Title: Comparison of S2 progeny and inbred tester methods for improvement of maize
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00099256/00001
 Material Information
Title: Comparison of S2 progeny and inbred tester methods for improvement of maize
Physical Description: viii, 64 leaves : graphs ; 28 cm.
Language: English
Creator: Ameha, Mesfin, 1942-
Copyright Date: 1977
 Subjects
Subject: Corn -- Breeding   ( lcsh )
Agronomy thesis Ph. D
Dissertations, Academic -- Agronomy -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
 Notes
Statement of Responsibility: by Mesfin Ameha.
Thesis: Thesis--University of Florida.
Bibliography: Bibliography: leaves 60-63.
General Note: Typescript.
General Note: Vita.
 Record Information
Bibliographic ID: UF00099256
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000010025
oclc - 02896569
notis - AAB2137

Downloads

This item has the following downloads:

comparisonofs2pr00ameh ( PDF )


Full Text











COMPARISON OF S PROGENY AND INBRED TESTER METHODS
FOR IMPROVEMENT OF MAIZE













By
MESFIN AMEHA


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









UNIVERSITY OF FLORIDA

















To

the memory of my mother, Weizero Zenebu Tirfe,

and father, Grazmach Ameha Meshesha.














ACKNOWLEDGMENTS


The author expresses his deepest thanks to Dr. Earl S. Horner, the

chairman of his supervisory committee, for his constructive guidance in

the research and preparation of the manuscript. The author also

expresses appreciation to the members of his supervisory committee,

Dr. Charles E. Dean, Dr. Wayne B. Sherman, Dr. Victor E. Green, and

Dr. Vincent N. Schroder for serving on the supervisory committee and for

their helpful suggestions for improving the content and structure of the

dissertation.

Thanks are extended to Dr. Frank G. Martin for his help in the

analyses of the data, Mr. Vernon R. Munden for his assistance in the field

work, and some colleagues who helped him in recording data. The author

is indebted to his fiancee Miss Birhan Tekabe for her encouragement and

assistance in typing.

The author thanks Dr. Dagnatchew Yirgou, who was the General

Manager of the Institute of Agricultural Research, Ethiopia, for allowing

him to continue his studies. Thanks are also due to the Food and

Agriculture Organzation of the United Nations for the financial

assistance during the study program.
















TABLE OF CONTENTS


Page

ACKNOWLEDGMENTS iii

LIST OF TABLES v

LIST OF FIGURES vi

ABSTRACT vii

INTRODUCTION I

LITERATURE REVIEW 3

Recurrent Selection 3
Gene Action 4
Half-sib Method and Effectiveness of Testers 5
Theory 8
Comparisons of Sl, S2, and Half-sib Methods 15

MATERIALS AND METHODS 19

Sources of Materials 19
Experimental Designs 23
Data Collection 24
Statistical Procedure 25

RESULTS 28

Testcrosses 28
Synthetics 31
Interpopulation Crosses 36
Summary of Evaluation Methods 36
Characteristics Other than Yield 41

DISCUSSION 44

CONCLUSIONS 49

APPENDIX 50

LITERATURE CITED 60

BIOGRAPHICAL SKETCH 64
















LIST OF TABLES


Table Page

1 Lines Used in Testcross Evaluations 21

2 Lines Used in Evaluation of Synthetics 22

3 Mean Squares and F Ratios for Testcrosses 29

4 Mean Performance for Testcrosses of Lines
Selected for High and Low Yield by Two Methods 30

5 Mean Squares and F Ratios for Synthetics 32

6 Grain Yield Performance of Synthetics of Lines
Selected for High and Low Yield 34

7 Mean Grain Yield of Selected Groups in 1975
and 1976 and Ratios of Differences Between
High and Low Means in the Two Years 35

8 Mean Squares and F Ratios for Interpopulation
Crosses 37

9 Grain Yield Performance of Interpopulation
Crosses of Lines Selected for High and Low
Yield 38

10 Differences in Grain Yield Between High and
Low Groups when Evaluated by Four Methods 40

11 Differences in Response for Characteristics
other than Yield Between High and Low Groups
Selected by Two Methods 42
















LIST OF FIGURES



Figure

I Population B Synthetics by Location
Interactions

2 Average Yield of Interpopulation Crosses
of High and Low Yielding Lines by the Two
Methods















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


COMPARISON OF S PROGENY AND INBRED TESTER METHODS
FOR IMPROVEMENT OF MAIZE

By

Mesfin Ameha

March, 1977

Chairman: Dr. Earl S. Horner
Major Department: Agronomy

The Inbred Tester and S2 Progeny Methods have been used effectively

for improving maize (Zea mays L.) populations. However, it is very important

for a breeder to know which one of these methods is more effective.

Theoretically, the S2 Progeny Method is more effective than the Inbred

Tester Method since the genotypic variances that are due to additive effects

are expected to be much larger among S2 progenies than among half-sib

progenies, but up to now, the two methods have not been compared together

in the field. In 1975, however, S2 lines of two populations, A and B, were

evaluated for yield by the Inbred Tester Method and by the S2 Progeny Method

and seeds were made available for study.

High and low yield groups of 10 lines each were selected by each method

from the A and B populations and seeds were grown to make the evaluation

by (1) out crossing to the original inbred testers and to two unrelated

hybrid testers, (2) yield of synthetics, and (3) interpopulation crosses.

Ears were harvested separately and seeds were bulked to make 40 composites

(24 for testcrosses, 8 for synthetics, and 8 for interpopulation crosses),


viI








The evaluation tests were planted at two locations near Gainesville

with 90 x 30 cm spacings. A split-plot design with 10 replications was

used to test the testcrosses, a randomized complete-block design with 15

replications was used to test the synthetics, and another randomized

complete-block design with 10 replications was used to test the inter-

population crosses at each of the two locations. Data on yield and other

characteristics were collected and analyzed.

The testcross results indicate that both methods were equally

effective in selecting lines for high combining ability, but the Inbred

Tester Method was more effective in selecting lines for low combining

ability with the unrelated doublecrosses. The synthetics and the inter-

population crosses showed no significant difference between methods.

These results show that lines selected by the Inbred Tester Method can

combine well with unrelated populations and crosses between lines of

Populations A and B should result in higher grain yields.

The inbred testers detected small differences in combining ability

among selected HS2 and LS2 groups better than the doublecross testers.

This indicates that the inbred testers would be effective in improving

frequency of favorable genes having additive effects.

Correlated responses of other characteristics to selection for high

and low grain yield were somewhat different for the two methods.















INTRODUCTION


During the period 1896 to 1917 mass selection and ear-to row methods,

both of which involve recurring cycles of selection, were frequently used

in corn (Zea mays L.) improvement. In the second decade of the 20th

century, however, these methods were found to be ineffective for improving

yield. These failures were probably due to visual selection for uniformity

of ear types, which resulted in inbreeding, and to poor experimental plot

techniques. When work began on hybrid development, population improvement

was neglected for many years. But,after plot techniques were improved and

adequate additive genetic variation for yield was found in most populations,

population improvement was re-emphasized. Gardner (1961), Webel and

Lonnquist (1967), and others reported significant improvement using the

mass and ear-to-row methods of selection.

During the 1910 to 1950 period, conventional inbreeding to produce

homozygous lines together with systematic determination of the best

combinations for hybrids (evaluation for general combining ability, GCA,

and specific combining ability, SCA) became important without emphasis on

rapidly recurring cycles of selection. However, Hull (1945) emphasized

the advantage of recurring cycles, which have multiplicative effects, for

high yield and suggested selection for SCA directly using an i:',i ii-

as a tester. He assumed that additive genetic variation in adapted

varieties used was less important than non-additive variation. Since then,

others followed with suggestions for alternative methods, among which

reciprocal recurrent selection (RRS), full-sib, half-sib, and selfed








progeny selection were important. Yet, there is no general agreement as

to which is the best for yield improvement.

In Gainesville, Florida, Robinson (1976) selected high and low

yielding S2 lines in two different populations based on (a) half-sib

progeny performance using an SI line of each population as a tester of

lines from the other and (b) S2 progeny performance per se. The results

showed that among the top yielding 15 lines, six lines in Population A and

one line in Population B and among the low yielding ten lines, two lines

in Population A and one line in Population B were selected by both methods.

This indicates that the two methods are quite different in identifying

lines for grain yield in corn.

The objective of this study was to evaluate the half-sib progeny with

an inbred tester and the S2 progeny per se methods for identifying lines

with high and low combining ability for grain yield. The evaluation was

made by forming high and low performing groups by each method in each

population using Robinson's (1976) data and (a) crossing each group to its

respective SI tester and to two unrelated commercial hybrids; (b) developing

synthetic sub-populations of each group; and (c) crossing different groups

from the two populations.














LITERATURE REVIEW


Recurrent Selection



Population improvement in corn is now being carried on by many

breeders, who use a system generally called recurrent selection (RS). The

improvement results from accumulation of favorable genes during recurring

cycles. Simple RS is performed by selecting superior plants and making

many intercrosses among selected plants or their progenies and is most

useful for characters having high heritability. Recurrent selection for

GCA (Jenkins, 1940) and SCA (Hull, 1945) employ testcross performances

with broad and narrow base testers, respectively. Reciprocal recurrent

selection (Comstock et al., 1949) uses two heterozygous populations, each

being the tester for lines from the other, simultaneously.

A breeding method similar to RS for GCA was first suggested by Hayes

and Garber (1919). They outlined a procedure for developing a synthetic

population with high protein content by selling, selecting, and subsequent

crossing. However, this method was not adopted at that time. Later, after

Jenkins (1940) presented the first detailed description of the methodology

for GCA, many breeders have stressed the importance of a broad base tester

since it has been considered superior to narrow base testers for identi-

fication of lines which provide greater average performance upon combination

with other lines.

Hull (1945) first proposed RS for SCA, and suggested the use of a

homozygous lines as a tester. He believed that overdominance was the major








type of gene action present in adapted varieties since additive gene action

was presumably exhausted by previous selection. Comstock et al. (1949)

proposed and outlined a method for RRS which was considered to be

effective regardless of level of dominance present, but it would have

maximum efficiency if overdominance or dominance by dominance interaction

was present in the populations used. Current theory further suggested

that RRS is appropriate whether additive or nonadditive gene action is

predominantly present.

