Warming decreases the predictability of methane emissions in a coastal wetland

Abstract Accurate ecological forecasts are increasingly vital for research and management in a changing world. However, the predictability of ecological dynamics differs across space and time, and the drivers of predictability remain poorly understood. Here, we leveraged 8 years of whole‐ecosystem warming experiments (+0 to +5.1°C) in a coastal wetland (Maryland, USA) to assess the impacts of warming on ecological predictability. We generated 6‐month‐ahead forecasts of methane fluxes from 24 experimental plots using 10 diverse forecast models. Across all models, forecast performance worsened as temperatures increased, with 5× higher forecast error under the highest level of warming. Notably, the effects of warming on predictability varied between plant communities, with greater effects in plots dominated by C 3 sedges compared to C 4 grasses. Amid ongoing climate change, our findings highlight the potential for increased temperatures to decrease our capacity to forecast ecological changes over time.