Understanding the response mechanism of ecosystem services (ES) to landscape patterns is of great significance for regional landscape planning and sustainable development. In this study, the landscape index and InVEST model were used to quantitatively analyze the spatio-temporal evolution of landscape patterns and ES in the Ganjiang River Basin of China from 1990 to 2020. Furthermore, the bivariate Moran's I method and spatial error model (SEM) were used to test the spatial correlation between landscape index and ES. The results showed that (1) cropland decreased and construction land increased, and the overall landscape tended to be fragmented, the patch shape tended to be complicated, and landscape diversity increased from 1990 to 2020. Water conservation (WC) and soil conservation (SC) capacity increased by 10.56 mm and 16.24 t hm–2 a–1, respectively, whereas carbon storage (CS) decreased by 1.22 t hm–2 a–1. (2) The responses of different typical ES to landscape patterns were different in landscape index and response degree. Typical ES negatively responded to Shannon’s diversity index and patch density. WC was sensitive to the Splitting Index, whereas SC and CS were more responsive to the average patch area. (3) The overall purpose of territorial spatial planning within a basin should be to reduce the fragmentation and heterogeneity of the landscape. According to four local aggregation patterns of landscape index and ES, corresponding measures can be taken according to local conditions in different regions. The results can provide a quantitative basis for landscape management and ecological construction in the Ganjiang River basin and scientific guidance for the Yangtze River conservation strategy.
Research Article
Impacts of landscape pattern evolution on typical Ecosystem Services in Ganjiang River Basin, China
https://doi.org/10.21203/rs.3.rs-2988438/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
landscape pattern
ecosystem service
InVEST model
spatial correlation
Ganjiang
- Gan River Basin landscape pattern and ES were evaluated
- Bivariate Moran's I methods and spatial regression were used for the analysis
- Impact of landscape patterns on ES were shown at the grid scale
- Landscape pattern optimization measures were proposed for four clustering methods
Ecosystem services (ES) are the benefits that human beings receive from ecosystems (Millennium Ecosystem Assessment, 2005). They form the basis of human existence and are closely associated with human welfare (Costanza et al., 1998; de Groot et al., 2012). Landscape patterns are composed of different land-use types which present different arrangements and combination forms in space (Tscharntke et al., 2012). Frequent human activities resulted in various changes in the area, shape, and spatial distribution of various landscapes that affect ecosystem structure and function. This will disturb the circulation of materials, the flow of energy, and ecological processes in the ecosystem; this ultimately affects the provision and maintenance of ES (Mitchell et al., 2015; Wang et al., 2019). The demand for resources has sharply increased with population expansion and economic development. Humans constantly allocate and utilize various resources, leading to the continuous degradation of ES (Fang et al., 2023). Therefore, it is a signal to elucidate the spatial correlation, influence, and response mechanisms of ES to landscape patterns for the construction of regional landscape ecology, landscape pattern optimization, and socioeconomic sustainable development.
The integration of landscape patterns and ES is a hot topic in landscape sustainable development and ES management (Yun et al., 2022). Most studies use the landscape pattern index to explore the effects of landscape pattern change on the ES value. A study of city clusters in the middle reaches of the Yangtze River showed that patch density, landscape shape index, and ecosystem service were significantly negatively correlated (Liu et al. 2020). High landscape connectivity and diversity positively impact ES in Wuhan (Chen et al. (2021)). Landscape patterns have a significant impact on all mapped ES and to varying degrees in different regions in the hill country of New Zealand (Tran et al. 2022)
However, these studies often carry out statistical analyses of the correlation between landscape patterns and ES based on static data. This has two problems. First, the landscape has spatial continuity and the spatial heterogeneity of variables is ignored (Chen et al., 2021; Liu et al., 2020; Shuangao et al., 2021; Tran et al., 2022). However, simply utilizing landscape indices to quantitatively describe the overall characteristics of the study area is to the disadvantage of reflecting the spatial change trend and laws of a continuous landscape (Baro et al., 2016; Zhai et al., 2018). Second, dynamic changes in the data were not considered; most previous studies have used static data from one or many years to study the relationship between them (Chicago et al., 2022; Xia et al., 2021; Yun et al., 2022). Nevertheless, dynamic change indicators can better reflect the driving effect of landscape pattern evolution on ES compared to static data.
In addition, regional research on landscape patterns and ES mostly involves certain provinces, cities, counties, and other administrative regions, whereas the geographical attributes of the study areas are often separate (Dadashpoor et al., 2019; Wang et al., 2021; Xu et al., 2022), and lacking the integrity of the geographical system. A consideration of the ecological environment and socioeconomic progress from the perspective of river basins is a more effective and systematic approach. Most current studies conducted at catchment scales have relatively small spatial scales owing to concerns regarding data integrity and computational effort (Gao et al., 2020; Zheng et al., 2014; Yushanjiang et al., 2018). However, the results are relatively weak in a study of large mesoscale watersheds with different natural environmental background characteristics owing to a large number of factors (Gao et al., 2021). Therefore, it is indispensable to consider spatial heterogeneity and use dynamic indicators to explore how ES responds to landscape pattern evolution at the mid-basin scale. A scientific understanding of the response mechanisms of ES to landscape patterns in large and mesoscale watersheds carries major implications to optimize landscape patterns and sustainably develop watershed landscape ecology.
The Ganjiang River Basin plays a critical role in conserving and recharging water sources in the middle and lower reaches of the Yangtze River (Liu et al., 2022). The Nanling, Wuyi, and Luoxiao Mountains outside the basin are important national mountain forest ecological barriers that play important roles in ecological protection. The past 30 years have given rise to the strategic revitalization and development of the Yangtze River Economic Belt and the Soviet Area, as well as the ecological protection projects such as the Great Protection and Restoration of the Yangtze River, the integrated protection of "mountains, water, forest, farmland, lake, grass, and sand,” and the conversion of farmland to forests and grasslands. These accelerated the evolution of the landscape pattern and ES in the Ganjiang River Basin.
This study quantitatively analyzed the effects of landscape pattern evolution on typical ES in the Ganjiang River Basin, China. The results are expected to benefit research on the response mechanism of ES to landscape patterns in the Ganjiang River Basin. Furthermore, this work provides a scientific basic foundation to optimize landscape patterns and continuously supply ES in other mesoscale basins in China and even the world.
2.1. Study area
The Ganjiang River is located on the southern bank of the middle and lower reaches of the Yangtze River in China. It originates from the west foot of Wuyi Mountain on the border between Jiangxi and Fujian provinces, runs from south to north in Jiangxi Province, and is approximately 766 km long. It is located between 113°42'–116°38' E and 24°30'–28°42' N (Fig. 1), covers an area of 83,500 km², with 44 counties under the jurisdiction, and with a large north-south span, and obvious regional differentiation. The Ganjiang River Basin has a humid subtropical monsoon climate and adequate rainfall with an average annual precipitation of 1400–1800 mm. Mountains and hills dominate the entire landform landscape of the basin, occupying 64.7% of the total area. Low hilly land below 200 m occupies 31.5% of the total area, and this pattern presents a stepped distribution from south to north as the altitude gradually decreases. The Ganjiang River Basin has abundant forestry resources with the forest coverage rate reaching 63.6%. The typical vegetation is subtropical evergreen broad-leaved forest. The main soil types are red, yellow, purple, and paddy soil.
