Yang et al. (2024). Diversifying crop rotation increases food production, reduces net greenhouse gas emissions and improves soil health

,

Yantai Gan

4,5

, Sien Li

1,2

, Longlong Xia

, Yanjun Shen

7

,

Steven Pacenka

8

, Tammo S. Steenhuis

8

, Kadambot H. M. Siddique

9

,

Shaozhong Kang

1,2

& Klaus Butterbach-Bahl

10,11

Global food production faces challenges in balancing the need for increased
yields with environmental sustainability. This study presents a six-year

ﬁ

eld

experiment in the North China Plain, demonstrating the bene

ﬁ

ts of diversify-

ing traditional cereal monoculture (wheat

–

maize) with cash crops (sweet

potato) and legumes (peanut and soybean). The diversi

ﬁ

ed rotations increase

equivalent yield by up to 38%, reduce N

2

O emissions by 39%, and improve the

system

’

s greenhouse gas balance by 88%. Furthermore, including legumes in

crop rotations stimulates soil microbial activities, increases soil organic carbon
stocks by 8%, and enhances soil health (indexed with the selected soil phy-
siochemical and biological properties) by 45%. The large-scale adoption of
diversi

ﬁ

ed cropping systems in the North China Plain could increase cereal

production by 32% when wheat

–

maize follows alternative crops in rotation

and farmer income by 20% while bene

ﬁ

ting the environment. This study

provides an example of sustainable food production practices, emphasizing
the signi

ﬁ

cance of crop diversi

ﬁ

cation for long-term agricultural resilience and

soil health.

Producing more nutritious food to alleviate world hunger

1

while safe-

guarding the environment

is a signi

ﬁ

cant challenge for humanity.

The challenge is much pronounced in highly-populated countries and
regions where agricultural resources are limited. Conventional inten-
si

ﬁ

ed food production systems

’

reliance on major inputs of synthetic

agrochemicals, such as fertilizers and pesticides, have boosted food
production signi

ﬁ

cantly since the

‘

Green Revolution

’

, but these

systems have also emitted great amounts of greenhouse gases (GHG)
and caused environmental degradation

yangxiaolin429@cau.edu.cn

. For instance, China

’

s crop

production has increased by 74% in the past 30 years (1986

–

2016), but

this came with a >300% increase in synthetic fertilizer use

7

. In 2019,

China

’

s GHG emissions associated with food production reached

2.4 Gt CO

2

-equivalent (CO

2

-eq)

8

, with about 50% emitted during the

crop production stage

8

,

9

. The loss of soil fertility which may go along

Received: 4 September 2023

Accepted: 14 December 2023

Check for updates

1

State Key Laboratory of Ef

ﬁ

cient Utilization of Agricultural Water Resources, Beijing 100083, China.

2

College of Water Resources & Civil Engineering, China

Agricultural University, Beijing 100083, China.

3

School of Tropical Agriculture and Forestry, Hainan University, Haikou 570228, China.

4

College of Life and

Environmental Sciences, Wenzhou University, Wenzhou, Zhejiang 325035, China.

5

The

μ

BC-Soil Group, Tallus Heights, Kelowna, BC, Canada.

6

State Key

Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, Nanjing 210008, China.

7

Key Laboratory of Agri-

cultural Water Resources, Centre for Agricultural Resources Research, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences,
Shijiazhuang 050022, China.

8

Department of Biological and Environmental Engineering, Riley-Robb Hall, Cornell University, Ithaca, NY 14853, USA.

9

The UWA

Institute of Agriculture, The University of Western Australia, Perth, WA 6001, Australia.

10

Land-CRAFT, Department of Agroecology, Aarhus University,

Aarhus, Denmark.

11

Institute of Meteorology and Climate Research, Atmospheric Environmental Research (IMK-IFU), Karlsruhe Institute of Technology (KIT),

Garmisch Partenkirchen, Germany.

e-mail:

;

dutaisheng@cau.edu.cn

;

juxt@cau.edu.cn

;

gary.gan@wzu.edu.cn

Nature Communications

| (2024) 15:198

1

12

3

4

5

6

78

12

3

4

5

6

7

8

with the intensi

ﬁ

cation of crop production further complicates food

production

10

12

and exposes it to climate risks

13

14

and environmental

health concerns

15

16

.

Innovative concepts like integrated farming systems with diver-

si

ﬁ

ed crop rotations

17

18

have emerged to address these food produc-

tion and environmental sustainability challenges. These systems offer
a range of food crops to meet consumer demand for plant-based,
healthy food

19

—

an increasing dietary trend in high and upper-middle-

income countries

20

—

while providing other agricultural products such

as animal feed, industrial

ﬁ

ber, or multi-purpose biofuels

21

,

22

. More-

over, integrated food production systems help increase farmers

’

income and deliver socio-economic bene

ﬁ

ts

20

,

23

. However, little is

known about how cash-crop and legume-diversi

ﬁ

ed cropping systems

can achieve the triple goals

—

increasing food yields, reducing envir-

onmental footprint, and bene

ﬁ

ting soil health.

In this context, we conducted a 6-year (2016

–

2022)

ﬁ

eld study in

the North China Plain

—

the food basket of China

—

one of the most

intensively cultivated regions in the world, where crop production is
dominated by simple winter wheat (

Triticum aestivum

L.) and summer

maize (

Zea mays

L.) double cropping (wheat

–

maize, WM; two crops 1

year) which occupies 70% of the area

’

s arable land, delivering about

23% of China

’

s total cereal food

24

. The speci

ﬁ

c objective of the case

study was to asses comprehensively several diversi

ﬁ

ed cropping sys-

tems in terms of food production, GHG balance, soil health bene

ﬁ

ts,

and farmers

’

income. We tested various diversi

ﬁ

ed cropping systems,

incorporating cash-crops, legumes, other cereals, or forages into the
conventional wheat

–

maize system. Major performance metrics inves-

tigated were: (i) plant biomass, grain and protein yields, and farmers

’

net incomes; (ii) soil GHG

ﬂ

uxes, soil carbon (C) sequestration, and net

GHG emissions; and (iii) soil health parameters including soil pH, bulk
density, soil water content, total nitrogen (TN), dissolved organic
carbon (DOC), soil nitrate-N (NO

3

–

-N), ammonium-N (NH

4

+

-N),

available phosphorus (AP), microbial biomass carbon (MBC), microbial
biomass nitrogen (MBN), and microbial community composition and
diversity. These measurements enabled us to test the hypotheses that
(a) cash-crop diversi

ﬁ

ed system increases farmers

’

net income without

jeopardizing crop yields, (b) legume diversi

ﬁ

ed systems reduce

ﬁ

eld

scale GHG emissions, and (c) integrating diversi

ﬁ

ed rotations increases

food production, reduces GHG emissions, and bene

ﬁ

ts soil health

(Fig.

1

). Our results demonstrate that (a) instructive

ﬁ

ndings from

newly designed, tested, and validated diversi

ﬁ

ed systems could guide

the North China Plain in establishing a more sustainable system to
maintain or increase grain and protein production with reducing the
damage to the environment and soil ecosystems, and (b) the results
from such a representatively intensive food producing region may
provide a guide for the countries/regions with similar agricultural
environments to follow on an expanded scale.

Results

Diversi

ﬁ

ed crop rotations increase ecosystem productivity

The food crops evaluated in the study are from different genera
families; thus, the yield of each crop was converted to the

‘

equivalent

’

product yield to wheat for a compatible comparison (detailed in
Methods).

The

sweet

potato

(

Ipomoea

batatas

L.)

→

winter

wheat

–

summer maize rotation (SpWM) increased the annual equiva-

lent yield by 38% compared to the conventional winter wheat

–

summer

maize double-cropping which yielded 13,185 kg ha

–

1

annually (Fig.

a).

Rotations diversi

ﬁ

ed with sweet potato (SpWM), peanut (

Arachis

hypogea

L.) (PWM), or soybean (

Glycine max

L.) (SWM) signi

ﬁ

cantly

increased the annual economic bene

ﬁ

t (net income) compared to the

winter wheat

–

summer maize control (Fig.

b), with sweet potato

→

-

winter wheat

–

summer maize rotation increasing 60%, and pea-

nut

→

winter wheat

–

summer maize rotation and soybean

→

winter

wheat

–

summer maize rotation increasing by 13

–

22% (

P

<

0.05

). Protein

Intensified cereal
monoculture

Diversified rotation systems

2

4

273

27

direct

GWP

N O

CH

indirect

i

GWP

I

C

CO

2

N

2

O CH

4

Respiration

0

c

t

T

d

M P

f

H

V P T

d

/

1

2

44 12

SOC

T

n

i

d u e

Topsoil

id
u

e

i

d u e

Carbon

Social economic and environmental benefit

Irrigation
fertilizer

Greenhouse
gas footprint

Economic
benefits and
soil health

Irrigation, fertilizer, transport,

field operations

Fig. 1 | Schematic illustration of system integration from issues to outcomes.

