Poeplau &amp; Don (2023). A simple soil organic carbon level metric beyond the organic carbon‐to‐clay ratio

). All those proper-

ties are directly coupled to the services that soils provide,

such as food security, water security or biodiversity protec-

tion (Kopittke et al.,

), making SOC an important in-

dicator of soil quality and health (Bünemann et al.,

).

The definition and differentiation or non- differentiation

of both terms vary across authors, but here we focus on

soil health as the ‘capacity of a soil to perform its key

functions and provide ecosystem services’ (Bonfante

et al.,

). In light of a steadily increasing global popula-

tion, severe soil degradation and climate change, healthy

soils and the definition of measurable and potentially

litigable soil health indicators are becoming increasingly

important (Allen et al.,

2011

; Lehmann et al.,

; van

Es & Karlen,

). The European Union is currently de-

veloping a Soil Health Law as a legal framework to achieve

the objectives of the European Soil Strategy (Panagos

et al.,

). The vision is that all soils in the European

Union will be healthy and that the protective and sustain-

able use of soils will be the norm by 2050. Possible soil

health indicators and respective threshold or reference

values are currently being discussed at multiple scientific

and political levels, with SOC content always among the

first to be highlighted (Bünemann et al.,

). However,

deriving a good, sufficient or even optimal level of SOC

is a complex task and remains soil- specific and hard to

generalize because (i) SOC is simultaneously relevant for

various biological, chemical and physical soil properties,

for all of which SOC content may have individual ranges

and thresholds, and (ii) SOC levels are site- specific and

depend on many abiotic factors such as soil texture and

climate (Drexler et al.,

). A definition of universal

SOC threshold values that could be developed using other

indicators, such as pH or contents of heavy metals, is thus

not feasible. SOC levels are, to a large extent soil- intrinsic

and can thus not be changed completely by management.

Any indicator that aims at evaluating the success of soil

management thus needs to consider the SOC variability

derived from non- changeable soil properties.

Clay content is an abiotic factor often found to be

among the most powerful explanatory variables of SOC

content at a regional scale, especially in agricultural soils

of near- neutral pH (Martin et al.,

;

; Rasmussen et al.,

). This can be re-

lated to clay's SOC- stabilizing properties, for example, its

high specific surface area and ability to form stable micro-

aggregates (Oades,

1988

; Wagner et al.,

2007

). At first

glance, it might thus make sense to normalize SOC by clay

content in a SOC:clay ratio to establish a clay- independent

indicator of the actual SOC level that allows soils to be

compared (Prout et al.,

A framework of different SOC:clay threshold values

for judging soil structural conditions has been introduced

by Johannes, Matter, et al. (

). According to the au-

thors, soils with a ratio of 1:8 or greater are considered to

have

very good

, between 1:8 and 1:10

good

, between 1:10

and 1:13

moderate

structural conditions, while SOC:clay

ratios smaller than 1:13 would indicate soils that are

likely to have a

degraded

soil structure (Johannes, Matter,

et al.,

; Prout et al.,

). These classes were de-

rived from an assessment of simple porosity parameters,

namely bulk volume, as well as water and air content at a

specific matric potential (−10 hPa), using a limited dataset

(161 data points from one broad soil group), with a simi-

larly limited range in SOC (8– 39 g kg

−1

) and clay content

(100– 340 g kg

−1

). It should also be noted that across the

full range of soils, SOC content alone showed a higher

correlation with all assessed structural parameters than

the SOC:clay ratio, suggesting a disconnect between the

SOC:clay ratio and the related structural soil property

(Johannes, Matter, et al.,

). Prout et al. (

) adopted

the framework and applied it to a large number of soils

(3809) from the UK, Poland and Switzerland, and sug-

gested that it is a meaningful index of the general state

of agricultural soils that should be used in other, similar

climate zones as well. However, statistically, the use of ra-

tios to control or eliminate the influence of the denomina-

tor can be problematic and needs to be handled with care

(Allison et al.,

1995

). They can be prone to extreme biases,

which is also likely in the case of the SOC:clay ratio. The

natural range in clay content is extensive, with anything

from

<

10 to

>

800 g kg

−1

being observed, even in managed

agricultural soils (Martin et al.,

In contrast, the range in SOC contents of mineral soils is

approximately one order of magnitude smaller, leading

to a foreseeable clay bias in the ratio. The present study

aimed to challenge the use of the SOC:clay ratio as a nor-

malized SOC level metric by applying it to 2958 topsoils

(0– 30 cm) from the German Agricultural Soil Inventory

dataset with a broad range of soil types, textures and SOC

contents. A further objective of this study was to develop

an unbiased alternative indicator based on the linear rela-

tionship between SOC and clay content that could serve as

a flexible, straightforward solution using clay to normalize

SOC contents at a regional scale.

