LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE

 

 

 

 

 

 

W5
MAY 1 2 2007

 

 

MAR 3 0 2013

 

h327 13

 

 

 

 

 

 

 

6/01 c-JCIRG!DatoDuo.p65-p.15

EFFECT OF WATER ACTIVITY AND HUMIDITY ON THE THERMAL
IN ACTIVATION OF SALMONELLA DURING HEATING OF MEAT

By

Tausha Rene’ Carlson

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements for the degree of
Master of Science

Department of Food Science and Human Nutrition

2002

ABSTRACT

EFFECT OF WATER ACTIVITY AND HUMIDITY ON THE THERMAL
IN ACTIVATION OF SALMONELLA DURING HEATING OF MEAT

By

Tausha Rene’ Carlson

The USDA-FSIS recently amended the regulations governing cooked meat and
poultry products, creating a shift to lethality performance standards, and a need for
inactivation models. Studies clearly show that many factors affect thermal inactivation of
pathogens; however, water has not been previously isolated as an intrinsic or extrinsic
factor. The objectives of this study were (1) to test the effects of meat moisture
content/water activity on thermal inactivation of Salmonella in a sealed environment, (2)
to test the effects of air humidity on thermal inactivation of Salmonella during convection
heating, and (3) to demonstrate the inclusion of a water term into a secondary
inactivation model. Ground turkey was inoculated with an 8-strain Salmonella cocktail
and heated isothermally either in a waterbath or in air convection oven. Survivors were
enumerated via serial dilutions and plated on Petrifilm®. The rate of thermal inactivation
of Salmonella decreased with decreasing meat water activity; however, in the air
convection oven, the same results were not observed for a corresponding decrease in
relative humidity. In conclusion, the water effect lies in the intrinsic property of the meat
(i.e., water activity), rather than the extrinsic process parameter (i.e., humidity), and
should be accounted for in inactivation models used to validate commercial convection

cooking systems.

ACKNOWLEDGEMENT

I would like to thank my major professor, Dr. Bradley Marks, for his constant
support, guidance, and positive attitude during my time here at Michigan State
University. I am truly blessed for having the opportunity to work for him.

I would also like to thank my committee members, Dr. Al Booren and Dr. Elliot
Ryser, for their useful insight. A big thanks to my lab group: Dr. Alicia Orta-Ramirez,
Ms. Kerri L. Harris, Mr. Scott Millsap, Mr. Adam Watkins, and Mr. Sang Jeong.

Also, thanks to my parents, Danny Burns and Jamie Burns, and brothers, Blake
and Bradley, for their unconditional love and support.

Finally, I would like to thank my husband, Chris Carlson, for making my short

time, and long winters, in Michigan all worthwhile. I love you.

iii

TABLE OF CONTENTS

LIST OF TABLES ................................................................................. vii
LIST OF FIGURES ............................................................................... viii
ABBREVIATIONS ................................................................................. x
CHAPTER 1: INTRODUCTION ................................................................ 1
1.1 Background and justification ...................................................... 1

1.1.] F oodborne disease ................. ‘ ........................................ 1

1.1.2 The aflected industry ...................................................... 2

1.1.3 Regulatory trends .......................................................... 3
1.1.4 Scientific needs ............................................................. 4

1.2 Hypothesis and objectives ........................................................... 5
CHAPTER 2: LITERATURE REVIEW ...................................................... 6
2.1 Salmonella .............................................................................. 6

2.2 Thermal inactivation modeling ................................................... 7

2. 2. 1 Primary models ............................................................ 8

2.2.2 Secondary models .......................................................... 9

2.3 Factors affecting thermal resistance ............................................. 11

2.3.1 Pathogen species and strains ............................................ 11

2.3.2 Inactivation media ........................................................ 11

2.3.3 Fat content ................................................................ 12

2.3.4 pH ........................................................................... 15

2.3.5 Salts and other common additives ..................................... 16

2.3.6 Water .............................. - ......................................... 1 7

2.3.6.1 Water activity ..................................................... 18

2.3.6.2 Humidity ......................................................... 20

CHAPTER 3: MATERIALS AND METHODS ............................................ 23
3.1 Overview .............................................................................. 23

3.2 Part 1 — Moisture effects-high range ............................................. 24

3.2.1 Inoculum .................................................................. 24

3.2.1.1 Bacterial strains .................................................. 24

3.2.1.2 Culture preparation ............................................. 25

3.2.2 Meat ........................................................................ 25

3.2.2.1 Ground turkey preparation ..................................... 25

3.2.2.2 Moisture content alteration ..................................... 26

3.2.3 Inoculation ................................................................ 27

3.2.4 Thermal inactivation ..................................................... 28

3.2.5 Enumeration ............................................................... 29

3.2.6 Statistics and modeling ................................................... 29

iv

3.3 Part 2 — Moisture effects-low range .............................................. 3O

3. 3.1 Inoculum .................................................................. 30
3.3.1.1 Bacterial strains ................................................. 30

3.3.1.2 Culture preparation ............................................. 30
3.3.2 Meat ........................................................................ 30

3.3.2.1 Ground turkey preparation and moisture content

alteration .......................................................... 30

3.3.2.2 Decreasing the particle size .................................... 32
3.3.3 Inoculation ................................................................ 32
3.3.4 Thermal inactivation ..................................................... 33
3.3.5 Enumeration .............................................................. 34
3.3.6 Statistics and modeling .................................................. 34
3.4 Part 3 —Humidity effects34
3. 4.1 Inoculum .................................................................. 35
3.4.1.1 Bacterial strains .................................................. 35

3.4.1.2 Culture preparation ............................................. 35

3.4.2 Meat ........................................................................ 35
3.4.2.1 Ground turkey preparation ..................................... 35
3.4.3 Inoculation ................................................................ 35
3.4.4 Thermal inactivation ..................................................... 36
3.4.5 Enumeration .............................................................. 38
3.4.6 Statistics and modeling .................................................. 38
CHAPTER 4: RESULTS AND DISCUSSION ............................................. 40
4.1 General background information ................................................ 40
4.1.1 Salmonella cocktail ...................................................... 40
4.1.2 Proximate composition .................................................. 41
4.1.3 Initial counts .............................................................. 43
4.1.3.1 Part 1 — Moisture effects-high range .................. . ....... 43
4.1.3.2 Part 2 — Moisture effects-low range ........................... 43
4.1.3.3 Part 3 — Humidity effects ....................................... 43
4.1.4 Inoculum distribution .................................................... 43
4.1.5 Thermal lag times ......................................................... 44
4.1.6 Additional test information for Part 1 .................................. 44
4.1.6.1 Changes during the experiment ............................... 44

4.1.6.2 Moisture content alteration ............................. ' ....... 45

4.1.7 Additional test information for Part 2 ................................ 45
4.1.7.1 Moisture content alteration .................................... 45

4.1.8 Additional test information for Part 3 ................................. 45
4.1.8.1 Moisture lost during heating ................................... 45
4.1.8.2 Oven adjustments ................................................ 45

4.2 Inactivation results .................................................................. 46
4.2.1 Part 1 — Moisture eflects-high range. . . .. ............................. 46
4.2.1.1 Data ............................................................... 46
4.2.1.1.1 Low fat .............................................. 46

4.2.1.1.2 High fat .............................................. 48

4.2.1.2 ANOVA .......................................................... 50

4.2.1.2.1 Raw data ............................................. 50

4.2.1.2.2 K values ............................................. 53

4.2.2 Part 2 - Moisture effects-low range ................................... 54
4.2.2.1 Data ............................................................... 54
4.2.2.2 ANOVA .......................................................... 55
4.2.2.2.] Raw data ............................................. 55

4.2.2.2.2 K values ............................................. 56

4.2.2.3 Modeling ......................................................... 61

4.2.3 Part 3 — Humidity effects ................................................ 62
4.2.3.1 Data ............................................................... 62
4.2.3.2 ANOVA .......................................................... 63
4.2.3.2.1 Raw data ............................................. 63

4.2.3.2.2 K values ............................................. 66

4.2.4 Parts 1 and 2 combined .................................................. 67
4.2.4.1 Data ............................................................... 67
4.2.4.2 ANOVA .......................................................... 68
4.2.4.2.1 Raw data ....................... . ...................... 68

4.2.4.2.2 K values ............................................. 69

4.2.5 Parts 1 and 3 combined .................................................. 70
4.2.5.1 Data ............................................................... 70
4.2.5.2 ANOVA ........................................................... 71
CHAPTER 5: CONCLUSIONS ............................................................... 72

CHAPTER 6: RECOMMENDATIONS FOR FUTURE RESEARCH.... . ........74

APPENDICES ...................................................................................... 76
APPENDIX A: Moisture lost during oven heating (Part 3) ......................... 77
APPENDIX B: Inactivation data (Part 1-3) ............................................ 81
APPENDIX C: Output from statistical analyses (Part 1-3) .......................... 85
APPENDIX D: Output from secondary modeling (Part 2) ........................ 102
APPENDIX E: Initial microbial counts (Part 2) ..................................... 105

REFERENCES ................................................................................... 106

vi

LIST OF TABLES

TABLE 2.] Effect of fat on thermal inactivation of vegetative cells ........................ 13
TABLE 3.1 Summary of experimental design ................................................. 24
TABLE 4.1 Salmonella counts ................................................................... 41
TABLE 4.2 Proximate composition of ground turkey ........................................ 42
TABLE 4.3 Oven settings for inactivation trials ............................................... 46

TABLE 4.4 P values from analyses of variance of the raw data in high fat and low fat
ground turkey at varying moistures, fats, times, and temperatures ........................... 52

TABLE 4.5 P values from analyses of variance of the k values in high and low fat
ground turkey at varying moistures, fats, times, and temperatures ........................... 54

TABLE 4.6 P values from analyses of variance of the raw data in low fat ground turkey
at varying moistures, water activities, and times, at a sample temperature of 60 °C ...... 56

TABLE 4.7 P values from analyses of variance of the k values in low fat ground turkey
at varying moistures, water activities, and times, at a sample temperature of 60 °C ...... 56

TABLE 4.8 k values (min'l) as a function of water activity ................................. 60

TABLE 4.9 Effect of the “water term " form on the root mean square error for a first-
order, modified Arrhenius-type model ............................................................ 62

TABLE 4.10 P values from analyses of variance of the raw data in low and high fat
ground turkey at varying relative humidities, at 60 °C ......................................... 64

TABLE 4.11 P values from analyses of variance of k values in low and high fat ground
turkey at varying relative humidities, at 60 °C ................................................... 66

TABLE 4.12 P values from analyses of variance of the raw data in low fat ground
turkey (raw vs. cooked) at similar moisture contents and water activities, and heated at

60 °C ................................................................................................... 69

TABLE 4.13 P values from analyses of variance of k values in low fat ground turkey
(raw vs. cooked) at similar moisture contents and water activities, and heated at

60°C ................................................................................................... 7O

vii

LIST OF FIGURES

FIGURE 3.] Meat samples rolled between two guides to achieve a uniform thickness..
.............. 24

FIGURE 3.2 Smokehouse operation schedule-temperature (0F) ........................... 31
FIGURE 3.3 Meat sample being spread to a uniform thickness onto a sterile screen. . .36
FIGURE 3.4 Meat sample entering the custom air convection oven ....................... 36

FIGURE 3.5 Sample in the heating chamber on a stand, which allows air to be blown
across the top and bottom surface of the sample ................................................ 37

FIGURE 4.1 Thermal inactivation of Salmonella in low fat ground turkey at 55 °C and

three different moisture contents (72.3-76.3 0/0) ................................................. 47
FIGURE 4.2 Thermal inactivation of Salmonella in low fat ground turkey at 60 0C and
three different moisture contents (72.3 -7 6.3 %) ................................................. 47
FIGURE 4.3 Thermal inactivation of Salmonella in low fat ground turkey at 65 °C and
three different moisture contents (72.3-76.3 %) ................................................. 48
FIGURE 4.4 Thermal inactivation of Salmonella in high fat ground turkey at 55 °C and
three different moisture contents (64.5-68.5%) ................................................. 48
FIGURE 4.5 Thermal inactivation of Salmonella in high fat ground turkey at 60 °C and
three different moisture contents (64.5 -68.5 %) ................................................. 49
FIGURE 4.6 Thermal inactivation of Salmonella in high fat ground turkey at 65 °C and
three different moisture contents (64. 5 —68.5 0/0) ................................................. 49
FIGURE 4.7 Thermal inactivation of Salmonella in low fat ground turkey at 60 °C and
three diflerent moisture contents (3 7. 1 -72.5%) ................................................. 55
FIGURE 4.8 Comparison of k values as a function of water activity ...................... 60

FIGURE 4.9 Thermal inactivation of Salmonella in low and high fat ground turkey at
60 °C and 90 and 96% relative humidity ......................................................... 63

FIGURE 4.10 Thermal inactivation of Salmonella in low fat ground turkey at 60 °C in
raw vs. cooked meat ................................................................................. 68

viii

FIGURE 4.11 Thermal inactivation of Salmonella in low fat ground turkey at 60 0C in
waterbath vs. air convection oven ................................................................ 71

ix

ANOVA
AVG

aw

b(T)

C0 to C4
CF U

-dN/dt

HF

LF

MC

n(T)

N(t)/No

SD

ABBREVIATIONS

frequency factor (Arrhenius equation)

slope of a line

analyses of variance

average

water activity

temperature dependent constant (Peleg and Cole, 1998)
empirical coefficients without biological significance (Cerf et al., 1996)
colony forming units

rate of inactivation of viable cells

activation energy

high fat

death rate (min'l)

low fat

moisture content

number of surviving cells

initial number of cells

temperature dependent constants (Peleg and Cole, 1998)
survival ratio

universal gas constant

standard deviation

absolute temperature (K)

temperature when the line was extrapolated to k=O (Zweitering et al.,
1990)

time

xi

CHAPTER 1

INTRODUCTION

1.1 Background and justification

In this section, four main points are emphasized. First, background
information on Salmonella is presented. Then, the food industry, and more specifically
the meat industry, is discussed. Thirdly, current changes in the federal regulations
affecting the industry are addressed. Finally, the resulting scientific needs are described.
1.1.1 F oodborne disease

Campylobacter, Salmonella, and Escherichia coli 0157:H7 are the most
commonly recognized causes of foodbome illness in the US (CDC, 2001). Over 2000
Salmonella strains have been identified (Jay, 1996). According to the Centers for
Disease Control and Prevention (CDC, 2001), there are 1.4 million cases of salmonellosis
in the United States per year, and of these, approximately 40,000 are culture-confirmed
cases that are reported to the CDC (CDC, 2000). People infected with Salmonella
develop fever, abdominal cramps, and diarrhea (sometimes bloody), which occurs 12-72
h after exposure and usually lasts 4-7 days (CDC, 2000). Most people recover without
treatment, although severe cases require hospitalization, and over 500 peOple die each
year in the United States from acute salmonellosis (CDC, 2000). Additionally, cases of
human salmonellosis impose a considerable economic burden on the economy. This
responsibility falls upon the industry (retail and wholesale), the infected people, and their

family (Roberts and Sockett, 1994).

Turkey is one of the most common vectors for pathogens, and Salmonella is one _
of the most prevalent pathogens found in turkey. FSIS reported combined prevalence
(small and large plants) of Salmonella from July 1999 to June 2000; broiler chicken was
9.9%, ground chicken was 14.4%, ground beef was 5.0%, and ground turkey was 30.0%
(USDA-FSIS, 2000).

Thermal processing is the main solution to eliminating bacteria in food products.
Salmonellae are obviously sensitive to heat, but their sensitivity varies greatly. The
composition of the heating menstrum has a strong influence on the thermal resistance of
bacteria (Murphy et al., 2000). Occasionally, some salmonellae may survive standard
food-processing techniques (Doyle and Mazzotta, 2000). This may result from outside
factors that affect the thermal resistance. In addition, some strains of Salmonella are
more heat resistant that others. Because of the various factors that affect thermal
resistance, the need exists to evaluate inactivation in meat and not rely on data developed
in model substrates.

1.1.2 The affected industry

The food industry is generally considered the nation’s largest manufacturing
sector and is one of the most stable. The meat and poultry industry contributes over $90
billion in annual sales to the US. Gross National Product (GNP) and is the largest
component of the US. agriculture sector (AMI, 2000). The United States Department of
Agriculture-Food Safety and Inspection Service (USDA-FSIS) (2001) reported that there
were 1,630 establishments producing ready-to-eat cooked or partially cooked meat and

poultry product in 1997 with the value of shipments totaling over $28.2 billion for that

year. Given consumer preferences for convenience, it is likely that the market for fully-
cooked products will continue to grow.

The focus of thermal processing is placed in three areas: 1) cooking methods in
homes and commercial kitchens; 2) processing methods in plants producing fully cooked
products; and 3) treatment of raw poultry (Doyle and Mazzotta, 2000). This thesis
focuses on the processing methods in plants producing fully cooked products.

1.1.3 Regulatory trends

Regulations are aimed to ensure that pathogens are destroyed and not present in
food products. For whole muscle products, the regulatory paradigm has shifted from
command-and-control regulations to performance standards (USDA-PSIS, 1999).
Performance standards require that commercial establishments meet specific food safety
objectives. USDA-FSIS has set regulations in Title 9 of the Code of Federal Regulations
for meat and poultry. The regulation states that any thermal processing procedure must
achieve 7.0- or 6.5-log10 reduction in Salmonella for whole-muscle poultry or beef,
respectively. Processors are not held to specific endpoint temperatures; however, they
must validate new or altered process schedules by “scientifically supportable means”
(USDA-FSIS, 1999).

A proposed regulation would extend these standards to all ready-to-eat products.
This regulation allows either challenge studies (i.e., inoculation of real products with
target organisms) or the use of models to document process lethality (USDA-F SIS,
2001). This regulation is advantageous because it allows flexibility in processing
procedures. However, this creates a problem, because pathogens cannot be intentionally

brought into processing facilities to conduct challenge studies. Furthermore, most

models are based on microbial thermal death time studies performed in a laboratory and
may not be valid for commercial processes. In regard to models, the regulation states,

“The establishment will need to demonstrate the relationships

between the lethality treatments and the specific characteristics of

a product, such as physical and chemical properties. This

demonstration could involve the use of heat transfer equations and

should account for all variables that would affect lethality (e.g.,

size of product, humidity, density, thermal conductivity, specific

heat, shape, product composition and strain of organism” (USDA-

F SIS, 2001).
1.1.4 Scientific needs

Studies clearly show that nearly “all variables,” including fat, salts, pH, and
additives (Chapter 2), affect the thermal inactivation of bacteria,. However, no current
inactivation model accounts for the effects of water (i.e., moisture content, water activity,
or humidity) on microbial inactivation in meat products. Water affects the lethality of
Salmonella in meat products, and more organisms generally survive in a dry
environment. However, the specific cause of the effect is unknown. Further research is
needed to determine whether the effect is best related to moisture content, water activity,
or process humidity. This effect must be fully understood to accurately model process
lethality for commercial systems. Incorporating accurate terms into a secondary model
would improve model performance and usefulness. Therefore, due to the regulatory
changes and economic importance of this industry, there is a need to directly test these

water effects.

1.2 Hypothesis and objectives
The hypothesis of this study was that the rate of thermal inactivation for
Salmonella decreases with decreasing meat moisture content and/or process humidity.
The objectives of this study were:
(1) To test the effects of meat moisture content/water activity on thermal
inactivation of Salmonella in a sealed environment,
(2) To test the effects of air humidity on thermal inactivation of
Salmonella during convection heating, and
(3) To demonstrate the inclusion of a water term into a secondary

inactivation model.

CHAPTER 2

LITERATURE REVIEW

2.1 Salmonella

Salmonellae are a small, gram-negative, non-spore forming rod shaped bacteria
that cause foodbome gastroenteritis (Jay, 1996). They are widely distributed in nature
and humans, with the intestinal tract of domestic livestock and wild animals being their
primary habitat (Jay, 1996). salmonellae are excreted in feces, then transmitted to other
living creatures in a variety of ways. The most common vectors associated with
salmonellosis in humans are eggs, poultry, and meat products (Jay, 1996).