In

the North China Plain

—

the case study area, traditional cereal monoculture (such as

wheat

–

maize double-cropping, i.e., two cereal crops per year) requires inputs of

synthetic agrichemicals and irrigation in food production, causing large green-
house gas (GHG) emissions; in contrast, rotation systems diversi

ﬁ

ed with cash and

legume crops can maintain crop yields, increase farmers

’

income, and reduce GHG

emissions due to the biological N

2

ﬁ

xation by legumes partly substituting for syn-

thetic N inputs. Legume-included rotations can also enhance soil health by stimu-
lating soil microbial activities, increasing carbon sequestration, and enhancing
nutrient cycles.

Article

https://doi.org/10.1038/s41467-023-44464-9

Nature Communications

| (2024) 15:198

2

yield varied among the rotation systems, with ryegrass (

Lolium multi-

ﬂ

orum

L.)

–

sorghum (

Sorghum bicolor

L.) rotation (RSWM) and soy-

bean

→

winter wheat

–

summer maize rotation having the highest

protein yields at 2274 and 1883 kg ha

–

1

yr

–

1

, respectively, being 8

–

31%

higher than winter wheat

–

summer maize rotation (Fig.

c). The

ryegrass-sweet sorghum

→

winter wheat

–

summer maize rotation, with

four crops in 2 years, produced the most aboveground biomass for
livestock, while the soybean

→

winter wheat

–

summer maize rotation

produced the most protein for human nutrition. In contrast, an
entirely

cereal-based

system

(i.e.,

spring

maize

→

winter

wheat

–

summer maize rotation, SmWM) decreased equivalent yield by

16% and protein yield by 23% compared to the winter wheat

–

summer

maize control.

The preceding crops in rotation signi

ﬁ

cantly affected grain yield

(Fig.

d), net income (Fig.

e), and protein yield (Fig.

f) of the suc-

ceeding winter wheat

–

summer maize crops as the rotation continued.

The diversi

ﬁ

ed crop rotations with sweet potato, peanut, and soybean

had a positive carryover effect on the productivity of the succeeding
winter wheat and summer maize. On average, winter wheat

–

summer

maize preceded by non-cereal crops in rotation showed increased
grain yield, economic bene

ﬁ

t, and protein yield by 26

–

32%, 39

–

46%,

and 25

–

29%, respectively, as compared to the winter wheat

–

summer

maize cereal monoculture.

Diversi

ﬁ

ed crop rotations reduce net GHG emissions

The amounts of supplementary irrigation and fertilizers applied to
each rotation varied among them, as did the N

2

O and CH

4

emissions

measured weekly from October 2016 to October 2022 (Supplementary
Figs. S1

–

2). The diversi

ﬁ

ed crop rotations signi

ﬁ

cantly reduced

annual cumulative N

2

O emissions compared to the wheat

–

maize

control which was at 8.9 ± 1.0 kg N ha

–

1

(Table S1, Fig.

a), decreasing

by 30%, 42%, and 49% in the peanut

→

wheat

–

maize rotation,

soybean

→

wheat

–

maize rotation, and sweet potato

→

wheat

–

maize

rotation, respectively. These rotations with peanut, soybean, and
sweet potato received a reduced dosage of N fertilizer as compared to
the wheat

–

maize control (Table S1). The soils of all crop rotations,

including wheat

–

maize, acted as net sinks for atmospheric CH

4

.

However, the diversi

ﬁ

ed crop rotations increased the sink strength by

33

–

76% above wheat

–

maize (Fig.

b). The wheat

–

maize control had

the highest annual global warming potential (GWP) for soil N

2

O and

CH

4

ﬂ

uxes (3764 kg CO

2

-eq ha

–

1

yr

–

1

, Fig.

c). The cereal-including

rotations (ryegrass-sweet sorghum

→

wheat

–

maize rotation and spring

maize

→

wheat

–

maize rotation) had 22% and 19%, respectively, lower

GWPs than wheat

–

maize rotation, while the rotations including

legumes (peanut, soybean) or sweet potato had 32%, 43%, and 51%,
respectively, lower GWPs than wheat

–

maize rotation. Besides

ﬁ

eld-

scale direct GHG emissions, indirect GHG emissions associated with
the use of agrochemicals and irrigation were highest for wheat

–

maize

rotation (8802 kg CO

2

-eq ha

–

1

yr

–

1

) and 34

–

41% lower for rotations with

sweet potato, peanut, soybean and spring maize (Fig.

d).

The soil carbon stocks (SOC) in the 0

–

90 cm layer signi

ﬁ

cantly

increased from 2016 to 2022 for all treatments, but the magnitude of
the change differed among rotation systems. Six years after the
initiation of the experiment (in 2022), the peanut

→

wheat

–

maize

rotation had the highest soil C sequestration (2.03 t ha

–

1

yr

–

1

),

followed by soybean

→

wheat

–

maize (1.91 t ha

–

1

yr

–

1

) and sweet

potato

→

wheat

–

maize

(1.44

t

ha

–

1

yr

–

1

),

while

ryegrass-sweet

sorghum

→

wheat

–

maize, spring maize

→

wheat

–

maize rotation, and

wheat

–

maize rotations had signi

ﬁ

cantly lower C sequestration rates

(0.21

–

0.69 t C ha

–

1

yr

–

1

) (Fig.

e). Soil C sequestration offset total GHG

WM

SpWM

PWM

SWM SmWM RSWM

0

5

10

15

20

25

30

E

qui

v

a

le

nt

y

ie

ld

(t

h

a

1

yr

1

)

a

b

bc

c

b

a

b

c

WM

SpWM

PWM

SWM SmWM RSWM

0

10000

20000

30000

40000

50000

60000

E

c

onomi

c

be

ne

fi

t

(Y

u

a

n

ha

1

yr

1

)

a

b

bc

c

bc

c

WM

SpWM

PWM

SWM SmWM RSWM

500

1000

1500

2000

2500

3000

3500

Pro

tein

y

ield

(k

g

h

a

1

yr

1

)

c

b

c

a

b

d

e

f

WM

SpWM

PWM

SWM SmWM RSWM

0

5

10

15

20

25

30

Grain

y

ield

(t

h

a

1

yr

1

)

Winter wheat
Summer maize

b

a

ab

b

WM

SpWM

PWM

SWM SmWM RSWM

0

10000

20000

30000

40000

50000

E

c

onomi

c

b

e

n

e

fi

t

(Y

ua

n

h

a

1

yr

1

)

Winter wheat
Summer maize

b

a

ab

b

WM

SpWM

PWM

SWM SmWM RSWM

0

500

1000

1500

2000

2500

3000

3500

P

ro

te

iny

ie

ld

(k

gh

a

1

yr

1

)

Winter wheat
Summer maize

b

a

ab

b

Fig. 2 | Agroecosystem productivity.

The entire rotation system productivity:

a

equivalent yield to wheat,

b

economic bene

ﬁ

t

–

net income, and

c

protein yield.

The productivity of WM preceded by alternative crops in rotation:

d

annual grain

yield,

e

economic bene

ﬁ

t, and

f

protein yield. In

a

–

c

, the black bar within violin

plots shows the 25 and 75 percentiles, the whiskers beyond the bar represent 95 and
5 percentiles, and the inner white dot indicates the median. The violin shaped area
reveals the data distribution. The dashed lines in

d

–

f

are the baseline values of the

WM rotation, and the error bars are the standard deviation of annual value summed
over winter wheat and summer maize. In

a

–

f

, one-way ANOVA with two-sided and

post-hoc test was conducted to determine signi

ﬁ

cant differences. Different

lowercase letters denote signi

ﬁ

cant differences between the rotations at

P

< 0.05.

The exact

P

values:

P

< 0.001 in

a

–

c,

P

= 0.001 in

d

, P

= 0.032 in

e

and

P

= 0.004 in

f

.

In

b

and

e

, $1 US = 6.95 Chinese Yuan (as of May, 2023). For

a

–

c

,

n

= 9;

d

–

f

,

n

= 18.

