Greenhouse Gas Emissions from a Full-Scale Vermifilter for Sewage Treatment: Effects of Seasonality and Sewage Parameters

Abstract

Biological processes may generate CO₂, CH_4, and N₂O. Few studies have evaluated the impact of vermifilters (VFs) on the generation of these gases. The objective of this study was to evaluate the GHG emissions of a full-scale VF used for sewage treatment, as well as the effects of seasonality and operational condition. The study monitored the influent and effluent of a VF in a rural area. Emissions fluxes were measured using the static chamber method in fall–winter and spring–summer. The results showed that in terms of annual per capita emissions (kgCO₂eq/cap·y), VFs generated less GHGs than conventional and non-conventional wastewater treatment plants (WWTPs), with CO₂, CH₄, and N₂O emissions ranging between 0.8 and 7.5 kg/cap·y, 0.1–0.5 kgCO₂eq/cap·y, and 5.7–9.5 kgCO₂eq/cap·y, respectively. Regarding the effects of seasonality, CO₂ increased by 139% in spring–summer compared to fall–winter, while N₂O increased by 139% in fall–winter compared to spring–summer. A positive correlation between influent COD concentrations and CO₂ emissions (r = 0.7) was observed, whereas the influent carbon/nitrogen ratio (C/N) and N₂O emissions (r = −0.6) presented a negative correlation. These results evidenced that seasonality and sewage characteristics influenced GHG emissions in a full-scale VF.

1. Introduction

Greenhouse gas (GHG) alters the temperature of the planet, affecting climate stability, society, and ecosystems [1,2]. Among the seven gases considered by the IPCC to be the main emissions, CO₂, CH₄, and N₂O are the most studied, since they represent 66%, 16%, and 7% of GHGs, respectively [3,4]. Regarding warming potential, CO₂ has a significant greenhouse effect on the planet; however, over a 100-year time horizon, CH₄ and N₂O have warming potentials 23 and 296 times greater than CO₂ [5]. GHGs from wastewater treatment plants (WWTPs) account for 1.5% of CO₂ and 5% of non-CO₂ GHG emissions and are expected to reach 42% of waste-related emissions by 2030 [6,7], meaning that the operation of these facilities has a significant effect on the climate [8]. Although these figures account for the impact of the wastewater treatment sector, most studies are based on estimates that consider conventional WWTPs such as activated sludge (AS) and anaerobic sludge (AN) facilities installed in urban areas, disregarding plants in rural areas [9].

In 2020, there was no safe treatment for 45% of the wastewater generated in the world, much of which is concentrated in rural areas with low-density, dispersed populations [10,11]. To avoid contamination in these low-population areas, decentralized technologies that treat wastewater at the source or near the source of generation are used [12]. Constructed wetlands (CWs), soil treatment units (STUs), and vermifilters (VFs) are nature-based decentralized technologies [13], with VFs being systems involving earthworms that, by interacting with microorganisms, change the physicochemical properties of wastewater [10]. Earthworms are used in vermifilters to accelerate processes that occur in soil filtration during wastewater treatment [14]. Ingestion, excavation, and excretion by earthworms reinforce this relationship and generate aerobic environments that promote the growth of microorganisms. Some studies have reported removal efficiencies of 83.4% for organic matter measured as chemical oxygen demand (COD), 54.1% for total nitrogen (TN), 96.6% for ammonium (NH₄⁺), and 52.2% for nitrate (NO₃⁻) [15,16,17]. However, few studies have evaluated the impact of these systems on GHG generation [18]. The role of earthworms in the generation of emissions is unclear. Some studies claim that earthworms could increase CO₂ and N₂O emissions from the soil by 33% and 42% [19]. While CO₂ and CH₄ are related to overall decomposition, N₂O occurs during a particular type of decomposition (denitrification) that requires anaerobic conditions and is also associated with intermediate steps of dissimilatory nitrate reduction [20]; however, it can be produced by ammonia-oxidizing archaea through nitrification [5]. In contrast, other studies have shown that earthworms inhibit methane generation due to the burrowing behavior of earthworms that increase oxygen and stimulate the methanotrophic community (i.e., −3.3 and −9.7 mg CH₄/day [21]). Regarding CO₂, they can reduce emissions by increasing carbon storage through decomposition of organic matter and carbon stabilization through the formation of microaggregates and macroaggregates [22]. Ingestion and subsequent peristalsis destroy the pre-existing microstructure, but during gut transit, clay minerals and organic materials are mixed and become encrusted with mucus to create a new nucleus for microaggregate formation [23,24]. However, stabilization depends on the organic load and earthworm density of the medium [22]. Zhao et al. [25] and Huang et al. [26] showed that CO₂, CH₄, and N₂O emissions in VFs depend on earthworm density and the C/N ratio of the influent. While decreasing C/N favors incomplete denitrification processes and generates N₂O, an increase in earthworm density can increase CO₂ and limit CH₄ generation to 3% of CO₂ [10]. Additionally, seasonal temperature variation is an important factor for the growth and metabolic activity of earthworms and microorganisms [27]. VFs operating during warm seasons (15–25 °C) have been reported to enhance the decomposition of organic matter by 15% due to increased microbial and earthworm activity [28]. In contrast, during cold seasons, low temperatures inhibit earthworm reproduction and induce hibernation, reducing overall metabolic performance [14]. These seasonal changes can lead to higher CO₂ emissions in warm seasons, driven by intensified metabolic activity and elevated respiration rates within the VF [27,29]. Some studies conducted in conventional WWTPs indicate that N₂O emissions increase the emissions factor by 1,8% in the cold season, possibly due to nitrification [30,31]. However, no studies have addressed these seasonal variations in vermifiltration systems (VFs). Therefore, GHG generation depends on operational conditions such as the earthworm density, temperature, C/N and the organic load of the influent (COD). Thus, the objective of this study is to evaluate the GHG emissions of a full-scale VF and determine the effects of seasonality and specific operational conditions of wastewater called sewage parameters (i.e., C/N and influent COD concentrations) on GHG emissions.

