King & Sokol (2025). Soil carbon formation is promoted by saturation deficit and existing mineral-associated carbon, not by microbial carbon-use efficiency

copyright © 2025 the
Authors, some rights
reserved; exclusive
licensee American
Association for the
Advancement of
Science. no claim to
original U.S.
Government Works.
distributed under a
creative commons
Attribution license 4.0
(cc BY).

Downloaded from https://www.science.org on October 19, 2026

King and Sokol

,

Sci. Adv.

11

, eadv9482 (2025) 13 June 2025

S c i e n c e A d vA n c e S

|

R e S e A R c h A R t i c l e

2 of 10

deficit, and microbial CUE affected the formation of new MAOC. We

selected a range of 118 plots across 15 agricultural sites in the conti-

nental US that varied in a wide range of properties, including their

existing MAOC concentration, as well as in their MAOC saturation

status and soil physicochemical properties (

18

) (fig. S1 and table S1).

These soils were then subjected to two distinct C addition treat-

ments in laboratory microcosms: a regular input of

13

C- DOC at

fixed location (generating a high C input treatment) versus a dis-

persed, irregular application of the same

13

C- DOC and at only half

the concentration, generating a low C input treatment. These treat-

ments simulated how C enters into some regions of the soil as a

more concentrated input (e.g., the area directly around the living

root—the rhizosphere) versus other regions of the soil as a more

diffuse input, spread across a greater area (e.g., the bulk soil) (

31

).

Treatments and observations were carried out over a 6- month peri-

od, commensurate with the growth cycle of an agricultural cover

crop, for a total of 25

13

C- DOC additions. Regular trace gas mea-

surements throughout the incubation allowed assessment of the

proportion of

13

C- DOC respired as CO

2

during the initial phase of

rapid microbial consumption of

13

C- DOC; formation of MAOC

from the

13

C- DOC and microbial CUE were assessed at the termi-

nation of the incubation. We asked: (i) Do soils with greater existing

MAOC concentrations support greater or lesser formation of new

MAOC? (ii) Do soils farther from their MAOC saturation deficit

support greater MAOC formation? and (iii) Does microbial CUE

influence the formation efficiency of MAOC under varying input

levels of C input?

RESULTS AND DISCUSSION

Formation of MAOC is regulated by existing MAOC

concentration and saturation deficit

Across 15 different agricultural sites and 118 plots, we found that

there was a strong positive relationship between the amount of new

MAOC formed and existing MAOC concentration (Fig. 1,

R

=

0.28

to 0.46;

P

<

0.01). Our results are consistent with isotope tracing

work in the field, where existing concentrations of total SOC pro-

mote formation of new MAOC (

32

). Our results are also consistent

with a previous isotope tracing work under laboratory conditions

where, in two contrasting soils, existing MAOC promoted forma-

tion of new MAOC (

20

). Our results provide more substantive evi-

dence for the role of existing MAOC in promoting new MAOC

formation, by showing this relationship through a controlled exper-

iment and across a much wider range of soils.

The formation of new MAOC also increased with MAOC satura-

tion deficit (Fig. 2, A and B). Although a role for saturation deficit in

regulating formation of new MAOC is widely expected, a quantifi-

cation of MAOC saturation deficit as a driver of MAOC formation

has not been clearly demonstrated. Our results indicate that, al-

though most soils were far from MAOC saturation, as is typical in

agricultural soils (

2

), MAOC saturation nevertheless influenced

MAOC formation across high versus low C input treatments. Al-

though saturation status has been theorized to influence the forma-

tion of particulate organic carbon (POC) as well (

7

), in this study,

the formation of new POC from the

13

C- DOC input was two orders

of magnitude lower than formation of new MAOC (fig. S2), as ex-

pected with the use of this nonstructural

13

C- DOC label, since it is

structural inputs that primarily enter the POC pool (

33

). For the

formation of new MAOC, our results suggest a dual role of both

mineral surfaces and existing MAOC as a base for formation of

new MAOC.

To accurately assess the relationship between existing MAOC con-

centrations and new MAOC formation, it is important to also consider

the soil physicochemical properties that covary with MAOC. For in-

stance, soils with greater quantities of Al

O

, Fe

O

, and Ca

ex

typically con-

tain higher MAOC or SOC levels (

). Thus, in our soils, it is

possible that soil physicochemical properties (such as concentrations

of Al

O

, Fe

O

, Ca

ex

, or the percent of silt and clay) may have also contrib-

uted to new MAOC formation to an equal or greater extent than exist-

ing MAOC concentration. To test this possibility, we constructed a

Fig. 1.

Patterns in MAOC across scales.

(

A

) Map of study sites in the United States. At each site, 6 to 10 subplots were sampled and incubated with 25 weekly applications

of

13

c- dOc under two c input treatments (low,

n

=

116; high,

n

=

115), after which (

B

) formation of new MAOc was quantified, here shown as a function of existing MAOc.

Downloaded from https://www.science.org on October 19, 2026

King and Sokol

,

Sci. Adv.

11

, eadv9482 (2025) 13 June 2025

S c i e n c e A d vA n c e S

|

R e S e A R c h A R t i c l e

3 of 10

structural equation model to test causal pathways influencing forma-

tion of new MAOC. In this model, soil physicochemical properties

were given an opportunity to increase MAOC, whereas MAOC satu-

ration deficit reduced existing MAOC, and all properties could influ-

ence formation of new MAOC. This choice of causal pathway was

informed by model fit indices, which indicated poor model fit when

MAOC influenced saturation deficit. Although silt

+

clay concentra-

tion did influence saturation deficit in these soils, it was removed from

the structural equation models due to high collinearity with MAOC

saturation deficit, which would have caused incorrect estimation of

path coefficients. Substituting silt

+

clay for saturation deficit resulted

in reduced explanatory power over new MAOC formation (fig. S3).

Structural equation modeling revealed that, after the influence of other

soil physiochemical properties on new MAOC had been accounted

for, across both high and low C input treatments, there was still a sig-

nificant effect of existing MAOC on new MAOC formation (Fig. 2B).

As with soil minerals, OC molecules have the capacity to form

associations with other OC molecules due to their surface charges—

known as “organo- organic interactions” (

21

,

36

). Because of these

organo- organic interactions and multilayer stacking of OC, a greater

amount of existing MAOC reasonably leads to greater new MAOC

formation. We also observed the C:N ratio of newly- formed MAOC

increasing with soil Ca

ex

(fig. S4). This relationship could be because

soils with greater matrix capacity to stabilize MAOC, e.g., through

cation bridging with Ca

ex

, formed MAOC via direct sorption of

DOC in organo- mineral or organo- organo interactions, rather than

through a microbial formation pathway. Calcium may have also been

associated with greater capacity for aggregate protection of DOC

(

35

). As a result, the newly formed MAOC may be less microbially

processed in soils higher in Ca

ex

as its C:N ratio is closer to that of the

added

13

C- DOC.

Formation efficiency of MAOC not explained by

microbial CUE

As hypothesized, a higher rate of C input led to lower proportion of

13

C- DOC that was retained as MAOC (Fig. 3A). In the low C input

treatment, 37

%

of

13

C- DOC applied was retained as MAOC, where-

as only 23

%

of

13

C- DOC applied was retained under the higher rate

of input. Microbial CUE likely did not explain these differences be-

cause CUE [measured via the

18

O- H

2

O method (

37

)] was not sig-

nificantly different between these treatments (

P

=

0.14; Fig. 3B).

Microbial community- level growth rates also did not explain differ-

ences in formation efficiency, as growth rates increased slightly in

the high C input treatment (

P

<

0.05; Fig. 3C) and did not explain

formation efficiency differences within treatments (Fig. 3E). Micro-

bial CUE and growth were poor predictors of absolute MAOC for-

mation (fig. S5,

R

2

<

0.05). It is possible that other differences in

microbial physiology influenced differential MAOC formation effi-

ciencies in the high versus low input treatments (such as extracel-

lular enzyme activity or extracellular polymeric substances) (

29

),

Fig. 2.

