Reciprocating subsurface-flow constructed wetlands: a microcosm treatability study

Chronically low dissolved oxygen (D) concentrations in subsurface-flow constructed wetlands (SSFW) adversely impact aerobic treatment processes including organic matter degradation and nitrification (Vymazal & Kropfelova 2008; Garcia et al. 2010). Oxygen dynamics in gravel-based SSFW are affected by saturated media profiles, low D solubility (8–16 mg/L), low convective mixing, negligible diffusion of oxygen at the air-water interface, and high community respiration rates. Net transfer of D via diffusion at the air-water interface in SSFW is marginal, ranging from 0.2 to 3.2 g/m²/day (Behrends et al. 1996; Tyroller et al. 2010; Nivala et al. 2013).

To overcome this problem, various ideas and concepts have been evaluated to enhance oxygen transfer in SSFW. García et al. (2005) obtained significant improvements in wastewater treatment by reducing the depth of the wetland substrate from 50 to 25 cm. However, this method requires doubling the land area requirements for a given hydraulic retention time, which adds to land and capital costs. Data suggests that aquatic macrophytes may or may not contribute significant D to the rhizosphere in SSFW, with results ranging from 0.005 to 12 g/m²/day (Brix & Schierup 1990; Steinberg & Coonrod 1994; Armstrong et al. 2000). However, based on empirical evidence gathered from field trials, other researchers have concluded that the amount of oxygen made available to the rhizosphere by aquatic macrophytes is at least an order of magnitude less than the amount needed for aerobic treatment of domestic and municipal wastewater (Nivala et al. 2013).

While oxygen diffusion at the air-water interface is rate limited as discussed above, oxygen diffusion in the atmosphere is up to 10,000 times faster. So Brix & Schierup (1990) and Watson & Danzig (1993) recommended intermittent flooding and draining of SSFW cells to improve atmospheric oxygen supply to plant roots and substrate biofilms. Green et al. (1997), discussed and proved the utility of a passive air-pump to enhance nitrification in gravel-based wetlands. Others have used the fill and drain process to improve oxygenation of plant roots and to significantly improve nitrification in aquaponic food production systems (Rakocy 2012).

Starting in 1993, TVA scientists began research and development of a novel reciprocating wetland, whereby coupled SSFW treatment cells were recurrently filled and drained (six to 12 times/day), such that dewatered substrate biofilms and plant roots were repeatedly exposed to atmospheric oxygen (Behrends 1999). During the dewatered drain phase, the roots and biofilms were exposed to a 21% oxygen atmosphere (300 mg/L O₂), for up to 2 hours per cycle. During the fill phase, these same roots and biofilms were exposed to an oxygen-deprived environment (anoxic), which promoted enhanced removal of nitrate via microbial denitrification. This fill and drain reciprocating process enhanced growth of select aquatic macrophytes, COD removal, and nitrification and denitrification (Behrends et al. 1996). Variations and modifications of the TVA reciprocating system have been referred to as fill and drain wetlands and tidal-flow wetlands (Austin & Nivala 2009; Wu et al. 2011).

To reiterate, the role of reciprocation is not to aerate the wastewater per se, but instead, it is a method for supplying an ever abundant and low-cost supply of atmospheric oxygen to the microbial and plant biota during the reciprocation drain phase. As will be verified in this batch-loaded microcosm study, oxygen dynamics, pH, and redox potential of paired SSFW systems can be controlled and optimized via reciprocation to enhance both aerobic and anoxic wastewater treatment processes. Discussions will be developed around these microcosm-based studies as well as later findings from pilot and demonstration scale reciprocating systems.

Two microcosm-based treatability studies were conducted concurrently to evaluate the reciprocation process and to quantify treatment dynamics and interactions over a 196-hour trial. Variables monitored included DO, temperature (°C), pH, redox potential (mV), electrical conductivity (EC), chemical oxygen demand (COD), total ammonia nitrogen (TAN), nitrate nitrogen (NO3-N), orthophosphorus (PO₄-P), and evapotranspiration (ET).

Table 1 summarizes information related to the composition of MRS, while Table 2 provides treatment designations, nutrient treatments, reciprocation regimes, and planting arrangements (plants vs. no plants).

