Yang, Shuhui; Li, Zhongkai; Zhou, Jianlin; Gao, Yancheng; Cui, Xue eng
A icle — Published Ve sion
E ol ing pa e ns o ag icul u al p oduc ion space in
China: A ne wo k-based app oach
Geog aphy and Sus ainabili y
P o ided in Coope a ion wi h:
Leibniz Ins i u e o Ag icul u al De elopmen in T ansi ion Economies (IAMO), Halle (Saale)
Sugges ed Ci a ion: Yang, Shuhui; Li, Zhongkai; Zhou, Jianlin; Gao, Yancheng; Cui, Xue eng (2024) :
E ol ing pa e ns o ag icul u al p oduc ion space in China: A ne wo k-based app oach, Geog aphy
and Sus ainabili y, ISSN 2666-6839, Else ie , Ams e dam, Vol. 5, Iss. 1, pp. 121-134,
h ps://doi.o g/10.1016/j.geosus.2023.11.007 ,
h ps://www.sciencedi ec .com/science/a icle/pii/S2666683923000755
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/281442
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
Geog aphy and Sus ainabili y 5 (2024) 121–134
Con en s lis s a ailable a ScienceDi ec
Geog aphy and Sus ainabili y
jou nal homepage: www.else ie .com/loca e/geosus
Resea ch A icle
E ol ing pa e ns o ag icul u al p oduc ion space in China: A
ne wo k-based app oach
Shuhui Yang
a , d
, Zhongkai Li
b , c
, Jianlin Zhou
a
, Yancheng Gao
a
, Xue eng Cui
a , ∗
a
School o Sys ems Science, Beijing No mal Uni e si y, Beijing 100875, China
b
S a e Key Labo a o y o Ea h Su ace P ocesses and Resou ce Ecology, Facul y o Geog aphical Science, Beijing No mal Uni e si y, Beijing 100875, China
c
School o Na u al Resou ces, Facul y o Geog aphical Science, Beijing No mal Uni e si y, Beijing 100875, China
d
Leibniz Ins i u e o Ag icul u al De elopmen in T ansi ion Economies (IAMO), Halle (Saale) 06120, Ge many
h i g h l i g h s g a p h i c a l a b s a c
•Ag icul u al p oduc ion space is cha ac-
e ized by bipa i e ne wo k.
•P oduc and p o ince spaces exhibi well
co e-pe iphe y and communi y s uc-
u es.
•The communi y changes show ela i ely
s able p oduc ion pa e ns.
•The s udy offe s a holis ic iew o unde -
s and ag icul u al p oduc ion sys em.
a i c l e i n o
A icle his o y:
Recei ed 10 Sep embe 2023
Recei ed in e ised o m 9 No embe 2023
Accep ed 27 No embe 2023
A ailable online 18 Decembe 2023
Keywo ds:
Ag icul u al sys em
Complex ne wo k
Ag icul u al p oduc ion space
P oximi y ma ix
P oduc ion capabili y
a b s a c
The ag icul u al p oduc ion space, as whe e and how much each ag icul u al p oduc g ows, plays a i al ole
in mee ing he inc easing and di e se ood demands. P e ious s udies on ag icul u al p oduc ion pa e ns ha e
p edominan ly cen e ed on indi idual o specific c op ypes, using me hods such as emo e sensing o s a is ical
me ological analysis. In his s udy, we cha ac e ize he ag icul u al p oduc ion space (APS) by bipa i e ne wo k
connec ing ag icul u al p oduc s and p o inces, o e eal he ela edness be ween di e se ag icul u al p oduc s
and he spa io empo al cha ac e is ic o p o incial p oduc ion capabili ies in China. The esul s show ha co e
p oduc s a e ce eal, po k, melon, and pome ui ; meanwhile he milk, g ape, and fibe c op show an upwa d
end in cen ali y, which is in line wi h die s uc u e changes in China o e he pas decades. The li le changes in
communi y componen s and s uc u es o ag icul u al p oduc s and p o inces e eal ha ag icul u al p oduc ion
pa e ns in China a e ela i ely s able. Addi ionally, iden ified p o incial communi ies closely esemble China’s
ag icul u al na u al zones. Fu he mo e, he obse ed g ow h in p oduc ion capabili ies in No h and No heas
China implies hei po en ial ocus a eas o u u e ag icul u al p oduc ion. Despi e he supe io p oduc ion capa-
bili ies o sou he n p o inces, ecen yea s ha e wi nessed a no able decline, wa an ing special a en ions. The
findings p o ide a comp ehensi e pe spec i e o unde s anding he complex ela ionship o ag icul u al p od-
uc s’ ela edness, p oduc ion capabili ies and p oduc ion pa e ns, which se e as a e e ence o he ag icul u al
spa ial op imiza ion and ag icul u al sus ainable de elopmen .
1. In oduc ion
As pa o he Uni ed Na ions Sus ainable De elopmen Goals (SDGs),
he Chinese Go e nmen has gi en p io i y o ending hunge and
ensu ing sufficien ood supply, achie ing ood sys em sus ainabili y
( Agnolucci e al., 2020 ). D i en by popula ion g ow h and economic
∗ Co esponding au ho a : No. 19 Xinjiekouwai S , Haidian Dis ic , Beijing, China.