Relationships between soil properties, including existing MAOC, and the formation of new MAOC.

(

A

) Multilevel Pearson’s correlation coefficients for rela-

tionships between formation of new MAOc (assessed as retention of a

13

c- dOc label in MAOc, mg

13

c- dOc- derived MAOc kg soil

−

1

) and existing MAOc (MAOc, g c kg

soil

−

1

), oxalate- extractable aluminum (Al

O

, g kg soil

−

1

), oxalate- extractable iron (Fe

O

, g kg soil

−

1

), exchangeable calcium (ca

ex

, cmol kg soil

−

1

), and saturation deficit (sat.

def.; g c kg soil

−

1

), across both levels of c input addition. (

B

) Structural equation model, in which standardized path coefficients were averaged across low and high c input

treatments after multilevel SeM testing indicated no significant difference between a constrained model (regression coefficients forced to be the same between treat-
ments) and an unconstrained model (regression coefficients could vary between treatments). Blue paths represent significant positive relationships (

P

<

0.05); red paths

represent significant negative relationships (

P

<

0.05); gray path represents nonsignificant relationships. Both models have comparative fit index (cFi)

>

0.95 and root

mean square error of approximation (RMSeA)

<

0.08. For both high and low c input treatments,

n

=

107 after uniform outlier cleaning procedure (see the section “Statisti-

cal analyses”).

Downloaded from https://www.science.org on October 19, 2026

King and Sokol

,

Sci. Adv.

11

, eadv9482 (2025) 13 June 2025

S c i e n c e A d vA n c e S

|

R e S e A R c h A R t i c l e

4 of 10

but these differences were not related to microbial CUE or growth

rate—two of the most commonly measured microbial traits related

to the microbial formation pathway of MAOC (

The low C input treatment may have stabilized MAOC with

greater efficiency than the high C input treatment if it favored a more

efficient, direct sorption pathway, compared to a less efficient, micro-

bial formation pathway, where C can be lost via microbial respiration.

We found that the low C input treatment was associated with reduced

microbial uptake of the

13

C- DOC input (fig. S6), as quantified by

13

C- DOC respiration, and more was retained as MAOC. Specifically,

in this experiment and in other experiments tracing isotopically-

enriched DOC, respiration of the DOC follows a biphasic pattern:

initial rapid respiration of the DOC, followed by slow respiration (

39

)

(fig. S7). In the initial phase, there may have been greater spatial prox-

imity of microbes to

13

C- DOC in the high C input treatment due

to more concentrated

13

C- DOC g soil

−

1

, or greater microbial bio-

mass (fig. S8), which could facilitate greater microbial uptake (

40

).

Such data support the idea that the low C input treatment may

have facilitated greater direct sorption versus microbial uptake of

13

C- DOC. Lower microbial uptake of

13

C- DOC in the low C input

treatment is also supported by slightly higher and more uniform C:N

ratio of newly formed MAOC in the low versus high C input treat-

ments [13.1 versus 12.7 C:N ratio; interquartile range (IQR) 1.93 ver-

sus 1.04; fig. S9,

P

=

0.047], indicating that MAOC in the low C input

treatment had a higher ratio of unprocessed DOC to microbial nec-

romass and a less consistent degree of microbial processing (

41

,

42

).

Lower microbial uptake of

13

C- DOC in the low versus high C input

treatment is also supported by lower proportional losses of

13

C- DOC

in the initial phase of respiration in the low C input treatment (fig. S6,

P

<

0.001). In this experiment, the high C input treatment lost 54

%

of applied

13

C- DOC in initial respiration phases, compared to 37

%

in the low C input treatments, which may have also been caused by

lower concentrations of DOC in soil solution, or other aspects of

microbial physiology not captured by the CUE assay. Overall, vari-

ability in microbial uptake between treatments and systems (

41

)

underscores a broad challenge of linking measures of microbial CUE

to MAOC formation.

Microbial CUE was highly variable within high and low C in-

put treatments (Fig. 3B), as was microbial community- level growth

rate (Fig. 3E), neither of which explained differences in MAOC

formation efficiency (

R

2

<

0.02;

P

>

0.22) or absolute MAOC for-

mation (fig. S5). Although the substrate- independent nature of

the

18

O- CUE approach is often highlighted as an advantage of this

CUE method (

37

), this feature may also be a limitation when ex-

amining the fate of a specific substrate. Microbial growth efficien-

cies can vary between different substrates (

43

), such as between

the existing SOC and newly added DOC. Consequently, if the aim

is to track the fate of a DOC label, then the substrate- independent

nature of the

18

O- CUE approach might obscure the microbial pro-

cessing of the label with the simultaneous consumption of existing

SOC. Our results supply an important rationale to reexamine the

assumption that substrate- independent tests of CUE are more reli-

able than substrate- specific tests. Revisiting substrate- specific mea-

sures of CUE—or other aspects of microbial metabolism—will be

particularly relevant if studies are to be manipulative (i.e., tracking

specific isotopically labeled compounds with microbial metabolism

Fig. 3.

Microbial CUE and growth rate do not explain differences in formation efficiency of MAOC between or within C input treatments.

(

A

) differences in forma-

tion efficiency of MAOc, assessed as

%

13

c- dOc recovered in MAOc (low,

n

=

115; high,

n

=

118), between c input treatments. (

B

) Microbial cUe (low,

n

=

111; high,

n

=

110) and (

C

) microbial growth rate (low,

n

=

108; high,

n

=

99). An asterisk between c input treatment boxplots indicates difference between treatments at

P

<

0.05.

(

D

) Formation efficiency of MAOc in both c input treatments as a function of cUe and (

E

) formation efficiency of MAOc in both c input treatments as a function of micro-

bial growth rate.

Downloaded from https://www.science.org on October 19, 2026

King and Sokol

,

Sci. Adv.

11

, eadv9482 (2025) 13 June 2025

S c i e n c e A d vA n c e S

|

R e S e A R c h A R t i c l e

5 of 10

as putative control on MAOC formation) rather than observation-

al (i.e., measuring microbial metabolism and its correlation with

MAOC pools). In this application of a widely used CUE measure-

ment to test its capacity to predict MAOC formation, we found

that features of the soil matrix—existing MAOC and MAOC satu-

ration deficit—emerged as stronger determinants of MAOC for-

mation than microbial CUE (Fig. 4).

Role of MAOC and microbial traits in MAOC accumulation

Understanding the primary controls on MAOC formation is critical

for determining how much C can be accrued in soils. To date, much

attention has focused on the availability of mineral surfaces in soil

that can adsorb OC, as well as the efficiency of the microbial com-

munity (often measured as microbial CUE) on controlling the for-

mation of new MAOC (

27

,

44

). Here, by assessing formation of new

MAOC across soils ranging in both MAOC and saturation deficit

and then assaying microbial CUE, we demonstrate that both exist-

ing MAOC concentrations and MAOC saturation deficit increase

new MAOC formation. In contrast with current theory, we found

that microbial CUE did not clearly influence MAOC formation.

Our results suggest that the concentration or rate of DOC inputs it-

self may be an important feature that moderates MAOC formation

efficiency by controlling microbial uptake versus direct sorption.

