1

1 2 A quantitative comparison of sample matrices for the detection of Campylobacter spp. in 3 .broiler houses 4 5 6 Received for publication, August 10, 2014 7 Accepted September 20, 2014 8 9 ROBERT H. MADDEN a, HYWEL BALL b, MIKE HUTCHISON c, 10 FIONA YOUNG, a AND MALCOLM TAYLOR a 11 Food Microbiology Branch, AFBI, Newforge Lane, Belfast BT9 5PX, Northern Ireland a; 12 Veterinary Sciences Division, AFBI, Stoney Road, Belfast BT4 3SD, Northern Ireland b; and 13 .,School of Veterinary Sciences, Bristol University Langford, BS40 5DU, UK c 14 Tel.: +44 (0)2890 255312; Fax: +44(0) 2890 255458; Email: [email protected] 15 16 Corresponding author email: [email protected] 17 Phone 44 (0)2890 255312 18 Fax 44(0) 2890 255458 19 20 Abstract 21 Most retail chicken in Europe, and in many other parts of the world, is contaminated with 22 Campylobacter spp., a major cause of food poisoning. There is therefore a need to detect this pathogen 23 on broiler farms in order to verify the efficacy of control measures which have been implemented, and 24 determine which flocks are infected. In this study eight potential matrices from broiler houses were 25 studied; bootswabs, cecal droppings, ceca, cloacal swab, dust, feces, litter, and ventral swab. They were 26 compared in a quantitative manner to determine which was best for the detection of Campylobacter, 27 using samples collected from commercial broiler houses. Campylobacter spp. recovered by the matrices 28 were enumerated using ISO 10272-1:2006. Bootswabs consisting of plastic overshoes (Tunika) 29 recovered the highest number of Campylobacter (p < 0.05) and showed the highest prevalence. When 30 bootswabs were stored chilled no loss of Campylobacter viability was seen for up to 48h allowing for 31 flexibility in sampling, and sample handling, since immediate analysis was not required. However, 32 overgrowth of the selective medium, mCCDA, by non-campylobacters was seen if samples were stored 33 at 22°C. Therefore transport of bootswabs cannot be undertaken at ambient temperature. Overall, in 34 .terms of Campylobacter recovery, ease of use and cost, bootswabs were the preferred sample matrix 35 36 .Keywords: Campylobacter, broiler, farm, sampling, bootswabs, enumeration 37 38 Introduction 39 40 Campylobacter are the principal causes of gastro-enteritis world-wide (4, 19) with 41 raw chicken an important source of Campylobacter spp. (7,20). Approximately 80% (3), to 42 over 90% (9, 16, 17) of UK retail chicken carries Campylobacter. The economic costs of 43 campylobacteriosis caused by broiler meat are estimated to be of up to €750 million in the 44 .(European Community (6 45 A Europe-wide study (5) found that 86% of broiler carcasses in UK abattoirs carried 46 campylobacters. In order to reduce the load at slaughter, the prevalence of campylobacters on 47 broiler farms must be decreased. In Denmark, intervention measures to achieve this aim have 48 been proposed (22), with monitoring of broilers for Campylobacter initially undertaken at 49 .(abattoirs, but subsequently extended to broiler houses (11, 23 50 Historically, microbiological sampling of broiler houses was mainly developed to 51 detect salmonellas, using qualitative procedures (8, 15). Qualitative detection of 52 campylobacters on farms has been developed based on these sampling methods (10, 11, 21), 53 with no single sampling methodology being universally applied. Accordingly, this study 54 aimed to compare eight sample matrices to determine which one gave the highest recoveries 55 .of Campylobacter in broiler houses 56 57 Material and Methods 58 59 .Broiler farms studied 60 61 A major chicken processor supported the project by allowing access to broiler 62 houses, and providing staff to assist during farm visits. Farms to be sampled were identified 63 by company staff and took place in previously thinned houses, which would shortly be 64 subjected to final clearance. For biosecurity reasons only one farm was visited on each 65 sampling trip. Sampling staff adhered to company biosecurity measures at all times. Once 66 inside the house to be sampled staff notionally divided the space into four equal quadrants, 67 based on the distribution of the normal equipment present. Lengthwise, the lines of drinkers 68 and feeders present provided lanes, along which samples could be taken. The house could be 69 