Investigation of the potential of buffalo and couch grasses to grow on AFIs and for removal of nutrients from paper mill wastewater

The pulp and paper industry (P&PM) is a large user of wood and fossil fuel energy, as well as requiring large volumes of fresh water for wood processing for pulp and paper production. Modern pulp mills typically consume about 10 to 50 m³ per tonne of paper produced (Toczyłowska-Mamińska 2017) and up to 70 m³ per tonne of paper (Hubbe et al. 2016). With the large use of water by the industry there are also large volumes of wastewaters produced, though modern mills in developed countries incorporate advanced wastewater treatment methods for water reuse purposes (Hubbe et al. 2016; Toczyłowska-Mamińska 2017). The P&PM industry is the world's third largest producer of industrial wastewater after primary metals and chemicals industries (Ashrafi et al. 2015). The industry uses a wide range of treatment processes including primary and secondary treatment as well as advanced tertiary processes to treat wastewaters for recycling purposes and for improved wastewater quality prior to discharge into the environment (Savant et al. 2006; Lewis 2012; Hubbe et al. 2016; Toczyłowska-Mamińska 2017). Nonetheless, P&PM wastewaters can contain a vast range of polluting constituents following treatment, such as nutrients (nitrogen (N) and phosphate (P)) that can lead to eutrophication in receiving waters. The discharge of P&PM wastewaters from the mill can significantly impact on the quality of receiving waters due to high loadings of remaining pollutants such as colour, nutrients (N and P), chemical oxygen demand (COD) and wide range of organic compounds, some of which are halogenated when chlorine-based bleaching of pulp is applied (Savant et al. 2006; van Leeuwen et al. 2012; Hubbe et al. 2016; Toczyłowska-Mamińska 2017).

Harmful algal blooms (i.e. blue-green algae or cyanobacterial) and their impacts are a recurring and significant issue worldwide (Puschner 2018). Factors supporting the extensive growth of cyanobacteria include warmer temperature, sufficient levels of reactive nitrogen and phosphorus, reduced rainfall, long hydraulic retention times within the water body (slow-moving water) and adequate sunlight (Smith 1983; Piontek et al. 2017; Puschner 2018). Cyanobacteria are photoautotrophic, and light is the key factor for their growth, followed by nutrients levels (Deas & Orlob 1999; Kirkwood et al. 2003). Kirkwood et al. (2003) reported that pulp and paper aerated stabilization basin (ASB) systems are severely light limited. Cyanobacteria have been reported to rarely occur in pulp and paper ASBs because of the dark colour of wastewaters and mixing of waters by aerators (Mahmood & Paice 2006). However, with the relatively low colour of paper mill wastewater (without integrated pulping processes) in ASB treatment with long hydraulic retention times (HRTs), there would be a risk of occurrence of cyanobacteria at bloom levels (Lewis et al. 2018).

Cyanobacteria can cause serious problems in water bodies, being potentially toxic to fish, invertebrates and vertebrates, as well as causing a reduction in plants and algae which form part of the food chain needed for healthy ecosystems in the water body (Kallas & Munter 1994; Piontek et al. 2017). Problems associated with the growth of cyanobacteria include formation of unpleasant odors and toxins that have the potential to harm people and wildlife that come into contact with contaminated water (Kallas & Munter 1994; Havens 2004; Piontek et al. 2017).

Artificial floating islands (AFIs) or floating treatment wetlands (FTW) are an emerging technology (Chang et al. 2015) used in remediating a variety of water types (Wu et al. 2015). AFIs employ a floating mat as a substrate to grow selected plants and grasses hydroponically. Plants are selected for their ability to remove nutrients, pollutants and contaminants from various water sources (Geng et al. 2017). Studies have also shown that the use of an AFI combined with selected plants can reduce nutrients in water and inhibit the growth of cyanobacteria (e.g. Chang et al. 2012). Water purification through fixed wetlands is a common practice (Webb et al. 2012) often installing a wetland is not feasible in an industrial setting due to the inappropriate surrounding substrate, amount of land needed, the cost and maintenance of the systems. The ability to tailor an AFI with selected plants for water remediation in industry has seen an increase in the use of AFIs in the last 20 years (Vymazal 2013). In this paper, we report the results of investigations on the potential growth of two grasses (couch and buffalo) under AFI conditions and their ability to remove nutrients, with an aim to apply AFI to improve the quality of wastewaters from paper mill operations.

