        
       
   

l
\

W H
‘H
\ z

W

WWW

H

l M

lac);
Iowa I
.moooo

.3_ L.
" \
't‘C\‘-|‘\

43

L'

’ lullllllllllL [LL "L LLLl

3 12930

LIBRARY

Michigan State
Unlverslty

 

 

 

 

This is to certify that the

thesis entitled

EVALUATION OF QTL ALLELES FROM THE WILD GLYCINE SOJA
THAT INCREASE PROTEIN CONTENT IN GLYCINE MAX

presented by
Audrey M. Sebolt

has been accepted towards fulfillment
of the requirements for

MS . CSS
degree in

/%w@/%

 

Major professor

5140—627

 

Date

0-7639

MS U is an Afﬁrmative Action/Equal Opportunity Institution

 

PLACE IN REI‘URN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECAUJED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1m mu

EVALUATION OF QTL ALLELES FROM WILD GL YCINE SOJA THAT
INCREASE PROTEIN CONTENT IN GL YCINE MAX

By

Audrey M. Sebolt

A THESIS

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

MASTERS
Crop and Soil Sciences

1999

ABSTRACT

EVALUATION OF QTL ALLELES FROM WILD GL YCINE SOJA THAT
INCREASE PROTEIN CONTENT IN GL YCINE MAX

By

Audrey M. Sebolt

Genes from Glycine soja that increase protein content were ﬁrst evaluated
to determine if they would stably increase protein content in a backcross (BC)
population. Provided a consistent response was observed, the G. soja genes would
be analyzed in the genetic backgrounds of soybean cultivars ‘Parker’, ‘Kenwood’,
and C1914. RFLP analyses were conducted using markers that map to regions on
Linkage Groups (LG) E and I, where the high protein genes from G. soja mapped.
The G. soja allele on LG I increased protein content after it was backcrossed into a
G. max background. Lines homozygous for this G. soja allele exhibited a 174 kg
ha" yield reduction when compared to the G. max allele.

One line, selected from this BC population, was crossed to Kenwood,
Parker, and C1914. The G. soja allele on LG I increased protein content in the
backgrounds of Kenwood and Parker but not in the C1914 background. These
data suggest that the high protein gene in G. soja gene is allelic to an allele in
C1914 but not in Parker or Kenwood. The high protein allele from G. soja was
associated with less yield in the Kenwood and Parker crosses but not in the C1914

CI'OSS.

ACKNOWLEDGMENTS

I would like to ﬁrst thank my committee members Drs. Brian Diers,
James Kelly, and Amy Iezzoni for their support and guidance. A special thanks to
Dr. James Kelly for agreeing to be my advisor at the eleventh hour. I would also
like to extend my gratitude to John Boyse for assisting me with my ﬁeldwork.
Thank you Deane Lehmann for your moral and technical support, and friendship.
Not only did she teach me the ﬁner points of proper lab technique (try not to break
the glassware and a mop is very handy) but also helped me through the stressful
times.

Special thanks to my “stat” buddies Kirsten Jaglo-Ottosen (the other
blonde), Maria Theresa Ospina, and Suzanne Downey. Not only did they help me
through two semesters of stats, but also through this entire experience of ﬁnishing
my thesis. Thanks also to Ann Gustafson, Holly Little, and Sarah Gihnour, my
nmning buddies, for providing an outlet for accumulated stress these last few
months.

I thank my father and mother for instilling in me an appreciation for
agriculture. But most of all, many thanks to my husband, Chris, who has been
there since the beginning and traveled this road with me. Thank you for all of

your love and support when I needed it most!

iii

TABLE OF CONTENTS

List of Tables ......................................................................... v

List of Figures ......................................................................... vi

Chapter 1: Literature Review ....................................................... 1
Introduction .................................................... 1
Conventional Plant Breeding ................................ 2
Glycine soja .................................................... 6
Molecular Analyses ............................................ 8
References ..................................................... 14

Chapter 2: Evaluation of QTL Alleles from Wild Glycine soja that Increase

Protein Content in Glycine max ..................................... 17
Introduction ................................................... 17
Material and Methods ....................................... 18
Results and Discussion ...................................... 20
Conclusions .................................................... 29
References ..................................................... 31

Chapter 3: Effect of a High Protein QTL Allele from Glycine soja in Three

Genetic Backgrounds ................................................. 33
Introduction .................................................. 33
Material and Methods ...................................... 34
Results and Discussion ..................................... 37
Conclusions ................................................... 49
References .................................................... 51

iv

LIST OF TABLES

Chapter 2:
Table 1: RF LP markers signiﬁcantly (P < 0.05) associated with

protein content for 1996 and 1997 ............................................ 23
Table 2: RF LP markers significantly (P < 0.05) associated with

oil content for 1996 and 1997 .................................................. 24
Table 3: RF LP markers signiﬁcantly (P < 0.05) associated with

yield for 1996 and 1997 ........................................................ 26
Table 4: RF LP markers signiﬁcantly (P < 0.05) associated with

maturity for 1996 and 1997 ................................................... 27
Table 5: RFLP markers signiﬁcantly (P < 0.05) associated with

weight per 100 seeds for 1996 and 1997 ..................................... 28

Chapter 3:
Table 6: RFLP markers signiﬁcantly (P < 0.01) associated with

protein content for the Parker population ................................... 38
Table 7: RF LP markers signiﬁcantly (P < 0.05) associated with

oil content for the Parker population ........................................ 39
Table 8: RF LP markers signiﬁcantly (P < 0.05) associated with

yield for the Parker population ............................................... 41
Table 9: Genotypic class means for weight per 100 seeds for the

Parker population ............................................................... 41
Table 10: RFLP markers signiﬁcantly (P < 0.01) associated with

protein content for the Kenwood population ............................... 42
Table 11: RF LP markers signiﬁcantly (P < 0.05) associated with

oil content for the Kenwood population ..................................... 43
Table 12: RFLP markers signiﬁcantly (P < 0.05) associated with

yield for the Kenwood population ........................................... 45
Table 13: Genotypic class means for weight per 100 seeds for the

Kenwood population .......................................................... 46
Table 14: Protein and oil content means of the genotypic classes

in the C 1914 population ...................................................... 48
Table 15: Yield means of the genotypic classes in the C 1914

population ...................................................................... 48

LIST OF FIGURES

Chapter 2

Figure 1: RFLP markers used in this study and their corresponding
Linkage Groups (LG) .......................................................... 21

vi

CHAPTER 1

LITERATURE REVIEW

Introduction

Prior to the 1920’s, soybean [Glycine max, (L.) Merr.] was imported into
the US. due to the lack of interest in the production of this ﬁeld crop. In the mid-
l920’s, the discovery that soybean meal provided a high protein supplement for
livestock and poultry feed led to an increase in the demand for soybean.
Thereafter, soybean acreage in the US. rapidly increased (Smith and Huyser,
1987)

Soybean breeders have attempted to achieve higher seed protein of elite
cultivars with limited success. The lack of diversity in elite populations has
hindered this effort and genes for increased protein content have been sought
elsewhere. Wild populations are one source of “new” genetic diversity and
genetic mapping has been useful in identifying genes for increased protein content
in the progeny of interspeciﬁc crosses.

In the following pages, researchers attempts to achieve higher seed protein
are reviewed. Whether genes that increase seed protein were successfully

transferred from G. soja to existing soybean cultivars will be examined as well. In

addition to these topics, the use of molecular techniques and the progress of

genetic mapping in soybean will also be discussed.