divided in two lengthwise based on the positioning of air ducts or lighting units, which were 70 usually evenly spaced. Hence each house was divided into four approximately equal 71 quadrants for sampling purposes. Officers prepared sketch plans of each house sampled, for 72 .subsequent reference 73 74 .Evaluation of sample matrices 75 76 Eight matrices were initially studied and samples, as described below, taken in the 77 four quadrants of a broiler house were placed in pre-labeled bags. On leaving the house 78 samples were stored in a coolbox with ice packs prior to returning to the laboratory within 79 two hours. Samplers wore a fresh pair of disposable gloves for each individual sampling 80 :exercise. The sampling procedures used were 81 Bootswabs. These were Tunika overshoes (Bowden and Knights, Thetford, UK). 82 Firstly, the sampler put on a fresh pair of disposable Tyvek overshoes (Arco, Hull, UK) to 83 prevent their footwear from contaminating the bootswabs. The Tunika bootswabs were then 84 put on over the clean overshoes and the sampler walked down the designated quadrant in one 85 lane and returned by another. The boot swabs were then placed in Seward closure bags 86 (BA6041/CLR, Seward Ltd, Worthing, UK), 10 cm3 of maximum recovery diluent (MRD, 87 .CM733; Oxoid, Basingstoke, UK) added, and the bag sealed 88 Fecal samples. The sampler walked the length of a quadrant in one lane and 89 collected 5 ‘pinches’ of fecal material in a sealable bag, then returned by a second lane 90 repeating the collection procedure. Litter samples. These were collected using the same 91 .procedures as the fecal samples 92 Ceca. A company representative walked through the house and selected four birds 93 meeting the normal criteria for culling. These were killed and placed in individual large 94 plastic bags, and transported to the laboratory in insulated coolbox containers with ice packs. 95 Once in the laboratory the birds were dissected and the ceca placed in Seward closure bags, 96 and stored refrigerated until required. This is the only sample type that was not collected from 97 .the designated quadrants 98 Ventral swab. Four birds per quadrant were sampled using pre-moistened (MRD) 99 standard cellulose carcass sampling swabs (TS/15-B:PSD. TSC, Heywood, UK). A bird was 100 caught, inverted to expose the ventral area which was subjected to five ‘wipes’ using one side 101 of a fresh cellulose sponge. This sponge was then used to sample a further three birds, using 102 .each side twice. The swab was placed in a Seward closure bag which was sealed and stored 103 Cloacal swab. After the ventral area of a bird had been swabbed a company staff 104 member inserted a transport swab (Amies charcoal, TS/5-10. TSC, Heywood, UK) into the 105 cloaca and turned it. The swab was then returned to its tube. After four birds had been 106 sampled, each with a fresh swab, the swabs were placed in a labeled bag which was then 107 sealed and stored. Cecal droppings. These were differentiated from normal feces by their 108 characteristic appearance; dark brown and glossy. Four individual droppings per quadrant 109 were sampled by sweeping a transport swab through a fresh dropping, using a fresh swab for 110 .each of the four droppings 111 Dust. Cellulose carcass sampling swabs, as used for the ventral swab, were used to swab the 112 upper surfaces of feeder and water lines. One side of the swab was used to wipe the chosen 113 line whilst walking down a quadrant, and a different line was swabbed using the other surface 114 .of the swab on the return walk 115 116 .Microbiological analysis 117 118 All samples were blended using a Seward 400 stomacher (Seward ltd, Worthing, 119 :UK) for 1 min. Samples were prepared for the enumeration of Campylobacter as follows 120 .Bootswabs: MRD (90 cm3) added and sample blended 121 Fecal samples: Pooled fecal samples were thoroughly mixed, and 10 g of the feces 122 .was added to 90 cm3 MRD and blended 123 Litter: This was again thoroughly mixed, and 10 g was added to 90 cm3 MRD. To 124 .prevent sample loss from punctured bags the sample was hand shaken for 15 s 125 Ceca: These were weighed, then MRD added to give 1:9 (w/w) dilution and ceca 126 .manually disrupted by hand squeezing, before being blended 127 .For ventral and dust swabs, MRD (90 cm3) was added and the sample was blended 128 Cloacal