Accurate and reliable predictions of invasive species distributions are urgently needed by land managers for developing management plans and monitoring new potential areas of establishment. Presence-only species distribution models are commonly used in these evaluations, however they are rarely tested with independent data over time or compared with presence-absence models fit with the same presence data. Using Maxent, we developed a presence-only model of invasive cheatgrass (Bromus tectorum L.) distribution in Rocky Mountain National Park, Colorado, USA in 2007 fit with limited data, and then tested the model with independent presence and absence data collected between 2008 and 2013. This model was verified using threshold dependent and threshold independent evaluation metrics. Next, we developed a Maxent model with cheatgrass presence data from 2007 through 2013 (i.e. Maxent 2013), and compared this model to a presence-absence method (i.e., generalized linear model; GLM 2013) using the same data. Threshold dependent and threshold independent evaluation metrics suggested Maxent 2013 outperformed GLM 2013, and a two-tailed Wilcoxon signed rank test indicated relative probability outputs were not significantly different between the models in geographic space. Based on known presences and absences of cheatgrass collected in the field, the Maxent 2013 and GLM 2013 relative probability outputs were highly correlated at absence locations but less correlated at presence locations. A Kappa comparison of Maxent 2007 and Maxent 2013 binary output provides evidence that Maxent is robust when fit with limited data. Our results indicate Maxent is an appropriate model for use when land management objectives are supported by limited resources and thus require a conservative, but highly accurate estimate of habitat suitability for invasive species on the landscape

National Parks are hallmarks of ecosystem preservation in the United States. The introduction of alien invasive plant species threatens protection of these areas. Bromus tectorum L. (commonly called downy brome or cheatgrass), which is found in Rocky Mountain National Park (hereafter, the Park), Colorado, USA, has been implicated in early spring competition with native grasses, decreased soil nitrogen, altered nutrient and hydrologic regimes, and increased fire intensity. We estimated the potential distribution of B. tectorum in the Park based on occurrence records (n = 211), current and future climate, and distance to roads and trails. An ensemble of six future climate scenarios indicated the habitable area of B. tectorum may increase from approximately 5.5% currently to 20.4% of the Park by the year 2050. Using ordination methods we evaluated the climatic space occupied by B. tectorum in the Park and how this space may shift given future climate change. Modeling climate change at a small extent (1,076 km2) and at a fine spatial resolution (90 m) is a novel approach in species distribution modeling, and may provide inference for microclimates not captured in coarse-scale models. Maps from our models serve as high-resolution hypotheses that can be improved over time by land managers to set priorities for surveys and removal of invasive species such as B. tectorum.

Early detection of invasive plant species is vital for the management of natural resources and protection of ecosystem processes. The use of satellite remote sensing for mapping the distribution of invasive plants is becoming more common, however conventional imaging software and classification methods have been shown to be unreliable. In this study, we test and evaluate the use of five species distribution model techniques fit with satellite remote sensing data to map invasive tamarisk (Tamarix spp.) along the Arkansas River in Southeastern Colorado. The models tested included boosted regression trees (BRT), Random Forest (RF), multivariate adaptive regression splines (MARS), generalized linear model (GLM), and Maxent. These analyses were conducted using a newly developed software package called the Software for Assisted Habitat Modeling (SAHM). All models were trained with 499 presence points, 10,000 pseudo-absence points, and predictor variables acquired from the Landsat 5 Thematic Mapper (TM) sensor over an eight-month period to distinguish tamarisk from native riparian vegetation using detection of phenological differences. From the Landsat scenes, we used individual bands and calculated Normalized Difference Vegetation Index (NDVI), Soil-Adjusted Vegetation Index (SAVI), and tasseled capped transformations. All five models identified current tamarisk distribution on the landscape successfully based on threshold independent and threshold dependent evaluation metrics with independent location data. To account for model specific differences, we produced an ensemble of all five models with map output highlighting areas of agreement and areas of uncertainty. Our results demonstrate the usefulness of species distribution models in analyzing remotely sensed data and the utility of ensemble mapping, and showcase the capability of SAHM in pre-processing and executing multiple complex models.