Abstract: Environmental DNA (eDNA) technology has emerged as a transformative, non-invasive, and highly sensitive analytical system, offering critical support for biological monitoring and invasive species management. Although stationary airborne eDNA sampling has been preliminarily applied in tropical regions, its limited detection range and spatial inflexibility hinder its ability to meet the large-scale and high-resolution monitoring requirements of expansive national parks. To address these technical bottlenecks and enhance monitoring efficiency and landscape-scale coverage, this study developed and validated an innovative vehicle-based mobile airborne eDNA collection system, specifically applying it to assess plant biodiversity dynamics and detect invasive species within the complex ecosystem of a tropical rainforest national park. Initial calibration experiments were conducted in a controlled campus environment with a well-documented plant inventory to determine the effective sampling range and sensitivity of traditional stationary collection towers. Subsequent field validation was performed in the Xian′an Stone Forest of Hainan Tropical Rainforest National Park, where a rigorous performance comparison was undertaken between the vehicle-mounted mobile system and stationary collectors deployed within the same spatial area. To capture temporal dynamics, mobile sampling campaigns were carried out during two distinct phenological periods: Labor Day (May) and National Day (October). Following high-throughput PE300 sequencing, the resulting metagenomic data were cross-referenced against global public databases and a dedicated Hainan regional invasive species database to elucidate community composition and species taxonomic diversity. The results indicated that the vehicle-mounted mobile system significantly outperformed the stationary collectors in both the total number of plant species detected (P < 0.05) and the capacity to identify invasive taxa (P < 0.001). Regarding species representation, mobile sampling successfully retrieved 96% of annotated species present in public databases and achieved a 100% detection rate for regionally known invasive species, confirming its excellent reliability and sensitivity. Temporal analysis revealed that the May sampling campaign detected 210 species (88 families and 140 genera), while the October campaign detected 167 species (72 families and 124 genera). Although community richness was significantly higher during the May period, potentially linked to seasonal peak flowering, no significant differences were observed in overall diversity indices between the two seasons. Furthermore, the system identified 58 and 54 invasive species during the two respective periods, demonstrating its precision in biosafety monitoring. In conclusion, this study confirms that vehicle-based mobile airborne eDNA technology represents an efficient and feasible approach for biodiversity monitoring in dense tropical rainforests. This method significantly enhances species detection rates and spatial sampling throughput, providing a reliable and scalable tool for constructing dynamic monitoring networks, as well as crucial data to support national park conservation management and proactive prevention of biological invasions.