Research report: Multi-scenario responses of ecological security networks in the Yangtze River Delta under a multi-model framework

A multi-model study of the Yangtze River Delta found that an ecological-protection scenario sharply constrained modeled urban expansion through 2035 and retained more forest, grassland, and water bodies, while barriers to ecological connectivity inherited from historically developed areas remained. The research compared business-as-usual, intensified urban development, and ecological-protection pathways by combining land-use simulation, landscape ecological risk assessment, morphological spatial pattern analysis, and circuit theory. The results indicate that regulating future land conversion can reduce ecological pressure, but the study also shows that effective connectivity would also require targeted action at persistent bottlenecks rather than reliance on protection alone.

The analysis used 2020 land use as its baseline and represented three alternative development patterns for 2035. The business-as-usual pathway continued land-use transitions observed from 2015 to 2020 without additional anthropogenic constraints. The urban-development pathway increased modeled transition probabilities from cropland, forest, and grassland to built-up land by 20% and incorporated constraints from designated development zones.

The ecological-protection pathway reduced all transition probabilities to built-up land by 30%, increased those to forest and grassland by 20%, and designated nature reserves as restricted-development areas. The authors treated the resulting maps as spatial representations of alternative pathways rather than deterministic forecasts because the simulations depend on historical patterns, selected driving factors, and policy parameters. Land-use maps were generated with the Patch-generating Land Use Simulation model.

Its land expansion analysis strategy was used to identify historical expansion of each land-use type, a random forest algorithm assessed the principal drivers of change, and a stochastic patch simulation represented the generation, expansion, and competition among land types. Inputs included land use, population density, gross domestic product, roads, rivers, railways, elevation, slope, mean annual temperature, annual precipitation, designated development zones, and nature reserves. All datasets were clipped, reprojected, reclassified, and resampled to a common 100-meter resolution.

Multi-period validation produced overall accuracies of 0.87 and 0.90 and Kappa coefficients of 0.81 and 0.85 for modeled 2018 and 2020 land use, respectively. These results indicate good agreement with the observed maps for those periods, but do not establish that the 2035 representations will occur as modeled. The three pathways generated markedly different modeled land-cover outcomes.

Under business as usual, built-up land increased by 5,080.43 square kilometers relative to 2020, while cropland declined by 4,417.07 square kilometers and forest by 428.55 square kilometers. Intensified urban development produced the largest built-up increase, 6,866.91 square kilometers, bringing the modeled total to 55,269.29 square kilometers; cropland decreased by 5,982.03 square kilometers. In contrast, the ecological-protection pathway allowed built-up land to increase by only 54.37 square kilometers.

It increased forest by 531.55 square kilometers, grassland by 411.07 square kilometers, and water bodies by 1,838.00 square kilometers. Built-up expansion in the two development-oriented pathways followed major transport and urban axes, including the Shanghai–Nanjing, Shanghai–Hangzhou, and Nanjing–Hangzhou corridors, as well as existing urban fringes and riverine or coastal plains. Landscape ecological risk remained spatially polarized across the 2020 baseline and all three modeled futures, with higher risk generally in the northeast and lower risk in the southwest.

Unsafe zones in 2020 were concentrated on the eastern coastal plain around Shanghai and in highly urbanized regions including the Suzhou–Wuxi–Changzhou and Nanjing–Zhenjiang–Yangzhou clusters. Safer areas occurred mainly in the mountains and hills of southwestern Zhejiang and southern and western Anhui. Under business as usual, the overall pattern changed only slightly, although some high-risk patches expanded along urban fringes.

Intensified development made relatively high-risk areas more prominent along major transport corridors and urban edges, whereas ecological protection left low-risk areas more continuous in the southwestern mountains and other areas with concentrated ecological land. The assessment combined fragmentation, separation, dominance, disturbance, and vulnerability measures across 3,877 units on a 10-kilometer grid, then classified them from safe to unsafe. Ecological source areas were selected from core patches of forest, grassland, and water rather than identified by size alone.

The researchers combined morphological spatial pattern analysis with the Integral Index of Connectivity, Probability of Connectivity, and delta Probability of Connectivity. The last index estimates a patch’s contribution to the wider network by comparing connectivity with the patch present and with it removed. The 100 patches with the highest delta Probability of Connectivity were retained as source areas under each scenario, and approximately 30 in each set were classified as important.

