Improvement of organic matter and nutrient removal from domestic wastewater by using intermittent hydraulic rates on earthworm–microorganism biofilters

Universal access to sanitation is one of the challenges established by the United Nations for 2030. However, global municipal wastewater production reaches 2,200 km³ year⁻¹ with the treatment coverage of developed and developing countries of up to 82% and 25%, respectively (Mateo-Sagasta et al. 2015; Singh et al. 2015).

Domestic wastewater (DW) is part of municipal wastewater and corresponds to the mixture between grey (cooking/washing, up to 80% v/v) and black (excreta/urine drainage, up to 20% v/v) water. The composition of DWs varies according to the consumption habits and socio-economic level of the population. Thus, urban wastewater is up to 50% more concentrated and generated faster than rural wastewater (Ghunmi et al. 2008). DWs are generated at variable rates, reaching up to 400 L inhab⁻¹ day⁻¹ (Villamar et al. 2018). Moreover, they are characterized by the presence of solids (total suspended solids: 100–350 mg SST L⁻¹), organic matter (total chemical oxygen demand: 250–1,600 mg COD L⁻¹, biochemical oxygen demand: 110–800 mg BOD₅ L⁻¹), nutrients (total nitrogen: 20–120 mg TN L⁻¹, total phosphorus: 2–23 mg TP L⁻¹), and micropollutants (pathogens and emergent pollutants) (Vera et al. 2013; Villamar et al. 2018). This condition restricts their discharge in water bodies and agricultural recycling (Bastian & Murray 2012).

A wide variety of treatment technologies has been successfully implemented around the world during recent centuries. For instance, conventional wastewater treatment plants (WWTPs) mainly based on activated sludge technologies are used at the urban level. Meanwhile, at the rural scale, non-conventional WWTPs are related to passive technologies (e.g. biofiltration) (Libralato et al. 2012). The main reason for this is related to the increasing cost (construction/operation) of conventional WWTPs in a decentralized context (<2,000 inhabitants) subjected to cost escalation (Singh et al. 2015). Specifically, biofiltration technologies based on microorganisms–organic support ‘biofilters’, plants–microorganisms ‘sub-superficial constructed wetlands’, and earthworms–microorganisms ‘vermifilters’ reproduce complex natural depuration processes (symbiotic biodegradation, precipitation, adsorption, among others), avoiding excess sludge generation and using up to 80% lower artificial energy resources (Lu et al. 2009; Shao et al. 2013). Biofilters based on earthworms–microorganisms (BEMs) are one of the most recently studied biofiltration technologies (since the 1990s), which offer an economical alternative for treating point and diffuse wastewater sources. They are structured by a biotic component (earthworms and microorganisms) and a non-biotic/biotic component (filtering material) that constitutes an engineered system based on symbiotic relationships. Microorganisms biochemically degrade waste materials, while earthworms regulate microbial biomass and activity.

BEMs represent an effective technology for wastewater treatment. They have been evaluated on DWs, agriculture and industry (Jiang et al. 2016; Samal et al. 2018a). BEMs treating DWs have reported high removal efficiencies of solids (>70% as SS), organic matter (>80% as BOD₅ and up to 90% as COD), and pathogens (>90% total coliforms) (Arora et al. 2014a; Lakshmi et al. 2014; Kumar et al. 2015; Jiang et al. 2016). However, some studies showed that nutrient (nitrogen and phosphorus) removal within BEMs did not exceed 60%, which could prevent its discharge into water bodies (Li et al. 2009; Zhao et al. 2012). The optimization of BEM operation is based on the ideal conditions for organisms (pH: 5–9, 15–28 °C), superficial organic loading (0.1–0.8 kg BOD₅/m² day and 0.1–0.6 kg COD/m² day) and hydraulic rate (0.3–3.0 m³/m² day) (Arora et al. 2014a; Kumar et al. 2014). An appropriate design and the use of adequate operational parameters could result in low cost, easy to maintain BEMs that are efficient for removing organic matter, nutrients and pathogens. Compound (organic matter and nutrient) removal within BEMs is carried out in the active and intermediate layers (first 10–50 cm). These layers are populated by saprophytes (earthworms) and heterotrophic/autotrophic microorganisms (associated bacteria) that cohabit in the organic/inorganic material (Arora et al. 2014a).