2

|

MATERIALS AND METHODS

The dataset used in this study was compiled in the first

German Agricultural Soil Inventory conducted be-

tween 2011 and 2018. Sampling and analyses have been

14752743, 2023, 3, Downloaded from https://bsssjournals.onlinelibrary.wiley.com/doi/10.1111/sum.12921 by Hayley Peter-Contesse , Wiley Online Library on [18/10/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

|

1059

POEPLAU and DON

described in detail by Poeplau et al. (

set is published online (

) and the data-

https://doi.org/10.3220/DATA2

02002 03151139

). A total of 3104 soil profiles in an 8

×

8 km

grid were sampled to a depth of 100 cm. In this study, the

analysis was restricted to mineral topsoils, i.e. the upper

30 cm, with average SOC contents of

<

87 g kg

−1

(

n

=

2958,

consisting of 2254 cropland and 704 grassland soils). The

relevant depth intervals for this study were 0– 10 and 10–

30 cm. Topsoil values (0– 30 cm) for SOC and clay contents

were derived by mass- weighted averaging of the 0– 10 and

10– 30 cm depths, using the respective fine soil (

<

2 mm)

amount of each depth increment. Derived topsoil SOC

and clay content ranges were 2.6– 122.9 g kg

−1

and 9.6–

822.5 g kg

−1

, respectively. SOC:clay ratios were calculated

for all sampling points and grouped into the above-

mentioned categories of Johannes, Matter, et al. (

),

Johannes, Weisskopf, et al. (

), i.e.

very good

(

>

1:8),

good

(1:8– 1:10),

moderate

(1:10–

1:13) and

degraded

(

<

1:13). Although initially developed in the context of soil

structure, the rationale behind this index is to provide a

basis for comparing actual SOC levels of differently tex-

tured soils. The developed score ranging from

degraded

to

very good

thus basically refers to a normalized SOC level

that has been used and could be used further completely

decoupled from soil structure (Prout et al.,

) but

rather as a reference system for judging the actual SOC

level. In the present study, these classes were also used to

evaluate SOC levels per se.

In the first part of the study, the clay dependency of the

SOC:clay ratio was tested by (i) performing a direct assess-

ment of the SOC:clay ratio as a function of clay content

and (ii) calculating the proportions of

degraded

,

moder-

ate

,

good

and

very good

SOC levels along the whole range

of clay contents in seven increment classes of 100 g kg

−1

each (

<

100, 100– 199, 200– 299, 300– 399, 400– 499, 500–

599,

>

600). In addition, differences in the normalized

SOC levels per land use (cropland and grassland), as well

as soil group based on the World Reference Base (WRB)

classification, were also evaluated. Permanent grasslands

were defined as sites that were used as grasslands for at

least five consecutive years. A certain proportion of crop-

lands and grasslands in the dataset had a land use change

from the respective other category in the last decades and

it can be expected that SOC contents will continue to rise

or fall in these cases (Springob et al.,

2001

). However, in

this clustering approach, land use history was ignored.

In the second part of the study, a less clay- biased ratio

was developed and tested as an alternative. To do this, first,

the regression equation for a linear regression between

SOC and clay content was derived. This was done to estab-

lish an average SOC value that can be expected at a certain

clay content in German agricultural topsoils (SOC

exp

). The

actual SOC content of each site was then divided by SOC

exp

to derive a ratio that would inform how much SOC is cur-

rently stored compared with what might be expected from

the site's clay content. The quartiles of this SOC:SOC

exp

ratio dataset were then used in a similar way to Johannes,

Matter, et al. (

) to determine the four previously intro-

duced SOC- level classes and their threshold values. In this

case, the threshold values were not derived in response to

any functional properties, but rather exemplarily in the

most pragmatic way. Soils with SOC contents below the

regression line of SOC and clay content were considered

to have

degraded

(1st quartile) or

moderate

(2nd quartile)

