Contamination of salmon fillets and processing plants with spoilage bacteria

International Journal of Food Microbiology 237 (2016) 98–108

Contents lists available at ScienceDirect

International Journal of Food Microbiology journal homepage: www.elsevier.com/locate/ijfoodmicro

Contamination of salmon fillets and processing plants with spoilage bacteria Trond Møretrø ⁎, Birgitte Moen, Even Heir, Anlaug Å. Hansen, Solveig Langsrud Nofima, The Norwegian Institute of Food, Fishery and Aquaculture Research, Aas, Norway

a r t i c l e

i n f o

Article history: Received 27 January 2016 Received in revised form 10 August 2016 Accepted 11 August 2016 Available online 12 August 2016 Keywords: Salmon fillets Cross-contamination Spoilage bacteria Pseudomonas Shewanella Photobacterium

a b s t r a c t The processing environment of salmon processing plants represents a potential major source of bacteria causing spoilage of fresh salmon. In this study, we have identified major contamination routes of important spoilage associated species within the genera Pseudomonas, Shewanella and Photobacterium in pre-rigor processing of salmon. Bacterial counts and culture-independent 16S rRNA gene analysis on salmon fillet from seven processing plants showed higher levels of Pseudomonas spp. and Shewanella spp. in industrially processed fillets compared to salmon processed under strict hygienic conditions. Higher levels of Pseudomonas spp. and Shewanella spp. were found on fillets produced early on the production day compared to later processed fillets. The levels of Photobacterium spp. were not dependent on the processing method or time of processing. In follow-up studies of two plants, bacterial isolates (n = 2101) from the in-plant processing environments (sanitized equipment/ machines and seawater) and from salmon collected at different sites in the production were identified by partial 16S rRNA gene sequencing. Pseudomonas spp. dominated in equipment/machines after sanitation with 72 and 91% of samples from the two plants being Pseudomonas-positive. The phylogenetic analyses, based on partial 16S rRNA gene sequencing, showed 48 unique sequence profiles of Pseudomonas of which two were dominant. Only six profiles were found on both machines and in fillets in both plants. Shewanella spp. were found on machines after sanitation in the slaughter department while Photobacterium spp. were not detected after sanitation in any parts of the plants. Shewanella spp. and Photobacterium spp. were found on salmon in the slaughter departments. Shewanella was frequently present in seawater tanks used for bleeding/short term storage. In conclusion, this study provides new knowledge on the processing environment as a source of contamination of salmon fillets with Pseudomonas spp. and Shewanella spp., while Photobacterium spp. most likely originate from the live fish and seawater. The study show that strict hygiene during processing is a prerequisite for optimal shelf life of salmon fillets and that about 90% reductions in the initial levels of bacteria on salmon fillets can be obtained using optimal hygienic conditions. © 2016 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction There is an increasing consumer demand for fresh, chilled fish. This is a result of increased consumption of fish eaten raw (e.g. sushi, sashimi) and a general increased consumer demand of fresh food (CBI Market Intelligence, 2016; Quested et al., 2010). Fresh fish has significant added value compared to frozen fish, but also requires increased attention to the sensory and microbial quality (Gram and Huss, 1996). Farmed salmon is a high volume product in the fresh, chilled fish product category. In Norway, the production of farmed Atlantic salmon in 2014 was about one million metric ton with an export value of about five thousand million Euro (Norwegian Seafood Council, 2015).

⁎ Corresponding author at: Nofima, The Norwegian Institute of Food, Fishery and Aquaculture Research, P.O. Box 210, N-1431 Aas, Norway. E-mail address: trond.moretro@nofima.no (T. Møretrø).

Microbial control during processing and storage is a key factor that determines the quality and shelf life of fresh fish. Bacteria on the product can originate from the raw materials or be introduced during processing by e.g. cross contamination from equipment or by food handlers. The microbial quality of the product is depending on the spoilage potential of the microorganisms present and the storage conditions that affect growth and formation of spoilage metabolites (Gram and Huss, 1996). The most commonly reported spoilage bacteria for aerobically stored chilled fish including salmon are species within the genera Pseudomonas (P.) and Shewanella (S.), while the CO2-resistant Photobacterium (Ph.) phosphoreum dominates on fish packed under modified atmosphere (Chaillou et al., 2015; Dalgaard et al., 1993; Emborg et al., 2002; Gram and Huss, 1996; Parlapani and Boziaris, 2016; Tryfinopoulou et al., 2002). Ph. phosphoreum is a producer of trimethylamine (TMA), a major spoilage product in fish (Dalgaard, 1995). The most important spoilage products of Shewanella spp. are volatile sulfides, but TMA may

http://dx.doi.org/10.1016/j.ijfoodmicro.2016.08.016 0168-1605/© 2016 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

T. Møretrø et al. / International Journal of Food Microbiology 237 (2016) 98–108

also be produced (Dalgaard, 1995; Joffraud et al., 2001). Pseudomonas spp. does not produce TMA but has been associated with quality changes and development of sweet, fruity off-odors in various species of chilled fish (Olafsdottir et al., 2006; Parlapani et al., 2015). Live salmon can harbour Pseudomonas spp., Shewanella spp. and Photobacterium spp. and can thus be considered an important primary source for these spoilage organisms on processed salmon (Cantas et al., 2011; Gram and Huss, 2000; Hovda et al., 2012; Navarrete et al., 2009). Although good hygienic practices are considered essential in all production of food, little is known about the importance of bacterial cross contamination from product contact surfaces to the fish during processing. Bagge-Ravn et al. (2003) studied the bacterial microbiota on equipment in four fish processing plants, including two smokehouses producing cold smoked salmon and found that Pseudomonas spp. and yeasts, followed by Acinetobacter and Neisseriaceae dominated after cleaning and disinfection. Photobacterium spp. was found at low prevalence in one of the two smokehouses after cleaning and disinfection. Pseudomonas is frequently isolated after cleaning and disinfection in other types of food industry, e.g. meat and dairy processing plants (Brightwell et al., 2006; Hultman et al., 2015; Mettler and Carpentier, 1998; Møretrø et al., 2013; Stellato et al., 2015). To our knowledge the prevalence of Shewanella spp. in fish processing plants has not been reported. The aim of this study was to identify the main sources of spoilage bacteria in salmon fillets. The microbiota of industrially processed salmon fillets and fillets processed by manual filleting under strict hygienic conditions from seven different processing plants were compared. Furthermore, the prevalence of spoilage bacteria along the processing line in two salmon processing plants was determined to detect high-risk sites for contamination from machines/equipment to products. 2. Materials and methods 2.1. Salmon fillets from Norwegian salmon processing plants Fillets were collected in June–September 2012 from seven Norwegian plants with pre-rigor processing of farmed Atlantic salmon. This was done to determine bacterial levels in pre-rigor processed salmon. Similar procedures for collection of the fillet samples were performed in each plant: Twice during a production day, four to six fillets were collected at the end of the processing lines. The first collection was performed early at the production day (within 1 h after start) and the second after mid shift (5–6 h after production start). In addition, from each salmon processor three salmon were collected prior to processing. These salmon were never in contact with processing plant surfaces. They were manually gutted under high hygienic conditions using clean knifes and cutting boards to avoid microbial crosscontamination. These salmon were used as controls representing salmon with optimal hygienic status. Fillets and gutted salmon (controls) were packed on ice in separate boxes (expanded polystyrene; EPS) and sent express to the laboratory. Within 24 h after reception in the lab, the gutted salmon were filleted and skinned under hygienic conditions to avoid cross-contamination. The in-plant produced fillets were skinned likewise. All fillets were stored on ice during the experiment (temperature b 1 °C, confirmed by temperature logging). 2.1.1. Culture-dependent bacterial analyses of salmon fillets Microbial sampling of salmon fillets (early, mid shift and control samples) from the seven processing plants was performed at Day 1–2 and Day 10. From each fillet, a sample of 3 × 3 cm × 0.5 cm depth (approximately 10 g) was diluted with peptone water (saline with 0.1% Bacto Peptone (Oxoid)) to obtain a 1/10 dilution. Bacterial quantification on homogenized suspensions after stomaching for 60 s was determined by cultivation on Long & Hammer agar (van Spreekens, 1974) incubated at 15 °C for 5–6 days. A total of 186 colonies, representing different plants, processing conditions and storage times were picked and

