The Effectiveness of an Artiﬁcial Floating Wetland to Remove Nutrients in an Urban Stream: A Pilot-Study in the Chicago River, Chicago, IL USA

: Ever expanding urbanized landscapes are increasingly impacting streams that run through them. Among other stressors, urban streams often are host to elevated concentrations of nutrients, salts, and heavy metals. The pollutants, coupled with high temperatures, are drivers of ecosystem degradation in urban streams. The installation of artiﬁcial ﬂoating wetlands (AFWs) has been successful in mitigating the effects of urbanization in lakes and wastewater treatment ponds, but rarely have they been tested in streams. This pilot-study examined the ability of an AFW to improve water quality in an urban stream. The small, 90 m 2 AFW was installed to improve the aquatic habitat and aesthetics of a small section of the Chicago River, Chicago, IL USA. Water samples and in-situ measurements were collected from the surface and at 0.3 m depth of upstream and downstream of the AFW. Samples were analyzed for nitrate-as-nitrogen, phosphate, chloride, and heavy metals. Comparison of upstream and downstream waters showed that the AFW lowered the concentrations of nitrate-as-nitrogen and phosphate during the growing season by 6.9% and 6.0%, respectively. Nitrate was also removed during the dormant season; however, phosphate was not removed during that time. Plant or microbial uptake of the nutrients are believed to be the dominant mechanisms in the growing season with denitriﬁcation serving as the primary pathway in the dormant season. Despite not having a measurable effect on the water temperature, the AFW was an effective means to reduce concentrations of nitrate and phosphorus, decreasing the potential for eutrophication.

Nitrogen and phosphorus, are key nutrients for plant growth and development of many photosynthetic organisms, and elevated concentrations of both contribute to eutrophication and hypoxic conditions in waterways [16][17][18]. Notorious algal blooms, such as the ones witnessed in Lake Erie [19], frequently occur at the mouths of waterways that drain urban and agricultural lands. Mitigating and preventing eutrophication requires the reduction of nutrients from the waters before algal growth can begin to negatively impact the proper functioning of ecosystems [20,21].
to improve the plant uptake potential by stimulating more plant growth [51,59]. While plants are most active during the summer, the role of nitrate abatement by plants in the winter has been observed in other environments, i.e., a saturated buffer [61]. The yearround presence of microbial biofilms could potentially provide an assimilation of the denitrification pathway even when the plants are dormant.
Several studies have proven that AFWs are effective on stagnant waters in removing nitrogen, phosphorus, and heavy metals from ponded water [8,50,62]. Many previous studies conducted on AFWs did so in different ecosystems and employed different species to act as phytoremediators, but it is significant to note that they unanimously observed the effects of AFWs in controlled systems. The present study will provide insight into the benefits of AFWs on flowing rivers. This work investigated the effectiveness of a small (pilot) AFW to remediate an urban stream in a small section of the Chicago River. The pilot AFW was designed to improve the aquatic habitat and enhance the aesthetics of the stream segment. However, we examined whether an AFW provided any in-situ benefits by addressing the following questions: (1) Does the AFW decrease nutrient concentrations, nitrate and phosphate, within the water column? (2) Are chloride concentrations of waters upstream from the AFW greater than chloride concentrations of waters downstream from the AFW? (3) Are heavy metal concentrations of waters upstream from the AFW greater than heavy metal concentrations of waters downstream from the AFW? (4) Are there seasonal differences in the effectiveness of an AFW to remove dissolved solutes?

