Ecosystems (2011) 14: 710–719

DOI: 10.1007/s10021-011-9440-z
 2011 Springer Science+Business Media, LLC

Identification of General Patterns

of Nutrient and Labile Carbon
Control on Soil Carbon Dynamics
Across a Successional Gradient
Alexandru Milcu,1* Angela Heim,2 Richard J. Ellis,3 Stefan Scheu,4
and Pete Manning1,5

1
Division of Biology, NERC Centre for Population Biology, Imperial College London, Silwood Park, Ascot SL5 7PY, UK; 2Department of
Earth and Environmental Sciences, University of Potsdam, Karl-Liebknecht-Str. 24-25, 14476 Potsdam, Germany; 3Veterinary
Laboratories Agency, VLA Weybridge, New Haw KT15 3NB, UK; 4J.F. Blumenbach Institute of Zoology and Anthropology,
Georg-August-University Göttingen, Berliner Str. 28, 37073 Göttingen, Germany; 5School of Agriculture, Food and Rural Development,
Newcastle University, Newcastle, Tyne and Wear NE1 7RU, UK

ABSTRACT
Carbon (C) inputs and nutrient availability are
known to affect soil organic carbon (SOC) stocks.
However, general rules regarding the operation of
these factors across a range of soil nutrient avail-
abilities and substrate qualities are unidentified.
‘‘Priming’’ (stimulated decomposition by labile C
inputs) and ‘preferential substrate utilization’ (re-
tarded decomposition due to shifts in community
composition towards microbes that do not mineral-
ize SOC) are two hypotheses to explain effects of
labile C additions on SOC dynamics. For effects of
nutrient additions (nitrogen and phosphorus) on
SOC dynamics, the stoichiometric (faster decompo-
sition of materials of low carbon-to-nutrient ratios)
and ‘microbial mining’ (that is, reduced breakdown
Received 28 September 2010; accepted 1 April 2011;
published online 12 May 2011
Author contributions: Alexandru Milcu and Pete Manning: conceived
and designed the study, Alexandru Milcu and Angela Heim: performed
research, Alexandru Milcu and Richard Ellis: analyzed data, Richard Ellis
and Stefan Scheu: contributed methods, Alexandru Milcu and Pete
Manning: wrote the paper. All authors discussed the results and the
structure of the paper, commented and revised the manuscript text.
*Corresponding author; e-mail: a.milcu@imperial.ac.uk
of recalcitrant C forms for nutrients under fertile
conditions) hypotheses have been proposed. Using
the natural gradient of soil nutrient availability and
substrate quality of a chronosequence, combined
with labile C and nutrient amendments, we explored
the support for these contrasting hypotheses. Addi-
tions of labile C, nitrogen (N), phosphorus (P), and
combinations of C and N and C and P were applied to
three sites: 2-year fallow grassland, mature grassland
and forest, and the effects of site and nutrient addi-
tions on litter decomposition and soil C dynamics
were assessed. The response to C addition supported
the preferential substrate hypothesis for easily
degradable litter C and the priming hypothesis for
SOC, but only in nitrogen-enriched soils of the forest
site. Responses to N addition supported the microbial
mining hypothesis irrespective of C substrate (litter or
SOC), but only in the forest site. Further, P addition
effects on SOC support the stoichiometric hypothesis;
P availability appeared key to soil C release (priming)
in the forest site if labile C and N is available. These
results clearly link previously contrasting hypotheses
of the factors controlling SOC with the natural gra-
dient in litter quality and nutrient availability that
exists in ecosystems at different successional stages. A
holistic theory that incorporates this variability of
responses, due to different mechanisms, depending

