Page 1
Nature Geoscience
| Volume 18 | July
2025 | 631–638
631
nature geoscience
https://doi.org/10.1038/s41561-025-01720-5
Article
Soil carbon accrual and crop production
enhanced by sustainable subsoil
management
Zheng-Rong Kan
1,9
, Zhenzhen Li
1,9
, Wulf Amelung
2,3
, Hai-Lin Zhang
4
,
Rattan Lal
5
, Roland Bol
3
, Xinmin Bian
1
, Jian Liu
6
, Yaguang Xue
6
,
Feng-Min Li
1,7,8
& Haishui Yang
1,7,8
Soil conservation practices such as no-till and straw mulching have been
practised worldwide, but they frequently show low potential to increase
organic carbon and crop grain production, especially in rice paddy systems.
Here a ditch-buried straw return technique is proposed to co-enhance soil
carbon and crop yield in paddy systems through injection of straw into the
subsoil. This technique can protect most of the surface soil and disturb only
10% of the whole field through deep tillage. A 15-year ditch-buried straw
return experiment in rice–wheat cropping has shown that compared to the
dominant rotary-tillage straw return management, grain yield was increased
under ditch-buried straw return by 15% without any additional fertilization
inputs. Ditch-buried straw return also increased soil organic carbon stocks at
0–40 cm depth by 46% (17.2 Mg ha
−1
) mainly through the enhanced conversion
of straw-derived carbon into mineral-associated fungal necromass. Overall,
ditch-buried straw return decreased net CO
2
equivalent emissions by 34% and
increased net economic benefits by 18%. Joint enhancements of soil organic
carbon and crop yield under ditch-buried straw return were further validated
using meta-analysis around China. We conclude that ditch-buried straw return
may work as an effective approach for subsoil management.
No-till (NT) with straw mulching has been well known for preventing
wind and water erosion in rain-fed agriculture
1
, with the additional
advantage of saving costs on labour and machinery
2
. However, a global
meta-analysis showed that such conservation agriculture techniques
can significantly decrease crop yield
3
, and similar findings were
also reported in China
4
and Europe
5
. Particularly for rice farming,
NT resulted even in larger yield declines than for other cereal
cropping
6
,
7
; consequently, conservation farming is scarcely adopted
in Southern China (Supplementary Fig. 1). Likewise, NT increases soil
organic carbon (SOC) content in the top layer but decreases it in the sub-
soil, thus contributing little to C sequestration
8
,
9
. What is more, paddy
rice management differs from that of upland cereal cropping, and the
context of wind or water erosion does not exist. When mulched with
crop straw under NT, the wet straw can even produce toxic substances
under anaerobic paddy conditions, which constrains rice growth and
the conversion of straw C into SOC
10
. Therefore, it remains imperative
Received: 5 July 2024
Accepted: 8 May 2025
Published online: 1 July 2025
Check for updates
1
College of Agriculture, Nanjing Agricultural University, Nanjing, China.
2
Institute of Crop Science and Resource Conservation (INRES), Soil Science and
Soil Ecology, University of Bonn, Bonn, Germany.
3
Institute of Bio- and Geosciences, Agrosphere Institute (IBG-3), Forschungszentrum Jülich, Jülich,
Germany.
4
College of Agronomy and Biotechnology, China Agricultural University, Beijing, China.
5
CFAES Rattan Lal Center for Carbon Management
and Sequestration, School of Environment and Natural Resources, The Ohio State University, Columbus, OH, USA.
6
Institute of Agricultural Sciences in
Yanjiang District of Jiangsu Province, Rugao, China.
7
Jiangsu Key Laboratory for Information Agriculture, Nanjing Agricultural University, Nanjing, China.
8
Jiangsu Collaborative Innovation Center for Modern Crop Production, Nanjing Agricultural University, Nanjing, China.
9
These authors contributed
equally: Zheng-Rong Kan, Zhenzhen Li. e-mail:
fmli@njau.edu.cn
;
yanghaishui@njau.edu.cn
Page 2
Nature Geoscience
| Volume 18 | July
2025 | 631–638
632
Article
https://doi.org/10.1038/s41561-025-01720-5
the 1980s in Southern China
6
. Thus, DB-SR can be regarded as a soil
conservation practice in paddy systems as it aligns with its basic prin-
ciples of minimizing soil disturbance to the maximum extent possible.
We hypothesize that the technique to bury crop straw into sub-
soil would not only be able to sequester SOC through feeding the
energy-limited microbes with fresh C but also elevate crop yields by
loosening the subsoil when alternating ditching positions.
Here we report on a 15-year (2008–2023) field experiment con-
ducted to: (1) assess the effects of DB-SR on SOC and crop yield; (2) elu-
cidate the underlying mechanisms of increase in SOC and the crop yield
and (3) evaluate the feasibility of DB-SR compared with rotary-tillage
straw return (RT-SR). To evaluate the generality of our findings, we
performed a meta-analysis with studies that more recently started
to test our technique, because our DB-SR paper was published in
2010
16
(with the concepts being communicated in 2008 at Shenyang
to develop a new conservation tillage strategy for co-enhancing SOC
and grain yield in paddy systems. Here, we suggest taking advantage
of the subsoil.
Subsoil still has more potential to sequester C due to its great
volume and high microbial C use efficiency
11
,
12
. However, the insuffi-
ciency of fresh C inputs limits the sequestration of SOC in the subsoil
13
.
Our team started in 2008 to experiment in burying crop straw into sub-
soil for enhancing SOC accumulation as feeding the starving microbes
and to gradually loosen soil through alternating ditching positions to
improve crop yields, termed as ditch-buried straw return (DB-SR)
14
(Fig.
1
). The land area under the ditch positions represents merely 10%
of the total field while the remaining 90% is under NT without puddling,
thus using energy very comparable to that under rotary tillage (RT)
15
.
The RT (about 10 cm depth) was used as a control because it has been
predominantly adopted in the paddy-based cropping regions since
6
9
5
4
10
6
10
4
8
2
5
9
3
7
1
6
8
2
3
7
1
1
7
3
9
5
2
8
4
10
Wheat straw
Rice straw
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Experiment
duration
(years)
, , and so on indicate the position of the first, second, third and so on ditches.
20 cm
200 cm
1 2 3
1
1
2
3
4
5
6
Positioning
Positioning
Ditching
Ditching
Burying
Burying
Repressing
Repressing
Covering
Covering
Repressing
Repressing
Positioning
10%
10%
Next season
a
b
c
Fig. 1 | Conceptual diagram for depicting DB-SR. a, Positions of straw ditches. b, Procedures for ditch-buried straw return. c, Machines for operating ditch-buried
straw return developed by Nantong Guangyi Electromechanical Co., Ltd in Jiangsu Province, China. The inset photo represents the ditching and straw burial devices
installed on the combine harvester.
Page 3
Nature Geoscience
| Volume 18 | July
2025 | 631–638
633
Article
https://doi.org/10.1038/s41561-025-01720-5
to the attendees of the Farming System Conference of China), to
broadly quantify the effects of DB-SR on SOC stock and crop yield
around China.
DB-SR substantially increased SOC stocks
The data from the 15-year field experiment showed that the imple-
mentation of DB-SR increased SOC stocks at 0–40 cm depth by 45.7%
(17.24 Mg ha
−1
, P < 0.01) relative to RT-SR based on equivalent soil mass
(Extended Data Fig. 1). The SOC concentrations of ten straw ditches
under DB-SR were all higher than RT-SR at 10–40 cm depth (P < 0.05;
Extended Data Fig. 2). The gain in SOC under DB-SR occurred mainly
at 10–20 cm and 20–30 cm soil depths, alongside a decrease in delta
carbon-13 isotope ratio (δ
13
C) values (Fig.
2a,b
). We attributed this find-
ing to three aspects. First, DB-SR buries straw into >20 cm soil depth,
which promotes SOC accumulation at 10–40 cm depth, whereas the
other tillage practices (for example NT; RT; plough tillage, PT) only
retain straw residues in the upper soil layers (0–15 cm). The simul-
taneous decline in soil δ
13
C values confirms that the detected SOC
largely originated from the crop straw with low
13
C content
17
. Second,
the deep injection of straw can improve the soil structure and pro-
mote soil infiltration, potentially facilitating deeper root growth and
the incorporation of root-derived C into the soil
15
. Third, the deeper
soil layers typically exhibit lower SOC content compared with topsoil,
thus possessing a higher ‘untapped’ potential for C sequestration
11
.