The temperature range for growth of salmonellae is between 5.5 and 45°C (Ng et
al., 1969). The temperature where the salmonellae begin to die and the maximum
temperature for growth depend on the strain, growth phase, food composition, test media,
other physical conditions, and competing microflora (Doyle and Mazzotta, 2000). The
pH for optimum growth is between 6.6-8.2, with values greater than 9.0 and less than 4.0
being bactericidal (Jay, 1996). Regarding moisture, Salmonella growth inhibition in
laboratory media (pH 7.0) has been reported at water activity values below 0.94 (Jay,
1996). Due to variations in these parameters, it is often difficult to compare data from
experiments using different conditions.

With only a few exceptions, most studies on pathogens in poultry were conducted
with single strains. However, a ‘real’ process is not necessarily limited to one strain,
because various pathogens may be concurrently encountered in products. Therefore,

regulations require that data and/or models used to document compliance be based on a

combination of Salmonella serotypes, referred to as a cocktail. The USDA does not
specify the serotypes to be used, but says that any blend should include strains that have
been implicated in foodbome outbreaks as well as strains that show fairly high heat
resistance (USDA-F SIS, 1999). Different cocktails result in different model parameters;

however, this problem could be eliminated if a universal cocktail were defined.

2.2 Thermal inactivation modeling

Predictive microbial models are mathematical representations of the growth,
survival, or inactivation of microbial populations. Such models can be used to describe
the behavior of microorganisms under different physical or chemical conditions. As
stated by Zwietering et a1. (1990), “these models allow the prediction of microbial safety
or shelf life of products, the detection of critical parts of the production and distribution
process, and the optimization of products and distribution chains.”

To be of practical value, predictive microbial models must account for the effects
of time and the various intrinsic and extrinsic factors affecting the microbial response.
Whiting and Buchanan (1993) classified microbial models into primary, secondary, and
tertiary types. Primary models describe the response of the microorganism with time to a
single set of conditions. Each population vs. time curve can be described by a set of
specific values for each of the parameters in the model (Whiting and Buchanan, 1993).
Secondary models describe the response of one or more parameters of a primary model to
changes in one or more of the cultural conditions (Whiting and Buchanan, 1993). These
models calculate the changes in primary model parameters with respect to changes in

temperature, pH, water activities, etc. (Whiting and Buchanan, 1993). Tertiary models

are computer programs that calculate microbial responses to varying conditions, compare
the effects of the conditions, or contrast the behavior of several microorganisms (Whiting
and Buchanan, 1993). Tertiary models make primary and secondary models “user-
friendly.” 7
2.2.1 Primary model

Several means are available to describe the relationship between microbial
populations and time during thermal inactivation, including reaction kinetics analogies,
simple D-values, and population-based models.

Chick (1908) proposed the following model:

N=Noe’k‘ (1)
where N0=the initial number of cells, N=the number of surviving cells, t=exposure time,
and k=death rate. The instantaneous rate of inactivation of viable cells is proportional to
the number of viable cells present at that time (Chiruta et al., 1997).

dN/dt=-kN (2)
where (-dN/dt)=rate of inactivation of viable cells, N=the number of surviving cells, and
t=time. According to this model, when bacteria are exposed to a constant temperature,
microbial death occurs following the kinetics of first-order reactions. Taking the
logarithm of equation (1) yields:

ln(N/No)=-kt, (3)
which is a log-linear equation with a slope of k, with k depending on factors such as
temperature, pH, or water activity.

The thermal reduction time, or “D-value,” describes the time dependence of

bacterial destruction at a given condition. Similar to reaction kinetics analogies, D-values

represent first-order, log-linear reduction models. The D-value is the time required to
decrease a bacterial population by 90% at a given temperature. When the D-value
increases, the culture becomes more heat resistant. From the equation above, the D—value
can be calculated as (Chiruta et al., 1997):

D=2.303/k (4)
where k=inactivation rate constant from equations 1 and 2. This measurement is often
used, but the variability among reported values is high, depending on the organism and
conditions. Also, this method has been criticized, because it can be confusing or can
obscure what should be simple mathematics of a first-order equation (Chiruta et al.,
1997). However, because D has the dimension of time, it is often better understood (than
k) in the food industry.

An example of a population-based model, where a non-linear relationship occurs,
is the Weibull distribution. Depending on the data, it can have a downward or upward
concavity, a “shoulder,” or sigmoidal shape (Peleg and Cole, 1998). Population-based
models assume that each cell in a bacterial population has a discrete resistance to thermal
inactivation. If resistance follows a Weibull distribution, then the number of survivors
can be modeled via the following model (Peleg and Cole, 1998):

10310[N(t)/No]='b(T)tn(T) (5)
where N(t)/No=survival ratio, and b(T) and n(T) are temperature dependent constants.
2.2.2 Secondary models

Various types of secondary models include Arrhenius, extended Arrhenius, and

square-root. While these are just a few of the most common secondary models, many

other secondary models (of various forms) exist that account for a variety of parameters.

The effect of temperature on the rate of microbial inactivation is often described

using the Arrhenius equation:

k=A (5‘3”T (6)
where A=frequency factor, Ea=activation energy, R=universal gas constant, and
T=absolute temperature. However, this model only accounts for temperature, and it has
been recognized for decades that other factors affect the death rate; however, few
attempts have been made to develop multifactorial models.

Reichart (1994) was the first to consider water activity in a semi-empirical model
for thermal inactivation of E. coli. Shortly after, Cerf et a1. (1996) proposed another five-
parameter, extended Arrhenius, model from the experimental data of Reichart (1994).
The Cerf model extends Davey’s (1978) model, and includes other parameters. The Cerf
et a1. (1996) model is as follows:

ln(k)=C0+(C1/T)+C2pH+C3pH2+C4aw2 (7)
where T=absolute temperature, and Co to C4 are empirical coefficients without direct
biological significance.

The square-root or Belehradek model is typically used for growth models, and is
based on the linear relationship between the square-root of the grth rate and
temperature (Zwietering et al., 1990). Biological zero, the value for temperature when
the growth rate was extrapolated to zero, was introduced here. The simplest version of
the model for temperatures below the optimum grth rate is:

\lk=a(T-TO) (8)

10

where k is the growth rate or other rate term, such as the reciprocal of the lag time, To is j
the temperature when the line is extrapolated to k=0, and a is the slope (Zweitering et al.,

1990).

2.3 Factors affecting thermal resistance

Variables affecting heat resistance of pathogens in meat include species, pH, fat
content, salts, and other environmental factors (Jay, 1996). In addition, experimental
approaches, serotypes, grth media, and enumerating procedures vary among
laboratories, and this makes comparison difficult and causes data to be relevant only to
the particular commodity tested (Skinner et al., 1994; Doyle et al., 2001; Doyle and
Mazzotta, 2000).
2.3.1 Pathogen species and strains

Heat resistance differences among species and strains exist (Doyle and Mazzotta,
2000); for the purpose of this literature review, various pathogens are examined.
2.3.2 Inactivation media

Salmonella tends to be more thermally resistant in actual food products than in
laboratory media (Murphy et al., 2002); moreover, food type also affects resistance
(Ahmed et al., 1995; Murphy et al., 2002). Numerous studies show that bacteria are more
resistant to heat when tested in food than in laboratory media (Doyle et al, 2001).

Bacteria attached to muscle tissue are more heat resistant than bacteria suspended
in liquid media (Murphy et al., 2002). Murphy et a1. (2000) compared D-values in meat

to those in a semi-liquid medium and found that the D-values were higher in ground

11

chicken breast than in a peptone-agar solution at 55 to 70°C. Therefore, there is a need to
evaluate Salmonella inactivation in meat and not rely on data (only) from model media.
2.3.3 Fat content

Fat content influences the thermal resistance of microorganisms in meat;
however, some inconsistencies have been observed. Some studies have shown higher D-
values in high fat meat, while other studies'have shown the opposite (Table 2.1).

However, in general, inconsistent trends between fat content and D-values have been

reported in the literature.

12

TABLE 2.1 Effect of fat on thermal inactivation of vegetative cells.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Organism Product Reference D-value (min) Temp (°C) Fat (7.)
70.41 50
6.37 55 3
turkey 0.55 60
115 50
9.69 55 11
E. coli 0157:H7 Ahmed et al., 0.58 60
1995 65.24 so
8.76 55 3
. 0.38 60
chicken 105.5 50
9.74 55 11
0.55 60
78.2 51.7
4.1 57.2 2
. Line et al., 0.3 62.8
E. CO/I O157.H7 beef 1991 115.5 51.7 p
5.3 57.2 30.5
0.5 62.8
423* 52
12.5’ 55 3
2.8' 57
E. coli 0157:H7 turkey 53:33: $337 32:5" :3
11* 55
2.4' 57 11
0.9* 60
81 .3* 51.7
Listeria 2.6 57.2 2
monocytogenes beef Falpggt1al., 7(1):. 2?:
Scott A ' .
5.8 57.2 30.5
1.2 62.8

 

* An increased D-value was not observed with increased fat content.

 

In some studies, D-values for pathogens were higher in high fat meat than in low
fat meat. According to Line et al. (1991), D-values for E. coli 0157:H7 in beef increased
in the heating range of 52 to 63°C as the fat content increased from 2.0 to 30.5%. Ahmed
et al. (1995) used a single strain of E. coli 0157:H7 and found that as the fat content

increased (3-30%) in different meat products (chicken, turkey, beef, and pork sausage),

l3

the D—values increased. Fain et al. (1991) inoculated ground beef with Listeria
monocytogenes and generally found that D-values increased as the fat content increased
(2-30.5%); however, this did not hold true on one occasion. Ben-Embarek and Huss
(1993) also reported higher D-values for L. monocytogenes in salmon than in cod and
attributed the greater heat resistance in salmon to the higher fat content.

Several explanations were given as to why D-values increased as the fat content
increased. Ahmed et al. (1995) stated that the higher D-values were likely due to the
decreased moisture content of the meat. They claimed that bacteria suspended in fat are
more difficult to destroy than in aqueous medium, due to a reduction of water activity.
Veeramuthu et al. (1998) observed higher D-values for S. Senftenberg in turkey
containing increased levels of fat and attributed this finding to the effect of fat on water
activity.

However, other authors did not find fat content to be a significant factor. Kotrola
and Conner (1997) did not see an increase in D-values for E. coli 0157:H7 as the fat
content increased in ground turkey, with the opposite being observed. Kotrola and
Conner (1997) reported D-values at 55°C ranging from 12.5 (3% fat) to 11 (11% fat) min
at 60°C. J uneja and Eblen (2000) found that the D-values of an 8-strain Salmonella
Typhimurium DT 104 cocktail in ground beef decreased with increasing fat content.
Maurer (2001) observed that higher fat levels significantly affected the D-value of S.
Senftenberg in turkey; however, no significant effect was observed with E. coli 0157:H7
in turkey or beef, or with a Salmonella cocktail in beef.

Several explanations were given as to why D-values decreased as the fat content

increased. Kotrola and Conner (1997) explained that finely grinding the meat and fat

14

together before heating could have affected the dispersal of the fat in the meat and
allowed it to emulsify. This could, in turn, have increased the solubility of water in fat
before the product was heated. Olson and Nottingham (1980) attributed not seeing an
increase in D-values with increasing fat content to a protective effect in higher fat
products.
2.3.4 pH

The pH describes the hydrogen ion concentration [Hi], and is often recognized as
one of the most important factors influencing the heat resistance of bacteria. Juneja and
Eblen (1999) showed that as the pH decreased, the D-values for L. monocytogenes
decreased. Abdul-Raouf et al. (1993) showed that E. coli 0157:H7 was less heat stable
in acidified ground beef slurries, as compared to non-acidified slurries, with stability
dependent on the type of acid used.

Davey et al. (1995) found that pH significantly affected the thermal inactivation
rate for E. coli. When experiments were performed in a test carrier liquid over a
temperature range of 54 to 62°C, the influence of pH was most significant at the lower
temperatures. Overall, D-values were highest at pH 7, and decreased as pH was reduced
below 7. Chiruta et al. (1997) tested the effect of pH on the rate constant for thermal
inactivation and generally found results consistent with Davey et al. (1995) for E. coli, L.
monocytogenes, and P. fluorescens. Temperatures ranged from 52 to 62°C, with the
effect most significant at the lower temperatures.

However, Foster and Hall (1991) showed that S. Typhimurium could be induced

to survive under more acidic conditions than expected. Also, Farber and Pagotto (1992)

15

demonstrated that HCl acidification actually increased thermal resistance of L.
monocytogenes.
2.3.5 Salts and other common additives

Salts, lactates, and phosphates are common additives in meat products. Primary
functions of salt in meat products are: (1) to solubilize muscle proteins to assist in
binding meat, moisture and fat; (2) to serve as a flavoring agent; and (3) to inhibit
growth of foodbome pathogens (Pearson and Gillett, 1996). In most cases, salt appears
to act as a protective agent, resulting in higher heat resistance, but this does not always
hold true.

Thermotolerance can be increased by incorporating salt or curing salt mixtures
(Juneja and Eblen, 1999). Juneja and Eblen (1999) found that by adding NaCl, this
protected L. monocytogenes against heat inactivation in beef gravy at 55 to 65°C. D-
values increased 2- to 5-fold after curing salts were added to meat (J unej a and Eblen,
1999). Maurer et al. (2000) found that the D-values for Salmonella increased as the salt
content increased from 0 to 2% in ground turkey.

Additional studies have assessed various combinations of additives. Kotrola and
Conner (1997) found that both sodium chloride and sodium lactate enhanced survival of
E. coli 0157:H7 in cooked turkey meat as compared to meat without additives at 52 to
60°C, with the highest D-values (greatest survival) observed when three additives
(sodium chloride, sodium lactate, and polyphosphate) were added to the turkey. The
authors attributed this increase to the reduction of water activity caused by the additives

binding water in the heating medium.

16

Other food additives, such as bacteriocins, EDTA, polyphosphates, hydrogen
peroxide and the lactoperoxidase system, make Salmonella more heat sensitive (Doyle
and Mazzotta, 2000). The effectiveness varies depending if the additive is in culture
media or a complex food, because it may interact with fat and protein and thereby be less
available to interact with bacterial cells (Doyle and Mazzotta, 2000). Goepfert et al.
(1970) and Corry (1975) tested several solutes and found that heat resistance of bacteria
varied widely with different solutes at the same water activities. Overall, sucrose had the
greatest protective effect, compared to glycerol, glucose, polyethylene glycol (Goepfert et
al., 1970) and glucose, fructose, sorbitol, and glycerol (Corry, 1975).

Because solutes and other additives affect the thermal resistance of bacteria, tests
should be run specific to the meat product and solute and/or additive of interest.

2.3.6 Water

Water is essential for all living processes. Due to its chemical and physical
properties, water is so unique that it is often considered one of the most important
compounds on earth (Gailani and Fung, 1987).

Water availability has an influence on the heat resistance of Salmonella in meat
products. However, the specific cause of the effect is unknown. There are several ways
to quantify water in a food system. Many studies have looked at meat moisture content
or water activity (intrinsic parameters), and a few have looked at humidity (extrinsic
parameter). This section will further investigate this issue.

Throughout the literature, a common theme is seen regarding various other factors
affecting thermal inactivation. Authors have often attributed the effects of other

parameters (specifically fat and salt) to changes in water activity (Blankenship, 1978;

17

Ghazala et al., 1995; Kotrola and Conner, 1997; Shelef and Yang, 1991; and O’Donovan
and Upton, 1999); however, others suggest that changes in water activity do not
completely explain the effects of these other factors (O’Donovan and Upton, 1999). The
present study includes tests specifically aimed at testing the impact of water activity on
thermal inactivation without changing other factors.
2.3.6.1 Water activity

Water activity (aw) describes the amount of available water and is defined as:

aW=Pi/Po (9)

where Pia—vapor pressure of water in equilibrium with the material, and Po=vapor pressure
of pure water at the same temperature (Gailani and Fung, 1987). Water activity controls
the movement of water between a food product and the environment (Gailani and Fung,
1987). The range of water activity for high moisture foods is 0.9 to 1.0 (Gailani and
F ung, 1987), with meats classified as high moisture foods.

In general, as water activity decreases, thermal fesistance of pathogens increases.
However, most studies have been performed in sugar solutions rather than in actual food
systems. As discussed in section 2.3.2, resistance varies depending on the media used;
therefore, it is crucial to perform the studies in actual food products.

Goepfert et al. (1970) studied the effect of water activity in sucrose solutions
(0.87-0.99), and found that heat resistance of Salmonella always increased as the water
activity of the heating menstruum (0.75—0.99) decreased. Riemann (1960) also
documented increased heat resistance with decreased water activity.

Cerf et al. (1996) used Reichart’s (1994) experimental data for thermal

inactivation of E. coli at both constant temperature (isothermic) and constantly varying

18

temperature (anisotherrnic). The tests were performed using laboratory media with
glycerol added to distilled water to reach the targeted water activities (Reichart, 1994).
The isothermic data encompassed the following conditions: 58°C, pH 3-9, and water
activity 0928-0995. The anisotherrnic data contained the same parameters, except the
temperature ranged between 52-63°C. Cerf et al. (1996) claimed that the additive, linear
Arrhenius model accurately predicted the combined effect of sterilizing temperature, pH,
and water activity on the thermal inactivation of E. coli. Cerf et al. (1996) suggested that
these models could be extrapolated over a limited range of environmental values;
however, sufficient published and independent data to test this were lacking. The
isothermic model (58°C) was as follows:
ln(k)s'l=-6.021-2.377pH+0.1994pH2+8.997aw2‘ (10)
The anisothermic model was as follows:
ln(k)s'l=86.49—0.3028*l0'5/T-0.5470pH+0.0494pH2+3.067aw2 (11)
O’Donovan—Vaughan and Upton (1999) investigated the survival of Salmonella
Typhimurium in four different carbohydrate solutions (glycerol, sucrose, glucose, and
polyethylene glycol) at three different water activities (0.45, 0.70, and 0.90). They found
that as the water activity of the solution was reduced, the heat resistance increased (55
and 65°C). Additionally, heat resistance depended on the nature of the solute used to
reduce the water activity; sucrose gave the greatest protection. The conclusion was that
the heat resistance depended on the solute used to reduce the water activity; however, this

result was not entirely consistent in the data reported.

19

2.3.6.2 Humidity

While water activity is the means to quantify the state of water in a food product,
humidity is the means to quantify the water state in the environment. Nevertheless, only
limited research has focused on evaluating the effects of process humidity on thermal
inactivation of foodbome pathogens.

Kirby and Davies (1990) evaluated humidity effects in a non-food system.
Salmonella Typhimurium LT2 received dehydration treatment by being placed in an
atmosphere controlled by a saturated salt solution of sodium bromide (BHD) (57%
equilibrium relative humidity (ERH)) at 37°C for 48 h, with this dehydration treatment
continued for up to 34 d (Kirby and Davies, 1990). After being heated at 135°C for 30
min, the thermal resistance of these dehydrated Salmonella cells were enhanced (Kirby
and Davies, 1990). By increasing the length of the dehydration treatment, the initial
count was reduced, but the shape of the curve was the same (triphasic death curve) (Kirby
and Davies, 1990). In addition, populations remained relatively constant when heated at
100°C for 1 h (Kirby and Davies, 1990).

Lethality of Salmonella during roasting of beef has been studied, and research
showed that the death rate depends on both where the bacteria are located and the heating
conditions (Goodfellow and Brown, 1978; Blankenship, 1978). Dry roasting of meat will
kill Salmonella on the interior, but allow for survival on the surface (Blankenship, 1978;
Blankenship, 1980; and Goodfellow and Brown, 1978). Goodfellow and Brown (1978)
found viable Salmonella on the surface of the meat after reaching an internal temperature

of 572°C in a dry environment with the oven at 107°C for 5.5 h. However, no survivors

were present after reaching an internal temperature of 544°C in a wet environment

20

(steam injection) at 794°C for 30 min. In a different study with dry heat, Blankenship
(1978) observed Salmonella survivors in meat that attained an internal temperature of
642°C. Blankenship et al. (1980) hypothesized that a possible explanation was that the
surface and near the surface of the meat probably had a lower water activity (compared to
the center part), due to drying and crust formation during cooking.

