Citation
Microbial and sensory assessment of milk with an electronic nose

Material Information

Title:
Microbial and sensory assessment of milk with an electronic nose
Creator:
Korel, Figen
Publication Date:
Language:
English
Physical Description:
xvii, 174 leaves : ; 29 cm.

Subjects

Subjects / Keywords:
Bacillus ( jstor )
Electronics ( jstor )
Microbial load ( jstor )
Microorganisms ( jstor )
Milk ( jstor )
Nose ( jstor )
Odors ( jstor )
Sensors ( jstor )
Storage time ( jstor )
Whole milk ( jstor )
Dairy products -- Sensory evaluation ( lcsh )
Dissertations, Academic -- Food Science and Human Nutrition -- UF ( lcsh )
Food Science and Human Nutrition thesis, Ph. D ( lcsh )
Milk -- Microbiology ( lcsh )
Olfactometry ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 2000.
Bibliography:
Includes bibliographical references (leaves 163-173).
General Note:
Printout.
General Note:
Vita.
Statement of Responsibility:
by Figen Korel.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
022943510 ( ALEPH )
45117388 ( OCLC )

Downloads

This item has the following downloads:


Full Text











MICROBIAL AND SENSORY ASSESSMENT OF MILK
WITH AN ELECTRONIC NOSE











By

FIGEN KOREL


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2000
























To my parents, Zerrin and Metin Korel, for all their love, encouragement, support and for the invaluable opportunity they gave me throughout my life to obtain the best education possible.













ACKNOWLEDGMENTS


I would like to express my sincere thanks and appreciation to my major advisor, Dr. Murat 0. Balaban, for his invaluable guidance, patience, encouragement, and friendship during my graduate studies. My appreciation is also extended to my committee members, Drs. Portier, Rodrick, Sims, and Williams for their guidance and recommendations for the success of this study.

I want to thank Celal Bayar University (Manisa, Turkey) for giving me the opportunity to conduct my graduate studies in the United States.

I would like to especially thank Dr. Diego A. Luzuriaga for all his invaluable assistance and unconditional patience during this project. I would like to thank Ash Z. Odaba i for her constant support and friendship throughout the completion of this study. My sincere and grateful esteem to Bee Mach for helping me anytime when I needed and Necla Demir for helping me in running some of the experiments. I also appreciate the panelists that helped in the sensory studies.














TABLE OF CONTENTS



ACKNOWLEDGMENTS............................................iii

LIST OF TAB LES .. ................................................. vii

LIST OF FIGURES .................................................. xiii

AB STRA CT .. ..................................................... xvi


CHAPTERS

1 INTRODU CTION ................................................ 1

2 LITERATURE REVIEW ........................................... 3

M ilk Quality Assessment ............................................ 3
M ilk Spoilage ................................................ 4
M ilk Volatiles and Off-Flavors .................................. 15
Factors Affecting Shelf Life of Milk .............................. 18
Rapid Methods for the Detection of Microorganisms ..................... 20
Immunomagnetic Separation (IMS) .............................. 21
Impedance Microbiology .................................... 21
Enzyme Immunoassays and Latex Agglutination Tests ................ 22
Bioluminescent Systems ....................................... 23
M iniaturized M ethods ........................................ 24
Other Biochemical M ethods .................................... 25
Electronic N ose ................................................. 26
Electronic Nose Technology .................................... 28
Sensor Technology ........................................... 30
Applications of Electronic Nose ................................. 34
Applications in Microbial Detection .............................. 35
Applications to Food Products .................................. 37
Current Shortcomings of the Electronic Nose ....................... 41
O bjectives ...................................................... 42









3 MATERIAL AND METHODS ......................


Milk Sampling, Inoculation, and Analysis .............................. 44
M ilk Sam ples ............................................... 44
Activation of Microorganisms .................................. 45
Inoculation of Microorganisms and Sample Treatments ............... 46
Electronic Nose Measurements .................................. 47
M icrobial Analysis ........................................... 48
Moisture Content Measurements ................................ 50
Fat Content M easurements ..................................... 50
pH M easurements ........................................... 50
Sensory Evaluation ........................................... 51
D ata Analysis ............................................ . 51

4 RESULTS AND DISCUSSION ..................................... 53

Milk Sampling, Inoculation, and Analysis .............................. 53
Moisture and Fat Content Measurements .......................... 53
M icrobial Analysis ........................................... 53
pH M easurements ........................................... 67
Sensory Evaluation ........................................... 74
Electronic Nose M easurements .................................. 81

5 CONCLUSIONS AND RECOMMENDATIONS ....................... 122

APPEN D ICES ..................................................... 124

A DATA FOR MICROBIAL ANALYSIS .............................. 124

B D ATA FOR pH ................................................ 135

C DATA FOR SENSORY ANALYSIS ................................ 149

REFEREN CES ..................................................... 163

BIOGRAPHICAL SKETCH ........................................... 174


............... 44













LIST OF TABLES


Table page

2-1. Residual enzyme activity from psychrotrophic organisms in HTST pasteurized m ilk .............................................. 12

2-2. Residual activity of extracellular enzymes of psychrotrophic organisms after UHT sterilization ......................................... 14

2-3. Flavor defects associated with psychrotrophic Bacillus spp. grown in milk at 7.20C .................................................... 17

4-1. Microbial load of all types of milk samples inoculated with Pseudomonas fluorescens or Bacillus coagulans ................................ 54

4-2. Microbial load of all types of milk inoculated with Pseudomonas fluorescens or Bacillus coagulans in accelerated study ................ 63

4-3. Microbial load of whole milk samples inoculated with Pseudomonas fluorescens and Bacillus coagulans in combination study ............... 65

4-4. Microbial load of whole milk samples inoculated with Pseudomonas fluorescens and Bacillus coagulans in accelerated study ................ 67

4-5. pH of whole milk inoculated with Pseudomonasfluorescens or Bacillus coagulans .................................................. 68

4-6. pH of reduced-fat milk inoculated with Pseudomonasfluorescens or Bacillus coagulans ............................................ 69

4-7. pH of fat-free milk inoculated with Pseudomonasfluorescens or Bacillus coagulans .................................................. 70

4-8. pH of all types of milk inoculated with Pseudomonasfluorescens or
Bacillus coagulans in accelerated study ............................ 72








4-9. pH of whole milk control and inoculated with P. fluorescens and
B. coagulans samples .......................................... 73

4-10. pH of whole milk control and inoculated with P. fluorescens and
B. coagulans samples in accelerated study .......................... 73

4-11. Sensory scores for whole milk inoculated with P. fluorescens or
B . coagulans ................................................ 75

4-12. Sensory scores for reduced-fat milk inoculated with P. fluorescens or
B. coagulans ................................................ 76

4-13. Sensory scores of fat-free fat milk inoculated with P. fluorescens or
B. coagulans ................................................ 77

4-14. Average sensory scores (rounded) of all types of milk inoculated with
P. fluorescens or B. coagulans ................................... 78

4-15. Sensory scores for whole milk inoculated with P. fluorescens and
B. coagulans in combination study ................................ 80

4-16. Average sensory scores (rounded) of whole milk inoculated with
P. fluorescens and B. coagulans in combination study ................. 81

4-17. Correct classification rates obtained from the DFA of electronic nose
sensor readings compared with microbial counts of whole milk
experiment 1 and experiment 2 samples separately .................... 83

4-18. DFA coefficients for microbial counts correlated to electronic nose
sensor readings. Whole milk experiment 1 samples inoculated with
Pseudomonasfluorescens or Bacillus coagulans ..................... 84

4-19. DFA coefficients for microbial counts correlated to electronic nose
sensor readings. Whole milk experiment 2 samples inoculated with
Pseudomonasfluorescens or Bacillus coagulans ..................... 90

4-20. DFA coefficients for microbial counts correlated to electronic nose
sensor readings. Whole milk experiment 1 and 2 samples inoculated
with Pseudomonasfluorescens or Bacillus coagulans ................. 93

4-21. Correct classification rates obtained from the DFA of electronic nose
sensor readings compared with microbial counts of whole milk
experiment 1 and experiment 2 samples pooled together ................ 95









4-22. Correct classification rates obtained from the DFA of electronic nose
sensor readings compared with microbial counts of reduced-fat milk
experiment 1 and experiment 2 samples separately .................... 98

4-23. Correct classification rates obtained from the DFA of electronic nose
sensor readings compared with microbial counts of fat-free milk
experiment 1 and experiment 2 samples separately .................... 99

4-24. Correct classification rates obtained from the DFA of electronic nose
sensor readings compared with sensory scores of whole milk
experiment 1 and experiment 2 samples separately ................... 101

4-25. DFA coefficients for sensory scores correlated to electronic nose sensor
readings. Whole milk experiment I samples inoculated with
Pseudomonasfluorescens or Bacillus coagulans .................... 102

4-26. DFA coefficients for sensory scores correlated to electronic nose sensor
readings. Whole milk experiment 2 samples inoculated with
Pseudomonasfluorescens or Bacillus coagulans .................... 103

4-27. Correct classification rates obtained from the DFA of electronic nose
sensor readings compared with sensory scores of whole milk
experiment 1 and experiment 2 samples pooled together ............... 107

4-28. DFA coefficients for sensory scores correlated to electronic nose sensor
readings. Whole milk experiment 1 and 2 samples inoculated with
Pseudomonasfluorescens or Bacillus coagulans .................... 108

4-29. Correct classification rates obtained from the DFA of electronic nose
sensor readings compared with sensory scores of reduced-fat milk
experiment I and experiment 2 samples separately ................... 112

4-30. Correct classification rates obtained from the DFA of electronic nose
sensor readings compared with sensory scores of fat-free milk
experiment 1 and experiment 2 samples separately ................... 113

4-31. Eigenvalues of the principal component factors obtained by PCA for
data sets of each type and experiment of milk ....................... 114

4-32. Correct classification rates obtained from the DFA of electronic nose
sensor readings compared with sensory scores of whole milk
samples inoculated with P. fluorescens and B. coagulans .............. 119








4-33. DFA coefficients for sensory scores correlated to electronic nose
sensor readings. Whole milk samples inoculated with Pseudomonas
fluorescens and Bacillus coagulans .............................. 120

A-1. Microbial load of whole milk inoculated with Pseudomonasfluorescens .... 124 A-2. Microbial load of whole milk inoculated with Bacillus coagulans ......... 125

A-3. Microbial load of reduced-fat milk inoculated with Pseudomonas
fl uorescens ................................................ 126

A-4. Microbial load of reduced-fat milk inoculated with Bacillus coagulans ..... 127 A-5. Microbial load of fat-free milk inoculated with Pseudomonasfluorescens ... 128 A-6. Microbial load of fat-free milk inoculated with Bacillus coagulans ........ 129

A-7. Microbial load of whole milk inoculated with Pseudomonasfluorescens
and Bacillus coagulans ....................................... 130

A-8. Microbial load of whole milk inoculated with Pseudomonasfluorescens or
Bacillus coagulans in accelerated study ............................ 131

A-9. Microbial load of reduced-fat milk inoculated with Pseudomonas
fluorescens or Bacillus coagulans in accelerated study ............... 132

A-10. Microbial load of fat-free milk inoculated with Pseudomonasfluorescens
or Bacillus coagulans in accelerated study ........................ 133
A-i 1. Microbial load of whole milk inoculated with Pseudomonasfluorescens
and Bacillus coagulans in accelerated study ....................... 134

B-I. pH of whole milk without inoculated with microorganisms (control) ....... 135 B-2. pH of whole milk inoculated with Pseudomonasfluorescens ............. 136

B-3. pH of whole milk inoculated with Bacillus coagulans .................. 137

B-4. pH of reduced-fat milk without inoculated with microorganisms (control) ... 138 B-5. pH of reduced-fat milk inoculated with Pseudomonasfluorescens ......... 139

B-6. pH of reduced-fat milk inoculated with Bacillus coagulans .............. 140








B-7. pH of fat-free milk without inoculated with microorganisms (control) ...... 141 B-8. pH of fat-free milk inoculated with Pseudomonasfluorescens ............ 142

B-9. pH of fat-free milk inoculated with Bacillus coagulans ................. 143

B-10. pH of whole milk inoculated with Pseudomonasfluorescens or Bacillus
coagulans in accelerated study ................................. 144

B-i 1. pH of reduced-fat milk inoculated with Pseudomonasfluorescens or
Bacillus coagulans in accelerated study .......................... 145

B-12. pH of fat-free milk inoculated with Pseudomonasfluorescens or Bacillus
coagulans in accelerated study ................................. 146

B-13. pH of whole milk inoculated with Pseudomonasfluorescens and Bacillus
coagulans ................................................. 147

B-14. pH of whole milk inoculated with Pseudomonasfluorescens and Bacillus
coagulans in accelerated study ................................. 148
C-1. Sensory data of whole milk without inoculated with microorganisms
(control) .................................................. 149

C-2. Sensory data of whole milk inoculated with Pseudomonasfluorescens ..... 150 C-3. Sensory data of whole milk inoculated with Bacillus coagulans .......... 151

C-4. Sensory data of whole milk for hidden control ........................ 152

C-5. Sensory data of reduced-fat milk without inoculated with microorganisms
(control) .................................................. 153

C-6. Sensory data of reduced-fat milk inoculated with Pseudomonasfluorescens . 154 C-7. Sensory data of reduced-fat milk inoculated with Bacillus coagulans ...... 155 C-8. Sensory data of reduced-fat milk for hidden control .................... 156

C-9. Sensory data of fat-free milk without inoculated with microorganisms
(control) .................................................. 157

C-10. Sensory data of fat-free milk inoculated with Pseudomonasfluorescens .... 158








C- 11. Sensory data of fat-free milk inoculated with Bacillus coagulans ......... 159

C-12. Sensory data for fat-free milk for hidden control ...................... 160

C- 13. Sensory data of whole milk inoculated with Pseudomonasfluorescens
and Bacillus coagulans ....................................... 161

C- 14. Sensory data of whole milk inoculated with Pseudomonasfluorescens
and Bacillus coagulans for hidden control ......................... 162













LIST OF FIGURES


Figure pag

4-1. Average microbial load of P. fluorescens for whole milk, both experiments stored at different temperatures over time. Error bars signify �1 standard
deviation ................................................... 56

4-2. Average microbial load of P. fluorescens for reduced-fat milk, both experiments stored at different temperatures over time. Error bars
signify �1 standard deviation .................................... 57

4-3. Average microbial load of P. fluorescens for fat-free milk, both experiments stored at different temperatures over time. Error bars signify �1 standard
deviation ................................................... 58

4-4. Average microbial load of B. coagulans for whole milk, both experiments stored at different temperatures over time. Error bars signify � 1 standard
deviation ................................................... 59

4-5. Average microbial load of B. coagulans for reduced-fat milk, both experiments stored at different temperatures over time. Error bars
signify � I standard deviation .................................... 60

4-6. Average microbial load of B. coagulans for fat-free milk both experiments stored at different temperatures over time. Error bars signify +1 standard
deviation ................................................... 6 1

4-7. Average microbial load of P. fluorescens and B. coagulans for whole milk in combination study, stored at different temperatures over time.
Error bars signify +1 standard deviation ........................... 66

4-8. DFA of odors of whole milk experiment 1 samples inoculated with P. fluorescens and stored at 1.7', 7.20, and 12.8�C based on microbial
counts and electronic nose readings ............................... 85








4-9. DFA of odors of whole milk experiment 1 samples inoculated with
B. coagulans and stored at 1.70, 7.20, and 12.8�C based on microbial
counts and electronic nose readings ............................... 86

4-10. DFA of odors of whole milk experiment 1 samples based on microbial
counts and electronic nose readings using discriminant function I
and 2 obtained from DFA of all temperatures analysis ................. 88

4-11. DFA of odors of whole milk experiment 1 samples based on microbial
counts and electronic nose readings using discriminant function I
and 2 obtained from DFA of all temperatures analysis ................. 89

4-12. DFA of odors of whole milk experiment 2 samples inoculated with
P. fluorescens and stored at 1.70, 7.20, and 12.8�C based on microbial
counts and electronic nose readings ............................... 91

4-13. DFA of odors of whole milk experiment 2 samples inoculated with
B. coagulans and stored at 1.7' and 7.2�C based on microbial counts
and electronic nose readings ..................................... 92

4-14. DFA of odors of whole milk experiment I and experiment 2 samples
inoculated with P. fluorescens and stored at 1. 7, 7.20, and 12.8�C
based on microbial counts and electronic nose readings ................ 96

4-15. DFA of odors of whole milk experiment 1 and experiment 2 samples
inoculated with B. coagulans and stored at 1.70, 7.20, and 12.8�C
based on microbial counts and electronic nose readings ................ 97

4-16. DFA of odors of whole milk experiment 1 samples inoculated with
P. fluorescens and stored at 1. 7, 7.20, and 12.8�C based on sensory
scores and electronic nose readings .............................. 104

4-17. DFA of odors of whole milk experiment 2 samples inoculated with
P. fluorescens and stored at 1.70, 7.20, and 12.8�C based on sensory
scores and electronic nose readings .............................. 105

4-18. DFA of odors of whole milk experiment 2 samples inoculated with
B. coagulans and stored at 1.70, 7.20, and 12.8�C based on sensory
scores and electronic nose readings .............................. 106

4-19. DFA of odors of whole milk experiment I and experiment 2 samples
inoculated with P. fluorescens and stored at 1. 70, 7.20, and 12.8�C
based on sensory scores and electronic nose readings ................. 109









4-20. DFA of odors of whole milk experiment 1 and experiment 2 samples
inoculated with B. coagulans and stored at 1.7', 7.20, and 12.8�C
based on sensory scores and electronic nose readings ................. 110

4-21. Change of shelf life with respect to DPC 1 and BPC I for whole milk
inoculated with P. fluorescens in accelerated study ................... 115

4-22. Change of shelf life with respect to DPC 1 and BPC I for reduced-fat
milk inoculated with P. fluorescens in accelerated study ............... 116

4-23. Change of shelf life with respect to DPC I and BPC I for fat-free milk
inoculated with P. fluorescens in accelerated study ................... 117

4-24. DFA of odors of whole milk samples inoculated with both microorganisms
and stored at 1.70, 7.20, and 12.8�C based on sensory scores and
electronic nose readings ....................................... 121













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

MICROBIAL AND SENSORY ASSESSMENT OF MILK WITH AN ELECTRONIC NOSE By

Figen Korel

May 2000

Chairperson: Murat 0. Balaban Major Department: Food Science and Human Nutrition

Important psychrotrophs encountered in raw milk are Gram-negative rods with

Pseudomonas spp. comprising 65 to 70% of the genera. Some Gram-positive bacteria are also present, with Bacillus being the most important genera. Spoilage bacteria found in raw milk produce heat-resistant lipases and proteinases that are not destroyed by pasteurization. In general, microbial counts in excess of x106 cfu/ml are enough to produce defects in milk.

Serious off-flavors in milk, such as bitter, putrid, unclean, rancid, and sour, have been associated with psychrotrophic microorganisms. Traditional microbial evaluation of milk is time consuming. Faster methods are desirable. The electronic nose is a promising technology that can be used as a fast screening tool. It enhances objectivity of flavor evaluation, requires minimal sample preparation, generates reproducible and reliable








results, is easy to operate, and results can be obtained rapidly. The objectives of this study were to test an electronic nose for odor assessment of milk inoculated with Pseudomonas fluorescens and/or Bacillus coagulans, to correlate microbial loads and sensory results with electronic nose readings, and to attempt to predict the shelf life based on microbial loads of milk samples in an accelerated study.

Parmalat� whole, reduced-fat, and fat-free milk were used. Sterile milk samples

were inoculated with P. fluorescens and/or B. coagulans, stored at 1.70, 7.20, and 12.8�C, and evaluated at days 0, 3, 5, 7, and 10 using an electronic nose. Counts for P. fluorescens were performed using aerobic plate count 3M Petrifilm. Those for B. coagulans were performed using nutrient agar plates. The odor of milk samples was evaluated by a 10-member untrained sensory panel. Electronic nose readings, microbial counts, and sensory data were analyzed using discriminant function analysis. The electronic nose discriminated differences in odor due to microbial load, storage temperature, and sensory data. This research demonstrated the potential use of electronic nose to detect odor differences in milk due to microbial loads. Electronic nose readings can be correlated with sensory panel perception. This may lead to a new rapid method for determining sensory evaluation and microbial loads of milk.














CHAPTER 1
INTRODUCTION


Milk is a good medium for growth of pathogenic and spoilage organisms. Raw milk contains varying numbers of microorganisms, depending on the care employed in milking, cleaning, and handling of milk utensils. Raw milk held at refrigeration temperatures for several days shows the presence of several bacteria of the following genera: Pseudomonas, Enterococcus, Lactococcus, Streptococcus, Leuconostoc, Lactobacillus, Microbacterium, Micrococcus, Propionibacterium, coliforms, Proteus, Bacillus, and some others. Those able to grow at low temperatures tend to increase in numbers (Bramley and McKinnon, 1990).

Defects in milk can arise from four sources: the growth of psychrotrophic

organisms prior to pasteurization, the activity of thermoresistant enzymes, the growth of thermoresistant psychrotrophic organisms, and post-pasteurization contamination. Psychrotrophic organisms, although mostly not thermoduric, are important because many produce extracellular thermostable proteolytic and lipolytic enzymes which can survive pasteurization and even ultra-high temperature (UHT) processing (Rowe and Gilmour, 1985). These extracellular enzymes hydrolyze milk proteins and lipids and cause offflavors in UHT milk. In general, psychrotrophic counts in excess of 1 x 106 colony forming units (cfu) /ml are required to produce defects on quality (Cousin, 1982).








2
The dairy industry needs a simple, rapid, sensitive, reliable, and economical method for assessing psychrotrophic organism populations in raw and pasteurized milk. A fast, non-complex, and inexpensive method for reliable detection of psychrotrophic organisms will be a valuable quality control tool for the dairy industry.

Each type of bacteria has a 'signature' of volatile products that form a unique

odor. The method of sensing the bacteria can be to 'smell' the bacterial metabolites by the use of sensor arrays, also known as electronic noses. Due to its sensitivity, the electronic nose has a great potential in microbiological analysis. Six types of bacteria (Clostridium perfringens, Proteus, Haemophilus influenzae, Bacteroidesfragilis, Oxford Staph, and Pseudomonas aeroginosa) were discriminated by using a 4-element metal oxide sensor array. Escherichia coli and Staphylococcus aureus were also discriminated (Gardner and Craven, 1996). Penicillium species, which produce different kinds of volatile metabolites, were separated by an electronic nose (Olsson et al., 1995). Electronic noses have many applications in the food industry, such as monitoring the deterioration of shrimp (Luzuriaga and Balaban, 1999b), tuna (Newman, 1998), and ground meat (Winquist et al., 1993). Therefore, electronic noses have the potential to detect and discriminate microorganisms based on volatile bacterial metabolites.













CHAPTER 2
LITERATURE REVIEW



Milk Quality Assessment



Milk is an important element of a balanced diet (Muir, 1990). The major

nutritional components of milk and their normal concentrations are 87.3% water, 4.6% lactose, 4.2% fat, 3.25% protein, and 0.65% minerals. It is a good source of B vitamins and minerals such as iron, copper, cobalt, and molybdenum (Frank, 1997). Indeed, milk is not only an excellent food for humans, it is also an ideal medium for the growth of microorganisms (Muir, 1990).

Carbon sources in milk are lactose, protein, and fat. Many microorganisms cannot utilize lactose, and therefore proteolysis or lipolysis must occur for them to obtain carbon and energy. However, some spoilage microorganisms may oxidize lactose to lactobionic acid. The amount of lactose present in milk is enough to support extensive microbial growth. The citrate in milk can be used by many microorganisms, but the amount is not sufficient to support significant growth. There is sufficient glucose in milk to initiate the growth of some microorganisms (Frank, 1997).

Two types of proteins, casein and whey proteins, are present in milk. Caseins are found in the form of highly hydrated micelles and are readily susceptible to microbial










proteolysis. Whey proteins (3-lactoglobulin, a-lactalbumin, serum albumin, and immunoglobulin) remain soluble in the milk after casein precipitates. In contrast to casein, whey proteins are less susceptible to microbial proteolysis (Frank, 1997).

Milk has significant fat content, but only few spoilage microorganisms utilize this fat as a carbon or energy source. The fat is in the form of globules surrounded by a protective membrane composed of glycoproteins, lipoproteins, and phospholipids. Unless the globule membrane is physically damaged or enzymatically degraded, it cannot be utilized by microorganisms (Alkanhal et al., 1985). Milk Spoilage


From a milk spoilage perspective, psychrotrophic microorganisms are the single

most important group. They are defined as bacteria that grow at 7'C or below, regardless of their optimal growth temperature. These organisms have constituted an important part of the microbial flora of raw milk since the introduction of bulk refrigerated storage. Growth of psychrotrophic microorganisms in raw milk can lead to quality and flavor defects in products made from that milk because of the residual activity of degradative heat-stable enzymes produced by these microorganisms (Bramley and McKinnon, 1990; Frank, 1997).

Milk, produced at ambient temperatures without refrigeration, must be cooled to about 3-50C at the farm. The initial microflora, the numbers, and types of microorganisms in milk immediately after production, reflect microbial contamination during production. The cooling after production inhibits general growth of bacteria. The temperature to








5
which raw milk has been cooled, the duration of milk storage and the storage temperature on the farm can affect the numbers and types of microorganisms present in raw milk (Bramley and McKinnon, 1990). Once milk leaves the farm, active refrigeration stops, and the temperature of the milk rises by at least I�C per day. Any temperature rise will enhance the growth of psychrotrophic microorganisms (Muir and Phillips, 1984).

The psychrotrophic microorganisms in milk produce a range of extracellular

enzymes which can readily degrade milk constituents. These microorganisms have either lipolytic or proteolytic or combined degradative ability. Even though psychrotrophic organisms can be killed by pasteurization or ultra-high temperature (UHT) process, their enzymes cannot be inactivated (Bramley and McKinnon, 1990). Growth of psychrotrophic microorganisms in milk can lead to spoilage because of the heat-stable degradative enzymes of these organisms. Consumers can detect these quality and flavor defects.

There are three main sources of microbial contamination of milk during

production: from within the udder, from the exterior of the teats and udder, and from the milking and storage equipment (Bramley and McKinnon, 1990). Soil, water, animals, and plant materials are sources of psychrotrophic microorganisms found in milk. The exterior of the teats and udder can harbor high levels of psychrotrophic bacteria, even after washing and sanitizing.

Water used on the dairy farm usually contains low populations of psychrotrophs; its use to clean and rinse milking equipment provides a direct contamination into milk.










Psychrotrophs isolated from water are often very active producers of extracellular enzymes, and grow rapidly at low temperatures (Cousin, 1982).

Plant materials used for animal feed have been found to contain 10' psychrotrophic organisms/g (Thomas, 1966). Milking and storage equipment are also major sources of psychrotrophic contamination of raw milk. Proper cleaning and sanitizing procedures can reduce contamination from such equipment. Good handling practices inhibit contamination and prevent high microbial counts, and the possible presence of undesirable bacterial enzymes (Bramley and McKinnon, 1990). Raw Milk

The main psychrotrophic microflora present in raw milk are aerobic Gram-negative rods with Pseudomonas spp. (P. fluorescens, P. putida, P. fragi, P. aeroginosa) forming 65 to 70% of the genera. P. fluorescens predominates (Bramley and McKinnon, 1990; Garcia et al., 1989; Reinheimer et al., 1990). Achromobacter, Acinetobacter, Aeromonas, Alcaligenes, Chromobacterium, Flavobacterium, and coliforms form the other genera present in milk (Bramley and McKinnon, 1990; Mikolajcik, 1979).

Some Gram-positive bacteria are also present in raw milk, but their numbers are much smaller than those of Gram-negative bacteria. Arthrobacter, Bacillus, Clostridium, Corynebacterium, Lactobacillus, Listeria, Microbacterium, Micrococcus, Sarcina, Staphylococcus, and Streptococcus are isolated from milk, and Arthrobacter and Bacillus are the most common (Bramley and McKinnon, 1990).

Psychrotrophs cause an "unclean" flavor in milk, and there is a significant

correlation between initial psychrotrophic counts and storage temperature of raw milk at










both 20C and 60C. A decrease in storage temperature from 60C to 20C leads to a 75% increase in storage life of raw milk. This was concluded by measuring the time taken for the microbial count to reach I x 106 cfu/ml (Griffiths et al., 1988a). Griffiths et al. (1988b) also investigated the effect of storage of raw milk at 20C and 6�C on the subsequent quality of pasteurized and UHT milk. Good quality pasteurized milk could be produced from raw milk stored at 20C for up to 5 days, but this quality could be achieved with the raw milk samples which had been stored at 6'C for only 2 days. The quality of UHT milk was much higher in products manufactured from raw milk stored at 2'C for 4 days than in those stored at 6�C for 4 days (Griffiths et al., 1988b).

Aerobic spore-forming bacteria can have frequent changes in minimum growth temperatures. Grosskopf and Harper (1974) stated that isolated Bacillus spp. subsequently lost their ability to grow at 7.2�C or below when they were stored at 21�C. However, mesophilic strains of Bacillus spp. have been adapted to grow at low temperatures by repeated transfers to media at colder temperatures over a long period of time.

Even though the majority of spore-forming organisms previously reported in milk were identified as Bacillus spp., anaerobic sporeformers are also present in milk. Micrococci and Microbacterium spp. are derived almost exclusively from milking equipment, and thermoduric counts in the milk sometimes exceed 5x104 cfu/ml. Most of the thermoduric organisms do not multiply appreciably in raw milk even at ambient temperatures, however, a high thermoduric count in milk is reliable evidence of cross contamination from milking equipment (Bramley and McKinnon, 1990).










The presence of coliforms and E. coli in raw milk is evidence partly of direct fecal contamination, and partly inadequate cleaning of milking equipment since coliforms can rapidly build up in moist, milky residues in milking equipment, and become the major source of contamination. Coliform counts higher than 100 cfu/ml are considered as evidence of unsatisfactory production hygiene. Unrecognized coliform mastitis may also cause high coliform counts (Bramley and McKinnon, 1990).

All important spoilage bacteria found in raw milk have the potential to produce extracellular degradative enzymes, heat-resistant lipases, and proteinases, regardless of growth conditions. When they are produced, they are not destroyed by simple heat treatment. These enzymes can play a role in the quality degradation of milk.

The fluorescing pseudomonads and bacteria in the genus Alcaligenes show the highest incidence of degradative action. P. flourescens and P. fragi are the major microorganisms causing lipolytic spoilage. It was found that P. fragi strains were more lipolytic than those of P. fluorescens, and grew faster at 5C (Shelley et al., 1987). Grampositive organisms show little proteolysis and only limited lipolytic activity. The strains of psychrotrophic Bacillus spp. are frequently only proteolytic, and their proteinases and lipases are less heat-stable than those from Pseudomonas spp. (Ewings et al., 1984). Pseudomonasfluorescens and Bacillus cereus also produce extracellular phospholipase C which can degrade the milk fat globule membrane leading to product defects. The phopholipase C from Bacillus cereus produces "bitty" cream, and exhibits similar effects in pasteurized milk.










The production of protease and lipase by psychrotrophic bacteria is not observed until all the cells are in the late exponential or stationary phase of their growth cycle. Such a stage with most cultures of dairy origin is reached when the cell population exceeds 10 cfu/ml. This concentration of microorganisms corresponds with the onset of detectable spoilage in raw milk (Muir et al., 1978).

A reduction in oxygen tension of the growth medium immediately precedes

protease synthesis by a strain of Pseudomonasfluorescens (Rowe and Gilmour, 1982). Griffiths and Phillips (1984) reported that protease synthesis by psychrotrophs in milk could be inhibited by maintaining high oxygen concentrations. However, other researchers stated that this effect may be strain specific, since they found that some bacteria continue to excrete both protease and lipase when the oxygen tension in the growth medium is high (Dring and Fox, 1983; Fox and Stepaniak, 1983).

The effect of growth temperature on the production of proteases and lipases by

psychrotrophic bacteria may also be strain specific. Growth temperature of 2"C depressed the synthesis of both protease and lipase by several pseudomonads (Griffiths et al., 1988a). This result shows that deep cooling of milk to 2'C extends its shelf life. Pasteurized Milk

Pasteurization ensures the microbiological safety of milk as well as commercially acceptable shelf life (Muir, 1990). However, increasing the pasteurization temperature does not necessarily result in an increased shelf life because of the activation of spores at the higher temperatures (Phillips and Griffiths, 1990).










In pasteurization, milk is heated to and retained at a temperature not less than

62.8'C, and not more than 65.6'C for at least 30 min, then immediately cooled to below 100C or below 6�C. Alternatively, the milk may be retained at a temperature of not less than 71.7'C for at least 15 sec, then immediately cooled to less than 10�C. This process is called the high-temperature short-time method (HTST). Another method, super pasteurization, is used in many places today. Milk is heated to and retained at a temperature of 82'C for 3 sec, then immediately cooled to I 0C (Bramley and McKinnon, 1990; Vela, 1997).

Pasteurization of milk was designed as a method of heating milk to kill

Mycobacterium tuberculosis, the organism that causes tuberculosis in humans. However, Coxiella burnetii, the organism that causes Q fever, is found to be more resistant to heat than M tuberculosis. Now the target organism for pasteurization is C. burnetii. The effectiveness of pasteurization is monitored by measurement of the residual level of alkaline phosphatase (Vela, 1997).

Microbial contamination of pasteurized milk occurs through post-pasteurization contamination by psychrotrophs and through survival of thermoduric psychrotrophic organisms during pasteurization. Many of the Gram-negative organisms do not survive pasteurization, in fact, only one species of Gram-negative bacterium, Alcaligenes tolerans, can survive since it is generally considered as thermoduric (Muir, 1990).

