INHERITANCE AND GENOTYPE . ENVIRONMENTAL
STUDIES OF TEST WEIGHT AND RELATED KERNEL
CHARACTERISTICS m SOFT WINTER'WHEAT. TRITICUM
AESTIVU‘M L.

Thesis for the Degree of Ph. D.
MICHIGAN STATE UNIVERSITY I
ARDESHIR GHADERI
1196.9

I'HESIS

'.'

This is to certify that the

thesis entitled

Inheritance and Genotype-Environmental
Studies of Test Weight and Related
Kernel Characteristics in Soft Winter

Wheat, Triticum aestivum 2.
presented by

Ardeshir Ghaderi

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Crop Science

5: 7%éwv

Major professor
r, E
Date MIA/£0 /l I 7(07
U T

0-169

ABSTRACT

INHERITANCE AND GENOTYPE-ENVIRONMENTAL STUDIES OF
TEST WEIGHT AND RELATED KERNEL CHARACTERISTICS
IN SOFT WINTER WHEAT, Triticum aestivum E.

 

BY

Ardeshir Ghaderi

Test weight is defined as the weight of grain
necessary to fill a given unit of volume. It is the
product of density and volume of grain occupying the
container. The latter component, when eXpressed as
percentage of the volume of the container, is referred
to as packing efficiency and this component was shown to
have a much greater effect on test weight than density.

A significant but negative correlation was found
between test weight and length.width ratio. Approxi-
mately 39 percent of the variation in test weight was
shown to be associated with this ratio and the remainder
was unexplained. Volume or weight of the grain was
shown to have no effect on test weight. Within varieties
pearling index and kernel protein were found to be related
to kernel size.

Test weight, packing efficiency, flour yield, and
other physical prOperties of the kernel such as weight,

density, length and width were affected by environment.

Ardeshir Ghaderi

With few exceptions, reduction in kernel weight was
usually accompanied by reduction in test weight. Kernel
length was more resistant to environmental factors than
kernel width at later stages of kernel develOpment. By
removing the effect of density and dealing with packing
efficiency alone, a definite gain was obtained in the
genetic variance. Using the analysis of variance pro-
cedure, the second order interaction, V x Y x L, was
demonstrated to be highly significant. The first order
interactions, i.e., V x L and V x Y, were not signifi-
cant. Heritability estimates, both on the basis of
single plots and line means differences, were made and

a much higher value was obtained for the latter than the
former.

Stability analysis for 22 lines under 15 environ-
ments revealed that some lines did not show genotype-
environment interaction. A negative but significant
correlation was obtained between the stability parameter
(8) and the average performance over all environments.

For the study of inheritance of test weight the
diallel technique was employed and high test weight was

shown to be dominant over low.

INHERITANCE AND GENOTYPE-ENVIRONMENTAL STUDIES OF
TEST WEIGHT AND RELATED KERNEL CHARACTERISTICS
IN SOFT WINTER WHEAT, Triticum aestivum E.

 

BY

Ardeshir Ghaderi

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of CrOp Science

1969

<§UOC775
7-3-47

ACKNOWLEDGEMENTS

The author wishes to express his sincere gratitude
to Dr. E. H. Everson for his encouragement and guidance
throughout the study.

Thanks are also due to Dr. C. E. Cress for his help-
ful suggestions and assistance in the statistical analysis,
to Dr. C. M. Harrison for his critical appraisal of the
manuscript, to Dr. F. C. Elliott and the other committee
members for their helpful suggestions.

Also the author wishes to express his sincere
appreciation to Dr. W. T. Yamazaki and the members of the
U.S.D.A. Soft Wheat Quality Laboratory, Wooster, Ohio for
their c00peration and assistance in determination of
quality characteristics.

The advice and assistance of my friends in the

Department of CrOp Science is also sincerely appreciated.

ii

TABLE OF CONTENTS

INTRODUCTION 0 O I O O O O O O O O O O O O O O 0

REVIEW OF LITEMTURE O O O O O O O O O 0 O O O 0

MATERIALS AND METHODS O I I O O O O O O G O C O O

l.

3.

Relationships among the Physical

PrOperties of the Kernel and Possible
Correlation of these PrOperties with
Quality in Soft Winter Wheat. . . . .

Effect of Environment on Test Weight
And Other Kernel Characteristics in
Soft Winter Wheat . . . . . . . . . .
a. Pattern of Change in Kernel
Characteristics Under Different
Environments. . . . . . . . . . .

b. Estimation of Variance Components
And Heritability of Test Weight .

c. Stability Study . . . . . . . . .

Inheritance of Test Weight. . . . . .

RESULTS AND DISCUSSION 0 O O O O O O 0 O O O 0 O

1.

Relationships Among Physical PrOperties
And Quality Characteristics of The
Kernel. O C O O O O 0 O O O O O O O 0

Effect of Environment on Test Weight
And Other Kernel Characteristics. . .

a. Pattern of Change in Kernel

Characteristics Under Different
Environments. . . . . . . . . . .

iii

l7

17

20

20

22

26

29

32

32

44

44

Page

b. Estimation of Variance Components
And Heritability. O O I O O O O O O 49

The Effect Of Environment
On Test Weight And Its

Components. . . . . . . . . . . 54

c. Stability Study . . . . . . . . . . 57

3. Inheritance of Test Weight. . . . . . . 62
SUMMARY AND GENERAL CONCLUSIONS. . . . . . . . . . 67
BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . 71
APPENDIX . . . . . . . . . . . . . . . . . . . . . 77

iv

Table

1.

10.

LIST OF TABLES
Page

Test Weight (gms/pint), Protein (%),

Flour Yield (%) And Pearling Index (%)

Data For 8 Varieties Separated Into 4

Groups; Large, Medium, Small, And Bulk. . . 39

The t Values For Comparisons of Different
Kernel Sizes For Test Weight, Flour Yield,
Pearling Index, And Kernel Protein. . . . . 40

Correlation Coefficients Of Differences
Between Test Weight And 1000 Kernel Weight
At Paired Locations . . . . . . . . . . . . 44

Correlation Coefficients Of Differences

Between Test Weight And Flour Yield; And

Of Differences Between 1000 Kernel Weight

And Flour Yield . . . . . . . . . . . . . . 46

Location Means For Flour Yield, 1000 Kernel
Weight And Test Weight From 30 Lines
Averaged Over 2 Replications. . . . . . . . 46

Mean 100 Kernel Length (cm.) And 100 Kernel
Width (cm.) For Locations Obtained From 30
Lines With 1 Replication. . . . . . . . . . 47

Correlation Coefficients Between Locations
For Length And Width Of Kernel. . . . . . . 48

Estimates Of Variance Comonents For Test
Weight In Set A And Set B . . . . . . . . . 49

Variance Components In Set A And Set B
Expressed As Percentage Of Their Corres-
ponding Total Phenotypic Variance . . . . . 50

Estimates Of Variance Components For Test
Weight, Packing Efficiency, And Density . . 55

Table

11.

12.

13.

14.

15.

16.

Estimates Of Variance Components Expressed
As Percentage Of Their Corresponding Pheno-
typic Variance. . . . . . . . . . . . . . .

Analysis Of Variance Table For Regression
Of Genotype-Environment Interaction Over
Environmental Index For Genesee . . . . . .

Analysis Of Variance Table For Regression
Of Genotype-Environmental Interaction
Over Environmental Index For Yorkstar . . .

Combined Regression Analysis For The 22
Lines Planted In 15 Environments. . . . . .

Average Performance, Slope, And Genotype-
Environmental Relationship Of 22 Lines
Planted In 15 Environments. . . . . . . . .

Micro Test Weight (gms per 47 ml. glass

F
jar) For A Set Of 7x7 Diallel Averaged Over
2 Replications. . . . . . . . . . . . . . .

Vi

Page

55

58

58

58

59

63

LIST OF FIGURES

Figure Page
l-a Relationship between (100) kernel
width (cm.) and (1000) kernel
volume (m1.). . . . . . . . . . . . . . . 35
l-b Relationship between (100) kernel

width (cm.) and (1000) kernel
weight (gms.) . . . . . . . . . . . . . . 35

l-c Relationship between density and
test weight (gsm/pint). . . . . . . . . . 35

1-d Relationship between test weight
(gms/pint) and packing efficiency (%) . . 35

2 Relationship between standard test
weight (gsm/pint) and micro test
weight (gms/47 m1. glass jar) . . . . . . 36

3 Wr-Vr Graph for Micro-Test Weight
from a set of 7 x 7 Diallel, F3 . . . . . 64

vii

INTRODUCTION

Test weight is widely recognized as an important
grading factor. Low test weight in soft wheats has been
a matter of concern to the farmer, breeder, and industry.
Farmers have been penalized for wheats of low test
weight because industry is reluctant to accept such
wheats since there is a belief that low test weight is
related to poor milling quality. The recent introduction
of semidwarf Norin 10 germplasm into newly released soft
wheat varieties has revived the controversy concerning
the relationship between test weight and quality,
especially flour yield.

Neither physical properties of the kernel contri-
buting to low or high test weight nor the relationship
between these prOperties and quality is clearly established.
Mode of inheritance, stability of performance, and the
environmental effect on test weight have received little
attention. The intent of this study was to establish the
relationship(s) between test weight and other physical
pr0perties of the kernel. Such information might lead to
a better understanding of test weight and may find its

usefulness in the breeding of high test weight varieties.

The relationships between physical properties of the kernel
and some quality characteristics were investigated. The
study was further oriented toward a better understanding

of environmental effect, its stability, and its mode of

inheritance.

REVIEW OF LITERATURE

Test weight, weight per bushel or bushel weight
of wheat was used by practical millers as a measure of
flour yield potential long before the establishment of
the present grading system. It is defined as the weight
of a volume of grain necessary to fill level full a unit
of volume, the volume used being a Winchester bushel
with a capacity of 2150.42 cubic inches (55, p. 7).
Because of the simplicity involved in the determination
of test weight, it was probably the first to be recognized
as an important grading factor (53). Boerner in 1916
emphasized the importance of test weight and described
an improved apparatus for its determination (12). He also
explained that test weight determinations should be carried
out under controlled conditions. The standard method of
test weight determination has been described (4, 21).
Micro determination of test weight was prOposed by
Harris and Sibbitt (23), Swanson (56), Aamodt and Torrie
(l), Roberts (49), and Yamazaki (60 and personal communi-
cation). They used containers of different sizes to
measure test weight. These measurements were reported to

be valuable when the amount of available seed is not

sufficient to run the standard test. The correlation
coefficients between micro-test and the standard method
were reported to be very high.

Roberts (49) discussed the importance of test
weight and the possibility of breeding for kernel types
with high packing quality and consequently, a high test
weight. In his comprehensive study on test weight, he
measured some physical properties of wheat kernels for
52 lines. The prOperties measured were: weight, volume,
density, length, width, packing quality, and length width
ratio. He employed a micro-test weight method and explained
packing quality as percentage of volume of the grain capable
of being packed into 100 cc. There was a high correlation
between packing quality and test weight. He also stated
that within limits, test weight was correlated with kernel
volume if length width ratio was kept constant. The
importance of kernel shape was emphasized with the hope
that the length width ratio and kernel volume would explain
shape differences. The importance of kernel shape as a
factor influencing test weight has also been emphasized
by many other workers.

Boshankian (ll) outlined the factors which influence
the shape of the wheat kernel. These factors were: rigi-
dity of the glumes; the size and form of the space within

which kernel develOpment occurs; number of kernels and

their locality in the spikelet; spike density; the amount
of pressure exerted by different growing parts of the
spike; and finally the species characteristics. He
stated:
"...that the grain of wheat throughout its period
of development is very soft and that it hardens
only after it attains its maximum development.
Hardening is a drying process and occurs during
the last few days of its period of maturation.
As the kernel is very soft before this period,
the slightest pressure on the grain through con-
tact is very apt to modify its shape."

Lamba (32) in a study of relationships between
kernel and glume dimensions reported that the highest
correlation was obtained between length of the kernel
and length of the glume. He concluded that the kernel
length is least affected by environment. Clark and
Bayles (14) stated that the wheat kernel reached its
maximum length several days before the kernel matures
and it is a valuable characteristic for taxonomical re-
search. Murphy and Frey (41) in a study of the components
of kernel weight in oats broke down groat weight into its
components as follows: G. W. = L (W/2)2 D K, where G. W.
is groat weight; L, length; W, width; D, density, and K
a constant related to shape factor.

Hlynka and Boshuk (26) discussed the relationship

between kernel size, packing, and density with test weight.

By presenting a model, they concluded that test weight is

independent of kernel size. Shape of the kernels and
their heterogeneity were the factors which influenced

the density of packing. They stated the possibility of
obtaining different packing densities from random pack-
ing of long and round plump kernels. They further stated
that mixing of small and large kernels may still lead to

a different test weight. Heizer and Johnson (25) obtained
a correlation coefficient of -0.19 between test weight and
kernel volume. Frequent reports have been published, indi-
cating low correlations between test weight and kernel
weight.

Yamazaki (60 and personal communication) used the
term packing efficiency rather than the term packing
quality as was used by Roberts (49) to express the per-
centage of the volume of the container being occupied by
the volume of the grain. Test weight is the product of
density and the volume of the grain occupying a certain
volume. The second component is different from packing
efficiency only by a constant. Yamazaki (60 and personal
communication) studied the relative importance of packing
efficiency and density_and obtained a higher correlation
between test weight and packing efficiency than between
test.weight and density. He also reported separation of
broken kernels of several samples and noted that broken
kernels always gave a lower packing efficiency due to their

irregular shape.

Shollenberger and Coleman (52) separated kernels
of a hard red winter wheat lot into 3 groups; dark, hard,
and vitreous; Spotted or mottled; and starchy or yellow.
The effect of kernel texture on test weight was very
small; mottled kernels had a tendency to be the highest
and starchy kernels to be the lowest in test weight.
Yamazaki (60 and personal communication) divided a sample
of Purkof wheat into vitreous and mealy kernels. The
third group consisted of a sub sample of the original lot.
Test weight for the mealy kernels tended to be the lowest
and for the vitreous kernels the highest. The original
lot fell between the two groups.

The amount of moisture present in the grain influ-
ences test weight. Bates, et. al. (8) tempered a quantity
of wheat in a temporary storage for 24 hours. A reduction
of 1.2 pounds per bushel as compared with the test weight
on entering the washer was reported. Swanson and Pence
(57) reported on the test weight of wheat with varying
moisture levels. An increase in moisture content was always
accompanied by a decrease in test weight. Low test weight
of the moist samples was attributed partly to lower density
of the moist kernels and partly due to swelling of the
kernels. The effect of moisture in lowering the density
of the grain is due to grain density being greater than

water density (26).

Sharp (3» showed that when kernels of wheat are
moistened and then redried, they will not regain their
original size. Swanson (55, p. 140) indicated that
scouring of the treated samples increased their test
weight. Shollenberger and Coleman (52) showed that scour—
ing of the kernels (without previous wetting) increased
the test weight of both hard red winter and hard red spring
wheats; however, they obtained opposite results from durum
wheat which they could not explain.

There has been a tendency to correlate test weight
with protein content of the kernel. This is mainly due
to higher protein content of vitreous kernels which usually
gives a higher test weight. Snyder (54) demonstrated that
within varieties dark kernels contained more protein than
the lighter ones. Shollenberger and Kyle (53) using a
multiple and simple correlation technique found a signi-
ficant correlation between test weight and protein in hard
red spring wheats. They noted a tendency for protein con-
tent to increase as test weight increased; however, this
tendency was for samples with test weight of less than 54
pounds per bushel. For wheats weighing more than 54 pounds
per bushel the correlation was reversed. They indicated
that 44 percent of variation in protein is due to the com-
bined effeCtIOf kernel texture and test weight, the former

being more important than the latter.