The observation that greater existing MAOC concentrations pro-

mote greater new MAOC formation opens key questions about how

soils can optimally accrue MAOC and how persistent new MAOC

will be over time. If existing MAOC stocks do not become more

vulnerable to destabilization as their size increases, or if this effect is

minimal, then soils with the greatest MAOC stocks hold substantial

potential to store more C. However, a simple positive relationship

between MAOC stocks and net MAOC formation and persistence is

not clearly supported by other lines of evidence, such as diminishing

returns in SOC accumulation across a gradient of C inputs (

45

). It is

likely that turnover of a portion of the MAOC pool increases with

greater accrual of MAOC, such that the formation of new MAOC is

partially or completely balanced out the by destabilization of exist-

ing MAOC. This possibility is consistent with the expectation that

more “stacked” MAOC is more vulnerable to destabilization (

36

)

and underscores emerging understanding of MAOC as a fraction

with heterogeneous turnover times (

46

)—i.e., some less vulnerable

MAOC at the base of MAOC stacks and some rapidly cycling

MAOC that attracts formation of new MAOC and is likely to be

destabilized through a priming process (

47

). A critical next step will

be to better understand how durable new MAOC is across soils with

a range of different MAOC concentrations and different proximities

to a theorized C saturation point.

Future work should also focus on untangling the mechanisms

controlling OC sorption to both vacant and occupied mineral sur-

faces. As existing MAOC can promote MAOC formation, as we

report here, mechanisms other than the absence of vacant mineral

surfaces are needed to explain the saturation, or nonaccumulation,

of MAOC over time. Continued reliance on bare, or pristine, min-

erals to study MAOC dynamics is likely to yield an incomplete pic-

ture of the drivers of MAOC cycling in natural soils, given the

common presence of OC coatings on soil minerals. Understanding

the extent to which the findings presented here generalize across

forest and grassland ecosystems and under in situ temperature and

moisture conditions will also be critical for predicting the accrual of

MAOC across different environmental contexts. It will also be im-

portant to refine CUE measurement approaches and test a wider

suite of microbial traits, beyond CUE and growth rate (such as the

amount and chemistry of extracellular polymeric substances) (

48

).

Together, these research avenues will clarify the features of soil mi-

crobial communities, the soil matrix, and existing OC that con-

strain net MAOC accumulation, in order to optimally manage soils

for C storage.

MATERIALS AND METHODS

Site selection and soil sampling

Soils from 15 agricultural sites across the United States (

18

) were

sampled in fall 2020 from the top 20 cm. Soils were transported to

Colorado State University campus on ice and transferred to 4°C

walk- in upon arrival. All soils were sieved to 8 mm while fresh, re-

moving only roots and rocks remaining on sieve surface. Air- dried

soils were root picked for 2 min for every 100 dry grams of soil. Soils

were stored air- dried until the incubation experiment was initiated

in March 2021.

Soil physicochemical properties

A suite of soil physicochemical properties was assessed on baseline

soils only and previously reported (

18

); a summary table of soil prop-

erties is provided in table S1, and the full dataset is available in a

Figshare repository (https://doi.org/10.6084/m9.figshare.25035818).

Among these properties, only soil pH and soil C were assessed again

on incubated soils. For both baseline and postincubation measure-

ments, soil pH was assessed in a slurry of 1:1 soil:water by mass after

10 min of rest. On baseline soils, soil texture was assessed using the

hydrometer method to determine clay content, and mass recovery of

sand particles (

>

53

μ

m) was used to determine sand content. Soil

exchangeable cations were extracted in ammonium acetate (

49

).

Briefly,

~

2.5 g of 2- mm sieved, air- dried soil was shaken for 30 min

with 25 ml of 1 M ammonium acetate at pH 7 and then filtered

Fig. 4.

Simplified structural equation model showing controls on new MAOC

formation.

Predictor variables shown include saturation deficit (g c kg soil

−

1

) of

MAOc, microbial cUe, and existing MAOc (MAOc). Standardized path coefficients
were averaged across low and high c input treatments after multilevel SeM testing
indicated no significant difference between a constrained model (regression coef-
ficients forced to be the same between treatments) and an unconstrained model
(regression coefficients could vary) (

n

=

115). Blue paths represent significant posi-

tive relationships (

P

<

0.05); gray path represents nonsignificant relationship. the

model has cFi

>

0.95 and RMSeA

<

0.08.

Downloaded from https://www.science.org on October 19, 2026

King and Sokol

,

Sci. Adv.

11

, eadv9482 (2025) 13 June 2025

S c i e n c e A d vA n c e S

|

R e S e A R c h A R t i c l e

6 of 10

through a Whatman #1 filter paper (nominal pore size

=

11

μ

m) and

analyzed via inductively coupled plasma optical emission spectrom-

eter (ICP- OES; Optima 7300 DV, PerkinElmer, Waltham, MA, USA)

for Ca

2

+

. Soil Fe and Al were extracted via the acid ammonium oxa-

late method following (

50

), which estimates Fe and Al in amorphous

phases (referred to as Fe

O

and Al

O

). For the acid ammonium oxalate

extraction,

~

0.5 g of ground, air- dried soil was mixed with 30 ml of

0.175 M ammonium oxalate at pH 3 and shaken for 2 hours in the

dark and then filtered through a Whatman #1 filter paper. Calcareous

soils were identified by reaction with HCl, as described below, and

their carbonates removed before the acid ammonium oxalate extrac-

tion by reacting the sample with 30 ml of 1 M ammonium acetate at

pH 5.5. Ammonium oxalate extracts were also run via ICP- OES. All

analyses performed on air- dried soils are reported in oven- dry

equivalents. Soil carbon saturation deficit was estimated following

Georgiou

et al

. (

15

), as previously reported for these soils (

2

), and

described below.

Preparation of dual- labeled DOC

Ryegrass was produced in a dual isotope labeling chamber (

51

), in

which

13

C- CO

2

and

15

N fertilizer were continuously supplied to cre-

ate uniformly labeled tissues, where

13

C- enriched was

~

4 atomic

%

(at

%

). Although formation of MAOC is the primary focus of this

study, we leveraged the

15

N label to quantify the C:N ratio of newly

formed MAOC. To prepare a DOC solution representative of solu-

ble root inputs, both shoots and roots were clipped to

~

2 cm and

extracted in hot water for 3 hours at 100°C, with a ratio of 0.7 g dry

litter to 35 ml of deionized (DI) water (

52

). Shoot inputs were in-

cluded to generate sufficient plant mass for C additions throughout

the entire experiment. This solution was filtered through sequen-

tially finer filters (20, 0.22, and lastly 0.1

μ

m) to remove solids that

could have clogged rhizons, which had a 0.15-

μ

m pore size. Two

DOC solutions were prepared in more and less concentrated forms

to correspond to high or low C input treatments: DOC for high C

input microcosms contained 1.37 mg C ml solution

−

1

and DOC

for low C input microcosms contained 0.68 mg C ml solution

−

1

(analyzed for C concentration on a Shimadzu–total organic carbon

analyzer for nonpurgeable organic carbon). The DOC solution was

freeze- dried and ground in a mortar and pestle prior to running on

via Elemental Analyzer–Isotope Ratio Mass Spectrometry (EA- IRMS;

vario ISOTOPE cube, Elementar, Langenselbold, Germany), which

retuned a

δ

13

C of 2958‰ (per mil) and C:N ratio of 34. Both low

and high DOC solutions were partitioned into containers por-

tioned for a single day’s application of DOC, frozen, and thawed

on the day of use to ensure consistent DOC quality throughout

the experiment.

Soil microcosms

Microcosms were constructed from sheets of 1/4- inch (6.35- mm)

hard acrylic with neoprene sides, with 100- cm

3

internal volume. To

determine exact soil mass for each microcosm, soils from each plot

were first assessed for their water holding capacity (WHC; g water

retained/g soil after 2 hours allowed to drain while covered). Given

the highly variable WHC between soils (0.42 to 0.76 g water/g soil),

and the need to maintain soils at a moderate level of soil water (45

%

WHC) to maintain optimal microbial activity, soils could not be

packed into microcosms to a consistent bulk density across soils.

