Evaluation of the occurrence of sporulating and nonsporulating pathogenic bacteria in manure and in digestate of five agricultural biogas plants

Abstract The number of agricultural biogas plants has been increasing in the past decades in some European countries. Digestates obtained after anaerobic digestion (AD) of manure are usually spread on agricultural land; however, their hygiene status regarding pathogens posing public health and/or animal health challenges has been poorly characterized up to now in France. In this study, three replicates of manure and digestate were collected from five farm biogas plants receiving animal manure in order to assess the occurrence and concentrations of sporulating (Clostridium botulinum, Clostridioides difficile, Clostridium perfringens) and nonsporulating (Listeria monocytogenes, thermotolerant Campylobacter spp., Salmonella, Escherichia coli, enterococci) bacteria. Concentrations of E. coli, enterococci, and C. perfringens in digestates ranged from 102 to 104, 104 to 105, and <103 to 7 × 105 CFU/g, respectively. Salmonella and C. difficile were detected in manure and digestate from the five biogas plants at concentrations ranging from <1.3 to >7 × 102 MPN/g and from 1.3 to 3 × 102 MPN/g, respectively. Thermotolerant Campylobacter, detected in all the manures, was only found in two digestates at a concentration of cells ranging from <10 to 2.6 × 102 CFU/g. Listeria monocytogenes and C. botulinum were detected in three manures and four digestates. The bacterial counts of L. monocytogenes and C. botulinum did not exceed 3 × 102 and 14 MPN/g, respectively. C. botulinum type B was detected at very low level in both the manure and digestate of farm biogas plants with no botulism history. The levels of pathogenic bacteria in both manure and digestate suggested that some bacteria can persist throughout AD.


| INTRODUC TI ON
Anaerobic digestion (AD) is a sustainable technology for converting livestock manure into biogas. Moreover, the stabilized residues of AD, called digestates, are usually spread on agricultural land as fertilizer. However, pathogenic microorganisms in manure and digestate can pose sanitary risks through land-spreading, such as the transmission of pathogens to vegetables. Pathogens such as Campylobacter jejuni, Salmonella spp., and Listeria monocytogenes are known to be responsible for major food-borne zoonotic diseases (EFSA, 2016) and to be excreted by farm animals that constitute a reservoir (Avrain, Humbert, Sanders, Vernozy-Rozand, & Kempf, 2004;Boscher, Houard, & Denis, 2012;Kempf et al., 2017;Milnes et al., 2008;Patterson, Kim, Borewicz, & Isaacson, 2016;Tadesse et al., 2011;Thépault et al., 2018). Moreover, these pathogens can persist in manure, soil and water (Cevallos-Cevallos, Gu, Richardson, Hu, & Bruggen, 2014;Erickson, Smith, Jiang, Flitcroft, & Doyle, 2014;Jäderlund, Sessitsch, & Arthurson, 2011). The fate of human and animal pathogens through the AD process is therefore of major concern. The pathogen die-off efficiency of this process depends on the feedstock composition as well as on operational parameters such as organic loading rate, hydraulic retention time, and temperature (mesophilic or thermophilic). Pathogen inactivation rates have been shown to be lower in mesophilic than in thermophilic AD plants (Watcharasukarn, Kaparaju, Steyer, Krogfelt, & Angelidaki, 2009).
The fate of pathogens, in particular Clostridium botulinum and Salmonella spp., during mesophilic AD appears to be a matter of public health concern, especially in Germany which has a high number of agricultural biogas plants (Froschle, Heiermann, Lebuhn, Messelhausser, & Plochl, 2015).
Up to now, no study had been conducted in France to assess the contamination of manure and digestate samples in agricultural biogas plants. The aim of this work was to assess the contamination of these organic products by sporulating (Clostridium perfringens, Clostridioides difficile, and C. botulinum) and nonsporulating (Escherichia coli and enterococci as biological indicators; Salmonella spp., thermotolerant Campylobacter and L. monocytogenes as major zoonotic bacteria) bacterial species. Besides pathogenic bacteria, fecal indicator bacteria (FIB), that is, E. coli, enterococci, and C. perfringens were monitored as they are commonly monitored to evaluate the sanitation efficiency of AD. This preliminary study provides a picture of the level of contamination of eight bacterial species in liquid manure and digestate from five biogas plants.