Study Location
This study site was located in a side canal on the north branch of the Chicago River flowing adjacent to Goose Island in Chicago, Illinois, USA ( Figure 1). The canal was constructed in the 1870s and for over a century hosted heavy industry along its banks. The commercial and industrial value of the canal meant it had undergone periods of channelization, widening, dumping, and dredging. Because of restrictions of commercial boat traffic and deprioritization by the Army Corps of Engineers, dredging of the canal has stopped, and several meters of fine, loose sediment sit atop a hard clay bed. The canal ranges from 1 m deep at the northern end to 2.5 m deep at the southern end and has a width of 24-37 m. In the portion of the Chicago River located around Goose Island, impermeable areas of urban landscape drain directly into the river. Two CSO outfalls near the north (upgradient) end of the canal can discharge into the canal during high rainfall events, sometime occurring with as little as a few centimeters of rain in the area.
Along the eastern edge of the canal, around 90 m 2 , approximately 3 m by 30 m, of AFWs were installed in 2017, hosting roughly 2000 plants representing 50 unique species native to Illinois ( Figure 1B,C). Installed to improve aquatic habitat and the aesthetics of the canal, the AFWs were constructed from interconnected tubes of coconut husk that provide a buoyant substrate for various plant species (Table 1; Figure 2). While some roots extended greater than 1 m into the water column, most roots grew to a depth of 0.5 m ( Figure 1D). The gardens were anchored to the eastern bank, which is entirely cement and steel lined. The bank opposite is mixed, with steel seawall and concrete riprap intermingled with natural banks with trees and plants growing along the riparian zone. Water typically enters the canal from the northwest, branching off from the river, and flowing to the southeast until it rejoins the north branch approximately 1.5 km south. Based upon the last 67 years of water data collected by the USGS, the canal has a mean discharge of 0.88 m 3 /s. The primary purpose of these AFWs is to assess improvement in aquatic habitat and species as the artificial habitat will provide an ecosystem for aquatic life and introduces both allochthonous and autochthonous organic matter to the river.

Water Chemistry
River water was collected upstream of the garden and downstream of the garden at two depths during sample events: at the surface and at 0.3 m below the surface. The 0.3 m depth was selected to capture waters that flowed through the root zone. From each depth, two samples were collected with a horizontal sampler before being filtered through a 0.4-µm membrane filter and transferred into acid washed 30 mL sample bottles. Samples to be processed for heavy metals were acidified to a pH of 2 with concentrated sulfuric acid. All samples were frozen prior to analysis. In-situ measurements of dissolved oxygen (DO) (mg/L), temperature ( • C), and specific conductance (SpC) (µS/cm) were recorded using a YSI-85.  Figure 1B,C). Installed to improve aquatic habitat and the aesthetics of the canal, the AFWs were constructed from interconnected tubes of coconut husk that provide a buoyant substrate for various plant species (Table 1; Figure 2). While some roots extended greater than 1 m into the water column, most roots grew to a depth of 0.5 m ( Figure 1D). The gardens were anchored to the eastern bank, which is entirely cement and steel lined. The bank opposite is mixed, with steel seawall and concrete riprap intermin-  [63]. Heavy metal concentrations (Al, As, Be, Cd, Cr, Cu, Pb, Mn, Se, Zn) were quantified using a PerkinElmer Optima 8300 Inductively Coupled Plasma Atomic Emission Spectrometer (ICP-AES). Quality assurance (QA) and quality control (QC) was maintained during analysis of each sampling event by running blank, duplicate, and replicate samples. The analytical error was less than 3%. the last 67 years of water data collected by the USGS, the canal has a mean discharge of 0.88 m 3 /s. The primary purpose of these AFWs is to assess improvement in aquatic habitat and species as the artificial habitat will provide an ecosystem for aquatic life and introduces both allochthonous and autochthonous organic matter to the river.   One-sided paired t-tests, α = 0.05, were used to assess the null hypothesis that upstream concentrations (or temperature) were equal to the downstream concentrations (or temperature) for the given ion, i.e., the difference in concentration will be zero (0). The alternative hypothesis was that the upstream ion concentration was larger than the downstream ion concertation. Statistical analyses were conducted on a complete data set and for the time periods representing the growing season (May-August) and the non-growing season (September-April).

Results
Between 29 April 2018, and 19 November 2019, 39 sampling events were completed. Of the 39 sampling events, 21 occurred during the growing season and 18 were completed during the non-growing season. The data that support the findings of this study are openly available (as a csv file) in Faculty Publications-Geography, Geology, and the Environment at https://ir.library.illinoisstate.edu/fpgeo/3 [64].

Nutrients
Nutrients, nitrate as nitrogen (NO 3 -N) and phosphate (PO 4 3− ) were observed in the Chicago River waters (Figure 3). Upstream of the AFWs, measured NO 3 -N concentrations in the surface waters ranged from 2.14 to 10.

Chloride
Chloride concentrations exhibited strong seasonal variability (Figure 3

Heavy Metals
For all of the heavy metals analyzed, we observed BDL concentrations, which were below the United States Environmental Protection Agency (USEPA) maximum contaminant levels (MCL) and the secondary maximum contaminant levels (SMCL) for the metals [65] ( Table 2). Thus, the concentrations for Cu, Pb, and Zn detected in this study were lower than what was found by Komínková, et al. [66] at combined sewage overflow sites, which implies that the four CSO locations on the canal are not significantly contributing to heavy metal contamination to this section of the Chicago River. The absence of such metals in the Chicago River could be influenced by the chemistry of the river itself, the chemistry of the individual metals, or the surface area and surface charge of sediments in the river. In the neutral to alkaline waters of the Chicago River, the solubility of metals is low. These metals may precipitate onto either sediments or suspended particles in the water [67], which could help to explain the absence of dissolved phases of the metals in the water. The metal concentrations measured for this work are consistent with those reported by the Chicago Metropolitan Water Reclamation District (CMWRD). The CMWRD has collected heavy metal and chloride data monthly at locations 3.2 km upstream on the North Branch of the Chicago River and 4.8 km downstream on the South Branch of the Chicago River of the canal. In 2018, monthly heavy metal concentrations for As, Cd, Cr, Cu, Pb, Mn, Se, and Zn were less than 2 µg/L on both the North and South Branches of the Chicago River. The lack of heavy metals in the river water is a positive; however, the role of the AFWs in the concentration of heavy metals at this section of the river could not be assessed.

Dissolved Oxygen
During

Temperature
For the growing season, the mean upstream and downstream surface water temperatures were 22.5 • C and 22.4 • C, respectively. For the dormant season, the upstream and downstream mean temperatures were 11.5 • C and 11.5 • C, respectively (Figure 4). No statistical difference was noted in the surface water temperatures during either season: growing (p = 0.09) or dormant (p = 0.24). At the 0.3 m depth, the mean temperatures upstream, 22.1 • C, and downstream, 22.1 • C, were not different (=0.36) during the growing season. Similarly, in the dormant season no difference was noted between the mean temperatures upstream, 11.5 • C and downstream, 11.5 • C.

Temperature
For the growing season, the mean upstream and downstream surface water temperatures were 22.5 °C and 22.4 °C, respectively. For the dormant season, the upstream and downstream mean temperatures were 11.5 °C and 11.5 °C, respectively (Figure 4). No statistical difference was noted in the surface water temperatures during either season: growing (p = 0.09) or dormant (p = 0.24). At the 0.3 m depth, the mean temperatures upstream, 22.1 °C, and downstream, 22.1 °C, were not different (=0.36) during the growing season. Similarly, in the dormant season no difference was noted between the mean temperatures upstream, 11.5 °C and downstream, 11.5 °C.

Discussion
The importance of plants and photosynthesis to nutrient reduction have been well documented in wetlands [68,69], in the hyporheic zone [70][71][72][73], in riparian forests [74][75][76][77], and in saturated buffer zones [61,78]. AFWs have been introduced into lake and stream systems to provide aquatic habitat, to enhance the aesthetics of aquatic systems, and to improve water quality [42][43][44][45]. The hydroponic nature of the AFWs provides a unique treatment mechanism; the combination of plant uptake and microbial transformations by biofilms on the roots that can contribute to decreases in nutrient concentrations [41][42][43][47][48][49]. In an AFW, the structural floating pontoons and plant root systems reduce the velocity of the water, allowing for extended interaction with the roots and the biofilms growing on the roots [47,79,80]. As the AFW matures, an extensive network of roots and attached biofilms develop within the water column [41][42][43]49].
While the primary purposes of the pilot garden on the Chicago River was to enhance aquatic habitat and the aesthetics of a formerly industrial canal, the presence of the AFW had a positive impact on the water quality. The AFW reduced the nitrate and phosphate concentrations during the growing seasons and the nitrate concentrations during the dormant season. During the growing season, downstream nitrate concentrations in the surface water were 6.9% lower than the upstream concentrations, while the phosphate concentrations were reduced 6.0%. In the dormant season, the waters at 0.3 m exhibited a significant decrease of 6.8% in nitrate, while the loss of nitrate in the surface water was smaller, at 4.2%. The observed reduction rates were much lower than rates observed in other AFW, 36.9% and 64.5% for nitrate and phosphate, respectively [81], but this may be a function of the residence time of the water within the root network of the garden or other factors specific to any study's particular materials and methods. In stream ecosystems, the loss of nitrate via uptake, denitrification, or microbial assimilation is thought to occur disproportionately in zones with long residence times that facilitate contact of reactive solutes with high biotic capacity for biogeochemical processing [82,83].
Despite a surface area of only 90 m 2 , the AFWs exhibited the ability to reduce nitrate and phosphate, which can limit eutrophication [52,53]. However, the residence time of the water within the root network is limited, and the footprint of the AFWs was small compared to the volume of water passing through as indicated by the minimal impact on the water temperature and the dissolved oxygen of the waters. Studies in systems with longer residence times for waters in the root system observed decreases in water temperature, up to 2 • C [45,84]. With an average discharge of 0.88 m 3 /s and a crosssectional area of 60 m 2 , the velocity for the Chicago River calculates to 0.015 m/s. With a length of 30 m, the waters would interact with the garden a minimum of 2000 s (33.3 min). The calculated residence time is a conservative estimate given that the roots will provide resistance, decreasing the velocity. The velocity coupled with the mass and heat capacity of the water limit the effectiveness of the garden to ameliorate the effects of urban heating on the stream waters. Increasing the size of the AFW would increase the residence time, which would further enhance the nutrient reduction and may allow for enhanced shading to decrease the water temperature.
Seasonally, the difference between upstream and downstream nitrate concentrations in the growing season, 6.8%, was slightly higher than during the dormant season, 4.2%, and a significant difference in phosphate uptake was only noted during the growing season. The higher reduction rates in the summer are consistent with those reported by [81], which they attributed to a combination of plant uptake and denitrification. Studies agree, however, that denitrification and vegetation uptake can occur simultaneously or independently depending on environmental conditions that may change within or between seasons [76,85,86]. However, the concomitant reductions in both nitrate and phosphate coupled with DO concentrations above levels suitable for dentification, suggest plant or microbial uptake as the primary mechanism of nutrient reduction. Seasonal changes in solar radiation influence growth of aquatic plants and algae by controlling photosynthesis [70,71,87]. While photosynthesis creates a complex set of interactions, the highest rates of plant and algal NO 3 -assimilation are reported during periods of greater sunlight [61,88], which occur during the growing season. Denitrification is also dependent upon the temperature of the environment [89]. The optimum temperature for denitrification is 30 • C, which was close to the average water temperature in the growing season at the pilot-scale site. Colder temperatures do not preclude denitrification from occurring but do tend to decrease microbial activity [90]. Denitrification has been observed during winter months in soils [75,91,92] and in groundwater [61,85]. The observed losses of nitrate during the dormant season for the AFW suggest that the biofilms on the roots may provide microsites suitable for denitrification despite the DO concentrations above 4.5 mg/L. Chloride concentrations at the study site did not experience significant changes upstream or downstream during either season. However, seasonal differences were detected with the concentration increasing from 117.9 mg/L to 124.8 mg/L in the growing season to 165.8 mg/L to 171.2 mg/L in the dormant season. Snow events and freezing temperatures occurred during the week of 5 November 2019 and road salt application took place throughout the city, contributing to the seasonal increase in Cl − concentrations. While some plants may incorporate Cl − [93], the lack of a significant decrease in Cl − during the growing season suggests that the plants are not removing the Cl − . However, the lower Cl − concentrations observed in the water exiting the AFW in the dormant season indicate some type of uptake is occurring, but the identifying the pathway is beyond the scope of this work.

Conclusions
A small, 90 m 2 AFW installed on the Chicago River reduced the concentrations of nitrate by 6.9% and phosphate by 6.0% in the river waters during the growing season. While phosphate was not removed in the dormant season, nitrate and chloride both exhibited reductions in concentration, 6.8% and 3.6%, respectively. The AFW was not effective in regulating the temperature of the water, but this may be a function of limited shading, approximately 30 min, as the waters travel below the garden for approximately 30 min. Despite the primary design of the AFW being to improve the aquatic habitat and the aesthetic, the AFW showed promise in improving water quality.

Data Availability Statement:
The data that support the findings of this study are openly available in Faculty Publications-Geography, Geology, and the Environment at https://ir.library.illinoisstate. edu/fpgeo/3 [64]. (accessed on 1 April 2021).