“-33.12; . .

v ,r,.,,.,.,,
. -. . . ’1“ V V‘
~. ""5.
1;, '1"'.‘r-: V i. V
“""""1..., ‘-
’» :A'l‘.,.,"- ‘
"~-"..7.V W" V
' 1-“ a;

my .

 

 

 

THE-3‘5

L‘.

mum =

This is to certify that the

dissertation entitled

MULTIVARIATE ANALYSIS AND HETEROTIC EFFECT STUDIES
IN CANOLA (Brassica napus L.)

 

presented by

Muhammad Ali

has been accepted towards fulfillment
of the requirements for ’ . _ .

 

 

Ph.D. degreein Crop & Soil Sciences -
Plant Breeding & Genetics

Program

, Major professor /

Date 4-7-3?

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

DATE DUE DATE DUE DATE DUE

-

movem“

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity Institution
dietitian-woman

MULTIVARIATE ANALYSIS AND HETEROTIC EFFECT
STUDIES IN CANOLA (Brassica nagus L.)

By

Muhammad Ali

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Crop and Soil Sciences

1994

JABSTTUMCT

DJUIJHVIUUUYUEAMNAIJEHSIUWDlflEITJuTTHCIflflflflTTS
IN CANOLA (firassica mpg L.)

By

Muhammad Ali

Thirty winter canola cultivars (Brassica gapug L.) from
various sources were analyzed and clustered/grouped based upon
their morphological traits, yield and yield component data.
Cultivars were placed in three distinct clusters. Genetic
distance between all of the thirty cultivars was also
estimated. Two cultivars from each group were selected as
parents and fifteen partial diallel inter- and intra-cluster
crosses were made between the six selected parents. These
crosses, along with their parents, were evaluated at two
locations: East lansing and Clarksville in Michigan in 1991.
The heterotic response for all of the fifteen crosses was
calculated as a percent of the mid-parent performance.

All of the fifteen crosses exceeded the mid-parent value
for yield, pods per plant and seeds per pod at both locations.
The correlation between parental distance and heterotic
effects for these traits was positive and highly significant.
No significant correlation was observed between genetic
distance and heterotic effect for morphological traits like,

crown diameter, leaf length, leaf width and branches per

 

plant. A lack of genetic variation and genotype x
environmental interaction was evident in these traits.
Dominance genetic variance was observed in favor of early
flowering in all crosses. Clustering based on yield and yield
component traits demonstrated that inter-cluster heterosis was
shown to be greater than intra-cluster heterosis in the

majority of cases.

Dedicated to my loving parents

to whom I stand in debt for my education and knowledge

iv

Acmnnywuiunaumuuvr

I wish to express my sincere thanks and gratitude to my
major professor Dr. James D. Kelly for his guidance,
encouragement, accurate criticisms and invaluable suggestions
throughout the preparation of this manuscript. Special
gratitude and appreciation is gratefully extended to my co-
professor Dr. L.O. Copeland for his kindness, invaluable
suggestions, continued moral and financial support and
providing me the opportunity to initiate my research project
on Canola. Thanks are also due to my other committee members
Drs. Dale D. Harpstead and Amy Iezzoni for their invaluable
comments and critiques. I would also like to thank Dr. Clay
Sneller and appreciate his help for analytical procedures.

I extend my thanks to the Government of Pakistan/USAID
who provided me the opportunity and financial support to
accomplish this study. I also take this opportunity to thank
my colleagues and friends Drs. Sabry G. Elias and Bryan R.
Brunner for their help and support. Thanks are also due to all
of my friends for sharing the moments of joy and interest with
me. Last but not the least I am greatly thankful to my loving
parents and other family members for their patience,

understanding and always inspiring and supportive attitude.

TABLE OF CONTENTS

LIST OF TABLES PAGE

1. INTRODUCTION . ............................ . .......... 1

2. LITERATURE REVIEW....... ............................. 5
2.1 Rapeseed vs. Canola. ..... ......... ............ 5
2.2 Genetic Distance......... ..................... 7

2.2.1 Estimation of Genetic Distance ...... 8

2.2.2 HeterOSiSooooooooooooooooooooooooooo 21

2.2.3 Gene Action in Heterosis ............ 23

3. MATERIALS AND METHODS ............................... 28
3.1 Development of Experimental Material ......... 28

3.2 Multivariate Analysis ................... ..... 31

3.2.1 Analysis of Variance................ 31

3.2.2 Principal Component Analysis ........ 31
3.2.3 Clustering Analysis ................. 31
3.2.4 Euclidean Distance .................. 32
3.3 Selection of Parents ......................... 32
3.4 Crossing Program ........................... .. 33
3.4.1 Floral Morphology ................... 34
3.4.2 Crossing Technique .................. 34
3.5 Evaluation of Crosses ........................ 36
3.5.1 Data Recording ...................... 36
3.5.2 Data Analysis .......... .... ......... 36

vi

 

 

 

PAGE
3.5.2.1 Analysis of Variance .............. 36

3.5.2.2 Correlation Studies............... 36

3.6 Estimation of Heterosis................ ...... 36
Mid Parent Heterosis. ............... 37

3.7 Chemical Analysis.... ....................... . 37
3.7.1 Oil Content... ....... ... ............ 37

3.7.2 Determination of ....... ........... . 37

Oil Content by NMR

4. RESULTS... .......................................... 43
4.1 Analysis of Variance... ........... ..... ...... 43
4.2 Cultivar Analysis ......... . ................... 61
4.3 Genetic Distance....... ...................... 61
5. DISCUSSION. ......................................... 71
APPENDIX... .......................................... .. 82
LIST OF REFERENCES ..................................... 85

vii

IJSTGDFTDNBLES

PAGE

Cultivars and origin of 30 winter canola
(Brassica napuS»L.) genotypes....................... 39

List of 15 canola (Brassica napus L.)
crosses made by using six selected parents
from three different clusters ............ ........ .. 41

List of morphological traits and yield
components selected for data recording..... ......... 42

Means for seven traits of 21 entries/crosses
of canola combined for both locations,
East Lansing and Clarksville........................ 49

Means for seven traits of 21 entries/crosses
of canola combined for both locations,
East Lansing and Clarksville........................ 50

Means for seven traits of 21 entries/crosses
of canola grown in East Lansing, MI, 1991...... ..... 51

Means for seven traits of 21 entries/crosses
of canola grown in East Lansing, MI, 1991..... ...... 52

Means of seven traits of 21 entries/crosses
of canola grown in Clarksville, MI, 1991.... ........ 53

Means of seven traits of 21 entries/crosses
of canola grown in Clarksville, MI, 1991.... ....... . 54

Percent mid-parent heterosis for 15 canola
crosses combined for both locations,
East Lansing and Clarksville, MI, 1991.............. 63

Percent mid-parent heterosis for 15 canola
crosses combined for both locations,
East Lansing and Clarksville, MI, 1991 ......... ..... 64

Percent mid parent heterosis for seven

traits in 15 canola crosses grown in
East Lansing, MI, 1991 ....... ... .................... 65

viii

10.

11.

12.

Percent mid parent heterosis for seven
traits in 15 canola crosses grown in
EastLanSing'MI,1991.00.00.000000000......OOOOOOOO 66

Percent mid-parent heterosis for seven
traits in 15 canola crosses grown in
ClarkSVille' MI, 1991. O O O O O O O O O ..... O O O OOOOOOOOOOOOO 67

Percent mid-parent heterosis for seven
traits in 15 canola crosses grown in
ClarkSVille, MI, 1991.0... ...... 0...... ....... .00... 68

Genetic distance between six parents
calculated based on their morphological
and.yield.component traits.. ....... ...... ........... 69

Eigenvalues of the five PC axes from PC

analysis of thirty canola cultivars with

the proportion of total variance accounted

for by each.axis............................ ..... ... 7O

Correlation coefficients between five
traits for thirty canola cultivars.................. 70

APPENDIX

A1.

A2.

A3.

Analysis of variance for 14 traits in
21 entries/crosses of canola in East Lansing,
MI 1991 OOOOOOOOOOOOOOOOOO .0... OOOOOOOOOOOOOOOOOOOOOO 82

Analysis of variance for 14 traits in
21 entries/crosses of canola in Clarksville,
MI 1991.... ......................................... 83

Analysis of variance for 14 traits in

21 entries/crosses of canola combined for

both locations, East Lansing and Clarksville,
MI1991.................................... ......... 84

ix

BWTR£HWUCHIOWI

During the last two decades rapeseed has developed into
a major cash crop in Canada and is also beginning a parallel
development in the United States. In 1987, 2,671,000 hectares
were seeded in Canada with a production of 1,442 kilograms per
hectare(FAO Production Yearbook 1987).

The: development. of neW' cultivars ‘with improved. oil
quality has been a major focus around the world, including
Canada. In 1968, the first low erucic acid.(5 percent or less)
cultivar produced in Canada, was Midas (Brassica napus). Span
and Torch are other examples of low erucic acid rapeseed
cultivars from Brassica campestris that were grown in Canada
(Daun 1983). Between 1972 and 1977, the erucic acid content of
Canadian rapeseed oil was further reduced through genetic
improvement to 2 percent or less.

The new commodity 'canola' (Canadian Oil low acid) was
adopted by Canola Crushers Association of Canada to
distinguish the new rapeseed with low erucic acid content in
the oil (2 percent or less) and low glucosinolate content in

the defatted meal (30p mol/g or less) from the old or common

rapeseed.
Importation of canola oil into the united States has
grown considerably in the last few years and demand is

increasing very rapidly. Only 12,000 metric tons of oil were

2

imported into the United States in 1985, compared to an
estimated 205,000 metric tons in 1989 and 400,000 in 1993.
Prior to 1985, there was less than 20,000 hectares of rapeseed
and canola cultivated annually in the United States. In 1985,
the U.S. Food, and. Drug'.Administration (FDA) granted. GRAS
(generally recognized as safe) status to low erucic acid
rapeseed (canola) oil (National Archives and Records 1985).
Since the fatty acid composition of canola oil has been
increasingly recognized by nutritionists as beneficial in
human diets, the consumption of imported canola oil in U.S.
has increased (Foreign Agricultural Service 1989).

It is expected that the development of hybrid rapeseed
cultivars might lead not only to increased seed yields, but
also to improvement in quality components such as oil
constituents. The successful development of hybrid canola
cultivars requires three components: an acceptable level of
heterosis that offsets the increased cost of hybrid seed, an
efficient mechanism of pollination, and sufficient pollen
transfer to seed parents to produce enough hybrid seed
economically.

Historically, the development of hybrid cultivars of
corn, sunflower, sorghum and many horticultural species has
led to significant increases in seed yield due to the
phenomenon of heterosis or hybrid vigor (Duvick 1984, Furgala
et al.1979, Miller & Kebede 1984). The yield potential of

single cross oilseed rape hybrids (Brassica napus L.) has

3
attracted considerable interest from oilseed breeders, seed
producers and growers around the world.

It is a common belief among plant breeders that
heterosis, measured as the superiority of hybrids over their
mid-parent, is proportional to genetic distance between their
respective parents (Ghaderi et al. 1979; Murty et a1. 1965;
and Ramanujam et a1 1974). Genetic distance between
populations can be defined either on the basis of known
relationships or their genetic composition (Jacquard 1974).
Moll et a1. (1962; 1965) reported that within a restricted
range of diversity, distant populations of corn.(;§§ gays L.),
as judged from their ancestral relationship and geographical
separation, produced higher levels of heterosis for yield in
their hybrid progenies compared to that of Fy's from closely
related genotypes.

However, there should be an optimum degree of genetic
divergence for maximum expression of heterosis (Moll et al.
1965). This optimum occurs within a range of divergence that
is narrow enough so that incompatibility barriers such as
those caused by cytological irregularities are not present.

The objectives of this research were:
1. To use multivariate statistics to investigate the genetic
relationships and distance among a set of 30 canola genotypes
and cluster genotypes based on genetic distance.
2. To select parents from these clusters and make crosses

within and between the groups and compare the resulting

 

heterosis of the crosses.
3. To study the level of heterosis for a set of agronomic

traits.

RENHEMVCHFLIFERAJTHUE

2.1 Rapeseed vs. Canola

The word "rapeseed" is derived from a Latin word "rapum",
which means turnip. All of the oilseed rape species belong to
the family Brassicaceae. Unlike most of the other oilseeds,
rapeseed originates from the various species belonging to the
genus Brassica. These species are B. napus, B. campestris and
B. junceae. All of these species are similar in their
phenotypic appearance but differ in chromosome and ploidy
levels. B. napus is an allotetraploid with a chromosome number
of 2n = 38, where as B. campestris (2n.= 10) and.B. olearaceae
(2n = 9) are the basic diploid species from which.B. napus was
derived. These species have been defined with only slight
differences in various parts of the world. In Canada and
Europe the seeds of rape and turnip rape are named rapeseed,
whereas in India, leaf mustard and turnip leaf is considered
rapeseed.

All traditional rapeseed cultivars contain high erucic
acid rapeseed oil (HEAR), ranging from 22 to 60 percent. Such
a high amount of erucic acid decreases the nutritional value
of the oil, while another chemical compound, glucosinolates,
limits the feeding value of the meal. These problems were
eliminated by the Canadian 'researchers who genetically

modified the traditional rapeseed cultivars by lowering the

6
amount of erucic acid to 2 percent or less and by reducing

the glucosinolate content in the meal to less than 30 p mol

glucosinolates / g or less of meal.

Since genetically modified cultivars of rapeseed have a
very low amount of erucic acid.and.glucosinolate content, they
have been designated officially named by the Canadians as
"canola" for Canadian oil with low acid. To better exploit the
high.quality oil of canola, there is an interest in.developing
high yielding hybrids. In Brassica the exploitation of
heterosis for oil content and yield productivity has not
reached the same commercial level as in corn, so efforts are
being made all over the world to utilize the heterozygous
advantage of the F1 hybrids (Shiga, 1976, Anand & Mishra,
1984). In order to formulate:a successful hybrid program it is
essential to estimate the extent of heterosis for seed and oil
yield in this crop.

Various efforts are currently underway to develop
cytoplasmic male sterility CMS and genetic restorer systems
for the production of hybrid cultivars. Ogura (1968) found a
male sterility inducing cytoplasm in radish (Raphanus sativus
L.). Bannerot et a1. (1977) transferred the nucleus of B.
napus into the radish cytoplasm using intergeneric crossing
followed by a series of backcrosses. Fu (1981) discovered male
sterile plants from the seed of the cultivar Polima (B. napus)
and this male sterile cytoplasm was designated as 29; by Fan

and Stefansson (1986). Shiga et a1. (1983) studied the nap CMS

7
types of some European B. napus cultivars and their ability to
restore male fertility in nap CMS lines. They also found that
a few cultivars could act as "maintainers" of the gap

cytoplasm since they possessed a male fertile cytoplasm but

had no male fertility restoration genes.

2.2 Genetic Distance

Genetic diversity plays a very crucial role in improving
plants to meet different.human needs. Modern.crop species were
selected from the vast array of species in the plant kingdom
and farmers have further selected among existing diverse
genotypes to improve these species. Later, plant breeders
began crossing different genotypes to create additional
diversity by recombining the genes of one parent with those of
the other parent. It is assumed that a cross will generate
successful progeny if both parents possess different genes
(i.e. exhibit adequate genetic distance between them).
Furthermore, plant breeders make the crosses with the aim of
creating new and improved genotypes for a specific trait,
which requires that the parents are at least genetically
distant for the trait to be improved.

The amount of genetic variability is very important in a
population and is a parameter which can be easily measured
since it is influenced by the genetic distance between the
parents (Cowen & Frey, 1987 a). The second parameter of

significant importance is the amount of mid-parent heterosis,

8

exhibited by the hybrid population (Falconer, 1981).