710
 Soil Carbon Dynamics in a Chronosequence
on nutrient availability and substrate quality is
essential for devising management strategies to safe-
guard soil C stocks.
INTRODUCTION
Soils currently act as a major carbon (C) sink and
play a vital role in climatic stability (Trumper and
others 2009), but there is increasing concern that
anthropogenically driven climate changes will con-
vert soils from functioning as a C sink to a C source,
exacerbating climate change (Cox and others 2000;
Solomon and others 2008). Effective management of
soil C is therefore vital, but our understanding of the
factors affecting soil C dynamics remains frag-
mented. Although nutrient availability and sub-
strate quality are recognized as important drivers of
soil C dynamics (van Groenigen and others 2006;
Bradford and others 2008), soils vary greatly in these
properties (for example, during ecosystem succes-
sion) and the focus of most studies on a single soil
type or site, limits generalization. Currently, there is
no universally accepted framework to explain how
these factors interact to affect soil organic carbon
(SOC) dynamics and multiple hypotheses have been
postulated to justify the often contrasting observa-
tions (Craine and others 2007).
A prominent question lacking consensus is the
effect of labile C on SOC stocks. This is a key issue
in the context of climate change because elevated
levels of atmospheric CO2 are known to enhance
rhizodeposition of plant-derived labile C (Phillips
and others 2006; Pollierer and others 2007).
Priming effects, that is, stimulated decomposition of
existing soil organic matter by the addition of labile
C substrates (Dalenberg and Jager 1989; Kuzyakov
2010), have been shown to increase soil respiration
(Heath and others 2005) and decrease SOC stocks
(Hoosbeek and others 2004). Conversely, several
studies report that labile C inputs to soils do not
accelerate SOC decomposition (Wu and others
1993; Fontaine and Barot 2005). To explain this
discrepancy the ‘preferential substrate utilization’
hypothesis has been proposed, which states that at
high labile C concentrations microbial community
composition shifts towards dominance by microbes
unable to mineralize complex SOC compounds,
and represses those that do (Fontaine and others
2004; Fontaine and Barot 2005). Indeed, C avail-
ability has been found to shape microbial commu-
nities (Fierer and others 2007). In an attempt to
reconcile our understanding of the effects of C
availability and the priming effect, Blagodatskaya
711
Key words: carbon sequestration; priming effect;
microbial mining; succession; microorganisms; lit-
ter decomposition.
and Kuzyakov (2008) suggest that priming may
occur at low C addition rates whilst preferential
substrate utilization at higher rates. However,
overall, there is little empirical evidence to support
the preferential substrate hypothesis and it is un-
known whether C entering the soil system as litter
(particulate C) and the old and more mineral-
bound/humified carbon (recalcitrant C) respond
similarly to increased additions of labile C.
Two contrasting hypotheses have also been put
forward to explain the impact of changes in nutrient
availability on SOC decomposition. The first
hypothesis, also known as the stoichiometry hypoth-
esis, assumes that microbial decomposition is driven
by the relative availability of nutrients. Although
relatively little is known about the role of P in stoi-
chiometric control of terrestrial ecosystem func-
tions (Craine and others 2007), it is common for the
C-to-N ratio to be used in predicting litter decay
dynamics. In N-limited decomposer systems
zincreased C input, resulting in a higher C-to-N ratio
should slow down decomposition, whereas N input
resulting in a lower C-to-N ratio should increase
decomposition and decrease soil C sequestration.
The second hypothesis, which is known as the
‘microbial mining hypothesis’, states that increased
nutrient availability slows down decomposition of
recalcitrant C by suppressing microbial activities
which break down recalcitrant litter compounds
such as lignin to acquire N. This response may be due
to changed enzyme production (Carreiro and others
2000; Waldrop and others 2004), altered microbial
community composition (Moorhead and Sinsab-
augh 2006), reduced microbial biomass (Treseder
2008) or a combination of these factors. Indeed,
there is evidence that mining for N depends on N
availability and litter quality (Knorr and others
2005; Craine and others 2007). According to this
hypothesis, excess N should increase ecosystem C
stocks in systems where litter inputs contain
recalcitrant compounds, but decrease it where litter
inputs are dominated by more decomposable com-
pounds (for example, Manning and others 2008). In
contrast to N there is no evidence for P mining be-
cause P is not associated with recalcitrant carbon
(McGill and Cole 1981). The implication of these
hypotheses, if supported, is that the factors control-
ling soil C turnover vary spatially depending
712 A. Milcu and others
on natural patterns of resource availability, and
temporally according to environmental changes (for
example, N and CO2 enrichment) and successional
time, in which changes in nutrient limitation and
litter quality take place (Walker and del Moral 2003).
To uncover general patterns in the regulation of C
dynamics by the availability of nutrients and labile C,
we conducted a multifactorial experiment exploring
how two forms of C substrate (fresh litter vs. soil
organic matter) are affected by the addition of labile
C and nutrients (N and P). These additions were re-
peated across the natural gradient of soil nutrient
availability and substrate quality of a secondary
succession chronosequence. Within a recently cul-
tivated field (early), a mid-successional grassland
(mid) and a 60-year-old deciduous woodland (late
successional stage), we established six subplots
receiving resource amendments (C, N, P, CN, CP and
control) where we assessed SOC dynamics and litter
decomposition. Using litterbags containing litters of
different qualities from the dominant plant species of
each successional stage (Raphanus raphanistrum,
Dactylis glomerata and Quercus petrea representing the
early, mid and late successional stage, respectively)
we were able to detect interaction effects between
litter quality, successional stage and nutrient avail-
ability on SOC dynamics. We then used these data to
explore whether the support for contrasting
hypotheses (priming hypothesis and preferential
substrate hypothesis for labile C additions, and the
stoichiometric and microbial mining for N and P
additions) depends on site nutrient availability and C
substrate type and quality.
METHODS
Study Sites
The study area was located at Silwood Park, South
East England (longitude 035¢W, latitude 51
Table 1. Chemical Soil Characteristics in the Early (Recently Cultivated Field), Mid (Grassland) and Late
(60-Year-Old Deciduous Woodland) Successional Sites
Characteristics/site
Soil pH
C% (C-to-N ratio)
DIN (mg/kg)
DON (mg/kg)
P% (N-to-P ratio)
Available P (mg/l)
The pH was measured using a standard soil water extraction. C% and N% were measured using a CNS elemental analyzer (Thermo Scientific Flash EA 1112 series). Nitrate
and ammonium N were extracted from the soil with potassium chloride solution. The nitrate nitrogen is measured by cadmium reduction to nitrite nitrogen, which is
determined spectrophotometrically after formation of a diazo compound between nitrite and sulphanilamide. The ammonium N is determined by formation of indophenol blue
by reaction with hypochlorite and phenol. Organic N was converted to inorganic N by persulphate oxidation and the resulting nitrate N was measured as above. Available P
was estimated using the Olsen method.
25¢N) at an elevation of 68 m. The three succes-
sional sites are found within 200 m of each other
and all lie on sandy deposits, the Bagshot sands.
They consist of a regularly tilled site left fallow for
1 year (the early successional site), mature grass-
land (mid successional site) and a deciduous forest
(late successional site) (Table 1). The early suc-
cession site was continuously cultivated since
1947 and left fallow in autumn 2006. The vege-
tation consists of a mixture of ruderal herbs
(Raphanus raphanistrum, Plantago lanceolata) and
early successional grassland species (Poa spp., Vicia
hirsuta). The mid-successional site has been
grassland since 1947 and is covered by typical
grassland vegetation dominated by the perennial
grass Dactylis glomerata. The late successional site
was also cultivated until 1947 then left fallow. A
myxoma virus outbreak in the 1960s resulted in
low grazing pressure by rabbits and tree recruit-
ment. Quercus petraea and Betula pendula are the
dominant species. Details of the soil chemical
characteristics for the three sites are given in Ta-
ble 1. Due to the closeness of the sites (200 m)
and their shared geology, differences in environ-
mental conditions between the sites are consid-
ered negligible and any variations in temperature,
soil moisture, shading and so on are assumed to
be caused by differences in successional develop-
ment.
Experimental Design
The experiment had a nested factorial design with
three treatments: (1) successional stage (S), with
three levels (early, mid and late), (2) resource
amendment (RA) with six levels (C, N, P, CN, CP
and control) and (3) litter type (L) with three levels
(Raphanus raphanistrum as high, Dactylis glomerata as
medium and Quercus petrea as low quality). L was
nested within RA and RA nested within S. In each
Early Mid Late
5.5 5.3 4.3
1.91% (12.1) 3.2% (12.0) 7.4% (13.3)
3.6 3.6 1.6
30.4 22.2 7.4
0.03% (13.4) 0.09% (3.0) 0.03% (5.2)
13.3 83.3 10.0
 Soil Carbon Dynamics in a Chronosequence 713

