Abstract: Agricultural landscapes are critical ecosystems for sustaining pollination services; however, 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 the Qiaosi Farm in Hangzhou, China, to evaluate 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, eight sampling events), documenting 376 individual bees across 15 species (four families), 73 individual butterflies from 6 species, and 132 syrphid flies. 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 that wildflower strips significantly enhanced the activity density of pollinating bees and syrphid flies but had no significant 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 such as Lasiloglossum mutilum and L. proximatum serving as wildflower strip indicators, whereas the exotic Apis mellifera ligustica was more associated with controls (Indval=0.364). The butterfly communities exhibited no structural differences (P=0.648). Co-occurrence network analysis demonstrated that wildflower strips increased pollinating bee network complexity (15 nodes, 80 edges, and density=0.762) compared to controls (13 nodes, 32 edges, and density=0.410), reflecting stronger species interactions and stability. In contrast, the butterfly networks remained poorly connected (two edges in the wildflower strips and zero in the controls). This study concluded that wildflower strips enhance pollinator biodiversity primarily by driving resource heterogeneity-mediated community turnover in bee populations, whereas 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 study provides a scientific foundation for balancing agricultural productivity with biodiversity conservation in human-modified landscapes, and offers actionable insights for sustainable agroecosystem management.