Page 1
B I O G E O C H E M I S T R Y L E T T E R S
Beyond clay: towards an improved set of variables
for predicting soil organic matter content
Craig Rasmussen
.
Katherine Heckman
.
William R. Wieder
.
Marco Keiluweit
.
Corey R. Lawrence
.
Asmeret Asefaw Berhe
.
Joseph C. Blankinship
.
Susan E. Crow
.
Jennifer L. Druhan
.
Caitlin E. Hicks Pries
.
Erika Marin-Spiotta
.
Alain F. Plante
.
Christina Scha¨del
.
Joshua P. Schimel
.
Carlos A. Sierra
.
Aaron Thompson
.
Rota Wagai
Received: 19 December 2017 / Accepted: 27 January 2018 / Published online: 3 February 2018
Ó
Springer International Publishing AG, part of Springer Nature 2018
Abstract
Improved quantification of the factors
controlling soil organic matter (SOM) stabilization
at continental to global scales is needed to inform
projections of the largest actively cycling terrestrial
carbon pool on Earth, and its response to environmen-
tal change. Biogeochemical models rely almost
exclusively on clay content to modify rates of SOM
turnover and fluxes of climate-active CO
2
to the
atmosphere. Emerging conceptual understanding,
however, suggests other soil physicochemical proper-
ties may predict SOM stabilization better than clay
content. We addressed this discrepancy by synthesiz-
ing data from over 5,500 soil profiles spanning
continental scale environmental gradients. Here, we
demonstrate that other physicochemical parameters
are much stronger predictors of SOM content, with
clay content having relatively little explanatory
power. We show that exchangeable calcium strongly
predicted SOM content in water-limited, alkaline
soils, whereas with increasing moisture availability
and acidity, iron- and aluminum-oxyhydroxides
emerged as better predictors, demonstrating that the
relative importance of SOM stabilization mechanisms
scales with climate and acidity. These results highlight
the urgent need to modify biogeochemical models to
Responsible Editor: Stuart Grandy.
Electronic supplementary material
The online version of
this article (
https://doi.org/10.1007/s10533-018-0424-3
) con-
tains supplementary material, which is available to authorized
users.
C. Rasmussen (
&
)
J. C. Blankinship
Department of Soil, Water and Environmental Science,
The University of Arizona, Tucson, AZ, USA
e-mail: crasmuss@email.arizona.edu
K. Heckman
US Forest Service Northern Research Station, Houghton,
MI, USA
W. R. Wieder
Climate and Global Dynamics Laboratory National Center
for Atmospheric Research & Institute of Arctic and
Alpine Research, University of Colorado, Boulder,
Boulder, CO, USA
M. Keiluweit
University of Massachusetts-Amherst, Amherst, MA,
USA
C. R. Lawrence
U.S. Geological Survey, Boulder, CO, USA
A. A. Berhe
University of California, Merced, Merced, CA, USA
S. E. Crow
University of Hawaii Manoa, Honolulu, HI, USA
J. L. Druhan
University of Illinois Urbana-Champaign, Champaign, IL,
USA
C. E. Hicks Pries
Dartmouth College, Hanover, NH, USA
123
Biogeochemistry (2018) 137:297–306
https://doi.org/10.1007/s10533-018-0424-3
Page 2
better reflect the role of soil physicochemical proper-
ties in SOM cycling.
Keywords
Soil organic matter
Biogeochemistry
Carbon cycle
Introduction
A clear description of soil organic matter (SOM)
stabilization mechanisms is needed to inform society’s
ability to manage carbon and meet desirable climate
targets (Minasny et al.
2017
). An established and
tested hypothesis is that clay content, i.e., the mineral
fraction
\
2
l
m in size, principally determines SOM
storage and stabilization by promoting the sorption of
organic matter to mineral surfaces and aggregate
formation (Oades
1988
; Schimel et al.
1994
; Wagner
et al.
2007
). Clay content may also alter SOM stability
indirectly, by influencing microbial community com-
position, soil hydrology, drainage, and O
2
availability
(Andrews et al.
2011
; Fierer and Schimel
2002
). The
small size of clay particles imbues them with high
specific surface area, and clay particles typically
exhibit a combination of permanent structural charge
and variable pH-dependent surface charge. The com-
bination of charge and high surface area allows clay
sized particles to dominate mineral–mineral, mineral-
organic, mineral-metal, and mineral–water interac-
tions in soils (Sposito et al.
1999
). Thus, total clay
content has been used as a proxy for the direct effects
of aggregation and sorption, and indirect controls of
soil water content on SOM stability (Kahle et al.
2003
;
Mikutta et al.
2006
; Oades
1988
). These insights are
broadly applied in biogeochemical models (e.g.,
Coleman and Jenkinson
1996
; Parton et al.
1987
;
Wieder et al.
2015
) in which clay or clay
?
silt
content describes variation in SOM storage by mod-
ifying carbon fluxes into particular pools, and their
rate of turnover. All of these models predict slower
SOM turnover and higher SOM storage in more finely
textured soils (Fig. S1). Although numerically tract-
able, these relatively simplistic considerations of SOM
stabilization may not apply across diverse soil systems
or capture the efficacy of different clay-sized particles
for stabilizing SOM.