Microcosms were instead standardized to a consistent total mass

of soil

+

water

=

175 g. Bulk density therefore ranged from 1.30 to

1.47 g/cm

3

, representative of available information on bulk density

from these agricultural soils. Half of all microcosms were designated

as low C input, whereas the other half were designated as high C

input. In high C input microcosms, microrhizon samplers, 10 cm in

rhizon length, 2.5 mm in diameter, and 0.15-

μ

m pores (SoilMoisture

Equipment Corp., Goleta, CA), were inserted midway into the mi-

crocosm to deliver

13

C- DOC to a consistent location. Microcosms

were maintained at 45

%

WHC through biweekly additions of DI

water. Although fluctuations in water content of microcosms were

small, additions of DI water occurred at least 48 hours prior to any

gas sampling to avoid potential CO

2

flushes associated with wetting.

Addition of

13

C- DOC to soil microcosms

13

C- DOC, described above, was applied weekly to each microcosm

for 25 weeks. Before the first

13

C- DOC addition, soils were incubated

at 45

%

WHC and 25°C for 1 week to enable determination of natural

abundance

δ

13

C- CO

2

from each microcosm (described below). Nat-

ural abundance

δ

13

C- CO

2

from the first week could be used through-

out the study because changes in

δ

13

C due to incubation are

relatively small (

53

) compared to changes induced by addition highly

enriched substrates, as in this study. Soils were maintained at 45

%

WHC during the incubation to allow soils to be within optimal range

for microbial activity. Addition of

13

C- DOC to high C input micro-

cosms was accomplished through a syringe to the microrhizon sam-

pler to enable consistent labeling of the same soil area each week for

25 weeks. Upon DOC application, the area of soil wetted was ob-

served and manually outlined with a marker onto both sides of the

acrylic surface to precisely trace DOC application area. For low C

input microcosms, custom needles, 140 mm in length (Hamilton

Company, Reno, NV), were inserted to a different location within

the microcosms each week to allow for diffuse

13

C- DOC labeling

across the entire volume of the microcosm. A consistent volume of

13

C- DOC was added to low and high C input microcosms of the

same plot, with a consistent scaling per g soil resulting in 1.4 to 1.6 ml

of

13

C- DOC added to each microcosm each week. Total DOC added

throughout the incubation was scaled to g dry soil per microcosm

and therefore varied on a microcosm basis between 24.5 and 27.6 mg

C for low C input treatments and between 48.9 and 55.2 mg C for

high C input treatments. Application rates were determined in refer-

ence to a previous

13

C- labeling incubation work (

31

) and reflected

in situ rhizodeposition from an agricultural cover crop (

54

).

CO

2

and

δ

13

C- CO

2

measurements

CO

2

flux rates and

δ

13

C of CO

2

fluxes were quantified periodically

throughout the incubation. On the basis of pilot studies with high

frequency gas sampling from a subset of soils, we observed that the

flux of CO

2

from added

13

C- DOC was low within the first hour after

DOC addition and then increased rapidly to a peak 7 to 9 hours af-

ter DOC addition and subsequently declined. To describe the emis-

sion of

13

C- DOC at both “phases” of emission: (i) “peak” and (ii) the

low, steady state of emission we term “off- peak,” we sampled micro-

cosms in a consistent pattern at 7- week intervals throughout the

incubation (weeks 1, 7, 13, 19, and 25). After an initial gas sampling

on the same day of soil wetting and packing into microcosms (week 0),

microcosms were incubated for 1 week without any

13

C- DOC addi-

tion. Hereafter, all gas sampling bouts consisted of (i) an initial off-

peak gas sampling, followed within 4 min by

13

C- DOC addition,

(ii) a peak gas sampling, which was conducted 7 hours after initial

gas sampling for weeks 1, 7, 13, and 19 and 1 hour after

13

C- DOC

Downloaded from https://www.science.org on October 19, 2026

King and Sokol

,

Sci. Adv.

11

, eadv9482 (2025) 13 June 2025

S c i e n c e A d vA n c e S

|

R e S e A R c h A R t i c l e

7 of 10

addition in week 25, and (iii) a follow- up off peak gas sampling,

4 min prior to the

13

C- DOC addition for the following week.

Although the exact peak rate of

13

C- DOC emissions for each mi-

crocosm across the duration of the study could not be assessed, we

chose 7 hours after

13

C- DOC addition for the first four bouts of gas

sampling as a timing that could be integrated into a workday and

could also capture

13

C- DOC flux behavior near peak emission rate.

The interval between

13

C- DOC addition and gas sampling was

shortened to 1 hour on the last week of sampling under the intuition

that time to peak

13

C- DOC flux had shortened over the course of

the incubation, possibly due to microbial adaptation to the supplied

substrate. Because of the time- intensive nature of individual gas

samplings, only the first three blocks (of three to five) from each site

were selected for gas sampling throughout the study.

To conduct each gas sampling interval, an airtight, custom, acrylic

chamber was used to sequentially enclose a microcosm in a closed-

loop system with a Picarro G2101-

i

gas analyzer (Picarro Inc., Santa

Clara, CA). The internal dimensions of this gas sampling chamber

were 4 cm by 27 cm by 13.3 cm. For each gas sampling interval, the

microcosm was kept in the sampling chamber for 240 s, while the

soil in the microcosm, which was open to the chamber, respired.

Of the 240- s sampling interval, 130 s was used to calculate both the

CO

2

flux rate and the

δ

13

C of CO

2

flux. After each sampling interval,

the gas analyzer was allowed to flush built- up CO

2

by sampling am-

bient air for 4 min. Concentrations of CO

2

in ambient air at the ini-

tiation of gas sampling intervals did not exceed 500 ppm (parts per

million). The proportion of CO

2

flux from

13

C- DOC versus existing

SOC,

f

A

, was calculated using a two- pool mixing model

where

δ

S

is the intercept of the Keeling plot for a given sampling

point after

13

C- DOC addition,

δ

B

is the average intercept of two

Keeling plots from the first two gas sampling intervals, both of

which occurred before any

13

C- DOC had been applied, and

δ

A

is the

δ

13

C of DOC, determined through EA- IRMS analysis on freeze-

dried

13

C- DOC.

Destructive harvest of microcosms

After the 25th weekly addition of

13

C- DOC, microcosms were al-

lowed to incubate for a final week and the bout of gas sampling oc-

curring over week 25 was completed before destructive harvest.

Small losses of water were corrected for by bringing microcosms to

45

%

WHC. For low C input microcosms, all soil was immediately

passed through an 8- mm sieved, and to minimize time spent sieving,

a subsample was 2- mm sieved and kept fresh for assays described

below. For high C input microcosms, customized templates for each

microcosm were produced to help delineate concentration area of C

application identified during

13

C- DOC labeling. Soils labeled with

13

C- DOC were gently separated from nonlabeled soils, and the en-

tirety of labeled soils were 8- mm sieved and then 2- mm sieved. Soils

sieved to 2 mm from low C input and high C input microcosms were

handled in the same manner for the CUE (including microbial bio-

mass) and SOC fractionation procedures described below.

Microbial CUE

Within 48 hours after destructive harvest of microcosms, 2- mm

sieved soils were assayed for microbial CUE using the

18

O labeling

method (

37

). This approach provides a substrate- independent assess-

ment of microbial necromass production and CUE and involves

microbial biomass measurements,

18

O labeling incubation with res-

piration measurement, followed by DNA extraction and quantification.

Microbial biomass via chloroform fumigation

From each microcosm,

~

5 g of fresh soil was weighed out twice,

for fumigated and unfumigated aliquots (

55

). Unfumigated sam-

ples were immediately extracted with 25 ml of 0.5 M K

2

SO

4

by

shaking for 2 hours and then filtered through Whatman #42 fil-

ter papers. Fumigated samples were subjected to chloroform fu-

migation for 72 hours prior to the same extraction procedure

with 0.5 M K

2

SO

4

. All K

2

SO

4

extracts were acidified with three

drops of phosphoric acid prior to freezing. Frozen extracts were

subsequently freeze- dried prior to analysis via EA- IRMS (de-

scribed above) for quantification of microbial biomass C and

δ

13

C. Microbial biomass was divided by a correction factor of

0.45 (

55

).

