ARTICLE
Received 11 Jan 2016 | Accepted 19 Oct 2016 | Published 28 Nov 2016
Direct evidence for microbial-derived soil organic
matter formation and its ecophysiological controls
Cynthia M. Kallenbach
1,2
, Serita D. Frey
1
& A. Stuart Grandy
1
Soil organic matter (SOM) and the carbon and nutrients therein drive fundamental
submicron- to global-scale biogeochemical processes and influence carbon-climate
feedbacks. Consensus is emerging that microbial materials are an important constituent of
stable SOM, and new conceptual and quantitative SOM models are rapidly incorporating this
view. However, direct evidence demonstrating that microbial residues account for the
chemistry, stability and abundance of SOM is still lacking. Further, emerging models
emphasize the stabilization of microbial-derived SOM by abiotic mechanisms, while the
effects of microbial physiology on microbial residue production remain unclear. Here we
provide the first direct evidence that soil microbes produce chemically diverse, stable SOM.
We show that SOM accumulation is driven by distinct microbial communities more so than
clay mineralogy, where microbial-derived SOM accumulation is greatest in soils with higher
fungal abundances and more efficient microbial biomass production.
DOI: 10.1038/ncomms13630
OPEN
1
Department of Natural Resources and the Environment, University of New Hampshire, Durham, New Hampshire 03824, USA.
2
Soil and Crop Sciences
Department, Colorado State University, Fort Collins, Colorado 80523, USA. Correspondence and requests for materials should be addressed to C.M.K.
(email: kallenbachcm@gmail.com).
NATURE COMMUNICATIONS | 7:13630 | DOI: 10.1038/ncomms13630 | www.nature.com/naturecommunications 1
Corrected: Author correction
F
or nearly a century, soil organic matter (SOM) formation in
conceptual and quantitative models has been depicted
primarily as a function of plant inputs and their chemis-
try
1–3
. As such, chemically diverse and stable SOM originates
from the preservation of biochemically recalcitrant complex plant
polymers, such as lignin derivatives and long-chain lipids
2,3
.
However, soil microbial communities are adept at decomposing a
wide range of plant compounds and using the carbon (C) to
synthesize their own biomass. The importance of soil microbes in
processing plant inputs and synthesizing SOM is not conceptually
new
4–6
, though until recently it has largely been overlooked as a
primary pathway of SOM formation
7
. In a significant departure
from the dominant plant-based models of the past, microbial
contributions to SOM formation have gained widespread
acceptance, in part due to advances in molecular analytical
techniques
1,7,8
. Mounting evidence demonstrates that some
decomposition-resistant SOM bears little chemical resemblance
to plant material, but is instead characteristic of microbial cells,
excretions and cytoplasmic materials
9–11
stabilized via organo-
mineral and organo-metal oxide interactions
12,13
. While the
evidence to support this alternative pathway of SOM formation is
compelling
8–11,14,15
, analytical challenges associated with
separating direct microbial and plant inputs to SOM create
fundamental uncertainties about the degree to which microbial
residues contribute to the formation of stable SOM and its
characteristic chemical diversity
11,16,17
.
Whether plant inputs are first converted to microbial residues
before stabilization influences how SOM responds to land use and
climate change
1,18,19
as well as how it should be modelled
20
and
managed to promote climate change mitigation
21
. Plant residues
that accumulate in soil through physical protection (e.g., inside
aggregates) or in zones with low biological activity are susceptible
to destabilization following disturbances such as cultivation
or in response to environmental change (e.g., temperature
increases)
19,22
. If, however, plant materials are synthesized into
microbial proteins, lipids or polysaccharides, the resulting
organo-mineral associations may include ligand bonds or other
strong interactions
22
that have lower temperature sensitivity
13,23
and may better withstand perturbations. Despite the potential
importance of microbially derived SOM, experimental evidence
for microbial contributions to SOM formation is constrained by
methods that select for a limited group of microbial biomarkers,
such as amino sugars or select lipids
9,21
, is typically correlative
and inferential
9,21,24,25
, or relies on visualization techniques that
cannot easily be scaled up to a whole-soil basis
10
. These
constraints have hindered our ability to quantitatively deter-
mine the importance of microbial-derived compounds as key
proximate inputs to SOM, and thus limit our understanding,
management and predictions of SOM dynamics.