Spore-forming bacteria of the genus Bacillus form the most important group of

microorganisms capable of surviving pasteurization, and growing in milk. The occurrence of psychrotrophic, spore-forming bacteria in pasteurized milk was first reported in 1969.








11
The quality loss and unpalatability of pasteurized milk (stored at 4'C for four weeks) was observed due to the outgrowth of Bacillus coagulans (Grosskopf and Harper, 1974). They also isolated Bacillus coagulans from pasteurized milk that had been stored at 2'C for 13 to 17 days. They stated that the generation time for B. coagulans was 24 to 30 hrs under these refrigeration conditions.

A seasonal variation in the numbers of the genus Bacillus in raw milk is observed, and these bacteria predominate in the period of June to October. B. licheniformis is the most common species (Bramley and McKinnon, 1990). During this period there is a peak in the number of spoilage incidents in milk with bitter taint, or with the defect known as "bitty" cream (Muir, 1990). The only pathogen of the genus Bacillus is B. cereus, and it can survive, and grow at refrigeration temperature (Bramley and McKinnon, 1990).

Researchers in the US reported that psychrotrophic sporeformers were isolated from 28-35% of freshly pasteurized milk (Ahmed et al., 1983). Others observed heat activation of Bacillus spores at pasteurization temperatures, and stated that more than 95% of spores could be activated at pasteurization temperatures (Phillips and Griffiths, 1990).

Coryneform bacteria which may form a substantial proportion of the flora of the heat-treated milk grow very slowly at refrigeration temperatures (Seiler et al., 1984). Streptococcus spp. such as Streptococcus thermophilus, Enterobacterfaecalis, and Streptococcus brevis which are thermoduric grow very slowly at refrigeration temperatures. Therefore these bacteria pose no great threat to pasteurized products which










are adequately refrigerated between heat treatment and the ultimate consumer (Muir, 1990).

The quality of pasteurized milk improves with the reduction in levels of postpasteurization contamination. The majority of the psychrotrophic bacteria are destroyed by pasteurization, and only thermoduric flora, and extracellular enzymes from psychrotrophic bacteria are left to cause spoilage. The thermostability of proteases, lipases, and phospholipase C of bacterial origin to HTST pasteurization is shown in Table 2-1.



Table 2-1. Residual enzyme activity from psychrotrophic organisms in HTST pasteurized
milk.
Type of enzyme Activity after HTST pasteurization (%)
Protease 66 Lipase 59 Phospholipase C 30 Source: Muir (1990).



About 60% of the protease and lipase activity of the psychrotrophic bacteria remains after HTST treatment, and a lower, but significant, proportion of the phospholipase C activity remains. This thermostability pattern is observed for the proteases of most fluorescent and non-fluorescent Pseudomonas spp. and for enzymes from many strains of A icaligenes, A cinetobacter, Achromobacter, Enterobacteriaceae, and Moraxella.










Many strains of Bacillus spp. are heat resistant, but not their enzymes. Even

though growth temperature has a significant effect on the rate of bacterial growth, and on the synthesis of extracellular enzymes, when they are excreted, they exhibit similar degree of thermostability. Psychrotrophic bacteria produce thermostable enzymes almost regardless of species, state, and temperature of growth. The thermostability of the extracellular enzymes of the psychrotrophic bacteria in milk has a minor importance for short shelf life products such as pasteurized milk because the shelf life of pasteurized milk is usually about 14 to 20 days, and during this period the products are refrigerated. The initial bacterial counts in raw milk for heat treatment is well below the threshold of 10' cfu/ml where significant amounts of extracellular degradative enzymes are synthesized (Frank, 1997; Muir, 1990).

UHT Milk

In sterile or UHT milk, the important point is to destroy bacteria while limiting the chemically-induced changes such as browning, cooked, and caramelized flavors. UHT milk is either heated in a closed container to yield sterilized milk, or it is heat-treated in a continuous flow, and packed aseptically. The UHT milk is heated at 138-142�C for 2 to 5 sec. The test for sterilized or UHT milk is that samples incubated at 30'C for 15 days will have a plate count of not more than 100 cfu/ml (Bramley and McKinnon, 1990).

The spore-forming bacteria are the organisms relevant to spoilage in sterilized products, but many spores are destroyed by the heat-treatment applied to sterilized or UHT milk. Spores of Bacillus stearothermophilus constitute the greatest hazard to










spoilage of sterilized dairy products since they are remarkably heat-resistant, and require extended heat treatments to ensure a reasonable kill (Muir, 1990).

Residual activity of the extracellular enzymes from psychrotrophic bacteria after UHT sterilization may be as high as 40% of that in the raw milk. The residual activity of protease, lipase, and phospholipase C are given in Table 2-2. Muir (1996) measured the residual proteinase and lipase activity found after treating cell free supernatants at 1400C for 5 sec, and found that Acinetobacter, Aeromonas, and Bacillus spp. had residual activities below 10%, and fluorescent pseudomonads had residual activity ranging between 14 and 51%. There could be degradation of milk protein and fat in the case of a product with a shelf life typically of six months at room temperature (Muir, 1990).



Table 2-2. Residual activity of extracellular enzymes of psychrotrophic organisms after
UHT sterilization.
Residual activity after heat treatment (%) Enzyme type
140'C for 5 sec 140'C for 5 sec + 55'C for I hr Protease 29 17 Lipase 40 7 Phospholipase C 0 - 57 n.d. n.d. = not determined
Source: Muir (1990).


Milk containing different concentrations of a psychrotrophic, proteolytic

Pseudomonas spp. was sterilized, and even though their UHT products were sterile, proteolysis and gelation of the milk occurred. The initial psychrotrophic count affected the shelf life before the onset of gelation. When the microbial count was less than 5xl 06








15
cfu/ml, there was little evidence of an effect on the product quality. However, when this threshold was reached, the decline in shelf life progressed rapidly. There were similar findings when skim milk with counts in excess of 5x106 cfu/ml was UHT sterilized (142'C/2 sec). Premature gelation and bitterness occurred within three months of storage at room temperature. The evidence of similar defects can be seen in commercial products from time to time (Muir, 1990). Similar results were observed in the samples from dairy plants and farms, and 70 to 90% of the samples contained heat-resistant proteases. After heat treatment at 149'C for 10 sec, these enzymes were still active. It was found that for UHT-treated milk to have a shelf life of I year, the raw milk must contain less than 0.1 unit of protease/ml, and this amount can be easily synthesized within a day for some highprotease-producing bacteria. If casein and whey proteins have been damaged by enzymes of Gram-negative psychrotrophs, they have a predisposition to denaturation and precipitation by UHT treatment. Gelation of UHT milk can occur as a result of the proteolytic activity of these enzymes (Adams et al., 1981).

Lipases can cause fat degradation in UHT products. Aeration affects lipase production by Pseudomonas spp. For P. fragi, aeration reduced lipase production, however, a high aeration was required for high lipase activity of P. meplitica var fipolytica (Champagne et al., 1994).


Milk Volatiles and Off-Flavors


During microbial spoilage, bacteria produce metabolites which cause off-odors.

Each type of bacteria has a 'signature' of volatile products that form a unique odor. These








16
odors can be qualified and/or quantified by humans and instrumental methods such as gas chromatography.

The possibility of characterizing some bacteria by gas chromatographic analysis of headspace vapors from milk cultures was investigated. Bassette et al. (1967) reported the characteristic patterns and rates of development of acetaldehyde, ethanol, dimethyl sulphide, and diacetyl in cultures ofAerobacter aerogenes, Escherichia coli, Streptococcus faecalis var. liquefaciens, Achromobacter lipolyticum, Streptococcus diacetylactis, E. freundii, Sterptococcus lactis, Lactobacillus acidophilus, Lactobacillus casei, and Pseudomonasfragi. Dimethyl sulphide, acetaldehyde, 2-methylpropanal, acetone, ethanol, 3-methylbutanal, butanone, 2-methylpropanal-l-ol, and 3-methylbutan1-ol in milk cultures of Streptococcus lactis var. maltigenes were reported (Morgan et al., 1966). Volatile compounds such as esters, ethanol, propan-2-ol are produced in refrigerated milk by pseudomonads, and increase in ethanol could be detected when the level of microflora in pasteurized milk stored at 4-10'C reached 10' cfu/ml (Urbach and Milne, 1987).

The number of psychrotrophs required to produce off-flavors varies between

species, and is determined by the length of the lag period and growth rate at the storage temperature, and by the proteolytic activity and heat resistance of the enzymes. In the case of Pseudomonas spp. 2.7 x 106 to 9.3 x 107 cfu/ml and Alcaligenes spp. 2.2 x 106 to

3.6x1O cfu/ml were required to produce off-flavors. In general, microbial counts in the excess of I x 106 cfu/ml are required to produce defects on product quality (Cousin, 1982).








17
Serious off-flavors such as bitter, putrid, unclean, stale, rancid, fruity, yeasty, and sour have been frequently associated with the presence of thermoduric psychrotrophs in milk. Isolated Bacillus spp. were inoculated into sterile milk at a level of 0.5%, and incubated at 7.2�C. Their flavor associated defects are shown in Table 2-3.



Table 2-3. Flavor defects associated with psychrotrophic Bacillus spp. grown in milk at
7.20C.
Species Days to appearance of defect Flavor defect B. macerans 4-10 Fruity, sour B. polymyxa 4-6 Sour, yeasty, gassy B. laterosporus 4-12 Sweet curdling, unclean, bitter B. subtilis 2-4 Sweet curdling, bitter B. lentus 6 Sour B. cereus 6 Sweet curdling, bitter B. sphaericus 12 Unclean, sour Source: Washam et al. (1977).



The occurrence of a bitter flavor in milk is related to the presence of heat stable protease. Hydrolysis of lactoalbumin and casein yields bitter peptides. When 300 ng of heat stable protease from P. fluorescens was mixed with pasteurized milk having low bacterial count, bitterness developed after storage of 7 days at 7C (Baker, 1983).

Lipases produced by thermoduric psychrotrophs produce off-flavors in milk. Free fatty acids give rise to off-flavors such as rancid, butyric, bitter, unclean, soapy, and astringent (Champagne et al., 1994). The fruity flavor arises from free fatty acid esterification. Some thermoduric psychrotrophs produce phopholipases, particularly phopholipase C which is responsible for a specific degradative action on the fat globule










membrane. This increases the susceptibility of exposed milkfat to the action of lipases (McKellar, 1989).

Thermoduric psychrotrophs were inoculated (200 to 100 cfu/ml) into sterile whole and skim milks. Sensory defects were observed in the populations of 3 to 4 x 106 cfu/ml. This cell population was reached within 6 days at 7.2�C. Obvious physical and flavor defects were observed when populations of 5 to 20 x 106 cfu/ml in milk were held at 6 to 20C (Phillips and Griffiths, 1990). The optimum temperature for the production of degradative enzymes by microorganisms is usually lower than the optimum temperature needed for cell division. Even though the microbial population might remain below that normally associated with formation of microbial defects, microbially produced enzymes develop off-flavors when milk is held at refrigeration temperatures for extended periods.


Factors Affecting Shelf Life of Milk


The "shelf life" or "consumable life" is defined as the period between

processing/packaging, and when milk becomes unacceptable to consumers (Bishop and White, 1986). The product needs to remain acceptable beyond the last date of sale.

The shelf life of milk is influenced by the quality of raw milk, milk handling, milk processing, and storage temperatures, product processing procedures, and postpasteurization contamination. Post-pasteurization contamination is most detrimental for keeping quality of pasteurized milk (Stepaniak, 1991).

Bacterial growth is responsible for the spoilage of milk, and a level of log 7.5

cfu/ml represents the end of shelf life (Griffiths et al., 1984). Spoilage organisms cause










biochemical changes in the substrate. The amount of substrate utilized, and product formed is proportional to the number of cells present. The substrate is stated as organoleptically spoiled.

When raw milk has a high population of psychrotrophic bacteria (> 10' cfu/ml), the UHT products obtained from it usually have a reduced shelf life. Collins et al. (1993) investigated the influence of the growth of psychrotrophic bacteria in raw milk on the acceptability and resultant shelf life of ultra-high temperature (UHT) processed skim milk. The psychrotrophic counts in the raw milk were highly correlated with the extent of proteolysis (r = 0.95), however, not with the extent of lipolysis (r = 0.18) in the stored UHT milk. It was also found that storage time, and storage temperature had a greater influence on the sensory acceptance, and resultant shelf life of the UHT skim milk than the psychrotrophic count of the raw milk since correlations between the extent of proteolysis and bitterness scores in the stored UHT milk were high at 30C (r = 0.92) and at 40'C (r = 0.90), but not at 20C (r = -0.23).

Shelf life prediction could be made by performing microbial enumeration methods, based on preincubation at 12-21 OC with or without selective inhibitors, and the results could be obtained after 2-3 days. On the other hand, Moseley keeping quality test (7-day count) can be performed, but the result can be obtained after 7-9 days. Due to these time constraints, there is a need for a shelf life test which could provide results in a short period of time, be accurate, simple, and economical to perform (Bishop and White, 1986).











Rapid Methods for the Detection of Microorganisms



The dairy industry experiences increased demands on manufacturing and

distribution efficiencies. Food safety and quality management are crucial in production and distribution. In order to avoid the sale of contaminated products, expensive inventories are held at the production site while samples are tested for microbial contamination, which often takes more than 3 days. Since the products have a short shelf life, they are released before microbial results are available. Rapid detection of pathogens, spoilage microorganisms, and other microbial contaminants in dairy products is important to ensure the safety of consumers and quality of foods.

Recent developments make microbial detection and identification faster, more convenient, more sensitive, and more specific than conventional assays. The most frequently used rapid methods in industry are immunomagnetic separation (IMS), enzymelinked immunosorbent assays (ELISA), impedance (or conductance), bioluminescence, miniaturized methods, and other biochemical methods (FDA Bacteriological Analytical Manual, 1995). These methods may be used on their own or in combination.

Rapid methods have been developed either to replace the enrichment step which

requires a prolonged growth period with a concentration step (such as in immunomagnetic separation) or to replace the end-detection method which is colony development that requires a prolonged incubation period (such as in impedance microbiology and bioluminescence).











Immunomagnetic Separation (IMS)


A separation step is normally required in order to discriminate the target organism from other cells. Superparamagnetic particles are used in immunomagnetic separation. They exhibit magnetic properties in the presence of an external magnetic field. These are coated with antibodies against the target organism to isolate the organism selectively.

Food poisoning bacteria can be magnetically separated from foods depending on availability and specificity of appropriate antibodies, but labeled magnetic particles are not available commercially. V haemolyticus K serotype from food and from a patient (Tomoyasu, 1992) and Yersinia enterocolitica from spiked food and water samples (Kapperud and Vardund, 1995) were isolated using IMS. Encapsulated S. aureus can also be isolated from milk using IMS (Johne et al., 1989). Impedance Microbiology


Impedance microbiology detects microorganisms either directly due to the production of ions from metabolic end products or indirectly from carbon dioxide production. The direct method monitors changes in impedance of the growth medium. Microorganisms produce ionic end products such as organic acids and ammonium ions from the growth medium, and increase the conductivity of the medium (Silley and Forsythe, 1996).

In the indirect method a potassium hydroxide bridge (solidified in agar) is formed across the electrodes. The sample is separated from the potassium hydroxide bridge by a








22

headspace. Carbon dioxide accumulates in the headspace during microbial growth. This dissolves in the potassium hydroxide. Decrease in conductance occurs since the resultant potassium carbonate is less conductive. This method is applicable to a range of microorganisms including S. aureus, L. monocytogenes, E. faecalis, B. subtilis, E. coli, P. aeruginosa, A. hydrophila, and Salmonella serovars (Bolton, 1990). This indirect technique is also most appropriate to detect C. tyrobutyricum which is a spoilage organism of high-pH cheeses due to its transformation of acid to carbon dioxide, hydrogen, and butyric acid (Druggan et al., 1993). Impedance microbiology has also been used to monitor the stability of lactic acid bacteria starter cultures in the dairy industry (Lanzanova et al., 1993).


Enzyme Immunoassays and Latex Agglutination Tests


The enzyme immunoassay (EIA) or enzyme-linked immunosorbent assays

(ELISA) are other techniques used in food microbiology. ELISA is performed using monoclonal antibodies coated microtitre trays to capture the target antigen. The captured antigen is detected using another antibody which may be conjugated to an enzyme. The presence of the target antigen is visualized by the addition of a substrate (Forsythe and Hayes, 1998).

These methods have been developed for the detection of specific pathogens,

toxins, and enterotoxins, antibiotics, drugs, and pesticide residues. Some are designed to detect specific organisms such as Salmonella enteritidis or Listeria monocytogenes from food or environmental samples (Vasavada, 1997). The technique generally requires the










target organism to be more than 106 cfu/ml. The conventional pre-enrichment and even selective enrichment might be needed prior to testing.

Reverse phase latex agglutination (RPLA) can be used to detect Staphylococcal enterotoxins. Latex particles coated with specific antisera to the enterotoxins, each on separate particles, are used in the Oxoid RPLA kit (Denka Seiken Co. Ltd.). The sensitivity limit is about 0.5 ng enterotoxin/g food (Forsythe and Hayes, 1998).

A number of enzyme immunoassays such as Tecra ELISA (Tecra Diagnostics) and VIDAS ELISA (bioMrieux) are available for staphylococcal enterotoxin detection. The detection limit for Tecra ELISA is less than 0.5 gg toxin/ 100 g food, and requires 7 hrs to obtain the results. Even though the latex agglutination test RPLA has a similar detection limit, it requires 21 hrs to obtain the results (Forsythe and Hayes, 1998).

The Vidas (bioMrieux) system contains predispensed disposable reagent strips.

The target organism is captured in a solid phase receptacle coated with primary antibodies, and transferred to the appropriate reagents automatically. This system can be used to detect most major food poisoning organisms (Forsythe and Hayes, 1998).


Bioluminescent Systems


Bioluminescent systems measure the presence of adenosine triphosphate (ATP) in a sample using an enzyme system, luciferin-luciferinase, from fireflies. The amount of light generated by this enzymatic reaction can be measured in a suitable luminometer, and is directly related to the ATP extracted, and thus to the number of microbial cells from which it came (Stanley et al., 1989).










The ATP-bioluminescence method is used to determine raw milk quality (Van

Crombrugge et al., 1989). The detection level is approximately lx 10' cfu/ml of raw milk. Predicting the shelf life of pasteurized milk can also be achieved using ATPbioluminescence after preincubating the milk at 15*C for 25 hrs or at 21*C for 25 hrs in the presence of crystal violet, penicillin, and nisin to inhibit the growth of Gram positive organisms (Bautista et al., 1992).

Bioluminescence assays often take just a few minutes to accomplish, but the

disadvangates of this system are that the method measures all the microbes present in the sample, it is not presently possible to selectively measure ATP from one microbial species in the presence of many species (Stanley et al., 1989).


Miniaturized Methods


Conventional methods for isolation and characterization of microorganisms,

especially pathogens, entail the use of special enrichment and cultivation, selective and differential media, and a wide range of biochemical tests. Recently, miniaturized methods, diagnostic kits, and sophisticated instruments have been developed that allow the rapid identification of foodborne pathogens. These methods show an improvement over a conventional test as well as savings in time (Adams and Hope, 1989). The API test (bioMrieux Vitek), Enterotube (Roche Diagnostic), Micro-ID system (OrganonTeknika), and Minitek, and Crystal Sytem (BBL Microbiology Systems) are examples of miniaturized kits currently available for use in the food industry. These systems are








25
convenient, efficient, economical, and easy to use. They are also 90-95% accurate when compared to conventional methods (Vasavada, 1997).


Other Biochemical Methods


Other biochemical methods include those based on the presence of the

lipopolysaccharide of a Gram-negative bacteria (LAL, or limulus amoebocyte lysate) and specific enzymes, such as catalase. The LAL test is a simple, rapid, and sensitive method with applications for rapid screening of Gram-negative spoilage bacteria in milk, meats, fish, turkey, and food ingredients (Vasavada, 1997). This test does not measure the Gram-positive bacteria, so some techniques or devices are needed to relate the LALdetermined Gram-negative bacteria to "total" bacterial number that may be determined by the ratio that exists between Gram-negative and Gram-positive bacteria (Adams and Hope, 1989). This will increase the cost of the experiment, takes time, and relies on prediction.

In spite of all of these methods, new rapid detection techniques which provide

reliable and accurate results in a short period of time that will allow for the performance of effective, corrective measures are needed. In addition to these, a test should be simple and fairly economical. Use of electronic noses may possibly accomplish this. Since results will be given in less than 10 min right after the incubation is completed for a specified time, there is no need for a qualified person to run the experiment, the cost of the experiment will be minimized and small sample size, which is the most important point for some of the food samples, will be enough to perform the experiment.











Electronic Nose



In the food industry, flavors are important from the raw ingredients to the final product. There are two components of flavor perception: taste and aroma. Taste forms from the presence of nonvolatile compounds that interact with sensors in the mouth and on the tongue, and appears as the basic tastes of sweet, sour, salty, and bitter. Although taste is important, the flavor of a food cannot be defined by taste alone.

Many volatile compounds that are responsible for the aroma of a food play an

important role in flavor. These volatile compounds contribute to the nature of a food, its product identity, and to consumer preferences between brands. They are also responsible for the occurrence of off-flavors and taints, which arise because of biochemical or chemical changes, microbial growth or contamination (Hodgins, 1997).

There are three sensory systems in humans that are responsible for the sensation of flavor. These are gustation (sense of taste), olfaction (sense of smell), and the trigeminal sense (responsive to irritant chemical species) (Gardner and Bartlett, 1994).

The sensation of smell depends on the interaction of odor molecules with a group of specialized nerve cells. The odor molecules go into the nasal cavity and across the olfactory area or epithelium. These molecules dissolve in an aqueous mucous layer covering the olfactory receptor cells. The olfactory receptor cells located at surface of the olfactory hairs or cilia have receptor binding proteins that bind with the odorous compounds (Breer, 1994; Clapham, 1996; Pearce, 1997). The number of olfactory










receptor cells is large (about 100 million); however, the number of distinct types of binding proteins is small (about 1,000). The same protein has to be found in many different olfactory cells (Bartlett et al., 1997). The receptor cells amplify the signal, and transmit it to the brain by means of the axons. The brain compares it with previous knowledge, and tries to define the odor.

Humans can detect a minimum of 10,000 odors, but the number of identifiable odors is approximately 50 (Bartlett et al., 1997). The specific combination of complex mixtures of many odorous molecules having different concentrations culminates in the recognizable flavors or odors. Odor molecules are generally small (molecular weight 20300 Daltons) and polar. They can be detected by humans below 1 ppb. Gas chromatography linked with mass spectrometry (GCMS) is used to detect complex odors at low levels, but the sample must be separated into its individual components to identify each odor. GC/MS is expensive, and requires a technician for operation and interpretation of the results. Due to these constraints, sensory analysis has been used for a long time for odor evaluation. Sensory analysis has restrictions also: panelists may not be sensitive to some flavors, some raw materials may be difficult to assess using panelists, training needs to be performed before the analysis, sensory analysis can be expensive, and the panelists are subject to fatigueness (Bartlett et al., 1997; Hodgins and Simmonds, 1995).

A single molecule can have a distinct odor. However, most natural smells or

flavors are a complex mixture of chemical species, and contain hundreds of constituents (Dodd et al., 1992). True aroma is related to the complex interaction of all volatile










compounds within foods. For example, using a gas chromatograph with a sniffer port, none of the individual aromatic compounds present in cocoa smell like cocoa to humans; however, when all of these compounds combine, the overall aroma is that of cocoa (Hodgins and Simmonds, 1995; Hodgins, 1997).

Due to the limitations of GS/MS and sensory analyses, there has been a need to

develop an instrument that can mimic the human sense of smell, and provide rapid sensory information at low cost.


Electronic Nose Technology


In 1961, Moncrieff started to develop an instrument to detect odors. In 1965, several researchers published studies of the redox reactions of odorants at an electrode, modulation of electrical conductivity, and contact potential by odorants. The concept of an electronic nose as a chemical array sensor system for odor classification was presented for the first time by Persaud and Dodd in 1982. Today, the electronic nose has various synonyms such as artificial nose, mechanical nose, odor-sensing system, and sensor array system.

Gardner and Bartlett (1994) defined an electronic nose as an instrument comprised of an array of electronic chemical sensors with partial specificity, and an appropriate pattern recognition system capable of recognizing simple or complex odors. The main components of an electronic nose are sample handling mechanisms, an array of chemical sensors, signal preprocessing and conditioning, and pattern recognition techniques.








29
The electronic nose technology simulates the human olfactory process with fewer sensors, and a suitable software designed to analyze the responses from the sensors. Each sensor represents a group of olfactory receptors, and produces electrical signals in response to an odor, and this electrical signal is time-dependent. Although the specificity of each sensor may be low, the combination of several specificity classes results in a very wide range of information. Any noise or sensor drift may be reduced using signal preprocessing techniques. Finally, the use of pattern recognition in the electronic nose is equivalent to the classification and memorization of odors in the brain (Gardner and Barlett, 1999).

The first step in the sample handling is to obtain the vapor above a sample, and to transport it to the sensor array. Currently there are two methods, static and dynamic sampling. In static sampling the headspace above a liquid or solid is measured. The system consists of a sample vessel and a sensor head (compartment with the sensor array). The sensor array and vapor of the sample remain in separate sealed compartments. The sample headspace and sensor head are purged to eliminate any foreign odors using compressed air or any other inert gas for a certain period of time before the analysis. Once the sample has reached equilibrium, the door between the two compartments is opened and the analysis starts. When conducting polymer sensors are used, an internal DC power supply maintains a constant current through the sensors. The sensor resistance changes when the sensors get in contact with the headspace of the sample. The corresponding voltage change across the sensor is measured. The resulting analog signal is digitized, and sent to the computer. The change in conductivity of the sensors is








30
acquired for a given period of time. Once the analysis is completed, the separating door is closed, and the sensor head and sample vessel are purged to get ready for the next analysis.

The electronic nose works similar to a GC in the dynamic sampling procedure. A sample is placed in a closed container. Once headspace equilibration is reached, a sample of the headspace is obtained, and injected into the sampling port of the electronic nose. The sample is carried by an inert gas to the sensor array. A sensor changes its electrical properties, and sends a signal to the computer. The rest is similar to the static system.

The information provided by the sensor signal is maximized by the signal

preprocessing and conditioning of the analog response of the sensor. This is done by using signal conditioning circuits, potential dividers, constant voltage sources, and an analog-to-digital converter (Corcoran, 1993). Change in current or voltage is optimized for system sensitivity. Noise is reduced by modulating the sensor signal which is amplified to a suitable level. System noise is affected by variations in the electronic circuitry as well as connections between the sensors and the circuit (Hodgins, 1997). The signal conditioning digitizes the response of the sensors, and generates an output that is then analyzed with pattern recognition techniques to define the sample odors.


Sensor Technology


Different types of materials such as conducting polymers, metal oxides, lipid layers, phthalocyanins, and piezoelectric technologies are being used to manufacture sensors that are useful for odor detection. The types of sensors that are being used










commercially in electronic noses are semiconductor metal oxides, conducting polymers, quartz-resonator sensors, and surface acoustic wave sensors (Bartlett et al., 1997; Hodgins, 1997). Other types of sensors that have potential or have been used are biosensors, enzyme sensors, electrolytic sensors, platinum hot wire detectors, and fiberoptic gas sensors (Shunner, 1990). The selectivity and the sensitivity of the sensors are determined by choice of the catalytic surfaces (Gardner and Bartlett, 1999). Sensor technology is changing very rapidly and more sensitive, stable, and fast response sensors are being developed.

Conducting polymers

Conducting polymers used as sensor materials in electronic noses have unique

electrical properties that make them suitable for gas detection. A wide range of materials can be synthesized. They respond to a broad range of organic vapors, and they operate at room temperature. The main types are poly-pyrrole and poly-anilines. The volatile compounds change the electrical conductivity of the polymer. This change occurs rapidly and reversibly at room temperature (Gardner and Bartlett, 1999). The adsorbed odor molecules are believed to cause a swelling of the polymers and to interfere with charge transfer within the polymer (Corcoran, 1993). Conducting polymer sensors are nonspecific. Different compounds will interact with the polymer material. These sensors are small, and have low power consumption since they operate at room temperature. They have quite a good sensitivity, typically between 0.1 and 100 ppm (Bartlett et al., 1997). The sensor responses are also rapid with rapid recovery of the baseline when the








32

volatile compound is removed. Conducting polymers are sensitive to humidity, therefore, caution should be taken when analyzing samples with different water activities. Semiconductor metal oxide chemoresistive sensors

These types of sensors are developed from chemoresistive arrays of inorganic

semiconducting materials such as oxides, and catalytic metals. Two main types have been developed: thick film metal oxides, known as Taguchi sensors, and thin film, which are commonly used in commercial electronic noses. These sensors comprise a ceramic support tube containing a platinum heater coil. Tin-dioxide is coated on the outside of the ceramic support tube along with the catalytic metal additives such as palladium or platinum. As current passes through the coil, the metal oxide heats up. The reaction between the vapor and the metal oxide causes a change in electrical resistance at a fixed temperature. This resistance change can be measured, and related to odors being monitored. In metal oxides, chemisorbed oxygen [0-] reacts with the odorant [R] irreversibly to produce combined molecules [RO] and liberated conducting electrons [e-]. Electron mobility increases, and electrical conductivity of the material changes (Tan et al., 1995). These sensors operate between 300-550'C to avoid interference from water, and to aid rapid response and recovery times (Gardner and Bartlett, 1999). They are sensitive to combustible materials, such as alcohols, but are less sensitive at detecting sulfur- or nitrogen-based odors.

Surface acoustic wave devices

Surface acoustic wave sensors have been in research and development for 5 to 10 years (Hodgins, 1997). The principle of operation is that a surface wave is generated in a








33
material that absorbs the compounds of interest. The surface wave is normally generated using a quartz resonator, and the frequency of operation is usually between 100 MHz and I GHz. The frequency of operation depends on the sensitivity required by the system. When the sensor is not exposed to a vapor, it will have a certain resonant frequency. When the sensor is in contact with the volatiles, there will be a change of mass in the sensor material, and therefore, a change in the resonant frequency (Balantine and Wohltjen, 1989; Hodgins, 1997). This frequency change is the response or output from the sensor to the volatiles present in the sample analyzed. These sensors have higher sensitivity than conducting polymers (Hodgins, 1997). However, they are more selective, and a larger number of these sensors are needed to cover all vapors that are likely to occur in food products.

Fiber-optic gas sensors

These sensors rely on the light guiding properties of the optical fiber to carry the light from the light source to the chemically sensitive layer, and then to return the light to the sensor. The optical properties measured include the optical path length, luminescence, absorption, fluorescence, and reflectance.

These types of sensors have potential advantages in that individual fibers can be as small as 2 jim in diameter, and large bundles of fibers are available which permit an attractive approach to the fabrication of miniature sensor arrays; video technology can be used to measure the responses from an array; the measurements can be made remotely because the fiber allows transmission of light over long distances; and the devices are not subject to electrical interference (Gardner and Bartlett, 1999).










In recent studies some researchers have shown that fiber-optic gas sensor arrays

can be used to sense a range of organic vapors (Dickinson et al., 1996; White et al., 1996). A fluorescent dye, Nile Red, was used in these studies since the fluorescence spectrum, and intensity of Nile Red is strongly dependent on the local solvation environment of the dye molecule. The fluorescence emission from the dye changes in the presence of organic vapors that sorb into the polymer films. This change can be measured. At present, the sensitivity of these sensors is not high, and there is not enough information about the lifetime, reproducibility, or stability of fiber-optic sensors.


Applications of Electronic Nose


The electronic nose is used for monitoring and control of industrial processes,

diagnosis in the medical field, environmental control, and for control of food quality. It is possible to classify various liquors, perfumes, tobacco brands, beers, and many more with this device.

In the environmental applications, the electronic nose was used in monitoring

sewage related odors. Canonical correlation was used to compare the multivariate data generated by the electronic nose (Neotronics Olfactory Sensing Equipment) with sensory panel analysis. A linear relationship can be obtained between the electronic nose data and the corresponding threshold odor numbers within the similar groups of data from the experiment (Stuetz et al., 1998). In another application, the odors from pig and chicken slurry were evaluated by using a photoionization detector and electronic nose based on polypyrrole sensors, and it was concluded that electronic nose was better at discriminating








35
between different odors through the pattern of sensor responses (Hobbs et al., 1995). In this case rapid and portable devices for odor measurement may be useful since some of the major odor compounds were chemically unstable.

In the paper industry, an electronic nose containing four Taguchi gas sensors, and one infrared CO2 sensor was used to examine the odors from five cardboard papers from commercial manufacturers. It was shown that the olfactory quality of cardboard papers could be recognized using an electronic nose (Holmberg et al., 1995).

Two different electronic noses (Neotronics e-Nose 400OTM equipped with

conducting polymer sensors and Alpha M.O.S. Fox 300OTM equipped with metal oxide sensors) were used to distinguish between different concentrations of two final flavor mixtures, ten different methanol samples, and nine tobacco samples. Both instruments distinguished between different concentrations of the same flavor for two different flavors for a given day. Ten methanol samples were distinguished from each other for a given day based on sensory analysis. Nine tobacco sample results showed discrimination between each other that were classified as good, borderline, or bad using both instruments (Robie, 1997).


Applications in Microbial Detection


Due to the sensitivity of the electronic nose, it has great potential in

microbiological analysis. Gardner and Craven (1996) reported on the use of a 4-element metal oxide system to discriminate between 6 types of bacteria (Clostridium perfringens, Proteus, Haemophilus influenzae, Bacteroidesfragilis, Oxford Staph, Pseudomonas










aeruginosa). An array of 4 Taguchi series-8 gas sensors was used to sample the headspace of the 6 different bacterial samples and a control of blood agar. The data was used to train a neural network by the backpropagation method using over 1,000 iterations with a set of 42 odor vectors. The results were reasonably encouraging with 64.3% of test vectors being correctly classified.