Microcosm units consisted of 30 opaque glass aquaria (70 litres volume, 60 cm × 38 cm × 30 cm). Each microcosm had a surface area equivalent to 0.228 m² and was divided into two equal volumes with a permanently fixed water tight vertical glass partition. Each of the two compartments within a microcosm were plumbed to accommodate recurrent draining and filling (reciprocation) via independent air-lift devices (Figures 1 and 2). Both sides of each microcosm were backfilled with 2.5 cm of limestone rock (2.5–4.0 cm), which was overlaid with 25 cm of river pea-gravel (0.5 to 1.0 cm). Porosity of the rock substate averaged 40 percent. Prior to adding the gravel substrates, 38 cm lengths of slotted PVC well-screen ports (5 cm diameter) were placed vertically on either side of the center partition (Figure 2). These ports enabled taking water samples, checking water quality with multi-parameter sondes, and checking redox potential with in-situ platinum-tipped probes.

Acclimation of planted microcosms consisted of batch loading 13 L of a synthetic wastewater solution into one side of each microcosm. The solution had 200 mg/L MRS as a source of organic carbon, and macro- and micro- nutrients (Table 1). The synthetic wastewater was amended with 50 ml of a ‘dirty’ water slurry collected from an existing SSFW. The slurry was assumed to have a diverse population of native microbial population that would supply seeding of the wetland microcosms. During acclimation (two weeks prior to microcosm trials), all microcosms remained static, with no reciprocation. The two-week acclimation period was intended for biotic establishment prior to the treatability studies. Designated treatments were planted with canary grass (Phalaris arundinacea) at 50 g wet weight, equivalent to 219 g/m². Other treatments were not planted and served as non-planted controls (Table 2).

Experimental designs consisted of two factorial experiments that were run concurrently. The first batch-loaded study evaluated three MRS concentrations (50,100, and 200 mg/L), equivalent to COD concentrations of 111, 238, 516 mg/L, two reciprocation treatments (six and 12 cycles/day), and a non-reciprocating control. All treatments in study one were planted with canary grass (P. arundinacea), a plant that has an anoxic rootzone which can enhance denitrification (Steinberg & Coonrod (1994); Zhu & Sikora (1995). The second concurrent study evaluated batch loading of a single MRS concentration (400 mg/L), equivalent to 988 mg/L COD, two reciprocation treatments (six and 12 cycles/day), a non-reciprocating control, and two planting regimes (with and without canary grass, P. arundinacea). In both studies, treatment combinations were replicated twice. See Table 2 footnotes regarding treatment definitions and designations for both treatability studies.

Four 200 L batches of synthetic wastewater were prepared using dechlorinated tap water, powdered MRS, ammonium chloride and potassium phosphate (Table 2). Reciprocation modes included a no-reciprocation control, a one-hour and a two-hour regime.

In the one-hour reciprocation regime water was air-lifted from Cell A to Cell B for twenty minutes followed by a 40-minute rest period. Twenty minutes of air lift was sufficient to move a 25 cm column of water in Cell A to Cell B and vice versa. Subsequently, water was air-lifted from Cell B to Cell A for 20 minutes followed by a 40-minute rest period. The two-hour reciprocation regime consisted of air-lifting wastewater from Cell A to Cell B for 20 minutes followed by a 100-minute rest period. Subsequently, water was air-lifted from Cell B to Cell A for 20 minutes followed by a 100-minute rest period. Thus, during a 24-hour period, there were 12 and 6 cycling events for the one- and two-hour reciprocation regimes, respectively.

After the two-week acclimation period, the microcosm units were drained and refilled with 13 L of simulated wastewater according to treatment designation (Table 2). Wastewater samples were collected at this time from each of the four batches of synthetic wastewater and analyzed to determine composition of synthetic wastewater solutions at time = 0. Subsequently, water samples were collected from microcosm units with syringes from the vertical PVC wells at elapsed times of 1, 3, 6, 12, 24, 48, 96, and 192 hours. Duplicate samples were preserved (or not) with 2.0 mL of concentrated sulfuric acid and stored at 4 C until analyzed (usually within 48 hours). Unpreserved samples were used for COD analyses using Hach kits (Hach Chemical Co, Ames, Iowa). TAN, nitrate, and ortho-P were analyzed using either methods of inductively coupled plasma (ICP), or an automated Latchet instrument.

