Land Use Policy 88 (2019) 104185

Contents lists available at ScienceDirect

Land Use Policy

journal homepage: www.elsevier.com/locate/landusepol

Influential factors and classification of cultivated land fragmentation, and implications for future land consolidation: A case study of Jiangsu Province T in eastern China ⁎ Jing Liua,b, Xiaobin Jina,b,c, , Weiyi Xua,b, Rui Suna,b, Bo Hana,b, Xuhong Yanga,b,c, Zhengming Gua,b, Cuilan Xud, Xueyan Suid, Yinkang Zhoua,b,c a School of Geography and Ocean Science, Nanjing University, 163 Xianlin Avenue, Qixia District, Nanjing, 210023, China b Key Laboratory of Coastal Zone Exploitation and Protection, Ministry of Land and Resources, 163 Xianlin Avenue, Qixia District, Nanjing, 210023, China c Natural Resources Research Center, 163 Xianlin Avenue, Qixia District, Nanjing University, Nanjing, 210023, China d Jiangsu Province Land Development and Consolidation Center, 58 Shuiximen Street, Jianye District, Nanjing, 210024, China

ARTICLE INFO ABSTRACT

Keywords: Cultivated land fragmentation (CLF) is one of the main obstacles hindering the development of agricultural Cultivated land fragmentation modernization and mechanization. Systematically exploring the general distribution characteristics, influential Regional differentiation factors and classification of CLF are of great significance for improving regional agricultural production capacity, fl In uential factors promoting resource conservation and intensive use, and ensuring national food security, especially at the re- Classification gional scale. In this study, we established a new conceptual index system using multivariable linear regression, Land consolidation geographical detectors, and magic cube model for CLF assessment as well as an analysis of the spatial differ- Jiangsu Province entiation characteristics, driving mechanism and management zoning trade-off of CLF in Jiangsu Province in eastern China based on multi-source data characterizing geographic, land-use and socio-economic information. The results showed that the connotation of CLF has spatial-scale characteristics due to the differences in func- tions positioning of cultivated land resources in macro-social security and micro-livelihood maintenance. At the national/regional scale, CLF mainly covers the natural (resource), spatial and utilization attributes of cultivated land. Based on this, the CLFI in Jiangsu presents a spatial pattern that gradually increases from north to south, with significant spatial differences. Besides, the CLFI within built-up areas is significantly higher than that outside built-up areas, and its fractal dimensions both within and outside the urban planning built-up areas show the spatial pattern of "spatial distribution > resource endowment > convenience of utilization". Furthermore, average plot size, the proportion of industry and service industry, gross domestic product, slope, grain output, and plot distance from town are the dominant factors driving the spatial differentiation of CLF, with the in- fluence power (q) is 0.472, 0.204, 0.133, 0.129, 0.097 and 0.084, respectively. Location conditions and socio- economic activities have significant effects on the spatial differentiation of CLF within the built-up areas while highlighting the role of rural settlements outside urban built-up areas on CLF. Finally, we propose a two-level zoning system for diminishing the CLF and optimizing the utilization of cultivated land resources in Jiangsu based on influencing factors and fragmentation characteristics. The findings of this study will assist the gov- ernment in developing appropriate regional context and land consolidation policies and coping strategies to CLF and food insecurity issues, and achieve sustainable development goals.

1. Introduction (Looga et al., 2018), which also implies that maximization of the pro- vision of products and services for human's increasing food demands More than half of the world's population (54%) currently lives in requires more rational and effective utilization of the finite global land urban areas, and this proportion is expected to grow to 66% by 2050 area and natural resources. (Masini et al., 2018). The rising population expects 70% higher food Cultivated land fragmentation (CLF) is a common agrarian feature production (by the year 2050) and more sustainable land management of many transition economies as well as developing countries (Niroula

⁎ Corresponding authors at: School of Geography and Ocean Science, Nanjing University, 163 Xianlin Avenue, Qixia District, Nanjing, 210023, China. E-mail addresses: [email protected] (J. Liu), [email protected] (X. Jin). https://doi.org/10.1016/j.landusepol.2019.104185 Received 14 April 2019; Received in revised form 11 August 2019; Accepted 27 August 2019 0264-8377/ © 2019 Elsevier Ltd. All rights reserved. J. Liu, et al. Land Use Policy 88 (2019) 104185 and Thapa, 2007; Falco et al., 2010). CLF has become a critical factor in et al., 2019), farms and households (Looga et al., 2018; Ntihinyurwa ensuring food security and sustainability of land resources in East Asia et al., 2019). While these studies work together to recognize the char- (Qi and Dang, 2018), especially for China where per capita cultivated acteristics of cultivated land resources and greatly expanded the land is only 0.08 ha, far below the global average of 0.20 ha per capita breadth and depth of this topic, there is still a great room for im- (Du et al., 2018; Jin et al., 2017). As a prominent feature in traditional provement in the following aspects due to limitations in technical agricultural production in China, CLF not only enriches production methods and data acquisition: diversification, reduces agricultural production risks, and increases farmers' income, but also causes a number of negative influences in- (1) The connotation of CLF needs to be improved. The social security cluding reducing the efficiency of agricultural production, wasting the function of cultivated land resources on macro-scale and the live- workforce in rural areas and improving production costs (Falco et al., lihood maintenance function on a micro-scale determine that the 2010), and then becomes a critical damper to China's agricultural characteristics of CLF vary significantly in different spatial scales modernization and scale development (Sun and Liu, 2010). Studies (e.g., national, regional or local scales, household levels). Currently, have shown that China's CLF wastes about 3%–10% of the effective the connotation of CLF mainly focuses on land property rights of the agricultural land area (Zhang et al., 1997), increases labor costs per ton farm, household, or individual levels. However, little attention has of grain production by 115 yuan and reduces land productivity by been paid to understanding the effects of natural, spatial, and uti- 15.3% (Bai et al., 2014). Meanwhile, the annual output of tuber crops lization properties of cultivated land resources on fragmentation, and wheat decreased by 9.8% and 6.5% respectively for each additional especially in the context of the spatial scales. plot (Wan and Cheng, 2001). (2) Indicators for assessing CLF are limited. Although CLF is mostly To further alleviate the negative influences of CLF, the Chinese understood as a high number of farmed plots or as a high number of government has focused on CLF control and management in recent plot co-owners (Ciaian et al., 2018), it is a more complex phe- years. The No.1 Document of the Central Committee put forward the nomenon. It includes the size distribution and shape of land plots, macro-strategic plan of "accelerating the development of land circula- the spatial distribution, and number of parcels, the distances be- tion and actively developing moderate-scale management" in 2017, tween plots and their uses or locations (Sundqvist and Lisa, 2006; marking a new era of transformation and development of China's tra- Latruffe and Piet, 2014). Because the quantification of several di- ditional agricultural production into the scale, intensive and modern mensions of CLF simultaneously is challenging (Ciaian et al., 2018), agriculture. As important land management and remediation instru- most studies measure CLF only based on one aspect (e.g., landscape ment, land consolidation (LC) has been shown to effectively tackle the patterns, the number of plots or their average size) (Sichoongwe CLF problem and promote agricultural modernization in many coun- et al., 2014), thereby the expansibility of CLF assessment is rela- tries across the world (FAO, 2004; Hartvigsen, 2015). The dominant tively weak. discourse is that fragmented land ownership and land use tend to be (3) The spatial scale of CLF associated with practical guidance needs to ineffective and unwanted, and LC is then a solution to this quandary be further expanded. Currently, the research on CLF is mostly fo- (Ntihinyurwa et al., 2019). Despite the close correlation between CLF cused on micro-scale constrained by data acquisition from house- and LC, little attention has been paid to the differences in regional holds surveys. Even if there are some individual case studies fo- patterns and directions of LC affected by regional differences of CLF and cusing on CLF across regions or countries (Niroula and Thapa, its influential factors, which thereby leading to the spatial mismatch 2005; Hartvigsen, 2014), to the best of our knowledge a compara- between LC practices and characteristics of arable land resources to tive study guiding the optimal utilization of cultivated land re- some extent, especially at the regional scale. With the new goal of sources and LC practices through the comprehensive measurement arable land protection of "quantity-quality-ecology", the aims of the LC of spatial differences and influencing mechanisms of CLF based on are shifting toward the integrated goals of increasing cropland, redu- multi-source data is still missing. cing fragmentation, improving agriculture infrastructure, promoting agricultural productivity, and improving environmental quality (Du To fill the gaps of existing CLF studies and the needs for sustainable et al., 2018). If this is the case, LC policies may focus on the regional LC practices, this paper presents an empirical case study of Jiangsu endowment difference and fragmentation of cultivated land resources, Province in eastern China. This study analyzes the spatial-scale char- including characteristics of scale, spatial pattern, and infrastructure acteristics of CLF from the perspective of spatial differences in func- conditions. Therefore, based on the deep understanding of the char- tional positioning of cultivated land. Then, we construct a compre- acteristics of CLF's spatial pattern, further excavation of the driving hensive CLF evaluation index system, including natural, spatial, and mechanism and influential factors of CLF will provide practical support utilization attributes of cultivated land resources. Next, based on Three- for LC practices to inform the policy-makers of more locally-adaptive dimensional magic cube, Geographical detectors, and Multivariable decisions. linear regression, the study analyzes the regional differences and in- Previous studies on CLF primarily have three foci. One focus is to fluential factors of CLF integrated with multi-source data that char- explore the connotation of CLF in terms of conceptual design (Sundqvist acterize geographic, land use, agricultural production, and socio-eco- and Lisa, 2006; de Vries, 2016) and classification (Sklenicka et al., nomic information. Finally, based on the above analysis, our study 2014). The second focus is to discuss the effects of CLF from the per- proposes a zoning system for the management of CLF in Jiangsu spectives of the governance of the commons (Qi and Dang, 2018; Zang Province based on the fusion of regional differences and influential et al., 2019), productivity (Sundqvist and Lisa, 2006; Looga et al., factors of CLF to offer insights into CLF management and LC practices 2018), and production diversification (Sichoongwe et al., 2014; Ciaian improvement. et al., 2018). The third focus is to analyze the driving factors of CLF in The information derived from this analysis might not only be used terms of cultural tradition (Bizimana et al., 2004; Hartvigsen, 2014), to support regional LC planning and policy-making, for example, to physical conditions (Ntihinyurwa et al., 2019), socio-economic (King identify the current cultivated land resources characteristics among and Burton, 1982), and land distribution (Sklenicka et al., 2014; natural, spatial and utilization attributes for tackling the problem of Jürgenson, 2016). Generally speaking, the previous studies have CLF, but also has theoretical and practical significance for formulation showed three characteristics: (1) research perspective focuses on dis- of regional cultivated land utilization strategies to promote agricultural persed land owners and dispersed plot locations; (2) research methods modernization and ensure food security. are mainly based on theoretical analysis or qualitative description; (3) research scales are mostly concentrated on middle- or micro- scales such as towns (Sun and Liu, 2010), villages (Ciaian et al., 2018; Zang

