The Weed Invasion and Management Simulator (WIMS) tool
The Weed Invasion and Management Simulator (WIMS) (currently in an offline prototype form) is an interactive digital tool designed to empower land managers, policy makers, and practitioners with robust, transparent and evidence-based insights for invasive weed management.
WIMS integrates over a decade of weed control and monitoring data for the Australian Capital Territory (ACT), advanced analytical models and expert knowledge on three key factors influencing the outcomes of weed control: the effectiveness of control in reducing weeds; weed recurrence following control and landscape context/habitat suitability.
Through its user-friendly interactive dashboards, WIMS enables users to seamlessly report on, and evaluate the value of, weed control efforts. The tool currently focuses on four priority weeds in the ACT that cause significant negative impacts on environmental values across south-eastern Australia: African lovegrass, Chilean needle grass, serrated tussock and St John’s wort.
Key features
- Integrates harmonised long-term weed control, monitoring, and environmental datasets.
- Integrates robust estimates of key parameters affecting weed population trends in actively managed landscapes derived using advanced statistical models.
- ‘Weed control history’ dashboard allows users to filter weed control activities by species, year and region and explore a range of summary metrics on weed control activities and weed cover assessments.
- ‘Weed density simulation’ dashboard allows users to predict and compare likely weed density trends for four priority weeds under alternative management scenarios:
- Management scenarios can be specified to reflect actual weed control effort, increased effort, decreased effort, or a counterfactual scenario (i.e., ‘no control effort’).
- Hindcasting functionality can be used to explore the value of past weed control effort.
- Forecasting functionality can be used to explore the value of future investments in weed control.
- Both ‘weed control history’ and ‘weed density simulation’ dashboards provide insights on spatial patterns, temporal trends and well as regional summaries of weed control and its outcomes.
A view of the options in the ‘Weed control history’ dashboard, here showing summaries for all control activities for all weeds across the ACT.
Jerrabomberra Reserve case study
The Jerrabomberra Grassland Reserve is home to many rare and threatened flora and fauna.
The ACT’s Jerrabomberra Reserve is an important conservation area and harbours some of the last remnants of Natural Temperate Grasslands, a threatened ecological community, alongside several flora and fauna species listed under the Environment Protection and Biodiversity Conservation Act. Controlling invasive weeds in this 261-hectare reserve is a priority and Reserve Managers need to effectively allocate limited resources to manage multiple threats.
The WIMS tool provides Reserve managers a single platform for viewing and reporting on past control efforts, as well as for exploring alternative control scenarios to demonstrate the value of their work and to guide future planning.
St John’s wort is a high priority weed invading and damaging native grasslands across the ACT and a major focus of weed control in Jerrabomberra Reserve.
Weed control history
The ‘Weed control history’ dashboard summarises the weed control history for St. John’s wort that occurred within Jerrabomberra Reserve over a user-specified time period (note that the dataset available for this research ended in June 2023). Reserve managers can review metrics that give insight into weed cover assessments and account for weed control activities over time.
The data shows that the area treated for St John’s wort in Jerrabomberra Reserve has steadily increased over the final three years in the dataset from July 2020 to June 2023. Yet, despite this sustained control effort, assessments conducted by weed control contractors before applying herbicide treatment reveal that St John’s wort cover steadily increased to mostly “dominant” levels.
This begs the question: “Was the considerable weed control effort worth it?”
A view of the ‘Weed control history’ dashboard, here showing control activities targeting St John’s wort in Jerrabomberra Reserve between July 2020 and June 2023.
Weed density simulation
[Disclaimer: while derived from high quality data using scientifically robust methods, CSIRO cannot guarantee the accuracy of the simulation results presented below, not least because it is impossible that all factors affecting weed densities in complex landscapes are taken into account in our models.]
Hindcasting function
Reserve managers can explore that question using the hindcasting function of the ‘Weed density simulation’ dashboard. For example, by simulating likely weed density outcomes of the actual control effort for St. John’s wort during this 3-year period (scenario 1) and comparing these to a counterfactual scenario where no control was undertaken (scenario 2).
In both scenarios, weed density was predicted to increase over these three years (the simulation results panel shows predicted values at the end of the simulation period, i.e. June 2023). However, without control, the model indicates that weed density would have increased by about twice as much to an average median density of 140 plants per hectare. This compares to an average median density of 73 plants per hectare with control.
The avoided increase in St. John’s wort density demonstrates the value of active weed control in Jerrabomberra Reserve.
A view of the ‘Weed density simulation’ dashboard, here showing a scenario in Jerrabomberra Reserve using the hindcasting function.
Forecasting function
Armed with that knowledge, Reserve Managers can then use the forecasting function to explore what level of ongoing control effort may allow them to truly get on top of St John’s wort in Jerrabomberra Reserve. For example, by simulating likely weed density outcomes over the following three years (July 2023 to June 2026) under three control scenarios: annual control with current efficacy levels (~75% of plants killed per year) (scenario 1); no future control (scenario 2); and annual control with increased (90%) efficacy levels (scenario 3).
Without any ongoing investment in weed control, the model predicts a further massive increase in St John’s wort to potentially thousands of plants per hectare. Even if the current control regime is maintained, this may not be enough to suppress the weed to acceptable densities. However, the model suggests that St. John’s wort density could be gradually reduced to an average median density of only 10 plants per hectare by June 2026 if control teams could find ways to remove a greater proportion of plants each year.
Our model also shows variability across the landscape. In locations that provide highly suitable conditions for St John’s wort, the weed may persist even with intensive and effective control effort.
A view of the ‘Weed density simulation’ dashboard, here showing a scenario in Jerrabomberra Reserve using the forecasting function.
Towards translation and broader application
The WIMS tool is currently a prototype whose purpose is to showcase the potential value of digital solutions for weed managers. WIMS delivers advanced analytical models based on high-quality long-term datasets through user-friendly interactive dashboards that have been co-designed to serve the reporting and evaluation needs of land managers and policy makers.
Translating WIMS from prototype to fully operational decision support tool is the exciting next step, and requires a separate, dedicated project to ensure full integration with government systems and workflows.
Importantly, the WIMS approach is not limited to the ACT context or to the four priority weeds that have been the focus of most control effort and monitoring in the ACT. The tool is designed to be adaptable to any context where rich, longitudinal data on weed control activities and trends in weed populations exist or are being collected.
For example, it can be applied to weed eradication programs, where it can be complemented by expert-elicited data on control efficacy to fill knowledge gaps. Our novel data integration and modelling framework can be extended to cross-jurisdictional and national scales, supporting collaborative, evidence-based weed management across Australia.
There are also many opportunities for integrating new features and research findings into the tool. For example, the likely native biodiversity outcomes of weed control could be simulated in addition to trends in weed density (see our research on this topic here). Landscape spread of weeds could be modelled and integrated as a fourth key parameter affecting weed population trends in actively managed landscapes. Given the right data is available, the opportunities are endless!
If you are interested in partnering with us to advance the application of the WIMS tool and realising its potential for adaptive, outcomes-focused weed management, please get in touch with Dr Ben Gooden or Dr Jens Froese.