PS 56-100
Macroecological forecasts of wildfire and invasion

Thursday, August 14, 2014
Exhibit Hall, Sacramento Convention Center
Amanda M. West, Natural Resource Ecology Laboratory and Bioagricultural Sciences and Pest Management Department, Colorado State University, Fort Collins, CO
Sunil Kumar, Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, CO
Background/Question/Methods

Imminent ecologic threats to the western United States include wildfires and invasive species. The wildfire-invasion cycle may occur in areas where wildfires promote invasive species, and these species in turn decrease wildfire intervals and increase wildfire intensity. One of the most conspicuous invasive species implicated with this cycle in the western US is Bromus tectorum(cheatgrass). Areas that are both suitable habitat for this grass and suitable for severe wildfires may be associated with higher fire suppression costs, habitat and forage loss, and degraded air, water, and soil resources. Macroecological approaches allow us to better understand the casual factors of landscape-level cycles such as wildfire-invasion and their interactions. Our objectives were to: (1) evaluate the environmental variables associated with severe wildfires and cheatgrass distribution in the Front Range of Colorado and Wyoming, (2) forecast spatial patterns and variation of potential wildfire-cheatgrass invasion risk, and (3) create maps highlighting areas that should be managed to alleviate future risk. We used macroecological models to examine the historical patterns of wildfires and suitable habitat for cheatgrass in the study area. We then compared the explanatory environmental covariates between models, evaluated patterns, and projected them into the same geographic space to understand areas at highest risk of the wildfire-cheatgrass invasion cycle.


Results/Conclusions

The maps we created from the models showed areas in the Front Range of Colorado and Wyoming currently suitable for both severe wildfire and cheatgrass occurrence. Both models had partial area under receiver operating characteristic curve (pAUC) values greater than 1.0. The importance of including both monthly and seasonal climatic variables in both models was highlighted in our results. These models may be used to develop future hypotheses of wildfire-cheatgrass invasion cycles in the region. The maps we created are valuable tools that will be available to land managers to prioritize specific areas for surveys, removal of invasive species, and targeted ecological restoration to reduce the risk of wildfires.