RELA‘HGNS EWEER WfiGH‘? AT FiRST
CALVlNfi AM) MSLZA’. PRCfiUCflv ‘
DURENG THE HRS? LACTA?§©N

Thesis for flu D099” of M. $.
MECHEGAR STATE UNEVERSITY
Robot? H. MEEM’
1958

LIBRARY

Michigan State
University

 

 

RELATIONS BETWEEN WEIGHT AT FIRST CALVING AND MILK PRODUCTION
DURING THE FIRST LACTATTON

BY
Robert H. Miller

AN ABSTRACT

Submitted to the College of Agriculture
Michigan State University of Agriculture and
Applied Science in partial fulfillment of
the requirements for the degree of

MASTER OF SCIENCE
Department of Dairy

Year 1958

Approved cam/v u. 7. 1 » . WJ

 

ABSTRACT ROBERT H. MILLER

First lactation milk and fat production records (includ-
ing age and weight at calving) for h677 Holsteins, lOOl
Guernseys, and 501 Jerseys compiled in the Michigan DHIA-IBM
program over a three and one-half year period were analyzed
to ascertain the relation between weight and production.

Least squares estimates of the independent influences of
age and weight at calving on first lactation production were
derived. Maturity, as measured by age, was negatively re-
lated to the economy of first lactation production. The ex-
pected change in production associated with a change in
weight at first calving was dependent upon the level at which
a change in weight occurred. In all three breeds, there were
instances where a decrease in production was associated with
increased weight. The production response to level of de-
velopment appeared to reach a maximum at weights 300-h00
pounds above the breed average for the age. When the value
of added production accompanying increased weight was con-
trasted to the added expense involved, heavier heifers had
little or no advantage in the first lactation. It was con-
cluded that dairy heifers should be bred as soon as suffi-

. cient size is attained to minimize harmful effects on the
length of productive life.

Partial regressions of production on age and weight were

computed on overall and intraherd bases. Weight was a more

ABSTRACT . ROBERT H. MILLER

important source of variation in production records in a group
of many herds than age. Large standard errors of estimate
suggested that the linear regression equation was inefficient
in predicting individual production records in the group stud-
ied. Weight was only about one—half as important in predict-
ing within-herd production as compared to production in a
group of many herds. The intraherd regressions of milk pro-
duction on age and weight were of the order of 75 lb. per
month and 200 lb. per 100 lbs., respectively. Partial corre-
lations indicated that age had little or no role in the asso-
ciation of weight and production in a group of many herds but
wasimportant in the intraherd relationship.

Large between herd correlations of weight with production
indicated that differences between herds contributed signifi-
cantly to the association. Irregular results were obtained
for Jerseys and Guernseys.

Estimates of the genetic correlation between weight and
production for Holsteins were of the order of .3, indicating
significant genetic contribution to the association between
weight and production.'

Individual differences in weight at first calving were
found to be heritable to the extent of .3 to .5. Heritabili-
ties of first lactation milk and fat production ranged from

.2 to .3 and .2 to .h, respectively.

ABSTRACT ' ROBERT H. MILLER

It was concluded that, while differences between herds
indicated that environmental factors were of primary impor-
tance, genetic contribution to the weight-production rela-
tionship was large enough to result in indirect increases in
weight through selection for milk production.

It was further concluded that correction for weight dif-
ferences may remove a fraction of the genetic variation in

production.

RELATIONS BETWEEN WEIGHT AT FIRSTPCALVING AND MILK PRODUCTION
DURING THE FIRST LAC TATTON

BY
Robert H. Miller

A THESIS

submitted to the College of Agriculture
Michigan State University of Agriculture and
Applied Science in partial fulfillment of
the requirements for the degree of

MASTER OF SCIENCE

Department of Dairy

Year 1958

 

 

 

 

 

i‘ )— . H .‘
_ _] I " 3 l
‘- , y l

.

ACKNOWLEGEMENTS

The writer wishes to express his gratitude to Dr. L. D.
McGilliard for the suggestion of the problem and for patient
guidance in the investigation.

Appreciation is due Dr. N. P. Ralston for the extension
of a graduate assistantship in connection with the author's
graduate work at Michigan State University.

Thanks are expressed to Dr. w. D. Baten for advice con-
cerning problems of analysis.

The author is indebted to Dr. D. E. Madden for friendly
counsel and to Mr. R. P. Witte and Mr. A. J. Thelen for invalu-
able assistance in the processing of the data.

Sincere appreciation is due the many faculty members of
Michigan State University whose thought-provoking tutelage made
the writer's graduate study meaningful.

Most of all, the author is grateful for the encouragement

and support of his mother and family.

11

\J a I

TABLE OF CONTENTS

INTRODUCTION . . . . . . . . . . . . . .
A REVIEW OF THE LITERATURE . . . . . . .
SOURCE OF DATA . . . . . . . . . . . . .
Description . . . . . . . . . . . .
Distribution and Means . . . . . .

METHODS AND RESULTS . . . . . . . . . .

Least squares estimates of influences

weight on production . . . . .
Analysis of variance and covariance
Heritabilities . . . . . . . . . .
Correlation analysis . . . . . . .
Multiple regression analysis . . .

DISCUSSION AND CONCLUSIONS . . . . . . .

O O O O O 0

Implications in dairy cattle management . . . . .

Implications in dairy cattle selection

Implications in statistical analysis of production

data . . . . . . . . . . . . .

Implications in future investigation

SU MIA BY 0 O O O O O O O O O O O O O O 0

REFERENCES 0 O O O O O O O O O O O O O 0

iii

12
13
17

17
29
33
35
1:5
51
51
SS

60
61
62

6’4

I
1
n W
q I
I a
I I
I o i
n I A
e . v
a o e
h. h |
e e p
a D t
w v a
. e D I
O I I
‘ . e
u C 1
Q . a
n O -

 

Table

10

11

12
13

1h

15

16

LIST OF TABLES

Distribution of Age at First Calving . . . . . .
Distribution of Weight at First Calving . . . .

Means and Standard Deviations of Age at Calv-
ing, Weight at Calving, Milk and Fat Production
For the First Lactation . . . . . . . . . . . .

Influence of Age of First Calving on Milk and
Fat Production in the First Lactation . . . . .

Influence of Age on Milk and Fat Production in
the First Lactation, Disregarding Weight . . . .

Influence of Weight at First Calving on Milk
and Fat Production in the First Lactation . . .

Value of Increased Milk Production in the First
Lactation From Older Freshening Ages, Contrasted
to Maintenance Cost Incurred . . . . . . . . . .

Value of Changes in Milk Production in the First
Lactation Due to Increased Weight Contrasted to
Cost of Gain and Maintenance . . . . . . . . . .
Number of Observations . . . . . . . . . . . . .

Components of Variance For Age, Weight, Milk,
and Fat . C O . C O O . C C O C O O O O O 0 O .

Components of Covariance Between Age, Weight,
Milk,andFat.................

Heritabilities . . . . . . . . . . . . . . . . .

Overall Phenotypic Correlations Between Age,
weight ’ Milk, and Fat 0 O O O 0 O O O O O O O O

Phenotypic Correlations Between Age, Weight,
Milk, and Fat Within HBPdS . e o o o O 0 O 0 e 0

Correlations Between Weight and Production of

Herds O O O O O O O O C O O O O O C O O O O O 0

Partial Correlations Between Age, Weight, Milk,
and Fat 0 O O O O O O O O O O O O O O O O O O 0

iv

Page
in
15

16

19

21

22

26

27

29

3O

31
33

35

36

37

38

Table
17

18

19

21

Page

Genetic Correlations Between Weight and Pro-
dUCtion e e e e e e e e e e e e e e e e e e e e ’41

A Comparison of Various Estimates of the
Genetic Correlation Between Weight and Produc-
tion e e e e e e e e e e e e e e e e e e e e e 143

Partial Regressions of Milk and Fat Production
onAgeandWeight............... ‘46

Within-Herds Partial and Standard Partial Re-
gressions of Production on Age and Weight . . . NB

Analysis of Variance of Regression Within Herds SO

LIST OF FIGURES
Figure Page

1 Influence of Age at First Calving on First
Lactation Milk Production . . . . . . . . 23

2 Influence of Weight at First Calving on First
Lactation Milk Production . . . . . . . . 23

vi

INTRODUCTION

How body weight and milk production are related has re—
ceived prominent attention by many workers. However, the em-
phasis which body size, as measured by weight, should receive
in dairy cattle selection and management programs and how it
should be used have not been definitely ascertained.

In practice, large animals are favored by breeders of
dairy cattle. Body size is a factor often considered impor-
tant in both inter- and intra-breed evaluation of dairy cat-
tle. The Unified Score-Card of the Purebred Dairy Cattle
Association (uh) subjects cows and bulls lacking size desired
for the breed to ”slight to serious discrimination" in the
show ring. Small animals are penalized in type classification.
Such evaluation is usually based on the theory that a defi-
ciency in size is to a large extent reflected in a deficiency
in the capacity for the consumption and storage of nutrients.
Size is also an important consideration in selecting the par-
ents of the next generation. However, the importance of size
to milk production and the factors responsible for the asso-
ciation are obscure.

If the relation of body weight pg; 33 to milk and fat
production is not large enough to be economically important,
size, as measured by weight, should be considered only in the
management of the dairy cow. If a close association of milk
and fat production with weight exists, but this connection is

primarily conditioned by environmental differences, selection

for gross size should not be a prominent consideration in a
dairy cattle selection program, except insofar as there may
be a genetic relation between size and other factors, such
as longevity. However, differences in size can be used to
remove environmental differences from production records.

If a close relationship between weight and production exists
and the genotype is important to the association, size should
receive attention in evaluating the genetic ability of ani-
mals fer milk production.

Closely associated with possible influences of weight on
milk yield is the energetic efficiency of milk production, a
topic which has been rigorously treated by Brody (6,7,8).
‘While the efficiency of production is an important subject
not necessarily synonymous with gross rate of production, a
discussion of its aspects is beyond the scope of the present
paper.

The objectives of this investigation are: (1) To assess
the direct quantitative effect of body weight upon milk and
butterfat production in the first lactation, and (2) To de-
termine the genetic or environmental nature of any associa-

tion which may exist.

A REVIEW OF THE LITERATURE

The general subject of the relationship of body size to
milk production has been reviewed by Beck and Turk (3) and by
Pfau and Bartlett (h3). In each case the conclusion was that
large cows tend to produce more than small cows. Gowen (20),
in a review of early work on the subject, concluded that the
large cow is a more economdcal producer than the small cow
because the small cow needs proportionately more feed for
maintenance. A review of work concerning the productive ef-
ficiency of milk production has been made by Brody (7).

Various analytical methods have been employed in an ef-
fort to discern any existing connection between live weight
and the amount of milk and butterfat produced by dairy cat-
tle.