Treatment abbreviations: WM winter wheat

–

summer maize (control case), SpWM

sweet potato

→

winter wheat

–

summer maize rotation, PWM peanut

→

winter

wheat

–

summer maize rotation, SWM soybean

→

winter wheat

–

summer maize

rotation, SmWM spring maize

→

winter wheat

–

summer maize rotation, RSWM

ryegrass

–

sorghum

→

winter wheat

–

summer maize rotation. Source data are pro-

vided as a Source Data

ﬁ

le.

Article

https://doi.org/10.1038/s41467-023-44464-9

Nature Communications

| (2024) 15:198

3

emissions by 75

–

89% in the rotations with sweet potato, peanut, and

soybean and 7

–

21% in the spring maize

→

wheat

–

maize, ryegrass-sweet

sorghum

→

wheat

–

maize and wheat

–

maize rotations (Fig.

f). Conse-

quently, wheat

–

maize rotation had the largest net GHG emissions

totaling 10,025 kg CO

2

-eq ha

–

1

yr

–

1

, and the diversi

ﬁ

ed rotations with

sweet potato, soybean and peanut lowered net GHG emissions by 83%,
90%, and 92%, respectively, as compared to wheat

–

maize (

P

< 0.05).

Diversi

ﬁ

ed crop rotations enhance soil health and microbial

diversity

Soil health is often disregarded in the assessment of food crop pro-
duction systems. In the present study, we used the Cornell Soil Health
Assessment (CSHA) scoring method and applied Principal Component
Analysis (PCA) to evaluate how crop diversi

ﬁ

cation in

ﬂ

uenced soil

health. A soil health score was calculated for each rotation based on
key indicators (detailed in Methods). The PCA revealed that the

ﬁ

rst

two principal components accounted for 70.5% of the cumulative
percent variability (PC1 = 51.1% and PC2 = 19.4%) in soil health (Fig.

a),

with the measured soil health indicators clustered in distinct groups.
At the end of the 6-year study, the peanut

→

wheat

–

maize rotation had

the highest soil health score (59.5), followed by soybean

→

wheat

–

maize

rotation (56.8) and sweet potato

→

wheat

–

maize rotation (52.9)

(Fig.

b), being 41

–

59% higher than wheat

–

maize (33.2).

A closer examination of the soil indicators that contributed to soil

health scores revealed that the soil in the peanut

→

wheat

–

maize rota-

tion had a 6.5% higher SOC concentration and 29.7% higher dissolved
organic carbon (DOC) but 5.7% lower total N and 15.4% lower available
P (AP) concentration than wheat

–

maize (Fig. S3). These changes were

due to reduced N fertilizer (

−

41%) and associated higher nutrient use

ef

ﬁ

ciencies by crops in the diversi

ﬁ

ed rotations. The ryegrass-sweet

sorghum

→

wheat

–

maize rotation had the lowest SOC and DOC due to

the removal of forage for silage during the growing season, which
stimulated soil microbial activity, leading to the highest MBC (Fig. S3).

Soil microbial community composition and diversity

—

an indi-

cator of the biological mechanism of soil health

—

were determined in

2016 and 2022 to evaluate the cumulative changes after 6 years

’

rotational experimentation. It showed that the diversi

ﬁ

cation of crop

rotations had positive effects (Fig.

). For example, the Shannon index

—

a popular index for quantifying microorganism diversity taking into

account the number of species and relative abundance of each species
in a sample

—

signi

ﬁ

cantly increased by 7

–

10% in the rotations

with sweet potato, peanut, and soybean rotations, but did not
change in wheat

–

maize or the graminaceous crop-based spring

maize

→

wheat

–

maize and ryegrass-sweet sorghum

→

wheat

–

maize sys-

tems from 2016 to 2022 (Fig.

a). Inclusion of the legumes peanut or

soybean in rotations increased bacterial community operational
taxonomic units (OTU) richness (Fig. S4a

–

f) and fungal community

Shannon index, Chao 1, and OTU richness (

P

< 0.05) compared to

wheat

–

maize (Fig. S4g

–

l). Many of the soil health indicators are

dynamic and may change weekly, seasonally, or annually. However,
after 6 years of

‘

rotation nourishment,

’

soils in the rotations with sweet

potato, peanut, and soybean had signi

ﬁ

cantly higher values for Chao 1,

abundance-based coverage estimator (ACE), and community richness,
compared to the cereal monoculture spring maize

→

wheat

–

maize sys-

tem (Fig.

b

–

d).

Redundancy analyses, depicting the association of soil physio-

chemical properties and crop rotations in the bacterial (Fig.

e) and

fungal (Fig.

f) communities, found that the

ﬁ

rst two sets of compo-

nents explained 54.6% (bacterial) and 43.5% (fungal) of the total var-
iation in the two communities, respectively. The relationship was

-15

-10

-5

0

5

10

15

20

GHG

e

m

is

s

io

n

s

(t

CO

2

-eq

h

a

1

yr

1

)

Direct GHGs
Indirect GHGs

a

e

cd

d

cd

b

a

d

e

c

b

a

b

c

d

e

f

WM

SpWM

SWM

PWM

RSWM

SmWM

a

d

e

f

Soil carbon sequestration
Net GHG emissions

WM

SpWM PWM

SWM SmWM RSWM

2

4

6

8

10

12

Annual

m

ean

c

umulative

N

2

O

emission

(kg

N

2

O-N

h

a

1

yr

1

)

a

b

bc

cd

d

b

c

WM

SpWM

PWM

SWM SmWM RSWM

-3

-2

-1

0

Annual

m

ean

c

umulative

C

H

4

emssion

(kg

C

H

4

-C

ha

1

yr

1

)

a

b

bc

c

WM

SpWM

PWM

SWM SmWM RSWM

1000

2000

3000

4000

5000

6000

7000

Annual

m

ean

G

WP

N

2

O+CH

4

(kg

C

O

2

-eq

h

a

1

yr

1

)

a

d

cd

bc

b

WM

SpWM PWM

SWM SmWM RSWM

0

2500

5000

7500

10000

N

P

2

O

5

K

2

O

Diesel fuel

Electricity

Pesticide

Seeds

Labor

Indir

e

ct

G

H

G

e

missions

(kg

C

O

2

-eq

h

a

1

yr

1

)

0-30cm
30-50cm
50-70cm
70-90cm

2016 2022 2016 2022 2016 2022 2016 2022 2016 2022 2016 2022

WM

SpWM

PWM

SWM SmWM RSWM

0

20

40

60

80

100

Soil

car

b

on

stock

o

f

0

-90

c

m

(t

h

a

1

)

**

Fig. 3 | Direct and indirect greenhouse gas (GHG) emissions in the diversi

ﬁ

ed

crop rotations. a

Annual mean N

2

O emissions;

b

CH

4

emissions;

c

GWP

N2O+CH4

;

d

indirect GHG emissions;

e

changes in soil carbon stock in the 0

–

90 cm depth;

f

net GHG emissions. In

a

–

c

, diamond box plots show triangles spanning 25 and 75

percentiles meeting at the median. Small squares indicate means. Whisker lines
extend to the minimum and maximum. In

e

and

f

, error bars in each column

represent the standard deviation of whole-pro

ﬁ

le sums of three replicates. In

a

–

c

,

e

, and

f

, one-way ANOVA with two-sided and post-hoc test was conducted for

statistical signi

ﬁ

cance; different lowercase letters denote signi

ﬁ

cant differences

between the rotations at

P

< 0.05. The exact

P

values:

P

< 0.001 in

a

–

c

and

f

. In

e

,

** denotes signi

ﬁ

cant changes in soil carbon stock in the 0

–

90 cm soil layer

between 2016 and 2022 at

P

< 0.01 (

P

= 0.005 for SpWM,

P

= 0.002 for PWM and

SWM). Treatment abbreviations: WM winter wheat

–

summer maize (control case),

SpWM sweet potato

→

winter wheat

–

summer maize rotation, PWM peanut

→

winter

wheat

–

summer maize rotation, SWM soybean

→

winter wheat

–

summer maize

rotation, SmWM spring maize

→

winter wheat

–

summer maize rotation, RSWM

ryegrass

–

sorghum

→

winter wheat

–

summer maize rotation. Source data are pro-

vided as a Source Data

ﬁ

le.

