He et al. (2025). Soil organic carbon improvement for mitigating crop yield losses under global warming

2021; Li et al., 2023a; Müller et al., 2017

agermeyr et al.,

). However, the understanding

of how soil properties impact crop yield responses to climate in these

models remain incomplete. Beyond serving structural supports, soil

properties directly influence the retention and release of water and

nutrients essential for crop growth (

Andrade et al., 2024; Cordeiro et al.,

2022

). High-quality soils (with medium-textured and high soil organic

matter) were found to reduce the sensitivity of crop production to

climate variability and improve yield stability (

Qiao et al., 2022

). Soil

organic carbon (SOC), a ‘master

’

indicator of soil health, is expected to

play a vital role in mitigating and adapting to the adverse effects of

climate warming (

Deng et al., 2023; Droste et al., 2020; Feng et al.,

2022; Luo et al., 2023; Teng et al., 2024

). In addition, a recent study

reported that yields of corn, soybean, cotton, and barley are most sen-

sitive to warming in coarse-textured soils compared to medium- and

fine-textured soils in the United States (

Huang et al., 2021

). Clay soils

are often characterized by a high water-holding capacity, while

coarse-textured soils tend to have higher rates of nitrogen mineraliza-

tion (

M

akinen et al., 2017; Soinne et al., 2020

). These differences in soil

properties are likely to result in soil-dependent responses of crop yields

under climate change.

Maize is widely cultivated in low-latitude regions, which are pro-

jected to face the most severe climate change impacts due to their

proximity to critical temperature thresholds (

J

Tigchelaar et al., 2018; Zabel et al., 2021

agermeyr et al., 2021;

). Several studies have explored

the buffering effects of soils on maize yield responses to historical

climate perturbations (

Chen et al., 2024; Deng et al., 2020; Feng et al.,

2022

). While the elevated atmospheric carbon dioxide associated with

climate change has rarely been considered. As a C4 crop, maize has

limited capacity to benefit from elevated [CO

2

] (

Kimball, 2016

);

conversely, elevated [CO

2

] directly stimulates C3 photosynthesis,

making the so-called [CO

2

] fertilization effect more likely to offset a

portion of climate-induced yield losses in C3 crops like soybean (

Jin

et al., 2017

). Moreover, C4 photosynthesis has greater carbon gain than

C3 under conditions of low [CO

2

] and high temperatures, whereas more

carbon gain in C3 than C4 plants under high [CO

2

] and cool tempera-

tures; and water availability also impacts C3 and C4 differently, either

independently or in concert with changes in temperature and [CO

2

]

(

Zhou et al., 2018

). Given these varied responses of different crops to

climate factors, the contribution of soil properties to crop yield is likely

to differ across crop types.

Although the negative impacts of climate change on crop yields are

widely recognized, actual yield responses to temperature increase vary

significantly across regions (

Asseng et al., 2015

). Moreover, previous

meta-analyses have shown that yield gains through conservation agri-

culture practices are often observed in arid regions, where improved soil

structure is particularly beneficial (

Pittelkow et al., 2015; Su et al.,

2021

). Thus, we hypothesize that spatial variation in soil properties

contributes to these region-specific yield responses under global

warming, and soil quality enhancement can reduce yield temperature

sensitivity more effectively in dryland than in non-dryland areas. To test

this hypothesis, we developed a yield-climate relationship at a global

grid scale using a multiple regression model across two periods: the

historical period (1980

–

2010) and a future period (2050

–

2080) under

three shared socio-economic pathways scenarios (SSP126, SSP245, and

SSP585). We then quantified the temperature sensitivity of maize and

soybean yields, and applied a random forest model to assess the

contribution of soil properties on the temperature sensitivity. Finally, we

proposed SOC improvements to reduce yield temperature sensitivity

under future climate scenarios, and compared these benefits between

current and future varieties and between dryland and non-dryland re-

gions. We aimed to answer the following questions: (1) What are the

spatial patterns of global warming impacts on maize and soybean yields?

(2) How do soil properties modulate the temperature sensitivity of these

crop yields? Our goal is to provide insights into potential soil-based

strategies for future climate change adaptation and mitigation to safe-

guard food production.

2. Materials and methods

2.1. GGCM emulators

The GGCM emulators based on Global Gridded Crop Model Inter-

comparison (GGCMI) Phase 2 simulation were developed by

Franke

et al. (2020b)

. The GGCMI Phase 2 experiment compared crop model

simulations under uniform perturbations in carbon, temperature, water,

nitrogen levels and adaptation to shifting growing seasons (CTWN-A)

(

Franke et al., 2020a

), which produced a structured training dataset for

statistical emulation. By fitting individual regression models for each

crop, model, and pixel to the regressors of the CTWN-A analysis, the

emulators provide a tool that can effectively capture crop yield response

for different climate scenarios. Such emulators combine the advantages

of both process-based crop models and statistical models, and have been

applied successfully for reproducing crop yields under climate model

projections (

Li et al., 2023a, 2023b; Zabel et al., 2021

).

We employed emulators to simulate maize and soybean yields at a

grid scale (0.5

◦

resolution) using atmospheric CO

2

, changes in growing-

season temperatures (

Δ

T) and precipitation (

Δ

P), and applied nitrogen.

Separate emulations were conducted for rainfed and irrigated yields,

both with no crop adaptation (existing varieties) and with adaptation

(future varieties). To focus on yield changes under climate change, the

rainfed and irrigated yields were aggregated for analysis using crop area

data from

Monfreda et al. (2008)

. A total of nine GGCMs (CARAIB,

EPIC-TAMU, JULES, GEPIC, LPJ-GUESS, LPJmL, pDSSAT, PEPIC, and

PROMET) were used for maize, and eight GGCMs (excluding

LPJ-GUESS) were available for soybean. Given that some GGCMs tend to

differ in their baseline crop productivity levels, we harmonized simu-

lated crop yields to match observed yield patterns, following

Müller

et al. (2021)

and

Li et al. (2023b)

:

Y

i

,

j

=

Y

sim

,

i

,

j

×

Y

obs

,

t

,

j

Y

sim

,

t

,

j

(1)

where

Y

i

,

j

and

Y

sim

,

i

,

j

are the crop yields of the year

i

in grid cell

j

after and

before correction, respectively.