Water quality parameters including temperature (°C), DO, EC, and pH were also checked at 1, 3, 6, 12, 244, 896 and 192 hours with a multi-parameter hand-held sonde (YSI 600 YSI Inc., Yellow Springs, OH). Redox potential was checked with a platinum tipped copper wire that were mounted inside the PVC monitoring wells, with the platinum tip 7.5 cm from the bottom. A millivolt (mV) meter was connected to the redox probe with a calomel reference electrode. Redox values were corrected by adding 244 mV to adjust values to a hydrogen reference (Zhu & Sikora 1995).

ET was quantified by recording the amount of dechlorinated tap water that was added back to each microcosm during the study to replace water lost to either evaporation (no-plant controls), or ET in planted treatments. These data were summed by treatment to illustrate water use as a function of MRS concentration, reciprocation treatment (0 vs. six vs. 12 cycles), and planting regime (plants vs. no plants).

Reaction rate coefficients (k) were calculated for removal of COD and TAN using the linearized first-order kinetic model: ln (C₀/C) = kt, where C and C₀ represent concentrations at time t and the starting concentration respectively (Crites & Tchobanoglous 1998). Means and standard deviations were calculated for tabular and graphic illustrations.

Average temperatures (°C) among treatments over the eight-day trial ranged from 28.2 to 31.3. (Figure 3). While there were marginal differences between treatments, there did not appear to be any significant effects due to reciprocation, organic loading rates, or planted vs. not planted.

Average pH values among treatments ranged from 6.3 to 7.3 and were influenced by reciprocation and to a much lesser extent planting regime (Figure 4). With reciprocation, average pH values were at least one pH unit greater than the non-reciprocating controls, while treatments with and without plants differed by 0.1 pH unit. Among the two reciprocating treatments (six vs. twelve cycles per day), there were only marginal differences. Among MRS loading rates (50, 100, 200, and 400 mg/L), there were no discernable differences in average pH values: 7.0 vs. 7.0 vs. 6.9 vs. 6.9, respectively.

Reciprocation was effective in striping dissolved CO₂ from the wastewater due to aeration induced by air-lifts and the fill and drain process. In this study, reducing CO2 concentrations via reciprocation affected the carbonate-bicarbonate alkalinity system leading to increases in pH. Average values for reciprocating treatments were all greater than 7.0, while non reciprocating treatments were near or below 6.5. Values of pH above 6.5 are often accompanied by precipitation of calcium salts including apatite and calcium carbonate (Boyd 1982).

EC, a measure of total dissolved salts, is measured in umhos/cm. In natural waters, EC ranges from 20 to 1,500 umhos/cm (Boyd 1982). Average EC values for treatments (Figure 5), ranged from 1087 to 1392, and revealed that neither MRS loading rates nor planting regimes influenced EC values. However, treatments with six or 12 reciprocation cycles per day had average EC values 14.4 and 17.4% less than overall treatment averages, respectively. Factors that may have influenced these reductions in EC include 1) nutrient removal (TAN and nitrate-N), via microbial-based oxidation /reduction reactions and plant uptake 2) pH induced precipitation of salts such as calcium carbonate and apatite. Figure 6 illustrates the strong relationship between EC and pH for all 15 microcosm treatments with and without reciprocation.

Notice that when pH was high, conductivity was low and vice versa. This relationship was induced primarily by reciprocation and its effective stripping of dissolved CO₂.

Within hours of starting the study, D concentrations in all treatments were reduced by 2.5–7.0 mg/L, indicating a robust microbial population and high community respiration rates (Figure 7). Irrespective of treatment, D concentrations reached their lowest point within the first 12 hours. After 12 hours, DO-related treatment dynamics rapidly diverged with oxygen concentrations increasing for all reciprocating treatments, while D concentrations for non-reciprocating treatments remained at <0.5 mg/L for the duration of the eight-day study.

Treatment D averages clearly revealed the effect of reciprocation vs. no reciprocation, and the lesser impact of MRS loading rates (Figure 8). While the reciprocating process resulted in higher average D concentrations, the differences among the reciprocating treatments (12 vs. six cycles/day) were marginal, amounting to less than 0.5 mg/L. With respect to MRS loading rates, average D concentrations were stepwise as expected, with 50 MRS > 100 > 200 > 400 (see trendline). The impact of plants vs. no plants on D concentration was marginal, amounting to less than 0.5 mg/L (see MRS 400 treatments).