Murphy et al. (2001c) studied thermal inactivation of Salmonella and Listeria in
inoculated ground chicken patties (N0~107CFU/ g) under varying conditions in an air
convection oven at an air temperature of 177°C. Thermal processing was conducted at
wet bulb temperatures (humidity conditions) of 48 and 93°C, with the endpoint center
temperature of the patties ranging from 65-75°C (Murphy et al., 2001c). Patties
processed at a wet bulb temperature of 93°C (high humidity) in a wet environment
showed no survivors. The patties processed at a wet bulb temperature of 48°C (low
humidity) in a dry environment contained more than 100 CFU/ g (both Salmonella and
Listeria) at the entire endpoint temperature range (Murphy et al., 20010). Therefore,
bacterial survival was enhanced at a lower humidity. In a high humidity environment, the
authors hypothesized that meat pores opened and the space was occupied with water
vapor, which created a wet environment that enhanced for bacterial inactivation (Murphy
et al., 2001c). In a low humidity environment, pores may have still opened, but the space
would have been occupied with dry air, which would create a dry environment that was
less effective in inactivating bacteria (Murphy et al., 2001c).

In another study, Murphy et al. (2001b) evaluated thermal inactivation of
Salmonella and Listeria in ground chicken patties processed in the same oven as the

previously reported study (Murphy et al., 2001c). The air humidity was controlled by

21

steam injection into the oven (Murphy et al., 2001b). Microbial inactivation decreased
with decreasing wet bulb temperature (39-98°C) (Murphy et al., 2001b). However, this
trend could be caused by the moisture content and water activity of the meat decreasing
during cooking, and not necessarily be a direct effect of wet bulb temperature (i.e.,
process humidity).

Murphy et al. (2001a) also used laboratory-based inactivation models to calculate
process lethality for chicken patties processed in an impingement oven (Murphy et al.,

2000). The air temperature was 149°C, wet bulb temperature ranged from 39 to 98°C,

and patty center temperature ranged from 55 to 80°C (Murphy et al., 2001a). The
cooking conditions affected the time-temperature history of the patties; therefore, the
cooking humidity affected predicted process lethality with a slight decrease in lethality
seen at higher wet bulb temperatures (Murphy et al., 2001a). According to the authors, '
this occurred so that the same final product temperature could be reached; therefore,

cooking time decreased with increasing wet bulb temperature (Murphy et al., 2001a).

22

 

 

 

 

 

 

 

CHAPTER 3

MATERIALS AND METHODS

3.1 Overview

This project was comprised of three different experiments (Table 3.1) involving
isothermal inactivation trials. For simplicity, the different experiments will hereafter be
referred to as Parts 1, 2, and 3. For Part 1, raw, ground, irradiated turkey breast was used.
The moisture content was either increased or slightly decreased, and the samples were
heated in a waterbath. However, after completing this experiment with a small range of
moisture contents, moisture content did not appear to influence the thermal inactivation
of Salmonella. Therefore, Part 2 consisted of a series of tests with a much wider moisture
content range, using cooked ground turkey breast. For Part 3, the same meat was used as
in Part 1, but the samples were heated in an air convection oven, with humidity as the
primary factor, to determine if increasing the moisture in the environment affected the
inactivation. See Appendix B for the details on treatment levels for every test in Parts 1,

2, and 3.

23

TABLE 3.1 Summary of experimental design.

Part 1 (Moisture Effects-High Range)

Temperature (°C) 55, 60, and 65

Moisture Content (°/o) 70.9-76.3 (LF) and 64.5-68.5 (HF)

Fat Content (°/o) 1 and 13

Time (min) 5 durations @ependent on temperature)

 

 

 

 

 

 

 

 

Part 2 (Moisture Effects-Low Range)
Temperature LC) 60

iMoisture Content (°/o) 37.1, 54.4, and 72.5
Time (min) 0, 0.75, 1.5, 2.25, and 3
Fat (°/o) 2 '

 

 

 

 

 

 

 

 

Part 3 (Humidity Effects)

Temperature (°C) 60

Relative Humidity (%) 90 and 96

Fat Content (%) 1 and 13

Timejmin) 0, 0.75, 1.5, 2.25, and 3

 

 

 

 

 

 

 

 

3.2 Part 1 — Moisture effects-high range

The purpose of Part 1 was to test the effect of meat moisture content (over a small
range) on the inactivation of Salmonella. in isothermal heating trials in a waterbath.
3.2.1 Inoculum
3.2.1.1 Bacterial strains

The inoculum consisted of eight Salmonella strains, obtained from Dr. V.K.
Juneja (Agricultural Research Service, Eastern Regional Research Center, USDA-ARS,
Philadelphia, PA). The strains were: S. Thompson FSIS 120 (chicken isolate), S.
Enteriditis H3527 and H3502 (clinical isolates phage type 13A and 4, respectively), S.
Typhimurium (DT104) H3380 (human isolate), S. Hadar MF 0404 (turkey isolate), S.
Copenhagen 8457 (pork isolate), S. Montevideo FSIS 051 (beef isolate), and S.
Heidelberg F5038BG1 (human isolate). Each strain was preserved at —80°C in a vial

containing tryptic soy broth (TSB) (Difco Laboratories, Detroit, MI) with 10% glycerol.

24

3.2.1.2 Culture preparation

To propagate the cultures, one loop of frozen culture was transferred to 9 ml of
TSB in 20 ml culture tubes. The cultures were transferred daily in TSB (37°C, 18-24 h),
with a minimum of two consecutive transfers before subsequent inoculation. Each
inoculum was prepared from an 18-24 h (assumed log phase (Maurer, 2001)) culture.
The eight strains were grown in separate culture tubes, and then equal volumes were
combined prior to centrifugation to produce a cocktail with a target total concentration of
108 CFU/ml. Anew series of cultures from the frozen stock was initiated every week.

On the day of each experiment, cultures were pelleted by centrifugation at 6,000 x
g for 20 min at 4°C and resuspended in sterile 0.1% peptone water. The cultures were
enumerated by plating in duplicate on Petrifi1m® aerobic count plates (3M, St. Paul,
MN) and incubating at 37°C for 24-36 h.
3. 2.2 Meat
3.2.2.1 Ground turkey preparation

Skinless turkey breast meat was obtained from Michigan Turkey Producers, Inc.
(Wyoming, MI) on the day of slaughter and transferred to the Michigan State University
Meat Laboratory at 0°C. The muscle was immediately chopped in a bowl chopper
(Hobart Mfg. Co., Model 841810, Troy, OH) until the temperature reached 13°C. The
turkey fat was chopped separately and then mixed back into half of the previously ground
turkey to create two lots, one with lower and one with higher fat content. Keeping the
two fat lots separate, the turkey was double-bagged in polyethylene-laminated nylon

pouches, vacuum packaged in approximately 100 g portions, and stored at -12°C.

25

The frozen meat was transported overnight on dry ice to Iowa State University
and irradiated to >30 kGy to eliminate indigenous microflora. The frozen meat was
transported back to Michigan State University on dry ice overnight. Samples of
irradiated turkey were tested for sterility to ensure negligible background microflora by
plating a 1:10 dilution in 0.1% peptone on Petrifilm® aerobic count plates.

Proximate analysis was performed in triplicate from three sub-samples taken from
each lot (i.e., low and high fat). Moisture, fat, and protein contents were determined by
AOAC (1996) methods 991.36, 981.1, and 950.46B, respectively. To determine the pH,
10 g of ground‘turkey were added to 90 g of distilled water and homogenized using a
Polytron homogenizer (Model PT 1035, Brinkman Instruments, Westbury, NJ) for 30 s
at speed setting 3. Three samples of both fat levels were prepared, and duplicate
measures were taken of each, using a combination electrode (Model 145, Coming,
Medfield, MA).

Twenty-four hours prior to performing each experiment, meat samples were
thawed in a refrigerator at 4°C.
3.2.2.2 Moisture content alteration

The overall purpose of this experiment was to manipulate the moisture content of
each sample before inoculation and thermal treatment, in the general range that might
occur during thermal processing of a fresh product to a ready-to-eat state. For “native
state” samples, no water was added or removed (other than that associated with the
inoculum). For increased moisture samples, 0.1% sterile peptone was pipetted dropwise
into the meat prior to inoculation. For decreased moisture samples, liquid was removed

by centrifugation at 6,000-9,000 x g for 10-40 min at 4°C. Liquid was poured off the

26

samples, and samples were weighed to determine the amount of liquid. The
centrifugation settings and time increments were varied in order to achieve the target
moisture content. While centrifugation decreased the moisture content, the degree of
reduction was fairly limited, because it was assumed that soluble proteins were also being
extracted with the liquid. After the manipulations, the moisture content and water
activity were determined for each sample using AOAC (1996) method 991.36 and an
electronic water activity meter (accuracy is :0003) (Decagon Devices, Inc., Model 3TE,
Pullman, WA), respectively.
3. 2.3 Inoculation

The inoculum (~1 ml) was added dropwise, using aseptic procedures, to obtain a
target concentration of 108 CFU/ g ground turkey. The inoculum added to the meat had a
minimal effect on the moisture content (<0.2%). The meat was manually mixed (using
sterile gloves) in a sterile bowl for 5 min to ensure even distribution of the inoculum.
Uniform distribution was visually verified using green food dye (McCormick and
Company, Inc., Hunt Valley, MD) in preliminary trials. Actual uniformity was verified
by plating sub-samples of the inoculated meat (Chapter 4).

For each sample to be heated, 1 g of inoculated meat was aseptically placed into a
5 x 25.5 cm polyethylene laminated nylon bag (Butcher and Packer Supply Co., Detroit,
MI). The bags were screened to ensure negligible background microflora by mixing 9 ml
of 0.1% peptone water in 10 random bags and plating on Petrifilm® aerobic count plates.
The bags containing meat were subsequently rolled between two guides, using a large
glass test tube, to a uniform thickness of <1 mm (Figure 3.1). This procedure was used

for two primary reasons; 1) to transfer heat as quickly as possible, thereby minimizing

27

the thermal lag time and 2) to consequently produce more accurate thermal inactivation
parameters (Orta-Ramirez and Smith, 2002). The bags were heat-sealed using a

soldering iron, refrigerated at 4°C, and subjected to thermal treatment within 4 h.

 

FIGURE 3.1 Meat samples rolled between two guides to achieve a uniform thickness.
3.2.4 Thermal inactivation

The sealed bags were placed in a rack and completely submerged in a
temperature-controlled waterbath (NESLAB Instruments, Inc., Newington, NH) set at
55.5, 60.5, or 655°C. The waterbath was set at 05°C above the treatment temperature to
obtain an actual water temperature of 55, 60, or 65°C.

The thermal lag time was defined as the time required for the meat temperature to
reach within 0.5°C of the waterbath temperature. The lag time was determined by
placing a T-type thermocouple in the geometric center of a sample, submerging the
sample in the heated water, and logging the sample temperature with a DuaLogRTM
thermocouple thermometer (Cole Parmer Instrument Company, Model # 01100-50,

Vernon Hills, IL). The test was performed in triplicate, and a lag time of 8 s was

28

determined. The end of the thermal lag was defined as the initial test time for
inactivation (“time zero”).

Samples were removed at five specific time intervals for each test temperature,
placed directly into an ice-water bath, and plated within 4 h. Samples at each moisture
content were heated in three replicate batches at each temperature. Replicate batches
were run on different days.

3.2.5 Enumeration

After treatment, each sample was aseptically transferred to a sterile WhirlpakTM
bag (18 oz, Nasco, Ft. Atkinson, WI) containing 9 ml of 0.1% sterile peptone water, and
manually homogenized for l min. Appropriate dilutions were prepared in 0.1% peptone
water, then 1 ml was pipetted onto Petrifilm® aerobic count plates. All samples were
plated in duplicate and incubated at 37°C for 24-36 h before enumeration. The minimum
detection level was 10 CFU/ g.
3.2.6 Statistics and modeling

Analysis of variance (ANOVA) was run to evaluate ln(N/No) of the cocktail as a
function of the main effects of meat temperature, time, moisture content, and fat content,
and all two-term interactions. N is number of survivors at the end of treatment, and N0 is
the initial inoculum. Linear regressions were run with the raw data to obtain k values
from the slope of equation 3. Then an ANOVA was conducted to evaluate the k values
of the cocktail as a function of the main effects of temperature, moisture content, fat
content, and all two-terrn interactions.

Statistical analyses were performed using J MP (version 4, copyright 2000-2001;

SAS Institute, Inc., Cary, NC.)

29

3.3 Part 2 - Moisture effects-low range

The purpose of Part 2 was to test the effect of meat moisture content (over a large
range) on the inactivation of Salmonella in isothermal heating trials in a waterbath. In
order to decrease the moisture content over a large range (as aseptically as possible), the
meat was cooked/dried in a smokehouse.

3.3.1 Inoculum
3.3.1 . 1 Bacterial strains

The same bacterial strains were used as described in Section 3.2.1.1.
3.3.1.2 Culture preparation

The culture was prepared as described in Section 3.2.1.2.

3. 3.2 Meat
3.3.2.1 Ground turkey preparation and moisture content alteration

Skinless turkey breast meat was obtained from Michigan Turkey Producers, Inc.
(Wyoming, M1) on the day of slaughter and transferred to the Michigan State University
Meat Laboratory at 0°C. The muscle was immediately chopped in a bowl chopper
(Hobart Mfg. Co., Model 841810, Troy, OH) until the temperature reached 13°C.

The turkey was stuffed into either permeable or impermeable casing, using a hand
stuffer (VOGT9, KOCH, Kansas City, MO). The meat to be held at native state moisture
content was stuffed in a non-permeable casing (Faserin #2, Teepak, Kansas City, MO)
measuring 6.5,cm in diameter and 68.58 cm in length. The meat to be dried was stuffed
into a permeable casing (Fiberous Securex #2, Teepak, Kansas City, MO) measuring 4.0

cm in diameter and 76.2 in cm length.

30

Meat was dried/cooked in a smoke house (CGI, Model A28-R0101, Cicero, IL),
to decrease the moisture content and to minimize the microbes in the product. All
samples were heated to an internal temperature of 739°C. Native state samples were
removed when the internal temperature reached 739°C; whereas, samples for the two
decreased moisture content levels remained in the smokehouse (at an internal temperature

of ~5 7°C) until the desired targeted moisture contents were obtained (Figure 3.2).

 

 

 

BumsRcasearch
Batch Oven #2
— _ — - - I -
Dry Bulb Temperature Wet Bulb Temperature Internal Tem-erature
180
\_i
170
. l
160 U i
- l
‘31 i
150 1 iii
140 i ‘9 mg.
I i ii “VA r' JA ka— \ 'I’yvma-flfpn-Rj \ ‘ (”x ' ‘‘‘‘‘ " ‘jfihlml V_-v——\\__fi
130 ' : ‘ “\AV‘ l L. I J \‘i-WW 4 ‘V V! x y '(N‘J‘ \‘H MC ‘Wfi/Vu five J m

 

{.
120 if; I _ V v v

100
90 f f
80

70

Temperature
at

60

12PM 1 Feb 12PM 2 Sat 12PM
Jan 2002 Time

FIGURE 3.2 Smokehouse operation schedule-temperature (°F).
When the turkey was removed from the smokehouse, it was immediately chilled
in a 2-3°C cooler, and placed in a polyethylene-laminated bag, vacuum packaged, and

stored at 2-3°C for approximately 24 h. In an aseptic environment, the turkey was then

31

cut into approximately 20 g sub-samples, double packaged in polyethylene-laminated
bags, vacuum-sealed, and stored at —12°C.

Indigenous microbial levels in the turkey were determined by manually
homogenizing a 1 g sample in 9 ml of 0.1% sterile peptone water for 1 min. Appropriate
dilutions were made in 0.1% peptone water, after which 1 ml was pipetted onto a
Petrifilm® aerobic count plate. All samples were plated in duplicate and incubated at
37°C for 24-36 h before enumeration.

Proximate analysis was performed in triplicate from three sub-samples taken from
each of the three lots. Moisture, fat, and protein contents were determined by AOAC
(1996) methods 991.36, 981 .l, and 950.463, respectively. To determine the pH, 10 g of
ground turkey were added to 90 g of distilled water and homogenized for 30 seconds
using a Polytron homogenizer (Model PT 10/35, Brinkman Instruments, Westbury, NJ) at
speed setting 3. Water activity was determined using an electronic water activity meter
(Decagon Devices, Inc., Model 3TB, Pullman, WA). Each sample was thawed by
placing in the refrigerator at 4°C for 4 h prior to performing the experiment.
3.3.2.2 Decreasing the particle size

Due to the low moisture content of the product, each sample was chopped
aseptically in a high-speed grinder (Tekmar Company, Cincinnati, Ohio) to a particle size
equivalent of powder.

3.3.3 Inoculation

The inoculum (~1 ml) was added dropwise (minimally affecting the moisture

content), using aseptic procedures, to obtain a target concentration of 108 CFU/g ground

turkey. The meat was manually mixed in a sterile bowl for 5 min using a sterile spatula

32

to ensure even distribution of the inoculum. Even distribution was visually verified using
food dye in preliminary trials. Actual uniformity was verified by plating sub-samples of
the inoculated meat (Chapter 4).

For each sample to be heated, 1 g of inoculated meat was aseptically placed into a
5 x 25.5 cm polyethylene laminated nylon bag (Butcher and Packer Supply Co., Detroit,
MI). The bags were screened to ensure negligible background microflora by mixing 9 ml
of 0.1% peptone water in 10 random bags and plating on Petrifilm® aerobic count plates.
The bags containing meat were subsequently rolled between two guides, using a large
glass test tube, to a uniform thickness of <1 mm (Figure 3.1). Again, this procedure was
used for two primary reasons; 1) to transfer heat as quickly as possible, thereby
minimizing the thermal lag time and 2) to consequently produce more accurate thermal
inactivation parameters (Orta-Ramirez and Smith, 2002). The bags were heat-sealed
using a soldering iron, refrigerated at 4°C, and subjected to thermal treatment within 1 h.
3. 3. 4 Thermal inactivation

To prevent water from entering the bags (through possible leaks at the seal), the
bags were sealed at the top, and the tops were held above the water line during treatment.
The samples were placed in a rack that was completely submerged in a temperature-
controlled waterbath (NESLAB Instruments, Inc., Newington, NH) set at 605°C, to
obtain an actual water temperature of 60°C.

The thermal lag time was defined as the time required for the meat temperature to
reach within 0.5°C of the target temperature (60°C). To determine this time, a
thermocouple was placed in the geometric center of a meat sample. The test was

performed in triplicate, and a lag time was determined. The lag times were 15, 30, and

33

60 s, for moisture contents of 73, 55, and 37%, respectively. The end of the thermal lag
was defined as the initial test time for inactivation (“time zero”).

Samples were removed from the waterbath at five specific time intervals and
placed directly into an ice-water bath, with the seals remaining above the ice-water line.
Duplicate batches were run on different days.

3.3.5 Enumeration

Within 1 h of treatment, each sample was aseptically transferred to a sterile
WhirlpakTM bag containing 9 ml of 0.1% sterile peptone water, and manually
homogenized for 1 min. Appropriate dilutions were made in 0.1% peptone water after

which 1 ml was pipetted onto Petrifilm® aerobic count plates. The experiment was

performed in duplicate, and all samples were plated in duplicate and incubated at 37°C
for 24-36 h before enumeration. The minimum detection level was 10 CFU/ g.
3.3.6 Statistics and modeling

Analysis of variance (ANOVA) was run to evaluate ln(N/No) of the cocktail as a
function of the main effects of time, moisture content/water activity, and all two-term
interactions. Linear regressions were run with the raw data to obtain k values from the
slope of equation 3. Then an AN OVA was conducted to evaluate the k values of the
cocktail as a function of the main effects of moisture content/water activity.

