Assembling plant diversity mitigates greenhouse gas emissions and achieves high nitrogen removal when treating the low-C/N wastewater by constructed wetlands

The low carbon-to-nitrogen (C/N) ratio in wastewater will inhibit pollutant removal, and more seriously, it will cause an increment of nitrous oxide (N2O) emissions of constructed wetlands (CWs). Raising the C/N ratio of wastewater is an effective way to solve this problem, while it may cause secondary pollution and is costly. Assembling plant diversity promotes N removal, while the effects of plant diversity and increasing C/N ratio on global warming potential (GWP) combined by N2O and methane (CH4) are lack of comparison. In this study, 108 CW microcosms were established to explore the effects of increasing the C/N ratio from 1 to 5 and assembling plant diversity on N removal and GHG emissions. Results showed that when the C/N ratio was 1, (1) increasing species richness reduced N2O and CH4 emissions then reduced the GWP by 70%; (2) the presence of Arundo donax in microcosms reduced GWP by 72%; (3) an A. donax × Tradescantia fluminensis × Reineckia carnea mixture resulted in a high N removal and decreased the GWP per g N removal by 92% with a cost increment of 0.05 USD per m3 wastewater treated; and (4) as the C/N ratio increasing to 5, the GWP per g N removal of monocultures was reduced by 96%, but the cost increased by at least 0.29 USD per m3 wastewater treated. In summary, configuring plant diversity in CWs is an efficient, clean, and cost-effective measure to treat wastewater with a low C/N ratio.


Introduction
In recent years, the demand for wastewater treatment around the world has increased sharply, with 80% of wastewater being discharged without any treatment (UNESCO 2021). The carbon-to-nitrogen (C/N) ratio of wastewater from different sources ranges from 1 to 132, and the C/N ratio will affect the performance of wastewater treatment facilities (Vymazal 2014;Sylla 2020). Theoretically, the N removal will be limited if the C/N ratio of wastewater was under 2.86 (Itokawa et al. 2001;Pelaz et al. 2018). In addition, wastewater treatment is accompanied by greenhouse gas (GHG) emissions (IPCC 2021). The total GHG emissions around the word by treating wastewater are 775 Mt CO 2 -eq (IPCC 2014). Low C/N ratio in wastewater also promotes nitrous oxide (N 2 O) emissions due to incomplete denitrification (Duan et al. 2021). Therefore, the measurement for treating wastewater with low C/N ratio urgently needs to be addressed.
Constructed wetlands (CWs) are the green measure to treat wastewater from dispersed sources (Vymazal et al. 2021). The N 2 O and methane (CH 4 ) emissions per kg N removed from CWs are only 3.4% and 0.1%, respectively, of those from wastewater treatment plants (Liu et al. 2012). However, CWs are commonly used to treat domestic wastewater with insufficient C sources (Sun et al. 2016;MEEPRC 2020). CWs are also used to treat piggery wastewater and landfill leachate, and the C/N ratio will be as low as 1 (Feng et al. 2020;Çakirgöz et al. 2021). The discharge of those wastewater with low C/N ratio is still increasing (Mander et al. 2014). Therefore, it is really urgent to optimize the CWs to meet the challenge posed by wastewater with a low C/N ratio.
Adding C sources is a common measure to enhance the wastewater treatment performance for wastewater with a low C/N ratio Çakirgöz et al. 2021). Increasing the C/N ratio will enhance N removal (Nguyen et al. 2018) and reduce N 2 O emissions (Duan et al. 2021). The lowest levels of N 2 O emissions were reported at the C/N ratio of 5 (Wu et al. 2009) or 10 (Zhao et al. 2014;Guo et al. 2020). However, an excessive C/N ratio (such as 20) will also increase N 2 O emissions (Wu et al. 2009). In addition, excessive C may promote CH 4 emissions (Bhullar et al. 2013). Moreover, increasing the C/N ratio of wastewater means continuous investment and transportation, which increases the cost (Hussain et al. 2019). For example, adding methanol increases the infrastructure cost by 25-31% (CDM 2007). Therefore, it is urgent to find a more economical and effective method to reach high N removal and low global warming potential (GWP) when treating the low-C/N wastewater.
As an important part of the ecological structure of CWs, plants play a key role in pollutant removal and GHG emissions (Brisson et al. 2020;Vymazal 2020). The CWs can remove 36-74% of N through plant uptake (Keizer-Vlek et al. 2014;Vymazal 2020). In addition, plant uptake reducing the substrate of nitrification and denitrification to mitigate the N 2 O emissions (Abalos et al. 2014;Kuypers et al. 2018). Optimizing the plant community may enhance N removal and reduce GHG emissions in CWs (Maucieri et al. 2017;Brisson et al. 2020). First, increasing plant richness promotes the complementary utilization of resources by plant communities and enhances N removal (Cardinale 2011;Brisson et al. 2020). Moreover, increasing plant species richness promotes plant biomass production to mitigate N 2 O and CH 4 emissions Luo et al. 2020). However, increasing plant richness may promote N 2 O emissions, because of the increase in plant C secretion . Notably, plant identities are also prominent (Abalos et al. 2014;Jesus et al. 2018). For example, CWs consisting of Arundo donax have higher N absorption than CWs consisting of other species (Kouki et al. 2012;Sylla 2020), and CWs with A. donax or Canna indica have lower N 2 O emissions than those without them ). Nevertheless, studies on the effects of plant diversity on CW ecosystem functioning have mostly been carried out in inorganic (C-free) wastewater treatment systems Han et al. 2021). One study showed that the effects of mixing 4 plant species on N removal as well as N 2 O emissions were equivalent to that of increasing the C/N ratio to 1 when treating C-free wastewater . Whether assembling plant diversity is a suitable alternative to increasing the C/N ratio to a higher level (5) to effectively solve the problem raised by treating wastewater at a low C/N ratio still needs to be explored.
In this study, 108 simulated vertical flow CW microcosms were set up to compare the effects of increasing C/N ratio from 1 to 5 and assembling plant diversity (species richness and species identity) on ecosystem functioning (N removal and GHG emissions) in CWs. The aims were to study (1) the changes in the N mass removal rates and GHG emissions of microcosms with an increment in C/N ratio from 1 to 5; (2) effects of plant diversity on the N mass removal rates and GHG emissions of microcosms at C/N ratios of 1 and 5; and (3) the costs and benefits of species combinations with high N removal and low GWP when treating wastewater at a C/N ratio of 1.

