Integrated Assessment of the Impact of Cropland Use Transition on Food Production Towards the Sustainable Development of Social–Ecological Systems

Abstract

Understanding the impact of changes in cropland on food production is crucial for economic development and social stability. In recent years, rapid economic growth and frequent population migration in Guangdong Province have significantly changed cropland use and patterns, posing challenges to cropland protection and food security. This study utilized Landsat-4/5/7/8 time-series imagery from the Google Earth Engine and combined it with deep learning techniques to identify long-term cropland use from 1991 to 2020. Then the Global Agro-Ecological Zones model was applied to assess the impact of various cropland use changes on grain production potential (GPP). On this basis, the intrinsic relationship between population, economic development, and food production was further explored using the center of gravity model and spatial mismatch model. The study finds that Guangdong Province’s cropland area has decreased by approximately 34.16%. The annual average loss due to non-agricultural use and abandonment is 2.75 thousand km² and 3.09 thousand km², respectively, while the average yearly compensated cropland area is 2.94 thousand km². The actual annual food loss could meet the needs of about 4.6 million people. Furthermore, non-agriculturalization is the main way of losing GPP, and the reduction of GPP caused by abandonment cannot be underestimated. When considering the GPP loss due to abandonment, new GPP has not fully compensated for lost GPP. Guangdong Province has rapidly decreased the coordination between food production, population, and economic development, leading to considerable contradictions in the social–ecological systems. Finally, the movement of cropland and population centers in opposite directions has intensified the decoupling phenomenon. The results can guide the development of refined cropland protection policies and promote sustainable development of social–ecological systems.

1. Introduction

Food production is crucial for human survival and advancement, as well as a fundamental guarantee for national security and social stability [1]. The latest Global Food Crisis Report from the United Nations Food and Agriculture Organization (FAO) states that over 280 million people are severely hungry, exacerbated by economic crises, extreme weather, and geopolitical conflicts [2,3]. Typically, changes in the cropland use patterns can directly affect final food production [4]. Since the reform and opening up, Guangdong Province has experienced a significant transition in cropland use due to rapid economic development and continuous urbanization [5]. The proportion of cropland in the province’s total land area has declined from 22.9% to 10.5%, representing a reduction of 22.27 thousand km². In addition, problems such as the quantity, quality, ecological imbalance, non-agriculturalization, and marginalization of cropland have appeared [6,7]. The contradiction between population growth and resource-environment supply and demand has become more pronounced in Guangdong Province, intensifying constraints on effective resource supply [8]. Moreover, rapid population growth and stable economic development have generated new demands for food production [9]. Therefore, it is essential to clarify the relationship between changes in the cropland use patterns and food production to ensure regional food security [10].

In general, non-agriculturalization, cropland abandonment, and cropland compensation are considered to be the most direct ways in which the cropland use transformation affects food production [11,12,13]. Scholars have extensively studied the impact of these three types of cropland use transition on food production. Non-agriculturalization is a major focus in the research on the cropland use transition [14], with large areas of high-quality cropland being permanently lost due to urbanization [15]. Urban expansion has led to an annual reduction of 12.45 million tons in China’s grain output. Additionally, the problem of cropland reduction due to ecological protection projects, such as the conversion of cropland to forest, cannot be ignored [16]. In recent years, the impact of cropland abandonment and cropland compensation on food production has gradually gained attention. Numerous studies have identified an increasing severity of cropland abandonment. From 1990 to 2019, the average area of abandoned farmland in China was recorded at 23,400 km², leading to an annual grain loss sufficient to meet the needs of approximately 20 million people [17]. However, these studies frequently concentrate on particular field survey samples, which may not accurately reflect the impact of abandonment on food production over larger regions. On the other hand, cropland compensation generally exerts a positive influence on regional food production. From 2002 to 2019, the total area of compensated farmland reached 27,900 km², significantly alleviating the negative effects of non-agriculturalization and cropland abandonment [18]. Nonetheless, some research suggests that cropland compensation in China’s mountainous regions has intensified non-food production. Up to 16% of compensated cropland is used for non-food purposes [19]. These studies primarily assess the quality and efficiency of the requisition–compensation balance of cropland. They provide limited quantification of the food production capacity of the newly added cropland.

Overall, previous studies have explored the effects of the cropland use transition on food production from various perspectives, such as horizontal measurement and driving mechanisms, forming a rich theoretical foundation [20,21,22]. However, these studies have lacked a thorough assessment and systematic calculation of the overall impact on food security, focusing instead on single aspects or types of the cropland use transition. They have not clarified which types of the cropland use transition primarily threaten regional food production. Although some studies have quantified the effects of non-agriculturalization, marginalization, and non-grain on GPP in the process of urbanization in China [23], they have overlooked the impact of afforestation on returned cropland, which is a significant ecological project over a certain period. Afforestation on returned cropland contributes as much as 33% to changes in cropland use. Additionally, cropland compensation has not been included in the GPP calculation framework [24]. Furthermore, many studies have used static cross-sectional data on cropland use, failing to accurately depict the spatiotemporal trends in food production [25,26]. Therefore, this study uses a long-time series of the cropland use datasets to systematically reveal the impact and spatial heterogeneity of various types of the cropland use transition on food production. It also takes into account the unique characteristics of rapid population growth and economic development in the study area to further explore the relationship between population size, economic development level, and food production.

Guangdong Province is currently actively constructing the Guangdong–Hong Kong–Macao Greater Bay Area, intending to establish a world-class bay area and a premier city cluster. In the future, Guangdong Province’s economy and population are expected to experience a second phase of rapid growth. Additionally, the proportion of agricultural employment has dropped from 70% in the early days of reform and opening up to 10% in 2020. A large number of people have shifted to non-agricultural activities, leading to further changes in cropland use. Therefore, exploring the impact of the cropland use transition on food production is particularly important to ensure regional food self-sufficiency. The objectives of this study include the following: (1) to characterize the spatiotemporal patterns of non-agriculturalization, cropland abandonment, and cropland compensation, and reveal their impact on food production and spatial heterogeneity; (2) to explore whether the loss and increase of food production caused by these cropland use transitions can be balanced; (3) to reveal the intrinsic relationship between population size, economic development level, and food production within the study area. The findings from this study can inform the development of targeted and nuanced cropland protection policies and explore effective paths for increasing food production.

