Salmonella is a foodborne pathogen commonly associated with poultry products. The aims of this work were to (i) estimate the impact of critical steps of the slaughter process on Salmonella detection from broiler chicken carcasses in two commercial poultry slaughter plants in Quebec, Canada; (ii) investigate the presence of Salmonella in the slaughter plant environment; (iii) describe, using a high-resolution melting (HRM) approach, the HRM Salmonella profiles and serotypes present on carcasses and in the slaughter plant environment; and (iv) evaluate whether the HRM flock status after chilling could be predicted by the flock status at previous steps of the slaughter process, the status of previous flocks, or the status of the processing environment, for the same HRM profile. Eight visits were conducted in each slaughter plant over a 6-month period. In total, 379 carcass rinsates from 79 flocks were collected at five critical steps of the slaughter process. Environmental samples were also collected from seven critical sites in each slaughter plant. The bleeding step was the most contaminated, with >92% positive carcasses. A decrease of the contamination along the slaughtering process was noted, with carcasses sampled after dry-air chilling showing ≤2.5% Salmonella prevalence. The most frequently isolated serotypes were Salmonella Heidelberg, Kentucky, and Schwarzengrund. The detection of the Salmonella Heidelberg 1-1-1 HRM profile on carcasses after chilling was significantly associated with its detection at previous steps of the slaughter process and in previously slaughtered flocks from other farms during a same sampling day. Results highlight the importance of the chilling step in the control of Salmonella on broiler chicken carcasses and the need to further describe and compare the competitive advantage of Salmonella serotypes to survive processing. The current study also illustrates the usefulness of HRM typing in investigating Salmonella contamination along the slaughter process.
Salmonella contamination of chicken carcasses was the highest after bleeding.
Most frequent Salmonella serotypes were Heidelberg, Kentucky, and Schwarzengrund.
Carcass status after chilling is associated with flock status at previous steps.
Results support chilling as being critical in the control of Salmonella.
Salmonella enterica is an important zoonotic pathogen, having a significant economic and health impact on animals and humans worldwide (3). In Canada, Salmonella remains one of the most burdensome foodborne pathogens, with more than 17.3 cases of salmonellosis per 100,000 inhabitants reported each year, with the number of affected people probably being up to 30 times higher due to underdiagnosis or underreporting (18, 42). More than 2,600 Salmonella serotypes have been identified, but a limited number of these serotypes are commonly linked to foodborne diseases in humans (38). Clinical manifestations of salmonellosis are usually characterized by fever, abdominal pain, nausea, and vomiting. Young children, the elderly, and immunocompromised people are at increased risk of complications from the infection (3).
The intestinal tract of farm animals represents the main habitat of zoonotic Salmonella. Various sources, namely feedstuff, rodents, litter, and visitors, can all contribute to the introduction of the pathogen into farming environments, probably explaining the high prevalence (44 to 59% depending on the year) of contaminated preharvest chicken farms reported in Canada (22, 24). Transmission of the pathogen to humans is most often occurring through the ingestion of contaminated foods, with meat, eggs, dairy products, and vegetables identified as the main contributing vehicles (24). Among foods of animal origin, poultry meat is recognized as a main source of human exposure to Salmonella (3). Several studies have reported that live birds originating from Salmonella-positive flocks are responsible for the introduction of the pathogen into poultry processing plants; the introduction can occur through plumage contamination or via damage to the gut during the slaughter activities, resulting in a leakage of Salmonella-contaminated intestinal content (34, 49). Although a limited number of Salmonella serotypes have been isolated from poultry, different prevalences and within-serotype diversities have been observed between geographic regions and farms (2, 27, 38).
The identification of critical control points to manage the microbiological risk in poultry slaughter plants is a key element of the slaughter process (49). A critical control point is defined as a step at which control measures can be applied to prevent or to control a hazard throughout the operations. Food processors have largely relied on the application of this approach to manage meat product contamination (9). The monitoring of Salmonella on broiler chicken carcasses at different steps along the slaughter line by using a carcass rinse approach has helped identifying critical control points impacting the meat product contamination (4, 25, 26, 39, 49).