Organic matter is removed by aerobic/anaerobic processes. Some studies have reported removals of more than 90% of the organic matter in the surface layers (5–35 cm), which could be related to a more active aerobic population (heterotrophic bacteria and earthworms) (Wang et al. 2011).

Nitrogen transformation/removal has been reported associated with adsorption mechanisms (up to 60%), assimilation/ammonification (earthwork: up to 5%), and nitrification (Wang et al. 2011; Krishnasamy et al. 2013). Nitrification is reinforced by soil support material and oxygen environments, increasing it up to 87% (Wang et al. 2011). Denitrification within the BEM environments (mainly oxygen-bearing) cannot be favored. Phosphorus removal is mainly related to assimilation and adsorption processes (Krishnasamy et al. 2013). However, earthworm metabolic action generates phosphorus-rich excrement which is not removed and may affect the phosphorus found in the effluent (Hait & Tare 2011). In fact, some studies have evidenced the increase of phosphates in the effluent (Kumar et al. 2014).

Organic matter and nutrient removal may be influenced due to operational changes such as hydraulic rate. Indeed, Kumar et al. (2014) studied the effect of varying the hydraulic rates on the removal efficiency of organic matter, solids, and coliforms. Their results showed that hydraulic rates up to 2.5 m³ m⁻² day⁻¹ allow the maintaining of over 80% removal efficiencies for these parameters (Kumar et al. 2014). That study also showed that nitrate and phosphate will increase, under optimum hydraulic rates, reaching average values of 45 mg L⁻¹ and 35 mg L⁻¹, respectively. Studies have reported strategies such as intermittency of the hydraulic rate on BEMs, but they have not studied nutrient removal (Wang et al. 2010). However, the effect on nitrification–denitrification has been studied in constructed wetlands. Indeed, Villamar et al. (2015) have reported increasing nitrification in spring (water level below 0.1 m) and denitrification in fall (water level above 0.1 m), treating swine slurry within constructed wetlands. The removal of phosphorus under intermittent hydraulic rates has been not studied, but modified conventional technologies (sequential batch reactors or SBR) have been shown to carry out denitrification/accumulation of phosphorus after nitrification/phosphate liberation within the oxic phase (He et al. 2018). Thus, the removal of phosphorus and nitrogen could be carried out jointly, if hydraulic intermittent rates are generated. Therefore, the aim of this study is to evaluate the effect of hydraulic intermittent rates on the removal of nutrients in BEMs treating domestic wastewater.

In this article, we study the effect of applying intermittent hydraulic loading rates on the performance of earthworm–microorganism biofilters, and their removal of organic matter and nutrients from domestic wastewater. The effects on earthworm development, biofilm density, redox potential (related to biological activity), and head losses (and therefore clogging) are also examined.

The columns were fed with synthetic domestic wastewater (SDW), which was prepared following the methodology developed by Almeida-Naranjo et al. (2017). SDW was prepared in lots of 10 L day⁻¹ and it was used immediately, to avoid changes in the TCOD. Table 1 details the physicochemical composition of the SDW.

The experimental model consists of three columns (55 cm high, and a 28.3 cm² cross-sectional area) made of PET. Each column was divided in three layers: active (chip and compost, ø = 1–5 mm, 10 cm high), intermediate (sand, ø = 1–2 mm, 20 cm high), and support (gravel, ø = 10–25 mm, 25 cm high). The active layer was inoculated with earthworms (Eisenia fetida, 8,592 individuals m⁻²) in accordance with the literature (Xing et al. 2012; Arora et al. 2014a; Zhao et al. 2014). Figure 1 details the experimental model at the laboratory-scale, showing both (a) external and (b) internal characteristics.