SOC levels, while SOC levels of soils with SOC contents

above this line were considered

good

(3rd quartile) or

very

good

(4th quartile). Second, these classifications, based

on the newly introduced operationally defined normal-

ized SOC levels, were compared with the classifications

based on the suitability of the SOC:clay ratio as an indica-

tor of soil structural conditions. This was done using the

inverse of bulk density (bulk volume, 1/BD, cm

3

g

−1

) that

describes soil porosity as an important structural param-

eter (Johannes, Matter, et al.,

; Johannes, Weisskopf,

et al.,

). Differences in bulk volume between SOC

level classes in each of the two schemes were tested using

the non- parametric Kruskal- Wallis test and the Wilcoxon

test as a post hoc test. Significance was assessed at

p

<

.05.

Calculations and plotting were performed using R version

4.2.1 (R Development Core Team,

2010

) and the pack-

ages tidyverse and ggplot2 (Wickham,

; Wickham

3

|

RESULTS AND DISCUSSION

3.1

|

Evaluation of the SOC:clay ratio

Using the SOC:clay ratio scheme, 37% of all cropland and

14% of all grassland topsoils in Germany were found to

have

degraded

SOC levels (

Table

1

). This might appear

reasonable since cropland soils are acknowledged to be

depleted in SOC as compared with soils under grassland

(Guillaume et al.,

). The proportions of soils with de-

graded SOC levels were similar to those found for soils in

the UK, Switzerland and Poland, where 38% and 7% of

all cropland and grassland soils had

degraded

SOC levels

(Prout et al.,

). However, a closer look at the different

soil reference groups revealed that the use of SOC:clay ratio

as an indicator was problematic and certainly misleading:

clay- poor Podzols, irrespective of land use, were always

found to have

very good

SOC levels, with median SOC:clay

ratios around 1:1. The ratio is thus not sensitive enough to

indicate degradation of sandy soils since all these sandy

soils were far beyond the defined threshold values of 1:8.

The opposite applied to more fine- textured soils: 97% of

14752743, 2023, 3, Downloaded from https://bsssjournals.onlinelibrary.wiley.com/doi/10.1111/sum.12921 by Hayley Peter-Contesse , Wiley Online Library on [18/10/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

1060

|

POEPLAU and DON

all clay- rich Vertisols had

degraded

SOC levels (median

SOC:clay ratio of 1:21) and it would require immense ef-

forts (

+

62% SOC) to achieve even a shift from

degraded

to

moderate

(threshold of 1:13). Moreover, Chernozems are

also fine- textures soils and are considered the most fertile

soils with generally high SOC contents and a favourable

soil structure (Šimansky & Jonczak,

2016

). Nonetheless,

according to the SOC:clay ratio, only 2% of all chernozems

under agricultural use in Germany had

very good

SOC

levels, while 61% only

degraded

. Even the one single

Chernozem under grassland was

degraded

. These cases

highlighted the over- dependence of SOC:clay ratios on

clay content, as visualized in

Figure

1a,b

.

