Partial Order Ranking of the Key Drivers of Grassland Conversion in the Urban–Grassland Interface: A Case Study of the Hohhot–Baotou–Ordos Region

Abstract

Identifying and ranking the key drivers of grassland conversion at the county level is crucial for developing targeted policies and improving protection efficiency. However, this process faces methodological challenges because of spatial and temporal variability. Partial order theory offers a robust framework for addressing these complexities. This study applies partial order theory (POT) combined with the Hasse diagram technique (HDT) to analyze grassland conversion in the Hohhot–Baotou–Ordos region during two time periods (2000–2010 and 2010–2020). First, patterns of grassland transformation are quantified, and the dominant driving factors of grassland conversion out (GCO) are identified and ranked, highlighting regional differences and temporal shifts. By integrating POT and HDT, this study offers a novel approach to handling complex, nonlinear, and hierarchical relationships among multiple drivers. The results provide scientific insight and policy recommendations for region-specific grassland management and sustainable land-use planning. The results show that (1) transitions between grasslands and other land-use types became more frequent across the two periods. Specifically, the rates of grassland conversion out and conversion increased from 2.1% and 3.5% during the period 2000–2010 to 4.7% and 4.8% during the period 2010–2020, respectively. (2) Urbanization was the primary driver of grassland conversion in 11 and 10 of the 18 counties during the first and second periods, respectively, followed by factors related to weather variables. (3) In the future, the eastern region of the study area needs to prioritize mitigating the impacts of urban development, while the western region should focus on enhancing ecological construction projects. This study recommends adopting region-specific ecological protection and economic strategies for balanced outcomes in conservation and development.

1. Introduction

Research on grassland conversion is essential to understanding the drivers of land-use change, as grasslands play a crucial role in ecosystem services and material production. Grassland conversion refers to the transformation of natural or semi-natural grasslands into land with other uses, such as agriculture, urban development, or energy production sites. This process is typically accompanied by disruptions to ecosystem structure and function and can be categorized into grassland conversion out (GCO) and grassland conversion in (GCI) [1]. This process encompasses both grassland loss and gain. GCO has an important effect on ecosystem services, including biodiversity loss [2], reduced carbon storage capacity [1], disruptions to water cycles [3], and soil degradation [4]. Investigating the drivers of grassland conversion is essential for understanding the underlying mechanisms of land-use change and formulating sustainable land management policies.

Grassland degradation and grassland conversion are two distinct concepts. Grassland degradation refers to the process by which grassland ecosystems are damaged due to physical stressors (e.g., overgrazing and trampling) or changes in growth conditions (e.g., climate change), leading to significant declines in grassland coverage or its transformation into other land-use forms [5,6]. In contrast, grassland conversion refers to the transformation of grassland ecosystems into other land-use types, such as croplands, construction lands, or barren lands, driven by changes in land-use practices [7]. Grassland conversion is often the result of trade-offs between regional development and ecological conservation [8]. Although grassland conversion may bring short-term economic benefits, its unsustainable expansion can disrupt the ecological balance, weaken ecosystem services, and adversely affect the sustainable development of pastoral social economies in the long term [9]. Grassland conversion is not only a manifestation of land-use change but also has a profound impact on ecosystem functions, regional economic development, and the effectiveness of social policies [10,11]. Therefore, investigating the spatiotemporal distribution patterns and driving factors of regional grassland conversion can provide a scientific basis for optimizing regional policy adjustments and offer theoretical support for identifying key conflicts and solutions in processes related to land-use and land-cover change (LUCC) [12,13].

The driving factors of land-use change exhibit five key characteristics. First, they demonstrate integrality, meaning that multiple factors are interconnected and jointly drive land-use change rather than the process being influenced by a single factor. Second, these drivers exhibit dynamism, evolving over time in response to changes in environmental conditions, policies, and socioeconomic contexts. Third, they display hierarchical organization, where different factors contribute according to an ordered structure based on their functions and significance. Fourth, synergistic effects often emerge among the driving factors, amplifying their overall impact and reflecting the complex interplay between societal, economic, and natural systems. Last, the spatial heterogeneity of these factors highlights regional variations in natural conditions and socioeconomic contexts, necessitating region-specific and customized land-use management strategies. Understanding these characteristics provides critical insights for formulating adaptive and sustainable land-use policies that are tailored to diverse regional contexts [14,15].

The existing methods for studying the driving factors of grassland conversion face several significant challenges. First, many studies rely on statistical analysis [16], machine learning [17], and field surveys [18], but these methods are often constrained by linear assumptions [19], heavy dependence on data [20], or inefficiency and high costs [21], which limit the exploration of complex nonlinear relationships in grassland conversion drivers. Second, the existing research often focuses on individual driving factors [22], neglecting interactions among multiple factors that hinder researchers from developing a comprehensive understanding of the intricate mechanisms behind grassland conversion. Finally, many approaches fail to account for the integrated effects of multiple factor groups, leading to incomplete and inaccurate interpretations of the conversion process.