2 J. Liu, et al. Land Use Policy 88 (2019) 104185

2. Theoretical framework characteristics of CLF from the perspective of spatial differences in functional positioning of cultivated land? 2) how do we measure it? 2.1. Literature review 2.2. Spatial-scale characteristics of CLF At present, there is no single commonly agreed definition of CLF (Ntihinyurwa et al., 2019). In this regard, Mcpherson (1982) defined Coping with this very complex issue needs a sound knowledge about CLF as a situation where household's land resources are divided among the level, forms, performances and causes of CLF along with all its ef- several spatially separated plots; King and Burton (1982) characterized fects at different spatial scales, and the components of food security at CLF as a fundamental rural spatial problem whereby farms are poorly both national and household levels and their relationships organized at different locations across space. Similarly, many authors (Ntihinyurwa et al., 2019). Because, the issue of CLF is connected to the (Sundqvist and Lisa, 2006; Hartvigsen, 2015; de Vries, 2016) con- dual attributes of resources and assets for social security and livelihood sidered CLF as the scattering of farmland or farming household pos- maintenance due to the spatial differences in functional orientation of sesses several non-contiguous land plots. Throughout the existing con- cultivated land, which has significant implications for resource utili- ceptual understanding of CLF, these conceptualizations focus however zation strategy and farmers' production choices both at national and only on dispersed land owners, the number of plots or dispersed plot household levels. locations held by single owners without considering the variety in size, Differences in the formation of cultivated land functions dominated use, shape and other aspects of the respective plots (Demetriou et al., by top-down institutional supply and bottom-up market demands lead 2012; Ntihinyurwa et al., 2019). In fact, the forms and causes of CLF to its hierarchical and scale dependence (Song et al., 2014), to a certain should be linked with each of the characteristics of cultivated land, extent, especially for different social groups (e.g., countries, states, including the features of natural (resource) property, i.e., focusing on governments, households, farmers, etc.). For the states or governments, natural endowment characteristics of cultivated land resources (such as cultivated land, as a precious and scarce natural resource, mainly its resource capacity, spatial distribution, soil qualities, size, patch possesses the function of macro-social security, including maintaining numbers and density, etc.); spatial property, i.e., emphasizing the social stability, ensuring food security, maintaining ecological security landscape pattern characteristics (such as the shapes of dispersed plots, and national economic contribution, etc.; while for households or their distance, agglomeration degree, and segmentation degree, etc.); farmers, it serves as an essential asset for their survival and mainly utilization property, i.e., emphasizing the utilization conditions (such as plays the role of micro-livelihood maintenance, such as basic living its ownership, infrastructures construction, accessibility, etc.), and security, family economic contribution, and employment security. other aspects (King and Burton, 1982; Muchová, 2017). Since their Therefore, the spatial hierarchy and scale differences of the cultivated causes and effects on the utilization and management of cultivated land land function positioning (Jiang et al., 2011) determine that CLF is a resources also vary from spatial-scale characteristics, which leads to the composite system (Fig. 1) that includes multi-scale (e,g., national scale, spatial-scale differences in CLF's management. However, most of the regional scale, household scale, etc.), multi-level (resource fragmenta- previous studies ignore the impact of natural attribute, spatial pattern, tion, spatial fragmentation, utilization fragmentation, ownership frag- and utilization attribute of arable land on CLF when analyzing the mentation, etc.), multi-factor (resource endowments, spatial patterns, spatial forms and causes of CLF in general, which in turn leads to the facilities construction, management and utilization, etc.), and then has development of broad and non-appropriate coping policies and strate- differences in guiding regional LC practices. gies (Ntihinyurwa et al., 2019), especially at the national or regional At the household/farmer scale, farmers are the basic production levels. decision-making units to obtain agricultural products and improve in- As for CLF measurement and its influential factors, CLF is re- come by working on cultivated land resources. Thus, the characteristics presented by the number of plots and/or their average size in most of of asset attributes, such as ownership status and economic output value, the researches mentioned above. However, these indices ignore critical make cultivated land resources as the necessary material to access spatial variables such as plot shapes as well as non-spatial variables economic benefits for farmers' survival and development. Farmers pay (Ciaian et al., 2018), for example, cultivated land production and uti- more attention to their livelihood maintenance function brought by lization conditions, infrastructures construction and accessibility of cultivated land. Due to the diversification of farmers' production and land parcels. Because they do not show all the information or dimen- livelihood needs, the micro-scale of CLF is mainly characterized by the sions related to CLF and may not capture all the constraints that CLF land ownership fragmentation in dispersed plots and the fragmentation imposes on agricultural production system (Latruffe and Piet, 2014; of farmland utilization modes, which is mostly caused by institutional Looga et al., 2018), and the influence mechanism of external (en- rights factors such as land reforms, land policies, and economic pro- vironmental) factors other than the characteristics of cultivated land on cesses (King and Burton, 1982; Bizimana et al., 2004). As such, CLF CLF, in particular in terms of distance, socio-economic development, based on micro-perspective can provide practical guidance for land use production and living conditions, land policies and land allocation, decision-makers to inform field planning, facility layout, ownership topography, and agricultural development. With the deepening of re- adjustment, and organization and implementation of LC projects. search endeavors, in addition to the above-mentioned sole variables, At the national/regional scale, cultivated land resources, as an es- some scholars have gradually recognized the multi-dimensional struc- sential natural resource to promote agricultural development, play a ture of CLF, and tried to characterize the information of CLF in distance, crucial role in stabilizing grain production patterns, ensuring national size, shape and dispersion of plots (Gonzalez et al., 2007), which would food security, and maintaining social stability, and possess multi-attri- enrich the principles of CLF. However, those elaborate measures were butes of natural, spatial, and utilization (as noted in Section 2.1). not tested on a regional scale or even a real sample of farms but instead However, the characteristics of natural property are more affected by were applied to a possible dataset (Latruffe and Piet, 2014). factors of topography, water, and soil matching patterns, and even CLF's measurement and its influence mechanism are the complex natural disasters, etc.; spatial property may be connected with the ele- and vital issue (Looga et al., 2018), and have significant implications in ments such as spatial segmentation of linear features, e.g., rivers and terms of food security, productivity improvement, rural development, roads, and human activities; while utilization property may be mainly agricultural modernization and sustainability of land resources (Qi and linked with the level of regional socio-economic development and land Dang, 2018). In the context of the call to use resources and the en- use patterns as well as facilities construction by the state. Based on this, vironment more efficiently and sustainably to attain the economies of the CLF at the regional scale mainly reflects the differences of resource scale of cultivated land and promote sustainable development for agglomeration, spatial distribution, and facilities construction, which human well-being, two questions arise: 1) what is the spatial-scale can play a guiding role in deepening the understanding of locally

3 J. Liu, et al. Land Use Policy 88 (2019) 104185

Fig. 1. Spatial-scale characteristics of CLF. The left and right blue circles respectively indicate the functional orientation of cultivated land at the national/regional scale and the household/farmer scale. Gray arrows represent the dominant function of cultivated land resources for different social groups. Red arrows indicate the main contents of CLF at different spatial scales. problems in arable land utilization, implementing differentiated re- (2) Calculating the CLF index gional LC strategies, and identifying the key issues and solutions of LC accordingly to diminish the CLF and realize the scale efficiency. This study constructed the cultivated land fragmentation index It is challenging to achieve scale management of cultivated land at (CLFI) with the standard value of "1-weighted linear sum of indicators" the household/farmer level under the constraints of scattered resources, to consistent with evaluation meanings and experience and knowledge. large population, and scarce land resources. Because the resource The calculation method of CLFI is shown in formula 1. fragmentation of cultivated land is the basis for the ownership frag- n m mentation, and the ownership fragmentation is the further extension of CLFI =−1 ∑∑Iij w ij w i i==11()j (1) resource fragmentation. Therefore, as for the goal of promoting the development of large-scale and modern agriculture, it is urgent to give Where, n denotes the number of dimensions; m represents the number priority to diminish the resource fragmentation at the national/regional of indicators in dimension i; Wij represents the weight of the j-th in- scale to further provide sufficient support for addressing the ownership dicator in i; Wi is the weight of dimension i; Iij is the normalized value of fragmentation at the household/farmer scale. the indicator j in dimension i. As a result, based on the above analysis, this study considers CLF at the macro-scale as a situation where cultivated land in specific region 3.2. Analyzing the impact factors of CLF possesses the comprehensive characteristics of insufficient resources endowment, spatial dispersion, and limited production and operation, (1) Influential factors of CLF and its basis which are mainly driven by the regional socio-economic development, production and living conditions, land distribution process, natural Cultivated land is a natural geography-socio-economic complex of environmental conditions, and agricultural development. dynamic integration of topography, climate, hydrology, soil, plants as well as human land-use behaviors and its impacts. As a result, culti- vated land utilization system (CLUS) is composed of endowment char- 3. Materials and methods acteristics of cultivated land resources, regional natural environmental conditions, and human socio-economic activities, where they interact 3.1. Multidimensional evaluation of CLF and co-exist with each other to jointly promote the sustainable land use and serve social demands. Among them, as the critical production (1) Indicators for assessing CLF factors of CLUS, the endowment of cultivated land such as its scale and spatial distribution, is always linked to the volume of agriculture pro- Based on theoretical basis of land resources science, land economics, duction and resource-supply capacity (Zhang et al., 2018) due to pro- and landscape ecology (Liu, 2010; Li, 2015), our study proposes a new viding the material basis and spatial carrier for human survival and conceptual index system for CLF assessment at regional scale aiming at development. In this process, the level of socio-economic development, the demands of "moderate-scale operation and intensive and efficient production and living conditions, land distribution mechanism, and development" in national agricultural modernization development natural environment characteristics are vital links and media of inter- strategy, with three dimensions of resource endowment, spatial dis- action between resources endowment and CLUS external environment. tribution, and convenience of utilization. The definitions and quanti- Their spatial heterogeneity, to a certain extent, influences the ways of tative methods of the indexes are shown in Table 1. utilization and development of cultivated land resources. CLF, as a multidimensional phenomenon in this interaction process, is

4 J. Liu, et al. Land Use Policy 88 (2019) 104185

highlighted with the mismatch between regional cultivated land re- sources and productivity development in terms of scale characteristics, c area c area c area fi fi fi c area c area spatial distribution patterns, etc. Its formation mechanism includes not fi fi only the division of natural environmental factors, e.g., topography, c area

fi rivers, but also human activities, e.g., land use, economic development. In summary, these factors can be broadly classified into five categories

denotes the number of in view of CLUS and previous studies (Hartvigsen, 2014; Jürgenson, m 2016; Zang et al., 2019), namely, socio-economic development level, production and living conditions, land distribution process, natural environmental conditions and agricultural development (Table 2).

