In addition to considering environmental sustainability, long-term sustainable biofuel production from switchgrass requires that high levels of biomass production be maintained over time. Field trials have shown that switchgrass yields are sensitive to spatial variation in temperature and precipitation (Casler and Boe 2003; Casler et al. 2004). Future climate change may similarly alter the capacity of biofuel production. Therefore, modeling biomass production under future climatic conditions and elevated atmospheric carbon dioxide concentrations is necessary to ensure maintenance of high yields over time without supplemental nutrients and irrigation.

Efforts to estimate future biomass of switchgrass using mechanistic models have been limited. Brown et al. (2000) used EPIC to predict the future yield of swtichgrass in four states (MO, IN, NE, and KS) using future weather data from the NCAR-RegCM2 model. Their results predicted that switchgrass yields would increase in this region by more than 8 Mg ha1. Behrman et al. (2013) used ALMANAC to estimate future yields across the entire central and eastern U. S. for current climate conditions under two climate change scenarios for the ten-year interval from 2080 to 2090 using the CCCMA-CGCM2 model. The A2 scenario was chosen to represent a pessimistic future scenario that predicts a large increase in atmospheric carbon dioxide and large increases in temperature. The B2 scenario was chosen as a "middle of the road" scenario that corresponds to a moderate increase in carbon dioxide levels and a smaller increase in temperature. Similar to Brown et al. (2000) the ALMANAC model also predicts increased yield for MO, IN, NE, KS. Furthermore, regions in Eastern Texas are predicted to have large decreases in yield by 2080s, whereas ND and SD are predicted to have large increases in yield. Modeling of future climate change scenarios predicts warmer minimum temperatures that will shift the USDA Hardiness Zones northward and may make conditions suitable for lowland switchgrass types to thrive further north in upland regions.

Mechanistic models can be used to locate areas that produce relatively high yields over a time. Areas with high long-term potential were determined using yields reported from Behrman et al. (2013) for three climate scenarios. Areas that continually produce yields greater than 10 Mg ha1 for all three climates are labeled as high long-term potential and areas with low long­term potential have high yields for only one of the three climate scenarios (Fig. 1). Assessing the potential of an area to sustain high productivity has the potential benefit of minimizing the amount of land conversion needed to meet production demands. Land-use change increases greenhouse gas emissions and is the primary factor responsible for the loss of biodiversity (Searchinger et al. 2008; Fletcher et al. 2011).