Article

https://doi.org/10.1038/s41467-023-44464-9

Nature Communications

| (2024) 15:198

4

complex, but it was evident that bacterial and fungal community
composition changed from 2016 to 2022. The rotations with sweet
potato, peanut, and soybean had distinct bacterial community com-
positions from the other three systems (Fig.

e), whereas the

ﬁ

ve stu-

died diversi

ﬁ

ed rotations had fungal community compositions

distinctly different from the wheat

–

maize control (Fig.

f). The 6 years

of rotational nourishment affected soil physiochemical properties
interactively with soil microbial communities. The rotations with sweet
potato, peanut, and soybean bacterial communities were closely
associated with SOC and AP, whereas those of wheat

–

maize, spring

maize

→

wheat

–

maize, and ryegrass-sweet sorghum

→

wheat

–

maize

rotations were closely associated with TN and microbial biomass car-
bon (MBC). In contrast, the rotations with sweet potato, peanut, and
soybean fungal communities were associated with SOC, pH, and DOC.

Multiple functions assessments of diversi

ﬁ

ed crop rotations

We used the comprehensive evaluation index concept (

CEI

) to assess

the synergies and trade-offs of the different crop rotations related to
yield, nutritional value, soil-related indicators (health, C sequestration,
microbial biodiversity), net GHG emissions, and economic bene

ﬁ

t

(Eqs.

in Methods) (Fig.

a, d). The results showed that the

wheat

–

maize rotation had the lowest

CEI

value, averaging 0.19

(Fig.

b), while the sweet potato

→

wheat

–

maize rotation had the

highest

CEI

value of 0.81, followed by peanut

→

wheat

–

maize rotation of

0.75, and soybean

→

wheat

–

maize rotation of 0.69. The diversi

ﬁ

ed

rotations with sweet potato, peanut, and soybean had

CEI

values more

than triple relative to wheat

–

maize and were signi

ﬁ

cantly higher than

the cereal-based spring maize

→

wheat

–

maize and ryegrass-sweet

sorghum

→

wheat

–

maize rotations (0.25

–

0.29). Signi

ﬁ

cant relation-

ships occurred among the key indicators across the rotation systems
(Fig.

c). Soil health was positively corelated with crop yield (

r

= 0.79),

economic bene

ﬁ

ts (

r

= 0.85), and nutrition score (

r

= 0.90), but was

negatively correlated with net GHG emissions (

r

=

–

0.83). Soil health

was also positively correlated with soil carbon sequestration (

r

= 0.79)

and soil biodiversity (

r

= 0.67), contributing substantially to the trade-

off against net GHG emissions (

r

=

–

0.83).

Discussion

The empirical evidence from the 6-year

ﬁ

eld experiment demon-

strated signi

ﬁ

cant positive impacts of diversi

ﬁ

ed crop rotations on

various agroecosystem functions and services. Based on the study

’

s

ﬁ

ndings and double-cropping planted area

25

, we estimated that

diversi

ﬁ

ed cropping systems could reduce net CO

2

-eq emissions by

106.8 ± 31.7 million tonnes annually, offsetting 5.6% of the annual GHG
emissions associated with China

’

s food system (1.9 billion tonnes in

2020 reported by FAO

26

). The primary reason behind this reduction is

the decrease in synthetic N fertilizer use by 3.6 million tonnes. Farmers
in the North China Plain region could bene

ﬁ

t from 20% increases in

annual net income, equivalent to 84 billion Yuan ($11.6 US billion) in
total. Moreover, these diversi

ﬁ

ed rotations will increase Cornell Soil

Health Assessment scores by 45%, due to increased C sequestration
rates compared to the winter wheat

–

summer maize rotation. Fur-

thermore, wheat and maize yields may increase by 32%, equivalent to
73.5 million tonnes per year if planted following alternative crops such
as sweet potato, soybean or peanuts; this would make about 36.1
million tonnes of additional straw biomass available annually for
alternative uses, such as feed, bioenergy, or enhancing soil carbon
stocks.

The three-dimensional bene

ﬁ

ts (grain and protein yields, GHG

emissions, and soil health) of crop diversi

ﬁ

cation in the winter

wheat

–

summer maize rotation system on the North China Plain will

add signi

ﬁ

cant

‘

social novelty

’

to the challenge of bringing essential

food nutrients to dining tables without adversely affecting soil health.
The social novelty is far beyond the agricultural bene

ﬁ

ts of crop

diversi

ﬁ

cation that are broadly recognized worldwide. For example,

long-term studies in North America show that rotational diversi

ﬁ

cation

increases crop yields, reduces the impact of adverse weather on eco-
system productivity

27

, and enhances system robustness

28

. Studies in

Europe and Africa demonstrate that

ﬁ

eld-scale diversity has a sub-

stitutive interaction with fertilization leading to increased yields at low
N fertilizer doses

29

–

32

. Several Chinese studies implementing crop

diversi

ﬁ

cation in various forms or scales show that farmers can adopt

multiple strategies to increase farming resilience and economic

b

PC1 (51.09%)

-1.0

1.0

AP

TN

SOC

pH

NO

3

-

MBC

MBN

BD

SWC

-0.5

0.0

0.5

-1.0

1

.0

-0.5

0.0

0.5

DOC

WM
SpWM
PWM

SWM
SmWM
RSWM

a

PC2

(1

9.

44%)

2016

WM

SpWM

PWM

SWM SmWM RSWM

0

20

40

60

80

100

Soil

health

score

f

e

c

a

b

d

c

+13%

+59%

+79%

+71%

+43%

+54%

+41%

+59%

+52%

+27%

+36%

Fig. 4 | Soil health assessment for different rotation systems. a

Principal com-

ponent analysis (PCA) for eigenvalues of the ten key soil properties, including soil
organic carbon (SOC), dissolved organic carbon (DOC), total nitrogen (TN), NO

3

–

-N,

available P (AP), pH, microbial biomass N (MBN), microbial biomass carbon (MBC),
bulk density (BD), and soil water content (SWC). Each soil property was normalized
as an individual CSHA (Cornell Soil Health Assessment) contributor to the overall
soil health score.

b

CSHA score for each rotation. In

b

, the whiskers in each column

represent the standard deviation of three replications; circles are the corre-
sponding data points for the bar chart,

n

= 3. One-way ANOVA with two-sided and

post-hoc test was conducted for the signi

ﬁ

cance test among different treatments.

Different lowercase letters denote signi

ﬁ

cantly different groups of crop rotations at

P

< 0.05. The red dashed line with red percentages is the baseline of the WM

rotation. The black dashed line with black percentages is the baseline of 2016.
Treatment abbreviations: WM winter wheat

–

summer maize (control case), SpWM

sweet potato

→

winter wheat

–

summer maize rotation, PWM peanut

→

winter

wheat

–

summer maize rotation, SWM soybean

→

winter wheat

–

summer maize

rotation, SmWM spring maize

→

winter wheat

–

summer maize rotation, RSWM

ryegrass

–

sorghum

→

winter wheat

–

summer maize rotation. Source data are pro-

vided as a Source Data

ﬁ

le.

Article

https://doi.org/10.1038/s41467-023-44464-9

Nature Communications

| (2024) 15:198

5

pH**

SOC*

DOC**

MBC

MBN

AP

MBC/MBN

TN

NO

3

-

-N

NH

4

+

-N

Fungi

1.0

RDA1 (25.81%)

-0.5

0.0

0.5

-0.8

0.0

-0.4

0.8

0.4

-1.0

2016

WM
SpWM
PWM
SWM

RSWM

SmWM

2022

RDA2 (17.70%

)

pH

SOC*

DOC

TN*

MBC*

MBN

MBC/MBN*

AP*

-1.0

1.0

RDA1 (29.40%)

RDA2 (25.15%

)

-0.5

0.0

0.5

-0.8

0.0

2016

WM
SpWM
PWM
SWM

RSWM

SmWM

NO

3

-

-N**

NH

4

+

-N

Bacteria

-0.4

0.8

0.4

2022

a

b

c

d

2022

e

f

2016

WM

SpWM

PWM

SWM

SmWM RSWM

5000

6000

7000

8000

c

ab

a

b

+18%

+20%

+22%

p

= 0.001

+12%

+14%

Chao1

2016

WM

SpWM

PWM

SWM

SmWM RSWM

5.0

5.5

6.0

6.5

7.0

d

cd

ab

abc

a

bcd

d

+3%

+8%

+7%

+10%

p

= 0.002

+4%

+1%

Shannon

2016

WM

SpWM

PWM

SWM

SmWM RSWM

4000

5000

6000

7000

c

ab

a

b

ab

+20%

+22%

+24%

p

= 0.0003

+16%

Richness

2016

WM

SpWM

PWM

SWM

SmWM RSWM

5000

6000

7000

8000

c

ab

a

b

ab

+18%

+20%

+22%

p

= 0.001

+13%

+14%

ACE

Fig. 5 | Soil microbe community diversity and soil properties in response to
rotation. a