2. Materials and Methods

2.1. Study Area

Guangdong Province is located in the southernmost part of mainland China, spanning latitudes 20°09′ to 25°31′ North and longitudes 109°45′ to 117°20′ East (Figure 1). It has held the top position in total GDP in China for 32 consecutive years. Guangdong Province possesses the most abundant light, heat, and water resources in China. The primary grain crops are rice and corn. The terrain is higher in the north and lower in the south, featuring hills and mountains in the north and plains in the south. By the end of 2020, Guangdong Province had a permanent population of 126.01 million, an urbanization rate of 71.7%, and only 18.99 thousand km² of arable land, with per capita arable land representing less than one-fifth of the national average, making it one of the largest grain-consuming regions in the country. However, over the past 30 years, rapid urbanization and industrialization have caused considerable reduction of cropland by urban expansion in Guangdong Province. Additionally, the loss of rural labor has resulted in the abandonment of many croplands [27]. Therefore, selecting Guangdong Province, with its dense population, limited cropland, and frequent cropland use changes, as the study area is advantageous for analyzing the impact of the cropland use transition on grain production and regional food security. To better explore the spatial heterogeneity of cropland use transitions on food production, Guangdong Province is divided into four regions according to national land spatial planning [28]: the Pearl River Delta (PRD) region, which includes Guangzhou, Shenzhen, Zhuhai, Foshan, Dongguan, Zhongshan, Huizhou, Jiangmen, and Zhaoqing; the eastern Guangdong (EG) region, encompassing Shanwei, Jieyang, Shantou, and Chaozhou; the western Guangdong (WG) region, comprising Zhanjiang, Maoming, and Yangjiang; and the northern Guangdong (NG) region, which covers Meizhou, Heyuan, Shaoguan, Qingyuan, and Yunfu.

2.2. Data Sources and Processing

This study uses a variety of data, including land use, soil, meteorological, and socio-economic data, as follows (

Table 1. Data description.

Date Type	Time Range	Spatial Resolution	Source
CNLUCC	1990, 1995, 2000, 2005, 2010, 2015, 2020	30 m	Resources and Environmental Science Data Center (https://www.resdc.cn/) accessed on 9 January 2024
Meteorological date	1991–2020	10 km	China Meteorological Data Network (https://data.cma.cn/) accessed on 12 January 2024
Statistical data	1991–2020	-	Guangdong Statistical Yearbook (http://stats.gd.gov.cn/gdtjnj/) accessed on 22 January 2024
DEM	-	30 m	Shuttle Radar Topography Mission data (http://srtm.csi.cgiar.org/srtmdata/) accessed on 9 January 2024
Soil data	-	1 km	Resources and Environmental Science Data Center (https://www.resdc.cn/) accessed on 9 January 2024

):

(1) Land use data: This primarily includes the China Multi-Temporal Land Use and Land Cover Remote Sensing Monitoring Dataset (CNLUCC) for 1990, 1995, 2000, 2005, 2010, 2015, and 2020. This dataset serves as a reference for expanding training samples and verifying the proposed methods. The data uses Landsat remote sensing images as the primary information source and is constructed through manual visual interpretation.

(2) Meteorological and soil data: Meteorological data includes monthly maximum temperature, average temperature, minimum temperature, precipitation, relative humidity, number of rainy days per month, wind speed, and solar radiation. Based on the monthly data of these eight key plant growth factors, this study interpolates meteorological data to a resolution of 10 km. Finally, this study clips the national interpolated meteorological data according to the administrative boundaries of Guangdong Province to obtain the meteorological data for the entire province. These interpolated monthly meteorological data are obtained based on daily observation data of meteorological element stations at more than 2400 stations nationwide and based on calculating the monthly values of each meteorological element. Soil data includes attributes such as composition, type, entire profile depth, and water-holding capacity. The Digital Elevation Model (DEM) is obtained from the Shuttle Radar Topography Mission (SRTM), a collaborative project between NASA and the National Geospatial-Intelligence Agency (NGA), providing coverage for all of China at a spatial resolution of 30 m.

(3) Socio-economic data: Actual yield data for various prefecture-level cities in Guangdong Province from 2000 to 2020, as well as population and economic data from 1991 to 2020, are all sourced from the Guangdong Statistical Yearbook.

2.3. GPP Measurement for Cropland Use Transition

This study categorizes cropland use transitions into three categories: non-agriculturalization, cropland abandonment, and cropland compensation, and systematically analyzes their impact on GPP. Non-agriculturalization refers to converting cropland into construction land, forest land, garden land, grassland, water bodies, or other non-agricultural uses, ultimately disrupting local agricultural ecosystems and crop cultivation and production [29]. Cropland compensation refers to converting non-agricultural land types into cropland. The Green Chlorophyll Vegetation Index (GCVI) is an indicator that linearly correlates with the content of vegetation chlorophyll and the Leaf Area Index (LAI), effectively reflecting the growth status and chlorophyll content of vegetation. The GCVI maintains high sensitivity even in dense vegetation, unaffected by the saturation of LAI in thick canopies. Therefore, the GCVI can be used as a key indicator to effectively distinguish land use types (grassland, farmland, and forest) with different vegetation coverage in the study area [30,31]. In this study, Landsat images processed via the Google Earth Engine are utilized to extract annual GCVI-max and GCVI-diff feature images. Then, CNLUCC data is combined with ground truth data from visual interpretation and a feature image library. On this basis, deep learning techniques are applied to generate a 30 m spatial resolution cropland distribution map for Guangdong Province from 1990 to 2020 [32]. After analyzing the annual changes in cropland, the spatial distribution of non-agriculturalization and cropland compensation is obtained. Finally, the Global Agro-Ecological Zones model (GAEZ) is used to estimate the GPP loss for each pixel (Figure 2).

