Identification of Production–Living–Ecological Spatial Conflicts and Multi-Scenario Simulations in Extreme Arid Areas

Abstract

“Production–Living–Ecological” spatial conflicts (PLECs) are critical issues arising from regional land development, affecting economic, social, and ecological security. Identifying and analyzing these conflicts’ spatiotemporal characteristics is essential for sustainable development. This study focuses on the Tuha region, which experiences an extremely arid climate, classifying the region’s “Production–Living–Ecological” (PLE) spaces into four types: living–production, ecological–production, production–ecological, and ecological spaces. A spatial conflict measurement model based on landscape patterns was developed to analyze the evolution of PLECs from 2000 to 2020. Additionally, the PLUS model was used to simulate PLEC patterns in 2030 under different development scenarios. The results indicate that between 2000 and 2020, the area proportions in the Tuha region ranked from largest to smallest as follows: ecological space, ecological–production space, production–ecological space, and living–production space. The area of living–production space increased, while production–ecological space first increased and then stabilized, and the areas of ecological and ecological–production spaces decreased. From 2000 to 2020, spatial conflicts in the region were predominantly characterized by mild weak conflicts. High–high PLEC clusters were concentrated in urban and surrounding areas of Gaochang District, Toksun County, Shanshan County, and Yizhou District, while low–low clusters were found in the Eastern Tianshan Mountains and northern Barkol Kazakh Autonomous County. NDVI, GDP, population, and proximity to roads positively influenced PLECs, while elevation, slope, aspect, and precipitation had inhibitory effects. Under different development scenarios, the natural development scenario leads to the most severe spatial conflicts, while the cropland protection scenario reduces PLECs and enhances regional welfare, making it the optimal pathway for future development.

1. Introduction

Arid zones are some of the most vulnerable regions in the global ecosystem, where land degradation and ecosystem degradation are becoming more and more prominent due to the double pressure of climate change and human activities [1]. In extreme arid zones, limited water resources and fragile ecosystems have intensified the competition between production, living and ecological spaces, leading to increased spatial conflicts and seriously threatening regional ecological security and sustainable development [2,3,4]. In response to this situation, the national government has explicitly proposed a new blueprint for spatial development, aiming to gradually establish efficient and concentrated production spaces, livable and suitable living spaces, and ecologically beautiful natural spaces [5]. Subsequently, the national government further strengthened the importance of ecological civilization construction, providing policy guidance for research on Production–Living–Ecological (PLE) spaces.

As a result, the issue of “ Production–Living–Ecological” spaces has rapidly become a hotspot of shared interest among national policymakers and the academic community. In recent years, numerous scholars have conducted systematic and in-depth research on Production–Living–Ecological (PLE) spaces. For example, Li et al. (2016) developed a classification framework based on land-use functions to distinguish among production, living, and ecological land [6,7,8]; Zhang and Fang (2019) analyzed the spatiotemporal evolution characteristics of PLE space in the Yangtze River Delta using remote sensing data [9,10,11,12]; Huang et al. (2021) proposed an index-based approach to assess the intensity of spatial conflicts among different land-use functions [13,14,15].

These research achievements have laid a solid theoretical foundation for understanding the PLECs (Production–Living–Ecological Spatial Conflicts) in the Turpan–Hami (Tuha) region. However, research on extreme arid zones is still limited; in particular, the spatial pattern of conflict evolution and its driving mechanisms are still unclear. In this context, conducting an in-depth study on spatial conflicts in the Tuha region, especially forecasting future conflicts under different scenarios, holds significant importance for the region’s sustainable development.

When exploring the connection between PLECs and land-use conflicts, it becomes evident that the two research domains are closely intertwined. Research on land-use conflicts dates back to 1977 when the Rural Society of the United Kingdom selected “land management, land-use relations, and conflicts” as one of the five core themes at an academic forum on urban fringes. Since then, “land-use conflicts” have become a central focus in the field. Scholars have systematically studied various aspects of land-use conflicts, including their origins [16], types [17], identification [18], evolution [19], and management strategies [20]. In China, research on this topic began relatively late, with initial studies emerging in 2001 [21]. These studies primarily examined conflicts between construction and residential land [22], construction and agricultural land [23], and issues such as land-type transformations and biodiversity loss [24]. Academic attention on spatial conflicts currently centers on several key areas: quantifying conflict intensity, analyzing characteristics of development and change, uncovering underlying influencing factors, and predicting potential future conflicts [25,26,27,28].

Early studies on land-use conflicts, constrained by technological limitations, have often relied on qualitative or empirical methods for conflict identification. Techniques such as expert evaluation, public participation surveys, and game theory were commonly employed. Some researchers also used the Pressure–State–Response (PSR) model to assess the degree of land-use conflicts. However, these approaches were limited in spatial accuracy and struggled to pinpoint the exact locations of conflicts [29,30,31]. With advancements in remote sensing and geographic information systems (GISs), methods such as landscape pattern indices and land-use change composite indices have been widely adopted in land-use conflict research [32,33,34]. These tools enable more precise spatial analyses, allowing researchers to better understand and address conflicts in land use. In terms of research focus, most studies have concentrated on plain regions and areas like the northern side of the Tianshan Mountains, which also share arid climate characteristics [25,35], with a notable lack of research on the PLECs in extreme arid regions.