Shannon Index;

b

Chao 1;

c

ACE index;

d

OTU richness measured in the

six crop rotations from 2016 to 2022;

e

Redundancy analysis (RDA) of bacterial OTU

and soil properties;

f

RDA of fungal OTU and soil properties. The percentage values

in

a

–

d

represent the increment from 2016 to 2022; dot is the mean value, whisper

line bars on each dot are the standard deviations for three replications. In

a

–

d

, one-

way ANOVA with two-sided and post-hoc test was conducted to determine sig-
ni

ﬁ

cant differences among treatments. Different lowercase letters below pillars

indicate signi

ﬁ

cant differences between rotations at

P

< 0.05. Charts

e

and

f

show

redundancy analysis with colored dots and triangles indicating rotation, vectors

indicating soil properties including soil organic carbon (SOC), dissolved organic
carbon (DOC), total nitrogen (TN), NO

3

–

-N, NH

4

+

-N

,

available P (AP), pH, microbial

biomass carbon (MBC), microbial biomass N (MBN), and the MBC to MBN ratio.
Treatment abbreviations: WM winter wheat

–

summer maize (control case), SpWM

sweet potato

→

winter wheat

–

summer maize rotation, PWM peanut

→

winter

wheat

–

summer maize rotation, SWM soybean

→

winter wheat

–

summer maize

rotation, SmWM spring maize

→

winter wheat

–

summer maize rotation, RSWM

ryegrass

–

sorghum

→

winter wheat

–

summer maize rotation. Source data are pro-

vided as a Source Data

ﬁ

le.

Article

https://doi.org/10.1038/s41467-023-44464-9

Nature Communications

| (2024) 15:198

6

bene

ﬁ

. Overall, cropping diversi

ﬁ

cation offers a comprehensive

systems

’

approach to enhance agroecosystem productivity and

adaptability to changing climates worldwide.

A major factor in designing diversi

ﬁ

ed cropping systems is to

ensure an increase in soil C sequestration

—

as this plays a crucial role

for soi health, in climate change mitigation and achieving C
neutrality

34

35

. We found that the diversi

ﬁ

ed rotations with sweet

potato, peanut and soybean signi

ﬁ

cantly increased soil C after 6 years

of rotation nourishment compared to the traditional winter
wheat

–

summer maize double-cropping system. Introducing sweet

potatoes and legumes into these rotations has been shown to stimu-
late microbial growth, increased C use ef

ﬁ

ciency

36

, and promoted the

formation of highly stable mineral-associated soil C complexes

37

Labile litter compounds from these rotational crops are the dominant

WM

SpWM

PWM

SWM

SmWM

RSWM

Soil
biodiversity
(+)

Nutrition
yield (+)

Soil health (+)

Net GHG
emissions (-)

Soil carbon
sequestration (+)

Equivalent yield (+)

Economic
benifit (+)

0

10

20

30

Equivalent
yield

Soil health

Soil
biodiversity

Soil carbon
sequestration

Net GHG
emssions

Nutrition
yield

Economic
benefit

WM

Equivalent
yield

Soil health

Soil
biodiversity

Soil carbon
sequestration

Net GHG
emssions

Nutrition
yield

Economic
benefit

SpWM

Equivalent
yield

Soil
biodiversity

Soil carbon
sequestration

Net GHG
emssions

Nutrition
yield

Economic
benefit

Soil health

PWM

Equivalent
yield

Soil
biodiversity

Soil carbon
sequestration

Net GHG
emssions

Nutrition
yield

Economic
benefit

Soil health

SWM

Equivalent
yield

Soil
biodiversity

Soil carbon
sequestration

Net GHG
emssions

Nutrition
yield

Economic
benefit

Soil health

RSWM

Equivalent
yield

Soil
biodiversity

Soil carbon
sequestration

Net GHG
emssions

Nutrition
yield

Economic
benefit

Soil health

SmWM

a

b

c

d

WM

SpWM PWM

SWM SmWM RSWM

0.0

0.2

0.6

0.8

1.0

CEI

Equivalent

yield

Economic

benefit

Nutrition

yield

Net GHG

emissions

Soil

carbon

sequestration

Soil

biodiversity

Soil

health

Equivalent yield

Economic benefit

Nutrition yield

Net GHG emissions

Soil carbon

sequestration

Soil biodiversity

Soil health

***

*

**

*

**

*

**

*

**

*

0.98

0.78

0.81

-0.43

-0.57

-0.77

0.56

0.45

0.77

-0.93

0.40

0.52

0.65

-0.95

0.93

0.79

0.85

0.90

-0.83

0.79

0.67

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

Equivalent
yied

Economic
benifit

Nutrition
yield

Net GHG
emissions

Soil carbon
sequestration

Soil
biodiversity

Soil
health

Fig. 6 | Comprehensive assessments of the different rotation systems. a

radar

map of multiple objective analysis to assess the various functions of the rotations;

b

comprehensive evaluation index (

CEI

, Eqs.

15

–

24 in Method, supplementary

material);

c

synergy and trade-off relationship among the key variables across the

different crop rotations;

d

, Nightingale Rose Charts of multiple objectives analysis

to assess the detailed functions 1for each crop rotation. In

a

, values are normalized.

(+), positive variables, (

–

), negative variables. Equivalent yield re-expresses crop

yield as equivalent to the grain yield of wheat; Economic bene

ﬁ

t denotes the crop

yield multiplied by its market price minus consumable cost; Nutrition yield includes
17 nutritional components of each crop such as crude protein, fat content, macro-
and micro-elements, and vitamins

68

. Soil carbon sequestration denotes changes in

soil carbon stock in the 0

–

90 cm layer from October 2016 to October 2022; Net

GHG emissions denotes the net GHG budgets including direct and indirect GHG
emissions minus soil carbon sequestration; Soil biodiversity is represented by the
Shannon Index of bacterial and fungal communities; Soil health score is calculated
from 10 indicators using the overall Cornell Soil Health Assessment (CSHA) matrix
combined with PCA

81

83

. Treatment abbreviations: WM winter wheat

–

summer

maize (control case), SpWM sweet potato

→

winter wheat

–

summer maize rotation,

PWM peanut

→

winter wheat

–

summer maize rotation, SWM soybean

→

winter

wheat

–

summer maize rotation, SmWM spring maize

→

winter wheat

–

summer maize

rotation, RSWM ryegrass

–

sorghum

→

winter wheat

–

summer maize rotation. Source

data are provided as a Source Data

ﬁ

le.

Article

https://doi.org/10.1038/s41467-023-44464-9

Nature Communications

| (2024) 15:198

7

source of microbial products that affect stable SOC with a high capa-
city for C stabilization

38

. Furthermore, our

ﬁ

ndings highlight a close

correlation between soil health and nutrition yield, underscoring the
interconnectedness of both dimensions and their reliance on agri-
cultural practices

39

A noteworthy aspect of the study is that we measured SOC con-

centrations to 90 cm depth. Researchers often focus on topsoil layers
(0

–

20 cm) when studying SOC status, potentially masking some

treatment effects, especially in crops with different root masses at
various soil depths. We found that SOC in the 0

–

20 cm soil layer

accounted for 51% of the total SOC, 0

–

30 cm layer accounted for 67%,

and 0

–

50 cm layer accounted for 80%. Our results highlight the

importance of subsoil C stocks when assessing soil C storage for a
more precise estimation of the contributions of soil C sequestration in
mitigating climate change.

Lowering GHG emissions without reducing crop yields is a major

challenge for global agriculture. We found that the legume-based
(peanut

→

wheat

–

maize, soybean

→

wheat

–

maize) rotations signi

ﬁ

cantly

reduced N

2

O emissions by 30

–

42% compared to the conventional

cereal-based wheat

–

maize rotation while signi

ﬁ

cantly increasing sys-

tem productivity (grain and protein yields). Legume crops improve soil
N cycling

40

, but may also increase N

2

O emissions, speci

ﬁ

cally following

residue incorporation

41

42

. However, legume cultivation requires sig-

ni

ﬁ

cantly less fertilizer than cereal crops such as maize or wheat.

