One of the major issues confronting management of parks and reserves is the invasion
of non-native plant species. Yosemite National Park is one of the largest and bestknown
parks in the United States, harbouring significant cultural and ecological
resources. Effective management of non-natives would be greatly assisted by information
on their potential distribution that can be generated by predictive modelling
techniques. Our goal was to identify key environmental factors that were correlated
with the percent cover of non-native species and then develop a predictive model
using the Genetic Algorithm for Rule-set Production technique. We performed a
series of analyses using community-level data on species composition in 236 plots
located throughout the park. A total of 41 non-native species were recorded which
occurred in 23.7% of the plots. Plots with non-natives occurred most frequently at
low- to mid-elevations, in flat areas with other herbaceous species. Based on the
community-level results, we selected elevation, slope, and vegetation structure as
inputs into the GARP model to predict the environmental niche of non-native
species. Verification of results was performed using plot data reserved from the
model, which calculated the correct prediction of non-native species occurrence as
76%. The majority of the western, lower-elevation portion of the park was predicted
to have relatively low levels of non-native species occurrence, with highest concentrations
predicted at the west and south entrances and in the Yosemite Valley.
Distribution maps of predicted occurrences will be used by management to: efficiently
target monitoring of non-native species, prioritize control efforts according to the
likelihood of non-native occurrences, and inform decisions relating to the management
of non-native species in postfire environments. Our approach provides a valuable
tool for assisting decision makers to better manage non-native species, which can be
readily adapted to target non-native species in other locations.
| 