Experimental design
In late March 2019, the total of 108 microcosms (46 cm × 35 cm × 22 cm) were set up in Sandun town (30° 21′ N, 120° 02′ E), Hangzhou City. The research site had a humid subtropical climate, with an annual average temperature of 18.0 °C, an annual sunshine duration of 1658 h, and an annual average relative humidity of 74% (NBSC 2020). The microcosm experimental system was set up indoor and is not affected by natural precipitation. At the beginning of the experiment, the experimental conditions of all microcosms were the same, except for the different combinations of plant species. The washed sand (diameter = 0.5-3 mm) was used as the substrate of each microcosm to ensure that there are no nutrients and propagules (Fig. S1).
The species richness in plant communities was 1, 3, or 4 ( Fig. 1; Fig. S2). Two frequently used species in CWs, A. donax L. and Oenanthe javanica (Blume) DC., and two ornamental species, Reineckea carnea (Andr.) Kunth. and Tradescantia fluminensis Vell., were chosen (Fig. S2). The four species are morphologically different, and have different resources utilization mode (Table S1). The nine plant combinations (4 monocultures of each species, 4 three-species mixtures, and 1 four-species mixture) and six replicates for each combination were set up based on a randomized block design. The planting density was 12 plants per microcosm. For the mixtures, the individual number of each species was the same, and different plant species were planted adjacent to each other to ensure the evenness of each species.
The synthetic wastewater was according to Hoagland and Arnon (1950), and sucrose was added to the water as the C source (Table 1). This study set the C/N ratio (calculated as COD/N) at 1 and 5 (Table 1). The research group has shown that plant diversity can improve N removal when treating low-C/N ratio wastewater, but there is a lack of research on GWP Du et al. 2020). Therefore, a C/N ratio of 1 was set in this study to further explore the various plant combinations in one treatment. The research group has also shown that in CWs, the promotion of increasing plant richness on N removal was equivalent to that of increasing the C/N ratio from 0 to 1 . Previous studies suggested the C/N ratio of 5 was conducive to N removal and N 2 O reduction at the same time (Wu et al. 2009). Therefore, in this study, a C/N ratio of 5 was set in another treatment for comparing the effect of plant diversity and increasing C/N ratio on N removal and GWP. The pH of  wastewater was approximately 6. Each microcosm was fed with wastewater (7 L) once every 10 days. The experiment lasted until the beginning of July.