Two bacteria types, Escherichia coli and Staphylococcus aureus, were also

examined. A backpropagation network was trained on the 90 odor vectors and in 87% of the test vectors the age of the bacteria were correctly classified. Another backpropagation network was trained on 180 vectors to predict the bacteria type, and in 91% of the 180 test vectors the type of bacteria of any age was correctly classified (Gardner and Craven, 1996). Electronic nose was successfully used to separate Penicillium species which produces different volatile metabolites (Olsson et al., 1995).

The electronic nose and neural network classifier were also used to detect and simultaneously identify pure plate cultures of a range of microorganisms. The overall classification rate for 12 different bacteria (E. Coli, Pseudomonas aeruginosa, Citrobacter freundii, Enterobacter aerogenes, Bacillus cereus, Klebsiella aerogenes, Staphylococcus aureus, Staphylococcus epidermis, Salmonella reading, Salmonella poona, Salmonella garinarium, Bacillus subtilis), and one pathogenic yeast (Candida albicans) was 93.4%. Three similar yeast cultures were also compared, and the correct classification rate was 96.3%. Principal component analysis gave good discrimination between water vapor and the test organisms (Gibson et al., 1997).










In medical applications the electronic nose is used to scan wounds to diagnose

infections and monitor the healing process. A hospital in Manchester, England has shown that the nose can detect early sings of wound infection, and can even distinguish between different infecting organisms. Electronic nose is also used in the analysis of breath of subjects. Researchers at the University of Pennsylvania detected pneumonia in hospitalized patients by collecting breath samples, and analyzing by electronic nose (Hanson, 1997).


Applications to Food Products


The applications of an electronic nose in the food industry are virtually unlimited. The instrument is currently used in the identification and classification of different food products based on their odor.

In most food companies raw materials are not checked frequently for aroma before processing. Final product is checked, and the batch is rejected if a taint or off-odor is present. This may cause difficulty in determining which supplier delivered the faulty material, and the faulty material may have cost a great deal to the manufacturer (Hodgins, 1997). However, if the company uses the electronic nose as a quality tool to check raw materials, this problem can be eliminated.

Electronic nose can also be used to monitor food odors during critical stages of

production to ensure that optimum processing conditions are being maintained, to monitor product deterioration during shelf life studies and during transport to retailers. It is concluded that this system is ideal for quick QC/QA checking (Hodgins, 1997).








38
In the dairy area, sensor arrays have been used to determine the role of fatty acids in the aroma profiles of Swiss cheese (Harper et al., 1996), and to differentiate enzyme modified cheese slurries (Jin and Harper, 1996). Korel et al. (1999) used the electronic nose to detect the odor differences in milk due to microbial load, storage time, and sensory panel perception. Sberveglieri et al. (1998) also obtained the selective discrimination of different heat treated commercially available milks (pasteurized, UHT, and sterilized) with an array of four semiconductor sensors. The sensors were formed by selected semiconductor thin film materials. The volatiles from the milk samples were collected by a dynamic headspace, and directly flowed into the test chamber where the sensor array was placed. The data was processed with simple principal component analysis. They stated that the results were promising for the industrial development of an electronic nose for the monitoring of the milk quality.

In the meat area, electronic nose was used for separation of ground raw and

cooked samples of pork-beef mixtures according to composition and freshness (Turhan et al., 1998), and used for estimation of quality of ground meat, stored under a polyethylene sheet, and gave a good possibility of predicting storage time (Winquist et al., 1993). Electronic nose was also used to monitor sausage fermentation following the changes in volatiles during the fermentation process, and to compare the electronic nose results with sensory panel results. From the sensor readings the fermentation time could be predicted, and sensory panel results were compared with the electronic nose sensor readings in the early stage of the process and on the final sausages (Ekl6v et al., 1998).










In the seafood area, applications of the electronic nose have been done in

differentiation of odors in shrimp stored on ice (Balaban and Luzuriaga, 1996), storage of tuna (Newman, 1998), and salmon (Luzuriaga and Balaban, 1999a) at different temperatures using conducting polymer sensors. Electronic nose was also used to monitor haddock and cod freshness (Olafsson et al., 1992), and to determine fish storage time (di Natale et al., 1996). The odor of decomposition in raw and cooked shrimp was evaluated based on electronic nose readings, sensory evaluation, and ammonia levels (Luzuriaga and Balaban, 1999b).

The fruits and vegetables area has also benefitted from the electronic nose. The electronic nose has been successfully employed for determination of harvest ripeness in cantaloupes (Benady et al., 1995), and quality assessment of packed blueberries (Simon et al., 1996). Volatiles of citrus juice (Hodgins, 1995; Hodgins and Simmonds, 1995), and fresh squeezed orange juice aroma volatiles (Bazemore et al., 1996) have been studied. Bazemore et al. (1997) reported that grapefruit juices of different cultivars were discriminated using metal oxide sensors. Maul et al. (1997) assessed the ability of an electronic nose to nondestructively identify and classify tomato fruit exposed to different harvesting and postharvest handling treatments. Werlein and Watkinson (1997) compared the sensory quality of conventionally processed carrots, green beans, and potatoes using metal oxide sensors and sensory panels.

In the area of grains and beans, B1rjesson et al. (1996) used an electronic nose to classify grains, and therefore reduce the inspector's exposure to grains that can be contaminated with aflatoxins. Jonsson et al. (1997) analyzed samples of oats, rye, and










barley with different odors and wheat with different levels of ergosterol, fungal, and bacterial loads. The odor of classes of good, mouldy, weakly, and strongly musty oats was predicted with a degree of accuracy using artificial neural networks. The percentage of moldy barley or rye grains in the mixtures of fresh grains was also indicated. It was also reported that there was a high degree of correlation between artificial neural network predictions, and measured ergosterol, fungal, and bacterial loads. Hofmann et al. (1997) used an electronic nose equipped with 4 metal oxide semiconducting sensors to follow the flavor generation during toasting of wheat bread, and to follow the roasting degree of toasted wheat bread slices. Some other work has been also done to discriminate among coffee cultivars, coffee from different origins, and coffee aromas (Aishima, 1991; Tan et al., 1995; Delaure et al., 1996).

In non-alcoholic and alcoholic drinks area there is a significant potential for the use of electronic noses. The flavor and aroma of beer and its raw materials were monitored using electronic nose technology (Pearce et al., 1993; Tomlinson, 1995; Tomlinson et al., 1995; Zimmermann and Leclercq, 1995). The aroma of pure hops and blends used in beer making were studied by Lucas and Castan (1995) and Weber and Poling (1996). Viaux and Robillard (1996) also used an electronic nose to help in the determination of the technical specifications of some additives and technological aids used in the sparkling wine process.

Since the electronic nose is rapid and objective in quantifying odors, there is a great potential in quality control applications in markets worldwide.










Current Shortcomings of the Electronic Nose


In electronic nose technology, the individual sensors need to be reproducible in their response to a given odor or chemical. It is important that a given sensor is reproducible throughout the lifetime of that sensor. However, this is an acute problem for electronic nose applications due to the difficulties of multivariate calibration, and the complex data sets that are required to train the pattern recognition software. The problem with the reproducibility of response over time arise from drift in the sensitivity of the sensors and the poisoning effects. Drift is a slow change in sensitivity, occurs with time, and can also be due to the effects of aging, slow morphological changes in the sensor material and some other long-term effects. Sensor poisoning arises when a sensor is exposed inadvertently to a material which irreversibly binds to, or interacts with, the sensing material leading to a reduction or even total loss of sensitivity. It is possible to avoid the poisoning problem by carefully selecting the right type of sensor for the particular application and by excluding poisons. Unfortunately, there is no chemical sensor which cannot be poisoned.

Another aspect of reproducibility is the response reproducibility between sensors of nominally the same type. This is important because if a sensor is poisoned then it can be replaced without recalibration and retraining the system. If the sensors are sufficiently reproducible in their response, then it becomes possible to train one sensor array, and to use the same training set for any nominally identical array in other instruments at different locations. If the transferability of data from one electronic nose to another (same










manufacturer) can be obtained, the use of electronic nose in quality control applications can spread to markets worldwide.

Changes in temperature, humidity and flow rate also play an important role in

sensitivity. These effects can be minimized by careful system design and sample handling, but this makes the electronic nose more expensive and complex (Ekl6v et al., 1998). However, it is necessary to overcome the effects of changes in temperature and humidity on the sensors' baseline and on the magnitude of their responses before portable instruments become available.


Objectives



The overall objectives of this study were to determine the ability of an electronic nose and sensory panels to detect the presence of pure cultures of P. fluorescens or B. coagulans in sterilized whole, reduced-fat, and fat-free milk samples stored at 1.70, 7.2, and 12.8�C; and, to determine the ability of an electronic nose to predict the shelf life of milk. The specific objectives were:

1. To inoculate whole, reduced-fat, and fat-free milk samples with known microbial

loads of P. fluorescens or B. coagulans, store the samples at 1.70, 7.20, and

12.8�C, and measure the electronic nose sensor response at 0, 3, 5, 7, and 10 days

in storage;

2. To conduct an odor sensory panel to determine whether the panelists can detect

the difference between a reference sample, and inoculated stored samples;










3. To determine and statistically correlate the relationship between electronic nose

readings and the microbial numbers for P. fluorescens or B. coagulans inoculated

into whole, reduced-fat, and fat-free milk, stored at 1.70, 7.2', and 12.8�C;

4. To determine and statistically correlate the relationship between electronic nose

readings and the sensory evaluations for whole, reduced-fat, and fat-free milk

inoculated with P. fluorescens or B. coagulans, stored at 1.70, 7.2', and 12.8�C;

5. To attempt to predict shelf life (based on microbial numbers) using electronic nose

readings from accelerated studies;

6. To inoculate whole milk samples with known microbial loads of P. fluorescens and

B. coagulans, store the samples at 1.7", 7.20, and 12.8�C, and measure the

electronic nose sensor response at 0, 3, 5, 7, and 10 days in storage;

7. To conduct an odor sensory panel to determine whether the panelists can detect

the difference between a reference sample and stored samples inoculated with both

microorganisms;

8. To determine and statistically correlate the relationship between electronic nose

readings and the sensory evaluations for whole milk inoculated with both

microorganisms, stored at 1.7', 7.20, and 12.80C.













CHAPTER 3
MATERIALS AND METHODS



Milk Sampling. Inoculation. and Analysis



Milk Samples


Whole, reduced-fat (2% milkfat) and fat-free milk (Parmalat, Teaneck, NJ) was purchased from a local supermarket in Gainesville, FL. The samples were aseptically packaged in 946 ml packages and the sell-by-dates and the lot numbers for each set of samples were the same. The packages were kept at 1.7�C until the experiments.

The sampling procedures were performed under a sterile laminar flow hood (Nuaire Biological Safety Cabinets, Plymouth, MN). The front surface of the milk package was swabbed thoroughly with 70% alcohol. A sterile, single use PrecisionGlide needle (B-D� 16 G 1 Becton Dickinson and Company, Cat. No. 305198, Franklin Lakes, NJ) of a sterile, single use 60 ml syringe ( B-D� Becton Dickinson and Company, Cat. No. 30966301, Franklin Lakes, NJ) was inserted through the previously swabbed package wall. A 50-ml milk sample was taken from the package, placed into a 300 ml sterile jar and the lid was closed.










Activation of Microorganisms


The milk samples were inoculated with Pseudomonasfluorescens (ATCC 948) (American Type Culture Collection, Manassas, VA) and/or Bacillus coagulans (ATCC 7050). P. fluorescens represents the Gram-negative spoilage organisms in milk and B. coagulans the Gram-positive microflora. The freeze-dried culture of P. fluorescens was activated by adding some of the culture into 10 ml nutrient broth (Difco Laboratories, Cat. No. 0003-01-6, Detroit, MI) tube and incubated at 28�C for 48 hrs. The nutrient broth with the activated organism was transferred to 150 ml nutrient broth flask and incubated at 28�C for 48 hrs. The freeze-dried culture of B. coagulans was activated by adding some of the culture into 10 ml tryptone soy broth (Oxoid Ltd., Cat. No. CM129, Hampshire, England) and incubated at 21 �C for 48 hrs. The tryptone soy broth with the activated organism was transferred to 150 ml tryptone soy broth flask and incubated at 21�C for 48 hrs.

The activated organisms were centrifuged by using a Sorvall� RC-5B refrigerated superspeed centrifuge (Du Pont Comp., Duluth, GA) at 4342 x g for 10 min for P. fluorescens and at 12061 x g for 10 min for B. coagulans. The supernatants were poured off and the cultures were resuspended with filter sterilized phosphate buffer saline (Harrigan, 1998). They were centrifuged at 4342 x g and 12061 x g for 10 min for P. fluorescens and B. coagulans, respectively. This procedure was repeated twice. Twenty ml of Butterfield's buffer solution (International BioProducts Inc., Redmond, WA) was added to both centrifuge tubes. Eleven ml of the aliquots from each centrifuge were








46
transferred to 99 ml Butterfield's buffer dilution bottle. This dilution was used for all the inoculations of RB. coagulans. One more dilution was performed for P. fluorescens, 1 ml from the previous dilution bottle was transferred to 99 ml Butterfield's buffer dilution bottle, and this dilution was used for all the inoculations. Inoculation of Microorganisms and Sample Treatments


The jars filled with 50 ml of milk were inoculated with I ml of the specific dilution for each microorganism and the control samples were inoculated with 1 ml of Butterfield's buffer solution. The samples were stored at 1. 7', 7.2', and 12.8�C for up to 10 days. The samples were evaluated at days 0, 3, 5, 7, and 10.

The samples for the accelerated study were prepared similarly, but the samples were stored at 1. 7�C for up to 8 days. At day 0, five jars for each treatment were incubated at 28�C for 24 hrs and these samples were analyzed at day 1. At days 3, 5, and 7, five jars for each treatment were taken out from the 1. 7�C refrigerator and incubated at 28�C for 24 h and they were analyzed at days 4, 6, and 8. The study was repeated twice for each type of milk using milk from the same lot.

Whole milk (Parmalat�, Teaneck, NJ) was used for the combination study. In this study, the milk samples were inoculated with I ml of each P. fluorescens and B. coagulans aliquots. The control samples were inoculated with 1 ml of Butterfield's buffer solution. The storage conditions and the analyses were the same as in the other studies.










Electronic Nose Measurements


An electronic nose (e-NOSE 4000 model, EEV Inc., Amsford, NJ) equipped with twelve conducting polymer sensors (sensor types: 483, 478, 464, 463, 462, 461, 460, 459, 458, 401, 298, 297) was used to quantify the sensor responses to differences in odor of milk samples inoculated with different microorganisms. The electronic nose measurements were done immediately (day 0), and at days 1, 3, 4, 5, 6, 7, 8, and 10. Five replicates were analyzed by the electronic nose for each treatment. Each replicate was kept at room temperature for 30 min prior to analysis in order to let the milk temperature equilibrate to room temperature (22.50 to 23.5'C). Five replicates were analyzed on each day for each storage temperature. The replicates were flushed with compressed air for about 10 sec prior to the electronic nose analysis. The milk jar was placed in the glass sampling vessel of the electronic nose. One day before the experiment started, the electronic nose was calibrated with 75% v/v propylene glycol solution (100% solution from Fisher Scientific, No. P-355-20, Fair Lawn, NJ), following the manufacturer's recommendation. Every day before the experiments the electronic nose was turned on and compressed air (CGA Grade D, Strate Welding Supply Inc., Jacksonville, FL) was passed through the sensors for at least 30 min. The vessel was purged with compressed air for 2 min to eliminate any foreign odor present in the vessel from the environment for each replicate, and then the sensor head was purged for 4 min with compressed air. During these 4 min, the sample volatiles were equilibrating in the headspace of the vessel. Sensor response data was acquired for 4 min. Total analysis time for each milk sample took 10










min. Readings at 4 min exposure of the sensors to the milk samples were used for data analysis. At the end of the day, electronic nose sensors were cleaned with compressed air for at least 30 min. Electronic nose raw data can be obtained from Dr. Murat Balaban, Food Science and Human Nutrition Department at the University of Florida (filenames:"\\thesis\e-nose data.txt").

Microbial Analysis


Microbial counts were performed for each treatment at each sampling day.

Dilutions were made using pre-filled sterile disposable Butterfield's buffer dilution bottles. Total aerobic count method using Petrifilm technique was used to enumerate pure P. fluorescens since it was the only microflora present. Inoculated aerobic plate count Petrifilms (3M Company, St. Paul, MN) were incubated at 28'C for 48 hrs.

Bishop and Juan (1988) used agar pour plates and Petrifilm dry medium culture plates to enumerate bacteria after preliminary incubation of milk samples at 13'C or 21C for 18 hrs. Results showed that the Petrifilm technique was not significantly different from agar pour plate methods. Ginn et al. (1984) reported that total aerobic bacteria data comparisons obtained by the Standard Plate Count and Petrifilm method produced a correlation of 0.971 and 0.946, respectively. Since the results of the Petrifilm technique and agar pour plate methods are similar, the Petrifilm technique may be preferred because they require no preparation, and are ready to use.

Pure B. coagulans cultures were enumerated using standard plate count agar (APHA) (Oxoid Ltd., Cat. No. CM463, Hampshire, England) by spread plate method.










They were incubated at 320C for 48 hrs. Colonies were counted and reported as logo cfu/ml.

Microbial counts for inoculated samples of milk were performed as follows: 6 ml of milk from one milk jar and 5 ml milk from another jar of the same treatment were taken and transferred to the same dilution bottle to assure homogeneity. From predetermined dilutions, I ml was taken and plated in the Petri films for P. fluorescens. They were incubated at 28�C for 48 hrs. For B. coagulans, 0.1 ml from predetermined dilutions was spread plated onto standard plate count agar and incubated at 32�C for 48 hrs. A I ml sample was taken from control samples, plated in Petri films and incubated at 32'C for 48 hrs. Microbial counts were performed on the control samples to see whether there was any contamination occurred during filling the jars, and any contaminated sets were discarded. All sets used for data analysis were free of contamination. Colonies were counted and reported as log,0 cfu/ml of milk.

For the combination study, 6ml and 5 ml milk samples from two replicates were

transferred to the dilution bottles. Serial dilutions were performed and 0.1 ml were spread plated onto crystal violet tetrazolium agar (Marshall, 1993) and thermoacidurans agar (Difco Laboratories, Cat. No. 0303-17-5, Detroit, MI) and incubated at 28'C and 32'C for 48 hrs, respectively. Crystal violet tetrazolium agar was used to enumerate P. fluorescens, and thermoacidurans agar was used to enumerate B. coagulans. For control samples, 1 ml milk sample was plated in Petri films and incubated at 320C for 48 hrs. Colonies were counted and reported as logo cfu/ml of milk.










Moisture Content Measurements


Moisture content was measured in triplicate for each type of milk and repeated

twice using the oven method (Bradley et al., 1993). This analysis was performed to have information on the composition of the milk in order to better describe the sample. A sample of approximately 3 g was placed in an aluminum weighing dish (50 mm diameter, Cat. No. 08-732, Fisher Scientific, Fair Lawn, NJ). The sample was placed in an oven at 103'C for 24 hrs. Moisture content was reported as percent wet basis. Fat Content Measurements


Fat content was measured according to Chilliard et al. (1991). This analysis was done to confirm that each type of milk had the same fat percentages as it was stated on the labels. Two ml of ethanol, 0.5 nl of 12.1 N HCI, and 25 ml of hexane were added to 5 ml of milk. The mixture was shaken and centrifuged at 500 x g for 5 min. The top organic layer was transferred to a clean tube, and the aqueous phase was reextracted with 25 ml of hexane. Organic layers were placed in the same tube and water was removed with sodium sulfate, and the solvent evaporated under a stream of nitrogen. The total fat percentage was determined gravimetrically.

pH Measurements


A 50 ml sample of milk in the jar was placed on a stirrer plate. The pH electrode (ROSS pH electrode, Model 81-02, Orion Research Inc., Beverly, MA) was connected to an Expanded Ion Analyzer, and was calibrated every day with pH 4.00 and 7.00 standards










(Buffer solution pH 4.00, SBIOI-500 and pH 7.00, SB107-500, Fisher Scientific, Fair Lawn, NJ). Measurements were performed for each treatment at every time interval and in duplicate.

Sensojy Evaluation


The odor of milk samples was evaluated by a 10-member untrained sensory panel consisting of students, 25-32 years of age, from the Food Science and Human Nutrition Department at the University of Florida. A difference-from-control test was performed at days 0, 3, 5, 7, and 10. Panelists were asked to smell milk samples and detect if there was any difference in odor among the treated samples and the reference sample. The reference sample was fresh Parmalat� milk from the same lot used for the experiments. Panelists rated the differences in a 0 to 10 scale (0 = no difference and 10 = very different). Samples were randomized and a hidden control was included in the test. The replicate number 1 was always taken out of the refrigerator 30 min before the sensory analysis at each day. All panelists smelled the same samples. Sensory tests were carried out on both experiments and in the combination study.

Data Analysis


Electronic nose sensor readings were analyzed in Statistica for Windows ('98

edition, StatSoft Inc., Tulsa, OK) using discriminant function analysis (DFA) as reported by other researchers (Corcoran, 1993; Gardner and Hines, 1997; Gardner and Bartlett, 1992). Microbial counts and sensory data were used as grouping variables and 12 electronic nose sensor outputs were used as independent variables. DFA was used to








52
develop predictive models for classification of samples based on grouping variables. The 12 sensor outputs were reduced to 2 discriminant functions. These functions were used to map the data in two dimensional plots and observe separation between groups. Correct classification rates and the coefficients for each function were calculated using Statistica.

Data of the compositional analysis, microbiological analysis, pH measurements and sensory scores for each treatment were subjected to analysis of variance with the general linear model (SAS, 1998). Least square means were obtained and separated using the least significant difference test procedures when significant (p < 0.01) F values were obtained.













CHAPTER 4
RESULTS AND DISCUSSION



Milk Sampling, Inoculation, and Analysis



Moisture and Fat Content Measurements


The moisture contents of whole, reduced-fat (2% milkfat), and fat-free milk

samples were 89.53% � 0.11, 90.39%/0 � 0.06, and 92.26% � 0.22, respectively. The fat contents of whole, reduced-fat, and fat-free milk samples were 3.20% � 0.07, 1.98% �

0.01, and 0.11% � 0.03, respectively. As the moisture content increased, the fat content decreased.

Microbial Analysis


Microbial counts for P. fluorescens and B. coagulans during 10 days of storage

for each type and two experiments of milk samples are given in Tables 4.1 and Appendices A for whole, reduced-fat and fat-free milk. Analysis of variance with the general linear model procedures was performed for each inoculated microorganism and for each experiment of each type of milk to see if there was any significant difference due to the storage temperature and time. It was expected that the storage temperature and time would have an effect on the microbial load. It was found that the storage temperature,









Table 4-1.


Microbial load of all types of milk samples inoculated with Pseudomonas
fluresensor Bacillus ca'~n


Microbial Load (log0 cfu/ml of milk)* Storage Time (Days) Experiment 1 Experiment 2
1.70C 7.20C 12.80C 1.70C 7.20C 12.80C
0 4,208 4.20 4.20a 3.602 3.60r 3.60' Inoculated - T
nu.57 4.63b 8.519 5.28r 5.08c 7.769 with 5 64~ T
wih 5 6.43r 7.28 d 8 .95h 4.85b 6.15d 9.08'
Pseudomona s 7 6.48c 7.62" 9.23' 6.430 7.769 9.234 fluorescens 10 8.30f 9.04h 9.28' 6.78 8.81 9.34' Whole Milk
Inoculated 5 4.95 4.95' 5.90" 5.90h 5.90"
with 2.90z 3.42 3.26b- 4.6od 4.48b 4.8T 5 2.42-b 3.43- 2.88 ; 74.26 5.08g 5. 23 Bacilus 7 1.95' 2.62a 2.11a 5.60X 4.78'76 5.15fi
coagulans 10 2.04' 2 2.88 7- 4' 3.40' 4.43b 4.81
0 3.23' 3.23' 3.23a 3.43b 3.43 b 343T
Inoculated ~-~- 59 .4
In cltd 3 3.88 b 5.79c 8.43' 2.54a 3.54 b 7.459
with
5 5.59- 7.71e 9.26h 3.87' 7.30f 9.11i Pudomonas - _ _7
7 4.70Td 8.42 9.28h T.18' 8.48hr 9.26& Reduced-fat fluorescens 10 7.53e 8.919 9.42h 6.40 9.04' 9.53k
Milk Inoculated 0 5.87h 5.87h 5.87h 5.11' 5.11g 5.11' 3 5.26'f 4.910 5.409 3.04de 3.4T2f 2.60' with 8 - 1TW 5 4.42 4.48 5.15f 1.30W 2.40' 3.63f Bacillus 4
7 3.97' j.88F- 4.89e <1.0" 2.28 2.40F coagulans 10 T74 3.36" 4.770 <1.0a 2.00b 0 3.99' 3.99' 3.99' 4.32b 4.37- T32b Inoculated 5.23b 6.11c 7.00 4.00' 5'51 7.77'
with
5 6.36d 7.4 3 9.20'm 4.71c 6.91e 9.11 7udomona 7 6.720 8.00i 9.11V 4.26 8.08h' 9.20'
Fat-free fluorescens 10 7.64' __891k 9.34- 7.26 8.98' 9.607Milk 0 5.439 5.349 5.34' 5.0 5.70 5.70
Inoculated 3 3.04e 2.88bc 3.78' 3.38 d 4.57' 4.28f
with 5 2.54' 3.00' 3.64f 2.40r 3.950 4.08f
Bacillus T T 7 - 3.850 399c coagulans 10 2.40' 2.7-0" 1.00' 3.38d 3.890 � Average of two readings
a-r : Superscripts within an experiment for each microorganism denote significant
difference at the p<0.01. Means were separated using LSD.










time and the interactions were significant (p<0.01) for each microorganism and for each experiment of each type of milk.

The microbial loads for P. fluorescens were plotted versus storage time and are shown in Figures 4.1- 4.3 for each type of milk and for each experiment. In general P. fluorescens counts increased gradually during storage at 1.7�C, but they increased rapidly at 12.8�C. The growth of P. fluorescens at 7.2'C showed a similar trend with the ones at 12.8�C except for whole milk experiment 2 and reduced-fat milk experiment 2. In whole milk experiment 1 and reduced-fat milk experiment 2, there were lag phases until day 3 and then exponential growth occurred. This could be due to the differences of inoculation loads. Regardless of the initial inoculation levels or the storage temperature, P. fluorescens counts exceeded 106 cfu/ml at day 10 for all types of milk. Since microbial counts in the excess of 1 x 106 cfu/ml are required to produce adverse defects on product quality (Cousin, 1982), formation of off-odors was expected during storage and these were detected by electronic nose.

The microbial loads for B. coagulans were plotted versus storage time and are shown in Figures 4.4- 4.6 for each type of milk and for each experiment. The initial inoculation levels were between 105 and 106 cfu/ml. However, this microorganism was unable to grow at 1.7', 7.2� and 12.8�C. The microbial counts were generally lower for the milk samples stored at 1.70C than the samples stored at 7.2' and 12.8�C except for reduced-fat milk experiment 1. Bacillus spp. were stated as thermoduric psychrotrophs in the literature and these may be variants of mesophilic organisms that have adapted to grow at lower temperatures (Grosskopf and Harper, 1974). B. coagulans used in this study













Experiment I 10.0000

9.0000 ,, '-" . "

8.00007.0000

6.0000 �
5.0000 "



4.0000 3.0000 2.0000
0 3 5 7 10 Storage Time (Days)




Experiment 2 10.0000.


9.0000 8.0000 7.0000 6.0000 5.0000 4.0000 3.0000 2.0000


0 3 5 7
Storage Time (Days)


Storage Temperatures


- 1.70C
-a- 7.20C
--o*- 12.8oC


Storage Temperatures


-I- 1.7C
-a- 7.20C 10 -o- 12.80C


Average microbial load of P. fluorescens for whole milk, both experiments stored at different temperatures over time. Error bars signify �1 standard deviation.


.o. ... . -o '


Figure 4-1.













Experiment 1


10.0000 9.0000 8.0000 7.0000 6.0000 5.0000

4.0000 3.0000


-p


* a.


* I
a.
*1


0 3 5 7 10
Storage Time (Days)




Experiment 2 10.0000

9.0000. .0

8.0000 " "

7.0000
6.0000 0

5.00(00 ,�


4.0000 .

3.0000

2000(


3 5
Storage Time (Days)


10


Storage Temperatures


--- 1.70C
-a- 7.20C
--o-- 12.80C



















Storage Temperatures


-a- 1.70C
-a- 7.20C � 12.80C


Average microbial load of P. fluorescens for reduced-fat milk, both experiments stored at different temperatures over time. Error bars signify �1 standard deviation.


Figure 4-2.













Experiment 1 0000.


9.0000 8.0000 7.0000

6.0000 5.0000 4.0000 3.0000


2.0000


0 3 5 7


Storage Time (Days)



Experiment 2 in A AAA


9.0000


8.0000 7.0000 6.0000 5.0000

4.0000 3.0000


2.


...._...-"


aw


* a


* a
a
* a


Storage Temperatures


I -D- 1.70C
-A- 7.20C 10 --o-- 12.80C


Storage Temperatures


-- 1.7oC
-a- 7.20C 10 --o- 12.80C


Storage Time (Days)


Average microbial load of P. fluorescens for fat-free milk, both experiments stored at different temperatures over time. Error bars signify �1 standard deviation.


a

a



*


Figure 4-3.


.0"0 1


u













Experiment 1
7 O0.


6.0000 5.0000

4.0000 3.0000 2.0000 1.0000 0.0000


Storage Time (Days)



Experiment 2 Aon'q


3 5
Storage Time (Days)


Storage Temperatures


EM 1.70C EM 7.20C Ei 12.80C


Average microbial load for B. coagulans for whole milk, both experiments stored at different temperatures over time. Error bars signify � 1 standard deviation.


Storage Temperatures


1.70C 7.20C 12.80C


10


6.0000 5.0000
U

.4.0000 3.0000 S2.0000

1.0000


0.0000'


Figure 4-4.













Experiment I


5
Storage Time (Days)


7.0000 6.0000 5.0000 4.0000 3.0000 2.0000 1.0000 0.0000


Experiment 2


3 5
Storage Time (Days)


Storage Temperatures


In E 1.70C
E 7.20C
10 M 12.80C


Average microbial load for B. coagulans for reduced-fat milk, both experiments stored at different temperatures over time. Error bars signify �1 standard deviation.


Storage

Temperatures


M 1.7-C EM 7.20C 10 MM 12.80C


7.0000 6.0000 5.0000 4.0000 3.0000 2.0000 1.0000 0.0000


Figure 4-5.













Experiment I


0 3 5 7
Storage Time (Days)


7.0000 6.0000 5.0000 4.0000 3.0000 2.0000 1.0000 0.0000


EM 1.70C M 7.20C 10 M 12.80C


Experiment 2


0 3


5
Storage Time (Days)


Storage

Temperatures


E 1.70C E 7.20C 10 M 12.80C


Figure 4-6. Average microbial load for B. coagulans for fat-free milk, both
experiments stored at different temperatures over time. Error bars signify
�1 standard deviation.


Storage Temperatures


7.UUU

6.0000 5.0000 4.0000 3.0000 2.0000 1.0000 0.0000L








62
(ATCC 7050) which was isolated from an evaporated milk most probably lost its ability to grow at refrigeration temperatures. Attempts were made to adapt this microorganism to grow at low temperatures. Initially, it was activated in tryptone soy broth at 35C and was transferred to a new sterile tryptone soy broth and incubated at 21C. Milk samples were inoculated with this and the organism was expected to grow at lower temperatures, but it did not proliferate at refrigeration temperatures. Accelerated Study

The microbial loads for all milk types, both experiments and two microorganisms for the accelerated study are given in Table 4.2. The microbial counts for P. fluorescens increased approximately to 10 cfu/ml for all milk types after 24 hrs of incubation at 28'C. The initial inoculation level had an effect on the microbial growth. Whole milk experiment

1 was inoculated with the microbial load of 4.20 log0 cfu/ml and the microbial load increased to 8.32 log0 cfu/ml after 24 hrs at 280C. On the other hand, whole milk experiment 2 was inoculated with 3.60 log0 cfu/ml and the microbial growth reached to

7.96 log0 cfiu/ml. This was observed in all types and experiments of milk.

The P. fluorescens counts increased to 108-109 cfu/ml for day 4, 6, and 8. Due to the possibility of insufficient substrate, the microbial loads did not increase more than 109 cfu/ml throughout the accelerated study regardless of the microbial load of the milk sample that was incubated at each time period. The type of milk did not have a significant effect on the microbial loads for this study.

The microbial counts for B. coagulans increased for all milk types after each 24 hrs of incubation at 28C. At day 0, the initial microbial load was 4.20 logo cfu/ml for










Table 4-2. Microbial load of all types of milk inoculated with Pseudomonas
fluorescens or Bacillus coagulans in accelerated stiidy


* : Average of two readings


Microbial Load (logo cfu/ml of milk)* Analyses Experiment 1 Experiment 2
(Days) (Inoculated Microorganisms) (Inoculated Microorganisms)

P. fluorescens B. coagulans P. fluorescens B. coagulans
1 8.32 4.36 7.96 6.00 WholeMlk 4 8.34 3.48 8.72 5.15
6 8.72 2.40 8.70 5.48 8 9.04 2.26 8.42 4.49 1 7.84 7.04 7.94 6.67 Reduced-fat 4 8.77 5.87 8.57 6.15
Milk 6 8.86 5.86 8.61 4.18
8 8.91 5.04 8.80 4.92 1 8.42 7.74 8.62 6.89 Fat-free 4 8.56 7.58 8.46 5.00 Milk 6 9.04 7.42 8.42 5.00
8 8.82 6.61 8.53 3.00










whole milk experiment 1 and it increased to 4.36 log0 cfu/ml after incubation. Even though only a slight increase in microbial number was observed in this case, in other cases, higher increases in the microbial loads were observed. For example at day 7 the microbial load for fat-free milk experiment 2 increased I log cycle after the incubation at 28�C. This also proved that this organism has not been adapted to grow at low temperatures.