Figure 9 illustrates average D concentrations at time zero vs. concentrations after 96 and 192 hours of treatment. This figure reveals marginal D recovery for non-reciprocating treatments even after 192 hours (8 days). Furthermore, among reciprocating treatments, most of the increases in oxygen concentration occurred with the first 96 hours. None of the treatments recovered to oxygen saturation during the study indicating significant residual community respiration.

ORP values for treatments without reciprocation decreased to less than −200 to −300 mV within 24 hours and remained below −100 mv for the duration of the eight-day study (Figures 10 and 11). These low redox values are near optimum for production of methane, a potent greenhouse gas (Wang et al. 1993) and may be problematic for passive SSFW (Brix et al. 2001). In contrast, treatments with reciprocation never dropped below +80, and beyond 48 hours all reciprocating treatments maintained average redox values within the range +200 to +300 mV (Figure 10). Among the fifteen treatments evaluated, there were significant correlations between D (mg/L) and redox potential (mV), and correlations between redox potential and pH (Figures 12 and 13, respectively).

Batch-loaded COD concentrations for MRS 50, 100, 200 and 400 treatments were equivalent to 2.8, 5.7, 11.4, and 22.8 (g/m²) respectively. Changes in COD were rapid and concentration dependent (Figure 14), with 46–95% reductions within the first 24 hours. COD removal among treatments with reciprocation (six vs. 12 cycles/day), averaged only 2 to 3% different. Thus, within the COD concentrations evaluated, there was a highly significant advantage to reciprocation, but no clear advantage to increasing the reciprocation regime from six to 12 cycles/day. This is an important finding, as pumping costs associated with reciprocation can be reduced by half by adopting the six cycles per day treatments.

COD removal as a function of planting regime (Y vs. N) was only evaluated at the 400 mg/L COD concentration. Results showed that COD removal was marginally more rapid for treatments without plants, which is consistent with the findings of Zhu & Sikora (1995), who found that canary grass roots, P. arundinacea, consistently released high concentrations of organic carbon. As will be seen later, this production of organic carbon can be important for enhancing removal of nitrate via microbial denitrification.

Figure 15 summarizes residual COD concentrations (mg/L) after 192 hours of treatment and final treatment averages expressed as percent removal. Residual COD concentrations ranged from 14 to 56 mg/L, with higher values associated with non-reciprocating treatments. Removal of COD expressed as a percent, ranged from 55 to 98%, with the higher removal rates associated with treatments receiving higher COD concentrations. This is consistent with first-order removal rates and has implications for placement of decentralized systems. Placement of decentralized reciprocating systems closer to the concentrated wastewater source provides higher removal rates and thus more economical treatment. The data also reveals that most of the COD removal in reciprocating systems occurred within the first 48 hours. This finding implies that a two-day retention time for reciprocating systems may be most economical and near optimum for COD removal within the COD concentrations tested (111, 238, 516, 988 mg/L).

Values of k (first-order removal rate coefficients) ranged from 0.10 to 0.55, and revealed progressively higher K values with increasing COD concentrations (Figure 16). Values of k were also higher for treatments with reciprocation. Note that in the 400 MRS treatments (988 mg/L C OD), rates of removal were marginally higher for treatments without P. arundinacea. For treatments without reciprocation and low D.O. concentrations, it is surmised that significant metabolism occurred under anoxic/anaerobic conditions with CO₂ and methane as end-products. Methanogenesis is a more efficient process for treating recalcitrant organic matter and produces less sludge than aerobic metabolism. However, anaerobic metabolism produces methane, a potent greenhouse gas and can require longer retention times to completely degrade recalcitrant organic compounds (Zitomer & Speece 1993).

TAN batch loading to each of the 15 treatments was equivalent to 5.7 g/m² with an initial starting concentration of 100 mg/L. Removal of TAN was rapid for all treatments within the first six hours, with reciprocating treatments exhibiting significantly faster removal as compared to non-reciprocating treatments (Figure 17). Removal rates, expressed as percentages, are illustrated for each treatment at 48 and 96 hours (Figure 19). TAN removal for reciprocating treatments ranged from 88–99% at 48 hours and greater than 99% at 96 hours. In contrast, non-reciprocating treatments removed from 52–64% at 48 hours and from 57–82% at 96 hours. The impact of canary grass, P. arundinacea, in non-reciprocating treatments receiving the highest organic loading (MRS 400), revealed 10 to 35% better removal of TAN at 48 and 96 hours respectively, as compared to the non-planted controls (Figure 18).