Sample collection and measurements
GHG mainly include carbon dioxide (CO 2 ), N 2 O, and CH 4 (IPCC 2021). In this study, GHG were represented by N 2 O and CH 4 without considering CO 2 , as CO 2 from herbaceous plants is a short-cycle gas and is not a contributor to the greenhouse effect (IPCC 2021). To compare the difference between the two treatments and among species combinations, gas samples were taken by closed static chamber method (Cheng et al. 2007) at the end of the experiment. Transparent PVC chambers (48 cm × 35 cm × 110 cm) enclosed all the plants and were made airtight by inserting the bottoms into water. During the preexperiment, one microcosm was randomly selected for sampling and determination for each block, and sampling was performed at 0, 3, 5, 10, 20, 30, 40, 50, and 60 min after the chambers were made airtight. Results showed that the concentrations of N 2 O and CH 4 increased linearly in the first 20 min. Therefore, during the formal test, the sampling time is determined as 20 min after the chambers were made airtight. A vacuum gas collection tube was used to collect gas samples from each microcosm. At the same time, the temperature and humidity in the chamber were measured by a hygrothermograph. Gas concentrations were determined by gas chromatography (Agilent 7820, Agilent Technologies Inc., USA). After gas sampling, a water sample (100 mL) was taken from each microcosm. The concentrations of ammonium (NH 4 + -N), nitrite (NO 2 − -N), and nitrate (NO 3 − -N) were determined by automated colorimetry (CleverChem-Anna, DeChem-Tech.GmbH, Germany). The total N (TN) and total organic carbon (TOC) concentration were analyzed by a nondispersive infrared detector (Multi N/C 3100, Analytik Jena AG, Germany). A multiparameter water quality meter (Bante 900P, Bante Instruments, China) was used to determine oxidation-reduction potential (ORP) and dissolved oxygen (DO).
After water sample collection, all the plants were harvested. The plants were oven-dried (65 °C) to a constant weight to measure the aboveground and belowground biomasses. The N concentrations in plant tissues were detected by an elemental analyzer (Flash HT2000, Thermo Finnigan, Germany). To determine the N pool in the substrate, a fresh substrate sample from each microcosm was weighed at 150 g, and the N concentration of each sample was measured after extract by KCl solution (2 mol L −1 ). Bulk density was determined with the ring-cutting method.

GHG emissions
The calculation of GHG emissions followed the method described by Cheng et al. (2007), where E i is the emissions rate of N 2 O or CH 4 (mg m −2 day −1 ), ρ is the density of every gas (mg m −3 ) under normal conditions, d c is the change in gas volume fraction (m 3 m −3 ), V is the volume (m 3 ) of the air in chamber, T 0 is the reference temperature (273 K), d t is the sampling duration (min), A is the bottom area (m 2 ) of the chamber, and T is the experimental temperature (K).
The total GWP (g CO 2 -eq m −2 day −1 ) was calculated as the sum of CH 4 and N 2 O emissions multiplied by 27 and 273, respectively (IPCC 2021).

N removal
The TN mass removal rate (NR, mg m −2 day −1 ) was calculated as, where c i is the influent TN concentration (mg L −1 ), V i is the total influent volume during the whole operation, c e is the effluent TN concentration of the end of the last hydraulic cycle (mg L -1 ), V e is the total effluent water volume (L), S is the microcosm's bottom area (m 2 ), and D is the total experimental time (days).
The N budget of the microcosm was calculated by the mass balance method (Geng et al. 2019 where M i is the amount of N supplied (mg), M e is the amount of N remaining in the effluent (mg), M p is the amount of plants N uptake (mg), M s is the amount of N adsorbed by the substrate (mg), M d is the amount of N used for denitrification, and M a is the amount of N lost by ammonia volatilization. Among these variables, M a was less than 1% of the total N removal in a previous study with the same influent N concentration, so it was ignored in this study .
The plant uptake (N p , mg N m −2 day −1 ), denitrification (N d , mg N m −2 day −1 ), and substrate adsorption (N s , mg N m −2 day −1 ) rates were calculated as, where S is the microcosm's bottom area (m 2 ), and D is the total experimental time (days).

Under-emission effect
The transgressive under-emission effect was calculated according to Luo et al. (2016), where Mix(Ei) is the GHG emissions or GWP in the mixtures and Min(Ei) is the lowest GHG emissions or GWP in the relevant monocultures. Similar to the transgressive under-depletion effect (Palmborg et al. 2005), D mix − GHG < 0 suggests that the GHG emissions in mixtures were lower than those in the monoculture with the lowest emissions, and this state is called transgressive under-emission ).

