Scenario-based prediction and optimization of greenspace ecological network under land-use dynamics: a case study of Nanjing metropolitan area, China - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Sci Rep . 2026 Mar 9;16:12761. doi: 10.1038/s41598-026-40732-y Search in PMC Search in PubMed View in NLM Catalog Add to search Scenario-based prediction and optimization of greenspace ecological network under land-use dynamics: a case study of Nanjing metropolitan area, China Wei Liu Wei Liu 1 School of Horticulture and Landscape Architecture, Jinling Institute of Technology, Nanjing, China 2 College of Landscape Architecture, Nanjing Forestry University, Nanjing, China 3 Nanjing Technology and Application Engineering Research Centre for the Management of Landscape Intelligence, Nanjing, China Find articles by Wei Liu 1, 2, 3 , Yijia Zhao Yijia Zhao 2 College of Landscape Architecture, Nanjing Forestry University, Nanjing, China Find articles by Yijia Zhao 2 , Xuefeng Bai Xuefeng Bai 4 College of Landscape Architecture, Northeast Forestry University, Harbin, China Find articles by Xuefeng Bai 4 , Hao Xu Hao Xu 2 College of Landscape Architecture, Nanjing Forestry University, Nanjing, China Find articles by Hao Xu 2, ✉ Author information Article notes Copyright and License information 1 School of Horticulture and Landscape Architecture, Jinling Institute of Technology, Nanjing, China 2 College of Landscape Architecture, Nanjing Forestry University, Nanjing, China 3 Nanjing Technology and Application Engineering Research Centre for the Management of Landscape Intelligence, Nanjing, China 4 College of Landscape Architecture, Northeast Forestry University, Harbin, China ✉ Corresponding author. Received 2025 Jul 8; Accepted 2026 Feb 16; Collection date 2026. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ . PMC Copyright notice PMCID: PMC13096483 PMID: 41803259 Abstract Land-use dynamics in rapidly urbanizing metropolitan regions profoundly reshape the spatial structure and functional connectivity of greenspace ecological network (GEN). However, how land-use policies influence GEN structure and resilience at the metropolitan scale remains unclear. To accurately predict GEN changes and support scenario-based optimization for sustainable planning, this study proposed a multi-scenario optimization framework that coupled multi-objective programming (MOP) model, patch-level land-use simulation (PLUS) model, morphological spatial pattern analysis (MSPA) model, and least-cost path (LCP) model. The Nanjing Metropolitan Area (NMA) was used to simulate four scenarios: business as usual (BAU), rapid economic development (RED), ecological land protection (ELP), and ecological and economic balance (EEB).The results showed that: (1) The MOP-PLUS model achieved high accuracy (overall accuracy of 84.58%, Kappa coefficient of 0.74, and FoM value of 0.27), effectively capturing regional land-use dynamics; (2) Land-use transitions and GEN structures significantly varied across scenarios, with RED causing severe ecological loss, whereas ELP and EEB scenarios effectively enhanced ecological connectivity; (3) The scenario-based GEN optimization highlighted that the EEB scenario provided the most practical balance, promoting both ecological stability and cost-efficient land management. These findings directly inform sustainable land-use policies and strategic planning decisions in metropolitan regions, revealing the methodological advantages of integrating scenario simulations with GEN analyzes for optimized greenspace management. Supplementary Information The online version contains supplementary material available at 10.1038/s41598-026-40732-y. Keywords: Greenspace ecological network, Land-use change, Scenario simulations, Nanjing metropolitan area Subject terms: Ecology, Ecology, Environmental sciences, Environmental social sciences Introduction Over the past half century, China has experienced substantial urbanization. From 1978 to 2022, China’s urbanization rate surged from 17.92% to 65.22%, marking an annual increase of 1.07 percentage points 1 . However, this rapid development has precipitated issues such as haphazard urban sprawl, predominant extensive economic growth models, and suboptimal land utilization 2 , 3 , engendering ecological challenges including unsustainable consumption of ecological resources, habitat quality diminishment, and biodiversity loss 4 – 6 . Metropolitan areas, with their dense populations, bustling economies, and vibrant human interactions, are central to regional economic growth 7 . Existing research reveals that metropolitan areas not only contribute to approximately three-quarters of China’s pollution 8 while facing a continuous contraction of ecological land 9 , coupled with a consistent decline in the quality of ecological space 10 . Hence, confronting ecological degradation and fostering sustainable development within metropolitan areas has become imperative. Establishing a regional greenspace ecological network (GEN) is recognized as an effective strategy for alleviating the conflict between ecological preservation and economic advancement in metropolitan areas 11 . As the cornerstone of the regional ecological infrastructure, the GEN focuses on enhancing greenspace connectivity. This is achieved by delineating, safeguarding, and establishing critical green habitats and corridors that promote energy and information exchange among terrestrial organisms and preserve the stability of terrestrial ecosystem functions and services 12 , 13 . The “source-resistance surface-corridor” conceptual framework, deeply rooted in landscape ecology theory, is the fundamental model guiding GEN research 14 . Recent methodologies for source identification have integrated morphological spatial pattern analysis (MSPA), connectivity indices, and ecosystem service hotspots to provide a quantitative basis for selection 15 , 16 . Least-cost path (LCP) analysis is a widely applied approach for identifying greenspaces ecological corridors by quantifying the relative difficulty of species movement across heterogeneous landscapes 17 . Modeling functional connectivity from a movement based perspective, it provides a practical framework for regional scale GEN construction. Changes in land-use can alter the original landscape structure and function, affecting the identification of regional GEN 18 , 19 . However, most studies have focused on current land-use analyzes, neglecting the dynamic impact of land-use changes on future regional GEN. This oversight poses substantial challenges for long-term sustainability and resource management in the region. Simulating land-use change scenarios can help to forecast future land-use patterns 20 , 21 . Integrating scenario simulations with GEN construction can facilitate predictions of ecological pattern shifts due to future urban growth under various scenarios 22 . This facilitates the strategic optimization of green-source layouts and corridor orientations in line with anticipated metropolitan developments. Moreover, facilitates the precise delineation of protection boundaries, proactively avoiding negative environmental consequences associated with regional development. Among the prominent land-use change scenario simulation models are the cellular automata (CA) model 23 , future land-use simulation (FLUS) 24 , conversion of land-use and its effects to a small extent (CLUE-S) model 25 , and the patch-generating land-use simulation (PLUS) model 26 . The PLUS model reveals the nonlinear drivers of land-use change and tracks transitions across different land-use types, thus attaining increased simulation accuracy on a regional scale 27 , 28 . However, given its focus on simulation, the PLUS model is limited in predicting land-use structures. To address this limitation, researchers have attempted to merge the PLUS model with land-use structural prediction models to develop a comprehensive framework 29 – 31 . The multi-objective programming (MOP) model integrates ecological and economic strategies to achieve optimal land-use structures in various development scenarios defined by precise and achievable parameter settings 32 . The coupled MOP-PLUS model not only addresses the shortcomings of the PLUS model in land-use structure prediction but also considers various spatial planning strategies, thereby establishing a spatial pattern simulation model that matches the macro-landscape with spatial unit layouts. Recently, the MOP-PLUS model has been employed to simulate future land-use changes under different scenarios 33 , estimate carbon storage capacities 