(2) Multivariable linear regression

CLFI is used as the dependent variable and indicators in Table 2 as independent variables. The general form of the multivariable linear

indicates the area of infrastructure land. regression equation is as follows: p H

nition Y=+ β βx + βx + βx ++..... βx ++ ..... βx + u fi 010 2 2 3 3 j j k k (2) denotes the distance between patches;

c area c area β i Where, Y is CLFI, the dependent variable; 0 is constant; k represents ecting the total number of cultivated land patches in a speci ecting the construction of agricultural infrastructures in a speci ecting the number of cultivated land patches per unit area ecting the segregation degree of cultivated land within a speci W β fl fl fl ; the number of independent variables; j represents the regression i e e Re fl Representing the accessibility of cultivated land patches in a speci Indicating the endowment of cultivated land resources inIndicating a the speci spatial agglomerationspeci fi degree of cultivated landCharacterizing within a the average distancespeci fi of cultivated land resources within a Indicating the regularity degree of patches shape in a speci coefficient, and μ is a random error perturbation term subject to a normal distribution. er zone of rural roads;

ff (3) Geographical detectors +R – – – – + + +Re +R cbu

fi A geographical detectors model (Wang et al., 2010) was im- ⎤ ⎥ ⎦

i fl

A plemented to analyze the impact magnitude of the major in uencing A () factors on CLFI, which was complementary to the multivariable linear × × i ⎞ ⎠ ⎤ ⎦

i … LA

i regression method. Assuming that A={A h =1,2, , L, where L is the P h, i is the circumference of object A P N i … P

0.25 fi

ln ( ) factor classi cation number} are the attributes associated with the + ; ⎛ ⎝ i 2 i 2ln( ) P ⎡ ⎢ ⎣ i 1

1 geographical stratum of a suspected CLFI, the power of the determinant m W = = N i n i 1 +LA …

1 A={A h =1,2, ,L}toCLFI is given by: = h, m i − /A 2 /A +∑ p =∑ cation method Expected sign Indicator de r (1) N 1 100 fi L nn =∑ ⎡ ⎣

() 1 = 2

= q =−1 nσh = ∑ h

NN 2 WMSI I nσ A A (3) M TNP TACL = LA PD = N/A ED = P/A PAIL=H LPC = F m h=1 is the area of patch

i Where, nh is the number of samples in the sub-region h of the de-

LA terminant Ah; n is the total number of samples of interest over the entire 2 2 region A; σ h and σ indicate the dispersion variance of sub-region h and the entire region A, respectively. The range of q is [0,1]. A larger value of q suggests more clear spatial differentiation of CLFI and denotes the area of cultivated land within a speci

r stronger spatial determination of the independent variable x to CLFI. F i

3.3. Trade-off judgment method for CLF zoning

First, we employ k-means clustering (Jain, 2010) to classify counties into several clusters to characterize CLF first-level zones based on the Total number of patches Total area of cultivated land Area-weighted mean shape index Patch density Mean nearest neighbor distance Edge density Proportion of agricultural infrastructure land Aggregation index Land patch accessibility geographic detection results of the major influencing factors on CLFI.

denotes the total number of patches; Then, we use magic cube model (Ye et al., 2017) to construct the fra- N mework for reflecting the spatial combination characteristics of CLF upon different dimensions, and finally, propose a two-level zoning system aiming at the management of CLF based on influencing factors and fragmentation characteristics. More concretely, the three sides of the magic cube represent resource endowment (x), spatial distribution (y) and convenience of utilization (z) of cultivated land, respectively.

represents the total circumference of patches; We further divide x, y, and z into four grades (i.e., lowest, lower, higher, Resource endowment (B1) Spatial distribution (B2) Convenience of utilization (B3) P and highest with numbered 1, 2, 3, 4, respectively) by using "Mean ± 0.5 × Standard Deviation" (Long et al., 2012). On this basis, 64 com- binations are merged by consulting relevant experts to minimize

is the total area of cultivated land; within-group variability and maximize its homogeneity, and the study A area can be divided into several categories, such as the utilization ef- ) Standard level (B) Indicator level Quanti

A ficiency improvement zone, spatial intensive merging zone, facility construction improvement zone, scale management and land circula- tion zone, resource allocation optimization zone, and comprehensive (A) In the table, improvement zone. The conceptual model and classification standards Cultivated land fragmentation Target level ( Table 1 Indicators for assessing CLF. Note: combinations of two patches; for second-level zones of CLF are shown in Fig. 2 and Table 3,

5 J. Liu, et al. Land Use Policy 88 (2019) 104185

Table 2 Influential factors of CLF at regional scale.

Categories of influential factors Influential factors Variables' description Relationship with CLF

Socio-economic development Land use degree (x1)Reflecting the comprehensive status of With the improvement of SEDL, the expansion of construction land level (SEDL) land resources utilization and and the development of the transportation network will have some development negative impacts on the layout and shape of cultivated land Gross domestic product Reflecting the level of regional economic

(x2) development The proportion of industry Reflecting the state of the regional

and service industry (x3) industrial structure Production and living conditions Plot distance from rural Indicating the interference degree of PLC reflects the possibility of cultivated land being occupied by

(PLC) settlements (x4) cultivated land by human activities urban construction, the disturbance degree caused by human Plot distance from town Reflecting the possibility of cultivated activities, the spatial segmentation degree of linear features and the

(x5) land being occupied by urban support for large-scale and mechanized farming construction Plot distance from nearest Reflecting the chance of cultivated land

water system (x6) being blocked or separated by linear features

Average plot size (x7)Reflecting the support to agricultural mechanization Land distribution process (LDP) Per capita cultivated land Reflecting the endowment status of LDP, determined by factors such as cultivated land quality, size,

area (x8) cultivated land resources distance, environment, and location, is closely linked with CLF

Cultivated land quality (x9)Reflecting the quality difference of regional cultivated land resources

Natural environmental Slope (x10)Reflecting the surface steepness of The slope is the essential elements reflecting the regional topographic conditions (NEC) regional cultivated land resources characteristics, while water network density indicates the division Water network density Reflecting the division degree of degree of cultivated land by rivers, lakes, etc. They are essential

(x11) cultivated land by rivers, lakes, etc. natural environmental factors for the formation of CLF

Agricultural development (AD) Grain output (x12)Reflecting the status of regional AD reflects the utilization efficiency of cultivated land in a specific agricultural development region, and to some extent is affected by CLF

Food supply capacity (x13)Reflecting the value of local agricultural production and the production diversification

Note: In the table, the calculation method of x1 refers to Zhuang and Liu (1997). For x12, “water network” includes rivers, canals, lakes, reservoirs, ponds, etc. More description and calculation method of x14 can be found in Sun et al. (2017).

Fig. 2. The conceptual model of second-level zones of CLF based on spatial differences. The green, red, and blue solid lines represent x (resource endowment), y (spatial distribution) and z (convenience of utilization) of CLF, respectively. The red ellipses represent second-level zones of CLF while blue ellipses indicate the statistical characteristics of the corresponding second-level zones in fractal dimension.

6 J. Liu, et al. Land Use Policy 88 (2019) 104185

Table 3 The classification standards and characteristics for second-level zones of CLF based on influencing factors and spatial differences.

Second- level zoning Classification standards and the corresponding Essential characteristics of second-level zones and the main directions of combination of magic cube units diminishing CLF through LC

Utilization efficiency improvement zone (4,4,4) (4,4,3) (4,3,4) (4,3,3) (3,4,3) (3,4,4) Possessing the conditions of resources and facilities for promoting large- (High values of x, y, and z) (3,3,4) (3,3,3) scale management and modern agriculture development endowed with superior resources, concentrated spatial distribution, and convenient utilization. Attention should be paid to agricultural technological innovation to improve the efficiency of cultivated land resources utilization Spatial intensive merging zone (Higher x and z (4,1,4) (4,2,4) (4,1,3) (3,2,3) (3,2,4) (3,1,4) Cultivated land resources have apparent advantages in scale and values while lower y value) (3,1,3) (4,2,3) (2,1,4) (4,1,2) (4,2,2) (2,2,4) convenience while its spatial layout is scattered and fragmented. It should focus on the optimization of the spatial structure of cultivated land combined with rural construction land consolidation to promote the centralized and continuous distribution Facility construction improvement zone (4,4,1) (4,4,2) (4,3,1) (4,3,2) (3,4,1) (3,4,2) Having great advantages in terms of resources scale and spatial (Higher x and y values while lower z value) (3,3,1) (3,3,2) (4,2,1) (2,4,1) agglomeration, but there are certain deficiencies in infrastructure construction and land accessibility. Attention should be paid to the improvement of regional agricultural infrastructure construction to enhance conditions for production Scale management and land circulation zone (1,4,4) (1,4,3) (1,3,4) (1,3,3) (2,4,4) (2,4,3) Having great advantages in spatial agglomeration and convenient (Higher y and z values while lower x value) (2,3,4) (2,3,3) (1,2,4) (1,4,2) (2,4,2) utilization, but there are certain deficiencies in the scale of resources. Attention should be paid to the market-oriented circulation of dispersed land plots and ownership adjustment to create conditions for promoting large-scale management Resource allocation optimization zone (3,1,2) (3,2,1) (2,1,3) (2,3,1) (1,2,3) (1,3,2) Resources scale, spatial agglomeration, and utilization convenience of (x, y, and z values are all medium) (2,2,3) (3,2,2) (2,3,2) cultivated land are all at a medium level, or two of the three dimensions are at a medium level, and the phenomenon of CLF is more serious. Differentiated LC strategies should be adopted to promote the optimal allocation of regional resource elements according to the characteristics of CLF in different regions Comprehensive improvement zone (1,1,1) (1,1,2) (1,2,1) (1,2,2) (2,1,2) (2,2,1) Resources scale, spatial agglomeration, and utilization convenience of (Low values of x, y, and z) (2,1,1) (2,2,2) (1,1,3) (1,1,4) (1,3,1) (1,4,1) cultivated land are poor, or two of the three dimensions are poor, and the (3,1,1) (4,1,1) phenomenon of CLF is the most serious. Attention should be paid to the global planning and comprehensive improvement of production factors such as cultivated land, water systems, roads, and rural settlements

Fig. 3. Study area. respectively. (RMB) in 1978 to 8590.094 billion RMB in 2017 with a national land area of 1.12% carrying a population of 5.78% and a total economic output of 10.4%. Nevertheless, the district also suffers from severe re- 3.4. Overview of the research area source shortage, with the per capita arable land area of only 0.057 hm2, which is only 60.96% of the average national level and approaching the Jiangsu Province is situated in the Yangtze River Delta on the warning line of 0.053 hm2 established by Food and Agriculture Orga- 2 coastal center of eastern China with an area of 107,000 km and a total nization of the United Nations. population of approximately 79.763 million (Fig. 3). As an economic- Agriculture in Jiangsu will continue to develop as planned by ally developed region in eastern China, the gross domestic product "Opinions on Further Promoting Agricultural Supply-side Structural (GDP) of the Jiangsu Province increased from 24.924 billion renminbi

7 J. Liu, et al. Land Use Policy 88 (2019) 104185

Table 4 Data sources and descriptions.

Data name Data source Time-series Resolution

Land Use/Land Cover Data Jiangsu Province Land Use Change Survey Database 2014 1:1,0000 Socioeconomic Data Jiangsu Statistical Yearbook 2016 County/township levels Cultivated Land Quality Data Results of National Survey of Cultivated Land Quality 2010 1:10,0000 DMSP/OLS Night Light Data http://www.noaa.gov/ 2013 1 km × 1 km DEM (ASTER GDEMV2) http://www.gscloud.cn/ 2009 30 m × 30 m

