Predicting daytime carbon isotope ratios of atmospheric CO2 within forest canopies

1. While measurements of leaf carbon isotope ratios (C-13/C-12) in terrestrial ecosystems have become more frequent, interpreting these data can remain a challenge in well developed canopies: the variation in leaf delta(13)C (delta(13)C(p)) values is influenced by both the variation in delta(13)C of source air (delta(13)C(a)) and by photosynthetic carbon isotope discrimination (Delta). However, source air information is often unavailable, limiting the interpretation of delta(13)C(p), particularly in dense stands. 2. In this synthesis we found that about 70% of the observed variation in delta(13)C(p) values within the canopy was influenced by changes in Delta, and that about 30% was determined by source air effects. Significant shifts in delta(13)C(a) occur in canopies with high leaf area, predominantly within 1 m above the forest floor. In complex canopies, particularly in the understorey, source air effects cannot be neglected if delta(13)C(p) measurements are used to calculate Delta and c(i)/c(a) ratios [ratio of internal CO2 concentration in the mesophyll air-spaces (c(i)) to the ambient atmospheric concentration of CO2 (c(a))]. 3. We modelled delta(13)C(a) of daytime source air for deciduous and coniferous forests in boreal, temperate and tropical biomes. An inverse regression model with easily available input variables accounted for about 90% of the variation in daytime delta(13)C(a) values throughout the canopy. 4. In open canopies with leaf area index (L) of <2.5 or at canopy heights >= 1 m, the within-canopy daytime delta C-13(a) differences are negligible, and variations in delta C-13(p) are associated primarily with changes in Delta. Then, one can use the easily available carbon isotope ratio of the troposphere (delta C-13(trop)) as a substitute for delta C-13(a) to calculate Delta to within +/- 0.4%. 5. In canopies with L values >2.5, and at canopy heights <1 in, our model is recommended for calculating canopy delta C-13(a) values when direct measurements are not feasible. Although delta C-13(a) is highly variable near the forest floor in those dense forests, our model is more accurate and precise for estimating delta C-13(a) within 1 in above the forest floor than using delta C-13(trop) throughout the canopy (-0.2% +/- 1.5 versus -1.4% +/- 1.1).