Y

obs

,

t

,

j

and

Y

sim

,

t

,

j

are the average values

of observed and simulated yields during the historical period

t

(1980

–

2010 in this study) in grid cell

j

.

2.2. Climate and soil data

Monthly temperature and precipitation data from simulations of 37

general circulation models (GCMs) across three Shared Socioeconomic

Pathways (SSP126, SSP245 and SSP585) were used to drive the GGCM

emulators (

Table S1

). The GCM data, sourced from the CMIP6 archives

(

https://aims2.llnl.gov/search/cmip6/

), were resampled to a 0.5

◦

Q. He et al.

European

Journal

of

Agronomy

170

(2025)

127739

2

resolution. The mean temperature and total precipitation during the

growing season of maize and soybean were calculated as weighted av-

erages of the monthly climate data. Thus, yield changes under different

SSP scenarios were attributed to changes in temperature, precipitation,

and [CO

2

] levels.

Gridded soil data were derived from the Harmonized World Soil

Database (HWSD version 1.2) (

Wieder et al., 2014

). The mapping raster

of the HWSD was aggregated from its native resolution of 30 arc-second

to a 0.5

◦

resolution for complying with the GGCMs. The HWSD combines

regional and national updates of soil characteristics worldwide, making

it available for global analyses. Soil properties were selected if they were

model parameters so that their impacts can be captured by modeling

methods; their characteristics did not change greatly under conventional

farming management practices; and they were commonly available in

gridded soil datasets (

Feng et al., 2022

). Finally, five key soil properties

at 0

–

30 cm depth, including soil organic carbon (kg C m

−

2

), bulk density

(g cm

−

3

), clay content (%), sand content (%) and pH, were extracted for

maize and soybean croplands.

2.3. Temperature sensitivity of crop yield

Temperature sensitivity of maize and soybean yields was estimated

using a panel data model, following the approach of previous studies

(

Deng et al., 2020; Feng et al., 2022; Zhu et al., 2019

). Growing-season

temperature changes (

T

,

℃

), growing-season precipitation changes (

P

,

%), and atmospheric CO

2

concentrations (

[

CO

2

]

, ppm) were used as

explanatory variables:

ln

(

Y

i

,

j

)

=

β

1

T

i

,

j

+

β

2

T

i

,

j

2

+

β

3

P

i

,

j

+

β

4

P

i

,

j

2

+

β

5

[

CO

2

]

i

,

j

+

β

6

[

CO

2

]

i

,

j

2

+

ε

(2)

where ln

(

Y

i

,

j

)

is the natural logarithm of crop yields of the year

i

in grid

cell

j

. The quadratic terms of

T

,

P

, and

[

CO

2

]

simulate the nonlinear

response of yield to temperature, precipitation, and CO

2

changes,

respectively.

β

1

to

β

6

are regression coefficients, and

ε

is the model error.

Then, the temperature sensitivity of crop yields, indicating the per-

centage of yield change per 1

℃

warming, was computed as the partial

derivative of yield with respect to temperature:

S

Y

,

T

=

∂

Y

∂

T

×

100%

=

(

β

1

+

2

β

2

T

t

,

j

)

×

100%

(3)

where

T

t

,

j

is the average temperature during period t (1980

–

2010 for

historical period and 2050

–

2080 for future period) in grid cell

j

.

β

1

and

β

2

are regression coefficients derived from

Eq. (2)

S

Y

,

T

denotes tem-

perature sensitivity, with positive values indicating yield increase and

negative values indicating yield decrease with warming. Yield sensi-

tivity to precipitation was excluded due to no significant precipitation

trends observed in most cropping areas (

b, e).

2.4. Relationship between temperature sensitivity and soil properties

We used a random forest (RF) model to evaluate the non-linear

relationship between temperature sensitivity and soil properties. RF is

an ensemble learning method that leverages multiple decision trees for

Fig. 1.

Linear trends in growing-season mean temperature (

T

r

), precipitation (

P

r

), and crop yield (

Y

r

) for maize (

a-c

) and soybean (

d-f

). Global maps depict the

median values derived from 37 general circulation models (GCMs) for the historical period 1980

–

2010. Boxplots represent future projections for the period

2050

–

2080 under three shared socioeconomic pathway scenarios (SSP126, SSP245, and SSP585). Stippling on the maps marks regions where more than half of the 37

GCMs exhibit statistically significant trends (

p

<

0.05).

Q. He et al.

European

Journal

of

Agronomy

170

(2025)

127739

3

prediction (

Breiman, 2001

). In addition to being faster and easier to

train, RF also has advantages to reduce overfitting problems and

improve model accuracy. We built the RF model using the ‘Ranger

’

package in R, with crop yield temperature sensitivity as the dependent

variable and five soil properties as independent variables. Model per-

formance was assessed using the coefficient of determination (

R

2

, esti-

mated using the internal out-of-bag samples) and the index of agreement

(

IA

, a standardized measure of the degree of model prediction error,

with value varies between 0 and 1, where a value of 1 indicates a perfect

match and a value of 0 indicates no agreement at all) (

Willmott, 1981

).

Key soil properties were determined by the variable importance which is

interpreted as the increase in the mean square error (

IncMSE

, %) due to

the random permutation of that variable in RF. The greater the increase

in error, the more important the variable is.

