Field-Scale Constructed Floating Wetland Applied for Revitalization of a Subtropical Urban Stream in Brazil

Abstract

Constructed floating wetland (CFW) is an ecotechnology used to improve water quality using emergent macrophytes on a floating mat structure. The goals of this research were to design and evaluate a low-cost field-scale CFW for revitalization of a polluted lentic section of an urban stream, located in the subtropical coastal region of South Brazil. To attain these goals (i), the design parameters were selected from field-scale applications in the literature, and (ii) the influence of the meteorologic and hydraulic data over the CFW performance to improve water quality were analyzed during five months. Macrophyte leaves grew 1 cm·day⁻¹. Biochemical oxygen demand (BOD) (72%), total phosphorus (TP) (52%), turbidity (53%), total solids (TS) (60%), dissolved oxygen (DO) (39%) and water temperature (WT) (0.4 °C) showed statistically significant reductions. The HRT was sufficient to reduce dissolved carbonaceous organic matter. HRT and solid particles-related parameters showed reductions both on high and low HRT. The resulting loading rates can be used for the design purposes of similar CFW field applications. The CFW promoted water quality improvement, attractiveness of fauna, temperature regulation, carbon sequestration, and is a potential ecotechnology towards the depollution of river basins in urban areas.

1. Introduction

The increasing urban population density has impacted the urban natural environment, mainly urban rivers. The impermeable surfaces harm the natural water cycle, causing urban flooding, droughts, and poor water quality in rivers [1].

In developing countries, where there is a lack of sanitation and effluents are still discharged untreated into stormwater conveys and rivers [2], in most rivers of Latin America, Africa, and Asia, water pollution has worsened by more than 50% since the 1990s in most rivers of Latin America, Africa, and Asia [3]. In Brazil, 55% of the urban population (92 mi) was served by safe sanitation; however, only 23% resided in municipalities that removed more than 60% of the organic load generated. Additionally, 57% of the urban population lacked enough watercourse flow to dilute the organic load in water bodies [4]. This scenario reveals the challenges to increase sanitation as well as the search for alternative approaches to deal with the water quality of urban stormwater and rivers. Ecological approaches have been developed concerning river remediation [5,6], sustainable stormwater management [7,8,9], and urban nature conservation [10,11].

The nature-based solutions (NBS) concept encompasses different ecological approaches, focused on improving nature, while emphasizing how bringing society closer to nature can contribute to communities facing social challenges and adapting to climate change [9]. However, to distinguish the NBS from other approaches, eight principles were listed by the IUCN [12] as follows: I—embrace nature conservation norms and principles; II—be implemented in alone or in combination; III—inserted in a natural and cultural context; IV—produce benefits to the whole society in a fair and equitable manner, and allow social participation; V—allow the maintenance of biological and cultural diversity, and the evolution of the ecosystem; VI—be applied at the landscape scale; VII—promote long-term benefits; VIII—be an integral part of other policies [12].

Constructed floating wetland (CFW) is an ecotechnology that can be applied to revitalize stormwater and urban rivers as a potential NBS [13]. It is a type of human-made wetland, similar in appearance to natural marshes, and included as a free water surface (FWS) hydrologic mode category [14]. Emergent macrophytes in these devices are supported by a floating medium, composed of materials that guarantee buoyancy, such as wood, plastic, inorganic matting, or fiberglass. The root growth in the water column provides an extensive surface area for microbial biofilm [15]. Roots and attached biofilms function as a natural filter providing an effective removal of pollutants by physical and biochemical processes, such as adsorption, sedimentation, and biodegradation [16,17]. The CFW ecotechnology can be installed inside the watercourses [18,19], even in drainage channels [20,21], and can be used in conjunction with all NBS principles, mainly with depollution of river basins in urban areas [13].

Despite the evolution of this ecotechnology in the last 30 years for stormwater, design parameters are still in development [15,16,21,22,23]. Field applications lacked detail on the experimental design and the physical parameters of CFW and the watercourse, such as hydraulic retention time (HRT) and water depth. Short circuiting occurs when a full-width pond is not covered, or water beneath the CFW passes without contact with roots, which causes a reduction in water treatment efficiency [15].

The high variability of meteorological characteristics of the subtropical climate in South Brazil challenges the application of ecotechnologies in urban streams, with sudden variations in flow, temperature, and radiation. Rain generation mechanisms are diverse in this coastal region, with frontal precipitation being common due to the complex topography (0 m to 1818 m close to the coast) and convective precipitation being frequent in summer [24]. In South America, CFWs were applied in the field [25] and pilot [26,27,28] scales for urban sewage treatment, and few studies at the mesocosm scale were applied for stormwater amelioration [13,29,30].

The goals of this research were to design and evaluate a low-cost field-scale CFW for revitalization of a polluted lentic section of an urban stream, located in the subtropical coastal region of South Brazil. To attain these goals, (i) the design parameters were selected from field-scale applications in the literature, and (ii) the influence of the meteorologic and hydraulic data over the CFW performance to improve water quality were analyzed during five months. This project was scaled based on the ratio CFW area by catchment contribution area, macrophyte type and density, minimum length and the surface BOD loading rate.

