UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Evaluation of indicator microorganisms in blueberry and raspberry production systems Aliphtiras, George J. 2001

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2001-0308.pdf [ 2.98MB ]
Metadata
JSON: 831-1.0089950.json
JSON-LD: 831-1.0089950-ld.json
RDF/XML (Pretty): 831-1.0089950-rdf.xml
RDF/JSON: 831-1.0089950-rdf.json
Turtle: 831-1.0089950-turtle.txt
N-Triples: 831-1.0089950-rdf-ntriples.txt
Original Record: 831-1.0089950-source.json
Full Text
831-1.0089950-fulltext.txt
Citation
831-1.0089950.ris

Full Text

EVALUATION OF INDICATOR MICROORGANISMS IN BLUEBERRY AND RASPBERRY PRODUCTION SYSTEMS by GEORGE J. ALIPHTIRAS B.Sc. (Microbiology) University of British Columbia, 1996 . A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES Food Science We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA May 2001 © George J. Aliphtiras, 2001 U B C Special Collections - Thesis Authorisation Form htrp://www.library.ubc.ca/spcoll/thesauth.html In presenting t h i s thesis i n p a r t i a l f u l f i l m e n t of the requirements for an advanced degree at the Un i v e r s i t y of B r i t i s h Columbia, I agree that .the Library s h a l l make i t f r e e l y a v a i l a b l e for reference and study. I further agree that permission for extensive copying of t h i s thesis for sc h o l a r l y purposes may be granted by the head of my department or by his or her representatives. It i s understood that copying or p u b l i c a t i o n of t h i s thesis for f i n a n c i a l gain s h a l l not be allowed without my written permission. Department of The U n i v e r s i t y of B r i t i s h Columbia Vancouver, Canada A B S T R A C T Recent outbreaks of foodborne disease have been attributed to the consumption of fresh produce. Enteric pathogens derived from animal and human faecal material have been identified as the dominant etiological agents in these outbreaks. The objective of this study was to identify significant sources of such microbial contamination in blueberry and raspberry production systems. Study participants (berry farms) were selected according to agronomic practices, including fertilization, harvesting, source of irrigation water and method of distributing irrigation water. The study was conducted over two berry-growing seasons. Enumeration of coliforms and Escherichia coli on fruit, fruit contact surfaces and environmental samples from growers and processors were conducted using traditional, violet red bile agar-based (TSA-VRBA M U G overlay, Petrifilm™ E coli count plates) and newly developed chromogenic assays (Chromocult® coliform agar and XM-G agar). The TSA-VRBAMUG overlay method, however, was found to be unsuitable and was discontinued after the first sampling year. Results suggested that contamination of berry fruit surfaces occurred at the grower level. Harvesting and processing practices did not contribute significant coliform bacteria (P > 0.05) to the fruit surface at the grower and processor levels. Based upon Kruskal-Wallis and multiple comparison analyses on E coli counts, ditch water-overhead sprinkler irrigation was deemed to be a significant source of microbial contamination in blueberry production. The inability to detect a difference in the sanitary or faecal indices of berry samples using coliform populations re-affirmed the unsuitability of coliform counts as indicators of sanitation for fresh produce. E coli isolates from berry samples were non-verocytotoxicogenic based upon PCR analyses. However, the frequency of E coli occurrence in berry samples indicates a potential risk from other enteric pathogens. Comparison of selective agars revealed that efficiencies of coliform and E coli recovery for CCA and PEC were significantly different (P < 0.05). However, E coli counts within the accuracy and estimated range on CCA and PEC correlated to each other (P < 0.05), likely due to the shared p-D-glucuronidase-based differentiation reaction. CCA is a suitable alternative to T S A - V R B A M U G overlay and PEC for coliform and E co//enumeration due to greater indicator recovery and differentiation capability. ii TABLE OF CONTENTS ABSTRACT ii TABLE OF CONTENTS iii LIST OF TABLES vi LIST OF FIGURES vii LIST OF APPENDICES viii LIST OF ABBREVIATIONS AND ACRONYMS ix ACKNOWLEDGEMENTS xi I. INTRODUCTION 1 II. LITERATURE REVIEW 3 2.1. Blueberry and raspberry industries in B.C 3 2.2. Foodborne disease and fresh produce 3 2.3. Indicator microorganisms 5 2.4. Sources of enteric pathogens 9 2.5. Microbiological examination of fresh produce 13 2.5.1. Selective agars for coliform and Escherichia coli enumeration: CCA, PEC and VRBA M U G 14 III. MATERIALS AND METHODS 18 3.1. Selection of study participants 18 3.2. Grower sampling during summer 1998 18 iii 3.2.1. Pre-harvest and post-harvest berry sampling.... 18 3.2.2. Leaf, equipment and water sampling 20 3.3. Processor sampling during summer 1998 '. 21 3.3.1. Pre-process and post-process berry sampling 21 3.3.2. Air, effluent and equipment sampling 22 3.4. Grower sampling during summer 1999 23 3.4.1. Post-harvest berry sampling 23 3.4.2. Soil and water sampling 24 3.5. Microbiological analyses 24 3.5.1. Sample preparation 25 3.5.2. Aerobe, coliform, Escherichia coli, yeast and mould enumeration in summer 1998 27 3.5.3. Coliform and Escherichia co//'enumeration in summer 1999 28 3.5.4. Escherichia coli enrichment 28 3.5.5. Coliform and Escherichia coli isolation 29 3.5.6. Identification of coliform and Escherichia coli isolates 29 3.6. Data analysis 30 IV. RESULTS AND DISCUSSION 32 4.1. 1998 sample analysis: enumeration of aerobes, coliforms and Escherichia coli 32 4.2. 1999 sample analysis: enumeration of coliforms and Escherichia coli. 42 4.3. Qualification of Escherichia coli 49 4.4. Identification of Escherichia coli isolates 51 V. CONCLUSIONS AND RECOMMENDATIONS 53 REFERENCES 55 APPENDIX 1 59 iv APPENDIX II APPENDIX III LIST OF TABLES Table 1. Coliforms and enterococci as indicators of food sanitary quality (Source: Jay, 2000) 8 Table 2. Differentiation of coliforms and E coli based upon (3-D-glucuronidase or p-D-galactosidase activity and possible fluorogenic or chromogenic enzyme substrates (Source: Manafi, 1992) 17 Table 3. Grower agronomic practices 19 Table 4. Medium utilization for aerobe, coliform, E coli, mould and yeast enumeration 26 Table 5. Populations of aerobes, coliforms and E coli on pre-harvest raspberries and blueberries from growers sampled in 1998 36 Table 6. Populations of aerobes, coliforms and E coli on post-harvest raspberries and blueberries from growers sampled in 1998 37 Table 7. Populations of aerobes, coliforms and E coli on berry bush leaves and harvesting equipment in 1998 38 Table 8. Populations of aerobes, coliforms and E coli on pre-processed raspberries and blueberries from processors sampled in 1998 39 Table 9. Populations of aerobes, coliforms and E coli on post-process raspberries and blueberries from processors sampled in 1998 40 Table 10. Populations of aerobes, coliforms and E co//'\n processing plant air, effluent and sorting belt samples in 1998 41 Table 11. Populations of coliforms and E co//on post-harvest raspberries and blueberries sampled from growers during the 1999 production season 45 Table 12. Populations of coliforms and E co//in irrigation water and soil obtained from raspberry and blueberry growers during 1999 sampling 46 Table 13. Absence or presence of fluorescence and confirmation of E a?//within grower and processor samples 50 Table 14. Serotype identity of E coli isolates from grower and processor samples 52 VI LIST OF FIGURES Figure 1. Sources and transmission routes responsible for fresh produce contamination 11 Figure 2. Relationship between PEC and CCA coliform plate counts within the accuracy (25 to 250 CFU) and estimated range (0 < and < 25 or 250 < CFU/plate) 47 Figure 3. Relationship between PEC and CCA E coli plate counts within the accuracy (25 to 250 CFU) and estimated range (0 < and < 25 or 250 < CFU/plate) 48 vii LIST OF APPENDICES A. 1. Populations of mould and yeast on raspberries and blueberries from growers and processors sampled in 1998 59 A. 2. Biochemical classification of coliform and presumptive E coli isolates from growers and processors sampled in 1998 60 A. 3. Coliform contribution of harvest practices of blueberry and raspberry growers (Wilcoxon signed rank test) 61 A. 4. Coliform contribution of processing lines of blueberry and raspberry processors (Wilcoxon signed rank test) , 62 A. 5. McNemar's test for significance of changes between CCA and PEC coliform and E coli counts 63 A. 6. Spearman's (p) correlation: rank of CCA and PEC coliform counts 65 A. 7. Spearman's (pj correlation: rank of CCA and PEC E coli counts 65 A. 8. Kruskal-Wallis test on CCA coliform counts from raspberry growers 66 A. 9. Kruskal-Wallis test on CCA coliform counts from blueberry growers 66 A. 10. Kruskal-Wallis test on CCA E coli counts from raspberry growers 66 A. 11. Kruskal-Wallis test on CCA E coli counts from blueberry growers 66 A. 12. Analysis of variance of ranked coliform data from raspberry growers 67 A. 13. Analysis of variance of ranked coliform data from blueberry growers 67 A. 14. Analysis of variance of ranked E a?//data from raspberry growers 68 A. 15. Analysis of variance of ranked E a?//data from blueberry growers 68 viii LIST OF ABBREVIATIONS AND ACRONYMS APHA American Public Health Association BCMAF British Columbia Ministry of Agriculture and Food BP phosphate buffered peptone °C degree Celsius CCA Chromocult® Coliform Agar CFU colony forming unit cm 2 centimetres squared (area) FCS food contact surface FDA Food & Drug Administration g gram GAP good agricultural practices glue fi-D-glucuronidase gene GMP good manufacturing practices GUD fj-glucuronidase h hour HACCP hazard analysis critical control point HGMF hydrophobic grid membrane filter L Litre I length, major axis of ellipse La ellipse surface area LSB M U G lauryl sulphate broth with 4-methylumbelliferyl-p-D-glucuronide u.L microlitre mL millilitre mm millimetre 4-MU 4-methylumbelliferone MUG 4-methylumbelliferyl-p-D-glucuronide ix Na 2S 2 0 3 sodium thiosulphate nm nanometre OD optical density PCA plate count agar PCAchi plate count agar with 0.01% (w/v) chloramphenicol PCR polymerase chain reaction PEC Petrifilm™ E coli count plate ppm parts per million (ug/mL) RTE ready-to-eat Salmon-GAL 6-chloro-3-indolyl-p-D-galactoside Tergitol®-7 sodium heptadecyl sulphate UBC University of British Columbia VRBA M U G violet red bile agar with 4-methylumbelliferyl-p-D-glucuronide w/v weight/volume (g/100 ml) X-GLUC 5-bromo-4-chloro-3-indolyl-p-D-glucuronide X ACKNOWLEDGEMENTS I. would like to thank Dr. Skura and Dr. Delaquis for their support during the study. Thank you for being my co-supervisors. I would like to also thank Dr. Copeman, the other member of my supervisory committee. Thanks to the support staff (Val, Sherman, Jeanette, Brenda and Angela) at Food Science for their technical help. Thanks to my fellow grad students (Ian, Andrea, Zoran, Bren, Azita, Hao and Tasos) and especially to Wendy for enduring my slave driver tendencies during the summer of 1998. In addition, a special thanks to Kim Ziebell at Health of Animals Laboratory in Guelph for performing the necessary serological analyses. I am indebted to the following agencies for their financial support, BC Blueberry Council, Lower Mainland Horticultural Industry Association, BC Investment Agriculture and Agriculture and Agri-Food Canada. Without their assistance, I would be a poor destitute grad student, but now I am only poor. Last but not least, I thank my parents, Demetre and Antonia, my brother Ari and my future sister-in-law Georgia for their love and support. A big thank you to all my friends, Greek (HSA) and non-Greek for all the good times - KAAA NA ITEPNATE! I. INTRODUCTION In the past decade, foodborne disease has continued to be a major concern for consumers, the food industry and regulatory agencies. Repeated outbreaks of gastroenteritis and other aliments caused by foodborne pathogens have helped perpetuate this concern. The emergence of Escherichia coli 0157:H7 as one of the etiological agents of foodborne disease and the life-threatening symptoms associated with diseases caused by this organism, have further heightened sensitivity to pathogens within our food supply. Fresh produce has long been recognized as a vehicle of transmission of foodborne pathogens. However, outbreaks of foodborne disease have only recently been attributed to fresh produce consumption. This has led to the re-evaluation of the safety of fresh produce. With the increasing concern regarding the safety of fresh produce, it is necessary to identify and to evaluate sources of contamination that are deemed significant health hazards. Contamination of fresh produce can occur at various stages during production, processing, storage and distribution. Production practices are responsible for introducing the majority of microorganisms to fresh and ready-to-eat (RTE) produce, such as blueberries and raspberries (Geldreich and Bordner, 1970; Beuchat, 1996). However, pathogen density within food is highly variable and is dependent upon the frequency of gastroenteritis within human and animal populations (Geldreich and Bordner, 1970). This high degree of variability could result in low pathogen incidence, which necessitates the utilization of indicator organisms. Coliform bacteria have been traditionally used as indices for the sanitary quality of food and water. Preference for coliforms has resulted from the observation that their incidence has been positively correlated with the sanitary quality of food and water. Coliform contamination occurs at the production level through natural events and through agronomic practice. Successful identification of sources of coliform contamination is necessary to ensure the safety and quality of blueberries and raspberries. Blueberry and raspberry production is of special concern to the B.C. economy because small fruit production represents a significant sector of the agriculture industry. The objective of the study was to evaluate the levels at which coliforms and E coll exist on blueberries and raspberries, and to identify factors that contribute most to overall berry sanitary quality. 1 The secondary objective was to compare the efficacy of coliform and £. co//enumeration techniques using fluorogenic and chromogenic substrate based direct plating methods. II. LITERATURE REVIEW 2.1. Blueberry and raspberry industries in B.C. Blueberry and raspberry cultivation have increased over the past several years because of strong consumer demand. The potential health benefit of consumption of anthocyanins (Bomser etal., 1996), potent antioxidants available in high amounts in blueberries (Kalt etal., 1999), has helped spur this increase in demand. Many parts of Canada, British Columbia, Ontario and Quebec, provide ideal growing conditions and marketed production and farm-gate value for blueberries have increased. B.C., which is the top producer of both berry types, produced 30% (14,700 tonnes) and 84% (13,200 tonnes) of Canada's blueberry and raspberry production in 1999, respectively (Canada, 2000). According to Statistics Canada (2000) data, farm gate values of B.C. blueberries and raspberries totalled $31.5 million and $24.1 million respectively, in 1999. The continued success of the berry industry in B.C. and in the rest of Canada is important due to its positive economic impact on the agricultural sector. 