Objective Against the backdrop of accelerating global urbanization, the conservation and restoration of urban biodiversity have become critical interdisciplinary issues in landscape architecture, ecology, and urban and rural planning. Spontaneous vegetation, as a key component of urban flora, has garnered increasing attention due to its adaptability to urban environments, low maintenance requirements, and potential in preserving regional biodiversity. However, the distribution and functional realization of urban vegetation, especially spontaneous vegetation, are shaped by the combined effects of macro-climatic conditions, meso-scale urban environments, and micro-scale habitat characteristics. Existing research often suffers from scale fragmentation, wherein macro-scale studies focus on regional climatic gradients but seldom integrate urban design needs, meso-scale analyses address urban structural impacts but fail to link regional ecological contexts with site-specific designs, and micro-scale investigations reveal how microhabitat conditions influence species establishment but rarely attribute these effects to broader climatic or urban-scale drivers. This lack of multi-scale linkage hampers a systematic understanding of the mechanisms underlying urban spontaneous vegetation diversity and the development of scale-adapted application strategies. To address this gap, this study aims to investigate the driving mechanisms by which climatic differences affect the distribution patterns of urban spontaneous vegetation diversity across multiple scales, and establish a corresponding near-natural ecological restoration technical system, thereby addressing the prevalent scale mismatch issue in current urban biodiversity restoration practices.

Methods This study was conducted in Jilin Province, which exhibits a distinct humid−semi-humid−semi-arid macro-climatic gradient from southeast to northwest. The research framework encompasses three macro-scale climatic subregions, nine meso-scale prefecture-level cities, and five micro-scale habitat types within urban built-up areas. A total of 3,267 sample plots were surveyed using the Braun-Blanquet phytosociological approach during the complete growing season (May to September 2024). All spontaneously occurring vascular plant species (including naturally colonized herbs, escaped cultivated individuals, and regenerated seedlings of woody plants) within each 1 m×1 m quadrat were recorded, with their height and coverage measured. Data analyses focused on both alpha diversity (Patrick richness index, Shannon-Wiener diversity index, and Pielou evenness index) and beta diversity (based on Jaccard and Bray−Curtis distance matrices). Statistical comparisons were performed using Kruskal−Wallis tests with post-hoc Dunn tests for alpha diversity, and PERMANOVA with principal coordinate analysis (PCoA) visualization for beta diversity. All analyses were conducted in R 4.5.1, with a significance threshold of p＜0.05.

Results 1) A total of 605 spontaneous plant species belonging to 85 families and 342 genera were recorded across the nine cities. Asteraceae, Fabaceae, Rosaceae, and Poaceae were identified as the dominant families. Perennial herbs constituted the most abundant life form (41.8%), and native species accounted for 78.2% of all recorded species. 2) At the macro scale, alpha diversity showed a pattern of MIDDLE > WEST > EAST. Beta diversity analysis revealed highly significant differences among the three climatic subregions (p＜0.001), although the explanatory power of inter-group variation was low (R²<1.5%). PCoA ordination based on Jaccard and Bray−Curtis distances indicated that community differentiation followed different ecological gradients: species presence−absence patterns varied mainly along PC1 for the MIDDLE and EAST, while the WEST varied along PC2. 3) At the meso-scale, Changchun exhibited significantly higher alpha diversity than all other cities. Beta diversity patterns differed between species presence−absence and abundance-based dimensions: based on Jaccard distance, Baishan and Tonghua showed high heterogeneity in species composition, whereas Bray−Curtis-based PCoA revealed two nearly orthogonal nested gradients, indicating that community abundance structures varied along two relatively independent ecological dimensions. 4) At the micro-scale, UG had the highest Shannon-Wiener diversity and Pielou evenness indices, while SG had the lowest values. RG and RAG exhibited highly similar species diversity composition and community structure. Beta diversity analysis further confirmed significant differences among most habitat types (p＜0.001), though habitat type explained only a small fraction of total variation (R²<1.13%). Notably, RG and RAG showed extensive overlap in PCoA ordination, suggesting that intensive and homogeneous management may override microclimatic filtering, leading to community convergence.

Conclusion The diversity patterns of urban spontaneous vegetation in Jilin Province are driven by a cascade of multi-scale climatic and anthropogenic factors. At the macro scale, climatic gradients shape regional species pools and broad life-form distributions, though their direct explanatory power for inter-regional variation is limited. At the meso-scale, urban functional characteristics and localized climate modifications (e.g., heat island effects) interact to reshape diversity patterns, sometimes decoupling them from macro-climatic expectations. At the micro-scale, habitat-specific microclimates and management intensities act as strong filters, with high-intensity management leading to functional homogenization across otherwise distinct habitats. Based on these findings, a scale-adapted near-natural restoration strategy system is proposed. At the macro-scale, implementing differentiated plant selection and community design according to climatic subregion characteristics. At the meso-scale, tailoring restoration approaches to urban functional types and local climate adaptations, and at the micro-scale, applying precise regulatory measures based on habitat-specific conditions and functional needs. This integrated multi-scale framework provides a scientific basis and practical guidance for biodiversity conservation and climate resilient ecological restoration in cold-temperate cities. Future research should incorporate multi-season dynamic monitoring to further elucidate the temporal dynamics and long-term sustainability of spontaneous vegetation communities in urban ecosystems.