A MODEL FOR CALCULATING NITROGEN FLUXES IN SOIL USING 15 N TRACING B. MARY, 1 * S. RECOUS 1 and D. ROBIN 2 1 INRA Unite d'Agronomie, rue Fernand Christ, 02007 Laon Cedex, France and 2 SCPA, 68700 Aspach-le-Bas, France (Accepted 8 April 1998) SummaryÐPublished methods for calculating gross N rates dier in their assumptions and the method of calculation (algebrical equations or numerical methods). The calculation model presented here called FLUAZ considers the major N processes occurring in soil and enables testing of the importance of the various assumptions. It combines a numerical model for solving the mass balance equations and a non linear ®tting program for optimizing the N rate parameters. It can be applied to a single or ``paired'' treatment(s) of an experiment in a bare soil. The model has been evaluated in two experiments made in the laboratory with wheat straw, each experiment involving two ``paired'' treatments. When FLUAZ was applied to the ``paired'' treatments, a good ®t was obtained between the simulated and measured values of 10 variables (amount of NH 4 + and NO 3  , isotopic excess of NH 4 + , NO 3  and organic N). This ®t validated the compartmental model and enabled calculation of six N ¯uxes: mineralisation (m), am- monium immobilisation (i a ), nitrate immobilisation (i n ), nitri®cation (n), volatilisation (v) or denitri®ca- tion (d) and remineralisation of recently immobilised N (r). Sensitivity analysis indicated that the classical assumptions of exclusive ammonium immobilisation (i n =0) and absence of N remineralisation (r = 0) had to be rejected. NH 4 + immobilisation appeared to be dominant when ammonium and nitrate were both present, but was not exclusive: a Langmuir-type relationship could be established between the immobilisation ratio i a /(i a +i n ) and the molar ratio of soil N concentrations NH 4 + /(NH 4 + +NO 3  ). Remineralisation of N occurred simultaneously with immobilisation during wheat straw decomposition and represented 7±18% of gross immobilisation. Taking into account small gaseous losses, volatilisation or denitri®cation, allowed a better ®t to be obtained between observed and simulated N and 15 N pools. Nitri®cation was better described by ®rst order than by zero order kinetics. The eventuality of direct assimilation of organic N by microbial biomass or N humi®cation could not be determined but had no signi®cant in¯uence on the calculation of other ¯uxes. When FLUAZ was applied to a single treatment (NH 4 + labelled), it also gave a good ®t but only m, i (=i a +i n ), n, v or d could be determined. The min- eralisation and immobilisation rates were slightly lower than those found with the paired treatments: this dierence was mainly due to the hypothesis r = 0 and disappeared when r was ®xed at the value obtained with the paired treatments. The ``apparent'' immobilisation rates (i  r) were then similar. The model is very useful to test the consistency of measurements, estimate several N rates simultaneously and quantify the importance of various assumptions. # 1998 Elsevier Science Ltd. All rights reserved INTRODUCTION: REVIEW OF CALCULATION METHODS Calculating the rates of gross N transformations in soil, particularly mineralisation and immobilisation, is essential to improve our understanding of N dynamics in soil and evaluate the concepts intro- duced in simulation models describing C and N transformations. However, the methods proposed in the literature to calculate gross N rates in soil using 15 N tracing dier by several aspects: the calculation method itself, the modelled system, the measured variables and the N rates which are determined. The main features of the methods dealing with N rates in aerobic soils (excluding marine sediments) are summarised in Table 1. In the pioneer work of Kirkham and Bartholomew (1954), the system modelled con- sidered only two pools, i.e. the mineral and organic N fractions. These authors showed that measuring mineral N and mineral 15 N vs time enabled determi- nation of the mineralisation rate m, using the isoto- pic dilution principle. In this work, the immobilisation rate i was deduced from a N bal- ance on the mineral N pool, assuming that no gas- eous loss took place either as ammonia volatilisation or denitri®cation and that reminerali- sation (r) of recently-labelled immobilised N was negligible. In a second paper, Kirkham and Bartholomew (1955) removed this latter assumption and gave a new analytical formulation of their biphasic system to calculate both the mineralisation rate m (in fact rate m + r) and the immobilisation rate i, supposing either m and i constant with time or m and i varying as ®rst order rates. The measure- ment of total organic N and organic 15 N is then required to use the equations derived by the Soil Biol. Biochem. Vol. 30, No. 14, pp. 1963±1979, 1998 # 1998 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0038-0717/98/$19.00+0.00 PII: S0038-0717(98)00068-6 *Author for correspondence. E-mail: mary@laon.inra.fr. 1963