Next Article in Journal
Impacts of Vegetation Ratio, Street Orientation, and Aspect Ratio on Thermal Comfort and Building Carbon Emissions in Cold Zones: A Case Study of Tianjin
Previous Article in Journal
Harnessing Machine Learning Algorithms to Model the Association between Land Use/Land Cover Change and Heatwave Dynamics for Enhanced Environmental Management
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Land
Volume 13
Issue 8
10.3390/land13081274
land-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Driving Forces behind the Reduction in Cropland Area on Hainan Island, China: Implications for Sustainable Agricultural Development

by
Jianchao Guo
1,
Shi Qi
2,*,
Jiadong Chen
1,* and
Jinlin Lai
2
1
Hainan Province Water Conservancy & Hydropower Survey, Design & Research Institute Co., Ltd., Haikou 571100, China
2
School of Soil and Water Conservation, Beijing Forestry University, Beijing 100083, China
*
Authors to whom correspondence should be addressed.
Land 2024, 13(8), 1274; https://doi.org/10.3390/land13081274
Submission received: 1 August 2024 / Revised: 7 August 2024 / Accepted: 9 August 2024 / Published: 13 August 2024
(This article belongs to the Section Land – Observation and Monitoring)
Browse Figures
Versions Notes

Abstract

:
Food security is a major challenge for China at present and will be in the future. Revealing the spatiotemporal changes in cropland and identifying their driving forces would be helpful for decision-making to maintain grain supply and sustainable development. Hainan Island is endowed with rich agricultural resources due to its unique climatic conditions and is facing tremendous pressure in cropland protection due to the huge variation in natural conditions and human activities over the past few decades. The purpose of this study is to assess the spatiotemporal changes in and driving forces of cropland on Hainan Island in the past and predict future cropland changes under different scenarios. Key findings are as follows: (1) From 2000 to 2020, the cropland area on Hainan Island decreased by 956.22 km2, causing the center of cropland to shift southwestward by 8.20 km. This reduction mainly transformed into construction land and woodland, particularly evident in coastal areas. (2) Among anthropogenic factors, the increase in the human footprint is the primary reason for the decrease in cropland. Land use changes driven by population growth, especially in economically active and densely populated coastal areas, are key factors in this decrease. Natural factors such as topography and climate change also significantly impact cropland changes. (3) Future scenarios show significant differences in cropland area changes. In the natural development scenario, the cropland area is expected to continue decreasing to 597 km2, while in the ecological protection scenario, cropland conversion is restricted to 269.11 km2; however, in the cropland protection scenario, the trend of cropland reduction is reversed, increasing by 448.75 km2. Our findings provide a deep understanding of the driving forces behind cropland changes and, through future scenario analysis, demonstrate the potential changes in cropland area under different policy choices. These insights are crucial for formulating sound land management and agricultural policies to protect cropland resources, maintain food security, and promote ecological balance.

1. Introduction

Cropland, as a coupled natural and artificial composite ecosystem, has been transformed by human activities, while retaining the characteristics of the original natural ecosystem [1]. China’s cropland area totals 1.43 billion km2, playing a crucial role in global food security and the sustainable development of agriculture [2]. Hainan Island, due to its unique geographical location and superior natural environment, possesses abundant agricultural resources and is one of China’s important agricultural production bases [3]. However, cropland resources are facing pressures from rapid socio-economic development and population growth. The continuously diminishing cropland area poses significant threats to agricultural supply security, ecological maintenance, and economic growth. This concerns not only food security on Hainan Island but also directly affects the stability of the entire ecosystem and regional sustainability [4,5].
Natural conditions, including topographic factors, soil texture, and climate, significantly impact the suitability, availability, and stability of agricultural production [6,7,8]. For example, elevation and slope affect soil erosion and the feasibility of cultivating land [9,10]. Soil texture determines the land’s capacity to retain moisture and nutrients, thereby influencing crop growth [11]. Climatic factors, particularly temperature and precipitation, directly impact crop growth cycles and yields [12]. In addition, anthropogenic activities are key factors influencing cropland changes. With population growth and economic development, the demand for cropland increases, leading to changes in cropland [13,14,15]. During urbanization, croplands are converted to residential and industrial lands, directly reducing the lands available for agricultural production [16]. At the same time, the application of modern agricultural technologies, such as irrigation systems, fertilizers, and pesticides, has transformed traditional cropland management practices. These changes have implications for the long-term productivity of the land and its ecological environment [17]. Numerous studies indicate that the impacts of natural conditions and human activities are not isolated; rather, they interact under specific circumstances, collectively determining the agricultural production potential of a region [18,19,20].
The spatiotemporal evolution of cropland resources has attracted extensive attention from scholars both domestically and internationally. Existing studies have focused on the spatiotemporal patterns of cropland use, ranging from macro scales such as national and river basin levels to micro scales including provinces, cities, and counties, to explore the distribution characteristics and trends of cropland resources across different regions and scales [21,22]. On the other hand, research has primarily concentrated on the driving mechanisms of changes in cropland resources, including natural conditions, socio-economic conditions, and policies [23]. However, past research has often focused on cropland issues in China’s northern arid and semi-arid areas and economically developed eastern coastal regions, examining the impacts of natural and anthropogenic factors on cropland changes. For example, Fu et al. [24] used correlation analysis to discover significant synergy between climate change and cropland area changes in northern arid regions. Wang et al. [25] used linear least squares regression to assess the impacts of socio-economic factors on the annual changes in cropland in developed coastal areas, highlighting the negative effects of human activities on the protection of cropland resources. Nouri et al. [26] used spatial statistics to find that slope and industrial output have significant negative impacts on the amount of cropland. Previous studies have often overlooked the combined effects of natural and anthropogenic factors on cropland changes and have not fully explored the driving forces. In particular, research on tropical island regions like Hainan has been scarce and lacks in-depth systematic analysis. Furthermore, research on cropland changes is not limited to the past; predicting future changes can help to provide a scientific basis for regional policy-making. Building on this clarification of influencing factors, some scholars have conducted land use simulation studies under various scenarios. The GeoSOS-FLUS model, based on the Cellular Automata (CA) method, is used to simulate different land use scenarios. It can predict urban growth, agricultural changes, and ecological impacts. This helps in planning and managing land use under various conditions [27].
This study focuses on a tropical island in China as its research subject, specifically addressing the following key issues: (1) the spatiotemporal characteristics of cropland changes on Hainan Island from 2000 to 2020; (2) the impacts of natural and anthropogenic factors on the spatiotemporal changes in cropland on Hainan Island, analyzing its driving forces; (3) and the future directions for land use transitions on Hainan Island under different policy contexts, discussing the spatial optimization of cropland to improve the structure and allocation of cropland resources. This study of spatiotemporal changes in cropland on Hainan Island provides insights and references for understanding land use changes in tropical island regions under global changes and the rational utilization and management of cropland resources in the future. Moreover, the research results offer an important perspective on understanding cropland changes in tropical island regions.

2. Study Area

Hainan Island is located in the northwestern part of the South China Sea and has a coastline approximately of 1855 km, predominantly consisting of natural shorelines. The island features low-lying terrain around its perimeter, with towering mountains and hills in the central area, centered around Wuzhi Mountain, descending in terraces in a circular pattern. The topography comprises mountains, hills, plateaus, and plains, displaying a distinct stepped structure (Figure 1). Hainan Island experiences a tropical monsoon climate with a clear distinction between dry and rainy seasons, high humidity, and rich vegetation and crop resources. However, in recent years, economic development and other human activities have significantly damaged the cropland ecosystems.