99

identified by partial 16S rRNA gene analysis. The picked colonies were resuspended in 50 μl of Tris-EDTA buffer in a microtiter plate well, followed by heat treatment at 99 °C for 10 min. After centrifugation at 4500 ×g for 3 min, 30 μl supernatant was transferred to a new microtiter plate, which was frozen at −20 °C until further analysis. For amplification of 16S rRNA gene, the supernatant was thawed and 1 μl used as template in a PCR reaction. Briefly, universal primers (Nadkarni et al., 2002) were used for PCR and sequencing. Amplification was performed using 0.25 μM of each primer, 10 μl Qiagen multipleks PCR kit (2 ×) (Qiagen) to a total volume of 20 μl. The cycling conditions, PCR purification and sequencing were performed as described (Omer et al., 2015). Genus was determined by search of approximately 400 bp in Ribosomal Database Project (RDP) (https://rdp.cme.msu.edu/seqmatch/ seqmatch_intro.jsp). 2.1.2. Culture-independent bacterial analyses of salmon fillets Culture-independent bacterial analyses of salmon fillet samples were done using Next generation sequencing (NGS) of the variable region 4 of the 16S rRNA gene. Twenty one samples from Day 10 fillets (early, mid shift and controls from seven plants) were prepared. For each sample, 8–12 ml of the salmon-peptone water stomaching suspensions from each of the four to six parallel fillets were combined. For control samples, 16 ml of three parallels were combined. Samples from Day 1 were not prepared due to low total counts. The suspensions were filtrated (20 μm Steriflip (Merck Millipore, Darmstadt, Germany) to remove salmon debris, aliquots of 4.5 ml were centrifuged (13,000 × g, 5 min), and pellets stored at −20 °C until DNA purification. The pellets were resuspended in 500 μl 2× Tris-EDTA (20 mM Tris-Cl, pH 8.0/2 mM EDTA)/1.2% Triton X-100 (Sigma Aldrich, St. Louis, USA), transferred to a FastPrep tube (Matrix B, MP Biomedicals, Solon, USA) and lysed in a FastPrep bead beater (MP Biomedicals) for 40 s at 6 m/s. The Fastprep tube was centrifuged for 5 min at 14,000 ×g, 360 μl of the supernatant was added 50 μl Proteinase K and 400 μl lysis buffer AL (DNeasy Blood and Tissue Kit, Qiagen, Valencia, CA), mixed and incubated for 30 min at 56 °C. 400 μl EtOH was added, mixed and transferred in two steps to a Qiagen column (DNeasy Blood and Tissue Kit, Qiagen). The manufacturer's protocol was followed from here. DNA was used as template for the NGS (MiSeq, Illumina Inc., San Diego, USA) analysis as previously described (Moen et al., 2015). Briefly, a portion of the 16S rRNA gene spanning the variable region 4 (V4) was amplified using the barcoded, universal primer set (515F/806R) (Caporaso et al., 2012). Of the 21 samples, two samples (F control and G control) were excluded due to low PCR product concentration. The library quantification and sequencing were performed at the Norwegian Sequencing Centre (https://www. sequencing.uio.no/). The pre-processing of the data was performed in QIIME (Quantitative Insights Into Microbial Ecology (version 1.6.0)) (Caporaso et al., 2010). The sequences were then demultiplexed in QIIME allowing zero barcode errors and a quality score of 20 (Q20). To remove short sequences (identified as Salmo salar mitochondrion), the minimum number of consecutive high quality base calls to include a read as a fraction of the input read length was increased from 0.75 to 0.9. Reads were assigned to their respective bacterial id using de novo operational taxonomic unit (OTU) picking workflow in QIIMEReads that did not match a reference sequence were discarded. In total 15,272 OTUs were written (5277 OTUs when not including singletons), each of these represents a phylotype and may be a representative of a bacterial species. The level 6 (genus level) table was used in further analysis. Only the dominating genera were represented (the other genera were represented as “other”). 2.2. Sampling of spoilage bacteria in two salmon processing plants The prevalence and contamination sources of common spoilage associated bacteria of fresh salmon were further investigated in two salmon processing plants (B and H). The two plants, both processing prerigor salmon, were visited in March (plant B) and November (plant H

100

T. Møretrø et al. / International Journal of Food Microbiology 237 (2016) 98–108

and second visit plant B) 2014, and sampled. Both processing plants were divided in a slaughter and a filleting department. The production steps were generally as follows: pumping of salmon from storage nets in ocean into the plant, stunning, desliming (plant B only), cutting of gills, bleeding, gutting, head cutting, filleting, trimming, skinning and packaging. At plant H some of the fillets were stored on ice for 2– 5 days before removing bones at a separate post rigor line. Both processing plants had a bleeding tank with refrigerated (0–3 °C), UV-treated sea water (RSW). Plant H had two other tanks with RSW (0–3 °C) used for fish storage during a production day. Operating procedures in both plants included salmon processing for 14–16 h followed by 6–8 h cleaning and disinfection, 5 days a week. The conveyor belts were cleaned when running, with dismantling and cleaning in weekends for plant B and less frequently for plant H. Bleeding tanks and the associated RSW systems were cleaned 1–2 times a week, while the additional storage tanks of plant H were cleaned daily and the associated RSW systems twice a week. Both plants used chloralkali based cleaning agents. Plant B alternated between quaternary ammonium compounds and peracetic acid for disinfection while plant H used peracetic acid. Machines, equipment and the processing environment were sampled after cleaning and disinfection and before start of production with sterile swabs (Mesoft, Mölnycke Health Care AB, Gothenburg, Sweden). Typically an area of 100–300 cm2 was swabbed. Sampling sites were product contact surfaces from equipment, machines and conveyor belts (Table 1). Salmon from 3 to 4 different locations in the slaughtering department were swabbed during production. Swabs from three salmon were combined to a pooled sample. Before gutting, skin and gills were swabbed, while after gutting also the stomach cavity was swabbed. Swabs were transferred directly after sampling to tubes with 15 ml Dey Engley neutralizing broth (Difco). After mixing, the suspension was serially diluted in peptone water,plated to Iron agar (Oxoid, Basingstoke, UK) and incubated at 15 °C. Water samples were retrieved from seawater pumped in together with salmon, water leading salmon to the stunner unit, bleeding tank, desliming (plant B) as well as two other storage tanks for plant H. Samples (20–50 ml) were filtered (0.45 μm, Microcheck II beverage monitor, Pall corporation, Ann Arbor, MI, USA), and the filter placed on Iron agar. With exception of bleeding water, 100 μl was also plated directly or after dilution to Iron agar. Fillets were removed directly from the production line in plant H after filleting, after trimming and at packaging. For plant B, fillets were not sampled in the main sampling (sampling 1), but during an additional sampling (sampling 2) of selected sites (five samples of equipment/ machines, four swab samples from salmon (pooled from three salmon) in the slaughter department, fillets from four processing steps stored on ice for 14 days) performed eight months after sampling 1. In sampling 2 from plant B also fillets after skinning were included. Fillets from the production line were wrapped in plastic, put on ice and sent to the lab for sampling and plating to Iron agar within 24 h. The remaining fillets