On the other hand, it was observed that B. coagulans counts for all types and experiments of milk samples were decreased over time. The milk samples were kept at 1.7�C before the accelerated study analysis, and this microorganism could not survive at low refrigeration temperatures as mentioned before. The types of milk had a significant effect in this study, but this could be observed because of not having the same initial inoculation levels at day 0 for each type and experiment of milk samples. Combination Study

In the combination study, P. fluorescens and B. coagulans counts are given in Table 4-3. The storage time and temperatures had significant effects on the microbial loads for both microorganisms. P. fluorescens counts increased rapidly for the samples stored at 12.8�C (Figure 4-7) and they increased gradually for the samples stored at 7.2�C until day 7, and had a rapid increase from day 7 to day 10. The microbial loads for P. fluorescens increased to 10' cfu/ml for the samples stored at 7.20 and 12.8�C at day 10. On the other hand, P. fluorescens in the whole milk samples stored at 1. 7�C multiplied, but did not show any increase in counts compared to the samples stored at the higher temperatures. This could occur due to the presence of B. coagulans together with P. fluorescens in the samples.










Table 4-3. Microbial load of whole milk samples inoculated with Pseudomonas
fluorescens and Bacillus coagulans in combination study.
Microbial Load
Storage (loglo cfu/ml of milk)* Time Pseudomonasfluorescens Bacillus coagulans (Days) Storage Temperature Storage Temperature
1.70C 7.20C 12.80C 1.70C 7.20C 12.80C
0 3.82b 3.82b 3.82b 5.60 5.60h 5.60h 3 3.082 5.43c 7.45f 4.49c 5.15fg 5.51b 5 2.90a 6.11d 8.779 4.34c 5.084 5.30 7 3.89b 6.53e 9.00g, 4.04b 4.94d 4.95d 10 3.70V 8.899 9.23h 3.682 4.86d 4.98d f
* : Average of two readings
Superscripts within each microorganism denote significant difference at the p<0.01.
Means were separated using LSD.



The microbial loads for B. coagulans for the samples stored at each storage

temperature decreased during the 10-day storage. However, for the samples stored at

7.20 and 12.80C, the decrease in B. coagulans counts was less compared to the decrease in the counts for the samples stored at 1.7'C (Table 4-3 and Figure 4-7). This was due to the lack of adaptation of B. coagulans at low refrigeration temperatures. Accelerated Study of the Combination Study

The microbial loads for P. fluorescens and B. coagulans for the accelerated study of the whole milk samples in the combination study are given in Table 4-4. After each incubation at 280C, P. fluorescens counts were increased to 10 cfu/ml. B. coagulans counts increased during incubation compared to the microbial load of the sample before it was incubated. However, B. coagulans counts decreased over time.













P. fluorescens in Combination Study


10.0000

9.0000 8.0000 7.0000


o �.� . =.6 � .-g,-


6.0000 5.0000 4.00001 3.0000


hAAAA.


0 3 5 7
Storage Time (Days)


Storage Temperatures


- -7.2oC
1--- 1 2701


10 *0-12.80C


B. coagulans in Combination Study


3 5 Storage Time (Days)


Storage

Temperatures


E 1.70C EM 7.2-C 10 M 12.8-C


Average microbial load for P. fluorescens and B. coagulans for combination study of whole milk, stored at different temperatures over time. Error bars signify �1 standard deviation.


7.0000 6.0000 5.0000

4.0000 3.0000

2.0000 1.0000 0.0000


Figure 4-7.


OM&,a %J'










Table 4-4. Microbial load of whole milk samples inoculated with Pseudomonas
fluorecen and Raeilu.� ienrd-mv in rr~lor~tPA tih,


Microbial Load
Storage Time (loglo cfu/ml of milk)*
Pseudomonasfluorescens Bacillus coagulans

1 8.34 6.69 4 7.95 5.70 6 8.49 5.04 8 8.59 3.85


* : Average of two readings



pH Measurements


The pH of all whole, reduced-fat and fat-free milk samples stored at different

temperatures during 10 days of storage are given in Tables 4-5, 4-6, and 4-7, respectively (Appendix B). Changes in the pH of whole, reduced-fat and fat-free milk during storage did not follow any specific trend. The pH of each type of milk and the experiments of each milk was significantly different (p<0.01). Therefore data for each milk type and experiments of these were analyzed independently. Overall, pH of the samples treated with P. fluorescens decreased during storage. This decrease was higher for the samples stored at 12.8�C than for the samples stored at 1.70 and 7.2'C, except for the fat-free experiment 2 samples stored at 7.2' and 12.8'C, where the pH increased. However, changes in pH were not as large as microbial loads. In most cases, there were minimum pH changes for control and samples inoculated with B. coagulans.










Table 4-5. pH of whole milk inoculated with Pseudomonasfluorescens or Bacillus
coagulas

Storage Time (Days) 1.70C 7.20C 12.80C C Pf Bc C Pf Bc C Pf Bc 0 Avg.* 6.73 6.71 6.71 6.73 6.71 6.71 6.73 6.71 6.71
St. dev. 0.06 0.01 0.01 0.06 0.01 0.01 0.06 0.01 0.01
3 Avg. 6.68 6.74 6.64 6.63 6.59 6.66 6.66 6.63 6.67
St. dev. 0.01 0.13 0.04 0.01 0.08 0.01 0.00 0.02 0.01
Exp. 1 5 Avg. 6.81F 6.82k 6.811 6.81b" 6.8 8 6.65' 6.78k St. dev. 0.01 0.01 0.04 0.01 0.01 0.05 0.01 0.02 0.00
7 Avg. 6.82' 6.75c 6.72b 6.80- 6.8W 6.80' 6.81' 6.62 6.80'
St. dev. 0.01 0.00 0.01 0.01 0.02 0.01 0.01 0.04 0.02
10 Avg. 6.76b 6.78 674b 676 6.62a 6.79 .78 6.57' -67
St. dev. 0.03 0.02 0.01 0.01 0.05 0.01 0.01 0.04 0.01
0 Avg. 6.76 6.76 6.74 6.76 6.76 6.74 6.76 6.76 6.74
St. dev. 0.00 0.06 0.01 0.00 0.06 0.01 0.00 0.06 0.01
3 Avg. 6.76 6.72 6.76 6.73 6.65 6.75 6.73 6.73 6.74
St. dev. 0.01 0.01 0.01 0.02 0.11 0.03 0.01 0.01 0.02
Exp. 2 5 Avg. 6.72 6.73 6.71 6.72 6.72 6.70 6.73 6.66 6.65
Exp St. dev. 0.07 0.08 0.02 0.01 0.02 0.00 0.01 0.01 0.04
7 Avg. 6.77' 6.76' 6.69b 6.75r 6.76' 6. 73 6.77' 6.63a 6.69b
St. dev. 0.01 0.01 0.00 0.00 0.00 0.02 0.02 0.03 0.01
10 Avg. 6.74b 6.73 6.72b 6.71b 16.72 6.73' 6.73b 6.6 6 6.61k
St. dev. 0.01 0.01 0.00 0.01 0.04 0.01 0.,1 0.01 0.02
Avg.* Average of two readings
St. dev. : Standard deviations of the means C Control samples
Pf Samples inoculated with P. fluorescens Be: Samples inoculated with B. coagulans a-e Superscripts within a row denote significant difference at the p<0.01. Means were
separated using LSD.










Table 4-6.


pH of reduced-fat milk inoculated with Pseudomonasfluorescens or Bacillus c'ln


pH
Storage Time (Days) 1.70C 7.20C 12.80C C Pf Bc C Pf Bc C Pf Bc 0 Avg.* 6.75 6.70 6.75 6.75 6.70 6.75 6.75 6.70 6.75
St. dev. 0.00 0.13 0.01 0.00 0.13 0.01 0.00 0.13 0.01
3 Avg. 6.72 b &V 6.32' 6727 6.70 6.78 6.72 b 6.67" 737
St. dev. 0.04 0.05 0.02 0.01 0.10 0.01 0.02 0.02 0.01
Exp. 1 5 Avg. 6.750 6.76 6.74k 6.73 6.676 6.73k 6.73" 6.53a 6.74b
E_ St. dev. 0.01 0.01 0.02 0.00 0.03 0.01 0.01 0.01 0.06
7 Avg. 6.71b 6.68a 6.71b 67.7 6.66a 6.73- 6.73b 6.51 6.71b
St. dev. 0.01 0.01 0.01 0.03 0.03 0.01 0.06 0.08 0.01
10 Avg. 6.7 6.69 6.71 6.73 6.0 6.70d 67 6
St. dev. 0.02 0.00 0.01 0.01 0.03 0.00 0.00 0.01 0.03
0 Avg. 6.75 6.75 6.75 6.75 6.75 6.75 6.75 6.75 6.75
St. dev. 0.01 0.00 0.00 0.01 0.00 0.00 0.01 0.00 0.00
3 Avg. 6.76 6.76 6.76 6.75 6.78 6.76 6.77 6.68 6.88
St. dev. 0.01 0.00 0.00 0.01 0.01 0.01 0.01 0.07 0.07
Exp. 2 5 Avg. 6.78b 6.77a 6.76b 6.74b 6.71& 6.73b 6.74b 6.67a 6.73b
Ep St. dev. 0.01 0.00 0.01 0.01 0.01 0.00 0.00 0.04 0.00
7 Avg. 66.75- 6.74k 6.7 6.73- 6.74 6.54a 6.74T
St. dev. 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.00
10 Avg. 6.69" 6.67' 6.71 6.69 6.5lb 6.69 6.77c 6.46 6.69e
St. dev. 0.01 001 001 0.01 0.01 0.00 0.01 0.01 0.01


Avg.* : Average of two readings St. dev. : Standard deviations of the means C Control samples
Pf Samples inoculated with P. fluorescens Bc: Samples inoculated with B. coagulans a-e Superscripts within a row denote significant difference at the p<0.01. Means were
separated using LSD.










Table 4-7. pH of fat-free milk inoculated with Pseudomonasfluorescens or Bacillus
coagun.

Storage Time (Days) 1.70C 7.20C 12.80C C Pf Bc C Pf Bc C Pf Bc 0 Avg.* 6.77 6.75 6.74 6.77 6.75 6.74 6.77 6.75 6.74
St. dev. 0.01 0.02 0.03 0.01 0.02 0.03 0.01 0.02 0.03
3 Avg. 6.72 6.74 6.75 6.72 6.74 6.72 6.78 6.76 6.76
St. dev. 0.01 0.01 0.02 0.01 0.02 0.00 0.04 0.04 0.02
Exp.1 5 Avg. 6.72 6 671b 6.72 6 6.72 6.618 6.72bc St. dev. 0.00 0.01 0.00 0.00 0.01 0.01 0.00 0.01 0.01
7 Avg. 6.82 6.75 6.74 6.72 6.70 6.72 6.69 6.71 6.70
St. dev. 0.09 0.05 0.02 0.01 0.03 0.01 0.04 0.01 0.00
10 Avg. 6.76 6.69 6.76 6.75 6.75 6.73 6.70 6.67 6.72
St. dev. 0.01 0.11 0.00 0.02 0.06 0.01 0.00 0.00 0.02
- Avg. 6.67& 6.70b 6.69b 6.67' 670W 69r 6.67' 6.70b 6.69b
St. dev. 0.01 0.00 0.00 0.01 0.00 0.00 0.01 0.00 0.00
3 Avg. 6.72" 6.72' 6.72" 6.74b 6.76b 6.76b 6.75b 6.75b 6.73b
St. dev. 0.00 0.01 0.01 0.00 0.01 0.00 0.04 0.01 0.01
Exp.2 5 Avg. 6.71 6.71 6.71 6.71 6.70 6.72 6.70 6.67 6.70
E. St. dev. 0.05 0.00 0.01 0.01 0.01 0.01 0.01 0.02 0.01
7 Avg. 6.69"' 6.68b 6.69b 6.70k 6.71c 6.7e 6.688b 6.79d 6.67
St. dev. 0.01 0.00 0.01 0.01 0.00 0.00 0.00 0.01 0.01
10 Avg. 6.71' 6.70' 6.70' 6.71a 6.78b 6.70 6.722 6.930 6.70
-t. dev. 000 0.01 0.00 0.01 0.01 0.00 0.00 0f01 0.01
*Avg: Average of two readings
St. dev. : Standard deviations of the means C Control samples
Pf Samples inoculated with P. fluorescens Bc: Samples inoculated with B. coagulans a-d: Superscripts within a row denote significant difference at the p<0.01. Means were
separated using LSD.











pH Measurements for Accelerated Study

In the accelerated study, the changes in pH for all types of milk during storage did not follow any specific trend (Table 4-8). According to the 10-day storage study, pH for the samples inoculated with P. fluorescens and stored at higher temperatures decreased more than for the samples stored at lower temperatures. However, the opposite occurred in the accelerated study. The incubation temperature for the accelerated study was 28'C and it was expected to have higher pH drops, but contrary to this, pH increased for all milk types and experiments except for whole milk experiment 2 samples. The reason behind this was not understood. In most cases there were slight pH changes for control and samples inoculated with B. coagulans. No research has been found in the literature on effects of growth of P. fluorescens and B. coagulans on pH. pH Measurements for Combination and Accelerated Studies

In the combination study, pH values for whole milk control and inoculated with P. fluorescens and B. coagulans samples are given in Table 4-9. The pH values of all control samples at all storage temperatures increased. On the other hand, pH values for samples inoculated with both microorganisms and stored at 1.7�C did not change, but pH values for samples stored at 7.2* and 12.80C decreased. The reason for this could not be explained.

The pH values for the whole milk control and inoculated with P. fluorescens and B. coagulans for the accelerated study are given in Table 4-10. For the control samples pH slightly dropped, but pH values increased for the other samples.










Table 4-8. pH of all types of milk inoculated with Pseudomonasfluorescens or
Bacillus caulninacceleratedstd


Avg.*: St. dev.


Average of two readings : Standard deviations of two readings


pH
Mlk Analyses Experiment I Experiment 2
Type Time (Inoculated microor ganisms) (Inoculated microor 4anisms)
(Days) Control Pseudomonas Bacillus Control Pseudomonas Bacillus fluorescens coagulans fluorescens coagulans
I Avg.* 6.64 6.81 6.73 6.70 6.78 6.72
St. dev. 0.01 0.01 0.03 0.03 0.09 0.03
4 Avg. 6.75 6.81 6.66 6.74 6.83 6.82 Whole St. dev. 0.01 0.02 0.01 0.01 0.03 0.01 Milk 6 Avg. 6.66 6.84 6.67 6.70 6.86 6.78
St. dev. 0.02 0.02 0.00 0.03 0.05 0.03
8 Avg. 6.68 6.88 6.67 6.66 6.77 6.71
St. dev. 0.04 0.01 0.01 0.01 0.02 0.01
1 Avg. 6.71 6.74 6.70 6.71 6.72 6.70
St. dev. 0.01 0.01 0.02 0.01 0.01 0.01
4 Avg. 6.66 6.73 6.69 6.76 6.84 6.79 Reduced- St. dev. 0.02 0.01 0.01 0.01 0.02 0.02 fat Milk 6 Avg. 6.70 6.79 6.69 6.54 6.84 6.72
St. dev. 0.01 0.00 0.01 0.03 0.01 0.00
8 Avg. 6.60 6.77 6.68 6.72 6.97 6.74
St. dev. 0.08 0.04 0.01 0.01 0.01 0.00
I Avg. 6.66 6.69 6.48 6.65 6.55 6.58
St. dev. 0.01 0.08 0.01 0.01 0.02 0.01
4 Avg. 6.56 6.84 6.50 6.71 6.74 6.69 Fat-free St. dev. 0.04 0.06 0.04 0.04 0.01 0.01 Milk 6 Avg. 6.42 6.79 6.69 6.66 6.78 6.68 St. dev. 0.01 0.03 0.00 0.03 0.01 0.01
8 Avg. 6.72 6.92 6.73 6.72 6.83 6.71
St. dev. 0.011 0.02 0.00 0.00 0.03 0.00










Table 4-9. pH of whole milk control and inoculated with P. fluorescens and B.
co ulans samples.
pH
Pseudomonasfluorescens
Storage Control and SampleCotl
Time Bacillus coagulans (Days) Storage Temperature Storage Temperature
1.70C 7.2-C 12.8-C 1.7�C 7.2�C 12.80C
0 Avg.* 6.65 6.65 6.65 6.70 6.70 6.70
St. dev. 0.05 0.05 0.05 0.01 0.01 0.01
3 Avg. 6.68 6.67 6.67 6.69 6.69 6.72
St. dev. 0.01 0.00 0.00 0.00 0.01 0.01
Avg. 6.68 6.70 6.68 6.64 6.63 6.57
5_ St. dev. 0.02 0.00 0.01 0.06 0.09 0.04 7 Avg. 6.70 6.71 6.70 6.72 6.71 6.56
St. dev. 0.01 0.01 0.00 0.00 0.01 0.00
10 Avg. 6.71 6.72 6.71 6.70 6.64 6.64
1 St. dev. 0.00 0.00 0.00 0.00 0.01 0.01
Avg.* Average of two readings
St. dev. : Standard deviations of two readings



Table 4-10. pH of whole milk control and inoculated with P. fluorescens and B.
co-ulans samples in accelerated study.
Analyses pH
Tmlyes Sal Pseudomonasfluorescens
Time Sample Control and
(Days) Bacillus coagulans
Avg.* 6.70 6.69
1 St. dev. 0.03 0.01
4 Avg. 6.66 6.67
St. dev. 0.02 0.08
6 Avg. 6.71 6.73
St. dev. 0.00 0.04
8 Avg. 6.68 6.80
St. dev. 0.01 0.02


Avg.* : St. dev.


Average of two readings : Standard deviations of two readings










Sensory Evaluation


Sensory data showed that in general panelists detected the odor differences due to the growth of P. fluorescens in all types of milk samples, but had difficulties in detecting the odor changes due to B. coagulans. This might be because B. coagulans could not grow at 1.70, 7.20, and 12.8�C and did not generate enough microbial metabolites which gave off-odors. Sensory scores given by ten panelists for all milk types are given in Appendix C. The average sensory scores for whole, reduced-fat, and fat-free milk control, inoculated with P. fluorescens or B. coagulans and hidden control samples and their standard deviations are shown in Tables 4-11, 4-12 and 4-13, respectively. The hidden control samples were the same as the reference samples. At day 0, hidden control samples were not presented to the panelists since all the samples were fresh and they all were assumed to have a difference of 0.

The average sensory scores for all milk types and for both experiments were

rounded off to the nearest integer to facilitate analysis using DFA. These scores are given in Table 4-14. In most cases, especially toward the end of the storage period and at higher storage temperatures, the sensory scores for control and hidden control samples are significantly different from the sensory scores of samples treated with P. fluorescens or B. coagulans. The scores for samples treated with P. fluorescens were significantly different from the rest of the sensory scores. Overall, the sensory scores for the controls and hidden control samples were not significantly different from each other. As the temperature and storage time increased, the sensory scores given to the samples











Sensory scores for whole milk inoculated with P. fluorescens or B.


Senor Scre
Inoculated Storage Time Experiment Sensory Scores xperiment 2
Microorganisms (Days) 1.70C 7.20C 12.80C 1.70C 7.20C 12.80C
0 Avg.* 0.00 0.00 0.00 0.00 0.00 0.00
0 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00
3 Avg. 1.00 1.30 0.80 0.70 0.60 1.30
St. dev. 0.82 1.16 0.92 1.25 0.97 1.49
Control 5 Avg. 1.20 0.60 1.00 0.30 1.60 1.50
St. dev. 1.03 0.70 1.05 0.48 0.48 0.97
7 Avg. 1.40 1.40 1.00 0.90 1.50 1.20
St. dev. 1.07 1.84 0.67 0.99 1.18 1.48
10 Avg. 1.30 1.20 1.00 0.70 2.00 1.20
St dL 1.16 0.79 1.05 0.95 1.49 1.03
0 Avg. 0.00 0.00 0.00 0.00 0.00 0.00
- St. dev. 0-00 0.00 000 0.00 000 0.00
3 Avg. 1.20 1.40 1.40 1.00 1.60 1.80
St- dev 114 0.97 0.97 0.94 1.07 L14 Pseudomonas 5 Avg. 1.80 0.80 4.10 0.60 0.60 4.30 fluorescens St. dev. 1.87 1.40 2.33 0.70 0.70 1.49 7 Avg. 2.60 2.30 6.90 1.60 2.00 5.80
St. dev. 1.84 0.82 1.37 1.35 1.15 2.57
10 Avg. 1.50 5.70 8.90 1.80 3.20 8.10
St. dev. 1.43 2.06 1.45 1.23 1.93 1.60
0 Avg. 0.00 0.00 0.00 0.00 0.00 0.00
St. dev. 0.00 0.00 0.00 0.00 0.00 0.00
3 Avg. 1.30 1.20 0.50 1.00 1.20 1.20
St. dev. 0.82 0.92 0.85 1.25 1.48 1.14
Bacillus 5 Avg. 0.20 0.70 1.20 1.80 1.20 4.30 coagulans St. dev. 0.63 0.67 1.03 0.99 1.14 0.95
7 Avg. 1.30 0.80 1.20 1.00 1.10 7.10
St. dev. 0.95 0.79 1.40 1.25 1.29 1.66
10 Avg. 1.30 1.30 1.40 1.30 2.40 7.70
St. dev. 1.16 0.95 1.07 1.06 2.01 1.34
0 Avg. - - -
St. dev. - - - -
3 Avg. - 0.20 0.20 - 0.50 0.50
St. dev. - 0.42 0.42 - 0.97 1.08 Hidden Control 5 Avg. - 0.50 0.40 1.10 1.10 1.10
St. dev. - 0.53 0.70 0.99 1.10 0.57
7 Avg. - 1.30 0.50 1.00 1.00 0.80
St. dev. - 0.67 0.71 0.94 1.41 1.14
10 Avg. - 1.00 0.80 0.60 0.70 0.40
St. dev. - 1.15 0.42 0.84 1.25 0.70


The 0 to 10 scale was used (0 = no difference and Avg.* : Average of ten readings St. dev. : Standard deviation often readings


10 = very different).


Table 4-11.











Table 4-12. Sensory scores for reduced-fat milk inoculated with P. fluorescens or B.

Senor Scre
Inoculated Storage Time Exi t Sensor Experiment 2
Microorganisms (Days) 1.70C 7.20C 12.80C 1.70C 7.20C 12.80C
0 Avg.* 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 1.00 1.00 0.80 0.91 2.72 1.19 St. dev. 0.82 0.94 1.23 0.94 2.08 0.88 Control 5 AvR. 0.90 1.00 0.20 2.09 0.69 1.44 St. dev. 1.29 0.94 0.42 1.64 0.84 0.97 7 Avg. 1.30 1.20 2.10 1.82 0.71 0.91 St. dev. 1.16 1.03 1.45 1.83 0.97 1.20 10 Av. 1.20 1.60 1.40 1.45 0.91 0.80 St dev. 0.63 1.90 1.07 2.98 0.82 0.00 0 Av. 0.00 0.00 0.00 0.00 0.00 0.00 St e. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 1.80 2.30 6.50 0.91 1.50 1.20 St. dev. 1.14 2.16 2.12 1.04 1.35 0.92 Pseudomonas 5 Avg. 1.20 3.40 6.80 2.27 2.20 5.20 fluorescens St. dev. 1.32 2.95 1.99 1.19 1.40 1.32 Avg. 3.30 3.00 8.00 1.91 3.30 8.00 St. dev. 0.67 1.25 1.25 2.07 1.34 1.33 10 Avg. 2.30 6.10 8.70 2.45 8.40 9.10 St. dev, 1.16 1.5 2.11 2.70 1.17 1.0 0 Avg. 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Av. 0.90 1.30 1.50 1.27 1.30 1.30 St. dev. 0.99 1.34 1.35 0.90 1.34 1.06 Bacillus 5 Avg. 1.90 2.00 1.20 2.18 1.90 1.90 coagulans St. dev. 1.66 0.94 0.79 1.54 1.10 0.32 Avg. 1.50 1.60 1.50 1.73 1.40 2.00 St. dev. 0.85 0.97 1.58 1.95 1.07 0.94 10 Avg. 0.90 1.50 1.80 2.55 3.00 3.60 qt- dev. 1.20 07 2.62 2.81 1.56 1.17
0 Avg. -- - -
St. dev. - - - -
3 Avg. 0.60 0.60 1.00 1.10 0.30 St. dev. 1.23 0.97 1.68 1.20 0.67 Avg. 1.10 1.20 0.73 0.80 0.70 Hidden Control - St. dev. 0.88 0.92 1.68 1.23 1.06
7 Avg. 0.70 0.90 1.64 0.60 0.90 St. dev. 0.82 1.10 2.11 0.70 1.66 10 Avg. 0.80 1.40 1.82 0.70 0.40 St. dev. - 0.79 2.07 2.82 1.16 0.70 The 0 to 10 scale was used (0= no difference and 10 = very different). Avg.* : Average of ten readings
St. dev. : Standard deviations of ten readings











Table 4-13. Sensory scores for fat-free milk inoculated with P. fluorescens or B.
Inoculaed StoageTimSensory Scores Inoculated Storage Time Experiment 1 Experiment 2
Microorganisms (Days) 1.70C 7.20C 12.80C 1.70C 7.20C 12.80C
0 Avjz.* 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00
3 Avg. 0.30 0.70 0.50 0.80 1.10 0.60 St. dev. 0.95 0.67 0.85 0.92 1.29 0.70 Control 5 Avg. 2.10 1.10 1.50 0.40 1.00 0.50 St. dev. 1.60 0.57 0.71 0.70 1.05 0.53 7 Avg. 0.90 2.10 3.20 1.00 1.00 1.30 St. dev. 0.88 1.10 2.49 0.82 0.82 1.16 10 1Ave. 0.80 1.30 2.60 0.40 0.90 0.70 1 t dev 0 .67 1.58 0.70 0.88 048
0 Avg. 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 0.40 0.90 2.40 1.30 1.10 3.80
1 St. dev. 0.52 1.00 0.97 0.95 1.10 1.32 Pseudomonas 5 Avg. 2.00 1.40 7.20 1.60 2.20 7.50 fluorescens St. dev. 0.94 0.70 1.03 1.58 1.62 1.08 7 Avg. 2.20 3.60 8.60 2.00 3.60 9.10 St. dev. 0.79 0.84 1.17 1.15 1.65 0.74 10 Avg. 3.10 5.10 9.80 3.60 6.60 9.80
-St dev 2.38 2-42 0.63 1.35 0-70 0.42
0 Avg. 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 0.90 0.80 1.20 1.50 1.70 1.50 St. dev. 0.88 0.92 1.14 1.27 0.67 0.71 Bacillus 5 Avg. 0.90 1.10 1.30 1.70 2.50 2.50 coagulans St. dev. 0.74 0.32 0.67 1.34 1.43 0.97
7 Avg. 1.70 2.50 5.90 1.50 1.80 4.90 St. dev. 0.48 0.97 0.74 0.85 1.23 1.66 10 Avg. 2.40 2.70 8.40 2.20 2.10 3.00 St de 1.78 1.70 1.17 1.03 0.99 0.82
0 Avg. - - -
St. dev. - - - - -
3 Avg. 0.90 0.60 0.80 0.40 0.20 0.20 St. dev. 0.57 0.70 1.40 0.97 0.63 0.63 Hidden Cont 5 Avg. 0.30 0.30 0.40 0.40 0.20 0.30 St. dev. 0.67 0.95 1.26 0.70 0.42 0.48 7 Avg. 0.20 0.20 0.60 0.30 0.10 0.40 St. dev. 0.63 0.63 1.26 0.67 0.32 0.84 10 Avg. 0.70 0.30 0.80 0.30 0.20 0.20 St. dev. 1.06 0.67 1.32 0.48 0.42 0.42


The 0 to 10 scale was used (0 = no difference and Avg.* : Average of ten readings St. dev. : Standard deviations of ten readings


10 = very different).










Table 4-14. Average sensory scores (rounded) of all types of milk inoculated with P.
fluresensor B. caua


Storage Time Sensory Scores
F (Days) 1.70C 7.20C 12.80C
C Pf Bc Hc C Pf Bc Hc C Pf Bc Hc 0 Avg.*0 0 0 - 0 0 0 - 0 0 0 3 Avg. I 1 1 - 1 1 1 0 1 1 1 0
Reply 5 Avg. 1' 2" F -- 1 1 1 1 1b 4a lb 0b
7 Avg. Ib 3a lb lb 2 lb I I 7a l I 10 Avg. 1 2 T - 6 1b lb lb Ib lb
- 0 0 0 O 0- 0 0 0 - 0 00
3 J Avg. I1 1 1 - b 2' lb 1 1 2 1 1
Rep 2 5 Avg. 0c 1b- 2a lab 2 2 1 1 2b 43 43 lb
7 Avg. 1 2 1 1 2 2 1 1 1 6 7 F 10 Avg. 1 2 1 1 2-b 3 2 1l b 8' 8& 0b - Av 0 0 0 - 0 0 0 - 0 0 0 3 Avg. lb 2a lb 1 2 1 1 lb 7a 2b lb
Rep 1 5 1 1 2 - b 3' 2ab lb 0b 7. 1b lb
7 Avg. 1bF 3a- 2 b -1b 3a Pb 1 2 8' 2 b '-IV
510 Avg. 1 2 1 -2' -I- V" 2b 1b I 8a -25b 77
0 Avg. 0 0 0 0 o o - 0 0 0 3 Avg. I I I 1 3 2 1 1 1 1 1 0
, Rep 2 5 Avg. 2a 2'" 2 b 2k 5a 2b 1�
7 Avg. I 1 1 2 lb 3a I1 lb lc 8a 2 b 1 10 Avg. 1 2 2 2 1P 8a 3b 1, 1c 9a 45 C
0 Avg. 0 0 0 - 0 0 0 - 0 0 0
3 Avg. 0 0 1 1 1 1 1 1 lb 27- lb 1
Rep 1 5 Avg. 2 a 2 a Ib 0b 1 I 1 0 2b 7 a b 0 c
7 Avg. P 2 2T 0 2b 4 3b 0F 3C 9a 6 1 10 Avg. lb 3' 2" 1b 1. 5a 3b 0 3c 10a 8b Id 0 Avg. 0 0 0 - 0 0 0 - 0 0 0 3 Avg. 1 1 2 0 1b - 42 12 4 r0c
Rep 2 5 Avg. 2 2 0 1b 2a 3a 0b 1P 8a 3 0C
7Avg. 1 2 214 2Oc 12 2 4 10 Avg. 0c 4' 2Tb 0c 1c 7' 2b; 0a 1c 10a 3b 06 The 0 to 10 scale was used (0 = no difference and 10 = very different). Avg.* : Average of two readings
C : Control samples
Pf Samples inoculated with P. fluorescens Bc� Samples inoculated with B. coagulans Hc Hidden control samples
a-d. Superscripts in each row within each storage temperature denote significant difference
at the p<0.01. Means were separated using LSD.










inoculated with P. fluorescens increased which meant that the odors of those samples were getting very different than the odor of the reference sample which was fresh milk. This was as expected since when the microbial counts exceeded 106 cfu/ml, the product started to have off-odors. This caused an adverse influence on product quality.

Sensory data for the whole milk inoculated with both microorganisms in the

combination study also showed that the panelists detected the overall odor changes caused by the microorganisms. The sensory scores given by ten panelists are given in Appendix C. The average sensory scores and their standard deviations for the combination study are shown in Table 4-15. The average sensory data on Table 4-15 was rounded of to the nearest integer to facilitate analysis with DFA as shown in Table 4-16. Panelists rated the control and hidden control samples from 0 to 1. There were no significant differences between these two samples according to the panelists except for the samples stored at 7.20 and 12.80C at day 10. They rated the control samples as 1 and hidden control samples as

0 (Table 4-16). This was a small difference since ratings were on a 0 to 10 scale.

Overall, the odor of the inoculated samples was significantly different from the odor of the control and hidden control samples. As the storage time increased, the odor differences increased, possibly due to the accumulation of bacterial metabolites based on the growth of P. fluorescens. The storage temperature had an effect on the sensory scores. The sensory scores for samples inoculated with both microorganisms and stored at 12.8'C were higher than the scores for the samples inoculated with both microorganisms and stored at 7.20C. This was as expected since the microbial loads were higher at higher storage temperatures.










Table 4-15. Sensory scores for whole milk inoculated with P. fluorescens and B.
coagulans in combination study.
Inoculated Storage Time Sensory Scores
Mcroorganisms (Days) 1.70C 7.2 12.8
0 Avg.* 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 3 Avg. 0.30 0.50 0.20 St. dev. 0.48 0.85 0.63 Control 5 Avg. 1.10 0.70 0.80 St. dev. 0.74 0.48 0.79 7 Avg. 1.00 1.30 0.70 St. dev. 0.94 0.95 0.67 10 Avg. 0.40 1.10 1.00 St. dev. 0.51 0.74 0.67 0 Avg. 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 3 Avg. 0.70 0.80 1.20 Psudons St. dev. 0.48 0.63 0.79 and 5 Avg. 0.70 1.40 7.00 and St. dev. 0.82 0.70 1.56
coagulans 7 Avg. 1.20 3.90 8.00 St. dev. 1.03 2.08 1.70 10 Avg. 3.20 6.50 10.00 St. dev. 1.40 1.08 0.00
0 Avg. - - St. dev.
3 Avg. 0.00 0.10 0.60 St. dev. 0.00 0.32 1.26 5 Avg. 0.40 0.60 0.60 St. dev. 0.70 0.84 1.07 7 Avg. 0.70 0.70 0.50 St. dev. 1.25 0.95 0.85 10 Avg. 0.40 0.20 0.10 St. dev. 0.70 0.42 0.32 The 0 to 10 scale was used (0 = no difference and 10 = very different). Avg.* : Average of ten readings St. dev. : Standard deviation often readings










Table 4-16.


