Sampling method influences Salmonella detection and quantification in pre-harvest commercial broiler production

Abstract

Effective pre-harvest Salmonella monitoring in broilers relies on accurate, reliable, and reproducible evaluation of pre-harvest Salmonella. In this study, six sampling methods were evaluated and compared to assess Salmonella prevalence and quantification during broiler production across three iterative experiments. In experiment one, bootsocks, electrostatic pad-rollers, feather swabs, cloacal swabs, fecal grabs, and litter grabs were collected from 24 houses across 10 farms (n = 288 samples). In the second experiment, bootsocks, bootsock-rollers, and feather swabs were collected in 16 houses on seven farms (n = 128). Bootsocks and bootsock-rollers were selected as the most reproducible sampling method. In experiment three both methods were performed in triplicate in 20 houses on 10 farms (n = 240). In all experiments, prevalence was determined by qPCR and by culture, then compared by Fisher's Exact test between and McNemar's test within methods. Salmonella was quantified by qPCR and Ct-values were compared using one-sided F-test. In experiment one, prevalence differed between methods by qPCR (p = 0.0150) only and the best performing sampling methods were bootsocks (42/48 culture and 41/48 qPCR positive), feather swabs (42/48 and 36/48), and electrostatic pad-rollers (35/48 and 34/48). In experiment 2, feather swabs differed by qPCR prevalence (p = 0.0004). Bootsocks (30/32 culture and 28/32 qPCR positive) and bootsock-rollers (31/32 and 32/32) performed best. In experiment 3, qPCR prevalence (210/240) was greater than culture (167/240) (p = 0.0021), but no differences were observed between methods or replicates. The average Ct-value for bootsocks and bootsock-rollers were 37.8 and 38.9, respectively and there was no difference in their variance (p = 0.8061). A linear mixed-effect model found that farm contributed 36.34 % of the variance observed while house, house-side, and replicate accounted for 24.69 %, 4.68 % and 3.85 %, respectively. This study shows that sampling methods directly influence both Salmonella detection and load recovery. For surveillance sampling, bootsocks and bootsock-rollers were found to best indicate pre-harvest Salmonella. These two methods were highly reproducible, user friendly, and provide the most reliable Salmonella results indicating the Salmonella prevalence in broiler flocks during production.

Keywords: Broiler; Pre-harvest; Salmonella; Surveillance sampling.

Fig. 1

Salmonella prevalence determined by culture and qPCR for the first experiment. In the first experiment, bootsocks, cloacal swabs, ESP-rollers, feather swabs, fecal grabs, and litter grabs were used to sample 24 houses on 10 farms. For each sampling method there were n = 48 samples. *: Significant differences were established at p< 0.05 for Fisher´s Exact test and Chi-square McNemar test. Bars with different alphabets (qPCR prevalence) show differences across sampling methodologies by the Fisher's Exact test. There were no differences between sample types by culture. McNemar test p-values (bottom) compare culture and qPCR outcomes within the same sampling method.

Fig. 2

Salmonella prevalence determined by culture and qPCR for the second experiment. In the second experiment, bootsocks, bootsock-rollers, and feather swabs were used to sample 16 houses on seven farms. For each sampling method there were n = 32 for Bootsocks and Bootsocks-rollers, and n = 64 for Feather swabs. *: Significant differences were established at p< 0.05 for Fisher´s Exact test and McNemar test. Bars with different alphabets (qPCR prevalence) show differences across sampling methodologies by the Fisher's Exact test. There were no differences between sample types by culture. McNemar test p-values (bottom) compare culture and qPCR outcomes within the same sampling method.

Fig. 3

Salmonella prevalence determined by culture and qPCR for the third experiment. In the third experiment, bootsocks and bootsock-rollerswere used to sample 20 houses on 10 farms in triplicate. For each sampling method there were Salmonella prevalence screened by culture and qPCR for the third experiment, n = 40 for Bootsocks and Bootsocks-roller per replicate. *: Significant differences were established at p< 0.05 for Fisher´s Exact test and McNemar test. McNemar test p-values (bottom) compare culture and qPCR outcomes within the same sampling method.

Fig. 4

Salmonella load in experiment three samples. Salmonella was quantified from bootsock and bootsock-roller samples using a commercial qPCR assay in the third experiment, and statistical comparisons were made using one-sided F-tests (p = 0.8061) for bootsock (red) and bootsock-roller (grey) samples. The average variance and Ct value with standard error for each methodology is provided in the bottom left and the variances with p-values derived from one-sided F-tests for each house along the bottom. No significant differences were established at p< 0.05 between sampling methodologies in each house.