In summary, the complexity of analyzing the drivers of grassland change lies not only in the multifaceted nature of the driving factors themselves but also in the limitations of the analytical methods used. To address these challenges, this study employs partial order theory (POT) and the Hasse diagram technique (HDT). POT establishes partial order relationships among elements based on multiple criteria, avoiding biases from linear assumptions and subjective weighting, thereby ensuring analytical objectivity and data-driven insights [23]. POT effectively handles multifactor interactions and nonlinear relationships, providing a robust tool for determining the drivers of grassland conversion. By integrating the HDT, POT can visually represent the hierarchical and relational structure of the driving factors, facilitating a deeper understanding of the dynamic processes underlying grassland conversion [10]. POT and the HDT have been widely applied in various fields, including the assessment of chemical hazards [24], public policy formulation [25], water quality evaluations [26], and the identification of drivers of land-use change [9].

The Hohhot–Baotou–Ordos (HBO) region in Inner Mongolia serves as an ideal case study for exploring grassland conversion. Characterized by diverse grassland ecosystems, arid and semi-arid climates, and severe water scarcity, the region is highly sensitive to climate change and human activities, as are its grassland ecosystems [27]. Grassland conversion in this region is extensive, and it is driven primarily by agricultural expansion, energy development, urbanization, and infrastructure construction. These changes have exacerbated vegetation loss, soil erosion, and ecosystem degradation while intensifying conflicts over limited pasture resources in traditional livestock farming [28]. As the most economically developed region in Inner Mongolia, HBO faces significant demand for grassland because of energy exploitation and urban expansion, further intensifying conflicts over limited pasture resources and grazing pressure in livestock farming [29]. Recent efforts have been made to combat these issues, such as restoring croplands to grasslands, grazing bans, and ecological compensation policies to mitigate grassland conversion and its ecological impacts; however, the rapid pace of regional economic development has limited the effectiveness of these measures [30]. Therefore, using HBO as a case study for analyzing the driving factors of grassland conversion not only allows us to determine the mechanisms of grassland-use changes and evaluate the effectiveness of ecological policies but also provides scientific guidance for optimizing regional land-use policies to achieve economic–ecological synergies. Additionally, this research offers valuable insights into grassland conservation and sustainable use in similar ecosystems globally [31].

On this basis, we hypothesize that the transformation of grasslands is driven by multiple interacting socioeconomic and environmental factors and that the application of POT and the HDT can allow us to identify and rank these complex interactions effectively. Based on this hypothesis, this study aims to systematically investigate the driving mechanisms of grassland-type conversion in the Hohhot–Baotou–Ordos region. The main objectives of this study are as follows: (1) to reveal the transformations that occurred in the grasslands of the Hohhot–Baotou–Ordos region between 2000 and 2020; (2) to identify and rank the key drivers of grassland conversion in the study area during the periods of 2000–2010 and 2010–2020; (3) to show how the combination of POT and the HDT is applied to the identification of the factors driving grassland conversion out (GCO).

2. Materials and Methods

2.1. Study Area

HBO is located in the middle and upper reaches of the Yellow River in Inner Mongolia, spanning 37°37′–42°44′ N and 106°31′–112°17′ E, with a total area of 1.32 × 10⁵ km² [32] (Figure 1). The HBO region has a semi-arid climate, with an average annual precipitation of 300 mm and an annual mean temperature of 8 °C [33]. Since 2000, Inner Mongolia has positioned the HBO region as a core area to promote the development of a regional economic cluster with distinct characteristics. Capitalizing on their respective resource advantages, the three cities have rapidly developed into the most dynamic urban economic cluster in Inner Mongolia, often referred to as the “Golden Triangle” [34]. Hohhot, the capital of the Inner Mongolia Autonomous Region, serves as a strategic hub connecting the Yellow River Economic Belt, the Eurasian Land Bridge, and the Bohai Economic Rim. It is also an important gateway for trade with Mongolia and Russia [35]. Baotou is renowned for hosting the world’s largest rare earth deposit, the Bayan Obo Mine, which contains 43.5 million tons of reserves, accounting for 83.7% of China’s and 37.8% of the world’s total [36]. Ordos is notable for its abundant coal resources, which account for one-sixth of China’s proven reserves, making it a key region with an economy centered on coal mining [37]. The primary land-use and land-cover type in the HBO region is grassland, accounting for 60% of the total study area. Bare land ranks second, covering 25% of the total area.

For ease of analysis, we use the abbreviations of the names of 18 banners, counties, and cities in the study area (hereafter, refer to Figure 1). Among them, Baotou, Hohhot, and Ordos refer to the three regions, including four, six, and eight counties or banners, respectively. HH (Hohhot) refers to the capital city, while BT (Baotou) and OD (Ordos) refer to the cities.

2.2. Framework

This study aimed to analyze the spatiotemporal characteristics of GC from 2000 to 2020, identifying GCO driver groups in the periods from 2000 to 2010 and 2010 to 2020. Based on this, our research was designed as follows: First, we analyzed the spatial and temporal LUCC patterns during 2000–2010 and 2010–2020 before exploring GCO and GCI in more detail based on LUCC processes. Second, the main drivers of GCO were placed into four categories: weather variables (temperature and precipitation), human activities (population density and sheep density), economy (GDP as well as primary, secondary, and tertiary industries), and urbanization (distance to urban center, rural areas, roads, and mining areas). We used POT and HDT to rank all driver groups and identify the dominant drivers for GCO in the two periods (Figure 2).

