Unraveling the microbial contamination characteristics of pork farms and disinfection efficacy assessment via high-throughput sequencing

Abstract

The contamination of pork during breeding can negatively impact both food quality and safety. Sodium dichloroisocyanurate (NaDCC), a chlorine-containing disinfectant, is widely used in animal environments. In this study, 16S rRNA sequencing was used to determine the bacterial communities in a pork farm. We also assessed the effectiveness of NaDCC disinfection by analyzing microbial diversity. The results revealed that the predominant genera in pork skin samples and environmental samples were Staphylococcus and Acinetobacter, respectively. Escherichia coli and Streptococcus equinus were present in all the samples, suggesting that NaDCC is not fully effective in preventing zoonotic pathogen contamination. The Chao1 and Shannon indices for sow skins increased after disinfection. No significant differences were observed in the microbiological composition of sow barn samples before and after disinfection (p > 0.1). Additionally, bacterial abundance in sow skin was strongly correlated with abundance on pen floors (r = 0.906, p < 0.05), indicating potential cross-contamination between these sample types. Conversely, the bacterial abundance in the floor samples was weakly correlated with that in the other samples, suggesting that NaDCC is effective as a decontaminant. This study provides valuable insights into microbial contamination on pork farms and underscores the importance of proper disinfection practices. This study also offers scientific recommendations for improving disinfection protocols.

Introduction

The substantial expansion of China's pig sector, which supplies almost 50% of the world's pork supply, is contributing to the extraordinary rise in the demand for animal proteins worldwide (Pan et al. 2024). Zoonoses are contagious illnesses that people can contract from animals. An estimated 2.5 billion infections and 2.7 million fatalities are attributed to zoonoses each year, according to public health estimates, and zoonotic pathogens are responsible for around 60% of newly emerging infectious diseases (Carpenter et al. 2022). Pork farms can harbor significant levels of pollutants, including zoonotic pathogens, which pose substantial risks to both environmental and human health (Gržinić et al. 2023). Owing to the direct contact between humans and pigs, cross-contamination is likely to occur. Campylobacter was detected in 100% of the fecal samples from farms, whereas Staphylococcus aureus was detected in 48% of the fecal samples (Fosse et al. 2011). Pigs that are free of Salmonella may get the disease if they are exposed to contaminated environments (Kim et al. 2024).

Disinfection typically involves the use of physical and chemical methods to eliminate microorganisms, with the goal of reducing the risk of infection in both humans and animals exposed to contaminated environments (Martelli et al. 2017). Chlorination-based disinfectants, such as sodium hypochlorite (NaClO), sodium dichloroisocyanurate (NaDCC), chlorine dioxide, and chloramine, have long been established as standard disinfection methods. In China’s wholesale pork marketplaces, a commonly used disinfectant known as “84 disinfectant” is extensively applied to sanitize various surfaces, including floors, vehicle wheels, and carts (Wu et al. 2023). This disinfectant comprises NaClO and surface-active agents (Xiao et al. 2024). NaDCC is a chlorine compound that can be used as a substitute for NaClO (Clance and Edmondson 2006). The key advantages of NaDCC over NaClO include its lower cost, greater stability, enhanced safety, better solubility, and ease of handling, measurement, transportation, and storage (Clance and Edmondson 2006; Sharafi et al. 2017). NaDCC can eliminate 99% of bacteria at a concentration of 20 ppm and has been used as an industrial deodorant and disinfectant in various environments, including hard surfaces and the breeding industry (Clance and Edmondson 2006). These authors concluded that the greater disinfection efficiency of NaDCC may stem from its greater germicidal power, slower decomposition, prolonged release of hypochlorous acid (HClO), better capacity to maintain available free chlorine, and increased resistance to pH fluctuations (Mazzola et al. 2003; Sharafi et al. 2017). To date, however, there has been little discussion about the impact of NaDCC disinfection on bacterial contamination in pork farms.