3. Date Sources and Methods

3.1. Data Sources

Land use data for Hainan Island were sourced from the Resource and Environmental Sciences and Data Center, with remote sensing monitoring data obtained for the years 2000, 2010, and 2020, with a spatial resolution of 30 m. Driving factors include natural factors (elevation, slope, temperature, precipitation, soil sand content, soil clay content, and water distance) and anthropogenic factors (population density, road distance, and human footprint). Specific data sources are detailed in Table 1. This study utilized ArcGIS 10.2 to resample data to unify the spatial resolution (30 m).

3.2. Methods

3.2.1. Center-of-Gravity and Standard-Deviation Ellipse Models

The change in the center of gravity reflected the spatiotemporal migration of cropland. During calculation, the area of the cropland patches was used as a weight, and the geometric center of each patch was considered its center of gravity. The center of gravity of the patches, combined with the weight factors, represented the shift in cropland’s center of gravity. The calculation formulas are as follows [28]:
X t = ∑ i n w i x i ∑ i n w i
Y t = ∑ i n w i y i ∑ i n w i
D = ( X t 1 − X t 2 ) 2 + ( Y t 1 − Y t 2 ) 2
where Xt and Yt are the latitude and longitude coordinates of the center of gravity of the cropland in year t, respectively; i is the grid serial number; n is the total number of grids; x and y are the latitude and longitude coordinates of the center of gravity of the ith grid; and wi is the weight of the ith patch. Assuming that the coordinates of the center of gravity of the cropland are (Xt1,Yt1) and (Xt2,Yt2) in the period of t1 and t2, respectively, then the migration distance of the cropland for the center of gravity in the period of t1–t2 is D.
The standard-deviation ellipse is a spatial statistical method for the multidimensional analysis of center of gravity, direction, distance, and shape, which can explore the global characteristics of cropland’s geographical distribution.

3.2.2. Kernel Density Analysis

Kernel density analysis effectively examines the regional differences in the dynamics of cropland changes. The calculation formula is as follows [29]:
f n ( x ) = 1 n h p ∑ i = 1 n K ( x − x i h )
where f(x) represents the kernel function, and h represents the bandwidth. In practical calculations, the basic principle for selecting the optimal bandwidth is to minimize the mean squared error. The specific expression is as follows:
K ( u ) = p ( p + 2 ) 2 S P ( 1 − u 1 2 − u 2 2 − ⋯ − u p 2 )
where Sp = 2πp/2/⊺(p/2), and the corresponding formula is
K h ( x − X i ) = p ( p + 2 ) 2 h S P 〔 1 − ( x 1 − x i 1 h ) 2 − ⋯ − ( x p − x i p h ) 2 〕

3.2.3. XGBoost

XGBoost (eXtreme Gradient Boosting) is an efficient gradient boosting tree model widely used across various machine learning tasks. It incrementally builds multiple decision trees and combines them with weights to enhance prediction accuracy, utilizes multi-threaded parallel computing to significantly accelerate training speed, and supports multiple objective functions and custom loss functions, catering to different types of prediction tasks. Feature selection is a crucial step in building high-performance models [30,31]. In this study, natural and anthropogenic factors from Table 1 were selected for model training, as they significantly impact cropland kernel density changes. Before training the model, we processed dynamic feature variables such as temperature, precipitation, population density, and human footprint. Temperature and precipitation changes from 2000 to 2020 were calculated using the Theil trend analysis method, while the differences in population density and human footprint between 2000 and 2020 were calculated. The formula for the Theil–Sen trend analysis method is as follows [32]:
β = M e d i a n x j − x i j − i ( ∀ j > i )
In the formula, β represents the trend in the time series of temperature or precipitation, with xi and xj being the values of temperature or precipitation at the i-th and j-th points in the time series, respectively.
The model training process included the following steps: (1) Data splitting: The dataset was divided into a training set and a test set, with an 80% training and 20% testing distribution to ensure model generalization capability. (2) Parameter tuning: Hyperparameters of the XGBoost model were optimized using grid search and cross-validation methods. Key parameters adjusted included the learning rate, the maximum depth, the number of estimators, the subsample ratio, and feature sampling by tree. (3) Model training: The XGBoost model was trained on the training set using the best hyperparameters. The process involved building multiple decision trees and weighting them collectively. (4) Model evaluation: The model’s performance was assessed on the test set by calculating the coefficient of determination (R2), root mean squared error (RMSE), and mean absolute error (MAE) [33,34].

3.2.4. SHapley Additive exPlanations

SHAP (SHapley Additive exPlanations) is a model interpretation method based on game theory used to explain the outputs of machine learning models. It assigns contribution values to feature variables, providing transparency about each feature’s impact on model predictions. Key features and benefits of SHAP include the consistency and additivity of feature importance scores, the provision of local explanations for each sample, and its applicability to various types of machine learning models, including tree-based models, linear models, and neural networks.
The calculation of SHAP values was derived from a game-theory-based distribution method. For a given predictive model and sample, SHAP values represented the contributions of feature variables to the predicted output. Initially, the marginal contributions of each feature were calculated using a trained XGBoost model across various combinations. By averaging over all possible combinations of feature variables, the SHAP values for each feature were obtained. Based on SHAP values, feature variables could be ranked by importance to determine which features most influenced the model’s predictions. The larger the SHAP value, the greater the impact of the feature on the model’s predictions. By averaging SHAP values across all samples, a global ranking of feature importance could be established. In machine learning models, there may be complex interactions among features. The SHAP method not only provides contribution values for individual features but also captures and explains the interactions between features [35].

3.2.5. GeoSOS-FLUS Model

This study applied the GeoSOS-FLUS model to simulate and predict the spatial distribution pattern of land use on Hainan Island by 2040. GeoSOS-FLUS is capable of geospatial simulation and supports decision-making, making it suitable for forecasting and analyzing land use scenarios [36,37]. This study identified driving factors for land use change primarily divided into natural factors (elevation, slope, temperature, precipitation, soil sand content, soil clay content, and water distance) and anthropogenic factors (population density, road distance, and human footprint).
Considering national policy responses, this study set scenarios of natural development, ecological protection, and cropland protection. In the natural development scenario, land use evolved according to historical patterns without policy or restrictive interventions. The ecological protection scenario followed Hainan’s development plans to maintain ecological zones, optimize economic layout, and combine the island’s land use structure, ecological carrying capacity, and development status to prioritize the protection of woodlands, grasslands, and water bodies, limiting their conversion and transitioning from high to lower levels of land use. Under the cropland protection scenario, the conversion of cropland to other land types was strictly restricted. The land use transition matrixes under different scenarios are presented in Table 2 [38,39].
To ensure that the model could be applied to simulate land use changes in the study area, the Kappa coefficient and overall accuracy were used to evaluate the simulation results. The larger the Kappa coefficient and overall accuracy, the higher the simulation accuracy of the model [37].

