Research Articles

Dynamic evolution and the mechanism of modern gully agriculture regional function in the Loess Plateau

  • QU Lulu , 1 ,
  • LI Yurui 2, 3 ,
  • WANG Yongsheng 2, 3 ,
  • DONG Shijie 2 ,
  • WEN Qi , 4, *
  • 1. School of Public Policy and Administration, Chongqing University, Chongqing 400044, China
  • 2. Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
  • 3. Key Laboratory of Regional Sustainable Development Modeling, CAS, Beijing 100101, China
  • 4. School of Geography and Planning, Ningxia University, Yinchuan 750021, China
*Wen Qi (1979-), PhD and Professor, E-mail:

Qu Lulu (1991-), PhD and Lecturer, E-mail:

Received date: 2022-01-26

  Accepted date: 2022-09-06

  Online published: 2022-11-25

Supported by

National Natural Science Foundation of China(42101202)

National Natural Science Foundation of China(42061037)

National Natural Science Foundation of China and National Science Foundation of the United States Sustainable Regional System Cooperation Research Project(T221101034)

China Postdoctoral Science Foundation Project(2022M710015)

Fundamental Research Funds for the Central Universities(2022CDJSKJC29)


The agricultural regional type and function are the key theoretical issues in agricultural geography research. Gully agriculture in the Loess Plateau is a new regional type of agricultural system with the coupling development of the modern gully human-earth relationship. The study of its functional changes is of great practical significance for food security, rural revitalization and sustainable development of regional agriculture in the region of interest. This paper analyses the multifunctional change of gully agriculture in the Loess Plateau and its dynamic mechanism by using large-scale remote sensing data, topographic relief amplitude model, and spatial econometric model to understand internal implications for evolution differentiation at the basin level. The results show that: (1) The spatial concentration of production and supply function of agricultural products (APF) in the gully of the Loess Plateau gully is high, while the ecological conservation and maintenance function (ECF), employment and social security function (ESF), cultural heritage and leisure function (CRF) are relatively low. The four functions’ spatial distribution has revealed an apparent regularity. (2) APF has been significantly enhanced, which is mainly distributed in point clusters and strips in the farming and pastoral areas in northern Shaanxi to the Yanhe river basin. The high-value areas of ESF are clustered around the urbanized metropolitan circles and urban-rural staggered areas along the Great Wall. ECF is concentrated in areas with significant natural endowments and excellent ecological conditions. CRF is significantly higher in the municipal districts and the surrounding regional central cities. (3) There are noticeable differences in the gully agriculture regional function (GARF) evolution process due to the geographical environment and socio-economic development stages. In this regard, natural factors have tremendously affected APF, ESF, and ECF, while socio-economic factors greatly differ in the four functions. There are still differences in the driving mechanisms of modern gully agriculture evolution types; hence many critical policies in the Loess Plateau can directly affect the function evolution paths. The dynamic evolution of GARF can reflect the general law of rural human-earth system transition in gully areas, thereby providing policy ideas for high-quality development of agriculture in the Loess Plateau.

Cite this article

QU Lulu , LI Yurui , WANG Yongsheng , DONG Shijie , WEN Qi . Dynamic evolution and the mechanism of modern gully agriculture regional function in the Loess Plateau[J]. Journal of Geographical Sciences, 2022 , 32(11) : 2229 -2250 . DOI: 10.1007/s11442-022-2045-y

1 Introduction

Rapid urbanization and agricultural modernization processes have brought about dramatic changes in the regional human-earth relationship and diversification of agricultural regional types (Rudel et al., 2009; Liu and Li, 2017). In the late 1980s and early 1990s, Japan first recognized the diversification and versatility of agriculture, and consequently, it was promoted globally (Yoshid, 2001). Scholars gradually began to pay attention to the multifunctional research of agriculture, pointing out that fully excavating the intangible value of agricultural production activities and combining agriculture with practically adequate resources such as agricultural products, natural conditions, and cultural concepts could expand agricultural functions and improve comprehensive benefits (Benoit et al., 2014; Sun et al., 2018). In the same context, China’s agriculture has entered a new development stage of rapid transformation and upgrading, where the differentiation of agricultural regional functions has become increasingly prominent. At present, the growing need for a better lifestyle puts forward new requirements for agricultural regions to provide more products and better services. Nevertheless, under the long-term strategic guidance of “giving priority to urban development”, the regional value dislocation caused by agricultural diseases has restrained the sustainable development of agriculture (Long et al., 2009; Liu et al., 2018). The remaining question on how to build a reasonable modern agricultural regional structure while promoting the multifunctional coordination is crucial challenging in order to solve these problems (Lu et al., 2019).
Modern agriculture is not solely an industrial form, yet it can also exist in various new forms of business such as characteristic industries, ecological industries, tourism and leisure industries, and cultural industries. It has many functions, such as economic, cultural, ecological, and employment security (Wu, 2007). An agricultural region is not only an area that supplies agricultural products for urban and rural areas while providing security for rural residents but also an ecological security-bearing area with a prosperous reserve of traditional agricultural culture. In the new era, the laws of agricultural regional differentiation and modern agricultural zoning aim to coordinate and arrange the direction of regional agricultural development, promote the integrated development of agriculture and secondary and tertiary industries, build advantageous agricultural zones with regional characteristics, and create complementary and competitive advantages (Liu et al., 2022). The national agricultural regional division of labor distribution system is an inevitable obligation to solve China’s “agriculture, rural areas, and rural people” problems while tapping new kinetic energy for rural development and opening up new fields for the development of agricultural geography.
At present, the research on agricultural regional function nationally and overseas mainly focuses on the research on agricultural multifunction, especially the research on agricultural multifunction management mode (Vanslembrouck et al., 2005), regional types and optimization strategies (Randall, 2002; Qu et al., 2020), multifunction characteristics and influencing factors (Rockström, 2017; Fang et al., 2019), emergence theoretical basis and origin of agricultural multifunction (Huang et al., 2015; Lu et al., 2019), its associated rural regional multifunction, landscape multifunction, and cultivated land multifunction (Liu et al., 2011; Peng et al., 2015; Wang et al., 2018; Yang et al., 2018). Overall, numerous achievements have been reached in terms of studying regional function with agricultural regions as the research topic. Based on theoretical analysis, this paper attempts to deeply grasp the differentiation law and evolution process of agricultural regional function in the of rural-urban integration process, investigate the factors affecting its change, and then reveal its dynamic mechanism to provide decision-making reference toward building a modern agricultural industry system with distinctive regional characteristics and complementary benefits. As different regional function types, calculation methods and data sources often lead to different research deductions, the comparability of relevant research is insufficient, hindering the in-depth development of this research field. Therefore, an explicit and concise theoretical framework, entirely considering the development stage and regional background characteristics of the research area, can overcome the issues mentioned above to a certain extent and improve the interpretation ability.
As the Loess Plateau is ecologically vulnerable, the contradiction between human and environment is exceptionally prominent, which is the most typical geomorphic unit and the embodiment area of ecological and economic contradictions in China (Liu et al., 2014; Wang et al., 2015). The land use and ecological governance of the Loess Plateau have attracted much attention (Fu et al., 2014). Moreover, it is a crucial area for consolidating the achievements of poverty reduction and connecting rural revitalization toward sustainability (Liu et al., 2018; Wen et al., 2019). The unique geographical characteristics have shaped the highly characteristic rural human-earth system (Li et al., 2019; Qu et al., 2021). Among them, the hilly and gully region of the Loess Plateau has various types of gullies, i.e., up, down, vertical, and horizontal. In the process of economic and social development and urbanization, the loess hilly and gully region suffered from double disturbances of natural ecology and human activities (Liu et al., 2020). In addition to the multifunctional interaction of land circulation, planting mode adjustment, characteristic agricultural production, recreation and health preservation, farming experience, and so on, the spatial difference in the path of regional functional layout evolution is noticeable. The landform pattern of millions of gullies in the loess hilly and gully region reveals that the ditch dam is the essence of the land resources in the area. Overall, it shows a transformation trend from traditional agriculture to modern agriculture, which can be summarized into three evolution stages: slope traditional agriculture stage, slope vegetation construction stage, and gully modern agriculture development stage. The experienced process has been caused by wide planting yet low agricultural production income for sustainable conservation of production practice.
Existing research findings have shown that slope land conversion and intensification coexist in the Loess Plateau (Li et al., 2019; Feng et al., 2021), agricultural and rural transformation (Cao et al., 2019; Lu et al., 2020), and significant changes have taken place in the regional landscape pattern (Fu et al., 2014). The implementation of the gully control and land reclamation projects in the recent decade has promoted the systematic transformation of the agricultural, industrial, and management structures. Such a transformation situation has created favorable conditions for the development of modern agriculture, focusing on gully in the hilly and gully region of the Loess Plateau, and accelerated the transformation of the gully human-earth relationship and gully agricultural land use mode, revealing new characteristics and issues. The main objectives of this study are the following: (1) identification and classification of the function evolution characteristics of the modern gully agriculture in the Loess Plateau; (2) presentation of temporal and spatial evolution pattern of GARF; and (3) exploration of the influencing factors of multifunctional differentiation of gully agriculture in the Loess Plateau. This paper analyzes the evolution law and driving mechanism of GARF in the Loess Plateau from the perspective of human-earth system science to provide practical support for the transformation and high-quality development of agriculture in the Loess Plateau region.