were wrapped in plastic and stored on ice (b 1 °C, confirmed by temperature logging). Samples were plated on Iron agar after 14 days. 2.2.1. Identification of bacteria from processing plants B and H From Iron agar, up to 20 colonies were picked randomly and identified to the genus level by partial 16S rRNA gene sequencing (variable region 3–4) as described above. The genus to which the highest number of colonies from a sample was identified was regarded as dominating in that sample. Sulfide producing bacteria were identified by picking up to three black colonies of each visually distinguishable type. A total number of 2101 colonies were identified. From the bleeding tank from plant B, colonies were not identified due to overgrown agar plates. Sequences for phylogenetic analyses were accessed with Sequence Scanner Software 2 (Applied Biosystems). Sequences of high quality (quality score N 50 and manually verified sequences with quality scores b50) were imported to CLC Main Workbench ((QIAGEN Aarhus A/S, Denmark) and used in the phylogenetic analysis. A separate alignment of unique sequences was made for each of the genera Pseudomonas, Photobacterium and Shewanella. For Pseudomonas and Shewanella, neighbor joining based cladograms were made and sequences of type strains of species most closely related to the strains from the processing plants as revealed by RDP search were retrieved from the NCBI database (http://www.ncbi.nlm.nih.gov/) and included in the alignment and in the cladograms. For Shewanella all unique sequence profiles were included in the cladogram, while for Pseudomonas only unique sequence profiles found in at least five samples in the processing plants were included. To distinguish between closely related Photobacterium, the housekeeping gene gyrB was sequenced for four isolates, selected to cover the diversity revealed by sequencing of the 16S rRNA gene, as well as including isolates from both processing plants. DNA from three colonies from each isolate was extracted as described above and amplified (using the 5 PRIME HotMasterMix) with the primers (22f (5′-GAAGTT ATCATGACGGTACTTC-3′) and 1240r (5′-AGCGTACGAATGTGAGAACC3′)) and amplification protocol described previously (Kaeding et al., 2007). Each PCR reaction was performed in duplicate and sequenced (Omer et al., 2015) using both the forward and reverse primer and the thermal profile in Kaeding et al. (2007). The sequences were aligned, edited and consensus sequences for each isolate were constructed using the CLC Main Workbench (QIAGEN Aarhus A/S). 2.3. Calculations and statistics Mean values and standard error of the means were calculated using log10 (log) transformed values of cell concentrations for each fillet. The statistical significance of fillet preparation (control, mid-day and early) was calculated using the general linear model and Tukey's test in Minitab v16 (Minitab Ltd., Coventry, UK) and using the differences between plants as error. The statistical significance within each production plant was tested using two-sample t-tests on log transformed cell concentrations.

Table 1 Prevalence of spoilage associated bacterial genera as percent positive of different types of samples. Genera

Plant Ba

Plant H b

Pseudomonas Shewanella Photobacterium a

c

d

e

Salmon swabs (3 )

Water (3)

Equipment (39)

Total (45)

Salmon swabs (4)

Water (5)

Equipment (45)

Total (54)

33f 67 67

67 67 0

62 26 0

61 29 4

25 100 75

80 100 40

49 18 0

51 33 9

Data shown from plant B are from sampling 1. Pooled sample of three salmon from the slaughter department, swabbed after stunning, after desliming, before gutting and after gutting. Number of samples. d Seawater pumped into plant, seawater stunner/desliming, bleeding tank, storage tanks. e All equipment/machines (product contact) Examples: stunning machine, head cutter, gutting machine, vacuum systems, filleting machine, trimming machine, bone remover, conveyor systems. f Percent positive samples, e.g. Pseudomonas identified in 1 out of 3 (33%) pooled samples. b c

T. Møretrø et al. / International Journal of Food Microbiology 237 (2016) 98–108

Statistical significance of fillet preparation for each bacterium was determined by paired t-tests in Minitab. The cell concentration of each bacterial genus in the fillet samples was calculated by multiplying the total counts (log cfu/g) in the sample with the fraction of each bacterial genus determined by NGS analyses (see 2.1.2). The total counts used were the mean of the cell concentration of single fillets constituting the pooled sample.

7

stored for 10 days followed by Shewanella spp., Janthinobacterium spp., Photobacterium spp. and Acinetobacter spp. (See Appendix Fig. A1 for details). There were higher numbers of Pseudomonas spp. and Shewanella spp. in industrially compared to manually produced fillets (p b 0.01, Fig. 2). There was no significant difference in numbers of Photobacterium spp. between industrially and manually produced fillets. After both 1–2 days and 10 days storage, there were significantly higher levels of bacteria on fillets produced early at the production day compared to fillets produced later the same day (p b 0.01, Fig. 1). Higher contamination levels correlated (p b 0.05) with higher numbers of Pseudomonas spp. and Shewanella spp. after 10 days of storage on ice. The presence/level of Photobacterium spp. was independent of total bacterial counts (p N 0.05) (Fig. 1B and 2). Further analyses of the NGS data revealed two dominating OTUs representing the genus Pseudomonas (based on 254 bp 16S rRNA gene sequence). In addition to the NGS, colonies were picked and identified from agar plates from Day 10 ice stored fillets. Based on the partial 16S rRNA gene sequences (380 bp) from 143 colonies of Pseudomonas spp., the isolates formed three clusters (data not shown), two of them the same groups as for NGS, in addition to a third group. Isolates representing these three clusters (MF05002, MF05004 and MF06398) were closely related to P. gessardi, P. lundensis and P. fluorescens, respectively (Fig. 3). Similarly, the majority of the 16S rRNA gene sequences identified as Shewanella spp. were separated in two clusters. Representative isolates of these groups (MF05008 and MF05009) clustered with the type strains of S. putrefaciens and S. baltica (Fig. 4).

6

3.2. Spoilage bacteria in two salmon processing plants

5

As the results from fillets from seven producers indicated that spoilage bacteria were transferred to salmon during processing, the presence of such bacteria in two salmon processing plants (B and H) was further investigated. Bacteria were detected from 75% (detection limit −0.3 to 0.3 log cfu/cm2 depending on area sampled) of machines/equipment after cleaning. The highest bacterial levels (N3.0 log cfu/cm2) were found on worn or difficult accessible conveyors, filleting machines and prisms for positioning salmon in gutting machines. Higher bacterial levels were found in the slaughter departments than in filleting departments. On all sampled conveyors in the filleting department of plant H, bacterial levels were below the detection limit. The total bacterial levels in seawater pumped in together with fish and water in stunner were 1.0–1.8 log cfu/ml. Bacterial levels in other water samples (desliming plant B, bleeding tanks, storage tanks plant H) were in the range 2.7–3.7 log cfu/ml. Total counts in swab samples from salmon in the slaughter department varied considerably from 2 log–4 log cfu/salmon, and there was no clear pattern with increasing or decreasing levels along the processing line. The bacterial levels in fillets stored on ice for 14 days were about 6 log cfu/g for fillets from all different steps in production tested from both processing plants.

3. Results 3.1. Bacteria on hand- and industrially filleted salmon from seven processing plants The bacterial levels on industrially produced fillets from the seven processors were higher than on manually processed fillets (p b 0.01, Fig. 1). From all seven processors (A-G), the bacterial levels of the manually gutted and filleted salmon were below or near the detection limit (1.3 log cfu/g) after 1–2 days ice storage. The mean total count after 10 days storage on ice was 2.6 log cfu/g. These manually produced salmon were never in contact with equipment or machines of the plants. Industrially processed fillets had bacterial levels in the range 1.4–2.9 log cfu/g 1–2 days after production and 3.3–5.9 log cfu/g after 10 days ice storage. Overall, Pseudomonas spp. dominated in fillets

Total counts (log cfu/g filet)

A

4 Early

3

Mid-shift

2 1 0 A

B

C

D

E

F

G

Salmon processing plant

Total counts (log cfu/g filet)

B

101

7 6 5 4 Control 3

Early

2

Mid-Shift

1 0 A

B

C

D

E

F

G

Salmon processing plant Fig. 1. Bacterial counts in ice-stored salmon fillets from seven different processing plants. Data from control fillets produced by manual gutting and filleting under optimized hygienic conditions, from fillets industrially processed early (within 1 h after production start) and mid-shift (5–6 h after production start) are shown as white, black and grey bars, respectively. Means for 4–6 fillets with standard error of means are shown. Bacterial analyses were performed at day 1–2 after salmon slaughter and filleting (A) and after 10 days storage on ice (B). Total counts from hygienic control fillets were not included in A as these were below the detection limit (1.3 log cfu/g; dotted line) for all fillets except one (1.6 log cfu/g).