For 2000–2020, the driving factors are defined using 1 × 1 km² pixels. IDW interpolation converts the weather variables and economic data into density values, the livestock density is converted to sheep units, and distances to urban, rural, road, and mining areas are calculated using ArcGIS. Precipitation and urbanization are negatively correlated with GCO, while other factors show positive correlations. All factors are processed using the WGS_1984_Albers projection.

2.3. Data Sources and Processing

2.3.1. Land-Use and Land-Cover Change Data and Processing

This study utilized the Global 30 m Land-Cover Dynamics Monitoring Product (GLC_FCS30D), developed by the research team led by Liu Liangyun, as the primary data source (https://data.casearth.cn/, accessed on 10 October 2024). The dataset integrates continuous change detection technology and Landsat imagery to capture the temporal points of pixel changes accurately. Updates are conducted using high-confidence, globally distributed training samples [38]. The dataset updates employ local adaptive classification models and temporal consistency optimization algorithms to enhance temporal stability, with rigorous validation applied to the results. Within the basic classification system (comprising 10 major land-cover types), the dataset achieved an overall accuracy of 80.88% (±0.27%). For the Level 1 classification system (including 17 land-cover types), the accuracy was 73.04% (±0.30%). This study analyzed grassland conversion in the HBO region using land-use/land-cover data from three key years: 2000, 2010, and 2020. Based on the land-use type definitions given by Liu J. Y. et al. [39], land-cover types in this study were categorized into six primary classes: cropland, forestland, grassland, bare land, water bodies, and impervious surfaces. Detailed classification results are presented in

Table 1. Descriptions of land-use classification and reclassification.

Basic Classification System	Level 1 Validation System	Fine Classification System	ID	Description	Merged Category
Cropland	Rainfed cropland	Rainfed cropland	10	Land on which crops are grown	Cropland
		Herbaceous cover cropland	11
	Irrigated cropland	Irrigated cropland	20
Forest	Deciduous broadleaved forest	Open deciduous broadleaved forest	62	Refers to the growth of trees, shrubs, bamboo, coastal mangrove forests and other forestry land	Forest
	Evergreen needle-leaved forest	Closed evergreen needle-leaved forest	71
		Open evergreen needle-leaved forest	72
	Deciduous needle-leaved forest	Open deciduous needle-leaved forest	82
Shrubland	Shrubland	Shrubland	120
Grassland	Grassland	Grassland	130	Refers to all kinds of grassland, mainly growing herbaceous plants and covering more than 5%, including shrubland grassland, mainly grazing and sparsely forested grassland with a canopy of less than 10%	Grassland
Bare areas	Sparse vegetation	Sparse vegetation	150	Land that has not yet been used, including land that is difficult to use
	Bare areas	Bare areas	200		Bare areas
		Unconsolidated bare areas	202
Wetland	Inland wetland	Marsh	182	Refers to natural land waters and water conservancy facilities	Waterbody
		Flooded flat	183
Waterbody	Waterbody	Waterbody	210
Impervious surface	Impervious surface	Impervious surface	190	Refers to urban and rural residential areas and other industrial, mining, transportation, and other land	Impervious surface

.

2.3.2. Identifying Potential Drivers

Driving factors refer to elements that directly or indirectly cause changes in biodiversity [40]. Grassland conversion is influenced by multiple factors, including weather variables, human activities, economic development, and urbanization (

Table 2. Collection and processing of data related to driving factors.

Groups of Driving Factors	Data Name	Unit	Year	Data Source	Orientation
Weather Variable-Related Driving Factors	Temperature	°C	2000, 2010, 2020	https://cds.climate.copernicus.eu/, accessed on 23 October 2024	The greater the value, the greater the impact on land degradation
	Precipitation	mm	2000, 2010, 2020		The greater the value, the smaller the impact on land degradation. Use 1—the corresponding value
Human Activity-Related Driving Factors	Population Density	Person/km2	2000, 2010, 2020	http://www.resdc.cn/, accessed on 10 November 2024	The greater the value, the greater the impact on land degradation
	Livestock Density	Sheep unit/km2	2000, 2010, 2016	Inner Mongolia Statistical Yearbook http://tj.nmg.gov.cn/, accessed on 15 November 2024
Economic Development-Related Driving Factors	Primary Industry GDP Density	CNY 104/km2	2000, 2010, 2020
	Secondary Industry GDP Density	CNY 104/km2	2000, 2010, 2020
	Tertiary Industry GDP Density	CNY 104/km2	2000, 2010, 2020
	GDP Density	CNY 104/km2	2000, 2010, 2020
Urbanization-Related Driving Factors	Distance to Urban Land	m	2000, 2010, 2020	http://www.resdc.cn/, accessed on 20 November 2024	The greater the value, the smaller the impact on land degradation. Use 1—the corresponding value
	Distance to Rural Settlements	m	2000, 2010, 2020
	Distance to Roads	m	2000, 2010, 2020
	Distance to Industrial and Mining Land	m	2000, 2010, 2020

). Specifically, weather variables have significantly exacerbated grassland degradation and desertification in the HBO region, undermining its function as an ecological security barrier [41]. The combined effects of population growth and high grazing intensity have driven grassland degradation, increasing the vulnerability of ecosystems in the region [42]. Economic growth has increased the demand for land resources, particularly for agricultural and commercial purposes, leading to the conversion of grasslands into agricultural or commercial land. Rapid urbanization has driven urban expansion, resulting in the development of large areas of grassland into residential, industrial, and infrastructural zones, significantly altering land-use patterns in these areas [43].