The considerable time and progeny testing needed to extract superior

homozygous lines encouraged the need for early testing, which essentially

helped RS to develop. Jenkins (1935) studied topcross yield results,

using lines which represented eight generations of selling and discarded

lines of each generation, and concluded that inbred lines that showed

superiority in the early generations of selling remained relatively stable.

Lonnquist (1950) presented conclusive evidence that continued selection

and testing after S1 would be most profitable for those lines exhibiting

highest topcross performance in early generations.



Gene Action



Partitioning of genetic components of variance based on type of gene

action for various mating designs has provided a scale for measuring the

inheritance of agronomically useful characters. Robinson et al (1949),

Gardner et al. (1953), and others found yield genes in the overdominance

range. However, later studies by these workers showed that these estimates

of overdominance were due to linkage disequilibrium. Kravenchko et al.

(1971) reported that overdominance as well as epistasis was apparent for

grain yield; partial to complete dominance was observed for days to silk,








moisture percentage in the grains at harvest, plant height, and ear height.

Heterosis occurred for length of growing season, resistance to smut, and

number of grains per row. Thompson et al. (1971) showed that additive

and dominance parameters were responsilbe for an average of 90% of the

variability in the inheritance of ear height. Tavares (1972) obtained the

highest heterosis from those hybrids connected with low yielding parents,

and heterosis for yield was not always accompanied by heterosis for

characters positively correlated with yield.



Half-sib Method and Effectiveness of Testers



The half-sib progeny test, where selected progenies are mated among

themselves, uses one-fourth of the additive genetic variance and requires

two generations per cycle of selection. Where the selected selfed male

or female parents are mated among themselves, it utilizes one-half the

additive variance and requires at least three generations per cycle.

Improvement by this method not only depends on the genetic constitution

of the population but also on the type of tester used. Rawlings and

Thompson (1962) crossed six lines classified as high, intermediate, and

low yielding to high and low yielding testers and found that a low yielding

tester would discriminate more effectively among combining abilities of

lines crossed with it than would a high yielding tester. In their

theoretical comparison, they showed that when gene frequency of a tester

is homozygous recessive at all loci, the genetic variance among testcross

progenies is larger and it has an increasing advantage as dominance

increases compared to heterozygous testers. Allison and Curnow (1966),

from a mathematical approach of the two allele locus, stated that if








partial to complete dominance is of primary importance, a low yielding

population selected from the parental variety would be the best tester.

Lonnquist (1968) showed that the highest yield resulted from the inter-

crosses of lines selected on the basis of performance with the parental

population. Selection on the basis of an unrelated tester resulted in

slightly better yield when selected high by low groups were crossed than

when high by high groups were crossed. This result suggests that the

parental tester emphasizes additive gene action while the unrelated tester

emphasizes dominance gene action.

Selection for GCA, which is primarily a function of additive gene

action, has historically used broad base testers while selection for SCA,

which is primarily a function of nonadditive gene action, used a narrow

base tester. However, recent results show that narrow base testers also

improve GCA, perhaps even more effectively than broad base testers.

Russell et al. (1973) evaluated populations developed from five cycles of

recurrent selection for combining ability for yield with the inbred tester

B14 in two populations, variety 'ALPH' and the F2 of (WF9 x B7), by using

1B4 and BSBB as testers. They obtained an average yield increase of about

3.1 q/ha per cycle in 'ALPH' and about 1.3 q/ha per cycle in (WF9 x B7)

using Bl4 as a tester and about 3.6 q/ha per cycle in 'ALPH' and about

1.5 q/ha per cycle in (WF9 x B7) using BSBB broad base tester. Horner

et al. (1973) completed five cycles of RS using an inbred line and the

parental population as testers and S2 progeny performance per se in

parallel programs. The 15 populations produced by these methods were

evaluated for average combining ability with the parental populations and

an unrelated broad base tester. They obtained a 4.4% yield gain per cycle

from the Inbred Tester Method compared with 2.4 and 2.0%/ for the parental

tester and S2 Progeny Methods, respectively. They indicated that the








inbred tester was evidently homuzygous recessive at many important loci

which would result in more successful selection for dominant favorable

alleles than would use of a broad base, heterozygous tester. Walejko and

Russell (1976) evaluated the progress of RS in which the inbred HY was the

tester for two open pollinated varieties and indicated that the yield

gain observed for the HY testcrosses and the population crosses were

expressed equally well in testcrosses with unrelated testers.

The significance of the findings [Russell et al. (1973), Horner et al.

(1973), and Walejko and Russell (1976)] is that (a) inbred testers are

very effective for improving the GCA of a population and (b) the fear

that narrow base testers improve only SCA in selected populations was

unfounded.

The earliest report on the effectiveness of RRS was made by Douglas

et al. (1961). They compared performances of first, second, and third

cycle composites of each variety and crosses among them and concluded

that selection was slightly effective in accumulating favorable alleles

in the first two cycles of 'Furguson's Yellow Dent' and only in the first

cycle of 'Yellow Surcropper'. Average combining ability of the varieties

was improved only in the first cycle of selection. Eberhart et al.

(1973) evaluated the progress from five cycle of RRS in varieties BSSS

and BSCBI and obtained a linear improvement of 2.7 q/ha (4.6/) per cycle

from BSSS(R)n x BSCBI(R)Cn population crosses. However, there was no

significant change in the parents themselves.









Theory



Selection based on Selfed Progeny Performance



Selfing is the major type of mating that changes the genotypic

frequency of a formerly random-breeding population. Selfing rapidly

reduces a population to homozygosity regardless of the number of hetero-

zygous loci initially present. Let us suppose that randomly selfed SO

plants were heterozygous (Aa) for all loci. In the first generation, S

progenies will be 50% homozygous (AA and aa) and 50% Aa. S2 progenies of

randomly selfed SI plants will be 75% AA and aa and 25% Aa. If AA and aa

in the SO generation are assumed to be at the same proportion as the Aa,

then randomly selfed SO plants will produce Sl progenies with 75% AA and

aa and 25% Aa and likewise SI plants will produce S2 progenies with 87.5%

AA and aa and 12.5% Aa. The gene frequency of S2 lines will be the same

as in the SO plants, but the genotypic frequencies will vary depending

on the gene frequency of the initial population.

When the S2 progeny means are compared,

AA will be better than Aa for d < 1.0

AA will be better than aa for all values of d

AA will be equal to Aa for d = 1.0

Aa will be greater than AA for d > 1.0

where d is a measure of dominance, d > 0.0.

It means that if dominance is less than complete, selection will favor

AA genotypes; if dominance is complete, selection will favor both AA and

Aa; and if there is overdominance (d > 1.0), selection will favor Aa alone.

But with selfing the frequency of Aa genotypes is low.









Examination of inbreeding and its effect on genetic variation among

the selfed progenies is very helpful for understanding the changes that

would occur in the selling series. Maximum inbreeding is obtained when

like gametes, A with A or a with a unite and under such conditions the

inbreeding coefficient (F) will be one because the gene controlling a

particular character is fixed. A random mating population on the other

hand has an F value of zero. If a base population with gene frequency

p = q = 0.5 is selfed, F values will be as follows:

selfing generation F values

SO 0.00

Sl 0.50

S2 0.75

So, 1.00

The equation for F with selling as given by Falconer (1960) is F = '(1 + F)

where F is the inbreeding coeficient of the parent.
P
It is observed that the increase of F value is inversely related to

the decrease of the frequency of heterozygotes in a selfed series. This

relationship can be further understood by looking at the genetic variation

of the selfed population. If the AA, Aa, and aa genotypes are coded to

have values of 2, 1, and 0, respectively (d = 0.0, additive gene action),

the variance of different generations can be shown statistically by the

formula,

f (x)2 ( .x )2/n
n

where f. is the genotypic frequency, i = I, 2, 3

x. is the genotypic value, j = 1, 2, 3

and n is the sum of the genotypic frequencies, n = 1.0









Therefore, if we start from a random mating population with p = q,


.25 (2)2 + .5(1)2 + .25(0)2 (1.0)2/1.0
V = .0 = 0.5
0 1.0


.375(2)2 + .25(1)2 + .375(0)2 (I.0)2/1.0


VI =



V2=



Voo=


= 0.75


.4375(2)2 + .125(1)2 + .4375(0)2 (1.0)2/1.0


1.0

p2(2)2 + 2pq()2 + q2(0)2 (1.0) 2/1.0
/1


=0.875


= 1.00


The relationship of

in the selfed series is,


V's, gene frequency, and inbreeding coeficients


V in terms of
gene frequencies

2pq

3pq

3.5pq

4pq


V in
terms of F

(1 + 0.00)VO, F = 0.00

(I + 0.50)v0, F = 0.50

(1 + 0.75)Vq, F = 0.75

(I + 1.00)V0, F = 1.00


Genetic variances for dominance and overdominance gene actions can

also be calculated similarly but with different codings. If complete

dominance is assumed, AA, Aa, and aa are coded 2, 2, 0 and if overdominance

is assumed, Aa has a larger value than AA.

The formula for additive variance in a random mating population is,

VA = 2pq[a + d(q p)]2

where 2a is the additive effect and d is the dominance effect at a locus.

The variance due to dominance is,

V = (2pqd)2 (Falconer 1960).

Horner et al. (1969) showed the fractions of the genetic variance


V in selfed
generations

V0

V

V2

Vco


V
values

0.500

0.750

0.875

1.000









among selfed progeny means that are due to additive and dominance gene

effects to be,




(I F) 2
pq(p + Fq) (q + Fp) + F)d respectively

where F is the inbreeding coeficient of the parental plant and the rest

are the same as before. They used d instead of d for the Aa genotype

since they were interested in the progeny means rather than individual

plants. They stated that the variance among selfed progeny means that is

due to additive gene effects is large compared with that due to dominance

effects. As the generation of selfing increases, the frequency of

heterozygotes decreases and the fraction of the genotypic variance that is

due to dominance decreases regardless of the level of d.



Selection Based on Testcross Performance



Selection based on testcross performance largely depends on the amount

of variance among testcross means. Horner et al. (1969) used the following

formula, similar to that of Rawlings and Thompson (1962), to estimate the

genetic variance in a two allele system.

