Feeney et al. (2024). Benchmarking soil organic carbon (SOC) concentration provides more robust soil health assessment than the SOC/clay ratio at European scale

www.elsevier.com/locate/scitotenv

Benchmarking soil organic carbon (SOC) concentration provides more

robust soil health assessment than the SOC/clay ratio at European scale

Christopher J. Feeney

Bridget A. Emmett

UK Centre for Ecology and Hydrology, Environment Centre Wales, Deiniol Road, Bangor, Gwynedd LL57 2UW, UK

b

School of Agriculture, Forestry, Food and Environmental Sciences, University of Basilicata, 85100 Potenza, Italy

c

European Commission

–

Joint Research Centre, Via Enrico Fermi, 2749, 21027 Ispra, VA, Italy

H I G H L I G H T S

G R A P H I C A L A B S T R A C T

•

Developed the O/T SOC indicator ac-

counting for climate, land cover and soil

texture

•

We model 12 pedo-climate zones across

Europe with associated typical SOC

values.

•

O/T SOC is less sensitive to soil texture,

land-use and climate than SOC/clay.

•

O/T SOC better reflects differing

porosity and SOC stock grades than

SOC/clay does.

•

O/T SOC is sensitive to changing SOC

under land management over years to

decades.

A R T I C L E I N F O

Editor: Fang Wang

Keywords:

Soil carbon

Clay

Pedo-climate zones

Soil structure

Soil monitoring

Sustainable Development Goals

A B S T R A C T

Increasing soil organic carbon (SOC) confers benefits to soil health, biodiversity, underpins carbon sequestration

and ameliorates land degradation. One recommendation is to increase SOC such that the SOC to clay ratio (SOC/

clay) exceeds 1/13, yet normalising SOC levels based on clay alone gives misleading indications of soil structure

and the potential to store additional carbon. Building on work by Poeplau

&

Don (2023) to benchmark observed

against predicted SOC, we advance an alternative indicator: the ratio between observed and

“

typical

”

SOC (O/T

SOC) for pan-European application. Here,

“

typical

”

SOC is the average concentration in different pedo-climate

zones, PCZs (which, unlike existing SOC indicators, incorporate land cover and climate, alongside soil texture)

across Europe, determined from mineral (

<

20 % organic matter) topsoils (0

–

20 cm) sampled during 2009

–

2018

in LUCAS, Europe

’

s largest soil monitoring scheme (

n

=

19,855). Regression tree modelling derived 12 PCZs,

with typical SOC values ranging 5.99

–

39.65 g kg

−

1

. New index classes for comparison with SOC/clay grades

were established from the quartiles of each PCZ

’

s O/T SOC distribution; these were termed:

“

Low

”

(below the

25th percentile),

“

Intermediate

”

(between the 25th and 50th percentiles),

“

High

”

(between the 50th and 75th

percentiles), and

“

Very high

”

(above the 75th percentile). Compared with SOC/clay, O/T SOC was less sensitive

to clay content, land cover, and climate, less geographically skewed, and better reflected differences in soil

porosity and SOC stock, supporting 2 EU Soil Health Mission objectives (consolidating SOC stocks; improving soil

* Corresponding author.

E-mail address:

chrfee@ceh.ac.uk

(C.J. Feeney).

Contents lists available at

ScienceDirect

Science of the Total Environment

journal homepage:

https://doi.org/10.1016/j.scitotenv.2024.175642

Received 10 March 2024; Received in revised form 19 July 2024; Accepted 17 August 2024

Science of the Total Environment 951 (2024) 175642

2

structure for crops and biota). These patterns held for 2 independent datasets, and O/T SOC grades were sensitive

enough to reflect land management differences across several long-term field experiments. O/T SOC used in

conjunction with several other physical, chemical and biological soil health indicators can help support the EU

Soil Monitoring Law and achieve several United Nations Sustainable Development Goals.

1. Introduction

Land degradation poses one of the greatest environmental challenges

to humanity, with threats to

“

soil health

”

(defined by the EU Soil Health

Mission as the physical, chemical and biological condition of the soil,

determining its continued capacity to function as a vital living system

and support ecosystem services (

Veerman et al., 2020

)) representing the

bulk of major land degradation pathways (

). The EU Soil

Observatory (EUSO) has identified

>

60 % of soils across Europe as

“

unhealthy

”

, with at least one land degradation process occurring in

affected areas (

Panagos et al., 2024

). In response to this, the European

Union (EU) proposed a Soil Monitoring Law in July 2023 to provide a

legal framework to achieve healthy soils by 2050 and ensure their long-

term sustainable use (

Directorate-General for Environment, 2023

). More

broadly, improving soil health is regarded by the EU Mission Board for

Soil Health and Food, among others, as essential for achieving several

United Nations Sustainable Development Goals (SDGs)

–

notably SDGs 2

(zero hunger), 6 (clean water and sanitation), 13 (climate action) and 15

(life on land) (

Bonfante et al., 2020

;

Bouma, 2014

;

Keesstra et al., 2018,

;

).

Given the myriad benefits soil organic carbon (SOC) confers upon

soil, particularly aggregation, which in turn promotes water retention

capacity and fluxes of water, air and gas, reduces risks of compaction,

sealing and erosion, and through creation of more stable aggregates,

promotes further long-term SOC storage (

Blanco-Canqui et al., 2013

), it

is widely regarded as one of the most important metrics for indicating

soil health (

Kopittke et al., 2022

). Additionally, soils store the most

organic carbon in the terrestrial biosphere (

Lal, 2004

;

Schmidt et al.,

2011

) and it is estimated that a third of soils globally are losing SOC

rapidly (

FAO and ITPS, 2015

). As a result, there have been growing calls

to promote not just

“

climate-smart soils

”

through initiatives such as

“

4per1000

”

to promote SOC storage for climate change mitigation, but

also

“

soil-smart agriculture

”

to promote sustainable land management

for the betterment of soils generally (

Moinet et al., 2023

). Because of

these concerns, soil properties including topsoil SOC concentration are

commonly monitored in national field surveys (

Reynolds et al., 2013

) to

assess status and trends. A persistent challenge however is deriving an

optimal or target SOC content for policymakers. Topsoil SOC content is

sensitive to several environmental and management factors and may

thus have individual ranges that are the product of site-specific config-

urations of land cover and inherent conditions such as soil type and

climate that are beyond the realm of manageable intervention.

Furthermore, the particular ecosystem services of interest will be an

important determining factor in defining thresholds or optimal levels of

SOC.

It has been established that intimate association with fine silt and

clay particles is a key storage mechanism for SOC (

;

;

). Furthermore, it has been posited that the physical

behaviour of mineral soils is controlled by the amount of complexed

organic carbon with clay (COCC) (

Dexter et al., 2008

). These observa-

tions have spurred the development of a SOC content metric that is

normalised by the relative clay content of the soil: the SOC/clay ratio.

Analysis of Swiss arable soils yielded 3 SOC/clay ratio thresholds: 1/13,

1/10 and 1/8, marking boundaries between the index classes,

“

Degraded

”

,

“

Moderate

”

,

“

Good

”

and

“

Very good

”

, respectively

(

Johannes et al., 2017

). These index classes were found to reflect a

gradient of soil structural quality scores, with recent analysis from En-

gland and Wales appearing to validate their wider applicability within

Europe (

Prout et al., 2021, 2022

). This has led to the inclusion of the

SOC/clay ratio as a recommended indicator to monitor SOC concen-

trations in the proposed EU Soil Monitoring Law (

Directorate-General

for Environment, 2023

). However, the suitability of the SOC/clay ratio

as a soil health indicator has been challenged. Analysis of German

agricultural soils revealed that soils with relatively low clay (

<

20 %)

were disproportionately likely to have a

“

Very good

”

SOC/clay ratio,

despite showing clear potential to store significantly more carbon, while

clay-rich soils (

>

60 %) were overwhelmingly likely to be classed as

“

Degraded

”

, even where other metrics would imply adequate soil

structure or fertility (

). Additionally, in-

consistencies have been shown between the soil carbon stock changes

reported by national greenhouse gas inventories and the proportions of

degraded soils according to SOC/clay across Europe (

), while

recommended an update to the SOC/

clay indicator that accounted for different soil types and climates.

As an alternative to the SOC/clay ratio,

developed a normalised SOC content metric based on the ratio of

observed SOC to the

“

expected

”

concentration. This

“

observed/expected

SOC ratio

”

was found to be less sensitive to clay content and more

adequately reflect soil porosity differences. However,

Poeplau and Don

(2023)

recommended that modifications are necessary for application at

European scale because they predicted SOC as a function of clay only,

and for soils from agricultural settings (i.e. no woodlands or semi-

natural grasslands) within a single country, which does not capture

the spectrum of climates across the continent (from the Boreal north to

the Mediterranean south). A new SOC-based soil health indicator ought

to apply across the breadth of possible mineral soils, from sandy soils

under arable land-use in relatively warm and dry regions (where SOC is

likely low), to clay-rich soils in semi-natural habitats in relatively cold

and wet regions (where SOC is likely high). Thus, the consideration of

such a high diversity of environmental conditions will likely require

multiple reference SOC content values (

;

Rabot

et al., 2024

The aim of this study was to improve the alternative normalised SOC

indicator proposed by

Panagos et al., 2012, 2022

for the assessment of soil

health for all mineral soils across Europe. The objectives included: (i)

deriving the

“

typical

”

SOC as a function of key environmental and

management controls on observed SOC concentrations across Europe;

(ii) defining index categories that usefully summarise differing grades of

topsoil porosity and SOC stocks; and (iii) demonstrate how the new in-

dicator is less sensitive to geography, clay content, land cover and

climate factors than SOC/clay and thus, more suitable for soil health

monitoring at large scale.

