Research on the Spatial and Temporal Distribution of Logistics Enterprises in Xinjiang and the Influencing Factors Based on POI Data

Abstract

Based on the POI data of logistics enterprises in Xinjiang in 2012, 2016, and 2020, the ArcGIS spatial analysis technique, geographic detector, and other methods were used for the quantitative analysis of the spatial and temporal distributions of logistics enterprises in Xinjiang during 2012–2020 and the influencing factors. The following findings were obtained in the present study: (1) there was a significant difference in the distributions of logistics enterprises in Xinjiang at different development stages, with unbalance among areas; further, there was a higher number of logistics enterprises in Northern Xinjiang compared with Southern Xinjiang; (2) the spatial distribution of logistics enterprises in Xinjiang was generally characterized by a “northeast–southwest” trend; there was a periodic shift in the distribution center from northeast to southwest; the distribution center remained in Bayingolin Mongol Autonomous Prefecture in 2012 and 2020, and shifted to Changji Hui Autonomous Prefecture in 2016, close to the junction of the two areas; (3) the agglomeration of logistics enterprises in Xinjiang was positively correlated with the scale; the kernel density analysis results revealed that there was obvious spatial differentiation characterized by “multi-center development with core agglomeration and patch distribution at the edge”, and the hotspot areas of logistics enterprises were distributed in major cities, with small variations; the Tianshan Mountain North Slope Economic Belt was the main agglomeration area of logistics enterprises; (4) the results from the geographic detector show that the regional GDP, regional total retail sales of consumer goods, regional utilization of foreign direct investment, and regional fixed assets investment were factors that influenced the spatial distribution of logistics enterprises in Xinjiang, thereby significantly promoting the stable and rapid development of logistics enterprises.

1. Introduction

As a productive service industry, the logistics industry is closely linked with a variety of different fields and, thus, is a notable factor in economic development. Further, as an organizer and significant basis of logistics space, logistics enterprises exhibit obvious spatial selection behaviors. In optimizing the spatial distribution of regional logistics and ensuring the rational allocation of resources, the location characteristics and differentiated strategy of the distribution of logistics enterprises are of considerable significance [1,2,3]. Although both domestic and international scholars have explored the spatial distribution and influencing factors of logistics enterprises, there has been much variation in terms of the research focus and data sources [4,5,6,7]. In developed countries, logistics is a highly market-oriented, centralized industry [8,9] in which logistics enterprises are concentrated within logistics nodes, and the research focus is on logistics agglomeration and site selection [10]. The data used in such research are primarily logistics statistics based on postal code and business survey data [11,12]. However, owing to the late start but rapid development of logistics enterprises, the focus of existing research in China has largely been on the location characteristics, influencing factors, and formation mechanisms [13,14,15,16]. In terms of scale, the research ranges from country [16,17] to urban agglomerations [14,18], provinces [19], and municipalities [20]. The research data are derived from government websites, enterprise network registration, and industrial and commercial registration, as well as other sources, with limited sample sizes and relatively complicated data acquisition and processing [21]. The spatial distribution of logistics enterprises is primarily explored using GIS technology, and the common methods adopted in prior research have included kernel density analysis, buffer analysis, the nearest neighbor index, spatial autocorrelation, and exploratory spatial data analysis [4,22,23,24]. Regarding the influencing factors of the spatial distribution of logistics enterprises, scholars have largely used multiple regression models [25], geographic weighted regression [26], and geographic detectors [27] to analyze the main influencing factors, such as the economic development level, traffic location conditions, the opening up of China, and policy support [28]. Through the open-source acquisition of Point of Interest (POI) data, Chinese scholars can adopt a new means [1,29] for investigating the spatial dynamics of logistics enterprises [14,18]. Each POI package contains information such as the name, category, coordinates, classification, and other pieces of information. Additionally, the open-source acquisition of POI data is characterized by the advantages of a sufficient data volume, rich and accurate information, a high degree of timeliness, and relatively easy acquisition. In addition to social and economic functions, such acquisition also incorporates the dual attributes of time and space, being widely used in the study of the spatial distribution of urban and rural facilities [29,30].

From the perspective of research methods, with the increasing maturity of GIS technology, the results of using kernel density estimation, standard deviation ellipse analysis, and other methods to study the spatial pattern evolution characteristics of logistics enterprises are increasing, which can effectively reflect the overall spatial structure characteristics of geographic elements and have been better applied; however, most of the research on influencing factors is mainly based on qualitative descriptive analysis, and the application of quantitative models is still relatively lacking; few scholars have used geodetector models in the analysis of the spatial distribution patterns and factors of logistics enterprises in Xinjiang. Few scholars have used the geographic probe model in the analysis of the spatial distribution pattern and factors of logistics enterprises in Xinjiang for empirical comparison and analysis. Therefore, we attempt to use the geodetector method to quantitatively analyze the influencing factors of their spatial differentiation and their interaction, which solves the limitations of the traditional method, in order to address the shortage of research on the spatial distribution pattern of logistics enterprises in Xinjiang. Most studies on the spatial distribution of logistics enterprises have involved the central and eastern regions of China, which have relatively developed economies. Meanwhile, the northwestern region has not received adequate attention. With a unique geographical location, Xinjiang borders numerous countries and has a large number of ports, enjoying conditions that are not available in other western provinces. Notably, the influence of land border ports on the distribution of logistics enterprises has also been ignored in other related studies. The aim of the present study was to analyze the changes in the spatial distribution of logistics enterprises in Xinjiang based on POI data, and to determine the potential factors that cause such changes. In the present study, a reference for the adjustment of logistics functions and layout optimization is provided, as well as a comparison of the distribution characteristics of logistics enterprises in other areas and supplemental empirical analysis of the location theory on logistics activities, which can facilitate further research and improvements in logistics location theory.

2. Materials and Methods

2.1. Studied Area

The Xinjiang Uygur Autonomous Region (hereinafter referred to as “Xinjiang”) is situated in Northwest China between 73°40′ E−96°23′ E and 34°25′ N−49°10′ N, with an area of 166 km², being the largest provincial administrative region in China. Xinjiang borders eight countries, including Russia, with a total land border of over 5600 km. Xinjiang is characterized by an undulating terrain with the Altai Mountains, the Junggar Basin, the Tianshan Mountains, the Tarim Basin, and the Kunlun Mountains, as shown in Figure 1.

In the present study, the research scope refers to the 14 areas, prefectures, and cities under the jurisdiction of Xinjiang as of 31 December 2020. Among such areas, the research data of Ili only included the data of the administrative districts at county level under the jurisdiction of Ili, excluding the data of Tacheng and Altay. As shown in

Table 1. List of cities in the studied area.

Area	Abbreviation	Area	Abbreviation	Area	Abbreviation
Urumqi City	Urumqi	Altay Prefecture	Altay	Tumushuk City	Tumushuk
Karamay City	Karamay	Changji Hui Autonomous Prefecture	Changji	Wujiaqu City	Wujiaqu
Turpan City	Turpan	Bortala Mongol Autonomous Prefecture	Bortala	Beitun City	Beitun
Hami City	Hami	Bayingolin Mongol Autonomous Prefecture	Bayingolin	Tiemenguan City	Tiemenguan
Akesu Prefecture	Akesu	Kizilsu Kirghiz Autonomous Prefecture	Kizilsu Kirghiz	Shuanghe City	Shuanghe
Kashi Prefecture	Kashi	Ili Kazakh Autonomous Prefecture	Ili	Kokdala City	Kokdala
Hotan Prefecture	Hotan	Shihezi City	Shihezi	Kunyu City	Kunyi
Tacheng Prefecturer	Tacheng	Alar City	Alar

, the data of 9 county-level cities directly under the autonomous region managed by the Xinjiang Production and Construction Corps were further included, so as to accurately identify the level of the logistics industry.