Specific gravity or density as a component of test
weight has been studied by many investigators. Density
of grain as defined by Swanson (55, p. 8) is "concentration
of matter in a unit of volume, the standard being a cubic~
centimeter of water at its greatest density at 4 C."
Shollenberger and Coleman (52) reported that there is a
definite relation between density and kernel texture.
Shollenberger and Kyle (53) also reported that there was
a definite increase in the amount of protein as the per-
centage of dark, hard vitreous kernels was increased.
Sharp (50) discussed the relationship between density and
test weight. He stated that a low test weight may not
necessarily be correlated with density. He emphasized the
fact that test weight depends on two factors: (1) density
of the grain and (2) the actual volume of the grains which
occupy a certain volume. He also studied the effect of
moisture on the vitreous and yellow berry kernels of the
same lot of wheat. It was reported that an increase in
moisture content of both types of kernels is accompanied
by a decrease in density; however the vitreous kernels
decreased in density at a higher rate. On the basis of
these results, Sharp concluded that there may be a ten-
dency for water to fill the air spaces in the yellow
berry kernel rather than expand it. Heizer and Johnson

(25) studied the effect of moisture on density of wheat

10

grain. They reported that an increase in moisture was
accompanied by a decrease in density. The Purkof variety
(with the highest density) tended to decrease in density
at the highest rate compared with other varieties. Moisture
content and moisture history, amount of protein, and the
kind of protein were reported by Sharp (50) to be the main
factors which determine density. When he ground vacuum
dried samples of kernels, the density of the two samples
(vitreous and yellow berry) was very close, indicating
that the lower density of yellow berry kernels was due to
a larger number of small air spaces. Yamazaki (60 and
personal communication) selected three varieties of soft
wheat with low, medium, and high density and samples of
each variety were subjected to different degrees of
cracking. He divided samples of each variety into four
categories as follows: 1) unground kernels, 2) coarse
ground kernels with kernels barely broken, 3) medium ground
kernels, more severe than the second category but not so
hard to produce flour, and finally 4) wheat meals of each
variety. For all three varieties of wheat, the wheat meal
samples had the highest density followed by medium ground,
coarse ground, and unground samples. Assuming the amount
of air spaces in wheat meal to be zero, he showed that
unground samples of the three varieties had 2.7, 5.5, and
8.3 percent air in the high, medium, and low density vari-

eties respectively.

11

Shape of the kernel has been reported as the main
factor influencing test weight (7, 26, 49, 55), but there
has been no success in quantifying this characteristic.
Other factors such as shriveling of the kernels, or the
condition of the kernel surface, depth of crease, extent
of brush, frequency of handling, have also been pointed
out as the factors affecting weight per bushel.

It has been an accepted concept that high weight
per bushel is associated with plumpness of the grain and
consequently a higher flour yield (38, 39). On the basis
of averages obtained for test weight and flour yield for
numerous samples Thomas (58) concluded that there was a
definite relation between test weight and flour yield.
However, when he considered individual samples this
relationship was not consistent. He found no relationship
between kernel weight and flour yield. Swanson (55, pp.
137-138) presented data which had been collected by the
Howard Testing Laboratories from many thousands of samples
over 15 years. The data showed that there was a relation-
ship between test weight and flour yield, but this relation
was not linear. An almost constant ratio between weight
per bushel and flour yield was obtained between the range
of 51 and 61 pounds per bushel. The data presented by
Bailey and Sherwood (6) showed that wheats of the same test

weight may differ considerably in their flour yield.

12

This inconsistency as described by Swanson (55, p. 138)

is due to the use of a smaller mill for flour extraction.
Swanson (55, p. 138) further pointed out that when milling
is carried out on a small laboratory mill the amount of
flour cannot be predicted from test weight. Bailey (5)
studied the relationship between kernel volume and actual
percentage of endosperm. He took samples of grain from a
field of the Bluestem variety starting 10 days after head-
ing and continuing until maturity. Endosperms of the
samples were dissected, weighed and dried. A progressive
increase in kernel volume was accompanied by an increase
in the percentage of the endosperm. On the basis of these
results, he concluded that within the same variety, kernels
with high volume would possess a higher flour yield poten-
tial due to their higher percentage of endOSperm. He also
pointed out that under equal conditions large kernels have
a higher Specific gravity than small kernels, indicating

a higher density for endosperm rather than the bran and
germ. Swanson (55, p. 6) explained the situation in another
way. He stated that within the same variety large kernels
would produce a higher flour yield, because the variation
caused by kernel volume would be more pronounced on total
size of the kernel than on the bran coat. Sheuy (51) pre-

sented data and reached the conclusion that wheats may be

13

as much as 9 pounds different in their test weight with

the same flour yield. He further pointed out that frequent
handling would increase test weight without any increase in
flour yield. He deve10ped a wheat sizer by which the kere
nels were separated according to their cross-sectional area.
He obtained a correlation of 0.957 between flour yields
calculated from this sizing procedure and commercial flour
yields. The correlation between test weight of the same
samples and flour yield was reported to be 0.747.:

Barmore and Bequette (7) stated that compactness of_
the head and distorted contours of the grain in club wheat
always led to a lower test weight than common varieties.
They pointed out that this special characteristic of club
wheats has resulted in (unfair) penalization of club wheat
producers. Further data indicated that club wheat varieties
may produce a 5 percent or higher flour yield than common
white varieties. They further concluded that in white
wheats of the Pacific Northwest, weight per bushel is a
poor predictor of flour yield, and test weight is a poor
index of flour yield between varieties, within varieties,
within a white wheat subclass, or between subclasses.

Pfiefer (46) without presenting data stated that a
correlation of 0.152 had been obtained between test weight
and flour yield. He made the statement that the millers

have no reason to expect a high or low extract in the 58-61

14

pound test weight range. He suggested that research should
be carried out and methods should be develOped so that
exact flour potential could be predicted. Johnson and
Hartsing (30) suggested that weighing of a Specific amount
of grain and then counting the number of kernels per weighed
sample would be a more precise measure of kernel weight.

He referred to this characteristic as kernel count and
obtained a significant correlation of -0.84 between this
prOperty and flour yield among hard wheats; however, the
correlation among soft wheats was reported to be of the
magnitude of -0.68.

Fisher and Halton (20) in a much earlier paper
reached the same conclusions with regard to sampling for
1000 kernel weight determinations. They pr0posed weighing
of samples of about the same amount, counting the number
of kernels per sample, and then converting the results
into 1000 kernel weights. Using this procedure they were
able to reduce the sampling error to a great extent.

Regarding the effects of various environmental fac-
tors on test weight, no comprehensive report has yet been
published; however, there have been some short reports on
the effect of lodging and delayed harvesting on test weight.
Pendleton (43) showed that different degrees of artificial
lodging at different dates had a definiteréffect in lower—

ing test weight. Laude and Pauli (33) accomplished

15

artificial lodging of different degrees at different dates
on winter wheat and concluded that the effect of lodging
on test weight is minimum if the lodging occurs 5 days
before heading. Lodging after or before this time was
more effective in reducing test weight. They pointed out
that lodging within 2 weeks immediately before heading
would lead to the production of small but fairly plump
kernels; however, shrinkage of the partly developed kernels
accounted for the low test weight from plots lodged during
heading and for two weeks after this time. Day (17) used
2 varieties of barley and employed 2 degrees of lodging
(45° and 90° from the perpendicular) at 3 stages of growth.
He showed that 90 degree lodging had a severe effect in
decreasing test weight and kernel weight at all stages

of heading. Weibel and Pendleton (59) demonstrated that

a 4.8 pound loss in test weight resulted from lodging at
heading time; however lodging at a later period, 3 weeks
after heading, during the hard dough stage resulted in a
reduction of only 1.1 pounds loss in test weight.

The effect of delayed harvesting on test weight was
reported by Pool, gt El' (48). They showed that delayed
harvesting resulted in some decrease in test weight. The
heads of 3 soft red wheat varieties were harvested at dif-
ferent times following maturity. They reported that delayed

harvesting had a pronounced effect in lowering test weight.

16

This decrease in test weight was attributed partly to
reduction in kernel weight and partly to some increase
in kernel volume brought about by weathering.

Date of seeding was reported by Pitman and Andrews
(47) to influence test weight. They showed that late and
early planting resulted in lower test weights. The low
test weight obtained from wheat planted late was probably
due to late maturity and greater effect of stem and leaf
rusts. When winter killing was severe, higher bushel
weights were obtained.

Bayles and Suneson (9) reported that bearded vari-
eties of wheat may have higher test weight than beardless
varieties. They made crosses between bearded and beard-
less varieties. Segregating pepulations then were divided
into bearded and beardless pOpulations. On the basis of
replicated plots over a period of 4 and 5 years they
reached the conclusion that the grains from the composite
of bearded plants were superior to that from the composite
of beardless plants both in kernel weight and test weight

for each cross.

MATERIALS AND METHODS

1. Relationships among the Physical PrOperties of
the Kernel and Possible Correlation of these
Properties with Quality in Soft Winter Wheat

A micro-test weight was develOped and was especially,
important in cases where the amount of seed was insufficient
for running the standard test weight determination. The
term ”standard test weight" in this study is defined as the
weight of the grain necessary to fill level full a pint
measure and is expressed as grams per pint. The container
used for the micro-test weight determinations was a small
glass jar with a capacity of 47 ml. The procedure of run-
ning the micro-test in this study and other studies (when
it was necessary to run the micro-test) can be summarized
as follows: 1) the container was filled until the grain
flowed over; 2) the container then was gently tapped on the
table to let the grain pack, but not so hard that the grain
on the tOp fell off; 3) starting from the edge of the glass
jar, with a stroker the grain was leveled with a zig zag
movement? 4) and finally the weight of the grain in the
glass jar was measured by a balance. All the data obtained
during the course of the study represent averages of 10

such determinations.

l7

18

A total number of 59 samples of soft winter wheat
representing a great amount of diversity among the avail-
able materials with regard to test weight, kernel weight,
and other physical pr0perties were studied. Test weight
(both micro and the standard) and other physical prOperties
of the kernel such as weight, length, and width were de-
termined on all the 59 samples. Length measurements were
made by taking 100 kernels from each sample at random and
placing them end to end on a measuring board somewhat simi-
lar to the device developed by Roberts (47). Width deter-
minations were also made on the same 100 kernels by placing
them crease down on another measuring board. Then length
width ratios were calculated by dividing the former over
the latter. Later samples of about one pound representing
all the 59 samples were sent to the U.S.D.A. Soft Wheat
Quality Laboratory in Wooster, Ohio to make measurements
on density, kernel volume, packing efficiency, and quality
characteristics.

Packing efficiency and kernel volume were determined
by a Beckman air comparison pycnometer, an instrument by
which volume of a certain weight of kernel can be measured
without damage. By using this inStrument it was possible
to make rapid measurements of test weight, packing efficiency,

and density at the same time. A cup with a known volume is

19

filled with grain and then the grain on the tOp of the cup
is leveled in the same manner as described for the micro-
test weight measurements. The weight of the grain in the
cup, also called micro-test weight, is a measure of test
weight and is highly correlated with test weights measured
by the standard method. Then the cup filled with grain

is placed inside of the instrument and the volume of the
grain is measured by air replacement. Having obtained
this volume, density can be calculated by dividing the
weight of the grains (micro-test weight) by their volume;
however, for volume determination of individual kernels,
the number of grains in the cup should be counted.

Among the determined quality characteristics, only
the data for flour yield, kernel protein, and pearling
index will be presented. Pearling index is a measure of
softness of the grain and is measured by subjecting a
sample of grain to a rotating carborundum wheel and metal
screen for a certain period of time. It is expressed as
percentage of reduction in weight of the sample (10).
Higher values represent softer textures.

In order to see if the kernel size within varieties
had any effect on test weight and quality, 8 commercial
varieties were taken. Each variety was separated into 4
groups.- This separation was based on sieving each variety

through 2 sieves of different size. The kernels remaining

20

on the tOp of the first sieve (with larger sieve size)
were graded as large, and the kernels passing through the
second sieve were graded as small. Medium sized kernels
consisted of kernels which could pass through the first
sieve, but not through the second. The fourth group was
composed of the natural bulk of the variety. Standard
test weights were made on all the groups within each of
the 8 varieties. The samples of about one pound from each
group were then sent to the U.S.D.A. Soft Wheat Quality
Laboratory in Wooster, Ohio for determination of flour yield,
kernel protein and pearling index.

2. Effect of Environment on Test Weight And Other

Kernel Characteristics in Soft Winter Wheat

a. Pattern of Change in Kernel Characteristics
Under Different Environments

This study was conducted to determine the varietal
behavior of test weight, kernel weight, length, width, and
flour yield under different environments.

Test weight and 1000 kernel weight measurements were
made on 30 lines replicated 3 times at 7 locations in 1968.
The 7 locations and their corresponding symbols which will
be used henceforth, are as follows: Ingham County, (I01);
Lenawee County, (L41); Ionia County,(SSl); Tuscola County,
(T61); Huron County, (H71); Kalamazoo County, (K81); and

Berrien County, (B91). Flour yield extractions were made

21

by the U.S.D.A. Soft Wheat Quality Laboratory on the same

30 lines, but only from 4 of the locations, namely 101,

L41, H71, and B91. For the_latter experiment only the first
2 replications were taken.

To study the effect of environment on kernel length
and width, only seeds from one replication of the 30 lines
from locations 101, L41, and H71 were uSed. Length and
width measurements followed the same procedure as previously
described.

Locations 101, L41, and H71 were deliberately selected
because lodging following heavy rainfall occurred in 101 at
maturity whereas prevalence of extremely unfavorable and
favorable environmental conditions occurred at ‘L41 and H71,
respectively.

All possible differences of test weight and kernel
weight for all pairs of locations for lines were calculated.
The lines at each location were represented by averageS‘
over replications. Differences of test weight were regressed
on kernel weight and flour yield differences. As a result
21 apprOpriate correlation coefficients of differences be-
tween test weight and kernel weight at paired locations were
obtained; however, only 6 correlation coefficients of this
type were obtained between test weight and flour yield and
another 6 between kernel weight and flour yield. An example

might be helpful in illustrating the procedure. Let test

22

weight in Huron County (H71) and Berrien County (B91) be
denoted by TH71 and TB91 respectively, and 1000 kernel
weight for the same 2 locations be represented by KH71 and
KB91 reSpectively. Then differences between corresponding
lines at these 2 locations were calculated both for test
weight (TH71 - TB91) and kernel weight (KH71 - KB91).
Finally the regression of the former on the latter was cal-
culated. Length and width comparisons were made only
between the means of the locations.

b. Estimation of Variance Components And

Heritability of Test Weight

The purpose of this study was to explore and evaluate
the preportion of phenotypic variance for test weight arising
from genetic differences among lines, interactions of geno-
type with environment and finally a random error associated
with plot error and a composite of errors due to sampling
within plots, measurements, etc.

The data used in this study came from yield evaluation
experiments in 1967 and 1968. Two sets of advanced breeding
lines and commercial varieties were planted in 7 locations
in both years. Each set hereafter will be referred to as
nursery I and nursery II. For any one year the same 30
lines appeared in nursery I in all locations. The same was
true for nursery II. However, the composition of nursery I
and II was not the same in both years. The 7 locations used

were: 101, L41, 851, T61, H71, K81, and B91.