2. Materials and Methods

2.1. Site Description

The monitoring site was located at Florianopolis coastal island, South Brazil, Cfa humid subtropical Koppen’s climate classification, 27°35′44″ S and 48°31′01″ W.

The average annual temperature in Florianopolis is 20.7 °C, with a temperature in the coldest month (July) of 16.2 °C and 25 °C in the hottest (February). Rainfall is well distributed throughout the year, with a total annual of 1.468 mm. The highest rainfall month is February with 174 mm. The average solar radiation is 171 W/m², and the highest solar radiation month is December with 222 W/m² [31].

The contributing catchment area is zoned as a high-density residential and commercial area of 115.970 m². The first unpiped segment of the urban stream is a retention pond (pond 1) and receives stormwater from one drainage convey (inlet). The CFW installation was located at the Pond’s 1 outlet (Figure 1) due to the insufficient water column depth at the inlet. The sediment in pond 1 was 30 cm deep close to the borders, and 100 cm deep at the center. The average water depth beneath the CFW was 0.40 m. The water quality of the stream was visibly deteriorated, with an absence of fishes, and a high proliferation of floating macrophytes, such as Pistia stratiotes, Lemna spp. and Hydrocotyle ranunculoides. This scenario indicated a low dissolved oxygen concentration and high eutrophication of the water body, due to the release of untreated effluents into stormwater conveys. Despite the pollution, a considerable amount of fauna inhabits the surroundings of this watercourse, such as the broad-snouted caiman (Caiman latirostris), red-eared turtle (Trachemys scripta elegans) and birds, as the snowy egret (Egretta thula) and the common gallinule (Gallinula galeata).

2.2. CFW Design Parameters

Design parameters used to scale the CFW were the ratio CFW area by catchment contribution area, macrophyte type and density, minimum length and surface mass loading rate.

The surface mass loading rate was 40 gBOD·m⁻²·d⁻¹, which resulted in a minimum surface area of 18.4 m². The BOD load in the pre-CFW installation period was 734 gBOD·d⁻¹ (1.0 L/s × 8.2 mgBOD·d⁻¹ in average). Bevenuti et al. [25] applied 126 gBOD·m⁻²·d⁻¹ for a field-scale CFW for raw urban sewage treatment (BOD around 300 mg·L⁻¹, surface of 289 m², HRT of 11 days) using Typha dominguensis in South Brazil. Von Sperling and Sezerino [32] recommended a surface BOD loading rate of 15 gBOD·m⁻²·d⁻¹ for the horizontal subsurface flow constructed wetland with low BOD concentration applied for sewage treatment in Brazil.

The CFW surface area (19.2 m²) resulted as 0.017% of the contributing catchment area, 90% smaller in percentage than used by Walker et al. [33]. Macrophytes were selected as recommended by Wang and Sample [34]: (1) native and non-invasive species, and (2) wetland plants with aerenchyma. The Typha dominguensis was selected due its native availability in Florianopolis, and its recognized performance in wastewater and stormwater treatment [35,36]. A density of 30 plants per m² or 6 plants per plastic crate was chosen, similarly tested by Rigotti et al. [13] in mesocosm in south Brazil. A total of 576 macrophytes were initially planted in CFW. The planting of seedlings in the CFW occurred with part of the clods of soil from the original site. Dried leaves and branches from around the pond were placed inside the crates to serve as substrate for the plants. Sixteen rafts were built. The length selected was 2.2 m, due the arrangement of two rows of rafts (Figure 2). The inside row had nine rafts, and the outside row seven rafts (outlet border). A similar length was chosen by Olguín et al. [22] with 1.32 m, and De Stefani et al. [19] with 1.5 m.

Each raft was composed by six plastic crates and four 20 L mineral water gallon to guarantee buoyance, with a total of 96 crates and 64 gallons, similar designed by Olguín et al. [22]

The CFW structure was developed for the lentic pattern of the pond, with low risk of damage by solid wastes or high flows, anchored with polypropylene ropes fixed to the banks. The materials used were low cost, reusable, and widely available, obtained from recycling cooperatives. Dried palm leaves were placed on the gallons to protect from sunlight.

The outlet section of pond 1 was fully covered by CFW to avoid short circuiting of flow, as recommended by Lucke et al. [15]. The pond 1 outlet was a small rectangular contracted weir, with 0.39 cm of crest width. The CFW width was 8.8 m in average.

2.3. Macrophyte Growth

Macrophyte growth was monitored four times in terms of leaf height growth and plant density in each raft. After sampling, the leaves were pruned at 60 cm. The roots were sampled four times from one raft. The dry and humid leaf matter was sampled twice. The dry matter was obtained after 5 days in 60 °C, according to Westlake [37].

2.4. Meteorological and Hydrologic Data, Sampling Parameters and Statistical Analysis

Hourly data rainfall, air temperature, and solar radiation were obtained from CIRAM-EPAGRI meteorological station, located 1.5 km far from the CFW project site, from 1 October 2022 to 30 April 2023 [38]. Air temperature and solar radiation were used to analyze macrophyte growth, while rainfall was used to correlate with water quality parameters, flow and HRT.