China has vigorously implemented a strategy for sustainable development and formulated relevant policies and regulations to protect natural resources and restore the ecology since the 1990s. The Law of the People's Republic of China on Conservation of Water and Soil was promulgated and implemented in 1991. The National Program for Ecological and Environmental Protection was issued in 2000, and has made steady ecological progress. Various ecological protection projects were launched in the Ganjiang River basin since 1990. For example, there was the gradual implementation of protected forest system projects in the middle and upper reaches of the Yangtze River, desertification control and plain afforestation, and key projects of water and soil conservation. At the same time, industrial structure adjustment in the 1990s increased land development activities in the Ganjiang River Basin with the goal of high-speed economic growth. Furthermore, the gradual acceleration of urbanization resulted in a significant change in landscape patterns. Therefore, the analysis of ES changes in Ganjiang River Basin from 1990 to 2020 and their responses to landscape patterns have good representativeness and important research value.
2.2. Data sources and data preprocessing
The land use data for 1990, 2000, 2010, and 2020 were derived from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences (http://www.resdc.cn/) with a spatial resolution of 30 m. The digital elevation model (DEM) data were obtained from Global Digital Elevation Model (GDEM) products of the Geospatial Data Cloud (http://www.gscioud.cn/) with a spatial resolution of 30 m. The meteorological data (including average daily rainfall, temperature, evaporation, and solar radiation) from 1990 to 2020 were sourced from the China Meteorological Sciences Data Sharing Service (http://cdc.cma.gov.cn) and the Jiangxi Meteorological Bureau. The station data were converted into the meteorological raster data of 30 m × 30 m using the kriging interpolation method. The 30-year average value of each meteorological index was used to represent the climatic background of the basin. This value eliminated the influence of fluctuations in meteorological factors on the assessment of ES in individual years and avoided interference with the judgment of the relevance of landscape patterns and ES. The soil data (including soil depth, root depth, and soil texture) were derived from the Chinese Soil Dataset of the Harmonized World Soil Database (HWSD).
3.1. Landscape pattern changes
3.1.1. Trend and status index of landscape types
The trend and status index of landscape types (PS) was used to represent the changing trend and state of each landscape type using the following formula (Du et al., 2023):
\({\text{P}}_{\text{s}}\text{ }\text{=}\text{ }\text{(}\text{∆}{\text{U}}_{\text{in}}\text{ }\text{}\text{ }\text{∆}{\text{U}}_{\text{out}}\text{)}/\text{(}\text{∆}{\text{U}}_{\text{in}}\text{ }\text{+}\text{ }\text{∆}{\text{U}}_{\text{out}}\) ) (1)
where ΔUin is the total area of other landscape types converted to this type in a certain period (hm2), ΔUout is the total area transferred to other landscape types in a certain period of time (hm2). The value range of this index was − 1 to 1. A PS value > 0 indicates that the landscape type area increased; otherwise, it decreased. A PS value that is close to 1 indicates that other landscape types are mainly transferred to this landscape type. A greater proportion of landscape types are transferred to other landscape types as the Ps value approaches − 1.
3.1.2. Landscape index
A series of indices were selected from the four aspects of landscape fragmentation, landscape shape and configuration, landscape agglomeration, and landscape diversity based on previous studies (Dadashpoor et al., 2019; Medeiros et al., 2022) to characterize the landscape pattern characteristics of Ganjiang River Basin. However, not all landscape indices provide useful information to describe landscape configuration patterns owing to the multicollinearity between these indicators and the instability across scales. Therefore, this study retained relatively independent and important indicators for landscape pattern analysis through a multicollinearity test, and finally selected a landscape index with a variance inflation factor (VIF) below 6 (Liu et al., 2020). Specifically, patch density (PD), mean patch area (AREA_MN), mean patch shape index (SHAPE_MN), interspersion and juxtaposition index (IJI), landscape separation index (SPLIT) and Shannon's diversity index (SHDI) were analyzed (see Supplementary Information Table S1 for details).
The PD and AREA_MN were used to reflect the landscape fragmentation. A higher value of PD or a lower value of AREA_MN, indicates that the landscape is more fragmented. SHAPE_MN reflects the landscape shape and configuration and a higher value corresponds to a more complex patch shape. Higher IJI and SPLIT values correspond to more equal adjacency probability between patch types, and a more discrete landscape. SHDI reflects landscape heterogeneity and a higher value shows an increase or balanced patch type distribution in the landscape. Fragstats 4.2 was used to calculate the landscape index.
3.2. The assessment of ES
Ganjiang River Basin has abundant forests and surface water resources. The upper reaches of the Ganjiang River are important ecological barrier areas. In addition, vegetation and soil are the main carbon sequestration units of terrestrial ecosystems that play important roles in mitigating climate change (Phillips et al., 2023). Accurate evaluation of the main ES such as water conservation (WC), soil conservation (SC), and carbon storage (CS) in the Ganjiang River Basin is of important meaning for realizing the harmonious development of humans and land in the area, and ensuring the ecological safety of the basin. Therefore, WC, SC, and CS were selected as the typical ES. The land-use data required for ES assessment were on the basis of the secondary classification data of the Chinese Academy of Sciences (see Supplementary Information Fig. S1 for details).
3.2.1. Water conservation
The formula for WC is as follows (Xue et al., 2022):
$$\text{WR}\text{ }\text{=}\text{ }\text{min(1,249/}\text{v}\text{)}\text{ }\text{× min(1,0.9}\text{TI}\text{/3)}\text{ }\text{×}\text{ }\text{(1,}{\text{K}}_{\text{s}}\text{/300)}\text{ }\text{×}\text{ }\text{Y}$$
2
where WR is WC (mm) and V is the velocity coefficient according to previous literature (Li et al., 2021). Ks is the soil saturated water conductivity (cm d–1), and Y is the water yield (mm).
Water yield was calculated using the InVEST model and the following formula (Li et al., 2021):
$$\text{Y}\text{ }\text{=}\text{ }\text{P}\text{ }\text{}\text{ }\text{AET}$$
3
where P is the average annual rainfall (mm), and AET is the average annual evapotranspiration (mm) (see Supplementary Information Fig. S2 for details).
The following topographic index (TI) formula was used (Li et al., 2021):
$$\text{TI}\text{ = lg(}\frac{{\text{D}}_{\text{a}}}{{\text{S}}_{\text{d}}\text{ }\text{×}\text{ }\text{S}{\text{l}}_{\text{p}}}\text{)}$$
4
where Da is the number of grids in the catchment area (dimensionless), Sd is the soil depth (mm), and Slp is the slope percentage (%).
3.2.2. Soil conservation
SC was calculated using the sediment delivery ratio (SDR) module in the InVEST model. SC is obtained by subtracting the actual soil erosion from the potential. The following formula was used (Song et al., 2023):
$$\text{SEDR}\text{ }\text{=}\text{ }\text{RKLS}\text{ }\text{}\text{ }\text{USLE}$$
5
where SEDR is soil retention (t hm–2 a–1), RKLS is potential soil erosion (t hm–2 a–1), and USLE is actual soil erosion (t hm–2 a–1).
