According to the United Nations 2019 World Population Prospects report, the world population will reach 8.5 billion in 2030 and 10.9 billion in 2100. Therefore, the demand for food and energy, and the amount of generated organic wastes (OWs) will continue to increase (Hoornweg et al., 2013). Currently, the burning of fossil fuels to produce energy generates about two-thirds of global greenhouse gas emissions (Energy and climate change — European Environment Agency, 2017). Moreover, the natural decomposition of OWs releases greenhouse gases directly into the atmosphere, which contributes to global warming. Thus, a large-scale application of ecological methods to treat OWs and produce energy without generating greenhouse gases is needed. Anaerobic digestion (AD) of OWs, which produces renewable energy while reducing OW volume and methane emissions, is well-suited for this purpose (Energy and climate change — European Environment Agency, 2017). The resulting methane (CH₄) can be converted into heat and electricity and the other main output (called digestate) can be used to amend agricultural land (Fagerström et al., 2018).

The AD process is driven by a diverse community of microbes (mainly bacteria and archaea). In the absence of oxygen, AD comprises a series of complex microbiological processes, including hydrolysis, acidogenesis, acetogenesis, and methanogenesis (Manyi-Loh et al., 2013; Abdelgadir et al., 2014). Many parameters, such as temperature, pH, volatile fatty acid (VFA) concentrations and ammonia concentrations, affect the ability of microorganisms to convert biomass to methane and impact the bacterial composition of the end products (Al Seadi et al., 2008). The bacterial community structure in digestates is also influenced by the origin of the raw OWs used in the digestion process. Aigle et al. (2021) showed that two different bacterial structures are present in raw OWs depending on their origin. In urban OWs (mainly activated sludge), Proteobacteria and Bacteroidetes were the most abundant phyla. In contrast, agricultural waste (mainly manure), contained Firmicutes, Proteobacteria, and Bacteroidetes. After AD, taxa from activated sludge increased among urban digestates. In contrast, taxa from raw agricultural OWs decreased in agricultural digestates (Aigle et al., 2021). This reshuffling of the bacterial community indicates that the origin of raw OWs plays an important role in the final bacterial content of digestates. However, the fecal matters frequently used as feedstock for AD (activated sludge and manure) are often contaminated with pathogenic bacteria (Hutchison et al., 2004; Sidhu and Toze, 2009). Therefore, it is important to understand the impact of the AD process on the fate of pathogenic bacteria in different OWs.

To address this question, many parameters, such as temperature, pH, and hydraulic retention time, have been investigated (Avadí et al., 2022). Temperature is crucial in AD processes because it affects the richness and diversity of the microbial community in digestates (Gou et al., 2014). Thermophilic digestion (55°C) is assumed to be more efficient than mesophilic digestion (35°C) in reducing pathogens such as Escherichia coli and Salmonella spp. (Smith et al., 2005). Also, with a high residence time, batch reactors ensure better sanitation than continuously fed processes (Jiang et al., 2018, 2020).

Beside pathogenic bacteria, other indicators such as helminth eggs can be used to evaluate the sanitary quality of digestates. Soil transmitted helminth infections are particularly widespread in tropical and sub-tropical countries where lack of drinking water and inadequate sanitary conditions favor their transmission (Riaz et al., 2020). Pathways of contamination may involve contaminated water, soil amended with contaminated digestate, and food grown on that soil (Nag et al., 2019). While full-scale thermophilic digestion seems to be sufficient to inactivate helminth eggs (Seruga et al., 2020), mesophilic digestion can be insufficient (Harroff et al., 2019). In some regions (e.g., Latin America), AD is often operated under mesophilic or psychrophilic conditions using, e.g., low-cost plastic tubular digesters (Garfí et al., 2016; Tavera-Ruiz et al., 2023). In these conditions, regardless of the feedstock (cow manure alone or in combination with cheese whey, pig manure, food waste), the AD process was insufficient to eliminate helminth eggs, and viable eggs were found in digestates at concentrations ranging from 5 to 50 per 4 g (Cucina et al., 2021; Parra-Orobio et al., 2021). However, Parra-Orobio et al. (2021) highlighted the effect of the anaerobic reactor configuration (single vs. two-stage configurations) on the parasitological properties of digestates. Interestingly, the best quality digestates were obtained using the anaerobic two-stage configuration. These studies highlight the importance of considering the configuration of anaerobic reactors but also the origin of organic waste and the region of their production in the choice of indicators that could be used to evaluate the health quality of digestates.