23

Among the 30 lines planted in nursery I in 1967
and 1968, 16 were common in both years and in all locations.
A total number of 3 replications was used at any one loca-
tion for any one year. Each plot consisted of 4 rows and
each row was 12 feet long with 12 inches between the rows.

A seeding rate of 80 grams per plot (2 bushels per acre)
was used. The two center rows were harvested and test

weight measured by the standard method, and expressed as
grams per pint. Because of insufficient seed from many
plots at L41 and S51 in 1968 the micro-test was employed
and the results then were converted into grams per pint.

For the present study test weight data from the above
mentioned 16 lines over both years and the 7 locations were
used. Locations 851 and L41 in 1968 were characterized by
heavy weed infestation and by lodging in 851. Because of
the prevalence of such unusual environments in these two
locations, the data was analyzed one time with and another
time without these two locations. In the latter case the
data for these two locations from 1967 should be ignored.
For the sake of simplicity the first set of data including
these 2 locations will be referred to as set A, and for the
next set, excluding these 2, as set B. The data were analyzed
for both sets in the same manner.

For statistical analysis, years, locations, and lines
were assumed to be random. Any phenotypic value was presented

as:

24
Yijk1==y+vi+lk+yj+rjk1+ (v1)ik+ (vy)ij+ (ly)jk+ (v1y)ijk+eijkl

In this model Yijkl being phenotypic value; u, grand mean
of the population; and vi, the average genotypic effect. 1

kl
and rjkl are the direct effects of locations, years, and

Yj:
replications respectively (15, 36). Interaction components
are presented as combinations between or among these main
factors and eijkl is an error term associated with plot
errors. Estimation of variance components was obtained
from expected mean squares as described by Comstock and

Robinson (15). The following estimates of variance compon—

ents were calculated:

2 . . . .
0v = variance component due to genetic differences among lines.
2 . . . . .
O = variance component due to lines x locations interaction.
v1

02 = variance component due to lines x years interation.
= variance component due to lines x locations x years
interaction.
02 = error variance arising from different plots within
locations within years.

Having calculated the estimated value of these vari-
ance components, heritability estimates of differences among
line means and single plots were calculated. The formula

used for estimation of the latter:

25

 

 

 

02
H- v
02 + 02 + 02 + 02 + 02
v v1 vy' vly e
and for the former:
°5
H:
02 c 2 02 02
v1 vy vly e
02 +——— + ——— + + ———
V 1 y 1y rly

In order to investigate the effect of environment
on the components of test weight, i.e., packing efficiency
and density, the same 30 liees from 3 of the locations, i.e.,
H71, B91, and L41 in 1968 were used. Packing efficiency
and desnity were measured. Test weight data are presented
in its original unit, grams per pcynometer cup and no
attempt was made to convert them to another unit. The data
for packing efficiency also were not converted into percent—
age of the container volume and they represented the real
volume of the grains occupying the pycnometer cup. Two
measurements per plot were made and the average of these
two was used to represent each plot. A randomized block
design with 2 replications in each location was used. Esti-
mates of heritability and the variance components were cal-

culated by the same procedure as explained before.

26

c. Stability Study

This study was designed to assess the stability of
22 lines of soft winter wheat for test weight and to obtain
more information about the interaction of individual geno-
types with environments. Analysis of variance procedure
as used before is useful for estimation of genotype—
environment interaction variance components; however, it
does not provide information about interaction of individual
genotypes with environments. Neither does it provide
information about the stability of the genotypes.

Yates and Cochran (61) using a pure statistical
analysis partitioned genotype-environment interaction into
linear and nonlinear components. This technique was modi-
fied by Finlay and Wilkinson (19), Eberhart and Russell
(18), and Perkins and Jinks (45). Finlay and Wilkinson
regressed the mean performance of each line onto the mean
performance of all lines for each environment, and the re-
gression lepe was suggested to be the measure of stability.
A stable line was described as one which performs the same
over all environments (b = 0). Eberhart and Russell (18)
developed a model and described a stable line which possesses
a sloPe of unity. They further employed two ancillary meas-
ures of stability, i.e., high mean and deviation from
regression of zero. Bucio-Alanis and Hill (13) developed

a model from which more information could be gained about

27

genotype-environmental interaction of paired lines. Perkins
and Jinks extended Bucio-Alanis' model to cover any number
of inbred lines. Their model also provides a better under-
standing of stability and they further subdivide genotype-
environment interaction of individual lines into linear

and non-linear components.

The data for this study was obtained from the yield
evaluation eXperiments in 1967 and 1968. The same 16 lines
and 7 locations used in the previous study for the esti-
mation of genotype-environment variance components, were
employed; however, an additional 6 lines which were planted
in nursery II in 1967 and in nursery I in 1968 at all
locations were also included in the analysis. The data
from Monroe County (M31) in 1968 were also incorporated
into the procedure. This gave a total number of 22 lines
planted in 8 locations in 1967 and in 7 of the same loca—
tions in 1968. These 15 environments were considered with-
out regard to their nature, i.e., separate environments.
This procedure resulted in 15 environments with 22 lines
within each environment. A total number of 3 observations
were made on each line at each environment. The means over
the three observations were used to represent the performance
of each line at each environment and for the variance analysis.

For the analysis the Perkins and Jinks' model was

employed. In this model we have:

= u + Di + Ej + gi.

yij J

28

yij being the performance of the ith line at the jth
environment; u, pOpulation mean; Di' additive effect of

genotype; Ej’ additive effect of environment; gi. the

3!
interaction of ith line with jth environment.

"’Di' Ej' and gij can be estimated as follows:

u = y../ ts

bi = Yi° /'8 'U

E y.j./ t -p

gij= yij - U ' Di ' Ej

t and s correspond to the number of lines and environments,
respectively. The next step in the analysis is the regres-
sion of gij for the individual lines onto the environmental
index (Ej). The analysis of variance for the regression

of qi on Ej requires an additional error term for testing

3'
the significance of the residuals. This error term arising
from within line within environment variation averaged over
all environments and lines was estimated on the assumption

of complete randomization of lines within each environment.
Non-significant regression and residual mean squares is the
indication of the absence of genotype-environment interaction.
Presence of genotype-environment interaction is evidenced by

significance of either regression or residual mean squares

or both. Significance of the regression mean squares alone

29

is the indication that genotype-environment interaction is
a linear function of the environment and depending on the
amount of error, prediction of genotype-environment inter-
action can be made. On the other hand, significance of
the residual mean square is the indication of no simple
relationship and as a result, no prediction can be made.
Comparison of the magnitude of mean squares would be of
value if both regression and residual mean squares are
significant. It can be shown mathematically that the re-

gression lepe 8 (obtained from regression of gi on Ej)

3'
is exactly one unit less than the stability parameter, 8,

(obtained from regression of yi on Ej) or B = l + B.

j
Analysis of variance for the regression of individual lines
on environment can be combined in one table as develOped
by Yates and Cochran.

3. Inheritance Of Test Weight

For studying the inheritance of test weight, the
diallel technique as developed by Jinks and Hayman (24,
27) was employed. Seven parental soft wheat lines were
deliberately selected and all pOSSible crosses were made
in the greenhouse in the Winter of 1966. Lack of recipro-
cal effects was assumed and seeds from each cross were
bulked without regard to the parental sex. F1 seeds were

harvested and along with the selfed seeds of the parents

were planted in peat pots in April, 1967. The F1 seedlings

30

were kept in the vernalization chamber for a period of 8
weeks. After this period, individual seedlings were trans-
planted in pots and were transferred to the greenhouse.
Because of insufficient seed and the abnormal effect of
greenhouse environment on test weight (and kernel weight)
no observations were made on F2 seeds arising from F1
plants. After maturity in the greenhouse a total number

of 200 seeds from each cross and each of the parents was
taken randomly for planting in the field at East Lansing.

A completely randomized design with 2 replications was

used. The F2 seeds from each cross and each of the parents
were divided into two lots. Each lot consisting of 100

seeds was used to plant a row 3 feet long. Planting was
effected in mid-October (later than usual planting time

in East Lansing). Winter killing of some seedlings in

each plot and a total loss of 3 plots occurred. Maturity

and harvesttime was characterized by heavy rains with con-
sequent lodging in many plots. After maturity each plot

was harvested separately and micro-test weight determinations
were made. Test weight results for the 21 crosses and the

7 parental lines were averaged over the two replications.
Prevalence of such environmental stresses during growth and
maturity produced kernels with low test weight and low ker-

nel~weight.

31

Due to the high stress situation described above,
it is unlikely that any inheritance pattern revealed by
analysis would have general application; however, it was
decided to make a brief genetical analysis by the Jinks-
Hayman Wr, Vr graph method. For such analysis, variances
of the arrays (Vr) and the covariance of each array with
the non-recurrent parents (Wr) were calculated and the
regression of the latter on the former was fitted. A
regression lepe of unity is the indication of the ful-
fillment of the hypotheses (such as lack of epistasis)
underlying this technique. Deviation from a slope of unity
would be due to the failure of one or more of these assump-
tions; however, this deviation can be tested by apprOpriate
statistical procedures. The arrangement of the points'
representing the parents along the regression line indicates
the dominance relationship of the parents; points toward
the origin having a higher proportion of dominant genes and
vice versa. The position of the line indicates the average
degree of dominance. The line passes through the origin if
dominance is complete (H = D). Intersection of the line
with Wr axis in the positive side is an indication of partial
dominance (H<D), and on the negative side is an indication
of over dominance (H>D). The parabola Wr2 = Vr Vp (Vp being
the variance of the parents) limits the area where the points

(Vr, Wr) may mathematically occur.

RESULTS AND DISCUSSION
1. Relationships Among Physical PrOperties And
Quality Characteristics Of The Kernel

Test weight (standard and micro), (1000) kernel
weight, (1000) kernel volume, (100)kernel length, (100)
kernel width, and the length width ratio appear in the
Appendix Table A. Packing efficiency (expressed as per-
centage of the air comparison chnometer cup volume
occupied by the grain), density, flour yield (%), pearl-
ing index (%), and protein (%) data are also included.

The correlation between micro-test weight and the
standard test weight was 0.99, suggesting a reliable pre-
diction of the standard test weight by the micro—test.

This relation is shown in Fig. 2.

Test weight is the product of density and the volume
of the grain occupying a certain volume. For the purpose
of simplicity, henceforth, the second component will be
represented as packing efficiency which is different from
that component by c/lOO, c being the volume of the container.
These two characteristics differing from each other only by I
a constant, will be used hereafter interchangeably. The
correlation between test weight and packing efficiency is

highly significant Fig. l—d, with r = 0.96 indicating the

32

33

minor importance of the other component, i.e., density.

A low correlation coefficient of about 0.16 was obtained
between test weight and density Fig. l-c. This relation-
ship was also evidenced by a very low standard partial
regression coefficient for density (.23), and a very high
value (.98) for packing efficiency, in a linear model using
these two as dependent and test weight as independent
variables; however, in the model:

Test Weight = bo + b1 (packing efficiency x density)
the correlation is near unity (r = 0.99). Packing efficiency
is associated to a higher degree with test weight and seems .
to be a varietal characteristic. For more accurate studies
of test weight, only this component should be taken into
consideration.

Among other properties, the regression of test
weight over length width ratio is highly significant
(r= -0.62). Stated in another way L/W ratio is responsible
for 39 percent of the variability in test weight. Roberts
(49) in his studies reached the same conclusion; however,
he indicated that with a constant L/W ratio, higher kernel
volumes would correspond to higher test weights. In the
linear equation:

Test Weight = bo + b1 (L/W) + b2 (kernel volume)

assuming L/W represents the kernel shape, the partial cor-

relation coefficient for L/W had a value of -0.57 and a

 

 

34

value of 0.0007 for kernel volume. Therefore, by keeping
the shape of the kernel constant (if shape is assumed to
be represented by L/W ratio) there would be no gain in
test weight if kernel volume is increased; however, more
than 60 percent of variation in test weight remains unex—
plained. Undoubtedly, factors such as depth and width of
the crease, shriveling of the kernel and other inherent
shape characteristics influence the test weight.

Concerning the relation betWeen other physical pro-
perties, the correlation between kernel width and kernel
volume is highly significant (r = 0.9D, Fig. l-a. A sig-
nificant high correlation was also obtained between kernel
weight and kernel width (r = 0.91), Fig. l-b; however, the
correlation between length and volume (or weight) is sig-
nificant but the r value is much lower (0.44). Kernel
volume as calculated by the formula:

V=L(W/2)2K1r
was significantly correlated with the kernel volume as
~measured by the pycnometer (r = 0.95). In this formula V
being volume of the grain, W and L representing width and
length respectively. K is a constant and is used because
the shape of the grain is not a cylinder.) This high cor-
relation indicates that by having values for length and
width we can closely estimate the volume of the grain. It
further indicates that in the study of components of kernel

volume (or kernel weight) such a breakdown is pertinent.

35

  
 

  
 

I
TO ”7‘200I 002! VI 5707900570!
I I

KERNEL 'IOTH
KERNE L WIDTH

 

JJ‘LLIJJAJIILI

4 (Illililnlilnlilil
I00 I” zlo m 2‘0 55 270 285 300 15.0 270 290 31.0 33.0 350 370 390 4L0 430 ‘50

111111

 

KERNEL VOLUNE KERNEL WEIGHT

d

YIO 79590 IZGX

Y' I 365 90.000“

  
  

564

PACKING EFFICIENCY

 

1LL111111lJJJlLlLL
7O 37‘ 3.. 394 ‘02 4I0 N. m ‘34 ‘42 Q”

 

TEST ICIGH‘Y TEST IEIGHY

Figure l-a Relationship between (100) kernel width (cm.) and
(1000) kernel volume (ml.); b, Relationship between
(100) kernel width (cm.) and (1000) kernel weight
(gms.); c, Relationship between density and test
weight (gms/pint); d, Relationship between test weight
(gms/pint) and packing efficiency (%)

MICRO TEST WEIGHT

4o.oI-

39.0

38.0

37.0

36.0

35.0

34.0

33.0

32 .0

3| .0

30.0

ro

TTTTT

I j

I T

 

\

I
L

I l
370

l

36

Y=|.8|7+0.077X

 

lllllllLlJLllllllll

376 386 394 402 4|O 4I8 426 434 442 450

STANDARD TEST WEIGHT

Figure 2. Relationship Between Standard Test Weight

(gms/pint) And Micro Test Weight (gms/
47 ml. glass jar).

37

As far as the relation between test weight and L/W
ratio and its practical implications is concerned, we should
either select for short or wide kernels. There are limits
to these values and small kernels with low L/W ratio would
not be of interest to the breeder because of the direct
effect that this type of kernel might have on the yield.

On the other hand, selection for large but plump kernels
would lead not only to a higher test weight but also a higher
yield. Visual selection of large and plump kernels at early
generations would be effective in increasing both test weight
and yield. Any cross giving shriveled and small kernels
should be discarded and only crosses with superior kernel
type should be maintained. At the beginning of develOpment
of short strawed varieties at the International Maize and
Wheat Improvement Center in Mexico, there were some problems
with the low test weight of the short strawed lines; however,
rigorous selection for large and plump kernels could very
well have been one of the effective procedures for overcoming
this problem.

