:. . ~ . . ,
darn.“ 1.9.3. I
to...»

x! Q

 

 

 

 

 

THfiwb

3100)

 

LIBRARY l
Michigan Sta 3
niaersity ]

 

 

"w

 

This is to certify that the
thesis entitled
VARIABILITY FACTORS AND USE OF MILK FAT-PROTEIN
CONCENTRATION RATIO IN MICHIGAN HOLSTEIN DAIRY COWS

presented by

William Raphael

has been accepted towards fulfillment
of the requirements for

M.S. degree in Large Animal Clinical
Sciences

Major professor

 

 

Date 8/9/2001

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

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

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/ClRC/DdoDue.pG5-pt 15

 

VARIABILITY FACTORS AND USE OF MILK FAT-PROTEIN CONCENTRATION
RATIO IN MICHIGAN HOLSTEIN DAIRY COWS

By

William Raphael

AN ABSTRACT OF A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Department of Large Animal Clinical Sciences

2001

Dr. Thomas Herdt

ABSTRACT

VARIABILITY FACTORS AND USE OF MILK F AT-PROTEIN CONCENTRATION
RATIO IN MICHIGAN HOLSTEIN DAIRY COWS

By

William Raphael

Descriptive statistics of and variability factors affecting the ratio of milk fat to
milk protein concentrations (F PR) of 4359 Holstein cows from 68 Mid-western US herds
were examined by the general-linear-models analysis. Peak mean FPR was 1.37 (SD =
0.33) and nadir mean FPR was 1.13 (SD = 0.22) and occurred at the first and tenth DHIA
test days after calving, respectively. The proportion of cows with FPR < l (“inversions”)
was 8.6% at the first test day and reached a plateau at approximately 21% after test day
four. A repeated measures. multivariate regression model identified test day number,
herd, parity, season of calving, and peak milk production as significant variability factors.
Proportion of variation due to herd was 44.0% and due to parity, season of calving, and
peak milk production was not greater than 1% at any test day. Mean FPR, milk fat, and
milk protein concentrations were calculated for strata of all variability factors and plotted
against test day number. Data from the study population were used to develop a model by
which FPR distribution of individual herds could be evaluated. This model adjusts test
day FPR for known variability factors. FPR differences among herds identified by this
model are probably explained by nutritional factors, so the model may be valuable for

dairy nutrition problem-solving and monitoring.

Copyright by
WILLIAM RAPHAEL
2001

DEDICATION

This thesis is dedicated to the world’s dairy veterinary practitioners. These men and
women are often faced with lack of reliable data to interpret the performance of their

client’s herds. This thesis will help fill this void.

iv

ACKNOWLEDGMENTS

I am indebted to my committee members, Drs. Tom Herdt, Paul Bartlett, Michelle
Kopcha, and Ron Erskine for their assistance throughout this research. Particularly, Dr.
Tom Herdt, for his weekly meetings; also Dr. Leslie Fowler, for access to the herds she
worked with, Joel Dobrzelewski, for his computer programming skills, Crystal Vierhout,
for her technical assistance, and Drs. Rob Tempelman and Ted Ferris, from Michigan
State University’s Department of Animal Science, who brought new insight to the

research.

I also acknowledge my family in Australia, who always keep me in touch with

what the cows really need, and my fiancee, Jennifer Stuhler, for her love and support.

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

KEY TO SYMBOLS OR ABBREVIATIONS

INTRODUCTION

PREVIOUS WORK WITH FPR

EFFECTS OF RUMEN FERMENTATION ON MILK FAT
CONCENTRATION
Trans-Fatty Acid Hypothesis
Insulin Hypothesis

EFFECTS OF RUMEN FERMENTATION ON MILK PROTEIN
CONCENTRATION
Propionic Acid Effects
Role of Insulin
Microbial Crude Protein Effects

OTHER FACTORS AFFECTING MILK COMPOSITION
Milk Fat Concentration

Milk Protein Concentration
CONCLUSIONS

CHAPTER ONE: VARIABILITY FACTORS AFFECTING MILK FAT-
PROTEIN CONCENTRATION RATIO TN MICHIGAN

HOLSTEIN DAIRY HERDS

ABSTRACT
INTRODUCTION

vi

viii

ix

xi

14

19
20
21
23
24
24
27
29

30

30
31

MATERIALS AND METHODS
Data Collection
Statistical Analysis

RESULTS

DISCUSSION

CONCLUSIONS

CHAPTER TWO: THE USE OF REGRESSION COEFFICIENTS TO
COMPARE MILK F AT-PROTEIN CONCENTRATION RATIOS
AMONG DAIRY HERDS

ABSTRACT

INTRODUCTION
MATERIALS AND METHODS
RESULTS

DISCUSSION

CONCLUSIONS

APPENDIX ONE: TABLES

APPENDIX TWO: FIGURES

APPENDIX THREE: MICROSOFT® VISUAL BASIC PROGRAMME
USED TO CREATE THE MICROSOFT® EXCEL

SPREADSHEET APPLICATION

BIBLIOGRAPHY

vii

33
33
34
35
38
42

44
44
45
47
50
51
54

57

66

83

103

LIST OF TABLES

Table 1: Reports of Effects of Propionic Acid on Dry Matter Intake,

Energy Intake, Milk Protein Concentration, Milk Protein
Yield, Milk Yield and Serum Insulin Concentration.

Table 2: Distribution of Milk Fat-Protein Concentration Ratio
(F PR) and Median Days in Milk (DIM) at Test Day
Numbers One to Ten After Calving (n = 4359).

Table 3: Probabilities of Type I Error in Univariate Regression
Models Examining the Variability Factors of Milk Fat-
Protein Concentration Ratio at Test Day Numbers One to
Ten (n = 4359).

Table 4: Intercept and Regression Coefficients from Repeated
Measures Multivariate Regression Model Examining
Variability Factors of Milk F at-Protein Concentration Ratio,
by Test Day Number (n = 4359).

Table 5: Data Format Required by the Microsoft® Excel1
Spreadsheet for Automatic Calculation of Adjusted Milk
F at-Protein Concentration Ratios.

Table 6: Coefficients from a Multivariate Regression Model
Created to Calculate the Effects of Within-Herd Variability
Factors of Milk F at-Protein Concentration Ratio (11 = 43 59).

1© Microsoft‘R: Corporation, One Microsoft Way, Redmond, Washington 98052-6399

viii

57

58

59

60

62

63

LIST OF FIGURES

Figure 1: Proportion of Test Day Milk Fat-Protein Concentration Ratios
< 1.0 (Inversions) vs. Test Day Number (11 = 4359).

Figure 2: Herd Mean Milk F at-Protein Concentration Ratio (FPR) (n =
68) and Mean of Herd Mean FPR (O) vs. Test Day Number.

Figure 3: Milk Fat-Protein Concentration Ratio vs. Test Day Number in
68 Michigan Dairy Herds (:1 = lactation 1, o = lactation 2, A =
lactation 23). a = lactation 1 vs. lactation 2 difference, b =
lactation 1 vs. lactation 23 difference, c = lactation 2 vs.
lactation 23 difference (P < 0.05).

Figure 4: Milk Fat-Protein Concentration Ratio vs. Test Day Number in
68 Michigan Dairy Herds (O = winter calving, I = spring
calving, 1:1 = summer calving, O = fall calving). a = winter vs.
spring difference, b = winter vs. summer difference, c = winter
vs. fall difference, (1 = spring vs. fall difference, e = summer vs.
fall difference (P < 0.05).

Figure 5: Milk Fat-Protein Concentration Ratio vs. Test Day Number in
68 Michigan Dairy Herds, Stratified by Peak Lactation Milk
Production (0 = <36.3 kg [category 1], I = 36.3 kg to 45.3 kg
[category 2], D = 45.4 kg to 54.3 kg [category 3], o = 254.4 kg
[category 4]). a = category 1 vs. 2 difference, b = category 1 vs.
3 difference, c = category 1 vs. 4 difference, (1 = category 2 vs. 3
difference, e = category 2 vs. 4 difference, f = category 3 vs. 4
difference (P < 0.05).

Figure 6: Milk F at-Protein Concentration Ratio vs. Peak Lactation Milk
Production (kg) (n = 43,590).

Figure 7: Milk Fat Concentration vs. Test Day Number in 68 Michigan
Dairy Herds (CI = lactation 1, o = lactation 2, A = lactation 23).

Figure 8: Milk Protein Concentration vs. Test Day Number in 68
Michigan Dairy Herds (CI = lactation l, o = lactation 2, A =
lactation 23).

Figure 9: Milk Fat Concentration vs. Test Day Number in 68 Michigan

Dairy Herds (O = winter calving, I = spring calving, 1:1 =
summer calving, o = fall calving).

ix

66

67

68

69

70

71

72

73

74

Figure 10: Milk Protein Concentration vs. Test Day Number in 68
Michigan Dairy Herds (O = winter calving, I = spring calving, c1
= summer calving, o = fall calving).

Figure 11: Milk Fat Concentration vs. Test Day Number in 68 Michigan
Dairy Herds, Stratified by Peak Lactation Milk Production (O =
<36.3 kg, I = 36.3 kg to 45.3 kg, 13 = 45.4 kg to 54.3 kg, 0 =
254.4 kg).

Figure 12: Milk Protein Concentration vs. Test Day Number in 68
Michigan Dairy Herds, Stratified by Peak Lactation Milk
Production (O = <36.3 kg, I = 36.3 kg to 45.3 kg, E1 = 45.4 kg to
54.3 kg, 0 = 254.4 kg).

Figure 13: Welcome Worksheet in Microsoft® Excel1 Spreadsheet
Application for Automatic Calculation of Adjusted Milk Fat-
Protein Concentration Ratios.

Figure 14: Results Worksheet in Microsoft® Excel1 Spreadsheet
Application for Automatic Calculation of Adjusted Milk Fat-
Protein Concentration Ratios.

Figure 15: Results Chart in Microsoft® Excel1 Spreadsheet Application
for Automatic Calculation of Adjusted Milk F at-Protein
Concentration Ratios. This Chart Plots Herd Mean Milk Fat-
Protein Concentration Ratio, Adjusted for Parity, Season of
Calving, and Peak Milk Production vs. Test Day Number.

Figure 16: Cow Identity Worksheet, Describing Cows Included, Cows
Excluded, and Reasons For Exclusion, in Microsoft® Excel1
Spreadsheet Application for Automatic Calculation of Adjusted
Milk Fat-Protein Concentration Ratios.

75

76

77

78

79

80

81

1© Microsoft}; Corporation, One Microsoft Way, Redmond, Washington 98052-6399

KEY TO SYMBOLS OR ABBREVIATIONS

CLA = conjugated linoleic acid

DHIA = Dairy Herd Improvement Association

DIM = days in milk

DMI = dry matter intake

F PR = ratio of milk fat concentration to milk protein concentration from individual cows
MFD = milk fat depression

R2 = coefficient of determination

xi

INTRODUCTION

Dairy veterinarians are continually looking for new ways to monitor health and
production. The DHIA records individual cow milk volume and composition on many
farms. These are invaluable data. They have been used successfully for many years in the
diagnosis and management of mastitis and for within-herd and among-herd comparisons
of cows’ production. New applications of these data have emerged, such as calculation of
FPR, and the use of this ratio in assessment of herd health and nutrition. However,
assessment of the distribution of FPR in North American herds is currently difficult
because standards of comparison are not available. Similarly, there have been no reports
identifying the variability factors affecting FPR in populations of commercial herds.

The objective of this thesis is to describe the distribution of FPR in Michigan
dairy herds, and identify variability factors, excluding nutritional factors, affecting FPR.

This will facilitate identification of abnormal FPR in cows and herds.

PREVIOUS WORK WITH FPR

The F PR has been examined to assess its usefulness in predicting energy balance,
displaced abomasa, and fertility in dairy cattle.

There is an inverse relationship between FPR and energy balance. In an analysis
of ten sets of data from four experiments conducted in the first three months of lactation,
breed, parity, and ration adjusted correlation coefficients between FPR and energy
balances were -0.36 to -0.74 (P < 0.05) (1). Regression analysis gave coefficients of -
27.17 MI to 6246 MJ net energy for lactation per unit increase in F PR. Stobbs and Brett
(1974) also report an important correlation between digestible organic matter intake and
milk protein-fat concentration ratio (r = 0.6473, P < 0.001) (2). Additionally, Hagert
(1991) calculated a correlation coefficient of r = -0.68 between FPR and energy supply
(3).

The relationship between ketosis, a disease associated with negative energy
balance, and FPR, further supports the relationship between energy balance and FPR.
Heuer et a1. (1999) calculated that the odds ratio for subsequent ketosis in cows with F PR
> 1.5 at the first test day after calving was 3.2 (P < 0.05) relative to cows FPR S 1.5 (4).
Additionally, FPR > 1.5 was associated with subsequent loss of body condition (odds
ratio = 1.8, P < 0.01).

Geishauser et a1. (1998) conducted a case-control study in Ontario dairy herds to
examine the FPR at the first DHIA test day after calving in cows that subsequently
developed a displaced abomasum (5). Each case (n = 27) was matched by calving date
and herd to three control cows. A FPR 2 1.39 was 8.2 times more likely (95% confidence

interval = 2.7 to 25.0) to come from a cow subsequently diagnosed with a displaced

abomasum than a F PR < 1.39. In another case-control study, but this time examining herd
differences, Geishauser (1996) compared 27-Gerrnan-Friesian herds with displaced
abomasa to 27-control herds, matched on milk production (6). Herds with displaced
abomasa had increased F PR in the year prior to displacement being diagnosed than in
herds with no displacements. No level of statistical significance is given and the source of
the milk sample (individual cow or bulk milk tank) is not described. The authors in both
of these reports speculate that displaced abomasa are preceded by negative energy
balance and this is the reason why F PR can be predictive of displaced abomasa.

The relationship between FPR and fertility has also been researched. Heuer et a1.
(1999) calculated odds ratios for reproductive outcomes by FPR at the first DHIA test
day after calving (4). Cows with F PR > 1.5 had an increased risk of cystic ovarian disease
(odds ratio = 1.7, P < 0.05), reduced first service conception rate (odds ratio = 0.6, P <
0.01), five more days between calving and conception (P < 0.05), and 0.22 more services
per conception (P < 0.01), relative to cows with F PR 3 1.5. In support of these findings,
large (>0.4) increases in the FPR from the test day immediately before first insemination
to the test day immediately after first insemination significantly reduced the first service
pregnancy risk (odds ratio = 0.49, P = 0.0002) (7). All other positive changes in FPR and
large negative changes (>-0.20) in FPR also significantly reduced the risk of pregnancy,
although the odds ratios were closer to one (odds ratios = 0.77 to 0.87, P 5 0.0128) (7). In
contrast, Kristula et a1. (1995) failed to demonstrate significance of first test day FPR in a
multivariable Cox proportional hazards model identifying the variability factors affecting
first insemination pregnancy rates, although it was significant in a bivariate Cox model

analysis (8). Also, small correlation coefficients between 100-day and 305-day mean FPR

and four fertility traits (days from calving to first insemination, number of inseminations,
number of days open, and 56-day non-return rate after first insemination) have been
published (9). This is not surprising, considering how milk composition changes over the
lactation.

It would seem that FPR can be a useful predictive tool to monitor herd health,
particularly energy balance in recently calved cows, but also, possibly, individual
diseases (ketosis, displaced abomasa) and infertility.

Interpretation of herd FPR data for nutritional and health monitoring is currently
difficult because there has been no report, to our knowledge, of studies in North America
directly examining the variability factors affecting FPR. There are several studies in
which FPR was measured, although not as the primary objective, and these are discussed
below. These reports provide some preliminary information about F PR variability factors.

Rodriquez et a1. (1985) calculated seasonal influences on FPR (10). Prediction
equations from the statistical analysis revealed that FPR increased at DHIA test days as
maximum environmental temperature increased above 29°C. It also was increased as
minimum environmental temperatures fell from 11°C to 0°C. These environmental
influences on FPR were small but significant, accounting for 0.30% to 0.51% of the
variability in FPR (P < 0.05). The effects of stage of lactation and stage of pregnancy on
FPR were also small but significant, accounting for 0.3% and 0.2% of the variation
respectively (P < 0.05). Prediction equations revealed that FPR decreased to the sixth
month of lactation then increased. Sharma et al. (1990) reported similarly small variation
in FPR attributable to stage of lactation and pregnancy (11). The trends observed from

prediction equations were a decrease in the ratio from the beginning to 4.3 months of

lactation, followed by a small increase during the remainder of the lactation. Leoffler et
al., 1999, found a small (r = -0.043) but significant (P < 0.01) negative correlation
between FPR and DIM (7).

Heuer et al. (1999) reports that FPR at the first DHIA test day after calving varies
significantly by parity (P < 0.001), being least in second parity cows (4). Additionally,
these authors report strong, but small, herd effects on FPR and small correlation

coefficients between milk yield and FPR at the first test day (r = -0.09).

EFFECTS OF RUMEN FERMENTATION ON MILK FAT

CONCENTRATION

Understanding the variability factors affecting milk fat concentration is integral to
a similar understanding of FPR. Feeding diets containing large proportions of
concentrates commonly leads to MFD. The possible reasons for this cause of MFD are
the focus of the following review.

The effect of feeding diets containing large proportions of concentrates (>50%
dry matter, (12)) on milk fat concentration is consistently negative and large. As an
example of the magnitude of MF D, Gaynor et al. (1995) classified cows as responders if
they experienced a drop in milk fat concentration of at least one percentage unit when fed
an 80% (dry matter) concentrate diet (13). The mean decrease in milk fat concentration

for this group was 38% (P = 0.0001).

T rans-F atty Acid Hypothesis

The specific biochemical changes that precede MFD are not known but
experimental evidence suggests that dietary fatty acid supply and hydrogenation are

important features.
Long-Chain Fatty Acid Hydrogenation in the Rumen

The majority of fatty acids present in ruminant diets are triglyceride esters of 18-
carbon (C13) chain fatty acids (14). Unsaturated forms of these undergo microbial
hydrogenation in the rumen to become intermediary unsaturated fatty acids or stearic acid
if hydrogenation is complete. Ruminant adipose tissue is composed of some fatty acids

originating in the rumen. Adipose tissue composition changes once gastric fermentation

begins in the young ruminant and is a reflection of rumen microbial fatty acid
hydrogenation. Demonstration of this is the increased depot fat concentration of stearic
acid in a normal lamb compared to a gnotobiotic lamb (15).

The process of fatty acid saturation in the rumen occurs in steps. Intermediate
products include various cis- and trans-fatty acid isomers with single or multiple double
bonds. Their production is demonstrated by the presence of trans-fatty acids in ruminant
adipose tissue. Experimental evidence of this is the more dilute concentration of trans-
fatty acids in gnotobiotic lambs’ depot fat compared to rumen-inoculated lambs (15).
Under normal feeding practices, rumen biohydrogenation results in a characteristic
pattern of positional isomers of C 181 where trans-1 1 comprises at least 80% (16).