2 Materials and methods

2.1 Study area

The Loess Plateau occupies a total area of approximately 640,000 km2, which belongs to a semi-arid region. The region is crisscrossed with gullies abundance and severe water and soil quality degradation. The annual mean precipitation rate ranges between 300 mm and 600 mm. The amount of water resources that can be used for agriculture are relatively scarce. In addition, the region’s ecological environment is vulnerable, with frequent occurrence of natural disasters occur frequently, threatening the region’s sustainable development. Since 1999, the Chinese government has begun implementing GGP in the middle and upper reaches of the Yangtze River and the Yellow River, remarkably improving vegetation density and vigor. The agricultural transformation in the mountainous and hilly regions of the Loess Plateau has been dynamic, especially in the hilly and gully region. The agricultural production has changed from hillside to gully and surface color from yellow to green. The hilly and gully region of the Loess Plateau covers a total area of approximately 140000 km2, including ten cities in Shaanxi, Shanxi and Inner Mongolia, including Yan’an, Yulin, Taiyuan, Datong, Shuozhou, Yizhou, Linfen, Luliang, Hohhot and Ordos, including 64 counties (Figure 1).
Figure 1 Geographical location of the study area (Loess Plateau, China)
With many interlaced beams and ravines, the area is considered the most typical geomorphological unit in the Loess Plateau. More than half of the land in this area has a slope of greater than 15°, with a gully density ranging between 2 and 7 km/km2. Most inclined farmland is dominated by surface erosion, and the gully erosion is dominated by gullies, with an aggravating trend. Under the background of high-quality development and urban-rural integration in the Yellow River Basin, it is typical and demonstrative to carry out research on the characteristics and mechanism of modern gully-type agricultural regional function in the Loess Plateau (Qu et al., 2021).

2.2 Gully agriculture identification and classification

Gully agriculture, also known as “modern dam agriculture”, is a new form of agricultural production land of different types (Liu et al., 2020). The identification rule of gully agriculture is as follows: First, the latest farmland range is extracted as the background data of gully farmland, and then the pixel (T) of other years is obtained based on the change in the range of the farmland. Besides, the suspected gully agriculture farmland range with the spe-cific research period is obtained based on the above judgment. Secondly, based on the range of gully agriculture in the year (t-1) and the overlapping part of the suspected gully agriculture land use in the year (t-2), and the suspected gully agriculture land use in the year (t-1) is extracted again. Based on the above method, the range of gully agriculture land in a continuous sequence of prescribed years is obtained.

