A Spatial Gradient Dose–Response Framework for Assessing Conservation Effectiveness in Mountainous Protected Area Networks: Evidence from Daxiangling Nature Reserve, China

Assessing conservation effectiveness in mountainous protected areas (PAs) is challenged by weak counterfactuals, spatial confounding, and limited historical baselines. This study develops a spatial gradient dose–response framework using linear mixed-effects models (LMMs) to evaluate conservation-associated ecological responses in Daxiangling Nature Reserve, China, controlling for topography, climate, and neighboring protected area effects. Using 1 km resolution time-series data (2003–2022), we assessed three indicators—Net Primary Productivity (NPP), Fractional Vegetation Cover (FVC), and a human-footprint-derived Ecosystem Integrity Index (EII)—each across state, trend, and stability dimensions. In the state dimension, NPP ( β = 0.99, p < 0.001) and EII ( β = 1.23, p < 0.001) were significantly higher inside the PA than in the peripheral reference zone, whereas FVC showed no significant difference, indicating indicator-specific sensitivity. NPP exhibited positive spillover to 15–20 km, while EII displayed a “boundary collapse” pattern—sharp decline within 0–5 km followed by distal recovery—suggesting edge effects and possible leakage. In the state dimension, high elevations amplified conservation effects on NPP and FVC. In the stability dimension, high elevations were associated with greater interannual variability across indicators, revealing a state–stability trade-off. Proximity to neighboring national PAs significantly enhanced EII across buffer zones, underscoring network connectivity. Cross-dimensional divergence was pronounced: NPP showed a state–stability trade-off in the PA, while EII exhibited decoupled state and trend signals across the boundary gradient. We recommend: (1) differentiated control in near-boundary zones (0–5 km); (2) prioritizing high-elevation core zones with stability monitoring; and (3) cross-PA patrols and ecological corridors. This framework offers a replicable, baseline-independent diagnostic approach for data-limited mountain systems where traditional counterfactual methods are difficult to apply.