The samples were analyzed for two different periods: (1) Pre-CFW installation period (1 October 2022–11 November 2023), and after installation or CFW monitoring period (12 November 2022–30 April 2023). Water quality parameters were subgrouped in: water quality 1 parameters—WQ1 (n = 55), comprised by pH, water temperature (WT), electric conductivity EC), and dissolved oxygen (DO)), and water quality 2 parameters—WQ2 (n = 22), comprised by WQ1 plus total phosphorus (TP), turbidity, total solids (TS), ammonium nitrogen (N-NH₄), nitrite N-NO₂, nitrate (N-NO₃), total coliforms (TC) and fecal coliforms (FC) (Figure 3,

Table 1. Collected data and methodology.

Group	Parameters	Unit	Methodology	Sampling Period
				Pre-CFW Installation	Monitoring CFW	Total
Water Quality 1 (WQ1)	Potential of hydrogen (pH)	-	pHmeter	19	32	51
	Water temperature (WT)	°C	Thermometer	24	31	55
	Electrical conductivity (EC)	μS·cm−1	Electrical conductivity meter	24	31	55
	Dissolved oxygen (DO)	mg·L−1	Dissolved oxygen meter	22	32	54
Water Quality 2 WQ2 = WQ1 + follow parameters	Biological oxygen demand—BOD5,20	mg·L−1	Manometric method (APHA)	2	19	21
	Total phosphorus (TP)	mg·L−1	Vanadomolybdophosphoric acid colorimetric method (APHA)	1	20	21
	Turbidity (Turb.)	NTU	Nephelometric method (APHA)	2	29	31
	Total solids (TS)	g·L−1	Gravimetric method (APHA)	2	20	22
	Ammonium nitrogen (N-NH4)	mg·L−1	Nessler colorimetric method (APHA)	2	20	22
	Nitrite (N-NO2)	mg·L−1	Alpha-naphthylamine colorimetric method (APHA)	2	20	22
	Nitrate (N-NO3)	mg·L−1	Brucine colorimetric method (APHA)	2	20	22
	Faecal coliforms (FC)	MPN·100 mL−1	Enzyme substrate coliform test (APHA)	2	18	20
	Total coliforms (TC)	MPN·100 mL−1	Enzyme substrate coliform test (APHA)	2	11	13
Macrophytes parameters	Leaves height	cm	Average height	2	3	5
	Plant density	plants·m−2	Counting all plants per crate	2	3	5
	Dry and humid matter	g.m−2·day−1	Weighting all pruned leaves above 60 cm. Dried at 60 °C for five days	0	2	2
Hydraulic parameters	Flow (q)—elevation head on the weir	L·s−1	Francis’s equation with two contractions for rectangular weir	16	35	51
	HRT—hydraulic retention time	h	Volume.flow−1 (considering root length of 0.25 m)	0	35	35
Meteorolo-gical data	Air temperature	°C	Hourly average data, organized in daily mean, minimum and maximum	41	169	210
	Rainfall	mm	Hourly total data, organized in daily total P24h	41	169	210
	Solar radiation	W·m−2	Hourly average data, organized in daily mean and maximum	41	169	210

). Flow data (n = 51) were estimated by Francis’s equation with two contractions by water head on the rectangular weir (0.39 cm crest width) at pond 1 outlet.

The Shapiro–Wilko test and histogram were used to analyze the normality of each water quality parameter distribution. The effect of pond 1 (P1- > P2) and the effect of CFW (P2- > P3) over water quality were analyzed using paired dependent mean or median statistical test for parametric (t-test) and non-parametric distributions (sign test) [39]. In the CFW monitoring period, sampling parameters were subgrouped into zero rainfall in 24 h (P24h = 0) and rainfall greater than zero in 24 h (P24h > 0). Independent non-parametric tests, the Mann–Whitney (MW) test and the Kolmogorov–Smirnov (KS) test, were used to analyze subgrouped changes influenced by rainfall in the same point (P1- > P1; P2- > P2; P3- > P3). Spearman’s rank order correlation was used to correlate the variations in water quality parameters, hydraulic retention time and rainfall. All tests were analyzed for 0.05 error, and null hypothesis of equality of medians or means.

3. Results

3.1. Macrophyte Growth

First and second rafts were set up in water in winter (07/22 and 08/22) and showed high plant losses. The last fourteen rafts were set up in September (1st, 15th and 27th). A total of 124 death plants were replaced (21%) until 27 September 2022 to keep the initial density of 30 plants/m².

The macrophyte leaves grew an average of 1 cm·day⁻¹ from 20 October 2022 to 15 March 2023. The highest growth of 1.21 cm·day⁻¹ occurred between 13 December 2022 and 25 January 2023, when the solar radiation and temperatures reached higher values, with 257 W/m² and 25 °C in January. The lowest growth of 0.6 cm.day⁻¹ occurred between 10 November 2022 and 13 December 2022, when consecutive ten days of unusual low temperatures for November (polar front) and consecutives low solar radiation periods (high rainfall) occurred (Figure 4). Four rafts showed no leaf growth and reduction in plant density (

Table 2. Average leaf height, growth per day and plant density per raft and sampling.