The correlation coefficient between test weight and
flour yield is of very low magnitude (r = 0.03) and nonsig-
nificant, supporting the reports that test weight is a poor
estimator of flour yield. The reports on the relation of
test weight and flour yield is so contradictory that any

further treatment of the subject involves more extensive

38

data and experience. This contradiction could be partly

due to differences between techniques and instruments that
have been used for flour extractions by different authors.
Low test weight could be either due to shriveling of the
kernels or to the inherent shape of the grain. As far as
low test weight is reflected by kernel shape, there would

be no reason to expect a lower test weight of this type to
give lower flour yield. The Yorkstar variety which was
developed in New York State has always given low test weight
even in the best environment where no kernel shriveling
occurred. Yorkstar does have some problems with consistency
of performance but flour yields obtained from this variety
on frequent occasions have been as good or better than other
commercial varieties with higher test weights (60). The
same situation has been reported to be true with the Blueboy
variety developed in North Carolina (42). A low correlation
was obtained between test weight and kernel protein (r=0.1l).
The correlation between test weight and pearling index was
also of a low magnitude (-0.21). These low correlations
were probably due to a greater variation of these prOperties
within the same bulk of a variety.

Concerning the relation between kernel size and test
weight and quality characteristics, the data for test weight,
flour yield (%), pearling index (S) and protein content
appear in Table 1. The t values for all the possible compari-

sons appear in Table 2.

39

 

 

 

 

 

 

 

 

 

 

 

7 .7 .7 7
7 7 .7 C. 9 9 9 9 T. T. T. T. Z Z Z Z
M. 0. .6 .o 0. 0. w .v 0. .1 .z .z .I. o E E
v. w w w m. w w v. m m w % u m. .1... 9. 582
.mm 0.00 0&0 900 0.m0 ~40 m.~0 m.m0 0.2 0.3 0.2 0.2 20 30 0M0 0m0 comm m
.00 m40 0.00 0.00 m.00 0.00 ~.~0 5.0.0 0.0.5 92 0.0L m.0H 000 30 $0 20 mm cozcm b
.m0 «.3 0.2 TS 0.2 0.8 TS 0.3 0.: T: t: T: 20 80 00 R0 2832 0
.00 0.0m m.m0 o.00 T00 o.00 m.n0 5.00 0.: mg: 0.: T: 30 30 30 ~00 monocow m
.3 SS 93 0.3 0.3 0.8 58 0.00 52 52 02 0.00 R0 80 R0 :0 3:50 0
.on 060 0.2 93 0.3 500 0.3 ~00 0.: 0.8 TB 5: 30 S0 20 :0 0033. m
.8 ER 0.0m 0.3 93 500 0.3 o.00 5.2 0.2 92 0.2 S0 030 3.0 20 1:5 N
.00 m.o0 0.00 T00 50 060 0.m0 0.0 NS: ~.: 0.: 0.3 0m0 30 000 0m0 umou com H
m m z a m m z a m m z a m m; z .0
meGH mcflanmom namww unon cwmuoum usmfioz puma .Hm> .oz
xasm can .HHmEm .Edflcoz
.mmumq «masono 0 oucH nwumummmm mmfiuoflum> 0 Mom muse va xoncH
mcHHHmmm pad va namflw usoam .va :wmuoum .Aucwm\mfimumv unmwmz umme .H manta

Table 2.

40

The t Values For Comparisons Of Different Kernel

Sizes For Test Weight, Flour Yield, Pearling

Index And Kernel Protein

 

 

Test Weight

 

Large v.
Large v.
Large v.
Medium v.
Medium v.
Small v.

Pearling Index

Medium
Small
+
Bulk
Small
+
Bulk
Bulk+

 

Large v.
Large v.
Large v.

Medium v.
Medium v.
Small v.

Medium
Small
Bulk+
Small
+
Bulk
Bulk+

(t)
0.143
2.935**
1.412N'S'
3.945**
2.081N'S'
3.476*

N.S.

(t)
0.467N'S'
7.280**
3.392*
5.197**
1.315N°S'
8.174**

Flour Yield

 

Large v.
Large v.
Large v.
Medium v.
Medium v.
Small v.

Kernel

 

Large v.
Large v.
Large v.
Medium v.
Medium v.
Small v.

+ Bulk - sample prior to size grading

** Significant at 1%

* Significant at 5%

N-S.Not Significant

(t)

Medium 1.119N°S°
Small 1.626N'S'
Bulk+ 1.510N°S'
Small 3.448*
Bulk+ 2.492*
Bu1k+ 1.260N‘S‘
Protein

(t)
Medium 8.277**
Small 12.4ss**
Bulk+ 6.479**
Small 9.335**
Bulk+ 3.193*
Bulk+ 3.08l**

 

41

For test weight, the three comparisons involving
the small sized kernels are significant. The 1000 kernel
weight means for medium and large size kernels over all the
8 varieties are 36.56 and 43.64 grams respectively; however
the difference between the test weight of these two groups
is not significant. These results support the previous
reports concerning the effect of kernel volume (or weight)
on test weight. Although the kernel size in the large
group is substantially higher than the medium group (119%)
the difference in test weight is not significant, mainly
because the kernel shape remains the same in both groups.
The lower test weight of small sized kernels could be either
I due to their shriveled condition or their lower density.
In a spike of wheat the small kernels arise from the late
develOped florets which get filled later in the season.
This may result in a slight shriveling of the small sized
kernels within a Spike and consequently within a variety.
No measurements on the density of these groups were taken
but the smaller kernels may have a lower density than the
larger. This is evidenced by the higher protein content
of the larger kernels when compared with small ones.

As far as the relationship between kernel size
(within varieties) and protein content of the kernel is

concerned, there seems to be a definite relation between

42

kernel protein and kernel size. The t values for all the
comparisons are significant.

In the wheat plant, flowering of the Spikelets begins
slightly above the middle of the Spike and then proceeds in
both directions (34, p. 310). Percival (44) stated that
there is a close relationship between the weight of the
grains and their position on the spike. The heaviest ker-
nels occur where flowering begins (44). The reports by
Levis and Anderson (35) and McNeal and Davis (37) showed
a progressive decrease in protein content from large to
small kernels. Levis and Anderson reported that the upper
Spikelets (where small sized kernels develop) had a decidedly
lower protein content than the remaining Spikelets. They
also reported that within a spikelet the middle kernel
tended to have a higher protein content. Within a spikelet,
there is a definite relationship between kernel weight and
protein content. The kernels heavier than the average, tend
to be higher in protein. McNeal EE.E£' (37) reported that
a higher, but non-significant protein content was obtained
from the central kernels. The protein content of the
earlier formed and matured kernels was the highest.

The results obtained in this study are in complete
agreement with previous studies. Smaller sized kernels
with lower protein content arise either from the later

develOped florets in the Spikelets located in the middle

43

of the spike or later develOped Spikelets at the top or
bottom of the Spike. In contrast larger kernels with
greater protein content originate either from the spike-
lets in the center of the spike or from the florets which
develop at an earlier time in the spikelet.

Concerning the relation between kernel size and
flour yield, the results are inconsistent from one variety
to another. One would expect to obtain a higher percent-
age of flour from the larger sized kernels than the smaller
sized ones; however, such differences were not obtained.
These results are contradictory to what was reported by
Bailey (5). The data is not extensive and the inaccuracy
of a small laboratory mill (55, p. 138) and small sample
size could very well have been responsible for such results.

As far as the relationship between kernel texture
and kernel size is concerned, there seems to be a definite
trend for smaller kernels to have lower pearling index
values. These results could be attributed to a higher bran-
endosperm ratio of the smaller kernels. Bran has more

elasticity than the endosperm and it resists attrition.

44
2. Effect Of Environment On Test Weight And Other
Kernel Characteristics

a. Pattern Of Change In Kernel Characteristics
Under Different Environments

The data for test weight and kernel weight averaged
over the 3 replications appear in Tables B and C. The
data for length and width are presented in Table D. Micro-
test weight, packing efficiency, density and flour yield
data appear in Table E, all in the Appendix. The 21 possible
correlation coefficients of differences between test weight
and 1000 kernel weight are presented in Table 3.
Table 3. Correlation Coefficients Of Differences

Between Test Weight And 1000 Kernel
Weight At Paired Locations

 

 

Pairs of Correlation Pairs of Correlation Pairs of Correlation
Locations Coefficient Locations Coefficient Locations Coefficient

 

T61 - K71 0.53 .H71 - B91 0.40 K81 - L41 0.90
T61 - K81 0.47 H71 - 101 0.57 B91 - 101 0.75
T61 - B91 0.13 H71 - 851 0.90 B91 - 851 0.89
T61 - 101 0.57 H71 - L41 0.74 B91 - L41 0.87
T61 - 851 0.90 K81 - B91 0.85 I01 - 851 0.93
T61 - L41 0.82 K81 - 101 0.82 101 - L41 0.81
H71 - K81 0.69 K81 - 851 0.95 851 - L41 0.88

 

Regression analysis indicated that reduction due to
regression is significant for all but one of the possible

pairs, i.e., T61-B91. The majority of correlation coefficients

45

are high, indicating a lower amount of error or deviation
from linearity. Therefore, in most environments the factors
affecting kernel weight also affect test weight in the same
direction. Reduction in test weight cannot be attributed

to reduction in kernel weight per se. Adverse environmental
conditions not only reduce weight of the grain, but may also
cause some shriveling of the grain leading to lower packing
efficiencies and consequently lower test weights. In some
cases environmental factors may alter the conformation and
shape of the grain without reducing its weight. Occurrence
of rain during the last stage of maturity could not affect
weight of the grain significantly but may result in kernel
expansion and thus a low test weight. The degree to which
this occurs would certainly determine the size of the cor-
relation coefficients.

The correlation coefficients of differences between
test weight-flour yield and 1000 kernel weight and flour
yield (%) at paired locations appear in Table 4. Location
means for flour yield (%), test weight, and 1000 kernel
weight are presented in Table 5.

The correlation coefficients obtained for both test
weight and kernel weight are inconsistent. A higher flour
yield was expected from locations with higher test weight
and kernel weight means. This expectation was mainly due to

a higher pr0portion of endosperm from larger sized kernels.

46

 

 

 

 

Table 4. Correlation Coefficients Of Differences
Between Test Weight And Flour Yield;
And Of Differences Between 1000 Kernel
Weight And Flour Yield
Pairs of Test Weight vs. 1000 Kernel Weight
Locations Flour Yield vs. Flour Yield
IOl - L41 -0.22 0.65
I01 - H71 0 04 0.46
L41 - H71 0.82 0.52
L41 - B91 0.89 0.79
H71 - B91 0.34 0.02
Table 5. Location Means For Flour Yield, 1000 Kernel

Weight, And Test Weight From 30 Lines
Averaged Over 2 Replications

 

1000 Kernel Test Weight

Location Flour Yield(%) Weight(gms) gms/pycnometer cup

 

101
L41
H71

B91

64.86 31.14 34.40
64.98 25.26 32.92
68.21 40.85 38.02

69.31 34.16 37.42

 

47

In the previous experiment in which kernels within varieties
were separated into different sizes, no definite relation
between flour yield and kernel size was established. By
considering the first 2 environments as one group (I01 and
L41) and the last 2 environments as another group (H71 and
B91) the trend is for larger flour yields from locations
with high test weight and kernel weight. This grouping
represents extremes of locations for test weight and kernel
weight and it is not known what the flour yields of the
intermediate group would be.

The means of 100 kernel length and 100 kernel width
for the 3 locations appear in Table 6. The correlation
coefficients between locations for both length and width

are presented in Table 7.

Table 6. Mean 100 Kernel Length (cm.) And 100 Kernel
Width (cm.) For Locations Obtained From 30
Lines With 1 Replication

 

 

 

Length Width
H71 61.32 33.20
101 59.91 28.58

L41 57.64 26.18

 

48

Table 7. Correlation Coefficients Between Locations
For Length And Width of The Kernel

 

Length Width
H71 101 L41 H71 101 L41
H71 1.00 H71 1.00
101 0.83. 1.00 101 0.49 1.00
L41 0.70 0.78 1.00 L41 0.02 0.33 1.00

 

. Although the location H71 is 102% of location 101
in kernel length, it differs substantially (116%) from this
location in kernel width. Clark and Bayles (14) stated
that kernel length is a genetical character and the kernel
reaches its full length several days before maturity. These
differences are readily explainable. In H71 and 101 kernels
developed and reached their full length. Heavy rains
followed by lodging of the lines at maturity in 101 prevented
maximum starch formation during the final days of maturation;
therefore the seed shrank as moisture content decreased at
maturity. The slight decrease of length in 101 compared
with H71 could be attributed to the extremely favorable en-
vironmental condition at H71 in 1968.

The picture at location L41 was completely different.

This location was characterized by plots heavily infested
with weeds. This situation resulted in a stress condition

fOr the wheat plants at the very beginning of growth and

49

all through the whole life of the plants. Prevalence of
such continuous environmental stress during the life his-
tory of the plant resulted in kernels small in both dimen-
sions. The high correlation coefficients between paired
locations for length are indicative of similarity in the
pattern of length development under different environments.
Such high correlations were not obtained for width. The
high correlation obtained between kernel volume (or weight)
and kernel width and these results indicates that environ-
ment plays an important role in determining kernel volume
(or weight). On the other hand length of the grain seems
to have higher heritability and any increase in kernel
volume or kernel weight (and as a result in yield) due to
fertilization, weed control, and other cultural practices
is reflected mainly in increasing the width of the grain.

b. Estimation Of Variance Components And
Heritability.

The magnitude of variance components for both
set A (that includes 851 and L41) and set B (that does not

include 551 and L41) are presented in the following table:

Table 8. Estimates Of Variance Components For Test
Weight in Set A and Set B.

 

2 2 2 2 2 2
Set °v °v1 °vy °vly 0e 0p
A 106.46 3.46 5.71 ' 40.68 34.74 191.05

B 87.65 3.13 3.38 19.20 22.76 136.12

 

50

For the purpose of comparison between the two sets,

each variance component was expressed as percentage of its

correSponding total phenotypic variance - a: as follows:

Table 9. Variance Components In Set A & Set B Expressed
As Percentage Of Their Corresponding Total
Phenotypic Variance.

 

Set‘ 2 2 2 2 2 2
av CV1 CIvy Ovly OIe Up

A .557 (H) 0.018 0.030 0.213 0.181 1.000

B .644 (H) 0.023 0.025 0.141 0.167 1.000

 

The data and the analysis of variance appear in Appendix
Tables F, G, and H. The differences between the lines are
highly significant. The effects of location and year in
both sets are highly significant. Among interaction
sources of variance, years x lines and locations x lines
are not significant in either set but the second order
interaction, i.e., lines x locations x years is highly sig-
nificant in both sets.

Elimination of L41 and 851 from the analyses did not
change the results of the analysis of variance. Consider-
ing the magnitude of variance components this elimination

had a great effect in decreasing the magnitude of oily

This is not surprising because the second order interaction
as explained by Comstock (16) "...has to be due to one or

more aspect of environment for which the pattern of variation

51

among locations differs from year to year." Plots at
locations L41 and 851 were heavily infested with weeds.
This was accompanied by some lodging in 851. Competition
for nutrients, water, light, and space resulted in stunted
plants which produced small shriveled kernels and con-
sequently a low test weight for all lines. The effect of
lodging in lowering test weight is a well known phenomenon
(17, 33, 43, 59).

Elimination of these 2 locations did not have a
major effect either on the value of 03y or the value of
031 expressed as percentage of their corresponding total
phenotypic variance. The effect on the former was higher
because both locations had a similar situation in the same

year, 1968. This elimination caused a decrease of about

2
vly

2
v1

and 03y , even though these locations have been eliminated.

