Inheritance of resistance to the soybean looper in soybean

MISSING IMAGE

Material Information

Title:
Inheritance of resistance to the soybean looper in soybean
Physical Description:
x, 102 leaves : ill. ; 29 cm.
Language:
English
Creator:
Kenty, Michael Montgomery, 1959-
Publication Date:

Subjects

Subjects / Keywords:
Agronomy thesis Ph.D
Dissertations, Academic -- Agronomy -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1994.
Bibliography:
Includes bibliographical references (leaves 95-101).
Statement of Responsibility:
by Michael Montgomery Kenty.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001986578
oclc - 32021356
notis - AKG3548
System ID:
AA00004723:00001

Full Text









INHERITANCE OF RESISTANCE TO THE SOYBEAN LOOPER
IN SOYBEAN















By

MICHAEL MONTGOMERY KENTY


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1994






























I would like to dedicate this work to the memory of the late Dr. Johnny D. Ouzts.

Thank you for your friendship and guidance; without it I would not have made it this far.














ACKNOWLEDGEMENTS


My deepest gratitude is extended to Dr. Kuell Hinson for serving as my mentor

throughout my program of study. I would also like to thank the remaining members of

my graduate committee, Drs. Joe Funderburk, Ken Quesenberry, John Strayer, and

David Wofford, for their insight and guidance. This study could not have been

completed without the support and aid of the USDA-ARS Soybean Production Research

Unit at Stoneville, Mississippi. I want to thank each and every member for their

friendship, patience, and support during this educational process. Also, I would like to

thank Dr. Clarence Watson and Debbie Boykin for their statistical advice. A special

thanks goes to Drs. Edgar E. Hartwig and Thomas C. Kilen for taking a chance on a

young kid years ago, and introducing me to the wonderful field of plant breeding.














TABLE OF CONTENTS


page


ACKNOWLEDGEMENTS .................

LIST OF TABLES ......................

LIST OF FIGURES ......................

ABSTRACT .............. ..........

CHAPTER 1 INTRODUCTION .............

CHAPTER 2 LITERATURE REVIEW .........


. . . iii

. . vi

. . viii

. . ix

S. . 1


CHAPTER 3 EVALUATION OF RATING METHODS .............. 13
Introduction ......................... ......... ............ 13
Literature Review ..................... ............. 13
Materials and Methods ....... ......................... 15
Results and Discussion .................................. 19
Summary and Conclusions ........... ................... 29

CHAPTER 4 INHERITANCE OF RESISTANCE TO SOYBEAN LOOPER IN
SOYBEAN ............................. ............. 31
Introduction .................................. ............ 31
Materials and Methods ............... ................. 32
1988 . . . . 32
1989 . . . . 33
1990 . . . .. . 34
1991 ... . . . .. 39
Results and Discussion ................................. 49
1990 .. . . . 49
1991 . . . . 51
Summary and Conclusions ....... ............. ......... 66

CHAPTER 5 SUMMARY AND CONCLUSIONS .................... 71
Evaluation of Rating Methods ....... ........... ......... 72
Inheritance Study ............... ..... ................. 74








APPENDIX A LIST OF DEFOLIATION SCORES GENERATED BY
WHOLE PLANT VISUAL, PARTITIONED PLANT VISUAL, AND
LEAF AREA OF THE PARTITIONED PLANT RATING METHODS
IN 1990. .......................................

APPENDIX B LIST OF DEFOLIATION SCORES GENERATED BY
WHOLE PLANT VISUAL, PARTITIONED PLANT VISUAL, AND
LEAF AREA OF THE PARTITIONED PLANT RATING METHODS
IN 1991. ................................. ........


APPENDIX C


GENERATION MEANS ANALYSIS SAS PROGRAM .....


APPENDIX D GENERATION MEANS ANALYSIS SAS PROGRAM .....


APPENDIX E


TEST FOR NORMALITY ......................


REFERENCES ......................................... 95

BIOGRAPHICAL SKETCH ................................. 102













LIST OF TABLES


page

Table 1. McNemar test for significant differences between the WP (whole plant
visual) and AV (partitioned plant visual) methods in 1990; and between
the WP and AV, WP and AL (leaf area of the partitioned plant), and AV
and AL methods in 1991 ............................... 28

Table 2. Populations used for generation means analysis of the inheritance of
resistance to the soybean looper. . .... ....... 46

Table 3. Ratings of leaf feeding by soybean looper on soybean parents, their F,
and F2 progeny, and the germplasm lines PI229358 and D75-10169 in the
field cage-1990..................................... 50

Table 4. Goodness of fit test for normality of soybean looper defoliation for
populations of D86-3429, Braxton, and D86-3429 x Braxton F2
populations in the 1990 cage study. ......................... 52

Table 5. Ratings of leaf feeding by soybean looper on soybean parents, their F1,
F2, F3, BCI, BSI, BC2, and BS2 progeny, and the germplasm lines PI
229358 and D75-10169 in the field cages-1991. ............... 54

Table 6. Estimates of variance components obtained from an analysis of the
homogeneous populations, D86-3429, Braxton, D75-10169, PI 229358,
for a location effect due to feedingt by soybean looper in field cages in
1991. .......................................... 55

Table 7. Means, standard errors, and variances of soybean looper defoliation
ratings for parental lines and seven populations of soybean arising from
the cross D86-3429 x Braxton grown in field cages at Stoneville, MS in
1991. . . . . . 57

Table 8. Mather's scaling test applied to the cross D86-3429 x Braxton to test
adequacy of additive-dominance model for resistance to soybean looper. 58








Table 9. Estimates of the additive, dominant, and epistatic effects in the
generation means for defoliation by soybean looper in the nine populations
of the D86-3429 x Braxton material grown in the field cages at
Stoneville, MS in 1991. ................... ............ 60

Table 10. Estimates of the additive, dominant, and epistatic effects in the
generation means for defoliation by soybean looper in the nine populations
of the D86-3429 x Braxton material grown in the field cages at
Stoneville, MS in 1991. ................... ............ 61

Table 11. Goodness of fit test for normality of soybean looper defoliation data
from 1991 cage study ................................. 63

Table 12. Average defoliation scores in soybean populations resistant to soybean
looper in 1991 cage study. ................... ........... 67














LIST OF FIGURES


page

Figure 1. Examples of photocopied leaflets used in determining the % defoliation
in the AV (leaf area of partitioned plant) method . ..... 17

Figure 2. 1990 frequency distribution of defoliation estimates by the whole plant
and partitioned plant average visual rating methods. . ... 21

Figure 3. 1991 frequency distribution of defoliation estimates by the whole plant,
partitioned plant average visual, and measured average leaf area of the
partitioned plant rating methods. . ... .. ...... 22

Figure 4. Schematic drawing of field cages used in the 1990 and 1991
experiment . ... . ..... 35

Figure 5. Mating scheme to derive populations required to estimate the genetic
effects of resistance to the soybean looper in soybean. . ... 45













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of
Doctor of Philosophy

INHERITANCE OF RESISTANCE TO THE SOYBEAN LOOPER
IN SOYBEAN

By

Michael Montgomery Kenty

April, 1994

Chairman: Kuell Hinson
Major Department: Agronomy

Phytophagous insects cause millions of dollars of damage to soybean [Glycine max

(L.) Merr.] throughout the southern United States. This study was conducted to: (i)

evaluate three methods for rating defoliation of soybean by insects and (ii) determine the

inheritance of resistance to the soybean looper [Pseudoplusia includes Walker] in

soybean. The insect-resistant breeding line D86-3429 was crossed with the cultivar

Braxton in 1988. Additional crosses were made in 1989 and 1990 to provide the

populations necessary to conduct rating methodology and genetic studies in field cages

in 1990 and 1991. The three rating methods evaluated were whole plant visual (WP),

partitioned plant average visual (AV), and measured average leaf area of partitioned plant

(AL). In 1990 the WP and AL were highly correlated (r=0.65, p < 0.001) and in 1991

all correlations (WP vs. AV, WP vs. AL, and AV vs. AL) were significant at p <








0.001. Based on the relative variation of each method, the precision of the three

methods was essentially equal and very good, well below the desirable level of 10.

These results indicated that relative estimates are suitable for genetic studies. In the

inheritance study defoliation was estimated by the WP method in 1990 and the AV

method in 1991. The data from the preliminary study in 1990 showed a trend towards

quantitative inheritance, therefore the 1991 data were analyzed quantitatively. Mather's

scaling test was applied to the data generated from each cage and the results indicated

that generation means depend only on additive gene effects. Utilizing Hayman's

methodology in terms of the generation means analysis, an epistatic effect was suggested,

but the primary effect is assumed to be additive, as was indicated in the scaling test.

Estimates of gene numbers indicate that the two parents differed by two genes for

resistance to soybean looper. Heritability for resistance was estimated to be 63%,

therefore a breeder should be able to make progress by selecting in the F2 or F3

populations. The potential exists for increasing resistance to the soybean looper, and

possibly other phytophagous insects, by pyramiding genes from new sources of

resistance.













CHAPTER 1
INTRODUCTION



Soybean [Glycine max (L.) Merrill] is a major world crop. Total world hectares

planted increased from 47.1 million in 1977-78 to 57.7 million in 1989-90 (2,3). This

22.5% increase was due primarily to demand for oil and meal products, and to a small

extent the whole bean product (75). In the United States, hectarage rose to a peak of

28.6 million hectares in 1978-79 with a steady decline to 24.0 million hectares in 1989-

90. This 16% decrease in soybean hectarage was primarily due to reduced plantings in

the southeastern United States (3).

The reduction in hectares planted to soybean in the southeastern United States is

attributable to the low market value of soybean. Another reason for the reduction is the

inability of farmers, in general, to effectively produce a profitable crop in the presence

of yield-reducing factors. These factors include fungal diseases, viral diseases, bacterial

diseases, nematodes, insects, and drought. In 1982, Mississippi suffered an estimated

$50 million loss in revenue from insects alone, of which $40 million could be attributed

solely to the soybean looper, Pseudoplusia includes Walker (15). Similar losses also

occurred in states neighboring Mississippi as a result of insect infestations (15,16). Pest

problems which can be costly to control, and a 10-year (1981-1990) average price of

$6.19 per bu (3), make it increasingly difficult for a grower to produce a profitable crop.








2
Several techniques are available for control of diseases, nematodes, and insects

in soybean. Metcalf and Luckmann (50) categorized these techniques into seven major

methods: cultural, mechanical, physical, biological, chemical, genetic, and regulatory.

All methods of control have merit and should be utilized in integrated pest management

programs (14,20,58,50). Although economics is the major limiting factor in achieving

control of any pest of soybean, human safety and environmental impact must be

considered when choosing an approach to control a particular pest. Before making a

decision on what approach to take, a grower should answer the questions: 'Will the end

results justify the cost of control?' and 'Is the control measure safe?'.

The soybean looper has become increasingly resistant to available insecticides

(10,18,48,58,60). For control of the soybean looper and other insect pests, a biological

approach should yield the best results. There are two biological approaches to insect

control: host plant resistance and the use of biocontrol agents (14,58). The latter

involves the identification and exploitation of natural enemies and pathogens. This

method of control usually is limited to a specific insect pest rather than the broad

spectrum of insects that damage soybean. As with conventional insecticides, resistance

to microbial insecticides has been observed (58).

Host plant resistance offers an effective method of control which is built into the

plant. It has relatively long stability, is compatible with other control tactics, is

environmentally safe (58), and development and subsequent release of a resistant cultivar

is far more socially acceptable (82). Lambert and Kilen (47) found that direct selection

for resistance in soybean to one species of insect resulted in the indirect selection for








3
resistance to certain other species. The development of an insect-resistant soybean

cultivar takes approximately the same amount of time as is required to develop a new,

effective insecticide, which can be as short as 10 to 12 years or as long as 20 years. The

insect-resistant cultivars 'Lamar' (28) and 'Crockett' (5) each took approximately 20

years to develop.

Initially, Van Duyn et al. (83) identified the three plant introductions PI 171451,

PI 227687, and PI 229358 from the USDA germplasm collection as having the highest

level of resistance to the Mexican bean beetle (Epilachna varivestis Mulsant). Schillinger

(69) identified additional sources of resistance to specific insects and described the

mechanism of resistance that each exhibits. Hatchett et al. (30) determined that breeding

lines identified as possessing resistance to multiple species of insects could serve as basic

germplasm for development of cultivars resistant to insects. To date, two soybean

cultivars and two soybean germplasm lines with resistance to insects have been registered

with the Crop Science Society of America (5,11,28,29).

Although there has been success in developing insect-resistant germplasm lines

and cultivars, there has been very little research conducted on the inheritance of insect

resistance in soybean (37,54,72). Of the three inheritance studies, two (54,72) were

conducted on the resistance to Mexican bean beetle and one (37) on the resistance to the

velvetbean caterpillar (Anticarsia gemmatalis Hubner). Since the soybean looper is a

major pest to soybean in the southeastern United States and is difficult to control with

insecticides, the inheritance of resistance to the soybean looper would aid breeders in the

development of resistant cultivars.








4
The objectives of this study were: (i) to evaluate three methods for rating

defoliation of soybean by soybean looper and (ii) to determine the inheritance of

resistance in soybean to the soybean looper.













CHAPTER 2
LITERATURE REVIEW



Although undocumented, plants that are resistant to insects have survived and

propagated based on the processes of adaptation and natural selection. These 'survivors'

were in turn selected by farmers to plant future crops (74). This practice of selection by

primitive farmers could be considered the beginning of the development of insect-

resistant cultivars.

According to Kogan (38), Pedigo (58), and Smith (74), in their brief histories of

insect-resistant plants, the first development and subsequent cultivation of insect-resistant

cultivars occurred in the late eighteenth and early nineteenth centuries. In 1792, J.N.

Havens identified the wheat (Triticum aestivium L. em. Thell.) cultivar 'Underhill' as

being resistant to the Hessian fly (Mayetiola destructor Say). In the 1830's, G. Lindley

recommended the cultivation of apple (Malus pumila Miller) cultivars resistant to the

woolly apple aphid (Eriosoma lanigerum Hausman). In the mid-nineteenth century, the

French wine industry was saved by grafting the highly susceptible scions of the French

varieties to the rootstocks of American grapes (Vitis L. spp.), which were resistant to the

grape phylloxera (Phylloxera vittifolae Fitch).

In the United States, R.H. Painter is considered to be the founder and pioneer of

the modern era of research on insect-resistant plants (58,74). In his book, Insect








6
Resistance in Crop Plants, which was the first book written on the topic, he described

mechanisms of resistance in plants and factors that affect them. Based on field

observations, he separated the mechanisms of resistance in plants into three distinct, but

interactive classes: nonpreference, later known as antixenosis (38), (for oviposition, food,

or shelter), antibiosis (adverse effect on the biology of the insect), and tolerance (ability

to withstand, repair, or recover from infestation) (57). Developing a cultivar that

exhibits a high level of all three mechanisms of resistance would be optimum. This is

seldom feasible, therefore developing cultivars with the antibiosis mechanism of

resistance is the most frequent objective of plant breeders. Although nonpreference and

antibiosis differ in their mechanisms of resistance, both involve the interaction of the

biology of the plant with the biology of the insect (58,74). There is often an overlapping

of the nonpreference and antibiosis classes which makes it difficult to determine the exact

mechanism of resistance exhibited in a plant to a particular species of insect (74).

