The process of agricultural land-use is both a cause and consequence of other environmental changes, such as climate change, that threaten the long-term well being of ecosystems and societies. This process is influenced by biophysical, economic and sociocultural influences that operate on a hierarchically ordered set of spatial scales. As such the importance of one of these factors – climate – to agriculture evolves over time and from place to place. Thus a central problem for researchers interested in the possible impacts of climate change on agriculture to characterize the conditions under which climate does and does not matter to agriculture. The objective of this dissertation is to examine the vulnerability of U.S. Great Plains agriculture to possible changes in climate, using a statistical framework capable of estimating region-wide trends alongside localized deviations, for 1969-1992. Three sets of important findings emerge.

First, the commonly studied factors that influence an important dimension of agricultural land-use – per acre land values – are broadly applicable to the entire region. However, the importance of factors excluded from the statistical models (due to lack of data availability or inability to quantify) varies significantly between two sub-regions: the portion of the Great Plains overlying the Ogallala Aquifer and the rest of the region. Great Plains farmers with access to Ogallala groundwater for irrigation purposes are relatively insulated from swings in large-scale macroeconomic factors. Compared to farmers without access to irrigation water, Ogallala farmers benefit less from improvements in market conditions, but they also suffer less from downswings in commodity prices. Second, there is strong evidence that for the region as a whole (i.e., without respect to the Ogallala), that land values – and by extension the variations in land uses that give rise to the differing land values – exhibit important localized clustering. The land uses of a given county are determined not only by characteristics of that county but also by characteristics of neighboring counties. This finding supports the longstanding theory supporting the between-farmer diffusion of agricultural practices. A complete understanding of climate change impacts and policies designed to mitigate those impacts will need to account for and leverage social communication networks rather than assume that information is instantaneously accessible to all farmers everywhere. Third, the economic importance of climate to Great Plains agriculture varies with the time at which the relationship is evaluated. Precipitation in both the growing season and preseason is found to be significantly and positively associated with land values for all years in the study period, and the relationship is non-linear and of variable importance. According to the models, a simulated 8 percent increase in annual precipitation would result in significant benefits in the early years of the study period (about $566-722 million per year, in 1992 dollars), but much more modest revenues for the later years (about $206-277 million per year). It is not clear how much these benefits would be offset by costs (or augmented by other benefits) associated with climate change. However, the likelihood that regional farmers would be able to withstand such changes economically varies strongly with the year of analysis.

This dissertation is a necessary first step in a larger research endeavor to identify how agriculture in an economically and ecologically marginal region is simultaneously affected by local and global influences. Future research on possible climate change impacts for the Great Plains should focus on the process by which an important and declining natural resource – groundwater – is regulated, to understand the relative costs and benefits of differing regulatory approaches. Such an understanding will inevitably require an appreciation of the large-scale contexts against which local-scale, place-based dynamics play out. colin_polsky@harvard.edu