successional stage two plots (Plot) of approximately Sampling and Analytical Procedures
5.5 9 3.5 m2 were fenced to prevent access by
The litter bags were collected after 12 months of
rabbits and deer. Within each plot of the succes-
field exposure, soaked in water and washed for
sional stage treatment six RA subplots of
1.1 9 1.1 m were established in October 2006. RA 30 min to minimize any mineral/soil contamina-
aimed to eliminate C, N and P nutrient limitation of tion, dried 3 days at 60C and weighed. Three 5-cm
deep soil cores (5-cm diameter) were used to sample
soil microbial community. The basis for calculating
each fertilizer treatment in October 2007, sieved
the RA amendments was the successional stage
through a 2-mm diameter mesh to remove any root
with the highest C input per year (the forest) using
fragments, homogenized and then used for micro-
the Jenkinson and others (1991) estimate of
bial biomass and soil chemistry analyses. Dry sam-
approximately 912 g m-2 y-1 in a cool temperate
ples of soil and litter for measuring total C and N
forest. The N and P additions were based on the
macronutrient ratio of microbial biomass: approx- were milled using a Retsch 400 MM ball mill and
imately 10:2:1 for C, N and P, respectively analyzed using a CNS elemental analyzer (Thermo
Fisher Scientific Inc). Microbial biomass (Cmic;
(Anderson and Domsch 1980). This resulted
lg C g-1 soil) was determined by substrate-induced
in annual rates of 182.4 g m-2 y-1 of N and
respiration (SIR; Anderson and Domsch 1978) after
90.12 g m-2 y-1 of P. Although these rates are
the addition of glucose to saturate the glycolytic
higher than those used in typical agricultural fer-
enzymes of microbes (8 mg glucose g-1 dry wt soil)
tilization they aim to compensate for the high
using an automated oxygen (O2) microcompensa-
nutrient losses that are to expected at this site. The
soils are freely draining and are derived from the tion apparatus (Scheu 1992). O2 production at the
Bagshot sands, which are known to have high rates maximum initial respiratory response (MIRR) was
used to calculate microbial biomass as Cmic =
of nutrient leaching and an extremely low nutrient
40 9 MIRR according to Anderson and Domsch
sorption capacity (Manning and others 2006).
(1978) and Beck and others (1997). Microbial
There were no signs that nutrient availability was
growth slopes (slopes of CO2 increase calculated
at toxic levels. The nutrients were applied in a 3-l
from O2 production after MIRR) were used to con-
solution fortnightly, starting in October 2006 until
firm that the fertilizer applications reduced the
October 2007. To eliminate any moisture effects the
control treatment was also sprinkled with 3 l of respective microbial nutrient limitations after addi-
water in every nutrient application. C was applied tional nutrients (C, CN, CP and CNP), as done pre-
viously by Scheu and Parkinson (1995).
as glucose monohydrate (C6H12O6ÆH2O) (105 g m-2
DNA was extracted from 250 mg frozen soil
per application), N as ammonium nitrate (NH4NO3)
using Powersoil DNA isolation kits (MO BIO Lab-
(22 g m-2 per application), and P as sodium phos-
oratories, Inc., USA) according to the manufac-
phate (NaH2PO4ÆH2O) (17 g m-2 per application).
turer’s instructions. The purified DNA was used as a
Nested within the nutrient addition treatment
template for PCR amplification of general bacterial
were the litter type treatments. Senescent litter
materials of the three plant species were collected ribosomal RNA genes (16S rRNA) using the fol-
in autumn 2006 and dried for 3 days at 50C. lowing primers, 341F (5¢-CCTACGGGAGGCAG
CAG) and 534R (5¢-ATTACCGCGGCTGCTGG)
Leaves and shoots were used for Raphanus (43.9%
(Muyzer and others 1993). The PCR mix (50 ll/
C, 7.7% lignin, C:N = 30.8, N:P = 5.9) and Dactylis
reaction) consisted of 1 ll of each primer (10 mM),
(45.6% C, 9.2% lignin, C:N = 34.9, N:P = 14.5)
5 ll 109 Immobuffer (Bioline, London, UK), 2 mM
and only leaves for Quercus (48.4% C, 23.7% lignin,
MgCl2, 0.4 mM deoxynucleoside triphosphates
C:N = 42.0, N:P = 12.8), thus simulating the usual
(dNTP) mix and 1 U of Immolase DNA polymerase
form of annual inputs. The treatments gave a total
of 216 litter bags [3 successional stages 9 2 sites per (Bioline, London, UK) and 1 ll of extracted DNA.
successional stage (S) 9 6 resource amendments The cycle conditions were 95C for 8 min, 30 cycles
of 95C 30 s, 50C 45 s, 72C 45 s and a final
(RA) 9 3 litter species (L) 9 2 litter bags litter per
extension of 72C for 5 min. Amplified fragments
litter species] with different levels of replication for
with different sequences (assumed to be repre-
the main effects of each treatment; 2 per treatment
senting different bacterial taxa) were separated
level for S, 6 for RA and 72 for L. The bags had a
using denaturing gradient gel electrophoresis
mesh size of 2 9 2 mm and contained 3-g dry
(DGGE) as described previously (Ellis and others
weight of plant material and were placed in the
field by securing them with a peg on the soil sur- 2003). The INGENY phorU2 system (Netherlands)
face in November 2006. was used. Samples were loaded onto 10% acryl-
714 A. Milcu and others

