Limitations of the N-15 natural abundance method for estimating biological nitrogen fixation in Arnazonian forest legumes

This study evaluates the N-15 natural abundance method for estimating biologicalnitrogen fixation (BNF) in 12 secondary and five primary forest sites in central Amazonia. Investigations were based on spatially systematic sampling of 20 potentially N-2-fixing legume tree and liana species, and of a range of differing references representing the 'soil-derived nitrogen'. The absence of systematic differences in the delta(15)N-signals of potential N-2-fixers and reference plants resulted in many invalid % Ndfa-estimates (i.e., the proportion of foliar N derived from BNF). Whereas the % Ndfa-estimates based on non-legumes were randomly distributed around zero, the use of non-N-2-fixing legumes resulted in Largely positive % Ndfa-estimates. Grouping of reference species into species types was a valid strategy to reduce the risk of atypical values. Leaf litter was an inadequate measure of soil-derived nitrogen. The impact of the B-value was low or moderate at best, the percentage of negative % Ndfa-estimates remained unaltered. The manner of individually pairing putative N2-fixers with reference plants only marginally affected the % Ndfa-estimates, the assumption of an overall, site mean reference delta(15)N-signal was sufficient. Since legumes were aggregated in clusters, BNF locally affected the surrounding soil nitrogen in secondary regrowth. Negative correlations between the vegetation shares of potential N-2-fixers and leaf litter delta(15) N-signals indicate N-15-dilution from the input side. Interpolated reference plant 815 N-signals appear low in the surroundings of legume clusters in some of the older regrowth sites, confirming such N-15-dilution. We conclude that, with our current state of knowledge, the N-15 natural abundance method fails to quantitatively estimate BNF in heterogeneous forests. Future research should focus on the search for a more direct measure of the plant-available delta(15)N-signal of the soil, as well as on plant-soil interactions and the resulting small-scale isotopic heterogeneity. (C) 2004 Elsevier GmbH. All rights reserved.