Average sensory scores (rounded) of whole milk inoculated with P. fluorescens and B. coaQ'ulan.' in comhination .tiidv


Sensory Scores
Storage Sample 1.70C 7.20C 12.80C
Time
(Days)C Pf+Bc Hc C Pf+Bc Hc C Pf+Bc Hc
0 Avg.* 0 0 - 0 0 - 0 0
3 Avg. 0 1 0 1 1 0 OF 1. 1A 5 Avg. 1 1 0 1 is lb lb 7a 1b 7 Avg. 1 1 1 I 4" I 1l 88 Pj 10 Avg. 0b 3a 0b lb 78 Oc 1b 10 OC
-11. _ / a 1 _ !. . . . 1 .,rt J , - -,, - -, .


Ine 0 to 1 scale was used ktu = no diference and 10o = very altterent).
*Avg : Average often readings C : Control samples
Pf + Bc : Samples inoculated with P. fluorescens and B. coagulans Hc: Hidden control
8-C : Superscripts in a row within each storage time denote significant difference storage at
the p<0.01. Means were separated using LSD.


Electronic Nose Measurements


Electronic nose sensor data showed very good classification results based on

microbial counts and sensory scores for whole, reduced-fat and fat-free milk inoculated with P. fluorescens or B. coagulans. Electronic nose sensor data was analyzed separately for each experiment. Experiments of whole milk study for each microorganism were also combined to observe the degree of classification on pooled data. In this section, only results of the whole milk study is explained in detail, reduced-fat and fat-free milk studies were explained briefly since these studies had similar trends with the whole milk study. Due to the bulk of the electronic nose sensor data, it can be obtained from Dr. Murat Balaban, Food Science and Human Nutrition Department at the University of Florida (filenames:"\\thesis\e-nose data.txt").










The electronic nose was able to discriminate the odor changes in whole milk

experiment 1 samples based on P. fluorescens counts with accuracies of 100% for storage temperatures of 1.70, 7.2�, and 12.8�C (Table 4-17). DFA was also capable of separating the data into different microbial loads for B. coagulans with correct classification rates of 100%, 100% and 96% for the storage temperatures of 1.70, 7.20, and 12.8�C, respectively (Table 4-17). By using DFA for whole milk experiment I samples stored at three different temperatures, the twelve sensor outputs were reduced to two discriminant functions (Table 4-18) to calculate coordinates of points which were mapped on the twodimensional plots. The two-dimensional plots generated from the discriminant functions showed a clear separation of the microbial counts for P. fluorescens and B. coagulans. Figure 4-8 shows results from the DFA for the whole milk experiment 1 samples inoculated with P. fluorescens and Figure 4-9 shows results from the DFA for the whole milk experiment 1 samples inoculated with B. coagulans. All samples showed perfect separation of the clusters.

The data of the whole milk experiment I samples at all storage temperatures were pooled together and by using DFA, the twelve sensor outputs were reduced to two discriminant functions (Table 4-18). These two functions (all temp 2)were used to calculate coordinates of points for data at each storage temperature which were mapped on the two-dimensional plots. For example, for P. fluorescens 12 sensor readings at

1.7�C were taken and these values were multiplied with the specific coefficients of discriminant function I and function 2 and summed with the constant. Then it was plotted on the two-dimensional plot. All data were processed this way. Data from each










Table 4-17. Correct classification rates obtained from the DFA of electronic nose
sensor readings compared with microbial counts of whole milk experiment
I and experiment 2 samples separately.
Correct Classification Rate (%)

MStorage Temperatures
" Microbial
Counts Experiment 1 Experiment 2

All All
1.70C 7.20C 12.80C l 1.70C 7.20C 12.80C Tem Temp Temp n 25 25 25 75 25 25 25 75
104 100.00 100.00 100.00 100.0 100.00 100.00 100.00 100.0 10, 100.00 100.00 - 80.00 100.00 100.00 - 80.00 106 100.00 - 90.00 80.00 100.00 - 70.00 107 - 100.00 - 60.00 100.00 - - 100.0
10, 100.00 100.00 100.00 33.00 - 100.00 100.00 90.00
10' - 100.00 100.00 70.00 - 100.00 100.00 80.00
Overall 100.00 100.00 100.00 72.00 96.00 100.00 100.00 85.33
n 25 25 25 75 25 25 25 75 102 100.00 100.00 100.00 90.00
103 100.00 100.00 90.00 93.33 100.00 - 100.0 104 - - - - 100.00 100.00 - 66.67 10 100.00 100.00 100.00 100.0 100.00 100.00 100.00 88.57 106 - - - - 100.00 100.00 100.00 75.00
Overall 100.00 100.00 96.00 93.33 100.00 100.00 100.00 81.33
n = number of electronic nose readings used to obtain the DFA functions




Full Text

PAGE 1

MICROBIAL AND SENSORY ASSESSMENT OF MILK WITH AN ELECTRONIC NOSE By FIGEN KOREL A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2000

PAGE 2

To my parents, Zerrin and Metin Korel, for all their love, encouragement, support and for the invaluable opportunity they gave me throughout my life to obtain the best education possible.

PAGE 3

ACKNOWLEDGMENTS I would like to express my sincere thanks and appreciation to my major advisor, Dr. Murat O. Balaban, for his invaluable guidance, patience, encouragement, and friendship during my graduate studies. My appreciation is also extended to my committee members, Drs. Fortier, Rodrick, Sims, and Williams for their guidance and recommendations for the success of this study. I want to thank Celal Bayar University (Manisa, Turkey) for giving me the opportunity to conduct my graduate studies in the United States. I would like to especially thank Dr. Diego A. Luzuriaga for all his invaluable assistance and unconditional patience during this project. I would like to thank Ash Z. Odaba§i for her constant support and friendship throughout the completion of this study. My sincere and grateful esteem to Bee Mach for helping me an)^ime when I needed and Necla Demir for helping me in running some of the experiments. I also appreciate the panelists that helped in the sensory studies. Ill

PAGE 4

TABLE OF CONTENTS page ACKNOWLEDGMENTS iii LIST OF TABLES vii LIST OF FIGURES xiii ABSTRACT xvi CHAPTERS 1 INTRODUCTION 1 2 LITERATURE REVIEW 3 Milk Quality Assessment 3 Milk Spoilage 4 Milk Volatiles and OfF-Flavors 15 Factors Affecting Shelf Life of Milk 18 Rapid Methods for the Detection of Microorganisms 20 Immunomagnetic Separation (IMS) 21 Impedance Microbiology 21 Enzyme Immunoassays and Latex Agglutination Tests 22 Bioluminescent Systems 23 Miniaturized Methods 24 Other Biochemical Methods 25 Electronic Nose 26 Electronic Nose Technology 28 Sensor Technology 30 Applications of Electronic Nose 34 Applications in Microbial Detection 35 Applications to Food Products 37 Current Shortcomings of the Electronic Nose 41 Objectives 42 IV

PAGE 5

3 MATERIAL AND METHODS 44 Milk Sampling, Inoculation, and Analysis 44 Milk Samples 44 Activation of Microorganisms 45 Inoculation of Microorganisms and Sample Treatments 46 Electronic Nose Measurements 47 Microbial Analysis 48 Moisture Content Measurements 50 Fat Content Measurements 50 pH Measurements 50 Sensory Evaluation 51 Data Analysis 51 4 RESULTS AND DISCUSSION 53 Milk Sampling, Inoculation, and Analysis 53 Moisture and Fat Content Measurements 53 Microbial Analysis 53 pH Measurements 67 Sensory Evaluation 74 Electronic Nose Measurements 81 5 CONCLUSIONS AND RECOMMENDATIONS 122 APPENDICES 124 A DATA FOR MICROBIAL ANALYSIS 124 B DATA FOR pH 135 C DATA FOR SENSORY ANALYSIS 149 REFERENCES 163 BIOGRAPHICAL SKETCH 174 V

PAGE 6

LIST OF TABLES Table page 2-1 . Residual enzyme activity from psychrotrophic organisms in HTST pasteurized milk 12 2-2. Residual activity of extracellular enzymes of psychrotrophic organisms after UHT sterilization 14 2-3. Flavor defects associated with psychrotrophic Bacillus spp. grown in milk at7.2°C 17 4-1 . Microbial load of all types of milk samples inoculated with Pseudomonas fluorescens or Bacillus coagulans 54 4-2. Microbial load of all types of milk inoculated with Pseudomonas fluorescens or Bacillus coagulans in accelerated study 63 4-3. Microbial load of whole milk samples inoculated with Pseudomonas fluorescens and Bacillus coagulans in combination study 65 4-4. Microbial load of whole milk samples inoculated with Pseudomonas fluorescens and Bacillus coagulans in accelerated study 67 4-5. pH of whole milk inoculated with Pseudomonas fluorescens or Bacillus coagulans 68 4-6. pH of reduced-fat milk inoculated with Pseudomonas fluorescens or Bacillus coagulans 69 4-7. pH of fat-free milk inoculated with Pseudomonas fluorescens or Bacillus coagulans 70 4-8. pH of all types of milk inoculated with Pseudomonas fluorescens or Bacillus coagidans in accelerated study 72 Vll

PAGE 7

4-9. pH of whole milk control and inoculated with P. fluorescens and B. coagulam samples 73 4-10. pH of whole milk control and inoculated with P. fluorescens and B. coagulans samples in accelerated study 73 4-11. Sensory scores for whole milk inoculated with P. fluorescens or B. coagulans 75 4-12. Sensory scores for reduced-fat milk inoculated with P. fluorescens or B. coagulans 76 4-13. Sensory scores of fat-free fat milk inoculated with P. fluorescens or B. coagulans 77 4-14. Average sensory scores (rounded) of all types of milk inoculated with P. fluorescens or B. coagulans 78 4-15. Sensory scores for whole milk inoculated with P. fluorescens and B. coagulans in combination study 80 4-16. Average sensory scores (rounded) of whole milk inoculated with P. fluorescens and B. coagulans in combination study 81 4-17. Correct classification rates obtained from the DFA of electronic nose sensor readings compared with microbial counts of whole milk experiment 1 and experiment 2 samples separately 83 4-18. DFA coefficients for microbial counts correlated to electronic nose sensor readings. Whole milk experiment 1 samples inoculated with Pseudomonas fluorescens or Bacillus coagulans 84 4-19. DFA coefficients for microbial counts correlated to electronic nose sensor readings. Whole milk experiment 2 samples inoculated with Pseudomonas fluorescens or Bacillus coagulans 90 4-20. DFA coefficients for microbial counts correlated to electronic nose sensor readings. Whole milk experiment 1 and 2 samples inoculated with Pseudomonas fluorescens or Bacillus coagulans 93 4-21. Correct classification rates obtained from the DFA of electronic nose sensor readings compared with microbial counts of whole milk experiment 1 and experiment 2 samples pooled together 95 viii

PAGE 8

4-22. Correct classification rates obtained from the DFA of electronic nose sensor readings compared with microbial counts of reduced-fat milk experiment 1 and experiment 2 samples separately 98 4-23. Correct classification rates obtained from the DFA of electronic nose sensor readings compared with microbial counts of fat-free milk experiment 1 and experiment 2 samples separately 99 4-24. Correct classification rates obtained from the DFA of electronic nose sensor readings compared with sensory scores of whole milk experiment 1 and experiment 2 samples separately 101 4-25. DFA coefficients for sensory scores correlated to electronic nose sensor readings. Whole milk experiment 1 samples inoculated with Pseudomonas fluorescens or Bacillus coagulans 102 4-26. DFA coefficients for sensory scores correlated to electronic nose sensor readings. Whole milk experiment 2 samples inoculated with Pseudomonas fluorescens or Bacillus coagulans 103 4-27. Correct classification rates obtained from the DFA of electronic nose sensor readings compared with sensory scores of whole milk experiment 1 and experiment 2 samples pooled together 107 4-28. DFA coefficients for sensory scores correlated to electronic nose sensor readings. Whole milk experiment 1 and 2 samples inoculated with Pseudomonas fluorescens or Bacillus coagidans 108 4-29. Correct classification rates obtained from the DFA of electronic nose sensor readings compared with sensory scores of reduced-fat milk experiment 1 and experiment 2 samples separately 112 4-30. Correct classification rates obtained from the DFA of electronic nose sensor readings compared with sensory scores of fat-free milk experiment 1 and experiment 2 samples separately 113 4-3 1 . Eigenvalues of the principal component factors obtained by PCA for data sets of each type and experiment of milk 114 4-32. Correct classification rates obtained from the DFA of electronic nose sensor readings compared with sensory scores of whole milk samples inoculated with P. fluorescens and B. coagulans 119 IX

PAGE 9

4-33. DFA coefficients for sensory scores correlated to electronic nose sensor readings. Whole milk samples inoculated with Pseudomonas fluorescens and Bacillus coagulans 120 A1 . Microbial load of whole milk inoculated with .... 124 A-2. Microbial load of whole milk inoculated with Bacillus coagulans 125 A-3. Microbial load of reduced-fat milk inoculated with Pseudomonas fluorescens 126 A-4. Microbial load of reduced-fat milk inoculated with 5ac/7/M5 coagw/a«5 127 A-5. Microbial load of fat-free milk inoculated with Pseudomonas fluorescens ... 128 A-6. Microbial load of fat-ffee milk inoculated with 5ac/7/M5 coagw/aA75 129 A-7. Microbial load of whole milk inoculated with Pseudomonas fluorescens and Bacillus coagulans 130 A-8. Microbial load of whole milk inoculated with Pseudomonas fluorescens or Bacillus coagulans in accelerated study 131 A-9. Microbial load of reduced-fat milk inoculated with Pseudomonas fluorescens or Bacillus coagulans in accelerated study 132 A10. Microbial load of fat-free milk inoculated with Pseudomonas fluorescens or Bacillus coagulans in accelerated study 133 A-1 1 . Microbial load of whole milk inoculated with Pseudomonas fluorescens and Bacillus coagtdans in accelerated study 134 B-1. pH of whole milk without inoculated with microorganisms (control) 135 B-2. pH of whole milk inoculated with Pseudomonas fluorescens 136 B-3. pH of whole milk inoculated with Bacillus coagulans 137 B-4. pH of reduced-fat milk without inoculated with microorganisms (control) ... 138 B-5. pH of reduced-fat milk inoculated with Pseudomonas fluorescens 139 B-6. pH of reduced-fat milk inoculated with Bacillus coagulans 140 X

PAGE 10

B-7. pH of fat-ffee milk without inoculated with microorganisms (control) 141 B-8. pH of fat-ffee milk inoculated with Pseudomonas fluorescens 142 B-9. pH of fat-free milk inoculated with 5ac;7/w5 coagw/a/75 143 B-10. pH of whole milk inoculated with Pseudomonas fluorescens or Bacillus coagulans in accelerated study 144 B-1 1 . pH of reduced-fat milk inoculated with Pseudomonas fluorescens or Bacillus coagulans in accelerated study 145 B-1 2. pH of fat-ffee milk inoculated with Pseudomonas fluorescens or Bacillus coagulans in accelerated study 146 B-13. pH of whole milk inoculated with Pseudomonas fluorescens and Bacillus coagulans 147 B-1 4. pH of whole milk inoculated with Pseudomonas fluorescens and Bacillus coagulans in accelerated study 148 C-1 . Sensory data of whole milk without inoculated with microorganisms (control) 149 C-2. Sensory data of whole milk inoculated with Pseudomonas fluorescens 150 C-3 . Sensory data of whole milk inoculated with Bacillus coagulans 151 C-4. Sensory data of whole milk for hidden control 152 C-5. Sensory data of reduced-fat milk without inoculated with microorganisms (control) 153 C-6. Sensory data of reduced-fat milk inoculated with Pseudomonas fluorescens . 154 C-7. Sensory data of reduced-fat milk inoculated with 5ac;7/w5 155 C-8. Sensory data of reduced-fat milk for hidden control 156 C-9. Sensory data of fat-free milk without inoculated with microorganisms (control) 157 C-10. Sensory data of fat-free milk inoculated with Pseudomonas fluorescens .... 158 xi

PAGE 11

C1 1 . Sensory data of fat-free milk inoculated with Bacillus coagulans 159 C12. Sensory data for fat-free milk for hidden control 160 C-13. Sensory data of whole milk inoculated Pseudomonas fluorescens zndi Bacillus coagulans 161 C-14. Sensory data of whole milk inoculated with Pseudomonas fluorescens and Bacillus coagulans for hidden control 162 xii

PAGE 12

LIST OF FIGURES Eigyre page 4-1 . Average microbial load of P. fluorescem for whole milk, both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation 56 4-2. Average microbial load of P. fluorescens for reduced-fat milk, both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation 57 4-3. Average microbial load of P. fluorescens for fat-ffee milk, both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation 58 4-4. Average microbial load of B. coagulans for whole milk, both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation 59 4-5. Average microbial load of B. coagulans for reduced-fat milk, both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation 60 4-6. Average microbial load of B. coagulans for fat-free milk both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation 61 4-7. Average microbial load of P. fluorescens and B. coagulans for whole milk in combination study, stored at different temperatures over time. Error bars signify ±1 standard deviation 66 4-8. DFA of odors of whole milk experiment 1 samples inoculated with P. fluorescens and stored at 1.7°, 7.2°, and 12.8°C based on microbial counts and electronic nose readings 85 xiii

PAGE 13

4-9. DFA of odors of whole milk experiment 1 samples inoculated with B. coagulans and stored at 1.7°, 7.2°, and 12.8°C based on microbial counts and electronic nose readings 86 4-10. DFA of odors of whole milk experiment 1 samples based on microbial counts and electronic nose readings using discriminant function 1 and 2 obtained from DFA of all temperatures analysis 88 4-11. DFA of odors of whole milk experiment 1 samples based on microbial counts and electronic nose readings using discriminant function 1 and 2 obtained from DFA of all temperatures analysis 89 4-12. DFA of odors of whole milk experiment 2 samples inoculated with P. fluorescens and stored at 1.7°, 7.2°, and 12.8°C based on microbial counts and electronic nose readings 91 4-13. DFA of odors of whole milk experiment 2 samples inoculated with B. coagulans and stored at 1.7° and 7.2°C based on microbial counts and electronic nose readings 92 4-14. DFA of odors of whole milk experiment 1 and experiment 2 samples inoculated With P. fluorescens md stored at 1.7°, 7.2°, and 12.8°C based on microbial counts and electronic nose readings 96 4-15. DFA of odors of whole milk experiment 1 and experiment 2 samples inoculated with 5. coagulans and stored at 1.7°, 7.2°, and 12.8°C based on microbial counts and electronic nose readings 97 4-16. DFA of odors of whole milk experiment 1 samples inoculated with P. fluorescens and stored at 1.7°, 7.2°, and 12.8°C based on sensory scores and electronic nose readings 104 4-17. DFA of odors of whole milk experiment 2 samples inoculated with P. fluorescens and stored at 1.7°, 7.2°, and 12.8°C based on sensory scores and electronic nose readings 105 4-18. DFA of odors of whole milk experiment 2 samples inoculated with B. coagulans and stored at 1.7°, 7.2°, and 12.8°C based on sensory scores and electronic nose readings 106 4-19. DFA of odors of whole milk experiment 1 and experiment 2 samples inoculated with P. fluorescens and stored at 1.7°, 7.2°, and 12.8°C based on sensory scores and electronic nose readings 109 XIV

PAGE 14

4-20. DFA of odors of whole milk experiment 1 and experiment 2 samples inoculated with B. coagulam and stored at 1.7°, 7.2°, and 12.8°C based on sensory scores and electronic nose readings 110 4-21. Change of shelf life with respect to DPC 1 and BPC 1 for whole milk inoculated with P. fluorescens in accelerated study 115 4-22. Change of shelf life with respect to DPC 1 and BPC 1 for reduced-fat milk inoculated with P. fluorescens in accelerated study 116 4-23. Change of shelf life with respect to DPC 1 and BPC 1 for fat-ffee milk inoculated with P. fluorescens in accelerated study 117 4-24. DFA of odors of whole milk samples inoculated with both microorganisms and stored at 1.7°, 7.2°, and 12.8°C based on sensory scores and electronic nose readings 121 XV

PAGE 15

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MICROBIAL AND SENSORY ASSESSMENT OF MILK WITH AN ELECTRONIC NOSE By Figen Korel May 2000 Chairperson: Murat O. Balaban Major Department: Food Science and Human Nutrition Important psychrotrophs encountered in raw milk are Gram-negative rods with Pseudomonas spp. comprising 65 to 70% of the genera. Some Gram-positive bacteria are also present, with Bacillus being the most important genera. Spoilage bacteria found in raw milk produce heat-resistant lipases and proteinases that are not destroyed by pasteurization. In general, microbial counts in excess of 1x10® cfli/ml are enough to produce defects in milk. Serious off-flavors in milk, such as bitter, putrid, unclean, rancid, and sour, have been associated with psychrotrophic microorganisms. Traditional microbial evaluation of milk is time consuming. Faster methods are desirable. The electronic nose is a promising technology that can be used as a fast screening tool. It enhances objectivity of flavor evaluation, requires minimal sample preparation, generates reproducible and reliable XVI

PAGE 16

results, is easy to operate, and results can be obtained rapidly. The objectives of this study were to test an electronic nose for odor assessment of milk inoculated with Pseudomonas fluorescens znA! or Bacillus coagulans, to correlate microbial loads and sensory results with electronic nose readings, and to attempt to predict the shelf life based on microbial loads of milk samples in an accelerated study. Parmalat® whole, reduced-fat, and fat-ffee milk were used. Sterile milk samples were inoculated With P. fluorescens and/or .S. coagulans, stored at 1.7°, 7.2°, and 12.8°C, and evaluated at days 0, 3, 5, 7, and 10 using an electronic nose. Counts forP. fluorescens were performed using aerobic plate count 3M Petrifilm. Those for B. coagulans were performed using nutrient agar plates. The odor of milk samples was evaluated by a 10-member untrained sensory panel. Electronic nose readings, microbial counts, and sensory data were analyzed using discriminant function analysis. The electronic nose discriminated differences in odor due to microbial load, storage temperature, and sensory data. This research demonstrated the potential use of electronic nose to detect odor differences in milk due to microbial loads. Electronic nose readings can be correlated with sensory panel perception. This may lead to a new rapid method for determining sensory evaluation and microbial loads of milk. xvii

PAGE 17

CHAPTER 1 INTRODUCTION Milk is a good medium for growth of pathogenic and spoilage organisms. Raw milk contains varying numbers of microorganisms, depending on the care employed in milking, cleaning, and handling of milk utensils. Raw milk held at refrigeration temperatures for several days shows the presence of several bacteria of the following genera; Pseudomonas, Enterococcus, Lactococcus, Streptococcus, Leuconostoc, Lactobacillus, Microbacterium, Micrococcus, Propionibacterium, coliforms, Proteus, Bacillus, and some others. Those able to grow at low temperatures tend to increase in numbers (Bramley and McKinnon, 1990). Defects in milk can arise from four sources: the growth of psychrotrophic organisms prior to pasteurization, the activity of thermoresistant enzymes, the growth of thermoresistant psychrotrophic organisms, and post-pasteurization contamination. Psychrotrophic organisms, although mostly not thermoduric, are important because many produce extracellular thermostable proteolytic and lipolytic enzymes which can survive pasteurization and even ultra-high temperature (UHT) processing (Rowe and Gilmour, 1985). These extracellular enzymes hydrolyze milk proteins and lipids and cause offflavors in UHT milk. In general, psychrotrophic counts in excess of 1 x 10* colony forming units (cfu) /ml are required to produce defects on quality (Cousin, 1982). 1

PAGE 18

2 The dairy industry needs a simple, rapid, sensitive, reliable, and economical method for assessing psychrotrophic organism populations in raw and pasteurized milk. A fast, non-complex, and inexpensive method for reliable detection of psychrotrophic organisms will be a valuable quality control tool for the dairy industry. Each type of bacteria has a ‘signature’ of volatile products that form a unique odor. The method of sensing the bacteria can be to ‘smell’ the bacterial metabolites by the use of sensor arrays, also known as electronic noses. Due to its sensitivity, the electronic nose has a great potential in microbiological analysis. Six types of bacteria {Clostridium perfringens, Proteus, Haemophilus influenzae, Bacteroides fragilis, Oxford Staph, and Pseudomonas aeroginosa) were discriminated by using a 4-element metal oxide sensor array. Escherichia coli and Staphylococcus aureus were also discriminated (Gardner and Craven, 1996). Penicillium species, which produce different kinds of volatile metabolites, were separated by an electronic nose (Olsson et al., 1995). Electronic noses have many applications in the food industry, such as monitoring the deterioration of shrimp (Luzuriaga and Balaban, 1999b), tuna (Newman, 1998), and ground meat (Winquist et al., 1993). Therefore, electronic noses have the potential to detect and discriminate microorganisms based on volatile bacterial metabolites.

PAGE 19

CHAPTER 2 LITERATURE REVIEW Milk Quality Assessment Milk is an important element of a balanced diet (Muir, 1990). The major nutritional components of milk and their normal concentrations are 87.3% water, 4.6% lactose, 4.2% fat, 3.25% protein, and 0.65% minerals. It is a good source of B vitamins and minerals such as iron, copper, cobalt, and molybdenum (Frank, 1997). Indeed, milk is not only an excellent food for humans, it is also an ideal medium for the growth of microorganisms (Muir, 1990). Carbon sources in milk are lactose, protein, and fat. Many microorganisms cannot utilize lactose, and therefore proteolysis or lipolysis must occur for them to obtain carbon and energy. However, some spoilage microorganisms may oxidize lactose to lactobionic acid. The amount of lactose present in milk is enough to support extensive microbial growth. The citrate in milk can be used by many microorganisms, but the amount is not sufficient to support significant growth. There is sufficient glucose in milk to initiate the growth of some microorganisms (Frank, 1997). Two types of proteins, casein and whey proteins, are present in milk. Caseins are found in the form of highly hydrated micelles and are readily susceptible to microbial 3

PAGE 20

4 proteolysis. Whey proteins (P-lactoglobulin, a-lactalbumin, serum albumin, and immunoglobulin) remain soluble in the milk after casein precipitates. In contrast to casein, whey proteins are less susceptible to microbial proteolysis (Frank, 1997). Milk has significant fat content, but only few spoilage microorganisms utilize this fat as a carbon or energy source. The fat is in the form of globules surrounded by a protective membrane composed of glycoproteins, lipoproteins, and phospholipids. Unless the globule membrane is physically damaged or enzymatically degraded, it cannot be utilized by microorganisms (Alkanhal et al., 1985). Milk Spoilage From a milk spoilage perspective, psychrotrophic microorganisms are the single most important group. They are defined as bacteria that grow at 7°C or below, regardless of their optimal growth temperature. These organisms have constituted an important part of the microbial flora of raw milk since the introduction of bulk refrigerated storage. Growth of psychrotrophic microorganisms in raw milk can lead to quality and flavor defects in products made from that milk because of the residual activity of degradative heat-stable enzymes produced by these microorganisms (Bramley and McKinnon, 1990; Frank, 1997). Milk, produced at ambient temperatures without refrigeration, must be cooled to about 3-5°C at the farm. The initial microflora, the numbers, and types of microorganisms in milk immediately after production, reflect microbial contamination during production. The cooling after production inhibits general growth of bacteria. The temperature to

PAGE 21

5 which raw milk has been cooled, the duration of milk storage and the storage temperature on the farm can affect the numbers and types of microorganisms present in raw milk (Bramley and McKinnon, 1990). Once milk leaves the farm, active refrigeration stops, and the temperature of the milk rises by at least 1°C per day. Any temperature rise will enhance the growth of psychrotrophic microorganisms (Muir and Phillips, 1984). The psychrotrophic microorganisms in milk produce a range of extracellular enzymes which can readily degrade milk constituents. These microorganisms have either lipolytic or proteolytic or combined degradative ability. Even though psychrotrophic organisms can be killed by pasteurization or ultra-high temperature (UHT) process, their enzymes cannot be inactivated (Bramley and McKinnon, 1990). Growth of psychrotrophic microorganisms in milk can lead to spoilage because of the heat-stable degradative enzymes of these organisms. Consumers can detect these quality and flavor defects. There are three main sources of microbial contamination of milk during production: from within the udder, from the exterior of the teats and udder, and from the milking and storage equipment (Bramley and McKinnon, 1990). Soil, water, animals, and plant materials are sources of psychrotrophic microorganisms found in milk. The exterior of the teats and udder can harbor high levels of psychrotrophic bacteria, even after washing and sanitizing. Water used on the dairy farm usually contains low populations of psychrotrophs; its use to clean and rinse milking equipment provides a direct contamination into milk.

PAGE 22

6 Psychrotrophs isolated from water are often very active producers of extracellular enzymes, and grow rapidly at low temperatures (Cousin, 1982). Plant materials used for animal feed have been found to contain 10* psychrotrophic organisms/g (Thomas, 1966). Milking and storage equipment are also major sources of psychrotrophic contamination of raw milk. Proper cleaning and sanitizing procedures can reduce contamination from such equipment. Good handling practices inhibit contamination and prevent high microbial counts, and the possible presence of undesirable bacterial enzymes (Bramley and McKinnon, 1990). Raw Milk The main psychrotrophic microflora present in raw milk are aerobic Gram-negative rods W\Xh Pseudomonas spp. {P. fluorescens, P. putida, P.fragi, P. aeroginosd) forming 65 to 70% of the genera. P. fluorescens predominates (Bramley and McKinnon, 1990; Garcia etal., 1989; Reinheimer et al., 1990). Achromobacter, Acinetobacter, Aeromonas, Alcaligenes, Chromobacterium, Flavobacterium, and coliforms form the other genera present in milk (Bramley and McKinnon, 1990; Mikolajcik, 1979). Some Gram-positive bacteria are also present in raw milk, but their numbers are much smaller than those of Gram-negative bacteria. Arthrobacter, Bacillus, Clostridium, Corynebacterium, Lactobacillus, Listeria, Microbacterium, Micrococcus, Sarcina, Staphylococcus, and Streptococcus are isolated from milk, and Arthrobacter and Bacillus are the most common (Bramley and McKinnon, 1990). Psychrotrophs cause an “unclean” flavor in milk, and there is a significant correlation between initial psychrotrophic counts and storage temperature of raw milk at

PAGE 23

7 both 2°C and 6°C. A decrease in storage temperature from 6°C to 2°C leads to a 75% increase in storage life of raw milk. This was concluded by measuring the time taken for the microbial count to reach 1x10® cfli/ml (Griffiths et al., 1988a). Griffiths et al. (1988b) also investigated the effect of storage of raw milk at 2°C and 6°C on the subsequent quality of pasteurized and UHT milk. Good quality pasteurized milk could be produced from raw milk stored at 2°C for up to 5 days, but this quality could be achieved with the raw milk samples which had been stored at 6°C for only 2 days. The quality of UHT milk was much higher in products manufactured from raw milk stored at 2°C for 4 days than in those stored at 6°C for 4 days (Griffiths et al., 1988b). Aerobic spore-forming bacteria can have frequent changes in minimum growth temperatures. Grosskopf and Harper (1974) stated that isolated Bacillus spp. subsequently lost their ability to grow at 7.2°C or below when they were stored at 21°C. However, mesophilic strains of Bacillus spp. have been adapted to grow at low temperatures by repeated transfers to media at colder temperatures over a long period of time. Even though the majority of spore-forming organisms previously reported in milk were identified as Bacillus spp., anaerobic sporeformers are also present in milk. Micrococci and Microbacterium spp. are derived almost exclusively from milking equipment, and thermoduric counts in the milk sometimes exceed 5x1 O'* cfu/ml. Most of the thermoduric organisms do not multiply appreciably in raw milk even at ambient temperatures, however, a high thermoduric count in milk is reliable evidence of cross contamination from milking equipment (Bramley and McKinnon, 1990).

PAGE 24

8 The presence of coliforms and E. coli in raw milk is evidence partly of direct fecal contamination, and partly inadequate cleaning of milking equipment since coliforms can rapidly build up in moist, milky residues in milking equipment, and become the major source of contamination. Coliform counts higher than 100 cfli/ml are considered as evidence of unsatisfactory production hygiene. Unrecognized coliform mastitis may also cause high coliform counts (Bramley and McKinnon, 1990). All important spoilage bacteria found in raw milk have the potential to produce extracellular degradative enzymes, heat-resistant lipases, and proteinases, regardless of growth conditions. When they are produced, they are not destroyed by simple heat treatment. These enzymes can play a role in the quality degradation of milk. The fluorescing pseudomonads and bacteria in the genus Alcaligenes show the highest incidence of degradative action. P. flourescens and P. fragi are the major microorganisms causing lipolytic spoilage. It was found that P. fragi strains were more lipolytic than those of P. fluorescens, and grew faster at 5°C (Shelley et al., 1987). Grampositive organisms show little proteolysis and only limited lipolytic activity. The strains of psychrotrophic Bacillus spp. are frequently only proteolytic, and their proteinases and lipases are less heat-stable than those from Pseudomonas spp. (Ewings et al., 1984). Pseudomonas fluorescens and Bacillus cereus also produce extracellular phospholipase C which can degrade the milk fat globule membrane leading to product defects. The phopholipase C from Bacillus cereus produces “bitty” cream, and exhibits similar effects in pasteurized milk.

