Microbiological characteristics of smoked and smoked–dried fish processed in Benin

Abstract This study aimed to assess the microbiological status of smoked fish (SF) and smoked–dried fish (SDF) processed in Benin, and to identify the contamination factors associated with these products. A total of 66 fish samples, including fresh fish and processed fish, were randomly collected from different processing sites and markets for microbial characterization using standard methods. The aerobic mesophilic bacteria (AMB) density varied from 2.9 to 9.5 Log10 CFU/g. Enterobacteriaceae, Escherichia coli, Bacillus cereus, Clostridium perfringens, yeasts, and molds were present in 63.9%, 27.8%, 55.6%, 58.3%, 61.1%, and 77.8% of samples, respectively, while no Salmonella spp., Listeria monocytogenes, and Staphylococcus aureus were found. The majority (66.7%) of SF samples and 22.2% of SDF samples were not compliant with the acceptable limit of <7.0 Log10 CFU/g recommended by the Health Protection Agency for AMB, whereas the Enterobacteriaceae counts exceeded the recommended level of 4.0 Log10 CFU/g for 50% of SF and 5.6% of SDF samples. Likewise, 38.9% of SF samples were not compliant for E. coli. Microbiological hazard analysis of practices allowed to identify the sensitive steps where hygiene measures need to be emphasized for an improved quality control.

Smoking consists in submitting fish to direct or indirect action of smoke during the incomplete combustion of certain trees used as fuel. Smoking of foodstuffs improves food organoleptic characteristics, induces water loss, and reduces the microbial load, thanks to heat and the presence of aromatic and bactericidal substances (Chakroborty & Chakraborty, 2017;Yusuf et al., 2015).
In Benin, fish products are the most important source of animal proteins (Kpodékon et al., 2014). Traditional smoking, one of the main methods used for fish preservation in the country (Dégnon et al., 2013), generates two types of end products, smoked fish (SF) and smoked-dried fish (SDF), used for local consumption or exported to neighboring countries.

| Follow-up of the manufacturing processing
The follow-up of the processing was performed on two fish species: S. scombrus (smoking) and C. cyanopterus (smoking-drying) identified as the most used by processors, according to a previous survey (data not shown). Twelve trials were performed with three experienced processors, each performing two smoking and two smoking-drying trials. Three types of samples were collected at sensitive steps: raw fresh fish, cleaned raw fresh fish, and SF (or SDF).

| Enumeration of the smoked and smoked-dried fish microflora
For each sample, 25 g was suspended in 225 ml of buffered peptone water (Bio-Rad, pH 7.0 ± 0.2) and homogenized (230 rpm for 2 min) using a stomacher (Lab Blender; Seward Medical) to obtain a 1/10 dilution. Decimal dilutions were prepared in BPW as described by ISO were investigated on Rapid'Salmonella Agar (Bio-Rad) after enrichment in buffered peptone water with addition of active supplement (capsules Bio-Rad). Confirmation was performed using Salmonella Latex Kit (Bio-Rad) for agglutination test according to the validated method BRD 07/11-12/05.
TA B L E 1 Microbial density (Log 10 CFU/g) in SF and SDF samples (n = 36)

Min.
Max. PCR products were purified using purification kit (GenElute™ PCR Clean-Up; Sigma-Aldrich). Identification of the bacteria was done by comparing their 16S rRNA sequence with those in databases using www.ncbi.nlm.nih.gov/BLAST.

| Data analysis
Statistical analyses were performed using STATISTICA 7.1. The analysis of data was performed with Student's t test, Mann-Whitney U test, one-way analysis of variance (ANOVA), and Kruskal-Wallis ANOVA. Significant difference was established at p < 0.05, and means were separated using Student, Newman, and Keuls range test. were not detected.

| Microbiological characteristics of SF and SDF samples
The minimum and maximum values recorded for each criterion showed important variability within samples of each type of product. This variation can be explained by the fact that the samples were collected from various processors and sellers where the quality of the raw material varied, as well as handling and hygiene practices.
Also, the density of AMB, Enterobacteriaceae, and E. coli was significantly higher (p < 0.05) in SF than in SDF (Table 1). This is probably due to the fact that SF samples have a higher moisture content (61 ± 11%) than the SDF ones (24 ± 11%). Also, in daily practices, SF are often more handled than SDF both by processors and customers.   also poses a risk, since they may produce mycotoxins during longterm storage (Job et al., 2016;Wogu & Iyayi, 2011). Table 2, no significant (p > 0.05) differences were observed between the microbial counts of samples from processing sites and those collected from markets regarding AMB, Enterobacteriaceae, B. cereus, C. perfringens, and yeasts in the case of SF. However, E. coli counts in SF samples collected from processing sites were significantly (p < 0.05) higher than those from markets.