Genetic distance between parents varies with the method
of estimation. One means of estimation is to compare the
parents on the basis of various heritable traits and attempt
to measure the genetic distance through differences in
phenotype. Such an approach has been used by various
researchers both for the qmantitative (e.g., morphological
traits), as well as qualitative traits. Another approach is
based on examining the pedigree data under a set of
assumptions concerning ancestral genealogy, ancestral
genotypes and unrestricted genetic transmission between the
parents and offsprings. These approaches, specifically for
quantitative traits and their application in breeding are
discussed below with the type of data that is used to
determine the genetic distance. Attempts have been made here
only to review the methods of estimating genetic distance and
research where the quantified genetic distances between
parents are related to some form of breeding behavior in the

hybrid population.

2.2.1 Estimation of Genetic Distancg

 

Morphological data can be used for estimating the
relationship ~between two populations. This is achieved by
obtaining a reliable estimate of the phenotype of the
population for univariate or multivariate traits and comparing

them statistically. This approach gives a statistical distance

9

between the population and any inferences about the genetic
distances are made with the assumption that the phenotypic
differences reveal underlying genotypic differences. Generally
speaking, a large phenotypic difference results in a large
genetic distance, regardless of the extent of underlying
genetic differences. More precise inferences are possible,
especially for highly heritable traits through the examination
of more traits, since this assays genetic differences at more
loci.

Several statistical methods have been used to estimate
the genetic distance between two populations with phenotypic
data. One of the commonly used approaches is to calculate the
Euclidian distance (ED) between populations (Goodman, 1972)
where the distance between the i‘'1 and j‘11 population is A

n
EDij = I E (Xik ' 1102110
k=l
Xi, Xjk = mean of the k1h of n traits for the ith and j‘h
population, respectively that have been
standardized by dividing by the standard deviation
of the k‘h trait.
Each population could be visualized as a point in 'n'
dimensions corresponding to 'n' traits where the mean for the
kth trait for the population coordinates that point along the
k‘11 axis. EDij is actually the distance between the i‘h and the
jth points in this multidimensional space, ED" is also

‘1

equivalent to the distance between the principal components

10

derived from an analysis of the correlation matrix between the
traits (Goodman, 1972) where

n
EDij = [ 2 (Ya; " ijlzlm

K=l

Yk, Yjk = Coordinates of the k-th of 'n' principal components
for the i-th and j-th population respectively.

In.case.of principal component.analysis, the original 'n'
axes corresponding to each trait under study are rotated in
space to correspond to new variates that are uncorrelated to
each other. Although the population points remain at the same
positions, their location in the new dimensional space are now
defined by the coordinates of the new axes. These new
coordinates for the ith population are called its principal
components score. The Euclidean distances such as Pearson's
(1926) coefficients or racial likeness or Sokal's (1961)
distance could be used generally when all the measured traits
are uncorrelated (Goodman, 1972). This would seem to be an
unlikely scenario, especially when a large number of traits
are compared, and ignoring the correlation may result in an
exaggerated genetic distance, especially when the correlation
has a genetic basis. This would be comparable to giving some
gene differences more weight than others in the distance
measured.

Another method used first by Mahalanobis (1936)
addressed the problems of the intercorrelated traits by

calculating a generalized distance (D) where the euclidean

11
distance is adjusted within the population correlation matrix
in such a way that the distance between i‘1| and j‘'1 population

is

C
II

ij [0% - Xi) R'l (xi - xj)]“2

n
fi [ E {(Ym "%02 /1J]”2
k=1

0
II

X“ )g = vector of the standardized means of the 'n' traits
for the i,h and jth populations.
R“1 = inverse of the correlation matrix of the 'n' traits.
k = eigenvalue of the kth principal components.

D is the multivariate generalization of the students 't' test
(Hotteling, 1954)and is commonly used for equality testing of
the two means and the EV could be used for calculating the
Hotteling's ”6“ test of the similarity of two mean vectors in
conjunction with a multivariate analysis of variance. All the
eigenvalues equal unity when the measured traits are not
correlated and 'D' then becomes equal to euclidean
distance(ED). The problem with the 'D' is that the principal
components which are with very small eigenvalues and which
accounts for very little of the overall variability may not be
biologically significant, and could have an inflated
contribution from more important principal components
(Goodman,1972).

An alternative would be to include only some of these

standardized principal components in the distance (Goodman,

12
1972). This approach projects the population point in a space
with the fewest but hopefully more biologically significant
dimensions. Goodman (1972) suggested using the k-th principal
components in the distance estimate only when the AK>=K,
whereas a less conservative approach could be to use all the
principal components where AK > = 1.

Calculations and interpretations on the distances becomes
more complex when there is heterogeneity between the
covariance matrices of the population being compared, since
this may result in midiffering from qi(Atchley et a1. 1982).
The heterogeneity of the covariance matrices reflects
different types of relationships of the traits within
different populations. This could be due to the biological
differences brought about by genetic and developmental
differences and the interaction of these factors with the
environment where the traits are measured.

When a plant breeder wants to make improvement for a
particular trait by crossing parents with high means for that
trait, the parents should not show excessive phenotypic
variability for that trait. In such case, the plant breeder
should measure the phenotypic variability for a substitute
morphological trait and hope that it is predictive of genetic
variability for the desired trait. Populations with a narrow
genetic base, normally do not have adequate variability for
other morphological traits for a successful application of

this method, particularly when the other traits are

13

correlated to the trait under improvement. The use of
quantitative traits has a great potential to assay genetic
distance at many loci because there are probably several
hundreds of gene differences between two populations at the
extreme ends of the expression range with multiple
morphological traits. However, it seems unlikely that this
type of extreme range would exist in an elite breeding
population.

Although genetic distances are pairwise comparisons
between populations, they could also be used to show'a broader
relationship which might exist among all genotypes. This may
be achieved by coupling the resulting distance matrix with a
clustering technique. This is a very common technique in
taxonomy and classification work, which could also be applied
to parental breeding populations.

Many reports exist of the use of quantitative trait
distance techniques in taxonomy, classification and evolution
studies. However, relatively few studies have related parental
distances to any form of breeding behavior exhibited by a
hybrid population. The morphological distance between parents
has been positively correlated to the mid-parent heterosis for
grain yield of hybrids of rapeseeds (Brassica napus L.)
(Lefort-Buson et al. 1987), soybean (Glycine max L. Merr.)
(Chauhan and Singh, 1982), wheat (Triticum aestivum L.)
(Shamsuddin 1985, Cox & Murphy, 1990), dry beans (Phaseolus

vulgaris L.) (Ghaderi et al. 1984), peanut (Arachis hypogaea

14

L.) (Arunachalam, 1984; Isleib & Wynne, 1983) , maize (Zea mays
L.) (Prasad & Singh, 1986) and tomato (Lycopersicum esculentum
Mill.) (Maluf et a1. , 1983) . No relationship was found in faba
bean (vicia faba L.) (Ghaderi et al., 1984) or oat (Avena
sativa L) (Cowen & Frey; 1987 a) ,however, Cox & Murphy (1990)
reported that the value of their morphological data in
predicting heterosis in a F2 wheat population depended on the
environment in which the data was collected.

There are a few studies relating distance based on
morphological data to genetic variance and the number of
transgressive segregants in segregating populations despite
the importance of these parameters in plant breeding. The
Euclidean distance between oat parents based on morphological
data was reported to be negatively correlated with the number
of transgressive segregants and the generalized variance
(Cowen & Frey, 1987 a) which is a measure of all measured
traits (Sokal, 1965; Goodman 1968).

It has now become a major concern of gene bank managers
and plant breeders that desirable material cannot be evaluated
effectively in large collections (Holden, 1984). Only a
relatively small portion of the germplasm on hand has been
fully described for agronomically useful characters (Duvick
1984). Many of the desirable characters are quantitative, so
multivariate statistical techniques have been suggested and
utilized to a limited extent to measure genetic and phenotypic

divergence among entries and genotypes to aid in planing

15

crosses among genotypes belonging to different clusters
(Bhatt, 1970: Goodman 1973; Sneath, 1976; Camussi et a1.
1983). Studies on several crops have indicated that a large
amount of variation occurs in these collections (i.e. Ashri,
1973; Narayan. & iMacefield, 1976; Hussaini et a1; 1977;
Camussi, 1979; Tolbert et al., 1979).

Ariyo (1987) conducted multivariate analysis studies for
30 genotypes of Okra of diverse ecogeographical origin. He
used.EF analysis technique to measure the genetic divergence
among these genotypes based on 14 different characters. He
identified five different clusters/groups and observed no
relationship between genetic divergence and ecogeographical
distribution. Cluster #1 contained the most geographically
diverse genotypes, including those from Nigeria, Ghana,
Turkey, and Zambia. Duddley & Davis (1966), Bhatt (1970),
Chedda.& Fatokun (1982), and.Dasgupta and Dass (1984), working
with cultivars of alfalfa, wheat, okra, and black gram,
respectively, noted that the cultivars clustered into
different groups irrespective of their countries of origin.

Peters et a1. (1989) did cluster analysis to classify
entries from barley germplasm.collections obtained from AFRC -
IPSR, based on both qualitative and quantitative descriptors
including disease resistance. They concluded that more
relatedness between entries exists within.a«given country gene
pool than between country gene pools. They also suggested that

cluster analysis could be a useful tool for the management of

16
the variation in germplasm collections. Results from crosses
of material between country gene pools were found to produce
more segregation than that from within the gene pools.

Dorothea et al. (1986) sampled the cultivars of sesame
from 20 countries for their morphological variability, and
used factor, cluster, and principal component analysis to
establish. the ‘patterns. of similarities and generate the
grouping taxa. The results indicated that eight major groups
can.beidiscerned, and they concluded that.genetic variation in
sesame can be used to identify the source of genetic material
for crop improvement.

Hillig et al. (1988) evaluated individuals in a sour
cherry germplasm collections from Eastern Europe for their
morphological traits by using principal component analysis.
They observed that families of cold hardy Russian cultivars
generally show a greater morphological resemblance to ground
cherry than did families of less-cold hardy cultivars. They
also reported a gradual range of diversity rather than two
distinct groups of sour cherry based on morphological
variation.

Buson et a1. (1987) studied different types of rapeseed
hybrids from similar and.different.geographic origins in Asia,
Europe and, mixed locations. They conducted the experiments
over two years at two locations and found that mixed group F1
hybrids were taller and more productive in seed yield than

both European and Asiatic groups alone. They did not conduct

17

any multivariate analysis to cluster or group the genotypes
involved in crossing through any genetic distance measuring
technique but assumed that genetic distance was predictable on
the basis of the originiof the cultivar, i.e. Europe and.Asia.
Grant et a1. (1985) conducted 6x6 diallel cross in spring
oilseed.rape.and.reported.positive heterosis for seed yield of
up to 72 percent over the high yielding parent in the hybrid
cross. Heterosis for seed weight, percent oil, plant height,
and lodging resistance‘was not significant. They also observed
negative heterosis for percent protein in some hybrids.

Shamsuddin (1985) performed multivariate analysis of ten
cultivars of spring wheat and grouped them according to
genetic diversity. During the correlation studies of genetic
diversity and heterosis, he found that the parents of most of
the best hybrids belonged to two divergent groups, which
indicates the positive relationship between genetic divergence
of parents and their hybrid performance. He also found that
the correlation between heterosis for grain yield of hybrids
and genetic diversity between parents was positive and
significant (r = 0.45, P = 0.01). Bhatt (1970) measured the
genetic distance quantitatively for 40 genotypes of wheat from
different ecogeographic regions of Australia, using D2
statistics analysis ‘techniquew The :results supported. the
findings of Moll, Salhuana & Robinson (1962), Timothy (1963)
and Murty and Arunchalam (1966) who could not find any

relationship between geographic distribution and genetic

18
divergence. In this study he observed that the genotypes from
different regions had clustered together in seven out of
twelve cases. Conversely, genotypes of the same origin were
distributed in three different clusters.

Bhatt (1973) compared four different systems to select
parents for hybridization in wheat based on: (a) the
conventional method (b) the random method (c) the multivariate
analysis method, and (d) ecogeographic diversity. He found the
multivariate analysis to be more efficient in selecting
parents than that based on ecogeographic diversity. Genetic
drift and selection in different environments could cause
greater diversity than that caused by geographical distance
(Murty & Arunchalam, 1966). Moreover, multivariate analysis is
a sound statistical technique for objective assessment of
parental diversity compared to other methods. The conventional
and random methods were found to be less efficient.
Spagnoletti et a1. (1985) assessed morphological and genetic
distances among six durum wheat lines, two from Algeria, two
from Italy and two from Ethiopia. Morphophysiological
characters were analyzed and multivariate statistics analysis
conducted to ascertain phenotypic and genotypic similarities
among entries. They found large variation among lines of
different origin compared with that found within origin. They
also observed a close relation between phenotypic and genetic
values, suggesting that diversity based upon phenotypic

analysis of morphological characters may be a reasonably good

19
indication of the choice of breeding material.

Balasch et a1. (1984) conducted multivariate analysis
studies on 60 tomato cultivars of different origin and based
their analysis on twenty characteristics. They used two
different methods: the principal coordinate technique, and
Mahalanobis (DU distance method to group cultivars based on
their genetic similarities. Large similarities of association
between both analytical methods were observed. They also
observed that the genetic divergence of the cultivars was in
no way influenced by geographic distribution, since they found
variation in the same geographic group. Rajana et a1. (1977)
studied 21 F, hybrids of tomato derived by crossing 21
cultivars with a common female parent 'A 142' to observe
heterozygote advantage as a function of genetic divergence.
Genetic distance was calculated by Mahalanobisly method. They
observed low yielding hybrids resulting from crossing closely
related or genetically less divergent parents. The poorest F,
hybrid 'A 142' x 'Gamed' had the smallest genetic distance
between parents (D2== 223.83) with a negative heterosis over
the control parent when compared to the high performance of F,
hybrid 'A142' x 'Sioux' (D2 = 2197.17) with a significant
positive heterosis of 71.6 percent over the control parent.

Peter et al. (1978) studied heterosis in relation to
genetic distance in an F, diallel set of ten homozygous
cultivars of tomato. They reported a 5.87 kg yield in an F,

hybrid Mobaci x EPH7 compared to 2.72 kg in Pusa Ruby, the

20
check parent. They noted the same type of results for most of
the crosses but did not find any clear correspondence between
the extent of the genetic divergence and the heterosis for
yield. The reason they explained could be due to the highly
domesticated and selected nature of the cultivars used in the
hybridization program.

Khanna. et. a1. (1977) reported. greater inter-cluster
heterosis than intra-cluster heterosis in the majority of the
cases in an experiment conducted on 50 cultivars of tomatoes.
Cultivars were grouped into clusters based on seven traits.
They pointed out.that.with some exceptions it was not.possible
to correlate the percentage of the heterosis exactly with the
cluster distances. Such a situation would not be unexpected
considering the complex nature of the expression of the
heterosis for any character and a relatively simple model of
statistical distance.

Moll et al. (1962) speculated that genetic differences
among cultivars have probably arisen through geographical
isolation accompanied by a combination of genetic drift
selection in a different environment. They did not use any
statistical analysis to determine genetic distance between
maize cultivars. In this study they chose six cultivars of
maize, two from each of the three geographical regions, i.e.
midwestern United States, southeastern United States, and
Puerto Rico. They observed that greater genetic diversity

based on the geographical separation of the parental cultivars

21
was associated with greater heterosis. They also suggested
that divergent material may have potential utility in yield
improvement in spite of poor local adaptation of cultivars

from distant sources.