E-mail add ess: [email p o ec ed] (X. Cui) .
de elopmen , die a y s uc u es ha e changed wi h people consuming
nume ous and di e se ood han e e be o e ( Sun e al., 2019 ). Simul-
aneously, he e is limi ed a able c opland in China, wi h o e 40%
o i classified as “mode a ely ”o “se e ely ” deg aded, due o issue
such as soil e osion, wa e pollu ion, o wa e sca ci y ( Somme e al.,
2023 ). The combina ion o hese ac o s pu s ce ain p essu e on ag i-
h ps://doi.o g/10.1016/j.geosus.2023.11.007
2666-6839/© 2023 The Au ho s. Published by Else ie B.V. and Beijing No mal Uni e si y P ess (G oup) Co., LTD. on behal o Beijing No mal Uni e si y. This is
an open access a icle unde he CC BY-NC-ND license ( h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/ )
S. Yang, Z. Li, J. Zhou e al. Geog aphy and Sus ainabili y 5 (2024) 121–134
cul u al p oduc ion in China ( Ye e al., 2022 ). The ag icul u al p oduc-
ion space, as whe e and how much each ag icul u al p oduc g ows,
plays an impo an ole in sa is ying he inc easing and di e sified ood
demands.
China’s ag icul u al p oduc ion capabili y s ands as a c i ical de e -
minan in main aining na ion’s u u e ood supply. I is affec ed by mul-
iple ac o s, including land and wa e esou ces ( Duan e al., 2021 ), cli-
ma e ( A o a, 2019 ), households’ land-use beha io s ( Liu e al., 2020 ),
e ilize and pes icide inpu s ( Seghezzo e al., 2020 ; Yu e al., 2021 ),
managemen p ac ices ( Cui e al., 2018 ), as well as na ional policies
and s a egies ( Lu e al., 2022 ). These ac o s a e p io no ions o wha
a e he impo an ac o s o ag icul u al p oduc ion. Howe e , i is no
easy o emb ace all ac o s and o quan i y all ac o s o componen s o
ag icul u al p oduc ion capabili y ( an I e sum e al., 2008 ). Ins ead,
acco ding o he agnos ic app oach, an ou comes-based app oach o -
e s an al e na i e way o measu e p oduc ion capabili y. This app oach
has been used in he in e na ional ade field, whe e a coun y’s ex-
po goods could e eal i s domes ic p oduc ion capabili ies and spe-
cializa ion pa e ns ( Scia a e al., 2020 ). In ag icul u al p oduc ion, he
ou pu o ag icul u al p oduc s can se e as a eflec ion o he p oduc-
ion capabili y o a pa icula egion. The ac ha he p oduc ion o
iden ical ag icul u al p oduc s ac oss diffe en egions may sugges a
sha ed p oduc ion capabili y o cul i a e hese p oduc s. Ag icul u al
p oduc s a e seen as a consequence o a egion’s o p o ince’s endow-
men s o capabili ies, possibly co e ing land, wa e esou ce, labo ,
echnology, knowledge and capi al ( An le e al., 2004 ; Gumbau Al-
be , 2017 ).
Analyzing ag icul u al p oduc ion sys em has been p o ed o be a
complex ask due o i s in e ela ed and he e ogeneous cha ac e is ic
( Jones e al., 2017 ). P e ious s udies such as c op dis ibu ion and i s
spa ial op imiza ion ( Hu e al., 2022 ; Q. Liu e al., 2022 ), c op yields
( Wang e al., 2022 ), c op p oduc ion condi ions ( Pickson e al., 2022 )
and so on ha e been ca ied ou . The e a e also nume ous esea ches
abou he s uc u e and space o ag icul u al p oduc ion in China, such
as spa io empo al cha ac e is ics o ag icul u al p oduc ion efficiency
( Guo e al., 2020 ), spa ial easibili y and cos -effec i eness o ag i-
cul u al land consolida ion ( Duan e al., 2021 ), p oduc i i y and en-
i onmen al cos s o diffe en -scale a ms ( Qi e al., 2018 ), he pa -
e n o p oduc di e sifica ion ( Em an and Shilpi, 2012 ). S a is ical
me ology ( Baldoni and Espos i, 2020 ) and emo e sensing wi h GIS
( Medenniko e al., 2020 ) a e commonly used in hese fields. How-
e e , hese s udies mainly ocus on single ag icul u al p oduc o some
specific p oduc s ( Geng e al., 2017 ). In e ela ion among di e se ag i-
cul u al p oduc s in he whole na ion is a ely explo ed. Fu he mo e,
cu en ly spa ial misma ch con inues o exis be ween majo p oducing
a ea, ag icul u al p oduc ion s uc u e and a mland esou ce in China
( Li e al., 2017 ). Analyzing single ag icul u al p oduc alone poses chal-
lenges in effec i ely in eg a ing a ious ag icul u al ac i i ies and p o-
duc ion capabili ies, he eby hinde ing he op imiza ion o ag icul u al
layou ( Lu e al., 2013 ).