PAGE 25

9 The production of protease and lipase by psychrotrophic bacteria is not observed until all the cells are in the late exponential or stationary phase of their grovvth cycle. Such a stage with most cultures of dairy origin is reached when the cell population exceeds 10^ cfli/ml. This concentration of microorganisms corresponds with the onset of detectable spoilage in raw milk (Muir et al., 1978). A reduction in oxygen tension of the growth medium immediately precedes protease synthesis by a strain of Pseudomonas fluorescens (Rowe and Gilmour, 1982). Griffiths and Phillips (1984) reported that protease synthesis by psychrotrophs in milk could be inhibited by maintaining high oxygen concentrations. However, other researchers stated that this effect may be strain specific, since they found that some bacteria continue to excrete both protease and lipase when the oxygen tension in the growth medium is high (Dring and Fox, 1983; Fox and Stepaniak, 1983). The effect of growth temperature on the production of proteases and lipases by psychrotrophic bacteria may also be strain specific. Growth temperature of 2°C depressed the synthesis of both protease and lipase by several pseudomonads (Griffiths et al., 1988a). This result shows that deep cooling of milk to 2°C extends its shelf life. Pasteurized Milk Pasteurization ensures the microbiological safety of milk as well as commercially acceptable shelf life (Muir, 1990). However, increasing the pasteurization temperature does not necessarily result in an increased shelf life because of the activation of spores at the higher temperatures (Phillips and Griffiths, 1990).

PAGE 26

10 In pasteurization, milk is heated to and retained at a temperature not less than 62.8°C, and not more than 65.6°C for at least 30 min, then immediately cooled to below 10°C or below 6°C. Alternatively, the milk may be retained at a temperature of not less than 71 ,7°C for at least 15 sec, then immediately cooled to less than 10°C. This process is called the high-temperature short-time method (HTST). Another method, super pasteurization, is used in many places today. Milk is heated to and retained at a temperature of 82°C for 3 sec, then immediately cooled to 10°C (Bramley and McKinnon, 1990; Vela, 1997). Pasteurization of milk was designed as a method of heating milk to kill Mycobacterium tuberculosis, the organism that causes tuberculosis in humans. However, Coxiella burnetii, the organism that causes Q fever, is found to be more resistant to heat than M. tuberculosis. Now the target organism for pasteurization is C. burnetii. The effectiveness of pasteurization is monitored by measurement of the residual level of alkaline phosphatase (Vela, 1997). Microbial contamination of pasteurized milk occurs through post-pasteurization contamination by psychrotrophs and through survival of thermoduric psychrotrophic organisms during pasteurization. Many of the Gram-negative organisms do not survive pasteurization, in fact, only one species of Gram-negative bacterium, Alcaligenes tolerans, can survive since it is generally considered as thermoduric (Muir, 1990). Spore-forming bacteria of the genus Bacillus form the most important group of microorganisms capable of surviving pasteurization, and growing in milk. The occurrence of psychrotrophic, spore-forming bacteria in pasteurized milk was first reported in 1969.

PAGE 27

11 The quality loss and unpalatability of pasteurized milk (stored at 4°C for four weeks) was observed due to the outgrowth of Bacillus coagulam (Grosskopf and Harper, 1974). They also isolated Bacillus coagulam from pasteurized milk that had been stored at 2°C for 13 to 17 days. They stated that the generation time for B. coagulam was 24 to 30 hrs under these refrigeration conditions. A seasonal variation in the numbers of the genus Bacillus in raw milk is observed, and these bacteria predominate in the period of June to October. B. licheniformis is the most common species (Bramley and McKinnon, 1990). During this period there is a peak in the number of spoilage incidents in milk with bitter taint, or with the defect known as “bitty” cream (Muir, 1990). The only pathogen of the genus Bacillus is B. cereus, and it can survive, and grow at refrigeration temperature (Bramley and McKinnon, 1990). Researchers in the US reported that psychrotrophic sporeformers were isolated from 28-35% of freshly pasteurized milk (Ahmed et al., 1983). Others observed heat activation Bacillus spores at pasteurization temperatures, and stated that more than 95% of spores could be activated at pasteurization temperatures (Phillips and Griffiths, 1990). Coryneform bacteria which may form a substantial proportion of the flora of the heat-treated milk grow very slowly at refrigeration temperatures (Seiler et al., 1984). Streptococcus spp. such as Streptococcus thermophilus, Enterobacter faecalis, and Streptococcus brevis which are thermoduric grow very slowly at refrigeration temperatures. Therefore these bacteria pose no great threat to pasteurized products which

PAGE 28

12 are adequately refrigerated between heat treatment and the ultimate consumer (Muir, 1990). The quality of pasteurized milk improves with the reduction in levels of postpasteurization contamination. The majority of the psychrotrophic bacteria are destroyed by pasteurization, and only thermoduric flora, and extracellular enzymes from psychrotrophic bacteria are left to cause spoilage. The thermostability of proteases, lipases, and phospholipase C of bacterial origin to HTST pasteurization is shown in Table 2 1 . Table 2-1. Residual enzyme activity from psychrotrophic organisms in HTST pasteurized milk. Type of enzyme Activity after HTST pasteurization (%) Protease 66 Lipase 59 Phospholipase C 30 Source: Muir (1990). About 60% of the protease and lipase activity of the psychrotrophic bacteria remains after HTST treatment, and a lower, but significant, proportion of the phospholipase C activity remains. This thermostability pattern is observed for the proteases of most fluorescent and non-fluorescent Pseudomonas spp. and for enzymes from many strains of Alcaligenes, Acinetobacter, Achromobacter, Enter obacteriaceae. and Moraxella.

PAGE 29

13 Many strains of Bacillus spp. are heat resistant, but not their enzymes. Even though growth temperature has a significant effect on the rate of bacterial growth, and on the synthesis of extracellular enzymes, when they are excreted, they exhibit similar degree of thermostability. Psychrotrophic bacteria produce thermostable enzymes almost regardless of species, state, and temperature of growth. The thermostability of the extracellular enzymes of the psychrotrophic bacteria in milk has a minor importance for short shelf life products such as pasteurized milk because the shelf life of pasteurized milk is usually about 14 to 20 days, and during this period the products are refrigerated. The initial bacterial counts in raw milk for heat treatment is well below the threshold of 10’ cfu/ml where significant amounts of extracellular degradative enzymes are synthesized (Frank, 1997; Muir, 1990). UHT Milk In sterile or UHT milk, the important point is to destroy bacteria while limiting the chemically-induced changes such as browning, cooked, and caramelized flavors. UHT milk is either heated in a closed container to yield sterilized milk, or it is heat-treated in a continuous flow, and packed aseptically. The UHT milk is heated at 138-142°C for 2 to 5 sec. The test for sterilized or UHT milk is that samples incubated at 30°C for 15 days will have a plate count of not more than 100 cfu/ml (Bramley and McKinnon, 1990). The spore-forming bacteria are the organisms relevant to spoilage in sterilized products, but many spores are destroyed by the heat-treatment applied to sterilized or UHT milk. Spores o^ Bacillus stearothermophilus constitute the greatest hazard to

PAGE 30

14 spoilage of sterilized dairy products since they are remarkably heat-resistant, and require extended heat treatments to ensure a reasonable kill (Muir, 1990). Residual activity of the extracellular enzymes from psychrotrophic bacteria after UHT sterilization may be as high as 40% of that in the raw milk. The residual activity of protease, lipase, and phospholipase C are given in Table 2-2. Muir (1996) measured the residual proteinase and lipase activity found after treating cell free supernatants at 140°C for 5 sec, and found that Acinetobacter, Aeromonas, and Bacillus spp. had residual activities below 10%, and fluorescent pseudomonads had residual activity ranging between 14 and 51%. There could be degradation of milk protein and fat in the case of a product with a shelf life typically of six months at room temperature (Muir, 1990). Table 2-2. Residual activity of extracellular enzymes of psychrotrophic organisms after UHT sterilization. Residual activity after heat treatment (%) Enzyme type 140°C for 5 sec 140°C for 5 sec + 55°C for 1 hr Protease 29 17 Lipase 40 7 Phospholipase C 0-57 n.d. n.d. = not determined Source: Muir (1990). Milk containing different concentrations of a psychrotrophic, proteolytic Pseudomonas spp. was sterilized, and even though their UHT products were sterile, proteolysis and gelation of the milk occurred. The initial psychrotrophic count affected the shelf life before the onset of gelation. When the microbial count was less than 5x10*

PAGE 31

15 cfli/ml, there was little evidence of an effect on the product quality. However, when this threshold was reached, the decline in shelf life progressed rapidly. There were similar findings when skim milk with counts in excess of 5x10® cfu/ml was UHT sterilized (142°C/2 sec). Premature gelation and bitterness occurred within three months of storage at room temperature. The evidence of similar defects can be seen in commercial products from time to time (Muir, 1990). Similar results were observed in the samples from dairy plants and farms, and 70 to 90% of the samples contained heat-resistant proteases. After heat treatment at 149°C for 10 sec, these enzymes were still active. It was found that for UHT-treated milk to have a shelf life of 1 year, the raw milk must contain less than 0. 1 unit of protease/ml, and this amount can be easily synthesized within a day for some highprotease-producing bacteria. If casein and whey proteins have been damaged by enzymes of Grram-negative psychrotrophs, they have a predisposition to denaturation and precipitation by UHT treatment. Gelation of UHT milk can occur as a result of the proteolytic activity of these enzymes (Adams et al., 1981). Lipases can cause fat degradation in UHT products. Aeration affects lipase production by Pseudomonas spp. For P.fragi, aeration reduced lipase production, however, a high aeration was required for high lipase activity of P. mepUtica var lipolytica (Champagne et al., 1994). Milk Volatiles and Off-Flavors During microbial spoilage, bacteria produce metabolites which cause off-odors. Each type of bacteria has a ‘signature’ of volatile products that form a unique odor. These

PAGE 32

16 odors can be qualified and/or quantified by humans and instrumental methods such as gas chromatography. The possibility of characterizing some bacteria by gas chromatographic analysis of headspace vapors from milk cultures was investigated. Bassette et al. (1967) reported the characteristic patterns and rates of development of acetaldehyde, ethanol, dimethyl sulphide, and diacetyl in cultures of Aerobacter aerogenes, Escherichia coli. Streptococcus faecalis var. liquefaciem, Achromobacter lipolyticum. Streptococcus diacetylactis, E. freundii, Sterptococcus lactis, Lactobacillus acidophilus, Lactobacillus casei, and Pseudomonas fragi. Dimethyl sulphide, acetaldehyde, 2-methylpropanal, acetone, ethanol, 3-methylbutanal, butanone, 2-methylpropanall-ol, and 3-methylbutanl-ol in milk cultures o^ Streptococcus lactis var. maltigenes were reported (Morgan et al., 1966). Volatile compounds such as esters, ethanol, propan-2-ol are produced in refrigerated milk by pseudomonads, and increase in ethanol could be detected when the level of microflora in pasteurized milk stored at 4-10°C reached 10^ cfu/ml (Urbach and Milne, 1987). The number of psychrotrophs required to produce off-flavors varies between species, and is determined by the length of the lag period and growth rate at the storage temperature, and by the proteolytic activity and heat resistance of the enzymes. In the case of Pseudomonas spp. 2.7 x 10® to 9.3 x 10’ cfu/ml and Alcaligenes spp. 2.2 x 10® to 3.6x10’ cfu/ml were required to produce off-flavors. In general, microbial counts in the excess of 1 x 10® cfu/ml are required to produce defects on product quality (Cousin, 1982).

PAGE 33

17 Serious ofF-flavors such as bitter, putrid, unclean, stale, rancid, fruity, yeasty, and sour have been frequently associated with the presence of thermoduric psychrotrophs in milk. Isolated Bacillus spp. were inoculated into sterile milk at a level of 0.5%, and incubated at 7.2°C. Their flavor associated defects are shown in Table 2-3. Table 2-3. Flavor defects associated with psychrotrophic Bacillus spp. grown in milk at 7.2°C. Species Days to appearance of defect Flavor defect B. macerans 4-10 Fruity, sour B. polymyxa 4-6 Sour, yeasty, gassy B. laterosporus 4-12 Sweet curdling, unclean, bitter B. subtilis 2-4 Sweet curdling, bitter B. lentus 6 Sour B. cereus 6 Sweet curdling, bitter B. sphaericus 12 Unclean, sour Source; Washam et al. (1977). The occurrence of a bitter flavor in milk is related to the presence of heat stable protease. Hydrolysis of lactoalbumin and casein yields bitter peptides. When 300 ng of heat stable protease from P. fluorescem was mixed with pasteurized milk having low bacterial count, bitterness developed after storage of 7 days at 7°C (Baker, 1983). Lipases produced by thermoduric psychrotrophs produce off-flavors in milk. Free fatty acids give rise to off-flavors such as rancid, butyric, bitter, unclean, soapy, and astringent (Champagne et al., 1994). The fruity flavor arises from free fatty acid esterification. Some thermoduric psychrotrophs produce phopholipases, particularly phopholipase C which is responsible for a specific degradative action on the fat globule

PAGE 34

18 membrane. This increases the susceptibility of exposed milkfat to the action of lipases (McKellar, 1989). Thermoduric psychrotrophs were inoculated (200 to 100 cfii/ml) into sterile whole and skim milks. Sensory defects were observed in the populations of 3 to 4 x 10® cfli/ml. This cell population was reached within 6 days at 7.2°C. Obvious physical and flavor defects were observed when populations of 5 to 20 x 10® cfu/ml in milk were held at 6 to 20°C (Phillips and Griffiths, 1990). The optimum temperature for the production of degradative enzymes by microorganisms is usually lower than the optimum temperature needed for cell division. Even though the microbial population might remain below that normally associated with formation of microbial defects, microbially produced enzymes develop off-flavors when milk is held at refrigeration temperatures for extended periods. Factors Affecting Shelf Life of Milk The “shelf life” or “consumable life” is defined as the period between processing/packaging, and when milk becomes unacceptable to consumers (Bishop and White, 1986). The product needs to remain acceptable beyond the last date of sale. The shelf life of milk is influenced by the quality of raw milk, milk handling, milk processing, and storage temperatures, product processing procedures, and postpasteurization contamination. Post-pasteurization contamination is most detrimental for keeping quality of pasteurized milk (Stepaniak, 1991). Bacterial growth is responsible for the spoilage of milk, and a level of log 7.5 cfli/ml represents the end of shelf life (Griffiths et al., 1984). Spoilage organisms cause

PAGE 35

19 biochemical changes in the substrate. The amount of substrate utilized, and product formed is proportional to the number of cells present. The substrate is stated as organoleptically spoiled. When raw milk has a high population of psychrotrophic bacteria (>10* cfu/ml), the UHT products obtained from it usually have a reduced shelf life. Collins et al. (1993) investigated the influence of the growth of psychrotrophic bacteria in raw milk on the acceptability and resultant shelf life of ultra-high temperature (UHT) processed skim milk. The psychrotrophic counts in the raw milk were highly correlated with the extent of proteolysis (r = 0.95), however, not with the extent of lipolysis (r = 0. 1 8) in the stored UHT milk. It was also found that storage time, and storage temperature had a greater influence on the sensory acceptance, and resultant shelf life of the UHT skim milk than the psychrotrophic count of the raw milk since correlations between the extent of proteolysis and bitterness scores in the stored UHT milk were high at 30°C (r = 0.92) and at 40°C (r = 0.90), but not at 20°C (r = -0.23). Shelf life prediction could be made by performing microbial enumeration methods, based on preincubation at 12-21°C with or without selective inhibitors, and the results could be obtained after 2-3 days. On the other hand, Moseley keeping quality test (7-day count) can be performed, but the result can be obtained after 7-9 days. Due to these time constraints, there is a need for a shelf life test which could provide results in a short period of time, be accurate, simple, and economical to perform (Bishop and White, 1986).

PAGE 36

20 Rapid Methods for the Detection of Microorganisms The dairy industry experiences increased demands on manufacturing and distribution efficiencies. Food safety and quality management are crucial in production and distribution. In order to avoid the sale of contaminated products, expensive inventories are held at the production site while samples are tested for microbial contamination, which often takes more than 3 days. Since the products have a short shelf life, they are released before microbial results are available. Rapid detection of pathogens, spoilage microorganisms, and other microbial contaminants in dairy products is important to ensure the safety of consumers and quality of foods. Recent developments make microbial detection and identification faster, more convenient, more sensitive, and more specific than conventional assays. The most frequently used rapid methods in industry are immunomagnetic separation (IMS), enzymelinked immunosorbent assays (ELISA), impedance (or conductance), bioluminescence, miniaturized methods, and other biochemical methods (FDA Bacteriological Analytical Manual, 1995). These methods may be used on their own or in combination. Rapid methods have been developed either to replace the enrichment step which requires a prolonged growth period with a concentration step (such as in immunomagnetic separation) or to replace the end-detection method which is colony development that requires a prolonged incubation period (such as in impedance microbiology and bioluminescence).

PAGE 37

21 Immunomagnetic Separation TIMSl A separation step is normally required in order to discriminate the target organism from other cells. Superparamagnetic particles are used in immunomagnetic separation. They exhibit magnetic properties in the presence of an external magnetic field. These are coated with antibodies against the target organism to isolate the organism selectively. Food poisoning bacteria can be magnetically separated from foods depending on availability and specificity of appropriate antibodies, but labeled magnetic particles are not available commercially. V. haemolyticus K serotype from food and from a patient (Tomoyasu, 1992) and Yersinia enterocolitica from spiked food and water samples (Kapperud and Vardund, 1995) were isolated using IMS. Encapsulated S. aureus can also be isolated from milk using IMS (Johne et al., 1989). Impedance Microbiology Impedance microbiology detects microorganisms either directly due to the production of ions from metabolic end products or indirectly from carbon dioxide production. The direct method monitors changes in impedance of the growth medium. Microorganisms produce ionic end products such as organic acids and ammonium ions from the growth medium, and increase the conductivity of the medium (Silley and Forsythe, 1996). In the indirect method a potassium hydroxide bridge (solidified in agar) is formed across the electrodes. The sample is separated from the potassium hydroxide bridge by a

PAGE 38

22 headspace. Carbon dioxide accumulates in the headspace during microbial growth. This dissolves in the potassium hydroxide. Decrease in conductance occurs since the resultant potassium carbonate is less conductive. This method is applicable to a range of microorganisms including S. aureus, L. monocytogenes, E. faecalis, B. subtilis, E. coli, P. aeruginosa, A. hydrophila, and Salmonella serovars (Bolton, 1990). This indirect technique is also most appropriate to detect C. tyrobutyricum which is a spoilage organism of high-pH cheeses due to its transformation of acid to carbon dioxide, hydrogen, and butyric acid (Druggan et al., 1993). Impedance microbiology has also been used to monitor the stability of lactic acid bacteria starter cultures in the dairy industry (Lanzanova et al., 1993). Enzyme Immunoassays and Latex Agglutination Tests The enzyme immunoassay (EIA) or enzyme-linked immunosorbent assays (ELISA) are other techniques used in food microbiology. ELISA is performed using monoclonal antibodies coated microtitre trays to capture the target antigen. The captured antigen is detected using another antibody which may be conjugated to an enzyme. The presence of the target antigen is visualized by the addition of a substrate (Forsythe and Hayes, 1998). These methods have been developed for the detection of specific pathogens, toxins, and enterotoxins, antibiotics, drugs, and pesticide residues. Some are designed to detect specific organisms such as Salmonella enteritidis or Listeria monocytogenes from food or environmental samples (Vasavada, 1997). The technique generally requires the

PAGE 39

23 target organism to be more than 10® cfli/ml. The conventional pre-enrichment and even selective enrichment might be needed prior to testing. Reverse phase latex agglutination (RPLA) can be used to detect Staphylococcal enterotoxins. Latex particles coated with specific antisera to the enterotoxins, each on separate particles, are used in the Oxoid RPLA kit (Denka Seiken Co. Ltd.). The sensitivity limit is about 0.5 ng enterotoxin/g food (Forsythe and Hayes, 1998). A number of enzyme immunoassays such as Tecra ELISA (Tecra Diagnostics) and VIDAS ELISA (bioMerieux) are available for staphylococcal enterotoxin detection. The detection limit for Tecra ELISA is less than 0.5 //g toxin/ 100 g food, and requires 7 hrs to obtain the results. Even though the latex agglutination test RPLA has a similar detection limit, it requires 21 hrs to obtain the results (Forsythe and Hayes, 1998). The Vidas (bioMerieux) system contains predispensed disposable reagent strips. The target organism is captured in a solid phase receptacle coated with primary antibodies, and transferred to the appropriate reagents automatically. This system can be used to detect most major food poisoning organisms (Forsythe and Hayes, 1998). Bioluminescent Systems Bioluminescent systems measure the presence of adenosine triphosphate (ATP) in a sample using an enzyme system, luciferin-luciferinase, from fireflies. The amount of light generated by this enzymatic reaction can be measured in a suitable luminometer, and is directly related to the ATP extracted, and thus to the number of microbial cells from which it came (Stanley et al., 1989).

PAGE 40

24 The ATP-bioluminescence method is used to determine raw milk quality (Van Crombrugge et al., 1989). The detection level is approximately 1x10’ cfu/ml of raw milk. Predicting the shelf life of pasteurized milk can also be achieved using ATPbioluminescence after preincubating the milk at 15°C for 25 hrs or at 21°C for 25 hrs in the presence of crystal violet, penicillin, and nisin to inhibit the growth of Gram positive organisms (Bautista et al., 1992). Bioluminescence assays often take just a few minutes to accomplish, but the disadvangates of this system are that the method measures all the microbes present in the sample, it is not presently possible to selectively measure ATP from one microbial species in the presence of many species (Stanley et al., 1989). Miniaturized Methods Conventional methods for isolation and characterization of microorganisms, especially pathogens, entail the use of special enrichment and cultivation, selective and differential media, and a wide range of biochemical tests. Recently, miniaturized methods, diagnostic kits, and sophisticated instruments have been developed that allow the rapid identification of foodbome pathogens. These methods show an improvement over a conventional test as well as savings in time (Adams and Hope, 1989). The API test (bioMerieux Vitek), Enterotube (Roche Diagnostic), Micro-ID system (OrganonTeknika), and Minitek, and Crystal Sytem (BBL Microbiology Systems) are examples of miniaturized kits currently available for use in the food industry. These systems are

PAGE 41

25 convenient, efficient, economical, and easy to use. They are also 90-95% accurate when compared to conventional methods (Vasavada, 1997). Other Biochemical Methods Other biochemical methods include those based on the presence of the lipopolysaccharide of a Gram-negative bacteria (LAL, or limulus amoebocyte lysate) and specific enzymes, such as catalase. The LAL test is a simple, rapid, and sensitive method with applications for rapid screening of Gram-negative spoilage bacteria in milk, meats, fish, turkey, and food ingredients (Vasavada, 1997). This test does not measure the Gram-positive bacteria, so some techniques or devices are needed to relate the LALdetermined Gram-negative bacteria to “total” bacterial number that may be determined by the ratio that exists between Gram-negative and Gram-positive bacteria (Adams and Hope, 1989). This will increase the cost of the experiment, takes time, and relies on prediction. In spite of all of these methods, new rapid detection techniques which provide reliable and accurate results in a short period of time that will allow for the performance of effective, corrective measures are needed. In addition to these, a test should be simple and fairly economical. Use of electronic noses may possibly accomplish this. Since results will be given in less than 10 min right after the incubation is completed for a specified time, there is no need for a qualified person to run the experiment, the cost of the experiment will be minimized and small sample size, which is the most important point for some of the food samples, will be enough to perform the experiment.

PAGE 42

26 Electronic Nose In the food industry, flavors are important from the raw ingredients to the final product. There are two components of flavor perception: taste and aroma. Taste forms from the presence of nonvolatile compounds that interact with sensors in the mouth and on the tongue, and appears as the basic tastes of sweet, sour, salty, and bitter. Although taste is important, the flavor of a food cannot be defined by taste alone. Many volatile compounds that are responsible for the aroma of a food play an important role in flavor. These volatile compounds contribute to the nature of a food, its product identity, and to consumer preferences between brands. They are also responsible for the occurrence of off-flavors and taints, which arise because of biochemical or chemical changes, microbial growth or contamination (Hodgins, 1997). There are three sensory systems in humans that are responsible for the sensation of flavor. These are gustation (sense of taste), olfaction (sense of smell), and the trigeminal sense (responsive to irritant chemical species) (Gardner and Bartlett, 1994). The sensation of smell depends on the interaction of odor molecules with a group of specialized nerve cells. The odor molecules go into the nasal cavity and across the olfactory area or epithelium. These molecules dissolve in an aqueous mucous layer covering the olfactory receptor cells. The olfactory receptor cells located at surface of the olfactory hairs or cilia have receptor binding proteins that bind with the odorous compounds (Breer, 1994; Clapham, 1996; Pearce, 1997). The number of olfactory

PAGE 43

27 receptor cells is large (about 100 million); however, the number of distinct types of binding proteins is small (about 1,000). The same protein has to be found in many different olfactory cells (Bartlett et al., 1997). The receptor cells amplify the signal, and transmit it to the brain by means of the axons. The brain compares it with previous knowledge, and tries to define the odor. Humans can detect a minimum of 10,000 odors, but the number of identifiable odors is approximately 50 (Bartlett et al., 1997). The specific combination of complex mixtures of many odorous molecules having different concentrations culminates in the recognizable flavors or odors. Odor molecules are generally small (molecular weight 20300 Daltons) and polar. They can be detected by humans below 1 ppb. Gas chromatography linked with mass spectrometry (GC/MS) is used to detect complex odors at low levels, but the sample must be separated into its individual components to identify each odor. GC/MS is expensive, and requires a technician for operation and interpretation of the results. Due to these constraints, sensory analysis has been used for a long time for odor evaluation. Sensory analysis has restrictions also: panelists may not be sensitive to some flavors, some raw materials may be difficult to assess using panelists, training needs to be performed before the analysis, sensory analysis can be expensive, and the panelists are subject to fatigueness (Bartlett et al., 1997; Hodgins and Simmonds, 1995). A single molecule can have a distinct odor. However, most natural smells or flavors are a complex mixture of chemical species, and contain hundreds of constituents (Dodd et al., 1992). True aroma is related to the complex interaction of all volatile

PAGE 44

28 compounds within foods. For example, using a gas chromatograph with a sniffer port, none of the individual aromatic compounds present in cocoa smell like cocoa to humans; however, when all of these compounds combine, the overall aroma is that of cocoa (Hodgins and Simmonds, 1995; Hodgins, 1997). Due to the limitations of GS/MS and sensory analyses, there has been a need to develop an instrument that can mimic the human sense of smell, and provide rapid sensory information at low cost. Electronic Nose Technology In 1961, Moncrieff started to develop an instrument to detect odors. In 1965, several researchers published studies of the redox reactions of odorants at an electrode, modulation of electrical conductivity, and contact potential by odorants. The concept of an electronic nose as a chemical array sensor system for odor classification was presented for the first time by Persaud and Dodd in 1982. Today, the electronic nose has various synonyms such as artificial nose, mechanical nose, odor-sensing system, and sensor array system. Gardner and Bartlett (1994) defined an electronic nose as an instrument comprised of an array of electronic chemical sensors with partial specificity, and an appropriate pattern recognition system capable of recognizing simple or complex odors. The main components of an electronic nose are sample handling mechanisms, an array of chemical sensors, signal preprocessing and conditioning, and pattern recognition techniques.

PAGE 45

29 The electronic nose technology simulates the human olfactory process with fewer sensors, and a suitable software designed to analyze the responses from the sensors. Each sensor represents a group of olfactory receptors, and produces electrical signals in response to an odor, and this electrical signal is time-dependent. Although the specificity of each sensor may be low, the combination of several specificity classes results in a very wide range of information. Any noise or sensor drift may be reduced using signal preprocessing techniques. Finally, the use of pattern recognition in the electronic nose is equivalent to the classification and memorization of odors in the brain (Gardner and Barlett, 1999). The first step in the sample handling is to obtain the vapor above a sample, and to transport it to the sensor array. Currently there are two methods, static and dynamic sampling. In static sampling the headspace above a liquid or solid is measured. The system consists of a sample vessel and a sensor head (compartment with the sensor array). The sensor array and vapor of the sample remain in separate sealed compartments. The sample headspace and sensor head are purged to eliminate any foreign odors using compressed air or any other inert gas for a certain period of time before the analysis. Once the sample has reached equilibrium, the door between the two compartments is opened and the analysis starts. When conducting polymer sensors are used, an internal DC power supply maintains a constant current through the sensors. The sensor resistance changes when the sensors get in contact with the headspace of the sample. The corresponding voltage change across the sensor is measured. The resulting analog signal is digitized, and sent to the computer. The change in conductivity of the sensors is

PAGE 46

30 acquired for a given period of time. Once the analysis is completed, the separating door is closed, and the sensor head and sample vessel are purged to get ready for the next analysis. The electronic nose works similar to a GC in the dynamic sampling procedure. A sample is placed in a closed container. Once headspace equilibration is reached, a sample of the headspace is obtained, and injected into the sampling port of the electronic nose. The sample is carried by an inert gas to the sensor array. A sensor changes its electrical properties, and sends a signal to the computer. The rest is similar to the static system. The information provided by the sensor signal is maximized by the signal preprocessing and conditioning of the analog response of the sensor. This is done by using signal conditioning circuits, potential dividers, constant voltage sources, and an analog-to-digital converter (Corcoran, 1993). Change in current or voltage is optimized for system sensitivity. Noise is reduced by modulating the sensor signal which is amplified to a suitable level. System noise is affected by variations in the electronic circuitry as well as connections between the sensors and the circuit (Hodgins, 1997). The signal conditioning digitizes the response of the sensors, and generates an output that is then analyzed with pattern recognition techniques to define the sample odors. Sensor Technology Different types of materials such as conducting polymers, metal oxides, lipid layers, phthalocyanins, and piezoelectric technologies are being used to manufacture sensors that are useful for odor detection. The types of sensors that are being used

PAGE 47

31 commercially in electronic noses are semiconductor metal oxides, conducting polymers, quartz-resonator sensors, and surface acoustic wave sensors (Bartlett et al., 1997; Hodgins, 1997). Other types of sensors that have potential or have been used are biosensors, enzyme sensors, electrolytic sensors, platinum hot wire detectors, and fiberoptic gas sensors (Shurmer, 1990). The selectivity and the sensitivity of the sensors are determined by choice of the catalytic surfaces (Gardner and Bartlett, 1999). Sensor technology is changing very rapidly and more sensitive, stable, and fast response sensors are being developed. Conducting polymers Conducting polymers used as sensor materials in electronic noses have unique electrical properties that make them suitable for gas detection. A wide range of materials can be synthesized. They respond to a broad range of organic vapors, and they operate at room temperature. The main types are poly-pyrrole and poly-anilines. The volatile compounds change the electrical conductivity of the polymer. This change occurs rapidly and reversibly at room temperature (Gardner and Bartlett, 1999). The adsorbed odor molecules are believed to cause a swelling of the polymers and to interfere with charge transfer within the polymer (Corcoran, 1993). Conducting polymer sensors are nonspecific. Different compounds will interact with the polymer material. These sensors are small, and have low power consumption since they operate at room temperature. They have quite a good sensitivity, typically between 0.1 and 100 ppm (Bartlett et al., 1997). The sensor responses are also rapid with rapid recovery of the baseline when the

PAGE 48

32 volatile compound is removed. Conducting polymers are sensitive to humidity, therefore, caution should be taken when analyzing samples with different water activities. Semiconductor metal oxide chemoresistive sensors These types of sensors are developed from chemoresistive arrays of inorganic semiconducting materials such as oxides, and catalytic metals. Two main types have been developed: thick film metal oxides, known as Taguchi sensors, and thin film, which are commonly used in commercial electronic noses. These sensors comprise a ceramic support tube containing a platinum heater coil. Tin-dioxide is coated on the outside of the ceramic support tube along with the catalytic metal additives such as palladium or platinum. As current passes through the coil, the metal oxide heats up. The reaction between the vapor and the metal oxide causes a change in electrical resistance at a fixed temperature. This resistance change can be measured, and related to odors being monitored. In metal oxides, chemisorbed oxygen [O'] reacts with the odorant [R] irreversibly to produce combined molecules [RO] and liberated conducting electrons [e'J. Electron mobility increases, and electrical conductivity of the material changes (Tan et al ., 1995). These sensors operate between 300-550°C to avoid interference from water, and to aid rapid response and recovery times (Gardner and Bartlett, 1999). They are sensitive to combustible materials, such as alcohols, but are less sensitive at detecting sulfuror nitrogen-based odors. Surface acoustic wave devices Surface acoustic wave sensors have been in research and development for 5 to 10 years (Hodgins, 1997). The principle of operation is that a surface wave is generated in a

PAGE 49

33 material that absorbs the compounds of interest. The surface wave is normally generated using a quartz resonator, and the frequency of operation is usually between 100 MHz and 1 GHz. The frequency of operation depends on the sensitivity required by the system. When the sensor is not exposed to a vapor, it will have a certain resonant frequency. When the sensor is in contact with the volatiles, there will be a change of mass in the sensor material, and therefore, a change in the resonant frequency (Balantine and Wohltjen, 1989; Hodgins, 1997). This frequency change is the response or output from the sensor to the volatiles present in the sample analyzed. These sensors have higher sensitivity than conducting polymers (Hodgins, 1997). However, they are more selective, and a larger number of these sensors are needed to cover all vapors that are likely to occur in food products. Fiber-optic gas sensors These sensors rely on the light guiding properties of the optical fiber to carry the light from the light source to the chemically sensitive layer, and then to return the light to the sensor. The optical properties measured include the optical path length, luminescence, absorption, fluorescence, and reflectance. These types of sensors have potential advantages in that individual fibers can be as small as 2 /im in diameter, and large bundles of fibers are available which permit an attractive approach to the fabrication of miniature sensor arrays; video technology can be used to measure the responses from an array; the measurements can be made remotely because the fiber allows transmission of light over long distances; and the devices are not subject to electrical interference (Gardner and Bartlett, 1999).