The proportion of soils with

degraded

SOC levels,

which was close to zero in coarse- textured soils, increased

exponentially with clay content (

Figure

1b

). It is not real-

istic for clay- rich soils to be particularly depleted in SOC,

e.g. by agricultural management, since there is a posi-

tive correlation between the fine particle content (clay

and fine silt) and the relatively stable mineral associated

SOC (Hassink,

1997

). In coarse- textured soils, particulate

organic matter is the dominant form of organic matter,

which is considered more labile and more responsive

to disturbances such as land use change or agricultural

management (Lavallee et al.,

; Poeplau & Don,

Vos et al.,

). This fits well with the fact that Prout

et al. (

) observed more pronounced SOC losses for ag-

ricultural soils in England and Wales with higher SOC:clay

ratios, and might suggest that the SOC:clay ratio could be

used as a rough proxy for the amount of unprotected SOC

prone to losses (Dexter et al.,

2008

proposed by (Johannes, Matter, et al.,

However, the indicator's extreme clay dependency

hampered both the comparison of soils within a similar

textural class as well as a comparison across all textural

classes. In both, coarse and fine- textured soils, the indi-

cator is insensitive to changes in SOC content. Losses or

gains in SOC needed to change class are beyond expect-

able SOC changes because of management, land use or

climate change within years to a few decades. For exam-

ple, in the Ultuna Frame Trial, a Swedish long- term fertil-

ization experiment that started in 1956 on soil with a high

clay content, different types of organic amendments were

TABLE 1

Proportions (%) of cropland and grassland topsoils (

n

) classified in the four different levels based on the SOC:clay ratio score

) by World Reference Base (WRB) soil group and land use, and median SOC:clay ratios.

Land use

Soil class

n

Degraded

Moderate

Good

Very good

Median

SOC:Clay

Cropland

Anthrosols

178

21

7

6

66

0.25 (1:4)

Cambisols

609

28

19

12

41

0.10 (1:10)

Chernozems

89

61

33

4

2

0.07 (1:14)

Fluvisols

12

50

33

8

0.07 (1:14)

Gleysols

175

21

12

4

63

0.19 (1:5)

Luvisols

321

39

26

17

19

0.08 (1:12)

Phaeozems

211

48

23

13

16

0.08 (1:12)

Podzols

84

0

100

0.74 (1:1.3)

Regosols

224

68

15

7

10

0.06 (1:17)

Stagnosols

293

35

18

14

33

0.09 (1:11)

Vertisols

58

97

0

3

0

0.05 (1:21)

All

2254

37

18

10

35

0.09 (1:11)

Grassland

Anthrosols

60

8

2

82

0.28 (1:4)

Cambisols

222

9

21

22

48

0.12 (1:8)

Chernozems

1

100

0

0.06 (1:14)

Fluvisols

3

33

0

0.10 (1:10)

Gleysols

106

8

12

9

70

0.19 (1:5)

Luvisols

28

11

39

18

32

0.10 (1:10)

Phaeozems

91

12

40

20

29

0.10 (1:10)

Podzols

20

0

100

1.01 (1:1)

Regosols

52

35

15

17

33

0.10 (1:10)

Stagnosols

103

17

18

26

38

0.11 (1:9)

Vertisols

18

61

33

6

0

0.07 (1:15)

All

704

14

21

17

48

0.12 (1:8)

14752743, 2023, 3, Downloaded from https://bsssjournals.onlinelibrary.wiley.com/doi/10.1111/sum.12921 by Hayley Peter-Contesse , Wiley Online Library on [18/10/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

|

1061

POEPLAU and DON

applied to the soil in doses of 4 Mg C every 2 years. The

organic amendments included straw, green manure, farm-

yard manure, sewage sludge, sawdust and peat (Kätterer

et al.,

2011

). At the start of the trial, the soil had an aver-

age SOC content of 15 g kg

−1

(SOC:clay ratio of 1:24). After

54 years of continuous organic fertilization, only peat, saw-

dust and sewage sludge amendments— all rarely used or-

ganic fertilizers on agricultural soil— resulted in a change

of SOC level from

degraded

to

moderate

, while SOC in all

other treatments remained at the

degraded

level. This is

in contrast to the results of the Woburn experiment high-

lighted by Prout et al. (

), in which farmyard manure

addition was able to change the SOC level from

degraded

to

good

within 7 years. The main difference between the

two experiments was the clay content, which was 78–

131 g kg

−1

in Woburn and 365 g kg

−1

in Ultuna.

It could be argued that only the extreme soils within

the whole texture range would be affected by this clay bias

and that those could simply be neglected when apply-

ing the SOC:clay ratio indicator. Nevertheless, Johannes,

Matter, et al. (

), Johannes, Weisskopf, et al. (

) de-

veloped the scheme for soils with clay contents of between

100 and 340 g kg

−1

. However, even within such a restricted

range of clay contents, the SOC:clay ratio is strongly bi-

ased: the same amount of additional SOC would lead to

a 3.4 times higher absolute increase in the SOC:clay ratio

for a 100 g kg

−1

clay soil than for a 340 g kg

−1

clay soil.