2. Materials and Methods

2.1. Design and Operating Conditions

This study was carried out in a rural WWTP at Copiulemu operated by the Municipality of Florida, Concepción, Biobío Region, Chile. The WWTP uses a sequential system that includes pretreatment for the removal of solids, followed by VF and ultraviolet radiation as a disinfection system. The VF is a vertical subsurface flow system. Wastewater enters a rotary filter that has a helical screw to sweep solids from an external container into the VF. Once treated, this wastewater is stored in an accumulator tank and transferred via two pumps to the upper part of the VF, from which it is distributed over the surface via 24 sprinklers. The VF of 276 m² (A) operates with a flow (Q), hydraulic loading rate (HLR), and organic loading rate (OLR) of 160 m³/d, 0.6 m³/m²·d, and 0.5 kgCOD/m²·d, respectively. The filter material is stratified, with a lower layer of stones and an upper layer of wood chips measuring 0.92 ± 0.12 m. The latter layer is the active layer, which contains earthworms of the species Eisenia foetida, along with heterotrophic and autotrophic microorganisms responsible for biological reactions [32,33]. The monitored earthworm density was between 1105 and 7221 earthworms/m³.

2.2. Monitoring Strategy

The VF was installed in 2020; however, this study involved monitoring during 2022 (Period I), 2023 (Period II), and 2024 (Period III), with a sampling frequency of two to three months during the fall–winter (April to August, N = 7) and spring–summer (October to March, N = 8) of the southern hemisphere. The wood chips were replaced at the end of Period II. Figure 1 shows a schematic of the VF and the emissions monitoring strategy. GHG samples (CO₂, CH₄, and N₂O) from between the active layer and the atmosphere were measured using the static chamber method, which involves the use of a cylinder with a lid that is 10 cm in diameter and 35 cm high to enclose a volume of atmosphere from the topsoil [34]. The cylinder was buried 5 cm deep to collect samples in vials using a 60 mL syringe attached to the chamber by a hose. The sampling frequency was every 15 min for 1 h. Solid samples of the active layer were analyzed at different times of use in Period II; and to determine the effect of wood chip replacement on the VF, samples were taken at different times in Period III (0, 2, and 8 months). For the analysis, 5 subsamples were taken at different points until 5 kg had been collected. Additionally, wastewater samples were taken from the influent (IN) at the pretreatment outlet and from the effluent (E) at the VF outlet. All samples were stored at 4 °C for less than 24 h before analysis.