2. Methods

2.1. Study area and selected datasets

The analysis used EU-wide soil survey data from the Land Use/Cover

Area Frame Survey (LUCAS) (

Orgiazzi et al., 2018

;

Soil property data, including soil organic carbon (SOC) concentration,

texture and land cover from the first three surveys (2009/12, 2015 and

2018) are publicly available from the European Soil Data Centre

(

) and were selected for data analysis. Soil

bulk density data for a subset of the 2018 LUCAS survey sites (

n

=

6059)

are also available (

Pacini et al., 2023

) and used in this study. All details

of data collection in the field, laboratory analysis and quality assurance

procedures can be found in the associated technical documentation

(

Fernandes-Ugalde et al., 2022

;

;

C.J. Feeney et al.

Science of the Total Environment 951 (2024) 175642

3

For the purposes of our analysis, we used an aggregated dataset that

combined data from all 3 extant LUCAS surveys, which had been pre-

pared for a recent analysis of SOC content changes across Europe (

De

Rosa et al., 2024

). As part of the development of this aggregated dataset,

rigorous checks were undertaken to validate recorded land cover classes

against a database of 874,646 geo-tagged close-up photos with land

cover and plant species labels (

D

’

Andrimont et al., 2022

). Where SOC

concentration data were available for multiple years, an average value

was calculated. The dataset was also filtered to exclude sites with either

missing SOC or texture information, or a change in land cover class

between survey years. The dataset was then filtered further to include

just cropland, grassland and forest sites because other land cover classes

were judged to be too rare for use in our modelling. After further refining

our data to exclude organic soils (

≥

120 g kg

−

1

SOC), we were left with a

total of 19,855 data points. Cropland and the Continental biogeo-

graphical region represent the dominant land cover and climate zone,

respectively in the LUCAS dataset (

Table 1

). More broadly, the pro-

portions of land cover classes and biogeographical regions in the LUCAS

dataset are generally similar to those categories that have been mapped

for the EU

+

UK land area (

Table 1

), demonstrating that the LUCAS data

closely reflects the environment across Europe.

2.2. Creating a new normalised soil carbon content metric for soil health

assessment

Before deriving a new soil carbon-based indicator of soil health, we

sought to establish whether the SOC/clay ratio exhibited a strong clay

content bias as had been previously identified for German agricultural

topsoils (

) and across Europe more widely

(

;

). The SOC/clay ratio was

calculated by dividing SOC by 10 to convert the units from g kg

−

1

to %

and then dividing by percentage clay content. Next, SOC/clay ratio was

plotted as a function of clay content to determine whether a clear bias

was observable (i.e. a decay curve from high SOC/clay ratios for sites

with minimal clay content to low SOC/clay ratios for sites much richer in

clay content, e.g.

>

35 % clay).

As an alternative to the SOC/clay ratio,

Jasiewicz and Stepinski, 2013

proposed the ratio between observed and

“

expected

”

SOC, with ex-

pected SOC estimated using a linear regression model to predict the SOC

concentration as a function of clay content. Developing a similar metric

for the LUCAS dataset required an alternative approach however, as the

broad range of land cover types and climate zones encompassed by

Europe means there is unlikely to be a single linear function of SOC and

clay content that would be universally applicable.

Here, we applied univariate regression tree modelling to stratify the

mineral soils from the LUCAS dataset into several discrete subsets

–

referred to herein as

“

pedo-climate zones

”

(PCZs)

–

each characterised

by a combination of key environmental predictors of SOC concentration.

Regression tree modelling was adopted here because it allows pre-

dictions to be generated from multiple potential predictors, including

categorical as well as continuous variables, and produces a transparent

decision tree which conveniently characterises discrete environments

(

Elith et al., 2008

). Potential predictor variables included elevation,

geomorphon landform class (

), land cover,

calcium carbonate content, coarse fragment content, biogeographical

region (as a proxy for climate) and soil type. Biogeographical regions

were assigned to the point locations from all LUCAS survey years

through an intersect operation performed in QGIS 3.32 [Lima] with a

polygon shapefile of the 2016 map of the biogeographical regions of

Europe (

European Environment Agency, 2017

). Elevation values from

the 30 m digital elevation model of Europe from EcoDataCube and

geomorphon land classes from a 90 m global map of landform classes

(

Amatulli et al., 2020

) were extracted to LUCAS point locations using the

“

Sample Raster Values

”

tool in QGIS. Soil types were defined by cate-

gorising the LUCAS observed mineral soils using the European Soil Map

(HYPRES) (

W

osten, 2000

) soil texture classification according to clay,

silt and sand content. Soil type classification was implemented with the

“

soiltexture

”

package in R (

Moeys et al., 2018

;

).

Once all predictor variables were assigned, a training dataset was

created by extracting a random sample of 80 % of the LUCAS dataset; the

remaining 20 % was kept as testing data. Univariate regression tree

modelling via recursive partitioning was implemented in R using the

“

rpart

”

package (

;

Therneau et al., 2022

) with the

training dataset as input and the natural logarithm of SOC concentration

as the response variable. SOC concentrations were log-transformed to

ensure values were approximately normally distributed because ANOVA

is used to compare the means of multiple groups in the

“

rpart()

”

function

(

Therneau et al., 2022

). Tree depth was controlled by optimising the

complexity parameter value via 10-fold cross-validation with the

“

caret

”

package in R (

Kuhn et al., 2023

;

), which ensured the

usual step of

“

pruning

”

decision tree sub-nodes to reduce model over-

fitting was unnecessary. Regression tree accuracy was assessed by

applying the model fitted to the training dataset to the testing dataset

and calculating the root mean square error (RMSE). A key disadvantage

of regression tree modelling is the tendency for a small change in the

dataset to induce a large change in the final tree (

Elith et al., 2008

Normally, this is overcome by fitting models to several unique training

samples of the original dataset and

“

bagging

”

(aggregating) the results.

However, for interpretive clarity, we wanted to produce a single deci-

sion tree that could be used to identify average or

“

typical

”

SOC con-

centrations based on identifiable combinations of environmental

variables. To overcome this issue, we modelled regression trees for 5000

unique permutations of training datasets and selected the one with the

lowest RMSE when evaluated against testing data. Additionally, the

number of times each complete branch to a

“

leaf

”

(terminal node) of the

selected regression tree occurred out of the 5000 models was calculated

and expressed as a percentage. Means and standard deviations were

recorded for each leaf and back transformed from logarithmic-scaled

SOC concentrations. Coefficients of variation were calculated for each

PCZ-level SOC distribution by dividing standard deviation by the mean

and multiplying by 100 to express as percentages. This was done to

illustrate the level of variability around the

“

typical

”

(mean) SOC value

in each PCZ.

Using the newly created decision tree, each of the LUCAS mineral soil

records was assigned a PCZ based on their specific combination of

relevant environmental characteristics. We then divided each observed

SOC concentration by the mean value predicted for the relevant PCZ to

derive our new normalised SOC metric: the observed/typical SOC ratio

(herein shortened to O/T SOC). There are 4 index classes established for

the SOC/clay ratio (

“

Degraded

”

,

“

Moderate

”

,

“

Good

”

and

“

Very good

”

,

separated by thresholds of 1/13, 1/10 and 1/8) (

Johannes et al., 2017

Table 1

Proportional make-up of land cover classes and biogeographical regions (BGRs)

in the LUCAS dataset. Land cover refers specifically to areas that have been the

same class for all 3 LUCAS surveys (2009/12, 2015

&

2018) whereas percentage

of EU

+

UK area represents means (covering CORINE land cover mapping years

2012

&

2018; no map exists for 2009 or 2015). Land cover percentages for

EU

+

UK area are calculated after removing other land cover classes (e.g. bogs,

urban areas, open water) to allow fair comparison against LUCAS.

Percentage of LUCAS

dataset

Percentage of EU

+

UK

area

Cropland land cover

52.7

43.6

Grassland land cover

24.4

16.5

Woodland land cover

22.9

40.0

Alpine BGR

4.4

8.7

Atlantic BGR

18.0

17.9

Boreal BGR

15.0

19.3

Continental BGR

33.6

29.5

Mediterranean BGR

24.2

20.7

Pannonian BGR

3.3

2.9

Other (Black Sea and Steppic)

BGRs

1.4

1.1

C.J. Feeney et al.

Science of the Total Environment 951 (2024) 175642

4

To compare our new O/T SOC indicator with the established SOC/clay

indicator, we calculated the 25th, 50th and 75th percentiles of O/T SOC

for each PCZ, to define index classes as follows:

“

Low

”

=

below the 25th

percentile;

“

Intermediate

”

=

between the 25th and 50th percentiles;

“

High

”

=

between the 50th and 75th percentiles;

“

Very high

”

=

above

the 75th percentile. Here, we opted to use descriptors like

“

low

”

and

“

high

”

because to judge a soil as

“

degraded

”

or

“

good

”

depends on

contexts that we do not capture in our methodology, such as which

ecosystem service(s)

–

as well as the balance of trade-offs and synergies

between these

–

that stakeholders wish to prioritise.

Classifications, based on the new O/T SOC metric were compared

with the established SOC/clay classes in two main ways. First, using the

LUCAS data, O/T SOC was plotted as a function of clay content to

determine whether a clear relationship with clay content exists. Second,

distributions of total soil porosity and SOC stocks from the 2018 LUCAS

survey were plotted for each of the SOC/clay and O/T SOC classes using

the

“

tidyverse

”

package in R (

;

Wickham, 2023

).