Therefore, the legume-based rotations with peanut and soybean
required 37% less fertilizer than conventional wheat

–

maize, explaining

the overall reduction in N

2

O emissions for the legume-based rotations

as N fertilizers were the main source of N

2

O emissions in our study

40

The manufacture of inorganic N fertilizer remained the most sig-
ni

ﬁ

cant contributor to total GHG emissions (about 36

–

39%). Similarly,

we found that the diversi

ﬁ

ed rotations with sweet potato, soybean, and

peanut signi

ﬁ

cantly reduced net GHG emissions by 75

–

92% compared

to the wheat

–

maize, spring maize

→

wheat

–

maize, and ryegrass-sweet

sorghum

→

wheat

–

maize systems. In a study in Western Canada, cereals

rotated with legumes decreased net GHG emissions by 17

–

35% com-

pared to cereal monocultures

43

due to improved N uptake associated

with organic N mineralization released from legume residues

44

. In

Europe, increasing legume production by introducing legumes into
cereal rotations effectively improved plant-based protein while redu-
cing environmental impacts

45

,

46

. In the US corn belt, diversi

ﬁ

ed maize

and soybean systems maintained or increased maize yields while
reducing N losses to the environment

47

.

Introducing shallow-rooted crops like sweet potato and annual

legumes in rotation with deep-rooted crops such as winter wheat
enables the use of soil nutrients across the entire rooting zone,
improving nutrient use ef

ﬁ

ciency and reducing soil nutrient losses

through leaching. Alternating deep- and shallow-rooted crops also
enhances root distribution, increases soil porosity and permeability,
reduces soil bulk density

48

, and improves soil aggregate stability

49

to

anchor SOC

50

. Furthermore, diversi

ﬁ

ed crop mixtures alter the quality

and quantity of organic matter entering the soil as plant litter,
enhancing the stabilization of microbial-derived compounds that
provide positive feedback for plant growth. The distinct microbial
communities associated with diverse crop mixtures can promote soil C
cycling

51

. In our study, diversi

ﬁ

ed legume

–

cereal and sweet

potato

–

cereal rotations signi

ﬁ

cantly increased the

α

-diversity of soil

bacterial and fungal communities after 6 years of rotation. The
increased fungal diversity resulted from a higher number of species,
while the increased Shannon

’

s diversity index in the bacterial com-

munity resulted from changes in the relevant abundance of certain
species. Diverse crops stimulate bacterial community activities in the
rhizosphere

52

, as the residue root exudates and plant litter from rota-

tional crops provide a greater diversity of residue carbon substrates,
supporting the growth of diverse soil microorganisms

53

. We found that

the rotations with sweet potato, peanut, and soybean signi

ﬁ

cantly

increased

fungal

community

diversity

compared

to

winter

wheat

–

summer maize, shaping soil nutrient pro

ﬁ

les with different

carbon substrates that impacted microbial diversity. Moreover,
diversi

ﬁ

ed crop rotations potentially reduce soil agrochemical con-

tamination due to decreased application of synthetic N fertilizer and
herbicides, favoring soil microenvironments

54

,

55

. Other studies have

demonstrated the positive in

ﬂ

uence of diversi

ﬁ

ed rotations on soil

health. Maize yields and SOC content in diversi

ﬁ

ed rotations were

more resilient than those found in monoculture

56

, potentially

increasing adaptability to climate change

57

. Diversi

ﬁ

ed agroecosys-

tems are considered an economically viable alternative to business-as-
usual maize

–

soybean rotations in the Midwest US

58

, with cover crops,

perennials, and small grain cereals enhancing soil health by 32

–

49%

and crop productivity by 16

–

29%

59

The novelty of this study is its highly diversi

ﬁ

ed crop rotations,

with cash crops and legumes replacing cereal monocultures, together
with rotating shallow-rooted crops with deep-rooted crops. It posi-
tively affects ecosystem services and functions, partly compensating
for the lower services in less diversi

ﬁ

ed cereal-based rotations. Crop

diversi

ﬁ

cation offers signi

ﬁ

cant bene

ﬁ

ts in reducing GHG emissions

and has a synergistic effect on plant biomass and protein production,
soil health, and microbial community biodiversity. Diversi

ﬁ

cation

reduced crop inputs, particularly N fertilizer and irrigation amounts,
minimizing environmental costs while increasing farmers

’

incomes.

Thus, diversi

ﬁ

ed crop rotations can serve as an effective strategy for

sustainable food production in areas worldwide with environments
similar to the North China Plain.

The scienti

ﬁ

c evidence presented in this study holds signi

ﬁ

cant

potential for informing and reinforcing current policies and incentive
schemes to improve the green transition of farming systems in China.
Speci

ﬁ

cally, China set objectives to promote soybean production

(initiated in 2019), implement an action plan for zero increase in fer-
tilizer and pesticide use (launched in 2015), and develop strategies to
reduce agricultural GHG emissions with the ultimate goal of trans-
forming agricultural systems into zero

‘

emitters

’

by 2060. When

designing effective diversi

ﬁ

ed cropping systems, it is important to

consider local environmental conditions to balance land use for food
production with the other functions of an ecosystem. Optimizing crop
con

ﬁ

guration is essential

—

through social, economic, and environ-

mental dimensional planning within a complex framework of various
indicators to ensure long-term sustainability. We recommend that
developing and adopting diversi

ﬁ

ed cropping systems should be a key

consideration in agricultural policy setting and a top priority for on-
farm decision-making, as they are critical for achieving long-term food
production resilience, maintaining soil health, and ensuring environ-
mental sustainability. Given that climate variability will increase, multi-
year assessments of the sustainability of agricultural production sys-
tems remain tentative, and funding should be sought for additional
observation sites for long-term monitoring.

Methods

The

ﬁ

eld experiment was established in October 2016 at the Luan-

cheng Agro-Ecosystem Station of the Chinese Academy of Sciences
(37°50

′

N, 114°40

′

E; altitude, 50.1 m) in Luancheng County, Hebei

Province (Fig. S5a), which represents the agricultural production and
climate conditions of the North China Plain. The experimental area
experiences a warm, temperate zone, semi-humid, monsoon climate
with a frost-free period of 200 days. The annual mean air temperature
was 14.7 ± 1.0 °C over the last 20 years, and the annual average pre-
cipitation was 472 ± 161 mm over the last 62 years (Fig. S5b). Fig. S5c
presents the average monthly precipitation and air temperatures from
2016 to 2020. The experimental site had loam soil with sandy loam in
the surface layers, light/medium loam at 40

–

80 cm depth, and light

clay below 80 cm. The 0

–

10 cm soil layer before the experiment start

has pH 7.6 ± 0.2, 240 ± 42

µ

S cm

–

1

EC, 1.49 ± 0.07 g cm

–

3

bulk density,

Article

https://doi.org/10.1038/s41467-023-44464-9

Nature Communications

| (2024) 15:198

8

0.42 ± 0.03 cm

3

cm

–

3

ﬁ

eld capacity, 0.10 ± 0.02 cm

3

cm

–

3

wilting point,

1.06 ± 0.14 g kg

–

1

total N, 11.5 ± 0.4 g kg

–

1

SOC, 9.3 ± 3.1 mg kg

–

1

available

phosphorous, and 109.6 ± 24.1 mg kg

–

1

available potassium. Prior to

the experiment, the core food crops on the same

ﬁ

elds were wheat and

maize grown in a 12-month double-cropping system that is popular
regionally. From the start of the experiment, the sites were managed
with optimized irrigation and seeding techniques, with all residues
returned to the soil during the winter wheat

–

summer maize year.

The experiment comprised eight crops: cereals winter wheat (cv.

Kenong 2009), summer maize (cv. Jundan 20), and spring maize (cv.
Jundan 20); legumes soybean (cv. Shidou 12) and peanut (cv. Jihua 4);
cash/forage: sweet potato (cv. Shangshu 19), ryegrass (cv. Dongmu 70)
and sorghum (cv. Jintianza 3). There were

ﬁ

ve newly designed diversi

ﬁ

ed

crop rotations: [sweet potato

→

winter wheat

–

summer maize rotation

(SpWM, 2-year cycle); peanut

→

winter wheat

–

summer maize rotation

(PWM, 2-year cycle); soybean

→

winter wheat

–

summer maize rotation

(SWM, 2-year cycle); ryegrass-sweet sorghum

→

winter wheat

–

summer

maize

rotation

(RSWM,

2-year

cycle);

spring

maize

→

winter

wheat

–

summer maize rotation (SpWM, 2-year cycle)]. The regionally

dominant winter wheat

–

summer maize double-cropping system (WM,

1-year cycle) was the control.

‘

→

’

denotes crop rotation across years and

‘–’

denotes crop rotation within a year. Unlike pure winter

wheat

–

summer maize, the alternative rotations included fallow seasons

for parts of their non-winter wheat

–

summer maize years (Fig. S6).