Further, many newly developed conceptual and quantitative
microbial models put considerable emphasis on the abiotic
stabilization
20,26,27
of microbial-derived SOM, yet the ecological
controls that regulate the transfer of microbial residues to
mineral-associated SOM have yet to be resolved
20
. While clay
mineralogy is known to regulate microbial-SOM accumu-
lation
22,28
, microbial community structure and physiology likely
also determine stable SOM accumulation rates due to differences
in microbial residue production
15,21,29
. For example, microbial
carbon use efficiency (CUE), the amount of C used for microbial
growth relative to total C uptake may have a direct impact on
microbial residue production
15,21
but can differ across resource
gradients
30
and microbial communities
31
. As CUE increases,
relatively more substrate-C goes towards biomass synthesis,
potentially increasing the amount of residues available for
stabilization
21,29
. Substrate chemistry can alter CUE due to its
direct effect on cellular metabolism, but may also drive CUE
indirectly, by selecting for distinct microbial communities with
different prevailing life histories
31
. For instance, highly reduced
substrate-C (high free energy) typically promotes higher CUE
within a community but may also select for a copiotrophic-
dominated community with an inherently lower CUE
31,32
. Thus,
microbial community composition, available substrates and CUE
are intimately connected, but the influence of their interactions
on microbial-SOM formation has not been well characterized
32
.
We use model soils to quantitatively assess whether microbial
processing of simple C (i.e., low-molecular-weight) substrates
alone, in the absence of complex plant compounds, can build
significant amounts of chemically diverse, stable SOM. Since our
model soils are initially C- and microbe-free, we eliminate the
difficulties of isolating microbial residues that occur when using
natural soils. We use a gradient of substrate-C inputs to represent
different microbial-available energy in order to facilitate the
development of diverging microbial communities and physiolo-
gies that are hypothesized to influence SOM chemistry and
accumulation rates. The substrate gradient includes monomeric
and dimeric sugars since they are abundant energy sources for
microbial metabolism in natural soils
33
, but also because their
rapid microbial uptake and intercellular breakdown should leave
little to no unaltered substrate in the soil that would interfere with
the detection of novel SOM molecules formed during the
experiment
33,34
. Further, we include a more recalcitrant lignin
monomeric substrate, as well as plant-derived dissolved organic
C (DOC), a natural analog to test whether substrate chemical
diversity is requisite to generating SOM chemical diversity. We
also compare two clay types (kaolinite or montmorillonite) to
investigate the effects of clay mineralogy on SOM accumulation
relative to microbial communities and substrate chemistry. We
characterize the composition of SOM after 18 months of
incubation using high-resolution molecular fingerprinting by
pyrolysis-gas chromatography/mass spectrometry (py-GC/MS)
to establish the chemical fingerprint of newly formed microbial
residues (including cell wall and cytoplasmic materials,
metabolites and extracellular excretions). Our model soils
accrue C concentrations between 1 and 1.4% C, with a
chemical diversity and stability characteristic of natural soils.
Substrate type has a stronger influence on SOM development
than clay mineralogy. However, this effect appears to be an
indirect consequence of diverging microbial communities, where
different substrates select for distinct microbial communities,
with microbial-SOM accumulation being greatest in soils where
fungal abundances are highest and microbial biomass production
is most efficient.
Results
Model soil systems. We created initially C- and microbe-free
model soils to study proximal drivers of SOM development.
Model soils were incubated with a natural soil microbial
community inoculum and received weekly C additions of glucose,
cellobiose, syringol (a lignin monomer) or plant-derived DOC,
combined with a nutrient solution for 15 months. Substrate
additions were terminated after 15 months, but the incubation
continued for an additional 3 months (18 months total) without
further C inputs in order to maximize endogenous C recycling.
