgrass volume over multiple generations. Simulations can be run with and without the incorporation of management procedures. The values that are used in these formulas are approximations from the literature. The actual values will need to be experimentally determined. The conceptual model continues to evolve as we gain more information about the competitive relationships among species. The computer program that will run the decision support system will be very complicated. A preliminary program was written prior to the current year. This program has improved during the current year, and it will continue to develop throughout the project. During the current year, output files that explain the results of simulations have been improved, and input procedures that are being developed will make the program user-friendlier. The equations that are used to run the life history model require the measurement of growth parameters in the absence of competition. To determine the growth rates and fecundity of isolated individuals, 20 seeds of either Kentucky bluegrass or western wheatgrass were seeded in 1 m2 plots and allowed to grow for 30 days in the summer 1998. Plants were thinned to a single isolated individual at that time. Leafy spurge root crowns were harvested from nearby infestations and introduced into the plots in the spring of 1999. Leafy spurge was planted after the grasses were well established, therefore all species should be equally mature. This should minimize the potential for artificially favoring one species over another because of their relative timing of establishment. Fifteen individuals of each species will be grown for three years. Plots will be arranged as a completely random design and replicated 15 times (3 species, 15 replications). In the summer of 1999 access tubes (3.8 cm diameter pvc pipe) were sunk in the center of each plot to a depth of 1.52 m. Neutron probes will be inserted into access tubes to measure soil water content during the growing season of 1999 and 2000. Soil water content information will be used to understand the influence of moisture availability on competition among the three species. Mixtures of leafy spurge, Kentucky bluegrass and western wheatgrass will be arranged in field plots. Kentucky bluegrass and western wheatgrass were established from seeds in 1m2 plots in 1998. Target densities at establishment were 0, 156, 1,250 and 10,000 plants/m2 for Kentucky bluegrass and 0, 156, 312, 1,250, 5,000 and 10,000 plants/m2 for western wheatgrass (Figure 3). These densities were used to ensure representation throughout a wide regression plane. Strips (0.91-m) of Kentucky bluegrass sod were planted around each plot to prevent leafy spurge from growing into adjacent plots. Plots were watered to ensure seedling establishment. In the summer of 1999 access tubes (3.8 cm diameter pvc pipe) were sunk in the center of each plot to a depth of 1.52 m. Neutron probes will be inserted into access tubes to measure soil water content during the growing season of 1999 and 2000. Soil water content information will be used to understand the influence of moisture availability on competition among the three species.

Question 6: What do you expect to accomplish, year by year, over the next 3 years?

The computer program will be fully developed by the end of the project. Ultimately, the information that is input by the user will be entered using a question/answer format. Simulation output will include predicted plant community and cost/benefit information. Output will be displayed in a clear, easily understood format.
The values necessary for developing a model include numerous life-history parameters and species competitive relationships. For example, the number of leafy spurge seeds that a reproductive shoot produces is dependent on the abundance and competitive relationships of various components of the plant community. This prediction of the number of leafy spurge seeds per shoot can be mathematically described as (Firbank and Watkinson 1985):

