论文信息
论文解读
AI 辅助整理- 方法
- 构建AHADR综合指数,分析全球时空格局、集聚特征与状态转移,并考察干旱及投入要
- 发现
- 全球AHADR总体改善但区域不平等加剧,低韧性国家易陷稳定低水平,极端干旱有负面影响。
- 机会点
- 可结合遥感干旱与气候情景,研究低韧性国家韧性跃迁的驱动机制与干预路径。
论文摘要
实现“零饥饿”可持续发展目标正变得愈发困难,原因在于气候变化和地缘政治冲突等日益加剧的不确定性。本研究构建了一个全球性指数——反饥饿农业发展韧性(Anti-Hunger Agricultural Development Resilience, AHADR),用以衡量农业系统抵御饥饿的韧性。2001—2019年间,AHADR虽有波动但总体呈改善趋势,而AHADR的区域不平等程度却在加剧。多数国家表现为低AHADR集聚(11.32%)或高AHADR集聚(59.41%)。被归类为中等、高或极高AHADR的国家往往在韧性状态之间经历更为动态的转换,而低AHADR国家通常停留在稳定的较低水平韧性状态。极端干旱对AHADR具有不利影响,而畜牧业和化肥使用与AHADR的关系可能呈非线性。
全文
Frontiers in Sustainable Food SystemsFrontiers in Sustainable Food Systems 01 frontiersin.org
Estimating agricultural
development resilience against
hunger in the 21 century
Zhuang Zhang 1, You-hua Chen 2* and Weiwei Cui 3*
1School of Management, Guangdong Ocean University, Zhanjiang, China, 2College of Economics and
Management, South China Agricultural University, Guangzhou, China, 3School of Business
Administration, Zhongnan University of Economics and Law, Wuhan, China
Achieving the Sustainable Development Goal of Zero Hunger is becoming more
difficult because of growing uncertainties such as climate change and geopo -
litical conflicts. This study develops a global index, Anti-Hunger Agricultural
Development Resilience (AHADR), to measure the resilience of agricultural sys -
tems against hunger. AHADR fluctuated throughout 2001–2019 but improved
overall, whereas regional inequality in AHADR increased. Most countries exhibit
either low-AHADR agglomeration (11.32%) or high-AHADR agglomeration
(59.41%). Countries categorized as having medium, high, or very high AHADR
tend to experience more dynamic transitions between resilience states, whereas
countries with low AHADR usually remain in stable, lower-level resilience states.
Extreme droughts have a detrimental effect on AHADR, while the relationships
of livestock and fertilizer use with AHADR may be nonlinear.
KEYWORDS
AHADR agglomeration, anti-hunger, anti-hunger agricultural development resilience,
generalized linear model, resilience-state transitions
1 Introduction
Global grain output has risen sharply over the past six decades. According to the FAO, world
cereal production increased from 87.69 million tons in 1961 to 305.96 million tons in 2022, and
average grain yields climbed from 1,353.2 kg per hectare to 4,182.4 kg per hectare. Y et this pro-
ductivity surge has not eliminated hunger. The 2021 edition of The State of Food Security and
Nutrition in the World reports that hundreds of millions of people worldwide remain chronically
food insecure, and since 2019 the number of undernourished people has grown by roughly
150 million. Regional conflicts, severe climate events, and public-health shocks repeatedly disrupt
production and supply and push vulnerable populations back into hunger. Raising average yields
alone is therefore no longer sufficient; agri-food systems must also be able to withstand, absorb,
and recover from shocks. This study is motivated by the need to measure such resilience directly.
The concept of resilience originated in materials physics and has since been adopted across
ecology, psychology, and supply-chain management (Holling, 1973; Luthans et al., 2006 ;
Christopher and Peck, 2004; Tendall et al., 2015) to describe an entity’s capacity to withstand,
adapt to, and reorganize in the face of disturbance without losing its essential functions. In
agriculture, resilience denotes the ability of food systems to continue providing sufficient and
accessible food for all despite unforeseen disruptions, from droughts and floods to market and
geopolitical shocks (Friel et al., 2020; Leach et al., 2020; Mu et al., 2021). As climate variability
and global conflict become more frequent and severe, agricultural resilience has become as
important as output growth itself for safeguarding food security. ( Kahiluoto et al., 2019 ;
Hasegawa et al., 2021).
TYPE Original Research
PUBLISHED 17 November 2026
DOI 10.3389/fsufs.2026.1930549
OPEN ACCESS
EDITED BY
Hanuman S. Jat,
Central Soil Salinity Research Institute
(ICAR), India
REVIEWED BY
Anis Omri,
Qassim University, Saudi Arabia
Ganeshkumar D. Rede,
Symbiosis International University, India
*CORRESPONDENCE
You-hua Chen
yhchen214@scau.edu.cn
Weiwei Cui
1099706581@qq.com
RECEIVED 07 July 2026
REVISED 13 September 2026
ACCEPTED 23 October 2026
PUBLISHED 17 November 2026
CITATION
Zhang Z, Chen Y-h and Cui W (2026)
Estimating agricultural development
resilience against hunger in the 21
century.
Front. Sustain. Food Syst. 10:1930549.
doi: 10.3389/fsufs.2026.1930549
COPYRIGHT
© 2026 Zhang, Chen and Cui. This is an
open-access article distributed under
the terms of the Creative Commons
Attribution License (CC BY). The use,
distribution or reproduction in other
forums is permitted, provided the
original author(s) and the copyright
owner(s) are credited and that the
original publication in this journal is
cited, in accordance with accepted
academic practice. No use, distribution
or reproduction is permitted which does
not comply with these terms.
Zhang et al. 10.3389/fsufs.2026.1930549
Frontiers in Sustainable Food Systems 02 frontiersin.org
Despite broad agreement on its importance, measuring agricul -
tural resilience remains difficult, and three approaches dominate
the literature. First, single indicators are commonly used: Zampieri
et al. (2017 , 2020) characterize resilience through the mean and
variance of grain production, while crop diversity, such as the
Shannon index, is sometimes used as a proxy ( Renard and Tilman,
2019). These measures are simple but fragile: an unusually low yet
stable yield can be misread as high resilience, and the effect of crop
diversity on overall production efficiency is itself ambiguous ( Chen
et al., 2021 ). Second, composite index systems have been devel -
oped, including the absorption-adaptation-transformation frame -
work of Smith and Frankenberger (2018) , the distinction between
specific and general resilience of Meuwissen et al. (2019) , and the
poverty-threshold probability index of Cissé and Barrett (2018) .