Separate RF models were built for maize (n

=

22,650) and soybean

(n

=

10,631). For potential SOC improvement in the future, we used the

medium-scenario data from

Zomer et al. (2017)

. The dataset provided a

spatially articulated estimate of the SOC distribution and increases in the

top 30 cm of soils over 20 years on all available cropland, achieved

through soil conservation practices like cover cropping, mulching,

conservation tillage, and organic manure application. We replaced the

original SOC input with the improved one in RF models to predict

changes in temperature sensitivity induced by SOC increases across

future climate scenarios. Since the aridity index (AI), defined as the ratio

of mean annual precipitation to potential evapotranspiration, is an

effective measure of long-term climate conditions, we calculated the AI

for the historical period of 1980

–

2010 at the grid scale to identify

dryland (AI

≤

0.65) and non-dryland (AI

>

0.65) regions in this study

(

Trabucco and Zomer, 2018

3. Results

3.1. Climate and yield trends

Significant trends in the temperature of crop growing season were

observed during the historical period (1980

–

2010), with average in-

creases of 0.28

℃

per decade for maize (66 % of grids showing signifi-

cant trends) and 0.29

℃

per decade for soybean (60 % of grids showing

significant trends). Large cropping areas of North America, Europe,

Russia and northeast China experienced above-average positive trends

in growing-season temperature (

a, d). For the future period

(2050

–

2080), projected average increase rates of temperature per

decade were 0.037

℃

, 0.23

℃

, and 0.61

℃

for maize, and 0.036

℃

,

0.24

℃

, and 0.61

℃

for soybean, under SSP126, SSP245, and SSP585,

respectively. However, no significant trends were detected in total

growing-season precipitation across most cropping areas during the

historical period, and precipitation projections for the future period

varied widely (

b, e).

For crop yields, maize showed a general increase in mid- to high-

latitude regions of the Northern Hemisphere, with Russia exhibiting a

respectively strong trend of 0.20 t ha

−

1

per decade, and 50 % of grids

showing statistically significant increases during the historical period

(

c). In contrast, most grids in the central United States and Central

Asia displayed negative yield trends. Under future climate scenarios, the

projected average maize yield trends were

−

0.01 t ha

−

1

,

−

0.07 t ha

−

1

,

and

−

0.19 t ha

−

1

per decade under SSP126, SSP245, and SSP585,

respectively

—

all lower than the historical trend of 0.05 t ha

−

1

per

decade. For soybean, more positive yield trends were observed during

the historical period, particularly in the mid-west United States, Europe,

and northeast China, with a global average increase of 0.08 t ha

−

1

per

decade (

f). However, future projections indicated declining yield

trends for soybean, with rates of

−

0.01 t ha

−

1

,

−

0.05 t ha

−

1

, and

−

0.13 t ha

−

1

per decade under SSP126, SSP245, and SSP585 scenarios,

respectively.

3.2. Temperature sensitivity of crop yields

Globally, maize yield responses to warming averaged

−

2.3 %

℃

−

1

during the historical period, with 26 % of grids showing positive values.

Under future climate scenarios, maize yield temperature sensitivity

declined further, averaging

−

6.6 %

℃

−

1

under SSP126 (9 % positive

grids),

−

7.4 %

℃

−

1

under SSP245 (6 % positive grids), and

−

7.5 %

℃

−

1

under SSP585 (4 % positive grids) (

a-d). Soybean yields responded

positively to warming during the historical period, with an average

temperature sensitivity of 3.0 %

℃

−

1

(39 % positive grids). However,

under future climate scenarios, soybean yield responses became

increasingly negative, with sensitivities averaging

−

8.9 %

℃

−

1

under

SSP126 (10 % positive grids),

−

10.2 %

℃

−

1

under SSP245 (7 % positive

grids), and

−

10.7 %

℃

−

1

under SSP585 (5 % positive grids) (

e-h).

While some northern high-latitude cropping regions exhibited enhanced

yield responses to warming, these areas also had high coefficients of

variation, reflecting significant uncertainty in model predictions

(

Fig. S1

). Moreover, uncertainty in the CO

2

effects contributed to much

overall model uncertainty, with greater impacts on soybean yield pre-

dictions compared to maize. Soybean yield predictions displayed the

greatest losses in the main-producer regions

–

the United States and

Brazil

–

paired with large gains across parts of China and other high-

latitude regions (

Fig. S2

).

3.3. Contribution of soil properties to yield temperature sensitivity

The Random Forest (RF) models explained 51 % and 59 % of the

variation in yield temperature sensitivity for maize and soybean,

respectively, with high indices of agreement (0.86 for maize and 0.88 for

soybean) during the historical period (

a, c). This suggests that the

spatially heterogeneous temperature sensitivity of both crops is largely

driven by soil heterogeneity, including factors such as SOC, bulk density,

pH, clay and sand content. Among these variables, SOC emerged as the

most important variable, with importance values ranging from 25 % to

28 % for maize and 26

–

28 % for soybean (

b, d). Increasing SOC

content notably reduced the negative impact of temperature on crop

yields when SOC levels were below approximately 13 kg C m

−

2

(

a, f). Current global SOC levels remain low (

Fig. S4

a), suggesting sig-

nificant potential for global soils to increase SOC before reaching the

optimal threshold, beyond which further increases in SOC generate no

additional mitigation in temperature sensitivity. Sand and clay content

also played critical roles, with relative importance values ranging from

17 % to 28 % (

b, d). The impact of sand content on temperature

sensitivity followed a hump-shaped pattern, with increasing sand con-

tent initially weakening the negative temperature sensitivity before

intensifying it (

c, h). In contrast, increasing clay content led to

greater yield losses per 1

◦

C increase in temperature before leveling off

(

Li et al., 2023b; Rose et al., 2016; Rosenzweig et al., 2013

b, g).