Soil erosion was evaluated using the revised universal soil loss equation (Song et al., 2023):
$$\text{RKLS}\text{ }\text{=}\text{ }\text{R}\text{ }\text{×}\text{ }\text{K}\text{ }\text{×}\text{ }\text{LS}$$
6
$$\text{USLE}\text{ }\text{=}\text{ }\text{R}\text{ }\text{×}\text{ }\text{K}\text{ }\text{×}\text{ }\text{LS}\text{ }\text{×}\text{ }\text{C}\text{ }\text{×}\text{ }\text{P}$$
7
where R is the rainfall erosivity factor (MJ mm hm–2 h–1 a–1) calculated using the correlation formula proposed by Wischmeier (Bai et al., 2022), K is the soil erodibility factor (t h MJ–1 mm–1) obtained using the EPIC method (Bai et al., 2022), L is the slope length factor (dimensionless), S is the slope factor (dimensionless), C is the vegetation cover factor (dimensionless) obtained by referring to relevant local studies, and P is the soil and WC measure factor (dimensionless). The corresponding values are assigned based on local land use and agricultural activities (see Supplementary Information Table S2 and Fig. S3 for details).
3.2.3. Carbon storage
CS was calculated using the InVEST model as follows (Gong et al., 2023):
$${\text{C}}_{\text{total}}\text{ }\text{=}\text{ }{{\text{C}}_{\text{above}\text{ }}\text{+}\text{ }\text{C}}_{\text{below}} {\text{+}\text{ }\text{C}}_{\text{dead}\text{ }}\text{+}\text{ }{\text{C}}_{\text{soil}}$$
8
where Ctotal is the total CS (t·ha–1), Cabove is the aboveground CS (t·ha–1), Cbelow is the underground CS (t·ha–1), Cdead is CS of dead organic matter (t·ha–1), and Csoil is soil CS (t·ha–1). This study analyzed soil CS data in Jiangxi Province and its adjacent regions using information from the relevant literature. Other carbon pool data were determined by modifying the relationship between carbon pools in the InVEST model user manual and related literature (Table 1).
| lucode | LUCC_name | C_above | C_below | C_soil | C_dead | References |
|---|---|---|---|---|---|---|
| 11 | Paddy field | 18.45 | 11.67 | 82.28 | 2.29 | Chuai et al., 2013; Fu et al., 2010; Tang et al., 2020 |
| 12 | Dryland | 18.45 | 11.67 | 82.28 | 2.29 | |
| 21 | Forest land | 35.86 | 7.37 | 148.87 | 3.24 | Chen et al., 2019; Diao et al., 2022; Wu et al., 2012 |
| 22 | Shrubwood | 28.51 | 6.14 | 109.69 | 2.89 | |
| 23 | Dredging land | 23.36 | 4.84 | 95.21 | 1.37 | |
| 24 | Other forest land | 31.05 | 6.88 | 132.32 | 2.21 | |
| 31 | High cover grassland | 16.49 | 21.06 | 112.78 | 8.46 | Balasubramanian et al., 2020; Bartholomée et al., 2018 |
| 32 | Medium cover grassland | 14.26 | 17.09 | 86.12 | 7.46 | |
| 33 | Low cover grassland | 11.57 | 14.98 | 69.78 | 5.17 | |
| 41 | Canals | 2.12 | 0.00 | 24.56 | 0.00 | Tang et al., 2020a; Xi et al., 2013 |
| 42 | Lake | 2.12 | 0.00 | 24.56 | 0.00 | |
| 43 | Reservoir pit | 2.12 | 0.00 | 24.56 | 0.00 | |
| 46 | Beach land | 2.12 | 0.00 | 24.56 | 0.00 | |
| 51 | Urban land | 7.61 | 1.52 | 34.33 | 0.00 | Chuai et al., 2013; Ke and Tang, 2019 |
| 52 | Rural residential area | 7.61 | 1.52 | 34.33 | 0.00 | |
| 53 | Other construction land | 7.61 | 1.52 | 34.33 | 0.00 | |
| 64 | Marshland | 16.49 | 27.87 | 162.68 | 6.09 | Tang et al., 2020b |
| 65 | Bare land | 7.61 | 1.52 | 34.33 | 0.00 | Chuai et al., 2013; Ke and Tang, 2019 |
| 66 | Bare rock | 7.61 | 1.52 | 34.33 | 0.00 |
3.3. Impacts of landscape patterns on ES
3.3.1. Spatial correlation test
A bivariate spatial analysis model was adopted and global Moran's I was used to explore the spatial correlation between the landscape index and ES using the following formula (Degefu et al., 2021):
where I indicates global Moran's I (dimensionless), n is the number of space units (dimensionless), wij is the spatial weight matrix, which is generated by the first-order adjacency of the queen (dimensionless), zxi is the standardized value of each ecosystem service function in the i-th space (dimensionless), and zyj is the standardized value of each landscape index in the j-th space (dimensionless).
Local indicators of spatial association (LISA) cluster maps were adopted for local spatial autocorrelation analysis to reflect local agglomeration and dispersion effects in the Ganjiang River Basin. LISA clustering indicates the spatial correlation between ES of a certain spatial unit and landscape indices of adjacent units at a certain level of significance. There are four types of clustering patterns: high-high, high-low, low-high and low-low. A 5 km × 5 km grid was created to be the minimum spatial unit based on data availability and computational efficiency considerations (Guo et al., 2022; Zhang et al., 2018; Zheng et al., 2023). The spatial correlation analysis was performed on GeoDa 1.12 software.
3.3.2. Spatial regression analysis
The degree of response of typical ES to landscape indices was analyzed using the spatial regression model, including the spatial lag model (SLM) and spatial error model (SEM). A more suitable model is selected by comparing the values of Lagrange multipliers (LM) and robust Lagrange multipliers (R-LM) (Liu et al., 2020).
The SLM assumes spatial autocorrelation, and exists in the dependent variables. The dependent variables show a neighborhood effect and the spatial diffusion among geographical units (the spillover effect) is considered using the following formula (Liu et al., 2020):
$$\text{y}\text{ }\text{=}\text{ }\text{β}\text{x}\text{ }\text{+}\text{ }\text{ρ}{\text{W}}_{\text{1}}\text{y}\text{ }\text{+}\text{ε}$$
10
where y is the vector of the dependent variable, x is the argument matrix, β is the independent variable coefficient reflecting the impact of landscape index on ES, W1 is the weight matrix of the dependent variable space, ρ is the spatial correlation coefficient, and ε is the residual vector.
A spatial correlation exists in the interference error term of the SEM according to the following formula (Liu et al., 2020):
$$\text{y}\text{ }\text{=}\text{ }\text{β}\text{x}\text{ }\text{+}\text{ }\text{λ}{\text{W}}_{\text{2}}\text{ε}\text{ }\text{+}\text{ }\text{u}$$
11
where the meanings of y, x, ε, and β are the same as those of Eq. (10), W2 is the residual space weight matrix, λ is the residual correlation coefficient, and u is the random error term.
