Abstract: Agricultural landscapes are critical ecosystems for sustaining pollination services, yet they face severe threats from urbanization and agricultural intensification, which have led to pronounced declines in pollinator biodiversity and ecosystem functions. To address this, a field experiment was conducted at Qiaosi Farm in Hangzhou, China, focusing on evaluating the effects of wildflower strip construction on pollinator communities and their co-occurrence networks. The experimental design included four replicate wildflower strips (1.5 m width × 80 m length) planted with a mixed assemblage of 10 wildflower species—selected for long flowering periods (e.g., Salvia farinacea from May to October), functional diversity (Lamiaceae, Apiaceae, Fabaceae), and ecological safety (e.g., native Viola philippica)—and four natural grass strip controls. Pollinator monitoring from June to August 2023 utilized standardized pitfall pan-trapping (yellow/white/blue traps, 8 sampling events), documenting 376 bee individuals across 15 species (4 families), 73 butterfly individuals from 6 species, and 132 hoverflies. Vegetation surveys in 1 m² quadrats (24 total) quantified plant coverage and species composition, revealing higher diversity and coverage in wildflower strips (e.g., Echium vulgare at 63.89% ± 1.20% coverage) compared to controls dominated by Rumex dentatus and Lactuca indica. Statistical analyses showed wildflower strips significantly enhanced activity density of pollinating bees (P=0.0097) and hoverflies (P=0.0015) but had no effect on butterfly species richness or density. Principal coordinate analysis (PCoA) with permutational multivariate analysis (PerMANOVA) revealed distinct pollinating bee community structures between treatments (P=0.05), with native bees like Lasiloglossum mutilum and L. proximatum serving as wildflower strip indicators, while the exotic Apis mellifera ligustica was more associated with controls (Indval=0.364). Butterfly communities showed no structural differentiation (p=0.648). Co-occurrence network analysis demonstrated wildflower strips increased pollinating bee network complexity (15 nodes, 80 edges, density=0.762) versus controls (13 nodes, 32 edges, density=0.410), reflecting stronger species interactions and stability. In contrast, butterfly networks remained poorly connected (2 edges in wildflower strips, 0 in controls). The study concludes that wildflower strips enhance pollinator biodiversity primarily by driving resource heterogeneity–mediated community turnover in bee populations, while butterfly responses are constrained by larval host specificity and landscape fragmentation. These findings highlight the importance of integrating functional plant groups (nectar sources and larval hosts) in wildflower strip design and advocate for long-term, landscape-scale monitoring to optimize agroecological restoration. By linking vegetation characteristics to pollinator dynamics, this research provides a scientific foundation for balancing agricultural productivity with biodiversity conservation in human-modified landscapes, offering actionable insights for sustainable agroecosystem management.