Few studies have described the dynamics of broiler chicken contamination by Salmonella serotypes along the poultry production chain and the influence of critical steps of the poultry slaughter process, from the incoming live birds to the processed carcasses, on these dynamics (25, 31, 47). Although rapid detection and differentiation of Salmonella serotypes still represent a challenge for the food industry, recent advances in molecular methods have made the detection of Salmonella more accurate and convenient (48). Many typing approaches used for the differentiation of Salmonella based on genomic characteristics are proposed (5, 7, 19, 25, 40, 51, 52). The high-resolution melting (HRM)–based Salmonella genotyping method is rapid, robust, easy to interpret, and affordable compared with other approaches such as serotyping, pulsed-field gel electrophoresis, and whole genome sequencing (7). Compared with a complete serotyping method, the greater discriminatory power of the HRM approach allows for the generation of subtypes among the serotypes identified. To our knowledge, no study using an HRM approach and aiming at understanding the distribution of Salmonella on broiler chicken carcasses along the slaughter process in commercial poultry processing plants has been conducted.
Therefore, the objectives of the present study were to (i) estimate the impact of five critical steps of the slaughter process on Salmonella detection from broiler chicken carcasses in two commercial poultry slaughter plants (abattoirs 1 and 2) in Quebec, Canada; (ii) investigate the presence of the pathogen in the environment of the two surveyed plants; (iii) describe the HRM Salmonella profiles and serotypes present on broiler chicken carcasses and in the slaughter plant environment; and (iv) evaluate whether, at the flock level, Salmonella carcass contamination detected at previous steps of the slaughter process, in the previously slaughtered flock, and/or in the slaughter plant environment can be predictive of the final meat product contamination by specific Salmonella HRM profiles after chilling.
MATERIALS AND METHODS
Broiler chicken flock selection
Sampling for this study was carried out in two different commercial poultry processing plants in the province of Québec, Canada. Each plant was visited eight times between February and July 2017. The characteristics of these surveyed abattoirs are presented in Table 1. For both slaughter plants, sampling visits were scheduled according to the number and origin of flocks slaughtered per day, to ensure that at least five broiler chicken flocks, all from different farms, were scheduled for slaughter at the beginning of the first shift of the day, after completion of the sanitation procedures. For each plant visit, these first five flocks slaughtered during a sampling day were sampled.
For each sampled flock, one carcass was sampled at each of five critical (C) steps identified as C1, after bleeding, with the feathers still attached to the carcass; C2, at time of transfer between the live receiving and the evisceration departments, before evisceration; C3, before chilling, after evisceration; C4, after water-immersion chilling; and C5, after dry-air chilling (Fig. 1). For each flock, five carcasses were selected from the last one-third of the slaughtered flock by collecting the first carcass arriving on the chain in front of the collection site once the research team was ready to proceed. Only four carcasses were sampled for flocks from which no air-chilled carcasses were available due to the production requirements of the surveyed slaughter plants.
Each sampled carcass was placed in a sterile plastic bag (Nasco poultry rinse sample bag, Fisher Scientific, Ottawa, Ontario, Canada), and a 550-mL volume of buffered peptone water (Biokar Diagnostic, Beauvais, France) was added. The carcass was vigorously shaken for 1 min before being removed from the bag with an approach preventing contamination of the sample. The remaining volume of rinsate was recovered, placed on ice, transported back to the laboratory, and stored overnight at 4°C. Samples were individually processed the next morning.
Environmental (E) samplings were carried out at seven critical sampling locations (CSLs) (23). CSLs were defined as follows: E1, the feather-plucking rubber fingers; E2, the conveyor belt between the live-receiving and the evisceration departments; E3, the eviscerating machine; E4, the floor surface in the evisceration department; E5, the conveyor belt before chilling; E6, the conveyor belt after chilling; and E7, a stainless steel equipment surface in contact with the meat product and located in the cut-up room (Fig. 1). Each of the seven CSLs was sampled twice during a same plant visit: after the sanitation procedures, before the slaughter activities (postsanitation [PS]) and at the end of the work shift (postoperation [PO]), for a total of 14 samples on each of the eight visits in each abattoir.
Sterile gauzes moistened in a 10-mL volume of neutralizing buffer (DE neutralising broth, Lab M, Ltd., Heywood, UK) were used to rub a surface (10 by 10 cm) on each CSL. After sampling, swabs were put back into their respective sterile bags, placed on ice, transported back to the laboratory, and stored overnight at 4°C. Samples were individually processed the next morning.