2.3 Functional accounting

The regional function classification of gully agriculture in the Loess Plateau follows the principles and standards of regional function classification of gully agriculture, and measures the evaluation values of four functional layers of gully agriculture on a regional scale from the four aspects of agricultural production, ecology, life and cultural leisure (Table 1).
Table 1 Evaluation index system of GARF in the Loess Plateau
Functional layer Functional index Calculation method Unit
Production and
supply function
of agricultural products (APF)
D1 Grain yield per unit area Total grain yield / grain planting area ton/ha
D2 Agricultural output value per unit area Total agricultural output value / total area of agricultural land, reflecting agricultural production efficiency ten thousand yuan
D3 Power of agricultural machinery per unit area It reflects the mechanization level of gully agricultural production kWh/ha
D4 Per capita non-food
(Oil + cotton + vegetables + medicinal materials + fruit output) / total population of the region kg/person
D5 Per capita share of
livestock products
(Meat + egg + milk production) / total population of the region kg/person
and maintenance function (ECF)
D6 Total value of ecological services Ecological service value per unit area of a certain type
of land * area of this type of land
ten thousand yuan
D7 Amount of chemical fertilizer per unit area The net amount of agricultural chemical fertilizer /
cultivated land area at the end of the year
D8 Pesticide consumption per unit area Pesticide use / cultivated land area at the end of the year, reflecting the negative impact of pesticides in
agricultural production
D9 Forest cover rate Forest coverage / total area of regional land %
Cultural heritage
and leisure
function (CRF)
D10 Leisure agricultural income Operating income of recreational agriculture such as sightseeing farms and picking gardens ten thousand yuan
D11 Number of agricultural cultural heritage scenic spots Number of main agricultural cultural heritages in towns Number
D12 Gully landscape value A certain type of land area entertainment cultural value * this type of agricultural land area ten thousand yuan
D13 Urban population above the county level in 100 kilometers around the township The point distance module tool in ArcGIS, and the
calculation and statistics functions of excel such as Vlookup
D14 Annual income per capita in cities above the county level in 100 kilometers around the township The point distance module tool in ArcGIS, and the
calculation and statistics functions of excel such as Vlookup
ten thousand yuan/person
and social
D15 The proportion of agricultural output value in GDP Gross agricultural output value / GDP, reflecting the income guarantee of farmers %
D16 Per capita agricultural output value Gross agricultural output value / agricultural labor force ten thousand yuan/person
D17 Proportion of
agricultural employees
Number of agricultural employees / total rural labor force, reflecting the service level of agricultural
D18 Rural population It reflects the carrying level of agricultural population Person
D19 Farmers’ per capita agricultural income It reflects the living security level provided by
agriculture for farmers
ten thousand yuan
(1) The agricultural product production and supply function layer mainly considers factors such as agricultural production level, agricultural land abundance, varieties of agricultural products, and output scale. Five indicators are selected: agricultural output value per unit area, grain yield per unit area, power of agricultural machinery per unit area, per capita non-food cultivation, and per capita share of livestock products.
(2) The ecological conservation and maintenance function layer is comprehensively evaluated by selecting positive indicators such as the total value of regional ecological services and forest coverage, as well as negative indicators such as the amount of chemical fertilizer and pesticide per unit area of cultivated land. Among them, the total value of regional ecological services is calculated according to the method proposed by Costanza (Costanza et al., 1997), using the theoretical value table of land average ecological function in Shaanxi Province and the relevant research of farmland ecosystem in different provinces of China (Hu et al., 2010; Xie et al., 2001), the value equivalent of distinguishing counties from hills and gullies in the Loess Plateau is calculated. Finally, combined with land use data at the township-level administrative unit scale, the total ecological service value of each township scale is obtained.
(3) The measurement of employment and social security function layer mainly considers the agricultural bearing population, employment and survival security, which is affected by the number of rural labor force, industrial output value and other factors. Five indicators such as the proportion of agricultural employment population, the number of rural population, and the total agricultural output value per capita are selected to perform the functional characterization.
(4) The cultural inheritance and leisure function layer is prominent in areas with specific production methods or regional culture. This function is affected by factors such as urbanization, traffic accessibility, local characteristic cultural endowments, and government financial investment. In this paper, the value of this function is calculated by the income of leisure agriculture, the number of agricultural cultural heritage scenic spots, the population of cities at or above the county level within 100 kilometers around the township resident in the gully basin and the per capita annual income.

2.4 Determining the weights of functional quantitative diagnostic indicators

The hierarchical structure and overall function of the gully agricultural area are analyzed based on the theory of complex watershed systems. Under the framework of the complex watershed system of gully, through the composition of system elements and the relationship between subsystems, this paper unveils the regional multifunction of gully agriculture from a systematic point of view. It constructs a multi-level contribution classification system and quantitative diagnosis index system of gully agricultural regional function from top to bottom, covering the “target layer-criterion layer-element layer”. Therefore, in the layer-by-layer function value calculation and aggregation process, the “one-to-one” regional function value calculation aggregation mode is selected. In this mode, the indicators are often pushed up layer by layer from bottom to top to contribute to the functional characteristic values of adjacent higher levels, as shown in Figure 2.
Figure 2 Multi-level correspondence model constructed by the index system
The indicators of GARF have different effects on the regional type function of gully, while there are also differences in the quantification of function. The index system of the gully agricultural regional function constructed in this paper considers the distribution of the strengths and weakness of redional agricultural regional multi-functionality, resulting in a “one-to-many” interaction between the function value and the index value as illustrated in Figure 2. The methods of index contribution weight mainly include subjective weighting like AHP and objective weighting like neural networks (Liu et al., 2019; Zhang et al., 2019), and the experimental data are required to correspond to the sample index. If only a single objective weighting method is selected, it is challenging to quantify the interaction between indicators. Therefore, this paper further introduces the AHP method of subjective weighting and comprehensively determines the contribution of each indicator to different functional levels through a combination of “subjective weighting” and “objective weighting”, as presented in Table 2.
Table 2 Evaluation index system of GARF in the Loess Plateau
Functional layer Weight Index layer Single sort weight value E Total ranking weight value F
D1 0.221 0.062
D2 0.213 0.059
APF 0.279 D3 0.132 0.037
D4 0.158 0.044
D5 0.185 0.052
D6 0.251 0.066
ECF 0.262 D7 0.234 0.061
D8 0.132 0.035
D9 0.382 0.100
D10 0.186 0.041
D11 0.272 0.060
CRF 0.222 D12 0.267 0.059
D13 0.133 0.030
D14 0.142 0.032
D15 0.126 0.030
D16 0.086 0.020
ESF 0.237 D17 0.158 0.037
D18 0.225 0.053
D19 0.405 0.096

2.5 Functional evolution model construction

2.5.1 Functional unit division

The research on the evolution characteristics of GARF involves the selection of the regional unit of the entity object. Theoretically, the regional unit should be a production entity, considering the availability of statistical data, the independence of the gully basin system unit, and the integrity of essential administrative boundaries. This study takes ten prefecture-level administrative units in the hilly and gully region of the Loess Plateau as the primary regional units of regional-scale research, including suburbs in 10 prefecture-level cities, 61 county (district) level administrative units, and 697 township-level administrative units. This study selects the township administrative unit as the grass-roots regional unit, which fits well with the gully basin production unit and fully reflects the type characteristics.