Date of set up raft in water		23/07	22/08	01/09	01/09	01/09	01/09	15/09	15/09	27/09	27/09	27/09	27/09	27/09	27/09	27/09	27/09
Period (days from start)		0	30	40	40	40	40	54	54	66	66	66	66	66	66	66	66
Raft id Sample data		B16	B15	B13	B12	B11	B10	B9	B5	B8	B7	B6	B14	B4	B3	B2	B1	Avg.
Leaves height per raft (cm)	In 20/10	80	70	70	50	80	40	50	70	30	40	50	50	50	60	50	50	56
	In 10/11	120	100	120	80	100	80	80	80	50	70	60	60	60	70	60	60	78
	In 13/12	120	120	100	80	80	80	80	80	80	70	80	70	60	60	60	60	80
	In 25/01	130	130	110	130	120	100	100	120	110	120	120	120	110	90	80	80	111
	In 15/03	120	120	120	105	100	110	110	120	120	100	130	120	100	90	90	80	108
	Avg. per raft	114	108	104	89	96	82	84	94	78	80	88	84	76	74	68	66	87
Growth (cm/day)	In 10/11	1.9	1.4	2.4	1.4	1.0	1.9	1.4	0.5	1.0	1.4	0.5	0.5	0.5	0.5	0.5	0.5	1.1
	In 13/12	1.8	1.8	1.2	0.6	0.6	0.6	0.6	0.6	0.6	0.3	0.6	0.3	0.0	0.0	0.0	0.0	0.6
	In 25/01	1.6	1.6	1.2	1.6	1.4	0.9	0.9	1.4	1.2	1.4	1.4	1.4	1.2	0.7	0.5	0.5	1.2
	In 15/03	1.2	1.2	1.2	0.9	0.8	1.0	1.0	1.2	1.2	0.8	1.4	1.2	0.8	0.6	0.6	0.4	1.0
	Avg. per raft	1.6	1.5	1.5	1.1	0.9	1.1	1.0	0.9	1.0	1.0	1.0	0.8	0.6	0.4	0.4	0.3	1.0
Plant density (plant/m2)	Start	30	30	30	30	30	30	30	30	30	30	30	30	30	30	30	30	30
	In 20/10	33	20	33	33	33	20	33	20	20	20	25	25	33	33	25	25	27
	In 10/11	33	33	33	33	33	28	33	15	28	18	18	23	18	23	18	13	25
	In 13/12	38	38	38	38	38	38	34	27	28	30	27	28	10	18	18	16	29
	In 25/01	43	43	29	43	33	32	24	23	27	30	25	30	10	18	22	22	28
	In 15/03	43	38	38	38	23	24	31	23	29	38	36	43	23	31	35	26	32
	Avg. per raft	38	34	34	37	32	28	31	22	26	27	26	30	19	24	23	20	28
Description

). This reduction allowed fauna to enter the crates and damage the leaves, by trampling. The most attracted fauna into the rafts was the common gallinule (Gallinula galeata). Other faunas were found around CFW such as the broad-snouted caiman (Caiman latirostris), pond turtle (Trachemys spp.), snowy egret (Egretta thula), or inside the CFW, such as the nest of small Vespula spp. These faunas did not damage the CFW.

The initial plant density of 30 plants·m⁻² reached 32 plants·m⁻² on average on 15/03, with raft density varying from 23 to 43 plants·m⁻², as a consequence of leaf damaging and pruning practices. The rafts set up before 15 September 2022 reached higher densities and leaf heights (highest = 220 cm). Macrophyte resilience was observed in all rafts, with new sprouts growing a few weeks after plant death. The increase in plant density between 25 January 2023 and 15 March 2023 was an average of four plants·m⁻². This growth reached an increase in dry matter for the pruning leaves above 60 cm of 15.3%, from 113 g·m⁻² on 25 January 2023 to 135 g·m⁻² on 15 March 2023 (Table 2).

The CFW roots showed constant development. On 10 November 2022, the start of the monitoring CFW period, the roots spread mainly inside the crates. On 13 December 2022, the roots were 10 cm in length beneath the CFW without biofilm. Additionally, on 15 March 2023 (182 days after set up, on average), they reached 25 cm beneath the crates (0.13 cm/day), with an initial barrier formation by roots and biofilm (

Table 3. Average root growth beneath CFW.

Date	Days since CFW Set up Average Data in Water (14 September 22)	Roots Length Beneath CFW (cm)	Average Growth (cm·day−1)
10 November 22	57	5	0.088
13 December 22	90	10	0.303
25 January 23	133	15	0.349
15 March 23	182	25	0.510

; Figure 5).

3.2. Hydrologic and Hydraulic Parameters

The average total monthly rainfall during this project was 213 mm, with a minimum of 80 mm in January and 560 mm in December. The two highest rainfall (P24h) periods were 20 December 2022 with 207 mm and 30 November 2022 with 78 mm.

The median flow rate in pond 1 outlet was 1.1 L/s (n = 51). A total of 80% of flow data were registered from the minimum of 0.12 L/s and 2 L/s, 17% between 2 L/s and 4 L/s, and only three measures (6%) above 5 L/s. One of these three measures occurred on 30 November 2022 with the maximum of 15.8 L/s. The other two measures occurred on 18/10 and 26/02 with zero rainfall in 2, which were confirmed caused by the overflow of a clogging untreated sewage pipe into stormwater.