3.4. Impact of SOC improvement on yield temperature sensitivity

The temperature sensitivity of maize and soybean yields was pro-

jected to benefit from increased SOC through improved farming and soil

conservation practices, regardless of future climate scenarios (

Figs. 4

and 5

). At the continental scale, the reduction in maize yield losses due

to warming was most pronounced in Africa, where SOC improvement

mitigated yield losses by 1.3 %

℃

−

1

, followed by North America (1.2 %

℃

−

1

) and South America (1.0 %

℃

−

1

), all surpassing the global average

of 0.9 %

℃

−

1

under SSP126 with current crop varieties. However, these

benefits diminished under more extreme warming scenarios (i.e.,

SSP585 compared to SSP126) but could be amplified by adopting new

crop varieties (

Fig. 4

). For soybean, SOC improvement significantly

alleviated widespread warming-induced yield losses, particularly in the

major production regions like North America and South America. In

these regions, SOC improvement avoided 1.4

–

4.9 % and 2.1

–

4.2 % of

yield losses per 1

◦

C warming with current and future crop varieties,

Q. He et al.

European

Journal

of

Agronomy

170

(2025)

127739

4

respectively (

Fig. 5

).

Under future climate conditions, the reduction in maize yield losses

due to warming was more pronounced in dryland regions with SOC

improvement, while adopting new crop varieties was more effective in

mitigating yield losses in non-dryland regions (

Fig. 6

a). For example,

SOC improvement increased maize yield temperature sensitivity by

averages of 1.2

–

1.9 %

℃

−

1

across different scenarios in dryland regions,

but the changes were negligible in non-dryland regions. For soybean, the

positive impact of SOC improvement on yield temperature sensitivity

was comparable between dryland and non-dryland regions (1.7

–

2.8 %

℃

−

1

versus 1.1

–

2.4 %

℃

−

1

), as well as between current and new crop

varieties (1.1

–

2.7 %

℃

−

1

versus 1.4

–

2.8 %

℃

−

1

) (

Fig. 6

b).

4. Discussion

4.1. Crop yield response to rising temperatures

Consistent with previous studies, our findings indicate that future

warming climate will substantially reduce maize and soybean yields,

particularly under the SSP585 scenario, with impacts varying across

regions (

During the historical period (1980

–

2010), rising temperatures enhanced

maize and soybean yields at mid to high latitudes, such as Eastern

Europe, Russia, and northeastern China (

a, e). In these regions,

temperatures were below crop optimal levels, and warming extended

the growing season, leading to positive yield impacts, consistent with

findings by

Iizumi et al. (2017)

and

Lin et al. (2021)

. However, under

Fig. 2.

Spatial pattern of temperature sensitivity for maize yield (a-d) and soybean yield (e-h). Maps show median values of temperature sensitivity derived from 37

general circulation models (GCMs) for the historical period 1980

–

2010 (a, e), and for future projections during 2050

–

2080 under SSP126 (b, f), SSP245 (c, g), and

SSP585 (d, h). Pie charts inserted within each map indicate the percentage of grids with positive or negative temperature sensitivities.

Q. He et al.

European

Journal

of

Agronomy

170

(2025)

127739

5

future scenarios (2050

–

2080), warming predominantly had adverse ef-

fects on yields for both crops across most regions, with hotspots in areas

already experiencing high-temperature stress, such as India and Brazil

(

Lin et al., 2021; Lobell and Gourdji, 2012

b-d, f-h), as reported by

Teixeira et al. (2013)

. Generally, the

negative impacts of warming were more pronounced for soybean than

maize in the tropics, likely due to the lower optimal leaf temperature for

photosynthesis of soybean (

).

Additionally, higher temperatures after the mid-century will shorten

growing season length and increase water demand through greater

evaporation, which cannot be mitigated fully by CO

2

fertilization effects

(

Harrison et al., 2014; Muleke et al., 2022; Rosenzweig et al., 2013

Elliott et al., 2014; Lesk et al.,

The slightly lower negative temperature sensitivity of maize yields

compared to soybean in future scenarios can also be attributed partially

to the higher proportion of irrigated maize areas (17.8 %) relative to

soybean (9.5 %). Rising temperatures increase atmospheric water de-

mand, leading to drought stress through elevated vapor pressure deficit,

which can be alleviated by irrigation (

2021; Lobell et al., 2014

). Furthermore, adopting future crop varieties

will become increasingly important to preserve yields under future

climate forcing scenarios (

Figs. 4 and 5

). The GGCMI assumed that

future cultivars were modified to maintain their original growing season

length under each temperature scenario (

Franke et al., 2020a

). By

incorporating traits such as delayed maturity and improved heat toler-

ance, breeding programs can develop crop varieties better equipped to

withstand the adverse effects of rising temperatures (

Asseng et al., 2019;

Tao et al., 2022; Zabel et al., 2021

4.2. SOC contribution to mitigate warming-induced yield losses

Our findings highlight the important role of soil in mitigating crop

yield losses and enhancing system resilience to future warming climates

on a global scale. As a fundamental component of the soil-plant-

atmosphere continuum, soil can influence crop-climate interactions, as

reported by

Feng et al. (2022)

and

Chen et al. (2024)

. In our analysis,

soil properties explained a large share of the variation in crop yield

temperature sensitivity, with SOC identified as the most influential

factor (

). SOC contributes to multiple soil-based ecosystem services

(

Lal, 2016

). For example, SOC enhances plant-available water by

increasing soil porosity (

Huang et al., 2021

), improves nutrient cycling

and availability through SOC mineralization (

Moinet et al., 2023

),

supports nutrient retention via increased cation exchange capacity (

Jian

et al., 2020

), and provides habitat for beneficial soil microbiota. These

mechanisms collectively bolster soil health and favor crop growth, even

under the stress of a warming climate (

Teng et al., 2024

). Moreover, soils

rich in organic carbon exhibit greater resilience to adverse weather

conditions, buffering against the impacts of climate extremes on crop

production (

Droste et al., 2020

), thereby reducing yield sensitivity to

temperature.