3.2.1. Qualitative bacterial analysis The 2101 bacterial colonies identified, belonged to 42 different genera, of which 21 genera were found in both plants. Gram-negative bacteria dominated (N 90%) in both plants. The dominating genera were Pseudomonas followed by Shewanella, Psychrobacter, Aeromonas, Photobacterium and Acinetobacter (see overview in Appendix Table A1). The prevalence of the spoilage associated bacteria Pseudomonas spp., Shewanella spp. and Photobacterium spp. in different sample types is described in detail below. 3.2.1.1. Pseudomonas. Pseudomonas spp. dominated in both processing plants, and was isolated from all sample types (Table 1). A qualitative overview of the prevalence of Pseudomonas spp. in the slaughter departments and filleting departments of the two processing plants is shown in Figs. 5 and 6, respectively. It was a tendency of higher prevalence of

102

T. Møretrø et al. / International Journal of Food Microbiology 237 (2016) 98–108

7 6

log cfu/g

5 4 3 2

A

B

C

Pseudomonas

D Shewanella

E

F

Mid-Shift

Early

Mid-Shift

Early

Mid-Shift

Early

Control

Mid-Shift

Early

Control

Mid-Shift

Early

Control

Mid-Shift

Early

Control

Mid-Shift

Early

0

Control

1

G

Photobacterium

Fig. 2. Levels of Pseudomonas spp., Shewanella spp. and Photobacterium spp. in fillets stored on ice for 10 days from processing plant A–G. Fillets were produced early and mid-shift at the production day in the processing plants. Control fillets (manually produced) were hygienically gutted and filleted by hand. The bacterial numbers are based on plate counts and relative proportion of the bacterial genera from the microbiota analysis.

Pseudomonas spp. on equipment/machines than on salmon in the slaughtering department. In swabs from equipment/machines with bacterial levels above the detection limit, Pseudomonas spp. were isolated from 91 and 72% of the sampling points in plant B and H, respectively. Pseudomonas spp. were detected on two out of seven pooled samples

from salmon in the slaughtering department. There were very low bacterial levels on fillets one day after production (b 2.0 log cfu/g). Pseudomonas spp. were detected in low levels (1.0–1.3 log cfu/g) on fillets from the two last processing steps for the second sampling in plant B. After 14 days storage on ice, Pseudomonas spp. dominated on

P. trivialis P33 P38 P42 P. veronii P. gessardii / MF05002 P46 P. libaniensis P45 P31 P. brennerii P30 P11 P13 P. fragi P16 P. lundensis / MF05004 P17 P. kilonensis P23 P. fluorescens / MF0638 P34 Fig. 3. Cladogram of Pseudomonas 16S rRNA gene sequences (393 bp) from salmon and salmon processing plants. P11–P46 represent the most common sequence profiles (found in at least 5 samples) of isolates from plant B and/or H, while MF05004, MF05002 and MF06398 are isolates representing the most commonly sequence profiles found in ice stored fillets from seven processing plants in an initial study. Sequences of type strains most closely related to the isolates from the processing plants are included. Symbols present to the right of P11–P46 indicate sources of Pseudomonas isolates with respective sequence profiles from plant B (squares) and H (triangles). One symbol represents one sample: ; Salmon from slaughterhouse, / ; water, ∎/▲ equipment/machines, / ; fillets.

T. Møretrø et al. / International Journal of Food Microbiology 237 (2016) 98–108

MF05008 S9 S. putrefaciens S8 S7 S. profunda S6 S.basaltis S5 S. gaetbuli S1 MF05009 S2 S3 S. baltica S4 S. glacialipiscicola S12 S13 S11 S. vesiculosa S10 S14 Fig. 4. Cladogram of Shewanella 16S rRNA gene sequences (394 bp). S1–S14 represent sequence profiles of isolates from processing plant B and H, while MF05008 and MF05009 are isolates representing the most common sequence profiles found in ice stored fillets from seven processing plants in the initial study. Sequences of type strains most closely related to the isolates from the processing plants are included. Symbols present to the right of S1-S14 indicate sources of Shewanella isolates with respective sequence profiles from plant B (squares) and H (triangles). One symbol represents one sample: / ; Salmon from slaughterhouse, / ; water, ∎/▲ equipment/machines, ; fillets.

fillets from all processing steps (filleting, trimming, skinning and packaging) from both plants at a concentration of about 6 log cfu/g. The diversity of the isolates within the Pseudomonas genus was high, with a total of 48 different profiles of the sequenced part of the 16S rRNA gene observed among the 466 isolates. Among these profiles, 25 were detected from both plant B and H. Nineteen of the sequence profiles were only found in a single sample while 14 were found in at least five samples. The two most dominating profiles (P45 and P46) were highly similar (separated by 2 out of 393 bp) and related to P. libaniensis and P. gessardii, respectively (Fig. 3). These two profiles were identified from 6 to 11 machine/equipment samples from both plants and also found in fillets from both plants. Also Pseudomonas spp. with four other sequence profiles were found both on equipment and fillets from both plants (Fig. 3). The sequences of Pseudomonas spp. isolates from plant B and H were compared to Pseudomonas spp. isolates representing the three dominating sequence profiles in fillets from the seven processing plants. All the three representative isolates from the seven processing plants (MF05002, MF05004 and MF06398) were 100% identical to sequence profiles (P46, P16 and P34, respectively) found on both equipment and fillets in processing plants B and H (Fig. 3). 3.2.1.2. Shewanella. The majority of bacteria forming black colonies on Iron agar were identified as Shewanella spp., followed by Aeromonas spp. Shewanella spp. were found on salmon in the slaughter department, seawater and refrigerated seawater from bleeding and storage tanks (Table 1 and Fig. 5). Shewanella spp. were present (about 5 log cfu/g) on ice-stored fillets from one out of four and two out of three processing steps for plant B and H, respectively (data not shown). Alignments of 16S rRNA gene sequences from 124 Shewanella isolates revealed 14 unique sequences that were compared phylogenetically. An alignment of partial 16S rRNA gene sequences (394 bp)