2.5.2 Functional statistics and spatial functional classification model

In order to grasp the differentiation characteristics and change laws of the four functional spaces as a whole, the spatial Gini coefficient representing spatial concentration is introduced to measure quantitatively (Lv et al., 2016). Its calculation model is as follows:
\[{{G}_{i}}=\frac{\sum\limits_{j=1}^{n}{\left| {{D}_{ij}}-\frac{{{D}_{i}}}{n} \right|}}{\sum\limits_{j=1}^{n}{{{D}_{ij}}+(n-2)\frac{{{D}_{i}}}{n}}}\text{ }i\le \text{4}\]
where Gi is the Gini coefficient of the function i; Dij is the functional value of item i in unit j (township); Di is the total value of i function of the whole region; n represents the number of evaluation units in the study area. In this paper, it refers to 697 township administrative units in the hilly and gully region of the Loess Plateau. The i value is the function type number. The spatial function classification adopts the systematic clustering method weighted by principal component distance, and the type structure of regional function is analyzed by comprehensive index weight. The specific steps are as follows:
Firstly, the index dimension of spatial function division is determined. By sorting out 19 functional indicators of 697 towns (watersheds) in the study area, they were placed in a unified 697×19 matrix with 697 rows and 19 columns.
\[X=\left[ \begin{align} & {{X}_{11}}\text{ }{{\text{X}}_{12}}\text{ }\cdots \text{ }\cdots \text{ }{{\text{X}}_{1\text{19}}} \\ & {{X}_{21}}\text{ }{{X}_{2\text{2}}}\text{ }\cdots \text{ }\cdots \text{ }{{\text{X}}_{2\text{19}}} \\ & \text{ }\cdots \text{ }\cdots \text{ }\cdots \text{ }\cdots \text{ }\cdots \\ & \text{ }\cdots \text{ }\cdots \text{ }\cdots \text{ }\cdots \text{ }\cdots \\ & {{X}_{6971}}\text{ }{{X}_{6972}}\text{ }\cdots \text{ }\cdots \text{ }{{\text{X}}_{697\text{19}}}\text{ } \\\end{align} \right]\]
Secondly, the relationship matrix of spatial function zoning is constructed. The spatial function establishes the correlation degree through the relationship matrix. The key factors are extracted by principal component analysis to reduce the dimension of the original index. Then the eigenvalues and eigenvectors of each correlation coefficient matrix are further calculated, and the contribution rate and cumulative contribution rate of each principal component factor are extracted, which are determined combined with the obtained factor load matrix.
For the index vector, let R=(rij) (i, j=1,2, …, p…, n) be the p-order principal component classification vector extracted by the n-dimensional vector, G1, G2, …, Gs be the s-order row vector of the township scale of the extracted research unit, and Rij be the j term principal component factor of the unit i (i = 1, 2, …, n; j = 1, 2, …, s). The corresponding eigenvalues and eigenweights corresponding to the principal component eigenvectors R1, R2, …, Rs are λ1, λ2, …, λs and β1, β2, …, βs, respectively:
Contribution rate βi: ${{\beta }_{i}}={{\lambda }_{i}}/\sum\limits_{k=1}^{p}{{{\lambda }_{k}}}\text{(}i\text{=1, 2,}...\text{, }p\text{)}$
Cumulative contribution rate βi: ${{{\beta }'}_{i}}=\sum\limits_{k=1}^{i}{{{\lambda }_{k}}}/\sum\limits_{k=1}^{p}{{{\lambda }_{k}}}\text{(}i\text{=1, 2,}...\text{, }p\text{)}$
The principal component sample distance for adding weights is:
\[{{d}_{ij}}=\sqrt{\sum\limits_{k=1}^{p}{{{\beta }_{k}}{{({{R}_{ik}}-{{R}_{jk}})}^{2}}}},\text{ (}i\text{=1, 2,}...,p;\text{ }j\text{=1, 2,}...,p;\text{ }k\text{=1, 2,}...,p)\]
where i and j are the evaluation units of functional zoning; k is the number of principal components extracted. Based on the weighted calculation of sample distance, 697 × 697 order weighting matrix Dij, and then evaluate the functional similarity of 697 township units:
\[{{D}_{ij}}=\left[ \begin{align} & 0\text{ }{{d}_{12}}\text{ }\cdots \text{ }\cdots \text{ }{{d}_{1\text{696 }}}\text{ }{{d}_{1\text{697}}} \\ & {{d}_{21}}\text{ }0\text{ }\cdots \text{ }\cdots \text{ }{{d}_{2\text{696}}}\text{ }{{d}_{2\text{697}}} \\ & \text{ }\cdots \text{ }\cdots \text{ }\cdots \text{ }\cdots \text{ }\cdots \text{ }\cdots \\ & \text{ }\cdots \text{ }\cdots \text{ }\cdots \text{ }\cdots \text{ }0\text{ }{{d}_{696\text{697}}} \\ & {{d}_{6971}}\text{ }{{d}_{6972}}\text{ }\cdots \text{ }\cdots \text{ }{{d}_{697\text{696}}}\text{ 0 } \\\end{align} \right]\]
For the systematic clustering of spatial function classification, the clustering analysis principle of weighted distance matrix is to first regard each research object unit as an independent individual, and then merge the category units Gi and Gj with similar distance between individuals into a new category unit Gu=(Gi, Gj), and so on. The calculation formula is as follows:
\[{{G}_{uv}}=\min \{{{G}_{iv}},{{G}_{jv}}\}\]
where u is the new category unit obtained by merging, and V is the category unit (township) to be merged. All category units are merged one by one, so as to obtain the clustering map of spatial functional units. Combined with expert opinions, the clustering results are qualitatively adjusted to finally realize the division of GARF in the Loess Plateau.
According to the contribution share of sub-functions within each type, following the vertical comparison of functional categories (total value of functions) and the horizontal sorting of functional sub-categories (sub-functions), the functional structure of the main categories of GARF is determined. The structural factor is:
where qij is the weight coefficient of j function of class i; Wij refers to multiple comprehensive weight coefficients of j functions of class i.

3 Regional function evolution and type identification of modern gully agriculture in the Loess Plateau

3.1 Descriptive statistics and spatial concentration analysis of the four functions

From 2000 to 2018, the mean values of APF, ECF and LFF in the hilly and gully region of the Loess Plateau increased to a certain extent, while the mean value of CRF slightly decreased, as illustrated in Table 3. The variation coefficient of ECF changed considerably, with a variation value of 0.085, whereas the other three functional values did not change significantly. The four functions’ standard deviation and spatial concentration index are quite low, with minimal change. Among them, the spatial concentration index of APF is relatively high, showing a growing tendency and a remarkable change. The functional gap between regions is gradually prominent and concentrated in key regions.
Table 3 Four functional values and concentration of gully agriculture in the Loess Plateau in 2000 and 2018
2000 2018 2000 2018 2000 2018 2000 2018
Maximum 0.633 0.675 0.530 0.509 0.848 0.848 0.693 0.730
Minimum 0.040 0.037 0.003 0.006 0.017 0.013 0.110 0.115
Average 0.219 0.233 0.219 0.256 0.210 0.206 0.348 0.374
Standard deviation 0.124 0.143 0.107 0.103 0.134 0.131 0.117 0.117
Coefficient of variation 0.568 0.613 0.489 0.404 0.638 0.634 0.338 0.312
Spatial concentration 0.228 0.252 0.193 0.160 0.247 0.244 0.137 0.131

3.2 Spatial and temporal evolution patterns of GARF at the basin level

3.2.1 Spatial and temporal evolution characteristics of supply function of agricultural products in gully agriculture

In 2000, the variation range of APF functional value in the hilly and gully region of the Loess Plateau varied between 0.031 and 0.616. The high-value distribution area (APF≥0.375) included 42 townships, which were distributed in clusters and strips in the Wubao area of northern Shaanxi, the agricultural and pastoral areas in Yanggao of Shanxi and the river valley in Shuozhou of Western Shanxi as shown in Figure 3. In 2018, the APF value ranged between 0.041 and 0.632, with an increasing high-value area in 66 townships, expanding to the Wubao Qingjian Jingbian area in Yulin Yanhe River Basin in northern Shaanxi.
Figure 3 Spatial pattern of agricultural product supply function in the Loess Plateau in 2000 and 2018