Although it is advantageous if a breeder knows which mechanism of resistance

is being expressed, knowledge of the mechanism is not essential to make advances in the

development of resistant germplasm. In fact, resistant cultivars and germplasm have

been developed for years with little concern for the underlying mechanism of resistance.

It has only been in the past two decades that mechanisms of resistance have been

identified in crop plants (39).

In soybean, Van Duyn et al. (83) identified three genotypes (PI 171451, PI

227687, and PI 229358) from the USDA germplasm collection of maturity groups VII

and VIII as resistant to the Mexican bean beetle. These three PI's have been used








7
extensively as sources of resistant germplasm and as standard checks in screening

additional germplasm for resistance. At the World Soybean Research Conference in

1975, Schillinger (69) reported additional sources of resistance to the Mexican bean

beetle. Gary et al. (22) screened 1108 lines from maturity groups VI, VII, VIII, and IX

of the USDA germplasm collection against velvetbean caterpillar initially, and

subsequently velvetbean caterpillar, soybean looper, corn earworm (Heliothis zea

Boddie), tobacco budworm (H. virescens Fabricius), and beet armyworm (Spodoptera

exigua Hubner). Based on the results of the screenings against these five species, PI

209837 and FC 31592 were identified as having levels of resistance that would be

suitable for developing resistant cultivars. The germplasm collection of 473 PI's of wild

soybean, Glycine soja Sieb. & Zucc., was screened against soybean looper, velvetbean

caterpillar, beet armyworm, and corn earworm to identify additional sources of resistance

equal to, or greater than, PI's 229358 and 171451. McKenna et al. (52) identified three

lines (PI's 464935-1, 366119, and 407301) from this study as having less defoliation than

the two standards, thus suggesting that there are additional genes for insect resistance.

In 1988, Kraemer et al. (42) identified several accessions of the USDA germplasm

collection as being resistant to the Mexican bean beetle. They were from maturity

groups VI (FC 31665, and PI's 379621, 416925, and 416937) and VIII (PI's 417061 and

417136).

Upon identification of a genotype that is resistant to a certain insect species, the

genotype usually is screened against other insect species. Resistance to more than one

species will enhance its usefulness in a breeding program. Hatchett et al. (30) conducted








8
a study utilizing breeding lines that derive their insect resistance from PI 229358 to

determine if selection for resistance to one or two insect species would result in the

indirect selection for resistance to other troublesome species. They found that no single

insect species could be utilized to select for resistance to multiple species of insects

among adapted genotypes. In a similar study, Lambert and Kilen (47) determined that

when PI 229358 was used as the donor parent, direct selection for resistance to one

insect species resulted in indirect selection for resistance to other foliar-feeding species.

Research to develop soybean genotypes resistant to insects is usually conducted

with species that are indigenous to a particular area (19,25,35,44,54,64,80,83). Lambert

and Kilen (45) evaluated PI229358, PI227687, and PI 171451 against five insect species

to determine their relative levels of resistance. Insect species tested were velvetbean

caterpillar, soybean looper, corn earworm, tobacco budworm, and beet armyworm.

Based upon the responses to these five species, they found that the mechanism for

resistance in PI 171451 was nonpreference, whereas that of PI 229358 or PI 227687 was

antibiosis. These same three PI's were evaluated against four important soybean

defoliating species (beet armyworm, Porthesia taiwania Shiraki, Anomala cupripes Bates,

and Orgyia species) to determine their usefulness in developing breeding lines resistant

to insects indigenous to Taiwan. No single accessions had a consistently high level of

resistance to all of the insect species. PI 227687 demonstrated the highest level of

resistance to the most prevalent species, the beet armyworm, and therefore would be

suitable to initiate a breeding program (80).








9
With the sources of resistance to various species of insects identified, researchers

have channelled their energy into determining the basis of the mechanism of resistance

in certain soybean accessions. Since the mid-1970's, researchers have investigated

various constituents of soybean to identify the underlying biochemical compound that

confers resistance to insects (9,53,73,81). If an association could be established between

a particular compound or class of compounds and resistance, then a new tool could be

developed to accelerate the selection process by breeders.

In studies with PI 227687, PI 229358, and two susceptible cultivars, Tester (81)

reported that resistant PI's had lower total nitrogen, more soluble carbohydrates, and

more total sterols at flowering and pod-fill than susceptible cultivars. Through a series

of grafting experiments, Lambert and Kilen (46) verified that constituents contributing

to resistance in PI 227687 and PI 229358 were formed in the leaves and translocated

throughout the plant. Smith (73) determined that the resistant factors in PI 227687

appeared to be chemical. He also found that resistance to the soybean looper in PI

227687 was an antibiotic effect due to the combined effects of a feeding deterrent and

a growth inhibitor. Chiang et al. (9) suggested that resistance to Mexican bean beetle

feeding in PI 227687 may be attributed to phenylpropanoid metabolites. In their

inorganic nutrient analysis of leaf tissue studies with soybean resistant to the Mexican

bean beetle, Mebrahtu et al. (53) found that the content of calcium, phosphorus, and

potassium was negatively correlated with pupal weight gain (r = -0.23", r = -0.26",

and r = -0.20'; p < 0.01 and p < 0.05, respectively).










In 1986, Kogan (39) published a treatise on the role of biochemicals in plant

resistance to insects. He examined the defensive patterns of three families: Solanaceae,

Cruciferae, and Leguminosae (specifically the common bean and soybean) and published,

as quoted below, a summation of all previous research findings that elucidated the

chemical bases of resistance in PI 171451, PI 227687, and PI 229358.

1. Foliar total nitrogen, soluble carbohydrates, organic acids
and sterols of both resistant and susceptible soybean vary
with stage of growth (Tester, 1977; Grunwald and Kogan,
1981).

2. Total nitrogen was generally lower and, at the end of the
season, soluble carbohydrates and sterols were higher in
the resistant lines; organic acids fluctuated without
correlation with resistance (Testor, 1977).

3. Sterol profiles were similar in both susceptible and resistant
lines, both in early and in late maturing soybeans
(Grunwald and Kogan, 1981).

4. Pinitol is a dominant cyclitol in soybean and it represents
an average of 60% of the total 80%-ethanol-soluble
carbohydrate fraction of soybean (Phillips et al., 1982).

5. Pinitol was found in large amounts in PI 229358, and in
smaller concentrations in the susceptible cultivar 'Davis'
(@ 1% dry weight). Added to artificial media at 0.8%,
pinitol reduced weight gain by corn earworm larvae by
about 63%. These findings suggest that pinitol is an
antibiotic factor in soybean but with no apparent antixenotic
property (Dreyer et al., 1979). As the reported results
could not be replicated elsewhere, this claim has been
recently disputed (Gardner et al., 1984).

6. Systematic fractionation of PI's 227687 (Smith and Fischer,
1983) and 229358 (Binder and Waiss, 1984) yielded
fractions with intermediate and with polar solvents that
caused slower developmental rates and high mortality in
soybean looper [Pseudoplusia includes (Walker)] and
Heliothis zea Boddie, respectively.









7. Seven phenolic acids were detected in PI 171451 and in cv.
'Forrest'; gallic, protocatechuic, p-hydroxybenzoic,
vanillic, caffeic, p-coumaric, and ferulic. The potentially
antibiotic caffeic and ferulic acids occurred at higher
concentrations in the resistant PI than in the susceptible
cultivar. In both types of plants, concentrations were
higher in injured than in uninjured tissue (Hardin, 1979).

8. Isoflavonoid phytoalexins in soybean cotyledons are potent
feeding deterrents for the Mexican bean beetle (Hart et al.,
1983). These phytoalexins are biochemically related to the
phenolics mentioned above, through the shikimic pathway.
(39, p.516-518).

Although a great deal of knowledge has been gained, there still remain many unanswered

questions as to what components comprise the various mechanisms of resistance.

A moderate amount of research has been conducted in the area of resistance in

soybean to insects, and much progress has been made in cultivar development.

However, there has been very little information published on the inheritance of resistance

to insects in soybean. In their work with the Mexican bean beetle, Sisson et al. (72)

hypothesized that resistance was inherited quantitatively and that the number of major

genes involved was small, perhaps two or three major genes. In further studies with the

Mexican bean beetle, Mebrahtu et al. (54) concluded that resistance appeared to be

primarily an expression of additive genetic variance. Kilen et al. (36), in their studies

with the soybean looper, suggested that susceptibility was partially dominant with a few

major genes conditioning resistance. In their studies with PI 171451, PI 227687, and PI

229358, and the F, plants derived from intercrosses among the three PI's, Lambert and

Kilen (45) suggested that genetic resistance in PI 171451 differed from that operating in

the other two lines. This was further supported by their work with the PI's and the F3








12
lines derived from the intercrosses among the PI's. They hypothesized that each PI

contained a resistance gene that differed from those of the other lines. To design a better

breeding program to develop cultivars resistant to the soybean looper utilizing breeding

lines that derive their insect resistance from PI 229358, it would be advantageous if the

mode of inheritance of this trait were known.














CHAPTER 3
EVALUATION OF RATING METHODS


Introduction

Throughout the past 25 years there have been advances in the identification and

development of soybean breeding lines resistant to phytophagous insects, especially

Lepidopterous species and the Mexican bean beetle (5,11,22,28,29,30,42,52,79,83).

Despite these advances, little effort has been devoted to improving the methods of

evaluating the amount of feeding injury of insects. Most researchers utilize a rating

system that estimates the level of resistance among cultivars, but little or no effort has

been made to determine whether or not the rating scales used were truly reflecting the

defoliation caused by insects. Hence, the objective of this study was to evaluate the

effectiveness of the widely used visual, or relative, assessment of insect defoliation as

compared to a quantitative, or absolute, assessment.

Literature Review

Plant pathologists have faced problems in the assessment of disease which is

dependent upon the causal organism and the host plant species. Many different methods

of disease assessment are available to plant pathologists (7). Shokes et al. (71) tested the

reliability of disease assessment procedures with late leafspot [Cercosporidium

personatum (Berk. and Curt.) Deighton] in peanut (Arachis hypogaea L.) and concluded








14

that plant pathologists would benefit if a standard rating system for foliar diseases of

peanut could be adopted as has been done in other crops. In a special report, Berger (4)

stated that a disease assessment procedure should have the following basic requirements:

easy to learn, quick and efficient to use, applicable across a broad spectrum of

conditions, and be accurate, precise, and reproducible. He concluded that the greater the

emphasis placed on the assessment method, the higher the quality of information obtained

by the method.

In 1980, Kogan and Turnipseed (41) described the various methods of quantitative

and qualitative assessments of defoliation that are currently used by researchers working

with phytophagous insects. These methods are: (i) photoelectric or electronic (measure

actual leaf area), (ii) gravimetric (regressing known leaf areas on the corresponding fresh

or dry weight of these areas, (iii) volumetric (volume displacement cylinder is used to

measure biomass), (iv) planimetric (mechanical measurement of area), (v) geometric

(estimates total nondefoliated leaf area so that percent defoliation can be calculated when

used in conjunction with a leaf area meter), and (vi) visual estimates (provide rapid

estimate of defoliation). Although descriptions of the methods were detailed by Kogan

and Turnipseed, no statistical comparisons were made among methods.

Some authors (58,78) refer to qualitative assessments as relative estimates and

quantitative assessments as absolute estimates. Relative estimates do not translate directly

into the amount of defoliation, whereas absolute estimates allow for direct translation into

the amount of defoliation caused by phytophagous insects. The major drawback of

absolute estimates is that they are usually time consuming and costly to make.








15
Conversely, relative estimates can be timely and are inexpensive which is why these

types of estimates are more widely used by researchers (74).

Funderburk et al. (19) determined that plot size or shape had little effect on the

relative rankings of soybean cultivars when measuring resistance to the velvetbean

caterpillar in terms of percentage of defoliation based on visual estimates. However,

they found that the mean and precision of the estimates were affected by plot size and

shape. The objective of this study was to compare two relative methods to one absolute

method for assessing defoliation caused by the soybean looper.

Materials and Methods

The three rating methods evaluated in 1990 and 1991 were the whole plant visual

rating (WP), the partitioned plant average visual rating (AV), and measured average leaf

area of the partitioned plant (AL). WP values were based on a 1-to-10 scale divided into

10% increments with a score of 1 signifying defoliation of 0 to 10% and a score of 10

signifying defoliation of 91 to 100%. Although other scales, such as a logarithmic scale

(33), may be better at estimating defoliation, the 1-to-10 scale was chosen since it is most

commonly used by participants in the Regional Host Plant Resistance Test which is

sponsored annually by the Southern Regional Information Exchange Group (SRIEG 32).

With the AV and AL methods, the canopies were visually partitioned into three sections

(top, middle, bottom) enabling a more detailed assessment of soybean looper defoliation

on each plant. The partitioning of the plants was based on a procedure that was

developed by Plaut and Berger (59) that was cited by Shokes et al. (71). For AV








16
estimates, each section of the plant was scored using the same scale as the WP with the

three scores averaged to obtain the AV.

In 1990, a LI-COR LI-3000 (Li-Cor, Inc., Lincoln, NE 68504) portable leaf area

meter was used to measure the middle leaflet of a representative trifoliate from each of

the sections and the three measurements were averaged to obtain the AL. In 1990 the

AL method only allowed an indirect comparison to the other two methods, therefore the

AL was modified in 1991 to permit a direct comparison with AV and WP. In the 1991

AL method, a representative trifoliolate leaf from each section of the partitioned plant

was removed and taken into a laboratory to determine the area of each sample. The

portable leaf area meter was used to determine the remaining portion of each defoliated

leaflet (DEF) of each trifoliolate. Following the procedure described by Kogan and

Turnipseed (41), each leaflet of the trifoliolates was then placed on a photocopy machine

and a copy was made. Examples of leaflet photocopies are shown in Figure 1. These

silhouettes were then cut out to represent the trifoliolates in a nondefoliated state and the

area (NONDEF) for each was determined as mentioned above. The two values obtained

for each trifoliolate were used in the following formula to determine the percent

defoliation (% def): (NONDEF DEF)/NONDEF 100 = % def. The percent

defoliation for each of the three trifoliolates from each plant was then converted to the

same 1-to-10 scale used by the other two methods and the values averaged, allowing

direct comparisons to be made among the three methods.

The plants used in the evaluation of the three methods were part of an inheritance

study that was conducted in field cages (43) in 1990 and 1991 at the Mississippi State




















































Figure 1. Examples of photocopied leaflets used in determining the % defoliation
in the AV (leaf area of partitioned plant) method.