Statistical analysis
The data were analyzed to investigate the impacts of plant species richness, identity, and composition on CW ecosystem functioning. At first, the data were subjected to the Kolmogorov-Smirnov test for normality and Levene's test for the variance equality. The GHG emissions data were natural logarithm transformed to meet the analysis of variance statistical assumptions. To meet the statistical requirements, the N 2 O emissions data at a C/N ratio of 5 were added by one when taking the logarithm. The effects of species richness and composition on the N mass removal rate and N removal pathways were analyzed by the Kruskal-Wallis test. Independent sample t-tests were used to define the significant differences in GHG emissions or N removal under two C/N ratios as well as the differences in combinations with or without a particular species. The difference between D mix-GHG and 0 was detected by a one-sample t-test. The analyses were performed in SPSS 20 (SPSS Inc., Chicago, IL, USA) and R 4.1.2. at a statistical significance level of α = 0.05.

Effects of plant diversity on N 2 O and CH 4 emissions under two C/N ratios
When the C/N ratio was 1, the N 2 O emissions of the microcosms decreased with increasing species richness (P < 0.05; Fig. 2a), and the lowest N 2 O emissions occurred in the microcosms with the T. fluminensis × R. carnea × A. donax mixture, while the highest N 2 O emissions were observed The open points, open bars, and thin lines indicate the treatments with a C/N ratio of 1, while the closed points, closed bars, and bold lines indicate the treatments with a C/N ratio of 5. Significant differences among combinations (at P < 0.05) are shown by letters above the bars: lowercase letters indicate a C/N of 1, and uppercase letters indicate a C/N of 5. The error bars mean the standard errors, and asterisks show significant differences between the two C/N ratios (* P < 0.05; ** P < 0.01) in the monoculture of T. fluminensis (P < 0.05; Fig. 2b). When the C/N ratio was 1, the average N 2 O emissions of mixture microcosms was only 1/3 of that of monoculture microcosms (P < 0.05; Fig. S3a). Allocating the A. donax and R. carnea in the community reduced the N 2 O emissions (P < 0.05; Table 2). When the C/N ratio was increased to 5, the N 2 O emissions of all microcosms were reduced (P < 0.05; Fig. 2b), but there were no differences among different species combinations, and there was no difference between the mean value of monocultures and mixtures (P > 0.05; Fig. 2b; Fig. S3a). Similarly, increasing species richness reduced CH 4 emissions at a C/N ratio of 1, but not 5 (Fig. 2c). Regardless of whether the C/N ratio was 1 or 5, the CH 4 emissions were lowest for the mixture of O. javanica × A. donax × R. carnea, while those of the R. carnea monoculture were the highest (P < 0.05; Fig. 2d). When the C/N ratio was 1, the average CH 4 emissions of mixture microcosms was only 1/2 of that of monoculture microcosms (P < 0.05; Fig. S3b). Allocainge the O. javanica and A. donax in the community reduced the CH 4 emissions (P < 0.05; Table 2). For the R. carnea monoculture, the CH 4 emissions declined when the C/N ratio was increased to 5 (P < 0.01; Fig. 2d). When the C/N ratio was 5, there was no difference between the average CH 4 emissions of monocultures and that of mixtures (P > 0.05; Fig. S3b).

Effects of plant diversity on GWP under two C/N ratios
To comprehensively evaluate GHG emissions, the N 2 O and CH 4 emissions were integrated into the GWP. When the C/N ratio was 1, the GWP negatively responded to species richness (P < 0.05; Fig. 3a). The GWP of the microcosms with the T. fluminensis × R. carnea × A. donax mixture was the lowest, and that of the T. fluminensis monoculture was the highest (P < 0.05; Fig. 3b). When the C/N ratio was 1, the average GWP of mixture microcosms was only 1/3 of that of monoculture microcosms (P < 0.05; Fig. S3c). Configuring the A. donax and R. carnea in community reduced the GWP Table 2 Emissions of N 2 O and CH 4 (mg m −2 day −1 ), the GWP (g CO 2 eq m −2 day −1 ), N removal rate (mg N m −2 day −1 ), plant uptake, denitrification, and substrate adsorption rate (mg N m −2 day −1 ) in the microcosms with or without the four plant species Arrows indicate a significant increase ( ), decrease ( ), or no effect ( ). A value of "1" indicates a C/N ratio of 1, while "5" indicates a C/N ratio of 5 when the C/N ratio was 1 (P < 0.05; Table 2). The N 2 O emissions composed the major part of GWP at a C/N ratio of 1, while CH 4 emissions were the driver at a C/N ratio of 5 (Fig. 3b). The GWPs of all the combinations was decreased as the C/N ratio increasing to 5 (P < 0.05; Fig. 3b). When the C/N ratio was 5, the lowest GWP appeared in the O. javanica × A. donax × R. carnea mixture (P < 0.05; Fig. 3b), while there was no difference between the average GWP of monocultures and that of mixtures (P > 0.05; Fig. S3c).
The trend of GWP per g N removal was similar to that of GWP. The GWP per g N removal negatively responded to species richness only when the C/N ratio was 1 (P < 0.05; Fig. 3c). The GWP per g N removal of the microcosms with the T. fluminensis × R. carnea × A. donax mixture was the lowest (P < 0.05; Fig. 3d). Increasing the C/N ratio to 5 decreased the GWPs of the combinations except for the mixture of T. fluminensis × R. carnea × A. donax. The GWP per g N removal of the T. fluminensis × R. carnea × A. donax mixture under a C/N ratio of 1 was as low as the average value of monocultures under a C/N ratio of 5 (P > 0.05; Fig. 3d).

