VARIABILITY IN TEMPERATURE SEASONALITY IN “EQUABLE” GREENHOUSE CLIMATES: STATISTICAL ANALYSIS AND EVIDENCE FROM CRETACEOUS WESTERN NORTH AMERICA

Since at least the 1960s, paleoclimatologists interested in warm periods like the Cretaceous have grappled with two seemingly irreconcilable sources of information regarding Earth’s greenhouse climate system: 1) lithologic, paleontologic and geochemical climate proxies and 2) climate model simulations. Proxy reconstructions of greenhouse climates have been interpreted as evidence that greenhouse Earth was characterized by higher global temperatures in general, amplified polar mean annual temperatures, and reduced temperature seasonality or mean annual range in temperature (MART). In contrast, climate models struggle to reproduce these conditions –especially reduced MART. Identifying the cause of these proxy/model discrepancies is difficult because MART estimates during greenhouse periods vary significantly depending on proxy type. Paleobotany-based proxies yield reduced MART relative to the modern, while geochemistry-based proxies (e.g., pedogenic carbonate clumped isotopes) suggest similar-to-modern MART. One potential explanation for these differing proxy-based MART reconstructions is that they reflect actual MART variability due to environmental differences unique to each proxy. To test this hypothesis, we employ a conditional autoregressive statistical model and global long-term landcover, topographic, and climate data to quantify the influence of different proxy formation environments on MART. We show that landcover differences, which reflect local differences in environmental conditions, have a statistically significant impact on MART, and that after considering location, topography, and precipitation, proxy MART differences are largely explained by differences in landcover. Cretaceous MART results from a multi-proxy case study are consistent with this statistical analysis. These findings suggest that the variable MART reconstructions for past greenhouse periods reflect real variations in MART, and that like the modern world, greenhouse climates were characterized by highly variable MART conditions at any given latitude. These findings have important implications for comparing proxy paleotemperature reconstructions to climate model output, and suggest that the discrepancy between paleoclimate proxy and model interpretations may need to be reassessed.