. . . 2
This high magnitude of ovly

differential response of the varieties to different environ-

7 percent in the second order interaction, i.e., 0

The magnitude of this component is still higher than c

is an indication of

ments. A number cf factors exist which affect test weight
seriously. Among the lines planted were those which had
field resistance to either mildew or leaf rust or both.
Prevalence of one or both of these diseases at one location
and in one year may have a profound effect in lowering the

test weight of susceptible lines in that location and in

52

that year and would lead to a larger second order inter-
action variance. This is also true with lodging, a serious
problem in rainy seasons. The environment in East Lansing
(101) in 1968 was characterized by heavy rains prior to
harvest. This resulted in severe lodging of many lines;
however, among the lines used in this study some, such

as Arthur, were resistant to lodging. Differential
reSponse of the lines to this particular environmental
stress may have contributed to a large second order inter-
action. Shriveling of the kernel is directly related to
test weight. Among the lines tested and used for this
study some lines like Yorkstar are very sensitive to
environmental stress and under such an environment Yorkstar
and lines similar to it shrivel drastically and will bring
about a differential response.

Test weight is a complex trait and differential
response of the components under different environments
may result in a differential response with regard to test
weight and hence a larger second order interaction (3).

Statistically non-significant locations x lines and
years x lines sources of variation and relatively small
values of their corresponding variances are indicative of
no consistent year and location effect on differential
varietal response. Non-significant lines x locations and

lines x years sources of variance may result from lines

53

under study which were either commercial varieties or ad—
vanced breeding lines selected for adaptation for many
years.

Considering the present results, little is gained
by including years and locations in the experiment in a
balanced manner. Thus when one considers only test weight
in a breeding program, he should include a sample of years
and locations which are likely to be encountered and carry
the analysis with respect to environment (36, 40).

In any breeding program attention is rarely focused
on test weight alone. Yield is of great concern to the

breeder and because of significant and large.values of “51

and 03y for yield, the breeder may have to replicate his
trial both in time and space.

Estimates of heritability for single plots for set
A and set B are presented in Table 9. Estimations for line
means are much larger and are of the magnitude of 0.94 and
0.95 for set A and B, respectively. A lower estimate of
heritability for single plots for set A is mainly due to a
substantially higher oily in this set than in set B. The
great amount of difference between the two estimates of
heritability~-on the basis of single plots and on the basis

of variety means is mainly due to the smaller contribution

of non-genetic variance to the line means (22).

54

The estimate based on differences among line means
is the characteristic of the environment and sets of
material used in this study, while the estimate based on
single plots is the characteristic of the trait-test
weight (22).

By having such estimates of heritability we would
be able to make predictions of gains in any hypothetical
population.

These estimates are characteristic of the environr
ments, and sets of materials used. Another experiment with
another set of materials under dissimilar environments may
result in different estimates of heritability. A knowledge
of heritability under any set of environments and known
materials would be helpful in arriving at a more efficient
selection scheme.

The Effect Of Environment On Test
Weight And Its Components

The analysis of variance for test weight, packing
efficiency, and density appear in the Appendix Tables I,

J and K. The data are presented in the Appendix Table E.

Differences between locations and lines for test
weight and its components are significant. Lines x locations
interaction source of variation is also significant for test
weight and its components. Estimates of variance components
for test weight, packing efficiency, and density appear in

the following table:

55

Table 10. Estimates Of Variance Components For
Test Weight, Packing Efficiency, And Density

 

 

2 2 2 2
av Ov1 0e 0p

Test Weight 0.6106255 0.5260645 0.1485356 1.2852256

Packing . ‘ -
Efficiency 0.3884833 0.2291525 0.0765506 0.6941864

Density 0.0000335 0.0000377 0.0000342 0.0001055

 

To make comparisons among variance components of
test weight and variance components of packing efficiency
and density, every variance component was converted into
the percentage of its corresponding total phenotypic

variance as follows:

Table 11. Estimates Of Variance Components Expressed
As Percentage Of Their CorreSponding Pheno-
typic Variance

 

O2 O2 O2 02
v v1 ‘e p
Test Weight 0.475 (H) 0.409 0.116 1.000
Packing
Efficiency 0.560 (H) 0.330 0.110 1.000
Density 0.317 (H) 0.358 0.325 1.000

 

Heritability estimates for all three factors appear
in the first column of the above table. Genetic variance

estimates contain interaction variances (2, p. 98, 20, 31)

56

and as a result larger estimates of heritability should

have been obtained. On the other hand, the peculiar char-
acteristic of L41 has led to a larger lines x locations
interaction variance and consequently a lower estimate for
this parameter resulted. The eXperiment was not conducted
solely to make estimates of heritability, but, within limits,
to make a comparison among the variance components of test
weight, packing efficiency, and density.

Table 11 shows that the prOportion of genetic vari-
ance for density is lower than the other two components,
i.e., 031 and 0:. The variance between plots within
locations, 0: , is greater than the variance of differ-
ences between the lines and this may imply a greater effect
of micro-environment, sampling error, or error of measure-
ment on density. The range in density among commercial and
advanced breeding lines is very narrow and density does not
seem to be a varietal characteristic. Low heritability for
density obtained in this study supports this idea. The con-
tribution of density as a component to test weight is very
low compared with the other component-packing efficiency.
Packing efficiency is a varietal characteristic and a major
pr0portion of variability in test weight is being controlled
by this component. Taking out the variability due to density
and dealing with packing efficiency alone, an increase of
about 18 percent in heritability was obtained when compared

with test weight.

57

The heritable portion of test weight is packing
efficiency and for genetical studies we should eliminate
the density factor and deal only with the former component.
For practical purposes and in selection programs test
weight should be the basis of selection because of the
simplicity involved in measuring this property and because
the correlation coefficient between test weight and packing
efficiency is very high (0.96).

c. Stability Study

The mean over the three observations for the
22 lines in the 15 environments appear in the Appendix
Table F. For the purpose of illustration, the analysis of
variance for the regression of individual genotypes on
environment are presented only for Genesee and Yorkstar,
Tables 12 and 13. The combined analysis of variance for
all 22 lines appear in Table 14. Table 15 summarizes the
results of the analysis for all the lines under study.

In this analysis the absence of genotype-environment
interaction will result in 8.3,0 and 8= 1. Values of 0
and l for these parameters respectively are not necessarily
an indication of the absence of genotype-environment inter-
action. Table 12 shows that with Genesee neither the re-
gression mean square nor the residual mean square are sig-
nificant, suggesting the absence of genotype-environment

interaction for this variety. Eleven of the lines under

58

Table 12. Analysis Of Variance for Regression Of
Genotype-Environment Interaction Over
Environmental Index For Genesee

 

 

 

Source DF MS
Regression l 3.71 N.S.
Residual 13 31.16N'S°
Error 660 49.31

 

Table 13. Analysis Of Variance For Regression Of
Genotype-Environment Interaction Over
Environmental Index For Yorkstar

 

 

Source DF MS

 

Regression 1 407.29**
Residual 13 41.49N'S°
Error 660 49.31

 

Table 14. Combined Regression Analysis For The
22 Lines Planted In 15 Environments

 

 

 

Source d.f. ss . MS
Lines ' 21 26990.70 1285.27**
Environments l4 93978.78 6712.77**
Lines x Env.: 294: 17144.08:
Heterogeneity
between regression 21 6174.65 294.03**
Residual 273 10969.42 40.18N'S'

Error 660 32546.00 49.31

 

59

Table 15. Average Performance, Slope, And Genotype-
Environmental Relationship Of 22 Lines
Planted In 15 Environments.

 

G x E Present

 

 

Av. , GXE
Line Performance 8=1 + 8 Absent Res. Reg.

Genesee 414.82 1.029 X

Avon 412.93 1.185 p X
Talbot 414.09 0.856 x

Monon' 414.27 1.106 x

Arthur 424.95 0.682 X
Reed 430.35 0.428 X
Yorkstar 393.67 1.309 X
.A1224 417.35 0.836 X

A2554 417.47 0.708 X
A2739 416.82 1.347 X
A2747 420.22 0.912 X

A3116 399.64 1.233

A3136 412.13 0.713 X
A3141 414.27 0.655 X
A4129 394.44 1.066 X

A4528 397.09 1.102 X

A5044 417.87 0.850 X

A5127 408.60 1.124 X

A5131 412.80 1.147 x

A5132 411.42 1.374 x
A5134 411.42 1.378 X
A5266 419.89 0.959 X

 

60

study fall into this category and do not Show a significant
genotype-environment interaction. The stability parameter,
8 for these lines does not deviate significantly from unity.
These results are in agreement with the past experience
with these lines. Genesee has been the standard variety
over many years and it has been more predictable with
respect to environment. Yorkstar and 10 other lines showed
a significant genotype—environment interaction whereas most
of the genotype-environment interaction for Yorkstar and 9
other lines (with significant genotype-environment inter-
action) was accounted for by the linear regression on the
environmental values. Only one line (A3116) had a signi-
ficant residual mean square. Nevertheless in the combined
analysis in Table 14, the mean square for the residual is
not significant, mainly because sum of squares and degrees
of freedom are pooled values. Mean square for heterogeneity
of regression lines, Table 14, is significant, indicating
definite differences among the stability parameter of the
lines.

Depending on the breeder's ideal, either criterion
of stability, i.e., B = l or 8= 0 may be adOpted. If
adaptation over a broader range of environments is desired
(B=0) such lines as Reed and Arthur would provide excellent
materials. In these lines, above average stability (B<l)

is accompanied by a superior performance over aU.environments.

61

This relation also exists among other lines and the regression
of average performance over all environments on the stability
parameter resulted in a negative but significant correlation.
(r =-.57). This might be a reflection of the fact that most
of the lines under study have been selected over many years
for superior performance over a wide range of environments.
These results are in contrast to previous reports that the
stability parameter is positively correlated to average
performance (18, 45). Lines such as Reed and Arthur
undoubtedly will Show some degree of genotype-environment
interaction. This will be mainly due to higher positive
interactions in the poorer environments and to a much

lesser extent due to negative interactions in the more
desirable environments. Variance or standard deviations

over all environments might be equally desirable if abso-
lute stability over all environments (B=0) is sought. To
reduce the effect of different means, coefficient of vari-
ation may provide a better index of stability.

If complete absence of genotype—environment inter-
action is desired lines with B=1 and high Di should be
developed. Such lines will keep their relative perform-
ance to other lines in all environments. The Di was shown
to be highly heritable for test weight. The 8 also has
been shown to be a heritable characteristic and it has

been reported that selection for B has been effective.

62

Crosses involving Genesee type for lack of genotype-environ-
ment interaction and Reed or Arthur for their performance
and selection in the segregating generations for lines with
B = 1 and superior performance may result in lines closer

to the breeder's ideal.

~‘TT
‘I

~;r~

3. Inheritance Of Test Weight

1,,

Micro-test weight data averaged over 2 replications

and the Wr, Vr graph appear in Table 16 and Fig. 3 respect-

ively. The slope of the line (0.97) does not deviate from

 

unity, an indication of the fulfillment of the assumptions
underlying the diallel cross analysis.p It also suggests
that the differences between the parents used in this study
are mainly due to additive and dominance effects, a system
uncomplicated by inter-allelic interactions. The arrange-
ment of the points representing the parents along the
regression line suggest that high test weight parents
(Genesee, Avon, and Monon) possess the higher proportion
of dominant genes and the other_four parents with low test
weight (10113, A4520, A1339, and A1290) possess the highest
proportion of recessive genes. There seems to be a definite
tendency for high test weight to be dominant over low test
weight.

The regression line intersects the Wr axis on the
positive side, an indication of partial dominance (H<D).

As mentioned before, the analysis was based on F3 seeds

3
6

 

 

 

 

00mm.mm omm0.m NNNO.N m0m.om om0.hm m0~.m~ m0a.0m omm.om oma.mm om0.mm ommad n
omh0.bm mhvo.0 mmm0.m o0m.0~ mNH.0~ 000.mm mm0.om onm.mm mmm.mm mmma< 0
m0mm.mm mmah.m onm0.m m00.mm omm.0m o0m.om ooa.m~ mm0.m~ omm0¢ m
0mmh.0m mo0m.m mvoa.m mmo.mm o0o.mm m0o.mm on.>~ MHHdH 0
mmov.om m0~m.H Hmm0.o omb.am osm.om mmm.mm coco: m
m0mm.mm 0mm~.~ vmmo.a oa0.mm mmm.om co>< m
0mam.mm ammo.m mmh~.H mmm.om commune H
mm.“ a. S . . s m N a .60.... Emma”
mcoflumowammm m Hm>o_pmwmum>¢ HDHHMHQ h x b
m0 new 4 Mom Assn mmmao .HE 50 non mEmv usmflmz umma OHoaz mm .0H manna

64

 

 

In
4.50—
I.
6
I— O
F . .5
3.60.. 7
_ 4
2'70” Y: 09724 + (0.9686 10.1251x
" 2
_ e
l
_ 0
L80—
3
" O
0.90
llilLllJlLlILillilll
0.00 0.90 I .80 2.70 3.60 4.50

Vr

Figure 3. Wr-Vr Graph for Micro-Test Weight From
a set of 7x7 Diallel, F3.

65

coming from F2 plants. Considering the fact that the
amount of H at any one generation is halved over the pre-
vious generation (29), such a degree of dominance is not
surprising. In fact it would have been possible for the
previous generations to denote some degree of overdominance
(H>D). As far as the breeder of self-pollinated crops is
concerned, this possibility of initial overdominance has
less practical importance, since the potential value of
any cross is evaluated by homozygous lines generated from
this cross (28).

A Wr, Vr analysis of kernel weight for the same
diallel set led to the same general conclusions obtained
for test weight. There were, however, very slight changes
in the position of points representing each parent. AS»
far as low test weight is associated with shriveling of the
kernels, visual evaluation of kernels in earlier generations
would be effective in selecting against the genes responsible
for shriveling.~ Large, plump kernels may also give low test
weight, because of their intrinsic shape. Past experience
has Shown that this type of low test weight is of minor
consequence. Varieties with this type.of low test weight
may be as superior as high test varieties in quality
features. As the wheat producers are always penalized for
low test weight, the breeders are concerned about both

types of low test weight. In the case where kernel shape

1..€:¢—.fl- s ’ "7’ . - .'—-.'*-“.-'-r—. “at?

v' "
gray. sir

”xi—00.13. .

66

causes low test weight, visual selection is not feasible.
In earlier generations where the amount of seed is not
sufficient to run the standard method of test weight
determinations, the micro-test weight determinations would
be very valuable to the breeder.

Visual selection against shriveled kernels not only
improves test weight, but also has a direct effect in in-
creasing yield, because kernel weight is one of the

components.

_. A‘. n'

Ll_‘I-nl“

I?"

3'7, V!

I‘. Irv—ht...
L
,._- .—

“ .
~L—msIlT-ISn.

SUMMARY AND GENERAL CONCLUSIONS

Test weight is the product of density and the total
volume of the grain in a given unit volume. The latter
component expressed as percentage of the volume of the con-
tainer is referred to as packing efficiency and this com-
ponent has a much greater effect on test weight than density.
Because of the minor importance of density and its narrow
range in genetical studies of test weight, the variation
due to density should be removed and the analysis should
be carried out with respect to packing efficiency. Using
the analysis of variance procedure, a definite gain in
genetic variance was obtained when the effect of density
was removed from the test weight and the data were analyzed
with regard to packing efficiency.