4. Results and Analysis

4.1. Land Use Change on Hainan Island

4.1.1. Characteristics of Land Use Change

Land use types on Hainan Island changed significantly from 2000 to 2020 (Table 3 and Figure 2). The land use structure of Hainan Island primarily consisted of woodland and cropland, accounting for 89.98%, 89.10%, and 87.36% of the total area in 2000, 2010, and 2020, respectively. Woodland, which occupied a dominant position, accounted for 63.70%, 63.83%, and 63.87%, respectively, showing a slight but consistent upward trend. In contrast, cropland areas, representing 26.28%, 25.26%, and 23.49%, respectively, continuously declined, with rate changes of −3.86% from 2000 to 2010 and −7.03% from 2010 to 2020. Grassland areas initially decreased and then increased, with rate changes of −6.15% and 3.66%, respectively. Water bodies first increased and then decreased in area, with changes of 17.30% and −2.09%, respectively. The area of construction land continued to grow, with rate changes of 26.10% and 59.99%, respectively. Unused land, which had the smallest proportion, continually decreased, with rate changes of −32.38% and −6.62%, respectively.
Overall, over the past 20 years, the areas of cropland, grassland, and unused land have shown a decreasing trend, with cropland experiencing the largest decrease, amounting to 956.22 km2. Meanwhile, woodland, water, and construction land areas have increased, with construction land showing the most significant increase, amounting to 793.27 km2.

4.1.2. Trajectory of Land Use Changes

This study analyzed the transformation of cropland from 2000 to 2020 through land use transition maps (Figure 3 and Figure 4). From 2000 to 2010, a total of 619.57 km2 of cropland was converted, primarily into woodland and construction land. Specifically, 262.61 km2 was converted to woodland, mainly in Danzhou and Lingao, while 227.92 km2 was transformed into construction land, concentrated in Haikou, Danzhou, and along the coasts of Sanya. A total of 241.58 km2 of cropland was gained, mostly from woodland, which contributed 128.53 km2, primarily in Wanning and Qiongzhong. From 2010 to 2020, 845.37 km2 of cropland was converted, with 441.47 km2 turned into woodland, predominantly in the northern part of the study area, and 304.78 km2 converted into construction land, notably in Haikou and Sanya. Cropland gains totaled 237.15 km2, primarily from woodland conversions, accounting for 155.15 km2. Overall, from 2000 to 2020, the cropland area decreased by 956.22 km2, mainly being transformed into woodland and construction land, with significant conversions of woodland into cropland as well. The extensive mutual conversion between woodland and cropland may be attributed to the tropical climate and demand for crop and woodland production. Notably, due to economic development and urban expansion, a significant amount of cropland was converted by construction land, especially in the coastal areas of Haikou and Sanya.

4.1.3. Center-of-Gravity and Standard-Deviation Ellipse Change in Cropland

Figure 5 reflects the changes in the center of gravity of cropland on Hainan Island from 2000 to 2020. The center of gravity of cropland was consistently located in Tunchang (2000: 109°55′29.02″ E, 19°21′58.87″ N; 2010: 109°54′45.79″ E, 19°20′42.25″ N; 2020: 109°51′57.23″ E, 19°19′5.16″ N). From 2000 to 2010, the center moved southwest by 2.64 km, and from 2010 to 2020, it moved southwest by 5.81 km, indicating a greater shift in cropland during the latter decade, with the distribution pattern moving southwestward. Overall, the center shifted southwest by a total of 8.20 km. Regarding the standard-deviation ellipse, from 2000 to 2010, its area decreased from 17,369.03 km2 to 16,660.82 km2, representing a reduction of 708.21 km2. From 2010 to 2020, the ellipse significantly contracted, decreasing by 2043.05 km2. The ellipse contracted towards the central part of the area, with its short axis remaining essentially unchanged, but its long axis became shorter, indicating a trend of cropland aggregating towards the central part.

4.2. Kernel Density of Cropland

Figure 6 illustrates the spatiotemporal changes in cropland kernel density (KD) on Hainan Island from 2000 to 2020. High-KD areas of cropland are mainly concentrated in coastal regions, indicating that these areas are the primary agricultural production zones on the island. The mountainous central regions exhibit lower cropland KD, reflecting less agricultural activity. From 2000 to 2020, the overall cropland KD on Hainan Island showed a declining trend, suggesting a reduction in cropland area, particularly in coastal regions, which is associated with urbanization and industrial expansion. Although the overall cropland area decreased, there was a slight increase in cropland KD in localized areas (such as Wanning), possibly reflecting agricultural development in certain regions, albeit on a modest scale.

4.3. Driving Forces of Changes in Cropland KD

4.3.1. SHAP Interpretability Analysis

The dataset of KD changes (KDC) from 2000 to 2020 and feature variables was divided into training and testing sets in an 80% to 20% ratio. The XGBoost model was used for training, with fit results showing an R2 of 0.79, RMSE of 0.02, and MAE of 0.01, indicating the good simulation performance of the model.
To further explore the drivers of KDC, SHAP values were used to analyze the importance and influence direction of each feature variable on the XGBoost model’s predictions. The SHAP value for Hf was the highest, significantly surpassing other features, and it played a dominant role in KDC. Ele was the second most important feature, followed by secondary features such as Tem, Rd, and Pre. Other features had lower SHAP values but still impacted the model (Figure 7a). The distribution of SHAP values revealed the range and intensity of the impact of each feature variable (Figure 7b). Hf had a wide range of SHAP values, indicating a strong and varied influence. Significant changes were also shown in the SHAP values of features such as Ele, Tem, and Rd. Overall, Hf was the most influential feature in the model, followed by Ele. The SHAP summary plot confirmed the significant and diverse impacts of these features. Additionally, this study found that compared to static feature variables such as soil composition and distance to water bodies, dynamic variables like human footprint and climate-related features contributed more significantly and had more pronounced effects.
Figure 8 illustrates the relationship between KDC and feature variables. Hf: KDC exhibits a negative correlation with increasing Hf, indicating that intensified human activities put pressure on cropland resources, leading to an increase in KDC amplitude. Ele: In low-altitude areas (0–200 m), KDC decreases with increasing elevation, but this effect stabilizes at higher altitudes. Tem: The impact of Tem on KDC is inconsistent, generally showing a negative correlation; higher temperatures may be detrimental to the effective use of cropland. Rd: Within a range of 0–4000 m, Rd shows a positive correlation with KDC, beyond which the effect stabilizes. Pre: Pre exhibits a fluctuating effect on KDC, with an overall negative trend, suggesting that reduced precipitation may not be favorable for maintaining cropland. Pop: An increase in population density, particularly above 200 people per square kilometer, has a significant negative impact on KDC, reflecting high land demand pressure in densely populated areas. Wd: The influence of Wd on KDC decreases with distance, indicating that areas closer to water bodies may maintain higher cropland densities due to their suitability for agriculture. Slope: The impact of slope on KDC is minor, but within certain slope ranges, there is a negative correlation; steeper slopes are less conducive to farming, affecting KDC. Soil texture (Clay and Sand): The effect of Clay on KDC is relatively stable, while a high Sand content significantly increases KDC, indicating that sandy soils have lower cropland utilization efficiency.