Based on an extensive literature review [13,44,45] and the dynamic, synergistic, holistic, hierarchical, and integrated characteristics of driving factors, this study identified twelve key factors and classified them into four categories: climatic factors (annual precipitation and mean annual temperature), human activity factors (population density and year-end livestock inventory), economic factors (total GDP, primary industry output, secondary industry output, and tertiary industry output), and urbanization factors (distance to urban land, rural settlements, roads, and industrial/mining land). To align with the grassland conversion process, all the driving factors were analyzed within the temporal range of 2000 to 2020. However, due to limitations in data availability, livestock density data for the HBO region were only available from 2000 to 2016 (Figure S1).

2.4. Methodologies

2.4.1. Assessment of Grassland Conversion Processes Between 2000 and 2020

This study analyzed GCO and GCI based on the land-use patterns from 2000 to 2020. GCO refers to the process of grassland being converted into other land-use types, such as cropland, forestland, water bodies, bare land, and impervious surfaces [6,46]. GCO is often accompanied by a reduction in the area of grassland and a loss of ecological functions. It is closely associated with agricultural expansion, urban development, and the exploitation of water resources. GCI refers to the process of other land-use types being converted into grassland, typically involving the transformation of cropland, forestland, water bodies, or bare land into grassland [47]. GCI represents a positive transition in land restoration, ecological rehabilitation, or land-use change, contributing to improved regional ecological stability and biodiversity.

2.4.2. Partial Order Theory

This study applied POT to analyze the complex relationships between driving factors and grassland conversion. Originating in mathematics, POT deals with sets equipped with partial order relationships, abstracting the intuitive concepts of ranking, order, or arrangement. This theory is particularly suitable for scenarios where elements are characterized by multiple comparative criteria [23].

In POT, each comparative relationship retains all the information about the elements being compared, eliminating the need to consider collinearity among elements. This approach allows for a more comprehensive consideration of the integrality, dynamism, hierarchy, synergy, and spatial heterogeneity of the drivers of grassland conversion.

In the present study, a poset: X = {a, b, c,… } is defined, where elements a, b, c represent counties or banners that can be mutually compared. The partial order relationships in this context must satisfy the following properties [48]:

• Reflexivity—a ≤ a, for all a ∈ X;
• Transitivity—if a ≤ b and b ≤ c, then a ≤ c, for all a, b, c ∈ X;
• Antisymmetry—if a ≤ b and b ≤ a, then a = b, for all a, b ∈ X.

2.4.3. Hasse Diagram Technique

A Hasse diagram is a visualization tool used to display partial order relationships [49]. Let X be a set containing a finite number of objects, and IB be a set of indicators qi (where i = 1, 2, …, |IB|): these objects and indicators can form a partially ordered set. In this study, the object set consists of GCO in the 18 counties of the HBO region during the periods 2000–2010 and 2010–2020, represented as P1 and P2, respectively. The indicator set contains 12 driving factors for grassland conversion, divided into four driving factor groups. The object sets are denoted as Counties_P1 and Counties_P2, and these four factor groups are denoted as IB_weather, IB_human, IB_economic, and IB_urbanization. Therefore, a total of 10 partial orders are formed: (Counties_P1, IB_weather), (Counties_P1, IB_human), (Counties_P1, IB_economic), (Counties_P1, IB_urban), (Counties_P2, IB_weather), (Counties_P2, IB_human), (Counties_P2, IB_economic), and (Counties_P2, IB_urban). The comparison between objects follows the following rules:

$$\begin{array}{cccc}\mathrm{x} \ge \mathrm{y}& <=>& q_i\left(\mathrm{x}\right) \ge { q}_i\left(\mathrm{y}\right) \forall & q_i\in \mathrm{IB}\\ \mathrm{x} \le \mathrm{y}& <=>& q_i\left(\mathrm{x}\right) \le { q}_i\left(\mathrm{y}\right) \forall & q_i\in \mathrm{IB}\\ & & \mathrm{x}|\mathrm{y} \mathrm{else}& \end{array}$$

(1)

Objects can be compared when x ≥ y or x ≤ y; when x|y, the objects cannot be compared.

For ease of comparison, all input data for the Hasse diagram need to be normalized so that the data input range is [0, 1]. The normalization formula is as follows [50]:

$$\mathrm{q}{\mathrm{n}}_{\mathrm{i}}\left(\mathrm{x}\right)={\displaystyle \frac{\left({\mathrm{q}}_{\mathrm{i}}\left(\mathrm{x}\right)-{\mathrm{q}}_{\mathrm{i}}\min \right)}{\left({\mathrm{q}}_{\mathrm{i}}\max -{ \mathrm{q}}_{\mathrm{i}}\min \right)}}$$

(2)

where qn_i is the value of the indicator, and q_imax and q_imin are the maximum and minimum values, respectively.