Edwards (lu) arranged the milk production data for 2h00
cows of various breeds tested at the London Dairy Show be-
tween the years 1922-193h according to body weight. He con-
cluded that within the breed there was a slight tendency for
live weight and the economy of production to be inversely re-
lated. '

Various workers have examined milk production and body
weight data for significant phenotypic correlation. Davis
and Willet (13) found no correlation'between weight gain from
birth to two years of age with milk or fat production for the
first lactation or for the lifetime average when the records

of 76 Holstein females were analyzed. Davis 33 31. (12)

-3-

reported correlations of .2h, .h5, .58, .60, and .83 for Guern-
sey, Ayrshire, Jersey, Holstein, and all breeds, respectively,
between.fat-corrected milk (FCM) for the first eight months
of the lactation and body weight measured in the first month
of the lactation of cows of various ages. Gaines (15) found
a correlation of .70 between the yield of h% FCM during the
first eight months of the lactation and body weight measured
in the first month of the lactation of 11 Holstein cows of
various ages. This correlation decreased as weight was meas-
ured at successive stages of the lactation. Gowen (21) re-
ported correlations of .51 for both milk and fat with weight
in a study of Jersey Registry of Merit (ROM) data for 8&0
cows of various ages. In the same paper, Gowen analyzed 365-
day milk and fat records from the Jersey ROM for 1371 cows
whose weights were known and 10,5h7 cows whose weights were
estimated. This sample exhibited correlations of .h7 and .h8
of weight with milk and fat, respectively. Gowen's data were
obtained from animals of diverse ages. Hofmeyr (27), in a
study of 93h records of Red Dane heifers from 7h8 different
herds, found a correlation of .30 between weight at calving
and the first lactation yield of h% FCM. Touchberry (h8),

in an analysis of data for 187 Holstein dampdaughter pairs in
one herd, found phenotypic correlations of-.0h and-.08 for
milk and fat, respectively, with weight. The records.stud-
ied by Touchberry were those which began nearest the cow's

third year of age, but the data were corrected to a mature-

equivalent basis. Thrner (N9) reported correlations of .32
and .33 for weight with milk and fat, respectively, in an
analysis of Advanced Registry (AR) records of 2700 Guernsey
cows of various ages for which actual and estimated body
weights were available. Thrner (50) also reported a correla-
tion between weight and fat of .11 in a study of 8h22 Jersey
ROM records of cows of various ages (the data were corrected
for age differences).

The above estimates of the phenotypic correlation be-
tween weight and production range from -.1 to .7. Most values
obtained appear to have been of the order of .h. While there
are many differences in ages of the animals studied and treat-
ment of the data in these investigations, there is strong evi-
dence of an important phenotypic relationship between weight
and production.

A number of workers have computed regression coefficients.
of milk and fat on body weight. Gowen (21), in an analysis
of 8h0 Jersey ROM records of cows of various ages, reported
regressions of milk and fat (pounds) on weight (pounds) of
9.5 and .5, respectively. When the data were corrected for
age differences, the corresponding regressions were only 6.1
and .h. In the same paper, Gowen analyzed 1371 Jersey ROM
365-day records of cows of various ages. This sample exhibited
regressions of milk and fat on weight of 7.2 and .h, respec-
tively. Johansson (29) found a regression of first lactatiai
250-day butterfat yield (kg.) on weight at calving (kg.) of

.15 in an analysis of records of Red Dane heifers tested in

special progeny testing stations. Turner (M9). in a study of
AR butterfat records and body weights (actual and estimated)
of 2700 Guernsey cows of various ages, reported a regression
of fat on weight of .77 where fat and weight were measured

in pounds and age at calving was ignored. When only three to
four year-old animals were considered, the regression was .22.
In earlier work, Thrner gt_gl. (51) computed a regression value
for butterfat (pounds) of 1.02, from Jersey ROM data. From
the results of these workers, the regression of production a1
weigflit appears to be decreased when age differences are re-
moved.

WOrk of the preceding nature has been questioned on sev-
eral grounds, but the chief point of dissension lies in the
problem of whether production is truly a linear function of
body weight. Brody (8), Kleiber (33), and Kleiber and Mead
(3h,35) maintain the negative position, while Gaines and his
associates (16,17,18) upheld the affirmative. Brody has ad-
vanced the "power formula," I 3 dub, as more closely repre-
senting the relationship between the two quantities (Y's milk
energy produced, W's body weight, b - slope of the fitted line
on the logarithmic grid). To the exponent b, Brody assigns
the value 0.7 t'O.l, the magnitude of the standard error de-
pending upon the "dairy merit" of the animal (presumably the
inherited productive capacity). This argument is based on
the assumption that maintenance requirement increases with
body weight raised to an exponential value less than 1.0 (9),

rather than with simple body weight. This represents a reversal

on Brody's part, for in earlier work with Jersey ROM data (10),
butterfat yield was expressed as a linear function of body
weight. As a basis for the linear relation, it was asserted
that increase in milk secretion with age follows the same
course as the increase of body weight with age, and that,
therefore, the two should be linearly related or directly pro-
portional for a group of animals of all ages during the grow-
ing period. To express the common course of the two quanti-
ties with increasing age, the function X . A(l - e‘kt) was
presented, where X 2 body weight or milk secretion at any age t,
A is a constant, e is the base of the natural logarithm system,
and k is the velocity constant. (The above is also the func-
tion expressing the course of a monomolecular chemical reac-
tion.) Kleiber and Mead (3h,35) employed a modification of

the "power formula," terming Brody's (W)b the "metabolic body
size" of the cow. Kleiber and Mead have contended that lac-
tation rate is proportional to (body weight in kilograms)3/u.
The basis of this expression is the theory that ”the capacity
for rate of production in animals is in general proportional

to their rate of metabolism, or in terms of the 3/h power of
their body weight." (35) This expression was formulated from
data obtained from 2h Holstein and h2 Jersey cows at the Cali-
fornia station. As a measure of the "inherent ability of cows
for milk production," Kleiber and Mead (3h) proposed the ratio
of average daily milk production during a ten-month period

(FCM or milk energy basis) to mean "metabolic body size" of

the animal in kilograms 3fl". Gaines (16) tested another

productive efficiency measure (FCMg/Wl, where FCM8 : h% fat-
corrected milk production for the first eight months of the
lactation, W1 : live weight measured in the first month fol-
lowing parturition) against Kleiber's efficiency ratio descrflnd
above, by equating both to the expression a +-le (a = a con-
stant, b : least squares regression coefficient) and comparing
the goodness of fit. From such a comparison on actual data,
the metabolic body size hypothesis was rejected.

A second Objection to phenotypic correlation and regres-
sion measurements of the association between weight and pro-
duction is that they fail to consider the role of age with
respect to weight and production. Several workers have meas-
ured this relationship by correlation. Hofmeyr (27) reported
a correlation between age and weight at first calving of .32
in data from 93h Red Dane heifers from 7h8 different herds.
Johansson (29) found a correlation of .26 between age and
weight at first calving for Red Dane heifers tested in Danish
progeny stations. Gowen (21) reported a correlation of .38
in Jerseys, while Turner (M9) found .32 in AR data from 2700
Guernsey cows (when both age and weight were measured at the
end of the test period). Animals of various ages were stud-
ied in the latter two investigations.

thh.of the confusion in determining the direct influence
of body weight on milk and butterfat production arises from
the lack of agreement by workers concerning valid methods of

disentangling the combined effects of age and weight. When

production data are converted to a mature-equivalent basis,

some of the effects of weight are eliminated also (17, 18).
From a linear regression analysis (17), Gaines 22.5As concluded
that the contribution of age to production with weight held
constant was negligible compared to that of weight with con-
stant age, and that, therefore, the correction of production
data for age differences is ”biologically unsound." Some
workers (h8,h9,50) have attempted to assess the relationship
between weight and production utilizing age-corrected data.

Multiple linear regression (2,17,18,2l,29) has been eme
ployed to estimate the change in production associated with
a change in weight, independent of age. The results of these
studies will be compared in a later section. However, in
general, the findings have indicated that weight is much more
important to production than age.

To determine the association of weight and production in-
dependent of the influence of age, the method of partial cor-
relation has been used (21,27,29,h9). The results of these
studies suggest that the association between weight and pro~
duction is little influenced by age, but that the correlation
between age and production is much smaller when weight is con-
stant.

Where body weight has been found to have a direct bear-
ing on the production of milk and butterfat, a few attempts
have been made to ascertain the primary nature of the rela-
tionship, i.e. environmental or genetic. Bailey and Broster
(2) compared regressions between sire-groups and concluded

that selection for live weight would improve production only

10

by chance. Gowen (22) deduced that, in Jerseys, the weight
of the sire was of slight importance in predicting the daugh-
ter's probable production. In another paper (23), Gowen, on
the basis of parent-offspring and sibship correlations in
Jersey cattle, stated that "inheritance accounts for most of
the variation in size of these cattle, such environmental dif-
ferences as do exist playing but little part in the ultimate
constitution of these animalsn (the relationship of weight to
production was not discussed). Mason g£_§l, (hO), in a study
of records of Red Dane cattle in special progeny testing sta-
tions in Denmark, found a significant between-sire relation-
ship of weight and milk-yield, and concluded that selection
for yield or efficiency would result in a slight increase in
body size. Blackmore (h), in a study of BSA Holstein dam-
daughter pairs, concluded that selection for milk production
and meat-type conformation simultaneously would be impractical.
Pfau and Bartlett (h3) found no evidence of a genetic rela-

- tion between size and milk production from a review of the
literature. Shrode and Lush (A7), in a review of the inheri-
tance of economic characters in farm animals, indicated that
the extent of pleiotropic action of genes controlling weight
and production might prove large enough to be important.
Touchberry (h8) found no evidence for a positive genetic con-
nection between size and production, but the data studied had
been corrected for differences in age. Thrner (30) inferred,
from a study of Jersey ROM fat records and body weights, that

the sire could cause significant changes in the daughter's fat

11

production over that of the dam, without materially changing
the level of the daughter's body weight over the dam's.

Thus, correlation and regression studies have indicated
that there is an important association between weight and pro-
duction. The results of these investigations have been ques-
tioned on the basis of whether production is actually a linear
function of body weight. An exponential expression has been
proposed as a more accurate expression of the relationship.
The exponential relationship is based on metabolic and ener-
getic considerations from the standpoint of physiology. Ap-
plications of the exponential formula to empirical data have
not yielded conclusive results. i

Partial correlation and regression have been employed
to separate the influences of age and weight on production.
Results of these studies indicate that weight is more impor-
tant than age in predicting production.

Efforts to ascertain the role of the genotype in the re-
lationship between weight and production have not yielded con-

clusive results.

SOURCE OF DATA

DESCRIPTION

 

The data were records completed in the Michigan DHIA-IBM
program during the period from January, 195h, through May,
1957 (no corrections for yearly differences were made). The
following measurements were obtained from individual first
lactation records: (1) actual milk production, (2) actual
fat production, (3) age at calving, and (h) body weight at
calving. Only completed lactations of 305 days or less were
used; lactations continuing beyond 305 days were cut off at
305 days. '

Numbers sufficient for analysis were available for only
three breeds. The breeds studied and the number of records
for each were: (1) h677 Holsteins, (2) 1001 Guernseys, and
(3) 501 Jerseys.

To facilitate computation, milk was coded in hundreds of
pounds, while fat and weight were coded in tens of pounds.
Age was recorded in months..

weight was estimated by means of taped chest girth meas-
urements based on the relationship between the two published
by the Bureau of Dairy Industry (32). Chest girth was meas-
ured by the DHIA supervisor on his first visit following the
date of calving (within 30 days following parturition). This
method of estimation is in wide use under practical conditions

due to the lack of cattle scales on commercial dairy farms (18L

-12-

13

Although there is a lack of data on the subject, estimates of
weight based on chest girth appear to provide suitable approx-
imations. Touchberry (h8) reported phenotypic and genetic
correlations of .81 and .88, respectively, between weight and
heart girth in Holstein cattle. Ragsdale and Brody (h5) and
Davis g£_gl. (11) reported a standard error of seven per cent
for weights estimated by chest girth measurements. On the
other hand, scale weights may show considerable variation
within a relatively short period of time. Lush g£_gl. (38)
found an "experimental error" of single scale weights in cat-
tle of six to twelve pounds. Johansson and Hildimmn (31) re—
ported a standard error of scale weights in cattle of one to

two per cent.