A significant but negative correlation was obtained
between test weight and length width ratio. Less than 40%
of variation in test weight was shown to be associated
with length width ratio and the remainder remained unex-
plained. Within varieties kernel size had no effect on
test weight, mainly because shape of the grain remains con-
stant. Kernel shape and degree of shriveling are some of

the factors which cause low test weight. Visual selection

67

68

in the segregating generations against shriveled kernels

and for large plump kernels was suggested to be an effect-
ive procedure for improving test weight. Use of micro-test
weight would be an effective tool for discrimination against
the lines which, because of their shape, give low test
weight.

No relationship between flour yield and test weight
either between varieties or within varieties was established.
Low test weight is either due to shriveling of the grain or
intrinsic shape of the kernel. As far as low test weight
is reflected by the shape of the grain, no relationship
between flour yield and test weight should be expected.
Inaccuracy of the small laboratory mill is another factor
which adds to the complexity of the relationship between
test weight and flour yield. A more definite relationship
between protein content of the grain and kernel size within
varieties was established. Higher protein content of
larger sized kernels is partly due to the fact that the
Spikelets located in the middle of the spike contain larger
kernels with higher protein (35, 37). Within a Spikelet,
the larger kernels have higher protein contents. Within
varieties pearling index was related to kernel size.

Harder texture (lower pearling index values) of small
sized kernels was attributed to higher proportions of bran

endosPerm ratios. More elasticity of the bran and its

xiv—0"! a."wmn._ ’j‘hu. Q‘ “I'm

.l'-’."- “p

‘ huh-4.2.

f

69

resistance to attrition was suggested as responsible for
such relationship.

As far as the effect of environment on test weight
and other properties of the grain is concerned, length of
the grain was more resistant to environmental factors at
later stages of kernel development. Width of the grain
was more vulnerable to environmental changes. Test weight,

kernel weight, packing efficiency, density, and flour

yield were also affected by environment. With few exceptions,
reduction in kernel weight is usually accompanied by reduction
in test weight. Reduction in kernel weight per se could not
be responsible for reduction in test weight.

Using the analysis of variance procedure, the second
order interaction, V x Y x L, was demonstrated to be highly
significant. The first order interactions, i.e., V x L
and V x Y, on the other hand were not significant. In
breeding programs, the analysis should include a number of
environments which are likely to be encountered. Heritability
estimates, both on the basis of single plots and differences
among line means were made and a higher value was obtained
for the latter than the former. The high heritability
estimates obtained in this study suggested that selection
for test weight would be effective and will result in

favorable amounts of gain in test weight.

’1' “1.“?- n
I
‘i
_'~jr_ a r

.P'\! '._

{flat-“‘1."

70

Stability analysis for 22 lines under 15 environ-
ments revealed that some lines did not show any genotype-
environment interaction. Genesee and 10 otherilines fell
into this category and they were characterized by slopes

which did not deviate from unity Significantly. Varieties

I:
a"

0 ."'
.40.. -—

Reed and Arthur showed genotype-environment interactions;

A

however, most of the interaction was accounted for by

linear regression on the environmental index. A negative

but significant correlation was obtained between the

- w .. urn—a. x‘e‘cJ‘hS‘.‘ «m:

.4
l-a .‘v- v.

stability parameter, 8, and the average performance over
all environments. Depending on the breeder's ideal either
definition of stability, i.e., 8=1 or B=0 may be employed.
Arthur and Reed represent the ideal types if the absolute
stability (B=0) is sought. In these lines superior per-
formance is accompanied by regression slopes significantly
lower than unity. If the other type of stability is
desired (6=l), crosses involving lines with no genotype-
environment interaction (8=l) such as Genesee and superior
performance such as Reed or Arthur, would provide initial
materials from which lines with high test weight and slope
of unity can be selected.

An inheritance study of test weight revealed that a
simple genetic system is involved. Additivity and dominance
with no complications of epistasis was evident from the

analysis.

lo.

11.

BIBLIOGRAPHY

Aamodt, O. S., and J. H. Torrie. 1934. A simple method
for determining the relative weight per bushel of grains
from individual wheat plants. Canadian Jour. Research.
11:589—593. ‘

Allard, R. W. 1964. Principles of Plant Breeding. 2nd
Printing. John Wiley and Sons, Inc. New York. pp. 485.

"I .”——.A I. d AI.HFF
'i
.- “v.3—

Allard, R. W. and A. D. Bradshaw. 1964. Implication
of genotype-environment interactions in applied plant
breeding. CrOp Science. 4:503-507.

:gw‘u ._~u -

American Association of Cereal Chemists. 1957. Cereal
Laboratory Methods. St. Paul, Minn. Ed. 6. 528 pp.

Bailey, C. H. 1916. The relation of certain physical
characteristics of the wheat kernel to milling quality.
Jour. Agr. Science 7:432-442.

Bailey, C. H., and R. C. Sherwood. 1924. Report of
the Operation of the State Testing Mill. Crop Season
of 1923. Minn. State Bd. of Agr. Bul. 37.

Barmore, M. A., and P. K. Bequette. 1965. Weight
per bushel and flour yield of Pacific Northwest white
wheat. Cereal Science Today. No. 3:72-77.

Bates, E. N., G. P. Bondar, and R. L. Baldwin. 1909.
Removing smut from Pacific Northwest wheat by washing.
U.S.D.A. Cir. 81' pp. 15-160

Bayles, B. B., and C. A. Suneson. 1940. Effect of
awns on kernel weight, test weight, and yield on wheat.
Jour. Am. Soc. Agronomy 32:382-387.

Bode, C. E. 1959. Progress and aims of research at
the Soft Wheat Quality Laboratory. Cereal Science
Today. Vol. 4. No. 9:259-263.

Boshankian, Sarkis. 1918. The mechanical factors de-

termining the shape of the wheat kernel. Jour. Am.

71

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

Boerner, E. G. 1916.

72

Improved apparatus for deter-

mining test weight of grain, with a standard method
U.S.D.A. Bulletin 472.

of making the test.

Bucio-Alanis, L., and J. Hill. 1966.
and genotype components of variability. I. Inbred
Lines. Heredity. 21:387-397.

Clark, J. A., and B.

Bul. 795, 145 pp.

Environmental

B. Bayles. 1942. Classification
of wheat in the United States in 1939. U.S.D.A. Tech.

Comstock, R. B., and H. F. Robinson.
parameters, their estimation and significance. Proc.
6th International Grasslands Congress. 1:284-291.
National Publishing Company, Washington, D. C.

Comstock, R. E., and R. H. Moll.
environmental interaction.
Plant Breeding, 164-

1952. Genetic

1963. Genotype-
Statistical Genetics and

196. NAS-NCR 982.

Day, A. D. 1957. Effect of lodging on yield, test
weight, and other seed characteristics of spring
barley grown under flood irrigation as a winter annual.
Agronomy Jour. 49:536-539.

Eberhart, S. A., and W. A. Russell.
parameters for comparing varieties.

6:36-40.

Finlay, K. W., and G. N. Wilkinson.

sis of adaptation in plant breeding

J. Agr. Res. 14:742-754.

Fisher, E. A., and P. Halton. 1931.

weight of a thousand kernels.

422.

1966. Stability
Cr0p Science.

1963. The analy-

programs. Australian

A note on the

Cereal Chemistry. 8:418—

Hand Book of Official Grain Standards of the United
States. Production and Marketing Administration-Grain

Branch. U.S.D.A.

Hanson, C. H., H. F. Robinson, and R. E. Comstock. 1956.
Biometrical studies of yield in segregating p0pulations

of Korean lespedeza.

Harris, R. H., and L. D. Sibbitt.

Agronomy Jour.

48:268-272.

1942. The utility

of micro methods of test weight determination with

hard Spring wheat.

Cereal Chemistry.

19:458-467.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

73

Hayman, B. I. 1954. The theory and analysis of the
diallel cross. Genetics. 39:378-809.

Heizer, H. K., and R. L. Johnson. 1963. A.Wheat and
Milling Wheat Density. Wooster Soft Wheat Quality
Laboratory. Ohio Ag. Exp. Sta. 60-72.

Hlynka, I. and W. Boshuk. 1960. The weight per
bushel. Cereal Science Today. No. 8. Vol. 4:239-
240.

Jinks, J. L. 1954. The analysis of quantitative
inheritance in a diallel cross of Nicotiana rustica
varieties. Genetics. 39:767-788.

 

Johnson, P. V. 1963. Application of the diallel-
cross techniques to plant breeding. Statistical
Genetics and Plant Breeding.' 561-578, NAS-NCR 982.

Johnson, L.P.V.and R. Askel. 1959. Inheritance
of yielding capacity in a fifteen-parent diallel
cross of barley. Canadian Jour. of Genetics and
Cytology. 1:208-265.

Johnson, R. M. and T. F. Hartsing. 1964. Kernel
count as a measure of flour milling yield. Northern
Miller. Vol. 269. No. 10:22-26.

Johnson, W. J., H. F. Robinson, and R. E. Comstock.
1955. Estimates of genetic and environmental vari-
ability in soybeans. Agronomy Jour. 47:314-318.

Lamba, P. S. 1949. The relation of glume measure-
ment to kernel shape and size in wheat. Jour. Am.
Soc. of Agronomy. 41:167-168.

VLaude, H. H. and A. W. Pauli. 1956. Influence of

lodging on yield and other characteristics in winter
wheat. Agronomy Jour. 48:452-455.

Leonard, W. H., and J. H. Martin. 1963. Cereal Crops.
pp. 824. Macmillan Co. New York.

Levis, I. and J. A. Anderson. 1950. Variation in
protein contents of plants, heads, Spikelets, and indi-
vidual kernels of wheat. Canadian Jour. Res., F, 28:
71—810

 

74

Matzinger, D. F. 1963. Experimental estimates of

genetic parameters and their applications in self-

fertilizing plants. Statistical Genetics and Plant
Breeding. 253-279. NA-NCR 982.

McNeal, F. H., and D. J. Davis. 1967. Protein
content of wheat kernels from different parts of
the spike. Agronomy Jour. 58:635-636.

Miller, B. S. and J. A. Johnson. 1954. A Review
of Methods for Determining the Quality of Wheat and
Flour for Bread Making. Kans. Ag. Exp. Sta. Tech.
Bul. 76.

Miller, B. S. and J. A. Johnson. 1957. Testing
Wheat for Quality. U.S.D.A. Research Report. No. 9.

Miller, P. A., H. F. Robinson, and O. A. Pope. 1962.
Cotton variety testing: Additional information on

variety x environment interaction. Cr0p Science.
2:349-352.

Murphy, C. F. and K. J. Frey. 1962. Inheritance
and heritability of seed weight and its components
in oats. Crop Science. 2:509-512.

Murphy, C. F. and J. G. Clapp. 1968. Wheat Market-
ing in the Southeast. A survey of wheat millers
Opinions on important problems. Personal communications.

Pendleton, J. W. 1954. Effect of lodging on spring
oat yields and test weight. Agronomy Jour. 46:265-267.

Percival, J. 1921. The wheat plant; a monograph.
E. P. Dutton and Company. New York. pp. 463.

Perkins, J. M. and J. L. Jinks. 1968. Environmental

and genotype environmental components of variability.
Heredity 23:339-356.

Pfeifer, R. P. 1968. Malting barley and wheat flour
extractions. Grist from the Mill. pp. 8 and 11.

Pitman, U. J. and J. E. Andrews. 1961. Effect of rate
of seeding on winter survival, yield, and bushel weight
of winter wheat grown in Southern Alberta. Canadian
Jour. of Plant Science. 41:71-80.

_- J. -__..1'

~.- .I 4 .4)... «P‘I‘KJ‘D—OH

“-9 .'.'—7‘. ".

ar-“.MJO . 0.; I‘m—“O g -
I

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

75

Pool, M., F. L. Patterson, and C. E. Bode. 1958.
Effects of delayed harvest on quality of soft red
winter wheat. Agronomy Jour. 50:271-275.

Roberts, H. F. 1910. Breeding for type of kernel

in wheat, and its relation to the grading and milling
of the grain. Kansas Ag. Exp. Sta. Bul. No. 170:
99-138.

Sharp, P. F. 1927. Wheat and Flour Studies IX.
Density of wheat as influenced by freezing stage of

development and moisture content. Cereal Chemist.

Sheuy, C. W. 1960. A wheat sizing technique for
predicting flour milling yield. Cereal Science
Today. 5:71-72.

Shollenberger, J. H. and D. A. Coleman. 1926. Relation
of kernel texture to physical characteristics, milling
and baking qualities and chemical composition of wheat.
U.S.D.A. Dept. Bul. 1420.

Shollenberger, J. H. and C. F. Kyle. 1927. Correl-
ation of kernel texture, test weight per bushel and
protein content of hard red spring wheat. Jour. Agr.
Science. 35:1137-1151.

Snyder, Harry. 1904. Glutenous and starchy wheat.
Minn. Ag. Exp. Sta. Bul. 85:179-188.

Swanson, C. O. 1937. Wheat and Flour Quality.
Burgess Publishing Company, Minneapolis, Minn. pp. 227.

Swanson, C. O. 1942. A micro method for determining
test weight. Cereal Chem. 19:468-470.

Swanson, C. O. and R. O. Pence. 1932. Moisture in
relation to yield, protein percent and test weight.
Association of Operative Millers Bulletin.

Thomas, L. M. 1917. A comparison of several classes
of American wheats and a consideration of some factors
influencing quality. U.S.D.A. Bul. No. 557.

Weibel, R. O. and J. W. Pendleton. 1964. Effect of
artificial lodging on winter wheat grain yield and
quality. Agronomy Jour. 56:487-488.

=2-. ‘K‘S' ‘1.an
4.1.29

. dun-“.38 ::

'§-K‘Afl'.
II...

60.

61.

76

Yamazaki, W. T. 1968. A Report of Proceeding of
Meeting on Test Weight in Soft Wheats. Ohio Agri-

cultural Research and Development Center, Wooster,
Ohio.

Yates, F. and W. G. Cochran. 1938. The analysis
of groups of experiments. Jour. of Ag. Science.
556-580.

a

28

ET;

 

"A a .; -flua-vi ‘ '
‘4“!

APPENDIX

 

Ou-

77

5|.Jfi3 I I
p:

I“! m”....

 

. .I..PId...I, IRA...

.T.. 3".) H07

.
.6.

.. . 22.143.00.44 u...