3.2.2 Spatial and temporal evolution characteristics of employment and social security function

In 2000, there were 53 townships units in the high-value (ESF≥0.529) in the hilly and gully region of the Loess Plateau, primarily distributed across Jingbian and Shenmu areas of Yulin in northern Shaanxi, Ordos agricultural and pastoral areas in southwest Mongolia and Hohhot metropolitan area, in a scattered cluster distribution pattern of minor points and large dispersion in strips. In 2018, the number of high-value areas increased to reach 94, and further gathered around the areas along the Great Wall in northern Shaanxi, Taiyuan, Datong Shuozhou and other urbanization metropolitan areas and urban-rural ecotones, as presented in Figure 4. By expanding the modern agricultural industry chain, the above-mentioned township basin units mainly focus on a combination of planting, animal husbandry, and eco-tourism and promote the integration of multi-functional industries such as processing, logistics, and tourism, and a new mode of modern agriculture and animal husbandry combined with regional comprehensive regional development. However, in the hilly and gully areas of the Jinxi river valley, the agricultural production conditions are limited, the employment opportunities provided by agriculture are scarce, and the farmers’ income guarantee ability is not robust. Comparing 2000 to 2018, the ESF in the hilly and gully areas of the Loess Plateau has seen an upward trend, where the ESF functions of the counties and municipal districts and their surrounding areas were pretty prominent. However, the Loess Plateau and mountainous and hilly areas in western Shanxi became the “low valley areas” of ESF.
Figure 4 Spatial pattern of employment and social security function in the Loess Plateau in 2000 and 2018

3.2.3 Spatial and temporal evolution characteristics of cultural inheritance and leisure function

The spatial agglomeration effect of CRF dominant areas in the hilly and gully region of the Loess Plateau is noticeable, and high-value areas are primarily distributed in Baota district, Ordos district, and its surroundings of Yan’an city, as presented in Figure 5. Regional central cities surround these areas. With the development of regional urbanization and community, urban modern agriculture in the suburbs of cities has risen, and the demand for rural leisure tourism has greatly expanded. Various rural cultural leisure tourism projects in the suburbs are developing rapidly. Combined agricultural tourism with agricultural business multifunctional industrial belt has begun to emerge, promoting the integrated development of primary, secondary and tertiary industries. Especially in the Baota district, where the urban area of Yan’an is located, the CRF advantage driven by red tourism is perspicuous. From 2000 to 2018, the CRF of all regions in the hilly and gully region of the Loess Plateau except the Qingshui River increased, and the number of watershed units in the original high-value region (CRF≥0.374) has also significantly grown from 79 in 2018. The number of township watershed units reached 93 in 2018, as well as other areas, but the value of leisure services was low, and the overall effect was not perspicuous.
Figure 5 Spatial pattern of cultural heritage and leisure function in the Loess Plateau in 2000 and 2018

3.2.4 Spatial and temporal evolution characteristics of ecological conservation and maintenance function

The ECF in the hilly and gully region of the Loess Plateau presents a gradient variation characteristic of the “Liangmao region of the Loess Plateau > Loess Plateau region > rolling mountain region > urban area”, as illustrated in Figure 6. The high-value areas (EPF≥0.363) between 2000 and 2018 include 101 and 112 townships, respectively, which are fundamentally concentrated in the Yulin-Shenmu area in northern Shaanxi, the Ganquan pagoda area in Yan’an in northern Shaanxi, the mountainous and hilly areas of Yizhou in Western Shanxi and other areas with more significant natural endowment and vigorous ecological conditions. The ECF of urban districts and their suburbs is weak due to the land occupied by the expansion of urban construction land expansion. However, ECF showed an overall upward trend from 2000 to 2018. The ECF of the Yan’an area in northern Shaanxi and Ordos area in Inner Mongolia increased significantly. In contrast, the ECF of urban areas, surrounding towns, and traditional agricultural areas decreased significantly, especially around the Shuozhou metropolitan area and the sand area along the Great Wall in northern Shaanxi.
Figure 6 Spatial pattern of ecological conservation and maintenance function in the Loess Plateau in 2000 and 2018

4 Analysis of influencing factors for GARF evolution in the Loess Plateau

4.1 Influence factor index selection

Based on the natural ecological background, socio-economic development characteristics and the availability of data in the hilly and gully region of the Loess Plateau, this paper discusses the formation mechanism of GARF from the aspects of nature, population, economy and society, technological progress, land use efficiency and location conditions.
(1) Natural factors affect the overall layout of agricultural production on a large scale, and usually have a significant impact on agricultural development. Topographic relief x1, altitude x2 (m), accumulated temperature x3 (°C) and dryness x4 are selected as factors to reflect the regional natural conditions. Due to the scale dependence of topographic relief, the best statistical unit is determined by moving window method, and the corresponding window is 21 × 21 pixel size, the best statistical unit size calculated by using DEM with 25 m resolution is 0.397 km2.
(2) The demographic factor causes the change of agricultural land use and agricultural production composition caused by the flow of urban and rural population, and then causes the change of agricultural regional function. It is characterized by the proportion of agricultural population x5 (%), urbanization rate x6 (%), and the number of rural population x7 (person).
(3) The socio-economic factors measure the regional economic development from the per capita resident savings, and characterize the social factors from the changes in the agricultural industrial layout caused by the adjustment of industries among different subjects and the impact of public financial investment on agricultural production capacity. Therefore, the per capita resident savings deposit balance x8 (yuan / person), the proportion of secondary industry x9 (%) per capita public financial agricultural expenditure x10 (yuan / person).
(4) The factors of technological progress make agricultural production break the original regional restrictions and improve the layout of crop production and the structure of agricultural products. The application intensity of chemical fertilizer x11 (t/ha) and mechanical power intensity x12 (kW·h/ha) are selected for characterization.
(5) Land use benefit factors affect crop yield. The multiple cropping index x13 (times) is selected to characterize the intensive utilization of cultivated land.
(6) Location factors select the distance from the Township resident to the county and district administrative center to represent the market-oriented location conditions x14 (km).
(7) The selection of policy factors is based on the impact of major project construction in the hilly and gully region of the Loess Plateau, such as the implementation of major ecological projects, farmland and water conservancy construction and other comprehensive measures, so the regional conversion of farmland to forest index x15 and agricultural capital investment x16 (yuan) are selected for characterization.
Table 4 Influencing factors for regional functional differentiation of gully agriculture in the Loess Plateau
Influencing factors Variable Influencing factors Variable
Natural factors x1 Topographic relief x9 Proportion of secondary production
x2 Altitude x10 Per capita fiscal expenditure
x3 Accumulated temperature Technological factors x11 Unit fertilizer consumption
x4 Dryness x12 Unit mechanical power
Demographic factors x5 Population structure Land benefit factors x13 Multiple crop index
x6 Urbanization rate Location factors x14 Geographic conditions
x7 Rural population Policy factors x15 Returning farmland to forest index
Socio-economic factors x8 Per capita household savings x16 Agricultural capital investment