Furthermore, a general SOC- level metric should only qual-

ify as such when it is broadly applicable. In regions with

glacial or glacio- fluvial sediments, coarse- textured soils

with low clay contents are common (Gebauer et al.,

Finally, and perhaps most importantly, sandy soils are

mostly low in SOC and also have a poor soil structure.

In addition to having a low water- holding capacity, they

also have a comparatively low pH, low cation exchange

capacity and low nutrient storage. With regard to soil

functions, ecosystem resilience and yield stability, sandy

soils may thus be most in need of SOC, which is able to

bring about positive change in most of the properties men-

tioned (Moinet et al.,

2023

; Murphy,

). However, the

SOC:clay ratio, as it is used now, would indicate the exact

opposite: sandy soils are generally classified as in

very good

structural condition and SOC level and improving them is

neither needed nor possible. According to the SOC:clay

indicator, clay- poor soils with, for example, 50 g kg

−1

clay

are in the

very good

SOC level with as little as 6.2 g kg

−1

SOC. Thus, the strong clay dependency of this indicator

makes it inappropriate for its use in a soil health context.

3.2

|

The newly introduced

SOC:SOC

exp

Ratio

Despite clay content being too strong a denominator for

the normalization of SOC, its use might still be meaning-

ful. It is a widely available standard soil property and is

often observed to be a strong predictor of SOC in regional-

scale case studies, at least in temperate regions (Martin

et al.,

2011

). One option that is proposed and tested here

is to use the correlation between SOC and clay content

FIGURE 1

(a) Soil organic carbon (SOC)- to- clay ratio as a function of clay content, using the four SOC level classes suggested by

Johannes, Matter, et al. (

), applied to German agricultural topsoils (0– 30 cm, 2958 sites under cropland and grassland use) and (b) the

proportion of each class as a function of clay content. In this case, the data were aggregated in steps of 100 g kg

−1

clay content (therefore

starting with 100 g kg

−1

as the first data point).

0

1

2

3

degraded

good

moderate

very good

SOC:clay rati

o

Clay content [g kg

-1

]

0 200 400 600 800

Clay content [g kg

-1

]

200 400 600

(a)

0

25

50

75

100

degraded

moderate

good

very good

Proportion [%

]

(b)

14752743, 2023, 3, Downloaded from https://bsssjournals.onlinelibrary.wiley.com/doi/10.1111/sum.12921 by Hayley Peter-Contesse , Wiley Online Library on [18/10/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

1062

|

POEPLAU and DON

to establish an average expected SOC content (SOC

exp

)

depending on the clay content as a denominator. This

normalization of SOC is less strong than clay content as

a denominator. When the SOC:SOC

exp

ratio is

>

1, the

SOC content is above expected SOC and thus above aver-

age SOC, and when the ratio is below 1 the soil is rather

SOC- depleted and below average. The correlation be-

tween clay and SOC contents for all German Agricultural

topsoils provided the new denominator with SOC

exp

=

c

lay

×

0.0288

+

13.674, where

R

2

was 0.11. Although the

correlation was highly significant (

p

<

.001),

R

2

was rela-

tively low because of a certain group of soils, which are

recognized as ‘black sands’ (Vos et al.,

) and devi-

ate strongly from the SOC

~

clay relationship (

Figure

2a

).

These coarse- textured soils are characterized by high

SOC contents, mainly in the form of recalcitrant plant

litter originating from historic land cover as heathland

or peatland (Springob et al.,

2001

; Vos et al.,

, there was no clay dependency of

).

However, as an important part of the whole ensemble of

German agricultural soils, they were not excluded from

the derivation of SOC

exp

here. The derived threshold

values of the SOC:SOC

exp

ratio for the classification of

German agricultural soils were 0.65 (threshold between

degraded

and

moderate

), 0.83 (

moderate

/

good

) and 1.16

(

good/very good

). Using these thresholds, this study found

that 32% of all cropland soils and only 3% of all grassland

soils had

degraded

normalized SOC levels (

Table

2

). This

was of a comparable magnitude to those observed for the

SOC:clay ratio for our dataset. However, of the 940 top-

soils that had degraded SOC levels using the SOC:clay

ratio, only 407 (43%) also had degraded SOC levels using

the SOC:SOC

exp

ratio. This indicates that the two ap-

proaches classified the soils differently. Importantly, as

depicted in

Figure

2b

the classification based on the SOC:SOC

exp

ratio. Also

for the Ultuna long- term experiment, on a fine- textured

soil, SOC:SOC

exp

was far more sensitive to management

changes than the SOC:clay ratio: Depending on the or-

ganic amendment treatment, the initially degraded SOC

level of this trial was changed into all possible classes

within the first 20 years. Of course, these thresholds are

operationally defined and specific to the set of samples

used. Nevertheless, this example showed that the metric

derived from German agricultural topsoils could produce

meaningful results also for an experiment in Central

Sweden.

