Performance of constructed wetlands with ornamental plants in the treatment of domestic wastewater under the tropical climate of South Mexico

Mexico, like many other places around the world, faces overexploitation of water resources. In 2012, Mexico had the infrastructure to treat only 47.5% of its wastewater (CONAGUA 2013), although this figure accounts only for collected sewage treated in centralized treatment plants and ignores wastewater generated in rural areas without sewers and treatment plants. The lack of infrastructure in rural areas is partly due to the high cost of implementing and maintaining conventional treatment systems. Moreover, in general there is little interest for the inhabitants of rural areas to implement treatment technologies for a number of reasons, including the lack of extra benefits, whether they are economic or otherwise. Constructed wetlands (CWs) are a potential solution for domestic sewage treatment, in particular for rural areas since they have a relatively low implementation and maintenance cost and are easy to operate (Kivaisi 2001; Ascuntar et al. 2009; Mesquita et al. 2012).

Several studies have shown that plants in CW contribute to water treatment through direct and indirect mechanisms (Brix 1997; Greenway 2002; Shelef et al. 2013; Vymazal 2011, 2013). The main function of the vegetation is related to physical effects, since roots provide a surface for biofilm formation and buffer temperature changes (Haberl et al. 2003; Vymazal & Kröpfelová 2005). The appropriate choice of plant species is still an important subject of study as it is essential for optimal CW performance (Brisson & Chazarenc 2009; Gagnon et al. 2012). Among the most studied and widely used species in CW are plants within the genus Phragmites, Scirpus and Typha. However, more recently the use of ornamental species has also been explored in order to obtain additional benefits (Konnerup et al. 2011; Vymazal 2011). For example, Zurita et al. (2009) reported that the use of ornamental plants (Zantedeschia aethiopica, Strelitzia reginae, Anthurium andraenum and Agapanthus africanus) did not affect a CW's performance. Other species of the genus Heliconia, Iris and Canna have also been studied. For instance, Konnerup et al. (2009) reported levels of chemical oxygen demand (COD) removal ranging from 42 to 83% when using Canna and from 42 to 79% when using Heliconia, depending on the hydraulic load. Meanwhile, Yousefi & Mohseni-Bandpei (2010) found up to 43.4% removal of total nitrogen (TN) and up to 12% removal of total phosphorus (TP) in CW planted with Iris pseudacorus. However, most studies so far have evaluated the efficiency of ornamental plants using polyculture CW, making it impossible to identify the actual efficiency of each species in the system. For example, Sohair & Mohamed (2012) reported a 91.5% removal of COD and 92.8% reduction of biochemical oxygen demand (BOD) in CW planted with Canna, Cyperus and Phragmites.

In the Soconusco region of Chiapas in southern Mexico, the use of ornamental plants in CW could be an attractive alternative to conventional plant species since, in addition to waste water treatment, there is potential for obtaining other benefits such as aesthetic enhancement of the sanitation systems, production of biomass for compost and forage, and even the possibility of marketing the harvested flowers. Heliconia stricta, Heliconia psittacorum and Alpinia purpurata are ornamental plants of economic importance to the Soconusco region where these are commercially cultivated and exported to European and Asian countries.

This study was aimed at evaluating the efficiency of ornamental plants, namely H. stricta, H. psittacorum and A. purpurata, in horizontal subsurface constructed wetlands (HSCWs) to remove organic matter and nutrients from domestic wastewater pretreated in an anaerobic digester. This paper presents performance results of a full-scale system installed in a coffee farm and operated under the tropical climate conditions of southern Mexico.

The anaerobic digester-HSCW system was installed in the coffee farm ‘La Concepción’ located in the small town of Felipe Carrillo Puerto (00′40.22″15 ° N and 92 °11′42.96'″ W; 542 masl), in the municipality of Tapachula, Chiapas in south Mexico. It was designed and built to full scale and received domestic wastewater discharges from laborers housing. The research was carried over a period of 6 months (March–August 2013).

