Overestimation of gross N transformation rates in grassland soils due to
non-uniform exploitation of applied and native pools
C.J. Watson
a,*, G. Travers
a,b, D.J. Kilpatrick
c, A.S. Laidlaw
d, E. O'Riordan
baAgricultural and Environmental Science Division, Department of Agriculture for Northern Ireland, Newforge Lane, Belfast BT9 5PX and School of Agriculture and Food Science, The Queen's University of Belfast, Newforge Lane, Belfast, Ireland
bTeagasc, National Beef Research Centre, Grange, Dunsany, County Meath, Ireland
cBiometrics Division, Department of Agriculture for Northern Ireland, Newforge Lane, Belfast BT9 5PX and School of Agriculture and Food Science, The Queen's University of Belfast, Newforge Lane, Belfast, Ireland
dApplied Plant Science Division, Department of Agriculture for Northern Ireland, Newforge Lane, Belfast BT9 5PX and School of Agriculture and Food Science, The Queen's University of Belfast, Newforge Lane, Belfast, Ireland
Accepted 3 May 2000
Abstract
The study tested the validity of some of the assumptions in the15N pool dilution technique in short-term soil incubations. Microbial N
transformation rates were calculated using15N pool dilution during 24 h in four grassland soils in April 1998. The change in concentration and enrichment of the NH41-N and NO32-N pools was determined at 0, 1.5, 4, 10, 16 and 24 h following application of differentially15N
labelled NH4NO3in solution at a rate of either 2 or 15 mg N kg21oven-dry soil and at an enrichment of 99.8 atom% excess. Rapid15N pool
dilution occurred in all soils. Rates of gross mineralisation and NH41consumption were not constant during the 24 h incubation in contrast to
nitri®cation rates. An application of 15 mg N kg21decreased gross mineralisation and NO32consumption and increased nitri®cation rates
compared to an application of 2 mg N kg21. Applied15NH41was rapidly nitri®ed with up to 55% of the added label recovered as 15
NO32after
24 h. This rapid conversion of15NH
41to15NO32occurred without a proportional and concurrent increase in the size of the unlabelled NO32
pool. Gross and net nitri®cation rates were signi®cantly different due to15NO32consumption. The results suggest that there was non-uniform
exploitation of the14N and15N pools by soil microorganisms, invalidating one of the key assumptions in the15N pool dilution technique.
Preferential consumption of applied NH41and NO32led to an overestimate of gross mineralisation and nitri®cation rates due to the greater rate
of decline of the15N enrichment of the added N pool. In future studies care should be taken to ensure that gross N transformation rates are not
altered by the method used to quantify them.q2000 Elsevier Science Ltd. All rights reserved.
Keywords: Ammonium consumption;15N pool dilution; Grassland soils; Gross nitrogen mineralisation; Nitrate consumption; Nitri®cation
1. Introduction
Net N mineralisation studies provide information on changes in overall nitrogen cycling but, do not give any indication of gross mineralisation±immobilisation turnover (MIT) which can only be studied using isotope techniques. There are several mathematical equations available to calculate gross N transformation rates using the data from experiments with15N (Barraclough, 1991; Bjarnason, 1988; Kirkham and Bartholomew, 1954). These calculations rely on certain key assumptions (Hart et al., 1994) namely: (1) all rate processes can be described by zero-order kinetics over the experimental period; (2) microorganisms do not
discriminate between 14N and 15N; (3) there is uniform mixing of added label with the soil inorganic N pool; and (4) labelled N immobilised over the experimental period is not remineralised.
Few15N pool dilution experiments have been undertaken in grassland soils over short (,3 d) time periods. Indigen-ous process rates can be studied by adding small concentra-tions of highly-enriched15NH41or15NO32to soil. Gross rates
of N mineralisation (NH41production) and NH41
consump-tion (immobilisaconsump-tion and nitri®caconsump-tion) can be calculated from the rate of dilution in 15N enrichment of the NH41
pool as organic 14N is mineralised to 14NH41and from the
change in the size of the total NH41pool. Gross nitri®cation
and NO32consumption are determined in a similar manner
with15NO32being applied to soil. As one of the assumptions
in the pool dilution technique is that microorganisms do not discriminate between14N and15N, consumption of NH41-N
0038-0717/00/$ - see front matterq2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 3 8 - 0 7 1 7 ( 0 0 ) 0 0 1 0 3 - 6
www.elsevier.com/locate/soilbio
* Corresponding author. Tel: 144-28-90-255359; fax: 1 44-28-90-662007.
and NO32-N will change the pool sizes but will not affect the 15
N enrichment allowing gross rates of production of NH41
and NO32 to be separated from concurrent consumption
rates. Gross immobilisation is the difference between NH41
consumption and nitri®cation.
A further key assumption in calculating gross N transfor-mation rates is that there is uniform mixing of added 15N label with the soil inorganic N pool. However, this is dif®-cult to achieve because ambient inorganic 14N is not uniformly distributed in soil (Hart et al., 1994). Any prefer-ential use of applied N by soil microorganisms would result in erroneously high gross MIT rates, due to the greater rate of decline of the 15N enrichment of the added N pool. High gross mineralisation and immobilisation rates have been reported in a range of soils (Bjarnason, 1988; Davidson et al., 1991; Schimel, 1986; Watson and Mills, 1998). The current study was undertaken to establish if there was any preferential use of applied N in short-term incubations by measuring 15N pool dilution at six time intervals during a 24 h period having applied highly enriched 15N to four grassland soils with different management histories. In addi-tion, the rates of microbial N transformations were calcu-lated to determine whether they were constant during the incubation period.