2.2. Foodborne disease and fresh produce Produce refers to a broad category of plant-derived products that includes any part of a plant, stem, root, fruit, leaf, bud, etc. Fresh produce has long been recognized as a potential vehicle for foodborne disease. Agronomic practices involving sanitation have changed little in the past decade; thus, the incidence of foodborne pathogens in fresh produce has likely remained static. Yet, the frequency of foodborne illness associated with these foods has increased. This apparent increase in frequency has risen because of several factors that include an increase in fresh produce consumption, the emergence of new pathogen variants with low infectious doses and an improvement in diagnostic and surveillance techniques (Tauxe etal., 1997). Pulse gel electrophoresis, PCR and serological techniques are examples of diagnostic techniques that have helped evaluate pathogen isolate relatedness and source reported outbreaks. An increase in reported outbreaks from all food groups would be expected if improved diagnostic techniques were solely responsible for the increase. Instead, only outbreaks associated with fresh produce have risen. Outbreaks attributed to consumption of fresh produce tripled to six percent from 1973 to 1991 and again increased substantially in 1995 (De Roever, 1999; Tauxe etal., 1997). 3 Therefore, foodborne disease outbreaks have likely increased as a result of greater fresh produce consumption; Canadian consumers have increased fresh fruit consumption by 15% in the past two decades (Canada, 1999), a statistic comparable to that reported for American consumers. The precise incidence of foodborne disease attributed to fresh produce remains unknown. This uncertainty likely results from disease underestimation. Difficulties encountered in outbreak investigations are responsible for this underestimation, including: wide dispersal of contaminated produce; intermittent, low levels of contamination; short product shelf-life; complex distribution systems; and failure to report cases of foodborne illness (Tauxe etal., 1997). Wide dispersal of contaminated product decreases the ability to identify and link related disease outbreaks from a common etiological agent. A short shelf life results in quick turn over with no leftover produce to confirm initial pathogen presence. Furthermore, produce distribution is quite complex, which leads to difficulties in isolation of contaminated produce lots. Most victims of foodborne illness exhibit mild gastroenteritis and neglect to report these cases to health officials, resulting in disease underestimation. Disease cases with severe symptoms, involving neurological, hepatic and renal syndromes, are predominantly reported. However, disease progression is dependent on the immune status of the consumer, inoculum size and pathogen type. Pathogens associated with fresh produce include bacteria, viruses and protozoa. Bacteria are considered to be of greatest concern due to their ability to repiicate prior to consumption and are the dominant identifiable etiological agent of foodborne illnesses (Tauxe etal., 1997). For bacteria to cause disease, they must be able to retain their viability within the food matrix. Inherent bacterial characteristics, physiological status of plant tissue, physical environment surrounding plant and bacteria and exposure to processing stresses will determine bacterial growth, survival and inactivation. RTE foods, such as blueberries and raspberries, are regularly eaten raw and do not receive a processing step designed to eliminate potential microbial hazards prior to consumption. Therefore, the reduction of microbiological hazards associated with berry products is of importance for ensuring continued consumer safety. 4 2.3. Indicator microorganisms The faecal-oral transmission route is primarily responsible for diseases associated with fresh produce (De Roever, 1999). Control of faecal contamination is necessary to reduce disease transmission, since human and animals faeces are the primary sources of enteric pathogens. Indicator bacteria are used to ascertain the degree of sanitation and the index of faecal and pathogen contamination of food. Indicator microorganisms are preferred over enteric pathogens because the latter typically have a low frequency and low concentration in fresh produce, making it difficult to quantify their presence without extensive sampling. The absence of enteric pathogens is not always indicative of a safe food product because not all foodborne pathogens are correlated with faecal contamination. Non-faecal correlated pathogens include Clostridium botu/inum, Clostridium perfringens, Bacillus cereus and Listeria monocytogenes. However, these pathogens are ubiquitous in nature, and thus are difficult to control. Ideal indicator microorganisms should closely emulate enteric pathogen characteristics. Jay (2000) suggests indicators should fulfill the following criteria: easily and rapidly detectable; easily distinguishable from other microflora; associated with enteric pathogens and faecal matter; detected when pathogens are present; population correlates well with enteric pathogen population; similar growth requirements, growth rates and kill-rates characteristic of enteric pathogens; and absence from foods that are enteric pathogen free. Coliforms, faecal coliforms, E coli, Enterobacteriaceae and faecal enterococci are five indicators of food safety; however, none of them fulfill all of the criteria listed. Therefore, indicator selection should be made in recognition of factors, including the determination of the enteric pathogen of concern, microflora constituents, food characteristics and any physical and chemical stresses potentially encountered by pathogens. The coliform group consists of Gram-negative, non-spore forming, aerobic and facultative anaerobic rods with the ability to produce gas within 48 hours at 32 - 37°C. Major coliform group constituents include, E coli, Citrobacter freundii, Enterobacter aerogenes, Enterobacter cloacae, and Klebsiella pneumoniae. Several of these species, including E coli, C. freundii, K. pneumoniae ss pneumoniae, naturally reside in the intestinal tract of mammals and birds, while E aerogenes and K. pneumoniae are inhabitants of soil. Limitations that restrict their application as indicators include multiplication at refrigeration temperatures, heat-sensitivity and lack of reliability as indicators of non-5 enteric pathogens. Numerous researchers have extensively debated the applicability of the coliform group as an indicator of poor sanitation. Spittstoesser (1970) concluded that the coliform test is suitable for milk and potable water, but not for fresh produce. Over 90% of the vegetable samples analyzed by Spittstoesser contained high levels of coliforms because of the inherent population of Enterobacter and Klebsiella spp. Thus, careful assumptions should be made on the safety index of food products when coliforms are used as indicators. Faecal coliforms are able to ferment lactose and produce gas at 44.5 °C. Their presence within food has been positively correlated with faecal contamination. However, not all group species are of enteric origin. Klebsiella spp. are thermo-tolerant and possess the defining faecal coliform group biochemical characteristics, but are of non-enteric origin. Greater selectivity is obtained with higher incubation temperatures. Examination of excrement from warm-blooded animals reveals that faecal coliforms comprise 93 - 98.7 % of total coliform content, of which E col/is the dominant constituent (Geldreich and Bordner, 1970). The group's sensitivity to adverse stimuli, ability to multiply on food and equipment surfaces are disadvantages. In some ways, coliforms are equivalent to faecal coliforms because both have similar drawbacks and habitats, but the latter is more indicative of faecal contamination because of increased method selectivity. The presence of E co//'is indicative of faecal and potential enteric pathogen contamination. However, many non-enteric pathogens are poorly indicated by E coli. Its absence may not reliably indicate a pathogen-free food. E co//has similar drawbacks reported for other faecal coliforms, specifically pertaining to assay sensitivities and potential growth. Not all strains of E coli are strictly indicator organisms; some can cause gastroenteritis. These disease causing variants are referred to as enteropathogenic E co//and several distinct strains exist: enteroaggregative (EAggEC), diffusely adherent (DAEC), enterotoxigenic (ETEC), enteroinvasive (EIEC), enterohemorrhagic (EHEC) and facultative enteropathogenic (EPEC) (Geraldine etal., 2000) The Enterobacteriaceae family is a broad group of bacteria used to indicate sanitary practices. Group members include many genera from faecal and non-faecal origin, soil and plant habitats. The ability to ferment glucose, instead of lactose, coupled with gas production are defining biochemical 6 characteristics of this group. Less selective growth requirements limits their use as indicators, and thus, they should only be used for products with low microbiological loads. The enterococci group has many species associated with humans, warm-blooded animals and birds. Two out of sixteen recognized species, E faecalis and E faecium, are commonly associated with food. Their ability to survive inhibitory physical stimuli better than coliforms makes them more attractive as indicators. Yet, the ability of enterococci to multiply on food, equipment surfaces and at refrigeration temperatures are drawbacks that limit their use as indicators. Of the five classes of microorganisms suggested as indicators, none fulfill all of the criteria as outlined by Jay (2000). Careful evaluation of indicator characteristics, as listed in Table 1, is required prior to final selection. The ease by which coliforms can be cultivated and differentiated makes them near ideal indicators for selected applications. However, the predominance of Enterobacterand Klebsiella spp., both coliform group constituents, on fresh produce restricts coliform determination as an indicator of microbiological safety because these species hold little sanitary significance in fresh produce (Nagle, 1999). Since coliform presence does not always indicate faecal or pathogen contamination, a more selective group composed of faecal coliforms and E col/is more appropriate for fresh produce. The Enterobacteriaceae family has similar drawbacks. Glucose, instead of lactose, fermentation is needed to allow growth of these bacteria, but at the expense of reduced selectivity. Enterococci based methods likely require further study because numerous enterococci hold little sanitary significance (Jay, 2000). Traditionally, food processors and regulatory agencies have used coliforms as an index of sanitation. The recent rise of food safety concerns has caused the expanded use of the coliform test to include evaluation of faecal and enteric pathogen in foods. This is an acceptable practice for water and processed foods such as dairy products, but this practice is not applicable to fresh produce. However, the food industry continues to use coliform-based methods likely because of the commercial availability of coliform test kits, test simplicity and government and industry standards. 7 Table 1. Coliforms and enterococci as indicators of food sanitary quality (Source: Jay, 2000). Characteristic Coliforms Enterococci Incidence in intestinal tract 107-109/g faeces 105-108/g faeces Incidence in faecal matter of various animal species Absent from some Present in most Specificity to intestinal tract Generally specific Generally less specific Occurrence outside of intestinal tract Common in low numbers Common in higher numbers Ease of isolation and identification Relatively easy More difficult Response to adverse environmental conditions Less resistant More resistant Response to freezing Less resistant More resistant Relative survival in frozen foods Generally low High Relative survival in dried foods Low High Incidence in fresh vegetables Low Generally high Incidence in fresh meats Generally low Generally low Incidence in cured meats Low or absent Generally high Relationship to foodborne enteric pathogens Generally high Lower Relationship to non-enteric foodborne pathogens Low Low 8 2.4. Sources of enteric pathogens The number and type of mesophilic microorganisms on fresh produce is highly variable and ranges from 10 3 - 10 9 CFU/gram (Nguyen-the and Carlin, 1994). The bacterial flora is epiphytic in nature and is mainly composed of Gram-negative rods (80 - 90%) (Nguyen-the and Carlin, 1994), while pathogens typically represent only a small portion of the microflora. Pseudomonas spp., Enterobacter spp., and Erwinia spp. are the most prevalent Gram-negative bacteria; frequent Enterobacteriaceae family members include Enterobacter agg/omerans, Erwinia herbicola and Rahnella aquatilis (Nguyen-the and Carlin, 1994). The composition and population of the microflora is largely determined during production, while processing generally alters the size of the microflora population (De Roever, 1999; Nguyen-the and Carlin, 1994). Contamination of the epiphytic surface requires direct contact with irrigation water, animals, insects, harvest equipment, air (dust), farm labourers and soil (Beuchat, 1996). Therefore, it is necessary to evaluate all stages in production systems to identify significant sources of enteric pathogens. The primary source of enteric pathogens is human and animal faeces. Numerous vehicles for their transmission have been identified and are typically common for all produce types (Figure 1.0). Raw agricultural products, such as blueberries and raspberries, are susceptible to occasional pathogen contamination because of close proximity to sources and vehicles of faecal matter. Practices that involve direct and indirect contact between faecal matter and fresh produce are responsible for introducing enteric pathogens. Complete elimination of enteric pathogens from produce is difficult. However, control measures can be implemented to reduce the frequency of gross faecal contamination, which poses an unacceptable health risk. Faecal transmission in production systems can occur through natural means and as a result of agronomic practices. Feral animals, especially birds, are significant contributors to faecal contamination of cultivated berries. Bird excrement has been shown to contain Salmonella spp. and enteropathogenic E coli at an incidence rate of 6.3% (Geldreich and Bordner, 1970) and between 0.9 to 2.9% (Beuchat and Ryu, 1997; Wallace etal., 1997), respectively. Pathogen incidence rates are variable and are dependent on the source of food consumed by birds. Faecal contamination occurs from direct defecation onto the plant. Insects are considered to contribute minimally to faecal contamination. Faecal coliforms 9 are not typically part of the inherent microbial flora of insects. Transient contamination occasionally occurs when insects come in contact with faecal matter or consume contaminated foods. Thus, insects may function as vehicles of pathogen transmission, while birds serve as a reservoir. Soil is an inhospitable environment for transient microorganisms not normally accustomed to such a harsh environment. Uncultivated soil, far from