As indicated in
This could be explained by the additional smoking of leftover products intended to extend their shelf life. Likewise, mold counts in SF samples collected from markets were significantly higher (p < 0.05) than those from processing sites. Klebsiella pneumoniae is known to possess histidine decarboxylase activity, enabling the bacterium to produce histamine in fish products (Visciano, Schirone, Tofalo, & Suzzi, 2012), which causes various health disorders to humans (Maintz & Novak, 2007). L. garvieae is found in aquatic environments (marine and freshwater aquaculture) and is a pathogen for fish (Vendrell et al., 2006). Wang et al. (2007) reported that it can be pathogenic for human with gastrointestinal disorder. E. faecalis has also been reported to cause endocarditis and diverse infections. Its transmission is nosocomial, but can be also done by food (Oprea & Zervos, 2007). Thus, beside the conventional microorganisms investigated for assessing the safety of ready-to-eat foods, SF and SDF samples also contained other potential pathogenic microorganisms exposing consumers to foodborne diseases. However, some of these bacteria can have positive effects. For instance, P. acidilactici produces a bacteriocin (Bacteriocin PA-1 or Pediocin AcH), which has an inhibitory effect on L. monocytogenes (Nieto-Lozano et al., 2010), and W. paramesenteroides secretes a bacteriocin with a broad spectrum of inhibition of spoilage bacteria and food pathogens such as Salmonella typhimurium, Vibrio parahaemolyticus, or L. monocytogenes (Pal & Ramana, 2010). Table 3 shows the evolution of the microbial loads during the processing of SF. AMB load in the raw frozen fish decreased significantly (p < 0.05) after the washing step (from 7.1 ± 0.5 to 6.2 ± 0.2), but remained stable after the smoking step (6.5 ± 0.4), close to the acceptable limit (<7 Log 10 CFU/g). This is surprising since a significant reduction in the microbial load of the fish was expected after the heat treatment. In addition, there were no significant (p > 0.05) changes in microbial loads for Hot smoking as carried out in traditional processing units can induce a reduction in the microbial contamination comparable to pasteurization (Plahar, Nerquaye-Tetteh, & Annan, 1999). During the follow-up trials, temperature values recorded in the core of fish remained above 70°C during the last 30 min of the average duration of 90 min of smoking (data not shown). Since this temperature is expected to reduce the microbial load, the hypothesis of recontamination of the product during and after smoking is therefore likely. Indeed, the follow-up trials revealed many practices that could contribute to fish recontamination: (a) processors do not wear appropriate clothes during processing (no clean apron and charlotte) (b) the processing is performed in unhygienic

Fresh fish (n = 3)
Cleaned fresh fish (n = 3) SF (n = 6) SDF (n = 6) AMB 7.4 ± 0.8 †a 6.5 ± 0.3 a,b 6.0 ± 1.2 a 5. TA B L E 4 Microbial load (Log 10 CFU/g) during the processing of smoked-dried fish (SDF) Cypselurus cyanopterus environment where the product is exposed to dust and flies, (c) processors use wastewater from raw fish washing to cool their hands, and (d) cement paper or frozen fish wrap is used to cover the end products. Table 4 shows the evolution of the microbial loads during the processing of SDF. AMB and LAB counts decreased significantly (p < 0.05) at the end of the smoking-drying period (from 7.4 ± 0.8 to 5.1 ± 1.1 and 6.1 ± 2.0 to 3.9 ± 1.1 Log 10 CFU/g, respectively). Moreover, Enterobacteriaceae were not detected at the end of the smoking-drying period. However, B. cereus, yeasts, and molds counts were not reduced significantly (p > 0.05), which could be explained by a recontamination of the product by these microorganisms during postprocess handling.
As for SF, potential pathogenic bacteria such as L. monocytogenes, Salmonella spp., E. coli, and C. perfringens were not detected along the process except S. aureus, which was detected in low amount in fresh fish (1.0 Log 10 CFU/g) and eliminated after the cleaning step.

| CON CLUS ION
This study revealed that SF and SDF processed in Benin are not always of satisfactory microbiological status and represent potential sources of foodborne diseases. The quality of raw material, poor hygiene practices, and inappropriate handling practices during processing and selling is factors that contribute to the unsatisfactory microbiological quality of these products.
Processors and sellers should be trained on good hygiene and handling practices in order to produce a safe and sound product for consumption.

ACK N OWLED G EM ENTS
This research was funded by ARES-CCD (Académie de Recherche et d'Enseignement Supérieur, Commission de la Coopération au Développement) through the QualiSani Project (PRD-2015). The authors are grateful for this financial support.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest to declare.

E TH I C A L R E V I E W
This study does not involve any human or animal testing.

I N FO R M E D CO N S E NT
Written informed consent was obtained from all study participants.