2.2. Data Sources and Processing

Based on the principles of data availability and typicality, as well as the principle of equal distribution, the data of 2012, 2016, and 2020 were selected to analyze the spatial and temporal distribution pattern of the Xinjiang logistics industry and the influencing factors. The POI data of logistics were mainly obtained from the AutoNavi map open platform in China in 2012, 2016, and 2020. The present study was based on the Application Programming Interface (API) interface of AutoNavi, and Python was used to search for the following terms: “logistics”, “freight”, “transportation”, and “warehousing”. The POI attributes of logistics included the name, type, address, and latitude and longitude coordinates of logistics enterprises. To ensure the integrity, accuracy, and validity of the POI data of logistics enterprises, EXCEL was used for the deletion of duplicate records and item-by-item data cleansing and review. Statistical Product and Service Solutions was used for the Pearson correlation analysis of the factors and quantifiable indicators that affect logistics enterprises.

Data Description

At present, in Chinese academic circles, there is no general consensus on the definition and classification system of logistics enterprises and logistics nodes. Despite the absence of a general consensus, there are classifications and definitions of various logistics indicators in the Standard of the People’s Republic of China, e.g., Logistics Terminology (GB/T 18354-2006) and the Classification and Evaluation Indicators for Logistics Enterprises of the People’s Republic of China (GBT 19680-2013), as well as a grading standard and classification methods for transportation and logistics enterprises in academic circles and the international community. Through the classifications and definitions provided, logistics enterprises can be classified into ordinary logistics enterprises, which mainly provide the conventional storage and transportation of large and medium-sized goods for enterprises, and express enterprises, which primarily deliver medium and small articles for individuals. In the present study, ordinary logistics enterprises rather than express enterprises were investigated.

The POI data covered logistics enterprises for which the name contained “logistics”, “freight”, “transportation”, “warehousing”, or other relevant terms, while express enterprises such as SF Express, STO Express, and Yunda Express were excluded.

Excluded logistics nodes: logistics parks, bases, logistics centers, delivery and distribution centers, freight yards; excluded logistics-related facilities: bonded areas or places under special customs supervision, airports, ports, railway stations, bus stations, transportation places, industrial parks, and comprehensive markets.

The social and economic data were mainly obtained from the Xinjiang Statistical Yearbook (2013–2021), China Urban Statistical Yearbook (2013–2021), and China Regional Economic Statistical Yearbook (2013–2021). With regard to specific missing data, the historical Statistical Bulletins of National Economy and Social Development were used for supplementary description. In addition, the vector boundaries of the Xinjiang administrative region were obtained from the latest database of the National Geomatics Center of China: https://www.ngcc.cn, accessed on 1 January 2021.

2.3. Research Method

2.3.1. Standard Deviation Ellipse

By calculating the semi-minor axis, semi-major axis, azimuth, and other parameters, standard deviation ellipses were formed to identify the trends of changes such as the spatial distribution center and expansion direction of logistics in Xinjiang, thereby revealing the spatial distribution patterns.

The area of the ellipse represents the main distribution range of logistics enterprises; the average center represents the relative position of distribution of logistics enterprises; the azimuth reflects the main trend direction of the distribution of logistics enterprises; and the long axis of the ellipse represents the degree of dispersion of logistics enterprises in Xinjiang in the main trend direction [31]. The calculation formula is as follows.

Average center:

$$SD{E}_x=\sqrt{\frac{{\displaystyle \sum _{i=1}^n{(X_i-\overline{X})}^2}}{n}}$$

(1)

$$SD{E}_y=\sqrt{\frac{{\displaystyle \sum _{i=1}^n{(Y_i-\overline{Y})}^2}}{n}}$$

(2)

X-axis standard deviation:

$${\sigma }_x=\sqrt{2}\sqrt{\frac{{\displaystyle \sum _{i=1}^n{({\tilde{x}}_i\cos \theta -{\tilde{y}}_i\sin \theta )}^2}}{2}}$$

(3)

Y-axis standard deviation:

$${\sigma }_y=\sqrt{2}\sqrt{\frac{{\displaystyle \sum _{i=1}^n{({\tilde{x}}_i\sin \theta +{\tilde{y}}_i\cos \theta )}^2}}{2}}$$

(4)

Here, ($X_i,Y_i$) are the spatial locations of logistics enterprises; $SD{E}_x$, $SD{E}_y$ are the distribution centers of logistics enterprises; ${\sigma }_x$ and ${\sigma }_y$ are the standard deviations of the major axis and minor axis of the ellipse, which reflect the spatial distribution trend and dispersion degree of logistics enterprises.

2.3.2. Average Nearest Neighbor Index

The actual average distance between logistics enterprises in Xinjiang was calculated using the Euclidean formula and compared with the expected average distance under the assumed random distribution mode, thereby allowing the distribution pattern of logistics enterprises within the area to be determined [32]. The calculation formula is

$$NNI=ND/PD=2\sqrt{A}{\displaystyle \sum _{i=1}^n{d}_i}$$

(5)

where $d_i$ is the distance from the point to the nearest neighbor point; $n$ is the number of logistics enterprises; $A$ is the area of the studied area; ND is the actual average distance; and PD is the expected average distance. If the nearest neighbor index (NNI) was greater than 1, the performance mode tended to be discrete; if the nearest neighbor index (NNI) was less than 1, the performance mode tended to be agglomerated. The smaller the nearest neighbor index (NNI), the higher the degree of agglomeration of logistics enterprises.

2.3.3. Kernel Density Analysis

Despite reflecting the distribution of geographical elements from the perspective of mathematical statistics, the nearest neighbor index cannot intuitively reflect the actual spatial distribution of elements. In overcoming such drawbacks, kernel density analysis can clearly reflect the spatial distribution and agglomeration characteristics of geographic elements. In kernel density analysis, with each grid point as the center, the spotty geographical elements within a specific radius are searched and counted, whereby the density of each grid point is obtained. The kernel density can better reflect the influence of a kernel on the surrounding area, with a higher kernel density indicating a denser distribution of sample points [33]. The calculation formula is as follows:

$$f\left(x\right)=1/nh{\displaystyle \sum _{i=1}^{n}k\left\{\left(x-x_i\right)/h\right\}}$$

(6)

where $f\left(x\right)$ is the estimated kernel density; $n$ is the number of farmhouses; ${\displaystyle \sum _{i=1}^{n}k\left\{\left(x-x_i\right)/h\right\}}$ is the kernel function; $h$> 0 is the bandwidth; and $\left(x-x_i\right)$ is the distance from the estimated value point $x$ to $x_i$.

2.3.4. Geographical Detector

A geographical detector was developed by Wang Jinfeng and other researchers of the Chinese Academy of Sciences [34], and has been extensively adopted to explore the characteristics and driving factors of spatial differentiation. The core concept of the geographical detector is that type quantity and numerical quantity, as two parameters that have significant influences on numerical quantity, share similarity in spatial distribution. The detector mainly consists of four modules: detection differentiation and factors, interaction, risk area, and ecology. In the present study, to identify the main driving factors of the spatial differentiation of logistics enterprises in Xinjiang and the interaction among the factors, the factor detection and interaction detection modules in the geographic detector model were used [35]. The model is as follows:

$$q=1-\frac{1}{N{\sigma }^2}{\displaystyle {\sum }_{h=1}^L{N}_h}{\sigma }_h^2$$

(7)

where N and ${\sigma }^2$ are the number of sample units and variance in the entire region, respectively; $N_h$ and ${\sigma }_h^2$ are the sample size and variance of category h influencing factors; L is number of classes of category h influencing factors; q is the size of the detection index, and the larger the value, the greater the influence of the detected factor. In the present study, ArcGIS 10.8 and the natural breakpoint method were adopted to convert driving factors into type variables; subsequently, the geographic detector model was used to measure the driving force for the development of logistics enterprises in Xinjiang.

