Foliar δ13C within a temperate deciduous forest: spatial, temporal, and species sources of variation.

Foliar 13C-abundance (δ13C) was analyzed in the dominant trees of a temperate deciduous forest in east Tennessee (Walker Branch Watershed) to investigate the variation in foliar δ13C as a function of time (within-year and between years), space (canopy height, watershed topography and habitat) and species (deciduous and coniferous taxa). Various hypotheses were tested by analyzing (i) samples collected from the field during the growing season and (ii) foliar tissues maintained in an archived collection. The δ13C-value for leaves from the tops of trees was 2 to 3%. more positive than for leaves sampled at lower heights in the canopy. Quercus prinus leaves sampled just prior to autumn leaf fall had significantly more negative δ13C-values than those sampled during midsummer. On the more xeric ridges, needles of Pinus spp. had more positive δ13C-values than leaves from deciduous species. Foliar δ13C-values differed significantly as a function of topography. Deciduous leaves from xeric sites (ridges and slopes) had more positive δ13C-values than those from mesic (riparian and cove) environments. On the more xeric sites, foliar δ13C was significantly more positive in 1988 (a dry year) relative to that in 1989 (a year with above-normal precipitation). In contrast, leaf δ13C in trees from mesic valley bottoms did not differ significantly among years with disparate precipitation. Patterns in foliar δ13C indicated a higher ratio of net CO2 assimilation to transpiration (A/E) for trees in more xeric versus mesic habitats, and for trees in xeric habitats during years of drought versus years of normal precipitation. However, A/E (units of mmol CO2 fixed/mol H2O transpired) calculated on the basis of δ13C-values for leaves from the more xeric sites was higher in a wet year (6.6±1.2) versus a dry year (3.4±0.4). This difference was attributed to higher transpiration (and therefore lower A/E) in the year with lower relative humidity and higher average daily temperature. The calculated A/E values for the forest in 1988-89, based on δ13C, were within ±55% of estimates made over a 17 day period at this site in 1984 using micrometeorological methods.