However, previous studies have shown conflicting results regarding the fate of pathogenic bacteria, even in similar digestion conditions. For example, Tápparo et al. (2020) showed that, during batch lab-scale mesophilic digestion using agricultural waste as feedstock, E. coli enumerated at 5–6 logs CFU/mL in raw waste and Salmonella at more than 6 logs CFU/mL. Both bacteria were removed in digestates within 8–10 days. However, in similar conditions, Alfa et al. (2014) found that 30 days were needed to reduce coliforms by 1 log and were insufficient to completely remove Salmonella from digestates. While E. coli and Salmonella are often used to estimate the hygienization effect of AD, the fate of other pathogens usually detected in raw OW has rarely been investigated in digestates. These pathogens include Mycobacterium tuberculosis complex (Santos et al., 2015), Yersinia enterocolitica, Yersinia pseudotuberculosis (Falcão et al., 2004), and Acinetobacter baumannii (Seruga Music et al., 2017). In addition to their pathogenicity, these bacteria may carry antibiotic resistance genes (ARGs), conferring resistance to antibiotics commonly used to treat food-production animals, such as tetracycline and aminoglycosides (van Hoek et al., 2011). Microbial communities mainly acquire ARGs via horizontal transfer of mobile genetic elements (MGEs), such as plasmids, transposons, and integrons (von Wintersdorff et al., 2016). Several studies have reported that AD processes have a variable effect on MGEs and ARGs, depending on the feedstock and digestion conditions. For example, Zou et al. (2020) showed that the abundance of class 1 integrons (intI1) decreased by 0.9 log copies/g dry manure after 60 days of mesophilic digestion in batch lab-scale reactors. In contrast, in a feedstock composed of food waste and sewage sludge, intI1 increased from 10⁻⁴ to 10⁻³ gene copies/16S rRNA copies after 33 days of mesophilic digestion (Zhang et al., 2016). Similarly, the total abundance of seven tet genes significantly increased after mesophilic AD of tetracycline-spiked livestock manure (Agga et al., 2020). However, the abundance of the resistome and MGEs was reduced in municipal sludge after thermophilic AD, even under high oxytetracycline concentrations (Tian et al., 2019).

Therefore, the fate and distribution of different pathogens, MGEs, and ARGs remain unclear, and more research is required to assess the biosafety of digestates. In this study, we assessed 77 sanitary indicators, including pathogenic bacterial species and ARGs, as well as new indicators of health hazards, such as virulence factors carried by MGEs. We first validated detection by real-time PCR and quantification by droplet digital PCR (ddPCR) for 40 and 31 sanitary indicators, respectively, in raw OWs and digestates. We then investigated the sanitary performance of AD according to the type of process (feeding mode, number of stages, temperature, pH) and the nature of the raw OWs (fecal matter with or without supplementation with straw and/or zeolite).

In France, raw OWs from different origins (agricultural or urban) can be mixed in the same anaerobic reactor (Avadí et al., 2022). Depending on their origin, these OWs may be contaminated by metallic and/or organic micropollutants and by pathogenic organisms (Aigle et al., 2021). In the anaerobic environment inside a reactor, opportunistic pathogenic bacteria can be enriched, especially if they have the functional traits necessary for their survival and/or growth. The main goal of this study was to evaluate the impact of the nature of raw OWs (using three laboratory-scale reactors Lab1, Lab2, and Lab3) and different AD processes (using three full-scale reactors Ter1, Ter2, and Ter3) on the levels of sanitary indicators in digestates. First, we selected a high number of health indicators (77) that are most often found in raw OWs including new indicators of health hazards, such as virulence factors carried by MGEs. At least, one gene from the most frequently detected ARG classes in livestock waste (sulfonamides, tetracyclines, β-lactams, macrolide-lincosamid-streptogramin B and fluoroquinolone) was selected (He et al., 2020; Zalewska et al., 2021; Tian et al., 2022). Similarly, the ARGs found frequently in activated sludge (e.g., sul1, sul2, tetA, tetO, tetC, qnrA, vanA, qacE) were also selected (Redhead et al., 2020; Shin et al., 2020; Mutuku et al., 2022).