Soil organic carbon is acknowledged to affect many soil

functional properties (Bagnall et al.,

; Murphy,

However, to qualify as an indicator in a soil health context,

also a SOC level metric should be related to soil functions.

The SOC:clay ratio was initially developed together with

soil structure indicators (Johannes, Matter, et al.,

Also here we exemplary use bulk volume as an indica-

tor of porosity to evaluate which SOC level metric can

be better linked to soil structural condition. As shown in

Figure

3

, SOC:SOC

exp

was a better indicator of soil struc-

ture than the SOC:clay ratio: In the case of SOC:clay, the

different classes differed only slightly in their average

porosity and in part not significantly: The class

good

had

the highest porosity, while

very good

and

moderate

were

not statistically different. In contrast, the classes based

on the SOC:SOC

exp

ratio were in the expected order and

FIGURE 2

(a) Soil organic carbon (SOC) content as a function of clay content in 2958 German agricultural topsoils (0– 30 cm depth)

with the four classes based on the SOC:SOC

exp

ratio, and (b) the proportion of soils in each class as a function of clay content. In this case,

the data were aggregated in steps of 100 g kg

−1

clay content.

Clay content [g kg

-1

]

125

100

75

50

25

0

SOC content [g kg

-1

]

200 400 600

0 200 400 600 800

degraded

moderate

good

very good

0

25

50

75

100

Proportion [%

]

(b)

(a)

Clay content [g kg

-1

]

14752743, 2023, 3, Downloaded from https://bsssjournals.onlinelibrary.wiley.com/doi/10.1111/sum.12921 by Hayley Peter-Contesse , Wiley Online Library on [18/10/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

|

1063

POEPLAU and DON

all statistically different from each other. Despite a cer-

tain overlap across SOC- level classes, the boxes were also

much smaller than for the classes based on SOC:clay, in-

dicating a better separation. This can be seen as an inde-

pendent validation of SOC:SOC

exp

being preferable over

SOC:clay even as a structural indicator.

3.3

|

Challenges of deriving SOC

classification schemes and potential

ways forward

The results of this study clearly suggest that a revised and

attenuated version of the SOC:clay ratio, that is, using

TABLE 2

Proportions (%) of the different normalized soil organic carbon (SOC) level classes for topsoil (0– 30 cm depth) based on the

quartiles of the SOC:SOC

exp

ratio by World Reference Base (WRB) soil group and land use, with the number of observations (

n

) and median

SOC:SOC

exp

ratio.

Land use

Soil class

n

Degraded

Moderate

Good

Very

good

Median

SOC:SOC

exp

Cropland

Anthrosols

178

18

19

22

40

1.00

Cambisols

609

34

23

8

0.72

Chernozems

89

20

35

43

2

0.78

Fluvisols

12

8

42

33

17

0.78

Gleysols

175

11

20

28

41

1.04

Luvisols

321

58

32

8

1

0.62

Phaeozems

211

30

37

24

9

0.73

Podzols

84

5

2

21

71

1.60

Regosols

224

34

28

30

7

0.75

Stagnosols

293

26

35

24

14

0.77

Vertisols

58

38

33

24

5

0.70

All

2254

32

30

23

15

0.75

Grassland

Anthrosols

60

5

12

25

58

1.32

Cambisols

222

4

10

32

55

1.21

Chernozems

1

0

100

0

0.96

Fluvisols

3

0

67

33

0.94

Gleysols

106

3

26

68

1.45

Luvisols

28

7

39

32

21

0.84

Phaeozems

91

1

7

33

59

1.23

Podzols

20

0

100

2.21

Regosols

52

4

8

23

65

1.37

Stagnosols

103

3

7

38

52

1.19

Vertisols

18

11

0

44

1.14

All

704

3

9

30

58

1.25

FIGURE 3

Soil porosity of 2958

soil samples as distributed into classes

derived according to (a) the soil organic

carbon- to- clay (SOC:clay) ratio (Johannes,

Matter, et al.,

) and (b) the quantiles

of the measured- to- expected SOC ratio

(SOC:SOC

exp

). Letters depict significant

differences between classes (

p

<

.05).