Various factors have been identified which determine the fate of dietary fatty
acids. Firstly, the mix of bacterial species within the rumen determines the extent of
hydrogenation and end products formed. Sachan and Davis (1969) demonstrated that a
Borrelia Sp. hydrogenated linoleic acid to trans—1 l-octadecenoic acid (17). A
Pseudomonad enzyme was shown to isomerize oleic acid to trans-IO-octadecenoic acid
(18). This reaction was pH dependent, the optimum pH being five. Strictly anaerobic
bacteria, able to hydrogenate fatty acids, were isolated from sheep rumens (19). These
were Ruminococcus albus, two Eubacterium spp. , and two F usocillus spp. The F usocillus
organisms were able to hydrogenate oleic acid and linoleic acid to stearic acid, and
linolenic acid to cis-lS-octadecenoic acid. The others did not hydrogenate oleic acid but
converted linoleic and linolenic acids to a mixture of octadecenoic acids; trans-11-
octadecenoic acid predominated but several isomeric cis- and trans-octadecenoic acids

were produced together with isomers of non-conjugated octadecenoic acids.

Secondly, the dietary supply of fatty acids also determines the extent of
hydrogenation and end products formed. Harfoot et a1. (1973) demonstrated that the fate
of linoleic acid in in vitro rumen fluid incubations was dependent on the initial
concentration of linoleic acid and availability of hydrogen donors (20). At concentrations
of linoleic acid less than 1.0 mg/ml of rumen contents, stearic acid was the major end
product. At concentrations greater than 1.0 mg/ml, the major end product of
hydrogenation was the trans-ll-monoenoic acid. Addition of sucrose as a hydrogen
donor throughout the incubation period increased the extent of conversion to stearic acid.
The authors implied that irreversible inhibition rather than competition for hydrogen
donors was involved.

Thirdly, hydrogenation is pH dependent. In vitro rumen fermentation systems
demonstrated that hydrogenation of oleic and linoleic acids to stearic acid is more
complete at pH = 7.2 compared to pH = 6.2 (21). In support of this, Latham et al. (1972)
demonstrated that hydrogenation of linolenic and linoleic acids from soybean oil was less
in media from the rumens of cows fed diets containing large proportions of concentrates
than diets containing large proportions of roughage, where pH is likely to be less (22).
This was explained by reduced numbers of Buryrivibrio fibrisolvens and Borrelia spp.
bacteria in the rumens of cattle fed diets containing large proportions of concentrates. Qiu
et al. (2000) also demonstrated inhibition of biohydrogenation of linoleic acid in vitro,
when pH was reduced from 6.5 to 5.8 (23). This conclusion was based on increased ciS-9-
trans-l l-octadecadienoic acid, trans-Cm, oleic, and linoleic acid outflow and reduced
stearic acid outflow from the more acidic in vitro fermentation systems compared to less

acidic systems.

In summary, biohydrogenation in the rumen is likely to be less complete when the
organisms capable of hydrogenation of dietary fatty acids are absent from the rumen,

dietary fatty acids are concentrate in the rumen, and when rumen pH is relatively low.
Trans-Fatty Acids and Milk Fat Concentration Relationship

Davis and Brown (1969) made the first report of a suspected association between
rumen fatty acid hydrogenation and MFD (24). Wide support for this theory has been
received since then.

Milk fat content of trans-C13] fatty acids is negatively associated with milk fat
concentration. Griinari et al. (1998) report R2 = 0.82 across 13 studies involving diets
containing large proportions of concentrates (25). When four studies involving diets
supplemented with unsaturated oils were added to the analysis, the R2 was 0.54. In a
study examining milk fat concentrations and milk fat trans-Cm concentrations while
feeding various oils, Wonsil et al. (1994) calculated an R2 of 0.72 across 16 individual
milk samples (26).

The effect of trans-Cm fatty acids on milk fat concentration has been tested in
cattle by feeding two diets that commonly result in MFD. These are diets containing large
proportions of concentrates and diets supplemented with polyunsaturated oils. Both are
thought to increase the flow of trans-C13] fatty acids to the intestine.

Gaynor et al. (1995) conducted an experiment examining the effects of increasing
the proportion of concentrate in the diet on the concentration of milk fat (13). Cows
which experienced at least one unit decrease in milk fat percentage (responders) were
compared to cows which experienced less than one unit decrease (non-responders).

Responders were found to contain a significantly increased concentration of trans-C131

fatty acids in their milk fat compared to non-responders. The correlation between the
concentrations of trans-C ,3] fatty acids in milk and the concentrations of fat in milk was r
= —0.486 (P = 0.0001). Responders also had increased DM1 and milk yield compared to
non-responders, when fed the diet containing a large proportion of concentrates.

One of the earliest reports of an effect of dietary supplied trans-C18] fatty acids on
milk fat concentrations was conducted in mice by Teter et al. (1990) (27). Various fat
supplemented diets were fed which differed in the ratio of cis'-C13;1-fatty acids to trans-
Cm-fatty acids and in the concentration of linoleic acid. Diets with trans-C13] fatty acids
decreased the concentration of fat in milk. When lactating mice raised on cis-diets were
crossed to trans-diets, the concentration of milk fat was decreased to concentrations
similar to that of nursing females raised continuously on the trans—diets. Conversely,
lactating females crossed from the trans-diets to the cis-diets produced milk with
increased fat concentrations similar to mice fed the cis-diets continuously.

Wonsil et al. (1994) fed diets supplemented with various oils to cattle and found a
significant negative association between the trans-C181 flow in duodenal chyme and
deviation in milk fat concentration, and also between trans-C13] concentration in milk fat
and deviation in milk fat concentration (26). Effects on DM1 and milk production were
variable.

Abomasal infusion of experimental treatments is performed to remove the rumen
effects on dietary fatty acids. Abomasal infusion of trans-C13: .-fatty acids resulted in
markedly reduced milk fat concentrations and increased milk fat trans-C131

concentrations relative to control and cis-Cm infused groups (28). Abomasal infusion of

10

either fatty acid decreased DMI but did not affect milk production, relative to the control
group.

In summary, the concentration of milk fat is decreased when trans-C134 fatty
acids are supplemented to the diet or when rumen production of trans-C18] fatty acids
increases, as occurs when diets containing small proportions of fiber are fed.

Additionally. the effects of the two dietary sources of trans-fatty acids are
additive. In a 2 x 2 factorial experiment designed to test the effects of dietary fat
(unsaturated vs. saturated) and rumen fermentation (diets containing large proportions of
concentrates vs. large proportions of fiber) on milk fat concentration, both unsaturated fat
diets and diets containing large proportions of concentrates suppressed milk fat
concentration (P < 0.05 and P < 0.01, respectively) (25). The interaction approached
significance at P < 0.1. DM1 and milk production were only reduced by the effect of large

proportions of concentrates in the diet.
Specific Isomers Involved

Since the association between MF D and trans-C181 fatty acids has been made,
research has been conducted to identify which isomers are involved. Advances in gas
chromatography have facilitated this.

Griinari et al. (1998) observed an interesting isomer effect in a study examining
the interaction between rumen fermentation and dietary fat (25). Milk fat concentrations
decreased 7% when the fiber content was reduced in a saturated fat diet compared to 26%
when the fiber content was reduced in an unsaturated fat diet. The effect of fiber and fat
were significant (P < 0.01 and P < 0.05 respectively). The interaction term approached

Significance (P < 0.1). Trans-ll-Clg] milk fat concentration decreased as the fiber

11

content of the diet was decreased (P < 0.10). T7'CIH.S"IO'C18;[ milk fat concentration
increased significantly by reducing the fiber content in the diet (P < 0.05). A consistent
increase in the milk fat concentration of trans-11 and trans-10 isomers of C13,, were
observed when unsaturated fat was added. Trans-lO-Clg] milk fat concentration
increased proportionally more when unsaturated fat was added to diets containing large
proportions of concentrates vs. diets containing large proportions of fiber (590% vs.
112%, interaction P < 0.05), which coincided with the larger drop in milk fat
concentration. Additionally, Piperova et al. (2000) demonstrated MFD feeding a 75%
concentrate diet without buffers (29). Cows fed this diet had similar trans-fatty acid
proportions as trans-10 and trans-ll-Clgj (25 to 30%). In contrast, cows fed diets
containing larger proportions of forages, which did not experience MFD, had
approximately three times the proportion of trans-1 l-C 18;; (approximately 30%)
compared to trans-lO-Clg]. Piperova et al. (2000) also observed that trans-lO-Cm was
the predominant trans-monoene during MF D (approximately 60% total trans-fatty acids)
when a 70% concentrate diet containing five percent soybean oil was fed, and that the
proportion of trans-l I-Clgq was decreased, relative to a 40% concentrate diet (30).

These findings led to the conclusions that specific isomers are likely involved in
MFD and that these may vary for different diets and vary in the magnitude of depression
they cause. Trans-lO-Cm may be involved specifically for diets containing large
proportions of concentrates. Griinari et al. (1997) cites an unpublished analysis that
illustrates the same association (31). Against this theory is the small R2 when milk fat

concentration was regressed on milk fat concentration of trans-IO-Clgj (R2 = 0.27, n =

10).

12

Conjugated Linoleic Acids

CLA have been incriminated in the pathogenesis of MP D. CLA refer to a mixture
of positional and geometric isomers of linoleic acid (C132) with conjugated double bonds.
Reports which document dietary induced MFD and increased milk fat CLA
concentrations (13, 25, 30) and MFD when CLA is infused abomasally (32-35) or
intravenously (36) support the relationship. In contrast, Teter et al. (1990) failed to
demonstrate an effect of CLA on milk fat concentration in mice in their convincing
experiment examining the effect of trans- and cis-isomers of C18,] (27).

The isomer positions of the trans-double bonds in CLA have specific effects on
milk fat concentration. Griinari et al. (1997) speculated that the trans-10, cis-12 isomer
might preferentially cause MFD over other isomers (31). In support, there was a greater
decrease in milk fat concentration when trans-10, Ci.S‘-12-C13;2 was infused into the
abomasum compared to cis-9, trans-1 I-Clgg (32, 33). This difference was significant in
the second trial and not reported in the first. Milk yield and DM1 did not differ between
treatments in either experiment. Milk fat concentration was decreased 43% relative to
controls and milk fat concentration of trans-10. cis—12-C [82 increased ten times, in a
dietary trial with cows fed five percent soybean oil and 70% concentrate rations (30). The
milk fat concentration of cis-9, trans—l 1-C1832 was decreased during MFD.

In vitro support for the effect of trans-10, cis-12-C 13:2 comes from a bovine
mammary cell-culture study, where MC-acetate incorporation into cellular lipids was

inhibited by 60% when this isomer was added to the culture (37).

13

Biochemical Pathogenesis

It is not known how trams-C18 fatty acids decrease milk fat concentration. Milk fat
concentrations of short- and medium-chain fatty acids (C10 to C16) decrease when MFD
takes place (13, 26, 28, 35). This indicates decreased synthesis of de-novo mammary fat.
Griinari et al. (1998) report a decreased milk yield of short—chain fatty acids in diets
containing large proportions of concentrates vs. diets containing large proportions of
fiber, but no effect of diet on the milk fat concentration of the same fatty acids (25).

Hypotheses to explain decreased milk fat concentration at the cellular level are
reduced mammary triacylglyceride synthesis secondary to reduced acyl-transferase
activity (28), direct effects of the physical and chemical properties of trans-C [3;] fatty
acids (25), and decreased A-9-desaturase activity (35). Other hypotheses are listed by

Romo et al. (1994) (14).
Insulin Hypothesis

An early theory for MFD involved the effects that propionate and serum insulin
concentrations have on the distribution of fat precursors between the udder and adipose
tissue. The hypothesis that diet could affect body lipid distribution was first postulated by
McClymont and Vallance in 1962 (38) and has largely been discounted by
hyperinsulinemic-euglycemic clamp studies (39, 40).

The rumen volatile fatty acid profile after fermentation of diets containing large
proportions of concentrates is well established. The molar proportional concentration of
propionic acid increases, and those of acetic and butyric acids decrease (41-48).

Experiments have been conducted to simultaneously monitor rumen propionic

acid production and milk fat concentrations when diets containing large proportions of

14

concentrates are fed, and the association with these variables and serum glucose and
insulin.

McCullough (1966) calculated that rumen propionic acid proportional
concentration accounted for 84% of the observed variation in milk fat percentages
recorded in 34 feeding trials (49). Jenny et a1. (1974) demonstrated decreased ruminal
acetate and increased ruminal propionate proportional concentrations, and increased
serum glucose and serum insulin concentrations in mid lactation cows fed rations
containing large proportions of concentrates, compared to cows fed a control diet (50).
Serum insulin was significantly correlated with milk fat concentration (r = -0.87), ruminal
acetate (r = -0.87), ruminal propionate (r = 0.85), and serum glucose (r = 0.71). Serum
glucose was significantly correlated with milk fat concentration (r = -0.77), ruminal
acetate (r = -0.76), and ruminal propionate (r = 0.73). Linear regressions of the effect of
serum insulin and glucose and ruminal propionate on milk fat concentration showed that
propionate was the most important independent factor (R2 = 82.8, P < 0.01). Diets, in this
study, containing large proportions of concentrates, had significant negative impacts on
DMI. Jenny and Poland (1975) demonstrated increased postprandial serum glucose and
insulin concentrations in cattle fed a ration containing a large proportion of concentrates
compared to cows fed a control ration (51). The cattle fed the ration containing a large
proportion of concentrates experienced a 25% to 50% decline in milk fat concentration.
Rumen propionic acid concentrations were not measured. Hurtaud et al. (1993) infused a
mix of volatile fatty acids or propionic acid, ruminally, in isoenergetic quantities and
observed that propionic acid decreased milk fat concentration compared to the mixed

infusion group (52). There was no significant difference in milk yield or DMI (P 2 0.10).

15

Sutton et al. (1988) concluded that the severity of MFD caused by feeding diets
containing large amounts of starch in two meals, daily, reflected the concentration of
volatile fatty acid production into a short period after a meal (53). MF D was a result of
simultaneous sharp increases in propionate production in the rumen and serum insulin
concentration. The severity of MFD was reduced with more frequent feeding.

From these and other experiments it was hypothesized that diets which increased
serum insulin led to the uptake of fat precursors by adipose tissue at the expense of the
udder, and hence prevented the udder from meeting its usual requirements for fat
synthesis.

Research has been conducted to investigate this hypothesis. Benson et al. (1969)
and Askew et al. (1971) observed little change in mammary lipoprotein lipase and
glyceride synthetase activity in cows when switched from normal rations to restricted
roughage rations (54. 55). In contrast, marked increases in adipose lipoprotein lipase and
glyceride synthetase enzyme activities were observed (54). Opstvedt et al. (1967)
observed that the enzymatic capacity for fat synthesis in abdominal adipose tissue was
increased in cows fed an all-concentrate ration compared to cows fed an all-hay ration
(56). In contrast, no large differences in enzyme activities were observed among
mammary samples from cows fed the same rations. In two separate experiments,
Laarveld et al. (1981 and 1985) demonstrated the refractory nature of the mammary gland
to the effects of insulin (57, 58). In the former report, the extraction of glucose was not
altered significantly by insulin. In the later report. insulin did not appear to alter the
extraction of acetate, B-hydroxybutyrate, and triglycerides (all important milk lipid

precursors) in the mammary gland. Additionally, Benson et al. (1972) demonstrated

16

increased adipose tissue glyceride synthetase and lipoprotein lipase activities and
simultaneous MFD when diets containing large proportions of concentrates (likely to
increase serum insulin) were fed (59). Also, Vernon (1980) summarized, in an extensive
review, that insulin is the principal adipose anti-lipolytic hormone in ruminants (60).

These experiments lend support to the hypothesis that diets which stimulate
insulin release are likely to stimulate fat synthesis in adipose tissue and not in mammary
tissue.

To study the effects of insulin on milk fat concentration, without the confounding
effect of hypoglycemia, a hyperinsulinemic-euglycemic clamp technique was employed
(39, 40). Both studies maintained the clamp for four-days, with no effect on milk fat
yield. Milk fat concentration was unchanged in the first study (P < 0.40) and reduced in
the second study (P < 0.01). but both these results are confounded by decreased feed
intake and the second by increased milk yield. Additionally, the second study
demonstrated changes in milk fat composition in direct contrast to those commonly seen
in MFD. The milk fat contained an increased proportion of medium-chain fatty acids and
decreased proportion of long-chain fatty acids. Additionally, Gaynor et al. (1995) failed
to demonstrate differences in serum insulin concentrations among cows suffering from at
least one unit decline in milk fat percentage and those experiencing less than one unit
decline, when diets containing large proportions of concentrates were fed (13). These
three studies cast serious doubt on the hypothesis that serum insulin is integral in the
MFD seen in dairy cows when diets containing large proportions of concentrates are fed.

This is not conclusive, however, since Mackle et al. (1999) observed significant

17

decreases in milk fat concentration (P < 0.01), increased milk yield (P < 0.02), and no

change in DMI during another hyperinsulinemic-euglycemic clamp (61).

18

EFFECTS OF RUMEN FERMENTATION ON MILK

PROTEIN CONCENTRATION

Since FPR is calculated from the milk fat and protein concentrations, it is
important to understand, in addition to the review of milk fat previously presented, the
variability factors affecting milk protein concentration.

Dietary factors that cause MFD also affect milk protein concentration. In contrast
to the large and negative effects on milk fat concentration, feeding diets containing large
proportions of concentrates or small proportions of fiber increases milk protein
concentration to a small degree. In the experiment by Gaynor et al. (1995), cited
previously for the large decreases in milk fat concentration when 80% (dry matter)
concentrate diets were fed, milk protein concentration increased only 1.8% (P = 0.06)
(13). Although the effect seems consistent and is well documented, the reason why it
occurs is not.

Two reports calculated correlations between milk protein concentration and
energy intake (MI/day) across multiple studies (62, 63). Emery (1978) used 13 reports
and 44 treatment groups in the calculation (62): Sporndly (1989) used 53 reports and 179
treatment groups (63). The correlation coefficients were similar (r = 0.42). Regression
indicated that an increased energy intake of one megajoule metabolizable energy per day
increased the percentage of milk protein by 0.0036 units and 0.003 units (R2 = 0.17)
respectively. Sporndly found that the proportion of roughage in the diet, which was
negatively associated with milk protein concentration in a simple correlation, ceased to
become significant when the effect of energy concentration in the diet was removed.

Emery noted that the energy association seemed to be independent of the ratio of

19

concentrate to roughage. Milk yield simultaneously increased with the protein
concentration in both datasets. These two studies concluded that increasing the daily

energy intake would increase the concentration of milk protein.