4.1. Landscape pattern changes
4.1.1. Dynamic changes of different landscape type areas
The trend of landscape area changes (Fig. 2) shows that the PS of cropland and grassland were −0.28 and −0.5, respectively from 1990 to 2000. This indicated that these landscape types were mainly transferred out, wherein farmland was abandoned due to a large amount of outgoing labor, and a large area was deserted, resulting in a decline in cropland area. The PS of construction land reached 0.97. This indicates that other land types were mainly transformed into construction land. The PS of the woodland and water areas were above 0. This indicated that they were mainly transferred in. The PS of croplands tended toward zero between 2000 to 2010. This indicated that there was a bidirectional equilibrium transition. The PS of woodlands and grasslands was below zero and transfer-out was the mainstay. In contrast, the PS of construction land reached 0.84. This indicated that it was mainly transferred in, owing to the use of forest land and grassland for urban construction. The continuous expansion of construction land from 2010 to 2020 led to a sustained reduction in forest land area, with the PS of construction land and forest land reaching 0.84 and − 0.28, respectively. The grassland area began to recover with the PS of grassland reaching 0.31. Grassland ecological projects such as the construction of standard artificial grasslands and the second round of returning farmland to grasslands facilitated. The PS of cropland was −0.12 indicating that there was a small transfer-out. The area of water fluctuates and continuously increases. This is closely related to the implementation of water conservancy facilities construction projects in the Ganjiang river basin and the restoration of water resources such as "returning farmland to lakes.” The PS of cropland, forest land, grassland, and water area in the Ganjiang River Basin was −0.11, −0.07, 0.01, and 0.17, respectively in the past 30 years. This tended to zero with two-way equilibrium conversion and small overall fluctuation; however, it showed some volatility with obvious expansion of construction land approaching the sub-extreme unbalanced development trend, and mainly involved transfer-in.
4.1.2. Change of landscape index
The landscape index of the Ganjiang River Basin from 1990 to 2020 showed that PD and SPLIT first decreased and reached their minimum values in 2000, then increased (Fig. 3). There was an overall increasing trend. This indicated that the landscape in the basin gradually tends to be fragmented and dispersed. AREA_MN first increased, peaked in 2000, then decreased from 177.18 to 174.12. The decrease in average patch area indicated that the landscape composition changed from relatively few large patches to staggered fine patches and reflected the gradual increase of landscape fragmentation. The gradual increase in SHAPE_MN manifested that the landscape was shaped toward complexity. The IJI rose from 48.62 in 1990 to 50.91 in 2020. This indicated an increased degree of restriction between patches and that the distribution between patches tended to be uniform. The SHDI slightly fell from 2.57 in 1990 to 2.56 in 2000 and gradually rose after 2000. This indicated that the dominant landscape types weakened, and landscape diversity gradually increased after 2000.
Landscape fragmentation in most areas of the Ganjiang river basin increased as the PD area increased by 76.78% and AREA_MN decreased by 69.45% (Fig. 4). The areas increased markedly in landscape fragmentation were mainly located in Gaoan, Fengcheng, Yushui, Fenyi, Jizhou, and Qingyuan in the middle of the basin, and Zhanggong and Nankang in the south of the basin. The global SHAPE_MN change distribution was relatively uniform. The variability of landscape shape in Zhanggong in the northeast and south of the basin significantly decreased and the area became more regular. SHAPE_MN increased significantly in Anfu, Yongxin, Jinggangshan, and Dayu in the west. The frequent transfer of arable land, woodland, and grassland landscapes caused the landscape shape of these areas more complex. The increased areas of the SPLIT, IJI, and SHDI accounted for 72.09%, 78.64%, and 79.25%, respectively. This indicated that the overall landscape aggregation in the Ganjiang River Basin decreased, the connectivity of the same landscape type gradually decreased, and landscape heterogeneity increased. IJI, SPLIT, and SHDI showed a certain similarity in spatial variation, and in the central urban regions such as Shanggao and Yuanzhou in the northern, Jizhou in the central, and Nankang and Zhanggong in the southern parts of the basin, IJI, SPLIT, and SHDI all increased significantly. Human activities cause homogeneous patches to change from the original aggregated state to a dispersed state. Furthermore, the degree of aggregation decreased, and the distributions of all landscape types were uneven.
4.2. Temporal and spatial changes of ES
WC in the Ganjiang River Basin initially increased in 1990, peaked in 2000, and then declined until 2020; however, the overall WC slightly increased from 430.79 mm in 1990 to 441.35 mm in 2020 (Fig. 5). SC increased from 151.29 t hm–2 a–1 in 1990 to 167.53 t hm–2 a–1 in 2020. CS initially increased, peaked in 2000, then decreased; there was an overall decreasing trend from 156.11 t hm–2 a–1 in 1990 to 154.89 t hm–2 a–1 in 2020. The rapid increase in ES since 2000 was closely related to the expansion of forest land since 1990. The implementation of the first phase of the ten-year national key treatment project rapidly reduced soil erosion in the southern red soil area and greatly improved WC and SC. WC and SC remained relatively stable between 2000 to 2010, while CS slightly decreased owing to the growth of construction land. Continued urban sprawl driven by more frequent human economic activity resulted in a prominent reduction in WC and CS in the region between 2010 and 2020.
The overall spatial distribution of each ES was higher in the south than in the north (Fig. 6). This is mainly owing to the mountainous and hilly landform in the south of Ganjiang River Basin, with interleaved forest, grassland, and excellent ecological resources. The terrain in the north of the basin is flat with widespread cultivated land and high human activity leading to low ecosystem service ability in this area. Spatial change analysis showed that 53.11% of the areas had increased WC with significant increases in Xingguo, Ningdu, Longnan, and Yongxin. The area of SC increased by 69.23% with significant increases mainly observed in Longnan, Yudu, Xingguo, Ningdu, and Luxi. This is due to a series of ecological protection policies executed in the region (Zhu et al., 2021). Regions with considerably reduced carbon reserves are mainly located in urban areas, including Nankang, Zhanggong, Jizhou, and Yuanzhou. The continuous improvement of urbanization had impacted the structure and process of ES leading to a sharp decrease in carbon reserves within the region.
4.3. Responses of ES to landscape patterns
4.3.1 Spatial correlation analysis
The correlation analysis with landscape pattern index and ES in Ganjiang River Basin from 1990 to 2020 was used by GeoDa software. It revealed a conspicuous spatial correlation in the landscape pattern index and ES using global bivariate Moran's I (Fig. 7). Moran's I values of the CS, SC, and WC were below 0 for SHDI, PD, and SPLIT and greater than 0 for AREA_MN, and all passed the significance test. These results indicated that these three ES showed a negative correlation with SHDI, PD, and SPLIT, and a positive correlation with AREA_MN. Therefore, more heterogeneous patches and fragmented landscapes are detrimental to the maintenance of the functions of ES. The correlation between SHAPE_MN and Moran's I for WC, SC, and CS was above 0, below 0, and did not pass the significance test, respectively. Therefore, patch shape complexity did not affect CS function. A complex patch shape tends to have a positive effect on WC but negatively affects SC. The IJI showed a negative correlation with WC and CS, and an insignificant correlation with SC.