Recen ly, complex ne wo k (CN) has been in oduced in o ag icul-
u al filed. Fo example, B anco e al. (2021) used ne wo k app oach
o op imize he spa ial dis ibu ion o new soybean and co n in B azil.
Ma e al. (2019) de eloped a ne wo k wi h 502 a ms in he UK o com-
pa e esilience o gene ically modified he bicide- ole an managemen
and con en ional weed managemen . Z. Chen e al. (2022) applied a
spa ial ne wo k amewo k o analyze he c oss- egional collabo a i e
mechanisms o ag icul u al g een de elopmen in China. Mo eo e ,
complex ne wo k app oach also has been used o measu e coun ies’
specializa ion pa e ns and analyze how hese pa e ns affec ood sup-
ply and ood secu i y in global ood sys em ( Campi e al., 2020 , 2021 ).
Combing complex ne wo k wi h global ag icul u al p oduc s is used
o explo e why and how a coun y p oduces goods, p oposing ha a
coun y’s abili y o p oduce new goods is cons ained by he combina-
ion o i s exis ing capabili ies and he acquisi ion o no el capabili ies
( Hidalgo and Hausmann, 2009 ). The global ag icul u al p oduc space
can also e eal he ela edness be ween di e se p oduc s and find diffe -
en s uc u es in diffe en ne wo k loca ions ( Hidalgo e al., 2007 ). Com-
plex ne wo k is ex ensi ely applied in a ie ies o eal-wo ld sys ems,
such as ag icul u e sys em ( Albe and Ba abási, 2002 ; Ma iani e al.,
2019 ), which can help o simula e in e connec ions among di e se ag i-
cul u al elemen s and cap u e dynamic cha ac e is ics o ag icul u al
sys em, p o iding a new pe spec i e and an effec i e ool o s udy com-
plex sys ems ( Lambio e e al., 2019 ).
In his s udy, we use complex ne wo k app oach o cha ac e -
ize China’s ag icul u al p oduc ion space, and iden i y co e-pe iphe y
s uc u e and communi y s uc u e o ag icul u al p oduc space and
p o ince space, e ealing he ela edness o mul iple ag icul u al p od-
uc s and he spa ial cha ac e is ic o p o incial p oduc ion capabili ies.
This wo k p o ides a sys ema ic and holis ic pe spec i e o in es iga e
ag icul u al p oduc ion space and s uc u e, offe s us a be e unde -
s anding o he complex ela ionship o ag icul u al p oduc ion space
and p oduc ion capabili y, and he ela edness o di e se ag icul u al
p oduc s. I is significan e e ence o upg ade and di e si y ag icul-
u al p oduc bundles in China o achie e he spa ial op imiza ion o
ag icul u al p oduc ion. We will add ess he ollowing ques ions in his
s udy:
(a) Wha cha ac e izes he “p o ince-p oduc ” ag icul u al p oduc ion
space in China om 1995 o 2019, and wha a e he s uc u es and
pa e ns o ag icul u al p oduc space and p o ince space?
(b) Wha is he ela edness o mul iple ag icul u al p oduc s and he spa-
ial cha ac e is ic o p o incial p oduc ion capabili ies unde cu en
p oduc ion s uc u e?
2. Ma e ials and me hods
2.1. Theo e ical backg ound o complex ne wo k
Complex ne wo k se es as a ype o g aph ha can abs ac ac i -
i ies o e en in o elemen s and hei ela ions ( Lacasa e al., 2008 ).
Wi hin he ne wo k amewo k, elemen s a e assigned as nodes, mean-
while connec ion o ela ionship be ween pai s o elemen s a e ega ded
as links. In his s udy, we use complex ne wo k connec ing p o inces and
ag icul u al p oduc s o build a se ies o ag icul u al p oduc ion space
in China. The ac ha diffe en p o inces p oducing he same p od-
uc s migh indica e ha hey sha e possession o he equi ed essen ial
capabili ies o p oducing hese goods ( Fig. 1 (a)).
When wo ag icul u al p oduc s sha e simila i ies in e ms o essen-
ial equi emen s such as clima e, soil, wa e , echnology, and o he
pe inen ac o s, hey end o be cul i a ed oge he in he same e-
gions. Con e sely, dissimila ag icul u al p oduc s exhibi a lowe like-
lihood o co-p oduc ion ( Fig. 1 (b)). Measu ing he simila i y o ca-
pabili ies equi ed o p oduc s and he simila i y o endowmen o
p o inces a e called “p oximi y ”in he complex ne wo k app oach.
P oximi y ma ix connec s pai s o ag icul u al p oduc s yielding he
“p oduc space ”( Hidalgo e al., 2007 ). Equally, p oximi y ma ix can
also link pai o loca ions, gi ing ise o “p o ince space ”( Baha e al.,
2014 ). The p oximi y o p oduc space and p o ince space is buil as
p oximi y ne wo k o e eal dis inc ag icul u al ne wo k cha ac e is-
ics in China ( Bö ne e al., 2012 ; Balland and Rigby, 2017 ). Despi e
hei specific s uc u es, p oximi y ne wo k se es as a ounda ion o
measu ing he ela edness among ag icul u al p oduc s ( Hidalgo e al.,
2007 ). In his s udy, ag icul u al p oduc ne wo k is p ojec ed in o
ag icul u al p o ince space and p oduc space by p oximi y ma ix
( Fig. 1 (c)).