Reform and Promoting Continuous Income Increase of Farmers" issued Yangtze River in Nanjing, Zhenjiang and Changzhou are scattered and by Jiangsu Provincial People's Government in 2017: Strengthen the poorly concentrated. The average value of the convenience of culti- infrastructure construction of farmland, actively develop the moderate- vated land use in the province is 0.392, which indicates that the ac- scale operation of agriculture, and construct modern agricultural pro- cessibility of cultivated land and infrastructure construction are poor duction and management system. This indicates that the regional (Fig. 4c), and the proportion of the townships at lowest, lower, higher, agricultural development will transform from the traditional agri- and highest level is 29.30%, 18.86%, 20.45%, and 31.39%, respec- culture of improved variety, intensive cultivation, fine management, tively. and more fertilizer to modern agriculture of green, ecological, efficient, and sustainable. Notably, in the context of comprehensively deepening 4.1.2. Characteristics of CLF and its fractal dimensions within and outside reforms to ensure food security and sustainable development, CLF is a the urban planning built-up areas severe obstacle existed in pushing forward scale management of farm- We further analyze the regional differences of CLF and its fractal land and agricultural mechanization and adopting agricultural in- dimensions within and outside the urban planning built-up areas of novations and new technologies. Hence, it is of great significance to different cities (Fig. 5), due to the differences in production and life- analyze the spatial disparities, influencing factors, and governance style, infrastructure construction, industrial structure characteristics, approaches of CLF to formulate state-regulated reasonable resource and intensity of human activities, which provides adequate support for utilization policies, improve LC modes and directions, and promote the determining locally strategies for arable land use. Fig. 5c shows the transformation and upgrading of agriculture. urban planning built-up areas of cities in Jiangsu Province. In summary, the spatial differences of CLF and its fractal dimensions 3.5. Data sources and pre-processing within and outside the urban planning built-up areas are significant. Overall, the CLFI in built-up areas is significantly higher than that In the quantitative evaluation, the cultivated land plot is used as the outside built-up areas (Fig. 5d), indicating that the CLF in the built-up research unit, and the research results are statistically analyzed at the area is more serious, which could be a result of a higher level of socio- township level (since townships are the most basic administrative units economic development, that is, the segmentation, erosion and spatial of LC strategy implementation) under the administrative division of crowding effect on agricultural space caused by high-intensity human Jiangsu Province in 2014, including 1389 sub-objects and 7.191 million activities, perfect infrastructure construction, and intensive demands sub-plots. We assembled data that characterize geographic, land use, for land use within built-up areas. The average provincial CLFI within and socio-economic information (Table 4). Due to the differences in the urban planning built-up areas is 0.423, while the average CLFI in data sources and spatial accuracy, this paper processes all data on the southern, central, and northern Jiangsu is 0.458, 0.450, and 0.377, ArcGIS 10.2 software platform. The weights of indicators in Table 1 are respectively, revealing a prominent regional difference. In terms of calculated by combining the entropy weight method and analytic fractal dimensions of CLF, the order of fractal dimensions both within hierarchy process method (Shemshadi et al., 2011), which can address and outside the urban planning built-up areas are the following: spatial the limitations of over-reliance on evaluation data and subjective ar- distribution (SD) > resource endowment (RE) > convenience of utili- bitrariness. zation (CU) (Fig. 5a/b), which indicates that reducing the number of land plots, promoting the centralized and continuous distribution of 4. Results and analysis arable land, and enhancing infrastructure construction to improve production conditions are still the critical contents of LC in the future. 4.1. Spatial differentiation characteristics of CLF Specifically, the values of RE and SD outside the built-up areas are generally higher than that within the built-up areas, indicating that the 4.1.1. Spatial differentiation of provincial CLF scale of the cultivated land resources outside the built-up area and the According to the measurement model of CLFI, we calculated the spatial agglomeration are superior; while the spatial difference of CU CLFI of Jiangsu, and divided it into four grades (i.e., lowest, lower, contradicts the RE and SD, and CU within the built-up areas is generally higher, and highest) by using natural break method. The CLFI in higher than that outside. Jiangsu Province presents a spatial pattern that gradually increases from north to south (Fig. 4d). The average provincial CLFI is 0.408, 4.2. Influencing factors of CLF while the average CLFI in northern, central, and southern Jiangsu is 0.372, 0.430, and 0.456, respectively, revealing obvious regional dif- 4.2.1. General analysis of influencing factors of CLF ferences of CLF. Judging from the quantity structure of each level of Collinearity diagnostics were performed assessing the variance in- townships, the CLFI in Jiangsu Province presents a "U-shaped" structure, flation factor (VIF) of each impact factor, and all calculated VIFs were and the proportion of towns with the highest and lowest grades of CLFI below five indicating that there were no multiple collinearities among is 31.53% and 30.45%, respectively. In general, the resource endow- factors. In the initial analysis, significance tests were made on linear ment (RE) attribute of Jiangsu is high, with the average provincial RE is regression relationship among all variables in Table 2, and there are

0.633, but the resource scale in southern and central Jiangsu is sig- only five variables, i.e., gross domestic product (x2), plot distance from nificantly lower than that in northern (Fig. 4a) with the average RE is town (x5), average plot size (x7), per capita cultivated land area (x8), 0.569, 0.592, and 0.687, respectively. The spatial difference of spatial and grain output (x12), passed the significant test at the 1% or 5% level distribution (SD) attribute is similar to that of RE (Fig. 4b), and culti- and the adjusted R2 of the multivariable linear regression analysis was vated lands around Taihu Lake, such as Suzhou and Wuxi, and along the 0.460, which was significantly inconsistent with expectations.

8 J. Liu, et al. Land Use Policy 88 (2019) 104185

Fig. 4. Spatial differentiation characteristics of CLF. The green shades indicate the lowest grade of CLF in the fractal dimension, and the red shades indicate the highest grade. The difference between green to red characterizes the increasing trend of CLF in each fractal dimension.

Dependent and independent variables have been further processed with multivariable linear regression and geographical detectors, which logarithm to limit the broad range of data changes with the adjusted R2 shows that the modeling-based cause analysis results had good relia- is 0.622 and passed the significance tests of F-test and t-test. In parti- bility and objectivity. cular, some variables, i.e., land use degree (x1), plot distance from nearest water system (x ), cultivated land quality (x ), and water net- 6 9 4.2.2. Identifying dominant factors for geographical differentiation of CLF work density (x11), are excluded from the modeling process due that they failed significance test in the t-test, while the other variables are (1) The dominant factors of regional differentiation of CLF at the pro- significant at the 1% or 5% level. The multivariable linear regression vincial scale model is shown in formula 4, while the regression results are expressed in Table 5. We further ranked the absolute values of q and Beta in descending

LnY=−0.709 + 0.044Lnx2+0.077Lnx3−0.016Lnx4−0.016Ln- order to cross-validate the results of multivariable linear regression and x5−0.132Lnx7−0.034Ln8+0.016Lnx10−0.009Lnx12−0.039Lnx13 (4) geographical detectors (Table 6). As indicated in Table 6, the results of the Beta analysis of the linear regression standardized coefficient fea- In the formula, the influential factors with larger regression coeffi- tured the importance of indicators with significantly positive and ne- cients are average plot size (x ), the proportion of industry and service 7 gative correlations with CLFI. Among all measured indicators for CLFI industry (x ), gross domestic product (x ), food supply capacity (x ), 3 2 13 in Jiangsu Province, the top six factors (ranked in descending order of per capita cultivated land area (x8) and slope (x10). absolute values of Beta) included average plot size (x7), the proportion In addition, geographical detectors were used to identify the impact of industry and service industry (x3), gross domestic product (x2), grain factors in Table 2 (x1∼x13)onCLFI; the influence power (q) on CLF is: output (x12), slope (x10), and plot distance from town (x5) are basically 0.067, 0.133, 0.204, 0.032, 0.084, 0.011, 0.472, 0.059, 0.039, 0.097, consistent with the results of influence power (q). These econometric 0.003, 0.129 and 0.016, respectively. There is no significant difference results fully indicate that these six indicators are the dominant factors in the effect of variables in Table 2 (x ∼x ) on CLF by the results of 1 13 affecting the geographical differentiation of the CLF in Jiangsu

9 J. Liu, et al. Land Use Policy 88 (2019) 104185

Fig. 5. The characteristics of CLF and its fractal dimensions within and outside the urban planning built-up areas of Jiangsu Province.

Table 5 Results of multivariable linear regression.

Categories of influential factors Influential factors Unstandardized B Std. Error t Sig. VIF

SEDL x1 0.004 0.061 0.060 0.952 1.720

x2 0.044*** 0.006 7.448 0.000 1.402

x3 0.077*** 0.012 6.249 0.000 1.540

PLC x4 −0.016*** 0.005 −3.309 0.001 1.237

x5 −0.016*** 0.004 −3.900 0.000 1.865

x6 −0.001 0.004 −0.020 0.984 1.068

x7 −0.132*** 0.006 −21.980 0.000 1.315

LDP x8 −0.034*** 0.006 −5.703 0.000 1.474

x9 −0.008 0.039 −0.194 0.846 1.158

NEC x10 0.016*** 0.009 1.776 0.007 1.109

x11 0.002 0.002 0.780 0.435 1.322

AD x12 −0.009** 0.005 −1.817 0.021 2.115

x13 −0.039*** 0.007 −5.969 0.000 1.872 Constant −0.709*** 0.119 −5.939 0.000

Note: ***, ** denotes significance level at 1% and 5%, respectively.

Province. Influence power (q) of factors in Table 2 on CLF are calculated and ranked as mentioned above, both within and outside the urban plan- (2) Factors of regional differentiation of CLF within and outside the ning built-up areas (Fig. 6). In general, the impact of factors on CLF is urban planning built-up areas basically consistent with the spatial detection results at the provincial

Table 6 Statistics of influencing factors on CLF in Jiangsu Province.

Influential factors x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13

Standardized Beta 0.002 0.320 0.350 −0.089 −0.137 0.000 −0.612 −0.046 −0.005 0.180 0.019 −0.093 −0.297 Ranking of Beta 12 3 2 8 6 13 1 9 11 5 10 7 4 Influence power q 0.067 0.133 0.204 0.032 0.084 0.011 0.472 0.059 0.039 0.097 0.003 0.129 0.016 Ranking of q 73210612189513411

Note: Ranking by the absolute values of Beta or q.

10 J. Liu, et al. Land Use Policy 88 (2019) 104185

Fig. 6. Differences of factors leading to CLF within and outside urban planning built-up areas. scale, especially the top six factors have not changed significantly but while the CLFI shows that higher in the south and lower in the north with slight changes in the ranking of individual factors. This indicates (Fig. 4a). This contradictory spatial pattern indicates that there may that the average plot size (x7), the proportion of industry and service be a negative correlation between the average plot size and CLFI, industry (x3), gross domestic product (x2), grain output (x12) and slope and the result of multivariable linear regression demonstrate a (x10) are still the dominant factors affecting the regional differentiation significant negative correlation between them with R = −0.738. of CLF both within and outside the built-up areas, but for different Especially in Xuhuai plain areas, such as Xinyi, Donghai, and regions, the influence intensity of each factor is different. Specifically, Suining, the endowment of cultivated land resources in these areas within the built-up areas, the order of each factor (ranked in descending is superior, and the scale agriculture and modern agriculture de- order of absolute values of q) is the following: average plot size (x7, velopment level is high, characterizing with a larger average plot 2 q = 0.427) > the proportion of industry and service industry (x3, size (1.362 hm ) and a lower level of CLFI (the average value of q = 0.240) > grain output (x12, q = 0.168) > plot distance from CLFI is 0.356); while in Ningzhenyang and Yili low mountain and town (x5, q = 0.128) > gross domestic product (x2, q = 0.122) > hilly areas in southwest Jiangsu Province, such as Jurong, Lishui, slope (x10, q = 0.086), which indicates that location conditions and and Liyang, the average plot size in these areas is smaller socio-economic activities have significant effects on the geographical (0.472 hm2), and the average value of CLFI (0.466) is obviously differentiation of CLF within the built-up areas. In comparison, the higher than that of other areas. order of each factor outside the urban planning built-up areas is the (2) The proportion of industry and service industry fully reflects the following: average plot size (x7, q = 0.491) > the proportion of in- characteristics of regional industrial structure, and presents a gra- dustry and service industry (x3, q = 0.188) > plot distance from rural dual increase from north to south (Fig. 7b), which is basically settlements (x4, q = 0.142) > gross domestic product (x2, consistent with the spatial patterns of CLFI, that is, the higher the q = 0.125) > grain output (x12, q = 0.084) > slope (x10, q = 0.076), proportion of industry and service industry, the higher the CLFI, which highlights the important impact of rural settlements outside otherwise the lower the CLFI. Multivariable linear regression result urban built-up areas on CLF. Further analysis shows that the influence demonstrates a significant positive correlation between them with power (q) of the same-ranking within the built-up areas is higher than R = 0.147. The secondary and tertiary industries in the southern that outside the built-up areas, indicating the mechanism of CLF in Jiangsu have made outstanding contributions to the economic de- built-up areas may be more complex. While production and living velopment of the province, especially in the areas along both sides conditions have a more significant impact on the CLF outside the built- of Yangtze River and the areas around Taihu Lake. These regions up areas. The smaller average plot size is still the main obstacle existed are horned as the pioneer demonstration areas of national moder- in pushing forward scale management of farmland in the vast rural nization construction with the rapid development of advanced areas. Meanwhile, the alternant distribution of rural settlements and manufacturing industry, modern service industry, and high-tech cultivated land also has an important impact on the development of industries. However, regional urban construction has a significant CLF. crowding effect on agriculture space to a certain extent, resulting in Further analysis of the mechanism of the dominant factors can a relatively high degree of CLFI. By contraries, the northern provide a reference for addressing the local obstacles of CLF. Thus, the Jiangsu, such as Jinhu, Jianhu, and Guannan, is a post-economic top six factors with the most significant influence of power (q)in development area in Jiangsu Province. The proportion of primary Table 6 are analyzed to inform regional agricultural development and industry in these regions is 12.04% with a higher development level LC policies. Fig. 7 shows the spatial differentiation characteristics of of large-scale agriculture and agricultural mechanization, which far each dominant factor. exceeds the provincial average (5.54%), and the CLFI is relatively low (0.372). (1) Average plot size is an important indicator reflecting the support of (3) The gross domestic product reflects the overall level of regional cultivated land resources for large-scale and mechanized agri- socio-economic development. On the whole, the socio-economic cultural development. Generally, the larger the average plot size, development of Jiangsu Province generally presents a spatial dis- the higher the appropriate degree of regional promotion of large- tribution pattern similar to that of the industrial structure. The scale agricultural development, and the smaller the degree of CLF. gross domestic product in southern is significantly higher than that From the perspective of geographical distribution, the average plot in central and northern Jiangsu (Fig. 7c), which is basically con- size in Jiangsu Province is characterized by the spatial distribution sistent with the spatial distribution trend of CLFI (Fig. 4d) while pattern of smaller in the south and larger in the north (Fig. 7a) contrary to the spatial pattern of resource endowment (Fig. 4a) and