3. Results

3.1. Research on Time-Series Distribution of Logistics Enterprises in Xinjiang

According to the comparison of the number of logistics enterprises in various regions of Xinjiang in Figure 2, the spatial distribution of logistics enterprises was uneven among a number of areas.

In terms of geographical distribution, the number of logistics enterprises should be evenly distributed in each city in space, but, in terms of supply and demand, logistics serves economic development, and the number of logistics enterprises in each city matches the economic and living needs. By comparing the number of POI of logistics enterprises in each city of Xinjiang in three time cross-sections, it can be observed that the spatial distribution of the logistics industry in Xinjiang has been unbalanced. The development momentum of logistics enterprises was largely concentrated in the provincial capital of Urumqi, Changji Hui Autonomous Prefecture, Bayingolin Mongol Autonomous Prefecture, Ili Kazakh Autonomous Prefecture, Akesu Prefecture, and Kashgar Prefecture. Kizilsu Kirghiz Autonomous Prefecture is an area with typical “low-lying land”. The increase in the number and size of local enterprises varied from place to place. From 2012 to 2020, Bayingolin Mongol Autonomous Prefecture, Kashgar Region, Shihezi City, Wujiaqu City, Shuanghe City, and Kokdala City maintained a positive development trend, while other prefectures and cities exhibited a decline after an increase.

3.2. Analysis of the Spatial Evolution Characteristics of Logistics Enterprises in Xinjiang

3.2.1. Analysis of the Standard Deviation Ellipses

ArcGIS was used to analyze the standard deviation ellipses of logistics enterprises in Xinjiang in 2012, 2016, and 2020, as shown in

Table 2. Parameters of standard deviation ellipses of logistics enterprises in Xinjiang.

Year	Length of Semi-Major Axis (km)	Length of Semi-Minor Axis (km)	Longitude of Central Point (E)	Latitude of Central Point (N)	Corner (θ)
2012	697.89	330.16	85.55	43.32	55.39
2016	579.93	270.92	86.24	43.44	65.68
2020	763.08	332.49	84.50	42.90	58.91

and Figure 3.

The center of the standard deviation ellipse represents the distribution center of logistics enterprises in Xinjiang; the distribution direction of logistics enterprises is represented by the long semi-axis of the ellipse, and the distribution range is represented by the short semi-axis of the ellipse. An observation can be made that the greater the oblateness, the more obvious the directionality of the spatial distribution of logistics enterprises in Xinjiang.

The standard deviation ellipses mainly covered Urumqi, Western Turpan, Changji, Shihezi, Fukang, Wujiaqu, Karamay, and Ili in Northern Xinjiang and partial areas of Bayingolin and Akesu in Southern Xinjiang. Regarding the distribution of logistics enterprises in Xinjiang, the long central axis was roughly within the vicinity of the line “Changji–Urumqi–Bayingolin–Akesu”, while the short central axis was roughly within the vicinity of the line “Bayingolin–Karamay–Tacheng”. The corner θ of the standard deviation ellipses varied between 55.39° and 65.68°, with certain fluctuations. The oblateness of the ellipses continued to increase, indicating an overall obvious “northeast–southwest” distribution of logistics enterprises in Xinjiang.

Specifically, the rotation angle θ between the major axis and the Y axis of the ellipse in 2016 increased from 55.39° in 2012 to 65.68°, with a reduced length of the major axis, reduced length of the minor axis, and increased oblateness, indicating that there was a shrinkage in the standard deviation ellipse in the “northeast–southwest” direction with the rapid increase and greater agglomeration of logistics industries within the ellipses.

In 2020, the rotation angle θ between the major axis and the Y axis of the ellipse of enterprises in Xinjiang was 58.91°. The difference between the major and minor axes of the ellipse was the largest—that is, the oblateness was the largest, indicating the strong centripetal force and obvious directionality of logistics enterprises in Xinjiang and the obvious “northeast–southwest” spatial distribution. In terms of the distribution center, the center of the ellipse remained in Bayingolin in 2012 and 2020, and shifted to Changji near the junction in 2016, without any significant change. The coordinates of the center point first shifted slightly from 85.55° E and 43.32° N in 2012 to 86.24° E and 43.44° N in 2016 in the northeast, and then shifted southwestward to 84.50° E and 42.90° N in 2020. In general, characterized by a trajectory “from northeast to southwest”, the position of the distribution center varied at different stages and there was an overall trend of shifting to the southwest. Since logistics enterprises are factors of economic development, the shifting direction of the respective geographical center was roughly the same. An observation can be made that with the economic development of Ili and Kashgar in the southwest, the logistics center was also shifted.

3.2.2. Analysis of Changes in Characteristics of Spatial Agglomeration

In the present study, the average nearest neighbor index was used to measure the agglomeration of logistics enterprises in Xinjiang, as shown in

Table 3. The nearest neighbor index of logistics enterprises in Xinjiang.

Year	Actual Nearest Neighbor Distance (km)	Theoretical Nearest Neighbor Distance (km)	Nearest Neighbor Index (NNI)	Z-Test Index	Distribution Type
2012	2.067	23.081	0.089	−66.595	Agglomeration
2016	1.117	12.679	0.076	−100.573	Agglomeration
2020	1.614	16.885	0.096	−85.622	Agglomeration

.

With respect to the interpretation of the nearest neighbor ratio, if the nearest neighbor index (NNI) was greater than 1, the spatial distribution of logistics enterprises was random; if the NNI was less than 1, the spatial distribution of logistics enterprises was agglomerated; if the NNI was equal to 1, the spatial distribution of logistics enterprises was uniformly discrete. The spatial distribution pattern could be observed through the significance of the Z-test results. If Z was less than −2.58, the spatial distribution was agglomerated with a 99% confidence level. If Z was greater than −2.58, the spatial distribution was uniform with a 99% confidence level.

An observation can be made from the nearest neighbor analysis of logistics enterprises that the actual nearest neighbor distance of logistics enterprises was smaller than the theoretical nearest neighbor distance from 2012 to 2020. At the three time nodes of 2012, 2016, and 2020, the average nearest neighbor indexes were 0.089, 0.076, and 0.096, respectively, which were all less than 1; and Z was also less than −2.58, indicating that the spatial distribution of logistics enterprises in Xinjiang was agglomerated—that is, there was agglomerated distribution. In 2012, the nearest neighbor index was 0.089, and the spatial distribution exhibited a trend of agglomeration. In 2016, the nearest neighbor index slightly declined to 0.076, and the spatial layout exhibited weak agglomeration. In 2020, the nearest neighbor index rebounded again to reach 0.096, exhibiting a trend of significant and consistently strong agglomeration. The agglomeration of logistics enterprises exhibited a significant U-shaped trend. The agglomeration of logistics enterprises was the highest in 2020, and the lowest in 2016.

3.2.3. Kernel Density Analysis

The overall spatial distribution trend and development characteristics of logistics enterprises were further explored. An observation can be made from the kernel density map in Figure 4 that logistics enterprises in Xinjiang demonstrated overall obvious spatial agglomeration.