| Microbiological analysis
Culture-based methods were used for pathogen detection and enumeration except for C. botulinum for which no selective medium is available and molecular methods were required after the enrichment step.
For each pathogen, except for C. difficile for which 1 g samples were used for detection and enumeration, 25 g samples were homogenized in a filter bag with 225 ml of the appropriate enrichment broth using a Pulsifier (Microgen, Surrey, UK) for 15 s.

| Enumeration of FIB
For each FIB, 25 g samples were homogenized in a filter bag with 225 ml of buffered peptone water (BPW; Thermo Fisher Diagnostics SAS, Dardilly, France). Serial 10-fold dilutions were then prepared using sterile BPW.

| Detection and enumeration of L. monocytogenes
For enumeration, 1 ml of a 10-fold dilution performed in half-  2005). The broths were streaked on ALOA plates.
All the plates were incubated at 37°C for 24 hr. The presence of L. monocytogenes was deduced from the following characteristics of the ALOA colonies, that is, green-blue colonies with an opaque halo. Enrichment in Preston broth was undertaken in parallel, at 41.5°C in a microaerobic atmosphere (5% O 2 , 10% CO 2 , 85% N 2 ) for 24 hr, followed by streaking on CASA.

| Detection and enumeration of thermotolerant Campylobacter
All of the plates were incubated at 41.5°C in a microaerobic atmosphere for 48 hr. The presence of typical colonies on the plates (small curved bacilli with spiraling "corkscrew" motility) was checked under a microscope. All the enrichments (from detection and enumeration) were streaked on Rapid'Salmonella plates (BioRad Laboratories, Inc.,

| Detection and enumeration of Salmonella
Marnes-la-Coquette, France). The plates were incubated at 37°C for 24 hr. The presence of Salmonella was deduced from the following characteristics of the Rapid'Salmonella agar colonies, that is, fuchsia colonies.

| Detection and enumeration of C. botulinum
Due to the absence of selective media for the detection or enumeration of C. botulinum, a strategy different from the one used in this study for other bacterial species was carried out for this pathogen by combining cultural and molecular methods.
For detection of C. botulinum, regardless of the form (vegetative or spore cells), 25 g of each sample were 10-fold diluted in prereduced trypticase peptone glucose yeast broth (TPGY) and ho-

| Detection and enumeration of C. difficile
For detection of C. difficile, regardless of the form (vegetative or spore cells), 1 g of each sample was 10-fold diluted in brain heart infusion (BHI; BioMérieux, Craponne) supplemented with 0.1% taurocholate (Sigma Aldrich, Lyon, France), cefoxitin (8 mg/L) and cycloserine (250 mg/L) (Oxoid). Tubes were incubated at 37°C in the anaerobic chamber. After 7 days of incubation, streaking from the enrichment was performed on ChromID C. difficile plates using a 10 µl loop. The plates were incubated for 48 hr at 37°C in the anaerobic chamber. Positive colonies were recognizable by their specific black color and/or form.
Optimal incubation time (7 days) of supplemented BHI was determined by comparing the rate of recovery of C. difficile after 7, 10, and 30 days of incubation (data not shown).
For enumeration, 1 g of each sample was 10-fold diluted in BHI supplemented with 0.1% taurocholate, cefoxitin (8 mg/L), and cycloserine (250 mg/L). It was homogenized using a vortex and was then 1:5 diluted in a serial dilution in 2 ml of BHI in triplicate using a 12-well microplate. After 7 days of incubation at 37°C in the anaerobic chamber, each well was streaked on a ChromID C. difficile plate.
Positive colonies were recognizable by their specific black color and form. The MPN/g value was estimated by an MPN calculator with a 95% confidence interval.