2.2.2 Heterosis

Heterosis has been reported among parents of diverse
origins (Sprague, 1966; Brown, 1967 and Heidrich-Sobrinho).
Beal (1880) was the first scientist to conduct experiments in
maize (fig; mags L.) that are analogous to present day methods
used for maize hybridization. It was not until later that the
modern concept of heterosis was established by Shull (1908).
He used the word heterosis to describe the yield stimulation
resulting from increased heterozygosity. He demonstrated that
certain lines when recombined produced F,yields that exceeded
those of the parental cultivars. The commercial use of F,
hybrids did not materialize until the 19305 (Lamkey &
Hallauer, 1984).

Earlier work with maize conducted by East and Hayes
(1912), East (1936), Hayes & Johnson (1939), Johnson & Hayes
(1940), and Wu (1939) suggested that greater heterosis is
observed as a result of the crosses made between the more
distantly related parents, compared to those which are closely
related. This may not be true for an entire range of genetic
diversity encountered in all species.

Niehaus et al. (1966) performed combining ability

22

analysis in sorghum crosses and reported that genetic
diversity could be the key to maximize hybrid vigor. They
also reported that geographically diverse parents, which
presumably were also genetically diverse, produced the highest
yielding hybrid. In a similar study in corn, Kinman and
Sprague (1945) suggested that the range in combining ability
among the inbred parents could indicate the value of parents
in a synthetic variety.

Ashton (1946) in a survey of many self pollinated crops
reported evidence that interspecific crosses showed better
heterosis compared to intraspecific crosses. Heterosis or
hybrid vigor has been defined by many researchers, such as
Hayman, 1957; Burton, 1980; Mather & Jinks, 1982; Mukherjee &
Saha 1982, Jinks, 1983. Shull (1952) defined it as "the
interpretation of increased vigor, size, fruitfulness, speed
of development, resistance to disease and to insect pests,
manifested by crossbred organisms compared with corresponding
inbreds, as the specific results of unlikeness in constitution
of the uniting parental gametes."

The term heterosis, is used to refer to either an
increase in performance over the better parent or an increase
over the mid-parent performance of a cross. Patterson and
Blitzer (1966) defined this term as the improvement of the
heterozygote over the better parent in a cross. Paterniani
(1973) and Burton (1980) suggested that hybrid vigor is of

practical importance only when the offspring exceeds the

23

better parent.

2.2.3 Gene Action in Heterosis

There are many theories, which explain the mechanism of
heterosis, and it might be quite possible that different
mechanisms operate in different organisms. Since hybrid vigor
was utilized first in corn, its success stimulated interest in
searching for the genetic basis for heterosis. Two main
hypotheses have been proposed (Jones, 1917; Brieger, 1950;

Hayman,,1957; Burton, 1980, Hallauer and Miranda, 1981):

1. Dominance Hypothesis

Davenport (1908) reported that hybrid vigor results from
the action and interaction of the favorable dominant genes.
This hypothesis has been challenged because the evidence does
not substantiate the proposed model (Burton, 1980). Jones
(1917) proposed that heterosis occurs as a result of the
dominance of linked factors. Dobzhansky (1941) and others have
proposed that during the evolutionary process, deleterious
recessive genes become accumulated in many species and reduce
the efficiency of the organism, whereas the deleterious
recessive genes effect is masked by their dominant alleles
when the genes are present in heterozygous form. Crow (1952)
discussed inbreeding depression and explained its recovery on
crossing. He suggested that this recovery is the result of

loci where the favorable allele is dominant and the recessive

24

allele is present at a low frequency.

2. Overdominance Hypothesis

Much of the early work on heterosis was done in maize,
which is a cross pollinated crop. In most of the cases, the
crosses, which involved inbred lines, showed heterosis, which
was associated with heterozygosity. As a result of this
investigation, the theory of overdominance was developed by
Shull (1908) who reported that heterozygotes were more
vigorous than either homozygotes. Hull (1952) presented the
same evidence by using the regression of progenies on parents
to show the overdominance in diallel crosses of maize. Mather
(1956) suggested that the genetic advantage of heterozygosity
lies in providing flexibility and greater stability. Robertson
and Reeve (1952), and Haldane (1954) proposed that heterosis
probably results from the physiological advantages of
heterozygotes.

Green and Green (1940) reported that the unit of specific
physiological action may not be one locus but a complex of
loci which are tightly linked together. Many investigators
believe that these complex loci produce the specific effects
of heterosis. Lerner (1954) concluded that overdominance is
widespread and quite important with respect to the characters,
which are closely connected with fitness. Mather (1955)
explained that overdominance in plants can be attributed to

epistasis. Rambaugh & Lonquist (1959) also reported that most

25
genetic effects were additive with a few examples of
overdominance from their studies of diallel crosses in maize.

The overdominance effects, especially in complex traits
like yield, could be due to the multiplicative action of the
components. This was explained in (Phaseolus vulgaris L.) by.
Adams and Duarte (1961) who found heterosis ix: a complex
trait, namely total leaf area in beans. They believed this was
because of the multiplicative relationship that prevailed
between the two components which were leaf area and leaflet
number.

Different planting methods and environments could also
influence heterosis. Santiago et al. (1968) found a higher
level of heterosis in 'hill' than in drill-planted winter
wheat. Pfahler (1966), in his experiments with rye, Secale
cereale found that the heterosis was dependent on the genotype
and environmental interactions. Sprague (1952) also confirmed
this fact in maize. Pawlisch and van Dijk (1965) reported
heterosis in four F,barley hybrids both for forage and grain
yield. Immer (1941) and Suneson (1962) also reported heterosis
for grain yield in barley.

Interest in the possibility of developing hybrid rape
cultivars with the potential for substantially higher seed
yield was stimulated by reports of cytoplasmic male sterility.
Shiga and Baba (1973) and Thompson (1972) discovered a
cytoplasm that induced male sterility in naturally occurring

Brassica napus. Seed yield in F,hybrids was 40 to 60 percent

26

greater than parental cultivars in oilseed forms of Brassica
napus and/or Brassica campestris. Shiga (1976) and Buson 1980)
noted significant heterosis for seed yield in F} Brassica
napus hybrids. Hutcheson et a1. (1981) discovered a naturally
occurring hybrid between Brassica campestris ssp. Sarson
Prain. Cultivar R-500 and Brassica campestris L. var oleifera
metzg which yielded 146 percent of the Brassica campestris var.
oleifera cv. Candle. Sernyk and Stefansson (1983) found that
certain rapeseed (Brassica napus) cultivars in crosses with
cultivar 'Regent' produced progeny that exhibited 38 to 43
percent more seed yield than Regent.

Populations that are widely differentiated through
adaptation to local conditions may fail to show heterosis and
may suffer a reduction of fitness in the F3 generation as has
been shown in studies of Drosophila populations (Wallace and
Vetukhiv, 1955).

Heterosis, like inbreeding depression, depends for its
occurrence on dominance 'd'. The amount of heterosis following
a cross between two particular lines or populations depends on
the square of the difference of gene frequency (y) between the
populations (Falconer 1982). If the populations crossed do not
differ in gene frequency, there will be no heterosis, and the
heterosis will be greatest when one allele is fixed in one
population and the other allele in the other population. Thus
heterosis in F,can be calculated as Hm = Edyfi

Genetic distances based on the genetic composition of

27
populations can 'be expressed either on the basis of the
frequency of different genotypes (genotypic distance) or the
frequency of the various alleles at a given locus (genic
distance). It is the later definition of the genetic distance
that is associated positively with heterosis (Cress 1966;
Falconer 1960). The association of genetic distance and
heterosis has been investigated in several crops species (Moll
et a1. 1965, Murty' et a1. 1965, Ramanujam. et al. 1974).
Heterotic responses have been reported to range from
significantly positive to significantly negative values. Cress
(1966), presented evidence that even in the absence of
epistasis, multiple alleles at a given locus could lead to
either positive or negative heterosis. Hutchinson et al.
(1930) concluded that not only dominance but epistasis or

interallelic gene actions may also be involved in heterosis.

LIATTHRLALS.Ahfl)hIETTflIDS

3.1 Development of Experimental Material

Thirty cultivars and/or breeding lines of winter canola
(Brassica pgpps L.) from different origins and breeding
programs were used for this study. The list is given in
Table 1. (Actual pedigree are unknown since this is
proprietary information.)

These cultivars were first planted on September 7, 1988
at Agronomy Farm of the Crop and Soil Sciences Department in
East Lansing Michigan in a randomized complete block design
with four replications. The plot size used in all experiments
was 6 m X 0.75 m (4.5 m5. There were five rows and row width
was 15 cm apart. The same thirty cultivars were planted the
following year in 1989 at two locations. One location was on
Botany Research farm on College Road at Michigan State
University, East Lansing. The planting date was September 5,
1989 and the statistical design used was randomized complete
block design with four replications. The second location was
the Horticultural Research Farm at Clarksville, Michigan.
Planting date was September 23, 1989 in a randomized complete
block design.

Since the duration of the crop is about 10 to 11 months,
the data on yield components was recorded on the 1988 crop

while data on morphological traits was recorded on the crop

28

29
planted in 1989 at two locations. The reason for recording
yield component data on the 1988 crop, was to complete the
multivariate analysis and allow selection of the parents prior
to the crossing, which had to be completed by May 1990.
Details of the data taken on morphological traits and yield

components are shown below.

Harvest Dates and Methods

Harvest date was July 15, 1990 in East Lansing and July
20, 1990 at Clarksville. A Hege Combine, 1.2 m wide with a
cylinder speed of 600 revolutions per minute was used for this
purpose at both locations. The harvest area was the same as

planting area, i.e. 4.5 m%

Data Collected
1. Crown Diameter

Crown diameter was recorded in millimeters with the help
of a Vernier Calliper one month after the emergence of plants

at both locations.

2. Leaf Length

Leaf length in centimeters was recorded in three
different stages at both locations. First, leaf length was
recorded from the base of the leaf (which is attached to the
stem) up to the first node. The second length was noted from

the base up to the second node, and the third and final length

30

was recorded from the base of the leaf up to the tip.

3. Leaf Width
Leaf width was also recorded at both sites. It was

measured from the center of the leaf to get the maximum width.

4. Number of Branches

Eight plants per cultivar at each location were selected
randomly and.the data on the number of branches were recorded.
Two plants per replication were selected in four replications

for this purpose.
5. Number of Pods
The same randomly selected eight plants were used to

collect data on pod numbers per plant.

6. Number of Seeds per Pod

 

Number of seeds per pod.were also recorded for all of the
thirty genotypes used for this study. The same eight randomly

selected plants were used for this purpose.

7. Hundred Seed Weight

One hundred seed weight (in grams) was recorded for the

same randomly selected plants for all the thirty genotypes.

31
3.2 Multivariate Analysis
Based on the morphological traits and yield components,
this multivariate data set, consisting of the thirty genotypes
of different origins as discussed above was analyzed using the

principal component and clustering techniques.

3.2.1 Analysis of Variance

Proc Anova procedure of the SAS program was used to
determine whether there was a significant difference among the
genotypes under evaluation. The result showed a significant
difference, which permitted clustering analysis in order to

group cultivars.

3.2.2 Principal Component Analysis
Principal component analysis was performed using the

PRINCOMP procedure of the SAS statistical package.

3.2.3 Clustering Analvsis

The purpose of the clustering analysis was to determine
if these thirty genotypes fell into different clusters.
Clusters are made in such a way that the genotypes in a given
cluster tended to be similar to each other while the genotypes
placed in different clusters, tended to be dissimilar.
Initially, nine traits including four yield component and five
morphological traits were used to make the clusters. However,

four morphological traits out of a total of nine were dropped.

32
These remaining included four yield components and only one
morphological trait, which was crown diameter. These five
traits finally contributed towards making three distinct
clusters.

PROC CLUSTER procedure of the SAS program was used for
this purpose. Only two genotypes, LD-9430 and Cascade were
placed in cluster No.1 and two genotypes, Cobra and Winfield,
were placed in cluster No.2. The rest of the twenty six

genotypes were placed in cluster No.3.

3.2.4 Euclidean Distance

Euclidean Distance among the parents used for the
crossing program was calculated using SAS. The distance among
the parents was estimated to allow correlation with the level
of heterosis observed from crossing the six selected parents

in this study.

3.3 Selection of the Parents

Based on the clustering analysis procedures, two
genotypes, LD-9430 and Cascade were placed in cluster No.1 and
selected. as parents for' use in crossing program. These
genotypes belonged to the Calgene and Cargill groups
respectively and exhibited a nonsignificant difference from
each other, so they were placed in the same group/cluster.

Cluster No.2 had also only two genotypes out of thirty

which were analyzed. These two genotypes were also selected as

33
parents to be used in the crossing program. These genotypes,
namely Cobra and Winfield exhibited a nonsignificant
difference from each other. Both of the genotypes belonged to
Calgene Inc. group.

Cluster No.3 comprised the rest of the twenty-six
genotypes out of thirty. Non of these genotypes were
significantly different from each other, since the actual
differences in morphological traits and yield components for
these 26 genotypes were very small. Thus, all were placed in
one cluster. The two genotypes, Ceres and CC-4, from Calgene
and Cargill, respectively were most similar to each other and
were selected as parents from this cluster. The distance found
between these:genotypes through.clustering analysis was almost
equal to the distance found among the parents selected from
the other two clusters. In order to make the selection
uniform, these parents were identified and selected from the
third cluster on the almost equal distance criteria as stated

above.

3.4 Crossing Program

Six parents representing two genotypes from each cluster
were then used in the crossing program. Cultivars used as

parents were as follows:

1. LD-9430 :

2. Cascade : Cluster No.1

34

3. Cobra :
4. Winfield : Cluster No.2
5. Ceres :
6. CC-4 } Cluster No.3

A half diallel crossing scheme was conducted between the
six parents using the formula n(n-1)/2 which is equal to 6(6-
1)/2=15 crosses. In this way these crosses were made within

and between the clusters.

3.4.1 Floral Morphology

The flower of Brassica m L. has six stamens, the
outer ones are distinctly shorter than the inner four, which
surround the stigma. The anthers are shorter than the stigma
at an early bud stage but prior to flower opening the filament
elongates and carries the anthers upward so that they become
as high as the stigma. The flower begins to open before 8 a.m.
and continues to open until about noon. The flower remains
open three to four days, after which.the sepals and petals are

shed.

3.4.2 Crossing Technigpe

Crossing work was started on May 4, 1990 and continued
through May 13, 1990 in the Botany Research Farm at Michigan

State University in East Lansing. It involved the following

35
steps:
1. Bagging the inflorescence approximately one day prior to
the bud opening.
2. Emasculation of buds was completed the following day after
bagging the plants.
3. Dusting the male parent pollen and covering it with a
butter paper bag until pod development started. This was done
to avoid contamination from any foreign pollen.

Five plants from each parent in each replication were
randomly selected for the crossing program. Unwanted or
matured flowering buds from the covered inflorescence were
clipped off before crossing with the help of scissors to avoid
any selfing. Such types of buds were clipped off and removed
and were not used in crossing. Forceps were used to emasculate
the buds by opening them and taking out the anthers. All
possible care was taken not to injure the stigma and ovary
after emasculation. Again the emasculated buds were covered
with butter paper bags until the mature pollen from the
desired plants was available for crossing purposes.

Most of the crossing was done at about 10 to 11 a.m. in
the field when the fresh flowers were in full bloom, and the
pollen dehiscence was at a maximum. Ten to fifteen buds were
emasculated and pollinated with one inflorescence of the male
plant. Two to three inflorescenses were used on one plant to
produce the maximum amount of seed. Details of the crosses

made are given in Table 2.