Figure 1. Schematic representation of a VF and gas measurement.

2.3. Analysis of Gas Samples

The sample vials were collected only in 2024 in fall–winter (May and August) and spring–summer (December and March). A gas chromatographic analysis was then performed [35]. The equipment used was a gas chromatograph Shimadzu with a methanizer, a flame ionization detector, and an electron capture detector. The flux of CO₂, CH₄, and N₂O between the soil surface and the atmosphere inside the chamber was calculated using Equation (1) [36]:

$$f={\displaystyle \frac{∆CPV}{∆tRTA}}$$

(1)

where f represents the flux of CO₂, CH₄, and N₂O (mg/m²·h), ∆C is the change in gas mass (mg) inside the chamber due to the change in time (∆t), P is atmospheric pressure (atm) inside the chamber (assumed to be 1 atm), V is the volume of the chamber (L), R is the ideal gas constant (0.08205 atm·L/mol·°K), T is the temperature (°K) inside the chamber, and A is the area of the chamber (m²). Table 1 shows the emissions factors resulting from the static chamber measurements. These factors and the removal efficiency of COD and TN in the VF influent were considered to estimate the GHGs for Periods I, II, and III.

Table 1. Vermifilter emissions factors in spring–summer and fall–winter (CO₂, CH₄, N₂O).

Season	kgCO2/COD Removed	kgCH4/COD Removed	kgN2O/TN Removed
Fall–Winter	1.3·10−1	2.3·10−5	2.0·10−2
Spring–Summer	3.1·10−1	2.4·10−4	6.2·10−3

The emissions factor (EF) is defined as the mass (kg) of CO₂, CH₄, or N₂O produced per kilogram of COD or TN removed, and the corresponding equations are as follows (Equations (2)–(4)):

$$EF C{O}_2 ={\displaystyle \frac{MC{O}_2}{Q(COD inf- CODeff)}}$$

(2)

$$EF C{H}_4={\displaystyle \frac{MC{H}_4}{Q(COD inf-CODeff)}}$$

(3)

$$EF N_{2}O={\displaystyle \frac{N_{2}O}{Q(TN inf-TNeff)}}$$

(4)

In the formula, COD_inf and COD_eff are inlet and outlet COD (mg/L); TNinf and TNeff are inlet and outlet total nitrogen (mg/L); MCO₂, MCH₄, and MN₂O, respectively, are CO₂, CH₄, and N₂O release amounts (kg/d).

To compare GHG generation in conventional WWTP, studies reporting GHG emissions normalized on per capita emissions per year were selected (kgCO₂ eq/cap·y). In particular, Ranieri et al. [8] analyzed emissions from more than 180 WWTPs, while Ross et al. [37] quantified gas fluxes and normalized emissions on a per capita per year basis. Regarding non-conventional technologies, less information is available; therefore, publications from 2011 to 2024 that reported emissions normalized on a per capita per year basis (kgCO₂ eq/cap·y) or provided information on system design and measured GHG fluxes were also included. For converting emissions fluxes to per capita GHG emissions (PC_GHG), the following equation was used (Equation (5)):

$${PC}_{GHG} ={\displaystyle \frac{fA}{PE}}$$

(5)

In the formula, f represents the average flux of CO₂, CH₄, and N₂O (mg/m²·h), A is the surface area of the VF, and PE is the population served. In this study, the system serves a population of 1000 people.

2.4. Analysis of Liquid Samples

The liquid samples, IN and E, were filtered using a Whatman membrane with a pore size of 0.7 µm and analyzed based on the protocol described in the standard method [38]. The organic matter present in the samples was identified in the form of COD (colorimetric method, 5220-D), biological oxygen demand (BOD₅) (azide-modified Winkler method, 5210-B), and total organic carbon (TOC) (catalytic combustion oxidation and non-dispersive infrared detection method). The analyzed nutrients were measured in the form of NH₄⁺-N, NO₃⁻-N, NO₂⁻-N (spectrophotometer UV–Vis Shimadzu UV 1800, Kyoto, Japan), and total nitrogen (TN) (Spectroquant-Nova 60, kits Merck, Darmstadt, Germany).