Statistical analyses were performed using I MP (Version 4, copyright 2000-2001;
SAS Institute, Inc., Cary, NO).
3.4 Part 3 - Humidity effects

The purpose of Part 3 was to test the effect of humidity on the inactivation of

Salmonella in isothermal heating trials in an air convection oven.

34

3.4.1 Inoculum
3.4.1.1 Bacterial strains

The same bacterial strains were used as described in Section 3.2.1.1.
3.4.1.2 Culture preparation

The culture was prepared as described in Section 3.2.1.2.
3.4.2 Meat
3.4.2.1 Ground turkey preparation

The ground turkey was prepared as described in 3.2.2.1.
3.4.3 Inoculation

The inoculum (~l ml) was added dropwise (minimally affecting the moisture
content), using aseptic procedures, to obtain a target concentration of 108 CFU/ g ground
turkey. The meat was manually mixed in a sterile bowl for 5 min using sterile gloves to
ensure even distribution of the inoculum. Even distribution was visually verified using
food dye in preliminary trials. Actual uniformity was verified by plating sub—samples of
the inoculated meat (Chapter 4).

For each sample to be tested, 1 g of inoculated meat was aseptically spread onto
an ~8 x 8 cm piece of sterile fiberglass screen (New York Co., Mt Wolf, PA) (Figure 3.3)
to a uniform thickness of <1 mm. For sterility testing, tenscreens were placed in a sterile
bag containing 9 ml of 0.1% peptone water with 0.1 ml plated on Petrifilm® aerobic

count plates.

35

 

FIGURE 3.3 Meat sample being spread to a unifome thickness onto a sterile screen.
3.4.4 Thermal inactivation

The samples were heated in a custom air convection oven at 60°C and 90 or 96%
relative humidity (Figure 3.4). The oven was capable of producing dry bulb temperatures
ranging from 25 to 200°C (: 1°C) and relative humidities ranging from 0 to 90% (i 1%)
and 90 to 100% (:2%). The unique heating system consisted of a sample chamber
connected to a mixing chamber, which supply an electronically controlled air/vapor

mixture for a programmed sample exposure.

 

FIGURE 3.4 Meat sample entering the custom air convection oven.

36

The oven contained 6 heat strips (350 watts each). Moisture was added from a
steam generator that injected steam in short bursts until the desired humidity was reached.
A centrifugal fan circulated air inside the heating chamber. The sample was placed in the
heating chamber on a stand, so that air was blown across the top and bottom surfaces of

the sample (Figure 3.5).

fl 12“" "NEW’W‘ “W‘Wfi'fifl‘fl‘ ‘ . l M

 

.‘lr' mum-z». . Hm“-
Ahhitl m '

FIGURE 3.5 Sample in the heating chamber on a stand, which allows air to be blown
across the top and bottom surface of the sample.

In Parts 1 and 2, the thermal lag time was defined as the time required for the
meat sample to reach within 0.5°C of the waterbath temperature. However, in Part 3,
using the oven, the samples did not reach oven temperature, due to the effects of
evaporative cooling, which limited the sample temperature to the oven wet bulb
temperature. Therefore, the oven setting was adjusted (Chapter 4) to ensure that the
samples reached the target temperature. The thermal lag time was defined as the time
required for the meat temperature to reach within 0.5°C of the target temperature (60°C).
To determine the lag time, a thin-wire thermocouple was woven in and out of the screen

in the middle of the meat sample. The test was performed in triplicate, with the thermal

37

lag time determined to be 20 s. The end of the thermal lag was defined as the initial test
time for inactivation (“time zero”).

Samples were removed after five specific test durations, aseptically placed
directly into sterile WhirlpakTM bags containing 9 ml of chilled 0.1% peptone water
(4°C), and plated within 30 min. The samples were heated in two replicate batches at
60°C and 90 or 96% relative humidity. Each replicate batch occurred on a different day.
The initial and final weights of each sample were recorded to determine the amount of
moisture lost during heating (Chapter 4 and Appendix A).

3.4.5 Enumeration

Each treated sample was aseptically placed in a WhirlpakIM bag and manually
homogenized for 1 min with 9 ml of 0.1% sterile peptone water. Appropriate dilutions
were made in O. 1% peptone water with 1 ml pipetted onto Petrifilm® aerobic count
plates. The experiment Was performed in duplicate, and all samples were plated in
duplicate and incubated at 37°C for 24-36 h before enumeration. The minimum detection
level was 10 CFU/g.

3.4.6 Statistics and modeling

Analysis of variance (ANOVA) was run to evaluate 1n(N/No) of the cocktail as a
function of time, relative humidity, final moisture content, and fat, and all two-term
interactions. Linear regressions were run with the raw data to obtain k values from the
slope of equation 3. Then an ANOVA was conducted to evaluate the k values of the
cocktail as a function of the main effects of relative humidity, final moisture content and

fat, and all two-term interactions.

38

Statistical analyses were performed using JMP (version 4, copyright 2000-2001;

SAS Institute, Inc., Cary, NC.)

39

CHAPTER 4

RESULTS AND DISCUSSION

As described in Chapter 3, this project was comprised of three parts. Section 4.1
will give some background information common to all three parts, related to the
inoculum, proximate composition, initial counts, inoculum distribution, lag time, and
then some background information pertinent for each specific part. Subsequent sections
(4.2.1 to 4.2.4) focus on the inactivation results specific to each of the three respective
parts, including graphs of the results, analyses of variance (both raw data and k values),
and inactivation modeling.

While much of the data obtained in this study appeared to be non-linear (e. g.,
Figure 4.9), the amount of data generated was insufficient to fit non-linear models. The
results and conclusions of this work would most likely be unaffected by this; however,

the precision and accuracy of the models would likely be affected.

4.1 General background information
4.1.1 Salmonella cocktail

The inoculum culture for each strain was plated in duplicate (Table 4.1). The
overall average, before mixing the cocktail, of all the strains was 1.40 x 109 CFU/ml in

the inoculum.

40

TABLE 4.1 Salmonella counts.

 

 

 

 

 

 

 

 

 

 

CFU/ml
Strain ' '
Indrvrdual AVG
reps

S. Thompson 1 .13E+09 1 .17E+09
FSIS 120 1,205+09

S. Enteritidis 1.55E+09 1.64E+09
H3527 1.73E+09

S. Enteritidis 3.95E+09 3.93E+09
H3502 3.91 E+09

S. Typhimurium 1.22E+09 1.24E+09
H3380 1.25E+09

S. Hadar 8.50E+08 8.50E+08
MF60404 8.50E+08

S. Copenhagen \4.30E+08 5.35E+08
8457 6.40E+08

S. Montevideo 6.70E+08 8.05E+08
FSIS 051 9,405+03

S. Heidelberg 8.40E+08 1.04E+09
F50388G1 1235-1-09

 

 

4.1.2 Proximate composition

41

The proximate composition of meat used (Part 1-3) is listed in Table 4.2.

Mm- -. -

 

 

 

 

5.0 3 Rd 0.8 a: in SS 3. v.25
8.0 me New o. 3. mm; in 8d 3 N to
mod 5 mad «.8. 8.0 QR 5o 3 22m «232
So we owe 02 Rd 08 «so of a”. :9:
n as... F

:3 no So QR m3 0% mod 2 an. 26..
om u>< om m>< am u>< am m><

tad on». “no.2

:a E 532“. 3.. 23%: 3.. an.

 

 

$815 hexeamxw zetwmomfieu Sefieaam N6. Hum—(H.

42

4.1.3 Initial counts
4.1.3.1 Part l—Moisture effects-high range

Samples of irradiated turkey were tested to ensure negligible background
microflora by plating a 1:10 dilution in 0.1% peptone water on Petrifilm® aerobic count
plates. All plate counts showed no growth.
4.1.3.2 Part 2——Moisture effects-low range

Samples of cooked turkey were tested for initial counts (immediately before
inoculation) by plating a 1:10 dilution in 0.1% peptone water on Petrifihn® aerobic count
plates. The average initial count was ~547.9 (SD=:1248.27) CF U/g (Appendix E).
Compared to the amount of inoculum added into the ground turkey, the initial count was
very small. Because non-selective Petrifilm® plates were used, it was not confirmed
what specific bacteria were actually in the ground turkey prior to inoculation. However,
only Salmonella-like colonies were observed and counted.
4.1.3.3 Part 3-—Humidity effects

The same meat was used as in Part 1 (See 4.1.2.1).
4. 1. 4 Inoculum distribution

Uniformity of inoculation was verified by plating unheated inoculated meat
samples diluted in 0.1% peptone water on Petrifihn® aerobic count plates. Sub-samples
were plated for all three parts to determine the uniformity of mixing. The means (:SD)
of the unheated inoculated meat samples in Part 1, 2, and 3 were 7.9 (i027), 6.9 (10.51),

and 7.8 (1:01 1) log(CFU/g), respectively.

43

The targeted total concentration was 108 CF U/ml. Raw ground turkey in Parts 1
and 3 was close to the targeted concentration; however, the cooked meat used in Part 2,
was approximately 1 log lower than the targeted concentration, and also more variable.
4.1.5 Thermal lag times

The thermal lag time was the time required for the meat temperature to reach
within 0.5°C of the target temperature. The end of the thermal lag was the initial test
time for inactivation (“time zero”). The thermal lag time for Part 1 and 3 were 8 and 20
s, respectively, whereas for Part 2, the thermal lag times for 73, 55, and 37% moisture
content were 15, 30, and 60 s, respectively. A
4.1.6 Additional test information for Part 1
4.1.6.1 Changes during the experiment

In Part 1 and 3, raw ground turkey was used. From the time the bag of meat was
opened until the experiment was completed, some moisture was lost from the meat during
processing. In order to evaluate this loss of moisture, 100 g of low fat ground turkey
(native state) sat at room temperature (22°C) for 2.5 h. During that time, 2.47 g of
moisture was lost. However, during actual inactivation trials, the meat was refrigerated
in a covered vessel, whenever it was not being used, in order to minimize moisture loss.

During the experiment, on any given day, meat inoculum levels fluctuated ~0.5
log(CFU/ g) from the time the experiment started until completion. However, all of the

inactivation analyses were based on actual “time zero” counts for each specific test.

44

4.1.6.2 Moisture content alteration

The moisture content ranged from 70.96 to 76.34% in the low fat ground turkey;
the water activity ranged from 0.993 to 0.997. The moisture content ranged from 64.49
to 68.49% in the high fat ground turkey; the water activity ranged from 0.991 to 0.994.
4.1.7 Additional test information for Part 2
4.1.7.1 Moisture content alteration

The moisture contents were 37.13, 54.37, and 72.51% in cooked ground turkey,
with water activities of 0.95, 0.98, and 0.99, respectively.
4.1.8 Additional test information for Part 3
4.1.8.1 Moisture lost during heating

Moisture lost during oven heating was determined by comparing the mass of the
sample before and after heating (Appendix A). During heating, the samples lost an
average of 0.0313 g (:0.10) of moisture per ~1 g of sample. For low fat ground turkey,
the average final moisture content was 71.5% (_+_0.01). For high fat ground turkey, the
average final moisture content was 62.4% (:008).
4.1.8.2 Oven adjustments

During convection heating, the wet bulb temperature of the air limited the
temperature of the meat. Because the goal was to keep the sample temperature the
constant for all treatments, the dry bulb temperature of the oven was adjusted to achieve
equal wet bulb (and therefore sample) temperatures for each treatment (Table 4.3). These
sample temperatures were verified by using a thin-wire thermocouple in preliminary tests

(as described for determining the thermal lag time, Chapter 3).

45

TABLE 4.3 Oven settings for inactivation trials.

 

 

 

Tdb (°C) RH (0/0) Tsample (°C)
62.3 90 , 60
60.6 96 60

 

 

 

 

 

4.2 Inactivation results

The inactivation results are presented as: graphs of the average raw results, as
analyses of variance (both raw data and k values), and as a secondary inactivation model.
Sections are also included to compare results from Parts 1 (raw) and 2 (cooked) at the
native state moisture content and to compare results from Part 1 (waterbath) to Part 3
(oven)
4. 2.1 Part 1 —— Moisture effects-high range
4.2.1.1 Data

Figures 4.1 to 4.6 depict the mean survivor data from all inactivation tests in
Part 1 (Appendix B). The lines are linear regressions (with slope=k). The goodness of fit
(R2) ranged from 0.05 to 0.99, with an average of 0.74 (:0.250). In Figures 4.2 and 4.5,
the data fit the line well. In Figures 4.1, 4.4, and 4.6, the data vary and even show an
increase in the number of survivors. In Figures 4.3, 4.4, and 4.6, the data exhibit
“tailing.”
4.2.1.1.1 Low fat:

Figures 4.1 to 4.3 depict the mean survivor data for low fat ground turkey heated

at 55, 60, and 65°C.

46

 

 

O 72.3%

D 74.3%

I 76.3%
- - -Linear(72.3 %)
— —Linear(74.3%)
—Linear (76.3%)

log (N/No)

 

 

 

 

 

 

 

0 20 40 60 80

Time (min)

FIGURE 4.1 Thermal inactivation of Salmonella in low fat ground turkey at 55 °C and
three different moisture contents ( 72.3- 7 6.3%).

 

OI—

 

. 72.3%
3° :1 74.3%
2
I; o 76.3%
.2
Linear (72.3%)

 

— —Linear (74.3%)

 

 

 

 

 

- - - Linear (76.3%)

 

 

0 0.5 1 1.5 2 2.5 3 3.5

Time (min)

FIGURE 4.2 Thermal inactivation of Salmonella in low fat ground turkey at 60 °C and
three different moisture contents (72.3-76.3%).

47

 

I
—L

 

 

 

 

 

2 O 72.3%
3 E1 74.3%
E O 76.3%
2 -3
E - - - Linear (72.3%)
"' .4 Linear (74.3%)
—- -Linear(76.3%)

 

I
U!

 

 

 

ch

T f r

0.05 0.1 0.15 0.2 0.25 0.3 0.35
Time (min)

0

FIGURE 4.3 Thermal inactivation of Salmonella in low fat ground turkey at 65 °C and
three different moisture contents ( 72.3-7 6.3 %).

4.2.1.1.2 High fat

Figures 4.4 to 4.6 represent mean survivor data for high fat ground turkey heated

at 55, 60, and 65°C.

 

 

 

O 64.5%
3 D 66.5%
g o 68.5%
E Linear (64.5%)

- - - Linear(66.5%)
— —Linear (68.5%)

 

 

 

 

 

 

 

0 20 40 60 80 100
Time (min)

FIGURE 4.4 Thermal inactivation of Salmonella in high fat ground turkey at 55 °C and
three different moisture contents (64.5 -68.5 %).

48

 

 

O 64.5%

D 66.5%

I 68.5%
— -Linear (64.5%)
Linear (66.5%)
- - - Linear (68.5%)

 

log NINo

 

 

 

 

 

 

 

0 0.5 1 1.5 2 2.5 3 3.5

Time (min)

FIGURE 4.5 Thermal inactivation of Salmonella in high fat ground turkey at 60 °C and
three different moisture contents (64.5-68.5%).

 

 

 

 

 

o 64.5%
3- n 66.5%
g o 68.5%
3 - - - Linear (64.5%)
9 Linear (66.5%)
— —Linear(68.5%)

 

 

 

 

 

-14

f

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Time (min)
FIGURE 4.6 Thermal inactivation of Salmonella in high fat ground turkey at 65 °C and
three different moisture contents (64.5 -68.5 %).

Variability occurred in the high fat meat samples at 55 and 65°C and the low fat
meat samples at 55°C. This variability could be caused because the fat may prevent the

inoculum from mixing well into the meat, as compared to the low fat meat. Fat “pockets

present in the high fat meat could have protective properties, as well. This could also

49

explain why an increase in CFU/ g was observed over time (Figure 4.4 at 66.5% MC and
Figure 4.6 at 64.5% MC). However, this was not always true in the high fat meat heated
at 60°C, since variability was small. Furthermore, this did not explain the variability in
the low fat meat (Figure 4.1), and the increased CFU/ g in low fat meat (Figure 4.1 at
74.3% MC).

While performing the trials in Part 2, it was noticed that moisture was leaking into
the pouches during the waterbath heating treatment. For Part 2, the problem was
corrected; however, it remains uncertain whether moisture was leaking into the pouches
in Part 1. If so, it was not noticed because of the moist state of the meat (compared to the
dry meat in Part 3), but the possibility cannot be eliminated. If water was actually
leaking into some bags, this could account for the high variability.
4.2.1.2 ANOVA
4.2.1.2.1 Raw data

For Part 1, the raw data were analyzed via analyses of variance (ANOVA) with
both fat levels together (Table 4.4, column a), and then at the low (Table 4.4, column b)
and high fat (Table 4.4, column c) levels independently. With all samples included, the
AN OVA included time, temperature, moisture content, and fat content, and all two-term
interactions (Table 4.4, column a). Time, temperature, moisture content, fat, and the time
* temperature interaction were significantly related to Salmonella survival (0t=0.05).

With only low fat samples included, the AN OVA included time, temperature, and
moisture content, and all two-term interactions (Table 4.4, column b). Time and
temperature, and. the time * temperature interaction were significantly related to

Salmonella survival (or=0.0001); however, moisture content did not affect inactivation.

50

With only high fat samples included, the AN OVA included time, temperature, and
moisture content, and all two-term interactions (Table 4.4, column c). Time, temperature,
moisture content, and the time "' temperature interaction was significantly related to
Salmonella survival (if (1:0.10).

Moisture content ranged from 64.5 to 76.3% (~12% range), and the maximum
moisture content of the high fat meat never exceeded the minimum moisture content of
the low fat meat (MCHF-max<MCu:-mm). Therefore, fat, rather than moisture content,
probably was the controlling factor for the observed difference in the ANOVA that
included both fat levels, given that fat was a significant factor in the lumped data set
(Table 4.4, column a). As fat increased, the lethality rate decreased, with others reporting
similar trends (Ahmed et al., 1995; Line et al., 1991; Fain et al., 1991; and Ben-Embarek
and Huss, 1993) (Chapter 2). However, for the high fat meat, moisture content was
significantly related (if or=0.10) to k values (Table 4.5, column (I); therefore, fat may not

have been the lone factor affecting inactivation.

51

TABLE 4.4 P values from analyses of variance of the raw data in high fat and low fat
ground turkey at varying moistures, fats, times, and temperatures.

 

 

 

 

 

 

 

 

 

 

 

TEST (a) (b) (c)
FACTORS HF and LF Only LF Only HF

Time <0.0001 <0.0001 <0.0001
Temperature <0.0001 <0.0001 <0.0001
MC 0.0177 0.1398 0.0772

Fat 0.0038 n/a* n/a
Time*Temp <0.0001 <0.0001 <0.0001
Time*MC 0.9520 0.9585 0.7847

Time*Fat 0.8592 n/a n/a
Temp*MC 0.7762 0.3289 0.1057

Temp*Fat 0.4424 n/a n/a

Fat*MC 0.6481 n/a n/a

 

 

 

*n/a=not included in the particular analysis represented by that column.

For Part 1, water activity was not a parameter in the analysis, because water
activity ranged from only 0.993 to 0.997. In these tests, moisture content was increased
by pipetting sterile 0.1% peptone water into the sample or decreased by centrifuging
liquid out of the sample. Controlling added moisture was uncomplicated and consistent;
however, removing moisture was more difficult. The entire water activity range for Part
1 was 0.004, including the increased and decreased moisture content range of 70.9 to
76.3%. Also, a single water activity reading was not possible for the decreased moisture
samples, because every sample was different, and the range of water activities was very

small due to limited ability to remove water via centrifugation.

52

4.2.1.2.2 K values

The k values, rates of thermal inactivation (assuming first order kinetics), were
calculated by linear regression of the ln(N/N0) data over time for each replication
(Appendix B).