36

3.5 W
3.5.1 Data Recording

Data on morphological traits and yield components were
recorded (Table 3). All of the 15 crosses, along with their
six parents were planted at East Lansing, Michigan on August
25 and, Clarksville on September 3, 1990. Harvesting of these
crosses and their parents was done in East Lansing on.July 24,

1991 and Clarksville on July 22, 1991.

3.5.2 Data Analysis
3.5.2.1 Analysis of Variancg

Analysis of variance for each character listed above was

performed for significance by MSTAT.

3.5.2.2 Correlation Studies

Correlation coefficients were calculated to establish the
relationship among the characters noted for all
cultivars/ crosses under study. The MSTAT program was used for

this analysis.

3.6 Estimation of Heterosis
Heterosis estimation which was the major part of the
study was also done. The heterosis for each character under

study was measured.

37

Mid-Parent Heterosis

This measurement was taken by comparing the performance
of F} with the average performance of its parents. The Spread
Sheet program of LOTUS 123 was used for this study and
following formula used to calculate the heterosis:

Mid-parent heterosis (%) = (F,- MP) / MP x 100

where F} = performance of hybrid.

MP = average performance of parent per se

(parent No.1 + parent No.2 / 2)

3.7 Chemical Analysis
3.7.1 Oil Content

Seed samples (5 grams each) were collected after the
harvest of the crop at a moisture level between eight to nine
percent. Samples were taken from all three replications at
East Lansing and Clarksville. After packing and labeling the
samples were sent to the American Pedigreed Seeds for

analysis.

3.7.2 Determination of Oil Content by NMR

Oil content of seed samples was determined by Nuclear
Magnetic Resonance (NMR) using 1.2 gram of clean, whole and
unbroken seeds for each sample dried in aluminum cups for two
hours at 130°C. After drying, the seed samples were
immediately transferred to plastic NMR vials to bring them to

equilibrium with NMR's room temperature. The NMR analyzer was

38
calibrated using samples containing 0, 50, 100 percent oil.
Oil content of all the seed samples was determined

individually.

 

 

39

Table 1. Cultivars and origin of 30 winter canola

(Brassica napus L.) genotypes

Cultivar

1.

2.

3.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

TH-lOl
TH-102
TH-103

Liraborn

Glacier
Crystal
SV0220
SV0238
SV0253
SV0255

SV0261

88-A
88-B
88-C
88-D
LD-943O
LD-9431
Allstar

Ceres

Source

Canola, Inc. 8910 Purdue Rd.

Indianapolis, IN 46268

Bonis & Co. Ltd., P.O. Box 317

Lindsay, Ont., Canada, K9V5Z4

Ameri-Can (Calgene) Pedigreed Seed

7664 Moore Rd., Memphis, TN 38120

20.

21.

22.
23.

24.
.25.
26.
27.
28.
29.

30.

Cobra

Diadem

Winfield
CC-OOl
CC-OO7
CC-l
CC-2
CC-4
Cascade
Libraska

Libravo

40

Ameri-Can (Calgene) Pedigreed Seed

7664 Moore Rd., Memphis, TN 38120

Cargill Hybrid Seeds, 2600

W. Galena Blvd., Aurora,

IL 60506

41
Table 2. List of 15 canola (Brassica napus L.) crosses made by

Using six selected parents from three different clusters

1. LD-9430 X Cascade
2. LD-9430 X Cobra

3. LD-9430 X Winfield
4. LD-9430 X Ceres

5. LD-9430 X CC-4

6. Cascade X Cobra

7. Cascade X Winfield
8. Cascade X Ceres

9. Cascade X CC-4

10. Cobra X Winfield
11. Cobra X Ceres
12. Cobra X CC-4

13. Winfield X Ceres
14. Winfield X CC-4

15. Ceres X CC-4

 

42

Table 3. List of morphological traits and yield components

Abbreviations

10.

11.

12.

13.

14.

DE

Cwn Dia
Lf L 1.nd
Lf L 2.nd
Lf L full
Lf Wd

DOF

DTM
Br/Plt
Pd/Plt
Sd/Pod
100 Sd/Wt
Yld Kg/ha

Oil %

selected for data recording

Traits

Days to emergence

Crown diameter in mm

Leaf length (up to lst node) in cm
Leaf length (up to 2nd node) in cm
Leaf length full (up to the tip) in cm
Leaf width in cm

Days of flowering

Days to maturity

Branches per plant

Pods per plant

Seeds per pod

One hundred seed weight in grams
Seed yield in kilograms per hectare

Percent oil content

RESULTS

4.1 Analysis of Variance

Analysis of variance and mean separation tests were
performed to measure the effects of different parents and
crosses on various morphological and yield component traits

(Tables 1 to 9).

Days to Emergence

No significant difference was detected between the
crosses and parents for this character at either East Lansing
or Clarksville, even when data was analyzed separately for

both locations.

Crown Diameter

Differences between the parents and crosses showed
significant effects on crown diameter using the combined
analysis for both locations. The same trend was observed when
analysis was performed separately for both locations.

The crosses LD-9430 x Ceres and Ceres x CC-4 shared the
highest mean value of 4.10 mm for each cross in a combined
analysis (Table 4a), whereas in East Lansing, Ceres x CC-4 had
the highest mean value of 5.13 mm, followed by the parent
Ceres with a mean value of 5.06 mm (Table 5a). In Clarksville

the cross LD-9430 x Cascade and Cascade as a parent shared the

43

44

first position with a mean value of 3.23 mm (Table 6a).

Leaf Length (1st, 2nd, & 3rd node)

This character, recorded at three different stages, i.e. ,

 

first observation up to first node, then up to second node,
and finally up to the tip of the leaf (third node), starting
from the basal attachment of the leaf with the stem, was
significantly affected by the difference between parents and
crosses, when a combined analysis for both locations was
performed. Similar results were observed when separate
analyses were conducted separately for both locations. It is
interesting to note that based on mean values the parent Ceres
exhibited the largest leaf length observations at three nodes
(Table 4a) in a combined analysis. The same trend also
persisted for the East Lansing location for this parent (Table
5a). However, LD-9430 at Clarksville had.the largest value for

this character analyzed (Table 6a).

Leaf Width

Significant differences for leaf width occurred between
the parents and the crosses for both combined and separate
analyses. The cross Ceres x CC-4 had the highest mean value of
6.18 cm, followed by one of its parents, CC-4, with a mean
value of 6.15 cm in combined analysis (Table 4a). In East
Lansing the parent CC-4 had the highest mean value of 7.56 cm

followed by the cross Ceres x-CC-4 with a mean value of 7.36

45
cm (Table 5a). However, LD-9430 had the highest leaf width
with a mean value of 5.23 cm followed by the cross LD-9430 x

Cascade with a value of 5.2 cm in Clarksville (Table 6a).

Days to Flower

Days to flower varied significantly between the crosses
and the parents in combined analysis. The cross Cascade x CC-4
was the earliest cross to flower with.a mean value of 233 days
from planting, followed by Cascade x Cobra with a mean value
of 235 days (Table 4a). In East Lansing, the parent Cascade
was the earliest to flower with a mean value of 230 days,
followed by Cobra x CC-4 with a value of 231 days (Table 5a).
In Clarksville no significant differences in flowering date
were found between Cascade, LD-9430 x Winfield, Cascade x
Winfield and Cobra x CC-4, all of which shared the highest

common mean value of 240 days (Table 6a).

Days to Maturity

The combined analysis of variance revealed a significant
difference at 0.05 alpha level between the parents and the
crosses for days to maturity. The cross Cobra x Ceres
exhibited the earliest maturity with an average value of 282
days, followed by CC-4 with a value of 290 days (Table 4b).
In East Lansing, Cascade x CC-4 exhibited the earliest
maturity with a mean value of 288 days, followed by CC-4 with

a value of 289 days (Table 5b). In Clarksville CC-4 was the

 

 

46
earliest maturing cultivar with a mean value of 291 days
followed by Cascade x CC-4 with a value of 293 days (Table

6b) .

Branches per Plant

The combined analysis of number of branches per plant
revealed significant differences between the parents and
crosses. LD-9430 had the highest number of branches per plant
with an average of 8.4 branches, followed by LD-9430 x
Winfield with an average number of 8.4 branches per plant
(Table 4b). In East Lansing LD-9430 and the cross LD-9430 x
Winfield both had a value of 8.3 branches per plant (Table
5b). In Clarksville LD-9430 and LD-9430 x Winfield again had
the largest value, both with the highest common value of 8.5

branches per plant (Table 6b).

Pods per Plant

Significant differences were found between the parents
and crosses using combined analysis of variance for both
locations. The cross Cobra x CC-4 had the highest number of
pods with a mean value of 231 followed by Ceres x CC-4 with a
mean value of 221 (Table 4b). In East Lansing, Cobra x CC-4,
exhibited the highest number of pods with.a mean value of 230,
followed by Winfield x CC-4 with a value of 226 (Table 5b).
However, in Clarksville, Cobra x CC-4 had the highest value

with an average of 233 pods per plant, followed by Cascade x

 

 

 

47

CC-4 with a mean value of 225 (Table 6b).

Seeds Per Pod

Significant differences were found between the crosses
and their parents in the combined analysis. The cross
Cobra x Ceres had the highest mean value of 32.8, which was
followed by the cross Cobra x CC-4 with an average of 32.3
seeds per pod. In East Lansing, again the cross Cobra x Ceres
had the highest mean value of 34.0 seeds per pod. This was
followed by the cross Ceres x CC-4 with an average of 32.6. In
Clarksville, the cross Cascade x CC-4 was the highest with an
average of 32.6, which was followed by the cross Cobra x CC-4

with a mean value of 32.3.

One Hundred Seed Weight

Significant differences occurred between the parents and
crosses both for combined and separate analyses at both
locations. In.the combined analysis, the cross Cascade x Cobra
had the highest one hundred seed weight with an average value
of 0.52 gms, followed by Cascade with.a mean value of 0.51 gms
(Table 4b). In East Lansing and Clarksville, the same trend
was observed, i.e. both crosses maintained their respective
positions with the highest and second highest average one
hundred seed weight at both locations separately (Tables 5b &

6b) .

 

48

MM).

Highly significant differences were found between the
crosses and parents for yield in the combined analysis.
Significant differences were also found when separate analysis
for both locations were performed. In the combined analysis,
the cross Cobra x Ceres had the highest mean yield of 3314
kg/ha followed by Cascade x CC-4 with 3258 kg/ha (Table 4b).
In East Lansing, Cobra x Ceres also produced the highest yield
at 3616 kg/ha followed by LD-9430 x Ceres at 3510 kg/ha (Table
5b). In Clarksville, Cascade x CC-4 gave the highest average
yield of 3258 kg/ha followed by Cobra x Ceres at 3012 kg/ha

(Table 6b).

Percent Oil Content

Significant differences between the parents and crosses
were observed when a combined analysis as well as separate
analyses of this trait for both locations were performed. The
cross Winfield x Ceres had the highest oil content in combined
analysis with a mean value of 45.1 percent followed by
Winfield x CC-4 at 44.2 percent oil content (Table 4b). In
East Lansing and Clarksville exactly the same trend for all of
these crosses occurred as for combined analysis (Tables 5b &

6b) .

49

 

 

 

Table 4.a Means for seven traits of 21 entries/crosses
of canola combined for both locations,
East Lansing and Clarksville, MI 1991
Entries/Crosses DE* Cwn Lf L Lf L Lf L Lf DOF
Dia 1.nd 2.nd full Wd

1.LD-9430 10 3.83 5.35 7.10 11.03 5.81 246
2.Cascade 8 3.78 5.88 7.16 11.53 5.63 235
3.Cobra 9 3.70 5.40 -6.96 11.93 5.56 241
4.Winfield 9 3.23 5.75 7.75 12.26 5.58 242
5.Ceres 10 4.03 6.28 8.00 13.93 5.80 246
6.CC-4 8 4.00 5.66 7.43 12.21 6.15 236
7.LD-9430xCasc. 9 3.95 5.60 7.13 11.50 5.83 237
8.LD-9430x Cobra 9 3.88 5.25 6.96 11.41 5.81 240
9.LD-9430XWin. 9 3.48 5.35 7.48 11.60 5.78 241
10.LD-9430XCeres 10 4.10 6.15 7.55 11.53 5.91 247
11.LD-9430xCC-4 9 3.96 5.76 7.16 11.75 5.80 238
12.Casc.xCobra 8 3.86 5.56 6.93 11.80 5.61 235
13.Casc.xWinf. 9 3.43 5.75 7.36 12.01 5.63 237
14.Casc.xCeres 9 3.91 6.26 7.50 12.88 5.86 241
15.Casc.xCC-4 8 3.86 5.73 7.23 11.78 5.93 233
16.CobraxWinf. 9 3.40 5.50 7.26 12.15 5.95 237
17.CobraxCeres 9 4.00 5.76 7.50 13.11 5.90 243
18.CobraXCC-4 8 3.80 5.51 7.00 12.13 5.93 235
19.Winf.xCeres 10 3.73 5.93 7.95 13.26 5.73 243
20.Winf.xCC-4 8 3.51 5.58 7.60 12.11 5.60 236
21.CerestC-4 10 4.10 4.70 7.68 13.21 6.18 242
Grand mean 9 3.79 5.65 7.36 12.15 5.81 240
LSD (0.05) ns 0.18 0.19 0.16 0.19 0.17 1.9

* Abbreviations & units of the characters

table 3, p. 42

are explained in

 

50

Table 4.b Means for seven traits in 15 canola crosses
combined for both locations,
East Lansing and Clarksville, MI 1991

 

Entries/Crosses DTM* Br/ Pd/ Sd/ 100 Yld Oil
Plt Plt Pod Sd/Wt kg/ha %

 

1.LD-9430 305 140 17.6 .47 2130 40.0
2.Cascade 292 136 20.8 .51 2246 41.3
3.Cobra 301 . 172 26.0 .49 2512 40.7
4.Winfield 301 158 21.8 .42 2253 42.3
5.Ceres 298 . 184 29.1 .38 2829 40.6
6.CC-4 290 . 205 27.8 .37 2634 41.6

152 20.8 .48 2465 42.6
186 25.5 .48 2816 42.4
170 22.1 .43 2548 42.8
207 29.6 .40 3240 40.9
212 27.5 .40 3022 41.3

7.LD9430XCaS. 300
8.LD9430XCOb. 302
9.LD9430XWin. 302
10.LD9430xCer. 300
11.LD9430XCC-4 295

m\l\l0\0\\l0\\l\lm\l\lm\l\lmm\lm\lm
fiUONQWflQHbOWQbQWQQMQN-h-

 

12.Cas.xCob. 297 180 26.5 .52 2756 41.0
13.Cas.xWin. 298 164 23.8 .46 2558 42.7
14.Cas.xCer 294 . 197 30.1 .42 3178 42.2
15.Cas.xCC-4 290 217 31.0 .44 3258 40.8
16.Cob.xWin. 299 181 26.6 .42 2698' 40.6
17.Cob.xCer. 282 217 32.8 .42 3314 42.7
18.Cob.xCC-4 293 231 32.3 .45 3181 42.8
19.Win.xCer. 297 . 197 30.6 .42 2932 45.1
20.Win.xCC-4 295 221 29.0 .43 2976 44.2
21.Cer.xCC-4 294 221 31.6 .40 3061 43.5

Grand mean 296 7.3 188 26.8 .44 2790 42.1
LSD (0.05) 14.9 0.67 30 7.2 ns 387 2.5

* Abbreviations & units of the characters are explained in
table 3, p. 42

51

 

 

 

Table 5.a Means for seven traits of 21 entries/crosses
of canola grown in East Lansing, MI 1991
Entries/Crosses DE* Cwn Lf L Lf L Lf L Lf DOF
Dia 1.nd 2.nd full Wd