18

University Delta Research and Extension Center, Stoneville, Mississippi. In the first

year of the study, seed were planted on 25 May 1990. In 1991, seed were planted on

25 May and 29 May in cage 1, and on 31 May in cage 2. Rainfall on 25 May prevented

completion of planting on the same day in cage 1.

To compare methods in 1990, ten soybean plants were chosen randomly from

each of 20 backcross families for a total of 200 plants. Fifty plants in each of two cages

were randomly chosen using PROC PLAN of PC-SAS (68) for use in the evaluation of

the three methods in 1991. These plants were chosen prior to release of soybean looper

moths in both years.

The 4000 three-day-old soybean looper moths were released on 25 June 1990

when plants were at the V5 to V6 stage of plant growth (17). Larvae from hatched eggs

were allowed to feed freely for approximately 12 days, at which time 'Centennial' (27),

the susceptible check, was estimated to have a rating of nine according to the WP

method. The area within the cage was then sprayed with methomyl [S-methyl-N-((methyl

carbamoyl) oxy)-thioacetimidate] at 0.5 kg ha' to kill the larvae. This allowed

defoliation ratings to be made without further feeding. The day following spraying, each

plant was rated by all three methods.

Approximately 4000 three-day-old soybean looper moths were released into each

cage on 8 July 1991. At the time of release, plants in cage 1 were V7 to V8 and in cage

2 they were V6 to V7. The larvae from the hatched eggs were allowed to feed freely

until the susceptible check, Centennial, received a rating of ten by the WP method, 16

days after release of the moths. To cease feeding at this stage of plant growth, R1 to R2








19

in cage 1 and beginning R1 in cage 2, the plants in both cages were sprayed with

methomyl at 0.5 kg ha"' to kill the larvae. The following day each of the 100 plants was

rated using the WP and AV methods and samples were collected for the AL.

Data derived from each rating method for the 200 plants in 1990 and the 100

plants in 1991 were subjected to analytical procedures in PC-SAS (68). A correlation

analysis was run to determine the correlation coefficient (r) between WP and AL, AV

and WP, and AL and WP rating methods each year. A paired difference comparison

(76) was made between WP and AV methods in 1990 and between WP and AL, WP and

AV, and AL and AV methods in 1991. McNemar's test for concordance (77) was used

to test for significant changes between WP and AL in 1990 and between WP and AL,

WP and AV, and AL and AV methods in 1991. This analysis was conducted to

determine if the methods differed in their estimates of defoliation. Precision, known as

the relative variation (RV), of the two relative methods for both years and the absolute

method in 1991 were calculated for each as follows:


RV = (SE/.)100


where SE is the standard error of the mean and X is the mean. This relationship,

described by Ruesink (66) and Pedigo (58), is considered useful in estimating precision

of a sampling method by entomologists.

Results and Discussion

The 1990 scores for each of the three rating methods by plant are presented in

Appendix A. Scores for AV and WP were based on a 1-to-10 scale, with lower numbers








20
indicating less defoliation. The measurements for AL in 1990 are presented as cm2 of

remaining tissue. Individual plant data for each of the three methods in 1991 are

presented in Appendix B. Unlike 1990, 1991 data for all three methods were based on

a 1-to-10 scale. Means and standard deviations for both years are also reported in

Appendices A and B.

Frequency distributions of the 1990 WP and AV rating methods are shown in

Figure 2. The AL method was omitted from this graph because it measured the

remaining undefoliated tissue in cm2, whereas the AV and WP methods were used to

estimate percent defoliation on a 1-to-10 scale. Frequency distributions of the AV and

WP methods suggest that either the evaluator underestimated defoliation by the WP

method or overestimated defoliation by the AV method. Without an absolute estimate

of defoliation, no substantial conclusion can be made as to the reliability of AV or WP

to accurately assess defoliation.

The 1991 frequency distributions of all three methods for both cages combined

are presented in Figure 3. Due to the modifications in the AL method, it was possible

to make direct comparisons between AL and the visual rating methods. The evaluator

overestimated defoliation to obtain too few plants below 30% or <4 in the 1-to-10 scale

when using either relative method, WP or AV, as compared to AL, the absolute method.

Also, the evaluator had a tendency to overestimate the amount of defoliation over 50%

or >5 in the 1-to-10 scale when using either relative method as compared to AL. Only

in classes 4, 5, and 6 was the evaluator's estimate of defoliation with the WP and AV

methods similar to that of the AL. The problem of overestimating might be avoided in









21




Number of Plants
20
Rating Methods
*WP HAV

60 -




40




20




1 2 3 4 5 6 7 8 9 10
Defoliation Scores



Figure 2. 1990 frequency distribution of defoliation estimates by the whole plant and
partitioned plant average visual rating methods. Defoliation scores based on 1-to-10
scale where 1 = 0 to 10% defoliation, 2 = 11 to 20%, 3 = 21 to 30%, 4 = 31 to 40%,
5 = 41 to 50%, 6 = 51 to 60%, 7 = 61 to 70%, 8 = 71 to 80%, 9 = 81 to 90%, 10
= 91 to 100%.














Number of Plants


Rating Methods
MWP HAV MAL


1 2 3 4 5 6 7 8 9 10


Defoliation Scores


Figure 3. 1991 frequency distribution of defoliation estimates by the whole plant,
partitioned plant average visual, and measured average leaf area of the partitioned plant
rating methods. Defoliation scores based on 1-to-10 scale where 1 = 0 to 10%
defoliation, 2 = 11 to 20%, 3 = 21 to 30%, 4 = 31 to 40%, 5 = 41 to 50%, 6 = 51
to 60%, 7 = 61 to 70%, 8 = 71 to 80%, 9 = 81 to 90%, 10 = 91 to 100%.








23
the future if a logarithmic rating scale, such as the Horsefall-Barratt scale to assess

disease (33), were used to estimate defoliation. A logarithmic scale may have advantages

for assessment of defoliation because the visual acuity of the human eye is more adapted

to the identification of small differences at each end of the scale.

The plant scores of the AV and the WP rating methods in 1990 were highly

correlated (r = 0.65, p < 0.001). AL measurements were in cm2 of nondefoliated leaf

area remaining, therefore not a measure of defoliation. As there was no absolute

measure of defoliation, it was not possible to make further comparisons with AV and

WP.

In 1991, the scores for cage 1 from the AV and WP methods were significantly

correlated with the AL method (r = 0.86, p < 0.001; and r = 0.88, p < 0.001,

respectively). The AV and WP methods correlated well with each other (r = 0.88, p

< 0.001). By squaring each of the correlation coefficients (r), both the WP vs. AV and

WP vs. AL relationships had r2 values of 0.77 meaning 77% of the variance in one rating

method can be explained by the variation in the other. The AV vs. AL correlation had

a r2 value of 0.74, indicating that 74% of the variance of the AV method can be

explained by the variation in the AL method (or vice versa).

Although the results from ratings on plants in cage 2 were similar, correlations

were not as high as those in cage 1. Scores from the AV and WP methods were

significantly correlated with the AL method (r = 0.75, p < 0.001; and r = 0.67, p <

0.001, respectively). The AV and WP methods were significantly correlated (r = 0.74,

p < 0.001) with each other. Calculated r2 values were: WP vs. AV r2 = 0.55, WP vs.








24
AL r2 = 0.49, and AV vs. AL r2 = 0.56. These values are lower than those found with

the cage 1 data. Differences may be attributed to differences in the plant growth at the

time of rating or to evaluator error. The cage 2 ratings were made in the rain which

hindered the evaluator's ability to accurately estimate defoliation. It was considered

more prudent to make ratings in the rain than to attempt to disregard the ensuing

regrowth which was increasing rapidly during a prolonged rainy period.

When data for both cages were pooled for analysis, scores for the AV and WP

methods were significantly correlated with the AL method (r = 0.79, p < 0.001; and

r = 0.78, p < 0.001, respectively). The AV and WP methods correlated well with each

other (r = 0.80, p < 0.001) as in the cages treated separately. The following r2 values

were obtained: WP vs. AV r2 = 0.64, WP vs. AL r = 0.60, and AV vs. AL r2 = 0.62

thus approximately 62% ( 2%) of the variation in one rating method could be attributed

to the variation in the other rating method.

In a similar study, a relative method and an absolute method were compared using

their estimates of defoliation of field plots infested primarily (90% of total insect

population) with the velvetbean caterpillar (G.R. Bowers, 1992, personal

communication). A 1-to-9 scale, with 1 signifying the least damage, was used as the

relative method. The absolute method followed a procedure described by Kogan and

Kuhlman (40) in which 20 leaflets each from the top, middle, and bottom thirds of plants

(for a total of 60) were sampled from plants located in the center two rows of a four-row

plot. These were then compared to diagrams of defoliated leaves to determine the

defoliation estimate of each sample. The two methods were significantly correlated, r








25
= 0.62, p < 0.001. Bower's correlation coefficients are somewhat similar to the pooled

correlation coefficients between the relative and absolute methods in this 1991 cage

experiment.

Correlations from the 1991 experiment, whether examined individually by cage

or combined, were all significant at p < 0.001. The magnitude of the r2 values indicate

that the evaluator obtained similar relative rankings of defoliation regardless the method

used. Although the correlations between the AL method and the WP and AV methods

were highly significant, it was not possible to determine if the evaluator estimated the

same amount of defoliation with the WP and AV methods (relative methods) as with the

AL method (the absolute method). To compare the accuracy of the evaluator using the

two relative methods and the absolute method, additional analyses were undertaken.

With the PROC MEANS procedure of PC-SAS, a paired difference comparison

was made between the WP and AV methods in 1990. A t-test was performed to test the

equality of the two methods based on the differences between the two methods for each

plant in the study. The mean difference of -0.36 between the WP and AV methods was

highly significant, p < 0.001, meaning the WP method yielded lower average percent

defoliation than the AV method. However, without some measure of the actual or

absolute defoliation across the plants, there was no way to determine if the evaluator

could obtain an accurate assessment of the true defoliation of a plant with either method.

A paired difference comparison was also made among the three rating methods

in 1991. The WP and AV methods had a mean difference of -0.55, p < 0.001, which

was somewhat similar in magnitude to the mean difference calculated in the 1990 study.








26
The mean differences standard deviation involving the AL method were 1.48 0.14

for WP AL and 2.03 0.12 for AV AL, respectively. Both differences were

significant at the p < 0.001 level. Although the evaluator estimated a higher average

percent defoliation with both relative methods than with the absolute method, the estimate

obtained with the WP method was numerically more similar to the measured defoliation

obtained with the AL method.

With the relative methods, WP and AV, the evaluator estimated somewhat similar

amounts of defoliation across all plants when compared with each other, even though the

difference between the two methods was highly significant. However, in comparisons

with an absolute method, AL, it becomes apparent that the evaluator's estimates of

defoliation with the WP method were closer to the actual than were the estimates of the

AV method. Although the AL method is accepted as the absolute measure of defoliation

for each plant, there is some evidence that it underestimated defoliation. Necrotic leaf

tissue due to insect injury was present. The leaf area meter could not distinguish viable

leaf tissue from necrotic tissue. The amount of necrotic tissue was not enough to

increase AL values to the level of WP value, however the best absolute estimate of

defoliation would have been to sample the entire plant (destructive sampling) and follow

the protocol of the AL method after cutting out necrotic tissue. In the 1991 experiment,

provisions were not made for destructive sampling; therefore it was not practiced.

McNemar's test for concordance was used to determine if the categorization of

plants having less than or equal to 50% defoliation and of plants having greater than 50%

defoliation differed with the WP and AV methods. In 1990, 36 of the 198 of the plants








27
were classified with the WP method as having at most 50% defoliation while with the

AV method these same plants were rated as having greater than 50% defoliation (Table

1). This percentage (18%) of plants was significantly different from zero, X2 = 34.03,

p < 0.001, supporting the earlier contention (Figure 2) that either the WP method

underestimated defoliation or the AV method overestimated defoliation. Without the

absolute method for reference it was not possible to determine which method was more

accurate.

All three methods were compared in 1991 to determine whether significant

differences among the methods existed. The results from McNemar's test for

concordance of the 1991 data are shown in Table 1. The two relative methods were

significantly different (X2 = 11.25, p < 0.001) with 20 of the 97 plants being classified

differently by WP than AV. The relative methods were both significantly different from

the absolute method (WP vs. AL X2 = 22.04, p < 0.001; and AV vs. AL X2 = 38.03,

p < 0.001) in the estimate of defoliation on a plant-by-plant basis. With both relative

methods, the evaluator had a tendency to overestimate defoliation as compared to the

absolute method. With the AV method 40 of 97 plants were classified as > 50%

defoliated while AL classified these same plants as < 50%, whereas with the WP

method only 24 of 97 plants were classified > 50% as compared to the AL method

classifying them as 5 50%. This indicates that the evaluator was able to estimate

defoliation closer to the absolute method with the WP method as was seen in the paired

difference comparison.










Table 1. McNemar test for significant differences between the WP (whole plant visual) and
AV (partitioned plant visual) methods in 1990; and between the WP and AV, WP and AL
(leaf area of the partitioned plant), and AV and AL methods in 1991.

1990
WP
AV 1 5t > 5t Totals X2
1-5 159 0 159
>5 36 3 39
Totals 195 3 198 34.03"'
1991
WP
AV 1-5 > 5
1-5 39 2 41
> 5 18 38 56
Totals 57 40 97 11.25'"
WP
AL 1-5 > 5
1-5 57 24 81
> 5 0 16 16
Totals 57 40 97 22.04'"
AV
AL 1-5 > 5
1-5 41 40 81
> 5 0 16 16
Totals 41 56 97 38.03**
*** Significant at the 0.001 level of probability.
t Defoliation ratings based on a 1-to-10 scale where 1 = 0 to 10% defoliation, 2 = 11 to
20%, 3 = 21 to 30%, 4 = 31 to 40%, 5 = 41 to 50%, 6 = 51 to 60%, 7 = 61 to 70%, 8
= 71 to 80%, 9 = 81 to 90%, 10 = 91 to 100%.








29
RV values for 1990 were 2.11 and 2.77 for WP and AV, respectively. According

to Pedigo (58), RV values less than 10 are desirable in research, but the closer the value

is to zero the more precise the method is at making estimates. Although the WP method

was numerically better than the AV method, the precision of both methods was

essentially equal and very good.

The RV values for each of the methods in 1991 were as follows: WP = 4.26, AV

= 3.49, and AL = 4.09. Based on the RV values, the AV method was numerically

better than either WP or AL. The AL method was expected to be the most precise since

it was the absolute measure of defoliation. However in this study, these values were

essentially equal and close to zero, thus the evaluator was fairly precise in making

estimates of defoliation with all three methods.



Summary and Conclusions

The objective of this study was to evaluate the effectiveness of the widely used

visual, or relative, assessment of insect defoliation as compared to a quantitative, or

absolute, assessment. In 1990 no absolute method comparisons were possible, however

the WP and AV methods were highly correlated (r = 0.65, p < 0.001). The evaluator

estimated a slightly lower amount of defoliation on the average with the WP method than

with the AV. The relative variation of these two relative methods were both less than

3.0 which according to Pedigo (58) is very desirable. Without an adequate absolute

method no determination could be made as to which relative method best reflected the

actual defoliation.