human habitation, normally has a low frequency of faecal contamination (De Roever, 1999). Contamination of uncultivated soil occurs through natural means by contact with faeces from feral animals. Enteric pathogens are normally relatively immobile at the site of defecation and require a vehicle, such as water, for their dispersal. Uncultivated soil is considered to be a poor source of enteric pathogens. However, not all foodborne pathogens are enteric in origin and some of these may be present in uncultivated soil. These non-faecal pathogens naturally reside within soil, but require specific conditions, rarely encountered, to present a serious health hazard. In contrast, cultivated soil contains enteric pathogens at a much higher frequency (Geldreich and Bordner, 1970; Roberts etal., 1982). The degree of contamination of cultivated soil is dependent on pre-production factors and agronomic practices. Irrigation, fertilizer and harvest practices have a direct effect on soil microflora. The use of land for intensive animal grazing leads to the concentrated deposition of faeces. Periods of flooding or heavy rainfall may result in the dispersion of faeces. Enteric pathogen survival in soil is dependent on the extrinsic properties of the soil, such as, soil type, temperature, rainfall, sunlight, foliage density and the surrounding microflora (Roberts etal., 1982). Survival times for enteric pathogens can range from days to months. However, favourable soil conditions may prolong survival of enteric bacterial for two months and longer (Geldreich and Bordner, 1970). Water is a potentially significant source of faecal contaminants. The diverse functional aspects of water at the agricultural level include irrigation, equipment, hand washing, produce washing, pesticide and fertilizer delivery. Water contamination can occur through point and non-point means. Point refers to direct introduction of sewage effluent, while non-point refers to the incorporation of faecal contaminated floodwater or ground water run-off from indirect sources. Irrigation methods that provide for maximum contact between produce and water can result in high levels of contamination (De Roever, 1999; Norman and Kabler, 1953). Therefore, produce irrigated by overhead irrigation can have substantially higher levels of contamination than those irrigated by drip irrigation (De Roever, 1999; 10 Figure 1. Sources and transmission routes responsible for fresh produce contamination. (Source: Beuchat, 1996). 11 Pabrua, 1999). Drip irrigation systems consist of a rubber tubing delivery system at the base of each plant, which allows for minimal produce-water contact. The source of water and the irrigation method used ultimately determines the extent of contamination. Produce physical attributes determine the epiphytic surface area on fruit available for enteric bacteria attachment. Greater surface area will allow for increased attachment. The faecal coliform indices of soil and produce typically reflect the faecal coliform content of irrigation water (Norman and Kabler, 1953). The time of irrigation water application, in relation to harvest, is another variable that may alter enteric pathogen load on produce. Geldreich and Bordner (1970) suggest produce irrigation should occur a minimum of four weeks prior to harvest, unless potable water is used. The use of potable water would help to alleviate the frequency of contaminated produce. Application of fertilizer to replenish diminishing nutrients in soil can provide a source of enteric pathogens. Raw and partially processed manures are known to harbour these microorganisms. Dairy cattle manure is a common source material for fertilizer production; therefore, evaluation of the prevalence of enteric pathogens is necessary for risk reduction. The prevalence of enteropathogenic E C0//'O157:H7 in dairy cattle herd populations is estimated to be less than 1 to 5%; however, young heifers and calves are more inclined to chronically shed this pathogen on an intermittent basis (Pell, 1997). More strict manure processing has been suggested because of the recent discovery that E coli 0157:H7 is able to survive longer than the 60-day holding period used in aging (Wachsmuth, 1997). Safety concerns associated with manure-based fertilizers has resulted in the use of chemical fertilizers, which are not typically regarded as a significant source of enteric pathogens. The inherent low water activity of concentrated chemical fertilizers is inhibitory to bacterial growth. However, the use of non-potable water for fertilizer delivery can introduce enteric pathogens on epiphytic surfaces. Dilute fertilizers and several pesticides (Blank etal., 2000), when not immediately applied to the field, can serve as a substrate for bacterial growth. Substrate availability will allow for pathogen growth and subsequent application of the fertilizer or pesticide will ultimately lead to contamination of epiphytic surfaces. Food contact surfaces (FCS) can serve as vehicles for transmission of enteric pathogens. Produce contamination by way of contact with unsanitary FCS can occur from produce production to final distribution. Agronomic practices that involve produce handling by either human or machine harvesters 12 can serve as points of significant contamination. Human harvesters are known to be a primary source (via the gastrointestinal tract) and vehicles of enteric pathogen distribution. Therefore, to reduce the risk of pathogen transmission, it is necessary for human harvesters to practice good personal hygiene. However, individual personal hygienic practices can vary. To counteract this variability and reduce the overall risk of pathogen transmission, Bracket (1999) and Beuchat (1996) suggest the implementation of hygienic practices involve the following: • Provide adequate hand-washing and sanitary facilities near the area of work. • Train personnel in personal hygiene, i.e. hand-washing techniques. In conjunction with improved personal hygiene, cleaning and sanitation of machine harvesters and other equipment that has direct contact with produce is necessary. Otherwise, the use of heavily contaminated harvesters and other FCS can result in gross contamination of produce as observed by Splittstoesser (1970). Fresh produce is often consumed without application of treatments for inactivation of pathogens. Therefore, production of microbiologically safe produce is necessary to reduce the frequency of foodborne illness. Investigation of where and how contamination occurs within production systems will help to identify microbial hazards. Implementation of good agricultural practices (GAP), a derivative of good manufacturing practices (GMP), is recommended to reduce microbial contamination (Pabrua, 1999). GAP are defined by the FDA as procedures used to reduce microbial safety hazards in production system constituents, such as, growing, harvesting, sorting, packing and storage. Safer fresh produce will be the end result of GAP implementation. 2.5. Microbiological examination of fresh produce There are four basic enumeration methods available to food microbiologists, but only two are commonly employed. These include both direct and indirect methods. Direct plating methods provide estimates of the number of viable cells within a sample; indirect methods statistically estimate the number of viable cells using a most probable number (MPN) technique. MPN offers greater enumeration sensitivity than direct plating methods, especially when indicators are at levels of < 10 per g. However, large amounts of glassware are needed for MPN determination; therefore, direct plating methods, 13 including spread and pour plate techniques, have gained in popularity. Both techniques are limited by the probability of overlapping colony growth, but this can be compensated for by the dilution scheme. Assay sensitivity is dependent on the volume of inoculum delivered; however, direct methods are insensitive in comparison to MPN procedures. Quantitative analysis of indicator bacteria is necessary to assess the sanitary and faecal indices of food and FCS. Indicator enumeration requires the use of selective media. Otherwise, non-indicator bacteria will compete for nutrients available in the media, leading to inaccurate estimation of the indicator population. Selective media contain compounds that facilitate the selection and differentiation of indicator versus non-indicator bacteria. Selective growth of indicators, such as, coliforms, faecal coliforms, E co//and Enterobacteriaceae, is obtained by the addition of surface-active-agents and dyes that inhibit other microorganisms, excluding their own. However, selective agents can be detrimental to the growth of indicator bacteria under certain conditions. Food ingredients can increase or decrease the selection pressures encountered by bacteria within a sample. Extremes in pH, a w , temperature, nutrient scarcity, etc., will cause physiological stress and decrease the percent recovery of indicator microorganisms (Ray, 1989). The type and number of non-indicator microorganisms that are part of the epiphytic microflora can interfere with indicator colony development and interpretation of data. Coliforms are the indicator of choice for the food industry. The ease with which coliforms can be cultivated, availability of rapid and inexpensive enumeration materials and their high correlation with sanitation indices, makes them near ideal indicators for many processed foods (Splittstoesser, 1983). However, coliforms are not good indicators of faecal and enteric contamination on fresh produce; instead, E co//is better suited for this task (Jay, 2000; Bell and Kyriakides, 1998). Many food distributors and secondary processors of fresh produce require primary processors, such as blueberry and raspberry processors, to conduct coliform and E coli counts. This has resulted in the implementation of indicator enumeration by many processors to meet supplier demands. 2.5.1. Selective agars for coliform and Escherichia coli enumeration: CCA, PEC and V R B A M U G The recent incorporation of chromogenic and fluorogenic substrates in media have increased the differentiation abilities of selective agars. Violet red bile agar (VRBA) supplemented with 4-14 methylumbelliferyl-p-D-glucuronide (MUG), Petrifilm™ E coli count plate (PEC) and Chromocult® coliform agar (CCA) are examples of these newly formulated selective agars. VRBA M U G is a derivative of VRBA, but contains the fluorogenic substrate MUG. E coli colonies that produce p-glucuronidase (GUD) cleave MUG to 4-methylumbelliferone (4-MU), which is responsible for the blue fluorescent halo around E coli colonies when viewed under UV light. GUD expression is specific to E coli, with over 96% of isolated strains expressing GUD activity (Feng and Hartman, 1982). However, fluorescence rapidly diffuses into the surrounding medium, necessitating plate interpretation between 12 to 24 h post-incubation. Incorporation of bile salts and crystal violet in VRBA formulation inhibits growth of non-Enterobacteriaceae. Addition of lactose encourages growth of lactose fermenters, which helps in colony differentiation. When lactose utilization occurs colonies appear dark-red with zones of bile salt precipitation. The American Public Health Association (APHA: Hitchins etal., 1992) recommends the use of VRBA for the enumeration of coliforms for water, sewage and dairy product testing. MUG supplemented VRBA permits enumeration of E co//and coliforms onto one medium. PEC is a derivative of VRBA agar but in a dry film form. PEC is based on the encapsulation of culture media and a cold-water-soluble gel onto two plastic films. As in VRBA, coliform and E coli selectivity in PEC is obtained with the addition of bile salts and crystal violet. Therefore, gas associated-red colonies are interpreted as coliforms, while colonies with a blue-halo are confirmed as E coli. E coli differentiation is based on its inherent GUD activity. The encapsulation of 5-bromo-4-chloro-3-indolyl-p-D-glucuronide (X-GLUC), a chromogenic substrate, and its subsequent catabolysis into indoxyl, result in the characteristic blue-coloured halo around E coli colonies. CCA is a selective chromogenic agar that is a suitable alternative for coliform and E coli enumeration. To inhibit the growth of non-Enterobacteriaceae, Tergitol®-7 (sodium heptadecyl sulphate), a surface-active-agent, is added to the agar formulation. The presence of pyruvate, a phosphate buffer system and sorbitol, encourage the growth of sublethally injured Enterobacteriaceae (manufacturer's literature). Pyruvate and phosphate facilitate the repair of freeze-injured E co//(Ray, 1989). The differentiation ability of CCA is greater than VRBA M U G and PEC because of its ability to differentiate between E coli, coliforms, other Enterobacteriaceae and several Salmonella spp. Differentiation is based on chromogenic reactions involving two chromogenic substrates, 6-chloro-3-15 indolyl-p-D-galactoside (Salmon-GAL) and 5-bromo-4-chloro-3-indolyl-p-D-glucuronide (X-GLUC: Manafi, 1992). The end result of substrate catabolysis is that coliforms appear salmon to red coloured while E co//appears dark-blue to violet coloured. The chromogenic products do not diffuse within the media, as does GUD (Frampton etal., 1988). Therefore, interpretation of coliform and E coli colony forming units (CFU) is made more convenient on plates of CCA. 16 Table 2. Differentiation of coliforms and £. coli based upon p-D-glucuronidase or p-D-galactosidase activity and possible fluorogenic or chromogenic enzyme substrates (Source: Manafi, 1992). Substrate End Product Colour 5-bromo-4-chloro-3-indolyl-p-D-glucuronide (X-GLUC) dibromo-dichloro-indigo (indoxyl derivative) Blue methylumbelliferyl-p-D-glucuronide (MUG) 4-methylumbelliferone (4-MU) Blue fluorescence 6-chloro-3-indolyl-p-D-galactoside (Salmon-GAL) dichloro-indigo (Salmon derivative) Pink to red III. MATERIALS AND METHODS 3.1. Selection of study participants Ten berry growers and processors were selected for sampling during the 1998 and 1999 summer harvests. Participant sampling was conducted over the two summer seasons, unless otherwise indicated. Selected participants included seven growers and three primary processors of either raspberries or blueberries. Growers were defined as farmers involved in either raspberry or blueberry production, while primary processors were involved in berry fruit washing, sorting, packing and freezing. The 1998 sampling consisted of one raspberry grower, one raspberry processor, two blueberry growers and two blueberry processors, while the 1999 sampling consisted of three raspberry growers and four blueberry growers. Processor participants were omitted from the 1999 sampling. Justification of processor exclusion was based on the analysis of trends from summer 1998 data and is further discussed in the Results and Discussion section. Participants were selected according to pre-determined criteria and recommendations from the Berry and Nut Specialist from the BCMAF. Criteria used for selection of growers were as follows: farm location, source of irrigation water, irrigation method, source of fertilizer and harvest practices. Attempts were made to select growers that differed from one another in at least more than one of the fore-mentioned criteria, as outlined in Table 3. In comparison, criteria used for processor selection were relatively simple. Processor requirements were as follows: Fraser Valley location, process fruit only from Fraser Valley growers, and considered as significant processors by the BCMA Berry and Nut Specialist. 