Sample treatment and microbiological analysis
Based on the Canadian Food Inspection Agency and Food Safety and Inspection Service requirements to which exporting Canadian poultry processors need to comply, the carcass rinsate volume submitted to a preenrichment step was limited to 50 mL, which represents 9% of the volume rinsate (15). To do so, from each carcass rinsate, a 200-mL volume was centrifuged for 20 min at 15,000 × g. The supernatant was removed, and a volume of 4 mL of buffered peptone water was added to the pellet and vortexed until complete resuspension of the pellet. The isolation protocol for Salmonella described by Larivière-Gauthier et al. (30) was used, with slight modifications. For each sample, 1 mL of the suspension was distributed into 9 mL of sterile buffered peptone water before being homogenized. Tubes were incubated at 37°C for 24 h. Three 100-μL equidistant drops of the preenriched culture were inoculated onto the surface of modified semisolid Rappaport-Vassiliadis agar plates (Lab M, Ltd.). Plates were incubated at 42°C for 24 h and then examined for any bacterial growth that was revealed by the development of a white migration zone. When no migration zone was observed, modified semisolid Rappaport-Vassiliadis agar plates were incubated for an additional 24 h and then reassessed. Typical migrations on modified semisolid Rappaport-Vassiliadis medium were subcultured on both brilliant green sulfa (BGS) agar plates (BD, Difco, Franklin Lakes, NJ) and on xylose lysine desoxycholate (XLD) agar plates (Biokar Diagnostics; International Organization for Standardization [ISO] 6579) that were incubated at 37°C for 24 h. When suspect colonies were available on BGS and/or XLD agar plates, one colony per selective agar plate was recovered. Suspect colonies on BGS and XLD plates were confirmed as Salmonella by using triple sugar iron agar (Lab M, Ltd.), urea agar (Lab M, Ltd.), followed by sero-agglutination by using Salmonella O antiserum Poly A-I C Vi (Statens Serum Institute, Copenhagen, Denmark). Positive cultures were subcultured on sheep blood agar plates (Oxoid, Nepean, Ontario, Canada) and stored at −80°C in a Brucella agar freezing medium (Difco, BD) containing 25% glycerol (Fisher Scientific).
For environmental samples, 20 mL of sterile buffered peptone water was added to each gauze-containing bag. Bags were homogenized for 1 min by using a stomacher and incubated at 37°C for 24 h. The same Salmonella detection protocol as described above was applied.
Preparation of genomic DNA
A DNA extraction protocol using a 10% Chelex 100 (Bio-Rad Laboratories, Mississauga, Ontario, Canada) solution in water was applied (23). In brief, three to five colonies of each pure Salmonella isolate grown on blood agar plates were suspended in 1 mL of sterile water, vortexed, and centrifuged at 12,000 × g for 3 min. After removal of the supernatant, 200 μL of the 10% Chelex 100 solution was added to the bacterial pellet that was vortexed and then incubated in a water bath at 100°C for 20 min. The supernatant (125 μL) was recovered after a second centrifugation at 12,000 × g for 3 min, transferred into sterile Eppendorf tubes, and stored at −20°C until used for molecular analysis.
InvA PCR-based Salmonella confirmation
The identity of all recovered Salmonella isolates was PCR confirmed by amplification of the invA gene (12). PCR amplifications were performed using a TProfessional Basic 96 thermocycler (Biometra GmbH, Göttingen, Germany). Each reaction was conducted in a 25-μL volume made of 1 μL of each primer (10 nmol/μL) (12), 3 μL of extracted genomic DNA, 2.5 μL of reaction buffer (10× ThermoPol reaction buffer, NEB, Whitby, Ontario, Canada), 2 μL of deoxyribonucleotide triphosphates (10 mM; Bio Basic Inc., Markham, Ontario, Canada), 0.5 μL of Taq DNA polymerase (NEB), and 15 μL of sterile water.
Reaction conditions were determined according to the protocol published by Chiu and Ou (12). The PCR program was as follows: an initial denaturation step at 94°C for 1 min, followed by 35 cycles of denaturation at 94°C for 30 s, annealing at 54.2°C for 30 s, extension at 72°C for 2 min, and final extension step at 72°C for 10 min.
PCR products were visualized and photographed under UV illumination following electrophoresis on a 1% agarose gel containing 0.01% SYBR Safe DNA gel stain (Invitrogen, Burlington, Ontario, Canada). A 100-bp ladder (Track It, Invitrogen) was used as a molecular weight marker.