18

O labeling and CUE respiration

From each microcosm, two aliquots of

~

500 mg of fresh, 2- mm

sieved soil were weighed into microcentrifuge tubes. Soil in one mi-

crocentrifuge tube was injected with a dilution of 97 at

%

18

O-

enriched water, calculated to bring soil to 60

%

WHC and final soil

water to

~

20

18

O at

%

. The other microcentrifuge received an equiv-

alent amount of natural abundance Milli- Q water. Immediately fol-

lowing addition of water (

18

O- enriched or natural abundance), each

microcentrifuge tube was placed in a 20- ml serum bottle and sealed

with aluminum crimp cap and rubber stopper. Every 10 samples, a

blank serum vial was sealed for estimation of baseline CO

2

concen-

trations. All samples were incubated in the dark for 72 hours at

25°C, at which point CO

2

headspace concentrations of

18

O- enriched

vials were measured on an LI 7810 with a 5- ml injection, and all soil

samples immediately frozen in liquid nitrogen and stored at

−

80°C

until DNA extraction.

Soil DNA was extracted using the FastDNA SPIN Kit for Soil,

with slight modifications from the manufacturer’s instructions: Cen-

trifugation after bead- beating was extended to 15 min; the whole of

the binding matrix suspension was transferred to the SPIN filter, and

molecular biology grade water was used in the elution step. Given

the low variability of

18

O in natural abundance DNA controls, a ran-

domized subset of 10

%

of control microcentrifuge tubes was ana-

lyzed to determine natural abundance of

18

O in DNA. The DNA in

extracts were quantified by fluorescence stain (Quant- iT PicoGreen

dsDNA Reagent) and assessed using a microplate reader, with stan-

dards and samples in triplicate.

18

O concentrations in DNA were

quantified by drying DNA extract at 60°C in silver capsules for

TC/EA- IRMS analysis (UC Davis Stable Isotope Facility). As the

18

O- CUE method assumes that microbes use only water- derived

oxygen for their DNA synthesis, it may underestimate CUE [see (

56

)

for further discussion].

Given information on microbial biomass (MBC), and informa-

tion on respiration, which is based on the amount of CO

2

- C pro-

duced during the incubation, and information on growth, which is

based on the amount

18

O uptake in DNA total C uptake, CUE was

calculated as

f

A

=

(

δ

S

− δ

B

)

(

δ

A

− δ

B

)

C

Uptake

=

C

Growth

+

C

Respiration

CUE

=

C

Growth

C

Growth

+

C

Respiration

Downloaded from https://www.science.org on October 19, 2026

King and Sokol

,

Sci. Adv.

11

, eadv9482 (2025) 13 June 2025

S c i e n c e A d vA n c e S

|

R e S e A R c h A R t i c l e

8 of 10

Given the assumption of steady- state conditions during the CUE

incubation (i.e., growth

=

death),

C

Growth

is interpreted as a rate of

microbial necromass production, expressed as unit C per g dry soil

hour

−

1

. For further discussion of CUE assumptions and interpreta-

tion, see (

56

).

Soil fractionation and formation efficiency of MAOC

SOC in baseline (preincubation) soils was assessed and reported in

). The same size fractionation method was applied to incubat-

ed soils. Soils were separated by size into sand

+

POC (

>

53

μ

m) and

silt

+

clay

+

MAOC (

<

53

μ

m) following (

57

). These fractions are

hereafter referred to as “POC” and “MAOC,” respectively. Briefly,

5.75 to 6.25 g of 2- mm sieved bulk soil dried at 60°C was shaken for

18 hours with 12 glass beads in 30 ml of 0.5

%

sodium hexameta-

phosphate to disrupt all aggregates. The resulting soil slurry was

rinsed with DI water over a 53-

μ

m sieve to isolate POC and remove

glass beads. Soil solution passing through the sieve was deemed

MAOC. Both the POC and MAOC fractions were dried at 60°C un-

til reaching constant mass. Recoveries of the initial soil masses in

the summed fractions were between 96 and 105

%

for all samples,

with a mean recovery of 101.7

%

. Soils and fractions containing car-

bonates (identified by effervescence after addition of 5

%

HCl, 36

samples) were treated to remove inorganic carbon via HCl fumigation

(

58

). SOC, MAOC, and POC were ground using a mortar and pestle

before running on an EA- IRMS (Elemental Analyzer–Isotope Ratio

Mass Spectrometry; vario ISOTOPE cube, Elementar, Langenselbold,

Germany). The average recovery of SOC in POC and MAOC frac-

tions was 89

%

.

A two- pool mixing model was used to estimate the proportion of

C in SOC, POC, and MAOC derived from the added

13

C- DOC

where

f

A

is the fraction C from

13

C- DOC in the C pool,

δ

13

C

incubated sample

is the

δ

value of that C pool after incubation,

δ

13

C

baseline sample

is the

δ

value of that C pool before incubation,

and

δ

13

C

-

DOC

is the

δ

value of the added DOC. Each

f

A

was mul-

tiplied by the g C kg soil

−

1

on the incubation fraction to arrive at

a quantity of DOC- derived C per mass soil. The recovery of the

13

C- DOC in the unlabeled portion of the high C input microcosms

was, on average, 4

%

of the applied label. Incorporating the unlabeled

area of high C input treatments to calculations of MAOC formation

efficiency increased MAOC formation efficiency from 22.6 to 25.6

%

and did not change comparison of C input treatment effects. We

therefore report formation efficiencies from the labeled area of high

C input treatments. Formation efficiency of MAOC was calculated as

where

ME

is MAOC formation efficiency,

f

A

is the fraction of MAOC

from

13

C- DOC (defined above),

MAOC

is the total mass of labeled

MAOC per microcosm, and

DOC

is the total mass of

13

C- DOC

applied per microcosm.

Estimation of peak and off- peak

13

C- DOC respiration

Total respiration of

13

C- DOC during off- peak respiration was esti-

mated by integrating the area under the gas curve, using the 10 off-

peak measurements for each microcosm. We first fit Michaelis- Menten

curves to describe off- peak respiration. Of 180 microcosms gas sam-

pled, 167 could be fitted with Michaelis- Menten functions; another

6 microcosms were suitably fit with linear regression models based

on visual inspection. Gas sampling series from 7 microcosms were

deemed of insufficient quality for integration and removed from sub-

sequent analyses, leaving 173 microcosms for gas sampling analyses.

Given the estimate of total off- peak respiration per microcosm, a

known quantity of

13

C- DOC added, and an estimate of

13

C- DOC

recovered in soil, respiratory losses of

13

C- DOC during peak fluxes

were calculated as the difference between

13

C- DOC added and

13

C- DOC otherwise accounted for (soil

+

off- peak respiration).

Estimation of MAOC saturation and saturation deficit

Following previous works (

2

,

15

), we estimated the theoretical C

saturation of the MAOC fraction in our soils. Briefly, to derive the

boundary line from the data in (

15

), we used the observations for g

MAOC kg soil

−

1

as a function of silt

+

clay, first identifying observa-

tions in the 95th percentile or higher MAOC concentration in each

of 10 segments of silt

+

clay content (i.e., the first segment of obser-

vations with

<

10

%

silt

+

clay, the next segment of observations with

greater than or equal to 10

%

silt

+

clay and

<

20

%

, etc.). Using sim-

ple linear regression of these boundary points with the intercept

forced through zero, on the assumption that soils without silt

+

clay

have no MAOC, we estimated the boundary line slope of MAOC

~

silt

+

clay to be 0.7323. To calculate theoretical MAOC saturation in

g MAOC kg soil

−

1

for all soils, we multiplied silt

+

clay by 0.7323

(our boundary line slope). We calculated the absolute MAOC satu-

ration deficit by subtracting observed g MAOC kg soil

−

1

from theo-

retical g MAOC kg soil

−

1

at saturation.