Based on the observed relationship between temperature sensitivity

and SOC (

a, f), our analysis suggested that declines in crop yields

due to warming could be mitigated through SOC improvement. This

finding aligns with results from field experiments (

Ma et al., 2023

),

data-driven approaches (

Deng et al., 2023

), meta-analyses (

Oldfield

et al., 2019

) and crop modeling studies (

Feng et al., 2022

). Current SOC

content in most maize and soybean croplands remained below 8 kg C

m

−

2

(

Fig. S4

), highlighting substantial potential for stabilizing crop

yields under warming conditions through SOC enhancement. Our results

also indicated that improving SOC had a greater positive impact on crop

yields in dryland areas compared to non-dryland regions (

Fig. 6

). This

can also be attributed to the ability of SOC to enhance soil water

retention, which is particularly beneficial in water-scarce regions.

Conversely, in wetter regions with consistently saturated soils or

high-water tables, especially in poorly drained areas, increased soil

water retention from SOC improvement may lead to unfavorable con-

ditions for crop growth (

Huang et al., 2021

Beyond SOC, soil sand and clay content also played a significant role

in modulating crop yield sensitivity to temperature (

).

Zhu et al.

(2022b)

demonstrated that soils with higher sand content provided

Fig. 3.

Comparison of temperature sensitivity and random forest (RF) model predictions across global grids for maize (

a

) and soybean (

c

) during the historical period

(1980

–

2010), along with the relative importance of soil properties based on model outputs (

b

for maize and

d

for soybean). The dashed line represents the 1:1

ratio line.

Q. He et al.

European

Journal

of

Agronomy

170

(2025)

127739

6

more favorable subsurface water supply and supported greater vegeta-

tion biomass compared to clay-rich soils in arid and semi-arid regions.

Similarly, our findings suggested that soils with approximately 25 %

sand and low clay content, which often classified as medium-textured

soils, exhibited lower yield sensitivity to increased temperature

(

Hijbeek et al., 2017; Moinet et al., 2023

b-c, g-h). Moreover, the positive effects of SOC improvement on

crop yields may be more pronounced in sandy soils than clay soils

(

), indicating a complex inter-

play among soil properties in modulating crop responses to climate

warming.

4.3. Uncertainties and limitations

A primary source of uncertainty in projecting climate impacts on

global crop yields stems from differences in the structures, assumptions,

and methodologies of the GGCMs used in the ensemble modeling

(

Asseng et al., 2013; Li et al., 2023a; Rosenzweig et al., 2013; Xiong

et al., 2020

). Variations in the GCMs and SSPs also contribute signifi-

cantly to the overall uncertainty (

Fig. S1

) (

Moss et al., 2010

). While

simulations consistently show a strong negative impact of warming on

crop yields in tropical and subtropical regions, uncertainty is notably

higher for mid- to high-latitude regions, as reported by

Xiong et al.

(2020)

. Despite efforts to reduce these uncertainties, such as parameter

optimization (

Wallach et al., 2021

), ensemble member selection (

Li

et al., 2023a

) and model structure improvement, challenges remain in

enhancing model accuracy. Addressing these uncertainties will require

more detailed empirical data from field experiments in future studies

(

Wang et al., 2020

et al., 2023; Huang et al., 2020; Yang et al., 2024

Furthermore, while some high-latitude regions are projected to

become more climatically suitable for crop production under warming

Fig. 4.

Improved SOC-induced temperature sensitivity changes for maize. The central map shows temperature sensitivity with improved SOC during 2050

–

2080

under SSP126. Each bar chart shows the mean values of temperature sensitivity under different scenarios globally and across six continents: Africa (

a

), Europe (

b

),

North America (

c

), South America (

d

), Asia (

e

), and Oceania (

f

) and globally (

g

). Error bars indicate standard deviations, and gray bars show the changes induced by

improved SOC under each scenario. BAU means scenario with no SOC improvement.

Q. He et al.

European

Journal

of

Agronomy

170

(2025)

127739

7

scenarios, further research is needed to evaluate whether soil quality in

these areas will be adequate to support sustained agricultural produc-

tivity. Soils, particularly SOC, play a critical role in sustaining agricul-

tural systems, as they are fundamental to maintaining soil fertility and

supporting crop production in the context of a changing climate. How-

ever, this study may have underestimated the contribution of SOC

improvement to crop yield temperature sensitivity due to the unac-

counted interconnections among various soil properties. The dataset

used here (

Zomer et al., 2017

) reflects SOC improvements achieved

through enhanced farming and soil conservation practices, but the

concurrent changes in other soil properties were not included in pro-

jections due to data limitations.

4.4. Policy and practical implications

The findings of this study reveal practical soil-based pathways for

farmers to combat climate change, and present important implications

for sustainable agricultural production. First, the critical role of SOC in

mitigating the adverse effects of warming on crop yields highlights the

need for policies prioritizing soil health and conservation. Governments

and agricultural institutions should encourage the adoption of soil

management practices that enhance SOC, such as conservation tillage,

cover cropping, residue retention, and crop rotation (

Bai et al., 2019; He

). These practices not

only improve nutrient availability and water retention to benefit SOC

sequestration, but also strengthen crop resilience to adverse weather

conditions. Second, our study underscores the potential of crop varietal

development as an adaptive strategy to address climate challenges.

Fig. 5.

Improved SOC-induced temperature sensitivity changes for soybean. The central map shows temperature sensitivity with improved SOC during 2050

–

2080

under SSP126. Each bar chart shows the mean values of temperature sensitivity under different scenarios globally and across six continents: Africa (

a

), Europe (

b

),

North America (

c

), South America (

d

), Asia (

e

), and Oceania (

f

) and globally (

g

). Error bars indicate standard deviations, and gray bars show the changes induced by

improved SOC under each scenario. BAU means scenario with no SOC improvement.