For Part 1, the k values were analyzed via analyses of variance (ANOVA) with
both fat levels together (Table 4.5, column a), and then with low (Table 4.5, column b)
and high fat (Table 4.4, column c) levels independently. With all data incorporated into
the model, the first ANOVA included temperature, moisture content, and fat, and all two-
terrn interactions (Table 4.5, column a). Temperature and the moisture content * fat
interaction significantly affected the k values (ct=0.05). The second ANOVA (Table 4.5,
column b) included only low fat samples and evaluated temperature, moisture content,
and the interaction of temperature * moisture content into the model, and both
temperature and moisture content were significantly related to k (if 0t=0.10). The third
AN OVA included only high fat samples and evaluated temperature, moisture content,
and the interaction of temperature * moisture content into the model, and only
temperature significantly affected the k values (if or=0. 10).

Therefore, these results are consistent with the AN OVA of the raw data. It
appears, given a very narrow range of moisture content (and therefore water activity),
that moisture content might affect thermal inactivation of Salmonella in turkey; however,
the results were not conclusive. Therefore, the next test series (Part 2) widened the range

of the moisture content/water activity in the meat.

53

TABLE 4.5 P values from analyses of variance of the k values in high and low fat
ground turkey at varying moistures, fats, times, and temperatures.

 

(a) (b) (C)
TEST
FACTORS Hi?!“ Only LF Only HF

 

Temperature <0.0001 <0.0001 <0.0001

 

 

 

 

 

MC 0.4985 0.0901 0.1310
Fat 0.5349 n/a n/a
Temp * MC 0.7305 0.6377 0.1278
Temp * Fat 0.8119 n/a n/a
MC * Fat 0.0395 n/a n/a

 

 

 

*n/a=not included in the particular analysis represented by that column.
4. 2.2 Part 2 - Moisture effects-low range
4.2.2.1 Data
Figure 4.7 depicts the mean survivor data from the entire inactivation trial in
Part 2 (Appendix B). The lines are linear regressions (with slope=k). The goodness of fit
(R2) ranged from 0.95 to 0.97 with an average of 0.97 (330.011). The R2 range in Part 2

was much smaller than in Part 1.

54

 

 

 

 

 

 

 

 

 

 

o 37.1%
’5 D 54.40/0
g A 72.5%
E Linear (37.1%)
- ~ ‘ \U — Linear (54.4%)
-4. ‘ e - ' Linear (72.5%)
‘ A
\
-5.
-5 , , , . . .
o 0.5 1 1.5 2 2.5 3 3.5

Time (min)

FIGURE 4.7 Thermal inactivation of Salmonella in low fat ground turkey at 60 °C and
three different moisture contents (3 7. 1-72.5%).

From the graph, the data appear to be fairly linear; therefore, it appears that any
background micro flora (Section 4.1.3.2) did not affect these results; otherwise, a tailing
phenomenon would be expected.
4.2.2.2 ANOVA
4.2.2.2.1 Raw data

For Part 2, one AN OVA was run with time and moisture content as the test
variables (Table 4.6, column a), and a second ANOVA was run with time and water
activity (Table 4.6, column b) as the test variables. Time, moisture content/water
activity, and their interactions all affected inactivation of Salmonella (0t=0.0001) (Table
4.6).

In contrast to Part 1, water activity in Part 2 was a parameter in the analysis,
because the experimental design contained three well-controlled water activities (0.950,

0.981, and 0.996). Therefore, an accurate water activity reading was possible for the

55

decreased moisture samples. The range of moisture content in Part 2 (35.4%) was much
larger than the range of moisture content in Part 1 (5.38%).

TABLE 4.6 P values from analyses of variance of the raw data in low fat ground turkey
at varying moistures, water activities, and times, at a sample temperature of 60 °C.

 

 

 

 

 

 

TEST (a) (b)
FACTORS LF—60°C LF-60°C
Time <0.0001 <0.0001
MC <0.0001 n/a
Time*MC <0.0001 n/a
aw n/a* <0.0001
Time “'aw n/a <0.0001

 

 

 

*n/a=not included in the particular analysis represented by that column.
4.2.2.2.2 K values

The k values, rates of thermal inactivation (assuming first order kinetics), were
calculated by linear regression of the 1n(N/No) data over time for each replication
(Appendix B).

For Part 2, the k values were analyzed by analysis of variance (AN OVA) with
moisture content (Table 4.7, column a) and water activity (Table 4.7, column b). The rate
of inactivation was significantly increased at both high moisture content and high water

activity (ct=0.01) (Figure 4.7 and Table 4.7).

TABLE 4.7 P values from analyses of variance of the k values in low fat ground turkey
at varying moistures, water activities, and times, at a sample temperature of 60 °C.

 

TEST 7 (a) (b)

 

 

FACTORS LF LF
MC 0.0022 n/a
aw n/a* 0.0019

 

 

 

*n/a=not included in the particular analysis represented by that column.

56

Three previous studies have reported an effect of water activity on thermal
inactivation of Salmonella or E. coli (Goepfert et al., 1970; Reichart, 1994; and
O’Donovan-Vaughan and Upton, 1999). Goepfert et al. (197 0) evaluated the relationship
between the heat resistance of Salmonella and water activity (0.75-0.99) in different
sugar solutions. Reichart (1994) and Cerf et al. (1996) attempted to model the
destruction of E. coli in a glucose solution as a function of pH (3-9) and water activity
(0928-0995). Lastly, O’Donovan-Vaughan and Upton (1999) assessed the survival of
Salmonella Typhimurium at three extreme water activities (0.45-0.90) and two
temperatures, in sucrose solution. All three studies were performed in carbohydrate
solutions (not food systems); and only one study attempted to model the data (while the
other two just reported data). All three studies showed that water activity caused an
effect, but other parameters (e.g., solute, pH) also affected the outcome.

First, Goepfert et al. (1970), used seven different Salmonella strains (S. Infantis,
S. Alachua, S. Typhimurium, S. Anatum, S. Anatum GF, S. Montevideo, S. Senftenberg
775W, and S. Tennessee), and evaluated the effects of water activity in different
carbohydrate solutions. Sucrose clearly provided a protective effect. Water activity
ranged from 0.87 to 0.99 in sucrose (pH 6.9) and 0.75 to 0.99 in glycerol (pH 6.9). The
media was heated in a flask to 572°C. In the study with glycerol, the same Salmonella
strains were used, except for S. Infantis. For the purpose of comparing the data from
Goepfert et al. (1970) with the current study, S. Senftenberg was removed from the
Geopfert et al. (1970) data prior to computing averages, because of the strain’s extreme

heat resistance (Goepfert et al., 1970) (Table 4.8 and Figure 4.8).

57

The current work was also compared to earlier data of Reichart (1994). E. coli B
200 was inoculated into a glycerol-water mixture to obtain three water activities (0.995-
0928) and heated at 58°C. The pH of the heating menstrum in that study ranged from 3-
9; however, only results obtained at pH 6 and 7 are being cited, because these values
were closest to the pH in the current work (pH 6.2) (Table 4.8 and Figure 4.8).

Lastly, O’Donovan—Vaughan and Upton (1999) tested the survival of S.
Typhimurium at water activity range of 0.45 to 0.9, at 55 and 65°C, and in different
solutes. Again, consistent with Goepfert et al. (1970), sucrose was most protective, with
heat resistance increasing as the water activity decreased. The overall conclusion was
that heat resistance depended on the nature of the solute used to lower the water activity.
In the following table (Table 4.8) and graph (Figure 4.8), the results of O’Donovan-
Vaughan and Upton (1999) are not included, because the water activity range was too
large (0.90-0.45), and minimum water activity was too small (0.45) to compare to the
current study.

The direct comparison between previously reported data and the current work is
limited, because heating occurred at different temperatures. When equal water activity
changes were compared (~0.95 to 0.99), k value reductions of 64, 42, 64, and 92% were
observed by the following: the current study, Reichart (1994) (pH 6), and Reichart
(1994) (pH 7), and Goepfert et al. (1970), respectively. Overall, heat resistance increased
as the water activity decreased, which was in agreement with this study.

To compare the current work with the previous work at similar temperatures, the

k values (at 60°C) needed to be transformed to equivalent values at 576°C. This was

58

accomplished by linear regression of 1n(k) vs. l/T data from Part 1 to get the Arrhenius
parameters (per equation 6); that regression (R2=0.9485) yielded:
ln(k)=169.64-56185(1/Tab5) (12)

From this model, a ratio was determined at two temperatures (57.6 and 60°C) and one
water activity, as follows:

kbooC/k57.6°c=e[(-56185/333) + (56185/330.6)] = 3.4 (13)
The actual k values for Part 3 at 60°C were divided by this ratio to obtain transformed k
values at 576°C, in order to make a more direct comparison with the published data

(Table 4.8 and Figure 4.8).

59

TABLE 4.8 k values (min!) as a function of water activity.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Test . . Temp Avg k value for 8 strain Salmonella
solute/product Citation a... (°C) pH cocktail (min'l)
Current 0.996 60 6.2 3.569
ground turkey work 0.981 60 6.2 2.648
(2002) 0.95 60 6.2 1.301
*Current 0.996 57.6 6.2 1.05
ground turkey work 0.981 57.6 6.2 0.779
(2002) 0.95 57.6 6.2 0.383
Avg k value for 6 Salmonella strains
(min'1)
sucrose ‘ 0.99 57.2 6.9 2.302
Gfafgfof 0.96 57.2 6.9 0.184
° 0.93 57.2 6.9 0.071
I cerol 0.99 57.2 6.9 2.224
9 y 0.9 57.2 6.9 0.972
Avg R value E. coli (mind)
0.995 58 6 0.762
Rei ch a rt 0.956 58 6 0.444
glycerol (1994) 0.928 58 6 0.288
0.995 58 7 1.314
0.956 58 7 0.468
0.928 58 7 0.306
*Transformed data
4
I Current work (2002)-
3.5 - ' Salmonella in turkey at
3 60C
- - Current work (2002)-
", 2,5 — Salmonella in turkey at
g D 57.6
g 2 A ElGoepfert et al. (1970)-
.: 1.5 , Salmonella in sucrose
I A at 57.20
1 “ _ ;' OReichart (1994)-E. coli
0 5 q ‘ in glycerol-pH 6 at 58C
. -
Cl
0 51 . . . A Reichart (1994)-E. coli
0.92 0.94 0.96 0.98 1 in glycerol-9H 7 at 580

 

 

 

water activity

FIGURE 4.8 Comparison of k values as a function of water activity.

60

Obviously, among these cited studies, tests were performed at different
temperatures, pH values, and water activities. The current study and that of Goepfert et
al. (1970) used Salmonella; however, the current study used a Salmonella cocktail, and
Goepfert et al. (1970) ran tests with individual Salmonella strains. Reichart’s (1994)
work was performed with E. coli, in glycerol, at different pH’s. The current study was
the only one that conducted tests in a food system and not a carbohydrate solution. In the
meat product, the solutes in the aqueous solution are probably dominated by electrolytes
(i.e., non-carbohydrate components).

However, in all these studies, the heat resistance of the organism increased as the
water activity decreased. The present data (in meat) were generally consistent with the
previous data in the carbohydrate system. However, meaningful differences between the
current data and those of Geopfert suggest that inactivation models should not be derived
from only media inactivation studies. In addition, specific substrate and test conditions
must also be considered when determining thermal inactivation models.
4.2.2.3 Modeling

Because water activity was significantly related to the survival of Salmonella
(Table 4.6 and 4.7), the data were applied to a secondary modified (additive) Arrhenius-
type model. Cerf et al. (1996) applied this model to previously published data, and
accounted for other parameters (Equation 7). The model used in the current work was as
follows:

1n(k)=ln(k0)-(Ea/R)(1/T)+”water term” (14)
where k=the inactivation rate, Ear-activation energy, =universal gas constant, and

T=absolute temperature. The three “water terms” added to the equation were as follows:

61

l/aw, aw, and aw+awz. Part 2 was only run at one temperature (60°C). Consequently, a
model parameter for the temperature term (Ea) could not be estimated from regression of
the present data, and a constant was arbitrarily applied for the activation energy (Ea)
(Table 4.9). All three “water terms” resulted in a root meat square error of the model of
less than 1 logo (CFU/ g) (Appendix D).

TABLE 4.9 Effect of the ”water term ” form on the root mean square error for a first-
order, modified Arrhenius-type model.

 

 

 

 

Error
“water term”
logm (CFU/g)
l/aw 0.747
aw 0.749
aw+ 21..2 0.750

 

 

 

4.2.3 Part 3 — Humidity effects

4.2.3.1 Data

Figure 4.9 depicts the mean survivor data collected from the entire inactivation
trial in Part 3 (Appendix B). The lines are linear regressions (with slope=k). The
goodness of fit (R2) ranged from 0.77 to 0.98, with an average of 0.89 (i0.081). The R2
range in Part 3 was greater than Part 2, but less than Part 1. In Figure 4.11, the data

appears to tail slightly.

62

 

 

 

 

 

 

 

 

 

 

e 96%-LP
n 90%-LF
3 -2. e 96%-HF
2 o -
2 -3 ‘ A 90/0 HF
45 Linear (96%-LP)
.2 44 — —Linear(90%-LF)
-5 « —Linear (96%-HF)
’6‘ \ . - - - Linear (90%-HF)
-7 . . , L , g
0 0.5 1 1.5 2 2.5 3 3.5

Time (min)

FIGURE 4.9 Thermal inactivation of Salmonella in low and high fat ground turkey at
60 °C and 90 and 96% relative humidity.

The hypothesis (Chapter 1) was that the rate of thermal inactivation of Salmonella
decreases with decreasing meat moisture and/or process humidity. However, it can be
seen (Figure 4.9) that, regarding humidity, the opposite was generally true. Therefore,
the part of the hypothesis that referred to humidity was rejected (based on statistical
analyses in the next section), because as the relative humidity increased, the thermal
inactivation was not increased, but rather, actually decreased (Figure 4.9). Consequently,
in the following analyses of variance of the raw data and k values, when the relative
humidity was significantly related to Salmonella survival, it was not significant in the
direction hypothesized.
4.2.3.2 ANOVA
4.2.3.2.1 Raw data

For Part 3, the raw data were analyzed by analyses of variance (ANOVA) with
both fat levels together (Table 4.10, column a), and then at the low (Table 4.10, column
b) and high fat (Table 4.10, column 0) levels independently. With all samples included,

the ANOVA included time, relative humidity, final moisture content, and fat, and all two-

63

term interactions (Table 4.10,. column a). Time, relative humidity, and the two-term
interaction between time * relative humidity affected Salmonella survival (if 01:01).

With the low fat samples only, the AN OVA included time, relative humidity, and
final moisture content, and all two-term interactions (Table 4.10, column b). Time and
relative humidity, and all two-term interactions significantly affected Salmonella survival
(or=0.05). With only high fat samples included, the ANOVA was run with time, relative
humidity, and final moisture content, and all two-term interactions (Table 4.10, column
c). Time, relative humidity, and the interaction between the two significantly affected
Salmonella survival (or=0.05).

TABLE 4.10 P values from analyses of variance of the raw data in low and high fat
ground turkey at varying relative humidities, at 60 °C.

 

 

 

 

 

 

 

 

 

 

 

TEST (a) (b) (c)
FACTORS HF and LF Only LF Only HF
Time <0.0001 <0.0001 <0.0001
RH <0.0001 0.0181 <0.0001
Final MC 0.6459 0.4740 0.2531
Fat 0.1710 n/a* n/a
Time * RH 0.0982 0.0462 0.0265
Time *

Final MC 0.3238 0.0282 0.8280
Time * Fat 0.3119 n/a n/a
RH * Final

MC 0.2423 0.0285 0.2450
RH * Fat 0.2234 n/a n/a
Final MC

* Fat 0.3488 n/a n/a

 

 

 

*n/a=not included in the particular analysis represented by that column.

64

A direct explanation cannot be given as to why the rate of thermal inactivation
increased with increasing relative humidity. However, the moisture lost was a controlled
parameter, and only a minimal amount of water was lost during heating (Appendix A).
In addition, thermal inactivation did not decrease with decreased humidity (as was
hypothesized). Therefore, it appears that meat water activity (Part 2), rather than process
humidity (Part 3), is the controlling water parameter affecting Salmonella resistance to
thermal inactivation in a meat product.

The experimental design (Table 4.3) of the humidity test could possibly explain
the significance in the results opposite of the hypothesis. The dry bulb temperature was
increased at the lower relative humidity (90%) to provide a consistent sample temperature
(60°C). In preliminary tests, the temperature set points and sample temperatures were
determined by using a thermocouple inserted in the samples. However, the thin-wire
thermocouple was nearly as thick as the meat sample. Hence it is possible that dry bulb
temperature set points did not achieve a sample temperature of exactly 60°C, and even
the slightest change in temperature could alter the inactivation rate. In order to maintain
an aseptic environment, using a thermocouple for every sample was not possible.

As discussed earlier in Chapter 4, a small amount of moisture was lost during
heating. When the humidity was increased from 90 to 96%, the average moisture lost
during the heating process decreased from 0.041 to 0.025 g. The slightly decreased
moisture content corresponded to an increased rate of thermal inactivation, which is

inconsistent with results from Part 2.

65

4.2.3.2.2 K values

The k values, rates of thermal inactivation (assuming first order kinetics), were
calculated by linear regression of the 1n(N/N0) data over time for each replication
(Appendix B).

The k values were analyzed by analyses of variance (ANOVA) with both fat
levels included (Table 4.1 1, column a), and then with low (Table 4.11, column b) and
high fat (Table 4.1.1, column 0) leVels independently. The first ANOVA included data
from both fat levels, and relative humidity and fat, and the two-term interaction, and none
of these factors were related to k (Table 4.11, column a). When low fat meat was
analyzed independently (Table 4.11, column b), the AN OVA included only relative
humidity, and it was not related to k. However, when the model was run with only high
fat meat, the ANOVA included relative humidity, which was related to k (or=0.05)
(Table 4.11, column c). However, k increased with decreasing humidity, which was
opposite of the hypothesis.

TABLE 4.11 P values from analyses of variance of k values in low and high fat ground
turkey at varying relative humidities, at 60 °C.

 

 

 

 

TEST (a) (b) (c)
FACTORS HF and LF Only LF Only HF
RH 0.1708 0.8580 0.0147
Fat 0.4044 n/a n/a
RH * MC 0.2735 n/a n/a

 

 

 

 

 

 

*n/a=not included in the particular analysis represented by that column.
Murphy et al. (2001b, c) evaluated thermal inactivation of Salmonella and
Listeria in ground chicken patties that were processed in an air-impingement oven. In

both studies, microbial inactivation decreased with decreasing wet bulb temperatures. A

66

reasonable explanation was given (Chapter 2); however, one very important parameter
was neglected. In all of these studies that assessed the effects of humidity, humidity was
not isolated as the sole factor. During cooking, the moisture content and water activity of
meat decreases; therefore, it was possible this was actually causing the effect, instead of
humidity.

The current study isolated the intrinsic and extrinsic water variables to determine
which parameters were actually controlling the effect. During the humidity test, the
moisture content changed minimally (0.03 g), and the samples were heated at an
isothermal dry bulb temperature of 60°C and two different humidities (90 and 96%). The
rate of thermal inactivation did not decrease with decreasing process humidity.

4. 2.4 Parts I and 2 combined

Parts 1 and 2 were combined to determine if the methodology (raw vs. cooked
meat) affected the inactivation of Salmonella.
4.2.4.1 Data

The low fat ground turkey data, heated to 60°C, was combined from Parts 1 and
2. The one main difference between Parts 1 and 2 was that Part 1 utilized raw, irradiated
ground turkey, whereas cooked, non-irradiated ground turkey was used in Part 2 (Figure

4.10).

67

 

 

0 Part 1 Turkey
0 Part 2 Turkey
Linear (Part 1 Turkey)
- - - Linear (Part2 Turkey)

 

log (NINo)

 

 

 

 

 

 

 

Time (min)

FIGURE 4.10 Thermal inactivation of Salmonella in low fat ground turkey at 60 °C in
raw vs. cooked meat.