 

_.05 0.50 0.5. 000.. 5.00 0.0.. 5.05 .50.. 0.00 0.00 50.00 0.050 05
5.00 0.00 0... 00_.5 0.00 0.05. 5.0. 550.. 0.00 0.05 50.00 0.050 0.
0.00 _.50 0.5. 000.5 5.50 5.5.. 5.05 050.. 0.00 5.05 50.00 0.500 0.
0.50 0.00 5.5. 05.5 5.00 0.0: 0.0. 000.. 0.00 _.55 00.00 0.000 5.
0.00 0.00 ..._ 000.5 0.00 _.05. 0.05 550.. 5..0 0.05 0..00 0.000 0.
..50 0.00 0.5. 500.5 0.00 0.55. 0..5 500.. 0.50 ..00 00.00 0.0.0 0.
0._0 0.00 0.5. 000.. 0.00 0.5.. 0.05 050.. 0.00 0.00 00.00 0.000 0.
0.00 _.00 0.5. 000.. 5.00 0.0.. 0.05 550.. 0.00 0.05 50.00 0.000 0.
5.00 0.00 0.5. .00.. _.50 0.0.. 0.05 000.. ..00 5.00 00.00 0.000 5.
0.00 0.00 0.5. 500.. 0.50 0._5. 0.05 550.. 0.00 5.00 00.00 0.000 __
5.50 0.00 0.5. 500.. 0.50 0.05. 5.05 500.. ..00 0.00 00.00 0.000 0.
0.00 0.00 0.0. 000.. 0.00 5.05. 0.05 050.. _.00 ..00 05.00 0.500 00
0.50 0.00 0.5. 000.5 ..00 0.05. 0.05 0.0.. 0.00 5.05 .5.00 0.5.0 00
5..0 0.00 0.0. 500.. 0.00 0.00. 0.0. 000.. 0.00 0.55 .0.00 0.050 50
0.00 0.00 0.0. .50.. 0.00 _.5__ 5.55 000.. 0.00 0._0 05.00 0..00 00
0.00 _.00 5.5. 000.. 5.00 5.50. 5.0. 000.. 0.00 0.55 05.00 0.500 00
5.00 _.50 5.5. 000.. 0.50 0.0.. 5.0. 0.0.. 5.00 0.05 00.00 0.000 00
0.00 ..50 0... 00..5 0.00 0.0.. 0.0. 500.. 0.00 0.55 00.00 0.0.0 00
0.50 0.00 0.5. 000.. 0.00 5.0: 0.0. 000.. 5.00 0.05 50.00 0.050 50
0.00 0.00 5.5. .00.. 0.00 0.0.. 5.55 5.0.. 0.50 0..0 05.00 0.500 .0
0.6.5 x600. .06.1 6.0.0 000.: 0.006. ._e> .060 .000 500.»: .06» 000_6: .6:
Lao—u ._smoa .0000: 3\4 .0000: .ocuox .ocuux xuma .mctox 0so_: amok

 

00m. :_ czouo 000:3 00 «0.0500 mm 000 ARV

v_o_> Lao—u one «ARV xooc. 0:__smua «ARV c_000sa .ocuux 5033.
000.3 0.0060 .5su0 0.0.: .oeeux 00. .5s60 000eo0 .oeeox 00. .5_s0
os:_o> .ocuux coo. 5>u_mc00 55x0 >oc0_u_wmm mc_xoma «Amsmv uzm_03

_ocuox 000— 55000 mmm_m\0Emv unm_oz amok cuu_z sauc_axmsmv u;m_03 amok 00 0.005

78

 

0.00 ..00 0.0. .00.. 0.50 0.00. 5.00 000.. 0.00 0.00 00.00 0.0.0 00
5.00 0.00 0.5. 000.5 0..0 5.05. 0.05 0.0.. 0.50 ..00 00.00 0.0.0 00
0.00 0.50 0.5. 000.. 0.00 0.05. 0.00 000.. 0.00 5.00 00.00 0.050 00
0.00 ..00 0.5. 000.5 0..0 0.05. ‘0.05 050.. 0.00 0.00 50.00 0.050 50
0.00 0.00 0.0. 000.5 0.00 0.05. 0.05 000.. 0.00 0.50 05.00 0.0.0 00
0.00 0..0 0.5. 000.5 0.00 , 0.05. 0.05 050.. 0.00 0.05 00.50 0.500 00
5.00 0.00 0.5. 550.. 0.00 0.0.. 0.00 5.0.. 0.00 0.00 05.00 0.000 00
0.00 0.00 5.5. 000.5 0.50 0.00. 0.05 0.0.. 0.50 5.00 00.00 0.000 00
0.00 0.00 0.5. 00..0 0.00 0.55. 0.05 000.. 5..0 0.50 05.50 0..00 50
0.00 ..00 0.0. 00..5 0.00 0.50. ..05 5.0.. 0..0. 0.50 00.50 0.000 .0
5.50 0.00 0... .5..5 0.00 0.0.. 5.5. 000.. 5.00 ..05 00.00 0.0.0 00
0.50 5.50 0.5. 0.0.5 0.00 0.5.. 0.05 500.. 5.00 0.05 00.00 0.5.0 00
0.00 0.00 0.5. 000.. ,.0.00 0.0.. 0.05 000.. 0..0 0.00 5..00 0.000 05
0.00 0.00 0.5. 050.. 5.00 0.0.. 0..5 050.. 0..0 ...0 .0.50 0.000 55
5.50 0.50 0.5. 00..0 0.00 0.55. 0.05 000.. 5.00 0.00 .0.50 04000 05
0.00 ..50 ..0. 500.. 0.50 0..5. 0.00 000.. 0.00 0.00 0..00 0.0.0 05
0.00 0.00 0... .0..5 0.00 ...0. 0.0. 050.. 0.00 0.55 05.50 0.000 05
..00 0.00 ..5. 05..5 0.00 0.05. ..0. 550.. 0.00 ..50 55.00 0.000 05
0.00 0.00 0.5. 500.5 0.00 0.0.. 0.0. 0.0.. ..00 0.05 00..0 0.000 55
0.50 0.50 0... 500.5 ..00 0.00. 0..5 0.0.. 0.50 0.00 0..00 0.000 .5
0.o.> x000. .00.0 0.000 000.: 000000 ..o> .coa .000 000.o: amok 000.»: .oz
Lao—...— . :mon— .093! 3\.. —oc.._0v_ pecuox —oc.._0v_ xumm —0c._0¥ 0503‘ umvh

 

0000.0000 .0 0.000

79

 

 

:.mo m.o: o... m~m.. m.~o m._~_ m.m~ m~:._ ~.:m _.:m ma.mm o.om: mm
...E TR m.: Rm; m.$ 59. :._~ 3.3.. :.mm mam mm.mm 903 mm
~.~m o.mm o.~_ Nmm._ m._m m.~__ ..NN m~:._ o.mm m._m fin.mm ‘m.mm: mm
:.~m o._: m... Nam._ ..No o.o~_ m.m~ ~m¢.. o.mm ~.mm um.mm o.om: mm
«.mo :.o: m... omm._ m.:m m._~_ o.m~ m~¢._ m.mm “.mm .m.mm m._:: mm
:.mo o._m ~.N_ m~m.. o.:o m.m~_ ~.m~ mo:._ o.:m m.mm mm.:m c._~: :m
m.mm m.om u... ~.m.. m._m o.~__ m._~ :o:._ m.mm ~.om .m.mm m.o_a mm
5.8 m.~m ~.~_ 5m.— m.$ 9:2 {cu mam; 9mm «in Rém m.m3 3
“.mo ..Nm o.~_ ::m._ m.~o ...N. c.m~ m.:._ ~.mm m.~m mm.mm m.~_: .m
:.:o m.~m _.~_ “mm._ m.~m ~.m__ o.- oo:.. o.mm m._m mm.mm m.m_: om
m.¢o m.m: ..N. umo.~ “.mm :.m~_ ~.- m_:._ ~.~m m._m om.mm m.mo: m:
“.mo m.~m o.~_ msm._ m.mm m.-_ m.- so:._ m.:m :.mm mm.:m o.:~: m:
..mo :.m: _.N_ m:m._ m.~o m.o__ :.m~ m_:._ o.mm ..mm mo.mm o.m~: 5:
..Jo :.om m... mam». m.~o ~.m__ m.:~ mo:._ 5,:m :.:m :o.:m m.:~: o:
«.mo m.m: o.~. ~mm.. m._e :.~_. m.- o_:._ “.mm :._m ~_.:m o.o~: m:
m._o o.~m :.~_ ~mm.. m._o ~.-_ :.m~ s_:._ m.:m _.mm mm.:m o.m~: ::
«.mm m.:: m... .Nm._ 5.:0 m.~__ m.:~ ~_:._ m.:m m.mm oo.:m o.m~: m:
:.~m m.m¢ _.~_ ~_m._ ~._m m.m__ ~.- m_:._ ~.mm m._m om.mm o.~_: N:
m.:o :.o: _.~_ mmm._ m.~m m.~__ m.m~ :_:._ m.:m m.mm am.:m m.n~: .3
u_u_> xouc_ .uo.m o_um¢ guu_: guano; ._o> .coa .mwu »;m_o: “mop “sane: .oz

L30 :— . 750m _0:LOM {4 —0c._ux —0c._ov_ —0c._0v_ xumm _ocL0v_ 0..0 m: amok
VODPZHCOU d. o—nmh

80

 

 

 

Table B Test Height (gms/pint) for the Mean of 3 Replications of 30 Lines
of "heat at 7 Locations in 1968
Site
No. Line
T6] H71 K81 ' 891 101 551 L01

0] Genesee 023.33 036.67 033.00 031.00 011.67 372.00 378.00
02 Avon 021.67 037.00 032.00 031.00 000.33 360.00 370.00
03 Talbot 011.67 021 67 031.67 030.67 017.33 372.67 379.00
00 Honon 028.33 007.67 013.67 031.67 380.00 391.00 363.00
05 Arthur 030.00 052.33 027 00 036.00 020.67 008.67 396.00
06 Reed 035.33 050.33 028.33 035.33 023.00 013.33 019.00
07 Yorkstar 013.00 023.00 016.33 017.33 377.33 366.67 392.33
08 A1220 023.67 038.00 025.00 031.33 010. 33 379.00 388.67
09 A2550 027.00 039.67 029.67 027.00 008. 33 396.67 390.67
10 A2739 028.00 003.00 027.00 030.67 398. 67 368.33 358.00
11' A2707 027.33 003.00 020.67 031.00 013.33 378.33 388. 33
12 A3116 011.67 023.670 1017.00 021.67 380.33 381.33 329. 33
13 A3136 026.00 038.33 020.67 020.33 002.33 003.33 378.33
10 A3101 020.67 039.33 022.33 020.00 006.67 399.00 387.67
15 A0129 009.33 017.33 003.33 011.33 380.67 363.33 306.00
16 A0528 012.00 026. 33 011.00 011.00 386.00 361.00‘ 309.33
17 A5000 030.67 006. 67 026.00 027.67 007.67 393.00 386.00
18 A5127 022.67 030. 67 029.33 027.67 005.33 380.00 353.00
19 A5131 023.67 036. 00 032.67 030.33 390.00 380.33 363.00
20 A5132 020.00 033.67 033.33 029.67 007.00 356.00 351.00
21 A5130 023.67 035.33 032.00 029.67 015.00 353. 67 350.00
22 A5266 027.67 001.00 037.33 036.33 013.67 398.67 369.67
23 A6620 001.00 050.67 026.67‘ 002.33 026.33 010. 67 399. 33
20 A6625 033.67 003.33 031.33 036.33 018.67 002. 33 385.67
25 A6626 031.00 005.00 030.33 038.33 010.00 001.67 378.00
26 A6628 000.67” 057.33 030.67 003.67 021.00 010.67 396.00
27 A6629’ 035.33 007.00 033.67 035.33 015.67 010.00 389. 67
28 A6630 038.00 052.67 036.33 036.00 019.00 015.00 006.33
29 A6631 036.33 051.67 032.67 038.67 013.67 000.67 389.67
30 A6632 036.00 052.67 026.33 035.33 005.33 002.00 380. 00

81

 

 

 

Table C 1000 Kernel Height (gms) for the Mean of 3 Replications of 30 Lines
of Wheat at 7 Locations in 1968
Site

"°' L‘"° T61 H71 x81 891 101 $51 L01

0| Genesee 00. 96 03.96 00.82 35.70 33.80 28.05 26.90
02 Avon 00. 65 00.16 39.90 37.10 31.53 27.30 25.71
03 Talbot 35 77 37.59 39.66 37.08 33.58 26.25 25 99
00. Honon 32.38 36.00 29.63 30.27 20.06 26.78 21.06
05 Arthur 37.50 01.10 35.13 30.81 30.18 32.97 26.62
06 Reed 39.89 03. 53 36.03 30. 90 33.07 35.65 32.07
07 Yorkstar 36.78 37.60 36.38 30. 00 27.72 28.72 20.70
08 A1220 30.73 38. 30 32.25 31.05 31.93 _26. 35 20.85
09 A2550 00.61 00. 00 38.96 30 95 33.03 30. 90 28.05
10 A2739" 02.31 06.00 38.10 37.33 33.90 29.00 25.63
11 A2707 35.30 39.56 32.61 32.56 . 32.32 26.12 25.30
12 A3116 36.01 38.20 35.61 30.73' 28.85 29.67 20.13
13 A3136 37.11 01.28 37.15 35.52 30. 78 30.11 25.77
10 A3101 36.30 00.10 30. 56 33.53 28.83 30:83 25.56
15 A0129' 30.35 36.51 30.57’ 30.00 28.88 27.38 21.83
16 A0528 36.65 39.87 36.09‘ 30.70 30.81 27.60 23.07
17 A5000 36.06 39.03 35.19‘ 32.23 29.82 29.29 20.80
18 A5127 00. 08 06.80 01.21 39.52 36.66 33.22 25. 77
19 A5131 02. 02 00.52 01.00 37.01 29.71 30.60 25.00
20 A5132 00.36 03.19 39.10 35.11 31.51 26.00 22.77
21 A5130 00.99 03.86 00.01 35.80 33.56 26.50 23.03
22 A5266 39.20 01.70 37.51 35.93 31.00 30. 70 20.00
23 A6620 35.55 00.22 32.28 30.51 31.15 31. 32 25.19
20 A6625 36.36 39.29 32.71 31.66 30. 85 31. 12 26.75
25 A6626 33.30 37.35 32.03 31.32 29.16 28.96 21.95
26 A6628 30.80 39.37 33.15 32.59 30.90 30.00 23.13
27 A6629 33.19 37.50 31.31 29.83 26.38 29.82 21.92
28 A6630 39.10 01.00 30.38 32.60 31.81 32.55 26. 78
29 A6631 32.83 37.20 32.10 31.12 28.51 28.35 21.70
30 A6632 30.39 39.23 32. 03 30.01 27.69 27.89 22.08

.‘c—w‘c-n“unh.——flfl

T I. I

.

1 .'

