I Abstract 9628819 Predictions of global climate change resulting from the buildup of greenhouse gases in the atmosphere are based on models that largely ignore climate-ecosystem interactions. Yet such interactions have the potential to generate large positive or negative feedbacks to the climate system, thereby either enhancing or alleviating the magnitude of climate change. To quantify these feedbacks, a detailed and mechanistic understanding of climate-ecosystem interactions on the scale of experimental plots is needed, along with reliable methods for extrapolating knowledge obtained from ecosystem experimentation to the landscape scale. A widely used technique for acquiring landscape-scale understanding of ecosystem responses to climate change is to develop from data acquired at sites across the landscape a set of correlations between the local ecosystem variable in question and the local climatic conditions. These correlations can then be used to generate predictions of response to climate change, but the value of such predictions hinges on the assumption that ecosystem variables track changing climate in the same way that such variables now track climate differences over space. Climate manipulation experiments provide a more mechanistically-based approach to forecasting responses to climate change, and can identify the most important contingent factors (such as interannual climate variability) that influence those responses, but practical difficulties preclude carrying these out at many sites over many landscapes. Moreover, the existence of contingent factors influencing ecosystem response to climate change precludes scaling up by simply assuming that information learned on experimental plots at one site will apply in a straightforward manner to all the similar habitat (based, say, on a vegetational classification) over a landscape. This project will combine manipulations with elevational transect studies to 1) deepen understanding of the mechanisms and contingent factors that shape climate-ecosystem feedbacks, and 2) evaluate and improve the use of transect studies for extrapolating results about ecosystem response and feedback to the landscape scale. The focus will be on two linked ecosystem responses to climate variation that have been observed over four years of an experimental simulation of climate warming in mixed shrub/ grass/forb habitat in Colorado: 1) a heating-induced net outflow of carbon dioxide from the ecosystem and 2) a shift in dominant vegetation from forbs to shrubs. Because carbon dioxide is a greenhouse gas and because a shift in vegetative cover will alter energy and water exchange with the atmosphere, these responses to warming, on a larger scale, could be a source of significant feedback to the climate. ??
