Johannes et al. (2017). Optimal organic carbon values for soil structure quality of arable soils. Does clay content matter?

analysed the

role of the clay:SOC ratio in the relation between bulk density (BD) and
SOC using soil databases from Poland and France that included soils of
several taxonomic orders. They introduced the concept of

“

Complexed

Organic Carbon

”

(COC) as the fraction of the SOC bound to clay, and

found highest coe

ﬃ

cient of determination of the linear relation

between COC and soil volume (1/BD) for clay:COC = 10, thus inter-
preted as the saturation of the clay surface by SOC. They found the
same optimum ratio for clay dispersibility and considered that the
fraction of SOC corresponding to a tenth of the clay content was the
maximum COC controlling the structure-related physical properties of
soils. In particular, they concluded that structural porosity was no
longer increasing with SOC above 10% of clay content, and that the
optimum SOC content of a soil would thus be 10% of the clay content.
They showed that the clay:SOC ratio 10 was separating the permanent
pasture soils from the cropped soils in their database. Maximum
correlation between 1/BD and SOC, however, does not mean optimum
or desirable values of soil physical quality. In further tests of this
approach,

Schjonning et al. (2012)

and

Getahun et al. (2016)

found that

non-complexed clay (with a clay:SOC ratio of 10) was a better predictor
of dispersible clay than total clay content, while aggregate strength was
shown to be better explained by clay content alone.

These di

ﬀ

erent analyses raise a number of issues. The main concern

is that soil structural quality was not described in these various articles,
which means that the used databases might include data of soils with

di

ﬀ

erent levels of structural degradation at the time of sampling, as is

often the case in cropped soils. Other factors than SOC, e.g., mechanical
stresses, in

ﬂ

uence the soil physical properties, which should not

interfere with the assessment of the e

ﬀ

ect of SOC. Moreover, even

when a speci

ﬁ

c sub order is considered, soil management, climate, and

parent materials may have di

ﬀ

ered strongly in the di

ﬀ

erent articles.

Also, the techniques used to determine the physical properties were not
described in detail. In particular for bulk density, a standardized matric
potential should be applied prior to the volume measurement to
eliminate the e

ﬀ

ect of the swelling state (

Goutal et al., 2012

).

This rapid overview of the literature on the connections between

SOM or SOC and various physical characteristics of soils, taking into
account or not the e

ﬀ

ect of the clay:SOC ratio, suggests that there might

be an optimum or desirable range of SOM, but as pointed out by

Loveland and Webb (2003)

, the evidence is

“

equivocal at best

”

. SOM

may be essential for a number of agricultural and environmental
aspects, but in the

ﬁ

eld, it is managed by farmers. In this general

context, the main objective of this research was to determine the SOC
content that should be targeted in the management of an agricultural
soil from a single soil group developed on the same parent materials,
and to determine if clay content should be taken into account to
identify this optimal SOC content. We focused on soil pore volumes and
worked at large scale. We used the Visual Evaluation of Soil Structure
(VESS,

Ball et al., 2007

) adapted to soil samples (

Johannes et al., 2016

)

to make distinction between soils with degraded soil structure and soil
with

“

good structure

”

, thus possibly not impacted by mechanical

stresses. On these later, the relationships between SOC or SOC:clay
ratio and soil volume (1/BD), and the relevance of the concept of clay
saturation by SOC proposed by

ce fédéral de topographie, 1984

were analysed. The

relationships between soil structure quality and the SOC:clay ratio were
then discussed. Throughout this article, the description will be in terms
of SOC only (not of SOM), to simplify the narrative.

Fig. 1.

Map of the soil sampling locations in Western Switzerland.

A. Johannes et al.

Geoderma 302 (2017) 14–21

15

2. Materials and methods

2.1. Sampling and analysis

The investigated soils are classi

ﬁ

ed as

“

Braunerde

”

soils according

to the Swiss soil map (

O

ﬃ

). They

correspond to an intermediate Cambi-Luvisol group in the World
Reference Base for soil resources (

IUSS Working Group WRB, 2006

)

and to Inceptisols according to the USDA taxonomy (

). Only soils developed on mixed moraine

–

molasses bedrocks

were considered. A total of 161 locations were selected to represent
agricultural soils with a large range of structural quality, including
permanent pasture, conventional plough tillage, and no-till practices.
When looking for soils without structure degradation, we discarded
sites with signs of stresses like wheel tracks, surface sealing or erosion.

The sampled area covered approximately 6000 km

2

and represented

a major part of the farmed soils on the Swiss Plateau (

Fig. 1

). Detailed

soil properties are presented in

Table 1

.