Furthermore, the gain in SOC was observed in both of the isolated
SOC fractions: DB-SR increased the particulate organic C (POC)
stock by 30.0% (4.69 Mg ha
−1
, P < 0.05; Extended Data Fig. 3a) and the
mineral-associated organic C (MAOC) stock by 56.8% (12.55 Mg ha
−1
,
P < 0.01; Fig.
2c
) at 0–40 cm depths. The larger gain in SOC thus
occurred as MAOC. Six et al.
18
proposed a simple calculation for estimat-
ing the maximum amount of C that can be retained by soil minerals, and
SOC saturability was 18.7% at 0–20 cm and only 6.1% at 20–40 cm depth
before the field experiment in 2008 (Supplementary Table 1). This low
saturation here confirms that there is a huge C sequestration potential
in the subsoil. Overall, these findings suggest that SOC sequestration
under DB-SR was sustainable, because any conversion of POC to MAOC
generally increases the residence times of SOC
19
. Besides, the soil fine
fraction (<0.053 mm) associated with MAOC is not yet ‘saturated’
18
, thus
still having the capacity to store further C from DB-SR management.
In general, saturation of MAOC is not reached in agricultural regions,
particularly not in subsoils, which can be attributed, at least in part,
to diminished rates of deep C inputs
20
,
21
. The DB-SR management thus
takes specific advantage of this effect: the inputs of straw C to horizons
below the tillage pan is an effective way to form stable MAOC through
microbial conversion and chemical interactions with minerals. It should
be noted, however, that our study adopted a size-based estimation of
MAOC. This method includes the risk that some fine POC attached to
minerals is also classified as MAOC. To avoid such risks, isolation of
MAOC can include a density-based separation, at best after ultrasonic
dispersion which additionally ensures aggregate breakdown
22
. Yet, our
findings show that microbial necromass C is the main component of
MAOC under DB-SR (Fig.
2d
).
2
4
6
8
10
2
4
6
8
10
−5
0
5
10
15
20
25
−4
0
4
8
12
16
−2
0
2
4
6
30–40
20–30
10–20
0–10
0 −5 −10 −15 −20 −25 −30 −35
0.004
0.001
< 0.001
0.003
30–40
20–30
10–20
0–10
0
3
6
9
12
15
18
0.674
0.011
0.009
0.026
Soil depth (cm)
30–40
20–30
10–20
0–10
0
1
2
3
4
0.053
0.001
0.003
0.027
BNC in MAOC
30–40
20–30
10–20
0–10
0
2
4
6
8
0.061
0.001
0.003
0.020
FNC in MAOC
δ
13
C
MAOC
30–40
20–30
10–20
0–10
0
4
8 12 16 20 24 28
0.121
<0.001
0.007
0.010
SOC
c
b
a
RT-SR
DB-SR
30–40
20–30
10–20
0–10
0
2
4
6
8
10
0.001
0.058
0.003
0.021
MNC in MAOC
f
e
d
g
i
h
SOC
DOC/(NH
4
+
+ NO
3
–
)
DOC/(NH
4
+
+ NO
3
–
)
2
4
6
8
10
DOC/(NH
4
+
+ NO
3
–
)
r
= –0.51*
r
= –0.70**
r
= –0.50*
MAOC
FNC in MAOC
Fig. 2 | Effects of DB-SR on SOC and its fraction stock. a, SOC stock (Mg ha
−1
).
b, δ
13
C abundance (‰). c, MAOC stock (Mg ha
−1
). d, Microbial necromass carbon
(MNC) stock in MAOC (Mg ha
−1
). e, Bacterial necromass carbon (BNC) stock in
MAOC (Mg ha
−1
). f, Fungal necromass carbon (FNC) stock in MAOC (Mg ha
−1
).
g–i, The relationship of DOC/(NH
4
+
+ NO
3
−
) to stocks (Mg ha
-1
) of SOC (g), MAOC
(h) and FNC in MAOC (i), respectively. Red and blue points represented DB-SR
and RT-SR, respectively. The stocks of SOC and its fractions were calculated
using equivalent soil mass. Data are presented with means ± standard error
across treatment replicates (n = 3). The comparisons in a–f were analysed using
two-sided paired-sample t-test. Values adjacent to paired bars denote P values for
differences between RT-SR and DB-SR treatments. The significance of the linear
regression model in d–h was assessed using F-test. *P < 0.05; **P < 0.01.
Page 4
Nature Geoscience
| Volume 18 | July
2025 | 631–638
634
Article
https://doi.org/10.1038/s41561-025-01720-5
Microbial necromass C (MNC) was the major component of
MAOC under DB-SR (Supplementary Table 2). Overall, the adoption
of DB-SR increased the MNC stock by 29.9% (2.25 Mg ha
−1
) in POC frac-
tions (P < 0.05; Extended Data Fig. 3b–d) and by 103.9% (9.19 Mg ha
−1
)
in MAOC ones (P < 0.05; Fig.
2d
). Specifically, the fungal necromass C
contributed more to POC and MAOC than bacterial necromass C (Sup-
plementary Table 2), and the MNC increase in MAOC mainly originated
from fungal necromass C under DB-SR (Fig.
2e,f
), as recently also found
for other organic matter injection into temperate rain-fed agriculture
23
.
The subsoil, characterized by a lower temperature than the topsoil,
is conducive to microbial anabolism
24
, and the cooler moist condi-
tions promote particularly the fungal growth
25
rather than leading to
an abundant C loss through respiration when the straw residues are
incorporated
24
,
26
. Fungi generally tend to have advantages in growth
when fresh crop residues are added to soil. Indeed, we previously
found that DB-SR exerted distinctive impacts on soil fungal composi-
tion, and 13–20% of fungal taxa were exclusively promoted by DB-SR,
whereas the respective proportion for bacteria was lower, and only
3.3–4.2% of all taxa
27
,
28
. These specific fungal taxa may thus be benefi-
cial to SOC formation from fungal necromass C. Noteworthily, several
components of fungal necromass, such as chitin, are more resistant to
decomposition than those of bacteria, and they can readily interact
with clay minerals through adsorption. This interaction can stabilize
the fungal residues, contributing to stable MAOC formation and to the
formation of larger soil aggregates, as typically needed for improved
root growth
29
,
30
. In addition, fungi can use less energy to produce more
biomass and subsequent necromass than bacteria when transforming
organic substrates
31
,
32
. Overall, more anticipation is thus attributed to
fungi in stabilizing SOC than bacteria, primarily through the formation
of fungal necromass C.
DB-SR consistently increased crop yield
The long-term field experiment also demonstrated that adoption
of DB-SR increased the grain yield consistently, on average by 16.5%
for wheat, 9.0% for rice and 14.8% for annual yield compared with
RT-SR (Fig.
3a,c
and Extended Data Fig. 4). Wheat yield was related to
spike number (P < 0.05; Fig.
3b
) and kernels per spike (Extended Data
Fig. 5) and rice yield to panicle number (P < 0.05; Fig.
3d
). Adoption
of DB-SR increased the wheat spike number by 10.7% and rice pani-
cle number by 16.4% compared with RT-SR (P < 0.05; Supplementary
Table 3). The increased spike number of wheat can be attributed to
great seedling emergence under DB-SR. Large amounts of crop straw
are retained in surface soil under RT, which suppresses wheat seed
emergence by decreasing the seed–soil contact with more soil pores
and seed embryo destruction by straw-carried pests (Extended Data
Fig. 6). These negative impacts are diminished when the straw is buried
beneath the plough horizon under DB-SR as the root systems of seed-
lings are confined within the topsoil. Besides, the sufficient N supply
under DB-SR promotes the formation of effective spikes per tiller. For
example, spatial niche segregation under DB-SR may decrease the N
competition for seedling roots from straw-decomposing microbes
33
.
The enlarged topsoil layers throughout the entire field after a
DB-SR cycle can additionally promote root growth and the attendant
uptake of N
15
. Collectively, strong wheat seedlings may be estab-
lished, and more productive tillers formed under DB-SR (Extended
Data Fig. 7a,b).
Notably, rice is cultivated by transplanting in this study, and
the seedling number per area is fixed, and therefore, panicle num-
ber increase should originate from tillering. Rice tillering was indeed
increased under DB-SR (Extended Data Fig. 7c). Balanced N–iron (Fe)
sufficiency boosts the grain yield of both rice and wheat by promot-
ing tillering via the inducement of tiller-related gene expressions
34
.
Plant uptake of N and Fe is determined by root development, which is
closely related to soil structure. The SOC increase under DB-SR (Fig.
2a
and Extended Data Fig. 2) is instrumental in fostering aggregation
35
,
36
,
thereby improving soil structure. Well-formed soil structure can pro-
mote root elongation and development through the improvement of
soil hydrothermal conditions, especially in the compacted subsoil
37
.