Following inoculation of model soils, microbial activity was
immediately detectable by a measurable CO
2
efflux, and by 6
months post-inoculation microbial extracellular enzymatic activ-
ities, CO
2
respiration rates and living microbial biomass all
indicated that an active microbial community had been
established in all treatments (Supplementary Fig. 1). Visually,
model soils progressively resembled natural soil, with significant
colour and structural development (Fig. 1a). The exception to this
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13630
2 NATURE COMMUNICATIONS | 7:13630 | DOI: 10.1038/ncomms13630 | www.nature.com/naturecommunications
Syringol-treated soils
Time 0 18 months
0
20
40
60
80
100
DOC-treated soils
Relative abundance (%)
0
20
40
60
80
100
Field soil
0
20
40
60
80
100
Sugar-treated soils
Relative abundance (%)
0
20
40
60
80
100
Time 0 18 months
Time 0 18 months
Kaol.
Mont. Mont.
Kaol. Mont.
57 93 82
82
42
86 82
17 57
Unspecified
Aromatics
Lignin derivatives
Phenolics
Polysaccharides
N-bearing
Chitin
Lipids
Protein
Time0 days
15 months
2.5 cm
a
bc
de
Figure 1 | Soil development and organic matter chemistry. Images of sugar-treated model soils over time (a); the far left panel is an uninoculated sterile
kaolinite and sand mixture, and the far right panel is the same mixture, inoculated and treated with weekly glucose additions for 15 months. Relative
abundance of chemical compound groups in substrate (Time 0) and model soils amended with (b) sugar, (c) syringol and (d) plant dissolved organic
carbon (DOC). These are compared to soil collected from an agricultural field (e). Glucose and cellobiose treatments were averaged since there were no
significant differences in their chemistry (ANOVA: P40.05). Numbers above bars are the total number of identified compounds.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13630 ARTICLE
NATURE COMMUNICATIONS | 7:13630 | DOI: 10.1038/ncomms13630 | www.nature.com/naturecommunications 3
was the kaolinite–syringol treatment, which after the initial 3
months did not exhibit an active microbial community and was
thus removed from the study.
SOM chemistry. We used py-GC/MS to examine the chemistry
of added substrates, along with SOM chemistry at the
termination of the study (18 months). Substrate chemistry at
Time 0 provides a control tocomparechangesinSOM
chemistry following mi crobial processing. The visual changes
we observed in soil development (Fig. 1a) coincided with the
creation of microbial residues, specifically microbial-derived
proteins and lipids (Fig. 1b–d). After 18 months, regardless of
initial s ubstrate chemistry or clay type, accumulated SOM
largely consisted of microbial product s (Fig. 1), and the SOM
molecular diversity (number of compounds) increased by 54–
83% across our model soil systems, with the number of novel
compounds varying from 73 to 100% of the total (Supple-
mentary Fig. 2).
Initially, the unprocessed glucose and cellobiose substrates
(Time 0) contained primarily polysaccharides (80% relative
abundance) while the syringol substrate was largely comprised
of lignin derivatives (80% relative abundance) (Fig. 1;
Supplementary Note 1). After 18 months, polysaccharide relative
abundance in sugar-treated soils declined to 18%. Similarly, in
syringol-treated soils, lignin derivatives declined to 1% over that
time period. In both the sugar- and syringol-treated soils, the
dominant chemical signature of the substrate was replaced by
proteins, non-proteinaceous nitrogen compounds and lipids.
Lipids and proteins, not initially present in unprocessed
substrates, collectively made up 37% relative abundance of all
compounds in newly formed SOM (Fig. 1 and Supplementary
Table 1). Chitin, originating primarily from fungal cell walls
6
, also
increased by 5% after 18 months. We also observed increases in
both the aromatic and unspecified compound classes after 18
months across all soils. The most abundant compounds within
these classes are consistent with abundant compounds observed
in fungal biomass derived from natural soil (Supplementary
Table 2 and Supplementary Note 1).
