Tuesday, August 5, 2008
Exhibit Hall CD, Midwest Airlines Center
James E. Bromberg, Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, Sunil Kumar, Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, CO and Cynthia S. Brown, Graduate Degree Program in Ecology, Bioagricultural Sciences and Pest Management, Colorado State University, Fort Collins, CO
Background/Question/Methods Bromus tectorum is an invasive weed that grows prolifically throughout the western United States. Land managers in the Rocky Mountains are concerned about its increase in abundance in higher elevation vegetation communities. Since land management agencies often do not have the funds to map entire ranges of species, distribution models can be a much more feasible option for making predictions. We used Maxent, a niche based species distribution model, to examine the range of B. tectorum from 1996 to 2007. Specifically, we wanted to know if the range of B. tectorum has expanded at high elevations, if the Maxent model could be used to make accurate predictions of the range of the species, and what environmental factors best predicted its distribution. Models were generated for 1996, 1999 and 2007 to look at changes in the predicted distributions over time.
Results/Conclusions
B. tectorum increased in each sampling period from 14 occurrences, 21 occurrences and 38 occurrences respectively. While the areas of high probability of occurrence did not change drastically, modeling the range in earlier sampling periods correctly predicted the new occurrences that appeared in later sampling periods. The environmental predictors that most influenced the models were the vegetation community, distance to roads and elevation. The model was able to predict the distribution of B. tectorum within the region of Rocky Mountain National Park that was sampled, but also predicted high probability of occurrence for other regions of the park. However, the predictions made for regions where no data were collected will need to be validated. Sampling of these other areas of the park will occur in 2008 to test whether the model can make inferences of large scale species distribution based on smaller areas of sampling. If the model has made accurate predictions of the entire park based on sampling within a smaller region, Maxent may be an extremely useful tool to land managers in making large scale species distribution predictions.