3.3. Water Quality Parameters

3.3.1. Pre-CFW Installation (From 1 October 2022 to 11 November 2023)

On Pre-CFW installation period, only water quality 1 parameters were statistically analyzed (n = 19). PH, water temperature and DO were normal distributions, while EC was log-normal distribution. The dependent tests between inlet (P1) and outlet (P3) showed no significant differences over pH and DO. Small significant reductions in EC and a rise of 0.5 °C in water temperature (p = 0.054) showed the lentic pattern influence of pond 1 over water quality.

The overflow of a clogging untreated sewage pipe into stormwater, registered between 17 October 2022 and 21 October 2022, resulted in low DO values, and unusual flow of 9 L/s (Figure 6). Despite the non-significant p-value (0.653) for DO t-test, many results showed an increase in DO into the pond 1 outlet. Most of samples were collected close midday where the water temperatures were higher. The rise of DO in the hottest time was due the photosynthetic activity of phytoplankton and macrophytes [40]. Two superficial macrophytes developed during this project: Pistia stratiotes and Hydrocotyle ranunculoides. On 01/10, these macrophytes covered only 10% of pond 1, while on 04/01, pond 1 was fully covered.

3.3.2. CFW Monitoring Period (From 12/11 to 30/04)

The effect of CFW (P2- > P3) on water quality for the monitoring CFW period showed statistically significant reductions (p-value < 0.05) of medians for the parameters BOD (72%; 7.9 mgBOD_5.20/L), TS (60%; 246 mgTS/L), turbidity (53%; 13 NTU), TP (52%; 0.11 mgP/L) and DO (39%; 1.6 mgDO/L).

The water temperature showed a median reduction of 0.4 °C (1.6%) when passing through the CFW. This behavior is relevant because before the CFW the temperature between P1 and P2 in pond 1 showed a median increase of 0.7 °C (p-value = 0.07). This increase was caused by the incident solar radiation on the water surface (1.235 m²), combined with the shallow water depth (<=1.0 m). The CFW, therefore, functioned as a water temperature regulator. For the pH, there was a reduction of 0.15 (2.1%; p-value = 0.03), from 7.20 in t P2 to 7.05 in P3, also functioning as a regulator of the availability of H⁺.

Electrical conductivity (EC), although significant (p-value = 0.00), increased 8 µS/cm in the median (221 µS/cm in P2 to 229 µS/cm in P3). This slight increase may be associated to permanence of inorganic dissolved solids in the CFW output, parameter not monitored in this research. According to De Sousa et al. [41], electrical conductivity is certainly an indirect indicator of pollution by domestic sewage discharges into water courses, due to its direct relationship with dissolved salts, mainly associated with the high consumption of chloride ion in the diet (Cl⁻ per day, or 15 mgCl⁻·L⁻¹ in raw sewage).

The increase in conductivity may also be associated with the increase in ionized ammonia N-NH₄⁺. Ammonium nitrogen showed a non-significant increase (p-value = 0.17) of 0.43 mgN-NH₄/L (21%), with a median in P2 of 2.02 mgN-NH₄/L and in P3 of 2.45 mgN-NH₄/L. The increase in ammonium may be associated with both ammonia from the sediments and from the ammonification process by the biofilm below the CFW. The ammonification is the first stage of organic nitrogen mineralization, that is, the breakdown of organic compounds with nitrogen to ammonia. It occurs both in aerobic and anaerobic environments, whose sources are tissues and dead cells of plants and animals, also direct excretion of urea and the raw sewage discharge into stormwater [14].

For fecal coliforms (FC), the difference in CFW was not significant (p-value = 0.81) due to the great variability that occurred in March and April. There was a median reduction of 20,000 FC·100 mL⁻¹ in this period, and this reduction or equality was observed in 14 of 19 campaigns carried out. In five campaigns, however, P3 was greater than P2, with the greatest difference of more than 100,000 FC·100 mL⁻¹. The attractiveness of the fauna for the CFW is one of the possible reasons for the great variability of FC, mainly the medium-sized birds inside and around the rafts (common gallinule). Recent studies on the ecology of E.coli in natural environments have shown these bacteria can survive and reproduce for long periods outside the intestinal tract of animals. E. coli may form a biofilm and this protects it from UV radiation, desiccation, protozoan predation, and chemicals including antibiotics and disinfectants [42].

In terms of dissolved oxygen dynamics, the results of DO, BOD, NO₂, NO₃ and HRT indicated an aerobic condition of organic matter degradation, with no theoretical conditions for nitrification in the water column (DO > 0.5 mg·L⁻¹) [43]. NO₂ and NO₃ showed no significant variation in CFW, with zero values in P2 and P3. However, as HRT increased, NO₂ rising (r_s = 0.59) showed significant Spearman’s correlation with HRT.

The DO median concentration reduction of 1.6 mgDO·L⁻¹ was significant (p = 0.00), from 4.2 mgDO·L⁻¹ in P2 to 2.7 mgDO·L⁻¹ in P3, also with great variability between minimums and maximums in P3. The minimum DO concentration in P3 was 1.4 mgDO·L⁻¹ and the maximum 7.2 mgDO·L⁻¹.