Current research on the PLECs in extreme arid zones still faces several key limitations: (1) the spatial and temporal dynamics of the conflict remain unclear; (2) there is a lack of in-depth investigation into the underlying mechanisms driving the conflict, particularly the interaction between natural and human-induced factors; and (3) the adaptability and accuracy of existing simulation and prediction methods in arid zones require significant improvement.

Based on this, the present study takes the Turpan–Hami (Tuha) region as a case study and utilizes landscape pattern theory to construct a model for identifying PLECs (Production–Living–Ecological Spatial Conflicts). The study systematically analyzes the evolution characteristics of PLECs in the Tuha region from 2000 to 2020 at the grid scale. Additionally, the geographic weighted regression (GWR) model is employed to explore the driving mechanisms behind the spatial patterns. Furthermore, the PLUS model is used to simulate the spatial conflict patterns under different scenarios in 2030. The research findings can provide scientific evidence and policy recommendations for land structure adjustment, ecological environment protection, regional spatial conflict mitigation, and spatial layout optimization in the Turpan–Hami region.

2. Study Area Overview and Data Sources

2.1. Study Area

The Tuha region, a collective term for Turpan City and Hami City, is located in the heart of the Eurasian continent, spanning latitudes 41°18′ to 43°43′ N and longitudes 86°40′ to 96°04′ E (Figure 1). Surrounded by mountains, the region covers a total area of 214,000 square kilometers, accounting for 12.59% of Xinjiang’s total area.

Climatically, the Turpan–Hami region exhibits a typical continental desert climate, characterized by significant diurnal temperature variations, extremely low annual precipitation (10–30 mm), and high annual evaporation (3000–4000 mm). These conditions present severe challenges for water resource management. From a socio-economic perspective, the population of the Turpan–Hami region increased from 1.043 million in 2000 to 1.367 million in 2020, while GDP surged from CNY 9.36 billion to CNY 98.13 billion. However, rapid economic growth has exacerbated ecological problems, including water scarcity, wetland degradation, and desertification, leading to the continued deterioration of the ecological environment. As a region uniquely defined by its distinct geography, extreme climate, and rapidly growing socio-economic activities, the Turpan–Hami region faces escalating conflicts in human–environment relationships. These challenges have made it a critical area for in-depth research and urgent resolution.

2.2. Data Sources

The datasets used in this study encompass various aspects, including land use, meteorological and climatic data, and socio-economic data. Detailed information about the data sources and specifications is provided in

Table 1. Data details and sources.

Category	Data	Year	Original Resolution	Data Resource
Land use data		2000, 2010, 2020	30 m	http://www.resdc.cn (accessed on 18 November 2024)
Socioeconomic driver	GDP	2022	1000 m
	Population	2022	1000 m
	Distance from road	2022	30 m	www.Webmap.cn (accessed on 18 November 2024)
	Distance from the railway
Natural driver	Digital Elevation Model (DEM) Slope	2020	30 m	http://www.gscloud.cn (accessed on 18 November 2024)
	Aspect
	Soil type	1995	1000 m	http://www.resdc.cn (accessed on 18 November 2024)
	NDVI	2020	30 m
	Annual precipitation	2020	1000 m	WorldClimv2.1 (www.worldclim.org/) (accessed on 18 November 2024)

.

3. Research Methods

3.1. Classification of “Production–Living–Ecological” (PLE) Space

Referring to the spatial classification methods for PLE spaces used by other researchers [9,11] and taking into account the specific land-use characteristics of the study area, this study divides PLE spaces into four categories: living–production space, ecological–production space, production–ecological space, and ecological space (

Table 2. PLE space dominant land-use function classification system.

PLES Classification	Meaning	Corresponding Land Use Types
Living–production space	Primarily meets basic human living and spiritual needs while also having diverse production functions, with the highest economic value.	Urban Land, Rural Residential Areas, Industrial and Transportation Construction Land
Ecological–production space	Primarily focused on ecological functions, while allowing for appropriate production activities to provide economic benefits.	Grassland, Forest Land, Reservoirs and Ponds
Production–ecological space	Primarily focused on agricultural production functions, while also serving ecological functions.	Paddy Fields, Dry Farmland
Ecological space	Provides ecological products and services, with functions such as climate regulation, carbon sequestration and oxygen release, and biodiversity protection.	Unused Land, Rivers and Canals, Lakes, Glaciers and Permanent Snow, Tidal Flats Beach Land

).

3.2. Construction of the PLECs Measurement Model

Based on relevant theories in landscape ecology and land-use spatial coordination development [27,36], this study constructs a spatial conflict measurement model to quantify and evaluate the degree of spatial conflicts in the study area. The model is developed with a focus on three aspects: complexity, stability, and vulnerability. The comprehensive index calculation method for PLECs (TSCI) is as follows:

$$TSCI={SC}_{CI}+{SC}_{FI}-{SC}_{SI}$$

(1)

The $TSCI$, or Conflict Risk Index, reflects the intensity LPECs, with higher values indicating greater regional risks of such conflicts.

(1)

Complexity index

$${SC}_{CI}=\sum _{i=1}^m\sum _{j=1}^n{\displaystyle \frac{2\ln \left(0.25p_{ij}\right)}{\ln a_{ij}}}\times {\displaystyle \frac{a_{ij}}{A}}$$

(2)

The ${SC}_{CI}$ has been demonstrated to effectively represent the complexity of landscape patterns influenced by human disturbances [37]. Higher values typically reflect more intricate landscape configurations and stronger human-induced impacts on LPECs.

