Gregory, K.M., 1996, Are paleoclimate estimates biased by foliar physiognomic response to elevated atmospheric CO2?: Palaeo-3, v. 124, p. 39-51.

Abstract.-- Physiognomic analysis of fossil angiosperm leaves has provided an important quantitative database of Tertiary terrestrial paleoclimate. However, atmospheric CO2 level, a critical control on plant growth, may have been higher in the Tertiary. It is thus crucial to investigate whether elevated CO2 affects leaf physiognomy. In this study, leaves were collected from white oak (Quercus alba L.) seedlings grown in open-top growth chambers at Oak Ridge National Laboratory. The only physiognomic change noted is an increase in length to width ratio with increasing CO2. In the literature, leaf size has been observed to increase, decrease or remain unchanged for woody C3 species grown in elevated CO2. Typically, one sees more variation due to microsite or phenotype than due to CO2 level. By applying these observed physiognomic trends to two fossil floras, it is argued that estimates of mean annual temperature and growing season precipitation may be biased on the order of 1°C and 20 cm, respectively. These are relatively small effects, as the values are similar to the standard errors of the regression models used to estimate paleoclimate. The lack of data, the variability of response to CO2 associated with microsite and phenotype, and the question of whether observed short-term trends with elevated CO2 are sustained make it impossible to propose a correction factor. Adequate sample size and sampling of several sites are the best way to attempt to compensate for CO2 effects on a given fossil flora until response to CO2 is better understood.