Floating wetland islands (FWIs) were tested in Pasco County, Florida, as a method of reducing total nitrogen (TN) in reclaimed water during reservoir storage. The Pasco County Master Reuse System (PCMRS) is a regional reclaimed-water transmission and distribution system providing wastewater effluent disposal for the county. Total daily mass loading from reclaimed water is limited by nitrogen content in the PCMRS watershed. To test TN reduction efficacy, 20 FWIs were constructed, installed, and monitored in a lined pond receiving PCMRS reclaimed water. In total, 149 m2 of FWIs were installed, distributed as a connected network covering 1,122 m2, or 7% of pond area. Pond hydraulic residence time averaged 15.7 days. Treatment performance was assessed during three consecutive periods: establishment (first 6 months of grow-in), performance (8 months immediately following grow-in), and control (3 months after the FWIs were removed from the pond). The FWIs enhanced pond nitrogen removal capacity by 32%. The primary effect of the FWIs was to decrease organic nitrogen in the pond outflow. By evaluating the difference between the performance and control periods, an incremental TN removal rate for the FWIs was calculated to be 4.2 kg N/m2 FWI per year.
The Pasco County Master Reuse System (PCMRS) is a regional reclaimed-water distribution system providing the sole wastewater effluent management mechanism for Pasco County, Florida, USA. This total reuse strategy is accomplished by the beneficial reuse of effluent from all wastewater treatment facilities (WWTFs) in Pasco County via a combination of irrigation customers and rapid-rate infiltration basin systems. In addition, the PCMRS includes a 235 megaliter (ML) storage pond (Lake Rita), an existing 379 ML reservoir at the Land O' Lakes WWTF, and an additional 1,893 ML reservoir under construction.
The Tampa Bay Nitrogen Management Consortium has developed total maximum daily loads for Tampa Bay. The PCMRS operates in the Hillsborough Bay watershed, a tributary to Tampa Bay. The 2012 Tampa Bay Reasonable Assurance Submittal (Tampa Bay Estuary Program 2012) includes a total nitrogen (TN) load limit for PCMRS discharges to the Hillsborough Bay Basin. The allocated basin load is 5.3 tons per year. Since the load was determined based on prior year loading (2008), the County already discharges its allocation during normal operations. The nitrogen in the PCMRS flow is predominantly in the form of nitrate.
Approximately 30% of the reclaimed water produced by Pasco County is reused in the Hillsborough Bay drainage basin. This basin also represents the area where most of the future growth in reclaimed-water customers is projected. Because the County is constructing large reclaimed-water storage reservoirs, installing FWIs for passive reduction of nitrogen could enable greater use of reclaimed water in nitrogen-limited areas.
Floating wetland islands (FWIs) offer a technology for improvement of surface water quality in existing or constructed water bodies where the water is being stored and conveyed. FWIs utilize emergent wetland species growing on a consolidated floating mat (Headley & Tanner 2006). Dodkins & Mendzil (2014) reviewed the literature and concluded that FWIs improve phosphorus removal by 2–55% and nitrogen removal by 12–42% relative to controls. A pilot experiment using FWIs in Hungary with additions of 5 mg/L of oxidized nitrogen showed an 85% reduction of TN (Headley & Tanner 2006). Van de Moortel et al. (2010) reported that floating mats significantly increased TN reduction in combined sewer overflows by an average of 33% relative to controls. Chang et al. (2013) reported a net increase in TN reduction in stormwater attributable to floating treatment wetlands. Borne et al. (2013a) attributed enhanced denitrification in nitrate-rich stormwater to low dissolved oxygen and increased organic carbon availability in the root zone below the FWIs when compared with a control pond without FWIs.
To investigate the efficacy of FWIs in reducing TN, FWIs were constructed and monitored in a pond receiving reclaimed water from the PCMRS for a period of 18 months. Reclaimed water was applied at rates designed to create relatively short hydraulic residence times (HRTs) consistent with actual reservoir residence times.
METHODS AND MATERIALS
Twenty FWIs were constructed and established within an existing 1.6 hectare plastic-lined pond at the Wesley Center WWTF. Reclaimed water from the PCMRS was delivered to the pond through a temporary pipe. Pond overflow spilled to the WWTF's adjacent reject pond, which was then pumped back to the WWTF headworks.