Effective soil porosity, which considers aggregation, connectivity be-

tween pores and pore morphology than total porosity, would likely be a

more suitable indicator of soil structure. However, it was not possible to

derive this metric from LUCAS data. Nevertheless, total soil porosity is

still an important factor influencing fluxes of water and gas through soils

(

Assouline and Or, 2013

), and microbial communities (

Or et al., 2007

),

and has been shown to be highly sensitive to changes in SOC with

environmental change (

Robinson et al., 2022

;

Thomas et al., 2024

). By

assessing total porosity and SOC stocks in relation to O/T SOC and SOC/

clay, we aim to show that O/T SOC index classes can more usefully

separate differing grades of physical structural quality and carbon

storage than would be the case for SOC/clay index classes.

Porosity (m

3

m

−

3

) was calculated following the approach of

Breun-

ing-Madsen et al. (2015)

:

Porosity

=

1

−

Bulk Density

Particle Density

(1)

Bulk density (g cm

−

3

) refers to the fine fraction of soil specifically

(

<

2 mm). For this, we used the corrected to fine earth (

<

2 mm) bulk

densities presented in

Pacini et al. (2023)

, and first removed extreme

bulk density values and any records associated with organic soils

(leaving

n

=

4642). Particle density (g cm

−

3

) was calculated as:

OM

pd

, Si

pd

and CaCO

3.pd

refer to the particle densities of organic

matter (1.30 g cm

−

3

), silicates (2.65 g cm

−

3

) and calcium carbonate

fractions (2.80 g cm

−

3

), respectively. OM

prop

, Si

prop

and CaCO

3.prop

refer

to the relative proportions of organic matter, silicates and calcium car-

bonate contents, respectively.

SOC stocks were calculated following the approach of

Poeplau et al.

(2017)

:

SOC St

o

ck

=

SOC

(

%

) ×

Bulk Density

×

Depth

×

(

1

−

Coarse

prop

)

(3)

Depth (cm) is set to 20 to align with the LUCAS survey design;

Coarse

prop

refers to the fraction by volume of coarse fragments in the soil

according to

Pacini et al. (2023)

. Differences in porosity and SOC stocks

between SOC level classes for each metric were tested using the Kruskall-

Wallis and the post hoc pairwise Wilcoxon tests, with significance

assessed at

α

=

0.05 for both tests.

As a further comparison of the 2 SOC-based soil health indicators,

percentages of observations with different SOC/clay and O/T SOC index

classes were summarised by HYPRES soil texture class, land cover and

biogeographical region. We also developed maps of the percentages of

LUCAS points in each NUTS 1 region with extreme SOC/clay and

extreme O/T SOC conditions. First, the number of LUCAS observations

in each NUTS 1 region was calculated using the

“

Count points in poly-

gons

”

tool in QGIS. This same step was repeated 4 more times, once for

each of the following conditions: (i) observations classed as

“

Degraded

”

SOC/clay, (ii) observations classed as

“

Very Good

”

SOC/clay, (iii) ob-

servations classed as

“

Low

”

O/T SOC, and (iv) observations classed as

“

Very High

”

O/T SOC. Counts from this second step were divided by the

total number of LUCAS observations in each NUTS 1 region, calculated

in step 1, and multiplied by 100 to express as percentages. Heavily

urbanised NUTS 1 regions in the UK, Belgium and Germany were

excluded from the analysis as these had very few or no LUCAS

observations.

2.3. Demonstrating the applicability of O/T SOC beyond the LUCAS

dataset

We used 2 independent datasets to evaluate the suitability of our O/T

SOC index classes for individual EU member states and regions to

benchmark their soil health status against comparable landscapes across

Europe. These 2 datasets include the Glastir Monitoring and Evaluation

Programme (GMEP) for Wales from UKCEH, and the Bodenzus-

tandserhebung (BZE) survey of agricultural soils for Germany from the

Thünen Institute. Both datasets were generated within the 2009

–

2018

timeframe of LUCAS and cover similar topsoil depths (0

–

15 cm for

GMEP; 0

–

30 cm for BZE) to LUCAS, so they should provide adequate

comparison. The 2 surveys also reflect contrasting land-use mixes, with

BZE predominantly covering cropland and GMEP covering pastures and

semi-natural grasslands and forests. Published location data for GMEP

and BZE datapoints were overlain with the biogeographical regions map

of Europe to extract climate zone information. These extracted data

were combined with recorded land cover and soil texture information in

the GMEP and BZE surveys to assign PCZ classes to all datapoints. Next,

O/T SOC ratios were calculated, porosities estimated according to Eqs.

(1) and (2)

, and the porosity distributions modelled for each index class.

Published location information in BZE and GMEP represent approximate

positions to protect landowner confidentiality. This may mean that a

small minority of BZE sites close to the Atlantic-Continental boundary

are assigned to an incorrect biogeographical region (and ultimately, PCZ

class). GMEP should be unaffected however, as all of Wales is covered by

the Atlantic biogeographical region. All data analyses and display items

were generated with R (4.3.0) (

) except where this is

stated otherwise.

To demonstrate the sensitivity of the O/T SOC indicator to changes in

land management, several datasets with data on SOC changes over time

under long-term field experiments (LTFEs) were sourced. These LTFEs

span multiple decades and represent semi-arid, temperate, and cold-

weather climates across Europe. Recorded SOC values were converted

to O/T SOC values based on the typical SOC value that corresponds to

each LTFE

’

s PCZ, determined from associated information on climate,

land cover and soil texture.

Particle density

=

OM

pd

×

Si

pd

×

CaCO

3

pd

(

S

i

pd

×

CaCO

3

pd

×

OM

prop

)

+

(

OM

pd

×

CaCO

3

pd

×

Si

prop

)

+

(

OM

pd

×

Si

pd

×

CaCO

3

prop

)

(2)

C.J. Feeney et al.

Science of the Total Environment 951 (2024) 175642

5

3. Results and discussion

3.1. Observed/typical SOC ratio as an alternative soil health indicator to

SOC/clay ratio

Plotting SOC/clay against clay content for mineral soils aggregated

across the LUCAS surveys from 2009/12, 2015 and 2018 revealed a

clear relationship between the 2 variables (Fig. S1a), comparable to that

found for German agricultural soils (

). This led to

coarse-textured soils (clay

<

18 %) being disproportionately classified as

likely to be

“

Very good

”

(SOC/clay

≥

1/8), while clay-rich soils (clay

>

35 %) were most likely to be

“

Degraded

”

(SOC/clay

<

1/13) (Fig. S1a).

This necessitated the development of a new normalised SOC content

ratio which would be less dependent on clay content (see

Methods

A total of 12 PCZs were defined from unique combinations of land

cover, biogeographical region (as a proxy for climate), and soil texture

–

the 3 variables determined to be significant predictors according to our

regression tree modelling (

). The PCZs are characterised by bi-

nary splits in land cover (Cropland vs. Grassland and Woodland) and soil

texture (Coarse vs all other texture categories according to the HYPRES

(

W

osten, 2000

) classification scheme), with biogeographical regions

generally clustered into variations of wet and cold vs warm and dry

groups. Two thirds of PCZs occur in

>

60 % of all 5000 modelled decision

trees, with the

“

Mediterranean arable sandy soils

”

and

“

Mediterranean

arable loamy and clayey soils

”

groups occurring 100 % of the time

(Fig. S2). This demonstrates a high level of overall confidence in our

delineation of PCZ classes.

“

Typical SOC

”

(predicted PCZ-level averages)

ranges from 5.99 to 39.65 g kg

−

1

, with higher SOC tending to occur for

combinations of semi-natural habitats, wetter and colder climates, and

more clay-rich soil textures (

), reflecting similar topsoil trends in

Germany and the UK (

;

). Coefficients of variation of SOC concentrations in each PCZ range

from 5.25 % (Cold climate semi-natural loamy and clayey soils) to 25.54

% (Mediterranean arable sandy soils), with most PCZs associated with a

coefficient of variation

<

10 %, suggesting that typical SOC contents are

generally well constrained (

). Dividing observed SOC concen-

tration by the PCZ-typical reference value yields a new

“

Observed/

Typical SOC

”

(O/T SOC) indicator. The O/T SOC thresholds, separating

the 4 index classes per PCZ are summarised in

and Fig. S3.

Plotting O/T SOC as a function of clay reveals no apparent sensitivity

to clay content in this new metric (Fig. S1b). This suggests that O/T SOC

should be applicable across mineral soils irrespective of their texture.

SOC is a key control on soil structure, which together can benefit soil

biogeochemical cycling and microbial activity (

Mattila, 2023

). There-

fore, a SOC-based indicator ought to be useful for differentiating be-

tween soils of differing structural condition. A comparison of the

porosity (as an indicator of soil structure) distributions reveals no dif-

ferences among the bottom 3 SOC/clay index classes (

Fig. 1

a). Only

when comparing the bottom 3 grades with the top class,

“

Very good

”

, is

there a statistically significant difference (

p

<

0.05; Table S1). By

contrast, porosity distributions appear to be much more clearly differ-

entiated by O/T SOC index classes (

Fig. 1

b), with statistically significant

differences (p

<

0.05) between all classes (Table S1).