0.50

0.75

1.00

1.25

1.50

(a)

SOC:clay ratio

(b)

SOC:SOC

exp

0.50

0.75

1.00

1.25

1.50

Bulk volume [cm³

g

-1

]

degraded moderate good very good

a

b

c

a

b

c

d

14752743, 2023, 3, Downloaded from https://bsssjournals.onlinelibrary.wiley.com/doi/10.1111/sum.12921 by Hayley Peter-Contesse , Wiley Online Library on [18/10/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

1064

|

POEPLAU and DON

the average regression equation of SOC vs. clay content

as a denominator instead of clay content, is preferable.

However, the first challenge is that both the SOC

~

clay

relationship, as well as SOC per se, are strongly depend-

ent on the pedoclimatic zone as well as on the spatial

scale. Rasmussen et al. (

) showed that the suitability

of clay as a predictor of SOC depends much on the type

of minerals present, as well as on soil pH. Also, on larger

scales (continental to global), climatic factors become in-

creasingly important and mask the clay effect to a certain

extent (Delgado- Baquerizo et al.,

; García- Palacios

et al.,

). Within the European Union, it will thus not

be possible to have one desirable SOC level and simple

SOC classification scheme (e.g. by clay content or texture

class) that fits equally well for all soils, e.g. Scandinavian

and Mediterranean soils. However, as a first step, the prag-

matic approach taken here could be adopted by others

and applied using regional SOC

~

clay relationships and

related quartiles. Apart from the fact that the SOC:SOC

exp

ratio is not clay- biased, it also has the advantage that the

absolute numbers can be compared across different pedo-

climatic regions, soil depths and land use types because it

resembles a normalization by what is observed on aver-

age in a given region at a given clay content. This means

that SOC

~

clay relationships in 0– 30 cm soil depth will

differ in intercepts and slopes from the same relationship

in 0– 10 cm, which is also true for different pedoclimatic

zones, for example. This is not the case for the SOC:clay

ratio, which does not take context into account at all.

Subsoil (30– 100 cm) data are not shown in this study, but

similar results for the SOC:SOC

exp

ratio were obtained.

Depending on the question, this might also be a disadvan-

tage of the SOC:SOC

exp

metric: it might not necessarily

be helpful to evaluate the absolute average SOC content

of a country or region compared with others; it does not

relate measured SOC to a certain fixed reference content

and is thus solely based on the population used to build

the model. In this respect, the approach is similar to that

taken by Drexler et al. (

), who defined benchmark

SOC contents for German agricultural soils using several

important explanatory variables such as texture, C:N ratio

and precipitation. This scheme was developed for farmers

and thus requires little soil data, but allows farmers' meas-

ured SOC contents to be compared with those of similar

sites. In their case, the benchmarking was also done with

current SOC contents found in German agricultural soils

that might not be optimal or desirable. Likewise, Chen

et al. (

) used a data- driven clustering approach to de-

rive carbon- landscape clusters in France and calculated

the highest possible SOC content within each individual

cluster using the percentile of 0.9, for example. This was

mainly done to estimate the SOC storage potential of ar-

able land in France, but could potentially be used to define

a cluster- specific range of SOC contents as related to a cer-

tain achievable SOC content. Again, the problem is that

the highest values within a specific cluster do not neces-

sarily have to resemble the highest achievable SOC con-

tent if, for example, the whole cluster is SOC depleted. In

each case, the chosen thresholds (quartiles in the present

study), or the 0.9 quartile are somewhat arbitrary, but (i)

we were able to show that the quartile thresholds used in

this study were able to separate soils with significantly

different porosities and (ii) a standardized methodology

across different countries or regions might be more im-

portant than the question at which exact threshold a soil

changes from one to the other SOC level.