(2)

Vulnerability index

$${SC}_{FI}=\sum _{i=1}^m\sum _{s=1}^r{f}_{is}\times {\displaystyle \frac{a_{is}}{A}}$$

(3)

The ${SC}_{FI}$ indicates the extent of ecological function loss and potential damage to the internal ecological environment caused by external disturbances to land patches [38]. Higher fragility implies greater ecological benefit losses due to neighboring influences and a diminished capacity to maintain ecosystem services.

(3)

Stability index

$${SC}_{SI}=1-PD={\displaystyle \frac{n}{A}}$$

(4)

The ${SC}_{SI}$ represents the stability of the land use system and can be assessed through patch density (PD) [39]. Higher PD values suggest a greater level of spatial fragmentation, which corresponds to reduced stability in the land use system.

In the formula, $p_{ij}$ refers to the perimeter of patch j in space type i, while $a_{ij}$ donates the area of patch j in the same space type i. A represents the total area of the spatial unit, and $f_{is}$ reflects the vulnerability of each space type. Additionally, $a_{is}$ represents the area of a specific space type within the spatial unit, m is the total number of spatial types, and r denotes the number of space types. PD stands for patch density, and n indicates the number of patches in the i-th space type.

3.3. Subdivision of Evaluation Units

This study uses square grids as evaluation units and applies the grid analysis method in Fragstats 4.2 to assess PLECs levels. Since landscape indices in Fragstats are sensitive to grid size [40], the study compares land-use conflict indices at 3 km × 3 km, 6 km × 6 km, and 10 km × 10 km scales. A 10 km × 10 km grid was too coarse to reflect a detailed PLEC distribution, while a 3 km × 3 km grid resulted in excessive fragmentation. Based on previous research results [41,42] and taking into full account the actual conditions of the study area. The 6 km × 6 km scale was found to be the most suitable, producing 5996 evaluation units. After standardizing the three conflict indices, they were integrated into a comprehensive PLEC model to quantify spatial conflict levels in the study area.

Following the approach of relevant scholars [26,33], the final PLEC values were standardized to a range of (0, 1). The natural breaks classification method was subsequently applied to classify the spatial conflict levels into five categories (

Table 3. Threshold and performance of each level of spatial conflict.

Conflict Type	Threshold Interval	Space Landscape Patch Performance
Weak spatial conflict	(0, 0.15]	Considered to be minimally disturbed, with a complete landscape patch structure and high stability
Mild weak spatial conflict	(0.15, 0.32]	Intersecting with areas of weak spatial conflict
Medium spatial conflict	(0.32, 0.43]	The fragmentation of landscape patches increases, with complex patch boundaries.
Mild strong spatial conflict	(0.43, 0.60]	The fragmentation, vulnerability, and complexity of landscape patches increase
Strong spatial conflict	(0.60, 1]	Landscape patches are fragmented and isolated, with complex structures and high fragmentation levels.

).

3.4. Geographically Weighted Regression Model

The Geographically Weighted Regression (GWR) model is widely used in studies exploring spatial heterogeneity. It serves as an optimization and enhancement to traditional regression models, which cannot quantitatively identify spatial variations in the influence of factors [43]. To investigate the driving mechanisms behind the distribution patterns of PLECs, this study employs the GWR model for analysis. The calculation formula is as follows:

$$y_i={\beta }_0\left(u_i,v_i\right)+\sum _j{\beta }_j\left(u_i,v_i\right)x_{ij}+{\epsilon }_i$$

(5)

In the formula, $y_i$ represents the dependent variable of grid unit i; $\left(u_i,v_i\right)$ are the geographical coordinates of grid unit i; ${\beta }_0$ is the constant term; ${\beta }_j$ is the coefficient of the j-th independent variable; $x_{ij}$ is the value of the j-th independent variable in grid unit i; and ${\epsilon }_i$ is the random error term.

3.5. PLUS Model

The PLUS prediction model is a land-use simulation model based on patch generation, used to simulate changes in land-use patches and analyze the potential driving factors behind land use/land cover (LULC) dynamics. This model improves upon traditional transfer rule mining methods, such as the Transfer Analysis Strategy (TAS) and Pattern Analysis Strategy (PAS), by introducing a new land expansion analysis strategy based on the Random Forest (RF) algorithm. The PLUS model retains the advantages of adaptive algorithms, utilizing mechanisms from the land-use simulation (FLUS) model, such as inertia competition and roulette wheel mechanisms, while integrating multi-type random patch seeds (CARSs) for fine-resolution modeling of various LULC types. As a machine learning method with strong fitting capabilities, the Random Forest algorithm is well-suited for uncovering the complex transfer rules of Cellular Automata (CA) models [44]. Several studies have shown that, compared to other models, the PLUS model more accurately simulates land-use conflict spatial patterns and evolution processes [45,46]. The specific process is shown in Figure 2.

3.5.1. Multi-Scenario Simulation of PLE Space

In the PLUS model’s multi-scenario simulation of PLE space, the first step is to convert the PLE space classification data into the required format. In the LEAS module, the expansion of PLE land is extracted, and eight driving factors (Figure 2) are input to explore the relationships between driving factors and spatial expansion using the Random Forest classification algorithm [44]. The Markov chain method is then employed to predict the future demand for PLE space in the watershed. Before simulating the future development of PLE space, the 2010 PLE land data for the region are used to model the PLE space pattern for 2020. The simulated results are compared with the actual data for 2020. After comparison, the kappa coefficient is found to be 0.867, indicating high simulation accuracy. Therefore, this model can be applied to simulate and predict the PLE space patterns.