The specific detection of 40 indicators including 13 pathogenic species/genera, four virulence factors, eight MGEs, and 15 genes implicated in antimicrobial resistance were validated in raw OWs, and their digestates, from agricultural and urban origin. We then assessed whether the origin of raw OWs affects the levels of sanitary indicators. We classified the prevalence of 40 indicators in raw OWs of different origin (agricultural vs urban) into three groups (high, medium, and low prevalence). From these results, three different profiles were identified.

Some sanitary indicators were highly prevalent in all raw OWs tested in this study, despite their diverse origin, including 23% (3/13) of the pathogenic species/genera (E. faecalis, E. faecium, and B. cereus group), 86.7% (13/15) of the antimicrobial resistance genes, and 62.5% (5/8) of the MGEs analyzed. These results are consistent with earlier studies reporting that both livestock manure and sludge can be an important reservoir of ARGs (Li et al., 2015; Karkman et al., 2018). Interestingly, the high concentrations of pharmaceutical products detected in our study could explain the high prevalence of ARGs which is likely associated with selection phenomena (Hughes and Andersson, 2012; Bengtsson-Palme et al., 2018) and horizontal gene transfers (Jutkina et al., 2018). In the mixture of activated sludge and cow manure (97:3) used to feed the Lab1 (100%), Lab2 (85%), and Lab3 (81%) reactors, three pharmaceutical classes were detected at high concentrations (antibiotics>antimicrobials>anti-inflammatory drugs). Fluoroquinolones were present at the highest concentrations, reaching values of up to 3.5 mg/kg for ofloxacin and 2.6 mg/kg for ciprofloxacin. Mailler et al. (2017) found similar concentrations of fluoroquinolones in activated sludge in France (up to 6.7 mg/kg for ofloxacin and 12.8 mg/kg for ciprofloxacin). In our study, antimicrobial products were also detected with a wide concentration range, from 160 μg/kg for triclocarban to 2.1 mg/kg for triclosan. This is lower than the concentration of triclocarban in activated sludge reported in a previous study (mean of 13 mg/kg) (Mejías et al., 2021).

Other sanitary indicators were only present in either agricultural or urban waste (Figure 5). While Bacteroides were not detected in the raw agricultural samples used in the three full-scale reactors, both Y. enterocolitica and enterotoxigenic E. coli (ETEC) were found only in raw agricultural wastes and were not detected in urban wastes. The primers used for Bacteroides are specific to the human HF8 cluster (Bernhard and Field, 2000), therefore, it is logical that this species complex was not found in agricultural waste. The bla_CTX-M gene was detected in urban and was not found in waste of agricultural origin. Moreover, IncQ and IncN prevalence were also associated with the OW origin because IncN plasmids were detected in urban wastes and IncQ in agricultural wastes. This is consistent with a previous study in which IncN plasmids containing a variety of antibiotic resistance (mainly streptomycin resistance) and transposition genes were isolated from a final effluent from a municipal WWTP (Eikmeyer et al., 2012). However, other authors detected IncN plasmids in animal-associated environments (Johnson et al., 2007; Moodley and Guardabassi, 2009), and therefore, we cannot exclude their potential presence in agricultural waste.

