15N in tree rings as a bio-indicator of changing nitrogen cycling in tropical forests: an evaluation at three sites using two sampling methods

摘要

Anthropogenic nitrogen deposition is currently causing a more than twofold increase of reactive nitrogen input over large areas in the tropics. Elevated~(15)N abundance (δ~(15)N) in the growth rings of some tropical trees has been hypothesized to reflect an increased leaching of~(15)N-depleted nitrate from the soil, following anthropogenic nitrogen deposition over the last decades. To find further evidence for altered nitrogen cycling in tropical forests, we measured long-term δ~(15)N values in trees from Bolivia, Cameroon, and Thailand. We used two different sampling methods. In the first, wood samples were taken in a conventional way: from the pith to the bark across the stem of 28 large trees (the “radial” method). In the second, δ~(15)N values were compared across a fixed diameter (the “fixed-diameter” method). We sampled 400 trees that differed widely in size, but measured δ~(15)N in the stem around the same diameter (20 cm dbh) in all trees. As a result, the growth rings formed around this diameter differed in age and allowed a comparison of δ~(15)N values over time with an explicit control for potential size-effects on δ~(15)N values. We found a significant increase of tree-ring δ~(15)N across the stem radius of large trees from Bolivia and Cameroon, but no change in tree-ring δ~(15)N values over time was found in any of the study sites when controlling for tree size. This suggests that radial trends of δ~(15)N values within trees reflect tree ontogeny (size development). However, for the trees from Cameroon and Thailand, a low statistical power in the fixed-diameter method prevents to conclude this with high certainty. For the trees from Bolivia, statistical power in the fixed-diameter method was high, showing that the temporal trend in tree-ring δ~(15)N values in the radial method is primarily caused by tree ontogeny and unlikely by a change in nitrogen cycling. We therefore stress to account for tree size before tree-ring δ~(15)N values can be properly interpreted.