A local bivariable LISA cluster map was used to represent the four spatial clusters modes between the landscape index and ES (Fig. 8). The SHDI, SPLIT, and PD landscape indices were spatially similar to the LISA cluster map of WC. The low-high cluster was mainly located on the east and west of the middle basin, and the high-low cluster was mainly located in the central area of Ji 'an and Ganzhou. The significant LISA agglomeration areas of landscape index and SC were mainly located in the middle and northern basins. The LISA agglomeration of SHDI, PD, and SC was mainly high-low agglomeration, while the AREA_MN and SC were mainly low-low agglomerations. This region showed an abundant landscape type, a high heterogeneity and fragmentation, and a low soil retention capacity. The distribution of the SHDI, PD, AREA_MN, and CS LISA agglomeration maps in the middle and northern basin showed some similarities with SC; however, they were very different in the southern part of the basin. The LISA agglomeration of SHDI, PD, and CS was mainly low-high agglomeration in the southern periphery of Ganjing River Basin, whereas the AREA_MN and CS were mainly high-low agglomerations, and the relationship with SC was not evident.
However, the correlation between landscape indices and ES in local areas such as Zhangshu, Taihe, and Nankang showed an agglomeration effect contrary to the overall trend. This may be on account of the high level of economic activity and intensive use of construction land in this region, relatively concentrated urbanization, and the building land landscape matrix. Therefore, the ecosystem service functions remain low, even if the areas were in high aggregation. The non-significant areas represent the areas that have no obvious aggregation in the landscape index and ES. This part of the area accounts for between 52.09–54.83% of the study region, and is mainly distributed in the southern basin. These regions are mainly forestlands with a low human disturbance and a stable landscape pattern resulting in a weak influence on ES.
4.3.2 Spatial regression analysis
The ordinary least squares (OLS) regression was performed on landscape indices and ES showed a significant spatial autocorrelation in the residual results (P < 0.001). This indicated that the spatial regression model such as SLM and SEM were more suitable than OLS. Spatial regression results showed that the LM (error) and R-LM (error) of each ES were larger than those of LM (lag) and R-LM (lag). Therefore, the spatial regression of each ES was analyzed by SEM (see Supplementary Information Table S3 for details). All ES show negative correlations with SHDI and PD and positive correlations with AREA_MN (Table 2). This suggested that ES were negatively affected by landscape diversity and fragmentation. This is consistent with the global correlation analysis results. The intensification of landscape fragmentation has led to the worse of ES. Specifically, PD increased by 1.00%, while WC, SC, and CS decreased by 0.16%, 0.16%, and 0.11%, respectively. A 1% reduction in AREA_MN led to a reduction in WC, SC, and CS by 0.07%, 0.09%, and 0.08% respectively.
| Explanatory variables | WC | SC | CS |
|---|---|---|---|
| PD | -0.16** | -0.16** | -0.11** |
| SHDI | -0.36** | -0.18** | -0.52** |
| AREA_MN | 0.07* | 0.13** | 0.08* |
| IJI | 0.06** | 0.09** | 0.082** |
| SPLIT | -0.10** | -0.02 | -0.06* |
| SHAPE_MN | 0.10** | -0.01 | 0.10 |
| R² | 0.75 | 0.79 | 0.73 |
| Breusch-Pagan test | 63.20** | 174.85** | 35.15** |
| Note: * represents significance at P < 0.01; ** represents high significance at P < 0.001. | |||
The IJI showed a significant positive correlation with all ES, contrary to the global correlation characteristics determined by correlation analysis. This might be caused by the hypothesis of the SEM which supposes that spatial correlation only preserves between the random errors of ES and its neighboring units. This may lead to a large difference between the regression coefficient and global correlation (Degefu et al., 2021). SPLIT showed a negative correlation with WC. Specifically, a rise in SPLIT by 1.00% resulted in a descent in WC by 0.10%. In addition, the Breusch-Pagan test passed the significance test at P < 0.001, which is indicative of heteroscedasticity. Therefore, factors other than landscape indices affect ES and more crucial variables may need to be introduced into the model.
5.1. Scale effects of landscape patterns on ES
Landscape patterns and heterogeneity vary with changes in the study scale. Therefore, there is scale heterogeneity in the impact of landscape patterns on ES. Analyzing the correlation between landscape patterns and ES at a single scale within a region may miss the real interaction between the two, and the results may vary during changes to the spatial scale of the study (Yang et al., 2022). Scholars often divide the study area into grids of different scales based on spatial size. For example, the study analyzing the impact of urban expansion on ES in Wuhan was based on a 3 km × 3 km grid (Zhang et al. 2018). A 10 km × 10 km grid was created to analyze the response of ES to human activities and landscape patterns in the upper and middle reaches of the Yellow River urban agglomeration (Guo et al. 2022). The study areas to compare the relationship between landscape fragmentation and habitat quality in the Yangtze River Economic Belt and the Yellow River Basin were divided into a 3 km × 3 km grid (Zheng et al. 2023). A relatively high grid resolution was adopted in a large spatial area to ensure research accuracy. This study used a 5 km × 5 km spatial unit on the grid scale to explore the spatial correlation between the landscape index and ES; this is in line with previous work.
5.2 Effects of landscape patterns on ES
A study on the response of integrated ES to landscape patterns showed that the composition and spatial configuration of landscape affect ES. Forests and grasslands are the dominant landscapes for ES. The increase in forest and grassland areas is beneficial to improve the CS of biomass and wood production and moderately promoting CS in soil (Winkel et al., 2022). It is conducive to intercepting surface runoff and improving WC (Wu, 2021). The Yangtze River shelterbelt and plain afforestation projects were comprehensively carried out between 1990 to 2000. This greatly increased the forest area and improved ES in Ganjiang River Basin.
The impact of each landscape index on ES was significantly different and the extent of the effect of the same landscape index on different ES was also varied. All ES were negatively correlated with the SHDI and PD, indicating that all ES were mainly negatively affected by landscape diversity and fragmentation, which is in keeping with previous studies (Chen et al., 2021; Guo et al., 2022; Liu et al., 2020; Zhou et al., 2021). The increase of forest land area directly led to the improvement of ES from 1990 to 2000, meanwhile, the landscape configuration also played a significant influence. Landscape fragmentation significantly decreased during this period, patches were closely connected, and the service capacity of each ecosystem improved. This indicated that maintaining and strengthening the continuity of the overall landscape in the region was more conducive to landscape ecological functions. Fragmented landscapes often obstruct the energy flow between regions and change the processes of material and nutrient cycling, thus affecting the maintenance of ES (Mitchell and Devisscher, 2022). The landscape in the Ganjiang River Basin has shown a trend of fragmentation since 2000, and landscape heterogeneity has significantly increased. The regional landscape has shown a state of continuous fragmentation with the continuous growth of urban areas in various counties (districts) and the construction of supporting transportation networks. This significantly negatively impacted ES. This also accounted for the dramatic decline of WC and CS between 2010 and 2020. The layout of construction land should be reasonably optimized during urbanization to avoid scattered and disorderly expansion of construction land. Appropriate land-use integration projects such as further leveling and integrating cultivated land through cultivated land transfer are crucial to reduce landscape fragmentation and improve ecosystem service capacity.