4.2 Model training and validation

In order to reveal the influencing factors of functional value, this paper selects multiple regression model to detect the influencing factors. In view of the possible spatial correlation effect between variables, spatial econometric regression model is introduced into the spatial effect analysis, including spatial lag model (SLM) and spatial error model (SEM), so as to measure the spatial correlation degree between regional functional differentiation and influencing factors.
The mathematical expression of SLM is:
\[Y=\rho Wy+X\beta +\varepsilon \]
where ρ is the spatial autoregressive correlation coefficient; W is the weight matrix; Wy represents spatial lag operator; β is the regression coefficient of independent variable; ε is a random error vector.
The mathematical expression of SEM is:
\[Y=X\beta +\nu,\text{ }\nu \text{=}\lambda W\nu +\varepsilon \]
where λ is the spatial error autocorrelation parameter, which represents the direction and degree of interaction in the region; V is the random error term; ε is the error vector of normal distribution; Wv is the spatial lag interference vector.
After calculation, among the four functional values, except for the Moran's I value of cultural heritage and leisure function (0.4739, p=0.318), which failed to pass the significance test of spatial dependence, the Moran's I value and LMLAG of the other three functions all passed the significance level test (p=0.000). In addition, the R-LMLAG of agricultural production functions, the R-LMERR of ecological conservation and maintenance function, and the R-LMERR of cultural inheritance and leisure function are at the significant level, and the LAMDA of the four functions have passed the significance test (Table 5), indicating that each spatial dependence of the indicator is significant, therefore, this paper further introduces SLM and SEM to verify the influencing factors.
Table 5 The calculation results of the four types of GARF evolution influencing factors in the Loess Plateau
Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
Natural factors x1 -0.2036*** -0.2159*** -0.1824*** -0.1259*** -0.1714*** -0.1213***
x2 -0.0455 -0.0104 -0.0221 0.0138 0.0064 0.0073
x3 0.0793*** 0.0901*** 0.0725*** 0.0292** -0.0103 0.0204
x4 -0.1016 ** 0.0223* -0.0267 -0.0324 -0.0191 -0.0308
Demographic factors x5 -0.0082 -0.0104* -0.0133 -0.0075 -0.0165 -0.0108
x6 -0.0767** -0.0963** -0.0829** -0.0220 -0.0685 -0.0378
Socio-economic factors x8 0.1212*** 0.0312** 0.0638* -0.0607* -0.0402 -0.0565*
x9 -0.0496 -0.0582** -0.0606** -0.0544* -0.0145 -0.0458
x10 0.0595 0.0156* 0.0403* 0.0588 0.0670** 0.0599*
Technological factors x11 0.1987*** 0.1666*** 0.1692*** 0.0565* 0.0142 0.0443
x12 -0.1652*** -0.1512*** -0.1570*** -0.0067 0.0482 0.0064
Land efficiency x13 0.1668*** 0.1364*** 0.1464*** 0.0703* 0.0207 0.0660**
Location factors x14 0.0244 0.0224** 0.0144* 0.0149 0.0223 0.0077
Policy factors x15 0.1170*** 0.1546*** 0.1533*** -0.0606 0.0033 -0.0392
x16 0.1078*** 0.0266** 0.0507** 0.1237*** 0.1374*** 0.1227***
C 0.1540*** 0.1649*** 0.0320*** 0.4542*** 0.4627*** 0.3549***
P (Spatial lag) 0.4207*** 0.2349*** 0.3518*** 0.2027
λ (Spatial lag) (0.1015) (0.1032) 0.1082 (0.1089)
Statistical test R2 0.3069 0.3917 0.3949 0.1363 0.2226 0.1640
Adjustment R2 0.2864 0.1107
F-statistic 14.9638 5.3322
LogL 592.7280 629.4719 634.7820 526.5900 555.7679 536.9280
AIC -1143.4600 -1216.9400 -1225.5600 -1011.1800 -1069.5400 -1029.8600
SC -1047.9700 -1121.4600 -1125.5300 -915.6970 -974.0530 -929.8280
Moran’s I (error) 0.6136*** 0.6582***
LMLAG 36.145*** 35.0842***
LAMDA 0.7382*** 0.6451***
Factors Variable MLRM MLRM SEM SLM
Coefficient Coefficient Coefficient Coefficient
Natural factors x1 0.2291*** -0.2502*** -0.2617*** -0.2153***
x2 0.0550 0.1120 0.0119 0.0004
x3 -0.0810*** 0.1120*** 0.0840*** 0.0967***
x4 -0.0332 -0.0207 0.0248 -0.0049
Demographic factors x5 -0.0205 0.0235 0.0276 0.0219
x6 -0.0114 -0.0072 -0.0337 -0.0200
x8 -0.0753*** 0.1561*** 0.1228*** 0.1358***
x9 -0.0504 -0.0344 -0.0213 -0.0350
x10 0.1428*** 0.0000 -0.0328 -0.0089
Technological factors x11 0.1349 0.1322*** 0.1066*** 0.1180***
x12 0.0082 -0.1389*** -0.1204*** -0.1358***
Land efficiency x13 -0.0757*** 0.0161 0.0014 0.0165
x14 0.1819 -0.0444** -0.0273 -0.0488**
Policy factors x15 -0.2119*** -0.0901*** -0.0306 -0.0643**
x16 0.2568*** 0.1870*** 0.1529*** 0.1705***
C 0.2194 0.4115*** 0.4045*** 0.3042***
P (Spatial lag) 0.5357 0.2743***
λ (Spatial lag) (0.0927) (0.0932)
Statistical test R2 0.3298 0.3403 0.3749 0.3686
Adjustment R2 0.3100 0.3208
F-statistic 16.6316 0.0967
LogL 550.8870 649.9640 663.6799 663.9680
AIC -1059.7700 -1257.9300 -1285.3600 -1283.9400
SC -964.2910 -1162.4500 -1189.8800 -1183.9100
Moran’s I (error) 0.4639 0.5437***
LMLAG 19.6297 39.5286***
R-LMLAG 1 1.0359
LAMDA 0.5357 ***

Note: ***, ** and * represent the significance levels of 0.01, 0.05 and 0.1, respectively.

The research findings have shown that the R2 value increases after considering the spatial correlation, indicating that the interpretability of the spatial econometric model to the influencing factors is significantly improved. From the regression results of SLM and SEM models of the three functions, the estimated values of spatial lag terms for APF, ECF, and ESF are 0.2349, 0.2027, and 0.2743, at a significance level of 1%. This indicates the strong spatial endogenous interaction effect among the three functions. Thus, an increase in the APF, ECF, or ESF of a certain watershed unit will positively promote the corresponding functional value increase of the adjacent watershed unit. The SEM spatial error terms of APF and ECF are 0.4207 and 0.3518, respectively, at a significance level of 1%, indicating a spatial interaction effect of error term between explanatory variables of the two functions, which might be omitted (error term).

