The CO2 emitted from the soil surface is produced by plant roots and soil microbes, and is called soil respiration. Soil respiration is one of the largest sources of CO2 to the atmosphere. Yet, our understanding of the mechanisms that control how much soil respiration occurs, and how soil respiration will respond to a changing climate, remains poor. I propose to synthesize existing soil respiration datasets that span different ecosystem types, with a particular emphasis on arid ecosystems. Arid ecosystems cover more than one-third of Earth’s land surface, and are relatively under-studied compared to tropical, temperate and boreal ecosystems. This work will increase our understanding of the basic factors that determine soil respiration across different ecosystem types, and improve how soil respiration is represented in global climate models. This information is needed to improve predictions of atmospheric CO2 concentrations and thus Earth’s future climate.