Polishing wastewater effluent using plants: floating plants perform better than submerged plants in both nutrient removal and reduction of greenhouse gas emission

Wastewater treatment plants (WWTPs) can account for up to 45% of the total nutrient loading in surface waters (e.g. Groenendijk et al. 2016). Because nutrient levels in domestic wastewater treatment effluents are relatively high (330–700 μmol/l N; 18–50 μmol/l P; Carey & Migliaccio 2009; CBS 2021), WWTPs contribute substantially to eutrophication of natural waterbodies (Carey & Migliaccio 2009), and furthermore contribute to high greenhouse gas (GHG) emissions from these waterbodies (Beaulieu et al. 2019). Reducing WWTP-derived nutrient loads can therefore reduce GHG emissions in receiving waterbodies. GHG emissions are set to be reduced by 55% in 2030 (European Commission 2021) and regional water authorities and wastewater managers can have a substantial role in this reduction.

Since the 1990s, the concept of ecological water treatment using aquatic plants (macrophytes) has gained interest (Wang 1987), but despite its many benefits, it is not yet widely used. Moreover, the focus has been on effluent polishing through nutrient uptake, but not yet on low GHG emission polishing. Next to constructed wetlands, a relatively new treatment of effluent has now started to gain interest, in which a closed system without a sediment layer is used for macrophyte growth, and effluent treatment takes place in a hydroponic way (e.g. Magwaza et al. 2020). By using the self-purification principle of natural waterbodies – the uptake and transformation of nutrients mediated by aquatic plants – effluent can be treated to reach nutrient concentrations below the critical values set by the European Water Framework Directive (WFD) (Norström et al. 2004).

Aquatic plants mediate nutrient removal both directly and indirectly. Directly, plants can extract inorganic phosphorus (P) and nitrogen (N) from wastewater, incorporating them into their biomass, and thus enabling N, P and carbon (C) harvesting and reuse (e.g. Norström et al. 2004). Indirectly, they alter conditions in water and sediment. For example, they alter oxygen concentrations and provide a surface for biofilm formation, thereby favouring coupled nitrification-denitrification and altering the production and emission of CH₄ (Danhorn & Fuqua 2007; Veraart et al. 2011; Law et al. 2012). During nitrification and denitrification, N₂O can be formed (Law et al. 2012), and aquatic plants have multiple ways in which they directly and indirectly affect this emission. At the same time, their photosynthesis removes CO₂ from the atmosphere or water layer while fixing C in their biomass.

Different aquatic plant growth forms have their own characteristics in removing nutrients and altering GHG emissions (Attermeyer et al. 2016; Christiansen et al. 2016). Submerged plant species can provide a large surface for epiphytic biofilm formation, altering microbial processes in these biogeochemically heterogeneous sites (Eriksson & Weisner 1999). At the same time, submerged macrophytes may inhibit denitrification by their oxygen leakage and by competing for nitrate with denitrifying bacteria (Toet et al. 2003). Floating plants, on the other hand, can form a dense mat on top of the water column, creating a reaeration barrier. Local conditions determine whether this barrier favours oxygen depletion or oxygen trapping. Although lower oxygen concentrations under such mats induce higher denitrification rates and CH₄ production (Veraart et al. 2011), the O₂ trapped under the macrophytes through radial oxygen loss (ROL) may enhance nitrification and CH₄ oxidation, making the outcome in terms of nutrient removal and GHG emission system specific (Kosten et al. 2016). Since floating plants can only cover the top layer of the water column, their growth easily becomes space-limited. Consequently, N and P uptake may stall when floating plants achieve full coverage (Si et al. 2019). Additionally, reduced surface for epiphytic biofilm formation potentially lowers the potential for microbial nutrient conversions, especially in systems without a sediment layer. Lastly, the floating fern Azolla filiculoides has a symbiosis with N-fixing microorganisms, which makes them less efficient in removing N, but highly efficient in removing P, because their growth does not stall once N is depleted in the water column (Brouwer et al. 2018).

The goal of this study was to explore the nutrient removal efficiency of two different macrophyte growth forms, floating vs. submerged, and their potential to lower GHG emissions when grown on WWTP effluent. We compared floating plants covering only the water column surface with submerged plants filling the entire water column. We expected that floating plants would stimulate denitrification, because they lower O₂ concentrations in the water column, and that submerged plants can stimulate nitrification because of their O₂ release. We expected the highest nutrient removal in systems with submerged plants because of their high uptake combined with a large surface area for biofilm formation. In addition, we expected that CO₂ uptake by photosynthesis would fully compensate for CO₂ production from respiration of organic carbon present in the wastewater effluent, leading to net CO₂ uptake. N₂O emission was expected during both nitrification and denitrification, where we expected highest emissions from systems covered by floating plants due to higher denitrification rates. Lastly, CH₄ emission was expected to be low in all cases, because of the lack of strictly anoxic habitats.