Enhancement of the hydrothermal regime is crucial for both nutrient
(for example, N) and mineral element (for example, Fe) absorption by
crops
34
. This hypothesis is supported by the positive correlation of
crop yield with the SOC stock (Extended Data Fig. 8). The balanced
N–Fe sufficiency under DB-SR may subsequently induce tillering-
related gene expressions, for example, MONOCULM 1, which controls
the initiation and growth of tillering buds
38
.
The increased accumulation of microbial necromass and elevated
yield can be both explained by the enhanced N availability under DB-SR.
DB-SR increased the concentrations of dissolved organic C (DOC),
available N (NH
4
+
+ NO
3
−
), microbial biomass C (MBC) and microbial
biomass N (MBN), and it decreased the DOC/(NH
4
+
+ NO
3
−
) and MBC/
MBN ratios at 10–40 cm depth (P < 0.05; Extended Data Fig. 9). Over-
all, the DOC/(NH
4
+
+ NO
3
−
) ratio was negatively related to SOC, MAOC
and fungal necromass C in MAOC (Fig.
2g–i
). These results indicate that
DB-SR optimizes the rates of DOC/(NH
4
+
and NO
3
−
) and may provide the
substrates (buried straw) with high quality for microbial utilization and,
thus, promoting the formation of microbial necromass C (refs.
39
,
40
)
and, thus also, microbial necromass N.
Crop straw with a high C/N ratio buried into subsoil increases
fresh C supply and microbial N immobilization, which can theoreti-
cally aggravate N limitation for the growing plant
41
. The rice plants,
however, did not show any signs of N deficiency at the DR-SR fields but
rather exhibited a greener colour than the controls under conventional
paddy management (Supplementary Fig. 2). The greener colour was
in line with an improved N supply at the DR-SR fields. This improved
N supply at lower overall DOC/(NH
4
+
and NO
3
−
) ratios (Fig.
2g–i
) prob-
ably reflected both, better N uptake by the plants and lower N loss. Two
main mechanisms may account for the latter results. First, the limited
oxygen supply decelerates the decomposition of crop straw
15
, thereby
2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024
2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024
4
5
6
7
8
9
10
7
8
9
10
11
7
8
9
10
11
280 300 320 340 360 380
280 300 320 340 360 380
400
4
5
6
7
8
9
10
Wheat yield (Mg ha
–1
)
Rice yield (Mg ha
–1
)
RT-SR
DB-SR
a
b
**
**
* *
**
**
**
**
**
**
c
d
*
ns ns
*
*
r
= 0.66
*
r
= 0.78
**
Year
Spike/panicle number (×10
4
ha
–1
)
Fig. 3 | Effects of DB-SR on the dynamics of crop yields and their relationship to
spike or panicle number. a, Wheat yield. b, The relationship between spike and
wheat yield. c, Rice yield. d, The relationship between panicle and rice yield. Data
in b and d were 2022 and 2023. *P < 0.05; **P < 0.01; ns, no significance. Red and
blue points represented DB-SR and RT-SR, respectively. Data are presented with
means ± standard error across treatment replicates (n = 3). The comparisons in
a and c were analysed using a two-sided paired-sample t-test. The significance of
the linear regression model in b and d was assessed using F-test.
Page 5
Nature Geoscience
| Volume 18 | July
2025 | 631–638
635
Article
https://doi.org/10.1038/s41561-025-01720-5
prolonging the period for microbes to mine N and incorporate it into
their necromass
40
. Second, the reduced oxidative decay of the straw
may prevent acidification. Indeed, the pH values in the individual
ditches under DB-SR varied from 6.9 to 7.5, whereas a slightly lower
pH of 6.8 prevailed in the RT-SR trials (Supplementary Fig. 3). Hence,
the variable charges in soil are more negatively charged under DB-SR
than under RT-SR management, leading to higher effective cation
exchange capacity and better NH
4
+
retention and decreasing N leach-
ing
42
. Certainly, also when calculating the N balance, it showed that
DB-SR reduced N losses by up to 56% (Extended Data Table 1).
Generality validation of positive DB-SR impacts in
China
The DB-SR management system has been widely experimented in China
since our idea was articulated in 2008 (published in Chinese in 2010
16
).
Hence, we performed a meta-analysis with the available data. It con-
firmed that straw return (SR) increased SOC stock by 9.9% at 0–20 cm,
16.3% at 20–40 cm and 8.1% at 40–100 cm soil depth compared with
straw removal (S0) (P < 0.05; Fig.
4a–c
). Specifically, the SOC stocks
under DB-SR were increased by 3.9% only at 0–20 cm depth but by
significantly 23.5% at 20–40 cm depth and by 14.1% at 40–100 cm depth
relative to S0 (P < 0.05). The SR increased crop grain yield by 7.4% com-
pared with S0 (P < 0.05; Fig.
4d
). Specifically, crop yield under SR using
DB ( + 9.4%) was increased more than using NT ( + 7.7%), RT ( + 6.7%) or
PT (7.5%), respectively, using S0 as reference.
Compared with the other tillage practices (no-tillage straw return,
NT-SR; RT-SR; plough-tillage straw return, PT-SR), DB-SR increased SOC
stocks, on average, by 40.3% at 20–40 cm and by 2.4% at 40–100 cm
depth (P < 0.05; Fig.
4e–g
) and crop yield by 10.0% on average (P < 0.05;
Fig.
4h
). Crop yield was variable among crops, and the greatest increase
was shown for maize (12.9% and 12.4%) compared with S0 and the
other tillage practices, respectively (P < 0.05; Supplementary Fig. 4).
Even though none of the considered studies lasted longer than five
years, they all consistently supported our observed benefits from the
DB-SR technique, rendering it a unique option for a more sustainable
paddy management in China and, thus, possibly also in other parts of
the world.
DB-SR/S0
Treatment/control
PT-SR/S0
RT-SR/S0
NT-SR/S0
Overall
5
10
15
20
0
10
20
30
40
50
0
10 20 30 40
0
5
10
15
20
0
DB/PT-SR
DB/RT-SR
DB/NT-SR
Treatment/control
Overall
10
20
30
0
10
20
30
0
60
120
180
240
0
3
6
9
12
0
(681)
(88)
(144)
(272)
(177)
0–20 cm
(226)
(27)
(33)
(43)
(123)
a
b
c
e
–10
–20 –10
f
g
h
20–40 cm
(82)
(5)
(16)
(61)
40–100 cm
(1,115)
(155)
(435)
(372)
(153)
Grain yield
(91)
(4)
(26)
(61)
Changes in crop yield (%)
Changes in SOC stock (%) Changes in SOC stock (%) Changes in SOC stock (%)
0–20 cm
(46)
(18)
(4)
(68)
20–40 cm
(14)
(14)
d
40–100 cm
(56)
(17)
(21)
(18)
Grain yield
Fig. 4 | Responses of SOC stock to DB-SR at different soil depths from a
meta-analysis. a–d, Responses of SOC stock at 0–20 cm (a), 20–40 cm (b) and
40–100 cm (c) and crop yield (d) to SR as compared with straw removal (S0)
using NT, RT and DB. e–h, Responses of SOC stock at 0–20 cm (e), 20–40 cm (f)
and 40–100 cm (g) and crop yield (h) to ditch burying as compared with the other
tillage practices (NT, RT and DB) under SR. The points and error bars indicated
the mean and 95% confidence intervals (CI), respectively. The 95% CI signifies
statistical significance when it does not overlap with the 0 line. The numbers in
parentheses represent the number of comparisons used in the meta-analysis.
RT-SR
DB-SR
Method
Method
–10
–5
0
5
10
15
20
Net CO
2
equivalent emissions (Mg CO
2
-eq ha
–1
)
Changes in SOC stock
N
2
O
CH
4
Harvest
Sowing
Tillage
Electrical energy
Seeds
Herbicides
Pesticides
Fertilizer
a
b
RT-SR
DB-SR
3,000
3,500
4,000
4,500
Net economic benefits (US$ ha
–1
)
Fig. 5 | Effects of DB-SR on net CO
2
equivalent emissions and net economic
benefits. a, Net CO
2
equivalent emissions. b, Net economic benefits.