4.2.4.2 ANOVA

Ideally, a difference between the meat used in Part 1 and the meat used in Part 2,
would be indistinguishable; however, in the current study, that was not the case.
4.2.4.2.1 Raw data

A slight, but significant difference existed between the raw turkey vs. cooked
turkey (only if 01:01), in terms of Salmonella inactivation (Table 4.12). Most likely a
difference was observed because the cooked meat contained a small amount of
background microflora, and the raw meat was irradiated. However, that was not the only
possible explanation. When the meat was dried in the smokehouse, the chemical
properties of the meat may have been altered in such a way that a protective mechanism
was afforded to the bacteria (possibly via fat), and this effect may not have occurred
when the raw meat was irradiated. This may have led to a slightly greater survival in

cooked meat compared to raw meat. The observed fat level at the native state moisture

68

content was 1.09 and 1.78% in the raw and cooked meat, respectively. Raw and cooked
meat were both obtained from the same source, but from different lots. The difference in
the fat level was approximately 0.7%.

TABLE 4.12 P values from analyses of variance of the raw data in low fat ground

turkey (raw vs. cooked) at similar moisture contents and water activities, and heated at

60 °C.

 

 

 

TEST FACTOR LF-60°C
Time ' <0.0001
Raw or cooked 0.0606

 

 

 

4.2.4.2.2 K values

The k values, rates of thermal inactivation (assuming first order kinetics), were
calculated by linear regression of the 1n(N/No) data over time for each replication. The
difference between the raw vs. cooked ground turkey did not significantly affect the k
value (Table 4.13), but it appeared to show a slight effect on the raw data (Figure 4.10).
Therefore, it appears that the use of different meat pre-treatments did not have a large

effect on thermal resistance of Salmonella.

69

TABLE 4.13 P values from analyses of variance of k values in low fat ground turkey
(raw vs. cooked) at similar moisture contents and water activities, and heated at 60 °C.

 

TEST FACTOR LF-60°C

 

Raw or cooked 0.5505

 

 

 

4.2.5 Parts 1 and 3 combined

Parts 1 and 3 were combined to determine if the different heating methods
(waterbath vs. air oven) affected the inactivation of Salmonella.
4.2.5.1 Data

The low and high fat ground turkey data (60°C) were combined from Parts 1 and
3 in order to test whether the different heat treatments affected inactivation. The one
main difference between Parts 1 and 3 was that samples in Part 1 were heated in a
waterbath, whereas in Part 3, samples were heated in an air convection oven.

Figure 4.11 shows an example of a graph of mean data for low fat ground turkey,

heated in a waterbath and air convection oven.

70

 

 

0 Part1
El Part3
Linear (Part 1)
— —Linear (Part 3)

 

log (NlNo)

 

 

 

 

 

 

 

0 0.5 1 1.5 ° 2 2.5 3 3.5

Time (min)

FIGURE 4.11 Thermal inactivation of Salmonella in low fat ground turkey at 60 °C in
waterbath vs. air convection oven.

4.2.5.2 ANOVA

No significant difference was observed between Parts 1 and 3, in terms of
Salmonella inactivation (data not shown). The k values, rates of thermal inactivation
(assuming first-order kinetics), were calculated by linear regression of the 1n(N/No) data
over time for each replication. The difference between Parts 1 and 3 did not significantly
affect the k values (data not shown). Therefore, no significant effect of heating method

on thermal inactivation could be determined from these results.

71

CHAPTER 5

CONCLUSIONS

The USDA-FSIS recently amended the regulations governing cooked meat and
poultry products, creating a shift to lethality performance standards. Studies clearly show
that many variables affect the thermal inactivation of bacteria. However, no current
inactivation models incorporate the effect of water on microbial inactivation in meat
products, and in particular, the intrinsic properties or extrinsic process. Therefore, the
current work isolated two very important variables, water activity/moisture content and
relative humidity, and studied the effect of both parameters on thermal inactivation.

The hypothesis of the current work was that the rate of thermal inactivation
decreases with decreasing meat moisture content and/or process humidity. The
objectives of the study were to test the effects of each parameter in either a sealed
environment (intrinsic properties) or air convection oven (extrinsic properties). The
experiment was divided into three parts to study thermal inactivation of Salmonella in
ground turkey (Part 1-3).

In Part 1, the moisture content was altered in raw turkey within a narrow range,
and a clear effect was not observed. Therefore, the moisture content/water activity range
was widened by drying/cooking the meat (Part 2), and as the moisture content/water
activity decreased, thermal inactivation decreased. Previous studies reported the same
trend; however, these earlier studies were performed in carbohydrate solutions, and when

compared to a food system, the k values differed. The data from the current study were

72

applied to a secondary modified Arrhenius-type model. The data fit the model, and
resulted in a error of less than 1 logo (CFU/ g).

Previous studies have Shown that the thermal inactivation rate decreased with
decreasing humidity; however, the effects were not isolated. It remained unclear whether
the intrinsic or extrinsic parameters were actually causing the effect. Therefore, the
effect of relative humidity was isolated (Part 3), and the rate of thermal inactivation did
not decrease with decreasing process humidity.

In conclusion, decreased moisture content/water activity, and not relative
humidity in the process environment, resulted in a decreased rate of thermal inactivation.
The intrinsic properties of meat should be included in inactivation models for commercial

convection cooking systems.

73

CHAPTER 6

RECOMMENDATIONS F ORFUTURE RESEARCH

The current study examined whether moisture content/water activity or relative
humidity decreased the thermal inactivation rate by decreasing the “water parameters.”
The following is a list of recommendations for future research:

0 A method Should be developed to test the effects of moisture content (alone) in a
meat sample, with only a small range of decreasing moisture. Decreasing the
moisture content would be more applicable to industry. The current study tested
the moisture content primarily by increasing moisture over a small range. The
range was broadened, and moisture was decreased; however, a small, decreased
range should be examined.

0 The thermal inactivation rate should be tested, isolating moisture content/water
activity and humidity, in a “real world” setting using processed turkey products.

0 The oven test should be performed at more temperatures. The current study only

tested the humidity effect at 60°C.

0 A means should be designed to aseptically monitor the temperature while
undergoing oven/humidity testing. Preliminary tests were performed to determine
set temperatures at which the experiment was run. However, the sample
temperature per each individual sample was unknown.

0 Both moisture content/water activity and humidity should be tested using a single
strain of Salmonella, instead of the cocktail, to reduce the range of resistance and

decrease tailing. '

74

o A larger data set at different water activities and temperatures (relevant to
commercial processes) should be created in order to develop and validate

secondary inactivation modeling.

75

APPENDIX

76

APPENDIX A: Moisture lost during oven heating (Part 3)

LF-96%-60C

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Final MC Moisture lost (g
72.3 0
72.3 0
72.5216 001
73.4352459 005
72.3 0
71 .72886598 0.02
70.58659794 0.06
69.22222222 0.13
71.746 0.02
71.16938776 0.04
68.92195122 0.1
71.0015625 0.03
LF—90%-60C
Final MC Moisture lost (g)_
71.80973451 0.02
72.56132075 -0.01
68.56516854 0.12
72.84851485 -0.02
72.55648148 -0.01
71 .42526316 0.03
71 .04090909 0.04
70.73207547 0.06
70.78909091 0.06
69.73518519 0.1
69.64787234 0.09
70.43781513 0.08

 

 

 

77

APPENDIX A, con'd

LF—96%-60C

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Final MC Moisture lost (g)_
72.02574257 0.01
72.3 0
72.52892562 -0.01
73.408 -0.05
72.8893617 003
72.52520325 -0.01
72.92954545 -0.03
71.82241379 0.02
73.82677165 -0.07
72.08854962 0.01
71.79636364 0.02
70.52110092 0.07
LF-90%-60C
Final MC Moisture lost (g)
72.00531915 0.01
71 .45204082 0.03
73.57356322 -0.04
72.91102941 -0.03
71 .75686275 0.02
72.87113402 -0.02
70.67058824 0.05
72.3 0
70.90336134 0.06
71 .42063492 0.04
70.37304348 0.08
70.62121212 0.06

 

 

 

78

APPENDIX A, con'd

HF -96°/o-60C

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Final MC Moisture lost (g)
64.11827957 0.01
63 .20909091 0.04

64.5 0

64.5 0
12.48837209 0.63
78.61180124 -0.64
63.88793103 0.02
63.70224719 0.02
63.89830508 0.02

63.1728972 0.04
64.22265625 0.01
53.18115942 0.22

 

 

 

HF -90°/o-60C

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Final MC Moisture lost (g)
63.74468085 0.02
63.71111111 0.02
64.88172043 -0.01
65.99473684 -0.04
64.20661157 0.01
64.14851485 0.01
63.38188976 0.04
63.09126984 0.05
63.83850932 0.03
63.86036036 0.02

60.24 0.12
60.89453125 0.13

 

 

 

79

APPENDIX A, con'd

HF -96°/o-60C

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Final MC Moisture lost (g)
63.69924812 0.03
64.5 0
64.87368421 -0.01
64.16190476 0.01
62.20967742 0.06
64.19130435 0.01
58.99137931 0.18
67.828125 -009
63.06565657 0.04
63.02083333 0.04
61.37912088 0.08
61.8375 0.09

 

 

HF -90°/o-60C

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Final MC Moisture lost (g)
63.76041667 0.027
64 0.01
64.0617284 0.01
64.91764706 -0.01
63.8826087 0.02
63.81730769 0.02
62.41176471 0.05
62.10674157 0.06
62.74257426 0.05
60.55555556 0.08
60.76315789 0.1
59.53763441 0.13

 

 

 

80

APPENDIX B: Inactivation data (Part 1-3)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PART 1
Temp C MC % Fat % K value R2 Regression.ofy=1n.(N/No)
(mm -1) against x=trme (mm)
65 72.3 1.0915 25.688 0.9227 y = -25.688x - 0.3578
65 72.3 1.0915 10.971 0.0939 y = -10.971x - 3.58
65 72.3 1.0915 11.653 0.6613 y = -11.653x - 0.8947
65 72.3 1.0915 61.31 0.998 y=-61.31x+0.133
65 72.3 1.0915 47.106 0.6752 y = -47.106x - 1.572
65 72.3 1.0915 65.652 0.9959 y = -65.652x - 0.2037
65 64.5 13.02 17.3 0.9769 y = -17.3x - 0.128
65 64.5 13.02 15.877 0.5652 y = -15.877x - 2.1077
65 64.5 13.02 16.976 0.8049 y = -16.976x - 1.1852
65 64.5 13.02 17.822 0.4428 y = -17.822x - 2.1011
65 64.5 13.02 32.822 0.5998 y = -32.822x - 4.3425
65 64.5 13.02 13.847 0.4125 = -13.847x - 3.9158
65 72.3 1.0915 47.106 0.6752 y = -47.106x - 1.572
65 64.5 13.02 32.822 0.5998 y = -32.822x - 4.3425
65 72.3 1.0915 11.873 0.471 y = -11.873x - 4.2585
65 64.5 13.02 13.847 0.4125 y=-13.847x - 3.9158
55 72.3 1.0915 0.2914 0.9999 y = -0.2914x + 0.0279
55 72.3 1.0915 0.2169 0.8154 = -O.2169x - 2.7798
55 74.3 1.0915 0.0674 0.268 y = -0.0674x - 3.0599
55 74.3 1.0915 0.0926 0.1798 = 0.0926x - 5.1375
55 76.3 1.0915 0.1753 0.8002 y = -0.1753x - 0.6317
55 76.3 1.0915 0.2069 0.6564 y = -0.2069x - 1.945
55 64.5 13.02 0.2925 0.9917 y = -0.2925x + 0.3475
55 64.5 13.02 0.1213 0.6634 = -0.1213x - 3.3312
55 66.5 13.02 0.2807 0.8321 y = -0.2807x - 1.6381
55 66.5 13.02 0.2102 0.9998 = -0.2102x - 0.0372
55 66.5 13.02 0.0675 0.1544 y = -0.0675x - 3.6031
55 68.5 13.02 0.2332 0.9499 y = -0.2332x - 0.6955
55 68.5 13.02 0.2128 0.8028 y = -0.2128x - 1.3698
60 72.3 1.0915 3.6189 0.8311 y = -3.6189x - 1.4108
60 72.3 1.0915 3.5743 0.9146 = -3.5743x - 0.4979
60 74.3 1.0915 4.4387 0.9613 = -4.4387x - 1.2101
60 74.3 1.0915 1.9279 0.7621 y = -1.9279x - 0.6053
60 74.3 1.0915 5.3411 0.9613 y = -5.3411x - 0.464
60 76.3 1.0915 3.7027 0.8571 = -3.7027x - 1.464
60 76.3 1.0915 3.906 0.8625 y = -3.906x - 1.071
60 76.3 1.0915 4.4181 0.9175 y = -4.4181x - 0.7706

 

 

 

 

 

 

81

 

APPENDIX B, Con'd

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

60 64.5 13.02 3.6836 0.9024 y = -3.6836x - 1.6104
60 64.5 13.02 3.5938 0.8881 y = -3.5938x - 0.9464
60 66.5 13.02 3.9753 0.924 y = -3.9753x - 1.6031
60 66.5 13.02 3.3075 0.8319 y = -3.3075x - 1.7134
60 66.5 13.02 5.6719 0.9859 y = -5.6719x - 0.2937
60 68.5 13.02 3.9887 0.8877 y = -3.9887x - 1.2742
60 68.5 13.02 5.5219 0.9399 y = -5.5219x - 1.1154
60 68.5 13.02 4.2253 0.9531 y = -4.2253x - 0.9784
65 72.3 1.0915 43.066 0.926 y = -43.066x - 1.1669
65 72.3 1.0915 18.294 0.153 y = -l8.294x - 6.3279
65 74.3 1.0915 41.404 0.8778 y = -41 .404x - 0.5975
65 74.3 1.0915 20.572 0.4096 y = -20.572x - 4.095
65 76.3 1.0915 29.201 0.7374 y = -29.201x - 2.4432
65 76.3 1.0915 44.03 0.9458 y = -44.03x - 1.5877
65 76.3 1.0915 17.433 0.2957 y = -17.433x - 3.2266
65 64.5 13.02 31.382 0.6846 y = -3l.382x - 2.0978
65 64.5 13.02 29.027 0.8822 = -29.027x - 1.6609
65 66.5 13.02 49.924 0.961 y = -49.924x + 0.0631
65 66.5 13.02 7.2754 0.0539 y = -7.2754x - 1.9004
65 68.5 13.02 37.727 0.9817 y = -37.727x - 0.6134
65 68.5 13.02 43.006 0.9059 y = -43.006x - 1.1311
65 68.5 13.02 32.87 0.9581 y = -32.87x - 1.1007
60 72.3 1.0915 3.9537 0.9264 y = -3.9537x - 1.2963
60 64.5 13.02 2.7795 0.6323 y = -2.7795x - 2.5999
55 72.3 1.0915 0.2435 0.9623 y = -0.2435x - 0.6258
55 64.5 13.02 0.1263 0.7552 y = -0.1263x - 3.0386
65 72.3 1.0915 61.31 0.998 y=-61.3lx+0.133
65 64.5 13.02 17.822 0.4428 y = -l7.822x - 2.1011
55 72.3626 1.0915 0.1474 0.5748 y = -0.1474x - 1.6467
55 71.454 1.0915 0.1564 0.9102 y = -O.1564x - 1.6007
55 71.9273 1.0915 0.1623 0.8964 y = -O.1623x - 1.4992
55 71.4248 1.0915 0.1971 0.9294 y = -0.197lx - 1.488
55 71.9598 1.0915 0.2335 0.9329 y = -0.2335x + 1.2427
55 71.7505 1.0915 0.0589 0.2791 y = -0.0589x - 4.0281
65 72.3626 1.0915 45.217 0.969 y = -45.217x - 0.0752
65 71.9273 1.0915 43.831 0.8184 y = -43.83lx - 1.9866
65 71.9598 1.0915 10.722 0.1122 y = -10.722x - 4.3436
65 71.7505 1.0915 27.914 0.536 y = -27.914x - 3.8563
55 72.493 1.0915 0.1917 0.8413 y = -0.19l7x - 2.0347
55 1.0915 0.1433 0.7011 y=-0.1433x -4.17l4

 

72.2029

82

 

APPENDIX B, con'd

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

55 72.3925 1.0915 0.2216 0.8617 y = -0.2216x - 2.4289
55 71.8723 1.0915 0.1874 0.8174 y = -O.1874x - 2.0627
55 70.9637 1.0915 0.1519 0.5911 y = -0.1519x - 3.3367
55 71.4545 1.0915 0.1633 0.7563 y = -0.1633x - 3.891
65 72.493 1.0915 41.059 0.8091 y = -41.059x - 2.8638
65 72.2029 1.0915 45.015 0.8666 y = -45.015x - 1.7957
65 72.3925 1.0915 38.849 0.8797 y = -38.849x - 2.1896
65 71.8723 1.0915 32.627 0.6878 y = -32.627x - 3.3337
65 70.9637 1.0915 45.986 0.8769 y = -45.986x - 1.7454
65 71.4545 1.0915 33.519 0.7139 y = -33.519x - 2.7798
PART 2
Temp C MC ”/0 Fat % aw K value R2

(mm -1)
60 37.12522 4.285198 0.95 1.315 0.9729
60 54.3668 3.036015 0.981 3.0081 0.9469
60 72.51162 1.780543 0.996 3.7763 0.9589
60 37.12522 4.285198 0.95 1.2877 0.9662
60 54.3668 3.036015 0.981 2.2889 0.9733
60 72.51162 1.780543 0.996 3.362 0.9746
PART 3
Temp C MC % Fat % RH % K Yam" R2

(mm -1)
60 72.3 1.0915 96 4.0202 0.8843
60 72.3 1.0915 90 3.5024 0.769
60 72.3 1.0915 96 3.0003 0.981
60 72.3 1.0915 90 3.7301 0.809
60 64.5 13.02 96 2.8326 0.9698
60 64.5 13.02 90 3.6867 0.8529
60 64.5 13.02 96 3.0196 0.9822
60 64.5 13.02 90 3.7248 0.9019

 

 

 

 

 

 

83

 

 

 

APPENDIX B, con'd

 

Regression of y=ln(N/No)

 

lgginst x=time (min)
y= -1.315x - 0.1762

 

y= -3.0081x - 0.118

 

y= -3.7763x - 0.5398

 

y = -l.2877x + 0.2682

 

y = -2.2889x - 0.3405

 

 

y = -3.362x - 0.403

 

 

 

Regression of y=ln(N/N0)
against x=time (min)

 

y: -4.0202x - 1.3034

 

y = -3.5024x - 3.0254

 

y = -3.0003x + 0.3819

 

y= -3.7301x - 1.8851

 

y = -2.8326x + 0.3463

 

y = -3.6867x - 0.739

 

y = -3.0196x + 0.4625

 

 

y= -3.7248x - 1.1297

 

 

84

APPENDIX C: Output from statistical analyses (Part 1-3)

#1 — Part 1 - Raw data

(For Table 4.4, column a)

Response in N/No

Summary of Fit

RSquare

RSquare Adj

Root Mean Square Error

0.403525
0.387273
3.992153
-6.87545

Mean of Response
Observations (01' Sum W gts)

Analysis of Variance

378

Source DF Sum of Squares
Model 10 3956.9374
Error 367 5848.9832

C. Total 377 9805.9207
Lack Of Fit

Source DF Sum of Squares
Lack Of Fit 169 3905.5658
Pure Error 198 1943.4174
Total Error 367 5848.9832

Parameter Estimates
Term

Intercept

Time (min)

Temp C

MC %

Fat %

(Time (min)-9.8093)*(Temp C-6l . 1243)
(Time (min)-9.8093)*(MC %-70. 1789)
(Time (min)-9.8093)*(Fat %-5.85659)

(Temp C-61.1243)*(MC %-70.1789)
(Temp C-61.1243)*(Fat %-5.85659)

(MC %-70.l789)*(Fat %-5.85659)

Effect Tests
Source

Time (min)

Temp C

MC %

Fat %

Time (min)*Temp C
Time (min)*MC %
Time (min)*F at %
Temp C*MC %
Temp C‘Fat %
MC %*Fat %