| RE SULTS
Results on the detection of pathogens in the manures and digestates of the five biogas plants are reported in Table 2 and the FIB and pathogenic bacterial concentrations are reported in Table 3.

| Quantification of FIB
The concentrations of E. coli, ranged from 3.1 × 10 4 to 4 × 10 5 CFU/g in manures, were 0.7 to 2.5 Log 10 lower in digestates. Enterococci counts were in the same order of magnitude in manures and digestates (1.2 × 10 4 to 2.7 × 10 5 CFU/g). Clostridium perfringens had the highest variations in concentrations which ranged between less than 10 2 CFU/g (BP2 digestate) to 7.6 × 10 5 CFU/g (BP3 digestate). As observed for enterococci, the difference of concentration between manure and digestate did not exceed 0.8 Log 10 . The proportion of spores, ranged from 1.5% to 12.9% in manure, was close to that observed in digestates (0.9%-5.6%).

| Detection and quantification of pathogenic bacteria
Thermotolerant Campylobacter was present in all manures but only in two out of five digestates. Regardless the matrix (manure or digestate), their concentration ranged between 9.7 × 10 1 and 2.5 × 10 2 CFU/g. Except for BP3, where the concentration in digestate was 0.3 Log 10 more than in manure, thermotolerant Campylobacter counts were higher in manure than in digestate.

| D ISCUSS I ON
Regardless the method of quantification (direct plate count, enrichment prior to selective plating, or enrichment prior to PCR), all the targeted bacteria have been detected at least in one of the manure or digestate samples. It is noteworthy that culture-based methods were used for pathogen detection and enumeration instead of molecular ones here except for C. botulinum for which no selective medium is available.

| Fecal indicator bacteria
The E. coli and enterococci counts in manure, ranged from 10 4 to 10 5 CFU/g, were consistent with those reported in bovine manure (Bonetta, Ferretti, Bonetta, Fezia, & Carraro, 2011) and pig manure (Masse, Gilbert, & Topp, 2011;Pourcher, Ziebal, Kervarrec, Bioteau, & Dabert, 2012). Concentrations of E. coli were 1 to 2 log 10 lower in digestates than in manures while those of enterococci were of the same order of magnitude in manures and digestates. Although manures and digestates were collected on one occasion only and both on the same day, this preliminary study shows a higher reduction in E. coli than in enterococci concentrations regardless the BP. While the lower concentrations of E. coli in digestates were consistent with previous studies (Bonetta et al., 2011;Masse et al., 2011;Orzi et al., 2015), the nonremoval of enterococci was not observed by Orzi et al. (2015) who reported the systematic removal of these bacteria in agricultural biogas plants.
Except in the BP2 digestate, C. perfringens were detected in all the samples at concentrations ranging from 2.4 × 10 3 to 2.5 × 10 5 CFU/g. This was consistent with previous studies reporting concentrations of C. perfringens ranging from <10 3 to 3.7 × 10 6 CFU/g in both manures and digestates (Bagge, Sahlstrom, & Albihn, 2005;Masse et al., 2011;Orzi et al., 2015). Except for one biogas plant (BP2), differences in total counts (vegetative and sporulated cells) or spore counts between manures and digestates were below 1 Log 10 . Orzi et al. (2015) reported variable removal of C. perfringens, the concentration of which was either reduced or had remained stable after mesophilic AD. In our study, mesophilic condition did not change the proportion of spores, which remained close, between manure and digestate of a same BP.