The next-generation sequencing era began with the advent of the first high-throughput sequencing platforms at the start of the twenty-first century, making it possible to quickly and affordably sequence complete genomes (Barba et al. 2014). Comprehensive estimations of epidemiological processes over comparatively short durations have been made possible by the use of genomes in the study of infectious disease transmission dynamics (Croucher and Didelot 2015). The application of genomic tools in One Health research offers exciting potential for reconstructing transmission events through genomic epidemiology and phylodynamics (Khoury and Holt 2021). Pathogen genome analysis can assist in determining the causes of outbreaks, locating transmission pathways, and identifying possible superspreading events when combined with qualitative and quantitative epidemiological data (Östlund et al. 2011). A detailed investigation of the porcine skin microbiome identified Staphylococcus as the core microbiota (Strube et al. 2018). Sui used 16S rRNA amplicon sequencing to compare bacterial diversity at various stages of pork slaughtering (Sui et al. 2023). Their study revealed that the disinfection procedure was partially effective in reducing microbial contamination in the environment, significantly decreasing bacterial diversity and favoring certain genera, such as Psychrobacter and Weissella confusa.

In response to the global need for “One Health” strategies, we conducted 16S rRNA amplicon sequencing to characterize the microbial composition of pork farms (Fig. 1). The impact of disinfection steps on microbial communities was assessed, and critical control points that may cause cross-contamination were also identified. This research aimed to elucidate the interactions among human activities, animals, and disinfected environments, providing guidance for policymakers and practitioners in effectively applying a “One Health” approach.

Fig. 1

Flow charts of this study. A Sodium dichloroisocyanate was used to sterilize three areas of the farm. B 16S rRNA sequencing of collected samples

Results

ASV characteristics of pork farms

A total of 5,202,770 high-quality effective sequences were clustered into 8,074 amplicon sequence variants (ASVs). The sequencing depth and number of samples met the requirements for analysis. The sequencing results revealed sufficient uniformity and abundance (Fig. S1). The overall characteristics of the ASVs in each sample were statistically summarized on the basis of their abundance and species annotation (Table 1). The greatest number of ASVs in the barn aisle was detected. After disinfection, there was a significant decrease in the ASVs in the living area floor. The sow skin had a greater ASV and readings than did the piglet skin.

Table 1 Sequence data of bacterial communities from each sample

Characteristics of microbial contamination on sample surfaces

The microbiota on the surfaces of the different samples varied, with 29 phyla, 63 classes, 139 orders, 257 families, 726 genera, and 550 species. The stacked bar chart shows the dominant phyla, genera and species in each sample. At the phylum level, the primary phyla identified were Firmicutes (42.6%) and Proteobacteria (24.2%). Proteobacteria was the dominant phylum in the sow barns, whereas Firmicutes was the dominant phylum in the farrowing houses and living areas (Fig. 2A). Firmicutes (87.5%) predominated in piglet skin samples, which aligns with the findings of previous studies (Choudhury et al. 2019; Buiatte et al. 2024). At the genus level, we also identified 20 genera as the most abundant colonizers; Acinetobacter, Staphylococcus, Weissella, Empedobacter, and Enterococcus presented average abundance levels of 14.3, 9.2, 8.3, 7.7 and 4.5%, respectively (Fig. 2B). The dominant genera in the sow barn, farrowing house and living areas were Acinetobacter, Staphylococcus and Exiguobacterium, respectively. At the species level, Staphylococcus sciuri, Enterococcus faecalis, and Escherichia coli (E. coli) presented average abundance levels of 2.2, 2.1 and 1.0%, respectively (Fig. 2C). The dominant species in the sow barn, farrowing house and living area were Streptococcus equinus (S. equinus), Enterococcus faecalis and Staphylococcus sciuri, respectively. Notably, E. coli and S. equinus were present in all the samples.

Fig. 2

A Relative abundance of bacterial taxa at the phylum level. B Relative abundance of bacterial taxa at the genus level. The bars represent the relative abundance of the 20 most relevant bacterial genera (average total abundance > 5%). C Relative abundance of pathogenic bacteria at the species level. The bars represent the relative abundances of the 10 most relevant bacterial species

Differences in taxonomic profiles following NaClO disinfection

The Chao1 index quantifies species richness, whereas the Shannon index assesses both richness and evenness. No significant difference in alpha diversity before and after disinfection was observed (Fig. S2). A decrease in the mean Chao1 and Shannon indices was detected in living areas (Fig. 3A and B). Both the Chao1 and Shannon indices for pork pen floors from the farrowing house decreased following disinfection. However, the Chao1 and Shannon indices for sow skins from the barn increased following disinfection (Fig. 3C and D). Principal coordinate analysis (PCoA) revealed that the microbial structures in the majority of samples were in close proximity to each other, with the exception of the farrowing house samples (pork pen floors, piglet skins), which formed a distinct cluster separate from the other samples (Fig. 3E). Disinfection significantly affects the microbiological composition of floors in living areas (Fig. S3). Notably, there were no significant differences in the microbiological compositions of the samples from the sows before and after disinfection (p > 0.1), indicating potential cross-contamination (Fig. 3F).

