Begill et al. (2023). No detectable upper limit of mineral‐associated organic carbon in temperate agricultural soils

).

To better understand soil organic carbon (SOC) dynamics, SOC is
conceptually divided into fractions of contrasting properties, func-
tions and residence times (von Lützow et al.,

2007

). In recent years,

the concept of having two specific, operationally defined fractions
becomes more and more accepted: those two are particulate organic

carbon (POC) and mineral- associated organic carbon (MAOC), which
are often separated by size (Lavallee et al.,

). POC is formed

by fragmentation and translocation of structural litter residues and
thus resembles the young and rather labile part of SOC, with turn-
over times of years to decades. In contrast, MAOC is formed either
by direct sorption of plant- derived dissolved OC to mineral surfaces
(ex vivo) or microbially mediated transformation and necromass as-
sociation with minerals (in vivo; Liang et al.,

2017

). MAOC was found

to contain less chemically complex compounds (e.g. polysaccharides,
proteins, lipids) as compared to POC (e.g. lignin, cellulose, hemicel-
lulose), have a longer turnover time (from decades to centuries), and
lower C:N ratios (Lavallee et al.,

Creative Commons Attribution-NonCommercial

). It is thus particularly the

MAOC fraction, that is SOC stored in the fine fraction of the soil
(clay and silt- sized particles), that should be targeted for long- term

Received: 29 March 2023

|

Accepted: 11 May 2023

DOI: 10.1111/gcb.16804

R E S E A R C H A R T I C L E

No detectable upper limit of mineral- associated organic carbon

in temperate agricultural soils

Neha Begill

|

Axel Don

|

Christopher Poeplau

This is an open access article under the terms of the

License, which permits use, distribution and reproduction

in any medium, provided the original work is properly cited and is not used for commercial purposes.
© 2023 The Authors.

Global Change Biology

published by John Wiley & Sons Ltd.

Thünen Institute of Climate- Smart

Agriculture, Braunschweig, Germany

Correspondence

Christopher Poeplau, Thünen Institute of
Climate- Smart Agriculture, Bundesallee

68, 38116 Braunschweig, Germany.

Email:

christopher.poeplau@thuenen.de

Funding information

Bundesministerium für Ernährung und
Landwirtschaft; Horizon 2020 Framework
Programme

Abstract

Soil organic carbon (SOC) sequestration is a promising climate change mitigation op-
tion. In this context, the formation of the relatively long- lived mineral- associated or-
ganic carbon (MAOC) is key. To date, soils are considered to be limited in their ability
to accumulate MAOC, mainly by the amount of clay and silt particles present. Using

the comprehensive German Agricultural Soil Inventory, we selected 189 samples with
a wide range of SOC (5– 118 g kg

−1

) and clay contents (30– 770 g kg

−1

) to test whether

there is a detectable upper limit of MAOC content. We found that the proportion
of MAOC was surprisingly stable for soils under cropland and grassland use across
the whole range of bulk SOC contents. Soil texture influenced the slope of the re-
lationship between bulk SOC and MAOC, but no upper limit was observed in any

texture class. Also, C content in the fine fraction (g C kg

−1

fraction) was negatively

correlated to fine fraction content (g kg

−1

bulk soil). Both findings challenge the notion

that MAOC accumulation is limited by soil fine fraction content per se.

K E Y W O R D S

carbon saturation, carbon sequestration, national soil monitoring, organic matter fractionation

|

4663

BEGILL

et

al

.

SOC sequestration in the context of climate change mitigation. This
does not imply that POC storage is ineffective, considering its im-
portance for soil structure and as a nutrient and energy source.