Conventional Plant Breeding

Approximately 4.6% of the amino acid composition soybean protein
consists of methionine and cysteine (Wilson, 1987). Unfortunately, the amount of
these amino acids in soybean does not meet the nutritional recommendation set by
FAO and thus, some foods that contain soy-protein may need to be supplemented
(Wilson, 1987). Because these amino acids are important for animal nutrition,
increasing the level of methionine and cysteine in soybean will increase its
nutritional value. At this time, no breeding efforts have successfully altered the
level of these amino acids, in part, due to the expense of the assays and the time
involved to evaluate lines for amino acid concentration. Breeding programs
simply attempt to increase seed protein by using traditional plant breeding methods
as a means to increase total seed protein concentration. While this attempt has
been made by numerous breeders during the last 20 years, conventional plant
breeding has unfortunately not increased percent methionine and cysteine (Burton
et a1, 1982) of the soybean seed.

Researchers who ﬁrst initiated the process of increasing seed protein in
soybean, used experimental breeding lines as a source of genes for increased
protein content. In a comparison of direct and indirect selection, Miller and Fehr
(1979) developed two populations with the goal of increasing protein

2

concentration. Whereas indirect selection was used to increase protein content by
selecting lines with low oil, direct selection was used as a means to select lines
with high protein.

Direct selection resulted in an increase in seed protein two times greater
than detected through indirect selection. Despite the improvement in seed protein,
there was a signiﬁcant decrease in oil concentration when direct selection was
attempted. Because both protein meal and vegetable oil from soybean are of
value, to increase both or to increase one without decreasing the other would be
advantageous (Miller and Fehr, 1979). Indirect selection was shown to be more
effective in increasing seed protein content without reducing yield.

Brim and Burton (1979) used recurrent selection as a means to increase
seed protein content. After ﬁve cycles of recurrent selection, seed protein content
was increased by 10.2% higher than the base population in one of the four
populations studied. Despite this success, three of the four populations exhibited
decreased yields after cycle ﬁve of recurrent selection. In addition, negative
correlations between protein and oil concentration and seed protein and yield were
observed.

Plant breeders have also used the backcross (BC) method to increase seed
protein of soybean. The BC method is used 'to transfer a desirable gene or group
of genes from one plant into another that has an overall good agronomic
background (Allard, 1960). After the initial cross, successive backcrosses to the

recurrent parent are made to recover the characteristics of the recurrent parent

while maintaining the gene or group of genes from the donor parent. Lines must
then be selfed after the ﬁnal BC in order for the gene (or genes) that were
transferred to become homozygous (Allard, 1960). The total number of
backcrosses is dependent upon the crop and the objective of the breeding program
and the heritability of the trait.

The backcross breeding method has been used for over a century. When
working with small grain crops, Harlan and Pope (1922) expressed a preference
for the BC breeding method because the desired character transferred was ﬁxed; in
other words, the character was not apt to easily segregate out of the population. In
addition, the BC method is considered of value because morphological and
agronomical features of the improved variety can be predicted in advance and
there is a high degree of genetic control of a population (Allard, 1960).

Utilizing the BC method, Wehrmann et a1. (1987) attempted to transfer
genes that increase protein content from the G. max accession Pando to three high
yielding soybean cultivars. Seed yields, equivalent to the recurrent parent, were
recovered after only two backcrosses; however, there was limited success in
increasing seed protein content. Pando was recorded as containing 480 g kg'1 seed
protein. The three high yielding recurrent parents had 361, 362, and 413 g kg'1
seed protein content while the three highest yielding BC; Fz-derived lines had only
slightly greater seed protein content averaging 373, 376, and 427 g kg". The

insigniﬁcant increase may have been attributed to difﬁculties in identifying a

donor parent that could successfully transfer loci controlling high seed protein to
progenies of consecutive backcrosses (Wilcox and Cavins, 1995).

With the use of the BC method, Wilcox and Cavins (1995) also attempted
to transfer genes for increased seed protein content from Pando to the high
yielding cultivar Cutler 71. With each BC, rapid progress was made in recovering
yield and other agronomic traits, while maintaining high seed protein. Results,
which were similar to Wehrmann et a1. (1987), also suggested that not all of the
genes controlling seed protein contents in Pando were transferred to the recurrent
parent. The protein content of Pando was recorded as 498 g kg'1 however, the line
with the highest seed protein content was only 472 g kg". Despite the signiﬁcant
increase in protein content, there was no dramatic decrease in yield; BC3 progenies
were equal to or greater in seed yield than Cutler 71.

Pleiotropic effects may explain the inverse relationship between seed
protein and oil contents found in the ”previous studies (Graef et al., 1989).
However, the basis for decreased yields when seed protein contents are
signiﬁcantly increased may be more complex. If genes controlling the trait studied
are linked to yield genes, there could be difﬁculties in separating the two traits,
and to further complicate matters, yield is generally a polygenic trait. In the
populations they examined, Brim and Burton (1979) attributed the negative
correlation between seed yield and protein content to pleiotropy, tight linkages, or

both.

Hartwig and Hinson (1972) were successful in transferring genes from the
high protein parent D60-7965 to Bragg, a high yielding cultivar, with out a
reduction of yield. After two backcrosses, one line was equivalent in seed yield to
Bragg and seed protein content of D60-7965. They concluded “that high protein
genes per se did not signiﬁcantly inﬂuence seed yield.” Despite a signiﬁcant
negative correlation between yield and seed protein content in the BC 1 lines, there
was a positive correlation in BC; lines. This was because lines were reduced from
25% donor gerrnplasm in the BC. generation to 12.5% in the BC; generation, thus
decreasing the number of deleterious genes that may have contributed to the

decrease in yield.

Glycine soia

The wild progenitor of G. max is G. sofa and both are diploid annual
species (2n=40x), however, their plant morphology differs dramatically
(Hymowitz and Singh, 1987). G. max, which has never been found in the wild,
exhibits an upright, sparsely branched, bush-type growth habit, with seeds
weighing 10 to 20 g (100 seeds)". In contrast, G. soja is known for its undesirable
characteristics such as its susceptibility to lodging, Vining, and colored seed coats
(Weber, 1950; Carpenter and Fehr, 1986). The species has a tendency to shatter,
grows prostrate with the ground, and seeds weigh roughly 0.1 g (100 seeds)”. G.

soja is found in the wild and is distributed throughout China and adjacent areas

such as Korea, Japan, Taiwan, and some countries of the former Soviet Union. In
the past, G. soja was referred to as G. ussuriensis (Hymowitz and Singh, 1987).

Sources of genes that increase protein content in elite U.S. soybean
cultivars are limited due to the lack of diversity present in this germplasm. U.S.
soybean cultivars derived from the hybridization of plant introductions (PI; Keim
et al., 1989) and few accessions have made large genetic contributions to the
pedigrees of elite cultivars. It is estimated that only ten accessions contribute 88%
to the Northern U.S. germplasm (Delanney et al., 1983; Fehr, 1987). Furthermore,
G. max contains less genetic diversity than its progenitor G. soja because of
bottlenecks during the domestication process. These bottlenecks resulted in a loss
of alleles during domestication and further losses occurred through modem plant-
breeding practices (Tanksley and McCouch, 1997). The reduced genetic variation
in elite germplasm, because of these bottlenecks, has resulted in a slow rate of
genetic improvement by plant breeders. When breeding populations have low
genetic variation, breeders are not likely to identify new and useful gene
combinations (Tanksley and Nelson, 1996). In order to recapture lost alleles, a
plant breeder should consider the wild ancestors of crop species as a source of
“new” genes.