Statistical analyses

We used multilevel, two- sided Pearson’s correlations with 95

%

confi-

dence levels to assess linear relationships between MAOC formation

(mg

13

C- DOC- derived MAOC kg soil

−

1

) and soil physicochemical

properties. Before all analyses, a uniform outlier cleaning procedure was

applied. Individual data points were flagged as outliers and removed

from subsequent analyses if they fell below the 25th percentile minus 1.5

times the IQR or above the 75th percentile plus 1.5 times the IQR.

We used structural equation modeling to assess inter- related driv-

ers of new MAOC formation in an initial full model and then in a

simplified model. In both cases, we avoided including soil silt

+

clay

contents and saturation deficit in the same model due to their high

correlation (

r

=

0.9). To assess whether relationships between variables

differed between high and low C input treatments, we standardized

variables included in the model and then compared multilevel SEM,

where high and low C input data were each assigned a level. An uncon-

strained model, where regression coefficients were freely estimated,

was first compared to a constrained model, where these coefficients

were set to be equal across groups. This comparison, assessed through

a chi- square difference test, aimed to determine whether the regres-

sion relationships differ significantly between groups. The resulting

P

values of

>

0.5 for the analysis of variance (ANOVA) of chi- square val-

ues indicated no significant difference between the constrained and

unconstrained models, suggesting that the relationships among driv-

ers of MAOC formation were consistent across C input levels. All

structural equation models met model fit indices (

59

) and were created

using global estimation in the lavaan package in R (

60

).

We tested for differences in the mean MAOC formation efficien-

cy,

%

13

C- DOC respired during peak, microbial CUE, and microbial

f

A

=

(

δ

13

C

incubated sample

− δ

13

C

baseline sample

)

(

δ

13

C

-

DOC

− δ

13

C

baseline sample

)

ME

=

f

A

MAOC

(

1

DOC

)

Downloaded from https://www.science.org on October 19, 2026

King and Sokol

,

Sci. Adv.

11

, eadv9482 (2025) 13 June 2025

S c i e n c e A d vA n c e S

|

R e S e A R c h A R t i c l e

9 of 10

growth between high and low C input treatments using Wilcoxon

rank sum tests. A nonparametric test, like Wilcoxon rank sum, was

preferred over a

t

test as Shapiro- Wilk tests and Q- Q plots estab-

lished that all pairs of comparisons had at least one set of non-

normally distributed data.

Supplementary Materials

This PDF file includes:

Figs. S1 to S9
tables S1 and S2

REFERENCES AND NOTES

1. n. W. Sokol, e. d. Whalen, A. Jilling, c. Kallenbach, J. Pett- Ridge, K. Georgiou, Global

distribution, formation and fate of mineral- associated soil organic matter under a
changing climate: A trait- based perspective.

Funct. Ecol.

36

, 1411–1429 (2022).

2. A. e. King, J. P. Amsili, S. c. córdova, S. culman, S. J. Fonte, J. Kotcon, M. d. Masters,

K. Mcvay, d. c. Olk, A. M. Prairie, M. Schipanski, S. K. Schneider, c. e. Stewart, M. F. cotrufo,
constraints on mineral- associated and particulate organic carbon response to
regenerative management: carbon inputs and saturation deficit.

Soil Till. Res.

238

,

106008 (2024).

3. B. Minasny, B. P. Malone, A. B. McBratney, d. A. Angers, d. Arrouays, A. chambers,

v. chaplot, Z.- S. chen, K. cheng, B. S. das, d. J. Field, A. Gimona, c. B. hedley, S. Y. hong,
B. Mandal, B. P. Marchant, M. Martin, B. G. Mcconkey, v. l. Mulder, S. O’Rourke,
A. c. Richer- de- Forges, i. Odeh, J. Padarian, K. Paustian, G. Pan, l. Poggio, i. Savin,
v. Stolbovoy, U. Stockmann, Y. Sulaeman, c.- c. tsui, t.- G. vågen, B. van Wesemael,
l. Winowiecki, Soil carbon 4 per mille.

Geoderma

292

, 59–86 (2017).

4. e. e. Oldfield, M. A. Bradford, S. A. Wood, Global meta- analysis of the relationship

between soil organic matter and crop yields.

SOIL

5

, 15–32 (2019).

5. d. A. Kane, M. A. Bradford, e. Fuller, e. e. Oldfield, S. A. Wood, Soil organic matter protects

US maize yields and lowers crop insurance payouts under drought.

Environ. Res. Lett.

16

,

044018 (2021).

6. M. F. cotrufo, M. d. Wallenstein, c. M. Boot, K. denef, e. Paul, the Microbial efficiency-

Matrix Stabilization (MeMS) framework integrates plant litter decomposition with soil
organic matter stabilization: do labile plant inputs form stable soil organic matter?

Glob. Chang. Biol.

19

, 988–995 (2013).

7. M. J. castellano, K. e. Mueller, d. c. Olk, J. e. Sawyer, J. Six, integrating plant litter quality,

soil organic matter stabilization, and the carbon saturation concept.

Glob. Chang. Biol.

21

,

3200–3209 (2015).

8. J. hassink, the capacity of soils to preserve organic c and n by their association with clay

and silt particles.

Plant Soil

191

, 77–87 (1997).

9. J. hassink, effects of soil texture and grassland management on soil organic c and n and

rates of c and n mineralization.

Soil Biol. Biochem.

26

, 1221–1231 (1994).

10. l. M. Mayer, Relationships between mineral surfaces and organic carbon concentrations

in soils and sediments.

Chem. Geol.

114

, 347–363 (1994).

11. J. hassink, A. P. Whitmore, A model of the physical protection of organic matter in soils.

Soil Sci. Soc. Am. J.

61

, 131–139 (1997).

12. R. Z. Abramoff, K. Georgiou, B. Guenet, M. S. torn, Y. huang, h. Zhang, W. Feng,

S. Jagadamma, K. Kaiser, d. Kothawala, M. A. Mayes, P. ciais, how much carbon can be
added to soil by sorption?

Biogeochemistry

152

, 127–142 (2021).

13. n. Begill, A. don, c. Poeplau, no detectable upper limit of mineral- associated organic

carbon in temperate agricultural soils.

Glob. Chang. Biol.

29

, 4662–4669 (2023).

14. W. Feng, A. F. Plante, J. Six, improving estimates of maximal organic carbon stabilization

by fine soil particles.

Biogeochemistry

112

, 81–93 (2013).

15. K. Georgiou, R. B. Jackson, O. vindušková, R. Z. Abramoff, A. Ahlström, W. Feng,

J. W. harden, A. F. A. Pellegrini, h. W. Polley, J. l. Soong, W. J. Riley, M. S. torn, Global stocks
and capacity of mineral- associated soil organic carbon.

Nat. Commun.

13

, 3797 (2022).

16. J. Six, S. doetterl, M. laub, c. R. Müller, M. van de Broek, the six rights of how and when

to test for soil c saturation.

SOIL

10

, 275–279 (2024).

17. A. R. Saidy, R. J. Smernik, J. A. Baldock, K. Kaiser, J. Sanderman, the sorption of organic

carbon onto differing clay minerals in the presence and absence of hydrous iron oxide.

Geoderma

209–210

, 15–21 (2013).

18. A. e. King, J. P. Amsili, S. c. córdova, S. culman, S. J. Fonte, J. Kotcon, M. liebig,

M. d. Masters, K. M. vay, d. c. Olk, M. Schipanski, S. K. Schneider, c. e. Stewart,
M. F. cotrufo, A soil matrix capacity index to predict mineral- associated but not
particulate organic carbon across a range of climate and soil ph.