1.LD-9430 9 4.70 5.90 8.20 13.60 6.40 242
2.Cascade 7 4.33 7.46 8.80 14.80 6.40 230
3.Cobra 8 4.63 7.36 9.50 16.03 6.40 235
4.Winfield 8 3.93 7.10 10.20 16.40 6.00 239
5.Ceres 9 5.06 8.30 10.75 19.83 7.06 243
6.CC-4 7 4.80 7.46 9.93 16.50 7.56 232
7.LD-9430XCaSC. 8 4.60 6.53 8.40 14.70 6.46 233
8.LD-9430XCobra 8 4.80 6.40 8.53 14.76 6.60 237
9.LD-9430XWin. 9 4.10 6.20 9.23 14.86 6.16 242
10.LD-943OXCeres 9 5.00 7.86 9.50 14.76 6.86 243
11.LD-9430XCC-4 8 4.76 7.13 8.93 15.10 6.40 233
12.Casc.xCobra 7 4.50 7.46 9.10 15.67 6.33 229
13.CaSC.XWinf. 8 4.03 7.20 9.43 15.80 6.23 233
14.Casc.xCeres 8 4.90 8.00 9.56 17.63 6.86 237
15.CaSC.XCC-4 7 4.63 7.43 9.23 15.43 7.10 229
16.CobraXWinf. 8 4.20 7.20 9.70 16.40 6.90 233
17.CobraxCeres 8 5.06 7.76 10.26 18.30 6.73 240
18.CobraxCC-4 7 4.66 7.36 9.50 16.13 6.93 231
19.Winf.xCeres 9 4.60 7.60 10.60 18.40 6.46 240
20.Winf.XCC-4 7 4.23 7.20 10.20 16.13 6.33 231
21.CerestC-4 9 5.13 5.06 10.16 18.43 7.36 240
Grand mean 8 4.60 7.14 9.51 16.17 6.48 236
LSD (0.05) ns 0.22 0.19 0.17 O 19 0.19 2.1

* Abbreviations & units of the characters

table 3, p. 42

are explained in

52

5.b Means for seven traits in 21 entries/crosses
of canola grown in East Lansing, Mi 1991

Table

 

 

 

Entries/Crosses DTM* Br/ Pd/ Sd/ 100 Yld Oil
Plt Plt Pod Sd/Wt kg/ha %
1.LD-9430 301 8.3 146 18.3 .47 2256 40.2
2.Cascade 290 7.2 142 21.6 .51 2411 41.3
3.Cobra 297 6.8 175 26.6 .49 2712 41.0
4.Winfield 296 7.8 162 21.6 .42 2435 42.3
5.Ceres 295 6.8 187 30.3 .38 3096 39.5
6.CC-4 289 6.7 205 28.0 .37 2858 41.4
7.LD9430xCas. 294 7.8 159 21.6 .50 2625 42.7
8.LD9430xCob. 299 7.7 190 26.3 .50 3002 42.4
9.LD9430xWin. 297 8.3 180 23.0 .45 2781 42.6
10.LD9430xCer. 298 7.5 212 31.0 .41 3510 41.6
11.LD9430xCC-4 291 7.5 219 28.3 .40 3302 41.4
12.Cas.xCob. 294 7.0 188 27.6 .52 2987 41.2
13.Cas.xWin. 294 7.5 171 24.3 .47 2804 42.8
14.Cas.xCer 291 7.0 206 31.3 .43 3459 42.2
15.Cas.xCC-4 288 6.8 209 29.3 .44 3258 40.9
16.Cob.xWin. 294 7.7 179 26.6 .44 2877 40.8
17.Cob.xCer. 295 6.8 221 34.0 .44 3616 42.9
18.Cob.xCC-4 290 6.7 230 32.3 .46 3405 42.9
19.Win.xCer. 293 7.5 201 29.0 .43 3146 45.2
20.Win.xCC-4 292 7.3 226 29.0 .44 3203 44.5
21.Cer.xCC-4 292 6.8 222 32.6 .40 3318 43.7
Grand mean 294 7.3 192 27.4 .44 3002 42.1
LSD (0.05) 2.2 0.83 33 6.3 ns 343.6 2.6

* Abbreviations & units of the characters are explained in
table 3, p. 42

 

53

 

 

 

Table 6.a Means for seven traits of 21 entries/crosses
of canola grown in Clarksville, MI 1991
Entries/Crosses DE* Cwn Lf L Lf L Lf L Lf Wd DOF
Dia 1.nd 2.nd full

1.LD-9430 12 2.96 4.80 6.00 8.46 5.23 251
2.Cascade 9 3.23 4.30 4.53 8.26 4.86 240
3.Cobra 10 2.76 3.43 4.43 7.83 4.73 247
4.Winfield 10 2.53 4.40 5.30 8.13 5.16 245
5.Ceres 11 3.00 4.26 5.23 8.03 4.53 250
6.CC-4 10 3.20 3.86 4.93 7.93 4.73 240
7.LD-9430XCaSC. 10 3.30 4.66 5.86 8.30 5.20 242
8.LD-9430XCobra 11 2.96 4.10 5.40 8.06 5.03 243
9.LD-9430xWin. 10 2.86 4.50 5.73 8.33 5.40 240
10.LD-9430xCeres 12 3.20 4.43 5.60 8.30 4.96 251
11.LD-9430XCC-4 10 3.16 4.40 5.40 8.40 5.20 242
12.Casc.xCobra 10 3.23 3.66 4.76 8.03 4.90 240
13.CaSC.XWinf. 10 2.83 4.30 5.30 8.23 5.03 240
14.CaSC.XCeres 11 2.93 4.53 5.43 8.13, 4.86 244
15.Casc.XCC-4 9 3.10 4.03 5.23 8.13 4.76 237
16.CobraXWinf. 10 2.60 3.80 4.83 7.90 5.00 242
17.CobraxCeres 11 2.93 3.76 4.73 7.93 5.06 247
18.CobraXCC-4 10 2.93 3.66 4.50 8.13 4.93 240
19.Winf.xCeres 11 2.86 4.26 5.30 8.13 5.00 247
20.Winf.XCC-4 10 2.80 3.96 5.00 8.10 4.86 241
21.CerestC-4 11 3.06 4.33 5.20 8.00 5.00 245
Grand mean 10 2.97 4.16 5.22 8.13 4.97 244
LSD (0.05) ns 0.14 0.20 0.20 0.20 0.16 1.8

* Abbreviations &
table 3,p. 42

units of the characters are explained in

54

6.b Means for seven traits in 21 entries/crosses
of canola grown in Clarksville, MI 1991

Table

 

 

 

Entries/Crosses DTM* Br/ Pd/ Sd/ 100 Yld Oil
Plt Plt Pod Sd/Wt kg/ha %
1.LD-9430 310 8.5 135 17.0 .47 2004 39.9
2.Cascade 295 7.2 130 20.0 .51 2081 41.3
3.Cobra 306 6.7 170 25.3 .49 2313 40.5
4.Winfie1d 306 7.8 155 22.0 .42 2072 42.3
5.Ceres 301 6.7 181 28.0 .38 2562 41.7
6.CC-4 291 7.0 205 27.6 .37 2416 41.7
7.LD9430xCas. 306 8.0 146 20.0 .47 2304 42.5
8.LD9430xCob. 305 7.8 182 24.6 .47 2631 42.4
9.LD9430xWin. 307 8.5 161 21.3 .41 2315 42.9
10.LD9430xCer. 301 7.7 203 28.3 .40 2971 40.3
11.LD9430xCC-4 300 7.5 206 26.6 .41 2745 41.3
12.Cas.xCob. 300 6.8 172 25.3 .52 2526 40.8
13.Cas.xWin. 302 7.3 157 23.3 .44 2311 42.6
14.Cas.xCer. 297 7.2 188 29.0 .41 2900 42.1
15.Cas.xCC-4 293 6.8 225 32.6 .44 3258 40.8
16.Cob.xWin. 305 7.8 183 26.6 .40 2520 40.7
17.Cob.xCer. 268 7.0 213 31.6 .40 3012 42.5
18.Cob.xCC-4 296 6.7 233 32.3 .44 2958 42.8
19.Win.xCer. 301 7.7 193 29.0 .40 2719 45.0
20.Win.xCC-4 299 7.3 215 29.0 .42 2748 43.9
21.Cer.xCC-4 297 6.7 221 30.6 .40 2803 43.3
Grand mean 299 7.36 184 26.2 .43 2579 41.9
LSD (0.05) 21.2 0.50 27.3 8.2 ns 437 2.5

* Abbreviations & units of the characters are explained in

table 3, p. 42

 

 

55

H_eiares_i_s

Heterosis was calculated as a percent increase or
decrease over the mid-parent performance for yield and a
series of morphological traits and yield components as
explained in materials and methods. The negative heterosis
value describes the decrease in trait performance while the
positive value indicates the greater heterosis compared to its

mid-parent value.

Days to Emergence

A range of negative to positive percent heterosis was
observed among the crosses for this trait for both combined as
well as separately for each location. The cross Ceres x CC-4
showed the highest percent increase with a mean value of 8.11
percent followed by Cascade x Winfield with a mean of 5.88
percent increase when calculated combined for both locations
(Table 7a).

In East Lansing, the same crosses ‘maintained their
respective positions (Table 8a), whereas in Clarksville,
Cascade x Ceres showed. the highest. performance with an
increase of 10 percent over the mid-parent value followed by
Cascade x Cobra and Cascade x Winfield, both with a common

increase of 5.26 percent (Table 9a).

 

56
Crown Diameter
The cross LD-9430 x Ceres showed the highest heterosis
with a 4.73 percent increase over mid-parent yield, calculated
across both locations. It was followed by Cobra x Ceres with
3.9 percent increase (Table 7a).
In East Lansing, Cobra x Ceres exhibited an increase of
4.4 percent followed by Ceres x CC-4 with 4.06 percent
increase (Table 8a). In Clarksville, Cascade x Cobra was found
with the largest increase of 7.85 percent followed by LD-9430

x Ceres with a 7.38 percent increase (Table 9a).

Leaf Length (lst, 2nd & 3rd node)

The cross LD-9430 x Ceres exhibited the largest percent
increase with a 10.76 percent value, followed by LD-9430 x
CC-4 with an increase of 6.74 percent for the leaf length up
to fist node, when combined for both locations. For leaf
length up to second node, Winfield x CC-4 was the best with a
1.34 percent increase, followed by Cobra x Ceres with an
increase of 1.28 percent. The cross LD-9430 x Cascade proved
to be the best with an increase of 3.52 percent (Table 7a) for
leaf length up to the third node (full leaf length).

In East Lansing, the highest percent increase for the
leaf length up to the first node occurred for the cross
LD—9430 x Ceres with an increase of 10.76 percent. In the same
location, Winfield x CC—4 exhibited the largest leaf length up

to the second node with 1.34 percent increase. For the final

 

57
leaf length (up to the 3rd node) LD-9430 x Cascade had the
largest 3.52 percent increase (Table 8a).

In Clarksville the cross Ceres x CC-4 had the largest
percent increase for the leaf length up to the first node with
6.65 percent, whereas LD-9430 x Cobra was next for leaf length
up to the 2nd node with 3.55 percent increase and for full
leaf length (up to the 3rd node) Cobra x CC-4 had the highest
percent increase of 3.17 Percent over the mid-parent value

(Table 9a).

Leaf Width

The cross Cobra x Winfield exhibited the largest increase
of 5.78 percent heterosis combined for both locations (Table
7a), whereas in East Lansing, it exhibited the largest
increase of 15.17 percent (Table.8a). In.Clarksville the cross
Cobra x Ceres had the highest increase of 9.29 percent (Table

9a).

Days to Floyg;

The cross Cobra x Winfield was the earliest maturing,
with a decrease of -1.76 percent, when heterosis was
calculated across both locations. This was followed by
LD-9430 x Cobra with -1.61 percent decrease (Table 7a).

In East Lansing, the cross Winfield X CC-4 was the
earliest maturing with -1.91 percent decrease, followed by

Cobra x Winfield with a -1.69 percent decrease (Table 8a). In

 

 

 

 

58
Clarksville, the Cross LD-9430 x Winfield was the earliest
maturing with a decrease of -3.22 percent, followed by LD-9430

X Cobra with -2.60 percent decrease (Table 9a).

Days to Maturity

When calculated across locations, the.cross Cobra.x Ceres
was the earliest maturing of all others with a decrease of
-5.95 percent, followed by Cobra x CC-4 with a decrease of
-0.87 percent (Table 7b). When calculated for East Lansing,
the cross LD-9430 x CC-4 was found the earliest maturing with
a decrease of -1.36 percent followed by Cobra x CC-4 with
-0.85 percent decrease (Table 8b). In Clarksville the cross
Cobra x Ceres was the earliest with -11.58 percent increase
followed by LD-9430 x Ceres with a decrease of -1.42 percent

(Table 9b).

Branches per Plant

The Cross Cobra x Winfield had the highest number of
branches per plant with 6.31 percent increase over the mid-
parent, followed by Winfield x Ceres, with a 3.98 percent
increase across both locations (Table 7b).

In East Lansing, Cobra x Winfield, again had the highest
number of branches with an increase of 4.50 percent. This was
followed by LD-9430 x Winfield with an increase of 3.09
percent (Table 8b).

In Clarksville, the same cross Cobra x Winfield

 

 

59
maintained its first. position, with. the Ihighest. percent
increase of 8.07 percent followed by Winfield x Ceres with a

5.73 percent increase (Table 9b).

Pods per Plant
The Cross LD-9430 x Ceres had the highest number of pods

per plant with 27.90 percent increase, followed by
Cascade x CC-4 with a 27.34 percent increase across both
locations (Table 7b).

,In East Lansing, LD-9430 x Ceres had the highest number
of pods per plant with a 27.47 percent increase, followed by
LD-9430 x CC-4 which.had the highest number of pods per plant,
with an increase of 25.04 percent (Table 8b). In Clarksville,
Cascade x CC-4 had the highest number of pods with an increase
of 34.59 percent, was followed by LD-9430 x Ceres with an

increase of 28.35 percent (Table 9b).

Seeds per Pod

The cross Cascade x CC-4 had the largest number of seeds
per pod with an increase of 27.41 percent. This was followed
by LD-9430 x Ceres with a 26.67 percent increase combined for
both locations (Table 7b).

In East Lansing, LD-9430 x Ceres had the largest number
of seeds per pod with an increase of 27.41 percent followed
by LD-9430 x CC-4 with a 22.30 percent increase (Table 8b). In

Clarksville, the cross Cascade x CC-4 was the largest with a

 

60
37.05 percent increase followed by LD-9430 x Ceres with an

increase of 25.91 percent (Table 9b).

One Hundred Seed Weight
The cross Winfiehi x CC-4 exhibited the highest one

hundred seed weight with an increase of 8.86 percent followed
by Ceres x CC-4 with a 8.11 percent increase across both
locations (Table 7b).

In East Lansing, the cross Winfield x CC-4 had the
largest percent increase of 11.39, followed by Winfield x
Ceres, with an increase of 7.50 percent (Table 8b). In
Clarksville, Ceres x CC-4 had the highest mean seed weight,
value with a 6.67 percent increase followed by Winfield x CC-

4 with a 6.33 percent increase (Table 9b).

Yield (kglha)

When calculated across locations Cascade x CC—4 had the
highest yield increase of 33.46 percent. It was followed by
LD-9430 x Ceres with a 30.69 percent increase (Table 7b).

LD-9430 x Ceres showed the greatest heterosis with a
31.18 percent yield increase at East Lansing, was followed by
LD-9430 x CC-4 with a 29.12 percent increase in yield (Table
8b). In Clarksville, Cascade x CC-4 had the best yielding
cross, with a 44.94 percent increase. It was followed by LD-
9430 x Ceres with an increase of 30.13 percent (Table 9b).