Microbial processing of sugars produced SOM similar in
chemistry and diversity to soils amended with more recalcitrant
syringol, and began to resemble soils treated with more
heterogeneous plant-derived DOC, as well as a natural soil
(Figs 1 and 2). Further, microbial processing of DOC also
increased chemical diversity by 54% (Fig. 1). In contrast to
traditional expectations that slowly decomposing plant
compounds accumulate in SOM
2,3
, lignin derivatives, specific to
plant-derived lignin macromolecules, declined from 10 to 2.5%
relative abundance in DOC-treated soils, and proteins and lipids
increased five-fold (Po0.05).
Significant differences in SOM chemistry due to mineralogy
emerged after 18 months (Monte Carlo P o0.05; multi-response
permutation procedure Po0.05) (Fig. 2 and Supplementary
Tables 1 and 3). For example, kaolinite sugar-treated soils
had higher relative abundances of polysaccharides and chitin
derivatives and a lower abundance of aromatic compounds
compared with montmorillonite (Po0.05). However, there was
little influence of the substrate treatment on final SOM chemistry,
with, for instance, montmorillonite, sugar- and syringol-treated
soils exhibiting similar SOM chemistries at 18 months (Fig. 2).
SOM accumulation and stability. We determined soil organic
carbon (SOC) accumulation rates and stability in the initially
C-free model soils. Across all model soils, SOC increased con-
sistently over time (Po0.05; Table 1). The efficiency with which
substrate-C was converted to total SOC—i.e., the amount of SOC
remaining relative to the total amount of substrate added—
declined from an average across all treatments of 32%
±
1.4 at 6
months to 21%
±
0.75 at 18 months Using SOC stocks as an
integrator of mass C balance, the majority (475%) of total
substrate-C added was lost via respiration across all treatments by
18 months (Table 1). As a comparison, estimated cumulative
respirations represented 50–75% of added C with the exception of
DOC-treated soils (Supplementary Table 4). These estimates are
based on flux measurements, which exhibited rapid and ephem-
eral responses to each substrate addition, especially for the
monomeric inputs. In DOC-treated soils, cumulative respiration
underestimated C loss, compared with calculations using SOC
stocks. We attribute this to the more chemically complex DOC
treatment having a longer respiration response time that was not
completely captured from fluxes collected immediately following
substrate additions.
The final total soil C concentrations at the end of the C
addition period (1–1.4%) are well within the range of many
natural soils
21
(Table 1). The syringol-treated montmorillonite
soils accumulated the most C (Po0.001); however, there were no
differences in SOC among the glucose- or cellobiose-treated soils
in either clay type. DOC-treated soils had higher SOC compared
with sugar-treated soils (12.8 mg C per g soil, Po0.001), and also
exhibited differences among clay types (Po0.05; Supplementary
Table 3).
We determined SOC biological and chemical stability to
evaluate its potential long-term persistence. We assessed
biological stability by adding a
13
C-labelled substrate mixture
(1:1 glutamic acid:glucose at 25 atom % and 50 mg C per g soil) to
a subsample of soil from the main experiment and incubating for
3 months (Table 2). The labelled substrate enabled us to use a
standard isotope mixing model
35
to determine the amount of
previously formed C vulnerable to decomposition by an active
microbial community. Chemical stability of accumulated SOC
was assessed with an acid hydrolysis fractionation, where the
unhydrolysable fraction is considered chemically stable
36
. The
majority of SOC was biologically stable (71%
±
2.5; Table 2) and
40%
±
1 was chemically stable to oxidation, which is within the
upper range observed for natural soils
36,37
. We observed greater
biological stability in montmorillonite than kaolinite model soils
only within the cellobiose and DOC treatments (Po0.05;
Supplementary Table 3).