Current understanding suggests other physico-
chemical parameters may be better predictors of
SOM content and stabilization. Observational and
manipulative studies indicate that SOM content may
be more related to the amount of extractable metals,
here defined to include measures of exchangeable
cations, pedogenic oxyhydroxides, short–range-order
(SRO) phases, and organically complexed metals
(Kaiser and Guggenberger
2000
; Lawrence et al.
2014
; Percival et al.
2000
; Rasmussen et al.
2006
;
Torn et al.
1997
). These studies highlight the fact that
not all clay-sized particles are equivalent in their SOM
stabilization capacity. The clay fraction includes some
combination of phyllosilicates, crystalline and SRO
oxyhydroxides and aluminosilicates, and organo-
metal complexes, each of which vary in their surface
area, reactivity, solubility, and sorptive capacity
(Kaiser et al.
1997
; Sposito et al.
1999
). Furthermore,
silicate clays are often coated with Al- and Fe-
oxyhydroxides that contribute to their SOM stabiliza-
tion capacity. Clay content as a sole measure can thus
mask the diversity of mineral properties and may not
effectively capture specific SOM stabilization mech-
anisms. It follows that other soil properties may serve
as better surrogates of SOM stabilization than clay
content alone.
E. Marin-Spiotta
Department of Geography, University of Wisconsin –
Madison, Madison, WI, USA
A. F. Plante
University of Pennsylvania, Philadelphia, PA, USA
C. Scha¨del
Center for Ecosystem Science and Society, Northern
Arizona University, Flagstaff, AZ, USA
J. P. Schimel
University of California – Santa Barbara, Santa Barbara,
CA, USA
C. A. Sierra
Max Planck Institute for Biogeochemistry, Jena, Germany
A. Thompson
Department of Crop and Soil Science and the Odum
School of Ecology, University of Georgia, Athens, GA,
USA
R. Wagai
National Agriculture and Food Research Organization,
Institute for Agro-Environmental Sciences, Tsukuba,
Japan
298
Biogeochemistry (2018) 137:297–306
123
Page 3
Recently, several attempts have been made to
quantitatively link soil physicochemical variables
with SOM abundance and stability across larger,
regional spatial scales (Augustin and Cihacek
2016
;
Doetterl et al.
2015
; Garrido and Matus
2012
; Mathieu
et al.
2015
). In many of these studies, clay content was
not a significant predictor of SOM content or stability,
and climatic variables were identified as secondary in
importance compared to soil mineralogical and mor-
phological characteristics. Here, we address whether
other soil physicochemical parameters are better
predictors of SOM using a comprehensive set of soil
profile data from the US National Cooperative Soil
Survey Database that spans continental scale gradients
in climates, ecoregions, and soil taxa.
Materials and methods
Global soil, climate, and ecoregion datasets
Soil data were compiled from the U.S. Department of
Agriculture’s National Cooperative
Soil
Survey
(NCSS)
Soil
Characterization
Database
(
http://
ncsslabdatamart.sc.egov.usda.gov/
). The data consist
of 392,710 individual soil horizons sampled from
62,843 pedons described as part of NCSS activities.
Data were assessed for quality assurance and control
purposes by analyzing the distributions of each data
type and errors corrected where possible or excluded
when no obvious correction could be applied. Histosols
and all organic horizons were excluded; only 240
horizons occurred in Gelisols and were not included in
soil order specific analyses. The soil properties ana-
lyzed included SOC (%) used as a measure of SOM,
horizon midpoint depth (depth; cm), clay (%), clay
?
silt (%), pH in H
2
O, and extractable metals that
included exchangeable Ca as determined by ammo-
nium acetate extraction at pH 7.0 (C
aex
; cmol
?
kg
-
1
),
dithionite
extractable
iron
(Fe
d
;
%),
oxalate
extractable iron (Fe
o
; %) and aluminum (Al
o
; %). The
selected physicochemical variables were chosen to
encompass a range of soil properties related to various
proposed SOM stabilization mechanisms. This selec-
tion of variables also maximized the number of soil
horizons that could be included in the analysis while
maintaining a broad spectrum of soil properties. Soil
organic carbon data were compiled following Wills
et al. (
2007
), correcting total reported C for carbonate
C, and converting Walkley–Black organic carbon
values to the equivalent of combustion C.
The soil data were paired with properties extracted
from global maps of ecoregion, climate, and soil order.
Ecoregion data were compiled from the World Wild-
life Foundation terrestrial ecoregions map (Olson et al.
2001
). Climate data, including mean annual temper-
ature (MAT;
°
C), mean annual precipitation (MAP;
mm year
-
1
), and potential evapotranspiration (PET;
mm year
-
1
) were extracted from the 0.5 degree CRU
climate dataset of New et al. (New et al.