Real-time PCR and HRM curve analysis
HRM-based genotypic characterization was used on all PCR-confirmed Salmonella isolates. DNA templates used for the conduct of the approach were diluted 1:10 in sterile water. Each 20-μL reaction volume contained 12.4 μL of sterile water, 0.8 μL of each primer, 4 μL of EvaGreen Mastermix (Montréal Biotech, Montréal, Quebec, Canada), and 2 μL of Salmonella genomic DNA.
Primers and linear normalization regions were selected according to the protocol published by Bratchikov and Mauricas (7). Real-time PCR and HRM curve analysis were performed using a LightCycler 96 real-time PCR thermocycler (Roche Diagnostics, Mannheim, Germany). Three genomic regions were amplified: CR1, CR2, and YohM.
For each HRM-typed strain, the combined analysis of the three curves generated by the three selected genes was attributed an HRM profile. HRM-typed strains were grouped according to their HRM profile and each profile was subsequently identified as a Salmonella serotype after submitting to serotyping at least one strain per HRM profile. Serotyping was performed by the Laboratoire d'épidémiosurveillance du Quebec (Ministère de l'Agriculture, des Pêcheries et de l'Alimentation du Québec, Saint-Hyacinthe, Quebec, Canada).
Descriptive statistics was used to present the data. A Salmonella-positive carcass or environmental sample status was established when a positive culture was obtained on either one or both culture media (BGS and XLD).
Multivariable logistic regressions were used to model the carcass Salmonella status (positive or negative) according to the critical step and slaughtering order, conducted separately for each slaughter plant. The sampling day was included in the model as a fixed effect to adjust for potential clustering, and the model was adjusted for repeated measures within flocks. A full model including all variables was built; however, nonstatistically significant variables (P > 0.05, type 3 Wald test) were then removed one at a time from the model (backward selection) (17). For statistically significant categorical explanatory variables, pairwise comparisons between categories were performed with P values adjusted using the Tukey-Kramer procedure for controlling the overall type 1 error rate at 5%. Odds ratios (ORs) were used to present the results (17).
A logistic regression was used to model Salmonella environmental sampling status (positive or negative) according to the type of sampling (PS versus PO), performed separately for each slaughter plant. Because of the limited sample size, no comparison was done between CSLs.
Finally, a logistic regression model was used to predict the flock status (positive or negative) after chilling for specific HRM profiles, that is, detection of a specific profile on carcasses of a same flock sampled at either C4 or C5. One model was built for each profile and slaughter plant, and models were limited to HRM profiles detected after chilling in at least five flocks from a same slaughter plant. Evaluated predictors were the detection of this HRM profile on carcasses from the same flock at each of the previous steps, the detection of this profile in previous flock(s) slaughtered the same day in the same plant, and the detection of this same profile in environmental samples at PS. The model was adjusted for repeated measures within slaughter visit. Because of the limited sample size, only univariable analyses were performed. Odds ratios are used to present the results.
All regression analyses were conducted in SAS 9.4 (SAS Institute Inc., Cary, NC) by using the Genmod procedure.
In total, 379 broiler chicken carcasses from 79 flocks were sampled between February and July 2017 in the two surveyed slaughter plants. On the eight sampling days in each abattoir, five flocks originating from a different farm were sampled, except for one sampling visit conducted in abattoir 2 for which only four different flocks were available at the time of plant visit. For the whole sampling period duration, 68 different broiler chicken farms in total were sampled, with some farms being sampled more than once. According to the selective medium used, 138 (36.4%) carcass samples were positive for Salmonella on both BGS and XLD agar plates, whereas 9 (2.4%) carcasses were positive only on BGS agar plates and 14 (3.7%) carcasses were positive only on XLD agar plates.
Salmonella carcass contamination varied among critical steps of the slaughter process
Salmonella was detected on at least 1 carcass from all flocks sampled in abattoir 1 (40/40) and from 35 of the 39 flocks sampled in abattoir 2. The Salmonella carcass positivity at each critical step of the slaughter process and according to the slaughtering order for both surveyed slaughter plants is presented in Table 2. For both plants, C1 was the most frequently detected as positive for the presence of Salmonella, with >89% of the carcasses sampled at this critical step being Salmonella positive. Salmonella was detected at least once in every sampling point of the processing line identified as critical, except at C5 (air chilling) in abattoir 2. Similarly, only one positive sample was found at this step in abattoir 1 (Table 2).