DISTRIBUTIONS AND MEANS
Tables 1 and 2 exhibit the distributions of age and weight,

respectively, in the samples studied.

1h

TABLE 1
DISTRIBUTION OF AGE AT FIRST CALVING

 

 

Age Group HoIsteig. guernggg £33§%%.
16-19 .5 2h .9 9 1.0 5
20-23 6.3 296 5.7 57 15.2 76
2h-27 3h.1 1597 no.3 h03 h9.7 2&9
28-31 31.2 lh59 29.h 29h 21.0 105
32-35 18.1 8R6 15.1 151 7.8 39
36-39 7.1 33h 6-1 61 3.8 19
uo-u3 2.0 92 2.0 20 1.2 6
hh-h7 .u 21 .2 2 0.0 o

h8 & above .2 8 .h h .h 2

 

TABLE 2

DISTRIBUTION OF WEIGHT AT FIRST CALVING

 

15

 

‘WSight Group Holstein §;323%%¥| Terese.
600-699 .1 h 1.2 12 17.8 89
700-799 .5 25 9.1 91 38.5 193
800-899 3.6 168 29.6 296 28.9 1&5
900-999 12.8 599 31.h 31h 9.3 A9
1000-1099 27.5 1286 20.5 205 3.u 17
1100-1199 26.9 1260 6.h 6h 1.0 5
1200-1299 17.8 833 1.7 17 .2 1
1300-1399 7.9 369 .2 2 .h 2
lhOO-lh99 2.1 98 --- O --- 0
1500-1599 .h 21 --- 0 --- 0
1600-1699 .3 1h --- 0 --- 0

 

Table 3 displays the means and standard deviations for

the four measurements studied.

coded units.

Milk, fat, and weight are in

16

TABLE 3
MEANS AND STANDARD DEVIATTONS
OF AGE AT CALVING, WEIGHT AT CALVING,
MILK AND FAT PRODUCTION FOR THE FIRST LAC TATI (ZN-z:-

 

 

Holstein 29 h.6 112 13.9 99 20.h 36 7.h
Guernsey 29 h.7 93 11.6 71 16.0 3A 7.6
Jersey- 27 h.2 79 12.0 62 13.5 32 6.9

 

* M : Mean, S sIStandard Deviation.

17

METHODS AND RESULTS

LEAST SQUARES ESTIMATES OF INFLUENCES OF AGE AND WEIGHT ON

 

PRODUCTION

 

There is a need to estimate the change in production ac-
companying a given change in the weight of the dairy cow.
From a management standpoint. information of this nature is
needed for the formation of sound decisions in such matters
as the level of development at which heifers should be bred
to achieve the most economical production during the first
lactation. Even if the genotype proves to be important in
the association of weight and production, the economic imp
portance of weight to milk production will have some bearing
on how weight is used in selection for milk production. Al-
though the analytical tools required for a study of this type
have been available for some time, the literature does not
disclose any estimates of this nature. The technical details
of the application of the method of least squares to non-
orthogonal data of the present nature have been well described
by Harvey (2h). The procedure of the present analysis has
closely followed that outlined by Harvey.

Since the effect of weight on production is closely as-
sociated with that of age, to obtain an estimate of the direct
effect of weight on production requires that effects of age
be analyzed also.

Let yijkl be the record made in the 1&3 herd by the 1&1

animal in the jig age group and the 23. weight group. If the

18

effects are additive and interaction is not an important source
of variation (an examination of the age-weight two-way tables
indicated such to be the case), the relationship between these
observations is yijkl 3,1 + h1 + a1 + wk + eijkl’ where}; is
the unknown population mean, hi is an effect common to all
observations in the iEE'herd, aJ is an effect for all records
in the jEll age group, wk is an effect for all observations in
the kEE'weight group, and eijkl is a random element peculiar

to the individual record.

The reduced normal equations were obtained with the above
model by the procedure described by Harvey (2h). For ease of
computation, age was grouped in four-month intervals and weight
in intervals of 100 pounds. The equations were solved simul-
taneously to estimate age and weight effects (see Tables h and
6, respectively). The constants are given in coded units and
are presented as deviations from the overall mean production
for mdlk and fat, respectively, for each breed (for the mean

production values, see Table 3).

TABLE h

INFLUENCE OF AGE OF FIRST‘CALVING

ON MILK AND FAT PRODUCTION IN THE FIRST LAC TATION

l9

 

 

 

 

FF Holstein Guernsey IUerseyA

éfiiup Ng§oég Milk Fat N8;033 Milk Fat N§;o§2 Milk Fat

16-19 2h -13.8 -5.h 9 -10.1 -h.8 5 -1o.5 - 5.9
20-23 296 - 7.5 -3.2 57 - u.1 -2.7 76 - 6.0 - 3.5
2u-27 1597 - u.a -2.3 h02 - 3.5 -1.9 2A9 - 3.u - 1.7
28-31 1u59 - 2.2 -l.2 29D - 1.7 -0.9 105 - 0.u - 0.2
32-35 8H6 0.2 -0.2 151 0.3 0.0 39 1.8 0.5
36-39 33h h.0 1.0 61 2.2 1.9 19 - 1.9 - 0.3
u0—h3 92 6.1 2.2 20 7.2 2.6 6 - s.u - 3.1
uu-u7 21 7.u 3.2 2 13.3 7.5 0 --- ---
hB-up 8 10.5 5.8 h - 3.6 -1.8 2 25.8 1b.?

The number of observations in many of the groups is

rather small.‘

The estimates for such classes must be used

with caution since sampling errors may be expected to be rather

large.

Thus the classes on both extremes, especially in Guern-

seys and Jerseys, are likely to be somewhat less reliable than
the central classes.

Although the age estimates were made primarily to separate
age effects from those of weight, some discussion of the mag-
nitude and behavior of the age effects seems appropriate. Such

effects, being independent of weight, may be interpreted as

representing the influence of the degree of maturity of body

20

functions upon production. In general, the age constants for
production exhibit a continuous positive direction with suc-
ceeding age groups.» The small numbers in the older age classes
for Jerseys and Guernseys may account for the exceptions noted
in those cases. The age constants for milk production are
plotted in Figure l. Departures from linearity are rather
large in the extreme classes for Jerseys and Guernseys; how-
ever, the approximation to a linear relationship is quite
striking in the Holstein example.

In Holsteins the largest increase in production between
two successive age classes occurs between the 18-22 and 22-26
month classes. In this instance delaying the calving of an
animal four months within this age range would be expected to
yield increases of 630 pounds of milk and 22 pounds of fat in
the first lactation. In Holsteins the increase in production
expected for animals calving in the median age group (28—31
months) as compared to those calving in the 16—19 month group,
is only 1170 pounds of milk and #3 pounds of fat. This is a
rather small increase in view of the 12 months delay in com-
mencement of productionrequired to obtain it. Similar ob-
servations may be made for the Jersey and Guernsey age con-
stants.

Since age correction factors are usually not derived in-
dependently of weight, the estimates in Table h can be com-
pared with those obtained when weight is not considered in
Table 5.

21

TABLE 5
INFLUENCE OF AGE ON MILK AND FAT PRODUCTION
IN THE FIRST LACTATTCN, DISREGARDING WEIGHT

 

 

[Es—Cioup MHOiSteigt Mguernsegt MiIEEEEFat
16-19 -17.3 -6.5 -13.6 -5.9 412.7 -6.2
20-23 -10.2 ~u.1 - 5.9 -3.5 - 7.0 -u.0
2h-27 - 6.1 -2.7 - h.6 -2.h - h.0 -2.0
28-31 - 2.h -l.3 - 2.3 -1.2 0.2 -0.1
32-35 0.7 0.0 0.5 0.0 2.2 0.7
36-39 5.2 1.h 2.8 2.1 - 0.6 0.0
hO-uB 7.5 2.6 8.1 2.9 - 3.7 -2.7
hh-h? 9.6 u.0 17.9 9.5 --- ---
h8 & above 12.9 6.6 - 3.0 -1.5 25.6 1h.l

 

The differences between the estimates of the effect of
age on production in Table 5 and the corresponding entries in
. Table u may be interpreted as due to the joint effects of age
and weight on production. The Joint contribution of age and
weight is generally quite small in the 28-31 month age group
but is progressively larger in the classes above and below
this group. The estimates for ages less than 28 months in
Table 5 are numerically smaller than the corresponding values
in Table h. This may be interpreted as due to the fact that
animals in these classes are smaller than average. Estimates

for ages greater than 32 months in Table 5 are numerically

larger than the corresponding entries in Table h.

This is

probably due to the fact that animals in these groups are

larger than the average.

is not considered, the estimates of the influence of age on

Table 5 indicates that when weight

production carry some correlated effects due to the animal

being smaller or larger than the average.

TABLE 6

INFLUENCE OF WEIGHT AT FIRST CALVING

ON MILK AND FAT PRODUCTION IN THE FIRST LAC TATION

22

 

 

 

 

 

Weight Nb. 1Eolstein No. iEuernsey N6. 1nJBrsel

Group Group Milk Fat Group, Milk Fat Groupg Milk Fat
600-699 A -7.6 -2.8 12 -7.u -3.u 89 -h.h -3.8
700-799 25 -7.0 ~2.9 91 -h.9 -2.h 193 -0.9 -2.2
800-899 168 -7.h -2.2 296' -0.u 0.1 1H5 1.0 -1.1
900-999 599 -3.2 -1.1 313 2.u 1.0 A9 1.8 -1.3
1000-1099 1286 -0.7 -0.2 205 1.8 0.7 17 3.6 -1.6
1100-1199 1260 1.7 0.6 6h 5.9 2.7 5 8.u h.9
1200-1299 833 11.3 1.5 17 2.5 1.3 l ~22.2 1.2
1300-1399 369 u.0 1.3 2 --- --- 2 :ULS' 3.9
1h00-1h99 98 7.h 2.9 0 --- --- 0 --- ---
1500-1599 21 2.8 l.u 0 --- --- 0 --- ---
1600-1699 1h 5.6 1.6 0 --- --- 0 --- ---

The significance of the weight constants (Table 6 and Fig-

ure 2) is somewhat more difficult to assess in view of there

being no continuous positive increase in production with

23

FIGURE I. INFLUENCE OF AGE AT FIRST CALVING ON FIRST
LACTATION MILK PRODUCTION

2500— /
2000 - ,'
bOO’

IOOO”

 

 

DEVIATIONS FROM MEAN MILK PRODUCTION

/.
“__ l l l l l l I

I6 I9 22. 25 28 3| 34 37 4‘0 4‘3 46 4‘9 52
AGE AT FIRST CALVING (MONTHSI

 

FIGURE 2. INFLUENCE OF WEIGHT AT FIRST CALVING ON FIRST
LACTATION MILK PRODUCTION

O
’T’

MILK PRODUCTION
8
9

8
‘3

 

MEAN
5
O
0

FROM
3
O
I

I ! LBGBNM

-———— HOLSTEIN
-— — — GUERNSEY
----- 'JERSEY

QJ J l l L l A 1 l l n

600 700 800 900 I000 I700 I200 I500 MOO I500 I600 I700

WEIGHT AT FIRST CALVING

I
N
O
O
O

l

 

 

DewAno
g
C)

2h

successive weight increments. Where the numbers in the clas-
ses concerned are small, this phenomenon may be attributed in
part to sampling error. However, in the case of the negative
increment of production associated with an increase from
1200-1299 pounds to 1300-1399 pounds in Holsteins, sampling
error would be expected to be small.