3.5.2. Parameter Settings

(1)

Restricted Conversion Area Settings

The restricted conversion area is represented by a binary raster image consisting of 0 s and 1 s, which impose restrictions on the land-use types in specific areas of the study region. A value of 0 indicates that land-use type conversion is not allowed in that area, while a value of 1 indicates that conversion is permitted [47]. In this research, the natural development scenario does not include any restricted areas.

(2)

Land Demand Parameter Settings

The Markov chain model is used in this study to obtain the “Production–Living–Ecological” (PLE) land demand data for the prediction phase. The demand for each land-use type is simulated using the Markov model and then input into the PLUS model to simulate and predict the PLE land pattern.

(3)

Neighborhood Weight Parameter Settings

Neighborhood weight is a key indicator used to assess the potential for expansion between different land-use types. The range of values typically falls around 0.1. The closer the neighborhood weight of a land-use type is to 1, the stronger its expansion ability; conversely, the closer the neighborhood weight is to 0, the weaker the expansion ability. Since the expansion abilities of different land-use types vary, the neighborhood weight values for the PLE land-use types in the watershed will differ. These values can significantly impact simulation accuracy and need to be repeatedly adjusted and compared. The neighborhood weight parameter settings in this study are based on related research findings [48,49] and were determined through multiple tuning experiments (

Table 4. Configuration of domain weighting parameters.

Space Type	Weights
Living production space	1
Ecological production space	0.3
Production ecological space	0.5
Ecological space	0.4

).

(4)

Transition Matrix Parameter Settings

The transition matrix reflects the mutual conversion and direction of various land-use types. In practice, the transition probabilities between land-use types vary. Typically, the values are represented as “0” or “1”, where “0” indicates that conversion is prohibited and “1” indicates that conversion is allowed.

3.5.3. PLUS Model Multi-Scenario Forecasting Settings

Future land-use changes are influenced by various factors, and land-use transitions have multiple possibilities. Based on related research on the PLUS model [50] and considering the regional context, this study constructs three development scenarios: natural development, cropland protection, and ecological protection, to simulate the PLE space pattern in the region for 2030.

In the natural development scenario, the influence of human factors on future PLE space changes is not considered. Water bodies are set as restricted conversion areas, and it is assumed that all land-use types can convert into one another. The conversion cost matrix is shown in

Table 5. Conversion rules between land use types.

Type	ND Scenario			CP Scenario				EP Scenario
	LP	EP	PE	E	LP	EP	PE	E	LP	EP	PE	E
LP	1	1	1	1	1	1	1	1	1	1	1	1
EP	1	1	1	1	1	1	1	1	0	1	0	0
PE	1	1	1	1	0	0	1	0	1	1	1	1
E	1	1	1	1	1	1	1	1	1	1	1	1

Note: In the table, LP stands for living–production space; EP stands for ecological–production space; PE stands for production–ecological space; E stands for ecological space.

. After multiple iterations, the simulated PLE space pattern for 2030 is obtained.

In the ecological protection scenario, forest and grassland are set as restricted areas, and ecological production land is prohibited from converting to other land-use types. The neighborhood factor for living and production land is appropriately increased. Specific settings are shown in Table 5.

In the cropland protection scenario, the protection of cultivated land resources is the primary development goal for ensuring national food security and the sustainable development of the ecological environment. In this scenario, permanent basic farmland is set as a restricted area, and production–ecological land is not allowed to be converted into other land-use types. The neighborhood factor for production–ecological land is appropriately increased. Specific settings are shown in Table 5.

4. Results

4.1. Changes in the PLE Space Pattern

Figure 3 shows the distribution of PLE spaces in the Tuha region from 2000 to 2020. The analysis reveals that the overall structure of PLE spaces in the Tuha region remained relatively stable without significant changes over the 20-year period, with ecological–production space and ecological space being dominant. The combined area of these two types of spaces accounted for 98% of the total area. Living–production space and ecological–production space were concentrated in areas with favorable natural conditions. As social development accelerated, these two space types gradually expanded into surrounding areas, while ecological space was distributed across various other spaces.

Looking at the quantity structure of the PLE space types (

Table 6. Statistics of the PLE space classification system of the research area based on land use function in the study area from 2000 to 2020.

Space Type	2000		2010		2020
	Area/km2	Proportion/%	Area/km2	Proportion/%	Area/km2	Proportion/%
LP	522.383	0.25	556.022	0.27	977.337	0.47
EP	39,396.8	19.04	39,152.1	18.92	39,126.5	18.91
PE	2691.02	1.30	3076	1.49	3053.33	1.48
E	164,336	79.41	164,162	79.33	163,785	79.15
Total	206,946.2	100.00	206,946.2	100	206,942.2	100

), the area of ecological space in the study area from 2000 to 2020 accounted for more than 79% of the total area, making it the largest proportion. Ecological–production space ranked second, maintaining a proportion of more than 18.5%. Production–ecological space ranked third among the four space types, with a share of less than 2% of the total area. Living–production space had the smallest area, accounting for around 0.5%.