At each location, two undisturbed soil samples were collected at

5

–

10 cm depth close to each other using a custom-made cylindrical

corer. The corer allowed easy removal without mechanical disturbance
of the sample from the cylinder so that sample quality could be checked
immediately after sampling. Removal from the corer was also required
to allow uncon

ﬁ

ned swelling of the soil during saturation in the

laboratory. This is important since the soils were collected at di

ﬀ

erent

water content, thus di

ﬀ

erent swelling state, which jeopardizes volume

comparisons (

Goutal et al., 2012

). Physico-chemical analyses were

performed on the

ﬁ

rst sample and the structural quality of the soil was

assessed on the replicate sample using CoreVESS scoring (

Johannes

et al., 2016

).

The samples were kept at 4 °C before analysis. They were

ﬁ

rst

equilibrated at

−

10 hPa matric potential in a sand box. After weighing

to determine the water content at

−

10 hPa, their volume was

measured using the plastic bag method (

Boivin et al., 1990

). After air

drying, weight and volume were measured again. Finally, the samples
were oven-dried at 105 °C until equilibrium, weighted and sieved to
2 mm. The weight and volume of the coarse (> 2 mm) fraction were
measured and removed from the sample volume and weight, to
calculate the physical properties of the < 2 mm fabric, namely gravi-
metric water content at

−

10 hPa (W

−

10 hPa

), speci

ﬁ

c volume (bulk

volume per g of dry mass) at

−

10 hPa (V

−

10 hPa

), speci

ﬁ

c (per g of dry

mass) air content at

−

10 hPa (Air

−

10 hPA

), and speci

ﬁ

c volume at air

dry state (V

dry

). Air

−

10 hPA

was calculated as V

−

10 hPa

- W

−

10 hPa

- 1/

ρ

,

with

ρ

the particle density. Air

−

10 hPa

and W

−

10 hPa

correspond to

the > 150

μ

m and < 150

μ

m pores in equivalent radius, respectively,

according to Jurin-Laplace's law. The < 2 mm fraction was analysed for

SOC with the method of

Walkley and Black (1934)

and for texture with

the traditional pipette method. The replicate samples were scored with
CoreVESS a method adapted from VESS (

Ball et al., 2007; Guimarães

et al., 2011

) to soil cores, and analysed for SOC and clay content as

well. For details, consult

Johannes et al. (2016)

. In brief, the scores (Sq)

range from 1 (

“

very good

”

) to 5 (

“

poor

”

). Scores of 1 and 2 represent

good structure, a Sq of 3 is a moderate structure quality, and Sq values
of 4 and 5 describe poor structures. When there was doubt between two
scores, half points were attributed. Before scoring, samples were
equilibrated to

−

100 hPa to ensure comparable scoring conditions;

samples were scored without knowing their origin for more objectivity.
The number of observations in each score is presented in

Table 2

. To

avoid interaction between physical stress and SOC e

ﬀ

ect, the sample

properties with replicate sample scores < 3 were used to analyse the
relationships between SOC and physical properties, and to test the
complexed organic carbon concept.

2.2. Relationships between SOC and physical properties

Three di

ﬀ

erent models (linear, semi-logarithmic, and broken-stick)

were

ﬁ

tted to the relationships between SOC or SOC:clay ratio and the

physical properties of the samples with CoreVESS < 3, with the R
software (version 3.1.0). The semi-logarithmic and broken-stick models
were selected to account for the decreasing e

ﬀ

ect of SOC on structural

properties at large SOC contents, with or without threshold, respec-
tively.

Broken-stick regression graphs and statistics (

Toms and Lesperance,

2003

) were

ﬁ

tted using the

“

segmented

”

package (

Muggeo, 2015

)

(version 0.5

–

1.4) of the R software. A simple piecewise-regression

model, which joins two straight lines at the breakpoint was used:

⎧

⎨

⎩

yi

x

e

x

α

e

for x

α

for x

α

=

β + β

+

β + β

+ β (

−

) +

≤

>

i

0

1

0

1

2

where

y

i

is the value for the

i

th observation,

x

i

is the corresponding

value for the independent variable,

α

is the breakpoint and

e

i

are

residual errors. The

ﬁ

rst slope is

β

1

and the second slope is

β

1

+

β

2

, so

β

2

can be interpreted as the di

ﬀ

erence between slopes. The statistical

signi

ﬁ

cance of the breakpoint was assessed using a Davies test (

Davies,

2002

), which tests for the di

ﬀ

erence in slope parameters in a piecewise

regression.