Close examination of the graphic portrayal of the con-
stants in Figure 2 yields two observations: (1) the departure
from a linear increase is not large over the range of 600-1000
pounds for Jerseys and Guernseys and from 800-1h00 pounds in
Holsteins, and (2) if the Jersey and Guernsey curves are
shifted 200 pounds to the right on the weight axis, the three
curves behave in a similar manner over the same general range
of increasing weight. More severe deviations of the Jersey
and Guernsey curves from the behavior of the Holstein curve
in Figure 2 may be attributed in part to smaller numbers in
the former breeds. In each breed, there is a consistent, al-
most linear increase in production over a wide range, then a
slight decrease, followed by an increase over a small range,
and finally, a rather sharp decrease.

The flatness of the Holstein curve between 600-900 pounds
indicates that weight has little or no effect on production
at this level of development. ‘While the estimates undoubtedly
contain some sampling error, there appears to be a type of
threshold at the 900-1000 pound weight level. This may indi-
cate a level at which growth needs no longer hold complete

priority over milk secretion requirements--above this level a

25

larger proportion of the nutrient intake may be channeled to
milk secretory usage. If more data were available, perhaps
the other breeds would exhibit similar reSponses at weights
less than 600 pounds.

The aberrations in the curves at higher levels of de-
velopment are more difficult to explain. The sharp decline
in production may indicate a level at which there is an in-
creasing tendency for the conversion of nutrients into fatty
tissue and storage. If such a condition interfered in some
way with milk secretion, one might expect to see a steady de-
cline in production beyond this weight level. Larger numbers
would need to be available in the extreme weight classes to
determine this. Certainly, one would not expect the produc-
tion to show a continuous increase with weight. At high lev-
els of weight, interference with foraging ability and a ten-
dency to sluggish activity might be expected.

,Based on the above observations, the following behavior
may generally be expected of the effects of weight on milk
production: first, a region of little or no response at rel-
atively low levels of development in size, then a rather con-
stant increase (approaching linearity) to a maximum level at
a weight of perhaps 300-h00 pounds above the breed average
for the age, and finally, a rather steady decrease in produc-
tion. Fat production would be expected to follow a similar
course.

A clearer picture of the importance of the independent

influences of age and weight on production may be obtained by

26

placing economic values upon them and contrasting these with
the cost of obtaining the added production. Tables 7 and 8
will serve to give a rough approximation of the economic worth
of the added milk production in the first lactation through
increased age and weight at the time of first calving (economic
values were placed only on the age constants independent of
weight). In Table 7, the left column lists the midpoints of
the two successive age classes between which a four-month in-
crease is made. For each breed, the first column shows the
value of the increment of production obtained by delaying age
at calving from the first age group to the second. The second
column under each breed shows the feed cost of maintenance of-
the animal over the four-month period required to obtain the

increased production.

TABLE 7
VALUE OF INCREASED MILK PRODUCTION IN THE FIRST LACTATION FROM OLDER
FRESHENING AGES, CONTRAS'IED TO MAINTENANCE COST INCURRED-3H!-

CEangeIn HolstEIn Guernse JErsey
Age at VaIue of FCCst of Value of Cost of Value of Cost of
Calving_ Add. Prod. Maint. Add. Prod. Maint. Add. Prod. Maint.

 

 

 

18-22 $25.83 $27.72 $25.01 $23.23 $18.h5 $21.65
22-26 11.07 27.72 2.05 2u.82 11.07 23.23
26-30 10.66 29.26 7.79 26.1h 12.30 2h.82
30-3u 9.8a 29.26 8.20 26.1u 9.02 2u.82
3u-38 15.58 29.26 7.79 26.1u lh.76* 26.1h
38-u2 8.61 29.26 20.50 26.1h 1h.35* 26.1u
h2-h6 5.33 29.26 25.01 27.72

u6-50 12.30 29.26 69.29* 27.72 128.33 52.27

 

* Value ofTProduction Lost
** Age independent of weight

TABLE 8

27

VALUE OF CHANGES IN MILK PRODUCTIOI~-.T IN THE FIRST LAC TAIION DUE TO

INCREASED WEIGHT'CONTRASIED IO COST OF GAIN AND MAINTENANCE

 

 

 

TCEangesiIn *Cost of—T Gest of* value Ef:Kdaed Prodfict
weight Gain Add. Maint. Holstein Guernsey Jersey
650-750 $7.77 Tu.70 I 2.u6 $10.25 9 1h.35
750-850 7.77 h.70 1.6a 18.h5 7.79
850-950 7.77 h.70 17.22 11.h8 3.28
950-1050 7.77 h.03 10.25 2.u6* . 7.38
1050-1150 7.77 h.03 9.8u 16.81 19.68
1150-1250 7.77 h.03 11.07 13.9u* 125.h6*
1250-1350 7.77 n.70 1.23* lh2.27
1350-lh50 7.77 h-70 13.9h
1h50-1550 7.77 h.03 18.86*
1550-1650 7.77 u.70 11.h8

 

* Value of Production Lost

Table 8 compares the value of the increased production

accompanying increased weight with the feed cost of obtaining

the increase in weight.

The left column gives the midpoints

of the two weight classes in which a 100-pound increase oc-

curs.

pound gain in weight.

The second column lists the cost of obtaining this 100

The third column presents the cost of

maintaining the 100 pounds additional weight for a 305 day

lactation period.

Under each.breed is listed the value of the

difference in production between the two successive weight

classes.

28

Both Tables 7 and 8 were computed using the 1956 average
price received by Michigan farmers for fluid milk ($h.10) and
the 1956 average price paid by Michigan farmers per ton of
alfalfa hay ($21.75). (1) Maintenance and growth require-
ments were determined from the standards of Morrison (hl) and
Knott g£_gl. (36), respectively.

From Table 7 it is seen that in all but two cases, the
maintenance cost exceeds the value of the added product from
the four-months delay in first calving. In Table 8, while '
there is some variation, the value of increased production due
to 100 pounds larger weight is generally slightly less than
the feed cost for the gain and maintenance of the added weight.
Thus, it appears that the production gained through older age
at calving is uneconomical, while that obtained through increased
weight at calving is slightly unprofitable. These comparisons
are intended to provide only a rough idea of the economy of
the increase in production due to increased age and weight at
calving. Many factors have not been considered, of course,
so in actual practice the costs involved are expected to be
greater than is shown here.

In making use of Tables h-8 and Figures 1 and 2 for other
groups of animals, these estimates of the effects of age and
weight should be applied as deviations from the average actual
production of the group in which they are to be used. Of
course, it is necessary to integrate the separate effects of
age and weight in actual application. Time is required to ob-

tain increased weight and, conversely, weight increases will

29

usually accompany increases in age over this range. The prob-
lem is one of determining the most advantageous level of age

and weight in regard to economical first lactation production.

ANALYSIS OF VARIANCE AND COVARIANCE

Before the association between weight and production can
be neasured, the variation and covariation of age, weight, and
production must be determined.

Let yijk signify the first lactation record made in the
iEE herd by the k32 daughter of the 359 sire. Then, the linear
model is

yijk ’7‘” hi "’ hs11+ pijk’
where/“is common to all observations, hi is a characteristic
of all records in the iEE.herd, hsij is common to all obser-
vations of daughters of the 332 sire in the 133 herd, and pijk

is peculiar to the record of the kth daughter of the 132 sire

in the iEE herd.
Table 9 lists the total number of observations and the

number of herds and sires within herds for each breed.

TABLE 9
NUMBER OF OBSERVATIONS

 

 

Breed Total IHerds Sires Within Herds
Holstein N677 651 2&51
Guernsey 1001 163 500

Jersey 501 82 2&3

 

30

Henderson (26) has described a method of obtaining com-
ponents of variance and covariance necessary for estimates of
genetic parameters in non-orthogonal systems (method 1). Ex-
cept for the constant/u, all elements in the model are uncor-
related variables with zeroIneans and variances H, HS, and P.

The results of the variance and covariance component analyses

are presented in Tables 10 and 11, respectively.

TABLE 10

COMPONENTS OF VARIANCE FOR AGE, WEICHT, MILK, AND FAT‘x'

 

 

 

 

 

 

 

BFBed _Component A e Weightfl Milk_g F—Fat
mfi camp. 3:, firm. 3 fimp. i
H 7.1 33 59.6 31 lhu.6 35 ' 21.6 39
HS 2.1 10 22.u 12 33.7 8 3.3
Holstein
P 12.u 57 110.8 57 236.6 57 30.3 55
Total 21.6 192.8 hlA.9 5522
H 12.1 h? 3h.0 25 100.0 39 2h.7 h3
Hs - 3.LI*“"' o 9.3 7 12.5 .5 2.8
Guernsey
P 13.5 53 91.9 68 luh.9 56 30.6 52
Total 22.2 135.2 257.h (58.1
H h.3 2h 50.7 35 61.1 33 16.6 35
RS .9 5 18.8 13 15.u 8 u.9 10
Jersey
P 12.9 71 75.9 52 107.5 59 26.0 55
T0133]. 18.1 1645.1} 181.100 “105

 

* Weight, Milk, and Fat in coded units.
** Negative HS component considered zero.

The estimation of the components of variance is a means

of apportioning the total variance among a group of contribu-

ting elements.

The H component represents the variance due

31
to differences between herds. These variations are thought
to be principally environmental in nature (28). The HS com-
ponent can be interpreted as portraying the variation between
sets of daughters by different sires in the same herd; this
would appear to be largely genetic, unless progeny groups
within a herd were treated differently. The P component is

an estimate of the variation between daughters of the same

sire within the herd.

TABLE 11
COMPONENTS OF COVARIANCE BETWEEN AGE, WEIGHT5 AND MILK AND FAT*

 

 

 

 

Age- weight- ‘Age- WeIght- Age-
Breed Component Weight Milk Milk Fat Fat
Covh Sch 16700 0.0 170“ 003
Covhs 5.1 9.1 2.7 3.0 0.3
Holstein
Covp 11.h 31.7 11.9 11.h 5.0
Total 21.9 87L8 14.6 31.8 5.6
Covh - 0.h 2h.6 - 6.7 19.8 - 2.7
Covhs 5.2 201 1400/ -2209 205
GuernSey
Covp 10.2 29.2 8.9 31.5 h.2
Total 15.0 55.9 7.1 28.h_g h.0
COVh '25.]. 39e3 -2901 “02 - 305
COVh 93el -5108 7902 - 0e 6 0.2
3
Jersey
COVp -5501 59. O ”’4506 110 0 he 6
TOtal 12e9 “6e; Ales 11.1.6 103

 

* weight, Milk, and Fat in coded units.

Age Components. Considerable between-breed variation is

 

observed in the components of age variation. The differences

32

between herds contribute only 2h% in Jerseys but account for
h7% in Guernseys. Although the HS is small, this component
is a larger part of the variance in age than was expected.
One might suppose that differences in age at first calving
would vary little from a group of daughters by one sire to a
group by another sire in the same herd. However, the results
estimate HS to be 0-10% of the total variation. To the ex-
tent that level of development of the heifers is a considera-
tion in the age at which heifers are bred, this phenomenon
may reflect genetic differences in rate of growth. Never-
theless, environmental differences undoubtedly are of prin-
cipal importance in determining the age at first calving.

The size of HS may indicate that breeders are more interested
in the daughters of one sire than those of another and are
breeding them at different times. Another possibility is
that breeders may be using different sires at different times
of the year. Breeding for fall freshening is a common prac-
tice and may cause sire progenies to differ in age at first
calving.

Weight Components. Differences among herds again are
quite important contributors to the variation in weight at
first calving, the proportion of H ranging from 25-35%. Gen-
etic differences as portrayed by HS are also large enough to
merit consideration (see heritability estimates).