4.3.2. Impact of Interactions on KDC

SHAP dependency plots are used to analyze the impact of interactions among feature variables on the outputs of the XGBoost model. The range of SHAP values indicates the type of impact of a feature: a SHAP value of 0 means that the feature has no effect on the model’s output under current conditions; however, a SHAP value greater than 0 indicates a positive impact, and a value less than 0 indicates a negative impact. Figure 9 displays the interactions among the top 15 feature variables. Hf: When Hf is below 5, the SHAP value is near 0, indicating a minor impact on cropland kernel density. However, when Hf exceeds 5, its negative impact significantly increases, especially in low-altitude and densely populated areas, reflecting the consumption of cropland resources due to high human activity intensity. Ele: In areas where Ele is below 200 m, the SHAP values are predominantly negative, indicating a significant negative impact of terrain on KD in low-altitude plains where human activities are concentrated. As altitude increases, this impact diminishes and SHAP values approach 0, suggesting that higher altitudes have a lesser impact on cropland. Tem: When Tem is below 0.06, SHAP values remain relatively stable; however, when Tem exceeds 0.06, SHAP values rapidly decline, indicating that an increase in temperature beyond a certain limit has a more pronounced negative effect on cropland, potentially limiting cropland development. Rd: When Rd is low, its SHAP values are negative and the impact is significant, indicating that proximity to roads may lead to the occupation of cropland or changes in land use. As Rd increases, this negative effect diminishes. This study also found that some features with lower SHAP values, in isolation, may exhibit more significant effects when interacting with other variables. For instance, Clay and Sand have more complex and significant impacts on model predictions in a multivariable interaction environment than when considered alone. This phenomenon suggests that future research should focus more on the complex interactions between different factors to gain a more comprehensive understanding of the dynamics of cropland changes.

4.4. Future Land Use Simulation of Hainan Island under Multiple Scenarios

Using the GeoSOS-FLUS model to simulate future land use changes, this study first simulated 2020 land use based on the 2010 land use data. The simulated results were compared with the actual 2020 land use data, showing a Kappa value of 0.85 and an overall accuracy of 89.45%, indicating that the simulation results are valid.
Subsequently, this study simulated land use on Hainan Island in 2040 (Table 4 and Figure 10). In the natural development scenario, land use mainly follows the current trends. The simulation results show that cropland area decreases significantly by 597 km2, primarily being converted into construction land and woodland. In this scenario, construction land increases by 491.27 km2, reflecting the ongoing trend of urbanization and infrastructure expansion. Woodland also increases, indicating possible afforestation in some areas. However, grassland and unused land show a decreasing trend, with reductions of 89.71 km2 and 40.04 km2, respectively. The ecological protection scenario emphasizes protecting existing ecosystems and limiting the expansion of construction land. The simulation results indicate that the increase in construction land is controlled, increasing only by 67.67 km2. Although the trend of cropland reduction persists, it is relatively minor at 269.11 km2, indicating that the loss of cropland resources is somewhat controlled under ecological protection policies. Woodland increases by 435.46 km2, reflecting the priority protection of ecological zones. Grassland and water areas also decrease, albeit to a lesser extent. In the cropland protection scenario, the focus is on maintaining and increasing cropland area. The results show an increase in cropland area by 448.75 km2, making it the only scenario with an increase in cropland. This indicates that policy interventions can effectively curb the conversion of cropland into other land use types. Woodland and grassland decrease by 213.06 km2 and 174.72 km2, respectively, indicating that some woodland and grassland are converted to cropland. The growth of construction land is also limited, increasing by only 76.03 km2 (Table 4).
Figure 10 illustrates the spatial distribution differences in land use types under different scenarios. In the natural development scenario, the process of urbanization is more pronounced in coastal areas, with significant expansion of construction land, especially in economic centers like Haikou and Sanya. This expansion severely encroaches on coastal cropland, leading to a significant reduction in cropland area. In the ecological protection scenario, the expansion of construction land in these areas is restricted, with strengthened protection of woodland, particularly in mountainous and coastal ecologically sensitive areas. The cropland protection scenario shows that under policy interventions, the cropland area not only is protected but also increases, particularly in areas previously categorized as woodland and grassland being converted to cropland.

5. Discussion

5.1. Spatial and Temporal Evolution of Cropland

Our study indicates that cropland is predominantly concentrated in coastal areas, where changes in cropland during the study period were more pronounced. From 2000 to 2020, cropland decreased by 956.22 km2, mainly converting to woodland and construction land. At the beginning of the 21st century, the national policy of returning farmland to forests led to afforestation on sloping lands prone to soil erosion, thereby restoring forest vegetation and promoting ecological balance. The implementation of this policy significantly reduced the cropland area, facilitating the conversion of sloping farmland into woodland [40,41,42]. It is noteworthy that the conversion of cropland to woodland in Danzhou was particularly evident, partly due to the implementation of the policy of returning farmland to forests and grasslands and partly due to industrial restructuring, where cropland was used for planting economic forests, especially rubber trees [43]. The local government introduced a series of policies supporting the development of the rubber industry, such as providing land subsidies, tax incentives, and technical support, which greatly contributed to the conversion of cropland to woodland in Danzhou [44]. Additionally, with rapid economic growth and accelerated urbanization in China, the urbanization rate in Hainan has been increasing, leading to increases in urban populations and urban expansion and growing demand for urban construction and infrastructure development [45]. To meet the needs of urban development, Hainan Province has continuously expanded the scale of construction land while also introducing a series of policies to encourage investment and promote economic development, including support and guidance for the real estate industry. This increase in construction land has significantly encroached on cropland, causing its loss, a phenomenon particularly evident in economically developed cities such as Haikou and Sanya [26,46,47].

5.2. Driving Forces of Cropland Change

This study used the XGBoost model and SHAP value analysis to deeply explore the factors driving cropland changes on Hainan Island, revealing complex interactions between natural and anthropogenic factors. Among human activities, the increase in human footprint is the primary driver of cropland reduction. Additionally, natural factors such as topography and climate significantly affect the spatiotemporal changes in cropland. The SHAP value analysis indicates that the increase in human footprint is the most significant factor influencing cropland reduction, suggesting that with the growth of the population and intensification of economic activities, cropland areas face increasing pressure from development. In particular, on Hainan Island, increases in tourism and rapid urbanization have led to the conversion of large areas of cropland into construction and commercial land, thereby accelerating cropland reduction. This trend has been observed in many regions globally, as supported by Parras et al. [48] in their study of certain regions in Turkey, which found that economic development is a key factor driving the conversion of cropland into non-agricultural land.
Additionally, climatic factors such as changes in precipitation and temperature also affect cropland changes, though their impact is relatively smaller. This study found that trends in temperature and precipitation changes have a direct impact on the suitability of cropland and agricultural productivity, consistent with Ahmad et al. [49], who noted that climate change has long-term effects on agricultural production. However, in tropical regions like Hainan Island, moderate changes in temperature and humidity can have bidirectional effects on crop growth cycles and yields, both positive or negative [50]. It is crucial to consider the combined effects of these driving factors in future land use planning and management. Effective management of cropland resources requires addressing the challenges posed by natural conditions while also considering the impacts of economic development [51]. Therefore, developing scientific land use policies, rationally allocating resources, and optimizing land use structures are key to achieving regional sustainable development. The findings of this study provide decision-makers with the data support and theoretical basis needed to more precisely identify and quantify these influencing factors, thereby aiding the formulation of cropland protection and sustainable land use strategies [52,53,54].