5.3 Tradeoffs and synergies between ES
WC, SC, and CS are the three typical ES in the Ganjiang River Basin. The relationships among the various ecosystem service types are complicated owing to the coupling effect of various factors. It is difficult to accurately analyze and compare the relationship between them which results in great limitations in the research of the tradeoffs and synergies among ES (Khosravi Mashizi and Sharafatmandrad, 2021). Currently, research on the tradeoffs and integrated application of multiple ES remains a challenge in the field of ES research (Xue et al., 2023). We studied the responses of WC, SC, and CS to landscape index respectively while avoiding the influence of complex coupling mechanisms between ES to clarify how landscape patterns affect ES in Ganjiang River Basin. Exploring the response of a single ES to landscape pattern and clarifying the extent to which each landscape index affects a single ES result in a clearer description of a single service such as WC, SC, and CS to landscape patterns. This leads to an optimization of the landscape patterns that can be used in practice.
5.4. Research suggestions
Ganjiang River Basin plays an important role in ecological maintenance in the middle and lower reaches of the Yangtze River which is characterized by accelerated economic development, landscape fragmentation, complex patch shapes, and increased landscape diversity. The results showed that increases in landscape fragmentation and diversity significantly negatively affected ES. References for spatial planning of the watershed can be provided according to the bivariate LISA agglomeration map of the landscape index and ES. For example, the LISA cluster map of the landscape index and SC can be divided into four types: "low diversity and fragmentation, low SC", "low diversity and fragmentation, high SC", "high diversity and fragmentation, high SC", and "high diversity and fragmentation, low SC". Measures can be taken following local conditions for four different types of regions:
(1) Areas in "low diversity and fragmentation, high SC" such as Chongyi, Shangyou, and Jinggangshan. These areas are advantageous for ES and the intensity of human disturbance is low. Moreover, they often overlap with the "ecological red line areas" in space. Therefore, strict ecological protection measures (such as establishing natural ecological reserves) should be implemented to maintain the ecological advantages of this region.
(2) Areas in "high diversity and fragmentation, high SC" such as the southwest of Shangyou and the northwest of Suichuan. The expansion of construction land should be appropriately limited, and the protection of forests and grasslands should be strengthened to avoid further fragmentation of the landscape. In addition, the lagging effect of landscape fragmentation on ES should not be ignored to avoid a significant reduction in ES and difficulties of their recovery owing to excessive development.
(3) Areas in "high diversity and fragmentation, low SC" such as the Jitai Plain in the middle of the Ganjiang River Basin and the alluvial plain in the north. The local SC can be improved by creating more parks and green spaces while maintaining the original natural patches in the built area.
(4) Regions like Ji'an and Ganzhou urban areas, which means "low diversity and fragmentation, low SC", may benefit from additional ecological restoration projects or urbanization that meet their development needs. However, it's crucial to avoid blind expansion, prioritize intensive use of construction land, and prevent landscape fragmentation.
The spatiotemporal evolution of landscape patterns and ecosystem functions in the Ganjiang River Basin from 1990 to 2020 were quantitatively evaluated using the landscape index method and the InVEST model. Bivariate spatial autocorrelation and SEM were used to analyze the effects of landscape indices on WC, SC, and CS. The three main conclusions are stated below.
(1) Cultivated land in the Ganjiang River Basin decreased and construction land increased over a large area between 1990 to 2020. The overall landscape tended to be fragmented, the patch shape tended to be complicated, landscape diversity increased, and the fragmentation of urban areas was more severe. WC and SC increased by 10.56 mm and 16.24 t hm–2 a–1, respectively, whereas CS decreased by 1.22 t hm–2 a–1. In the Ganjiang River Basin, 53.11% of the regional WC and 69.23% of the regional SC showed an increasing trend, whereas CS generally decreased. The regions with significant increases in WC and SC were mainly located at the boundary zone of Longnan, Ningdu, and Xingguo, and the boundary zone of Lianhua and Yongxin. CS most obviously decreased in urban areas.
(2) ES are negatively affected by landscape fragmentation, and the intensification of landscape fragmentation leads to the degradation of ES. However, different ES have different responses to landscape patterns that are reflected in landscape indicators and the degree of response. SHDI and PD are the main indices affecting WC, SC, and CS in Ganjiang River Basin. WC also exhibited a strong response to SPLIT, whereas SC and CS exhibited strong responses to AREA_MN. Local indicators of spatial association maps of different landscape indices and ES have their characteristics, but retain certain spatial similarities. There is high fragmentation and low ES in the northern and central plain regions of the basin. Meanwhile, the western parts of the basin exhibited low fragmentation and high ES.
(3) Landscape fragmentation should be reduced to the maximum possible extent during territorial spatial planning within the basin. The bivariate LISA cluster map identified four clustering models of landscape patterns and ES that could be adapted to the local conditions in different regions in the Ganjiang River Basin. For example, strict ecological protection measures should be implemented in "low diversity and fragmentation, high ES" areas. Conversion of other lands into construction land in the "high diversity and fragmentation, high ES" areas should be avoided. More ecological space can be created in the "high diversity and fragmentation, low ES" areas. Furthermore, additional ecological restoration projects should be considered and disorderly expansion of construction land should be prevented in "low diversity and fragmentation, low ES" areas.
AREA_MN, mean patch area; CS, carbon storage; DEM, digital elevation model; ES, ecosystem services; IJI, interspersion and juxtaposition index; LISA, local indicators of spatial association; LM, Lagrange multipliers; PD, patch density; Ps, trend and status index of land use types; R-LM, robust Lagrange multipliers; SHAPE_MN, mean patch shape index; CS, soil conservation; SDR, sediment delivery ratio; SEM, spatial error model SHDI, Shannon's diversity index; SLM, spatial lag model; SPLIT, landscape separation index; TI, topographic index; VIF, variance inflation factor; WC, water conservation.
CRediT authorship contribution statement
Wei Wan,Yun Huang, and Bofu Zheng conceived the research. Shiwen Wu, Xu Song, and Wei Wan wrote the manuscript. Shiwen Wu, Yun Huang, Wei Wan, Bofu Zheng, and Hanqing Wu edited the manuscript. Zhong Liu and Jinqi Zhu contributed to the data collection. All authors reviewed the manuscript.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was supported by the Key Research and Development Program of Jiangxi Province (Grant No. 20223BBG74S01; 20223BBG71013).
- Millennium Ecosystem Assessment: Ecosystems and human well-being: General Synthesis. World Resources Institute, Washington, DC.
- Bai J, Zhou Z, Li J, et al. Predicting soil conservation service in the Jinghe River Basin under climate change. J. Hydrol. 2022, 615: 128646. https://doi.org/10.1016/j.jhydrol.2022.128646.