Blue-green algae blooms have occurred sporadically at small scale in all ponds of an ASB system that treats the wastewaters of a paper mill in South Australia. Being a paper mill only (no pulping processes), the wastewaters are low in colour and with primary treatment by a settling clarifier, turbidity is low. The growth of cyanobacteria is due to sufficient levels of nitrogen and phosphorus compounds in the wastewaters, long HRTs within the ASB ponds and due to favourable weather conditions (i.e. warmer temperature and adequate sunlight). The ASB comprises three ponds originally designed to function as (1) a continuous stirred tank reactor, (2) plug flow pond and (3) a maturation pond, respectively. The ASB has a surface area of ∼23 hectares, a total holding capacity of 870 mL, and under current operations, has a wastewater residence time of ∼80 days. The discharge of treated wastewater is to a coastal lake. The three ponds of the ASB system are clay lined, up to ∼4 m deep, have steep, rock-lined sides and therefore are not suitable for redesign as a fixed wetland system.

Low cost, highly robust and rapidly growing grasses were sought as potential candidates for AFI for paper mill wastewater treatment. Two types of lawn grasses, buffalo grass (Stenotaphrum secundatum) and couch grass (Cynodon dactylon), were selected due to their reported nutrient removal rates in soil (Fresenberg & Miller 2006) and suitability for growth in South Australian climate conditions (Jewell et al. 2012). Both grasses have been shown to have resistance to drought and grow well under high nutrient conditions (Tran et al. 2016). In this study, these grasses were tested for growth in synthetic water prepared to simulate the wastewater quality of the ASB.

A small-scale trial was conducted inside a greenhouse, predominantly during summer (Adelaide, South Australia) for a period of 14 weeks, equivalent to 1.2 × residence time of the full-scale ASB. Air temperature (°C) and humidity (%) levels inside the greenhouse were measured weekly using a Digitech QP6014 temperature and humidity data logger and are presented in Table 1. Solar radiation data were obtained from the Bureau of Meteorology (BoM) and the Queensland Government's Scientific Information for Land Owners (SILO) service. Recorded temperature (°C) and humidity (%) inside the greenhouse for this period were as follows: minimum ∼16 °C and 6%, maximum ∼47 °C and 93% with an average of ∼27 °C and 45%. Local temperature (°C), humidity (%) and solar radiation (MJ/m²) data sourced from BoM during the study period are as follows: minima, ∼8 °C, 7% and 9 MJ/m²; maxima ∼44 °C, 86% and 32 MJ/m².

Eighteen mesocosms were established using 54 L black plastic containers (64 cm × 42 cm × 18 cm). Nine were filled with 30 L of reverse osmosis (RO) water and nine were filled with synthetic water made with ammonium dihydrogen phosphate, potassium nitrate, sodium sulphate, calcium chloride dehydrate, magnesium sulphate heptahydrate, potassium sulphate and sodium chloride salts to the following concentrations (in mg/L): TP 0.41; TN 0.65; sulphate 129; calcium 70; magnesium 25; sodium 106, potassium 75 with a pH of 7.24 and conductivity of 1,571 μS/cm. For this study, data acquired of the quality of wastewaters of the paper product manufacturing mills during and after the closer of the pulping process between January 2011 and March 2016 were used to guide the preparation of synthetic water. Ranges of nutrient and mineral concentrations (in mg/L) in discharge treated wastewaters were as follows: TP, 0.02–0.88; TN, 1.8–7.4; NOx: 0.011–2.19; sulphate, 58–912; calcium, 60–823; magnesium 23–155; sodium 84–214 with a pH of 6.6–9.1 and conductivity of 826–2,331 μS/cm.