103

showed that most of the isolates from an individual sample had identical sequences, an exception was water tanks from plant H where isolates with different sequences were present. The sequence profiles were related to different species of Shewanella (Fig. 4). Of the 14 sequence profiles (S1–S14), four were found in both plant B and H (Fig. 4). The diversity among isolates from equipment, water samples and salmon from slaughter department were high with six and seven different profiles from equipment, two and four for water and two and six different profiles from salmon from slaughterhouse from plant B and H, respectively. The profile found most frequently in plant B was present on salmon from slaughterhouse and on a conveyor belt on both sampling dates. An isolate from fillet from plant H had the same sequence profile as isolates from equipment (water tank surface). The sequence profile (S2) dominant on most equipment samples from plant H (six different equipment), was not found in other sample types from plant H, but in a water sample from plant B (Fig. 4). Two isolates (MF05008 and MF05009) representing the two most common sequence profiles as revealed by NGS analysis of fillets from seven processing plants were 100% identical to isolates found on both equipment, raw salmon and water samples in plant B and H. These showed highest identity to type strains of S. putrefaciens and S. baltica, respectively (Fig. 4). 3.2.1.3. Photobacterium. Photobacterium spp. were found on salmon directly after stunning, and on salmon from other early processing steps at the slaughter departments in both processing plants (Table 1, Fig. 5). Also, the bacterium was isolated from seawater pumped in together with salmon and from the bleeding tank of plant H (Fig. 5). Photobacterium spp. were absent in the 89 swab samples from machines and equipment taken after cleaning and disinfection (Table 1), and was not found on fillets on the processing day or on fillets stored on ice for 14 days from any of the plants. The 36 isolates identified constituted only two 16S rRNA gene sequence profiles differing by a single nucleotide. Both profiles shared 100% identity to both Ph. phoshoreum and Ph. kishitanii strains after BLAST search (398 bp of the 16S rRNA gene). The isolates from salmon from the slaughter department in plant B, had different sequence profiles in March and in November, while all salmon isolates from plant H had identical sequences. Both sequence profiles were present in the bleeding tank of plant H. To give further insight into the species identity of the isolates, the housekeeping gene gyrB of four isolates, representing the two 16S rRNA gene profiles present, was sequenced. The results showed that the gyrB (1168 bp) of three of the isolates were almost identical (99.6%) with 99–100% identity to Ph. phosphoreum in the database. The last isolate (from salmon from plant B in November) was 100% identical to Ph. Iliopiscarium. 3.2.1.4. Other bacterial genera. Aeromonas spp. were present on 47% of conveyor belts in processing plant B, but not detected from conveyors in plant H. Most isolates formed black colonies (sulfide production) on Iron agar. The Aeromonas spp. isolates had three different profiles based on partial 16S rRNA gene (393 bp) sequences with the dominating profile, found in 22 samples, showing 100% identity to the type strain of A. salmonicida. The two other profiles had 100% similarity to the type strains of A. sobria (found in 4 samples) and A. hydrophila (found in 1 sample). The A. salmonicida like profile was found in all sample types in plant B: salmon from slaughterhouse, water samples and equipment (from both departments, including knife in filleting machine). Levels of Aeromonas spp. on ice-stored fillets from plant B were about 5 log cfu/g from three out of four different production steps after 14 days storage. The dominating sequence profile was also found in two samples from plant H, including a fillet sample. Acinetobacter was identified from all sample types including fillets from two out of four processing steps from plant B after 1 day storage on ice.

104

T. Møretrø et al. / International Journal of Food Microbiology 237 (2016) 98–108

B

Seawater

Conveyors

Stunning S

PS

P Ph S Desliming

P

S

PS

H

Ph P

Stunning S

S Ph

Ph

Bleeding tank

Ph S

Ph

SP Gutting machine

P

P

Gutting machine

S

S

P

S P

P

S

Grader

Storage tank

P

S

P

S Ph

S PS

S Storage tank

P

S

Fig. 5. Overview of prevalence of Pseudomonas spp. (P), Shewanella spp. (S) and Photobacterium spp. (Ph) in the slaughter departments of plant B and H. Bold large letter indicates dominance in sample point. Cylinders indicate samples from seawater.

The following genera of Enterobacteriaceae were present: Morganella, Kluyvera, Serratia, Buttiauxella and Yersinia. From both samplings in plant B, Morganella spp. with 100% identical 16S rRNA genes (424 bp) were detected on surfaces from the filleting machine. Bacteria within the genera Vibrio, Pseudoalteromonas and Psychromonas were present on unprocessed fish and in seawater but not found on surfaces in the processing plants after cleaning and disinfection. The numbers of identified Gram-positive bacteria were low. From the main sampling of plant B and plant H, only 6 of 1650 colonies were Gram-positive. In sampling 2 from plant B, Gram-positive bacteria were present in three out of four samples from equipment, with Carnobacterium spp. most commonly isolated. Lactococcus spp. was found in one of the equipment samples as well as in a sample from salmon from the slaughtering department.

4. Discussion The present study showed that spoilage associated bacteria contaminated salmon during processing. Higher bacterial levels were found in fillets that were processed industrially than in manually gutted and filleted salmon.

The results of the initial survey on bacterial levels of salmon fillets indicated that insufficient cleaning and disinfection resulted in contamination of the product during industrial processing of salmon. Starting with clean equipment, one would expect increased contamination during the production day, as bacteria from raw materials would contaminate equipment and growth would be supported by introduction of water and nutrients. Surprisingly, higher bacterial levels were found on fillets sampled early compared with mid-shift. This observation indicated that bacteria surviving cleaning and disinfection were transferred to salmon during processing, and that the bacterial contamination of processing equipment and machines decreased during the production day. Up to 5000 salmon were processed per hour and the high amount of salmon in contact with moving parts of equipment/machines and water remaining after sanitation may have a “cleansing” effect on equipment/machines. This may lead to high initial contamination of products, followed by reduced bacterial levels on product contact surfaces and less contamination to products after continued production. It should be noted that the fish processed early and mid-shift were originating from the same nets, thus it was unlikely that the observed differences in bacterial levels in the final products were due to varying bacterial levels in live fish. From the results in the initial study, it was hypothesized that bacteria able to survive sanitation contribute to contamination of salmon

T. Møretrø et al. / International Journal of Food Microbiology 237 (2016) 98–108

B

105

From slaughter

From slaughter Conveyors

Filleting machine

P

Trimming machine

P

Skinning machine

Packaging

*

P

P P

S

P

P

Filleting machine

Trimming machine *

P P

*

*

H

P

Skinning machine

P

Packaging

P

P

P

S

Ice stored fillet 14 d

Ice stored fillet 14 d

P

Post rigor line

P

Fig. 6. Overview of prevalence of Pseudomonas spp. (P) and Shewanella spp. (S) in the filleting department of plant B and H. Bold large letter indicates dominance in sample point. Photobacterium was not detected in any samples. *No bacteria isolated (b1/cm2).

fillets. To test this, the microbiota on fillets was identified and compared to the microbiota from processing lines of two salmon processing plants. In accordance with previous findings (Gram and Huss, 2000), Pseudomonas spp. dominated on ice- stored fillets while Shewanella spp. were also present. Also, higher levels of Shewanella and Pseudomonas were often found on industrially processed fillets compared to hand gutted/filleted fish. For Pseudomonas spp., the sequence profiles (partial 16S rRNA gene) most commonly found in fillets were also found among the isolates dominating on machines after cleaning. Together, the findings indicate that there is a net transfer of Pseudomonas spp. from equipment to fillets during processing. The 16S rRNA gene sequencing separates at the genus/species level. Thus, it cannot be excluded that the isolates from fillets and equipment were different. Typing using molecular fingerprinting or whole genome sequencing would be needed to verify this. The sequencing indicated that the contaminating microbiota partly originated from the raw materials as one of the most common sequence profiles (P45) on salmon fillets was also found on skin/gills of salmon early in the process (before desliming) in plant B. This was not surprising, since it is well known that Pseudomonas spp. can be found in seawater and live fish, including salmon (Cantas et al., 2011; Gram and Huss, 2000; Moore et al., 2006; Pascual et al., 2012). Many of the Pseudomonas spp. established in the processing plants probably originally entered the plants via live salmon or seawater. In general, Pseudomonas spp. growing to high numbers in food will cause spoilage.

In literature, species determination of Pseudomonas is often not performed (Gram and Huss, 1996), or is uncertain due to unclear and evolving taxonomy of Pseudomonas. The most common sequence profiles of isolates from fillets were similar to P. gessardi, P. libaniensis, P. fluorescens and P. lundensis. The involvement of these species in spoilage of salmon was not evaluated, however P. fluorescens and P. lundensis are known spoilers of food (Edwards et al., 1987; Ternstrøm et al., 1993; Tryfinopoulou et al., 2002). Similarly to Pseudomonas spp., higher numbers of Shewanella spp. were found on industrially produced fillets than manually prepared fillets, and isolates with identical sequence profile dominating on fillets from the seven plants were correspondingly found on equipment after cleaning in plant B and H. Together this indicated that fillets were contaminated from the production environment. However, contamination from seawater and the raw materials themselves was probably also important sources since Shewanella spp., including isolates with the two particular sequence profiles dominating in fillets in the initial study, were also frequently found on salmon in the slaughter departments and in tanks with sea water. This finding was expected as Shewanella spp. are widespread in the marine environment including fish and seawater (Gram and Huss, 2000; Nealson and Scott, 2006; Vogel et al., 2005). In seawater tanks but also at other sites, contact between skin and fish flesh was observed. Thus, there may be a direct transfer of Shewanella spp. from fish to fish during processing.