$${\mathrm{G}\mathrm{P}\mathrm{P}}_{\mathrm{c}\mathrm{n}\mathrm{a}}={\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{C}\mathrm{N}\mathrm{A}\mathrm{S}}_{\mathrm{i}}\times {\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{i}}\hspace{1em}\mathrm{i}=1,2,3,\dots ,$$

(1)

$${\mathrm{G}\mathrm{P}\mathrm{P}}_{\mathrm{c}\mathrm{c}}={\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{C}\mathrm{C}\mathrm{S}}_{\mathrm{i}}\times {\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{i}}\hspace{1em}\mathrm{i}=1,2,3,\dots ,$$

(2)

where ${\mathrm{G}\mathrm{P}\mathrm{P}}_{\mathrm{c}\mathrm{n}\mathrm{a}}$ and ${\mathrm{G}\mathrm{P}\mathrm{P}}_{\mathrm{c}\mathrm{c}}$ represent the GPP loss caused by non-agriculturalization and the GPP from cropland compensation, respectively. ${\mathrm{C}\mathrm{N}\mathrm{A}\mathrm{S}}_{\mathrm{i}}$ and ${\mathrm{C}\mathrm{C}\mathrm{S}}_{\mathrm{i}}$ represent the areas of non-agriculturalization and cropland compensation for the i-th pixel, respectively. The GAEZ model estimates the climatic suitability for growing specific crops based on climate conditions and then uses an empirical formula to calculate the crop’s production potential, considering both irrigation and rain-fed scenarios [33]. Under rain-fed conditions, the influence of precipitation on crop yield is solely considered, while under irrigation conditions, it is assumed that water is sufficient, and thus the impact of water on crops is not considered [34]. Guangdong Province is a primarily rain-fed agricultural area. ${\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{i}}$ represents the unit area GPP for the i-th pixel under rain-fed conditions, calculated using the GAEZ model. The main crops considered are rice and maize, accounting for about 90% of the total grain production in Guangdong Province. To accurately depict the impact of cropland use transitions on GPP, the climatic conditions, elevation, soil, and other factors for calculating ${\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{i}}$ remain constant. The input land use data represents three different cropland use transitions. Additionally, the climate conditions are based on the average climate values from 1991 to 2020.

Cropland abandonment refers to the process whereby previously cultivated land progressively reverts to natural vegetation due to a lack of agricultural management for a period of at least two to five years [35]. Based on the obtained annual cropland maps for 1990–2020, this study combines the difference between the annual Normalized Difference Snow Index (NDSI) and the Normalized Difference Vegetation Index (NDVI), along with NDVI feature images. Using prior knowledge and natural laws, we applied a pixel-based sliding window time-series correction algorithm to generate a 30 m spatial resolution map of cropland abandonment in Guangdong Province [36]. The following formula was used to estimate the GPP loss caused by abandonment for each pixel:

$${\mathrm{G}\mathrm{P}\mathrm{P}}_{\mathrm{c}\mathrm{a}}={\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{C}\mathrm{A}\mathrm{S}}_{\mathrm{i}}\times {\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{i}}\hspace{1em}\mathrm{i}=1,2,3,\dots ,$$

(3)

where ${\mathrm{G}\mathrm{P}\mathrm{P}}_{\mathrm{c}\mathrm{a}}$ represents the GPP loss caused by cropland abandonment; ${\mathrm{C}\mathrm{A}\mathrm{S}}_{\mathrm{i}}$ represents the area of abandoned cropland for the i-th pixel; and ${\mathrm{y}\mathrm{i}\mathrm{e}\mathrm{l}\mathrm{d}}_{\mathrm{i}}$ is as previously defined.

This study estimates the impact of different cropland use transitions on GPP in Guangdong Province based on high-precision, long-term land cover data. To verify the accuracy of the estimation results, the GPP of various prefecture-level cities in Guangdong Province from 2000 to 2020 was compared with the actual yield. As shown in Figure 3, the Pearson correlation coefficient is 0.96, indicating a strong correlation. Therefore, the trend of GPP largely reflects the changes in actual yield.

2.4. Spatiotemporal Balance Calculation for Newly Added GPP and Lost GPP

Guangdong Province has maintained a requisition–compensation balance of cropland area for 23 consecutive years. However, this balance must also consider the quality of cropland. GPP directly reflects the production capacity of cropland, serving as a key indicator of cropland quality. Non-agriculturalization and abandonment decrease GPP, while cropland compensation increases GPP. To determine whether the newly added GPP in Guangdong Province can offset the negative impact of GPP loss, this study developed the following formula:

$$\mathrm{B}\mathrm{V}={\mathrm{G}\mathrm{P}\mathrm{P}}_{\mathrm{c}\mathrm{c}}-({\mathrm{G}\mathrm{P}\mathrm{P}}_{\mathrm{c}\mathrm{a}}+{\mathrm{G}\mathrm{P}\mathrm{P}}_{\mathrm{c}\mathrm{n}\mathrm{a}})$$

(4)

where BV represents the balance value of the newly added and lost GPP, and the other parameters are as described in Section 2.3. To more accurately reflect the relationship between the newly added GPP and the lost GPP in a prefecture-level city, this study further categorizes them, as shown in the

Table 2. Categorization.

Status	Serious Imbalance	Less Balanced	Basic Equilibrium	Perfect Balance
BV (104t)	−10<	−10~−1	−1~1	>1

.