Milk and Fat Components. The H component indicates that
differences between herds account for 33-h3% of the variation

in production. The contribution from differences between

33

sires within herds is rather stable, ranging from 5-8% of the
total variation in milk and 5-10% of the total variance in fat.
The total variation in milk for Holsteins is much greater than
that of the other breeds, yet there is little breed difference
in the total variation in fat. The larger average milk pro-

duction for Holsteins may be associated with this phenomenon.

HERIIABILITIES

 

Estimates of the heritability of the measurements of weight
and production are useful to evaluate the genotypic contribu-
tion to observed differences. These values may be obtained

by the following ratio from Table 10: H—T—%§§T_P° The her-

itabilities are presented in Table 12.

 

 

TABLE 12
HERIIABILIIIES
Breedi Weight MilET Fa?
Holstein .h6 .32 .2h
Guernsey .28 .20 .19
Jersey .52 .38 .hl

 

The estimates of the heritability of weight at first
calving range from .28 to .52, i.e. differences in this meas-
urement appear to be transmitted to a relatively high degree.
Most of the previous estimates of the heritability of weight,
while they were estimated for various stages of development,

fall within the range of the present estimates. Johansson (29)

3h

reported a heritability of weight at first calving of .3h for
Red Dane heifers tested in special progeny testing stations.
Nbson g£_gl. (MO) reported heritabilities of weight in Red
Dane heifers at two different stages of development. The
heritability of September weight (prior to parturition) was
.hl, while the heritability of March weight (after five months'
lactation) was .37. Touchberry (u8) found a heritability of
.37 for weight in 187 Holstein daughter-dam comparisons, where
the measurements were obtained at the calving date nearest

the animal's third birthday. Turner (50) reported an intra-
sire regression of daughter's body weight on dam's body weight
of .28 in Jerseys. Turner's weight data were obtained from
animals of various ages but were corrected to a "mature-
equivalent" weight basis.

Most heritability estimates for milk and fat production
have been on the order of .2 to .3 (MT). The range in the
present study is from .2 to .h. However, most previous esti-
mates were based on lactations at various stages of life,
whereas these estimates apply specifically to first lacta-
tions. Although the data are usually corrected for age dif-
ferences, such estimates need not correspond precisely to those
obtained in the present example.

Johansson (30) reported a heritability of BOO-day first
lactation.fat yield of .33 for Swedish Red and White cattle.
In an analysis of the records of Red Dane heifers tested in

progeny stations in Denmark, Johansson (29) found heritabilities

35

of 250-day first lactation milk and fat yield of .58 and .56,
respectively. Mason 33 El. (NO), in a similar analysis of
progeny test station records, reported a heritability of first
lactation milk yield of .57. The heritability of fat of .33
found by Johansson agrees generally with the present results,
but the values reported from progeny testing station analyses
are larger than the present findings. However, the phenotypic
variation seems to be reduced when animals are tested under
similar environmental conditions, as in the Danish testing
stations. The ratio between the genetic variance and the
total variance would be increased, thus yielding estimates

of heritability larger than estimates based upon records made

under field conditions (29).

CORRELATION ANALYSIS

 

The extent 0f the association between two measurements
in the same animal is given.by the phenotypic correlation.
Table 13 presents the overall phenotypic correlations found

in the present study.

TABLE 13
OVERALL PHENOTYPIC CORRELATIONS BETWEEN AGE, WEIGHT, MILK, AND FAT

 

 

 

 

 

Holstein Guernsey Jersey
Trait Weigfit Milk Fat’ WéIghE' Milk TFat Weight Milk Fat
Ago 03,-} .15 .16 .28 .10 e11 025 .08 .0”.
Weight "" .3]. e31 ""' e30 e32 "" e28 018

 

The phenotypic correlations between weight and production

are all of the order of .3, except for the correlation of fat

36
and weight in Jerseys. Previous estimates by various workers
have ranged widely, but most seem to have been of the order
of .h (see previous discussion). The correlation between age
and weight appears to be about .3, while the correlation of
age and production appears to be around .1. The extent of
the correlation between age and weight demonstrates that, had
the production data been corrected for age differences, the
variation in production due to age~and weight jointly would
also have been removed. I

The within-herd phenotypic correlation measures the de-
' gree of association present in the herd group. Any associa-
tion due to herd differences is removed, thus permitting a

clearer view of the fundamental relationships (see Table 1h).

TABLE 1h
PHENOTYPIC CORRELATIONS
BETWEEN AGE, WEIGHT, MILK, AND FAT WITHIN HERDS

 

 

 

 

 

Holstein Guernsey. JEraey
Trait weight Milk *Fat’ Height Milk Fat Wéight Milk Fat
Age .36 .23 .2h .h3 .30 .32 .28 .2h .2h
Weight '"""’ .2]. e21 "‘"' .25 .26 --- .22 .21

 

The phenotypic correlation between age and weight appears
to be slightly larger within the herd group than over a group
of many herds. 0n the other hand, the correlation between
weight and production within the herd is less than on an over-
all basis, the association within herds being on the order of

.2. The correlation of age and production is distinctly larger

37

when the differences between herds are removed. The intra-
herd correlation is of the order of .2 to .3, whereas the
correlation between age and production is only .1 on an over-
all basis.

In view of the effects apparently being exerted by herd
differences, an estimate of the correlation between weight and
production due to these effects is needed. Since the princi-
pal differences between herds are thought to be environmental
(28), this correlation might be termed the "environmental”
correlation. Estimates of this association may be obtained
by employing the between-herd components of variance and co-
variance previously obtained (see Tables 10 and 11). Table
15 presents the estimates of the association due to the in-

fluence of herd differences.

TABLE 15

CORRELATIONS BETWEEN WEIGHT AND PRODUCTION OF HERDS

 

 

Breedf MiIk 'TFat
Holstein .51 .h8
Guernsey .h2 .68
Jersey .71 .18

 

Except for the correlation between weight and fat pro-
duction in Jerseys, the correlations between weight and pro-
duction due to herd differences are quite large--of the order
of .5. Thus, it appear that there are factors peculiar to
different herds which influence weight and production in a

positive direction.

38

The importance of herd differences in the relationship
between weight and production has been investigated by Hofmeyr
(27). In a study of Red Dane heifers from 7&8 different herds,
Hofmeyr classified the herds from.which the animals came into
low, medium, and high groups, according to the contemporary
herd average production. He found that the phenotypic cor-
relation between weight at calving and first lactation yield
of h% FCM declined from the low to high herd levels. The cor-
relation between weight and yield was .33 for low herds, .27
for medium herds, and .17 for high herds. He concluded that
weight at calving in relation to age is especially important
in regard to heifers from low-yielding herds and that the level
of nutrition prior to calving has a pronounced influence on
lactation yield.

The preceding measures of correlation have ignored the
common effects of age and weight upon production. The partial
correlation coefficients listed in Thble 16 are estimates of
the association between weight and production independent of
age, and of age and production independent of weight. The

correlations are presented on overall and intraherd bases.

TABLE 16
PARTIAL CORRELATIONS BETWEEN AGE, WEIGHT, MILK, AND FAT

 

r'13.2 P117.2 I‘23.1 T52u.1
Breed T H T H T H T H

 

Holstein .05 .17 .06 .18 .28 .1h .27 .18
Guernsey .01 .22 .05 .2u .29 .1h .30 .1h

Jersey .01 .19 .00 019 027 017 017 015

1 : age, 2 3 weight, 3 : milk, h : fat; Ti: overall basis,
H : intraherd.

 

39

'When many herds are considered,the correlation of age with
production is close to zero when weight is held constant. Thus,
on the aggregate basis, age has little or no relationship to
production independent of weight. The correlation of weight
and production with constant age appears to be of the order
of .3. Comparison of the overall partial correlations with
the phenotypic correlations in Table 13 reveals that the par-
tial correlations of weight and production are about the same
as the overall phenotypic correlations. Thus, when differences
between a large number of herds are ignored, the association
between weight and production is largely independent of age.

The overall partial correlations are similar to those of
earlier studies of records from many different herds. Gowen
(21). in a study of Jersey ROM 365-day records of cows of vari-
ous ages, reported partial correlations of .23 and .22 for
weight with milk and fat, respectively. Turner (M9) found a
correlation of .25 between age and fat production when weight
was held constant in an analysis of 2700 AR Guernsey records
of animals three to four years of age. In addition, Turner
reported a partial correlation of .25 between weight and fat
production. Hofmeyr (27) computed partial correlations of age
and weight with production at three different levels of herd
production averages, as well as on an overall basis, in a
study of the records of 93h Red Dane heifers from 7&8 herds.
The partial correlations between age and yield of u% FCM were
.12, .08, .01, and .08, for low, medium, and high herds, and

the total basis, respectively. Thus, the association of age

no

and production adjusted for weight was small in any case, but
especially so where the herd conditions were relatively supe-
rior. The association of weight with production adjusted for
age also declined with increasirg level of herd environment.
Weight was only one-half as closely associated with production
in the highest herds as for the lowest group of herds. The
findings of the above workers agree well with the present re-
sults, with the exception of Turner's estimate of .25 for the
partial correlation of age and fat production. The two re-
sults need not correspond due to differences in the data em-
ployed, but this may not explain all of the disagreement.

The intraherd partial correlations in Tuble 16 indicate
that age is slightly more closely associated with production
than is weight. A comparison of overall and.intraherd partial
correlations shows that the association between age and pro-
duction is much larger when herd differences are removed. The
association between weight and production appears to be only
about one-half as large within.tte herd as in a group of many
herds. A comparison of the intraherd partial correlations
with the within-herds phenotypic correlations (Thble 1h) in-
dicates that the correlations of age and weight with produc-
tion are reduced by a similar amount by holding the remaining
factor constant.

The preceding discussions have contained estimates of the
phenotypic, partial, and "environmental" correlations of weight
and production. The role of genotypic differences as a common

source of variation in weight and production is as yet

bl

undetermined in these data. Hazel g£_§l, (25) have advanced
a method of obtaining estimates of this relationship. Accord-

ing to unese authors, the genetic correlation between traits

: Cov Gng, where Cov GlGZ’ SGl' and 3G2 are
G1 G2

the genetic covariance and standard deviations, reapectively.

1 and 2 is r
G1G2

Lush (37) has discussed the interpretation of genetic corre-
lations. Such a correlation may be thought of as measuring
the degree to which genes controlling one trait also contri-
bute to the expression of the other measurement in question.
The most important cause of genic associations between traits
is pleiotropy, i.e. genes which influence one trait also in-
fluence the other. Linkage (39) and variable selection pres-
sure in isolated sub-groups may account for minor genetic as-
sociations between traits. Lush concluded that negative pleio-
tr0pic effects are expected to be much more frequent than posi-
tive ones.

As an estimate of genetic variance and covariance, the
components between sires within herds may be employed. The
genetic variance and covariance are u HS and h Covhs. Table

17 presents the genetic correlations between weight and pro-

 

 

duction.
TABLE 17
GENETIC CORRELATIONS BETWEEN WEIGHT AND PRODUCTION
Bieed MilE’ Fat
Holstein .33 .35
Guernsey .19 -h.h9

Jersey —3.05 -0.20

 

142

Estimates outside normal limits are obtained for weight
and milk in Jerseys and for weight and fat in Guernseys
(~15r911). Reference to Teble 11 shows that the estimates of
Covha in Jerseys, and to a lesser extent in Guernseys, are
rather erratic, with large positive and negative values occur-
ring. This is the source of the aberrant behavior of the gene-
tic correlations. Since the covariance estimates are more
stable the larger the number in the breed sample, small num-
bers may be a source of inaccuracy. The results may reflect a
a lack of resolving power in the method of estimation of the
variance and covariance components where the numbers involved
are comparatively small (the summed components agree closely
with the computed total variances and covariances).