1999
). The
climate data were used to calculate an aridity index for
each location, defined as the ratio of annual PET/
MAP, where ratios
[
1 indicate water-limited systems
and values
\
1 indicate energy-limited systems. The
aridity index data were further divided into seven
humidity classes using hierarchical clustering, with
the number of classes chosen to maximize the number
of locations in each class and to correspond with
climate zones. Soil order information was extracted
from the NCSS database when available. For locations
lacking taxonomic information, soil order was
extracted from the NRCS Global Soil Region map
available at:
https://www.nrcs.usda.gov/wps/portal/
nrcs/detail/soils/use/?cid=nrcs142p2_054013
. Spatial
data were handled using ArcMAP 10.3.1 (ESRI,
Redlands, CA) and all statistical analyses performed
using JMP Pro v12.1.0 (SAS Institute, Cary, NC).
Statistical analyses and model development
The data were filtered such that only horizons with a
complete set of soil and climate variables were
included, providing a total of 5587 pedons with
28,819 horizons. To standardize variation among
variables and meet normality assumptions of the
applied statistical tests, all continuous predictors and
the soil organic carbon (SOC) data were transformed to
normal distributions using Box-Cox transformations,
followed by standardization to a mean of 0 and
standard deviation of 1. The transformed and stan-
dardized data allowed for direct comparison of regres-
sion coefficient estimates among predictor variables.
Regression analyses
Linear mixed model (LMM) regression was imple-
mented to determine the soil and climate parameters
that best explained the variance in SOC content. LMM
Biogeochemistry (2018) 137:297–306
299
123
Page 4
was selected to address the non-independent nature of
multiple horizons within one pedon. The models were
structured with pedon ID as a random effect, allowing
the slope and intercept of the relationship between soil
depth and SOC to vary by each pedon based on the a
priori knowledge of a relationship between soil depth
and SOC content (Jobbagy and Jackson
2001
). Three
LMM models were run: (1) using just soil depth as a
fixed effect, (2) using soil depth and the entire set of
the selected continuous physicochemical and climate
variables, including two-way interaction terms among
all variables, as fixed effects, and (3) a reduced final
model that included only those fixed effects that
contributed the most to the prediction of SOC based on
regression coefficient estimates and F-values. Specif-
ically, the final model only kept terms with regression
coefficient estimates
[
0.1 or
\
-
0.1 that corre-
sponded with F-values
[
400. The fixed-effect vari-
ables in the final model included depth, Al
o
, Ca
ex
, pH,
PET/MAP, and the pH by Al
o
interaction term
(Table
1
). The relative importance and contribution
of fixed effects was determined based on regression
coefficient estimates and F values.
Additionally, LMM modeling of SOC was per-
formed across pH classes. Soil-pH data were grouped
using hierarchical clustering, with the number of
classes chosen to maximize the number of samples in
each class. We then ran LMM for each pH class, using
pedon ID as the random variable and allowing both the
slope and intercept to vary with depth, and included
depth, Al
o
, and Ca
ex
as fixed parameters. The F-values
for each fixed parameter were reported in Fig.
2
b to
demonstrate the relative importance of each parameter
to account for SOC variance across pH classes. The
relative contribution of each fixed parameter by pH
class, in combination with soil physicochemical
property change by pH class (Fig.
2
a), was used to
develop the conceptual figure of change in dominant
SOM stabilization mechanism with pH (Fig.
3
).
Correlations by environmental categorical
variables
The conditional residuals of the soil depth-only LMM
were related to all soil physicochemical and climate-
continuous variables, parsed among three categorical
variables, using Spearman’s
q
(Fig.
1
). The categories
were humidity class ecoregion, and soil order. This
approach allowed us to include collinear variables left
out of the LMM. Significance was evaluated based on
a Bonferroni’s corrected
a
value of 0.0002 (
a
=
0.05/
250 individual correlations). Only categories with
n
[
400 soil samples were used for correlation
determination.
Results and discussion
Soil organic carbon and physicochemical
parameters
Linear mixed model (LMM) regression indicated that
soil depth was the strongest predictor of SOC, with a
significant trend of decreasing SOC content with
increasing depth. After accounting for depth, Al
o
and
Ca
ex
were the best predictors, followed by pH, the
interaction of pH and Al
o
, and PET/MAP (potential
evapotranspiration over mean annual precipitation).
Al
o
and Ca
ex
showed a positive relationship with SOC,
whereas pH, the pH-Al interaction term, and PET/
MAP exhibited weaker, negative relationships with
SOC (Table
1
).
Table 1
Fixed-effect parameters from the linear mixed model
Term
Degrees of
freedom
Regression
coefficient
95% lower
95% upper
F ratio
Prob
[
F
Depth
5782
-
0.579
-
0.589
-
0.569
12,616
\
0.0001
Al
o
26,054
0.306
0.298
0.314
5135
\
0.0001
Ca
ex
25,934
0.342
0.330
0.354
3202
\
0.0001
pH
23,444
-
0.258
-
0.270
-
0.245
1725
\
0.0001
pH
9
Al
o
27,852
-
0.141
-
0.148
-
0.134
1492
\
0.0001
PET/MAP
6468
-
0.108
-
0.121
-
0.096
279
\
0.0001
Intercept
5198
-
0.022
-
0.036
-
0.009
–
–
300
Biogeochemistry (2018) 137:297–306
123
Page 5
Correlation analysis of the conditional residuals
from a LMM using only depth as the fixed effect was
performed to further explore SOC relationships among
the categorical variables of humidity class, ecoregion,
and soil order (Fig.