We quantified nutrient removal and GHG emissions of two floating plant species (Azolla filiculoides (hereafter: Azolla) and Lemna minor (Lemna)) and two submerged species (Ceratophyllum demersum (Ceratophyllum) and Callitriche platycarpa (Callitriche)). Additionally, control of effluent without plants was included, resulting in a total of five experimental treatments, each consisting of four replicates. The experiment was performed at the Radboud University greenhouse facility, in glass aquaria of 24 × 24 × 30 cm, distributed in a randomised block design to avoid confounding microclimatic effects in the greenhouse. We maintained a light/dark cycle of 16 h/8 h, by using 400 W high-pressure sodium lamps (Hortilux-Schréder, Monster, The Netherlands), which turned on when the natural daylight intensity fell below 250 W/m².

Wastewater effluent originated from the municipal wastewater treatment plant in Remmerden, the Netherlands, which has a 2,100 m³/h hydraulic capacity and serves 46,000 households. It is a UCT (University of Cape Town) carrousel (Østgaard et al. 1997) which had the following effluent concentrations in 2021, ranging between 80 and 500 μmol/l ⁠; 20 and 220 μmol/l ⁠; 1 and 30 μmol/l ⁠; 17 and 56 mg/l chemical oxygen demand (COD) and 1.9 and 8.9 mg/l biological oxygen demand (BOD₅) (Hoogheemraadschap de Stichtse Rijnlanden, WWTP Rhenen).

At the start of the experiment, we added 15 l of domestic wastewater effluent to each aquarium and introduced the assigned plant species to this effluent. Because of their different growth strategies and different morphological traits, floating plants were introduced to a surface area coverage of 25%, whereas submerged plants started at 25% volume in the water column. For each of the treatments, the wet weight of this 25% cover was determined, and an extra batch of plants was used to obtain the wet to dry ratio, to estimate initial dry biomass.

In each aquarium, we monitored nutrient concentrations and GHG emissions for 14 days, as well as physical–chemical properties of the water (temperature, pH, dissolved O₂). We measured three times on the first day, once a day during days 2–5, and every other day in the remaining period. On the last day, all plants were harvested to determine wet and dry biomass and plant nutrient content. We additionally harvested the filamentous green algae that started to grow in some of the treatments.

Concentrations of ⁠, and were measured colorimetrically in rhizon-filtered samples (membrane pore size 0.12/0.18 μm, Rhizon SMS 10 cm, Rhizosphere Research, Wageningen, The Netherlands) on an auto analyser III (Bran and Luebbe GmbH, Norderstedt, Germany) after being stored at −20 °C. Total phosphorus was measured in acidified water (0.1 ml 10% nitric-acid) on an ICP-OES (IRIS Interpid II, Thermo Fisher Scientific, Franklin, MA, USA) after being stored at 4 °C. Total inorganic carbon (TIC) was measured in unfiltered samples (ABB Advance optima Infrared Gas Analyzer (IRGA), Frankfurt, Germany) immediately after sample collection. The pH, temperature (°C) and dissolved O₂ (mg/l) concentrations in the water column of each aquarium were measured using a Portable Multi Meter (HQ2200, HACH, Loveland, CO, USA).

The plants that were harvested at the end of the experiment as well as the extra batch of each plant species at the beginning of the experiment were dried at 70 °C for 4 days, after which they were ground manually. The same was done for the filamentous algae that were collected on the last day. N and C contents were determined in plant material (3 mg) using an elemental CNS analyser (Vario Micro Cube, Elementar, Langenselbold, Germany). P content was determined on the ICP-OES after microwave digestion, adding 4 ml HNO₃ (65%) and 1 ml H₂O₂ (35%) to 200 mg dried plant material in Teflon vessels, followed by heating in an EthosD microwave (Milestone, Sorisole Lombardy, Italy).

GHG fluxes (CO₂, CH₄, N₂O) were measured using a Greenhouse Gas Analyser (G2508, Picarro, Santa Clara, CA, USA) connected to a transparent acrylic glass floating chamber (7.1 dm³ headspace). In each aquarium, we measured diffusive fluxes of CO₂, CH₄ and N₂O over a period of 4 min, counted from when concentrations started to change. In between the measurements, the chamber was aerated until gas concentrations returned to atmospheric levels.