Each rotation was replicated three times, resulting in 18 plots

established using a randomized complete block design (Fig. S6). All
cycles were repeated on their respective plots from October 2016 to
October 2022. The winter wheat

–

summer maize double-cropping

control rotation completed six cycles, while the other

ﬁ

ve rotations

completed three rotation cycles. Supplementary Table S1 lists the total
amounts of inputs, including N, P, K, irrigation, diesel for tillage,
electricity for irrigation, pesticide, seeds, and labor for each crop in
each rotation. Supplementary Data 1 details fertilizer and irrigation
amounts and timings for each crop in each rotation in each year. Basal
N applications (as urea) were broadcast and plowed (20 cm depth) into
the soil before seeding. Topdressing of N was performed at the key
growth stages with speci

ﬁ

c amounts summarized in Supplemen-

tary Data 1.

Equivalent yield and economic bene

ﬁ

t

Crops were harvested at maturity with a conventional combine har-
vester with an automated weighing system, which measured the total
crop yield of each plot. After air-drying to a constant moisture content
(13%), wheat and maize grain yields were recorded

60

61

. Sorghum bio-

mass yield (fresh weight) per plot was weighed to represent silage
moisture of 30% (70% dry matter)

61

. Ryegrass dry matter was deter-

mined in 1 × 1 m sub-sample after drying at 85 °C until constant
weight

62

. Sweet potato yield (fresh weight) was obtained by weighing

the fresh root tubers harvested from each plot using a single row
harvester

63

. All mature pods in each plot were collected to determine

peanut yield accurately. Ten uniform peanut plants were used for
investigating the pod numbers per plant. All harvested peanut pods
were air-dried to a water content of approximately 14% and weighed to
calculate the pod yield

64

. After harvesting each plot, soybean seeds

were cleaned and weighed, with the grain yields adjusted to 13.0%
moisture content

65

For yield comparisons among different crop products, the

equivalent yields of the six crop rotations were calculated by multi-
plying the crop yield by the market price relative to that of winter
wheat in the same year

66

67

. Each cropping system

’

s equivalent yield,

economic bene

ﬁ

t, and protein yield were determined as follows:

EY

ij

=

Y ield

ij

×

Pr

ice

ij

W heat price

j

ð

1

Þ

EB

ij

=

Y ield

ij

× Pr

ice

ij

×

CPI

j

CPI

2008

Cos

t

ij

ð

2

Þ

PC

i

=

Y ield

i

×

β

i

ð

3

Þ

where

EY

ij

is equivalent yield of crop

i

at year

j

(kg ha

–

1

), wheat price

j

is

price of winter wheat in year

j

adjusted for China National Consumer

Price Index (CPI, relative Yuan), economic bene

ﬁ

t (

EB

) is within-year

crop yield multiplied by its corresponding price in Chinese Yuan minus
crop production costs (Yuan ha

–

1

),

j

is a given year,

PC

i

is total protein

content of crop

i

(kg ha

–

1

), and

β

i

is protein concentration per kg crop

product (%) referred from previous studies

67

,

68

. (For comparison, the

following exchange rate was used: $1 US = 6.95 Chinese Yuan and 1

€

= 7.60 Yuan in May 2023.) Each indicator for one crop rotation was

summed across the crops within one crop rotation cycle. The CPI was
used to adjust crop prices across years to eliminate in

ﬂ

ation or

de

ﬂ

ation as a source of variability between cases in different years.

Table S2 lists the crop prices and CPI values for individual years and
crops, and Table S3 lists each crop

’

s protein content.

Gas

ﬂ

ux and net GHG emissions

Soil GHG (N

2

O and CH

4

)

ﬂ

uxes were measured using the static cham-

ber method

69

. Plastic frames with water-sealing grooves were inserted

5 cm into the soil between crop rows, and chambers (20 × 20 × 30 cm)
were placed on top of the frames during measurements. For

ﬂ

ux

measurements, gas samples (>30 mL per sample) were taken between
9:00 and 10:00 am at speci

ﬁ

c time intervals (0, 12, 24, and 36 min after

chamber closure) using 100 mL plastic syringes attached to a three-
way stopcock. Gas samples were collected about once every 7 days
from sowing to harvest, with more frequent measurements (every
3 days) after N fertilizer application and precipitation events. The
headspace air temperature in the chamber was recorded with a ther-
mometer. Gas samples were analyzed using a gas chromatograph
(Agilent 7890 A, Agilent Inc., USA) equipped with a

ﬂ

ame-ionization

detector for measuring CH

4

at 200 °C and an electron capture detector

for measuring N

2

O at 330 °C. The column temperature was set to 55 °C.

CH

4

was separated with a Porapak Q column and detected by the

ﬂ

ame-ionization detector. N

2

O was detected by a micro electron cap-

ture detector (

μ

ECD). The carrier gas was high-purity N

2

(99.999%)

with a

ﬂ

ow velocity of 20 mL min

–

1

. A purge gas (5% CO

2

in N

2

) was used

to avoid interference with CO

2

. The gas

ﬂ

ux was calculated using the

linear approach. Soil N

2

O and CH

4

ﬂ

uxes for each rotation across the 6

years are illustrated in Figs S1 and S2.

For the net GHG emissions (kg CO

2

-eq ha

–

1

) estimation, we con-

sidered indirect GHG emissions of crop inputs from a life cycle
assessment, direct N

2

O and CH

4

emissions from the soil, and changes

in soil carbon stock from 0

–

90 cm during the 6-year experiment. The

system boundary in the life cycle assessment was set from manu-
facture, storage, and delivery of external inputs such as fertilizer to
crop harvest, including soil carbon storage change (Fig. S7). Potential
emissions related to the logistics of transporting, exporting, or mar-
keting grain products beyond the farm gate were excluded as they
were considered outside the system boundary.

Indirect GHG emissions (

CE

) result from manufacturing, storing,

transporting, and delivering agricultural inputs, such as fertilizer (N, P,
K), pesticides, diesel, and electricity to the farm gate, as follows:

CE

=

X

n

k

= 1

ð

D

k

×

C

k

Þ

ð

4

Þ

where

CE

(kg CO

2

-eq ha

−

1

yr

–

1

) is the indirect GHG emissions for one

crop within a given rotation,

D

k

is the amount of the

k

th

input (Table S1),

and

C

k

is the corresponding CO

2

-eq emission coef

ﬁ

cients (see Table S4

for the manufacture, storage, and delivery of agricultural inputs).

Article

https://doi.org/10.1038/s41467-023-44464-9

Nature Communications

| (2024) 15:198

9

Direct N

2

O and CH

4

ﬂ

uxes from soil were calculated from October

2016 to October 2022, as follows:

f

=

M

V

0

T

0

T

P

0

H

d

c

d

t

× 60

ð

5

Þ

where

f

(ug m

–

2

h

–

1

) is GHG emission

ﬂ

uxes of N

2

O and CH

4

emissions,

M

is the molecular mass of the speci

ﬁ

ed gas (44 g mol

–

1

for N

2

O and

16 g mol

–

1

for CH

4

),

V

0

(m

3

mol

−

1

) is the molar volume of an ideal gas,

T

0

(K) is the ideal gas temperature,

T

(K) and

P

(kPa) are the air tem-

perature and pressure in the chamber at the sampling time,

P

0

(kPa) is

the standard air pressure,

H

(m) is the height of chamber,

d

c

/d

t

(ppm

min

–

1

) is the slope of the linear regression curve for the change of

headspace gas concentrations during times of chamber closure, and
60 is min h

−

1

. Total GHG emissions during each crop growth and fallow

period were calculated using linear interpolation

70

.

Cumulative N

2

O and CH

4

emissions during each crop

’

s growing

season or fallow period were calculated as:

F

=

X

n

i = 1

ð

f

i

+ 1

+

f

i

Þ

2

×

ð

t

i

+ 1

t

i

Þ

× 24

ð

6

Þ

where

F

is the cumulative emission of one of the two gases from one

experimental plot (kg ha

–

1

) over a speci

ﬁ

ed period,

f

is the gas emission

ﬂ

ux (kg ha

–

1

h

–

1

),

t

is the sampling time in days,

i

is the

i

th

sampling from

the plot, and 24 is the conversion coef

ﬁ

cient between hours and days

(h day

–

1

).