106

T. Møretrø et al. / International Journal of Food Microbiology 237 (2016) 98–108

The Shewanella spp. isolates from equipment and fillets with the most common 16S rRNA gene sequence profiles were similar to S. putrefaciens, S. baltica and S. vesiculosa. The two former species are known spoilers of fish (Dalgaard, 1995; Gram and Dalgaard, 2002; Hozbor et al., 2006), while we are not aware of published studies on spoilage potential of S. vesiculosa. Of three isolates tested from the present study closely related to the type strain of S. vesiculosa, none formed black colonies (H 2 S not produced) on Iron agar or reduced trimethylamine oxide to TMA (Undrum, 2015), which may indicate low spoilage potential. In contrast to what was found for Pseudomonas spp. and Shewanella spp. there were no clear differences in Photobacterium spp. levels between hand filleted salmon and industrially produced fillets. Photobacterium spp. were isolated from unprocessed salmon and apparently eliminated by cleaning and disinfection. The results suggest that contamination from the production environment was not an important source of Photobacterium spp. on the products. Photobacterium spp. are widely present in marine environments (Urbanczyk et al., 2011), and transfer of Photobacterium spp. from fish to fish during processing by flesh-skin contact or indirectly via seawater in tanks cannot be ruled out. Photobacterium spp. were identified by NGS analysis after 10 days ice-storage of fillets from all seven producers in the first part of the project, but not from colonies from 14 days ice stored fillets from plant B and H. The reason behind this is unclear. Two different methods were used for identifying Photobacterium spp. From the industrial processed fillets stored on ice for 10 days, Photobacterium spp. were detected at b15% of the total sequences by culture independent NGS analysis. From the 14 days fillets, culture-dependent identification of a maximum of 20 colonies per sample was performed. Thus Photobacterium spp. may have been present below the detection level (5%). The majority of the isolates was most closely related to Ph. phosphoreum, known for fish spoilage (Dalgaard, 1995), while some isolates were more related to Ph. iliopiscarium/Ph. kishitanii. Photobacterium iliopiscarium has been found in cold smoked salmon (Olofsson et al., 2007). In previous studies it has been difficult to evaluate the spoilage potential of various species of Photobacterium due to the previous challenges for reliable species identification (Ast and Dunlap, 2005). As found in other studies, species within the genus Pseudomonas seemed to be particularly fit to survive in food production environments as it dominated after cleaning in both processing plants sampled. The dominance of certain species/profiles of Pseudomonas indicates that certain species or types have an increased ability to survive and establish themselves in the processing plants. Pseudomonas spp. grow at low temperature, have low growth requirements, are resistant to antimicrobials and readily form biofilm (Drenkard and Ausubel, 2002; Moore et al., 2006). These abilities fit well with the humid production environment, 10–12 °C ambient temperature, and use of cleaning and disinfection agents in the processing plants. Bagge-Ravn et al. (2003) studied the bacteria microbiota in three fish processing plants and found that Pseudomonas spp. and yeast were most dominant after cleaning. In another study, Pseudomonas spp. dominated in a shrimp processing plant, while Pseudomonas was the second most dominant group after Enterobacteriaceae in a fish processing plant (Guδbjornsdottir et al., 2005). Pseudomonas spp. are also frequently isolated from surfaces after cleaning and disinfection in other types of food production (Brightwell et al., 2006; Liu et al., 2013; Mettler and Carpentier, 1998; Møretrø et al., 2013). We have not found other reports on the presence of Shewanella spp. on equipment/machines in processing plants after cleaning. Shewanella spp. were detected from equipment and machines after cleaning in the slaughter departments but not in the filleting departments. The reason for the different prevalence of Shewanella spp. in the two types of department is not clear. In general there are more blood, slime and water during production in the slaughter department than in the filleting department. Thus the bacterial levels before sanitation are likely to be higher, which may also lead to higher levels after sanitation in the slaughter department. Some surfaces in the slaughtering

department of plant H were not visually clean after sanitation. As the prevalence of Shewanella spp. was high on the surface of salmon in the slaughter department, it may be hypothesized that surfaces of equipment are contaminated with Shewanella spp. from salmon during production, leading to high levels on surfaces where a certain fraction survives cleaning and disinfection. While Photobacterium spp. were not found after cleaning and disinfection in the present study, Photobacterium spp. were found at low prevalence in one of two salmon smokehouses after cleaning and disinfection in another study (BaggeRavn et al., 2003), indicating that the bacterium may survive in some production environments. The above mentioned bacteria dominated in the processing plants and on fillets, however also other bacteria were present. Bacteria closely related to Aeromonas salmonicida colonized plant B. Aeromonas spp. have been found on surfaces after sanitation in fish- and shrimp processing plants (Bagge-Ravn et al., 2003; Guδbjornsdottir et al., 2005). Aeromonas spp. may spoil fish (Joffraud et al., 2001) and be human and fish pathogens (Beaz-Hidalgo and Figueras, 2013). The very low numbers of Gram-positive bacteria found in the processing plants are in accordance with the findings from a shrimp- and a fish processing plant (Guδbjornsdottir et al., 2005). However, BaggeRavn et al. (2003) found a prevalence of 6–31% Gram-positive bacteria after cleaning and disinfection of salmon smokehouses and that the relative presence of Gram-positive bacteria was reduced after cleaning and disinfection compared to under production. While higher prevalence of Gram-positive bacteria have been reported in less humid environments like smokehouses, Gram-negative bacteria often dominate in processing environments like slaughtering and filleting lines having low temperatures and humid conditions (Brightwell et al., 2006; Gibson et al., 1995; Liu et al., 2013). In general Gram positive bacteria are more resistant to air-drying than Gram negative bacteria (Kramer et al., 2006). To our knowledge there have been no other studies focusing on the processing environment as a possible source of spoilage bacteria of fish. This is in contrast to the food pathogen Listeria monocytogenes, where cross contamination from equipment/machines is the main contamination source of ready to eat foods, including fish products (Holch et al., 2013; Møretrø and Langsrud, 2004; Rørvik et al., 1995). Listeria monocytogenes was not identified in any sample in the present study. However, to isolate L. monocytogenes from processing plants, an enrichment step should be included, as L. monocytogenes is normally outnumbered by other bacteria (ISO, 1998). The present study gives new information about the contamination routes of the most important spoilage bacteria of salmon as well as indications that improved hygiene may enhance the quality of salmon fillets. A potential to reduce the initial levels of bacteria on fresh salmon fillets with about 90% (one log) using optimal hygienic conditions was indicated. Thus improved hygiene may increase the quality and shelf life of ice stored salmon. However it must be taken into consideration that growth of L. monocytogenes may be a factor limiting the shelf life of such products. In conclusion, salmon may be contaminated by species within the spoilage associated genera Pseudomonas and Shewanella during processing. There is a potential for better quality of salmon fillets by improving hygiene in the processing plants. Acknowledgements The work was funded by The Norwegian Seafood Research Fund (grant no. FHF-900792 and FHF-900938) and the Research Council of Norway (grant no. 194050/F40). Thanks to Hanne Tobiassen, Rudi Jakobsen, Kurt Olav Oppedal, Pål Storø and Kristian Prytz for valuable advices and discussions through their participation in the steering group of the project. The authors wish to thank Janina S. Berg, Merete Rusås Jensen, Signe Marie Drømtorp, Tove Maugesten, Marius Normann, Elin Røssvoll and Emma Undrum for excellent technical assistance.