3.2. Grower sampling during summer 1998 3.2.1. Pre-harvest and post-harvest berry sampling Collection of berry samples began at the end of June and proceeded until the end of September. Sample collection coincided with the harvest schedule determined by each grower. A typical sampling in 1998 consisted of the collection of a minimum of eight berry samples. These samples were partitioned 18 Table 3. Grower agronomic practices. Grower Location3 Irrigation H20 Fertilizer11 Harvest6 Source11 Distribution0 Practices R-a f g Fraser Valley Aquifer Overhead Manure Hand R-p9 Fraser Valley Aquifer Overhead Manure Hand/Machine R-Y9 Fraser Valley Aquifer Overhead Chemical Machine B - a f g Fraser Valley Municipal Drip/Trickle Chemical Hand B -p f 9 Fraser Valley Ditch Overhead Chemical Machine B-Y 9 Fraser Valley Aquifer Overhead Chemical Hand B -5 9 Fraser Valley Ditch Drip/Trickle Chemical Hand a Farms from either Abbotsford, Abbotsford-Matsqui, Langley and Pitt Meadows. b Aquifer, ditch, or municipal water sources. c Drip/trickle, or overhead irrigation systems. d Manure, or chemical fertilizers. e Hand, machine, or both for harvest practices. f 1998 participant. 9 1999 participant. into two sample subgroups, a pre-harvest and a post-harvest group. A pre-harvest sample referred to berries collected by the researcher just prior to harvest, while a post-harvest sample referred to berries collected from either a machine harvester or the farm's hired labourers. Berry samples were partitioned into subgroups in an attempt to quantify the contribution of harvest practices on berry microflora. Attempts were taken to ensure rows selected for post-harvest samples corresponded to the rows from which pre-harvest berry and leaf samples were collected. Rows of berry bushes were randomly assigned to researchers. Berry samples were collected manually, using vinyl chloride gloves sanitized with 70% (w/v) ethanol to prevent microbial contamination from the researcher's hands. A new pair of sanitized gloves was worn for each sample collection to prevent any carry-over of microorganisms from preceding samples. Sterile sampling bags (Baxter Scientific with write-on strip/Weber Scientific "Whirl-Type" with write on strip) were used for each sample collection. Fruit was collected from each row until sampling bags were filled without inducing physical damage to the berries (approximately 250-350 g of fruit, depending on berry variety). Full bags were stored on frozen ice packs in a styrofoam cooler after collection (1-2 h) and during transit, approximately 1-1.5 h, from the Fraser Valley to the Food Science Building at UBC. After arrival at UBC, samples were stored at 4°C in a walk-in cooler for a period of 12-24 h until microbiological testing could be conducted. 3.2.2. Leaf, equipment and water sampling Leaves were collected from raspberry canes or blueberry bushes in rows designated for fruit sampling. Care was taken to avoid physical damage to the leaves. Sanitized vinyl chloride gloves were worn when collecting leaves. Leaf samples were stored on ice packs during transit and under refrigeration after arrival at UBC. A total of three leaf samples were collected over the entire study. Further leaf sampling was halted as a result of limited laboratory resources. The sponge contact method was used for equipment swabbing. Generic sponges, 10 cm x 7 cm x 1 cm, without anti-microbial additives were purchased from a neighbourhood grocery store. In preparation for environmental swabbing, the sponges were soaked in distilled water to remove residual detergents and were squeezed to remove excess water. Sponges were then individually wrapped in aluminium foil and autoclaved to ensure sterility. Generic sponges were later replaced with a commercial sponge kit (Nasco Whirl-pack® "speci-sponge" bags with one 18 oz sponge). Ten ml of sterile 0.1% (w/v) peptone water was added to each sponge just prior to use. Once the added peptone was distributed evenly, the researcher swabbed the equipment while wearing sanitized gloves. Used sponges were placed in sterile sampling bags and stored in styrofoam coolers under conditions used for berry and leaf samples. Two swabs were collected during the grower 1998 sampling. One sample was from a pail used for berry collection and the other from a belt loader on a machine picker used to load totes. Totes were the containers used to collect and store harvested berries. Swabbed areas measured 490cm 2 and 342cm 2, respectively. Water samples were collected from farms that utilized aquifer or ditch water for irrigation. Farms that utilized municipal water were omitted from sampling because governing regulations require potable water to be free of all coliforms. Opened sampling bags were submerged into ditch water, or held underneath spigots at water pump stations. A 2 m long plastic pole, with a small plastic beaker (10 cm (diameter) x 10.5 cm (height)) attached to the end of the pole, was used for collecting samples from inaccessible ditches. The attached beaker was rinsed briefly with 70% ethanol prior to water sample collection. Residual ethanol within the beaker was removed by allowing sufficient time for evaporation and by submerging the container into the ditch water prior to sample collection. The container was once again submerged, filled and removed. Approximately 300 ml of water was then poured into each sterile sample bag. Water samples were stored upright in a styrofoam cooler under conditions used for berry, leaf and equipment swab samples, during transit and storage. 3.3. Processor sampling during summer 1998 3.3.1. Pre-process and post-process berry sampling Grower identity, farm location and harvest practices were recorded for each sample taken to ensure participants met study requirements, sample differentiation and data analysis. However, the production staff was not always able to provide such information because of the hectic work environment. When such a situation was encountered, sample collection proceeded as normal, but 21 information regarding agronomic practices was not available. Samples were partitioned into two subgroups, a pre-process and a post-process sample group. A pre-process sample referred to fruit collected from bins delivered directly from the fields, while a post-process sample referred to fruit that had been sorted and washed with chlorinated or non-chlorinated water, prior to sample collection. The only exception arose with processor P-y, on Tuesday, August 17,1998, which had only dry sorting lines in operation. Sample partitioning was used in an attempt to quantify whether processing lines contributed to the overall microbiological quality of the fruit. Berry samples were collected manually, using sanitized vinyl chloride gloves. Eight samples were collected during every sampling. Pre-process samples were collected from totes typically stored outside the plant for several hours because of space limitations within the plant. Collection of post-process fruit samples occurred at the post-sorting stage along the processing line and at the end of the sotting belt, just prior to box packaging. Selection of berry sampling sites was based on processing line accessibility. Enough fruit was collected from each site until the sampling bags were filled, approximately 250-350 g of fruit depending on berry variety. 3.3.2. Air, effluent and equipment sampling A solid agar surface impaction method was used for microbiological analysis of plant air quality. Quantitative determination of total aerobes, coliforms, and yeast and mould were determined using petri-plates (100 x 15 mm) filled with plate count agar (PCA), violet red bile agar supplemented with MUG (VRBA M U G ) and PCA supplemented with 0.01% choramphenicol (PCAo,i), respectively. A total of nine air samples were collected during summer 1998 sampling. Air exposure times ranged from five to thirty minutes. Air sampling was conducted at two locations along the processing lines within each processing plant. The areas surrounding the loading hopper and near the end of the sorting belt were selected as sites suitable for air sampling because of low worker traffic. Petri-plates, one from each agar type, were placed with their lids opened at the selected sampling sites. Once air samples were collected, the petri-plates were stored at ambient temperatures away from direct sunlight during transit to UBC. Plates required no further preparation after air sample collection, which resulted in their immediate incubation at the appropriate temperature post arrival at UBC. 22 Analysis of effluent was included to assess the residual microflora from epiphytic surfaces. Effluent represents the wastewater from the chlorinated or non-chlorinated washing steps used for berry fruit washing. Sterile sampling bags were used to collect approximately 300 ml of effluent. One ml of filter sterilized (Millex-ha, 0.45 pM sterile filter unit) 1.0% (w/v) Na 2 S 2 0 3 was added to each collected effluent sample in order to reduce any residual oxidative chlorine compounds, which may have led to underestimation of total colony forming units (CFU). A total of five effluent samples were collected during the 1998 sampling. Collected samples were stored under conditions outlined in section 3.2.2 for irrigation water. Samples were not collected during 1999 because of exclusion of processors from the study at the time. The sponge contact method was used for equipment swabbing as outlined in section 3.2.2. A total of seven equipment swabs were collected from participating processors. Sorting belts along the processing lines between the sorting and packing stages were selected for swabbing because of easy access. The length of the belt in contact with the swab was recorded for each sample taken to enable calculation of the total area swabbed. Used sample sponges were stored under conditions outlined in section 3.2.2. 3.4. Grower sampling during summer 1999 3.4.1. Post-harvest berry sampling Collection of berry samples proceeded as in summer 1998, from the end of June until the end of September. As before, sample collection coincided with the harvest schedule determined by each grower. The sample regime involved collection of eight berry samples. Samples were not partitioned into pre-harvest or post-harvest groups. Instead, only post-harvest fruit, harvested using either machine or hand picked by farm labourers, were collected using sanitized vinyl chloride gloves. Approximately, 250-300 g of fruit were collected in each sampling bag. Full sampling bags were stored on ice packs immediately following collection and during transit to UBC. Upon arrival at UBC, samples were refrigerated overnight prior to microbiological analysis. 23 3.4.2. Soil and water sampling Rows designated for fruit harvest were selected for soil sample collection. A potting shovel was used to collect the top few centimetres of topsoil, less than 5 cm in soil depth, as is required for sampling using the thin surface horizon protocol (Carter, 1993). Prior to each sampling session, the potting shovel was washed with dishwashing soap (Palmolive®, Colgate-Palmolive Inc.) and water to remove any visible remnants of soil particulates to reduce the likelihood of cross-contamination. The shovel was later sanitized with 70% ethanol before sample collection. Soil beneath the plant foliage was targeted for sampling because of the close proximity to the mature berry fruit above. Approximately 100-200 g of soil was transferred to each sampling bag. Full bags were stored at ambient temperatures and away from direct sunlight during transit and storage to preserve soil humidity. Water sample collection proceeded as outlined in section 3.2.2. 3.5. Microbiological analyses A total of 291 samples were collected and analysed for coliform and E coli content during the summer harvests of 1998 and 1999. Sample analysis was initiated within 24 h of collection. Direct plating techniques, such as spread plate and pour plate methods were employed for microorganism enumeration during 1998 and 1999, respectively. A most probable number technique, the ISO-GRID HGMF (QA Life Sciences Inc.) VRBA M UG (Fluorocult®, Merck KGaA, Darmstadt, Germany) method was used in preliminary laboratory studies, but was found to be unsuccessful because of low sample analysis output and the inability of the method to allow for the observance of positive fluorescent reactions. The ISO-GRID HGMF method was used for the analysis of both soil and water samples for summer 1999, once XM-G agar (Nissiu Pharmaceutical Co., Japan) replaced the previous VRBA M UG- The continued use of the ISO-GRID HGMF method was advantageous for water analysis because inoculum volumes of greater than one ml could be accommodated in order to increase sensitivity of the assay. Concerns about microorganism injury induced by exposure to molten agar resulted in the selection of the spread plate method for sample analysis in 1998. However, it was later replaced with the pour plate method in 1999 because of the desire to increase assay sensitivity. The presence of 24 pyruvate in the CCA (Merck KGaA, Darmstadt, Germany) formulation helped alleviate previous concerns about injury. Standard enumeration media (PCA, PCAchi, VRBA M U G and PEC (3M Inc., Minneapolis, MINI)) were selected for summer 1998. PEC utilization was intermittent and was used to confirm corresponding coliform and E coli counts derived from VRBA M U G plates. CCA and XM-G replaced VRBA M U G in summer 1999 because the latter was found to be inadequate for coliform enumeration on fresh produce. Justification of its replacement is discussed in the Results and Discussion section. The use of XM-G agar was limited to soil and water sample analyses. Sparingly used in the previous year, PEC utilization was expanded to include all collected samples, thereby allowing the comparison of media (CCA and PEC) efficiencies for coliform and £. co//enumeration. Enumeration of aerobes, yeast and mould were initially included as part of the analyses in 1998; however, it was decided that continued inclusion of those analyses in summer 1999 would unnecessarily consume limited laboratory and human resources. 3.5.1. Sample preparation Berry samples were removed from refrigeration just prior to analysis. Approximately 250 g of berries were blended with 2250 ml of sterile 1.0% (w/v) phosphate buffered peptone (pH 7.0; BP: Szabo, 1997) in a stomacher bag (Seward Stomacher® Model 3500 Bags 6042). Weights of berries and of the BP were recorded for accurate determination of the initial dilution. Full stomacher bags were processed in a stomacher (Seward Stomacher® 3500 Lab System) for a period of two minutes on medium or high speed for raspberry and blueberry samples, respectively. All berry homogenates were prepared with 1.0% (w/v) BP to protect against pH injury of microorganisms upon release of organic acids. Leaves were transferred with the aid of sterile stainless steel forceps into stomacher bags (Seward Stomacher® Model 400 Bags 6041) from the refrigerated leaf sampling bags. The number of leaves transferred and the weight of the leaves in the stomacher bag were recorded. Approximately 30 g and 1.0 L of 1.0% (w/v) BP were added to each stomacher bag. The weight of the BP was recorded for accurate determination of the initial dilution. Full stomacher bags were processed (Seward Stomacher® 400 Lab System) for one minute. Approximation of total leaf surface area was determined by recording the lengths (I) and widths (w) of a minimum of twenty or more leaves randomly selected from a previous Table 4. Medium utilization for microbiological analyses. Medium Medium Utilization Year Sample Microorganisms C C A 1999 Berry Coliforms, E. coli LSBMUG 1999 Berry Coliforms, E. coli PCA . 