A three-chamber compartmentalized anaerobic digester was installed with a nominal volume of 37 m³ (Figures 1 and 2). The effluent from the digester was conducted through PVC tubes to the CW. The system included four HSCWs, each with a dimension of 2.5 × 11 m (Figure 3). The four wetlands had the same support structure (60 cm height) which consisted of a layer of gravel of 2 cm in diameter topped by a 20-cm layer of sand. In the inlet and outlet sides of the wetlands this materials were replaced by river rocks of about 20 cm in diameter.

Planting was done by seeding three plant tillers per square meter. Twenty-four points were plotted on each CW and two to three plants were planted at each point to ensure the survival of at least one plant. Only one species was planted in each CW (1) Heliconia stricta, (2) Heliconia psittacorum, (3) Alpinia purpurata and (4) Typha domingensis, which was used as a control (Figure 3).

The HRT of the anaerobic digester had an average value of 8 h during the sampling period. The HRT of the CWs at the beginning of the study was calculated at 24 h, but from day 112 of operation this was reduced to 12 h in two of the wetlands (H. stricta and H. psittacorum) because the other two wetlands (those planted with A. purpurata and T. domingensis) were taken out of operation to correct leaking problems that appeared in the basins.

The sampling points of the system were the inflow and outflow of the anaerobic digester and the outflow of the wetlands. Temperature, turbidity, pH, conductivity and dissolved oxygen (DO) were measured in each of the sampling points by means of a field meter (Horiba Water Quality Checker). In addition, 1,000 ml of sample were collected in amber bottles at each sampling point and transported to the laboratory, in a cooler at 4 °C, for chemical analysis. Weekly samplings were conducted over a period of 6 months.

Four parameters were analyzed in the laboratory to determine water quality TN; by the micro-kjeldahl method; TP, by the stannous chloride method; COD, by the closed reflux colorimetric method and BOD, by the Iodometric method. All analyzes were conducted according to the standard methods described by the American Public Health Association, APHA (1998). The determination of COD and BOD was done the same day that the samples were collected and brought in to the laboratory. In the case of TN and TP, 500 ml of sample were frozen at −20 °C for subsequent analysis.

Descriptive statistics were used to analyze water quality data in order to compare the efficiency of ornamental plants and T. domingensis during the 6 months of sampling. The average removal efficiency of each system was calculated with their corresponding standard deviation. Microsoft^® Excel^® for Mac 2011 was used to carry out statistical analyzes and to plot the data.

H. stricta and H. psittacorum showed normal growth and development in the CW stratum indicating tolerance to the stress of water saturation (Figure 4). After 2 months of being planted, their average height was 40 cm for H. psittacorum and 47 cm for H. stricta. At the end of the study, i.e. 6 months after being planted, H. psittacorum had an average height of 70 cm while the height of H. stricta was 79 cm. With the passing of time, new shoots were observed to emerge from the gravel support. However, only one inflorescence was observed in each species, which was to be expected since it has been reported that species of this genus present flowers after 8 months of being planted and only after a year begin to produce flowers of good quality (Rodríguez 2013). This suggests that these species reach maturity after about 1 year from planting. A. purpurata was not able to tolerate the harsh conditions of the CWs whereas T. domingensis, being an emergent aquatic species, was established without any apparent problem (Figure 5).

Water quality parameters showed important variation along the study period in part due to uncontrollable environmental factors such as rainfall and the hurricane season. Climate considerations have special relevance in the tropics where the magnitude, frequency and duration of rainfall can influence the performance of CWs (Harrington & McInnes 2009). Soconusco's coffee zone has a rainy season from April to August with cumulative rainfall (the combined amount of total rainfall for any given period) ranging from 2,300 to more than 2,500 mm by the end of August (CONAGUA 2009). An extraordinary cumulative rainfall of 3,004 mm was recorded by the end of this study (Figure 6).

Rainfall and other factors such as an increase in the number of on-farm laborers as the coffee harvest season started by the end of the study, resulted in important variations in the composition of the wastewater and in the hydraulic load received by the system which undoubtedly affected its performance.

The physical characteristics of the water improved as a result of its passing through the CW (Table 1). For example, the pH of the water in the effluent of the digester was slightly acidic, but this parameter increased to about neutral values in the effluent of all wetlands. Similarly, turbidity was considerably lower in the CW's effluents than in the raw wastewater and in the effluent of the anaerobic digester.

DO at the inlet of the system showed little variation during almost the entire study period (0.69–0.91 mg/l), except for the last sampling month (August) when it increased to 3.88 mg/l (Figure 7). DO in the anaerobic digester effluent was not measured in the first 3 months (March–May) due to sampling problems for this particular parameter. However, from the measurements made from June to August it was evident that DO was greatly reduced due to biological activity (i.e. oxygen uptake) in the digester.

An increase in DO was observed during the entire testing period as the effluent passed from the anaerobic digester through the CW. Similar levels of DO were recorded at the outlet of the four CW. For example, H. stricta showed DO concentrations ranging from 1.72 to 2.77 mg/l while for H. psittacorum DO levels ranged from 1.5 to 2.59 mg/l (Figure 7). These results seem to suggest that the presence of plants in the CW may have favored the increase of DO in the treated effluents due to oxygenation through the plants roots (Zurita et al. 2006). Heliconias have a stem with a vigorous rootstock and strong fibrous roots which can facilitate the transportation of oxygen downward (Jerez 2007). Some surplus of oxygen can be released from the roots into their immediate environment. However, the release of oxygen through the plant roots has been reported to be minimal and usually only generates a thin aerobic layer of just a few millimeters around the root (Sorrell & Armstrong 1994). Therefore, more detailed studies are required to identify the oxygen transportation capacity through the roots of these ornamental plants and to assess the contribution of other oxygenation mechanisms such as oxygen diffusion through substrate pores (Shelef et al. 2013) or the aeration of the wastewater in the distribution box previous to its introduction to the CW.

As expected, the highest level of COD removal was observed in the CW planted with T. domingensis after the 4th month of operation (Figure 8). The COD removal efficiency of the CW planted with H. psittacorum also increased over the first 4 months of operation but then remained at a similar level for the rest of study period. Despite the various uncontrollable factors observed throughout the sampling period, the average COD removal obtained was 52.7% with a HRT of 24 h and 63.8% with a HRT of 12 h; while for H. stricta the average removal efficiency was 55.4% with a HRT of 24 h and 45.1% with a HRT of 12 h (Table 2). These results compare well with previous studies where H. psittacorum was evaluated. For example, Konnerup et al. (2009) reported removal efficiencies of 42% at a HRT of 12 h and 58% at 24 h of TRH. Other researchers found 53.6% COD removal with a HRT of 24 h (Gutiérrez et al. 2010). It should be noted however that both of these studies were carried out in laboratory-scale wetlands. Differences are usually found when a vegetal species, previously assessed in the laboratory, is brought to a full-scale plant, resulting in lower removal percentages due to scale effects and lower control of environmental conditions (Brisson & Chazarenc 2009).

The COD removal efficiencies of both heliconia wetlands did not show marked differences as the HRT changed from 24 to 12 h, in contrast to previous reports showing that COD removal levels generally improve with an increase of the HRT (Belmont & Metcalfe 2003; Kadlec & Wallace 2009). As a matter of fact, the average COD removals observed in the CW with H. psittacorum seems to contradict this reported trend. This result is more likely due to differences in plant maturity as the study progressed rather than the effect of the HRT. Some authors have warned that the use of young plants can underestimate the real efficiency of systems that are meant to operate for several years (Vymazal & Kröpfelová 2005). Even though, the transformation and degradation of the water pollutants is brought about largely by microbial activity, the efficiency of micro-organisms is influenced by their interaction with the plants and the substrate (Huang et al. 2012). In the radicular zone, the micro-organisms can be exposed to compounds such as sugars, fatty acids, sterols, aminoacids, growth factors, enzymes etc. all excreted by the plant roots (Menon et al. 2013; Peña-Salamanca et al. 2013). This can regulate or even improve the microbial activity which in turn can result in the effective degradation of organic contaminants (Weber & Legge 2011). However, for this to occur plants must develop an extensive radicular zone where abundant microbial communities can thrive and this can take time (Montoya et al., 2010). Possibly, this is the reason why after 3–4 months the heliconia, now with a more developed radicular system, were able to maintain or even increase the COD removal levels even after a decrease of the HRT to 12 h.

Similarly to COD, the BOD removal efficiency of the wetland planted with H. psittacorum increased with time reaching its highest level in June (Figure 9) when the HRT changed and the highest rainfalls of the season occurred. The overall removal percentage for H. psittacorum was 48% at HRT of 24 h and 43.9% at HRT of 12 h, whereas for H. stricta the average removal was 48.9% with HRT of 24 h and 35.6% for an HRT of 12 h (Table 2). As discussed before, the BOD removal performance of the wetlands improved as the plants grew up and got more established in the CW. The COD to BOD ratio during most of the sampling was close to 1:1, suggesting that the organic matter in the wastewater was largely biodegradable (Sánchez 2008). Also, for most of the study period the effluents of the heliconia CW met the maximum allowable values of BOD (200 mg/l) according to Mexican standards (NOM-001-SEMARNAT-1996).

The efficiency of heliconia species to remove TN and TP was evaluated starting from May (operation day 84). The average TN removal was very similar between species, being 38.9% for H. psittacorum and 37.5% for H. stricta with a HRT of 24 h in both cases (Table 2). The highest TN removal levels were observed in June, reaching values of up to 47.9% for H. psittacorum and 41.7% for H. stricta (Figure 10). These results are superior to those previously reported for H. psittacorum at a laboratory level, where a TN removal efficiency of 4% was obtained with a HRT of 12 h (Konnerup et al. 2009). Nevertheless, the TN removal levels observed in this study are in agreement with the performance of wetlands with other plant species. Vymazal & Kröpfelová (2008) reported that the N removal in the majority of CWs is low, ranging between 40 and 55% and rarely is higher. These percentages are due to the inability of the system to provide simultaneous oxic conditions for nitrification and anoxic conditions for denitrification. The nitrification–denitrification process is widely recognized as the main route for the removal of N (Vymazal 2008). In the future, the TN removal efficiency shown by the heliconia wetlands could be expected to increase since the plants would have had more time to grow and to increase their foliage which would require higher levels of nutrient uptake, especially N and K (Rodríguez 2013). This has not been clearly observed in this study, possibly due to factors such as changes in hydraulic load and the immaturity of the plants.

In the case of TP removal, H. psittacorum showed an average percentage of 38.9% while H. stricta registered 28.5% (Table 2) but removal levels of up to 44.3 (H. psittacorum) and 36.7% (H. stricta) were recorded (Figure 11). These results compare well with those found by Konnerup et al. (2009), who reported a removal efficiency of 4% for H. psittacorum under similar conditions (HRT 12 h).

The concentration of TN in the effluent of the CW planted with H. psittacorum (Figure 10) was always below the maximum allowable value (60 mg/l) established by the corresponding Mexican norm (NOM-001-SEMARNAT-1996). Similarly, the concentration of TP in the effluents of both heliconia CW (Figure 11) met the Mexican standards (<30 mg/l) (NOM-001-SEMARNAT-1996).

The low levels of TP removal observed in this study are in agreement with previous results by other authors which have shown that vegetation removes phosphorus from wastewater through plant absorption but the amount removed is usually small compared to the phosphorus load entering the system (Browning & Greenway 2003; Vymazal 2004, Konnerup et al. 2009). Phosphorus removal is also variable as it depends on the hydraulic load rates, plant species and climate among other factors (Menon & Holland 2013; Vymazal 2004, 2007).

In the last month of the study, the concentration of TP was higher in the effluent of the CWs than in the effluent of the biodigester, resulting in negative percentage removal values (Figure 11). This could be due to litter decomposition and phosphorus release back into the system (Akratos & Tsihrintzis 2007). Furthermore, it could have been the effect of the increase of the hydraulic load that washed out some materials previously retained in the wetlands.

During the conduction of this research, several factors influenced the performance of the treatment system due to the difficulty in maintaining suitable conditions and in controlling variables that can affect performance in a full-scale system. The major limitations for the system were changes in the HRT, problems during the start-up of the anaerobic digester and leakages in two CW during the sampling period. The removal of organic matter by the anaerobic digester was very low probably because of the problems encountered in the start-up stage. The digester was not seeded with inoculum sludge and therefore the micro-organisms had to grow and colonize the digester and this usually can take up to 14 weeks (Álvarez et al. 2006) which is half the sampling period of this study. The lack of inoculum coupled to the variations in the hydraulic load might have affected reactor stabilization during the study period. Furthermore, the short study period could have repercussions for plant evaluation, meaning that a longer sampling period might be necessary. In the literature, one can find reports of wetlands with over 5 years or even decades of operation time (Chazarenc et al. 2009; Mustafa & Scholz 2010; Vera et al. 2011; Zhi & Ji 2012). It is also important that when evaluating and comparing plant species the plants are healthy and mature (Brisson & Chazarenc 2009).

It should be noted that the CWs showed a good performance even though the pre-treatment by the anaerobic digester presented a poor efficiency. In other words, the process of removal or transformation of contaminants was made largely by the CWs. In addition, despite having twice the hydraulic load in the last 2 months of operation, the removal percentages of COD and BOD remained without major changes. The growth of heliconias under the CW conditions is also worth noting since they showed the ability to adapt and survive in an environment saturated with water and a gravel support, probably due to the development of arenchyma or air-space tissue in leaves, stems and roots. Arenchyma tissue is also present in Typha domingensis providing an intercellular pathway for rapid flux of oxygen. This is a growth strategy of plants in flooded soils (Chabbi et al. 2000; De la Cruz et al. 2012). Furthermore, although the mechanism was unclear, the CWs were able to achieve removal efficiencies comparable to the levels observed in laboratory scale systems (Konnerup et al. 2009).

A preliminary measurement of TC and FC concentrations in the effluent of the H. psittacorum CW was conducted in order to evaluate the possibility of manipulating and marketing the produced flowers in the future. The values obtained for TC were 24 × 10⁴MPN/100 ml and for FC the level was found to be above the maximum allowable value of 1,000 MPN/100 ml in treated water for public service use with indirect or occasional human contact, according to Mexican standards (NOM-003-SEMARNAT-1997). However, no repetitions of the analysis were performed and the results are inconclusive. In addition, the measurement was made on day 134 of operation when the CWs operated at twice the original value of the hydraulic load (i.e. at 12 h HRT). For that reason, the shorter HRT may have influenced the results. It has been reported that retention times from 3 to 7 days are necessary to decrease the coliform numbers by one or two logarithmic units whereas HRTs greater than 14 days would be needed to achieve reductions of three to four logarithmic units (Lara 1999).

H. psittacorum and H. stricta are ornamental plants able to adapt and grow well in CWs with a gravel support, with subsurface horizontal flow and fed with domestic sewage. This ability makes them an option for improving the aesthetic appearance of this type of technology and by this increasing public acceptance of wastewater treatment systems.

Despite the unfavorable conditions that occurred during the conduction of this study, the system achieved important removal levels when compared to studies done with H. psittacorum at a laboratory scale. Similar removal percentages of COD, BOD were observed and the results were even better in the case of TN (30% more) and TP (15% more).

This research provides important information regarding the performance of ornamental plant species for the treatment, at full scale, of domestic wastewater under the climate conditions of the Socunusco region in Chiapas, Mexico. The evaluated heliconias seem to play a role in the removal of pollutants, however it is unclear what mechanisms are involved. Therefore, further studies are necessary to understand the removal mechanisms of specific pollutants.

The authors would like to thank ‘La Concepción’ coffee farm for the facilities provided during the conduction of this study. ASMM gratefully acknowledges the scholarship granted by Consejo Nacional de Ciencia y Tecnología (CONACyT), Mexico. The authors are also indebted to Dr Tom Cochrane (University of Canterbury) for his valuable comments and English editing of the manuscript.