PAGE 50

34 In recent studies some researchers have shown that fiber-optic gas sensor arrays can be used to sense a range of organic vapors (Dickinson et al., 1996; White et al., 1996). A fluorescent dye, Nile Red, was used in these studies since the fluorescence spectrum, and intensity of Nile Red is strongly dependent on the local solvation environment of the dye molecule. The fluorescence emission from the dye changes in the presence of organic vapors that sorb into the polymer films. This change can be measured. At present, the sensitivity of these sensors is not high, and there is not enough information about the lifetime, reproducibility, or stability of fiber-optic sensors. Applications of Electronic Nose The electronic nose is used for monitoring and control of industrial processes, diagnosis in the medical field, environmental control, and for control of food quality. It is possible to classify various liquors, perfumes, tobacco brands, beers, and many more with this device. In the environmental applications, the electronic nose was used in monitoring sewage related odors. Canonical correlation was used to compare the multivariate data generated by the electronic nose (Neotronics Olfactory Sensing Equipment) with sensory panel analysis. A linear relationship can be obtained between the electronic nose data and the corresponding threshold odor numbers within the similar groups of data from the experiment (Stuetz et al., 1998). In another application, the odors from pig and chicken slurry were evaluated by using a photoionization detector and electronic nose based on polypyrrole sensors, and it was concluded that electronic nose was better at discriminating

PAGE 51

35 between different odors through the pattern of sensor responses (Hobbs et al., 1995). In this case rapid and portable devices for odor measurement may be useful since some of the major odor compounds were chemically unstable. In the paper industry, an electronic nose containing four Taguchi gas sensors, and one infrared COj sensor was used to examine the odors from five cardboard papers from commercial manufacturers. It was shown that the olfactory quality of cardboard papers could be recognized using an electronic nose (Holmberg et al., 1995). Two different electronic noses (Neotronics e-Nose 4000™ equipped with conducting polymer sensors and Alpha M.O.S. Fox 3000™ equipped with metal oxide sensors) were used to distinguish between different concentrations of two final flavor mixtures, ten different methanol samples, and nine tobacco samples. Both instruments distinguished between different concentrations of the same flavor for two different flavors for a given day. Ten methanol samples were distinguished from each other for a given day based on sensory analysis. Nine tobacco sample results showed discrimination between each other that were classified as good, borderline, or bad using both instruments (Robie, 1997). Applications in Microbial Detection Due to the sensitivity of the electronic nose, it has great potential in microbiological analysis. Gardner and Craven (1996) reported on the use of a 4-element metal oxide system to discriminate between 6 types of bacteria {Clostridium perfringens, Proteus, Haemophilus influenzae, Bacteroides fragilis, Oxford Staph, Pseudomonas

PAGE 52

36 aeruginosa). An array of 4 Taguchi series-8 gas sensors was used to sample the headspace of the 6 different bacterial samples and a control of blood agar. The data was used to train a neural network by the backpropagation method using over 1,000 iterations with a set of 42 odor vectors. The results were reasonably encouraging with 64.3% of test vectors being correctly classified. Two bacteria types, Escherichia coli and Staphylococcus aureus, were also examined. A backpropagation network was trained on the 90 odor vectors and in 87% of the test vectors the age of the bacteria were correctly classified. Another backpropagation network was trained on 180 vectors to predict the bacteria type, and in 91% of the 180 test vectors the type of bacteria of any age was correctly classified (Gardner and Craven, 1996). Electronic nose was successfully used to separate Penicillium species which produces different volatile metabolites (Olsson et al., 1995). The electronic nose and neural network classifier were also used to detect and simultaneously identify pure plate cultures of a range of microorganisms. The overall classification rate for 1 2 different bacteria {E. Coli, Pseudomonas aeruginosa, Citrobacter freundii, Enterobacter aerogenes. Bacillus cereus, Klebsiella aerogenes. Staphylococcus aureus. Staphylococcus epidermis. Salmonella reading. Salmonella poona. Salmonella garinarium. Bacillus subtilis), and one pathogenic yeast {Candida albicans) was 93 .4%. Three similar yeast cultures were also compared, and the correct classification rate was 96.3%. Principal component analysis gave good discrimination between water vapor and the test organisms (Gibson et al., 1997).

PAGE 53

37 In medical applications the electronic nose is used to scan wounds to diagnose infections and monitor the healing process. A hospital in Manchester, England has shown that the nose can detect early sings of wound infection, and can even distinguish between different infecting organisms. Electronic nose is also used in the analysis of breath of subjects. Researchers at the University of Pennsylvania detected pneumonia in hospitalized patients by collecting breath samples, and analyzing by electronic nose (Hanson, 1997). Applications to Food Products The applications of an electronic nose in the food industry are virtually unlimited. The instrument is currently used in the identification and classification of different food products based on their odor. In most food companies raw materials are not checked frequently for aroma before processing. Final product is checked, and the batch is rejected if a taint or off-odor is present. This may cause difficulty in determining which supplier delivered the faulty material, and the faulty material may have cost a great deal to the manufacturer (Hodgins, 1997). However, if the company uses the electronic nose as a quality tool to check raw materials, this problem can be eliminated. Electronic nose can also be used to monitor food odors during critical stages of production to ensure that optimum processing conditions are being maintained, to monitor product deterioration during shelf life studies and during transport to retailers. It is concluded that this system is ideal for quick QC/QA checking (Hodgins, 1997).

PAGE 54

38 In the dairy area, sensor arrays have been used to determine the role of fatty acids in the aroma profiles of Swiss cheese (Harper et al., 1996), and to differentiate enzyme modified cheese slurries (Jin and Harper, 1996). Korel et al. (1999) used the electronic nose to detect the odor differences in milk due to microbial load, storage time, and sensory panel perception. Sberveglieri et al. (1998) also obtained the selective discrimination of different heat treated commercially available milks (pasteurized, UHT, and sterilized) with an array of four semiconductor sensors. The sensors were formed by selected semiconductor thin film materials. The volatiles from the milk samples were collected by a dynamic headspace, and directly flowed into the test chamber where the sensor array was placed. The data was processed with simple principal component analysis. They stated that the results were promising for the industrial development of an electronic nose for the monitoring of the milk quality. In the meat area, electronic nose was used for separation of ground raw and cooked samples of pork-beef mixtures according to composition and freshness (Turhan et al., 1998), and used for estimation of quality of ground meat, stored under a polyethylene sheet, and gave a good possibility of predicting storage time (Winquist et al., 1993). Electronic nose was also used to monitor sausage fermentation following the changes in volatiles during the fermentation process, and to compare the electronic nose results with sensory panel results. From the sensor readings the fermentation time could be predicted, and sensory panel results were compared with the electronic nose sensor readings in the early stage of the process and on the final sausages (Eklbv et al., 1998).

PAGE 55

39 In the seafood area, applications of the electronic nose have been done in differentiation of odors in shrimp stored on ice (Balaban and Luzuriaga, 1996), storage of tuna (Newman, 1998), and salmon (Luzuriaga and Balaban, 1999a) at different temperatures using conducting polymer sensors. Electronic nose was also used to monitor haddock and cod freshness (Olafsson et al., 1992), and to determine fish storage time (di Natale et al., 1996). The odor of decomposition in raw and cooked shrimp was evaluated based on electronic nose readings, sensory evaluation, and ammonia levels (Luzuriaga and Balaban, 1999b). The fruits and vegetables area has also benefitted from the electronic nose. The electronic nose has been successfully employed for determination of harvest ripeness in cantaloupes (Benady et al., 1995), and quality assessment of packed blueberries (Simon et al., 1996). Volatiles of citrus juice (Hodgins, 1995; Hodgins and Simmonds, 1995), and fresh squeezed orange juice aroma volatiles (Bazemore et al., 1996) have been studied. Bazemore et al. (1997) reported that grapefruit juices of different cultivars were discriminated using metal oxide sensors. Maul et al. (1997) assessed the ability of an electronic nose to nondestructively identify and classify tomato fruit exposed to different harvesting and postharvest handling treatments. Werlein and Watkinson (1997) compared the sensory quality of conventionally processed carrots, green beans, and potatoes using metal oxide sensors and sensory panels. In the area of grains and beans, Bbijesson et al. (1996) used an electronic nose to classify grains, and therefore reduce the inspectorÂ’s exposure to grains that can be contaminated with aflatoxins. Jonsson et al. (1997) analyzed samples of oats, rye, and

PAGE 56

40 barley with different odors and wheat with different levels of ergosterol, fungal, and bacterial loads. The odor of classes of good, mouldy, weakly, and strongly musty oats was predicted with a degree of accuracy using artificial neural networks. The percentage of moldy barley or rye grains in the mixtures of fresh grains was also indicated. It was also reported that there was a high degree of correlation between artificial neural network predictions, and measured ergosterol, fungal, and bacterial loads. Hofmann et al. (1997) used an electronic nose equipped with 4 metal oxide semiconducting sensors to follow the flavor generation during toasting of wheat bread, and to follow the roasting degree of toasted wheat bread slices. Some other work has been also done to discriminate among coffee cultivars, coffee from different origins, and coffee aromas (Aishima, 1991; Tan et al., 1995; Delaure et al., 1996). In non-alcoholic and alcoholic drinks area there is a significant potential for the use of electronic noses. The flavor and aroma of beer and its raw materials were monitored using electronic nose technology (Pearce et al., 1993; Tomlinson, 1995; Tomlinson et al., 1995; Zimmermann and Leclercq, 1995). The aroma of pure hops and blends used in beer making were studied by Lucas and Castan (1995) and Weber and Poling (1996). Viaux and Robillard (1996) also used an electronic nose to help in the determination of the technical specifications of some additives and technological aids used in the sparkling wine process. Since the electronic nose is rapid and objective in quantifying odors, there is a great potential in quality control applications in markets worldwide.

PAGE 57

41 Current Shortcomings of the Electronic Nose In electronic nose technology, the individual sensors need to be reproducible in their response to a given odor or chemical. It is important that a given sensor is reproducible throughout the lifetime of that sensor. However, this is an acute problem for electronic nose applications due to the difficulties of multivariate calibration, and the complex data sets that are required to train the pattern recognition software. The problem with the reproducibility of response over time arise from drift in the sensitivity of the sensors and the poisoning effects. Drift is a slow change in sensitivity, occurs with time, and can also be due to the effects of aging, slow morphological changes in the sensor material and some other long-term effects. Sensor poisoning arises when a sensor is exposed inadvertently to a material which irreversibly binds to, or interacts with, the sensing material leading to a reduction or even total loss of sensitivity. It is possible to avoid the poisoning problem by carefully selecting the right type of sensor for the particular application and by excluding poisons. Unfortunately, there is no chemical sensor which cannot be poisoned. Another aspect of reproducibility is the response reproducibility between sensors of nominally the same type. This is important because if a sensor is poisoned then it can be replaced without recalibration and retraining the system. If the sensors are sufficiently reproducible in their response, then it becomes possible to train one sensor array, and to use the same training set for any nominally identical array in other instruments at different locations. If the transferability of data from one electronic nose to another (same

PAGE 58

42 manufacturer) can be obtained, the use of electronic nose in quality control applications can spread to markets worldwide. Changes in temperature, humidity and flow rate also play an important role in sensitivity. These effects can be minimized by careful system design and sample handling, but this makes the electronic nose more expensive and complex (Eklov et al., 1998). However, it is necessary to overcome the effects of changes in temperature and humidity on the sensors’ baseline and on the magnitude of their responses before portable instruments become available. Objectives The overall objectives of this study were to determine the ability of an electronic nose and sensory panels to detect the presence of pure cultures of P. fluorescens or B. coagidam in sterilized whole, reduced-fat, and fat-free milk samples stored at 1.7°, 7.2°, and 12.8°C; and, to determine the ability of an electronic nose to predict the shelf life of milk. The specific objectives were: 1. To inoculate whole, reduced-fat, and fat-free milk samples with known microbial loads oi P. fluorescens ox B. coagulans, store the samples at 1.7°, 7.2°, and 12.8°C, and measure the electronic nose sensor response at 0, 3, 5, 7, and 10 days in storage; 2. To conduct an odor sensory panel to determine whether the panelists can detect the difference between a reference sample, and inoculated stored samples;

PAGE 59

43 To determine and statistically correlate the relationship between electronic nose readings and the microbial numbers for P.fluorescem or B. coagulans inoculated into whole, reduced-fat, and fat-free milk, stored at 1.7°, 7.2°, and 12.8°C; To determine and statistically correlate the relationship between electronic nose readings and the sensory evaluations for whole, reduced-fat, and fat-free milk inoculated with P. fluorescem or B. coagulans, stored at 1.7°, 7.2°, and 12.8°C; To attempt to predict shelf life (based on microbial numbers) using electronic nose readings from accelerated studies; To inoculate whole milk samples with known microbial loads of P. fluorescens and B. coagulans, store the samples at 1.7°, 7.2°, and 12.8°C, and measure the electronic nose sensor response at 0, 3, 5, 7, and 10 days in storage; To conduct an odor sensory panel to determine whether the panelists can detect the difference between a reference sample and stored samples inoculated with both microorganisms; To determine and statistically correlate the relationship between electronic nose readings and the sensory evaluations for whole milk inoculated with both microorganisms, stored at 1.7°, 7.2°, and 12.8°C.

PAGE 60

CHAPTER 3 MATERIALS AND METHODS Milk Sampling. Inoculation, and Analysis Milk Samples Whole, reduced-fat (2% milkfat) and fat-free milk (Parmalat®, Teaneck, NJ) was purchased from a local supermarket in Gainesville, FL. The samples were aseptically packaged in 946 ml packages and the sell-by-dates and the lot numbers for each set of samples were the same. The packages were kept at 1.7°C until the experiments. The sampling procedures were performed under a sterile laminar flow hood (Nuaire Biological Safety Cabinets, Plymouth, MN). The front surface of the milk package was swabbed thoroughly with 70% alcohol. A sterile, single use PrecisionGlide* needle (B-D® 16 G 1% Becton Dickinson and Company, Cat. No. 305198, Franklin Lakes, NJ) of a sterile, single use 60 ml syringe ( B-D® Becton Dickinson and Company, Cat. No. 30966301, Franklin Lakes, NJ) was inserted through the previously swabbed package wall. A 50-ml milk sample was taken from the package, placed into a 300 ml sterile jar and the lid was closed. 44

PAGE 61

45 Activation of Microorganisms The milk samples were inoculated with Pseudomonas fluorescens (ATCC 948) (American Type Culture Collection, Manassas, VA) and/or Bacillus coagulans (ATCC 7050). P. fluorescens represents the Gram-negative spoilage organisms in milk and B. coagulans the Gram-positive microflora. The freeze-dried culture of P. fluorescens was activated by adding some of the culture into 10 ml nutrient broth (Difco Laboratories, Cat. No. 0003-01-6, Detroit, MI) tube and incubated at 28°C for 48 hrs. The nutrient broth with the activated organism was transferred to 1 50 ml nutrient broth flask and incubated at 28°C for 48 hrs. The ffeeze-dried culture of B. coagulans was activated by adding some of the culture into 10 ml tryptone soy broth (Oxoid Ltd., Cat. No. CM129, Hampshire, England) and incubated at 21°C for 48 hrs. The tryptone soy broth with the activated organism was transferred to 150 ml tryptone soy broth flask and incubated at 21°C for 48 hrs. The activated organisms were centrifuged by using a Sorvall® RC-5B refrigerated superspeed centrifuge (Du Pont Comp., Duluth, GA) at 4342 x g for 10 min for P. fluorescens and at 12061 x g for 10 min for B. coagulans. The supernatants were poured off and the cultures were resuspended with filter sterilized phosphate buffer saline (Harrigan, 1998). They were centrifuged at 4342 x g and 12061 x g for 10 min for P. fluorescens and B. coagidans, respectively. This procedure was repeated twice. Twenty ml of Butterfield’s buffer solution (International BioProducts Inc., Redmond, WA) was added to both centrifuge tubes. Eleven ml of the aliquots from each centrifuge were

PAGE 62

46 transferred to 99 ml Butterfield’s buffer dilution bottle. This dilution was used for all the inoculations of B. coagulam. One more dilution was performed for P. fluorescem, 1 ml from the previous dilution bottle was transferred to 99 ml Butterfield’s buffer dilution bottle, and this dilution was used for all the inoculations. Inoculation of Microorganisms and Sample Treatments The jars filled with 50 ml of milk were inoculated with 1 ml of the specific dilution for each microorganism and the control samples were inoculated with 1 ml of Butterfield’s buffer solution. The samples were stored at 1.7°, 7.2°, and 12.8°C for up to 10 days. The samples were evaluated at days 0, 3, 5, 7, and 10. The samples for the accelerated study were prepared similarly, but the samples were stored at 1.7°C for up to 8 days. At day 0, five jars for each treatment were incubated at 28°C for 24 hrs and these samples were analyzed at day 1. At days 3, 5, and 7, five jars for each treatment were taken out from the 1.7°C refrigerator and incubated at 28°C for 24 h and they were analyzed at days 4, 6, and 8. The study was repeated twice for each type of milk using milk from the same lot. Whole milk (Parmalat®, Teaneck, NJ) was used for the combination study. In this study, the milk samples were inoculated with 1 ml of each P. fluorescens and B. coagulam aliquots. The control samples were inoculated with 1 ml of Butterfield’s buffer solution. The storage conditions and the analyses were the same as in the other studies.

PAGE 63

47 Electronic Nose Measurements An electronic nose (e-NOSE 4000 model, EEV Inc., Amsford, NJ) equipped with twelve conducting polymer sensors (sensor types: 483, 478, 464, 463, 462, 461, 460, 459, 458, 401, 298, 297) was used to quantify the sensor responses to differences in odor of milk samples inoculated with different microorganisms. The electronic nose measurements were done immediately (day 0), and at days 1, 3, 4, 5, 6, 7, 8, and 10. Five replicates were analyzed by the electronic nose for each treatment. Each replicate was kept at room temperature for 30 min prior to analysis in order to let the milk temperature equilibrate to room temperature (22.5° to 23.5°C). Five replicates were analyzed on each day for each storage temperature. The replicates were flushed with compressed air for about 10 sec prior to the electronic nose analysis. The milk jar was placed in the glass sampling vessel of the electronic nose. One day before the experiment started, the electronic nose was calibrated with 75% v/v propylene glycol solution (100% solution from Fisher Scientific, No. P-355-20, Fair Lawn, NJ), following the manufacturer’s recommendation. Every day before the experiments the electronic nose was turned on and compressed air (CGA Grade D, Strate Welding Supply Inc., Jacksonville, FL) was passed through the sensors for at least 30 min. The vessel was purged with compressed air for 2 min to eliminate any foreign odor present in the vessel from the environment for each replicate, and then the sensor head was purged for 4 min with compressed air. During these 4 min, the sample volatiles were equilibrating in the headspace of the vessel. Sensor response data was acquired for 4 min. Total analysis time for each milk sample took 10

PAGE 64

48 min. Readings at 4 min exposure of the sensors to the milk samples were used for data analysis. At the end of the day, electronic nose sensors were cleaned with compressed air for at least 30 min. Electronic nose raw data can be obtained from Dr. Murat Balaban, Food Science and Human Nutrition Department at the University of Florida (filenames:”\\thesis\e-nose data.txt”). Microbial Analysis Microbial counts were performed for each treatment at each sampling day. Dilutions were made using pre-filled sterile disposable Butterfield’s buffer dilution bottles. Total aerobic count method using Petrifilm technique was used to enumerate pure P. fluorescens since it was the only microflora present. Inoculated aerobic plate count Petrifilms (3M Company, St. Paul, MN) were incubated at 28°C for 48 hrs. Bishop and Juan (1988) used agar pour plates and Petrifilm dry medium culture plates to enumerate bacteria after preliminary incubation of milk samples at 13°C or 21°C for 1 8 hrs. Results showed that the Petrifilm technique was not significantly different from agar pour plate methods. Ginn et al. (1984) reported that total aerobic bacteria data comparisons obtained by the Standard Plate Count and Petrifilm method produced a correlation of 0.971 and 0.946, respectively. Since the results of the Petrifilm technique and agar pour plate methods are similar, the Petrifilm technique may be preferred because they require no preparation, and are ready to use. Pure B. coagulans cultures were enumerated using standard plate count agar (APHA) (Oxoid Ltd., Cat. No. CM463, Hampshire, England) by spread plate method.

PAGE 65

49 They were incubated at 32°C for 48 hrs. Colonies were counted and reported as logjo cfu/ml. Microbial counts for inoculated samples of milk were performed as follows: 6 ml of milk from one milk jar and 5 ml milk from another jar of the same treatment were taken and transferred to the same dilution bottle to assure homogeneity. From predetermined dilutions, 1 ml was taken and plated in the Petri films for P. fluorescem. They were incubated at 28°C for 48 hrs. For 5. coagulans, 0.1 ml from predetermined dilutions was spread plated onto standard plate count agar and incubated at 32°C for 48 hrs. A 1 ml sample was taken from control samples, plated in Petri films and incubated at 32°C for 48 hrs. Microbial counts were performed on the control samples to see whether there was any contamination occurred during filling the jars, and any contaminated sets were discarded. All sets used for data analysis were free of contamination. Colonies were counted and reported as log,o cfli/ml of milk. For the combination study, 6ml and 5 ml milk samples from two replicates were transferred to the dilution bottles. Serial dilutions were performed and 0. 1 ml were spread plated onto crystal violet tetrazolium agar (Marshall, 1993) and thermoacidurans agar (Difco Laboratories, Cat. No. 0303-17-5, Detroit, MI) and incubated at 28°C and 32°C for 48 hrs, respectively. Crystal violet tetrazolium agar was used to enumerate P. fluorescem, and thermoacidurans agar was used to enumerate B. coagulans. For control samples, 1 ml milk sample was plated in Petri films and incubated at 32°C for 48 hrs. Colonies were counted and reported as logjo cfu/ml of milk.

PAGE 66

50 Moisture Content Measurements Moisture content was measured in triplicate for each type of milk and repeated twice using the oven method (Bradley et al., 1993). This analysis was performed to have information on the composition of the milk in order to better describe the sample. A sample of approximately 3 g was placed in an aluminum weighing dish (50 mm diameter, Cat. No. 08-732, Fisher Scientific, Fair Lawn, NJ). The sample was placed in an oven at 103°C for 24 hrs. Moisture content was reported as percent wet basis. Fat Content Measurements Fat content was measured according to Chilliard et al. (1991). This analysis was done to confirm that each type of milk had the same fat percentages as it was stated on the labels. Two ml of ethanol, 0.5 ml of 12.1 N HCl, and 25 ml of hexane were added to 5 ml of milk. The mixture was shaken and centrifuged at 500 x g for 5 min. The top organic layer was transferred to a clean tube, and the aqueous phase was reextracted with 25 ml of hexane. Organic layers were placed in the same tube and water was removed with sodium sulfate, and the solvent evaporated under a stream of nitrogen. The total fat percentage was determined gravimetrically. pH Measurements A 50 ml sample of milk in the jar was placed on a stirrer plate. The pH electrode (ROSS pH electrode. Model 81-02, Orion Research Inc., Beverly, MA) was connected to an Expanded Ion Analyzer, and was calibrated every day with pH 4.00 and 7.00 standards

PAGE 67

51 (Buffer solution pH 4.00, SB 101-500 and pH 7.00, SB 107-500, Fisher Scientific, Fair Lawn, NJ). Measurements were performed for each treatment at every time interval and in duplicate. Sensory Evaluation The odor of milk samples was evaluated by a 10-member untrained sensory panel consisting of students, 25-32 years of age, from the Food Science and Human Nutrition Department at the University of Florida. A difference-ffom-control test was performed at days 0, 3, 5, 7, and 10. Panelists were asked to smell milk samples and detect if there was any difference in odor among the treated samples and the reference sample. The reference sample was fresh Parmalat® milk from the same lot used for the experiments. Panelists rated the differences in a 0 to 10 scale (0 = no difference and 10 = very different). Samples were randomized and a hidden control was included in the test. The replicate number 1 was always taken out of the refrigerator 30 min before the sensory analysis at each day. All panelists smelled the same samples. Sensory tests were carried out on both experiments and in the combination study. Data Analysis Electronic nose sensor readings were analyzed in Statistica for Windows (‘98 edition, StatSoft Inc., Tulsa, OK) using discriminant function analysis (DFA) as reported by other researchers (Corcoran, 1993; Gardner and Hines, 1997; Gardner and Bartlett, 1992). Microbial counts and sensory data were used as grouping variables and 12 electronic nose sensor outputs were used as independent variables. DFA was used to

PAGE 68

52 develop predictive models for classification of samples based on grouping variables. The 12 sensor outputs were reduced to 2 discriminant functions. These functions were used to map the data in two dimensional plots and observe separation between groups. Correct classification rates and the coefficients for each function were calculated using Statistica. Data of the compositional analysis, microbiological analysis, pH measurements and sensory scores for each treatment were subjected to analysis of variance with the general linear model (SAS, 1998). Least square means were obtained and separated using the least significant difference test procedures when significant (p < 0.01) F values were obtained.

PAGE 69

CHAPTER 4 RESULTS AND DISCUSSION Milk Sampling. Inoculation, and Analysis Moisture and Fat Content Measurements The moisture contents of whole, reduced-fat (2% milkfat), and fat-free milk samples were 89.53% ± 0.1 1, 90.39% ± 0.06, and 92.26% ± 0.22, respectively. The fat contents of whole, reduced-fat, and fat-free milk samples were 3.20% ± 0.07, 1.98% ± 0.01, and 0. 1 1% ± 0.03, respectively. As the moisture content increased, the fat content decreased. Microbial Analysis Microbial counts for P. fluorescem and B. coagulam during 10 days of storage for each type and two experiments of milk samples are given in Tables 4. 1 and Appendices A for whole, reduced-fat and fat-free milk. Analysis of variance with the general linear model procedures was performed for each inoculated microorganism and for each experiment of each type of milk to see if there was any significant difference due to the storage temperature and time. It was expected that the storage temperature and time would have an effect on the microbial load. It was found that the storage temperature. 53

PAGE 70

54 Table 4-1 . Microbial load of all types of milk samples inoculated with Pseudomonas fluorescens or Bacillus coagidans. Storage Time (Days) Microbial Load (log,o cfli/ml of milk)* Experiment 1 Experiment 2 1.7°C 7.2°C 12.8°C 1.7°C 7.2°C 12.8°C Whole Milk Inoculated with Pseudomonas fluorescens 0 4.20* 4.20* 4.20* 3.60* 3.60* 3.60* 3 4.57*’ 4.63” 8.518 5.28' 5.08' 7.768 5 6.43' 7.28“ 8.95” 4.85” 6.15“ 9.08' 7 6.48' 7.62' 9.23' 6.43' 7.768 9.23'J 10 8.30*^ 9.04” 9.28‘ 6.78* 8.81” 9.34^ Inoculated with Bacillus coagulans 0 4.95^ 4.95^ 4.95* 5.90” 5.90” 5.90” 3 29 oabce 3.42' 3.26”“ 4.60”'“ 4.48”'“ 4.81“* 5 2.42*” 3.43“ 2.88*”“ 4.26” 5.08'*8 5.23” 7 1.95* 2.62*”' 2.11* 5.608”^ 4.78”'“'* 5.15‘8 10 2.04* 2.48*” 2.48*” 3.40* 4.43”'“ 4.81'“'* Reduced-fat Milk Inoculated with Pseudomonas fluorescens 0 3.23* 3.23* 3.23* 3.43” 3.43” 3.43” 3 3.88” 5.79' 8.43* 2.54* 3.54” 7.458 5 5.59' 7.71' 9.26” 3.87' 730* 9.11' 7 4.70*' 8.42^ 9.28” 5.18“ 8.48” 9.26" 10 7.53' 8.918 9.42” 6.40' 9.04' 9.53” Inoculated with Bacillus coagulans 0 5.87” 5.87” JZ 00 5.118 5.118 5.118 3 5.26** 4.91' 5.408 3.04“' 3.42'* 2.60'“ 5 4.42“ 4.48“ 5.15* 1.30* 2.40”' 3.63* 7 3.97' 3.88”' 4.89' <1.0* 2.28”' 2.40”' 10 3.74” 3.36* A.IT <1.0* 2.00” <1.0* Fat-ffee Milk Inoculated with Pseudomonas fluorescens 0 3.99* 3.99* 3.99* 4.32” 4.32” 4.32” 3 5.23” 6.11' 7.00* 4.00* 5.51“ 7.778 5 6.36“ 7.43” 9.20“" 4.71' 6.91' 9.11'j 7 6.72' 8.00* 9.11' 4.26” 8.08” 9.20< 10 7.64' 8.91” 9.34"' 7.26* 8.98' 9.60” Inoculated with Bacillus coagulans 0 5.43« 5.348 5.348 5.70” 5.70” 5.70” 3 3.04“' 2.88”'“ 3.78' 3.38“ 4.578 00 5 2.54* 3.00“ 3.64* 2.40' 3.95' 4.08'* 7 2.65*”' 2.54*” 2.70*”'“ 2.00” 3.85' 3.99' 10 2.40* 2.70*”'“ 2.54*” 1.00* 3.38“ 3.89' * ; Average of two readings : Superscripts within an experiment for each microorganism denote significant difference at the p<0.01 . Means were separated using LSD.

PAGE 71

55 time and the interactions were significant (p<0.01) for each microorganism and for each experiment of each type of milk. The microbial loads for P. fluorescens were plotted versus storage time and are shown in Figures 4. 14.3 for each type of milk and for each experiment. In general P. fluorescens counts increased gradually during storage at 1.7°C, but they increased rapidly at 12.8°C. The growth of P. fluorescens at 7.2°C showed a similar trend with the ones at 12.8°C except for whole milk experiment 2 and reduced-fat milk experiment 2. In whole milk experiment 1 and reduced-fat milk experiment 2, there were lag phases until day 3 and then exponential growth occurred. This could be due to the differences of inoculation loads. Regardless of the initial inoculation levels or the storage temperature, P. fluorescens counts exceeded 10* cfu/ml at day 10 for all types of milk. Since microbial counts in the excess of 1 x 10* cfu/ml are required to produce adverse defects on product quality (Cousin, 1982), formation of off-odors was expected during storage and these were detected by electronic nose. The microbial loads for B. coagulans were plotted versus storage time and are shown in Figures 4.44.6 for each type of milk and for each experiment. The initial inoculation levels were between 10* and 10* cfli/ml. However, this microorganism was unable to grow at 1 .7°, 7.2° and 12.8°C. The microbial counts were generally lower for the milk samples stored at 1.7°C than the samples stored at 7.2° and 12.8°C except for reduced-fat milk experiment 1 . Bacillus spp. were stated as thermoduric psychrotrophs in the literature and these may be variants of mesophilic organisms that have adapted to grow at lower temperatures (Grosskopf and Harper, 1974). B. coagulans used in this study

PAGE 72

Microbial Load (Logio cfii/ml) Microbial Load (Logjo cfu/ml) 56 0 3 5 7 10 Storage Temperatures -o1.7°C 7.2°C "O" 12.8°C Storage Time (Days) Experiment 2 Storage Temperatures -D1 70Q 7.2“C “O" 12.8°C Storage Time (Days) Figure 4-1 . Average microbial load of P. fluorescem for whole milk, both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation.

PAGE 73

Microbial Load (Log jq cfli/ml) Microbial Load (Log,o cfij/ml) 57 Experiment 1 10.0000 9.0000 8.0000 7.0000 6.0000 5.0000 4.0000 3.0000 2.0000 0 3 5 7 10 12.8°C Storage Time (Days) Experiment 2 10.0000 9.0000 8.0000 7.0000 6.0000 5.0000 4.0000 3.0000 2.0000 0 3 5 7 10 12.8°C Storage Time (Days) Figure 4-2. Average microbial load of P. fluorescem for reduced-fat milk, both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation.

PAGE 74

Microbial Load (Logic cfii/ml) Microbial Load (Logic cfu/ml) 58 Experiment 1 10.0000 9.0000 8.0000 7.0000 6.0000 5.0000 4.0000 3.0000 2.0000 0 3 5 7 10 12.8°C Storage Time (Days) Experiment 2 Storage Temperatures -D1.7°C 7.2°C "> 12.8°C Storage Time (Days) Figure 4-3. Average microbial load of P. fluorescens for fat-ffee milk, both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation.

PAGE 75

Microbial Load (Logio cfli/ml) Microbial Load (Log,o cfii/ml) 59 Experiment 1 7.0000 6.0000 5.0000 4.0000 3.0000 2.0000 1.0000 0.0000 3 5 7 Storage Time (Days) 10 Storage Temperatures 1.7°C ^ 7.2°C mil 12.8°C Experiment 2 Storage Temperatures 1.7°C IS 7.2°C mni i2.8°c Figure 4-4. Average microbial load for B. coagulam for whole milk, both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation.

PAGE 76

Microbial Load (Logio cfu/ml) Microbial Load (Log,o cfWml) 60 Experiment 1 7.0000 1— 6.0000 5.0000 4.0000 3.0000 2.0000 1.0000 0.0000 Storage Temperatures 1.7°C 7.2°C 12.8”C Storage Time (Days) Experiment 2 7.0000 6.0000 5.0000 4.0000 3.0000 2.0000 1.0000 0.0000 Figure 4-5. Average microbial load for B. coagulans for reduced-fat milk, both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation. Storage Temperatures 1.7“C S 7.2“C mu i2.8°c

PAGE 77

Microbial Load (Log i o cfii/ml) Microbial Load (Log , o cfii/ml) 61 Expjeriment 1 3 5 7 Storage Time (Days) Storage Temperatures 1.7°C 7.2°C 12.8°C Experiment 2 0 3 5 7 10 Storage Temperatures 1.7°C 7.2°C 12.8°C Storage Time (Days) Figure 4-6. Average microbial load for B. coagulans for fat-free milk, both experiments stored at different temperatures over time. Error bars signify ±1 standard deviation.