11 J. Liu, et al. Land Use Policy 88 (2019) 104185

Fig. 7. Spatial distribution of the main factors influencing CLF in Jiangsu Province.

spatial agglomeration (Fig. 4b) of cultivated land. In the meantime, economic growth and social stability. In terms of grain output, there multivariable linear regression result also shows that there is a are significant regional differences in Jiangsu Province, and the significant statistical positive correlation between gross domestic grain output in northern and central Jiangsu is significantly higher product and CLFI with R = 0.320, indicating that the higher the than that in southern (Fig. 7d). The spatial distribution of grain level of regional socio-economic development, the lower the scale output is contrary to the CLFI, indicating that the better the regional of cultivated land resources, the more dispersed the spatial dis- agricultural development, the lower the CLFI. Further analysis finds tribution and the higher the CLFI, on the contrary, the lower the that the spatial distribution trend of grain production is similar to CLFI. that of resource endowment and spatial agglomeration of cultivated (4) Agricultural development. As an essential element of agricultural land, especially in the Xuhuai plain and the Lixia river plain. These development, grain production is the fundamental guarantee for areas are essential bases for production, processing, and marketing

12 J. Liu, et al. Land Use Policy 88 (2019) 104185

of agricultural and sideline products throughout the province and intensity. While promoting socio-economic development and improving the country. The original superior agricultural production condi- quality of life, rapid urbanization also has a special impact on CLF, due tions characterized by the superior endowment of cultivated land to that rapid expansion of construction land has occupied arable land resources, right scale conditions and spatial distribution of ag- and the spatial separation by urban infrastructure construction, such as glomeration, and the implementation of large-scale LC projects in high-speed lines, highways, and railways, etc. the regions have reduced the degree of CLF to a certain extent. Production and living conditions dominant influence area. This (5) Slope effectively reflects the topographical features of the area. cluster is greatly influenced by factors such as the plot distance from Generally, the gentler the slope, the weaker the spatial division on rural settlements and towns, and the average plot size, with q of 0.424, cultivated land resources. It can be seen from Fig. 7e that the slope 0.382 and 0.449, respectively, which are significantly higher than SEDL in Jiangsu Province shows remarkable spatial connectivity char- (0.150), LDP (0.148), NEC (0.128) and AD (0.135). As the main pri- acteristics with a gradual increase from east to west and the slope of mary division of the CLF in Jiangsu Province, it is mainly concentrated the western region is significantly higher than that in the central in the central and eastern coastal areas of Jiangsu, including 49 coun- and eastern areas. Especially in the low mountain and hilly areas of ties. These areas are dominated by plains with abundant arable land southwest Jiangsu, e.g., Lishui, Jurong, Liyang, and Yixing. The resources, concentrated spatial distribution and apparent advantages in topographic fluctuation in the region varies significantly due to the agricultural production, and are essential commodity production bases higher land's slope. To a certain extent, the spatial separation of for rice, wheat, cotton, and vegetables in the province. Further analysis cultivated land resources is more obvious, resulting in a relatively finds that there are significant differences in the mechanism of the high degree of CLF. county-level objects located within or outside the urban built-up areas. (6) Plot distance from town reflects the location of cultivated land re- Specifically, for the county-level objects within the urban built-up area, sources, and the possibility of occupying on it by urban construction such as Nanjing, Suzhou, and Xuzhou, the q of the above three factors land expansion to some extent. From Fig. 7f, plot distance from on the CLF is 0.378, 0.434 and 0.451, respectively; while the q outside town in Jiangsu Province is gradually decreasing from north to the built-up areas is 0.425, 0.324 and 0.494, respectively, e.g., Sheyang, south, which is contrary to the CLFI. In addition, the result of Hai'an, and Guanyun. These results fully indicate that the smaller multivariable linear regression demonstrates a significant negative average plot size is an important production factor affecting the re- correlation between them with R = −0.319. In terms of spatial gional differentiation of CLF in Jiangsu Province. In addition, plot distribution, the low-value areas of plot distance from town are distance from towns and rural settlements has a significant impact on basically consistent with the spatial extent of the high-value areas the CLF within the built-up areas and outside, respectively. This also of CLFI, especially in southern Jiangsu. These statistical results confirms the spatial detection results of the dominant factors on CLF in show that the smaller the distance between cultivated land re- Section 4.2.2. sources and towns, the more severe the interference of cultivated Natural environmental conditions dominant influence area. Natural land by human activities, and the higher the CLFI; otherwise, the environmental factors have a significant impact on the spatial differ- lower the CLFI. entiation of CLF, e.g., slope. This cluster includes 14 county-level ob- jects. Agricultural planting methods and structures are restricted to 4.3. Zoning trade-off of CLF and implications for future land consolidation some extent due that low mountains and hills dominate landforms in this cluster, and their topography vary greatly. The average q of the 4.3.1. Classification of CLF based on the major influencing factors slope on CLF is 0.624, which is significantly higher than that of SEDL On the basis of quantitative detection of the dominant influencing (0.367), LDP (0.350), PLC (0.380) and AD (0.239). factors of provincial CLF, geographical detectors are used to carry out Multi-factor comprehensive influence area. The q of SEDL, LDP, multi-level geographic detection at the county-level city, and the PLC, NEC, and AD on CLF is 0.733, 0.796, 0.739, 0.691, and 0.791, samples are grouped into four clusters by k-means clustering (Jain, respectively. This cluster consists of 9 county-level objects, mainly 2010), which varied in different influential factors, hence revealing distributed in urban built-up areas and suburban areas with complex four first-level zones (Table 7; Fig. 8). We named the four clusters based impact mechanisms and numerous in fluencing factors. Therefore, given on dominant categories of influential factors. this cluster, the mechanism of CLF is more complicated, and it is more Socio-economic dominant influence area. This cluster is mainly difficult to address the problem of CLF accurately. It is more necessary concentrated in the rapid economic development of southern Jiangsu to adopt differential measures according to local conditions and sci- and the superior endowment of agricultural resources in northern entific evaluation. Jiangsu, including 39 counties. The average q of the SEDL on CLF in the cluster is 0.512, which is significantly higher than that of PLC (0.358), 4.3.2. Zoning scheme for CLF management based on influencing factors and LDP (0.316), NEC (0.366) and AD (0.367). The level of regional socio- fragmentation characteristics, and suggestions for addressing CLF economic development is relatively high, especially in southern The same types of first-level zones based on the major influencing Jiangsu. The CLF in the region is greatly influenced by social and factors on CLF incorporate different sub-categories featuring different economic factors such as industrial structure and land development cultivated land resource endowment, spatial distribution patterns,

Table 7 Average influence power (q) for each influencing factor on CLFI within each first-level zone. The number of counties in each cluster is denoted by n.

Clusters SEDL PLC LDP NEC AD Name

x2 x3 x4 x5 x7 x8 x10 x12 x13

Ⅰ(n = 39) 0.487 0.537 0.328 0.302 0.443 0.316 0.363 0.363 0.370 Socio-economic dominant influence area Average q 0.512 0.358 0.316 0.363 0.367 Ⅱ(n = 49) 0.152 0.147 0.424 0.382 0.449 0.148 0.128 0.164 0.105 Production and living conditions dominant influence area Average q 0.150 0.418 0.148 0.128 0.135 Ⅲ (n = 14) 0.460 0.274 0.393 0.378 0.368 0.350 0.624 0.229 0.249 Natural environmental conditions dominant influence area Average q 0.367 0.380 0.350 0.624 0.239 Ⅳ(n = 9) 0.753 0.713 0.674 0.759 0.785 0.796 0.691 0.792 0.789 Multi-factor comprehensive influence area Average q 0.733 0.739 0.796 0.691 0.791

13 J. Liu, et al. Land Use Policy 88 (2019) 104185

Fig. 8. Classification of CLF in Jiangsu Province based on the major influencing factors on CLFI. The number of counties in each cluster is denoted by n. The petals of different colors in a, b, c, and d represent different categories of influential factors. The size of the petals represents the average q of the influencing factors on CLF upon different dimensions. production, and utilization status. Therefore, we use the magic cube planting and production. The organic, ecological, facility and resource- model to further identified 24 s-level zones (Fig. 9) that were associated intensive agriculture should be actively developed. with fractal dimensions of cultivated land utilization, aiming at pro- Spatial intensive merging zone. This type of areas accounts for viding practical support for regional LC practices for diminishing CLF. 10.51% of the total towns, and mainly distributes in the areas around Utilization efficiency improvement zone. This type of areas accounts Taihu Lake. The average CLFI is 0.404, and the mean value of the re- for 24.76% of the total towns and is mainly distributed in the Xuhuai source, spatial, and utilization properties in this type is 0.624, 0.332, plain and the western Lixia river plain. The average CLFI is 0.338, and 0.516, respectively. The cultivated land resources in the region which is lower than the average level of provincial CLFI (0.408). The have better scale conditions and entire infrastructure, yet the scattered mean value of the resource, spatial, and utilization properties in this and fragmented spatial layout. Rural settlements have a higher degree type of areas is 0.704, 0.739, and 0.535, respectively. As a critical main of separation for cultivated land, and the average distance between grain-producing area in Jiangsu Province, cultivated land area and them is only 85.48 m. Attention in LC should be paid to the spatial grain output in this region account for 30.22% and 32.38% of the integration of regional cultivated land resources by reducing the province respectively, with rich in cultivated land resources, con- number of ridges and combining the dispersed plots to achieve the centrated spatial distribution and complete facilities. Therefore, LC purposes of lowering CLF and expanding the scale of cultivated land. At practices in this region should focus on further improving the utiliza- the same time, for the rural settlements with the scattered and dis- tion efficiency of cultivated land. In reality, the government could ordered spatial layout in the region, rural construction land con- further improve the quality of resource utilization by: (1) actively solidation should be implemented to promote the centralization of promote the construction of high-standard basic farmland and com- farmers' residence and the centralized and continuous distribution of modity grain bases; (2) appropriately adjust regional planting habits arable land. Especially, the spatial structure and layout of rural settle- and planting methods to create conditions for large-scale management ments should be renovated and reconstructed, as well as the infra- and mechanized farming; (3) actively highlight the promotion and structure construction. application of modern new technologies in regional agricultural Facility construction improvement zone. This type of areas accounts