2.2. Da a
Da a on ag icul u al p oduc ion om 1995 o 2019 o all 31
p o inces a e sou ced om he Na ional Bu eau o S a is ics in China
122
S. Yang, Z. Li, J. Zhou e al. Geog aphy and Sus ainabili y 5 (2024) 121–134
Fig. 1. Theo e ical map o he complex ne wo k in ag icul u al p oduc ion.
Table 1
Ca ego ies o ag icul u al p oduc s a Le els 1, 2 and 3.
Le el 1 Le el 2 Le el 3
Ce eal Whea
Maize
Rice
So ghum
Ba ley
Mille
Vege able and melon Melon Wa e melon Muskmelon
F ui and nu T opical and sub opical ui Banana Pineapple
Ci us
G ape
Be y S awbe y
Pome ui Apple Pea Jujube Pe simmon
Oil c op Soybean
Peanu
Tempo a y oilseed c op Flaxseed Rapeseed Sesame Sunflowe seed
Roo c op Po a o
I i an and spice c op I i an c op Tea
Bean Mung
Adzuki
Suga c op Suga bee
Suga cane
Li es ock p oduc Bee
Mu on
Po k
Milk
Poul y egg
Honey
Aqua ic p oduc Seawa e p oduc
F eshwa e p oduc
O he s Fibe c op Co on Sisal Ju e Flax Ramie
Tobacco
Sou ces : Bu eau o S a is ics in China (
h ps://da a.s a s.go .cn/easyque y.h m?cn = E0103 ); he Wo ld Ag icul u e Census P og am in 2020 o he FAO
( h ps://www. ao.o g/wo ld-census-ag icul u e/wca ounds/wca2020/en/ ).
(NBSC). P oduc classifica ion is conduc ed ollowing he Re ised In-
dica i e C op classifica ion om he Wo ld Ag icul u e Census P og am
in 2020. This classifica ion ca ego ized all ag icul u al p oduc s in o
h ee main ypes, as ou lined in Table 1 . The Le el 1, consis s o 11
g oups, while he mo e de ailed Le el 2 includes 31 specific p oduc s.
Fo he pu poses o his s udy, Le el 2 p oduc s a e used due o hei
inc eased specifici y. In case whe e NBSC does no di ec ly p o ide da a
on ce ain p oduc s, we calcula e hem om da a in Le el 3. As shown
in Table 1 , he p oduc s a e melon, opical and sub opical ui , be y,
pome ui , empo a y oilseed c op, i i an c op, and fibe c op. All
da a o he selec ed ag icul u al p oduc s a e comple ely eco ded in 31
p o inces om 1995 o 2019. Du ing his pe iod, he eco ded da a o
each p oduc emains s able, wi h a e missing alues. The p opo ion
o each ag icul u al p oduc is also demons a ed, which is in line wi h
China’s ood p oduc ion si ua ion (Fig. S1, Fig. S2).
2.3. Me hods
2.3.1. Ag icul u al p oduc ion space
To explo e how p o ince g ows ag icul u al p oduc ion, we adop
he complex ne wo k o build ag icul u al p oduc ion space, as in o-
duced by Campi e al. (2020) . This app oach enables he ep esen a ion
o he ela ionship be ween p o inces and ag icul u al p oduc s. The
specific ype o complex ne wo k in his s udy is bipa i e ne wo k which
is desc ibed as a bipa i e ma ix M . Wi hin his amewo k, bipa i e
ne wo k, designa ed as a iple G = ( C, P, X ), consis s o h ee compo-
nen s: C , ep esen ing he se o op nodes co esponding o p o inces;
P , signi ying he se o bo om nodes co esponding o ag icul u al p od-
uc s; X , deno ing he se o links. The specific elemen 𝑋𝑡
𝑖𝑘
indica es ha
he p o ince 𝑖 p oduces p oduc k in yea . We cons uc ed he APS wi h
ag icul u al p oduc s a Le el 2 om 1995 o 2019 o e 31 p o inces.
123
S. Yang, Z. Li, J. Zhou e al. Geog aphy and Sus ainabili y 5 (2024) 121–134
2.3.2. Link weigh
The link be ween wo nodes wi hin he APS indica es ha he spe-
cific p o ince p oduces he specific ag icul u al p oduc s. Mo eo e , he
weigh assigned o each link se es as an indica o o he p o ince’s ca-
pabili ies o abili ies conce ning ha pa icula p oduc . Wi h he he-
o y o an ou comes-based me hod, we use he da a o final ag icul u al
p oduc s o calcula e capabili ies o pe o mance o his p o ince in p o-
ducing a pa icula ag icul u al p oduc . I is iden ified as ollows:
𝑀𝑡
𝑖,𝑘
=𝑥(
𝑖, 𝑘
)
∑𝑘
𝑥(
𝑖, 𝑘
)
×∑𝑖
𝑥(
𝑖, 𝑘
)
∑𝑖,𝑘
𝑥(
𝑖, 𝑘
)
(1)
whe e, x is p oduc ion; 𝑥 (𝑖, 𝑘 ) is he p oduc ion o p oduc 𝑘 in p o ince
𝑖 in yea 𝑡 . The fi s ac ion is he a io o he p oduc ion o p oduc
k in ha o all p oduc s in p o ince 𝑖 , and he second ac ion is he
p opo ion o a gi en p oduc k among all p oduc s in China.