2.3. Test of the complexed organic carbon concept

We applied, on the soil samples with CoreVESS < 3, the same

procedure as

Dexter et al., 2008

, which we recall hereafter. It is based

on the assumption that the physical properties are determined by the
part of SOC complexed to the clay particles, and that above a maximum
SOC:clay ratio the clay is saturated. Hence, they calculated the
correlation between the bulk density and the Complexed Organic
Carbon (COC), a fraction of the SOC that does not exceed 1/n of the
clay content (Eq.

1

) to

ﬁ

nd n corresponding to the maximum correlation

coe

ﬃ

cient.

Table 1

Soil characteristics.

SOC

pH

CEC

Clay

Fine silt

Coarse silt

Silt

Fine sand

Coarse sand

Sand

(%)

(cmol

c

kg

−

1

)

< 2

μ

m

2

–

20

μ

m

20

–

50

μ

m

2

–

50

μ

m

50

–

200

μ

m

0.2

–

2 mm

0.05

–

2 mm

(%)

Mean

1.9

6.6

13.7

20.5

22.2

16.2

38.3

24.2

16.5

40.7

Median

1.8

6.7

13.8

20.1

21.8

15.0

37.4

24.3

16.2

41.7

SD

0.6

0.7

4.9

5.1

4.9

4.6

8.4

5.3

8.5

11.9

Min.

0.8

5.0

5.3

9.9

10.6

7.4

20.3

11.0

2.1

14.4

Max.

3.9

8.0

26.3

34.3

34.7

28.1

57.7

39.4

41.4

66.7

SD: standard deviation, SOC: soil organic carbon, CEC: cation exchange capacity.

Table 2

Number of observation in each score.

CoreVESS score

1

1.5

2

2.5

3

3.5

4

4.5

5

Number of observations

10

36

25

33

17

16

10

4

A. Johannes et al.

Geoderma 302 (2017) 14–21

16

⎡
⎣⎢

⎤
⎦⎥

⎡
⎣⎢

⎤
⎦⎥

COC

IF SOC

clay

n

THEN SOC ELSE

clay

n

=

<

[

]

Eq. 1

Note that the number of samples with

SOC

<

clay

n

and thus with

COC = SOC increases as n decreases. In a second part of their paper
dedicated to clay dispersibility,

Manrique and Jones (1991)

omitted all the

samples with SOC <

clay

n

from their analysis, de

ﬁ

ning:

⎡
⎣⎢

⎤
⎦⎥

⎡
⎣⎢

⎤
⎦⎥

COC

IF SOC

clay

n

THEN not available EL E

clay

n

=

<

[

]

S

Eq. 2

In the case of Eq.

1

, one should expect a correlation coe

ﬃ

cient

converging to a

ﬁ

nal value with decreasing n since the smaller n the less

the data set changes. In the case of Eq.

2

, the number of considered data

decreases, and the remaining samples should be the sandier ones, in
which a small n is more frequently observed. We used both Eqs.

1 and 2

to plot the correlation coe

ﬃ

cient between the physical properties as a

function of n. We refer to COC

1

and COC

2

when Eqs.

1 or 2

were used,

respectively.

3. Results and discussion

In the description of results, we shall

ﬁ

rst focus on the samples with

good structure (CoreVESS < 3) to examine the relationships between
SOC and porosities, and test the complexed organic carbon concept.
Then we considered samples of all the structural qualities to determine
the relationships between CoreVESS scores and the SOC:clay ratio.

3.1. Relationships between SOC, SOC:clay and porosities

Linear, logarithmic and broken-stick regressions were

ﬁ

t to the

experimental data of SOC, SOC:clay ratio and the physical properties
V

−

10 hPa

, V

dry

, W

−

10 hPa

, and Air

−

10 hPa

obtained at the 87 locations

with good structural quality (Sq < 3) (

Table 3

). The relationships

between SOC and physical properties are always better determined than
those between the SOC:clay ratio and physical properties. The values of

R

2

range from 0.37 to 0.79. A slightly smaller R

2

was obtained with the

BD to SOC relationship (not shown) instead of volume (1/BD) to SOC.
The break point of the broken stick model is never signi

ﬁ

cant, which

means that there is no threshold in the SOC values or the SOC:clay
ratios above which the e

ﬀ

ect of SOC decreased. The R

2

of the linear

model (

Fig. 2

) is larger or equal to the R

2

of the other models. It

describes best the relation between SOC or SOC:clay and the physical
properties of non-degraded soils (Sq < 3). The relationships between
porosities and SOC appear proportional, up to the highest value
observed in our study. According to this, the general rule is

“

the more

SOC the better the physical properties

”

, whatever the clay content.