Q. He et al.

European

Journal

of

Agronomy

170

(2025)

127739

8

Breeding programs focused on improved heat tolerance, drought resis-

tance, and traits that enhance soil quality, such as deeper root biomass

(

Cotrufo et al., 2024

), could help mitigate yield losses in vulnerable

regions. Finally, the observed spatial variation in crop yield responses to

warming emphasizes the need for region-specific adaptation strategies.

To ensure food security under climate change, integrated measures that

combine enhanced soil management, advanced crop varieties, and

tailored regional interventions will be essential (

Bilotto et al., 2023

).

Effective implementation of these strategies will require coordinated

efforts among governments, researchers, agricultural practitioners, and

stakeholders to build resilient and sustainable farming systems.

5. Conclusions

This study investigated the effects of soil properties on crop yield

responses to increased growing season temperatures across global

croplands, and quantified their contributions using statistical analysis

and random forest modeling. During the historical period (1980

–

2010),

each 1

◦

C increase in temperature resulted in yield changes of

−

2.3 % for

maize and

+

3.0 % for soybean. However, under future warming sce-

narios (2050

–

2080), maize and soybean yields were projected to decline

significantly, with losses ranging from

−

6.6 % to

−

7.5 % for maize and

−

8.9 % to

−

10.7 % for soybean per 1

◦

C of warming, across the three SSP

scenarios. The spatial variability in yield sensitivity to temperature was

influenced by soil properties, which explained 51 % of the variation in

maize and 59 % in soybean. Among these properties, SOC emerged as

the most influential factor, exhibiting a positive relationship with yield

temperature sensitivity before reaching optimal thresholds (around

11

–

15 kg C m

−

2

). Given that current SOC levels are far below this

optimal level, SOC improvements through farming and soil conservation

practices are projected to mitigate yield losses under future climate

scenarios. SOC improvement could help reduce widespread yield de-

clines induced by warming, particularly in dryland areas. Integrating

SOC improvements with the adoption of future crop varieties could

further alleviate these yield losses under warming climate. This study

highlights the potential of soil-based pathways to enhance the resilience

of global crop production to climate change impacts.

CRediT authorship contribution statement

Ke Liu:

Writing

–

review

&

editing, Resources.

Shengwei Zhang:

Writing

–

review

&

editing, Funding acquisition.

De Li Liu:

Writing

–

review

&

editing, Formal analysis.

Zikui Wang:

Writing

–

review

&

editing.

Matthew Tom Harrison:

Writing

–

review

&

editing.

Sien Li:

Writing

–

review

&

editing.

Qiang Yu:

Writing

–

review

&

editing.

Qinsi

He:

Writing

–

review

&

editing, Writing

–

original draft, Visualization,

Methodology, Conceptualization.

Kadambot H.M. Siddique:

Writing

–

review

&

editing.

Linchao Li:

Writing

–

review

&

editing, Methodology,

Data curation, Conceptualization.

Yu Shi:

Writing

–

review

&

editing,

Methodology.

Puyu Feng:

Writing

–

review

&

editing, Methodology.

Declaration of Competing Interest

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influence

the work reported in this paper.

Acknowledgements

We acknowledge financial support from the Fundamental Research

Funds for the Inner Mongolia Universities-Science Foundation; Natural

Science Foundation of Inner Mongolia Autonomous Region, China

Fig. 6.

Temperature sensitivity changes induced by improved SOC in dryland (aridity index

≤

0.65) and non-dryland (aridity index

>

0.65) regions for maize (

a

) and

soybe

a

n (

b

) during 2050

–

2080. Box plots show the 25th and 75th percentiles across 37 general circulation models (GCMs), with whiskers representing the 10th and

90th percentiles. The line and circle within each box indicate the multi-model median and mean values, respectively.

Q. He et al.

European

Journal

of

Agronomy

170

(2025)

127739

9

(2024ZD08); National Natural Science Foundation of China

(52279017); and the Program for Innovative Research Team in Uni-

versities of Inner Mongolia Autonomous Region (NMGIRT2313), inno-

vation team of ‘grassland talents

’

are jointly funded.

Appendix A. Supporting information

Supplementary data associated with this article can be found in the

online version at

doi:10.1016/j.eja.2025.127739

.

Data availability

Data will be made available on request.

References

Andrade, M.Gd.O., Cordeiro, C.Fd.S., Calonego, J.C., Franzluebbers, A., Rosolem, C.A.,

2024. Soil amendment and N fertilization strategies to improve C sequestration and

storage in soil aggregates. Soil Sci. Soc. Am. J. 88, 1767

1777

Asseng, S., et al., 2013. Uncertainty in simulating wheat yields under climate change.

Nat. Clim. Change 3, 827

832

Asseng, S., et al., 2015. Rising temperatures reduce global wheat production. Nat. Clim.

Change 5, 143

Asseng, S., et al., 2019. Climate change impact and adaptation for wheat protein. Glob.

147

.

Change Biol. 25, 155

Bai, X., et al., 2019. Responses of soil carbon sequestration to climate-smart agriculture

173

.

practices: a meta-analysis. Glob. Change Biol. 25, 2591

2606

Bilotto, F., Christie-Whitehead, K.M., Malcolm, B., Harrison, M.T., 2023. Carbon, cash,

cattle and the climate crisis. Sustain. Sci. 18, 1795

https://doi.org/10.1023/A:

1811

.

Breiman, L., 2001. Random forests. Mach. Learn. 45, 5

–

32.

1010933404324

.

Chen, F., et al., 2024. Soil buffering capacity enhances maize yield resilience amidst

climate perturbations. Agric. Syst. 215, 103870

.

Cordeiro, C.Fd.S., Rodrigues, D.R., Silva, G.Fd, Echer, F.R., Calonego, J.C., 2022. Soil

organic carbon stock is improved by cover crops in a tropical sandy soil. Agron. J.