V = 0.5pq(l + F) [a + d(Q p)]2

where P and Q are the frequencies of A and a, respectively, in the tester

and the other symbols are the same as before. They stated that if S0 plants

or SI lines are tested, F = 0.0. If S plants, which are equivalent to

S2 lines, are tested, F= 0.5. In the absence of epistasis, the total

variance is the sum over all loci. The expression shows that the variance

of testcross means is a function of the additive effect of a gene or the









average effect of a gene substitution [a + d(Q P)], gene frequency in

both the tested and the tester populations, and inbreeding of the tested

plants. If a further assumption is made that A is also the favorable

allele over a in the tester, the theoretical effectiveness of a tester can

be visualized.

1. When d = 0, selection for high performance of the individuals

of the population being tested will not be influenced by gene

frequency of a tester, because d(Q P) = 0 in the expression.

2. When d < 1.0, selection for high performance of the individuals

of the population being tested favors AA genotypes. If the

population being tested has a low frequency of AA genotypes, the

rate of improvement will be large when Q is large.

3. When d = 1.0, selection for high performance is ineffective if

P = 1.0, because, in such circumstances, selection by testcross

performance will not discriminate between individuals of the

population tested, since AA = Aa. However, as the frequency of

P gets smaller, more individuals with aa genotypes will be

discarded, because a gametes of the tester are likely to unite

with a of the population tested.

4. When d > 1.0 and P = 1.0, selection will favor aa genotypes.

Aa genotypes will also be selected but less frequently, because

a gametes from Aa genotypes can unite with the tester gamete.

AA genotypes, however, will be discarded. As P decreases,

selection for AA individuals increases and at P = 0.5 the

frequency of selecting each genotype will be the same since either

A or a gametes of the tester will only show overdominance when

united with a and A gametes of the population tested, respectively.









Horner et al. (1969) presented a very descriptive figure showing a

comparison of the expected genotypic variances that are due to additive

effects for selfed progenies and for testcross progenies based on the

expressions shown for selfing and testcross progeny methods. In this model,

the parental population was used as a tester. Their figure showed that

the expected additive variance among the selfed progenies is much larger

than among testcross progenies. Up to p = 0.2, the expected variance is

similar for SI and S2 progenies, but very low for testcross progenies.

At p > 0.2, S2 progenies show the greatest variation and testcross progenies

the least variation. At p = 0.7, variance among S2 progeny means would

be 14 times larger than among testcross progenies, and variance among SI

progeny means would be about four times larger than testcross progenies

at p = 0.5. This shows that S2 progenies provide more opportunity for

selection than both SI and testcross progenies while testcross progenies

provide the least opportunity.



Reciprocal Recurrent Selection



RRS has remained an effective method for improving two populations

for combining ability with each other simultaneously at all levels of

dominance. It was designed for characters that exhibit very low herita-

bilities and high heterosis. Comstock et al. (1949), who proposed the

method, compared theoretical limits of improvement using RS for SCA and

GCA, and RRS and concluded that:

1. When d < 1.0, the improvement limit is the same for RS for GCA

and RRS, but lower for SCA. The superiority of Aa genotype

declines as the value of d is reduced, and so RS for SCA will be

less effective.









2. When d = 1.0, RS for GCA, SCA, and RRS will be equally effective.

If the broad base tester is homozygous dominant for many favorable

alleles, selection for GCA will be less efficient, because all

testcross genotypes would have the same value.

3. When d > 1.0, RS for SCA and RRS will be the same, but RS for GCA

will be inferior to both methods, because the mean of the

individuals selected based on performance with the broad base

tester is the average of all genotypes which would result in low

response relative to the other two methods.

Cress (1966a) compared RRS with within population selection (WPS) based

on theoretical considerations. WPS is the average response of the two

populations, X and Y, for selection within populations (similar to RS for

GCA). The rate of progress was measured by,

C= M ( M + M )
xy yx xx xx yy yy

where C is the difference which is the comparison of the rate of improvement

between RRS and WPS,

M is the mating system for RRS, and
xy yx

M and M are mating systems within population X and Y,
xx xx yy yy

respectively.

If A is the favorable gene over its allele, a, in both populations,

and p and P are the gene frequencies for favorable allele in the X and Y

populations, respectively, the theoretical rate of improvement is:

1. When (p + P) < 1.0, RRS is better than WPS for all positive

dominance.

2. When (p + P) > 1.0, WPS is faster in the rate of the improvement

than RRS for levels of dominance including complete dominance

because the larger proportion of the favorable alleles contributes









larger additive variance with the WPS method.

3. If overdominance is present, RRS is better when (p + P) > 1.0

and the difference between p and P must be large enough to see

pronounced result. If p = P, RRS = WPS, and the rate of progress

in RRS is dependent on the additive variance in the testcrosses.

Cress (1967), using the simulation method, also compared RRS as

proposed by Comstock et al. (1949), RRSs, modified by additional selling

prior to testcrossing, and RRSc, using the original parents as testers

over 30 cycles. Generally, the results showed that RRS and RRS methods

provided high and linear mean response in hybrid combinations between

populations at all level of dominance with RRS being superior over RRS

in the rate of progress in early cycles.



Comparisons of 51, S2, and Half-Sib Methods



Selfed and half-sib progeny selection methods have been compared more

frequently since the early 1960's, but the question as to which method is

most effective still remains unanswered. The first attempt was made by

Davis (1934) in which he compared selfed lines and their crosses with an

unrelated open-pollinated variety tester. He indicated that average

yield of the first and second inbred generations was important for

selection. Center and Alexander (1962) made a comparative performance

test between SI progenies and progenies selected on the basis of testcross

performance with single cross testers. They found more dispersed means

and less environmental effects from SI progenies than from testcrosses.

Koble and Rinke (1963), Lonnquist and Lindsey (1964), Torregroza and

Harpstead (1965), and Lonnquist and Castro (1967) all made similar reports









that the SI progeny method was more effective for selection than testcross

methods with related and unrelated testers. Duclos (1967) and Duclos and

Crane (1968), however, found no significant difference between S

progeny and topcross methods.

Comparisons of advanced populations developed by different methods

of recurrent selection have been made in order to evaluate the effective-

ness of the methods. Center (1966) found mean yield increases of 31% and

18% with two cycles of S, and testcross selection, respectively. In

1973, he obtained a population yield of 9.0 q/ha (11%) more from the

second cycle of SI selection than from the testcross populations.

However, Center and Eberhart (1974) obtained the same yield response from

both VCBS(S)C4 and VCBS(HT)C3 developed from Virginia Corn Belt-Southern

Synthetic by four cycles of RS for SI progeny yield and by three cycles

of half-sib selection, respectively. BSK(S)C4, developed from Krug Hi

syn-I by four cycles of SI selection, and BSSS(HT)C7, developed from

'Stiff Stalk Synthetic' by seven cycles of half-sib selection, failed to

show any improvement. Carangal (1967) stated that RS on the basis of S

and topcross evaluation was effective in both cases in increasing the

frequency of favorable genes for yield in synthetic A. However, SI

evaluation showed greater improvement in the first cycle synthetic made

from the selected lines. Later, Carangal et al. (1971) indicated that

SI progeny evaluation was more efficient than testcross evaluation even

for GCA. Burton et al. (1971), using a double cross tester, obtained about

6% yield increase from the testcross series, but 16% from the selling

series for yield of populations per se. Combining ability with four

testers was also improved more by the S, method.









Horner (1963) compared SI lines and S1 plants (which are equivalent

to S2 lines) and stated that selection for combining ability can be done

more effectively among individual SI plants than among SO plants or SI

lines. The variance component estimate for crosses involving SI plants

was larger than for crosses of S lines. This shows that subsequent

selfings do provide an opportunity for more effective selection. Horner

et al. (1969) found that the S2 progeny method was as effective as the

parental tester method for population improvement and suggested that

it places more emphasis on contribution of homozygous loci than hetero-

zygous loci, whereas the parental tester method emphasizes the contribution

of heterozygous loci to a greater extent, resulting in a higher yielding

syn-3 population. In 1973, from two additional cycles, they reported a

gain in GCA of 4.47% per cycle from an inbred tester method compared with

2.4% and 2% for the parental and S2 progeny methods, respectively.

RRS, regardless of the type of gene action involved, theoretically

is considered effective in producing lines with SCA for yield and vigor

between two populations. Thomas and Grissom (1961) evaluated RRS for popping

volume, grain yield, and resistance to root lodging in popcorn and

concluded that RRS was effective in improving the mean of the two populations

simultaneously. However, nearly the same improvements were made when

S4 lines were developed from direct selling in the two populations.

Moll and Stuber (1971) compared full-sib family selection and RRS for

higher grain yield in the varieties 'Jarvis' and 'Indian Chief' following

six cycles of selection, and stated that heterosis in the variety hybrid

increased about 30% more after RRS than after full-sib family selection

within each population.






18

Darrah et al. (1972) obtained an 8 q/ha larger gain in yield from

RRS compared with modified ear-to-row method in variety H611, but there

was no significant difference between methods in varieties Kil and Ec573.

Russel and Eberhart (1975) evaluated three crosses among BSCB(R)C5,

BSSS(R)C5, and BSSS(HT)C6 populations improved by five cycles of RRS

and six cycles of testcross selection with a double-cross than from the

population crosses. There was no significant difference among the population

crosses. Robinson (1976) obtained a higher estimate of genetic variance

from S2 line evaluation than from half-sib progeny evaluation in the

initial cycle of a selection program comparing the RRS and S2 progeny

methods.
















MATERIALS AND METHODS


Sources of Materials



The S2 lines used in this study were derived from Populations A

and B of FSHMR, a synthetic variety developed recently by the Florida

Agricultural Experiment Stations for resistance to the fungus Helminthos-

porium maydis N. and M. These lines had been evaluated by Robinson (1976)

for combining ability with an S line tester and performance of S2 lines

per se. The testers used were a line (Tester A) from Population A to

evaluate lines from Population B and a line (Tester B) from Population B

to evaluate lines from Population A.