Effects of plant diversity on N removal and pathways under two C/N ratios
The TN mass removal rate was slightly improved with increasing species richness only when the C/N ratio was 5 (P < 0.001; Fig. 4a). The TN mass removal rate of O. javanica was lower than that of the other plant combinations (P < 0.05, Fig. 5a). Increasing the C/N ratio to 5 reduced the TN mass removal rate in each combination (P < 0.05, Fig. 5a). When the C/N ratio was 5, the monoculture of A. donax and the mixtures containing A. donax had the highest TN mass removal rates (P < 0.05; Fig. 5a). Regardless of whether the C/N ratio was 1 or 5, the presence of A. donax increased the TN mass removal rate (P < 0.05; Table 2).
Plant uptake, denitrification, and substrate adsorption are three major N removal pathways. Species richness was positively correlated with plant uptake under the two C/N ratios (P < 0.001; Fig. 4b). Under the two C/N ratios, the average plant uptake of mixtures was about twice that of monocultures (P < 0.001; Fig. S3d). When the C/N ratio was 1, the plant uptake was highest in the A. donax monoculture and the mixtures containing A. donax (P < 0.05; Fig. 5b). Increasing the C/N ratio to 5 resulted in a decrease in plant uptake in the A. donax monoculture and all the mixtures containing A. donax (P < 0.05; Fig. 5b). When the C/N ratio was 5, the A. donax × T. fluminensis × O. javanica mixture had the highest plant uptake (P < 0.05; Fig. 5b). Regardless of whether the C/N ratio was 1 or 5, the presence of A. donax increased plant uptake but inhibited denitrification (P < 0.001; Table 2). Under the two C/N ratios, increasing species richness decreased denitrification (P < 0.001; Fig. 3 Effects of plant diversity on GWP in CWs. a Relationship between GWP and richness; b GWP for species combinations, where the inset panel shows the contribution ratios of N 2 O and CH 4 to GWP under the two C/N ratios; c relationship between GWP per unit N mass removal (GWP/NR) and richness; d GWP/NR among species combinations under two C/N ratios, and the dash line represent the average of GWP/ NR at the C/N ratio of 5. The abbreviations of plant species (O, T, R, and A), letters, points, and bars are the same as those in Fig. 2. The asterisks indicate significant differences between the two C/N ratios (* P < 0.05; ** P < 0.01) . 4c). When the C/N ratio was 1 or 5, the average denitrification of mixtures was 24% and 50% lower than that of monocultures, respectively (P < 0.05; Fig. S3e). Increasing the C/N ratio from 1 to 5 decreased denitrification in each combination (P < 0.01; Fig. 5c). Regardless of whether the C/N ratio was 1 or 5, the highest levels of denitrification occurred in the monocultures of T. fluminensis and R. carnea (P < 0.05; Fig. 5c). Substrate adsorption slightly decreased with increasing species richness only when the C/N ratio was 5 (P < 0.05; Fig. 4d), and increasing the C/N ratio to 5 led to a reduction in substrate adsorption (P < 0.01; Fig. 5d).

