Testing a Nitrogen-Cycling Model of a Forest Stream by Using a Nitrogen-15 Tracer Addition Robert O. Hall, Jr., 1 * Bruce J. Peterson, 2 and Judy L. Meyer 1 1 Institute of Ecology, University of Georgia, Athens, Georgia 30602, and Institute of Ecosystem Studies, Box AB, Millbrook, NY 12545; and 2 Ecosystems Center, Marine Biological Laboratory, Woods Hole, Massachusetts 02543, USA A BSTRACT Cycling of nitrogen (N) is commonly studied in aquatic ecosystems; however, most studies examine only parts of the N cycle, such as budgets, N uptake lengths, or oxidative transformations. To integrate conceptually and experimentally several aspects of the N cycle in a stream, we combined a N-cycling model and a tracer addition of nitrogen-15 ( 15 N) to Hugh White Creek, a second-order forested moun- tain stream in North Carolina (USA). We calibrated a steady-state box model for N cycling in 5-m stream segments that included dissolved, detrital, and biotic compartments. This model was parameterized based on prior studies and used to predict the expected distribution of tracer 15 N in all compartments through both time and distance downstream of the addition site. We tested the model results with a 23-day continuous addition of 15 N-NH 4 + to the stream. Deviations of field data from model predic- tions suggested areas in which we lacked under- standing of the N cycle. Downstream distribution of 15 N in epilithon and moss matched model predic- tions, indicating that our prior estimations of N uptake rates were correct. Leaves and fine detritus contained less label than predicted by the model, yet their consumers had both higher  15 N than pre- dicted and higher  15 N than the detritus itself, suggesting selective assimilation of microbial N from ingested detritus. Splitting fine benthic organic N (FBON) into a microbial and recalcitrant pool gave better predictions of FBON and seston  15 N values relative to field data, yet overestimated invertebrate consumer  15 N possibly because our estimates of the fraction of invertebrate N derived from microbes were too high. We predicted that much of the labeled N would move downstream via FBON sus- pension and transport. We found that most of the 15 N remained near the addition site 33 days after the addition was stopped, suggesting that the stream is highly retentive of particulate N. Key words: nitrogen cycling; macroinvertebrates; stream; nitrogen-15; tracer; model; detritivory; Coweeta Hydrologic Laboratory (North Carolina). INTRODUCTION Nitrogen (N) is a commonly studied element in aquatic ecosystems because it often limits productiv- ity in marine systems, streams, and lakes (Grimm and Fisher 1986; Howarth 1988; Elser and others 1990). Humans are currently impacting the global N cycle: anthropogenic N fixation exceeds natural N fixation, which can alter primary and secondary productivity in a range of aquatic ecosystems (Vi- tousek and others 1997). Most studies of N cycling in an ecosystem examine a portion of the entire N cycle, such as oxidative transformations, biotic N uptake, or N budgets. These individual process studies, in themselves, cannot conceptually link all or most of the important controls and fluxes of N in an ecosystem. On the other hand, mechanistic models can potentially link disparate aspects of N cycling. Until models are tested, however, we do not know whether our conceptual integration is correct. The goal of this study was to test a model of N Received 7 October 1997; accepted 13 February 1998. *Corresponding author’s current address: Institute of Ecosystem Studies, Mill- brook, Box AB, New York 12545, USA. e-mail: hallro@aol.com Ecosystems (1998) 1: 283–298 E CO SYSTEM S  1998 Springer-Verlag 283