For both slaughter plants, only the critical step variable was retained in the final multivariable logistic regression modeling the Salmonella status of carcasses, as the slaughtering order and date of slaughter were not significantly associated with carcass positivity during model building (P > 0.05). For abattoir 1, the proportion of Salmonella-positive carcasses was significantly associated with the critical step (P < 0.001, 3 df). According to pairwise comparisons, this proportion was significantly higher at all steps than for C5 (after dry-air chilling) (Table 2). It was also significantly higher at C1 (after bleeding) than at subsequent C2 (OR = 11, adjusted P < 0.01), C3 (OR = 20, adjusted P < 0.01), and C4 (OR = 15, P < 0.01) steps, respectively. For abattoir 2, the proportion of Salmonella-positive carcasses was also statistically significantly associated with the critical step (P < 0.001, 3 df). For this abattoir, the two chilling steps (C4 and C5) were combined to allow for model convergence, considering the very low number of Salmonella-positive carcasses from these two steps. Carcasses at both water-immersion and dry-air chilling steps were significantly less contaminated than at any other of the three previous steps (Table 2). The odds of positivity to Salmonella was significantly higher in carcasses sampled at C1 (bleeding) than at C2 (OR = 11, adjusted P < 0.01) and C3 (OR = 14, adjusted P < 0.01) steps, respectively.
Slaughter plant environmental contamination increases after operations
In total, 224 environmental samples were collected (112 in each abattoir) from the surveyed plants during the visits. Overall, 54 (24.1%) samples were found to be positive for Salmonella on both BGS and XLD agar plates, whereas 10 (4.5%) samples were positive only on XLD agar and 10 (4.5%) samples were positive only on BGS agar. The distribution of Salmonella-positive environmental samples is presented in Table 3. For both plants, the proportion of Salmonella-positive samples was significantly higher (P < 0.001) for samples collected after the slaughter activities (PO) than after the sanitation procedures (PS), before the slaughter activities. The odds of positivity at PS was 3.7 (1.3 to 10.4) times higher for abattoir 1 and 10.9 (95% confidence interval: 4.6 to 26.2) times higher for abattoir 2, respectively, than at PO.
Identification of Salmonella HRM profiles and serotypes
In total, 432 Salmonella isolates were recovered (302 isolates from carcasses and 130 isolates from the environmental sampling). Each isolate was attributed a three-digit number corresponding to the HRM profile obtained following analysis of the fusion curves for CR1, CR2, and yohM genomic regions (Supplemental Fig. S1). Among the 192 samples from which isolates were recovered from both BGS and XLD culture media, a same HRM profile was identified for the two isolates in 159 (82.8%) samples, whereas the analysis of the two isolates generated different HRM profiles corresponding either to a same Salmonella serotype for 7 (3.6%) samples or to a distinct serotype for 26 (13.5%) samples.
In total, 40 distinct HRM profiles corresponding to 15 different Salmonella serotypes were identified among all isolates. Up to eight different HRM profiles per serotype were identified (Table S1). Among these HRM profiles, 17 were identified only from the environmental samples, 17 profiles were found exclusively in carcass rinsates, and 6 profiles were found in both types of samples. Twenty-two HRM profiles were observed only in abattoir 1, 12 were unique to abattoir 2, and 6 were common to both plants.
The most commonly identified Salmonella serotypes in carcasses were Heidelberg, Schwarzengrund, and Kentucky in both abattoirs (Table 4). Salmonella serotypes Heidelberg, Schwarzengrund, Kentucky, Mbandaka, and Thompson were the only serotypes identified from the environmental samples collected at PS. Salmonella Heidelberg was isolated at least once from all environmental surfaces sampled, except from E6. For abattoir 1, the presence of a single serotype and of a single HRM profile was noted for 16 of the 40 flocks sampled, whereas analyses revealed the presence of two different serotypes and of two distinct HRM profiles in 17 and 16 flocks, respectively (Table 5). In this same abattoir, seven and eight flocks were found positive for the presence of three distinct serotypes and HRM profiles, respectively. Similarly, a single serotype and HRM profile was identified from 14 and 12 flocks of the 39 flocks slaughtered in abattoir 2, respectively. Fifteen and 17 flocks sampled from abattoir 2 revealed the presence of two distinct serotypes and genotypes, respectively, whereas three different serotypes and HRM profiles were found on the carcasses of 6 flocks sampled in this plant.
