The lower layers of experimental constructed wetlands removed more than 99% of caffeine from stormwater.
An Australian study found that different pollutants were removed most effectively at different depths.
What researchers studied
Researchers at RMIT University investigated horizontal subsurface-flow constructed wetlands, a nature-based treatment approach for polluted runoff.
Their experiment compared planted and unplanted systems, including three treatment stages, under outdoor conditions.
The team monitored water quality at different depths over 191 days.
They examined ammonia, phosphorus, suspended solids and turbidity alongside caffeine, a chemical of emerging concern.
Their central question was whether a wetland functions as a uniform filter or contains layers with distinct treatment roles.
What they found
Caffeine removal exceeded 99% in bottom layers.
Surface layers favoured ammonia and phosphorus removal, while deeper layers performed better for suspended solids, turbidity and caffeine.
The study observed vertical differences in both vegetated and unvegetated configurations.
These patterns show why a single average pollutant-removal figure may conceal substantial differences within a treatment system.
They also indicate that vegetation is only part of the treatment process: the substrate and depth-dependent conditions deserve attention.
What it means
Stormwater engineers could benefit from sampling water at multiple depths, rather than assessing only inlet and outlet conditions.
Understanding where different pollutants are removed could inform maintenance, vegetation selection and the design of treatment systems tailored to local contamination.
Caffeine served as a useful example of an emerging contaminant, but its behaviour cannot be assumed to represent all pharmaceuticals or household chemicals.
The findings offer design guidance rather than a universal guarantee of pollutant removal.
Important limitation
The experiment involved a particular three-stage setup and synthetic stormwater under outdoor conditions.
Results cannot automatically be transferred to every climate, hydraulic load, wetland age or mixture of real-world pollutants.
Longer field trials and comparisons among operating wetlands would be needed before promising similar caffeine removal at municipal scale.
Source Information
Study: Vertical treatment efficiency of constructed wetlands: vegetation influence and caffeine removal
Authors: Harini Harish; Biplob Kumar Pramanik; Oliver A. H. Jones; Vincent Pettigrove; Veeriah Jegatheesan
Journal / Source: Frontiers in Environmental Science
Published: 9 November 2026