| Nonsporulating pathogens
Campylobacter spp., L. monocytogenes and Salmonella were considered in this study due to their importance to public health (EFSA, 2016).
The higher prevalence in manures (100%), than in digestates (40%) suggests that thermotolerant Campylobacter spp. poorly persists through AD. Thermotolerant Campylobacter is known to be highly prevalent in intestinal contents of pigs and cattle, which can lead to contamination of manure. Their prevalence can reach 53.8% to 75.4% for pig intestinal contents (Avrain et al., 2004;Kempf et al., 2017;Milnes et al., 2008;Tadesse et al., 2011) and 54.6% to 69.1% for cattle intestinal contents (Milnes et al., 2008;Thépault et al., 2018). Other studies have demonstrated the occurrence of Campylobacter in 36.5% of pig manure (Farzan, Friendship, Cook, & Pollari, 2010). In our study, Campylobacter counts in manure were slightly lower (between 9.7 × 10 1 and 2.5 × 10 2 CFU/g) than in some previous studies with values of 10 3 to 10 4 CFU/g were reported in pig or dairy manure (Manyi-Loh et al., 2014;Masse et al., 2011). The origin of such differences could be related to many parameters such as livestock feeding, farm management, animal health, or manure storage. The role of these parameters in the presence of pathogens is, however, quite difficult to evaluate.
On the day of sampling, the prevalence of L. monocytogenes was higher in digestates (80%) than in manures (60%) ( Table 2).
However, the level of contamination was below 10 CFU/g in digestates (Table 3). This result may be related to the initial contamination of the manure, which feeds the biogas plant and which may vary during the year. Indeed, it has been shown that the prevalence of L.
monocytogenes in pig feces was significantly higher in autumn/winter (Boscher et al., 2011).
Nevertheless, our result seems to be concordant with available literature where prevalence was reported higher for digestates. The prevalence of L. monocytogenes in pig feces or manure has been reported to be low, at respectively 11%, 18.2%, and 3.3% depending on the study (Boscher et al., 2012;Farzan et al., 2010;Pourcher et al., 2012). A higher prevalence of this pathogen in digestates than in manures had previously been reported (Bonetta et al., 2011;Orzi et al., 2015). Indeed, Bonetta et al. (2011), who analyzed bovine manure and digestate from one mesophilic biogas plant over a 1-year period, detected L. monocytogenes in one of the five manure samples (20%) and in three of the 12 digestate samples (25%). Orzi et al. (2015) also detected L. monocytogenes in three out of eight digestate samples (37.5%) and two out of eight manure samples (25%), suggesting the ability of these bacteria to persist throughout mesophilic AD. The occurrence of L. monocytogenes in digestate is not surprising, as these ubiquitous bacteria can persist for up to 6 months in stored dairy slurry (Nicholson, Groves, & Chambers, 2005) and up to 40 days during the storage of digestates under microcosm conditions (Maynaud et al., 2016).