‘I;_‘,_‘_¢ _- _ '."I’§.If'
I

82

 

 

 

Table I) 100 Kernel Length (cm) and 100 Kernel Hidth (cm)
for 30 Lines Planted in 3 Locations
in 1968

No. Line Length Width

H71 101 L01 H71 101 L01
01 Genesee 61.0 60.5 59.1 33.9 28.6 27.2
02 Avon 60.6 59.1 55.7 30.6 27.1 20.7
03 Talbot 61.0 59.8 56.7 32.6 29.8 26.1
00 Honon 59.0 58.0 57.6 31.8 26.6 27.0
05 Arthur 60.2 59.6 57.0 33.5 29.5 28.1
06 Reed 65.8 65.0 61.0 32.3 27.6 27.0
07 Yorkstar 61.0 59.9‘ 56.7 32.7 26.1 22.3
08 A1220 60.9 59.2 55.3 30.0 29.5 25.9
09 A2550 61.1 59.0 58.9 30.2 28.3 26.6
10 A2739' 63.7 63.0 61.5 35. 29.6 27.1
11 A2707 61.9 62.0 58.8 33.8 30.0 26.8
12 A3116 61.1 60.7 56.2 32.5 28.0 23.9
13 A3136 61.8 60.0 58.8 32.7 28.0 26.3
10 A3101 60.6 58.3 58.6 33.8 29.2 28.0
15 A0129 61.5 62.3 60.2 31.0 27.6 25.1
16 A0528 63.3 63.3 60.3 33.3 28.7 25.7
17 A5000 59.5 60.3 59.2 33.3 29.8 28.5
18 A5127 60.5 62.5 59.2 35.8 32.3 26.0
19 A5131 65.2 62.9 59.7 30. 29.1 20.9
20 A5132 60.8 60.0 60.5 33.7 26.0 25.3
21 A5130 63.9 61.0 56.9 30.5 28.5 23.2
22 A5266 6l.5 59.7 56.6 30.2 28.8 25.9
23 A6620 60.5 '56.5 56.1 32.5 28.7 27.8
20 A6625 59.6 58.0 56.8 32.3 28.6 27.0
25 A6626 59.0 58.2 55.8 31.7 2923 20.0
26 A6628 59.8 58.8 55.5 32.0 29.6 26.1
27 A6629' 58.9 57.6 57.3 31. 26.3 26.6
28 A6630 58.6 56.0 53.3 33.7 29.3 28.0
29 A6631 59.5 57.0 50.6 31.6 28.7 26.2
30 A6632 59.3 57.9 53.6 32.0 27.0 26.1

83

 

 

0:0 LoaoEOcoxa com.cmqeou L.m ecu mc.>a=ooo mc.mcm ozu mo 05:.o>«
.N..o m~.mm 00.50 m~.~0 m.:.. 5.0.. .00.. 00.m~ m5.m~ ow..~ .m.0m 04.0m mm.0m 0N.:< m.
0..00 00.00 mm.50 m~.mo m0:.. 003.. 003.. 5~.0~ m5.0~ om.m~ mm.0m M5.5m 0:.mm .:.m< :.
3.3 00.50 3.00 00.3 3... 53.. 00.... 5m.m~ 05.8 .m.m~ 8.05 mm.5m ...mm of? m.
m..mo m..m0 00.50 mm.00 o.:.. mo:.. ..:.. :~.0~ m~.0~ mo.0N 0..5m 5m.0m :..m~ 0..m< N.
m0.mo 05.mo m0.mo m:.m0 .~:.. 0.3.. ..:.. .:.o~ mm.m~ wo.m~ Nm.5m ~..mm 0:.mm 535~< ..
0:.N0 mm.mo mm.50 0:.N0 5.0.. 000.. 504.. 5m.0~ N..5~ 05.~N 00.5m w..mm mm..m .mn5N< 0.
m:.m0 05.00 m0.mo mm.50 m.:.. :.:.. .00.. .m.0~ 0m.0~ «5.:N m..5m :0.mm :0.:m amm~< mo
00.30 00.00 mm.5m mm.mm .~:.. .~:.. 503.. No.0~ 00.0N ::.:~ .m.5m um.5m mm.:m :NN.< :0
00.~0 00.00 m~.mo 00.00 0N:.. ..4.. 003.. 00.m~ :m.mN 00..~ no.0m 00.0m 00.0w Lmumxgo> 50
0..00 00.00 mm.50 0~.mm m.:.. 0.0.. mua.. 55.0w mm.5N mm.mN1-mm.5m 05.0m m..mm 000: 00
mm.m0 m0..5 mm.mo m..mo 0.0.. 5.0.. 500.. 05.m~ ~..5~ .m.:~ M5.5m m:.mm .m:.:m LaauL< m0
mm.:0 m0.05 mm.mo mm.mm 0.4.. m~:.. 055.. .:.0N ...5~ m0.m~ m~.5m mm.mm m5..m coco: :0
mm.5o 00.05 0m.mo m0.oo .~:.. m.:.. 0.0.. m:.o~ mm.m~ um.mu mm.5m 00.0m .5M.mm uoa.mh m0
00.00 mm.05 03.00 0m.00 m~:.. 0.0.. 0.3.. m~.0~ 0m.m~ :m.m~ ~m.5m 0:.5m ~5.mm :o>< «0
0~.m0 mm.mo 0..50 00.:0 -:.. ~.:.. ~0:.. .m.m~ 0:.0N ~:.m~ 0:.5m 0:.5m aw.~m oumocou .0

.0. .mm .5: .4. .mm .5: .J. .00 .5: .4. .mm .5: .0.
. oc.. .oz

v.o.> Lao.u >u.mcoa awe:.o> usm.o: amok oLo.:

 

L.< ogu 00.500000 mc_mL0 ozu mo oe:.o> 050:0 Louoao:u>q 00m.Lmano L.m\m30v uam.oz amok oLo.: 3m manikw

mco.uooo. acoLomm.0 um anus: mo moc.. on no»

meo.uao..aom N .o momoLo>< «Ago 0.o.> Lao.u 00m 05:300 .A.EV 0:0 Louoeoco>m com.Lm6560

84

0:0 LouveOcu>a com.Lmano L.o osu m:.>a=uoo mc.mLm ogu mo oE:.o>«

m2.

 

 

 

05.00 00.m0 0m.05 05.00 5mm.. 0.0.. 05m.. 50.5N 00.5N mm.m~ 00.0m .0.mm 0m00< on
00.00 5.00 3.00 3.00 000.. 0.0.. 50m. 8.5a 50.5N .~.0~ 095m 0..0m mmém .m00< mu
00.50 00.00 0..05 00.50 mmm.. 0.0.. .00.. mm.5~ 00.5w 0..m~ 00.5m 50.0m 0.40m 0m00< mu
m5.m0 0...5 mm.m0 00.00 mmM.. 5.0.. .00.. 00.0N .m.5~ .0.0~ Nm.5m .5.0m 00.nm .m~00< 5N
05.00 3.00 00.00 mm.m0 0mm. 30.. 0mm. m0.5~ 0m.5~ K.0~ ~m.0m 5min 3.0... 0~00< 0N
05.00 0mn00 m0.00 00.00 000.. 0N0.. ~00.. mm.0~ .~.5~ m5.m~ 00.5m ~0.mm 0~.mm 0~00< mu
m. .00 3.00 3.00 3.00 0%.. 000.. «mm. 00.5N 0~.5~ mm.m~ 55.5m 0m.0m 0m.mm m~00< 0N
00.50 m0.m0 $.00 $.00 0%.. 0.0.. mam. 0m.5~ m5.5~ 55.0N 5. .mm 0. .mm .m.0m 0~00< MN
mm.00 0m.m0 mu.50 00.00 0.0.. ..0.. ~.0.. mw.0~ 00.5w 5m.m~ 00.0m m..mm mm.Nm 00~m< -
0m.m0 00.00 00.00. mm.00 0N0.. 0.0.. 0.0.. m..0~ ~0.0~ 0m.- 0m.5m «0.5m mm..m 0M.m< .N
m..m0 0m.50 00.00 0...0 .N0.. 0.0.. 0.0.. 0m.0~ 00.0N om..~ 5m.5m 00.5m 5m.0m um.m< 05
00.00 R50 $.00 3&0 0.0.. 50.. 0.0.. 5~.0~ 0m.0~ 05.3 8.2 m0.5m mm.~m .m.m< m.
mm..0 05.50 00.00 00.~0 0.0.. 5.0.. 0mm.. m..0~ m0.0~ mm.- m..5m 00.5m 00..m 5N.m< 0.
m0.m0 05.05 mm.00 00.00 500.. 0.0.. 0mm.. ~0.0~ 0N.5~ 00.0N 5..5m 00.0n mm.mm 0000< 5.
0m.m0 mm.m0 0m.00 00.00 0.0.. ~.0.. «00.. om.m~ m0.0~ 05».~ mm.mm m0.0m 00.0m 0~m0< 0.

.0. .00 .5: .0. .00 .5: .0. .mm .5: .0. .mm .5: .0.
02.. .oz

0.o.> Lao.m >u.mcoo h...0..=...o.. u:0.o) amok oLo.:
000:.uco0 m 0.00h

85

 

00.000 00..00 00.000 00.0.0 00.000 00.000 00.000 00.000 00.000 00.000 00.000 00..00 00:
00..00 00.000 00.000 00..00 00.0.0 00.000 00.000 00..00 00.000 00..00 00..00 00.000 000
00.000 00.000 00.000 00.000 00.0.0 00.000 00.000 00.0.0 00..00 00.000 00.000 0..000 000
00.000 00.000 00.000 00.000 00.000 00.000 00.000 00.000 00..00 00.000 00.000 00.000 00:
00.000 00.000 00.000 00.000 00.0.0 00.000 00.000 00.000 00...0 00..00 00.000 0..000 000
00.000 00.000 00.000 00.000 00.000 00.0.0 00.000 00..00 00.000 00.000 00.000 0...00 000
mm.000 00.0mm 50.000 50.000 mm.m~m 00.0.4 00.0mm 00.n0m 00.050 00.05n 00.050 40.000 000
00.0.4 50.000 00.004 00.4.4 00.550 00.m~4 50.004 00.400 mm.5.4 mm.004 50...4 00.004 00.
00.000 00.000 00.000 00.000 00.000 00.000 00.000 00.000 00.0.0 00.000 00.000 00.0.0 000
00.000 00..00 00.0.0 00.000 00..00 00.000 00.000 00.0.0 00.0.0 00...0 00.0.0 00.0.0 000
00.000 00.000 00.0.0 00.000 00..00 00..00 00.000 00.000 00.0.0 00.000 00.0.0 00.0.0 00:
00.000 00.000 00.0.0 00.000 00.000 00.000 00.000 00.000 00..00 00.000 00.000 00.0.0 000
00.0.0 00.0.0 00.0.0 00.0.0 00.000 00.000 00.0.0 00.000 00.0.0 00.000 00.000 .0.000 000
00.000 00..00 00.0.0 00.0.0 00.000 00.000 00.0.0 00.0.0 00.0.0 00.0.0 00.0.0 00...0 000
00..00 00..00 00...0 00...0 00.000 00.000 00.000 00.0.0 00.0.0 00.0.0 00...0 00.000 00.
00000 0000< 0000< 000.< “00” 0000 .000.< 0000: 000.00 co>< oomoco0 00oz .000
u go . .>cm

 

000.4 «N go>0 memo!

“00.0 000 030.

.mucoacoL.>cm.vc<

u:0&:00.>cm Lou n:0.u~0..000 m Lo>0 vomago>< 000.4 «N 000 050.0: 0000

m 0.000

6
8

 

00.004 00.404 00.004 50.004 50.5.4 00.004 00..04 00.000 50.004 00.m.4 00.0.4 50:
00.004 50.004 50.004 00.004 50.504 50.504 00...4 00...4 00.404 00.404 50..04 000
00.504 00.004 00.004 50.004 00.004 00.004 00...4 00.004 00.004 50.004 00.5.4 002
00..44 00.004 50.004 00.004 50.004 50.044 00.004 00.5.4 00.004 00.004 50.004 00:
50.504 50.004 00.404 50.004 50.004 50.004 00.0.4 00.004 50.404 00.004 50...4 00h
00.004 50.000 00.000 00.000 00.000 00.000 00..00 00.000 00.000 00.004 00..00 000
50.000 00.000 00..00 00.000 00.000 00.000 00.040 00.040 50.500 00.050 00.000 004
50.0.4 00.0.4 00.504 00.000 00.004 50.504 00.000 50.000 50.004 00.004 00.400 00.
00.404 00.0.4 00.004 00.0.4 50...4 50.004 50.004 50.0.4 50.0.4 50.0.4 00...4 500
50.5.4 00.5.4 50.0.4 00.0.4 50.004 50.5.4 00.404 00.004 00.0.4 50.004 50.004 502
00.004 00.004 50.004 00.404 00.0.4 00..04 000000 :004000 50.0.4 .0005.4 00.000 50:
50.004 00.0.4 00.004 00.004 00.5.4 00.004 00.000 00.404 00.0.4 00.m.4 50.004 50h
00.0.4 00.5.4 00.004 50.5.4 00.504 00.504 00..00 00.400 00...4 50.504 00.000 500
50.5.4 00.004 50.004 00.504 00...4 00.4.4. 00.000 00.000 00.0.4 00.004 00.004 504
50.004 00...4 00.0.4 00.004 50.000 50.504 50.000 50.000 00.004 00.004 50.000 50.
00Nm< 40.m< 00.m< .m.m< 50.m< 440m< mNm4< mN.4< .4.0< 00.0< 0..0< .>cu

 

02.0.0000 .m 0.00...

87

 

.50 .m00h
Nb 0045.40 004 4 a > =_gu_x
a x >
H_>b0 +.Mb «00005.0m_ cm A x > x >
_>b0 + _>bm + mu m 20000.550 om 4 x >
>> 5.) o .
>020+50b0+0b mzmmm0050 m. _>x>
> —.5. —> 0.0 - .
0b04 + 050 + 0b_0 + . 0b0 + 0b «£0050.0054 0. .>. o:_4
00 .a. 4 an» >
=_gu_: co_uau__aoa
h..._...0000.00...\.0. 0 4 x >
«#0000.0000. 0 .40 co.u~004
«#0500..000 . A». Lao»
mxu m: ma ou.:om

 

aco.ua004 5 new
who.» 0 c. moa.h 0 vouao..ao¢ no:.4 0. Law

Au:_0 Lon.»300 unm.ox «map 500 0.0mh uuqm..m> mo m.m».~¢< o 0.0mh

88

 

 

054 .0005
Ma 5005.00 com 4 a 5 :_;u_3
x x 5
5.> 0 -- . ,
5 0b0 + 0b «#0000 00 00 4 x 5 x >
.5 .> 0 .
5>0b0 + 5 0b0 + 0b .m.z440. 00 00 4 x 5
.5 0 .
5> 0am. + 5 0b0 + 0b .m.z05.0 .0. 0. -5 x 5
> .> .> 0 -- .
0b00 + 0b0. + 0b0 + .000 +.0b «#00400550 0. - -5500:_u
00 .0. 4 0:0 5
c.:u.3 co.wwu..00¢
005000.5050 4 - 4 x 5
«#0004.0040. 4 .40 00.00004
«#0000.4004 . .50 0005
0:0 0: 00 005500
000.00004 0 new
«0005 0 a. 003.5 0 noumu..00¢ 000.4 0. Low

500.0 500 0300 u:0.0) 0005 Low 0.005,0uco..05

.0 u.m5.~=< m o_n~5

89

Table I Analysis of Variance Table for Packing Efficiency
for 30 Lines Replicated 2 Times

In 3 Locations

 

 

 

 

Source DF HS ENS
. **
Location (L) 2 213.59896l57
Rep. (R) V 3 0.2010 ‘
Line (v) 29 2.865755%“ 0: + 273' + 60':
V x L 58 0.53h85603ff a: + 203]
V x R 8 Error 87 0.07655] 0:
Total 179
Table J” Analysis of Variance Table for Density
for 30 Lines Replicated 2 Times
In 3 Locations
Source DF HS EH5
Location (L) 2 0.00l7155fiaf
Rep. (R) i 3 0.00mi» . .
** ,
Line (V) 29 0.0031053“ a: + 2‘31 + 60':
fl
V x I. w 58 0.000975 _ a: + 2‘31
v x a .-. Error ‘87 0.000305 0':

Total

179

90

Table K Analysis of Variance Table for Test Height
for 30 Lines Replicated 2 Times
In 3 Locations

 

 

Source DF HS EHS
Location (L) 2 466.6h5j**

Rep. (R) l 3 .
Line (V) 29 #.86hhf* a: + 203] + 603
” .** 2 2
V x L 58 1'2007__ we + 20v]

V x R = Error 87 0.i‘+85 0'2

Total 179

(D

 

 

M'llllfilmflllillillill fillilfillfllllll Willi)“
3 1293 0106] 2992