Propionic Acid Effects

One hypothesis to explain this increase in milk protein concentration is the amino
acid sparing effect of providing gluconeogenic precursor (in the form of propionate)
when concentrates are fed in large quantities (64, 65).

Studies have been conducted to evaluate the effects of propionic acid on milk
protein concentration. In support, Rook and Balch (1961) and Rook et al. (1965)
demonstrated significant increases (up to eight percent) in the concentration of milk
protein with rumen propionic acid infusion (66, 67). In contrast, no significant change in
milk protein concentration and a quadratic response in milk protein yield were reported in
a trial designed to test the effects on milk composition of various quantities of propionate
infused into the rumen (68). Milk yield increased linearly as propionate was infused in
increasing quantities. Also in support of the amino acid sparing effect was the increase in
plasma concentrations of Ala and Gln with increasing propionate infusions in this trial,
and decreased concentrations of Gly and branched-chain amino acids. The later possibly
indicates decreased tissue mobilization (68). In further support, Seal and Parker (1996)
measured elevated total amino acid, essential amino acid, and non-essential amino acid
concentrations in arterial plasma, mesenteric venous, and portal venous plasma of steers,
with ruminal infusion of propionic acid (69). In a Latin square trial investigating
supplementary rumen volatile fatty acids and protein types fed to cows, Huhtanen et al.

(1998) found increased milk and protein yields with propionic acid infusion, but no

20

significant changes in milk protein concentration (70). Propionate infusion tended to
increase plasma concentrations of Met and Thr (P < 0.10) and decrease the concentration
of Gly (P < 0.05). The later is hypothesized to reflect decreased muscle amino acid
turnover and net improvement in protein status. In an experiment designed to explore the
combined effects of the source of energy (propionic acid vs. volatile fatty acid mixtures)
and protein amounts (abomasal casein) on milk composition, Hurtaud et al. (1993) failed
to demonstrate changes in either yield or concentration of milk protein with rumen
propionic acid infusion (52). Casse et al. (1994) demonstrated decreased splanchnic
release of glucose and increased release of acetate and Ala with intramesenteric venous
infusion of sodium propionate (71). This is evidence that the liver was sparing
gluconeogenic amino acids, but the experiment failed to demonstrate changes in milk
protein concentration.

The conclusion from these studies is that milk protein concentration may be
increased by rumen infusion of propionic acid mixtures. However, the effect is
inconsistent. The effect may be mediated by increased availability of circulating amino
acids for mammary protein synthesis.

Experiments are confounded by energy intake and DM1 (Table 1). This may

explain why results differ.
Role of Insulin
It is apparent from recent experiments that insulin may be involved in the
regulation of milk protein synthesis. In the same experiment that placed doubts on the

role of insulin in milk fat synthesis, Griinari et al. (1997) observed increased milk protein

concentration (P < 0.05) with a hyperinsulinemic-euglycemic clamp (72). The effect was

21

larger with simultaneous abomasal casein infusion, indicating that the response to insulin
is limited by absorbed amino acids. Milk protein concentration increased 0.03 and 0.29
percentage units from the baseline period for the clamp and clamp-casein groups
respectively. Milk yield was increased and feed intake reduced by casein and the clamp
respectively. The clamp reduced plasma branched-chain and essential amino acid
concentrations and plasma urea nitrogen; the later indicating decreased amino acid
oxidation. Together these results indicate improved efficiency of amino acid use for milk
protein synthesis with hyperinsulinemia.

In a similarly designed trial, but this time fortifying casein with branched-chain
amino acids, a hyperinsulinemic-euglycemic clamp increased milk protein concentration
and milk yield (61). Milk protein concentration increased 0.17 and 0.33 percentage units
for the clamp and clamp-protein groups respectively, compared to the control group.
Insulin treatments reduced the concentration of milk urea nitrogen and plasma urea
nitrogen and essential and branched-chain amino acids. DMI was not affected in this
study in either treatment group. In contrast, McGuire et al. (1995) showed only small
increases in the milk protein concentration (P < 0.09) by the end of a four-day
hyperinsulinemic-euglycemic clamp (39). There was no effect on milk yield and modest
increases in the yield of milk protein (P < 0.05). They did, however, demonstrate
significantly decreased plasma urea nitrogen (P < 0.001) and essential amino acid
concentrations (P < 0.01).

These reports offer another explanation of why diets containing large proportions

of concentrates, which are known to increase serum insulin (50, 53), also increase milk

22

protein concentration. Interestingly, the magnitude of the response in milk protein

concentration in two out of three of these trials is large.

Microbial Crude Protein Effects

A theory which has received less attention in the literature relative to others is the
possible association between milk protein concentration and increased duodenal flow of
bacterial crude protein, when diets containing large proportions of concentrates are fed.
Replication of rumen bacteria is known to be dependent on supply of, among many
things, fermentable carbohydrate to the rumen (73). Poore et al. (1993) conducted an
experiment to test the effects of rumen starch degradation on milk composition and yield
using diets that differed in the rate of starch degradation (dry rolled or steam flaked
sorghum (milo)) (74). Milk yield remained unchanged but milk protein concentration
increased significantly (P < 0.04) with increased rumen starch degradability. DMI was
similar between diets. The flow of bacterial nitrogen to the duodenum was greater on
diets with more degradable starch (P < 0.01).

In an experiment designed to test and compare this theory with the amino acid
sparing effect of propionate, cows were either infused with glucose ruminally or
propionate duodenally (75). Total carbon infusion and DMI were similar between groups.
Increased milk protein concentration (P < 0.08) and increased duodenal microbial crude
protein flow (P < 0.25) were observed with ruminal glucose, compared to duodenal
propionate. Milk yields were similar between the two groups.

It would seem that reasonable support exists for a positive relationship between
milk protein concentration and microbial crude protein flow to the intestine of cows fed

diets containing large proportions of concentrates.

23

OTHER FACTORS AFFECTING MILK COMPOSITION

The previous discussion focuses on the effect of feeding diets containing large
proportions of concentrates and small proportions of forage based fiber on milk fat and
protein concentrations.

There are other important factors that affect milk fat and protein concentrations,
and so also affect F PR. These have been reviewed extensively by Crabtree, 1984, Emery,
1988, Sutton, 1989, Sutton and Morant, 1989, Palmquist et al., 1993, and Fredeen, 1996,
among others (12, 64, 76-79). It is beyond the scope of this review to repeat these

authors, but rather, list the major factors involved.

Milk Fat Concentration

The factors affecting milk fat concentration can be classified as either dietary or
non—dietary. The predominant dietary influence has been the focus of the previous
review, but a significant other dietary factor is the effect of dietary fat on milk fat
concentration.

Dietary fat has a variable effect on milk fat concentration, depending on the
saturation of the fatty acids fed and the effects on rumen fermentation. The negative
effect that unsaturated fatty acids have on milk fat concentration is proposed to occur
secondary to trans-fatty acid formation. Saturated and unsaturated fatty acids are equally
capable of decreasing rumen fiber fermentation, especially when fed over six to eight
percent dry matter (12), and this also leads to MFD. The mechanism by which this occurs
has not been elucidated, but seems to be reduced if physical contact between the fat and
fiber in the ration can be prevented (77). Perhaps associated with this is a known

antimicrobial effect of dietary fat (79).

24

If the dietary fat can be protected from microbial actions so that trans-fatty acids
are not synthesized or fed in a manner so that rumen fermentation is not impaired, then
milk fat concentration may increase (12). Examples include feeding rumen-protected fat
(12) and reducing fiber and fat physical interaction within the rumen by mixing the fat
and roughage separately at feed preparation (77).

The effects of dietary fat on milk fat concentration are frequently confounded by
the positive effects on milk yield (12, 64, 79).

Other factors which affect milk fat concentration are those which affect rumen pH
and hence the biohydrogenation of dietary fatty acids. There are a myriad of such factors,
but the most significant are the type of dietary fiber (77), the physical characteristics of
forages (12, 77, 79), and dietary buffer supply (77). Rumen pH is elevated or maintained
when rumination takes place or when dietary ingredients buffer fermentation acids.
Rumination is effective because it stimulates flow of salivary bicarbonate to the rumen
and also rumen mixing, and hence absorption of fermentation acids. Feeding various
types of fibrous feeds facilitates this, such as long stem forages. By-product feedstuffs are
also fibrous (e.g. distillers’ grain) and when substituted for concentrates in the diet as a
source of fermentable carbohydrate, help normalize milk fat concentrations (64).

There are several significant non-dietary factors that affect milk fat concentration.
The most important are stage of lactation, energy balance, the cows’ environment, and
lactation number.

Reports of the effects of stage of lactation differ because of the effects of other

non-dietary and dietary factors, but the gross effects are similar among reports. These are

a decline in fat concentration to a nadir around the time of peak milk volume production,

25

then a gradual increase to the end of lactation. The nadir was reached at four-months of
lactation in Holsteins (11). The peak fat percentage was noted at the commencement of
lactation in Holsteins by Schutz et al. (1990) and the nadir at around 50 DIM (80).
Rodriquez et al. (1985) quantified the variation in milk fat concentration attributable to
stage of lactation at 3.7% for Holsteins, larger than the effects attributable to climate
(2.9%) (10).

Negative energy balance is associated with increased milk fat concentrations (1,
2) and compositional changes reflecting decreased de-novo synthesis (reduced
proportions of milk fat short-chain fatty acids and increased proportions of long-chain
fatty acids) (2). In this report, the best correlation between digestible organic matter
intake (100%, 75%, and 50% ad lib intake) and various production variables, including
fat and protein percentages and milk yield, was with fatty acid composition of milk fat
(C4 to C16 r = 0.7317, C13,; r = -0.7429) and protein2fat ratio (r = 0.6473). These
coefficients were much greater than blood non-esterified fatty acid concentration (r = -
0.2451) and milk yield (r = 0.5670).

Milk fat concentration decreases in the summer (77). Milk fat concentration
decreased as the DHIA test day maximum temperature increased from 94°C to 361°C
and decreased as relative humidity increased to 80% (10). Month of calving also affects
milk fat concentration. Schutz et al. (1990) report increased test day fat percentages in
cows calving from April to August, in Minnesota, relative to other months (80). Crabtree
(1984) and Emery (1988) both report decreased milk fat concentration with increasing

parity, but the confounding effect of increased milk production is not clear (76, 77).

26

There is a negative correlation between milk volume and milk fat concentration (81)

although this is frequently measured as a stage of lactation effect (10, 1 1).

Milk Protein Concentration

An important dietary factor decreasing milk protein concentration is lipid (12, 64,
65, 79, 82). This effect is often confounded by increased milk yield (12, 64, 65, 79, 82).
The mechanism of action has not been elucidated (64).

Milk protein concentration varies with stage of lactation, declining rapidly to a
nadir at five to ten weeks after calving and increasing slowly from then until the end of
lactation (l 1, 76, 80, 82, 83). Protein concentrations are increased in the first seven days
of lactation, relative to later lactation, largely reflecting increased milk globulin
concentrations (76).

There is a direct relationship between milk protein concentration and energy
balance. Stobbs and Brett (1974) report a decline in milk protein concentration from
4.01% to 3.67% with a reduction in feed intake from ad lib to 50% of ad lib quantities
(2). Grieve et al. (1986) calculated the correlation coefficient between milk protein
concentration and energy balance to be r = 0.12 to r = 0.47, depending on the dataset (1).

Climate also influences milk protein concentration. Although the effect of stage of
lactation was greater in magnitude, the combined effect of DHIA test day maximum and

minimum temperatures and relative humidity was significant, accounting for 6.33% of
the variability in Holsteins’ milk protein concentration (10). Stage of lactation accounted
for 9.3% of the variability. Milk protein concentration decreased as the maximum
temperature increased from 94°C to 361°C and decreased as relative humidity increased

10 80%. Ng-Kwai-Hang et al. (1982) concluded that seasonal variation may be ill defined

27

once stage of lactation, age of cow, and milk somatic cell score are accounted for (83).
However, in a review by DePeters and Cant (1992) it is concluded that hot environmental
temperatures reduce total protein content of milk (82). The reports by Schutz et al. (1990)
and Crabtree (1984) are in direct contrast to this, although the former is an analysis of the
effect of season of calving (not season of test day) and the later is based on bulk tank
milk analyses (76, 80).

The protein content of milk shows a small decline with increasing age (76, 83).
Lactation curves from Mid-western United States’ Holsteins indicate that milk protein
concentration varies very little amongst parities early in lactation, but after 225 DIM,
shows disparity (80). It is greatest in second lactation cows and least in first lactation
cows. There is a negative correlation between milk volume and milk protein

concentration ( 81) although this is frequently measured as a stage of lactation effect (10,

11).

28

CONCLUSIONS

The FPR is likely to be influenced by all factors affecting the milk components
individually. Though many of these are nutritional factors, non-nutritional factors also
affect milk composition. The purpose of this thesis is to 1) identify these non-nutritional
variability factors affecting FPR and 2) describe how clients’ FPR data can be compared

among each other. after adjustment for known significant variability factors.

29

CHAPTER ONE

VARIABILITY FACTORS AFFECTING MILK FAT-
PROTEIN CONCENTRATION RATIO IN MICHIGAN

HOLSTEIN DAIRY HERDS

W. Raphael, P. Bartlett, M. Kopcha, R. Erskine, and T. Herdt

ABSTRACT

The objective of this report is to examine distribution of and variability factors
affecting the ratio of milk fat concentration to protein concentration (FPR) in Holstein
cows in 68 Michigan commercial herds. Peak mean FPR was 1.37 (SD = 0.33) and
occurred at the first DHIA test day after calving. Nadir mean FPR was 1.13 (SD = 0.22)
and occurred at the tenth test day after calving. The proportion of cows with FPR < 1
(inversions) was 8.6% at the first test day and reached a plateau at approximately 21%
after test day four. In a repeated measures multivariable regression model with FPR at the
first ten test days as the dependent variables, FPR varied significantly by herd, parity,
season of calving, peak milk production, and test day number. The effect of peak milk
production was quadratic. All main effects had a significant interaction with test day
number. Proportion of variation due to herd was 44.0% and due to parity, season of
calving, and peak milk production did not exceed 1%. Therefore, diagnostic assessment
of a herd’s FPR distribution must take days in milk, parity, season of calving, and peak

milk production of cows into consideration, in addition to nutrition.

30

INTRODUCTION

The FPR may be a useful index of nutritional status and health. Several reports
have examined the relationship between energy balance and FPR, with correlation
coefficients in the range r = -0.36 to 074 (P < 0.05) (l, 2). This, and research indicating
increased risk of subsequent ketosis (odds ratio = 3.2, P < 0.05) and increased subsequent
body condition loss (odds ratio 1.8, P < 0.01) if FPR > 1.5 at the first test day after
calving (4), suggest that FPR is a useful measure of energy balance in dairy cattle.

The FPR has also been demonstrated to be predictive of displaced abomasa,
although with low sensitivity and specificity (80% and 68% respectively, FPR Z 1.39,
odds ratio = 8.2, P < 0.05) (5). There are also reports indicating a negative relationship
between FPR and fertility (4, 7, 8).

Interpretation of herd FPR data for nutritional and health monitoring is currently
difficult because there has been no report, to our knowledge, of studies in North America
examining the variability factors affecting FPR. There are several studies, as discussed
below, in which F PR was measured, although not as the primary objective. These reports
provide some preliminary information about FPR variability factors.

Rodriquez et al. (1985) calculated seasonal influences on FPR (10). Prediction
equations from the statistical analysis revealed that FPR was increased at DHIA test days
when maximum environmental temperature exceeded 29°C. It also was increased when
minimum test day environmental temperature was in the range 0°C to 11°C. These
environmental influences on FPR were small but significant, accounting for 0.30% to
0.51% of the variability in FPR (P < 0.05). The effects of stage of lactation and stage of

Pregnancy on FPR were also small but significant, accounting for 0.3% and 0.2% of the

31

variation respectively (P < 0.05). The F PR decreased from the start of lactation to the
sixth month of lactation, then increased. Sharma et al. (1990) reported similarly small
variation in FPR attributable to stage of lactation and pregnancy (11). The trends
observed from prediction equations were a decrease in the ratio from the beginning to 4.3
months of lactation, followed by a small increase during the remainder of the lactation. In
a study examining predictors of fertility, there was a small (r = -0.043) but significant (P
< 0.01) negative correlation between FPR and DIM (7).

Heuer et al. (1999) reports that F PR at the first DHIA test day after calving varies
significantly by parity (P < 0.001 ), being lowest in second parity cows (4). Additionally,
these authors report strong, but small, herd effects on FPR and a small correlation
coefficient between milk yield at the first test day and F PR at the first test day (r = -0.09).

The FPR is directly proportional to milk fat concentration and inversely
proportional to milk protein concentration. Therefore. the variability factors affecting
these milk components are likely to be also variability factors affecting FPR. Milk fat
concentration is affected by diet, particularly diets containing large proportions of
concentrates and containing unsaturated lipid or large quantities of any lipid (12, 13, 25).
Milk protein concentration varies with level of energy intake (62, 63), dietary
carbohydrate characteristics (13, 74), and dietary lipid content (12, 79). Both milk fat and
milk protein concentrations vary by parity or age (76, 80, 83), stage of lactation (l 1, 76,
80), seasonal factors (10, 77, 82), and level of milk production (81).

The objective of this report is to describe the variability factors, excluding
nutritional factors, affecting FPR in Michigan Holstein cows and discuss why these

associations might exist.

32

MATERIALS AND METHODS

Data Collection

Production data from 89 herds that participated in a study measuring the
prevalence and severity of lameness in Michigan Holstein herds were made available
from Dairy Records Management Systems]. Herds used in that study were selected
randomly from respondents (11 = 388) to a descriptive letter sent to all DHIA subscribing
herd owners within a l20-km radius of Lansing. Michigan (11 = 612), but who were
milking 50 to 300 cows.

Production data were gathered for lactations commencing between July 1, 1997
and June 30, 1998. If cows had calved more than once during this interval, the last
lactation was selected. Milk was sampled repeatedly during these lactations by DHIA on
occasions commonly referred to as “test days". Test days typically occur at monthly
intervals, hence there usually is a maximum of ten test days in a 305-day lactation.
Lactations with milk fat and protein concentration data for the first ten test days (test day
one to test day ten) after calving were selected.

Since test day one is always the first DHIA test after calving, and test day ten, the
tenth after calving, the test day number approximates stage of lactation. However,
considerable variation in DIM among cows at each test day is expected, due to different
time intervals between calving and test day one and different inter-test day time intervals.
To minimize this variation during the stage of lactation when milk fat and protein
concentrations change markedly, cows with DIM at the second or third test days outside

of the 2.5th to 90th percentile range were excluded. The first test day was examined in a

33

similar fashion although the lower DIM limit was increased. Lactations with DIM at the
first test day less than 7 days or greater than the 90'h percentile were excluded. If the milk
composition was estimated, rather than measured, on any test day, that entire lactation
was deleted from the analysis. Examples of such test days are when milk samples become
frozen or sour. Lactations were also excluded if any part of any test day sampling
procedure was recorded as abnormal.