Page 6
Nature Geoscience
| Volume 18 | July
2025 | 631–638
636
Article
https://doi.org/10.1038/s41561-025-01720-5
DB-SR decreased C emissions and increased
economic benefit
We had reported in our early previous studies, now also supported by
the meta-analysis, that DB-SR is technologically feasible and described
the technological details along with the short-term effects on straw C
accrual and rice yields
43
,
44
. However, from these short-term studies,
neither any conclusion on the ultimate sustainability of DB-SR nor
its role in climate change mitigation could be drawn. Our long-term
observations document that the DB-SR does enhance both SOC and
crop yield, thus offering a solution to sustainably manage our planet.
While future efforts might still be needed to popularize the technol-
ogy of DB-SR into large areas and to evaluate farmers’ willingness to
adopt it, we can already practically offer a well-tested all-in-one farm
equipment for ditch burial by modifying the harvester, achieving the
integrated operations of harvesting, threshing, ditching, burying straws
and recovering soil (Fig.
1c
).
Compared with RT-SR, adoption of DB-SR decreased net CO
2
equivalent emissions by 5.4% and 34.1%, respectively, without and with
considering SOC stock changes for the rice–wheat systems (Fig.
5a
).
Among the components of GHG emissions, we did not find any evidence
that DB-SR greatly increased GHG emissions (Fig.
5a
); we rather found
a tendency even lowering emissions of CH
4
by 10.6% and of N
2
O by 5.6%.
Yet, DB-SR showed a trend of increasing CO
2
equivalent emissions
from diesel by 2.3%. The CH
4
emissions are usually correlated with the
concentrations of dissolved organic C (DOC)
45
,
46
. Given the consider-
able differences in straw placement, mixing and depths between the
DB-SR and RT-SR approaches, it seems reasonable to assume that DOC
produced from the decomposition of straw was both readily utilized by
microbes and converted into necromass and absorbed by unsaturated
minerals when the straw is placed in the subsoil, as also evidenced
by enhanced MAOC contents (Fig.
2
). Microbial N immobilization
and binding of excess NH
4
+
to the straw by electrostatic forces, which
may lower the intensity of nitrification and denitrification and, thus,
reduce N
2
O emissions
42
. Furthermore, any GHGs produced from the
buried straw in the subsoil are probably more difficult to emit due to
physical barriers from the overlying, compacted paddy surface soil.
Overall, the increased SOC stock under DB-SR management contributes
significantly to the reduction of net CO
2
equivalent emissions. These
findings demonstrate that DB-SR is a climate-friendly and thus more
sustainable environmental management practice than the other related
agricultural production systems.
The diesel consumption under DB-SR results from tillage in the
rice season whereas even less diesel is needed in the wheat season
when compared with RT-SR (Supplementary Table 4). The possible
10
20
40
50
10
20
40
50
Straw layer
Rotary tillage
Ditch burying
Fungal taxa are more specific to DB-SR, and fungi have higher efficiency to produce necromass.
FNC readily interacts with clay minerals through adsorption.
Subsoil exhibits the largest undersaturation of MAOC.
Enhanced N adsorption alleviates N limitation, benefiting microbial utilization.
Deep injection process facilitates deeper root growth and root-derived C into the soil.
Tillering
Seed emergence
Straw
C
Straw
C
Limited C inputs to subsoil
Root
C
Root
C
Increased FNC contributing to SOC
Leaching
NH
4
+
NH
4
+
NH
4
+
MAOC
MAOC
Poor seed emergence
Root C
FNC
FNC
SOC
SOC
SOC
SOC
Improved
structure
1
2
3
4
5
4
3
2
1
5
-
(Adsorption)
Increased spikes contributing to grain yield
Fig. 6 | Conceptual diagram illustrating the influencing process of DB-SR on SOC and crop yields. DB-SR improved wheat and rice yields because of spike or panicle
contribution. DB-SR enhancedSOC stock at subsoil through promoting the conversion of straw-derived carbon into mineral-associated fungal necromass.
Page 7
Nature Geoscience
| Volume 18 | July
2025 | 631–638
637
Article
https://doi.org/10.1038/s41561-025-01720-5
explanation for this trend may be that soil is water saturated after
irrigation and under the swelling status during the rice season. When
employing RT for mechanical tillage, soil structure is susceptible to
disruption. However, DB-SR has to be operated before irrigation and
thus consumes more diesel. However, as DB-SR increases crop grain
yields (Fig.
3
), the elevated income and subsequent SOC sequestra-
tion compensate for the increased diesel consumption in comparison
with RT-SR management. The revenue from grain production finally
generates a net profit under DB-SR (Fig.
5b
). Consequently, beyond
ecological benefits, DB-SR shows greater agronomic advantages over
conventional RT-SR, establishing it as a superior eco-economic strat-
egy. This paddy soil conservation approach enhances straw utilization,
subsoil SOC sequestration, grain yield and advances sustainable food
production (Fig.
6
).
Online content
Any methods, additional references, Nature Portfolio reporting sum-
maries, source data, extended data, supplementary information,
acknowledgements, peer review information; details of author con-
tributions and competing interests; and statements of data and code
availability are available at
https://doi.org/10.1038/s41561-025-01720-5
.
References
1. Gebhardt, M. R., Daniel, T. C., Schweizer, E. E. & Allmaras, R. R.
Conservation tillage. Science
230
, 625–630 (1985).
2. Jat, M. L. et al. Conservation agriculture for sustainable
intensification in South Asia. Nat. Sustain.
3
, 336–343 (2020).
3. Pittelkow, C. M. et al. Productivity limits and potentials of the
principles of conservation agriculture. Nature
517
, 365–368 (2015).
4. Zhao, X. et al. Crop yields under no-till farming in China: a
meta-analysis. Eur. J. Agron.
84
, 67–75 (2017).
5. Achankeng, E. & Cornelis, W. Conservation tillage effects on
European crop yields: a meta-analysis. Field Crops Res.
298
,
108967 (2023).
6. Wang, Y., Ying, H., Yin, Y., Wang, H. & Cui, Z. Benefits and
trade-offs of tillage management in China: a meta-analysis.
Agronomy
11
, 1495 (2021).
7. Pittelkow, C. M. et al. When does no-till yield more? A global
meta-analysis. Field Crops Res.
183
, 156–168 (2015).
8. Du, Z., Angers, D. A., Ren, T., Zhang, Q. & Li, G. The effect of
no-till on organic C storage in Chinese soils should not be
overemphasized: a meta-analysis. Agric. Ecosyst. Environ.
236
,
1–11 (2017).
9. Zhao, X., Zhang, R., Xue, J. F., Zhang, X. Q. & Zhang, H. L.
Management-induced changes to soil organic carbon in China:
a Meta-analysis. Adv. Agron.
134
, 1–50 (2015).
10. Gao, S., Tanji, K. K. & Scardaci, S. C. Impact of rice straw
incorporation on soil redox status and sulfide toxicity. Agron. J.
96
, 70–76 (2004).
11. Button, E. S. et al. Deep-C storage: biological, chemical and
physical strategies to enhance carbon stocks in agricultural
subsoils. Soil Biol. Biochem.
170
, 108697 (2022).
12. Li, J., Pei, J., Dijkstra, F. A., Nie, M. & Pendall, E. Microbial carbon
use efficiency, biomass residence time and temperature
sensitivity across ecosystems and soil depths. Soil Biol. Biochem.
154
, 108117 (2021).
13. Fontaine, S. et al. Stability of organic carbon in deep soil layers
controlled by fresh carbon supply. Nature
450
, 277–280 (2007).
14. Zhu, F., Lin, X., Guan, S. & Dou, S. Deep incorporation of corn
straw benefits soil organic carbon and microbial community
composition in a black soil of Northeast China. Soil Use Manag.
38
, 1266–1279 (2022).
15. Yang, H. et al. Ditch-buried straw return: a novel tillage practice
combined with tillage rotation and deep ploughing in rice-wheat
rotation systems. Adv. Agron.
154
, 257–290 (2019).
16. Liu, C. X., Cao, Y., Wang, X. J., Wang, W. B. & Bian, X. M. Research
on absorption of wheat straw to nutrient in urea solution. Acta
Agric. Jiangxi
22
, 125–127 (2010).
17. Lange, M., Eisenhauer, N., Chen, H. & Gleixner, G. Increased soil
carbon storage through plant diversity strengthens with time and
extends into the subsoil. Glob. Chang. Biol.
29
, 2627–2639 (2023).
18. Six, J., Doetterl, S., Laub, M., Müller, C. R. & Van de Broek, M. The
six rights of how and when to test for soil C saturation. Soil
10
,
275–279 (2024).
19. Derrien, D. & Amelung, W. Computing the mean residence time of
soil carbon fractions using stable isotopes: impacts of the model
framework. Eur. J. Soil Sci.
62
, 237–252 (2011).