Values of k (first-order removal rate coefficients, ranged from 0.12 to 0.96 (Figure 19). Removal rate coefficients for reciprocating treatments were 2.6 to 8.0 times greater than non-reciprocating treatments. Organic loading rates did not appear to significantly affect TAN k-values among reciprocating treatments. However, in non-reciprocating treatments at the highest loading rate (MRS 400), TAN removal was greater for planted vs. not planted; k = 0.26 vs 0.12, respectively.

It is proposed that greater TAN removal rates among reciprocating treatments was due primarily to the repeated exposure of the microbial biofilms to atmospheric oxygen and the ensuing positive impact of oxygen availability for nitrifying bacteria. Furthermore, reciprocation supplied sufficient oxygen to meet both the oxygen requirements for removal of COD and removal of TAN. Uptake of ammonium by plants was also clear (Figure 18), as revealed by the direct comparison of TAN removal rates in treatments in the high organic loading treatments (400 MRS). TAN removal rates for treatments with P. arundinacea were 18 to 35% higher than for treatments without plants.

There was no nitrate added to the initial synthetic wastewater solutions, and thus any nitrate detected in the treatment systems was due to in situ nitrification. Denitrification is defined as the ‘process in which nitrate is converted to dinitrogen gas (N₂), via intermediates nitrite, nitric oxide, and nitrous oxide’ (Vymazal & Kropfelova 2008). Denitrification is a microbial driven process and occurs in anoxic and anaerobic environments and requires a bioavailable dissolved organic carbon source (Garcia et al. 2010).

In the present study, total nitrogen dynamics was influenced by nitrification and denitrification, reciprocation and the presence or absence of canary grass, P. arundinacea, a plant known for its ability to reduce rhizosphere redox potential (Steinberg & Coonrod 1994; Zhu & Sikora 1995). Canary grass is a valuable wetland plant as it can uptake ammonium and nitrate as nutrient sources and supply a root-exuded organic carbon source, which supplements the carbon needed for denitrification (Zhu & Sikora 1995). Figure 20 illustrates that in 13 of the 15 treatments evaluated (all with canary grass), nitrate was only detected at low concentrations (<4.0 mg/L) during the first 48 hours of the study and were less than one mg/L after 96 hours of treatment. This shows that while nitrification was ongoing in these treatments at a high rate (Figure 17), denitrification, microbial immobilization, and plant uptake were removing nitrate as fast as it was being produced, and thus there was no significant buildup of nitrate. This is also consistent with the findings of Sikora et al. (1995a). Conversely, treatments with reciprocation (six and 12 cycles per day), high organic inputs (400 MRS), and no canary grass showed nitrate concentrations of 17 and 27 mg/L after 96 hours of operation (Figure 20). As noted earlier, microbial denitrification requires anoxic conditions and a labile organic carbon source. While reciprocation was effective at increasing oxygen supply and rapidly reducing COD, there was no canary grass to supplement the organic carbon, and therefore denitrification was carbon limited. If denitrification is needed in reciprocating systems, there will be a need for carbon supplementation or use of plants that supply a labile carbon source (Zhu & Sikora, 1995).

Phosphorus batch loading to each of the 15 treatments was equivalent to 2.85 g/m² with an initial starting concentration of 50 mg/L. Removal of P was rapid during the first 12 hours and then tended to stabilize. The rapid removal was influenced by sorption of P to the rock substrate as a function of exchange sites (Sikora et al. 1995a; Zhu et al. 1997). After eight days, average removal rates ranged from 92 to 100% (Figure 21). Figure 22 illustrates treatment averages and trendlines at 24 and 192 hours. Data shows that after 24 hours, treatments with reciprocation removed significantly more P than treatments without reciprocation. Also, at the highest COD concentration (400 MRS), P removal was greater with plants than without plants.