2.3.3. P oximi y ma ix
In he ne wo k-based app oach, ela edness cap u es he in e ac ions
among elemen s ha a e simila , which quan ifies hese ela edness o
simila i y call ‘p oximi y’ ( Hidalgo e al., 2007 ). P oximi y ma ix can
e alua e a p o ince’s simila capabili ies in p oducing an ag icul u al
p oduc ela ed wi h i s o he p oduc s. In ag icul u al p oduc ion space,
i wo ag icul u al p oduc s a e ela ed because hey equi e simila cli-
ma e, soil, wa e , echnology, o some combina ion he eo , hey a e
likely o be p oduced oge he in he same p o inces. Me ics o p ox-
imi y connec pai s o ag icul u al p oduc s o pai s o p o inces. The
p oximi y be ween he pai s o p oduc s k and
𝑘 is he minimum p oba-
bili ies o a p o ince p oducing a specific p oduc gi en ha i p oduces
ano he ( Hidalgo, 2021 ). To analyze he e olu ion o p oduc i e s uc-
u e, he APS is p ojec ed in o “p oduc space ”and “p o ince space ”by
defining hei simila i y wi h p oximi y ma ix. Wi hin he ag icul u al
p oduc space, each pai o nodes is linked wi h hei deg ee o simi-
la i y. Acco ding o he Hidalgo (2021) , he simila i y 𝑁𝑘
𝑘
is defined as
ollows:
𝑀𝑖
=∑
𝑘
𝑀𝑖𝑘
(2)
𝑀𝑘
=∑
𝑖
𝑀𝑖𝑘
(3)
𝑁𝑘
𝑘
=∑
𝑖
𝑀𝑖𝑘
𝑀𝑖
𝑘
𝑀𝑖
𝑀𝑘
(4)
whe e, 𝑀𝑖
and 𝑀𝑘
ep esen he numbe o p oduc s p oduced by ha
p o ince, and he numbe o p o inces p oducing a gi en p oduc s, e-
spec i ely. 𝑁𝑘
𝑘
is he simila i y o he pai s o p oduc s k and
𝑘 . Fol-
lowing he same s a egy, an ag icul u al p o ince space defined which
nodes a e p o inces, and links be ween p o inces 𝑖 and
𝑖 a e weigh ed
by 𝑁𝑖
𝑖
:
𝑁𝑖
𝑖
=∑
𝑘
𝑀𝑖𝑘
𝑀
𝑖 𝑘
𝑀𝑖
𝑀𝑘
(5)
whe e
𝑁𝑖
𝑖
ep esen he simila i y o he pai s o p o inces 𝑖 and
𝑖 .
2.3.4. Co e-pe iphe y s uc u e de ec ion
The de ec ion o co e and pe iphe y wi hin he ne wo k can e eal
localized ne wo k s uc u es ( Mallia os e al., 2020 ; Gallaghe e al.,
2021 ). Va ious me hods and algo i hms ha e been employed o
ex ac ne wo k’s co e and pe iphe y, such as s a is ical in e ence
( Kojaku and Masuda, 2017 ; Peixo o, 2019 ), spec al decomposi ion
( Cucu ingu e al., 2016 ; Tudisco and Higham, 2019 ), diffusion mapping
( Rossa e al., 2013 ) o mo i coun ing ( Ma e al., 2018 ), geodesic ac-
ing ( Cucu ingu e al., 2016 ), and ich clubs ( Ma and Mond agón, 2015 ).
The e is also a new app oach o de ec he co e-pe iphe y s uc u e o
weigh ed and undi ec ed ne wo k. Unlike o he me hods de e mining
whe he a node belongs o co e wi h “yes ”o “no ”, his me hod uses a
sco ing sys em anging be ween 0 and 1 o measu e he deg ee o co e.
A highe sco e indica es close p oximi y o he co e. No ably, a node
is mo e likely in he ne wo k’s co e posi ion which no only densely
connec s among o he co e nodes and has high s eng h, bu also has
links wi h pe iphe al nodes ( Rombach e al., 2017 ). Mo eo e , co e-
pe iphe y s uc u e can also be nes ed wi h ne wo ks’ communi y s uc-
u e ( Lesko ec e al., 2009 ; Yang and Lesko ec, 2012 ).