Total dissolved inorganic N (TDIN) was obtained by summing and ⁠. Total dissolved P (TDP) concentrations were obtained from elemental ICP analysis of the filtered water samples (μmol/l).

GHG fluxes (mg/m²/day) were calculated according to Almeida et al. (2016). A global warming potential of 29.8 for CH₄ and 273 for N₂O was used (100-year time frame; IPCC 2021) to convert fluxes to CO₂ equivalents (g CO₂-eq/m³/day).

For element stocks (C, N and P), the plant content was multiplied by the dry weight of the plants. The total uptake of C, N and P (in μmol) was then obtained by subtracting the total mass of each element at the end of the experiment from the total mass at the start.

Plant growth was calculated by the difference in dry weight between the start and end of the experiment. The dry weight of filamentous algae harvested on the last experimental day was added to the plant growth data. Differences in plant growth and C, N and P plant uptake between treatments were analysed using ANOVA with a Tukey post hoc test (R 4.1.1 (R Core Team 2021), stats::aov; multcompView::TukeyHSD (Graves et al. 2019)).

The efficiency of N and P removal by plant uptake for the different plant species was calculated from the change in plant N and P content compared to dissolved inorganic N and P uptake from the water column. Efficiency was shown as a percentage, in which 100% indicated a complete removal due to plant uptake. A negative percentage showed a net release of N or P to the water column.

Dissolved O₂ concentrations and pH were stable (4–6 mg/l and 7.3, respectively) until day 4 when filamentous algae started to appear (Fig. S1). After this, pH rose to 8.5 for the floating plants, 8.7 for Callitriche and 9.5 for the Ceratophyllum and the control treatment. Dissolved O₂ concentrations increased as well and ended at concentrations of 8–9 mg/l for the floating plants, 10 mg/l for Callitriche, and 13–15 mg/l for Ceratophyllum and the control treatment.

concentrations were below 0.3 μmol/l after 8 days for Lemna, Azolla and the control treatment (Figure 1(c)). Ceratophyllum treatments started with higher concentrations, yet after 10 days all was removed. Treatment with Callitriche resulted in increase in the first 3 days, followed by uptake. However, after 2 weeks still, a considerable amount of (average 2.3 μmol/l) was present. Total dissolved phosphorus (TDP) concentrations were lowered from 9.0 (±1.3 sd.) to 3.3 (±0.8 sd.) μmol/l in treatments with Azolla, Lemna and Ceratophyllum, while Callitriche initially showed an increase and later a decrease in TDP concentration, but plateaued around 7.8 (±4.7 sd.) μmol/l (similar as the start concentration) after 2 weeks (Figure 3(b)).

GHG flux measurements started on day 3 (after 49 h). Fluxes of CH₄ were low in all treatments (max. CH₄ flux 0.04 mg/m²/day), and from day 4 onwards no CH₄ fluxes were observed (Figure 2(a)). At the start of the measurements, only Lemna and Azolla were taking up CO₂ and had highest CO₂ uptake during the whole experiment (Figure 2(b)). Control treatments had similar CO₂ uptake as Ceratophyllum and Callitriche. CO₂ uptake only took place after day 4 for these treatments, which was at the onset of algal growth. N₂O emissions were low overall, with only a small peak after day 4 for Callitriche and the control treatment (Figure 2(c)). Lemna and Azolla showed a net GHG uptake, having negative fluxes in CO₂ equivalents, while the other three treatments first emitted GHGs, followed by net uptake (Figure 3(c)).

The plants differed significantly in how well they grew on wastewater effluent (P < 0.001; Figure 3(d)). Azolla and Lemna had the highest biomass increase, although Lemna did not differ significantly from Ceratophyllum (P = 0.06). The Callitriche treatment had little to no growth, with algae accounting for 46 (11–75)% of its total biomass gain. In control treatment aquaria, an increase in algal biomass was observed as well (up to 0.15 g).

N and P sequestration by the different plant species corresponded with N and P removal from the water column for the floating plants, whereas N and P removal in the submerged macrophyte treatments can only for a small part be explained by plant uptake (Table 1).