30
Adjustments in the method of calculating percent defoliation by the AL method

for 1991 allowed for direct evaluations of the two relative methods. All correlations

(WP vs. AV, WP vs. AL, and AV vs. AL) were significant at the p < 0.001 level,

whether the cage data were treated separately or pooled. The two relative estimates of

defoliation were significantly correlated with the absolute estimate of defoliation. These

results are similar to the findings of a another relative versus an absolute estimation of

defoliation conducted in 1985 by G.R. Bowers (G.R. Bowers, 1992, personal

communication).

Paired comparisons were made among the methods to test the equality of the

methods to rate defoliation. In 1991, the estimates of defoliation by all three methods

were significantly different. These findings were validated by McNemar's test for

concordance. The WP defoliation estimates were more similar to the AL results, than

the AV ratings were to AL.

Based on the RV's of each method, the precision of the AV and WP methods was

essentially equal in 1990. Similar results were obtained in 1991 with the AV, WP, and

AL methods. In both years the precision of the methods was essentially equal and well

below the value of 10 that Pedigo (58) states as being desirable. These results indicate

that either relative method is suitable for estimating defoliation.

The results of these studies demonstrate that relative estimates of defoliation can

be used effectively in genetic studies or routine screenings of advanced breeding lines.

Since the two relative methods yielded similar results, it would be advantageous to use

the most economical method.












CHAPTER 4
INHERITANCE OF RESISTANCE
TO SOYBEAN LOOPER IN SOYBEAN


Introduction

During the past three decades great strides have been made in the identification

and development of insect-resistant breeding lines of soybean (22,30,42,52,79,83). This

research has led to the release of several germplasm lines and cultivars (5,11,28,29) that

are resistant to phytophagous insects. Although advances have been made, very little

information has been published on the inheritance of resistance to foliar-feeding insects.

Sisson et al. (72) reported that resistance to the Mexican bean beetle was

quantitatively inherited with two or three major genes involved. In other inheritance

studies, Mebrahtu et al. (54) concluded that resistance to the Mexican bean beetle was

primarily an expression of additive genetic variance. Kilen et al. (36) suggested that

susceptibility to the soybean looper was partially dominant and a few major genes

conditioned resistance. Based on their work with PI 171451, PI 227687, and PI229358

and the FI progeny derived from the intercrosses among the three PI's, Lambert and

Kilen (45) suggested that the genetic basis for insect resistance in PI 171451 differed

from that in the other two PI's. This was further supported by their work with the three

PI's and the F3 lines derived from intercrosses among the three PI's (37). They

hypothesized that each PI contained a distinct resistance gene.








32
To facilitate the design of breeding programs to develop cultivars resistant to the

soybean looper, it would be beneficial to understand more completely the mode of

inheritance of this trait. Therefore, the objective of this study was to determine the mode

of inheritance of resistance to the soybean looper in genotypes derived from crosses

involving PI 229358.

Materials and Methods

All aspects of this experiment were conducted at the Mississippi State

University, Delta Research and Extension Center at Stoneville, MS on a Bosket fine

sandy loam (Mollic Hapludalfs).

1988

To study the inheritance of resistance to soybean looper in soybean, the breeding

line D86-3429 was crossed with the cultivar Braxton (32). D86-3429 has white flowers

(wl), gray pubescence (t), sensitivity to metribuzin [4-amino-6-(1,1-dimethylethyl)-3-

(methylthio)-1,2,4-triazin-5,(4H)-one] (hm), resistance to Phytophthora rot (induced by

Phytophthora megasperma Drechs. f. sp. glycinea T. Kuan & D.C. Erwin) (Rpsl-c), and

resistance to soybean looper. The donor of insect resistance was the germplasm line

D75-10169 (29), which derived its resistance from PI 229358 (83). The male parent,

Braxton, has purple flowers (WI), tawny pubescence (7), tolerance to metribuzin (Hm),

susceptibility to Phytophthora rot (rps), and susceptibility to soybean looper. Both

parents were of maturity group VII. Hand pollinations (1,60) made in the field produced

seven seeds of the cross D86-3429 x Braxton.










1989

The seven F, seeds produced in 1988 were space-planted 30.5 cm apart in 91.5-

cm rows in the F, nursery. Several rows of each parent, D86-3429 and Braxton, were

planted at a rate of 14 seeds per meter for crossing purposes. Four F, plants emerged

and these plants were used as the pollen parents for two backcrosses, D86-3429 x

F,(D86-3429 x Braxton) (BC1) and Braxton x FI(D86-3429 x Braxton) (BC2). The

original cross, D86-3429 x Braxton, and its reciprocal were also made to provide

additional F, seed for the following year. Approximately 60 hand pollinations were made

for the two backcrosses and the two crosses with the expectation that an adequate number

of seed would be produced. The two backcrosses, BC, and BC2, produced fourteen and

six seed, respectively. At maturity, the seed from the four F, plants were harvested to

provide F2 seed for use in 1990. In addition, seed from both parents were harvested for

use in 1990. Seven seeds were produced from the cross D86-3429 x Braxton and five

seeds were produced from the reciprocal cross. The seed of the backcrosses, BC, and

BC2, were scarified, treated with metalaxyl [N-(2,6-dimethylphenyl)-N-

(methoxyacetyl)alanine methyl ester] to optimize rapid emergence of healthy plants, and

planted one seed per pot in 15 L pots filled with soil treated with Bradyrhizobium

japonicwn inoculum in the greenhouse during the winter of 1989-90. Plants were

allowed to self-pollinate, thus providing seed of the BS, (the selfed generation of BC,)

and BS2 (the selfed generation of BC2) generations. At maturity the plants were

harvested and threshed with an ALMACO (Allen Machine Co., Nevada, IA 50201)

single-plant thresher.









1990

The inheritance study was conducted in a screened field cage 2.4 m high x 19.2

m wide x 32.0 m long (43). A schematic of the field cage design is shown in Figure

4. In mid-April the soil was disked twice in opposite directions with a disk harrow. On

23 May the soil was tilled with a spring-tooth harrow to smooth the surface. This was

followed with a preplant-incorporated herbicide treatment of clomazone 2-(2-

chlorophenyl) methyl-4,4-dimethyl-3-isooxazolidinone and trifluralin [2,6-dinitro-N,N-

dipropyl-4-(trifluoromethyl)benzenamine] at 1.12 kg ha-' and 0.84 kg ha', respectively,

on 24 May. The area within the cage was marked with rows spaced 68.5 cm apart to

be used as a guide for planting with push planters.

The seed for the study were planted on 25 May. A susceptible cultivar,

'Centennial' (27), was planted at 27 seed per meter of row in border rows as well as a

1 m section at the ends of each row of experimental material. Seed to produce

experimental plants were scarified, treated with metalaxyl, and space planted 25 cm

apart. The parents, D86-3429 and Braxton, were planted in three 12.7 m rows for a

total of 150 plants each. Also, 50 seed from each of the fourteen BC, and six BC2 plants

were planted in 12.7 m rows. Following the backcross progeny, five F, seed each of the

cross D86-3429 x Braxton and its reciprocal were planted. These were followed by the

360 F2 seed from the F, (D86-3429 x Braxton) plants grown in 1989. The insect-

resistant germplasm lines D75-10169 (29) and PI 229358 (83) were planted in the middle

and at the end of the experimental material for a total of 70 plants each.































/


iLJ'


i


,It


r
i
I
i








36
Normal production practices were used to maintain healthy plants. Soil was

cultivated on 7 and 20 June to control weeds, and bentazon [3-(1-methylethyl)-(1H)-

2, 1,3-benzothiadiazin-4(3H)-one 2,2-dioxide] was applied at 0.84 kg ha' on 19 June with

a hand held pneumatic sprayer. Supplemental irrigation was not required during this

time. A Saran cage top (43) was installed on 21 June as the final step in preparation for

the release of insects.

Approximately 4000 laboratory-reared (26) soybean looper moths were released

in the cage at 0830 h on 25 June. The absence of physical restrictions made all areas of

the cage equally accessible to the moths. Water was applied by furrow irrigation to

elevate the relative humidity within the cage to enhance egg laying. All plants were at

the V5 to V6 growth stage at the time of release. Larvae emerged from eggs after an

incubation period of approximately three days.

Larvae were allowed to feed until susceptible Centennial was determined to be 85

to 90 % defoliated which occurred on 6 July or 11 days after moths were released. At

that time, methomyl was applied at 0.5 kg ha' with a backpack sprayer to kill the

loopers and stop further feeding. On 9 July, the Saran cage top was removed to facilitate

regrowth by plants that were to be harvested for use in 1991, and all plants in the study

were evaluated for defoliation. Each plant was assigned a score of 1-to-10 (AV method)

based on the visual estimate of defoliation. The 1-to-10 rating scale represented 10%

increments, with a score of 1 representing defoliation of 0 to 10% and a score of 10

representing defoliation of 91 to 100%.








37

Ratings for the two parents, D86-3429 and Braxton, and their F, and F2 progeny

were grouped by their defoliation scores and plotted on a graph. The distribution of the

parental and F, populations were tested for normality following the procedure described

by Leonard et al. (49). Initially the defoliation data of these populations were

summarized in frequency distributions by using the upper class limits. The mean i,

variance s2, and standard deviation s of each population were calculated from the class

centers (i.e. 0.5, 1.5, 2.5, ... 9.5) using the following formulas:


S= n1(0.5) + n2(1.5) + ... + nio(9.5)/(ni + n2 + ... + n0)

s2 = n(0.5 i) + n1(1.5 i) + ... + nj(9.5 i)/(n.. 1)

s = (s2)12


where n, was the number of plants in each of the respective classes and n.. was the grand

total. Following the calculation of these variables, the variable x was calculated for each

class in the following manner:


x = (i upper class limit)/s


After determining the x for each class, a table of normal probability integrals was used

to determine the corresponding Z value of each class. The theoretical percentage of each

population was calculated by subtracting Z from 1.000 and multiplying by 100. The

theoretical number of each class was calculated as follows:


n.. x theoretical % = theoretical n








38
For each succeeding class, the cumulative percentage in all prior classes must be

subtracted to obtain the theoretical percentage to calculate the theoretical n of that class.

The observed distribution versus theoretical distribution was tested with a X2 test. This

was done to gain insight into the mode of inheritance of resistance to soybean looper and

to facilitate the design of the 1991 experiments. The information from these four

populations was used to determine what additional crosses, if any, would be required to

fulfill the objectives of this study.

The original cross, D86-3429 x Braxton, was repeated to generate seed to be

used as the FI generation in 1991. The two backcrosses, BCi and BC2, were repeated

for the same purpose as the original cross, as well as to provide seed for planting in a

winter nursery to generate the BSi and BS2 populations, the selfed generations of the

backcrosses. The bulk of the pollinations were made on plants within the cage. The

remainder were on plants located in the F2 nursery. Plants within the cage were irrigated

as needed throughout the remainder of the growing season to insure adequate seed set on

F1 and F2 plants for use the following year.

Twelve seeds were produced from the D86-3429 x Braxton cross. The two

backcrosses, BC, and BC2, yielded 49 and 29 seeds, respectively. The 12 F, seeds were

combined with remnant D86-3429 x Braxton F, seed generated from the 1988 and 1989

crosses for use in 1991. Twelve seeds and 11 seeds, from BC, and BC2, respectively,

were scarified and treated with metalaxyl and sent to the winter nursery at Mayaguez,

Puerto Rico to produce the BSi and BS, generations. The single plants from the various

generations in the cage were harvested and threshed with a single-plant thresher between








39
31 October and 2 November. The soil in the cage was subsoiled on 9 November in

preparation for the 1991 experiment.

1991

Two field cages were used to complete the inheritance study. Soil in each cage

was disked with a disk harrow and treated with trifluralin at 0.84 kg ha-, which was

incorporated with a spring-tooth harrow on 26 April. To control a broad spectrum of

weeds, a tank mix of clomazone at 1.12 kg ha' and imazaquin [2-(4,5-dihydro-4-methyl-

4-(1-methylethyl)-5-oxo-1H-imidazol-2-yl)-3-quinolinecarboxylicacid] at 0.11 kg ha1 was

applied and incorporated with a spring-tooth harrow on 13 May. On 25 May, soil was

reworked with a spring-tooth harrow and rows spaced 68.6 cm apart were marked to

serve as guides for the push planter.

Each row consisted of 10 plants spaced 25 cm apart except for the Fi, BC,, and

BC, generations, which had only one plant per row. There were three rows, consisting

of 10 plants each, of the insect-resistant germplasm lines D75-10169 and PI 229358.

The planting order for each cage was generated by randomizing the 277 rows with PC-

SAS PROC PLAN (68). Each cage contained the entire complement of plants and was

treated as a replication over environments.

The border rows were planted to Centennial at 27 seed per meter in each cage.

The rows of experimental populations were centered 68.6 cm apart. Rainfall of 43 mm

halted planting on 25 May in cage 1 after one third of the rows were planted. The

remainder of cage 1 and all of cage 2 was lightly tilled with a spring-tooth harrow on 29

May. The remaining rows in cage 1 were planted on 29 May. Approximately 1 m of








40
Centennial was planted at the ends of all rows to serve as a border. The experimental

populations were surrounded by a susceptible cultivar.

On 31 May, soil in cage 2 was rolled with a water-filled roller pulled by a four-

wheel all-terrain vehicle. This operation smoothed and firmed the seed bed, which

decreased the rate of evaporation of moisture from the soil. The rows were then marked

as in cage 1. The susceptible cultivar Centennial was planted by the same method as in

cage 1. The entire area within cage 2 was planted on 31 May. Short sections of the

Centennial border rows were replanted on 14 June.

Soil in the cages was cultivated on 4, 14, 20, and 26 June with a self-propelled

garden tiller. Supplemental soil moisture was provided by furrow irrigation on 4 and 21

June to ensure that the plants would have sufficient vegetative growth at the time of

insect release. The Saran tops were put on the cages on 26 June as final preparation for

the release of insects.

Approximately 4000 laboratory-reared (26) soybean looper moths were released

in each cage at 0800 h on 8 July. A rainfall of 19 mm on 4 July provided sufficient soil

moisture to maintain an elevated relative humidity within the cages during the egg-laying

period. At the time of release, the plants were at the V7 to V8 stage of growth in cage

1, and V6 to V7 in cage 2 (17). Larvae emerged after an incubation period of

approximately three days.