The FWIs were BioHaven™ floating mats purchased from Martin Ecosystems, Inc., Baton Rouge, LA, USA (www.martinecosystems.com). Each mat measured 2.4 × 3.0 m. The surface area of FWIs totaled 149 m2. The FWIs were interconnected with stainless steel cables through PVC pipes and distributed in a network of FWIs covering a total area of 1,122 m2, or 7% of the total pond area (Figure 1). Weighted anchors resting on the pond bottom held the FWIs in their target configuration after initial installation. The pond bottom sloped from approximately 1 m deep near the west end to 2 m deep near the east end. The FWIs were located near the outfall where the pond bottom is approximately 1.5 m deep.
Wetland plant species selection
Wetland plants comprising 18 local species were obtained as potted plants and bare root propagules from local commercial nurseries. A local seed mix was applied to two islands to compare with the 18 planted islands.
The pond volume totaled approximately 19 ML. A temporary 10 cm diameter pipe with a flow meter was installed on the west side of the pond near the southwest berm corner to provide continuous flow. Meter readings were recorded daily by WWTF operators. The fixed-elevation outflow pipe provided continuous water level control and eliminated the need to monitor the pond stage. Recorded flows and pond stage–storage relationships were used to complete a daily water balance of the pond.
Water quality sampling
Water samples were collected every 2 weeks at the pond inflow and outflow and analyzed for ammonia nitrogen, oxidized nitrogen (nitrate plus nitrite), organic nitrogen (ON), and TN for three distinct project phases: the grow-in period (July 2012 through December 2012), the performance period (January 2013 through August 2013), and the control period after the islands were removed (September 2013 through November 2013). Laboratory analysis was performed according to Standard Methods SM-4500 (APHA 2000) by the Pasco County Environmental Laboratory, which is certified by the Florida Department of Health Laboratory Certification Program in accordance with the National Environmental Laboratory Accreditation Conference.
Tissue sample collection
Tissue analysis of the planted vegetation was performed to quantify plant nutrient uptake during the study period. Plants were harvested quarterly and analyzed for dry weight and percent TN. During FWI installation, six tissue sampling sites were placed evenly throughout each FWI. During each sampling event, at least one sample was collected from all islands. Plants for tissue sampling were selected randomly.
Samples were collected in September 2012, November 2012, April 2013, and August 2013. Root length, shoot length, and media depth (island matrix) were measured upon collection. Plant samples were packaged in coolers and shipped to the University of Florida Wetland Biogeochemistry Laboratory, Gainesville, FL, for analysis in accordance with Standard Methods (APHA 2000).
A total of six sites were prepared before planting for tissue samples. These included the use of a small section of 75 cm polyvinyl well screen installed through the mat for easy removal. Sites were planted randomly and in the same way as the rest of the mat. One sample from each island was collected quarterly from the pre-prepared sampling sites. The root zone of each sample was washed to remove all soil, packaged, and sent to the University of Florida Wetland Biogeochemistry Laboratory for analysis of TN. The minimum detection limit and practical quantitation limit for TN were 0.23 mg/kg and 0.27 mg/kg, respectively.
Ten samples of calcified algal deposits were taken at the study conclusion, each with a surface area of 0.37 m2. Groups of two samples were taken at 9 m increments inward of the west berm. Within these groups, the first sample was taken between 23 and 30 m north of the first sample in the previous sample group. The second sample was taken at varying distances north or south of the first sample. Samples were analyzed for wet weight and TN.
Lithium chloride tracer study
After 11 months of island establishment, a tracer study was performed involving a one-time slug application of lithium chloride at the pond influent pipe. Lithium ion was monitored at the pond outfall in order to determine the time and concentrations of exiting tracer.
Floating wetland islands removal and relocation
The FWIs were removed from the pond at the Wesley Center WWTF on 27 and 28 August 2013. The FWIs were fully grown and saturated with water at the completion of the study. The FWIs were relocated and permanently installed at the Lake Rita reclaimed water storage facility.
Pond hydraulic characteristics
The nominal hydraulic residence time (nHRT) averaged 25 days based upon an average daily inflow rate and an estimated pond volume at the operating water depth. The nHRT varied from 5.6 days after extreme rain events to 158 days during short periods when the flow to the pond was halted for operational reasons outside of the control of this study.