Each mesocosm was initially filled with 30 L of the tested water and the mesocosms were topped up weekly with either RO water or synthetic water to maintain the initial volume of the mesocosms (30 L). Each AFI was made from polyethylene foam (23 cm × 29 cm × 2.5 cm covering 25% of the surface area of the container) with 48 evenly spaced planting holes, as shown in Figure 1. The grasses were rinsed (x5) with RO water to remove soil contaminants. The buffalo and couch grasses with similar size (roots: 4 cm each; shoots: 5 cm each) were placed into the AFI's holes (buffalo grass: one stem per hole; couch grass: 5–10 blades per hole). The experiment was performed in triplicate.

For nutrient and mineral analysis, 600 mL PET sample bottles were thoroughly washed with RO water. Following air drying, water samples were then collected from each mesocosm, and composites from each triplicate were sent for analysis. The samples were stored at <4 °C until analyses which was conducted by a National Association of Testing Authorities (NATA) accredited laboratory. The following methods were applied, TN – APHA 4500 Norg/NO₃, TP – APHA 4500 P –F and major cations (potassium, sodium, calcium, magnesium) – APHA 3120B (APHA 2012). The pH, conductivity and temperature were measured on a weekly basis using a TPS 90-FL field laboratory analyzer. The instrument was calibrated daily when in use using Ajax Finechem Pty Ltd laboratory chemical buffer solutions pH 4 and 7 and 2.76 mS/cm conductivity standard. Organics (measured as total organic carbon (TOC) and dissolved organic carbon (DOC)) were measured in water samples collected at the end of the trial period using the standard method of APHA 5310 B (APHA 2012).

Root and shoot growths for each AFI were measured at the end of the study period. Two holes for each row on the grid of the AFI (12 holes in total from the original 48) were randomly selected. For consistency, grass shoots were measured if they had grown (both existing and new shoots) over 5 cm and new roots that had grown through the duration of the experiment.

Statistical analyses were performed using IBM SPSS Version 24. Kolomogorov's test of normality was performed to determine parametricity of the data. If the data were parametric, the t-test was performed and if non-parametric, the Mann-Whitney test was performed. Test for differences between tests and controls were performed as described by (Keizer-Vlek et al. 2014).

The survival of the sprigs was assessed for each grass type and no difference was found in the overall survival rates between buffalo and couch grasses (buffalo: 93% for RO water and 93% for synthetic water; couch: 95% for RO water and 92% for synthetic water). Using the wet weight values, the percentage increase of buffalo and couch grasses weight and the relative growth rate (RGR) during the study period were calculated and the average values are presented in Table 2. In this study and despite the grass types, addition of nitrogen and phosphorus at low levels were found to have no effect on the grass growth measured by the RGR (buffalo: 0.009 for RO water vs 0.009 for synthetic water; couch: 0.008 for RO water vs 0.007 for synthetic water), as shown in Table 1. This finding indicates that these plants acquired essential elements from airborne sources and have the ability to grow in differing nutrient conditions. Office of the Gene Technology Regulator Department of Health Australian Government (2018) reported that Stenotaphrum secundatum is tolerant of low nutrient levels presented in the soil.

For both RO and synthetic water, buffalo grass was found to have slightly higher percentage increase in weight gain compared to that for the couch growth at the corresponding water (RO water: 148 ± 26% for buffalo vs. 125 ± 36% for couch; synthetic water: 133 ± 34% for buffalo vs. 112 ± 46% for couch). Menzel & Broomhall (2006) reported that buffalo grass is less sensitive to the relatively lower nutrient levels in soil compared to the couch grass. In contrast, the RGR data showed a different trend with growth values found to be similar for both buffalo grass and couch grass (Table 2).

Root and shoot lengths were measured for both grass types after 14 weeks (at the end of the experiment) and average values are presented in Figure 2. In this study, the addition of nitrogen and phosphorus at low levels was found to have no effect on the growth of roots and shoots for both grasses (based on the Mann-Whitney U-test results, p-value >0.05: 0.66 and 0.40 for buffalo and couch roots, respectively, 0.53 and 0.61 for buffalo and couch shoots, respectively), Figure 1. In contrast, despite the water types, average root length for buffalo grass was found to be significantly higher (with p-value = 0.01, <0.05) than that for the couch grass (RO water: 30 ± 2.0 cm for buffalo vs. 3.9 ± 0.1 cm for couch; synthetic water: 32 ± 2.0 cm for buffalo vs. 4.0 ± 1.0 cm for couch). However, the buffalo grass was found to have lower number of new roots compared to couch grass (RO water: 109 for buffalo vs. 201 for couch; synthetic water: 123 for buffalo vs. 36 for couch). Further, despite the water types, no significant difference (p-value >0.05) was found in the average shoot lengths between both buffalo and couch grasses (RO water: 16 ± 0.4 cm for buffalo vs. 15 ± 0.3 cm for couch; synthetic water: 16 ± 0.3 cm for buffalo vs. 15 ± 0.3 cm for couch), Figure 1. While the buffalo grass was found to have a higher number of new shoots compared to couch grass (RO water: 127 for buffalo vs. 201 for couch; synthetic water: 140 for buffalo vs. 36 for couch). Further, buffalo grass was found to increase organics to the water compared to couch grass (TOC: 20 mg/L for buffalo vs. 7 mg/L for couch; DOC: 7 mg/L for buffalo vs. 6 mg/L for couch).

Average weekly pH and conductivity levels of the water samples during the study period were measured and are presented in Table 1. No differences were found in pH levels of water samples collected from buffalo and couch grass mesocosms (buffalo: 7.5 ± 0.3 for RO water and 7.5 ± 0.3 for synthetic water; couch: 7.6 ± 0.4 for RO water and 7.7 ± 0.3 for synthetic water). Compared to control mesocosms (mesocosms with RO water or synthetic water without grasses: 7.3 ± 0.5 for RO water and 7.4 ± 0.2 for synthetic water), water samples collected from mesocosms with buffalo or couch grasses were found to have higher pH levels (Table 1). Average conductivity levels (in μS/cm) followed the same trend with values higher in water samples collected from mesocosms with buffalo or couch grasses compared to control mesocosms (buffalo: 284 ± 185 for RO water and 1,684 ± 456 for synthetic water; couch: 326 ± 196 for RO water and 1,682 ± 450 for synthetic water; control: 78 ± 14 for RO water and 1,389 ± 297 for synthetic water). The conductivity levels in water samples collected from mesocosms with buffalo or couch grasses increased significantly from week 6 onwards (buffalo: 113 μS/cm during the first five weeks vs 455 μS/cm from week 6 onwards for RO water and 1,247 μS/cm vs. 2,121 μS/cm for synthetic water; couch: 210 μS/cm vs. 441 μS/cm for RO water and 1,270 μS/cm vs. 2,093 μS/cm for synthetic water, see Table 1).

Weekly additions of synthetic water (L), nutrients and minerals (mg) to the mesocosms over the 14-week study are shown in Table 3. Total nutrients and minerals added and final masses (mg) in the mesocosms are as follows: For buffalo grass: TN, 61/33; TP, 39/7; sulphate, 12,141/6,559; calcium, 6,588/5,168; magnesium, 2,352/1,543; sodium, 9,976/14,385; potassium, 7,059/324. For couch grass, TN, 60/33; TP, 38/6; sulphate, 11,905/8,042; calcium, 6,459/5,700; magnesium, 2,307/1,800; sodium, 9,780/11,820; potassium, 6,921/480. In open RO control mesocosms, there was an average increase in the sodium by 129 mg/L and sulphate by 71 mg/L. These indicated airborne sources. For other water quality parameters tested, airborne sources were minor. Percentage uptakes of nutrients and minerals by the buffalo and couch grasses over the 14-week experiment are shown in Figure 3.

Similar to the RGR values, no difference was found between the potassium uptake by buffalo grass compared to that by couch grass (Figure 2) with average effluent potassium concentration decreased by 95% and 93% in buffalo and couch grasses, respectively. In contrast, sulphate, calcium and magnesium uptakes were found to be higher by buffalo grass compared to the corresponding mineral by couch grass (sulphate: 46% vs. 32%; calcium: 22% vs. 12%; magnesium: 34% vs. 22%).

Similar to the potassium uptakes, no difference was found between the TP uptake by buffalo grass compared to that by couch grass, Figure 3. Overall, average effluent TP concentrations were decreased by 82% and 83% in buffalo and couch grasses, respectively. The net phosphorus uptakes by buffalo and couch grasses were found to be 0.47 g P/m² and 0.48 g P/m² of the AFIs, respectively. The TN uptake data showed a similar trend with the percentage N uptakes by buffalo grass (47%) found to be similar to that by couch grass (45%). The net nitrogen uptakes by buffalo and couch grasses were found to be 0.43 g N/m² and 0.41 g N/m² of the AFIs, respectively. This finding is consistent with those of previous studies (Wu et al. 2013; Wu et al. 2015; Geng et al. 2017) for various emergent wetland plants, indicating the capability of both buffalo and couch grasses to treat wastewaters, i.e. reduce the nutrients concentrations present in wastewater. Geng et al. (2017) reported that average effluent TP concentration was reduced by ∼78% by four plant species (Rumex japonicas, Oenanthe hookeri, Phalaris arundinacea and Reineckiacarnea) in hydroponic microcosm. Wu et al. (2013) reported that the net phosphorus uptake was 0.27–1.48 g P/m² in constructed wetland planted by four emergent wetland plants (Phragmites australis, Typha orientalis, Scirpus validus, Iris pseudacorus) while Wu et al. (2015) reported that the wetland plants decreased the effluent TP and TN concentration by 24–80% and 15–80%, respectively. Compared to soil-based buffalo and couch grasses, the results found in the present study are similar to that previously reported. Milandri et al. (2012) reported that soil-based buffalo grass reduced TP level presented in urban stormwater by 91% while the study conducted by Menzel & Broomhall (2006) indicated the ability of both soil-based buffalo and couch grasses to uptake nutrient from the soil.

As both grasses were chosen for their ability for nutrient uptake in soil (Menzel & Broomhall 2006; Milandri et al. 2012) and growth in South Australia (Jewell et al. 2012), the results reported here indicated that these grasses are able to reduce the nutrients levels in a hydroponic setting. Residual TP concentrations in synthetic waters treated by buffalo and couch grasses were 0.23 mg/L and 0.21 mg/L, respectively, while residual TN concentrations were less than 1.5 mg/L (1.1 mg/L for both grass types), indicating a potential of buffalo and couch grasses on AFI to lower nutrient levels in ASB. Previous studies (Chua et al. 2012; Borne et al. 2014) have shown that artificial island structure can limit light penetration to the water column and, consequently, might influence growth of cyanobacteria. However, the significance of the physical restriction of light to the water column by the AFI infrastructure will depend on the surface area of the AFI to the surface area of the water body.

Oxygen levels in a wastewater will be highly influenced by influent biochemical oxygen demand (BOD) of the wastewater, artificial aeration such as air diffusers and surface mechanical mixers, and ambient wind conditions, as well as temperature.

As the growth of both grasses were successful in the synthetic water, further trials are now being conducted using secondary treated paper mill wastewater under greenhouse and ASB conditions. Both grasses grow well under greenhouse conditions while buffalo grass grows extensively on AFI under field conditions in warm waters and in warm weather conditions.

Results from the experiments conducted showed that buffalo and couch grasses are able to grow well under the AFI test conditions applied, with high percentage growths for both the roots and shoots, over the test period. The growth of both grasses in the RO water suggests these plants acquired essential elements from airborne sources and have the ability to grow in very low nutrient conditions. Both buffalo and couch grasses were able to reduce the nutrients (measured as TP and TN) levels in a hydroponic setting, with percentage N and P uptakes of 82% and 45%, respectively. Uptake of TP and TN achieved by both buffalo and couch grasses on AFIs indicate these have potential capacity to limit cyanobacteria growth in ASBs when under the test conditions applied.

The authors thank Mr Tim Golding and the technical team from the University of South Australia for support in the experiments conducted. The authors sincerely thank Mr Graham Burch for his support in conducting this study.