2.5. Spatial Mismatch Analysis

Guangdong Province has consistently been at the forefront of China in terms of population and economic development. Despite having only 1.9% of the national land area, it supports 8.9% of the population and contributes 10.9% to China’s total economic output. The PRD region exhibits high economic development and population density but relatively low grain production, whereas the WG has abundant grain production but an underdeveloped economy. In addition to calculating the impact of the cropland use transformation on GPP, this study explores the spatial mismatch and evolutionary trends among grain production, population, and economic development by introducing the Spatial Mismatch Index [37]. GDP is utilized as an indicator of economic development. The specific formula is:

$${\mathrm{S}\mathrm{M}\mathrm{I}}_{\mathrm{i}}={\displaystyle \frac{1}{\mathrm{G}}}\left\{\left[\left({\displaystyle \frac{{\mathrm{P}}_{\mathrm{i}}}{\mathrm{P}}}+{\displaystyle \frac{{\mathrm{E}}_{\mathrm{i}}}{\mathrm{E}}}\right)/2\right]\times \mathrm{G}-{\mathrm{G}}_{\mathrm{i}}\right\}\times 100$$

(5)

where P, E, and G represent the total population, total economic development level, and total GPP of each prefecture-level city, respectively; ${\mathrm{P}}_{\mathrm{i}}$, ${\mathrm{E}}_{\mathrm{i}}$, and ${\mathrm{G}}_{\mathrm{i}}$ represent the population, economic development level, and GPP of city i. The absolute values reflect the spatial mismatch intensity between GPP and population, and economic development level in city i. The larger the absolute value, the more pronounced the spatial mismatch. A positive value indicates that population growth and economic development are progressing more rapidly than GPP, whereas a negative value suggests that population growth and economic development are lagging behind GPP development.

After measuring the spatial mismatch degree of each prefecture-level city, we calculate the overall spatial mismatch index of Guangdong Province. Finally, we decompose the average contribution rate of each prefecture-level city to the overall spatial mismatch index from 1991 to 2020. The formula is as follows:

$$\mathrm{S}\mathrm{M}\mathrm{I}={\displaystyle \frac{1}{21}}\sum _{\mathrm{i}=1}^{21}{\mathrm{S}\mathrm{M}\mathrm{I}}_{\mathrm{i}}$$

(6)

$${\mathrm{R}}_{\mathrm{i}}={\displaystyle \frac{1}{30}}\sum _{\mathrm{k}=1}^{30}{\displaystyle \frac{\left|{\mathrm{S}\mathrm{M}\mathrm{I}}_{\mathrm{i}}\right|}{{\sum }_{\mathrm{i}=1}^{21}\left|{\mathrm{S}\mathrm{M}\mathrm{I}}_{\mathrm{i}}\right|}}$$

(7)

where $\mathrm{S}\mathrm{M}\mathrm{I}$ represents the overall spatial mismatch index; ${\mathrm{R}}_{\mathrm{i}}$ represents the average contribution rate of the i-th prefecture-level city to the overall spatial mismatch index from 1991 to 2020; k represents the number of years since the starting year of the study.

3. Results

3.1. Overall Characteristics of Cropland Use Transition in Guangdong Province

From 1991 to 2020, the total area of cropland in Guangdong Province showed a decreasing trend, which was opposite to the trend of the cropland use transition rate (Figure 4). Specifically, during the study period, the total cropland area decreased by 9.85 thousand km², a decrease of about 35%. The reduction in cropland area occurred in three distinct phases: 1992–1994, 1998–2000, and 2004–2016. During the first phase, the cropland area decreased by 8.42 thousand km², while the cropland use transition rate increased by 4.31%. This phase saw the highest loss of cropland due to abandonment and non-agricultural use in nearly 30 years, averaging an annual loss of 8.74 thousand km². The cropland use transition rate continued to decrease after 1994, reaching its lowest point of 20.18% in 2001. During the second phase, the cropland area decreased by 4.52 thousand km², with only 1.32 thousand km² lost to abandonment in 1999. The period from 2004 to 2016 marked the longest continuous decline in cropland area. Economic development triggered a new wave of urbanization, resulting in a total reduction of cropland area by 10.4 thousand km². Meanwhile, the cropland use transition rate rose rapidly, reaching its maximum in the study period in 2013 at 34.14%, highlighting the intensity of the cropland use transition. Additionally, from the spatial distribution of the cropland use transition (Figure 5c), the most frequent occurrence of cropland use transition was in Zhanjiang and Maoming in WG, as well as Jieyang in EG, where the cropland use transition rate in many areas exceeded 60%. The PRD region followed, with an overall cropland use transition rate of 20–40%, and the region with the smallest cropland use transition changes was NG. Overall, the annual average loss of cropland area due to non-agriculturalization and abandonment was 2.76 thousand km² and 3.09 thousand km², respectively, while the annual average replenishment of cropland area was 2.94 thousand km².

Based on the distribution map of cropland and population gravity centers (Figure 5a,b,d), it’s clear that their gravity centers shifted in opposite directions. The longitude and latitude changes show negative correlations of −0.9 and −0.79, indicating more pronounced changes in longitude. Specifically, from 1991 to 2020, the cropland gravity center shifted northwestward by 21.2 km at a rate of 0.71 km/a, moving 20.18 km westward and 6.52 km northward. The study also finds that the center of gravity of cropland is shifting toward the northwest hilly area. This may lead to a decrease in the soil fertility of newly added cropland, thus affecting food production. Meanwhile, the population gravity center shifted southeastward by 11.35 km at a rate of 0.38 km/a, moving 9.22 km eastward and 6.62 km southward. The population center of gravity has further shifted to the economically developed coastal areas. This indicates an intensifying decoupling between cropland and population distribution in Guangdong province.