SPS = SPSmax [1 + amax(FS + _1VS + _2VB + _3VW)]b

were SPS is the mean number of seeds produced by an individual leafy spurge shoot. SPSmax is the number of seeds that a leafy spurge shoot produces when it is not influenced by competition. amax represents the maximum area of leafy spurge plants that are not influenced by competition. FS and VS are the final number of leafy spurge flowering and vegetative shoots/m2, respectively, in the current generation. VB and VW are the final volumes of Kentucky bluegrass and western wheatgrass, respectively, in the current generation. _1, _2, and _3 are the competition coefficients for the effects of vegetative shoots of leafy spurge, Kentucky bluegrass volume, and western wheatgrass volume on SPS, respectively. b describes the efficiency of resource use by the community. The life-history parameters that are used in the computerized model described in Objective 1 are approximations from the literature. These parameters need to be experimentally determined before this model can be developed into a decision support system. The following experiments will be used to estimate these parameters and assess their accuracy across sites and years. The volume of isolated Kentucky bluegrass and western wheatgrass plants will be determined each month during the growing season of 2000 and 2001. The volume, shoot number and shoot type (seedling, vegetative shoot from seedling, reproductive shoot from seedling, crown shoot, reproductive shoot from crown, root shoot, reproductive shoot from root) of leafy spurge will also be determined. These measurements will be used to estimate growth rates in the absence of competition. The number of seeds per shoot of leafy spurge will be counted in order to estimate maximum seed production. Absolute and relative growth rates will be calculated by fitting data for each species into either linear or curvilinear regression models. Models will be chosen based on sums of squares, residual error and the coefficient of determination (R2). The independent variable will be time. All other variables will be dependent. Growth rates of the species will be compared using the extra sums of squares procedure (Snedecor and Cochran 1980). The maximum area of isolated individuals and the maximum seed production of leafy spurge will be incorporated into the model equations. Field plots will be sampled during the spring and fall of 2000 and 2001. Fall sampling of leafy spurge will occur when seed production is complete. Fall sampling of grasses will be conducted at peak standing crop. The spring sampling will be conducted when leafy spurge shoot and seedling emergence is complete. Hand-drawn maps will be made that show the position and shoot type (seedling, crown shoot, root shoot) of leafy spurge shoots during the spring sampling periods. The volume of Kentucky bluegrass and western wheatgrass will be determined. A digital camera and computer software may be used for this data collection process. The maps will be updated during the fall sampling periods to reflect the growth stage of leafy spurge individuals. The volume of Kentucky bluegrass, western wheatgrass and leafy spurge and the leafy spurge shoot type (seedling, vegetative shoot from seedling, reproductive shoot from seedling, crown shoot, reproductive shoot from crown, root shoot, reproductive shoot from root, dead shoot) will be recorded. Because seedpods of leafy spurge can propel seeds several feet from the plant, seedpods will be harvested prior to seed rain to prevent seeds from entering adjacent plots. The number of seedpods per shoot and seeds per pod will be determined from ten randomly selected shoots and ten randomly selected seedpods, respectively, in each plot. The number of seeds collected from plots will be dispersed in each plot in the fall of 2000. One 5-cm diameter soil core will be randomly taken from each plot in the spring and fall of 2000 and 2001, and the number of viable seeds per plot will be determined using a standard tetrazolium test. All aboveground plant material will be harvested from the addition series plots after sampling in 2001 to determine the relationship between grass volume and biomass. Plant material will be separated by species and dried, and the dry weight per plot of each species will be determined. The relationship between volume and biomass will be quantified using regression. The data from these experiments will be used to generate parameter estimates for the equations in the life history model. The parameter estimates will be used in the computer program for simulating plant community attributes. Data collected from naturally occurring plant communities will be used to determine the accuracy of predictions made with the multi-species life-history model and to assess the magnitude of site to site and year to year variation. Preliminary search for field sites was conducted. Three sites codominated by leafy spurge, Kentucky bluegrass and western wheatgrass were located throughout Montana. Fences were erected around these sites to exclude grazing. Twenty-nine permanent plots (1 m2) were established at each site. Twenty plots contain varying densities of leafy spurge, Kentucky bluegrass and western wheatgrass. Nine plots contain isolated individuals of the three species (3 plots per species). Glyphosate was used to remove surrounding vegetation from plots containing isolated plants. In the summer of 1999 access tubes (3.8 cm diameter pvc pipe) were sunk in the center of each plot to a depth of 1.52 m. Neutron probes will be inserted into access tubes to measure soil water content during the growing season of 1999 and 2000. Soil water content will be used to understand the influence of moisture availability on competition among the three species. Data collection will be the same for the competition and validation experiment. Data from natural communities will be compared to each other and to the competition experiment. Comparisons of data from natural communities will be used to estimate the magnitude of site to site variation. Comparisons of data from natural communities to data from the competition experiment will be used to estimate how well the competition experiment mimics naturally occurring communities.

Question 7: What science and/or technologies have been transferred and to whom? When is the science and/or technology likely to become available to the end user (industry, farmer, other scientists)? What are the constraints, if known, to the adoption and durability of the technology products?

Because the decision support system is currently being developed, it has only been used as a teaching tool. The decision support system will continue to be developed and improved for the next several years, but the first version will be available near the end of 2001.
The decision support system will not be useful in some habitat types and will only be equipped to make temporal predictions by the end of 2001. Over the next several years additional habitat types will be added to the decision support system, and further development will allow for spatial predictions of leafy spurge infestations.

Question 8: List your most important publications in the popular press and presentations to non-scientific organizations and articles written about your work.

Our TEAM-Leafy Spurge work was presented at the 1999 Leafy Spurge Symposium in Medora, ND, in Roger Sheley’s rangeland weed ecology class during a lecture on life history models, to Matt Rinella’s (graduate student) graduate committee, at a Montana State University LRES departmental seminar and at various Montana State University weed tours. A description of the model will be used in a book chapter that is currently being written describing techniques for modeling biological control efficacy.

Question 9: Scientific Publications.

None to date.
Back to Index