Leaf form and the reconstruction of past climates



Climate is accepted as the primary control of plant distribution, as is the idea that leaf morphology reflects adaptation to the environment. Because of this relationship, climate change is expected to fundamentally alter present vegetation patterns. Certainly it is the case that shifts in climate in the past have altered vegetation distribution; changes that in part are documented through examining the pattern of leaf forms observed in fossil deposits around the world. Understanding leaf form and climate may inform about past climates, and it is this connection that makes the paper by Traiser et al. (pp. 465–484 in this issue) of interest to both plant and earth scientists interested in climate change.

‘What stands out is the corroboration of the relationship between the proportion of woody species with untoothed leaf margins and mean annual temperature as a truly global phenomenon’

Part of the debate about whether currently observed global climate change is due to anthropogenic or natural influences is the extent to which natural variation in climate – and the various factors that influence climate – are contributing to increasing temperatures. One of the ways in which climate change science addresses these issues is to examine the recent and geological past, when global temperatures were much warmer than they are now due to the influence of factors wholly independent of human activity (Shellito et al., 2003). For example, during the Eocene geological epoch (34–55 Myr ago), the poles were ice-free and palms, crocodiles and other tropical organisms were found at middle and high latitudes throughout Europe, North and South America, Asia and Australia (Greenwood & Wing, 1995).

Leaf form can be used to reconstruct climates of the past

Palaeobotanical analyses of the Eocene have also shown that temperatures shifted alternately cooler and warmer by 5–10°C, often over geologically short intervals of time (Wilf, 2000; Greenwood et al., 2003). As discussed by Traiser et al. in their paper, key tools in reconstructing climates of the geological past using plant fossils are methods that use observations in the modern world of close correlations between leaf form, or physiognomy, and climate parameters (Bailey & Sinnott, 1915; Wolfe, 1979, 1995; Wilf, 1997; Greenwood et al., 2004).

The best known leaf physiognomy method is leaf margin analysis (LMA), where the proportion of woody dicots with uninterrupted (or ‘entire’) leaf margins, later refined to mean ‘untoothed’, was found to be correlated with mean annual temperature (MAT) (Fig. 1). Leaf margin analysis has a long pedigree, having been first proposed in the early 20th century (Bailey & Sinnott, 1915), and developed further in the past 30 years (e.g. Wolfe, 1979; Wilf, 1997; Burnham et al., 2001; Kowalski & Dilcher, 2003; Greenwood et al., 2004; Traiser et al.). Wolfe (1995) also proposed a multivariate approach, called climate leaf analysis multivariate program (CLAMP) that considered multiple leaf traits and climate variables for sites across the Northern Hemisphere and a small number of Southern Hemisphere sites. The application of the CLAMP data set to fossil Australian and European floras, however, often shows a systematic underestimation of MAT and temperature minima in comparison to other climate estimation methods, perhaps indicating that leaf form and climate in these areas reflected a different climate and vegetation history (Greenwood & Wing, 1995; Utescher et al., 2000; Greenwood et al., 2004).

Figure 1.

This figure plots data showing the proportion of woody dicot species with untoothed leaves vs mean annual temperature. The leaf cartoons serve to illustrate the difference between a toothed and an untoothed leaf. For most of the data shown, forest site census data was used to score LMP. In some studies, woody vines or climbers as well as trees would have been included in the census, whereas in others only trees were scored for leaf margin type. The data from Traiser et al. (this issue, pp. 000–000) are based on synthetic species lists for grid squares. The graph shown here reverses the axes of Fig. 5 from Traiser et al. to match the style used in most other papers on the topic, and is largely based on a similar figure in Greenwood et al. (2004). The CLAMP 3B data is from Wolfe (1995) (and sources cited in Greenwood et al., 2004) and reflects modern vegetation sites principally from across the Northern Hemisphere, but includes a small number of sites in the Southern Hemisphere. All other data is from sources cited in Greenwood et al. (2004).

Traiser et al. analysed synthetic species lists for 0.5° × 0.5° latitude/longitude grid squares across Europe, and demonstrated that leaf traits there, including leaf margin type, are correlated with climate, as they are elsewhere in the world. Their study looked at a range of leaf traits, and several climate variables, in simple linear regression (as used in LMA) and in multiple linear regression. Of most interest, however, was how their observations fit into the broader literature on leaf margin analysis.

What is most important in determining leaf form: climate, soils or phytogeography?