Similarly, O/T SOC index classes appear to capture sharper contrasts

in SOC stocks (medians range from 22.38 to 62.54 t ha

−

1

) than their

SOC/clay counterparts (medians range from 26.83 to 52.59 t ha

−

1

; see

Fig. 1

c

&

d and Table S2). Additionally, there is no statistically signif-

icant difference (p

<

0.05 level) between the

“

Moderate

”

and

“

Good

”

SOC stock distributions for SOC/clay (Table S1). Together, these results

demonstrate that compared to the SOC/clay ratio, the O/T SOC ratio

Table 2

Pedo-climate zones (PCZs,

n

=

12) derived from univariate regression tree modelling of SOC concentration recorded in the LUCAS data. Columns 2

–

4 outline the

biogeographical region(s), land cover class(es) and soil texture class(es) that constitute each PCZ. Column 5 indicates the predicted mean, standard deviation and

coefficient of variation SOC concentration for each PCZ. Columns 6

–

8 indicate the O/T SOC index category thresholds derived from the LUCAS data.

Pedo-climate zones

derived from

regression tree

modelling

EEA Biogeographical

regions (2016

classification)

Land cover

classes

(LUCAS

scheme)

Soil texture

classes

(HYPRES

scheme)

Typical (mean) SOC, g kg

−

1

;

[standard deviation, g kg

−

1

];

{coefficient of variation, %}

Low/

intermediate

threshold

Intermediate/

high threshold

High/very

high

threshold

Mediterranean arable

sandy soils

Mediterranean

Cropland

Coarse

5.99 [1.53] {25.54}

0.68

0.96

1.53

Mediterranean arable

loamy and clayey

soils

Mediterranean

Cropland

Fine; Medium;

Medium fine;

Very fine

12.18 [1.71] {14.04}

0.71

0.99

1.38

Central arable sandy

soils

Black Sea; Continental;

Pannonian; Steppic

Cropland

Coarse

10.59 [1.75] {16.53}

0.76

0.93

1.28

Atlantic arable sandy

soils

Atlantic

Cropland

Coarse

19.11 [1.68] {8.79}

0.7

0.97

1.41

Atlantic and Central

arable loamy and

clayey soils

Atlantic; Black Sea;

Continental;

Pannonian; Steppic

Cropland

Fine; Medium;

Medium fine;

Very fine

16.61 [1.55] {9.33}

0.75

0.96

1.26

Alpine and Boreal

arable soils

Alpine; Boreal

Cropland

Coarse; Fine;

Medium;

Medium fine;

Very fine

23.34 [1.87] {8.01}

0.71

0.95

1.31

Southern semi-natural

sandy soils

Black Sea;

Mediterranean;

Pannonian; Steppic

Forest;

Grassland

Coarse

12.55 [1.94] {15.46}

0.58

0.96

1.57

Continental semi-

natural sandy soils

Continental

Forest;

Grassland

Coarse

19.11 [2.07] {10.83}

0.62

0.96

1.57

Southern semi-natural

loamy and clayey

soils

Black Sea;

Mediterranean;

Pannonian; Steppic

Forest;

Grassland

Fine; Medium;

Medium fine;

Very fine

23.81 [1.81] {7.6}

0.64

0.99

1.57

Continental semi-

natural loamy and

clayey soils

Continental

Forest;

Grassland

Fine; Medium;

Medium fine;

Very fine

29.96 [1.76] {5.87}

0.7

1.02

1.45

Cold climate semi-

natural sandy soils

Alpine; Atlantic;

Boreal

Forest;

Grassland

Coarse

29.96 [2.28] {7.61}

0.61

1.01

1.72

Cold climate semi-

natural loamy and

clayey soils

Alpine; Atlantic;

Boreal

Forest;

Grassland

Fine; Medium;

Medium fine;

Very fine

39.65 [2.08] {5.25}

0.69

1.02

1.5

C.J. Feeney et al.

Science of the Total Environment 951 (2024) 175642

6

should be: (i) less sensitive to clay content, and (ii) more appropriately

reflect gradients in soil structure and carbon stocks with increasing SOC

concentration. Thus, the O/T SOC indicator should be more effective for

monitoring progress to 2 of the EU

’

s Soil Health Mission objectives for

2030: conservation and enhancement of SOC stocks; improving soil

structure to support soil biodiversity and crops (

European Commission,

2023

).

LUCAS mineral soils classed as

“

coarse

”

under the HYPRES texture

classification are disproportionately categorised (

>

80 %) as having

“

very good

”

soil structure with SOC/clay

>

1/8, whereas similar pro-

portions of

“

fine

”

and

“

very fine

”

soils are described as having

“

degraded

”

soil structure with SOC/clay

<

1/13 (

Fig. 2

a). By contrast,

there are roughly equal proportions of all 4 O/T SOC index classes for

“

coarse

”

,

“

medium

”

and

“

medium fine

”

soils, with a slight skew towards

higher proportions of

“

high

”

and

“

very high

”

O/T SOC ratios for

“

fine

”

and

“

very fine

”

soils (

Fig. 2

b). This suggests that soil texture (which

cannot be readily modified by landowners) is not the prevailing factor

that determines soil health status according to O/T SOC. Furthermore, a

critical weakness of the SOC/clay indicator is that croplands are

disproportionately likely to be classed as

“

degraded

”

, while forests are

disproportionately likely to be classed as having

“

very good

”

soil

structure

–

trends which are clearly visible from LUCAS data (

Fig. 3

a).

The absence of any apparent bias in O/T SOC ratios by land cover class

(

Fig. 3

b), coupled with similar patterns among biogeographical regions

(Fig. S4), suggests that the O/T SOC ratio ought to capture soil health

differences more effectively within common environments. This means

Fig. 1.

Porosity distributions by (a) SOC/clay index and (b) O/T SOC index; SOC stocks by (c) SOC/clay index and (d) O/T SOC index. All plots were generated from

LUCAS 2018 data with available bulk density measurements. The 25th, 50th and 75th percentiles of all 16 boxplots are listed in Table S2). Numbers on the boxplots

refer to the number of LUCAS observations with bulk density measurements in each category.

C.J. Feeney et al.

Science of the Total Environment 951 (2024) 175642

7

that the soil health of arable land in the Mediterranean region (where

SOC is likely to be lower than Europe-wide average levels) can be

benchmarked more fairly against comparable landscapes; meanwhile,

the soil health of forest in cold climates (where SOC is likely to be

relatively high) can also be benchmarked against comparable land-

scapes as well.

3.2. Spatial patterns and further evaluation of the O/T SOC ratio across

Europe

Mapping the proportions of LUCAS samples per NUTS 1 region

highlights the degree to which the 2 extreme SOC/clay index classes

(

“

Degraded

”

and

“

Very Good

”

) are skewed across Europe spatially.

>

50

% of LUCAS points per NUTS 1 region are classed as having

“

Degraded

”

SOC/clay in most regions of southern Europe (

a), with 3 of the

most extreme examples found in Romania (Table S3). By contrast,

northern Europe is overwhelmingly represented by

“

Very Good

”

SOC/

clay observations (

b), with nearly 100 % of observations in Swe-

den

’

s

“

Norra Sverige

”

region classed as

“

Very Good

”

and

>

90 % of ob-

servations in 3 regions of the British Isles classed as

“

Very Good

”

(Table S3). These geographical patterns partly reflect the contrast be-

tween relatively warm and dry climate conditions in the south and

wetter and colder conditions farther north, which are not accounted for

in the established SOC/clay index classes. These patterns also seem to

reflect the distributions of clay, silt and sand mapped for Europe, with

the authors of that study attributing these patterns to the influence of

Late Glacial loess deposition (

Ballabio et al., 2016

Repeating the same mapping process for O/T SOC classes reveals no

clear geographical trends in either

“

Low

”

(

c) or

“

Very Good

”

(

d) O/T SOC condition. Compared to the SOC/clay maps, there are

far fewer regions where the number of observations with extreme O/T

SOC exceeds 50 %. Additionally, the 5 regions with the highest pro-

portions of

“

Low

”

O/T SOC levels have between 48.1 and 63 % of ob-

servations rated as

“

Low

”

, well below the equivalent range for

“

Degraded

”

SOC/clay: 72.7

–

81.5 % rated as

“

Degraded

”

(Table S3).

Similarly, the 5 regions most dominated by

“

Very High

”

O/T SOC ob-

servations have between 47.8 and 60 % of samples rated as

“

Very High

”

compared to the 91.8

–

97.3 % range for the top 5 regions with

“

Very

Good

”

SOC/clay (Table S3). These figures illustrate both the clustering

of soils with extreme SOC/clay ratios compared to extreme O/T SOC

ratios, as well as the narrowness between the

“

Degraded

”

and

“

Very

Good

”

SOC/clay categories compared to the gap between

“

Low

”

and

“

Very High

”

O/T SOC. Thus, incremental improvements in SOC storage

and soil structure should be much clearer to track over time and with

much less geographical sensitivity using the O/T SOC indicator than the

SOC/clay indicator.

As a further demonstration, data from national soil monitoring

programmes in Germany (BZE) and Wales (GMEP) were used to calcu-

late O/T SOC and model porosity distributions by index class in these 2

countries. Results presented in

Fig. 5

a illustrate that porosity increases

with O/T SOC and differences in porosity between index classes are all

statistically significant (

p

<

0.05 level) for both countries. Median

porosity increases from 0.45 m

3

m

−

3

in

“

Low

”

-rated soils to 0.52 m

3

m

−

3

in

“

Very high

”

-rated soils in Germany, while in Wales, the medians

range from 0.54 to 0.72 m

3

m

−

3

across the same index classes (

Fig. 5

a).

These results illustrate how porosity can continue to improve from

relatively SOC-poor, largely arable-dominated soils through to relatively

SOC-rich, predominantly semi-natural habitat covered soils (

Fig. 5

b).