These are conceptually richer, but they combine many country- or
household-specific indicators into scores that are not readily com -
parable across nations. 1 Third, resilience is sometimes inferred
indirectly: Zampieri et al. (2020) simulate food losses under climate
scenarios, and Birthal and Hazrana (2019) show that crop diversi -
fication buffers heat and water shocks. Such studies establish that
resilience matters, but they do not provide a universal, quantifiable
metric. As a result, evidence on agricultural resilience, especially at
the national level and over long periods, remains scarce.
Agriculture performs at least three functions: production,
ecology, and development ( Foley et al., 2011 ; FAO, 2021; Tilman
et al., 2011 ); however, eradicating hunger and ensuring food secu -
rity constitute its most fundamental function. This paper therefore
develops the Anti-Hunger Agricultural Development Resilience
(AHADR) index, grounded in feasible development theory, to
measure a country’s probability of sustaining a level of grain pro -
duction sufficient to prevent hunger despite shocks. Built on the
mean and variance of grain output and anchored to the FAO’s
minimum per capita consumption threshold, AHADR is less dis -
torted by single-year extremes than yield- or diversity-based prox -
ies and is sufficiently standardized for cross-country comparison.
We compute the index for 161 countries from 2001 to 2019.
On this basis, the study makes three contributions. First, we
estimate how key production and shock variables, including labor,
capital, fertilizers, machinery, livestock, arable land, irrigation,
extreme floods, and severe droughts, affect AHADR ( Table 1 ).
Second, we characterize its spatiotemporal distribution and spatial
agglomeration (Figures 1, 2). Third, we classify countries into resil -
ience states and trace transitions between these states over time
(Figure 3).
2 Method
2.1 Data
This paper compiles country-level panel data for 2001–2019 in
three main areas: agricultural production and conditions, agricul -
tural trade, and global extreme climate events. Data from the Food
and Agriculture Organization (FAO) include total cereal output,
imports, exports, farmland area, irrigation, livestock, fertilizer, and
1 As a supplementary analysis, Appendix Figure A5 compares GFSI with AHADR.
machinery; some variables were cross-referenced against addi -
tional databases. The FAO Statistical Database (FAOSTAT) con -
tains approximately 3 million time-series and cross-sectional
datasets for more than 210 countries and territories, covering nine
categories, including trade, supply, food security, prices, raw-mate -
rial inputs, forestry, and fisheries. 2 Data on extreme drought and
flood events were obtained from EM-DAT, which contains infor -
mation on more than 26,000 large-scale disasters worldwide since
2000. EM-DAT compiles information from sources including
United Nations agencies, nongovernmental organizations, reinsur -
ance companies, research institutions, and news agencies.
Descriptive statistics for the key variables are presented in
Appendix Tables A2, A3, and the full list of countries appears in
Appendix Tables A4.
2.2 Measurement of AHADR
The mean and variance are fundamental parameters of cumula-
tive probability distributions and are therefore essential for model -
ing sustainable grain production. We estimate the mean and
variance of grain output using established equations. Following
Cissé and Barrett (2018) , we incorporate lagged grain output into
the baseline equation to account for the effect of farmers’ expecta -
tions on current production decisions. Statistically, using a one-
period lag captures the most recent production state, minimizes the
loss of observations caused by lagging, and helps address serial cor -
relation in panel data.
The mean and variance are essential for defining cumulative
probability distribution functions. We aim to predict the mean and
variance of grain output using established mean and variance
equations. Following Cissé and Barrett (2018) , we incorporate
lagged grain output into the basic equation to account for the
effects of farmers’ expectations. The mean equation is expressed as
follows Equation 1:
( ),1g , , it M i t it M M it MitLG LG X X uβδ−= ++ (1)
The scatter plot shows that there is a linear relationship between
the lagged and current variables (see Appendix Figure A1). Therefore,
a linear specification is more appropriate. In theory, a polynomial
lag specification may improve model fit, but the appropriate lag
order must be tested, as shown in Appendix Tables A4. In general,
the Akaike information criterion (AIC), R-squared, and root mean
squared error (RMSE) are used to evaluate the model specification.
AIC, proposed by Akaike (1974), is a statistical criterion for evaluat-
ing the complexity and accuracy of statistical models. Generally, a
larger AIC value indicates a better fit for the lagged-period model.
Appendix Tables A4 shows that model (2) has the highest AIC value.
However, compared with model (1), model (2) shows no significant
changes in R-squared or RMSE. Therefore, model (1) is selected as
the optimal functional form, and the mean equation can be
expressed as follows.
,1it M i t M it MitLG LG X uβδ −= ++ (2)
2 https://www.fao.org/faostat/en/
Zhang et al. 10.3389/fsufs.2026.1930549
Frontiers in Sustainable Food Systems 03 frontiersin.org
TABLE 1 Estimated equations for AHADR.
(1) (2) (3) (4) (5) (6) (7) (8)
GLM Driscoll-Kraay GLM Driscoll-Kraay GLM Driscoll-Kraay GLM Driscoll-Kraay
Variables Mean Variance AHADR
TLU_lag
−2.671***
(0.642)
−2.671 (1.718) 6.127 (5.221) 6.127 (6.145)
−1.655***
(0.227)
−1.655***
(0.372)
−1.571***
(0.220)
−1.571***
(0.354)
Labor
−0.221**
(0.110)
−0.221* (0.124)
−0.383
(0.891)
−0.383 (1.199) 0.021 (0.039) 0.021 (0.018)
Labor_sq 0.009 (0.009) 0.009 (0.009) 0.032 (0.071) 0.032 (0.088)
−0.001
(0.003)
−0.001 (0.002)
Capital 0.068 (0.101) 0.068 (0.112)
−0.051
(0.816)
−0.051 (0.345)
−0.040
(0.036)
−0.040**
(0.018)
Capital_sq 0.001 (0.006) 0.001 (0.007)
−0.031
(0.052)
−0.031 (0.031) 0.002 (0.002) 0.002 (0.002)
Fertilizer
−0.288***
(0.107)