swabs and swabs of cecal droppings: sets of four swabs were added to 10 129 .cm3 MRD in a sterile plastic universal and vortex mixed for 10 s 130 Enumeration of Campylobacter. The sample suspensions prepared as above were 131 utilized as described in ISO EN 10272-1:2006 Microbiology of food and animal feeding 132 stuffs-Horizontal method for detection and enumeration of Campylobacter spp.. Briefly, the 133 suspensions were used to prepare decimal dilution series in MRD. Each dilution was plated 134 out, in duplicate, with 0.1 cm3 being applied to modified charcoal cefoperazone deoxycholate 135 agar (mCCDA; Oxoid CM0739 plus SR0155). Plates were incubated at 41.5°C in a 136 microaerobic atmosphere (85% N2, 10% CO2 and 5% O2, all v ⁄ v) in a Don Whitley MACS 137 workstation (Don Whitley Scientific, Shipley, UK). Typical colonies were enumerated and 138 confirmed by determination of cell morphology, characteristic motility in a hanging drop, 139 .oxidase reaction, aerobic growth at 41.5°, and microaerophilic growth at 25°C 140 141 .Bootswab storage 142 143 To assess the effect of storage of bootswabs on the recovery of campylobacters three 144 farms were visited and a total of ten houses sampled, using the Tunika bootswab sampling 145 procedures described above. In each house the sampler walked the full length of the house 146 and returned. This conducted seven times, wearing a fresh pair of bootswabs and in a 147 different lane each time. Previous work had shown that such replicate samples did not differ 148 significantly, p < 0.05, in terms of Campylobacter enumeration. On return to the lab each pair 149 of bootswabs was assigned a random number, based on which they were allocated to one of 150 seven treatments. One pair was immediately subjected to Campylobacter enumeration whilst 151 three pairs were placed in a chill (3°C) and three in an incubator (22°C). One pair of 152 bootswabs from the chill, and one from the incubator, was subjected to Campylobacter 153 .enumeration after 1, 2 and 4d. sampling was replicated in 10 different houses 154 155 .Statistical analysis 156 157 Prior to analysis the Campylobacter counts for each of the matrix samples were 158 transformed by taking a logarithm to base ten for each sample measured. Variation in the 159 measured count was attributable to either the farm, house or quadrant from which the sample 160 .was taken. Additionally for the ceca samples this could be split into farm, house and bird 161 In order to ascertain the percentage variation attributable to each of this factors listed 162 above a REML variance components model was fitted to each count in turn. For the boot 163 swab, feces and dropping counts, quadrant nested within house nested within farm was fitted 164 as the random term in the model. A similar model was fitted for the ceca counts with bird 165 replacing quadrant in this case. Individual variance components for the factors fitted to each 166 count were saved for each variable and used to calculate the percentage variation for each 167 .factor in the model in question 168 As the method of analysis carried out above requires that each variable being 169 analyzed follows a normal distribution, this was assessed using the Anderson-Darling, 170 .Cramer-von Mises and the Watson statistic in every case 171 All analyses were carried out using GenStat for Windows Release 14.1For all 172 .analyses, a value of p < 0.05 was considered significant 173 174 175 Results and Discussions 176 177 One farm was visited and samples of the eight matrix types taken in two houses, in 178 four quadrants. Mean Campylobacter counts (Log10 cfu) for cloacal swab, litter, ventral swab, 179 and dust were 6.40, 4.92, 3.37 and 2.78 respectively. Counts were per gram for litter samples, 180 and per swab for the other three samples. Only three dust samples yielded campylobacters. 