According to further investigation into the spatial distribution of logistics enterprises in Xinjiang in 2012, logistics enterprises in Northern Xinjiang demonstrated obvious agglomeration, with high density in patches, which consisted of Urumqi as the main core and several surrounding areas. The sub-core of agglomeration in Northern Xinjiang included Changji, Hami, Karamay, Ili, Bortala, and others, which formed a large annular sub-core area surrounding the main core of Urumqi; a sub-core agglomeration of logistics enterprises was formed in Southern Xinjiang, consisting of Bayingolin, Akesu, Kashgar, and Hotan, which were relatively discrete. In addition to the core and sub-core areas, there were several relatively obvious spatial distribution agglomeration centers, such as the two agglomeration centers in Turpan and Tacheng. In the remaining areas, logistics enterprises were characterized by an isolated, spotty distribution. In 2016, the overall agglomeration of logistics enterprises in Xinjiang took the form of a belt-like area, with Urumqi as the main core, based on agglomeration in 2012; the kernel densities of the sub-cores of Hami, Changji, Karamay, Ili, Bayingolin, Akesu, and Kashgar were slightly reduced, and Urumqi was still the region with the highest density of logistics enterprises. In terms of the spatial distribution characteristics of logistics enterprises in Xinjiang in 2020, the main core mostly consisted of areas around Urumqi, which formed multiple sub-core areas, such as Hami, Changji, Shihezi, Karamay, Ili, Bayingolin, Akesu, and Kashgar; the regional agglomeration was most prominent in Northern Xinjiang, where Urumqi witnessed strengthened agglomeration with the highest density of logistics enterprises; as well as the core and sub-core areas, Tacheng, Bortala, and Hotan had become agglomeration centers, and, in the remaining areas, logistics enterprises were characterized by an isolated, spotty distribution.

In general, the spatial distribution of logistics enterprises in Xinjiang was generally characterized by “multi-center development with core agglomeration and patch distribution at the edge”. In Northern Xinjiang, the distribution of logistics enterprises was mainly in belts and clusters, including the belt consisting of Urumqi, Shihezi, Hami, Karamay, Changji, and Ili. At the same time, logistics enterprises in nearby areas were also intensive, suggesting cross-region development in Northern Xinjiang. However, logistics enterprises were still relatively closed in Southern Xinjiang, and the linkage among regions was not as strong as that in the north. Potentially due to the joint effect of the regional economy, transportation, and other factors, there was a significant difference in the kernel density of logistics enterprises between Southern and Northern Xinjiang. The logistics industries in Urumqi, Changji, Shihezi, and Karamay were all distributed in clusters and belts in different years. Comparatively, in several areas in Western Xinjiang, such as Kizilsu, there was a smaller difference in kernel density with insignificant variation. Such results could be attributed to the weak development of logistics enterprises, and the influences of economic development, transportation, output value, and policies.

The following findings were also obtained through the analysis.

The layout of logistics enterprises in Xinjiang was based on cities and towns, with the metropolitan area consisting of Urumqi and the surrounding areas as the center. At the same time, with a longer distance from Urumqi, the number and area of core areas gradually decreased. The urban agglomeration consisting of Urumqi and surrounding areas was the place with the highest density of logistics enterprises. Furthermore, the farther the distance from Urumqi, the smaller the number and scope of core areas; the spatial distribution of logistics enterprises consistently exhibited a strong tendency of agglomeration toward Urumqi. Such findings could be attributed to the fact that Urumqi is the provincial capital, with a concentrated population and various complete infrastructures. Thus, there was a greater demand for logistics enterprises than that in other regions. Additionally, logistics enterprises also pursue high timeliness and maximum transportation convenience.

There was a significant difference in spatial layout between Northern and Southern Xinjiang. Xinjiang belongs to a typical oasis economy, and, due to the special natural topography thereof, cities are separated by a long distance and mainly connected by several important routes. Since no dense distribution network had been established, logistics enterprises were mostly distributed along the main transportation lines and the edges of basins. Under spatial dispersion, logistics enterprises were distributed in belts, batches, and clusters. The Tianshan Mountain North Slope Economic Belt was the first urban economic belt in Xinjiang with a considerable scale, due to the developed highway and railway network and dense cities and towns. Northern Xinjiang ranked highest in terms of urban economic development in Xinjiang, and approximately 65% of the entire urban population also lived in Northern Xinjiang. Comparatively, there were fewer cities in Southern and Eastern Xinjiang, thereby resulting in weak urban economic development and industrial foundations. As a result, logistics enterprises also lagged behind. Therefore, the efficiency of transport lines in Xinjiang should be strengthened.

3.3. Selection of Influencing Factor Variables and Correlation Analysis

3.3.1. Selection of Variables

Based on the results of existing research [14,36,37,38,39], the availability of data and the actual development of the studied area were considered in the present study, with a focus on the dimensions of the economic environment, social environment, market environment, and special location. The geographical detector was used to detect the main influencing factors of the distribution of logistics enterprises in Xinjiang according to the following indicators: X1 regional GDP, X2 regional total industrial output value, X3 regional per capita GDP, X4 regional fixed assets investment, X5 regional population size, X6 regional total local fiscal budget expenditure, X7 regional use of foreign direct investment, X8 regional total import and export trade volume, X9 regional total retail sales of social consumer goods, and X10 regional number of ports. Through geographical detection, the influencing factors of the spatial differentiation of logistics enterprises in Xinjiang and the driving mechanism were revealed. The selection and calculation of indicators are detailed in

Table 4. Selection of variables and indicators influencing the spatial and temporal distribution of logistics enterprises.

Dimension	Indicator	Code	Unit
	Regional GDP	(X1)	RMB 10,000
Economic environment	Total industrial output value	(X2)	RMB 10,000
	Regional GDP per capital	(X3)	RMB 10,000
	Total fixed assets investment in transportation, warehousing, and post industry	(X4)	RMB 10,000
Social environment	Resident population at year end	(X5)	RMB 10,000
	Local fiscal budget expenditure	(X6)	RMB 10,000
	Actual use of foreign investment	(X7)	USD 10,000
Market environment	Total import and export trade volume	(X8)	USD 10,000
	Total retail sales of social consumer goods	(X9)	RMB 10,000
Locational factor	Number of ports	(X10)	Each

.

3.3.2. Correlation Analysis

Although the geographical detector could better detect the extent to which a variable factor X explained the spatial differentiation of attribute Y, the direction of influence could not be determined. Therefore, Pearson correlation analysis was performed before analysis with the geographical detector.

The correlation coefficient was calculated to identify whether the 10 indicators were related to the scale of logistics enterprises in Xinjiang and to determine the influence direction of influencing factors. SPSS Statistics 25.0 was used for the statistical analysis of Pearson correlations, and the results are shown in

Table 5. The correlation coefficient between logistics enterprises in Xinjiang and the Pearson correlation.

Indicator	X1	X2	X3	X4	X5	X6	X7	X8	X9	X10
2012	0.922 **	0.764 **	0.058	0.714 **	0.504 *	0.640 **	0.935 **	0.817 **	0.893 **	0.137
2016	0.910 **	0.777 **	0.111	0.783 **	0.449 *	0.587 **	0.072	0.649 **	0.962 **	0.136
2020	0.963 **	0.744 **	0.427	0.663 **	0.623 **	0.593 **	0.746 **	0.676 **	0.977 **	0.168

Note: * indicates significant correlation (two-sided) at the 0.05 level; ** indicates significant correlation (two-sided) at the 0.01 level.

.

The results reveal that there was a significant positive relationship between the 10 indicators and the scale and quantity of enterprises, thereby verifying the rationality of the selected indicators to a certain extent. Among the indicators, seven influencing factors passed the significance test, with relatively high correlation coefficients within the three periods, including X1 regional GDP, X2 regional total industrial output value, X4 regional fixed assets investment, X5 regional population size, X6 regional total local fiscal budget expenditure, X8 regional total import and export trade volume, and X9 regional total retail sales of social consumer goods. The significance of X3 regional GDP per capital increased year by year. X7 regional use of foreign direct investment in 2012 and 2020 also passed the significance test with relatively high correlation coefficient. The correlation coefficients of X10 regional number of open ports were 0.137, 0.136, and 0.168, which were low, indicating a low correlation between the number of ports at prefectures and the distribution of logistics enterprises.

3.4. Analysis of Influencing Factors

3.4.1. Analysis of Main Influencing Factors and Action Intensity

Spatial differentiation is one of the basic features of geographical phenomena. The basic meaning of the spatial differentiation of the logistics industry is that, in the process of development of logistics enterprises, the spatial layout along certain “points” or “lines” shows the characteristics of spatial circle expansion, and the areas near these “points” or “lines” have more development opportunities and thus can obtain more economic benefits. On the contrary, the number of logistics enterprises is small, the agglomeration capacity is weak, they obtain fewer development opportunities, and the overall development will be relatively lagging. This study draws on previous research results, takes into account the actual development of Xinjiang, and considers the availability of data, with specific values from the Xinjiang Statistical Yearbook and the official statistical bulletin of Xinjiang. In the present study, the geographical detector was used to analyze the spatial differentiation of logistics enterprises in Xinjiang. First, ArcGIS 10.8 was used for the hierarchical processing of various independent variables. The ArcGIS natural breakpoint method was used to classify the collected influencing factors into five categories, and the independent variables were converted from numerical quantity to type quantity. Subsequently, the type quantity was read into the geographical detector to calculate the influence q of each influencing factor on the distribution of logistics enterprises in Xinjiang in 2012, 2016, and 2020, and the calculation results are shown in

Table 6. Geographical detection results of influencing factors of distribution of logistics enterprises in Xinjiang.

Indicator	X1	X2	X3	X4	X5	X6	X7	X8	X9	X10
2012	0.9411	0.5596	0.0454	0.6049	0.5050	0.5300	0.8927	0.5348	0.8923	0.3891
2016	0.5990	0.4908	0.1935	0.6589	0.5990	0.4579	0.6180	0.5380	0.9609	0.3438
2020	0.9551	0.0836	0.4242	0.6437	0.4819	0.4953	0.7899	0.6131	0.9549	0.2918

.

From the time series of 2012, 2016, and 2020, the specific influence of each factor could be described as follows: X3 regional GDP per capita, X8 regional total import and export trade volume, X4 regional fixed assets investment, and X9 regional total retail sales of social consumer goods exhibited steady growth. X2 regional total industrial output value, X1 regional GDP, and X10 regional number of open ports exhibited a decline. X1 regional GDP, X5 regional population size, X6 regional total fiscal budget expenditure, and X7 regional utilization of foreign direct investment first declined and then rebounded.

Specifically, the core influencing factors of the spatial distribution of logistics enterprises in Xinjiang in 2012 were X1 regional GDP, X7 regional use of foreign direct investment, X9 regional total retail sales of social consumer goods, X4 regional fixed assets investment, and X2 total industrial output value, of which the q values were 0.9411, 0.8927, 0.8923, 0.6048, and 0.5596, respectively.

In 2016, the core influencing factors of the spatial distribution of logistics enterprises in Xinjiang were X9 regional total retail sales of consumer goods, X4 regional fixed assets investment, X7 regional use of foreign direct investment, X1 regional GDP, and X5 regional population size, for which the q values were 0.9609, 0.65892, 0.6180, 0.5990, and 0.5990, respectively.

In 2020, the core influencing factors of the spatial distribution of logistics enterprises in Xinjiang were X1 regional GDP, X9 regional total retail sales of social consumer goods, X7 regional use of foreign direct investment, X4 regional fixed assets investment, and X8 regional total import and export trade volume, for which the q values were 0.9609, 0.65892, 0.6180, 0.5990, and 0.5990, respectively.

In conclusion, X1 regional GDP, X9 regional total retail sales of social consumer goods, X7 regional use of foreign direct investment, and X4 regional fixed assets investment had the strongest influences on the distribution of logistics enterprises in all areas.

3.4.2. Analysis Based on Detected Factors

With respect to the influence q of detected factors in 2012, 2016, and 2020, the main influencing factors of the spatial distribution of logistics enterprises in Xinjiang were X1 regional GDP, X9 regional total retail sales of social consumer goods, X7 regional use of foreign direct investment, X4 regional fixed assets investment, X2 regional industrial output value, X5 regional population size, and X8 regional total import and export trade volume, which largely determined the spatial differentiation.

(1) Regarding X1 regional GDP, X2 regional total industrial output value, and X3 regional GDP per capital, logistics enterprises generally prefer to be located in areas with relatively high economic development, where regional economic strength is a significant foundation for the development of logistics enterprises. Here, regional economic strength not only largely determines the development of the logistics market, but also directly leads to regional differences in the layout of logistics enterprises. During the studied period, the influence of regional economic strength on the distribution of logistics enterprises was consistently high, with q values of 0.9411, 0.5990, and 0.9551, respectively. The economic development in Xinjiang exhibited a downward trend from north to south, which determined the basic distribution pattern of logistics enterprises, which gradually spread from the north to the south. At the same time, the distribution pattern of logistics enterprises also reflected the economic differences between Northern and Southern Xinjiang to a certain degree. The regional total industrial output value also had a significantly high influence on the layout and distribution of logistics enterprises in Xinjiang, with a greater regional total industrial output value indicating a greater demand for logistics services and the more favorable agglomeration of logistics enterprises. During the three periods, the q values were 0.5596, 0.4908, and 0.4552, respectively. With the changes in regional development, the force q value of each test factor on the distribution of the logistics industry in the region also changed to varying degrees, and the acting force q of each tested factor on the regional distribution of logistics enterprises was also gradually reduced. As such, the regional total industrial output value somewhat stabilized the growth of regional logistics enterprises. Logistics enterprises grow rapidly with fast regional economic development, and are closely related to regional industrial planning and industrial development. With the general improvement in social industrialization, the influence q of industrialization on logistics enterprises also declined from 0.5596 in 2012 to 0.4908 in 2016. An observation can be made that the higher the regional GDP per capital, the stronger the economic strength and the more favorable for the agglomeration of logistics enterprises. The q values were 0.0454, 0.1935, and 0.4242, respectively, exhibiting an increasing trend. The influence was also increased, which, to some extent, reflected that regional economic development promoted the growth of regional logistics enterprises.

(2) With regard to X4 regional fixed assets investment, X5 regional population size, and X6 local total fiscal budget expenditures, such factors had significant influences on the layout of logistics enterprises. The influences were relatively high during the three different periods, with q values of 0.6048, 0.658, and 0.6437, respectively. The q values of X5 regional population size were 0.5050, 0.5990, and 0.4819, respectively. The agglomeration of logistics enterprises benefits from the size of the population, human production and economic activities, and other activities, which are closely related to logistics enterprises. For instance, Urumqi, Changji, and Ili were consistently in the leading positions in terms of the number of logistics enterprises, since there were large numbers of service enterprises. The regional population emerged as a significant influencing factor of the spatial pattern of the logistics industry, and could promote the layout of logistics enterprises to a large extent. With the increased population and convenient transportation, the economic connections within and outside Xinjiang were further strengthened, resulting in a slight decline in the q value in 2020. X6 local total fiscal budget expenditure had a significant influence on the layout of logistics enterprises, and the influence q values during different periods were 0.5300, 0.4579, and 0.4953, respectively, with relatively little difference. By formulating logistics policies and development plans that meet regional conditions, building relevant infrastructure to serve logistics enterprises, and building logistics industrial parks, the local government can guide logistics enterprises to grow larger and stronger, which is of considerable significance for the development of logistics enterprises.

(3) The q values of X7 regional use of foreign direct investment were 0.8927, 0.6180, and 0.7899, respectively. The q values of X8 regional total import and export trade volume were 0.5348, 0.5380, and 0.6131, respectively. The q values of X9 regional total retail sales of social consumer goods were 0.8923, 0.9609, and 0.9549, respectively. X7 regional use of foreign direct investment had a significant influence on the layout of logistics enterprises in Xinjiang. During the three periods, there was little difference in this influence, with q values of 0.8927, 0.6180, and 0.7899, respectively. The agglomeration of logistics enterprises could be attributed to the increased regional total import and export trade volume (X8). There are considerable activities relating to logistics enterprises in the import and export trade. As an example, Urumqi had the highest number of logistics enterprises since there were more foreign enterprises. A satisfactory market scale is a significant basis for the layout of logistics enterprises, and can significantly reduce the less interactive transportation cost and improve the logistics organization efficiency of enterprises. Through comparison of the years 2012, 2016, and 2020, the influence q values of total retail sales of social consumption goods on the layout of logistics enterprises were 0.8923, 0.9609, and 0.9549, respectively. An observation can be made that the influence on the layout of logistics enterprises was maintained at a high level on the whole and that the q value was increased from 0.8923 in 2012 to 0.9549 in 2020.

(4) For the dimension of the special locational factor (number of land ports), the q values of X10 number of land ports were 0.3891, 0.3438, and 0.2918, respectively. As such, the factor had no significant influence on the layout of logistics enterprises in Xinjiang. Border ports have multiple roles in aspects such as transportation hubs and resource transportation, support for conventional and modern industries, culture, and security [39]. However, due to issues in Xinjiang’s land border ports, such as the limited opening of ports and reliance on traditional geographical advantages, the role in promoting border logistics was affected.

3.4.3. Detection of Interaction of Influencing Factors

To explore the influence of any two factors on the spatial differentiation of logistics enterprises when working conjunctively, the interaction detector in the geographical detector software was further used, and the results are shown in

Table 7. Interaction detection results of different influencing factors of logistics enterprises in Xinjiang.

Factors	2012	2016	2020	Factors	2012	2016	2020
X1 ∩ X2	BE	BE	BE	X3 ∩ X10	NE	BE	NE
X1 ∩ X3	BE	NE	BE	X4 ∩ X5	BE	BE	NE
X1 ∩ X4	BE	BE	BE	X4 ∩ X6	BE	NE	NE
X1 ∩ X5	BE	BE	BE	X4 ∩ X7	BE	NE	BE
X1 ∩ X6	BE	BE	BE	X4 ∩ X8	BE	NE	NE
X1 ∩ X7	BE	BE	BE	X4 ∩ X9	BE	BE	BE
X1 ∩ X8	BE	BE	BE	X4 ∩ X10	BE	NE	NE
X1 ∩ X9	BE	BE	BE	X5 ∩ X6	BE	BE	BE
X1 ∩ X10	BE	BE	BE	X5 ∩ X7	BE	BE	BE
X2 ∩ X3	NE	BE	BE	X5 ∩ X8	BE	BE	BE
X2 ∩ X4	BE	NE	BE	X5 ∩ X9	BE	BE	BE
X2 ∩ X5	BE	NE	NE	X5 ∩ X10	NE	NE	NE
X2 ∩ X6	BE	NE	NE	X6 ∩ X7	BE	BE	BE
X2 ∩ X7	BE	BE	BE	X6 ∩ X8	BE	BE	BE
X2 ∩ X8	BE	BE	BE	X6 ∩ X9	BE	BE	BE
X2 ∩ X9	BE	BE	BE	X6 ∩ X10	NE	NE	NE
X2 ∩ X10	BE	BE	BE	X7 ∩ X8	BE	BE	BE
X3 ∩ X4	BE	NE	NE	X7 ∩ X9	BE	BE	BE
X3 ∩ X5	NE	NE	NE	X7 ∩ X10	BE	BE	BE
X3 ∩ X6	NE	NE	NE	X8 ∩ X9	BE	BE	BE
X3 ∩ X7	BE	NE	BE	X8 ∩ X10	BE	NE	NE
X3 ∩ X8	NE	NE	BE	X9 ∩ X10	BE	BE	BE
X3 ∩ X9	NE	BE	BE

Note: Nonlinear enhancement is denoted as NE; bi-factor enhancement is denoted as BE

.

In 2012, there were eight groups of interactions that generated a nonlinear enhancement (NE); in particular, the influence of the interaction between different influencing factors was greater than the sum of the influence of two individual factors, namely X2 regional total industrial output value ∩ X3 regional GDP per capital, X3 regional GDP per capital ∩ population size, X3 regional GDP per capital ∩ X6 local fiscal budget expenditure, X3 regional GDP per capital ∩ X8 regional total import and export trade volume, X3 regional GDP per capital ∩ X9 regional total retail sales of social consumer goods, X3 regional GDP per capital ∩ X10 regional number of open ports, X5 regional population size ∩ X10 regional number of open ports, and X6 local fiscal budget expenditure ∩ X10 regional number of open ports.

Thus, the influence of cargo transportation capacity and X3 regional GDP per capita on the distribution of logistics enterprises during the studied period was particularly notable; the interaction between the remaining factors generated a bi-factor enhancement effect, which was less significant compared with the nonlinear enhancement.

In 2016, there were 16 groups of interactions that generated nonlinear enhancement effects (NE), namely X1 regional GDP ∩ X3 regional GDP per capital, X2 regional total industrial output value ∩ X4 regional fixed assets investment, X2 regional total industrial output value ∩ X5 regional population size, X2 regional total industrial output value ∩ X6 local fiscal budget expenditure, X3 regional GDP per capital ∩ X4 regional fixed assets investment, X3 regional GDP per capital ∩ X5 regional population size, X3 regional GDP per capital ∩ X6 local fiscal budget expenditure, X3 regional GDP per capital ∩ X7 regional use of foreign direct investment, X3 regional GDP per capital ∩ X8 regional total import and export trade volume, X4 regional fixed assets investment ∩ X6 local fiscal budget expenditure, X4 regional fixed assets investment ∩ X7 regional use of foreign direct investment, X4 regional fixed assets investment ∩ X8 regional total import and export trade volume, X4 regional fixed assets investment ∩ X10 regional number of open ports, X5 regional population size ∩ X10 regional number of open ports, X6 local fiscal budget expenditure ∩ X10 regional number of open ports, and X8 regional total import and export trade volume ∩ X10 regional number of open ports.

In 2020, there were 13 groups of interactions that generated nonlinear enhancement effects (NE), namely X2 regional total industrial output value ∩ X5 regional population size, X2 regional total industrial output value ∩ X6 local fiscal budget expenditure, X3 regional GDP per capital ∩ X4 regional fixed assets investment, X3 regional GDP per capital ∩ X5 regional population size, X3 regional GDP per capital ∩ X6 local fiscal budget, X3 regional GDP per capital ∩ X10 regional number of open ports, X4 regional fixed assets investment ∩ X5 population size, X4 regional fixed assets investment ∩ X6 local fiscal budget expenditure, X4 regional fixed assets investment ∩ X8 regional total import and export trade volume, X4 regional fixed assets investment ∩ X10 regional number of open ports, X5 regional population size ∩ X10 regional number of open ports, X6 local fiscal budget expenditure ∩ X10 regional number of open ports, and X8 regional total import and export trade volume ∩ X10 regional number of open ports.