1998 Berry, Environmental Aerobes PCA C h | 1998 Berry, Environmental Mould, Yeast P EC 1998,1999 Berry Coliforms, E. coli T S A - V R B A M U G overlay 1998 Berry, Environmental Coliforms, E. coli VRBAMUG 1998 Environmental Coliforms, E. coli XM-G 1999 Environmental Coliforms, E. coli 26 sample. The /and k/variables corresponded to the major and minor axes from an ellipse and were used to calculate ellipse surface area, as outlined in Equation 1. Equation 1. La-n I w/2 Average values of /and ivwere calculated and were subsequently used to determine the average leaf surface area (La). The La value was multiplied by the number of leaves within the stomacher bag to determine total leaf surface area. Sponges used for equipment swabbing were aseptically transferred from sampling bags into sterile stomacher bags. Approximate volumes (150 ml) of 1.0% (w/v) BP were added into each stomacher bag. The weight of the added BP was recorded for initial dilution determination. Prepared stomacher bags were subjected to stomacher agitation for one minute (Seward Stomacher® Model 400). Processing plant effluent and irrigation water samples were not subjected to any specific preparation procedures, except for sample agitation prior to dilution set-up or HGMF inoculation. 3.5.2. Aerobe, coliform, Escherichia coli, yeast and mould enumeration in summer 1998 The blended samples were used to prepare serial dilutions with sterile 0.1% (w/v) peptone ranging from 10"2 to 10"4. Wide mouth pipets were used for all aliquot transfers from the berry homogenate to avoid pipet tip obstruction by suspended berry particulates. Aliquots of 150 u.L from each dilution were pipetted into duplicate sets of PCAchi and PCA petri-plates. To increase protocol sensitivity for coliforms and E. coli, an inoculum of 0.5 ml versus the previous 150 ^ L , was used for TSA petri-plate inoculation, while aliquots of 1.0 ml were used for PEC. Sterile glass spreaders were used to distribute the delivered inoculums evenly across the agar surfaces. Inoculated petri-plates and PEC were incubated for 24 h and 48 h, respectively. However, prior to final incubation at 35°C, TSA petri-plates were overlaid with V R B A M U G after a two-hour incubation at room temperature. The inclusion of a two-hour-incubation period of TSA plates prior to V R B A M U G exposure allowed for recovery of injured coliforms and £. coli'Way, 1989; Speck etal., 1975). After 18 to 20 h of incubation, T S A - V R B A M U G overlay plates were 27 viewed under UV light (312 nm: FOTO/Prep® UV Transilluminator, Fotodyne Inc., Hartland, WI) to test for fluorescence. 3.5.3. Coliform and Escherichia coli enumeration in summer 1999 Serial dilutions ranging from 10"2 to 1 0 3 were prepared with sterile 0.1% (w/v) peptone. Aliquots (1.0 ml) from each dilution were used to inoculate duplicate sets of CCA and PEC plates. Subsequent to agar solidification, inoculated plates were incubated at 35°C for 24 h, while PEC were further incubated for 24 h to differentiate between coliforms and E coli. Ten grams of soil sample was diluted in 95.0 ml of sterile 0.1% (w/v) peptone (10 1 ) . A 10*2 serial dilution was also prepared. The dilutions were shaken vigorously and large soil particulates were allowed to settle prior to inoculum removal. Aliquots of 1.0 ml from prepared dilutions were used to inoculate ISO-GRID HGMF filters. Inoculated HGMF filters were transferred onto XM-G agar followed by incubation at 35°C for 24 h. Ten ml of water sample was diluted in 90.0 ml of sterile 0.1% (w/v) peptone (10 1 ) . Aliquots of 10.0 ml from the undiluted (10°) sample and 10"1 dilution and were used to inoculate ISO-GRID HGMF filters. Inoculated HGMF filters were transferred onto XM-G agar followed by incubation at 35°C for 24 h. 3.5.4. Escherichia co//enrichment To increase the detection limit for E coli, an enrichment broth of LSB supplemented with MUG (Fluorocult®, Merck KGaA, Darmstadt, Germany) was included as part of the microbiological analysis in summer 1999. One hundred ml of berry homogenate was aseptically transferred into a corresponding volume of sterile LSB M U G , in a one-to-one ratio. Flasks containing inoculated L S B M u G were incubated at 35°C for 24 h. Post-incubated flasks were observed for media turbidity and blue fluorescence. Blue fluorescence was observed under long wave ultraviolet light. Media turbidity and blue fluorescence were indicators of E coli presence and constituted a positive reaction. 28 3.5.5. Coliform and Escherichia colir isolation As a confirmation of £. coli presence, attempts were made to isolate pure cultures from all positive plates and enrichment flasks. Numerous other colonies exhibiting typical coliform characteristics were isolated from coliform selective media to determine species identity. Coliform identification was not attempted during summer 1999 sampling, and isolation was reserved for presumptive E coli colonies. Streaks were prepared using either V R B A M U G or CCA petri-plates from all positive flasks and enumeration media, respectively. V R B A M U G plates were solely used for isolation streaks in 1998 because CCA had yet to be introduced into the protocol. Repeated streakings did not necessarily result in the isolation of pure cultures. In summer 1999, further failures resulted in the use of a LSB M UG enrichment step followed by filtration through an ISO-GRID HGMF, which was subsequently transferred onto XM-G agar. Any resulting turquoise coloured colonies were subsequently streaked onto CCA plates. All incubations were conducted at 35°C for 24 h. Isolates were streaked onto CCA and V R B A M U G plates, during 1998 and 1999 respectively, and were stored at 4°C until needed for further characterization. 3.5.6. Identification of coliform and Escherichia coli isolates Isolate identification was conducted using The Biolog GN MicroPlate™ system (Biolog, Inc., Hayward, CA). A total of 221 isolates were processed, 164 of them originating from 1998 samples and the remaining 57 from 1999 samples. The 1998 isolates were identified with the GN MicroPlate™ system, while identification of 1999 isolates were conducted with the GN2 MicroPlate™ system. Unsuccessful identifications resulted in re-testing, but this practice was limited to two tests per isolate. Attempts were made to follow all necessary precautions and procedures as outlined in the GN/GN2 MicroPlate™ instruction manual. As recommend by the instruction manual, Gram stains were performed using either a rapid 3.0% KOH (a Gram stain alternative: Bamarouf etal., 1996) or a 4-step Gram stain kit (BBL® Microbiological Systems, Cockeysville, MD). All isolates originating from 1998 samples were cultivated on BUGM without the addition of 5% defibrinated sheep blood because they were considered to be environmental isolates. This practice was discontinued with the implementation of the GN2 MicroPlate™ system. Inoculated GN/GN2 MicroPlates™ were incubated at 35°C and the resulting purple well patterns were recorded at intervals of 4-6 h and 16-29 24 h. A microplate reader (Thermo Labsystems iEMS Reader MF, Franklin, MA) set at 595 nm was used for all optical density (OD) recordings of the wells. The experimenter visually confirmed the well patterns and saved the data in ASCII format with the help of Thermo Labsystems Genesis® program. Biolog's MicroLog computer program was later used to access the saved files to determine species identification. Unfortunately, implementation of the GN2 MicroPlate system necessitated the manual entry of the well patterns into the revised MicroLog 2 System version 4.0. Identified E co//isolates were stored on TSA slants at 4°C until needed for serological testing. Isolates were once again transferred onto TSA slants and individually wrapped in Parafilm® in preparation for transport to Health Canada (Health of Animals Laboratory, Guelph, ON). Prepared isolates were stored under cool temperatures with the addition of frozen icepacks to the styrofoam container, used for overnight transport. Isolates were transported according to IATA Dangerous Goods Regulations 1.3.3.1 set for infectious substances (Risk Group II) affecting humans. At Health Canada, purity of E coli isolates was confirmed with MacConkey agar and veal infusion yeast extract (VIYE) agar streaks. Serological analysis was initiated on pure cultures and comprised of somatic (0) and flagellar (H) antigen identification (Identification of Escherichia co//Serovars Version 3.0: Health of Animals Laboratory, Guelph, ON). Verocytotoxigenic activity was determined using PCR analysis (Read, 1996). 3.6. Data analysis Plate counts for coliform and E a?//were converted to logio values and were reported according to APHA guidelines (Swanson etal., 1992). Results were analyzed with Minitab® for Windows software, release 12.1 (Minitab Inc., State College, PA). The McNemar's test and Spearman correlation coefficients (Conover, 1999) were used to determine interdependence and correlation between CCA and PEC, respectively. However, only counts within the accuracy range (25-250 CFU/plate) or estimates (0 < and < 25 or 250 < CFU/plate) were used for correlation analysis because of the test requirement for interval 30 data. The Wilcoxon rank signed test and Kruskal-Wallis test (Conover, 1999) were used to evaluate grower and processor practices. For purposes of coliform and E coli population comparisons amongst raspberry and blueberry growers, indeterminate results were assumed to be at the appropriate limit of sensitivity (i.e., < 1 log CFU/g was treated as 1 log CFU/g for statistical analysis when using the Kruskal-Wallis test). The level of statistical significance was set at 5% or P < 0.05 for all tests. 31 IV. RESULTS AND DISCUSSION Microbiological analysis focused primarily on the evaluation of microorganisms on the surface of berry fruit (Tables 5, 6, 8, 9 and 11). Therefore, microbial populations reported in succeeding tables represent both endophytic and exophytic populations of relevant microorganisms. The assumption was made that the vast majority of the microorganisms were on the epiphytic surface of the blueberries and raspberries. Analysis of environmental samples from processor and grower groups was included but was limited to a few F C S , air, soil and water samples (Tables 7, 10 and 12). Information regarding populations of indicator microorganisms was required to ascertain the sanitary and faecal indices of berry fruit. Thus, coliform and E co//enumerations were conducted to quantify the effect that production practices had on berry fruit sanitation. 4.1. 1998 sample analysis: enumeration of aerobes, coliforms and Escherichia coli The geometric means of total aerobe, coliform and E coli populations are reported in Tables 5 to 10. Mould and yeast counts are excluded from succeeding tables because they hold little sanitary significance but are included as part of Appendix I (A. 1.). Aerobes are not typically used as indicators, but served as a reference point for the number of mesophiles present on berry fruit surfaces of which coliforms comprised a significant portion of the total mesophile population. E coli counts on V R B A M U G were reported qualitatively (Table 13) because of the low incidence of E co//within the samples and the difficulty encountered in differentiating between fluorescent and non-fluorescent colonies. As Ray (1989) suggested, plates with V R B A M U G were checked for fluorescence at 12 to 20 h post-incubation, to avoid 4-MU diffusion. However, VRBA M UG plates contained large numbers of non-coliform colonies (non-violet coloured colonies with no bile precipitation), which interfered with plate interpretation. The prevalence of these non-coliforms likely caused the discolouration of surrounding violet-coloured colonies, thereby hampering the ability to isolate and enumerate fluorescent and non-fluorescent colonies. The possible dilution of selective agents in the T S A - V R B A M U G overlays may be responsible for excess coliform and non-coliform growth. Hartman etal. (1975) suggested the use of double strength selective agar to counteract dilution effects when selective agar overlays are used. 32 Initially, the ISO -GRID H G M F method was used to enumerate coliforms and E coli. It was dropped after the first grower sampling of R-a, where eight pre-harvest and post-harvest samples were analysed. Low sample analysis output and the inability to observe fluorescence on or under the H G M F filters resulted in the replacement of H G M F by direct plating alternatives. This inability to observe fluorescence was not reported by Venkateswaran etal. (1996) who tested a variety of food samples with H G M F - V R B A M U G -Therefore, direct plating methods were more suitable to the present analysis because of improved sample output and successful observation of fluorescence. Total aerobes ranged between 3.3 to 5.2 kxjio CFU/g (Tables 5 to 6 and 8 to 9) and were within the expected range of 3 to 9 logio CFU of aerobes/g fresh produce (Nguyen-the and Carlin, 1994). Coliform populations were considerably lower than those reported for aerobes because the latter included dominant microorganisms, such as, coliforms, and Pseudomonas sop. (Lund, 1992; Nguyen-the and Carlin, 1994). The disparity between coliform populations determined by V R B A M U G and PEC methodologies was apparent for all samples analysed, even though statistical analysis was not applied. The discrepancy between methods likely resulted from differences in enumeration protocols. V R B A M U G and PEC have similar selectivity and differentiation abilities. However, the inclusion of a 2 h resuscitation step with TSA (Speck etal., 1975) was likely responsible for higher coliform counts on the TSA-VRBA M U G overlay. Silk etal. (1997) reported coliform detection levels improved (approximately 1 log 1 0 unit) when a TSA-VRBAMUG overlay method was used. In addition, dilution of selective agents is another plausible explanation for higher coliform recovery on V R B A M U G . However, both factors are likely to effect coliform recovery. Interpretation of results on PEC was easier than on VRBA M UG because PEC allowed for greater colony differentiation. The use of a tetrazolium dye, formation of gas-associated colonies and indoxyl accumulation by E coli facilitated differentiation. Coliform populations on processed berries were lower than corresponding populations on pre-processed berries. Processed berries were either washed with or without chlorinated wash water to remove extraneous matter and to reduce the epiphytic microflora. Chlorine concentration in wash water ranged from 0-35 ppm of free C l 2 and was set according to buyers' specifications. As a result of the washing step, a statistically significant (P < 0.05) reduction in the number of coliforms (0.5 log 1 0 CFU/g; arithmetic mean) on processed berries was observed (Wilcoxon signed rank test; Appendix III: A . 4.). 