In terms of the changes in the four space types over time, the area of living–production space showed a continuous upward trend, while the area of production–ecological space first increased and then stabilized. From 2000 to 2020, the proportion of living–production space increased from 0.25% to 0.47%. From 2000 to 2010, the proportion of production–ecological space increased from 1.30% to 1.49%, and from 2010 to 2020, it changed slightly from 1.49% to 1.48%, remaining almost unchanged. With the acceleration of development, continuous improvements in living conditions, medical conditions, and infrastructure and the increasing population, the expansion of living–production space and production–ecological space was inevitable to ensure food production and living security. On the other hand, the area of ecological–production space continuously decreased, with its share dropping from 19.04% in 2000 to 18.91% in 2020. This decrease was partly due to the encroachment of living–production space, as well as production–ecological spaces. Additionally, harsh weather and overgrazing caused a significant loss of grasslands.

The area of ecological space also showed a continuous downward trend, with its share dropping from 79.41% in 2000 to 79.15% in 2020. This was primarily due to the conversion of ecological land types into other types of land, as the Tuha region, a critical area for national development (“three bases and one corridor”), exploited a large amount of natural resources, such as oil, natural gas, and photovoltaic energy. Additionally, sand control forests and protective forests were built, ultimately leading to a decrease in ecological space area.

4.2. PLEC Change Analysis

The conflict index of the PLE space in the study area from 2000 to 2020 was measured (

Table 7. PLE spatial conflict comprehensive index table.

Conflict Type	Conflict Classification	Number of Spatial Conflict Units			Proportion of Conflict Units (%)
		2000	2010	2020	2000	2010	2020
Weak spatial conflict	(0, 0.15]	709	696	711	11.82	11.61	11.86
Mild weak spatial conflict	(0.15, 0.32]	4112	4060	3691	68.58	67.71	66.06
Medium spatial conflict	(0.32, 0.43]	694	699	793	11.57	11.66	13.23
Mild strong spatial conflict	(0.43, 0.60]	360	384	393	6	6.40	6.55
Strong spatial conflict	(0.60, 1]	118	154	135	1.97	2.57	2.25
Total		5996	5996	5996	100	100	100

and Figure 4), and the results show the following:

In 2000, the level of PLE spatial conflict in the Tuha region was mainly characterized by mild weak spatial conflicts, which accounted for 68.58% of the total conflict units. This was followed by weak spatial conflicts and medium spatial conflicts, which accounted for 11.82% and 11.57%, respectively. The proportion of mild strong and strong spatial conflicts was the smallest. Most of the mild weak spatial conflict units were distributed in the southern part of the Tianshan Mountains and the northeastern region of the study area. The grid units mainly consisted of unused land types, which are primarily ecological, such as wastelands, sandy land, saline–alkali land, and bare land, and are relatively less disturbed by human activities. Therefore, the conflict level was mostly mild. Weak spatial conflicts were found in the northern mountainous areas and grasslands of the study region, with most conflict units located in Barkol and Yiwu counties. Medium spatial conflict units are distributed across various spaces in the study area, showing a scattered spatial distribution. Mild strong and strong spatial conflicts were mainly found in the densely populated oases of the study area, especially around Hami Yizhou District and the northern part of Turpan. These two types of spatial conflicts were mainly concentrated in areas with lower elevations where arable land and construction land were distributed, with living–production and production–ecological spaces being dominant. These areas were heavily disturbed by human activities, leading to higher conflict intensities.

In 2010, the level of PLECs in the Tuha region was still predominantly characterized by mild weak spatial conflicts, but the proportion decreased to 67.71% compared to 2000. Weak spatial conflicts decreased by 0.21%, while medium spatial conflicts increased by 0.09%. The combined proportion of mild strong and strong spatial conflicts increased by 1%, and these conflicts spread to surrounding towns and villages. Notably, some medium and mild strong spatial conflicts evolved into strong spatial conflicts by 2010 in towns such as Huoyanshan Town, Ilahu Town, Nanhu Township, Huayuan Township, Daqianwan Township, Qijiaojing Village, Kagetuoer Village, and some towns in Toksun County.

Compared to 2010, the proportions of medium, mild strong, and strong spatial conflicts in the study area increased in 2020. The number of medium spatial conflict units increased by 94, but mild weak spatial conflicts still dominated, accounting for 66.06%. When examining this conflict pattern evolution in relation to the changes in the PLE space, the increase in conflict intensity was mainly due to resource development and population growth, which led to the external expansion of living–production and production–ecological spaces such as towns, villages, and farmland. This expansion increased external pressure on the space, resulting in higher conflict levels. In summary, from 2000 to 2020, the overall PLECs in the Tuha region exhibited an increasing trend.

4.3. Spatial Clustering Characteristics of PLECs in the Study Area

Using GeoDa and ArcMap software for spatial autocorrelation analysis, the global Moran’s I values of spatial conflicts in 2000, 2010, and 2020 were 0.5623, 0.5670, and 0.5731 (p < 0.01), respectively. These results indicate a significant spatial positive correlation in the distribution of spatial conflicts within the study area, with an increasing clustering tendency over time. It is expected that this clustering will further intensify by 2030.

An analysis of the local spatial autocorrelation clustering patterns reveals (Figure 5) that, at a 99% confidence level, spatial conflicts are primarily characterized by high–high and low–low clustering. High–low and low–high clustering are rare, scattered, and exhibit no significant patterns. Based on the conflict maps and the study area’s overview map (Figure 1) for the period 2000–2020, it is evident that high–high clustering areas are mainly located in the urban centers and surrounding areas of Gaochang District, Toksun County, Shanshan County, and Yizhou District. These high–high clustering zones show a gradual expansion trend, especially in the transitional areas between urban areas and farmland in Gaochang and Yizhou. Low–low clustering areas are distributed across most parts of the Tianshan Mountains and the northern parts of Barkol County. These regions primarily consist of grasslands and forested ecological production spaces, where ecosystems are relatively stable, human activities are minimal, and the ecological carrying capacity is strong. Regions with insignificant clustering are dominated by ecological space but also feature a mix of other space types.