1
). There were few correlations of
SOC with climate variables within each category;
rather, a combination of either Fe
o
and Al
o
, or Ca
ex
and
clay or clay
?
silt best explained SOC. Exchangeable
Ca and soil texture exhibited the strongest correlation
in water-limited environments, whereas Al- and Fe-
oxyhydroxides exhibited the strongest correlations in
humid environments.
Among soil taxa, Entisols and Inceptisols, with
minimal pedogenic development, presented strong
positive correlations with nearly all soil properties
(Fig.
1
), likely owing to their large range of soil
properties, and the diversity of environmental condi-
tions in which these soils are found (Buol et al.
2011
).
At the other soil development and weathering extreme,
no significant correlation to any soil property was
observed for Oxisols. This may be due in part to the
rapid litter and SOC decomposition associated with
the warm, wet tropical climates typical of Oxisols
(Tiessen et al.
1992
). Oxisols are highly weathered and
dominated by quartz and crystalline, variable charge
Fe- and Al-oxides and kaolinite, with few to no
weatherable minerals (Buol and Eswaran
2000
). In
these
soils,
Fe-oxide
cementations
form
very
stable concretions (El Swaify
1980
) that serve as the
dominant mechanism for protecting SOM from
decomposition (Tiessen et al.
1994
). However, these
concretions also confound soil textural analysis by
resisting dispersion, biasing results towards a coarse
classification while physically behaving more simi-
larly to fine-textured soils. Taken together, these
observations suggest a general decrease in the range of
soil properties that can facilitate SOM stabilization
with increased soil weathering (Lawrence et al.
2015
;
Torn et al.
1997
).
Prevalence of extractable Ca and texture in dry
climates
The regression and correlation analyses reflected
strong positive correlation of SOC to Ca
ex
, clay
?
silt,
and clay in water-limited systems, that included
deserts, grasslands, savannas, and shrublands, largely
characterized by Aridisols, Mollisols and Alfisols
(Fig.
1
). These ecoregions with less precipitation
maintain circum-neutral to alkaline pH values, and
retain non-hydrolyzing base cations. These conditions
favor the formation of 2:1 phyllosilicates with an
abundance of negatively charged exchange sites
(Douglas
1989
; Kittrick
1971
). When divalent ions,
such as Ca
2
?
, occupy these exchange sites, they can
bind negatively charged organic moieties, such as the
carboxylic acid functional group common on SOM,
effectively forming a bridge with the underlying clay
Fig. 1
Heat maps indicating the Spearman correlation between
the conditional residuals of a mixed regression model using
pedon ID as a random factor and horizon depth as a fixed effect,
and soil physicochemical properties and climate parameters by
the categorical variables of
a
humidity class,
b
ecoregion, and
c
soil order. Spearman correlations are scaled from dark blue for
negative correlations to dark red for positive correlations. Those
correlations with P
[
0.0002 are considered not significant
(NS), based on Bonferroni’s corrected
a
value
Biogeochemistry (2018) 137:297–306
301
123
Page 6
(Mikutta et al.
2007
). Such cation bridging is widely
recognized as an important SOM stabilization mech-
anism (von Lu¨tzow et al.
2006
). In addition, exchange-
able Ca can cross-link two negatively charged organic
functional groups together, causing SOM aggregation
and immobilization (Kunhi Mouvenchery et al.
2012
).
Calcium has a stronger influence on SOM stabilization
in dry climates than other divalent cations because of
its higher relative abundance (Smith et al.
2014
) and
greater ionic radius, which facilitates stronger binding
energies (Muneer and Oades
1989
). The strong
correlation of SOC to exchangeable Ca and texture
in water-limited systems is consistent with previous
work highlighting a central role for texture in mod-
ulating SOM cycling (Amato and Ladd
1992
; Burke
et al.
1989
; Hassink
1997
; Nichols
1984
), work that
served as the foundation for current biogeochemical
models.
Prevalence of extractable Al and Fe in humid
climates
Soil carbon content in humid, forested ecoregions
(generally represented by Inceptisols, Ultisols, Andis-
ols, and Spodosols) exhibited strong positive correla-
tions with Fe
o
and Al
o
, a negative correlation with pH,
and little to no correlation with clay or clay
?
silt
(Fig.
1
). In these systems, the combination of water
availability, an abundance of low molecular weight
organic acids, or distinct primary mineral assem-
blages, facilitates high rates of mineral weathering that
may release Fe, Al, and Si faster than crystalline
minerals can precipitate (Shoji et al.
1993
). This can
lead to soil solutions that are oversaturated with
respect to SRO phases, that once formed, have high
surface areas and high densities of reactive hydroxyl
sites (Harsh et al.
2002
), allowing them to immobilize
large quantities of SOM. These SRO phases have also
been implicated as important drivers of aggregation,
which further protects SOM against decomposition
(Asano and Wagai
2014
; Rasmussen et al.