114, 1546

Cotrufo, M.F., Haddix, M.L., Mullen, J.L., Zhang, Y., McKay, J.K., 2024. Deepening root

1556

.

inputs: potential soil carbon accrual from breeding for deeper rooted maize. Glob.

Change Biol. 30, e17591

Deng, X., et al., 2023. Building soil to reduce climate change impacts on global crop

yield. Sci. Total Environ. 903, 166711

.

Deng, X., Huang, Y., Qin, Z., 2020. Soil indigenous nutrients increase the resilience of

maize yield to climatic warming in China. Environ. Res. Lett. 15, 094047

.

Droste, N., et al., 2020. Soil carbon insures arable crop production against increasing

adverse weather due to climate change. Environ. Res. Lett. 15: 124034

.

Elliott, J., et al., 2014. Constraints and potentials of future irrigation water availability

on agricultural production under climate change. Proc. Natl. Acad. Sci. 111,

Elliott, J., et al., 2015. The global gridded crop model intercomparison: data and

modeling protocols for Phase 1 (v1.0). Geosci. Model Dev. 8, 261

277

Feng, P., et al., 2022. Soil properties resulting in superior maize yields upon climate

warming. Agron. Sustain. Dev. 42, 85

.

Franke, J.A., et al., 2020a. The GGCMI Phase 2 experiment: global gridded crop model

simulations under uniform changes in CO2, temperature, water, and nitrogen levels

(protocol version 1.0). Geosci. Model Dev. 13, 2315

Franke, J.A., et al., 2020b. The GGCMI Phase 2 emulators: global gridded crop model

2336

.

responses to changes in CO2, temperature, water, and nitrogen (version 1.0). Geosci.

Model Dev. 13, 3995

Fu, J., et al., 2023. Extreme rainfall reduces one-twelfth of China

4018

.

’

s rice yield over the last

two decades. Nat. Food 4, 416

Godfray, H.C.J., et al., 2010. Food security: the challenge of feeding 9 billion people.

423

.

Science 327, 812

Harrison, M.T., 2021. Climate change benefits negated by extreme heat. Nat. Food 2,

818

.

Harrison, M.T., Liu, K., 2024. Holistic systems analyses accelerate progress towards

Sustainable Development Goals. Nat. Food 5, 544

Harrison, M.T., Tardieu, F., Dong, Z., Messina, C.D., Hammer, G.L., 2014. Characterizing

545

.

drought stress and trait influence on maize yield under current and future

conditions. Glob. Change Biol. 20, 867

He, Q., et al., 2023. Modelling interactions between cowpea cover crops and residue

878

.

retention in Australian dryland cropping systems under climate change. Agric.

Ecosyst. Environ. 353, 108536

Hijbeek, R., et al., 2017. Do organic inputs matter

a meta-analysis of additional yield

effects for arable crops in Europe. Plant Soil 411, 293

303

Huang, Y., et al., 2020. Assessing synergistic effects of no-tillage and cover crops on soil

carbon dynamics in a long-term maize cropping system under climate change. Agric.

For. Meteorol. 291, 108090

.

Huang, J., Hartemink, A.E., Kucharik, C.J., 2021. Soil-dependent responses of US crop

yields to climate variability and depth to groundwater. Agric. Syst. 190, 103085

.

Iizumi, T., et al., 2017. Responses of crop yield growth to global temperature and

socioeconomic changes. Sci. Rep. 7, 7800

.

J

agermeyr, J., et al., 2021. Climate impacts on global agriculture emerge earlier in new

generation of climate and crop models. Nat. Food 2, 873

Jian, J., Du, X., Reiter, M.S., Stewart, R.D., 2020. A meta-analysis of global cropland soil

885

.

carbon changes due to cover cropping. Soil Biol. Biochem. 143, 107735

Jin, Z., et al., 2017. The combined and separate impacts of climate extremes on the

current and future US rainfed maize and soybean production under elevated CO2.

Glob. Change Biol. 23, 2687

Kimball, B.A., 2016. Crop responses to elevated CO2 and interactions with H2O, N, and

2704

.

temperature. Curr. Opin. Plant Biol. 31, 36

43

Lal, R., 2016. Soil health and carbon management. Food Energy Secur. 5, 212

Lesk, C., et al., 2021. Stronger temperature-moisture couplings exacerbate the impact of

222

.

climate warming on global crop yields. Nat. Food 2, 683

Li, L., et al., 2023a. The optimization of model ensemble composition and size can

691

.

enhance the robustness of crop yield projections. Commun. Earth Environ. 4, 362

Li, L., et al., 2023b. Integrating machine learning and environmental variables to

constrain uncertainty in crop yield change projections under climate change. Eur. J.

Agron. 149, 126917

Lin, T.S., Song, Y., Lawrence, P., Kheshgi, H.S., Jain, A.K., 2021. Worldwide maize and

soybean yield response to environmental and management factors over the 20th and

21st centuries. J. Geophys. Res. Biogeosciences 126, e2021JG006304

.

Liu, K., et al., 2023. Silver lining to a climate crisis in multiple prospects for alleviating

crop waterlogging under future climates. Nat. Commun. 14, 765

.

Lobell, D.B., et al., 2014. Greater sensitivity to drought accompanies maize yield increase

in the US Midwest. Science 344, 516

Lobell, D.B., Gourdji, S.M., 2012. The influence of climate change on global crop

519

.

productivity. Plant Physiol. 160, 1686

Luo, N., et al., 2023. China can be self-sufficient in maize production by 2030 with

1697

.

optimal crop management. Nat. Commun. 14, 2637

Ma, Y., et al., 2023. Global crop production increase by soil organic carbon. Nat. Geosci.

akinen, H., Kaseva, J., Virkaj

arvi, P., Kahiluoto, H., 2017. Shifts in soil-climate

combination deserve attention. Agric. For. Meteorol. 234-235, 236

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

246

.

for carbon. Glob. Change Biol. 29, 2384

2398

Monfreda, C., Ramankutty, N., Foley, J.A., 2008. Farming the planet: 2. Geographic

distribution of crop areas, yields, physiological types, and net primary production in

the year 2000. Glob. Biogeochem. Cycles 22. GB1022

.