3.2. Influence of Different Cropland Use Transitions on GPP

3.2.1. Loss of GPP Due to the Cropland Non-Agriculturalization

Throughout the study period, the non-agricultural area in Guangdong Province exhibited an overall decreasing trend (Figure 6), with a total decreasing area of 82.7 thousand km² and a non-agriculturalization rate of 1.53%. The regions mainly affected by non-agriculturalization are the PRD and WG, losing 28.7 thousand km² and 25 thousand km² of cropland, respectively. The cities with the largest non-agriculturalization areas are Zhanjiang, Qingyuan, and Maoming, while those with the highest non-agriculturalization rates are Guangzhou, Dongguan, and Zhongshan. Temporally, the most significant non-agriculturalization of cropland occurred from 1991 to 2000, reaching 37.8 thousand km², whereas 2010–2020 saw the least, with only 20.3 thousand km². This change is likely attributed to the stabilization of urbanization processes in Guangdong Province and a shift in economic development from extensive to intensive growth.

From 1991 to 2020, the non-agriculturalization of cropland in Guangdong Province resulted in a total GPP loss of 34.11 million tons (

Table 3. Statistics on the area and GPP of cropland non-agriculturalization.

Districts		Area (km2)			GPP (104t)			Area (km2)	GPP (104t)
		1991–2000	2001–2010	2011–2020	1991–2000	2001–2010	2011–2020	1991–2020	1991–2020
PRD	Dongguan	1158.99	469.63	361.06	57.20	22.59	19.87	28,745.96	1240.26
	Foshan	1361.72	757.19	635.23	75.66	43.78	42.33
	Guangzhou	2420.80	1233.28	1015.31	104.70	56.55	52.50
	Shenzhen	627.67	334.45	285.74	25.35	12.54	11.83
	Zhongshan	687.18	383.91	324.66	30.55	17.40	16.26
	Zhuhai	481.80	267.52	268.34	7.00	3.67	4.10
	Huizhou	2486.67	1459.87	1288.00	99.05	63.73	67.82
	Jiangmen	2104.44	1868.60	1531.27	79.22	71.34	75.69
	Zhaoqing	2242.41	1505.76	1184.45	76.23	51.53	51.75
EG	Shantou	712.80	431.87	396.04	28.08	17.67	18.96	9188.98	222.60
	Shanwei	1077.09	623.03	753.06	36.94	23.22	33.22
	Jieyang	1592.06	991.40	863.10	65.88	40.97	45.34
	Chaozhou	944.57	470.23	333.74	31.83	17.03	15.64
WG	Zhanjiang	2970.04	3381.26	2721.06	186.57	212.95	205.66	19,787.06	1056.41
	Yangjiang	1773.93	1524.91	928.59	61.19	59.63	43.54
	Maoming	2636.22	2334.55	1516.52	103.76	100.50	82.61
NG	Meizhou	2693.25	1149.06	1270.54	73.08	33.74	51.18	24,987.57	891.54
	Yunfu	1675.31	1187.77	706.88	41.43	31.00	28.23
	Shaoguan	2371.74	1210.99	1158.23	106.58	49.14	60.63
	Qingyuan	3308.60	2057.36	1563.65	117.00	72.53	72.69
	Heyuan	2447.41	1006.60	1180.17	72.91	32.89	48.50

). The spatial distribution of this loss is similar to that of non-agricultural cropland, primarily concentrated in the PRD and WG, losing 12.4 million tons and 10.56 million tons of GPP, respectively. There is considerable variation in GPP loss among different cities; Zhanjiang recorded the highest loss at 6.05 million tons, representing 17% of the province’s total; whereas Zhuhai had the lowest at 0.15 million tons, reflecting the lowest per-unit area yield in the study region.

3.2.2. Loss of GPP Due to the Cropland Abandonment

During the study period, the area of abandoned cropland in Guangdong Province remained relatively stable, totaling 92.7 thousand km², with an abandonment rate of 13.75% (Figure 7). The most severe abandonment occurred in the WG, particularly in cities like Zhanjiang and Maoming, accounting for a total area of 35.1 thousand km². By contrast, cities such as Shenzhen, Zhongshan, and Zhuhai had the least abandoned cropland, with Zhongshan accounting for only 301.87 km². It’s worth noting that the spatial pattern of cropland abandonment in the PRD exhibited lower abandonment rates in the core area and high rates in the surrounding areas. This pattern may be due to the rapid economic development in the PRD, which created a significant “siphoning effect”, attracting many people from surrounding areas to migrate to the core region for non-agricultural activities. This population movement significantly reduced the agricultural labor force in surrounding areas, leading to gradual cropland abandonment and exacerbating the abandonment phenomenon.

Throughout the study period, cropland abandonment in Guangdong Province resulted in a decrease in GPP of 21.45 million tons, averaging an annual loss of 0.71 million tons (

Table 4. Statistics on the area and GPP of cropland abandonment.

Districts		Area (km2)			GPP (104 t)			Area (km2)	GPP (104 t)
		1991–2000	2001–2010	2011–2020	1991–2000	2001–2010	2011–2020	1991–2020	1991–2020
PRD	Dongguan	251.34	155.68	252.20	5.62	3.86	6.40	22,891.76	511.33
	Foshan	312.87	284.48	473.18	7.79	7.56	13.22
	Guangzhou	1271.63	1065.69	1108.78	25.95	25.44	29.55
	Shenzhen	237.85	156.49	179.22	4.22	2.73	3.39
	Zhongshan	87.45	71.61	142.81	1.76	1.40	3.09
	Zhuhai	119.51	114.91	161.45	0.74	1.02	1.29
	Huizhou	2233.60	1886.40	2254.87	42.57	48.17	59.62
	Jiangmen	1961.82	1901.88	1889.00	39.99	44.66	47.31
	Zhaoqing	1667.43	1145.09	1504.50	30.87	20.98	32.10
EG	Shantou	566.59	449.93	662.83	9.74	10.94	15.23	12,326.04	158.57
	Shanwei	1304.80	1062.78	1406.83	24.53	23.08	33.66
	Jieyang	2085.24	1387.45	1502.56	45.94	35.55	42.81
	Chaozhou	876.09	477.51	543.43	16.61	11.84	12.93
WG	Zhanjiang	5201.53	5912.45	5052.79	156.31	186.23	206.78	35,124.60	1005.33
	Yangjiang	2856.83	2411.39	1848.11	57.62	54.46	45.06
	Maoming	4455.34	4498.46	2887.70	104.24	113.01	81.63
NG	Meizhou	1608.55	1050.54	1657.29	22.98	17.81	31.86	22,360.76	469.98
	Yunfu	1327.34	1391.09	1213.15	21.33	24.41	26.75
	Shaoguan	1157.28	897.27	1415.19	29.26	21.91	44.20
	Qingyuan	2107.13	1698.49	2129.01	47.30	38.81	56.26
	Heyuan	1943.87	1038.97	1725.60	32.60	18.14	36.35