Tanner and Headley [44] stated that the reduction in oxygen below the CFW was surprising due to the known release of dissolved oxygen by the roots of macrophytes. However, the reason is linked to the demand for oxygen by the biofilm and sediments below the CFW, in addition to the demand of the roots of macrophytes during the night (breathing). It resulted in a total DO demand higher than the DO available in the water column and to the OD provided by the plants during the day, a condition responsible for the formation of anoxic and anaerobic microzones below the CFW.

The median HRT was 1.64 h, varying between 0.085 h and 11.43 h, considering the root length of 0.25 m. According to Pavlineri, Skoulikidis and Tsihrintzis [16], the total nitrogen removal is directly related to HRT (r = 0.57) and mesocosms studies reported HRTs on the scale of days, with removals of 8% total nitrogen on 1 day to 90% removal in 7 days. De Stefani et al. [19] in Italy showed an average HRT of 17 s and obtained median removals of 52% BOD, 65% TP and 10% of TN.

The HRT of the present research was sufficient to reduce dissolved carbonaceous organic matter (represented by dissolved oxygen consumption and BOD reduction), and sufficient to function as a filter for solid particles, such as reducing turbidity, TP, and TS. Although Spearman’s rank order correlation showed low results for DO and BOD (Figure 7) and no-correlation for TP and TS. The HRT has not correlated with parameters related with solid particles, because there were reductions both at higher flows (low HRT) and at lower flows (high HRT) (Figure 8). This result was also observed by Pavlineri, Skoulikidis and Tsihrintzis [17] with a Pearson correlation (r) of −0.056 between HRT and TP variation.

The variation in TP, TS and turbidity formed a group of positive correlation with each other while HRT, N-NH₄, N-NO₂ and EC formed another correlation group. FC, water temperature and EC formed a third group (

Table 4. Spearman’s rank order correlation of variation in parameters in CFW (P3 less P2), rainfall and flow. Description: Red highlighted p-values were significant for 0.05. The red hatched box indicates a positive correlation. The green hatched box indicates a negative correlation.

	P24h	HRT	Q	pH	EC	WT	DO	BOD	Turb.	TS	TP	NH4	NO2	NO3	TC	FC
P24h
HRT	0.27
Q	−0.27	−1.00
pH	0.00	0.16	−0.16
EC	0.09	0.41	−0.41	−0.02
WT	0.01	−0.09	0.09	0.08	−0.29
DO	−0.14	0.28	−0.28	0.07	−0.06	0.00
BOD	0.28	0.03	−0.03	0.10	−0.14	0.20	−0.01
Turb.	0.34	0.24	−0.24	−0.19	0.09	0.15	0.21	0.37
TS	−0.05	0.00	0.00	−0.06	−0.06	−0.35	0.26	−0.29	0.41
TP	0.33	0.08	−0.08	−0.25	0.21	−0.46	−0.04	0.07	0.60	0.45
NH4	−0.01	0.46	−0.46	0.06	0.07	0.02	0.24	−0.15	−0.20	−0.27	−0.30
NO2	0.03	0.59	−0.59	0.20	0.25	−0.04	0.16	−0.42	−0.23	−0.08	−0.24	0.44
NO3	−0.18	−0.16	0.16	−0.09	−0.01	0.11	0.19	0.06	0.14	0.13	−0.34	0.20	−0.34
TC	0.46	0.26	−0.26	0.00	0.49	−0.27	−0.14	−0.77	−0.31	0.26	0.26	0.37	0.66	−0.70
FC	0.46	−0.41	0.41	0.19	−0.51	0.51	0.15	0.23	−0.07	−0.08	−0.03	−0.01	−0.11	−0.06	−0.60
	Description:

).

In terms of the influence of total rainfall in 24 h over CFW effect parameters (P2- > P3), only nitrate (N-NO₃) showed divergent results for the two subgroups; that is, in the condition P24h = 0, N-NO₃ showed an increase, while in P24h > 0. there was a reduction. However, both results were statistically not significant (p-value > 0.05).

In terms of the influence of total rainfall in 24 h over the same point, only EC, water temperature and N-NH₄ resulted in significant differences at least at one point. Water temperature was 1.1 °C higher with rainfall, in P2 (p-value = 0.055) and 1.0 °C in P3 (p-value = 0.040). For N-NH₄ the concentration in P2 and P3 were both higher with rainfall, however not significant (

Table 5. Statistical tests under null hypothesis of non-differences at 0.05 during CFW monitoring period. Legend: SD—standard deviation. dConc.—absolute median variation on concentration. dLoad—absolute median variation on load = concentration x flow. MW—Mann–Whitney test. KS—Kolmogorov–Smirnov test. Red highlighted p-values were significative for 0.05. The red hatched box indicates significative rising in absolute concentration or load. The green hatched box indicates significative reduction in absolute concentration or load. P1- > P2 statistical test indicates pond effect test. P2- > P3 statistical test indicates the CFW effect test. P1- > P1, P2- > P2 and P3- > P3 tests indicate tests for the same point with rainfall (P24h > 0) and without rainfall (P24 = 0).