amide-bisacrylamide (37.5:1) gels with denaturing terms, followed by Tukey’s HSD posthoc test. This
gradients from 30 to 60% (where 100% is 7 M urea allowed us to compare levels of treatments with
and 40% [v/v.] deionized formamide) in 19 TAE more than two levels. For graphical representation
electrophoresis buffer. Electrophoresis was per- of data, we used Statistica 8 package (StatSoft Inc).
formed at 100 V at a temperature of 60C for 18 h.
Gels were then stained with SBYR Gold (Cam-
RESULTS
bridge BioScience) in 19 TAE for 30 min at room
temperature and visualized under UV illumination. The slope of the microbial respiration rates (see
Digital images of DGGE gels were analyzed using ‘‘Methods’’) was significantly lower when the
GelCompar II V5.10 software (Applied Maths, additional substrate amendments (SA) matched the
Belgium). The software accounts for slight run respective resource amendment (RA), thus con-
anomalies and will allow multiple gels to be nor- firming that the RA applications reduced the
malized using standard lanes run on each gel. The respective microbial nutrient limitation (significant
position and intensities of the bands for each RA 9 SA interaction, F15,90 = 5.34, P < 0.001;
sample were used for comparisons between them. Figure 1A).
Pairwise similarities between samples were calcu-
lated using the dice coefficient, allowing for 0.5%
variation in band position. Graphical representa-
tions of similarity matrices were produced using the
unweighted pair group method using arithmetic
averages (UPGMA). We tried to perform a similar
analysis for fungal community composition, how-
ever, despite several attempts the resulting gels
were of poor quality and the results unpublishable.

Statistical Analysis
Mixed effects models, as implemented in the R
Statistical package (lme function, R version 2.10.1;
Pinheiro and Bates 2009), were used to analyze the
fixed effects of successional stage (S), resource
amendment (RA), litter type (L) and their interac-
tion on the amount of litter mass remaining, soil C
and microbial biomass. The random-effects of the
initial (maximal) model, which was fit by maxi-
mum likelihood, had a RA within S within plot
structure to account for pseudoreplication result-
ing from the nested design [R model: lme
(y  S *RA *L, random = 1/Plot/S/RA)]. The
slopes of respiration rate during microbial growth
were compared in a similar way, but replacing the
RA treatment with the additional substrate
amendment treatment (SA) as an explanatory
categorical variable with four levels (C, CN, CP and
CNP). These maximal models were reduced to
minimum adequate models, containing only sig- Figure 1. Slope of substrate induced microbial respira-
nificant terms by sequentially removing non-sig- tion (SIR), as affected by resource amendments with
nificant terms (starting with highest-order labile carbon (C) and nitrogen (N) and phosphorus (P)
interactions) from the maximal model and com- nutrients in the field and substrate additions (C, CN, CP
and CNP) in the laboratory (A). Percentage litter
paring each model with its predecessor using like-
remaining after 12 months of field exposure in litterbags
lihood ratio tests (Crawley 2007), until only as affected by B resource amendments to sites repre-
significant terms remained. To explore interactions senting different successional stages and C litter types.
that had been found to be significant in the mini- Means presented are averaged across other treatments
mum adequate models we performed factorial not shown, therefore, for each bar n = 6 in (A) and
ANOVAs on subsets of significantly interacting n = 12 in (B, C). Error bars represent ±SE.
 Soil Carbon Dynamics in a Chronosequence 715