Larvae were allowed to feed in each cage until 22 July, at which time Centennial

was approximately 95% defoliated. The cage tops were then removed to facilitate the

defoliation ratings and to promote regrowth of the plants. Both cages were sprayed with








41

methomyl at 0.5 kg ha'- to kill larvae. This insured that additional feeding would not

occur once the ratings had begun. Cage 1 was rated on 23 and 24 July using the AV

rating method. This method used the same 1-to-10 scale previously described for the WP

method in 1990. In the AV method the plants were visually partitioned into three

sections (top, middle, bottom). Each section was scored for defoliation and these three

scores were averaged to get the mean score for each plant. This method enabled the

evaluator to make a more detailed assessment of defoliation by the soybean looper on

each plant. Cage 2 was rated during a rainfall by the same method on 25 and 26 July.

All ratings in both cages were made by the same evaluator.

Generation means and variances were calculated for each of the nine generations

on a cage basis. These means and variances were employed in Mather's scaling tests

(51) to determine the adequacy of an additive dominant model and to test for epistasis.

The formulas for the different scaling tests were as follows:


A = 2RBC1 XPI RxF VA = 4ViBCI + ViPI + ViFI

B = 2xBC2 RP2 F, VB = 4VxBC2 + VXP2 + ViFi

C = 4XF2 2iF, ixP iP2 Vc = 16ViF2 + 4ViFi + ViP, + ViP,

D = 8RF3 3iPi 3iP2 2iF1 VD = 64VRF3 + 9VixP + 9VPP2 + 4VRFI


where i was the mean of the respective generations and Vi was the variance of the mean

of the same respective generations. The variance of the mean of each generation was

calculated as Vi = o2/n, where o2 was the variance of the generation and n was the

number of individuals observed in that generation.








42
To determine if each of the scaling tests had the expected value of zero, each was

tested in the following manner:

tA = A/(VA)C

t, = B/(VB,)

tc = C/(Vc)'

to = D/(VD)H

where (VA)", (VB)', (Vc)', and (VD)" were the standard errors of each tests. The t

values obtained were compared to a table of t distribution (76) to determine the level of

significance for each of the tests. The probability was found in the table of t using the

sum of the degrees of freedom of each generation in each test as the number of degrees

of freedom.

Since the preliminary results indicated that resistance to soybean looper is

inherited quantitatively, it was necessary to select an appropriate method of analysis that

would allow for a detailed study of the inheritance of resistance. After reviewing the

statistical methods available for the analysis of a quantitative trait, a generation means

analysis was selected. This analysis was chosen over other methods because of the

following advantages (23):

1. Errors are inherently smaller since means (first order

statistics) are used, rather than variances (second order

statistics);

2. Smaller experiments allow the same degree of precision;










3. Additive (a), dominant (d), and epistatic (aa, ad, and dd)

effects are estimated using means rather than variances;

and

4. It is applicable to both cross- and self-pollinating crops.

In utilizing the generation means analysis, it is assumed that the parents are homozygous

and there is no linkage among genes influencing the trait being studied. If this

assumption is correct and it appears that the trait in question is not quantitatively

inherited, these generations can be used to analyze for Mendelian ratios.

Hallauer and Miranda (23) cited the following disadvantages to the generation

means analysis:

1. An estimate of heritability cannot be obtained; and

2. Genetic advances cannot be predicted because genetic

variances are not available.

Although these disadvantages are inherent to the generation means analysis, they are of

little consequence since additional calculations can be made to obtain the variances

necessary to estimate heritability.

Generation means analysis, as described by Hayman (31), has been used

extensively in other crops such as corn (21,34,70), cotton (Gossypium hirsutum L.)

(55,56), and wheat (8). In most cases, Gamble's (21) notation is used in conjunction

with Hayman's (31) methodology. Generation means analysis apparently has not been

used previously in soybean to analyze quantitative traits. Because of the advantages of








44

this analysis, it was chosen to determine the mode of inheritance of resistance to soybean

looper.

Populations used were derived from the cross D86-3429 x Braxton (Figure 5).

A minimum of six populations are required in a generation means analysis with a six

parameter model, but this experiment was expanded to include nine populations. The

number of individuals comprising each generation are listed in Table 2. To determine

the inheritance of resistance to soybean looper in soybean according to Hayman's (31)

model, programs were written in SAS to analyze the data by cage (see Appendix C) and

combined over cages where cages were treated as blocks (see Appendix D). Using

Hayman's methodology and Gamble's (21) notation, the models for the generation means

analysis were as follows:

P, = m + a d/2 + aa ad + dd/4
P2 = m a d/2 + aa + ad + dd/4
F, = m + d/2 + dd/4
F2 = m
BC, = m + a/2 + aa/4
BC = m a/2 + aa/4
F3 = m d/4 + dd/16
BSi = m + a/2 d/4 + aa/4 ad/4 + dd/16
BS2 = m a/2 d/4 + aa/4 + ad/4 + dd/16


where m = the overall mean, and a, d, aa, ad, and dd represent the additive, dominance,

additive x additive, additive x dominance, and dominance x dominance genetic effects,

respectively.

Each SAS program fitted two regression models which were set up using matrix

notation according to the procedures outlined by Jennings et al. (34). The first












D86-3429 (P)




x




BCI


BS
BS,


Braxton (P2)




x




BC2




BS2


the genetic


Figure 5. Mating scheme to derive populations required to estimate
effects of resistance to the soybean looper in soybean.










Table 2. Populations used for generation means analysis of the inheritance of resistance to
the soybean looper.
Row Plants Total
Population Gen. Number per Rowt Plants
D86-3429 P1 1 11 10 110
Braxton P, 12-22 10 110
D86-3429 x Braxton F, 23 -35 1 13
D86-3429 x Braxton F2 36 85 1 500
D86-3429 x Braxton F3 86 210 10 1250
D86-3429(2) x Braxton BCi 211 -224 1 14
D86-3429(2) x Braxton BS, 225- 249 10 250
Braxton(2) x D86-3429 BC2 250 255 1 6
Braxton(2) x D86-3429 BS2 256 271 10 160
D75-10169 272 274 10 30
PI 229358 275 277 10 30
tRows were 2.5 cm long except where there was only 1 plant/row.
tInsect-resistant germplasm lines used as checks.








47

regression model, defined as Model 1, consisted of the three parameters m, a, and d.

The second regression model, defined as Model 2, consisted of the epistatic effects, aa,

ad, and dd, in addition to the parameters in Model 1. Model 2 is used only if a

significant additive or dominant effect is detected and to determine if significant epistatic

effects exist that are contributing to the significance in Model 1. To adjust for the

unequal population sizes comprising each generation, the models were weighted using

reciprocals of the standard errors of the generation means as suggested by Rowe and

Alexander (65).

In addition, Powers' partitioning method of genetic analysis (49,61) was

incorporated to test proposed genetic models. Initially the frequency distributions of the

PI, P2, FI, and F2 generations were tested for normality by testing the observed versus

theoretical distribution with a X2 test following the procedure previously outlined.

Frequency distributions (observed and theoretical) of the F2 population were converted

to percentages in each class in the following manner:


%(in each class) = n,/n..


where ni was the number of plants in each class and n.. was the total number of plants

in the F2 population. After conversion, the observed distribution of the F2 population

was examined for modes. Finally, the F2 population of each cage was tested for

goodness of fit to a two-gene additive model. The proposed genetic ratio for the model

was as follows: 1:2:1:2:4:2:1:2:1. To test the model, the ratio was simplified to 1:14:1

by combining the defoliation classes in the two cages as follows: 2-4, 5-8, and 9-10 in








48
cage 1, and 2-3, 4-8, and 9-10 in cage 2. The end classes were defined by a distinct

transition between defoliation class in the distribution of the F2 population of each cage.

The homozygous populations D86-3429, Braxton, D75-10169, and PI 229358

were used to determine if there was a significant location effect within each cage. An

analysis of variance was performed on each population to obtain the row and plant within

row mean squares. Estimates for plant to plant variability within a row (o2) and the

additional variability of plants in different rows (or) were obtained from the expected

values of the mean squares. Also, F-tests for significant row effects (Ho:o2r=0) were

performed on each of the homozygous generations within each cage.

An estimate for the number of genes (n) involved in resistance to soybean looper

was obtained for the progeny in each cage by the formula:


n = (iPi iP2)2/8[(oF2) ((v1o2Pi + v2P2 + v3o2F)/v1 + v2+ v)]


In the formula, iP, and xP2, were the mean defoliation ratings for the parents, and o0F2,

02P1, 0P2, and o2Fi were the variances of the respective generations. This formula is

a modification of the formula presented by Poehlman (60). In the modification, o2F, is

replaced with (vo2P+v22P2 +v 32F,)/(v+ v2+v3). This modification was used in order

to better estimate the environmental variance, since o2Fi, a2P1, and o2P2 were essentially

estimates of environmental variation. The degrees of freedom (v,, v2, and v3) for the

associated generations were employed to lessen population size effects on the estimate

of environmental variance. This method of estimating genes assumes that the genes have

equal effects without significant dominant or epistatic effects.








49
Heritability of insect resistance, in the broad sense, was estimated for the progeny

in each cage from the formula:


H = (Vo/Vp) x 100%


where Vo was the genetic variance and Vp was the phenotypic variance. The genetic

variance was:


V, = VP VE


where VE was the environmental variance. The VE was estimated by obtaining the

weighted average of the variances of the F, population and the populations of the parents

P, and P2:


(vor2P, + v2o2P2 + v32F,)/(v, +v2+v,)


In the formula the respective degrees of freedom v were used to weight the variances of

each population. The Vp was estimated by utilizing the variance of the F2 population.

Results and Discussion

1990

The defoliation data of the PI, P2, Fi, and F2 populations from the preliminary

cage study are presented in Table 3. Also presented in this table are the defoliation

ratings of PI 229358 and D75-10169, from which D86-3429 derived its resistance to

soybean looper. D86-3429 was visibly more resistant than the original source of










Table 3. Ratings of leaf feeding by soybean looper on soybean parents, their F, and F2
progeny, and the germplasm lines PI 229358 and D75-10169 in the field cage-1990.
Upper class limits of leaf feeding ratings
1 2 3 4 5 6 7 8 9 n
D86-3429(PI) 4 60 64 11 2.6 139
Braxton(P2) 1 32 48 38 19 1 6.3 139
PI x P2(F) 5 3.0 5
P2 X PI(Fi) 1 4 3.8 5
PI x P2(F2) 1 33 78 154 39 6 2 1 3.7 314
PI 229358 21 25 1 1 3.6 48
D75-10169 4 34 18 3.3 56
tl = 0 to 10% defoliation, 2 = 11 to 20%, 3 = 21 to 30%, 4 = 31 to 40%, 5 = 41 to
50%, 6 = 51 to 60%, 7 = 61 to 70%, 8 = 71 to 80%, 9 = 81 to 90%, 10 = 91 to
100%. None of the plants received a rating of 10.








51
resistance, PI 229358. This suggests that additional genes for resistance might be

contributing to the increased level of resistance.

Resistance to soybean looper was interpreted as being quantitatively inherited

based on the defoliation data of the Pi, P2, FI, and F2 populations (Table 3). Distribution

of the F2 population (n=314) was continuous, with a mean defoliation rating of 3.7

which was slightly less than the parental midpoint of 4.5. Although a high number of

individuals fell in the fourth rating class (31 to 40% defoliation), the F2 population

appeared to have a normal distribution. The PI, P2, and F2 population distributions were

tested for normality using the procedure outlined by Leonard et al. (49) (see Appendix

E). Both parental populations had low X2 values (Table 4) indicating they were normally

distributed. The frequency distribution of the F2 ratings failed the normal distribution

test (Table 4). Rating class 4 had too many plants, whereas classes 3 and 5 had too few

(Table 3 and Appendix E).

Although the defoliation ratings of the F2 population did not fit a normal curve

the distribution was continuous with only one mode. Therefore use of a method of data

analysis for quantitative traits seemed appropriate in order to define the mode of

inheritance of soybean looper resistance. A generation means analysis (31,34) was

chosen and additional crosses were made to supply the generations necessary for the

experiment.

1991

Larvae were allowed to feed until susceptible Centennial was 95% defoliated.

Plants were found in all rating classes except class one. All populations receiving larger










Table 4. Goodness of fit test for normality of soybean looper defoliation for
populations of D86-3429, Braxton, and D86-3429 x Braxton F2 populations in the 1990
cage study.
Population Degrees of Freedom X2 P
D86-3429 3 3.6628 0.50- 0.25
Braxton 5 7.1877 0.50-0.25
F2 7 22.3204 < 0.005
t Test follows the procedure of Leonard et al. (49).








53

mean defoliation scores in 1991 (Table 5) than in 1990 (Table 3), when larvae were

killed when Centennial was only 85% defoliated.

In 1991 the AV rating method was substituted for the WP method used in 1990.

The AV method accounted for feeding preferences of soybean looper (24). It was

expected to yield a better estimate of defoliation than the WP method, because each score

of the AV method was the average of three observations per plant, whereas the WP

method was based on a single observation. The AL method was too time consuming to

be used in an experiment of this magnitude with the resources available.

Ratings of defoliation by soybean looper on the parents, Pi and P2, and their F,,

F2, and F3 progenies grown in cages 1 and 2 are shown in Table 5. The F2 and F3

progenies, in both cages, exhibited a normal distribution for ratings of defoliation. In

both cages the F, population exhibited a bimodal distribution instead of the expected

single mode. All of the FI plants exhibited purple flowers and tawny pubescence, ruling

out the possibility that the bimodal distribution resulted from selfs in D86-3429.

The Fi defoliation data distribution was expected to exhibit a narrow range and

single mode intermediate to the two parental modes. Instead the distribution had a wide

range and was bimodal in both cages (Table 5). Since the Fi data did not fit expected

distributions, it was suspected that there was a location effect within each cage interfering

with the defoliation ratings. The four homozygous populations (D86-3429, Braxton,

D75-10169, and PI 229358) included in the study were tested to determine if there was

a significant location effect within each cage. A significant location effect was detected

for the populations of D86-3429, Braxton, and D75-10169 in each cage (Table 6).










Table 5. Ratings of leaf feeding by soybean looper on soybean parents, their F1,
BSI, BC2, and BS2 progeny, and the germplasm lines PI 229358 and D75-10169


F2, F3, BCI,
in the field


cages-1991.
Upper Class Limits of Leaf feeding ratings
2 3 4 5 6 7 8 9 10 it n


D86-3429(P,)
Braxton(P2)
P, x P2(F1)
PI x P2(F)
P x P2(F3)
P, x (PI x P2)(BCI)
P1 x (PI x P2)(BS,)
P2 x (PI x P2)(BC2)
P2 x (PI x P2)(BS2)
PI 229358
D75-10169

D86-3429(P,)
Braxton(P2)
P1 X P2(F,)
Pl X P2(F2
P1 x P2(F3)
PI x (P1 x P2)(BCi)
PI x (PI x P)(BSI)
P2 x (PI x P2)(BC2)
P2 x (P, x P2)(BS2)
PI 229358
D75-10169


Cage 1
3 30 44 19 2
2
1 2 2
1 8 31 58 102
7 37 118 194 232


5
2 34



1
2 7


8 32 43
1
1 2 1
2 17 43


28
2
5 23


103
5
59


1
3
3 9


1
19

10
11
2
Ca2e 2


2
58 82
89 238
3 2
51 37

5 12
16 8
9 6


18
6
111
201
1
8

13
1


5
3
120
273
2
16

17
1


3.5
30 40 4 8.0
5.8
67 36 4 6.2
148 131 18 6.0
2 4.4
6 4.2
3 1 7.9
52 45 9 7.9
5.1
3.6


93 30 5
187 114 21

7 1
2
33 43 11


tBased on the mean of three observations per plant.