Fourteen herds were excluded because the data selection criteria resulted in them
contributing less than 20 observations to the dataset. This reduced the final dataset to

4359 cows from 68 herds.

Statistical Analysis

A repeated measures, multivariable regression model was created, based on the
significance of Type III sum of squares (84). The FPR at the first ten test days after
calving (numbered one to ten) was the repeated. dependent variable. Independent
variables considered were test day number, herd, parity, season of calving, and peak milk
production. Parity was classified categorically at three levels (I, 2 or 3), as first, second,
and third or greater lactation. Parity was also considered as a continuous variable and a
categorical variable without grouping of parity three or greater. Season of calving was
classified categorically at four levels; summer (June 21 to September 21), fall (September
22 to December 21), winter (December 22 to March 19), and spring (March 20 to June
20). Peak milk production was the greatest test day milk weight during the lactation.
Linear and quadratic effects of peak milk production were evaluated. Peak milk

Production was also evaluated as a categorical variable at levels <36.3 kg, 36.3 kg to 45.3

 

 

‘ 313 Chapanoke Rd., Raleigh, NC 27695

34

kg, 45.4 kg to 54.3 kg, and 254.4 kg. Interaction terms considered were parity x peak
milk production, season of calving x peak milk production, and parity x season of
calving.

The model was created by selecting independent variables in a manual, forward,
step-wise manner, excluding variables with a level of significance greater than P = 0.05.
Shapiro-Wilk tests of normality were performed on dependent variables and residuals
from the model (84).

The probabilities of Type I error for between-subject effects in the multivariate
model and also the R2, probabilities of Type I error, and regression coefficients at each
test day (univariate model analysis) were computed using the GLM procedure within
SAS (84). The mixed procedure within SAS was also used, but abandoned because the
computation exceeded the capability of desktop hardware (84). Proportion of total FPR
variation due to each variable was approximated by calculating (1)2 (85). Probabilities of
Type I error for within-subject effects in the multivariate model were computed using
Greenhouse-Geisser adjustments of univariate tests (84) and multivariate analysis of
variance. Tests for sphericity (transformed variates and orthogonal components) were

computed (84). Specific contrasts were by the Seheffé method (84).
RESULTS

The FPR at each of the first ten test days are shown in Table 2. The median
number of lactations per herd was 53 (range = 21 to 188). Frequencies of lactations in
first, second, and third or greater parities were 1672, 1185, and 1502 respectively.
MCdian parity number was four (range = three to eleven) in the parity three or greater

91388. Frequencies of lactations with season of calving as summer, fall, winter, and spring

35

were 1222, 1233, 928, and 976, respectively. Means of peak milk production in the entire
data set, for parity classes 1, 2, and 3, and summer, fall, winter, and spring calving cows
were 44.6 kg, 38.6 kg, 47.3 kg, 49.3 kg, 43.6 kg, 45.0 kg, 46.3 kg, and 43.9 kg,
respectively. Standard deviations were 9.2 kg, 6.8 kg, 8.3 kg, 8.6 kg, 9.4 kg, 9.2 kg, 9.4
kg, and 8.8 kg, respectively. The proportion of first, second, and third or greater parity
cows that had peak milk production on test days one to four was 47%, 81%, and 83%,
respectively. The median DIM at each test day is described in Table 2.

The repeated measures, multivariable regression model includes the fixed effects
of test day number, herd, parity, season of calving, and the parity x season of calving
interaction, and also the linear and quadratic components of the peak milk production
effect. Parity is classified categorically at three levels (1, 2 or 3), as first, second, and
third or greater lactation, although regression coefficients for all other main effect
variables did not change appreciably when parity was classified categorically at eleven
levels or when parity was considered as a continuous variable. Peak milk production is a
continuous variable in the model, although all other regression coefficients did not
change appreciably when peak milk production was classified categorically.

Shapiro-Wilk tests indicated that FPR was normally distributed at all test days (W
= 0.83 to 0.99). Residuals from the model were also normally distributed (W = 0.80 to
0.99).

The effect of herd (P < 0.0001), parity (P < 0.0001), season of calving (P =
0.0091), the linear (P < 0.0001) and quadratic terms (P < 0.0001) for peak milk
Production, and the interaction term for parity x season of calving (P = 0.0264) were all

Significant as between-subject effects in the multivariate model.

36

Tests for sphericity (transformed variates and orthogonal components) were
significant (P < 0.0001) so the within-subject effects were determined from multivariate
analysis of variance. However, this yielded similar probabilities of Type I errors to using
Greenhouse-Geisser adjustments of univariate tests. The test day number (P < 0.0001),
test day number x herd (P < 0.0001), test day number x parity (P < 0.0001), test day
number x season of calving (P < 0.0001), and the linear (P < 0.0001) and quadratic (P <
0.0001) peak milk production x test day number interaction terms were significant. The
interaction term for test day number x parity x season of calving was not significant (P =
0.1736).

All univariate models were significant (P < 0.0001 ). Coefficients of determination
were highest at the second test day (R2 = 0.29) and lowest at test day ten (R2 = 0.14). The
significant variables are described in Table 3. The proportion of total FPR variation due
to herd factors was 44.0% in the multivariable model, and varied from 1 1.3% in the non-
repeated measures model at test day ten to 25.7% at test day two. The proportion of
variation due to parity, season of calving, and peak milk production did not exceed 1% at
any test day. Regression coefficients are described in Table 4.

The proportion of test day FPR data with a value less than one (commonly
referred to as "inversions") is illustrated in Figure 1. Individual herd mean and mean of
herd mean FPR are plotted against test day number in Figure 2.

Parity effects are illustrated in Figure 3. First parity cows have a significantly
lower FPR (P < 0.05) compared to the other parities for the first six test days. Second
parity cows have significantly lower FPR (P < 0.05) compared to third or greater parity

cows at the first two test days.

37

Season of calving effects are illustrated in Figure 4. Fall and winter calving cows
had significantly higher FPR (P < 0.05) at the first test day compared to spring and
summer calving cows. Winter calving cows had significantly higher FPR (P < 0.05) than
spring, summer or fall calving cows at test days two and four to seven.

Peak milk production effects are illustrated in Figures 5 and 6. Cows with 254.4
kg peak milk production had significantly lower FPR (P < 0.05) compared to cows with
peak milk production of 36.3 kg to 45.3 kg and 45.4 kg to 54.3 kg for all test days except
the first and fifth (Figure 5). Cows with <36.3 kg peak milk production had significantly
lower FPR (P < 0.05) compared to all other peak milk production groups at the first test
day.

The parity, season of calving, and peak milk production variations in milk fat and

protein concentrations are illustrated in Figures 7 to 12.
DISCUSSION

The data used in this report are from herds that were recruited into a separate
study measuring the prevalence and severity of lameness in Michigan. The high response
rate (63%) to the letter requesting expression of interest in that study and random
selection of herds from respondents suggest that the effect of bias is likely to be minimal
and that the herds selected for the study are representative of the Mid-westem United
States. However, it is possible that recruitment was more successful among herds
experiencing lameness. This would affect inference of the results of this study, because of
the association between lameness and nutritional circumstances that influence milk

composition.

38

The significant change in FPR through the lactation (P < 0.0001) indicates that
interpretation of herd FPR requires consideration of the distribution of DIM among cows
within the herd. The prevalence of FPR values less than one (“inversions”) is lowest early
in lactation and plateaus after the fourth test day. It is possible that primary stage of
lactation effects (11. 76, 80), negative energy balance (1, 2), or other nutritional factors,
may be the cause of high FPR values in early lactation, through increased milk fat
concentration. Possible nutritional factors influencing FPR are lipid feeding (12, 25) or
high levels of forage in the diet (13, 25). Only the former is likely to be fed to cows in
early lactation, which suggests that either primary stage of lactation effects, energy
balance effects or dietary lipid are involved.

The individual herd FPR vs. test day plots, in Figure 2, being approximately
parallel, indicate that between-herd differences are approximately uniform over the
lactation. This indicates that there were few changes in within-herd factors such as
nutrition and housing over the time period during which the research was conducted.
However, a significant test day number x herd interaction (P < 0.0001) indicates that this
is not so. A reasonable conclusion from this is that within-herd factors, such as nutrition,
did change over the study period, but to a small degree.

Figure 2 also illustrates the significant difference among herds‘ mean FPR (P <
0.0001). The proportion of total variation of FPR due to herd factors is large (44.0%).

The most likely explanation for this is the among-herd variation in nutrition, because it is
an important determinant of milk composition and differs markedly among herds,
although other herd factors, which could be responsible, are genetics and housing. The

most likely nutritional factors are the proportions of the diet as concentrate, forage, and

39

lipid, and the characteristics of each of these ingredients. The authors of this report
suggest using the distribution of values in Figure 2 to determine whether a herd of
interest has extreme mean FPR.

The F PR increases with increasing parity number at the first six test days (Figure
3). This is because milk fat concentrations increase and milk protein concentrations
decrease with increasing parity (Figures 7 and 8). These changes in milk composition
suggest that a negative relationship between parity and energy balance may be the reason
parity three or greater cows have increased FPR compared to other parities. Negative
energy balance has previously been associated with increased milk fat concentration and
F PR, and decreased milk protein concentration (1, 2). Alternatively, the parity effect seen
here may be due to differences inherent to parity. A third possibility is that there are
confounding factors responsible for the effect, such as different diets being fed to
different parity cows. Regardless of the etiology of the effect, it is important to appreciate
that FPR differs by parity for much of the lactation, and so parity must be taken into
consideration in order to accurately interpret cows’ FPR data.

The FPR is significantly higher in early lactation in cows calving in fall and
winter compared to cows calving in spring and summer (Figure 4, P < 0.05). This effect
is due to increased milk fat concentration in early lactation (Figure 9). Milk protein
concentration in these cows also tends to be increased at this stage of lactation (Figure
10). The effects of season on milk fat and milk protein concentrations early in lactation
are consistent with those reported in the literature (10, 77, 82).

It is apparent from the reversal in relationship between warm and cold season

milk composition between early lactation test days and later lactation test days, that cows

40

calving in cool seasons experience warm season effects later in their lactation. This may
be the reason fall calving cows have decreased FPR relative to other seasons, later in
lactation —— in contrast to early lactation. However, this change in relationship does not
affect winter FPR, which remains significantly increased compared to other seasons at
test days two and four to seven (P < 0.05).

The magnitude ofthe difference in FPR among seasons of calving is small, except
at the first test day. although statistically significant at test days throughout the lactation
(P < 0.05). The regression coefficient for summer calving cows at test day one is
relatively large (r = -0.075), compared to other seasons and test days (r = -0.039 to 0.035,
Table 4). Therefore it is especially important to consider season of calving when
interpreting herd F PR data in the summer and when the mean DIM is low, suggesting a
high proportion of recently calved cows in the herd.

The mean FPR increases linearly with parity in cows calving in all seasons except
summer, where second parity cows have the lowest mean FPR. This explains the
significant parity x season of calving interaction (P = 0.0264).

The effect of peak milk production on FPR is quadratic (Figure 6, P < 0.0001).
This is also illustrated in Figure 5, where FPR increases for most of the later test days
with increasing peak milk production but the highest peak milk production category has

lowest F PR. This may be because increased peak milk production is associated with high
levels of concentrate or lipid feeding, which are known to influence milk composition
(12, 13, 25). Lipid feeding, in particular, would explain the simultaneous depression in

milk fat and protein concentrations (Figures 1 1 and 12) (12, 79).

41

The F PR and milk fat concentration are increased and milk protein concentration
decreased in cows with peak milk production 236.3 kg relative to <36.3 kg at the first test
day (Figures 5, 11, and 12). The FPR is also increased and milk protein decreased at the
second test day. The FPR and milk fat concentration are known to increase and milk
protein concentration decrease with negative energy balance (1, 2). Therefore FPR may
differ between peak milk production categories at the first and second test days because
of differing energy balances. The effect seems restricted to early lactation because cows
with the highest peak milk production have the lowest F PR at test days three to ten. This
seems plausible because energy requirements are known to exceed energy intake for the
first eight to ten weeks of lactation (86).

For these reasons, individual cows‘ peak milk production must be considered
when interpreting a herd’s FPR. It is safe to assume that some nutritional factors are
confounding the peak milk production effect seen here.

In summary, there is likely to be misinterpretation of a herd’s FPR data without
prior consideration of the test day number (or DIM), parity. season of calving, and peak
milk production of cows. In order to use these results in the field, the authors suggest
identifying F PR data in herds of interest by these variables, and comparing herd means to

those of typical herds, illustrated in Figures 2 to 5.
CONCLUSIONS

Milk F PR varies significantly among herds, likely because of differences in herd
nutrition. In addition, it is affected by test day number, parity, season of calving, and peak
milk production. The relationship between FPR and peak milk production is quadratic.

Therefore, assessment of a herd’s FPR distribution must take DIM, parity, season of

42

calving, and peak milk production of cows into consideration, in addition to the major
dietary ingredients fed in the herd that are known to affect milk composition. The
proportion of total FPR variation due to herd factors is large, but small for parity, season
of calving, and peak milk production effects. There are several suggestions that a

negative relationship between F PR and energy balance exists in early lactation.

43

CHAPTER TWO

THE USE OF REGRESSION COEFFICIENTS TO
COMPARE MILK FAT-PROTEIN CONCENTRATION

RATIOS AMONG DAIRY HERDS

W. Raphael, P. Bartlett, J. Dobrzelewski, and T. Herdt.

ABSTRACT

This report describes how regression coefficients, calculated from a statistical
model including known significant variability factors (test day number, parity, season of
calving, and peak milk production) affecting the ratio of milk fat concentration to protein
concentration (FPR) of dairy cows. are applied to actual FPR data, within common
spreadsheet software. By doing so, this application adjusts actual F PR for the effects of
these variability factors. DHIA test day mean of adjusted FPR can be used to assess risk
of negative energy balance, metabolic disease, and infertility in a herd. Differences in
adjusted mean FPR among client’s herds can be identified with the application and likely
exist because of differences in nutrition, particularly nutritional factors affecting milk
composition. Hence the application, which is available from the first author, will be

useful in problem solving and monitoring herd nutrition.

44

INTRODUCTION

The F PR has been reported to be associated with dairy cows’ nutritional status.
Strong negative correlations between energy balance and FPR (1, 2) suggest that FPR
may be useful to monitor nutritional status of cows in early lactation — a time when
negative energy balance is common (86).

Associations between FPR and displaced abomasa (5, 6) and F PR and ketosis (4)
suggest that F PR may also be useful as a predictor of the risk of these diseases in a herd.
Additionally, there is a negative relationship between FPR and fertility (4, 7, 8).

To illustrate the use of FPR for diagnosis of displaced abomasa, cows’ with first
test day FPR 2 1.39 were 8.2 more likely to be subsequently diagnosed with displaced
abomasa than cows’ with FPR < 1.39 (95% confidence interval 2.7 to 25.0, P < 0.05) (5).
This study design accounted for herd, parity, season, and DIM variation in F PR. Test day
refers to the day on which DHIA takes milk samples from cows and these are numbered
in the order in which they occur after calving, so are proportional to DIM. Sensitivity of a
FPR value of 21.39 in diagnosis of subsequent displaced abomasum was 80% and
specificity was 68% (5). Also, the odds ratio for subsequent ketosis, after first test day
FPR > 1.5, was 3.2 (P < 0.05), with consideration of herd and parity effects (4).

In summary, individual herd’s FPR may be a useful index by which to measure
energy balance, fertility, and the risk of metabolic disease in recently calved cows.
However, the herd mean FPR cannot be compared to other herds’ mean FPR without
prior adjustment for factors by which it is known to vary. These factors are test day
number, parity, season of calving, and peak milk production (Chapter One). The effect of

peak milk production is quadratic and the effect of parity varies by season of calving

45

(Chapter One). The difference among herds’ mean F PR after adjustment for these factors
is likely because of herd determinants of milk composition, the most important of which
is nutrition. Other determinants are genetics and environment.

Important nutritional factors affecting milk fat concentration are the proportion of
concentrates, unsaturated lipid or total lipid in the diet (12, 13, 25). Milk protein
concentration varies with level of energy intake (62, 63), dietary carbohydrate
characteristics (13, 74), and dietary lipid content (12, 79). There has been no research
directly examining the affect of these diets on FPR. However, it is safe to extrapolate
from milk composition research that FPR is commonly low within a herd when the
proportion of concentrates, unsaturated, or total lipid in the diet is great (because milk fat
concentration is often decreased). The mechanisms by which this occurs have not been
fully elucidated, but are associated with either increased dietary supply or rumen
production of trans-isomers of oleic acid (25). The later is likely to occur with rumen
acidosis (22). Predictions of how nutrition will affect FPR are not always clear, because
milk fat and protein concentrations are often affected simultaneously. For example, both
milk fat and protein concentrations can be decreased by dietary lipid, which if occurs to
the same extent, would not change FPR.

The objective of this report is to describe the creation of a spreadsheet that will
adjust individual cow FPR for known, significant variability factors, excluding nutritional
and other herd factors, and calculate basic statistics of this adjusted FPR data at each of

the first ten test days.

46

MATERIALS AND METHODS

Production data from 4359 Holstein cow lactations, in 68 Michigan herds, were
used to generate a repeated measures multivariable regression model. The purpose of the
model was to identify variability factors affecting FPR at the first ten test days after
calving, excluding nutrition-related variables. The model is described in Chapter One.

The herds used in the model were volunteers in another study that measured the
prevalence and severity of lameness in Michigan Holstein herds. The herds in that study
were selected randomly from respondents (11 = 388) to a descriptive letter sent to all
DHIA subscribing herd owners within a l20-km radius of Lansing, Michigan (n = 612),
but who were milking 50 to 300 cows.

The last lactation for cows calving between July I, 1997 and June 30, 1998, with
milk fat and protein concentration data for the first ten test days, was selected. Cows were
eliminated if the DIM at any of the first three test days were extreme for that test day (i.e.
less than 7 days for the first test day, less than the 2.5’h percentile for the second and third
test days, and greater than the 90th percentile for the first three test days). Cows were
excluded if milk composition was estimated, rather than measured, on any test day, or if
the sampling procedure was recorded as abnormal. Examples of when milk composition
is estimated are when samples freeze or sour in transit. Fourteen herds were excluded
because these criteria reduced the number of cows in each of them to less than 20.

Parity was classified categorically at three levels in the model (parity one, parity
two, and parity three or greater). Season of calving was classified categorically at four

levels; summer (June 21 to September 21), fall (September 22 to December 21), winter

47

(December 22 to March 19). and spring (March 20 to June 20). Peak milk production was
the greatest milk weight at any of the first ten DHIA test days.