T. Møretrø et al. / International Journal of Food Microbiology 237 (2016) 98–108

107

Appendix A

100

Other

90

Gammaproteobacteria Relative abundance (%)

80

Brochothrix Aliivibrio

70

Verrucomicrobiaceae

60

Staphylococcus 50

Sphingobacteriales Proteobacteria

40

Comamonadaceae 30

Rickettsiales (f__mitochondria)

20

Acinetobacter Janthinobacterium

10

A

B

C

D

E

F

Stramenopiles Mid-shift

Early

Early

Mid-shift

Early

Mid-shift

Control

Early

Mid-shift

Control

Mid-shift

Early

Control

Mid-shift

Early

Control

Mid-shift

Early

Control

0

Shewanella

Photobacterium Pseudomonas

G

Fig. A1. Dominating bacterial taxa (genus level) in salmon fillets from seven producing plants after storage for 10 days on ice. The fillets were hygienically produced outside the plants (control) or industrially produced in the plants early and mid-shift on the production day. Taxa represented above 0.5% across all samples are shown, the remaining taxa are represented in “Other”.

Table A1 Prevalence of bacterial genera as percent positive in different types of samples from two processing plants. Generab

Plant Ba

Plant H c

Pseudomonas Shewanella Aeromonas Acinetobacter Janthinobacterium Psychrobacter Chryseobacterium Comamonas Flavobacterium Morganella Serratia Yersinia Photobacterium Pseudoalteromonas Kluyvera Vibrio Psychromonas Staphylococcus a

d

e

f

Salmon swabs (3 )

Water (3)

Machines (39)

Total (45)

Salmon swabs (4)

Water (5)

Machines (45)

Total (54)

33g 67 33 33 0 33 33 0 67 0 0 0 67 33 0 0 0 0

67 67 67 33 33 33 0 0 33 0 0 33 0 33 0 0 0 0

62 38 23 21 13 10 10 13 5 8 8 5 0 0 0 0 0 0

60 31 27 22 13 13 11 11 11 7 7 7 4 4 2 0 0 0

25 100 25 0 0 75 0 0 25 75 0 0 75 75 0 100 100 0

80 100 0 0 0 60 0 0 40 0 0 0 40 60 0 40 20 0

49 18 0 11 11 16 9 4 4 0 13 4 0 0 7 0 0 9

50 31 2 9 9 24 7 4 9 6 11 4 9 11 6 11 9 7

Data shown from plant B are from sampling 1. Bacterial genera identified from N5% of the samples in at least one of the plants are shown. c Pooled sample of three salmon from the slaughter department, swabbed after stunning, after desliming, before gutting and after gutting. d Number of samples. e Seawater pumped into plant, seawater stunner/desliming, bleeding tank, storage tanks. f All machines/equipment (product contact) examples: stunning machine, head cutter, gutting machine, vacuum systems, filleting machine, trimming machine, bone remover, conveyor systems. g Percent positive samples, e.g. Pseudomonas identified in 1 out of 3 (33%) pooled samples. b

108

T. Møretrø et al. / International Journal of Food Microbiology 237 (2016) 98–108

References Ast, J.C., Dunlap, P.V., 2005. Phylogenetic resolution and habitat specificity of members of the Photobacterium phosphoreum species group. Environ. Microbiol. 7, 1641–1654. Bagge-Ravn, D., Ng, Y., Hjelm, M., Christiansen, J.N., Johansen, C., Gram, L., 2003. The microbial ecology of processing equipment in different fish industries - analysis of the microflora during processing and following cleaning and disinfection. Int. J. Food Microbiol. 87, 239–250. Beaz-Hidalgo, R., Figueras, M.J., 2013. Aeromonas spp. whole genomes and virulence factors implicated in fish disease. J. Fish Dis. 36, 371–388. Brightwell, G., Boerema, J., Mills, J., Mowat, E., Pulford, D., 2006. Identifying the bacterial community on the surface of Intralox((TM)) belting in a meat boning room by culture-dependent and culture-independent 16S rDNA sequence analysis. Int. J. Food Microbiol. 109, 47–53. Cantas, L., Fraser, T.W.K., Fjelldal, P.G., Mayer, I., Sorum, H., 2011. The culturable intestinal microbiota of triploid and diploid juvenile Atlantic salmon (Salmo salar) - a comparison of composition and drug resistance. BMC Vet. Res. 7. Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., Fierer, N., Pena, A.G., Goodrich, J.K., Gordon, J.I., Huttley, G.A., Kelley, S.T., Knights, D., Koenig, J.E., Ley, R.E., Lozupone, C.A., McDonald, D., Muegge, B.D., Pirrung, M., Reeder, J., Sevinsky, J.R., Tumbaugh, P.J., Walters, W.A., Widmann, J., Yatsunenko, T., Zaneveld, J., Knight, R., 2010. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336. Caporaso, J.G., Lauber, C.L., Walters, W.A., Berg-Lyons, D., Huntley, J., Fierer, N., Owens, S.M., Betley, J., Fraser, L., Bauer, M., Gormley, N., Gilbert, J.A., Smith, G., Knight, R., 2012. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 6, 1621–1624. CBI Market Intelligence, 2016. CBI trade statistics: Fish and seafood. Available at: https:// www.cbi.eu/sites/default/files/trade-statistics-fish-seafood.pdf (Assessed January 26th, 2016). Chaillou, S., Chaulot-Talmon, A., Caekebeke, H., Cardinal, M., Christieans, S., Denis, C., Desmonts, M.H., Dousset, X., Feurer, C., Hamon, E., Joffraud, J.-J., La Carbona, S., Leroi, F., Leroy, S., Lorre, S., Mace, S., Pilet, M.-F., Prevost, H., Rivollier, M., Roux, D., Talon, R., Zagorec, M., Champomier-Verges, M.-C., 2015. Origin and ecological selection of core and food-specific bacterial communities associated with meat and seafood spoilage. ISME J. 9, 1105–1118. Dalgaard, P., 1995. Qualitative and quantitative characterization of spoilage bacteria from packed fish. Int. J. Food Microbiol. 26, 319–333. Dalgaard, P., Gram, L., Huss, H.H., 1993. Spoilage and shelf-life of cod fillets packed in vacuum or modified atmospheres. Int. J. Food Microbiol. 19, 283–294. Drenkard, E., Ausubel, F.M., 2002. Pseudomonas biofilm formation and antibiotic resistance are linked to phenotypic variation. Nature 416, 740–743. Edwards, R.A., Dainty, R.H., Hibbard, C.M., 1987. Volatile compounds produced by meat pseudomonads and related reference strains during growth on beef stored in air at chill temperatures. J. Appl. Bacteriol. 62, 403–412. Emborg, J., Laursen, B.G., Rathjen, T., Dalgaard, P., 2002. Microbial spoilage and formation of biogenic amines in fresh and thawed modified atmosphere-packed salmon (Salmo salar) at 2 degrees C. J. Appl. Microbiol. 92, 790–799. Gibson, H., Taylor, J., Hall, K., Holah, J., 1995. Biofilms and their Detection in the Food Industry. Chipping Campden, Gloucesterchire. Gram, L., Dalgaard, P., 2002. Fish spoilage bacteria - problems and solutions. Curr. Opin. Biotechnol. 13, 262–266. Gram, L., Huss, H.H., 1996. Microbiological spoilage of fish and fish products. Int. J. Food Microbiol. 33, 121–137. Gram, L., Huss, H.H., 2000. Fresh and processed fish and shellfish. In: Lund, B.M., BairdParker, T.C., Gould, G.W. (Eds.), The Microbiological Safety and Quality of Food. Aspen Publishers Inc., Gaithersburg, Maryland. Guδbjornsdottir, B., Einarsson, H., Thorkelsson, G., 2005. Microbial adhesion to processing lines for fish fillets and cooked shrimp: Influence of stainless steel surface finish and presence of gram-negative bacteria on the attachment of Listeria monocytogenes. Food Technol. Biotechnol. 43, 55–61. Holch, A., Webb, K., Lukjancenko, O., Ussery, D., Rosenthal, B.M., Gram, L., 2013. Genome sequencing identifies two nearly unchanged strains of persistent Listeria monocytogenes isolated at two different fish processing plants sampled 6 years apart. Appl. Environ. Microbiol. 79, 2944–2951. Hovda, M.B., Fontanillas, R., McGurk, C., Obach, A., Rosnes, J.T., 2012. Seasonal variations in the intestinal microbiota of farmed Atlantic salmon (Salmo salar L.). Aquac. Res. 43, 154–159. Hozbor, M.C., Saiz, A.I., Yeannes, M.I., Fritz, R., 2006. Microbiological changes and its correlation with quality indices during aerobic iced storage of sea salmon (Pseudopercis semifasciata). Lwt-Food Science and Technology 39, 99–104. Hultman, J., Rahkila, R., Ali, J., Rousu, J., Bjorkroth, K.J., 2015. Meat processing plant microbiome and contamination patterns of cold-tolerant bacteria causing food safety and spoilage risks in the manufacture of vacuum-packaged cooked sausages. Appl. Environ. Microbiol. 81, 7088–7097. ISO, 1998. Microbiology of Food and Animal Feeding Stuffs - Horizontal Method for the Detection and Enumeration of L. monocytogenes. pp. 11290–11292. Joffraud, J.J., Leroi, F., Roy, C., Berdague, J.L., 2001. Characterisation of volatile compounds produced by bacteria isolated from the spoilage flora of cold-smoked salmon. Int. J. Food Microbiol. 66, 175–184.