Crosses Made to Compare Methods



On the basis of data obtained by Robinson (1976), eight groups of

10 lines each were chosen as follows:

1. AHT High yielding Population A lines (Testcross data)

2. ALT Low yielding Population A lines (Testcross data)

3. AHS2 High yielding Population A lines (S2 line data)

4. ALS2 Low yielding Population A lines (S2 line data)

5. BHT High yielding Population B lines (Testcross data)

6. BLT Low yielding Population B lines (Testcross data)









7. BHS2 High yielding Population B lines (S2 line data)

8. BLS2 Low yielding Population B lines (S2 line data)

The individual lines making up the above groups and their performance

data either in testcrosses or as S2 progenies per se are shown in Table 1.

In addition, a similar series of eight six-line groups were established

with the limitation that the lines within groups would not be related to

each other according to pedigree records. These lines and their

performance records are listed in Table 2. Some lines not listed in

Table I had to be used in the second series to avoid inclusion of related

lines. This resulted in slightly lower mean performances of the groups

in Table 2 than in Table 1.

All of the lines listed in Table 1 and 2, together with Tester A,

Tester B, and the doublecrosses Coker 71 and Greenwood 471, were grown

near Homestead, Florida during the Winter of 1975-1976 in sufficient

quantities to make the following crosses using standard controlled-

pollination techniques: For lines in Table 1, (a) each line was crossed

with Coker 71 and Greenwood 471, using one plant of each hybrid as the

female parent, (b) each line from Population A was crossed with Tester B

and each line from Population B was crossed with Tester A, using two

plants of each line as the female parents, and (c) interpopulation crosses

between A and B populations within breeding methods were made, AHT x BHT,

ALT x BLT, AHS2 x BHS2, ALS2 x BLS2, AHT x BLT, BHT x ALT, AHS2 x BLS2,

and BHS2 x ALS 2. For each cross one group was designated the female

parent. Two or three ears of each female line was pollinated by pollen

from each line in the male group. For lines in Table 2, all possible

intra-group crosses were made, including reciprocals, to produce first

generation synthetics of the eight groups.

















I N.-


CI CZ )




a
o
O C

6-






OC (
') C











0
-1 41





IZ 0

















a a
00



C 0
a
a)








SCO

O

0




DC 0

uC





O




Sc


0
-
0PC-


4o Cm -o C 0 7 -




N Q UNNo CN m r N W -
I I I DI I I rI I

N1- 0 mN-- 0 N

NNNC NN NN O- -








- N - - n m "
IN Ic N n n C n N 'n


N LN -T N ND UN NO -I

r- 0CDN 00 ON 0 0\ o 0
- N - -



NNNO \ N- NNN)
NNN-- --- - N N



















rN 0 NO- N 0 \o oN 0
T 4- N-z L(,% -4- Ln NrD NZ D N T 4- n





























14 I i I I I "AI
N r--t4T -N-T--
N-4N-- NNNNN \o










ro N---o -No N








4- N N N N N N 0C
I l l f i lld ^ '^ c i f l C L T
N a dr r\u0Lr O ~ rm -Nam o(
a m m m m m m m so m -t--. t^-v>


o0 n o --T u- rN N- N- O-e 0o
eO- N0 N' r LN ~- eN N- 4-~ 0
- CO -0 CN --




N- -M -1rCM m-. 0- --C


T- d -" ^ 1.-4 N

--NNNNNNN- N








No-N- - NO








-o 0 co 0-0 N ,C 0N




















I I w I I I I M
---CD----t--1---
000o400 c 0 NM -

--oNN- NN cN0C-










0 - N N -








Sm- 4-- m mD r--- 0-m-
----m m n -
N 0- 4- 4- NM 0-

- N_ U-N N- NO N N N N




























C
0



a
0
0

















-- I

S0- i
(7 5' -


cC

C
O

o
In

3

a
C-


1 0 o 3 C 0





















-I 00 oC h M
NN N -N N
I I I I I I








ms c, om




mon m -
CNN CO- 0




N--N-7































\ NO 0 N- 0 N,
--4 Lm -4








r- \- m c D C D
N---N--




NNNNNN










-o NN\D










Lnm e mse\ o \o (
Nt N4 4- Nd d d










N N NNNN
444444f-r


NN --- N
i I I I i f




JN -Nco
N L LrN NN N
C NN NN









oo co 0f) r-- \n LT\ o







I I
N0 N 0 N N



N---






0 o- o--- C



























oo oI-III I
N .0 N 0 Nr N -
-Cl---- -
























I II III M
So 00 L-
NNL 0 N r- 0 N


















N- -- N N -


0N4444
co o mno
- m N s

























a
0
a


C 0



O C :)


l-
U
ct-


-J 0

O 0
J C

-J









When the seeds were mature the ears were harvested and taken to

Gainesville for drying and shelling. After sorting according to group

and type of cross, composites were made by mixing approximately equal

numbers of seed from each individual cross within the various group x

tester, population x population, and intra-group combinations. A total

of 40 such composites were prepared (24 group x tester, eight synthetics,

and eight population x population).



Experimental Designs



The 40 composites described above were evaluated in two-row plots

at two locations, the Agronomy Farm adjacent to the University of Florida

Campus in Gainesville and the Agronomy Green Acres Farm 12 miles west of

Gainesville. Row spacing was 90 cm and the plants were spaced 30 cm

apart in the rows. Two seeds were planted per hill and the plants were

thinned to one per hill when they reached 15 cm height. Thirty plants

per plot were used on the Campus Farm and 26 plants on the Green Acres

Farm. Missing hills were replanted with a purple-stalk hybrid to provide

uniform competition to all plants.

A split-plot experimental design was used to evaluate the testcrosses,

with the three testers as main plots and the four selected groups of a

population (A or B) as subplots. Separate experiments were conducted for

the A and B populations. The main plots were assigned to each block at

random and the subplots were randomized within main plots. Ten

replications were used at each location.

The four synthetics from each population were also evaluated in

separate experiments. A randomized complete block design was used with









15 replications at each location.

The eight interpopulation crosses were tested in a randomized

complete block experiment with 10 replications at each location.

The plots at both locations received uniform applications of

fertilizer and pesticides, and were irrigated when necessary to prevent

excessive drought stress. Additional nitrogen was applied at the Green

Acres location about 6 weeks after planting when pronounced deficiency

symptoms developed.



Data Collection



Prior to data collection, deformed and purple-stalked plants were

discarded from the field. Plants badly damaged by smut were also removed

from the Green Acres field. Data on total number of plants, average ear

height, husk score, and lodging were taken from every plot. Ear height

was taken on a plant that represented the average ear height in the plot;

the measurement was made from ground level to the node of attachment of

the ear. A husk score for each plot was determined on one to nine rating

scale, one being best with relatively tight husk and well-covered ears

and nine lacking such characteristics. Root lodging was also taken on a

one to nine rating scale where one was assigned to a plot with less than

10% of the plants lodged and nine with greater than 90' of the plants

lodged.

At harvest, data on number of plants erect, ear quality score, total

number of ears, and number of ears rotten were recorded. Yield per plot

was taken by weighing the ears to the nearest 0.1 pound (45g). Finally,

yield was adjusted to full stand by multiplying the yield at harvest by









the ratio of the number of plants at full stand to the number of plants

harvested corrected by an adjustment factor. The factors used for each

missing plant were 0.75 for Green Acres and 0.50 for Gainesville. The

factors were chosen to minimize yield advantage that a plot with a poor

stand might have compared with those plots with full stand, since plants in

a plot with a poor stand have less competition for light and soil nutrients.



Statistical Procedure



Since our primary interest at this stage was on the yield performance

of the groups, analysis of variance (AOV) was made for the adjusted yield

only. For the other characteristics measured, means were calculated.

AOV for the testcrosses was based on the model,

X jkmn= p + Ai + Bj(i) + Tk + AT.i + E (ijk) + S + AS. +
ijktmn i j(i) k ik L(ijk) m im
TS + ATS. + E
Tkm ASikm+ En(ijkm)

where p is the overall mean

A. is the effect of the ith location, i = I, 2

B (i) is the effect of the jth block in the ith location, j = I,...,20

Tk is the effect of the kth tester, k = 1, 2, 3

ATik is the interaction effect of the ith location and the kth tester

E .(ijk) is the error term of the Lth unit in ith location, jth

block, and kth tester

S is the treatment effect of the mth group, m = I, 2, 3, 4
m
AS. is the interaction effect of the ith location and the mth group
Im
TS is the interaction effect of the kth tester and the mth group
km
ATS.km is the interaction effect of the ith location, kth tester,
and mth group
and mth group









En(ijkm) is the error term of the nth sub-unit in the ith location,

jth block, kth tester, and mth treatment.

The combined AOV for the adjusted yield of testcrosses was done as follows:


L- I

(r1 1) + (r2 1)

a -

(, 1) (a 1)

[(r, 1) + (r2 1

b 1I

(t 1) (b 1)

(a 1) (b 1)

(A 1) (a 1) (b

a[(rI 1) + (r2 -


Sources

Locations

Blocks/Ls

Testers

LxT

Error (A)

Groups

LxG

TxG

LxTx G

Error (B)


AOV for the

the model,


synthetics or the interpopulation hybrids were based on


Xijkt = p + Ai + Bj + Sk + ASik + E t(ijk)

where p is the overall mean

A. is the effect of the ith location, i = 1, 2

B. is the effect of the jth block in the ith location

j = 1, ..., 30 for the synthetics

j = 1, ..., 20 for the hybrids

Sk is the effect of the kth treatment

k = 1, 2, 3, 4 for the synthetics

k = 1, ..., 8 for the hybrids

ASik is the interaction effect of the ith location and kth treatment
ik


)] x (a 1)







- I)

1)] x (b 1)


SS

SSI

SS2

SS3

SS4

SS5

SS6

ss7

Ssg
SS
SS
SS 9

SS10


MS

MS

MS2

MS3

MS4

MS5

MS6

MS7

MS8

MS9

MS10


--









E (ijk) is the error term of the tth unit in the ith location,

jth block, and kth treatment. All factors except the blocks


are fixed.