The model demonstrated that FPR varied significantly among herds (P < 0.0001),
by test day number (P < 0.0001), by parity of cows (P < 0.0001), season of calving of
cows (P = 0.0091), and peak milk production (P < 0.0001). The effect of peak milk
production was quadratic. The effect of parity varied by season of calving (P = 0.0264).
Shapiro-Wilk tests indicated that FPR was normally distributed at all test days (W = 0.83
to 0.99) (84). Residuals from the model were also normally distributed (W = 0.80 to
0.99).

A second model was created, with the intent of calculating regression coefficients
for significant variability factors from the first model, excluding the herd effect. The
model specification was identical to the first model, but excluded the independent
variable herd.

The mathematical relationship that describes the use of these regression
coefficients to adjust measured FPR for the effect of parity, season of calving, and peak
milk production, is described in Equation 1. The coefficients A, B, C, D, and E refer to
the regression coefficients for the parity, season of calving, peak milk production (a), the
square of peak milk production, and the parity x season of calving effects, respectively, at
test day number i after calving (i = one to ten). Peak milk weights are in English units
(pounds) because that is what DHIA uses. Adjustment of measured FPR allows
comparison of cows or groups of cows when they differ in parity, season of calving, and

peak milk production.

48

Adjusted FPR, = Measured FPR, —- A, —- 13,—(ci x a.) — (D, x (011)?) — E,
(Equation 1)
A Microsoft° Excel2 spreadsheet application was created to automate the
calculation of adjusted F PR (Equation 1) for the most recent test day of cows in a herd of
interest, and also to calculate basic descriptive statistics of these values. This application
is designed to import cow data in comma-delimited, fixed length format, contained within
files named Drmscows.dl and Drmscows.d2. These files are created by PCDART3
software when report number 112 is run within that programme, and are stored in the
herd folder within the PCDART directory. However, the application is not specific to this
software and can be used with any comma-delimited, fixed length data providing that the
data are contained within two files, similarly named, and in the format of Table 5.
Numerical values are padded to the left with zeros (e.g. 00012) and text values are
padded to the right with spaces (e.g. “f__”). Each line of data is terminated by a semicolon
(ASCII character 59), carriage-return (ASCII character 13), linefeed (ASCII character 10)
(i.e. ;CRLF) combination. Lines are read one at a time. Each line in the file Drmscows.dl
is 1384 bytes long, plus 3 bytes for the ;CRLF combination. Therefore the total size of
this file should be evenly divisible by 1387. Each line in Drmscows.d2 is 62 bytes long,
plus 3 bytes for the ;CRLF combination. Therefore the total size of this file should be
evenly divisible by 65.
The application categorizes parity and season of calving in a manner similar to
that used in the identification of variability factors in the first model. Peak milk

production is the greatest test day milk weight for first parity cows with eight to ten test

 

: © Microsoft® Corporation, One Microsoft Way, Redmond, Washington 98052-6399
Dairy Records Management Systems, 313 Chapanoke Rd., Raleigh, NC 27695

49

days and is the greatest test day milk weight for second or greater parity cows with four
to ten test days. The means of these values, for parity x season of calving groups, are used
to estimate peak milk production in first parity cows with fewer than eight test days and
second or greater parity cows with fewer than four test days. Data from the first ten test
days from cows’ previous lactations are also used, if available, to calculate mean peak
milk production.

Data exclusion criteria are similar to those used in the identification of variability
factors in the first model, but in addition, cows are excluded if they have zero or missing
values for parity number or calving date for the lactation, or zero or missing values for
DIM, milk weight, milk fat or milk protein concentrations for any test day in the
lactation. Also, cows with an eleventh or greater test day are excluded. Cows excluded by
these criteria are not used in the calculation of mean peak milk production. If no data are
available to calculate the mean peak milk production for a parity x season of calving
group, cows which require it in Equation 1 are also excluded by the application. Once
adjusted FPR is calculated for all eligible cows, the frequency, mean, and standard

deviation of the mean of these values, at test days one to ten, are calculated.
RESULTS

The welcome worksheet for the application is illustrated in Figure 13. When the
user clicks on the welcome panel in this worksheet, a window will open, which requests
the directory path of the Drmscows.dl and Drmscows.d2 files. The application then
proceeds to calculate adjusted FPR for the most recent test day of appropriate cows.
Results are listed in a worksheet that is illustrated in Figure 14. Individual cows’ adjusted

FPR are listed, as well as the frequency, mean, and the standard deviation of adjusted

50

FPR at each test day. The mean is plotted against test day number (Figure 15). To
identify individual cows’ adjusted FPR values, cow numbers are placed in a worksheet
separate from results, but in a similar location within that worksheet to the location in the
results worksheet. Cows excluded, and reason for exclusion, are also listed in this
worksheet (Figure 16).

The Microsoft° Visual Basic4 programme, used to create the Microsoft°° Excel
spreadsheet application, is described in appendix three. An electronic copy of the
spreadsheet is available from the first author. The regression coefficients used within the
application (parity, season of calving, peak milk production, the square of peak milk

production, and the parity x season of calving interactions) are listed in Table 6.

DISCUSSION

This Microsoft® Excel spreadsheet application is a simple and accurate tool to
adjust a herd’s FPR for known, significant variability factors. excluding nutritional and
other herd factors.

The herds used to generate the regression coefficients were part of a study
measuring the prevalence and severity of lameness in Michigan. The high response rate
(63%) to the letter requesting expression of interest in that study and random selection of
herds from respondents suggest that the effect of bias is likely to be minimal and that the
herds selected for the study are representative of the Mid-western United States.
However, bias may be present, if, for example, herd owners experiencing lameness
problems were more willing to participate than those who were not. This is important

because of the associations between nutrition, milk composition, and hoof health.

 

4 © Microsoft° Corporation, One Microsoft Way, Redmond, Washington 98052-6399

51

The data used to generate the regression coefficients did not contain cows with
extreme DIM at the first three test days. The aim of this was to improve calculation of the
variability factors affecting FPR at a time when milk composition is relatively dynamic,
by excluding cows first enrolled in DHIA past the typical 30 DIM, and excluding cows
with unusually long inter-test day time intervals early in lactation.

Approximately 80% of first parity cows in the data used to identify the variability
factors affecting FPR had peak milk production prior to the ninth test day. A similar
proportion of second or greater parity cows had peak milk production prior to the fifth
test day. So, the application used actual peak milk production in lactations with eight to
ten test days for first parity cows and four to ten test days, for second or greater parity
cows. In shorter lactations, the application estimates peak milk production with the herd
mean value, for the appropriate parity x season of calving group. This prevents under-
estimation of peak test day milk production, in short lactations.

The regression coefficients in this application are calculated from a model
containing known significant variability factors affecting FPR, excluding herd. Herd was
excluded because regression coefficients for other variables are likely to less indicative of
the true effect of those variables when herd is included in the model. This is because of
the confounding effect of herd on these other effects. Additionally, there would be no use
for herd regression coefficients in a model used diagnostically at a herd level, unless the
herd of interest was one of the 68 herds used to generate the regression coefficients.

The spreadsheet excludes cows with test day eleven or greater data. This is to

ensure test day ten data, from the most recent DHIA test day, reflect cows currently at

52

that stage of lactation, and also, because there are only regression coefficients for the first
ten test days.

Calculation of mean F PR, adjusted for variability factors within a herd, is
important, because FPR may be useful in the diagnosis of negative energy balance,
metabolic disease, and infertility. For example, according to Geishauser et al. (1998), a
value of FPR 2 1.39 at test day one, adjusted for herd, season, parity, and DIM
differences, indicates 8.2 times the odds of subsequent displaced abomasa (P < 0.05) (5).

More importantly, this application will facilitate fair comparisons among clients’
herds, by adjusting for herd differences in DIM, parity, season of calving, and peak milk
production of cows. Remaining herd differences will largely reflect nutritional factors
that affect milk composition, so the application will be useful in herd nutrition trouble-
shooting and also as a tool to monitor changes made to herd nutritional programmes. In
large herds, groups of cows could be compared among each other and also within each
other, over time. Sub-normal mean FPR warrants investigation of the proportions of
concentrates and lipid in the diet, and may precipitate further diagnostic procedures to
confirm that a nutritional problem exists, such as ration analysis or rumenocentesis (87).
This type of diagnostic investigation, combined with intensive nutritional
experimentation examining dietary effects on FPR, will confirm nutritional risk factors
for different levels of FPR among a herd of cows. Caution is required with interpretation
of results from this application in small herds, where the frequency of test day data can be

low.

53

CONCLUSIONS

This Microsoft° Excel spreadsheet application will allow interpretation of F PR to assess
risk of negative energy balance, metabolic disease, and infertility in North American
dairy herds. It will also allow fair comparisons among herds’ mean FPR. The difference
among herds, after their milk composition data have been processed by this application,
is likely to reflect differences in herd nutrition. For this reason, it is envisaged that this

application will be useful in dairy nutrition problem solving and monitoring.

54

APPENDICES

55

APPENDIX ONE

Tables

56

2o< otbzm 2028322 68:23.5 «Z .> 2.52882: .2o< 200002 0:00:20. .8390 .8280 28 E2508...
:ooBHom 8:20.225 5,2 cotm 2 09C. :0 3253050 63305 3.02235 “oz 00 205082 82 02.2mm .2o< 2:285 228sz
.sosmbcoocou 22:05 880m 0 .22» 2:20 .22 2 288m 2:_>r. .coumbcoocou 5800A. 2:2M .0202: xwaocmm .0285 5302 ED.

 

 

_.0 A a 8.0 v a 8.0 v a _.0 A 2 <2 _.0 A .4 <2 <m2 at 82 .._e a 00003::
_.0 A .2 _.0 v a 8.0 v a __0 A a _.0 A a _.0 A .2 <2 N_<m2 $00 002
528232 was 52:22

_.0 A a _.0 A .2 <2 _.0 A a _.0 A a _.0 A a .252 00:8 st 000. .0... a 325
<2 0:0 1 2 E0 1 a 000.01 a 2.0.1. 2 _.0 1 2 <22 2.0 035 as 82 .00 a 0:352
<2 8.0 A 2 <2 8.0 v 0. <2 <2 <2 ea.» 5 ea
<2 8.0 A 2 <2 80 A 2 <2 <2 <2 as; = axm
<2 80 A .2 <2 8.0 v 2 <2 <2 <2 baa _ axm
€00 80. .20 a 2002

<2 8.0 A 2 <2 4. _00.0 v 2 <2 MW<2 222 an; @0202 Sam 0% 2002
0m; A22 022 M€222 ~_.2 ~~2Q macaw E2508... macaw .0550 ooSom

 

 

.cocmbcoocou 220:— EEom 20:0 22 2 2:2
1222 280i 2:2 .cocmtcoocou 2805 2:2 .285 mwcocmifififi 5:02 CD so 20< 2:2qu 20 muoobm— mo mtomom 2 230,—.

 

57

Table 2: Distribution of Milk Fat-Protein Concentration Ratio (FPRland Median Days

 

in Milk (DIM) at Test DayI Numbers One to Ten After Calving (n = 4359).

 

 

 

DIM FPR

Test Da . 25th . 75th

Numbei] Medran Mean SD] Percentile Median Percentile
1 20 1.37 0.33 1.15 1.32 1.54
2 51 1.29 0.32 1.08 1.25 1.46
3 82 1.23 0.31 1.03 1.21 1.39
4 113 1.20 0.29 1.03 1.19 1.37
5 144 1.18 0.28 1.00 1.17 1.33
6 175 1.17 0.27 1.00 1.15 1.31
7 205 1.16 0.26 1.00 1.14 1.29
8 237 1.15 0.24 1.00 1.14 1.27
9 268 1.14 0.23 1.00 1.13 1.26
10 299 1.13 0.22 1.00 1.13 1.25

 

‘Standard deviation

58

Table 3: Probabilities of Type I Error in Univariate Regression Models Examining the
Variability Factors of Milk Fat-Protein Concentration Ratio at Test Day Numbers One
to Ten (11 = 4359).

 

 

 

 

 

Variable
Test Da . Season of Season of
Numbery Herd Panty Calving Peak! Peak x Peak Calving x Parity
1 *** *** *** *** *** NS
2 *** *** * *** *** NS
3 *** ** NS *** *** NS
4 *** * * *** *** NS
5 *** *** NS NS NS NS
6 *** *** ** NS * **
7 *** NS * 1’ ’t *
8 *** NS NS ’r * **
9 *** NS NS NS NS NS
10 *** * NS NS NS NS
lPeak Milk Production
NSNot Significant
1P < 0.10
*P < 0.05
**P < 0.01
***P < 0.001

59

 

 

 

 

 

 

 

 

 

 

ooooooooood ooooooooood ooooooooood cocoooooood ooooooooood NM a 20k
oooooocoood ooooooooood ooooooooood ooooccooood ooocooooood N a 20k
ooocooooood ooooooooood ooooooooood ooooooooood ooooooooood ~ a 20k
ooooooooood ooooooooood ooooooooood ocooooooood ooooooooood MIA. a $552
ovomm ;mmod- oomnmcg 3.: mgmncomvod- mNmNoonNNod- Boom—:39? N a 20552
33335.: $3385.: $232 5.: wNNSVVmwood 333 3.8.: N abEEzw
ooooooooood ooocooooood ooocooooood ooooooooood ooooooooood m 1A. a MSEW
ocooofmmod- oomowomflod- wmoomfl No.0- wmonoommcod- .mmhcmmwmod- N a 3:22
cm. mnwotod omcommnoood- 2320025.: 22828.? owvomooovod- N x Nahum
ooooooooood ooooooooood ooooooooood ooooooooood ooooooooood mm a 222.5:
33838.? omNmNVmoSd cam—0033.0- mNmoNSoood- Somme—Sod- N Remake:
omnmcmwmood ownwmowoood- 2.330036- moowmvowood- 83333:. N M2233

2%:

x w:_>_mu .6 58%

80300000- 38020000. 22020000- 008020000- 000820000- N20: 28%
Swmwmmoood ononmmvfiod $33356 33:135.: Sham—03.: 2:2 28:
ooooooooood ooooooooood ooooooooood oooocoooood ooooooooood NM
owooogvood- momm _ $0.86- ow _ 32.88.: Nemommommod- n82: Smod- N
ooommovccod- mvwmovovmod- 33388:. SE Svmwod- momnvnwmmod- \

0:...
ooooooooood ooooooooood ooooooooood ooooooooood ooooooooood =01:
ommowmmm _ o.o mmono _ mm _ o.o woe—RN08: 233mm _ od- vvmwcnomnod- 2253a.
ownocmvwood Nonmvmovmcd w _ 08mm _ Nod 30 _ cam _ Nod mmwmm _ vo _ ed- 3:me
91me 5.8.: mommvcmomcd momm _ Rammed Nmmwnwnmmcd wowofl memo: 220$:

”2:200 (.0 :83m
$08033.— » _ w _ moom _ 5o ooowwmmaood 2394330.: moooowoomwd 228:2:—
m v m N _
.80 :32.

 

89001..

 

 

.33: .1. 5 .2252 >00 0812. 31.2002 :ocm::oo:ou 2205-812 2:2 yo

 

2:223 2:285 .2002 :282 02 22:03:22 35032 233 oz :8: 9:22:80 :28“: oz 20:: 5082:. HV 2:31

60

 

 

 

 

 

 

 

 

 

 

ooooooooood ooooooooood ooooooooood ooooooooood ooocooooood mm a Net
ooooooooood ooooooooood ooooooooocd ooooooooood ooooooooood N a NNenN
ooooooooood ooooooooood ooooooooood ooooooooood ocooooooood N x NNeuN
ooooooooood ooooooooood oooooooocod ooooooooood ooooooooood NM a 025.3%
3230086 oownmmofiod mowmwSN—od ommmvmniod- wane—$856- N 0008530
ego—Nowmod oENv—mmocd 823—386 322 NS: Sommvvpod N 022532
ooooooooood oooooooocod ooooooooood ooooooooood ooooooooood NM x MENNNW
233356- ovwocomomod Nonooooflod- wawmmwwood Madam—336- N RMSNNM.
83338.: goon—wowed ooNNNbivod mmmovtomod mmoNOmode N 025%.
ooooooooood ooooooooood ooooooooood ooooooooood ooooooooood mm a 222.2
2 ~18»:de oVNVNNwOmod wmcmoomnood- Gomwoowood- mmomNrbomNod- N 0022.2
033055.: 8205056 .0332 .86. Not}: 536. Ntvmmwoood- N 0me5.:

20:00

x w:_>_0U No 50%

532886 283825.: 8303890- 2030886. 323886. N2:2 200A:
0.302089? wovmommvood- Nccomcovood comm—38o: wmmvvwmvood 2:2 200:
ooooooooood ooooooooood ooooooooood ooooooooood ooooooooood Mm
one: _ v _ wood- ogmomoomod- of 33m _ od- cmvmvioood- m3 _ wnNroood N
92.3 _ o _ cod owcvwoommod- $312 :w _ od- momonooomod- ohm. mvmcnod- N

5:0:
ooooooooood ooooooooood ooooooooood ooooooooood ooooooooood NNerN
300mm _ 3 :6. ©3038 _ od- cemwnwommod- 2o _ Ntmvvoood $52 3086. 225:2.
9.ch— mmood 3323.86- 523.9: 56. cemom Swood- mvwmmbow 3.: M2120
o. Roomm—od- o_ Pogo—ed- avmmNNoMood Nmmmommmmod £9003: mod 20:5:

w:_>_0U mo 80%
033332 ._ 2.30332.— 2N0m<ooomod cow—mwmoowd wmmmoconod 22082:—
A: a w 0 0
>00 08,—.

 

 

23:03 0 2:01:

61

Table 5: Data Format Required by the Microsoft?” Excel! Spreadsheet for
Automatic Calculation of Adiusted Milk Fat-Protein Concentration Ratios.

