Young, Nicholas E., authorStohlgren, Thomas J., advisorKelly, Eugene Francis, committee memberGraham, James J., committee memberEvangelista, Paul Harrison, committee member2007-01-032007-01-032010http://hdl.handle.net/10217/45982Species distribution models are frequently used to predict species occurrences in novel conditions, yet few studies have examined the effects of extrapolating locally collected data to regional scale landscapes. Using boosted regression trees, I examined the issues of spatial scale and errors associated with extrapolating species distribution models developed using locally collected abundance data to regional extents for a native and alien plant species across a portion of the central plains in Colorado. Topographic, remotely sensed, land cover and soil taxonomic predictor variables were used to develop the models. Predicted means and ranges were compared among models and predictions were compared to observed values between local and regional extent models. All models had significant predictive ability (p < 0.001). My results suggested: (1) extrapolating local models to regional extents may restrict predictions; (2) modeling species abundance may prove more useful than models of species presence; (3) multiple sources of predictors may improve model results at different extents; and (4) regional data can help refine and improve local model predictions. Regional sampling designed in concert with large sampling frameworks such as the National Ecological Observatory Network, Inc (NEON) may improve our ability to monitor changes in local species abundance.born digitalmasters thesesengCopyright and other restrictions may apply. User is responsible for compliance with all applicable laws. For information about copyright law, please see https://libguides.colostate.edu/copyright.abundanceextrapolationcentral plainsboosted regression treesPlant ecology -- ColoradoPhytogeography -- ColoradoPlants -- Habitat -- Research -- MethodsSpatial ecology -- ColoradoText