Weber (1950) published a study in which a segregating population was
created by a cross between G. soja and G. max. Results showed there was a
strong negative correlation between percent protein and percent oil in the
population, which is consistent with ﬁndings in G. max. Weber (1950) also

7

indicated that if a breeder attempted to transfer high protein content from the G.
soja parent to G. max, there would be difﬁculty in recovering seed size and oil
content. Harlan (1976) reported that transferring genes that increase seed protein
content from G. soja into G. max should be possible and the deleterious
characteristics of G. soja can be selected against in a backcrossing program.
Ultimately, yields could be recovered that are similar to the recurrent parent.

Using G. soja in breeding programs is simpliﬁed because G. soja and G.
max are interfertile, therefore, genes from G. soja can easily be transferred to G.
max through crossing. To obtain lines similar to the recurrent G. max, Ertl and
Fehr (1985) found that three backcrosses to G. max were required. Carpenter and
Fehr (1986) conﬁrmed that three backcrosses were needed and resistance to
lodging and absence of Vining were most difﬁcult to recover in early BC

generations.

Molecular Analyses

Molecular tools such as isozyme or restriction fragment length
polymorphism (RF LP) markers assist in the mapping of genes and the selection of
lines that contain transferred genes. Genetic markers can indicate genetic diversity
through the selection of marker alleles from the wild species. Molecular markers
have also proven beneﬁcial in helping to decrease the number of lines evaluated in
the ﬁeld because continuous selection of breeding lines can be imposed as lines
are advanced (Suarez et al., 1991). Despite these beneﬁts, this process is often

8

more costly when compared to traditional plant breeding (Tanksley and Nelson,
1996) because different resources and expertise are required compared to
traditional plant breeding.

Genes that control quantitative traits are referred to as quantitative trait loci
(QTL) (Hartl, 1994). Manipulation of QTLs could eventually lead to superior
varieties, however, genes controlling the trait studied must ﬁrst be mapped with
genetic markers (Hartl, 1994). The number of QTLs linked to markers and the
amount of recombination between the marker loci and the QTL are critical when
selecting for a trait such as increased protein content.

Successful application of marker-assisted selection has been accomplished
in tomato. Quantitative trait loci controlling soluble solids (SS) content were
mapped in a BC population using genetic markers. The genes that increased SS
were transferred from the wild to the cultivated tomato (Osborn et al., 1987).
Researchers ﬁrst screened a high SS derived BC line for variation in RFLPs
compared to the low SS recurrent and high SS donor parent. Two cDNA clones
that hybridized to RFLPs were identiﬁed and the authors concluded that the
RFLPs had been introduced into the BC line from the high SS donor parent
(Osborn et al., 1987).

To determine if RF LPs identiﬁed by the two cDNA clones were linked to a
gene(s) that controlled SS content in tomato, Osborn et a1. (1987) developed a
population in which the previously mentioned high SS derived BC line was
crossed to a low SS tomato processing line. Analysis of variance revealed that a

9

RF LP locus was linked to one or a group of loci affecting SS content. The RF LP
allele from the high SS BC derived line was associated with signiﬁcantly higher
SS content suggesting this linkage relationship could be used in a tomato breeding
program. Lines with increased SS content could be identiﬁed by selecting for the
RFLP allele.

Tanksley and Nelson (1996) proposed a new breeding method known as the
advanced backcross QTL analysis. Their purpose of the method was to compare
QTL analyses of traditional balanced populations with advanced backcross
populations. Results, through computer simulations, indicated that genetic
mapping should be done no later than the BC; or BC3 generation. Researchers
also concluded that the genotype and phenotype of lines in the BC; and BC3
generation resemble the recurrent parent more so than earlier BC generations
because the frequency of deleterious or undesirable alleles from the donor parent
is diminished.

In soybean, Graef et a1. (1989) and Suarez et al. (1991) studied the
association between isozyme markers and quantitative traits. Both groups of
investigators used the same two G. max by G. soja populations and found that
vegetative traits, including seed protein and seed oil content, were signiﬁcantly
associated with isozyme loci, although, associations were population speciﬁc.
Suarez et a1. (1991) found there were a limited number of polymorphic isozyme

loci between the parents of both crosses, a detriment to using marker-assisted

10

selection. Only six isozyme markers were polymorphic for the ﬁrst cross and
eight for the second cross.

To be useful across environments, QTLs must be found that are stable in
different environments. Environmentally sensitive QTLs are present only when
environmental conditions are similar to the environment in which the QTL was
ﬁrst identiﬁed (Mian et a1. 1996). Therefore, environmentally sensitive QTLs lack
consistency across environments. Over three locations, Mian et al. (1996)
discovered several molecular markers associated with seed weight in two soybean
populations. Effective marker-assisted selection for seed weight was feasible and
could be applied to other breeding programs because useful QTLs were identiﬁed
in several populations and environments.

QTL mapping of several traits in interspeciﬁc soybean populations has been
successful. Keim et a1. (1990) mapped QTL for seed hardness using RFLP
markers using a population created by a cross between the G. max experimental
line A81-356022 and the wild G. soja accession PI 486916,. Five QTL were
identiﬁed that combined explained 70% of the variation for seed hardness in the
population. Markers mapped to these regions could be used to develop genotypes
with varying levels of seed-coat hardness.

The locations of genes that increase protein content in soybean were
mapped by Diers et a1. (1992) in a population derived from a cross between the G.
max experimental line A81-356022 and the G. soja accession PI 468916. Several

markers associated with signiﬁcant variation for protein content were identiﬁed.

11

The RF LP maker K011 explained 42% of the variation for protein content and this
marker mapped to LG I. The most signiﬁcant markers (P < 0.01) mapped to LGs
E and I (Shoemaker and Specht, 1995), suggesting that important genes for protein
and oil content are located within these linkage groups. Furthermore, all G. soja
alleles at loci signiﬁcant for protein content were associated with greater protein
content than G. max alleles.

Brummer et a1. (1997) mapped genes controlling protein and oil content in
eight different soybean populations. One particular population exhibited a very
strong QTL for protein on LG I. A noteworthy difference between this population
and the other seven populations was that one parent for this population was 25%
G. soja. The RFLP marker A144 was used to detect this strong QTL and the
marker explained 27.5% of the genetic variation for protein content in the
population. QTL for protein and oil content were also found in other populations,
but these QTL were population speciﬁc. Data suggested that marker-assisted
selection could be used as a tool to pyramid different protein QTLs into a common
background, creating a population higher in seed protein content and superior to
populations already derived.

Most researchers value the use of interspeciﬁc crosses to gain knowledge of
where genes are located. Mansur et al. (1993), however, concluded that
segregating populations resulting from interspeciﬁc crosses should not be used to
study agronomic characteristics. Furthermore, populations that result from an
interspeciﬁc cross can not be evaluated in a meaningful way because these

12

populations have agronomic characteristics that are not desirable, such as trailing
vines and pods that have a tendency to shatter. A breeding population that results
from similar phenotypic parents is more ideal for QTL mapping and evaluation of
traits such as yield.

Further investigations of populations with G. soja and G. max as parents
should eventually lead to a better understanding of where genes that increase
protein content are located and their relationship with other traits. More
importantly, a backcross population in which genes from G. soja that increase
protein content were transferred to G. max needs to be conducted to evaluate the
stability of G. soja genes. Finally, the correlation between seed yield and protein
content is an economic issue that, with continued research, could be better

understood and possibly resolved.

13

REFERENCES

Allard, R.W. 1960. Principles of plant breeding. John Wiley and Sons, Inc.
New York, N. Y. p. 150-165.

Brim, CA, and J .W. Burton. 1979. Recurrent selection in soybean. II.
Selection for increased percent protein in seeds. Crop Sci. 19:494-498.

Brummer, E.C., G.L. Graef, J. Orf, J.R. Wilcox, and RC. Shoemaker. 1997.
Mapping QTL for seed protein and oil content in eight soybean
populations. Crop Sci. 37 1370-378.

Burton, J .W., A.E. Purcell, and W.M. Walter, Jr. 1982. Methionine
concentration in soybean protein from populations selected for increased
percent protein. Crap Sci. 23:744-747.

Carpenter, J.A., and W.R. Fehr. 1986. Genetic variability for desirable
agronomic traits in populations containing Glycine soja germplasm. Crop
Sci. 26:681-686.

Delanney, X., D.M. Rodgers, and R.G. Palmer. 1983. Relative genetic
contribution among ancestral lines to North American soybean cultivars.
Crop Sci. 23:944-949.

Diers, B.W., P. Keim, W.R. Fehr, and RC. Shoemaker. 1992. RF LP
analysis of soybean seed protein and oil content. Theor. Appl. Genet.
83:608-612.

Ertl, BS, and W.R. F ehr. 1985. Agronomic performance of soybean
genotypes from Glycine max X Glycine soja crosses. Crop Sci. 25:589-
592.

Fehr, W.R. 1987. Breeding methods for cultivar development. p. 249-293.

In J .R. Wilcox (ed) Soybean: improvement, productions, and uses. 2"d edn.
Agron. Monogr. 16. ASA, CSSA, and SSSA, Madison, WI.

14

Graef, G.L., W.R. Fehr, and SR. Cianzio. 1989. Relation of isozyme
genotypes to quantitative characters in soybean. Crop Sci. 29:683-688.

Harlan, H.V., and MN. Pope. 1922. The use and value of back-crosses in
small-grain breeding. Jour. Heredity 13:319-322.

Harlan, J .R. 1976. Genetic resources in wild relatives of crops. Crop Sci.
16:329-332.

Hartl, UL. 1994. Genetics. Jones and Bartlett Publishers. Boston, Ma. p.
245-247.

Hartwig, BE, and K. Hinson. 1972. Association between chemical
composition of seed and seed yield of soybean. Crop Sci. 12:829-830.

Hymowitz, T., and RI Singh. 1987. Taxonomy and speciation. p. 23-48. In
J .R. Wilcox (ed) Soybean: improvement, productions, and uses. 2nd edn.
Agron. Monogr. 16. ASA, CSSA, and SSSA, Madison, WI.

Keim, P., B.W. Diers, and RC Shoemaker. 1990. Genetic analysis of
soybean hard seededness with molecular markers. Theor. Appl. Genet.
79:465-469.

Keim, P., R.C. Shoemaker, and R.G. Palmer. 1989. Restriction fragment
length polymorphism diversity in soybean. Theor. Appl. Genet. 77:786-
792.

Mansur, L.M., K.G. Lark, H. Kross, and A. Oliveira. 1993. Interval
mapping of quantitative trait loci for reproductive, morphological, and
seed traits of soybean (Glycine max L.). Theor. Appl. Genet. 86:907-913.

Mian, M.A.R., M.A. Bailey, J.P. Tamulonis, E.R. Shipe, T.E. Carter Jr.,
W.A. Parrott, D.A. Ashley, R.S. Hussey, and HR. Boerma. 1996.
Molecular markers associated with seed weight in two soybean
populations. Theor. Appl. Genet. 93: 1 101-1016.

Miller, J .E., and W.R. Fehr. 1979. Direct and indirect selection for protein
in soybean. Crop Sci. 19:101-106.

Osborn, T.C., D.C. Alexander, and J .F. Fobes. 1987. Identiﬁcation of

restriction fragment length polymorphisms linked to genes controlling
soluble solids content in tomato fruit. Theor. Appl. Genet. 73:350-356.

15

Shoemaker, RC, and J.E. Specht. 1995. Integration of the soybean
molecular and classical genetic linkage groups. Crop Sci. 35:436-446.

Smith, K.J., and W. Huyser. 1987. World distribution and signiﬁcance of
Soybean. p. 1-22. In J.R. Wilcox (ed) Soybean: improvement, production,
and uses. 2lrml ed. Agron. Monogr. 16. ASA, CSSA, and SSSA, Madison, WI.

Suarez, J.C., G.L. Graef , W.R. Fehr, and SR. Cianzio. 1991. Association
of isozyme genotypes with agronomic and seed composition traits in
soybean. Euphytica 52: 137-146.

Tanksley, SD, and SR. McCouch. 1997. Seed banks and molecular
maps: unlocking genetic potential from the wild. Science 277: 1063-1066.

Tanksley, SD, and J.C. Nelson. 1996. Advanced backcross QTL analysis:
a method for the simultaneous discovery and transfer of valuable QTLs
from unadapted germplasm into elite breeding lines. Theor. Appl. Genet.
92:191-203.

Weber, CR. 1950. Inheritance and interaction of some agronomic and
chemical characters in an interspeciﬁc cross in soybean, Glycine max x G.
ussuriencsis. Ames, Iowa: Research Bulletin 374.

Wehrrnann, V.K., W.R. Fehr, S.R. Cianzio, and J.F. Cavins. 1987.
Transfer of high seed protein to high-yielding soybean cultivars. Crop
Sci. 27:927-931.

Wilcox, J .R., and J .F. Cavins. 1995. Backcrossing high seed protein to a
soybean cultivar. Crop Sci. 35:1036-1041.

Wilson, RF 1987. Seed metabolism. p. 643-686. In J .R. Wilcox (ed) Soybean:
improvement, productions, and uses. 2“d edn. Agron. Monogr. l6. ASA, CSSA,
SSA, Madison, WI.

CHAPTER 2

EVALUATION OF QTL ALLELES FROM THE WILD GL YCINE SOJA

THAT INCREASE PROTEIN CONTENT IN GL YCINE MAX

Introduction

Soybean [Glycine max (L.) Merr.] meal protein and oil are currently the
most widely produced, traded, and utilized protein meal and vegetable oil source
in the world. The protein meal constitutes approximately sixty percent of the total
soybean based products and the market for soybean is more dependent upon meal
than oil products (Smith and Huyser, 1987). With the global interest in soybean
products, research to improve the protein content of soybean is justiﬁed.

Previous research established that genes from the wild G. soja could
increase protein content in G. max. In a population developed by crossing a G.
max experimental line and a G. soja plant introduction (PI), Diers et a1. (1992)
identiﬁed two major quantitative trait loci (QTL) from G. soja that increase
protein content. The two QTL were mapped with restriction fragment length
polymorphism (RFLP) markers on Linkage Groups (LG) E and I of the soybean
linkage map (Shoemaker and Specht, 1995). Brummer et al. (1997) also mapped a
QTL for increased protein content to the same region of LG I as Diers et a1.

(1992). The population used by Brummer et a1. (1997) had one parent that was

17

25% G. soja, suggesting that their high protein gene also came from G. soja.

Indeed, there has been several attempts to increase soybean protein content
in existing populations, unfortunately researchers have found a negative
association between seed protein concentration, yield, and oil concentration
(Hartwig and Hinson, 1972; Miller and Fehr, 1979; Brim and Burton, 1979). A
better understanding of the interaction of these traits may improve breeding
strategies. Breeders may eventually increase protein or oil content while
simultaneously increasing yield, using marker-assisted selection (Brummer et al.,
1997)

The objective of this study was to analyze regions from G. soja on LG E
and I to determine if they will stably increase protein content in a backcross (BC)
population. The relationship of agronomic traits were also examined with markers

that map to these regions.

Materials and Methods

The genetic population used in this study was obtained from the population
derived from the interspeciﬁc cross between the G. soja accession Pl 468916 and
the G. max experimental line A81-356022 developed by Diers et al. (1992). For
our study, one line was selected from this original F; population because it was
homozygous for the G. soja regions associated with increased protein content on
LGs E and I. This line was the donor parent for three successive backcrosses to
A81-356022. During the backcrosses, the pb gene, a gene conferring pubescence

l8

tip (Palmer and Kilen, 1987), on LG E and the RF LP marker A144 on LG I were
used as selection criteria to recover the G. soja regions. The BC; F. plant, that had
both regions based on the RF LP marker and pb, was selfed and the population was
inbred to the F4 generation through single seed descent to develop F4 derived lines.

For each BC3, F4-derived line, leaf tissue was collected from several plants
and DNA extractions were conducted according to Kisha et a1. (1997) and
Southern blotting, hybridization, and autoradiography were performed as
described by Diers and Osborn (1994). A total of ten soybean RF LP markers
(Figure 1) were screened against parental DNA and digested with ﬁve restriction
endonucleases (EcoRI, Hindlll, EcoRV, Dral and Taql) to identify
polymorphisms.

Fifty-three BC3, F4-derived lines were evaluated during the smnrners of
1996 and 1997 near East Lansing and Britton, MI with one replicate at each
location. Plots were 4 m long, 2-rows wide, with 76 cm row spacing and sown at
a rate of 23 seeds rn'l row. During 1996, F436 lines were sown on 16 May in East
Lansing and on 23 May in Britton. In 1997, F4; lines were sown at East Lansing
on 4 June and Britton on 10 June.

Plots were rated both years for weight (100 seeds)", plant height, maturity
date, and lodging. Maturity date was rated as the number of days after 31 August
when 95% of the plants in a plot reached their mature pod color (R8) (F ehr et al.,
1971). Plant height and lodging where recorded when the plots were mature.
Plant height was measured as cm from the ground to the average terminal node of

19

plants. Plots were rated for lodging on a scale of one to ﬁve with one designated
as plants standing erect and ﬁve as plants lying prostrate to the ground.

Plots were harvested with a combine to measure seed yield and were not
end-trimmed during the growing season. Seed yields were not reported for the E.
Lansing location in 1997 because harvest equipment was unable to enter the ﬁeld
due to an unusually wet fall. However, seed was sampled from each plot for
protein and oil analyses and weight (100)'1 seeds.

Seed protein and oil content was measured at the USDA Northern Regional
Research Center at Peoria, IL, by using a Paciﬁc-Scientiﬁc NIR grain analyzer.
Measurements were taken on a 21 to 25 g sample for each plot. All data collected
were analyzed by standard analysis of variance procedures (ANOVA; SAS
Institute, 1987). The R2 value was used to describe the proportion of the genetic

variance in the population explained by individual markers in the population.

Results and Discussion

Molecular analyses were conducted using RF LP markers that mapped
regions approximately 53 centimorgans (cM) in length on LG E, and
approximately 26 cM in length on LG I (Figure 1). Marker loci on LG B were
found monomorphic, suggesting that this region was not successfully backcrossed
into the population. The RFLP marker B214, which mapped to LG I, was also
monomorphic. In Figure 1, the RFLP markers A144 and A688 are shown as
completely linked. Results in this study for the two RFLP markers were not

20

 

*WEEIII
Algal:
*eeoialll
Alleeoa
e233]
amid/l _ Eon.
$8.1 I
1.32 Also:
lEood
TEQVN.
.33me III
I:

29.: .283 Ea cexeeeeemv

Swill

 

cog—anon mmocoxomn E @068 Boo—BE...
$8820 Eoomonsnnna mcawcogacoouﬁo

Alﬁe”:

 

lilies:
also:
/ as:
.523 leeod
Also;
:82 I I
Also:
a ll
Also:
Bani II I
allege
£32 I. l
m on
ADI: 3:80 owed—=5

wﬁvcoamotoo :05 can beam was 5 tom: 33:3: a35— ”_ 233m

21

identical because the mapping population used to generate these linkage groups
(Figure 1) was a population other than the interspeciﬁc population used in this
study.

Only four markers that mapped to LG I were polymorphic. These four
RFLP markers were A144, A407, A515, and A688. Genotypic class means for the
homozygous G. soja and G. max classes and the heterozygous class were
examined, however, results are only reported for the G. soja and G. max
homozygous class means. The heterozygous class means are not reported due to
the small number of heterozygous lines; of the 53 lines in the population,
approximately 6% were heterozygous.

All four RFLP markers were signiﬁcantly (P < 0.01) associated with protein
content for the combined analyses of both locations for 1996 and 1997 (Table 1).
The most signiﬁcant marker for protein content was A144 (Table 1). This marker
explained as high as 76% of the variation for protein content, however, R2 values
were considerably lower in 1997 compared to 1996 data. Oil content was only
signiﬁcant (P --' 0.05) for both locations in 1996 (Table 2).

Despite lower R2 values in 1997 for A144, protein and oil means of the
genotypic classes for the G. soja allele were similar, and in some cases higher, to
the 1996 results. For seed protein content, lines ranged from 421 to 485 g (kg
seed)" in 1996 and 420 to 482 g (kg seed)" in 1997 for the population. For both

years of the study, the line with the lowest protein content was still greater than

22

Amee v A: Edema—mi 6: n m: a
.320 exec. .O mzowhnoEo: mega—$63 mm H awn—o BE .D meowtﬁoEo: Examine 22 JP

 

 

 

 

 

 

 

 

 

 

 

 

 

5% one 93. 5% New N: m3. mm
the m3 m3. wmv 3% N: emv 22
7903. we: w
em.e mme E .e mme ewe wee woe mm
Neeee Seed 2 Se Nmeee _ee.e .ee.e _eee.e k A & wwe<
a: v3 eev mi. New N; va mm
>3 9% mmv wmv mmv 2% Re 3:2
7:53 wee w
ewe 2 .e ve.e NNe med Nee mme NE
meeee wheee m: emeee _eee.e .eeee _eee.e k A m m _ m<
9% me Ev 3% New N3 N3 mm
wmv eve «3 wmv mmv 2% EV 3:2
73on mu: m
wme nme woe ewe Que 8e 8e Nm
_eee.e mBee eevee Neeee _eee.e _eee.e _eee.e k A K nev<
9% one new 9% new NS v9. “mm
0? w: vmv hmv came 2% Re e22
730% mu: m
eve «me :.e nme ere mhe wee E
_eee.e meeee mm: meeee _eee.e _eee.e _eee.e K A K 33..
359.80 eoEeEoU mam—=3 .m 52.5 eoEeEoU 3653 .m .855 coo—BE

 

“6-er

82
.32 Ba 92: E seas seen a? 8:688 Roe v e baseman; cease Sui ._ use.

wee—

23

.ﬁed v a: Emu—mama 8: n m: m
.350 exec. My gowaoEo: mug—wave mm w .330 BE .D Segue—:0: mouaawmmoe 3:2 .2.