14 J. Liu, et al. Land Use Policy 88 (2019) 104185

Fig. 9. The zoning scheme for CLF management in Jiangsu Province based on influencing factors and fragmentation characteristics. Among them, the first-class zones fully reflect the impacts of dominant influencing factors on CLF, while the second-class zones highlight the combined characteristics of fractal dimensions of CLF within the same type of first-class zones and refine the spatial development directions of optimal utilization of cultivated land resources for CLF management and LC practices improvement in Jiangsu Province. for 21.81% of the total towns, and mainly distributes in some townships the spatial distribution characteristics along both sides of Yangtze in the interior of central Jiangsu and the eastern coastal plain regions in River, around Taihu Lake and along the coast. However, there is a vast northern Jiangsu. The average CLFI is 0.386, while the mean value of spatial difference in combination characteristics of fractal dimensions the resource, spatial, and utilization properties in this type is 0.708, of CLF, due to the vast regional differences in natural environment 0.792, and 0.304, respectively. There are abundant arable land re- characteristics, socio-economic development level, and production and sources with concentrated spatial distribution in these areas, but with living conditions, etc. Accordingly, LC in this region should embrace the large potential for improvement in the construction of agricultural differentiated strategies to adapt to local conditions and promote the infrastructure. LC practices in this region should focus on improving optimal allocation of resource elements. For example, in plain areas, LC production and living conditions, including water conservancy facil- should aim at expanding the resource scale of cultivated land to attain ities, production facilities, and farmland protection facilities. On the the economies of scale, and pay more attention to address the problems other hand, attention should also be paid to the spatial layout and of small scale of cultivated land resources and scattered spatial layout; planning of production factors in the regional LC planning, such as field while in low mountains and hilly areas, LC should focus on improving planning, field road planning, field irrigation, and drainage system the output efficiency of cultivated land through the agricultural infra- planning, and field protection forest planning. structure construction to enhance the convenience of cultivated land Scale management and land circulation zone. This type of areas utilization. accounts for 8.28% of the total towns, and mainly distributes in the Comprehensive improvement zone. This type of areas accounts for plains along the Yangtze river and Lixia river. The average CLFI is 23.18% of the total towns with mean CLFI of 0.491, which is con- 0.405, while the mean value of the resource, spatial, and utilization centrated in the areas of along Yangtze River, e.g., Jurong, Yizheng and properties in this type is 0.431, 0.747, and 0.530, respectively. The Jiangyin, and the areas around Taihu Lake, e.g., Wujiang. The phe- spatial distribution of cultivated land in the region is concentrated, and nomenon of CLF is serious, which is characterized by the inadequate the agricultural infrastructure is well equipped, but there are certain resources scale, dispersed spatial layout, and poor utilization con- deficiencies in resources scale. The average plot size is only 0.621 hm2, venience of cultivated land, with the average values of 0.470, 0.357, which is significantly lower than the provincial average (0.943 hm2). and 0.342, respectively. Socio-economic factors are the dominant fac- LC practices in the region should focus on promoting moderate-scale tors affecting the regional differentiation of CLF, such as the high-in- management of agricultural land and expanding its resources scale. In tensity human activities and urban construction land expansion, ac- practice, our study suggests that: (1) expand the scale of cultivated land counting for 47.2%. Furthermore, the spatial distribution pattern of management through the measures of leveling abandoned ditches and arable land is also affected by the diverse topography and the scattered ridges to meet the needs of irrigation, drainage and mechanized rural settlements, and the average distance between the latter and plots farming; (2) merge fragmented small fields and promote land owner- is only 77.59 m. LC in this region should pay more attention to the ship adjustment simultaneously to create conditions for moderate-scale global planning and overall improvement of production factors such as management of farmland and agricultural modernization. farmland, water system, and roads. Such as, expanding the scale of Resource allocation optimization zone. This type of areas accounts farmland management by digging high and filling low in the same area for 11.45% of the total towns with mean CLFI of 0.440, which presents and leveling abandoned ditches and ridges; strengthening the

15 J. Liu, et al. Land Use Policy 88 (2019) 104185 construction of agricultural infrastructure to promote the effective use research methods, etc., and have made some efforts to address it. of cultivated land resources; and implementing the rural construction As noted above, this study attempts to break through the traditional land consolidation to achieve the centralization of farmers' residence research idea of CLF, that is, average plot size is used to measure the and agricultural production scale. degree of regional CLF, and takes it as a factor to characterize regional production and living conditions to explore its impact mechanism and 5. Discussion and policy implications intensity on spatial differentiation of CLF. The result shows that smaller average plot size is indeed the main factor affecting the spatial differ- 5.1. The contribution of this study and the quantitative basis for academic entiation of CLF, whether based on the results of multivariable linear debate regression (the ranking of influence degree of the average plot size on CLF is 1 st with the absolute value of Beta is 0.624) or geographical 5.1.1. Contribution of the analytical framework for interaction between detectors (also ranks 1 st with the absolute value of q is 0.472), which CLF's spatial-scale characteristics and LC provides a quantitative basis for academic debate. Based on this, we Ensuring food security and promoting sustainable socio-economic have reason to believe that the expansion of the average plot size can development within a sustainable land use framework are still con- diminish the CLF, which has important implications for LC practices. sidered the challenging policy targets owing to limited land space and cultivated land resources along with the rapid growth of population. In 5.2. Policy implications of land consolidation and rural land use system in this case, improving agricultural technologies and policies through land China management tools including land consolidation are effective measures to meet food demands (Hartvigsen, 2015; Du et al., 2018), especially for Our research shows that CLF is the result of the comprehensive ef- ensuring food security at regional or national levels. This requires sta- fect of regional resource endowment and the rural land system (Section keholders (e.g., national, local/provincial governments, ministries, 2.2). This enlightens us that reducing CLF requires not only strength- council) to make a sound knowledge about the characteristics of cul- ening the systematic planning and top-level design of land consolida- tivated land in the multi-dimensional aspects of the resource scale, tion planning at the macro-scale to further highlight the spatial guiding spatial distribution and facilities allocation, and then make the most role of planning, but also innovating the policies of effective utilization realistic decision-making of resources utilization and agriculture de- and management of rural land resources at the micro-scale to create velopment planning. Unfortunately, prior to the lack of crucial knowl- policy environment for the implementation of land consolidation edge and the literature gap about the above problems, the majority of planning and the efficient utilization of cultivated land resources after the governmental programs to tackle CLF have emerged, including the consolidation. Farm Land Use Consolidation (LUC) program (Ntihinyurwa et al., 2019) and the National Land Remediation Plan. The uniform agricultural land 5.2.1. Improving operation modes and structural system of land consolidation models have caused a disconnect between the resource consolidation in China utilization targets at larger scales and the cultivated land utilization 5.2.1.1. CLF should be embedded in LC planning and decision-making to improvements at smaller scales due to that they do not account for the promote the optimal allocation of critical resources, especially at the regional regional differences in resource and environment utilization, and to a scale. State-regulated consolidation is often perceived as a critical certain extent have damaged the local characteristics. measure to tackle the CLF problem (Ciaian et al., 2018). LC planning, Therefore, expanding the spatial-scale characteristics of CLF for as the primary content of reasonable arrangement and classification different social groups and exploring its spatial differentiation char- guidance of land consolidation activities, is an essential basis for the acteristics, diversity mechanisms, and solutions on a larger spatial scale spatial allocation of critical resources, resources utilization, and will be conducive to human well-being. The analytical framework regional agricultural development strategies. However, it mainly presented in this study has the potential to make an important con- focuses on local natural environmental conditions and agricultural tribution to the existing literature as it enriches the connotation of CLF production methods currently, such as topography, geomorphology, to a certain extent and expands its content and research scale. In terms meteorology and hydrology, soil conditions, cropping systems, of practical application, on the basis of scientifically measuring the irrigation, and drainage methods, while paying little attention to the spatial differentiation characteristics and its mechanisms, this study effects of natural, spatial and utilization properties of cultivated land brings in a new idea of considering both natural (resource) property, resources on fragmentation. Regional differences in resources size, spatial property and utilization property of CLF when guiding or im- spatial patterns, and facilities of cultivated land, to a certain extent, plementing the practice, planning and management of regional LC, have caused the serious spatial mismatch among the consolidation which is of great significance to break the dilemma of CLF, improve the modes, the engineering measures and the characteristics of regional utilization efficiency of cultivated land resources and ensure national cultivated land. Thus the benefits of LC need to be improved. food security. This work provides valuable reference and guidance for As the result of a combination of natural, economic, and social cultivated land use optimization and LC practices in China as well as factors, CLF can intuitively reflect the human disturbance to cultivated other developing countries worldwide. land resources, and provide reference and basis for coordinating the relationship among human activities, socio-economic development and 5.1.2. Smaller average plot size is indeed the primary cause of CLF cultivated land protection. Therefore, it is necessary to improve the For a long time, there is agreement on the quantitative evaluation, national LC planning system and incorporate the CLF into the planning causes, and effects of CLF. It is generally believed that CLF is mainly and decision-making of policies to achieve higher quality and sustain- characterized by dispersed plot locations. This conceptualization fo- able of cultivated land utilization. Firstly, governments should regard cuses however on the landscape pattern characteristics of cultivated the region's CLF as the scientific basis of planning and improve its en- land resources and considers that smaller average plot size is the typical forcement to enhance the guidance of national resources utilization. spatial form of CLF. However, this viewpoint is mainly derived from Specifically, governments should be responsible for organizing the strict qualitative experiences summarization, but few quantitative studies and comprehensive investigation of CLF yearly or at regular intervals to verified it. With the deepening of research endeavors, academic circles clarify the characteristics of the CLF, including the level, forms, causes, have aroused heated discussions about whether the average plot size is mechanisms along with all its effects, especially at the regional scales the manifestation of CLF, or the influencing factor leading to its spatial (e.g., administrative regions, states, provinces). On this basis, regional differentiation, or both expression and cause. Scholars have been con- differences in CLF should be taken as guidance for the allocation of fronted with this debate for various reasons, e.g., data acquisition, critical resources, such as finances, markets, agricultural facilities and