The acclimatization period of BEMs was carried out in two stages. During the first stage, earthworms (220 individuals obtained from an organic worm farm located in Quito, Ecuador) and the BEMs substrate (chip, sand, and gravel without earthworms) were acclimated using real domestic wastewater (5 cm³ min⁻¹ every four hours and every two days) for six days. This real wastewater was collected at 1 min intervals during 45 min, from a sewer line of a classroom building located in Quito, Ecuador. During the second stage, earthworms (24 individuals) were acclimated to conditions similar to those of the BEMs (chip and compost, 5 cm high) and SDW (200 mL day⁻¹) for six days (Arora et al. 2014a). Temperature (15.4 ± 0.5 °C) under dark conditions and pH (7.5 ± 0.1) were maintained during this process. Each BEM was inoculated with three adult and 21 juvenile Eisenia fetida earthworms with a total weight between 6.4 and 7.2 g and an average weight of approximately 0.3 g/individual.

The operational strategy involved the selection of the hydraulic loading rate for which optimum results were observed by Kumar et al. (2015), when it was applied to similar vermifilters containing Eisenia fetida earthworms. A nominal hydraulic loading rate of 2.5 m³ m⁻² day⁻¹ was selected for this work. The biofilters operated in parallel for 70 days, during which they worked with an average flow of 4.9 cm³ min⁻¹ (7.1 L d⁻¹) corresponding to an average hydraulic loading of 2.48 m³ m⁻² day⁻¹ with an average standard deviation of 0.25 m³ m⁻² day⁻¹. The wastewater was applied to each biofilter (column) for 24 hours (B_{1–24 h}), 8 hours (B_{2–8 h}), and 4 hours (B_{3–4 h}), in a 24 h period. Therefore, an intermittent operational condition was used for the B_{2–8 h} and B_{3–4 h} biofilters. The optimal operational conditions such as humidity (40%–90%), temperature (15–20 °C), and pH (6.6–7.9) were monitored and controlled three times a day. Table 2 summarizes the operational conditions of the BEMs.

Influent and effluent characteristics were evaluated according to Standard Methods for the Examination of Water and Wastewater (APHA-AWWA-WEF 2005). The parameters considered were total/volatile solids (TS/VS; method: 2540B), color (method: 8025), chemical oxygen demand (COD, method: 5220), ammonium (NH₄⁺, method: 8038), nitrate (NO₃⁻, method: 8039), nitrite (NO₂⁻, method: 8507), and total phosphorus (TP, method: 8190). Sampling was carried out once a week (70 days) and the analyses were performed in duplicate. The BEM removal efficiencies were calculated as the arithmetic mean (for COD) and as the average (for NH₄⁺, NO₃⁻, NO₂⁻ and TP) of the efficiencies calculated using the influent and effluent concentrations determined for the samples weekly collected.

The biofilms attached to materials within the biofilters were evaluated at the beginning and at the end of the operation. Biofilm was extracted at a 2 cm depth (active layer), a 15 cm depth (intermediate layer), and a 35 cm depth (support layer). Samples (up to 200 g of material) were suspended in a buffer solution (7.47 g K₂HPO₄ L⁻¹, 1.43 g KH₂PO₄ L⁻¹), and sonicated (Branson 1800 ultrasonic) for 20 min (Caselles-Osorio et al. 2007). The quantification of the attached biofilm was performed by determining volatile solids according to Standard Methods for the Examination of Water and Wastewater (APHA-AWWA-WEF 2005). Visual characteristics were determined using scanning electron microscopy (Aspex PSEM eXpress microscope). Monitoring of redox potential (oxidation–reduction) was performed in the active and intermediate layers, using two stainless-steel electrodes (20 cm long, 1 cm wide, and 0.1 cm thick), with an Ag/AgCl electrode used as reference. Electric potential (mV) was measured using a DTR DT9205A voltmeter every 30 minutes for eight hours in the last stage of the operation (Szögi et al. 2004; DeLaune & Reddy 2005).