- Balasubramanian D, Zhou W-J, Ji H-L, et al. Environmental and management controls of soil carbon storage in grasslands of southwestern China. J. Environ. Manage. 2020, 254: 109810. https://doi.org/10.1016/j.jenvman.2019.109810.
- Baro F, Palomo I, Zulian G, et al. Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land Use Pol. 2016, 57: 405-417. https://doi.org/10.1016/j.landusepol.2016.06.006.
- Bartholomée O, Grigulis K, Colace M-P, et al. Methodological uncertainties in estimating carbon storage in temperate forests and grasslands. Ecol. Indic. 2018, 95: 331-342. https://doi.org/10.1016/j.ecolind.2018.07.054.
- Chicago L Q, Echeverría C, Pizarro C J. Ecosystem services trade-offs in landscapes: trends, areas of greatest impact, and temporal evolution of the scientific field. Landsc. Ecol. 2022, 37: 2225-2239. https://doi.org/10.1007/s10980-022-01488-8.
- Chuai X, Huang X, Lai L, et al. Land use structure optimization based on carbon storage in several regional terrestrial ecosystems across China. Environ. Sci. Policy. 2013, 25: 50-61. https://doi.org/10.1016/j.envsci.2012.05.005.
- Chen L-C, Guan X, Li H-M, et al. Spatiotemporal patterns of carbon storage in forest ecosystems in Hunan Province, China. For. Ecol. Manage. 2019, 432: 656-666. https://doi.org/10.1016/j.foreco.2018.09.059.
- Chen W, Zeng J, Chu Y, et al. Impacts of Landscape Patterns on ES Value: A Multiscale Buffer Gradient Analysis Approach. Remote Sens. 2021, 13(13). https://doi.org/10.3390/rs13132551.
- Costanza R, d'Arge R, de Groot R, et al. The value of the world's ecosystem services and natural capital. Ecol. Econ. 1998, 25(1): 3-15. https://doi.org/10.1016/S0921-8009(98)00020-2.
- Dadashpoor H, Azizi P, Moghadasi M. Land use change, urbanization, and change in landscape pattern in a metropolitan area. Sci. Total Environ. 2019, 655: 707-719. https://doi.org/10.1016/j.scitotenv.2018.11.267.
- de Groot R, Brander L, van der Ploeg S, et al. Global estimates of the value of ecosystems and their services in monetary units. Ecosyst. Serv. 2012, 1(1): 50-61. https://doi.org/10.1016/j.ecoser.2012.07.005.
- Degefu MA, Argaw M, Feyisa GL, et al. Dynamics of urban landscape nexus spatial dependence of ES in rapid agglomerate cities of Ethiopia. Sci. Total Environ. 2021, 798: 149192. https://doi.org/10.1016/j.scitotenv.2021.149192.
- Diao J, Liu J, Zhu Z, et al. Active forest management accelerates carbon storage in plantation forests in Lishui, southern China. For. Ecosyst. 2022, 9: 100004. https://doi.org/10.1016/j.fecs.2022.100004.
- Du L, Dong C, Kang X, et al. Spatiotemporal evolution of land cover changes and landscape ecological risk assessment in the Yellow River Basin, 2015–2020. J. Environ. Manage. 2023, 332: 117149. https://doi.org/10.1016/j.jenvman.2022.117149.
- Fang G, Sun X, Liao C, et al. How do ecosystem services evolve across urban-rural transitional landscapes of Beijing-Tianjin-Hebei region in China: patterns, trade-offs, and drivers. Landsc. Ecol. 2023, 38(4): 1125-1145. https://doi.org/10.1007/s10980-022-01582-x.
- Fu Q, Zhao X, Yuan F, et al. Analysis of Organic Carbon Reserves in Cultivated Land Topsoil of Jiangxi Province. Chinese Journal of Soil Science. 2010, 41(4): 835-838. https://doi.org/10.19336/j.cnki.trtb.2010.04.015.
- Gao J, Du F, Zuo L, et al. Integrating ecosystem services and rocky desertification into identification of karst ecological security pattern. Landsc. Ecol. 2020, 36: 2113-2133. https://doi.org/10.1007/s10980-020-01100-x.
- Gao X, Shen J, He W, et al. Spatial-temporal analysis of ecosystem services value and research on ecological compensation in Taihu Lake Basin of Jiangsu Province in China from 2005 to 2018. J. Clean Prod. 2021, 317: 128241. https://doi.org/10.1016/j.jclepro.2021.128241.
- Gong W, Duan X, Sun Y, et al. Multi-scenario simulation of land use/cover change and carbon storage assessment in Hainan coastal zone from perspective of free trade port construction. J. Clean Prod. 2023, 385: 135630. https://doi.org/10.1016/j.jclepro.2022.135630.
- Guo S, Wu C, Wang Y, et al. Threshold effect of ecosystem services in response to climate change, human activity and landscape pattern in the upper and middle Yellow River of China. Ecol. Indic. 2022, 136: 108603. https://doi.org/10.1016/j.ecolind.2022.108603.
- Ke X, Tang L. Impact of cascading processes of urban expansion and cropland reclamation on the ecosystem of a carbon storage service in Hubei Province, China. Acta Ecologica Sinica. 2019, 39(2): 672-683. https://doi.org/10.5846/stxb201712042177.
- Khosravi Mashizi A, Sharafatmandrad M. Investigating tradeoffs between supply, use and demand of ecosystem services and their effective drivers for sustainable environmental management. J. Environ. Manage. 2021, 289: 112534. https://doi.org/10.1016/j.jenvman.2021.112534.
- Li M, Liang D, Xia J, et al. Evaluation of water conservation function of Danjiang River Basin in Qinling Mountains, China based on InVEST model. J. Environ. Manage. 2021, 286: 112212. https://doi.org/10.1016/j.jenvman.2021.112212.
- Liu L, Chen X, Chen W, et al. Identifying the Impact of Landscape Pattern on ecosystem services in the Middle Reaches of the Yangtze River Urban Agglomerations, China. Int. J. Environ. Res. Public Health. 2020, 17(14). https://doi.org/10.3390/ijerph17145063.
- Liu X, Zhang L, She D, et al. Postprocessing of hydrometeorological ensemble forecasts based on multisource precipitation in Ganjiang River basin, China. J. Hydrol. 2022, 605: 127323. https://doi.org/10.1016/j.jhydrol.2021.127323.
- Medeiros A, Fernandes C, Gonçalves JF, et al. A diagnostic framework for assessing land-use change impacts on landscape pattern and character – A case-study from the Douro region, Portugal. Landsc. Urban Plan. 2022, 228: 104580. https://doi.org/10.1016/j.landurbplan.2022.104580.
- Mitchell MGE, Devisscher T. Strong relationships between urbanization, landscape structure, and ecosystem service multifunctionality in urban forest fragments. Landsc. Urban Plan. 2022, 228: 104548. https://doi.org/10.1016/j.landurbplan.2022.104548.
- Mitchell MGE, Suarez-Castro AF, Martinez-Harms M, et al. Reframing landscape fragmentation's effects on ecosystem services. Trends Ecol. Evol. 2015, 30(4): 190-198. https://doi.org/10.1016/j.tree.2015.01.011.
