Fractal Evolution and Associated Factors of Urban Green Space in Zhangzhou: A Multi-Scenario Land-Use Simulation Based on the PLUS Model

Loss, marginalization, and fragmentation of urban green space (UGS) under urbanization are well documented. Still, the pathway of loss, the nature of the pattern change, and the leverage of policy scenarios are seldom tested. Using 30 m land-use data for 2000–2023 in the central urban area of Zhangzhou, China, we tested three propositions with land-use transfer analysis, class-level landscape metrics, three fractal dimensions reported with their estimation uncertainty, and PLUS-model simulation of four 2030 scenarios. (1) Structural UGS fell by 40.1%; 93.0% of the loss went to cropland and only 6.1% directly to built-up land, whereas 90.6% of built-up gains came from cropland—an indirect displacement chain (built-up land → cropland → green space) rather than direct encroachment. (2) Patch number fell by 44% while mean patch area, patch density, and aggregation were stable or improved, and the largest patch index declined: a spatial reconfiguration, not fragmentation. (3) The box-counting dimension declined (1.556 → 1.488), the radial-distribution dimension rose across 2 (1.922 → 2.369), and the perimeter–area dimension remained near-Euclidean (1.31–1.36) and declined; green space became center-depleted and increasingly regular, the signature of planned form. (4) PLUS contribution values identified an elevation–temperature–woodland gradient and road proximity as spatial correlates of expansion location, not temporal drivers. (5) UGS declined by 1763–1892 ha under all scenarios; the 129 ha spread among scenarios and their statistically indistinguishable fractal structure show that conversion-probability controls only fine-tune an outcome fixed by projected land demand. Protecting UGS in such cities requires managing the cropland-displacement chain and land-demand targets, not conversion constraints alone.