Atmospheric CO₂ — Last 160 Million Years

Reading this chart. Each scatter point is one published CO₂ estimate from a single proxy site/sample; hover for the reported low–high range. The dark blue line is the Foster 2017 LOESS fit to all proxies combined; the surrounding shading shows the 68 % (inner) and 95 % (outer) confidence intervals on that fit. Use the legend to hide proxy methods or the CI bands. The CO₂ axis defaults to linear for direct comparison with the Berner 2001 (GEOCARB III) style; switch to log to better resolve low-CO₂ intervals such as the late Cenozoic.

Dashed black line: GEOCARB III (Berner & Kothavala 2001). A forward geochemical model of the long-term carbon cycle (weathering, volcanism, organic-carbon burial), evaluated at 10-Myr steps and published as RCO₂, the ratio to the pre-industrial value. Converted to ppm here by RCO₂ × 280 ppm. This is an independent model reconstruction, not a fit to the proxy data above, so divergence between the dashed line and the blue LOESS curve reflects genuine model–proxy disagreement (most notably in the mid-Cretaceous).

Hollow markers below ≈150 ppm flag estimates that fall under the rough C3-plant photosynthesis floor (Gerhart & Ward 2010). These are mostly pre-2009 paleosol studies (Ekart, Nordt, Cleveland, Montañez et al.) that Foster 2017 recomputed downward using the Breecker 2009 standard soil-respiration value S(z) = 2000 ppm; a few later studies (Nordt 2015) report similarly low values directly. They are kept in the dataset for fidelity to the source compilation but should not be interpreted as evidence of sub-150 ppm atmospheric CO₂.