16 J. Liu, et al. Land Use Policy 88 (2019) 104185 knowledge, and the decision-making of regional development strategies institutions, etc.), and recreational facilities (e.g., landscape during the adjustment, revision or redesign of relevant planning. Since greening, parks, etc.), etc., so as to improve the economical and these administrative units have the authority, through policies, reg- intensive land use and reconstruct rural living space. ulations, and plans, to make the most efficient arable land use and LC strategies that are most consistent with regional resources character- 5.2.2. Innovative policies for effective utilization and management of rural istics, e.g., financial support, projects arrangement and development land resources strategies. 5.2.2.1. Deepening the reform of rural land property rights system. Although LC can eliminate certain types of land 5.2.1.2. Implementing rural land comprehensive consolidation project for fragmentation through the projects of farmland improvement and rural revitalization and agricultural modernization. The importance of rural settlement optimization (Zhou et al., 2019), thus creating traditional agriculture in China also means that rural areas play a conditions for modern agriculture, however, the fragmentation of the significant role in the country's sustainable development (Cheng et al., ownership of cultivated land at the micro-scale is to some extent the key 2019). Facing the current rural decline in China (Long, 2014; Yang factor that restricts farmers' enthusiasm for production and inefficient et al., 2015; Liu and Li, 2017; Liu, 2018a, b; Liu, 2019), such as of cultivated land utilization. Especially, the scattered and flower- depopulation, non-agriculturalization of rural elements, hollowed arrangement ownership of land plots has further led to the villages, deterioration of rural residential environment, and fragmentation of farmland management. Meanwhile, promoting the inefficiently used or abandoned farmland, etc., the Chinese modernization of agriculture and reconstructing the production and government proposed Rural Revitalization Strategy in 2017 aimed at living space of rural areas through LC are still faced with issues of realizing agriculture and rural modernization. LC is an important unsmooth transfer of land and imperfect development of rural land platform for implementing the Rural Revitalization Strategy. market caused by the ambiguous rural land property rights, incomplete However, CLF is the critical obstacle in the process of implementing power, and the vacancy of ownership subjects (Long et al., 2016; Liu Rural Revitalization Strategy and promoting agricultural et al., 2018). Furthermore, the lack of scientific and efficient modernization. It is also the result of the interaction between the management system of land resources in rural areas has also led to human-land relationship and the external environment in rural areas the parallel management of rural land by agriculture, forestry, animal and is affected by various agricultural production factors such as roads, husbandry, and other departments, as well as a large number of land ditches, and settlements, etc. Against this background, the disputes and conflicts of interests. In this context, there is an urgent comprehensive land consolidation projects covering fields, water, need to push forward the reform of rural land property rights system to roads, forests, and villages should become the key content of rural create policy support for rural land consolidation (Long et al., 2016). land consolidation in the new era to support rural revitalization and China's rural land property rights system mainly includes the agricultural modernization. Based on our study, we suggest as follows: contractual management right of agricultural land and the collective construction land use right. Accordingly, our study suggests as follows: (1) Fields: farmland improvement project (FIP). Under the strategic goal of rural revitalization, the focus of the FIP should be to opti- (1) Further clarify the subject of rural land property rights under the mize the spatial layout of cultivated land and reduce the CLF by framework of collective ownership of rural land. Governments strengthening the construction of high-standard basic farmland and should speed up the construction of laws and regulations for the improving the quality of cultivated land, so as to create conditions definition of rural land property rights, and clarify the social group for the construction of modern agricultural production and man- of "collective ownership", such as villager committees, village col- agement system. lectives, village groups, or grass-roots governments, etc. (2) Water: farmland water conservancy project (FWCP). The goal of (2) Gradually establish a standardized and efficient mechanism for the FWCP should be to build a good irrigation and drainage system to separation and parallel operation of land ownership, land con- enhance the disaster resistance of agricultural production, espe- tracting right and land management right (hereinafter referred to as cially in areas with scarce water resources. Specifically, it includes "three rights"). Effective forms of collective ownership of rural land the transformation and design of field branch canals, lateral canals, should be explored constantly and the respective functions and and bucket canals, the construction of irrigation wells, and the overall effectiveness of the "three rights" should be given full play promotion of water-saving irrigation instruments. through the implementation of collective ownership, stabilization (3) Roads: rural and field road project (RFRP). The goal of RFRP is to of farmers' contractual rights and liberalization of land manage- improve the transportability of agricultural products and materials ment rights. In short, rural land property rights system reform in rural areas, facilitate farmers' travel, and then provide infra- should focus on the protection of farmers' rights and interests and structure support for mechanized agriculture and modern rural the improvement of land use efficiency, by emphatically addressing communities, including field roads, production roads, bridges, and the finely divided and scattered land use pattern by land con- recreational trails, etc. solidation adapted to local conditions (Liu et al., 2014a, 2018). (4) Forests: farmland protection and eco-environmental protection project (FPEPP). The goal of FPAPP should be to build farmland 5.2.2.2. Promoting the transfer of rural land management rights. It is also ecosystem barriers through engineering and biological measures to necessary to recognize that rural transformation, which is characterized enhance soil and water conservation capacity and improve the rural by non-agriculture farmers, loss of young and middle-aged rural labors, ecological environment. Among them, engineering measures and abandonment of farmland, determines that the fragmentation of mainly include farmland shelterbelt networks engineering, bank cultivated land management and the vacant and abandoned rural slope protection engineering, ditch protection engineering, and homestead will exist for a long time in China (Li et al., 2014; Long, ecological engineering of slope surface, etc.; while biological mea- 2015; Yang et al., 2018). More seriously, the subjectivization of the sures mainly include farmland shelterbelts, road protection forest, aged, children, and female groups in rural areas makes it impossible to and forest belt for shore protection, etc. utilize the restored cultivated land effectively. In this case, the transfer (5) Villages: rural settlement optimization project (RSOP). The goal of of contractual management right of agricultural land and collective the RSOP should be to rationally plan and orderly integrate the construction land use right will have important impacts on the rural residential space through strengthening municipal facilities sustainable rural development and modern agriculture (Liu et al., (e.g., water supply, electricity, environmental sanitation, etc.), 2014b, 2018). Consequently, the government could further stimulate public service facilities (e.g., medical service, education, financial the transfer of rural land management rights by: (1) scientifically

17 J. Liu, et al. Land Use Policy 88 (2019) 104185 promoting the registration of rural collective land rights to lay the geographical detectors and magic cube model to explore the spatial foundation for the unified management system of urban and rural land differentiation characteristics of regional CLF using three aspects: nat- (Liu et al., 2013), because clear land property right is the premise of ural (resource) property, spatial property and utilization property, and orderly land transfer; (2) perfecting the relevant laws and regulations the driving mechanism using multi-source data that characterize geo- and technical systems to standardize rural land transfer. To this end, the graphic, land use and socio-economic information. After this, our study national legislature should actively build a sound legal system of land further constructs the guiding zoning scheme for the management of transfer based on extensive investigations to supervise the rural land CLF in Jiangsu Province in eastern China, which offers insights into transfer according to law, and also provide the basis for resolving the sustainable cultivated land utilization and LC practices. disputes. It should clearly stipulate various aspects of the rural land The CLFI in Jiangsu shows a gradual increase from north to south, transfer process, including the subject, scale, mode, price, time and indicating the noticeable regional difference. The mean value of pro- duration, purpose, and division of powers and responsibilities; (3) vincial CLFI is 0.408, and the quantity structure of each level of the developing and improving a unified rural land transfer market to grade of CLFI shows a "U-shaped" structure, which dominated by activate the contractual management right of agricultural land and lowest- and highest-level, revealing greater polarization in CLF. The the rural collective construction land use right (mainly referred to as CLFI in built-up areas is significantly higher than that outside built-up homestead), including the inefficient utilization, abandoned, and idle areas, and its fractal dimensions both within and outside the urban rural residential land transfer market and the scattered, fragmented planning built-up areas show the spatial pattern of "spatial distribu- agricultural land transfer market. The former focuses on the tion > resource endowment > convenience of utilization", which in- establishment of rural homestead withdrawal with compensation, and dicates that promoting the centralized and continuous distribution of market-oriented circulation mechanism, but it should be noted that the arable land, enhancing infrastructure construction, and improving transfer of rural collective construction land use rights into the market production conditions are still the critical contents of LC in the future. must be carried out under the premise of land use planning and land use The results of this study show that x7, x3, x2, x12, x10, and x5 are the control (Liu et al., 2018); while the latter focuses on promoting the dominant factors affecting the geographical differentiation of the CLF, large-scale operation and industrialization of agricultural land by with the influence power (q) is 0.472, 0.204, 0.133, 0.129, 0.097 and cultivating new-type agricultural management mainbody 0.084, respectively. The spatial detection results of the impact factors of (e.g.,agricultural leading enterprises, farmer cooperatives, family CLF both within and outside the urban planning built-up areas are farms, large and specialized family businesses, etc.) and adjusting the basically consistent with these in provincial but with slight changes in ownership of fragmented fields (Long et al., 2016; Liu et al., 2019). In the ranking of individual factors, which indicates that the influence addition, the land administration department shall strengthen the intensity of each factor is different for different regions. In summary, supervision and management of the transfer of land management location conditions and socio-economic activities have significant ef- rights (Li et al., 2018) to protect the interests of farmers. fects on the spatial differentiation of CLF within the built-up areas, while highlighting the role of rural settlements outside urban built-up 5.3. Limitations and future prospects areas on CLF. Finally, this study proposes a two-level zoning plan for optimizing CLF is the result of dynamic processes driven by natural, economic, the utilization of cultivated land resources and diminishing the CLF in social, cultural, and institutional impacts. In addition to the five cate- Jiangsu based on influencing factors and fragmentation characteristics, gories of influential factors mentioned in Section 3.2, CLF is also subject which includes 4 first-level zones and 24 second-level zones. Each level to local traditional culture, e.g., inheritance customs (Niroula and of zones features characteristics and relationships among different CLF Thapa, 2005), marriage, and dowry culture (Hartvigsen, 2014), etc.; in fractal dimensions and influencing factors. The proposal of trade-off institutional context, e.g., land market, land transactions, land con- zoning of CLF management can provide a feasible approach for LC tracting, management and use rights, and the duration and stability of spatial planning, optimal utilization of cultivated land resources and these rights (Zang et al., 2019); and socio-economic changes, e.g., wars, implementation of land management policies in Jiangsu Province, as population growth (Niroula and Thapa, 2007), etc. Particularly the well as a pathway to ensure food security and achieve sustainable de- tradition of land inheritance prevailing in many countries, that is, the velopment. Egalitarian principle applied in the distribution of land among house- hold heirs (Falco et al., 2010; Sklenicka et al., 2014), has made land Acknowledgements fragmentation an on-going process, resulting in landholdings and land parcels getting smaller and smaller, and dispersed over successive This work was supported by Nanjing University Innovation and generations (Mearns and Sinha, 1999). Although it is beyond the scope Creative Program for PhD candidate (CXCY18-21) and National Science of this paper to indulge in a full-scale discussion on the above issues due Technology Support Plan Projects of China (2015BAD06B02). to the limitations of data sources, they are also important factors af- fecting the CLF. Because these factors are based on fine-scale local data References that need to be sourced locally, without which it is challenging to re- search on a regional scale. Bai, Z.Y., Chen, Y., Xie, B.P., Wu, W., 2014. The relationship between landscape frag- The above challenges imply that field surveys and interviews can be mentation and cultivated land use efficiency-Case of Kangle Country, Gansu Province. J. Arid Land Res. Environ. 28 (4), 42–47 in Chinese. conducted in typical areas to quantitatively explore the mechanism of Bizimana, C., Nieuwoudt, W.L., Ferrer, S.R., 2004. Farm size, land fragmentation and CLF's regional differentiation under the interaction of natural, social, economic efficiency in southern Rwanda. Agrekon 43 (2), 244–262. cultural, and institutional factors in further studies. Meanwhile, based Cheng, M.Y., Liu, Y.S., Zhou, Y., 2019. Measuring the symbiotic development of rural fl housing and industry: a case study of fuping county in the taihang mountains in on the current understanding of the spatial pattern and in uencing China. Land use Pol. 82, 307–316. factors of CLF, it is still necessary to further reveal the spatial-temporal Ciaian, P., Guri, F., Rajcaniova, M., Drabik, D., Paloma, S.G., 2018. Land fragmentation evolution law and trend of CLF in a long time series and to improve the and production diversification: A case study from rural Albania. Land use Pol. 76, – analytical framework of CLF management that combines national/re- 589 599. de Vries, W.T., 2016. Post Disaster Consolidation of Land, Memory and Identity. FIG gional macro-scale with household/farmer micro-scale. Working Week 2016. Recovery from Disaster, Christchurch, New Zealand pp. 12. Demetriou, D., Stillwell, J., See, L., 2012. Land consolidation in Cyprus: why is an in- 6. Conclusion tegrated planning andDecision support system required? Land use Pol. 29 (1), 131–142. Du, X.D., Zhang, X.K., Jin, X.B., 2018. Assessing the effectiveness of land consolidation for In this study, we employ multivariable linear regression, improvingagricultural productivity in China. Land use Pol. 70, 360–367.

