Evidence continues to accumulate that humans are significantly increasing atmospheric CO 2 concentrations, resulting in unprecedented changes in the global climate system. Experimental manipulations of terrestrial ecosystems and their components have greatly increased our understanding of short-term responses to these global perturbations and have provided valuable input to ecosystem, dynamic vegetation, and global scale models. However, concerns exist that these initial experimental responses may be transitory, thereby limiting our ability to extrapolate short-term experimental responses to infer longerterm effects. To do these extrapolations, it will be necessary to understand changes in response patterns over time, including alterations in the magnitude, direction, and rate of change of the responses. These issues represent one of our largest challenges in accurately predicting longer-term changes in ecosystems and associated feedbacks to the climate system. Key issues that need to be considered when designing future experiments or refining models include: linear vs. non-linear responses, direct vs. indirect effects, lags in response, acclimation, resource limitation, homeostasis, buffers, thresholds, ecosystem stoichiometry, turnover rates and times, and alterations in species composition. Although experimental and landscape evidence for these response patterns exist, extrapolating longer-term response patterns from short-term experiments will ultimately require a unified multidisciplinary approach, including better communication and collaboration between theoreticists, experimentalists and modelers.