Is southern Africa different? An investigation of the relationship between leaf physiognomy and climate in southern African mesic vegetation

Abstract

Leaves from 24 South African vegetation sites, including 3 Fynbos sites, exhibiting high levels of endemism, were assessed by both LMA (Leaf Margin Analysis) and CLAMP (Climate Leaf Multivariate Program) to determine the effect of endemism on these palaeoclimate proxies. We examined whether existing calibrations using either locally recorded climate data or globally gridded climate data are appropriate for South Africa, or whether new calibrations specific to the region provide more accurate results. The results suggest that calibrations using gridded data yield slightly cooler estimates using LMA when the percentage of entire margined species is over 25. Overall, however, the differences are small and both gridded and local climate station calibration data can be used with equal accuracy. Xeric sites differ from mesic sites regarding the relationship between leaf margin proportion (LMP) and MAT, but with the exception of Fynbos sites, the differences are small when the percentage of entire margined species is < 65. Fynbos sites plot up to 8 °C cooler for a given leaf margin percentage (LMP) than do other sites. LMA was also sensitive to the number of taxa scored and in general should not be attempted on less than 15 leaf morphotypes per site. For CLAMP, the African non-Fynbos sites plotted within the parameters characterized by the existing PHYSG3BRC data set, indicating that they had a leaf physiognomic response to climate consistent with that of the rest of the world. However, the African non-Fynbos sites did fill a previously unoccupied void within that space, indicating the existence of a regional variance from the global pattern. This suggests that endemism per se does not prevent CLAMP from yielding reliable climate predictions. The inclusion of African non-Fynbos sites into the calibration improved the ability of CLAMP to predict Fynbos site climate, although this remained poor for non-coastal xeromorphic Fynbos vegetation. The addition of the African non-Fynbos vegetation did not degrade significantly the PHYSG3BRC calibration, particularly regarding key climatic variables such as enthalpy. Enthalpy appears to be a particularly robust variable with which to test climate model performance against CLAMP palaeoclimate predictions.