PAGE 78

62 (ATCC 7050) which was isolated from an evaporated milk most probably lost its ability to grow at refrigeration temperatures. Attempts were made to adapt this microorganism to grow at low temperatures. Initially, it was activated in tryptone soy broth at 35°C and was transferred to a new sterile tryptone soy broth and incubated at 21°C. Milk samples were inoculated with this and the organism was expected to grow at lower temperatures, but it did not proliferate at refrigeration temperatures. Accelerated Study The microbial loads for all milk types, both experiments and two microorganisms for the accelerated study are given in Table 4.2. The microbial counts for P. fluorescens increased approximately to 10* cfu/ml for all milk types after 24 hrs of incubation at 28°C. The initial inoculation level had an effect on the microbial growth. Whole milk experiment 1 was inoculated with the microbial load of 4.20 logio cfu/ml and the microbial load increased to 8.32 log,o cfu/ml after 24 hrs at 28°C. On the other hand, whole milk experiment 2 was inoculated with 3.60 logjo cfti/ml and the microbial growth reached to 7.96 logio cfii/ml. This was observed in all types and experiments of milk. The P. fluorescens counts increased to 10*10’ cfu/ml for day 4, 6, and 8. Due to the possibility of insufficient substrate, the microbial loads did not increase more than 10’ cfu/ml throughout the accelerated study regardless of the microbial load of the milk sample that was incubated at each time period. The type of milk did not have a significant effect on the microbial loads for this study. The microbial counts for B. coagulans increased for all milk types after each 24 hrs of incubation at 28°C. At day 0, the initial microbial load was 4.20 logn, cfii/ml for

PAGE 79

Table 4-2. Microbial load of all types of milk inoculated with Pseudomonas fluorescens or Bacillus coa^lans in accelerated study. Milk Type Analyses Time (Days) Microbial Load (log,o cfu/ml of milk)* Experiment 1 (Inoculated Microorganisms) Experiment 2 [Inoculated Microorganisms) P. fluorescens B. coagulans P. fluorescens B. coagulans Whole Milk 1 8.32 4.36 7.96 6.00 4 8.34 3.48 8.72 5.15 6 8.72 2.40 8.70 5.48 8 9.04 2.26 8.42 4.49 Reduced-fat Milk 1 7.84 7.04 7.94 6.67 4 8.77 5.87 8.57 6.15 6 8.86 5.86 8.61 4.18 8 8.91 5.04 8.80 4.92 Fat-free Milk 1 8.42 7.74 8.62 6.89 4 8.56 7.58 8.46 5.00 6 9.04 lAl 8.42 5.00 8 8.82 6.61 8.53 3.00 * : Average of two readings

PAGE 80

64 whole milk experiment 1 and it increased to 4.36 logio cfu/ml after incubation. Even though only a slight increase in microbial number was observed in this case, in other cases, higher increases in the microbial loads were observed. For example at day 7 the microbial load for fat-free milk experiment 2 increased 1 log cycle after the incubation at 28°C. This also proved that this organism has not been adapted to grow at low temperatures. On the other hand, it was observed that B. coagulam counts for all types and experiments of milk samples were decreased over time. The milk samples were kept at 1.7°C before the accelerated study analysis, and this microorganism could not survive at low refrigeration temperatures as mentioned before. The types of milk had a significant effect in this study, but this could be observed because of not having the same initial inoculation levels at day 0 for each type and experiment of milk samples. Combination Study In the combination study, P. fluorescem and B. coagulam counts are given in Table 4-3. The storage time and temperatures had significant effects on the microbial loads for both microorganisms. P. fluorescem counts increased rapidly for the samples stored at 12.8°C (Figure 4-7) and they increased gradually for the samples stored at 7.2°C until day 7, and had a rapid increase from day 7 to day 10. The microbial loads for P. fluorescem increased to 10® cfu/ml for the samples stored at 7.2° and 12.8°C at day 10. On the other hand, P. fluorescem in the whole milk samples stored at 1.7°C multiplied, but did not show any increase in counts compared to the samples stored at the higher temperatures. This could occur due to the presence of B. coagulam together with P. fluorescem in the samples.

PAGE 81

65 Microbial load of whole milk samples inoculated with Pseudomonas fluorescens and Bacillus coagulans in combination study. Storage Microbial Load (logio cfii/ml of milk)* Time Pseudomonas fluorescens Bacillus coagulans (Days) Storage Temperature Storage Temperature 1.7°C 7.2°C 12.8°C 1.7°C 7.2°C 12.8°C 0 3.82** 3.82” 3.82” 5.60” 5.60” 5.60” 3 3.08* 5.43= 7.45^ 4.49= 5.15** 5.51” 5 2.90* 6.11“ 8.77* 4.34= 5.08=^ 5.30* 7 3.89** 6.53= 9.00*” 4.04” 4.94“= 4.95“= 10 * . A, 3.70'’ 8.89* 9.23” 3.68* 4.86“ 4 98“=f * ^ Average of two readings : Superscripts within each microorganism denote significant difference at the p<0.01. Means were separated using LSD. The microbial loads for B. coagulans for the samples stored at each storage temperature decreased during the 10-day storage. However, for the samples stored at 7.2° and 12.8°C, the decrease in B. coagulans counts was less compared to the decrease in the counts for the samples stored at 1 .7°C (Table 4-3 and Figure 4-7). This was due to the lack of adaptation of 5. coagulans at low refrigeration temperatures. Accelerated Study of the Combination Study The microbial loads for P. fluorescens and B. coagulans for the accelerated study of the whole milk samples in the combination study are given in Table 4-4. After each incubation at 28°C, P. fluorescens counts were increased to 10* cfli/ml. B. coagulans counts increased during incubation compared to the microbial load of the sample before it was incubated. However, B. coagulans counts decreased over time.

PAGE 82

Microbial Load (Logjo cfu/ml) Microbial Load (Logio cfu/ml) 66 P. Jluorescem in Combination Study 10.0000 9.0000 8.0000 7.0000 6.0000 5.0000 4.0000 3.0000 2.0000 0 3 5 7 10 12.8°C Storage Time (Days) B. coagulcms in Combination Study 7.0000 6.0000 5.0000 4.0000 3.0000 2.0000 1.0000 0.0000 3 5 7 Storage Time (Days) 10 Storage Temperatures 1.7°C ^ 7.2°C Emu i2.8°c Figure 4-7. Average microbial load for P. Jluorescem and B. coagulam for combination study of whole milk, stored at different temperatures over time. Error bars signify ±1 standard deviation.

PAGE 83

67 Table 4-4. Microbial load of whole milk samples inoculated with Pseudomonas fluorescens and Bacillus coagulans in accelerated study. Storage Time (Days) Microbial Load (logio cfu/ml of milk)* Pseudomonas fluorescens Bacillus coagulans 1 8.34 6.69 4 7.95 5.70 6 8.49 5.04 8 8.59 3.85 * : Average of two readings pH Measurements The pH of all whole, reduced-fat and fat-ffee milk samples stored at different temperatures during 10 days of storage are given in Tables 4-5, 4-6, and 4-7, respectively (Appendix B). Changes in the pH of whole, reduced-fat and fat-free milk during storage did not follow any specific trend. The pH of each type of milk and the experiments of each milk was significantly different (p<0.01). Therefore data for each milk type and experiments of these were analyzed independently. Overall, pH of the samples treated with P. fluorescens decreased during storage. This decrease was higher for the samples stored at 12.8°C than for the samples stored at 1.7° and 7.2°C, except for the fat-free experiment 2 samples stored at 7.2° and 12.8°C, where the pH increased. However, changes in pH were not as large as microbial loads. In most cases, there were minimum pH changes for control and samples inoculated with B. coagulans.

PAGE 84

68 Table 4-5. pH of whole milk inoculated with Pseudomonas fluorescens or Bacillus coa^lans. Storage Time (Days) pH 1.7°C 7.2°C 12.8'^C C Pf Be C Pf Be C Pf Be Exp. 1 0 6.73 6.71 6.71 6.73 6.71 6.71 6.73 6.71 6.71 St. dev. 0.06 0.01 0.01 0.06 0.01 0.01 0.06 0.01 0.01 3 Avg. 6.68 6.74 6.64 6.63 6.59 6.66 6.66 6.63 6.67 St. dev. 0.01 0.13 0.04 0.01 0.08 0.01 0.00 0.02 0.01 5 Avg. 6.81*^ 6.89“ 6.82”' 6.81”' 6.81”' 6.85'“ 6.75” 6.65“ 6.78”' St. dev. 0.01 0.01 0.04 0.01 0.01 0.05 0.01 0.02 0.00 7 Avg. 6.82'^ 6.75”^ 6.72*’ 6.80" 6.78”'“ 6.80" 6.81' 6.62“ 6.80" St. dev. 0.01 0.00 0.01 0.01 0.02 0.01 0.01 0.04 0.02 10 Avg. 6.76'’ 6.78” 6.74” 6.76” 6.62‘ 6.79” 6.78” 6.57“ 6.79” St. dev. 0.03 0.02 0.01 0.01 0.05 0.01 0.01 0.04 0.01 Exp. 2 0 Avg. 6.76 6.76 6.74 6.76 6.76 6.74 6.76 6.76 6.74 St. dev. 0.00 0.06 0.01 0.00 0.06 0.01 0.00 0.06 0.01 3 Avg. 6.76 6.72 6.76 6.73 6.65 6.75 6.73 6.73 6.74 St. dev. 0.01 0.01 0.01 0.02 0.11 0.03 0.01 0.01 0.02 5 Avg. 6.72 6.73 6.71 6.72 6.72 6.70 6.73 6.66 6.65 St. dev. 0.07 0.08 0.02 0.01 0.02 0.00 0.01 0.01 0.04 7 Avg. 6.77'= 6.76' 6.69” 6.75' 6.76' 6.73”' 6.77' 6.63“ 6.69” St. dev. 0.01 0.01 0.00 0.00 0.00 0.02 0.02 0.03 0.01 10 Avg. 6.74” 6.73” 6.72” 6.71” 6.72” 6.73” 6.73” 6.66“ 6.61“ St. dev. 0.01 0.01 0.00 0.01 0.04 0.01 0.01 0.01 0.02 Avg. * ; Average of two readings St. dev. ; Standard deviations of the means C ; Control samples Pf ; Samples inoculated with P. fluorescens Be ; Samples inoculated with B. coagulans : Superscripts within a row denote significant difference at the p<0.01 . Means were separated using LSD.

PAGE 85

69 Table 4-6. pH of reduced-fat milk inoculated with Pseudomonas fluorescens or Bacillus coa^dans. Storage Time (Days) pH 1.7°C 7.2°C 12.8°C C Pf Be C Pf Be C Pf Be Exp. 1 0 Avg.* 6.75 6.70 6.75 6.75 6.70 6.75 6.75 6.70 6.75 St. dev. 0.00 0.13 0.01 0.00 0.13 0.01 0.00 0.13 0.01 3 Avg. 6.72'> 6.70” 6.32* 6.72” 6.70” 6.78” 6.72” 6.67” 6.73” St. dev. 0.04 0.05 0.02 0.01 0.10 0.01 0.02 0.02 0.01 5 Avg. 6.75*= 6.76" 6.74”" 6.73”" 6.67” 6.73”" 6.73”" 6.53* 6.74”" St. dev. 0.01 0.01 0.02 0.00 0.03 0.01 0.01 0.01 0.06 7 Avg. 6.71” 6.68* 6.71” 6.73” 6.66* 6.73” 6.73” 6.51* 6.71” St. dev. 0.01 0.01 0.01 0.03 0.03 0.01 0.06 0.08 0.01 10 Avg. 6.75“ 6.69" 6.71"“ 6.73"“ 6.60” 6.70"“ 6.70"“ 6.49* 6.72"“ St. dev. 0.02 0.00 0.01 0.01 0.03 0.00 0.00 0.01 0.03 Exp. 2 0 Avg. 6.75 6.75 6.75 6.75 6.75 6.75 6.75 6.75 6.75 St. dev. 0.01 0.00 0.00 0.01 0.00 0.00 0.01 0.00 0.00 3 Avg. 6.76 6.76 6.76 6.75 6.78 6.76 6.77 6.68 6.88 St. dev. 0.01 0.00 0.00 0.01 0.01 0.01 0.01 0.07 0.07 5 Avg. 6.78” 6.77* 6.76” 6.74” 6.71* 6.73” 6.74” 6.67* 6.73” St. dev. 0.01 0.00 0.01 0.01 0.01 0.00 0.00 0.04 0.00 7 Avg. 6.78“ 6.76"“ 6.75" 6.74”" 6.71” 6.73”" 6.74”" 6.54* 6.74" St. dev. 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.00 10 Avg. 6.69“* 6.67" 6.71“ 6.69"“ 6.51” 6.69"“ 6.77" 6.46* 6.69"“ St. dev. 0.01 0.01 0.01 0.01 0.01 0.00 0.01 0.01 0.01 Avg.* ; Average of two readings St. dev. : Standard deviations of the means C : Control samples Pf : Samples inoculated with P. fluorescens Be ; Samples inoculated with B. coagulans ; Superscripts within a row denote significant difference at the p<0.01 . Means were separated using LSD.

PAGE 86

70 Table 4-7. pH of fat-free milk inoculated with Pseudomonas fluorescens or Bacillus coa^ilans. Storage Time (Days) 1.7°C 7.2°C 12.8°C C Pf Be C Pf Be C Pf Be Exp. 1 0 Avg.* 6.77 6.75 6.74 6.77 6.75 6.74 6.77 6.75 6.74 St. dev. 0.01 0.02 0.03 0.01 0.02 0.03 0.01 0.02 0.03 3 6.72 6.74 6.75 6.72 6.74 6.72 6.78 6.76 6.76 St. dev. 0.01 0.01 0.02 0.01 0.02 0.00 0.04 0.04 0.02 5 Avg. 6.72*~ 6.73** 6.71" 6.74'=“ 6.72* 6.76“ 6.72"'= 6.61* 6.72"^ St. dev. 0.00 0.01 0.00 0.00 0.01 0.01 0.00 0.01 0.01 7 Avg. 6.82 6.75 6.74 6.72 6.70 6.72 6.69 6.71 6.70 St. dev. 0.09 0.05 0.02 0.01 0.03 0.01 0.04 0.01 0.00 10 Avg. 6.76 6.69 6.76 6.75 6.75 6.73 6.70 6.67 6.72 St. dev. 0.01 0.11 0.00 0.02 0.06 0.01 0.00 0.00 0.02 Exp. 2 0 Avg. 6.67* 6.70" 6.69" 6.67* 6.70" 6.69" 6.67* 6.70" 6.69" St. dev. 0.01 0.00 0.00 0.01 0.00 0.00 0.01 0.00 0.00 3 Avg. 6.72* 6.72* 6.72* 6.74" 6.76" 6.76" 6.75" 6.75" 6.73" St. dev. 0.00 0.01 0.01 0.00 0.01 0.00 0.04 0.01 0.01 5 Avg. 6.71 6.71 6.71 6.71 6.70 6.72 6.70 6.67 6.70 St. dev. 0.05 0.00 0.01 0.01 0.01 0.01 0.01 0.02 0.01 7 Avg. 6.69*" 6.68*" 6.69"^ 6.70"“ 6.7ff 6.70"^ 6.68*" 6.79“ 6.67* St. dev. 0.01 0.00 0.01 0.01 0.00 0.00 0.00 0.01 0.01 10 Avg. 6.71* 6.70* 6.70* 6.71* 6.78" 6.70* 6.72* 6.93' 6.70* St. dev. 0.00 0.01 0.00 0.01 0.01 0.00 0.00 0.01 0.01 *Avg : Average of two readings St. dev. : Standard deviations of the means C ; Control samples Pf : Samples inoculated with P. fluorescens ^ ; Samples inoculated with B. coagulans : Superscripts within a row denote significant difference at the p<0.01 . Means were separated using LSD.

PAGE 87

71 pH Measurements for Accelerated Study In the accelerated study, the changes in pH for all types of milk during storage did not follow any specific trend (Table 4-8). According to the 10-day storage study, pH for the samples inoculated with P. fluorescem and stored at higher temperatures decreased more than for the samples stored at lower temperatures. However, the opposite occurred in the accelerated study. The incubation temperature for the accelerated study was 28°C and it was expected to have higher pH drops, but contrary to this, pH increased for all milk types and experiments except for whole milk experiment 2 samples. The reason behind this was not understood. In most cases there were slight pH changes for control and samples inoculated with B. coagulans. No research has been found in the literature on effects of growth of P. fluorescens and B. coagulans on pH. pH Measurements for Combination and Accelerated Studies In the combination study, pH values for whole milk control and inoculated with P. fluorescens and B. coagulans samples are given in Table 4-9. The pH values of all control samples at all storage temperatures increased. On the other hand, pH values for samples inoculated with both microorganisms and stored at 1.7°C did not change, but pH values for samples stored at 7.2° and 12.8°C decreased. The reason for this could not be explained. The pH values for the whole milk control and inoculated with P. fluorescens and B. coagulans for the accelerated study are given in Table 4-10. For the control samples pH slightly dropped, but pH values increased for the other samples.

PAGE 88

72 Table 4-8. pH of all types of milk inoculated with Pseudomonas fluorescens or Bacillus coagulans in accelerated study. Milk Type Analyses Time (Days) P H Experiment (Inoculated microor 1 ganisms) Experiment (Inoculated microor 2 ganisms) Control Pseudomonas fluorescens Bacillus coa^dans Control Pseudomonas fluorescens Bacillus coagulans Whole Milk 1 Avg.* 6.64 6.81 6.73 6.70 6.78 6.72 St. dev. 0.01 0.01 0.03 0.03 0.09 0.03 4 Avg. 6.75 6.81 6.66 6.74 6.83 6.82 St. dev. 0.01 0.02 0.01 0.01 0.03 0.01 6 Avg. 6.66 6.84 6.67 6.70 6.86 6.78 St. dev. 0.02 0.02 0.00 0.03 0.05 0.03 8 Avg. 6.68 6.88 6.67 6.66 6.77 6.71 St. dev. 0.04 0.01 0.01 0.01 0.02 0.01 Reducedfat Milk 1 Avg. 6.71 6.74 6.70 6.71 6.72 6.70 St. dev. 0.01 0.01 0.02 0.01 0.01 0.01 4 Avg. 6.66 6.73 6.69 6.76 6.84 6.79 St. dev. 0.02 0.01 0.01 0.01 0.02 0.02 6 Avg. 6.70 6.79 6.69 6.54 6.84 6.72 St. dev. 0.01 0.00 0.01 0.03 0.01 0.00 8 Avg. 6.60 6.77 6.68 6.72 6.97 6.74 St. dev. 0.08 0.04 0.01 0.01 0.01 0.00 Fat-ffee Milk 1 Avg. 6.66 6.69 6.48 6.65 6.55 6.58 St. dev. 0.01 0.08 0.01 0.01 0.02 0.01 4 Avg. 6.56 6.84 6.50 6.71 6.74 6.69 St. dev. 0.04 0.06 0.04 0.04 0.01 0.01 6 Avg. 6.42 6.79 6.69 6.66 6.78 6.68 St. dev. 0.01 0.03 0.00 0.03 0.01 0.01 8 Avg. 6.72 6.92 6.73 6.72 6.83 6.71 St. dev. 0.01 0.02 0.00 0.00 0.03 0.00 Avg. * : Average of two readings St. dev. ; Standard deviations of two readings

PAGE 89

73 pH of whole milk control and inoculated with P. fluorescem and B. coa^ilam samples. pH Storage Time Sample Control Pseudomonas fluorescem and Bacillus coa^lam (Days) Storage Temperature Storage Temperature 1.7°C 7.2°C 12.8°C 1.7°C l.TC 12.8°C 0 Avg.* 6.65 6.65 6.65 6.70 6.70 6.70 St. dev. 0.05 0.05 0.05 0.01 0.01 0.01 3 Avg. 6.68 6.67 6.67 6.69 6.69 6.72 St. dev. 0.01 0.00 0.00 0.00 0.01 0.01 5 Avg. 6.68 6.70 6.68 6.64 6.63 6.57 St. dev. 0.02 0.00 0.01 0.06 0.09 0.04 7 Avg. 6.70 6.71 6.70 6.72 6.71 6.56 St. dev. 0.01 0.01 0.00 0.00 0.01 0.00 10 . A. Avg. 6.71 6.72 6.71 6.70 6.64 6.64 St. dev. 0.00 0.00 0.00 0.00 0.01 0.01 Avg. * : Average of two readings St. dev. : Standard deviations of two readings Table 4-10. pH of whole milk control and inoculated with P. fluorescem and B. coa^lans samples in accelerated study. Analyses Time (Days) pH Sample Control Pseudomonas fluorescem and Bacillus coa 2 idans 1 Avg.* 6.70 6.69 St. dev. 0.03 0.01 4 Avg. 6.66 6.67 St. dev. 0.02 0.08 6 Avg. 6.71 6.73 St. dev. 0.00 0.04 8 . A, Avg. 6.68 6.80 St. dev. 0.01 0.02 Avg. * : Average of two readings St. dev. : Standard deviations of two readings

PAGE 90

74 Sensory Evaluation Sensory data showed that in general panelists detected the odor differences due to the growth of P. fluorescens in all types of milk samples, but had difficulties in detecting the odor changes due to B. coagulans. This might be because B. coagulans could not grow at 1.7°, 7.2°, and 12.8°C and did not generate enough microbial metabolites which gave off-odors. Sensory scores given by ten panelists for all milk types are given in Appendix C. The average sensory scores for whole, reduced-fat, and fat-free milk control, inoculated with P. fluorescens or B. coagulans and hidden control samples and their standard deviations are shown in Tables 4-11,4-12 and 4-13, respectively. The hidden control samples were the same as the reference samples. At day 0, hidden control samples were not presented to the panelists since all the samples were fresh and they all were assumed to have a difference of 0. The average sensory scores for all milk types and for both experiments were rounded off to the nearest integer to facilitate analysis using DFA. These scores are given in Table 4-14. In most cases, especially toward the end of the storage period and at higher storage temperatures, the sensory scores for control and hidden control samples are significantly different from the sensory scores of samples treated with P. fluorescens or B. coagulans. The scores for samples treated with P. fluorescens were significantly different from the rest of the sensory scores. Overall, the sensory scores for the controls and hidden control samples were not significantly different from each other. As the temperature and storage time increased, the sensory scores given to the samples

PAGE 91

75 Table 4-11. Sensory scores for whole milk inoculated with P. fluorescens or B. coagulans. Inoculated Microorganisms Storage Time (Days) Sensory Scores Experiment 1 Experiment 2 1.7°C 7.2°C 12.8°C 1.7°C 7.2°C 12.8°C Control 0 Avg.* 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 1.00 1.30 0.80 0.70 0.60 1.30 St. dev. 0.82 1.16 0.92 1.25 0.97 1.49 5 Avg. 1.20 0.60 1.00 0.30 1.60 1.50 St. dev. 1.03 0.70 1.05 0.48 0.48 0.97 7 Avg. 1.40 1.40 1.00 0.90 1.50 1.20 St. dev. 1.07 1.84 0.67 0.99 1.18 1.48 10 Avg. 1.30 1.20 1.00 0.70 2.00 1.20 St. dev. 1.16 0.79 1.05 0.95 1.49 1 03 Pseudomonas fluorescens 0 Avg. 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 000 3 Avfi. 1.20 1.40 1.40 1.00 1.60 1.80 St. dev. 1.14 0.97 0.97 0.94 1.07 1 14 5 Avg. 1.80 0.80 4.10 0.60 0.60 4.30 St. dev. 1.87 1.40 2.33 0.70 0.70 1.49 7 Avg. 2.60 2.30 6.90 1.60 2.00 5.80 St. dev. 1.84 0.82 1.37 1.35 1.15 2.57 10 Avg. 1.50 5.70 8.90 1.80 3.20 8.10 St. dev. 1.43 2.06 1.45 1.23 1.93 1.60 Bacillus coagulans 0 Avg. 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 1.30 1.20 0.50 1.00 1.20 1.20 St. dev. 0.82 0.92 0.85 1.25 1.48 1.14 5 Avg. 0.20 0.70 1.20 1.80 1.20 4.30 St. dev. 0.63 0.67 1.03 0.99 1.14 0.95 7 Avg. 1.30 0.80 1.20 1.00 1.10 7.10 St. dev. 0.95 0.79 1.40 1.25 1.29 1.66 10 Avg. 1.30 1.30 1.40 1.30 2.40 7.70 St. dev. 1.16 0.95 1.07 1.06 2.01 1.34 Hidden Control 0 Avg. . St. dev. _ . 3 Avg. 0.20 0.20 0.50 0.50 St. dev. 0.42 0.42 • 0.97 1.08 5 Avg. 0.50 0.40 1.10 1.10 1.10 St. dev. 0.53 0.70 0.99 1.10 0.57 7 Avg. 1.30 0.50 1.00 1.00 0.80 St. dev. 0.67 0.71 0.94 1.41 1.14 10 Avg. 1.00 0.80 0.60 0.70 0.40 St. dev. 1.15 0.42 0.84 1.25 0.70 The 0 to 10 scale was used (0 = no difference and 10 = very different). Avg.* : Average of ten readings St. dev. : Standard deviation of ten readings

PAGE 92

76 Table 4-12. Sensory scores for reduced-fat milk inoculated with P. fluorescens or B. coa^lam. Inoculated Microorganisms Storage Time (Days) Sensory Scores Experiment 1 Experiment 2 1.7°C 7.2°C 12.8°C 1.7°C 7.2°C 12.8°C Control 0 Avg* 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 1.00 1.00 0.80 0.91 2.72 1.19 St. dev. 0.82 0.94 1.23 0.94 2.08 0.88 5 Avg. 0.90 1.00 0.20 2.09 0.69 1.44 St. dev. 1.29 0.94 0.42 1.64 0.84 0.97 7 Avg. 1.30 1.20 2.10 1.82 0.71 0.91 St. dev. 1.16 1.03 1.45 1.83 0.97 1.20 10 Avg. 1.20 1.60 1.40 1.45 0.91 0.80 St. dev. 0.63 1.90 1.07 2.98 0 82 1 03 Pseudomonas fluorescens 0 Avg. 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 1.80 2.30 6.50 0.91 1.50 1.20 St. dev. 1.14 2.16 2.12 1.04 1.35 0.92 5 Avg. 1.20 3.40 6.80 2.27 2.20 5.20 St. dev. 1.32 2.95 1.99 1.19 1.40 1.32 7 Avg. 3.30 3.00 8.00 1.91 3.30 8.00 St. dev. 0.67 1.25 1.25 2.07 1.34 1.33 10 Avg. 2.30 6.10 8.70 2.45 8.40 9.10 St. dev. 1.16 1.45 2.11 2.70 117 1 20 Bacillus coagulans 0 Avg. 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 0.90 1.30 1.50 1.27 1.30 1.30 St. dev. 0.99 1.34 1.35 0.90 1.34 1.06 5 Avg. 1.90 2.00 1.20 2.18 1.90 1.90 St. dev. 1.66 0.94 0.79 1.54 1.10 0.32 7 Avg. 1.50 1.60 1.50 1.73 1.40 2.00 St. dev. 0.85 0.97 1.58 1.95 1.07 0.94 10 Avg. 0.90 1.50 1.80 2.55 3.00 3.60 St. dev. 1.20 097 2.62 2.81 I 56 1 17 Hidden Control 0 Avg. St. dev. . _ 3 Avg. 0.60 0.60 1.00 1.10 0.30 St. dev. 1.23 0.97 1.68 1.20 0.67 5 Avg. 1.10 1.20 0.73 0.80 0.70 St. dev. 0.88 0.92 1.68 1.23 1.06 7 Avg. 0.70 0.90 1.64 0.60 0.90 St. dev. 0.82 1.10 2.11 0.70 1.66 10 Avg. 0.80 1.40 1.82 0.70 0.40 St. dev. 0.79 2.07 2.82 1.16 0.70 The 0 to 10 scale was used (0 = no difference and 10 = very different). Avg.* ; Average of ten readings St. dev. : Standard deviations of ten readings

PAGE 93

77 Table 4-13. Sensory scores for fat-free milk inoculated with P. fluorescens or B. coagidans. Inoculated Microorganisms Storage Time (Days) Sensory Scores Experiment 1 Experiment 2 1.7°C 7.2°C 12.8°C 1.7°C 7.2°C 12.8°C Control 0 Avg.* 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 0.30 0.70 0.50 0.80 1.10 0.60 St. dev. 0.95 0.67 0.85 0.92 1.29 0.70 5 Avg. 2.10 1.10 1.50 0.40 1.00 0.50 St. dev. 1.60 0.57 0.71 0.70 1.05 0.53 7 Avg. 0.90 2.10 3.20 1.00 1.00 1.30 St. dev. 0.88 1.10 2.49 0.82 0.82 1.16 10 Avg. 0.80 1.30 2.60 0.40 0.90 0.70 St. dev. 0.79 0.67 1.58 0.70 0.88 0.48 Pseudomonas fluorescens 0 Avg. 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 0.40 0.90 2.40 1.30 1.10 3.80 St. dev. 0.52 1.00 0.97 0.95 1.10 1.32 5 Avg. 2.00 1.40 7.20 1.60 2.20 7.50 St. dev. 0.94 0.70 1.03 1.58 1.62 1.08 7 Avg. 2.20 3.60 8.60 2.00 3.60 9.10 St. dev. 0.79 0.84 1.17 1.15 1.65 0.74 10 Avg. 3.10 5.10 9.80 3.60 6.60 9.80 St. dev. 2.38 2.42 0.63 1.35 0.70 0.42 Bacillus coagulans 0 Avg. 0.00 0.00 0.00 0.00 0.00 0.00 St. dev. 0.00 0.00 0.00 0.00 0.00 0.00 3 Avg. 0.90 0.80 1.20 1.50 1.70 1.50 St. dev. 0.88 0.92 1.14 1.27 0.67 0.71 5 Avg. 0.90 1.10 1.30 1.70 2.50 2.50 St. dev. 0.74 0.32 0.67 1.34 1.43 0.97 7 Avg. 1.70 2.50 5.90 1.50 1.80 4.90 St. dev. 0.48 0.97 0.74 0.85 1.23 1.66 10 Avg. 2.40 2.70 8.40 2.20 2.10 3.00 St. dev. 1.78 1.70 1.17 1.03 099 0.82 Hidden Control 0 Avg. _ St. dev. . . 3 Avg. 0.90 0.60 0.80 0.40 0.20 0.20 St. dev. 0.57 0.70 1.40 0.97 0.63 0.63 5 Avg. 0.30 0.30 0.40 0.40 0.20 0.30 St. dev. 0.67 0.95 1.26 0.70 0.42 0.48 7 Avg. 0.20 0.20 0.60 0.30 0.10 0.40 St. dev. 0.63 0.63 1.26 0.67 0.32 0.84 10 Avg. 0.70 0.30 0.80 0.30 0.20 0.20 St. dev. 1.06 0.67 1.32 0.48 0.42 0.42 The 0 to 10 scale was used (0 = no difference and 10 = very different). Avg.* ; Average of ten readings St. dev. : Standard deviations of ten readings

PAGE 94

78 Table 4-14. Average sensory scores (rounded) of all types of milk inoculated with P. fluorescem or B. coa^lans. Milk Type Storage Time (Days) Sensory Scores 1 . 7 °C 7 . 2 °C 12 . 8 °C C Pf Be He c Pf Be He c Pf Be He 0 Avg.* 0 0 0 0 0 0 0 0 0 _ 3 Avg. 1 1 1 1 1 1 0 1 1 1 0 Rep 1 5 Avg. 1* 2* 0^* 1 1 1 1 1'’ 4 “ 1'’ o'’ 7 Avg. I** l'’ l'’ 2* I** 1'’ 1'’ T l'’ 1'’ 10 Avg. 1 2 1 l'’ 6 l'’ I** 1'’ 9a 1'’ 1'’ o 0 Avg. 0 0 0 0 0 0 0 0 0 _ § 3 Avg. 1 1 1 l'’ 2* P 1'’ 1 2 1 1 Rep 2 5 Avg. 0*= |bc 2* 2ab 2 2 1 1 4a 4 * 1'’ 7 Avg. 1 2 1 1 2 2 1 1 1 6 7 1'’ 10 Avg. 1 2 1 1 2ab 3 2 1'’ 1'’ 8“ 8* o'’ 0 Avg. 0 0 0 0 0 0 0 0 0 _ 3 Avg. l'’ 2* I** 1 2 1 1 1'’ T 2'’ 1*’ Rep 1 5 Avg. 1 1 2 l'’ 3 * 2**’ 1'’ o'’ T 1'’ 1'’ 7 Avg. I** 3 " 2‘’ 2 be 3 * 2'’ P 2'’ 8* 2'’ 1'’ 1 10 Avg. 1 2 1 2'’ 6’ 2'’ 1'’ 1'’ 8" 2'’ I*O 0 Avg. 0 0 0 0 0 0 0 0 0 3 13 3 Avg. 1 1 1 1 3 2 1 1 1 1 1 0 Rep 2 5 Avg. 2* 2' 2* l'’ l'’ 2“ T 1'’ 2bc 5 “ 2'’ p 7 Avg. 1 1 1 2 l'’ 3 * l'’ 1'’ P 8* p 10 1 2 2 2 r 8* 3b P P 9a 4b 0' 0 Avg. 0 0 0 0 0 0 0 0 0 _ 3 0 0 1 1 1 1 1 1 1'’ 2* 1'’ 1'’ Rep 1 5 Avg. 2* 2* l^* o'’ 1 1 1 0 2'’ T 1'’ 0' 7 Avg. V 2* 2^ O'* 2'’ 4‘ 3 '’ o'’ 3 ' 9a 6'’ I** -<