Biogeochemistry

165

,

1–14 (2023).

19. J. Six, R. t. conant, e. A. Paul, K. Paustian, Stabilization mechanisms of soil organic matter:

implications for c- saturation of soils.

Plant Soil

241

, 155–176 (2002).

20. J. Kang, c. Qu, W. chen, P. cai, c. chen, Q. huang, Organo–organic interactions

dominantly drive soil organic carbon accrual.

Glob. Chang. Biol.

30

, e17147 (2024).

21. A. R. Possinger, M. J. Zachman, A. enders, B. d. A. levin, d. A. Muller, l. F. Kourkoutis,

J. lehmann, Organo–organic and organo–mineral interfaces in soil at the nanometer
scale.

Nat. Commun.

11

, 6103 (2020).

22. S. A. Schweizer, c. W. Mueller, c. höschen, P. ivanov, i. Kögel- Knabner, the role of clay

content and mineral surface area for soil organic carbon storage in an arable
toposequence.

Biogeochemistry

156

, 401–420 (2021).

23. d. S. e. Bramble, S. Ulrich, i. Schöning, R. Mikutta, l. Brandt, c. Poll, e. Kandeler, c. Mikutta,

A. Konrad, J. Siemens, Y. Yang, A. Polle, P. Schall, c. Ammer, K. Kaiser, M. Schrumpf,
Formation of mineral- associated organic matter in temperate soils is primarily controlled
by mineral type and modified by land use and management intensity.

Glob. Chang. Biol.

30

, e17024 (2024).

24. d. M. O. elias, K. e. Mason, t. Goodall, A. taylor, P. Zhao, A. Otero- Fariña, h. chen,

c. l. Peacock, n. J. Ostle, R. Griffiths, P. J. chapman, J. holden, S. Banwart, n. P. Mcnamara,
J. Whitaker, Microbial and mineral interactions decouple litter quality from soil organic
matter formation.

Nat. Commun.

15

, 10063 (2024).

25. c. liang, J. P. Schimel, J. d. Jastrow, the importance of anabolism in microbial control over

soil carbon storage.

Nat. Microbiol.

2

, 17105 (2017).

26. c. M. Kallenbach, A. S. Grandy, S. d. Frey, A. F. diefendorf, Microbial physiology and

necromass regulate agricultural soil carbon accumulation.

Soil Biol. Biochem.

91

, 279–290

(2015).

27. F. tao, Y. huang, B. A. hungate, S. Manzoni, S. d. Frey, M. W. i. Schmidt, M. Reichstein,

n. carvalhais, P. ciais, l. Jiang, J. lehmann, Y.- P. Wang, B. Z. houlton, B. Ahrens, U. Mishra,
G. hugelius, t. d. hocking, X. lu, Z. Shi, K. viatkin, R. vargas, Y. Yigini, c. Omuto, A. A. Malik,
G. Peralta, R. cuevas- corona, l. e. di Paolo, i. luotto, c. liao, Y.- S. liang, v. S. Saynes,
X. huang, Y. luo, Microbial carbon use efficiency promotes global soil carbon storage.

Nature

618

, 981–985 (2023).

28. M. e. craig, K. M. Geyer, K. v. Beidler, e. R. Brzostek, S. d. Frey, A. Stuart Grandy, c. liang,

R. P. Phillips, Fast- decaying plant litter enhances soil carbon in temperate forests but not
through microbial physiological traits.

Nat. Commun.

13

, 1229 (2022).

29. n. W. Sokol, M. M. Foley, S. J. Blazewicz, A. Bhattacharyya, n. didonato, K. estera- Molina,

M. Firestone, A. Greenlon, B. A. hungate, J. Kimbrel, J. liquet, M. lafler, M. Marple,
P. S. nico, l. Paša- tolić, e. Slessarev, J. Pett- Ridge, the path from root input to mineral-
associated soil carbon is dictated by habitat- specific microbial traits and soil moisture.

Soil Biol. Biochem.

193

, 109367 (2024).

30. e. d. Whalen, A. S. Grandy, K. M. Geyer, e. W. Morrison, S. d. Frey, Microbial trait

multifunctionality drives soil organic matter formation potential.

Nat. Commun.

15

,

10209 (2024).

31. n. W. Sokol, M. A. Bradford, Microbial formation of stable soil carbon is more efficient

from belowground than aboveground input.

Nat. Geosci.

12

, 46–53 (2019).

32. M. l. haddix, e. G. Gregorich, B. l. helgason, h. Janzen, B. h. ellert, M. Francesca cotrufo,

climate, carbon content, and soil texture control the independent formation and
persistence of particulate and mineral- associated organic matter in soil.

Geoderma

363

,

114160 (2020).

33. M. F. cotrufo, J. l. Soong, A. J. horton, e. e. campbell, M. l. haddix, d. h. Wall, W. J. Parton,

Formation of soil organic matter via biochemical and physical pathways of litter mass
loss.

Nat. Geosci.

8

, 776–779 (2015).

34. c. Rasmussen, K. heckman, W. R. Wieder, M. Keiluweit, c. R. lawrence, A. A. Berhe,

J. c. Blankinship, S. e. crow, J. l. druhan, c. e. hicks Pries, e. Marin- Spiotta, A. F. Plante,
c. Schädel, J. P. Schimel, c. A. Sierra, A. thompson, R. Wagai, Beyond clay: towards an
improved set of variables for predicting soil organic matter content.

Biogeochemistry

137

, 297–306 (2018).

35. M. c. Rowley, S. Grand, É. P. verrecchia, calcium- mediated stabilisation of soil organic

carbon.

Biogeochemistry

137

, 27–49 (2018).

36. M. Kleber, P. Sollins, R. Sutton, A conceptual model of organo- mineral interactions in soils:

Self- assembly of organic molecular fragments into zonal structures on mineral surfaces.

Biogeochemistry

85

, 9–24 (2007).

37. M. Spohn, K. Klaus, W. Wanek, A. Richter, Microbial carbon use efficiency and biomass

turnover times depending on soil depth—implications for carbon cycling.

Soil Biol. Biochem.

96

, 74–81 (2016).

38. c. M. Kallenbach, S. d. Frey, A. S. Grandy, direct evidence for microbial- derived soil

organic matter formation and its ecophysiological controls.

Nat. Commun.

7

, 1–10 (2016).

39. J. e. hobbie, Microbes in nature are limited by carbon and energy: the starving- survival

lifestyle in soil and consequences for estimating microbial rates.

Front. Microbiol.

4

, 1–11

(2013).

40. K. Schneckenberger, d. demin, K. Stahr, Y. Kuzyakov, Microbial utilization and

mineralization of [

14

c]glucose added in six orders of concentration to soil.

Soil Biol. Biochem.

40

, 1981–1988 (2008).

41. e. d. Whalen, A. S. Grandy, n. W. Sokol, M. Keiluweit, J. ernakovich, R. G. Smith, S. d. Frey,

clarifying the evidence for microbial- and plant- derived soil organic matter, and the path
toward a more quantitative understanding.

Glob. Chang. Biol.

28

, 7167–7185 (2022).

Downloaded from https://www.science.org on October 19, 2026

King and Sokol

,

Sci. Adv.

11

, eadv9482 (2025) 13 June 2025

S c i e n c e A d vA n c e S

|

R e S e A R c h A R t i c l e

10 of 10

42. Y. chang, n. W. Sokol, K. J. van Groenigen, M. A. Bradford, d. Ji, t. W. crowther, c. liang,

Y. luo, Y. Kuzyakov, J. Wang, F. ding, A stoichiometric approach to estimate sources of
mineral- associated soil organic matter.

Glob. Chang. Biol.

30

, e17092 (2024).