20. Cotrufo, M. F., Ranalli, M. G., Haddix, M. L., Six, J. & Lugato, E. Soil
carbon storage informed by particulate and mineral-associated
organic matter. Nat. Geosci.
12
, 989–994 (2019).
21. Georgiou, K. et al. Global stocks and capacity of mineral-
associated soil organic carbon. Nat. Commun.
13
, 3797 (2022).
22. Krause, L. et al. Microaggregate stability and storage of organic
carbon is affected by clay content in arable Luvisols. Soil Tillage
Res.
182
, 123–129 (2018).
23. Du, L., Bauke, S. L., Mörchen, R., Schmittmann, O. & Amelung, W.
Fungal necromass is vital for the storage of subsoil C after deep
injection of compost. Soil Tillage Res.
245
, 106325 (2025).
24. Zhao, D. et al. The incorporation of straw into the subsoil
increases C, N, and P enzyme activities and nutrient supply by
enriching distinctive functional microorganisms. Land Degrad.
Dev.
34
, 1297–1310 (2023).
25. Bell, C. W. et al. Linking microbial community structure and
function to seasonal differences in soil moisture and temperature
in a Chihuahuan Desert Grassland. Microb. Ecol.
58
, 827–842
(2009).
26. Xu, Y. et al. Long-term residue returning increased subsoil carbon
quality in a rice–wheat cropping system. J. Environ. Manage.
360
,
121088 (2024).
27. Yang, H. et al. Long-term ditch-buried straw return alters soil
water potential, temperature, and microbial communities in a
rice-wheat rotation system. Soil Tillage Res.
163
, 21–31 (2016).
28. Yang, H. et al. Long term ditch-buried straw return affects soil
fungal community structure and carbon-degrading enzymatic
activities in a rice-wheat rotation system. Appl. Soil Ecol.
155
,
103660 (2020).
29. Veloso, M. G., Angers, D. A., Chantigny, M. H. & Bayer, C. Carbon
accumulation and aggregation are mediated by fungi in a
subtropical soil under conservation agriculture. Geoderma
363
,
114159 (2020).
30. Six, J., Bossuyt, H., Degryze, S. & Denef, K. A history of research on
the link between (micro)aggregates, soil biota, and soil organic
matter dynamics. Soil Tillage Res.
79
, 7–31 (2004).
31. Cotrufo, M. F., Wallenstein, M. D., Boot, C. M., Denef, K. & Paul, E.
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. Change Biol.
19
, 988–995 (2013).
32. Kallenbach, C. M., Frey, S. D. & Grandy, A. S. Direct evidence
for microbial-derived soil organic matter formation and its
ecophysiological controls. Nat. Commun.
7
, 13630 (2016).
33. Yang, H. et al. Waterlogging reduction and wheat yield increase
through long-term ditch-buried straw return in a rice-wheat
rotation system. Field Crops Res.
209
, 189–197 (2017).
34. Song, Y. et al. Balanced nitrogen–iron sufficiency boosts grain
yield and nitrogen use efficiency by promoting tillering. Mol.
Plant.
16
, 1661–1677 (2023).
35. Mustafa, A. et al. Soil aggregation and soil aggregate stability
regulate organic carbon and nitrogen storage in a red soil of
southern China. J. Environ. Manage.
270
, 110894 (2020).
Page 8
Nature Geoscience
| Volume 18 | July
2025 | 631–638
638
Article
https://doi.org/10.1038/s41561-025-01720-5
36. Kan, Z. R. et al. Mechanisms of soil organic carbon stability
and its response to no-till: a global synthesis and perspective.
Glob. Change Biol.
28
, 693–710 (2022).
37. Zhang, P., Wei, T., Jia, Z., Han, Q. & Ren, X. Soil aggregate and
crop yield changes with different rates of straw incorporation in
semiarid areas of northwest China. Geoderma
230
, 41–49 (2014).
38. Li, X. et al. Control of tillering in rice. Nature
422
, 618–621 (2003).
39. Camenzind, T., Mason-Jones, K., Mansour, I., Rillig, M. C. &
Lehmann, J. Formation of necromass-derived soil organic carbon
determined by microbial death pathways. Nat. Geosci.
16
, 115–122
(2023).
40. Cui, J. et al. Carbon and nitrogen recycling from microbial
necromass to cope with C:N stoichiometric imbalance by
priming. Soil Biol. Biochem.
142
, 107720 (2020).
41. Li, S. et al. The mechanism of the dose effect of straw on soil
respiration: evidence from enzymatic stoichiometry and
functional genes. Soil Biol. Biochem.
168
, 108636 (2022).
42. Yang, H. et al. Effects of ditch-buried straw return on water
percolation, nitrogen leaching and crop yields in a rice–wheat
rotation system. J. Sci. Food Agric.
96
, 1141–1149 (2016).
43. Zhai, S. et al. Long-term ditch-buried straw return alters soil carbon
sequestration, nitrogen availability and grain production in a
rice–wheat rotation system. Crop Pasture Sci.
72
, 245–254 (2021).
44. Wang, X. et al. Effects of ditch-buried straw return on soil organic
carbon and rice yields in a rice–wheat rotation system. Catena
127
, 56–63 (2015).
45. Kan, Z. et al. Soil organic carbon regulates CH
4
production
through methanogenic evenness and available phosphorus
under different straw managements. J. Environ. Manage.
328
,
116990 (2023).
46. Kan, Z. et al. Tillage after irrigation decreases soil N
2
O but
increases CO
2
emissions in a rice–wheat rotation system. Sustain.
Prod. Consum.
44
, 114–122 (2024).
Publisher’s note
Springer Nature remains neutral with regard
to jurisdictional claims in published maps and institutional
affiliations.
Springer Nature or its licensor (e.g. a society or other partner) holds
exclusive rights to this article under a publishing agreement with
the author(s) or other rightsholder(s); author self-archiving of the
accepted manuscript version of this article is solely governed by the
terms of such publishing agreement and applicable law.
© The Author(s), under exclusive licence to Springer Nature Limited
2025
Page 9
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Methods
Field experiment
Experimental site
.
The field experiment began in November 2008
at the Institute of Agricultural Sciences in Yanjiang District, Jiangsu
Province (32° 13′ N, 121° 03′ E). This region experiences a subtropical
monsoonal climate, with a mean annual temperature of 14.5 °C and
precipitation of 1,057.0 mm. Soil, classified as Gleyic Cambisols, has a
sandy loam texture. Before the experiment, soil parameters at a depth
of 0–20 cm were as follows: SOC, 11.9 g kg
−1
; total nitrogen, 1.6 g kg
−
¹;
available phosphorus, 12.8 mg kg
−1
; available potassium, 44.4 mg kg
−1
;
and pH 7.5. The cropping system comprises a rice–wheat rotation, with
rice cultivated from June to October and wheat from November to the
subsequent May.
Experimental design
.
The experiment adopted a randomized com-
plete block design, comprising two treatments: RT-SR (adopted in over
80% of regions in China) and DB-SR. Each treatment was replicated
thrice, totalling six plots, and each experimental plot covered an area of
112 m
2
(14 × 8 m). For RT-SR, straw was chopped into 5–10 cm pieces and
incorporated into the soil at 0–10 cm depth using a rotary cultivator.
For DB-SR treatment, straw was bundled, weighed and transported to
the field ridge for three days of sun-drying and buried into ditches of
20-cm depth, 20-cm width and neighbouring distance of 2 m (Fig.
1a
).
The land area under the ditch positions represents only 10% whereas
the remaining 90% of the total area is adopted NT in each season. In
preliminary short-term (4 years) studies, we had investigated different
straw burial depths (10, 20, 30 and 40 cm) with different amounts (half,
total and double) on SOC and crop yield under DB-SR
44
. The results
had shown that full amount of straw buried to 20-cm depth showed
the highest SOC and rice yield, and thus this amount and depth were
selected in our study.
Considering the annual double cropping of rice and wheat, a com-
plete DB-SR cycle will be completed in five years. The procedures for
DB-SR included arranging straw bundles, positioning and creating
straw ditches, placing straw into the ditches and compacting, recover-
ing soil from digging the ditch and then overburdening it (Fig.
1b
). The
positions of straw ditches were rotated after each cropping season. We
designated the ditches created in November 2008 with rice straw as
the first, and those made in June 2009 with wheat straw as the second
and so on. Ditches marked with odd numbers were filled with rice
straw, whereas those with even numbers were filled with wheat straw.
Additional information regarding fertilizer, irrigation, pests and weed
management was presented in our previous studies
45
.
Soil sample and analysis
.