A third profile of sanitary indicators was found, for which it was not possible to identify an association between functional traits and species occurrence according to OW origin. The cpe and ces genes encoding toxins produced by C. perfringens and B. cereus, respectively, were not detected in agricultural OWs even though 44 and 78% of samples were positive for C. perfringens and B. cereus, respectively. However, in urban wastes, cpe was detected in 100% of tested samples (100% prevalence of C. perfringens) and ces was detected in 33% of samples (78% prevalence of B. cereus). Our results are consistent with earlier studies, which reported a very low prevalence of cpe-positive C. perfringens and ces-positive B. cereus in OWs of agricultural origin (Altayar and Sutherland, 2006; Cui et al., 2016; Derongs et al., 2020). In contrast, many other studies suggested that humans are a potential reservoir for cpe-positive C. perfringens, which can, therefore, be detected in urban sludge (Heikinheimo et al., 2006; Carman et al., 2008). These data, together with our results, suggest that WWTP sludge could be a selective environment for growth of C. perfringens and B. cereus carrying the cpe and ces genes and underline the importance of exploring the associations between bacterial functional traits and OW origins for a more precise assessment of the health hazards associated with OWs.

After AD treatment, the fate of sanitary indicators was also evaluated. Of the 40 sanitary indicators, only one (S. agalactiae) was completely removed in all digestate samples, regardless of the AD process or the nature of the raw waste. S. agalactiae has a better ability to form biofilms at acidic pH, whether in vitro or in the host vaginal tract (Ho et al., 2013; D’Urzo et al., 2014). The reactors sampled in this study operated at neutral to basic pH, which could explain the failure of S. agalactiae to persist during the AD process. Interestingly, a significant decrease was also observed in the prevalence of two other indicators (bla_CTX-M and vanA) but only in the digestate from urban origin.

As revealed by real-time PCR analysis, the prevalence of many sanitary indicators monitored in our study was similar in OWs before and after AD treatment. Therefore, to better assess the sanitary quality of digestates, we performed an absolute quantitation of 31 indicators, as well as the total bacterial abundance by ddPCR. Bacterial abundance was similar in the raw OWs from agricultural and urban origin and not impacted by any of the AD treatment processes (Ter1, Ter2, and Ter3) or by the nature of the raw wastes tested in this study (Lab1, Lab2, and Lab3). We then investigated whether the nature of the raw OWs used to feed the Lab1, Lab2, and Lab3 reactors affected the abundance of 27 sanitary indicators in the digestates. AD led to a moderate increase in tetA in Lab3, but no other enrichment was observed in Lab2 and Lab3. In contrast, seven sanitary indicators (E. faecalis, E. faecium, M. tuberculosis complex, intI1, intI2, IS26, and bla_TEM) were significantly decreased in the digestates from the Lab2 and Lab3 reactors, where 15% wheat straw was added to the OW inputs. This was in direct contrast to Lab1, fed with urban waste without wheat straw, where intI2 increased in digestates. These results suggest that adding wheat straw may allow a better sanitizing effect during AD of urban wastes (Figure 5). The hydraulic retention time (HRT) was extended by 20 days in Lab2 and Lab3, compared with Lab1. This is due to the lignocellulosic properties of the added wheat straws, leading to slower processing by bacterial communities within the reactors. This longer retention time may have contributed to the decrease in certain sanitary indicators observed in the Lab2 and Lab3 reactors, such as different ARGs as suggested by (Haffiez et al., 2022). Also, Song et al. (2017) showed that mesophilic reactors using 30% straw in addition to pig manure allowed the reduction of ARGs by 4.23 log (Song et al., 2017). Furthermore, cellulose-based materials are often used to adsorb residues of antibiotics, heavy metals, and organic pollutants in wastewater treatment (Hokkanen et al., 2014; Jamshaid et al., 2017; Lu et al., 2021). A similar effect was observed in our study. We found that 11 PHPs and 11 PAHs were more concentrated in the reactor without wheat straw (Lab1), compared with the reactors (Lab2 and Lab3) supplemented with wheat straw. Many studies have reported the effects of various parameters (e.g., solids content, surface area, pH, temperature, and physico-chemical properties of the different micropollutants) on the biosorption processes within a solid matrix (Rathod et al., 2015; Semblante et al., 2015; Madariaga-Segovia et al., 2023). In our conditions (pH = 7.3, temperature = 37.5°C), the physicochemical characteristics of micropollutants can mainly influence their partition between the aqueous and particulate phase. For example, the high hydrophobicity of many compounds such as triclosan have an impact on their interactions with a solid matrix. However, other PHPs such as tetracyclines and fluoroquinolones have polar functional groups and can highly adsorb onto organic matter via electrostatic interactions regardless of their hydrophobic nature. As sub-inhibitory concentrations of antibiotics and heavy metals are known to promote horizontal gene transfer (Jutkina et al., 2018; Zhang Y. et al., 2018), the adsorption of these molecules during the AD process could reduce their bioavailability. This may lower the selection pressure linked to their concentration and, therefore, limit the dissemination of resistance associated genes. Thus, the lack of improvement in sanitary quality and the significant increase in intI2 in Lab1 may be associated with the high concentration of organic micropollutants present in this reactor. For example, the triclosan concentration in Lab1 was 2.1 mg/kg of dry sample, compared with 0 mg/kg and 0.5 mg/kg in Lab2 and Lab3, respectively. Triclosan concentrations between 0.02 μg/L and 2 mg/L (e.g., concentrations found in urban sludge or wastewater) can promote the dissemination of certain ARGs, such as tetA, tetR, and aphA, via horizontal transfer of the plasmid carrying these genes (Lu et al., 2018). A high triclosan concentration can lead to increased levels of reactive oxygen species (ROS), favoring an increase in the permeability of cell membranes and promoting horizontal transfer. In a more recent study, Lu et al. (2022) showed that exposure of the aquaculture pathogen Edwardsiella piscicida to triclosan concentrations between 2 and 20 μg/L promoted conjugative transfer among donor and E. coli recipient strains (Lu et al., 2022).