Litter Decomposition
Successional stage (S) significantly affected litter
decomposition after 12 months, with the amount
of litter remaining in the mid-successional site
(40.6%) being significantly lower than in the early
(49.1%) and late successional sites (50.1%) (Tukey
HSD < 0.05; Table 2). Although litter mass loss
depended on the interaction between S and re-
source amendment (S 9 RA, Table 2; Figure 1B),
the C and CP treatments generally retarded
decomposition (48.5 and 61.7% litter remaining)
compared to the control (34.5%, Tukey’s
HSD < 0.05, Figure 1B), in line with the prefer-
ential substrate hypothesis. Alleviating N limitation
with the addition of N and CN generally retarded
decomposition in the late successional site (52.4
and 54.9% litter remaining compared to 34.8% in
control, Tukey’s HSD < 0.05) (Figure 1B), which
is in line with the microbial mining hypothesis.
Different litter types decomposed at different rates,
with Raphanus decomposing fastest: 37.9% litter
remaining versus 53.4% and 48.7% for Dactylis and
Quercus, respectively, but no significant interaction Figure 2. Soil carbon (C) percentage (A) and C-induced
between litter and resource amendment was found soil microbial biomass (Cmic) (B) as affected by site
(L 9 RA, Figure 1C; Table 2). (successional stage) and resource amendment treatments
after 12 months of field exposure. For each bar n = 6.
Error bars represent ±SE.
Soil Carbon
Resource amendment only affected SOC content in HSD < 0.05), thus supporting the microbial min-
the late successional stage (S 9 RA, F10,15 = 4.22, ing hypothesis, but only for soils where more
P < .01; Figure 2A), with the low SOC soils of the recalcitrant, lignin-rich inputs were the norm. N
early and mid successional sites being unaffected addition also led to more similar (convergent)
(P > 0.05). At the late site, additional N signifi- microbial community compositions (Figure 3),
cantly increased SOC content from 6.2% in the suggesting that changes in bacterial community
control (Ctr) to 7.6% in the N treatment (Tukey’s composition were contributing to the reduced
degradation of recalcitrant C under N addition.
These N effects were absent when N addition was
Table 2. Significance of the Treatment Effects on
Litter Decomposition combined with labile C addition, which singly did
not affect SOC content but when combined with N
Source/treatment Litter remaining or P reduced it. Adding both C and N (CN treat-
after 12 months ment) decreased the soil C content from 6.2% in
Ctr to 5.0% (Tukey’s HSD < 0.05) suggesting a
Intercept F1,144 = 2832.97***
S F2,3 = 11.04*
priming effect of C when N is available. Adding P
RA F5,15 = 17.23*** also reduced SOC content to 4.0% in the late suc-
L F2, 144 = 27.25*** cessional soil that had the highest N-to-P ratio of
S:RA F10,15 = 3.07* the three sites, indicating a stoichiometric response.
S:L F4,144 = 15.15*** The addition of CP similarly decreased soil C con-
RA:L F10,144 = 1.82+ tent to 4.4%, and there was no significant differ-
S:RA:L F20,144 = 1.52+ ence between the P and CP treatments (Tukey’s
HSD > 0.1, Figure 2A).
***P < 0.001, **P < 0.01; *P < 0.05; +P < 0.10.
F values shown are from an analysis of variance on the mixed effects model,
showing the estimated significance of succession (S), resource amendments (RA), Microbial Biomass
litter type (L) and their interactions on the remaining mass of litter after
12 months of field exposure and percentage carbon (C) (expressed as percentage of Analysis of SIR microbial biomass C (Cmic)
initial).
showed successional (F2,3 = 36.40, P < 0.01) and
716 A. Milcu and others

Dice(fuzzy)(area) (Opt:0.75%) (Tol 1.3%-1.3%) (H>1.0% S>1.0%) [0.0%-100.0%] Figure 3. Similarity of

Bacterial 16S Bacterial 16S
bacterial community
composition based on

100
30

40

50

60

70

80

90
DDGE profiles of bacterial
Mid Control 16 s DNA as affected by
Mid P site (successional stage)
Mid C P and resource amendment
Mid C treatments.
Early C P
Early P
Early C
Early Control
Late C P
Late P
Late C
Late Control
Late C N
Late N
Mid C N
Mid N
Early C N
Early N

RA effects (F5,15 = 47.74, P < 0.001), though the DISCUSSION

pattern of response was generally consistent
across successional stages, with the CN treatment
causing the biggest gains to microbial biomass
and the N treatment reducing it the most (Fig-
ure 2B). The interaction between S and RA was
also significant (F10,15 = 4.12, P < 0.01), with the
highest Cmic in the mid-successional stage when
CN was added and the lowest in the early and
mid successional stage when N was added (Fig-
ure 2B). However, the pattern of response in Cmic
did not relate to any RA effects on SOC or litter
decomposition, suggesting that there was a weak
relationship between microbial biomass and its
activity.
Bacterial Community Composition
Comparison of bacterial communities in soil sam-
ples from the experimental plots, as assessed by
DGGE of 16S rRNA gene amplicons, demonstrated
that both successional stages and nutrient addition
affected community structure. Figure 3 shows that
samples from the same successional stage are
more similar to each other than to communities
from other successional stages. Furthermore, this
result is not markedly affected by the addition of
C, P or CP. However, the addition of N (whether
or not in combination with C) causes a shift in
bacterial community composition in all succes-
sional stages.
Our results support our central proposal that con-
trasting observations and support for various
hypotheses in studies investigating the impact of
additional labile C and nutrients on soil organic
carbon (SOC) stocks, results mainly from natural
variations in soil nutrient limitation and the type of
C substrate (litter vs. soil C). We show that despite
the complexity of the interactions between the
addition of labile C, nutrients, successional stage
and C substrate, general patterns emerge (Table 3).
However, the results should be interpreted with
caution because like the majority of other studies
trying to indentify changes in soil C pools, the
statistical power of our study is insufficient (<0.80)
to detect subtle changes (Strickland and others
2009).
Effects of Labile C Additions
There is now substantial evidence that the domi-
nant form of belowground C input is labile C
compounds from rhizodeposition (van Hees and
others 2005; Högberg and Read 2006; Boddy and
others 2007). However, little is known about how
rhizodeposits affect soil C turnover, or to what
extent their effects are modified by nutrient avail-
ability. In our experiment, the addition of labile C
alone did not impact soil C levels after 1 year.
However, in combination with additional nitrogen
 Soil Carbon Dynamics in a Chronosequence 717