34 , evaluate landscape risks 35 , and assess the evolution of ecosystem service values 36 . However, its application to GEN analysis has rarely been reported. This research addresses this knowledge gap by proposing an integrated approach combining spatial partitioning with MOP and PLUS models to improve the accuracy of regional GEN simulations under different future scenarios. In addition, the theoretical discussion on how land-use policies influence GEN structure and resilience at the metropolitan scale remains unclear. By analyzing policy-driven land-use scenarios, this study contributes to a better understanding of how different development strategies, balancing economic and ecological objectives, reshape regional GEN connectivity and resilience. As a national-level metropolitan area in China, the Nanjing Metropolitan Area (NMA) has experienced rapid urbanization through consistent policy support, resulting in increasingly severe human impacts on the regional ecological environment. Researchers have investigated the evolution of ecosystem service values within the NMA 37 , identified areas of ecological and economic conflict 38 , and developed functional land resource zoning strategies for the NMA based on ecological efficiency 39 . These studies have predominantly focused on past or present ecosystem structures and functions. However, they lack research integrating future dynamic land-use changes with GEN construction. This limits their ability to inform proactive policymaking for the sustainable use of land resources and ecological planning. In this study, we aimed to address current knowledge gaps by integrating the MOP and PLUS models within a spatial partitioning framework to improve the accuracy of regional land-use simulations under different future scenarios, considering the NMA as the study region. Based on the simulated land-use patterns, the LCP model was applied to identify regional GEN for each scenario and explored the variances in these networks under different development scenarios. Overall, this study aims to explore the relationships among land-use policies, GEN structures, and system resilience at the metropolitan scale, focusing on how different policy orientations change ecological connectivity under competing economic and ecological objectives. Materials and methods Study area Located on the lower reaches of the Yangtze River, the NMA comprises nine cities in the Jiangsu and Anhui provinces and covers an area of approximately 66,000 km 2 (Fig. 1 ). Geographically, the NMA serves as a transitional zone that not only connects the eastern and central regions of China but also serves as a critical factor that fosters economic synergy between the southern and northern regions. Economically, the NMA is a powerful growth engine within the Yangtze River Delta, asserting its importance in the broader economic landscape of Eastern China. By 2020, the metropolitan area’s GDP had reached 41,289 trillion yuan, accounting for 16.90% of the Yangtze River Delta’s regional GDP and 4.07% of the national GDP. The demographic characteristics of the NMA are also noteworthy, with a total population of 35.31 million and an urbanization rate of 70.51%, which is higher than the national average of 63.89%. Notably, the NMA is in a period of rapid growth, with rapid expansion and industrial development 40 . Fig. 1. Open in a new tab Location of study area. The map was created using ArcGIS Desktop 10.6 ( https://www.esri.com/ ). Data The data used in this study are described in Appendix A1 and primarily consisted of land-use and cover data, digital elevation model (DEM), slope, annual average temperature and precipitation, GDP, population density, and traffic networks data. All raster data were pre-processed by projecting, clipping, and resampling to obtain a standard dataset for the NMA with a resolution of 30 m. Methodology The methodological innovation of this study lies in coupling the MOP model with PLUS model, enhanced by spatial partitioning, to comprehensively simulate complex ecological dynamics at the metropolitan scale. First, focusing on the NMA, four development scenarios, business as usual (BAU), rapid economic development (RED), ecological land protection (ELP), and ecological and economic balance (EEB), were established using the MOP model to determine optimal land-use structures. Second, the PLUS model was used to simulate the spatial patterns of land-use change for each scenario. The model was calibrated and validated by simulating the 2020 land-use pattern using data from 2010 to 2015 and comparing the results with the actual 2020 map. After validation, the verified parameters were applied to predict the land-use pattern for 2035. Finally, MSPA and LCP analyses were applied to identify green habitat sources and ecological corridors, and network structure indicators were used to evaluate and compare GEN configurations across scenarios. The overall workflow is shown in Fig. 2 . Fig. 2. Open in a new tab Study workflow. Scenario setting The MOP model refers to a multi-objective programming model used to determine the quantitative land-use structure under different objectives 41 . This model defines land-use structures by formulating objective functions and constraints that integrate economic, environmental and social factors. Four development scenarios were designed for the NMA for the year 2035, in line with regional spatial planning: BAU scenario: This scenario represents trend-continuing land-use change without explicit policy intervention. The land-use structure for 2035 was generated by extrapolating historical land-use transitions using a Markov chain model. RED scenario: This scenario is predicted on the assumption of maximizing economic benefits from land-use. This gives preference to the allocation of land to support and enhance economic growth, potentially changing land-use patterns to favor economically productive activities. Under this scenario, the MOP model was configured with a single objective function aimed at maximizing economic benefits from land-use. ELP scenario: This scenario emphasizes ecological protection and environmental enhancement through strict land-use planning and management policies that guide land-use towards conservation and ecological enhancement. The MOP model was configured with a single ecological objective to maximize ecological benefits. EEB scenario: This scenario seeks a harmonious balance between economic development and ecological preservation by integrating both objectives into a single composite objective function using a weighted-sum approach. Following existing studies on integrated land-use planning and sustainability oriented optimization 32 , 42 , 43 , economic and ecological benefits were assigned equal weights to reflect their comparable importance. Under this assumption, the objective function aims to maximize the overall comprehensive benefits of land-use allocation. The Markov chain model was applied only to the BAU scenario because it is suitable for simulating land-use change under historical trends, whereas the RED, ELP, and EEB scenarios are goal-oriented and therefore require the MOP model to determine the land-use structure according to planning objectives. To refine the objective functions for each scenario, this study uses quantitative indicators for economic and ecological land-use benefits, following previous research 32 . The economic benefit indicator ( ) and ecological benefit indicator ( are defined as 1 2 where xi denotes the area of land-use type i ; di and pi represent economic and ecological benefit coefficients, respectively. Economic coefficients reflect sectoral productivity, whereas ecological coefficients are based on the China Ecosystem Services Value Equivalence Factor, adjusted for local context. The parameter settings for each scenario are detailed in Table 1 . Table 1. Optimization goals for four scenarios. Scenarios Optimization goals Parameter setting BAU A continuation of the historical process = + + + + + + 1488.00 +0 +46.29 +966.36 +233.25 +616.24 + +583.31 RED Max( ) ELP Max( ) EEB Max( , ) Open in a new tab is water area; construction land area; agricultural land area; forest area; grassland area; wetland area; other land area; urban green space area. Land-use structure constraints This study set upper and lower limits for changes in different types of land-use based on current spatial planning and environmental policies. Given the expansive metropolitan scale coupled with challenges such as disparate development levels and policy diversity, single constraints could neglect