4.3 Impact mechanism analysis

The regression results show that several factors such as location, urbanization rate, and non-agricultural industry development boost or inhibit the performance of various gully agricultural functions. They are transmitted and diffused between different functions, ultimately affecting gully agriculture regional function (Figure 7). In the past 20 years, the economic and ecological win-win goal orientation of the Loess Plateau has promoted the continuous optimization of the regional pattern of agriculture. In particular, driven by policies such as the grain for the green project (GGP), agricultural structure adjustment, gully land consolidation (GLC), and economic forest and fruit planting, a characteristic industrial layout along the hilly and gully region has been formed, which has driven the green and high-quality development of typical districts and counties. In the long term, attention must be paid to efficiently dealing with the internal relationship between humans and nature, ecology and economy, mainly to promote gully ecological industrialization and industrial ecology.
Figure 7 Impact mechanism of GARF in the Loess Plateau
(1) APF is closely related to the natural environment, and areas with flat terrain, sunny valleys, and high accumulated temperatures are more ambient conditions for crop cultivation. The urbanization rate and population structure have an adverse impact on APF. The higher the urbanization rate and the proportion of secondary industry proportion, the more non-agricultural employment opportunities and the lesser the proportion of agricultural income.
(2) Per capita public financial expenditure and population structure are negatively correlated with ESF. Increasing the per capita financial expenditure in agriculture can alleviate the financial difficulties of agricultural production, promote agricultural development and related jobs and improve ESF. The larger the terrain relief and the wetter the climate, the stronger the ECF. Increasing capital investment in returning farmland to forests, gully and slope water and soil conservation and gully consolidation is conducive to improving ECF.
(3) Urbanization rate, geographic conditions, and population employment structure have a steady and positive impact on CRF. In order to effectively coordinate the relationship between town and city and improve the agricultural area, it is necessary to strengthen the financial investment policy support for supporting agriculture and guiding social capital to invest in it. In addition, it is crucial to encourage farmers to increase the investment level of gully agricultural land, improve the investment structure and efficiency, and promote the technical level of productive power facilities and equipment.
Overall, the promotion of advanced agricultural technology and market put APF in a state of mutual promotion among gully basins. Watersheds with ideally optimal ecological conditions will have apparent spatial spillover effects and promote ECF improvement in the surrounding watersheds. With the development and utilization of agricultural leisure resources and cultural exchange and integration, CRF has a radiative driving effect on surrounding areas.

5 Discussion

The multifunctional coordination of modern gully agriculture is not only an efficient approach to build a new pattern of coordinated development of modern agriculture with mutual integration of advantages, benefit sharing, and risk sharing, but also a practical way to deepen the agricultural industry value chain and achieve farmers’ income and common prosperity. The Loess Plateau is in a new period of green transformation and high-quality development (Yang et al., 2018; Guo and Liu, 2021). How to optimize the agricultural industrial structure, explore its multifunctional value, and promote the deep integration of new industries, new formats, and traditional industries, while forming a new model of production, lifestyle, ecology, culture, and regional rural-urban cooperation is a frontier issue that must be addressed urgently (Liu et al., 2021). It is necessary to comprehensively conduct the research on the regional multifunction of gully agriculture, refine the regional function integration scheme and control path of gully agriculture, build a modern agricultural system with regional characteristics and synergy and complementarity, and formulate differentiated development strategies (Cao et al., 2019; Liu et al., 2020).
Specifically, it is necessary to promote the adjustment of agricultural production supply structure, elevate the agricultural product quality, coordinate the planting structure of grain, economy and feed, and improve large-scale operation. At the same time, social capital must be encouraged to turn to agriculture to ensure the effective supply of agricultural products and regional food security. Secondly, urbanization and non-agricultural industry development in gully areas should be reasonably promoted, the transfer of agricultural surplus labor should be promoted, and the pressure on agricultural employment and social security should be alleviated. It is also necessary to strengthen the profitable industries of the gully, extend the industrial value chain, promote the deep processing of modern agricultural products and e-commerce platforms, and expand the gully for farmers to increase their income. In addition, it is necessary to pay attention to the role of modern geographic engineering technology in the maintenance of ecological environment such as gully and slope treatment and ecological restoration. There is an urgent need to rationally allocate hilly, slope, and gully resources, promote the development of three-dimensional gully agriculture, and support the chain development of characteristic modern gully agriculture. Combined with the region’s unique natural and cultural advantages, develop leisure gully sightseeing agriculture in various forms of cultural experience, tourism and health, to realize the green water and mountains on the Loess Plateau and prosper the people.
There are still some limitations to research methods. This paper uses only a multiple regression model to measure the correlation robustness between regional functional differentiation and influencing factors, which might raise spatial error interaction of variables. However, in the model construction and test process, spatial errors can be reduced by mutual verification with a spatial econometric model to ensure the reliability of the study findings. Overall, the evaluation of the evolution of the regional function of gully agriculture in this context is not comprehensive. With the recent progress in modern technical means and the refinement of statistical methods, as well as changes in the new development environment. In future studies, it is urgent to focus on the regularity of the evolution of GARF in the Loess Plateau, and deeply investigate its regional geographical background, influence mechanism, and the double cycle and correlation effect of the Yellow River Basin (Liu, 2021; Qu et al., 2021). This research recommends exploring the multifunctional expansion and differentiation mechanism of modern gully agriculture with the integration of rural and urban development based on the advantageous conditions, promoting the realization mode, standardized mode and suitable area of gully agriculture and multifunctional agriculture, and supporting the agricultural modernization and sustainable high-quality development of the Loess Plateau.

6 Conclusions

This study analyzes the characteristics of gully agriculture in the Loess Plateau from the perspective of the human-earth system, investigates the multifunctional change and mechanism, and proposes the path selection and theoretical mode of green development to guide the high-quality development in the Loess Plateau. The main conclusions are as follows:
(1) The spatial concentration of production and supply function of agricultural products (APF) in the gully of the Loess Plateau is high. In contrast, the ecological conservation and maintenance functions (ECF), employment and social security functions (ESF), cultural heritage and leisure function (CRF) are low. Overall, the spatial distributions of the four functions presented an apparent regularity.
(2) APF has been significantly enhanced, which is mainly distributed in point clusters and strips in the farming and pastoral areas of Wubao in northern Shaanxi and the river valley of Shuozhou in western Shanxi to the Yanhe river basin in northern Shaanxi. The high-value areas of ESF are clustered around the urbanized metropolitan circles and urban-rural staggered areas along the Great Wall in northern Shaanxi. ECF is concentrated in areas with dense natural endowments and vigorous ecological conditions in northern Shaanxi. However, CRF is most significant in the municipal districts and the surrounding regional central cities.
(3) There are noticeable disparities in GARF due to the geographical environment and socio-economic development stages. Among them, natural factors significantly impact on the supply of agricultural products, employment and the formation of social security and ecological conservation functions, while socio-economic factors significantly differ in the four functions. The dynamic evolution of the functional difference of gully agriculture can reflect the general law of rural human-earth system transition in gully areas, and thereby assisting in decision-making for high-quality development of agricultural in the Loess Plateau.
Benoit A, Todd H, Allison M, 2014. Identifying the extent of the urban fringe and its impact on agricultural land values. Land Economics, 90(4): 587-600.