 

Filename File Item Byte Range Description

Drmscowsd l 2
1 1-5 Cow Number
188 1064-1065 Lactation Number

Drmscows.d23

 

1 1-5 Cow Number

2 7_14 Last Calving Date
(yyyymmdd format)

3 16-19 Test Day DIM4

4 21-25 Test Day Milk (lbs)

5 27-29 Test Day Fat °/o

6 31-33 Test Day Protein %

8 39-40 Test Day CAR Code5

 

l© Microsoftw Corporation, One Microsoft Way, Redmond, Washington 98052-
6399

2Data Sorted by Increasing Cow Number

3Data Sorted by Increasing Cow Number, Lactation Number, then Test Day
Number

4Days in Milk

5Indicates Milk Composition was Estimated or Sampling Procedure Abnormal if
CAR Code = A, AF, E, F, HF, I or L

62

 

 

 

 

 

 

 

 

 

 

 

ooooooooood ooooooooood ooooooooood ooooooooood ooooooooood mm a Nch
ooocooooood ooooooooood ooocooooood ooooooooood ooooooooood N a NNunN
ooooooooood ooooooooood ooooooooood oooooocoood ooooooooood N a NNunN
oooocoooood ooooooooood ooooooooood oooocoooood ooooooooood mm a $5554
wonvwmflvod- Swommmmood- ommmmomeod- «Summammod- ego—2.39? N Hamfifizm
anonmmmoood nvommomflod owvcmmmmood mNmnmvooood chomoommsd N Ramsszm.
ooooooooood ooooooooood ooooooooood ocooooooood ooooooooood Mm a MEBW
conmwmwmmod- Ego—owniod- gamma—god- ohmamtmood- omvmmovaod- N Rmztow
$3.26— _o.o- Noomongod- gown—336- cenflmmmmod- ovowmmovnod- N RMEKNW
ooooooooood ooooooooood ooooooooood oocoooooood ooooooooood mN x $35:
cvmmnomnmod- Nmmnovonood- oncomvnoood- 2N2 Nvmmod- ovnfomvmod- N RENEE
owmmwnoiod- Ngmfoomod- Exemvmoood- cmvovooomod- 2.033956- N RSNNS:

are

x $530 No 58%

omowmmooood- nmmmmvooood- coomwvooood- mnwvmmooood- oNovaooood- NOE—2 xmon:
gemwowvood Svmmotood mom—vowwood enumocooood onmmgnwood x52 xmom
ooooooooood ooooooooood oooocoooood ooooooooood ooooooooood mm
owe—0.386 mcmvooimcd- Nomwmooflod _m_vw$o_o.o- choowmnmmod- N
3363 _ mod- ooconwmmod- N. _ moommnmod- Nw _ who _ wwod- 23328.0- N

32$
ooooooooood ooooooooood coooooooocd ooooooooood ooooooooood NNanN
VMSmNNoNod .mwmo _ owmcd anmmmmmvod whoommovood- commemomood- 82:35
3839.56 33 _ mvonmod nvomccmnmod Somoomomod ocnmwwww _ od- 3me
momoovmmmod omogoommod mmwnmmgmod movoomwgod oooomoowmod EENE

@330 No 53%
m v m N _
$0 $3.

 

 

.83, u 5 25. Sceeaocoo 5395-5 55

 

No 222E 3:732; EoI-:E:3 No 88km 2: 22330 8 3880 38.2 :23»:qu 8253232 a 80+. £56580 6 2me

63

 

 

 

 

 

 

 

 

 

 

oooooooood oooooooood ooooooooood ooooooooood ooooooooood NM x NNan
oooooooood ocoooooood ooooooooood ooooooooood ooooooooood N a NNurN
oooocooood oooooooood ooooooooood ooooooooood ooooooooood N x NNanN
oooooooood oooooooood ooooooooood ooooooooood ooooooooood mN a mean
Sonownood- w. Nomcvood Nwwomwooood mmvomimmod- Saws—Nod- N xxwfiszm
emmcvmiod Sam—coed 23.3286 330.3206 593385 N meEsz
oooooooood oooooooood ooooooooood ooooooooood ooooooooood mm a mztmm
omonflmmod- woKooEod moommoS—od- wvafwomood mecca—$36- N RMENNW
mmOmmmvood oSocmNNod oomowmogod 2.02 Somod mute: $5.0- N 232me
oooooooood oooooooood ooooooooood ooooooooood ooooooooood mm a 532%:
333256 303.836 $23336- 3038—86- mucowmvomod- N kgmNEN:
mnwoooSod $03355 vomommbmmod- Own—mmvnmod- omwnwoommod- N H§E§

beam

x 3360 No 833

wvmflooood- wvvflooood- vaomoocod- mmncvgoood- mow—32:56- NOE—2 xmmflv
onvwoflood Sommomood mnovmmaood maowonmcood 2: 38306 x52 xmom
oooooooood oooooooood ooooooooood ooooooooood ooooooooood mm
mNmm _ Vmcod- nmvntvmod- o _ outmmood- moo—owNoNbod mnemotm _ od N
830386 wmoomtmod- mmemmVNSd- NNONOmmNSd- $33.33.? N

as».
oooooooood oooooocood oooooocoood ooooooooood ooooooooood NNouN
wmvmnoaood- E 3830.? Nme $8.86. E «58% N o.o NM: mvvwomod- LmEEzm
o _ emoOONod Nnmmm _ w _ o.o- ;vooocoood- hm» _ v3 Sod- Soowv _ oNod MEEW
mmmwmmmood- $33.86- $033206 Emmooovmod 308N886 hi3:

wEimU No 530m
2 o w h 0
b5 “3%

 

 

dicooq 9 Bomb

64

APPENDIX TWO

Figures

65

:58: Z Nam amok
o m N c m

 

 

 

S

m—

cm

 

 

 

.Smmv H 5 828:2

 

Ham “mob .m> @8652: o._ v mega coumbcoocoo EBBméE x22 35 “mot—L No 53.59on “N Esflm

mm

luaomd

66

68) and Mean of Herd Mean FPR

 

Figure 2: Herd Mean Milk F at-Protein Concentration Ratio (FPR) (n

(O) vs. Test Day Number.

 

 

X75:
‘ I’M' l i
“CW4 / «i
M iii/it
/ i “(at i /
" / / I / I

+ ‘i «it
/ ‘ /
M
7»

 

M
H 3
NW
(ta/9177
1%):le ‘/;’//
/”/:I/’;/4/

i

\

 

 

 

 

 

 

'— —I v—I4 '— v— v—fl '— v—Q

oneu uouenuaouog utaton-m :1 mm

67

moi—1F‘0‘w.

10

Test Day Number

 

 

 

 

 

 

mA :oNNfiomN .m> N coNNSomN H AN .855th N 8:88. .m> N 56802 n a . mA .5588. H 4 N 83802 .I. o N

 

 

 

.God v a: oocoNotNu mm :ozflmfl .m> N :oNNSomN H o 60:20th

 

 

 

 

53802 H E $53580 53:22 we 5 52:52 GD 33. .m> 23% 5:828:00 :NoNoNnNamflNNNNZ ”m oSwNuN

55:52 b5 NmorNr
N: n c m v N N
a _ N
can nan saw a 0.3m 04m
[itti t t r i i t \ N.N
I/UIO,’ I I
: /p
0 I
i i l r i ‘ iii ti .o i i I N.N
I
- I F. I r- 2
I
O
i i i t i i i i i l: vN
mN

open uoumiuaouog U!9101d-19;I )ll!W

68

 

 

 

33E: Z N80 30%
N: o. m N. o w v m N N
0 ob o.“ 06 o a Redd

 

N.N

, N.N

m.N

vN

 

m.N

 

 

.Nmod v 5 8:20th
=th .m> 888% u o 60:08th :fl .m> macaw n v 60:20th :fl .m> .853 u o docoNotNNo NoEEzm .m>

52.3» H n HoocobNNNNv mctmm .m> SENS u a .NwEZwo :fl u o .wEZmo .NoEEsm H D .chZmocham H I «ma—3N8
N853 n 3 mNzo: NCNmD :meoN—z me E 82:32 Mao amok .m> oNNmyN coNNmbcoocoU EBONANNNHN «:2 ”v uNawE

 

 

 

open uouenuoouog ugaton-wfl myw

69

 

 

 

 

 

 

59:52 b5 33.
ON 0 m N o m w m N N
macaonn “no “no “nuns Mounvnp “up “song “Nona?“ mnoogaw hnfinonnnm
./ it tiltilitlilii , it
n /.l/.“ |||T
0‘ il.’ in. Mu o t

H
0

r N.N

N.N

 

 

 

I N.N

v2

 

 

 

 

 

mN

 

 

Ago v a Savage v
go n .N docoNobNNo NV .m> N bomflmo H o .855th m .m> N Domflmo H No 60:98wa v .m> N bomflfi o
doocobbfi m .m> N @0330 H n 60:80th N .m> N @0330 H 33x NiamN n 0 Am be;
claw. v.8 u a rNN bog? n? owwximam u uquNIEIOwanUVN 03v u 8 8:885 Va: 85%; N85
an NuoENmbm ANNE: EN§mmNNNoN§ we E 32:32 Nam 80% .m> 23% :oNNmibcoocoU :Nouoififlflamfl

 

 

 

 

 

 

 

 

omen uouanuaouog uremid-md mm

70

co

 

Na: 88395 5:2 8:903 Ned
ow ON co ow ow cm om ON

N L N N

 

 

 

ad

36

L 2:

.T N.N

mNN

‘ N.N

mNN

 

MN

 

 

.AOGW.MVH~C Adv: COCOS—VONQ x12 COCNHUNJ v—mok .m> 05mm COthCOUGOU azouOhn—uumn— x12 “0 UNBWWL

open uopenuaouog utmoxd-te :1 WW

71

oN

 

89:32 3Q 50%

c

m

 

 

 

om

airmm

3

cm

00
(’1

O
V

N
v

<1.
v

 

 

 

 

0v

(Sn/3) uopenuaouog 1e :1 mm

72

 

 

 

 

Nongz N30 NmoNr
ON a , w n o w v m N N
N _ N N r N N e N 0N
\ t t i i i t l i i t i r i i i i i i t i r, NlN
‘ i r i i i i i i t t i i i i 1 MN

 

 

mm

 

 

 

.NmN :oNNmNomN n 4 “N 53803 n o
4N «55802 u 9 £53 CED :wuwNaNoNNN/N mo 5 NBSZAMQ Nmoh .m> :onmbcoocoU £8on VNHNNSN ”m Emmi

(Sn/8)uope11uaouo;) upwxd )mw

73

oN

 

595.2 b5 Nmok

o m N w m v m N

 

 

 

 

 

 

.3538 NNE H a $528 5883 u U 42:33 datum H I 43318 SE? H 8

 

mNzoI tan 53:32 we 5 82:52 5Q NmorNt .m> zoNNmibNNoocoU NmnN VNNNE ”m oSMNnN

Om

Nm

.. : 3m

0m

, mm

Tow

Nv

3

cv

(fin/8) uopenuaouog 1e :1 XIFW

74

oN

 

59:52 Nam 58.
m N o m NV m N

 

 

 

 

 

.Addm>_mo =mm H O 4W~£>~NU Noegm I D .Mflm>_mo “Gin—m U I “MGMNr—mo HQHGMNS H 8

 

mNBoNN EEO :mwNNoNZ me E 8:832 Nam 30% .m> :oNNmb:oo:oU :NBSAN wNNNSN 5N oBuNnN

0N

NN

wN

0N

Om

Nm

Nm

mm

Nam

mm

mm

(8)1/8) uopenuaouog uraioid mm

75

oN

 

5:852 N89 Nmo‘Nt
m m N o w v m N N

 

 

 

 

 

 

 

 

3 3me n . .5 2% 2mm «.3 u Did]: a? 2 a. SM n u a: no? u 0N 8:885 :3 8:303

 

IdeoaN >9 nocNNfiNm £202 :5de :wwEoNNN/N we :N 59:52 :5 NmoNt .m> :oNNwMNSogU NmflNNNE “N N oSuNnN

om

Nm

iv».

,3.

.mm

ow

NV

3

ow

(Sn/8) uopenuaouog 1e 5 mm

76

oN

5:622 Nam 58.
o m N c w v m N N

 

 

 

 

d3

 

v VmAH o .qu m Vm 8 3N v. mvn D aux m. Q 8 qu m. omI I I .MN m. omVI I 0N :oNNoswon VNNNN>N :oJmNomIN xmonN

 

0N

NN

IwN

oN

om

Nm

mm

3”

mm

om

(fin/3) uopepuaouog urotoid )"rw

77

Fi ure 13: Welcome Worksheet in Microsoft® ExcelI S readsheet A lication for

Automatic Calculation of Ad'usted Milk Fat-Protein Concentration Ratios.

E] Elle Edlt {new Insert Format Iools Data ‘fllndow Help 4511]:
B C 0 E F R H , V I J K '

 

 

ft

Click this button to process
files DRMSCOWSDl and
DRMSCOWSDZ.

atnmwmmnimm

ueens un :1 8506 _

NNNNNgNN
unworthy-Lou
.

30 .N
31'; ,I
H 4 v N\W‘ELCONE 1. Coefficients [ PESL‘LTS CHART 1 RESLLT'S DATA [Peaks]! 1 l vlf—

 

l© Microsofi® Corporation, One Microsoft Way, Redmond, Washington 98052-6399

78

Figure 14: Results Worksheet in Microsoft® Excel1 Spreadsheet Application for
Automatic Calculation of Adiusted Milk F at-Protein Concentration Ratios.

 

 

 

 

 

 

 

 

 

 

'1'] Elle Edit mew Insert Fgrmat Iools Data \flmdow Help $5111.
- A .B .4: p, 70 E. .. Fr: 6 H , I J 1411-:

1 ‘ Fat/Protein ~—

2 ‘ Test1 Test? T8913 Testd TestS TestE Tesl7 TestB Test9 Test 10

3; 1336308 0901022 0848999 0919154 1201194 0880352 1351252 0609144 1449977 1096067

4 f‘ 090412 0907616 0742102 0739693 0953817 0878614 1017475 1289591

5? 1812722 1011286 1026718 1061:1324 0.754147 1.209021 1.269154

6 0921917 090583 1197658 0880635 1050318 1.140184 1.621442 ,

7 0839604 089333 0802897 09138301 0.89454 1.11931 k",

8 0.859412 0972641 0790375 1239491 1220309 g

9 ; 1221813 1101972 1064336 0868107 . m

10 0.969585 1036414 11132485 1142978 ' Q

11 ; 1964237 0928649 1 170619 3 F

123 1722978 1053286 ‘ Q 1

133 1396841 ' g “

14; 1762115 ' 3

15. 1304972 "

161 1413966

17; 10905

18; 1026366

l9; 0969522

20; 1989434

215 1483167

22,;Coun1 19 6 5 1 4 1 2 9 8 10

23 Mean 1315188 0931954 0923675 0919154 0970586 0880952 1.152534 0901993 1.143439 1.185086

2461867 0385452 0048316 0186291 #DMO' 0202258 110M131 0281029 0160387 0165485 0194855

‘15

3‘5;

2 .

285

29;, N.

ms.

14 g1) va;\ WELconE [6612111661115 _,( RESULTS CHART \Resuuspara/ 12.383114] . ., .. _| )lf— _

1© Microsoft® Corporation, One Microsoft Way, Redmond, Washington 98052-6399

79

 

 

Figure 15: Results Chart in Microsoft® Excell Spreadsheet Application for Automatic
Calculation of Adjusted Milk Fat-Protein Concentration Ratios. This Chart Plots Herd

Mean Milk Fat-Protein Concentration Ratio. Adjusted for Paritv, Season of Calving, and
Peak Milk Production vs. Test Day Number.

 

 

 

1] E119 Edit Mew insert Fgrmat Iools Qhart \mndow fielp 4513;]?
‘1
Herd Mean Milk Fol-Protein Concentration Ram, Adjusted for Parity, Season of Calving mi Peak Milk
Producmnvs Test DayNumber
14 a. .
; IQ
' 8
l2 : (I)
1 2‘
8 : =
a: 1 ————— v 9
9 <0
g ‘2? <1)
2 E :1
g E 0.8 12.“.
3:8 5
11%.? 0.6 1
8.: 1
< I?"
g 04 1
E 1
02 “
0
1 2 3 4 5 1s 7 s 9 10
Te" Day “vegan Area]
7‘}
14 1:» .1 \7 WELCOME ( Coefficrants \RESULTS CHART/ 111593130414 [Peaksh I, p p I y]—

 

l© Microsoft® Corporation, One Microsoft Way, Redmond, Washington 98052-6399

80

Figure 16: Cow Identity Worksheet, Describing Cows Included, Cows Excluded, and
Reasons For Exclusion, in Microsoft® Excel1 Spreadsheet Application for Automatic
Calculation of Adjusted Milk Fat-Protein Concentration Ratios.

11‘] E118 Edit 118w insert Format 10015 Data Window Help

 

 

 

 

 

 

 
  

 

 

    

1 A 5..-: We“; 8 r. E i F 1 G. ‘; 11 1. .1
1 Cow ID Number:
V;_]Excluded Test 1 __ Test 2 Test 3 Test 4 Test 5 _ Test 8 Test 7 Test 8
2 4%, catGiida=E " 69 1 97 43']
. _. 45:11:91.1. '111'm:=6 . 395 236 402 779
tests coum=11>10 419 417 418 785
'_ “,8 tests. count=16>10 422 512 426 796
__ 94155;.1816858. -_ 428 803 826
_ 1., 2319819 comit=‘13>10 433 851
_ 3118311511 dim=3 439
006218515 Count=11>10 443
p 1 ‘135.tests count=14>10 445 1277
2_ ’294 test 1 dim=64 448
_ 3:;37818815 count=11>10 783
711119181! peak-=0 3 839
_ . j_'ode_='A ~ 844
- “:1. 8111115 stefipeak=0 920
,_ awedjusted peak=0 ' 955
1% adjust“ peak=1J » 1398
45318519 count=14>10 1442
j .‘403 adjusted peak=0 1717

_2 ,fitests count=15>.10. 3040
_: 14161881 1 dim—4.

' 1452 tests. count=11>1fl
.5 545318815. count=16>10
.I_ 1485 tests ebum=12>10
' :48? Oncie ‘ ‘~‘ ‘I' 10
p; 7 ‘520tests. c_uur1_t=.11>10
“13620191511 dim=4

'2 l

 

 

 

   

 

 

39339181861164“) ‘1]
M. 1 2 21‘! RESULTS CHART I.RE$LLTSDAT.A [PeasxRESULTs cows.[ 0191M] 22.1? 2....

 

'© Microsoft® Corporation, One Microsoft Way, Redmond, Washington 98052-6399

81

APPENDIX THREE

Microsoft® Visual Basic1 Programme Used to Create the
Microsoft® Excel1 Spreadsheet Application

 

l© Microsoft® Corporation, One Microsoft Way, Redmond, Washington 98052-6399

82

 

Attribute VB_Name = "modMilk"
'milk module

'2001 by Will Raphael and Joel Dobrzelewski
'Michigan State University
'requires additional classes: clsCow, clsHerd, clsMilkTest

Option Explicit
Global Const MaximumNumberOtTests = 10

Public Sub GenerateResultsO

Dim h As clsHerd
Dim p As String

p = InputBox("Path", "Data Path", ActiveWorkbook.path)
If p = "" Then Exit Sub

 

Set h = New clsHerd
With ActiveWorkbook
h.CalculateResults p, .Sheets("Results Data"), .Sheets("Results Cows"),
.Charts("Results Chart")
.Charts("Results Chart").Activate
End With
Set h = Nothing

End Sub

Public Function GetCoefficient(ByVal coefficient As Integer, ByVal row As Integer,
ByVal column As Integer) As Double

Const leftcol = 2
Dim toprow As Long
Dim sheetrow As Long

Dim sheetcol As Long

Select Case coefficient

Case 1
toprow = 2

Case 2
toprow = 6

83

 

Case 3

toprow = 9
Case 4

toprow = 10
Case 5

toprow = 1 1
Case Else

'trying to retreive an invalid coefficient
'look at the calling routine
Stop

End Select

sheetrow = toprow + row — l
sheetcol = leftcol + column - l

GetCoefficient = Sheets("Coefficients").Cells(sheetrow, sheetcol)
End Function
Public Sub HighlightRegion(rangel As range, color] As Long)

range] .Borders(xlDiagonalDown).LineStyle = xlNone
rangel .Borders(xlDiagonalUp).LineStyle = xlNone
With rangel .Borders(xlEdgeLeft)

.LineStyle = xlContinuous

.Weight = leedium

.colorlndex = xlAutomatic
End With
With rangel .Borders(xlEdgeTop)

.LineStyle = xlContinuous

.Weight = leedium

.colorlndex = xlAutomatic
End With
With rangel .Borders(xlEdgeBottom)

.LineStyle = xlContinuous

.Weight = leedium

.colorlndex = xlAutomatic
End With
With rangel .Borders(xlEdgeRight)

.LineStyle = xlContinuous

.Weight = leedium

.colorlndex = xlAutomatic
End With
rangel .Borders(xllnsideVertical).LineStyle = xlNone
rangel .Borders(xlInsideHorizontal).LineStyle = xlNone
With range1.Interior

84

 

.Color = colorl
.Pattern = xlSolid
End With
End Sub
Public Function MilkCDate(s As String) As Date
MilkCDate = CDate(Mid(s, 5, 2) & "/" & Mid(s, 7, 2) & "/" & Mid(s, 1, 4))

End Function

 

VERSION 1.0 CLASS
BEGIN

MultiUse = -1 'True
END
Attribute VB_Name = "clsCow"
Attribute VB_GlobalNameSpace = False
Attribute VB_Creatable = False
Attribute VB_Predeclaredld = False
Attribute VB_Exposed = False
'cow class
'2001 by Will Raphael and Joel Dobrzelewski
'Michigan State University

Option Explicit

Public reason As String

Public id As Long

Public lactationNumber As Long
Public lactations As Collection
Public herd As clsHerd

Public Sub addTest(test As clsMilkTest)
Dim I As clsLactation

'is this a new calving date?
If test.calvingDate <> lastCalvingDate Then
'yes this is a new calving date and a new series of tests
Set I = New clsLactation
l.calvingDate = test.calvingDate
Set l.cow = Me
lactations.Add l
EndIf

85

 

lastLactation.addTest test
End Sub
Public Function adj usteder() As Double

Dim c1 As Double
Dim c2 As Double
Dim c3 As Double
Dim c4 As Double
Dim c5 As Double

Dim fpr As Double
Dim Peak As Double
Dim t As Long

1 = getTestNumber
Ift = 0 Then Exit Function

cl = GetCoefficient(l , season, 1)

c2 = GetCoefficient(2, parity, 1)

c3 = GetCoefficient(3, 1, t)

c4 = GetCoefficient(4, l, t)

c5 = GetCoefficient(5, SeasonParity, t)

fpr = realer()

Peak = adjustedPeak()

'peak and fpr should never be zero

'if they are, then these cows should be excluded from herd data
'and you should not call this function

If Peak = 0 Then Stop
If fpr = 0 Then Stop

adjusteder = (fpr - cl) - (c2) - (Peak * c3) - (Peak * Peak * c4) - c5
End Function
Private Function adjustedPeak() As Double

Dim p As Double

If lastLactation.goodLength Then

adjustedPeak = lastLactation.Peak
Else

86

p = herd.Peak(parity, season)
adjustedPeak = p
End If
End Function
Public Function hasExcludedLactations() As Boolean
Dim lactation As clsLactation
For Each lactation In lactations
If lactationexcluded Then GoTo YesExcluded
Next

hasExcludedLactations = False
Exit Function

YesExcluded:
hasEchudedLactations = True
End Function
Public Function isExcludedFromHerdO
If lactationscount = 0 Then reason = "no lactations": GoTo YesExcluded
If (adjustedPeak = 0) Then reason = "adjusted peak=O": GoTo YesExcluded

If lastLactation.excluded Then reason = lastLactation.reason: GoTo YesExcluded

isExcludedFromHerd = False
Exit Function

YesExcluded:
isExcludedFromHerd = True

End Function

Public Function getTestNumber() As Long
getTestNumber = lastTest.number

End Function

Private Function lastCalvingDate() As Date

87

If lactationscount > 0 Then lastCalvingDate = lastLactation.calvingDate
End Function
Public Function lastLactation() As clsl.actation
Set lastLactation = lactations(lactationscount)
End Function
Public Function lastTest() As clsMilkTest
Set lastTest = lastLactation.lastTest
End Function
Public Function parity() As Long
parity = lastLactation.parity
End Function
Private Function realPeakO As Double
realPeak = lastLactation.Peak
End Function
Public Function realer() As Double
realer = lastTest.fpr
End Function
Public Sub RemoveNullTests()

Dim lactation As clsLactation
Dim i As Long

For Each lactation In lactations
lactation.RemoveNullTest
Next

i=1

Do While i <= lactationscount
If lactations(i).tests.count = 0 Then

88

lactations.Remove i
Else

i=i+1
End If
Loop
End Sub
Public Function season() As Long
season = lastLactation.season
End Function
Public Function SeasonParity() As Long
SeasonParity = (season - 1) * 3 + parity
End Function
Private Sub Class_lnitialize()
Set lactations = New Collection
End Sub
Private Sub Class_Terminate()

Set lactations = Nothing

End Sub

 

VERSION 1.0 CLASS
BEGIN

MultiUse = -1 'True
END
Attribute VB_Name = "clsHerd"
Attribute VB_GlobalNameSpace = False
Attribute VB_Creatable = False
Attribute VB_PredeclaredId = False
Attribute VB_Exposed = False
'herd class
'2001 by Will Raphael and Joel Dobrzelewski
'Michigan State University

89

Option Explicit

Const seasons = 4
Const parities = 3

Private cows As Collection
Private peakMilk(l To parities, I To seasons) As Double
Private peaksCalculated As Boolean

Private Sub CalculatePeaks()

Dim cow As clsCow

Dim lactation As clsLactation
Dim p As Long

Dim s As Long

Dim peaktotal(l To parities, I To seasons) As Double
Dim cowcount(l To parities, 1 To seasons) As Double

For Each cow In cows
If (cow.lactations.count > 0) And (Not cow.hasExcludedLactations) Then
p = cow.parity
s = cow.season

For Each lactation In cow.lactations
If lactation. goodLength Then
peaktotal(p, s) = peaktotal(p, s) + lactationPeak
cowcount(p, s) = cowcount(p, s) + 1
End If
Next
Endlf
Next

For p = 1 To parities
For 5 = 1 To seasons
If cowcount(p, s) > 0 Then peakMilk(p, s) = peaktotal(p, s) / cowcount(p, s)
Sheets("Peaks").Cells(p + 2, s + l) = peakMilk(p, s)
Sheets("Peaks").Cells(p + parities + 5, s + 1) = cowcount(p, 3)
Next
Next

peaksCalculated = True

End Sub

9O

Public Sub CalculateResults(path As String, resultsheet As Worksheet, cowsheet As
Worksheet, resultchart As Chart)

Const tests = 10
Const header = 2

Dim cow As clsCow
Dim last(O To tests)
Dim t As Long

Dim r As String
Dim footer As Long

'load data into herd and cow objects
LoadHerd path

'clear the results sheets
footer = 0
resultsheet.Cells.Clear
cowsheet.Cells.Clear

'process the herd
For Each cow In cows

'make sure the cow is valid
If Not cow.isExcludedFromHerd Then

1 = cow. getTestNumber

’keep track of the number of cows in this test
last(t) = last(t) + 1

'display the adjusted FPR value
resultsheet.Cells(last(t) + header, t + 1) = cow.adjustedF pr

'display the cow ID number
cowsheet.Cells(last(t) + header, t + l) = cow.id

'keep track of where the footer will go
If last(t) > footer Then footer = last(t)
Else
'this cow is rejected
last(O) = last(O) + l
cowsheet.Cells(last(O) + header, 1) = cow.id & " " & cow.reason
End If
Next

91

footer = footer + header + 1
'display the test results

For t = I To tests

'temporary range for statistical calculations
r = Chr(t + 65) & "2:" & Chr(t + 65) & footer - 1

resultsheet.Cells(2, t + l) = "Test " & t
cowsheet.Cells(Z, t + 1) = "Test " & t

resultsheet.Cells(footer + 0, t + 1) = "=count(" & r & n)"
resultsheet.Cells(footer + 1, t + 1) = "=average(" & r & ")"
resultsheet.Cells(footer + 2, t + 1) = "=stdev(" & r & n)"

Next
'finish up the result sheet

With resultsheet
.Cells(1, 2) = "Fat / Protein"
.Cells(l, 2).Font.Bold = True

.range(.Cells(l, 2), .Cells(l, tests + 1)).Merge
.range(.Cells(l, 2), .Cells(l, tests + 1)).HorizontalAlignment = xlCenter

.Cells(footer + O, 1) = "Count"
.Cells(footer + I, I) = "Mean"
.Cells(footer + 2, 1) = "Stdev"

HighlightRegion .range(.Cells(I, 2), .Cells(header, tests + 1)), RGB(255, 255, 153)

HighlightRegion .range(.Cells(footer, 2), .Cells(footer + 2, tests + l)), RGB(204,
255,204)
End With

'finish up the cow id sheet

With cowsheet
.Cells(1, 2) = "Cow ID Numbers"
.Cells(2, 1) = "Excluded"
.Cells(l, 2).Font.Bold = True

.range(.Cells(l, 2), .Cells(l, tests + 1)).Merge
.range(.Cells(l , 2), .Cells(l , tests + 1)).HorizontalAlignment = xlCenter

HighlightRegion .range(.Cells(l, 2), .Cells(header, tests + 1)), RGB(255, 255, 153)

92

HighlightRegion .range(.Cells(header + 1, 1), .Cells(last(0) + header, 1)), RGB(255,
153,204)
End With

Dim oldblanks As Long

'update the graph
With resultchart
'tricky problem: can't change formula if the series is blank and visible
oldblanks = .DisplayBlanksAs
.SeriesCollection("This Herd").FormulaR1Cl = _
"=SERIES(""This Herd"",,’Results Data'!R" & footer + 1 & "C2:R" & footer + 1
& "C" & tests + 1 & ",1)"
'retum old visibility
.DisplayBlanksAs = oldblanks
End With

End Sub
Private Sub LoadCows(path As String)

Dim f As Integer
Dim s As String
Dim c As clsCow

'get a free file handle
f = F reeF ile

'open the cow file

On Error Resume Next

Open path & "\DRMSCOWS.DI " For Input As f

If Err.number <> 0 Then
Err.Clear
Open path & "\DRMSCOWS.D1.csv" For Input As f
If Err.number Then Exit Sub

End If

On Error GoTo 0

Do Until EOF(f)
'read a line at a time

Line Input #f, 8

'add the cow to the herd
Set c = New clsCow

With c

93

 

.id = CLng(Mid(s, 1, 5))
.lactationNumber = CLng(Mid(s, 1064, 2))
Set .herd = Me

End With

cows.Add c, "i" & c.id
Loop

'close the file
Close f

End Sub
Private Sub LoadHerd(path As String)

LoadCows path
LoadTests path
RemoveNullTests
UpdateLactationNumbers
CalculatePeaks

End Sub
Private Sub LoadTests(path As String)

Dim c As clsCow
Dim f As Integer
Dim s As String

Dim t As clsMilkTest

'get a free file handle
f = FreeFile

'open the milk test file

On Error Resume Next

Open path & "\DRMSCOWS.D2" For Input As f

If Err.number <> 0 Then
Err.Clear
Open path & "\DRMSCOWS.D2.csv" For Input As f
If Err.number <> 0 Then Exit Sub

Endlf

On Error GoTo 0

Do Until EOF(f)

’read a line at a time
Line Input #f, s

94

'create a new milk test
Set t = New clsMilkTest

t.cow = CLng(Mid(s, 1, 5))
t.calvingDate = MilkCDate(Mid(s. 7, 8))
t.daysInMilk = CDbl(Mid(s, 16, 4))
t.milk = CDbl(Mid(s, 21, 5))

t.fat = CDbl(Mid(s, 27, 3))

t.protein = CDbl(Mid(s, 31, 3))
t.carCode = Trim(Mid(s, 39, 2))

'add it to the collection of tests

On Error Resume Next

Set c = cows("i" & t.cow)

If Err.number > 0 Then
'found a test with no corresponding cow
'this should never happen
'all cows should be listed in drmscowsdl
Stop

Endlf

On Error GoTo 0

 

cows("i" & t.cow).addTest t
Loop

Close f
End Sub
Public Function Peak(parity As Long, season As Long) As Double
'should not be trying to retreive the peak if they have not been calculated yet
'something must have gone wrong
If Not peaksCalculated Then Stop
Peak = peakMilk(parity, season)
End Function

Private Sub RemoveNullTests()

Dim cow As clsCow
Dim i As Long

i=1

95

 

Do Until i > cows.count
cows(i).RemoveNullTests
If cows(i).lactations.count = 0 Then
cows.Remove i
Else
i = i + 1
End If
Loop

End Sub

Private Sub UpdateLactationNumbers()
Dim cow As clsCow
Dim lactation As clsLactation

Dim i As Long

For Each cow In cows
i = cow.lactationNumber - cow.lactations.count + 1

For Each lactation In cow.lactations
lactation.number = i
i = i + 1
Next
Next
End Sub
Private Sub C lass_lnitialize()
Set cows = New Collection
End Sub
Private Sub Class_Terminate()

Set cows = Nothing

End Sub

 

 

VERSION 1.0 CLASS
BEGIN

MultiUse = -1 'True
END

96

Attribute VB_Name = "clsLactation"
Attribute VB_GlobalNameSpace = False
Attribute VB_Creatable = False

Attribute VB_PredeclaredId = False
Attribute VB_Exposed = False

'lactation class

’2001 by Will Raphael and Joel Dobrzelewski
'Michigan State University

Option Explicit
Public reason As String

Public calvingDate As Date
Public Peak As Double
Public tests As Collection

Public cow As clsCow
Private num As Long

Public Sub addTest(test As clsMilkTest)

test.number = tests.count + l
tests.Add test
If test.mi1k > Peak Then Peak = test.milk

End Sub
Public Function excluded() As Boolean
Dim test As clsMilkTest

If tests.count < 1 Then
reason = "tests.count< l "
GoTo YesExcluded

End If

If tests.count > MaximumNumberOfTests Then
reason = "tests.count=" & tests.count & ">" & MaximumNumberOfTests
GoTo YesExcluded

End If

For Each test In tests
If Not test.isValid Then
reason = test.reason
GoTo YesExcluded

97

 

 

End If
Next

excluded = False
Exit Function

YesExcluded:
excluded = True
End Function
Public Function goodLength() As Boolean

'assume failure
goodLength = False

If tests.count <= MaximumNumberOfTests Then
Select Case parity
Case 1
If tests.count >= 8 Then goodLength = True
Case 2, 3
If tests.count >= 4 Then goodLength = True
Case Else
'invalid parity - should not happen
'could be a file problem
Stop
End Select
End If

End Function
Public Function lastTest() As clsMilkTest
Set lastTest = tests(tests.count)

End Function

Public Property Get number() As Long
'all tests should be numbered
'must be a logic problem when tests were added

If num = 0 Then Stop

number = num

98

End Property

Public Property Let number(value As Long)
num = value

End Property

Public Function parity() As Long

Select Case num

Case 1
parity = 1
Case 2
parity = 2
Case Is >= 3
parity = 3
Case Else
parity = 0
End Select

End Function

Public Sub RemoveNullTest()
If lastTest.fat = 0 And lastTestprotein = 0 And lastTest.milk = 0 Then
tests.Remove tests.count
Endlf
End Sub
Public Function season() As Long
Const winter = 356 'dec 22
Const spring = 79 'mar 20
Const summer = 172 'j un 21
Const autumn = 265 'sep 22
Dim (1 As Long
d = DatePart("y", lastTest.calvingDate)
Select Case d

Case winter To 366, 1 To spring - 1
season = 1

99

Case spring To summer - 1

season = 2

Case summer To autumn - 1
season = 3

Case autumn To winter - 1
season = 4

Case Else
season = 0
'invalid calving date?
Stop

End Select

End Function
Private Sub Class_lnitialize()

Set tests = New Collection
Peak = 0

End Sub
Private Sub Class_Terminate()
Set tests = Nothing

End Sub

 

VERSION 1.0 CLASS
BEGIN

MultiUse = -1 'True
END
Attribute VB_Name = "clsMilkTest"
Attribute VB_GlobalNameSpace = False
Attribute VB_Creatable = False
Attribute VB_PredeclaredId = False
Attribute VB_Exposed = False
'milk test class
'2001 by Will Raphael and Joel Dobrzelewski
'Michigan State University

Option Explicit
Public calvingDate As Date

Public carCode As String
Public cow As Long

100

Public daysInMilk As Double
Public fat As Double

Public milk As Double
Public number As Long
Public protein As Double
Public reason As String

Function fpr() As Double

If protein <> 0 Then
fpr = fat / protein
Else
fpr = 0
End If

End Function

Public Function isValid() As Boolean

 

'check for out-of-range values for days in milk
If number = 1 Then
If daysInMilk < 7 Or daysInMilk > 38 Then
reason = "test 1 dim=" & daysInMilk
GoTo NotValid
End If
ElseIf number = 2 Then
If daysInMilk < 32 Or daysInMilk > 75 Then
reason = "test 2 dim=" & daysInMilk
GoTo NotValid
End If

Elself number = 3 Then
If daysInMilk < 63 Or daysInMilk > 113 Then
reason = "test 3 dim=" & daysInMilk
GoTo NotValid
End If
End If

'check for zero/blank values

If daysInMilk = 0 Or milk = 0 Or fat = 0 Or protein = 0 Then
reason = "dim milk fat or protein=0"
GoTo NotValid

End If

'check for invalid condition codes

101

 

If carCode = "AF" Or carCode = "HF" Or carCode = "A" Or carCode = "E" Or
carCode = "F" Or carCode = "1" Or carCode = "L" Then
reason = "carCode=" & carCode

GoTo NotValid
End If
isValid = True

Exit Function

NotValid:

isValid = False

End Function

102

10.

ll.

BIBLIOGRAPHY

Grieve DG, Korver S, Rijpkema YS, Hof G. Relationships Between Milk
Composition and Some Nutritional Parameters in Early Lactation. Livest Prod Sci
1986;14:239-254.

Stobbs TH, Brett DJ. Milk Yield and Composition of Milk and Blood as
Indicators of Energy Intake by Jersey Cows. Aust J Agric Res 1974;25:657-666.