			CFW Monitoring Period—All Data Test							Subgrouped Test						Same Point Test
										P24h > 0				P24h = 0		(P24h > 0/P24h = 0)				MW	KS
Parameter		Point	N	Mean	SD	Median	Test	dConc.	p	N	dConc.	p	N	dConc.	p	test	N1	N2	dConc.	p	p
pH		P1	32	7.65	0.9	7.45										P1- > P1	17	15	−0.2	0.497	>0.10
		P2	32	7.38	0.9	7.20	P1- > P2	−0.25	0.003	17	−0.35	0.289	15	−0.20	0.074	P2- > P2	17	15	0.0	0.396	>0.10
		P3	32	7.34	0.9	7.05	P2- > P3	−0.15	0.026	17	−0.15	0.289	15	−0.20	0.248	P3- > P3	17	15	−0.1	0.720	>0.10
EC (µS/cm)		P1	31	261	83	234										P1- > P1	16	15	15	0.252	>0.10
		P2	31	219	31	221	P1- > P2	−13	0.003	16	−14	0.724	15	−12	0.683	P2- > P2	16	15	17	0.038	>0.10
		P3	32	234	43	229	P2- > P3	8	0.000	16	8	0.724	15	12	0.221	P3- > P3	17	15	24	0.290	>0.10
Water temp. (°C)		P1	31	24.0	1.1	24.0										P1- > P1	16	15	0.7	0.031	>0.10
		P2	31	24.4	1.3	24.7	P1- > P2	0.7	0.072	16	0.1	0.131	15	0.5	0.617	P2- > P2	16	15	1.1	0.055	<0.025
		P3	31	24.1	1.1	24.3	P2- > P3	−0.4	0.002	16	−0.2	0.617	15	−0.3	0.371	P3- > P3	16	15	1.0	0.040	<0.05
DO (mg/L)		P1	32	4.3	0.8	4.3										P1- > P1	17	15	0.0	0.835	>0.10
		P2	32	4.2	1.2	4.1	P1- > P2	−0.2	0.591	17	0.0	0.131	15	−0.6	0.683	P2- > P2	17	15	−0.5	0.220	>0.10
		P3	32	2.7	1.2	2.5	P2- > P3	−1.6	0.000	17	−1.6	0.013	15	−1.4	0.041	P3- > P3	17	15	−0.4	0.126	>0.10
BOD (mg/L)		P1	19	4.1	4.9	3.4										P1- > P1	10	9	−4.9	0.246	>0.10
		P2	19	10.9	6.5	11.0	P1- > P2	7.6	0.006	10	5.4	0.289	9	11.7	0.041	P2- > P2	10	9	1.5	0.744	>0.10
		P3	19	4.4	4.1	3.1	P2- > P3	−7.9	0.001	10	−5.5	0.077	9	−9.9	0.221	P3- > P3	10	9	−3.0	0.540	>0.10
Turbidity (NTU)		P1	29	11.2	10.2	9.2										P1- > P1	15	14	−0.8	0.663	>0.10
		P2	29	26.1	16.7	21.9	P1- > P2	12.6	0.000	15	11.7	0.077	14	14.4	0.041	P2- > P2	15	14	1.9	0.570	> 0.10
		P3	29	10.5	5.5	8.7	P2- > P3	−13.1	0.000	15	−12.3	0.013	14	−13.47	0.041	P3- > P3	15	14	0.8	0.616	>0.10
TS (mg/L)		P1	20	300	224	243										P1- > P1	10	10	−30	0.162	>0.10
		P2	20	480	215	464	P1- > P2	221	0.001	10	303	0.077	9	252	0.041	P2- > P2	10	10	−95	0.199	<0.10
		P3	20	215	48	218	P2- > P3	−246	0.000	10	−342	0.013	9	−296	0.041	P3- > P3	10	10	19	0.326	>0.10
TP (mg/L)		P1	20	0.15	0.10	0.14										P1- > P1	10	10	−0.02	0.791	>0.10
		P2	20	0.27	0.18	0.27	P1- > P2	0.13	0.007	10	0.03	0.289	10	0.16	0.683	P2- > P2	10	10	0.11	0.307	<0.10
		P3	20	0.13	0.08	0.16	P2- > P3	−0.11	0.000	10	−0.09	0.013	10	−0.11	0.074	P3- > P3	10	10	0.09	0.140	>0.10
N-NH4 (mg/L)		P1	20	2.23	1.02	2.04										P1- > P1	10	10	0.86	0.031	>0.10
		P2	20	1.95	0.59	2.02	P1- > P2	−0.03	0.502	10	0.10	0.289	10	−0.59	0.221	P2- > P2	10	10	0.17	0.198	>0.10
		P3	20	2.35	0.71	2.45	P2- > P3	0.43	0.169	10	0.58	1.000	10	0.47	0.683	P3- > P3	10	10	0.08	0.307	>0.10
N-NO2 (mg/L)		P1	20	0.25	0.14	0.23										P1- > P1	10	10	0.00	0.910	>0.10
		P2	20	0.10	0.10	0.07	P1- > P2	−0.16	0.001	10	−0.13	0.289	10	−0.19	0.074	P2- > P2	10	10	−0.06	0.129	>0.10
		P3	20	0.09	0.07	0.07	P2- > P3	0.00	1.000	10	0.00	0.683	10	−0.00	0.617	P3- > P3	10	10	−0.07	0.149	>0.10
N-NO3 (mg/L)		P1	20	1.92	1.52	1.88										P1- > P1	10	10	−0.87	0.427	>0.10
		P2	20	0.50	0.72	0.13	P1- > P2	−1.75	0.034	10	−2.32	0.450	10	−1.64	0.683	P2- > P2	10	10	−0.20	0.668	>0.10
		P3	20	0.53	0.96	0.00	P2- > P3	−0.13	0.773	10	−0.21	1.000	10	0.11	0.617	P3- > P3	10	10	0.11	0.562	>0.10
FC (MPN·100mL−1)		P1	19	1.9 × 104	2.7 × 104	8.6 × 103										P1- > P1	10	9	7.4 × 103	0.744	>0.10
		P2	18	4.4 × 104	4.8 × 104	3.1 × 104	P1- > P2	2.2 × 104	0.015	8	3.4 × 104	0.131	9	2.8 × 104	0.221	P2- > P2	8	10	−6.9 × 103	0.477	>0.10
		P3	20	4.0 × 104	6.6 × 104	1.1 × 104	P2- > P3	−2.0 × 104	0.814	8	3.1 × 102	1.000	10	−2.2 × 104	0.221	P3- > P3	10	10	−1.1 × 104	0.571	>0.10
TC (MPN·100mL−1)		P1	11	2.0 × 103	3.0 × 105	1.0 × 106										P1- > P1	8	3	−8.3 × 104	0.850	>0.10
		P2	13	3.0 × 103	8.8 × 105	2.4 × 106	P1- > P2	8.0 × 105	0.131	5	2.1 × 105	0.371	2	1.2 × 106		P2- > P2	6	7	7.5 × 105	0.826	>0.10
		P3	10	8.2 × 103	6.0 × 105	1.7 × 106	P2- > P3	−6.9 × 105	0.371	5	8.0 × 104	1.000	2	−8.0 × 105		P3- > P3	6	4	−3.2 × 105	0.076	>0.10
Description:

).

3.4. Resulting Loading Rates

Hydraulic and mass loading rates are key design parameters for treatment plants [45]. However, these loading rates in CFW field applications in urban streams are not controllable due to variations in flow and concentration of pollutants. Further, the root length increases over time, and the transverse area changes. Comparisons between different projects must consider the loading rates per unit of volume and per unit of surface. The surface unit can also be in terms of the transverse dimension (width × length of roots, or width × water depth) or the top dimension (width × length). This information is crucial for comparison with other floating systems or other effluent treatment systems, and for the evolution of the ecotechnology.

Table 6. Resulting loading rates.

Description	Value	Unit
BOD median input concentration P2	11.0	mg·l−1
BOD median output concentration P3	3.1	mg·l−1
BOD median concentration reduction	72.0	%
w—CFW width average	8.80	m
hr—Roots length	0.25	m
Atr—Transverse area with roots	2.20	m−2
L—CFW Length	2.20	m
hc—Average water depth beneath CFW	0.40	m
Atwc—Transverse area watercourse	3.52	m−2
Asurf—CFW Surface area	19.2	m−2
hr/hwc—Relation root length by water depth	62.5
Qwc—Median flow	71.0	m3·day−1
BOD load	779	g·day−1
Vr—Roots volume	4.84	m3
Vwc—Water volume	7.74	m3
OLRtr—Organic transverse loading rate (roots)	354.2	gBOD·m−2·day−1
OVLRvr—Organic volumetric loading rate (roots)	73.2	gBOD·m−3·day−1
HLRtr—Hydraulic transverse loading rate (roots)	32.2	m3·m−2·day−1
HLRvr—Hydraulic volumetric loading rate (roots)	14.6	m3·m−3·day−1
HRTr—Hydraulic retention time (roots)	1.64	h
OLRtwc—Organic transverse loading rate (watercourse)	221.4	gBOD·m−2·day−1
OLRvwc—Organic volumetric loading rate (watercourse)	100.6	gBOD·m−3·day−1
HLRtwc—Hydraulic transverse loading rate (watercourse)	20.1	m3·m−2·day−1
HLRvwc—Hydraulic volumetric loading rate (watercourse)	9.1	m3·m−3·day−1
HRTwc—Hydraulic retention time (watercourse)	2.62	h
vwc—average flow (watercourse)	0.84	m·h−1
OLRsurf—Organic surface loading rate (top)	40.6	gBOD·m−2·day−1
HLRsurf—Hydraulic surface loading rate (top)	3.69	m3·m−2·day−1

presents the resulting loading rates during the CFW monitoring period.

4. Conclusions

The low-cost field-scale CFW applied for the revitalization of the urban stream in a lentic section showed statistically significant reductions in water quality parameters and contributes to the development of this ecotechnology in subtropical climates.

The watercourse full-width coverage, the CFW length, the type, the initial density, and the resilience of emergent Typha dominguensis, and the surface BOD loading rate applied were determining factors in the results found. The HRT was sufficient to reduce dissolved carbonaceous organic matter and solid particles-related parameters showed significant reductions both on high and low HRT.

The resulting loading rates can be used for the design purposes of similar CFW field applications.

This ecotechnology can be used in conjunction with all NBS principles for the depollution of river basins in urban areas.