Table 3. Simplified Summary Table of the Support for the Different Hypotheses Concerning the Impact of
Resource Amendments (RA) Supplying Excess of Labile Carbon (C), Nitrogen (N) and Phosphorus (P) when
Applied to Simple (Easily Degradable) Litters Originating from Early and Mid-Successional Stages, Complex
(Recalcitrant) Litters from the Late Successional Stage and Soil Organic Carbon (SOC) Derived from These
Sources
C substrate/RA Hypothesis supported

Labile C N P

Non-recalcitrant litters Preferential substrate (P) No detectable effect No detectable effect

Recalcitrant litters Preferential substrate (P) Microbial mining Stoichiometric
SOC originating from simple litters No detectable effect No detectable effect No detectable effect
SOC originating from recalcitrant Priming (N) Priming if labile C available Priming (labile C)
litters (labile C)
Microbial mining

In brackets we note the next limiting resource.

(CN treatment) it enhanced soil C loss, but only at
the late successional site. This suggests that the
priming effect of additional C depends on N avail-
ability, with our results indicating that the priming
effect is more important in late successional soils,
where a larger proportion of the SOC is derived
from recalcitrant material such as lignin. In con-
trast, the addition of readily available C (in the C
and CP treatments) retarded the decomposition of
litter, providing support for the preferential sub-
strate hypothesis. We acknowledge that the lack of
data on shifts in microbial community structure
and enzyme activity at the litter level prevents
inferences about whether these changes were due
to changes in community composition or func-
tioning.
Effects of N Additions
The microbial mining hypothesis predicts that the
decomposition of recalcitrant litter compounds is
stimulated by increased availability of labile C un-
der low N availability (Craine and others 2007).
Our results confirm the findings of Sinsabaugh and
others (2002) and Craine and others (2007) in
showing that N mining operates in systems with
slowly decomposing litters rich in lignin and tannin
and other condensed carbon compounds. Interest-
ingly, we found support for microbial N mining of
both litter and soil C, but only at the late succes-
sional site, where inputs of lignin-rich compounds
and their derivatives are likely to dominate the
SOC; the percentage of soil C in the late succes-
sional stage increased from 6.5 to 7.5 in just 1 year
of N addition. However, when C and N were added
together they contrastingly affected the soil C (was
reduced from 6.5 to 4.4%). This suggests that the
concurrent availability of both labile C and mineral
N results in a strong priming effect on soil C,
resulting in marked C losses from soil organic
matter. In contrast, we found that in the more early
successional stages, where inputs and their deriv-
atives are more inherently decomposable that the
availability of labile C and N is less important for
SOM dynamics. Although we detected significant
shifts in bacterial composition under the N treat-
ment of these sites it appeared to have little con-
sequence for SOM dynamics. This may be related to
the fact that recalcitrant compounds, such as lignin
and humified materials, make up a far lower pro-
portion of the SOC in these systems in comparison
to systems dominated by woody plants, or due to
the fungal dominance of decomposition processes,
though this would be expected of later successional
ecosystems, which tend to be more fungal domi-
nated (Wardle 2002).
Several non-exclusive mechanisms have been
proposed to explain the mechanisms underlying
microbial mining. (1) Increased N availability re-
duces the production of enzymes, such as phenol
oxidase and peroxidase, responsible for the
decomposition of recalcitrant C compounds (Craine
and others 2007). (2) N mining is modified by
competition between microorganisms for available
N and labile C resulting in a microbial community
shift from fungal (especially white rot basidiomy-
cetes) to bacterial at elevated N availability (Wal-
drop and others 2004). Microbial data indicate
significant shifts in microbial community compo-
sition associated with N addition. However, we
cannot tell if this was caused by biotic (competition
between microorganisms) or abiotic (for example,
718 A. Milcu and others
the reduction in soil pH caused by ammonium ni-
trate additions) drivers. In particular, soil acidifi-
cation caused by N fertilizer applications has been
shown to strongly impact bacterial community
composition (O’Donnell and others 2001; Rousk
and others 2010).
There was also evidence that nutrient amend-
ments alleviated nutrient limitations, as indicated
by changes in microbial biomass in response to
fertilizer addition and shallower slopes of microbial
growth when lab fertilization matched the respec-
tive field addition. However, no clear relationships
between microbial biomass or the slopes of micro-
bial growth and decomposition or N mining were
evident. This lack of correlation is most likely a
result of not capturing information on microbial
functional responses. Measurements of microbial
enzyme activity may be better indicators of