−0.288* (0.147)
−0.451
(0.868)
−0.451 (1.337)
−0.012
(0.038)
−0.012 (0.030)
Fertilizer_sq
0.016***
(0.005)
0.016** (0.006) 0.021 (0.041) 0.021 (0.063) 0.001 (0.002) 0.001 (0.001)
Machinery
−0.403***
(0.131)
−0.403***
(0.137)
1.574 (1.062) 1.574 (1.102)
−0.090*
(0.046)
−0.090**
(0.037)
Machinery_sq
0.014***
(0.005)
0.014*** (0.005)
−0.052
(0.040)
−0.052 (0.040) 0.003* (0.002) 0.003** (0.001)
Livestock 0.113 (0.196) 0.113 (0.194)
−1.663
(1.591)
−1.663 (1.562)
0.146**
(0.069)
0.146 (0.093)
Livestock_sq
−0.006
(0.013)
−0.006 (0.012) 0.110 (0.102) 0.110 (0.089)
−0.013***
(0.004)
−0.013**
(0.005)
Cropland
0.538***
(0.178)
0.538* (0.259)
−1.951
(1.448)
−1.951* (1.047)
−0.118*
(0.063)
−0.118 (0.097)
Cropland_sq 0.008 (0.013) 0.008 (0.018) 0.120 (0.103) 0.120 (0.082)
0.021***
(0.004)
0.021*** (0.006)
Irrigation
−0.135*
(0.070)
−0.135* (0.068) 0.515 (0.571) 0.515 (0.394) 0.008 (0.025) 0.008 (0.007)
Irrigation_sq 0.010 (0.008) 0.010 (0.007)
−0.034
(0.067)
−0.034 (0.042)
−0.001
(0.003)
−0.001 (0.001)
Flood_event 0.005 (0.005) 0.005** (0.002)
0.092**
(0.041)
0.092** (0.034)
−0.001
(0.002)
−0.001 (0.001)
−0.000
(0.002)
−0.000 (0.001)
Drought_event
−0.111***
(0.019)
−0.111***
(0.021)
0.673***
(0.154)
0.673*** (0.175)
−0.024***
(0.007)
−0.024**
(0.011)
−0.026***
(0.007)
−0.026**
(0.011)
Constant
38.566***
(5.072)
0.000 (0.000)
−28.395
(41.532)
0.000 (0.000)
12.435***
(1.809)
0.000 (0.000)
11.827***
(1.692)
0.000 (0.000)
Y ear effects Ye s Ye s Ye s Ye s Ye s Ye s Ye s Ye s
Country/
territory effects
Ye s Ye s Ye s Ye s Ye s Ye s Ye s Ye s
Observations 2,898 2,898 2,898 2,898 2,898 2,898 2,898 2,898
Robust standard errors are reported in parentheses. ***p < 0.01; **p < 0.05; *p < 0.10. Table 5A reports bias estimates for the mean and variance equations based on 200 bootstrap replications;
most estimation deviations lie between −0.1 and 0.1.
Zhang et al. 10.3389/fsufs.2026.1930549
Frontiers in Sustainable Food Systems 04 frontiersin.org
FIGURE 1
AHADR in different years, income groups, areas, and types. (A) Presents the mean AHADR and the total number of extreme drought events;
(B) illustrates AHADR across countries with different income levels over time; (C) shows AHADR across regions over time; (D) presents the distribution
of AHADR states over time; and (E) and (F) display the Gini coefficient and its decomposition over time.
FIGURE 2
Spatial agglomeration. LL denotes low-AHADR agglomeration, HH denotes high-AHADR agglomeration, and HL and LH indicate no clear AHADR
agglomeration. Part (A) maps local AHADR agglomeration across countries in 2019; Part (B) shows the annual global Moran’s I and Z-values; and Part
(C) shows the distribution of agglomeration types across regions.
Zhang et al. 10.3389/fsufs.2026.1930549
Frontiers in Sustainable Food Systems 05 frontiersin.org
In Equation 2, −,1itLG is lagged total grain output, itX is a vector of
control variables that includes basic production conditions and exter-
nal climate shocks, äM represents the corresponding coefficients, Mitu
is the estimated residual. Using Equation 2, we estimate a generalized
linear model (GLM) with a Gaussian distribution to obtain the sample
mean ( ) ,1it M i t M itE LG LG Xβδ −= + . In this case, we can also derive
the variance equation ( ) ( ) ( )= −= 2 2it it it MitVar LG E LG E LG E u .Because ( )2MitEu is not normally distributed, we replace it with
ln( ( )2MitEu ) (see Appendix Figure A2). The variance equation is then
expressed as follows.
( )( )
2 ,1Ln Mit V i t V it VitE u LG X β δµ −= ++
(3)
Using Equation 3 and a GLM (Generalized Linear Model)
with a Gaussian distribution, we obtain the expected value
( )( )( ) 2 ,1Ln Mit V i t V itE E u LG X βδ −= + .
Furthermore, we can calculate
( )( ) ( )βδ −= +,1exp .it V i t V itE Var LG LG X Because itLG follows the
normal distribution ( ) ( )( ),it itN E LG Var LG , we can calculate the
probability that it exceeds the established threshold itz . This prob-
ability represents agricultural development resilience
against hunger.
( ) ( )= >= − 1it it it G itp P LG z F z (4)
The Food and Agriculture Organization of the United Nations
(FAO) defines the food-security threshold as 400 kilograms per
person per year (Eurasia Media Network, 2024).3 Considering food-
trade relationships between countries, the established threshold is
defined as follows Equation 5.
( )( )= ×− −400 itz Ln population Food imports Food exports (5)
Furthermore, we can predict agricultural development resilience
against hunger based on an assessment of existing production capac-
ity. The estimation equation is as follows.
,1it R i t R it Ritp LG Xβ δµ −= ++ (6)
Using Equation 6, we can estimate the expected AHADR value,
E(p). In addition, the AHADR estimates change little when we use a
Gamma distribution (see Appendix Figure A3).
2.3 Spatial relationships in AHADR
This study uses global and local Moran’s I statistics to exam -
ine whether AHADR values are spatially correlated across neigh -
boring countries. Global Moran’s I measures overall spatial
autocorrelation across the study area, whereas local Moran’s I
3 The results are robust when thresholds of 350, 400, and 450 kilograms are
used for comparison (Figure 4A).
FIGURE 3
State distribution and evolution. Parts (A–F) show AHADR levels in selected years; Part (G) presents the distribution of resilience levels across regions;
and Part (H) shows transitions among the low, medium, high, and very-high resilience states. The percentages on the arrows and looped paths indicate
transition and persistence probabilities, respectively.
Zhang et al. 10.3389/fsufs.2026.1930549
Frontiers in Sustainable Food Systems 06 frontiersin.org
identifies specific patterns of spatial agglomeration. Together,
these statistics describe both global and localized spatial pat -
terns in AHADR.
2.3.1 Global Moran’s I
( )( )
( )
= =
−−
= ×
−
∑∑
∑
11
2
1
nn
ij i j
ij
n
i
i
wy yy y
nI S
yy
(9)
In Equation 9,
= =
=∑∑
nn
ij
i 1j 1
Sw , n is the number of spatial units; iy and
jy represent AHADR in units i and j , y is the mean AHADR across
units, ijw is a spatial weight that captures the inverse distance between
national capitals and is defined as follows.