These patterns also illustrate further the importance of benchmarking

soil health by appropriately defined landscape contexts, as what makes a

“

Very high

”

-rated soil in German agricultural landscapes would most

likely be classed among the

“

Low

”

-rated mineral soils occurring in

Wales. Importantly, we have demonstrated how O/T SOC status within

smaller European regions could be assessed against EU-wide data. This

potentially offers EU policymakers an indicator which can be used

within future

“

Soil Health Districts

”

(discrete geographical regions

within Europe for soil health monitoring and assessment) for soil health

reporting.

For a soil health indicator to be useful, it should be sensitive enough

to detect change over time, including responses to land management

interventions (

Bünemann et al., 2018

). Several long-term field experi-

ments (LTFEs), beginning as early as the 19th Century in some cases,

have been undertaken throughout Europe. For those LTFEs where SOC

concentration data are available, we can test the sensitivity of the O/T

SOC status to change over time under land management interventions.

Rothamsted Research contains a rich archive of long-term data from

LTFEs in the UK. The LTFEs of Broadbalk and Highfield (

) present

Fig. 2.

Percentages of soil types in the LUCAS mineral soils data per SOC/clay

index (a) and O/T SOC index (b).

Fig. 3.

Percentages of land cover classes in the LUCAS mineral soils data per

SOC/clay index (a) and O/T SOC index (b).

C.J. Feeney et al.

Science of the Total Environment 951 (2024) 175642

8

contrasting types of land management and trends in SOC contents over

time. As well as being agricultural plots from the same country, both

LTFEs consist of similar soil textures (~23

–

33 % clay), so share the same

PCZ class,

“

Atlantic and Central arable loamy and clayey soils

”

that al-

lows the 2 LTFEs to be more easily compared.

In the Broadbalk example (

Rothamsted Research, 2021

a), 6 fertiliser treatments (including

a control with no fertiliser application,

“

nil

”

) were trialled, with treat-

ments starting in 1843 (or later, if stated) with SOC concentration

measured at regular intervals (

). In this

example, the O/T SOC status can shift from

“

low

”

to

“

very high

”

O/T

SOC status over a few decades of farmyard manure (FYM) application.

Under FYM treatments, the O/T SOC status of the soil can improve in as

little as 9 years in the case of FYM since 1885 and decline in as little as 5

years when FYM applications cease, as in the case of FYM

+

N inputs

(

a). The Highfield bare fallow LTFE by contrast recorded the

decline of SOC content in multiple plots where long-term grassland

(since 1838) was ploughed up in 1959 and left as bare fallow and

sampled at regular intervals for multiple decades thereafter (

Poulton

et al., 2022

). Here, O/T SOC declined from

“

very high

”

to

“

high

”

status

across all 4 monitored plots in 12 years (1959

–

1971), with plot 2

declining to

“

intermediate

”

status in another 8 years after that (

b).

By 2000, all 4 plots had declined to

“

low

”

status and remained there for

the additional 14 years of the experiment (

b).

Changes in O/T SOC status are also visible in other environments

across Europe. The Ultuna LTFE in Sweden (

K

atterer et al., 2011

), rep-

resenting the

“

Alpine and Boreal arable soils

”

PCZ, shows that over 60

years, experimental plots under 8 different organic amendments

exhibited a shift from

“

low

”

O/T SOC to one of the higher O/T SOC

grades (Fig. S5).

Rothamsted Research, 2021

identified a similar result for

their observed/expected SOC ratio in the Ultuna LTFE and remarked

that by contrast, only 3 treatments produced a shift in SOC/clay status

from

“

degraded

”

(and in these cases, only up to

“

Moderate

”

status).

Monitoring of forest succession following abandonment of cropland in

northeast Spain (

Bell et al., 2021

) shows that after early-stage succes-

sion, O/T SOC reaches at least

“

high

”

status on average (Fig. S6).

3.3. Potential future applications and limitations

The derivation of an

“

expected

”

or

“

typical

”

SOC value depends

entirely on the population of data used for prediction (

Poeplau and Don,

2023

). In this sense, the higher and lower recorded SOC concentration

values do not necessarily indicate a good or poor-quality soil, but

instead, what might reflect typical or atypical conditions for a particular

environment, following previous benchmarking studies (

). This population-driven issue applies to the

O/T SOC indicator too, but the main advantages of the benchmarking

approach used here are that newly monitored soils can in future can be

assessed more fairly against landscapes with similar characteristics, and

that each of the derived O/T SOC thresholds separated differing grades

of porosity and SOC stocks across Europe much more clearly than

established SOC/clay thresholds did. Thus, O/T SOC index classes

should be more meaningful than SOC/clay index classes in describing

soil structure and below ground carbon stocks across Europe. Further-

more, a benchmarking approach offers much flexibility, especially for

metrics where larger values are generally better than lower values (e.g.

SOC concentration) or vice versa (e.g. in the case of bulk density). If a

landowner for instance, learns that their soil shows atypical conditions

against representative benchmarks (e.g. abnormally low SOC or abnor-

mally high bulk density for their environment), they can better under-

stand which aspects of soil health to improve most urgently.

Policymakers meanwhile can set objectives and appropriate timetables

(e.g. X percentage of soils with a

“

low

”

O/T SOC status should improve

by at least 1 O/T SOC grade by year, Y). Additionally, provided that

grades still reflect differences in soil condition (e.g. porosity),

>

4 O/T

Fig. 4.

Percentages of

“

Degraded

”

SOC/clay (a),

“

Very Good

”

SOC/clay (b),

“

Low

”

O/T SOC (c) and

“

Very High

”

O/T SOC (d) mapped across the EU

+

UK by NUTS 1

region. Percentages refer to proportions of LUCAS samples points per NUTS 1 region.

C.J. Feeney et al.

Science of the Total Environment 951 (2024) 175642

9

SOC grades could be derived to increase the sensitivity of O/T SOC to

detect change in response to land management influences. Determining

the maximum number of potential O/T SOC grades will require a more

thorough sensitivity analysis of soil monitoring datasets and is suggested

as an important direction for future research.

Developing a harmonised approach for the EU using the 3 currently

available LUCAS surveys (2009/12, 2015

&

2018) as a baseline dataset

is arguably important given that ultimately, LUCAS is the most

comprehensive and up to date dataset publicly available at European

scale. Additionally, there is ample scope to apply our approach to

generate PCZs and associated O/T SOC thresholds that are appropriate

to other geographical contexts beyond Europe. For instance, datasets,

including national soil survey data and a map of North American

Ecoregions to represent different bioclimates (

Koop et al., 2023

), com-

bined with land cover mapping and other contextual spatial data (e.g.

elevation) could be applied to determine O/T SOC ratios and establish

benchmarks across the USA. Key challenges for any application beyond

Europe will include ensuring that PCZs are i) representative of all min-

eral soil environments across the region; ii) O/T SOC grades are suffi-

ciently constrained to be sensitive enough to detect change over time in

response to land management interventions; iii) can be readily identified

by stakeholders to benchmark third-party data against regionally

defined O/T SOC grades (e.g. following the decision tree approach used

here for Europe). More widespread monitoring of O/T SOC ratios

internationally could contribute to realising key UN SDGs worldwide (e.

g. climate action and life on land), so adapting methods to determine O/

T SOC beyond Europe is an important direction for further research.

The EUSO recently introduced the

“

Soil Health Dashboard

”

, an

interactive web portal of map layers, which provides a convergence of

evidence assessment of threats to soil health across Europe (

Panagos

et al., 2024

). Several soil health indicators have been mapped at pan-

European scale and are fed into the Soil Health Dashboard to highlight

where soils may be classed as either

“

healthy

”

(no soil health threats

present) or

“

degraded

”

(at least 1 soil health threat is present). This

system provides a powerful tool for policymakers to see which soils are

most under threat and is the basis of the estimate that

>

60 % soils across

Europe are rated

“

degraded

”

(

Panagos et al., 2024

). The O/T SOC could

be incorporated in this system and used in conjunction with maps of

other soil health indicators to help refine understanding of which soils

are degraded across Europe. Details of how O/T SOC could be mapped

are included in Text S1 (Supplementary Materials) and we have pro-

duced maps of PCZs and O/T SOC grades using high-resolution digital

soil maps for Europe from EcoDataCube for illustration (Fig. S7). While

in principle, it is straightforward to map O/T SOC grades from digital

soil maps, in practice, key limitations should be considered, including: i)

the distribution of PCZs (and with them, O/T SOC thresholds) will

change over time as land cover changes; ii) predicted PCZ classes and O/

T SOC levels are sensitive to the accuracy of the digital soil maps used.

Digital soil maps often do not predict extreme values very effectively

(

Feeney et al., 2022

), so areas of PCZs defined by coarse textured soils

for instance, may be significantly under-estimated because predicted

clay contents at the lower end of the distribution may be over-estimated;

conversely, many soils that are mapped as having high mineral SOC

concentrations may in reality be organic soils (and therefore should not

be considered in O/T SOC assessment) because SOC values at the upper

end of the distribution have been under-estimated. Maps of any soil

health indicator should be sufficiently robust to provide reasonably

accurate assessments of soil health at regional scales, otherwise users

Fig. 5.

O/T SOC index classes and associated porosity distributions in Germany and Wales according to national monitoring datasets, BZE and GMEP, respectively

(a); PCZs represented in each dataset broken down into percentages (b). Numbers on the boxplots refer to the number of observations in each O/T SOC index class.

C.J. Feeney et al.