2.5. Analysis of Solid Samples

Solid samples of the active layer were analyzed at different times of use. Organic carbon in the wood chip active layer was measured via calcination loss at 550 °C [39]. NH₄⁺-N and NO₃⁻-N were measured using the KCl 2 mol/L extraction method and colorimetry in a Skalar autoanalyzer [40]. Total nitrogen was measured using the Kjeldahl method [41].

2.6. Mass Balance Analysis

The mass balances used the average of the data for 2022 (Period I), 2023 (Period II), and 2024 (Period III). Furthermore, they were calculated in terms of TOC and TN using Equation (6) [42]:

$$Ci·Qi-\left(Storage\right)- Emission=Ce·Qe$$

(6)

where Ci is the influent concentration of COT or TN (kg/L), Qi is the influent flow (L/d), Storage is the biomass growth and retention, Emission is the the mass of carbon or nitrogen emitted in gaseous form (specifically as C-CO₂, C-CH₄, or N-N₂O (kg/d)), Ce is the effluent concentration of COT or TN (kg/L), and Qe is the effluent flow (L/d). Carbon and nitrogen concentrations were determined from liquid samples, as described in Section 2.4. Gaseous emissions were quantified using gas chromatography, as detailed in Section 2.3. To express emissions in terms of carbon or nitrogen content, the molecular weights of CO₂ (44 g/mol), CH₄ (16 g/mol), and N₂O (44 g/mol) were used, converting them to C-CO₂, C-CH₄, and N-N₂O, respectively, based on the molar mass of the element within each compound [42].

2.7. Statistical Analysis

To evaluate the effects of each period and seasonality on COD and TN storage, statistical analyses were carried out using R software 4.4.2 and RStudio version 2024.9.1.394, with a significance level of p = 0.05. The Shapiro–Wilk and Levene tests were performed to analyze normality and homogeneity of variance, respectively. Subsequently, an ANOVA test was performed, and this was followed by the Kruskal–Wallis test for data without a normal distribution [43]. Furthermore, a principal component analysis (PCA) was carried out with the first two principal components to evaluate the influence of different physicochemical parameters of the wastewater influent (i.e., C/N and influent, COD and NT concentrations) on the GHG emissions of the VF (CO₂, N₂O, CH₄). Given a set of variables, PCA can be used to construct new variables or principal components (PCs) that are linear combinations of the initial variables. The first and second PC is the combination that maximizes the explained variance in the data set (66.9%). Moreover, a correlation matrix was performed to detect correlations between physicochemical parameters and GHG emissions. In the correlation matrix, values range from −1 (perfect negative correlation) to +1 (perfect positive correlation); values close to 0 indicate no linear relationship [44].

3. Results and Discussion

3.1. CO₂, N₂O, and CH₄ Formation in WWTPs

Table 2 shows the mass flux of gas per capita. While conventional technologies generate emissions ranging from 67.9 to 122.6 kgCO₂eq/cap·y, non-conventional technologies generate emissions that range from 1.1 to 75.5 kgCO₂eq/cap·y. In WWTPs, the main source of gas emissions is biological transformations involving microorganisms [42]. Generally, in these systems, CO₂ is generated by oxidation in aerobic processes and CH₄ is generated by reduction in organic matter in anaerobic processes [17]. N₂O is generated by nitrification and denitrification of nitrogen compounds in wastewater [8].

Table 2. Comparison of per capita GHG emissions among different WWTPs (kgCO₂eq/cap·y).

Type	Name	Emissions (kgCO2 eq/cap·y)				Reference
		CO2	N2O	CH4	Total
Conventional	AO	14.6	3.3	64.2	82.1	[37]
	ANS	30.1	36.0	56.5	122.6	[8]
	AS	30.3	36.7	12.5	79.5	[8]
	A2O	18.0	2.0	48.2	67.9	[45]
Non-conventional	CW	1.4	5.5	13.3	20.2	[45]
	CW	26.0	3.7	45.9	75.5	[46] *
	STU	6.3	0.9	2.8	10.0	[9]
	SSTU	−2.8	0.6	2.7	3.3	[9]
	VF	0.8	9.5	0.5	10.8	[25] *
	VF	0.6	-	0.5	1.1	[26] *
	VF	7.5	5.7	0.1	13.1	This study

Note: AO: anoxic–oxic; ANS: anaerobic sludge; AS: aerobic sludge; A²O: anaerobic–anoxic–oxic; CW: constructed wetlands; STU: soil treatment unit; SSTU: secondary treatment-soil treatment unit; VF: vermifilter. * Converting with emissions fluxes, surface area, and estimated population.