 

 

 

 

 

 

 

 

 

 

 

 

 

52 m2 «3 >2 wE e2 e2 mm
e2 e2 mm: mm: ew— eS mE 3:2
7:53. mu: m
Nee wee ee.e ee.e ee.e mme one N:
m: m: m: 8 Seed _eee.e _eee.e k A m wwe<
e2 m2 e2 e2 e2 N2 e2 mm
wﬂ mm. m2 wﬂ e5 3: 3: 3:2
7:58 wccw
_e.e ee.e wee _e.e Se and . ee.e N:
m: m: m: 8 Seed _eee.e _eee.e k A m m _m<
02 N2 we: em: me_ _e_ e2 mm
of wﬂ m2 wﬂ em: m2 #3 3:2
780% mu: m
Nee ee.e wee Nee ewe ewe hne -~
8 m: m: m: _eee.e _eee.e :eeee k A & nev<
e2 NE 32 e2 mm: e2 of «mm
e2 e2 mm: e2 3: >2 mm— 23:2
7803 mu: m
mee _.e wee wee ee.e ee.e ee.e -~
m: m: me am: _eee.e _eee.e _eee.e K A m 33‘
eoEeEoU eoEeEoU mama“: .m 55.5 359800 mamas“: .m newt: H33:2

 

ee-eea

hee—

wee:

 

.32 as. £2 E .528 .6 a? 3283.. ES v .e assuage nine 3,; .N 0:3

24

that of the G. max parent, which produced 406 g (kg seed)" in 1996 and 462 g (kg
seed)" in 1997. The G. soja parent was previously reported to contain 471 g (kg
seed)" (Diers et al., 1992).

G. soja alleles were associated with greater protein content than G. max
alleles for all markers signiﬁcant for protein content on LG I. All G. max alleles at
signiﬁcant loci for oil were associated with greater oil content than G. soja alleles.
This general inverse relationship has been recorded in previous research in several
G. max populations and G. max by G. soja populations (Brim and Burton, 1979;
Miller and Fehr, 1979; Weber, 1950). G. soja marker alleles on LG I were also
signiﬁcantly associated with a decrease in seed yield (Table 3). For the combined
analysis across years, the marker A144 revealed that lines homozygous for G. soja
alleles yielded 174 kg ha" less than lines homozygous for G. max alleles (Table 3).

All markers on LG I were signiﬁcant (P < 0.05) for maturity in 1996 and
1997 (Table 4). At these loci, G. soja alleles were associated with earlier maturity
than G. max alleles. G. soja alleles were also associated with smaller seed size
(Table 5) when compared to G. max alleles. When comparing seed size between
the class means for the combined 1996 and 1997 data, G. soja alleles were

associated with a reduction of seed weight of 0.9 g (100 seed)".

25

.283 v e 6.3%? a: u a. m
.820 exec. U mnoweNoEoe Examine mm H .820 BE 0 maowﬁoEon meg—mace 3:2 +

 

 

 

 

 

 

 

 

 

 

 

 

 

 

33 33. eeem Rem e2 mm mm
Sum Nenm evem M: _ m SEN 3:2
.2 we.
_
ee.e ee.e . Nme vme Rd N:
m: m: _eee.e feed feed k A & wwe<
emem emem Neem mmwm 3mm mm
ewnm 3mm Sew wwem berm 3:2
7e: wx
ee.e me.e eme w: ..e m: .e N:
m: 3 Seed eweee wmmee K A m 2 m<
enem ehem gem wva 2 mm mm
nmbm mErN Dem meem mesa 3:2
7a: m:
ee.e mod and e_ .e mme N:
m: m: _eee.e mmeee eeeee k A & nev<
Gem Sen Gem eNmN mevm “mm
m mwm eebm enem e2 m e_ em 23:2
.2 we
2 .e 2e mve ewe hme N:
Seed mm: _eee.e Eeee .eeee k A & v3<
veg—5:80 :ottm 3:22:50 mega: .m cost: $2.83:
Sbee: see 092

 

.32 e5 32 é sea a? 3383 «83 v .e 3:82:23 waxes Sb. .m 2e?

26

awn—o exec. .D meowaoEo: 88:?er mm H
3220 BE .D meowaoEo: 83:282. 3:2 2.

 

mN

 

 

 

 

 

2. 2. 2. 2. em 3 mm

2. 2. 2. 2. mm 3 2 22

3S :3 3o 32 2.3 3.3 8o .2
583 883 583 583 383 383 .883 23 22..

2. 2. 2. 2. 8 em 3 mm

2. 2. 2. 2. mm 3 3m 22

23 ”no 33 e2 Moo Moo .2 .2
so: 883 383 883 583 583 e83 2A2 22

2. 3. 2. 2. em em 3 mm

2. 2. 2. 2. mm 3. 3m 22

£3 93 3o 33 3.3 2.3 28 E
883 583 883 58.3 583 883 383 2A2 33.

2. 3. 2. 2. em 3 3 ”mm

2. 2. 2. 2. mm 3 3m :22

32 ee.e 8° 23 ”no m2 ”No a:
883 883 883 883 38.0 82:. 23 2A2 2.2
8:328 852:8 2223 .m Seem 8:528 2:23 .m 8:5 .32.:

3-32 32 32

 

.32 2:. 32 c2 bases .23 3282183 v A: 228%? 28:2: 25. .2. 2.3

27

.233 v .2 3:32:23 8: u m: a
.320 2.2:. .D 283382— 822232. mm H 3220 :85 .D mzowzmoﬁo: 88:28.3 22 .2.

 

 

 

 

 

 

 

 

 

 

 

 

 

28

 

2.3 23 _.2 _.3 N3 2: 3.2 mm
23 22 22 3.3 22 .2 N2 22
:88 33: w
333 33 23 33 33 33 33 mm
m: 333 m: m: 3333 3333 3333 2A2 82.
N3 23 22 23 N3 23 22 mm
3.3 22 22 23 22 3.2 22 22
:88 33: w
233 333 333 83 33 2 3 83 :2
m: m: m: 2: 2.3333 233 3333 2A2 22.
.3 3.3 22 23 3.3 3.3 3.2 mm
23 22 22 3.3 22 2 N2 22
L 82 33: w
23 33 23 23 33 33 £3 :2
2:33 3333 383 _233 3333 3333 3333 2A2 32.2.
3.3 23 22 3.3 23 3.3 22 “mm
3.3 3.2 22 3.3 22 _.2 N2 :22
:82 33: w
33 23 23 £3 83 33 23 N2
3333 2333 :33 3333 8333 2333 3333 2A2 3.3..
852:8 85:58 MW5222 .m 828 85558 3523 .m 83.8 8.82
3-332 32 332

 

.32 3:: 332 :2 88: 32 83 522: 58 8.58%: A23 v 5 235.82% 2885 35.2 .n 232.

22212192

Our results show that the QTL on LG I continued to increase protein
content after it was backcrossed into a G. max background. Furthermore, a QTL,
once mapped in an interspeciflc soybean population, can have a similar genetic
effect after it is backcrossed into G. max. The effect of the protein QTL in LG I
during 1996 was similar to the ﬁndings of Diers et a1. (1992). However, this QTL
had a lesser effect in 1997. This may be due to the late sowing date of the
locations in 1997.