Heterosis ranged from 12.0 to 33.5 percent across locations.

 

61

Percent Oil Content

When the heterosis was calculated across both locations,
the cross Winfield x Ceres had the highest oil content with a
8.81 percent increase, followed by Ceres x CC-4 with a 5.90
percent increase. At East Lansing, Winfield x Ceres had a
10.47 percent increase in oil content, followed by Ceres x CC—
4 with a 8.07 percent increase (Table 8b). In Clarksville, the
cross Winfield x Ceres showed the highest oil content with a
7.18 percent increase followed by LD-9430 x Cobra with a 5.64

percent increase (Table 9b).

4.2 Cultivar Analysis

The first three PCs of the cultivar data accounted for
the largest amount of variation, which was 76 percent of the
total variance for crown diameter and yield components among
the cultivar means; .PC,, PC2 and PC3, accounted for 32 percent,
23 percent and 21 percent of the variance, respectively (Table
11). No significant correlation between the 12 traits of the
thirty canola cultivars occurred except that between pods per

plant and seed per pod (r = 0.43) (Table 12).

4.3 anetic Distance

The genetic distance between the six parents used in the
crossing program was based upon their morphological and yield
component traits. The distance between the parents ranged from

1.104 to 7.497 (Table 10). The minimum distance was between

 

 

 

62
Cobra and Winfield which belong to the second cluster. In the
first cluster comprised of LD-9430 and Cascade with genetic
distance of 2.128. In the third cluster the distance between
the two parents was 2.501. This was compared with a maximum
distance of 7.497 between the parents of two different
clusters. One parent, LD-9430, belonged to cluster number one,
whereas the other parent, Ceres, belonged to cluster number

three.

 

63

Table: 7.a IPercent.mid-parent.heterosis for 15 canola crosses
combined for both locations,
East Lansing and Clarksville, MI 1991

 

 

Crosses DE* Cwn Lf L Lf L Lf L Lf Wd DOF

Dia 1.nd 2.nd full
1.LD-9430xCasc. -2.70 3.81 -0.27 0.00 1.95 1.92 -1.42
2.LD-9430xCobra. -2.56 3.05 -2.33 -1.00 -0.61 1.22 -1.61
3.LD-9430xWin. -2.56 -1.42 -3.60 0.74 -0.39 1.49 -1.36
4.LD-9430xCeres 2.44 4.73 5.76 0.00 -7.61 1.81 0.07
5.LD-9430xCC-4 -5.26 1.15 4.63 -1.45 1.12 -3.01 -1.48
6.Casc.xCobra. 0.00 3.21 -1.42 -1.84 0.60 0.27 -1.33
7.Casc.xWinf. 5.88 -2.14 -1.12 -1.27 0.97 0.45 -0.73
8.Casc.xCeres 5.56 0.51 2.96 -1.06 1.18 2.54 -0.03
9.Casc.xCC-4 -3.03 -0.77 -0.69 -0.89 -0.76 0.68 -1.13
10.CobraxWinf. 0.00 -1.88 -1.35 -1.29 0.45 5.78 -1.76
11.CobraxCeres 0.00 3.90 -1.37 0.27 1.39 2.88 -0.10
12.CobraxCC-4 -2.86 -1.30 -0.36 -2.71 0.50 0.34 -1.33
13.Winf.xCeres 5.26 3.18 -1.41 0.95 1.26 0.70 -0.41
14.Winf.xCC-4 -2.86 -2.90 -2.19 0.13 —1.02 -4.52 -1.29
15.CerestC-4 8.11 2.50 -1.27 -0.45 1.07 3.43 0.45
Mean 0.36 1.04 0.26 -O.62 0.01 1.06 -0.90

 

* Abbreviations & units of the characters are explained in

table 3,

p.

42

 

 

64

 

 

Table 7.b Percent mid-parent heterosis for seven traits
in 15 canola crosses combined for both locations,
East Lansing and Clarksville, MI 1991

Crosses DTM* Br/ Pd/ Sd/ 100 Yld Oil
Plt Plt Pod Sd/Wt %
1.LD-9430xCasc. 0.42 1.54 10.18 8.21 -2.04 12.66 4.72
2.LD-9430XCobra -0.55 2.18 18.72 16.78 0.00 21.33 5.09
3.LD-9430XWin. -0.47 3.51 14.14 12.20 -3.37 16.27 3.96
4.LD-9430xCeres -0.66 -0.07 27.90 26.67 -5.88 30.69 1.45
5.LD-9430XCC-4 -0.87 -1.64 23.08 20.88 -4.76 26.85 1.40
6.Casc.xCobra -0.03 -0.65 16.66 13.18 4.00 24.32 -0.09
7.Casc.xWinf. 0.45 -1.13 11.13 11.72 -1.08 13.70 2.18
8.Casc.xCeres -0.42 1.80 22.91 20.66 -5.62 25.28 2.92
9.Casc.xCC-4 -0.23 -2.36 27.34 27.41 0.00 33.46 -1.42
10.CobraxWinf. -0.58 6.31 9.41 11.52 -7.69 13.27 -1.82
11.CobraxCeres -5.95 2.37 21.74 19.04 -3.45 24.10 4.89
12.CobraxCC-4 -0.87 -1.91 22.63 20.12 4.65 23.56 4.07
13.Winf.xCeres -0.86 3.98 15.14 20.26 5.00 15.41 8.81
14.Winf.xCC-4 0.00 0.00 21.38 16.79 8.86 21.70 5.34
15.CerestC-4 0.09 -0.59 13.81 11.14 8.11 12.01 5.90
Mean -0.70 0.89 18.41 17.11 -0.22 20.97 3.16

 

* Abbreviations &
table 3,

p. 42

units of the

characters are explained in

 

 

 

65

 

 

 

Table 8.a Percent mid-parent heterosis for seven traits
in 15 canola crosses grown in East Lansing, MI 1991
Crosses DE* Cwn Lf L Lf L Lf L Lf Wd DOF
Dia 1.nd 2.nd full

1.LD-9430xCasc. 0.00 1.88 -2.45 -1.18 3.52 7.40 -1.27
2.LD-9430XCobra -5.88 2.89 -3.47 -3.62 -0.37 10.37 -0.56
3.LD-9430XWin. 5.88 -4.98 -4.62 0.33 -0.93 2.84 0.55
4.LD-9430xCeres 0.00 2.46 10.76 0.21 -11.70 12.46 0.07
5.LD-9430XCC-4 0.00 0.21 6.74 -1.49 0.33 1.99 -1.48
6.Casc.xCobra -6.67 0.45 0.67 -0.55 0.94 5.85 -1.29
7.Casc.xWinf. 6.67 -2.42 -1.10 -0.74 1.28 4.01 -0.43
8.Casc.xCeres 0.00 32.61 1.52 -2.25 1.82 12.46 0.21
9.Casc.xCC-4 0.00 1.42 -0.40 -1.44 -1.41 13.15 -0.94
10.CobraXWinf. 0.00 -1.87 -0.41 -1.52 1.14 15.19 -1.69
11.CobraxCeres -5.88 4.44 -0.89 1.28 2.06 10.33 0.35
12.CobraxCC-4 -6.67 -1.17 -0.67 -2.21 -0.83 10.44 -1.01
13.Winf.xCeres -5.88 2.34 -1.30 1.15 1.57 9.49 -0.48
14.Winf.xCC-4 -6.67 -3.09 -1.10 1.34 -1.95 4.20 -1.91
15.CereSXCC-4 8.11 4.06 -5.79 -l.79 1.46 11.43 0.98
Mean -1.13 2.62 -0.17 -0.83 -0.26 8.77 -0.59

* Abbreviations & units of the characters are explained in

table 3, p. 42

 

 

Table 8.b Percent mid-parent heterosis for seven traits

66

in 15 canola crosses grown in East Lansing, MI 1991

 

 

 

Crosses DTM* Br/ Pd/ Sd/ 100 Yld Oil
Plt Plt Pod Sd/Wt %
1.LD-9430XCasc. -0.40 1.10 10.16 8.33 2.04 12.51 4.75
2.LD-9430XCobra 0.06 1.06 18.26 17.05 4.17 20.85 4.55
3.LD-9430XWin. -0.56 3.09 16.63 15.03 1.12 18.59 3.38
4.LD-9430XCeres 0.11 -1.06 27.47 27.41 -3.53 31.18 4.47
5.LD-9430XCC-4 -1.36 0.07 25.04 22.30 -4.76 29.12 1.43
6.CaSC.XCObra 0.11 0.07 18.40 14.49 4.00 16.61 0.19
7.Casc.xWinf. 0.28 0.07 12.01 12.33 1.08 15.76 2.46
8.Casc.xCeres -0.51 0.07 24.93 20.52 -3.37 25.60 4.49
9.Casc.xCC-4 -0.40 -1.16 20.12 18.12 0.00 23.68 -1.09
10.CobraxWinf. -0.84 4.50 5.92 10.35 -3.30 11.85 -1.97
11.CobraxCeres -0.17 0.00 22.13 19.32 1.15 24.57 6.54
12.CobraxCC-4 -0.85 -1.33 21.02 18.29 6.98 22.26 4.13
13.Winf.xCeres -0.79 2.32 15.13 11.56 7.50 13.79 10.47
14.Winf.xCC-4 -0.11 1.10 23.01 16.79 11.39 21.03 6.26
15.CereSXCC-4 0.00 1.19 12.98 11.98 6.67 11.49 8.07
Mean -0.36 0.74 18.21 16.26 2.08 19.93 3.87

* Abbreviations & units of the characters are explained in

table 3, 42

p.

 

 

Table 9.a Percent mid-parent heterosis for seven traits

67

 

 

 

 

in 15 canola crosses grown in Clarksville, MI 1991
Crosses DE* Cwn Lf L Lf L Lf L Lf Wd DOF
Dia 1.nd 2.nd full

1.LD-9430XCaSC. -4.76 6.62 2.42 1.65 -0.72 3.07 -1.56
2.LD-9430XCobra 0.00 3.50 -0.36 3.55 -1.04 1.00 -2.60
3.LD-9430xWin. -9.09 4.19 -2.17 1.42 0.42 3.95 -3.22
4.LD-9430XCeres 4.35 7.38 -2.21 -0.27 0.67 1.64 0.07
5.LD-9430XCC-4 -9.09 2.60 1.62 -1.19 2.50 4.42 -1.49
6.Casc.XCobra 5.26 7.85 -5.34 -4.42 -0.19 2.19 -1.37
7.Casc.xWinf. 5.26 -1.74 -1.15 -2.12 0.43 0.40 -1.03
8.Casc.xCeres 10.00 -5.94 5.84 0.93 -0.18 3.51 -0.27
9.Casc.xCC-4 -5.26 -3.58 -1.23 0.00 0.43 -0.73 -1.32
10.CobraxWinf. 0.00 -1.70 -2.94 -0.72 -1.00 1.11 -1.82
11.CobraxCeres 4.76 1.74 -2.21 -2.07 0.00 9.29 -0.67
12.CobraxCC-4 0.00 -1.68 0.41 -3.85 3.17 4.23 -1.57
13.Winf.xCeres 4.76 3.44 -1.62 0.66 0.62 3.20 -0.33
14.Winf.XCC-4 0.00 -2.27 -4.12 -2.25 0.87 -1.72 -0.69
15.CerestC-4 4.76 -1.29 6.65 2.36 0.25 7.99 -0.20
Mean 0.73 1.27 -0.43 -0.42 0.42 2.90 -1.20

* Abbreviations & units of the characters are explained in

table 3, 42

p.

68

Table 9.b Percent mid-parent heterosis for seven traits

 

 

in 15 canola crosses grown in Clarksville, MI 1991
Crosses DTM* Br/ Pd/ Sd/ 100 Yld Oil
Plt Plt Pod Sd/Wt %

1.LD-9430XCaSC. 1.21 2.17 10.19 8.11 -4.08 12.83 4.73

2.LD-9430XCObra -1.13 3.30 19.22 16.51 -2.08 21.88 5.64

3.LD-9430XWin. -0.38 4.10 11.49 9.31 -7.87 13.61 4.54

4.LD-9430xCeres -1.42 1.06 28-35 25.91 -5.88 30.13 -1.25

5.LD-9430xCC-4 -0.39 -3.23 21.06 19.39 -2.38 24.23 1.35

6.CaSC.XCObra -0.17 -1.16 14.76 11.76 4.00 14.97 -0.33

7.Casc.xWinf. 0.61 -2.20 10.18 11.10 -5.38 11.29 1.88

8.Casc.xCeres -0.34 3.62 20.77 20.83 -7.87 24.89 1.37

9.Casc.xCC-4 -0.06 -3.53 34.59 37.05 0.00 44.94 -1.76

10.CobraxWinf. -0.33 8.07 12.70 12.66 -12.09 14.93 -1.69
11.CobraxCeres -11.58 5.11 21.33 18.73 -8.05 23.54 3.30
12.CobraxCC-4 -0.89 -2.49 24.05 22.02 2.33 25.09 4.03
13.Winf.xCeres -0.93 5.73 15.16 16.00 0.00 17.35 7.18
14.Winf.xCC-4 0.11 -1.15 19.70 16.79 6.33 22.49 4.42
15.CerestC-4 0.17 -2.49 14.65 10.17 6.67 12.63 3.76
Mean -1.03 1.13 18.55 17.09 2.42 20.98 2.48

 

* Abbreviations & units of the characters are explained in

table 3, p. 42

 

69

Table 10. Genetic distance between six parents
calculated based on their
morphological and yield components traits*

 

Cluster No.1 Cluster No.2 Cluster No.3

LD-9430 Cascade Cobra Winfield Ceres CC-4

 

LD-9430 - - - - _ -
Cascade 2.128 - - — - -
Cobra 3.576 2.402 - - - -
Winfield 4.266 2.927 1.104 - - -
Ceres 7.497 6.180 4.592 4.376 - -
CC-4 5.610 4.785 3.341 3.545 2.501 -

 

* Five traits including four yield components and one
morphological trait (crown diameter) were used to make
three clusters.

 

 

70

Table 11. Eigenvalues of the five PC axes from PC analysis
of thirty canola cultivars, with the proportion
of total variance accounted for by each axis

 

 

Principal Eigenvalue Proportion Cumulative
Component of Variance Variance

1 1.609 0.322 0.322

2 1.169 0.234 0.555

3 1.056 0.211 0.767

4 0.774 0.154 0.922

5 0.390 0.078 1.000

 

 

 

 

Table 12. Correlation coefficients between five traits
for thirty canola cultivars

 

 

Br/ Pd/ Sd/ 100
Plt* Plt Pod Sd/Wt
Cwn -0.040 0.184 -0.002 0.119
Dia
Br/ _ 0.290 -0.065 0.063
Plt
Pd/ _ _ 0.435 -0.213
Plt
Sd/ _ _ _ -0.190
Pod

 

* Abbreviations & units of the characters are explained in
table 3, p. 42

 

 

DISCUSSION

The heterotic response was calculated as a percent of the
mid-parent performance for yield, yield components and other
morphological traits. The:data reveals that all of the fifteen
hybrids exceeded the mid-parent value for seed yield at both
locations during 1991. Twelve out of fifteen crosses showed
higher performance than their mid-parent value for oil
content. Likewise, all the 15 crosses exhibited higher
performance compared to their mid-parent value for pods
per plant and seeds per pod.