Science of the Total Environment 951 (2024) 175642

10

may be presented with misleading assessments that reflect weaknesses

in the maps themselves rather than current soil health status. Judge-

ments of what makes a soil health map

“

robust

”

, alongside other con-

siderations including the flexibility to choose different combinations of

soil health indicators and rank order their relative importance depend-

ing on user priorities, are all important areas for further research to

support tools for policymakers and other stakeholders.

Some limitations should be noted for O/T SOC. First, this new metric

only applies to mineral topsoils. In large part, this was done for com-

parison with the SOC/clay ratio, which was also designed for mineral

topsoils. There is also much wider variability of SOC concentrations for

organic (

≥

120 g kg

−

1

) compared to mineral soils (

<

120 g kg

−

1

), and it is

unclear whether O/T SOC index classes could usefully separate differing

grades in organic soil structure.

Robinson et al. (2022)

identified a

reduction in the rate of porosity increase with SOM content around the

mineral-organic soil transition (20 % soil organic matter). It is also likely

that a different configuration of environmental factors will best predict

SOC concentration deeper in the soil (

Vos et al., 2019

); thus, a different

set of

“

typical

”

SOC concentrations may be necessary for subsoils.

Second, data from all 3 extant LUCAS surveys were aggregated

together to develop the O/T SOC indicator. This was done primarily to

maximise the data available to generate PCZs and

“

typical

”

(average)

SOC concentrations in as statistically a robust way as possible. While this

does prevent any change over time analysis from being performed with

LUCAS, it has been demonstrated from several LTFE plots, spanning

semi-arid to cold-weather environments, that changes in SOC content

can be reflected by changes in O/T SOC status over management time-

scales (multiple years to decades).

Third, additional data are required to confirm whether increases in

O/T SOC would translate into increases in soil porosity and SOC stocks.

Fig. 6.

Changes in O/T SOC over multiple decades under contrasting land management interventions for 2 LTFEs in the UK: Broadbalk (a) and Highfield (b). The

Broadbalk LTFE consisted of a series of field plots under winter wheat cover, where different forms of fertiliser application were trialled. Dotted lines indicate where

starting SOC contents were estimated; dashed lines show where farmyard manure application paused during fallow periods (

). The

Highfield LTFE represents a bare fallow experiment across 4 plots along a north-south gradient (1 is most northerly; 4 is most southerly); here long-term grass (since

1838) was ploughed out in 1959 and kept in bare fallow by cultivation to suppress weeds (

Poulton et al., 2022

). The PCZ for both LTFE examples is

“

Atlantic and

Central arable loamy and clayey soils

”

; limits between the 4 O/T SOC categories for this PCZ are indicated by the horizontal grey lines with thresholds labelled.

C.J. Feeney et al.

Science of the Total Environment 951 (2024) 175642

11

Moreover, additional metrics related to soil structure (e.g. aggregation

and effective porosity) should be compared across O/T SOC grades to

further validate their robustness. This will require different datasets to

LUCAS, potentially from a wide range of pilot sites across Europe.

Lastly, SOC is not a universal indicator of soil health. Indeed, soils

can have a

“

Very High

”

O/T SOC condition but also be afflicted with

other issues such as excess nitrogen and phosphorous or high concen-

trations of metallic elements such as lead and copper. O/T SOC should

therefore be complimented with other indicators, reflecting various

physical, chemical and biological aspects of soils to assess soil health,

similar to the CASH framework (

Moebius-Clune et al., 2017

) in the USA.

4. Conclusion

The EU Soil Strategy 2030 aims to increase SOC to improve soil

health and promote carbon sequestration to help mitigate climate

change. Tracking progress towards these aims under the proposed EU

Soil Monitoring Law requires a SOC-based indicator that is applicable

across Europe, irrespective of environmental and management factors,

can separate differing grades of soil structure and carbon storage

effectively, and is sensitive enough to reflect the effects of land man-

agement interventions. Building on an existing blueprint to benchmark

observed SOC against a predicted value, which hitherto was based on

using clay alone as a predictor and for agricultural landscapes (

Poeplau

and Don, 2023

), we present the

“

observed/typical SOC

”

(O/T SOC)

ratio. O/T SOC represents an important advancement, as it is derived

from modelling that considers land cover and climate, in addition to soil

texture, as important predictors, which allows appropriate SOC levels to

be determined in different pedo-climatic zones (PCZs). Applying

regression tree modelling to Europe

’

s largest active soil monitoring

dataset (LUCAS), 12 PCZs, representing the full breadth of mineral soil

environments across Europe, were delineated, each with a predicted

average SOC concentration which defined the

“

typical

”

SOC. Dividing

observed SOC values in each PCZ by the corresponding typical SOC

value and splitting each population into quartiles defined the O/T SOC

grades, which were termed

“

low

”

,

“

intermediate

”

,

“

high

”

and

“

very

high

”

.

We successfully demonstrated that O/T SOC better reflects differing

grades of soil porosity, as a proxy for physical structure, and SOC stocks

than the SOC/clay ratio, establishing O/T SOC as a more useful indicator

of the technical potential for future SOC storage for both soil structure

enhancement and land-based carbon sequestration. The flexibility of our

indicator design means that O/T SOC is applicable to datasets

throughout Europe, which we demonstrated by assessing porosity dis-

tributions from national soil monitoring data in 2 European countries.

Analysis of multi-decadal time-series of SOC concentration changes from

several long-term field experiment plots in temperate, semi-arid and

cold-weather climate environments shows that O/T SOC can be useful

for detecting responses to various land management interventions. This

opens the possibility of incorporating O/T SOC as a soil health indicator

for the EU and its member states to track progress under the proposed EU

Soil Monitoring Law, or for guidance to landowners to engage in

voluntary monitoring. With adaptation to other datasets, the O/T SOC

ratio could be incorporated into decision support tools such as the EUSO

Soil Health Dashboard for policymakers. It could also serve as an

important indicator to guide efforts to enhance soil health beyond

Europe, ultimately contributing to the delivery of several UN SDGs

worldwide.

CRediT authorship contribution statement

Christopher J. Feeney:

Writing

–

review

&

editing, Writing

–

original draft, Methodology, Investigation, Formal analysis, Data cura-

tion, Conceptualization.

Laura Bentley:

Writing

–

review

&

editing,

Methodology.

Daniele De Rosa:

Writing

–

review

&

editing, Data

curation.

Panos Panagos:

Writing

–

review

&

editing, Data curation.

Bridget A. Emmett:

Writing

–

review

&

editing.

Amy Thomas:

Writing

–

review

&

editing.

David A. Robinson:

Writing

–

review

&

editing,

Supervision, Project administration, Methodology, Funding acquisition,

Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influence

the work reported in this paper.

Data availability

LUCAS soils data are available from the JRC European Soil Data

Centre (ESDAC) for 2009

–

12 from

https://esdac.jrc.ec.europa.

eu/content/lucas-2009-topsoil-data

, 2015 from

https://esdac.jrc.ec.

europa.eu/content/lucas2015-topsoil-data

and 2018 including bulk

density data from

https://esdac.jrc.ec.europa.eu/content/lucas-2018

-topsoil-data

. The EcoDataCube digital elevation map and Geo-

morphon90 landform classes can be accessed from OpenGeoHub:

https://opengeohub.org/

and

OpenTopography

https://portal.

opentopography.org/dataspace/dataset?opentopoID

=

OTDS.012020.43

26.1

, respectively. The land cover, biogeographical regions and EU

NUTS administrative boundaries are available from

https://land.

copernicus.eu/en/products/corine-land-cover/clc2018

,

https://www.

eea.europa.eu/data-and-maps/figures/biogeographical-regions-in-euro

pe-2

and

https://ec.europa.eu/eurostat/web/gisco/geodata/reference-

data/administrative-units-statistical-units/nuts

, respectively. GMEP

soil survey data for Wales is available from the Environmental Infor-

mation

Data

Centre:

https://catalogue.ceh.ac.uk/documents

/0fa51dc6-1537-4ad6-9d06-e476c137ed09

and BZE soil survey data

for Germany is available from the Thünen Institute:

https://www.

openagrar.de/receive/openagrar_mods_00054877

. Data for the Broad-

balk and Highfield long-term field experiments are available from the

Electronic Rothamsted Archive:

https://www.era.rothamsted.ac.

uk/index.php

.

Acknowledgements

This analysis was undertaken as part of the Horizon Europe project

AI4SoilHealth (Grant No. 101086179). We would like to thank the de-

velopers of the datasets used in our analysis, particularly the LUCAS,

BZE and GMEP soil monitoring teams. Additional comments on the

original version of the manuscript from Dr. Fabrizio Albanito (UKCEH)

and from 4 anonymous reviewers are also gratefully acknowledged.

Appendix A. Supplementary data

Supplementary data to this article can be found online at

https://doi.

org/10.1016/j.scitotenv.2024.175642

https://doi.org/10.1038/s41597-020-

References

Amatulli, G., McInerney, D., Sethi, T., Strobl, P., Domisch, S., 2020. Geomorpho90m,

empirical evaluation and accuracy assessment of global high-resolution

geomorphometric layers. Sci. Data 7 (1).

0479-6

.

Assouline, S., Or, D., 2013. Conceptual and parametric representation of soil hydraulic

properties: a review. Vadose Zone J. 12 (4), vzj2013-07

https://doi.org/10.2136/

vzj2013.07.0121

org/10.1016/j.geoderma.2015.07.006

Ballabio, C., Panagos, P., Monatanarella, L., 2016. Mapping topsoil physical properties at

European scale using the LUCAS database. Geoderma 261, 110

–

123.

https://doi.

.