Predicting the final flock status
For abattoir 1, 18 flocks had at least one carcass positive for Salmonella after chilling (C4 and/or C5) (see Table S2). Among these, 17 flocks also showed a Salmonella-positive carcass in at least one of the previous steps of the slaughter process. The Salmonella status of carcasses sampled at previous steps (C1, C2, and/or C3) for these 17 flocks was distributed as follows: 3 flocks were exclusively positive to the same HRM profile in at least one of the previous steps, 10 flocks were positive to both the same and different HRM profiles, and 4 flocks were positive for a different profile only. When a same flock was found Salmonella positive at both C1 and C4/C5 in this same abattoir, Salmonella Heidelberg 1-1-1 and Kentucky 6-6-1 were the sole HRM profiles identified. For abattoir 2, only two flocks were positive after chilling and neither of these flocks showed similar HRM profiles between the carcasses sampled at C1 and those sampled at C4/C5.
Based on these results, the model predicting the flock status after chilling was only performed for abattoir 1 with Salmonella Heidelberg HRM profile 1-1-1 for which a sufficient number of positive samples was available to allow for model convergence. Overall, the flock status for Salmonella Heidelberg 1-1-1 at any of the previous steps (C1, C2, C3) of the slaughter process was predictive of the flock status for this same HRM profile after chilling (Table 6). Of note, for 10 of the 12 flocks found positive for Salmonella Heidelberg HRM profile 1-1-1 at C4/C5, this profile was also isolated from the carcass rinsates in at least one of the previous steps. The Salmonella Heidelberg 1-1-1 status of the previously slaughtered flock(s) in the same day was also predictive of a flock carcass status for this same HRM profile after chilling. No statistically significant association was found or could be tested with the status of environmental samples (Table 6). The distribution of positive samples by flock and sampling day is illustrated in Table S3.
Salmonella-positive broiler chicken carcasses were identified at each critical step of the slaughtering line, as reported in a similar study by Rivera-Perez et al. (49) conducted in Costa Rica. They found that the critical step showing the highest level of contamination was C1 (after bleeding), which is not surprising considering that the feathers, skin, crop, and cloaca of birds entering the slaughter plant can all carry significant bacterial loads (21). Salmonella contamination at this step could originate from the farm environment, from the transport coops, or even from an increased shedding of the pathogen from carrier birds during the stressful transport conditions (27, 46). Some of the Salmonella contamination might also have originated from the first steps of the slaughter process, such as through the contamination carried on the hands of abattoir workers, the knife used for bleeding, or the shackles on which birds were hung (32, 47). In contrast to other studies reporting the evisceration step as critical for magnifying poultry carcass contamination by Salmonella, usually following a rupture of the intestinal tract, this critical step appeared to be under control in abattoirs in the this study, because a decrease in the prevalence of Salmonella-positive carcasses was observed for both plants at this step compared with C1 (25, 49).
Water chilling was associated with a reduction in Salmonella carcass contamination only in abattoir 2. Chilling equipment and chemical antimicrobial agents used for carcass sanitation can vary considerably among processing establishments (26). For the present study, the water renewal rate, the carcass residence time in the chiller, and the incorporation of a chemical processing aid into the cooling water are three major factors that could have contributed to the higher prevalence of Salmonella-positive carcasses in abattoir 1. Indeed, whereas abattoir 1 was using a non-countercurrent flow system with a water turnover of 35 L/min, carcasses sampled in abattoir 2 were chilled in a countercurrent flow system for which the water turnover was 110 L/min. It is well established that the water renewal rate contributes to reducing carcass bacterial loads and preventing bacterial multiplication as well as controlling the microbiological status of the cooling water (28, 37). Although the carcass residence time in the chilling system was close to 2 h for abattoir 2, this time was limited to 1.5 h in abattoir 1, mainly due to the smaller size of the prechill reservoir for which the temperature control was more challenging owing to its limited capacity. Moreover, peracetic acid was added to the chilling water in abattoir 2, which was not the case for abattoir 1. Even if water chilling has been reported to reduce the mean aerobic bacterial load of poultry carcasses, immersing carcasses in a sanitizer-free chilling reservoir was also correlating with an increased frequency of Salmonella detection among these carcasses, probably reflecting the cross-contamination occurring at this step (8, 32). Indeed, the 45% Salmonella prevalence observed for carcasses exiting the chiller tank in abattoir 1 does not support the hypothesis that the sole physical effect of washing during immersion chilling is the primary mode of action for chilling to remove Salmonella from broiler chicken carcasses at this step (37). By contrast, the 5.1% prevalence obtained from the sampling of carcasses at C4 in abattoir 2 supports a significant contribution of chemical processing aids in managing the contamination of poultry meat products by Salmonella: the 37.5 and 38.5% Salmonella positivity identified just before entering the chilling reservoirs was highly similar between the two plants (8, 26). Although most of these sanitizers show great efficacy for controlling foodborne pathogens, some evidence seems to indicate that the different Salmonella serotypes found in poultry would respond differently to the chemical action of the various sanitizers used during chilling, with chlorine-based products reported to be the most effective for Salmonella control in commercial conditions (26, 33, 54).