80
Field soil
Syringol time 0
Glucose
Cellobiose
Syringol
DOC
Montmorillonite
Kaolinite
Time 0
DOC
time 0
Cellobiose time 0
Glucose time 0
0
40
80
Axis 1 (74%)
40
Axis 2 (6%)
0
Figure 2 | Differences in soil organic matter chemistry between
substrate and clay types. Non-metric multidimensional scaling (NMDS)
ordination of the relative abundance of chemistry compounds at 18 months
for substrate and clay treatments. Open symbols are kaolinite and closed
symbols are montmorillonite. For comparison, unprocessed substrates
(Time 0) and an agricultural field soil are indicated by a star symbol
(Stress ¼ 8.1, Monte Carlo: Po0.05).
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13630
4 NATURE COMMUNICATIONS | 7:13630 | DOI: 10.1038/ncomms13630 | www.nature.com/naturecommunications
Microbial community composition and physiology. Given its
potential influence on SOM accumulation rates and chemistry, we
examined microbial community composition using phospholipid
fatty acid (PLFA) biomarkers. We observed significant differences
in broad microbial groups due to substrate treatment (Po0.05;
Supplementary Table 3 and Supplementary Fig. 3). Montmor-
illonite, syringol-treated soils had lower Gram-negative and
Gram-positive bacteria and higher fungal relative abundance
compared with sugar-treated soils (Po0.001; Fig. 3). We also
observed an effect of mineralogy on the microbial community;
fungal abundances were generally higher in kaolinite than in
montmorillonite soils, while Gram-negative bacteria were more
abundant in montmorillonite soils (Po0.001).
We evaluated community-level CUE to determine whether
the distinct community composition we observed between
substrates was related to differences in microbial physiology
and subsequently SOC accumulation. We estimated the CUE
using
13
C-labelled glutamic acid uptake into microbial biomass
and respired
13
C-CO
2
as a proxy for determining the proportion
of new C inputs used for microbial biomass synthesis
30,32
. There
was a strong effect of substrate type, where syringol-treated soils
exhibited the highest CUE (Po0.001; Fig. 3). Microbial
community composition, CUE and SOC accumulation were
highly correlated across substrate treatments (Fig. 4). Soils with
the highest SOC accumulation exhibited higher fungal relative
abundances and CUE (Fig. 4a,c,d). Accordingly, higher fungal
abundance was also positively correlated with CUE (Fig. 4b).
Though syringol-treated soils exhibited the highest CUE, fungal
relative abundances and SOC, the relationships among these
variables remain when we analysed only sugar-treated soils
(Supplementary Fig. 4). We also observed that soils with higher
SOC concentrations had greater lipid relative abundances and
lower protein abundances (Fig. 4a).
Discussion
Model soil systems provide a platform for directly manipulating
specific abiotic and biotic controls on soil processes. They allow
for insight into fundamental questions about soil aggregate
development
38
, organic matter turnover
39
and mineralogical
influences on microbial communities and decomposition
40–42
.
Here we use model soils for real-time monitoring of microbial-
SOM formation and demonstrate that microbial processing of
simple C substrates produced an abundance of stable, chemically
diverse SOM dominated by microbial proteins and lipids,
comparable to natural soils. However, clay mineralogy, a well-
known control over SOM dynamics
13,28
, had little effect on SOM
abundance and stability. Instead the microbial community and its
associated physiology was a stronger driver of SOM development.
This suggests that the community’s physiology and their residue
inputs to SOM may be as or more important than the more
widely recognized effects of soil mineral structure on soil C
stabilization.
Considerable SOM accumulated (8–13 mg C per g soil) within
18 months in our model soils (Table 1). The high microbial
activity (Supplementary Fig. 1), cumulative respiration
(Supplementary Table 4) and especially SOM chemistry (Fig. 1)
in our model soils indicate that substrates were first processed by
microbes before becoming SOM. Further, consistent with studies
Table 2 | Percentage of stable soil carbon.