CO₂ emissions in the VF were lower than in other WWTPs, with 0.6–7.3 kgCO₂/cap·y. In particular, STU emissions were −2.8 kgCO₂/cap·y. However, Knapper et al. [9] suggest that this is due to low organic matter removal efficiency in the biofilter. It has been reported that in some VFs with other influents (i.e., dairy and swine waste), CO₂ decreases as worm density increases [21,47]. Consequently, low CO₂ concentrations are linked to the role of earthworms in mineralization, carbon stabilization, and the generation of recalcitrant vermicast [22]. In the case of wetlands, although plants assimilate atmospheric CO₂ through photosynthesis and transfer carbon compounds into the soil, it returns to the atmosphere during respiration [5]. This explains the higher CO₂ concentrations, ranging between 1.4 and 26 kgCO₂/cap·y. For conventional WWTPs such as ANS, AS, A²O, and AO plants, emissions range between 14.6 and 30.3 kgCO₂/cap·y. With these technologies, organic matter is degraded through aerobic and anaerobic processes, which explains the increase in CO₂ levels [37]. In aerobic processes, however, heterotrophic microbial kinetics are higher in conventional plants [48], leading to increased emissions.

CH₄ emissions in the VF were lower than other WWTPs, with −0.04 and 0.5 kgCO₂eq/cap·y. CH₄ reduction is attributed to earthworm and microorganism activity that enhances VF aeration [49]. These systems generate aerobic environments, with ORP between 200 mV and 300 mV [13,50], which reduces CH₄ [25]. On the other hand, Luth et al. [21] reported emissions of −3.3 and −9.7 mg/d CH₄ with another type of influent. Negative concentrations suggest the presence of methanotrophic microorganisms [46]. Wetlands have higher CH₄ than VFs, with 13.3–45.9 kgCO₂eq/cap·y. Unlike in VFs [51], anaerobic microorganisms (i.e., hydrolytic, fermentative, and acetogenic) in wetlands are not inhibited and can degrade organic matter into simpler compounds (e.g., acetate, CO₂, and H₂). These compounds are subsequently transformed into CH₄ by methanogenic bacteria [52]. Conversely, when aerobic environments are favored, methanotrophic bacteria can reduce CH₄ levels by more than 50% [53]. Therefore, CH₄ generation depends on the quantity, composition, and activity of methanogenic and methanotrophic microorganisms impacted by the operational conditions, i.e., temperature, UV radiation, and organic load [54]. In conventional WWTPs, such as ANS, A²O, and AO plants, emissions range between 48.2 and 64.2 kgCO₂eq/cap·y. In this type of treatment, anaerobic digestion is part of the main processes and methane is generated as a by-product of methanogenesis [9]. For AS, methane emissions are much lower, at 12.5 kgCO₂eq/cap·y. In these systems, the O₂ is injected by pumps at a concentration of over 2 mg/L, which energetically favors aerobic degradation and generates a toxic effect on methanogenic microorganisms [41].

N₂O emissions in the VF were between 5.7 and 9.5 kgCO₂eq/cap·y. Such emissions are generated by nitrification and denitrification processes, depending on the operational parameters of the VF [13]. Zhao et al. [25] observed increases in denitrification and N₂O generation when the influent C/N ratio dropped from 10 to 2.5. In the present study, the influent C/N ratio was between 8 and 10. This ratio, and the aerobic environment generated by the worms, inhibits the generation of N₂O via a denitrification process [54]. In wetlands, N₂O is between 3.7 and 5.5 kgCO₂eq/cap·y. These systems have aerobic, anaerobic, and anoxic microbial sites; therefore, N₂O originates in nitrification and denitrification processes [46]. Even though there is no consensus as to what the primary source is, as with VFs, there are operational conditions that cause an increase in N₂O, i.e., increased nitrogen, low organic load, O₂, and pH [5]. In conventional WWTPs, such as ANS, AS, A²O, and AO plants, emissions are between 2 and 36.7 kgCO₂eq/cap·y. N₂O can be generated under aerobic or anoxic conditions. Under aerobic conditions, oxidation of ammonium and nitrite generates N₂O [37]. Under anaerobic conditions, only ammonium oxidation generates N₂O. In activated sludge, 90% of emissions are generated during sludge aeration and the remaining 10% are generated in sludge storage tanks [55]. Factors that increase N₂O generation are low pH levels, the presence of toxic compounds, or low oxygen concentrations [56].

