Plastisphere microbes differ markedly from surrounding aquatic communities, meta-analysis shows

Every year, billions of plastic fragments drift through the world’s lakes, rivers, and oceans, and each one of them carries a microscopic passenger manifest. Scientists have known for roughly a decade that plastic particles in water are never bare surfaces: within hours, they acquire a coating of organic molecules, and within days they are colonized by dense microbial biofilms. This distinctive microbial habitat has been dubbed the plastisphere, and it has quietly become one of the most intensively studied ecological niches in environmental microbiology. Now, a new large-scale synthesis published in Current Research in Biotechnology has delivered the most statistically rigorous verdict yet on just how different these plastic-riding communities really are from the free-living microbes in the water around them. By re-analyzing raw genetic sequencing data from fifteen independent studies totaling 1,091 samples, researchers led by Qu-Yi Zhao and Bin-Yuan Gao have quantified, with unprecedented consistency, the magnitude of the microbial divide separating plastic surfaces from their watery surroundings.

The motivation for such a synthesis stems from the sheer scale of the problem. An estimated 7 billion metric tons of plastic waste have been produced globally since the mid-twentieth century, yet only about 9 percent has ever been recycled, leaving the overwhelming majority in landfills or dispersed into the environment. Microplastics, defined as particles smaller than 5 millimeters, are now effectively ubiquitous in aquatic systems, from freshwater lakes and rivers to coastal shelves and the open ocean. Beyond their well-documented physical and toxicological effects, these particles act as chemically and physically distinct interfaces that adsorb organic pollutants, heavy metals, and plastic additives, while simultaneously supporting microbial biofilm formation. Once colonized, these biofilms can alter the surface hydrophobicity, roughness, and sorption behavior of the particles, changing the environmental fate of both the plastics themselves and the contaminants they carry. Understanding whether the plastisphere’s microbes are simply a passive snapshot of the surrounding water, or a selectively filtered community in their own right, is therefore central to evaluating the ecological risks of microplastic pollution.

Earlier work laid important groundwork but left key questions unanswered. A previous meta-analysis of 2,229 samples from 35 studies showed that several potentially plastic-degrading microbial groups, notably the hydrocarbonoclastic orders Oceanospirillales and Alteromonadales, were consistently enriched in plastisphere communities relative to control biofilms, but that effort focused primarily on broad taxonomic patterns rather than paired, study-level comparisons of diversity and dispersal. A more recent global comparison of plastisphere and natural-habitat microbiomes across freshwater, seawater, and terrestrial systems confirmed that the plastisphere hosts compositionally distinct and highly heterogeneous communities, apparently shaped by relatively strong deterministic assembly processes and organized into loose, specialized ecological networks. What remained missing was a paired aquatic meta-analytical framework that could quantify, across independent studies, the effect sizes of microplastic colonization on multiple diversity indices, the consistency of enriched and depleted taxa when all raw data are processed identically, and the parameters governing microbial immigration into plastic biofilms.

The new study addresses exactly those gaps. The research team systematically searched the Web of Science Core Collection covering the period from 2000 to 2025, screening for studies that sampled both the plastisphere and the surrounding water, employed high-throughput sequencing of the 16S rRNA gene, deposited raw data in public repositories, and included more than 20 samples. Crucially, the authors imposed no restriction on polymer type, allowing polypropylene, polyethylene, and unspecified polymers alike, because their central question concerned the general effect of the plastisphere habitat rather than the properties of any single plastic. Studies were also retained regardless of whether the microplastics had been experimentally incubated for a known duration or retrieved from the field after an unknown residence time, an inclusive design that maximizes real-world relevance while introducing methodological heterogeneity that the statistical framework was built to handle.

Methodological standardization was the technical backbone of the analysis. Rather than pooling processed community tables from different papers, which risks conflating biological signals with pipeline artifacts, the team downloaded the raw FASTQ sequences for every eligible study and reprocessed them from scratch in USEARCH version 11. Primers were stripped, sequences were quality-filtered, and chimeric artifacts were detected and removed using the uchime3 algorithm. Amplicon sequence variants, or ASVs, the finest-resolution units of modern amplicon sequencing, were then inferred with the unoise3 command and assigned taxonomy against the RDP 16S rRNA gene reference database. To neutralize the distorting effect of unequal sequencing depth, each dataset was rarefied to the minimum depth within that study, with datasets falling below 10,000 sequences excluded entirely. Because different studies targeted different hypervariable regions of the 16S rRNA gene, ASVs were never merged at the sequence level; instead, all cross-study comparisons were performed at the genus level after independent taxonomic assignment. Fifteen studies survived this stringent gauntlet, yielding 1,091 paired samples for the final synthesis.