% Stable C
Chemically stable C Biologically stable C
Kaolinite
Glucose 37.39
Aa
76.61
Aa
Cellobiose 37.37
Aa
63.34
Aa
DOC 37.14
Aa
74.13
Aa
Montmorillonite
Glucose 44.12
ABa
82.76
Aa
Cellobiose 36.33
Aa
87.65
Bb
Syringol 48.03
Ba
93.42
C
DOC 38.65
Aa
93.00
Cb
Field soil 32–66* 77.60
±
3
w
Chemically stable SOC is the per cent of non-hydrolysable C, and biologically stable SOC is the
per cent of previously accumulated SOC not mineralized during a 3-month incubation at 6
months. For comparison, natural field soil results are also presented. Significance among
substrates within clay type is indicated by capital letters. Significant pair-wise comparisons
among clay types within a substrate group are indicated by lowercase letters (ANOVA:
Po0.05) (experimental replication n ¼ 5). DOC, dissolved organic C.
*Range is the acid unhydrolysable fraction using 6 M HCl from 22 soils from cultivated and
grassland soils at depths from 0 to 20, 25 to 50 and 50 to 100 cm (ref. 25).
wPer cent of non-mineralizeable SOC from a 588 day laboratory incubation on eight cultivated
and native grassland soils
24
.
Table 1 | Soil carbon accumulation.
Months Substrate added SOC (mg C per g soil)
Kaolinite Montmorillonite Field soil
Glucose Cellobiose DOC Glucose Cellobiose Syringol DOC
6 16.80 5.10
Ab
5.33
Ab
6.21
Bb
4.01
Aa
3.91
Aa
7.96
C
5.55
Ba
9.18
±
0.04*
9 27.30 7.21
Ab
6.55
Aa
8.54
Ba
6.2
Aa
6.72
Aa
14.36
C
8.83
Ba
12 37.10 8.05
Aa
9.21
Ba
11.10
Ca
7.76
Aa
8.86
Aa
12.98
B
10.73
Ba
15 46.90 11.08
Aa
11.33
Aa
12.88
Bb
10.43
Aa
10.17
Aa
14.47
C
11.98
Ba
18 46.90 8.61
Aa
8.88
Aa
10.45
Ba
8.36
Aa
7.98
Aa
13.11
B
11.75
Bb
Amount lost 38.29 38.02 36.45 38.54 38.92 33.79 35.20
SOC conversion efficiency (mg SOC per g total substrate-C added)
6 16.80 0.30
Ab
0.32
Ab
0.37
Bb
0.24
Aa
0.23
Aa
0.47
C
0.33
Ba
0.03–0.33
w
9 27.30 0.26
Ab
0.24
Ba
0.31
Ca
0.23
Aa
0.29
Aa
0.49
C
0.32
Ba
12 37.10 0.22
Aa
0.25
Ba
0.30
Ca
0.21
Aa
0.24
ABa
0.35
B
0.29
Ba
15 46.90 0.24
Aa
0.24
Aa
0.27
Bb
0.24
Aa
0.23
Aa
0.47
C
0.26
Ba
18 46.90 0.18
Aa
0.19
Aa
0.22
Ba
0.18
Aa
0.10
Aa
0.28
B
0.25
Bb
Fraction lost 0.82 0.81 0.78 0.82 0.90 0.72 0.75
The soil organic carbon (SOC) concentration and conversion efficiency (the proportion of added substrate-C converted to SOC) at 6, 9, 12, 15 and 18 months in soils treated with glucose, cellobiose,
syringol or plant leachate DOC. For comparison, field soil results are also presented. Significance among substrates within clay type is indicated by capital letters. Significant pair-wise comparisons among
clay types within a substrate group is indicated by lowercase letters (ANOVA: Po0.05) (experimental replication n ¼ 5). DOC, dissolved organic C.
*Soybean bulk soils 0–7 cm depth, collected from the W.K. Kellogg Biological Station, Michigan
16
.
wData synthesized from 15 field and laboratory incubation experiments
7
.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13630 ARTICLE
NATURE COMMUNICATIONS | 7:13630 | DOI: 10.1038/ncomms13630 | www.nature.com/naturecommunications 5