microbial functional responses than biomass mea-
sures as they contain information on both physio-
logical and community composition shifts, but we
did not cover this aspect.
Effects of P Additions
We did not find any support for microbial P mining.
These results are in line with those of Craine and
others (2007) who suggested that the impacts of P
addition on organic matter decomposition follow
the stoichiometric hypothesis. In our study, P was
not sufficiently limiting for its addition to affect
microbial processes in the early and mid-succes-
sional stage. However, we found a strong interac-
tion between labile C and P addition in the late
successional stage. This led to contrasting effects on
litter mass loss (which decreased under CP) and the
release of C from litter and soil (which increased
under CP). The results support the findings of
Cleveland and Townsend (2006) and Bradford and
others (2008), who found that P availability is key
to soil C release (priming) when N is not limiting
and additional labile C enters the soil.
Conclusions and Implications
In conclusion, our findings confirm the overarch-
ing idea that contrasting results regarding how lit-
ter decomposition and soil C dynamics are affected
by labile C are a result of different decomposer re-
source limitations at different sites. Although some
of the emerging patterns are in line with previous
studies, this study links previously contrasting
hypotheses with the natural gradient in litter
quality and nutrient availability that exists in eco-
systems at different successional stages (Figure 3).
It shows that many of the hypothesized mecha-
nisms operate in tandem under natural conditions,
but that their relative importance depends greatly
upon initial soil conditions. Further studies
encompassing a wider range of sites with different
soils, vegetation, nutrient limitations and C sub-
strates are necessary for further investigation of the
observed patterns and should allow us to gain a full
mechanistic understanding of the factors affecting
SOC stocks. Such an understanding may allow
modelling soil C dynamics in a way that synthesizes
these fragmented hypotheses by incorporating the
amounts of C, N and P in a variety of forms with
the dynamics of the decomposer community. The
acquisition of such knowledge is becoming
increasingly important as anthropogenic impacts
such as soil fertilization with N and P, and the input
of labile C via root exudates are predicted to in-
crease in future. Understanding how SOC stocks
from different ecosystems will respond to these
changes is essential if we are to reliably predict the
C sink potential of soils and devise management
strategies to more effectively sequester soil C.
ACKNOWLEDGMENTS
We thank the NERC Centre for Population Biology
for funding and Tom Sloan, Melanie Wessels, Mark
Saunders, Kim Prior, Gary Needham and Callum
Brandon for helping with the fertilizer applications
and sample analyses.
REFERENCES
Anderson J, Domsch K. 1978. A physiological method for the
quantitative measurement of microbial biomass in soils. Soil
Biol Biochem 10:215–21.
Anderson J, Domsch K. 1980. Quantities of plant nutrients in
the microbial biomass of selected soils. Soil Sci 130:211.
Beck T, Joergensen RG, Kandeler E, Makeschin F, Oberholzer
HR, Nuss HW, Scheu S. 1997. An inter-laboratory comparison
of ten different ways of measuring soil microbial biomass C.
Soil Biol Biochem 29:1023–32.
Boddy E, Hill P, Farrar J, Jones D. 2007. Fast turnover of low
molecular weight components of the dissolved organic carbon
pool of temperate grassland field soils. Soil Biol Biochem
39:827–35.
Blagodatskaya E, Kuzyakov Y. 2008. Mechanisms of real and
apparent priming effects and their dependence on soil
microbial biomass and community structure: critical review.
Biol Fertil Soils 45:115–31.
Bradford M, Fierer N, Reynolds J. 2008. Soil carbon stocks in
experimental mesocosms are dependent on the rate of labile
carbon, nitrogen and phosphorus inputs to soils. Funct Ecol
22:964–74.
Carreiro M, Sinsabaugh R, Repert D, Parkhurst D. 2000.
Microbial enzyme shifts explain litter decay responses to
simulated nitrogen deposition. Ecology 81:2359–65.
 Soil Carbon Dynamics in a Chronosequence
Cleveland C, Townsend AR. 2006. Nutrient additions to a trop-
ical rain forest drive substantial soil carbon dioxide losses to
the atmosphere. Proc Natl Acad Sci USA 103:10316.
Cox PM, Betts RA, Jones CD, Spall SA, Totterdell IJ. 2000.
Acceleration of global warming due to carbon-cycle feedbacks
in a coupled climate model. Nature 408:184–7.
Craine J, Morrow C, Fierer N. 2007. Microbial nitrogen limita-
tion increases decomposition. Ecology 88:2105–13.
Crawley M. 2007. The R book. New York: Wiley.
Dalenberg J, Jager G. 1989. Priming effect of some organic
additions to 14C-labelled soil. Soil Biol Biochem 21:443–8.
Ellis R, Morgan P, Weightman A, Fry J. 2003. Cultivation-
dependent and-independent approaches for determining
bacterial diversity in heavy-metal-contaminated soil. Appl
Environ Microbiol 69:3223.
Fierer N, Bradford MA, Jackson RB. 2007. Toward an ecological
classification of soil bacteria. Ecology 88:1354–64.