3.2. Vermifilter Performance and Mass Balance

Figure 2 shows the mass balance of TOC. In the effluent, TOC is reduced to 8.3 ± 3.4 kg/d, representing 33% of the influent. Additionally, the system emits 4.9 ± 3 kg/d of C-CO₂ and 7.2·10⁻³ ± 7.4·10⁻³ kg/d of C-CH₄, which together account for 19% of the influent. C-CO₂ and C-CH₄ emissions are associated with the degradation of organic matter related to earthworm and aerobic microorganism activity [10]. Low C-CH₄ emissions (7.2·10⁻³ kg/d) indicate prevailing aerobic conditions and methanotrophic activity [5,25]. Therefore, the remaining 48% of TOC is likely retained as worm biomass, vermicast, microbial biomass, and carbon and absorbed into the wood chips of the active layer [42].

Figure 2. Mass balance in a full-scale vermifilter. Note: TOC: total organic carbon; TN: total nitrogen; N-N₂O: nitrogen from N₂O; C-CO₂: carbon from CO₂; C-CH₄: carbon from CH₄.

Table 3 shows the physicochemical characteristics of the active layer analyzed in Periods II and III. The results show that during the first 8 months of Period III with the new wood chips, there was a rapid decrease in TOC of 12% (i.e., 23.074 kg to 20.482 kg). TOC decreases in the active layer have been seen in other VFs [40,57] and are related to VF acclimatization and an increase in earthworm biomass [58]. Earthworms feed on organic matter, grow, and reproduce [59], changing the microstructure and compacting the active layer with an increase in microbial enzymes in the biofilms [60] and a loss of some compounds present in the wood chips, such as lignin, cellulose, and hemicellulose [61,62].

Table 3. Characterization of the active layer in periods of operation II and III.

Parameter	Unit	Period
		II-48 Months	III-0 Months	III-2 Months	III-8 Months
TOC	kg	17.138	23.074	22.364	20.482
TN	kg	835	84	217	534
NH4+-N	kg	0.8	0.5	13	8
NO3−-N	kg	41	0.1	12	12
N-org	kg	793	83	192	514
C/N	-	21.9	243.5	109.1	38.9

Note: TOC: total organic carbon; TN: total nitrogen; NH₄⁺-N: ammonium–nitrogen; NO₃⁻-N: nitrate–nitrogen; N-org: organic nitrogen; C/N: carbon/nitrogen ratio.

The mass balance of TN can be observed in Figure 2. In the effluent, TN is reduced to 8.9 ± 3.3 kg/d, accounting for 58% of the influent. In addition, the system emits 5.8·10⁻² ± 0.05 kg/d N₂O. N₂O emissions depend on the removal of the nitrogen that is related to nitrification and denitrification processes [59]. In the cited study, the denitrification process is inhibited by the aerobic environment of the VF [63]. However, the earthworm gut functions as a bioreactor that generates N₂O through incomplete denitrification when the C/N ratio decreases [62], as evidenced in Period III (i.e., after 48 months of active layer usage; see Table 3). N₂O emissions account for 0.4% of the TN in the influent. Thus, like TOC, the remaining 42% of TN is retained as worm biomass, worm vermicast, microbial biomass, and nitrogen and adsorbed into the wood chips of the active layer [42]. Table 3 shows that, unlike TOC, TN increases from 84 kg to 534 kg in the first 8 months of Period III. This is related to the adsorption of organic nitrogen, which increases from 83 kg to 514 kg, and the nitrification of NH₄⁺-N, which increases NO₃⁻-N from 0.1 kg to 12 kg. In Period II (48 months), the biofilm is well developed, and these processes intensify, with N-org reaching 793 kg and NO₃⁻-N reaching 41 kg.