Some sanitary indicators, including an MGE (IncP) and three ARGs (qacH/E, oqxA, and sul1) were significantly decreased after AD only if wheat straw and zeolite powder were added (Lab3). This is consistent with a previous study showing that zeolite enhanced ARG reduction during mesophilic AD of swine manure, possibly due to passivation of heavy metals by zeolite, which, unlike antibiotics, cannot be degraded during AD (Zhang et al., 2012; Zhang J. et al., 2018). As animal manures are a source of heavy metals, this passivation could lower the long-term co-selection pressure exerted by heavy metals on ARGs during AD (Zhang et al., 2012; Zhang J. et al., 2018). Another study investigated the impact of zeolite on ARGs in aquaculture wastewater. Yuan et al. (2022) showed that zeolite had a significant impact on the absolute abundance of sulfonamide resistance genes (sul1 and sul2). Zeolite removed up to 90% of sulfonamide ARGs, which the authors suggested was due to altered microbial community structure and abundance (Yuan et al., 2022). Many other studies have reported that 16S rRNA and ARG copy numbers are positively correlated (Yi et al., 2017; Song et al., 2018). In our study, the effect of AD on ARG copy numbers could not be correlated to a decrease in total bacterial abundance. We saw no decrease in total bacterial abundance in digestates, regardless of the type of AD process or the nature of the raw wastes.

We also investigated the impact of different AD processes on the abundance of sanitary indicators. None of the 27 indicators monitored in OWs of agricultural origin was significantly reduced by AD in either of the semi-continuous feeding reactors: Ter1 (two-stage, 44.5°C) and Ter2 (single-stage, 39°C). The batch feeding reactor, Ter3 (45°C), had a better hygienization effect, with complete removal of ybtA and a significant reduction in intI2 and tetO. ybt is part of the genetic island called the High Pathogenicity Island (HPI) and is involved in regulation of siderophore-mediated iron acquisition (Yersiniabactin) (Carniel, 2001). Virulence HPI genes are essential for expression of the virulence phenotype (Carniel et al., 1987, 1989). Two hypotheses may explain the reduction of the ybtA gene. First, in a basic reactor such as Ter3 (pH = 9), the microorganisms will undergo very high selection pressure and only those which possess the most effective siderophores with a very strong affinity to chelate traces of ferric iron will be able to survive. Second, in an environment characterized by the probable presence of certain antibiotic residues (livestock waste), the production of siderophores is unfavorable for the survival of bacteria. The first hypothesis is based on the fact that bacteria undergo ferric stress in a basic medium (pH = 9) where ferric iron is less soluble mainly in the form of ferric hydroxide (Fe(OH)₃) (Boukhalfa and Crumbliss, 2002; Hider and Kong, 2010; Caza and Kronstad, 2013). Therefore, only bacteria producing siderophores with a very strong affinity that can chelate traces of ferric iron Fe³⁺ can survive. The second hypothesis is based on the work of Braun (1999), Mislin and Schalk (2014), and Pramanik and Braun (2006), which showed that iron acquisition pathways using siderophores allow the internalization of certain antibiotic molecules, becoming a disadvantage for bacteria (Braun, 1999; Pramanik and Braun, 2006; Mislin and Schalk, 2014). However, despite the hygienization effect observed in this reactor, a significant enrichment of 3.5 log10 copies/g was observed for the B. cereus group. The presence of B. cereus is desirable in agricultural soil due to their ability to fix nitrogen (Iniguez et al., 2004; Yousuf et al., 2017) but their enrichment in digestates, which are usually spread on agricultural land, could pose a health problem through the food chain.