The soil was stratified into depth intervals
of 0–10 cm, 10–20 cm, 20–30 cm and 30–40 cm and sampled using a
soil auger (3.5-cm diameter) at the rice maturity stage in October 2023.
The experiment had been going on for 15 years at the sampling time. For
RT-SR, soil samples were collected from five points along an S-shaped
linear transect within each plot. For DB-SR, straw ditches were precisely
marked over different years in each plot. Three straw ditches were built
after each crop season, given a total of 30 straw ditches in each plot
after a DB-SR cycle (Fig.
1a
). One soil core was randomly collected from
each straw ditch, and three soil cores from the same season in each plot
were mixed thoroughly for determining the SOC concentration in each
ditch. Subsamples of ten straw ditches with different time were mixed
thoroughly with equivalent soil mass for determining SOC fractions
and soil properties. Recognizable gravels, debris and straw residues
(including straw layers) were manually picked out before analysis. Soil
bulk density was determined using a core ring (100 cm
3
) method at the
same four soil depth ranges at the rice maturity stage in October 2023.
In each replication plot, randomized two core rings were collected for
RT-SR whereas three representative core rings for DB-SR (among ditch
4 and 10; 2 and 5; 3 and 7; Supplementary Fig. 5). The air-dried soil was
sieved through a 2-mm mesh after removing visible debris.
Measurement of SOC fractions and soil properties
.
Soil NH
4
+
and
NO
3
−
were extracted by K
2
SO
4
and then analysed by the colorimetric
method. Soil DOC was analysed from the same supernatant using a
dissolved C autoanalyser (TOC-L, Shimadzu). The MBC and MBN were
measured using the chloroform fumigation–extraction methods
47
. The
SOC content was measured in the air-dried soil samples after remov-
ing visible debris through a 2-mm mesh using the K
2
Cr
2
O
7
oxidation
method
48
. Soil natural δ
13
C values were measured using an isotope mass
spectrometer (CF-IRMS, Bremen). Hexametaphosphate extraction
was used for POC and MAOC measurement
49
. Briefly, 10 g of air-dried
soil was dispersed in a centrifuge tube containing hexametaphos-
phate solution (5 g ml
−1
) and shaken for 18 h. Dispersed samples were
sieved through the 53-μm mesh and subsequently rinsed with deion-
ized water to separate POC (>53 μm) from MAOC fractions (<53 μm).
These isolated fractions were dried in an oven at 60 °C until a constant
weight. The dried soil fractions were finely ground using a ball mill
for SOC analysis. Amino sugars (glucosamine [GluN], galactosamine
[GalN] and muramic acid [MurA]) in POC and MAOC fractions were
determined by high-performance liquid chromatography following
hydrolysis, purification, derivatization and derivative purification
50
.
The GluN and MurA were used to calculate the microbial (fungal and
bacterial) necromass C
51
.
Bacterial necromass C
=
MurA
×
45
(1)
Fungal necromass C
= (
GluN
179
.
2
−
2
×
MurA
251
.
2
) ×
179
.
2
×
9
(2)
where 45 and 9 are the conversion coefficients for MurA to bacterial-
derived C and for GluN to fungal-derived C, respectively. The molecular
weights of MurA and GluN are 251.2 and 179.2, respectively. Microbial
necromass C was determined as the amounts of bacterial and fungal
necromass C.
The stock of SOC and its fractions (Mg ha
−1
) were determined
based on the equivalent soil mass
52
considering the changes in soil bulk
density (Supplementary Fig. 6).
SOC stock
=
n
∑
i
=
1
[
M
soil
,
i
×
conc
i
+ (
M
m
,
i
−
M
soil
,
i
) ×
conc
i
+
1
] ×
0
.
001 (3)
M
soil
,
i
=
BD
i
×
H
i
×
10
,
000
(4)
where M
soil,i
represents the soil mass of layer i (Mg ha
−1
); conc
i
and conc
i+1
refer to the SOC concentrations of layer i and i + 1 (g kg
−1
), respectively;
and M
soil,i
is the reference soil mass of layer i (Mg ha
−1
). The maximum
value chosen was M
m,i
. i
1-5
represents the 0–10, 10–20, 20–30, 30–40
and 40–50 cm soil layers, respectively. The BD
i
is field bulk density of
layer i (g cm
−3
); and H
i
is the soil thickness of layer i (m). To calculate the
SOC stock at 30–40 cm, soils at 40–50 cm layer were also collected.
Grain yield
.
Grain yield for wheat and rice was determined on three
1 × 1 m
2
quadrats in each plot and normalized to 13.0% moisture con-
tent. The average value was used as the grain yield of one replication
plot. The coefficients of variation for subsample in each plot, and
between plots showed similar values, indicating that the representa-
tive quadrats were sampled (Supplementary Fig. 7). Kernel weight was
determined using three random subsamples, each containing 1,000
kernels for grain samples from each replication plot. The spike number
per hectare was estimated by counting all spikes from each harvested
area. Furthermore, the kernel number per spike was determined from
30 spikes randomly sampled from each treatment.
Soil greenhouse gas emissions
Soil greenhouse gas (GHG) fluxes, including N
2
O and CH
4
, from the
rice–wheat system were measured using the static chamber method
Page 10
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
during a cycle of DB-SR. The gas samples were collected from 9:00
to 11:00 a.m. once per week, with more frequent sampling following
precipitation or fertilizer application events. The chambers, sized
either 50 cm × 50 cm × 50 cm or 50 cm × 50 cm × 100 cm based on the
crop height, were mounted on a PVC frame. A 5-cm deep groove along
the top edge of the frame was filled with water to seal the chamber.
Each chamber was equipped with a fan to ensure gas mixing and was
insulated with a layer of material covered in aluminium foil to minimize
temperature fluctuations. Two chambers (A and B) were installed in
each replicate plot of DB-SR treatment. Chamber A covered the ditch
and some of the area on either side of it, and Chamber B covered the
non-ditch area. The position of chamber A was adjusted according
to the position of the latest straw ditches. The values from the latest
straw ditches were used to convert the GHG from straw ditches in
each season. The GHG emissions from DB-SR were determined by the
sum of ditch and non-ditch considering the area each occupies
53
. We
recognize the risk that the measured GHG can overestimate the actual
value because fresh straws are hotspots for GHG emissions, for which
the data are extrapolated to the old ditches; yet, GHG emissions from
DB-SR still had a lower value compared with RT-SR.
Cumulative GHG emissions were determined as follows
46
.
Cumulative GHG
(
kg ha
−
1
) =
n
∑
1
[
F
i
+
F
i
−
1
2
×(
t
i
−
t
i
−
1
)]
(5)
where F
i
and F
i-1
are GHG fluxes at the ith and i − 1th sampling, respec-
tively; t
i
and t
i-1
are the ith and i − 1th sampling time, respectively; and
n is the number of gas collected.
Calculation of N balance
For the calculation of N balance, N loss was estimated by subtracting
changes in soil N pool and harvested grain N pool from the N inputs
(equation (
6
))
54
. As all the straw was returned into the soil, the grain N
was the only biomass N taken away from the field. The N inputs com-
prised inputs from fertilization, atmospheric deposition and irrigation
N (equation (
7
)). The N deposition was taken from Zhu et al.
55
; for the
irrigation estimate, we applied the irrigation amount with the N con-
centration in the irrigation water (Extended Data Table 1). Biological
N fixation can be ignored in a wheat–rice system because no legumes
are incorporated
54
, and if N fixation by cyanobacteria in the flood
water occurs, it was probably similar in both DB-SR and RT-SR trials.
Also, gaseous N losses were not considered because differences in N
2
O
rates between treatments were negligibly small (Fig.
5
), and N
2
losses
remained unknown. Changes in the soil N pool were thus calculated
from the difference in total soil N stocks between the time of current
sampling (2023) and the start of experiment (2008). Grain yield was
estimated according to equation (
8
).
N loss
(
Mg ha
−
1
yr
−
1
) =
N inputs
−
changes in soil N stock
−
harvested grain N pool
(6)
N inputs
=
Fertilizer N
+
irrigation N
+
deposition N
(7)
Grain N
=
wheat yield
×
N content in wheat grain
+
rice yield
×
N content in rice grain
(8)
Eco-economic evaluation
.