The data in the present study show, for example, by

the relatively low regression coefficient, that it is a sim-

plification to derive an expected SOC content for a site

solely by the parameter clay content. Numerous other soil

properties determine the SOC content, with groundwater

level among the most prominent (Poeplau et al.,

The high fraction of groundwater- influenced Gleysols

classified as

good

and

very good

indicates that beyond clay

content, the water regime also positively influences SOC

contents (

Table

2

). Furthermore, Doetterl et al. (

)

highlighted the strong impact of geochemistry, so basically

bedrock properties, on SOC contents. This might be more

accounted for, when, for example, certain landscape units,

pedo- climatic zones or similar are used as a grouping vari-

able (Chen et al.,

). At the same time, such spatial

clusters can also be highly heterogeneous so that such a

clustering might not necessarily lead to a clear separation

either (Chen et al.,

). It is therefore questionable and

certainly context- specific, how the denominator in such

a SOC:SOC

exp

should optimally look like. Even more so-

phisticated and potentially applicable on large spatial

scales is the reciprocal modelling approach using machine

learning, as introduced by Schneider et al. (

). Here,

beyond clay also many other drivers could be used to

model a pedo- climatic reference SOC. Such reference SOC

content would also need to be defined, e.g. that of grass-

land soils, and a statistical model used to estimate how

much the SOC content of a given cropland soil deviates

from its hypothetical grassland SOC content. The result-

ing residue could then potentially be scaled into an indi-

cator of degradation. Such an approach is data hungry and

also has its weaknesses, mainly in the critical assumptions

that need to be made (Schneider et al.,

). However,

this approach should be investigated further for data- rich

regions as a potential way forward towards a comprehen-

sive evaluation scheme for SOC. Finally, we argue that

clay content is a soil property that (i) has been shown to

strongly affect SOC storage, (ii) is mostly available and

(iii) does not severely change over time and thus quali-

fies to be used in such a normalization approach. More

14752743, 2023, 3, Downloaded from https://bsssjournals.onlinelibrary.wiley.com/doi/10.1111/sum.12921 by Hayley Peter-Contesse , Wiley Online Library on [18/10/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

|

1065

POEPLAU and DON

dynamic soil properties that also influence SOC, such as

groundwater level or pH, might be better suited as explan-

atory variables for changes in the numerator. For example:

If drainage is introduced in a certain area, leading to an

increased mineralization of SOC (Castellano et al.,

),

this will be reflected in the SOC:SOCexp over time.

To overcome the problem of SOC:SOC

exp

only pro-

viding meaningful insights in relation to the current sit-

uation of the whole population, it is proposed that the

SOC:SOC

exp

- derived classes be understood as a simple

baseline benchmarking system (e.g. country- specific) that

could be used to monitor and evaluate SOC over time. The

regression is established with the given baseline dataset

at a certain time, in which the whole dataset is classified

into equal shares (quartiles) of

degraded

,

moderate

,

good

and

very good

. The quartile thresholds will then be fixed

for the country (or similar spatial unit) and changes over

time will always be related to exactly those thresholds of

initial SOC conditions. Then, in repeated soil invento-

ries, for example, the development in the proportions of

the different classes over time can be evaluated. Specific

goals such as ‘no soils with

degraded

SOC levels by year

x’, or ‘25% more soils with

very good

SOC levels by year

y’ could be set as political targets. The major advantage

of a scoring approach of this kind over simply following

the trends in SOC per se would be the direct comparability

across countries (e.g. within the EU), which have their in-

dividual baseline levels, established sampling depths and

also different analytical methods. It could potentially also

be embedded in more complex soil health- scoring func-

tions including other indicators, comparable to the CASH

framework (Fine et al.,

).

However, a more general problem is that an

optimal

,

sufficient

or

degraded

SOC content is difficult to define

and the definition may greatly depend on the desired func-

tions of SOC. For biological and chemical functions, such

as fuelling soil biota and increasing nutrient storage, there

could be something like an optimum value that should not

be exceeded in order to avoid negative environmental im-

pacts such as nitrate leaching or increased N

2

O emissions

(Yanai et al.,

2003

). In contrast, for soil physical proper-

ties such as water- holding capacity and structural stabil-

ity, there seems to be a positive linear relationship with

SOC content (Chaney & Swift,

1984

; Johannes, Matter,

et al.,