5.3. Implications for Future Policies

This study delved into the spatiotemporal changes in cropland on Hainan Island from 2000 to 2020, exploring the underlying natural and anthropogenic factors while also predicting future changes under different land use scenarios. These findings are not only significant for understanding the complex mechanisms of cropland change but also provide empirical foundations for formulating scientific land management policies.
(1)
Future land use planning: Facing the dual challenges of global change and regional development needs, future land use planning must be more forward-looking and adaptable. This study simulated land use changes under different policy scenarios, revealing potential changes in land area under natural development, ecological protection, and cropland protection scenarios. This suggests that policymakers should consider long-term ecological, economic, and social impacts when formulating land use policies, allocating land resources reasonably to promote sustainable land use.
(2)
Cropland protection and ecological balance: As awareness of ecological protection increases and food security becomes a concern, future land policies should emphasize the coordinated development of cropland and ecological environment protection. The simulation results under the cropland protection scenario indicate that appropriate policy interventions can effectively prevent the overdevelopment and unreasonable conversion of cropland, thus safeguarding agricultural productivity and ecosystem services. Policymakers need to focus not only on the quantity of cropland but also on enhancing its quality and ecological functions.
(3)
Flexible policy mechanisms and technical support: Future policies should provide flexibility to adapt to rapidly changing environmental and socio-economic conditions. Policy mechanisms should promote technological innovation, support sustainable agricultural practices, and develop land management technologies such as precision agriculture, ecological agriculture, and resource-efficient utilization technologies. Furthermore, policies should encourage multidisciplinary research and interdepartmental collaboration to create an integrated governance policy framework, enhancing the scientific and effective management of cropland.
Future land use policies on Hainan Island need to strike a balance between ensuring food security, protecting the ecological environment, and supporting economic development. Through scientific land planning, strengthened policy support, the promotion of technological innovation, and increased public participation, cropland resources can be effectively managed and protected, driving regional sustainable development.

5.4. Study Limitations

In this study, we provided important insights for regional sustainable development by analyzing the spatiotemporal changes in and driving forces of cropland on Hainan Island from 2000 to 2020. However, there are several limitations to this study:
(1)
The data used in this study cover the period from 2000 to 2020. While this time period captures trends over nearly two decades, it may not fully reveal characteristics over longer time scales. Additionally, the spatial resolution of the land use data is 30 m. Although this resolution is sufficient to capture major trends at a larger scale, it may be limited in detecting finer changes, especially in small-scale land use changes.
(2)
Although we considered both natural factors and anthropogenic factors in this study of cropland changes, there may still be potential influencing factors not included in the model. For example, the impacts of policy changes, shifts in economic development models, and technological advancements might not be fully reflected in the current model.
(3)
The GeoSOS-FLUS model used for simulating future land use changes is based on the correlation between driving factors and land use types in historical data. The model assumes that these correlations will remain valid in future scenarios; however, socio-economic policy and environmental factors could significantly change in the future, altering these correlations. For instance, policy adjustments, economic model shifts, or climate change could change land use patterns, which are challenging to predict accurately using models based solely on historical data.
Future research should consider time series data from longer periods, higher spatial resolution, and a broader range of driving factors. Additionally, using more diverse models and methods, such as integrating big data analysis and artificial intelligence technologies, can enhance our understanding and the predictive capabilities of cropland changes. This approach will provide more accurate and comprehensive scientific evidence for the formulation of sustainable land use policies.

6. Conclusions

This study systematically investigated the spatiotemporal changes in and driving factors of cropland on Hainan Island from 2000 to 2020, using a combination of geographic statistical analysis and machine learning models. It also predicted cropland evolution under different future land use scenarios. The main conclusions are as follows: (1) From 2000 to 2020, the cropland area on Hainan Island decreased by 956.22 km2, causing the center of cropland to shift southwestward by 8.20 km. The reduction in cropland was primarily due to its conversion into construction land and woodland, a phenomenon particularly noticeable in coastal areas. (2) Among anthropogenic factors, the increase in the human footprint was the main reason for the reduction in cropland. Land use changes driven by population growth, especially in economically active and densely populated coastal areas, were key factors in cropland reduction. Natural factors such as topography and climate also significantly influenced cropland changes. (3) Future scenarios show significant differences in cropland area changes. Under the natural development scenario, cropland is expected to continue decreasing, while the ecological protection and cropland protection scenarios can effectively mitigate or reverse the trend of cropland reduction. This indicates that through reasonable policy interventions and land management strategies, cropland resources can be effectively protected and restored. The in-depth analysis provided by this study offers a comprehensive understanding of the spatiotemporal changes on Hainan Island’s cropland resources and provides data support and scientific evidence for formulating effective land protection and management policies. This result has important reference value for sustainable land use strategies in similar regions in China and globally.

Author Contributions

Conceptualization, J.G. and J.C.; methodology, J.L. and S.Q.; software, J.G.; validation, J.G., S.Q. and J.C.; formal analysis, S.Q.; investigation, J.L. and S.Q.; resources, J.C.; data curation, S.Q.; writing—original draft preparation, J.L. and J.C.; writing—review and editing, S.Q. and J.G.; visualization, J.L. and S.Q.; supervision, J.C.; project administration, J.C.; funding acquisition, J.C. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Hainan Province Science and Technology Special Fund, grant number ZDKJ2021033.

Data Availability Statement

The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding authors.

Acknowledgments

We acknowledge the reviewers and academic editors for their positive andconstructive comments and suggestions. We are grateful to the assistant editor and English editor forprocessing our manuscript efficiently.

Conflicts of Interest

Authors jiadong Chen and jianchao Guo were employed by the company Hainan Province Water Conservancy & Hydropower Survey, Design & Research Institute Co., Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References