However, it is claimed that the capacity of a soil to stabilise ad-

ditional C as MAOC is limited by the amount of fine fraction and its
mineral surfaces (Cotrufo et al.,

; Hassink,

). It is a common

perception that the fine fraction ‘saturates’ at a certain point, which
would restrict measures fostering C- sequestration to so- called ‘un-

dersaturated’ soils (Georgiou et al.,

). The theory originated from

observations that the amount of C and N associated with fine silt
and clay particles was positively correlated with the amount of such
particles in uncultivated and grassland soils (Hassink,

). Based

on this, numerous studies have estimated the ‘saturation deficit’ and
subsequently the SOC sequestration potential of soils, leading to
the conclusion that mineral surface area must be one of the most
limiting factors in SOC stabilisation (Angers et al.,

2011

; Georgiou

et al.,

). The original saturation limit of mineral soils, as sug-

gested by Hassink (

), has been changed by several authors who

have detected ‘oversaturated’ soils (Feng, Plante, et al.,

; Feng,

Xu, et al.,

; Fornara et al.,

; Wenzel et al.,

) that is

soils with more MAOC than the theoretical upper limit would allow.
This indicates that the upper limit of MAOC storage as derived from
upper quantiles (Feng, Plante, et al.,

; Feng, Xu, et al.,

) of

the correlation of MAOC and fine fraction mass is strongly context
specific, that is dependent on factors such as mineralogy, sampling
depth or land use types (Beare et al.,

; Georgiou et al.,

). It

might also raise the question of whether an upper limit set by min-
eral surface area for SOC stabilisation actually exists at all, which
has however not been investigated systematically. Furthermore, re-
cent studies have suggested that OC accrual in MAOC is decoupled
from the availability of ‘free’ mineral surfaces and that it might even
preferentially pile up on surfaces already loaded with OC (Schweizer
et al.,

; Vogel et al.,

). In that sense, evaluating and ex-

plaining the content of C in the fine fraction (g C kg

−1

fine fraction)

rather than the content of MAOC (g MAOC kg

−1

bulk soil) might be

key to a better understanding of MAOC formation and its potential
constraints.

Another, and much more direct approach to test the saturation

concept or detect a potential MAOC saturation limit of a soil is to
add different rates of OC and investigate the increase in MAOC as a
function of C inputs (Abramoff et al.,

; Feng, Plante, et al.,

Feng, Xu, et al.,

; Six et al.,

; Stewart et al.,

). This ap-

proach has the disadvantages that (i) such experiments require high
amounts of organic amendments until saturation might be reached
and (ii) MAOC formation is a slow and steady process so that such
experiments need to be run for many years. However, the concept
that soils might have a certain universal capacity to store carbon as
MAOC and any excess carbon will be stored as POC (Carter,

2002

)

can also be tested across various soils, by investigating the MAOC–
SOC relationship. A recent study (Cotrufo et al.,

) on European

forest and grassland soils from the Land Use/Land Cover Area Frame
Survey (LUCAS) database predicted that soils cannot generally store

more than 45 g C kg

−1

bulk soils as MAOC (fraction

<

53

μ

m) and that

above a total SOC content of 50 g kg

−1

only the relatively labile POC

can accumulate further (original data plotted in

Figure

S1

). This sug-

gested upper limit contradicts findings of other authors, who found

MAOC values of up to 100 g MAOC kg

−1

bulk soil (Beare et al.,

;

Georgiou et al.,

). The distinct upper limit shown by Cotrufo

et al. (

) might be attributed to the scarcity of soils with SOC

contents higher than 50 g kg

−1

in their specific dataset, which led to a

strong prediction bias of the machine learning model. It is highly un-
likely that the same inflection point would occur for all soils, regard-
less of soil texture, mineralogy and potentially other soil properties.

Here, we used a comprehensive national soil archive to deter-

mine (i) whether an upper limit of MAOC storage in temperate ag-
ricultural soils can be detected or not and (ii) which factors explain
the OC content in the fine fraction. To do so, a systematic random
sample (

n

=

189) with a wide range of SOC (5– 118 g kg

−1

) and clay

content (30– 770 g kg

−1

) was taken from a total of 3104 topsoil sam-

ples from the German Agricultural Soil Inventory. In contrast to the

natural distribution of SOC in soils, an equal number of samples was

selected in each SOC content class in 10 g kg

−1

intervals, ensuring

an unbiased evaluation of the existence of a potential upper limit of
MAOC in temperate agricultural soils.