Little can‘be said of the other Guernsey and Jersey es-
timates which are not invalid by inspection, in view of the
accompanying disturbances. How much confidence can be placed
upon the estimates of genetic correlation.for Holsteins is
questionable. The degrees of freedom for Holsteins exceed
those for Jerseys and Guernseys by factors of 9 and u, re-
spectively, so the Holstein estimates should at least pro-
vide a general picture of the parameters involved. As ex-

pected, the correlations for milk and fat agree rather closely.

 

h3

. TABLE 18
A COMPARISON OF VARIOUS ESTIMATES OF THE GENETIC CORRELATION

BE TWEEN WEI GH T AND PR ODUC TI 01?

 

 

Source Breed Milk Fat
Touchberry (N8) Holstein -.53 .2h
Mason et al. (NC)

a. IniIIaI—weight Red Dane .25 ---
b. March weight Red Dane .02 ---
Blackmore (h) Holstein -.02 ---
Present Study Holstein .33 .35

 

The present results for Holsteins are compared with those
of other workers in Thble 18. The estimate for milk agrees
generally with that of Mason g£_gl. for initial weight (weight
prior to first calving), while that for fat is of the general
order of Touchberry's result. The data used by Touchberry
and Blackmore were corrected.for age differences. Tbuchberry
used the laCtation record nearest the third birthday, while
Blackmore employed the lifetime record of 3.5% FCM. The data
employed by Mason.gt_gl. were obtained from first lactation
records, so this study is more comparable to the present analy-
sis than those of Touchberry and Blackmore.

The positive results of Touchberry and Mason g£_gl,,
coupled with the present findings, point to a positive gene-
tic relationship between weight and production, at least for
the first lactation. How may such results be interpreted?

As mentioned previously, most of any observed genetic

MI

correlation must be the product of the manifold effects of
genes. ‘Where phenotypic differences in two traits are markedly
controlled by genotypic differences and the number of genes
affecting each trait is large, some pleiotropy may well exist.
Milk production and body weight apparently fit these require-
ments. Intraherd heritabilities of production and weight have
been found to be on the order of .2 to .3 and .5, respectively
(see previous discussion). Though the evidence is scant, both
production (A7) and weight (h2) appear to be controlled by a
large number of genetic factors. However, a factor affecting
two different traits might well have opposite effects on the
two scales of measurement. Lush (37) believes that this is
the more common association.

If the genetic correlation between weight and production
is as large as .3 and if the heritabilities 0f production
and weight are as large as have been estimated, it would seem
that selection for one would achieve visible gain in the other
as well. Concurrent selection for milk production and size
(as reflected by beef characteristics) is an important feature
of improvement programs in the dual-purpose breeds of cattle.
Blackmore (h) concluded that simultaneous selection for milk
production and meat-type characteristics would be impractical.

Information concerning the amount of genetic progress made
by simultaneous selection for weight and milk production would
be of value in further study of the problem. On the commercial
basis, it is doubtful if size has received consistent degrees

of emphasis in selection programs, (with the possible exception

145

of the dual-purpose breeds). Vacillating degrees of impor-
tance attached to weight, therefore, would appear to preclude
obtaining much useful information on the degree of progress
which has been made by simultaneous selection for the two

traits among commercial dairy herds.

MULTIPLE REGRESSION ANALYSIS

 

The method of multiple linear regression has been em-
ployed in previous efforts to determine how production changes
with weight, independent of age.

The exponential relationship, Y = a Wb, suggested by Brody
to be more appropriate than a linear relationship, has been
tested by Gaines (15) and by Bailey and Broster (2). From data
on 11 Holstein cows Gaines reported regression coefficients
of production on weight on the logarithmic grid ranging from
.28 to l.h9, depending on the stage of lactation at which
weight was measured. Bailey and Broster found an exponential
value of .82 3 .22 in data from 99 first lactations of Dairy
Shorthorn cattle. They concluded that the deviation from
unity was not serious and that a linear function could ade-
quately express the relationship. Turner (50) tested the
linearity of the regression of fat production on weight in
Guernsey cattle of three to four years of age by a test in-
volving the correlation ratio (1p. He concluded that the de-
parture from linearity was not great. These results indicate
that there is some basis for the assumption of a linear rela-

tionship.

M6

Multiple regression equations were calculated for both
milk and fat. Partial regression coefficients (b) of produc-
tion on age and weight, standard partial regression coeffi-
cients (b'), standard errors of the partial regression coeffi-
cients, and standard errors of estimate are presented in
Table 19 for the overall basis.

Age and weight appear to be rather poor predictors of
production. In milk for Holsteins, for example, the standard
error of estimate is almost 2000 pounds, while the standard
deviation of milk production is only slightly more than 2000
pounds. Age appears to be a rather unimportant source of
variation in production, since the age coefficients are not

significantly different from zero for Jerseys and Guernseys.

TABLE 19
PARTIAL REGRESSIONS OF MILK AND FAT PRODUCTION ON AGE AND WEIGHT

 

 

 

 

 

BFeed ITonsfarfi SEE Partial CoefficIents Ttan. Part. Coefi‘:
(a) by1.2 SET by2.1 SE» b'y1.2 b'y2.1
A. Milk
Holstein hh.65 19.36 ' .2h .06 .h3 .02 .06 .29
Guernsey 31.81 15.25 .05 .11 .hl .0h .01 .29
Jersey 35.96 12.98 .03 .1h .32 .05 .01 .28
B. Fat
Holstein 15.70 7.05 .10 .02 .15 .01 .06 .29
Guernsey 13.55 7.22 .0h .05 .20 .02 .03 .31
Jersey 2h.27 7.2h .00 .08 .10 .02 .00 .18

 

1 : age,T2 - weight, y : preduction, SEE : standard error of
estimate, SE : standard error of regression coefficient.

h?

In every case the weight coefficient is much larger than the
age coefficient. The standard partial regression coefficients
give a better comparison of the importance of age and weight
since these coefficients measure the part of the standard de-
viation of the dependent variable contributed by the independ-
ent variable in question. From these entries in Table 19,
weight is seen to be a more important influence on individual
production records over a group of many herds than is age.

The variation upon which the estimates above are based
contains a large amount of herd differences. Although no
practical method of analyzing these herd variations is avail-
able, environmental characteristics peculiar to individual
herds may be expected to be the chief components. To obtain
a clearer picture of any existing relationships between weight
and production, these environmental fluctuations were elimin-
ated by computing regressions within herds. Most practical
applications are made within herds.

Within herds the relative contributions of age and weight
are virtually the reverse of the overall situation. Age is
now a slightly more important factor in production than is
weight. A comparison of the overall and intraherd partial
coefficients shows that weight is a smaller source of vari-
ation in production within herds than of overall variation.
Two observations may be made from these results: (1) the re-
moval of differences between herds (chiefly environmental)
also removes much of the variation in production associated

with weight, and (2) variation in production due to differences

u8

TABLE 20
WITHIN-HERBS PARTIAL AND STANDARD PARTIAL REGRESSIONS
OF PRODUCTION ON AGE AND WEIGHT

 

 

 

 

Predicted? Part. Coeff. Stan. Part._Coef?T
Quantity Breed by1.2 by2.1 b'y1.2 bfy2.1
Milk Holstein .76 .21 .18 .15

" Guernsey .90 .18 .2u .15

" Jersey .57 .20 .19 .17

Fat Holstein .30 .07 .19 .1h

" Guernsey .h5 .08 .26 .1h

" Jersey .29 .09 .20 .15

 

1 = age,’2 = weight, y : production.

in age at first calving is much more important relative to
weight within herds than on an overall basis. The latter ob-
servation is due in part to a decrease in the total variation,
however, all of the increase cannot be attributed to this re-
duction. There is apparently a tendency for variation due to
age to cancel out when the influence of differences between
herds (probably largely environmental) is included.

Several previous studies in multiple regression are avail-
able. Bailey and Broster (2), in an intraherd study of 99
first lactation records of Dairy Shorthorns, found partial re-
gressions of h% FCM (tens of pounds) on weight (pounds) and
age of .51 and -.11, respectively. Gaines gt_gl. (17), in an
investigation of 231 lactations of 57 Holstein cows, reported
a within-herd, within-cow partial regression of .Oh, where the

dependent variable was FCM in pounds and weight (pounds) was

in

recorded within the first month following parturition. In
earlier work, Gaines 2£.2l° (18) analyzed 199 Holstein and lhO
Jersey eight-month lactation records made in Illinois DHIA
herds, for which age and estimated body weights were avail-
able (weight was estimated by taped chest girth measurements).
The regressions of FCM in pounds on weight in thousands of
pounds were 13 and 26 for Holsteins and Jerseys, respectively.
The corresponding age partial regressions were .96 and .37.
Johansson (29) found partial regressions of first lactation
butterfat yield (kg.) on weight (kg.) and age at calving (days)
of .1M and .13, respectively, in an analysis of records of Red
Dane heifers in Danish progeny stations.

The work of Bailey and Broster is most comparable to the
present study. A comparison shows that their regression of
milk on weight is more than twice as large as the present
intraherd regressions. However, their data were in terms of
FCM and were collected over a longer time interval than in the
present case.

Tests of the significance of the reduction in within-herd
variation due to fitting regression coefficients for age and
weight are presented in Table 21. The reductions in residual
variation due to fitting age and weight simultaneously, due
to fitting weight alone, and due to fitting both age and weight
as compared to weight alone, are shown. It is evident that
the fitting of partial regression coefficients for age in ad-
dition to weight reduces the within-herd variation to a sig-

nificant degree, as compared to fitting weight alone. Again,

50

this is in contrast to the overall situation where the fitting

of partial regression coefficients to age was of little value.

ANALYSIS OF VARIANCE OF

TABLE 21

REGRESSION WITHIN HERDS

 

 

 

 

 

Breed Source BF Milk Fat
Mean Square Mean Square
Due Tb Fitting Age & Weight 2 37502** 5016**
Due To Fitting Weight 1 h6391** 5807**
Holstein ,.
Due To Fitting Age After Weight 1 2861u*“ u226**
Residual 402} 2h} 30
Due To Fitting Age & Weight 2 7117** l6u2**
Due To Fitting Weight 1 8008** 1783**
Guernsey " '
Due To Fitting Age After Weight 1 6226w% 1500**
Residual 835 138 29
Due To Fitting Age And Weight 2 2082** h80**
Due To Fitting Weight 1 2h62** 528**
Jersey v ’
Due To Fitting Age After Weight 1 1703*" u32**
Residual Q16 109 27

 

** P<.OI.

51

DISCUSSION AND CONCLUSIONS

The influence of weight upon production, the factors con-
trolling such influences, and the importance which should be
attached to body weight in the management and selection of
dairy cattle can be evaluated partially from the information

obtained in previous sections.

IMPLICATIONS IN DAIRY CATTLE MANAGEMENT

We have two pictures of the quantitative influence of weight
on production in the first lactation. First, partial regres-
sions and phenotypic correlations indicated that the influence
of weight on first lactation records over the many herds stud-
ied is much larger than that of age. Weight accounted for ap-
proximately 8% of the variation between first lactation pro-
duction records, while age was not a measurable source of vari-
ation. Partial correlations indicated that age was not impore
tant in the influence of weight on production when differences
between herds were ignored. Secondly, the importance of age
and weight to intraherd production estimated by least squares
and partial regression indicated that larger heifers had lit-
tle or no economic advantage in the first lactation. Weight
apparently should receive only 50% as much emphasis in pre-
dicting intraherd production as for predicting production in
a group of many herds. Only 2% of the intraherd variation fin

production was directly attributable to weight.