≠ −=
=
1 ,
0,
ijij
ij
distance distancew
ij
(10)
In Equation 10 , > 0I denotes positive spatial autocorrelation,
indicating that units with similar AHADR values cluster together;
< 0I denotes negative spatial autocorrelation, indicating that units
with dissimilar AHADR values cluster together; if I is close to 0,
AHADR is randomly distributed, indicating little or no spatial
autocorrelation.
2.3.2 Local Moran’s I
≠
= ∑2
n
ii ij j
ji
ZI wZ
T
(11)
In Equation 11, = −iiZ yy , = −jjZ yy , ( )== −∑ 22 11 n iiT yy n . To measure the relationship between each unit and its surrounding geo-
graphical units, we use iI to determine whether countries or regions
display spatial agglomeration.
0iI denotes positive spatial autocorrelation, whereby a high-
AHADR unit is surrounded by high-AHADR units (HH) or a low-
AHADR unit is surrounded by low-AHADR units (LL); < 0iI denotes
negative spatial autocorrelation, whereby a low-AHADR unit is sur-
rounded by high-AHADR units (low-high) or a high-AHADR unit is
surrounded by low-AHADR units (high-low). The results are shown
in Appendix Figure A4.
2.4 Decomposition of regional differences
Compared with the traditional Gini and Theil indices, the Dagum
Gini coefficient can effectively decompose regional differences. The
Dagum Gini coefficient accounts for the distributions of subsamples
and their overlap, thereby yielding more accurate conclusions.
Following Dagum (1997) and the associated subgroup decomposition
method, the Gini coefficient is defined as follows.
= = = =
−
∑∑∑∑
1 11 1
22
kkkk
ji hr
jhir
yy
G
ny
⛂⛂
(12)
In Equation 12, G represents the overall Gini coefficient and thus
the overall gap. k indicates the total number of regions, while i and r
index countries within regions. ( )jhnn represents the number of
countries in region ( )jh , and ( )ji hyy is the AHADR value of a coun-
try in region ( )jh . n represents the total number of countries, and y is
the mean AHADR. For the Gini coefficient decomposition, the
regional mean AHADR values are first ordered as shown in
Equation 13.
≤≤h jkYYY (13)
Here, wG represents differences in the distribution of AHADR
within region j (h). nbG represents the distributional gap in AHADR
between regions j and h. The relative-density contribution tG captures
the effect of overlapping AHADR distributions across regions on over-
all inequality. If no overlap exists, the relative-density contribution is
zero. W e define wG , nbG , and tG in detail in Equations 12–19.
= =
−
∑∑
11
2
1
2
jjnn
ji jr
j irjj
j
yyY
G
n
⛂⛂
(14)
==∑ 1kw jj j jjG G ps (15)
( )
= =
−
∑∑
11
j hn n
ji jr
irjh
jh j h
yy
G
nn Y Y
⛂⛂
(16)
( )
−
= == +∑∑ 121kjnb jh j h h j jhjhG G ps ps D (17)
( )( )
−
= == +−∑∑ 121 1kjnb jh j h h j jhjhG G ps ps D (18)
−= +jh jhjh jh jhdpD dp (19)
( ) ( ) ( )
∞
= −∫∫00
y
jh j hd dF y y x dF x (20)
( ) ( ) ( )
∞
= −∫∫00
y
jh h jp dF y y x dF x (21)
From Equations 14–18, = /jjp nn and = /j jjs n Y nY (j = 1, 2, ., k),
jhD is the relative influence of AHADR between regions j and h and
is defined in Equation 19. jhd represents the difference in AHADR
between j and h, specifically the weighted mean difference when
−> 0ji hryy holds for regions j and h. The term jhd is defined in
Equation 20. jhp in Equation 21 represents the first-order moment and
the expected AHADR difference when −> 0hr jiyy holds in regions j
and h; ( )jhFF is the cumulative distribution function for region j (h).
Zhang et al. 10.3389/fsufs.2026.1930549
Frontiers in Sustainable Food Systems 07 frontiersin.org
3 Results
3.1 Factors affecting AHADR
Table 1 presents the estimated AHADR equations. Models (1) to
(8) report the mean, variance, and resilience equations, respectively.
Several factors significantly affect the mean, variance, and
AHADR of grain output. Table 1 reports coefficients estimated using
generalized linear models (GLMs) with conventional and Driscoll-
Kraay standard errors. Lagged TLU may have a significant negative
effect on mean grain output, but this effect appears only in the GLM
specification. When we account for autocorrelation using Driscoll-
Kraay standard errors, the effect disappears. Several production fac-
tors show significant U-shaped relationships with mean grain output.
Specifically, fertilizer and machinery exhibit significant U-shaped
relationships with mean output, suggesting that the law of diminishing
marginal returns does not apply uniformly to these inputs.
However, their impacts on conditional variance and AHADR
differ markedly. Labor, fertilizer, and livestock exert no statistically
significant influence on the variance of grain output. Notably, irrigated
area does not exhibit a statistically significant effect on production
variance. Irrigation nevertheless represents a pivotal adaptation strat-
egy for coping with climate change and safeguarding agricultural pro-
ductivity. For populous developing countries such as China and India,
failure to mitigate drought risks via climate adaptation measures will
amplify food insecurity (Birthal and Hazrana, 2019).
Agricultural mechanization has a significant U-shaped effect on
AHADR. At low levels of mechanization, farmers cannot fully offset
the effects of extreme climate events on sowing, crop growth, and har-
vesting, even when they receive early warnings from weather forecasts.
By contrast, adequate mechanization allows farmers to adapt machin-
ery operations to weather conditions and reduce climate-induced
losses in grain production. Notably, our results show that the adverse
effect of extreme drought events weakens by 7.69% after controlling
for fertilizer and agricultural machinery inputs.
Lastly, severe drought events significantly reduce mean grain
output while increasing production variance and reducing
AHADR. By contrast, severe flood events significantly increase pro-
duction variance but have only a negligible effect on AHADR and no
robust effect on mean output. Consistent with Lesk et al. (2016), this
study finds that extreme drought events lower average grain produc-
tion by 10.6%. Quantifying the impact of floods on grain output
remains difficult because of limited fine-scale topographic data. As
Lesk et al. (2016) note, flood impacts are jointly shaped by meteoro-
logical conditions and terrain features, which complicates the identi-
fication of flood-induced changes in grain yields. This constraint
prevents us from fully capturing the heterogeneous effects of extreme
floods. Additionally, extreme drought events increase the variance of
grain production. Because drought shocks are stochastic, they gener-
ate large differences in grain yields across years and regions and cause
excessive production volatility. For example, single-crop systems have
limited capacity to buffer climate change and extreme weather shocks.