2
"U
a:
5

H~HHH~HHHH

O

dep—fib—‘j—l—l—‘p—‘p-‘p—‘m

Mean Square

395.694

15.937

Mean Square

23.1099
9.8152

Estimate
399.26394
-4.27912
-6.97236
0.3124975
0.270407
-0.671289
0.0005768
0.001097
0.0116665
0.0201702
0.0101322

Sum of Squares

85

2182.6007
2225.0077

90.5238
134.9998

2062.7271

0.0579
0.5023
1.2897
9.4249
3.3259

F Ratio
24.8282
Prob > F

<.0001

F Ratio
2.3545
Prob > F
<.0001
Max RSq
0.8018

Std Error
36.40635
0.365657
0.590094
0.131 121
0.092909
0.059006
0.009572
0.006179
0.041011
0.026229

0.02218

F Ratio
136.9493
139.6102

5.6800

8.4707
129.4278

0.0036

0.0315

0.0809

0.5914

0.2087

t Ratio
10.97
-11.70
-11.82
2.38
2.91
-11.38
0.06
0.18
0.28
0.77
0.46

Prob > F
<.0001
<.0001
0.0177
0.0038
<.0001
0.9520
0.8592
0.7762
0.4424
0.6481

Prob>|t|
<.0001
<.0001
<.0001
0.0177
0.0038
<.0001
0.9520
0.8592
0.7762
0.4424
0.6481

APPENDIX C, con’d

#2 — Part 1 - Raw data

(For Table 4.4, column b)

Response In N/No

Whole Model

Summary of Fit

RSquare 0.417101
RSquare Adj 0.401204
Root Mean Square Error 3.915196
Mean of Response -7. 17878
Observations (or Sum Wgts) 227

Analysis of Variance

Source DF Sum of Squares
Model 6 2413.1140
Error 220 3372.3277

C. Total 226 5785.4417
Lack Of Fit

Source DF Sum of Squares
Lack Of Fit 129 2736.1469
Pure Error 91 636.1808
Total Error 220 33723277

Parameter Estimates
Term

Intercept

Time (min)

Temp C

MC %

(Time (min)-l 1.9431)*(Temp C-60.7489)
(Time (min)-11.9431)*(MC %-73.049)
(Temp C-60.7489)*(MC %-73.049)
Effect Tests

Source

Time (min)

Temp C

MC %

Time (min)*Temp C
Time (min)*MC %
Temp C*MC %

Nparm D

Hp—‘p—‘p-‘y—np—n
pdpnpdpnp—aflm

Mean Square F Ratio
402.186 26.2373
15.329 Prob > F
<.0001
Mean Square F Ratio
21.2104 3.0340
6.9910 Prob > F
<.0001
Max RSq
0.8900
Estimate Std Error
471 .62286 57.51517
-3.857806 0.463262
-8. 126511 0.962308
0.2497615 0.168525
-0.64191 1 0.07979
0.0005812 0.011151
-0.049699 0.050796
Sum of Squares F Ratio
1063.0018 69.3469
1093,1682 71.3148
33.6688 2.1964
992.1099 64.7221
0.0416 0.0027
14.6739 0.9573

86

t Ratio
8.20
-8.33
-8.44
1.48
-8.05
0.05
-0.98

Prob > F
<.0001
<.0001
0.1398
<.0001
0.9585
0.3289

Prob>|t|
<.0001 .
<.0001
<.0001
0.1398
<.0001
0.9585
0.3289

APPENDIX C, con’d

#3 — Part 1 - Raw data
(For Table 4.4, column c)

Response In N/N 0

Summary of Fit
RSquare

RSquare Adj

Root Mean Square Error
Mean of Response
Observations (or Sum Wgts)
Analysis of Variance
Source DF
Model 6
Error 144

C. Total 150
Lack Of Fit

Source DF
Lack Of Fit 37
Pure Error 107
Total Error 144

Parameter Estimates
Term

Intercept

Time (min)

Temp C

MC %

0.405
0.380208
4.049244

-6.4 1946
15 1

Sum of Squares
1607.1183
2361.0778
3968.1961

Sum of Squares
1053.8413
1307.2365
2361.0778

(Time (min)-6.60155)*(Temp 061.6887)
(Time (min)-6.60155)*(MC %-65.8642)
(Temp C-6l.6887)*(MC %-65.8642)
Effect Tests

Source

Time (min)

Temp C

MC %

Time (min)*Temp C
Time (min)*MC %
Temp C*MC %

Nparrn D

.‘p—ap—e—n—ep—‘T’

Mean Square F Ratio
267.853 16.3361
16.396 Prob > F
<.0001
Mean Square F Ratio
28.4822 2.3313
12.2172 Prob > F
0.0004
Max RSq
0.6706
Estimate Std Error
287.42852 40.45771
-5.014653 0.592644
-5. 103426 0.606911
0.3706851 0.208294
-0.723031 0.087321
-0.005029 0.01837
0.1124595 0.069086
Sum of Squares F Ratio
1173.9288 71.5969
1 159.3687 70.7089
51.9283 3.1671
1124.1390 68.5602
1.2286 0.0749
43.4471 2.6498

87

t Ratio
7.10
-8.46
-8.41
1.78
-8.28
-0.27
1.63

Prob > F
<.0001
<.0001
0.0772
<.0001
0.7847
0.1057

Prob>|t|
<.0001
<.0001
<.0001
0.0772
<.0001
0.7847
0.1057

APPENDIX C, con’d

#4 - Part 1 - k values
(For Table 4.5, column a)

Response R value

Summary of Fit

RSquare

RSquare Adj

Root Mean Square Error '
Mean of Response
Observations (or Sum Wgts)
Analysis of Variance

Source DF Sum of Squares Mean Square
Model 6 19371.800 3228.63
Error 80 10001.897 125.02

C. Total 86 29373.697

Lack Of Fit

Source DF Sum of Squares Mean Square
Lack OfFit 33 3161.181 95.793
Pure Error 47 6840.716 145.547
Total Error 80 10001 .897

Parameter Estimates

Term Estimate
Intercept -l43.586
Temp C 3.3322601
MC % -0.507665

Fat % -0.330035
(Temp C-60.8621)*(MC °/o-70.3188) 0.0631256
(Temp C-60.8621)*(Fat %-5.6161) -0.027434
(MC %-70.3188)*(Fat %-5.6161) 0.2709161
Effect Tests

Source Nparrn DF Sum of Squares
Temp C 1 1 17597.531

MC % 1 1 57.813

Fat % 1 1 48.565
Temp C*MC % l 1 14.933
Temp C*Fat % l 1 7.131

MC %*Fat % l 1 547.786

0.659495
0.633957
11.1814
16.18663
87

88

F Ratio
25.8242
Prob > F
<.0001

F Ratio
0.6582
Prob > F
0.8954
Max RSq
0.7671

Std Error
58.91756
0.280872
0.746551
0.529535
0.182653
0.114869
0.129427

F Ratio

140.7535

0.4624
0.3884
0.1194
0.0570
4.3815

t Ratio
-2.44

1 1.86
-0.68
-0.62
0.35
-0.24
2.09

Prob > F
<.0001
0.4985
0.5349
0.7305
0.8119
0.0395

Prob>lt|
0.0170
<.0001
0.4985
0.5349
0.7305
0.81 19
0.0395

APPENDIX C, con’d

#5 - Part 1 - k values
(For Table 4.5, column b)

Response R value

Summary of Fit

RSquare

RSquare Adj

Root Mean Square Error
Mean of Response
Observations (or Sum Wgts)
Analysis of Variance

Source DF Sum of Squares Mean Square
Model 3 14752.730 4917.58
Error 50 7466.603 149.33

C. Total 53 22219.334

Lack Of Fit

Source DF Sum of Squares Mean Square
Lack Of Fit 27 2215.371] 82.051
Pure Error 23 5251.2324 228.314
Total Error 50 7466.6035

Parameter Estimates

Term Estimate
Intercept -67.831 1 1
Temp C 3.5752004
MC % -l.797851
(Temp C-60.6481)*(MC %-72.9859) -0.118713
Effect Tests

Source Nparrn DF Sum of Squares
Temp C l 1 13619.305

MC % l 1 446.143
Temp C*MC % 1 1 33.516

0.663959
0.643797
12.22015
17.77034

54

89

F Ratio
32.9305
Prob > F
<.0001

F Ratio
0.3594
Prob > F
0.9941
Max RSq
0.7637

Std Error
79.2747
0.374369
1.040144
0.25058

F Ratio
91.2015
2.9876
0.2244

t Ratio
-0.86
9.55
-1.73
-0.47

Prob > F
<.0001
0.0901
0.6377

Prob>|t|
0.3963
<.0001
0.0901
0.6377

APPENDIX C, con’d

#6 - Part 1 - k values
(For Table 4.5, column c)

Response k value

Summary of Fit

RSquare 0.658678

RSquare Adj 0.623369

Root Mean Square Error 8.944401

Mean of Response 13.5951

Observations (or Sum Wgts) 33

Analysis of Variance

Source DF Sum of Squares Mean Square
Model 3 4477.2288 1492.41
Error 29 2320.0672 80.00

C. Total 32 6797.2960

Lack Of Fit

Source DF Sum of Squares Mean Square
Lack Of Fit 5 730.5840 146.117
Pure Error 24 1589.4832 66.228
Total Error 29 2320.0672

Parameter Estimates

Term Estimate
Intercept -259.5 101
Temp C 2.8552049
MC % 1.4992956
(Temp C-61.2121)*(MC %-65.9545) 0.3789683
Effect Tests

Source Nparrn DF Sum of Squares
Temp C l l 4286.1626

MC % 1 1 193.2031
Temp C*MC % l 1 196.6433

90

F Ratio
18.6546
Prob > F
<.0001

F Ratio
2.2063
Prob > F
0.0870
Max RSq
0.7662

Std Error
72.65463

0.39008
0.964787
0.241721

F Ratio
53.5755
2.4150
2.4580

tRatio
-3.57
7.32
1.55
1.57

Prob > F
<.0001
0.1310
0.1278

Prob>|t|
0.0013
<.0001
0.1310
0.1278

APPENDIX C, con’d

#7 - Part 2 - Raw data

(For Table 4.6, column a)
Response In N/No
Summary of Fit
RSquare 0.950327
RSquare Adj 0.947666
Root Mean Square Error 0.769737
Mean of Response -3.97773
Observations (or Sum Wgts) 60
Analysis of Variance
Source Sum of Squares Mean Square F Ratio
Model 634.78981 211.597 357.1279
Error 33.17974 0.592 Prob > F
C. Total 667.96955 <.0001
Lack Of Fit
Source Sum of Squares Mean Square F Ratio
Lack Of Fit 14.632414 1.33022 3.2274
Pure Error 18.547323 0.41216 Prob > F
Total Error 33.179738 0.0026

Max RSq

0.9722

Parameter Estimates
Term Estimate Std Error t Ratio Prob>|t|
Intercept 5.8261487 0.413525 14.09 <.0001
Time (min.) -2.506345 0.093689 -26.75 <.0001
MC 01 10565 0.006878 -16.08 <.0001
(Time (min.)-1.5)*(MC-54.6679) -0.063971 0.006485 -9.87 <.0001
Effect Tests
Source Nparrn DF Sum of Squares F Ratio Prob > F
Time (min) 1 1 424.0191] 715.6497 <.0001
MC 1 1 153.11004 258.4156 <.0001
Time (min.)*MC 1 1 57.66066 97.3183 <.0001

91

APPENDIX C, con’d

#8 — Part 2 — Raw data
(For Table 4.6, column b)

Response In N/No

Summary of Fit

RSquare

RSquare Adj

Root Mean Square Error
Mean of Response
Observations (or Sum Wgts)
Analysis of Variance

0.95163
0.949039
0.759576
-3.97773

60

Source DF Sum of Squares Mean Square F Ratio
Model 3 635.66002 21 1.887 367.2493
Error 56 32.30953 0.577 Prob > F
C. Total 59 667.96955 <.0001
Lack Of Fit
Source DF Sum of Squares Mean Square F Ratio
Lack Of Fit 11 13.762207 1.25111 3.0355
Pure Error 45 18.547323 0.41216 Prob > F
Total Error 56 32.309530 0.0041

Max RSq

0.9722

Parameter Estimates
Term Estimate Std Error t Ratio Prob>lt|
Intercept 81.27421 1 4.997822 16.26 <.0001
Time (min) -2.506345 0.092453 -27.11 <.0001
Aw -83.52486 5.11951 -16.32 <.0001
(Time (min.)-l .5)*(Aw-0.97567) 48.42232 4.82672 -10.03 <.0001
Effect Tests
Source Nparrn DF Sum of Squares F Ratio Prob > F
Time (min) 1 1 424.01911 734.9246 <.0001
Aw I 1 153.57389 266.1796 <.0001
Time (min.)*Aw 1 1 58.06702 100.6438 <.0001

92

APPENDIX C, con’d

#9 - Part 2 - k values
(For Table 4.7, column a)

Response k value

Summary of Fit

RSquare 0.923759

RSquare Adj 0.904699

Root Mean Square Error 0.325197

Mean of Response 2.506333

Observations (or Sum Wgts) 6

Analysis of Variance

Source DF Sum of Squares Mean Square F Ratio

Model 1 5.1253581 5.12536 48.4654

Error 4 0.4230121 0.10575 Prob > F

C. Total 5 5.5483701 0.0022

Lack Of Fit

Source DF Sum of Squares Mean Square F Ratio

Lack Of Fit 1 0.07819285 0.078193 0.6803

Pure Error 3 0.34481921 0.114940 Prob > F

Total Error 4 0.42301206 0.4700
Max RSq

0.9379

Parameter Estimates

Term Estimate Std Error t Ratio Prob>|t|

Intercept -0.990789 0.519584 -1 .91 0.1292

mc 0.0639703 0.009189 6.96 0.0022

Effect Tests

Source Nparrn DF Sum of Squares F Ratio Prob > F

are 1 1 5.1253581 48.4654 0.0022

93

APPENDIX C, con’d

#10 — Part 2 - k values
(For Table 4.7, column b)

Response k value

Summary of Fit

RSquare 0.930272
RSquare Adj 0.91284
Root Mean Square Error 0.310997
Mean of Response 2.506333

Observations (or Sum Wgts)
Analysis of Variance

6

Source DF Sum of Squares Mean Square F Ratio

Model 1 5.1614932 5.16149 53.3657

Error 4 0.3868769 0.09672 Prob > F

C. Total 5 5.5483701 0.0019

Lack Of F it

Source DF Sum of Squares Mean Square F Ratio

Lack Of Fit 1 0.04205768 0.042058 0.3659

Pure Error 3 0.3448 1921 0.114940 Prob > F

Total Error 4 0.38687689 0.5879
Max RSq

0.9379

Parameter Estimates

Term Estimate Std Error t Ratio Prob>lt|

Intercept -44.73762 6.468426 -6.92 0.0023

water act 48.422229 6.628473 7.31 0.0019

Effect Tests

Source Nparrn DF Sum of Squares F Ratio Prob > F

water act 1 1 5.1614932 53.3657 0.0019

94

APPENDIX C, con’d

#11 - Part 3 — Raw data
(F or Table 4.10, column a)

Response In N/No

Summary of Fit

RSquare 0.856687
RSquare Adj 0.834294
Root Mean Square Error 1.767196
Mean of Response -6.13153
Observations (or Sum W gts) 75

Analysis of Variance

Source DF Sum of Squares
Model 10 1 194.7705
Error 64 199.8708

C. Total 74 13946413
Lack Of Fit

Source DF Sum of Squares
Lack Of Fit 63 199.86015
Pure Error 1 0.01060
Total Error 64 199.87076

Parameter Estimates

Term

Intercept

Time (min)

RH %

final me %

Fat %

(Time (min)-l.53)*(RH %-92.8)
(Time (min)-l.53)*(final me %-67.2343)
(Time (min)-1.53)*(Fat %-7.13527)
(RH %-92.8)*(final mc %-67.2343)
(RH %-92.8)*(Fat %-7.13527)

(Fat %-7.l3527)*(final me %-67.2343)

Effect Tests

Source

Time (min)

RH%

final me %

Fat %

Time (min)*RH %

Time (min)*frnal me %

Time (min)*Fat %
RH %*frnal me %
RH %*Fat %

Fat %‘final me %

Z
'0
ca
.3.

H##H_H--

“#thydydfl—‘de’n

Mean Square

t Ratio
-3.34
-13.69
5.44
0.46
1.38
1.68
-0.99
-1.02
1.18
1.23
-0.94

Prob>lt|
0.0014
<.0001
<.0001
0.6459
0.1710
0.0982
0.3238
0.3119
0.2423
0.2234
0.3488

F Ratio
119.477 38.2574
3.123 Prob > F
<.0001
Mean Square F Ratio
3.17238 299.2562
0.01060 Prob > F
0.0459
Max RSq
1.0000
Estimate Std Error
-44.74747 13.38503
-3.465307 0.253064
0.3889134 0.071496
0.0818721 0.177325
0.1833283 0.132424
0.1524578 0.09085
-0.127683 0.128414
-0.099054 0.097182
0.0744757 0.06311 1
0.0594761 0.048376
-0.031204 0.033059
Sum of Squares F Ratio
585.58976 187.5099
92.40874 29.5899
0.66574 0.2132
5.98543 1.9166
8.79463 2.8161
3.08754 0.9887
3.24446 1.0389
4.34896 1.3926
4.72066 1.5116
2.78229 0.8909

95

Prob > F
<.0001
<.0001
0.6459
0.1710
0.0982
0.3238
0.31 19
0.2423
0.2234
0.3488

APPENDIX C, con’d

#12 — Part 3 - Raw data
(For Table 4.10, column b)

Response In N/No

Summary of Fit

RSquare

RSquare Adj

Root Mean Square Error
Mean of Response
Observations (or Sum Wgts)
Analysis of Variance

Source DF
Model 6
Error 30
C. Total 36
Parameter Estimates
Term

Intercept

Time (min)

RH %

final mc %

(Time (min)-1.52027)*(RH %-92.7568)

0.835362
0.802435
2.062154
-6.84133

37

Sum of Squares
647.30535
127.57440
774.87975

(Time (min)-l.52027)*(final me %-71.6382)

(RH %-92.7568)*(final me %-7l.6382)

Effect Tests

Source

Time (min)

RH %

final mc %

Time (rnin)*RH %
Time (min)*final mc %
RH %‘final mc %

Nparrn

~——ep——n—

D

p—‘p—np—oudp—A—m

Mean Square F Ratio
107.884 25.3697
4.252 Prob > F
<.0001
Estimate Std Error
-50.53718 23.71511
-3.480053 0.394869
0.3153437 0.126101
0.2589522 0.357111
0.3841554 0.184746
-0.830827 0.360285
0.406191 0.176535
Sum of Squares F Ratio
330.29953 77.6722
26.59345 6.2536
2.23601 0.5258
18.38672 4.3238
22.61361 5.3177
22.51338 5.2942

96

t Ratio
-2.13
-8.81

2.50
0.73
2.08
-2.31
2.30

Prob > F
<.0001
0.0181
0.4740
0.0462
0.0282
0.0285

Prob>|t|
0.0414
<.0001
0.0181
0.4740
0.0462
0.0282
0.0285

APPENDIX C, con’d

#13 - Part 3 — Raw data
(For Table 4.10, column c)

Response In N/No

Summary of Fit

RSquare

RSquare Adj

Root Mean Square Error
Mean of Response
Observations (or Sum Wgts)
Analysis of Variance

0.920269
0.904838
1.224489
-5.44041

38

Source DF Sum of Squares
Model 6 536.48950
Error 31 46.48055

C. Total 37 582.97005
Lack Of Fit

Source DF Sum of Squares
Lack Of Fit 30 46.469949
Pure Error 1 0.010601
Total Error 31 46.480550

Parameter Estimates
Term

Intercept

Time (min)

RH %

final mc %

(Time (min)-1.53947)*(RH %-92.8421)
(Time (min)-l .53947)*( final me %-62.9463)
(RH %-92.8421)*(final me %-62.9463)
Effect Tests

