Assessing effects of variation in global climate data sets on spatial predictions from climate envelope models

Journal of Fish and Wildlife Management
By: , and 

Links

Abstract

Climate change poses new challenges for natural resource managers. Predictive modeling of species–environment relationships using climate envelope models can enhance our understanding of climate change effects on biodiversity, assist in assessment of invasion risk by exotic organisms, and inform life-history understanding of individual species. While increasing interest has focused on the role of uncertainty in future conditions on model predictions, models also may be sensitive to the initial conditions on which they are trained. Although climate envelope models are usually trained using data on contemporary climate, we lack systematic comparisons of model performance and predictions across alternative climate data sets available for model training. Here, we seek to fill that gap by comparing variability in predictions between two contemporary climate data sets to variability in spatial predictions among three alternative projections of future climate. Overall, correlations between monthly temperature and precipitation variables were very high for both contemporary and future data. Model performance varied across algorithms, but not between two alternative contemporary climate data sets. Spatial predictions varied more among alternative general-circulation models describing future climate conditions than between contemporary climate data sets. However, we did find that climate envelope models with low Cohen's kappa scores made more discrepant spatial predictions between climate data sets for the contemporary period than did models with high Cohen's kappa scores. We suggest conservation planners evaluate multiple performance metrics and be aware of the importance of differences in initial conditions for spatial predictions from climate envelope models.

Publication type Article
Publication Subtype Journal Article
Title Assessing effects of variation in global climate data sets on spatial predictions from climate envelope models
Series title Journal of Fish and Wildlife Management
DOI 10.3996/072012-JFWM-056
Volume 5
Issue 1
Year Published 2014
Language English
Publisher U.S. Fish and Wildlife Service
Contributing office(s) Wetland and Aquatic Research Center
Description 12 p.
First page 14
Last page 25
Online Only (Y/N) N
Additional Online Files (Y/N) N
Google Analytics Metrics Metrics page
Additional publication details