2.3.5. Communi y de ec ion
Communi y de ec ion may be a help ul ool o analyzing he diffe -
ences o p oduc s g oup and p o inces g oup. The adi ional s a egy o
communi y de ec ion ocuses on op imizing modula i y ( Es ahlani e al.,
2021 ), wi h example including Guime a’s modula i y ( Guime a e al.,
2007 ) and Ba be ’s modula i y ( Ba be , 2007 ). QuanBiMo is he fi s
algo i hm o maximize weigh ed modula i y in bipa i e ne wo k. How-
e e , i s sensi i i y o specific inpu a iables ha migh no be ac-
cessible can lead o de ia ions om expec ed ou comes ( Do mann and
S auss, 2014 ). Fu he mo e, DIRTLPAwb + , an ex ension o QuanBiMo,
pe o ms well on small ne wo ks, which can sea ch he op imal mod-
ula i y in whole possible space ( Becke , 2016 ). In his s udy, he
DIRTLPAwb + algo i hm was used o maximize modula i y sco es in
he ne wo ks. DIRTLPAwb + has mo e meaning ul inpu pa ame e s and
exhibi s enhanced pe o mance. I s implemen a ion is also mo e s able
han ha o QuanBiMo on he es ( Becke , 2016 ).
3. Resul s
3.1. O e iew o he “p o ince-p oduc ” ag icul u al p oduc ion space
The “p o ince-p oduc ” ag icul u al p oduc ion space has been buil
o exhibi how p o inces g ow ag icul u al p oduc s om 1995 o 2019.
Fig. 2 shows he case o 2019 and i clea ly demons a e ha he ne -
wo k’s cen al nodes include maize, whea , ice, po k, suga cane, pome
ui , and melon, indica ing he c i ical ole o hese p oduc s in he ood
p oduc ion sys em. Addi ionally, his ne wo k effec i ely depic s he ge-
og aphical dis ibu ion o ag icul u al p oduc s ac oss China. Maize p e-
dominan ly links wi h p o inces in he no he China, such as Hebei,
Henan, Heilongjiang, Liaoning, Jilin, Shanxi, and he Inne Mongolia
Au onomous Region. Meanwhile, ice exhibi s s ong connec ions wi h
sou he n p o inces like Anhui, Jiangsu, Jiangxi, Hunan, Hubei, Sichuan,
and Guangdong. Simila ly, whea p ima ily connec s wi h Henan, Shan-
dong, Hebei, Shaanxi, and Anhui. These findings a e well consis en wi h
p e ious esea ches conduc ed by Ye e al. (2020) , Bai e al. (2021) ,
L. Zhang e al. (2022) . Suga cane, a ypical opical and sub opical
p oduc , is logically plan ed in Hainan, Guangdong, Guangxi, and Yun-
nan. Blue p o incial nodes ep esen adi ional ag icul u al p oduc ion
a eas ha ha e di e se p oduc s and highe ou pu , which include Shan-
dong, Henan, Hebei, Heilongjiang, Shanxi, Jilin, and Liaoning. In con-
as , p o inces like Qinghai, Hainan and Xizang Au onomous Region
occupy a mo e pe iphe al posi ion, in keeping wi h he ac ual s a e o
ag icul u al p oduc ion ( Guo e al., 2020 ). The da ke -colo ed links con-
nec ing maize, ice, whea , melon, and po k o hei p o inces signi y
ha hese p oduc s ha e conside able impo ance in hei p o inces.
P oduc s loca ed in he ou e ing exhibi spa se and weake links, likely
due o hei equi emen o specialized p oduc ion condi ions, o en
connec o specific p o inces wi hin a limi ed geog aphic a ea. Ou sub-
sequen in es iga ion will del e in o he co e-pe iphe y s uc u e and
communi y s uc u e wi hin bo h he ag icul u al p oduc space and he
p o ince space.
3.2. Analysis o he co e-pe iphe y s uc u e
3.2.1. Co e-pe iphe y s uc u e o ag icul u al p oduc space
Fig. 3 (a) shows he bubble ma ix diag am ha illus a es he co e
sco e anking o each p oduc om 1995 o 2019. Among all p oduc s,
maize, whea and ice consis en ly main ain he highes posi ion, wi h
maize g adually su passing whea and ice, which he e is no doub ha
124
S. Yang, Z. Li, J. Zhou e al. Geog aphy and Sus ainabili y 5 (2024) 121–134
Fig. 2. Ag icul u al p oduc ion space in 2019.
No e: Nodes a e wo disjoin se s ha ep esen
p oduc s and p o inces. P oduc nodes a e o -
ange, and p o ince nodes a e blue. The size o
he nodes ep esen s hei impo ance in he
ne wo k. Links a e colo coded wi h hei RCA
alue. Links a e always be ween he node o
he p oduc se and a node o he p o ince se ,
and he e is no a link be ween wo nodes in
he same se . The da ke colo o links indica es
he s onge capabili ies o his p o ince o p o-
duce a specific p oduc .