4.4. Geographically Weighted Regression (GWR) Model

The GWR model can better reflect the local patterns of explanatory variables that influence the intensity of PLECs. Following the previous step, where the spatial clustering characteristics of PLECs in the study area were identified through correlation analysis, the GWR model was employed to delve deeper into the driving mechanisms underlying this pattern. In Figure 6, the regression coefficients of the explanatory variables reflect the extent to which these variables explain the PLECs. Positive coefficients indicate that an increase in the explanatory variable will intensify the PLECs, while negative coefficients suggest that an increase in the explanatory variable will reduce the PLECs. In analyzing the influencing factors, this study selects eight driving factors from both natural and socio-economic aspects.

Table 8. GWR model results.

Factors	R2	R2 Adjusted	Regression Coefficient
NDVI	0.76	0.67	0.00132
DEM	0.41	0.38	−1.8848
GDP	0.58	0.54	0.01297
POP	0.57	0.53	0.00256
Aspect	0.73	0.62	−1.7135
Road	0.77	0.68	0.08904
Slope	0.73	0.63	−0.00272
Precipitation	0.38	0.36	−0.00027

shows the overall goodness of fit (R²) for each factor.

Through the regression coefficient map of the driving factors in Figure 6 and the LISA clustering map, it can be observed that factors such as A2, A3, A6, and A7 have a positive impact on PLE spatial conflict. A7 indicates that areas with denser vegetation cover tend to have higher levels of PLECs. This is because the Tuha region is a typical oasis city, especially evident in Yizhou District, Gaochang District, and the urban areas of Toksun County and their surroundings. The positive regression coefficients of A3 and A6 are mainly concentrated in the central areas of the two districts and four counties, suggesting that areas with active economic development and high population density are generally more prone to spatial conflicts. In contrast, negative regression coefficients are distributed in ecological spaces and ecological production spaces. Factors A1, A4, A5, and A8 are negatively correlated with spatial conflicts, indicating that low-elevation areas are more suitable for construction, agricultural, or industrial activities, which leads to higher land use demand. As the elevation rises, PLEC decreases.

4.5. PLECs Pattern Multi-Scenario Simulation

The results obtained from the PLUS model simulation (Figure 7), combined with the spatial conflict measurement model, show the distribution pattern of PLECs in the Tuha region under different development scenarios for 2030. The results exhibit high similarity to the characteristics of the PLECs pattern in Tuha from 2000 to 2020, but with a noticeable overall increase in conflict (Figure 8).

In the 2030 scenario under different development conditions, the mild weak PLEC zones are mainly distributed in the higher altitude areas of the Tianshan Mountains and the grasslands in the northwest of Barkol County. The stronger and high-level PLEC zones are still primarily concentrated in the urbanized areas and their surrounding regions of each district and county, with significant intensification in the southern and southeastern parts of Yizhou District. This is due to the conflict between ecological space and living–production space caused by natural resource development. The distribution of PLEC conflicts overlaps with the concentrated areas of socio-economic development, and its spatial diffusion is significantly limited by topography. Influenced by the natural environmental baseline, PLECs cannot overcome the altitude limitations and spread into higher terrain areas, instead expanding into relatively flat regions around the existing conflict zones.

The ND scenario continues the historical development trend of the region. The mild strong spatial conflict and strong spatial conflict zones expand outward in all directions, with their area continuously increasing. Compared to 2020, the combined proportion of these zones increased by 6.39%, reaching 15.19% (

Table 9. Conflict index of PLE space under different scenarios.

Conflict Type	Conflict Classification	2020 (%)	ND (%)	cp (%)	EP (%)
Weak spatial conflict	(0, 0.15]	11.86	12.06	12.88	11.86
Mild weak spatial conflict	(0.15, 0.32]	66.06	52.69	48.15	32.10
Medium spatial conflict	(0.32, 0.43]	13.23	20.06	27.35	39.44
Mild strong spatial conflict	(0.43, 0.60]	6.55	10.67	9.31	13.98
Strong spatial conflict	(0.60, 1]	2.25	4.52	2.32	2.62
Total		100	100	100	100

). Meanwhile, the areas with mild weak spatial conflict showed a significant reduction trend. In the ND scenario, the regional economy and society develop according to the historical evolutionary trend (Table 9), with the polarization effect of regional economic and social development continuing to intensify. Due to topographical constraints, the mild strong spatial conflict and strong spatial conflict zones spread outward, increasing landscape vulnerability and, consequently, the intensity of conflicts. This phenomenon is most apparent in the central urban areas and the central regions of each district and county, showing a clear pattern of expansion overall (Figure 8).

In the CP scenario, the mild weak spatial conflict dominates in the Tuha region, followed by medium spatial conflict. From 2020 to 2030, the area of weak spatial conflict significantly decreases, primarily in the central and eastern parts of the study area. This scenario focuses on ensuring food security and restricts non-agricultural land use from encroaching on farmland. There is no significant increase in mild strong spatial conflict and strong spatial conflict zones in the study area under this scenario.