The statistical machinery then layered three complementary approaches. First, for each study, four alpha-diversity indices were computed from the rarefied ASV tables: observed richness, the Chao1 richness estimator, Shannon diversity, and Pielou’s evenness, which together capture the number of species, the robustness of richness estimates, the balance of abundances, and the evenness of the community. The effect of plastisphere colonization was expressed as the log response ratio, lnRR = ln(mean plastisphere / mean water), and synthesized across studies using random-effects meta-analysis with restricted maximum-likelihood estimation in the metafor package. Between-study heterogeneity was quantified with the I² statistic, which expresses the proportion of total variation attributable to genuine differences among studies rather than sampling error. Second, genus-level compositional differences were tested by calculating log2 fold changes between plastisphere and water for each genus, followed by Wilcoxon rank-sum tests with false-discovery-rate correction to control the false-positive inflation that arises from testing thousands of taxa simultaneously. Third, random forest classification, an ensemble machine-learning method, was deployed to screen for core discriminatory taxa capable of reliably distinguishing plastic-associated from water-column communities.

The third analytical pillar tackled the thorny question of community assembly. Whether the plastisphere is assembled passively, through random dispersal from the regional microbial pool, or deterministically, through selective filtering imposed by the plastic surface, has divided the field, with different studies reaching different conclusions. The team applied the neutral community model, a widely used theoretical framework in which the probability that a taxon occupies a given habitat depends on its regional abundance and the rate of dispersal between communities. Within this model, the immigration parameter, often denoted m, estimates the proportion of a community that arrives from the regional pool rather than being generated locally. The authors are careful to note that this parameter should be interpreted specifically as an estimate of immigration under neutral assumptions rather than as a direct measure of the overall balance between deterministic and stochastic forces, but comparing it between plastisphere and water communities across all fifteen studies nonetheless yields a valuable window on whether recruitment from the regional pool differs systematically between the two habitats.

The results, as the study’s title signals, reveal significant microbial community differentiation between the plastisphere and the surrounding water, and they do so with a level of cross-study consistency that individual investigations could never achieve on their own. Importantly, the findings support a nuanced picture rather than a simple one. The plastic habitat does not uniformly reduce microbial diversity; instead, the direction and magnitude of the diversity response proved to be context-dependent, shaped by local environmental conditions, polymer properties, exposure history, and experimental design. What does appear stable is the identity of the plastisphere’s characteristic taxa: specific genera were consistently enriched or depleted across independent aquatic datasets when the raw sequences were reprocessed through a single unified workflow, and the random forest classifier could identify core discriminatory taxa that reliably mark the plastic habitat. This combination of context-sensitive diversity and stable compositional signature is precisely what the authors hypothesized at the outset, and it reframes the plastisphere as a genuine ecological niche rather than a statistical echo of the water column.

The implications ripple outward in several directions. For ecologists, a well-defined set of discriminatory taxa provides a practical biomarker toolkit for detecting and monitoring plastisphere development across lakes, rivers, and coastal seas. For environmental chemists, the confirmation that biofilms modify surface properties and sorption behavior strengthens the case that microplastics function as active reaction platforms, coupling plastic ageing, contaminant transformation, antimicrobial resistance dissemination, and biogeochemical cycling on a single mobile interface. For risk assessors, quantified effect sizes and honest measures of between-study heterogeneity offer a defensible basis for comparing hazards across environments, while the immigration parameters help clarify how readily potentially harmful microbes, including pathogens, disperse onto floating plastic. The authors argue that by integrating effect-size meta-analysis, machine-learning classification, and neutral-model inference into a single quantitative framework, their approach distinguishes general plastisphere signals from study-specific noise, setting a methodological template for future syntheses. As plastic production shows no sign of slowing, the invisible ecosystems coating every discarded fragment are now measurable with a precision that may ultimately determine how, and how safely, humanity manages its plastic legacy in the water that sustains it.