33 This level of reduction was within the expected range reported in the literature, 0.25-2.0 logi 0 CFU/g for washed fresh produce (Lund, 1992; Beuchat, 1996). Processing of berry fruit alters the epiphytic microflora. If GMP's are implemented, processing should not introduce a significant number of indicator microorganisms or pathogens on the berry fruit surface. Yet, processing steps have the potential to introduce new pathogens or induce growth of indicator microorganisms and pathogens (Brackett, 1999). Poor processing plant sanitation protocols would likely be the cause of gross contamination. Plant sanitation by participating processors is difficult to ascertain since the small available sample size limited the ability to evaluate the effectiveness of sanitary practices in the plants. However, indicator microorganism populations in environmental samples were comparable to those obtained from berry samples. Microorganisms in belt swab and water effluent samples were likely derived from direct contact with berry fruit surfaces. Negligible aerobe and coliform counts in air samples indicated acceptable plant air quality. The presence of presumptive E coli'm belt swab sample, P-xe, suggested that this species was present at a low frequency. Upon further biochemi-cal analysis, the isolate was identified as Buttiauxella agresti. This glue positive organism has been isolated from aquatic environments, but not from humans or animals (Brenners, 1984), and thus, has little sanitary significance. Therefore, the effectiveness of sanitation procedures was not investigated further. Relatively high aerobe and coliform populations reported in Table 8 for pre-processed berries indicated that contamination of berry fruit likely occurs in the field. Similar aerobe, coliform and E coll populations were also reported for pre-harvest and post-harvest berry samples (Tables 5 and 6). Therefore, berry fruit surface contamination by indicator microorganisms mainly occurred prior to fruit harvest. Grower practices and pre-production factors are likely responsible for indicator microorganism and pathogen introduction. Contamination through natural events, such as contact with bird/feral animal excrement (Wallace etal. 1997), is also known to be a significant source of indicator microorganisms and pathogens, but is difficult to reduce or eliminate. Harvesting practices were evaluated as a possible source of microorganisms. Comparison of coliform populations, using the Wilcoxon signed rank test, in pre-harvest (Table 5) and post-harvest (Table 6) fruit revealed no statistical difference (P > 0.05) between the two (Appendix III: A. 3.). Therefore, harvesting did not significantly increase coliform 34 levels. Previous studies have indicated that gross contamination of fruit with indicator microorganisms and pathogens can occur if machine harvesters (Splittstoesser, 1970) or human workers (Pichner and Gareis, 1999) are heavily contaminated. For this reason, implementation of sanitation protocols should remain a proactive measure to limit the possibility of transmission by such means. Analyses carried out in 1998 indicated that microbiological hazards were more likely to be acquired during production than during processing. Therefore, a more comprehensive sampling regime at the grower level was undertaken in 1999 to identify the sources of coliforms and E. coffin berry production systems. 35 Table 5. Populations of aerobes, coliforms and E colion pre-harvest raspberries and blueberries from growers sampled in 1998. Grower 3 Aerobes Coliforms E. coli n c l og 1 0 CFU/gb VRBAMUG P EC PEC n c l og 1 0 CFU/gD n c l og 1 0 CFU/g D n c l og 1 0 CFU/gD R - a 8 4.7 4 2.1 d N A E N A 8 N A E N A 8 B - a 8 3.3 8 2.4 1 1.7 6 <1* B-3 4 4.4 4 3.3 2 <1' 2 <1' 3 R, raspberry grower; B, blueberry grower. b Geometric means, where applicable; log 1 0 C FU per gram of berry fruit. 0 Number of samples with estimated plate counts or counts within the acceptable range (25-250 CFU) . It can refer to the number of samples with no observed colonies, when the succeeding column reports an organism population density of < 1 log 1 0 C F U per gram of berry fruit. d l og 1 0 MPNGU/g, sample analysis comprised of the HGMF method. 8 NA, not available or not applicable. f<1, below the detection limit of < 1 log 1 0 C FU per gram of berry. 36 Table 6. Populations of aerobes, coliforms and £. coli on post-harvest raspberries and blueberries from growers sampled in 1998. Grower 3 Aerobes Coliforms E. coli n c l og 1 0 CFU/gb VRBAMUG PEC PEC n c l og 1 0 CFU/gD n c l og 1 0 CFU/g D n c log 1 0 CFU/g° R-a 18 4.8 13 2.9D/4.0 1 2.2 4 <1E B-a 7 3.3 7 2.4 6 <1E 6 <1E B-P 4 4.4 3 3.6 2 <1E 2 <1E 3 R, raspberry grower; B, blueberry grower. b Geometric means, where applicable; log 1 0 C FU per gram of berry fruit. c Number of samples with estimated plate counts or counts within the acceptable range (25-250 CFU). It can refer to the number of samples with no observed colonies, when the succeeding column reports an organism population density of < 1 log 1 0 C FU per gram of berry fruit. d log 1 0 MPNGU/g. Three out of the thirteen samples were analysed using the HGMF method. 6 <1, below detection limit of < 1 l o g 1 0 CFU per g of sample. 37 Table 7. Populations of aerobes, coliforms and E co/f on berry bush leaves and harvesting equipment 1998. Aerobes Coliforms E. coli V R B A M U G PEC PEC Grower 3 n c l og 1 0 CFU/gb n c l og 1 0 CFU/gD n c l og 1 0 CFU/gD n c log 1 0 CFU/g° R-a e 2 2.8 2 <1' NA h NA h NA h NA h B-a e 2 1.5 2 <1' NA h NA h NA h NA h B-Pe 1 2.2 1 2.1 1 <1j 1 <r B-a f 1 6.4 1 4.4 NA h NA h NA h NA h B-p 9 2 4.4 2 3.7 NA h NA h NA h NA h 3 R, raspberry grower; B, blueberry grower. b Geometric means, where applicable; log 1 0 C FU per gram berries or c m 2 leaf. 0 Number of samples with estimated plate counts or counts within the acceptable range (25-250 CFU). It can refer to the number of samples with no observed colonies, when the succeeding column reports an organism population density of < 1 log 1 0 C FU per gram of berry fruit. d log 1 0 MPNGU/g, sample analysis comprised of the HGMF method. \ 6 Berry bush leaves. f Pail swab. 9 Machine harvester belt loader swab. h NA, not available or not applicable. ' <1, Below the detection limit of < 1 log 1 0 C FU per g of sample. Table 8. Populations of aerobes, coliforms and E co//on pre-processed raspberries and blueberries from processors sampled in 1998. Processor 3 Aerobes Coliforms E. coli n c l og 1 0 CFU/gb VRBAMUG PEC PEC n c l og 1 0 CFU/gD n c l og 1 0 CFU/g D n c l og 1 0 CFU/gD P-co 6 5.1 6 4.9 2 <1 d 2 <1 d P-JC 18 4.4 18 4.0 2 2.7 2 <1d P-y 5 5.2 5 4.5 2 2.7 3 <1 d a P-co, raspberry processor; P-%, P-y, blueberry processors. b Geometric means, where applicable; log 1 0 C FU per gram of berry fruit. 0 Number of samples with estimated plate counts or counts within the acceptable range (25-250 CFU). It can refer to the number of samples with no observed colonies, when the succeeding column reports an organism population density < 1 log 1 0 C F U per gram of berry fruit. d <1, Below the detection limit of < 1 l og 1 0 CFU per g of sample. 39 Table 9. Populations of aerobes, coliforms and £. coli on post-process raspberries and blueberries from processors sampled in 1998. Aerobes Coliforms E. coli VRBAMUG P EC PEC Processor 3 n c l og 1 0 CFU/gb n c l og 1 0 CFU/gD n c l og 1 0 CFU/g D n c l og 1 0 CFU/gD P-co 4 4.5 4 4.5 NA d N A D N A D N A D P-X 8 4.4 8 3.7 3 2.5 2 1.2 P-V 5 5.0 5 4.3 2 2.3 1 2.4 3 P-co, raspberry processor; P-%, P-y, blueberry processors. b Geometric means, where applicable; log 1 0 C FU per gram of berry fruit. 0 Number of samples with estimated plate counts or counts within the acceptable range (25-250 CFU). It can refer to the number of samples with no observed colonies, when the succeeding column reports an organism population density of < 1 log 1 0 C FU per gram of berry fruit. d NA, Not available or not applicable. Table 10. Populations of aerobes, coliforms and E a?//in processing plant air, effluent and sorting belt samples in 1998. Aerobes Coliforms E. coli VRBAMUG PEC PEC Processor 3 n c l og 1 0 CFU/gb n c l og 1 0 CFU7g° n c l og 1 0 CFU/gD n c l og 1 0 CFU/gD P-CO* 2 4.8 2 4.5 N A H N A H N A H N A H P-x d 2 3.7 2 3.8 1 1.3 2 <v P-y" 1 5.8 N A H N A H 1 <1' 1 <1' P-co8 1 1.8 1 <1' N A H N A H N A H N A H P-%9 2 5.8 2 4.8 1 <r 1 3.8 P V 3 6.0 3 5.0 1 <v 1 <r 9 1.5" 8 0 9 N A H N A H N A H N A H a P-co, raspberry processor; P-%, P-y, blueberry processors. b Geometric means, where applicable; log 1 0 C FU per gram of berry fruit. c Number of samples with estimated plate counts or counts within the acceptable range (25-250 CFU). It can refer to the number of samples with no observed colonies, when the succeeding column reports an organism population density of < 1 log 1 0 C FU per gram of berry fruit. d Water effluent. 8 Sorting belt swabs. f Air samples. 9 Organism frequency is reported in CFU/minute of air exposure. h NA, Not available or not applicable. ' <1, Below the detection limit of < 1 log 1 0 C FU per g of sample. 41 4.2. 1999 sample analysis: enumeration of coliforms and Escherichia coli Based upon trends observed in 1998, microbiological analyses during 1999 were focused on the enumeration of coliform bacteria and E coli on berry fruit, water and soil in the field (Tables 11 and 12). The number of grower participants was increased from the previous year (from three to seven), as was the number of samples collected from each grower. Coliform populations determined on CCA ranged from 2.8 to 3.9 and 3.3 to 5.7 log 1 0 CFU/g of raspberry and blueberry fruit, respectively. PEC was also used in conjunction with CCA for coliform and E co//enumeration, but both indicator populations were found to be lower when the former medium was used. Coliforms enumerated with PEC ranged from 1.6 to 2.2 and 1.7 to 2.6 logi 0 CFU/g for raspberry and blueberry samples, respectively. E coll populations on CCA ranged from < 1 to 2.5 and 1.1 to 2.6 logio CFU/g, while populations were < 1 logio CFU/g and < 1 to 1.8 logio CFU/g on PEC, for raspberry and blueberry samples, respectively. Coliform populations were in agreement with the range expected on fresh produce as reported in literature sources. Garcia-Villanova etal. (1987) reported 82.6% (285 samples) of fresh vegetable produce had coliform populations greater than 10 CFU/g. If these results are compared to other foods, such as fresh meat, coliform populations on berry fruit are similar to high coliform counts on fresh meat (2.7 log 1 0 CFU/g) (Attala etal., 2000). Unlike in fresh meat, most coliforms on blueberries and raspberries are not likely from faecal origin (Appendix II: A. 2.) Interpretation of results with CCA and PEC was less problematic than with VRBA M U G . Simplification of coliform and E co//"enumeration was made possible by colony accumulation of chromogenic end products, Salmon and indoxyl (Manafi, 1992), and reduced overgrowth of coliforms and non-coliforms. Both CCA and PEC plates were easy to interpret, although an advantage of CCA was the ability to enumerate E coli colonies after 24 h of incubation, unlike the 48 h required by PEC. When coliform and E coli populations for each medium were compared, the geometric means from both grower groups were consistently higher on CCA. Based upon McNemar's test, coliform and E coli recoveries on CCA and PEC were not equivalent (Appendix III: A. 5.). Further analysis using the Spearman's rank correlation on counts within the accuracy range, 25 to 250 CFU/plate, revealed coliform populations were not positively correlated (p = 0.20; P > 0.05; Appendix III: A. 6.), as shown in Figure 2., whereas E coli populations were positively correlated (p = 0.68; P < 0.05; Appendix III: A. 7.), as 42 shown in Figure 3. A positive correlation for £ co//was a result of similar differentiation abilities. Both PEC and CCA rely on glue expression for £ coli colony differentiation. However, the lack of correlation for coliforms between the two methods may have been attributable to differences in agar formulation. Selectivity of both agars relies on surface active agents, but CCA contains the added compounds pyruvate, phosphate and sorbitol, which help in resuscitation of injured cells (manufacturer's instructions; Ray, 1989). Turner etal. (2000) reported strong correlations for coliforms (0.89) and £. GO//(0.86) when CCA was compared to PEC. Differences in the type of samples analysed, meats versus fresh fruit, may account for the disagreement between results reported here and by Turner etal. In addition, Turner et al. (2000) reported better recovery of £. coli on CCA than on PEC, which agrees with results reported here. Further assessment of microbial hazards in production systems was evaluated with CCA because of the greater recovery of coliforms and £ coli. Differences in grower practices (Table 3) were likely responsible for some of the apparent variation in coliform and £ coli populations between berry growers (Table 11). Except for soil analysis, random effects from natural events and pre-production factors were not evaluated, even though the latter are known to be a significant contributor of microorganisms. Based upon the Kruskal-Wallis test, median coliform (Appendix III: A. 8. and A. 9.) and £ coli (Appendix III: A. 10. and A. 11.) populations differed significantly (P < 0.05) from one another. Subsequent analysis with Tukey's multiple comparison test on ranked coliform (Appendix III: A. 12. and A. 13.) and £ coli (Appendix III: A. 14 and A. 15.) populations means, revealed grower comparisons Ra-Rfi, Ba-By, Ba-Bp, B5-BP and B8-By and Ba-Bp, B5-BP and By-Bp differed significantly from one another for coliform and £ co//data, respectively. Once grower comparisons were cross indexed with defined grower characteristics, as listed in Table 3, specific practices were associated with microbial hazards. Blueberry samples from grower Bp, where machine harvesting and ditch water-overhead sprinkler irrigation were practiced, had not only the greatest coliform and £ coli populations, but £ coli populations were found to differ significantly from growers Ba, By and BS. Therefore, ditch water applied by overhead sprinkler irrigation appeared to account for the differences in £ coli populations. However, the effect of natural and pre-production factors, such as, historical land usage and current usage of proximal lands, should not be discounted. 