ABSTRACT Urbanization affects vegetation cover primarily through altering landscape patterns. Understanding how vegetation growth responds to urban landscape patterns is therefore crucial for sustainable urban planning. However, previous studies have neither systematically revealed the non‐linear relationship between urban landscape patterns and vegetation growth, nor adequately explored the long‐term dynamics of landscape patterns and their spatiotemporally non‐stationary effects on vegetation during urbanization. In addition, spatial heterogeneity in natural conditions may cause traditional vegetation cover degree (VCD) to mask the effects of urbanization on vegetation growth. This study integrated multi‐source remote sensing and geospatial data across 19 urban agglomerations (UAs) in China to develop a similar‐habitat‐based model for calculating the vegetation cover potential achievement degree (VCPAD), thereby effectively mitigating the influence of natural conditions. Furthermore, the eXtreme Gradient Boosting model and the Geographically and Temporally Weighted Regression model were employed to investigate the non‐linear and spatiotemporally heterogeneous effects of urban landscape pattern on vegetation growth. Over the past two decades, both the proposed VCPAD and VCD in UAs exhibited award trends, increasing by 18.97% and 18.18% respectively. However, vegetation growth dynamics varied markedly across regions. Aggregation‐related (COHESION and AI) and area‐related (LPI) urban landscape pattern metrics demonstrated marked non‐linear effects on vegetation growth. Specifically, LPI maintained a negative yet weakening influence. In contrast, COHESION and AI exhibited unimodal effects, indicating optimal threshold values for urban landscape pattern. Furthermore, VCPAD displayed greater sensitivity to urban landscape pattern than VCD, particularly in northern UAs compared to southern UAs. These findings can provide a reference for incorporating ecological planning into sustainable urban development strategies.
Cui et al. (Sun,) studied this question.