Our results highlight dry-air chilling as critical in reducing the contamination with Salmonella, with the proportion of positive carcasses decreasing from 37.5 to 2.5% and from 38.5 to 0% in abattoir 1 and 2, respectively. It can be hypothesized that the absence of the mechanical washing effect (as provided by water-immersion chilling), the close contact between hung birds, the use of water sprays in the air chilling room, or even the carriage of cetylpyridinium chloride (a quaternary ammonium compound) resistance genes such as qacEΔ1 by some of the Salmonella strains present on the air-chilled carcasses sampled might not have been significant contributors to a residual contamination by the pathogen (13, 26). It can therefore be hypothesized that the use of cetylpyridinium chloride combined with a prolonged contact time and the desiccation effect of the forced-air cooling system can significantly reduce Salmonella contamination found on broiler chicken carcasses.
The great serotype diversity observed for carcasses at C1 suggests that multiple sources of the pathogen contribute to this contamination. Salmonella diversity was previously reported to be greatest at the broiler chicken farm level than at other stages of the production chain (15). It is assumed that the diversity of Salmonella present at this step is representative of the farm and transport (49). According to the Canadian Integrated Program for Antimicrobial Resistance Surveillance, the most frequently detected serotypes at the broiler farm level are Salmonella serotypes Heidelberg, Kentucky, and Enteritidis, with varying predominance over years (10). Results from the current study are thus in line with what was observed at the national level. They also highlight the emergence of Salmonella Schwarzengrund, which has been identified as such in the United States, in addition to having been incriminated in the occurrence of human disease outbreaks attributed to the consumption of poultry meat products (11, 14). Salmonella Heidelberg was the most frequently identified serotype in the present study. According to the FoodNet Canada annual report (42), Salmonella Heidelberg consistently ranks as one of the top three most prevalent causes of human salmonellosis in the country. Also, Salmonella Heidelberg was associated with a more invasive form of human disease and with a higher case fatality rate than other nontyphoidal Salmonella serotypes (28). In 2017, a Salmonella Heidelberg outbreak was responsible for nine cases of foodborne illness in six Canadian provinces and territories, and frozen raw breaded chicken products were identified as the source of infection (44, 45).
Even if the use of next-generation sequencing approaches such as whole genome sequencing would represent the most discriminant approach when it comes to tracking Salmonella, the use of an HRM typing approach on the isolates recovered during the present study was discriminant enough to define with more precision the role of critical steps of the slaughter process, of the slaughter plant environmental contamination, and of previous flocks slaughtered during a same sampling day, which contrasts with other studies reporting results on the dynamics of Salmonella carcass contamination within the slaughter process by using positivity and/or serotype data only (6, 25, 39, 49).
Interestingly, the CSL for which a residual Salmonella contamination was detected at PS in both surveyed plants is closely related to the plucking step, which is recognized as one of the most critical steps contributing to the cross-contamination of carcasses moving along the kill line (36, 49). The evisceration step may represent an increased risk for the contamination of broiler chicken carcasses, as supported by the six of eight samples collected at this step (E3) in abattoir 2 being found positive for Salmonella at PO. The absence of any Salmonella-positive environmental samples collected after chilling at PO in abattoir 2 supports the hypothesis that the contaminated carcasses act as a primary source of cross-contamination (25). The five of eight Salmonella positivity for E5 at PO and a 45% Salmonella positivity identified for carcasses sampled at C4 in abattoir 1 further support this hypothesis. These observations emphasize the importance of controlling Salmonella at the live production stage, before broilers reach the slaughter plant. The Salmonella Heidelberg 1-1-1 HRM profile distribution among critical steps of the slaughter process further supports the hypothesis that the incoming birds act as a primary source of Salmonella contamination for carcasses of a same flock and for subsequently slaughtered flocks through contamination of the processing equipment.