43 A similar trend with coliform populations was not clearly observed because E coli\s a better indicator of the sanitary and faecal indices of fresh produce. Unsuitability of coliforms, as a sanitary indicator, is further supported by biochemical analysis of the 163 coliform isolates from 1998 samples (Appendix II: A. 2.). Rahnella, Klebsiella and Enterobacterspecies were the predominant microorganisms, constituting over 50% of the coliforms classified. However, they are of little public health significance in the context of foodborne disease, according to Brenners (1984). Microbiological analysis of soil was necessary to evaluate the effects from previous land usage and grower fertilization practices. Coliform populations in soil ranged from 3.2 to 4.6 logio CFU/g, whereas £ co//levels were lower, ranging between < 0.69 to 4.1 log 1 0 CFU/g (Table 12). £ co//was used as the lone indicator of sanitation because several Enterobacter and Klebsiella spp. are known to be common inhabitants of soil (De Roever, 1999). Therefore, high coliforms counts were not indicative of high levels of faecal contamination of soil. Confirmation of presumptive £ coli colonies was unsuccessful for all 11 samples tested. The failure to isolate £ co/imay have resulted because of competing glue positive organisms or loss of cell viability because of prolonged refrigerated storage of the inoculated HGMF-XM-G agar plates. Therefore, the sanitary and faecal indices of soil and its contribution of microbial hazards remained inconclusive. Water analysis (Table 12) revealed that aquifer water sources did not harbour substantial levels of coliforms and £ coli. However, irrigation water from ditches had coliform populations between 4.0 to 4.3 logio MPNGU/100 ml, whereas £ co//was below the detectable limit of 0.69 logio MPNGU/100 ml. Norman and Kabler (1953) and Nguyen-the and Carlin (1994) reported coliform counts on produce decrease substantially when irrigation water with low levels of coliforms is used. Therefore, in the present study, ditch water was a potential source of microbial hazards, particularly where overhead sprinkler irrigation was practiced. 44 Table 11. Populations of coliforms and E coli on post-harvest raspberries and blueberries sampled from growers during the 1999 production season. Grower 3 CCA PEC Coliforms E. coli Coliforms E. coli n c l og 1 0 CFU/gD n c l og 1 0 CFU/gD n c l og 1 0 CFU/gD n c l og 1 0 CFU/gD R-a 18 2.8 18 <1 d 4 1.6 6 <1 d R-P 19 3.9 4 2.5 6 1.7 19 <1 d R-Y 19 3.0 6 1.7 8 2.2 1 <1 d B-a 17 3.3 1 2.6 10 2.3 1 1.6 B-P 24 5.7 19 2.6 23 2.6 17 1.8 B - Y 18 4.9 2 1.1 7 1.7 18 <1 d B -5 16 4.0 1 1.1 6 1.9 16 <1 d 3 R, raspberry grower; B, blueberry grower. b Geometric means, where applicable; log 1 0 C F U per gram of berry fruit. 0 Number of samples with estimated plate counts or counts within the acceptable range (25-250 CFU). It can refer to the number of samples with no observed colonies, when the succeeding column reports an organism population density of < 1 log 1 0 C F U per gram of berry fruit. d <1, Below the detection limit of < 1 log 1 0 C FU per g of sample. 45 Table 12. Populations of coliforms and E co/I'm irrigation water and soil obtained from raspberry and blueberry growers during 1999 sampling. Grower 3 N Coliforms E. coli log 1 0 MPNGU/(100 ml or g) b R-a d 1 0.3 C <0.69f B-(3D 11 4.3 <0.69 c f B- Y d 3 <0.69f <0.69 c f B-8 d 3 4.0 <0.69 c f R-a e 1 4.5 2.5 R - P e 3 4.3 2.5 R-Y9 2 3.2 1.3 B-a e 2 3.8 <0.69 c f B Y 2 4.6 4.1 B-5 e 1 3.4 <0.69f 3 R, raspberry grower; B, blueberry grower b Geometric means, where applicable; per 100 ml for water samples or per g for soil samples. 0 Represents samples with HGMF scores below the detection limit and/or under the recommended limit of 20 counts (< 1.3 log 1 0 MPNGU/g). d Water samples. e Soil samples. f <0.69, Below the detection limit of < 0.69 log 1 0 MPNGU per g of sample. 46 Figure 2. Relationship between PEC and CCA coliform plate counts within the accuracy (25 to 250 CFU) and estimated range (0 < and < 25 or 250 < CFU/plate). • Coliforms * Expected 0.0 1.0 2.0 3.0 4.0 5.0 CCA ( log 1 0 CFU coliforms/g berries) 6.0 7.0 47 Figure 3. Relationship between PEC and CCA E CD//plate counts within the accuracy (25 to 250 CFU) and estimated range (0 < and < 25 or 250 < CFU/plate). • • • • E coli • Expected 0.0 1.0 2.0 3.0 CCA ( log 1 0 CFU E coli/g berries) 4.0 5.0 48 4.3. Qualification of Escherichia coli c Recovery of presumptive £. co//from fluorescent plates and flasks was not always successful (Table 13). Fluorescent isolates on V R B A M U G from processor P-to were the only ones conclusively identified as £. co//"\n 1998. Three £ a?//and one Hafnia a/veiwere confirmed to be the organisms responsible for fluorescence in processor P-to samples. To increase the detection limit for £ coli, L S B M U G was introduced in 1999 to increase the detection limit from the previous 10 CFU/g of berries to 1 CFU/10 g of berries, a 100-fold improvement. In response, £ coli recovery from fluorescent media increased from the previous 12.5% (VRBAMUG) to 37.3% ( L S B M U G ) . For most samples, confirmed £ a?//from fluorescent flasks corresponded to £ coli populations within detectable limits, while unconfirmed £ coli from fluorescent flasks corresponded to £ coli levels below detectable limits for CCA and PEC. Isolation attempts may have failed because of loss of £ co//viability or other glue expressing microorganisms present in the enrichment broth were responsible for GUD production (false positive reactions). Moberg (1985) reported Staphylococcus species are capable of producing fluorescence, even though they show poor growth in LSBMUG- In addition, attempts may have failed because bacteriocin-like compounds produced by competing bacteria are known to be antagonistic towards coliforms (Weaver and Boiter, 1951). Grower BB had the highest incidence of fluorescence and isolations of £ coli (Table 13). These results confirm previous observations implicating ditch water-overhead sprinkler irrigation as a source of microbial contamination in berry production. In addition, LSBMUG utilization increased the recovery of £ co/Zfrom berry fruit and confirmed quantitative results from enumeration methods. 49 Table 13. Absence or presence of fluorescence and confirmation of E a?//within grower and processor samples. VRBAMUG LSBMUG Participant 3 n + - E. coll" + - E. coll" R -a 32 2 24 0 1 5 0 R-P 19 N A N A N A 7 12 1 R-Y 19 N A N A N A 7 12 2 B -a 41 1 22 0 1 17 1 B-p 34 3 7 0 18 6 6 B-Y 18 N A N A N A 5 13 3 B-8 18 N A N A N A 4 14 1 P-co 13 4 9 3 N A N A N A P-JC 39 4 35 0 N A N A N A P-y 13 4 9 0 N A N A N A 3 R, raspberry grower; B, blueberry grower; P-co, raspberry processor; P-%, P-\\i, blueberry processors. b E coli confirmed through biochemical and serological means. LSBMUG/ applicable to 1999 samples only. n, number of samples analysed. NA, not available or not applicable. V R B A M U G , applicable to 1998 samples, only. += Fluorescent samples - = Non-fluorescent samples 50 4.4. Identification of Escherichia coli isolates E col/is used not only as an indicator of faecal contamination, but also for the possible presence of variants that are pathogenic in nature. Of the 64 E coli isolates sent for serological analysis, 33 distinct serotypes were identified (Table 14). All serotypes were negative for verocytotoxins. However, several of these serotypes have been previously isolated from cattle and beef products and confirmed as verocytotoxigenic (Health Canada, Health of Animals Laboratory, 1985-93: Butler and Clarke, 1994, 1994). The absence of virulence factors through loss of either plasmid or phage encoded entero-haemolysins and toxins can render isolates non-verocytotoxigenic (Lior, 1994; Ziebell, 2001). Therefore, the E coli isolates recovered in this study are not commonly associated with disease, but served to indicate that low levels of faecal contaminants and potential enteric pathogens were present in the blueberry and raspberry production systems investigated. The two £. coli serotypes belonging to processor P-x were originally classified as Salmonella subspecies 1G/1F using the Biolog GN Microplate™ system (Appendix II: A. 2.). Once serology was performed, these isolates were identified as £. coli. Therefore, inconsistencies in isolate identification through biochemical-means suggest the necessity to confirm presumptive Salmonella identities using serology and support recommendations outlined in Biolog's MicroLog database. The greatest number of £. coli isolates was from grower B-(3, where ditch water applied by overhead sprinklers was used for irrigation purposes. In contrast, grower B-y utilized overhead irrigation with water from an aquifer, while B-8 utilized ditch water applied by drip/trickle irrigation for water delivery, but both had substantial lower recovery of £. coli. These results indicate that overhead sprinkler irrigation and ditch water by themselves were not substantial contributors to £. coli contamination of blueberries. However, the combination of ditch water and overhead sprinkler irrigation was identified as one of the sources responsible for £. coli introduction onto berry fruit surfaces. 51 Table 14. Serotype identity of £. coli isolates from grower and processor samples. Participant 3 Serotypes R-p R-y B-a B-P B-y B-8 P-co P-% O?:H10 - - + - - - -0?:H14 . . - + - -0?:H16 . - - + - -0?:H19 . . . + - - + -0?:H2 + . . . 0?:H34 + . . . 0?:H7 . + - - - + - -011:H25 - - - + - - -011:H5 . . . + - -O113:H10 . - - + - -0117:H5 . - - + - -0117:NM . . - - - - + -O120:H10 . . . + O120:H5 . . . + - - - + 0137:H6 . . - + - -0138:H48 + - - . . . O140:H14 . - + - - - - + 014V.H? . . - - - - + -0147:1-132 . - + - - -0147:NM . - - + - -0149:H12 - - - + - - -0149:H5 . . - + - -0156:H7 . . . + O160:H11 . . - + - -018:NM . . . + - -027:1-121 . . . + - - - -051:H42 + . . . 053:NM + - - - + -07:H6 . . . . + . 08:NM - + - . . . 081:H21 . . . + 086:H51 . + - . . . 09:H5 - - - _jt • - : — Totals 2 6 3 16 4 1 4 2 3 R-a and P-y are absent because E. coli was not found in the samples analysed. + = Present - = Absent V. CONCLUSIONS AND RECOMMENDATIONS The results of this study suggest contamination of blueberry and raspberry production systems occurred mostly in the field. Aside from natural events and pre-production factors, grower practices, such as irrigation water source and method of distributing irrigation water helped define the microflora on epiphytic surfaces. Minimal processing and harvesting did not contribute substantially to the number of coliforms on berry fruit, although this potential cannot be overlooked. Ditch water-overhead sprinkler irrigation was found to introduce significant levels of £. co//onto blueberries, and thus was a significant source of microbial contamination. However, none of the identified £ coli isolates were found to be verocytotoxigenic. Of the three selective agars used for coliform and £. coli enumeration, TSA-VRBA M U G overlay was found to be problematic and was subsequently replaced by CCA and PEC. Coliform and £ coli enumeration efficiencies of CCA and PEC were found to be statistically dissimilar. Coliform and £. coli recoveries were greater on CCA than on PEC. Therefore, utilization of CCA is recommended as an index of sanitation and safety for fresh produce. Although ditch water-overhead sprinkler irrigation was identified as a source of microbial contamination in only blueberry production systems, the impact of this practice on the safety of all fruits should not be overlooked. Other irrigation sources or distribution methods that limit fruit-water interactions would be able to reduce this sort of contamination. The implementation of GAP (Pabrua, 1999), a comprehensive and proactive measure, is recommended to reduce the inherent microbial contamination in blueberry and raspberry production systems. £ coli is a better indicator of potential microbial contamination for fresh berries than coliforms because the latter are always present as part of the epiphytic microflora. Therefore, industry utilization of coliforms as the preferred sanitary indicator for fresh fruit should be re-evaluated. Utilization of more selective indicator microorganisms, such as faecal coliforms, is encouraged because they better indicate epiphytic surface sanitation. However, high faecal coliform counts are not always indicative of either faecal contamination or a health hazard. 53 Finally, a more thorough sampling plan, including sample enrichment and greater assay selectivity (i.e., higher incubation temperatures) is encouraged for water sample analysis. These recommendations would increase the likelihood of isolating E coli, thereby allowing for the comparison of E coli serotypes isolated from both berry fruit and water samples. 54 REFERENCES Attala, H.N., Johnson, R., McEwen, S., Usborne, R.W. and Gyles, C L 2000. Use of a Shiga toxin (Stx)-enzyme-linked immunosorbent assay and immunoblot for detection and isolation of Stx-producing Escherichia a?//from naturally contaminated beef. Journal of Food Protection. 63:1167-1172. Bamarouf, A., Eley, A. and Winstanley, T. 1996. Evaluation of methods for distinguishing gram-positive from gram-negative anaerobic bacteria. Anaerobe. 2:163-169. Bell, C. and Kyriakides, A. 1998. Introduction, p. 1-14. In E coir. A Practical Approach to the Organism and its Control in Foods. Blackie Academic and Professional, London, UK. Brenners, D J . 1984. Facultatively anaerobic Gram-negative-rods: Family I. Enterobacteriaceae, p. 409-505. In Bergey, D.H. Bergey's Manual of Systematic Bacteriology. Krieg, N.R. and Holt, J.G. (Ed). Williams and Wilkens, Baltimore, MD. Beuchat, L.R. 1996. Pathogenic microorganisms associated with fresh produce. Journal of Food Protection. 59:204-216. Beuchat, L.R. and Ryu, J.H. 1997. Produce handling and processing practices. Emerging Infectious Disease. 3:459-464. Blank, G., Guan, T.Y. and Ismond, A. 2000. Pathogen survival in liquid pesticide formulations. CIFST Symposium: Food Safety, July 16-19, 2000. Winnipeg, MB. Bomser, J . , Madhavi, D.L., Singletary, K. and Smith, M.A.L. 1996. In vitro anticancer activity of fruit extracts from Vaccinium species. Planta Medica. 62:212-216. Brackett, R.E. 1999. Incidence, contributing factors, and control of bacterial pathogens in produce. Postharvest Biology and Technology. 15:305-311. Butler, P.G. and Clarke, R.C. 1994. Diarrhoea and dysentery in calves, p. 91-116. In Escherichia co/iin Domestic Animals and Humans. Gyles, C L . (Ed). CAB International, Wallingford, UK. Canada. Statistics Canada. 1999. Food Consumption in Canada - Part II. Catalogue no. 32-230. Canada. Statistics Canada. 2000. Fruit and Vegetable Production. Agriculture Division: Horticultural Crops Unit. February. Catalogue no. 22-003-X1B. Carter, M.R. 1993. Soil sampling and methods of analysis, p. 16. Lewis Publishers, Boca Raton, FL. Conover, W J . 1999. Some variations of the sign test, p. 166-170. Several independent samples, p. 288-297. The one-sample or matched-pairs case, p. 350-364. In Conover, W.J. Practical Nonparametric Statistics. 3 r d ed. John Wiley and Sons, Toronto, ON. De Roever, C. 1999. Microbiological safety evaluations and recommendations on fresh produce. Food Control. 10:117-143. Feng, P.C.S. and Hartman, P.A. 1982. Fluorogenic assays for immediate confirmation of Escherichia coll. Applied and Environmental Microbiology. 43:1320-1329. 55 Frampton, E.W., Restaino, L. and Blaszko, N. 1988. Evaluation of the beta-glucuronidase substrate 5-bromo-4-chloro-3-indolyl-beta-D-glucuronide (X-GLUC) in a 24 hour direct plating method for Escherichia coli. Journal of Food Protection. 51:402-404. Garcia-Villanova, B., Galvez Vargas, R. and Garcia-Villanova, R. 1987. Contamination on fresh vegetables during cultivation and marketing. International Journal of Food Microbiology. 4:285-291. Geldreich, E.E. and Bordner, R.H. 1970. Fecal contamination of fruits and vegetables during cultivation and processing for market. A review. Journal of Milk and Food Technology. 34:184-195. Geraldine, A., Willshaw, T.C. and Smith, H.R. 2000. Escherichia coli, p. 1136-1177. In The Microbiological Safety and Quality of Food. Lund, B., Baird-Parker, T.C. and Grahame, G.W. (Ed.) Aspen Publishers Inc., Gaithersburg, MA. Hartman, P.A., Harman, P.S. and Lanz, W.W. 1975. Violet red bile 2 agar for stressed coliforms. Applied Microbiology. 