1 Department of Geosciences and NASA Astrobiology Institute. E–mail: achim@geosc.psu.edu

2 The Pennsylvania State University. University Park, PA 16802 USA. E–mail: brachio@geosc.psu.edu

Key words: Mass Extinction. Ocean modeling. Climate modeling. Obliquity cycles. Ice sheet model.

Due to its association with a major glaciation and absence of evidence for extraterrestrial causes, the Late Ordovician mass extinction has generally been attributed to environmental perturbations that were the result of this glaciation. However, the nearly complete subduction of Ordovician oceanic crust limits the data set of proxies from deep ocean sea floor sediments that can be used to study the Late Ordovician climate system. In particular we are left with mostly geochemical data from shallow epeiric seas that might not reflect the global climate characteristics. Numerical models of Earth’s atmosphere–ocean interactions therefore provide a tool to constrain those geochemical results within reasonable boundary conditions and assess the response of the climate system to different perturbations. We used numerical models to investigate the response of the Late Ordovician climate system to different perturbations. More specifically, we evaluated changes in sea level, solar insolation cycles (obliquity cycles), paleogeography and atmospheric pCO2.

We performed sensitivity experiments using Caradocian and Ashgillian paleogeographies with the atmospheric general circulation model (AGCM) GENESIS under a range of atmospheric pCO2 values (8–18x PAL; pre–industrial atmospheric level), high and low sea level, and two values of poleward ocean heat transport. We then coupled a 3–dimensional ice sheet model to the AGCM in order to investigate the necessary boundary conditions for ice sheet formation. All simulations with a high sea level and normal heat transport (i.e., modern day values) remain free of ice sheets, even with pCO2 levels as low as 8x PAL. Assuming that pCO2 did not fall below 8x PAL, a minimum value for this time period based on geochemical modeling and geochemical data from paleosols, a drop in pCO2 and the paleogeographic evolution coupled with an ice–albedo feedback can therefore be regarded as only preconditioning factors for the Late Ordovician glaciation. In order for ice sheets to form, other factors must have changed such as a drop in sea level from its generally high Late Ordovician levels and/or a reduction in poleward ocean heat transport.

We also used the ice sheet model to investigate the sensitivity of waxing and waning of these ice sheets to changes in atmospheric pCO2 and orbital forcing at the obliquity timescale (30 to 40 k.y.). Our simulations indicate that large ice sheets, grown during extreme periods of low pCO2 (8x PAL), can subsequently be sustained during periods of higher pCO2 (9–10x PAL) that would otherwise prevent the growth of ice from ice–free starting conditions. Thus, if atmospheric pCO2 was one of the main driver of climate during the Late Ordovician then atmospheric pCO2 must have risen to greater than 10x PAL to melt the ice–sheets in order to end glaciation.

The AGCM results were also used to produce the forcing boundary conditions and initial conditions for an ocean general circulation model (OCGM) (MOM v.2.2) in order to investigate oceanic feedbacks. In particular, we looked at changes in ocean heat transport in response to continental drift and changes in atmospheric pCO2 and sea level. In all simulations, a drop in sea level led to a reduction in poleward ocean heat transport. This indicates a possible positive feedback that could have led to enhanced global cooling in response to falling sea level before or in the early stages of glaciation. Alterations in poleward ocean heat transport linked to changes of atmospheric pCO2 also indicate that there is a threshold of 10x PAL, above which changes in poleward ocean heat transport can not be responsible for glaciation in the Late Ordovician. While continental drift can not explain the initiation of the glaciation by itself, it could explain the observed global cooling trend in the Late Ordovician through a combined poleward ocean heat transport feedback and increased ice–albedo effect if atmospheric pCO2 was low during the entire Late Ordovician. The OGCM results indicate that due to vigorous meridional overturning, a stagnant global ocean could not have existed during the Late Ordovician, even under pCO2 levels as high as 18x PAL when the thermal gradient between the poles and the equator where relatively low. In addition, prior estimates of a sea surface temperature drop of ~11ºC during the glaciation, based on oxygen isotopes, are not supported by the OGCM results. In our simulations a change of atmospheric pCO2 from 18x PAL to 8x PAL only leads to a ~5ºC drop in surface ocean temperatures.

Overall, these results help constrain environmental changes that caused the Late Ordovician mass extinction.