Pollutant removal and growth dynamics of macrophyte species for faecal sludge treatment with constructed wetland technology

On-site sanitation systems (OSS) are currently the prevalent excreta management solution in developing countries of sub-Saharan Africa. Currently, they serve between 2.1 and 2.6 billion people in low- and middle-income countries (Strande et al. 2014). With over 80% level of coverage (Diener et al. 2014), there is consequently continuous accumulation of large volumes of faecal sludge in the pits and vaults of the OSS installations. Management of faecal sludge is, however, characterised by collection and treatment of only a small fraction while the remainder is discharged indiscriminately into the environment, posing various degrees of environmental and health risk. In Ouagadougou, the capital city of Burkina Faso, about 88% of the entire population is served with on-site sanitation installations (Bassan et al. 2013; Williams & Overbo 2015). Meanwhile, less than 40% of faecal sludge in Ouagadougou is collected and treated while the rest is discharged openly into the environment. This situation is noted to account for the high prevalence of waterborne diseases (Bassan et al. 2013). Faecal sludge (FS) management interventions are thus needed to squarely address the challenge of establishing safe and hygienic systems for the management of sanitation infrastructure. Yet low-cost FS treatment options that allow for biosolid reuse with minimum environmental risk are generally lacking. Studies over the years have emphasised the positive potential of constructed wetland (CW) treatment technology for FS management (Kengne et al. 2008; Kengne & Tilley 2014; Kouawa et al. 2015; Andrade et al. 2017). The system is composed of a filter layer of sand, supported at the base with gravels and planted with emergent macrophytes. Dehydration and mineralisation are thus important hydraulic and biochemical processes of the CW system. Studies by Stefanakis & Tsihrintzis (2011) found evapotranspiration to be responsible for almost 58 to 84% of total water losses, while draining accounts for 13 to 41% with a remnant of 1 to 4% in the sludge layer. The dewatered sludge hence undergoes the process of stabilisation, disinfection and mineralisation, where organic matter is stabilised by microbial decomposition for which microbiologically active macrophyte root zone layers play a significant role. Over a considerable period of operation, a CW has the potential to produce biosolid of good quality, which can be utilised for soil amelioration and nutrient management purposes. However, technology transfer and application of the CW technology within the local context is limited. A major restraining factor is the identification of indigenous plant species adaptable to the noxious conditions of faecal sludge. Several plant species have over the years proven to withstand the harsh conditions of FS, among which Phragmites sp. (reeds), Typha sp. (cattails), Cyperus papyrus (papyrus) and Echinochloa sp. are paramount (Kengne & Tilley 2014). Results of the premier study in the Sudano-Sahelian context showed that the species of Oryza longistaminata and Sporobolus pyramidalis are not suitable for the CW system after total withering and subsequent death when loaded with faecal sludge (Kouawa et al. 2015). Another study by Sawadogo et al. (2016) assessed the potential of Andropogon gayanus and Cymbopogon nardus in the same context. However, the study was limited to the evaluation of effluent quality without taking account of the species' physiological development. A follow-up pilot-scale experiment by Osei et al. (2017) revealed a satisfactory macrophyte adaptability potential of four native species, Setaria verticillata, Cymbopogon nardus, Bambusa vulgaris, and Typha latifolia L. to wastewater pollution with potentials for faecal sludge load. This study assessed the macrophyte potentials of Bambusa vulgaris and Cymbopogon nardus species for faecal sludge treatment.

This study was undertaken at the International Institute of Water and Environmental Engineering (2iE), Ouagadougou, Burkina Faso, in the Sudano-Sahelian ecological zone. Mean annual rainfall is reported within the range of 600 to 900 mm from May to October. Temperatures between 42 and 43 °C are experienced during the hottest season of the Harmattan, whereas the normal range is between 25 and 28 °C (Ibrahim et al. 2012). The zone is characterised by a long period of dry season principally influenced by the monsoon and Harmattan. The study was conducted from October 2017 to May 2018 where there was virtually no influence of precipitation.

Bambusa vulgaris (BV) and Cymbopogon nardus (CN), locally known as bamboo and citronella, were planted and monitored for the study after the period of 8 and 11 weeks respectively for wastewater acclimatisation (Osei et al. 2017). Both species from the family of Poaceae (Table 1) are noted to have important economic potential with positive environmental impact.

