Estimation of mean annual temperature from leaf and wood physiognomy

Abstract

There are distinct relationships among the anatomy of wood, the morphology of leaves, and the climate in which woody plants are growing. The relationships between leaf morphological characters and climate have been known for many years, but wood characters as climate indicators are less well studied. In this article, we use measurements of wood anatomy and leaf morphology from woody dicotyledonous plants, growing in Florida and Connecticut, to determine the accuracy to which statistical models can predict climate. The strength of the relationship between climate and physiognomy is important because it allows us to evaluate the phenotypic plasticity that woody plants express under various climates. In this study we use canonical correspondence and regression models to examine how precisely wood anatomical and leaf morphological characters are related to climate. For leaves, canonical correspondence analysis (CCA) using 31 characters gave the closest estimate of mean annual temperature (MAT) in Connecticut, whereas a regression equation using only a single leaf character (leaf margins with no teeth) as predictor gave the closest estimate in Florida. For wood, CCA using 13 wood characters gave the closest estimate in Florida, whereas a regression equation using only a single wood character (the occurrence of vessels smaller than 100 pm) gave the closest estimate in Connecticut. CCA showed that, although MAT has a smooth and continuous relationship with leaf physiognomy, this is not the case for wood. Temperate woods form a different physiognomic population than subtropical and tropical woods, in which the physiognomy of temperate woods is more strongly influenced by MAT than is the physiognomy of subtropical and tropical woods.