Leaf and wood carbon isotope ratios, specific leaf areas and wood growth of Eucalyptus species across a rainfall gradient in Australia

Leaves and samples of recent wood of Eucalyptus species were collected along a rainfall gradient parallel to the coast of Western Australia between Perth in the north and Walpole in the south and along a southwest to northeast transect from Walpole in southwestern Australia, to near Mount Olga in central Australia. The collection included 65 species of Eucalyptus sampled at 73 sites and many of the species were collected at several sites along the rainfall gradient. Specific leaf area (SLA) and isotopic ratio of 13C to 12C (delta13C) of leaves that grew in 2002, and tree ring growth and delta13C of individual cell layers of the wood were measured. Rainfall data were obtained from the Australian Bureau of Meteorology for 29 locations that represented one or a few closely located collection sites. Site-averaged data and species-specific values of delta13C decreased with decreasing annual rainfall between 1200 and 300 mm at a rate of 1.63 per thousand per 1000 mm decrease in rainfall. Responses became variable in the low rainfall region (< 300 mm), with some species showing decreasing delta13C with rainfall, whereas delta13C increased or remained constant in other species. The range of delta13C values in the low rainfall region was as large as the range observed at sites receiving > 300 mm of annual rainfall. Specific leaf area varied between 2 and 6 m2 kg(-1) and tended to increase with decreasing annual rainfall in some species, but not all, whereas delta13C decreased with SLA. The relationship between delta13C and SLA was highly species and soil-type specific. Leaf-area-based nitrogen (N) content varied between 2 and almost 6 g m(-2) and decreased with rainfall. Thus, thicker leaves were associated with higher N content and this compensated for the effect of drought on delta13C. Nitrogen content was also related to soil type and species identity. Based on a linear mixed model, statistical analysis of the whole data set showed that 27% of the variation in delta13C was associated with changes in SLA, 16% with soil type and only 1% with rainfall. Additionally, 21% was associated with species identity. For a subset of sites with > 300 mm rainfall, 43% of the variation was explained by SLA, 13% by soil type and only 3% by rainfall. The species effect decreased to 9% because there were fewer species in the subset of sites. The small effect of rainfall on delta13C was further supported by a path analysis that yielded a standardized path coefficient of 0.38 for the effect of rainfall on SLA and -0.50 for the effect of SLA on delta13C, but an insignificantly low standardized path coefficient of -0.05 for the direct effect of rainfall on delta13C. Thus, in contrast to our hypothesis that delta13C decreases with rainfall independent of soil type and species, we detected no statistically significant relationship between rainfall and delta13C in leaves of trees growing at sites receiving < 300 mm of rainfall annually. Rainfall affected delta13C indirectly through soil type (a surrogate for water-holding capacity) across the rainfall gradient. Annual tree rings are not clearly visible in evergreen Eucalyptus species, even in the seasonally cool climate of SW Australia. Generally, visible density transitions in the wood are related not to a strict annual cycle but to periods of growth associated mainly with rainfall. The relationship between delta13C of leaves and the width of these stem increments was not statistically significant. Analysis of stem growth periods showed that delta13C in wood responded to rainfall events, but carbohydrate storage and reallocation also affected the isotopic signature. Although delta13C in wood of any one species varied over a range of 2 to 4 per thousand, there was a general relationship between delta13C of the leaves and the annual range of delta13C in wood. We conclude that species-specific traits are important in understanding the response of Eucalyptus to rainfall and that the diversity of the genus may reflect its response to the large climatic gradient in Australia and to the large annual and interannual variations in rainfall at any one location.