Past research has clearly established that two groups of plants respond differently to elevated carbon dioxide levels, with one group gaining substantially more biomass when carbon dioxide is more abundant; however, a new study monitoring plots over a 20-year period reveals that this well-established pattern is in fact reversed over long time scales. Scientists categorize plants based on the way in which they process carbon, the two most common classes being C3 (e.g., rice, wheat, trees) and C4 (e.g., corn, sugarcane, most other grasses). Theory and experimental evidence have suggested that C3 grasses are more sensitive to carbon dioxide levels than C4 species and thus will grow and gain more biomass in response to rising carbon dioxide levels - a pattern that is critical to understand to accurately model future climate. But, past experiments have only looked at C3 and C4 responses over relatively short periods. Peter B. Reich et al. now report results from a 20-year study monitoring 88 plots in Minnesota, U.S., which are part of the BioCON project. They found that, during the first 12 years of the study, C3 plots averaged a 20% increase in total biomass in response to elevated carbon dioxide levels, compared to ambient conditions, while C4 plots averaged a 1% increase, changes that were in line with expectations. However, during the subsequent 8 years, the pattern reversed: C3 plots averaged 2% less than their ambient counterparts, and C4 plots averaged 24% more biomass. The researchers found that variables such as rainfall and net photosynthesis of the plants had little correlation with this reversal, while, mysteriously, the mineralization of nitrogen did. Mark Hovenden and Paul Newton provide more context in a related Perspective, illustrating the value of longer-term research in revealing the complexities of ecological patterns. They note that because C4 plant species contribute 25% of land biomass globally, provide an important forage source for grazing animals, and are overrepresented among weeds, it is especially important to correctly estimate the future distribution of these plants.