Hagert C. Continuous Control of Energy and Protein Supply of the Dairy Cow at
Peak Lactation by Means of Acetone, Urea, Protein, and Fat Concentrations in
Milk. Thesis: Ludwig Maximilians Universitat Munchen, Germany, 1991:126.

Heuer C, Schukken YH, Dobbelaar P. Postpartum Body Condition Score and
Results from the First Test Day Milk as Predictors of Disease, Fertility, Yield, and
Culling in Commercial Dairy Herds. J Dairy Sci 1999;82:295-304.

Geishauser TD, Leslie KE, Duffield TF, Edge VL. An Evaluation of Protein/Fat
Ratio in First DHI Test Milk for Prediction of Subsequent Displaced Abomasum
in Dairy Cows. Can J Vet Res 1998;62:144-147.

Geishauser T. Investigations on Etiology and Pathogenesis of Abomasal
Displacement in Adult Cattle: Heritability, FatzProtein Quotient, and Motility
Disorders. Bov Pract 1996;30:72-74.

Loeffler SH, DeVries MJ, Schukken YH. The Effects of Time of Disease
Occurrence, Milk Yield, and Body Condition on Fertility of Dairy Cows. J Dairy
Sci 1999;82:2589-2604.

Kristula MA, Reeves M, Redlus H, Uhlinger C. A Preliminary Investigation of
the Association Between the First Postpartum Milk Fat Test and First
Insemination Pregnancy Rates. Prev Vet Med 1995;23:95-100.

VanArendonk JA, Hovenier R, DeBoer W. Phenotypic and Genetic Association
Between Fertility and Production in Dairy Cows. Livest Prod Sci 1989;21:1-12.

Rodriquez LA, Mekonnen G, Wilcox CJ, Martin FG, Krienke WA. Effects of
Relative Humidity, Maximum and Minimum Temperature, Pregnancy, and Stage
of Lactation on Milk Composition and Yield. J Dairy Sci 1985;68:973-978.

Sharma AK, Wilcox CJ, Martin FG, Thatcher WW. Effects of Stage of Lactation

and Pregnancy and their Interactions on Milk Yield and Constituents. J Dairy Sci
1990;73:1586-1592.

103

Tw-

 

12.

l3.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

Sutton JD. Altering Milk Composition by Feeding. J Dairy Sci 1989;72:2801-
2814.

Gaynor PJ, Waldo DR, Capuco AV, Erdman RA, Douglass LW, Teter BB. Milk
Fat Depression, the Glucogenic Theory, and T rans-C18;1 Fatty Acids. J Dairy Sci

1995;78:2008-2015.

Romo G, Erdman R, Teter B, Casper D. Potential Role of Trans Fatty Acids in

Diet Induced Milk Fat Depression in Dairy Cows. Maryland Nutrition Conference
for Feed Manufacturers. College Park, MD, 1994.

Leat WM, Kemp P, Lysons RJ, Alexander TJ. Fatty Acid Composition of Depot
Fats from Gnotobiotic Lambs. J Agric Sci 1977;88:175-179.

Kemp P, Lander DJ, Gunstone FD. The Hydrogenation of Some Cis- and Trans-
Octadecenoic Acids to Stearic Acid by a Rumen F usocillus sp. Br J Nutr
1984;52:165-170.

Sachan DS, Davis CL. Hydrogenation of Linoleic Acid by a Rumen Spirochete. J
Bacteriol 1969;98:300-301.

Mortimer CE, Niehaus WG. Enzymatic Isomerization of Oleic Acid to Trans-
AlO-Octadecenoic Acid. Biochem Biophys Res Commun 1972;49:1650-1656.

Kemp P, White RW, Lander DJ. The Hydrogenation of Unsaturated Fatty Acids
by Five Bacterial Isolates from the Sheep Rumen, Including a New Species. J Gen
Microbiol 1975;90:100-1 14.

Harfoot CG, Noble RC, Moore JH. Factors Influencing the Extent of
Biohydrogenation of Linoleic Acid by Rumen Micro-organisms In Vitro. J Sci
Food Agric 1973;24:961-970.

Romo G, Teter B, Erdman R, Sampugna J. Effect of pH and Substrate on the In
Vitro Biohydrogenation of 18 Carbon Unsaturated Fatty Acids. J Dairy Sci
1992;75, Suppl. 1:170.

Latham MJ, Storry JE, Sharpe ME. Effect of Low-Roughage Diets on the
Microflora and Lipid Metabolism in the Rumen. Appl Microbiol 1972;24:871-
877.

Qiu X, Eastridge ML, Griswold KE, F irkins JL. Effect of Solid Passage Rate, pH,
and Level of Linoleic Acid on the Production of C is-9, Trans-1 l-octadecadienoic

acid (CLA) in Continuous Culture. J Dairy Sci 2000;83, Suppl. 1:285.

Davis CL, Brown RE. Low-Fat Milk Syndrome. In: Phillipson AT, ed.
Physiology of Digestion and Metabolism in the Ruminant: Proceedings of the 3rd

104

 

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

International Symposium. Cambridge, England: Oriel Press, Newcastle, England,
1969:545-565.

Griinari JM, Dwyer DA, McGuire MA, Bauman DE, Palmquist DL, Nurmela
KV. T ram-Octadecenoic Acids and Milk Fat Depression in Lactating Dairy
Cows. J Dairy Sci 1998;81:1251-1261.

Wonsil BJ, Herbein JH, Watkins BA. Dietary and Ruminally Derived Trans-18:1
Fatty Acids Alter Bovine Milk Lipids. J Nutr 1994;124:556-565.

Teter BA, Sampugna J, Keeney M. Milk Fat Depression in C57B1/6J Mice
Consuming Partially Hydrogenated Fat. J Nutr 1990;120:818-824.

Gaynor PJ, Erdman RA, Teter BB, Sampugna J, Capuco AV, Waldo DR, Hamosh
M. Milk Fat Yield and Composition During Abomasal Infusion of Cis or Trans
Octadecenoates in Holstein Cows. J Dairy Sci 1994;77:157-165.

Piperova L, Sampugna J, Teter B. Comparison of Trans Octadecenoic Isomer
Profiles in Duodenal and Milk Lipids of Cows Fed Different Diets. J Dairy Sci
2000;83, Suppl. 1:277.

Piperova LS, Teter BB, Sampugna J, Yurawecz MP, Bruckental I, Erdman RA.
Dietary Milk Fat Depression and Trans-18:1 and CLA Isomer Distribution in
Milk of Lactating Cows. J Dairy Sci 2000;83, Suppl. 1:277.

Griinari JM, Chouinard PY, Bauman DE. Trans Fatty Acid Hypothesis of Milk
Fat Depression Revised. Cornell Nutrition Conference for Feed Manufacturers.
Ithaca, NY, 1997:208-216.

Loor JJ, Herbein JH. Exogenous T ranle, Cisl2-l 8:2 Reduces De Novo
Synthesis and Desaturation of Milk Fatty Acids in Cows Fed Diets Supplemented
with High-oleic or High-linoleic Oil. J Dairy Sci 2000;83, Suppl. 1:162.

Baumgard LH, Corl BA, Dwyer DA, Saebo A, Bauman DE. Identification of the
Conjugated Linoleic Acid Isomer that Inhibits Milk Fat Synthesis. Am J Physiol,
Regulatory Integrative Comparative Physiology 2000;278:R179-R184.

Thorson J C, Erdman RA, Piperova LS, Teter BB, Sampugna J, Auchtung TL. The
Effect of Abomasal Infusion of Conjugated Linoleic Acid on Milk Fat of
Lactating Dairy Cows. J Dairy Sci 2000;83, Suppl. 1:279.

Chouinard PY, Corneau L, Barbano DM, Metzger LE, Bauman DE. Conjugated

Linoleic Acids Alter Milk Fatty Acid Composition and Inhibit Milk Fat Secretion
in Dairy Cows. J Nutr 1999;129:1579-1584.

105

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

Viswanadha S, Hanson TW, Giesy JG, McGuire MA. Response of Milk Fat to
Intravenous Administration of the Trans-10, Cis-l2 isomer of Conjugated
Linoleic Acid. J Dairy Sci 2000;83, Suppl. 1:163.

Matitashvili E, Bauman D. Effect of Different Isomers of C18:1 and C18:2 Fatty
Acids on Lipogenesis in Bovine Mammary Epithelial Cells. J Dairy Sci 2000;83,
Suppl. 1:165.

McClymont GL, Vallance S. Depression of Blood Glycerides and Milk-Fat
Synthesis by Glucose Infusion. Proc Nutr Soc 1962;21:xli.

McGuire MA, Griinari JM, Dwyer DA, Bauman DE. Role of Insulin in the
Regulation of Mammary Synthesis of Fat and Protein. J Dairy Sci 1995;78:816-
824.

Griinari JM, McGuire MA, Dwyer DA, Bauman DE, Palmquist DL. Role of
Insulin in the Regulation of Milk Fat Synthesis in Dairy Cows. J Dairy Sci
1997;80:1076-1084.

Eadie JM, Hyldgaard-Jensen J, Mann SO, Reid RS, Whitelaw FG. Observations
on the Microbiology and Biochemistry of the Rumen in Cattle Given Different
Quantities of a Pelleted Barley Ration. Br J Nutr 1970;24:157-177.

Mann SO. Some Effects on the Rumen Micro-organisms of Overfeeding a High
Barley Ration. J Appl Bacteriol 1970;33:403-409.

Latham MJ, Sharpe ME, Sutton JD. The Microbial Flora of the Rumen of Cows
Fed Hay and High Cereal Rations and its Relationship to the Rumen
Fermentation. J Appl Bacteriol 1971;34:425-434.

Latham MJ, Sutton JD, Sharpe ME. Fermentation and Microorganisms in the
Rumen and the Content of Fat in the Milk of Cows Given Low Roughage
Rations. J Dairy Sci 1974;57:803-810.

VanBeukelen P, Wensing T, Breukink HJ. Studies on Mechanisms in Milk Fat
Depression in Dairy Cows: A Comparison of Chopped Hay Rations and High
Concentrate Rations. J Anim Physiol Anim Nutr 1986;55:224-238.

Sutton JD, Hart IC, Broster WH, Elliott RJ, Schuller E. Feeding Frequency for
Lactating Cows: Effects on Rumen Fermentation and Blood Metabolites and
Hormones. Br J Nutr 1986;56:181-192.

Grummer RR, Jacob AL, Woodford JA. Factors Associated with Variation in

Milk Fat Depression Resulting from High Grain Diets Fed to Dairy Cows. J Dairy
Sci 1987;70:613-619.

106

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

Goad DW, Goad CL, Nagaraja TG. Ruminal Microbial and Fermentative
Changes Associated with Experimentally Induced Subactue Acidosis in Steers. J
Anim Sci 1998;76:234-241.

McCullough ME. Relationships Between Rumen Fluid Volatile Fatty Acids and
Milk Fat Percentage and Feed Intake. J Dairy Sci 1966;49:896-898.

Jenny BF, Polan CE, Thye FW. Effects of High Grain Feeding and Stage of
Lactation on Serum Insulin, Glucose and Milk Fat Percentage in Lactating Cows.
J Nutr 1974;104:379-385.

Jenny BF, Polan CE. Postprandial Blood Glucose and Insulin in Cows Fed High
Grain. J Dairy Sci 1975;58:512-514.

Hurtaud C, Rulquin H, Verite R. Effect of Infused Volatile Fatty Acids and
Caseinate on Milk Composition and Coagulation in Dairy Cows. J Dairy Sci
1993;76:3011-3020.

Sutton JD, Hart IC, Morant SV, Schuller E, Simmonds AD. Feeding Frequency
for Lactating Cows: Diurnal Patterns of Hormones and Metabolites in Peripheral
Blood in Relation to Milk-Fat Concentration. Br J Nutr 1988;60:265-274.

Benson JD, Askew EW, Emery RS, Thomas JW. Effects of a Restricted
Roughage Ration on Lipoprotein Lipase and Glyceride Synthetase Activites in
Bovine Mammary and Adipose Tissues. Fed Proc 1969;28:623.

Askew EW, Benson JD, Thomas JW, Emery RS. Metabolism of Fatty Acids by
Mammary Glands of Cows Fed Normal, Restricted Roughage, or Magnesium
Oxide Supplemented Rations. J Dairy Sci 1971;54:854-862.

Opstvedt J, Baldwin RL, Ronning M. Effect of Diet Upon Activities of Several

Enzymes in Abdominal Adipose and Mammary Tissues in the Lactating Dairy
Cow. J Dairy Sci 1967;50:108-109.

Laarveld B, Christensen DA, Brockman RP. The Effect of Insulin on Net
Metabolism of Glucose and Amino Acids by the Bovine Mammary Gland.
Endocrinology 1981;108:2217-2221.

Laarveld B, Chaplin RK, Brockman RP. Effects of Insulin on the Metabolism of
Acetate, B-hydroxybutyrate and Triglycerides by the Bovine Mammary Gland.
Comp Biochem Physiol 1985;828:265-267.

Benson JD, Askew EW, Emery RS, Thomas JW. Metabolism of Fatty Acids by

Adipose Tissue and Liver of Cows Fed Normal, Restricted Roughage or MgO
Supplemented Rations. J Dairy Sci 1972;55:83-92.

107

60.

61.

62.

63.

64.

65.

66.

67.

68.

69.

70.

71.

Vernon RG. Lipid Metabolism in the Adipose Tissue of Ruminant Animals. Prog
Lipid Res 1980;19:23-106.

Mackle TR, Dwyer DA, Ingvartsen KL, Chouinard PY, Lynch JM, Barbano DM,
Bauman DE. Effects of Insulin and Amino Acids on Milk Protein Concentration
and Yield from Dairy Cows. J Dairy Sci 1999;82:1512-1524.

Emery RS. Feeding for increased Milk Protein. J Dairy Sci 1978;61:825-828.

Sporndly E. Effects of Diet on Milk Composition and Yield of Dairy Cows with
Special Emphasis on Milk Protein Content. Swed J Agric Res 1989;19:99-106.

Sutton JD, Morant SV. A Review of the Potential of Nutrition to Modify Milk Fat
and Protein. Livest Prod Sci 1989;23:219-237.

Murphy JJ, O’Mara F. Nutritional Manipulation of Milk Protein Concentration
and its Impact on the Dairy Industry. Livest Prod Sci 1993;35:117-134.

Rook JA, Balch CC. The Effects of Intraruminal Infusions of Acetic, Propionic
and Butyric Acids on the Yield and Composition of the Milk of the Cow. Br J
Nutr 1961;15:361-369.

Rook JA, Balch CC, Johnson VW. Further Observations on the Effects of
Intraruminal Infusions of Volatile Fatty Acids and of Lactic Acid on the Yield
and Composition of the Milk of the Cow. Br J Nutr 1965;19:93-99.

Miettinen H, Huhtanen P. Effects of the Ratio of Ruminal Propionate to Butyrate
on Milk Yield and Blood Metabolites in Dairy Cows. J Dairy Sci 1996;79:851-
861.

Seal CJ, Parker DS. Effect of Intraruminal Propionic Acid Infusion on
Metabolism of Mesenteric- and Portal-Drained Viscera in Growing Steers Fed a
Forage Diet: II. Ammonia, Urea, Amino Acids, and Peptides. J Anim Sci
1996;74:245-256.

Huhtanen PJ, Blauwiekel R, Saastamoinen I. Effects of Intraruminal Infusions of
Propionate and Butyrate with Two Different Protein Supplememts on Milk
Production and Blood Metabolites in Dairy Cows Receiving Grass Silage-Based
Diet. J Sci Food Agric 1998;77:213-222.

Casse EA, Rulquin H, Huntington GB. Effect of Mesenteric Vein Infusion of
Propionate on Splanchnic Metabolism in Primiparous Holstein Cows. J Dairy Sci
1994;77:3296-3303.

108

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

Griinari JM, McGuire MA, Dwyer DA, Bauman DE, Barbano DM, House WA.
The Role of Insulin in the Regulation of Milk Protein Synthesis in Dairy Cows. J
Dairy Sci 1997;80:2361-2371.

Sniffen CJ, Robinson PH. Microbial Growth and Flow as Influenced by Dietary
Manipulations. J Dairy Sci 1987;70:425-441.

Poore MH, Moore JA, Eck TP, Swingle RS, Theurer CB. Effect of Fiber Source
and Ruminal Starch Degradability on Site and Extent of Digestion in Dairy Cows.
J Dairy Sci 1993;76:2244-2253.

Wu Z, Sleiman FT, Theurer CB, Santos F, Simas JM, Francolin M, Huber JT.
Effect of Isocaloric Infusion of Glucose in the Rumen or Propionate in the
Duodenum. J Dairy Sci 1994;77:1556-1562.

Crabtree RM. Milk Compositional Ranges and Trends. In: Castle ME, Gunn RG,
eds. Milk Compositional Quality and its Importance in Future Markets.
Edinburgh: British Society of Animal Production, 1984:35-42.

Emery RS. Milk Fat Depression and the Influence of Diet on Milk Composition.
Vet Clin North Am, Food Anim Pract 1988;4z289-305.

Palmquist DL, Beaulieu AD, Barbano DM. Feed and Animal Factors Influencing
Milk Fat Composition. J Dairy Sci 1993;76:1753-1771.

F redeen AH. Considerations in the Nutritional Modification of Milk Composition.
Anim Feed Sci Technol 1996;59:185-197.

Schutz MM, Hansen LB, Steuemagel GR, Kuck AL. Variation of Milk, Fat,
Protein, and Somatic Cells for Dairy Cattle. J Dairy Sci 1990;73:484-493.

Hargrove GL, Mbah DA, Rosenberger J L. Genetic and Environmental Influences
on Milk and Milk Component Production. J Dairy Sci 1981;64:1593-1597.

DePeters EJ, Cant JP. Nutritional Factors Influencing the Nitrogen Composition
of Bovine Milk: A Review. J Dairy Sci 1992;75:2043-2070.

Ng-Kwai-Hang KF, Hayes JF, Moxley JE, Monardes HG. Environmental
Influences on Protein Content and Composition of Bovine Milk. J Dairy Sci
1982;65:1993-1998.

SAS OnlineDoc. Version 8 ed. SAS Inst., Inc., Cary, NC, 1999.

Oakes MW. Statistical Inference. Chestnut Hill, MA: Epidemiology Resources,
1986.

109

 

86.

87.

Subcommittee on Dairy Cattle Nutrition and Committee on Animal Nutrition,
National Research Council. Nutrient Requirements of Dairy Cattle. 7th ed.
Washington, DC: National Academy Press, 2001.

Nordlund KV, Garrett EF, Oetzel GR. Herd-Based Rumenocentesis: A Clinical

Approach to the Diagnosis of Subacute Rumen Acidosis. Comp Contin Educ
Pract Vet 1995;17:548-556.

110

 

   

IIIIIIIIIIIIIIIIIIIII

11111111111111111111,111