18 J. Liu, et al. Land Use Policy 88 (2019) 104185

Falco, S.D., Penov, I., Aleksiev, A., Rensburg, T.M.V., 2010. Agrobiodiversity, farm profits use Pol. 79, 285–292. and land fragmentation: evidence from Bulgaria. Land use Pol. 27 (3), 763–771. Masini, E., Tomao, A., Barbati, A., Corona, P., Serra, P., Salvati, L., 2018. Urban growth, FAO, 2004. Operations Manual for Land Consolidation Pilot Projects in Central and land-use efficiency and local socioeconomic context: a comparative analysis of 417 Eastern Europe (No.). Land Tenure Studies. FAO, Rome. metropolitan regions in Europe. Environ. Manage. 1–16. Gonzalez, X., Marey, M., Alvarez, C., 2007. Evaluation of productive rural land patterns Mearns, R., Sinha, S., 1999. Social Exclusion and Land Administration in Orrisa-India. with joint regard to the size, shape and dispersion of plots. Agric. Syst. 92, 52–62. World Bank, Washington. Hartvigsen, M., 2015. Experiences With Land Consolidation and Land Banking in Central Mcpherson, M.F., 1982. Land fragmentation: a selected literature review. Development and Eastern Europe After 1989 (No. 26). Food and Agriculture Organization of the Discussion Papers. 141. Harvard University, pp. 85. United Nations, Rome. Muchová, Z., 2017. Assessment of land ownership fragmentation by multiple criteria. Hartvigsen, M., 2014. Land reform and land fragmentation in Central and Eastern Europe. Surv. Rev. 1–8. Land use Pol. 36, 330–341. Niroula, G.S., Thapa, G.B., 2007. Impacts of land fragmentation on input use, crop yield Jain, A.K., 2010. Data clustering: 50 years beyond K-means. Pattern Recogn. Lett. 31 (8), and production efficiency in the mountains of Nepal. Land Degrad. Dev. 18 (3), 651–666. 237–248. Jiang, G.H., Zhang, F.R., Kong, X.B., Zhao, H.F., Zhou, D.Y., 2011. The different levels and Niroula, G.S., Thapa, G.B., 2005. Impacts and causes of land fragmentation, and lessons the protection of multi-functions of cultivated land. China Land Sci. 25 (8), 42–47 in learned from land consolidation in South Asia. Land use Pol. 22, 358–372. Chinese. Ntihinyurwa, P.D., de Vries, W.T., Chigbu, U.E., Dukwiyimpuhwe, P.A., 2019. The po- Jin, X.B., Shao, Y., Zhang, Z.H., Resler, L.M., Campbell, J.B., Chen, G., Zhou, Y.K., 2017. sitive impacts of farm land fragmentation in Rwanda. Land use Pol. 81, 565–581. The evaluation of land consolidation policy in improving agricultural productivity in Qi, X., Dang, H., 2018. Addressing the dual challenges of food security and environmental China. Sci. Rep. 7 (1), 2792. sustainability during rural livelihood transitions in China. Land use Pol. 77, 199–208. Jürgenson, E., 2016. Land reform, land fragmentation and perspectives for future land Shemshadi, A., Shirazi, H., Toreihi, M., Tarokh, M.J., 2011. A fuzzy VIKOR method for consolidation in Estonia. Land use Pol. 57, 34–43. supplier selection based on entropy measure for objective weighting. Expert Syst. King, R., Burton, S., 1982. Land fragmentation: notes on a fundamental rural spatial Appl. 38, 12160–12167. problem. Prog. Hum. Geogr. 6 (4), 475–494. Sichoongwe, K., Mapemba, L., Ng’ong’ola, D., Tembo, G., 2014. The Determinants and Latruffe, L., Piet, L., 2014. Does land fragmentation affect farm performance? A case study Extent of Crop Diversification among Smallholder Farmers. IFPRI Working Paper 05. from Brittany, France. Agr.Syst. 129, 68–80. International Food Policy Research Institute. Li, X.L., 2015. Fragmentation Evaluation and Spatial Pattern Analysis of Basic Farmland. Sklenicka, P., Janovska, V., Salek, M., Vlasak, J., Molnarova, K., 2014. The Farmland East China University of Technology, Fuzhou. rental paradox: extreme land ownership fragmentation as a new form of land de- Li, Y.H., Wu, W.H., Liu, Y.S., 2018. Land consolidation for rural sustainability in China: gradation. Land use Pol. 38, 587–593. practical reflections and policy implications. Land use Pol. 74, 137–141. Song, X.Q., Wu, Z.F., Ou, Y.Z., 2014. Changes of cultivated land function in China since Li, Y.R., Liu, Y.S., Long, H.L., Cui, W.G., 2014. Community-based rural residential land 1949. Acta Geogr. Sinica 69 (4), 435–447 in Chinese. consolidation and allocation can help to revitalize hollowed villages in traditional Sun, Y., Liu, Y.Z., 2010. Evaluation of land use sustainability based on land fragmenta- agricultural areas of China: Evidence from Dancheng County, Henan Province. Land tion: a case on Fenyi county, Jiangxi Province. J. Nat. Res. 25 (5), 802–810 in use Pol. 39, 188–198. Chinese. Liu, J., Jin, X.B., Xu, W.Y., Fan, Y.T., Ren, J., Zhang, X.L., Zhou, Y.K., 2019. Spatial Sun, Y.J., Ren, Z.Y., Zhao, S.N., Zhang, J., 2017. Spatial and temporal changing analysis couplingdifferentiation and development zoning trade-off ; of land space utilization of synergy and trade-off between ecosystem services in valley basins of Shaanxi efficiency in eastern China. Land use Pol. 85, 310–327. Province. Acta Geogr. Sinica 72 (13), 521–532 in Chinese. Liu, L.M., 2010. Land Resource Science. China Agriculture Press, Beijing. Sundqvist, P., Lisa, A., 2006. A Study of The Impacts of Land Fragmentation on Liu, Y.Q., Su, C.G., Long, H.L., Hou, X.G., 2013. The institutional innovation of rural land Agricultural Productivity in Northern Vietnam. Bachelor Thesis. Department of management in the context of urban-rural integration development. Econ. Geogr. 33 Economics, Uppsala University, Sweden Autumn. (10), 138–144 in Chinese. Wan, G.H., Cheng, E.J., 2001. Effects of land fragmentation and returns to scale in the Liu, Y.S., 2019. Research on the geography of rural revitalization in the new era. Geogr. Chinese farming sector. Appl. Econ. 33 (2), 183–194. Res. 38 (3), 461–466 in Chinese. Wang, J.F., Li, X.H., Christakos, G., Liao, L.Y., Zhang, T., Gu, X., Zheng, X.Y., 2010. Liu, Y.S., 2018a. Research on the urban-rural integration and rural revitalization in the Geographical Detectors-Based health risk assessment and its application in the neural new era in China. Acta Geogr. Sinica 73 (4), 637–650 in Chinese. tube defects study of the heshun region. China. Int. J. Geogr. Inf. Sci. 24 (1), 107–127. Liu, Y.S., 2018b. Introduction to land use and rural sustainability in China. Land use Pol. Yang, R., Liu, Y.S., Long, H.L., 2015. The study on non-agricultural transformation co- 74, 1–4. evolution characteristics of population-land-industry": Case study of the Bohai Rim in Liu, Y.S., Li, J.T., Yang, Y.Y., 2018. Strategic adjustment of land use policy under the China. Geogr. Res. 34 (3), 475–486 in Chinese. economic transformation. Land use Pol. 74, 5–14. Yang, Y.Y., Liu, Y.S., Li, Y.R., Du, G.M., 2018. Quantifying spatio-temporal patterns of Liu, Y.S., Li, Y.H., 2017. Revitalize the world’s countryside. Nature 548, 275–277. urban expansion in Beijing during 1985-2013 with rural-urban development trans- Liu, Y.S., Fang, F., Li, Y., 2014a. Key issues of land use in china and implications for policy formation. Land use Pol. 74, 220–230. making. Land use Pol. 40, 6–12. Ye, J., Xie, Q.Q., Tan, N.Y., 2017. National land spatial pattern distribution method based Liu, Y.S., Long, H.L., Wang, J.Y., et al., 2014b. Agricultural Modernization and Farmers in on ecological carrying capacity. Trans. Chin. Soc. Agric. Eng. (Trans. CSAE) 33 (11), China. Science Press, Beijing 2014. 262–271 in Chinese. Long, H.L., 2014. Land consolidation: an indispensable way of spatial restructuring in Zang, L.Z., Araral, E., Wang, Y.H., 2019. Effects of land fragmentation on the governance rural China. J. Geogr. Sci. 24 (2), 211–225. of the commons: theory and evidence from 284 villages and 17 provinces in China. Long, H.L., 2015. Land use transition and land management. Geogr. Res. 34 (9), Land use Pol. 82, 518–527. 1607–1618 in Chinese. Zhang, L.X., Huang, J.K., Rozelle, S., Cooperat, O.E., Dev, 1997. Land Policy and Land Use Long, H.L., Tu, S.S., Ge, D.Z., Li, T.T., Liu, Y.S., 2016. The allocation and management of in China. Organization Economic Cooperation & Development, pp. 71–77. critical resources in rural China under restructuring: problems and prospects. J. Rural Zhang, Y.N., Long, H.L., Ma, L., Ge, D.Z., Tu, S.S., Qu, Y., 2018. Farmland function Stud. 47 (47), 392–412. evolution in the Huang-Huai-Hai Plain: Processes, patterns and mechanisms. J. Long, H.L., Zou, J., Li, T.T., Liu, Y.S., 2012. Study on the characteristics and terrirorial Geogr. Sci. 28 (06), 759–777. types of rural transformation development: the case of “Southern Jiangsu-Northern Zhou, Y., Guo, L.M., Liu, Y.S., 2019. Land consolidation boosting poverty alleviation in Shaanxi” transect. Geogr. Res. 31 (3), 495–506 in Chinese. China: Theory and practice. Land use Pol. 82, 339–348. Looga, J., Jürgenson, E., Sikk, K., Matveev, E., Maasikamäe, S., 2018. Land fragmentation Zhuang, D.F., Liu, J.Y., 1997. Study on the model of regional differentiation of land use and other determinants of agricultural farm productivity: The case of Estonia. Land degree in China. J. Nat. Res. 12 (2), 10–16 in Chinese.

19