The combined AOV


for the adjusted yield was as follows:


Sources Df SS MS

Locations (L) S- 1 SS MS

Blocks/Ls (r- ) + (r ) SS2 MS2

Treatments (T) t 1 SS MS3

L x T ( ) (t 1) SS4 MS4

Error [(r1 I) + (r2 I)] x (t I) SS5 MS5


The two breeding methods were compared using Duncan's new

range test and single degree of freedom contrast following the

shown in Steel and Torrie, 1960.


multiple-

procedure
















RESULTS


Testcrosses



Mean squares and F ratios from the analysis of testcross yield data

involving four selected groups of inbred lines from each population in

crosses with three testers are shown in Table 3. In both populations

there were highly significant differences among groups averaged over

testers, and the tester x group interaction was significant at the 0.05

level in Population A and at the 0.01 level in Population B. The

tester x location interaction was highly significant in Population B

but not in A.

Means for yield of individual group x tester combinations and for

groups over testers are shown in Table 4. The crosses involving HT

groups were not significantly different from those involving HS2 groups

for all testers except Coker 71 in Population A or for means over testers.

In contrast, the LT crosses were significantly lower in combining ability

on the average than the LS2 crosses and there was a significantly larger

difference in average combining ability between HT and LT than between

HS2 and LS2 in both populations.

The interaction of testers with groups in Population A was due

primarily to the fact that HS2 and LS2 combined more poorly with Coker 71

and better with Greenwood 471 than would be expected on the basis of

the other tester x group yields. In Population B, the significant

















Table 3
Mean Squares and F Ratios for Testcrosses.


Sources of
variation


Population A


Df Mean
square


Main plots

Locations (L)

Blocks/Locations

Testers (T)

Tx L

Error (a)


169.780

3.864

2.384

1.004

1 .000


F ratio


43.9389""

3.8640 ."'

2.3840

1.0040


Sub-plots

Groups (G)

Lx G

TxG

LxTx G

Error (b)

Total

' Significant


5.385

0.565

0.864

0.775

0.379


14.2084""

1.4907

2.2797

2.0448


at the 0.01 level.


* Significant at the 0.05 level.


Population B


Mean
square


308.205

1.575

0.815

3.418

0.473


F ratio


195.6857

3.3298""

1.7230

7.2262""


3.713

0.382

0.755

0.193

0.238


15.6008'"

1.6050

3.1723

0.8109















Table 4
Mean Performance for Testcrosses of Lines Selected
for High and Low Yield by Two Methods.


Tester
Selected group I rnkr CrenwnnA


tester


71


471


kg/ha


Population A


5067a"

4188c

4982a

4637b


5306a

4715b

4886b

4564b


5136a

4703b

5301a

5124a


Population B


5312a

4548b

5038a

4540b


5167a

4734b

5176a

5220a


5041 a

4618b

5070a

5020a


Average


5170a

4535c

5056a

4775b


5173a

4633c

5095ab

4927b


SMeans in columns within populations followed by different letters are
significantly different at the 0.05 level according to Duncan's New
Multiple-range Test.








tester x group interaction was caused mostly by the failure of the two

double-cross testers to differentiate among the HT, HS2, and LS2 groups,

while the inbred tester combined significantly more poorly with LS2 than

with HS2.



Synthetics



Mean squares and F ratios from the analyses of yield data involving

synthetics from the four selected groups of inbred lines from each

population are presented in Table 5. In both populations, differences

among groups were highly significant. The highly significant synthetics

by locations interaction in Population B was due to the fact that the

HS2 and LS2 synthetics did not yield as well at Green Acres as they

did at Gainesville relative to the HT and LT synthetics (Figure 1).

Means for yield of the four synthetics from each population are

shown in Table 6. The yield of the HT synthetic was not significantly

different from the HS2 synthetic and likewise the LT synthetic was not

significantly different from the LS2 synthetic within populations.

However, the high and low synthetics within populations were signi-

ficantly different at the 0.01 level for each method of selection.

To obtain measures of repeatability for combining ability with the

two inbred testers used by Robinson (1976) and to estimate heritability

with both methods of selection, differences between means of high and

low selections in 1975 were compared with the 1976 evaluations of the

various selected groups. The upper half of Table 7 shows comparisons

of the two methods provided by crosses of selected groups with the two

inbred testers, and the lower half shows comparisons by use of synthetics




















Table 5
Mean Squares and F Ratios for Synthetics.


Population A Population B
Sources of
variation Df Mean F ratio Mean F ratio
square square

Locations (L) 1 113.294 271.6882"' 101.875 156.4900..

Blocks/Locations 28 0.417 1.2991 0.651 2.3759

Synthetics (S) 3 4.515 14.0654. 3.987 14.5511

S x L 3 0.101 0.3146 1.227 4.4781".

Error 84 0.321 0.274

Total 119

** Significant at the 0.01 level.














Kg/ha


6000
















5000





LS2 \ LT


4500 \


\
\




4000 \






Gainesville Green Acres
Locat ions

Figure I
Population B Synthetics by Location Interactions.

























Table 6
Grain Yield Performance of Synthetics of Lines Selected
for High and Low Yield.
--=== -= -===== --==---============= --== === == = = == = =
Method of Population Population
evaluation A B


kg/ha

Synthetics

HT 4963a* 5477a

LT 4458b 4891b



HS2 5003a 5360a

LS2 4240b 4717b

SMeans within columns followed by different letters
are significantly different at the 0.01 level.



















Ln C
\O N

0 0


c











LD













0 a
0



















CC
'-






















0o
, OH




O*-
00









am
-o 0
c







0m 0
3

r~-






--




,-a
c2c


-O 04








-0 20

0







'--a











0
'4C


-J -J



0 I












M D

n- -


















0 n

C U) *-
0-3. ---

























00
co
'i- >







0









0



Slm
4- 0 cm

0 0


























cJ
-i -
a0 u
o-0












c c







m 0
-i-




I I




0









4,
- 0 cm

0-j







00



-D4-
>-
















00
T3 -


r .- o 10 r,
r- oO0 ,0 .-
C 0 0L















S-- O mN
4 cm cm r-
0\,-- N--


o -

0 0


r-0 00
0O N CO cm4
0 -CO c\0 m-
c-m4 4


- CO 01 '- 0~\ CN
-- cN -- A-mN
044.0 --04o
SLnc- m- N


0
a
u
c
a)
0
a)

14-


0 -1


u
c
0)
a
0


U-
N N-

-0


L


a0
0


!--

-J


0.0 _0










c m 4 0 N








OD Lr" LD C- I
mc-- 0n o
SN\ 0 o N 2 r-

f cm- cm--


0


,I




o--
4i-,


0
0
c
0
L
a



N N-

-0
(^1 Csl
I_ 13


L


01n
0l

--





0



(-3








C
o-


u



/1
0)


-C






01:










cm
>-
4-J




_I


00 : 0 C
N CO- COc

















--. Ln r o c0 _
- -T r N--
M7 :: M7 *- CO
LU '\- N -


Ln





c
0







:4









tot
0












u
c
'U



0
a0




0
0



c
u



uO -




a 4,

0 0.
L-01
0 L













0





0
--
0
;_;














ca
UC (








D >.-








0 ,r
4-) -M





m 2
0-














ac
04,
03
0 U

r- 4-4
cm

--



0
0> ~

0)0

00



-- 0

^.0 -,
cr r










-~ E
0O









00Q
C C









000
Q CC



3









>-
C

u o
aj o

(J


TO C ;
+- T
T U (U
Q 2:

r;; ^








developed from each selected group. Repeatability of combining ability

with the same tester in the two years was 59% in Population A and 65%

in Population B (ratios of 0.59 and 0.65, respectively, of high-low in

1976 to high-low in 1975). When selection was based on S2 progeny

performance and the 1976 evaluation was for combining ability with the

inbred testers, realized heritabilityy" was only 17% in Population A

and 27% in Population B (ratios of 0.17 and 0.27). When evaluation

was done by producing synthetics, the Inbred Tester Method resulted in

ratios of 0.36 and 0.54 and the S2 Progeny Method in ratios of 0.42 and

0.46 for Populations A and B, respectively. These results indicate that

the two methods were equally effective with respect to yield of synthetics.



Interpopulation Crosses



Mean squares and F ratios from the analysis of yield data of eight

interpopulation crosses are shown in Table 8. Yield differences among

crosses were highly significant (Table 9). The High x High crosses were

significantly higher yielding than the Low x Low crosses with both

methods, and the average of the two High x Low crosses for each method

was not significantly different from the average of the High x High and

Low x Low crosses as shown in Figure 2.



Summary of Evaluation Methods



Differences in average performance between high and low groups of

lines selected by the two methods for four methods of evaluation are

summarized in Table 10. Larger yield differences were obtained from
















Table 8
Mean Squares and F Ratios for Interpopulation Crosses.


Sources of variation Of Mean square F ratio


Locations (L) 1 75.282 41.6844"

Blocks/Locations 18 1.806 4.4373"

Crosses (C) 7 1.757 4.3169'.

L x C 7 0.177 0.4349

Error 126 0.407

Total 159

Significant at the 0.01 level.
Significant at the 0.05 level.
















Table 9
Grain Yield Performance of Interpopulation Crosses
of Lines Selected for High and Low Yield.


Cross Grain yield


AHT x BHT

AHT x BLT

Mean

ALT x BHT

ALT x BLT

Mean

AHS2 x BHS2

AHS2 x BLS2

Mean


kg/ha

5383a

5281ab

5332

4830cd

4689d

4759.5


5529a

51llabc

5320


ALS2 x BHS 4948bcd

ALS2 x BLS2 4937bcd

Mean 4942.5

* Means followed by different letters are
significantly different at the 0.05 level.
































S Progeny Method --J- -
J


--L--Inbred Tester Method


HxL


HxH


Figure 2
Average Yield of Interpopulation Crosses of High and
Low Yielding Lines by the Two Methods.


Kg/ha

6000 i


5500 -


5000 -


4500 -


Lx L


I L I



























CDm en O g
SNO o c c
-f N- U' e


en e
-t


-s cm


0 C

-L' -

0)0



a 0




a o




CL
O
0
0-


4 c cc
'CD N CD
N- 0 00










D r- -f Ln





CO LA LC


n
0



4-- U


o a
u c
<0


- 4




N N




N- -
CO CD
f^. U-


o,


LO Cmi eni r- eni
-t' -3" CD r- en
i~^ (^ ~r-~ ^













- 0 CO CD CO
C 1- 4 4 4
CO -4 e LCO


C)
QJ

c

O
C








C
0~






O
C
in






- cC



CD
0
a

tD

a
a
Q-


CD
-C
N-


o
IC
-a
-c






3m
CO



En

-
-
L-..