Increasing plant richness alleviates GHG emissions under a low C/N ratio
Low C/N ratio in wastewater, theoretically defined as lower than 2.86, usually leads to high N 2 O emissions (Itokawa et al. 2001;Duan et al. 2021). In this study, when the C/N ratio was 1, the average N 2 O emissions of monoculture microcosms accounted for 5% of the inflow TN mass (Table S2). This contribution is slightly higher than the scope of previous reports (0.001-4%; Table S2). Fortunately, increasing plant richness in CWs can make the mean N 2 O emissions of mixtures (species richness at 3 or 4) 70% lower than that of monoculture microcosms when the C/N ratio is 1 (Fig.  S3a). This result is different from the fact that increasing plant richness in CWs from 1 to 4 increased N 2 O emissions by 68-267% when treating C-free wastewater (Fig. S6), but is consistent with the trend reported by Du et al. (2020) that the mean N 2 O emissions of mixtures were 46% lower than those of monocultures at a C/N ratio of 1. Community denitrification is one of the important factors affecting N 2 O emissions (Maucieri et al. 2017). In this study, increasing plant richness reduced the denitrifying N indirectly by promoting plant uptake by 2-fold compared with that of monocultures and then reduced N 2 O emissions (Fig. 3c, d; Fig.  S7). In addition to reducing N 2 O emissions by increasing the richness, increasing the C/N ratio from 1 to 5 also reduces N 2 O emissions, and the N 2 O emissions of all microcosms were declined to the same low level, accounting for less than 0.4% of influent TN (Fig. 2b). The further increase of species richness under C/N ratio of 5 will no longer affect N 2 O emissions (Fig. 2a). Studies have shown that increasing the C/N ratio is conducive to complete denitrification, thus reducing the gaseous N emissions in the form of N 2 O (Pan et al. 2013). By increasing the C/N ratio from 3 to 8, the N 2 O-N emissions in the influent TN is significantly reduced Fig. 4 Effects of plant species richness on N removal under the two C/N ratios. a TN mass removal rate, b plant uptake, c denitrification, d and substrate adsorption in response to species richness under two C/N ratios. The points and lines are the same as those in Fig. 2. The asterisks indicate significant differences between the two C/N ratios (** P < 0.01) from 2 to 0.1% (Yan et al. 2017). The direct effect of C/N ratio on N 2 O emissions may uncouple the negative correlation between species richness and N 2 O emissions. The above results show that when CWs are used for treating wastewater with a low C/N ratio, improving plant species richness has great potential for N 2 O emissions reduction. In this study, the N 2 O emissions of CWs account for 90% of GWP (calculated as the combination of N 2 O and CH 4 emissions) when the C/N ratio is 1 (Fig. 3b). This result is similar to the values of 84% reported by Du et al. (2020) and 88% reported by Luo et al. (2020). Another study pointed out that N 2 O emissions can account for up to 75% of the C footprint in wastewater treatment plants (Daelman et al. 2013), indicating that controlling N 2 O emissions is the key factor in reducing GWP. However, there is often a tradeoff between N 2 O and CH 4 emissions (Saha et al. 2017), so reducing the emissions of one gas may be at the cost of adding another one. For example, Yao et al. (2012) found that although the application of N fertilizer in rice fields reduced CH 4 emissions by 53%, it increased N 2 O emissions by 6-fold and was not conducive to reducing GWP. However, in this study, increasing species richness reduced CH 4 emissions by 56% on the premise of reducing N 2 O emissions at a C/N ratio of 1 in the CW microcosms ( Fig. 2c; Fig. S3b). This may be because when the C/N ratio is 1, increasing species richness reduces the TOC in the system (Fig. S4c). The content of TOC often promotes the CH 4 production (Mander et al. 2014). In this study, CH 4 emissions were positively correlated with the TOC in the system under a C/N ratio of 1 (Fig. S4a). In addition, DO concentration and CH 4 emissions are negatively correlated under the low C/N ratio (Fig.  S4b); the community DO concentration increased with the increase of species richness (Fig. S4d), which is conducive to CH 4 oxidation (Maucieri et al. 2017). Therefore, a simultaneous reduction in the emissions of N 2 O and CH 4 was realized, so the system has great potential for a final reduction in GWP.