2. Materials and methods
2.1. Site characteristics
Samples from four grassland soils were collected in April 1998 from the Central Nitrogen Experimental Site (CENIT) at the Agricultural Research Institute for Northern Ireland (ARINI), Hillsborough, Co. Down and the Teagasc, Grange Research Centre, Co. Meath. The CENIT grassland site at ARINI was established in 1987 on a relatively free draining clay-loam soil. The grass sward received an annual input of 300 kg N ha21, applied in six equal dressings between April and August and was continuously grazed by beef steers from April to October to maintain a constant sward height of 7 cm. The three grassland swards at Grange Research Centre were established in 1994, on a moderately well drained Brown Earth soil and consisted of two grass swards which were cut at 4-week intervals between April and September. Nitrogen fertiliser was applied to one sward after each cut giving a total N application of 300 kg ha21y21, while the other cut sward did not receive
any N fertiliser. The remaining grass sward at Grange Research Centre was a rotationally grazed grass/clover sward (21 d cycle; beef steers) which received no N fertili-ser. All swards received a single annual application of P and K, according to soil analysis and standard recommenda-tions. Selected soil properties (average of three replicates) are given in Table 1.
2.2. Sampling and incubation procedure
Prior to N fertiliser application, soil cores (2.5 cm diameter£5 cm deep) were collected randomly and bulked for each of 3 replicate swards of the 4 soils. The freshly collected soil was coarsely sieved through a 6.7 mm sieve to remove large root and shoot material. Fresh soil (equivalent to 72 g on an oven-dry weight basis) was weighed into 500 cm3Kilner jars and acclimatised in a controlled envir-onment cabinet at 13.58C for 24 h. The surface area of exposed soil was 81 cm2.
Differentially15N labelled NH4NO3was applied at a rate
of either 2 or 15 mg15N kg21oven-dry soil to allow paired soil incubations. Half of the Kilner jars received15NH4NO3
and the other half received NH415NO3, each of the labelled
moieties being at an enrichment of 99.8 at% excess. The15N labelled substrates were applied uniformly over the soil surface in a solution (10 ml) using a ®ne tipped pipette. The average moisture content of the soils was 30% (g g21) initially and increased to 40% (g g21) after substrate addi-tion.
The jars were sealed with glass lids and incubated in a temperature controlled cabinet at 13.58C in the dark. This temperature was selected as it was the mean soil tempera-ture at the Grange Research Centre at a depth of 5 cm during the period April to September 1997. The soil in the jars was destructively sampled at 0, 1.5, 4, 10, 16 and 24 h. There were 288 jars in total (4 soils£2 labels£2 concentrations£6 times£3 replicates).
2.3. Chemical analysis
At each sample time 3 replicates per treatment were destructively harvested. The soil in the jars was shaken with 200 ml of 2 M KCl for 1 h and ®ltered (Whatman GF/C). The NH41-N and NO32-N concentration in the ®ltrate
was determined using a Technicon Random Access Auto-mated Chemistry System (TRAACS 8001) (Bran and Luebbe, 1995) and expressed as mg N kg21oven-dry soil. The KCl extracts were stored at 48C and analysed for
C.J. Watson et al. / Soil Biology & Biochemistry 32 (2000) 2019±2030
Table 1
Average physical and chemical characteristics of the soils (5 cm depth)
Grass sward Location N (%) C (%) pH Silt (%) Clay (%) Sand (%)
Grazed grass/clover Grange (Grange GC) 0.39 3.91 6.1 42 23 35
Cut 0 kg N ha21y21 Grange (Grange 0) 0.37 3.76 6.2 42 24 34
Cut 300 kg N ha21y21 Grange (Grange 300) 0.40 4.11 5.9 44 25 31
mineral N within 24 h of extraction and for15N within one week. The extraction at time zero occurred instantaneously after application of the15N label.
Determination of the 15N enrichment of the NO32-N in
the KCl soil extracts was based on the production of N2O from nitrite and hydroxylamine intermediates
during reduction with Cd/Cu (Stevens and Laughlin, 1994). The 15N enrichment of NH41-N in the KCl
extracts was determined by ®rstly generating NH3 by
addition of MgO. The NH3 was absorbed by a CuSO4/
H2SO4 solution, which was later dried to a residue. The
N2O produced as a side reaction on the addition of
sodium hypobromite was analysed by isotope ratio mass spectrometry (Laughlin et al., 1997).
2.4. Calculation of gross mineralisation, consumption and nitri®cation rates
Rates of gross mineralisation, NH41consumption,
nitri®-cation and NO32consumption were calculated for each of
the ®ve possible time periods (0±1.5, 1.5±4, 4±10, 10±16, 16±24 h) using Eqs. (1) and (2) (Kirkham and Bartholo-mew, 1954), separately for each of the three replicates.
m M02M1=tlog H0M1=H1M0=log M0=M1 1
and where m±c: Kirkham and Bartholomew (1954) provided another equation for the condition when mc
(i.e. when the mineral N pool size stays constant with time), which did not occur in this study.
For samples that received15NH41the NH41pool was used
forMandH. For samples that received15NO32the NO32pool
was used forMandHin Eqs. (1) and (2), to give the rate of nitri®cation (mg kg21h21) and NO32consumption,
respec-tively. Gross immobilisation was the difference between NH41consumption and nitri®cation.
The average mineral N concentrations of the NH41-N and
NO32-N labelled moieties were calculated for each of the
three replicates for each soil and concentration at each time. There were three replicates for all determinations of 15N enrichment and calculation of gross N transformation rates. The data were analysed as a split-plot design with treatments as the main plot factor and concentration and time as sub-plot factors. However, two problems were iden-ti®ed with this approach when calculating and analysing
gross N transformation rates:
1. There was considerable variation between the replicates leading to large standard errors for the mean rates. This is a common statistical problem due to the calculation being based on the means of ratios, which tends to produce highly variable results, rather than the more stable ratio of means.
2. The rates ¯uctuated erratically between the various time periods. This was particularly true for the calculation of gross mineralisation. Examination of the data showed that this was due to erratic variability in theM andH
values.
Accordingly another approach was investigated. The ®rst problem was addressed by basing the calculation on the means for M andHover the three replicates. A bootstrap technique was then used to estimate standard errors for the mean rates. As described by Manly (1997), the bootstrap technique allows the distribution of values in a population to be investigated in the absence of any prior knowledge. The method is to repeatedly resample the sampled values and calculate the parameter of interest for each resample. This resampling is done ªwith replacementº i.e. some values may appear two or more times in the resample while others may not appear at all. If a large number of independent resamples are taken, then the overall mean and standard deviation of the parameter provides unbiased estimates of the parameter and its standard error. In relation to the current dataset, the bootstrap resampling procedure involves randomly select-ing a sample of size 18 with replacement from the 18 actual values (6 times £3 replicates) for bothMandH.
The second problem was addressed by ®tting smoothing curves to the meanMandHvalues over time (t). A random bootstrap resample of size 18 was selected as described in the previous paragraph. Three types of curve were ®tted to these values Ð (1) linear ya1bt;(2) exponentialy
a1brt; and (3) spline which does not have a functional
form but corresponds to an iterative mathematical procedure to ®t cubic functions to segments of the curve between adjacent time points constrained to be ªsmoothº at the junc-tions between segments. The Kirkham and Bartholomew (1954) equations were applied to both the original and the ®tted M and H values from each of these three types of curve. This provided estimates of the rates of gross miner-alisation, consumption, immobilisation and nitri®cation both for each time period and for the overall 24 h period. Net nitri®cation was estimated from direct linear regression of the NO32pool size against time. The difference between
the gross nitri®cation rate, calculated from the Kirkham and Bartholomew (1954) equations, and the net nitri®cation rate was also calculated. Net mineralisation was estimated from direct linear regression of the total mineral N pool size against time. The resampling procedure was repeated 1000 times. The means and standard errors of the various rates over these 1000 resamples were calculated. These
were used to compare the rates in successive time periods and also to test whether the difference between gross and net nitri®cation was equal to zero. The difference between daily gross mineralisation and immobilisation should indicate the net production of N. This calculated value was compared with the measured change in total mineral N (netm) over the 24 h incubation.
All random re-sampling and calculations were carried out using the Genstat (1993) statistical package.
3. Results
3.1. Soil mineral N concentrations
Figs. 1 and 2 show the changes in NH41-N and NO32-N,
respectively, during the 24 h incubation for soils that received (a) 2 mg 15N kg21 or (b) 15 mg 15N kg21. When data for all soils were analysed together there was a signi®-cant decrease P,0:001 in NH41-N and a signi®cant
increase P,0:001in NO32-N with time at both N
appli-cations. The NH41-N and NO32-N content of the CENIT soil
was signi®cantly P,0:05greater than the Grange soils at the start of the incubation. The decrease in NH41-N and
increase in NO32-N content was greater when 15 mg
N kg21 was applied than when 2 mg N kg21 was applied. Net NO32-N production after 24 h was signi®cantly P,
0:001greater with the CENIT soil than with the Grange soils.
C.J. Watson et al. / Soil Biology & Biochemistry 32 (2000) 2019±2030
Fig. 1. Change in NH41-N during a 24 h incubation with (a) 2 mg N kg21
and (b) 15 mg N kg21. (soil
£time£concentration sem0.66).VCENIT; BGrange GC;KGrange 0; £ Grange 300.
Fig. 2. Change in NO32-N during a 24 h incubation with (a) 2 mg N kg21and (b) 15 mg N kg21. (soil£time£concentration sem0.35).VCENIT;BGrange
3.2. Atom% excess 15N in NH41-N and NO32-N
Fig. 3 shows the change in atom% excess of (a) 15NH41
and (b) 15NO32 during 24 h when 2 mg 15
N kg21 was applied. There was a highly signi®cant P,0:001decline in both labelled moieties with time; however, the 15NH41
decreased exponentially from an average of 24.0 at% excess at time zero to 1.9 at% excess after 24 h, whereas 15NO32
decreased linearly over the time period from an average of 28.8 at% excess to 14.3 at% excess. When 15 mg 15N kg21 was applied the atom% excess of both 15N moieties decreased in a linear manner (Fig. 4). The rate of decline in atom% excess 15NO3was greater P,0:001at 15 mg 15
N kg21 than at 2 mg 15N kg21, being 0.94 and 0.60at% excess h21, respectively, averaged for all soils. At both application rates there was a highly signi®cant P,
0:001difference between soils and a signi®cant soil£time interaction for both labelled moieties. This was because the CENIT soil had a higher initial NH41-N and NO32-N content
than the other soils, which resulted in a signi®cantly lower atom% excess at time zero.
Fig. 5 shows the signi®cant P,0:001 appearance of
15
NH41 during the experimental period in soils that had
received 15NO3 labelled NH4NO3 at (a) 2 mg N kg21 and
(b) 15 mg N kg21. With 15 mg15N kg21the atom% excess
15
NH41continued to increase over the duration of the study,
however, with 2 mg15N kg21the15NH41peaked at 4 h in the
Grange GC soil and at 1.5 h in the CENIT soil. There was a signi®cantly P,0:001 higher atom% excess 15NH41-N
with 15 mg N kg21than with 2 mg N kg21which, averaged for the duration of the incubation and soils, was 0.63 and
C.J. Watson et al. / Soil Biology & Biochemistry 32 (2000) 2019±2030
Fig. 3. Change in atom% excess of (a)15NH
41(soil£time£concentration sem2.11) and (b)15NO32(soil£time£concentration sem1.08) with an
application of 2 mg N kg21.
VCENIT;BGrange GC;KGrange 0; £ Grange 300.
Fig. 4. Change in atom% excess of (a)15NH 4
1(soil£time£concentration
sem2.11) and (b)15NO 3
2(soil£time£concentration sem1.08) with
an application of 15 mg N kg21.
VCENIT;BGrange GC;KGrange 0; £
0.36 at% excess, respectively. There was a signi®cant P,
0:05difference between soils and a signi®cant P,0:001 soil£time interaction. There was an increase in 15NH41
throughout the incubation in the Grange 0 soil which after 24 h was 0.76 and 1.63 at% excess with 2 and 15 mg N kg21, respectively. The other soils were more variable. When the15NH41-N content was expressed as a % of
15
NO32
-N applied the recovery was small and did not exceed 3.3% when 2 mg 15NO32-N was applied. When 15 mg 15NO3-N
was applied no more than 1.7% of the 15N was recovered as 15NH41-N.
The ®xation of NH41to clay minerals can occur in some
soils. Davidson et al. (1991) suggest that abiological reac-tions occur rapidly and that initial 14N and 15N pool sizes should be adjusted by undertaking a time zero extraction. In
the current study the % recovery of 15NH41and 15NO32 at
time zero was 98.1 and 99.9%, respectively with the CENIT soil when 2 mg N kg21was applied and 102.4 and 95.4% when 15 mg N kg21 (Table 2) was applied. Abiotic NH41
®xation did not occur in the CENIT soil, in contrast to the Grange soils where the recovery of 15NH41 at time zero
averaged 88.1 and 89.4% with an application of 2 and 15 mg N kg21, respectively.
There was a high recovery of 15NO3-N in soils that
received 15NH41-N (Fig. 6). The % recovery was highest
with the CENIT soil and after 24 h was equivalent to 55.2 and 45.1% of the 15NH41-N applied with 2 and 15 mg
N kg21, respectively. Table 2 shows the size of the labelled and unlabelled NH41and NO32moieties at the start and end
of the incubation, when 15NH41 was applied at the rate of
15 mg N kg21. The rapid conversion of 15NH41into 15NO32
occurred in the Grange soils without a concurrent increase in the size of the unlabelled NO32pool. For example, in the
Grange GC soil 69.6% of the NH41at the start (0 h) was15N
labelled. If the 14NH41and 15NH41pools were exploited in
proportion to their size the expected increase in14NO32and 15
NO32pools after 24 h would be 1.34 and 3.06 mg N kg2 1
, respectively. The observed increase in the 14NO32 pool
(0.1 mg N kg21) was considerably lower than expected whereas, the increase in the 15NO32 pool of 4.30 mg
N kg21 was greater than expected. A similar ®nding was observed with the lower rate of N application (results not shown). Although there was an increase in the unlabelled NO32 pool in the CENIT soil after 24 h, the increase in
labelled NO32was proportionately greater.
When 2 mg15NO32-N kg21was applied to the soils there
was a signi®cant P,0:001 decrease in the recovery of
15
NO32(expressed as a % of the time zero value) after 24 h
(Table 3). The % recovery of15NO32was higher P,0:01
in the CENIT soil compared to the Grange soils and was signi®cantly greater P,0:001 at the higher application rate.
3.3. Hourly gross N transformation rates
Gross N transformation rates were calculated at each time using the equations of Kirkham and Bartholomew (1954), having smoothed the data using a curve ®tting procedure and the results were expressed as mg kg21h21. For gross mineralisation and NH41consumption an exponential ®t was
C.J. Watson et al. / Soil Biology & Biochemistry 32 (2000) 2019±2030
Fig. 5. Appearance of 15NH 4
1 in soils that received15NO
32at (a) 2 mg
N kg21and (b) 15 mg N kg21(soil
£time£concentration sem0.12). VCENIT;BGrange GC;KGrange 0; £ Grange 300.
Table 2
Labelled and unlabelled NH41and NO32pool sizes (mg N kg21) at the start (0 h) and end of the incubation (24 h) after linear smoothing when15NH41was
applied at the rate of 15 mg N kg21. Figures in brackets are the standard errors of the means;n3
Pool size (mg N kg21) CENIT Grange GC Grange 0 Grange 300
0 h 24 h 0 h 24 h 0 h 24 h 0 h 24 h
15NH 4
1 15.36 (0.615) 3.46 (0.329) 13.05 (0.585) 5.79 (0.741) 13.48 (0.105) 4.02 (0.724) 13.71 (0.017) 6.23 (0.906) 14NH
4
1 8.04 (0.837) 10.44 (1.668) 5.71 (1.272) 6.65 (1.932) 5.52 (0.323) 11.92 (2.006) 8.00 (0.746) 8.29 (0.619) 15NO
3
2 0.00 6.92 (0.762) 0.00 4.30 (0.556) 0.00 4.54 (0.341) 0.00 3.09 (0.219)
14NO 3
best. Table 4 shows the hourly gross mineralisation rates for the different soils receiving either 2 or 15 mg 15N kg21. Rates varied signi®cantly with time and were generally higher with the low N application than with the high N application. Gross NH41 consumption rates also varied
signi®cantly with time, generally decreasing (Table 5). However, the rate of N application had little or no effect. Gross mineralisation and NH41 consumption rates were
highest in the CENIT soil. The estimate of gross mineralisa-tion and NH41consumption during the incubation obtained
using the zero and 24 h smoothed data from the exponential curve ®t, agreed reasonably well with the values calculated using the raw data at time zero and time 24 h (Tables 4 and 5).
In contrast, gross nitri®cation rates were generally constant with time and were higher when 15 mg N kg21
was applied than when 2 mg N kg21 was applied (Table 6). The gross nitri®cation rate was higher in the CENIT soil than in the other soils. The estimate of hourly gross nitri®cation rate from the linear, exponential and spline smoothing procedures agreed well with each other and with the calculation using the 0 and 24 h raw data. As neither the exponential nor spline smoothing gave a signi®-cantly better ®t than the linear smoothing, only the linear results are shown in Table 6. Nitrate consumption was generally constant with time so only the hourly rates from the linear smoothed data (0±24 h) are shown in Table 7. Nitrate consumption was generally higher when 2 mg N kg21was applied than when 15 mg N kg21was applied and was higher in the Grange soils than in the CENIT soil. The rate of consumption in the CENIT soil receiving 15 mg N kg21was not signi®cantly different from zero (Table 7).
3.4. Daily N transformation rates
Daily gross mineralisation and immobilisation rates are shown in Table 8. Gross immobilisation was calculated as the difference between NH41consumption and gross
nitri®-cation. Daily net mineralisation was determined from linear regression of the change in the total mineral N pool with
C.J. Watson et al. / Soil Biology & Biochemistry 32 (2000) 2019±2030
Fig. 6. Recovery of15NO
32in soils that received15NH41at (a) 2 mg N kg21and (b) 15 mg N kg21(soil£time£concentration sem1.80).VCENIT;B
Grange GC;KGrange 0; £ Grange 300.
Table 3
Percentage recovery of applied15NO32in each soil after 24 h (sem2.31; n3)
N applied (mg N kg21) CENIT Grange GC Grange 0 Grange 300
2 83.1 71.5 56.9 66.9
time. Net mineralisation was greater in the CENIT soil than in the Grange soils where there was little or no net increase in total mineral N during the 24 h incubation (Table 8). The difference between daily gross mineralisation and immobi-lisation should indicate the net production of N. However, this calculated net production did not always agree with the measured change in total mineral N over the 24 h incubation (Table 8), with signi®cant differences found for Grange 0 at both N concentrations and for Grange 300 at the lower N concentration.
Daily net nitri®cation was determined from linear regres-sion of the change in the NO32pool with time. Table 9 shows
that daily gross nitri®cation rates were signi®cantly higher (at leastP,0:01than net nitri®cation rates, except in the CENIT soil receiving 15 mg N kg21, where there was no signi®cant difference.
4. Discussion
The calculation of gross N transformation rates using the
15
N pool dilution technique relies on certain key assump-tions (Hart et al., 1994)
1. All rate processes can be described by zero-order kinetics over the experimental period.
2. Microorganisms do not discriminate between 14N and
15
N.
3. There is uniform mixing of added label with the soil
inorganic N pool.
4. Labelled N immobilised over the experimental period is not remineralised.
These assumptions will be appraised in turn for this study.
4.1. Zero-order kinetics
The current study has shown that gross mineralisation and gross NH41consumption rates cannot be described by
zero-order kinetics during a 24 h incubation. Calculated hourly rates varied with time, although the best estimate of the hourly rate from the smoothed data agreed reasonably well with the hourly rate calculated using the raw data at time zero and 24 h. As the rates of gross mineralisation and NH41 consumption generally decreased with time, rates
calculated over the ®rst 24 h would likely be higher than if a longer incubation interval had been used. The daily rates calculated in this study were considerably higher than other reported studies with grassland soils (Jamieson et al., 1998; Ledgard et al., 1998; Murphy et al., 1999), where15N pool dilution was measured several days after15N application.
Nitrogen transformation rates were also affected by the amount of N applied. The current study has shown that generally an application of 15 mg N kg21 decreased gross mineralisation and NO32consumption and increased
nitri®-cation rates compared to an applinitri®-cation of 2 mg N kg21. Nitri®cation is known to be stimulated by the addition of an NH41-N substrate (Recous et al., 1999; Willison et al.,
C.J. Watson et al. / Soil Biology & Biochemistry 32 (2000) 2019±2030
Table 4
Hourly gross mineralisation rates (mg N kg21h21) calculated after exponential smoothing at different times during a 24 h incubation. Figures in brackets are
the standard errors of the means estimated from the bootstrap technique n3
Time (h) Application rate of 2 mg N kg21 Application rate of 15 mg N kg21
CENIT Grange GC Grange 0 Grange 300 CENIT Grange GC Grange 0 Grange 300
1.5 0.89 (0.433) 1.18 (0.262) 0.84 (0.237) 0.97 (0.148) 0.64 (0.244) 0.61 (0.307) 0.14 (0.148) 0.50 (0.159) 4 1.24 (0.151) 1.06 (0.126) 1.01 (0.112) 0.91 (0.118) 0.53 (0.236) 0.50 (0.211) 0.24 (0.136) 0.47 (0.128) 10 1.22 (0.076) 0.82 (0.075) 0.89 (0.043) 0.76 (0.062) 0.75 (0.089) 0.36 (0.077) 0.53 (0.104) 0.40 (0.075) 16 0.89 (0.085) 0.48 (0.145) 0.45 (0.059) 0.49 (0.072) 0.73 (0.112) 0.16 (0.074) 0.67 (0.048) 0.29 (0.062) 24 0.42 (0.075) 0.30 (0.193) 0.11 (0.038) 0.12 (0.079) 0.63 (0.175) 0.07 (0.052) 0.80 (0.150) 0.13 (0.089) 0±24 h (smoothed) 0.94 (0.061) 0.61 (0.133) 0.57 (0.048) 0.44 (0.050) 0.70 (0.093) 0.19 (0.057) 0.65 (0.045) 0.25 (0.020) 0±24 h (raw) 0.99 (0.042) 0.73 (0.100) 0.65 (0.069) 0.50 (0.034) 0.74 (0.100) 0.26 (0.076) 0.75 (0.053) 0.29 (0.029)
Table 5
Hourly gross NH41consumption rates (mg N kg21h21) calculated after exponential smoothing at different times during a 24 h incubation. Figures in brackets
are the standard errors of the means estimated from the bootstrap technique n3
Time (h) Application rate of 2 mg N kg21 Application rate of 15 mg N kg21
CENIT Grange GC Grange 0 Grange 300 CENIT Grange GC Grange 0 Grange 300
1998). Although nitri®cation rates may have been overesti-mated in the current study, due to the addition of NH41, they
may represent the potential nitrifying activity of the soil. The CENIT soil would appear to have a higher nitrifying potential than the Grange soils, which may re¯ect its previous grazing management. In the case of gross miner-alisation, because the product pool is labelled with 15N rather than the substrate pool, rates of NH41 production
should not be affected by the amount of N applied (Hart et al., 1994). However, this was not the case.
4.2. Isotopic fractionation and uniform mixing
The assumption that microorganisms do not discriminate between14N and15N is not strictly true. Delwiche and Steyn (1970) showed some discrimination in favour of14N in the ®xation of N, the oxidation of NH41 to NO22 and in the
assimilation of NH41. However, the error due to
fractiona-tion during an incubafractiona-tion of a few days is small relative to the large decreases in 15N enrichment of the product pool that occur from production (Hart et al., 1994). The current study suggests that microorganisms exploit the indigenous and applied N pools at different rates. For example 15NH41
was rapidly nitri®ed with 24.5±55% of the added label recovered as 15NO32 after 24 h. This rapid conversion of 15
NH41 to 15NO32 occurred without a concurrent increase
in the size of the unlabelled NO32pool. This suggests that
there was non-uniform mixing of the14N and15N pools. The newly applied15NH41in solution would appear to be more
accessible to nitri®ers compared to indigenous soil NH41
located or produced at microsites. Preferential consumption of applied NH41-N leads to an overestimate of gross N
mineralisation rates due to the greater rate of decline in
the enrichment of the added15NH41-N pool. The magnitude
of this overestimation is dependent on the soil and the concentration of N applied.
The net ¯ux of14NH41and15NH41between the native soil
solution and the added15N labelled solution will depend on the concentration difference between the two solutions. Prior to N application the NH41 pool size in the CENIT
soil averaged 8.9 mg N kg21and the moisture content was 32.2% on an oven-dry weight basis. The concentration in the soil solution was 28 mg NH41-N l21. In comparison the
concentration of NH41-N applied was 14 and 108 mg NH41
-N l21 at the low and high application rates, respectively. Homogeneous mixing of the indigenous and applied N pools would take longer at the low N than at the high N application rate due to a less pronounced concentration gradient. One way of ensuring uniform mixing of the added label with the soil inorganic N pool would be to use soil suspensions. This could be useful for comparative purposes but the MIT rates obtained could not be extrapo-lated to the ®eld. Application of 15N label in solution has been found to stimulate N transformation processes compared to dry application techniques (Murphy et al., 1999; Willison et al., 1998).
The signi®cant difference between gross and net nitri®ca-tion rates observed in the current study was due to 15NO32
consumption. Gross nitri®cation rates would be overesti-mated if 15NO32 consumption takes place. For example
substantial NO32 consumption occurred in the Grange 0
soil receiving 2 mg N kg21, which resulted in daily gross nitri®cation rates being 3.6 times higher than net nitri®ca-tion rates. There was no evidence that the addinitri®ca-tion of NO32
stimulated NO32consumption (Stark and Hart, 1997), as the
rate was lower when 15 mg N kg21was applied than when
C.J. Watson et al. / Soil Biology & Biochemistry 32 (2000) 2019±2030
Table 6
Hourly gross nitri®cation rates (mg N kg21h21) calculated after linear smoothing at different times during a 24 h incubation. Figures in brackets are the
standard errors of the means estimated from the bootstrap technique n3
Time (h) Application rate of 2 mg N kg21 Application rate of 15 mg N kg21
CENIT Grange GC Grange 0 Grange 300 CENIT Grange GC Grange 0 Grange 300
1.5 0.29 (0.053) 0.17 (0.017) 0.18 (0.016) 0.18 (0.023) 0.55 (0.084) 0.28 (0.023) 0.30 (0.013) 0.27 (0.046) 4 0.29 (0.054) 0.17 (0.017) 0.18 (0.016) 0.18 (0.023) 0.55 (0.084) 0.29 (0.023) 0.30 (0.013) 0.28 (0.047) 10 0.30 (0.054) 0.18 (0.017) 0.20 (0.018) 0.20 (0.025) 0.55 (0.083) 0.29 (0.023) 0.31 (0.014) 0.29 (0.048) 16 0.31 (0.055) 0.20 (0.018) 0.22 (0.020) 0.22 (0.028) 0.55 (0.083) 0.29 (0.023) 0.32 (0.016) 0.30 (0.052) 24 0.32 (0.056) 0.22 (0.018) 0.27 (0.023) 0.24 (0.032) 0.55 (0.082) 0.30 (0.022) 0.34 (0.018) 0.32 (0.056) 0±24 h (smoothed) 0.30 (0.055) 0.19 (0.017) 0.22 (0.019) 0.21 (0.027) 0.55 (0.084) 0.29 (0.023) 0.32 (0.016) 0.30 (0.051) 0±24 h (raw) 0.33 (0.068) 0.19 (0.022) 0.21 (0.025) 0.21 (0.033) 0.53 (0.096) 0.29 (0.026) 0.31 (0.016) 0.27 (0.059)
Table 7
Rate of NO32consumption (mg N kg21h21). Figures in brackets are the standard errors of the means estimated from the bootstrap technique n3
Time interval 0±24 h Application rate of 2 mg N kg21 Application rate of 15 mg N kg21
CENIT Grange GC Grange 0 Grange 300 CENIT Grange GC Grange 0 Grange 300
C.J.
Watson
et
al.
/
Soil
Biology
&
Biochemistry
32
(2000)
2019
±
2030
Table 8
Daily gross N mineralisation and immobilisation rates (mg N kg21d21) calculated using 0±24 h data after exponential smoothing. Figures in brackets are the standard errors of the means estimated from the
bootstrap technique n3. (Daily net mineralisation (mg N kg21d21) was determined from linear regression of the change in the total mineral N pool with time; ns, not signi®cant; *P
,0:05;**P,0:01 and ***P,0:001;any small discrepancy in scaling up from hourly to daily rates is due to rounding of the means to 2 decimal places)
Application rate of 2 mg N kg21 Application rate of 15 mg N kg21
CENIT Grange GC Grange 0 Grange 300 CENIT Grange GC Grange 0 Grange 300
Gross mineralisation (m) 22.60 (1.459) 14.57 (3.197) 13.72 (1.147) 10.49 (1.188) 16.69 (2.234) 4.62 (1.358) 15.50 (1.075) 6.08 (0.478) Gross immobilisation (i) 18.06 (2.372) 9.76 (2.120) 10.18 (1.238) 7.53 (1.110) 11.14 (2.716) 3.67 (1.092) 12.16 (1.678) 5.28 (1.878)
m2i 4.54 (1.815) 4.80 (2.323) 3.54 (0.610) 2.97 (1.173) 5.55 (2.391) 0.94 (1.813) 3.34 (1.289) 0.81 (1.734)
Net mineralisation (netm) 4.96 (1.975) 1.97 (1.015) 20.168 (0.480) 0.52 (1.061) 5.52 (2.258) 1.28 (1.711) 21.24 (0.593) 0.64 (1.025) Signi®cance of difference
between grossm2iand netm
ns ns *** * ns ns *** ns
Table 9
Daily gross nitri®cation rates (mg N kg21d21) calculated using 0±24 h data after linear smoothing. Figures in brackets are the standard errors of the means estimated from the bootstrap technique n3. (Daily
net nitri®cation (mg N kg21d21) was determined from linear regression of the change in the NO 3
2-N pool with time; ns, not signi®cant; **P,0:01;***P,0:001;any small discrepancy in scaling up from hourly to daily rates is due to rounding of the means to 2 decimal places)
Application rate of 2 mg N kg21 Application rate of 15 mg N kg21
CENIT Grange GC Grange 0 Grange 300 CENIT Grange GC Grange 0 Grange 300
Gross nitri®cation (n) 7.31 (1.310) 4.64 (0.408) 5.28 (0.466) 5.08 (0.646) 13.21 (2.006) 7.05 (0.542) 7.63 (0.382) 7.14 (1.217) Net nitri®cation (netn) 5.70 (1.529) 2.45 (0.550) 1.46 (0.463) 2.36 (0.576) 13.36 (2.143) 5.44 (0.749) 4.37 (0.262) 3.25 (1.099)
n2netn 1.61 (0.554) 2.19 (0.245) 3.81 (0.218) 2.71 (0.238) 20.15 (0.362) 1.61 (0.382) 3.26 (0.391) 3.90 (0.485)
Signi®cance of difference betweenn and netn
2 mg 15NO32-N kg21was applied. This would explain the
higher % recovery of15NO32after 24 h when 15 mg N kg21
was applied. The CENIT soil had a lower rate of NO32
consumption compared to the Grange soils, which resulted in a higher % recovery of15NO32at the end of the
incuba-tion. Nitrate consumption was negligible in the CENIT soil receiving 15 mg N kg21and with this treatment there was no signi®cant difference between gross and net rates of nitri®cation. Consumption of NO32would include
denitri®-cation, dissimilatory NO32reduction and microbial
assimi-lation. Although gaseous losses were not measured in the current study, it is unlikely that denitri®cation alone would have resulted in the observed loss of NO32as the soils were
aerated and their moisture content was well below ®eld capacity. Dissimilatory NO32 reduction is also unlikely as
this pathway occurs in strictly anaerobic environments such as sediments (Cole, 1988). Evidence that rapid microbial assimilation of15NO32occurred in the current study comes
from the appearance of 15NH41 within 1.5 h in soil that
received15NO32. There are a number of recent reports that
indicate that rapid microbial assimilation of NO32 is an
important process in undisturbed forest soils (Stark and Hart, 1997) and in aquatic (Caraco et al., 1998) and marine (Kirchman and Wheeler, 1998) ecosystems.
Recent workers have taken the initial extraction time as 24 h after15N application and have calculated daily gross N transformation rates using the time interval from 24 h (time zero) to 72 h (time 1) (Murphy et al., 1999). However, unless it can be established that preferential use of applied N is not occurring after 24 h, calculated gross N transforma-tion rates will still be overestimated. The rapid decrease in enrichment of the15NH41pool observed in the current study
after applying 2 mg N kg21meant that after 24 h there was no further 15N pool dilution. Signi®cantly increasing the
15
NH41-N pool size, by applying 15 mg N kg21, ensured
continued pool dilution after 24 h but stimulated the rate of nitri®cation. Nitri®cation inhibitors could be used to prevent the conversion of NH41 to NO32. However, their
use would maintain an elevated NH41pool size that might
stimulate immobilisation or decrease mineralisation by feedback inhibition. There was evidence that the rate of gross mineralisation was lower with an application of 15 mg N kg21compared to 2 mg N kg21. Due to rapid15N pool dilution in some soils it may not be possible to deter-mine indigenous N transformation rates at time intervals greater than 24 h. However, information on gross N trans-formations could be obtained in response to a simulated fertiliser application. In this case differentially labelled NH4NO3would be the preferred N source.
4.3. Remineralisation
Remineralisation of immobilised15N can lead to substan-tial error in estimating mineralisation±immobilisation rates, but it is not believed to be a major process if incubations are less than one week (Bjarnason, 1988). The ®xation of NH41
to clay minerals can be allowed for by undertaking a time zero extraction (Davidson et al., 1991). The % recovery of
15
NH41at time zero was close to 100% in the CENIT soil.
However, abiotic NH41®xation occurred in the Grange soils.
Trehan (1996) noted that where ®xation of NH41occurred at
time zero the loss of14NH41from the soil solution via
nitri-®cation remobilised the ®xed15NH41from the clay minerals
into the soil solution. This could alter calculated mineralisa-tion rates if nitri®camineralisa-tion was rapid (Scherer and Werner, 1996).
The current study has shown that preferential consump-tion of applied 15NH41and 15NO32by soil microorganisms
invalidated some of the assumptions used in the 15N pool dilution technique. This led to an overestimate of gross mineralisation and nitri®cation rates, due to the greater rate of decline of the15N enrichment of the added N pool. In future studies it will be important to establish that prefer-ential use of applied N is not occurring during the experi-mental period and that steady-state conditions have been reached following 15N application. Care should be taken to ensure that process rates are not altered by the methods used to quantify them.
Acknowledgements
Gerard Travers would like to thank Teagasc for receipt of a Walsh Fellowship. The authors would also like to thank Mr P. Poland and Mr R.J. Laughlin for analysis of samples and Dr R.J. Stevens for helpful discussions.
References
Barraclough, D., 1991. The use of mean pool abundances to interpret15N
tracer experiments. Plant and Soil 131, 89±96.
Bjarnason, S., 1988. Calculation of gross nitrogen immobilization and mineralization in soil. J. Soil Sci. 39, 393±406.
Bran, Luebbe, 1995. Traacs 8001 instruction manual and application notes. Methods 54501605-A and 54501605-B. Bran and Luebbe, Brix-worth, Northants, UK.
Caraco, N.F., Lampman, G., Cole, J.J., Limburg, K.E., Pace, M.L., Fischer, D., 1998. Microbial assimilation of DIN in a nitrogen rich estuary: implications for food quality and isotope studies. Mar. Ecol. Prog. Series 167, 59±71.
Cole, J.A., 1988. Assimilatory and dissimilatory reduction of nitrate to ammonia. In: Cole, J.A., Ferguson, S.J. (Eds.). The Nitrogen and Sulphur Cycles, Cambridge University Press, Cambridge, UK, pp. 281±329.
Davidson, E.A., Hart, S.C., Shanks, C.A., Firestone, M.K., 1991. Measur-ing gross nitrogen mineralization, immobilization and nitri®cation by
15N isotopic pool dilution in intact soil cores. J. Soil Sci. 42, 335±349.
Delwiche, C.C., Steyn, P.L., 1970. Nitrogen isotope fractionation in soils and microbial reactions. Environ. Sci. Technol. 4, 929±935. Genstat, 1993. Genstat 5 Release 3 Reference Manual, Clarendon Press,
Oxford, UK.
Hart, S.C., Stark, J.M., Davidson, E.A., Firestone, M.K., 1994. Nitrogen mineralization, immobilization and nitri®cation. Methods of Soil Analysis: Microbiological and Biochemical Properties. SSSA Book Series No. 5, pp. 985±1018.
dynamics in a natural calcareous grassland under a changing climate. Biol. Fertil. Soils 27, 267±273.
Kirchman, D.L., Wheeler, P.A., 1998. Uptake of ammonium and nitrate by heterotrophic bacteria and phytoplankton in the sub-Arctic Paci®c. Deep-Sea Res. Part 1 45, 347±365.
Kirkham, D., Bartholomew, W.V., 1954. Equations for following nutrient transformations in soil, utilizing tracer data. Proc. Soil Sci. Soc. Am. 18, 33±34.
Laughlin, R.J., Stevens, R.J., Zhuo, S., 1997. Determining nitrogen-15 in ammonium by producing nitrous oxide. Soil Sci. Soc. Am. J. 61, 462± 465.
Ledgard, S.F., Jarvis, S.C., Hatch, D.J., 1998. Short-term nitrogen ¯uxes in grassland soils under different long-term nitrogen management regimes. Soil Biol. Biochem. 30, 1233±1241.
Manly, B.F.J., 1997. Randomization, Bootstrap and Monte Carlo Methods in Biology, Chapman and Hall, London (399pp.).
Murphy, D.V., Bhogal, A., Shepherd, M., Goulding, K.W.T., Jarvis, S.C., Barraclough, D., Gaunt, J.L., 1999. Comparison of15N labelling
meth-ods to measure gross nitrogen mineralisation. Soil Biol. Biochem. 31, 2015±2024.
Recous, S., Aita, C., Mary, B., 1999. In situ changes in gross N transforma-tions in bare soil after addition of straw. Soil Biol. Biochem. 31, 119± 133.
Scherer, H.W., Werner, W., 1996. Signi®cance of soil microorganisms for the mobilization of nonexchangeable ammonium. Biol. Fertil. Soils 22, 248±251.
Schimel, D.S., 1986. Carbon and nitrogen turnover in adjacent grassland and cropland ecosystems. Biogeochemistry 2, 345±357.
Stark, J.M., Hart, S.C., 1997. High rates of nitri®cation and nitrate turnover in undisturbed coniferous forests. Nature 385, 61±64.
Stevens, R.J., Laughlin, R.J., 1994. Determining nitrogen-15 in nitrite or nitrate by producing nitrous oxide. Soil Sci. Soc. Am. J. 58, 1108±1116. Trehan, S.P., 1996. Immobilisation of15NH
4
1in three soils by chemical and
biological processes. Soil Biol. Biochem. 28, 1021±1027.
Watson, C.J., Mills, C.L., 1998. Gross nitrogen transformations in grass-land soils as affected by previous management intensity. Soil Biol. Biochem. 30, 743±753.
Willison, T.W., Baker, J.C., Murphy, D.V., Goulding, K.W.T., 1998. Comparison of a wet and dry 15N isotopic dilution technique as a