  1. Gao, H.; Gong, J.; Ye, T.; Maier, M.; Liu, J. Constructing Cropland Ecological Stability Assessment Method Based on Disturbance-Resistance-Response Processes and Classifying Cropland Ecological Types. Sci. Total Environ. 2024, 930, 172673. [Google Scholar] [CrossRef]
  2. Yang, R.; Xu, S.; Gu, B.; He, T.; Zhang, H.; Fang, K.; Xiao, W.; Ye, Y. Stabilizing Unstable Cropland towards Win-Win Sustainable Development Goals. Environ. Impact Assess. Rev. 2024, 105, 107395. [Google Scholar] [CrossRef]
  3. Li, T.; Hong, X.; Liu, S.; Wu, X.; Fu, S.; Liang, Y.; Li, J.; Li, R.; Zhang, C.; Song, X.; et al. Cropland Degradation and Nutrient Overload on Hainan Island: A Review and Synthesis. Environ. Pollut. 2022, 313, 120100. [Google Scholar] [CrossRef]
  4. Qi, X.; Feng, K.; Sun, L.; Zhao, D.; Huang, X.; Zhang, D.; Liu, Z.; Baiocchi, G. Rising Agricultural Water Scarcity in China Is Driven by Expansion of Irrigated Cropland in Water Scarce Regions. One Earth 2022, 5, 1139–1152. [Google Scholar] [CrossRef]
  5. Wang, Y.; Tao, F.; Chen, Y.; Yin, L. Interactive Impacts of Climate Change and Agricultural Management on Soil Organic Carbon Sequestration Potential of Cropland in China over the Coming Decades. Sci. Total Environ. 2022, 817, 153018. [Google Scholar] [CrossRef]
  6. Han, Y.; Yi, D.; Ye, Y.; Guo, X.; Liu, S. Response of Spatiotemporal Variability in Soil pH and Associated Influencing Factors to Land Use Change in a Red Soil Hilly Region in Southern China. CATENA 2022, 212, 106074. [Google Scholar] [CrossRef]
  7. Wang, Y.; Chen, F.; Zhao, H.; Xie, D.; Ni, J.; Liao, D. Characteristics of Agricultural Phosphorus Migration in Different Soil Layers on Purple Soil Sloping Cropland under Natural Rainfall Conditions. Front. Environ. Sci. 2023, 11, 1230565. [Google Scholar] [CrossRef]
  8. Du, B.; Ye, S.; Gao, P.; Ren, S.; Liu, C.; Song, C. Analyzing Spatial Patterns and Driving Factors of Cropland Change in China’s National Protected Areas for Sustainable Management. Sci. Total Environ. 2024, 912, 169102. [Google Scholar] [CrossRef]
  9. Chen, W.; Yang, L.; Zeng, J.; Yuan, J.; Gu, T.; Liu, Z. Untangling the Increasing Elevation of Cropland in China from 1980 to 2020. Geogr. Sustain. 2023, 4, 281–293. [Google Scholar] [CrossRef]
  10. Tao, J.; Zhu, J.; Zhang, Y.; Dong, J.; Zhang, X. Divergent Effects of Climate Change on Cropland Ecosystem Water Use Efficiency at Different Elevations in Southwestern China. J. Geogr. Sci. 2022, 32, 1601–1614. [Google Scholar] [CrossRef]
  11. Deluz, C.; Sebag, D.; Verrecchia, E.; Boivin, P. Soil Organic Matter Thermal Pools as Influenced by Depth, Tillage, and Soil Texture—A Rock-Eval® Analysis Study on the Cropland Soils of the Swiss Plateau. Geoderma 2024, 445, 116871. [Google Scholar] [CrossRef]
  12. Sun, Y.; Zhong, H.; Ding, Y.; Cai, H.; Peng, X. Exploring the Diverse Response of Cropland Vegetation to Climatic Factors and Irrigation across China. Agronomy 2024, 14, 188. [Google Scholar] [CrossRef]
  13. Sanogo, N.D.M.; Dayamba, S.D.; Renaud, F.G.; Feurer, M. From Wooded Savannah to Farmland and Settlement: Population Growth, Drought, Energy Needs and Cotton Price Incentives Driving Changes in Wacoro, Mali. Land 2022, 11, 2117. [Google Scholar] [CrossRef]
  14. Ullah, K.M.; Uddin, K. The Relationships between Economic Growth and Cropland Changes in Bangladesh: An Evidence Based on Annual Land Cover Data. Environ. Chall. 2021, 5, 100252. [Google Scholar] [CrossRef]
  15. Ibarrola-Rivas, M.J.; Granados-Ramírez, R.; Nonhebel, S. Is the Available Cropland and Water Enough for Food Demand? A Global Perspective of the Land-Water-Food Nexus. Adv. Water Resour. 2017, 110, 476–483. [Google Scholar] [CrossRef]
  16. Wu, J.; Jin, S.; Zhu, G.; Guo, J. Monitoring of Cropland Abandonment Based on Long Time Series Remote Sensing Data: A Case Study of Fujian Province, China. Agronomy 2023, 13, 1585. [Google Scholar] [CrossRef]
  17. Yang, C.; Zhai, H.; Fu, M.; Zheng, Q.; Fan, D. Multi-Scenario Simulation of Land System Change in the Guangdong–Hong Kong–Macao Greater Bay Area Based on a Cellular Automata–Markov Model. Remote Sens. 2024, 16, 1512. [Google Scholar] [CrossRef]
  18. Xie, E.; Chen, J.; Peng, Y.; Yan, G.; Zhao, Y. Historical and Future Dynamics of Cropland Soil Organic Carbon Stocks in an Intensive Human-Impacted Area of Southeastern China. Agric. Ecosyst. Environ. 2024, 372, 109098. [Google Scholar] [CrossRef]
  19. Lyu, F.; Pan, Y.; Yu, L.; Wang, X.; Yu, Z. The Impact of Anthropogenic Activities and Natural Factors on the Grassland over the Agro-Pastoral Ecotone of Inner Mongolia. Land 2023, 12, 2009. [Google Scholar] [CrossRef]
  20. Wang, J.; Feng, C.; Hu, B.; Chen, S.; Hong, Y.; Arrouays, D.; Peng, J.; Shi, Z. A Novel Framework for Improving Soil Organic Matter Prediction Accuracy in Cropland by Integrating Soil, Vegetation and Human Activity Information. Sci. Total Environ. 2023, 903, 166112. [Google Scholar] [CrossRef]
  21. Ye, Y.; Qin, Y.; Yu, R.; Wu, Q. Optimization of Chinese Land Spatial Pattern in the Transformation Process of Resource-Based Cities: A Case Study in Tongling City, China. Sci. Rep. 2024, 14, 6092. [Google Scholar] [CrossRef] [PubMed]
  22. Li, M.; Wei, X.; Li, B. Reconstruction of Spatial–Temporal Changes in Cropland Cover from 1650 to 1980 in Taiyuan City. Land 2023, 13, 36. [Google Scholar] [CrossRef]
  23. Wu, Z.; Li, S.; Wu, D.; Song, J.; Lin, T.; Gao, Z. Analysis of Characteristics and Driving Mechanisms of Non-Grain Production of Cropland in Mountainous Areas at the Plot Scale—A Case Study of Lechang City. Foods 2024, 13, 1459. [Google Scholar] [CrossRef] [PubMed]
  24. Fu, C.; Cheng, L.; Qin, S.; Tariq, A.; Liu, P.; Zou, K.; Chang, L. Timely Plastic-Mulched Cropland Extraction Method from Complex Mixed Surfaces in Arid Regions. Remote Sens. 2022, 14, 4051. [Google Scholar] [CrossRef]
  25. Wang, L.; Zhang, S.; Xiong, Q.; Liu, Y.; Liu, Y.; Liu, Y. Spatiotemporal Dynamics of Cropland Expansion and Its Driving Factors in the Yangtze River Economic Belt: A Nuanced Analysis at the County Scale. Land Use Policy 2022, 119, 106168. [Google Scholar] [CrossRef]
  26. Pan, S.; Liang, J.; Chen, W.; Peng, Y. Uphill or Downhill? Cropland Use Change and Its Drivers from the Perspective of Slope Spectrum. J. Mt. Sci. 2024, 21, 484–499. [Google Scholar] [CrossRef]
