Impacts of Vegetation Management on Pollination Network Structure and Robustness in Urban Parks: Insights from Beijing Wenyu River Park

Research Hypotheses: 1. The near-natural management model can form more cohesive and robust pollination networks by promoting specialized interactions. 2. Although exotic plants in artificially planned areas can increase network nestedness by providing abundant resources, the lack of functional complementarity may make the network more vulnerable.data files：There are nine files in total, among which 'alien plant' is a list of alien cultivated and invasive plants in two locations; 'plant_insect_matrix' is a matrix of plant-insect interactions for each season, with insects as columns, plants as rows, and the intersections representing the number of interactions; 'Pollination Network' details the specific interactions of pollinating insects, including data on flowering plant cover, number of flowers, flower diameter, flower color, and other information.result：Results indicated that APM areas exhibited higher floral richness but lower community evenness and the dominance of alien species, resulting in diffuse, modular, and fragile network. In contrast, NNM areas, dominated by native flora, supported more diverse and specialized pollinators, with networks exhibiting higher nestedness, connectivity, and pollinator robustness. Mechanism analysis revealed that plant origin was the primary driver: NNM fostered native-specific specialization, while APM promoted resource-driven foraging. Although alien plants enhanced nestedness, their networks were less stable overall, highlighting limitations for long-term ecosystem integrity.数据采集：Pollinator data: Monthly surveys (Mar-Oct 2025) during favorable weather. Plant-pollinator interactions were recorded via visual census and sweep netting. Insect taxonomy, abundance, and visitation frequency were noted upon floral contact. Plants identified to species; insects released after field ID or capture for lab examination.Plant data: Entomophilous plants were recorded for taxonomy, cover, abundance, and floral traits (diameter, density). Floral units were defined as single flowers or, for dense inflorescences (e.g., Asteraceae), as entire functional units.Data Analysis:Analyses used R, ArcGIS, and Origin. Management regime differences were tested with t-tests or Wilcoxon tests. Community dissimilarity drivers were identified via SIMPER analysis. Dominant plants and pollinators were determined using Importance Value Index (IVI) and relative metrics (abundance/visitation), respectively.Beta regression modeled alien plant proportions; flower density used Negative Binomial GLM. Drivers of pollinator visitation were analyzed with Bayesian Negative Binomial GLMMs, with visitation frequency as the response, plant traits and origin as predictors, and plant species/site as random effects (MCMC estimation).