43. S. d. Frey, J. lee, J. M. Melillo, J. Six, the temperature response of soil microbial efficiency

and its feedback to climate.

Nat. Clim. Change

3

, 395–398 (2013).

44. M. e. craig, M. A. Mayes, B. n. Sulman, A. P. Walker, Biological mechanisms may contribute

to soil carbon saturation patterns.

Glob. Chang. Biol.

27

, 2633–2644 (2021).

45. S. Gulde, h. chung, W. Amelung, c. chang, J. Six, Soil carbon saturation controls labile

and stable carbon pool dynamics.

Soil Sci. Soc. Am. J.

72

, 605–612 (2008).

46. M. Schrumpf, K. Kaiser, large differences in estimates of soil organic carbon turnover in

density fractions by using single and repeated radiocarbon inventories.

Geoderma

239- 240

, 168–178 (2015).

47. F. Bastida, c. García, n. Fierer, d. J. eldridge, M. A. Bowker, S. Abades, F. d. Alfaro,

A. Asefaw Berhe, n. A. cutler, A. Gallardo, l. García- velázquez, S. c. hart, P. e. hayes,
t. hernández, Z.- Y. hseu, n. Jehmlich, M. Kirchmair, h. lambers, S. neuhauser,
v. M. Peña- Ramírez, c. A. Pérez, S. c. Reed, F. Santos, c. Siebe, B. W. Sullivan, P. trivedi,
A. vera, M. A. Williams, J. luis Moreno, M. delgado- Baquerizo, Global ecological predictors
of the soil priming effect.

Nat. Commun.

10

, 3481 (2019).

48. n. W. Sokol, e. Slessarev, G. l. Marschmann, A. nicolas, S. J. Blazewicz, e. l. Brodie,

M. K. Firestone, M. M. Foley, R. hestrin, B. A. hungate, B. J. Koch, B. W. Stone, M. B. Sullivan,
O. Zablocki, llnl Soil Microbiome consortium, J. Pett- Ridge, life and death in the soil
microbiome: how ecological processes influence biogeochemistry.

Nat. Rev. Microbiol.

20

, 415–430 (2022).

49. G. W. thomas, “exchangeable cations” in

Methods of Soil Analysis, Part 2

(Wiley, 1982),

vol. 9, pp. 159–165.

50. R. h. loeppert, W. P. inskeep, “iron” in

Methods of Soil Analysis, Part 3. Chemical Methods

(Soil Science Society of America and American Society of Agronomy, 1996), pp. 639–664.

51. J. l. Soong, d. Reuss, c. Pinney, t. Boyack, M. l. haddix, c. e. Stewart, M. F. cotrufo, design

and operation of a continuous

13

c and

15

n labeling chamber for uniform or differential,

metabolic and structural, plant isotope labeling.

J. Vis. Exp.

83

, 1–8 (2014).

52. technical Association of the Pulp and Paper industry (tAPPi), Water solubility of wood

and pulp. test method t204 (or 207) (tAPPi, 1981).

53. R. F. Follett, e. A. Paul, e. G. Pruessner, Soil carbon dynamics during a long- term

incubation study involving

13

c and

14

c measurements.

Soil Sci.

172

, 189–208 (2007).

54. e. e. Austin, K. Wickings, M. d. Mcdaniel, G. P. Robertson, A. S. Grandy, cover crop root

contributions to soil carbon in a no- till corn bioenergy cropping system.

GCB Bioenergy

9

,

1252–1263 (2017).

55. e. d. vance, P. c. Brookes, d. S. Jenkinson, An extraction method for measuring soil

microbial biomass c.

Soil Biol. Biochem.

19

, 703–707 (1987).

56. K. M. Geyer, P. dijkstra, R. Sinsabaugh, S. d. Frey, clarifying the interpretation of carbon

use efficiency in soil through methods comparison.

Soil Biol. Biochem.

128

, 79–88 (2019).

57. c. A. cambardella, e. t. elliott, Particulate soil organic- matter changes across a grassland

cultivation sequence.

Soil Sci. Soc. Am. J.

56

, 777–783 (1992).

58. d. harris, W. R. horwáth, c. van Kessel, Acid fumigation of soils to remove carbonates

prior to total organic carbon or cARBOn- 13 isotopic analysis.

Soil Sci. Soc. Am. J.

65

,

1853–1856 (2001).

59. d. hooper, J. coughlan, M. Mullen, “evaluating model fit: A synthesis of the structural

equation modelling literature,” in

7th European Conference on Research Methodology for

Business and Management Studies

(eRcM, 2008), pp. 195–200.

60. Y. Rosseel, lavaan: An R package for structural equation modeling.

J. Stat. Softw.

48

, 1–36

(2012).

61. M. Wiesmeier, l. Urbanski, e. hobley, B. lang, M. von lützow, e. Marin- Spiotta,

B. van Wesemael, e. Rabot, M. ließ, n. Garcia- Franco, U. Wollschläger, h. J. vogel,
i. Kögel- Knabner, Soil organic carbon storage as a key function of soils—A review of
drivers and indicators at various scales.

Geoderma

333

, 149–162 (2019).

Acknowledgments:

We thank members of the colorado State University Soil innovation

laboratory, particularly e. Riley, for logistical support, as well as M. Kleber for feedback on an
early version of the manuscript and K. esquivel for statistical input. A.e.K. was supported by the
national institute of Food and Agriculture Postdoctoral Fellowship 202067034- 31762. Work at
llnl by n.W.S. was supported by the dOe Office of Biological and environmental Research,
Genomic Science Program “Microbes Persist” Scientific Focus Area, #ScW1632, and performed
under the auspices of US dOe under contract de- Ac52- 07nA27344.

Funding:

this work was

supported by the national institute of Food and Agriculture, 202067034- 31762 (A.e.K.) and
department of energy, Office of Biological and environmental Research, Genomic Science
Program “Microbes Persist” Scientific Focus Area, #ScW1632 (n.W.S.).

Author contributions:

conceptualization: A.e.K. and n.W.S. Funding acquisition: A.e.K. Methodology: n.W.S.
investigation: A.e.K. Supervision: A.e.K. Project administration: A.e.K. Formal analysis: A.e.K.
visualization: A.e.K. and n.W.S. Writing—original draft: A.e.K. Writing—review and editing:
A.e.K. and n.W.S.

Competing interests:

the authors declare that they have no competing

interests.

Data and materials availability:

All data needed to evaluate the conclusions in the

paper are present in the paper and/or the Supplementary Materials. data used in analysis are
archived in a Figshare repository (10.6084/m9.figshare.28791830).

Submitted 13 January 2025
Accepted 8 May 2025
Published 13 June 2025
10.1126/sciadv.adv9482

Downloaded from https://www.science.org on October 19, 2026

Soil carbon formation is promoted by saturation deficit and existing mineral-
associated carbon, not by microbial carbon-use efficiency

Alison E. King and Noah W. Sokol

Sci. Adv. 11 (24), eadv9482. DOI: 10.1126/sciadv.adv9482

View the article online

https://www.science.org/doi/10.1126/sciadv.adv9482

Permissions

https://www.science.org/help/reprints-and-permissions

Use of this article is subject to the

Terms of service

Science Advances (ISSN 2375-2548) is published by the American Association for the Advancement of Science. 1200 New York Avenue
NW, Washington, DC 20005. The title Science Advances is a registered trademark of AAAS.

Copyright © 2025 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim
to original U.S. Government Works. Distributed under a Creative Commons Attribution License 4.0 (CC BY).

Downloaded from https://www.science.org on October 19, 2026

King & Sokol (2025). Soil carbon formation is promoted by saturation deficit and existing mineral-associated carbon, not by microbial carbon-use efficiency

Page 1

Page 2

Page 3

Page 4

Page 5

Page 6

Page 7

Page 8

Page 9

Page 10

Page 11

Document Outline