Global surface-water microparticle carrier-load patterns across aquatic ecosystems: a spatial framework for prioritizing future plastisphere-associated antimicrobial resistance surveillance

Microplastic-associated biofilms can concentrate microorganisms, mobile genetic elements, and co-selective chemicals, but globally comparable measurements of plastisphere antimicrobial resistance (AMR) remain sparse. We therefore asked where surface-water particle carrier-load is sufficiently high to define candidate locations for paired, confirmatory plastisphere-AMR surveillance, without treating particle abundance as a proxy for AMR burden. We harmonized two worldwide surface-water datashares collected during 2013-2017. After removing three exact duplicate freshwater records and reserving 18 depth-profile samples for sensitivity analysis, 2,399 samples were classified as freshwater (n=1,006), estuarine (n=45), coastal (n=663), or open ocean (n=685). A two-part hurdle framework modeled particle detection and, among positive samples, log concentration, using 2-degree spatial-cell cluster-robust standard errors and adjustment for sampling year, platform, and container material. Operational surveillance-priority cells required at least three samples, at least 50% detection, and a median concentration at or above the ecosystem-specific 90th percentile among eligible cells. Detection increased from 49.5% in freshwater to 75.6% in estuaries, 79.3% at the coast, and 94.5% in the open ocean. Relative to freshwater, adjusted odds ratios for detection were 2.28 (95% CI 1.03-5.05), 2.20 (1.51-3.19), and 4.78 (2.62-8.75), respectively. The estuarine estimates should be interpreted cautiously because only 45 strict estuarine samples were available, and classification was based on text-derived source information. Among positive samples, adjusted geometric mean ratios were 1.45 (0.79-2.67), 1.40 (1.14-1.72), and 2.99 (2.33-3.84). Filaments constituted 82.7%-95.1% of classified particles. Twenty-eight operational hotspot cells were identified, although several open-ocean cells reflected concentrated campaign sampling. This analysis provides a transparent carrier-load layer for selecting candidate locations for future paired measurements of resistance genes, resistant organisms, antibiotics, metals, and mobile genetic elements. The identified cells are substrate-rich sampling priorities, not microbiologically verified AMR hotspots, and their value for AMR surveillance remains to be empirically validated.