181 Campylobacters were enumerated from all of the other matrices. No further sampling based 182 on cloacal swabs, litter, ventral swabs, or dust was undertaken. A further eight farms were 183 then visited, and the remaining four matrices were sampled in twenty three houses, in each of 184 four quadrants, to give an overall total of 100 samples per matrix. Compared with the other 185 matrices, significantly higher numbers of Campylobacter were recovered from bootswabs, 186 .p<0.05. In addition, bootswabs had the highest number of positive samples, Table 1 187 To investigate the persistence of Campylobacter on bootswabs samples were stored 188 in a chill, and at 22°C to mimic ambient temperature. For samples stored at 3°C the mean 189 Campylobacter counts (log10 cfu/sample) after 0, 1, 2 and 4d were 7.31 ±0.67, 7.06 ±0.67, 190 7.29 ±0.56 and 6.52 ±0.62, respectively. Each result is the mean of 10 samples. Results are 191 not presented for the samples stored at 22°C as considerable overgrowth by non- 192 campylobacters was observed on some of the mCCDA plates, which prevented enumeration 193 of Campylobacter colonies; after 1 d three samples could not be enumerated, after 2 d seven, 194 .and after 4 d nine samples could not be counted 195 This study aimed to compare eight matrices to determine which would recover the 196 highest numbers of Campylobacter from broiler houses containing infected birds. The lowest 197 recoveries were from cloacal swabs, litter, ventral swabs, and dust. Since Campylobacter are 198 highly susceptible to desiccation (2,13) the low recoveries from dust are not surprising. Poor 199 recoveries of Campylobacter from litter were previously reported (1), supporting the decision 200 .to discontinue work with this matrix 201 Cloacal swabs have been used for the qualitative detection of campylobacters in 202 broilers (12, 14), but with the involvement of veterinarians. Routine sampling with cloacal 203 swabs would require staff training, and could cause stress to the birds. However, similar 204 numbers of campylobacters were recovered from fecal samples (Table 1), which were much 205 easier to collect, therefore sampling with cloacal swabs was discontinued. Ventral swabbing 206 was evaluated as a simpler, non-invasive alternative to cloacal swabs, but recovered 207 approximately 1000-fold fewer campylobacters per sample and hence sampling was 208 .discontinued 209 210 Table 1 211 Positive samples Standard

(%) deviation (Campylobacter log10 (cfu Sample matrix 95 0.69 7.61a Bootswab 89 1.32 7.38b Ceca 92 1.25 6.49c Cecal dropping swabs 92 0.87 6.41c Feces 212 213

214 Table 1. Mean recovery of campylobacters, as log10 cfu, from four sample matrices. Numbers with 215 different suffixes are significantly different, p < 0.05. Samples (n=100) were collected from four discrete areas 216 (quadrants) in each of twenty five broiler houses. For swab samples results are log10 (cfu) per set of swabs, 217 .(whilst for ceca and feces results are log10 (cfu/g 218 219 Of the four remaining matrices, bootswabs recovered significantly higher numbers 220 of campylobacters than the other three matrices, p < 0.05, and also gave the highest number 221 of positive samples (Table 1). They were also relatively cheap and simple to use, and allowed 222 sampling to be undertaken quickly. A further advantage of boot swabs is that farmers 223 routinely walk the broiler houses several times per day to remove injured or deceased birds 224 and samples can be collected as part of that practice. For these reasons bootswabs have been 225 used previously in several UK studies (8). Reactions regarding the practicality of bootswabs 226 .from poultry farmers and production staff observing the trial were favorable 227 For the remaining matrices, it was noted that removal of the ceca from cull birds 228 required the use of trained staff. Further, although high numbers of campylobacters were 229 recovered from these samples, 6% of samples were false negatives, with regards to the house 230 status. Increasing the sensitivity of the method would require that more birds were culled, 231 further reducing the acceptability of this matrix. In addition shipping of the birds to the 232 laboratory would be costly due to their weight. Alternatively farm staff would have to be 233 taught how to routinely eviscerate the birds and remove the ceca for shipping, which could 234 .prove problematic 235 236 Recoveries of campylobacters in fecal samples and cecal droppings were not 237 significantly different, p > 0.05 (Table 1). However, collecting feces was a relatively simple 238 procedure whereas locating the characteristic cecal droppings required careful examination of 239 the litter surface. This was difficult in houses with no natural daylight. As birds neared the 240 maximum stocking density floor space was minimized and birds kicked litter over the 241 droppings and obscured them. Therefore sampling of feces would be the preferable method. 