). Among them, cropland abandonment in the WG had a significant impact on grain production, directly causing a loss of 10.53 million tons of GPP. Conversely, the eastern part of Guangdong was surrounded by mountains on three sides and by the sea on one side. Most of the cropland was distributed in the plains of Shantou and Shanwei. The favorable topographic conditions were conducive to reducing the occurrence of the phenomenon of cropland abandonment. As a result, this area experienced the least impact from cropland abandonment, with GPP losses accounting for only 7.39% of the total, despite cropland abandonment covering 13.29% of the total area.

3.2.3. Increase in GPP Due to Cropland Compensation

Guangdong Province has accumulated a total of 89.7 thousand km² of compensated cropland, experiencing significant increases in cropland area across all regions except for EG (Figure 8). Among them, the PRD has seen the largest increase in cropland area, reaching 27.6 thousand km². However, the cities with the largest increases in cropland area are Zhanjiang and Maoming in the WG, with 13.3 thousand km² and 8.24 thousand km², respectively, accounting for 24.07% of the total increase. Examining different periods, the additional cropland area decreased from 33.9 thousand km² in 1991–2000 to 25.6 thousand km² in 2010–2020. This represents an average annual decrease of 276.83 km², aligning with the trend observed in non-agriculturalized areas.

Unlike the other two types of cropland use transitions, complemented cropland can enhance crop planting area and GPP (

Table 5. Statistics on the area and GPP of complemented cropland.

Districts		Area (km2)			GPP (104t)			Area (km2)	GPP (104t)
		1991–2000	2001–2010	2011–2020	1991–2000	2001–2010	2011–2020	1991–2020	1991–2020
PRD	Dongguan	666.75	449.26	386.02	29.59	22.45	23.19	27,646.26	1212.26
	Foshan	832.73	780.58	696.76	45.40	43.14	49.49
	Guangzhou	1500.35	1400.05	1307.57	64.09	62.65	65.98
	Shenzhen	438.76	343.76	319.11	15.47	12.92	13.49
	Zhongshan	413.48	386.75	364.47	17.59	16.84	17.72
	Zhuhai	308.98	276.86	332.72	4.15	3.94	4.96
	Huizhou	2065.61	1837.81	1759.69	89.06	83.49	93.92
	Jiangmen	1959.56	2148.80	1755.18	77.42	84.47	88.62
	Zhaoqing	1816.27	1722.22	1376.17	65.61	53.55	63.07
EG	Shantou	644.49	535.57	512.59	26.15	21.86	24.73	10,265.30	264.08
	Shanwei	1157.29	913.34	1027.29	44.88	36.52	46.38
	Jieyang	1564.35	1249.34	1073.70	70.22	56.56	59.38
	Chaozhou	649.11	515.66	422.58	24.54	19.46	19.56
WG	Zhanjiang	4828.42	4440.01	4087.52	296.44	282.80	316.81	26,817.51	1520.07
	Yangjiang	2059.52	1909.23	1250.46	82.04	77.35	60.27
	Maoming	3171.66	3217.11	1853.58	141.26	147.28	115.83
NG	Meizhou	1693.05	1351.37	1425.61	54.02	39.81	58.91	24,991.77	1006.29
	Yunfu	1490.85	1411.38	1027.13	50.55	40.33	44.87
	Shaoguan	1925.13	1602.81	1257.45	100.03	66.32	74.64
	Qingyuan	2864.81	2491.39	1913.23	121.56	90.52	101.41
	Heyuan	1854.51	1231.11	1451.94	64.67	40.12	58.55

). From 1991 to 2020, Guangdong Province increased its GPP by 40.02 million tons due to additional cropland, with an average annual increase of 1.33 million tons. Among them, the PRD saw the largest increase in GPP, reaching 12.12 million tons. Specifically, Zhanjiang and Maoming, having the largest increases in cropland area, contributed significantly to the increase in GPP, with increments of 8.96 million tons and 4.04 million tons, respectively. Overall, additional compensated cropland has played a crucial role in stabilizing cropland areas and enhancing agricultural production capacity in Guangdong Province.

3.3. Spatial and Temporal Comparison of Added and Lost GPP

As depicted in Figure 9, from 1991 to 2020, Guangdong Province lost 55.6 million tons of GPP and gained 40 million tons, resulting in a deficit of 15.5 million tons. Most prefecture-level cities failed to achieve a balance in GPP. Cities that maintained a basic GPP balance were mainly in the PRD, including Zhongshan, Zhuhai, Foshan, Dongguan, and Shenzhen. Zhanjiang experienced a serious imbalance, losing up to 261 thousand tons in 2010. From 1991 to 1995, 15 prefecture-level cities faced serious imbalances, driven by the reform policies that spurred significant economic and social development, resulting in extensive cropland being encroached upon or abandoned. Since the early 21st century, China has implemented the strictest cropland protection measures. During the 2000–2005 period, GPP was significantly supplemented, with 11 prefecture-level cities achieving basic equilibrium or even perfect balance. These well-compensated areas were mainly located in the western Guangdong region. After this period, the GPP balance improved compared to the 1990s. According to China’s policy of the requisition–compensation balance of cropland, Guangdong Province has generally achieved a dynamic balance in cropland area and grain production. However, considering the impact of cropland abandonment on GPP, the newly added GPP has not completely offset the lost GPP.