Extreme heat shortens the grain-filling stage of wheat and rice,
depressing average yields and increasing production volatility (Zhang
et al., 2022). As Zampieri et al. (2020) point out, agricultural systems
have become less adaptive to gradual climate change and extreme
weather in recent decades. When exposed to extreme drought, average
grain yields decline, output volatility rises, and adverse production
outcomes become more pronounced.
3.2 Spatiotemporal distribution of AHADR
First, over the study period AHADR followed a “decline–rise–
decline” trajectory, reaching its highest value in the early 2000s, then
entering a sustained decline that reached its lowest point around 2013,
followed by a partial recovery and a shallower renewed decline in the
later period (see Part (A) of Figure 1). Before 2013, variations in
AHADR were closely associated with the frequency of severe
droughts: in years with fewer severe droughts, resilience values were
generally higher, while in periods of frequent drought, values were
lower. Between 2001 and 2013, extreme weather conditions—particu-
larly severe droughts—were likely the primary drivers of AHADR
fluctuations. After 2012, however, this association weakened, indicat-
ing that broader and more complex factors began to influence
AHADR. Since the early 2010s, AHADR has come under renewed
pressure, driven mainly by ongoing local conflicts (van Weezel, 2018)
and the severe disruptive effects of a strong El Niño event (Thirumalai
et al., 2017). The average number of armed conflicts was below 100
between 2005 and 2010, but exceeded 148 after 2011. The 2015–2016
El Niño event brought a particularly difficult period, during which
AHADR remained persistently depressed. However, the 2016 World
Humanitarian Summit, which emphasized the critical role of interna-
tional assistance, marked a turning point. Thereafter, AHADR gradu-
ally improved, reflecting the global commitment to strengthening
humanitarian and development responses.
Second, AHADR trends among countries in different income
groups broadly resemble the overall pattern shown in Part (A),
although the ranking across income strata is not as clear-cut as a
strict income hierarchy. Here, HI, LI, MI-U, and MI-L denote high-
income, low-income, upper-middle-income, and lower-middle-
income countries, respectively (see Part (B) of Figure 1 ). In the
earlier period, both low-income (LI) and lower-middle-income
(MI-L) countries exhibited relatively higher AHADR values com -
pared with the other income groups, while high-income (HI) coun-
tries consistently occupied the lowest position among the four
groups. This pattern suggests that the relationship between national
income and AHADR is not a simple linear hierarchy, but is instead
jointly shaped by a country’s economic strength, the stability of its
agricultural trade, and its exposure to production risk—factors that
play a key role in climate-adaptive behavior and the climate adapt -
ability of agricultural practices. Since 2010, the co-movement among
the four income groups has weakened further, and their relative
positions have become more unstable. By 2013, the proportion of
internal (intra-regional) conflicts had declined from 61.24% in 2000
to 54.46%, indicating a shift in the nature of global conflict. In addi-
tion, the intensity of agricultural policy interventions in the 2010s
was significantly greater than in the early 2000s ( Wuepper et al.,
2024). During this period, notable disparities emerged among coun-
tries at different income levels in terms of agricultural policy and
armed conflict, which may have contributed to the weakening and
reconfiguration of the relationship between income and resilience.
As a result, armed conflict, agricultural policy, and other factors have
made the relationship between economic status and agricultural
disaster risk more complex. In short, the association between
national income and agricultural disaster risk shifted from a rela -
tively stable pattern before 2013 to a more volatile and less predict -
able one thereafter.
Third, we divide the world into seven regions: ASIA (Asia),
CW ANA (Central and West Asia and North Africa), EUROPE
Zhang et al. 10.3389/fsufs.2026.1930549
Frontiers in Sustainable Food Systems 08 frontiersin.org
(Europe), LAC (Latin America and the Caribbean), NORTH
AMERICA (North America), OCEANIA (Oceania), and SSA (Sub-
Saharan Africa) (see Part (C) of Figure 1 ). Since 2016, AHADR
levels have continued to decline in most regions. In CW ANA, scarce
rainfall makes the agricultural system particularly vulnerable to
climate change, population growth, and inadequate land-use plan -
ning, leaving the region exposed to a persistent risk of famine. Local
conflicts, agricultural systems, and environmental fragility further
compound this situation. A 2017 FAO study found that conflict
alone increases hunger by 1.4–4.4%, while national institutions,
environmental fragility, and local conflict together increase hunger
by 11–18%. The region’s harsh climate, complex resource distribu -
tion, and persistent ethnic tensions have consistently kept its
AHADR at a low level.
Fourth, following the FAO’s definition, AHADR can be divided
into four resilience states based on the 25th, 50th, 75th, and 99th per-
centiles: low, medium, high, and very high (see Part (D) of Figure 1).
The results show that the low category remained at a low level
throughout all years, while the medium, high, and very high categories
all showed a continuous decline after 2016.
Fifth, inequality across regions shows an overall upward trend,
albeit with considerable fluctuation. We use the Dagum Gini coef -
ficient (Dagum, 1997 ) to measure and decompose this inequality.
From the early to mid-2000s, the coefficient remained at a relatively
low and stable level, before beginning to rise around 2010, with
year-to-year fluctuations, reaching its highest point around 2017
(see Part (E) of Figure 1). Before 2010, local conflict and extreme
climate events affected agricultural production and trade, keeping
the Gini coefficient relatively stable. Between 2010 and 2017,
regional disparities in resilience widened, increasing the explana -
tory power of the Dagum Gini coefficient; during this period, dif -
fering positions on local disputes may have intensified internal
divisions. After 2017, although climate-driven disparities in cereal
production across countries remained unevenly distributed, inter -
national humanitarian assistance helped mitigate the upward trend
in regional inequality.
Finally, the contribution structure of the Dagum coefficient
remained stable over time. Intra-group variation accounted for 18
to 21% of the coefficient, while inter-group variation—the primary
source—accounted for 38 to 52% (see Part (F) of Figure 1), and the
transvariation (hypervariation) density accounted for 28 to 42%.
Owing to differences in geographical endowments, agricultural
technology, and climatic conditions, AHADR varies across regions.
Furthermore, we find that the gap between CW ANA and the other
four major regions significantly influences the Dagum Gini
coefficient.
3.3 Spatial agglomeration of AHADR
The local Moran’s I decomposition shows that 11.32% of countries
fall into the low-resilience agglomeration (LL) category, whereas
59.41% are classified as high-resilience agglomeration (HH) (see Part
(A) of Figure 2). Within the HH group, 19.24% of countries are in
ASIA, 10.72% in CW ANA, 34.31% in Europe, 17.81% in LAC, 1.37%
in NORTH AMERICA, and 16.56% in SSA. The LL group contains
fewer countries and is concentrated in CW ANA (46.88%) and SSA
(39.06%) (see Part (C) of Figure 2).
