An analysis of the spatial evolution and influencing factors of rural settlements along the Shandong section of the Grand Canal of China

Since 1980s, the weakening of the transportation function of the Grand Canal and rapid urbanisation have generated significant changes in the spatial pattern of rural settlements along the Grand Canal. Analysing the changing characteristics of this spatial pattern, and exploring the natural, social, and economic factors influencing these changes are key to understanding the regional spatial structure and development law, and grasping the degree of influence of these factors. This study selected approximately 13,000 rural settlements in 27 county‐level units in the Shandong section of the Grand Canal, from 1980 to 2018, as the research object. The Grand Canal's distance buffer zone is divided into canal‐side, near‐canal, far‐canal, and further‐canal settlements. The correlation model with the canal is constructed through the controlled experiment. The distribution, scale, and form of the settlement are analysed quantitatively by applying the change of gravity centre model (CGC), average nearest neighbour analysis (ANN), landscape metrics (LM), and other methods. The quantitative analysis of geographic detectors in spatial pattern differentiation factors shows the relative importance and interaction within them. This study indicates that the spatial distribution shows ‘large dispersion and small concentration’ and ‘small‐scale agglomeration’ characteristics. The larger the scale of the settlement expands, the more the number of settlements decreases. The closer the settlement nears the river, the more complex the shape clusters. The total output value of agriculture, forestry, animal husbandry, fisheries, the total power of agricultural machinery, and the effective irrigation area are the main influencing factors. The grain area and the per capita disposable income of rural residents are the main auxiliary influencing factors.


| INTRODUCTION
Since the Reform and Opening, the social economy has developed rapidly, and people's living standards have improved. With rapid urbanisation, there was a noticeable difference in the spatial distribution between urban and rural structures (Tian, Kong, Liu, & Wang, 2016;M. M. Zhang, Chen, Cai, & Li, 2019). The development of rural areas was showing slow growth under the double pressure of urban development expansion and its development limit (Bao & Fang, 2012). The distribution and form of rural settlements have also changed dramatically (Tu, Long, Zhang, & Qu, 2017). Since 2010, rural development has been gradually emphasised by introducing China's new urbanisation and rural revitalisation strategy. Achieving efficient and sustainable rural development, especially the development of rural settlements in specific areas, has become a new research focus.
As a world cultural heritage, the Grand Canal is the crystallisation of the wisdom of ancient Chinese civilisation. As the main waterway running from north to south, it has driven the rise and prosperity of settlements along the route in Chinese history (Y. R. . However, with the completion of China's first railroad from north to south in the 1980s, the transport function of the canal was primarily replaced, leading to the lagging development of the canal and the settlements along its route. As a unique part of China's cultural landscape, the government attaches great importance to the development of the canal and the settlements and has promulgated the Outline for the Protection and Utilisation of the Grand Canal Culture (2019). Shandong Province has put forward documents accordingly, such as the Implementation Plan for the Protection, Heritage, and Utilisation of the Grand Canal Culture in Shandong Province (2020). Also, the study of the settlements along the canal is highly valued, particularly regarding the rural settlements. As a rural population settlement influenced by the natural environment, social economy, and policies, it has been given a special canal imprint under the unique geographical location, bearing the social production life of rural residents (Long, Zhang, & Tu, 2019). The evolution and development of their spatial patterns have witnessed human-land relations along the canal and the formation of their heterogeneity since the Reform and Opening. The evolution and development have also explored the characteristics of rural settlements' location, scale, distribution, morphology, and other differentiations, and their influencing factors and mechanisms. Through the research, fully understanding the dynamic evolution of the countryside for the merging points, village relocation and other relocating issues in the future, and issues of resource sharing public facilities allocation are possible. It helps to identify the dominant and auxiliary factors affecting the development of villages. It also helps to find the optimal entry point for industrial layout, structure optimisation, and spatial layout in village planning. It is of great theoretical value and practical significance to maximise the inputbenefit ratio of rural development, optimise the spatial pattern of rural settlements, promote sustainable rural development, and revive canal settlements (Y. S. Liu, Fang, & Li, 2014).
Hence, the authors selected 27 county-level rural settlements in the Shandong section of the Grand Canal as the research object. The authors innovatively made divisions according to four levels: canalside zonal settlement (<1 km), near-canal zonal settlement (1-4 km), far-canal zonal settlement (4-10 km), and further-canal zonal settlement (>10 km). Based on land use data and socioeconomic data in 1980, 1990, 2000, 2010, and 2018, the authors applied the CGC model, kernel density, ANN analysis, Getis Ord Gi*, and land metrics to analyse the settlement patches in rural areas quantitatively. Based on the quantitative analysis, the authors studied the spatial characteristics of settlements' size, distribution, and morphology. The paper also analysed the dominant and auxiliary factors influencing differentiation from both natural and socioeconomic perspectives.
The study has provided theoretical references for the development of rural settlements along the canal, including: 1. Defining the characteristics of scale and the regular pattern of focus-shift, that becomes a guiding role in the degree control and migration site selection in village planning; 2. Defining the characteristics of distribution, which provides references for sharing resources and managing public facilities in regional planning; 3. Clarifying the characteristics of morphology, that assists to find the development direction of rural settlements along the canal according to local conditions.