  27. Wang, Z.; Gong, X.; Zhang, Y.; Liu, S.; Chen, N. Multi-Scale Geographically Weighted Elasticity Regression Model to Explore the Elastic Effects of the Built Environment on Ride-Hailing Ridership. Sustainability 2023, 15, 4966. [Google Scholar] [CrossRef]
  28. Şanlıtürk, İ.H.; Kocabaş, H. Precise Calculation of Inverse Kinematics of the Center of Gravity for Bipedal Walking Robots. Appl. Sci. 2024, 14, 3706. [Google Scholar] [CrossRef]
  29. Qian, T.; Shi, Z.; Gu, S.; Xi, W.; Chen, J.; Chen, J.; Bai, S.; Wu, L. A Water Shortage Risk Assessment Model Based on Kernel Density Estimation and Copulas. Water 2024, 16, 1465. [Google Scholar] [CrossRef]
  30. Xu, Y.; Lin, K.; Hu, C.; Wang, S.; Wu, Q.; Zhang, J.; Xiao, M.; Luo, Y. Interpretable Machine Learning on Large Samples for Supporting Runoff Estimation in Ungauged Basins. J. Hydrol. 2024, 639, 131598. [Google Scholar] [CrossRef]
  31. Hee Shim, S.; Hyun Choi, J. Building an XGBoost Model Based on Landscape Metrics and Meteorological Data for Nonpoint Source Pollution Management in the Nakdong River Watershed. Ecol. Indic. 2024, 165, 112156. [Google Scholar] [CrossRef]
  32. Lai, J.; Zhao, T.; Qi, S. Spatiotemporal Variation in Vegetation and Its Driving Mechanisms in the Southwest Alpine Canyon Area of China. Forests 2023, 14, 2357. [Google Scholar] [CrossRef]
  33. Cai, Y.; Feng, J.; Wang, Y.; Ding, Y.; Hu, Y.; Fang, H. The Optuna–LightGBM–XGBoost Model: A Novel Approach for Estimating Carbon Emissions Based on the Electricity–Carbon Nexus. Appl. Sci. 2024, 14, 4632. [Google Scholar] [CrossRef]
  34. Peng, Z.; Lu, W.; Yuan, L.; Zhang, Y. The Rise of Syndicates: Social Network Analyses of Construction Waste Haulers in Hong Kong Using a Novel InfoMap-XGBoost Method. Resour. Conserv. Recycl. 2024, 206, 107663. [Google Scholar] [CrossRef]
  35. Liu, Y.; Fu, Y.; Peng, Y.; Ming, J. Clinical Decision Support Tool for Breast Cancer Recurrence Prediction Using SHAP Value in Cooperative Game Theory. Heliyon 2024, 10, e24876. [Google Scholar] [CrossRef] [PubMed]
  36. Wang, Y.; Shen, J.; Yan, W.; Chen, C. Backcasting Approach with Multi-Scenario Simulation for Assessing Effects of Land Use Policy Using GeoSOS-FLUS Software. MethodsX 2019, 6, 1384–1397. [Google Scholar] [CrossRef] [PubMed]
  37. Varga, O.G.; Pontius, R.G.; Singh, S.K.; Szabó, S. Intensity Analysis and the Figure of Merit’s Components for Assessment of a Cellular Automata—Markov Simulation Model. Ecol. Indic. 2019, 101, 933–942. [Google Scholar] [CrossRef]
  38. Gao, H.; Qin, T.; Luan, Q.; Feng, J.; Zhang, X.; Yang, Y.; Xu, S.; Lu, J. Characteristics Analysis and Prediction of Land Use Evolution in the Source Region of the Yangtze River and Yellow River Based on Improved FLUS Model. Land 2024, 13, 393. [Google Scholar] [CrossRef]
  39. Chen, J.; You, J.; Wei, J.; Dai, Z.; Zhang, G. Interpreting XGBoost Predictions for Shear-Wave Velocity Using SHAP: Insights into Gas Hydrate Morphology and Saturation. Fuel 2024, 364, 131145. [Google Scholar] [CrossRef]
  40. Huang, X.; Liu, Y.; Cong, Y.; Zhang, Y.; Zhao, X.; Huang, L.; Li, Q.; Li, L.; Hou, F. Soil Moisture Stability of Rangeland Is Higher than That of Woodland and Cropland in the Loess Plateau, China. Ecol. Indic. 2022, 144, 109543. [Google Scholar] [CrossRef]
  41. Zhang, Y.; Yuan, J.; You, C.; Cao, R.; Tan, B.; Li, H.; Yang, W. Contributions of National Key Forestry Ecology Projects to the Forest Vegetation Carbon Storage in China. For. Ecol. Manag. 2020, 462, 117981. [Google Scholar] [CrossRef]
  42. Han, N.; Yu, M.; Jia, P.; Zhang, Y.; Hu, K. Influence of Human Activity Intensity on Habitat Quality in Hainan Tropical Rainforest National Park, China. Chin. Geogr. Sci. 2024, 34, 519–532. [Google Scholar] [CrossRef]
  43. Ren, H.; Chen, J.; Yuan, X.; Liu, J. Composition and Diversity of the Spider Fauna in Rubber Forest and Natural Forest in Hainan Island, China. Acta Ecol. Sin. 2016, 36, 1774–1781. [Google Scholar] [CrossRef]
  44. Zhai, D.-L.; Cannon, C.H.; Slik, J.W.F.; Zhang, C.-P.; Dai, Z.-C. Rubber and Pulp Plantations Represent a Double Threat to Hainan’s Natural Tropical Forests. J. Environ. Manag. 2012, 96, 64–73. [Google Scholar] [CrossRef] [PubMed]
  45. Gu, K.; Wall, G. Rapid Urbanization in a Transitional Economy in China: The Case of Hainan Island. Singap. J. Trop. Geogr. 2007, 28, 158–170. [Google Scholar] [CrossRef]
  46. Shu, R.; Wang, Z.; Guo, N.; Wei, M.; Zou, Y.; Hou, K. Multi-Scenario Land Use Optimization Simulation and Ecosystem Service Value Estimation Based on Fine-Scale Land Survey Data. Land 2024, 13, 557. [Google Scholar] [CrossRef]
  47. Lu, Y.; He, T.; Yue, W.; Li, M.; Shan, Z.; Zhang, M. Does Cropland Threaten Urban Land Use Efficiency in the Peri-Urban Area? Evidence from Metropolitan Areas in China. Appl. Geogr. 2023, 161, 103124. [Google Scholar] [CrossRef]
  48. Parras, R.; De Mendonça, G.C.; Da Costa, L.M.; Rocha, J.R.; Costa, R.C.A.; Valera, C.A.; Fernandes, L.F.S.; Pacheco, F.A.L.; Pissarra, T.C.T. Land Use Footprints and Policies in Brazil. Land Use Policy 2024, 140, 107121. [Google Scholar] [CrossRef]
  49. Ahmad, N.; Younas, F.; Ahmad, H.R.; Sarfraz, M.; Ayub, M.A.; Maqsood, M.A.; Rasul, F.; Sardar, M.F.; Mehmood, T.; Ajarem, J.S.; et al. Soybean (Glycine Max) Cropland Suitability Analysis in Subtropical Desert Climate through GIS-Based Multicriteria Analysis and Sentinel-2 Multispectral Imaging. Land 2023, 12, 2034. [Google Scholar] [CrossRef]
  50. Dou, X.; Guo, H.; Zhang, L.; Liang, D.; Zhu, Q.; Liu, X.; Zhou, H.; Lv, Z.; Liu, Y.; Gou, Y.; et al. Dynamic Landscapes and the Influence of Human Activities in the Yellow River Delta Wetland Region. Sci. Total Environ. 2023, 899, 166239. [Google Scholar] [CrossRef]
  51. Wang, Y.; Zhang, Y.; Xian, Y.; Zhang, J.; Tan, B.; Zhang, Y. Horizontal and Vertical Variations in Soil Fertility in Response to Soil Translocation Due to Tillage-Induced Erosion on Sloping Cropland. CATENA 2024, 242, 108089. [Google Scholar] [CrossRef]
  52. Lu, D.; Wang, Z.; Li, X.; Zhou, Y. Evaluation of the Efficiency and Drivers of Complemented Cropland in Southwest China over the Past 30 Years from the Perspective of Cropland Abandonment. J. Environ. Manag. 2024, 351, 119909. [Google Scholar] [CrossRef] [PubMed]