All countries in North America belong to the HH group, while
Europe is dominated by HH (63.91%), with the remainder in HL
(36.09%). Oceania shows a mixed distribution: HH accounts for
37.04%, LH for 1.85%, LL for 30.86%, and HL for 30.25%. In Asia, HH
accounts for 35.10%, LH for 5.96%, LL for 15.78%, and HL for 43.56%,
indicating that a large share of Asian countries exhibit high-low resil-
ience patterns. LAC presents a different pattern: HH accounts for
61.63%, LH for 13.38%, LL for 1.85%, and HL for 23.26%. In Sub-
Saharan Africa (SSA), HH is the dominant category at 78.01%, fol-
lowed by LH (4.35%), LL (8.70%), and HL (8.94%). This regional
distribution reveals widely divergent resilience profiles across conti-
nents and highlights the need for targeted strategies that address the
specific drought risks faced by each region. Additionally, 39.29% of
the population in CW ANA lives in poverty, predominantly in West
Asian countries located north of the Sahara and near the Indian
Ocean, including Cyprus, Israel, Jordan, Kuwait, Lebanon, Libya,
Oman, Palestine, Qatar, the United Arab Emirates, and Y emen (see
Part (C) of Figure 2). These regions are significantly affected by the
Somali cold wave.
Moran’s I is significantly different from zero at the 1% level from
2002 to 2019, indicating spatial agglomeration in AHADR across
countries. Moran’s I exhibits three stages: a prolonged decline, a sus-
tained rise, and a subsequent high plateau (see Part (B) of Figure 2).
From 2002 to 2005, Moran’s I decreased, indicating weaker spatial
agglomeration in AHADR. The index then increased rapidly from
2005 to 2015, indicating stronger spatial agglomeration. It remained
high from 2015 to 2019.
3.4 State distribution and evolution of
AHADR
In 2001 (A), nearly all countries worldwide are colored dark
green, indicating a globally very high resilience baseline at the start
of the study period. A sharp decline is evident by 2005 (B): North
America, most of Europe, and the Russian Federation shift to
yellow (Low), while only scattered nations in Africa and South
America retain medium-to-high resilience. This low-resilience
footprint persists through 2009 (C) and 2013 (D), with large por -
tions of Asia and the Americas remaining in the Low category,
interspersed with patchy Medium zones in Latin America and sub-
Saharan Africa. From 2017 (E) onward, a partial recovery is visible:
extensive regions in North America, northern Asia, and Oceania
move from yellow to light green (Medium). By 2019 (F), the global
landscape is dominated by Medium and High tiers, although iso -
lated pockets of Low resilience remain in Central Africa, parts of
the Middle East, and Central America.
Part (G) is a 100% stacked horizontal bar chart showing the dis-
tribution of countries across four resilience levels (low, medium, high,
and very high) in seven regions: Asia, CW ANA, Europe, LAC,
America, Oceania, and SSA. Oceania has the largest share of very-
high-AHADR countries (52.44%), whereas SSA has the largest share
of low-AHADR countries (44.74%). CW ANA has the smallest share
of low-AHADR countries (5.26%), with medium and high AHADR
each accounting for 39.47%. Europe is dominated by medium
AHADR (57.90%), and very high AHADR accounts for 33.80% in
Asia. America and LAC have relatively even distributions across the
four resilience tiers. Together, these patterns demonstrate substantial
regional heterogeneity in national AHADR.
Among countries categorized as having low resilience, 82%
remain in that state over the study period, indicating substantial per-
sistence (see Part (H) of Figure 3). Only 18% transition to medium
(17.42%) or high (0.58%) resilience. Among countries in the high-
resilience state, 63.54% remain there, 16.98% advance to very high
Zhang et al. 10.3389/fsufs.2026.1930549
Frontiers in Sustainable Food Systems 09 frontiersin.org
resilience, and 16.98% decline to medium resilience. Of countries
initially in the medium-resilience state, 20.35% transition to high
resilience, 6.64% to very high resilience, and 16.11% to low resilience.
Finally, 71.89% of countries in the very-high-resilience state remain
there, whereas 17.44% move to high resilience, 4.03% to low resilience,
and 0.44% to medium resilience. These data highlight the challenges
countries face in strengthening agricultural development resilience
against hunger.
4 Discussion and implications
Against the backdrop of a projected 30% increase in global
hunger risk between 2010 and 2050 (Van Dijk et al., 2021), alleviating
hunger and strengthening agricultural development resilience are
urgent global priorities. The AHADR framework developed in this
study offers a standardized, cross-country metric for this purpose.
Built on the mean–variance properties of grain production and the
Gamma distribution and anchored to the FAO’s minimum per capita
consumption threshold, it measures a country’s probability of sus-
taining a grain-production level sufficient to prevent hunger during
shocks and is less distorted by single-year outliers than single-indi-
cator measures. For the 161 countries observed from 2001 to 2019,
the index shows that global AHADR fluctuated markedly and ended
the period only modestly above its initial level, even as its geographi-
cal distribution became substantially more unequal.
Our estimates identify severe drought as the most destabilizing
force affecting AHADR. Consistent with Lesk et al. (2016) , severe
drought reduces mean grain output by about 10.6%, increases pro -
duction variance, and lowers AHADR. By contrast, severe floods
increase variance but leave AHADR essentially unchanged because
flood damage is governed by fine-grained topography that national
data cannot capture. This asymmetry explains why AHADR tracked
drought frequency so closely before the early 2010s and why
drought, rather than flooding, should be the primary concern for
anti-hunger policy. Our results also identify meaningful coping
capacity. After controlling for fertilizer and agricultural machinery
inputs, the adverse effect of drought on AHADR weakens by 7.69%.
Fertilizer use follows an inverted-U path: moderate applications
raise resilience, whereas overuse erodes it, presumably through soil
degradation and diminishing marginal returns. Mechanization
instead has a U-shaped effect: at low levels, farmers cannot translate
weather forecasts into timely sowing, field operations, or harvest -
ing, whereas adequate mechanization permits site-specific opera -
tions that absorb climate shocks. Together with farmers’ own
adaptations, such as soil improvement and adjustments to sowing
dates and cultivars ( Qiao et al., 2022 ; Minoli et al., 2022 ), these
results imply that adaptive capacity, rather than input levels alone,
mediates the translation of shocks into resilience.
Three spatial patterns have important implications. First,
AHADR is strongly and positively clustered: global Moran’s I is
significantly different from zero at the 1% level throughout 2002–
2019, with 59.41% of countries falling into a high-resilience clus -
ter (HH) and 11.32% into a low-resilience cluster (LL). The LL
group is concentrated primarily in CW ANA (46.88%) and SSA
(39.06%). Second, regional inequality has widened rather than
narrowed: the Dagum Gini coefficient trended upward from
around 2010 to a peak near 2017, and inter-group differences,
rather than within-group dispersion, account for most of this
inequality (38–52%), with the gap between CW ANA and other
regions contributing disproportionately. After 2012–2013, the ear -
lier co-movement between national income and AHADR weak -
ened as ongoing local conflicts ( Van Weezel, 2018 ), the severe
2015–2016 El Niño event ( Thirumalai et al., 2017 ), and more
intensive agricultural policy intervention ( Wuepper et al., 2024 )
reshaped the map. Third, this divergence is geographically con -
centrated: SSA has the largest share of low-AHADR countries
(44.74%), while CW ANA remains in a low-resilience state because
of scarce rainfall, conflict, and institutional fragility.
The state-transition analysis sharpens the policy message. Low-
resilience states are remarkably sticky: roughly 82% of countries
starting in the lowest tier remained there, and fewer than one-fifth
moved upward. By contrast, high- and very-high-resilience states
are far more dynamic, with about 72% of very-high countries
retaining that status and the rest shifting only to adjacent tiers.
Resilience thus behaves as a stock rather than a flow: once a country
falls into the low tier, climbing out is rare, whereas already-resilient
systems continue to adapt. Early, sustained support therefore mat -
ters more than late rescue, and the marginal return of additional
investment is highest for countries poised to transition.
These findings have direct policy implications, particularly
under pressure to reduce carbon emissions ( Crippa et al., 2022 ;
Laborde et al., 2021 ). Climate-smart agriculture and nature-
based solutions offer a credible way to manage this trade-off
because they can increase yields and stabilize output simultane -
ously ( Lipper et al., 2014 ). More specifically, investment should
prioritize breeding drought-tolerant cultivars and conducting
heat- and drought-stress experiments ( Vadez et al., 2024 ; Xiong
et al., 2021 ); fertilizer policy should target the moderate range
of the inverted-U relationship; and mechanization support
should focus on countries below the turning point of the
U-shaped relationship, where adaptive operations remain out of
reach. Because low-AHADR countries are spatially clustered and
concentrated in SSA and CW ANA, aid should be geographically
targeted and should reinforce the humanitarian-assistance
momentum established at the 2016 World Humanitarian Summit.
Several limitations qualify these conclusions. Our analysis is
conducted at the national level, which obscures subnational het -
erogeneity; analysis at regional or township levels should improve
both measurement and targeting. Flood impacts remain imper -
fectly identified because fine-scale topographic data are limited.
Like any probability-based metric, AHADR depends on the
chosen consumption threshold and distributional assumptions
and should be considered alongside indicators of nutrition and
market access. Our AHADR index mainly reflects absorptive
capacity but does not fully capture adaptive or transformative
resilience. Nor does it cover non-grain food sources or the acqui -
sition and utilization dimensions of food security. Finally, the
index is descriptive rather than causal; future research could
exploit within-country variation and quasi-experimental changes
in adaptation policy to identify the mechanisms through which
resilience is built.
Data availability statement
The original contributions presented in the study are included in
the article/Supplementary material, further inquiries can be directed
to the corresponding author/s.
Zhang et al. 10.3389/fsufs.2026.1930549
Frontiers in Sustainable Food Systems 10 frontiersin.org
Author contributions
ZZ: Methodology, Writing – original draft, Conceptualization,
Writing – review & editing, Data curation. Y-hC: Writing – original
draft, Investigation, Methodology, Writing – review & editing, Formal
analysis. WC: Writing – original draft, Conceptualization, Writing –
review & editing, Methodology, Formal analysis.
Funding
The author(s) declared that financial support was not received for
this work and/or its publication.
Acknowledgments
We sincerely thank the FAO for providing the necessary data for
this research.
Conflict of interest
The author(s) declared that this work was conducted in the
absence of any commercial or financial relationships that could be
construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that Generative AI was not used in the
creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this
article has been generated by Frontiers with the support of artificial
intelligence and reasonable efforts have been made to ensure accuracy,
including review by the authors wherever possible. If you identify any
issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors
and do not necessarily represent those of their affiliated organizations,
or those of the publisher, the editors and the reviewers. Any product
that may be evaluated in this article, or claim that may be made by its
manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary material for this article can be found online
at: https://www.frontiersin.org/articles/10.3389/fsufs.2026.1930549/
full#supplementary-material
References
Akaike, H. (1974). A new look at the statistical model identification. IEEE Trans. Autom.
Control 19, 716–723. doi: 10.1109/tac.1974.1100705
Birthal, P . S., and Hazrana, J. (2019). Crop diversification and resilience of agriculture to
climatic shocks: evidence from India. Agric. Syst. 173, 345–354. doi: 10.1016/j.
agsy.2019.03.005
Chen, J., Engbersen, N., Stefan, L., Schmid, B., Sun, H., and Schöb, C. (2021). Diversity
increases yield but reduces harvest index in crop mixtures. Nat. Plants 7, 893–898. doi:
10.1038/s41477-021-00948-4
Christopher, M., and Peck, H. (2004). Building the resilient supply chain. Int. J. Logist.
Manage. 15, 1–14. doi: 10.1108/09574090410700275
Cissé, J. D., and Barrett, C. B. (2018). Estimating development resilience: a conditional
moments-based approach. J. Dev. Econ. 135, 272–284. doi: 10.1016/j.jdeveco.2018.04.002
Crippa, M., Solazzo, E., Guizzardi, D., Tubiello, F . N., and Leip, A. (2022). Climate goals require
food systems emission inventories. Nat. Food 3, 1–1. doi: 10.1038/s43016-021-00450-2
Dagum, C. (1997). A new approach to the decomposition of the Gini income inequality
ratio. Empir. Econ. 22, 515–531. doi: 10.1007/bf01205777
Eurasia Media Network. (2024) Global Food Security: Lessons from China's Achievements
[EB/OL]. (2024-10-22)[2026-09-12]. Available online at: https://eurasiamedianetwork.
com/global-food-security-lessons-from-chinas-achievements/.
FAO (2021). In Brief The State of Food and Agriculture 2021. Rome: Food and Agriculture
Organization.
Foley, J. A., Ramankutty, N., Brauman, K. A., Cassidy, E. S., Gerber, J. S., Johnston, M.,
et al. (2011). Solutions for a cultivated planet. Nature 478, 337–342. doi: 10.1038/
nature10452
Friel, S., Schram, A., and Townsend, B. (2020). The nexus between international trade,
food systems, malnutrition and climate change. Nat. Food 1, 51–58. doi: 10.1038/
s43016-019-0014-0
Hasegawa, T., Sakurai, G., Fujimori, S., Takahashi, K., Hijioka, Y ., and Masui, T. (2021).
Extreme climate events increase risk of global food insecurity and adaptation needs. Nat.
Food 2, 587–595. doi: 10.1038/s43016-021-00335-4
Holling, C. S. (1973). Resilience and stability of ecological systems. Annu. Rev. Ecol. Syst.
4, 1–23. doi: 10.1146/annurev.es.04.110173.000245
Kahiluoto, H., Kaseva, J., Balek, J., Olesen, J. E., Ruiz-Ramos, M., Gobin, A., et al. (2019).
Decline in climate resilience of European wheat. Proc. Natl. Acad. Sci. 116, 123–128. doi:
10.1073/pnas.1804387115
Laborde, D., Mamun, A., Martin, W ., Piñeiro, V ., and Vos, R. (2021). Agricultural subsi-
dies and global greenhouse gas emissions. Nat. Commun. 12:2601. doi: 10.1038/
s41467-021-22703-1
Leach, M., Nisbett, N., Cabral, L., Harris, J., Hossain, N., and Thompson, J. (2020).
Food politics and development. World Dev. 134:105024. doi: 10.1016/j.
worlddev.2020.105024
Lesk, C., Rowhani, P ., and Ramankutty, N. (2016). Influence of extreme weather disasters
on global crop production. Nature 529, 84–87. doi: 10.1038/nature16467
Lipper, L., Thornton, P ., Campbell, B. M., Baedeker, T., Braimoh, A., Bwalya, M., et al.
(2014). Climate-smart agriculture for food security. Nat. Clim. Chang. 4, 1068–1072. doi:
10.1038/nclimate2437
Luthans, F ., Vogelgesang, G. R., and Lester, P . B. (2006). Developing the psychological
capital of resiliency. Hum. Resour. Dev. Rev. 5, 25–44. doi: 10.1177/1534484305285335
Meuwissen, M. P . M., Feindt, P . H., Spiegel, A., Termeer, C. J. A. M., Mathijs, E., de Mey, Y .,
et al. (2019). A framework to assess the resilience of farming systems. Agric. Syst.
176:102656. doi: 10.1016/j.agsy.2019.102656
Minoli, S., Jägermeyr, J., Asseng, S., Urfels, A., and Müller, C. (2022). Global crop yields
can be lifted by timely adaptation of growing periods to climate change. Nat. Commun.
13:7079. doi: 10.1038/s41467-022-34411-5
Mu, W ., van Asselt, E. D., and van der Fels-Klerx, H. J. (2021). Towards a resilient food
supply chain in the context of food safety. Food Control 125:107953. doi: 10.1016/j.
foodcont.2021.107953
Qiao, L., Wang, X., Smith, P ., Fan, J., Lu, Y ., Emmett, B., et al. (2022). Soil quality both
increases crop production and improves resilience to climate change. Nat. Clim. Chang.
12, 574–580. doi: 10.1038/s41558-022-01376-8
Renard, D., and Tilman, D. (2019). National food production stabilized by crop diversity.
Nature 571, 257–260. doi: 10.1038/s41586-019-1316-y
Smith, L. C., and Frankenberger, T. R. (2018). Does resilience capacity reduce the negative
impact of shocks on household food security? Evidence from the 2014 floods in northern
Bangladesh. World Dev. 102, 358–376. doi: 10.1016/j.worlddev.2017.07.003
Zhang et al. 10.3389/fsufs.2026.1930549
Frontiers in Sustainable Food Systems 11 frontiersin.org
Tendall, D. M., Joerin, J., Kopainsky, B., Edwards, P ., Shreck, A., Le, Q. B., et al. (2015).
Food system resilience: defining the concept. Glob. Food Secur. 6, 17–23. doi: 10.1016/j.
gfs.2015.08.001
Thirumalai, K., DiNezio, P . N., Okumura, Y ., and Deser, C. (2017). Extreme temperatures
in Southeast Asia caused by El Niño and worsened by global warming. Nat. Commun.
8:15531. doi: 10.1038/ncomms15531
Tilman, D., Balzer, C., Hill, J., and Befort, B. L. (2011). Global food demand and the
sustainable intensification of agriculture. Proc. Natl. Acad. Sci. 108, 20260–20264. doi:
10.1073/pnas.1116437108
Vadez, V ., Grondin, A., Chenu, K., Henry, A., Laplaze, L., Millet, E. J., et al. (2024). Crop
traits and production under drought. Nat. Rev. Earth Environ. 5, 211–225. doi: 10.1038/
s43017-023-00514-w
Van Dijk, M., Morley, T., Rau, M. L., and Saghai, Y . (2021). A meta-analysis of projected
global food demand and population at risk of hunger for the period 2010–2050. Nat. Food
2, 494–501. doi: 10.1038/s43016-021-00322-9
Van Weezel, S. (2018). Food security and armed conflict: a cross-country analysis. FAO
Agricultural Development Economics Working Paper 18-03. Rome, FAO. 28. Licence:
CC BY-NC-SA 3.0 IGO. Available online at: https://www.fao.org/agrifood-economics/
publications/detail/zh/c/1153220/
Wuepper, D., Wiebecke, I., Meier, L., Vogelsanger, S., Bramato, S., Fürholz, A., et al.
(2024). Agri-environmental policies from 1960 to 2022. Nat. Food 5, 323–331. doi:
10.1038/s43016-024-00945-8
Xiong, W ., Reynolds, M. P ., Crossa, J., Schulthess, U., Sonder, K., Montes, C., et al. (2021).
Increased ranking change in wheat breeding under climate change. Nat. Plants 7,
1207–1212. doi: 10.1038/s41477-021-00988-w
Zampieri, M., Ceglar, A., Dentener, F ., and Toreti, A. (2017). Wheat yield loss attributable
to heat waves, drought and water excess at the global, national and subnational scales.
Environ. Res. Lett. 12:064008. doi: 10.1088/1748-9326/aa723b
Zampieri, M., Toreti, A., Ceglar, A., Naumann, G., Turco, M., and Tebaldi, C. (2020).
Climate resilience of the top ten wheat producers in the Mediterranean and the Middle
East. Reg. Environ. Chang. 20:41. doi: 10.1007/s10113-020-01622-9
Zhang, T., He, Y ., DePauw, R., Jin, Z., Garvin, D., Yue, X., et al. (2022). Climate change
may outpace current wheat breeding yield improvements in North America. Nat.
Commun. 13:5591. doi: 10.1038/s41467-022-33265-1
论文原文与出版信息
Frontiers in Sustainable Food Systems