Source

Time (min)

RH %

final me %

Time (min)*RH %
Time (min)*final me %
RH %‘final me %

p—ep—Ap—‘p—ap—ep—a

Nparm D

y—‘y—dp—lp—‘p—dp—Im

Mean Square F Ratio
89.4149 59.6349
1.4994 Prob > F
<.0001
Mean Square F Ratio
1.54900 146.1196
0.01060 Prob > F
0.0654
Max RSq
1.0000
Estimate Std Error
-24.31497 14.10394
-3.633116 0.316625
0.4177081 0.066414
-0.227224 0.195133
0.2277787 0.097777
0.0230355 0.105119
0.0653434 0.055137
Sum of Squares F Ratio
197.41326 131.6639
59.31084 39.5571
2.03309 1.3560
8.13693 5.4269
0.07200 0.0480
2.10582 1.4045

97

tRatio
-1.72
-11.47
6.29
-1.16
2.33
0.22
1.19

Prob > F
<.0001
<.0001
0.2531
0.0265
0.8280
0.2450

Prob>|t|
0.0947
<.0001
<.0001
0.2531
0.0265
0.8280
0.2450

APPENDIX C, con’d

#14 - Part 3 — K values
(For Table 4.11, column a)

Response K value

Summary of Fit
RSquare

RSquare Adj

Root Mean Square Error
Mean of Response
Observations (or Sum Wgts)
Analysis of Variance
Source DF
Model 3
Error 4

C. Total 7
Parameter Estimates
Term

Intercept

RH %

Fat Content

(RH %-93)*(Fat Content-7.05575)

Effect Tests
Source

RH % 1
Fat Content 1
RH %*Fat Content 1

0.567857
0.24375
0.375577
3.439588
8

Sum of Squares
0.7414294
0.5642320
1.3056614

Mean Square
0.247143
0.141058

Estimate
10.449669
-0.073 804
-0.020734
-0.009412

Sum of Squares
0.39218796
0. 12233931
0.22690216

98

F Ratio
1.7521
Prob > F
0.2947

Std Error
4.121517
0.044262
0.022264
0.007421

F Ratio
2.7803
0.8673
1.6086

t Ratio Prob>|t|
2.54 0.0643
-1.67 0.1708
-0.93 0.4044
-1.27 0.2735
Prob > F
0.1708
0.4044
0.2735

APPENDIX C, con’d

#15 - Part 3 — K values
(For Table 4.11, column b)

Response K value

Summary of Fit

RSquare

RSquare Adj

Root Mean Square Error
Mean of Response
Observations (or Sum Wgts)
Analysis of Variance

Source DF

Model 1

Error 2

C. Total 3
Parameter Estimates

Term Estimate
Intercept 5.20625
RH % -0.017667
Effect Tests

Source Nparrn DF
RH % 1 1

0.020163
-O.46976
0.522504
3.56325
4

Sum of Squares
0.01123600
0.54602165
0.55725765

Std Error
8.103031
0.087084

Mean Square
0.01 1236

0.273011

t Ratio
0.64
-0.20

Sum of Squares
0.01 123600

99

Prob>|t|
0.5864
0.8580

F Ratio
0.0412

F Ratio
0.0412

Prob > F

0.8580

Prob > F
0.8580

APPENDIX C, con’d

#16 - Part 3 - K values
(For Table 4.11, column c)

Response K value

Summary of Fit

RSquare

RSquare Adj

Root Mean Square Error
Mean of Response
Observations (or Sum Wgts)
Analysis of Variance

Source DF

Model 1

Error 2

C. Total 3
Parameter Estimates

Term Estimate
Intercept 15 .4005
RH % -0. 129942
Effect Tests

Source Nparrn DF
RH % 1 1

0.970913
0.95637
0.095421
3.315925
4

Sum of Squares
0.60785412
0.0182 1031
0.62606443

Std Error
1.479794
0.015903

Mean Square
0.607854

0.009105

t Ratio
10.41
-8.17

Sum of Squares
0.60785412

100

Prob>|ti
0.0091
0.0147

F Ratio
66.7594

F Ratio
66.7594

Prob > F

0.0147

Prob > F
0.0147

APPENDIX C, con’d

#17 — Part 1 and 2 - Raw values

(For Table 4.12)

Response In N/No

Summary of Fit

RSquare 0.915669

RSquare Adj 0.910399

Root Mean Square Error 1.229352

Mean of Response -6.17514

Observations (or Sum W gts) 35

Analysis of Variance

Source DF Sum of Squares Mean Square
Model 2 525.11568 262.558
Error 32 48.36181 1.511

C. Total 34 573.47748

Lack Of Fit

Source DF Sum of Squares Mean Square
Lack Of Fit 7 34.037903 4.86256
Pure Error 25 14.323906 0.57296
Total Error 32 48.361809

Parameter Estimates

Term Estimate Std Error
Intercept -O.785543 0.361165
Time (min) -3.631953 0.195914
raw vs. cooked[cooked] 0.4083415 0.209952
Effect Tests

Source Nparrn DF Sum of Squares
Time (min) 1 l 519.39878
raw vs. cooked 1 1 5.71690

101

F Ratio
173.7290
Prob > F

<.0001

F Ratio
8.4868
Prob > F
<.0001
Max RSq
0.9750

tRatio
-2.18
-18.54
1.94

F Ratio

343.6753

3.7828

Prob>lt|
0.0371
<.0001
0.0606

Prob > F
<.0001
0.0606

 

APPENDIX D: Output from secondary modeling (Part 2)
Part 2-raw model-nonlinear

#1 - Water term = 1/aw
Nonlinear fit

Control panel

Report

Coverged in the gradient

 

Criterion Current Stop limit
Iteration 1 89 200
Shortening 0 15

Obj change 1.13E-10 1E-07
pmr change 3.82E-05 1E-07
Gradient 3.652 0.000001

Parameter Current value lock

3 2.0176 SSE 32.327
Ea 332479 N 60
bl -21.949

Confidence Limits
Convergence Criterion 0.05

Solution
SSE DFE MSE RMSE
32.327 58 0.557 0.747

Parameter Estimate ApproxStdErr

 

a 2.02E+62 2.89E+62
Ea 332479 0
b1 -21.949 1.413

 

102

APPENDIX D, con'd

#2 - Water term = aw

 

 

Nonlinear fit

Control panel

Report

Coverged in the gradient

Criterion Current Stop limit
Iteration 1 77 200
Shortening 0 1 5

Obj change 2.90E-08 1E-07
pmr change 0.000396 1E-O7
Gradient 9.44E—07 0.000001

Parameter Current value lock

a 5.88E+42
Ea 332479
b1 23.04

Confidence Limits
Convergence Criterion 0.05

Solution
SSE DFE MSE
32.518 58 0.561

SSE

RMSE
0.749

Parameter Estimate ApproxStdErr

a 5.88E+42 8.59E+42
Ea 332479 0
b1 23.04 1.48

32.518
60

 

103

 

APPENDIX D, con'd

#3 - Water term = aw+aw2

 

 

Nonlinear fit
Control panel

Report

Coverged in the gradient

Criterion Current Stop limit
Iteration 47 200
Shortening 0 1 5

Obj change 6.58E+10 1E-07
pmr change 0.000136 lE-07
Gradient 2.19E-08 0.000001

Parameter Current value lock

3 4.49E+47 SSE
Ea 332479 N
b1 1 1.799

Confidence Limits
Convergence Criterion 0.05

Solution
SSE DF E MSE RMSE
32.64. 58 0.563 0.75

Parameter Estimate ApproxStdErr

a 4.49E+47 3.32E+47
Ea 332479 0
b1 1 1.799 0.757

32.64
60

 

104

 

APPENDIX E: Initial microbial counts (Part 2)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CFU/g
72.5% MC Plate 1 Plate 2 AVG
Sample A 20 O 10
Sample B O 0 0
Sample A 550 360 455
Sample B 610 510 560
AVG 256.25
SD 293.297

CFU/g
54.4% MC Plate 1 Plate 2 AVG
Sample A 20 0 10
Sample B 0 0 0
Sample A 4200 4600 4400
Sample B 620 1220 920
AVG 1332.5
SD 2090

CFU/g
37.1% MC Plate 1 Plate 2 AVG
Sample A 20 80 50
Sample B 150 50 100
Sample A 50 4O 45
Sample B 40 10 25
AVG 55
SD 31.8852
OVERALL AVG 547.917
OVERALL SD 1248.27

105'

REFERENCES

106

REFERENCES

Abdul-Raouf U.M., Beuchat, LR, and Ammar, MS. 1993. Survival and growth of
Escherichia coli 0157:H7 in ground, roasted beef as affected by pH, acidulates,
and temperatures. Appl. Envir. Microbiol. 59: 2364-2368.

Ahmed, M.N., Conner, DE, and Huffman, D.L. 1995. Heat-resistance of Escherichia
coli 0157:H7 in meat and poultry as affected by product composition. J. Food
Sci. 60: 606-610.

American Meat Institute. 2000. The role of meat and poultry industry in the US.
economy. American Meat Institute. http://www.meatami.org/indupg02.htm July
16, 2002.

Ben-Embarek, P.K., and Huss, H.H. 1993. Heat resistance of Listeria monocytogenes in
vacuum packaged pasteurized fish fillets. Int. J. Food Microbiol. 20: 85-95.

Blankenship, LC. 1978. Survival of a Salmonella typhimurium experimental
contaminant during cooking of beef roasts. Appl. Envir. Microbiol. 35: 1160-
1165.

Blankenship, L.C., Davis, CE, and Magner, G.J. 1980. Cooking methods for
elimination of Salmonella Typhimurium experimental surface contaminant from
rare dry-roasted beef roasted. J. Food Sci. 45: 270-273.

CDC. December, 2001. “Salmonellosis.”
http://www.cdc.gov/ncidod/dbmd/diseaseinfo/salmonellosis_t.htm July 9, 2002.

CDC. June, 2001. “What are the most common foodbome diseases?” .
http://www.cdc.gov/ncidod/dbmd/diseaseinfo/foodbomeinfections_g.htm#mostco
mmon July 9, 2002.

Cerf, O., Davey, K.R., and Sadoudi, A.K. 1996. Thermal inactivation of bacteria-a new
predictive model for the combined effect of three environmental factors:
temperature, pH and water activity. Food Research Int. 29: 219-226.

Chick, H. 1908. An investigation of the laws of disinfection. J. Hyg. Cambridge 8, 92-
158. Cited in: Reichart, Oliver. 1994. Modeling the destruction of Escherichia
coli on the base of reaction kinetics. Intl. J. Food Microbiol. 23: 449-465.

Chiruta, J ., Davey, K.R., and Thomas, C]. 1997. Thermal inactivation kinetics of three

vegetative bacteria as influenced by combined temperature and pH in a liquid
medium. Food Bioproducts Processing. 75: 174-180.

107

Corry, J.E.L. 1975. The effect of water activity on the heat resistance of bacteria.
In Duckworth, R.B., ed. Water Relations of Foods, Proceeding of an
international symposium held in Glasgow, Sept. 1974. pp. 325-337.

Davey, K.R., Lin, SH, and Wood, DC. 1978. The effect of pH on continuous high-
temperature/short-time sterilization of liquid. Amer. Inst. Chem. Eng. J. 24:
537-540.

Davey, K.R., Hall, RF, and Thomas, C.J. 1995. Experimental and model studies of the
combined effect of temperature and pH on the thermal sterilization of vegetative
bacteria in liquid. Trans. Inst. Chem. Eng. Part C, Food Bio Products Process.

73: 127-132.

Doyle, M.E., and Mazzotta, AS. 2000. Review of studies on the thermal resistance of
Salmonellae. J. Food Prot. 63: 779-795.

Doyle, M.E., Mazzotta, A.S., Want, T., Wiseman, D.W., and Scott, V.N. 2001. Heat
resistance of Listeria monocytogenes. J. Food Prot. 64: 410-429.

Fain, Jr., A.R., Line, J .E., Moran, A.B., Martin, L.M., Lechowich, R.V., Carosella, J .M.,
and Brown, W.L. 1991. Lethality of heat to Listeria monocytogenes Scott A: D-
value and Z-value determinations in ground beef and turkey. J. Food Prot. 54:
756-761.

Farber, J .M., and Pagotto, F. 1992. The effect of acid shock on the heat resistance of
Listeria monocytogenes. Lett. Appl. Microbiol. 15: 197-201.

Foster, J .W., and Hall, H.K. 1991. Inducible pH homeostasis and the acid tolerance
response of Salmonella Typhimurium. J. Bacteriol. 173: 5129-5135.

Gailani, M.B., and Fung, D.Y.C. 1987. Critical review of water activities and
microbiology of drying of meats. CRC Critical Reviews in Food Sci. and Nutr.
25: 159-183.

Ghazala, S., Coxworthy, D., and Alkanani, T. 1995. Thermal kinetics of Streptococcus
faecium in nutrient broth/sous vide products under pasteurization conditions. J.
Food Process. and Preserv. 19: 243-257.

Goepfert, J .M., Iskander, I.K., and Amundson, CH. 1970. Relations of the heat
resistance of Salmonellae to the water activity of the environment. Appl.

Microbiol. 19: 429-433.

Goodfellow, S.J., and Brown, W.L. 1978. Fate of Salmonella inoculated into beef for
cooking. J. Food Prot. 41: 598-605.

Jay, J.M. 1996. Modern Food Microbiology, 5th ed. Chapman & Hall, New York.

108

J uneja, V.K., Snyder, J r., OR, and Marmer, BS. 1997. Thermal destruction of
Escherichia coli 0157:H7 in beef and chicken: determination of D- and z-values.
Intl. J. Food Microbiol. 35: 231-237.

J uneja, V.K., Foglia, T.A., and Manner, BS. 1998. Heat resistance and fatty acid
composition of Listeria monocytogenes: Effect of pH, acidulant, and growth
temperature. J. Food Prot. 61: 683-687.

J unej a, V.K., and Eblen, BS. 1999. Predictive thermal inactivation model for Listeria
monocytogenes with temperature, pH, NaCl, and sodium phrophosphate as
controlling factors. J. Food Protect. 62: 986-993.

Juneja, V.K., and Eblen, BS. 2000. Heat inactivation of Salmonella Typhimurium
DT104 in beef as affected by fat content. Lett. Appl. Microbiol. 30: 461-467.

J uneja, V.K., Eblen, BS, and Ransom, GM. 2001. Thermal inactivation of Salmonella
spp. in chicken broth, beef, pork, turkey, and chicken: determination of D and 2
values. J. Food Sci. 66: 146:152.

Kirby, R.M., and Davies, R. 1990. Survival of dehydrated cells of Salmonella
Typhimurium LT2 at high temperatures. J. Appl. Bacteriol. 68: 241-246.

Kotrola, J .S., and Conner, DE. 1997. Heat inactivation of Escherichia coli 0157:H7 in
turkey meat as affected by sodium chloride, sodium lactate, polyphosphate, and
fat content. J. Food Prot. 60: 898-902.

Line, J .E., Fain, A.R., Moran, A.B., Martin, L.M., Lechowich, R.V., Carosella, J .M. and
Brown, W.L. 1991. Lethality of heat to Escherichia coli 0157:H7: D-value and
z-value determinations in ground beef. J. Food Prot. 54: 762-766.

Maurer, J .L. 2001. Environmental effects on the thermal resistance of Salmonella,
Escherichia coli 0157:H7, and triose phosphate isomerase in ground turkey and
beef. MS. Thesis. Michigan State University, East Lansing, MI.

Murphy, R.Y., Marks, B.P., Johnson, ER, and Johnson, MG. 2000. Thermal
inactivation kinetics of Salmonella and Listeria in ground chicken breast meat and
liquid medium. J. Food Sci. 65: 706-710.

Murphy, R.Y., Duncan, L.K., Johnson, ER, and Davis, MD. 2001a. Process lethality
and product yield for chicken patties processed in a pilot-scale air-steam

impingement oven. J. Food Prot. 64: 1549-1555.

Murphy, R.Y., Johnson, E.R., Duncan, L.K., Davis, M.D., Johnson, M.G., and Marcy,

109

J .A. 2001b. Thermal inactivation of Salmonella spp. and Listeria innocua in the
chicken breast patties processed in a pilot-scale air-convection oven. J. Food Sci.
66: 7 34-741.

Murphy, R.Y., Johnson, E.R., Marcy, J.A., and Johnson, M.G. 2001c. Survival and
growth of Salmonella and Listeria in chicken breast patties subject to time and
temperature abuse under varying conditions. J. Food Prot. 64: 23-29.

Murphy, R.Y., Duncan, L.K., Johnson, E.R., Davis, M.D., and Smith, J .N. 2002.
Thermal inactivation D- and z-values of Salmonella serotypes and Listeria
innocua in chicken patties, chicken tenders, franks, beef patties, and blended beef
and turkey patties. J. Food Prot. 65: 53-60.

Ng, H., Bayne, H.G., and Garibaldi, J .A. 1969. Heat resistance of Salmonella: the
uniqueness of Salmonella Senftenberg 775W. Appl. Microbiol. 17: 78-82.

O’Donovan-Vaughan, CE. and Upton, ME. 1999. Food microbiology and food safety
into the next millennium. Chapter 1: Food Safety. The combined effect of
reduced water activity (aw) and heat on the survival of Salmonella Typhinurium.
85-87. Ponsen and Looyen, Wageningen, The Netherlands.

Olson, J .C., and Nottingham, RM. 1980. Temperature, p. 1-37. In ICMSF, microbial
ecology of foods, vol. 1. Factors affecting life and death of microorganisms.
Academic Press, New York.

Orta-Ramirez, A., and Smith, D.M. 2002.‘ Thermal inactivation of pathogens and
verification of adequate cooking in meat and poultry products. Adv. Food Nutr.
Research. 44: 146-194.

Peleg, M., and Cole, MB. 1998. Reinterpretation of microbial survival curves. Crit.
Rev. Food Sci. 38: 353-380.

Pearson, A.M., and Gillett, TA. 1999. Processed meats, 3rd ed. Chapman & Hall,
Gaithersburg, Maryland.

Reichart, Oliver. 1994. Modeling the destruction of Escherichia coli on the base of
reaction kinetics. Intl. J. of Food Microbiol. 23: 449-465.

Riemann, H. 1960. Effect of water activity on the heat resistance of Salmonella in dry
materials. Appl. Microbiol. 16: 1621-1622.

Roberts, J .A., and Sockett, RN. 1994. The socio-economic impact of human Salmonella
enteritidis infection. Intl. J. Food Microbiol. 21: 117-129.

Shelef, L.A., and Yang, Q. 1991. Growth suppression of Listeria monocytogenes by
lactates in broth, chicken, and beef. J. Food Prot. 54: 283-287.

110

USDA-F SIS. 1999. Performance standards for the production of certain meats and

poultry products. US. Department of Agriculture Food Safety Inspection
Service, Washington, DC. Federal Register 64: 732-749. January 6.

USDA-F SIS. 2000. “Interim progress report on Salmonella testing of raw meat and
poultry products.” http://www.fsis.usda.gov/ophs/haccp/salmraw.htm October 9.

USDA-F SIS. 2001. Performance standards for the production of processed meat and
poultry products. 9 CFR Parts 301, 303, et al. Food Safety and Inspection
Service, US Dept. of Ag., Washington, DC. February 27.

Veeramuthu, G.J., Price, J.F., Davis, GB, Booren, A.M., and Smith, D.M. 1998.
Thermal inactivation of Escherichia coli 0157:H7, Salmonella Senftenberg, and.

enzymes with potential as time-temperature indicators in ground turkey thigh
meat. J. Food Prot. 61: 171-175.

Whiting, RC, and Buchanan, R.L. 1993. A classification of models in predictive
microbiology-a reply to K.R. Davey. Food Microl. 10: 175-177.

Zwietering, M.H., Jongenburger, I., Rombouts, F.M., and Van’t Riet, K. 1990.
Modeling of the bacterial growth curve. Appl. Envir. Microbiol. 56: 1875-1881.

111

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

' lllllllllllllllljlllllflllllllllll