the diverse developmental variations across subregions, potentially leading to simulation results that differ from the actual conditions. To address this issue, the study introduces a spatial partitioning approach that assigns different constraint thresholds to different cities, more accurately reflecting their individual development contexts. This methodological improvement significantly enhances the modelling power of ecological informatics and its relevance to real-world spatial planning. The specific constraint conditions are listed in Appendix A2. The upper and lower limits of various types of land-use changes are detailed through analyzes of total land area, forest cover share, wetland protection rate, water protection rate, development intensity of construction land, agricultural land protection area, and urban green space rate. Land-use change simulation The PLUS model (Patch-generating Land Use Simulation Model, https://github.com/HPSCIL/Patch-generating_Land_Use_Simulation_Model ) integrates a land expansion analysis system (LEAS) for rule extraction with cellular automata using random seeds (CARS) for dynamic simulation. Initially, the LEAS model is responsible for extracting expansion pattern for each land type from the land-use maps at two time points, together with key spatial drivers, applies a random forest regression to estimate expansion probabilities. The CARS model then simulates future land-use patterns based on these probabilities 26 . Based on previous research on the drivers of land-use change in the NMA and data availability, 10 driving factors were selected, including DEM, slope, annual average precipitation, GDP, population density, total food production, distance from built-up areas, distance from roads, financial expenditure on forestry and water conservation, and the area of nature reserves 40 . These factors represent major natural and socio-economic drivers that have been widely used in previous land-use simulation studies 44 , 45 . Policy-related constraints were mainly incorporated through scenario settings and the random forest mechanism of the PLUS model helps to reduce the influence of potential multicollinearity among diving factors. Two key parameters were defined in the CARS model. The transition matrix defines whether one land-use type can convert into another, with 0 indicating disallowed conversions and 1 indicating allowed conversions. This parameter was set based on historical land-use transition likelihoods in the study area 46 . Transitions accounting for more than 0.1 percent of the total change for a specific land-use type were assigned a value of 1, while those below this threshold were assigned a value of 0. The neighborhood weights quantified the expansion tendency of each land-use type, ranging from 0 to 1, with higher value indicating a higher probability of expansion. The neighborhood weights were calculated from the absolute dimensionless land-use change values using the following formula: 3 where is the neighborhood weight of land-use type i , is the area change of type i during period T, and and are the minimum and maximum area changes among all land-use types, respectively. Spatial partitioning was introduced to account for the heterogeneous development landscapes of different cities within the NMA. This approach involved city-specific adjustments to the transition matrix and neighborhood weight, ensuring that the simulation is tuned to the different development characteristics of each city. The detailed parameters of the transition matrix and neighborhood weights have been listed in Appendix A3 and A4, respectively. GEN identification and assessment This study used MSPA and LCP models to identify regional GEN. Land-use data were transformed into binary maps in ArcGIS (ESRI ArcGIS Desktop 10.6, https://www.esri.com ), classifying forest, grassland, wetland and urban green space as foreground and other land-use types as background. GuidosToolbox (GuidosToolbox v3.0, https://forest.jrc.ec.europa.eu/en/activities/lpa/gtb/ ) was then used to categorize these maps into MSPA landscape types. Core areas larger than 10 km2 with high connectivity were selected as ecological sources 47 . Corridors were identified using Linkage Mapper (Linkage Mapper v2.0.0, https://linkagemapper.org/ ), which calculates cost-weighted minimum paths between ecological sources, and redundant paths were removed to construct the corridor network 48 . To accurately construction of GEN, landscape resistance, which represents the relative difficulty encountered by organisms when moving across different spatial units, must be reasonably estimated. Following our previous studies 47 , the resistance surface was constructed based on multiple factors, including land-use type, MSPA landscape type, topography, and human activities. Resistance values were assigned based on the relative suitability of each factor for ecological movement, with higher values indicating greater resistance. Then, the analytical hierarchy process (AHP) was used to determine the weight of each resistance factor, reflecting their relative importance in influencing ecological connectivity. Finally, the weighted resistance layers were overlaid to generate the final comprehensive resistance surface used for corridor identification. To ensure methodological consistency and comparability across scenarios, the same resistance values and weights were applied to the baseline year of 2020 and to all simulated land-use scenarios for 2035 (Table 2 ). Table 2. Resistance value and weight of resistance factors in the NMA (adapted from References 44). Resistance factor Weight Tiered factor Value Resistance factor Weight Tiered factor Value MSPA landscape type 0.38 Core Hubs 1 Topography 0.15 Elevation (m) < 50 5 Other cores 5 50–175 20 Bridge 10 175–350 60 Other six MSPA types 50 350–650 200 Background 600 > 650 600 Land-use type 0.25 Forest, grassland, wetland, urban green space 5 Slope (°) < 5 5 water (km 2 ) River width > 100 m, and lakes area > 50 1000 5–15 20 15–25 200 > 25 600 Human activities 0.22 Distance from built-up area (km) > 2.5 5 River width 50–100 m, lakes area < 50 and > 10 400 1.5–2.5 100 1.0–1.5 500 0.5–1.0 800 < 0.5 1000 Lakes area < 10 40 Distance from traffic artery (km) > 3.0 5 2.0–3.0 100 Agriculture land 60 1.5–2.0 500 Construction land 1000 0.5–1.5 800 Other land 600 < 0.5 1000 Open in a new tab The structural stability of the GEN was assessed using four indices: network closure index (α), line-point ratio index (β), network connectivity index (γ), and cost ratio (δ). The α index quantified the proportion of loops within the network, with higher values indicating more pathways for species migration and a lower probability of ecological disturbance. The β index identified the network structure by calculating the mean connectivity per node. β < 1 indicated a dendritic, tree-like architecture, β = 1 typified a singular loop structure, whereas β > 1 indicated intricate connectivity within the network. This index reflected the level of connectivity between nodes, which was essential for maintaining ecological connectivity. Notably, the δ index provided an economic–ecological perspective by correlating the cost and ecological return, with lower values indicating a more economically viable GEN construction. The indexes were calculated as follows: 4 5 6 7 where L is the number of corridors, V is the number of sources, and d is the total resistance in all corridors of the network. Results Change in land-use from 2010 to 2020 From 2010 to 2020, land-use types in the NMA were dominated by agricultural land, forest, construction land, and water, which collectively accounted for more than 98% of the total area. In contrast, grassland, wetland, urban green space, and other land types collectively occupied less than 2%. As shown in Fig. 3 A, land-use change during this period was primarily characterized by a significant decline in agricultural land and rapid expansion of construction land. The proportion of agricultural land decreased from 63.37 to 59.32%, whereas construction land increased significantly from 8.68 to 12.08%, representing the most notable structural change. In contrast, changes in other land-use types were relatively moderate. Forest area increased slightly from 19.24 to 19.52%, and urban green space expanded from 0.20 to 0.37%. Water and grassland exhibited a pattern of initial increase followed by a slight decline, whereas wetlands and other land types showed a U-shaped trajectory. However, the proportional changes in all these land-use types were minor, with variations