Significance Although plate tectonics is generally considered to be the main contributor of CO ₂ production, despite decades of paleoproxy and solid-earth research, this very first-order assumption remains unproven. Based on estimates of total subduction-zone lengths, from a combined analysis of seismic tomography models and plate tectonic reconstructions, we have bridged a major gap between solid-earth sciences and the atmosphere and biosphere sciences by providing understanding of one of the most fundamental controls of climate: variation of plate tectonic activity of our planet. By using the subduction zone length input into the GEOCARBSULF climate model, we obtain a better fit with atmospheric CO ₂ proxies compared with previously used plate tectonic assumptions.