Unfortunately, with the increase in protein content, lines with the G. soja
QTL allele on LG I had lower yield. Because G. soja alleles were associated with
earlier maturity, this earlier maturity may have resulted in the yield decrease. This
may explain the decrease in yield to some extent; however, reasons for this
interaction may be more complex. The association between high protein content
and low yield may also be caused by the allele for lower yield being tightly linked
with the QTL that increased protein content, or caused by the protein QTL directly
reducing yield. With more backcrosses to the G. max recurrent parent and/or a
larger population size, recombination between the gene(s) that decrease yield and
the high protein gene may be found. If however, the yield reduction was due to a
pleiotrophic effect, then the yield reduction will be impossible to separate from
high protein content.

In the study conducted by Diers et a1. (1992), QTL for higher protein and

oil content were mapped to LG E and I (Shoemaker and Specht, 1995). During the

29

backcross process, the pb gene on LG E was used to select hybrids with the LG E
QTL. Although the pb gene was segregating in the BC3 population, the other
closely linked markers were not. Because pb was not signiﬁcantly associated with
seed protein content in the BC3 population, recombination between pb and the
QTL for higher seed protein content likely occurred during backcrossing resulting
in the loss of this QTL. Most likely, this was a double crossover event since the
RF LP markers that map to LG E surrounding the pb gene were monomorphic
while the BC3, F4-derived lines continued to segregate for this trait.

Flu'ther studies should be performed to evaluate the protein QTL in
populations other than BC populations, such as in crosses with cultivars. The
purpose of these studies would be to determine whether the high protein QTL is
effective in other genetic backgrounds and whether the association between low

yield and high protein content can be broken.

3O

REFERENCES

Brim, C.A., and J .W. Burton. 1979. Recurrent selection in soybean. II.
Selection for increased percent protein in seeds. Crop Sci. 19:494-498.

Brummer E.C., G.L. Graef, J. Orf, J.R. Wilcox, and RC. Shoemaker. 1997.
Mapping QTL for seed protein and oil content in eight soybean
populations. Crop Sci. 37:370-378.

Diers, B.W., P. Keim, W.R. Fehr, and RC. Shoemaker. 1992. RFLP
analysis of soybean seed protein and oil content. Theor. Appl. Genet.
83:608-612.

Diers, B.W., and TC. Osborn. 1994. Genetic diversity of oilseed Brassica
napus germplasm based on restriction fragment length polymorphisms.
Theor. Appl. Genet. 88:662-668;

Fehr, W.R., C.E. Caviness, D.T. Burmood, and 1.8. Pennington. 1971.
Stage of development descriptions for soybeans, Glycine max (L.) Merrill.
Crop Sci. 11:929-931.

Hartwig, EB, and K. Hinson. 1972. Association between chemical
composition of seed and seed yield of soybean. Crop Sci. 12:829-830.

Kisha, T.J., C.H. Sneller, and B.W. Diers. 1997. Relationship between
genetic distance among parents and genetic variance in populations of
soybean. Crop. Sci. 37:1317-1325.

Miller, J .E., and W.R. Fehr. 1979. Direct and indirect selection for protein
in soybean. Crop Sci. 19:101-106.

Palmer, R.G., and TC. Kilen. 1987. Qualitative genetics and cytogenetics.
p. 23-48. In J .R. Wilcox (ed) Soybean: improvement, production, and uses.
2“d ed. Agron. Monogr. 16. ASA, CSSA, and SSSA, Madison, WI.

SAS Institute. 1987. SAS/STAT guide for personal computers, Version 6
ed. SAS Institute, Cary, NC.

Shoemaker, RC, and J .E. Specht. 1995. Integration of the soybean
molecular and classical genetic linkage groups. Crop Sci. 35:436-446.

31

Smith, K.J., and W. Huyser. 1987. World distribution and signiﬁcance of
Soybean. p. 1-22. In J .R. Wilcox (ed) Soybean: improvement, production,
and uses. 2nd ed. Agron. Monogr. l6. ASA, CSSA, and SSSA, Madison, W].

Weber, CR. 1950. Inheritance and interaction of some agronomic and
chemical characters in an interspeciﬁc cross in soybean, Glycine max x G.
ussuriencsis. Ames, Iowa: Research Bulletin 374.

32

CHAPTER 3

EFFECT OF A HIGH PROTEIN QTL ALLELE FROM GL YCINE SOJA IN

THREE GENETIC BACKGROUNDS

Introduction

Soybean [Glycine max, (L.) Merr.] meal is an important source of protein in
many products such as livestock feed, baked goods, and adhesives (Smith and
Huyser, 1987). Increasing the protein content of existing breeding lines would
prove beneﬁcial and researchers have attempted to increase this (Brim and Burton,
1979; Miller and Fehr, 1979) in G. max derived populations. Each of these studies
report an increase in seed protein content was associated with a decrease in yield.

A source of genes for increased protein content can be found in G. 'soja, the
wild progenitor of G. max (Hymowitz and Singh, 1987). Diers et a1. (1992)
examined a population derived from a cross between the G. max experimental line
A81-356022 and the G. soja plant introduction (PI) 468916 (Diers et al., 1992).
Major QTLs from G. soja that were signiﬁcantly associated with an increase in
protein content were mapped to linkage groups (LG) E and I using restriction
fragment length polymorphism (RFLP) markers. The marker associated with the
greatest effect was K011 on LG I, which explained 42% of the variation for

protein content in the population.

33

The two G. soja QTLs associated with increased protein content were
further analyzed in a backcross population created by selecting one Fz-derived line
from the population used by Diers et al. (1992). This line, which canied both
QTLs, was a donor parent in three backcrosses, using A81-356022 as the recurrent
parent. The new population was developed to determine if the high protein alleles
from G. soja would have a stable effect in a backcross population. In this
backcross population, the QTL allele from G. soja on LG 1 increased protein
content (Sebolt and Diers, 1998). Lines with the G. soja allele, however, exhibited
signiﬁcantly lower yields than lines without this allele (Sebolt and Diers, 1998).
To further evaluate the high protein QTL allele, it was concluded that this allele
should be tested in other genetic backgrounds.

The objective of this study was to determine whether the QTL allele from
G. soja that increases protein content would also increase protein content in
crosses with the cultivars ‘Parker’ and ‘Kenwood’ and the experimental line
C1914. Agronomic traits were analyzed in these populations to determine the

effect of the high protein allele on these traits.

Materials and Methods

A population was developed from the interspeciﬁc cross between the G.
soja accession PI 468916 and the G. max experimental line A81-356022 (Diers et
al., 1992). For our study, one line from the original F2 population was selected
because it was homozygous for the G. soja regions associated with increased

34

protein content on LGs E and I (Diers et al., 1992). This line was used as the
donor parent in three backcrosses using A81-356022 as the recurrent parent.
During the backcrosses, the pb gene on LG E (Palmer and Kilen, 1987), and the
RFLP marker A144 on LG I were used to select for the G. soja regions where the
protein QTL were mapped. A BC3 F1 plant, that carried both G. soja regions, was
selfed and the population was inbred to the F4 generation using single seed descent
to develop F4-derived lines.

One BC3F4-derived line from the population was selected, because it was
homozygous for the G. soja region associated with increased protein content on
LG I, and crossed to the cultivars ‘Parker’ and ‘Kenwood’ and to the experimental
line C1914. The populations were inbred to the F3 generation using single seed
descent. The Parker population included 100 lines, while both the Kenwood and
the C1914 populations had 98 lines each.

Leaf tissue was collected from several plants from each F34 line from the
three populations and DNA extractions were conducted according to Kisha et al.
( 1997) and Southern blotting, hybridization, and autoradiography were performed
as described by Diers and Osborn (1994). Four RFLP markers, A144, A407,
A515, and A688 were screened against parental line DNA that was digested with
four restriction endonucleases (EcoRl, HindIII, Dral and T an). The region
analyzed was approximately 24 centimorgans (cM) on LG I.