of less than 0.20% over the study period. Fig. 3. Open in a new tab Land-use change and land cover conversion of 2010–2020. Figure 3 B further reveals the dominant land-use conversion probabilities from 2010 to 2020. Agricultural land was primarily converted from grassland and wetland, with conversion probabilities of 36.97% and 30.45%, respectively. Forest was primarily derived from other land (14.88%) and grassland (14.67%). Wetland was an important source for water, with 28.81% converted to water, whereas 6.72% of grassland transitioned to wetland. Construction land was largely converted from urban green space (34.00%) and other land (28.73%). In addition, 5.28% of other land was converted into grassland. Verification of land-use change simulation accuracy Land expansion information from the NMA from 2010 to 2015 was extracted and used to simulate land-use development in 2020. A comparative analysis was conducted between the simulation results and the actual land-use situation in 2020. As shown in Fig. 4 , the simulated distribution of land-use was similar to the observed landscape patterns. The simulation results showed a spatial pattern characterized by a mixture of agricultural land, forest, and grassland, which closely matched the actual landscape pattern. In addition, the boundaries of the construction land were generally consistent with the actual situation, showing the effective preservation of urban green spaces. Fig. 4. Open in a new tab Land-use actual and simulation results of NMA in 2020. Simulation accuracy was evaluated using overall accuracy, the Kappa coefficient, the figure of merit (FoM), and class-based precision, recall, and F -score (Table 3 ). Overall accuracy reflects the proportion of correctly predicted overall land-use classes, whereas the Kappa coefficient (values of > 0.6 were considered satisfactory) measures the agreement between simulated and actual classifications 49 . The FoM assesses the accuracy of land-use change at the pixel level, with values above 0.2 indicating good model performance 50 . Precision and recall quantify the reliability and completeness of class-level predictions, respectively, and the F -score integrates both metrics to provide a balanced measure of classification performance 51 . Table3. Validation of predictive performance of model. Land- use types Area of simulated land-use in 2020 (km 2 ) Recall WA AL FO GL WL CL OL UG Area of actual land-use in 2020 (km 2 ) WA 3726.72 778.15 64.56 17.41 45.31 74.43 7.10 4.08 0.79 AL 730.83 33,630.80 2304.77 49.63 48.87 2027.08 33.20 9.76 0.87 FO 45.47 1896.88 10,681.20 20.35 15.37 106.95 7.79 5.84 0.84 GL 18.13 49.29 17.71 242.27 14.06 20.18 8.89 0.55 0.65 WL 45.65 15.58 20.11 12.06 306.69 9.36 0.10 0.18 0.75 CL 82.06 1068.81 189.79 16.78 15.33 6475.58 26.33 24.31 0.82 OL 2.92 23.43 23.38 4.79 0.15 10.61 130.17 0.58 0.66 UG 8.76 25.09 13.82 2.48 0.22 18.97 0.87 169.29 0.71 Precision 0.80 0.90 0.82 0.66 0.69 0.74 0.61 0.79 Overall accuracy: 0.8458 F score 0.79 0.88 0.82 0.66 0.72 0.78 0.63 0.75 Kappa:0.74 FoM:0.27 Open in a new tab WA, water; AL, agricultural land; FO, forest; GL, grassland; WL, wetland; CL, construction land; OL, other land; UG, urban green space. In addition, the model in this study achieved an overall accuracy of 84.58%, kappa coefficient of 0.74, and FoM of 0.27, indicating high simulation reliability for the NMA. Dominant land-use types, including agricultural land, forest, construction land, and water, exhibited high precision and recall, thereby reflecting robust predictive performance for major land-use components. Although grassland and other land types showed relatively lower F -scores, their limited area resulted in minimal influence on overall accuracy. Therefore, these results indicate that the model effectively captures both dominant land-use patterns and key land-use change processes in the NMA. Changes in land-use under four scenarios in the NMA The MOP-PLUS model was employed to predict and simulate land-use change for the year 2035. Given the unique objectives associated with each scenario, notable differences existed in land-use structure and spatial distribution (Table 4 and Fig. 5 ). Table 4. Area and proportion of each land-use type in the NMA under different scenarios in 2035. Number in parentheses are land-use shares in %, and land-use area in km 2 . Land-use type 2020 2035 BAU RED ELP EEB Water 4717.76 4752.68 3746.79 5315.89 5312.62 (7.21) (7.27) (5.73) (8.13) (8.12) Agricultural land 38,789.92 36,884.36 37,702.69 35,416.33 35,803.78 (59.31) (56.40) (57.65) (54.15) (54.74) Forest 12,779.84 12,835.00 12,663.03 15,752.24 13,629.28 (19.54) (19.63) (19.36) (24.09) (20.84) Grassland 371.07 391.45 528.11 383.70 461.52 (0.57) (0.56) (0.81) (0.59) (0.71) Wetland 409.73 434.13 252.47 661.02 653.37 (0.63) (0.66) (0.39) (1.01) (1.00) Construction land 7898.99 9644.31 9900.64 7543.20 9177.42 (12.08) (14.75) (15.14) (11.53) (14.03) Other land 196.03 203.91 196.94 0.00 0.00 (0.30) (0.31) (0.30) (0.00) (0.00) Urban green space 239.49 256.99 412.16 330.47 364.85 (0.37) (0.39) (0.63) (0.51) (0.56) Open in a new tab Fig. 5. Open in a new tab Land-use change in the NMA in 2035 under the four development scenarios. A, B, C, and D represent the scenarios of BAU, RED, ELP, and EEB in 2035, respectively. BAU scenario The BAU scenario was simulated based on observed land-use changes in the NMA over the last 5 years, reflecting unrestricted development without policy or regulatory constraints. Under this scenario, economic benefits increased by approximately 19.5% compared with those in 2020, whereas environmental benefits remained largely unchanged. The land-use structure shifted significantly, with a substantial increase in building land and a marked decrease in agricultural land, whereas other land use categorize experienced only minor changes. Spatially, land conversion was concentrated around existing urban areas, primarily through the conversion of non-urban land into building land on the periphery of cities. Cities along the Yangtze River, such as Nanjing, Zhenjiang, Yangzhou, Ma’anshan and Wuhu, experienced strong urban growth driven by the Yangtze River Economic Belt. However, the lack of ecological policies resulted in significant encroachment on ecological spaces, including water bodies and forests along the river (Fig. 5 A). RED scenario The RED scenario was designed to maximize economic benefits from land-use across the metropolitan area. Compared to 2020, economic benefits increased by 22.38%, whereas environmental benefits decreased by 7.25%, reflecting a clear prioritization of economic growth over environmental protection. Construction land increased rapidly to 6,822.67 km2, the largest of all scenarios, primarily because of the conversion of agricultural land. Although legal constraints limited the total loss of agricultural land, significant areas of water, forest and grassland were also converted to agricultural and construction uses, resulting in a significant reduction in natural and semi-natural resources. Compared with that in the BAU scenario, the trend towards urban expansion was more pronounced in the RED scenario. However, environmental protection policies still ensured the preservation of large green spaces within and between cities (Fig. 5 B). ELP scenario The ELP scenario prioritized ecological protection and the maximization of ecological benefits. Compared with those of RED scenario, ecological benefits increased by 17.63%, whereas economic benefits decreased by 6.05%. Notable expansions occurred in forests, wetlands, and water areas by 2972.40 km 2 , 251.29 km 2 , and 598.13 km 2 , respectively. These were the largest increases among all scenarios. Much of this growth resulted from converting agricultural land, especially in central, hilly regions, where ecological value is higher (Fig. 5 C). However, the expansion of construction land was limited in this scenario, with its scale and spatial distribution remaining almost unchanged compared with 2020. It can be asserted that the scenario trades off economic benefits for ecological benefits. EEB scenario In the EEB scenario, economic and ecological benefits increased by 10.10% and 12.46%, respectively. Similar to the ELP scenario, the areas of agricultural land and other land decreased, whereas those of water, forest, grassland, and wetland increased. However, the increase in forest area under this scenario was significantly lower than that under the EEL scenario. Consequently, a significant amount of agricultural land was converted to construction land, resulting in an increase