Experimental evidence for species-dependent responses in leaf shape to temperature: Implications for paleoclimate inference

Abstract

In many woody dicot plant species, colder temperatures correlate with a greater degree of leaf dissection and with larger and more abundant leaf teeth (the serrated edges along margins). The measurement of site-mean characteristics of leaf size and shape (physiognomy), including leaf dissection and tooth morphology, has been an important paleoclimate tool for over a century. These physiognomic-based climate proxies require that all woody dicot plants at a site, regardless of species, change their leaf shape rapidly and predictably in response to temperature. Here we experimentally test these assumptions by growing five woody species in growth cabinets under two temperatures (17 and 25°C). In keeping with global site-based patterns, plants tend to develop more dissected leaves with more abundant and larger leaf teeth in the cool treatment. Overall, this upholds the assumption that leaf shape responds in a particular direction to temperature change. The assumption that leaf shape variables respond to temperature in the same way regardless of species did not hold because the responses varied by species. Leaf physiognomic models for inferring paleoclimate should take into account these species-specific responses.

Fig 1. Sensitivity of tooth abundance variables to temperature.

(A-C) Comparisons of the species grown in the growth cabinet experiment for tooth abundance variables. Values are means ± 1 s.e.m. * indicates a significant difference (P < 0.05) based on estimated marginal means (see Materials and Methods). Global site-based patterns [6] would predict higher values in the cool treatment for all variables. (D) Representative leaves from species in this experiment that show significant differences in tooth abundance between the two treatments. Scale bar = 1 cm.

Fig 2. Sensitivity of tooth size variables to temperature.

(A-D) Comparisons of the species grown in the growth cabinet experiment for tooth size variables. Values are means ± 1 s.e.m. * indicates a significant difference (P < 0.05) based on estimated marginal means (see Materials and Methods). Global site-based patterns [6] would predict higher values in the cool treatment for all variables. (E) Representative leaves from species in this experiment that show significant differences in tooth size between the two treatments. Scale bar = 1 cm.

Fig 3. Sensitivity of leaf dissection variables to temperature.

(A-D) Comparisons of the species grown in the growth cabinet experiment for leaf dissection variables. Values are means ± 1 s.e.m. * indicates a significant difference (P < 0.05) based on estimated marginal means (see Materials and Methods). Global site-based patterns [6] would predict higher values in the cool treatment for perimeter ratio, feret diameter ratio, and fractal dimension; and lower values for circularity. (E) Representative leaves from species in this experiment that show significant differences in leaf dissection between the two treatments. Scale bar = 1 cm.