| LITERATURE REVIEW
In recent years, many scholars have studied the distribution and evolution of the settlements along the Grand Canal and the rural settlements in China, which have made rich achievements. Many have further explored the spatial pattern and morphology of the canal's settlements, primarily relying on historical excavation and spatial characterisation methods. In the overall spatial pattern of canals, a three-level structure of city-town-village is commonly seen. However, there are regional adaptations and differences in spatial layout, morphological patterns, functional characteristics, and cultural attributes of settlements along the Grand Canal (Z. Y. Wen, Zhang, Zhu, Xie, & Xing, 2020). Since 1984, urban settlement distribution has tended to be discrete and dispersed (Shi & Huang, 2019). Since the Reform and Opening, the spatial pattern of towns along the canal has changed from closed and monotonous to open and diversified (Niu, 2012). Villages along the canal are distributed in a regular and skewed matrix with natural distribution (M. Zhang & Tang, 2018), mainly influenced by natural geographical location, socioeconomic policies, and canal functions (C. Li, 2007; C. Zhang, 2019a;Y. Y. Zhang, 2019b). However, under the influence of towns and roads, the rural settlements, as essential elements of the Grand Canal, are mainly driven by agricultural production and road networks, indicating that rural settlement has changed from natural to sudden growth (C. B. Wen, 2019;Zhao, Dong, & Wang, 2021).
Studies on rural settlements originated and focused on qualitative research in Europe. Since the revolution of econometric geography in the West, scholars have applied the cellular automation (CA) model, Voronoi diagram, binary model, and fractal theory to quantitative studies of the spatial evolutionary trends of rural settlements and their influencing factors. Researchers urgently requested spatial data in the fields, especially for applying probability, computational geometry, and geographic information science, that could establish standardisation and be more objective than qualitative analysis (Atsuyuki, Barry, Kokichi, Nok, & Kendall, 2000;Duyckaerts & Godefroy, 2000).
Studies on rural settlements in China have been more fruitful in recent years (Tan, Zhang, Wang, & Ni, 2021;Yang, Song, & Xue, 2020). The evolution of Chinese rural settlements follows a life cycle of formation, development, stabilisation, decline, and revival .
The scale, pattern, and morphology of rural settlements reflect the evolution of nature, socioeconomic policy changes, and human-land relations in current society (G. R. Yao & Xie, 2016). Furthermore, the studies have found that the distribution pattern of Chinese villages is the result of the interaction of environmental, geographical, social, and economic factors (Zhou, Huang, & Liu, 2020), with rapid population growth and accelerated economic growth being the main reasons for changes in rural settlements. The continuation of the rural land system and changes in family structure have also influenced the evolution of rural settlements (H. H. . In the research regarding the evolution of rural settlements' spatial pattern and its influencing factors, a large number of scholars have used spatial analysis methods, geographically weighted regression models, and geographic detectors to investigate rural settlements in suburban counties (C. C. Liu, 2020); traditional settlements in Youshui Basin (Du & Guo, 2020); rural settlements in Hexi, Gansu (R. Y. Li, 2020); rural settlements in wetland landscapes (K. Y. ; rural settlements on urban islands (Cui & Yan, 2020); rural settlements in the poverty belt of Beijing, Tianjin, and Hebei (Tao, Zhang, Xu, Zhang, & Zhang, 2020); rural valley settlements (Xu & Zhao, 2020); rural settlements in near urban areas (Chen, 2020), Dingziwan area in Jiaodong (J. B. Zhang, 2020); villages in poor areas in southwest China (L. Yao, Li, Chen, Wang, & Peng, 2020); oasis villages in Xinjiang (Lin, Lei, Wu, Yang, & Li, 2020); and various other types of rural settlements in terms of size, density, and shape. Scholars have analysed the evolution of size, density and shape, and the relationship between their evolution and indicators such as slope elevation, agricultural technology, non-agricultural population, industrial output value, and net income per capita. Researchers have found that the natural environment (Chen, 2020), socioeconomic, and cultural factors (Gulizre et al., 2020) influence the evolutionary diversity of rural settlements. However, the dominant factors have significant regional differences (Long et al., 2019). For example, the spatial distribution of rural settlements in Xinjiang Oasis is influenced by road accessibility at the townscape and county level, the road's slope, the distance to rivers, temperature, elevation, but insignificantly by social and economic factors. Land relief, average annual temperature, and average annual precipitation are the main natural forces within the Loess Plateau region. Equally, food production, total population, and proportion of the rural population are the critical humanistic driving forces. However, a massive change in the waterway caused a water shortage in 1855, seriously affecting the regular operation of water transport. The delayed canal dredging disrupted grain transport towards the end of the Qing dynasty, and the associated management agencies and personnel also disappeared due to diplomatic difficulties. In addition to some sections of the canal, it also began to dry up. However, with its rich historical connotation, crucial geographical location, and unique human, the Grand Canal still occupies a significant position on China's regional cultural map (C. Zhang, 2019a;Y. Y. Zhang, 2019b).