Notably, previous studies have shown that Salmonella can persist in broiler processing environments despite intensive cleaning procedures and disinfection of contaminated equipment and surfaces, an observation made for both plants investigated in the current study (47). Indeed, HRM profiles of Salmonella serotypes Heidelberg, Kentucky, Schwarzengrund, and Thompson that were not found on sampled carcasses were recovered from both the plucker and the eviscerating machine (Table S1). Similarly, other HRM profiles were observed in the environment of processing plants at PO, whereas these profiles were not observed from the carcasses. These observations suggest that some Salmonella would enter the slaughter plant along with the incoming birds, but owing either to their lower numbers or to their differential ability to survive the slaughter process, would not contaminate the final meat product. Although the few HRM profiles that were found on carcasses sampled at C4/C5 were also present at C1, 14 other profiles persisted from C1 to C2, before not being detectable from the carcasses sampled at C3, C4, and C5 (Table S1). However, because only one carcass per flock was sampled for each of the critical steps selected, with a maximum of two distinct colonies kept for molecular typing, it is likely that the full diversity of Salmonella profiles present was not revealed (13). The laboratory protocol used has allowed showing the presence of two distinct Salmonella HRM profiles on 22 and 28% of the Salmonella-positive carcasses sampled at C1 for abattoir 1 and 2, respectively. When adopting a similar isolation protocol combined with a serotyping and pulsed-field gel electrophoresis approach on one Salmonella strain per positive sample, Vinueza-Burgos et al. (53) showed that two or three distinct sero-genotypes could be recovered from a same flock with the cecal sampling of 25 chickens per flock at the slaughter plant level (53). However, Cox et al. (14) recently showed that broiler chicken carcasses sampled at prechill can harbor an average of five distinct Salmonella serotypes and that a preenrichment step of the whole carcass rinsate volume yields even more diversity.
Salmonella strains can evolve through the acquisition of genetic elements to survive hostile environments created by the use of antibiotics, acid-based alternatives, carcass sanitizers, high and low temperatures, and the pH and osmolarity variations that are encountered in poultry facilities and processing plants (1, 13, 35, 41). A repeated exposure of some Salmonella strains to various sublethal stresses seems to increase the tolerance and survival ability of these strains, without neglecting the ability of some serotypes of Salmonella to preferentially adhere to chicken skin (20, 50). These observations highlight the importance for poultry processors to optimize the control of Salmonella cross-contamination along the slaughter line, but more importantly to the prevention of this contamination at the farm level.
In conclusion, our results reinforce the importance of the Salmonella status of the incoming birds on the contamination of the final product. They also illustrate the importance of both the water-immersion and dry-air chilling steps in the control of Salmonella on broiler carcasses and underline the importance of the design and features of the water-immersion chilling system in the final carcass Salmonella status. The HRM approach allowed for the description of the distribution of Salmonella in commercial poultry processing plants. This approach also allowed for the identification of some Salmonella types that persist through the slaughter process, an observation that would deserve particular attention. This study highlights the importance of better documenting the survival and dissemination ability of some specific Salmonella types at the slaughter plant level to optimize the microbiological quality of poultry meat products.
This study was co-supported by an industry funding from Olymel S.E.C./L.P. and by a collaborative and research grant from the Natural Sciences and Engineering Research Council of Canada (494530-2016). The authors thank Nicole Trottier, Saoussen Sfaxi, Micaela Miyauchi, Dominique Grohman, and Alexandre Quessy for technical assistance. A special thanks to food safety quality assurance personnel of the two participating slaughter plants. Olymel S.E.C./L.P. partially funded this project and contributed to the sampling design and results interpretation but was not involved in laboratory analyses and statistical analyses.
Supplemental material associated with this article can be found online at: https://doi.org/10.4315/JFP-20-250.s1