2005
). The
generally high SOM content in humid systems also
promotes Al and Fe-organic complexation and co-
precipitation of organo-metal complexes, which can
compete and occur in conjunction with SRO formation
(Dahlgren et al.
2004
). Both phases are important to
SOM stabilization (Wagai and Mayer
2007
), but the
analyses herein were based on SOC correlations with
oxalate-extractable Al or Fe, which includes both
organo-metal complexes and SRO phases, so we are
not able to directly separate these factors. We did not
include more direct measures of organo-metal com-
plexation because of the limited extent where it was
measured. However, as indicated below (Fig.
2
), for
those locations with such measures, we observed
strong gradients in the relative concentration of
organo-metal complexes and SRO phases with soil
pH.
Soil pH and stabilization mechanisms
Our results suggest a complicated and important
influence of soil pH on SOM content. The LMM
results indicated a significant role for pH, and the
interaction of pH and Al
o
in accounting for SOC
content, but pH was not significantly correlated with
SOC across most categories. Soil pH reflects the
overall chemical state of the soil system and dictates a
number of geochemical gradients, including the
speciation of dissolved metals, the reactivity and
charge of minerals and organic molecules, and the
predominant type of organo-mineral bond (Deng and
Dixon
2002
). Each of these influences SOM stabiliza-
tion, resulting in an overall influence of pH that is
nonlinear and driven by thresholds in aqueous speci-
ation and mineral stability, which are in turn modu-
lated by water availability and other soil-forming
factors.
Soil pH corresponded with distinct gradients in soil
properties (Fig.
2
a). In particular, Ca
ex
increased
significantly above pH 6.5, whereas Al
o
was greatest
at pH values
\
6.5. Although there were not enough
data to include them in the full analysis, two additional
values, the coefficient of linear extensibility (COLE),
and pyrophosphate extractable Al (Al
p
) were summa-
rized by pH class to further elucidate soil property
changes with pH. COLE provides a measure of soil
shrink-swell capacity that corresponds with clay
content and presence of expansible 2:1 phyllosilicates;
Al
p
is a common metric for organically complexed Al.
COLE increased above pH 6.5, whereas the ratio Al
p
/
Al
o
, which provides an estimate of the proportion of
organo-metal versus organo-mineral Al species,
increased at pH values less than 5.5, with Al
p
dominant at low pH.
Regression analyses across pH classes (Fig.
2
b),
using Al
o
to represent SRO phases and organo-metal
complexes and Ca
ex
to represent 2:1 clays and
302
Biogeochemistry (2018) 137:297–306
123
Page 7
exchangeable cations, indicated that, after accounting
for soil depth, Ca
ex
was the most important factor at
pH
[
6.5, and Al
o
was the dominant variable at
pH
\
6.5. The Ca-bridging mechanism is particularly
important in soils enriched in 2:1 phyllosilicates
(Laird
2001
), corresponding with the increase in
COLE at pH
[
6.5 (Fig.
2
a). The dominance of Al
o
and the increase in the Al
p
/Al
o
noted at pH
\
5.5
(Fig.
2
a) suggest that SRO is the dominant factor at pH
5.5–6.5, whereas at low pH, organo-metal complex-
ation is the dominant factor controlling SOC content
(Wagai and Mayer
2007
).
Here, we propose a conceptual model whereby the
dominant SOM stabilization mechanism varies as a
function of pH (Fig.
3
). The SOM stabilization
mechanisms vary with increasing soil pH from
predominantly organo-metal complexation, to associ-
ation with SRO phases, to Ca complexation and cation
bridging with phyllosilicates. Soil pH varies nonlin-
early with soil depth and soil-forming factors
(Fig. S5), and soil physicochemical properties vary
nonlinearly with pH (Fig.
2
a), leading to a complex
and indirect relationship between pH and SOM
content.
Implications for model improvement
Our analyses demonstrated that variables other than
clay content serve as better predictors of SOM content
in many systems. This observation implies that models
for continental to global scale SOM content may be
improved by inclusion of alternate mineralogical
proxies. Variables of particular importance include
exchangeable Ca, SRO Al- and Fe-oxyhydroxides,
and Al-, Fe- organo-metal complexes, and the relative
contribution of each parameter can be scaled with pH
(Fig.
3
). A step to improve models would be to
develop explicit functions that relate these variables to
specific rate and transfer coefficients.
Soil pH
4
5
6
7
8
9
Fi
xed Param
e
ter F
-Val
u
e
0
1000
2000
3000
4000
Al
o
Ca
ex
Depth
Soil pH
4
5
6
7
8
9
Al
p
/Al
o
[%]
0
20
40
60
80
100
Al
o
[%]
0.08
0.12
0.16
0.20
Ca
ex
[cmol
+
kg
-1
]
0
10
20
30
40
COLE
0.015
0.020
0.025
0.030
0.035
0.040
Alp/Alo
Alo
Caex
COLE
(A)
(B)
Fig. 2 a
The change in soil
physicochemical properties
important for predicting
SOC content by soil-pH
class, including the soil
content of oxalate
extractable Al (Al
o
), the
fraction of Al
o
that is
partitioned to Al- organo-
metal complexes as
measured by Na-
pyrophosphate
extractable Al (Al
p
), the
amount of Ca on the cation
exchange complex (Ca
ex
),
and the coefficient of linear
extensibility (COLE)
(parameter values are
medians within each pH
class).
b
Fixed parameter
F-values for Al
o
, Ca
ex
and
horizon depth from linear
mixed effect models using
pedon ID as the random
effect performed for each
pH class
Biogeochemistry (2018) 137:297–306
303
123
Page 8
For broad applicability in Earth system modeling, a
new set of global-scale data products are needed that
include the variables identified in our analysis. Cur-
rently, only data products on clay content and pH are
available at the global level (Hengl et al.
2017
), but as
this analysis shows, extractable metal data do exist for
a wide range of locations and soil taxa, and some
globally gridded information about soil mineral com-
position has been derived (Ito and Wagai
2017
;
Journet et al.
2014
). Given the large number of soil
profiles that contain these data, compiling this infor-
mation into large-scale data products should be a
priority. With the current state of soil data products,
global grids of soil pH may provide the most
immediate and tractable variable for modifying bio-
geochemical models and their treatment of different
SOM stabilization mechanisms (Fig.
3
). Implement-
ing such changes would strengthen the relationships
among our conceptual understanding of SOM stabi-
lization, model design and structure, and common
measurements of soil physicochemical properties.
Acknowledgements
This work was conducted as a part of the
‘‘What Lies Below? Improving quantification and prediction of
soil carbon storage, stability, and susceptibility to disturbance’’
Working Group supported by the John Wesley Powell Center for
Analysis and Synthesis, funded by the U.S. Geological Survey.
Additional support was provided by NSF EAR-1331408 and
EAR- 1123454 to Rasmussen, NSF CAREER BCS-1349952 to
Marin-Spiotta,
US
Department
of
Agriculture
NIFA
2015-67003-23485 and US Department of Energy TES DE-
SC0014374 to Wieder, and USDA-NIFA Hatch project
HAW01130-H to Crow. Any use of trade, firm, or product
names is for descriptive purposes only and does not imply
endorsement by the U.S. Government.
Author Contributions
This work is the result of two
workshops sponsored by the USGS John Wesley Powell
Center for Analysis and Synthesis in May of 2016 and May of
2017. The motivation and ideas for this work were generated
collaboratively among all authors during these workshops.
Rasmussen led manuscript development, data compilation, and
analysis. All authors contributed to writing/editing, statistical
analyses, and figure development.
References
Amato M, Ladd JN (1992) Decomposition of C-14-labeled
glucose and legume material in soils—properties influ-
encing the accumulation of organic residue-c and microbial
biomass-c. Soil Biol Biochem 24(5):455–464
Andrews DM, Lin H, Zhu Q, Jin LX, Brantley SL (2011) Hot
spots and hot moments of dissolved organic carbon export
and soil organic carbon storage in the shale hills catchment.
Vadose Zone Journal 10(3):943–954
Asano M, Wagai R (2014) Evidence of aggregate hierarchy at
micro- to submicron scales in an allophanic Andisol.
Geoderma 216:62–74
Augustin C, Cihacek LJ (2016) Relationships between soil
carbon and soil texture in the northern great plains. Soil Sci
181(8):386–392
Buol SW, Eswaran H (2000) Oxisols. In: Sparks DL (ed)
Advances in agronomy, vol 68. Academic Press, San
Diego, pp 151–195
Buol SW, Southard RJ, Graham RC, McDaniel PA (2011) Soil
genesis and classification. Wiley, Hoboken
Burke IC, Yonker CM, Parton WJ, Cole CV, Flach K, Schimel
DS (1989) Texture, climate, and cultivation effects on soil
organic-matter content in Us grassland soils. Soil Sci Soc
Am J 53(3):800–805
Coleman D, Jenkinson DS (1996) RothC-26.3 - A model for the
turnover of carbon in soil. In: Powlson DS, Smith P, Smith
JU (eds) Evaluation of soil organic matter models : using
existing long-term datasets. Springer, Berlin; New York,
pp 237–246
Dahlgren RA, Saigusa M, Ugolini FC (2004) The nature,
properties and management of volcanic soils. Adv Agron
82(82):113–182
Deng Y, Dixon JB (2002) Soil organic matter and organo-
mineral interactions. In: Dixon JB, Schulze DG (eds) Soil
Mineralogy with Environmental Applications. Soil Sci-
ence Society of America, Madison, pp 69–108
Doetterl S, Stevens A, Six J, Merckx R, Van Oost K, Pinto MC,
Casanova-Katny A, Munoz C, Boudin M, Venegas EZ,
Boeckx P (2015) Soil carbon storage controlled by inter-
actions between geochemistry and climate. Nat Geosci
8(10): 780
Douglas L (1989) Vermiculites. In: Dixon JB, Weed SB, Din-
auer RC (eds) Minerals in soil environments. vol no 1. Soil
Science Society of America, Madison, Wis., USA. p 635-
674
El Swaify SA (1980) Physical and mechanical properties of
Oxisols. In: Theng BKG (ed) Soils with variable charge.
New Zealand Society of Soil Science, Lower Hutt,
pp 303–324
0
50
100
4.0
5.0
6.0
7.0
8.0
9.0
Al-, Fe-complexes
SRO
Ca+clay
Relave contribuon (%)
Soil pH
Fig. 3
Proposed conceptual model of how horizon-level soil
pH controls the relative importance of SOM stabilization
mechanisms grading from Al- and Fe- organo-metal complexes
in acidic systems, to relative dominance by short-range-order
(SRO) materials in moderately acid soils, to exchangeable
calcium and clay content in circum-neutral to basic soils. The
conceptual model is based on results of mixed regression
modeling of soil organic carbon content across pH classes
304
Biogeochemistry (2018) 137:297–306
123
Page 9
Fierer N, Schimel JP (2002) Effects of drying-rewetting fre-
quency on soil carbon and nitrogen transformations. Soil
Biol Biochem 34(6):777–787
Garrido E, Matus F (2012) Are organo-mineral complexes and
allophane content determinant factors for the carbon level
in Chilean volcanic soils? (vol 92C, pg 106, 2012).
CATENA 95:184–184
Harsh J, Chorover J, Nizeyimana E (2002) Allophane and
imogolite. In: Dixon JB, Schulze DG (eds) Soil Mineralogy
with Environmental Applications. Soil Science Society of
America, Madison, pp 291–322
Hassink J (1997) The capacity of soils to preserve organic C and
N by their association with clay and silt particles. Plant Soil
191(1):77–87
Hengl T, de Jesus JM, Heuvelink GBM, Gonzalez MR, Kili-
barda M, Blagotic A, Shangguan W, Wright MN, Geng
XY, Bauer-Marschallinger B, Guevara MA, Vargas R,
MacMillan RA, Batjes NH, Leenaars JGB, Ribeiro E,
Wheeler I, Mantel S, Kempen B (2017) SoilGrids250 m:
Global gridded soil information based on machine learn-
ing. Plos One 12(2)
Ito A, Wagai R (2017) Global distribution of clay-size minerals
on land surface for biogeochemical and climatological
studies. Sci Data 4:
Jobbagy EG, Jackson RB (2001) The distribution of soil nutri-
ents with depth: Global patterns and the imprint of plants.
Biogeochemistry 53(1):51–77
Journet E, Balkanski Y, Harrison SP (2014) A new data set of
soil mineralogy for dust-cycle modeling. Atmos Chem
Phys 14(8):3801–3816
Kahle M, Kleber M, Jahn R (2003) Retention of dissolved
organic matter by illitic soils and clay fractions: Influence
of mineral phase properties. Journal of Plant Nutrition and
Soil Science-Zeitschrift Fur Pflanzenernahrung Und Bod-
enkunde 166(6):737–741
Kaiser K, Guggenberger G (2000) The role of DOM sorption to
mineral surfaces in the preservation of organic matter in
soils. Org Geochem 31(7–8):711–725
Kaiser K, Guggenberger G, Haumaier L, Zech W (1997) Dis-
solved organic matter sorption on subsoils and minerals
studied by C-13-NMR and DRIFT spectroscopy. Eur J Soil
Sci 48(2):301–310
Kittrick JA (1971) Montmorillonite Equilibria and Weathering
Environment. Soil Sci Soc Am Pro 35(5):815–900
Kunhi Mouvenchery Y, Kucerik J, Diehl D, Schaumann GE
(2012) Cation-mediated cross-linking in natural organic
matter: a review. Rev Environ Sci Bio 11(1):41–54
Laird D (2001) Nature of clay-humic complexes in an agricul-
tural soil: II. Scanning electron microscopy analysis. Soil
Sci Soc Am J 65(5):1419–1425
Lawrence C, Harden J, Maher K (2014) Modeling the influence
of organic acids on soil weathering. Geochim Cosmochim
Acta 139:487–507
Lawrence CR, Harden JW, Xu XM, Schulz MS, Trumbore SE
(2015) Long-term controls on soil organic carbon with
depth and time: A case study from the Cowlitz River
Chronosequence, WA USA. Geoderma 247:73–87
Mathieu JA, Hatte C, Balesdent J, Parent E (2015) Deep soil
carbon dynamics are driven more by soil type than by
climate: a worldwide meta-analysis of radiocarbon pro-
files. Glob Change Biol 21(11):4278–4292
Mikutta R, Kleber M, Torn MS, Jahn R (2006) Stabilization of
soil organic matter: Association with minerals or chemical
recalcitrance? Biogeochemistry 77(1):25–56
Mikutta R, Mikutta C, Kalbitz K, Scheel T, Kaiser K, Jahn R
(2007) Biodegradation of forest floor organic matter bound
to minerals via different binding mechanisms. Geochim
Cosmochim Acta 71(10):2569–2590
Minasny B, Malone BP, McBratney AB, Angers DA, Arrouays
D, Chambers A, Chaplot V, Chen ZS, Cheng K, Das BS,
Field DJ, Gimona A, Hedley CB, Hong SY, Mandal B,
Marchant BP, Martin M, McConkey BG, Mulder VL,
O’Rourke S, Richer-de-Forges AC, Odeh I, Padarian J,
Paustian K, Pan GX, Poggio L, Savin I, Stolbovoy V,
Stockmann U, Sulaeman Y, Tsui CC, Vagen TG, van
Wesemael B, Winowiecki L (2017) Soil carbon 4 per mille.
Geoderma 292:59–86
Muneer M, Oades JM (1989) The Role of Ca-Organic Interac-
tions in Soil Aggregate Stability. 3. Mechanisms and
Models. Aust J Soil Res 27(2):411–423
New M, Hulme M, Jones PD (1999) Representing twentieth
century space-time climate variability. Part 1: development
of a 1961–90 mean monthly terrestrial climatology. J Clim
12:829–856
Nichols JD (1984) Relation of organic-carbon to soil properties
and climate in the southern great plains. Soil Sci Soc Am J
48(6):1382–1384
Oades JM (1988) The retention of organic-matter in soils.
Biogeochemistry 5(1):35–70
Olson DM, Dinerstein E, Wikramanayake ED, Burgess ND,
Powell GVN, Underwood EC, D’Amico JA, Itoua I, Strand
HE, Morrison JC, Loucks CJ, Allnutt TF, Ricketts TH,
Kura Y, Lamoreux JF, Wettengel WW, Hedao P, Kassem
KR (2001) Terrestrial ecoregions of the worlds: a new map
of life on Earth. Bioscience 51(11):933–938
Parton WJ, Schimel DS, Cole CV, Ojima DS (1987) Analysis of
factors controlling soil organic-matter levels in great-
plains grasslands. Soil Sci Soc Am J 51(5):1173–1179
Percival HJ, Parfitt RL, Scott NA (2000) Factors controlling soil
carbon levels in New Zealand grasslands: Is clay content
important? Soil Sci Soc Am J 64(5):1623–1630
Rasmussen C, Torn MS, Southard RJ (2005) Mineral assem-
blage and aggregates control carbon dynamics in a Cali-
fornia conifer forest. Soil Sci Soc Am J 69(6):1711–1721
Rasmussen C, Southard RJ, Horwath WR (2006) Mineral con-
trol of organic carbon mineralization in a range of tem-
perate
conifer
forest
soils.
Glob
Change
Biol
12(5):834–847
Schimel DS, Braswell BH, Holland EA, Mckeown R, Ojima DS,
Painter TH, Parton WJ, Townsend AR (1994) Climatic,
edaphic, and biotic controls over storage and turnover of
carbon in soils. Global Biogeochem Cycles 8(3):279–293
Shoji S, Nanzyo M, Dahlgren R (1993) Volcanic ash soils—
genesis, properties and utilization. Elsevier, Amsterdam
Smith D, Cannon WF, Woodruff LG, Solano F, Ellefsen KJ
(2014) Geochemical and mineralogical maps for soils of
the conterminous United States. In: Survey USG (ed). U.S.
Geological Survey. p 386
Sposito G, Skipper NT, Sutton R, Park SH, Soper AK, Great-
house JA (1999) Surface geochemistry of the clay miner-
als. Proc Natl Acad Sci USA 96(7):3358–3364
Biogeochemistry (2018) 137:297–306
305
123
Page 10
Tiessen H, Salcedo IH, Sampaio EVSB (1992) Nutrient and Soil
Organic-Matter Dynamics under Shifting Cultivation in
Semiarid Northeastern Brazil. Agric Ecosyst Environ
38(3):139–151
Tiessen H, Cuevas E, Chacon P (1994) The Role of Soil
Organic-Matter in Sustaining Soil Fertility. Nature
371(6500):783–785
Torn MS, Trumbore SE, Chadwick OA, Vitousek PM, Hen-
dricks DM (1997) Mineral control of soil organic carbon
storage and turnover. Nature 389(6647):170–173
von Lu¨tzow M, Ko¨gel-Knabner I, Ekschmitt K, Matzner E,
Guggenberger G, Marschner B, Flessa H (2006) Stabi-
lization of organic matter in temperate soils: mechanisms
and their relevance under different soil conditions—a
review. Eur J Soil Sci 57(4):426–445
Wagai R, Mayer LM (2007) Sorptive stabilization of organic
matter in soils by hydrous iron oxides. Geochim Cos-
mochim Acta 71(1):25–35
Wagner S, Cattle SR, Scholten T (2007) Soil-aggregate for-
mation as influenced by clay content and organic-matter
amendment. J Plant Nutr Soil Sci 170(1):173–180
Wieder WR, Grandy AS, Kallenbach CM, Taylor PG, Bonan
GB (2015) Representing life in the Earth system with soil
microbial functional traits in the MIMICS model. Geosci
Model Dev 8(6):1789–1808
Wills SA, Burras CL, Sandor JA (2007) Prediction of soil
organic carbon content using field and laboratory mea-
surements of soil color. Soil Sci Soc Am J 71(2):380–388
306
Biogeochemistry (2018) 137:297–306
123
Document Outline