The impact of climatic changes that occurred during the last glacial maximum and the Holocene on vegetation changes in the Amazon Basin and the Guiana Shield are still widely debated. The aim of our study was to investigate whether major changes in vegetation (i.e. transitions between rainforests and C4 savannas) occurred in northern French Guiana during the Holocene. We measured variations in the ä13C of soil organic matter at eight sites now occupied by forest or savannah. The forest sites were selected to cover two regions (forest refugia and peneplains) which are thought to have experienced different intensities of disturbance during the latest Pleistocene and the Holocene. We found that none of the forest sites underwent major disturbances during the Holocene, i.e. they were not replaced by C4 savannahs or C4 forest savannahs for long periods. Our results thus suggest that tropical rainforests in northern French Guiana were resilient to drier climatic conditions during the Holocene. Nevertheless, geographical and vertical variations in the 13C of SOM were compatible with minor changes in vegetation, variations in soil processes or in soil physical properties.

Despite the progress made in understanding the ecophysiology of tropical plants during the past two decades (Lüttge 1997, Mulkey et al. 1996), questions regarding relationships between the environment and physiological diversity remain. It is now recognized that tropical climate can be quite variable (see Coen 1983) which could lead to significant functional diversity (increased variation in life history traits) among species due to the tight association between gas exchange physiology and the environment (see Enquist & Leffler 2001, Guehl et al. 1998, Huc et al. 1994, Martinelli et al. 1998, Sobrado 1993). It remains unclear, however, how the subtleties of variation in tropical climate and tree life history traits are related to the functional diversity of tropical communities (Borchert 1994, 1998).

Hakea psilorrhyncha R. M. Barker seedlings were subjected to a two-way interaction experiment with two levels of water availability and two levels of light. Physiological drought was imposed by adding 6000 MW polyethylene glycol (PEG), at an osmotic potential of −0·56 MPa, in a continuous-flow drip irrigation system. Unstressed plants (-PEG) were watered with distilled water (osmotic potential of −0·01 MPa). Seedlings were grown under natural light (120–500 μmol m−2 s−1 sunlight at midday +L) or shaded (<150 μmol m−2 s−1 sunlight at midday; −L. Plant morphology tissue water relations and carbon isotope composition (δ13C) were measured after 12 wk growth. The smallest leaves had the highest mass per unit area (LMA, an index of sclerophylly) and were present in the +PEG+L treatment, whereas the largest, thickest leaves had the lowest density and were produced in the −PEG+L treatment. Plants in shade were smaller and less sclerophyllous. There was a linear decrease in osmotic potential at full turgor and turgor loss point, and an increase in elastic modulus and δ13C, as level of sclerophylly increased, with −PEG−L leaves at one extreme and +PEG+L at the other. We conclude that high levels of sclerophylly induced by low water availability and high light intensity are associated with substantial drought tolerance in H. psilorrhyncha.