In this study, we tested the nutrient removal efficiency of floating and submerged macrophytes grown on WWTP effluent, and their potential to capture CO₂ and suppress CH₄ and N₂O emission. We compared the effects of two floating plants, Azolla filiculoides and Lemna minor, and two submerged macrophytes, Ceratophyllum demersum and Callitriche platycarpa. In this experiment, systems covered by the floating plants Azolla or Lemna, had the highest N and P removal efficiency – resulting from plant uptake – and captured most CO₂ while emitting the least CH₄ and N₂O, thus resulting in net GHG uptake. Submerged plants Ceratophyllum and Callitriche did not grow well on WWTP effluent, and therefore contributed less to both nutrient removal and CO₂ uptake.

All treatments, including unvegetated controls, caused TDIN concentrations to decrease to on average ≈130 μmol/l, which is well below the average concentrations observed in the waterbodies to which WWTPs discharge their effluent (≈285 μmol/l; Carey & Migliaccio 2009; Puijenbroek et al. 2010). However, coverage by Lemna caused the largest decrease, resulting in almost complete TDIN removal after 2 weeks (Figure 2(a)).

For water treated with Azolla, Lemna and Ceratophyllum, P concentrations were reduced to on average 3.3 (±0.8 sd.) μmol/l P after 2 weeks, which is similar to P concentrations occurring in the potential receiving waterbodies (Carey & Migliaccio 2009; Puijenbroek et al. 2010). In both floating plant treatments, plant P uptake resulted in immediate P removal from the water column, whereas Ceratophyllum P uptake could not explain all P removal from the water column. concentrations increased in systems with Callitriche, likely due to plant senescence, observed from its lack of growth and visible signs of decay.

While submerged macrophytes were hampered in their growth by algal dominance, and presumably also by the high pH leading to very low CO₂ concentrations in the water layer, floating macrophytes showed high growth rates of 4.9 (±1.2 sd.) and 3.3 (±1.0 sd.) g/m²/day for Azolla and Lemna, respectively. This is in line with, and for Azolla even in the high range of, maximum growth rates found for these species (Reddy & DeBusk 1985).

After 6 days, all treatments resulted in net GHG uptake, with systems covered by Azolla or Lemna showing the highest uptake (Figure 2(c)). In treatments containing Ceratophyllum and Callitriche, CO₂ uptake only took place after 4 days, similar to the control treatment and starting at the moment filamentous algae became visible. Combined with the poor growth of these submerged plant species, we expect at least part of the CO₂ uptake to be due to algal growth rather than macrophyte growth. Little to no CH₄ emission was detected in all treatments, which can be explained by the high O₂ concentrations in the water and lack of sediment. A small peak in N₂O emission occurred at the time when concentrations were at its highest and was depleted. Yet, the highest emission of 1.88 mg N₂O/m²/day (occurring in Callitriche treatments) was still well below emissions observed in constructed wetlands, which can reach 3.12 mg N₂O/m²/day (e.g. Mander et al. 2014).

After 4–5 days, in all plant treatments all was removed. It was expected that due to their larger surface area and thus expected higher biofilm production, submerged plants would facilitate a higher removal, which was not the case. Because similar removal rates were found for the control treatment, in the absence of algae, the removal is most likely caused by nitrification performed by microorganisms in the water column and in biofilms on the aquaria walls, rather than by plant uptake. The coincidence with an increase in in these first days confirms this. This is in line with other hydroponic systems, in which nitrification was also the predominant process of removal (Vaillant et al. 2003).

Our calculations show that all removal from aquaria treated with Lemna as well as those with Azolla can be explained by plant N uptake (Figure 5), which is contrasting to other studies where Lemna and Azolla species only take up a fraction of (Singh et al. 1992). In our systems, denitrification was not significantly contributing to N removal from the effluent. Moreover, Azolla coverage resulted in higher plant N uptake than N removal from the effluent, which indicates N₂ fixation from the atmosphere by the Azolla–Nostoc symbiosis.

Whereas Lemna had up to 100% (and thus TDIN) removal after 2 weeks, Azolla hardly removed any of the produced ⁠. This is most likely because of its symbiosis with the cyanobacterium Nostoc azollae that fixates nitrogen from the atmosphere (Brouwer et al. 2018). Normally, N fixation is a costly process which only takes place when N is limited. Yet, it is found that N fixation by the microbial symbiont occurs even when Azolla is grown on water containing substantial amounts of inorganic N, and N fixation is only inhibited by much higher concentrations of nitrogen than present in our experiment (Ito & Watanabe 1983). Azolla showed the highest N plant uptake (Figure 5) combined with the lowest TDIN removal, suggesting that almost all N that Azolla took up was derived from N-fixation from the atmosphere.