In conclusion, our data suggest that the mesophilic AD process may not fully eliminate major pathogenic bacteria and associated virulence and resistance genes, and that pollutants may interfere with hygienization. Furthermore, adding co-substrate, such as wheat straw may overcome the detrimental effect of pollutants (Figure 5). To our knowledge, this study represents the most complete survey of sanitary indicators of public health concern in digestates derived from various raw OWs and anaerobic digestion processes in France. Based on the One Health concept as defined by the World Health Organization (WHO), we chose indicators that address health risks at the animal-human-ecosystems interface, including: (1) chemical products (or residues) linked to human activity, including veterinary care, (2) well-known bacterial pathogens including, but not limited to, zoonotic agents, and (3) indirect health indicators such as ARG and MGE, which are considered gold standard indicators for studies of global health concern. These latter were selected based on the most frequently detected ARG classes in organic wastes of agricultural and/or urban origin (He et al., 2020; Mutuku et al., 2022). Nevertheless, our study has some limitations: (1) some ARGs (sugE, pepA, qnrB, qacA, qnrA, fabK, tetM and vanB), initially selected to be monitored in our study, could not be specifically amplified in the OWs samples from agricultural and urban origin, and (2) another limitation concerns four MGEs (intI3, IScr1, IS100 and IncW trwAB), which could not be amplified specifically in our study. The specific amplification of each target gene was investigated by real-time PCR using primers previously published. To select the primers, we considered the number of samples used to validate these primers and the origin of samples, preferring to have organic waste as matrices if possible. Despite the precautions taken, among the primers selected for our study: (1) some were previously validated in samples of clinical origin where the microbial communities were different from the microbial communities found in organic waste; (2) some were validated in a limited number of samples, and (3) detection technologies used previously to amplify some indicators have advanced. To our knowledge, this is the first report to use the ddPCR technique to quantify so many health indicators in environmental samples. This technique has been initially used as a reliable and ultraprecise method in the diagnosis of infectious disease (Cheng et al., 2019; Yan et al., 2019; Zeng et al., 2020).

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

CW: Conceptualization, Data curation, Formal analysis, Methodology, Validation, Writing – original draft, Writing – review & editing. BC: Conceptualization, Validation, Writing – review & editing, Funding acquisition, Investigation, Project administration, Resources. CM: Funding acquisition, Writing – review & editing, Methodology. MB: Methodology, Writing – review & editing, Data curation, Formal analysis. LL: Data curation, Methodology, Writing – review & editing. FL: Data curation, Writing – review & editing, Funding acquisition. KF: Data curation, Writing – review & editing. NS: Data curation, Writing – review & editing, Conceptualization, Formal analysis, Methodology. C-SH: Data curation, Writing – review & editing. SH: Data curation, Writing – review & editing, Funding acquisition, Methodology. DP: Data curation, Funding acquisition, Methodology, Writing – review & editing, Conceptualization, Formal analysis, Resources. GG: Funding acquisition, Methodology, Writing – review & editing, Data curation, Project administration. WG: Data curation, Funding acquisition, Methodology, Project administration, Writing – review & editing, Conceptualization, Formal analysis, Investigation, Resources, Software, Supervision, Validation, Visualization, Writing – original draft.