1 = 0 to 10% defoliation, 2 = 11 to


20%, 3 = 21 to 30%, 4 = 31 to 40%, 5 = 41 to 50%, 6 = 51 to 60%, 7 = 61 to 70%, 8 =
71 to 80%, 9 = 81 to 90%, 10 = 91 to 100%. None of the plants received a rating of 1.
:Derived from the mean defoliation of all plants without regard of the upper class limits.


98
94
11
418
1086
14
194
4
130
21
28

97
94
12
450
1153
14
199
2
122
28
27










Table 6. Estimates of variance components obtained from an analysis of the homogeneous
populations, D86-3429, Braxton, D75-10169, PI 229358, for a location effect due to
feeding by soybean looper in field cages in 1991.
F
Population do2,* 2 + 2,1' Ho:e,=0


Cage 1
D86-3429 (PI) 0.31 0.72 12.55"'
Braxton (P) 0.25 0.84 20.78"*
D75-10169 (BD) 0.29 1.50 40.29"*
PI 229358 (D) 0.58 0.96 5.37'
Cage 2
D86-3429 (P,) 0.37 0.54 5.03**
Braxton (P) 0.69 0.80 2.35'
D75-10169 (BD) 0.42 0.94 12.03'"
PI 229358 (D) 0.39 0.41 1.41
Significant at the 0.05 and 0.001 probability levels respectively.
t Based on defoliation ratings made with a 1 to 10 scale divided into 10% increments with
a 10 reflecting 91-100% damage.
* Variance of plants within same row.
* Variance of plants among rows.








56

The location effect for the PI 229358 population was significant (p < 0.05, Table 6) in

cage 1, but was nonsignificant in cage 2. Although the population size of PI 229358 is

approximately the same for each cage, different randomization plans would produce

different results. It is apparent that the defoliation ratings were affected by location

within the cages.

The mean ratings for soybean looper defoliation and their standard errors for the

nine generations grown in each cage are presented in Table 7. In addition to the

observed mean ratings for defoliation, the predicted means determined from relationships

described by Mather and Jinks (51) are also presented. In all instances except the BC2

generation in cage 2, the predicted means were within + one standard deviation of the

observed means in each cage. The observed mean for the BC2 generation in each cage

was larger than the predicted mean. This deviation from the predicted mean could be

associated with the small sample size available for defoliation ratings. Variances of the

generations from each cage are presented in Table 7, and are used later in Mather's

scaling test (51) and to determine heritability.

The results of Mather's scaling test (51) for the cross D86-3429 x Braxton are

shown in Table 8. According to Mather and Jinks (51), if the values of A, B, C, and

D do not differ from zero, then additive gene effects are indicated. The significance of

any of these scales indicates the presence of epistasis. In cage 1, nonsignificant values

for A, B, C, and D, indicated that an additive-dominance model was adequate. Cage 2

data yielded similar t-tests for A, C, and D, supporting the conclusions from cage 1.

The B scale test could not be applied to cage 2 since the BC2 population had a variance










Table 7. Means, standard errors, and variances of soybean looper defoliation ratings for
parental lines and seven populations of soybean arising from the cross D86-3429 x
Braxton grown in field cages at Stoneville, MS in 1991.
Mean
-------- :i


Population Gen.


D86-3429
Braxton
P1 XP2
P x P2
Pi x P2
Pi x (PI
Pi X (PI
P2 X (P1
P2 X (PI:


P1
P2
F1
F2
F3
BCI
BSI
BC2
BS2


D86-3429
Braxton
P x P2
PI XP2
P1 XP2
Pi x (PI x P2)
P, x (PI x P2)
P2 X (PI X P,)
P2 x (P1 x P2)
tBased on relationships
fStandard error


P,
P2
F,
F2
F3
BCi
BS,
BC2
BS2
described


Number Actual
Cage 1
98 3.49
94 7.97
11 5.76
418 6.19
1086 6.02
14 4.36
194 4.22
4 7.92
130 7.89
Cage 2
97 3.34
94 8.28
12 5.42
450 6.13
1153 6.18
14 4.45
199 4.52
2 9.33
122 7.69


Predt S.E.t a2


5.73
5.75
5.96
4.63
4.62
6.87
6.86




5.81
5.62
5.97
4.38
4.48
6.85
6.95


0.084
0.092
0.427
0.072
0.052
0.524
0.087
0.394
0.090

0.074
0.092
0.617
0.074
0.046
0.376
0.093
0.000
0.117


by Mather and Jinks, 1971.


0.685
0.797
2.002
2.160
2.916
3.837
1.452
0.620
1.055

0.528
0.788
4.568
2.490
2.478
1.976
1.708
0.000
1.655








58

Table 8. Mather's scaling test applied to the cross D86-3429 x Braxton to test adequacy of
additive-dominance model for resistance to soybean looper.

Test Cage 1 Cage 2

tA -0.125 0.039
tB 0.918
tc 0.268 0.267
tD 0.157 0.273








59
of zero. Since C was not significantly different from zero in either cage (Table 8), this

indicated that generation means depended only on the additive gene effects and no

epistasis was present.

The results of the generation means analysis of each cage are reported in Table

9. Table 10 reports the results of the combined analysis where the cage effects are

treated as blocks. Significant negative additive effects were detected with Model 1,

which measures only additive and dominant effects, for both Cages 1 and 2 (p < 0.06

and p < 0.02, respectively; Table 9) and the combined cages (p < 0.001, Table 10).

The sign of this effect is dependent upon the parent designated as P,. The sign of the

effect is a function of the fact that an inverted scale was used and also a reflection of the

relationship between the mid-parent and the means of the F,, F2, and F3 generations

indicating which parent was contributing to the additive variation (51). In Cage 1 (Table

7) the means of the Fi, F,, and F3 generations were between the mid-parent (5.73) and

P2. The means of the F, and F3 in Cage 2 were skewed away from the mid-parent (5.81)

in the same direction, but the F, mean was oriented towards the other parent. Although

the Fi's mean was on the opposite side, it was within one standard deviation of the mean.

All of the progeny means skewed towards P2, in addition to a significant negative

additive effect indicated that Braxton might be contributing to the additive variation.

Since it has been reported that Braxton does have some level of resistance (64), it stands

to reason that it would contribute to the additive effect.

The full model, Model 2, was then fitted to estimate the epistatic effects as well

as the remaining additive and dominant effects. None of the epistatic (aa, ad, and dd),










Table 9. Estimates of the additive, dominant, and epistatic effects in the generation
means for defoliation by soybean looper in the nine populations of the D86-3429 x
Braxton material grown in the field cages at Stoneville, MS in 1991.
Standard
Effects Estimate error t value
Cage 1
Model 1
F2 mean (m) 5.503 0.655 8.399"*
Additive (a) -1.830 0.805 -2.273t
Dominance (d) -1.287 2.071 -0.621
Model 2
F2 mean (m) 5.839 0.775 7.538"
Additive (a) 0.859 2.525 0.340
Dominance (d) -2.639 3.103 -0.851
Additive x Additive (aa) -2.182 2.642 -0.826
Additive x Dominance (ad) 3.212 2.737 1.174
Dominance x Dominance (dd) 3.484 9.370 0.372
Cage 2
Model 1
F2 mean (m) 5.387 0.575 9.374*"
Additive (a) -2.206 0.667 -3.308'
Dominance (d) -1.385 1.796 -0.771
Model 2
F2 mean (m) 5.613 0.590 9.507"
Additive (a) -0.031 1.929 -0.016
Dominance (d) -2.673 2.525 -1.059
Additive x Additive (aa) -2.191 1.953 -1.122
Additive x Dominance (ad) 2.534 2.063 1.228
Dominance x Dominance (dd) 4.302 7.742 0.556
,","" Significant at the 0.05, 0.01, and 0.001 probability
levels,respectively.
t Significant at the 0.06 probability level.










Table 10. Estimates of the additive, dominant, and epistatic effects in the generation
means for defoliation by soybean looper in the nine populations of the D86-3429 x
Braxton material grown in the field cages at Stoneville, MS in 1991.
Standard
Effects Estimate error t value
Cages 1 and 2 Combined
Model 1
F2 mean (m) 5.414 0.503 10.760"'
Additive (a) -2.020 0.487 -4.150"*
Dominance (d) -1.327 1.281 -1.040
Model 2
F2 mean (m) 5.690 0.418 13.630'"
Additive (a) 0.424 1.165 0.360
Dominance (d) -2.669 1.468 -1.820
Additive x Additive (aa) -2.195 1.197 -1.830
Additive x Dominance (ad) 2.883 1.254 2.300'
Dominance x Dominance (dd) 3.900 4.464 0.870
*,*" Significant at the 0.05 and 0.001 probability levels,respectively.








62
additive, or dominant effects were significantly different from zero in either cage.

Although unexpected, ad was significant (p < 0.05) in the combined analysis and neither

the additive or dominant effects were significant. It is interesting and informative that

ad effect is significant. However, it introduces a dimension that cannot be examined

further with currently available data. In Model 1 the epistatic effects are included in the

additive effects. For this reason, Model 1 will be assumed to treat the data adequately.

With an indication that inheritance of resistance to soybean looper was controlled

by an additive genetic system, the data from each cage were tested to fit a two-gene

model. Since no modes were evident in the F2 or F3 populations in either cage (Table

5), Power's partitioning analysis was employed to analyze the data and test the predicted

gene models following the examples of Deren and Quesenberry (12) and Sage and de

Isturiz (67).

Initially, the frequency distributions of the defoliation data from the parental, F,,

and F2 populations from each cage were tested for normality (Table 11). As was seen

in 1990, both parental populations had low X2 values which indicated that these

populations were normally distributed about a central mean. In cage 1, the Fi population

did not have a normal distribution, but the F2 population's distribution did fit the normal

curve. The opposite situation occurred with the Fi and F2 populations in cage 2. The

differences between the two cages with respect to the F, and F2 populations may be due

to a location effect present within the cages.

In the hypothesized two-gene additive model the middle classes were combined

resulting in three classes for the model to be tested. Partitioning the middle classes










Table 11. Goodness of fit test for normality of soybean looper defoliation data from 1991
cage study.

Population Degrees of Freedom X2 P

Caege 1


1.519

7.767

10.667

6.528


0.211

4.584

4.000

26.599


0.90 0.75

0.25- 0.10

0.025 0.010

0.75 0.50



0.990 0.975

0.500 0.250

0.75 0.50

0.005 >


D86-3429


Braxton

F,

F2


Cage 2


D86-3429


Braxton

F1

F2








64
would have been arbitrary, whereas the end class ratios were easily manipulated to test

the model. The F2 data for each cage were tested for goodness of fit to the ratio of

1:14:1 for the two-gene model. The chi-square probabilities were as follows: cage 1,

0.005 > p; and cage 2, 0.10 > p > 0.05. These probabilities are reflective of the test

for normality of the F, defoliation data in both cages as shown in Table 11. The higher

the probability of a normal distribution of the F2 defoliation data, the lower the

probability that the F2 defoliation data fit the two-gene model. Neither F2 population fit

the two-gene model well. The three-gene model gave a better fit. However, because

there is good reason to believe that the scores assigned in this experiment should not be

assumed to correspond to genotype effects, the Power's partitioning analysis probably has

no utility in these analyses.

An estimate for the number of genes, n, in this cross contributing to the

expression of resistance to soybean looper was obtained for both cages using a

modification of the formula presented by Poehlman (60). The number of genes

contributing to the expression of resistance to the soybean looper in the cross D86-3429

x Braxton was estimated at n = 1.87 and n = 1.91, cages 1 and 2, respectively.

Broad sense heritability was estimated for resistance to soybean looper in the

progeny from both cages. Although it has already been concluded that resistance to

soybean looper is controlled by additive gene action, narrow-sense heritabilities could not

be calculated because of the size and variances of the BCI and BC, populations in each

cage. Although broad sense heritabilities were calculated, it is assumed that these

heritabilities are attributed to only the additive genetic variance and are in fact narrow-








65
sense heritabilities. Since the F, generations in both cages exhibited a bimodal

distribution for defoliation scores, the environmental variance was estimated by the

following formula:

(vP,2P, + vpP,2 + vF,2Fl)/(vp, + v,2 + VF)

as was done in estimating the number of genes. This gives a better estimate of the

environmental variance than just using the Fi because the number of individuals

comprising the parental generations were relatively large (Table 5). Also, parents are

genetically uniform and should only exhibit variability due to the environment. The

heritabilities were estimated at 62.3% for cage 1 and 64.3% for cage 2. As has been

demonstrated in all previous analyses the results from the two cage studies are in

agreement for the estimate of broad sense heritability. Based on the high heritabilities

for resistance to soybean looper, it is apparent that the phenotypes expressed in this

population were determined by the genetic variation more than by the environmental

variation.

Although the estimates for both cages are in agreement, the estimate is probably

an underestimation of the total number of genes contributing to soybean looper resistance

because these data only estimate the number of genes controlling soybean looper

resistance by which D86-3429 and Braxton differ. In both cages, Braxton exhibited a

low level of resistance with 80% defoliation, whereas Centennial was almost completely

defoliated. When compared to Centennial, which is known for its high susceptibility

(47), Braxton has repeatedly demonstrated 15 to 20% less defoliation. Since the

inception of this study, resistance of Braxton to foliar-feeding insects has been








66
documented by Rowan et al. (64) in a study to compare the resistance of 46

recommended soybean cultivars in Maturity Groups V, VI, VII, and VIII. In that study,

Braxton was cited as being one of the most resistant cultivars in Maturity Group VII.

As in 1990, the insect-resistant germplasm lines, PI229358 and D75-10169, were

included in both cages to provide additional checks and to document the level of

resistance in the resistant parent, D86-3429. In both cages, D86-3429 was visibly and

statistically (Table 12) more resistant than the original source of resistance, PI 229358.

D86-3429 also exhibited more resistance than did D75-10169 in both cages, although it

was significantly different at p = 0.01 (LSD) only in cage 2 (Table 12). This suggests

that during the five cycles of crossing and selection that occurred in the development of

D86-3429 from PI 229358, additional genes were gained from 'Tracy-M', a parent in the

fourth cycle. Tracy-M has subsequently been identified as being measurably resistant to

foliar-feeding insects (47).

Summary and Conclusions

This research was conducted to determine the mode of inheritance of resistance

to soybean looper in soybean. The results from Mather's scaling test (51) applied to the

data from each cage study indicated that generation means depend only on the additive

gene effects for resistance to soybean looper. Utilizing Hayman's (31) methodology in

terms of the generation-means analysis to analyze the data from each cage separately,

only the additive genetic effects were significant. When the data from the two cages

were combined for analysis the ad epistatic effect was significant (t = 2.30, p < 0.005).