The eco-economic evaluation comprised
the net CO
2
equivalent emissions and net economic benefits. Net CO
2
equivalent emissions (kg CO
2
-eq ha
−1
) were the sum of indirect CO
2
equivalent emissions from agricultural inputs and direct soil N
2
O and
CH
4
emissions minus changes in SOC stock based on a life-cycle assess-
ment. The system boundary for the life-cycle assessment encompassed
the entire life cycle of external inputs, from manufacture, storage
and delivery to crop harvest, including changes in SOC stock
56
. Soil
CO
2
emissions were included in the changes in SOC stock and thus
were not calculated in this study
57
. Net CO
2
equivalent emissions were
calculated as follows
58
:
Net CO
2
equivalent emissions
=
indirect
+
direct emissions
−
changes in SOC stock
(9)
Indirect emissions
=
n
∑
1
(
A
i
×
δ
i
)
(10)
Direct emissions
=
CH
4
×
29
.
8
+
N
2
O
×
273
(11)
Changes in SOC stock
=
SOCstock
2023
−
SOC stock
2008
15
×
44
12
(12)
where indirect CO
2
equivalent emissions arise from agricultural inputs,
including fertilizers, pesticides, diesel and electricity. A
i
is the annual
quantity of the ith agricultural input used in rice–wheat production
(kg ha
−1
; Supplementary Table 4) and δ
i
the GHG emissions factor for
the ith agricultural input (kg CO
2
-eq kg
−1
; Supplementary Table 5).
Global warming potential of N
2
O and CH
4
is 273 and 29.8 estimated from
a period of 100 years, respectively
59
. The coefficient 44/12 is used to
convert C into CO
2
. SOC stock
2023
and SOC stock
2008
are the SOC stock
at the end and beginning of the long-term experiment, respectively.
The number 15 is the experiment duration.
Net economic benefit was calculated based on the within-year
crop yield (Yield) multiplied by its corresponding price in Chinese Yuan
(Price) minus the costs of crop production (Cost) (US$ ha
−1
, US$1 = 6.95
Chinese Yuan).
Net economic benefit
=
Yield
×
Price
−
Cost
(13)
Data analysis for the field experiment
.
Normal distribution was
performed first, and variables exhibiting non-normal distribution
underwent log transformation. A paired-sample t-test was used to
analyse the effects of DB-SR on SOC and its fractions and grain yield per
year compared with RT-SR from the field experiment. For data on the
SOC concentration of ten straw ditches, analysis of variance (ANOVA)
was adopted with treatment (RT-SR and ten ditches of DB-SR) as fixed
factors and block as a random factor. Analysis of variance for repeated
measures was conducted by a general linear model to compare the
time-series data (for example, crop yield) among DB-SR and RT-SR.
Linear regression analysis was conducted to determine the relation-
ship of crop yield to spike (panicle) number, kernel number per spike
(panicle) and thousand-kernel weight.
Meta-analysis
Data collection
.
The literature search was conducted on the Web of
Science (
http://www.webofknowledge.com
) and the China National
Knowledge Infrastructure (CNKI;
https://www.cnki.net
) to screen
studies involving the effects of tillage and straw management on SOC
and crop yield. The last search was done in December 2023 with the
equation: (‘till*’ OR ‘residue’ OR ‘ditch’ OR ‘straw’ OR ‘residue’) AND
(‘yield’) OR (‘carbon’). Publications were screened based on the spe-
cific criteria: (1) the experiment was conducted under field conditions
including side-by-side comparisons; (2) the data on SOC or crop yield
data were available; (3) the initial climatic conditions, soil properties
and primary crop management practices have been documented;
(4) the experiment location was provided. An attempt was made to
collect global-scale datasets, but the experiments about DB-SR were
all in China.
DB-SR involved two management practices (straw and tillage),
and two datasets were constructed to compare the effects of SR ver-
sus S0 using DB, NT, RT and PT (Dataset I) and DB versus the other
tillage practices (NT, RT and PT; Dataset II) with the former (SR and
Page 11
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
DB) as treatment and the latter (S0 and the other tillage practices) as
control. Finally, 989 comparisons were performed on SOC stocks in
SR versus S0, while we obtained 173 data points for the comparison
of DB versus the other tillage (NT, RT and PT) practices. For the crop
yields, we compiled 1,115 data points for the comparison of SR with S0
and 56 for those of DB versus the other tillage options. A total of 249
publications were collected, and 38 out of 249 were associated with
DB-SR (Supplementary publications for meta-analysis). The locations
of these experiment stations are shown in Supplementary Fig. 8a,b.
When data were presented graphically, values were extracted using
the GetData software (
http://getdata-graph-digitizer.com/
).
Data analysis
.
For cases not providing bulk density, we estimated it
by equation (
14
)
60
.
BD
=
1
.
377 exp
(−
0
.
0048
×
SOC
)
(14)
where exp represents exponential function with base e. A random-
effects meta-analysis was used to evaluate the effects on SOC and crop
yield under DB-SR compared with the other management practices.
The effect size was calculated using the natural logarithm (lnR) of the
response ratio
61
.
lnR
=
lnX
t
−
lnX
c
(15)
where X
t
and X
c
are the mean values of SOC or yield under treatment and
control, respectively. Because standard deviations were not provided
in many studies, the weight (w) for each natural logarithm ratio (lnR)
was calculated using the number of experiment replications, as shown
in equation (
16
)
3
.
w
= (
n
t
×
n
c
)/(
n
t
+
n
c
)
(16)
where n
c
and n
t
denote the number of observational replications in the
control and treatment groups, respectively. When a study contributed
multiple observations to a category, weights were adjusted by dividing
them with the total number of observations
3
.
The quality of lnR data for SOC and yield were assessed using fre-
quency distribution. A Gaussian function was applied to fit the normal
distribution of each observed variable using the following equation.
y
=
a
×
exp
(−(
x
−
x
0
)
2
/
2b
2
)
(17)
where x
0
was the mean values of lnRs, x and y denote the mean value
of lnRs within their respective intervals and the frequency within each
interval and a and b stand for a coefficient and the variance of the fre-
quency distribution, respectively. The lnR of SOC stock and crop yield
followed a Gaussian normal distribution (P < 0.0001; Supplementary
Fig. 8c,d). Bootstrapping (4,999 iterations) generated the 95% con-
fidence interval (CI) and weighted mean effect size (lnR+). For ease
of interpretation, the lnR+ results were reverse-transformed as the
percentage change using [(R − 1) × 100].
Data availability
All data used and source data for figures and tables in this study are
available via figshare at
https://doi.org/10.6084/m9.figshare.28334369
(ref.
62
). Source data are provided with this paper.
References
47. Vance, E. D., Brookes, P. C. & Jenkinson, D. S. An extraction
method for measuring soil microbial biomass C. Soil Biol.
Biochem.
19
, 703–707 (1987).
48. Walkley, A. & Black, I. A. An examination of the Degtjareff method
for determining soil organic matter, and a proposed modification
of the chromic acid titration method. Soil Sci.
37
, 29–38 (1934).
49. Cambardella, C. A. & Elliott, E. T. Particulate soil organic matter
change across a grassland cultivation sequence. Soil Sci. Soc.
Am. J.
56
, 777–783 (1992).
50. Zhang, X. & Amelung, W. Gas chromatographic determination of
muramic acid, glucosamine, mannosamine, and galactosamine in
soils. Soil Biol. Biochem.
28
, 1201–1206 (1996).
51. Liang, C., Amelung, W., Lehmann, J. & Kästner, M. Quantitative
assessment of microbial necromass contribution to soil organic
matter. Glob. Change Biol.
25
, 3578–3590 (2019).
52. Ellert, B. H. & Bettany, J. R. Calculation of organic matter and
nutrients stored in soils under contrasting management regimes.
Can. J. Soil Sci.
75
, 529–538 (1995).
53. Hu, N. et al. Effects of different straw returning modes on
greenhouse gas emissions and crop yields in a rice–wheat
rotation system. Agric. Ecosyst. Environ.
223
, 115–122 (2016).
54. Zhang, X. et al. Managing nitrogen for sustainable development.
Nature
528
, 51–59 (2015).
55. Zhu, X. et al. Nitrogen emission and deposition budget in an
agricultural catchment in subtropical central China. Environ.
Pollut.
289
, 117870 (2021).
56. Chen, X. et al. Identifying the main crops and key factors
determining the carbon footprint of crop production in China,
2001–2018. Resour. Conserv. Recycl.
172
, 105661 (2021).
57. Gan, Y. et al. Improving farming practices reduces the carbon
footprint of spring wheat production. Nat. Commun.
5
, 5012
(2014).
58. Yang, X. et al. Diversifying crop rotation increases food
production, reduces net greenhouse gas emissions and improves
soil health. Nat. Commun.
15
, 198 (2024).
59. IPCC Climate Change 2021: The Physical Science Basis
(eds Masson-Delmotte, V. et al.) (Cambridge Univ. Press, 2021).
