Collin W Ahrens1, Thomas H Meyer2, Carol A Auer3. 1. Royal Botanic Gardens Melbourne, Birdwood Avenue, Private Bag 2000, South Yarra, VIC 3141, Australia. 2. University of Connecticut, Department of Natural Resources and the Environment, 1376 Storrs Road, Unit 4087, Storrs, Connecticut 06269 USA. 3. University of Connecticut, Department of Plant Science, Agricultural Biotechnology Laboratory, 1390 Storrs Road, Unit 4163, Storrs, Connecticut 06269 USA.
Abstract
PREMISE OF THE STUDY: Expanded area cultivated with the bioenergy crop Panicum virgatum (switchgrass) could alter the genetics of native populations through gene flow, so understanding current and future species distribution is a first step toward estimating ecological impacts. We surveyed switchgrass distribution in the northeastern United States and generated statistical models to address hypotheses about current distribution relative to historical records and responses to climate change. METHODS: Surveys were conducted on 1600 km of road verges along environmental gradients. Switchgrass abundance became the training data for two multivariate generalized linear models that generated maps representing the probability of switchgrass in road verges. Models were evaluated and the superior model was used with variables from three climate change scenarios for 2050 and 2099. KEY RESULTS: Switchgrass populations were found in 41% of roadside plots and up to 188 km from the coast. The environmental variables temperature, urban areas, and sandy soils were positively correlated with switchgrass presence, while elevation, soil pH, and distance to the coast were negatively correlated. The model without spatial autocorrelation performed better. Models and maps incorporating climate change predictions showed a sharp northward shift in suitable habitat. CONCLUSIONS: Switchgrass populations in the northeastern United States occur on inland road verges, supporting the idea that species distribution has expanded relative to historical descriptions of a restricted coastal habitat. The optimal model showed that mean temperature, elevation, and urban development were important in switchgrass distribution today, and climate change will increase suitable habitat for future bioenergy production and wild populations.
PREMISE OF THE STUDY: Expanded area cultivated with the bioenergy crop Panicum virgatum (switchgrass) could alter the genetics of native populations through gene flow, so understanding current and future species distribution is a first step toward estimating ecological impacts. We surveyed switchgrass distribution in the northeastern United States and generated statistical models to address hypotheses about current distribution relative to historical records and responses to climate change. METHODS: Surveys were conducted on 1600 km of road verges along environmental gradients. Switchgrass abundance became the training data for two multivariate generalized linear models that generated maps representing the probability of switchgrass in road verges. Models were evaluated and the superior model was used with variables from three climate change scenarios for 2050 and 2099. KEY RESULTS:Switchgrass populations were found in 41% of roadside plots and up to 188 km from the coast. The environmental variables temperature, urban areas, and sandy soils were positively correlated with switchgrass presence, while elevation, soil pH, and distance to the coast were negatively correlated. The model without spatial autocorrelation performed better. Models and maps incorporating climate change predictions showed a sharp northward shift in suitable habitat. CONCLUSIONS:Switchgrass populations in the northeastern United States occur on inland road verges, supporting the idea that species distribution has expanded relative to historical descriptions of a restricted coastal habitat. The optimal model showed that mean temperature, elevation, and urban development were important in switchgrass distribution today, and climate change will increase suitable habitat for future bioenergy production and wild populations.
Keywords:
biofuel; climate change; generalized linear model; minimum predicted area; northeastern United States; spatially explicit model; species distribution model; switchgrass