3.3. Effects of Seasonality

Figure 3 shows a comparison of emissions compositions between seasons, revealing that CO₂ emissions were 4177 ± 1798 kg/y in fall–winter and 9992 ± 1714 kg/y in spring–summer. The significant increase (p = 0.002) of 139% in spring–summer is related to earthworm and microorganism activity. Earthworms are poikilothermic species; therefore, they are affected by temperature [14]. The optimal life range of earthworms is between 15 °C and 20 °C. In this study, the temperature range was between 11 ± 2 °C in fall–winter and 18 ± 4 °C in spring–summer. These differences affect the metabolism, growth, reproduction cycle, and activity of earthworms and microorganisms that can degrade organic matter [28]. The amount of CH₄ was 17 ± 7 kg/y in fall–winter and 175 ± 30 kg/y in spring–summer. A significant increase (p = 0.001) of 90% may be related to higher organic loading due to reduced rainfall [64]. When the system receives a high input of organic matter, substrate transformations depend on the abundance of earthworms [65], and the system may experience decreased aeration. However, CO₂ emissions are approximately 50 times higher than CH₄. Therefore, aerobic processes represent the main mechanism of organic matter degradation in VF [63]. Although the optimal range for aerobic degradation is 25–35 °C [66], it has been observed in environments with earthworms that exothermic processes of organic compound bioconversion can increase the temperature by 1–5 °C relative to the environment [67]. This is consistent with other publications that have recorded increases in heterotrophic activity responsible for organic matter degradation in spring–summer [28]. Furthermore, temperature can also affect air diffusion in the active layer, increasing the extent of biological reactions [66].

Figure 3. Effects of seasonality on GHG in a full-scale vermifilter. (a) CO₂ eq emissions due to CO₂ (■) and N₂O (■). (b) CO₂eq emissions due to CH₄. (■). Note: significant differences, p < 0.01.

Figure 3 also shows that N₂O emissions were 9555 ± 3763 kgCO₂eq/y in fall–winter and 3992 ± 1279 kgCO₂eq/y in spring–summer. The significant increase (p < 0.001) of 139% in fall–winter may be related to variations in the C/N ratio [13]. In fall–winter, the C/N ratio was 8 ± 3 and in spring–summer it was 10 ± 2. These changes may be related to variations in TN concentrations generated by rain events that sweep along nitrogen compounds [68].

3.4. Effects of Other Parameters in GHG Emissions

Figure 4a shows a PCA, in which 66.9% of the variance is explained by component 1 (PC1) and component 2 (PC2). Component 1 is represented by N₂O (0.43) and C/N (0.44) and component 2 by CO₂ (0.41), COD (0.44), and TN (0.51). Figure 4b shows the correlation matrix of the physicochemical parameters measured in the influent and GHG. In addition, Table 4 presents a mass balance for different active layer usage times.

Figure 4. Relationship between physicochemical parameters and greenhouse gas emissions: (a) principal component analysis and (b) correlation matrix. Note: TN: total nitrogen; NH₄⁺-N: ammonium nitrogen; NO₃⁻—N: nitrate—nitrogen; NO₂⁻—N: nitrite—nitrogen; C/N: carbon/nitrogen ratio; N₂O: nitrous oxide; CH₄: methane; CO₂: carbon dioxide. In the correlation matrix (Figure 4b), values range from -1 (perfect negative correlation) to +1 (perfect positive correlation); values close to 0 indicate no linear relationship. The color scale represents the strength and direction of the correlations, with blue tones indicating positive correlations and red tones indicating negative ones.

Table 4. Mass balance in periods of operation I, II, and III of the vermifilter.

Parameter	Unit	Period
		I	II	III *
Organic matter
Influent TOC	kg/d	29.2 ± 3.5	24.7 ± 4.1	22.6 ± 6.8
Effluent TOC	kg/d	9.7 ± 2.3	10.7 ± 2.7	4.6 ± 2.8
Storage TOC	kg/d	13.2 ± 3.6	10.0 ± 5.0	13.8 ± 5.8
	%	45%	40%	61%
C-CO2	kg/d	6.3	4.0	4.2
Nitrogen
Influent TN	kg/d	17.8 ± 5.1	11.6 ± 2.3	17.0 ± 3.4
Effluent TN	kg/d	9.3 ± 1.7	7.2 ± 2.1	10.6 ± 5.4
Storage TN	kg/d	8.4 ± 4.5	4.3± 1.1	6.3 ± 2.7
	%	47.4	37.2	37.1
N-N2O	kg/d	8.0·10−2	4.0·10−2	5.2·10−2

Note: TN: total nitrogen; TOC: total organic carbon; N-N₂O: nitrogen from N₂O; C-CO₂: carbon from CO₂; Period I: 2022; Period II: 2023; Period III: 2024. Storage: mass retained in the VF. * Replacement of wood chips.

