Spatial-Temporal Evolution of GPP-LAI Coupling Degree in Urban Agglomeration from the Perspective of Ecological Network : A Case Study of Forest Urban Agglomeration in the Pearl River Delta

The Pearl River Delta (PRD) was constructed as China’s first "National Forest City Cluster" in 2020. While ecological restoration projects have significantly increased vegetation coverage, a critical question remains: has the increase in "green quantity" (Leaf Area Index, LAI) effectively translated into "carbon quality" (Gross Primary Productivity, GPP)? Existing studies have often focused on pixel-scale vertical processes, neglecting horizontal ecological flows within the landscape. This study introduced an Ecological Spatial Network (ESN) perspective to construct an analytical framework for carbon sequestration in urban ecosystems. The primary objective was to evaluate the spatiotemporal evolution of the degree of coupling between GPP and LAI from 2015 to 2020 and to quantify how network topological characteristics drive these functional changes, specifically differentiating between Ecological Construction Sources (artificial/restored areas) and Natural Development Sources (remnant natural patches). This study utilized multi-source data, including MODIS GPP, high-quality reprocessed LAI (HiQ-LAI), and meteorological data. First, Morphological Spatial Pattern Analysis (MSPA) and Minimum Cumulative Resistance (MCR) models were employed to identify ecological sources and extract corridors, constructing ESNs for 2015 and 2020. Complex network theory was applied to calculate topological indices (e.g., Betweenness Centrality, Modularity). Second, the degree of coupling between GPP and LAI was quantified using linear regression. Finally, a Random Forest (RF) algorithm combined with the SHapley Additive exPlanations (SHAP) model was used to quantitatively attribute the variations in GPP-LAI coupling to network topology and environmental factors. The findings reveal significant spatiotemporal heterogeneity: (1) Network Optimization: The PRD forest city construction significantly optimized the network structure. From 2015 to 2020, the number of nodes and corridors increased by more than 20%, and the average path length decreased by 18.6%. This indicates that the new small-scale patches acted as "stepping stones," enhancing the overall physical connectivity of the landscape. (2) Decoupling Phenomenon: Although regional LAI and GPP showed an overall increasing trend, a "quantity-quality decoupling" phenomenon was observed, particularly in ecological barrier and buffer zones. In these areas, rapid LAI growth did not synchronously enhance the GPP. (3) Source Heterogeneity: There is a distinct divergence between the source types. Ecological Construction Sources effectively improved GPP in low-vegetation areas but exhibited diminishing returns and stronger decoupling in high-greenness areas owing to immature community structures. In contrast, Natural Development Sources demonstrated a higher carbon sink potential and stability in the high-greenness zones. (4) Driving mechanisms:–The RFA-SHAP analysis identified distinct drivers for different sources. For Ecological Construction Sources, the degree of coupling is primarily constrained by precipitation (environmental stress) and network modularity (spatial isolation), characterizing a "physiologically limited" model. Conversely, Natural Development Sources are driven by GPP base levels and Closeness Centrality, characterizing a "topologically enhanced" model where central network positions facilitate stability and resource flow. This study confirms that simply increasing green quantity is insufficient for maximizing carbon sequestration in urban agglomerations. The effectiveness of carbon fixation is regulated nonlinearly by the topological structure of the ecological network. The study provides theoretical support for a differentiated optimization strategy: future ecological planning should prioritize reducing the modularity (fragmentation) of artificial construction sources by adding connecting corridors to break "ecological islands," while strictly protecting the central hub status of natural sources to maintain their functional stability. This network-based approach offers a precise scientific basis for transitioning urban ecological restoration from "increment-oriented" to "quality-oriented" management.