0 >N D
ID CD








m(3
CD


CD
C







LL

4-

cO<


mt









group x inbred testers evaluation of lines selected from Populations A

and B by the Inbred Tester Method compared with lines selected by S2

Progeny Method. On the other hand, larger differences were obtained

between synthetics developed from lines selected by the 52 Progeny

Method compared with lines selected by the Inbred Tester Method. Yield

differences between groups for combining ability with two double-crosses

were larger for lines selected by the Inbred Tester Method than for

lines selected by the S2 Progeny Method. Results of interpopulation

crosses indicated that the two methods were similar. For the average

over all evaluation methods, the Inbred Tester Method showed relatively

greater yield differences between high and low selections than the S

Progeny Method.



Characteristics Other than Yield



Differences in response between high and low groups for other

characteristics and their averages over the four methods of evaluation

are shown in Table 11. The groups were selected primarily for high and

low yield. Numbers preceded by negative signs show that selections

for low yield were associated with higher mean responses for the

characteristic measured than the corresponding high yield group.

Positive ear weight and negative ear height differences were shown for

all evaluations of the lines selected by the Inbred Tester Method in

both populations. In contrast, positive differences for ears per plant

were shown for all evaluations of the lines selected by the S2 Progeny

Method in both populations. On the average over the two populations,

ear weight difference was much greater for groups selected by the









table 11
Differences in Response for Characteristics other than Yield
Between High and Low Groups Selected by Two Methods


Characteristics
Ears Ear Root
Ear Ears Ear Husk Ear Ears Root Plants
tper .quaIl ry lodging
weight per height score quali rotten erect
plant score score


Inbred Tester Method


Population A
Inbred Tester B
Avg. double
cross testers
Synthetics
Interpopulation
crosses
Average
Population B
Inbred Tester A
Avg. double
cross testers
Synthetics
Interpopulation
crosses
Average


27.00 0.01 -3.0

3.00 0.10 -3.0
18.00 -0.05 -2.0

7.00 0.09 -3.5
13.75 0.04 -2.9


0.55

-0.30
1.30

0.70
0.56


-0.05

0.18
0.16


0.004 -0.15

-0.010 -0.20
-0.010 0.04


0.50 -0.004 -0.22
0.20 -0.005 -0.13


22.00 0.03 -7.0 0.60 -0.05 -0.010

13.00 0.03 -2.0 -0.57 -0.07 0.010
22.00 0.01 -4.0 -1.63 -0.66 -0.010

16.00 -0.05 -0.5 -0.95 -0.35 0.004
18.25 0.00 -3.4 -0,64 -0.28 -0.001


0.00

-0.02
-0.04

0.00
-0.01


-0.15

-0.10
0.02

-0.22
-0. 11


S2 Progeny Method
Population A
Inbred Tester B 9.00
Avg. double
cross testers -5.00
Synthetics 13.00
Interpopulation
crosses -7.00
Average 2.50


0.03 3.0


0.95


0.07 -3.0 0.30
0.06 -1.0 -0.07

0.14 -3.5 0.60
0.07 -1.1 0.44


0.10 -0.001 -0.05


0. 10
0.30

0.00
0.12


0.003
-0.008

0.009
0.001


-0.07
-0.07

0.10
0.02


Population B
Inbred Tester A 9.00 0.08 8.0 0.45
Avg. double
cross testers 2.00 0.00 1.0 0.62
Synthetics -13.00 0.16 -1.0 0.03
Interpopulation
crosses 7.00 0.00 0.5 -0.20
Average 1.25 0.06 2.1 0.22


0.05

-0. 10
0.20

-0.05
0.02


0.004 -0.05 0.02

0.003 0.03 0.00
-0.004 -0.14 0.03

-0.004 0.15 0.03
0.000 -0.08 0.02

low groups exceed


the high groups in measurement of the characteristics.


Methods of
evaluation


0.01

0.03
-0.01

0 01
0.01


* Numbers preceded by negative signs show that the






43

Inbred Tester Method than for groups selected by the S2 Progeny Method.

Except for ear weight, ears per plant, and plants erect, average

differences were negative for groups selected by the Inbred Tester Method

whereas differences except for root lodging score were positive for

groups selected by S2 Progeny Method.















DISCUSSION


Although the Inbred Tester Method was more effective than the S2

Progeny Method for separating high and low combining ability groups, the

two methods appeared to be equally effective for selecting high combin-

ing lines for grain yield. This is evident from the fact that the

HT and HS2 groups in both populations were not significantly different

in (1) combining ability with the two unrelated doublecross testers,

(2) yield of synthetics developed from them, or (3) yield of HT x HT

and HS2 x HS2 crosses between populations. These results indicate that

the breeder should select lines that produce good grain yields as inbreds,

which is very desirable from the hybrid seed production point of view.

Selection for low yield with the S2 Progeny Method, on the other hand,

did not result in significantly lower combining ability with unrelated

doublecross testers than did selection for high yield of S2 lines.

This suggests that low yield of some inbred lines may have been due to

homozygosity for a few deleterious recessive genes rather than to a low

frequency of dominant favorable genes over the entire genotype. In

crosses with the doublecross testers, such deleterious genes probably

were masked by dominant genes in the testers, resulting in higher than

expected yields. When the Population A LS2 and Population B LS2 groups

were crossed, yield of the hybrid was significantly lower than yield of

the HS2 interpopulation cross. Also, synthetics produced by inter-

crossing lines within the two LS2 groups were significantly lower than








those from HS2 groups and were comparable to the LT synthetics.

Apparently there was less masking of homozygous recessive deleterious

genes in low x low combinations than in crosses between the LS2 groups

and the doublecross testers.

The inbred testers were more effective than the doublecross testers

in measuring differences in combining ability between selected high

and low groups. As mentioned above, the doublecross testers showed no

significant difference between the HS2 and LS2 groups in either

population, but the inbred tester for each population indicated signi-

ficant differences between these groups for combining ability. Data on

the synthetics and intergroup crosses show conclusively that the HS2

and LS2 groups in each population differed in frequency of favorable

genes. We can conclude, therefore, that heterogeneous, heterozygous

testers such as the two commercial doublecrosses used in this study are

ineffective for detecting small differences in combining ability. They

were effective in showing significant differences between the HT and LT

groups, however, where combining ability differences were larger than

those between the HS2 and LS2 groups. These results are in agreement

with those of Horner et al. (1973), who found that an inbred tester

was about twice as effective as a heterozygous population for improving

frequency of genes having additive effects. The crosses between the

HT groups and the inbred testers had grain yields similar to the means

of the same crosses when tested individually in 1975 by Robinson (1976).

Differences between the high and low groups were 59 and 65% as large

in 1976 as in 1975 for Population A and B, respectively. This level of

repeatability of yield measurement in different years is good.

A perfect correlation between the 1975 and 1976 results would not be









expected because experimental error and genotype x environment interaction

cause variation in grain yield from year to year.

Estimates of realized heritability, based on yield of synthetics

developed from the high and low groups, were rather low compared with

the repeatability estimates. For the Inbred Tester Method, comparisons

of differences between the high and low group means in 1975 and of the

synthetics tested in 1976 indicate that heritability was 36 and 54%,

respectively, for the A and B Populations. Comparable estimates for

the S2 Progeny Method were 42 and 36%. These results show, however,

that both methods were effective in differentiating among S2 lines for

additive genetic differences that affect the yield of synthetics. There

appears to be no significant difference between the two methods in this

respect.

The yields of the high x high, high x low, and low x low inter-

population crosses suggest that gene action for yield was predominantly

additive. With each method of selection, mean yield of the high x low

crosses was not significantly different from the mean of the high x high

and low x low crosses. If dominance is of major importance and two

alleles per locus are assumed, the high x low crosses would be expected

to yield more than the mean of the high x high and low x low combina-

tions. It is known, however, that dominance is very important to the

yield of maize. In the populations studied here, Robinson (1976) found

that the mean yield of S2 lines was only about 46% as much as mean

yield of half-sib families. This degree of inbreeding depression can

only be explained by a high level of dominance of favorable genes.

It seems likely that multiple alleles were involved in these inter-

population crosses between groups of 10 lines. Lines selected in the









high groups may have had a higher Frequency of dominant favorable genes

that could be designated the A' type, while the low groups may have had

a high frequency of dominant, less favorable alleles of the A" type.

Combinations such as A' A" on the average may exhibit little or no

dominance in crosses between different populations. Cress (1966b) has

shown algebraically that with multiple alleles some loci might make

negative contributions to heterosis in crosses between populations,

when the dominance relationships of all pairs of alleles are positive.

Since the two methods were positively correlated for selecting

lines for high and low yield, some lines should be selected by both

methods. In Population A, there were five lines common to both the

HT and HS2 groups and two lines common to both the LT and LS2 groups.

In Population B, there were two lines common to the high groups and one

line common to the low groups. These results indicate that even though

the two methods generally gave the same results, different lines were

selected, because the above situation, except for the Population A

high groups, could simply arise from chance alone.

Correlated responses of other traits to selection for high and low

grain yield were apparently somewhat different for the two methods.

Average ear weight differences between high and low selections were

16 grams for the Inbred Tester Method and only 2 grams for the S2

Progeny Method. On the other hand, there was a larger difference

between high and low groups for number of ears per plant with the S2

Progeny Method (6.5 ears per 100 plants) than with the Inbred Tester

Method (2 ears per 100 plants). Selection for high yield resulted in

an average of 3 cm lower ear height than selection for low yield with

the Inbred Tester Method, but there was no difference for this trait









with the S2 Progeny Method. Differences between methods for response of

husk, ear quality, and root lodging scores and of percentage rotten ears

and erect plants were very small.

Theoretically, the S2 Progeny Method should be more effective than

methods involving a tester for changing frequencies of favorable genes,

because dominant genes in the tester tend to mask expression of genes

in the plant being tested. In S2 progenies at least 75% of the loci

are homozygous, so there is a much lower probability of recessive alleles

being masked by dominant alleles than with half-sib progenies. The

results of this study show, however, that low yielding S2 lines are not

necessarily low in combining ability with unrelated testers. This

suggests that contributions of homozygous loci are in some way different

from contributions of heterozygous loci, and that for combining ability

improvement a tester should be used. This was a conclusion reached by

Horner et al. (1973), who found that the Inbred Tester Method was more

effective than the 52 Progeny Method for improving general combining

ability.

The relatively high yields (Table 9) of the interpopulation crosses,

AHT x BHT and AHS2 x BHS2, indicate that the two populations combine

well with each other. Additional cycles of selection by both methods

should result in improved combining ability between the two populations

and should provide some additional information on the relative efficiency

of the methods.
















CONCLUSIONS


The S2 Progeny and the Inbred Tester Methods were equally effective

in selecting lines for high combining ability, but the latter method

was superior in selecting lines for low combining ability with unrelated

doublecross testers. Evaluations based on synthetics and inter-

population crosses indicated no significant difference between methods.

These results show that lines selected by the Inbred Tester Method

can be expected to combine well with unrelated populations. There-

fore, this method should result in selecting lines for higher combining

ability between the A and B populations.

The two doublecross testers were not as effective as the inbred

testers for detecting small differences in combining ability among

selected groups. The testcross data with the HS2 and the LS2 groups

showed no significant difference when the doublecrosses were the testers.

This shows that the inbred testers would be more effective than the

heterozygous testers in improving frequency of genes having additive

effects, because inbred testers can better discriminate between high

and low combiners in the populations tested.









































APPENDIX



















0 0 0






o LA LA
0 -

- 0


C U







01
fo










c

LU
E0 43
L


0 -t
- 0
0 0

0 0


C







-C

0)

CO









o









u










0 1
-



C









0 m
S-c




c .






C-
M- C





L0.
0L 4-





0)-



-c
o U-

















L

c


S0
>


U
co m
00 r
- c0
4-3


0)


C)
I-

h-

-0 -C
) 01n 3
I .- o

C


- r- L
0 _0


C)-



0 -C


(N
Cm


0 0 0





o o 0

o c









ooco





- o C



o o a
c -
0 0 0












00 c
(N C)
Ln o -



r~ o e
CN) -



cNj Oi ^



c~O r^


-t
C)
C


S0
Q1








m3 U



.C
I- c0



4-*




kLm
--j 1- _
) a-















L CL

IC)-D


0 LA\ LA
LA cc LA




Lno 0















- o








n cc LA (N


r-~ cc
(N- LA0
00O
0-


Cc

















CU







n,



0 0 u
2a

cto


1-




-U
U-











-i 0-
mn







i 0
CU -











c
-m
- -c


SU-






qL
ui


UC
N -- 43



U CV




U-C









en-
o N-




(U-
3




Lii
4-r






E t


CDN

- 0






0c en cc\
cCD


c e -
o c

Ln 4


-C
m 3

r -I


.D cD 0
cc C o;








o o 0


















a N
-e 000





NN m















I
04 4 I


N 3
>-
c
Di

0 -C
L- en 3
at - 0
C: -J
04


\D 0D


CD 0 0
o a o






CD
0 144C Ln >



SLn


o

a

0 04 0 I





o0 0 0 U%






CD


ct














C
t-
CD




Mi

m0 0 I U



Seo




CI Cf





U-


'- o -




Di

4 -


C -D c I 1u
ci


0







J J
ti.0

SU,






4 -C
D --

C-













4-D




w3


Cj 0




c Na


Co (o)
cc cc


















mo o






ooo
o o C


-D-


o
B.
0


Co -
U



- c
Q)


ma)
a-)
L






O



- 0
OC





O
a-




L.
a)














: 0
a0 )














>
3 -















E
u -


.4- n
3L-

-c

r)
- 0


So
c )

00 00






cu
7, 0

L 4-i








IJ
LL 0U

0u








ol


I-
C>


















m o








LO
3 Ia






c.D


0m -









L ,-


a)


4-1
(U

-c


0) an 0

0= 32 -J
c


cm cm 0


N NO 0







- 0







E cm -









N 0







CO N -t
cr 0 N I


'0 0 a-
m -t
mo


Un -t -


o o o





0 0 0
- C













N--O
N 0N
o 0 0








CO O

N N 0







N OD -(D








o N N








m '0 N


- C


0-0






( 0 C I




- -:- rN
o C I
rr\ -_ -
iLr\ L


au
U
C

L-




S0





























L



-C


(U




0)









o 3



CC
, U0











-1
4-
Cc
0 m







L 43
CC









C- 0




Q)



3
I-

C
UC'







a)


c C





C)

0 e U
0 -0 u







CIn a)



LU 0





3L1-

rO--
43C







40
Ln C








-C
a C




LC.





LU a




C In
CC









C XI



C3C


L-

tC
+J



C C
L C
Si


C


r r-


o o 0







o i-n Lfl










0 CD 0
o o o









0I 0 0

Ln o
N co









In In )







E Ln I
U N. In I









- 0



cn r^ L\ s


0o0 Y
D LrI


0 -n

Lrn


C


C -C

al - o
r -l
N
C)


n0 r N













o
o o o







0 0 C0
oo o







Lr -t-
- 0







o o' 0














N NC

















00
r--r~- o


CD f
N] In
.- 4 :
In 4



















L: Q)
0- La





ca


So 0






C0 D



wog
L 0
t..O Ul


CI 4


O
0

-0







o



0 0
a)







[- )
C-




m O



O

O U

n C



a -
CL
0 -CO












.- a-
0 0
>o 0


o C)







0
u-1-
-- C














IU 0


a-





In





a -
U



CU-I




0-C


0)
U
C

00
0 C


-c 0- O _C
En 3 Q i n
0- 0- 0-
S -i -J

Li


LA\ LfA 0









cm cm o
o o 0
















-A 0 CA
o o 0

o o o











0 LA LA
(m cm o

N (N a





LA LA 0










0 CO CO O0







LA N- c
-Z N 0

e- 0















S '- -f
-, D .
-n 4 L
.2 LA4


,0
(U


_3U



L- 0)

Li (U
0 -)

C -





n LC



0 CL2-





D -

3




C -o


L- -
C3 >


L-
a)


0

vi
0)
~0
-C
Ct


N- c -


0 0 0









N 0






- 0


o o 0















o LA LA
- cO

L N 0
I














N cm -
00 CO I







.LA -
LA LA 0

- 0
CDm


OT cT
LA LA


10


N I
LP>

















a o C1 _

S. .
0 0 0







0 U- IA

C-


-3-
o

0


0



o
03
I1) -


C r


0
0









cog






S-- C-
03

0- 0



u




o 0C

Q-





0 0
Q- 4)
un U




c -C





-3
U0










c
rI


In -








uC


0 S


U 0





04
I5 J


o ul 0
0 C 0
C) 0






L +-
0





oU
U 0


L-





m
I) I






m
S* LL


















I 0
Wc
O
r_ r_0
S-E








.L
S-






00
.1C




















01












-o 4-




00 I
guC


Im










0 0







UND
0 D0 0
- o






-. m 0o
o o
0 0 0



























OD
0 0 0
N NO

N No0

























m m



















00 0
N N 0





















LIA U>t


o C


44 0 0
01 > 0
S 0 0C

n0 C *- O IC

I 0 -- O
OI J I- J
C N
l/l


0 -t

SOm o









N N 0















0D CO CN
4 O ) 0
N N





0 0 0



n 3o
U) 2 !


- 0 CC
41 N
0 tO 4
CLn


























cn


0 0
on





a o









LC


SJ Qo
L m












'O
flu
CT


r n




-c

Li

4-

r- L
a0-

-c "
; or
an v
4a

U1 1-
I- L) (DQ.-


U)
U


L >
a) c
'4- 0n
-c *- O .c
Cn 3 0 L CD 3
*- 0 a- O


-)


o o o





SCN -





0 0 0




















































"0 ro _o
r --I or

r N -







E -- a cN










- 0







or co










2 \Dn


0
L.




U






























u 0
-c




r-
































L)
3
a
>a
0



gU


Sr
N- 04

w a




o

a)
-a '













C-)
.- t/
r~ 41
in l'
L).
O- i
LQQ
TO4J X


t- U


N o -


















o or o
- 0







SCO CO

S0

0 00















'0 or or



N 0

















o N- N0






or or m
mo or













-? 0













o '-



















C,
Ca)






oi


0 -u L
-0 U






c

ro 44








44)
w 0







t_ .- L

n-
Sou






S 0


.C





r-






U C,--
04-C














rn
C u
-nc






CLa)
-c


-i-











S-C
L- c


03


















0 E
-i-j
C



0







CQ






C
0







aQ)

I 0





G0 C



OCL
- t
In
















C:
-D
a)




>,
(U
(/









i_


a,


Ga
Ga

I-

-u r

4I o
OI 'C -J
C


OCD
0 0 0


- 0


o 0


0 0 0

























NN









-E 0













N
47 co cs


N
C
Ga
en
C -C
44 Cn S
l. - 0
'C -2
N
(IC


00 o












- 0







o o 0
000

o o 0


















mm









N- -












N 0

















o N N


'U-- 0 n
CO -C

-4 cc LA
LA 4
















mL
N r o --t N m 1) 0 0
0 C O \ o0 0 m 0 mm m

o 0 0 0 0 0 0 0 0 C 3 C


-4-












= 0
Co









C


Ciu )
L 0
0-








r.
44' ))
LA -C!
U -I LA0
LLJ )D
0 y


0 a m 0 o m n m- N -
0 o D mo o o
- CD m m m o o oo m

0 0 0m 0 0 m0


0 Co r)-'
J -
coo
000
0 0 C)
a a


m o L m \ o m Ln m o N L O N
N- o O N o o m

N N O N NO m0 N N N NN


S0 o m r o om
- .- o r L -

CA N o N rA rI A N CA


0




0
L.



C
0
O









































u
L
0

C

IC

0









L
I---



rC




o
LA










c


LA

u-


CO N 0
O0O m0


m m o -
NNO N m N A -N

- 0 -O -


CO _I Co C0 a
) r-- \ m N ) o


-o

NN

m m
_0-


(N NU

x) I C

X )< 1U


nr -
< <


\ r- N-) ) r- r-


-u o

-f N
m 0
_:- m


(N N

- -
co co




c -c


0 20


Ln 0

C,


Lm o A\ o o o o o r-
um C- N rtt 4 N-
- ,


4_A




c
ru ._







w4-


TO a-







f .-









Lua
i_ *


O mm
C L\ 1N


I


E m m- -I m -m mm
o co i mm( I mmm


Y LAO -T
N- CO Co m

a 4















LITERATURE CITED


Allison, J. C. S., and R. N. Curnow. 1966. On the choice of tester
parent for the breeding of synthetic varieties of maize
(Zea mays L.). Crop Sci. 6:541-544.