52

In practice, the separate pictures of the importance of
age and weight must be integrated since they are entangled in
nature. In application, when we consider age, we are also
considering weight to some extent, and vice versa.

The quantitative influence of weight upon first lactation
production is not large enough to show economic advantage when
contrasted to the feed cost of obtaining the weight gain and
the additional cost of maintaining the heavier animal. There
is an additional point to be considered in that larger ani-
mals may have more economy of scale, i.e. fewer animals re-
quired to produce a given amount of milk, fewer replacements
needed, etc.

The primary consideration of body weight in management
would seem to be to develop dairy heifers as rapidly as pos-
sible to calve them at least at an earlier age than is the
case in the sample studied. Since these samples have been
taken from the more progressive herds in Michigan, there well
may have been a tendency to withhold breeding in order to ob-
tain larger first lactation records. In such situations, the
marginal value of the increased milk production obtained is
greater than has been shown here due to the increased value
placed on breeding stock by higher production. Nevertheless,
it seems that commercial dairymen need to reconsider the level
of develOpment at which heifers are commonly bred.

The results indicate that a dairy heifer should be bred
as soon as sufficient body development is reached that birth

complications, interference with growth, and unfavorable effects

53

on later lactations will be precluded. There may be some need
to place more emphasis on size rather than on age, since the
latter appears to decrease the economy of first lactation pro-
duction.

The early entrance of heifers into productive life is
Justifiable on grounds other than the economy of production.
One of the chief obstacles to rapid genetic improvement of
dairy cattle is the length of the generation interval. The
earlier animals are production tested, the more rapidly their
genetic worth can be ascertained, as well as that of their
parents. For example, the average age at calving in the Hol-
stein sample was 29 months. This would mean that the worth
of a sire could not be evaluated until the youngest member of
the progeny group was 39 months of age if completed records
are used and the average conditions hold. If the average age
at calving were 23 months, the sire proof could be obtained
6 months earlier. It seems reasonable to assume that some
time gain can be made in progeny testing through earlier calv-
ing of heifers without hindering the economy of first lacta-
tion production.

The conclusion reached in the preceding discussion finds
support in earlier work. TUrner (A9) concluded that early
breeding and a rapid succession of pregnancies are necessary
for the most rapid gland development and the most economical
production, since the greater production of large cows only
slightly exceeds the cost of obtaining the additional product.

He further concluded that the sires whose daughters are above

514

the average for the breed in fat production without exceeding
the average in body weight are especially desirable because
their daughters are increasing the economy of fat production
in the breed.

The work of Reid 33 31. (M6) with different levels of nu-
trition in heifers adds further strength to the conclusions of
the present study. In a lifetime experiment three groups of
Holstein heifers were raised from birth to first calving on
low, medium, and high planes of nutrition. After the date of
first calving, they were placed on comparable normal rations.
The low-plane group averaged over 200 pounds smaller at first
calving than either of the better-fed groups. However, the
low—nutrition group produced only slightly less than the other
groups during the first lactation. Moreover, in later lacta-
tions, the low-plane heifers equalled and then excelled the
production of the larger, better-fed heifers. Thus, there is
an indication that heifers calving at comparatively low levels
of develOpment produce Just as well as those calving at high
levels of development.

Gethin (19) has reviewed the relation of age at first
calving to later production. He concluded that earlier first
calvers are more economical producers than later calvers.
Gethin further concluded that 2h months is the minimum age at
first calving to avoid growth complications.

The effect of earlier breeding and lower levels of devel-
opment at the time of breeding upon the length of the produc-

tive life is an area where future study is needed. No final

55

conclusions on standards for the proper level of deve10pment
at which heifers should be bred can be made until the effect
upon longevity is ascertained. However, from the standpoint
of production in the first lactation, the breeding of heifers
at slightly younger ages and lower levels of development than

is the common practice appears justifiable.

IMPLICATIONS IN DAIRY CATTLE SELECTION

Comparison of overall and intraherd regressions and cor-
relations indicated that differences between herds made sig-
nificant contributions to the association of weight and pro-
duction. Conversely, differences between herds appeared to
neutralize the influence of age on production since age was
far more important within the herd than on an aggregate basis.

The role of differences between herds in the overall con-

tribution of weight to production was measured by the "

en-
vironmental" correlation between weight and production. With
the single exception of the association between weight and fat
production in Jerseys, these correlations were all of the order
of .5. This indicated that herd differences, which are pre—
sumably chiefly environmental in nature, are the most impor-
tant causes of the association of weight and production.
Hofmeyr's work (27) indicates that the nature of differences
between herds may be such that smaller heifers have a much

better opportunity to reach their productive potential in the

higher producing herds, as compared to the poorer herds.

56

To ascertain the role of the genotype in the relationship
between weight and production, genetic correlations were com-
puted. Although irregular results were obtained for Jerseys
and Guernseys, probably because of small numbers, the corre-
lations for Holsteins of .3 for both milk and fat indicate that
the genotype may make an important contribution to correlated
changes in weight and milk production.

Some of the difficulty in evaluating the nature of the
factors controlling the association between weight and produc-
tion lies in how we choose to view the mechanisms through which
the association is brought about. Speaking of weight as a
"source" of variation in production may be improper as the
word source may be a misnomer. Milk production is usually
pictured as a result of the forces of heredity and environment
and the interplay between the two. Thus it would seem more
nearly correct to view associations of weight with production
as being an expression of some function of heredity and en-
vironment. weight, as such, may have little to do with mdlk
production. However, an association between weight and pro-
duction may be produced by environmental and hereditary factors
which are reflected by both measurements. Thus, to learn the
nature of the relationship, a partition of the weight-produc-
tion association between hereditary and environmental sources
seems to be necessary.

The most important factors affecting the relationship be-
tween weight and production appear to be those which are ex-

pressed through differences among herds. These are thought

57

to be chiefly environmental in nature, such as herd to herd
variations in level of nutrition. The work of Reid gt_§l.

(M6) indicates that these differences act upon both weight

and production in a similar manner, rather than on production
through weight. Reid's experiment also suggests that the plane
of nutrition of growing heifers may not be as important to
later production as has been believed. His results indicate
that the plane of nutrition during the lactation period is

more important than that prior to the productive period. In
the present data the plane of nutrition would probably be lit-
tle changed over the period of life studied, whereas the three
groups of animals in Reid's experiment were treated differently
prior to first freshening but similarly thereafter. Ideally,
one would like a mass of field data similar to Reid's experi-
mental data, so that the environmental differences prior to
first calving could be evaluated. The present study provides
no clue as to whether the plane of nutrition in the growing
period or plane of nutrition in the lactation period is more
important.

The genetic correlations obtained, while they are not con-
clusive, in view of previous results, give further indication
that genes which cause the heifer to grow well also cause her
to produce well. Previous estimates may have been smaller due
to consideration of several lactations simultaneously.

What does the presence or absence of a genetic correlation
mean in practice? If there is a marked degree of positive cor-

relation between the genotypes for weight and production,

58

selection for one characteristic would achieve some gain in
the other measurement, even without paying attention to the
second at all. If there is no correlation, selection for one
alone would achieve no genetic gain for the other except by
chance. Gain in both traits could be achieved by concurrent
selection, but the amount of gain possible for each would be
less than if selection were solely for one or the other. If
there is a significant negative correlation, selection for one
would result in a net genetic loss in the other.

Further light might be shed on the subject if we could
learn whether there has been a consistent trend in size in
connection with prolonged selection for milk production. If
the correlation is as high as .3, we should be increasing the
size of our animals when we select for greater milk production.

Mason 33 31. (M0), in a study of the relationship of body
size and first lactation production in Red Dane cattle, con-
cluded that selection for milk yield alone would produce a
taller cow with less fleshing and a tendency to convert flesh
into milk during lactation. The bases for this conclusion were
genetic correlations of -.M5 between milk production and weight
gain during the lactation and .31 between wither height and
production.

Until the existence of a genetic correlation between weight
and production is further substantiated, no final disposition
of the role of weight in estimating breeding value for milk
production can be made. In the meantime, it would be unwise

to give much consideration to weight in choosing the parents

of the future generation. Should there be no genetic correla-
tion, in fact, the amount of genetic improvement possible in
milk production would be decreased by paying attention to weight.
Should a negative genetic correlation exist, actual losses in
the ability for milk production might result from strong em-
phasis upon weight.

Even if a genetic correlation is established with some
degree of certainty, the course of action to be taken is in
doubt. The present results indicate that the larger heifer
is not a more economical producer than the smaller heifer (in
the first lactation). If such a situation prevails throughout
productive life, it is questionable whether body size should
be emphasized any more than is absolutely necessary to prevent
deleterious effects on the length of productive life. If the
genetic correlation is large (.3 or greater), increasing the
size of dairy cattle would be concomitant with increasing their
ability for production. To select for both traits would be
to accentuate increases in size.

In conclusion, weight at first calving does not appear
to be a factor of economic importance in first lactation pro-
duction. (The primary consideration of weight in the first
lactation lies in evaluating the level of development at which
heifers should be bred. This stage of development should be
established in such a way that growth is not hindered materialLy
and the length of productive life is not decreased.

The principal causes of the association between weight

and production appear to be environmental differences such as

60

those which are manifested between herds. In the present ex-
ample, the principal factor is probably herd to herd differences
in the level of nutrition.

The investigation indicates that the genetic association
between weight and production merits further study. The results
suggest that the selection for milk production may also indi-

rectly be selection for size.

IMPLICATIONS IN STATISTICAL ANALYSIS OF PRODUCTION DATA

Several authors (5,50) have suggested that production data
be corrected for differences in weight. The present results
indicate that the intraherd direct influence of weight on pro-
duction is only about 2% of the total variation in production.
If this association is proven to be entirely environmental,
correction for weight differences may be worthwhile. If a
marked genetic correlation exists, some of the variation "due"
to weight is heritable. Under this condition, corrections for
weight differences would remove some of the genetic variatioi
in production.

The present results seem to contradict Gaines' conclusions
that age has no effect on production independent of weight and
that age correction is unsound. Gaines' premise appears to
be correct for a group of many herds, but not entirely sound
on an intraherd scale. Age appears to account for M-5% of the
intraherd variation in first lactation production, independent
of weight. Age corrections of high precision would remove this

fraction plus most of the joint variation due to age and weight.

61

IMPLICATIONS IN FUTURE INVESTIGATION

There are many aspects of the weight-production associa-

tion which require further attention if we are to accurately

assess the importance of weight in dairy cattle. Of course,

the present type of analysis needs to be repeated with larger

numbers of observations and for later lactations.

However, the following topics, which for one reason or

another could not be examined in the present analysis, appear

to offer

(1)

(2)

(3)

(h)

(5)

(6)

(7)

(8)

promise of yielding useful information:

Interaction between age and weight

Analyses of variance in production due to independent
effects of age and weight

Effect of calving at younger ages and smaller size

on longevity

Independent influence of weight on production at very
low and very high levels of weight

Heritable variation in production associated with
weight (estimation of heritability independent of
weight)

Linearity of relationship between weight and produc-
tion (application of curvilinear regression)
Estimation of simultaneous genetic gains for weight
and production in dual-purpose breeds of cattle
Apparent canceling of intraherd production variation

due to age differences by differences between herds.