These observations are in good agreement with those of

Heuscher et al.

(2005)

amd

who found that clay and silt

are poor predictors of the bulk density of Inceptisols, which are close to
Cambi-Luvisols, whereas SOC was a good predictor.

3.2. Application of the

“

complexed organic carbon

”

concept

According to

, COC determines the physical

properties of soils. The correlation coe

ﬃ

cients between the speci

ﬁ

c

volume (1/BD) of the dry soil and COC

1

or COC

2

as a function of the

clay:SOC ratio n are presented in

a and b, respectively, for the 87

soils in good structural state (Sq < 3). In both cases, the correlation
coe

ﬃ

cient continuously increased up to a maximum that is reached by

n = 6 to 7 corresponding to SOC:clay ratios of 0.14

–

0.16, that is when

COC is equal to SOC and total SOC is, therefore, taken into account
(

a). This ratio can be considered as the observed maximum,

di

ﬃ

cult to exceed in the

ﬁ

eld.

a shows that total SOC must be

taken into account to fully explain the measured physical properties.

When only the samples with SOC > COC

2

are considered (

b),

the number of samples included in the analysis decreases with
decreasing n, and the average texture of the samples becomes sandier.
This procedure, thus lead to an increasing bias with respect to the
representation of the entire sample population. Irrespectively of this,
though ranges of SOC and clay content in our study were larger than in

Table 3

Parameters of the linear, logarithmic and broken-stick models for several physical soil properties (speci

ﬁ

c volume at

−

10 hPa (V

−

10 hPa

), speci

ﬁ

c volume of dry soil (V

dry

), water content

at

−

10 hPa (W

−

10 hPa

) and speci

ﬁ

c air content at

−

10 hPa (A

−

10 hPa

)) as a function of soil organic carbon (SOC) content or SOC:clay ratio.

Predictor

Model

Equation

R

2

Adj. R

2

SOC

Linear

V

−

10 hPa

0.574

+ 0.113

SOC

0.788

0.786

V

dry

0.593

+ 0.064

SOC

0.632

0.627

W

−

10 hPa

0.197

+ 0.079

SOC

0.724

0.720

A

−

10 hPa

0.000 + 0.034

SOC

0.404

0.396

Logarithmic

V

−

10 hPa

0.641

+ 0.249

log (SOC)

0.775

0.772

V

dry

0.632

+ 0.149

log (SOC)

0.634

0.629

W

−

10 hPa

0.241

+ 0.179

log (SOC)

0.748

0.744

A

−

10 hPa

0.0226

⁎⁎

+ 0.070

log (SOC)

0.353

0.345

Broken-stick

V

−

10 hPa

0.470

+ 0.190 SOC - 0.081 (SOC - 1.43) for SOC > 1.43

0.792

0.784

V

dry

0.470

+ 0.159 SOC - 0.097 (SOC - 1.41) for SOC > 1.40

0.644

0.631

W

−

10 hPa

0.160

+ 0.101

SOC - 0.056 (SOC - 2.66) for SOC > 2.66

0.746

0.737

A

−

10 hPa

0.029 + 0.017 SOC + 0.039 (SOC - 2.57) for SOC > 2.57

0.442

0.421

SOC:clay ratio

Linear

V

−

10 hPa

0.638

+ 1.627

ratio

0.365

0.358

V

dry

0.605

+ 1.197

ratio

0.453

0.447

W

−

10 hPa

0.225

+ 1.301

ratio

0.437

0.430

W

−

10 hPa

0.037

⁎⁎

+ 0.322

⁎⁎

ratio

0.082

0.071

Logarithmic

V

−

10 hPa

1.219

+ 0.178

log (ratio)

0.368

0.360

V

dry

1.036

+ 0.132

log (ratio)

0.467

0.461

W

−

10 hPa

0.698

+ 0.146

log (ratio)

0.467

0.461

W

−

10 hPa

0.143

+ 0.031

⁎

log (ratio)

0.065

0.054

Broken-stick

V

−

10 hPa

0.601

+ 2.012

ratio - 2.321 (ratio - 0.157) for ratio > 0.157

0.401

0.379

V

dry

0.579

+ 1.472

ratio - 1.414 (ratio - 0.156) for ratio > 0.156

0.490

0.471

W

−

10 hPa

0.193

+ 1.623

ratio - 1.981 (ratio - 0.157) for ratio > 0.157

0.487

0.468

W

−

10 hPa

0.097

⁎

−

0.436 ratio + 0.988 (ratio - 0.093) for ratio > 0.093

0.128

0.095

V: Speci

ﬁ

c volume (cm

3

g

−

1

); W: Gravimetric water content (g g

−

1

); A: Gravimetric air content (cm

3

g

−

1

); SOC: Soil organic carbon (%).