in urban construction land of 1278.43 km2 compared with that of 2020. It is worth noting that the area of other land became zero in both the ELP and EEB scenarios. Under strengthened ecological constraints, the other land was mainly transformed into forest and grassland. In the EEB scenario, a proportion was additionally converted into construction land to meet urban development demand. Compared with the RED scenario, EEB scenario effectively preserved ecological areas, particularly in water-rich regions such as the Jianghuai Plain and the ancient Danyang Lake, where minor water bodies and wetlands were substantially restored (Fig. 5 D). In contrast to the ELP scenario, the EEB permitted moderate urban expansion to accommodate development needs. The EEB scenario is characterized by a balanced land-use structure, which fosters both rapid economic development and ecological stability within the NMA. Changes in GEN under different scenarios Source changes and distribution MSPA and connectivity indicators were used to identify ecological source areas. The results showed that although the total source area increased under all four scenarios by the year 2035, the number of sources decreased in the BAU and RED scenarios, but increased in the ELP and EEB scenarios. This divergence reflected the simultaneous agglomeration and fragmentation of green sources (Fig. 6 ). In the BAU and RED scenarios, the number of ecological sources decreased, but their total area increased. This phenomenon was mainly due to the consolidation of small and fragmented green patches into fewer but larger core sources during scenario simulation. Fig. 6. Open in a new tab Comparison of number and area of sources and corridors. A, B, C, and D represent the scenarios of BAU, RED, ELP, and EEB in 2035, respectively. To improve the comparative analysis of the dynamics of green sources across two different periods and scenarios, this study implemented connectivity indicators to quantify the importance of green sources. Sources were classified high, medium, or low was determined by the natural break method, which was based on 2020 dI values. As shown in Fig. 7 , the spatial distribution of green sources under the BAU and RED scenario exhibited significant imbalances, with all high-quality sources located exclusively in the southern highlands. Despite the notable preservation of the Gaoyou wetland and Baohua mountain patches (Fig. 7 A,B), the loss of connectivity has led to a decline in source function and a reduction in source area within the northern and central regions. In contrast, the ELP and EEB scenarios showed a more balanced spatial pattern of sources, with a notable increase in the size of medium-quality source (Fig. 7 C,D). Fig. 7. Open in a new tab Distribution of green sources in the NMA in 2020 and 2035. A, B, C, and D represent the scenario of BAU, RED, ELP, and EEB in 2035, respectively. Corridor changes and distribution This study employed LCP model to identify the spatial distribution of green corridors under different scenarios. For each potential corridor, the cost-weighted distance (CWD) and path length (PL) were calculated, and the CWD/PL ratio was used to assess corridor connectivity 52 . This ratio reflects the cumulative movement resistance per unit distance along a corridor and provides a comprehensive indicator of corridor connectivity for multiple species, rather than an exact simulation of movement for a specific species. Corridors classification was conducted using the natural break method based on the CWD/PL values from the year of 2020. To ensure comparability, and this same classification was consistently applied across various scenarios. Higher-grade corridors were characterized by lower CWD/PL ratios, indicating lower movement and stronger ecological connectivity. As shown in Fig. 6 , the number and length of corridors both decreased under the BAU and RED scenarios, while both increase under the ELP and EEB scenarios. The spatial distribution of corridors is closely related to the pattern of sources, with denser and shorter corridors found in the southern and western regions, where there are more sources, and sparser and longer corridors occurring in the north, where sources are fewer. By the year 2020, the fundamental structure of the corridor network in the NMA had been established. In the BAU scenario, the GEN network suffered degradation due to the loss of source status for the Gaoyou wetland and Baohua mountain patch. This loss resulted in the absence of corridors in the north and a marked reduction in central corridors, forming a simple ring structure. However, the western network structure was relatively complex because the Zhimaling patch is well-preserved. (Fig. 8 A). In the RED scenario, corridors displayed significant imbalances in their distribution, with a concentration in the southern region and a paucity in the western sector. These corridors were distributed in an isolated, “bead-like” pattern, with minimal presence in other regions (Fig. 8 B). In contrast, the ELP and EEB scenarios exhibited a more balanced corridor distribution, including the addition of new corridors, particularly in the highly urbanized areas surrounding Nanjing (Fig. 8 C,D). Fig. 8. Open in a new tab Distribution of green corridors in NMA in 2020 and 2035. A, B, C, and D represent the scenario of BAU, RED, ELP, and EEB in 2035, respectively. Evaluation of GEN structure In the four development scenarios, the α indices for the ELP and EEB scenarios with emphasis exceeded the 2020 reference values, indicating an improvement in the GEN loops (Table 5 ). This enhancement promoted more efficient species movement and energy dynamics under these scenarios. Conversely, the BAU and RED scenarios showed a decrease in the α indices, indicating a constricted network, thus inhibiting species migration pathways. Table 5. Indicators of green space ecological network structure in NMA in 2020 and 2035. α indicator β indicator γ indicator δ indicator 2020 0.368 1.680 0.583 0.951 2035 BAU 0.313 1.568 0.548 0.954 RED 0.237 1.406 0.500 0.893 ELP 0.419 1.787 0.616 0.951 EEB 0.414 1.776 0.613 0.951 Open in a new tab The β and γ indices were similar to the trend observed in the α index, further supporting the variation in network robustness between scenarios. The GEN structure under the ELP scenario emerged as the most complex, with excellent node connectivity, which is a clear indication of its well-integrated structure. This was closely followed by the EEB scenario. In contrast, the RED scenario exhibited the least connectivity. The δ index refected the cost dynamics of GEN construction, revealing a cost-effective nature in the RED scenario. This efficiency was due to a streamlined corridor layout, primarily concentrated in the southern and western regions, which were characterized by less resistant terrain. By contrast, the cost ratios in the other scenarios were approximately 0.95, reflecting higher investment requirements. This increase was attributed to the metropolitan extent of the corridors, which crossed several urban areas with significant landscape resistance, and the inherent complexity of the resulting network structures. A detailed analysis of the ELP and EEB scenarios revealed minor differences among the four network structural indices. This observation emphasizes the structural similarity of GEN in these two scenarios. Discussion Incorporating administrative-based spatial heterogeneity to improve the applicability of the MOP-PLUS model for simulating regional land-use change This study developed an integrated MOP-PLUS framework for predicting the quantitative composition and spatial distribution of land-use change at the metropolitan scale. Previous applications of the MOP-PLUS model have predominantly assumed spatial uniformity by employing identical constraints and transformation protocols 32 , 36 . However, land-use dynamics are naturally spatially heterogeneous, which is particularly prominent at the metropolitan scale owing to differing ecological conditions, socioeconomic processes, and policy settings. Ignoring this spatial variability can result in inaccurate estimates of land-use change 53 . In this study, spatial heterogeneity was represented through an administrative-based spatial partitioning approach, in which different parameter constraints and transition rules were assigned based on to city-level administrative units. This approach does not constitute functional or ecological zoning in a strict sense, but instead reflects differences in land-use policies, planning regulations, and