However, the primary effect is assumed to be additive based on the results of the analysis










Table 12. Average defoliation scores in soybean populations resistant to soybean looper in
1991 cage study.

Population n Meant

Cage 1

D86-3429 98 3.49 a

D75-10169 28 3.64 a

PI 229358 21 5.13 b

Cage 2

D86-3429 97 3.34 a

D75-10169 27 4.35 b

PI 229358 28 5.42 b

t Means followed by the same letter are not significantly different at the p = 0.01 (LSD).








68
of the data from cage 1 (t = -2.273, p < 0.06), cage 2 (t = -3.308, p < 0.05), and

combined (t = -4.15, p < 0.001), as was indicated by the scaling test. When Power's

partition analysis (49,61) was applied to the data from each cages the F2 data did not fit

a two-gene additive model for resistance to soybean looper.

A modification of Poehlman's (60) formula for determination of gene number was

used to estimate the number of genes controlling soybean looper resistance by which the

parents D86-3429 and Braxton differ. Two genes were estimated to contribute to the

expression of resistance to soybean looper in this cross. Since the formula used only

estimates the number of genes in which the two parents, the actual number of genes for

resistance to soybean looper is probably greater than two. It would take further research

with PI 229358 and/or its derived breeding lines to determine exactly how many genes

are controlling resistance to soybean looper. Also, the information generated from this

research could be integrated with the results of research involving other sources of

resistance (36,45,54,72,83) to determine the number of genes that not only control

resistance to soybean looper, but other insect species as well.

Heritability was estimated to be 63% for resistance to soybean looper. Broad

sense heritabilities were calculated because the size and variances of the BC, and BC2

populations were insufficient, thus prohibiting calculation of narrow-sense heritability.

The previous analyses suggests that resistance to soybean looper in this cross is

controlled by two genes which behave additively, therefore it can be assumed that this

heritability is due only to the additive genetic variance. Since the genetic variation








69
influences resistance to soybean looper more than environmental variation, a breeder

should be able to make selections in the F2 population with confidence.

By utilizing a population from a cross between a resistant breeding line and a

highly susceptible breeding line, it should be possible to determine if there are only three

genes for resistance to soybean looper in breeding lines derived from PI 229358. The

single seed descent method of breeding (60) could be used to determine the number of

genes involved by advancing a large population to the F6 generation and evaluating

replicated hills for reaction to soybean looper. For example, assuming three segregating

gene pairs, 91 % of the F6 lines should be homozygous (6) for their respective genotypes,

thus allowing for replications of each genotype. Replicating the F6 lines, as well as their

parents, increases the precision and accuracy of the measured reaction to soybean looper

for each genotype. Examination of results from this type of experiment should indicate

whether or not there are only three genes controlling resistance to soybean looper.

Another way to determine if there are only three genes for resistance to soybean

looper in PI 229358 derived genotypes would be to utilize restriction fragment length

polymorphism (RFLP) markers. RFLP markers have been widely used in other plant

species to construct linkage maps and to place genes that control qualitative and

quantitative traits onto these maps. Recently, RFLP markers have been used to map

phytophthora resistance loci in soybean (13). This methodology could be used to map

the genes and possibly determine any linkage groups that may be associated with the

genes controlling resistance to soybean looper.








70
Since the results of this study have shown that resistance to soybean looper was

a quantitative trait that was additive, it should be possible to increase the level of

resistance in future breeding lines. This might be accomplished by utilizing new sources

of resistance which have additional genes for insect resistance. Another method of

increasing the level of resistance to soybean looper would be by interspecific

hybridization which would allow genes for resistance in other species to be incorporated

into the soybean plant.

To fully exploit the benefits of insect resistant breeding lines, it will become

necessary to understand the mechanisms of resistance that the plants possess. Depending

on the source of resistance, the mechanism of resistance may be different in the various

sources. Several researchers (39,53,63,73,81) have already conducted studies on the

mechanisms of resistance, but further research is needed. As new technology becomes

available, the mysteries of the mechanisms of insect resistance and the genes that control

their expression are closer to being solved.













CHAPTER 5
SUMMARY AND CONCLUSIONS



Soybean is grown worldwide primarily for its oil and meal products, and to a

lesser degree the whole bean product. As in any crop, soybean is exposed to biotic and

abiotic yield-reducing factors such as diseases, nematodes, insects, and drought

throughout the growing season. To produce a profitable crop a grower must modify the

effects of these yield-reducing factors by one or more of the methods described by

Metcalf and Luckmann (50). Environmentally, the best approach to any pest problem

would be to utilize genes for resistance in available germplasm to develop resistant

cultivars (82).

Soybean looper is a major pest to soybean in many areas of the southeastern

United States. The soybean looper, more so than any other arthropod pest, has become

increasingly resistant to available insecticides (10,18,48,58,60) making it difficult to

control. By utilizing soybean germplasm lines that possess genes for resistance to foliar-

feeding insects, it is possible to increase the host plant responses for control as has been

demonstrated by the development of the insect-resistant cultivars Lamar (28) and

Crockett (5).

The objectives of this study were: (i) to evaluate three methods of rating for

defoliation of soybean by soybean looper, and (ii) to determine the inheritance of








72
resistance in soybean to the soybean looper. The latter was the most important objective

since knowledge of inheritance of a particular trait increases the efficiency for

incorporating genes for that trait into an agronomically desirable plant type. Rating

methodology can affect the accuracy of classification and therefore was evaluated

separately to offer guidelines for future insect defoliation studies.

Evaluation of Rating Methods

The three methods of defoliation assessment evaluated in 1990 and 1991 were the

whole plant visual rating (WP), the partitioned plant average visual rating (AV), and

measured average leaf area of the partitioned plant (AL). The WP and AV were visual

assessments of defoliation based on a 1-to-10 scale divided into 10% increments with a

score of 1 signifying defoliation of 0 to 10% and a score of 10 signifying defoliation of

91 to 100%. The WP method was based on one observation per plant, whereas the AV

method was based on the mean of three observations per plants. The absolute method

(AL) utilized a LI-COR LI-3000 leaf area meter to measure the area of three trifoliolates,

one each from the top, middle, and bottom of the plant.

In 1990 no absolute method comparisons were possible, however the WP and AV

were highly correlated (r=0.65, p < 0.001). Adjustments in the method of calculating

percent defoliation by AL for 1991 allowed for direct comparisons with the WP and AV

methods. All correlations (WP vs. AV, WP vs. AL, and AV vs. AL) were significant

at the p < 0.001 level whether the cages were examined separately or pooled. The

results of these correlations indicate that the evaluator could obtain similar estimates of

defoliation regardless of which of the three methods is utilized. These findings are








73
similar to the results of another relative versus absolute estimation of defoliation study

conducted in 1985 by G.R. Bowers (G.R. Bowers, 1992, personal communication).

A paired difference comparison was performed to test the equality of the

defoliation assessment between the relative methods in 1990 and among the three

methods in 1991. The results indicated that estimates of defoliation by the two relative

methods in 1990 and by all three in 1991 were significantly different. With the relative

methods, WP and AV, the evaluator estimated somewhat similar amounts of defoliation

across all plants when compared with each other, even though the difference between the

two methods was highly significant. However, in comparisons with an absolute method,

as in 1991, it becomes apparent that the evaluator's estimates of defoliation with the WP

method were closer to the actual than were the estimates of the AV method. These

findings were validated with McNemar's test for concordance.

The relative variation (RV) was calculated for each method as a measure of

precision. In 1990 the precision of the WP and AV methods were essentially equal.

Similar results were obtained in 1991 by all three methods. Both years the precision of

the methods was essentially equal and well below 10, the value that Pedigo states as

being desirable in research. These results indicate that either relative method, WP or

AV, would be suitable for estimating defoliation by insects.

Based on the results of these studies it is apparent that relative estimates of

defoliation can be used in genetic studies or routine screenings of advanced breeding

lines, in lieu of an absolute estimate. Since the two relative methods yielded similar

results it would advantageous to use the simplest and most economical method for








74
making estimates of defoliation. With increased experience in visual estimates of

defoliation, researchers should be able to estimate defoliation with reliability and

repeatability.

Inheritance Study

This study was undertaken to determine the mode of inheritance of resistance to

the soybean looper in soybean. The study consisted of two years of field cage

evaluations of progeny derived from the cross D86-3429 x Braxton. The results of the

preliminary study in 1990 suggested that the genes controlling resistance to soybean

looper might be inherited quantitatively. The 1991 experiment was designed so that

inheritance of resistance to soybean looper could be examined quantitatively as well as

qualitatively. Also, the 1991 experiment was duplicated in two field cages so that the

experiment would be repeated in two separate environments to strengthen the results and

conclusions of the study.

In the 1991 study, estimates of defoliation were made for each plant using the AV

method. Although this method was more time consuming than the WP method, it was

chosen because its estimate of defoliation was based on three observations per plant

rather than one observation. After assessing each plant in each cage for defoliation by

the soybean looper the data generated were analyzed by several methods of genetic

analysis.

Mather's scaling test (51) was applied to the defoliation data from each cage study

and it was determined that generation means depend only on the additive gene effects and

no epistasis was present. Utilizing Hayman's (31) methodology in terms of the








75
generation-means analysis to analyze the data from each cage separately, only the

additive genetic effects were significant. When the data from the two cages were

combined for analysis the ad epistatic effect was significant (t = 2.30, p < 0.005).

However, the primary effect is assumed to be additive based on the results of the analysis

of the data from cage 1 (t = -2.273, p < 0.06), cage 2 (t = -3.308, p < 0.05), and

combined (t = -4.15, p < 0.001), as was indicated by the scaling test. Applying

Power's partitioning analysis (49,61) to the data from each cage it was determined that

in this cross the F, data did not fit the two-gene additive model for resistance to soybean

looper. There is good reason to believe that the defoliation scores assigned in this

experiment should not be assumed to correspond to genotype effects, therefore the

Power's partitioning analysis has no utility in these analyses.

Based on a modification of the formula presented by Poehlman (60), it was

determined that two genes contribute to the expression of resistance to soybean looper

in this cross. However this formula only estimates the number of genes for resistance

to soybean looper by which the two parents differ. In actuality the number of genes for

resistance to soybean looper is probably greater than two, but it would take further

research to determine this.

Broad sense heritability based on the defoliation data was estimated to be 63% for

resistance to soybean looper. Since the previous analyses suggest that resistance to the

soybean looper is controlled by two genes which behave additively, it can be assumed

that this heritability is due only to the additive genetic variation. A heritability of this








76
magnitude suggests that a breeder should be able to make progress by selecting for

resistance to soybean looper in early generations.

This research could be continued with breeding lines derived from PI 229358 or

by incorporating additional sources of resistance, such as PI 227687, PI 171451, PI

417061, or PI 507301 into agronomically adapted breeding lines. These elite genotypes

could then be cross-pollinated with a highly susceptible cultivar, such as Centennial, to

develop large F2 populations for future evaluations. In order to determine the number

of genes controlling resistance to the soybean looper and/or other species, these F2

populations could be screened by conventional methods or some of the newer methods,

such as RFLP analysis.

Since resistance to soybean looper in this study was determined to be additive, it

should be possible to increase resistance in future cultivars by pyramiding additional

genes from other sources of resistance. This can be done by conventional plant breeding

techniques by utilizing soybean germplasm lines that show resistance to leaf-feeding

insects and incorporating these genes into adapted cultivars or advanced breeding lines.

The same goals could also be accomplished by any of the new methods of biotechnology

that are currently available.

Additional research will need to be conducted on sources of resistance other than

PI 229358-derived lines to determine how many other genes for resistance are available

to breeders for utilization in their breeding programs. In order to fully exploit the

benefits of insect resistance in a breeding program, it will become necessary to

understand the mechanisms of resistance as well as the underlying genes that code for the








77

mechanism. As new technology becomes available, the mysteries of the mechanisms of

insect resistance will be solved. Ideally, the future promises the development of a

resistant cultivar that has many genes for resistance that code for all three mechanisms

of resistance.































APPENDIX A
LIST OF DEFOLIATION SCORES GENERATED BY WHOLE PLANT VISUAL,
PARTITIONED PLANT VISUAL, AND LEAF AREA OF THE PARTITIONED
PLANT RATING METHODS IN 1990.










List of defoliation scores generated by whole plant visual, partitioned plant visual, and leaf
area of the partitioned plant rating methods.
Whole Avg. Avg.
Observation Plant Plant Visual Leaf Area
1 BBE-5-3 1 2.67 57.55
2 BBE-5-4 2 2.00 66.45
3 BBE-5-5 4 1.67 72.74
4 BBE-5-6 3 2.33 65.45
5 BBE-5-7 3 3.00 81.61
6 BBE-5-8 2 2.33 75.80
7 BBE-5-9 2 2.33 111.43
8 BBE-5-10 3 2.67 66.73
9 BBE-6-1 4 3.00 84.04
10 BBE-6-2 2 2.33 47.86
11 BBE-6-3 2 2.33 66.22
12 BBE-6-4 4 2.33 32.03
13 BBE-6-5 3 2.00 87.55
14 BBE-6-6 3 2.67 86.85
15 BBE-6-7 2 2.33 52.67
16 BBE-6-8 4 3.33 76.31
17 BBE-6-9 4 3.33 71.35
18 BBE-6-10 4 2.00 64.61
19 BBE-7-1 3 4.33 100.27
20 BBE-7-2 3 3.00 71.86
21 BBE-7-3 4 3.33 68.00
22 BBE-7-4 3 2.67 75.74
23 BBE-7-5 4 1.33 61.07
24 BBE-7-6 3 2.00 77.64
25 BBE-7-7 4 2.67 67.08
26 BBE-7-8 4 4.33 74.19
27 BBE-7-9 2 1.67 73.05
28 BBE-7-10 4 2.00 63.85
29 BBE-8-1 3 3.33 58.87
30 BBE-8-2 3 1.00 88.32
31 BBE-8-3 2 2.00 72.49
32 BBE-8-4 4 2.00 70.66
33 BBE-8-5 4 2.67 73.00
34 BBE-8-6 4 2.00 79.00





Observation
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70


Plant
BBE-8-7
BBE-8-8
BBE-8-9
BBE-8-10
BBE-9-1
BBE-9-2
BBE-9-3
BBE-9-4
BBE-9-5
BBE-9-6
BBE-9-7
BBE-9-8
BBE-9-9
BBE-9-10
BBE-10-1
BBE-10-2
BBE-10-3
BBE-10-4
BBE-10-5
BBE-10-6
BBE-10-7
BBE-10-8
BBE-10-9
BBE-10-10
BBE-11-1
BBE-11-2
BBE-11-3
BBE-11-4
BBE-11-5
BBE-11-6
BBE-11-7
BBE-11-8
BBE-11-9
BBE-11-10
BBE-12-1
BBE-12-2