hese h ee p oduc s domina e ag icul u al p oduc ion in China. I is
also wo h no ing ha melon (wa e melon and muskmelon), po k, and
pome ui (apple, pea , jujube, and pe simmon) a e in he ou h, fi h
and six h places, espec i ely. The co e sco e o po k emained ela-
i ely s able om 1995 o 2018 bu wi nessed a sha p decline in 2019
due o he impac o he A ican swine e e and “Enhanced Pig Cycle ”
(EPC). This e en led o a significan educ ion in domes ic po k p o-
duc ion, plumme ing om 54.075 million ons o 42.553 million ons
( Song e al., 2022 ). Melon, as one o he mos consumed ood in China,
expe ienced an expansion in i s plan ing a ea om 1,013.52 ha in 1995
o 1,894.05 ha, ep esen ing app oxima ing 40% o he global yield om
China ( Luo e al., 2018 ). Pome ui s, including apple, pea , jujube, pe -
simmon, se e as necessa y ood in ou daily die . Thei plan ing a eas
in China a e concen a ed in he no he n p o inces, main aining s able
g ow h in hei ou pu ( Wang e al., 2020 ; Qi e al., 2023 ). Fo exam-
ple, cul i a ion a ea o apple has expanded om semi humid (i iga ed)
o semi-a id ( ain- ed) a ea ( Zhang e al., 2023 ). Fibe c ops (flaxseed,
apeseed, sesame, and sunflowe ) and g ape ha e expe ienced a consis-
en inc ease in hei anking since 1999, while milk began o ise in
2004. On he con a y, oil c ops ( empo a y oilseed c op, peanu , and
soybean) displayed a declining end un il 2019. O he p oduc s wi h
in isible colo such as ba ley and mu on a e usually away om he
co e a ea in he ne wo k excep some ex eme yea s. Fo example, ci us
achie ed ela i ely highe ankings in 2015, 2018, and 2019 compa ed
o o he yea s, mainly owing o a apid su ge in ci us ou pu du ing
hese yea s ( om 245.25 million ons o 274.01 million ons).
Fig. 3 (b) shows he box diag am p esen ing he co e sco e o each
p oduc om 1995 o 2019, offe ing a clea e unde s anding o he
o e all diffe ences among ag icul u al p oduc s. The mean co e sco es
o maize, whea , and ice a e 0.9443, 0.9187, and 0.9015, espec-
i ely. Following hese, melon (0.8428), po k (0.7944), and pome ui
(0.67769) hold ela i ely high a e age co e sco es. These p oduc s,
as highligh ed by Wang e al. (2019) , a e undamen al oods in he
Chinese die and a e cul i a ed in ela i ely high-p oducing egions,
which domina e Chinese ood sys em. In he second o de , wi h sligh ly
lowe co e sco e compa ed o he fi s o de , a e ag icul u al p oduc s
such as po a o, empo a y oilseed c op, milk, peanu , fibe c op, g ape,
and soybean. These p oduc s a e s ill subs an ially impo an in ou
daily die o li ing ( Liu e al., 2021 ). P oduc s limi ed o specific e-
gions wi h specific condi ions a e posi ioned on he ne wo k’s pe iph-
e y, wi h ela i ely lowe sco es. Fo ins ance, suga bee , mille , adzuki,
mung, and so ghum a e mos ly cul i a ed in no he n China ( Lin e al.,
2023 ). Ci us, i i an c op, eshwa e p oduc , seawa e p oduc , op-
ical and sub opical ui a e mainly in sou h China ( Duan e al., 2020 ;
Yan e al., 2020 ). Tobacco has widesp ead dis ibu ion in sou he n hilly
a ea ( Liu e al., 2018 ). Bee , mu on, and honey a e concen a ed o
p oduce in no hwes China ( H. Zhang e al., 2022 ).
3.2.2. Co e-pe iphe y s uc u e o ag icul u al p o ince space
As depic ed in Fig. 4 (a), p o inces such as Shanxi, Jilin, Hebei,
Inne Mongolia Au onomous Region, Liaoning, Heilongjiang, Henan,
and Shandong occupy ela i ely highe ankings. These p o inces ex-
hibi high s eng h in main ag icul u al p oduc s, pa icula ly in maize,
whea , ice and pome ui . Addi ionally, hese p o inces a e well-
connec ed o each o he o e en o some pe iphe al p o inces. Fo
example, Shanxi no only sha e connec ion wi h Henan, Hebei, Shan-
dong, Inne Mongolia Au onomous Region, bu also demons a e s ong
links wi h Gansu, Shaanxi due o hei mu ual cul i a ion o specific
ag icul u al p oduc s. The main ag icul u al p oduc s o high- anking
p o inces ha e been ou lined in Table S1. Liaoning, Tianjin and Hei-
longjiang ha e demons a ed a g adual upwa d end in hei ankings.
Con e sely, Shaanxi, Ningxia, and Gansu show a dec easing end. An-
hui and Xinjiang Uygu Au onomous Region exhibi no iceable fluc-
ua ions. While p o inces like Xizang Au onomous Region and Qing-
hai, si ua ed on he Qingzang Pla eau wi h limi ed cul i a ed land and
lowe empe a u e, main ain lowe ankings. Hainan, Guangdong, and
125
S. Yang, Z. Li, J. Zhou e al. Geog aphy and Sus ainabili y 5 (2024) 121–134
Fig. 3. Co e-pe iphe y s uc u e in ag icul u al
p oduc space. (a) The bubble ma ix diag am o
co e sco e ankings o each p oduc ; (b) Box dia-
g am o co e sco es om 1995 o 2019 o each
p oduc .
No e: Do s wi h da ke colo s and wide sizes ha e
highe co e sco es. The line in he box ep esen s
he median alue, and he solid do indica es he
mean alue, which is linked by a black solid line.
The as e isk deno es he ou lie s. The wo ends o
he box a e he fi s and hi d qua iles.