Kaeding, A.J., Ast, J.C., Pearce, M.M., Urbanczyk, H., Kimura, S., Endo, H., Nakamura, M., Dunlap, P.V., 2007. Phylogenetic diversity and cosymbiosis in the bioluminescent symbioses of “Photobacterium mandapamensis”. Appl. Environ. Microbiol. 73, 3173–3182. Kramer, A., Schwebke, I., Kampf, G., 2006. How long do nosocomial pathogens persist on inanimate surfaces? A systematic review. BMC Infect. Dis. 6, 130. Liu, N.T., Lefcourt, A.M., Nou, X., Shelton, D.R., Zhang, G., Lo, Y.M., 2013. Native microflora in fresh-cut produce processing plants and their potentials for biofilm formation. J. Food Prot. 76, 827–832. Mettler, E., Carpentier, B., 1998. Variations over time of microbial load and physicochemical properties of floor materials after cleaning in food industry premises. J. Food Prot. 61, 57–65. Moen, B., Røssvoll, E., Måge, I., Møretrø, T., Langsrud, S., 2015. Microbiota formed on attached stainless steel coupons correlate with the natural biofilm of the sink surface in domestic kitchens. Can. J. Microbiol. 10. Moore, E.R.B., Tindall, B.J., Dos Santos, V.A.P.M., Pieper, D.H., Ramos, J.L., Palleroni, N.J., 2006. Nonmedical: Pseudomonas. Prokaryotes 6, 646–703. Møretrø, T., Langsrud, S., 2004. Listeria monocytogenes: biofilm formation and persistence in food processing environments. Biofilms 1, 107–121. Møretrø, T., Langsrud, S., Heir, E., 2013. Bacteria on meat abattoir meat production process surfaces after sanitation: characterisation of survival properties of Listeria monocytogenes and the commensal bacterial flora. Adv. Microbiol. 3, 255–264. Nadkarni, M.A., Martin, F.E., Jacques, N.A., Hunter, N., 2002. Determination of bacterial load by real-time PCR using a broad-range (universal) probe and primers set. Microbiology 148, 257–266. Navarrete, P., Espejo, R.T., Romero, J., 2009. Molecular analysis of microbiota along the digestive tract of juvenile Atlantic Salmon (Salmo salar L.). Microb. Ecol. 57, 550–561. Nealson, K.H., Scott, J., 2006. The ecophysiology of the genus Shewanella. In: Falkow, S., Rosenberg, E., Schleifer, K.H., Stackebrandt, E. (Eds.), The Prokaryotes. Springer Science & Business Media, pp. 1133–1151. Norwegian Seafood Council, 2015. Norwegian seafood exports 2014. Available at: http:// en.seafood.no/News-and-media/Key-figures (Assessed January 27th 2016). Olafsdottir, G., Lauzon, H.L., Martinsdottir, E., Kristbergsson, K., 2006. Influence of storage temperature on microbial spoilage characteristics of haddock fillets (Melanogrammus aeglefinus) evaluated by multivariate quality prediction. Int. J. Food Microbiol. 111, 112–125. Olofsson, T.C., Ahrne, S., Molin, G., 2007. The bacterial flora of vacuum-packed coldsmoked salmon stored at 7 degrees C, identified by direct 16S rRNA gene analysis and pure culture technique. J. Appl. Microbiol. 103, 109–119. Omer, M.K., Hauge, S.J., Ostensvik, O., Moen, B., Alvseike, O., Rotterud, O.-J., Prieto, M., Dommersnes, S., Nesteng, O.H., Nesbakken, T., 2015. Effects of hygienic treatments during slaughtering on microbial dynamics and contamination of sheep meat. Int. J. Food Microbiol. 194, 7–14. Parlapani, F.F., Boziaris, I.S., 2016. Monitoring of spoilage and determination of microbial communities based on 16S rRNA gene sequence analysis of whole sea bream stored at various temperatures. Lwt-Food Sci. Technol. 66, 553–559. Parlapani, F.F., Verdos, G.I., Haroutounian, S.A., Boziaris, I.S., 2015. The dynamics of Pseudomonas and volatilome during the spoilage of gutted sea bream stored at 2 °C. Food Control 55, 257–265. Pascual, J., Lucena, T., Ruvira, M.A., Giordano, A., Gambacorta, A., Garay, E., Arahal, D.R., Pujaltel, M.J., Macian, M.C., 2012. Pseudomonas litoralis sp nov., isolated from Mediterranean seawater. Int. J. Syst. Evol. Microbiol. 62, 438–444. Quested, T.E., Cook, P.E., Gorris, L.G.M., Cole, M.B., 2010. Trends in technology, trade and consumption likely to impact on microbial food safety. Int. J. Food Microbiol. 139, S29–S42. Rørvik, L.M., Vaugant, D., Yndestad, M., 1995. Contamination pattern of Listeria monocytogenes and other Listeria spp. in a salmon slaughterhouse and smoked salmon processing plant. Int. J. Food Microbiol. 25, 19–27. Stellato, G., De Filippis, F., La Storia, A., Ercolini, D., 2015. Coexistence of lactic acid bacteria and potential spoilage microbiota in a dairy processing environment. Appl. Environ. Microbiol. 81, 7893–7904. Ternstrøm, A., Lindberg, A.-M., Molin, G., 1993. Classification of the spoilage flora of raw and pasterurized bovine milk, with special reference to Pseudomonas and Bacillus. J. Appl. Bacteriol. 75, 25–34. Tryfinopoulou, P., Tsakalidou, E., Nychas, G.J.E., 2002. Characterization of Pseudomonas spp. associated with spoilage of gilt-head sea bream stored under various conditions. Appl. Environ. Microbiol. 68, 65–72. Undrum, E., 2015. Identification and Characterization of Sulphide Producing Bacteria from Processing Plants for Salmon MSc Thesis Norwegian University of Life Sciences, Ås, Norway, p. 80. Urbanczyk, H., Ast, J.C., Dunlap, P.V., 2011. Phylogeny, genomics, and symbiosis of Photobacterium. FEMS Microbiol. Rev. 35, 324–342. van Spreekens, K.J.A., 1974. The sustainability of a modification of Long and Hammer's medium for the enumeration of more fastidious bacteria from fresh fishery products. Arch. Leb. 10, 213–219. Vogel, B.F., Venkateswaran, K., Satomi, M., Gram, L., 2005. Identification of Shewanella baltica as the most important H2S-producing species during iced storage of Danish marine fish. Appl. Environ. Microbiol. 71, 6689–6697.