In the EP scenario, the area of medium spatial conflict in the PLECs significantly increases (from 13.23% in 2020 to 39.44%), concentrating in the northeastern part of Barkol, the northern part of Yiwu County, the southeastern part of Yizhou District, and the southern part of Shanshan County. At the same time, the area of mild weak spatial conflict decreases, with a 26.62% reduction compared to 2020. This scenario prioritizes the protection of ecological production space, with grasslands and forests as restricted areas. As a result, living–production space and production–ecological space are further compressed, driving the expansion of medium spatial conflict zones.

5. Discussion

5.1. Identification of PLECs

This study adopts a 6 × 6 km grid scale to characterize PLECs, as the choice of scale in landscape pattern analysis is critical for studying changes in PLECs. Both excessively small and overly large grid scales may compromise the reliability and interpretability of results. A smaller grid scale (e.g., 3 × 3 km) can provide finer spatial resolution but may introduce greater random fluctuations, making it difficult to capture broader PLEC pattern [51]. Conversely, a larger grid scale [50] may be overly coarse, failing to adequately reflect land-use changes and local conflicts within the region [52]. Therefore, the 6 × 6 km scale represents a reasonable compromise between capturing detailed information and reflecting macro trends. Additionally, research indicates that land-use conflicts often exhibit spatial clustering. A 6 × 6 km grid scale can effectively capture the spatial distribution characteristics of conflicts without losing critical information due to excessively large or small scales [53].

The results indicate that from 2000 to 2020, PLECs exhibited high intensity in areas characterized by intensive human activity and high levels of economic development, with conflict intensity continuously increasing over the 20-year period. These findings indicate that the changes in PLECs are influenced by both human activities and natural environmental factors, consistent with the conclusions of prior research [54]. The results also indicate that the expansion of conflict areas within the study region is concentrated in urban-rural transitional zones and agro-pastoral ecotones, reflecting conflicts between the primary and secondary industries. On the one hand, the national emphasis on food security and increasing agricultural production scales has intensified conflicts between living–production spaces and production–ecological spaces [54,55]. On the other hand, the region’s abundant resources, such as petroleum, natural gas, and photovoltaic energy, have driven extensive natural resource development, leading to significant encroachment of ecological spaces by living–production spaces.

From 2000 to 2020, the proportion of production–ecological spaces in the study area exhibited a continuous decline. This trend can be attributed to two primary factors: first, extreme arid weather conditions, such as reduced vegetation cover and accelerated glacial retreat; second, issues including unregulated tourist site development, overgrazing, and low water resource utilization efficiency. These findings are consistent with the results of studies [30,56] on land-use changes in arid regions.

5.2. Key Factors Influencing PLECs

The results in Section 4.4 indicate that the primary driving factors of PLECs include natural environmental conditions and socioeconomic development activities. This highlights the interrelationship between PLE spatial patterns and both natural environments and economic activities, resulting in differing degrees of PLECs across various periods and domains. These findings align with the conclusions of prior studies [57].

Among natural environmental factors, elevation and slope exert a significant inhibitory effect on PLECs. The findings show that PLECs intensify and then gradually alleviate as slope increases, with these dynamics primarily observed in plain areas, consistent with findings from previous research [58]. Low-elevation and low-slope regions are suitable for urban construction, industrial development, and agricultural production, which may explain the low regression coefficients for elevation in areas with high PLECs within the study region. This suggests that PLECs decrease as elevation rises, aligning with studies indicating that the expansion of living–production spaces intensifies PLECs [59,60].

Additionally, the normalized difference vegetation index (NDVI) demonstrates a promotion effect on PLECs, attributed to the unique geographic and climatic conditions of the Turpan–Hami region. High evaporation rates and low precipitation necessitate engineering solutions such as karez systems to improve water-use efficiency. These factors contribute to high vegetation coverage in residential areas and oasis construction on flat terrains. Moreover, much of the region consists of barren land and sandy areas, resulting in low overall vegetation coverage, which explains the high NDVI regression coefficients in areas with elevated PLECs. Precipitation is concentrated in the eastern Tianshan Mountains and northern areas, while the southern region experiences minimal and unevenly distributed rainfall. This uneven distribution creates spatial competition for water resources, contributing to elevated PLECs. The negative mean regression coefficients for precipitation in the GWR results (Table X) reflect this pattern. In these areas, changes in precipitation significantly impact primary economic activities like agriculture and animal husbandry. Increased precipitation may improve economic returns and subsequently reduce competition for urban spaces. This observation is consistent with findings from [61], which demonstrated the influence of spatiotemporal precipitation distribution on urbanization processes. Temperature and evaporation also act as driving factors of PLECs in arid regions. Drought accelerates desertification in inland arid areas [62,63]. Although human activities have relatively limited direct effects, they exacerbate the already harsh ecological conditions.

Among socioeconomic factors, GDP, population, and distance from roads all exert a positive driving effect on PLECs. This aligns with previous studies indicating that economic development is a key driving force behind the occurrence of conflicts [64,65]. As regional economies develop, the demand for land resources increases, and the excessive expansion of living–production spaces threatens the stability of landscape patterns, accelerating fragmentation and complexity [66,67,68], which in turn leads to prominent human–environment conflicts. These findings are consistent with earlier research [59]. It is noteworthy that although population growth exacerbates changes in the PLE spatial pattern, the implementation of scientific spatial planning can optimize national territorial spatial patterns [69,70], promoting sustainable land development. This human intervention approach is considered beneficial. Analyzing PLECs changes from 2000 to 2010 reveals that the growth of conflicts during this period was less intense than in the previous decade, suggesting that population growth and land sustainability were better coordinated during this time.