29:537-539. Hitchins, A.D., Hartman, P.A. and Todd, E.C.D. 1992. Coliforms-Escherichia co/i and its toxins, p. 325-369. In American Public Health Association (APHA). Compendium of Methods for the Microbiological Examination of Foods. Vanderzant, C. and Splittstoesser, D.F. (Ed). American Public Health Association Inc., Washington, D.C. Jay, J.M. 2000. Indicators of food microbial quality and safety, p. 387-406. In Jay, J.M. Modern Food Microbiology. 6 t h ed. Aspen Publishers Inc., Gaithersburg, MD. Kalt, W., Forney, C.F., Martin, A. and Prior, R.L. 1999. Antioxidant capacity, vitamin C, phenolics, and anthocyanins after fresh storage of small fruits. Journal of Agricultural and Food Chemistry. 47:4638-4644. Lior, H. 1994. Classification of Escherichia coli, p. 31-72. In Escherichia coli'm Domestic Animals and Humans. Gyles, C L . (Ed). CAB International, Wallingford, UK. Lund, B.M. 1992. Ecosystems in vegetable foods. Journal of Applied BacteriologySymposium Supplement. 73:115S-126S. Manafi, M. 1992. Fluorogenic and chromogenic enzyme substrates in culture media and identification tests. International Journal of Food Microbiology. 31:45-58. Moberg, L.J. 1985. Fluorogenic assay for rapid detection of Escherichia coli'm food. Applied and Environmental Microbiology. 50:1383-1387. Nagle, N.E. 1999. Understanding produce safety. Dairy, Food and Environmental Sanitation. 19:598, 604. Nguyen-the, C. and Carlin, F. 1994. The microbiology of minimally processed fresh fruits and vegetables. Critical Reviews in Food Science and Nutrition. 34:371-401. Norman, N.N. and Kabler, P.W. 1953. Bacteriological study of irrigated vegetable. Sewage and Industrial Wastes: The Journal of the Federation of Sewage Works and Association. 25:605-609. Pabrua, F.F. 1999. Good agricultural practices: methods to minimize microbial risk. Dairy, Food and Environmental Sanitation. 19:532,526. Pell, A.N. 1997. Manure and microbes: public and animal health problem? Journal of Dairy Science. 80:2673-2681. 56 Pichner, R. and Gareis, M. 1999. Clinically symptomless VTEC/EHEC excreting personnel in meat processing factories. Mitteilungsblatt der Bundesanstalt fuer Fleischforschung, Kulmbach. 38:270-278. Ray, B. 1989. Introduction and Enumeration of Injured Indicator Bacteria From Foods, p. 1-54. In Injured Index and Pathogenic Bacteria: Occurrence and Detection in Foods, Water and Feeds. Ray, B. (Ed). CRC Press, Boca Raton, FL. Read, D. 1996. Identification of presumptive positive verocytotoxigenic Escherichia coli by polymerase chain reaction. Compendium of Analytical Methods: Health Protection Branch, Health Canada. Warburton, D. (Ed). MFLP-86. Polyscience Publications Inc., Laval, PQ. Roberts, D., Watson, G.N. and Gilbert, R J . 1982. Contamination of food plants and plant products with bacteria of public health significance, p. 169-191. In Bacteria and Plants. Rhodes-Roberts, M.E. and Skinner, F.A. (Ed). Academic Press, Toronto, Canada. Silk, T.M., Ryser, E.T. and Donnelly, C W . 1997. Comparison of methods for determining coliforms and Escherichia coli levels in apple cider. Journal of Food Protection. 60:1302-1305. Speck, M.L., Ray, B. and Read Jr., R.B. 1975. Repair and enumeration of injured coliforms by a plating procedure. Applied Microbiology. 29:549-550. Splittstoesser, D.F. 1970. Predominant microorganisms on raw plant foods. Journal of Milk & Food Technology. 33:500-505. Splittstoesser, D.F. 1983. Indicator organisms on frozen vegetables. Food Technology. 37(6): 105-106. Swanson, K.M.J., Busta, F.F., Peterson, E.H. and Johnson, M.G. 1992. Colony counts methods, p. 75-95. In American Public Health Association (APHA). Compendium of Methods for the Microbiological Examination of Foods. Vanderzant, C. and Splittstoesser, D.F. (Ed). American Public Health Association, Washington Inc., D.C. Szabo, R.A. 1997. Determination of Enterobacteriaceae. Compendium of Analytical Methods: Health Protection Branch, Health Canada. Warburton, D. (Ed). MFLP-43. Polyscience Publications Inc., Laval, PQ. Tauxe, R., Kruse, H., Hedberg, C , Potter, M., Madden, J . and Wachsmuth, K. 1997. Microbial hazards and emerging issues associated with produce. A preliminary report to the national advisory committee on microbiologic criteria for foods. Journal of Food Protection. 60:1400-1408. Turner, K.M., Restaino, L. and Frampton, E.W. 2000. Efficacy of Chromocult® coliform agar for coliform and Escherichia coli \n detection in food. Journal of Food Protection. 63:539-541. Venkateswaran, K., Murakoshi, A. and Satake, M. 1996. Comparison of commercially available kits with standard methods for the detection of coliforms and Escherichia coli'm foods. Applied and Environmental Microbiology. 62:2236-2243. Wallace, J.S., Cheasty, T. and Jones, K. 1997. Isolation of Vero cytotoxin-producing Escherichia coli 0157:H7 from wild birds. Journal of Applied Microbiology. 82:399-404. Wachsmuth, I.K. 1997. Escherichia co//0157:H7 - harbinger of change in food safety and tradition in the industrialized world. Food Technology. 51(10):26. Weaver, R.H. and Boiter, T. 1951. Antibiotic-producing species of Bacillus from well water. Transactions of the Kentucky Academy of Science. 13:183-188. 57 Ziebell, K. 2001. Personal communication. Health Canada: Health of Animals Laboratory. Guelph, ON. 58 APPENDIX I A . 1. Populations of mould and yeast on raspberries and blueberries from growers and processors sampled in 1998. Participant3 Mould Yeast n l o g 1 0 C F U / gD n l o g 1 0 C F U / gD R-a 26 4.8 26 . 4.7 B-a 16 3.6 16 4.5 B-p 8 4.3 8 4.9 P-co 10 4.8 9 5.5 P-X 24 4.3 23 4.8 P-y 9 4.5 9 5.4 3 R, raspberry grower; B, blueberry grower; P-co, raspberry processor; P-%, P-y, blueberry processors. b Geometric means; log 1 0 cfu per gram of berry fruit. A P P E N D I X I I A. 2. Biochemical classification of coliform and presumptive E coli isolates from growers and processors sampled in 1998. Microorganism P-co R-a Source of Isolates P-% B-a P-H/ Total Burkholderia cepacia 0 0 1 0 1 2 Buttiauxella agrestis 0 0 1 0 5 6 Cedecea lapagei 0 0 1 0 0 1 Citrobacter 0 0 1 0 0 1 Citrobacter freundii 0 0 0 1 0 1 Enterobacter spp. 1 0 0 0 0 1 Enterobacter aerogenes 0 0 1 0 0 1 Enterobacter amnigenus 0 0 0 1 1 2 Enterobacter cloacae A/B 0 0 2 1 2 5 Enterobacter gergoviae 0 1 2 4 1 8 Enterobacter intermedius 0 0 1 0 0 1 Enterobacter sakazaki 0 0 1 0 0 1 Escherichia coli 8 0 3 a 0 0 11 Gilardi pink Gram negative rod 0 0 0 0 1 1 Hafnia alvei 2 0 0 0 0 2 Klebsiella 0 0 10 1 0 11 Klebsiella planticola 0 0 8 0 0 8 Klebsiella pneumoniae SS pneumoniae 0 0 4 0 0 4 Kluyvera ascorbata 0 0 4 1 0 5 Pantoea agglomerans 0 2 3 3 1 9 Pseudomonas fluorescens type B 0 0 0 3 0 3 Salmonella subspecies 1G/1 F b 0 2 2 5 3 12 Serratia fonticola 0 0 1 0 0 1 Serratia liquefaciens/grimesii 0 0 8 0 0 8 Serratia marcescens 0 0 2 0 0 2 Rahnella aquatilis 3 1 12 1 16 33 Unidentified 3 0 13 3 4 23 Total Tested 17 6 81 24 35 163 a Originally classified as Salmonella subspecies 1G/1F, but were later identified as E a?//once serology was performed. b Isolates were biochemically classified as Salmonella subspecies 1G/1F, but upon further analysis, three of the isolates were identified as E coli. Therefore, Salmonella classification is doubtful for the remaining twelve unconfirmed isolates. 60 APPENDIX III A. 3. Coliform contribution of harvest practices of blueberry and raspberry growers (Wilcoxon signed rank test). Pre-harvested Harvested n (logio coliform counts) Difference (£>/) Rank of |£V| 1 2.4 3.1 0.7 5 5 2 1.4 3.4 2.0 11 11 3 1.8 1.3 -0.5 3.5 0 4 1.8 1.8 0.0 1 1 5 1.5 2.1 0.6 3.5 3.5 6 2.8 2.6 -0.2 2 0 7 1.6 3.5 1.9 10 10 8 2.4 3.2 0.8 6 6 9 3.8 2.3 -1.5 8 0 10 5.4 4.6 -0.8 7 0 11 2.8 4.3 1.5 9 9 ZR{+) 45.5 a Rank of positive £>/values. H0: There was no difference in coliform counts between pre-harvested and harvested berry samples (m = 0). Ha: There was a difference in coliform counts between pre-harvested and harvested berry samples {m * 0). T = ZRi+ = 45.5. Two-tailed hypothesis, where a = 0.05 and n = 11; therefore, utilizing a/2 and 1 - a/2\x\ Table A12 (Quantiles of the Wilcoxon signed ranks tests statistic: Conover, 1999). Tc = n(n + l)/2 - wx Tc(11,o.o2S) = U and Tc(nA975) = 55. Tc> T; therefore, accept / / 0 . 61 A. 4. Coliform contribution of processing lines of blueberry and raspberry processors (Wilcoxon signed rank test). Pre-processed Processed Rank of \D\ n (log 1 0 coliform counts) Difference (D/) R? 1 6.3 5.6 -0.7 10 0 2 5.2 4.9 -0.3 4 0 3 5.1 3.3 -1.8 16.5 0 4 5.1 4.5 -0.6 8 0 5 4.7 4.3 -0.4 5.5 0 6 2.9 2.8 -0.1 1 0 7 3.1 3.3 0.2 2.5 2.5 8 3.3 1.5 -1.8 16.5 0 9 6.3 4.8 -1.5 15 0 10 4.0 4.9 0.9 12 12 11 6.1 5.2 -0.9 12 0 12 5.1 4.7 -0.4 5.5 0 13 6.0 5.4 -0.6 8 0 14 4.3 4.0 -0.3 2.5 0 15 5.0 3.8 -1.2 14 0 16 5.0 4.4 -0.6 8 0 17 3.0 3.9 0.9 12 12 Mean -0.5 ZRK+) 26.5 a Rank of positive £>/values. Ha. Pre-processed berries have equivalent or lower coliform counts than processed berries 'm > 0). Ha: Pre-processed berries have higher coliform counts than processed berries (m < 0). T= ZRi+ = 26.5. One-tailed hypothesis, where a = 0.05 and n = 17; therefore, utilizing Table 7. (Conover, 1999), TC(17A05) = 42. Tc > T; therefore, reject H0. 62 A. 5. McNemar's test for significance of changes between CCA and PEC coliform and £. coli counts. Within refers to values within the accuracy range (25-250 CFU/plate), while outside refers to values considered to be estimates (0 < and < 25 or 250 < CFU/plate). Coliform counts. Classification PEC-coliform counts Condition Within Outside Total CCA-coliform counts | Within 52 62 114 Outside 12 13 25 139 H0: P(CCA-coliform counts = outside) = P(PEC-coliform counts = outside) Ha: P(CCA-coliform counts = outside) *• P(PEC-coliform counts = outside) 7} = (b- cf/b + c(McNemar's test equation: Conover, 1999) b and c. the number of non-ties from the two available groups. 7} = (62 - 12)2/62 + 12 = 33.8 Aos.i= 3.84 < 33.8 (TO Therefore, reject H0. £. coli counts. Classification P E C - E co/icounts Condition Within Outside Total C C A - E co//counts | Within 19 14 33 Outside 0 94 94 127 H0: P(CCA-E coli counts = outside) = P(PEC- E coli counts = outside) Ha: P(CCA- E coli counts = outside) * P(PEC- E coli counts = outside) Ti = (T2 - cf/T2 + c(McNemar's test equation: Conover, 1999) T2 = b, if b + c<20; 14 + 0 < 20. 63 Rearrange testing, reject H0\fT2<torT2>n-t, otherwise, accept H0. Using the following parameters, p = 0.5, n = 14 (n = b + c) and a2 = 0.025 (ai = a/2), obtain f (equivalent to y) value from Table A3 (Binomial Distribution: Conover, 1999). From Table A3, t = 3, 14 < 3 is false, while 14 > (14 - 3) is true; therefore, if either statement is true, reject H0. 64 A. 6. Spearman's (p) correlation: rank of CCA and PEC coliform counts (Spearman's) Pearson correlation of Rank CCA and Rank PEC = 0.20 H0: PEC and CCA coliform counts are mutually independent (no correlation). Ha\ PEC and CCA coliform counts are positively correlated to each other. A large PEC coliform count corresponds to a large CCA coliform count. Where n = 54, a = 0.05 and p = 0.20. Using Table A10 and equation wp = xp/^(n-\X (1.6449/^(53) (Quantiles of the Spearman test statistic: Conover, 1999), Spearman's test statistic pC(54,0.95) = 0.23. p c (0.23) > r(0.20); therefore, accept H0. A. 7. Spearman's correlation: rank of CCA and PEC E coli counts. (Spearman's) Pearson correlation of EC Rank CCA and EC Rank PEC = 0.68 H0: PEC and CCA E coli counts are mutually independent (no correlation). Ha: PEC and CCA E coli counts are positively correlated to each other. A large PEC E coli count corresponds to a large CCA E coli count. Where n = 19, a = 0.05 and p = 0.68. Using Table A10 (Quantiles of the Spearman test statistic: Conover, 1999) Spearman's test statistic Pc(19,0.95) = 0.39. p c (0.39) < /-(0.68); therefore, reject 65 A. 8. Kruskal-Wallis test on CCA coliform counts from raspberry growers. Growers N Median Ave Rank Z R-a 18 2 .900 19 .4 -2 .89 R-P 19 3 . 900 39.5 3 .62 R-y 19 3 .400 26.2 -0. 77 Overall 56 28.5 H = 14.69 DF = 2 P = 0.001 H = 14.72 DF = 2 P = 0.001 (adjusted f o r ties) A. 9. Kruskal-Wallis test on CCA coliform counts from blueberry growers. Growers N Median Ave Rank Z B-ct 17 3.600 16.8 -4.56 B -P 24 5.300 57.0 5.17 B-8 16 4.050 26.5 -2.37 B - Y 18 4.950 42.9 1.09 Overall 7 5 38.0 H = 39.59 DF = 3 P = 0.000 H = 39.90 DF = 3 P = 0.000 (adjusted f o r t i e s ) A. 10. Kruskal-Wallis test on CCA E coli counts from raspberry growers. Growers N Median Ave Rank Z R-a 18 1.000 23 .5 -1 . 58 R-P 19 1.000 29.7 0 .39 R-y 19 1.000 32 .1 1. 17 Overall 56 28.5 H=2.69 D F = 2 P=0.260 H=6.04 D F = 2 P = 0.049 (adjusted f o r t i e s ) A. 11. Kruskal-Wallis test on CCA E coli counts from blueberry growers. Growers N Median Ave Rank Z B-a 17 1.000 29.3 -1 .88 B -P 24 2.500 57 .4 5 .29 B-8 16 1.000 28.7 -1 92 B - Y . 18 1.000 28 . 6 -2 . 09 Overall 75 38.0 H = 27 . 96 DF = 3 P = 0.000 H = 43.21 DF = 3 P = 0.000 (adjusted f o r t i e s ) A. 12. Analysis of variance of ranked coliform data from raspberry growers. Source DF Raspberry 2 Error Total Level R-a R-P R-Y 53 55 N 18 19 19 SS 3907 10689 14596 Mean 19.36 39.50 26.16 MS 1953 202 StDev 12 .11 15 .31 14.87 Pooled StDev = 14.20 Tukey's pairwise comparisons Family error rate = 0.0500 Individual error rate = 0.0194 C r i t i c a l value = 3.41 Intervals for (column l e v e l mean) R-P R-Y R-a -31.40 -8.88 -18 . 06 4.47 R-P 2 .23 24.45 F 9 . 68 P 0.000 Individual 95% CIs For Mean Based on Pooled StDev ( * ) (-+ --20 -- + -30 (row l e v e l mean) A. 13. Analysis of variance of ranked coliform data from blueberry growers. Source DF SS MS F P Blueberry 3 18803 6268 27.69 0.000 Error 71 16073 226 Total 74 34876 Individual 95% CIs For Mea Based on Pooled StDev Level N Mean StDev + + + _ B-a 17 16.82 13 .81 ( * ) B -P 24 56.98 12 .58 B-8 16 26.53 17 .22 ( * ) B-Y 18 42 . 89 17 . 04 ( * _ _ . Pooled StDev = 15 . 05 15 30 45 Tukey's pairwise comparisons Family er r o r rate = 0.0500 Individual error rate = 0.0105 C r i t i c a l value = 3 .72 Intervals f o r (column l e v e l mean) - (row l e v e l mean) B-a B-P B-8 B -P -52 .70 -27 . 61 B-8 -23 .49 4.08 17 . 67 43 .22 B-y -39.45 -12 . 68 1.75 26 .43 -29.96 -2 .76 A. 14. Analysis of variance of ranked E a?//data from raspberry growers. Source Raspberry-Error Total Level R-a R-P R-Y DF 2 53 55 N 18 19 19 Pooled StDev SS 716 5805 6522 Mean 23 . 50 29 . 68 32 . 05 10.47 MS 358 110 Tukey's pairwise comparisons Family error rate = 0.0500 Individual error rate = 0.0194 C r i t i c a l value = 3.41 Intervals f o r (column l e v e l mean) R-a R-P R-P -14 .48 2 .12 F 3 .27 P 0 . 046 Individual 95% CIs For Mean Based on Pooled StDev StDev + + + + 0.00 ( * ) 12.40 ( * ) 12.99 ( * ) 20.0 25.0 30.0 35.0 (row l e v e l mean) R-Y -16.85 -0.25 -10.56 5.82 A. 15. Analysis of variance of ranked E co//data from blueberry growers. Source DF SS MS F P Blueberry 3 13282 4427 33 .22 0.000 Error 71 9461 133 Total 74 22743 68 Individual 95% CIs For Mean Based on Pooled StDev Level N Mean StDev _ + + + + _ B-a 17 29.26 9.34 ( - - - * ) B-3 24 57.40 16.86 ( - - - * - - -B-8 16 28.72 6.87 ( * ) B-y 18 28.64 6.95 ( * _ _ _ ) _ + + + + _ Pooled StDev = 11.54 24 36 48 60 Tukey's pairwise comparisons Family error rate = 0.0500 Individual error rate = 0.0105 C r i t i c a l value = 3.72 Intervals for (column l e v e l mean) - (row l e v e l mean) B-a B -P B -8 B - P -37 .76 -18 .51 B-8 -10 .03 18.88 11 .12 38.48 B-Y -9 .64 19.29 -10.35 10 .90 38.22 10.51 69 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0089950/manifest

Comment

Related Items