Fig. 3

A Chao1 indices of farrowing houses, sow barns, and living areas before and after disinfection. B Shannon index of farrowing houses, sow barns, and living areas before and after disinfection. C Chao1 index of the bacterial flora at each sampling site. D Shannon indices of the bacterial flora at each sampling site. E Principal coordinate analysis of each sample type before and after disinfection. F Analysis of the principal coordinates of sow barns before and after disinfection

Correlation analysis among pork skins and environmental samples

Network inference was used to investigate patterns of co-occurrence between sampling-indicated types and genera and between sampling-indicated types and pathogenic bacteria. At the genus level, the results revealed that Aerococcus and Lactococcus were the main hosts in piglet skin. Weissella and Empedobacter were the main hosts in the environmental samples (Fig. 4A). Among the species-level pathogens, Enterococcus faecalis was the predominant host in piglet skin (Fig. 4B). Spearman correlation analyses were used to examine correlations between bacterial contamination at different sampling site types. The bacterial abundance in the skin of sows was highly positively correlated with that in their pen floors (r = 0.906, p < 0.05), suggesting that these two sample types were potential sources of bacterial cross-contamination. The abundance of bacteria in the living area floor samples had a lower correlation with that in the other samples, indicating that NaDCC was an effective decontaminant (Fig. 4C). Furthermore, given the limitations of the discriminant analysis described above, Figure 4D shows the evolutionary relationship from phylum to genus differential species after disinfection, and linear discriminant analysis (LDA) was used to identify specific bacterial genera related to postdisinfection. With a log LDA score threshold of 2.0, 41 differential genera were identified, with 17 belonging to group N and the highest score attributed to Bacteroides, and 24 belonging to group D, with Bacillus showing the highest score (Fig. 4E). Thus, NaDCC disinfection has a significant effect on Bacteroides but has little effect on Bacillus. We found that disinfection decreased the relative abundance of drug resistance via the antimicrobial pathway (Fig. S5).

Fig. 4

A Heatmap of correlation network analyses for sampled site types and dominant genera. The red lines indicate positive correlations, and the blue lines indicate negative correlations. The solid lines indicate the presence of significant effects, and the dashed lines indicate insignificant effects. The strength of the correlation is indicated by the thickness of the lines and the different colors. B Heatmap of correlation network analyses of sampling site types and pathogens at the species level. C Correlation analysis of bacterial taxonomy across all samples before and after disinfection. D Cladogram of the microbial communities. E LDA score of size differentiation using a threshold of 2

Discussion

The quantity of information offered by genetic data is crucial for pathogen characterization, illness surveillance, and the development of preventive measures. As genomic databases increase, they are becoming crucial big data resources in infectious disease research (Catalano et al. 2024). Several factors contribute to the widespread use of Illumina platforms, including high-throughput sequencing capabilities, cost-effectiveness, and ease of use. These platforms allow investigators to increase pathogen detection and characterization, even at low titers, and undertake larger-scale surveys (Reuter et al. 2015). According to our research, Staphylococcus was the most prevalent genus in piglet skin samples from farrowing houses, whereas Acinetobacter predominated in sow skin samples (Figure S4). Members of the genus Staphylococcus are common colonizers of mammalian skin, and some species, such as Staphylococcus spp., are known for their antibiotic resistance. These bacteria are often implicated in infections caused by multiresistant microorganisms, both in humans and animals (Mancuso et al. 2021). Acinetobacter poses a significant public health concern because of its high level of antibiotic resistance and ability to be transmitted through contact with contaminated surfaces (Carvalheira et al. 2021). Notably, E. coli and S. equinus were present in all the samples. E. coli is a common inhabitant of the gastrointestinal tract and feces of warm-blooded animals and reptiles (Ramos et al. 2020). It is a common gut commensal and a multipurpose pathogen that causes both extraintestinal and intraintestinal diseases that claim the lives of over 2 million people each year (Ramos et al. 2020; Leekitcharoenphon et al. 2021). S. equinus is part of the gastrointestinal microbiome in 6-9% of humans and animals and is the dominant Streptococcus in the gastrointestinal tract of horses. It has also been isolated from bovine milk, rumen, livestock feces, and ileostomy fluids. While S. equinus can cause acute rumen acidosis and flatulence in cows and is associated with mastitis, it is also of interest in food preservation research because of the broad-spectrum antibiotics it produces (Mcauliffe et al. 2001). A study in France from 2012-2017 reported 16 cases of infective endocarditis caused by S. equinus (Glajzner et al. 2021).