There were 37 groups of interactions in the three time series that generated a nonlinear enhancement (NE), accounting for 37/135 of the total number of interaction situations, which were mainly concentrated in X2 regional total industrial output value, X3 regional GDP per capital, and X4 regional fixed assets investment. Such factors generated a nonlinear enhancement (NE) after interaction with the remaining factors.

X3 regional GDP per capital ∩ X5 regional population size, X3 regional GDP per capital ∩ X6 local total fiscal budget expenditure, X5 regional population size ∩ X10 regional number of open ports, and X6 local total fiscal budget expenditure ∩ X10 regional number of open ports generated a nonlinear enhancement within consecutively three time series. Such findings indicate that X3 regional GDP per capital ∩ X5 regional population size, X3 regional GDP per capital ∩ X6 local total fiscal budget expenditure, X5 regional population size ∩ X10 regional number of open ports, and X6 local total fiscal budget expenditure ∩ X10 regional number of open ports provided significant support and drive, which effectuated the fast development of Level-A logistics enterprises. The significant influences of X3 regional GDP per capital, X5 regional population size, X6 local total fiscal budget expenditure, and X10 regional number of open ports on the layout of logistics enterprises were particularly notable.

X2 regional total industrial output value ∩ X5 regional population size, X2 regional total industrial output value ∩ X6 regional total fiscal budget expenditure, X3 regional GDP per capital ∩ X4 regional fixed assets investment, X4 regional fixed assets investment ∩ X6 regional total fiscal budget expenditure, X4 regional fixed assets investment ∩ X8 regional total import and export trade volume, X4 regional fixed assets investment ∩ X10 regional number of open ports, and X8 regional total import and export trade volume ∩ X10 regional number of open ports generated a constant enhancement for two consecutive time series. As such, an observation can be made that X2 regional total industrial output value, X4 regional fixed assets investment, and X8 regional total import and export trade volume generated increasingly significant influences on the layout of logistics enterprises, and more attention should be paid to the role of such factors. The interaction between X3 regional GDP per capital ∩ X8 regional total import and export trade volume was reduced, indicating that the distribution of logistics enterprises was influenced by more factors.

4. Discussion

To summarize, logistics enterprises in Xinjiang have formed an industrial pattern of “multi-core agglomeration and coverage relying on central cities”.

Urumqi, the number one logistics center in Xinjiang, enjoys relatively rapid and steady development of logistics enterprises with a relatively concentrated scale. Despite such development in Urumqi, other secondary logistics agglomeration centers are quite different, with weaker logistics agglomeration. The logistics industry in Xinjiang faces problems such as large regional differences, few inter-regional linkages, an unreasonable structure, imbalance, and inadequacy. As such, in terms of improving the logistics scale system in Xinjiang, the development of secondary logistics agglomeration centers is an effective method, while promoting high-quality development to form new development momentum and determining how to reasonably arrange the development of logistics industry and make better use of existing spatial advantages and greater development room is integral. Based on the present research, the following feasible optimization paths are proposed:

(1) Importance should be attached to the multi-factor synergistic effect in the logistics industry, regional differentiated development strategies should be implemented, and complementary advantages with differentiated positioning and development patterns should be formed. On the one hand, for Northern Xinjiang, with highly developed logistics enterprises, stable economic development should be actively maintained to form more larger industrial agglomeration cores, thereby driving the development of other areas. As the economic and logistics center of Xinjiang, Urumqi should continue its infrastructure construction, such as the modern logistics industry, be a leader within Central Asia and other regions, and support the construction of Xinjiang’s logistics system as a city with the primacy ratio. Regarding other cities in Northern Xinjiang, such as the port cities of Ili Kazak Autonomous Prefecture and Bortala, with a prosperous and convenient import and export trade, the advantages of land ports such as Horgos Port and Alataw Port should be fully utilized. Further, such areas should coordinate and cooperate with Urumqi and other cities to form a stronger logistics carrying capacity, thereby providing a more effective service for the Chinese and Central Asian economies. Additionally, the guidance on the development of the logistics industry should be strengthened, and the development of the basic economy in Southern Xinjiang should be promoted; the regional abilities of Akesu and Kashgar, with relatively weak economic foundations in the layout of the logistics industry, should be improved, thereby bridging the gaps in the logistics industry among different areas. Cities in Southern Xinjiang should make use of the respective advantages in logistics agglomeration and ports, and cooperate with cities such as Urumqi, Ili, and other cities in Northern Xinjiang to form a driver for logistics development in Southern Xinjiang, thereby providing a more effective service for economic trade between China and countries in Central Asia and South Asia.

(2) There is a need to make plans, provide guidance, strengthen linkages, and optimize the spatial pattern of the logistics industry. First, medium- and long-term strategies of logistics development should be implemented to form a cooperative relationship of coordinated development among various areas; overall arrangement should be implemented for emerging problems in the development of logistics, and efficient communication should be promoted between areas. Second, specific planning for logistics development should be conducted within the areas based on local conditions, and development strategies should be formulated as per the respective development characteristics and the industrial characteristics. Logistics enterprises are an effective carrier of regional economic services. Currently, logistics enterprises in Xinjiang have undergone a certain level of development and basically meet the existing economic demand. However, the agglomeration and development of logistics enterprises are not balanced among the areas in Xinjiang. Therefore, relying on the regional advantages, Ili and Kashgar should proactively develop the local logistics industry, support the improvement of transportation, and deepen the synergetic coordination between logistics enterprises and the local industry, so as to meet the demand for more prosperous economic development in the future.

(3) Transportation facilities and regional connectivity should be improved. Transportation is a key condition for the further promotion of logistics enterprises and economic development. The improvement of backward transportation in cities such as Ili, Kizilsu, and Kashgar will help to improve the local logistics efficiency and promote economic trade with Central Asian countries.

Although the evolution of the spatial pattern of logistics enterprises and the driving factors were analyzed in the present study, due to a lack of business attributes such as finance and taxation, employees, and input and output of the logistics industry, there is only a small amount of quantitative analysis with regard to the theories and methods of fusion of the logistics industry with other industries. Further, the suitable areas for logistics industries in Xinjiang were not clarified. Therefore, in-depth research should be conducted on such spatial and temporal adaptability.

5. Conclusions

In the present study, ArcGIS and other software were used for analyzing the evolutionary characteristics of the spatial and temporal distribution of logistics enterprises in Xinjiang in 2012, 2016, and 2020 with data analysis methods such as nearest neighbor analysis, kernel density analysis, and standard deviation ellipse analysis. Further, the geographic detector model was adopted to analyze the influencing factors of the evolution of the spatial pattern and the interactions therebetween. The main conclusions are as follows:

(1) The development of logistics enterprises in Xinjiang was unbalanced at different development stages, with obvious different characteristics. With respect to the development dynamics, the distribution of logistics enterprises in Xinjiang was generally characterized by an “inverted U-shaped” trend. Regarding the distribution range, there was a higher number of logistics enterprises in Northern Xinjiang than Southern Xinjiang, with significant differences among different areas.

(2) The spatial distribution of logistics enterprises in Xinjiang was generally characterized by a “northeast–southwest” trend, and the ellipses covered the main economically significant areas in Xinjiang. The overall distribution center of logistics enterprises in Xinjiang was not greatly deviated, with the center of the ellipse always within Bayingolin and close to the junction of Bayingolin and Shihezi, without cross-regional movement. Xinjiang’s unique topography had a considerable influence on the coverage of elliptical axis cities. Notably, there was a small number of towns around the axis cities owing to the large distance among cities caused by the topography.

(3) During the studied period, the spatial distribution of logistics enterprises in Xinjiang was overall agglomerated, and significantly characterized by agglomeration. The overall spatial distribution was characterized by “multi-center development with core agglomeration and patch distribution at the edge”. The degree of agglomeration had a positive relationship with the number and scale of enterprises, and underwent an increase before declining. The kernel density of logistics enterprises was reliant on cities and towns, and primarily characterized by distribution in belts and clusters. In Northern Xinjiang, the kernel density was distributed in patterns of clusters, belts, and spots; in Southern Xinjiang, the kernel density was still largely distributed in patterns of clusters and spots. Despite undergoing certain changes, the spatial distribution of logistics enterprises in Xinjiang was generally stable. The logistics enterprises were mainly concentrated on the northern slope of the Tianshan Mountains and Northern Xinjiang, with a dense population and high economic development, but Southwestern Xinjian was still the secondary hotspot area. At present, the hierarchical structure of the spatial distribution of logistics enterprises in Xinjiang is not reasonable enough. The logistics enterprises in the agglomeration center are in a weakly balanced state, reflecting the low integration of the logistics agglomeration system in Xinjiang. With the city circle of Urumqi as the core, the spatial structure of logistics enterprises in Xinjiang was attenuated toward other areas. The logistics agglomeration structure in Urumqi was relatively compact, with a certain degree of spatial density. As the largest city in Xinjiang, Urumqi is the main transit location for materials entering and leaving Xinjiang, and has a strong influence on other logistics center cities. The other cities have limited influence as secondary logistics centers, but the logistics agglomeration structure is stable.

(4) With respect to the influencing factors concerning the evolution of the spatial pattern of logistics enterprises in Xinjiang and the combinations thereof, X3 regional GDP per capital, X8 regional total import and export trade volume, X4 regional fixed assets investment, and X9 regional total retail sales of social consumption goods exhibited steady growth. X2 regional total industrial output value, X10 regional number of open ports, X1 regional GDP, X5 regional population size, X6 local total fiscal budget expenditure, and X7 regional use of foreign direct investment declined and then rebounded, showing a “high–low–high” trend. As indicated by the gradually reduced difference in the influential effect, multiple elements had a greater influence on the spatial distribution of the logistics industry, while the influence of a single factor was weakened and the influence on the logistics industry also became balanced. The influence of the interaction between different factors on the spatial distribution of logistics enterprises in Xinjiang was greater than the influence of each single factor, and there were no mutually independent influencing factors. X1 regional GDP, X9 regional total retail sales of social consumption goods, X7 regional use of foreign direct investment, and X4 regional fixed assets investment had the greatest influence on the distribution of logistics enterprises, and could be regarded as the main driving forces for the rapid development of logistics enterprises. Based on the foregoing analysis on influence, the detection results of single factors were as follows: X1 regional GDP, X9 regional total retail sales of social consumption goods, X7 regional use of foreign direct investment, and X4 regional fixed assets investment were highly influential factors, and had the strongest influence on the distribution of logistics enterprises in all areas; the detection results of the interactions between two factors indicate an enhanced action effect, revealing that the spatial distribution of logistics enterprises in Xinjiang was the result of the synergistic effect of multiple factors.

6. Suggestions

To summarize, logistics enterprises in Xinjiang have formed an industrial pattern of “multi-core agglomeration and coverage relying on central cities”.

Urumqi, the number one logistics center in Xinjiang, enjoys the relatively fast and steady development of logistics enterprises with a relatively concentrated scale. However, other regions, as secondary logistics agglomeration centers, are quite different from Urumqi, with weaker logistics agglomeration. The logistics industry in Xinjiang faces problems such as large regional differences, few inter-regional linkages, an unreasonable structure, imbalance, and inadequacy. Therefore, the development of secondary logistics agglomeration centers is an effective way to improve the logistics scale system in Xinjiang. While promoting high-quality development to form a new development momentum, it is critically important to determine how to reasonably arrange the development of the logistics industry and make better use of the existing spatial advantages and greater development room. Based on the study, the following feasible optimization paths are proposed:

(1) Attach importance to the multi-factor synergistic effect in the logistics industry, implement regional differentiated development strategies, and form complementary advantages with differentiated positioning and development patterns. On the one hand, for Northern Xinjiang, with highly developed logistics enterprises, it is necessary to actively maintain stable economic development to form more larger industrial agglomeration cores, thus driving the development of other areas. As the economic and logistics center of Xinjiang, Urumqi should continue its infrastructure construction, such as a modern logistics industry, play a leading role within Central Asia and other regions, and support the construction of Xinjiang’s logistics system as a city with the primacy ratio. As to other cities in Northern Xinjiang, such as the port cities of Ili Kazak Autonomous Prefecture and Bortala, with a prosperous and convenient import and export trade, the advantages of land ports such as Horgos Port and Alataw Port should be fully utilized. Furthermore, they should coordinate and cooperate with Urumqi and other cities to form a stronger logistics carrying capacity, thus providing more effective services for the Chinese and Central Asian economies. Furthermore, it is necessary to strengthen the guidance on the development of the logistics industry, promote the development of the basic economy in Southern Xinjiang, and improve the regional ability of Akesu and Kashgar, with relatively weak economic foundations in the layout of the logistics industry, thus bridging the gap in the logistics industry among different areas. Cities in Southern Xinjiang should give full play to their respective advantages in logistics agglomeration and ports, and cooperate with cities such as Urumqi, Ili, and other cities in Northern Xinjiang to form a driver for logistics development in Southern Xinjiang, thus providing more effective services for economic trade between China and countries in Central Asia and South Asia.

(2) Make plans, provide guidance, strengthen linkages, and optimize the spatial pattern of the logistics industry. First, it is necessary to implement medium- and long-term strategies of logistics development to form a cooperative relationship of coordinated development among various areas; to make overall arrangements for emerging problems in the development of logistics, and to promote efficient communication among areas. Second, it is necessary to carry out specific planning for logistics development within the areas based on local conditions, and formulate development strategies as per the respective development characteristics and the industrial characteristics. Logistics enterprises are an effective carrier of regional economic services. At present, logistics enterprises in Xinjiang have witnessed certain development and basically meet the existing economic demand. However, the agglomeration and development of logistics enterprises are not balanced among the areas in Xinjiang. Therefore, Ili and Kashgar should, relying on their regional advantages, proactively develop the local logistics industry, support the improvement of transportation, and deepen the synergetic coordination between logistics enterprises and the local industry, so as to meet the demand for more prosperous economic development in the future.

(3) Improve transportation facilities and regional connectivity. Transportation is a key condition for the further promotion of logistics enterprises and economic development. The improvement of backward transportation in cities such as Ili, Kizilsu, and Kashgar will help to improve the local logistics efficiency and promote economic trade with Central Asian countries.

Although this study analyzes the evolution of the spatial pattern of logistics enterprises and the driving factors, there is not much quantitative analysis with regard to the theories and methods of fusion of the logistics industry with other industries due to a lack of business attributes such as finance and taxation, employees, and the input and output of the logistics industry. Furthermore, the suitable areas for logistics industries in Xinjiang are not clarified. Therefore, in-depth research should be carried out on such spatial and temporal adaptability.