| Sporulating pathogens
The detection of C. botulinum in biogas plants was quite unexpected considering the available studies in the literature (Bagge, Persson, & Johansson, 2010;Froschle, Messelhausser, Holler, & Lebuhn, 2015;Neuhaus, Schrodl, Shehata, & Kruger, 2015). It had previously been shown that C. botulinum can be detected when the manure comes from farms with chronic botulism (Neuhaus et al., 2015). While the occurrence of C. botulinum had previously been demonstrated in bovine and hog intestinal contents (Dahlenborg, Borch, & Radström, 2001, our results, showed that when C. botulinum are present in manure, they can be detected at very low loads in fresh digestates, even in farms with no botulism history. The C. botulinum loads in our study were very low, whether in manures or in digestates, which was consistent with the results of DahlenborgBorch and Radström (2001), DahlenborgBorch and Radström (2003) who reported that 71% of positive pig fecal samples and 64% of positive cattle fecal samples had a spore load of less than four spores per gram.
The most common gene (100% of the positive samples) was that encoding BoNT type B, both in the manure and digestate samples.
Indeed, it had already been shown that C. botulinum type B is common in pig and bovine intestinal contents (Dahlenborg et al., 2001(Dahlenborg et al., , 2003. Although the prevalence of C. botulinum in pig and bovine intestinal contents has not been investigated in France to date, our preliminary results suggest that C. botulinum type B can frequently be detected in cattle and pig manure. Clostridioides difficile were detected in all the biogas plants (100%), which appeared consistent with previous studies. These bacteria are common on dairy and pig farms (Bandelj et al., 2016;Rodriguez et al., 2012) and are reported to be isolated more frequently from calves and newborn piglets than from adults (Alvarez-Perez et al., 2009;Hoffer, Haechler, Frei, & Stephan, 2010;Rodriguez et al., 2012).
Moreover, C. difficile are frequently detected in digestate from agricultural biogas plants  and in sludge samples (Romanazzi et al., 2016;Xu, Salsali, Weese, & Warriner, 2016 Clostridioides difficile loads ranged from less than 1.3 to 350 MPN/g with few differences between counts in manures and digestates among the same BP. Load was slightly higher in digestates compared to manures in BP1 (0.3 Log 10 ) and BP2 (0.5 Log 10 ) and lower in digestates than in manure collected in BP4 (1.6 log) and BP5 (0.8 log).  reported similar levels of C. difficile loads in their study (between less than 3 and 43 MPN/g) in agricultural biogas plants. In wastewater treatment plants, C. difficile loads in digested sludge samples vary between studies, with reported loads of 10 2 to 10 3 CFU/ml (Romanazzi et al., 2016), 10 1 to 10 2 CFU/ml (Xu et al., 2014), and around 10 4 CFU/ml (Viau & Peccia, 2009), showing their survival through AD as observed in our study.
Results on C. botulinum and C. difficile show that these two anaerobic spore-forming pathogenic bacteria may persist during AD but that there is no multiplication.

| Effect of operating conditions
We observed the highest concentrations of thermotolerant  (Kearney, Larkin, Frost, & Levett, 1993) described that C. jejuni could survive in a full-scale anaerobic digester operated at 28°C with naturally contaminated samples, which is consistent with our results on Campylobacter. However, it is noteworthy that BP3 manure also contained high concentrations of these pathogenic bacteria compared to the other BPs.

| CON CLUS ION
Considering the results obtained in our preliminary study, it can be suggested that spore-forming bacteria, as well as L. monocytogenes, Salmonella spp. and enterococci, have the ability to persist during AD. On the contrary, Campylobacter spp. was less commonly detected in digestate from mesophilic AD than in manure. No growth trend was detected through AD for these bacterial species. Overall, this study shows that concentration of the pathogens studied here were similar or lower in digestates than in liquid manures.
Determination of the occurrence and concentrations of these pathogens during the AD process over a longer period and with temporal replicates will allow the confirmation of these preliminary results by implementing statistical analyses and full comparison of the contamination of manures and digestates. Further questions like the characterization of pathogenic strains isolated from biogas plants to assess whether digestates are potential reservoirs of human and animal-pathogenic strains, or the evaluation of inhibition or potential regrowth of these pathogens during digestate storage, posttreatment or spreading also needs to be explored to better assess the risk of contamination.

ACK N OWLED G M ENTS
The authors are grateful to the participating farmers. This research was financially supported by the French Environment and Energy Management Agency (ADEME) (agreement number: 1606C0022).

CO N FLI C T O F I NTE R E S T S
None declared.

AUTH O R CO NTR I B UTI O N S
The authors collected the samples on farms and analyzed them.
CLM, MD, and AMP analyzed and interpreted the results. CLM, AMP, CD, and MD wrote the manuscript and acquired the funding for this study (CloDia project). AMP is the coordinator of the CloDia project. MD is the general supervisor of the research group (HQPAP unit). All of the authors read and approved the final manuscript.

E TH I C S S TATEM ENT
None required.

DATA ACCE SS I B I LIT Y
All data are provided in full in the results section of this paper and in tables. The authors adhere to all policies on sharing data and materials described in the guidelines for authors.