All of the fifteen crosses exceeded.the high.parent value
for seed yield which is an important trait and major focus of
the study. High parent heterosis ranged from 3-23 percent
compared.tO'mid-parent.heterosis which showed.a:range of 13-33
percent. Since high parent heterosis was demonstrated,
however, the value of the low parent does not appear to
negatively effect dominance genetic variance contributing to
yield. The low parent might be used for combining some other,
important qualitative traits like disease resistance and oil
content with the yield.

The data agrees with findings of other researchers.
Paterniani (1973) observed that 61 out of 63, or 97 percent of
the crosses exceeded their mid-parental value for yield in

maize. Upadhaya (1967) observed that in 28 barley crosses, 14

71

 

 

 

72
exceeded the corresponding mid-parent value significantly for
yield, 7 out of 28 exceeded the mid-parent for kernel weight,
11 for kernels per head and 5 out of 28 exceeded mid-parent
for heads per plant.

The correlation between heterotic effects for seed yield
and parental distance as estimated by the Euclidean distance
method was positive and highly significant (r = 0.70) for
combined analysis at both individual and combined locations.
The correlation coefficient r = 0.86 was estimated for the
East Lansing location and r = 0.60, for the Clarksville
location. This positive and significant correlation supports
the strong association of seed yield heterosis with genetic
distance between. parents. These data agree ‘with earlier
reports by Khanna et al. (1977) in tomato, Moll et al. (1962)
in maize and Shamsuddin (1985) in wheat. In a study of genetic
diversity in relation to heterosis and combining ability in
spring wheat, Shamsuddin (1985) observed a significant
positive correlation for genetic diversity between the parents
with heterosis for grain yield of hybrids (r = 0.45).

Estimated heterosis for the number of pods per plant and
seeds per pod was also highly significant and positively
correlated with genetic distance. The correlation for number
of pods per plant calculated for both locations was r = 0.79,
which was positive and highly significant for both East
Lansing at r = 0.71 and Clarksville at r = 0.63. Likewise, the

correlation for number of seeds per pod for combined analysis

73

at both locations highly significant and.positive atr= 0.80.
It was also positive and highly significant for each location
separately; Values for East Lansing were r = 0.86 and r = 0.61
for Clarksville. From individual and combined locations it is
concluded that both pods per plant and seeds per pod have a
strong association with the genetic distance. This association
is therefore, an indication that there was a positive
association between seed yield as deduced from the number of
pods per plant, seeds per pod and the genetic distance. This
supports the findings of other researchers in several crops
(Moll et al. 1962, Grant et al. 1985, Hutcheson et al. 1981,
Khanna et al. 1977) in maize, rapeseed and tomato
respectively. In a similar study in dry beans Ghaderi et al.
(1984) found a positive correlation between genetic distance
and number of pods per plant, seeds per pod, and seed yield.
However, they did not find any correlation for seed yield and
number of seeds per pod with the genetic distance in faba
bean. They presumed that this could be due to the complete
additivity of genes and the common genes shared by the
parents.

Relatively little variation occurred between the crosses
for traits like crown diameter, leaf length, leaf width and
branches per plant. The data for crown diameter deviated from
mid-point values for all the fifteen crosses from -0.11 to
0.14, for leaf length -0.95 to 0.23, for leaf width it ranged

from -0.27 to 0.38, and —0.17 to 0.29 for branches per plant.

 

 

74
No significant correlation occurred between mideparent
heterosis for these. traits and the estimates of genetic
distance. The data suggests that the observed variation was
due to environmental effects rather than genetic effects.
Falconer (1982) showed the.occurrence.of heterosis (H) in
the F3 generation for a given quantitatively inherited trait

depends on the following:

H = E dy2

where "d" is the deviation of the heterozygote from the mean
of the homozygote parents and "y" is the difference of gene
frequency between the two intermated populations. The positive
association between. the above-mentioned characters under the
study could be attributed to the following:

Two possible factors could be contributing to a positive
association between the genetic distance and heterosis in a
set of crosses. 1) Either dominance has to exist, which should
increase in magnitude and proportion directly with an increase
in distance of the parental population and/or 2) the parental
populations should be different in gene frequencies which are
controlling these characters and this difference should also
have to increase as the distance among the parents becomes
greater. For characters such as yield, numbers of pods, and
seeds per pod which are inherited quantitatively in

canola, the heterotic effect depends on the number of

 

 

75
contrasting loci and their dominance effects.

As discussed earlier, almost all of the morphological and
phenological characters except yield and yield components
showed no correlation with the genetic distance estimates.
This could be explained by the complete additivity of genes
controlling these characters, i.e. (d = 0) and/or the parental
populations involved in crossing shared the same genes and
thus lacked adequate genetic variation for these traits

2:= 0). A second possible reason for lack of association

(Y
could be due to an unpredictable and random expression of the
heterosis. The data indicates that certain crosses have zero
heterotic effects, perhaps because of the cancellation of
positive and negative dominance effects, which occurs in both
directions. Moreover, the interallelic interactions or
epistasis may also produce certain kinds of specific effects
which may lead to the lack of association between the
heterosis and genetic distance.

A lack of genetic variation in the morphological traits
in the parental material used for this study is apparent. For
a successful hybridization program some long term strategies
are needed to maximize the genetic diversity in the material.
New germplasm could be obtained from geographically diverse
origins and international germplasm centers. This would
enhance the genetic diversity and broaden the genetic base of

the material, which should lead towards the maximum heterosis

and ultimately improvement in both yield and specific useful

76
morphological traits.

For phenological traits such as days to flower and days
to maturity, there was no significant correlation occurred
between heterosis and genetic distance. Thus, is based on
combined and separate analysis for both locations, almost all
of the crosses had negative heterosis. These results are also
consistent with those of Khanna et al. (1977) who observed
maturity as an earliness character with negative heterosis in
a similar study conducted with tomatoes. This kind of
heterosis implies dominance of genes favoring early flowering
which would give a negative "d". Based on this assumption, it
may be concluded that as the parents became more distantly
related, their hybrid's performance moved towards the dominant
early flowering parents due to dominance genetic variance.

Since the average seed yield of canola is rather low,
selection for larger seed size could be an important
consideration for increasing the seed yield. The task of a
canola breeder should be to improve the genetic capacity for
maximum yield while maintaining a desirable oil and meal
quality. The major emphasis in a rapeseed breeding program
should be on improving the agronomic performance of the crop
to get the maximum yield, which our data indicates, could be
achieved by making hybrids. In various hybridization programs,
high levels of heterosis for seed yield, both in spring and
winter forms of B. napus have been found (Sernyk & Stefansson,

1983 ; Lefort-Buson and Dattee, 1982) . The mid-parent heterosis

 

 

77

was up to 40 percent for yield in summer rape and 60 to 70
percent for winter forms of Brassica. The average heterosis
found was 10 to 20 percent, a level which would justify the
efforts directed towards the production of hybrid cultivars.

In order to maximize genetic diversity, the crossability
of rapeseed and mustard could be important for the breeder
because of the excellent possibilities to transfer the
economically important traits from the related species to the
commercial crop. Another major breeding objective for canola
is to obtain higher levels of oil with improved nutritional
value for human consumption and industrial use. Brassica oil
with low erucic acid is good for human consumption, whereas
high erucic acid oil is used for industrial purposes. High
erucic acid, which is fractioned from the oil, is first
converted to an amide for use in plastic products as a slip
agent. This is done to avoid any sticking problem of plastic
products with each other. High yielding cultivars developed
through hybridization are required to serve both edible and
industrial markets. Our data showed a small increase in the
percent oil content for the crosses ranging from -1.82 to 8.81
percent. This could be due to limited genetic variation in the
parental material for this trait. However, considering the
importance of oil content in this crop, the parents with a
wider genetic range for this trait could be selected in future
hybridization to improve oil content.

In this study, as in the majority of other studies, the

78
inter-cluster heterosis was shown to be greater than the
intra-cluster heterosis. These results support the findings of
other researchers, i.e. Khanna et al. (1977), Moll et al.
(1962) in other crops like tomatoes and maize.

Since the association between heterosis and genetic
distance observed in this study for a series of morphological
traits, yield, and yield components was variable, one should
not always expect a significant correlation between genetic
distance and heterosis. Such situations could be expected by
considering the. complex: genetic. nature of expression. of
heterosis for a given character.

Since pods per plant and seeds per pod showed a highly
significant positive correlation with yieLd in this study,
more importance should be given to these components when
selecting the parents for hybridization, provided adequate
variability exists for both traits.

Calculating the genetic distance between the parents
based upon morphological traits, yield, and yield components
is one approach to determine the relationship between the
parental material prior to hybridization. However, genetic
distance could also be estimated.by using the allele frequency
data from simply inherited polymorphic traits. Suitable data
in this regard could be derived from isozyme, RFLP and RAPD
markers. A general approach to calculating genetic distance
from marker loci data is to determine the frequency of

different marker loci alleles in the parental population and

 

79
compare these marker loci profiles using the same distance
equation.

Some researchers using isozymes (Hunter & Kannenberg,
1971; Hadjinov et al. 1982) and RFLPs (Godshalk et al., 1990)
in maize have found insignificant correlation between hybrid
yield and parental distance. Frei et al. (1986) found the
marker loci distance between maize parents to be predictive of
hybrid performance only when there was a known pedigree
relationship between the parents. Despite some limitations,
the marker loci comparison could also give useful additional
insight into the genetic relationship between the parental
material.

The highly significant correlation of genetic distance
with yield demonstrated that the greater the genetic distance
between the parents, the higher would be the yield potential.
Thus, multivariate analysis and clustering techniques play a
very crucial role in the selection of parents. Based on the
results of this study, it could be recommended that
multivariate statistics and clustering procedures could be
adopted to determine the genetic distance among the cultivars.
This provides the breeder with objective information to help
narrow down the initial parental selection for a successful
hybridization program. While measuring the genetic distance
between the cultivars only those traits should be included in
the clustering procedure, which are useful in forming distinct

clusters. In this study, yield components like branches per

 

80

plant, pods per plant, seeds per pod and one hundred seed
weight are the examples of the traits which placed the
cultivars in distinct clusters prior to parental selection.
Other traits did not have any significant impact in cluster
formation. This suggests that yield components in particular
are very important traits to be utilized in clustering
procedure if yield is the ultimate objective within the
breeding program.

Up to 33 percent heterosis for yield in this study
justifies the effort to develop a successful system like
cytoplasmic male sterility for production of hybrid cultivars
in such an important crop. Several attempts in this regard are
still underway and some male sterility, inducing cytoplasms
for .B. napus have already been found. For 'example the
previously described a CMS cytoplasm could be used to
prevent self pollination of the female parent of the cross,
and the restorer genes could be incorporated to restore the
male fertility of the F, hybrid. In other systems self
incompatibility could also be investigated, since some self
incompatible plants have been found in B. napus. Thompson
(1983) proposed a scheme for the production of a self
compatible three way cross hybrid of B. napus utilizing self
incompatibility. A major drawback of this scheme is the
difficulty of producing commercial quantities of selfed seed
on the self incompatible parent lines. Despite these

difficulties, efforts could still be made to find efficient

 

81
large-scale methods of producing the hybrid seed in order to
minimize the cost of the hybrid seed to the farmer and thus

permit the exploitation of heterosis demonstrated in the crop.

APPENDIX

ANALYSIS OF VARIANCE TABLES

 

82

Table A1. Analysis of variance for 14 traits in 21
entries crosses of canola in East Lansing, MI 1991

 

 

Name of Trait ms
Days of emergence ns
Crown diameter **
Leaf length (lst node) **
leaf length (2nd node) **
Leaf length full (up to tip) **
Leaf width **
Days to flower **
Days to maturity **
Branches/plant **
Pods/plant **
Seeds/pod **
100 seed weight ns
Yield in Kg/ha **
oil content (%) *

 

*, ** Indicate significance at 0.05 and 0.01 level of
probability respectively according to the F-test

 

 

83

Table A2. Analysis of variance for 14 traits in 21
entries/crosses of canola in Clarksville, MI 1991

 

 

Name of Trait ms
Days of emergence ns
Crown diameter **
Leaf length (lst node) **
leaf length (2nd node) **
Leaf length full (up to tip) **
Leaf width **
Days to flower **
Days to maturity **
Branches/plant **
Pods/plant **
Seeds/pod **
100 seed weight ns
Yield in Kg/ha **
oil content (%) *

 

*, ** Indicate significance at 0.05 and 0.01 level of
probability respectively according to the F-test

Table A3. Analysis of variance for 14 traits in 21
entries/crosses of canola combined for both
locations, East Lansing and Clarksville, MI 1991

 

 

Name of Trait ms
Days of emergence ns
Crown diameter **
Leaf length (lst node) **
leaf length (2nd node) **
Leaf length full (up to tip) **
Leaf width **
Days to flower **
Days to maturity **
Branches/plant **
Pods/plant **
Seeds/pod **
100 seed weight ns
Yield in Kg/ha **
oil content (%) *

 

*, ** Indicate significance at 0.05 and 0.01 level of
probability respectively according to the F-test

List of References

Adams, M.W. & Duarte R. 1961. The nature of heterosis for a
complex trait in a field bean cross. Crop Sci. 1: 380

Ariyo, O.J. 1987. Multivariate analysis and the choice of
parents for hybridization in Okra (Abelmoschus esculentus
L. Moench) .

Arunchalam, V.A. Bandyppadhyay, S.N. Nigam & R.W. Gibson 1984.
Heterosis in relation to genetic divergence and specific
combining ability in groundnut (Arachis hypogaea L.)
Euphytica 33:33-39

Ashri, A.. 1973. Divergence and evolution in the safflower
genus. Cartahamus L.) Final report, USDA P.L. 480
project Jerusalem.

Ashton, T. 1946. The use of heterosis in the production of
agricultural and horticultural crops. Imp. Bur. Plant
Breed. and Genetics, Schl. of Agr. Cambridge, England

Atchley, W.R., E.V. Nordheim, F.C. Gunsett, and P.L. Crump.
1982. Geometric and.probabilistic aspects of statistical
distance functions. Syst. Zool. 31:445-460.

Balasch, S., F. Nuez, G. Palomares, and J. Cuartero. 1984.
Multivariate analysis applied to tomato hybrid
production. Theor. Appl. Genet. 69:39-45.

Beal, W.J. 1880. Indian.corn. Rep. Michigan State Board.Agric.
19:279.

Bedigian D., C.A. Smyth, and J.R. Harlan. 1986. Patterns of
morphological variation in Sesamum indicum. Econ. Bot.
40:353-365.

Bhatt, G.M. . 1970. Multivariate analysis approach to selection
of parents for hybridization aiming at yield improvement
in self-pollinated crops. Aust. J. Agric. Res. 21:1-7.

Bhatt, G.M.. 1973. Comparison of various methods of selecting
parents for hybridization in common bread wheat (Triticum
aestivum L.). Aust. J. Agric. Sci. 24:457-464.

Brieger, F.G.. 1950. The genetic basis of heterosis in maize.
Genetics 35:420-444.

85

86

Brown, W.. 1967. Results of non-selective inbreeding in maize.
Zuechter 37:155-159.

Burton, G.W.. 1980. Utilization of hybrid vigor. p. 89-107. lg
D.R. Wood (ed.) Crop breeding. Am. Soc. Agron., Madison,
WI.

Buson, M.. 1980. Heterosis and genetic parameters in winter
rape (Brassica napus L.). Cruciferae Newsletter No. 5.
pp. 13-14.

Camussi, A.. 1979. Numerical taxonomy of Italian populations
of maize based on quantitative traits. Maydica 24:161-
174.

Chheda HR. Fatokun CA. 1982. Numerical Analysis of variation
patterns in Okra (Abelmoschus Asculentus L. Moench). Bot.
Gaz. Chicago. 143:253-261.

Cowen, N.M. and K.F. Frey. 1987a. Relationships between three
measures of genetic distance and breeding behavior in
oats (Avena sativa L.). Genome 29:97-106.

