Low phosphorus (P) supply frequently has been shown to limit the abundance and activity of nitrogen (N)-fixing organisms, potentially constraining N inputs to ecosystems. Previous research in a montane Hawaiian forest has shown that ground-level P-fertilization led to significant increases in the population size of epiphytic N-fixing lichens (cyanolichens), as well as a shift in community composition from crustose to leafy species. In this study, we ask whether these changes in the cyanolichen community have resulted in increased N inputs to the forest, and also whether the very high levels of P in the canopy of P-fertilized forest stimulate individual lichen fixation rates over those of lichens from a nearby unfertilized reference forest. We used acetylene reduction (AR) assays to measure the fixation rates of 14 cyanolichen species from P-fertilized forest, and calibrated these rates by measuring N-15(2) fixation incorporation in four species. We found that the ratio of acetylene reduced to N fixed ranged from 2.4 +/- 0.4 in Pseudocyphellaria crocata to 9.3 +/- 2.4 in Leptogium denticulatum. Nitrogen fixation rates in the P-fertilized forest ranged from 0.64 +/- 0.05 nmol N cm(-2) h(-1) in Nephroma helveticton to 3.97 +/- 1.48 nmol N cm(-2) h(-1) in Parmeliella nigrocincta. Fixation rates did not vary greatly among species from P-fertilized forest. We compared these P-fertilized rates to those of 10 species from the reference forest, and found that mass-based fixation rates of P-fertilized lichens were not greater than those of lichens from the unfertilized forest. Using the measured AR rates, we estimate that the P additions increase cyanolichen N inputs to the forest 30-fold, from similar to 0.3 kg N ha(-1) y(-1) to similar to 9 kg N ha(-1) y(-1). These results suggest that P additions to this ecosystem increase N inputs primarily by increasing the abundance of cyanolichens, and that shifts in cyanolichen community composition and changes in individual fixation rate were of lesser importance in determining ecosystem N inputs.