126
S. Yang, Z. Li, J. Zhou e al. Geog aphy and Sus ainabili y 5 (2024) 121–134
Fig. 4. Co e-pe iphe y s uc u e in ag icul u al p o ince space. (a) The bubble ma ix diag am o co e sco e ankings o each p o ince; (b) Box diag am o co e
sco es om 1995 o 2019 o each p o ince.
No e: Do s wi h da ke colo s and wide sizes ha e highe co e sco es. The line in he box ep esen s he median alue, and he solid do indica es he mean alue,
which is linked by a black solid line. The as e isk deno es he ou lie s. The wo ends o he box a e he fi s and hi d qua iles.
127
S. Yang, Z. Li, J. Zhou e al. Geog aphy and Sus ainabili y 5 (2024) 121–134
Fig. 5. Communi y p esen ed in ag icul u al p oduc space om 1995 o 2019.
Guangxi which p ima ily ocus on opical and sub opical p oduc s,
ha e ewe connec ions o p oduc s o o he p o inces. The e o e, hey
a e a he lowe ankings o he ne wo k. The box diag am in Fig. 4 (b)
shows a clea e insigh o each p o ince’s co e sco e om 1995 o 2019.
Shanxi, Jilin, Hebei, and Inne Mongolia a e in he high posi ions, wi h
a e age sco es o e 0.8. Following closely a e Liaoning, Heilongjiang,
Henan, and Shandong, wi h sco es fluc ua ing be ween 0.1 o 1. Con-
e sely, o he p o inces ha e sco es ha a e app oxima ely below 0.3,
and in some case, nea ly 0. Despi e hese ela i ely lowe sco es, hese
p o inces s ill play i al oles in some specific p oduc s. Fo example,
Fujian, Guizhou, and Yunnan a e significan o ea p oduc ion; Qinghai,
Xinjiang Uygu Au onomous Region and Xizang Au onomous Region a e
impo an o mu on p oduc ion ( Yan e al., 2021 ).
3.3. Changes in he ela edness among ag icul u al p oduc s
The e ol ing ela edness among diffe en ag icul u al p oduc s a e
in es iga ed by communi y de ec ion and he ou comes a e p esen ed
wi hin he ag icul u al p oduc space. These communi ies ga he p od-
uc s ha sha e simila p oduc ion condi ions o capabili ies. As shown
in Fig. 5 , he node size co esponds o he cen ali y o each p oduc ,
de e mined by i s co e sco e. The modula i y anges om 0.4749 o
0.5957, which can suppo he possibili y o p esen ne wo k subs uc-
u es. A o al o h ee o ou communi ies a e de ec ed om 1995 o
2019. F om 1995 o 2002 and om 2003 o 2011, he ne wo ks a e di-
ided in o h ee communi ies. The pu ple communi y mainly includes
maize, whea , melon, pome ui soybean, empo a y oilseed c ops and
some pe iphe al p oduc s ha a e mainly sui able in no h o China.
The o ange communi y is domina ed by po k and ice, bo h o which
equi e mo e hea and wa e esou ces o hei p oduc ion compa ed
o he p oduc s in he pu ple communi y. The g een communi y includes
suga cane and opical and sub opical p oduc s. The communi y com-
posi ions in 2003 esemble hose in 1995, wi h he excep ion o soybean,
which has mo ed o he o ange communi y wi h po k and ice. Since
2012, ou communi ies a e de ec ed, and his configu a ion emains
ela i ely s able un il 2019. No ably, maize and soybean a e obse ed
wi hin he same communi y a e 2012, while o he communi ies exhibi
minimal changes o e hese yea s.
The Sankey g aph p esen ed in Fig. 6 shows he dynamic changes
in communi y componen s o e ime, wi h co e p oduc s highligh ed
in bold. The co e p oduc s wi hin each communi y emain almos un-
changed om 1995 o 2019, sugges ing a ela i ely s able ne wo k s uc-
u e in his pe iod. This finding implies p oduc s uc u e in China’s
ag icul u al p oduc ion demons a es a ema kable s abili y ( Huang and
Tian, 2019 ; Guo e al., 2020 ), in line wi h he ac ual si ua ion o ag i-
cul u al p oduc ion in China, because p oduc ion condi ions such as
soil, c opland, clima e, cul i a ion habi s a e long- e m elemen s and in-
equen changes. Addi ional no ewo hy finding is he ansi ion om
h ee communi ies du ing 1995 o 2003 o ou communi ies a e 2012.
Mo e impo an ly, soybean is sepa a ed om he ice and po k g oups
o he communi y wi h maize, milk, pome ui , g ape, and fibe c ops
un il 2019. Fu he mo e, i is also impo an o poin ou ha he po-
si ion o ba ley equen ly changes, influenced by fluc ua ions in ou -
pu wi hin i s p ima y p oduc ion egions. Ba ley ou pu declines om
0.805 million ons o 0.239 million ons in Jiangsu, 0.123 million ons
o 0.002 million ons in Zhejiang, 0.369 million ons o 0.109 million
ons in Henan, 0.125 million ons o 0.0254 million ons in Hubei, and
0.546 million ons o 0.133 million ons in Gansu. Howe e , an inc ease
128