Shi et al. (2017), evaluated P-removal in intermittently aerated wetlands and found superior plant growth and P removal as compared to non-aerated controls. Their research also evaluated several different substrates and found fly-ash brick to be significantly better than gravel or red clay brick. Ilyas & Masih (2018), found tidal flow wetlands (like reciprocating wetlands) to be superior to other wetland types for P removal and reported removal rates of 88% and 2.6 g/m²/day. They attributed the superior performance to ‘a positive effect of tidal flow on rejuvenating the wetland with fresh air, thus enhancing dissolved oxygen (DO) in the system, and augmenting phosphorus precipitation and adsorption to the substrate.’

While research results reported above are encouraging, sustained phosphorus removal in substrate based SSFW will be difficult to achieve, because removal dynamics in gravel-based systems are controlled primarily by adsorption of P to limited sorption sites on the substrates surface. Once these sites are saturated, P removal will be significantly diminished. While P uptake by plants and microbes can be significant, they are seasonal and not sustainable in the long term unless plant and microbial biomass are harvested and removed from the wetland site (Vymazal & Kropfelova 2008).

ET is the joint loss of water from the landscape due to evaporation from surface substrates and transpiration, the loss of water through vegetation. ET is influenced by plant biomass, prevailing wind speed, and seasonal variations in temperature and humidity (Vymazal & Kropfelova 2008).

Figure 23 illustrates ET rates (mm/m²/day) for 15 treatments during an eight-day batch-loaded treatability study. The trendline indicates that the influence of MRS loading rate was insignificant, while the influences of plants and reciprocation were additive and significant. Treatments with plants but no reciprocation averaged 2.7 mm/m²/day, while treatments with plants and reciprocation (six and 12 cycles/day), averaged 8.5 and 10.8 mm/m²/day, respectively.

Data for the MRS 400 series illustrates the influence of two interacting treatment variables: with and without reciprocation, and with and without plants. This data reveals a strong positive interaction for planting and reciprocation regimes. Average ET rates (mm/day) were low for treatments without plants or reciprocation (1.2) and moderately low for treatments with no plants but with reciprocation (4.0). However, with both plants and reciprocation, average ET was equivalent to 10.9 mm/m²/day).

The ET values reported in this study are within the range of values reported in the literature (Vymazal & Kropfelova 2008; Milani et al. 2019). While the data in the present studies were derived from a short-term treatability study with low plant biomass, the among-treatment values reveal the positive interaction of reciprocation and planting regimes on ET and water budgets.

The research team designed and fabricated the 30-tank microcosm system on-site using off-the-shelf items. During this eight-day study and four later studies, the microcosm system ran admirably with few problems. Use of solenoid valves, digital timers, regenerative blowers, and one-way aquaria valves allowed air-lift operations to distribute discrete and uniform volumes of water between treatment cells on a timed basis. Therefore, between tank experimental error was well controlled as corroborated by low coefficients of variation (CVs), for measured variables. For example, the CV for TAN over 120 determinations was 12.4%.

This microcosm system was cost-effective with respect to materials and operating costs and enabled complex experimental designs in which main effects and interactions could be quantified. Future studies should be designed to further evaluate (1) experimental substrates and substrate combinations, (2) aquatic and terrestrial plants for applications of phytoremediation with- and without reciprocation (3) optimized cycle times for high strength anaerobic lagoon wastewater, and (4), engineered underdrain designs for long term storage and treatment of bio-solids. Because reciprocation aids movement of sluffed biosolids to engineered underdrains, this may enhance anaerobic degradation of solids and mitigate clogging issues and extend the life of the treatment system.

The recurrent fill and drain process provided mixing of the wastewater and passive aeration such that nutrients and oxygen were distributed uniformly to the biofilms and to the plant roots in the rhizosphere. It is envisioned that reciprocation and attendant mixing will diminish, if not eliminate, short circuiting which is a problem in conventional surface-flow and subsurface-flow constructed wetlands (Dierberg et al. 2005).

With reciprocation, atmospheric oxygen was available during the drain phase. Nivala et al. (2013) monitored oxygen consumption rates in fifteen different pilot-scale wetland treatment systems and reported maximum oxygen consumption of 89.3 g/m³/day for a 2-cell reciprocating system treating municipal wastewater. This was the highest rate among the fifteen different treatment systems examined and testifies to the utility of the reciprocating process to provide a near-inexhaustible supply of atmospheric oxygen for aerobic wastewater treatment processes. Conversely, because of the anoxic environment that develops during the fill phase of reciprocation, denitrification, and TN removal were also significantly greater than the other 14 pilot-scale systems being evaluated. Furthermore, in the present study it was revealed that reciprocation allowed supersaturated metabolic gases, such as CO₂, to escape such that pH values were increased by one full unit as compared to non-reciprocating controls. This may have utility for enhancing calcite precipitation and removal of phosphorus.

The exceptionally high removal rates for COD, TAN, and P during the first 3 to 12 hours was due to a combination of factors including high initial concentrations, microbial immobilization, plant uptake of nutrients, and the adsorption of TAN COD, and P to rock substrates due to cation and metal binding exchange sites. Sikora et al. (1995a, 1995b), provide relevant and detailed interpretations and discussions of cation exchange dynamics for gravel-based constructed wetland systems and notes that with respect to ammonium, adsorption can be rapidly reversible due to the dynamic equilibrium with nitrification. It should also be noted that the rapid removal of DO during the first 12 hours is emblematic of high microbial community respiration rates associated with COD removal and nitrification (Nivala et al. 2013).

It is surmised that having an aerobic rhizosphere in reciprocating systems will allow culture of select terrestrial plant species for purposes of phytoremediation, including flowers, woody ornamentals, tree saplings, and fodder crops. It should be noted that these plants have economic and aesthetic values beyond their wastewater treatment functions.

While these short-term batch-loaded microcosm treatability studies provided significant treatment of synthetic wastewater, later collaborative and third-party pilot-scale and commercial-scale demonstrations have corroborated the sustainability, scalability, and energy efficiency (Austin & Nivala 2009), of deep-cell reciprocating systems for treating a diversity of domestic, industrial, and agricultural wastewaters. Examples include airport deicers (Arendt et al. 2002), acid mine drainage (Sikora et al. 1995b), swine lagoon wastewater (Behrends et al. 2004), dairy lagoon wastewater (Choperena 2010; Henneman 2011), pretreated high-density aquaculture wastewater (Behrends et al. 1999, 2007) and domestic wastewater (Nivala et al. 2019). Reciprocating wetlands have also been evaluated to control malodors (swine) and greenhouse gas emissions (dairy) and found to have significant mitigation capabilities (Schiffman et al. 2008; Henneman 2011).

Conceptually, reciprocating wetlands may have even greater economic value for wastewater treatment in the tropics and at high altitudes, where oxygen solubility is reduced due to elevation and wastewater temperatures respectively (Baquero-Rodrigues et al. 2022). It should be noted that reciprocation requires electricity for rapidly moving water (fill and drain), among paired treatment cells. However electrical costs can be significantly reduced by using auto-siphons and axial flow pumps that are designed for low head-high volume applications Yoo & Boyd (1994). Clogging of wetland gravel substrates is also a potential problem for all gravel-based systems, including reciprocating wetlands (Behrends et al. 2006). Regarding control and management of clogging, Nivala et al. (2012) concluded ‘the best approach moving forward will be to minimize the negative effects of clogging through improved pretreatment, better systems design, and new operation and management practices.

This short term eight-day microcosm study quantified the impacts of reciprocation with and without plants (canary grass, P. arundinacea) on the treatment dynamics of COD, TAN, nitrate, phosphorus, and ET. Reciprocation induced passive aeration and the development of recurrent aerobic and anoxic environments six to 12 times per day. During the drain phase, the microbial biofilms were exposed to high atmospheric oxygen concentrations (21%), which promoted rapid microbial oxidation of ammonium and COD and removal of phosphorus. During the fill phase, the biofilms were covered by oxic-anoxic water, which enhanced denitrification and the removal of nitrate. The combination of reciprocation with P. arundinacea significantly improved biological denitrification, ET, and other wastewater treatment dynamics.

Among treatments with six and 12 reciprocation cycles per day there were significant improvements with respect to removal of COD, ammonium, nitrate, and phosphorus. Treatments with reciprocation and P. arundinacea removed 81–98% of COD, greater than 95% of ammonium and nitrate, and greater than 98% of phosphorus. Treatments with both reciprocation and plants (P. arundinacea), revealed a strong positive interaction with respect to ET, with treatment averages ranging from 8.3 to 12.7 mm/day.

This work was funded and supported by the Tennessee Valley Authority's Environmental Research Center (TVA), and by grants from the U.S. Environmental Protection Agency (EPA).

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

The authors declare there is no conflict.