Species identity surpasses species richness in terms of the effect on GHG mitigation under a low C/N ratio
The plant species also affects community GHG emissions in CWs (Maucieri et al. 2017). A. donax was the most conducive to GHG emissions reduction among the species in this study ( Table 2, Table S3). Compared with the microcosms without this species, the presence of A. donax in microcosms reduced N 2 O by 71% under a low C/N ratio ( Table 2, Table S3). The allocation of A. donax in the community also reduced N 2 O emissions by 42% in the treatment of C-free wastewater . Configuring species with high plant uptake in the community is conducive to N 2 O emissions reduction (Oram et al. 2020). A. donax has a high plant N uptake capacity that is 40% (Kouki et al. 2012) or 77% (Meng et al. 2015) higher than that of Typha latifolia, and its productivity is higher than that of 19 other species (Liu et al. 2012). The plant N uptake of A. donax was 2-to 23-fold higher than that of the other species in this study ( Fig. 4b; Fig. S8). The allocation of A. donax in the community increased plant N uptake by 200% (Table S3) and indirectly reduced the N 2 O emissions reflected by structural equation modeling (Fig. S7). In addition, A. donax has developed roots and a high oxygen secretion capacity, which is conducive to reducing denitrification and reducing  (Lai et al. 2011;Kuypers et al. 2018). In this study, A. donax had the largest underground biomass production, and the presence of A. donax improved the DO concentration while reducing denitrification ( Fig. S1; Table S3). An increase in DO may promote CH 4 oxidation (Bhullar et al. 2013), and the presence of A. donax reduced CH 4 emissions by 63% (Table 2; Table S3). In conclusion, A. donax is an excellent species for practical use since it has a high plant N uptake capacity and mitigates the emissions of GHG in the plant community.
It is worth noting that the total biomass production of per plant of A. donax in mixtures was always higher than that in monoculture under the two C/N ratios (Fig. S9, S10). This result means that the mixtures are more conducive to the growth of A. donax. Previous studies have also shown that the complementary effect between plant species makes the biomass production of per plants in mixture greater than that in monoculture (Siebenkäs et al. 2016). This result means that A. donax can play a greater role in GHG emissions reduction in mixed communities than in monocultures. In fact, A. donax is also recommended as a high-yield biofuel production species to further reduce the GWP worldwide (Corno et al. 2014). Moreover, the presence of A. donax enhanced the TN mass removal by 8 mg m −2 day −1 in community ( Fig. 4a; Table 2). Therefore, the combination containing A. donax is suitable for the wastewater treatment with a low C/N ratio to improve N removal while reducing GWP.
GHG emissions vary greatly among plant species, and monocultures increase the risk of high emissions. When the C/N ratio was 1, the N 2 O emissions of the T. fluminensis monoculture were higher than those of other species in previous studies (1.3-293.8 mg m −2 day -1 , Fig. S5, Table S2). In this study, NO 2 − -N was positively correlated with N 2 O emissions (Fig. 6a). The accumulation of NO 2 − -N in this monoculture (4.6 mg L −1 , Fig. 6b) indicated incomplete denitrification so that N was released in the form of N 2 O (Itokawa et al. 2001;Pan et al. 2013). Regarding CH 4 emissions, the monoculture of R. carnea yielded a high level in previous studies (−3.3-110.6 mg m -2 day −1 , Table S2). However, no high emissions were observed in mixed communities containing these two species (Fig. 2b, c; Fig. 3b). Du et al. (2020) reported that species with a small biomass proportion in a mixture had little effect on community functioning, supporting the mass ratio hypothesis (Grime 1998). Similarly, in this study, the biomass proportions of T. fluminensis and R. carnea in mixtures only accounted for 8% and 3% on average, respectively (Fig. S9). Mixing plant species eliminated the disadvantage of high GHG emissions of T. fluminensis and R. carnea and was conducive to the ornamental value of the plant community (Hu and Gill 2015) (Table S1). Therefore, although monocultures of T. fluminensis and R. carnea are not recommended for treating low-C/N ratio wastewater, they can be used in mixtures to ensure clean treatment.  Fig. 2. When the C/N ratio was 5, the effluent NO 2 − -N concentration, N 2 O emissions, and GWP of the microcosms were nearly zero, so the analysis was not performed. The error bars mean the standard errors, and asterisks show the significant differences between the D mix value and zero (** P < 0.01) Interestingly, A. donax combined with two high-GHGemissions species (T. fluminensis and R. carnea) resulted in the most effective plant combination when treating a low-C/N wastewater (Fig. 2b, d). The plant species composition had a greater effect than plant species richness on GHG emissions reduction (Table 3; Table S6). This result is similar to the findings of Abalos et al. (2014) in grassland; that is, plant species composition explained a greater proportion of N 2 O emissions. In this study, when the C/N ratio was 1, compared with the average GWP of monocultures, the mixture of T. fluminensis × R. carnea × A. donax reduced the GWP of the microcosm by 92% (Fig. 3b). The N 2 O emissions of T. fluminensis × R. carnea × A. donax were 76% lower than those of the A. donax monoculture, leading to transgressive under-depletion of GWP in this community (Fig. 6c, d). Similarly, the NH 3 volatilization of Rumex japonicus × Cichorium intybus × Lolium perenne was found to be 60% lower than that in the monoculture with the lowest value . In general, allocating appropriate species combinations, rather than only improving richness, has a higher potential for GWP reduction.