242 However, 3% of fecal samples gave false negative results with regard to the presence of 243 campylobacters in the house, and Campylobacter counts were 93% lower than those obtained 244 using bootswabs. It should be noted that a fecal sample comprised 10 discrete ‘pinches’ of 245 material, whilst in a typical broiler house the bootswabs of a sampling officer were estimated 246 .to make contact with approximately 2.5m2 of litter 247 Since bootswabs were clearly the best sampling matrix the effect of shipping them 248 from a farm to the laboratory, using chilled containers and at ambient temperature, was 249 investigated. At chill temperature there was no significant loss of viability of campylobacters 250 for up to 48 h, allowing for flexibility in the times of sampling, shipping and laboratory 251 analysis. However, ambient temperature storage resulted in a proliferation of non- 252 campylobacters which subsequently prevented the enumeration of Campylobacter spp. on 253 mCCDA. Similar overgrowth was found to be due to the proliferation of antibiotic resistant 254 E. coli, able to degrade cefoperazone (18). Since E. coli grows readily above 10°C, but not 255 below 6°C (24), it is probable that similar organisms were causing the overgrowth observed. 256 .Hence it was essential that samples were kept chilled when being shipped to the laboratory 257 To control Campylobacter spp. in broilers it is essential that the colonization status 258 of a broiler house is determined accurately so that the efficacy of biosecurity measures can be 259 determined, and the appropriate fate of the infected birds in the food chain chosen. 260 Bootswabs gave the highest recoveries of Campylobacter spp., in terms of numbers and 261 prevalence, of eight matrices studied. They were seen to be cheap, simple and quick to use, 262 and samples could be stored for up to 48h with no loss of campylobacter viability. They are 263 .therefore the recommended sampling matrix 264 265 Acknowledgements 266 267 This project was funded by the Food Standards Agency as project M01060. The authors 268 thank various members of the British Poultry Council for donations of staff time and farm 269 .access 270

271 References 272 Berghaus, R. D., S. G. Thayer, B. F. Law, R. M. Mild, C. L. Hofacre, and R. S. Singer. 2013. .1 273 Enumeration of Salmonella and Campylobacter in environmental farm samples and processing 274 plant carcass rinses from commercial broiler chicken flocks. Appl. Env. Microbiol. Published 275 .ahead of print 26 April 2013, doi: 10.1128/AEM.00836-13 276 Berrang, M. E. and J. K. Northcutt. 2005. Use of water spray and extended drying time to lower .2 277 .bacterial numbers on soiled flooring from broiler transport coops. Poult. Sci. 84:1797–1801 278 Bolton, F.J., A. D. Sails, A. J. Fox, D. R. A. Wareing and D. L. A. Greenaway. 2002. Detection of .3 279 Campylobacter jejuni and Campylobacter coli in foods by enrichment culture and polymerase 280 .chain reaction enzyme-linked immunosorbent assay. J. Food Prot. 65:760-767 281 EFSA. 2007. Trends and sources of zoonoses, zoonotic agents and antimicrobial resistance in the .4 282 European Union in 2005. European Food Safety Authority, Parma. Available at: 283 .http://www.efsa.europa.eu/en/efsajournal/doc/94r.pdf. Accessed 02 September 2013 284 EFSA. 2010. Analysis of the baseline survey on the prevalence of Campylobacter in broiler .5 285 batches and of Campylobacter and Salmonella on broiler carcasses in the EU, 2008. Part A: 286 Campylobacter and Salmonella prevalence estimates. The EFSA Journal, 8:1503. Available at: 287 .http://www.efsa.europa.eu/en/efsajournal/doc/1503.pdf Accessed 02 September 2013 288 EFSA. 2011. Scientific Opinion on Campylobacter in broiler meat production: control options and .6 289 performance objectives and/or targets at different stages of the food chain. EFSA Journal 2011 9, 290 141 pp. Available at: http://www.efsa.europa.eu/en/efsajournal/doc/2105.pdf Accessed 30 August 291 .2013 292 Forbes, KJ., F. J. Gormley, J. F. Dallas, O. Labovitiadi, M. MacRae, R. J. Owen, J. Richardson, N. .7 293 J. C. Strachan, J. M. Cowden, I. D. Ogden and C. C. McGuigan. 2009. Campylobacter immunity 294 and coinfection following a large outbreak in a farming community. J. Clin. Microbiol. 