62

SUMMARY

First lactation milk and fat production records (including ,
age and weight at calving) for M677 Holsteins, 1001 Guernseys,
and 501 Jerseys compiled in the Michigan DHIA-IBM program over
a three and one-half year period were analyzed to ascertain
the relation between weight and level of production.

Least squares estimates of the independent influences of
age and weight at calving on first lactation production were
presented in Tables M and.6. Maturity, as measured by age,
was negatively related to the economy of first lactation pro-
duction. The expected change in production associated with a
change in weight at first calving was dependent upon the level
at which the change in weight occurred. In all three breeds,
there were instances where a decrease in production was asso-
ciated with increased weight. The production response to level
of development appeared to reach a maximum at weights 300-M00
pounds above the breed average for the age. When the value
of added production accompanying increased weight was contrasted
to the added expense involved, heavier heifers had little or
no advantage in the first lactation.

Partial regressions of production on age and weight were
computed on overall and intraherd bases. When many herds were
considered, weight was more important to production than age.
The intraherd regressions of milk production on age and weight
at calving were on the order of 75 pounds per month and 200

pounds per 100 pounds, respectively. Weight was only about

63

one-half as important in predicting production records within
the herd as compared to production records among many herds.

Large between-herd correlations of weight and production
indicated that differences between herds contributed signifi-
cantly to the association.

Estimates of the genetic correlation between weight and
production were on the order of .3, indicating significant
genetic contribution to the association between weight and
production. Irregular results were obtained for Jerseys and
Guernseys.

Individual differences in_weight at first calving were
found to be heritable to the extent of .3 to .g. Heritabili-
ties of first lactation milk and fat production ranged from
.2 to .3 and .2 to .h, respectively.

It was concluded that dairy heifers should be bred as
soon as sufficient size is attained to minimize harmful ef-
fects on the length of productive life.

The results indicated that selection for increased milk
production may indirectly result in increased body size in
dairy cattle.

It was further concluded that correction for weight dif-
ferences may remove a fraction of the genetic variation in

production.

(1)

(2)

(3)

(h)

(S)

(6)

(7)

(8)

(9)

(10)

(11)

6h

REFERENCES

Agricultural Economics Department, Michigan State Univer-
sity. Compilation of Average Prices of Farm.Commodities.
Mimeo. 1957.

Bailey, G. L., and Broster, W. H. The Influence of Live
weight on Milk Production During the First Lactation.
J. Dairy Research, 21: S. 19Sh.

Beck, G. H., and Thrk, K. L. Size and Its Influence on
Milk Production. Purebred Dairy Cattle Association
14117160. 19,480

Blackmore, D. Phenotypic and Genetic Relationships Between
Body Measurenents and Milk Production in Holstein Cat-
tle. Unpublished Manuscript. 1958.

Brenton, C., and Stone, E. J. Interrelationships of Age,
Body Weight, and Milk Production of Holstein-Friesian
Cows in Louisiana. J. Animal Sci., 16: 1068. 1957.

Brody, S. B. Growth and Development With Special Refer-
ence to Domestic Animals. X. The Relation Between tha
Course of Growth and the Course of Senescence With
Special Reference to Age Changes in Milk Secretion.
Missouri Agr. Expt. Sta., Research Bull. 105. 1927.

Brody, S. B. Factors Influencing the Apparent Energetk:
Efficiency of Productive Processes in Farm Animals.
J. Nutrition, 17: 235. 1939.

Brody, S. B. Lactational Performance and Body Weight.
Science. 95: h85. l9h2.

Brody, S. B., and Procter, R. C. Growth and Development
With Special Reference to Domestic Animals. XXXV.
Energetic Efficiency of Milk Production and the In-
fluence of Body Weight Thereon. Missouri_Agr. Expt.
Sta., Research Bull. 222. 1935.

Brody, S. B., Ragsdale, A. C., and Turner, C. W. The
Rate of Growth of the Dairy Cow. III. The Relation
Between Growth in Weight and Increase of Milk Secre-
tion With Age. J. Gen. Physiol., 6: 21. 192k.

Davis, H. P., Morgan, R. P., Brody, S. B., and Ragsdale,
A. C. Relation of Height at Withers and Chest Girth
to Live Weight of Dairy Cattle of Different Breeds
and Ages. Nebraska Agr. Expt. Sta., Research Bull.
910 19370

(12)

(13)

(1h)

(15)

(16)

(17)

(18)

(19)

(20)

(21)

(22)

(23)

(2h)

65

Davis, H. P., Morgan, R. F., and Gaines, W. L. Live
Weight and Milk Energy Yield in the Nebraska Station
Dairy Herd. J. Dairy Sci., 26: 625. l9h3.

Relation Between Rate
J. Dairy Sci.,

Davis, H. P., and Willett, E. L.
of Growth and Milk and Fat Production.
21: 637. 1938.

Edwards, J. Effect of Breed, Size of Cow, Yield of Milk,
and Stage of Lactation Upon Efficiency of Milk Produc-
tion. J. Dairy Research, 7: 211. 1936.

Live Weight of Cow at Various Stages of
J. Dairy

Gaines, W. L.
Lactation in Relation to MilkéEnergy Yield.
Sci., 2h: 795. 19u1.

Live Weight Versus Metabolic Body Size
J. Dairy Sci., 29: 259.

Galnfis, W. L.
in Dairy Cows and Goats. 19u6.

Gaines, W. L., DEVlS, H. P., and Morgan, R. F. Within-
Cow Regression of Milk-Energy Yield on Age and Live
Weight. J. Dairy Sci., 30: 273. l9h7.

Gaines, W. L., Rhoda, C. S., and Cash, J. G. Age, Live
Weight, and MilkéEnergy Yield in Illinois Cows. J.
Dairy Sci.. 23: 1031. 19u0.

Gethin, R. H. The Age at First Calving of Dairy Cattle
in Relation to Subsequent Performance. Animal Breed-
ing Abstracts, 18: 133. 1950.

Gowen, J. W. Studies on Conformation in Relation to Pro-
ducing Ability in Cattle. IV. The Size of the Cow in
Relation to the Size of Her Milk Production. J. Agr.
Research, 30: 865. 1925.

Gowen, J. W. Conformation of the Cow as Related to Milk
Secretion, Jersey Registry of Merit. J. Agr. Sci..
23: hag. 19330

Gowen, J. W. The Conformation of the Parents as Related
to the Milk Secretion of the Daughter, Jersey Record
Of Merit. Jo Agr. $01., 23: 51).}. 1933.

Gowen, J. W. On the Genetic Constitution of Jersey Cat-

tle, As Influenced by Inheritance and Environment.
Genetics, 18: h15. 1933.

The Analysis of Non-Orthogonal Data With
Mimeo., Proc. Am. Soc. Animal Production. 1953.

Harvey, W. R.
IBM.

(25)

(26)

(27)

(28)

(29)

(30)

(31)

(32)

(33)

(3h)

(35)

(36)

(37)

(38)

(39)

66

Hazel, L. N., Baker, M. L., and Reinmuller, C. F. Gene-
tic and Environmental Correlations Between Growth Rates
of Pigs at Different Ages. J. Animal Sci., 2: 118.

19MB.

Henderson, C. R. Estimation of Variance and Covariance
Components. Biometrics, 9: 226. 1953.

Hofmeyr, Jan. A Study of Danish and Swedish Progeny Test-

ing Methods for Dairy Bulls.
Annaler, 22: M25. 1955.

The Heritability of Milk and Butgerfat
l9 0.

An Analysis of Data From the Danish Bull
2. f. Tierzuchtg. u.
195M.

Johansson, I. The First Lactation Yield as a Basis of
Selection Compared to the Second and Third Lactations.
Proc. Brit. Soc. Animal Prod., 79. 1955.

Johansson, I., and Hildiman, S. E. The Relationship Be-
tween Certain Body Measurements and Live and Slaughter
‘Weight in Cattle. Animal Breeding Abstracts, 22: 1.

195M.

Kendrick, J. F.
From Heart Girth Measurements.

Kungl. Lantbrukshogskolans

Johansson, I.
Yield. Animal Breeding Abstracts, 18: 1.

Johansson, I.

Progeny Testirg Stations.
Zuchtungsbiol., 63: 105.

Estimating the Weights of Dairy Cows
USDA BDIM, 695. 1936.

Kleiber, M. Body Size and Metabolic Rate.
27: M. 19h7.

Kleiber, M., and Mead, S. W.

Physiol. Revs.,

Body Size and Milk Produc-

tion. J. Dairy Sci.. 28: M9. 19M5.
Kleiber, M., and Mead, S. W. Body Size and Lactation
Rate. J. Dairy Sci., 28: M9. 19M5.

Knott, J. C., Hodgson, R. E., and Ellington, E. V.
Methods of Measuring Pasture Yields With Dairy Cattle.
Wash. Agr. Expt. Sta., Bull. 295. 193M.

The Genetics of Populations.
19MB.

Lush, J. L., Christensen, F. W., Wilson, C. V., and
Black, W. H. The Accuracy of Cattle Weights. J. Agr.
Research, 36: 551. 1928.

LUSh, J. L.
Lecture Notes.

Unpublished

MacArthur, J. W. Selection for Small and Large Body Size
in the House Mouse. Genetics, 3M: 19M. 19M9.

(M0)

(M1)

(M2)

(M3)

(MM)

(M5)

(M6)

(M7)

(M8)

(M9)

(50)

(51)

67

Mason, I. L., Robertson, A., and Gjelstad, B. The Gene-
tic Connexion Between Body Size, Milk Production, and
Efficiency in Dairy Cattle. J. Dairy Research, 2M:

135. 1957.

Morrison, F. B. Feeds and Feeding.
Publ. Co., Ithaca, N. Y. 1956.

22nd ed. Morrison

 

Pease, M. S.
in RabbitS.

Experiments on the Inheritance of Weight
J. Genetics, 20: 261. 1928.

Pfau, K. 0., and Bartlett, J. W. The Relationship of

Body Size to Milk Production in Dairy Cattle. Pure-
bred Dairy Cattle Association Mimeo. 195M.
Purebred Dairy Cattle Association Type Committee. Uni-

‘fied Score Card.
19MB.

Ragsdale, A. C., and Brody, S. B. Estimating Live Weights
of Dairy Cattle. Missouri Agr. Expt. Sta., Bull. 35M.
1935.

Reid, J. T., Loosli, J. K., Turk, K. L., Asdell, S. A.,
and Smith, S. E. Progress Report on a Study of the
Effect of Plane of Early Nutrition Upon Reproductive
and Productive Performance of Holstein Cattle. J.
Dairy Sci., M0: 610. 1957.

Shrode, R. R., and Lush, J. L.
Advances in Genetics, 1: 209.

Purebred Dairy Cattle Association.

The Genetics of Cattle.
19M7.

Touchberry, R. W. Genetic Correlation Between Five Body
Measurements, Weight, Type, and Production in the Same
Individual Among Holstein Cows. J. Dairy Sci., 3M:
2M2. 1951-

Turner, C. W. The Relation Between Weight and Fat Pro-
duction of Guernsey Cattle. J. Dairy Sci., 12: 60.
1929.

Turner, C. W. The Inheritance of Body Weight in Rela-
tion to Milk Secretion. Missouri Agr. EXpt. Sta.,
Research Bull. 1M7. 1930.

Turner, C. W., Ragsdale, A. C., and Brody, S. B. The
Relation Between Age, Weight, and Fat Production in

Dairy Cows. Missouri Agr. Expt. Sta., Bull. 221. 192M.

nn'ia UbE. 0'? ELY

""9" 21:33“

‘1

CIR.
.5
(’2.

Date Due