Fontaine S, Barot S. 2005. Size and functional diversity of mi-
crobe populations control plant persistence and long-term soil
carbon accumulation. Ecol Lett 8:1075–87.
Fontaine S, Bardoux G, Abbadie L, Mariotti A. 2004. Carbon input
to soil may decrease soil carbon content. Ecol Lett 7:314–20.
Heath J, Ayres E, Possell M, Bardgett RD, Black HIJ, Grant H,
Ineson P, Kerstiens G. 2005. Rising atmospheric CO2 reduces
sequestration of root-derived soil carbon. Science 309:1711–13.
Högberg P, Read D. 2006. Towards a more plant physiological
perspective on soil ecology. Trends Ecol Evol 21:548–54.
Hoosbeek M, Lukac M, van Dam D, Godbold D, Velthorst E,
Biondi F, Peressotti A, Cotrufo M, de Angelis P, Scarascia-
Mugnozza G. 2004. More new carbon in the mineral soil of a
poplar plantation under Free Air Carbon Enrichment (POP-
FACE): cause of increased priming effect? Glob Biogeochem
Cycles 18:GB1040.
Jenkinson D, Adams D, Wild A. 1991. Model estimates of CO2
emissions from soil in response to global warming. Nature
351:304–6.
Knorr M, Frey S, Curtis P. 2005. Nitrogen additions and litter
decomposition: a meta-analysis. Ecology 86:3252–7.
Kuzyakov Y. 2010. Priming effects: interactions between living
and dead organic matter. Soil Biol Biochem 42:363–1371.
Manning P, Putwain PD, Webb NR. 2006. The role of soil phos-
phorus sorption characteristics in the functioning and stability of
lowland heath ecosystems. Biogeochemistry 81:205–17.
Manning P, Saunders M, Bardgett R, Bonkowski M, Bradford M,
Ellis R, Kandeler E, Marhan S, Tscherko D. 2008. Direct and
indirect effects of nitrogen deposition on litter decomposition.
Soil Biol Biochem 40:688–98.
McGill W, Cole C. 1981. Comparative aspects of cycling of or-
ganic C, N, S and P through soil organic matter. Geoderma
26:267–86.
Moorhead D, Sinsabaugh R. 2006. A theoretical model of litter
decay and microbial interaction. Ecol Monogr 76:151–74.
Muyzer G, de Waal E, Uitterlinden A. 1993. Profiling of complex
microbial populations by denaturing gradient gel electropho-
resis analysis of polymerase chain reaction-amplified genes
coding for 16S rRNA. Appl Environ Microbiol 59:695.
719
O’Donnell A, Seasman M, Macrae A, Waite I, Davies JT. 2001.
Plants and fertilisers as drivers of change in microbial com-
munity structure and function in soils. Plant Soil 232:135–45.
Phillips D, Fox T, Six J. 2006. Root exudation (net efflux of
amino acids) may increase rhizodeposition under elevated
CO2. Glob Change Biol 12:561–7.
Pinheiro JC, Bates DM. 2009. Mixed-effects models in S and S-
PLUS. Berlin: Springer.
Pollierer M, Langel R, Korner C, Maraun M, Scheu S. 2007. The
underestimated importance of belowground carbon input for
forest soil animal food webs. Ecol Lett 10:729–36.
Rousk J, Bååth E, Brookes PC, Lauber CL, Lozupone C, Caporaso JG,
Knight R, Fierer N. 2010. Soil bacterial and fungal communities
across a pH gradient in an arable soil. ISME J 4:1340–51.
Scheu S. 1992. Automated measurement of the respiratory re-
sponse of soil microcompartments: active microbial biomass in
earthworm faeces. Soil Biol Biochem 24:1113–18.
Scheu S, Parkinson D. 1995. Successional changes in microbial
biomass, respiration and nutrient status during litter decompo-
sition in an aspen and pine forest. Biol Fertil Soils 19:327–32.
Sinsabaugh R, Carreiro M, Repert D. 2002. Allocation of extra-
cellular enzymatic activity in relation to litter composition, N
deposition, and mass loss. Biogeochemistry 60:1–24.
Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt K,
Tignor M, Miller H. 2008. Climate change 2007: the physical
science basis. Cambridge, New York, Melbourne, Madrid,
Cape Town, Singapore, São Paulo, Delhi: Cambridge Univer-
sity Press.
Strickland MS, Lauber C, Fierer N, Bradford MA. 2009. Testing
the functional significance of microbial community composi-
tion. Ecology 90:441–51.
Treseder KK. 2008. Nitrogen additions and microbial biomass: a
meta analysis of ecosystem studies. Ecol Lett 11:1111–20.
Trumper K, Bertzky M, Dickson B, van der Heijden G, Jenkins
M, Manning P. 2009. The natural fix? The role of ecosystems
in climate mitigation. UNEP-WCMC. Cambridge: UNEP Rapid
Response Assessment. United Nations Environment Pro-
gramme.
Van Groenigen K, Six J, Hungate B, De Graaff M, Van Breemen
N, Van Kessel C. 2006. Element interactions limit soil carbon
storage. Proc Natl Acad Sci USA 103:6571–4.
van Hees P, Jones D, Finlay R, Godbold D, Lundström U. 2005.
The carbon we do not see—the impact of low molecular
weight compounds on carbon dynamics and respiration in
forest soils: a review. Soil Biol Biochem 37:1–13.
Waldrop M, Zak D, Sinsabaugh R, Gallo M, Lauber C. 2004.
Nitrogen deposition modifies soil carbon storage through
changes in microbial enzymatic activity. Ecol Appl 14:1172–7.
Walker L, del Moral R. 2003. Primary succession and ecosystem
rehabilitation. Cambridge: Cambridge University Press.
Wardle DA. 2002. Communities and ecosystems: linking the
aboveground and belowground components. Princeton:
Princeton University Press.
Wu J, Brookes P, Jenkinson D. 1993. Formation and destruction
of microbial biomass during the decomposition of glucose and
ryegrass in soil. Soil Biol Biochem 25:1435–41.