5.3. Optimal Allocation Scheme for PLE Spaces

Multi-scenario simulation methods can assist policymakers in selecting and implementing scientifically sound spatial planning to address the increasingly growing PLECs [24,70]. This study indicates that under the CP scenario, the degree of PLECs is the lowest and most stable (Figure 8). In the ND scenario, as the area of living–production spaces continues to expand, high-value PLECs progressively spread across the space. In the EP scenario, the protection of ecological production land is ensured, with strong and very strong conflicts being controlled, yet the proportion of moderate-conflict areas significantly increases. Therefore, the CP scenario is considered the direction for future development. Regulating regional PLECs should adhere to the principle of maximizing overall regional benefits, aiming to improve the general well-being of regional development. At the same time, targeted, multidimensional land-use management policies should be formulated based on national and regional policies.

5.4. Strategic Recommendations

Based on an analysis of local policies, from 2000 to 2010, rapid regional economic growth, population increase, expansion of construction land, acceleration of urbanization, and encroachment on ecological land all contributed to the exacerbation of PLECs. However, from 2010 to 2020, the rate of PLECs in the Tuha region significantly slowed down. The deceleration of PLECs during this period can be attributed to the adjustment of national policies, including increased regional investments and optimization of land structure. At the same time, policies on ecological civilization, afforestation, and population migration, issued by Xinjiang, have contributed to the restoration of the region’s ecosystem [71,72,73]. These developments suggest that PLECs in the Tuha region did not continue to deteriorate, with better coordination achieved between population growth, socioeconomic development, and land sustainability. However, it is important to recognize that PLECs are a cyclical phenomenon, requiring continuous monitoring to ensure their mitigation. Therefore, regional governments should prioritize PLEC issues in future national land spatial utilization and related planning, establish a PLEC mechanism suited to the region, and incorporate it into the planning and supervision system. Additionally, place-specific guidance for production and living activities should be provided.

To mitigate PLECs, future strategies should integrate regional and conflict-specific approaches. Urban areas should enforce zoning and expand green infrastructure; rural and agro-pastoral zones need sustainable land use and compensation mechanisms; ecologically sensitive areas must be strictly protected. Different conflict types can be managed through ecological redlines, compact development, and mandatory environmental assessments. At the policy level, establishing monitoring systems, encouraging participatory planning, and applying adaptive land-use tools will help shift from conflict identification to active governance and long-term prevention.

5.5. Limitations of the Study

This study, based on landscape ecology theory, adopts a landscape pattern perspective to construct a spatial conflict measurement model using landscape pattern indices, and assesses and predicts the spatial conflict levels in the Tuha region. However, when measuring spatial conflict levels, the study did not integrate multiple perspectives, nor did it employ both qualitative and quantitative methods for in-depth analysis. To provide a comprehensive evaluation of conflict levels, future research should attempt to combine qualitative and quantitative approaches and introduce indicator-based evaluation systems, in order to develop a more comprehensive spatial conflict assessment framework. In addition, future studies should address spatial land-use patterns, socioeconomic development, ecological environmental protection, and other dimensions to formulate corresponding regulatory strategies. Furthermore, there is a need to explore more effective ways to apply these strategies in spatial optimization and simulation research [74], which will be an important direction for future studies.

6. Conclusions

This study takes the Turpan and Hami regions as the research area and builds a LPECs measurement model based on landscape patterns. The temporal and spatial evolution characteristics and patterns of LPECs are analyzed at the grid scale. The PLUS model is used to simulate the evolution of LPECs in the Tuha region under different scenarios for 2030. The main conclusions are as follows:

(1)

From 2000 to 2020, the area of ecological space in the study area was the largest, accounting for 79% of the total area, followed by ecological–production space, production–ecological space, and living–production space.

(2)

From 2000 to 2020, the conflict level in the study area was predominantly mild weak conflict, with strong conflicts initially increasing and then decreasing. Medium and mild strong conflicts continued to grow, while weak conflicts consistently decreased.

(3)

The spatial conflict distribution in the study area from 2000 to 2020 shows a significant positive correlation. Analysis of the local spatial autocorrelation clustering map and the results from the GWR driving factors indicates that high–high clustering areas are mainly located in the built-up areas of Gaochang District, Toksun County, Shanshan County, and Yizhou District, along with their surrounding areas, with slight expansion. Low–low clustering areas are distributed across most regions of the Tianshan Mountains and the northern part of Barkol Kazak Autonomous County. Areas with non-significant clustering are primarily ecological spaces, which also intersect with other spatial types. Factors such as NDVI, GDP, population, and distance to roads have a positive impact on PLECs, while factors such as elevation, slope, direction, and precipitation exhibit inhibitory effects.

(4)

In the cropland protection scenario under different development contexts, the spatial conflict level in the study area is the lowest, with weak and moderate conflict zones accounting for 75.5%. This scenario is considered the optimal direction for future regional development. The regulation of land use conflicts in the region should adhere to the principle of maximizing overall regional benefits, with the goal of improving the overall welfare of regional development. Additionally, it is crucial to formulate a targeted and multidimensional land use management policy system based on national and regional policies.