Manipulating plant combinations is a win-win measure compared to increasing the C/N ratio
Increasing the C/N ratio was found to reduce the N 2 O emissions of monoculture CWs by 28-85% (Table S2). In this study, when the C/N ratio was increased from 1 to 5, N 2 O emissions of monoculture CWs decreased by 1-fold in average (Fig. 2b). However, in this study, the mass of N released in the form of N 2 O from microcosms with a high C/N ratio was lower than 0.4%, which is at the low discharge level of CWs (0.001-4%, Table S2). Moreover, on the basis of reducing N 2 O emissions, raising the C/N ratio to 5 reduced the average CH 4 emissions of monocultures by 60% (Fig. S3b). This may be because the DO concentration in monoculture with high C/N ratio is higher than that with low C/N ratio, which is conducive to CH 4 oxidation (Fig. S3b). In general, increasing the C/N ratio from 1 to 5 reduced GWP by 98% in this study (Fig. S3c). This effect is almost equivalent to the optimal emissions reduction obtained with plant combinations. The GWP per unit N removal is a suitable index to comprehensively measure the treatment performance of CWs ). The T. fluminensis × R. carnea × A. donax mixture reduced the GWP per unit N removal in the treatment of wastewater with a C/N ratio of 1 by 96%, and the value was as low as that of the monocultures with a C/N ratio of 5 (Fig. 3d). This mixture also provided a win-win combination of efficient N removal and low GHG emissions under a low C/N ratio (Fig. 7a). The N removal efficiency of this combination was higher than 90%, which was higher than the average value of previous studies in CWs (55% ,  Table S4). These results showed that for the treatment of wastewater with a low C/N ratio, the configuration of an appropriate species combination has great potential and can be comparable to increasing the C/N ratio.

Plant diversity is a low-cost route to treat wastewater with a low C/N ratio in CWs
In the practical application of CWs, the trade-off between strengthening water treatment efficiency, reducing environmental impact, and increasing cost must be considered (Resende et al. 2019). The cost of wastewater treatment will rise when increasing the influent C/N ratio, which is generally achieved by adding external C sources (Sun et al. 2010). In wastewater treatment, the commonly used C sources include methanol, sucrose, glucose, and acetic acid, and their costs vary greatly (Table S5). Methanol is the cheapest (0.29 USD per m 3 wastewater treated; Table S5) among the commonly Table 3 Results of two-way ANOVA on the effects of plant species richness (SR), species composition (SC), C/N ratio, and their interactions on N 2 O and CH 4 emissions (mg m −2 day −1 ), GWP (g CO 2 eq m −2 day −1 ), and GWP per g N removal (GWP/NR) (g CO 2 -eq g N −1 ) used C sources; however, it is toxic (Ramírez et al. 2006). Sucrose is nontoxic and relatively economical (0.45 USD per m 3 wastewater treated; Table S5). Under the same infrastructure, energy, and labor costs, when the C/N ratio was increased from 1 to 5 by adding C, the cost (seedlings, planting labor, and sucrose cost) per m 3 wastewater treated reached 0.55 USD, while the cost (seedlings and planting labor) for assembling the best plant composition was only 0.15 USD per m 3 wastewater treated (Table S5). This means that increasing plant diversity can save 73% of the cost compared with adding sucrose or save 62% of the cost compared with adding methanol (Table 4, Table S5). In addition, perennial plants can operate continuously in CWs (Table S1), and the harvested plant biomass can further effectively produce bioenergy (Liu et al. 2012;Tanaka et al. 2017). Therefore, assembling plant combinations can be more economical in the long terms. However, adding C requires continuous investment (Hussain et al. 2019). These findings showed that optimizing plant diversity will solve the problems, which raised by the low C/N ratio in wastewater, more economically than increasing C/N ratio.

Conclusions
When treating wastewater with a C/N ratio as low as 1, both increasing the C/N ratio and assembling plant communities can make CWs efficient and green. Reasonable species combinations can mitigate the GWP of wastewater treatment by mitigating N 2 O, and the reduction in GWP per g N removal achieved by this method is almost equivalent to that achieved by increasing the C/N ratio to 5. The excellent practical species A. donax has a high plant uptake ability, which provides an opportunity to improve efficiency while mitigating the environmental impact of treating wastewater with a low C/N ratio. Considering the economic costs of CWs, assembling optimal plant combinations, such as T. fluminensis × R. carnea × A. donax, is more cost-effective than increasing the C/N ratio. The biomass of high-yielding A. donax can be further considered to produce bioenergy to enhance the benefits of CWs. Future studies should expand the C/N ratio gradient, consider the time dynamics of GHG emissions, and explore species combinations suitable for different C/N ratios to optimize the treatment effect of CWs.

Supplementary Information
The online version contains supplementary material available at https:// doi. org/ 10. 1007/ s11356-022-22088-8. Data availability All data generated or analyzed during this study are included in this published article and its supplementary information files.

Fig. 7
Synergy between mitigating GHG emissions and improving N removal when treating wastewater with two C/N ratios. a Relationship between GWP and TN mass removal rate at a C/N ratio of 1; the mixture of T. fluminensis × R. carnea × A. donax (in the red circle) had high N mass removal with low GWP; b relationship between GWP and TN mass removal rate at a C/N ratio of 5