Study location and Pilot plant design

During March and April of 2008, a small pilot-scale integrated CW wastewater treatment facility was designed, built and put into operation at the Biological Reserve CAOBA, which is located 15 km from the city of Santa Marta (Colombia), in the Manzanares river basin, a warm region, which is characterized by a complex mosaic of habitats and very high levels of biodiversity. The design of this integrated system was based in part on in-depth studies of shallow horizontal subsurface-flow wetlands, HSSF (^{Caselles & García, 2006}; ²⁰⁰⁷). The systems provided primary and secondary wastewater treatment for 6 people. The system operated in serial flow (Figure 1), and consists of a 2.6 m³ sedimentation tank for primary clarification, followed by a series of 5 subsurface-flow wetland cells (1m x 1m x 0.5m), a single surface-flow wetland (2m x 0.9m x 0.5m), and a lagoon reservoir (1 x 0.9 x 0.5) polishing of the wastewater and water reuse options (irrigation). The lagoon provided an area for culture of native floating plants and nutrient polishing. The multicell design enabled frequent redistribution of flow to mitigate short-circuiting and to enable periodic removal of sludge from cells 1 and 2.

Figure 1 Schematic representation of the pilot plant. Note both wetlands types and a terminal holding reservoir for water reuse options. The sampling points where put in the corner of each cell, assuming the same possibility to examination the wastewater quality 

Multiple cell designs with appropriate plumbing also allow for bypassing influent cells to facilitate periodic removal of sludge. The clarifier and cells were constructed of hollow cement blocks and sealed with commercial epoxy paint. Each of the subsurface-flow cells were backfilled to a depth of 0.5 m with screened 10 mm river gravel harvested from the nearby river. The wetland system was backfilled to a depth of 0.1 m with river’s sand to facilitate rooting of aquatic plants. Plastic standpipe distribution headers at the entrance and exit of each cell facilitated wastewater distribution and maintenance of water level at approximately 0.4 m depth. In addition, each of cells had slotted vertical plastic pipes installed at strategic locations to allow for in situ measurements and collecting of discrete whole column water samples

Influent wastewater flow rates (L/minute), were estimated using the bucket and stop watch method at one to two hour intervals during daylight hours. It is envisioned that integrated systems such as portrayed in Figure 1 can be planted with adapted vegetable crops, such as tomatoes, and that treated wastewater can be stored in the reservoir and reused as toilet flush water or nutrient rich water for irrigating adjacent crop lands, fish ponds or gardens. These objectives are similar to those as described for high diversity wastewater gardens and irrigated ecoscopes (^{Nelson et al., 2008}).

Planting tomatoes (Lycopersicum sculentum)

Seeds of tomatoes were germinated in culture flats and grown for 30 days prior to being transplanted into the treatment system. At the time of transplanting, the plants ranged from 10-15 cm high and were planted on 20 cm. It should be noted that the first treatment cell in the series was not planted due to severe organic clogging. After 94 days of culture, the tomato biomass, including roots, shoots and fruit were harvested in each cell and processed. Harvested tomato fruits were stored in a cardboard box and transferred to laboratory. Two samples of fruit from each treatment cell were rinsed in dechlorinated tap water and the surface and flesh analyzed for coliform bacteria. The remaining tomatoes were weighed in Ohaus TR6RS balance for total biomass calculations.

Pilot plant performance monitoring

Water quality monitoring was carried out between October of 2009 and May of 2010. During the period December/09 to March/10, tomatoes were cultivated in the treatment cells. During the study, four (4) water samples were collected from each of the 5 treatment cells and four of influent and effluent. The average values were reported. All samples were preserved in an insulated cooler and temperatures were maintained at near 5ºC. Subsequently, and within 24 hours, the samples were analyzed for organic matter (COD, BOD₅), ammonium nitrogen (NH⁴⁺-N), nitrate nitrogen (NO₃⁻-N), nitrite nitrogen (NO₂⁻-N) and ortho-phosphate (PO₄⁻³-P) using methods as described in (^{APHA-AWWA-WEF, 2005}). Water temperature, dissolved oxygen (DO), pH and conductivity measurements were obtained by monitoring the water within the vertical perforated PVC tubes (Figure 1). Measurements were taken at the midpoint of the water depth and water temperature, conductivity, pH and DO measurements were recorded with WTW 315i Set probes. Paired comparisons of inlet and outlet concentrations for each treatment cell were evaluated for COD, BOD₅, (PO₄⁻³-P), (NO₃⁻-N), (NO₂⁻-N) and (PO₄⁻³-P).