The dynamics of the earthworm population was measured at the beginning and at the end of the operation by tracking allometric parameters related to growth (number of individuals, live weight), maturity (clitellum presence) and reproduction (cocoon number) (Schuldt et al. 2005).

An analysis of variance (ANOVA) with p = .05 was carried out to compare the operational behavior data of the biofilters. The evaluated factors were B_{1–24 h}, B_{2–8 h}, and B_{3–4 h}, while the variables were VS, COD, color, NH₄⁺, NO₃⁻, NO₂⁻, TP, redox potential, ΔH, and K. Sixteen data points (two per week, for eight weeks) were collected per variable and for each BEM. Therefore, a total of 48 data points per variable was used during the analysis of variance. Before the ANOVA test, normality and variance homogeneity were tested using the Shapiro–Wilk and the Levene tests, respectively. Parametric (Duncan's) or non-parametric (Kruskal–Wallis) tests were used, depending on the compliance or not of the distributional assumptions (normality and homogeneity). The statistical software used was Infostat, version 2017.

Figure 2 summarizes effluent concentrations and organic matter removal (measured as total chemical oxygen demand, or TCOD) within the BEMs. As can be observed, the effluent arithmetic mean TCOD concentrations for the B_{1–24 h}, B_{2–8 h}, and B_{3–4 h} biofilters were 249.5 (±75.7) mg L⁻¹, 115.9 (±24.1) mg L⁻¹, and 172.5 (±35.7) mg L⁻¹, respectively. Other studies have reported similar COD effluent concentrations (up to 200 mg L⁻¹) at a continuous hydraulic rate of 1.0 m³ m⁻² day⁻¹ (Arora et al. 2014b). Therefore, intermittence did not affect the BEMs' performance regarding COD final concentration. However, the TCOD removal efficiencies for the B_{1–24 h}, B_{2–8 h}, and B_{3–4 h} biofilters reached arithmetic mean values of 35%, 74%, and 61%, respectively. This removal was obtained by the symbiotic relationship between earthworms and aerobic/anaerobic microorganisms (Kumar et al. 2015; Lourenço & Nunes 2017; Singh et al. 2019a). Annelids can transform insoluble organic matter into soluble, and they also add digestive enzymes present in the mucus of the earthworm intestine. These products serve as food for microorganisms and also favor oxygen diffusion (Singh et al. 2017). The results obtained show significant differences (p < .05) between continuous (B_{1–24 h}) and intermittent (B_{2–8 h}/B_{3–4 h}) operation. Thus, intermittence within the BEMs doubled the removal of TCOD, maintaining effluent quality below 200 mg L⁻¹. Moreover, intermittence could increase the hydraulic loading rate and dissolved oxygen within the BEMs. In fact, this work reached a similar removal of TCOD compared with other studies that used lower hydraulic loading rates. Arora et al. (2014b) showed similar efficiencies at a continuous hydraulic rate of 1.0 m³ m⁻² day⁻¹.

Effluent concentrations and nutrient removal, measured as (a) NH₄⁺, (b) NO₃⁻, (c) NO₂⁻, and (d) TP within the BEMs are shown in Figure 3. The nutrient average removal efficiencies for B_{1–24 h} were 9.6%, 48.8%, −19.7% and 9.6% for NH₄⁺, NO₃⁻, NO₂⁻ and TP, respectively. Meanwhile, B_{2–8 h} showed average removal efficiencies up to 57.3%, −15.5%, −260.3% and 20.1% for the same variables. Finally, B_{3–4 h} showed average removal efficiencies of 61.3%, −19.3%, −319.1% and 19.5% for NH₄⁺, NO₃⁻, NO₂⁻ and TP, respectively.