Only the second equation of the broken-stick model is shown here. None of the breakpoints were signi

ﬁ

cant.

⁎

Indicates signi

ﬁ

cance level at p < 0.05.

⁎⁎

Indicates signi

ﬁ

cance level at p < 0.01.

⁎⁎⁎

Indicates signi

ﬁ

cance level at p < 0.001; adj. R

2

: adjusted R

2

.

A. Johannes et al.

Geoderma 302 (2017) 14–21

17

the databases used by

, none of the methods showed

an optimum of the correlation between COC and structural soil
properties, in contrast to the

ﬁ

ndings of these authors. This means that

our results do not support the hypothesis that complexed organic
carbon controls the speci

ﬁ

c soil volume or the soil bulk density. Our

data show instead that total SOC controls physical properties regardless
of clay content. Similar results were obtained with the other physical

parameters V

−

10 hPa

, Air

−

10 hPa

and W

−

10 hPa

(not shown). The main

di

ﬀ

erences between the two studies are that we (i) worked with a single

soil group, (ii) selected only samples with good structure, (iii) used a
speci

ﬁ

c sampling and volume measurement procedure, in particular

with free to swell samples equilibrated at standard matric potential, and
(iv) chose a sample population with a larger range of SOC and clay
content.

Fig. 2.

Linear relationship between speci

ﬁ

c volume at

−

10 hPa (V

−

10 hPa

), speci

ﬁ

c volume of dry soil (V

dry

), water content at

−

10 hPa (W

−

10 hPa

), air content at

−

10 hPa (Air

−

10 hPa

),

and soil organic carbon (SOC) content (left) or SOC:clay ratio (right) for samples with CoreVESS scores < 3.

A. Johannes et al.

Geoderma 302 (2017) 14–21

18

3.3. SOC:clay ratio and structure quality

If the pore volumes are proportional to SOC alone for soils with

good structural quality, is there a relation between SOC:clay and soil
structure quality in general? In other words, if the clay saturation by
SOC at a good structural state is not explaining the physical properties,
does this parameter account for the structural state?

a suggest that the average SOC:clay ratio is decreasing with

increasing score. The linear regression lines with no intercept for
Sq < 2 (very good structure), Sq = 3 (medium structure) and Sq >

4 (very poor structure) correspond to SOC:clay ratios (signi

ﬁ

cant

slopes) of 0.12, 0.09 and 0.08, respectively. The

“

good structure

”

domain (Sq < 3) corresponds to average SOC:clay ratios larger than
0.10 (

Figs. 4

a and

5

). Therefore, on average, the larger the clay content,

the larger the SOC content must be to observe given structure quality,
especially a good one. This is in agreement with the

ﬁ

ndings of

Feller

and Beare (1997)

who showed that at larger clay content a larger SOC

content was required to reach a given aggregate stability level, and
suggested that the concept of clay saturation is relevant for soil
structural quality.

3.4. SOC content criteria

b presents the SOC values as a function of clay content for the

good structured soils (Sq < 3), distinguishing between the di

ﬀ

erent

soil management practices. The distribution of the permanent grass and
tilled soils is close to that reported by

. In the area

with a SOC:clay ratio larger than 1:10 (

b, full line), mainly

samples from permanent grass can be found, but also some from no-
tillage and conventional tillage. Conventional tillage samples, however,

hardly overshoot the 1:10 ratio compared to permanent grass and no-
till, their highest ratio being 1:8 (

b, dashed line). Therefore, a

SOC:clay ratio of 1:10 seems to be a reasonable target to obtain a good
structure whatever the soil management. The ratio of 1:8 is the mean
ratio of the very good structured samples (Sq = 1,

Fig. 5

), and we

consider it as a

“

top value

”

. Conversely, the regression line for the

Fig. 3.

(a) Correlation coe

ﬃ

cient (r) between the speci

ﬁ

c volume of dry soil (V

dry

) and

complexed organic carbon (COC

1

) calculated following

, expressed as

a function of n (clay:SOC, Eq.

1

; black dots). The right ordinate corresponds to the

average percentage of SOC present as COC (open circles). (b) Correlation coe

ﬃ

cient (r)

between the speci

ﬁ

c volume of dry soil (V

dry

) and COC

2

calculated with Eq.