Microbiological examination for total coliform (TC) and fecal coliform (FC),

Samples of wastewater were collected in labeled plastic bottles, placed in a cooler and transported directly to the laboratory where they were stored under refrigeration. The samples were analyzed within 24 h of collection for the presence of coliform bacteria group (total and faecal coliform) using the multiple-tube fermentation technique (^{APHA-AWWA-WEF, 2005}).

Water Quality Dynamics

Tables 1 and 2 and Figure 2 provide summary information related to design criteria, water quality and treatment dynamics. Anoxic / anaerobic conditions prevailed in all treatment cells during the study, as evidenced by high NH₄⁺-N concentrations (>30 mg/L) and low DO conditions (< 0.2 mg/L). High hydraulic and organic loading rates encouraged high respiration rates leading to anoxic conditions, poor nitrification and no a NH₄⁺-N removal (Table 2). Chronically low DO concentrations and low nitrification rates have been commonly reported for horizontal sub surface flow systems (^{EPA, 2002}; ^{Zhang et al., 2015}). Where required, enhanced rates of nitrification and denitrification can be achieved in SSF wetlands by decreasing the depth of the system (^{Garcia et al., 2014}), although it decreases the HRT. However, these design changes often require additional land and capital expenses which may not be justified (^{EPA, 2002}). Poor removal of phosphorus during the second year was also evident (Figure 2), and was most likely related to rapid saturation of gravel micro-sites (^{Drizo et al., 1999}). The stable tropical temperature regime observed in this study (near 26 ºC), was viewed as being beneficial as it provided year-round opportunities for plant production (^{Trang et al., 2010}). pH values were relatively stable throughout the system, and always less than 7 which helped to minimize losses of ammonia to volatilization. Electrical conductivity values increased progressively throughout the system, probably due to high rates of evapotranspiration and release of nutrients and elements via organic mineralization (^{Coleman et al., 2001}) Scirpus validus L., and Typha latifolia L.. The pH and electrical conductivity values range within Colombian environmental standards for agricultural wastewater (Resolution 1207 of 2014).

Figure 2 Spatial treatment of wastewater contaminants in pilot plant in Santa Marta, Colombia 

Influent COD concentrations (326 mg/L ±176), BOD₅ (124 mg/L ± 33), and nutrients were within the ranges reported for septic tank effluents (^{Crites & Tchobanoglous, 1998}; ^{EPA, 2000}). COD and BOD₅ removal efficiency were modest and progressive throughout the system (Figure 2) and averaged 42 and 41% respectively. These levels of removal are low for the tropics as compared to other studies in which values ranged from 54 to 85% ( ^{Machado, et al., 2016}; ^{Caselles et al., 2017}) since they can remove a broad range of contaminants by a combination of physical, chemical and biological processes with a low cost. Therefore, CWs can be successfully applied for decentralized wastewater treatment in regions with low population density and/ or with large land availability as Brazil. The present work provides a review of thirty nine studies developed on CWs implemented in Brazil to remove wastewater contaminants. Brazil current sanitation data is also considered to evaluate the potential role of CWs as decentralized wastewater treatment. Performance of CWs was evaluated according to (i. It is inferred that the modest removal rates in this study were influenced by the relatively high hydraulic and organic loading rates. NH₄⁺-N and PO₄⁻³-P removal within the system (Figure 2) were very low and in fact ammonium concentrations increased within the system. This behavior is not uncommon in subsurface-flow wetlands (^{EPA, 2000}) and can be explained on the basis of poor nitrification due to anoxic conditions, while the addition of new ammonium in cells 4 and 5 resulted from mineralization of stored organic matter. During year I, ^{Caselles et al. (2008)} reported moderate levels of nitrification and NH₄⁺-N removal, but this was during system start up and before anoxic conditions and cell clogging became chronic. (NO₃⁻-N) and (NO₂⁻-N) concentrations remained less than 3 mg/L, and were probably controlled by a lack of nitrification and near optimum conditions for denitrification.

Nitrate and coliform effluent concentration values were low and in compliance with Colombian environmental standards for agricultural wastewater. (Resolution 1207 of 2017). Removal of coliform bacteria was progressive throughout the system (Figure 2), and averaged 46 and 81% for fecal and total coliforms, respectively. It is anticipated that the short hydraulic retention time (< 2 days), did not allow for higher coliform removal rates.

It is know that in small human settlements, the relationship between mean flows and peak flows is higher than in medium and big cities (^{Crites & Tchobanoglous, 1998}). These higher flow variations affected the organic load variations and pollutants concentrations, as COD and nitrogen species (Table 2).

Although the process of clogging was not monitored, it was especially evident in cell 1 where water surfaced and caused short-circuiting. Eventually, during the course of the study, cell 1 was cleaned out in Dec. 2009 and returned to service. Clogging is a chronic problem in SSFW ( ^{Knowles et al., 2011}; ^{Corbella et al., 2016}) leading to surface ponding, odors, mosquitoes and short circuiting. Mechanical and non-invasive techniques have been evaluated for remediating clogged cells (^{Pedescoll et al., 2012}), and additional studies will be required to validate cost-effective remediation strategies. Note however, that several studies have reported no loss of treatment efficacy, even with severe clogging (^{Drizo et al., 1999}; ^{Caselles & García, 2007}; ^{Aiello et al., 2016};). In retrospect, it is surmised that clogging phenomena could have been avoided by installing a shallow surface-flow wetland as the first treatment cell in the series. This would have allowed much of the particulate organic matter in the septic tank effluent to settle in the surface-flow wetland, thus removing some of the particulate load, and delaying clogging in the following sub-surface flow wetland cells (^{Behrends et al., 2007})

Tomatoes production

Although the influence of tomatoes plants on nutrient removal was not representative, this does not detract from the potential for producing valuable crops that are adapted to saturated conditions. It is possible that growth in waterlogged soils is achieved by exploiting the availability of nutrients. Traditionally tomato cultivation reaches is completed in three months (Ricardo et al., 2013). Tomatoes were cultivated in cells 2 to 5, and resulted in production of 2.2 kg of tomatoes fruit (Figure 3), it is 0.73 Kg/m². Annual production of tomatoes in a traditional crop in one tropical country like Colombia, is on average, 3.7 kg/m² (^{Agronet, 2014}). ^{Cirelli et al., (2012)}, reported a maximum tomato crop of 8.97 Kg/m² during a two-year trial of tertiary-treated municipal wastewater use in tomato crop irrigation. Our results are lower than those of ^{Cirelli et al., (2012)}. We produced tomatoes in just four months.

Figure 3 Total fresh weight of tomatoes fruit (g/cell) in SSF CW. Note that Cell 1 was not planted due to clogging and cell 3 productions were negligible for unknown reasons 

Although production in the experiment of this study is lower respect a traditional crop, the hydroponic conditions could affect the total production. Growth performance and crop yield of tomato plants were studied by ^{Rana et al. (2011)} with different wastewater concentrations. These authors reported that maximum crop production was recorded at 100% wastewater. The Cell 3 production was nil, probably due to not enough oxygen to nitrification of ammonia and (NH⁴⁺-N) high levels (>30 mg/L), which could be toxic for tomatoes (^{Martínez et al., 2013}). Additionally, tomatoes prefer nitrate as a source of nitrogen and may explain the zero growth of tomatoes in cell 3 where the plants were short and thin and produced no fruits (Figure 4).

Figure 4 Picture of the pilot plant with planted tomatoes. Note growth of tomatoes in most cells except cell 3 

Soil conditions in HSSF constructed wetlands can limit use of terrestrial plants, like tomatoes. Nevertheless, several studies have shown that some plants adopt strategies to grow in stress conditions (^{Hough et al., 2015}). One the major biological consequences of soil flooding is oxygen deficiency. According with ^{Lin et al. (2004)}, most tomato varieties are not able to tolerate flooding stress, but they have evolved a complex antioxidative defense system composed of both antioxidant enzymes and metabolites to survive oxidative stress. Then, it may be possible to utilize wastewater treatment in CW for the production of crop tomatoes.

Testing tomatoes for coliform bacteria indicated that there was negligible contamination of tomatoes fruits and suggests that with care, treated wastewater can be used for cultivating and drip-irrigating vegetable crops. ^{Al-Lahham et al. (2003)} reported similar results in tomatoes drip irrigated with treated wastewater, but with contamination on the tomato skin. These authors suggested that the treated wastewater can be used as an alternative for irrigation, and that the tomatoes can be eaten after cooking.

Finally, the experiments showed that treating wastewater with CW could become a profitable business in rural areas with little space limitation; instead, the occupied surface could generate income by tomatoes production.