PALEOBOTANY AND CRETACEOUS ARCTIC WARMING: A COMBINED EMPIRICAL AND MECHANISTIC PERSPECTIVE

A warm Cretaceous Arctic has posed problems for Earth System Models (ESMs). ESM simulations that use only geologically realistic levels of pCO₂ do not adequately reproduce high-latitude warmth. Low-albedo polar forests provide additional warmth, but this can be limited, especially in the absence of polar ice. Model-data discrepancies could indicate major problems with ESMs, but also could indicate inadequate knowledge of boundary conditions and processes that contributed to Cretaceous polar warmth.

Recent studies of Cretaceous vegetation, plant physiology, and climate physics suggest additional mechanisms of high-latitude warming and new methods for reconstructing Cretaceous Arctic boundary conditions. Analysis of vein density—which correlates with maximum transpiration rate—indicates that Cretaceous angiosperms had low maximum transpiration rates relative to Cenozoic and Recent angiosperms. Low transpiration rates in Cretaceous angiosperms could be an important component of high-latitude warming for times when angiosperms were common in Arctic vegetation. Altered properties of liquid clouds in a pre-anthropogenic atmosphere are another possible mechanism of Arctic warming, along with polar stratospheric clouds resulting from higher methane flux during the Cretaceous. Model simulations for the Maastrichtian that combine elevated atmospheric trace gases, low-albedo polar forests, and altered liquid clouds provide an improved simulation of polar warmth and latitudinal temperature gradients.

Full validation of models requires a spatially and stratigraphically resolved record of vegetation and climatic indicators. An abundant dispersed cuticle record from the North Slope of Alaska has the potential for detailed land-surface reconstructions and a stratigraphically resolved record of atmospheric pCO₂. This record has the potential to provide more accurate boundary conditions for model simulations and independent assessment of model output.