development intensities across cites. Accordingly, it should be understood as a policy-oriented spatial representation instead of a zoning scheme based on ecological processes or land-use functions. From a modeling perspective, this represents a practical extension of the MOP-PLUS framework by explicitly incorporating policy-driven spatial heterogeneity, thereby improving its applicability to metropolitan regions characterized by multi-jurisdictional governance. This study prioritized city-level partitioning, recognizing that national spatial planning policies have a significant impact on regional land-use development and that the city scale is central to the implementation of spatial strategies in China. In particular, cities had significant differences in requirements for forest cover, wetland conservation, farmland protection, and urban development boundaries. Besides administrative partitioning, spatial zoning based on spatial function 54 or socio-economic development 55 represents a more process-oriented approach. Recent studies have also highlighted the value of spatial clustering methods to improve the validity of spatial zoning and the accuracy of land use change models. For example, Xu et al. 56 used the area-weighted frequency of land-use change to define zoning in the Dianchi Basin, while Ke et al. 57 modeled urban growth in Wuhan by subdividing the city using k-means and k-nearest neighbor clustering algorithms. The present study relies on administrative units as a policy-relevant spatial proxy, while recognizing that future work could integrate functional or ecological zoning to better capture underlying landscape processes. Insights from the construction of GEN under different development scenarios To understand how the construction of GEN respond to different regional development pathways, we compared GEN structure under four scenario simulations. Variance in the distribution of sources and corridors reflected land-use diversity driven by different development strategies, highlighting the inherent trade-offs between economic expansion and ecological protection 19 . Across the four scenarios, land-use strategies primarily affect the GEN structure through ecological source integration, landscape resistance redistribution, and corridor redundancy adjustment. Scenarios that emphasize ecological protection or balanced development, such as the ELP and EEB scenarios, promoted the consolidation of fragmented green spaces into larger and more continuous core sources, while maintaining multiple alternative corridors, thereby enhancing network connectivity and complexity. In contrast, regional development pathways dominated by rapid urban expansion, such as the RED scenario, tend to simplify GEN topology by increasing movement resistance and removing secondary corridors. Although this configuration reduces construction costs, the resulting lack of corridor redundancy substantially weakens overall network resilience. These interacting processes are reflected in the observed changes in GEN structure indicators, with higher α, β, and γ values under the ELP and EEB scenarios indicating improved connectivity and robustness, whereas lower values under the RED scenario indicating a simplified and more vulnerable GEN structure. The EEB scenario represents a relatively balanced development pathway that moderates urban expansion while preserving key green ecological spaces. This balance helps to maintain GEN connectivity and accommodates necessary urban growth. Development strategies that integrate ecological and economic objectives are generally more conducive to sustaining metropolitan-scale GEN 58 . However, the GEN connectivity of local areas is highly sensitive to development pathways, even under scenarios that perform well at the metropolitan scale. For example, the Putang forest patch becomes fragmented under the EEB scenario, resulting in the loss of the corridor with the Marenshan patch, whereas this connection is retained under the BAU scenario. This contrast indicates that optimizing GEN at the metropolitan scale does not necessarily guarantee stable local connectivity. Similar studies have been reported in Miami 59 and Tokyo metropolitan areas 60 , emphasizing the challenge of relying on a single growth model to fully address the diverse requirements within metropolitan regions. Consequently, regional ecological space planning should move beyond a single-scenario approach, and integrate attributes from multiple scenarios to refine the GEN structure and improve the regional ecological resilience. Policy implications for metropolitan land-use planning and ecological protection Metropolitan land-use policies should aim to expand beyond single-objective planning and toward an integrated policy framework that jointly addresses both economic development and ecological conservation. Multi-scenario simulations reveal that some sources, such as large forest patches in the southern highlands and western regions (e.g., the Zhimaling patch), remain of high ecological importance across different development scenarios. These areas therefore represent zones of stable ecological significance and should be prioritized for long term protection through strict land-use control and ecological redline enforcement. In contrast, corridor analyses reveal that ecological connectivity in rapidly urbanizing areas, particularly in the northern Jianghuai Plain and along the Yangtze River (e.g., the Gaoyou wetland–Baohua mountian corridor), is highly sensitive to development pathways. In these areas, GEN degradation is mainly driven by corridor fragmentation associated with urban expansion rather than by the loss of core green patches. This indicates the need for policies that regulate development intensity and actively maintain or restore ecological corridors. From a planning implementation perspective, the multi-scenario GEN maps provide a spatially tool for guiding ecological conservation and enhancing connectivity. By identifying ecological sources and corridors that remain stable or vulnerable under different development pathways, these maps enable policymakers to target priority areas for protection, regulation, and restoration more precisely 46 , 61 . Linking multi-scenario GEN results with spatial planning decisions supports more targeted ecological redline policies and green infrastructure planning at the metropolitan scale 62 , thereby improving the alignment between regional development strategies and long-term ecological resilience goals. Study limitations This study presents a framework for assessing GEN through simulations of future land-use changes, providing a strategic tool for ecological conservation and sustainable metropolitan area development. However, there are some limitations. First, although the advanced patch level land-use simulation model accurately reflects development trends in the NMA, the study did not include comparative analyzes with other models. Future research should include models such as FLUS and CLUE-S in comparative studies to improve the accuracy of land-use simulations. Second, although the integration of MSPA and connectivity indices, successfully identifies large regional green sources, it tends to ignore smaller and more fragmented sources 22 , potentially compromising the integrity of the GEN. Future research could further refine the identification of ecological sources using machine learning methods or deep learning algorithms. Third, the weighting of economic and ecological benefits in the EEB scenario was based on an assumption specific to the study area and supported by policy and the literature, rather than being a universally applicable solution. In other regions or planning scenarios, alternative weight arrangements may be more suitable and could result in varied land use patterns and GEN structures. Conclusion This study integrated the MOP-PLUS model with MSPA and LCP to simulate land-use dynamics and the evolution of regional GEN in the NMA under four development scenarios. The results demonstrate that the MOP-PLUS model possesses high predictive accuracy, and scenario analysis reveals substantial differences in regional GEN. Specifically, intensive urban expansion under RED significantly weakened ecological connectivity, whereas ELP and EEB scenarios significantly improved habitat integration and network complexity. In particular, the EEB scenario achieved a more effective balance between development needs and ecological resilience, resulting in cost-effective improvements in GEN structures. According to our finding, policy orientations systematically change ecological sources, corridors, and network topology, thereby influencing metropolitan-scale GEN connectivity. Comparative scenario analysis further indicated that integrated policy strategies that balance ecological protection and economic development are more conducive to sustaining robust GEN than single-objective approaches. We provide policy-oriented insights for GEN planning, emphasizing the critical role of land-use policy design in guiding GEN configuration and enhancing ecological resilience in metropolitan areas. These insights offer conceptual guidance for aligning spatial planning decisions with long-term ecological sustainability goals. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 (71.9KB, rar) Author contributions Wei Liu: Conceptualization, Methodology, Investigation, Formal analysis, Writing—original draft, Writing—review & editing and Funding acquisition. Yijia Zhao: Data curation, Investigation and Visualization. Xuefeng Bai: Resources, Software and Validation. Hao Xu: Conceptualization, Methodology, Supervision, Project administration and Funding acquisition. Funding This research was funded by Jiangsu Higher Education Institutions Philosophy and Social Sciences Research General Projects (2025SJYB0422), the Research Initiation Programme for High-level Talents of Jinling Institute of Technology (jit-b-202426), the Postgraduate Research & Practice Innovation Program of Jiangsu Province (KYCX20_0870) and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD). Data availability The data that support the findings of this study are available from the first author, [Wei Liu, Email: [email protected]], upon reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References 1. National Bureau of Statistics. Statistical Bulletin of the People’s Republic of China on National Economic and Social Development 2022 , https://www.stats.gov.cn/sj/zxfb/202302/t20230228_1919011.html?eqid=ec830cf1000c5d7100000002642ee4fa (2023). 2. Liu, Y., Fang, F. & Li, Y. Key issues of land use in China and implications for policy making. Land Use Policy 40 , 6–12. 10.1016/j.landusepol.2013.03.013 (2014). [ Google Scholar ] 3. Liu, T., Liu, H. & Qi, Y. Construction land expansion and cultivated land protection in urbanizing China: Insights from national land surveys, 1996–2006. Habitat Int. 46 , 13–22. 10.1016/j.habitatint.2014.10.019 (2015). [ Google Scholar ] 4. Felipe-Lucia, M. R. et al. Land-use intensity alters networks between biodiversity, ecosystem functions, and services. Proc. Natl. Acad. Sci. U. S. A. 117 , 28140–28149. 10.1073/pnas.2016210117 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Antwi, E. K., Krawczynski, R. & Wiegleb, G. Detecting the effect of disturbance on habitat diversity and land cover change in a post-mining area using GIS. Landscape Urban Plann. 87 , 22–32. 10.1016/j.landurbplan.2008.03.009 (2008). [ Google Scholar ] 6. Zheng, Z., Wu, Z., Chen, Y., Yang, Z. & Marinello, F. Exploration of eco-environment and urbanization changes in coastal zones: A case study in China over the past 20 years. Ecol. Indic. 119 , 106847. 10.1016/j.ecolind.2020.106847 (2020). [ Google Scholar ] 7. Fang, Y., Wang, J., Fu, S., Zhai, T. & Huang, L. Changes in ecological networks and eco-environmental effects on urban ecosystem in China’s typical urban agglomerations. Environ. Sci. Pollut. Res. 29 , 46991–47010. 10.1007/s11356-022-19226-7 (2022). [ DOI ] [ PubMed ] [ Google Scholar ] 8. Fang, C. & Liu, X. Comprehensive measurement for carrying capacity of resources and environment of city clusters in central China. Chin. Geogr. Sci. 20 , 281–288. 10.1007/s11769-010-0281-z (2010). [ Google Scholar ] 9. Zheng, Z. & He, Q. Spatio-temporal evaluation of the urban agglomeration expansion in the middle reaches of the Yangtze River and its impact on ecological lands. Sci. Total Environ. 790 , 148150. 10.1016/j.scitotenv.2021.148150 (2021). [ DOI ] [ PubMed ] [ Google Scholar ] 10. Chen, X., Li, F., Li, X., Hu, Y. & Hu, P. Quantifying the demographic distribution characteristics of ecological space quality to achieve urban agglomeration sustainability. Environ. Res. Lett. 16 , 094025. 10.1088/1748-9326/ac109b (2021). [ Google Scholar ] 11. Dai, L., Liu, Y. & Luo, X. Integrating the MCR and DOI models to construct an ecological security network for the urban agglomeration around Poyang Lake, China. Sci. Total Environ 754 , 141868. 10.1016/j.scitotenv.2020.141868 (2021). [ DOI ] [ PubMed ] [ Google Scholar ] 12. Cui, L., Wang, J., Sun, L. & Lv, C. Construction and optimization of green space ecological networks in urban fringe areas: A case study with the urban fringe area of Tongzhou district in Beijing. J. Clean Prod. 276 , 124266. 10.1016/j.jclepro.2020.124266 (2020). [ Google Scholar ] 13. Kong, F., Yin, H., Nakagoshi, N. & Zong, Y. Urban green space network development for biodiversity conservation: Identification based on graph theory and gravity modeling. Landsc. Urban Plan. 95 , 16–27. 10.1016/j.landurbplan.2009.11.001 (2010). [ Google Scholar ] 14. Staccione, A., Candiago, S. & Mysiak, J. Mapping a green infrastructure network: A framework for spatial connectivity applied in Northern Italy. Environ. Sci. Policy 131 , 57–67. 10.1016/j.envsci.2022.01.017 (2022). [ Google Scholar ] 15. An, Y., Liu, S., Sun, Y., Shi, F. & Beazley, R. Construction and optimization of an ecological network based on morphological spatial pattern analysis and circuit theory. Landsc. Ecol. 36 , 2059–2076. 10.1007/s10980-020-01027-3 (2021). [ Google Scholar ] 16. Huang, K., Peng, L., Wang, X. & Deng, W. Integrating circuit theory and landscape pattern index to identify and optimize ecological networks: A case study of the Sichuan Basin, China. Environ. Sci. Pollut. Res. 29 , 66874–66887. 10.1007/s11356-022-20383-y (2022). [ DOI ] [ PubMed ] [ Google Scholar ] 17. Etherington, T. R. Least-cost modelling and landscape ecology: Concepts, applications, and opportunities. Curr. Landsc. Ecol. Rep. 1 , 40–53. 10.1007/s40823-016-0006-9 (2016). [ Google Scholar ] 18. Salviano, I. R., Gardon, F. R. & dos Santos, R. F. Ecological corridors and landscape planning: A model to select priority areas for connectivity maintenance. Landsc. Ecol. 36 , 3311–3328. 10.1007/s10980-021-01305-8 (2021). [ Google Scholar ] 19. Li, L., Huang, X., Wu, D. & Yang, H. Construction of ecological security pattern adapting to future land use change in Pearl River Delta, China. Appl. Geogr. 154 , 102946. 10.1016/j.apgeog.2023.102946 (2023). [ Google Scholar ] 20. Jantz, C. A., Goetz, S. J. & Shelley, M. K. Using the SLEUTH urban growth model to simulate the impacts of future policy scenarios on urban land use in the Baltimore-Washington metropolitan area. Environ. Plan. B Plan. Des. 31 , 251–271. 10.1068/b298 (2004). [ Google Scholar ] 21. Chen, G. et al. Global projections of future urban land expansion under shared socioeconomic pathways. Nat. Commun. 11 , 537. 10.1038/s41467-020-14386-x (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Xie, J. et al. Impacts of landscape pattern on ecological network evolution in Changsha-Zhuzhou-Xiangtan Urban Agglomeration, China. Ecol. Indic. 145 , 109716. 10.1016/j.ecolind.2022.109716 (2022). [ Google Scholar ] 23. Batty, M., Couclelis, H. & Eichen, M. Urban systems as cellular automata. Environ. Plan. B-Plan. Des. 24 , 159–164. 10.1068/b240159 (1997). [ Google Scholar ] 24. Liu, X. et al. A future land use simulation model (FLUS) for simulating multiple land use scenarios by coupling human and natural effects. Landsc. Urban Plan. 168 , 94–116. 10.1016/j.landurbplan.2017.09.019 (2017). [ Google Scholar ] 25. Verburg, P. H., Eickhout, B. & Van Meijl, H. A multi-scale, multi-model approach for analyzing the future dynamics of European land use. Ann. Reg. Sci. 42 , 57–77. 10.1007/s00168-007-0136-4 (2008). [ Google Scholar ] 26. Liang, X. et al. Understanding the drivers of sustainable land expansion using a patch-generating land use simulation (PLUS) model: A case study in Wuhan, China. Comput. Environ. Urban Syst. 85 , 101569. 10.1016/j.compenvurbsys.2020.101569 (2021). [ Google Scholar ] 27. Zhang, S. et al. Landscape ecological risk projection based on the PLUS model under the localized shared socioeconomic pathways in the Fujian Delta region. Ecol. Indic. 136 , 108642. 10.1016/j.ecolind.2022.108642 (2022). [ Google Scholar ] 28. Zhang, S. et al. Land use/land cover prediction and analysis of the middle reaches of the Yangtze River under different scenarios. Sci. Total Environ. 833 , 155238. 