The global warming potential (GWP) of seasonal soil N

2

O and CH

4

emissions, combining Eq. (

7

) results and re-expressed in CO

2

-eq

71

, was

calculated as:

GW P

N

2

O

+

CH

4

=

N

2

O

× 273 +

CH

4

× 27

ð

7

Þ

The annual change in SOC stock (

Δ

C

; kg CO

2

-eq ha

–

1

yr

–

1

) from

October 2016 to October 2022 was estimated for each crop rotation, as
follows:

Δ

C

=

ð

T

1

T

2

Þ

×

44

12

n

ð

8

Þ

where

T

1

and

T

2

(t ha

–

1

) are the total soil carbon storage values in the

0

–

90 cm soil layer in October 2016 and October 2022, respectively,

n

is

the number of years of the study period, and 44/12 is the conversion
coef

ﬁ

cient from C into CO

2

.

Net GHG emissions were equal to the sum of indirect (

CE

) and

direct (GWP of N

2

O and CH

4

) GHG emissions minus soil carbon

sequestration (

Δ

C

) when soil carbon was the sink, calculated as follows:

Net GHGs

=

CE

+

GW P

N

2

O

+

CH

4

Δ

C

ð

9

Þ

Soil physiochemical properties, microbial diversity and soil
health scoring

Soil samples (0

–

20 cm depth) were collected at the beginning of the

experiment (October 2016) and at the end of the summer maize har-
vest in October 2022 to measure soil physicochemical properties:
including soil pH, bulk density, soil water content, total nitrogen (TN),
dissolved organic carbon (DOC), soil nitrate-N (NO

3

–

-N), ammonium-N

(NH

4

+

-N), available phosphorus (AP), and microbial biomass carbon

and nitrogen (MBC and MBN, respectively).

Soil pH was measured in a 1:5 (w/v) soil: water suspension as

described by McLean (1982)

72

. TN was determined using the mod-

i

ﬁ

ed Kjeldahl method

73

. MBC and MBN were determined using

the chloroform fumigation and extraction method

74

. DOC con-

centrations extracted from fumigated and unfumigated soil

samples were determined using a Multi 3100 N/C TOC analyzer
(Analytik Jena, Germany). A modi

ﬁ

ed Bremner

’

s standard protocol

was used to extract soil NO

3

–

-N and NH

4

+

-N using 2 M KCl

75

, mea-

sured with a UV-1800 spectrophotometer (Mapada Instruments,
Shanghai, China). Soil available P concentrations were determined
using 0.5 mol L

–

1

NaHCO

3

(pH 8.5) and the molybdenum-blue col-

orimetric method

76

.

Soil organic carbon (SOC) concentrations in the 0

–

90 cm soil

layer (0

–

10, 10

–

20, 20

–

30, 30

–

50, 50

–

70, and 70

–

90 cm increments)

were measured after each crop harvest from October 2016 to October
2022, in samples collected by auger at

ﬁ

ve random locations per plot.

Soil samples from each sampling depth within a plot were combined,
dried at 105 °C for 24 h, sieved (2 mm), pretreated with 0.5 M HCl to
remove carbonates

77

, and ball-milled. SOC concentrations were

determined using a vario Macro CNS Analyzer (Elementar, Germany).
Soil bulk density in all soil layers down to 0.9 m in all plots was mea-
sured using the gravimetric method

78

Soil samples were also collected before the start of the

experiment (October 2016) and at the

ﬁ

nal summer maize harvest

(October 2022) for DNA sequencing and microbial community
composition and diversity measurements. Subsamples for DNA
sequencing were immediately placed on dry ice and stored at

–

80 °C

until DNA extraction, while the other subsamples were taken to the
laboratory for chemical analysis. Soil DNA was extracted from 0.5 g
freeze-dried soil using the Powersoil DNA Isolation Kit (Mo Bio
Laboratories, Carlsbad, CA, USA) according to the manufacturer

’

s

protocol. DNA quality was assessed according to the 260/280 nm
and 260/230 nm absorbance ratios using a NanoDrop ND2000
spectrophotometer (NanoDrop, ND2000, Thermo Scienti

ﬁ

c, 111

Wilmington, DE) and then stored at

–

80 °C for further molecular

biology analysis.

Primer pairs

—

barcode-515F/806 R (5

′

-GTGCCAGCMGCCGCGGTAA-

3

′

/5

′

-GCACTACHVGGGTWTCTAAT-3

′

)

—

were used to amplify the V3 + V4

region of the bacterial 16 S rRNA gene

79

, yielding accurate taxonomic

information with few biases among various bacterial taxa. Primers

—

ITS1

(5

′

-CTTGGTCATTTAGAGGAAGTAA-3

′

) and ITS2 (5

′

-TGCGTTCTTCATC-

GATGC-3

′

)

—

were used to amplify the ITS1 region of the fungal rRNA

gene. The PCR was carried out in a mixture (

ﬁ

nal volume, 50

µ

L) com-

prising 2

µ

L DNA template (1

–

10 ng), 5

µ

L of each 2

μ

M primer, 25

μ

L

Premix Ex Taq (Takara Biotechnology), and 13

μ

L sterilized water.

Thermal-cycling conditions were as follows: initial denaturation of 3 min
at 94 °C, six touchdown cycles of 45 s at 94 °C, 60 s from 65 to 58 °C,
70 s at 72 °C, followed by 22 cycles of 45 s at 94 °C, 60 s at 58 °C, 60 s at
72 °C, with a

ﬁ

nal elongation of 72 °C for 10 min. Tag-encoded high-

throughput sequencing of the 16 S and ITS genes from puri

ﬁ

ed and

quanti

ﬁ

ed PCR products was performed by the Magigene Company

(Beijing, China) using the Illumina MiSeq platform.

Raw sequences were quality processed using the Quantitative

Insights into Microbial Ecology (QIIME) pipeline (version 2.1)

79

.

Resampling operational taxonomic units (OTUs) was based on the
minimum sequence numbers to correct the sampling effort before
further analysis. The OTUs were clustered with a 97% similarity cutoff
using the UPARSE pipeline (version 7.1)

80

. Taxonomic assignment was

carried out using OTUs with SILVA (16 S) and Unite (ITS).

Alpha diversity was characterized by six indices (Shannon-Wea-

ver, Chao1, ACE, Richness, Simpson, and Pielou), calculated using the
OTU numbers for bacteria and fungi. (see Supplementary Information
for detailed calculations).

The Cornell Soil Health Assessment (CSHA)

81

–

83

combined with

principal component analysis (PCA)

84

85

was used to determine health

scores for soils sampled in 2016 and 2022, which allowed comparing
each plot

’

s changes in soil health after 6 years of

‘

rotation nourish-

ment

’

. The ten soil indicators (physical attributes: bulk density, soil

water content; chemical attributes: soil pH, TN, SOC, DOC, NO

3

–

-N, AP;

biological attributes: MBC and MBN) were normalized as individual

Article

https://doi.org/10.1038/s41467-023-44464-9

Nature Communications

| (2024) 15:198

10

CSHA scores

81

. The weights of individual CSHA scores were based on

PCA of all soil indicators (Table S5), representing the sum of the
eigenvectors derived from the

ﬁ

rst three principal components, which

were selected based on the in

ﬂ

ection point from a Scree plot and

Kaiser

’

s cut-off (eigenvalues > 1) (Fig. S8). These

ﬁ

rst three principal

components accounted for 51.09%, 19.44%, and 15.04% of the data

’

s

total variation, with a cumulative variance of 85.57%, capturing most of
the variation among the soil indicators. The soil health overall score (%)
was computed as a weighted average of all individual CSHA scores,
calculated as follows:

Soil health score =

ð

A

1

×

w

1

Þ

+

ð

A

2

×

w

2

Þ

+

ð

A

n

×

w

n

Þ

w

1

+

w

2

+

w

n

ð

10

Þ

where

A

is the CSHA score for each individual soil indicator, and

w

is

the weighting factor for each soil indicator (Table S5).

Entropy-TOPSIS for multiple function assessment of diversi

ﬁ

ed

crop rotations

The

CEI

was determined using crop yield, nutritional yield (detailed in

Eqs.

1

in SI Methods), economic bene

ﬁ

t, soil health, C sequestration,

microbial biodiversity, and net GHG emissions based on Entropy-TOPSIS
(Technique for Order Preference by Similarity to an Ideal Solution).

Entropy-TOPSIS is a multi-objective decision analysis tool for

identifying the most feasible solution from a set of options. It
de

ﬁ

nes the most positive and negative ideal solutions of a problem,

with the most feasible solution representing the solution that is the
most positive and furthest from the negative ideal solution

86