Cox, T.S., and J.P. Murphy. 1990. The effect of parental
divergence on F2 heterosis in winter wheat crosses.
Theor. Appl. Genet. 79:241-250.

Cress, C.E.. 1966. Heterosis of the hybrid related to gene
frequency differences between two populations. Genetics
53:269-274.

Crow, J.F.. 1952. Dominance and overdominance. Heterosis pp.
282- 297. Iowa State College Press. Ames, Iowa.

Dasgupta, T. and Das P.K. 1984. Multivariate analysis and
selection of parents for hybridization in black gram.
Philipp. Agric. 67:86-92.

Daun, J.K. 1983. The introduction of low erucic acid requested
variety into Canadian production, in high and low erucic
acid rapeseed oils. Kramer, J.K.G.; Sauer, F.D.; and
Pigden, W.J., (eds.) Toronto: Academic Press. 161-180.

Davenport, C.G. 1908. Degeneration, albinism, and inbreeding.
Science 28:454-455.

Dobzhansky, Th. 1941. Genetics and the origin of species (2nd
ed.) Columbia University Press, New York, 446 pp.

Duddley J.W., and R.L. Davies 1966. Preliminary groupings of
plant introductions of alfalfa (Medicago sativa L.)
heterosis studies. Crop Sci. 6:597-601.

87

Duvick, D.N.. 1984. Genetic diversity in major form crops on
the farm and in reserve. Econ. Bot. 38:161-178.

East, E.M. 1936. Heterosis. Genetics 21:375-397.

East & H.K. Hayes. 1912. Heterozygosis in evolution and plant
breeding. USDA Bur. Plant Ind. Bull. 243: 1-58

Falconer, D.S.. 1982. Heterosis pp. 230. ;p Introduction to
quantitative genetics. Longman London & New York.

Food and Agriculture Organization of the United Nations. 1987.
FAO Production Year Book. 161.

Foreign Agricultural Service. October 1989. World oilseed
situation and market highlights. USDA Circular Series FOP
10-89.

Frei, O.M., C.W. .Stuber & M.M. Goodman. 1986. Use of allozymes
as genetic markers for predicting performance in maize
single cross hybrids. Crop Sci. 26: 37-42.

Furgala, B., Noetzel, D.M. and Robinson, R.G. 1979.
Observations on the pollination of hybrid sunflowers.
Proc. 4th Int. Symp. on Pollination. Md. Agric. Exp.
Sta. Spec. Misc. Publ. 1: 45-48.

Ghaderi, A., M. Shishegar, A. Rezai, and B. Ehdaie. 1979.
Multivariate analysis of genetic diversity for yield and
its components in mung bean. J. Amer. Soc. Hort. Sci.
104:728-731.

Godshalk, E.B., M. Lee & K.R. Lamkey. 1990. Relationship of
restriction fragment length polymorphisms to single cross
hybrid performance of maize. Theor. Appl. Genet. 80: 273-
280.

Goodman, M.M., 1968. A measure of "overall variability" in
populations. Biometrics 24:189-192. Can. J. Genet. Cytol.
27:472-478.

Goodman, M.M.. 1972. Distance analysis.in biology. Syst. Zool.
21:174-186.

Goodman, M.M.. 1973. Genetic distances: Measuring
dissimilarity among populations. Yearb. Pys. Anthropol
17:1-38.

Grant, I., and W.D. Beversdorf. 1985. Heterosis and combining
ability estimates in spring oilseed rape (Brassica napus
L.) Can. J. Genet. Cytol. 27:472-478.

88

Green, M.M. & K.C. Green 1949. Crossing over between alleles
at the lozenge locus in Drosophila melanogaster. Proc.
Nat. Acad. Sci. 35: 586-591.

Hadjinov, M.I., V.S. Scherbak, N.I., Benko, V.P. Gusev, T.B.
Sukhorzhevskaya & L.P. Voronova 1982. Interrelationship
between isozymic diversity in combining ability in maize
lines. Maydica 27:135-149.

Haldane, J.B.S.. 1954. The statistics of evolution. Ip
Evolution as a process. pp. 109-121. Allen & Unwin,
London.

Hallauer, A.R., and J.B. Miranda, Fo. 1981. Quantitative
genetics in maize breeding. Iowa State University Press,
Ames.

Hayes, H.K. & I.J. Johnson. 1939. The breeding of improved
selfed lines of corn. J. Amer. Soc. Agron. 31: 710-724.

Haymen, B.I. 1957. Interaction, heterosis, and diallel cross.
Genetics 42:336-355.

Heidrich-Sobrinho, E., and A.R. Cordiero. 1975. Codominant
isoenzymic alleles as markers of genetic diversity
correlated with heterosis in maize (Zea mays L.). Theor.
Appl. Genet. 46:197-199.

Hillig, K.W., and A.F. Iezzoni. 1988. Multivariate analysis of
a sour cherry germplasm collection. J. Amer. Soc.
Hort. Sci. 113(6):928-934.

Holden, J.H.W.. 1984. The second ten years. p. 276-285. lp
J.H.W. Holden and J.T. Williams (eds.) Crop genetic
resources: Conservation and evaluation. Allen & tmwhy
Inc., Winchester, MA.

Hotelling, H.. 1954 Multivariate analysis. P 67-80 in O.
Kempthorne, T.A. Bancroft, J .W. Gowan, and J .L. Lush
(eds.). Statistics and. Mathematics in. Biology. Iowa
State University College Press, Ames.

Hull, F.H.. 1952. Recurrent selection and overdominance. p.
451- 473. _I_p J.W. Gowen (ed.) Heterosis. Iowa State
University Press, Ames.

Hunter, R.B. & L.W. Kannenberg. 1971. Isozyme characterization
of corn (Zea mays) inbred and its relationship to single
cross hybrid performance. Can. J. Gent. Cytol. 13: 649-
655.

89

Hussaini, S.H., M.M. Goodman, and D.H. Timothy. 1977.
Multivariate analysis and the geographical distribution
of the world collection of finger millet. Crop. Sci.
17:257-263.

Hutcheson, D.S., R.K. Downey, and S.J. Campbell. 1981.
Performance of a naturally occurring subspecies hybrids
in B. campestris L. var oleifera Metzg. Can. J. Plant
Sci. 61:895-900.

Hutchinson, J.B., V.G. Panse & G.K. Govande. 1938. Studies in
plant breeding technique IV.The inheritance of
agricultural characters in three inter-strain crosses of
cotton. Indian J. Agr. Sci. 8: 757-775.

Immer, F.R.. 1941. Relation between yielding ability and
homozygosis in barley crosses. J. Amer. Soc. Agron. 33:
200- 206.

Isleib, T.G., and J.C. Wynne. 1983. Heterosis in testcrosses
of 27 exotic peanut varieties. Crop Sci. 23:832-841.

Jacquard A.. 1974. Genealogies et distance entre populations.
pp. 23-40. In J.F. Crow and C.Denniston (eds). Genetic
distance. Plenum Press, New York.

Jinks, J.L. 1983. Biometrical genetics of heterosis. p. 1-46.
IQ R. Frankel (ed.) Heterosis. Springer Verlag, Berlin.

Johnson, I.J. & H.K. Hayes. 1940. The value in hybrid
combinations of inbred lines.of corn selected from single
crosses by the pedigree method.of breeding. J. Amer. Soc.
Agron. 32: 479-485.

Johnes, D. F.. 1917. Dominance of linked factors as a.means of
accounting for heterosis. Genetics 2: 446-479.

Khanna, K.R., and C.H. Misra. 1977. Divergence and heterosis
in Tomato. Sabrao Journal. 9(1):43-50.

Kinman, M. Lu & G.F. Sprague. 1945. Relation between number of
parental lines and theoretical performance of synthetic
varieties in corn. J. Amer. Soc. Agron. 37: 341-351.

Lamkey, K.R., and A. R. Hallauer. 1984. Comparison of maize
populations improved by recurrent selection. Maydica
29:357-374.

Lerner, I.M.. 1954. Genetit:Homeostasis. Edinburgh: Oliver and
Boyd. 134 pp.

90

Mahalanobis, P.C.. 1936. On the generalized distance in
statistics. Proc. Nat. Inst. Sci. India. 2:49-55.

Maluf, W.R., P.E. Ferreira, and J.E.C. Miranda. 1983. Genetic
diversity in tomatoes and its relationship with heterosis
for yield in F1 hybrids. Rev. Brasil Genet. 6:453-460.

Mather,K.. 1956. Response to selection. Cold Spring Harbor
Symposia on Quant. Biol. 30: 158-165. '

Mather, K. and J.L. Jinks. 1982. Biometrical genetics.
University Press, Cambridge.

Miller, F.R., and Kebede, Y. 1984. Genetic contributions to
yield.gain in Sorghum, 1950-1980 Pgs. 1-12 in: W.R. Fehr,
ed. Genetic contribution to yield grains of five major
crop plants. CSSA 7. AM. Soc. Agron., Madison,
Wisconsin.

Moll, R.H., J.H. Lonnquist, J. Velez Fortuno, and E.C.
Johnson. 1965. The relationship of heterosis and genetic
divergence in maize. Genetics 52:139-144.

Moll,R.H., W.S. Salhuana, and H. F. Robinson. 1962. Heterosis
and genetic diversity in variety crosses of maize. Crop
Sci. 2:197-198.

Mukherjee, B.K., and.B.C. Saha. 1982. An analysis of heterosis
and heterosis manifest for yield characteristics in
intervarietal crosses of maize (Zea mays L.). Egypt. J.
Genet. Cytol. 13:41-52.

Murty, B.RJ and. I.J. .Anand. 1965. Combining' ability' and
genetic diversity in some varieties of Linum
usitatissimum. Indian J. Genet. and Plant Breed. 26:21-
36.

Murty, B.R., and V. .Arunchalam ‘V. 1966. The nature of
divergence in relation to breeding system in some crop
plants. Indian J. Genet. A 26:188-198.

Narayan, R.K.J., and A.J. Macfield. 1976. Adaptive responses
and genetic divergence in a world collection of chickpea
(Cicer arietinum L.). Theor. Appl. Genet. 47:179-187.

National Archives and Records Administration. 1985. Food and
drugs. Code of federal regulations, Title 21:184.155.

Niehaus, M.H. & R.G. Pickett. 1966. Heterosis and combining
ability in a diallel cross in Sorghum vulgare. Crop Sci.
6: 33-36.

91

Paterniani, M E.. 1973. Recent studies on heterosis. p. 1-22.
L13 R. Moav (ed.) Agricultural genetics. John Wiley &
Sons, New York.

Patterson, F.1n &.M. J. Bitzer. 1966. Hybrid.wheat research.at
Purdue University. Purdue Univ. Agr. Exp. Sta. Res.
Progress Report 222.

Pawlisch, P.E. & A.H. Van Dijk. 1965. Forage and grain
production of four F1 barely hybrids and their parents.
Crop Sci. 5: 135-136.

Pearson, K.. 1926. On the coefficient of racial likeness.
Biometrika 18:105-117.

Peter K.V., B.Rai. 1978. Heterosis as a function of genetic
distance in tomato. Indian J, Genet. Plant Breed.
38:173-178.

Peters, J.P., and J.A. Martinelli. 1989. Hierarchical cluster
analysis as a tool to manage variation in germplasm
collections. Theor. Appl. Genet. 78:42-48.

Pfahler, P.L.. 1966. Heterosis and homeostasis in rye Secale
cereale, L. 1. Individual plant production of varieties
and intervarietal crosses. Crop Sci. 6: 339-340.

Prasad, S.K. & T.P. Singh. 1986. Heterosis in relation to
genetic divergence in maize (Zea mays L.). Euphytica
35:919-924.

Rajana, A., Gulshan Lal, and K.V. Peter. 1977. Heterozygote
advantage as a function of genetic divergence in tomato.
Indian J. Agric. Sci. 47(9):434-437.

Ramanujam, S.-, A.S. Tiwari, and R.B. Mehra. 1974. Genetic
divergence and hybrid performance in mung bean. Theoret.
Applied Genet. 45:211-214.

Rambaugh, M.D., and J.H. Lonnquist. 1959. Inbreeding
depression of diallel crosses of selected lines of corn.
Agron. J. 51:407-412.

Robertson , F . W. & E . C . Reeve . 1952 . Heterozygosity,
environmental variation and heterosis. Nature (London)
170: 296.

Santiago, Fonseca.& F.L. Peterson. 1968. Hybrid.vigor in seven
parent diallel cross in common winter wheat, (Triticum
aestivum, L.) Crop Sci. 8:85-88

92

Sernyk, J.L., and B.R. Stefansson. 1983. Heterosis in summer
rape (Brassica napus L.) Can. J. Plant Sci. 63:407-413.

Shamsuddin, A.K.M.. 1985. Genetic diversity in redation to
heterosis and combining ability in spring wheat. Theor.
Appl. Genet. 70:306-308.

Shiga, T.. 1976. Cytoplasmic male sterility and its
utilization in rapeseed, (Brassica napus L.) Jap. Agric.
Res. Quart. 10:178-182.

Shiga, T. and S. Baba. 1973. Cytoplasmic male sterility in
oilseed rape and its utilization to breeding. (Japanese
with Englilsh summary.) Jap. J. Breed. 23:187-197.

Shull, G.H.. 1908. The composition of a field of maize. Rep.
Am. Breeders Assoc. 4:296-301.

Shull, G. H.. 1952. Beginnings of the heterosis concept. p.
14-48. ;p_J.W. Gowen (ed.) Heterosis. Iowa State College
Press, Ames.

Sneath, P.H.A.. 1976. Some applications of numerical taxonomy
to plant breeding. Z. Pflanzaenzuecht. 76:19-45.

Sokal, R.R.. 1961. Distance as a measure of taxonomic
similarity. Syst. Zool. 10:70-79.

Spagnoletti Zeuli, P.L., C. De Pace, and E. Porceddu. 1985.
Variation in Durum wheat populations from different
geographical origins. Z. Pflanzenzuechtg. 94:177-191.

Spagnoletti Zeuli, P.L. and C.O. Qualset. 1987. Geographical
diversity for quantitative spike characters in a world
collection of durum wheat. Crop. Sci. 27:235-241.

Sprague, G.F. 1952. Early testing and.recurrent selection. pp.
400-417. Ip: J.W. Gowen, (ed.), heterosis. Iowa State
College Press, Ames, Iowa.

Sprague, G.F. 1966. Quantitative genetics in. plant
improvement. p. 315-354. 1p K.J. Frey (ed.) Plant
breeding. Iowa State University Press, Ames.

Suneson, C.A. 1962. Hybrid barely promises high Yields. Crop
Sci. 2: 410-412.

Thompson, K.F.. 1972. Cytoplasmic male sterility in oilseed
rape. Heredity 29:253-257.

93

Timothy, D.H.. 1963. Genetic diversity, heterosis and the use
of exotic stocks in maize in Columbia. Symp. Statist.
Genet. Pl. Breed., Raleigh, N. Carolina, 1961, 581-91.

Tolbert, D. M., C.O. Qualset, S. K. Jain, and J.C. Craddock.
1979. A diversity analysis of a world collection of
barley. Crop Sci. 19:789-794.

Wallace, B. and M. Vetukhiv. 1955. 'Adaptive organization of
the gene pools of Drosophila population'. Cold Spring
Harbor Symp. Quant. Biol., 20:303-309.

Wu, S.K.. 1939. The relationship between the origin of selfed
lines of corn and their value in hybrid combination. J.
Amer. Soc. Agron. 31: 131-140.

 

MICHIGQ

iii

A.

iiiiiiiiiiiii

find '3’- ...
> . .. ,

. . y . - ..3: