Predicting habitat suitability for invasive velvet grass (Holcus lanatus) in Yosemite, Kings Canyon, and Sequoia National Parks
Graduation Date
2015
Document Type
Thesis
Program
Other
Program
Thesis (M.S.)--Humboldt State University, Natural Resources: Environmental and Natural Resource Sciences, 2015
Committee Chair Name
Alison O'Dowd
Committee Chair Affiliation
HSU Faculty or Staff
Keywords
Invasive plants, Humboldt State University -- Theses -- Natural Resources, Holcus lanatus, GIS modeling, Habitat suitability model
Abstract
In order to restore and maintain natural ecosystems in wilderness areas, it is crucial that land managers understand which ecosystems are most threatened by invasive plant species and where to search for new infestations. Limited personnel resources and vast areas of rugged terrain make targeted early detection surveys an urgent need. Habitat suitability modeling is a spatial analysis tool that provides further understanding and graphical representation of the potential distribution and spread of invasive plant species. Velvet grass (Holcus lanatus) is a non-native perennial grass that aggressively invades wet meadows in the Sierra Nevada mountain range of California. This study used 2,865 presence locations of velvet grass in Sequoia, Kings Canyon, and Yosemite National Parks. Maximum Entropy (Maxent) modeling software was used to develop three different habitat suitability maps: 1) a 909-meter resolution map based on climate data, 2) a 10-meter resolution map based on terrain and vegetation features, and 3) a 10-meter resolution map with terrain and vegetation features, also using elevation data as a surrogate for climate data. All models were tested for area under the receiver operating curve and Aikake's Information Criterion. The vegetation and terrain-based models showed high habitat suitability for un-infested areas of all three national parks. The climate-based model, which could be used to predict habitat suitability in a changing global climate, was limited by its coarse spatial resolution of 909 square meters in an area with heterogeneous topographic features. The 10-meter resolution map using elevation as a surrogate served as a combination between the other two models. All models had high areas under the receiver operating curves, but the high-resolution models using fewer predictor layers had much higher Akaike Information Criterion scores. Separate maps of Yosemite National Park were also generated to test whether adding soil data improved model performance. Adding soil data improved the area under the receiver operating curve of the Yosemite model, but decreased the Akaike Information Criterion score. These models reveal characteristics that are common in areas infested by H. lanatus, such as wetter areas, flatter slopes, and higher temperatures, and highlights un-infested areas that could be future suitable habitat. Because of the high correlation between elevation and different climate variables in the three National Parks, the map using elevation as a surrogate for climate variables provides useful information at a fine enough resolution (10 meters) that it could be used by managers and field crews to help prioritize early detection surveys. Additional analysis is needed to determine the impact of potential vectors (i.e., pack stock, hikers, and rivers) on the location and likelihood of future infestations.
Recommended Citation
Degenstein, Erin Rose, "Predicting habitat suitability for invasive velvet grass (Holcus lanatus) in Yosemite, Kings Canyon, and Sequoia National Parks" (2015). Cal Poly Humboldt theses and projects. 1829.
https://digitalcommons.humboldt.edu/etd/1829
https://scholarworks.calstate.edu/concern/theses/79408061c