In treatments containing submerged plants, as well as the unvegetated controls, algae started to appear after 4 days, which was facilitated by the abundance of light and nutrients in these treatments. Likely, light-limitation suppressed algal growth in the floating plant treatments. As a result, N and P uptake by submerged plants was negligible (Figure 5 and Table 1). Most likely, in these treatments, N removal took place via algal uptake and coupled nitrification-denitrification by the microbial community, while P removal was mostly caused by algal uptake, especially in the Callitriche treatments.

Our systems including Azolla and Lemna were more efficient in the removal of N and P than other hydroponic systems (Shah et al. 2014) as well as constructed wetlands (Tang et al. 2017; Hernández et al. 2018), and are performing better than, or similar to floating treatment wetlands (Prajapati et al. 2017). In line with these findings, floating macrophytes were more efficient in removing nitrogen and phosphorus than emergent macrophytes in floating treatment wetlands (Prajapati et al. 2017) and are therefore considered good candidates for the treatment of wastewater effluent.

Where constructed wetlands can emit up to 500 mg/m²/day CH₄ and 25 mg/m²/day N₂O (Hernández et al. 2018), our systems did not show any significant CH₄ emissions (lower than 0.04 mg/m²/day) and N₂O emissions of only 1.5 mg/m²/day at one specific point in time. While some studies also indicate CO₂ emission in constructed wetlands (Badiou et al. 2019), our treatments showed CO₂ uptake, resulting in a total net uptake of GHG. Although our measurements are based on treated wastewater effluent, while constructed wetlands often deal with untreated wastewater – inherently having a higher potential for GHG emission – our data show the potential to mitigate part of the WWTP emissions in the process of hydroponic effluent polishing. Moreover, nutrient reduction in WWTP effluent most likely lowers GHG production in receiving waterbodies, by decreasing eutrophication effects (Beaulieu et al. 2019).

Ideally, plants used in effluent polishing are used in added-value applications, to contribute to the circular economy. One prerequisite for growing plants on wastewater effluent is that algal growth should be limited unless algae are the main product to be cultured. Floating plants that prevent light penetration in the water column can suppress algal growth. When using other plant types, algal growth can be suppressed by using UV light or by adding aquatic animals such as snails to counteract the formation of floating algae beds; while zooplankton or mussels can be used to minimise phytoplankton density. But, it remains to be tested if such animals can also be used in wastewater effluent polishing systems.

Both floating plants tested in our experiment have economic value. Azolla and Lemna are rich in proteins and amino acids, potentially containing even more protein than soybeans (Brouwer et al. 2018). However, non-food applications are preferred because plants grown on domestic wastewater may contain contaminants such as heavy metals and traces of pharmaceuticals. Azolla can be used to produce potting soil for ornamental plants, substituting peat, thereby contributing to the protection of C-storing peatland ecosystems (Khomami et al. 2019). Both species can be digested into bioethanol or biogas as well. Although this would offset the negative carbon footprint of phytoremediation, saving on fossil fuels is always beneficial.

Based on our results, we conclude that the floating plants Azolla and Lemna are promising for use in effluent polishing, due to their ability to lower nutrient concentrations in the effluent while at the same time sequestering carbon and limiting the emission of other greenhouse gases. Where the growth of submerged macrophytes was strongly affected by competition with algae, both of the floating plants showed the highest biomass production, and were most efficient in removing nitrogen and phosphorus from the water column. Note, however, that nutrients taken up by the plants are only permanently removed after harvesting. When combining Azolla with Lemna, or other high value floating plants, excess P and N can be removed from wastewater effluent, while taking up GHGs and producing plant biomass with commercial value, contributing to a circular economy. Moreover, by lowering the nutrient load derived from discharged WWTP effluent, effluent polishing can also contribute to the mitigation of eutrophication and GHG emissions from natural waterbodies.

We would like to thank Koos Janssen, Harry van Zuijlen and Walter Hendrickx from the greenhouse facility at Radboud University for their help during the experiment. Thanks to Jiry de Waal (Adviesbureau de Waal/Vijvermeester.nl) for plant advice and delivery. We thank Sebastian Krosse and Paul van der Ven from the Radboud General Instrumentation for performing the elemental analyses, and Germa Verheggen en Roy Peters for help during other laboratory analyses.

This work was funded by the Dutch water authorities Hoogheemraadschap de Stichtse Rijnlanden, Waterschap Rivierenland and Hoogheemraadschap Hollands Noorderkwartier (Grant Aquafarm 2.0).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.