Burton, J. W., L. H. Penny, A. R. Hallauer, and S. A. Eberhart. 1971.
Evaluation of synthetic populations developed from a maize variety
(BSK) by two methods of recurrent selection. Crop Sci. 11:361-365.

Carangal, V. R. 1967. The effectiveness of S and topcross evaluation
in a recurrent selection program in corn (Zea mays L.). PI. Breed.
Abstr. 39(1):77.

Carangal, V. R., S. M. Ali, A. F. Koble, E. H. Rinke, and J. C. Sentz.
1971. Comparison of S with testcross evaluation for recurrent
selection in maize. C op Sci. 11:658-661.

Comstock, R. E., H. F. Robinson, and P. H. Harvey. 1949. A breeding
procedure designed to make maximum use of both general and specific
combining ability. Agron. J. 41:360-367.

Cress, C. E. 1966a. A comparison of recurrent selection systems.
Genetics 54:1371-1379.

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

Cress, C. E. 1967. Reciprocal recurrent selection and modifications in
simulated populations. Crop Sci. 7:561-567.

Darrah, L. L., S. A. Eberhart, and L. H. Penny. 1972. A maize breeding
methods study in Kenya. Crop Sci. 12:605-608.

Davis, R. L. 1934. Maize crossing values in second-generation lines.
Jour. Agr. Res. 48:339-359.

Douglas, A. G., J. W. Collier, M. F. El-Ebrashy, and J. S. Rogers. 1961.
An evaluation of three cycles of reciprocal recurrent selection in
a corn improvement program. Crop Sci. 1:157-161.

Duclos, L. A. 1967. Testcross versus S progeny performance as a basis
of selecting superior genotypes from a synthetic population of
(Zea mays L.). Pl. Breed. Abstr. 38(1):74.









Duclos, L. A. and P. L. Crane. 1968. Comparative performance of top-
crosses and S progeny for improving populations of corn
(Zea mays L.). Crop Sci. 8:191-194.

Eberhart, S. A., Seme Debela, and A. R. Hallauer. 1973. Reciprocal
recurrent selection in the BSSS and BSCBI maize populations and
half-sib selection in BSSS. Crop Sci. 13:451-456.

Falconer, D. S. 1960. Introduction to quantitative genetics. The
Ronald Press Co. New York pp. 365.

Gardner, C. 0. 1961. An evaluation of effects of mass selection and seed
irradiation with thermal neutrons on yield of corn. Crop Sci.
1:241-245.

Gardner, C. 0., H. P. Harvey, R. E. Comstock, and H. F. Robinson. 1953.
Dominance of genes controlling quantitative characters in maize.
Agron. J. 45:186-191.

Center, C. F. 1966. Development and selection of productive S, inbred
lines of corn (Zeaa mays L.). Crop Sci. 6:429-431.

Center, C. F., and M. W. Alexander. 1962. Comparative performance of
SI progenies and testcrosses in corn. Crop Sci. 2:516-519.

Center, C. F., and S. A. Eberhart. 1974. Performance of original and
advanced maize populations and their diallel crosses. Crop Sci.
14:881-885.

Hayes, H. K., and R. J. Garber. 1919. Synthetic production of high
protein corn in relation to breeding. Jour. Amer. Soc. Agron.
11:309-319.

Horner, E. S. 1963. A comparison of S line and S, plant evaluation
for combining ability in corn (Zea mays L.). rop Sci. 3:519-522.

Horner, E. S., H. W. Lundy, M. C. Lutrick, and W. H. Chapman. 1973.
Comparison of three methods of recurrent selection in maize.
Crop Sci. 13:485-489.

Horner, E. S., W. H. Chapman, M. C. Lutrick, and H. W. Lundy. 1969.
Comparison of selection based on yield of topcross progenies and
of S2 progenies in maize (Zea mays L.). Crop Sci. 9:539-543.

Hull, F. H. 1945. Recurrent selection of specific combining ability
in corn. J. Amer. Soc. Agron. 37:134-145.

Jenkins, M. T. 1935. The effect of inbreeding and of selection within
inbred lines of maize upon the hybrids made after successive
generations of selfing. Iowa State Coll. Jour. Sci. 3:429-450.

Jenkins, M. T. 1940. The segregation of genes affecting yield of
grain in maize. J. Amer. Soc. Agron. 32:55-63.









Koble, A. F. and E. H. Rinke. 1963. Comparative S line and topcross
performance in maize. Amer. Soc. Agron., Agron. Abstr. p. 83.

Kravenchko, V. A., M. C. Korolchuk, and G A. Morshed. 1971. Heterosis
in maize in respect to some economically valuable characters after
interline crossing. Pl. Breed. Abstr. 45(8):498.

Lonnquist, J. H. 1950. The effect of selection for combining ability
within segregating lines of corn. Agron. J. 42:503-508.

Lonnquist, J. H. 1968. Further evidence on testcross versus line
performance in maize. Crop Sci. 8:50-53.

Lonnquist, J. H., and M. Castro G. 1967. Relation of intra-population
genetic effects to performance of S, lines of maize. Crop Sci.
7:361-364.

Lonnquist, J. H., and M. F. Lindsey. 1964. Topcross versus S, line
performance in corn (Zea mays L.). Crop Sci. 4:580-584.

Moll, R. H., and C. W. Stuber. 1971. Comparisons of response to
alternative selection procedures initiated with two populations
of maize (Zea mays L.). Crop Sci. 11:706-711.

Rawlings, J. 0., and D. L. Thompson. 1962. Performance level as
criterion for the choice of maize testers. Crop Sci. 2:271-220.

Robinson, H. F., R. E. Comstock, and P. H. Harvey. 1949. Estimates of
heritability and the degree of dominance in corn. Agron. J.
41:353-359.

Robinson, S. L. 1976. Selection with an inbred tester versus S2 line
selection in maize. University of Florida, Master's thesis.

Russell, W. A., and S. A. Eberhart, 1975. Hybrid performance of
selected maize lines from reciprocal recurrent and testcross
selection programs. Crop Sci. 15:1-4.

Russell, W. A., S. A. Eberhart, and Urbano A. Vega 0. 1973. Recurrent
selection for specific combining ability for yield in two maize
populations. Crop Sci. 13:257-261.

Steel, R. G. D., and J. H. Torrie. 1960. Principles and procedures of
statistics. The McGraw-Hill Book Co. New York. pp. 481.

Tavares, F. C. A. 1972. Yield components related to heterosis in
inter-varietal hybrids of maize (Zea mays L.). PI. Breed. Abstr.
46(2):107.

Thomas, W. I., and D. B. Grissom. 1961. Cycle evaluation of reciprocal
recurrent selection for popping volume, grain yield, and resistance
to root lodging in pop corn. Crop Sci. 1:197-200.






63


Thompson, D. L., W. D. Hanson, and A. W. Shaw. 1971. Ear height
inheritance estimates and linkage bias among generation means of
corn. Crop Sci. 11:328-331.

Torregroza, M., and D. D. Harpstead. 1965. Performance of SI lines of
maize per se and as testcrosses to related and unrelated varieties.
Amer. Soc. Agron., Agron. Abstr. p. 21.

Walejko, R. N., and W. A. Russell. 1976. Evaluation of progress in
recurrent selection for specific combining ability in two
open-pollinated maize varieties. Amer. Soc. Agron., Agron. Abstr.
p. 66.

Webel, 0. D., and J. H. Lonnquist. 1967. An evaluation of modified
ear-to-row selection in a population of corn (Zea mays L.).
Crop Sci. 7:651-655.
















BIOGRAPHICAL SKETCH


Mesfin Ameha was born on April 30, 1942 in Huruta, Arusi, Ethiopia.

He graduated from Jimma Agricultural and Technical High School, Ethiopia,

in 1965. The same year, he enrolled at Haile Selassie First University,

now called Addis Ababa University, the College of Agriculture, Alemaya.

After completing the third year study, he participated in the Ethiopian

University Service program working for the Swedish Mission in Backo,

Shoa, as a school teacher and farm manager for one year. The following

year, in July, 1970, he received the Bachelor of Science Degree in

Agriculture.

Since graduation, he worked for the Institute of Agricultural

Research, Jimma, Ethiopia, as an Assistant Research Officer in coffee.

In December, 1972, he enrolled in the Agronomy Department of the

University of Florida, where he received the Master of Science Degree in

Agriculture in 1974 and is now a candidate for the Degree of Doctor

of Philosophy.











I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is fully
adequate, in scope and quality, as a dissertation for the degree of
Doctor of Philosophy.


/ ..-' 7 U I

Earl S. Horner, Chairman
Professor of Agronomy



I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is fully
adequate, in scope and quality, as a dissertation for the degree of
Doctor of Philosophy,





Charles E. Dean
Professor of Agronomy



I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is fully
adequate, in scope and quality, as a dissertation for the degree of
Doctor of Philosophy,



*I IO r7 - I / k ,. ..

Wayne B/ Sherman
Associate Professor of Fruit Crops



I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is fully
adequate, in scope and quality, as a dissertation for the degree of
Doctor of Philosophy.





Victor E. Green/
Professor of Agronomy









I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is fully
adequate, in scope and quality, as a dissertation for the degree of
Doctor of Philosophy.




Vincent N. Schroder
Associate Professor of Agronomy



This dissertation was submitted to the Graduate Faculty of the
College of Agriculture and to the Graduate Council, and was accepted as
partial fulfillment of the requirements for the degree of Doctor of
Phi losophy,

March, 1977




Dea College of Agrc Iture




De n, Graduate School




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs