(1987) Seasonal Variation of Potentially Mineralizable Nitrogen in ...

Seasonal Variation of Potentially Mineralizable Nitrogen in Four Cropping Systems 1
ABSTRACT
The concept of potentially mineralizable N is discussed and data
that support the identification of /V () as an active fraction of soil
organic N are presented. Aerobic medium-term incubations (13 weeks)
were used to measure the potentially mineralizable nitrogen (N a ) in
four cropping systems on four different occasions from early spring
to late autumn. The cropping systems consisted of barley (Hordeum
distichum L.) with no N fertilizers (BO), barley receiving 120 kg N
ha~' yr~' (B120), meadow fescue (Festuca pratensis Hudson) ley
receiving 120 + 80 kg N ha ~' yr~' (GL), and lucerne (Medicago
sativa L.) ley with no N additions (LL). The amounts of N mineralized
during the 13-week incubations (37 °C) ranged from 188 to
483 kg N ha ', and the cropping systems ranked in the order GL
> LL > B120 > BO and showed a steady decline in all systems
from spring to harvest and a subsequent increase from harvest to
autumn. The seasonal differences were as large as the differences
between systems. Three related models were employed to describe
the kinetics of N mineralization during the incubations: (i) firstorder,
(ii) two-component (sum of two first-order models), and (iii)
a simplified special case of the two-component model. In all cases,
the special case of the two-component model offered the best description
of the curves of accumulated mineral N. Based on the mineralization
course and an observed relationship between literature
values of /V 0 and the rate constant (k), it is hypothesized that the
mineralization is mediated by two distinct populations: an opportunistic,
mainly immobile, population (bacteria) and a generalistic,
mobile one (fungi), and that the k value is influenced both by the
amount of substrate and its quality. The fact that the amount of
mineralizable N decreases during the growing season and increases
in autumn as a result of organic matter input provides evidence for
the existence of an active fraction of soil organic matter.
Additional Index Words: barley, lucerne, meadow fescue, N models,
microbial processes, soil organic matter, first-order kinetics.
Bonde, T.A., and T. Rosswall. 1987. Seasonal variation of potentially
mineralizable nitrogen in four cropping systems. Soil Sci. Soc.
Am. J. 51:1508-1514.
THE FIRST FORMULATION of the "nitrogen mineralization
potential C/V 0 )" concept and its associated
rate constant (k), representing a description of
the N-mineralization course during a long-term (30
weeks) aerobic incubation, was presented by Stanford
and Smith (1972). They regarded N Q as a definable
1
Contribution from the project Ecology of Arable Land. The Role
of Organisms in Nitrogen Cycling at the Swedish Univ. of Agricultural
Sciences, Uppsala, Sweden. Received 11 Mar. 1986.
2 Graduate Research Associate and Professor, Dep. of Water in
Environment and Society, Univ. of Linkoping, S-581 83 Linkoping,
Sweden, respectively.
TORBEN A. BONDE AND THOMAS ROSSWALL 2
quantity, and its proportion of total soil N as a variable
affected by cropping practices, whereas the rate
constant (k) was a true constant (0.054 week"') for the
range of soils studied. These findings are generally accepted
(Campbell et al., 1984), though there has been
some discussion as to what N 0 represents and the universality
of the k value (Juma et al., 1984). Nitrogenmineralization
potentials have been widely used as a
means to determine the effect of various agricultural
practices on soil fertility, such as N fertilization, tillage,
crop rotations, and manure additions (Doran,
1980; Campbell and Souster, 1982; Carter and Rennie,
1982; El-Haris et al., 1983; Griffin and Laine, 1983).
Similar aerobic incubations also proved useful in
determining the effect pf temperature on N mineralization
during long-term (Stanford et al., 1973a, 1975;
Campbell et al., 1981) and moisture during short-term
incubations (Stanford and Epstein, 1974; Cassman and
Munns, 1980; Myers et al., 1982). The determination
of the effects of temperature and moisture led to an
attempt to calculate N mineralization in the field, using
maximum mineralization rates corrected for field
temperature and moisture (Stanford et al., 1973b, 1977;
Smith et al., 1977; Marion et al., 1981; Campbell et
al., 1984).
A more sophisticated employment of this concept
would be the use of N 0 as an estimate of an active
fraction in models describing temporal fluctuations in
soil organic matter both with regard to its quantity
and quality. Such models have been developed, and
in order to obtain realistic results, the soil organic
matter has been divided into three or four components
(i.e., biomass-active, nonbiomass-active, slow,
and passive) and usually two litter components (i.e.,
labile and structural). The fractionation is based on
turnover rates and used to model mineralization and
immobilization of soil N (Paul and Juma, 1981; van
Veen and Frissel, 1981; Parton et al., 1983). There
have, however, been difficulties in developing methods
that can realistically quantify the components of
the model.
Most authors have considered N 0 to be characteristic
for a certain cropping system and soil, and have
not investigated possible seasonal patterns. If N Q is
equivalent to a measure of an active fraction of soil
organic matter, seasonal fluctuations should occur. If
this is the case, the fraction of soil organic N represented
by N 0 might supply a major portion of plant
available N for crop growth. El-Haris et al. (1983),
however, found twice as high N 0 values when sampling
took place in mid-September as compared with

BONDE & ROSSWALL: NITROGEN IN FOUR CROPPING SYSTEMS 1509
sampling in mid-March. The spring k values were a
factor three higher than those from the autumn. Stanford
et al. (1977) calculated a 20% reduction in N 0
from April to September for a number of soils, but
neither publication has analyzed the seasonal variation
in any detail.
The present study was conducted to (i) investigate
whether N 0 could be considered a representative of an
active fraction of soil organic matter, (ii) investigate
whether the first-order kinetics represent a correct description
of the mineralization course, and (iii) interpret
the biological implications of the NQ concept based
on the mineralization course and published values of
NQ and k.
MATERIALS AND METHODS
Soils and Sampling
Soil was collected from field experiments of the Ecology
of Arable Land project, the research site of which is located
at Kjettslinge, 40-km north of Uppsala, Sweden. The field
is situated in a flatland area 25-m above sea level. It had
been cropped for ca. 100 yr before initiation of the experiment
in 1980. The climate is cold-temperate and semihumid.
Mean annual precipitation is 520 mm with the highest
precipitation in August (70 mm). Mean annual
temperature is 5.4 °C and monthly mean temperatures are
— 5.3 and 16.7 °C for January and July, respectively (Steen
et al., 1984). The top soil is a loam (Cumulic Haploboroll,
illitic, frigid. USDA, Soil Survey Staff, 1975) with 19% clay,
a pH of 6.3, and total C and N contents of 22 and 2.3 g
kg~', respectively, at the start of the experiment. Detailed
description of the experimental site is given by Steen et al.
(1984).
Four cropping systems were investigated: barley (Hordeum
distichum L.) with no addition of N fertilizers (BO);
barley receiving 120 kg N ha~' yr~' as Ca(NO 3 ) 2 (B120) on
16 May 1984; meadow fescue (Festuca pratensis L.) ley receiving
120 and 80 kg N ha -' yr~' (GL) on 16 May and 2
July, respectively; and lucerne (Medicago saliva L.) ley without
N fertilizer additions (LL).
Each treatment had four replicates. From each individual
plot, two soil cores were taken to a depth of 27 cm (plough
layer) on 21 Apr., 7 June, 7 Aug., and 22 Oct. 1984. The
samples were air-dried, mixed, and sieved (2 mm) to produce
a replicate sample from each treatment.
Nitrogen-mineralization Procedure
The apparatus employed in this study was similar to the
one used by MacKay and Carefoot (1981), i.e., polystyrene
filter units (150 mL; Filter unit 7103, Falcon, Oxnard, CA,
USA) fitted with cellulose based filter membranes of 0.22-
jum pore size with a bubbling point pressure (380 kPa) well
above the applied suction (80 kPa).
Soil subsamples (15 g) were mixed with equal amounts of
washed sand (0.6-mm mean grain size) and rewetted with 3
mL of distilled water followed by gentle mixing. The mixtures
were transferred to the filter units and covered by two
layers of nylon nets (1-mm mesh size), and during the leaching
procedure by a 2-cm layer of glass-wool pads held together
by two layers of nylon net. The glasswool protected
the soil from disaggregation during leaching. The leaching
solution consisted of 100-mL 0.01 MCaCl 2 , followed by 25-
mL minus-N nutrient solution (Stanford and Smith, 1972).
In an initial experiment, the 5-mL successive additions of
extractant recommended by Stanford and Smith (1972) dispersed
the soil, and, in the end, partly clogged the filters.
Such clogging would lead to erroneous moisture adjustments.
To prevent this, a simple drip irrigation system was
constructed consisting of 150-mL funnels connected with
the top center of the filter units. The tubings were fitted with
clamps for dripping rate adjustment. A vacuum pump was
connected to a manometer and through manifolds to the
filter units. A leaching rate of 250 mL h~' and a consistent
suction of 80 kPa (600-mm Hg), extended by 15 min after
the end of leaching, conserved soil structure and prevented
filter clogging. This procedure ensured water contents close
to optimum (20% of soil plus sand dry weight; Munro and
MacKay, 1964; Keeney and Bremner, 1966). MacKay and
Carefoot (1981) recommended an "overnight" procedure for
adjusting soil moisture, but a stable water content was
reached almost immediately after leaching, probably due to
the sand amendments. The soil samples were incubated at
37 °C and leached 0, 1, 2, 3, 4, 6, 8, 10, and 13 weeks after
the start of the experiment.
Chemical Analyses
Mineral N (NH 4 , NO 2 , and NO 3 ) was determined by flow
injection analysis using an automated FIA 06 (Tecator AB,
Hoganas, Sweden). Organic N in the leachates was measured
according to Jaenicke (1974) after perchlorate digestion.
Mineralization Models
The first-order model (Stanford and Smith, 1972) is represented
in product-appearance form as follows:
N m = N 0 [l - exp(-M] [1]
where N m is the amount of N mineralized at time t, N 0 is
the initial amount of potentially mineralizable N, and k^ is
the first-order rate constant. Because different fractions of
the organic N in the soil may be differentially susceptible to
mineralization, the first-order model is sometimes modified
to a two-component model
N m = N A [l - exp(-/2f)] + N R (l - [2]
where Nj and NK are the amounts of organic N initially
present in the available and resistant fractions, respectively,
and h and k are the first-order rate constants for the two
fractions. The sum of N. t plus N R equals N 0 or N A = N&S
where 5 is the fraction of N 0 belonging to the available pool
(Lindeman and Cardenas, 1984). A model intermediate in
complexity between Eq. [1] and [2] can be derived as a special
case, or as a degenerate form, from Eq. [2]
N m = N A [\ - exp(-/zO] + Ct [3]
where C = kN R . Equations [1], [2] and [3] were the models
for N mineralization evaluated in this study.
Statistical Analysis
The three models were fit to data on N mineralization by
nonlinear regression using the method of Marquardt in the
SAS (SAS Inst., 1982) software package. The model offering
the most appropriate description of the data was selected on
the basis of the residual sum of squares (RSS) left unexplained
by the regression. However, the model leaving the
lowest RSS was not always selected as the most justifiable.
Because Eq. [3] is merely a special case of the two-component
model (Eq. [2]), the latter will never leave an RSS higher
than that of Eq. [3]. The first-order model can also be regarded
as an approximation of Eq. [2], applicable in situations
when either N R or k are effectively zero. Consequently,
the first-order model could not be expected to leave a lower
RSS than Eq. [3]. Therefore, the significance of the difference
of the RSS left by the three models was subjected to an F
test as described by Robinson (1985).

1510 SOIL SCI. SOC. AM. J., VOL 51, 1987
25-i
_20-
1 2 3 4 5 6 7
TIME (WEEKS)
1 2 3 4 5 6 7
TIME (WEEKS)
9 10 11 12 13
10 11 12 13
Fig. 1. Rate of N mineralization during 13 weeks of incubation of
soils sampled from the grass ley (GL) cropping system on 7 June
and 7 August. The horizontal bars represent the time period for
which rates were calculated. The solid line represents GL 7 June
and the broken line represents GL 7 August.
RESULTS AND DISCUSSION
Mineralization Rates
An example of the weekly mineral-N production rate
during the actual incubation period is presented in
Fig. 1. In all cases, the (NO 3 + NO 2 )-N production
rate declined rapidly to about 10% of the initial rate
during weeks 10 to 13, but in some cases there was
no decline between weeks 3 and 4. The NH 4 -N production
rate showed a sharp decline from week 1 (4
mg kg" 1 week" 1 ) to very low levels during weeks 2
and 3 (1 mg kg~' week^ ')> then a rapid increase during
week 4, followed by a leveling off throughout the rest
of the incubation period. Standard error of means
amounted to 8 to 12% of total mineralized N on leachings
performed at weeks 1 to 2, but declined to 2 to
5% on subsequent leachings, resulting in a cumulated
standard error of 3 to 6% of cumulated mineralized
N. The error was mainly caused by low amounts of
mineral N in a few replicates. The amounts of organic-
N in the leachates were 10 to 20% of inorganic N as
measured on the initial leachings. This was not further
analyzed and excluded from the following calculations.
Future studies using this methodology should,
however, include analyses of organic N since we do
not know if its proportion to total N in the leachate
remains constant over time.
Nitrate plus NO 2 -N constituted 80% of the total
mineral N in week 1, increasing to 90% in week 2,
followed by a gradual decrease to 15 to 20% in week
13 (Fig. 1). The reason for this shift from NO 3 as the
major form of inorganic N may be related to the general
finding that the temperature optimum for nitrification
is very close to 35 °C (Bremner, 1965). Lower
optima (20 °C) have been reported (Malhi and McGill,
1982). In a number of temperate soils, Keeney and
Bremner (1966) found a complete shift in the appearance
of mineral N in favor of NH 4 when the incubation
temperature was changed from 35 to 40 °C. P.
Berg (1986, personal communication) did not, however,
find 37 °C to be suboptimal for nitrification rates
on the same soil as we used. The incubation temperature
of 37 °C could be slightly above optimum for
nitrification and cause the incomplete conversion of
NH 4 to NO 3 . This temperature was, however, preferred
to decrease length of incubation time. Temperature
optima for mineralization is not likely to be <
37 °C (Myers, 1975). Therefore, mineralization rates
were probably not affected negatively by the incubation
conditions.
As the aggregate size did not exceed 2 mm and the
size of the individual sample was small, it did not
seem likely that anaerobic conditions inhibited mineralization
or nitrification due to restricted O 2 diffusion
into aggregates (Sexstone et al., 1985) or that it
caused significant denitrification.
The Amount of Mineral N Produced during a 13-
week Incubation
The amounts of N mineralized from the samples
collected in April were equal to 321, 371, 483, and 420
kg ha" 1 for BO, B120, GL, and LL, respectively (Table
1). The percentage of mineralized N to total organic
N was 4.2, 3.6, 5.0, and 4.8%. This can be compared
with the estimate by Paul (1984), that the microbial
biomass accounts for 4 to 6% of total organic N,
whereas the active nonbiomass fraction accounts for
6 to 10% of total organic N.
The highest amount of mineral N was produced in
soils from the grass ley irrespective of sampling date
(Table 1). The lucerne ley, which fixed 291 kg N ha~'
in 1983 (Wivstad et al., 1986), suffered from an unusually
cold winter in 1983 to 1984 and only managed
to partially recover during the growing season, which
is probably the cause for the relatively low mineralization
in samples collected in June and August. The
results, thus, cannot be considered representative of
legume leys.
The influence of fertilization and cropping practices
on the amounts of mineralizable N is in accordance
with the findings of Stanford and Smith (1972), Doran
(1980), and El-Haris et al. (1983). Increasing additions
of organic matter and N fertilizer increases the mineralization
potentials and the proportion of total N
present in the more available form of soil N.
The decrease in mineralized N from April to August
was 133, 103, 153, and 165 kg ha~' for BO, B120, GL,
and LL, respectively (Table 1), which is similar to cal-

BONDE & ROSSWALL: NITROGEN IN FOUR CROPPING SYSTEMS 1511
Table 1. Total mineral N produced during a 13-week incubation
of top soil (0-27 cm) collected from four cropping systems
at four sampling occasions.
Sampling
date
21 April
7 June
7 August
22 October
Form of
mineral N
Total mineral N
(NO, + NOJ-N
NH.-N
Total mineral N
(NO, + NOJ-N
NH.-N
Total mineral N
(NO, + NOJ-N
NH.-N
Total mineral N
(NO, + NOJ-N
NH 4 -N
Cropping systemf
BO B120 GL LL
321adi
190
(120)§
259ad
117
142
188ae
68
121
269ad
140
129
Kg iia
371abd 483cd
232 338
(139) (145)
354bcde
210
144
268abe
119
150
SOOabde
177
124
405cde
255
150
330be
185
144
364bde
222
141
420bcd
254
(166)
279abde
104
175
255be
89
167
299abde
142
157
t BO = Barley without addition of N fertilizer; B120 = barley receiving
120 kg N ha-' yr 1 ; GL = grass ley receiving 200 kg N ha' 1 yr 1 ; LL =
lucerne ley receiving no N fertilization.
t Means that do not differ significantly (p = 0.05) are followed by the same
letter; a, b, c rows; d, e columns;
§ An estimate based on the mean of the last three sampling dates.
culated net mineralization rates of 105, 128, 257, and
211 kg ha~' yr- 1 for BO, B120, GL, and LL, respectively
(Rosswall and Paustian, 1984). If N 0 is a definable
quantity and the only major source of mineralizable
N, a decrease in N 0 during the growing season
should equal the amount of N mineralized. On a yearly
basis this will, however, be an underestimation, since
autumn mineralization is important despite an increase
in A^ due to addition of root litter and aboveground
litter plowed under. In addition, our data
probably underestimate net N mineralization during
the growing season, since the 13-week incubations were
too short to allow a statistically justifiable determination
of both N A and N R .
Fertilizer N (120 kg) was applied on 16 May to the
B120 and GL treatments, and the grass ley further
received 80 kg of N on 2 July. Application of fertilizer
N has been demonstrated to add to the mineralizable
soil N and, in particular, to the readily available pool
of TV,, (El-Haris et al, 1983). Fertilization would then
tend to prevent a decrease in the amount of mineralizable
N during the growing season. To estimate net
N mineralization by a substrate disappearance method,
it is thus necessary to take into account possible additions
to the substrate pool during the period investigated.
The amounts of organic N added through the system
by root and aboveground litter input have been
estimated to be 40 and 55 kg ha~' yr" 1 for BO and
B120, respectively (A.C. Hansson, 1986, personal
communication). This amount is not sufficient to explain
the observed increase of the mineralizable N
fraction in BO from August to October (81 kg ha" 1 ).
The larger amount might be caused by a period of
stimulated microbial activity after harvest, due to
added C-rich crop residues, which would result in
greater immobilization of N and thus larger amounts
of N in live and dead microbial cells and microbial
metabolites (Campbell and Biederbeck, 1982). Nitrogen
in these forms are thought to be a large part of
the readily mineralizable soil N (7V 0 ). This apparent
5 6 7
TIME (WEEKS)
9 10 11 12 13
Fig. 2. Accumulated amounts of mineralized N during 13 weeks of
incubation of soils collected from four cropping systems on 7 June.
Error brackets denote accumulated standard deviation.
rise in the active part of soil N took place during a
period with net mineralization and total N loss from
the system (Bergstrom, 1986). The net effect in the N
transformations during this period is thus a decrease
of the slow fraction rather than the active one. The
smaller increase in the mineralizable fraction in B120
(32 kg), in spite of a larger amount of litter added to
the system, is surprising, but may be caused by a greater
net immobilization of mineral N in these plots.
Cumulative N Mineralization Patterns
Patterns of cumulative N mineralization from soil
samples collected on 7 June are shown for all four
treatments in Fig. 2. The shapes of the mineralization
curves for soil samples collected on the other three
dates were very similar to those in Fig. 2 and are not
shown. The initially high rates of N mineralization in
soil samples collected from all four treatments at all
four dates declined to low, fairly constant rates by
week 6 (Fig. 1), resulting in the apparently linear increase
in total cumulative inorganic N after about week
4 (Fig. 2).
Stanford and Smith (1972) calculated an initial N 0
value and the k value on the basis of the curve of
accumulated mineralized N from week 2 to week 30.
The mineralized N during the first 2 weeks was added
to the initially calculated N 0 value, giving the final 7V 0 .
This was motivated by a poor model fit to the data
set if the first 2 weeks were included, which was probably
a result of pretreatment of soil samples. Numerous
reports on the effect of air drying, sieving, grinding,
and rewetting are available (e.g., Birch, 1960;
Agarwal et al., 1970). Soils respond to drying and rewetting
with a flush of C and N mineralization of short
duration, the magnitude depending on soil characteristics.
To overcome the problem of an initially rapid
mineralization rate, a two-component model (a sum
of two first-order equations; Molina et al., 1980; Griffin
and Laine, 1983), a three-component model (Richter
et al., 1982), or a modified one-component model
(Marion et al., 1981) have been used. These models

1512 SOIL SCI. SOC. AM. J., VOL 51, 1987
give good data fit, but are often not based on attempts
to understand the underlying biological mechanisms.
The first-order and two-component models, including
the simplified special case of the latter, were fit by
nonlinear regression to the data on soils from all four
treatments on three of the four collection dates. The
first sampling date was omitted due to difficulties in
NH 4 analyses (Table 1), and the data from this set of
samples were not used in the statistical analyses. The
first-order model (Eq. [1]) never offered as close a fit
to the data as the special case of the two-component
model (Eq. [3]). The latter always gave a statistically
significant reduction in RSS over the variance left
unexplained by the first-order model. Moreover, the
residuals left by fits of the first-order model were highly
nonrandom. Nonrandom residuals are produced when
data are fitted to a wrong model (Robinson, 1985).
The values of N,,, predicted using the first-order model
were always lower than the first two or three data points
and the last two or four points, and predicted values
for intermediate sampling times were almost always
higher than those observed. Therefore, the first-order
model was judged to be an inadequate description of
the kinetics of N mineralization observed in these experiments.
Nonlinear regression using the two-component
model (Eq. [2]) invariably left unexplained variance
slightly lower than that of the special case (Eq. [3]),
but the improvement in fit achieved by the two-component
model was never statistically significant by the
F test. The residuals left by the regressions using both
models appeared to be random.
At least two reasons can be offered to account for
the inability of the two-component model to provide
a significant improvement in fit over that offered by
its special case. The special case is a good approximation
to the more complex two-component model
when the length of time, over which data were collected,
is short compared to the half-life of the resistant
fraction of soil organic N. That is, the initial N
mineralization from a large resistant fraction of soil
N could appear to be linear rather than gradually declining.
It is therefore possible that if the duration of
the incubations had been doubled or tripled, the special
case would no longer have closely approximated
the two-component model and the latter would have
proved statistically superior. Alternatively, if the
number of sampling times during the incubations had
been greatly increased, the difference in RSS left by
the two models might have become statistically significant,
and, again, the two-component model could
have emerged as the more justifiable description of
the observed patterns of N mineralization. Therefore,
the special case of the two-component model was
judged to offer the most statistically justifiable description
of the kinetics of N mineralization in these experiments.
Accordingly, the smooth curves in Fig. 2
were calculated using the special case.
It is also possible that the model, which has hitherto
been considered merely a special case of the two-component
model, may actually reflect more accurately
the biological determinants of N-mineralization kinetics
in these experiments. A constant mineralization
rate may be due to a constant amount of enzymes (or
no limitation due to enzymes) working on a constant
amount of substrate. Filamentous fungi are able to
grow on nutrient poor substrate, but hyphal outgrowth
is then mainly accounted for by an increase in amounts
of ghost hyphae (hyphae without cytoplasm; Paustian
and Schniirer, 1987). An on-going colonization of new
substrate by extension of fungal hyphae through the
soil aggregates, resulting in maintenance of the active
fungal population (i.e., no growth in terms of cytoplasmic
content) would result in a constant mineralization
rate.
An alternative explanation of the straight line phase
would be the existence of an upper limit of the active
microbial biomass due to, for example, protozoan
grazing rather than limitations to growth caused by
substrate. Protozoa have been proposed to serve as
mineralizing agents by consuming bacteria with concomitant
release of NH 4 (Clarholm, 1984). Under certain
circumstances they could prevent a given bacterial
population to expand, e.g., out of protected
microsites (van Veen et al, 1984), and thus limit the
number to a constant level. An abiotic factor, such as
clay fixation, might interfere with the method employed
for the study of mineralization kinetics and
should be taken into account on NH 4 -fixing soils.
The size of the available soil organic matter (SOM)
fraction (N A ) could be affected by air-drying and sieving
(thus partial grinding) of the soil samples. In addition
to particular available SOM, the N A fraction
probably consists of killed microbial biomass and SOM
rendered decomposable by the pretreatment (Jenkinson
and Powlson, 1976; Powlson, 1980). However,
Bottner (1985) identified a dormant (or protected) part
of the microbial biomass, which survived drying,
whereas an active (or exposed) part corresponding to
25% of total biomass suffered from drying but was able
to recover. The surviving population probably consists
mainly of bacteria and the killed population primarily
of fungi. Based on these observations and the
mineralization pattern observed in this study we suggest
the N A fraction to be decomposed by a surviving
opportunistic group of organisms (r strategists; Andrews,
1984). The main part of the protected biomass
with limited access to organic matter will probably die
during the incubation due to energy limitations and
increase the amount of easily decomposable organic
matter. Hence, it is likely that a large part of the microbial
biomass will decompose in a few weeks in a
Stanford incubation and that the composition of the
microbial biomass will change in favor of the fungi.
The fungal dominated biomass will be constant over
a long period.
Relationships between 7V 0 and k 0 Values
The first-order equation was originally used to describe
the effect of cultivation of virgin soil on total
soil N levels (Stevenson, 1965). This assumes that "the
biological or biochemical potential to decompose organic
matter is not limited" and that "the organic N-
pool is initially kinetically homogenous and remains
so" (van Veen et al., 1981). Jansson (1958), among
others, showed that the latter assumption is wrong,
and proposed the existence of an active fraction of soil
N, which is probably equivalent to the potentially mi-

BONDE & ROSSWALL: NITROGEN IN FOUR CROPPING SYSTEMS 1513
neralizable pool. "Limited" in the former assumption
implies that the biomass is at a maximum in relation
to the amount of substrate, so that no increase in biomass
occurs at the start of the experiment.
To investigate whether mineralization rates are limited
during incubation according to Stanford and Smith
(1972), literature values (Stanford and Smith 1972,
Campbell et al., 1984) of A^ vs. k 0 for long-term (30
weeks) incubations have been plotted, but no significant
correlation was found.
An inverse relationship between A^ and k 0 values
was evident when data from El-Haris et al. (1983) medium-term
incubations (12 weeks), were plotted. Small
N 0 values were always associated with large k 0 values
when sampling took place in spring as compared with
values obtained when sampling took place in autumn.
As the half-lives of the micrpbial biomass and the active
nonbiomass soil organic matter have been estimated
to be 0.5 and 1.5 yr, respectively (Paul, 1984),
a relatively large amount of biomass N compared to
nonbiomass active N would affect the k 0 value positively.
An inverse relation between N 0 and ko values
can thus be accomplished if the microbial biomass
sustains itself through winter and early spring at the
expense of active nonbiomass soil organic matter.
Schnurer et al. (1986) have not observed any marked
seasonal dynamics of the microorganisms at the Kjettslinge
field, and the variation in N 0 would then be
caused mainly by changes in the amount of nonbiomass
substrate.
Carter and Rennie (1982) compared the flush of C
and N after CHC1 3 fumigation concomitant with a determination
of A^o and t\ j2 (the "half-life" of N 0 ) on
tilled and nontilled soils. This provides an experimental
test of the proportionality between the biomass
N/./VO ratio and k 0 . Biomass N can be calculated
based on C and N flushes (Voroney, 1983), and the
rate constant ko from the formula t\ /2 = In2/k 0 . Analysis
of the data of Carter and Rennie (1982) showed
no significant correlation between the biomass N/A'o
ratio and k 0 , but did show a positive one for conventionally
tilled soils. If one data point, representing a
topsoil of a no-till treatment, was excluded, the correlation
became significant (r = 0.530, p = 0.05, « =
15) for the total data set. These data suggest that a
major part of the microbial biomass is easily mineralizable
and provides some explanation for the observed
relation between N 0 and ko values.
A two- or multicomponent model, however, would
be a better description of the mineralization course,
since several fractions of organic matter are present
in the soil. The single k 0 value associated with the firstorder
model is therefore proposed to be a mixture of
h and k values belonging to a two-component model,
as is NO a mixture of N A and N R . The mineralization
course in this study was well fitted by a simplified twocomponent
model, making it fallacious to determine
a traditional N 0 and fc 0 value. However, the same tendency
of having a small ko value whenever a large N 0
value occurs, is recognizable in the N A and h values.
These parameters are comparable to the N 2 , k 2 values
in the two-component model of Richter et al. (1982),
in the data set of which the same pattern occurred
(Fig. 3). Because the simplified two-component model
was the most justifiable for the data presented in this
4-1
3-
2-
1-
0^
•- DATA FROM THIS PAPER
x-DATA FROM RICHTER ET AL.I1982 )
50 100 150
N 2 or N A (kg ha ,-1i
Fig. 3. Plot of the rate constant h vs. N A for the data from this paper,
and * 2 and N, from Richter et al. (1982).
paper, N A and h values should not suffer from errors
caused by fitting of an inappropriate model. Thus, there
may be a real biological mechanism generating the
observed relationship.
The findings of Paris et al. (1981), that first-order
rate constants were proportional to bacterial numbers,
might explain the relationship between h and N A values
presented in Fig. 3, if bacterial density varied between
treatments and sampling dates. Based on their
findings, a high bacterial density would be associated
with a high measured value for h. In addition, a high
bacterial density could develop at the expense of available
soil organic matter, and therefore be associated
with low N A values. However, an inverse relationship
between bacterial density and available soil organic
matter is not inevitable. Alternatively, a biological
limitation on mineralization imposed by the growth
of active organisms in the early stages of incubation
would account for the relationship between h and N A ,
even if a lag phase did not occur.
ACKNOWLEDGMENT
We are greatly indebted to Dr. S. Simkins for many valuable
discussions regarding the statistical treatment. T.A.
Bonde received financial support from the Inst. of Genetics
and Ecology, Univ. of Aarhus, Denmark. The investigation
formed part of the project, Ecology of Arable Land. The
Role of Organisms in Nitrogen Cycling, supported by the
Swedish Council for Planning and Coordination of Res., the
Swedish Council for Forestry and Agric. Res., the Swedish
Natural Science Res. Council, and the Swedish Environment
Protection Board.

1514 SOIL SCI. SOC. AM. J., VOL. 51, 1987