To characterize pond hydraulics, the tracer response curve was analyzed for the number of tanks in series (N), dimensionless variance, wetland dispersion number, Peclet number, and volumetric efficiency. Residence time distribution was analyzed as a first-order gamma distribution to determine N and mean residence time as described in Kadlec & Wallace (2009).
The pond N was determined to be 1.04, just greater than 1, the value for a continuously stirred tank reactor (CSTR). The wetland dispersion number was 8.0. Highly sensitive to N values in this range, pond dispersion values are typically higher, indicating rapid lateral and reverse dispersion as inflow water enters the lined pond, in contrast to wetlands, which typically range much lower from 0.07 to 0.33 and are consistent with laminar flow (Kadlec & Wallace 2009).
The measured HRT for the pond averaged 15.7 days. The volumetric efficiency, calculated by dividing the actual HRT by the nominal HRT, was 0.63, meaning that the actual HRT was less than expected, presumably through preferential flow. The tracer response curve showed tracer exiting the pond within 4 hours after application, along with a long descending limb, indicating the presence of both short-circuits and ‘dead’ zones. The presence of algae and floatable debris in the northeast and northwest pond corners indicated stagnant water in these areas. A Peclet number (Pe) of 0 represents a CSTR, and Pe = ∞ represents a plug-flow reactor. Reported values for free water surface wetlands range from Pe = 5 to 20 (Kadlec & Knight 1996). The Pe for the test cell was 0.13, which indicates significant short-circuiting.
With a duration of 90 days, the control period was approximately equivalent to six measured HRTs. Given the general understanding that three residence times is considered sufficient to characterize a tracer impulse (Kadlec & Wallace 2009), the data collected during the control period represent the pond conditions with no lingering effect from the islands.
Plant growth response
In September 2012, 5 months after planting, plant shoot length averaged 94 cm (range: 30–157 cm) and root length averaged 28 cm (range: 10–46 cm). In August 2013, just before the islands were removed, root and shoot lengths were remeasured. Shoot length averaged 132 cm (range: 9–238 cm) and root length averaged 38 cm (range: 6–305 cm).
Nitrogen water quality monitoring results
Inflow TN averaged 6.1 mg/L and ranged from 3.4 to 9.6 mg/L (Figure 2). Average (±1 standard error) and range of TN concentration reductions were 54 ± 5% (range: 43–69%) during the grow-in period, 67 ± 7% (range: 31–83%) during the performance period, and 25 ± 12% (range: 6–35%) during the control period.
Oxidized nitrogen (nitrite + nitrate) inflow concentrations for the pond averaged 5.3 mg/L and ranged from 2.8 to 7.8 mg/L (Figure 3). In general, the outflow concentrations were <1.5 mg/L, with most measurements below detection during all monitoring periods.
Ammonia concentrations were either below the detection limits of the analytical method or between the detection limit and the practical quantitative limit. Based on the results for ammonia, nitrogen species conversions involving ammonia were considered negligible.
Approximately 87% of the inflow nitrogen form throughout the study was oxidized nitrogen, while the dominant outflow nitrogen form was ON. Inflow ON concentrations averaged 0.6 mg/L and ranged from non-detected to 1.8 mg/L (Figure 4). Outflow ON concentrations were significantly greater, averaging 2.2 mg/L and ranging from 0.4 to 4.0 mg/L. The difference between inflow and outflow ON indicates the conversion of inorganic nitrogen to ON. Average (±1 standard error) outflow ON concentrations for the grow-in period, performance period, and control periods were 3.0 ± 0.2 mg/L, 1.5 ± 0.3 mg/L, and 2.9 ± 0.4 mg/L, respectively.
This difference in pond performance is not related to differences in season or system hydraulics. During this time, monthly average temperatures showed similar ranges for the performance (19.2–29.2 °C) and control (21.8–30.7 °C) periods, and average period temperatures were not statistically significantly different (p = 0.28). The decreased concentration of ON in the pond outflow during the performance period and the subsequent increase after the removal of the islands points to reduction in algal growth in the presence of the islands.
Tissue sampling results
Total estimated vegetation mass per island was calculated by averaging the plant samples collected for each event, multiplying the average by the number of plugs per FWI, and then summing the totals for each zone together. Average TN sample concentration was used to estimate the mass of TN removed that can be attributed to plant uptake. Based on these estimates, approximately 2.2 kg of TN was estimated to be bound up in plant tissue mass, which accounts for 0.2% of the TN removed.