| Defining rural settlement
X and Y are the longitude and latitude of the centre of gravity of the rural settlement patch in the study area in the data year. X i and Y i are the longitude and latitude of the rural settlement patch n, respectively, and a i is the patch's area.

| Kernel density analysis
Kernel density analysis expresses the continuity of discrete spatial data and analyse the settlements' spatial distribution density (Tan et al., 2021). Thus, it directly reflects regional differences in the distribution of rural settlements in different regions.

| Spatial hot spot detection
Spatial hot spot detection uses the Getis-Ord Gi* statistics in ArcGIS 10 to identify statistically significant hot and cold spots. It is used to test whether there are statistically significant high and low values in some areas (Tan et al., 2021;M. M. Zhang et al., 2019). The area visualisation method can be used to identify hot spots and cold spots.
where w ij (d) is the spatial weight defined by the distance rule, and x i and x j are the variable values of the i and j regions, respectively.

| Landscape metrics
This study uses Fragstats4.2 to calculate various indices and selects the average patch area according to three levels of scale, agglomeration, and shape characteristics of the rural settlement space (R. Y. Li, 2020). The mean patch size (MPS) and patch size standard deviation (PSSD) were selected to quantify the scale characteristics of rural settlements, whereas the aggregation index (AI) and patch density (PD) quantified their agglomeration characteristics. The Landscape shape index (LSI) and perimeter area fractal dimension (PAFRAC) quantified the shape characteristics. The main calculation formulae are as follows:  The settlement patch's total area has increased annually from 2,774.61 km 2 to 3,329.06 km 2 (Figure 4a), but the number of patches has decreased from 12,986 to 12,239. The patch area increase rate was 8.18%, which was the most significant from 2000 to 2010 (Figure 4b), and the number of patches also decreased considerably (Figure 4c). The distance between the rural settlement patch and the canal is inversely proportional to the settlement patch area's rate increase. The further the distance between the rural settlement patch and the canal, the higher the settlement patch's area increase rate (Figure 4d).
The results show that the rural settlement spatial scale is in con- The closer the rural settlements are to the canal, the earlier and more apparent the differentiation. In short, from 1980 to 2018, the rural settlement patches in the study area showed a partial merger trend, and the closer to the canal, the more evident that trend.

| Hot spot analysis
The hot spots and cold spots of the scale distribution of rural settlements from 1980 to 2018 were obtained by considering the rural settlement patch area as the statistical attribute. The results show obvious regional differentiation in the distribution of cold and hot spots in rural settlements along the canal. The hot spots are more evident in some areas than others, with cold spots relatively concentrated in southern areas. By 2010, the number of hot spots on the north and east sides of the study area had decreased, and the cold spots on the south side had also decreased. These results show that the area of corresponding cold and hot spots has been reduced, and the scale of settlement patches tends to be different (Figure 6b).

| Kernel density analysis
The kernel density analysis and search radius of the rural settlements along the canal were set to 30 km, the kernel density maps of rural settlements in 1980, 1990, 2000, 2010, and 2018  The closer the rural settlement to the canal, the lower and more concentrated the degree of settlement boundary meander.
The rural settlement boundary meander degree increased and was more dispersed 10 km away from the canal (Figure 7b). The closer the village settlement to the canal, the more complicated the shape due to its long history. The construction of the canal and the railway's geographical location also led to less human intervention.

| Identifying the influencing factors
The evolution of rural settlements is affected by multiple factors. It can be summarised into natural factors and socioeconomic factors (Wang & Hu, 2012;Zhou et al., 2020). Based on the above analysis, the availability of data, the integrity of the time series, and the independence and integrity of each index were considered , and 1990, 2000, 2010, and 2018 were selected as sample years for impact detection, and refer to related literature and selected 13 detection factors (Table 1) Li, 2020;Tan et al., 2021;Yang et al., 2020). This paper explores different factors' influence on the rural settlement space along the Grand Canal in Shandong province and clarifies the canal's influence on rural settlement space through a comparative analysis of each settlement zone.

| Geodetector quantitative analysis
From the above, due to the adaptability of spatial pattern characteristics detection and the representativeness of Canal Regional type detection, the gathering characteristics of canals have the most spatial attributes and are most closely related to canals. Therefore, the dependent variable Y is the spatial attribute data of rural settlements, and the independent variable X is the exploration influencing factor data. Obtain spatial data consistency, each village settlement unit's centroid was extracted using ArcGIS 10.8 software, and each centroid's attribute value was obtained via sampling; the exact matching of X and Y data in space was realised (Cao et al., 2019;Wang & Xu, 2017). This study aims to determine the main influencing and auxiliary influencing factors of the spatial pattern of rural settlements and their correlation characteristics with the canal.

| Single-factor detection
For the rural settlement patches in the whole settlement in the study area, the factor detection results show that, in short, economic, and industrial factors have a more noticeable impact on rural settlements, followed by policy and living standards. On the contrary, the impact of natural environmental factors was the smallest (Figure 8a). The closer the rural settlements are to the canal, the more pronounced the settlement area's growth rate and the higher the incidence of settlement mergers. The Grand Canal attracted the migration of rural settlements along its two banks from 1980 to 2018. Therefore, it is imperative for village planning to clarify the scale characteristics and the law of gravity centre migration.
2. Agglomeration characteristics of rural settlements: The spatial distribution characteristics along the Grand Canal showed an agglomeration pattern from 1980 to 2018, which was a significant lowcluster. The proportions of higher-density and highest-density areas in the overall settlements decreased: The highest-density areas were from 13.35% to 11.13%; higher-density areas were from 27.92% to 23.48%. The lower-density and lowest-density areas increased: The lowest-density areas were from 4.69% to 5.88%; the lower-density areas were from 17.74% to 21.85%. The The further the rural settlements are from the canal and the greater the number of villages per unit area, the more obvious the growth rate, and the settlements' overall connectivity increases year by year; the closer the settlements are to the canal, the higher the connectivity. The apparent distribution characteristics can provide a reference for resource sharing and public facilities layout in regional planning.
3. From 1980 to 2018, the overall boundary tortuosity and shape complexity of rural settlement patches increased. Between 2000 and 2010, the integration of rural settlements is the most obvious, and the degree of tortuosity and complexity increased significantly.
The closer the patch is to the canal, the more concentrated the dis- the proximity to the canal, the more dependent settlements are on primitive agricultural production and improvements in rural science and technology facilities. Clarify the leading factors and find the optimal focus for the planning and designation of the spatial industrial layout, such as the development of modern agriculture in the canal settlements in this article to promote rural development.
2. Double-factor effects are more substantial than single-factor effects. Rural residents' food area and per capita disposable income are the auxiliary influencing factors for the spatial differentiation of rural settlements. These factors can significantly enhance the impact of index factors on the spatial pattern evolution of rural settlements. In addition, the rural working population, the total output value of agriculture, forestry, animal husbandry and fishery, general public budget expenditures, and an adequate irrigation area also have a particular supporting role. Clarify the auxiliary influence factors, find the optimal auxiliary development point, and promote the development of the settlement at multiple aspects.

| Discussions
This study adopts various analysis methods to study the spatial- the spatial-temporal evolution characteristics and its leading factors of the area along the Grand Canal.
The changes are related to the transformation of rural areas to urban areas in the context of urbanisation and industrialisation after China's Reform and Opening (Bao & Fang, 2012;Tian et al., 2016).
Then, the impact of village relocation and Poverty Alleviation Policies in Shandong Province and the overlapping of the Grand Canal and railway construction lead to the relatively backward development of the canal region. Meanwhile, the summary of characteristics and study of influential factors provide a particular reference value for the government to develop the Grand Canal settlement and sustainable rural development.
However, the research is still preliminary, only following the natural environment and socio-economic factors mentioned many times in the relevant studies. Moreover, the historical and cultural factors of the grand canal are represented and have not been quantified that need further discussion in the future. Future research needs a longer historical review to summarise the characteristics of rural settlement succession and make an appropriate prediction. Finally, the authors suggest that this study should be introduced to the specific villages along the canal, combined with the specific situation. Furthermore, it should be adopted in other canal sections to promote determination of spatial industrial structure layout, grasp the direction of relocation of villages, merging points and migration, clarify the layout of resources sharing and public facilities allocation, to prove further the significance and external effectiveness of this study in practice.