A criticism with applying leaf margin analysis to reconstructing past MAT was that the original correlation was based on one geographical area, East Asia (Wolfe, 1979). Part of the concern was the possible role that phytogeography may have played in shaping leaf form and climate relationships. Principally, are some plant lineages – such as the predominantly temperature deciduous Acer, Betulaceae and Ulmaceae – phylogenetically predetermined to having toothed leaves? Contributing to this concern was a poor understanding as to why some leaves have teeth, and others do not (Mosbrugger & Roth, 1996; Wilf, 1997; Greenwood et al., 2004).

Another concern about leaf margin analysis was the influence on leaf form of local edaphic controls, such as differences between riparian and lakeside vegetation (where water was not likely to be limiting to the trees) and sites removed from water bodies, such as ridges, where water supply may be a limiting factor influencing leaf form (Kowalski & Dilcher, 2003). In local forest habitats such as stream-edges vs forest interiors, life forms such as woody vines may preferentially have toothed or untoothed leaves, shifting the relative proportion of toothed species between these habitats (Burnham et al., 2001).

Leaf form and climate relationships do show globally consistent patterns

What stands out from the study of European vegetation by Traiser et al. is the corroboration of the relationship between the proportion of woody species with untoothed leaf margins (LMP) and MAT as a truly global phenomenon (Fig. 1). Greenwood et al. (2004) had shown that the LMP vs MAT relationship held in Australia, but that it was different to that for East Asia and the Americas. Studies across the Northern Hemisphere or in South America had generally found the same relationship as had been previously demonstrated for East Asia (Wilf, 1997; Greenwood et al., 2004). Limited studies in Africa and New Zealand, however, found a poor or no correlation between LMP and MAT. Traiser et al. however, show that this relationship is the same in Europe – albeit with a weaker correlation – than shown for East Asia, North America and South America, and is of the same character as has been shown for Australia.

The close correspondence between the European vegetation LMA regression line and that for the other Northern Hemisphere data sets is extraordinary (Fig. 1). This has implications for (1) applying the LMA regression equation to palaeotemperature analyses across the world, and (2) for understanding broad-scale relationships between leaf form and climate – it shows convergence for some leaf characters globally towards a common functional morphology by woody plants, but also highlights differences between Europe and elsewhere.

The weaker correlation between LMP and MAT for the European data than for other Northern Hemisphere data (i.e. lower r2 value) matches the result for the Australian LMA regression. The additional scatter in the Australian data, relative to the East Asian and Americas data (Fig. 1), and thus the weaker correlation, was attributed to the influence of highly infertile soils included in the Australian sites, and perhaps more so differences in the climatic history of that continent relative to the others, preventing the acquisition of a cold-adapted temperate deciduous toothed-leaf flora (Greenwood et al., 2004).

The European data reflects a predominantly cold-adapted deciduous flora – in common with the other Northern Hemisphere studies. It also used synthetic floras – floral lists based on all taxa known from a grid square – rather than site floral census data, and as Traiser et al. point out in their paper, their method likely sums potential differences in the proportion of woody species with toothed margins (or other leaf traits) between different local habitats, such as stream-side and ridge-top vegetation (Burnham et al., 2001; Kowalski & Dilcher, 2003). Traiser et al. point out that, in the highly human-modified landscape of Europe, obtaining site census data across a sufficient range of climates is difficult because very little unaltered vegetation remains. That their study still demonstrated significant correlations between leaf traits and climate variables is therefore all the more impressive and informative.

In their paper, Traiser et al. also demonstrated that temperature minima had the highest significance in determining leaf traits, in part matching the Australian study which found comparable correlation levels between LMP and both MAT and the mean minimum temperature of the coldest quarter (essentially analogous to Tmin/coldest month minimum temperature used in the European study). This is an important observation from two perspectives: (1) it further highlights global convergence in leaf-climate responses, at least for leaf margin teeth; and (2), it points in the direction of the functional significance of teeth – temperature minima would act as a stressor during leaf expansion, whereas MAT is likely a proxy for the causative selection pressure for toothed leaf margins.


The recent papers that addressed some of the areas of concern for LMA (e.g. Burnham et al., 2001; Kowalski & Dilcher, 2003; Greenwood et al., 2004; Traiser et al.) were not all good news for palaeobotanists and palaeoclimatologists; however, they have all shown leaf margin analysis to be a simple but effective tool.

As they assert in their paper, Traiser et al. provide a methodology for additional studies of leaf form and climate, including leaf margin type, for other parts of the world where human activity has also removed or greatly modified the natural vegetative cover, such as much of mainland China and South-east Asia. Their approach also offers a different path for multivariate techniques to reconstruct past climates. What may still be required from Europe, however, is an analysis of whether a specialized stream-side or ‘wetland’ vegetation may differ in its leaf margin response to forests elsewhere in that landscape.