Algal productivity and deposits
Throughout the study, algae flourished in the pond and left calcified deposits. Cyanobacterial calcium carbonate precipitates when pH increases in response to photosynthesis (Riding 2011). Algal deposits were limited to shallower depths of the pond. The highest line of calcification was a reliable high-water level indicator for this study.
Calcified deposits on the pond liner were sampled at the end of the study and were estimated to have accumulated a total of 96 kg TN over the course of the study.
Nitrogen mass balance and transformation
The pond nitrogen cycle, performance of each nitrogen species, and the potential for FWIs to reduce TN concentrations within reclaimed-water storage facilities was assessed. Figure 5 shows the principal components of the nitrogen cycle in the pond with the FWIs. Principal processes transforming nitrogen in aquatic systems included ammonification, nitrification, denitrification, plant uptake, algal assimilation and burial. A nitrogen mass balance was estimated for the performance and control periods by quantifying pond input and output, and the storage in plant and cyanobacterial biomass, and estimating the denitrification rates based on the water balance and water quality monitoring.
During the performance period, 61% of the nitrogen mass was removed, of which 56% was estimated as loss by denitrification (Figure 5, left). Based on the biomass samples, system storage was estimated to be approximately 4.3 g N/m2 × year, corresponding to 4.2 g N/m2 × year in calcified cyanobacteria and 0.1 g N/m2 × year in plant tissue. During the establishment period, system storage corresponded to 5% of TN lost. Only 0.2% of the TN removed during this period was attributed to plant uptake and storage, a calculable but negligible amount.
In the control period, nitrogen mass removal was 13.9 g N/m2 × year, a 30% reduction, of which 23% was estimated to be gasification or denitrification (Figure 5, right). Based on the liner cyanobacteria samples, approximately 4.2 g N/m2 × year accumulated in system storage. Because the islands were removed during this period, plant uptake is removed from the mass balance. During the establishment period, system storage corresponded to 7% of TN removed.
Almost all (96%) of the nitrogen in the pond outflow was composed of ON, indicating that algal uptake was responsible for converting the remaining nitrate to ON in algal biomass, which was then exported in the pond outflow. By removing the FWIs, the mass balance suggests that the ability of the system to denitrify and convert oxidized nitrogen to ON in the absence of FWIs was reduced, and the majority of the nitrogen was exported as algal solids in the outflow.
By comparing the denitrification rates of the control period with the performance period, a total mass reduction of 630 kgN/year can be attributed to the FWIs. This corresponds to a mass removal rate of 4.2 kgN/m2 FWI mat × year. This value is greater than 90% of the range of TN removal rates observed for free water surface treatment wetlands (Kadlec & Wallace 2009). When calculated for the area-defined FWIs and intermediate open areas (1,122 m2), the removal rate is 562 gN/m2 FWI area × year, corresponding to the 80th percentile of treatment wetlands (Kadlec & Wallace 2009). The greater reduction of ON (in the form of algal solids) during the performance period may be related to attachment of suspended solids to root surface areas, as noted by Borne et al. (2013b). Enhanced denitrification may also be attributed to the FWIs, given the greater oxygen depletion in floating treatment wetlands, particularly in the presence of plants (Tanner & Headley 2011).
The results from this study indicate that the TN mass removal efficiency of the reclaimed-water pond was 61% during the performance period with FWIs present, compared to 30% mass removal efficiency during the control period with the FWIs removed. Based upon the difference in TN performance during the performance and control periods, the FWIs accounted for 32% more reduction of TN. Relatively high TN removal rates of 4.2 kg/m2 FWI mat × year or 562 gN/m2 FWI area × year were estimated. Tracer test results suggest that performance might be improved if a pond were designed for greater hydraulic efficiency.
These results suggest that the removal of TN in reclaimed-water reservoirs may be enhanced by FWIs. The FWIs may be capable of enhancing pond TN removal by limiting algae activity and enhancing denitrification. Similar studies are recommended for FWI pond applications, to develop general sizing criteria and removal rate constants. Ponds differ in depth, size and loads and must be assessed uniquely to understand the capacity for FWIs to enhance TN removal.
The authors gratefully acknowledge the assistance of the staff of the Wesley Center WWRF, Pasco County Utilities Engineering, and Pasco County Environmental Laboratory. Project funding was provided by the Southwest Florida Water Management District to Pasco County through Cooperative Funding Agreement No. 12C00000079.