Whole
Plant
3
3
2
4
4
3
5
4
4
3
4
3
3
4
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
2
3
3
2
4
3
3


Avg.
Visual
2.00
1.67
2.00
2.67
4.67
3.67
5.00
5.00
5.33
3.00
2.67
1.67
2.67
3.67
1.33
2.67
3.67
3.00
1.67
2.67
2.00
2.67
1.33
3.33
7.00
4.67
4.00
3.67
4.33
3.33
3.67
5.33
1.67
6.67
4.33
2.33


Avg.
Leaf Area
80.88
94.90
60.38
31.06
70.52
66.67
47.64
69.57
57.64
87.20
74.37
75.70
66.17
74.00
45.08
77.06
63.51
67.74
71.65
70.88
76.32
89.68
52.45
29.95
64.62
75.10
55.21
65.95
62.94
88.50
74.01
63.24
97.20
74.45
62.75
78.93





Observation
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106


Plant
BBE-12-3
BBE-12-4
BBE-12-5
BBE-12-6
BBE-12-7
BBE-12-8
BBE-12-9
BBE-12-10
BBE-13-1
BBE-13-2
BBE-13-3
BBE-13-4
BBE-13-5
BBE-13-6
BBE-13-7
BBE-13-8
BBE-13-9
BBE-13-10
BBE-14-1
BBE-14-2
BBE-14-3
BBE-14-4
BBE-14-5
BBE-14-6
BBE-14-7
BBE-14-8
BBE-14-9
BBE-14-10
BBE-4-1
BBE-4-2
BBE-4-3
BBE-4-4
BBE-4-5
BBE-4-6
BBE-4-7
BBE-4-8


Whole
Plant
3
3
3
2
3
3
3
2
4
2
2
4
3
4
3
3
3
4
3
3
2
2
4
3
4
4
2
5
3
3
4
3
3
5
4
2


Avg.
Visual
2.67
2.00
2.33
2.33
4.33
3.00
4.33
4.00
5.00
1.67
1.67
4.00
5.00
4.00
4.00
2.33
3.00
5.00
4.67
5.00
4.33
2.00
5.00
3.67
6.33
4.67
3.00
5.00
2.67
2.67
3.33
2.00
4.67
5.67
4.33
2.33


Avg.
Leaf Area
72.66
92.53
93.98
106.10
72.10
86.99
98.32
60.94
53.08
91.42
73.57
75.45
46.18
63.52
68.61
91.37
51.28
58.01
57.90
65.62
65.58
83.45
57.62
84.46
58.82
50.44
79.47
42.37
69.64
72.35
74.24
80.57
49.60
60.68
80.55
72.02










Whole Avg. Avg.
Observation Plant Plant Visual Leaf Area
107 BBE-4-9 3 2.67 58.72
108 BBE-4-10 3 5.67 65.01
109 BBE-1-1 5 6.67 77.55
110 BBE-1-2 3 2.33 92.04
111 BBE-1-3 3 2.67 72.01
112 BBE-1-4 4 5.00 63.01
113 BBE-1-5 4 5.67 64.55
114 BBE-1-6 4 6.00 51.66
115 BBE-1-7 4 5.33 82.70
116 BBE-1-8 3 2.67 64.59
117 BBE-1-9 3 4.33 60.43
118 BBE-1-10 4 3.67 56.59
119 BBE-3-1 2 3.33 69.57
120 BBE-3-2 2 3.33 64.69
121 BBE-3-3 3 2.67 69.09
122 BBE-3-4 2 1.67 60.31
123 BBE-3-5 2 3.00 82.66
124 BBE-3-6 2 2.33 68.51
125 BBE-3-7 2 2.33 70.63
126 BBE-3-8 2 2.33 65.65
127 BBE-3-9 3 2.67 80.47
128 BBE-3-10 3 2.33 69.33
129 BBE-2-1 2 2.00 80.21
130 BBE-2-2 2 2.33 81.01
131 BBE-2-3 2 2.00 67.84
132 BBE-2-4 2 2.67 52.42
133 BBE-2-5 2 2.00 73.62
134 BBE-2-6 3 4.00 82.98
135 BBE-2-7 3 2.33 77.75
136 BBE-2-8 2 2.00 69.91
137 BBE-2-9 2 2.00 64.00
138 BBE-2-10 3 3.00 79.92
139 BBE-16-1 4 5.33 51.63
140 BBE-16-2 4 6.33 45.94
141 BBE-16-3 5 6.00 40.06
142 BBE-16-4 4 4.00 62.02





Observation
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178


Plant
BBE-16-5
BBE-16-6
BBE-16-7
BBE-16-8
BBE-16-9
BBE-16-10
BBE-15-1
BBE-15-2
BBE-15-3
BBE-15-4
BBE-15-5
BBE-15-6
BBE-15-7
BBE-15-8
BBE-15-9
BBE-15-10
BBE-17-1
BBE-17-2
BBE-17-3
BBE-17-4
BBE-17-5
BBE-17-6
BBE-17-7
BBE-17-8
BBE-17-9
BBE-17-10
BBE-18-1
BBE- 18-2
BBE-18-3
BBE-18-4
BBE- 18-5
BBE-18-6
BBE-18-7
BBE-18-8
BBE-18-9
BBE-18-10


Whole
Plant
5
6
5
7
4
7
5
3
4
3
3
4
4
3
4
4
4
5
3
4
3
4
3
4
4
4
4
4
5
4
4
5
4
4
4
4


Avg.
Visual
5.67
7.00
4.67
5.67
5.33
7.33
5.33
4.00
4.67
3.67
3.33
5.33
5.00
4.33
4.67
5.67
6.00
6.00
4.00
6.00
4.33
5.33
5.33
4.67
4.33
5.33
5.67
5.33
5.00
5.33
5.33
5.33
5.00
3.00
5.33
5.00


Avg.
Leaf Area
45.71
35.54
75.60
37.47
59.21
62.08
74.96
62.52
64.42
67.52
76.06
50.96
70.10
74.26
64.68
46.88
44.94
39.03
80.14
39.94
52.10
42.98
78.97
68.50
52.05
58.14
56.14
61.43
51.84
54.93
58.81
53.30
75.94
62.40
61.60
78.89










Whole Avg. Avg.
Observation Plant Plant Visual Leaf Area
179 BBE-19-1 5 5.67 56.32
180 BBE-19-2 3 3.33 74.48
181 BBE-19-3 5 4.00 49.36
182 BBE-19-4 5 6.00 58.93
183 BBE-19-5 3 4.67 55.07
184 BBE-19-6 3 4.33 88.80
185 BBE-19-7 5 5.67 50.52
186 BBE-19-8 4 5.67 57.10
187 BBE-19-9 4 6.33 48.33
188 BBE-19-10 4 3.00 74.01
189 BBE-20-1 4 3.67 53.10
190 BBE-20-2 3 2.67 71.80
191 BBE-20-3 4 4.33 63.00
192 BBE-20-4 3 4.00 60.55
193 BBE-20-5 2 2.67 66.57
194 BBE-20-6 4 3.33 69.70
195 BBE-20-7 3 3.33 72.31
196 BBE-20-8 4 3.67 56.56
197 BBE-20-9 3 4.33 64.67
198 BBE-20-10 4 2.67 53.80
Mean (n= 198) 3.32 3.68 67.13
Variance 0.98 2.07 201.92
Std.Dev. 0.99 1.44 14.21
t Based on a 1 to 10 scale divided into 10% increments with a 10 reflecting 91-100% defoliation.
$ LI-COR LI-3000 portable leaf area meter measurements in cm2 of leaf tissue.































APPENDIX B
LIST OF DEFOLIATION SCORES GENERATED BY WHOLE PLANT VISUAL,
PARTITIONED PLANT VISUAL, AND LEAF AREA OF THE PARTITIONED
PLANT RATING METHODS IN 1991.










List of defoliation scores generated by whole plant visual, partitioned plant visual,
of the partitioned plant rating methods.


and leaf area


Observation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39


Cage
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1


Plant
9
197
214
15
250
235
63
180
61
42
223
150
186
26
225
187
3
248
17
196
222
275
8
158
166
140
141
138
85
244
122
144
105
92
126
67
51
75
80


Whole plant
3
5
4
8
-
4
6
2
3
4
3
6
8
2
3
7
2
8
9
8
4
5
2
7
7
8
4
2
5
4
3
8
5
6
8
7
6
7
5


Avg.
visual
2.00
4.00
3.00
7.00

4.33
6.67
2.00
3.33
3.67
3.67
5.00
7.67
3.00
2.67
5.67
3.33
6.67
7.67
8.00
3.67
7.00
3.67
5.67
7.67
8.00
5.33
3.67
5.00
4.33
5.33
8.33
5.33
6.67
8.33
7.00
8.67
8.33
6.33


Avg.
LICOR
1.67
2.00
3.00
5.00

2.00
5.00
1.33
2.67
2.00
4.00
4.00
5.00
2.33
2.33
4.67
1.33
6.33
7.00
6.00
3.00
5.00
1.67
4.33
4.00
5.67
2.33
2.33
3.33
3.00
2.00
6.00
3.00
4.67
6.00
5.00
6.00
4.67
3.33





--------


Observation
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80


Cage
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2


Plant
178
179
22
110
270
212
100
213
25
103
79
248
213
17
270
150
100
194
186
61
79
9
250
26
222
214
244
103
178
22
85
180
42
110
212
275
158
51
75
25
8


Whole plant
4
5
8
7
9

6
3
2
5
3
4
2
8
8
5
4
6
6
7
4
4
9
2
7
1
2
5
3
8
7
3
3
5

5
6
4
5
5
3


Avg.
visual
5.33
6.00
7.67
6.33
8.67

5.67
4.33
4.00
5.67
3.67
3.33
3.33
8.33
9.00
5.67
5.00
6.67
5.67
7.67
4.00
3.00
9.33
2.33
6.00
3.33
4.00
6.33
4.67
8.67
5.33
5.00
6.33
4.67

5.00
7.00
4.33
6.33
2.00
3.00


Avg.
LICOR
3.33
4.00
4.67
5.00
7.00

2.67
2.00
3.00
2.67
3.00
1.67
3.33
5.67
4.33
4.00
3.00
3.33
4.00
5.00
2.67
4.33
6.33
1.67
5.00
2.00
2.00
3.00
3.00
5.33
5.33
2.67
5.00
3.33

3.33
3.00
3.00
3.33
1.67
1.67










Avg. Avg.
Observation Cage Plant Whole plant visual LICOR
81 2 140 8 4.67 2.33
82 2 15 3 8.67 5.00
83 2 138 3 2.67 1.33
84 2 67 6 6.67 3.33
85 2 126 7 8.00 6.00
86 2 63 8 7.67 5.33
87 2 179 6 7.33 5.33
88 2 80 7 7.67 2.67
89 2 92 6 6.67 2.00
90 2 3 3 4.33 2.00
91 2 223 4 6.67 3.33
92 2 166 5 7.00 4.00
93 2 122 2 3.67 2.33
94 2 225 3 4.67 2.00
95 2 196 9 9.00 5.33
96 2 144 9 9.00 5.00
97 2 187 3 5.33 2.33
98 2 105 4 6.33 3.00
99 2 141 3 4.33 3.00
100 2 235 5 7.67 3.00
Mean (N=97) 5.10 5.65 3.62
Variance 4.70 3.88 2.19
Std.Dev. 2.17 1.97 1.48
t Based on a 1 to 10 scale divided into 10% increments with a 10 reflecting 91-100% defoliation.
$ Based on LI-COR LI-3000 portable leaf area meter measurements that have been converted to %
defoliation and categorized into the 1-to-10 scale.
Missing data points.













APPENDIX C
GENERATION MEANS ANALYSIS
SAS PROGRAM


OPTIONS LS=80 PS =60;
/*This is a Generation Means Analysis based on Hayman's methodology, Heredity
12:371-390 */
DATA;
INFILE 'A:Cagel.dat';
/* The INPUT statement will vary according to the data set, you need generation
("GEN") and a dependent variable */
INPUT row plant gen $ fc$ rating l rating rating;
MEANRATE = (rating 1+rating2+rating3)/3;
PROC SORT; BY GEN;
PROC MEANS; BY GEN; VAR MEANRATE;
OUTPUT OUT=NE MEAN=Y VAR=V STDERR=S;
DATA NEW; SET NE; RS=1/S; IF GEN='D' OR GEN='BD' THEN DELETE;
PROC PRINT;
/* You need a minimum of 6 generations to conduct this analysis, if the number of
generations used are different than the amount (9) in this example, refer to Gamble's
paper (Canadian J. Plant Sci. 42:339-348) to obtain the proper coefficients. Another
source of information is Jennings, et al. (IA State J. Research, 48:267-280) */
DATA COEFCNTS;

INPUT GEN $ X1 X2 X3 X4 X5;
CARDS;
BC1 0.5 0.0 0.25 0.0 0.0
BC2 -0.5 0.0 0.25 0.0 0.0
BS1 0.5 -0.25 0.25 -0.25 0.0625
BS2 -0.5 -0.25 0.25 0.25 0.0625
F1 0.0 0.5 0.0 0.0 0.25
F2 0.0 0.0 0.0 0.0 0.0
F3 0.0 -0.25 0.0 0.0 0.0625
P1 1.0-0.5 1.0 -1.0 0.25
P2 -1.0-0.5 1.0 1.0 0.25

DATA FINAL; MERGE NEW COEFCNTS;
PROC PRINT;









/* This model tests the significance of the additive (X1) and the dominant (X2) effects
*/
TITLE 'PROC REG WEIGHTED';
PROC REG;
MODEL Y = X1 X2;
WEIGHT RS;
/* The model must be weighted to account for the unequal population sizes among the
generations. See Rowe and Alexander, Crop Sci. 20:109-110, for further detail. The next
model tests for the epistatic effects, additive-additive (X3), additive-dominant (X4), and
dominant-dominant (X5), as well as the additive (X1) and dominant (X2) effects. */
PRO REG;
MODEL Y = X1 X2 X3 X4 X5;
WEIGHT RS;
RUN;

/* This next set of models are tested with PROC GLM instead of PROC REG */
TITLE 'PROC GLM WEIGHTED';
PROC GLM;
MODEL Y = X1 X2;
WEIGHT RS;
PROC GLM;
MODEL Y = X1 X2 X3 X4 X5;
WEIGHT RS;
RUN;

/* The next series of models are the same as the previous models, except that they
assume equal population size among the generations, and are therefore not weighted. */
TITLE 'PROC REG NOWEIGHT';
PROC REG;
MODEL Y = X1 X2;
PROC REG;
MODEL Y = X1 X2 X3 X4 X5;
RUN;
TITLE 'PROC GLM NOWEIGHT';
MODEL Y = X1 X2;
PROC GLM;
MODEL Y = Xl X2 X3 X4 X5;
RUN;
/* Depending on whether or not the population sizes among the generations are equal,
report the model that has the most significant effects from either PROC GLM or PROC
REG. */