- Phillips CL, Wang R, Mattox C, et al. High soil carbon sequestration rates persist several decades in turfgrass systems: A meta-analysis. Sci. Total Environ. 2023, 858: 159974. https://doi.org/10.1016/j.scitotenv.2022.159974.
- Shuangao W, Padmanaban R, Mbanze AA, et al. Using Satellite Image Fusion to Evaluate the Impact of Land Use Changes on ES and Their Economic Values. Remote Sens. 2021, 13(5). https://doi.org/10.3390/rs13050851.
- Song S, Yu D, Li X. Impacts of changes in climate and landscape pattern on soil conservation services in a dryland landscape. Catena. 2023, 222: 106869. https://doi.org/10.1016/j.catena.2022.106869.
- Tang L, Ke X, Zhou Q, et al. Projecting future impacts of cropland reclamation policies on carbon storage. Ecol. Indic. 2020a, 119: 106835. https://doi.org/10.1016/j.ecolind.2020.106835.
- Tang L, Ke X, Zhou T, et al. Impacts of cropland expansion on carbon storage: A case study in Hubei, China. J. Environ. Manage. 2020b, 265: 110515. https://doi.org/10.1016/j.jenvman.2020.110515.
- Tran DX, Pearson D, Palmer A, et al. Quantifying spatial non-stationarity in the relationship between landscape structure and the provision of ecosystem services: An example in the New Zealand hill country. Sci. Total Environ. 2022, 808. https://doi.org/10.1016/j.scitotenv.2021.152126.
- Tscharntke T, Tylianakis JM, Rand TA, et al. Landscape moderation of biodiversity patterns and processes - eight hypotheses. Biol. Rev. 2012, 87(3): 661-685. https://doi.org/10.1111/j.1469-185X.2011.00216.x.
- Wang C, Wu D, Shen Z, et al. How do physical and social factors affect urban landscape patterns in intermountain basins in Southwest China? Landsc. Ecol. 2021, 36: 1893-1911. https://doi.org/10.1007/s10980-020-01182-7.
- Wang K., Zhang C., Chen H, et al. Karst landscapes of China: patterns, ecosystem processes and services. Landsc. Ecol. 2019, 34: 2743-2763. https://doi.org/10.1007/s10980-019-00912-w.
- Winkel G, Lovrić M, Muys B, et al. Governing Europe's forests for multiple ES: Opportunities, challenges, and policy options. Forest Policy Econ. 2022, 145: 102849. https://doi.org/10.1016/j.forpol.2022.102849.
- Wu D, Shao Q, Li J, et al. Carbon fixation estimation for the main plantation forest species in the red soil hilly region of southern-central Jiangxi Province, China. Acta Ecologica Sinica. 2012, 32(1): 142-150. https://doi.org/10.5846/stxb201011241669.
- Wu J. Landscape sustainability science (II): core questions and key approaches. Landsc. Ecol. 2021, 36(8): 2453-2485. https://doi.org/10.1007/s10980-021-01245-3.
- Xi X, Li M, Zhang X, et al. Research on soil organic carbon distribution and change trend in middle east plain and its vicinity in China. Earth Science Frontiers, 2013, 20(1): 154-165. https://www.earthsciencefrontiers.net.cn/CN/Y2013/V20/I1/154.
- Xia H, Kong W, Zhou G, et al. Impacts of landscape patterns on water-related ecosystem services under natural restoration in Liaohe River Reserve, China. Sci. Total Environ. 2021, 792. https://doi.org/10.1016/j.scitotenv.2021.148290.
- Xu Z, Peng J, Zhang H, et al. Exploring spatial correlations between ecosystem services and sustainable development goals: a regional-scale study from China. Landsc. Ecol. 2022, 37: 3201-3221. https://doi.org/10.1007/s10980-022-01542-5.
- Xue C, Chen X, Xue L, et al. Modeling the spatially heterogeneous relationships between tradeoffs and synergies among ES and potential drivers considering geographic scale in Bairin Left Banner, China. Sci. Total Environ. 2023, 855: 158834. https://doi.org/10.1016/j.scitotenv.2022.158834.
- Xue J, Li Z, Feng Q, et al. Spatiotemporal variations of water conservation and its influencing factors in ecological barrier region, Qinghai-Tibet Plateau. J. Hydrol.-Reg. Stud. 2022, 42: 101164. https://doi.org/10.1016/j.ejrh.2022.101164.
- Yang D, Zhang P, Jiang L, et al. Spatial change and scale dependence of built-up land expansion and landscape pattern evolution—Case study of affected area of the lower Yellow River. Ecol. Indic. 2022, 141: 109123. https://doi.org/10.1016/j.ecolind.2022.109123.
- Yun J, Liu H, Xu Z, et al. Assessing Changes in the Landscape Pattern of Wetlands and Its Impact on the Value of Wetland ES in the Yellow River Basin, Inner Mongolia. Sustainability. 2022, 14(10). https://doi.org/10.3390/su14106328.
- Yushanjiang A, Zhang F, Yu H, et al. Quantifying the spatial correlations between landscape pattern and ecosystem service value: A case study in Ebinur Lake Basin, Xinjiang, China. Ecol. Eng. 2018, 113: 94-104. https://doi.org/10.1016/j.ecoleng.2018.02.005.
- Zhai R, Li W, Zhang C, et al. The transiogram as a graphic metric for characterizing the spatial patterns of landscapes. Landsc. Ecol. 2018, 34: 2103-2121. https://doi.org/10.1007/s10980-018-0760-7.
- Zhang F, Zhan J, Zhang Q, et al. Impacts of land use/cover change on terrestrial carbon stocks in Uganda. Phys. Chem. Earth. 2017, 101: 195-203. https://doi.org/10.1016/j.pce.2017.03.005.
- Zhang Y, Liu Y, Zhang Y, et al. On the spatial relationship between ecosystem services and urbanization: A case study in Wuhan, China. Sci. Total Environ. 2018, 637-638: 780-790. https://doi.org/10.1016/j.scitotenv.2018.04.396.
- Zheng L, Wang Y, Li J. Quantifying the spatial impact of landscape fragmentation on habitat quality: A multi-temporal dimensional comparison between the Yangtze River Economic Belt and Yellow River Basin of China. Land Use Pol. 2023, 125: 106463. https://doi.org/10.1016/j.landusepol.2022.106463.
- Zheng Z, Fu B, Hu H, et al. A method to identify the variable ecosystem services relationship across time: a case study on Yanhe Basin, China. Landsc. Ecol. 2014, 29: 1689-1696. https://doi.org/10.1007/s10980-014-0088-x.
- Zhou Y, Yue D, Guo J, et al. Spatial correlations between landscape patterns and net primary productivity: A case study of the Shule River Basin, China. Ecol. Indic. 2021, 130: 108067. https://doi.org/10.1016/j.ecolind.2021.108067.
- Zhu Q, Guo J, Guo X, et al. Relationship between ecological quality and ecosystem services in a red soil hilly watershed in southern China. Ecol. Indic. 2021, 121: 107119. https://doi.org/10.1016/j.ecolind.2020.107119.
No competing interests reported.