2.4.4. Application of Partial Order Theory

The Hasse diagram encompasses several conceptual elements, as detailed below [48,50].

Level: in a Hasse diagram, elements that share the same vertical position are grouped into a level, reflecting their equivalent rank within the partial order structure (e.g., three levels are shown in Figure 3).

Chain: a chain refers to a sequence of elements that are entirely comparable, where each pair has a defined order. Chains are often used to trace the primary progression of influencing factors in the input dataset (e.g., WC → QSH → HH in Figure 3).

Maximal element: this is an element that has no other element ranked above it. It is displayed at the top of the diagram, representing a dominant position that is not exceeded by any other (e.g., HH, TKT, and TMZ in Figure 3).

Minimal element: an element with no other element ranked below it, shown at the bottom of the diagram, indicating a foundational or initiating position in the hierarchy (e.g., WC in Figure 3).

3. Results

3.1. Spatial and Temporal Analysis of Grassland Conversion Between 2000 and 2020

In the period from 2000 to 2010, the GCO area accounted for approximately 2.1% of the total area. At the same time, approximately 3.5% of the study area experienced GCI. GCO mainly occurred in Hohhot, as well as the surrounding BT, DS, ETK, and WS of Ordos (Figure 4a). GCI mainly occurred in the western counties of the study area (Figure 4a). There were significant conversion processes between grassland and cropland, as well as between grassland and bare land (Figure S2, Table S1, highlighted in yellow and green). During this period, the proportion of GEI is larger than GCO, referring, respectively, to the grassland that is restored and degraded (Figure 4a and Figure S3).

During the period 2010–2020, the trends in grassland-use change intensified, with the proportion of GCO rising to 4.8% and GCI increasing to 4.7%. Compared with the previous period, the spatial distribution of GCO expanded significantly, particularly in the resource development zones of Ordos, where transitions to cropland, bare land, and impervious surfaces became more pronounced (see Figure 4b and Figure S2, Table S2). Simultaneously, the spatial distribution of grassland conversion further expanded, concentrating in the northern and central parts of the study area, with cropland and bare land being the primary sources of this change (Figure 4b and Figure S3).

3.2. Partial Order Ranking of Grassland Conversion out Drivers Between 2000 and 2020

3.2.1. Partial Order Ranking of Grassland Conversion out Drivers in the Period from 2000 to 2010

In an attempt to identify major drivers for GCO, we systematically analyzed the layers, chains, and structures to evaluate and rank the impact of the four driving factors on GCO.

Regarding the effect of the weather variable-related driver group on GCO in the period from 2000 to 2010, a total of five levels were identified (Figure 5a). Weather variables have the strongest effect on GCO in DMJ, HJ, DLT, WS, EKTO, and EKT; Figure 5a indicates that all six of these counties are located in the upper level of the Hasse diagram. Meanwhile, weather variables have the smallest effect in WC, QSH, and HL, which are located in the lower level (the lowest level in Figure 5a). Overall, weather variable drivers have the largest effect on GCO in Ordos and the smallest effect in Hohhot.

Figure 5b and Table S3 show that GCO was least strongly affected by human drivers in the western region of Ordos, including EKT, ETKQ, and HJ. Human activities also had the smallest effect on DMJ, which is part of Baotou. The major land-use type in these four counties is grassland. The effect of population is larger than that of sheep density in HH, BT, and DS, and the population effect is decreased as follows: {HH > BT > DS}. Meanwhile, the effect of sheep density is larger than the population in the following counties: WC, TKT, TMZ, QSH, HL, TMY, GY, HJ, DLT, YJHL, WS, and ETKQ. A total of 74 related chains were identified, and the above counties are a “sheep-density-dominant chain” (Table S4).

For the economic driver group, six levels were identified in this period (Figure 5c). From a spatial perspective, the economic effect was larger in four of the six counties in the Hohhot region (HH, TKT, TMZ, and HL), the eastern four counties of Ordos (DS, DLT, ZGE, and YJHL), and two counties in the southern part of Baotou (BT and TMY) (the upper level of the Hasse diagram in Figure 5c). Meanwhile, economic factors have a smaller effect in the western region of Ordos (ETKQ, ETK, and HJ).

Regarding the urbanization effect during this period, a total of three levels were identified; this means that the partial order relationship for each county is simple. Figure 5d shows that urbanization has the strongest effect in four counties in Hohhot (HH, WC, TKT, and HL), three counties in Baotou (BT, TMY, and GY), and only one county in Ordos (DS). This result indicates that urbanization has the largest effect in Hohhot, and then in Baotou, while it has the smallest effect in Ordos. Specifically, urbanization has the smallest effect in the eastern region of Ordos: HJ, WS, ETKQ, and ETK. Overall, the urbanization effect was larger around the three major cities of HBO (HH, BT, and DS).

3.2.2. Partial Order Ranking of Grassland Conversion out Drivers in the Period from 2010 to 2020

Regarding the effect of weather variables in the period of 2010–2020, a total of four levels were identified (Figure 6a). For the spatial characteristics, the effect of weather variables seems the same as in the previous period (2000–2010). HJ, ETKQ, and ETK, which are part of Ordos, continued to suffer the largest effect from weather variables. Moreover, for WC, HH, and HL, which are part of Hohhot, the impact of weather variables on GCO was minimal. Overall, the effect of weather variables in Ordos decreased in this period (Figure 6a).

Regarding the effect of human activities in the period of 2010–2020, 10 levels were identified; compared with the previous period, the structure of the Hasse diagram is more complex, and the effect of humans on GCO in each county is more intricate (Figure 6b). Overall, BT, HH, DS, TMY, TMZ, and TKT are still located on the upper level, and DMJ, ETK, ETKQ, WC, and HJ are still located on the lower level of the Hasse diagram (Figure 6b). In this period, HH and DS represent a unique chain: the “population-density-dominant” chain. Meanwhile, the effect of sheep density is larger than that of the population in the following counties: WC, TKT, TMZ, QSH, HL, TMY, GY, HJ, DLT, WS, ETKQ, and ETK. A total of 15 chains were identified, which we called “livestock-dominant” chains (Tables S5 and S6).

Compared with the previous period of 2000–2010, there was no obvious spatial characteristic for the economic effects. There were still six levels for economic effects, and HH, BT, and TMY suffered the largest effect in these three counties. Meanwhile, economic factors had the smallest effect in WC, DMJ, ETKQ, ETK, and HJ (Figure 6c).

As for the effect of urbanization in this period, the eastern parts of Ordos (HJ, WS, ETKQ, and ETK) and DMJ still suffered the least in these five counties. However, in the regions previously most affected by urbanization, this phase exhibited significant changes. Compared with the previous period, the impact of urbanization in TKT, HL, DS, and GY decreased to a notable degree (Figure 6d).

3.3. Ranking of All Driver Groups for the Two Periods (2000–2010 and 2010–2020)

In an effort to identify the dominant driver for each county in the two periods, we transformed all partial order ranking results for all driver groups into two maps (Figure 7 and Figure 8). The identification of the levels helps us to compare the major drivers for each county.

4. Discussion

In the period of 2000–2010, the influence of the four driver groups on GCO is ranked as follows: urbanization > weather variables > economic factors > human activities. Urbanization was the dominant driver group in 11 counties (HH, WC, TKT, QSH, HL, BT, TMY, GY, DS, ZGE, and YJHL; see Figure 7). Urbanization is the unique driver group in five counties (WC, QSH, HL, GY, and DS; see Figure 7). Weather variables also had a stronger effect in eight counties (DMJ, HJ, DLT, ZGE, YGHL, WS, ETKQ, and EKT), seven of them being part of Ordos. Among these eight counties, weather variables were the unique drivers in six: DMJ, HJ, DLT, WS, ETKQ, and ETK. Economic factors constituted the dominant driver group in seven counties: HH, TKT, TMZ, BT, TMY, ZGE, and YJHL. Human activities constituted the major driver group in four counties: HH, TKT, TMZ, and TMY. Ultimately, the combination of human activity, economic factors, and urbanization represented the dominant driver groups in HH, TKT, and TMY. Both human activities and economic factors were the major driver groups in TMZ. The dominant driver group in BT is made up of economic factors and urbanization. Weather variables, economic factors, and urbanization were the major driver groups in ZGE and YJHL.

After 2010, the overall ranking of the impact of each driver group on grassland conversion remained unchanged compared with the period from 2000 to 2010: urbanization > weather variables > economic > human activities. Compared with the previous period, the effect of urbanization was decreased; it was the dominant effect in ten counties (HH, WC, TKT, TMZ, QSH, HL, BT, TMY, DS, and DLT; see Figure 8). The urbanization effect increased in Hohhot and decreased in both Baotou and Ordos. The effect of weather variables was still the dominant driver group in seven counties, and it had the largest effect in Ordos. The economic effect was still the major driver group in seven counties, and the economic effect decreased in Hohhot and increased in Ordos (see Figure 8). The effect of human activities increased during this period, especially in BT and DS. Overall, the partial order ranking for all drivers seems unchanged compared with the previous period. However, for the urbanization-related drivers, the effect decreased, and the effect of human activities increased.

4.1. The Intensity of Land-Use and Land-Cover Change and Grassland Conversion in Hohhot–Baotou–Ordos Has Increased Since 2000

The research results indicate that during both study periods, the intensity of land-use change and the proportion of GCO and GCI gradually increased, showing significant spatial characteristics. During the period from 2000 to 2010, 6.9% of the study area experienced changes; these changes were mainly concentrated in the three major cities in the western part of the study area and their surrounding areas. This value increased to 11% in the 2010–2020 period, with significant LUCC changes occurring across the entire study area but still primarily around the three major cities and their surroundings. GCO accounted for 2.1% and 4.8%, while GCI accounted for 3.5% and 4.7% during the two study periods (Figure 4 and Figure S4). The increase in the intensity of land-use change suggests more frequent and drastic transformations in land-use patterns within the region. Previous studies have shown that human activities such as urban expansion, industrial development, and resource exploitation are key drivers of increased intensity in land-use change in this region [51,52], as is consistent with the findings of this study. This change has exacerbated the occupation and disturbance of grassland resources, putting greater pressure on the protection of grassland ecosystems [53,54]. On the one hand, the risk of grassland degradation has increased, and the ecosystem service capacity is threatened [55]; on the other hand, intensified land-use competition has highlighted the conflict between grassland protection, accelerated urbanization, and economic development [56]. On this basis, the region should continue to strengthen land-use planning and zoning management, optimize land-use structures, and strictly control grassland conversion [57]. At the same time, ecological compensation and incentive mechanisms should be promoted, supporting the development of green industries and ecological economic models to ensure the balance between grassland protection and local economic development [58]. The intensive use of construction land should be enhanced to reduce grassland occupation [59], and grassland ecological restoration projects should be implemented along with dynamic monitoring to track real-time changes in grasslands [60].

4.2. The Effect of Urbanization on Grassland Conversion out from 2000 to 2020

This study shows that urbanization-related driving factors were the primary drivers of GCO in the HBO region during both research periods (2000–2020). Notably, in Hohhot, the key driving factors include HH, WC, TKT, QSH, and HL; in Baotou, BT and TMY; in Ordos, DS. Overall, the impact of urbanization in the western part of the region is significantly greater than in the eastern part, reflecting spatial imbalances (see Figure 7 and Figure 8). This difference is mainly attributed to the rapid urbanization in the eastern region and the marked acceleration of population concentration (Figure S5). Population migration is one of the core drivers of urbanization, and as an important economic center of Inner Mongolia, the HBO region’s rapid urbanization is closely linked to the migration of populations from rural areas to cities. Notably, during the period from 2000 to 2020, a large number of rural residents migrated to cities such as Hohhot, Baotou, and Ordos, driving the expansion of urban areas and the rapid growth of urban land. Figure S5a indicates that between 2000 and 2020, the urban population in the study area consistently exceeded the rural population, and the gap between them gradually widened. The urban populations in Hohhot and Baotou were greater than those in Ordos. During this period, urban construction land and rural settlements in the study area increased by 332 km² and 102 km², respectively, significantly changing the use of grassland resources. These changes were particularly evident in the development of wide roads, large squares, new urban areas, and the encroachment of expanding industrial and mining land. This unplanned expansion directly occupied grasslands, destroyed surface vegetation and the ecological environment and led to a significant increase in the intensity of GCO. Previous studies have shown that the construction land area in the study area exhibits the characteristics of disorderly expansion, especially with the excessive pursuit of wide roads, large squares, new urban areas, and industrial and mining land, resulting in changes to the surface landscape pattern and the ecological environment, destroying natural vegetation, and weakening the stability of the land structure [61].

HBO is a resource-based region dominated by industry and mining, and the expansion of industrial and mining land has had a significant negative impact on the grassland ecosystem. Since 2000, the study area has gained 1270 km² of industrial and mining land, and mining activities have combined the direct occupation and indirect destruction of grasslands, further exacerbating the loss of grassland resources [62]. Industrial and mining development not only leads to a reduction in grasslands [63] but also triggers habitat fragmentation [64], soil erosion, and pollution [65]. These destructive activities have weakened the ecological functions of grasslands, such as carbon sequestration, water conservation, and the maintenance of biodiversity [66]. Furthermore, the expansion of mining sites is accompanied by the construction of roads, with a total increase of 6606 km in road length. These roads directly occupy grasslands and, by facilitating human activities in the surrounding areas, such as agricultural expansion [67], industrial development, and tourism [68], further exacerbate grassland degradation [69].

4.3. The Effect of Weather Variables on Grassland Conversion out from 2000 to 2020

This study found that changes related to weather variables were the second-largest driver of grassland conversion during both study periods, surpassing the influence of human activities and economic development (see Figure 7 and Figure 8). The area of grassland converted to bare land was particularly significant in the six counties in the western part of Ordos (HJ, DLT, YJHL, WS, ETKQ, and ETK) and the DMJ district in Baotou (Figure S2). These areas have been severely affected by frequent droughts, abnormal precipitation patterns, and rising temperatures because of weather variable-related change [70], exacerbating ecosystem vulnerability, reducing grassland cover, and significantly expanding bare land [71]. This indicates that, in most counties in Ordos and the DMJ district in Baotou, the primary challenge for grassland conversion to bare land is the persistent influence of weather variable-related factors. Weather variables have undermined the ecological functions of grasslands, such as water conservation and soil retention, exacerbating ecological degradation [72]. Therefore, in these areas, it is crucial to strengthen further the implementation of ecological projects, such as restoring degraded grasslands, controlling desertification, and promoting vegetation recovery, to enhance the ecosystem’s resilience to weather variables.

4.4. Regional Variability in Grassland Degradation Drivers

To verify the reliability of our findings, we compared the results of this study with those from other regions in China, as well as from Europe and Africa. Our study shows that urbanization is the primary driver of grassland conversion in the study area, which aligns with the findings of Batunacun et al., who identified urban expansion as a key cause of grassland degradation in Xilingol since 2000 [10]. However, the dominant drivers of grassland degradation vary significantly across regions. For instance, approximately 90% of the UK’s semi-natural species-rich grasslands have been degraded because of intensive agriculture and land conversion since the 1940s [60]. In Inner Mongolia, overgrazing, reclamation, and rodent activity led to the degradation of about 3.9 million hectares of typical steppe, accounting for 62% of the total grassland area by 2010 [73]. Likewise, nearly 65% of savannas in arid and semi-arid regions of South Africa are threatened by extreme droughts. Although the drivers of grassland degradation in Inner Mongolia differ from those in Europe and the United States, the key driving factors in the Inner Mongolian region remain consistent overall. For instance, before 2000, human activities were the primary drivers of grassland degradation, whereas, after 2000, the key driver of grassland degradation in the study area shifted from human activities to urbanization and other factors.

4.5. Methodological Strengths and Limitations of Partial Order Theory

In the present study, we analyzed the dominant drivers of GCO from 2000 to 2010 and 2020–2020, based on the hypothesis that the GCO was driven by multiple interacting socioeconomic and environmental factors. POT and the HDT were integrated to conduct this analysis, given that only limited data pertaining to the drivers were available. Compared with traditional statistical methods, such as principle component analysis and regression analysis, POT does not ignore the complex relationship between GCO and its drivers, and it was used to identify the major drivers of GCO. The framework effectively overcomes the limitations of traditional methods in handling complex interrelationships among drivers. Through the analysis of grassland conversion out drivers in the HBO region from 2000 to 2020, POT was successfully used to identify the dominant influences of weather variables, human activities, economic factors, and urbanization across different county-level units.

Overall, this study establishes an innovative framework for identifying the driving factors of grassland conversion based on POT. By integrating POT with the HDT, the framework is well-suited for analyzing multi-indicator and nonlinear systems, effectively overcoming the limitations of traditional methods in handling complex interrelationships among variables. Additionally, the use of HDT clearly visualizes the hierarchical structure of the drivers, providing a solid foundation for future policy adjustments. The advantage of POT lies in its ability to effectively integrate and rank multiple drivers, even under the conditions of limited or missing data [74].

However, POT also has certain limitations. The structure of the partial order is highly dependent on the selection and number of driving factors, as these determine the multidimensional representation of each element and subsequently affect the comparability among elements and the complexity of the partial order structure [51]. Therefore, in practical applications, it is essential to carefully select and combine driving factors based on regional characteristics to ensure the scientific validity of the partial order structure.

5. Conclusions

This study provides an example of how to use POT to rank the potential drivers of GCO and visualizes the results using HDT. The combination of POT and HDT can address the limitations encountered in previous studies, which often overlook the dynamic, integrative, synergistic, hierarchical, and spatially heterogeneous characteristics of driving factors. POT effectively handles complex interactions and nonlinear relationships among the multiple driving factors of grassland conversion, providing a comprehensive and objective analysis. By avoiding biases from linear assumptions and subjective weighting, POT ensures more accurate results and captures the hierarchical and dynamic structure of these factors through Hasse diagrams. Additionally, POT accounts for spatial heterogeneity, making it ideal for developing region-specific land-use policies tailored to diverse ecological and socioeconomic contexts. This approach not only provides a novel perspective for identifying the drivers of grassland change but also offers region-specific recommendations for grassland management. The main conclusions are as follows:

(1)

Between 2000 and 2020, land use in the HBO region underwent significant changes. From 2000 to 2010, 6.9% of the study area experienced land-use changes, which were primarily concentrated around Hohhot and Baotou. In the period from 2010 to 2020, the proportion of land-use change increased to 11%, with a notable rise in the intensity of land-use change in the region.

(2)

The spatiotemporal dynamics of GCO showed significant regional differences. From 2000 to 2010, approximately 2.1% of the study area underwent GCO, mainly from grassland to cropland, bare areas, or impervious surfaces. During this period, 3.5% of the land experienced GCI, primarily from bare areas, cropland, or forest. From 2010 to 2020, the proportion of GCO increased, reflecting the intensifying impact of land-use change on the grassland ecosystem.

(3)

From 2000 to 2020, the primary driving factors for GCO in the HBO region were urbanization, climate change, economic development, and human activities. Urbanization was the most significant driver of large-scale GCO. Climate change, particularly drought, also played a significant role, along with economic development and human activities.

This research suggests that in the future, the eastern region of the study area should prioritize addressing urban development impacts, while the western region should focus more on implementing ecological construction projects. Adopting locally tailored ecological protection and economic development strategies within the study area is crucial for achieving a win–win outcome in regional ecological conservation and economic development. Future research on POT should focus on further refining its application to complex, nonlinear systems, particularly in dynamic environmental and socioeconomic contexts. There is a need to enhance the integration of POT with other advanced analytical methods, such as AI and machine learning, to improve the prediction and visualization of land-use change processes. Additionally, expanding its application to a broader range of ecological and socioeconomic drivers across other global regions will allow for a more comprehensive understanding of regional variations and interactions.