For CO₂, the PCA showed a positive relationship with COD. The correlation factor between COD and CO₂ was 0.71 (see Figure 4b). Generally, in all systems, CO₂ is directly related to overall organic matter decomposition rates [19]. In wetlands, Gagnon et al. [69] found that high organic matter concentrations influence bacteria activity, generating an increase in CO₂. In a VF, the variations in organic loads favor earthworm activity and therefore, affect the respiration of aerobic microorganisms [43,70]. Thus, the organic matter concentration in sewage affects earthworm and microorganism activity and induces a CO₂ increase in a VF [13].

For N₂O, the PCA showed a positive relationship with TN and a negative one with C/N. The correlation factors of N₂O with TN and C/N were 0.51 and -0.57, respectively (see Figure 4b). Denitrification activity in earthworms can be 1000 times more than in the soil. This produces N₂ together with N₂O [24]. However, when the C/N ratio decreases, incomplete denitrification in earthworms increases, along with the release of N₂O. Meanwhile, when the C/N ratio increases, complete denitrification increases, along with the release of N₂ [71]. Generally, denitrification is carried out by heterotrophic bacteria that grow when organic carbon is abundant in the system [20]. The mucus secreted by earthworms has gut microbes, including heterotrophs, mixed with the active layer [10]. Zhao et al. [25] found similar results when increasing the influent TN and reducing the C/N ratio. When the C/N ratio was 10, N₂O emissions were between 2.03 and 2.41 mg/m²·h. However, when the ratio was 2.5, N₂O emissions increased to 12.24–12.34 mg/m²·h.

Table 4 shows a slightly significant increase in TOC storage (p = 0.06), from 40% in Period II to 61% in Period III. This is due to the replacement of the wood chips of the active layer. The C-CO₂ increase from 4.0 to 4.2 kg/d was not significant (p > 0.05). Therefore, in the first year with new wood chips, the main processes may be related to wood chip adsorption and an increase in worm biomass [32,57]. TN storage is higher in Period I at 47.4%. However, the difference is not substantial relative to Period II and III. It is possible that the rainy seasons and changes in the C/N ratio could generate significant changes within the periods, thus explaining the lack of significance of the results [13].

Overall, the results highlight the relevance of VF systems as low emissions alternatives for wastewater treatment in rural areas. The observed relation between operation conditions (i.e., earthworm density, temperature, C/N and the organic load of the influent, COD) and GHG emphasizes the need for optimizing design and operation to minimize CO₂ and N₂O release while maintaining high pollutant removal efficiencies. These findings demonstrate that the incorporation of VF into existing wastewater treatment infrastructure could reduce the carbon footprint of the sanitation sector and provide a sustainable upgrading pathway for decentralized systems.

4. Conclusions

Considering the results of this study, the following can be concluded:

-

VFs generate lower GHGs than other WWTPs: CO₂ 7.5 kgCO₂eq/cap·y, CH₄ 0.1 kgCO₂eq/cap·y, and N₂O 5.7 kgCO₂eq/cap·y;

-

According to the mass balance, 19% of the influent TOC and 0.4% of the influent TN were converted into CO₂ and N₂O emissions, respectively, in the full-scale VF system;

-

Seasonality affects GHG emissions of the VF. CO₂ emissions were 4052 ± 1722 kg/y in fall–winter and 9630 ± 1640 kg/y in spring–summer. N₂O emissions were 11,423 ± 6330 kgCO₂eq/y in fall–winter and 3992 ± 1279 kgCO₂eq/y in spring–summer.

-

Influent COD concentrations and the C/N ratio were factors that determined GHG production. COD concentrations were positively correlated with emissions, with higher COD concentrations in the influent generating higher CO₂ emissions. In contrast, the C/N ratio was negatively correlated with N₂O emissions, with higher C/N ratios resulting in lower emissions of GHGs.

-

This study is the first attempt to establish GHG emissions from a full-scale VF, providing a basis for future research in this area.