3.4. Spatial Mismatch Analysis of Population, Economy, and GPP

From 1991 to 2020, Guangdong Province experienced a rapid increase in the overall spatial mismatch index, indicating that the degree of spatial mismatch has intensified, and the coordination between food production and the levels of population and economic development has declined (Figure 10). In terms of the directions of the mismatch, the positive mismatch areas are primarily located in the PRD and EG regions, whereas the negative mismatch areas are mainly found in the NG and WG regions. Specifically, 10 prefecture-level cities have consistently been in the negative mismatch zone, with Zhanjiang showing the highest negative mismatch index. By the end of the study period, Zhanjiang’s mismatch index reached −23.15, showing a change range of 6.57% over the period. Most prefecture-level cities have moved towards high negative mismatch values, indicating that food production capacity is increasingly outpacing the demands of population growth and economic development. Only Jiangmen and Huizhou maintained stable spatial mismatch indices. By contrast, 11 prefecture-level cities have long been in the positive mismatch zone. Guangzhou, Shenzhen, and Dongguan quickly reached high positive mismatch values, indicating that regional food production increasingly lags behind population growth and economic development. Shenzhen’s spatial mismatch index increased by 14.86%, primarily due to its strong attraction to the external population and rapid economic growth over recent decades. Meanwhile, food production lagged significantly behind, resulting in poor spatial coordination. Conversely, Jieyang and Shantou gradually decreased to low positive mismatch values. Only three prefecture-level cities, Zhaoqing, Yunfu, and Jieyang, experienced fundamental changes in their spatial mismatch indices, shifting from positive to negative values. This indicates a shift from food production being relatively advanced to relatively lagging behind population and economic development, with low levels of spatial mismatch. Due to differences in geographical location, cropland resources, population density, and socioeconomic levels among cities, their contributions to the overall spatial mismatch index of Guangdong Province vary. Most prefecture-level cities have low contributions to the overall spatial mismatch, maintaining around 3%. Only Zhanjiang, Guangzhou, and Shenzhen have contributions of 20.98%, 14.14%, and 12.78%, respectively, with Guangzhou and Shenzhen experiencing significant human–grain relationship conflicts.

4. Discussion

4.1. Main Cropland Use Transitions Causing GPP Loss in Guangdong Province

China’s non-agriculturalization process has experienced rapid development followed by gradual stabilization [38]. Non-agriculturalization, which sacrifices cropland, poses a significant challenge to food production [39]. In China, two-thirds of urbanization has directly contributed to cropland loss, with 37% of the lost cropland being high-quality [40]. Our study results are similar, showing that from 1991 to 2020, non-agriculturalization caused a GPP loss of about 34.1 million tons, which is significantly higher than the GPP loss of 21.5 million tons caused by cropland abandonment, with a difference of about 12.7 million tons. This indicates that cropland non-agriculturalization remains the dominant factor affecting food production (Figure 11). Using the “400 kg per capita food security line” proposed by the FAO to quantify food loss, we find that the annual loss could meet the needs of approximately 4.6 million people, equivalent to the population of Zhongshan in Guangdong province. Notably, from 1991 to 2020, the per-unit area production potential of abandoned cropland was generally lower than that of non-agriculturalized cropland, indicating lower quality of abandoned cropland during this period. This is because abandonment typically occurs in hilly or mountainous areas with poor natural conditions, such as northern Guangdong, where cropland is fragmented, slopes are steep, soil fertility is low, and agricultural infrastructure is weak, severely restricting regional grain production capacity [41]. Compared to flatlands, hillside cropland demands greater labor input, increasing costs and reducing marginal profits. This makes cropland abandonment more likely and decreases the likelihood of its reuse after abandonment [42]. By contrast, non-agriculturalization often occurs alongside urban expansion. Cropland in these areas have been cultivated for a long time, are conveniently located, and have fertile soil. Only in 2001, 2017, and 2018 did the production potential per unit area of supplemented cropland exceed that of the reduced cropland.

4.2. Impact of Population and Economy on GPP Changes

Maintaining the stability of cropland area, quality, and structure is vital for ensuring regional grain productivity [43]. The quantity and structure of cropland directly influence grain crops, while the spatial transfer of cropland alters natural production conditions, including temperature, water, light, and soil, thereby affecting production potential. This study primarily investigates the impact of the cropland use transition on GPP, including abandonment, non-agriculturalization, and cropland compensation. Non-agriculturalization and abandonment are largely driven by population and economic development [44]. Due to policy support and an advantageous location adjacent to Hong Kong and Macau, Guangdong province has experienced rapid economic development, promoting urbanization and industrialization processes. This economic growth has created numerous non-agricultural employment opportunities. It has attracted rural populations within Guangdong to migrate to cities and drawn labor from other provinces, such as Hunan and Jiangxi [45]. This has formed a strong talent market and labor advantage, which contributes to economic growth. This manifests as a spatial pattern of population and economic agglomeration, mutually promoting each other, and this trend is anticipated to persist. However, the rapid increase in population and economic development has driven soaring demand for construction land, leading to the non-agriculturalization of high-quality cropland near cities and significantly threatening food production [46]. Additionally, the reduction in rural population and agricultural labor in Guangdong province, coupled with higher income and opportunities brought by non-agricultural employment, has increased the opportunity cost of agricultural production. This has resulted in a swift increase in agricultural labor costs and overall agricultural expenses, causing large-scale cropland abandonment, which negatively impacts food production.