Cao Z, Liu Y, Li Y, et al, 2019. Rural transition in the loess hilly and gully region: From the perspective of “flowing” cropland. Journal of Rural Study,

Costanza R, Arge R, Groot R, et al, 1997. The value of the world’s ecosystem services and natural capital. Nature, 386: 253-260.

Fang Y, Liu B, Liu J, 2019. Territorial types and optimization strategies of agriculture multifunctions: A case study of Jilin Province. Progress in Geography, 38(9): 1349-1360. (in Chinese)


Fu B, Zhao W, Zhang Q, et al, 2014. Landscape Pattern Change and Soil Erosion in the Loess Plateau. Beijing: Science Press.

Hu X, Wang L, Xue L, et al, 2010. Study on the sustainable development of agricultural ecosystems in western China based on emergy analysis method. Journal of Southwest University, 32(2): 7-12. (in Chinese)

Huang J, Tichit M, Poulot M, et al, 2015. Comparative review of multifunctionality and ecosystem services in sustainable agriculture. Journal of Environmental Management, 149: 138-147.


Li Y, Li Y, Fan P, et al, 2019. Impacts of land consolidation on rural human-environment system in typical watershed of the Loess Plateau and implications for rural development policy. Land Use Policy, 86: 339-350.


Li Y, Zhang X, Cao Z, et al, 2021. Towards the progress of ecological restoration and economic development in China’s Loess Plateau and strategy for more sustainable development. Science of The Total Environment, 756: 143676.


Liu Y, 2021. Urban-Rural Transformation Geography. Singapore: Springer Nature.

Liu Y, Li Y, 2017. Revitalize the world’s countryside. Nature, 548: 275-277.


Liu Y, Liao H, Li T, et al, 2019. Spatio-temporal diversity and influencing factors of multi-functionality of land use in mountainous regions. Transactions of the Chinese Society of Agricultural Engineering, 35(21): 271-279. (in Chinese)

Liu Y, Liu X, Liu Z, 2022. Effects of climate change on paddy expansion and potential adaption strategies for sustainable agriculture development across Northeast China. Applied Geography, 141: 102667.


Liu Y, Liu Y, Chen Y, 2011. Territorial multi-functionality evaluation and decision-making mechanism at county scale in China. Acta Geographica Sinica, 66(10): 1379-1389. (in Chinese)


Liu Y, Long H, Li Y, 2021. Human geography research based on the new thinking of global rural-urban relationship. Acta Geographica Sinica, 76(12): 2869-2884. (in Chinese)


Liu Y, Zhang Z, Wang J, 2018. Regional differentiation and comprehensive regionalization scheme of modern agriculture in China. Acta Geographica Sinica, 73(2): 203-219. (in Chinese)


Long H, Liu Y, Zou J, 2008. Assessment of rural development types and their rurality in eastern coastal China. Acta Geographica Sinica, 64(4): 426-434. (in Chinese)

Lu D, Fan J, 2009. Regional Development in China in 2050. Beijing: Science Press.

Lu D, Yang X, Shi Y, et al, 2020. Rural regime shifts and transformation development on the Loess Plateau. Acta Geographica Sinica, 75(2): 348-364. (in Chinese)


Lu S, Liu Y, Qin F, 2019. Spatio-temporal differentiation of agricultural regional function and its impact factors in the Bohai Rim region of China. Acta Geographica Sinica, 74(10): 2011-2026. (in Chinese)


Lv Y, Li P, 2016. A clustering analysis method of weighted principal component distance. Statistical Research, 33(11): 102-108.

Peng J, Lv H L, Liu Y X, et al, 2015. International research progress and perspectives on multifunctional landscape. Advances in Earth Science, 30(4): 465-476. (in Chinese)


Qu L, Huang Y, Yang L, et al, 2020. Vegetation restoration in response to climatic and anthropogenic changes in the Loess Plateau, China. Chinese Geographical Science, 30(8): 89-100.


Qu L, Liu Y, Chen Z, 2019. Spatio-temporal evolution of ecologically-sustainable land use in China’s Loess Plateau and detection of its influencing factors. Journal of Mountain Science, 16(5): 1065-1574.


Qu L, Wang Y, Li Y, et al, 2021. Spatial evolution of gully agricultural production function and its enlightenment in loess hilly and gully region. Transactions of the Chinese Society of Agricultural Engineering, 37(21): 259-268. (in Chinese)

Rockström J, Williams J, Daily G, et al, 2017. Sustainable intensification of agriculture for human prosperity and global sustainability. Ambio, 46(1): 4-17.

Rudel T, Schneider L, Uriarte M, et al, 2009. Agricultural intensification and changes in cultivated areas, 1970-2005. Proceedings of the National Academy of Sciences, 106(49): 20675-20680.


Sun Y, Wu Y, Chen R, et al, 2018. Path analysis of agricultural project design based on function expansion: A case study of Naya Mountain Villa. Asian Agricultural Research, 10(2): 22-31.

Vanslembrouck I, Huylenbroeck G, Meensel J, 2005. Impact of agriculture on rural tourism: A hedonic pricing approach. Journal of Agricultural Economics, 56(1): 17-30.


Wang C, Peng Q, Tang N, et al, 2018. Spatio-temporal evolution and the synergy and trade- off relationship of cultivated land multi-function in 2005-2015: A case of Shapingba District, Chongqing City. Scientia Geographica Sinica, 38(4): 590-599. (in Chinese)

Wen Q, Zheng D, 2019. Identification and revitalization of rural poverty-stricken areas in northwest China. Geographical Research, 38(3): 509-521. (in Chinese)


Wu C, 2007. Economic Geography of China. Beijing: Science Press. (in Chinese)

Xie G, Zhang Y, Lu C, et al, 2001. Service value of natural grassland ecosystem in China. Journal of Natural Resources, 16(1): 47-53. (in Chinese)

Yang R, Pan Y, 2021. Rural vulnerability in China: Evaluation theory and spatial patterns. Journal of Geographical Sciences, 31(10): 1507-1528.


Yoshid A. 2001. Economic valuation of multifunctional roles of agriculture in hilly and mountainous areas in Japan. Proceeding of International Seminar on Multi-functionality of Agriculture, 191-200.

Yu M, Yang Y J, Chen F, et al, 2019. Response of agricultural multifunctionality to farmland loss under rapidly urbanizing processes in Yangtze River Delta, China. Science of the Total Environment, 666: 1-11.


Zhang X, Jin X, Fan Y, et al, 2019. Spatial-temporal characteristics and coordination status of the land use function transition in Jiangsu province from 1995 to 2015. Journal of Natural Resources, 34(4): 689-706. (in Chinese)


Zhou Z, Li M, 2017. Spatial-temporal change in urban agricultural land use efficiency from the perspective of agricultural multi-functionality: A case study of the Xi’an metropolitan zone. Journal of Geographical Sciences, 27(12): 1499-1520.