60. Song, G., Li, L., Pan, G. & Zhang, Q. Topsoil organic carbon
storage of China and its loss by cultivation. Biogeochemistry
74
,
47–62 (2005).
61. Hedges, L. V., Gurevitch, J. & Curtis, P. S. The meta-analysis of
response ratios in experimental ecology. Ecology
80
, 1150–1156
(1999).
62. Yang, H. Raw data-NGS-2024-07-01565.xlsx (Nature Geoscience).
figshare
https://doi.org/10.6084/m9.figshare.28334369
(2025).
Acknowledgements
The study was supported by the National Key R&D Program of China
(2023YFD2302400) (Z.-R.K.) and the Natural Science Foundation of
Jiangsu Province (BK20241553) (Z.-R.K.). W.A. acknowledges support
from the Chinese Academy of Sciences President’s International
Fellowship Initiative and the Germany’s Excellence Strategy
(EXC-2070–390732324–PhenoRob) funded by German Research
Foundation. The determination of microbial necromass was performed
by BaiHui Biotechnology Co., Ltd at Chengdu, China.
Author contributions
Z.-R.K., F.-M.L. and H.Y. designed the study. Z.L., J.L. and Y.X. performed
the field experiment. Z.-R.K., Z.L., F.-M.L. and H.Y. collected data for
field experiment and meta-analysis. Z.-R.K. and H.Y. wrote the paper.
H.-L.Z., R.L., R.B., W.A. and F.M.L. improved and revised the paper.
Competing interests
The authors declare no competing interests.
Additional information
Extended data
is available for this paper at
https://doi.org/10.1038/s41561-025-01720-5
.
Supplementary information
The online version contains supplementary
material available at
https://doi.org/10.1038/s41561-025-01720-5
.
Page 12
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Correspondence and requests for materials
should be addressed to
Feng-Min Li or Haishui Yang.
Peer review information
Nature Geoscience thanks Tuomas Mattila
and the other, anonymous, reviewer(s) for their contribution to the
peer review of this work. Primary Handling Editors: Xujia Jiang and
James Super, in collaboration with the Nature Geoscience team.
Reprints and permissions information
is available at
www.nature.com/reprints
.
Page 13
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Extended Data Table 1 | Nitrogen balance of RT-SR and DB-SR after 15 years’ application
N input
Soil N pool
N output
N fertilizer (kg ha
-1
yr
-1
)
N deposition (kg ha
-1
yr
-1
)
Irrigation N (kg ha
-1
yr
-1
)
Changes after trials
(kg ha
-1
yr
-1
)
Grain N (kg ha
-1
yr
-1
)
N loss (kg ha
-1
yr
-1
)
RT-SR
480.00 a
35.90 a
38.00 b
55.10 b
303.08 b
195.72 a
DB-SR
480.00 a
35.90 a
38.00 b
70.68 a
319.11 a
164.11 b
N loss
(
Mg ha
−
1 yr
−
1
) =
N inputs
−
Changes in soil N stock
−
harvested Grain N pool N inputs
=
Fertilizer N
+
Irrigation N
+
deposition N
Grain N pool
=
wheat yield
×
N content in wheat grain
+
rice yield
×
N content in rice grain Calculation of soil N pool was similar to SOC pool. DB-SR, ditch-buried straw return; RT-SR,
rotary-tillage straw return. Different lowercase letters indicate significant differences (P<0.05) between RT-SR and DB-SR. The comparisons were analyzed using two-sided paired-sample T test.
Page 14
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Extended Data Fig. 1 | Effects of 15-year ditch-buried straw return on soil
organic carbon stock at 0-40 cm depth based on fixed soil depth and
equivalent soil mass. DB-SR, ditch-buried straw return; RT-SR, rotary-tillage
straw return. Data are presented with means ± standard error across treatment
replicates (n = 3). The comparisons were analyzed using two-sided paired-sample
t-test. Values adjacent to paired bars denote P values for differences between
RT-SR and DB-SR treatments.
Page 15
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Extended Data Fig. 2 | Soil organic carbon concentration of ten straw ditches
for ditch-buried straw return at 0-50 cm depth. DB-SR, ditch-buried straw
return; RT-SR, rotary-tillage straw return. Number at X-axis represents individual
ditches under different burial duration after a complete (5 years) DB-SR cycle.
Data are presented with means ± standard error across treatment replicates
(n = 3). The difference was analyzed using one-sided F-test and Duncan’s multiple
comparison. Different letters mean the significant differences at P < 0.05 level;
ns, no significance. P = 0.020, <0.001, <0.001, <0.001, and 0.430 in (a), (b), (c),
(d), and (e), respectively.
Page 16
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Extended Data Fig. 3 | Effects of ditch-buried straw return on POC fraction
stock at different soil depths. (a)-(d) represent POC (Mg ha
-1
), MNC in POC
(Mg ha
-1
), BNC in POC (Mg ha
-1
), and FNC in POC (Mg ha
-1
). POC, particulate
organic carbon; MNC, microbial necromass carbon; BNC, bacterial necromass
carbon; FNC, fungal necromass carbon. DB-SR, ditch-buried straw return; RT-SR,
rotary-tillage straw return. The stocks of POC, MNC, BNC, and FNC in POC were
calculated using equivalent soil mass. Data are presented with means ± standard
error across treatment replicates (n = 3). The comparisons were analyzed using
two-sided paired-sample t-test. Values adjacent to paired bars denote P values for
differences between RT-SR and DB-SR treatments.
Page 17
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Extended Data Fig. 4 | Effects of ditch-buried straw return on wheat (a),
rice (b), and annual (c) yields on average from 2008 to 2023. DB-SR, ditch-
buried straw return; RT-SR, rotary-tillage straw return. n, sample number. **,
P < 0.01. The solid and dashed lines in box represent median and mean values,
respectively. Box minima and maxima represent the 25th and 75th percentile,
and whisker minima and maxima represent the 25th and 75th percentile minus
and plus 1.5× the interquartile range respectively. Analysis of variance for
repeated measures was conducted by a general linear model to compare the yield
effects among DB-SR and RT-SR. P = 0.003, 0.003, and <0.001 in (a), (b), and (c),
respectively.
Page 18
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Extended Data Fig. 5 | Relationship of crop yield to kernels per spike (a, c) and thousand kernel weight (b, d). Red and blue points represented DB-SR and RT-SR,
respectively. DB-SR, ditch-buried straw return; RT-SR, rotary-tillage straw return. The significance of the linear regression model in (a)-(d) was assessed using F-test.
**, P < 0.01.
Page 19
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Extended Data Fig. 6 | Photos showing the seed embryo destruction by straw-carried pests under rotary-tillage straw return. Photos taken by Xinhu Guo in
December 2023.
Page 20
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Extended Data Fig. 7 | Photo showing the spike number of wheat under rotary-
tillage straw return and ditch-buried straw return. (a), photo of spike number
taken on 27/05/2023; (b) wheat emergence rate; (c) rice tillering number. DB-SR,
ditch-buried straw return; RT-SR, rotary-tillage straw return. Data in b and c are
presented with means ± standard error across treatment replicates (n = 3). The
comparisons in (b) were analyzed using two-sided paired-sample t-test.
Page 21
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Extended Data Fig. 8 | Relationship between changes in soil organic carbon
stock and crop yield at 0-20 cm (a) and 20-100 cm (b) depths. No significant
correlation when lnR of SOC ≥ 0.4 may be due to the limitation of data amounts.
LnR
=
LnX
t
−
LnX
c
, X
t
represents the values of experimental treatment, and X
c
for the values of control. The significance of the linear regression model in (a)
and (b) was assessed using F-test.
Page 22
Nature Geoscience
Article
https://doi.org/10.1038/s41561-025-01720-5
Extended Data Fig. 9 | Effects of DB-SR on soil available C and N contents.
(a)-(f) represent DOC (mg kg
-1
), NH
4
+
+ NO
3
-
(mg kg
-1
), DOC/(NH
4
+
+ NO
3
-
), MBC
(mg kg
-1
), MBN (mg kg
-1
), and MNC/MBN, respectively. DB-SR, ditch-buried straw
return; RT-SR, rotary-tillage straw return; DOC, dissolved organic carbon; MBC,
microbial biomass carbon; MBN, microbial biomass nitrogen. ns, no significance.
Data are presented with means ± standard error across treatment replicates
(n = 3). The comparisons were analyzed using two-sided paired-sample t-test.
Values adjacent to paired bars denote P values for differences between RT-SR and
DB-SR treatments.
Document Outline