Our investigation revealed that disinfection of pork farms with NaDCC led to an increase in the relative abundances of Weissella, Pseudomonas, and Aerococcus. The row barns presented the highest levels of bacterial cross-contamination. Bacterial survival in disinfectant products may result from contamination with intrinsically resistant bacteria, such as Bacillus cereus spores in ethyl alcohol solution, bacteria that have acquired resistance, as observed with Serratia marcescens in a 2% aqueous chlorhexidine solution, or from the use of ineffective disinfectant concentrations due to improper application (Maillard and Pascoe 2024). Furthermore, the presence of organic matter is a critical factor to consider when laboratory results are translated to on-farm disinfection practices. The effective removal of organic matter is essential for eliminating pathogens from pork farm buildings, but this removal can be hindered by cracks and crevices in floors, walls, and ceilings, as well as the formation of biofilms. The efficacy of disinfectants in eliminating pathogens is influenced by both the type and concentration of disinfectant and may be significantly compromised by organic matter (Makovska et al. 2024). Wales et al. (2006) noted that it is more important to perform cleaning and disinfection to a high standard than to rely solely on the use of the best disinfectant after an inadequate cleaning procedure. In routine cleaning and disinfection, protocols recommend allowing sufficient drying time between detergent use and disinfectant application. However, in practice, there is often not enough time to fully adhere to this protocol before the next batch of pigs arrives.

NaDCC may have potential toxic effects on animals, including humans, cattle and rats (Hu et al. 2024). NaDCC presents thermal hazards, oral toxicity, and inhalation toxicity, and it can be harmful to humans if inhaled (Xu et al. 2022; Yoo et al. 2022). Furthermore, the use of humidifier disinfectants in South Korea has been linked to 1,256 fatalities out of 5,790 individuals (Seo et al. 2021). The Material Safety Data Sheet for NaDCC warns of potential irritation to the nose, mouth, trachea, and lungs if inhaled (Gosling et al. 2017). With respect to oral toxicity, Clasen and Edmondson reported that NaDCCs exhibit low acute oral toxicity and are not genotoxic or carcinogenic (Clasen and Edmondson 2006). Chlorination, chlorine dioxide, ozonation, ultraviolet radiation (UV), adsorption, membrane filtration, and coagulation are some of the methods of bacterial disinfection that have been developed to control bacterial levels within safe limits (Bharti et al. 2022). However, these conventional methods are plagued by multiple drawbacks. They may generate secondary toxic waste, entail high maintenance expenses, fail to completely eliminate pollutants in wastewater, exhibit poor recyclability, and involve the use of toxic chemicals. In contrast, advanced oxidation processes, typified by photocatalysis and piezocatalysis, have emerged as promising alternatives. These processes are capable of fully degrading organic waste pollutants and generating fewer toxic by - products. Specifically, photocatalysis harnesses visible light and piezocatalysis utilizes ultrasonic vibration. Both mechanisms rely on potent oxidants, such as hydroxyl radicals and superoxide anions, to break down organic contaminants (Zhang et al. 2024).

Conclusion

In conclusion, samples collected from farrowing, sowing, and living areas before and after disinfection presented distinct microflora profiles. NaDCC disinfection altered the microbial composition of pork farms. However, NaDCC is not adequate to completely prevent Escherichia coli contamination. Environmental factors could play a significant role in the dissemination of pathogens to pork. Our findings provide epidemiological and analytical context for genomic investigations on the transmission and diversity of zoonotic diseases across several One Health domains. One disadvantage of most microbiota investigations is that 16S rRNA gene analysis lack adequate resolution to characterize the microbiota at the strain or species level.