The pilot-scale vertical flow CW was made up of a 27 litre plastic reactor (pot), filled with sand up to 15 cm and supported at the base with a 5 cm layer of gravels to mimic a typical CW condition (Table 2). Both media were shifted and thoroughly washed for removal of a mixture of clay and organic impurities. A freeboard of 5.5 cm was allowed for sludge loading. An under-drain system which consisted of a perforated pipe was attached to a regulator tap at the base for dewatering and monitoring of effluent quality. The experiment was a completely randomised design with four experimental units (pot) per plant species (three treatments and a control setup) (Figure 1). Planting of species was in the order of three tufts and four plants per pot for bamboo and citronella respectively.

The study employed APHA standard methods for the examination of raw sludge and WW quality parameters. Laboratory analysis was carried out for NH₄-N, PO₄³-P, total suspended solids (TSS), total volatile solids (TVS) and chemical oxygen demand (COD). The oven and muffle furnace, DR 5000 spectrophotometer and DRB200 reactor aided in the determination of the solids, nutrients and organic contents respectively, while a Hanna instrument (HI 9828) was utilised for on-site analyses of pH, redox potential (Eh), electrical conductivity (EC) and total dissolved solids (TDS).

Pollutant removal efficiencies were evaluated based on differences between influent and effluent concentrations. The analyses were for FS/WW load due to inadequate effluent liquid volumes from the FS load for laboratory analyses.

Species growth parameters were monitored: the number of plants (NP) and number of leaves (NL) by count, and plant height (PH) using a standard (3 m/10 ft) tape measure at weekly intervals.

The development of selected growth parameters was subjected to trend analysis at monthly interval while an independent-sample T-test at p ≤ 0.05 was performed to determine the level of statistical significance between treatment (FS) and control (WW). The relationship between pollutant removal and plant growth was determined by Pearson correlation at p ≤ 0.05.

PH development for CN generally realised an increasing trend despite the slight decrease in the third month. Both the treatment and control progressively increased from a starting average of >50 cm to over 65 cm/pot at the end of the study (Figure 2). The variation between treatment (FS = 56.9 ± 9.4 cm) and the control (WW = 57.3 ± 9.1 cm) was statistically insignificant at p = 0.88. The result suggests that the development of CN PH was not significantly affected by the faecal sludge load. Scutti (2013) noted Citronella nardus as a perennial, clump-forming tropical grass which could grow up to 1.5–1.8 m tall with cane-like stems.

For PH of BV species, the treatment setup increased from a starting average of 116.4 ± 0.7 cm to a maximum of 127.6 ± 7.4 cm/pot in the third month, followed by an appreciable decrease to a minimum of 117.4 cm/pot in the fourth month. The subsequent months recorded a virtually stabilised height development of over 120 cm/pot (Figure 2). The control similarly increased from 87.2 ± 0.7 cm to a maximum of 106.5 ± 1.4 cm in the third month, after which a gradual decrease was observed from 101.7 ± 10.5 cm in the fourth month to a minimum of 84.8 ± 1.3 cm/pot in the seventh month. The final month (8), however, recorded an appreciable increase of 92.9 ± 6.8 cm/pot. Variation between the treatment (FS = 121.4 ± 6.5 cm) and the control (WW = 93.5 ± 8.7 cm) was statistically highly significant at p = 0.00. Vigorous PH development is a significant physiological characteristic of bamboo species. Height of culms could reach up to 30 m at a potential growth rate of 1.2 m per day, coupled with high consumption of carbon dioxide while a significant quantity of oxygen is produced (Bamboo Information Network 2011).

The number of CN plants for the control increased from a minimum of 13 ± 1 in the first month to a maximum of 31 ± 1 plants/pot at the end of the study (Figure 3). The treatment averagely recorded an appreciable increase from 19 ± 2 from the beginning to a maximum of 25 ± 3 plants/pot at the second month, after which there was a gradual decrease from the third month to a minimum of 14 ± 1 plants/pot at the end of the study. The observed trend was due to gradual wilting of a treatment setup from a maximum of 18 plants in the ninth week to a minimum of 5 plants from the twenty-fifth week to the end of the study. This could be attributed to the prevalence of noxious conditions detrimental to plant health. The difference between treatment (FS = 19.1 ± 4.3) and the control (WW = 23.68 ± 6.5) was statistically highly significant at p = 0.003.

The average development of BV culms/shoots for the treatment gradually decreased from a maximum of 11 ± 1 at the beginning to a minimum of 5 ± 1 plants/pot at the end of the study. Similar to CN, the observed reduction was due to a gradual wilting of a treatment setup. The control setup recorded a slight variation in the NP from 9 ± 1 plants in the first month to a maximum of 10 plants/pot at the end of the study, after a slight reduction to a minimum of 8 plants/pot at the fifth and sixth month (Figure 3). The observed variation between the treatment (FS = 8 ±2) and control (WW = 9 ±1) was statistically significant at p = 0.003.

The monthly development of CN leaves from a starting average of 87 ± 10 for the treatment setup appreciably decreased from a maximum of 162 ± 5 in the third month to a minimum of 70 ± 1 leaves/pot at the end of the study (Figure 4). This might be a coupling effect of the observed decrease in the NP within the same period. The control setup progressively increased from 69 ± 9 to 195 ± 5 leaves/pot at the end of the study. The difference between the treatment (FS = 119.7 ± 37.7) and control (WW = 124.3 ± 45.9) was not statistically significant at p = 0.685.

For the BV species, the treatment setup increased from an average of 533 ± 20 to a maximum of 700 ± 44 leaves/pot in the fourth month. Defoliation of over 250 leaves was, however, observed in the fifth month while a vigorous development up to 567 ± 65 leaves/pot occurred from the sixth to the end of the study. The control demonstrated a rapid leaf development to a maximum of 756 ± 72 leaves, after a considerable reduction to a minimum of 197 ± 144 within the third and fourth month (Figure 4). The phenomenon of bamboo defoliation could be influenced by the senescence of old leaves. However, the rapid loss of soil moisture through evapotranspiration due to the high intensity of the Harmattan may have affected the control setup within the second and fourth month while sludge build-up on the treatment setups served as mulch. However, the rapid leaf recovery by the control demonstrates a positive stress recovery potential of the bamboo species. There was no significant difference between the treatment (FS = 559 ± 100) and the control (WW = 492 ± 195) at p = 0.09. The ability of BV to adapt to difficult wetland conditions is explained by its morphological internal air spaces feature for transportation of oxygen to the roots and rhizomes (Brix 2003) together with the presence of aerenchyma structures for carbon sequestration (Zachariah et al. 2016).

Dynamics of selected physico-chemical parameters (Table 3) were monitored for assessment of pollutant reduction for the treatment setup.

The pH for the treatment loads of FS and FS/WW generally decreased from alkaline to acidic and nearly neutral in the effluents of CN and BV respectively. A lower pH for FS (9.1 ± 0.7) was recorded by Kouawa et al. (2015). Kengne et al. (2014) similarly realised a pH of 7.13 ± 0.58 in leachate from FS drying beds planted with Echinochloa pyramidalis.

The Eh increased from negative to positive in the effluents of both species. The variation among average Eh was statistically significant at Fpr (F probability) of 0.011. Positive Eh indicates the tendency of chemical species to gain electrons and thereby be reduced while negative reduction potential implies the tendency to lose electrons and thereby be oxidised (Szymczycha & Pempkowiak 2016).

Dilution of the raw FS with WW decreased EC for the FS/WW load while the effluent of BV was relatively higher than that of CN. Kouawa et al. (2015) recorded higher EC for raw FS (6,600 ± 2,200 μS/cm) while the effluent of CN was closer to what Kengne et al. (2014) recorded.

The effluent of CN recorded the lowest COD level as compared to BV. The concentration was, however, higher than 652 ± 558 and 515 ± 1,316 mg/l in FS percolate as recorded by Kengne et al. (2014) and Andrade et al. (2017) respectively. The maximum COD for the raw FS was reduced by dilution with wastewater (FS/WW) and was comparatively higher than those reported by the different authors (Kouawa et al. 2015; Sawadogo et al. 2016; Andrade et al. 2017).

A relatively lower concentration of NH₄-N was observed for the effluent of CN than BV. Kengne et al. (2014) and Andrade et al. (2017) comparatively realised lower NH₄-N concentrations of 38.70 ± 20.50 and 29 ± 23 mg/l respectively. The variation among the various treatment loads was statistically highly significant at Fpr of 0.001.

The concentration of PO₄³-P for raw FS reduced in the FS/WW treatment composition. A lower effluent concentration was realised for BV as compared to CN. However, the results were higher than the percolate concentration of 10.5 ± 9.31 mg/l by Kengne et al. (2014).

The selected solid compositions of TVS, TSS, TDS, and TS were generally higher for the effluent of CN as compared to BV. Except for TDS, variation among the various compositions was statistically highly significant at Fpr ≤ 0.005.

The study realised pollutant removal efficiencies between the ranges of 45.6–95.7 and 37.7–92.8% for BV and CN respectively. Both species essentially recorded an appreciable efficiency (>80%) for the selected pollutant except for TDS (<50%). The performance of BV was slightly better than that of CN, with relatively higher removal efficiencies for the selected pollutant with the exception of COD and NH₄-N (Figure 5). Using Echinochloa pyramidalis, Kengne et al. (2009) noted similar efficiencies of 77, 86, 90 and 95% for NH₄⁺, TSS, TS and COD respectively. On the contrary, results of Sawadogo et al. (2016) were slightly lower (<80%). Andrade et al. (2017), after realising lower efficiencies of 51, 52 and 65% respectively for TS, TVS, and NH₄⁺-N, but 82% for COD, noted lower influent concentration and larger grain size of the filter (2.4–12.5 mm) as possible reasons, for Cynodon dactylon Pers. species.

The CW functions through complex physical, chemical and biological processes which may occur by a series of interactions for removal of pollutant (Stefanakis et al. 2014). Solid contents are removed by the processes of filtration and sedimentation in which particles are mechanically strained by flow constrictions or settle into stagnant micro pockets (Kadlec & Wallace 2008). Removal of organic matter is attributed to aerobic and anaerobic degradation by microorganisms (Moshiri 1993; UN-HABITAT 2008) as well as sedimentation, filtration, sorption and volatilisation (Rai et al. 2015). Nitrogen is removed mainly by the processes of ammonification, nitrification, and denitrification (that is, the oxidisation of ammonia to nitrate by nitrifying bacteria in aerobic zones, followed by denitrification to dinitrogen gas by denitrifying bacteria in anoxic and anaerobic zones). Other important removal processes include volatilisation, plant uptake and matrix adsorption (Rai et al. 2015). Phosphorus is removed by the processes of filtration, adsorption, complexation, precipitation, storage, plant uptake and biotic assimilation (UN-HABITAT 2008; Stefanakis et al. 2014).

The results from the Pearson correlation generally revealed a negative linear relationship between species development and pollutant removal with coefficients ranging from strong negative (r > −0.9) to moderate positive (r < 0.6), with the exception of TDS and TSS which demonstrated a strong positive relationship (r = 0.81) with CN-PH (Table 4) accounting for 65.61% (coefficient of determination) of the observed efficiencies. The correlations were statistically not significant at p < 0.05. The different mechanisms of the CW might have significantly influenced the overall processes of pollutant removal. Cheng et al. (2009) noted a lower correlation with plant growth indexes and COD removal while Read et al. (2009) similarly recorded a strong negative correlation between plant traits (especially root size development) and effluent N and P concentrations.

Sludge loading rate is the mass of solids dried on a square metre of bed for a year (Dodane & Mariska 2014). Total annual sludge loading rate of 31.4 and 103.4 kg TS/(m²·yr) were determined for FS/WW (acclimatisation phase) and FS load respectively (Figure 6). Kengne et al. (2008), however, applied a higher annual sludge loading rate (100–300 kg TS/(m²·yr)) at a typical yard-scale dewatering experiment for the different species of Echinochloa pyramidalis and Cyperus papyrus.Sonko et al. (2014) in a similar study adopted a sludge load of 50 to 150 kg TS/(m²·yr) for the acclimatisation of Echinochloa pyramidalis species.

Despite the varied trends for species growth parameters, both species demonstrated satisfactory performance for survival in polluted conditions of faecal sludge. The variation between the control and treatment showed significant difference for the NP, but not for the NL. However, the average pH for the treatment setup of BV was significantly higher than the control. Both species recorded appreciable pollutant removal efficiency (>80%) for the organic (COD) nutrients (PO₄^3_P and NH₄-N) and solid (TSS and TVS) compositions; however, the performance of BV was slightly better than CN for all parameters except for COD and NH₄-N. The study revealed a weaker relationship between species development and pollutant removal. Regardless of the different variations exhibited between species, the results generally showed satisfactory performance of resistance to faecal polluted wetland conditions and thus suggest positive macrophyte potential for the faecal sludge CW treatment technology.

The authors express gratitude to the International Institute for Water and Environmental Engineering (2iE), University for Development Studies and the World Bank, Africa Centres of Excellence (ACE) Project for their financial support for this study.