Moss, R.H., et al., 2010. The next generation of scenarios for climate change research and

assessment. Nature 463, 747

756

Muleke, A., et al., 2022. Earlier crop flowering caused by global warming alleviated by

irrigation. Environ. Res. Lett. 17, 044032

.

Müller, C., et al., 2017. Global gridded crop model evaluation: benchmarking, skills,

deficiencies and implications. Geosci. Model Dev. 10, 1403

1422

Müller, C., et al., 2021. Exploring uncertainties in global crop yield projections in a large

ensemble of crop models and CMIP5 and CMIP6 climate scenarios. Environ. Res.

Lett. 16, 034040

.

Oldfield, E.E., Bradford, M.A., Wood, S.A., 2019. Global meta-analysis of the relationship

between soil organic matter and crop yields. SOIL 5, 15

32

Pittelkow, C.M., et al., 2015. Productivity limits and potentials of the principles of

conservation agriculture. Nature 517, 365

Qiao, L., et al., 2022. Soil quality both increases crop production and improves resilience

368

.

to climate change. Nat. Clim. Change 12, 574

580

Ray, D.K., Gerber, J.S., MacDonald, G.K., West, P.C., 2015. Climate variation explains a

third of global crop yield variability. Nat. Commun. 6, 5989

.

Rose, G., Osborne, T., Greatrex, H., Wheeler, T., 2016. Impact of progressive global

warming on the global-scale yield of maize and soybean. Clim. Change 134,

417

Rosenzweig, C., et al., 2013. Assessing agricultural risks of climate change in the 21st

428

.

century in a global gridded crop model intercomparison. Proc. Natl. Acad. Sci. 111,

3268

Seppelt, R., Klotz, S., Peiter, E., Volk, M., 2022. Agriculture and food security under a

3273

.

changing climate: an underestimated challenge. iScience 25, 105551

.

Shi, Y., et al., 2022. Building social resilience in North Korea can mitigate the impacts of

climate change on food security. Nat. Food 3, 499

511

Soinne, H., et al., 2020. Soil organic carbon and clay content as deciding factors for net

nitrogen mineralization and cereal yields in boreal mineral soils. Eur. J. Soil Sci. 72,

1497

Su, Y., Gabrielle, B., Makowski, D., 2021. The impact of climate change on the

1512

.

productivity of conservation agriculture. Nat. Clim. Change 11, 628

Tao, F., Zhang, L., Zhang, Z., Chen, Y., 2022. Designing wheat cultivar adaptation to

633

.

future climate change across China by coupling biophysical modelling and machine

learning. Eur. J. Agron. 136, 126500

Teixeira, E.I., Fischer, G., van Velthuizen, H., Walter, C., Ewert, F., 2013. Global hot-

spots of heat stress on agricultural crops due to climate change. Agric. For. Meteorol.

170, 206

Teng, J., et al., 2024. Conservation agriculture improves soil health and sustains crop

215

.

yields after long-term warming. Nat. Commun. 15, 8785

Tigchelaar, M., Battisti, D.S., Naylor, R.L., Ray, D.K., 2018. Future warming increases

probability of globally synchronized maize production shocks. PNAS 115,

6644

Trabucco, A., Zomer, R.J., 2018. Global aridity index and potential evapotranspiration

6649

.

(ET0) climate database v2. CGIAR Consort. Spat. Inf. 10 (m9)

Wallach, D., et al., 2021. How well do crop modeling groups predict wheat phenology,

given calibration data from the target population? Eur. J. Agron. 124, 126195

Wang, X., et al., 2020. Emergent constraint on crop yield response to warmer

temperature from field experiments. Nat. Sustain. 3, 908

916

Wieder, W.R., Boehnert, J., Bonan, G.B., Langseth, M., 2014. Regridded harmonized

Q. He et al.

European

Journal

of

Agronomy

170

(2025)

127739

10

world soil database v1. 2. ORNL DAAC

Willmott, C.J., 1981. On the validation of models. Phys. Geogr. 2, 184

Xiong, W., et al., 2020. Different uncertainty distribution between high and low latitudes

194

.

in modelling warming impacts on wheat. Nat. Food 1, 63

Yang, X., et al., 2024. Diversifying crop rotation increases food production, reduces net

69

.

greenhouse gas emissions and improves soil health. Nat. Commun. 15, 198

Zabel, F., et al., 2021. Large potential for crop production adaptation depends on

available future varieties. Glob. Change Biol. 27, 3870

Zhou, H., Helliker, B.R., Huber, M., Dicks, A., Akçay, E., 2018. C4 photosynthesis and

3882

.

climate through the lens of optimality. PNAS 115, 12057

Zhu, P., et al., 2022a. Warming reduces global agricultural production by decreasing

12062

.

cropping frequency and yields. Nat. Clim. Change 12, 1016

Zhu, X., et al., 2022b. Soil coarsening alleviates precipitation constraint on vegetation

1023

.

growth in global drylands. Environ. Res. Lett. 17, 114008

.

Zhu, P., Zhuang, Q., Archontoulis, S.V., Bernacchi, C., Müller, C., 2019. Dissecting the

nonlinear response of maize yield to high temperature stress with model-data

integration. Glob. Change Biol. 25, 2470

2484

Zomer, R.J., Bossio, D.A., Sommer, R., Verchot, L.V., 2017. Global sequestration

potential of increased organic carbon in cropland soils. Sci. Rep. 7, 15554