Understanding variations in temperature and atmospheric CO2 concentration in Earth’s deep past is critical for understanding how global climate will respond to future changes in CO2 concentration. Here, I evaluate the assumptions of two well-known methods for reconstructing paleoclimate from fossil leaves: leaf-physiognomy climate-models for reconstructing temperature and the Franks model for reconstructing CO2 levels. Leaf-physiognomy climate-models assume that all plants, regardless of species, change their leaf shape rapidly in response to temperature, with colder temperatures often selecting for larger and more abundant leaf teeth (the serrated edges on leaf margins) and a greater leaf dissection. The Franks model assumes that the leaf temperature at which fossilized leaves developed does not affect the CO2 estimates. I grew six dicotyledonous tree species (two from seed, two from saplings, and two from seed and saplings) in growth cabinets at two temperatures (time-weighted mean = 25 and 17 °C) but under the same CO2 concentration (500 ppm). I measured the size and shape of the leaves to assess differences between the temperature treatments. I also ran the Franks model on the leaves to determine the validity of the model given its assumptions.

Measurements revealed that the plants tended to develop more dissected leaves with more abundant and larger leaf teeth in the cool treatment. This upholds the assumptions that leaf shape changes predictably and rapidly in response to temperature change. However, a third assumption, that leaf shape responds the same way to temperature regardless of species, was not upheld because temperature did not have the same effect on each species. Therefore, these models should take into account how the leaf morphology of different species react to changes in temperature.

For the Franks model, neither the measured inputs (stomatal density, stomatal size, and δ13C of the leaf tissue) nor the CO2 estimates were significantly different between the two temperature treatments. The mean CO2 estimate for the species was close to the target value of 500 ppm in both temperature treatments (625 and 534 ppm in the warm and cool treatment respectively). This gives paleobotanists greater confidence in the CO2 estimates using the Franks model.