4.3. Impact of Cropland Protection Policies on Agricultural Production

Since the 1980s, China has implemented cropland protection policies to ensure national food security. The basic national policy of “cherishing and rationally utilizing every inch of land, and effectively protecting cropland” was first proposed in 1986 [47]. After 1990, social phenomena such as ‘development fever’ and ‘real estate fever’ emerged, causing a surge in construction land demand and accelerated the loss of cropland in Guangdong Province. Between 1992 and 1994, 8420 km² of cropland was lost, primarily due to non-agricultural conversion. To curb this phenomenon, the government adopted the “Regulations on the Protection of Basic Farmland” in 1994, introducing protection red lines to regulate agricultural practices [48]. However, as urbanization accelerated at the beginning of the 21st century, the pressure on cropland protection intensified, prompting the government to introduce a policy of balancing cropland with compensation through the “dynamic balance of the total amount of cropland”. Despite these efforts, studies indicate that much of the compensated cropland is located in mountainous areas and is of lower quality. Furthermore, cropland protection policies appear to have a limited impact on curbing cropland abandonment, especially in areas with poor agricultural conditions [49,50].

To improve the quality of cropland, a comprehensive cropland protection system focusing on land use control and special protection of permanent basic farmland was established in 2012 [51]. After this period, the amount of cropland in Guangdong Province remained relatively stable, and the GPP capacity of newly added cropland has improved. Additionally, the government recognizes the severity of abandoned farmland and has sought to address this issue by encouraging reclamation, improving agricultural production conditions, and promoting cropland transfers. Overall, the state has gradually established a three-in-one cropland protection framework encompassing quantity, quality, and ecology. This has been achieved through measures such as delineating permanent basic farmland, constructing high-standard basic farmland, establishing a mechanism for the compensation of cropland occupation in different areas, and strengthening the supervision of the quality of cropland occupied and compensation. The cropland protection policy in Guangdong Province has played a role in mitigating cropland loss and supporting the sustainable development of agriculture. However, challenges such as declining cropland quality and widespread abandonment persist. It is essential to adopt scientifically informed measures tailored to local conditions for effective cropland protection.

4.4. Implications for Cropland Protection

Urbanization can be categorized into reversible and irreversible processes [52]. The irreversible transformation of cropland into construction land results in the destruction of the topsoil structure and fundamental changes in the soil’s physical and chemical properties, leading to permanent GPP loss [53]. During urbanization, the government can leverage high-precision remote sensing technology to improve the monitoring of cropland loss hotspots, establishing a dynamic, comprehensive, and full-range cropland protection monitoring system. This allows for timely tracking of cropland occupation, effectively guiding urban expansion patterns, and achieving coordinated development of cropland protection and the economy. Furthermore, the issue of cropland abandonment in Guangdong Province is becoming progressively more severe, and the GPP loss caused by abandonment cannot be ignored. The government should closely address the issue of cropland abandonment. Under the guidance of the “three-in-one” new cropland protection pattern, comprehensive strategies for preventing and managing abandoned land should be implemented based on the driving factors and agricultural development trends [54]. This involves institutional reform, innovation in mechanisms, and supporting policies. Specifically, to address the primary cause of abandonment-loss of the rural labor force [55]—a well-established land transfer market needs to be developed. For abandoned land with good production conditions, priority should be given to its use through rotation and reclamation. In recent years, despite the government taking various policy measures to ensure food security, such as land development, land consolidation, and land reclamation projects to supplement cropland area, it is also important to focus on supplementing cropland based on the quantity, quality, and structure of the cropland to ensure balanced food productivity [56,57]. Our research findings indicate significant regional differences in the quality of supplemented cropland. Areas with better-supplemented cropland quality are mainly distributed in the WG, whereas the EG has poorer-quality supplemented cropland.

4.5. Limitations and Perspectives

This study also has certain limitations and uncertainties. Firstly, our cropland use data is obtained from remote sensing monitoring. Inherent scale effects and mixed pixel issues in remote sensing pixels may introduce uncertainties in identifying fragmented cropland in mountainous and hilly areas of Guangdong Province, potentially leading to an underestimation of calculated results, which is difficult to avoid completely. Secondly, for a more refined framework of grain production estimation, we did not consider the impact of non-grain on food production. In some areas, non-grain cultivation is severe, up to 50% [58]. However, non-grain is generally considered an adjustment of cropland planting structure and thus not classified as a type of cropland use transition. Finally, we did not account for yield differences caused by the conversion between paddy fields and drylands. Typically, paddy fields have higher productivity than drylands. Future research should incorporate the impact of non-grain into the food production estimation framework and refine yield differences between paddy fields and drylands.

5. Conclusions

This study utilizes continuous long-term land cover data from 1991 to 2019 to comprehensively assess the impact of various cropland use transitions on GPP, revealing their spatiotemporal patterns and regional differences. Integrating population and socio-economic data, it employs spatial mismatch analysis to explore the intrinsic relationships among population, economic development, and food production.

Results indicate that the total area of cropland in Guangdong Province has decreased overall, largely opposite to trends in the cropland use transition rate. Specifically, the total cropland area decreased by 9.85 thousand km². Non-agriculturalization and abandoned cropland accounted for annual average losses of 2.76 thousand km² and 3.09 thousand km², respectively, while the annual average new cropland area was 2.94 thousand km². From the perspective of centroid migration, the cropland centroid has shifted northwestward, whereas the population centroid has moved southeastward, indicating a diverging trend in their spatial distribution, exacerbating the decoupling phenomenon between cropland spatial distribution and population.

Furthermore, from 1991 to 2020, Guangdong Province lost 55.6 million tons of GPP while only gaining 40 million tons, resulting in a deficit of 15.5 million tons annually, which roughly translates to meeting the food needs of about 4.6 million people per year. Under China’s policy of the requisition–compensation balance of cropland, the cropland area and food production in Guangdong Province have achieved a dynamic equilibrium, but considering the impact of cropland abandonment on GPP, the newly added GPP has not fully compensated for the lost GPP.

The study also found that the overall spatial mismatch index in Guangdong Province showed a rapid upward trend, indicating an increase in the spatial mismatch between food production, population size, and the level of economic development. In conclusion, government departments should formulate targeted policies based on different land use types to ensure food security.