Bell, S.M., Terrer, C., Barriocanal, C., Jackson, R.B., Rosell-Mel

´

e, A., 2021. Soil organic

carbon accumulation rates on Mediterranean abandoned agricultural lands. Sci.

Total Environ. 759

https://doi.org/10.1016/j.scitotenv.2020.143535

.

Blanco-Canqui, H., Shapiro, C.A., Wortmann, C.S., Drijber, R.A., Mamo, M., Shaver, T.M.,

Ferguson, R.B., 2013. Soil organic carbon: the value to soil properties. J. Soil Water

Conserv. 68 (5)

https://doi.org/10.2489/jswc.68.5.129A

.

C.J. Feeney et al.

Science of the Total Environment 951 (2024) 175642

12

Bonfante, A., Basile, A., Bouma, J., 2020. Targeting the soil quality and soil health

concepts when aiming for the United Nations Sustainable Development Goals and

the EU Green Deal. Soil 6 (2), 453

–

466.

https://doi.org/10.5194/soil-6-453-2020

.

Bouma, J., 2014. Soil science contributions towards Sustainable Development Goals and

their implementation: linking soil functions with ecosystem services. J. Plant Nutr.

Soil Sci. 177 (2), 111

–

120.

https://doi.org/10.1002/jpln.201300646

.

Breuning-Madsen, H., Kristensen, J. A.,

&

Balstrom, T. (2015). Final report on the

establishment of a spade 14 soil profile analytical database connected to the EU Soil

Map at Scale 1

:1.000.000

.

Bünemann, E.K., Bongiorno, G., Bai, Z., Creamer, R.E., De Deyn, G., de Goede, R.,

Fleskens, L., Geissen, V., Kuyper, T.W., M

¨

ader, P., Pulleman, M., Sukkel, W., van

Groenigen, J.W., Brussaard, L., 2018. Soil quality

–

a critical review. Soil Biol.

Biochem. 120, 105

–

125.

https://doi.org/10.1016/j.soilbio.2018.01.030

.

D

’

Andrimont, R., Yordanov, M., Martinez-Sanchez, L., Haub, P., Buck, O., Haub, C.,

Eiselt, B., Van Der Velde, M., 2022. LUCAS cover photos 2006-2018 over the EU: 874

646 spatially distributed geo-tagged close-up photos with land cover and plant

species label. Earth System Science Data 14 (10), 4463

–

4472.

10.5194/essd-14-4463-2022

.

De Rosa, D., Ballabio, C., Lugato, E., Fasiolo, M., Jones, A., Panagos, P., 2024. Soil

organic carbon stocks in European croplands and grasslands: how much have we lost

in the past decade? Glob. Chang. Biol. 30 (1)

https://doi.org/10.1111/gcb.16992

https://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 (3

–

4),

620

–

627.

https://environment.ec.europa.eu/publications/proposal-directive

Directorate-General for Environment, 2023. Proposal for a Directive on Soil Monitoring

and Resilience.

-soil-monitoring-and-resilience_en

.

Drexler, S., Broll, G., Flessa, H., Don, A., 2022. Benchmarking soil organic carbon to

support agricultural carbon management: a German case study. J. Plant Nutr. Soil

Sci. 185 (3), 427

–

440.

https://doi.org/10.1002/jpln.202200007

.

Elith, J., Leathwick, J.R., Hastie, T., 2008. A working guide to boosted regression trees.

J. Anim. Ecol. 77 (4), 802

–

813.

https://doi.org/10.1111/j.1365-2656.2008.01390.

x

.

European Commission, 2023. A soil deal for Europe.

Https://Research-and-Innovation.

Ec.Europa.Eu/Funding/Funding-Opportunities/Funding-Programmes-and-Open-Ca

lls/Horizon-Europe/Eu-Missions-Horizon-Europe/Soil-Deal-Europe_en

.

European Environment Agency, 2017. Biogeographical regions in Europe. Environmental

Information Systems.

https://www.eea.europa.eu/data-and-maps/figures/biogeo

graphical-regions-in-europe-2

.

FAO, ITPS, 2015. Status of the World

’

s Soil Resources (SWSR) - Main Report

Feeney, C.J., Cosby, B.J., Robinson, D.A., Thomas, A., Emmett, B.A., Henrys, P., 2022.

Multiple soil map comparison highlights challenges for predicting topsoil organic

carbon concentration at national scale. Sci. Rep. 12, 1379.

https://doi.org/10.1038/

s41598-022-05476-5

https://doi.org/10.1016/j.scitotenv.2023.163973

Feeney, C.J., Robinson, D.A., Keith, A.M., Vigier, A., Bentley, L., Smith, R.P., Garbutt, A.,

Maskell, L.C., Norton, L., Wood, C.M., Cosby, B.J., Emmett, B.A., 2023. Development

of soil health benchmarks for managed and semi-natural landscapes. Sci. Total

Environ. 886 (April), 163973

.

Fernandes-Ugalde, O., Scarpa, S., Orgiazzi, A., 2022. LUCAS 2018 Soil Module.

https://

doi.org/10.2760/215013

https://www.researchgate.net/publication

Hassink, J., 1997. A model of the physical protection of organic matter in soils the

capacity of soils to preserve organic C and N by their association with clay and silt

particles. Plant and Soil 191, 77

–

87.

/40154117

org/10.1016/j.geomorph.2012.11.005

Jasiewicz, J., Stepinski, T.F., 2013. Geomorphons-a pattern recognition approach to

classification and mapping of landforms. Geomorphology 182, 147

–

156.

https://doi.

Johannes, A., Matter, A., Schulin, R., Weisskopf, P., Baveye, P.C., Boivin, P., 2017.

Optimal organic carbon values for soil structure quality of arable soils. Does clay

content matter? Geoderma 302 (September 2016), 14

–

21.

https://doi.org/10.1016/

j.geoderma.2017.04.021

.

Jones, A., Fernandez-Ugalde, O., Scarpa, S., 2020. LUCAS 2015 Topsoil Survey.

https://

doi.org/10.2760/616084

https://doi.org/10.1016/j.agee.2011.02.029

K

¨

atterer, T., Bolinder, M.A., Andr

´

en, O., Kirchmann, H., Menichetti, L., 2011. Roots

contribute more to refractory soil organic matter than above-ground crop residues,

as revealed by a long-term field experiment. Agric. Ecosyst. Environ. 141 (1

–

2),

184

–

192.

.

Keesstra, S., Mol, G., de Leeuw, J., Okx, J., Molenaar, C., de Cleen, M., Visser, S., 2018.

Soil-related sustainable development goals: four concepts to make land degradation

neutrality and restoration work. Land 7 (4).

https://doi.org/10.3390/land7040133

.

Keesstra, Saskia, Bouma, J., Wallinga, J., Tittonell, P., Smith, P., Cerd

`

a, A.,

Montanarella, L., Quinton, J.N., Pachepsky, Y., Van Der Putten, W.H., Bardgett, R.D.,

Moolenaar, S., Mol, G., Jansen, B., Fresco, L.O., 2016. The significance of soils and

soil science towards realization of the United Nations sustainable development goals.

Soil 2 (2), 111

–

128.

https://doi.org/10.5194/soil-2-111-2016

https://doi.org/10.1016/j.

Koop, A.N., Hirmas, D.R., Billings, S.A., Li, L., Cueva, A., Zhang, X., Wen, H., Nemes, A.,

Souza, L.F.T., Ajami, H., Flores, A.N., Rudick, A.K., Guthrie, A., Klamm, L.M.,

Unruh, M., Sullivan, P.L., 2023. Is macroporosity controlled by complexed clay and

soil organic carbon? Geoderma 437.

geoderma.2023.116565

.

Kopittke, P.M., Berhe, A.A., Carrillo, Y., Cavagnaro, T.R., Chen, D., Chen, Q.-L., Rom

´

an

Dobarco, M., Dijkstra, F.A., Field, D.J., Grundy, M.J., He, J.-Z., Hoyle, F.C., K

¨

ogel-

Knabner, I., Lam, S.K., Marschner, P., Martinez, C., McBratney, A.B., McDonald-

Madden, E., Menzies, N.W., Minasny, B., 2022. Ensuring planetary survival: the

centrality of organic carbon in balancing the multifunctional nature of soils. Crit.

Rev. Environ. Sci. Technol. 52 (23), 4308

–

4324.

https://doi.org/10.1080/

10643389.2021.2024484

https://cran.r-project.org/web/packages/caret/index.

Kuhn, M., Wing, J., Weston, S., Williams, A., Keefer, C., Engelhardt, A., Cooper, T.,

Mayer, Z., Kenkel, B., R Core Team, Benesty, M., Lescarbeau, R., Ziem, A.,

Scrucca, L., Tang, Y., Candan, C., Hunt, T., 2023. caret: Classification and Regression

Training (6.0-94), CRAN.

html

a, R., Menichetti, L., Martinez-Garcia, E., T

Lal, R., 2004. Soil carbon sequestration impacts on global climate change and food

security. Science 304 (5677), 1623

–

1627.

https://doi.org/10.1126/

.

orm

anen, T., Lehtonen, A., 2024. Is the

organic carbon-to-clay ratio a reliable indicator of soil health? Geoderma 444,

116862

.

Mattila, T.J., 2023. Redox potential as a soil health indicator

–

how does it compare to

microbial activity and soil structure? Plant and Soil.

https://doi.org/10.1007/

s11104-023-06305-y

Matus, F.J., 2021. Fine silt and clay content is the main factor defining maximal C and N

accumulations in soils: a meta-analysis. Sci. Rep. 11 (1), 1

–

17.

10.1038/s41598-021-84821-6

Moebius-Clune, B.N., Moebius-Clune, D.J., Gugino, B.K., Idowu, O.J., Schindelbeck, R.R.,

Ristow, A.J., van Es, H.M., Thies, J.E., Shayler, H.A., McBride, M.B., Kurtz, K.S.M.,

Wolfe, D.W., Abawi, G.S., 2017. Comprehensive Assessment of Soil Health - The

Cornell Framework (3.2). Cornell University

org/web/packages/soiltexture/index.html

Moeys, J., Shangguan, W., Petzold, R., Minasny, B., Rosca, B., Jelinski, N., Zelazny, W.,

Souza, R.M.S., Safanelli, J.L., Caten, A. Ten, 2018. soiltexture: functions for soil

texture plot, classification and transformation (1.5.1), CRAN.

https://cran.r-project.

.

Moinet, G.Y.K., Hijbeek, R., van Vuuren, D.P., Giller, K.E., 2023. Carbon for soils, not

soils for carbon. Glob. Chang. Biol. 29 (9), 2384

–

2398.

https://doi.org/10.1111/

gcb.16570

.

Or, D., Smets, B.F., Wraith, J.M., Dechesne, A., Friedman, S.P., 2007. Physical constraints

affecting bacterial habitats and activity in unsaturated porous media

–

a review. Adv.

Water Resour. 30 (6

–

7), 1505

–

1527.

https://doi.org/10.1016/j.

advwatres.2006.05.025

.

Orgiazzi, A., Ballabio, C., Panagos, P., Jones, A., Fern

´

andez-Ugalde, O., 2018. LUCAS

soil, the largest expandable soil dataset for Europe: a review. Eur. J. Soil Sci. 69 (1),

140

–

153.

https://doi.org/10.1111/ejss.12499

Pacini, L., Yunta, F., Jones, A., Montanarella, L., Barr

`

e, P., Saia, S., Chen, S., Schillaci, C.,

2023. Fine earth soil bulk density at 0.2m depth from Land Use and Coverage Area

frame Survey (LUCAS) soil 2018. Eur. J. Soil Sci. May, 1

–

11.

https://doi.org/10.1016/j.landusepol.2011.07.003

10.1111/ejss.13391

.

Panagos, P., Van Liedekerke, M., Jones, A., Montanarella, L., 2012. European Soil Data

Centre: response to European policy support and public data requirements. Land Use

Policy 29 (2), 329

–

338.

https://doi.org/10.1111/ejss.13315

Panagos, P., Van Liedekerke, M., Borrelli, P., K

¨

oninger, J., Ballabio, C., Orgiazzi, A.,

Lugato, E., Liakos, L., Hervas, J., Jones, A., Montanarella, L., 2022. European Soil

Data Centre 2.0: soil data and knowledge in support of the EU policies. Eur. J. Soil

Sci. 73 (6), 1

–

18.

https://doi.org/10.1111/ejss.13466

Panagos, P., Borrelli, P., Jones, A., Robinson, D.A., 2024. A 1 billion euro mission: a Soil

Deal for Europe. Eur. J. Soil Sci. 75 (1)

.

Poeplau, C., Don, A., 2023. A simple soil organic carbon level metric beyond the organic

carbon-to-clay ratio. Soil Use Manag. 1057

–

1067.

https://doi.org/10.1111/

sum.12921

(December 2022).

Poeplau, C., Vos, C., Don, A., 2017. Soil organic carbon stocks are systematically

overestimated by misuse of the parameters bulk density and rock fragment content.

Soil 3 (1), 61

–

66.

https://doi.org/10.5194/soil-3-61-2017

Poulton, P.R., MacDonald, A.J., Gregory, A.S., Glendining, M.J., Ostler, R.J., 2022.

Highfield Bare Fallow Soil Chemical Properties 1959

2014. Electronic Rothamsted

Archive, Rothamsted Research

https://doi.org/10.1016/j.earscirev.2021.103689

Pr

˘

av

˘

alie, R., 2021. Exploring the multiple land degradation pathways across the planet.

Earth Sci. Rev. 220 (May)

https://doi.org/10.1111/ejss.13012

Prout, J.M., Shepherd, K.D., McGrath, S.P., Kirk, G.J.D., Haefele, S.M., 2021. What is a

good level of soil organic matter? An index based on organic carbon to clay ratio.

Eur. J. Soil Sci. 72 (6), 2493

–

2503.

Prout, J.M., Shepherd, K.D., McGrath, S.P., Kirk, G.J.D., Hassall, K.L., Haefele, S.M.,

2022. Changes in organic carbon to clay ratios in different soils and land uses in

England and Wales over time. Sci. Rep. 12 (1), 1

–

13.

https://doi.org/10.1038/

s41598-022-09101-3

https://www.r-project.org/

R Core Team, 2023. R: A language and environment for statistical computing (4.3.0),

CRAN.

https://doi.org/10.1016/j.

Rabot, E., Saby, N.P.A., Martin, M.P., Barr

´

e, P., Chenu, C., Cousin, I., Arrouays, D.,

Angers, D., Bispo, A., 2024. Relevance of the organic carbon to clay ratio as a

national soil health indicator. Geoderma 443.

geoderma.2024.116829

.

Reynolds, B., Chamberlain, P.M., Poskitt, J., Woods, C., Scott, W.A., Rowe, E.C.,

Robinson, D.A., Frogbrook, Z.L., Keith, A.M., Henrys, P.A., Black, H.I.J., Emmett, B.

A., 2013. Countryside survey: national

“

Soil Change

”

1978

–

2007 for topsoils in

Great Britain-acidity, carbon, and total nitrogen status. Vadose Zone J. 12 (2),

vzj2012.0114

https://doi.org/10.2136/vzj2012.0114

.

Robinson, D.A., Thomas, A., Reinsch, S., Lebron, I., Feeney, C.J., Maskell, L.C., Wood, C.

M., Seaton, F.M., Emmett, B.A., Cosby, B.J., 2022. Analytical modelling of soil

porosity and bulk density across the soil organic matter and land-use continuum. Sci.

Rep. 12 (1)

https://doi.org/10.1038/s41598-022-11099-7

Rothamsted Research, 2021. Dataset: Broadbalk Soil Organic Carbon Content

1843

2015. Electronic Rothamsted Archive, Rothamsted Research

Six, J., Conant, R.T., Paul, E.A., Paustian, K., 2002. Stabilization mechanisms of SOM

C.J. Feeney et al.

Science of the Total Environment 951 (2024) 175642

13

Schmidt, M.W.I., Torn, M.S., Abiven, S., Dittmar, T., Guggenberger, G., Janssens, I.A.,

Kleber, M., K

¨

ogel-Knabner, I., Lehmann, J., Manning, D.A.C., Nannipieri, P.,

Rasse, D.P., Weiner, S., Trumbore, S.E., 2011. Persistence of soil organic matter as an

ecosystem property. Nature 478 (7367), 49

–

56.

https://doi.org/10.1038/

nature10386

.

implications for C saturation of soils. Plant and Soil 241 (2), 155

176

https://cran.r-project.org/package

Therneau, T., Atkinson, B., Ripley, B., 2022. rpart: Recursive Partitioning and Regression

Trees (4.1.19), CRAN.

=

rpart

.

Thomas, A., Seaton, F., Dhiedt, E., Cosby, B.J., Feeney, C., Lebron, I., Maskell, L.,

Wood, C., Reinsch, S., Emmett, B.A., Robinson, D.A., 2024. Topsoil porosity

prediction across habitats at large scales using environmental variables. Sci. Total

Environ. 922

https://doi.org/10.1016/j.scitotenv.2024.171158

.

T

´

oth, G., Jones, A., Montanarella, L., 2013a. LUCAS Topsoil Survey: Methodology, Data,

and Results.

https://doi.org/10.2788/97922

T

´

oth, G., Jones, A., Montanarella, L., 2013b. The LUCAS topsoil database and derived

information on the regional variability of cropland topsoil properties in the

European Union. Environ. Monit. Assess. 185 (9), 7409

–

7425.

10.1007/s10661-013-3109-3

.

Veerman, C., Correia, T.P., Bastioli, C., Biro, B., Bouma, J., Cienciala, E., Emmett, B.,

Frison, E.A., Grand, A., Filchew, L.H., Kriau

ˇ

ci

¯

unien

˙

e, Z., Pogrzeba, M., Soussana, J.

F., Olmo, C.V., Wittkowski, R., 2020. Caring for Soil Is Caring for Life

–

Ensure 75%

of Soils Are Healthy by 2030 for Food, People, Nature and Climate.

10.2777/611303

.

Visser, S., Keesstra, S., Maas, G., de Cleen, M., Molenaar, C., 2019. Soil as a basis to create

enabling conditions for transitions towards sustainable land management as a key to

achieve the SDGs by 2030. Sustainability 11 (23).

https://doi.org/10.3390/

su11236792

https://doi.org/10.1111/ejss.12787

Vos, C., Don, A., Hobley, E.U., Prietz, R., Heidkamp, A., Freibauer, A., 2019. Factors

controlling the variation in organic carbon stocks in agricultural soils of Germany.

Eur. J. Soil Sci. 70 (3), 550

–

564.

.

Wickham, H., 2023. tidyverse: Easily Install and Load the ‘Tidyverse

’

(2.0.0), CRAN.

htt

ps://cran.r-project.org/web/packages/tidyverse/index.html

W

osten, J.H.M., 2000. The HYPRES database of hydraulic properties of European soils.

Advances in GeoEcology 32, 135

143