10.1016/j.scitotenv.2022.155238 (2022). [ DOI ] [ PubMed ] [ Google Scholar ] 29. Liu, Y., Jing, Y. & Han, S. Multi-scenario simulation of land use/land cover change and water yield evaluation coupled with the GMOP-PLUS-InVEST model: A case study of the Nansi Lake Basin in China. Ecol. Indic. 155 , 110926. 10.1016/j.ecolind.2023.110926 (2023). [ Google Scholar ] 30. Wang, Z. et al. Dynamic simulation of land use change and assessment of carbon storage based on climate change scenarios at the city level: A case study of Bortala, China. Ecol. Indic. 134 , 108499. 10.1016/j.ecolind.2021.108499 (2022). [ Google Scholar ] 31. Zhang, J., Zhang, C., Dong, H., Zhang, L. & He, S. <article-title update=“added”>Spatial–temporal change analysis and multi-scenario simulation prediction of land-use carbon emissions in the Wuhan Urban Agglomeration, China. Sustainability 15 , 11021. 10.3390/su151411021 (2023). [ Google Scholar ] 32. Gao, L. et al. Multi-scenario simulation and ecological risk analysis of land use based on the PLUS model: A case study of Nanjing. Sust. Cities Soc. 85 , 104055. 10.1016/j.scs.2022.104055 (2022). [ Google Scholar ] 33. Guo, P., Wang, H., Qin, F., Miao, C. & Zhang, F. Coupled MOP and PLUS-SA model research on land use scenario simulations in Zhengzhou Metropolitan Area, central China. Remote Sens. 15 , 3762. 10.3390/rs15153762 (2023). [ Google Scholar ] 34. Li, Y. et al. Spatial-temporal evolution and prediction of carbon storage: An integrated framework based on the MOP-PLUS-InVEST model and an applied case study in Hangzhou, east China. Land 11 , 2213. 10.3390/land11122213 (2022). [ Google Scholar ] 35. Du, M., Liu, C., Wang, W. & Yang, H. Simulation of ecological risk in Beijing using Mop-Plus model. Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci. 48 , 15–22. 10.5194/isprs-archives-XLVIII-3-W2-2022-15-2022 (2022). [ Google Scholar ] 36. Han, J. et al. Spatio-temporal evolution and optimization analysis of ecosystem service value-A case study of coal resource-based city group in Shandong, China. J. Clean. Prod. 363 , 132602. 10.1016/j.jclepro.2022.132602 (2022). [ Google Scholar ] 37. Chen, S., Liu, X., Yang, L. & Zhu, Z. Variations in ecosystem service value and its driving factors in the Nanjing Metropolitan Area of China. Forests 14 , 113. 10.3390/f14010113 (2023). [ Google Scholar ] 38. Cao, J. et al. The identification and driving factor analysis of ecological-economi spatial conflict in Nanjing Metropolitan Area based on remote sensing Data. Remote Sens. 14 , 5864. 10.3390/rs14225864 (2022). [ Google Scholar ] 39. Song, S. et al. Functional regionalization of land resources considering eco-efficiency in Nanjing Metropolitan Area, China. Ecol. Indic. 155 , 110964. 10.1016/j.ecolind.2023.110964 (2023). [ Google Scholar ] 40. Liu, W., Li, H., Xu, H., Zhang, X. & Xie, Y. Spatiotemporal distribution and driving factors of regional green spaces during rapid urbanization in Nanjing metropolitan area, China. Ecol. Indic. 148 , 110058. 10.1016/j.ecolind.2023.110058 (2023). [ Google Scholar ] 41. Wenbo, X., Hengzhou, X., Xiaoyan, L., Hua, Q. & Ziyao, W. Ecosystem services response to future land use/cover change (LUCC) under multiple scenarios: A case study of the Beijing-Tianjin-Hebei (BTH) region, China. Technol. Forecast. Soc. Change 205 , 123525. 10.1016/j.techfore.2024.123525 (2024). [ Google Scholar ] 42. Tong, H., Dong, X., Liu, J. & Gao, W. Composite optimization of land use structure and layout in the middle and upper reaches of the Yellow River. Trans. Chin. Soc. Agric. Eng. 40 , 252–262. 10.11975/j.issn.1002-6819.202401220 (2024). [ Google Scholar ] 43. PRC National Development and Reform Commission. Nanjing Metropolitan Area Development Plan , https://zfxxgk.ndrc.gov.cn/web/iteminfo.jsp?id=17873 (2021). 44. Cao, R., Hu, Y. & Tang, B. Balancing economic development and forest ecosystem services: Multi-scale spatial land use optimization modeling for Guiyang, China. Ecol. Indic. 183 , 114616. 10.1016/j.ecolind.2026.114616 (2026). [ Google Scholar ] 45. Zhang, Z. et al. Simulating land use change for sustainable land management in rapid urbanization regions: A case study of the Yangtze River Delta region. Landsc. Ecol. 38 , 1807–1830. 10.1007/s10980-023-01657-3 (2023). [ Google Scholar ] 46. Men, D. & Pan, J. Ecological network identification and connectivity robustness evaluation in the Yellow River Basin under a multi-scenario simulation. Ecol. Modell. 482 , 110384. 10.1016/j.ecolmodel.2023.110384 (2023). [ Google Scholar ] 47. Liu, W., Xu, H., Zhang, X. & Jiang, W. Green infrastructure network identification at a regional scale: The case of Nanjing metropolitan area, China. Forests 13 , 735. 10.3390/f13050735 (2022). [ Google Scholar ] 48. Proctor, M. F. et al. Grizzly bear connectivity mapping in the Canada–United States trans-border region. J. Wildl. Manag. 79 , 544–558. 10.1002/jwmg.862 (2015). [ Google Scholar ] 49. Congalton, R. G. & Mead, R. A. A quantitative method to test for consistency and correctness in photointerpretation. Photogramm. Eng. Remote Sens. 49 , 69–74 (1983). [ Google Scholar ] 50. Pontius, R. G. et al. Comparing the input, output, and validation maps for several models of land change. Ann. Reg. Sci. 42 , 11–37. 10.1007/s00168-007-0138-2 (2008). [ Google Scholar ] 51. Shi, X. et al. A scenario-and spatial-downscaling-based land-use modeling framework to improve the projections of plausible futures: A case study of the Guangdong–Hong Kong–Macao Greater Bay Area, China. Sustain. Sci. 16 , 1977–1998. 10.1007/s11625-021-01011-z (2021). [ Google Scholar ] 52. Lai, J. & Qi, S. Construction of ecological security pattern based on ecological protection red line, machine learning algorithms, and circuit theory: A case study of Kunming city, China. J. Clean. Prod. 520 , 146117. 10.1016/j.jclepro.2025.146117 (2025). [ Google Scholar ] 53. Zhou, Y., Huang, C., Wu, T. & Zhang, M. A novel spatio-temporal cellular automata model coupling partitioning with CNN-LSTM to urban land change simulation. Ecol. Modell. 482 , 110394. 10.1016/j.ecolmodel.2023.110394 (2023). [ Google Scholar ] 54. Zhao, Y., Leng, H., Sun, P. & Yuan, Q. A spatial zoning model of municipal administrative areas based on major function-oriented zones. Sustainability 10 , 2976. 10.3390/su10092976 (2018). [ Google Scholar ] 55. Domingo, D., Palka, G. & Hersperger, A. M. Effect of zoning plans on urban land-use change: A multi-scenario simulation for supporting sustainable urban growth. Sust. Cities Soc. 69 , 102833. 10.1016/j.scs.2021.102833 (2021). [ Google Scholar ] 56. Xu, Q., Wang, Q., Liu, J. & Liang, H. Simulation of land-use changes using the partitioned ANN-CA model and considering the influence of land-use change frequency. ISPRS Int. J. Geo-Inf. 10 , 346. 10.3390/ijgi10050346 (2021). [ Google Scholar ] 57. Ke, X., Qi, L. & Zeng, C. A partitioned and asynchronous cellular automata model for urban growth simulation. Int. J. Geogr. Inf. Sci. 30 , 637–659. 10.1080/13658816.2015.1084510 (2016). [ Google Scholar ] 58. Artmann, M., Kohler, M., Meinel, G., Gan, J. & Ioja, I.-C. How smart growth and green infrastructure can mutually support each other—A conceptual framework for compact and green cities. Ecol. Indic. 96 , 10–22. 10.1016/j.ecolind.2017.07.001 (2019). [ Google Scholar ] 59. Al Rifat, S. A. & Liu, W. Predicting future urban growth scenarios and potential urban flood exposure using artificial neural network-markov chain model in Miami Metropolitan Area. Land Use Policy 114 , 105994. 10.1016/j.landusepol.2022.105994 (2022). [ Google Scholar ] 60. Wang, R., Derdouri, A. & Murayama, Y. Spatiotemporal simulation of future land use/cover change scenarios in the Tokyo metropolitan area. Sustainability 10 , 2056. 10.3390/su10062056 (2018). [ Google Scholar ] 61. Gao, J., Gong, J., Li, Y., Yang, J. & Liang, X. Ecological network assessment in dynamic landscapes: Multi-scenario simulation and conservation priority analysis. Land Use Policy 139 , 107059. 10.1016/j.landusepol.2024.107059 (2024). [ Google Scholar ] 62. Jia, Z. et al. Integrating ecological networks and multi-scenario optimization: A novel framework for constructing ecological security patterns. Agric. Ecosyst. Environ. 10.48130/aee-0025-0007 (2025). [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplementary Material 1 (71.9KB, rar) Data Availability Statement The data that support the findings of this study are available from the first author, [Wei Liu, Email: [email protected]], upon reasonable request. Articles from Scientific Reports are provided here courtesy of Nature Publishing Group ACTIONS View on publisher site PDF (7.9 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top