2

|

MATERIALS AND METHODS

2.1

|

Selection of soil samples

Soil samples, sieved to 2 mm, were selected from the archive of the
first German Agricultural Soil Inventory where soil sampling was per

-

formed in a fixed grid of 8

×

8 km grid covering the whole agricultural

area of Germany. Agricultural soils comprise croplands, permanent

grasslands (defined as sites with more than five consecutive years of
grassland use) and permanent crops (vineyards and orchards). Out

of a total of 3104 topsoil samples (0– 10 cm), 189 soils were selected
with a wide range of SOC (5– 118 g kg

−1

) and clay content (3%– 77%)

by applying a systematic random approach. After excluding soils with

a C:N ratio

>

13 as a clear sign of heath or peatland history or similar

reasons for large amounts of recalcitrant plant material (including
plaggen soils; Poeplau et al.,

; Springob & Kirchmann,

2010

; Vos

et al.,

), an equal number of samples were collected in each SOC

content class in 10 g kg

−1

intervals (

Figure

1

). Up to a SOC content

of 80 g kg

−1

, 20 samples were selected per content class (

n

=

160).

Within each SOC content class, the clay content distribution of the
randomly selected 20 samples was visually inspected and compared
with the clay content distribution of all samples within that content
class. This was done to ensure a selection of samples that represent
the full range of clay content within each SOC content class. Random
selection per SOC content class was repeated up to three times.

Above 80 g kg

−1

of SOC, only 29 samples were available, which were

all included without further selection. Ultimately, 71 cropland and
118 grasslands soils were selected. The SOC content range in crop-

lands was comparatively lower, ranging from 5 to 79 g kg

−1

, in con-

trast to grasslands which exhibited a broader range (28– 118 g kg

−1

).

13652486, 2023, 16, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/gcb.16804 by Hayley Peter-Contesse - Oregon State University , Wiley Online Library on [19/10/2026]. 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

4664

|

BEGILL

et

al

.

2.2

|

SOC fractionation and analysis

Physical fractionation was conducted to derive MAOC and POC

using a 50

μ

m sieve. In brief, 10 g of soil was immersed in 150 mL of

distilled water and ultrasonic dispersion (100 J/mL) was performed

according to Just et al. (

), followed by wet sieving. For wet siev-

ing, a dispersed soil sample was placed on top of the sieve and then a
continuous flush of deionised water was applied. The coarse fraction
was collected as the fraction that remained on the sieve (

>

50

μ

m)

and was oven- dried at 60°C. However, the fraction that passed
through the sieve (

<

50

μ

m) was treated with 0.8 g CaCl

2

L

−1

to obtain

the maximum fine fraction (

<

50

μ

m) from the soil– water solution,

followed by centrifugation for 15 min at 3800 g and drying at 60°C.

The fractionated soil samples were then milled to obtain an homoge-
neous sample for analysis. SOC and total nitrogen (N) were analysed
by dry combustion using an elemental analyser (LECO TRUMAC and
RC612). POC refers to SOC in the coarse fraction, while MAOC re-
fers to SOC in the fine fraction. Average C recovery was 95%. When
recovery was less than 85% or more than 100%, the bulk soil sample
was reanalysed. Thus, a few samples changed the SOC content class
from which they had initially been selected.

2.3

|

Further soil properties

Soil pH was determined using a soil- to- water ratio of 1:5 (5 g of soil
with 25 g of water). The pH value was measured using a pH meter

(ProLab 4000, SI Analytics) in combination with a pH combination
electrode (Blue Line 28 pH, SI Analytics). Soil texture was deter-
mined based on sieving and sedimentation of suspended fine par-
ticles classifying clay

<

2 mm, silt 2– 63 mm and sand

>

63 mm but

<

2000 mm. To determine pedogenic metal oxides (Fe, Al, Mn) in

the fine fraction (

<

50

μ

m), 1 g of dried and sieved soil sample was

shaken with 50 mL of 0.2 M ammonium oxalate solution for 4 h under

protection against UV radiation. The mixture was then centrifuged

for 10– 15 min. Therefore, the easily soluble, poorly crystalline oxides

and hydroxide compounds of iron and aluminium were dissolved by
complexation and brought into solution. Fe and Al oxides were then
determined via inductively coupled plasma optical emission spec-
troscopy (iCap7000, Thermo Scientific) in this extract. At the same
time, the plant- available proportions of phosphorus and sulphur
were also mobilised and analysed in the same extract (Batjes,

2012

;

Utermann et al.,

2000

).

2.4

|

Carbon input data

Data on average annual C inputs, including total C inputs, organic
fertiliser inputs, as well as aboveground and belowground C in-
puts, were derived from Jacobs et al. (

), in which the farmers'

questionnaire of the German Agricultural Soil Inventory and a crop-

specific C allocation coefficient framework were used to estimate
C inputs.

2.5

|

Statistical analyses

All data visualisation as well as data analysis was performed using
R version 4.1.1 (R Core Team,

). Linear regressions were fit-

ted to describe the relationships between total SOC and MAOC.

Generalised linear models were used to explain the variation in SOC

content of the fine fraction (

<

50

μ

m). Due to the large number of po-

tential explanatory variables, multi- model inference was applied for
a priori model selection based on the Akaike information criterion
using the MuMIn package (Barton & Barton,

2015

).

3

|

RESULTS AND DISCUSSION

3.1

|

Linear relationship between SOC and MAOC

For both cropland and grassland soils, linear relationships between
MAOC and SOC were observed up to the highest investigated SOC

content of 118 g kg

−1

(

Figure

). Particularly for croplands, a very close

positive relationship was observed across the whole range of SOC
and clay contents (

y

=

0.86

x

,

R

2

=

.99). This indicates that irrespective

of soil properties and total SOC content, intensively managed tem-
perate cropland soils store about 86% of their total SOC as MAOC
and 14% as POC. This is well in line with the 83% observed for a
wide range of cropland, grassland and forest soils in a recent meta-
analysis (Matus,

). Grassland soils had a higher scatter and thus

more variability in the fraction of MAOC and POC. They had also a
slightly lower slope than cropland soils with 72% of SOC being stored
as MAOC and 18% as POC (

y

=

0.72

x

,

R

2

=

.77). As indicated by the

size of the bubbles (

Figure

), the scatter in the grassland data was

partly related to fine fraction content, with the lowest MAOC/SOC
ratios observed for coarse- textured grassland soils (

Figure

). To test

F I G U R E 1

Distribution of soil organic carbon (SOC) contents

at 0– 10 cm depth of all samples in the German Agricultural Soil

Inventory (BZW- LW) and the selected samples.

13652486, 2023, 16, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/gcb.16804 by Hayley Peter-Contesse - Oregon State University , Wiley Online Library on [19/10/2026]. 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

|

4665

BEGILL

et

al

.

for potential texture- specific inflection points, the MAOC/SOC ratio
was further investigated in the fine fraction content classes

<

300,

300– 600 and

>

600 g kg

−1

(

Figure

3

), which revealed linear relation-

ships in all cases. Indeed, the proportion of MAOC increased with
fine fraction content, indicating that a relatively higher proportion
of SOC is stored as POC in coarser- textured soils. The mechanisms
behind these relationships are not entirely clear, but might be related
to texture- driven differences in POC turnover time, or a lower ef-
ficiency of converting POC to MAOC in more coarse- textured soils
(Poeplau et al.,

2015

). Certainly, it cannot be explained by saturation

of the fine fraction: the three soils with the highest SOC and MAOC

contents had an average fine fraction SOC content of 172 g kg

−1

in the

low fine fraction class, of 136 in the medium fine fraction class and of
89 in the high fine fraction class, indicating a much higher C loading in
the fine fraction of the coarser- textured soils.

In fact, in contrast to Cotrufo et al. (

), no inflection point and

thus no upper limit of MAOC storage was detected in the present
dataset (

Figures

and

3

). The simplest explanation for the very clear

inflection point in the previous study (Cotrufo et al.,

) might be

that the authors did not select a sample set that allowed a predic-

tion of soils with more than 45 g kg

−1

MAOC. None of the soils in

the sample set that was fractionated and analysed contained more

than 47 g kg

−1

MAOC (Cotrufo et al.,

). Additionally, (supervised)

machine- learning models are generally optimised in a way that the
values within the central tendency of the target distribution greatly
impact their performance (Ribeiro & Moniz,

). Thus, extreme

values are normally underrepresented in their predictions. In other
words, the natural distribution of SOC on a national to global scale
is strongly right skewed, with high SOC contents being rather rare
exemptions. A random sample selection may thus not be suitable for
answering the question if MAOC formation generally has an upper
limit. Our dataset included soils of extremely high SOC contents.
Such high amounts of MAOC will certainly not be achievable under
every condition, especially not under average agricultural manage-
ment regimes. For example, many of the soils with high SOC con-

tents studied here were Gleysols with high groundwater level and
the vast majority of the soils with more than 60 g C kg

−1

were used

as permanent grasslands. However, the clear linear relationship as
well as the fact that the average MAOC proportion was very similar
to the one found in the global meta- analysis of Matus (

) raises

the question if and by which factors soils should be limited to accrual
even higher than observed MAOC contents. Apart from one excep-
tion, the C:N ratio of the observed MAOC fractions was well con-
strained across all soils (

Figure

5

) and in the typical value range for

MAOC (10.5

±

1.2; Hassink,

1994

). We can thus exclude the possi-

bility of artefacts, for example, contaminations of MAOC with large
amounts of fine POC, charcoal particles or similar.

3.2

|

What drives the content of OC in the fine

fraction?

The amount of fine particles, and with that the total mineral sur-
face area, has been central to the theory of MAOC saturation
(Hassink,

). Indeed, in the present study too, the content of fine

fraction was significantly correlated with total MAOC content in
croplands and grasslands (

Figure

4a,b

). However, fine fraction con-

tent explained only 21% (cropland;

y

=

0.02

x

+

4.9,

R

2

=

.21) and 19%

(grassland;

y

=

0.03

x

+

26,

R

2

=

.19) of the variability in MAOC across

the 189 sites. The positive correlation is expected, because MAOC
is calculated by the amount of fine fraction and the C content of the

F I G U R E 3

Relationship between

soil organic carbon (SOC) and mineral-
associated organic carbon (MAOC)
(

<

50

μ

m) for cropland and grassland

soils grouped by fine fraction content;

<

300 g kg

−1

(

n

=

21; a) 300– 600 g kg

−1

(

n

=

58; b)

>

600 g kg

−1

(

n

=

110; c).

F I G U R E 2

Relationship between soil organic carbon (SOC) and

mineral- associated organic carbon (MAOC) for cropland (

n

=

71)

and grassland (

n

=

118) soils. The size of the bubbles is directly

proportional to the fine fraction content of the soils.

13652486, 2023, 16, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/gcb.16804 by Hayley Peter-Contesse - Oregon State University , Wiley Online Library on [19/10/2026]. 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

$Figures and page graphics, page 5$

4666

|

BEGILL

et

al

.

fine fraction. As observed in this and previous studies, the content
of fine fraction can vary from close to zero to close to 100% (Beare
et al.,

), which is not the case for the carbon content in the fine

fraction, which ranged from 9 to 260 g kg

−1

in this study (

Figure

5

).

Thus, even when a soil is strongly C depleted, it can have a relatively
high MAOC content just because it is fine textured. Furthermore,

Figure

4

indicates that the upper limit of this relationship, which is

often assumed to be the saturation limit, was not well constrained
by our observations, especially for the grassland soils. Also soils with
low fine fraction contents could have relatively high MAOC contents
(

Figure

4b

), which can also be seen in the datasets of previous stud-

ies (Beare et al.,

). Fitting an upper linear interval to such ob-

servations might thus not lead to the ultimate upper limit of MAOC
storage, although it can be helpful to understand interactive effects
of fine fraction content and other factors, such as mineralogy or land

use, on the observed MAOC contents of soils (Georgiou et al.,

).

However, what is often overlooked in studies investigating

the MAOC storage capacity is the OC content of the fine fraction
(

<

50

μ

m), which might help to understand the actual physicochemical

capacity of the clay and silt particles to bind and stabilise C. In this
study, the OC content in the fine fraction was highly variable, ranging

from 9 to 260 g kg

−1

fine fraction (

Figure

5

). The silt and clay fraction

can thus become highly enriched in OC. Agricultural management
alone cannot fully explain such variation. Instead, this C content is
influenced by mineralogy, hydrology, pedogenesis, land- use history
and potentially other factors (Rasmussen et al.,

; Schneider

et al.,

). Therefore, multiple regression models were applied to

elucidate the most important factors influencing the content of OC
in the fine fraction. The models could explain up to 67% of the vari-
ability, with important explanatory variables being the annual carbon
inputs to the soil via crop residues, root litter and organic amend-
ments as well as pedogenic oxides, groundwater level, amount of fine
fraction, soil pH (ranges from 4– 8) and C:N ratio (8– 12.8;

Table

S1

).

Interestingly, the fine fraction content was negatively correlated

with the C content in the fine fraction in both models (

Table

1

). This

was particularly true for grassland soils, as depicted in

Figure

4d

, and

indicates a certain concentration, or piling up of OC in the fine frac-
tion when the binding sites are limited. For a limited range of soils
with similar SOC but variable clay contents, Schweizer et al. (

)

found a similar negative relationship between clay content and C
stock as well as its proportion in the fraction

<

6.3

μ

m in temperate

cropland soils. NanoSims imaging revealed that MAOC was associ-
ated with a similar extent of surface across the whole clay gradient,
pointing towards organic matter piling up on a relatively low surface
area (Schweizer et al.,

). This implies that MAOC can form and

grow even when mineral surfaces are limited, which is in contrast to
the saturation concept, predicting that no additional MAOC can form
once all mineral surfaces are covered. Metaphorically spoken, MAOC
in soils might either resemble a mega city with a lot of skyscrapers
on a small island (high OC content piling up in the fine fraction of
a coarse- textured soils), or, when space is not limiting, many small
distant settlements with an overall low population density (low OC
content in the fine fraction of a fine- textured soil). Another NanoSims
study actually confirms this settlement analogy, since it showed that
MAOC was not homogeneously distributed across the available min-
eral surfaces but was found in organo- mineral clusters. Interestingly,
newly added carbon was mainly recovered at those clusters (Vogel
et al.,

). This, together with the fact that the proportion of SOC

stored as MAOC was found to be surprisingly stable across the whole
range of investigated soils, challenges the assumption that every soil

F I G U R E 4

Relationship between mineral- associated organic

carbon (MAOC) and fine fraction content (a, b) and between
organic carbon (OC) in the fine fraction and the fine fraction
content (c, d).

F I G U R E 5

Fine fraction organic carbon (C) and total nitrogen (N)

contents of the selected samples. The size of the bubbles is directly
proportional to the fine fraction content of the soils.

13652486, 2023, 16, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/gcb.16804 by Hayley Peter-Contesse - Oregon State University , Wiley Online Library on [19/10/2026]. 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

|

4667

BEGILL

et

al

.

has, determined by the amount of silt and clay particles, a certain

maximum capacity to store MAOC (Georgiou et al.,

). At least for

temperate agricultural soils, in which high activity clays prevail, and

at least for SOC contents of up to 118 g C kg

−1

, the ‘saturation deficit’

might not be a helpful concept to judge upon the ability to further
build up the MAOC pool of a specific soil.

For instance, if megacities with skyscrapers can be built on

small islands, why should they not be able to last on larger land
surfaces? In the end, it might be primarily a question of material
to build up such skyscrapers, thus biomass to build up MAOC. If

more than 100 g kg

−1

of carbon in the fine fraction is possible in

rather coarse- textured soils, why should it not be possible to en-
rich the silt and clay fraction of fine- textured soils in a similar way?
At the same time, the question comes up why such an enrichment
is rarely observable in nature. A clayey soil with a hypothetical

fine fraction OC content of 100 g kg

−1

fine fraction, 80% of fine

fraction, an average MAOC proportion of 79% (as found in this

study) and a hypothetical bulk density of 1.3 g cm

−3

, would lead to

a SOC stock of 394 Mg ha

−1

and a MAOC stock of 312 Mg ha

−1

in

0– 30 cm. This is a multiple of SOC of what is usually observed in

topsoil across the globe, so along the lines of the saturation con-

cept all mineral soils in the world are ‘undersaturated’ (Georgiou

et al.,

). However, our data indicate that mineral surfaces are

not the major limitations of SOC stabilisation. Certainly, an im-
portant limitation for MAOC formation is the availability of bio-
mass carbon (Janzen et al.,

). Substrate quality, which can

affect microbial carbon use efficiency and priming of native SOC,
might also play an important role for the rate of MAOC formation

(Guenet et al.,

; Ridgeway et al.,

). On the other hand,

it might also be that the degree of stabilisation of the compos-
ite fraction, that is commonly isolated as MAOC, is not as homo-
geneous throughout this fraction as conceptualised (Ridgeway
et al.,

). For example, if MAOC is piling up in organo- mineral

clusters, there might be gradients in decomposability within each
pile (Kleber et al.,

2007

; Schweizer,

https://orcid.org/0000-0003-3229-5600

) with organo– organo

interactions being less stable than direct organo– mineral inter-
actions. Such gradients, or growth conditions of organo– mineral

clusters might again strongly depend on abiotic soil properties,
such as mineralogy, pH or oxygen supply. The findings of our
study call for further mechanistic investigations to better under-
stand the potential constraints of MAOC formation. One import-
ant unsolved question evolving from our study is, whether the
C content in the fine fraction limits further MAOC formation or
not. Yet, we conclude that the surprisingly constant proportion
of MAOC in a large variety of soils and up to very high bulk SOC
contents questions if the saturation concept can help to estimate
the SOC sequestration potential of soils.

AUTHOR CONTRIBUTIONS

Christopher Poeplau and Axel Don designed the study,
Christopher Poeplau selected the samples, Neha Begill carried
out the fractionation work and data analysis, Neha Begill and
Christopher Poeplau wrote the manuscript, which was reviewed
and edited by Axel Don.

ACKNOWLEDGMENTS

We are grateful to everyone involved in the German Agricultural
Soil Inventory, which was funded by the German Federal Ministry

of Food and Agriculture. This particular study has received funding
from the European Unions' Horizon 2020 research and innovation
programme under grant agreement no. 862695 EJP SOIL. Open
Access funding enabled and organized by Projekt DEAL.

CONFLIC T OF INTEREST STATEMENT

The authors declare to have no competing interests.

DATA AVAIL ABILIT Y STATEMENT

The data of this manuscript are available at

https://doi.org/10.5281/

zenodo.7966076

.

ORCID

Neha Begill

Axel Don

https://orcid.org/0000-0001-7046-3332

Christopher Poeplau

https://orcid.org/0000-0003-3108-8810

TA B L E 1

Best models for explaining soil organic carbon (SOC; g kg

−1

bulk soil) in the fine fraction (

<

50

μ

m) with multiple explanatory

variables, including total inorganic carbon (TIC; g kg

−1

), carbon inputs, which includes

C

fert

(average annual carbon inputs from organic

fertilisers in Mg C ha

−1

),

C

roots

(average belowground organic carbon inputs in Mg C ha

−1

(Jacobs et al.,