  53. Wang, J.; Wang, Z.; Li, K.; Li, C.; Wen, F.; Shi, Z. Factors Affecting Phase Change in Coupling Coordination between Population, Crop Yield, and Soil Erosion in China’s 281 Cities. Land Use Policy 2023, 132, 106761. [Google Scholar] [CrossRef]
  54. Zuo, C.; Wen, C.; Clarke, G.; Turner, A.; Ke, X.; You, L.; Tang, L. Cropland Displacement Contributed 60% of the Increase in Carbon Emissions of Grain Transport in China over 1990–2015. Nat. Food 2023, 4, 223–235. [Google Scholar] [CrossRef] [PubMed]
Land 13 01274 g001
Figure 1. The geographic location of the study area.
Figure 1. The geographic location of the study area.
Land 13 01274 g001
Land 13 01274 g002
Figure 2. Spatial distribution of land use types.
Figure 2. Spatial distribution of land use types.
Land 13 01274 g002
Land 13 01274 g003
Figure 3. Cropland transformation from 2000 to 2020.
Figure 3. Cropland transformation from 2000 to 2020.
Land 13 01274 g003
Land 13 01274 g004
Figure 4. Land transfer Sankey diagram.
Figure 4. Land transfer Sankey diagram.
Land 13 01274 g004
Land 13 01274 g005
Figure 5. Spatial distribution of center-of-gravity and standard-deviation ellipses of cropland.
Figure 5. Spatial distribution of center-of-gravity and standard-deviation ellipses of cropland.
Land 13 01274 g005
Land 13 01274 g006
Figure 6. Spatial distribution of cropland KD in (a) 2000, (b) 2010, and (c) 2020 and (d) change from 2000 to 2020.
Figure 6. Spatial distribution of cropland KD in (a) 2000, (b) 2010, and (c) 2020 and (d) change from 2000 to 2020.
Land 13 01274 g006
Land 13 01274 g007
Figure 7. Feature variable importance (a) and SHAP summary (b). Note: Ele, Slop, Clay, Sand, Rd, and Wd are static variables, representing elevation, slope, clay content, sand content, road distance, and water distance, respectively. Tem, Pre, Hf, and Pop are dynamic variables, representing changes in temperature, precipitation, human footprint, and population density, respectively.
Figure 7. Feature variable importance (a) and SHAP summary (b). Note: Ele, Slop, Clay, Sand, Rd, and Wd are static variables, representing elevation, slope, clay content, sand content, road distance, and water distance, respectively. Tem, Pre, Hf, and Pop are dynamic variables, representing changes in temperature, precipitation, human footprint, and population density, respectively.
Land 13 01274 g007
Land 13 01274 g008
Figure 8. (a–j) Correlation between KDC and feature variables.
Figure 8. (a–j) Correlation between KDC and feature variables.
Land 13 01274 g008
Land 13 01274 g009
Figure 9. (a–o) Interactions between feature variables.
Figure 9. (a–o) Interactions between feature variables.
Land 13 01274 g009
Land 13 01274 g010
Figure 10. Land use in 2040 on Hainan Island.
Figure 10. Land use in 2040 on Hainan Island.
Land 13 01274 g010
Table 1. Sources of analyzed data.
Table 1. Sources of analyzed data.
CategoriesDriving FactorsTimeResolutionSource
Land use types 2000–202030 mhttps://www.resdc.cn, accessed on 20 March 2024.
Natural factorsElevation 30 mhttps://www.resdc.cn, accessed on 20 March 2024.
Slope 30 m
Soil clay content 1 km
Soil sand content 1 km
Water distance202030 m
Temperature2000–20201 km
Precipitation2000–20201 km
Anthropogenic factorsHuman footprint2000–20201 kmhttps://www.geodata.cn, accessed on 22 March 2024.
Population density2000–20201 km
Road distance202030 m
Table 2. Land use transition matrixes under different scenarios.
Table 2. Land use transition matrixes under different scenarios.
ScenarioNatural development scenario
Land use typesCroplandWoodlandGrasslandWatersConstruction
land		Unused
land
Cropland111111
Woodland111111
Grassland111111
Waters111111
Construction Land111111
Unused Land111111
ScenarioEcological protection scenario
Land use typesCroplandWoodlandGrasslandWatersConstruction LandUnused Land
Cropland111110
Woodland010000
Grassland011100
Waters010100
Construction Land000010
Unused Land111111
ScenarioCropland protection scenario
Land use typesCroplandWoodlandGrasslandWatersConstruction LandUnused Land
Cropland100000
Woodland111111
Grassland111111
Waters111111
Construction Land000100
Unused Land111111
Note: Here, 1 indicates that conversion is allowed, and 0 indicates that conversion is not allowed.
Table 3. Areas and change rates of land use types.
Table 3. Areas and change rates of land use types.
Land Use TypesArea (km2)Change Rate (%)
2000201020202000–20102010–20202000–2020
Cropland9004.078656.088047.85−3.86%−7.03%−10.62%
Woodland21,827.4821,872.1221,885.660.20%0.06%0.27%
Grassland1229.961154.291196.51−6.15%3.66%−2.72%
Water1281.391503.041471.6317.30%−2.09%14.85%
Construction Land780.69983.761573.9626.01%59.99%101.61%
Unused Land141.1395.4389.11−32.38%−6.61%−36.85%
Table 4. Land use changes under different scenarios in 2040.
Table 4. Land use changes under different scenarios in 2040.
Land Use TypesNatural DevelopmentEcological ProtectionCropland Protection
Area
(km2)		Area Changes
(km2)		Area
(km2)		Area Changes
(km2)		Area
(km2)		Area Changes
(km2)
Cropland7450.85−597.007778.85−269.118496.60448.75
Woodland21,992.60126.9422,321.12435.4621,672.60−213.06
Grassland1106.80−89.711120.07−76.441021.79−174.72
Water1600.17108.541367.05−104.581379.05−92.58
Construction Land2065.23491.271641.6367.671649.9976.03
Unused Land49.07−40.0436.01−53.1044.68−44.43
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Guo, J.; Qi, S.; Chen, J.; Lai, J. Driving Forces behind the Reduction in Cropland Area on Hainan Island, China: Implications for Sustainable Agricultural Development. Land 2024, 13, 1274. https://doi.org/10.3390/land13081274

AMA Style

Guo J, Qi S, Chen J, Lai J. Driving Forces behind the Reduction in Cropland Area on Hainan Island, China: Implications for Sustainable Agricultural Development. Land. 2024; 13(8):1274. https://doi.org/10.3390/land13081274

Chicago/Turabian Style

Guo, Jianchao, Shi Qi, Jiadong Chen, and Jinlin Lai. 2024. "Driving Forces behind the Reduction in Cropland Area on Hainan Island, China: Implications for Sustainable Agricultural Development" Land 13, no. 8: 1274. https://doi.org/10.3390/land13081274

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop