T

'I
V i :‘i
5 '7 $1135.};
. . I Iv *3 ‘“
, I» "'E 5.“? 1&2: §§£fig .3941.
; ' ' “Mimi ”aw-151"”, WWW,
>2 :35 fig. I":'Z:.1;iitl 1 {I *3?“
. 53'2’33551‘33 in 555.55% 5551 3:5
. I. 11's“??? '-' ‘eu'hun E5 5.I. ‘51} NYE: 4-”
‘ 0-.51‘5 . ‘5: 4 2:! 1?: v' fiifil
5 1' ’ 5 5:355:31" 31%;“: 555: 'i' 11%! .155: 4; 152"} ‘39; i5?
. , I w, n .4 w; ----
'5 %.5’§é$k§5;§5 ‘54; gflxfi'h‘k‘iifi 5"5'5‘}. 553*! {i g“: t; .55
. \‘ 4 '5 2 : 4 ‘ ' ' ,
5 5‘ ‘5, égfi‘i‘“ ‘5 (53? k}: ‘19.?35ty fig“ 13;» fisé giggfi .
? ‘0 4 -.I ‘ ‘i‘wi’ ‘ 17“ ’ g
.5 bigi { ifi'§!i§i§i ‘gfih’fifi 3'11: $1? 21?"!3
35.

r O 31,9 . 7

4:15: 5 lk55555155555§555555fi12 1135;55547

v z :’4& 5.. 51. , :iiéz’fikgé --I§:§:gk

1 4, mg: «53.. hi: '55535553555555‘45fiégg

3': h‘ 2 "x, is! w?“

an: $11.11.} ,x“ §‘}|:E'i mfg.“

'4 5 55:55 film an; ‘32.

12:55 gww ..
112. I.

5,155! “$55552 5‘ 10.55152?

"1533'

I. .
,'*’ $92"-

... _,
”$.44." n: A
. 4,..ud-‘g: .
v uni... I
d l'll‘flvv
3' “'11”...—
4.4"

 
 
  
 
 
  
   

  

. .4

<~ ".2.
. 4-
44

pal-1N
n«
4 '49-.
v4

   
    
   
 
  
 
  
   
     

‘4'3'.’
'. .4 4"» .4.
J... .4-
.

L4.

          
 
 
  

.54”
4"“
W4.“

input-J

'r,‘ 41
:1?»

   

939$ :Lli {Iii‘

5'3: 3 i?

‘5 '1! i“ “it“;
1
”if"

 

'. 4 _,
«:54. u

4",...-
.44 #-

    

4.447474%
.4 '4... J"

v"
14W
4..
we.
“or"

mg'%

on

4

143‘

..

“:Wwa'

' we» ' .
mg, 'I I u v, ‘ I!

4!...
.o-ov

'7'
44..

 
   
        
     

  
 
 
 

1'03:-

      
      

V1.7”
.4.
”44-:

4
d-I‘.
4:44:14
941'.

4!?»

.4
43:4“
4
.
3:41?-
.4.
'and:

 
  
 

  
 
  
   
  

5 5:21!" 5555551555555555'" .5155!" E1 3.215th? 1532*; iv“
v I 513'5‘355 5 . ‘ 9 "3' . I
5 “53.15“- 52:55 (1'55 5;. £5555. £553 5:11‘3535-mfifi 5:55“ 4
.515 2‘4) 5" ‘1‘; gfi

  
 
  
   

44'

I

11.3% 55115“.
I. 5.211%; . 13,453?

  

*v
w
r I’M-r
”fl”???

.31.”

35.55
fig?

4—.-

   
 

”2‘30”
M“
£23..."
md‘w‘
"‘f‘fii‘nu
”I‘M
71?;
«:4
r43
4
4“”
w‘
m'

figgfifi;
II
4
W4.
4

.33“

,..“4“- <
”.4 “a... ‘
rm 4-3;" . 4..

.00

4151.55:

.3; 5‘
5. 933145“;ng
5_

  
  

1"».
33%;.
$43?- 442. was
‘ “flush:-

I g-I‘E' i " giggz;

    
     
  

  

Oca-
M

   
   
    

.2 . i5 :52;
2‘! 1"h.$§' 645‘ ,
$3” 2? an 51%;?

’1 ‘ ‘33:“?
:4...“ 4.5555
rt 7! $3.2,”
g? 9 L‘éé‘: iégié .

.~ ME“:‘ .3, -
fl :5: 32%;:Qfl’ :‘ 9:}-

’§2 5‘ ‘

5% 55111 ‘1
55.2.

     

   

   

J :04
"Nag

u.

””1
4:33er
44..
2:33.34“
4:: 10013;;
. .
' .‘2‘

‘ - Qt‘!‘ 3.9;
: ' v’fi.‘ 13:1. 1.
I 441555511 "5555
' . E’ E55353; KI.
'1 1
£52355": 555‘; iii“
"5:

ma; . . ‘ ., 5 i .
£55.25 42% E I " 9 j .
.55335553312 35‘ 55 5‘
5-3“ wag it
5;: 344‘

4r. . .
£3.31; g2: ‘2
u. “in”: ‘5‘ ’tI-fi 1%
9t; Liz 531-51531 it“ ~ .
(4E. ‘

in
wt.” ”as?

v-
w n;-

    

.-
I
m
14
4‘
-

    

      

 
   
   
      
   
 

  
 
  

“'1”
£53354??-
'23.? N‘
’ ’ "mu-’-

4....

rant.»
'5'.”

       
 

figfiifii .

5 k 1 ‘ 4 S

3255‘ 7‘3
5

 

 
 
  
 

     
  
  

  

55 2?}???
5.555 5411 mm

’1:

.N3
.4».
m

. . 4‘
if: 55mi- Eu. .Ii 21:; LE; : .3: 93;? 4
-: I 5g 522' :4 23745 ‘5'
‘53 3‘5'355555 5‘1 ' 5.552% 452.5-

§Au :5: Iixtéfi =15;‘155353& liofifli‘t'izg 'L a gag. iii??? ifiij
3‘2!

        
   
 
  
 
  
  

   
   

     
      
 

f;

‘51:,LIM'3V

"-0

A.

$1“ stristég

    
  

        
    
  
  
   

 
 
 

       
      

  

    
  

     

         
  
  

 
 
 

  

          
  

 
      
    
   
 
 

  
   

 
  

“I 5533 = 3 1 ~
.55sz I ’~ . i I “2:;
£53. Irwin? 1155‘
M ‘ “‘5: 1515 Iii: $35 . ME?” 4 I...
¥ ‘1 .14 554?: F532; 4* :55; E‘ét‘ét; 5t“. W‘, ii." gilt?! in |1§§1|§§ii H :55 1:3 5 ,
kill§gi§11fifigfz 51“.}? 5.21.55555 Epgth 311345155 55.1%. {it 5 :§:: 15 .53. #:EE:- {3 If.“
555 *‘i 11454-5215 53 25555555“ ‘5‘ 5-5 '3‘ "'1‘" 555 =
.. I5._.t IIWFIII $53.5 . . I g. ._ - . 1 '5' $149519; 2:51 {‘5 I
. _ 1+2: 5 $53.4 Mm: ; ”5555““ _ " 13. 5% 55‘ W5 5'
- 556515; ' E'S‘ ii?“ 5355, b: I511“’iu 5'2 1*‘553552555 ' , “ '5 55551.1: 55%;. $5555
:1 in: Win 5‘ 51:55 115554115: <tg5~k1fi§5faa5fiw= at £2555 E‘ ‘2‘“5" ‘ 5f
E 5' ‘- . "z '5? - SS .38: ”$15.55: 55it‘555{§95£t 151‘? ’v" ”£555”? £515 5‘?
_.; ,4, Z 1, 5‘55 amp; .5333; 1:51;}? 55555555555555.5553;55555555555555;I Y 91:3. W 132““;
:~. I 5:4. . E}: 55““ 5.55.“ ‘ ‘55.“:‘2, .4: “:73“! 56..
4 :5“ I III ’EIJIIII‘II‘I. 5523.511: 5.5"5555555‘ 15:; 2;: i3; " I“ »I
' 55.155511: “5?? 5. * 55555.52 5wa153 2255555" “55555“
_ If g1:} ‘Is’IiIIII ‘ ‘54" “E 9- 5% ‘51253555. ,EIE‘I’Ifi‘iifgi’i'siéfik 55.551151? I551
t c§§.¢'. 1'5thqu 11%. 1.133%?31’5-izi‘tffiilt-zfi‘g}, ,5
‘ 5 ‘Izéfiifisi‘té‘éigétgizégé .égérfigzsgifiéi‘g‘3212:5155 1,52%:
.‘tbk‘1‘t'.l“ I: - K. “5 ‘fl 5 II'V ' ‘5‘.
. I ‘.§§’,§%§§%§;g§§§ {9,1332% 5%:55555555355,‘.535554 41;».
‘iE-fiu" ‘Enéi‘t‘I.-: f"; ‘ 5555
i5f‘.r.?§5+fi55555551 5&3516 i I: ._ III? ,

HESIS

This is to certify that the

thesis entitled

GENETICS AND DEVELOPMENTAL INTER-RELATIONSHIPS

OF ARCHE’I‘ECTURAL TRAITS IN BEANS (PHASEOLUS VULGARIS L.)

presented by

RHEA HOWARD HARMSEN

has been accepted towards fulfillment
of the requirements for

Master's degeein Plant Breeding

and Genetics

%4 615%

Major professor
Date fi/Z/é/ygj

0-7639 MS U is an AflirmativeAction/Equal Opportunity Institution

Mickie; }‘« Sitate
University

_, , t,-_ I

KW“ if, fig Y

I

 

 

 

hViESI.] RETURNING MATERIALS:
Place in book drop to

LIBRARIES remove this checkout from

your record. FINES will

\—

 

 

 

be charged if book is
returned after the date
stamped below.

 

Fr!-

90 m1 cmcu u:

muswm'

{an

Q NW Sigma”!!!

 

 

PLEASE NOTE:

In all cases this material has been filmed in the best possible way from the available copy.
Problems encountered with thisdocument have been identified here with a check mark J .

—l
.

990899990!”

dd
‘0

12.
13.
14.

15.

Glossy photographs or pages __

Colored illustrations, paper or print__

Photographs with dark background __

Illustrations are poor copy __

Pages with black marks, not original copy __

Print shows through as there is text on both sides of page___
Indistinct, broken or small print on several pages—

Print exceeds margin requirements __

Tightly bound capy with print lost in spine _

Computer printout pages with indistinct print

 

 

Page(s) lacking when material received, and not available from school or
author.

Page(s) 7] seem to be missing in numbering only as text follows.

Two pages numbered . Text follows.

 

Curling and wrinkled pages

Other

University
Microfilms
International

*“‘A~*‘i‘-FL‘~(<~--"

/ /— «96.37

GENETICS AND DEVELOPMENTAL INTER—RELATIONSHIPS
OF ARCHITECTURAL TRAITS IN BEANS(EflA§§QLfl§ yuLQABLS L.)
WITH EMPHASIS ON BRANCHING

BY

Rhea Howard Harmsen

A THESIS

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

MASTER OF SCIENCE

Department of Crop and Soil Sciences

1983

ABSTRACT
GENETICS AND DEVELOPMENTAL INTER-RELATIONSHIPS

OF ARCHETECTURAL TRAITS IN BEANS (PHASEOLUS VULGARIS L.)

 

BY

Rhea Howard Harmsen

An investigation of the genetic and developmental rela-
tionships of 25 architectural traits of bean (Phaseolus
vulgaris L.) was conducted. Special effort was devoted to
understanding the role of branches in yield. Both linear
and multiplicative component compensation systems were
identified among architectural traits, by using decapitation
studies. Difference in branching patterns of two contrast-
ing variation was associated with indole-3-acetic acid
levels in the shoot tip, as measured by Double Standard
Isotope Dilution Assay. Analyses of generation means and
genetic variance components revealed the presence of both
additivity and dominance, and large amounts of heterosis
were identified. Principal components factor analysis
pointed to several clusters among the architectural traits,
for which underlying developmental and genetic interpreta-

tions were discovered.

To my parents, Jane and Edward Howard,

who taught me...

"Be anxiously concerned with the needs
of the time ye live in, and center
your deliberations on its exigencies
and requirements." — Baha'i Writings

ACKNOWLEDGEMENTS

Words fail me in expressing how priviledged I feel to
have had Dr. Wayne Adams as my major professor. I enjoyed
his sense of humor, and was inspired by his insight. I was
touched by his caring, and uplifted by his exemplary ethics.
His subtle guidance encouraged me on the path of independent
thinking, and I have discovered in myself a love of science
so deep I dare say it may match his own. This is the
priceless gift of the great teacher, one I shall cary with
me forever, and though I shall try, it is a gift I can never
repay.

Members of my guidance committee included Drs. Robert
Bandurski, Ardeshir Ghaderi, Don Penner, and Frank Dennis.
To Dr. Bandurski, Mrs. Aga Schulze, and their excellent
research group many thanks are extended for meticulous gui-
dance in performing the IAA extraction procedures of
Chapter 3. I am truly grateful to Dr. Ghaderi, whose
invaluable counsel and preliminary research enabled the
achievement of the genetics of-architectural traits pre-
sented in Chapter 5. I extend warm thanks to Dr. Penner,
whose advice was always pesitive and encouraging. To Dr.
Dennis, whose guidance was instrumental in the early stages

of this work, merci beaucoup.

ii

 

I extend thanks to Dr. Charlie Cress, a truly excellent
teacher, who was helpful as a statistical consultant and
reviewed a portion of this manuscript. Dr. Dave Reicoski,
who overextended himself in giving me computer assistance,
has my unreserved gratitude. Others I could not omit a
thanks to are Ms. Susan Neuhausen, Dr. Tom Isleib, and Dr.
Jim Kelly, whose discussion was more helpful than they know.

I am humbly grateful for the support of my family and
many precious friends. To my husband, Eric, whose love and
encouragement were my constant companions, I owe an unrede—
emable debt.

I gratefully acknowledge the National Science Founda-

tion's support through a Minority Graduate Fellowship.

iii

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . .
LIST OF FIGURES . . . . . . . .
INTRODUCTION . . . . . . . . . .
CHAPTER 1: LITERATURE REVIEW . .
Plant Architecture and Yield

Negative Associations,
and Morphogenic Development

Genetics of Architectural Traits

Strategies for Improvement Yield Through

Architecture . . . . . . . .

Decapitation Studies, Apical Dominance,

Branching . . . . . . . . .
Hormones and Apical Dominance
CHAPTER 2:
Introduction . . . . . . .
Materials and Methods . . .

Results and Discussion . . .

in Beans

Component

FIELD DECAPITATION EXPERIMENT

o

o

and

Correlations of Yield and Yield Components

and Changes with Decapitation

CHAPTER 3: IAA LEVELS AND BRANCH
Introduction . . . . . . . .
Materials and Methods . . .

Extraction . . . . . .

Partitioning . . . . .

iv

GROWTH

0

Compensation

DEAE Sephadex Column Separation . . . . .

High Pressure Liquid Chromatography

Methylation . . . . . . . . . . . . . . .
Gas Chromatography . . . . . . . . . . .
The Equations . . . . . . . . . . . . . .
Analysis of Results . . . . . . . . . . .
Results and Discussion . . . . . . . . . . . .
CHAPTER 4: GENETICS 0F ARCHITECTURAL TRAITS . . .
Introduction . . . . . . . . . . . . . . . . .
Materials and Methods . . . . . . . . . . . .
Statistical Analyses . . . . . . . . . . . . .
Generation Means Analysis . . . . . . . .
Heterosis . . . . . . . . . . . . . . . .
Genetic Variance Components Analysis . .
Heritability . . . . . . . . . . . . . .
Results and Discussion . . . . . . . . . . . .
Generation Means Analysis . . . . . . . .
Heterosis . . . . . . . . . . . . . . . .
Genetic Variance Components Analysis . .

On the Agreement Between the Results
of Generation Means Analysis and Genetic
Variance Components Analysis . . . . . .
Heritabilities . . . . . . . . . . . . .
CHAPTER 5: MULTIVARIATE ANALYSES ON F3 . . . . . .
Introduction . . . . . . . . . . . . . . . .
Methods . . . . . . . . . . . . . . . . . . .

Principal Components Analysis . . . . . .

45

47

47

48

53

53

54

6O

6O

61

63

64

64

65

67

68

68

82

85

94
97
98
98
99
99

Linear Correlations and Multiple Regression
of Yield and Its Components . . . . . . .

Results 0 O O O O O O O O O O O O O O O O O O 0
Principal Components Analysis . . . . . . .
Phenotypic Correlations . . . . . . . . . .

Multiple Regression of Yield and
Yield Components . . . . . . . . . . . . .

SUMMARY AND CONCLUSIONS . . . . . . . . . . . . . . .
APPENDIX 0 O O I O O O O O O O O O O O O 0

LITERATURE CITED 0 O O O O O O O O O O O O O O O O 0

vi

100
101
101

108

114
120
124

127

LIST OF TABLES

CHAPTER 2 Page

2.1 Results of Analysis of Variance conducted
on 23 variables measured on decapitated
and control beans of varieties Single Stem
and Seafarer. . . . . . . . . . . . . . . . . 24

2.2 Grouping of variables according to results of
AOV. O O O O O O O O O O O O O O O O O O O O O 26

2.3 Deployment of nodes on main stem and branches. 28

2.4 Deployment of racemes on main stem and
branCheSO O O O O O O O O O O O O O O O O O O 29

2.5 Deployment of pods on main stem and branches. 29

2.6 Deployment of stem growth on main stem and
branCheS. O O O O O O O O O O O O O O O O O O 30

2.7 Pattern of branching and internode length and
number. 0 0 O O 0 O O O O O O O O O O O O O O 31

2.8 Composition of seed yield. . . . . . . . . . . 32
2.9 Correlations among W(yield) and yield com-

ponents X(pods), Y(seeds per pod), and Z(seed
weight). 0 O O O O O O 0 O O O O O O O O O O O 38

CHAPTER 3
3.1 Number of plants, fresh and dry weight, and
amount of (l4)C-IAA added per sample. . . . . 45
3.2 Analysis of variance table for levels of IAA
per shoot tip, per gram of dry weight, and per
gram of fresh weight. . . . . . . . . . . . 54
3.3 Amount of IAA in each variety standardized
over shoot tip, dry weight, and fresh weight 55
3.4 Ratio of IAA in Single Stem to IAA in

seafarer. O O O O O O O O O O O O O O O O O O 55

vii

CHAPTER 4

4.1

4.2

Number of plots per generation of each cross
which were planted and harvested. . . . . . . 62

Partitioning of generation effects into
additive and dominance effects. . . . . . . . 64

Sources of variation and expectation of mean
squares for parents and F3 families. . . . . . 66

Generation means and percent heterosis for
each variable(V) in cross l(Seafarer*Single
Stem) C O O O O O O O O O O O O O O O O O O O 0 69

Generation means and percent heterosis for

each variable(V) in cross 2(Seafarer*Architype
790780). 0 O O O O O O O O O O O O O O O O O 0 7O

Sums of squares from analysis of variance on
generation means over replications for cross 1.
Experiment analysed as completely randomized
block design. . . . . . . . . . . . . . . . . 74

Results of analysis of variance on generation
means over reps for cross 2. The experiment

was analysed as a completely randomized block
design. . . . . . . . . . . . . . . . . . . 75

Partitioning of generation sums of squares

into additive, dominance, and residual or
epistatic effects, each with one degree of
freedom, in cross 1. . . . . . . . . . . . . 77

Estimates of genetic parameters and standard
errors for architectural traits in cross 1. . 78

Partitioning of generation mean squares into

additive, dominance, and residual or epistatic
effects for cross 2. . . . . . . . . . . . . 80

Estimates of genetic parameters and their
respective standard errors, and R2 values for
the simple model fit in cross 2. . . . . . . . 81

Heterosis, variance of heterosis, and t value
for significance of heterosis. . . . . . . . . 84

Mean squares for analysis of variance conducted

on parents(M4) and F3 families(Ml,M2,M3)
in cross 1. O O O O O O O O O O O O O O O O O 86

viii

C HAPTER 5

5.1

5.2

Mean squares for analysis of variance conducted
on parents(M4) and F3 families(Ml,M2,M3) in
cross 2. O O O O O O O O O O O O O O O O O O O 87

Calculation of coefficients for expected mean

squares in analysis of genetic variance
components of crosses l and 2. . . . . . . .

Variance and heritability of F3 families for
variables(V) in cross 1. . . . . . . . . . . 90

Variance and heritability of F3 families in
cross 2. l O I O O O O O O O O O O O O O C O 91

Estimates of additive(D) and dominance(H)
genetic variance components for crosses 1
and 2. O O O O O O O O O O O O O O O O O O O 93

Percentage of variance accounted for by each
factor in crosses l, 2, and 3. . . . . . . . 102

Grouping of 23 variables from cross 1(Sea*SS)
into seven factors using principal components
anaIYSis. O O O O O O O O O O O O O 0 O O O O 103

Grouping of 23 variables from cross 2(Sea*
Architype) into five factors using principal
components analysis. . . . . . . . . . . . . 104

Grouping of 23 variables in cross 3(SS*

Architype) into six factors using principle
components analysis. . . . . . . . . . . . . 105

Architectural traits grouped into factors
or principal components for crosses 1,
2 and 3. O O O O O O O O O O O O O O O O O 108

Linear correlations of architectural traits
with yield and yield components pods per
plant(X), and weight of seed per pod(YZ). . 110

Maximum R2 improvement model selected for
dependent variable yield when primary yield
components were included in the set of
independent variables. . . . . . . . . . . 115

Maximum R2 improvement model selected for

dependent variable yield when primary
yield components were excluded form the set
of independent variables. . . . . . . . . . 116

ix

APPENDIX

A

Maximum R2 improvement model selected for
dependent variable pods per plant when seed
weight per pod and pods on the main stem
and branches were excluded from the set of

independent variables.

Maximum R2 improvement model selected for
variable
when primary yield components were excluded

dependent

seed

weight

per

from the set of independent variables.

Derivation of the Rittenberg and Foster

equation.

Analysis of variance for amount of IAA per
shoot tip, per gram dry weight, and per gram

fresh weight.

Correlation between values of metric traits

and amount of IAA per shoot tip.

traits).

0

(Three
replications were used as levels of IAA and
correlated with the corresponding replication
means of measurements made on architectural

pod

117

119

127

128

129

CHAPTER 2

2.1

2.2

CHAPTER 3

3.1

305
CHAPTER 5

5.1

LIST OF FIGURES

Page
Young bean plant at the time of decapitation 22

Graphic presentation of variables grouped
according to results of AOV with varieties

(A and B) on the vertical axis and treatments
1(control) and 2(decapitation) on the

horizontal axis. . . . . . . . . . . . . . 27

Compensation among components to total branch
growth in control and decapitated(D) plants of
Single Stem and Seafarer. . . . . . . . . . 34

Compensation among components of pods per

plant in control and decapitated(D) plants of
Single Stem and Seafarer. . . . . . . . . . 35

Internal standards used in the Double Standard
Isotope Dilution Assay for endogenous IAA levels
in beanS. O O O O O O O O O O O O 0 O O O O O 44

Standard curve and retention times of indole-3-
acetic acid(IAA) and indole—3-butyric acid(IBA),
as determined by gas Chromatography, (nitrogen

specific detector). . . . . . . . . . . . . . 50

Gas Chromatographic curve and retention times
obtained by nitrogen specific detector for
Single Stem variety of beans. . . . . . . . . 51

Gas chromatographic curve and retention times
obtained by nitrogen specific detector for
Seafarer variety of beans. . . . . . . . . . 52

Flow Chart of partitioning procedures. . . . 46

Effect of three bean cultivars on the factor
position of selected architectural traits, as
inferred from principal components analysis 113

xi

I NTRODUCTION

The concept of architecture breeding is a fairly new
one. In beans it involves the construction of a plant which
displays its structural characters in a manner which opti—
mizes the interception and distribution of incident radia—
tion. An increase in photosynthecizing capacity leads to
greater ability to support reproductive structures and thus
to increased grain yield. Ideally, the leaves on the bean
plant should be displayed in such harmony that they do not
shade each other, and permit a canopy sufficiently aerated
to preclude disease development due to incubation condi-
tions. This leaf arrangement requires that stems and pe-
tioles be organized and angled to minimize shading within
the plant canopy. In this context, the length and number of
internodes, and the number of branches, are important fea—
tures to be able to manipulate. First, however, the nature
of their relationship to yield and to each other must be
understood.

The above considerations have the goal of increasing
single plant productivity. But the goals of architecture
breeding also include maximizing productivity per unit area.
Increasing productivity per unit area entails greater sowing

density, which intensifies interplant competition. Plants

2
respond to increased interplant competition by growing tal—
ler, sacrificing numbers of branches, and sometimes aborting
reproductive structures. In this context the usefulness of
a strong hypocotyl to support a taller, spindlier plant is
obvious. Also, it would be important to understand the
relationship of branch growth to yield, and to discover to
what extent their productivity loss will be compensated for
by increased number of plants per area. The percentage of
racemes and pods on branches versus main stem, and their
relative adjustments under increased competition must be
understood. Furthermore, increasing the number of pods per
raceme and seeds per pod could be a means of insuring yield
stability if the number of reproductive structures must be
reduced due to loss of branches.

A number of architectural traits have been mentioned
above: internode number and length, branch number and
length, racemes and pods on main stem and branches, hypoco—
tyl diameter, pods per raceme, and seeds per pod. Many more
can be identified. Though for many years breeding work has
concentrated on selection for yield per se, breeders are
becoming increasingly aware of the usefullness of targetting
more particular components of yield for selection.

In this thesis an investigation of the developmental
relationships of architectural traits is conducted. A deca—
pitation treatment is imposed on two contrasting plant types
and the subsequent adjustment in structural components moni

tored. Special effort is devoted to understanding the role

3

of branch structures in the yield construct. The possible
hormonal involvement in branching is probed through an
investigation of shoot tip indole-3-acetic acid levels.
Genetic analysis of the mode of inheritance of a large set
of architectural traits is undertaken. Phenotypic correla—
tions and multiple regression are employed to discover the
degree of association with yield. Finally, principal compo-
nents factor analysis is used, to reduce the complex set of
developmental interactions and extensive genetic information
obtained to a simplified number of factors;

The overall objective of the thesis was to obtain both
developmental and genetic information which might assist in

designing and breeding for improved architecture in beans.

CHAPTER 1

LITERATURE REVIEW

PLANT ARCHITECTURE AND YIELD IN BEANS

In an article entitled "The breeding of crop ideo—
types", Donald (1968) commented on the widespread use of
models in the process of invention and suggested that this
approach could also be applicable to the biological scien—
ces. The construction of a tentative model plant, designed
for yield superiority, coupled with model testing, and in—
tensive selection for yield, was suggested as an alternative
to conventional defect elimination breeding procedures. The
term "ideotype" was coined for this plant construct which
originates from the breeder's ideal.

Adams (1974) defined an ideotype for beans grown under
conditions of monoculture. Some of the features included
were: plant height of approximately 75 centimeters, deter-
minate growth or short indeterminate vine, high number or
nodes, few branches(0—4), small leaves, erect profile, ra-
cemes with 8 florets each, pods with 8-10 seeds, medium
sized seeds, maximally overlaping vegetative and repro-
ductive growth phases, a high partitioning ratio(.6 to .7),
and a number of desirable physiological traits. It also

includes, of course, disease resistance considerations. The

5

concept of plant architecture has arisen (Adams, 1982), re-
ferring to the number, size, shape, structure, arrangement,
and display of plant parts. These structures are relevant
to yield only insofar as they are associated with improved
physiological functions which confer productive superiority.
NEGATIVE ASSOCIATIONS, COMPONENT COMPENSATION, AND MORPHO—
GENIC DEVELOPMENT

The presence of negative associations between some
components of yield has been widely documented in beans.
(Duarte,l961) (Denis,l967). Archibong (1963) looked at
simple correlations between yield components in relation to
plant spacing and noted changes in these. Duarte (1966)
reported the effect of selection for high and low yield
components as producing a compensatory effect on the re-
maining yield components and a "homeostatic stability" in
yield itself. Path coefficient analysis was used to dis—
cover the relationship of two other architectural components
(number of leaflets per plant and size of leaflets) to the
primary yield components, (Duarte and Adams, 1972). Here
again causative relationships between architectural traits
were found and negative correlations were widespread.

Adams (1967) distilled these findings into a unified
theory of component compensation. He proposed that the
demand created by the primary yield components on a limited
metabolic pool over the time sequence of developement

engenders an intraplant competition at the reproductive

6
level. It is not the performance of any one of the indivi-
dual components which limits yield but the intrinsic level
of that metabolic pool. The ability to produce, or to uti-
lize, those metabolic resources is the determinant of yield.
Negative associations among the components are merely mani—
festations of an adjustment mechanism subordinate to the
pre-destined level of productivity. On a wider scale, Weiss
(1939) states that organisms are systems and these systems
tend to preserve their state and Character by their own
intrinsic forces.

Grafius (1978) pointed out that correlated response
between Characters can be linkage induced, physiological, or
due to allometry of development. He recognized, however,
that the "developmental sequence is started as a result of
the genetic code and is at least partially controlled by
it."

Sinnot (1921) is credited with having established the
fundamental principle that "the size of an organ is propor—
tional to the size of the meristem from which it develops."
Grafius (1978) defined two corollaries to Sinnot's rule,
namely, 1. that the ability to manipulate related traits is
inversely proportional to their proximity in time of origin,
and 2. that an inverse relationship also prevails between
numbers and size of organs.

Sinnot (1936), in discussing multi—dimensional growth
of an organ, noted that when the logarithm of length was

plotted against that of width the points fell along a

7
straight line. Though rate of growth for each dimension
could change throughout developemnt, the ratio, as denoted
by the slope of the line, was constant. Sinnot suggested
that the fact that growth of a character could be expressed
by a single constant, unchanging throughout development and
independent of size, was "an important step toward a simpler
statement of what is actually inherited." He concluded that
it is dimensional relationships which are inherited instead
of absolute dimensions themselves. Grafius (1978) observed
that this developmental allometry is related to the need for
morphological balance, hormonal control, and competition for
environmental resources.

Went (1953), in discussing gene action in relation to
growth and development, asserts that "hormonal control of
growth insures proportionality between different parts of
the organism" and again that absolute dimensions are less
directly gene controlled but that proportions, by contrast,
are under gene control. He asserts, moreover, that for a
particular set of growing conditions(environment) the maxi—

mal size of an organism is definitely a heritable trait.
GENETICS OF ARCHITECTURAL TRAITS

Generation means analysis, as explained by Mather, has
been employed classically to elucidate the mode of inheri-
tance of desirable quantitative traits. Powell (1950) em—
ployed this technique on yield components in tomato and

Ghaderi and Lower (1978) illustrated its use for several

8
traits in cucumber crosses. Through use of generation means
analysis, the mode of gene action for a number of archi-
tectural traits has been preliminarily established by
Ghaderi and Adams (1981) in beans, using early generations of
the same crosses as employed in this thesis. They found
plant height and hypocotyl diameter to be governed by both
additive and dominance gene actions in both crosses of
Seafarer vs. Single Stem and Seafarer vs. Architype. In the
first cross, number of branches, racemes, pods, and seeds
per plant were also found to have both types of heritable
gene action. Only dominant effects were present in nodes
below 15 cm and yield, while additive effects were present
for number of nodes above 15 cm and pod length. In the
second cross, number of nodes above 15 cm and pod length
showed both types of effects while number of branches showed
neither. Nodes below 15 cm showed additivity and the
remaining characters all showed dominance.

Dickson (1967) reported that number of pods per plant,
number of seeds per pod, and number of seeds per plant were
determined by a simple additive gene system. Pod number
showed incomplete degree of dominance and mostly recessive
gene behavior.

Heterosis was reported for seed yield and pods per
plant by Evans (1970), and for leaf number and total leaf
area by Duarte and Adams (1963). The later authors explain—
ed their heterotic findings as resulting from multiplicative

interaction between components of complex traits.

9

Heritabilities of the yield components were reported to
be 51.1%, 82.9%, and 84.8%, for pods per plant, seeds per
pod, and seed size, respectively, by some workers, and as
being very low, by others. (Voysest, 1970).

In the study by Ghaderi and Adams (1981) mentioned
above, broad sense heritabilities for each trait were found
to be quite high and consistent over both crosses. For
crosses l and 2, respectively, they were as follows: .80 and
.93 for plant height, .81 and .85 for nodes above 15 cm(main
stem), .51 and .78 for nodes below 15 cm(main stem), .84 and
.73 for hypocotyl diameter, .88 and .93 for racemes, .75 and
.74 for number of branches, .91 and .95 for pods, .97 and
.94 for pod length, .74 and .67 for seeds per pod, .94 and
.97 for seeds per plant, and .96 and .97 for biological
yield.

Analysis of genetic variance components is an additio-
nal method of elucidating the type and magnitude of gene
action in quantitative traits. (Mather, 1977). Duarte and
Adams (1963) used it to discover the type of gene action in
the leaf area trait and its components, number of leaflets
and leaflet size. Ghaderi and Lower (1981) applied it to
obtain estimates of genetic variances from between and
within F3 family variances of several yield characters in

cucumber.

 

10

STRATEGIES FOR IMPROVEMENT OF YIELD THROUGH ARCHITECTURE

Denis (1971) has done one of the few studies on the
relationships of a large number of plant architectural
traits and productivity in beans. He conducted a factor
analysis of 22 variables measured on 16 varieties and iden—
tified two main factors of importance in yield, namely, pod
weight, and pod number related characteristics. He also
identified plant types where one or the other of these
factors is more important to yield and delineated strategies
for the improvement of each type. He concluded that a third
factor, plant growth, is of relatively less importance in
determining individual plant productivity although it is of
more relevance when considering per area productivity where
inter—plant competition becomes important.

Kueneman (1978) studied relationship of growth habit to
yield potential in beans when plant density was increased
with closer spacing. He identified two plant types which
responded successfully to the increased competition created
by closer plant spacing. The first was determinate, short,
compact, with numerous vertical branches with small, verti-
cally displayed leaves, The other density responsive plant
type is indeterminate, non—twining, erect, with few bran-
ches. The interesting observation was that in the latter
plants pods and racemes were set mostly on the main stem.
For this reason reduction in branch number due to increased

competition did not lead to as severe a reduction of yield

11
as for genotypes which set their pods on lateral branches.
He made the general observation that pods below the 5th node
are found mostly on branches, whereas pods above the 5th
node are found mainly on the main stem. Also, successful
genotypes are those which maintain a high harvest index(HI)
with increased plant density. He made no distinction as to
which component of the ratio(economic yield, biological
yield, or both) is adjusted to produce this effect.

In cowpea(Vigna unqgiculata), Aryeetey and Laing (1973)
studied the inheritance of yield components and their geno-
typic correlation with yield. They concluded that for
selection based on yield components to be effective these
components must be highly heritable, strongly genotypically
correlated with yield, and non-negatively correlated among
themselves.

Voysest (1970) attempted to overcome the yield dampen-
ing effect of negative correlations among the primary yield
components in beans by recurrently backcrossing and select—
ing for increased seed size, the most highly heritable
component, while holding the seed number components constant.
The approach was deemed successful.

Grafius and Thomas (1971) suggested that we think of
traits not as individually significant entities but "as
sumps which trap environmental resources." The outcome of
development, which is yield, depends in their view upon the
strategy of deployment of environmental resources by the

plant. They claimed that a "successful variety(high yield-

12
ing) has a markedly different strategy than an unsuccesfull

one."

DECAPITATION STUDIES, APICAL DOMINANCE, AND YIELD

The importance of branch growth as a secondary
yield component has long been the object of curiosity. A
few studies have been conducted where decapitation was used
as a means of delaying maturity or changing the plant archi-
tecture. The effect of this treatment on yield and some
yield Characters was measured. Tayo (1982) measured a number
of vegetative growth characters and reproductive Characters
in decapitated pigeon peas(§ajagus gaian L.). He noted that
decapitation encouraged development of primary and secondary
branches "thereby providing more loci for pod development."
Some compensation between number of pods and seeds per pods
was observed. Number and area of leaves was also enhanced
by decapitation. Overall, greater vegetative matter pro-
duction and higher yields were achieved through decapita—
tion. However, the results reported were only for one dwarf
variety of pigeon pea.

Gehriger et a1. (1981) decapitated Vicia faba plants
with the aim of interrupting the competition between vegeta-
tive and reproductive growth. They observed that even
though number of pods increased, seed weight was low and
yields inferior to the control. GA3 was applied at the
beginning of flowering and gave greater pod set and yield.

The interesting observation was made that harvest-index

13
remained stable over all experiments.

Nagl (1981) reported on selection for reduced apical
dominance in a Vicia_fiaba_mutant. These plants produced 50%
more pod bearing nodes and 26% more pods per plant than the
original parent. There was a 30% reduction in seeds per
pod, and this was negatively associated with number of stems
per plant. The result was that yield decreased in the lines
selected for reduced apical dominance. The author also
presented some interesting comparisons of effect of se-
lection for high and low levels of a trait upon the other
components. When this effect was plotted(high versus low)
it showed almost opposite influences. These "mirror like"
images are a good example of the compensatory relationships

and associations which weave together architectural traits.
HORMONES AND APICAL DOMINANCE

In 1933, Thimann and Skoog published their results on
the inhibition of lateral bud growth by a "growth substan—
ce"(auxin), diffusing from the apical growing point. The
approximate amount of "growth substance" diffusing out of
the terminal bud was determined by Aygna coleoptile assay.
Developing lateral buds were shown also to produce "growth
substance" and to inhibit adjacent or lower buds. "Growth
substance" promoted stem elongation too, in lower concentra-
tions than for bud inhibition.

Snow (1934), writing on the nature of the correlative

inhibition, proposed an indirect theory, claiming that auxin,

14
which only moves basipetally, stimulated an inhibitor which
could move acropetally into the lateral bud and inhibit
growth.

Since these initial efforts, a barrage of minutial and
sometimes repetitive articles have been published concerning
the further action of hormones in the apical dominance
process. Phillips (1975) provided an extensive review of
this work. Thimman (1977) also provided a good discussion
of this work. At one point eight or nine theories had been
proposed on the nature of this correlative inhibition.
(Wickson and Thimman, 1958). Most studies aiming to measure
the level of IAA activity made use of bioassays.

Since the elucidation of the structure of the auxin
indole acetic acid, many authors have attempted to substi-
tute the shoot tip inhibitor with biosynthetic IAA. White
(1976) tediously investigated the response of decapitated
plants at different ages, times of year, amount of IAA, and
type of lanolin paste used for IAA application. A major
question of the validity of hormone application studies is
that concentrations of hormone required to produce satis-
factory results are much higher than in intact plants(as
obtained from diffusion experiments), (Patrick and Woolley,
1973).

Recently, some workers have actually attempted the
isolation and quantification of endogenous IAA in apical
dominance studies. White et al. (1975) collected IAA from

shoot tips and young leaves by agar diffusion and measured

it by bioassay, chromogenic assay, and gas chromatography—
mass spectrometry. Comparison of these methods showed that
their estimates of IAA levels obtained by GC-MS were more

variable than bioassay(Avena coleoptile). Levels of IAA

 

obtained from the latter were .09-.66 nanograms per shoot

t ip, by extraction, and .10 nanogram, by bioassay(agar
collection). Because successive defoliation of trifoliates
very near the apex led to increasing levels of lateral bud
outgrowth, these researchers concluded that young leaves
were responsible for the correlative inhibition of bud out—
growth.

Hillman et a1. (1977) used combined gas chromatography—
mass spectrometry to determine IAA levels of lateral buds in
Phaseolus. They showed a rise in IAA level in the buds
following decapitation. Mean levels of IAA were 6.4 and
15.6 picograms per bud in intact and decapitated lateral
buds, respectively.

McDougall and Hillman (1980) reported on levels of IAA
in leaves and internode sections of Phaseolus seedlings.
Again GC-MS was the technique used. They found that IAA
levels in the leaves (2.0 micrograms per kilogram of fresh
weight in primary leaves, and 8.3 micrograms per kilogram in
the first trifoliates) were significantly lower than apex
IAA levels (38.3 micrograms per kilogram).

Lately, a technique for quantification of IAA using two
isotopes as internal standards has been perfected (Cohen and

Schulze, 1980). Its reliability for detecting free IAA and

l6
IAA derivatives is very high. Vegetative tissue of Phaseolus
has not, until the present study, been assayed by use of
this technique.

No studies were found where investigation of auxin ef-
fects in more than one variety of the Phaseolus vulgaris
species was conducted. Genetic variation in maize coleoptile
growth in relation to the auxin regulatory system was,
however, reported by Moll et al. (1969).

As new hormones were discovered, the effect of these on
the apical dominance process was tested. Wickson and Thimann
(1958) reported an antagonism between the newly discovered
cytokinin and auxin in peas. When cytokinin was applied
with auxin the auxin inhibition was released. Cytokinin
also released inhibition in intact plants. Gibberellic acid
promoted bud elongation too, but only after inhibition was
removed.

A number of experiments have been performed where
exogenous hormone applications have been made to beans in a
variety of systems. Shein and Jackson (1971) applied gib-
berellic acid (GA3), kinetin, and indole-Byl-acetic acid(IAA)
to roots of beans under two different light intensities and
when either young or old leaves were removed. The results
suggested that exogenous GA promotes stem and lateral
growth. IAA and kinetin reduced and eliminated this
growth, respectively. Removal of young and mature leaves
reduced main stem growth; removal of young leaves promoted,

and of mature leaves reduced, lateral shoot growth. Jackson

l7

and Field (1972) applied the same hormones, each at four
different concentrations, in all possible combinations to
the same system. Again, GA increased growth of main stem
and laterals, reducing apical dominance. Kinetin reduced
apical dominance but still lowered GA induced growth. IAA
slightly decreased growth of main stems and laterals, and
slightly increased apical dominance. The authors suggested
that a homone balance system was responsible for apical
dominance and lateral bud elongation. Shein and Jackson
(1973) then expanded the experiment applying those combina-
tions to decapitated stems, petioles, and buds. GA, IAA,
and kinetin showed some of the same effects but not con-
sistently. On occasions, GA3 and IAA acted synergistically
to promote and sometimes to inhibit lateral shoot growth.
Kinetin applied with GA often increased GA—induced growth.
The diversity of results was attributed in part to the point
and concentration of hormone application, age of the plant
or bud, and to light intensity and nutrition. In another
system, IAA was applied to decapitated stems while GA and
kinetin were applied to roots. (Field Jackson, 1974). Here
again a hormone balance system seemed to be operative.
Levels of hormones seemed to be relevant only in relation to
levels of other hormones.

A few experiments have been conducted where effects of
organs were observed without addition of exogenous hormones.
These were primarily experiments where mature or immature

leaves were removed and subsequent effect on lateral branch

18
development observed. (Kigel, 1980) (Guven and Vardar,
1977). In general, young leaves are thought to be inhibi-
tors of bud growth. Primary leaves inhibit cotyledonary bud
growth. The inhibitory effect of the stem on cotyledonary

bud growth was shown in soybean plants. (Peterson and Fletcher,

1975).

CHAPTER 2

FIELD DECAPITATION EXPERIMENT

INTRODUCTION

How important are branches to yield? Is there an
optimum number of branches for a productive plant? What
percentage of reproductive structures are borne on branches?
What is the relationship of branch growth to other architec-
tural traits? What happens to the other metric traits when
the branching configuration is Changed in a plant? Does
greater or lesser number of branches confer competitive
superiority with increased interplant competition(such as
that produced by greater plant density)?

Stem growth(of which branch growth is a major part) in
plant architecture can be equated in function to the structu—
ral skeleton of a skyscraper. Upon it are stacked in perfect
order the phytomeric units, like the many floors and apart-
ments of its steel and glass counterpart. Each unit consists
of a source-sink team of leaves and pod-bearing racemes. In
designing and breeding for an efficient bean architype under—
standing of the role of branches in the yield product is
crucial. Not only because of considerations of single plant
productivity, but because maximizing productivity on a per

area basis is a key agronomic goal.

I9

20

The experiment which follows was designed with the
objective of obtaining some understanding of the complex
interrelations between reproductive and structural cha-
racters of plant architecture. By imposing an artificial
change in the configuration of stem growth through decapita—
tion "architectural isolines" were, in effect, obtained.
That is to say, plants with two different configurations or
architectures but essentially identical genetic backgrounds
were created due to the release from apical dominance.
A comprehensive array of Characters was monitored, and this
was essential to a wholistic understanding of fundamental
plant allometry. Two contrasting plant types were used to

asses possible genotypic differences in response.
MATERIALS AND METHODS

Bean plants of the variety Seafarer and of the
Single Stem breeding line 791515 were grown in the field in
E. Lansing, MI in 1982. Plants were grown at 15 centimeter
spacing in plots of four 5 meter-long rows. The experiment
was replicated four times and included two treatments 1.
control, and 2. decapitation. The variety Seafarer is a
highly branched Type I variety with weak apical dominance.
The Single Stem breeding line 791515 was derived originally
from an irradiated mutant of Seafarer(M17) which was crossed
to BTS, selected in the F3 generation, crossed to Nep-Z, and
then selected again in an advanced generation. It is

generally few-branched with strong apical dominance.

21
One month after sowing the plants were decapitated.
At this time the plants were approximately 10 centimeters in
length and had three or four nodes on the main stem. The
lateral buds which had not visibly elongated previously had
just begun to show some growth. Figure 2.1 shows the ap—
proximate status of the young bean plant at the time of
decapitation. At physiological maturity 10 plants were
sampled randomly from the two central rows of the plot. The
only criteria used for sampling was that each plant chosen
be flanked by two other plants at a 15 centimeter distance
in the row. This precaution was taken to guard against the
known interaction of branching patterns with plant popula-
tion in cases of irregular plant stand. The following
measurements were made on each plant:
(l)Number of branches
(2)Branch growth(sum, in cm)
(3)Height(cm)
(4)Nodes below 10 cm on the main stem
(5)Nodes above 10 cm on the main stem
(6)Hypocotyl diameter(in 1/32 inch)
(7)Racemes on the main stem
(8)Pods on the main stem
(9)Nodes on branches
(10)Racemes on branches
(11)Total pods
(12)Five pod length(cm)

(13)Seeds per five pods

 

 

 

 

 

 

 

 

 

 

”Wt, t‘ Apical had.
an era a a
hat [1 ' Di
,’ A First a
second -
fl ‘ ~ % trifoliates .
_ " ‘ _ Main stern
laterals ‘ ,/ .
. \ / . ' .
/ a
\
I i Primary
Botylerlurrs ) \ leaves

Figure 24. Young bean plant at the time of decapitation (—> ).

23

(l4)Yield(gm)
Several other variables were calculated from the original
ones:

(15)Average branch length(cm)

(l6)Nodes on the main stem

(l7)Tota1 stem growth(branches + height)

(18)Total nodes

(l9)Average internode length on branches(cm)

(20)Average nodes/branch

(21)Total racemes

(22)Pods on branches

(23)Pods/raceme

RESULTS AND ISCUSSION

The results of the analysis of variance conducted
on the 23 variables is depicted in Table 2.1. The level
of significance is shown for treatments, varieties, and the
treatment*variety interaction. The results are further
summarized in Table 2.2, where variables have been divided
into five groups according to their AOV results. The general
response of the variables is graphically represented in
Figure 2.2.

Group A - Variables in this group did not show any sig-
nificant difference either across varieties or treatments.
Only one variable fell into this category, it was

(13)seeds/pod.

Group B - This group was Characterized by variables which

24

Table 2.1 - Results of Analysis of Variance conducted on 23
variables measured on decapitated and control beans on vari-
eties Seafarer and Single Stem.

VARIABLE TREATMENT VARIETY TMT*VAR
(l)Number of branches N.S. ** N.S.
(2)Length of all branches * ** *
(3)Height ** ** **
(4)Nodes below 10 cm ** ** *
(5)Nodes above 10 cm ** ** **
(6)Hypocoty1 diameter N.S. ** N.S.
(7)Racemes on the main stem ** ** **
(8)Pods on the main stem ** N.S. **
(9)Nodes on branches ** N.S. *
(10)Racemes on branches ** ** **
(ll)Total pods N.S. ** N.S.
(12)Five pod length N.S. ** N.S.
(13)Seeds per five pods N.S. N.S. N.S.
(l4)Yield N.S. ** N.S.
(15)Average branch length ** ** **
(16)Nodes on the main stem ** ** **
(17)Total stem growth * ** N.S.
(18)Total nodes N.S. ** N.S.
(l9)Ave. internode length(br.) ** ** *
(20)Average nodes/branch ** ** **
(21)Total racemes N.S. ** N.S.
(22)Pods on branches ** ** **
(23)Pods/raceme N.S. ** N.S.

N.S.,*,** = Non-significant and significant at the 5%, and 1%
level,respectively.

25
were initially distinct for each variety. Both varieties
reacted similarly to the treatment(no treatment*variety
interaction) by showing no significant response. The group
included the following variables: (l)number of branches,
(6)hypocotyl diameter, (ll)total pods, (12)pod length,
(l4)yield, (21)total racemes, and (23)pods/raceme.
Group C - This category depicted a variable which was also
initially different across varieties and where response by
varieties was parallel. Contrary to results in group B,
however, this variable responded significantly to the treat—
ment. Variable (17), total stem growth was of this type.

Group D - In this group were included variables which
were not initially different between varieties but for which
the varieties showed dissimilar but significant reactions to
the treatment. This disimilarity may have been due to
different trends in the means or due to a difference in
degree of response even though mean trends were similar.
Variables (8), pods on the main stem, and (9), nodes on
branches, were in this category.

Group E - This final and largest group consists of
variables which showed different but significant response to
treatments in the two varieties. Initially the varieties
differed significantly for these traits. Included in this
group were the variables: (2)total branch length, (3)height,
(4)nodes below 10 cm, (5)nodes above 10 cm, (7)racemes on
the main stem, (10)racemes on branches, (15)average branch

length, (16)nodes on the main stem, (19)average node length

26
on branches, (20)average nodes/branch, and (22)pods on

branches.

Table 2.2 - Grouping of variables according to results of
AOV.
INTERACTION VARIETY TREATMENT VARIABLE GROUP

N.S. 11611-1112! B
,,,//””’ 14,21,23,18

SIG 17 C

N.S.¢:::::::::
’,,/’/’ SIG 8,9 D
‘\\\\\\\ N.S.
SIG==:::::::::
SIG 2,3,4,5,7, E

10,15,16,19
20,22

*No distinction is made here between 5% and 1% levels of
significance.

SIG

The means for each variety and treatment were compared
using LSD. An LSD value was calculated for each variable.
The means for several related "clusters" of variables are
depicted separatly in Tables 2.3—2.8 in order to facilitate
understanding of the complex interactions of variables in

the total plant growth.

In Table 2.3, the relationship of nodes on main stem
and branches to total nodes on the plant is shown. Decapi-
tation did not produce a meaningful Change in the number of

nodes below 10 cm on the main stem in either variety. A

27

amocv b @205. m oaocv O

. ./ .

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m
m

_ N . ~ . m

omocv a 232. m
>
. Po 0
Po m
. n _ n -. n . m

39:5 Nu. mania came-58:6: as Sammie» mac—Gen Andaman 8 33.3
2 >O< 5:: 553.8 3 2:. E o: :5 53:8. 9:... as: "33338
a .833: can N Agog—9.832: o: :5 gag—.8. ex?

28
drastic change in number of nodes above 10 cm, however, was
imposed by the decapitation treatment. Plants of both va-
rieties responded by significantly increasing the number of
nodes on branches. The resultant was that there was abso-
lutely no change in total nodes on each variety attributable
to decapitation. The plant appears to be flexible in its

deployment of nodes on main stem or branches.

Table 2.3 - Deployment of nodes on main stem and branches.

VARIABLE SEAFARER SINGLE STEM
CONTROL DECAPITATBD CONTROL DECAPITATED

(4)Nodes below 4.50b* 4.40b 4.13b 3.18a
10 cm on
main stem

(5)Nodes above 3.90b .20a 12.28c .03a
10 cm on
main stem

(9)Nodes on 21.45a 27.38b 20.03a 32.220
branches

(18)Total 29.85a 31.98a 36.43b 35.43ab

nodes

* letters refer to LSD's between treatments and varieties
within a single variable.

A similar pattern was apparent for racemes, as shown in
Table 2.4. Racemes on the main stem was drastically reduced
by decapitation but racemes on branches increased propor-
tionately. The result was that total racemes for each
variety was no more and no less than in the control. Thus
it seems the plant achieved its full developmental potential
for number of nodes and racemes and this flexibility in
placement of these structures was the mechanism employed to

production stability.

29

Table 2.4 - Deployment of racemes on main stem and branches.

VARIABLE SEAFARER SINGLE STEM
CONTROL DECAPITATED CONTROL DECAPITATED
(7)Racemes on 4.15b* .88a 7.25c .OOa
main stem
(10)Racemes on 16.78c 19.40d 7.28a 14.28b
branches
(21)Racemes 20.93b 20.28b 14.53a 14.28a
total

* letters refer to LSD's between treatments and varieties
within a single variable.

The plants demonstrated flexibility in deployment of
pods also. As shown in Table 2.5, pods on the main stem
decreased with decapitation while pods on branches in-
creased. The result was that control and decapitated plants
at maturity did not differ for number of total pods. Seed
yield, as a consequence of this pattern of compensation, was

not affected in either variety by the decapitation treatment.

Table 2.5 - Deployment of pods on main stem and branches.

VARIABLE SEAFARER SINGLE STEM
CONTROL DECAPITATED CONTROL DECATITATED
(8)Pods on 8.52c* 2.15b 10.80d .03a
main stem
(22)Pods on 29.53c 35.00d 8.68a 20.05b
branches
(ll)Total 38.05b 37.15b 19.48a 20.08a
pods
(14)Seed 28.20b 28.70b 24.03a 24.70a
yield

* letters refer to LSD between varieties and treatments
within a single variable.

30

In Table 2.6 the distribution of stem growth on the
plant is shown. Here the varieties reacted differently to
the treatment. Single Stem displayed the same flexibility
discussed previously in its stem growth deployment. Stem
growth was aborted on the main stem due to decapitation.
The branches responded by significantly increasing their
growth so that the sum of height of main stem and total
branch length ended up to be equal. Seafarer, however, did
not compensate for its reduced main stem height by an in—
crease in branch growth and this resulted in much reduced
total stem growth in the decapitated plants. Seafarer did
not appear to have flexibility in the placing of stem growth
because the branches grew as normal even though main stem

growth was abnormally low.

Table 2.6 - Deployment of stem growth on main stem and

branches.

VARIABLE SEAFARER SINGLE STEM
CONTROL DECAPITATED CONTROL DECAPITATED

(2)Branch 163.95d* 159.52d 94.35a 133.77b
growth

(3)Height of 32.13b 9.13a 58.05c 7.40a
main stem

(l7)Branches 196.08c 168.65b 152.40ab 141.18a

+ height

* letters refer to LSD between varieties and treatments
within a single variable.

It has already been mentioned that branch length re-

mains the same for Seafarer and changes in the case of

31
Single Stem. Although the number of branches does not
change for either variety the structure of these branches
does change with the treatment. In Single Stem the change
in average branch length is mainly accounted for by the
increase in the number of nodes, because the average inter-
node length does not change appreciably. In Seafarer, number
of nodes per branch also increases. The branch length
remains the same only because the average internode length

is decreased.

Table 2.7 - Pattern of branching and internode length and

number.

VARIABLE SEAFARER SINGLE STEM

CONTROL DECAPITATED CONTROL DECAPITATED

(l)Number of 8.33c * 7.73b 2.95a 3.05a
branches

(20)Nodes/per 2.53a 3.63b 7.20c 10.70d

branch

(l9)Internode 7.83c 5.93b 4.82a 4.18a

length on

branches(ave.)

(15)Branch 19.80a 20.50a 33.47b 44.48c
length(ave)

*The LSD value for comparison of these treatments was .52.
Since the two Seafarer treatments vary by only .60 the
difference, although significant, is not considered meaning-
ful.

The complex interactions of pod and seed dimensions as
they relate to yield is depicted in Table 2.8. Pod length,

seeds per pod, and pods per raceme were not changed by the

decapitation treatment for either variety. Total pods and

32
seed yield were not, therefore, affected. Though pod length
is greater for Single Stem than for Seafarer the number of
seeds per pod for both varieties is the same. And although
pods per raceme were only slightly lower for Single Stem the
number of total racemes was much lower producing less pods

per plant and reduced seed yield.

Table 2.8 - Composition of seed yield.

VARIABLES SEAFARER SINGLE STEM!
CONTROL DECAPITATED CONTROL DECAPITATED

(12)Pod 37.38a* 38.10a 49.70b 49.33b
length(5)

(13)Seeds/ 24.70a 24.50a 24.58a 25.28a
5 pods

(23)Pods/ 1.8Sb 1.83b 1.38a 1.25a
raceme

(ll)Pods 38.05b 37.15b 19.48a 20.08a
total

(l4)Seed 28.20b 28.70b 24.03a 24.70a
yield

* letters refer to LSD between varieties and treatments
within a single variable.

Aside from the linear relationships where compensation
in growth is apparent, there are several multiplicative
relationships(both vegetative and reproductive) where
compensation is a rule in development. Two examples are
given below and the data is depicted in Figures 2.3 and 2.4.
Branch length = (branches/plant)(nodes/branch)(length/inter-

node)

Pods per plant = (nodes/plant)(racemes/node)(pods/raceme)

33

Compensation among primary yield components was postu-
lated by Adams to originate from a time sequence of demand
by plant components on a limited input of resources. I
suggest the obvious, that compensation has broader applica—
bility than just to immediate components of yield. It is a
widespread phenomenon of morphogenic development, the me-
chanism by which the genotype expresses its ability to
produce a prescribed measure of biological material in the
environment and the set of stresses impingent upon it.

In Adams' compensation theory, "resources" had a broad
definition but was usually taken to mean metabolic substrates.
The data presented in this chapter suggest an additional
inference. Although compensatory effects are clearly visible
in intact plants, in at least one case compensation did not
occur when the plant was decapitated. More specifically, in
the linearly additive case of total stem growth(which is the
sum of height of the main stem and branch growth). Seafarer
did not increase its branch growth to replace lost main stem
growth while Single Stem did. This suggests that the "power"
to compensate is a unique characteristic of each genotype
and, in this trait, was related to the presence or absence
of the shoot tip. Since the shoot tip has never been as—
sociated with providing metabolic substrates to other de-
veloping parts of the plant but has, however, been as-
sociated with providing hormonal stimuli, it could be infer-
red that the compensation phenomenon has a hormonal basis.

This hormonal input hypothesis and that of metabolic

314

£9.23»
new 58m $9.5 *0 3:2: .9 253383 tea 35:8
5 «:2: 8a «to; so 35:859. 9555 20285588 .?~ 952”.

 

 

 

 

 

 

2:82 one: Eu...
«coal $882 mono:
- — -
\\\F/ 1 m
8 . o mm 1 00..
m. u mm Legumm I on; U
hm _. o <wm / owhuomufiOON m.
on . 5m // .. mm m
llddnmlm a 52m gosvaW/ .. on
.8 San. 52m 2&5! 1 mm
L 0v

.8338
15 :85 29.5 we 35.: E. tagging... new 35:8
5 Eu... .2. upon ho 3:22.88 9.25 stanza—:00 .Q.N 2:2”.

 

 

 

 

258.. one: :63
30% $888 $8:
- . _
\\\. , 1 m

.6, ONuomm

 

m. . mm Lot—zoom 1 on; u

h .. oLoLofiomJbON n

on . 4mm /% 1 mm m
lllmfl a 52m o_oc_m..\W/ 1 on
.8 «Ba 52m mama/L 1 mm

 

Low

36
substrates are not mutually exclusive. Indeed, they could
easily coexist if the size of the substrate pool were consi-
dered a manifestation of the hormone status of the genotype
and mobilization of substrates assumed to be under hormonal
control.

It is not possible to extrapolate from this one case to
all compensatory relationships in the plant's morphology but
other plant organs may be the source of hormones which
influence similar compensatory relationships. Certainly in
the XYZ complex(pods per plant, seeds per pod, and seed
weight) where compensation is a result of physiological
processes of pod and flower abcission, and of seed abortion,
hormonal mechanisms are not too difficult to imagine. This
is a difficult hypothesis to test. The purpose and results
of the investigation in Chapter 4 are relevant here.
CORRELATIONS AMONG YIELD COMPONENTS AND CHANGES WITH DECAPI-
TATION

Correlations among yield components X, Y, and 2, were
calculated. Variables 11(pods per plant), and 13(seeds per
pods) correspond to yield components X and Y, respectively.
Component Z(seed weight) was calculated by the equation
Z=W/(X*Y) where w is variable 12(seed yield per plant).

Linear correlations between yield and the yield com-
ponents were also obtained. The correlation between yield
and pods per plant was further broken down into that of

yield and pods on the main stem or pods on branches.

37

Correlations among pods per plant and seeds per pod
were not significant in either variety or treatment. No
significant correlation was found in Seafarer between pods
per plant and seed weight. Decapitation produced no change
here. Decapitation in Single Stem, however, produced a low
but significant negative correlation between pods per plant
and seed weight showing that the use of early resources to
set a higher number of pods per plant had to be compensated
for by the decreased filling of the seed later on. It must
be remembered that the stress imposed by decapitation could
not be anticipated ‘by the plant when it projected the
number of pods to be produced. Since decapitation did not
reduce pods per plant it would appear that the number of
pods per plant is pre-determined by the genetic blueprint of
each genotype. That is to say, for each environment there
is a prescribed "potential" for pod production. It is
curious, however, that the location of the pods(whether on
main stem or branches), does not appear to be pre-established.

Significant negative correlations were found between
seeds per pod and seed weight in both varieties and treatments.
In Single Stem this negative correlation increased in magnitude
upon decapitation showing that decapitation heightened compe-
tition between the process of seed set and seed filling. It
appears that decapitation had the effect of increasing inter-
plant competition due to increased number of stems(shifing

from one dominant axis with smaller branches to several

38
Table 2.9 - Correlations among W(yield) and yield components

X(pods), Y(seeds per pod), and Z(seed weight).

SEAFARER SINGLE STEM
CONTROL DECAPITATED CONTROL DECAPITATED
x11 1:13;" ""333-" 3:32“ ""311?"
X2 .154 -.174 -.255 -.3l6*
YZ —.765** -.667** -.384* -.757**
wx .885** .868** .703** .861**
WX(main stem) .441* .137 .559** .196
WX(branches) .784** .812** .451* .861**
WY -.093 .300 —.038 .146
wz .351* .130 .385* .002

dominant lateral branches and no central axis). Increased
interplant competition creates a situation of increased
intraplant competition and limited resources available for
plant growth over a time sequence of demand (Adams, 1967).
In this situation one would expect that negative correla—
tions among yield components would become more pronounced.
In Single Stem this is the case because decapitation did not
reduce the amount of total vegetative growth(as measured by
total stem growth). In Seafarer, however, no increase in
competition was created because growth in branch length did
not compensate for loss of the main stem length.

In the correlations of the components with yield signi-
ficant correlations with pods per plant were present in both

varieties and both treatments. When the X component was

39
broken down into pods on the main stem and pods on branches
significant correlations with yield were found in both
varieties for both traits. The magnitude of the correlation
was higher in Seafarer for pods on branches and in Single
Stem for pods on the main stem showing the greater association
of yield with a dominant main stem in Single Stem as opposed
to branches in Seafarer. Decapitation produced, as would be
expected, an increase in the magnitude of the correlation of
pods on branches and a loss of significance in the correla-
tion of pods on the main stem with yield.

No correlation of seeds per pod with yield was found in
either variety or treatment. Significant correlations of
seed weight with yield was found in both varieties but the
correlation disappeared upon decapitation. Pods moved to
branches and seed weight was no longer associated with

yield.

CHAPTER 3

IAA LEVELS AND BRANCH GROWTH

INTRODUCTION

Seafarer and Single Stem varieties of bean exhibit
contrasting branching patterns. Upon decapitation the va-
rieties show different degrees of developmental compensation
(Chapter 2). The objective of this study was to determine
if the difference in branching and compensation patterns
could have a hormonal basis.

There is ample evidence in the literature that auxins
play a role in apical dominance. It is most likely that it
does not act alone but is one of a number of factors govern;
ing the process. Although the majority of apical dominance
work is done with beans, no comparative studies have been
discovered by the author where within-species variability in
auxin(indole-3—acetic acid) levels was measured.

The science of plant breeding revolves around the exis-
tence of species variability. From a cursory examination of
any crop's germplasm it can be seen that large within—
species differences exist for agronomic traits which are the
manifestations of a physiological process. Seed abortion,
pod abscission, and apical dominance are a few examples in

beans. Yet, hormone workers, for the most part, have ne—

l+0

41
glected to investigate genotypic variation in physiological
processes. This sort of information is essential if we are
to begin bridging the most fundamental sequence in biology,
i.e. the pathway from agronomic trait, to physiological
process, to hormone, to enzyme or polypeptide, and lastly,
to the gene.

The method of assay of indole-3-acetic(IAA) employed
herein involves isotope dilution analysis, a prefered method
of quantitative measurement of labile natural products. It
makes use of radiolabeled IAA as the first standard and a
chemically related radiolabeled compound, indole-3-butyric
acid, as the second. The second standard, added in known
amounts at the last step before gas-chromatography, permits
the accurate determination of the final specific activity of
the first standard. The degree of dilution of the first
internal standard by endogenous IAA permits quantification
of the IAA present in the sample. Nanogram sensitivity is
obtained with this method. This is, to my knowledge, the
first instance of its application to assaying IAA levels in

vegetative bean tissue.
MATER IALS AND METHODS

Bean plants of the variety Seafarer and of the Single
Stem breeding line 791515 were grown in the field in East
Lansing, MI in 1982. Plants were grown at 15 centimeter
spacing in plots of four 5 meter-long rows spaced 50 cm

apart. Four plots of each variety were grown. The variety

42
Seafarer is a highly branched(8 branches) determinate bush
with weak apical dominance. Single stem is generally few—
branched(0—3 branches) with strong apical dominance and
indeterminate growth habit. These are the same experimental
materials as described in Chapter 2.

One month after sowing the plants were decapitated. At
this time the plants were approximately 10 centimeters in
length and had three to four nodes on the main stem. The
lateral buds which had not visibly elongated previously had
just begun to exhibit some growth. The decapitated shoot
tips were one to two centimeters long and included the
smallest, unexpanded trifoliate, which was at most one cen-
timeter in diameter. The shoot tips from each four row
plot(sample) were bulked and placed immediatly on dry ice,
100-200 plants were collected per plot. Each sample was
then weighed and stored in a liquid nitrogen refrigerator
until extraction. The number of plants per plot(sample) was
recorded and is shown in Table 3.1 together with weight per
sample.

The endogenous indole acetic acid(IAA) content of each
sample was assayed using the Double Standard Isotope Dilu-
tion method(Cohen and Schulze, 1981), with minor modifica-
tions. The steps in the procedure include extraction, puri-
fication by partitioning, DEAE Sephadex Column separation,
and high pressure liquid chromatography. The final step is

the use of gas chromatography.

43

EXTRACTION

Each sample was homogenized in a Sorval Omnimix blender
with enough pure acetone to make up the final acetone con—
centration of 70%. Another aliquot (100-150 ml) of 70%
acetone was added prior to homogenization. A quantity of
carbon—2 labelled (l4)C-Indole Acetic Acid(IAA) was added
to the homogenate as an internal standard. The chemical
structures of the two internal standards used in the assay
are shown in Fig. 3.1. The initial specific activity of the
IAA was 42,200 microCi/mmole. The homogenate was stored at
room temperature for 20-24 hours, then filtered in vacuo and
partitioned. At this point several aliquots of the filtrate
were counted on a scintilation counter and recorded as the

initial amount of IAA added.

PARTITIONING

After filtration the extract was dried in vacuo in a
rotovapor until the aqueous fraction was obtained. Fig. 3.5
shows the flow of steps in the partitioning procedure. The
aqueous fraction was then acidified to pH 3, using 4 M phos-
phoric acid(H3PO4) and extracted with redistilled ether.
This fraction was dried down to half volume and extracted
twice with a l M solution of sodium bicarbonate(NaHC03) at
pH 10. The sodium bicarbonate fraction was acidified to
pH 3 and extracted with ether again. The ether extract was

dehydrated by adding anhydrous sodium sulfate, and then

1,1.

INDOLE " 3" ACETIC ACID

.. i?
/ CH2 -c-OH

Wu

 

 

 

 

IZ

INDOLE' 3‘ BUTYRIC ACID

 

 

 

 

o
/ l I CHz-CHz-g-OH
V\N/*

H

Figure 31. Internal standards used in the Double Standard IsotOpe
' Dilution Assay for endogenous IAA levels in beans.
( * indicates radioactive label).

45
filtered and taken to dryness on the rotovapor. It was

subsequently re—eluted in 1 ml of 50% 2-propanol.

Table 2.1 - Number of plants, fresh and dry weight, and
amount of (l4)C—IAA added per sample.
VARIETY REP #PLANTS FRESH WT DRY WT (l4)C-IAA added
(grams) (grams) (uCi) (mg)
Seafarer 1 178 14.21 1.6312 .0732 .3074
2 94 11.00 1.2156 .0388 .1630
3 95 9.43 1.0936 .0631 .2650
Single 1 114 14.08 1.6851 .0464 .1949
stem
2 131 14.08 1.6648 .0531 .2230
3 113 20.50 1.7310 .0231 .0970

DEAE SEPHADEX COLUMN SEPARATION

A DEAE Sephadex column was prepared using 10 cm of DEAE
Sephadex in a glass column .6 cm in diameter. The column
was washed with approximately 10 m1 of 50% 2—propanol after
which the sample was placed over it. The column was then
washed until almost clear of chlorophylaceous pigments with
50% 2-propanol. When the column was clear the IAA was
eluted with a gradient of 50% 2-propanol and 5% glacial
acetic acid. Fractions of approximately 1.5 ml were collect-
ed on a fraction collector. The IAA peak was usually found
between fractions 7 and 20. The fractions were collected
and dried down completely on the rotovapor and re-eluted

with 100 microliters of 100% ethanol.

46

water-acetone extract

1 acetone volatilized on rotovapor
aqueous fraction
acidified to pH 3

ether added

aqueous fraction
discarded

 

.1\/

ether fraction

dried down to half volume on
rotovapor

NaHC03 added

 

ether fraction
discarded

AZ

NaHC03 fraction

acidified to pH 3

ether added

 

NaHC03 fraction
discarded

AZ

ether fraction

sodium sulfate
(anhydrous) added

filtered

XL

dried down on rotovapor

taken up in 1 ml
50% 2-propanol

 

J

IAA in 50% 2-propanol

Fig. 3.5 - Flow chart of partitioning procedures.

47

HIGH PRESSURE LIQUID CHROMATOGRAPHY(HPLC)

The sample was then placed on a reverse phase PARTISIL—
10 ODS HPLC column of 4.0 mm x 25 cm dimensions. The elu-
ting solvent was 30% ethanol + 1% acetic acid. Fractions
were again collected on a fraction collector. The IAA was
usually eluted at three to twelve mililiters. IAA containing
fractions were collected and dried down on the rotovapor.
They were then taken up in 100 microliters of methanol and

transferred to a 2 ml screw cap vial.

METHYLATION

The sample was methylated under a hood by adding 100-
150 microliters of diazomethane until a faint yellow color
appeared. This was then dried down using a nitrogen stream.
100 microliters of methanol were again used to solubilize
the solvent. This solution was stored in a freezer until
being injected on the gas chromatograph. Storage of the
sample in methanol insured stability of the IAA in solution.

A standard containing (l4)C—Indole Butyric Acid(IBA) at
a concentration of 100 ng per microliter was also methylated
and stored in the same manner. The initial specific activity
of the IBA was 508 microCi/mmole. Immediately prior to in—
jecting on the gas chromatograph the sample was again dried
on the nitrogen stream and re-eluted with 50 microliters of
tetra hydro fluran(THF). The IBA standard was re-eluted in

100 microliters of THF. At this point 5 microliters of the

48
methylated IBA standard was added to each sample. A stand—
ard containing synthetic IAA and IBA was also made up and

methylated for use in determining retention times on the GC.

GAS CHROMATOGRAPHY

The Nitrogen Specific Detector(NSD) and the Flame
Ionization Detector(FID) were both used in the assay. GC
columns were packed according to the following procedure:
two columns of dimensions 6' x 2mm were sylonized and
packed under vacuum with 3% OV-17 gas chrom. Q 100/120. The
columns were conditioned in the CC for 48 hours at 275 C by
running a continuous stream of nitrogen gas through them.
Several injections of IAA standard were made to saturate
sites on the column with affinity for IAA.

The conditions for both FID and NSD injections were as
follows:

Column Temperature - 180 C

Detector Temperature - 240 C

Injector Temperatures — 240 C

Nitrogen Gas Pressure - 60 PSI

Hydrogen Gas Pressure - 12 PSI for FID, 42 PSI for NSD

Air Pressure - 50 PSI
Nitrogen Specific Detector - The IAA + IBA standard was
injected several times to obtain approximate retention times
for IAA and IBA. The retention times under the conditions

specified were:

S IAA — 5:00 minutes to 6:35 minutes.

49

IBA - 10:15 minutes to 12:30 minutes.

Figure 3.2 shows the peaks and retention times obtained
for the IAA + IBA standard. The samples were injected next
at 1x10‘(12) attenuation and retention times recorded for
all peaks. Three replicate injections of one microliter were
made per sample. The IAA and IBA peaks were easily distin-
guishable, not only due to agreement with standard retention
times but also because the curves were fairly clean with no
other interfering peaks. Figures 3.3 and 3.4 show examples of
Seafarer and Single Stem curves. In general, Single Stem
curves were slightly dirtier. The recorded peaks were pho-
tocopied, cut out and weighed. The ratio of IBA to IAA
peaks for each sample was recorded. This number is the
molar ratio of IBA to IAA since each molecule contains one
nitrogen atom which is detected by the NSD.

Flame Ionization Detector - The IAA + IBA standard was again
used to obtain retention peaks. Since the IAA and IBA
concentrations in the samples were too low to be recorded on
the FID chart, retention times were estimated by adding 30
seconds to retention times of the standard peaks. Two injec-
tions of 5 microliters each were made for each sample.
Immediately after each sample injection the flame was extin-
guished. At the appropriate times the IAA and IBA were
collected by placing glass tubes over the detector jet. The
tube dimensions were .3 cm inside diameter and approximately
24 cm in length. Each tube had been stuffed with glass

wool at the upper end. After collection, each tube was

50

 

 

 

 

 

 

 

STANDARD
IAA
IBA
w L - fij Afi— ‘2:
' 1 I l I

 

 

.5'00 6'35 IO*I5 I260
RETENTION TIME

Figure 32. Standard curve and retention times‘of Indole-3-Acetic Acid
(IAA) and lndole-3-Butyric Acid (IBA), as determined by
gas chromatography, (nitrogen specific detector).

51

 

SEAFAR ER

IBA

 

 

 

IAA

 

 

 

 

 

J J 1 1
5:00 645 10:00:24:

RETENTION TIME

 

Figure 34. Gas chromatographic curve and retention times obtained by
nitogen specific detector for Seafarer variety of beans.

 

SINGLE
’ STEM

 

 

 

TI IBA

L.

~ I 1 1 1
5:00 6:35 l0tOO l2'30

RETENTION TIME

 

 

 

 

 

 

 

 

 

Figure 33. Gas chromatographic curve and retention times obtained by
nitrogen specific detector for Single Stem variety of beans .

53
inverted and rinsed with 5 m1 of ACS cocktail fluid and
counted on a scintillation counter. The ratio of IAA to IBA

radioactivity was recorded.

THE EQUATIONS

The Rittenberg and Foster equation is used to determine

the amount of IAA in the sample. It is of the form:

Y ((Ci/Cf)-l) X

Y = amount of endogenous IAA in the sample
C’ = initial specific activity
C = final specific activity
X = amount of (l4)C-IAA initially added
The utility of the first internal standard, (l4)C-IAA,
is seen in the above equation. The second internal stand—

ard, (l4)C-IBA, is used to find the final specific activity

of IAA:

Cf = (DPM IAA / DPM IBA)(moles IBA / moles IAA)IBA S.A.

where IBA S.A. is IBA Specific Activity

The initial specific activity of IBA is known and the
assumption is made that it is not diluted by IBA present in
the sample. It can, therefore, be used to obtain the final
specific activity of the IAA. The derivation of the equa-

tion is outlined in Appendix A.
ANALYSIS OF RESULTS

The ratios obtained from FID collections were averaged

 

54

and used to calculate endogenous IAA estimates for each peak
ratio obtained from an NSD sample injection. The three
estimates per sample were averaged and divided by number of
plants, fresh weigth and dry weight. Each set of means was
subjected to an analysis of variance in order to partition
out the portion of variance due to experimental error. An F
test was conducted to determine the significance of varietal

differences.

Table 3.2 - Analysis of variance table for levels of IAA per

shoot tip, per gram of dry weight, and per gram of fresh

weight.

""""""""" 111311313};—6ETHREE-"m"""""""""
SOURCE DF MEAN SQUARE EXPECTED MEAN SQUARES
1325"" "E31; """ (1.27?" QIQEESEJIQIZEEQEEZISQ
Variety v—1=l MS 2 V(error)+V(variety)
Error (v—1)(r-l)=2 MS 3 V(error)

F test = M8 2/MS 3

RESULTS AND DISCUSSION

The results obtained for nanograms of free IAA per
plant, per gram of dry weight, and per gram of fresh weight
are shown in Table 3.3. Although there is variability in
the data, several conclusions can be drawn from the numbers
presented. First, in each repetition of the experiment the
amount of IAA in the Single Stem variety exceeds that in

Seafarer. This is consistent over individual plants(shoot

55

tips), dry weight, and fresh weight.

Table 3.3 - Amount of IAA in each variety standardized over

shoot tip, dry weight, and fresh weight.

VARIETY REP IAA/SAMPLE IAA/SHOOT TIP IAA/DRY WT. IAA/FR. WT.

(09) (09) (09/9) (09/9)

EE;"" "1' ""SEZ"' ""‘1T§£6”" ”"I;;"" ""§3T;"‘
2 240 2.553 197 21.8
3 561 5.905 513 59.5
MEAN 375 3.426 303 34.7
58 1 691 6.061 410 49.1
2 471 3.595 283 33.4
3 1997 17.672 1154 97.4
MEAN 1053 9.109 616 60.0

The ratios of Single Stem IAA to Seafarer IAA for each
repetition are pre—sented in Table 3.4. In the raw data
(Table 3.3) the third repetition shows values quite a bit
higher than the other two repetitions. When the data are

converted to ratios, however, the third repetition does not

Table 3.4 - Ratio of IAA in Single Stem to IAA in Seafarer.

REPETITION IAA/SHOOT TIP IAA/DRY WT. IAA/FRESH WT.
""’I"‘-- --"3T336"-" _"'ETSE-" "”3TIE""
2 1.408 1.44 1.53
3 2.993 2.24 1.64

56
differ at all from the other two. The ratios are in fact
intermediate. There is, therefore, in the data a strong and
consistent suggestion that Single Stem has a greater amount
of free IAA in its shoot tips than Seafarer.

The analysis of variance(AOV) results for all three
sets of numbers are given in appendix B. The results of the
analysis are as follows: the difference between the two
varieties is non- significant for IAA per plant(shoot tip)
and IAA per gram of dry weigth. The probability of a larger
F by chace alone was .2. If there were no true differences
between the two varieties chance fluctuations would produce
an F value this large or larger about 20% of the time. For
IAA per gram of fresh weight, however, the difference was
significant at the .08 level.

The low significance of the F tests can be attributed
to two main causes. First, the extremely small number of
samples gave a low number(two) of error degrees of freedom.
Because of the low number of degrees of freedom the tabular
F values for such a test are exceedingly high. Thus, even
though the means show a two-fold difference in IAA levels
between the two varieties the low number of samples would
almost preclude a significance in the difference.

The second reason for the low significance of the
results is the much larger values obtained for the third
Single Stem repetition. These values caused a larger error
sum of squares and a smaller F. This is demonstrated by the

results of the AOV for IAA per gram of fresh weight. Due to

57
the larger fresh weight of this sample compared to the
others, the values were closer in magnitude to the others.
Reduction in the variabilility of this one value was suf—
ficient to produce significance at the 10% level in the
variety difference. As mentioned above, however, when taken
as a ratio these values do not differ from the other two
repetitions.

Some of the variability between replication means of
the same variety may be due to experimental error and due to
the fact that analysis of each repetition was done at a
different time(block effect). However, the major portion of
the variabilitity should be attributed to intrinsic differe-
nces in the plant material analysed. When the three repeti-
tions of IAA per shoot tip are regressed on the replication
means of variables measured on plants from the same experi-
ment very high correlations were obtained for both varieties
(Appendix C).

These correlations although not all individually signi-
ficant(because only three levels were used on each axis),
show a definite trend. The replications with the higher
levels of IAA per shoot tip were the ones with higher values
for many metric traits. The point here is that "bigger
plants"(as denoted by higher numbers of nodes, pods, ra-
cemes, stem growth, etc.) have more hormone. This is con—
sistent with the notion that hormone levels, more than
promoting a linear response in any one metric trait, main-

tain a constant proportionality between those traits(Went,

58
1956). A "bigger plant" needs a larger amount of hormone to
maintain this proportionality. This point is made only to
propose that the variability within replications may be
arising from intrinsic differences in the plant material
used, and not from experimetal error. It is suggested,
though, that higher numbers of repetitions would be desira-
ble to produce statistical significance as well the accepta-
ble "eyeball" results.

It was shown in chapter 2 that only a few traits chan—
ged in magnitude(a1though many shifted relative position on
the plant) due to decapitation. Removal of the shoot tip
affected only one morphological process, namely, branch
length(manifested also by changes in internode number and
length). Branch number did not change upon decapitation. If
IAA is the shoot tip compound responsible for the change in
branch growth then the results of the IAA assay clearly show
the following association: Seafarer has a lower amount of
IAA and a higher amount of branch growth. Single Stem has a
higher amount of IAA and a lower amount of branch growth.
This implies that IAA is either directly or indirectly
responsible for suppressing branch growth.

A direct effect is ruled out by the fact that Seafarer
has IAA in its shoot tip and yet when this shoot tip was
removed the growth of branches was not stimulated. If IAA
alone was directly responsible for the branch growth effect
then both of the varieties should have reacted similarly to

decapitation.

59

Indirect action of IAA could be of two types: either
by IAA in the shoot tip translocating to some other plant
organ to induce increased levels of GA or cytokinin(for
example) which would trigger increased branch growth, or, by
simply being a component in a larger "hormone balance"
picture, which when upset could cause increased growth, also
due to the action of other hormones(Shein and Jackson,
1972).

If the first type of indirect control is opperative we
should expect the same response from Seafarer and Single
Stem upon removal of IAA. The fact that the two varieties
reacted differently points to the following explanation:
Seafarer IAA levels are already low, so low in fact that
removal of the shoot tip IAA was not sufficient to upset the
hormone ratio significantly, as reflected by lack of branch
growth. In Single Stem the reverse is the case. Removal of
the high levels of IAA in the shoot tip did significantly

affect the hormone balance, allowing branch growth to take

place.

CHAPTER 4

GENETICS OF ARCHITECTURAL TRAITS

INTRODUCTION

The unique feature of architectural traits as a group
is that they exhibit continuous variation and, therefore,
are not amenable to explanation by simple Mendelian gene-
tics. With continuous variation we cannot distinguish indi—
vidually the genes contributing to a metrical trait. In such
cases our best hope for understanding the inheritance of
these traits is by using statistical methods of estimation
of gene action. Highly sophisticated methods have been
developed for this purpose, and some are illustrated in this
chapter.

Two types of genetic effects can be distinguished,
those due to additive effects of genes(the average effect of
a single gene substitution) and those due to non-additivi-
ty(dominance). Similarly, genetic variance may be parti—
tioned into a portion due to fitting a regression 1ine(or
due to the variation of expected values from their mean),
and a second portion due to deviation of the phenotype from
that regression line. The first type is termed additive
genetic variance and the latter is variance due to domi-

nance. This chapter reports upon the use of generation

60

61
means analysis, genetic variance components analysis, and
estimation of heterosis and heritabilities, to determine the
mode of inheritance of a large number of architectural
characters.

The objective of these studies was to obtain knowledge
of the type of gene action governing a trait, the degree of
phenotypic variation attributable to environmental and gene—
tic influences, and the expected behavior of the trait in
progeny generations. More succinctly, to gain any genetic
information which might assist the breeder in designing a
more efficient architype, and enable him, by more judicious

selection strategies, to achieve his breeding goal.
MATERIALS AND METHODS

A cross of Seafarer and Single Stem 791515(cross l) was
performed and F1, F2, and F3 generations obtained. Similar-
ly, a cross of Seafarer and architype 790780(cross 2) and a
cross of Single Stem and architype 790780(cross 3) were
performed. In the summer of 1981, parents, a random sample
of F2 seeds, and F3 families of each cross were planted in
the field in East Lansing. The experimental design was a
randomized block with four replications. Each 1.2 m row had
twelve seeds and rows were spaced 50 cm apart. F3 families
were planted in one—row plots, F2 plants were planted in
four row plots and parents were planted in two row plots.

Due to certain time limitations only the following

portions of the crosses were harvested: four replications

62
of the parents for all three crosses, four replications of
the F2 for cross 1 and three for cross 2, and three replica—
tions of F3 families for crosses 1 and 2. In cross 3 only
one replication of F2 plots and F3 family plots was col-
lected. Table 4.1 shows the number of plots per generation

which were planted(and collected) for each cross.

Table 4.1 — Number of plots per generation of each cross

which were planted and harvested.

CROSS PLOTS/REP REPS HARV. PLANTS
PAR F2 F3 PAR F2 F3 MEASURED

TQQQEQEQIJQISSIQ’QEQQ "3 '1' 3'7" "2 ’2 "3 "-1713?"

2 Seafarer*Architype 2 l 46 4 3 3 1,532

3 Single Stem*Architype 2 1 31 4 1 1 467

At physiological maturity 5-10 plants were harvested

per plot and the following variables measured:

(l)Number of branches

(2)Branch growth(sum, in cm)

(3)Height(cm)

(4)Nodes below 10 cm on the main stem

(5)Nodes above 10 cm on the main stem

(6)Hypocotyl diameter(in 1/32 inch)

(7)Racemes on the main stem

(8)Pods on the main stem

(9)Nodes on branches

(10)Racemes on branches

(11)Total pods

63
(12)Five pod length(cm)
(13)Yield(gm)
From these variables another set of variables was generated:
(14)Average branch length(cm)
(15)Nodes on the main stem
(16)Tota1 stem growth(branches + height)
(l7)Total nodes
(l8)Average internode length on branches(cm)
(l9)Average nodes/branch
(20)Total racemes
(21)Pods on branches
(22)Pods/raceme
(23)Average internode length on the main stem
(24)Average internode length

Variable 13(yie1d per plant) was only measured in cross 1.
STATISTICAL ANALYSES

Crosses 1 and 2 were used in performing analysis of
generation means, estimation of genetic variance components,
and estimation of heterosis and heritabilities. The F3
families of these two crosses were also used in a Principal
Components Analysis. Cross 3(F3 families) was used only in
the principal components analysis. In addition, cross 1 was
used correlation and multiple regression analysis. The re-
sults of principal components analyses, correlations, and

multiple regression are reported in Chapter 5.

 

64

GENERATION MEANS ANALYSIS

Crosses l and 2 were analysed as follows. An analysis
of variance was conducted on replication means of each of
the 24 variables. In variables where significant differen-
ces were found between generations, the generation sum of
squares was further partitioned into additive, dominance,
and residual components. The sources of variance are shown
in Table 4.2. Other variables(with non-significant genera-
tion differences) were discarded at this point.

Tests of the significance of the model and of additive,
dominance, and epistatic effects were made. Estimates of a
and d were obtained from a regression analysis and R2 was

used to assess the goodness of fit of the model.

Table 4.2 - Partitioning of generation effects into additive

and dominance effects.

SOURCE DF
Generation 3
Additive effect 1
Dominance effect 1
Residual 1
Error 11
Total 14
HETEROSIS

Heterosis for each trait was calculated by subtracting

the midparent(MP) value from the mean of the F2. Its signi-

65

ficance was tested by a t-test, t=H/s(h), where H is hetero-
sis and s(h) is its standard deviation. The variance of
heterosis(VH) was determined by the equation:

VH = VF2 + 1/4 VPl + 1/4 VP2
where VF2, VPl, and VP2 are the variances of the means for
the F2, parent 1, and parent 2, respectively. The degrees of
freedom used for the t-test was 51 for cross 1 and 67 for
cross 2, these being the n—l number of observations in the
parent with the least observations. This was a conservative
approach to selecting the degrees of freedom for the test.
Percent heterosis was found by:

%H = {(F2 - MP)/MP} * 100

Heterosis in the F1 equals h and in the F2 it equals 1/2 h.
I consider this estimate(using the F2) to be just as useful
to the breeder in the practical sense as would be a strict

F1 heterosis estimate.

GENETIC VARIANCE COMPONENTS ANALYSIS

Genetic variance components were obtained for crosses 1
and 2. The source of variation and expectation of mean
squares for a genetic study where parents and F3 families
are used are shown in Table 5.3.

The significance of family and plot mean squares was
tested by the following F tests:

F = MSl/MSZ (for testing difference in family means)

where the degrees of freedom for numerator and denominator

in cross 1 and 2 were 36 and 72, and 45 and 76, respectively.

66

F = MSB/MS4 (for testing differences within plots)

where the degrees of freedom for the numerator and the

denominator were 827 and 100, and 1185 and 127 in crosses 1

and 2, respectively.

Table 4.3 - Sources of variation and expectation of mean

squares for parents and F3 families.

SOURCE DF MS

F3 Families f—l MSl

Family*Rep (f—l)(r—1) M82

Within Families fr(n-l) MS3
Error

Within Plot 2r(n—l) M84

Error(Parents)

VEw + VF3 + cl VEb + c2 VF3

VEw + Tit-‘3 + c3 VEb

Because of unequal numbers of plants per plot all three

coefficients for the variance terms differ. They are esti-

mated using a method for unbalanced designs. First, a

number of k values are estimated such that:

k1= 2 Ni2 /N
k2= _ N.2/N
1 1
k3=ZZN../N.
1 j 13 1.
where i=1...37 or i=1...46

l and 2, respectively.

for F3 families in crosses

j=l...3 for replications

Next, a number of v values are estimated such that

v1=N-kl
v2=k3-k2
v3=N-k3
v4=a-1
v5=b-a

 

67
where a=37 or a=46, the number of replicated cells in
crosses l and 2, respectively;
b=110 or b=138, the number of unreplicated cells;

N=919 or N=1305, the total number of observations.

From these v values the coefficients cl, c2, and c3 are
found:

cl=v2/v4

c2=v1/v4

c3=v3/v5

The estimates of within F3 family variance, VF3, and
between F3 family variance, VF3, can then be found from the
equations:

VF3= MS3-MS4

VEb=(MSZ-MS3)/c3

VF3=(MSl-(M83 + c1 VEb))/c2

The genetic variance components, D, and H, can be esti-
mated algebraically from the classical relationships:

VF3=1/2 D+l/l6 H

VF3=1/4 D+l/8 H

by solving the two equations simultaneously for the unknowns.

Thus:
D = 2.67*VF3 — 1.33*VF3 and
H = 8*(VF3 - D)

HERITABILITY

Narrow sense heritability on a family basis was calcu-

lated by use of the following equation:

 

 

68
h2= (3/40) / (vF3 + MSZ/n)
where n=8.42 for cross 1 and 10 for cross 2, the average of

coefficients c1 and c3 from Table 4.3.

RESULTS AND DISCUSSION

GENERATION MEANS ANALYSIS

The overall generation means for each variable in cros-
ses 1 and 2 are presented in Tables 4.4 and 4.5. A compa-
rison of parental generation means and F2 means revealed
very few variables where complete additivity seemed indi-
cated. They were variables 3(height), 5(nodes above 10 cm
on the main stem), 6(hypocotyl diameter), 11(total pods),
12(pod length), 13(yield), and 22(pods per raceme) in cross
1. In cross 2 variables 6(hypocotyl diameter), 11(tota1
pods), 18(average internode length on branches), and 24(ave-
rage internode length) were indicated.

The remaining variables all seemed to indicate partial,
complete, or overdominant gene effects except for variables
1(number of branches) and 20(total racemes) in cross 1 where
parent means were not sufficiently different to permit any
inference. Significant and highly significant heterotic
effects for almost all remaing variables further pointed out
the existence of dominance. Since F2-MP=1/2h, two
times this quantity was added to the F2 and then a compari-
son was made to the parent in whose direction dominance was

indicated. Overdominance was apparent in almost every case.

69

Table 4.4 — Generation means and percent heterosis for each

variable(V) in cross 1(Seafarer*Single Stem).

V P1 P2 F2 F3 MP %Heterosis
1 "3:62 "EB; "3:3; "I; "IS; 7233""
2 117.07 142.69 238.77 200.16 129.88 83.84**
3 32.30 94.98 73.04 70.20 63.64 14.77

4 5.21 4.27 5.11 5.06 4.74 7.80*
5 3.18 11.65 7.78 7.72 7.41 4.99

6 7.86 12.08 9.76 9.41 9.97 -2.11

7 4.75 7.24 3.17 3.66 5.99 —47.08

8 10.07 12.57 7.31 7.96 11.32 -35.42**
9 16.58 19.51 30.51 27.49 18.04 69.12**
10 10.86 8.44 12.75 11.19 9.65 32.12**
11 32.08 24.44 29.34 27.01 28.26 3.82
12 35.92 46.44 40.92 40.68 41.18 —.63
13 24.36 30.79 28.33 25.98 27.57 2.76
14 23.96 28.13 43.56 38.31 26.04 67.28**
15 8.39 15.92 12.89 12.78 12.15 6.09*
16 149.26 235.38 310.96 271.10 192.32 61.69**
17 25.19 35.02 41.65 40.04 30.10 38.37**
18 7.07 7.13 8.04 7.89 7.10 13.30**
19 3.41 3.91 5.50 5.29 3.66 50.27**
20 15.44 15.43 15.86 14.85 15.44 2.72
21 21.96 11.57 21.68 18.92 16.76 29.35**
22 2.07 1.58 1.85 1.82 1.82 1.65
23 3.88 5.95 5.54 5.35 4.91 12.83**
24 5.92 6.72 7.47 6.77 6.32 18.20

Table 4.5 - Generation means and percent heterosis for each

variable(V) in cross 2(Seafarer*Architype 790780).

10

11

12

14

15

16

17

18

19

20

21

8.59
17.45
10.28
31.73
34.57
20.90

8.22

142.66

25.67

177.28

80.97
6.03
9.60

10.69
4.53

10.51

30.76

11.62

31.39

42.82

30.34

15.63

258.25

46.40

244.82

69.47
6.11
7.85
9.28
3.35
6.65

40.40

14.70

31.34

37.22

35.50

13. 96

314.29

54.36

226.91

72.40

6.88
30.41
12.67
29.45
39.35
36.40

12.96

299.48

43.37

145.23

55.22

24.10
10.95
31.56
38.69
25.62

11.92

200.45

36.03

68.57**
25.80**
9.30**
24.01**
2.77
-17.08**
-30.36**
67.63**
34.25**
-.01
3.80*
11.42*
17.11**
56.79**
50.87**
2.44
43.47**
20.41**
16.50
17.51

13.24**

significant at the 5% and 1% level, respectively.

 

72

In cross 1 dominance in the direction of P1(Seafarer)
was indicated for traits 4(nodes below 10 cm on the main
stem), 8(pods on the main stem), 10(racemes on branches),
and 21(pods on branches). Seafarer's dominance over Single
Stem caused an increase in nodes below 10 cm on the main
stem(probably due to a shorter hypocotyl), a decrease pods
on the main stem (probably due to loss in height), and an
increase in racemes and pods on branches. In cross 2,
Seafarer(Pl) was dominant in decreasing racemes and pods on
the main stem, and pod length. Overdominance was apparent
in all cases except for pod length in cross 2 where complete
dominance seemed to be the case.

Single Stem was dominant over Seafarer in cross 1
traits, where it increased branch growth, nodes on branches,
average branch length, nodes on the main stem, total stem
growth, total nodes, average internode length on branches,
average nodes per branch, and average internode length on
the main stem. For all these traits overdominance was
manifested.

Architype 790780 in cross 2 was dominant and had the
effect of increasing branch growth, height, nodes above and
below 10 cm on the main stem, nodes and racemes on branches,
average branch length, nodes on the main stem, total stem
growth, total nodes, average nodes per branch, total ra-
cemes, and average internode length on the main stem. It
was also dominant but decreased racemes and pods on the main

stem. In all cases overdominance was indicated.

 

73

Analysis of variance on generations showed significant
generation differences for all variables in cross 1 except
variables 1(number of branches), 18(average internode length
on branches), 20(total racemes), and 22(pods/raceme). These
results are depicted in Table 4.6. Table 4.7 shows the
results of analysis of variance on generations in cross 2.
Non-significant generation differences were found for the
same variables of the preceding cross and also for variables
4(nodes below 10 cm on the main stem), 7(racemes on main
stem), 10(racemes on branches), 11(total pods), 21(pods on
branches), and 24(average internode length).

In variables where significant generation differences
were found the generation sums of squares was further parti-
tioned into additive, dominance, and residual or epistatic
effects.

In cross 1 five variables were found to have only
significant additive effects(Table 4.8). They were variables
4 and 5(nodes below and above 10 cm on the main stem), 6(hypo-
cotyl diameter), 11(tota1 pods), and 12(five pod length).
The magnitude of the a and d coefficients and their standard
errors when a simple additive-dominance model was fit con—
firmed these results. The coefficients for the additive
effects had comparatively low standard errors and were sig-
nificant while for the dominance effects the reverse was

true(Table 4.9).

73

Analysis of variance on generations showed significant
generation differences for all variables in cross 1 except
variables 1(number of branches), 18(average internode length
on branches), 20(total racemes), and 22(pods/raceme). These
results are depicted in Table 4.6. Table 4.7 shows the
results of analysis of variance on generations in cross 2.
Non-significant generation differences were found for the
same variables of the preceding cross and also for variables
4(nodes below 10 cm on the main stem), 7(racemes on main
stem), 10(racemes on branches), 11(total pods), 21(pods on
branches), and 24(average internode length).

In variables where significant generation differences
were found the generation sums of squares was further parti—
tioned into additive, dominance, and residual or epistatic
effects.

In cross 1 five variables were found to have only
significant additive effects(Table 4.8). They were variables
4 and 5(nodes below and above 10 cm on the main stem), 6(hypo—
cotyl diameter), 11(total pods), and 12(five pod length).
The magnitude of the a and d coefficients and their standard
errors when a simple additive-dominance model was fit con-
firmed these results. The coefficients for the additive
effects had comparatively low standard errors and were sig—
nificant while for the dominance effects the reverse was

true(Tab1e 4.9).

 

 

74
Table 4.6 — Sums of squares from analysis of variance on
generation means over replications for cross 1. Experiment

analysed as completely randomized block design.

 

VARIABLE REPLICATION GENERATION ERROR
= 3 3 8

;_______ ""3333; ""1335; ""331"
2 2678.59ns '35454.20** 4760.61

3 240.34ns 8142.73** 168.29
4 1.23ns 2.12** .68
5 .95ns 144.15** 1.31

6 1.49ns 36.08** 2.03
7 .48ns 38.13** 11.77
8 .26ns 66.49* 26.30
9 35.83ns 489.92** 89.05
10 11.88ns 38.37* 16.49
11 66.89ns 132.53* 59.37
12 8.02ns 221.81** 9.46
14 87.79ns 785.76** 66.79
15 .27ns 115.3l** 2.00
16 2185.34ns 56185.79** 5722.79
17 31.36ns 750.80** 109.24
18 1.58ns .98ns 2.34
19 .35ns 10.38** .77
20 11.43ns 1.29ns 29.11
21 61.15ns 274.76** 41.38
22 .38ns .57ns .40
23 1.21* 10.81** .79
24 1.04ns 2.70* 1.33

 

ns' *, **

nonsignificant, significant at 5%

and 1% level.

 

 

75

Table 4.7 — Results of analysis of variance on generation
means over reps for cross 2. The experiment was analysed as
a completely randomized block design-

 

VARIABLE REPLICATION GENERATION ERROR
= 3 3 7
1m" "”7333; ""3333; ""33?
2 7825.56** 35069.19** 870.36
3 376.53ns 5813.17** 672.60
4 .27ns 1.87ns 3.41
5 3.47ns 86.40** 5.73
7 2.25ns 4.00 2.19
8 19.84ns 38.00* 15.45
9 146.82ns 682.50* 204.19
10 1.45ns 37.54ns 51.13
11 63.00ns 15.26ns 94.15
12 15.04ns 157.79** 12.07
14 116.52* 522.12** 45.82
15 4.46ns 107.99** 12.03
16 10408.44** 62976.28** 2071.07
17 172.89ns 1259.00** 294.19
18 13.81ns 3.59ns 9.77
19 2.76ns 9.31* 4.11
20 1.46ns 29.50ns 51.35
21 26.03ns 25.96ns 66.61
22 .37ns 1.14ns .85
23 1.53ns 8.63** 1.27
24 8.41* 3.76ns 4.50

 

ns' t, **

nonsignificant, significant at 5% and 1% level.

76

Six variables in cross 1 showed significant sums of
squares for dominance(Table 4.8). They were variables
2(branch growth), 9(nodes on branches), 10(racemes on bran-
ches), 19(average nodes per branch), and 24(average inter-
node lenght). Again the magnitude of the a and d coeffi—
cients confirmed these results. The a coefficients were not
significant since their standard errors were quite large by
comparison. The d coefficients had comparatively small
standard errors and were significant.

Eight variables had both significant additive and domi-
nance mean squares. They were variables 3(height), 7(racemes
on the main stem), 8(pods on the main stem), 15(nodes on the
main stem), 16(total stem growth), 17(tota1 nodes), 21(pods
on branches), and 23(average internode length on the main
stem). In six of these variables(7,15,l6,l7,21, and 23) the
magnitudes of the coefficients and their standard errors
were in agreement with the mean squares results. For
variable 3, however, only the coefficient for the additive
effects proved significant. Variable 8, by contrast, had
only a significant dominance coefficient. In cross 1 signi—
ficant epistatic effects were only found in variable 19(ave—
rage nodes per branch).

The R2 value for the simple model fit was high in
general and greater than .70 for 14 of the variables(Table

4.9).

 

77
Table 4.8 - Partitioning of generation sums of squares into
additive, dominance, and residual or epistatic effects, each

with one degree of freedom in cross 1.

VARIABLE a d residual
DF= I l 1 1

3------" ""ESIETEIS; ”33563311; MERGE?
3 7857.56** 255.69* 586.97ns
4 1.75** .39ns .04ns
5 143.57** .40ns .03ns
6 35.57** .Zlns .49ns
7 12.45** 24.04** 2.01ns
8 12.45* 48.30** 4.31ns
9 17.11ns 457.67** 24.13ns
10 11.66ns 26.25** .00ns
11 116.89** 1.85ns 7.49ns
12 221.55** .27ns .32ns
14 34.35ns 736.80** 11.00ns
15 113.55** 1.59* .15ns
l6 15592.89** 39939.56** 732.85ns
17 218.82** 513.33** 28.06ns
19 .46ns 9.43** .54*
21 205.64** 69.07* .43ns
23 8.87** 1.64* .lOns
24 .94ns l.72* .Olns
3?,”I:’17"1'133;};BEEI;ESE:”£1133121';332’QE‘EE'ESE'IQ’IEJEIT

 

 

78
Table 4.9 — Estimates of genetic parameters and standard

errors for architectural traits in cross 1.

V m st.dev. a st.dev. d st.dev. R2
2 'ISET'iT'SSTES ’3236""“;§3 22331177131 T53
3 63.91 18456 -3l.34** 9.29 19.34 9.19 .95
4 4.76 1.38 .47* .19 .76 .52 .53
5 7.43 2.15 4.23** 1.23 .76 .58 .98
6 9.90 2.86 -2.10** .64 -.57 .70 .90
7 5.86 1.73 -l.24* .52 —5.93* 2.15 .72
8 11.12 3.26 -1.25 .68 -8.41* 3.13 .66
9 18.52 5.47 —1.46 1.27 25.87* 8.53 .77
10 9.65 2.83 1.21 .65 6.20* 2.59 .57
11 28.00 8.16 3.82* 1.64 1.65 4.16 .46
12 41.13 11.88 —5.26 1.58 —.63 1.47 .93
14 .26 .08 —.02 .11 .33** .10 .82
15 12.20 3.53 -3.77** 1.09 1.53* .70 .98

16 196.11 57.34 -44.15* 15.86 241.76** 76.91 .87

17 30.72 8.95 -5.23* 1.98 27.40* 9.06 .82
19 .04 .01 -.00 .00 .04** .01 .86
21 16.88 4.97 5.07* 1.79 10.05* 4.57 .73
23 .05 .01 .01** .00 .02* .01 .82
24 .06 .02 -.00 .00 .02* .01 .52

All mean effects were found to be highly significant.
*, ** = significant at 5% and 1% level, respectively.

 

79
In cross 2(Table 4.10) six variables showed significance

for the additive mean squares only. They were variables
3(height), 5(nodes above 10 cm on the main stem), 6(hypoco-
tyl diameter), 12(five pods length), 23(average internode
length on the main stem) and 15(nodes on the main stem).

The estimates of genetic parameters and their standard er-
rors comfirmed these results in variables 3, 6, 12, and 23

by showing significance only for the additive coefficient

(Table 4.11). In variables 5 and 15 the dominance parame—

 

ters were also significant but their magnitudes were much
smaller by comparison with the additive coefficient.

Only one variable showed significance for the dominance
mean squares alone. This was variable 8(pods on the main
stem), and the estimates of genetic parameters confirmed
this result.

The remaining variables in cross 2 showed both additive
and dominance sums of squares that were significant and
these results were confirmed in every case by the estimates
of the genetic parameters and their respective standard
errors. The variables were: 2(branch growth), 9(nodes on
branches), 14(average branch length), 16(total stem growth),
17(total nodes), and 19(average nodes per branch).

In cross 2 variables 14(average branch length) and
23(average internode length on the main stem) had significant
epistatic effects.

The R2 value for the fit of the simple model was high

in general, the lowest value was .68 for variable 23.

 

80
Table 4.10 - Partitioning of generation mean squares into

additive, dominance, and residual or epistatic effects for

cross 2.
17312211131; """"""" Q """"""" 5""""'""§;;ESJQI"
DF= 1 1 l

_________ L__------_-__ -__-__-_-_--_ ---------__-
2 7750.75** 27510.30** 2495.06ns
3 4761.90** 491.95ns 348.84ns
5 78.75** 4.10ns 1.26ns
6 22.41** .05ns .08ns
8 7.24ns 30.96** 2.17ns
9 305.66** 411.13** .07ns
12 l49.90** 2.62ns 3.51ns
14 180.93** 320.89** 62.69*
15 97.30** 7.30ns .26ns
16 24663.09** 35359.86** 4709.78ns
17 747.88** 527.98** .6lns
19 6.40** 3.34* .00ns
23 4.79** 1.35ns 2.17*

 

 

81
Table 4.11 - Estimates of genetic parameters and their
respective standard errors, and R2 values for the simple

model fit in cross 2.

 

 

V m st.dev. a st.dev. d st.dev. R2
2- 148.95** 14.66 —31.13** 7.33 2.19** .29 :92
3 63.20** 8.30 —24.40** 4.15 .35ns .16 .84
5 7.26** .69 3.14** .34 .03* .01 .92
6 8.69** .32 -1.67** .16 .00ns .00 .93
8 9.30** 1.05 -.95ns .52 -.07** .02 .76
9 24.15** 3.57 -6.18** 1.79 .27** .07 .80
12 37.28** .94 -4.33** .47 —.03ns .01 .92
14 25.66** 2.60 -4.75** 1.29 .23** .05 .84
15 12.98** .89 —3.49** .44 .04* .02 .90
16 212.15** 21.18 -55.52** 10.59 2.54** .41 .90
17 37.13** 4.30 -9.67** 2.15 .31** .08 .83
19 4.16** .51 -.89** .25 .02* .01 .74
23 4.67** .48 -.77* .24 .02ns .01 .68

 

*, ** = significant at the 5% and 1% level, respectively.

 

82

HETEROSIS

Heterosis, variance of heterosis, and the corresponding
t values for crosses 1 and 2 are shown in Table 4.12. Per-
cent heterosis is shown in Tables 4.4 and 4.5 for crosses l
and 2, respectiyely. Crosses 1 and 2 had eleven variables
in common where heterosis was considered significant or
highly significant. They were: 2(branch growth), 3(height),
8(pods on the main stem), 9(nodes on branches), 10(racemes
on branches), 14(average branch lenght), 15(nodes on the
main stem), 16(total stem growth), 17(total nodes),
19(average nodes per branch), and 23(average internode
length on the main stem). The percent heterosis for these
traits ranged from 6.09 to 83.84% in cross 1 and from 9.30
to 68.57 in cross 2. For pods on the main stem heterosis
was in the negative direction in both crosses. Racemes on
the main stem also showed heterosis in the negative di—
rection. The loss of height and nodes on the main stem
could partially explain the decrease in racemes and pods on
the main stem. However, even though the main stem was
taller than the shortest parent(73 and 32 cm,respectively),
fewer racemes appeared on the progeny main stem than on the
parent with the lowest number of racemes(F2 had 3.17 and P1
had 4.75 racemes on the main stem).

In addition to these eleven common variables, cross 1
showed a significant 13.30% heterosis for variable 18(ave—

rage internode length on branches) and 29.35% for variable

 

83
21(pods on branches).

In cross 2 five additional variables showed significant
heterosis: variables 3(height) at 25.80%, 5(nodes above
10cm on the main stem) at 24.01%, 7(racemes on the main
stem) at 17.08%, 12(five pod length) at 3.80%, and 20(tota1
racemes) at 20.41%. The presence of such large amounts of
heterosis in many traits has several implications in a plant
breeding program. In the absence of epistatsis it indicates
the presence of dominance in the gene effects, either partial
or complete(Ghaderi and Lower, 1979). These effects are
"unfixable"(by selection) in a program where true breeding
lines are the goal (Mather, 1949). If a breeder uses se-
lection in ignorance of large heterotic effects the apparent
gain will vanish as populations approach homozigocity.
Perhaps the best approach here would be to advance genera—

tions until F3 or F4 and then make selections.

 

W.»

84

4.12 — Heterosis, variance of hets'-sis,

for significance of heterosis.

10
11
12
14
15
16
17
18
19

20

significant at the 5% and 1% level, respectively.

HETEROSIS

154.93**
.94
.37*
.37

-.21
-2.82
-4.01**
12.47**

3.10**

1.08

-.26
17.52**

.74*
118.64**
11.55**
.94**
l.84**
.42
4.92**
.03
.63**

1.15

n
i
ll

CROSS 1
VARIANCE

258.75
I

8.67

.02

.10

.10
293.90

4.35

.08

t

3.6
1.9

and t value

CROSS 2
HETEROSIS VARIANCE
""771" "-31"
99.59** 252.24
14.25** 10.69
.52** .03
1.52** .13
.25 .09
-.69** .04
-2.90** .26
l6.30** 8.14
3.75** 1.31
-.22 5.18
-1.47* .32
3.64* 3.13
2.04** .17
113.84** 288.13
18.33** 9.34
.16 .16
1.83** .15
3.06* 1.38
3.59 4.49
-.38 .00
.58** .04
.43 .07

.0
2.9

1.6

 

85

GENETIC VARIANCE COMPONENTS ANALYSIS

The mean squares for analyses of variance conducted on
parents and F3 families for all variables in crosses 1 and 2
are shown in Tables 4.13 and 4.14.

In cross llall variables except three showed highly
significant family differences when tested by the family*rep
interaction mean squares. The exceptions were variables
1(number of branches), 21(pods on branches) and 22(pods per
raceme). All but eight of the variables also showed signi— 1
ficant within plot(family) differences when tested by the
parental plot error. The variables with non-significant
within family differences where 1(number of branches),
4(nodes below 10 cm on the main stem), 6(hypocotyl diameter),
7(racemes on the main stem), 11(tota1 pods), 20(tota1 racemes),
and 21(pods on branches).

In cross 2(Tab1e 4.14) seven variables did not show
significant family differences. They were variables 1(number
of branches), 2(branch growth), 10(racemes on branches),
11(total pods), 20(tota1 racemes), 21(pods on branches), and
22(pods per raceme). These variables also did not show
significant within plot(family) mean squares, with the
exception of variable 2. Variable 7(racemes on the main
stem) did not show significant within—plot differences either.
The remaining variables all showed highly significant or

significant differences in both categories.

 

86

Table 4.13 — Mean squares for analysis of variance conducted

on parents(M4) and F3 families(Ml,M2,M3) in cross 1.

M1 M2 M3 M4
V FAMILY FAMILY*REP FAM ERROR PLOT ERROR
I" ' 72:33 " -1333 ""739 """ETBS
2 74084.16*F 29175.81 12069.35** 6529.61
3 88l7.27** 1411.43 561.55** 280.00
4 8.78** 2.21 .72 .67
5 124.38** 17.68 5.77** 3.16
6 1874.32** 674.17 354.15 450.20
7 25.94** 6.93 2.77 5.71
8 160.81** 33.51 14.24 19.51
9 2478.39** 752.85 265.60** 124.72
10 287.36** 110.75 55.17* 42.47
11 807.53** 241.36 162.08 187.82
12 202.74** 41.91 12.67** 8.68
14 2705.62** 615.93 268.59** 153.87
15 149.25** 20.58 6.89** 3.56
16 110220.95** 34163.42 13437.73** 7669.25
17 3354.32** 881.81 306.98** 141.95
18 44.48** 18.20 12.30** 3.03
19 73.10** 19.83 6.13** 2.57
20 10228.63** 107.01 58.25 55.14
21 677.49 498.35 145.39 131.34
22 3.37 2.22 .23** .15
23 18.72** 3.53 1.72** .88
24 19.90** 6.32 2.72** 1.27

 

Table 4.14 - Mean squares for analysis of variance conducted

on parents(M4) and F3 families(Ml,M2,M3)

.---—-—---~‘”-‘---—“~----’_—_-----—-—“-~—------’_--~*--n~—~-

10
11
12
14
15
16
17
18
19
20

21

22477.40;
7972.08**
10.91**
91.28**
26.50**
14.92**
72.78**
1746.02**
125.19
338.89
176.21**
1585.32**
12l.24**
91865.97**
2656.35**
45.72**
43.46**
125.10
336.35
1.10
19.14**
17.44**

M2

29030.06
1942.57
5.41
14.46
11.48
6.72
29.33
608.56
127.20
321.26
24.13
623.93
24.26
38429.93
776.93
25.77
11.49
130.33

299.73

M3

FAM ERROR

16110.30**

716.98**

12.74**
340.73**
65.87
221.49**
12.95**
267.39**
7.37**
18432.86**
387.85**
15.65**
5.93**
66.12
217.78
.17
1.79**

in cross 2.

M4
PLOT ERROR

5886.11

189.56

12.48
211.03

48.82

6717.57
254.35
6.10
2.88
54.17
177.52
.14

.63

_.-.-. III-l—F :-

88
The coefficients employed in extracting the variance
between and witnin F3 families are shown in Table 4.15

below.

Table 4.15 - Calculation of coefficients for expected mean

I

squares in analysis of genetic variance components of

crosses l and 2.

k v c
v ERBEE'TEEBEE’S BEBE-17%;"; ERBQQ'ETQBEEE
'1" "33:32 "33315 $333 123%}; "7525 "ESTES
2 8.72 10.41 310.19 456.85 24.82 28.35
3 318.91 467.26 600.09 837.74 8.22 9.86
4 36.00 45.00
5 73.00 85.00

Variance within and between F3 families and total genetic
variance for the F3 families is shown in Tables 4.16 and
4.17. Heritabilities on a family basis are also displayed
in these tables.

In cross 1, four variables had 100% of the variance
between families due to negative values for estimates of
within-family variance. These negative values were inter-
preted as zero variances since they are generaly attributed
to sampling error(Ghaderi and Lower, 1981). The four
variables were: 6(hypocoty1 diameter), 7(racemes on the main
stem), 8(pods on the main stem), and 11(tota1 pods). These

results were a confirmation of the mean squares results

 

89
where these variables showed no significant within plot(fa—
mily) differences. Variable 1(number of branches) showed
negative(zero) estimates of both within- and between—family
variance, as would be expected form the previous mean
squares results" Six variables had larger between-family
variance than within-family variance, a greater portion of
the genetic variance being generated by differences between
family means as opposed to segregation within those fami—
lies. These variables included 4(nodes below 10 cm on the
main stem), 5(nodes above 10cm on the main stem), 12(five
pod length), 15(nodes on the main stem), and 20(total ra-
cemes). Variables 2(branch growth), 10(racemes on bran-
ches), 16(tota1 stem growth), 17(total nodes), 18(average
internode length on branches), 19(average nodes per branch),
22(pods per racemes), and 24(average internode length) had
higher within—family variance while the remaining variables
had more or less equal portions of the genetic variance
distributed within and between families. Variables of this
type were: 3(height), 21(pods on branches), and 23(average
internode length on the main stem).

In cross 2(Tab1e 4.17), two variables showed only
between family variance due to negative values for within-
family variance. They were: 1(number of branches), and
4(nodes below 10 cm on the main stem). Negative values for
between- family variance were responsible for 100% of the

heritable variance being estimated as within-families in

90

Table 4.16 - Variance within and between and heritability of

F3 families for variables(V)

in cross 1.

 

10

11

12

14

15

16

17

18

19

20

21

22

23
24

 

 

VARIANCE

BETWEEN FAM W/IN FAM TOTAL
-.37 -1.73 -2.10
1777150 5539.74 7317.24
296.73 281.55 578.28
.26 .05 .31
4.28 2.61 6.89
47.73 —96.05 47.73
.76 -2.94 .76
5.09 -5.27 5.09
68.57 140.88 209.45
7.01 12.70 19.71
22.66 -25.74 22.66
6.42 3.99 10.41
83.52 114.72 198.24
5.16 3.33 8.49
3025.74 5768.48 8794.22
98.50 165.03 263.53
1.05 9.27 10.32
2.12 3.56 5.68
407.71 3.11 410.82
6.53 14.05 20.58
.04 .08 .12
.61 .84 1.45
.54 1.45 1.99

HERITABILITY

.67
.90

.94

.07

.78
.34
.92
.17

.16

.16

.37

 

Table 4.17 - Variance within and

F3 families of cross 2.

91

between and heritability of

10
11
12
14
15
16
17
18
19
20
21

-244C54
211.41
.19
2.70

.53

-.13
.52
5.35
33.54
3.40
1864.12

65.89
.69

1.12

10224.19
527.42

-.13

.84
.26
129.70
17.05
3.30
4.18
169.32
2.48
11715.29
133.50
9.55
3.05
11.95
40.26
.03

1.16

10224.19
738.83
.19

6.40

2.96

169.54
17.05
3.82
9.53
202.86
5.88
13579.41
199.39
10.24
4.17
11.95
41.47

.03

.70

.41

.62

.84

.74

92
variables 2 (branch growth), 10(racemes on branches), 20(to-
tal racemes), and 22(pods per raceme). One variable, 8(pods
on the main stem), had greater between-family variance
while two variables, 5(nodes above 10cm on the main stem),
12(five pod length), and 15(nodes on the main stem), had
approximately equally ditributed variances. The remaining
13 variables had greater variance within families than
between families. They were variables 3,6,7,9,ll,l4,16,l7,
18,19,21,23, and 24.

The additive and dominance components of genetic va-
riance for cross 1 and 2 are presented in Table 4.18. Many
of the variables had negative values for one or the other of
the components; these values were interpreted as zero. Cros-
ses l and 2 had five variables in common where only additive
genetic variance was shown. They were variables 1(number of
branches), 4(nodes below 10 cm on the main stem), 8(pods on
the main stem), 12(five pod length), and 15(nodes on the
main stem). Other variables containing only additive gene—
tic variance in cross 1 were: 3(height), 5(nodes above 10 cm
on the main stem), 6(hypocoty1 diameter), 7(racemes on the
main stem), 11(tota1 pods), and 20(total racemes).

Dominant genetic variance was found for four variables
common to both crosses. The variables were: 2(branch
growth), 9(nodes on branches), 18(average internode length
on branches), and 21(pods on branches). In cross 1 one
additional variable had only dominant genetic variance,

variable 24(average internode length). In cross 2 twelve

Table 4.18

- Estimates of

93

Additive(D) and

Genetic Variance Components for cross 1 and 2.

CROSS l

v ""6 """"""
I" """ITEI '—
2 -2621.93

3 417.81

4 .63

5 7.96

6 255.19

7 5.94

8 20.60

9 -4.29

10 1.83

11 94.74

12 11.83

14 70.42

15 9.35

16 406.64

17 43.51

18 -9.53

19 .93

20 1084.45

21 -1.25

22 0

23 .51

24 -.49

35195.44
-968.62

-2.94
-29.41
-l659.64
-4l.44
~124.08
528.87
41.47
-576.61
-43.32
104.79
-33.51
42894.70
439.96
84.60
9.56
-5413.91
62.25

.32

-27951.51
-137.00

.68

—135.64
5.78
-10604.14
-1.63
~10.86
-1.07
-16.56
-50.32
-.09
-.15

Dominance(H)

221655.75
2787.31

-3.92

-17.54
847.75
183.15
28.16
-27.00
1353.47
-19.04
99746.08
540.15
92.39
17.49
130.49

412.20

------------------—---—------------------------------’-‘-’--

94
variables had only dominant effects. They were 3(height),
6(hypocoty1 diameter), 7(racemes on the main stem), 10(ra-
cemes on branches), 11(total pods), 14(average branch
length), 16(total stem growth, 17(total nodes), 19(average
nodes per branch), 20(tota1 racemes), 22(pods per raceme),
and 23(average internode length on the main stem).

Variables which contained both additive and dominant
genetic variances were 10(racemes on branches), 14(average
branch length), 16(total stem growth), 17(total nodes),
19(average nodes per branch), 22(pods per raceme), and

23(average internode length on the main stem) in cross 1.
In cross 2, only variables 5(nodes above 10 cm on the main
stem) and 24(average internode length) were in this cate—
gory.

Because standard errors for the variance components
have not been calculated caution should be observed in the
conclusions drawn. Their reliability increases, however,
when viewed in conjunction with the results of generation
means analysis.

ON THE AGREEMENT BETWEEN RESULTS OF GENERATION MEANS
ANALYSIS AND GENETIC VARIANCE COMPONENTS ANALYSIS

A few traits were identified which exhibited similar
gene action for both crosses and where the generation means
analysis results were in agreement with the estimation of
genetic variance components. Pod length, for instance,

showed additive gene action in the generation means analysis

 

95
coupled with moderate additive variance.

Several cases where gene effects were not the same for
both crosses but for which there was agreement between the
two methods of analysis were also present. Examples in
cross 1 included height and hypocotyl diameter which showed
additive effects and high additive variances. Pods per plant
showed additive effects and moderate additive variances and
nodes above and below 10 cm had additive effects and low but
significant variances. Branch growth and nodes on branches 1
both exhibited dominant effects coupled with very large
dominant variances, while average internode length had low
dominant variance and dominant effects. Total stem growth
and total nodes had both significant a and d coefficients
and appreciable variances though dominance variance was
larger than the additive variance. For average internode
length on the main stem, both types of gene effects and
genetic variance were present but variances were small. In
cross 2, nodes above 10 cm was shown to have significant
effects in the additive and dominance categories and low
variances also in both categories.

In the remaining results, the mode of inheritance indi—
cated showed some disagreement. The relative merits of each
type of analysis must be considered. Generation means ana-
lysis has the advantage of less compounding of error due to
first degree statistics(Ghaderi and Lower, 1979). Its de-
triment is that dominant and additive gene effects are

subject to cancelation in the averaging process. Estimates

96
of genetic variance components, on the other hand, can be
influenced by the distribution of increasing and decreasing
alleles in the parents and the scale used for measurements.
(Ghaderi and Lower, 1981).

On the whole, the architectural complex of traits can
be assumed to be influenced by additivity and dominance(and
overdominance) in these crosses. An overwhelming majority
of the architectural traits analysed showed heterotic effects.
It is difficult to determine by simple examination of the
data if there are patterns or groups of traits for which a
particular type of gene action(indicating the underlying
genetic unity of the group) is more prevalent than another.
In trying to establish such patterns multivariate analyses
such as factor or principal components analysis would be
useful. Principal components analysis is employed with this
objective in Chapter 5.

Regarding breeding strategy and the proper use of
available genetic variance, the variables analysed herein
are far too numerous to be addressed on an individual
basis. Neither is it possible to extrapolate from these two
crosses to the behavior of the 23 characters in other cros-
ses. It must suffice to say that when significant amounts
of additive variance exists, breeding progress can be made
by selfing followed by selection. In the cases where high
amounts of dominance are present, little can be gained since
hybrids are not at present feasible in beans. It may be

possible, however, to breed for a desirable trait indirectly

97
through selection for a related trait. The degree of deve-
lopmental and genetic correlation between architectural
traits should be understood and the compensatory nature of
the relationships between these traits should be kept in

mind. I
HERITABILITIES

Estimates of narrow sense heritability on a family
basis were obtained for several variables and these are
shown in Tables 4.16 and 4.17. Heritability in crosses l
and 2 were: .90 and .70 for nodes above 10 cm on the main
stem, .94 and .41 for nodes below 10 cm on the main stem,
.78 and .84 for pod length, and .92 and .74 for nodes on the
main stem, respectively. In addition, heritability in cross
1 was .67 for height, .34 for average branch length, .17 for
total stem growth, .16 for total nodes, .16 for average
nodes per branch, and .37 for average internode length on
the main stem. In cross 2 number of branches had a herita-

bility of .32.

 

CHAPTER 5

MULTIVARIATE ANALYSES ON F3

INTRODUCTION

It is impossible for any researcher to examine a large
body of data such as the set of 23 architectural traits
measured in the preceding experiments and discover all the
common aspects between some traits as distinct from others,
or to arbitrarily group them into clusters having some
biological meaning. Techniques such as factor and principal
components analyses are extremely useful for such purposes.
Denis and Adams (1978) used factor analysis to identify
patterns of morphological traits associated with yield.
Adams and Wiersma (1978) employed principal components ana-
lysis to assess genetic distance of bean cultivars. In this
chapter the objective was to reduce the large set of archi-
tectural traits measured to a smaller set of factors or
clusters with a common foundation, either developmental or
genetic. Particularly, the intent was to discover if any
trends existed in the genetic information obtained form the
studies reported in Chapter 4.

Principal components analysis involves several steps,
(Adams and Wiersma, 1978). The initial data matrix consists

of measurements made on architectural traits for a set of

98

 

99

cultivars or families. The raw data matrix is standardized
by subtracting the mean and dividing by the standard devia-
tion. This transformed matrix is multiplied by its transpose
and divided by n—1(where n is the number of observations) to
obtain a matrix,of correlations, from which a set of eigen-
roots and eigenvectors are extracted. Multiplication of the
correlation matrix by these roots rotates the original axes
into a new set of orthogonal axes or principal components.
Each principal component accounts for a portion of the total
variance and represents an independent genetic contribution
to variance.

Multiple regression and correlation are also employed
in this chapter as a means of extracting further information
on the relationship of variables to one another and their

predictive value in a yield equation.
METHODS
PRINC I PAL COMPONENTS ANALYS IS

All available replications of F3 families of crosses 1,
2, and 3 were used in 3 individual runs of principal compo-
nents analysis. The materials and methods section of
Chapter 4 outlines the genotypes used in each cross, the
number of plants harvested per cross and the variables
measured. All measured and calculated variables excluding
yield were included in the raw data matrix, so that variables

had some degree of correlation with each other.

100
A SPSS(Statistical Program for the Social Sciences)(Nie et
al., 1975) factor program was used to perform the appro-
priate calculations. The objective of the analysis was to
reduce the 23 variables to a smaller group of variables or
factors which cpuld have some biological reason for cluster-
ing, and then to examine the patterns of clustering and the
relative importance of these factors in each population.

The rational for using the F3 population data was the
following: in a segregating population many associations
usually present are broken. These associations form a milieu
against whose background traits are expressed. It may modify
or influence their expression to a greater or lesser degree.
These associations are developed over long periods and during
the process of increasing homozygosity and natural selection.
In an early segregating population where new variability
has been generated and broken associations have not had a
chance to completely reform the influence of each trait on
the genotype should be easier to ascertain. The same rational
applies to the use of the F3 population in a multiple regres-
sion of yield and yield components on a set of independent
variables.

LINEAR CORRELATIONS AND MULTIPLE REGRESSION USING YIELD AND
ITS COMPONENTS

Replicated F3 families of cross 1 were used in linear
correlations and in a multiple regression procedure. The

correlation of each variable with yield and the two primary

101

components of yield were obtained. For the multiple regres-
sion the data matrix of independent variables consisted of
all measured and calculated variables excluding yield(varia-
ble 13) and total pods(variable 11). In some cases varia-
bles 8(pods on the main stem) and 21(pods on branches) were
also excluded. Three dependent variables used were: yield,
total pods, and pod weight. The latter two are primary
components of yield in the relationship:

(pods/plant) * (weight/pod) = (weight/plant)
Pod weight is a product of (seeds/pod) * (weight/seed) and
was used because the two variables above were not available.
A "best fit" model was discovered by using Maximum R2
Selection technique (Barr et al., 1976). The objective of
the first procedure was simply to determine the relative
importance of each measured variable to yield and the prima—
ry yield components. The objective of the second procedure
was to determine the secondary yield components most closely
associated with and predictive of yield and each primary
yield component. This knowledge is important in determining

overall breeding strategy for architecture.

RESULTS

PRINCIPAL COMPONENTS ANALYSIS

In cross 1 seven factors were identified and these ac—
counted for 88.3% of the variance. In cross 2 five factors

emerged accounting for 81.8% of the variance, and in cross 3

102
six factors were found for which the cummulative percentage

of variance amounted to 83.3.(Table 5.1).

Table 5.1 — Percentage of variance accounted for by each

factor in crosses 1, 2, and 3.

.............. qb———-——--_-__-_____-—---___________-__-___-_-
FACTOR PERCENT OF VARIANCE
CROSS 1 CROSS 2 CROSS 3

1 38.4 40.4 39.8

2 12.8 14.7 14.0

3 12.3 11.3 11.2

4 8.9 9.3 7.0

5 5.7 6.0 5.9

6 5.6 5.4

7 4.6
CUMULATIVE 88.3 81.8 83.3

The variable loadings in the principal components or
factors for each cross are shown in Tables 5.2—5.4. The
results are summarized in Table 5.5, and are as follows:

Factor 1 - This factor was very similar for all three
crosses. In cross 1, 2, and 3 it included the following
variables: branch growth, nodes on branches, racemes on
branches, pods on branches, and four totals, total stem
growth, total nodes, total racemes, and total pods. In
crosses 2 and 3 number of branches was added to this factor
and in cross 1 and 2 hypocotyl diameter was also present.
This factor appears to be dominated by branch traits.
Within it are found both vegetative and reproductive traits
associated with branches, and also the totals, in which
branch contribution is the major percentage. This indicates

the developmental unity of the traits and the basic gene set

 

103

Table 5.2 - Grouping of 23 variables from

cross 1(Sea*SS)

into seven factors using principal components analysis.

v FACTOR: 1 2 3 4 5 6 7

I 7363- :62? 7622’ :15?" 3332; :63? 3:663-
2 .802* .196 .236 -.251 .103 -.023 -.019
3 .132 .891* .067 .066 .117 .294 -.087
4 .208 .272 —.191 .158 -.131 -.786* .159
5 .242 .898* -.096 .060 .114 .043 -.051
6 .738* .186 .015 .180 -.042 .118 .231
7 -.018 .150 -.037 .920* .005 -.023 -.120
8 -.021 .010 -.049 .924* —.004 -.094 .213
9 .867* .242 -.231 —.210 .133 —.047 -.055
10 .897* .043 .049 -.023 -.004 -.028 -.316
11 ‘ .877* .006 -.031 .325 -.005 -.069 .165
12 .187 .237 .091 .024 -.131 .604* .299
14 .515 .247 .240 -.116 .724* .058 -.002
15 .282 .902* —.115 .102 .066 -.192 .000
16 .826* .378 .231 -.214 .119 .044 -.037
17 .837* .374 -.229 -.174 .133 -.076 -.050
18 .002 .096 .946* -.018 -.011 .017 —.031
19 .539 .279 -.278 -.097 .691* -.001 -.025
20 .880* .078 .039 .196 -.003 -.034 -.340
21 .933* .003 —.015 .013 —.003 -.039 .098
22 -.099 -.128 —.067 .071 .001 —.025 .907*
23 -.091 .499 .204 -.031 .111 .666* -.119
24 .039 .020 .953* -.074 -.019 .244 -.043

* = indicates variables highly associated with the factor.

“ii-—

 

104

Table 5.3 - Grouping of 23 variables from cross 2(Sea*
Architype) into five factors using principal components
analysis.
G“”EXEE~6§I””I"""""E"""’S""'""ATM—"'3 """"
"I ""3362? "3333' ""33? "7633" "3:163-
2 [.834* .359 .264 -.198 .091
3 .093 .943* .080 .094 -.043
4 .233 .070 -.412 .284 -.321
5 .137 .888* -.174 .030 -.096
6 .720* .038 -.161 .278 .265
7 .018 .112 -.023 .869* .060
8 .054 -.085 .038 .877* .222
9 .834* .406 -.133 —.226 .114
10 .923* .152 -.004 -.031 -.195
11 .893* .036 -.005 .283 .234
12 .187 .257 .037 .161 .472
14 .429 .621* .203 -.259 .267
15 .120 .832* -.290 .118 -.190
16 .758* .543 .252 -.152 .070
17 .786* .525 -.l76 -.l79 .065
18 .042 -.l33 .908* .031 -.020
19 .390 .638* -.355 -.260 .263
20 .906* .174 -.009 .176 -.175
21 .929* .062 -.016 .048 .184
22 -.022 -.234 .018 .170 .824*
23 -.036 .765* .373 .035 .064
24 .161 .143 .954* .028 .027

* = indicates variables highly associated with the factor.

105
Table 5.4 - Grouping of 23 variables in cross 3(SS*Architype

into six factors using principle components analysis.

V FACTOR: 1 2 3 4 5 6
’1 ":33; "T632- "T362- "3331' ":63? "3322-
2 "713* .570 .329 -.083 .064 .066
3 .209 .146 .682* .318 .543 .051
4 .127 -.016 .300 .014 -.752* .024
5 .348 .073 .836* .151 .024 .043
6 .135 .779* -.050 -.027 -.233 .041
7 .047 -.019 .074 .901* .172 .108
8 .119 .035 .113 .903* .063 -.108
9 .851* .059 .357 -.155 -.120 .217
10 .926* .047 .136 .103 -.009 .093
11 .881* .057 .124 .355 -.021 -.215
12 .115 .175 .108 .230 .180 -.l48
14 .436 .772* .039 .158 .053 .303
15 .360 .060 .865* .142 -.240 .047
16 .688* .546 .408 -.127 .143 .068
17 .817* .063 .462 -.114 -.147 .200
18 -.167 .787* .062 -.003 .063 -.108
19 .577 .037 -.003 .120 -.200 .614
20 .878* .040 .144 .309 .031 .112
21 .937* .052 .102 .106 -.044 -.204
22 .053 .016 -.017 .063 -.052 -.787*
23 -.060 .083 .145 .294 .874* .017
24 .041 .897* .090 .065 .330 -.113

indicates variables highly associated with the factor.

 

106
from which the factor arises. This is strikingly borne out
by the fact that most traits in this factor are governed by
dominant gene action. So, in addition to being a "branch
factor" this appears to be a ”dominant gene action factor."
Table 5.5 showslthe genetic action connected with each
factor. As shown in Chapter 2 this is the conglomerate of
traits for which an antagonism or tradeoff between main stem
and branches was identified. Therefore, the hypothesis may
be made that this is the factor upon which apical dominance
exerts its influence, and could in fact, be named the "IAA
factor." IAA could be affecting many traits through a
cascade of secondary messages.

Factor 2 — In crosses 1 and 2, height, nodes above 10
cm on the main stem, and nodes on the main stem(total) were
in common. Average branch length was present in crosses 2
and 3. In cross 2, average nodes per branch and internode
length on the main stem were also part of factor 2. In
cross 3, internode length and internode length on branches
were also present. Factor 3 - In crosses l and 2, internode
length and internode length on branches were the only compo—
nents of the factor. In cross 3 the factor was composed of
height, nodes above 10 cm on the main stem, and nodes on the
main stem(total).

From the above, two more clusters of traits can be
identified. A cluster of main stem vegetative traits is
found in factor 2 for crosses l and 2 and in factor 3 for

cross 3. It is uncany that all the traits in this cluster

 

 

 

107
exhibited mostly additive gene action. It is strong valida-
tion that a fundamental genetic unity exists for the clus-
ter. An internode length factor was identified and found to
be governed mostly by dominance.

Factor 4 -7This factor was identical for all three
crosses and was composed of racemes and pods on the main
stem. This is clearly a "reproductive traits of the main
stem factor," the counterpart of the "vegetative main stem
factor" already discussed. The separation of main stem
traits into two distinct factors suggests the existence of
different genetic systems governing their expression. The
fact that they exhibit different types of gene action (ad-
ditivity vs. dominance) supports this conclusion.

The remaining factors account for a much smaller per—
centage of the variance and show no easily identifiable
pattern.

Further interpretation of the principal components
analysis results was based on the differences in clustering‘
pattern between the three crosses. A sort of "triangle
approach" was used whereby unique dissimilarities of cross 1
from crosses 2 and 3 were interpreted as relative effects of
the parent common to the two latter crosses and uniquely
absent from cross 1. An attempt was made to understand the
genotypic effect on the trait in the context of this ”trian-
gular' set of crosses. The placing of a trait into a factor
indicates that that trait is correlated with the axis

represented by that factor. The initial factor accounts for

 

108

Table 5.5 - Architectural traits clustered in each factor or

principal component(PC) for crosses 1, 2, and 3. The type of
gen action for each trait is also shown.

CROSS 1 CROSS 2 CROSS 3
SEA/SS SEA/ARCHITYPE SS/ARCHITYPE
PCI
I
br gr h,H #br D #br
hyp d dh,H br gr dh,H br gr
nodes br h,H hyp d d,H nodes br
racemes br h,DH nodes br dh,H racemes br
pods br dh,H racemes br H pods br
stem gr DH pods br H stem gr
nodes DH stem gr dh,H nodes
racemes D nodes dh,H racemes
pods D racemes H pods
pods H
PC II
height d,D height d,H internode lbr
nodes d,D nodes dh,DH internode 1
above 10 cm above 10 cm
nodes ms dh,D nodes ms dh,D hyp d.
ave. br 1. dh,H ave. br 1.
PC III
internode 1 br H internode 1 br H height
internode l h,H internode l H nodes above
10 cm
nodes ms
PC IV -
racemes ms dh,D racemes ms H racemes ms
pods ms h,D pods ms h,D pods ms
d additive gene effects, h dominant gene effects

D = additive genetic variance H = dominant genetic variance

109
the largest portion of the variance and each subsequent
factor attempts to identify an orthogonal axis accounting
for another portion of the remaining covariation. The
”triangle" is depicted in Figure 5.1, and genotypic effects
on the position of traits is indicated.

The fact that number of branches appears in factor 1 in
both crosses where architype was present but is relegated to
the fifth factor when architype is not one of the parents
suggests that architype influences the covariation involving
this trait, when used in these crosses. By the same token,
when architype was one of the parents average branch length
was positioned in factor 2, as opposed to factor 5 when it
was not.

Seafarer, by the same criterion, appears to have in-
creased slightly the covariance of main stem characters when
used in a cross, so that they were associated with the
second orthogonal axis instead of the third. These cha—
racters include height of the main stem, nodes above 10 cm
on the main stem, and total nodes on the main stem and were
positioned in factor 2 when Seafarer was present or factor 3
when it was not. Hypocotyl diameter followed a similar
pattern of greater prominence (factor 1) when Seafarer was a
parent and relatively less importance(factor 2) when it was
not. Seafarer's caused internode length to be found in
factor 3 in Seafarer crosses and factor 2 in cross 3.

Single Stem influenced the position of average nodes per

branch and internode length on the main stem. These traits

110

«A 53386 5.5:. 163.:
53386 .253 26:98 283 0093 .0 2:. w
a 13682. $6322 mMDflbmmw 288 an... «33 .

     

   

060
9
9
.»

  

030mm u

min-um mam: . w Dania—{um
« >533 333333: 2:33.. 03:33 «
o 5333. .263 3a.: :2: >833 03:2. .2622

meCHm muH. mmmmon om nSHmm com: ocHnH<mHm on «Sm mmonOH
wOmHnHon om mmHmnnma mHorwnmnnCHmH nnmwnm. mm
95mmHHma mnoa vnwnowpmw cosponmbnm mmonou
msmwwmwm. A¢u Hdnflmmmm. mam @ u amonmmmm HS
powwnwod. Swn: mmonow H mm nrm now mam mmonou
N Na nsm gonnoa om n50 momwmv.

111
were found in factor 2 in the cross where Single Stem was
absent and in factors 5 and 6 when it was present.

In all other cases the three genotypes appeared to be
essentialy similar in their effects upon the positioning of
the traits. I

The usefulness of principal components analysis for
reducing a large number of variables into more easily under—
standable clusters cannot be disputed. The biological mean-
ing attributed to the clusters obtained here is strengthened
by the remarkable consistency of clustering patterns over
all three crosses, and by corroboration with results of gene
action investigations. When the principal components method
is combined with another sort of analysis, such as the
genetic studies, it becomes an extremely powerful analytical

tool.

PHENOTYPIC CORRELATIONS

All variables, with the exception of three, showed
positive and significant correlations with yield. They were
internode length on branches and main stem, and average
internode length. Among the most highly correlated with
yield were variables 11(total pods), 21(pods on branches),
22(pods per raceme), 10(racemes on branches), 6(hypocotyl
diameter), 16(total stem growth), and 2(branch growth) in
that order. The magnitude of these correlations ranged from
.621 to .860.

Most variables were significantly correlated with pods

 

112
per plant but the magnitudes of these correlations were more
polarized than in the previous case, most being either
reasonably high or quite low. This demonstration of greater
specificity in the correlations is a natural shift. Since
pods per plant ds only a component of yield one would expect
to see some variables with high correlations and others with
low correlations because they are more closely associated
with other yield components. Among the highest correlations
were variables 21(pods on branches), 22(pods per raceme),
20(total racemes), and 10(racemes on branches). Next were
9(nodes on branches), 17(total nodes), 2(branch growth),
6(hypocotyl diameter), and 16(total stem growth). Correla-
tions with weight of seed per pod were very low and only
half of them were significant. The highest correlations
were variables 12(five pod length), and 3(height) with r
values of .383 and .214, respectively. The presence of so
little correlation can be attributed to the lack of varia—
tion in the seed weight per pod character. When we consider
that the cross made was between two small seeded lines
great variability could not have been expected.

The importance of phenotypic correlations, and especial-
ly those ones arising from negative associations between
architectural traits, cannot be overemphacized in an
architecture breeding scheme. As mentioned previously in
Chapter 3, compensation is a widespread phenomena in bean

plant development. It occurs not only among the primary

 

113

Table 5.6 - Linear correlations of architectural traits with
yield and yield components pods per plant(X), and weight of

seed per pod(YZ).

VARIABLE YIELD(W)
""1"" ""3163"--
2 .621**
3 I.236**
4 .215**
5 .319**
6 .633**
7 .208**
8 .288**
9 .593**
10 .656**
11 .860**
12 .268**
14 .420**
15 .357**
16 .622**
17 .599**
18 .030
19 .385**
20 .695**
21 .807**
22 .679**
23 .032
24 .069
YZ .424**

.059
.040
-.003
-.008
-.009
.383**
.152**

.144**

t, *t g significant at the

5% and 1% level,

respectively.

114
yield components as suggested by Adams (1967) and (1982) but
among many architectural traits. Genotypic correlations are
reflective of genetic linkage among traits. Phenotypic
correlations in addition to including a genetic component
are indicators pf environmental influences and the develop—
mental interplay of architectural traits. Two fundamental
rules must be observed in this type of breeding. The first
is that a desirable trait cannot be selected without consi—
deration of its other partners in a compensatory relation-
ship. If this is neglected an improvement in "source" or
"sink" may not be taking place but merely an adjustment in
the "system", as alluded to by Weiss (1939). Secondly,
since correlations are subject to change with the incident
stress provided by the environment(decapitation was an exam-
ple shown in Chapter 3)it is extremely important for archi—
tectural breeding to be conducted over several environments,
lest the breeder be biased into selecting an artificial

improvement.
MULTIPLE REGRESSION OF YIELD AND YIELD COMPONENTS

The model selected for yield when all variables inclu—
ding the primary yield components were in the input matrix
was expected. It included only the primary yield compo-
nents, pods per plant and seed weight per pod(or pod weight),
which gave an R2 of .926. These results are depicted in
Table 5.7 The F values shown can be used to compare the

importance of variables in the model relative to each other.

 

115

The partial regression coefficients should not, however, be
compared in this sense, since they are not standardized.

When primary yield components were excluded from varia-
bles going into the model, the best model selected included
seven characters and gave an R2 value of .68.(Table 5.8)
The variables included were number of branches, hypocotyl
diameter, branch growth, racemes on branches, pod length,
total racemes, and pods per raceme, in increasing order of
importance(according to the magnitude of the F values).
Number of branches and racemes on branches showed negative
partial regression coefficients, though total racemes was
positively loaded.

When number of pods was regressed on the other archi-

tectural traits (seed weight per plant and pods on the main

Table 5.7 - Maximum R2 improvement model selected for depen-
dent variable yield when primary yield components were in—

cluded in the set of independent variables.

VARIABLE PARTIAL REG. COEF. (b) F
$325252 """" 11322 """""""
Pods per plant .94 8999.99
Seed weight per pod 20.79 2255.48
SOURCE DF MEAN SQUARE F
EEQEEQEESS "E "'SSSQSTSE 3313:3911
Error 897 17.30

 

 

116
Table 5.8 — Maximum R2 improvement model selected for
dependent variable yield when primary yield components were

excluded from the set of independent variables.

VARIABLE PARTIAL REG. COEF.(b) F
$325225; , """"" 33E}; """"""
Number of branches -.35 6.77
Branch growth .03 53.71
Hypocotyl diameter .09 18.41
Racemes on branches -1.09 40.93
Five pod length .46 49.47
Total racemes 2.09 177.84
Pods per raceme 6.09 201.96
SOURCE DF MEAN SQUARE F
£25255; "3 WEEKS? 3363311
Error 892 75.22

117
Table 5.9 — Maximum R2 improvement model selected for depen-
dent variable pods per plant when seed weight per pod and
pods on the main stem and branches were excluded from the

set of independent variables.

 

 

VARIABLE I PARTIAL REG. COEF. (b) F
£32522; 41.42 """"
Branch growth .02 105.93
Racemes on branches -l.00 88.75
Total racemes 2.24 543.60
Pods per raceme 7.94 852.25
Average internode length -.64 40.38
SOURCE DF MEAN SQUARE F
;;;;;;;;;; "E "'SSEEETE 35332;;
Error 931 32.80

R2 = .832

 

stem and branches were excluded) a model including 5 charac—
ters with an R2 value of .832 emerged.(Table 5.9).

Branch growth, total racemes, and pods per raceme were
most important(as indicated by their F values) and were
positively loaded in the equation. Racemes on branches and
average internode length were smaller in importance and
negatively loaded in the prediction equation. It follows
naturally that with shorter internode length the more nodes
are possible, providing more loci for pod-bearing racemes.

Regarding the negatively loaded branch racemes, this could

 

118
be interpreted two ways: either one should have more racemes
on the main stem and fewer on branches, or attention should
be given to increasing number of racemes on branches, since
branch growth seems to be positively associated with yield.
Kueneman (1978)” in a study of effects of plant density on
yield stated that one of the most successful plant types
under increased competition had most of its pods and racemes
on the main stem and thus suffered less yield reduction due
to decreased number of branches.

Attempts to identify a best fit model for seed weight
per pod were unsuccesful.(Table 5.10). The R2 value was
.201. This was perhaps an indication that pod weight is much
more closely associated with the pod number traits excluded
from the input variables. Or, perhaps as implied by the low

correlations the variability in the trait was too small.

19

 

119
Table 5.10 - Maximum R2 improvement model selected for de-
pendent variable seed weight per pod when primary yield com-

ponents were excluded from the set of independent variables.

VARIABLE PARTIAL REG. COEF.(b) F

$322223; , """WTES """"""""

Number of branches -.03 34.28

Hypocotyl diameter .0008 1.73ns

Racemes on branches -.003 1.65ns

Five pod length .02 113.74 E
Nodes on the main stem .004 2.05ns 5
Total stem growth .0005 15.95

Average nodes per branch -.02 16.63

Pods per raceme -.04 8.58

SOURCE DF MEAN SQUARE F

EEQZEQQISE "é """ZES' 353511

Error 891 .08

SUMMARY AND CONC LUS I ONS

The major findings of this thesis can be summarized as

follows:

1. A high number of significant correlations was found
between architectural traits and yield and its components.
2. A high degree of developmental inter—relationships
was found between architectural traits, as demonstrated by
the many cases of component compensation in the decapitation
study. Both linear and multiplicative component compensa-
tion systems were identified where increase in one variable
was coupled with decrease in another, creating a homeostatic
stability in their sum or product. Two such examples are
the racemes on branches vs. racemes on the main stem rela—
tionship, and the branch number vs. internode number vs.
internode length relationship. The sum and product in these
relationships are total racemes and total branch growth,
respectively. The total "potential" for production of a
particular architectural trait seems to be determined by the
genetic code, but the placing of these structures on the
plant does not appear to be pre—established. An antagonism
between placement on the main stem or branches exists. The
main stem is dominant over branches in receiving a certain

percentage of nodes, racemes, and pods. When a dominant

120

 

121
central axis is not present, however, the branches compensate
for its absence by producing those structures. This compen—
sation ability was shared by both Single Stem and Seafarer
in most cases, with the notable exception being total stem
growth(the sum of height of the main stem and branch
growth). I
3. The difference in branching pattern and ability to
compensate in the total stem growth sum was shown to be
attributable at least in part to a shoot tip hormonal in—
fluence. Single Stem had high levels of IAA in its shoot
tip and less branch growth. Upon removal of the shoot tip
it compensated for loss of main stem growth by increased
branch growth. Seafarer had low levels of IAA concomitant
with much greater degree of branching. It did not increase
branch growth to compensate for loss of main stem, showing
that little shoot tip inhibition had been present in the
intact plant.
4. Both additive and dominant types of gene action were
identified for the complex of architectural traits. Addi-
tive gene action can be utilized by selection following
selfing. An overwhelming majority of architectural traits
showed heterotic effects, leading to the conclusion that F3
or F4 generation selection and several selection environ-
ments may be advisable. Proper regard should be given to
the compensatory nature of architectural traits when target-
ting any particular one for enhancement through selection.

It would be of interest to discover, by assaying the F1

 

122

and segregating generations, if the presence of heterosis in
the architectural complex is due to an increased or de—
creased hormone content in the progeny, relative to the
parents. To my knowledge such physiological assays have not
been conducted on genetic material and very little is known
on the inheritahce of hormonal characteristics.
5. Principal components analysis identified several clus—
ters of architectural traits, for which both developmental
an genetic explanations were found. The first was a branch
traits factor governed mostly by dominant gene action. The
second was a main stem(vegetative) factor dominated by addi—
tive inheritance. Two less prominent factors were: main
stem(reproductive), governed by both dominance and additivi-
ty, and internode length, with mostly dominant gene action.

The first factor, composed mostly of branch traits, is
affected by released from apical dominance. This has led to
the hypothesis that it is an IAA(indole-3-acetic acid) factor.
It is entirely possible that groups of architectural traits
are influenced by one hormone and its network of secondary
messengers and effects. Thus, the branch factor could be an
IAA factor, while the main stem vegetative factor could be a
GA(gibberelic acid) factor, the main stem reproductive factor
a cytokinin factor, and so on. Future genetic-hormonal

work must investigate this hypothesis.

Breeding strategy for improvement of yield through

architecture must include consideration of component compen-

123
sation, heterosis, and the existence of additivity. The
fact that large numbers of architectural traits have domi-
nant gene action may present a problem. However, some

traits, when broken into components, may present one or more

of these which had additivity and can therefore be enhanced
through selectibn. An example is total pods which is gover-
ned by both types of action. The pods on the main stem
component of this trait exhibits additivity, while pods on
branches is governed by dominance. It would seem then that
direct selection for greater numbers of pods on the main
stem instead of pods in general would be a more judicious
effort.

Branches exist with the function of bearing nodes,
racemes, and pods. All these traits are involved in linear
compensatory relationships with their counterparts on the
main stem. To the degree that these structures can be
transfered to, or increased on the main stem at the expense
of branches, then branches would seem to be expendable.
Reduced numbers of branches might provide for less inter-
plant competition, allowing sowing density to be increased

to obtain higher grain yield.

APPENDIX

 

 

124

Appendix A

The derivation of the equation can be illustrated by
analogy to the simple case where one wishes to dicover the
amount(X) of 10M HCl needed to make up 100 ml of a 5M HCl

I
solution.

10M (X) = SM (100 m1)

X SM (100 m1) / 10M

X 50 ml i

Therefore 50 ml of 10M HCl must be used in making up
the desired solution.

In order to obtain the amount of endogenous IAA the
same principle is applied. If the initial(Ci) and fina1(Cf)
specific activity of the IAA are known, and likewise the
amount(X) of IAA added, the amount of IAA in the sample can

be obtained from the equation:

Ci (X) = Cf (Y + X)
Ci/Cf = (Y +1X) / X
Ci/Cf = (Y/X) + 1

Y = ((Ci/Cf) - 1) X

The analogy is made between initial and final molarity
and the initial and final specific activity. The amount of
IAA is analogous to the mililiters of HCl with the differ-
ence that it is split up into two components, X and Y. By
defining these two components the IAA initially added can be

separated from the IAA present in the sample.

 

125

Appendix B — Analysis of variance for amount of IAA per

shoot tip, per gram dry weight, and per gram fresh weight.

 

NG IAA/SHOOT TIP

SOURCE MEAN SQUARE
REP 7 46.09
VARIETY 48.45
ERROR 15.15

NG IAA/G DRY WEIGHT

SOURCE MEAN SQUARE
REP 212081.17
VARIETY 146640.67
ERROR 42379.17

NG IAA/G FRESH WEIGHT

SOURCE MEAN SQUARE
REP 1487.23
VARIETY 957.60

ERROR 86.86

 

Ilv

126

Appendix C - Correlation between values of metric traits and

amount of IAA per shoot tip. (Three replications were used
as levels of IAA and correlated with the corresponding

replication means of measurements made on architectural

traits). ,

CREE; """"""""""" SEAss """""
r %PROB. r %PROB

nodes on branches --:995- --6:57 ---------------

racemes on the main —.986 11.22 -.997 4.87

stem

racemes on branches .986 11.22 .997 4.84

pods total 1.000 .72 .997 5.18

total nodes .996 5.32

racemes total .986 11.22 -.997 4.84

pods on branches .930 24.76 .961 18.67

average intenode -.879 32.53 .986 10.92

length

pods/raceme .986 10.92

number of branches .938 23.42

branch length .785 42.79 .986 14.57

LITERATURE CITED

LITERATURE CITED

I

Achibong, D. 1963. On the correlation among yield
components in the navy bean as influenced by spac-
ing and interstrain competition. M.S. Thesis,
Michigan State University, East Lansing, Michigan.

Adams, M.W. 1967. Basis of yield component compensa-
tion in crop plants with special reference to the
field bean, Phaseolus vulgaris. Crgp Science,
Vol. 7, September-October, pp. 5055510.

 

 

Adams, M.W. 1974. Plant architecture and yield in the
grain legumes. Food and Agricultural Organization
Sponsored Grain Legume Workshop, All India Agri-
cultural Research Institute, New Delhi, India,
October, pp. 1-16.

Adams, M.W. 1982. Plant architecture and yield breed-

ing. Iowa State Journal of Research, Vol. 56,
No. 3,_February, pp. 225-254.

 

Adams, M.W. and J.W. Wiersma. 1978. An adaptation of
principal components analysis to an assessment of
genetic distance. Research Report from the
Michigan State University Agricultural Experiment
Station, East Lansing, Michigan, pp. 1-7.

Aryeetey, A.N. and E. Laing. 1972. Inheritance of
yield components and their correlation with yield
in cowpea [Vigna unguiculata (L.) Walp.].
Euphytica, Vol. 22, pp. 386-392.

 

Barr, A., J.H. Goodnight, J.P. Sall and J.T. Helwig.
1976. A user's guide to SAS 76. SAS Institute,
Inc., Raleigh, North Carolina.

Cohen, J.D. and A. Schulze. 1981. Double-standard iso-
type dilution assay 1. Quantitative assay of
indole-3-acetic acid. Analytical Biochemist_y,
pp. 249-257

 

127

 

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

128

Denis, J.C. 1967. Estimacion de la hereditabilidad del
rendimiento y sus componentes primarios en el
frijol comun (Phaseolus vulgaris L.); correlaciones
fenotipicas y genotipiEas entre estos caracteres.
Instituto Interamericano de Ciencias Agricolas de
la OEA, Centro de Ensenanza e Investigigacion
Turrialba, Costa Rica.

 

Denis, J.C. 1971. Factor analysis of plant-type vari-
ables related to yield of dry beans (Phaseolus
vulgaris L.). Ph.D. Dissertation, Michigan State
University, East Lansing, Michigan.

 

 

Denis, J.C. and M.W. Adams. 1978. A factor analysis
of plant variables related to yield in dry beans-
I. Morphological traits. Crop Science, 18:74-78.

 

Dickson, M.H. 1967. Diallel analysis of seven eco-

nomic characters in snap beans. CropgScience,
7:121-124.

 

Donald, C.M. 1968. The breeding of crop ideotypes.
Euphytica, Vol. 17, pp. 385-403.

Duarte, R.A. 1961. Component interaction in relation
to mean expression of complex traits in a field
bean cross. M.S. Thesis, Michigan State Univer-
sity, East Lansing, Michigan.

Duarte, R.A. and M.W. Adams. 1963. Component inter-
action in relation to expression of a complex

trait in a field bean cross. Crop Science, Vol.
3, No. 3, pp. 185-186.

 

Duarte, R.A. 1966. Responses in yield and yield
components from recurrent selection practiced in
a bean hybrid population at three locations in
North and South America. Ph.D. Dissertation,
Michigan State University, East Lansing, Michigan.

Duarte, R.A. and M.W. Adams. 1972. A path coefficient
analysis of some yield component interrelations in
field beans (Phaseolus vulgaris L.). Crop Science,
Vol. 12, September-OctoBEr, pp. 579-5827

 

 

Dudley, J.W. and R.H. M011. 1969. Interpretation and
use of estimates of heritability and genetic vari-
ances in plant breeding. Crgp Science, 9:257-262.

Evans, A. 1970. Heterosis for yield in Phaseolus
vulgaris crosses. Annual Report of_the Bean Im-
provement Cogperative, 13:52454.

 

 

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

129

Field, R.J. and D.I. Jackson. 1974. A hormone balance
theory of apical dominance. In Mechanisms of
Regulation of Plant Growth (R.L. Bieleski, A.R.
Ferguson, M.M. Cresswell, eds.), Bul. 12, The
Royal Society of New Zealand, pp. 655-657.

 

Gehriger, W., S. Belluli and Keller. 1977. Influence
of decapitation and growth regulations on yield
components and yield of Vicia faba L. In Vicia

faba: ’Ph $1010 and Breedin (ed. by R. Thompson

and Martinus Nishoff for ECC), (1981), pp. 421—435.
Ghaderi, A. and R.L. Lower. 1979. Analysis of genera—

tion means for yield in six crosses of cucumber.

Journal of the American Society for Horticultural
Science, Vol. 104(4), pp. 567-572.

 

Ghaderi, A. and R.L. Lower. 1981. Estimate of genetic
variances for yields in pickling cucumbers.
Journal of the American Society for Horticultural

Science, Vol. 106(2), March, pp. 237—239.

Ghaderi, A. and M.W. Adams. 1981. Preliminary studies
on the inheritance of structural components of plant
architecture in dry beans (Phaseolus vulgaris L.).

Annual Re ort of the Bean Im rovement Cogperative,
Vol. 24, pp. 35-31.

Grafius, J.E. and R.L. Thomas. 1971. The case for in-
direct genetic control of sequential traits and
the strategy of development of environmental re-
sources by the plant. Heredity, Vol. 26, Part 3,
pp. 433-442.

 

 

 

Grafius, J.E. 1978. Multiple characters and correlated
response. Cro Science, Vol. 18, November-
December, pp. 831-934.

Guven, A. and Y. Vardar. 1977. Correlations between
the primary leaves and the cotyledonary-axillary
buds of Phaseolus vulgaris L. Z. Pflvelen-
physiol.de., 97. 377 PP. 153-160.

 

 

Hillman, J.R., V.B. Math and G.C. Medlow. 1977. Apical
dominance and the levels of indole acetic acid in
Phaseolus lateral buds. Planta, Vol. 134, pp. 191-
193.

Jackson, D.I. and R.J. Field. 1972. Light and hormone
interaction in apical dominance in Phasedlus
vulgaris L. Ann. Bot., 36 525—532.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

130

Kigel, J. 1980. The role in the leaves in the regula-
tion of internode elongation in Phaseolus vulgaris.
Physiol. Plant., 49:161-168.

 

 

Kueneman, E.A. 1978. Evaluation of the yield potential
of growth habits of dry beans (Phaseolus vulgaris
L.) and determination of plant types for high
density planting. Ph.D. Dissertation, Cornell
University, Ithaca, New York.

 

Mather, K. 1949. Biometric genetics--the study of con-
tinuous variation. Dover Publications, Inc.

Mather, K. and J.L. Jinks. 1977. Introduction to bio-
metrical genetics. Cornell University Press,
Ithaca, New York, 231 pp.

McDougall, J. and J.R. Hillman. 1980. Distribution of
indole-3-acetic acid in shoots of Phaseolus

vul aris L. Z. Pflanzenphysiol. Bd., 97. 5.,
367-371.

Moll, R.H., C.S. Levings, III and C.W. Stuber. 1969.
Genetic variation in maize coleoptile growth and
its relationship to the auxin regulatory system.
Crop Science, Vol. 9, May—June, pp. 262-264.

 

 

Nagl, K. 1981. Some reflection and data about useful
plant types in Vicia faba breeding. In Vicia faba:
Physiology and Breeding (ed. by R. Thompson and
Martinus Nishofffifor EEG), (1981), pp. 193-209.

 

 

Nie, N., H. Hull, J. Jenkins, K. Steinbrenner and D.
Bent. 1975. Statistical package for the social

sciences. ‘McGraw-Hill Book Company, New York, pp.
468-508.

Patrick, J.W. and D.J. Woolley. 1973. Auxin physiology
of decapitated stems of Phaseolus vulgaris L.
treated with indole-3yl-acetic acid. Journal of
Expgrimental Botany, Vol. 21, No. 82, October,
pp. 949-957.

Peterson, C.A. and R.A. Fletcher. 1975. Lateral bud
growth on excised stem segments: effect of the
stem. Can. J. Bot., Vol. 53, pp. 243—248.

 

 

 

 

Phillips, I.D.J. 1975. Apical dominance. Ann. Rev.
Plant Physiol., 26:341-367.

 

Powers, L. and L.F. Locke. 1950. Partitioning method
of genetic analysis applied to quantitative char-
acters of tomato crosses. U.S. Department of Agri—
culture Technical Bulletin No. 998, January.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

131

Shein, T. and D.I. Jackson. 1971. Hormone interaction
in apical dominance in Phaseolus vulgaris L. Ann.
Bot., Vol. 35, pp. 555-564.

 

Shein, T. and D.I. Jackson. 1972. Interaction between
hormones, light and nutrition on extension of
lateral buds in Phaseolus vulgaris L. Ann. Bot.,
36:791-780.

 

 

Shelagh, M.H. and J.R. Hillman. 1975. Correlative in-
hibition of lateral bud growth in Phaseolus vulgar-

is L. timing of bud growth following decapitation.
PIanta., Vol. 123, pp. 137-143.

 

Sinnot, E.W. 1921. The relation between body size and
organ size in plants. The American Naturalists,
55:385-403.

 

Sinnot, E.W. 1936. The development of developmental
relationships. The American Naturalists, No. 733,
pp. 113-119.

 

Snow, R. 1937. On the nature of correlative inhibition.
New Phytology, 36, pp. 283-300.

 

Tayo, T.O. 1982. Growth, development and yield of
pigeon pea [Cajanus ca'an (L) millsp.] in the low-
land tropics. Journal of Agricultural Science
(Cambridge), V01? 98, pp. 79-84.

Thimann, K.V. and F. Skoog. 1934. On the inhibition
of bud development and other functions of growth
substance in Vicia faba. Royal Society B., 114:

317-339.

 

Thimann, K.V. 1977. Apical dominance in hormone action
in the whole life of plants. The University of
Massachusetts Press, Amherst, Massachusetts, pp.
289-326.

Foysest, O.V. 1970. An attempt to overcome the yield-
dampening effect of negative correlations among
yield components in beans (Phaseolus vulgaris L.).
Ph.D. Dissertation, Michigan StateUniversity,
East Lansing, Michigan.

 

Weiss, P. 1939. Principles of development. Henry Holt
and Company, New York, p. 601.

Went, F.W. 1953. Gene action in relation to growth and
development -Il. Phenotypic variability. Proc.
N.A.S., Vol. 39, pp. 839-848.

132

54. White, J.C., G.C. Medlow, J.R. Hillman and M.B. Wilkins.
1975. Correlative inhibition of lateral bud
growth in Phaseolus vulgaris L. isolation of
indoleacetic acid from the inhibitory region. J;
Exp. Bot., Vol. 26, No. 92, pp. 419-424.

 

55. White, J.C. 1976. Correlative inhibition of lateral
bud growth in Phaseolus vulgaris L. effect of ap-
plication of ifido-3y1-acetic aEid to decapitated
plants: Ann. Bot., 40:521-529.

56. Wilkson, M. and Thimann. 1958. The antagonism of auxin
and kinetin in apical dominance. Physiolggia

Plantarum, 11:62-74.

 

“'71711711111((111111111Es