2

, as a

function of n (clay:SOC), as black dots, above which the number of observations is given.
The right ordinate corresponds to mean sand content represented in the diagram by the
symbol

“

x

”

. Samples with CoreVESS scores < 3.

Fig. 4.

Soil organic carbon content (SOC) as a function of clay content (a) for di

ﬀ

erent

CoreVESS scores (Sq) and (b) for di

ﬀ

erent soil management practices (PG: permanent

grass, NT: no-till, CT: conventional tillage) within soils of good structural state (Sq < 3).
The dashed line indicates a SOC:clay ratio of 1:8, the full line a SOC:clay ratio of 1:10, and
the dotted line a SOC:clay ratio of 1:13.

Fig. 5.

Boxplots of soil organic carbon (SOC) to clay ratio for di

ﬀ

erent CoreVESS scores.

Boxplots show mean values (cross), median values (solid horizontal line), 50th percentile
values (box outline), minimum and maximum values (whiskers) and outliers (open
circles). The dashed line indicates a SOC:clay ratio of 1:8, the full line a SOC:clay ratio of
1:10, and the dotted line a SOC:clay ratio of 1:13.

A. Johannes et al.

Geoderma 302 (2017) 14–21

19

relationship between SOC and clay for CoreVESS Sq > 4 samples
shown in

a corresponds to a SOC:clay ratio of 1:13 (see

Goutal-Pousse et al., 2016; Schä

b).

At this SOC:clay ratio, 63% of the samples show degraded (Sq > 3) or
moderate (Sq = 3) structure. This 1:13 ratio can be considered as a

“

minimum value

”

below which the structure is likely to be degraded.

Most of these samples are from conventionally-tilled

ﬁ

elds (55%), but a

considerable fraction also is from

ﬁ

elds under NT (33%). The proposed

SOC:clay criterions are summarized in

Table 4

.

As can be seen in

Figs. 4 and 5

, a high SOC:clay ratio is not su

ﬃ

cient

to have a good structure, since a soil with a large SOC content can be
mechanically compacted. On the other hand, the smaller the SOC
content, the more the soil can be a

ﬀ

ected by mechanical stresses

(

ﬀ

er et al., 2008

). It is, therefore,

logical to

ﬁ

nd an increasing VESS score with decreasing SOC, though

the structure degradation can also be due to low SOC (

Kay, 1998

;

Kay

and Munkholm, 2004

). On the contrary, some samples with small

SOC:clay ratio received a good score. An important factor in this
context is local heterogeneity, which is particularly high in tilled soils,
resulting in large scoring heterogeneity at the local scale (

Johannes

et al., 2016

). These uncertainties, however, are likely to be averaged

out due to the large number of data used to calculate the SOC:clay
ratios for the di

ﬀ

erent scores, and the proposed SOC:clay criteria are,

therefore, relevant. This is supported by the good partition between
permanent pasture and tillage obtained with the 1:10 ratio. Based on a
large survey of arable soil properties the criteria represent realistic
goals for application. The present results were established on samples
collected in the 5

–

10 cm layer and apply to the topsoil. Below the

plough layer, because SOC content drops dramatically, changes in clay
content would most likely become the leading parameter controlling
the physical properties.

Our results were obtained from samples from Cambi-Luvisols

developed on mixed moraine-molasses bedrocks. This is a wide group
covering a large area of Europe. Moreover, apart from the methodolo-
gical improvements that may explain some of the discussed discrepan-
cies with the literature, our results are generally in line with previously
reported observations, in particular regarding the importance of a
SOC:clay ratio of 1:10 and the relation between SOC and physical
properties. They may apply, therefore, to a much larger range of soils,
which should be tested.

4. Conclusions

For soils showing no evidence of physical stress or structure

degradation, the linear relation between SOC and soil pore volume
(or 1/BD) means that a SOC increase will result in a proportional
increase of soil porosity regardless of how much SOC is complexed to
clay. Contrarily to

org/10.1016/j.soilbio.2008.09.015

, we found no optimum in the

correlation between the physical properties and a COC fraction of the
SOC proportional to the clay content. The largest correlations were
observed when the SOC content was fully taken into account indicating
that total SOC controls physical properties rather than COC.

The SOC:clay ratio, however, appears to be a relevant criteria when

considering soil structure quality. Soils with visually evaluated good
structure quality have higher SOC:clay ratios than soils of poor

structural quality, and the di

ﬀ

erent structure quality scores correspond

in average to di

ﬀ

erent SOC:clay ratios. This allows the establishment of

criteria for SOC management. We proposed a ratio of 1:8 as a

ﬁ

eld

optimum for good structure quality, and 1:10 as a reasonable goal for
farmers, reachable even with tillage. Finally, 1:13 is a ratio below
which the structure quality is most likely unacceptable and needs
improvement. Nevertheless, the 1:8 and 1:10 SOC:clay ratios do not
guarantee a good soil structure, since mechanical damage may occur
regardless of SOC content. These results support the idea that com-
plexed organic carbon is a relevant concept for soil structural quality,
and that clay content has to be taken into account in the de

ﬁ

nition of

objectives for SOC content.

Acknowledgments

Funding provided by the Swiss Federal O

ﬃ

ce of Environment for

the STRUDEL project (13.001.KP/M044-1527) is gratefully acknowl-
edged. The authors would like to thank Léonie Givord, Tania Ferber,
Elisabeth Busset and Quentin Chappuis for their assistance in the
laboratory and in the

ﬁ

eld.

References

Abiven, S., Menasseri, S., Chenu, C., 2009. The e

ﬀ

ects of organic inputs over time on soil

aggregate stability

–

a literature analysis. Soil Biol. Biochem. 41, 1

–

12.

http://dx.doi.

Ball, B.C., Batey, T., Munkholm, L.J., 2007. Field assessment of soil structural quality

development of the Peerlkamp test. Soil Use Manag. 23, 329

a

Boivin, P., Brunet, D., Gascuel-Odoux, C., 1990. Densité apparente d'échantillon de sol:

337

.

méthode de la poche plastique (in french). Milieux Poreux Transf. Hydr. Bull. GFHN
28, 59

71

Davies, R.B., 2002. Hypothesis testing when a nuisance parameter is present only under

the alternative: linear model case. Biometrika 89, 484

–

489.

http://dx.doi.org/10.

1093/biomet/89.2.484

http://dx.doi.org/10.1016/j.geoderma.2008.01.022

Dexter, A.R., Richard, G., Arrouays, D., Czyz, E.A., Jolivet, C., Duval, O., 2008.

Complexed organic matter controls soil physical properties. Geoderma 144, 620

–

627.

http://dx.doi.org/10.1016/S0016-7061(97)00039-6

Feller, C., Beare, M.H., 1997. Physical control of soil organic matter dynamics in the

tropics. Geoderma 79, 69

–

116.

http://dx.doi.org/10.1016/j.geoderma.2015.10.017

Fernandez-Ugalde, O., Barre, P., Virto, I., Hubert, F., Billiou, D., Chenu, C., 2016. Does

phyllosilicate mineralogy explain organic matter stabilization in di

ﬀ

erent particle-

size fractions in a 19-year C-3/C-4 chronosequence in a temperate Cambisol?
Geoderma 264, 171

–

178.

http://dx.doi.org/10.1016/j.

Getahun, G.T., Munkholm, L.J., Schjønning, P., 2016. The in

ﬂ

uence of clay-to-carbon

ratio on soil physical properties in a humid sandy loam soil with contrasting tillage
and residue management. Geoderma 264, 94

–

102.

geoderma.2015.10.002

.

Goutal, N., Ranger, J., Boivin, P., 2012. Assessment of the natural recovery rate of soil

speci

ﬁ

c volume following forest soil compaction. Soil Sci. Soc. Am. J. 76, 1426

1435

http://dx.doi.org/10.1111/ejss.12323

Goutal-Pousse, N., Lamy, F., Ranger, J., Boivin, P., 2016. Structural damage and recovery

determined by the colloidal constituents in two forest soils compacted by heavy
tra

ﬃ

c. Eur. J. Soil Sci. 67, 160

–

172.

http://dx.doi.org/10.1111/j.1475-2743.2011.00354.x

Guimarães, R.M.L., Ball, B.C., Tormena, C.A., 2011. Improvements in the visual

evaluation of soil structure: visual evaluation of soil structure. Soil Use Manag no-no.

Heuscher, S.A., Brandt, C.C., Jardine, P.M., 2005. Using soil physical and chemical

properties to estimate bulk density. Soil Sci. Soc. Am. J. 69, 51

56

IUSS Working Group WRB, 2006. World reference base for soil resources 2006. A

framework for international classi

ﬁ

cation, correlation and communication. In: FAO

(Ed.), World Soil Reports. Food and Agriculture Organization of the United Nations,
Rome,

.

Je

ﬀ

rey, D., 1970. A note on use of ignition loss as a means for approximate estimation of

soil bulk density. J. Ecol. 58, 297.

http://dx.doi.org/10.2307/2258183

.

Johannes, A., Weisskopf, P., Schulin, R., Boivin, P., 2016. To what extent do physical

measurements match with visual evaluation of soil structure? Soil Tillage Res.

http://

dx.doi.org/10.1016/j.still.2016.06.001

Kay, B.D., 1998. Soil structure and organic carbon: a review. In: R. Lal, J.M.K. (Ed.), Soil

Processes and the Carbon Cycle, Advances in Soil Science. CRC Press, Boca Raton, FL,
pp. 169

197

Kay, B.D., Munkholm, L.J., 2004. Management-induced soil structure degradation

organic matter depletion and tillage. In: Schjønning, P., Elmholt, S., Christensen, B.T.
(Eds.), Managing Soil Quality: Challenges in Modern Agriculture. CABI Pub,
Wallingford, Oxon; Cambridge, MA, pp. 185

197

http://dx.doi.org/10.1071/SR11174

Keller, T., Dexter, A.R., 2012. Plastic limits of agricultural soils as functions of soil texture

and organic matter content. Soil Res. 50, 7.

1016/S0167-1987(02)00139-3

Loveland, P., Webb, J., 2003. Is there a critical level of organic matter in the agricultural

soils of temperate regions: a review. Soil Tillage Res. 70, 1

–

18.

http://dx.doi.org/10.

.

Manrique, L.A., Jones, C.A., 1991. Bulk density of soils in relation to soil physical and

Table 4

Expected structure quality as a function of the SOC:clay ratio.

SOC:clay

Expected structural quality

CoreVESS

⁎

> 1:8

Very good

< 2

1:10 < SOC:clay < 1:8

Good

2

–

3

1:13 < SOC:clay < 1:10

Improvement suggested SOC:clay of
1:10

3

–

4

< 1:13

Bad

> 4

⁎

Johannes et al., 2016

.

A. Johannes et al.

Geoderma 302 (2017) 14–21

20

chemical properties. Soil Sci. Soc. Am. J. 55, 476

–

481.

http://dx.doi.org/10.2136/

sssaj1991.03615995005500020030x

Muggeo, V.M., 2015. Package

.

ce fédéral de topographie, 1984. Atlas de la Suisse, 2e édition.

Saini, G., 1966. Organic matter as a measure of bulk density of soil. Nature 210, 1295.

http://dx.doi.org/10.1038/2101295a0

.

Schä

ﬀ

er, B., Schulin, R., Boivin, P., 2008. Changes in shrinkage of restored soil caused by

compaction beneath heavy agricultural machinery. Eur. J. Soil Sci. 59, 771

783

http://dx.doi.org/10.2136/vzj2011.0067

Schjonning, P., de Jonge, L.W., Munkholm, L.J., Moldrup, P., Christensen, B.T., Olesen,

J.E., 2012. Clay dispersibility and soil friability-testing the soil clay-to-carbon
saturation concept. Vadose Zone J. 11.

http://dx.doi.org/10.1016/j.still.2004.03.008

Six, J., Bossuyt, H., Degryze, S., Denef, K., 2004. A history of research on the link between

(micro)aggregates, soil biota, and soil organic matter dynamics. Soil Tillage Res. 79,
7

–

31.

.

Soane, B.D., 1990. The role of organic matter in soil compactibility: a review of some

practical aspects. Soil Tillage Res. 16, 179

–

201.

http://dx.doi.org/10.1016/0167-

, 1999. Soil taxonomy: a basic system of soil classi

ﬁ

cation for making and

interpreting soil surveys. In: Natural Resources Conservation Service. U.S.
Department of Agriculture Handbook, second ed. pp. 436

.

Stock, O., Downes, N.K., 2008. E

ﬀ

ects of additions of organic matter on the penetration

resistance of glacial till for the entire water tension range. Soil Tillage Res. 99,
191

–

201.

http://dx.doi.org/10.1016/j.still.2008.02.002

http://dx.doi.org/10.1890/02-0472

Toms, J.D., Lesperance, M.L., 2003. Piecewise regression: a tool for identifying ecological

thresholds. Ecology 84, 2034

–

2041.

.

Walkley, A., Black, I.A., 1934. An examination of the Degtjare

ﬀ

method for determining

soil organic matter, and a proposed modi

ﬁ

cation of the chromic acid titration

method. Soil Sci. 37, 29

–

38.

http://dx.doi.org/10.1097/00010694-193401000-

00003