Australia has numerous fossil floras suitable for paleoclimatic analysis, with potential to improve understanding of Southern Hemisphere climatic evolution. Leaf-margin analysis (LMA) is a widely used method that applies present-day correlations between the proportion of woody dicot species with untoothed leaves and mean annual temperature to estimate paleotemperatures from fossil megafloras. Australia's unique history and vegetation imply that its leaf-margin correlation might differ from other regions; these possible differences are investigated here to improve paleoclimatic interpretations.

Australian rainforest vegetation shows nearly the same regression slope as recorded in East Asia and the Americas, indicating a globally convergent evolutionary response of leaf form to temperature. However, Australian sites tend to have fewer toothed species at localities with the same temperature as Asian and American sites. The following factors, singly or in combination, may account for this difference: (1) Australia's Cenozoic movement into lower latitudes, insulation from global cooling, and isolation from high-latitude cold-tolerant vegetation sources; (2) lack of high mountains as sources and refuges for cold-adapted taxa; (3) Pleistocene extinctions of cold-adapted taxa; and (4) the near absence of a cold-climate forest ecospace in Australia today.

Application of Australian LMA to Australian Cenozoic floras resulted in cooler temperature estimates than other LMA regressions. However, Australian paleotemperature estimates should account for the relative importance of cold-deciduous taxa. The timing and magnitudes of the extinctions of cold-adapted lineages are not known, and the most conservative approach is to use Australian LMA as a minimum and non-Australian LMA as a maximum temperature estimate.