B_{1–24 h} removed significantly (p < .05) less NH₄⁺ than B_{2–8 h} and B_{3–4 h} (at least six times) and TP (at least two times). Moreover, B_{1–24 h} produced less NO₃⁻ (at least three times) and NO₂⁻ (at least 13 times) compared with B_{2–8 h} and B_{3–4 h}, with significant differences (p< .05) only for NO₂⁻. Other studies working with a constant hydraulic loading rate of 1.5 m³ m⁻²day⁻¹ have reported effluent concentrations of NH₄⁺ (11.8–22.5 mg L⁻¹), NO₃⁻ (19.1–31.2 mg L⁻¹) and TP (13.3–18.1 mg L⁻¹), whose nutrient removal efficiencies reached up to 74%, −19.8% and −2.5%, respectively (Kumar et al. 2015). Therefore, intermittence improved the nitrification up to 85%, favoring also the increase of the hydraulic loading rate without affecting effluent quality. Moreover, it would generate spaces between substrate pores that supplant air, promoting heterotrophic (aerobic) and autotrophic (nitrification) respiration. This phenomenon has been demonstrated in biofilters based on plants and microorganisms (constructed wetlands). In these constructed wetlands, nitrification–denitrification attributed to intermittent water-level due to seasonality has been reported (Samal et al. 2018b). Additionally, the burrowing carried out by earthworms increases oxygen concentration inside the BEMs, favoring the presence of aerobic nitrifying bacteria (Singh et al. 2019a). Nevertheless, the differences in the decrease and/or increase of the NH₄⁺/NO₃⁻ ratio within the effluents of B_{2–8 h} and B_{3–4 h} are minimal (0.75) with respect to B_{1–24 h} (0.79). Therefore, increasing the hydraulic loading rate decreases wastewater leaching through the BEMs, decreasing the nitrification–denitrification processes (Luth et al. 2011). BEMs operating with intermittent hydraulic loading rates reached approximately 66% more phosphorus removal than continuous BEMs, which has not been previously reported. Other authors have proposed that the low rate of phosphorus removal on BEMs is due to its rapid mineralization, produced by earthworm enzymatic activities combined with microorganisms' action (Kumar et al. 2014; Samal et al. 2018b).

According to Figure 4(a), the biofilm concentration per area showed a heterogeneous distribution between the BEMs and layers. The intermittent condition (B_{2–8 h,} B_{3–4 h}) increased the biomass density up to 28 times (3.5 − 13.5 g VS m⁻²) compared with the non-intermittent biofilter (up to 0.1 g VS m⁻²). In general, 50%, 99%, and 71% of the biofilm in B_{1–24 h}, B_{2–8 h}, and B_{3–4 h} was concentrated in the active–intermediate layers, respectively. The highest biofilm density (13.5 g VS m⁻²) was reached by B_{2–8 h}. It is possible that this phenomenon is related to the presence of earthworms. In fact, Figure 4(b) describes the earthworm density per layer in each BEM. The results showed that the non-intermittent condition (B_{1–24 h}) does not support earthworm survival, while intermittent operation (B_{2–8 h,} B_{3–4 h}) allowed it. There are several factors involved, such as: metabolic relationships of a symbiotic type, medium oxygenation by earthworm peristaltic movements, etc. (Arora et al. 2014a, 2014b). Operational factors, such as intermittence and moisture content also influence biofilm growth, washing biomass, and avoiding/promoting optimum growth conditions of earthworms, which can contribute to microbial growth (Kumar et al. 2014, 2015). Arora et al. (2014b) have found more microbial diversity in BEMs with earthworms than in BEMs without them. Our study highlights the presence of bacillary forms (Clostridium sp., Bacillus sp., Lactobacillus sp., Pseudomonas sp., among others), these bacillary forms being the ones that have been detected to a greater extent using scanning electron microscopy in B_{2–8 h} and B_{3–4 h}.

Figure 4(b) and 4(c) show the earthworm behavior regarding density measured as live weight m⁻² and individuals m⁻², respectively. Non-intermittence conditions (B_{1–24 h}) resulted in a total loss of biomass, where the average moisture content in the BEMs was higher than 80%, occurring about 90% of the operation time. In this regard, the moisture content range for earthworm growth is between 50% and 90%, with an average of around 70% (Zhao et al. 2012). Therefore, keeping a continuous hydraulic loading rate does not provide adequate conditions for survival, growth, development and reproduction of earthworms (Kumar et al. 2015).

Figure 4(b) and 4(c) also show that the earthworms' growth availability during long times is given only inside BEMs with adequate intermittence (B_{2–8 h}). In fact, B_{2–8 h} (from 2.3 to 2.6 kg live weight m⁻²) shows up to 12% higher biomass at the end of the operation, with 60% in the active layer. Moreover, under these operating conditions, adults and cocoons increased at values of 3,534/11,780 (233%) and 12,014/40,047 (100%) individuals m⁻²/individuals m⁻³, respectively. Only juveniles decreased by 29% (5,300 individuals m⁻² or 17,667 individuals m⁻³). The earthworms (adults, juvenile and cocoons) were mainly distributed (up to 60%) in the intermediate layer (12,368 individuals m⁻² or 41,227 individuals m⁻³). However, B_{3–4 h} shows up to 57% lower biomass with up to 71% in the intermediate layer. Adults and cocoons increased at values of 2,474/3,221 (133%) and 3,180/10,600 (100%) individuals m⁻²/individuals m⁻³, respectively. Juveniles decreased by 57% (3,180 individuals m⁻² or 10,600 individuals m⁻³) and were mainly found (up to 77%) in the intermediate layer (2,473 individuals m⁻² or 12,365 individuals m⁻³). It is possible that long periods without water supply do not generate sustainable humidity conditions. Consider for this that Eisenia fetida has a moisture tolerance that ranges between 40% and 90% (Arora et al. 2014a,, 2014b). B_{3–4 h} showed some periods of operation with humidity values of 10%. Therefore, moisture was the most influential parameter in the survival and growth of earthworms, which is directly related to the intermittence. In practice, it has been reported that BEMs require harvesting the excess earthworm population (up to 50%) once a year every three to four months, considering the survival times (approx. 90 days) of the annelids (Singh et al. 2017).

The biological activity within the active and intermediate layers was indirectly measured by daily monitoring of the RP (electrochemical measurements), Figure 5. Thus, intermittence (B_{2–8 h}, B_{3–4 h}) would favor more electrochemical activity than continuously operating biofilters (B_{1–24 h}). B_{1–24 h} reported values from −40 to −0.9 mV for both layers all the time, while B_{2–8 h} registered electrochemical activity differences between active (−50.5 to −6.5 mV) and intermediate (−22.6 to +9.6 mV) layers with negative values at least 75% of the time. On the other hand, B_{3–4 h} showed the highest electrochemical activity in both the active (−64.4 to +52.1 mV) and intermediate (−15.5 to +1.4 mV) layers. In the B_{3–4 h} active layer, positive charges were reported 82% of the time, while negative charges were dominant (94%) in the intermediate layer. Therefore, intermittence caused the filtering medium to be interspersed with electron acceptors (organic matter, nutrients) and electron donors (O₂), alternately. However, the redox potential ranges reported by this study may correspond to only facultative conditions as mentioned by DeLaune & Reddy (2005). In fact, nitrification via production of NO₂⁻ may have taken place.

The head losses and hydraulic conductivities in the biofilters studied are presented in Figure 6. The values obtained for the head losses in the B_{1–24 h} biofilter showed a significant variation during the period in which they were recorded. Having observed this, it was decided to calculate accumulated head loss values, which are the sum of the losses recorded during the last six days of operation of each biofilter (20 measurements were taken). The analyses performed showed a relationship between intermittence and the accumulated head losses calculated. The intermittent operation also resulted in smaller variability ranges for the head losses.

The variability of the head losses could be influenced by the earthworm presence. B_{1–24 h} showed a positive head loss at the beginning, which decreased to zero on the third day due to clogging. The variability range of the head losses for the B_{1–24 h} biofilter goes from −1.6 to +0.2 cm, and a negative accumulated head loss of −11.7 cm was reached. B_{2–8 h} started its operation with head loss values close to zero and showed a variability range going from −0.1 to +0.1 cm. In this case, the accumulated head losses reached a slightly negative value of −1.0 cm. Finally, B_{3–4 h} maintained a positive head loss value during the measurement period with a variability ranging between +0.03 and +0.2 cm. The accumulated head losses reached a value of 2.8 cm. Therefore, B_{1–24 h} (without earthworms) showed the most variable head loss and the highest accumulated loss value, which was also statistically significant (p < 0.05) compared with B_{2–8 h} and B_{3–4 h}. BEMs with earthworms (B_{2–8 h}, B_{3–4 h}) were only affected by intermittence. In this regard, other authors have mentioned that earthworms help with preventing and delaying clogging, because the annelids favor tunneling and burrowing (Singh et al. 2018,, 2019a , 2019b).

Hydraulic conductivity results also show a relationship with intermittence and earthworm growth. The hydraulic conductivity for B_{1–24 h}, B_{2–8 h}, and B_{3–4 h} show average values of −90, −433, and +835 mm h⁻¹, respectively. The ranges for B_{1–24 h}, B_{2–8 h}, and B_{3–4 h} were from −472 to +263, −1,083 to +649, and +288 to +1,299 mm h⁻¹, respectively. In fact, intermittence significantly (p < .05) influenced BEMs (B_{1–24 h}, B_{3–4 h}), when the earthworm population was less than 8,000 individuals m⁻² (inoculums). Therefore, hydraulic conductivity values for BEMs (B_{2–8 h}) with earthworm populations greater than 8,000 individuals m⁻² will be mainly influenced by the annelid population. It should also be mentioned that the international legislation on biofilter design establishes ranges of hydraulic conductivity between +36 and +360 mm h⁻¹ (ÖNORM B2506-1 2000). Only B_{3–4 h} would have values within this range.

The selection of adequate operating parameters for BEMs can help with improving their wastewater treatment performance. Intermittent hydraulic loadings (B_{2–8 h}, B_{3–4 h}) increased the total organic matter removal (up to 74%). This condition also improved nitrification up to 85% (up to 61% ammonium removal) and TP removal up to 66%; this is up to 20% higher when compared with the continuous hydraulic rate (B_{1–24 h}).

Intermittent operation also influenced BEM internal behavior. B_{2–8 h} reached the highest biofilm density (13.5 g VS m⁻²) and the best earthworm growth (2.6 g live weight m⁻², 1,060 adults m⁻², 2,827 cocoons m⁻²), which is mainly observed in the active layer (humidity between 32% and 97%).

The highest activity (higher amount of VS), measured by monitoring the redox potential, was observed in intermittent hydraulic loading BEMs. In fact, the intermittency (B_{2–8 h}, B_{3–4 h}) created changing facultative environments (−64.4 to +52.1 mV) in both active and intermediate layers, which can be related to the removal of total organic matter and nutrients.

Intermittency (B_{2–8 h}, B_{3–4 h}) also prevented clogging, resulting in the lowest head losses (up to −1.0 cm) and variability (−0.1 to 0.2 cm). It also favored the increase of hydraulic conductivity (up to 1,299 mm h⁻¹) in some specific conditions (B_{3–4 h}). This fact could be reviewed during further studies, considering that the available international legislation on biofilter design defines specific hydraulic conductivity ranges.

BEMs are a non-conventional technology more robust for rural domestic wastewater treatment, due to being easy to operate, and having high efficiencies and low construction, equipment and maintenance costs. Therefore, one can say that the application under real conditions of the operational strategy studied in this work could be relatively simple and not increase the cost of technology, which would not limit its overcrowding. Further studies should analyze alternatives for the optimization of wastewater treatment by selecting the appropriate design and operating parameters for BEMs, including combinations of filter media and earthworm load.

This work was supported by FONDECYT (Grant project 11190352) and Universidad de Santiago de Chile (Grant project 091918VA). Moreover, this research thanks the support of Escuela Politécnica Nacional (Grant project PIS-18-01).