47:111- 295 .116 296 FSA 2004. Recommendations on the best practical procedures to sample and test poultry flocks for .8 297 Salmonella. FSA project Reference B15003, final report. Available at: 298 .www.ukmeat.org/pdf/FinalTechProjectReport.pdf Accessed 02 September 2013 299 Gormley, F. J., M. MacRae, K. J. Forbes, I. D. Ogden, J. F. Dallas and N. J. C. Strachan. 2008. .9 300 Has retail chicken played a role in the decline of human campylobacteriosis? Appl. Env. 301 .Microbiol. 74:383-390 302 Hansson, I., I. Vagsholm, L. Svensson and E. Olsson Engvall. 2007. Correlations between .10 303 Campylobacter spp. prevalence in the environment and broiler flocks. J. Appl. Microbiol. 304 .103:640–649 305 Krause, M., H. Josefsen, M. Lund, N. R. Jacobsen, L. Brorsen, M. Moos, A. Stockmarr and J. .11 306 Hoofar. 2006. Comparative, collaborative, and on-site validation of a TaqMan PCR method as a 307 tool for certified production of fresh, Campylobacter-free chickens. Appl. Env. Microbiol. 308 .72:5463-5468 309 Kudirkien, E., M. Malakauskas, A. Malakauskas, A.M. Bojesen and J.E. Olsen. 2009. .12 310 Demonstration of persistent strains of Campylobacter jejuni within broiler farms over a 1-year 311 .period in Lithuania. J. Appl. Microbiol. 108:868-877 312 Kusumaningrum, H. D., G. Riboldi, W. C. Hazeleger, R. R. Beumer. 2003. Survival of foodborne .13 313 pathogens on stainless steel surfaces and cross-contamination to foods. Int. J. Food Microbiol. 314 .85:227-236 315 McDowell, S. W. J., F. D. Menzies, S. H. McBride, A. N. Oza, J. P. McKenna, A. W. Gordon and .14 316 S. D. Neill. 2008. Campylobacter spp. in conventional broiler flocks in Northern Ireland: 317 .Epidemiology and risk factors. Prev. Vet. Med. 84:261-276 318 Mead, G., A. M. Lammerding, N. Cox, M. P. Doyle, F., Humbert, A. Kulikovskiy, A. Panin, V. P. .15 319 Do Nascimento, M. Wierup, and the Salmonella on raw poultry writing committee. 2010. 320 Scientific and technical factors affecting the setting of Salmonella criteria for raw poultry: a global 321 .perspective. J. Food Prot. 73:1566–1590 322 Moran, L., C. Kelly and R.H. Madden. 2009a. Factors affecting the recovery of Campylobacter .16 323 spp. from retail packs of raw, fresh chicken using ISO 10272-1:2006. Lett. Appl. Microbiol. 324 .48:628-632 325 Moran, L., P.J. Scates and R.H. Madden. 2009b. Prevalence of Campylobacter spp. in raw retail .17 326 .poultry on sale in Northern Ireland. J. Food Prot. 72:1830-1835 327 Moran, L., C. Kelly, M. Cormican, S. McGettrick and R. H. Madden, 2011. Restoring the .18 328 selectivity of Bolton broth during enrichment for Campylobacter spp. from raw chicken. Lett. 329 .Appl. Microbiol. 52:614-618 330 Olson, C. K., S. Ethelberg, W. van Pelt and R. V. Tauxe. 2008. Epidemiology of Campylobacter .19 331 jejuni infections in industrialized nations, p. 163-189. In I. Nachamkin, C. M. Szymanski and M. J. 332 .,Blaser, (ed.), Campylobacter third edition. ASM Press, Washington, DC 333 Pearson A.D., M.H. Greenwood, J. Donaldson, T.D. Healing, D.M. Jones, M. Shahamat, R.K.A. .20 334 Feltham and R.R . Colwell. 2000. Continuous source outbreak of campylobacteriosis traced to 335 .chicken. J. Food Prot. 63:309-314 336 Ridley, A., V. Morris, J. Gittins, S. Cathraw, J. Harris, S. Edge and V. Allen. 2011. Potential .21 337 sources of Campylobacter infection on chicken farms: contamination and control of broiler- 338 .harvesting equipment, vehicles and personnel. J.Appl. Microbiol. 111:233-244 339 Rosenquist, H., L. Boysen and B. Borck, 2008. Interventions to control Campylobacter in the .22 340 broiler production. Report of an International Expert Consultation, Copenhagen, Denmark, 26-27 341 November 2007. National Food Institute, Technical University of Denmark. Available at: 342 http://www.food.dtu.dk/~/media/Institutter/Foedevareinstituttet/Publikationer/Pub- 343 2008/Interventions%20to%20control%20Campylobacter%20in%20the%20broiler 344 .%20production.ashx Accessed 30 August 2013 345 Rosenquist, H., L. Boysen, C. Galliano, S. Nordentoft, S. Ethelberg and B. Borck. 2009 Danish .23 346 strategies to control Campylobacter in broilers and broiler meat: facts and effects. Epidemiol. 347 .Infect. 137:1742–1750 348 Shaw, M.K., A.G. Marr and J.L. Ingraham. 1971. Determination of the minimal temperature for .24 349 .growth of Escherichia coli. J. Bacteriol. 105: 683-684