These areas were concentrated principally in the southwestern mountains, with additional strategic patches in mountain transition zones and river–lake junctions. Some relatively small and medium-sized patches occupied positions between larger sources and made substantial contributions to east–west and north–south connections. The analysis therefore identified them as potential stepping stones whose network position could matter more than their area alone.

Circuit theory was then used to estimate potential movement across a composite resistance surface containing land use, landscape ecological risk, elevation, and slope. Land use carried the largest weight at 0.5, ecological risk carried 0.2, and elevation and slope carried 0.15 each. Low-resistance values were assigned to ecologically strong or relatively safe areas, forests, water bodies, and gentle slopes, while built-up land, high-risk areas, and steep slopes received higher values.

The resulting network contained 194 modeled corridors in 2020, compared with 193 under business as usual, 175 under urban development, and 188 under ecological protection. Total corridor length ranged from 58,192.25 kilometers in the ecological-protection scenario to 62,788.76 kilometers in 2020. These figures describe the extent of the modeled network rather than its ecological effectiveness.

Corridors also cannot be ranked by count alone: business as usual had 136 important corridors, whereas urban development and ecological protection each had 113. The analysis further located where potential ecological flows concentrate and where reducing resistance could yield comparatively large connectivity gains. The 2020 network contained 89 pinch points and 42 barrier points.

These rose to 111 and 43 under business as usual, 93 and 47 under urban development, and 95 and 48 under ecological protection. Pinch points marked narrow connections, corridor intersections, and other bottlenecks, especially along intercity links such as Nanjing–Zhenjiang–Yangzhou and Wuhu–Ma’anshan. Barrier points occurred more often in densely built landscapes affected by major transport obstructions, including Shanghai, the core cities of southern Jiangsu, and the coastal urban belt.

They were modeled locations where local resistance reduction could provide relatively large improvements in connectivity, not observed barriers confirmed through tracking of individual species. Cross-scenario overlays also separated recurrent features, which appeared under multiple pathways, from scenario-sensitive locations whose relevance depended more strongly on a particular land-use pattern. Ecological protection therefore produced a clear trade-off.

It constrained new built-up conversion and increased the modeled extent of several ecological land types, yet it did not markedly reduce the number of barrier points. The authors linked this outcome to resistance inherited from historically developed land and existing transport infrastructure, because the scenario chiefly constrained future conversion rather than removing established development. Some ecological patches could also remain isolated or positioned at abrupt interfaces among protected land, cropland, and existing built-up surfaces, creating local edge effects and fragmentation even when broader ecological land remained continuous.

The authors propose a network-based management hierarchy: strict protection of source areas, maintenance of priority corridors, stringent regulation around pinch points, and targeted restoration of barrier points. Suggested measures included wildlife passages, ecological bridges, culverts, roadside buffer forests, riparian-wetland restoration, and urban green wedges. These are planning implications derived from the model, not interventions tested in the study.

The network was not validated with field observations of individual species or long-term biodiversity monitoring, and source areas were selected from structural connectivity without explicitly measuring habitat quality. Circuit theory estimates random-walk connectivity probabilities rather than actual movement routes, while alternative resistance weights, policy adjustments, major infrastructure projects, or extreme climatic events could alter the results. The study therefore presents a model-based framework for identifying regional planning priorities, not a confirmed map of species-specific conservation pathways.

Subject of Research: Modeled ecological connectivity under alternative land-use pathways in the Yangtze River Delta through 2035

Article Title: Multi-scenario responses of ecological security networks in the Yangtze River Delta under a multi-model framework

Article References: Wei, H., Li, X., Tao, J., He, L., & Zhu, H. (2026). Multi-scenario responses of ecological security networks in the Yangtze River Delta under a multi-model framework. Frontiers in Ecology and Evolution, 14, Article 1977267. https://doi.org/10.3389/fevo.2026.1977267

Image Credits: AI Generated

DOI: 10.3389/fevo.2026.1977267

Keywords: Yangtze River Delta, ecological connectivity, land-use scenarios, urban expansion, landscape ecological risk, circuit theory, PLUS model, ecological security networks, habitat corridors, ecological restoration, Multi-scenario, responses