Methods

Sample collection

The trial was place on a breeding day in July 2023 at the Wuyi pork farm in Jinhua, China. The pork farm had a capacity of about 500 pigs, and a disinfecting regimen was followed every two days. Sampling was carried out during a single day to acquire 84 samples from various locations (Table 1). Before disinfection (N), 12 pork skin samples were collected from different age groups: piglet skin (Ds, n = 6) and sow skin (Ns, n = 6). Additionally, a total of 30 environmental samples were collected, including from farrowing areas (barn aisles, Dc, n = 6; pen floors, Dg, n = 6), sow barns (barn aisles, Nc, n = 6; pen floors, Ng, n = 6), and living areas (living area floors, L, n = 6). A disinfection strategy was followed using a NaDCC solution with an effective chlorine concentration of 100 mg/L to spray the surface. The time interval between the collection of environmental samples was approximately 4 h. Samples were collected after disinfection (D) from pork skin (piglet skin, aDs, n = 6; sow skin, aNs, n = 6), the farrowing area (barn aisle, aDc, n = 6; pen floor, aDg, n = 6), the sow barn (barn aisle, aNc, n = 6; pen floor, aNg, n = 6), and the living area (living area floor, aL, n = 6).

Sterile absorbent gauze premoistened with phosphate-buffered saline was used. The sampling involved wiping approximately 0.25 m × 0.25 m of available surfaces (pork barn aisles, pork pen floors, living area floors) horizontally, vertically, and diagonally in each direction with rotation of the swab. Pork skin samples were acquired by making three successive ~3 cm strokes at the same spot on the skin's surface. Disposable gloves were worn during the sampling procedure and replaced after each sample. Sample swabs were kited in a refrigerator with ice packs and transferred to the laboratory for processing within two hours (He et al. 2024).

DNA extraction

Microbial DNA was isolated from samples collected from farrowing houses, sow barns, and living quarters using the E.Z.N.A.® DNA Kit (Hernandez et al. 2021). The DNA concentration was measured using a Nanodrop One spectrophotometer, and the DNA quality was validated using 1% agarose gel electrophoresis. The isolated DNA was kept at −20°C for future analysis (Goumon et al. 2022).

PCR amplification, sequencing and analysis

For 16s rRNA, the PCR primer amplification sequences were 1492R (5'-RGYTACCTTGTTACGACTT-3') and 27F (5'-AGRGTTYGATYMTGGCTCAG-3') (Johnson et al. 2019). The PCR products were purified using the AxyPrep DNA Gel Extraction Kit according to the manufacturer's instructions (Li et al. 2024). SMRTbell libraries were generated from the amplified DNA using blunt-ligation (Callahan et al. 2019). The reads were then filtered in SMRT Portal on the basis of length and quality criteria (Kanwar et al. 2021). ASVs were created using paired sequence merging and chimera filtering (Amato et al. 2013).

To determine the alpha diversity indices, a Mothur-based rarefaction analysis was performed, and Origin was used to create a box diagram (Schloss et al. 2009). The Beta diversity analysis was performed via UniFrac using the community ecology software R-forge (Lozupone et al. 2011). ADONIS was used to measure the Bray‒Curtis distances. Phyla and ASVs with relative abundance levels > 1% and 0.1%, respectively, were defined as predominant and sorted for further comparison. LDA effect size (LEfSe) analysis was calculated to determine significant differences in the relative abundance of bacteria after disinfection. LEfSe analysis of microbial abundance was performed by pooling all samples between the N (predisinfection) and D (postdisinfection) groups to differentiate microbial communities, and online tools were used to draw graphs (Lyu et al. 2023; Segata et al. 2011). In addition, the LefsE figure was made with different locations before and after disinfection. The chi-square and Kruskal-Walli’s rank-sum tests were used to investigate the differences resulting from the distributions of multiple populations. LDA effect size (LEfSe) was applied to discover biomarker panels that possessed organismal features for differentiating the microbial communities specific to a particular treatment (Ijaz et al. 2018). In addition, we used the PICRUSt program (Lyu et al. 2023), which is based on the KEGG database, to predict functional changes in the microbiota (Liu et al. 2020). Relevance network heatmap analyses were performed via the NetCoMi software package in R, which uses online tools to draw graphs (Yao et al. 2024; Lyu et al. 2023).

Statistical analysis

The R software package was used for statistical analysis. Correlations were analyzed using Spearman’s coefficient (Yao et al. 2024). Figures were visualized with GraphPad Prism 8.