The populations were evaluated as F34 lines in 1997 and as F35 lines in
1998. Plots were sown in 1997 on 9 June near East Lansing, MI. Plots were 1 m

35

long, one-row wide and sown at a rate of approximately 30 seeds m'l row, with a
76 cm row spacing. In 1998, lines were sown in Urbana, IL and near E. Lansing,
MI. Lines at the Urbana, Illinois location were sown on 27 April in plots 3.2 m
long, two-rows wide, with a 76 cm row spacing and at a rate of 39 seeds m'l row.
The E. Lansing location was sown on 12 May in plots 4.3 m long, six-rows wide,
with a 38 cm row spacing and at a rate of 25 seeds m'l row. The center four rows
in E. Lansing were harvested for yield estimation.

The 1997 and 1998 plots were evaluated for weight (100 seeds)’1 and seed
protein and oil contents. In addition, plots were evaluated for plant height,
lodging, matruity, and seed yield in 1998. Maturity was rated as the nmnber of
days after 31 August when 95% of the plants in a plot reached their mature pod
color (R8) (Fehr et al., 1971). Once plants reached maturity, plant height and
lodging where recorded. Plant height was measured, in centimeters from the
ground to the average terminal node of plants in each plot. Plots were rated for
lodging on a scale of one to ﬁve with one designated as plants standing erect and
ﬁve plants lying prostrate to the ground.

All plots were harvested with a combine and seed yield was measured only
in 1998. The plots were not end-trimmed during either year. Seed protein and oil
content were measured using a Paciﬁc-Scientiﬁc NIR grain analyzer at the USDA
Northern Regional Research Center in Peoria, IL. Measurements were taken on 21

to 25 g samples.

36

All data collected were analyzed by standard analysis of variance
procedures with PROC GLM of SAS (SAS Institute, 1987). The R2 value was

used to describe the proportion of genetic variance explained by each marker.

Results and Discussion

Molecular analyses were conducted on the populations derived from Parker,
Kenwood, and C1914 using the RFLP markers A144, A515, and A688 that
mapped to the region approximately 24 centimorgans (cM) in length on LG I
(Figure 1). For the three populations studied, marker loci on LG I were
polymorphic. Genotypic class means for the homozygous G. soja and Parker,
Kenwood, and C 1914 classes and the heterozygous class were examined, however,
results are only reported for the homozygous class means. The heterozygous class
means are not reported due to the small number of lines derived from heterozygous

plants; approximately 6% were heterozygous for the three populations.

Parker Population:

All three RF LP markers were found to be signiﬁcant (P - 0.05) (Tables 6
and 7) for seed protein and oil content. The most signiﬁcant marker for the
combined analyses of 1997 and 1998 was A144. This marker explained as much
as 44% of the variation for protein content and 15% for oil. R2 values were mostly
consistent across locations and years for protein content. For A144, the lines
homozygous for the G. soja allele had 20 g (kg seed)’l greater protein content that

37

.326 38. .D msowxwoﬁo: mouacwaoe mm H
awn—o Luv—Em. mzowznoEo: 3:2—wave mm +

 

mew emv mve Seed 2% wmv Nae _eee.e eve 36 Sue _eee.e 359600 weuhea
mew eNv eve _eee.e mew va e_.e Neeee eve emv mve _eee.e eoEanU
mg. emv 3e ﬁeeee 9% N3 e_.e meeee eve emv eve _eee.e maﬁa—«Am
5% wmv wNe Seed 9% N? :e need 3% wmv Nme _eee.e «52:3 wee—

mvv .Nv hme ~eee.e emv vmv Ce Seed 2% :3 Nme _eee.e wﬁmgqm hee—

 

 

 

mm me E KAK mm mm mm “Fm mm mm «m RAK 8:804 30>
7958 we w HGoomwv—ew hmeuom we: w

 

wwe< 2 m< v3<

 

douﬂsgm Exam 2: Se 88:3 5889 5;» 3583“ Wed v at bane—mama Eco—SE “SHE 6 63m...

38

.83 v e Eocene 8.. u 8 a
.320 axon .D mzowxnoEo: moscwuoe mm H
.320 Luv—Sm. mzowaoEo: megawaoe mm L.

 

 

 

 

 

:— 05 2e meee m2 wE med ~ee.e e: 02 2e weeee cop—5:80 $-32

NM: 2: Se neeee mm: 3: Se weed of e2 _~.e _eee.e eocﬁﬁou

mm: 3: :e Se 3: 5— :.e meee S. 2: 2e weee wEmSS .m

m: m2 Se oeeee D: S. ee.e mee :2 mo. eNe Neeee «Swat: woo—

wﬂ NE Se .0... of _e_ ee.e m: wﬂ N2 .ee mm: wimqﬁm nee.

wmmwgm mam—WW mm ink .Amembum mam—Wm Nm RAE wmwwmvm $5,"me mm kAm 5:804 50>
wwe< m~m< vE<

 

dots—smog 8x25 65 Se 23:8 :0 53> 3.2083 Woe v 5 3282.3an £8.35 “:5— K 2an

39

lines homozygous for the Parker allele across locations (Table 6). The protein
content of Parker was measured as 422 g (kg seed)’1 for E. Lansing in 1998 and
400 g (kg seed)’1 in Urbana.

The RFLP markers A144 and A688 were found to be signiﬁcant (P 0.05)
for yield in Urbana but not in E. Lansing (Table 8). For A688, the lines
homozygous for the G. soja allele yielded 251 kg ha'1 less than the lines
homozygous for the Parker allele in Urbana. Results were based on four harvested
rows, which were 4.3 m long, for E. Lansing and two harvested rows, 3.2 m long
for Urbana. The Parker parent yielded 3998 kg ha'1 in E. Lansing and 1909 kg ha'1
in Urbana.

RFLP markers were found to be non signiﬁcant (P - ' 0.05) for seed weight
with the exception of the marker A515 for Urbana (Table 9). Lines homozygous
for the G. soja allele had a 0.5 g (100 seed)’l less seed weight than the lines
homozygous for the Parker allele. RFLP markers were not signiﬁcant (P < 0.05)

for maturity, lodging, and height (data not shown).

Kenwood

All three markers, A144, A515, and A688, were signiﬁcantly (P 0.01)
associated with protein and oil content (Tables 10 and 11) for the Kenwood
population. The combined analyses of both locations for 1997 and 1998

demonstrated that the RFLP marker A144 explained 41% of the variation for

40

200.0 v 0: 0082.000 000 n 00 0
.0020 6.80 U msowaoEo: 0000:2020 mm H" .0020 100.009 mzowzmoEo: 0000:3020 mm .0

 

Num— e.m_ Nee 0: v.2 e.e_ wee 0: m.m_ wa— _e.e 0: 202.5800 makeo—
ve— v.2 med 0: QB ee— mee 0: _.e_ we— ee.e 0: “00320080
N.w_ WE ee.e 0: _.w_ v.2 .ee 0: v.3 N.w_ ee.e 0: maﬁa“; .m
men— e.v_ mee 0: 11 bé. wee wood :2 0.3 med 0: 05303 wee—

0: 0.: 00.0 00 0.: a: 00.0 00 0.: 0.: _0.0 000 05003.0 32
00 00 N00 07¢ 00 00 000 u& 000 2: 000 .040 00083 08>

 

 

 

09000 00:.01 78000 Se 0 09000 00 e 0

 

wwe< 2 m< vE<
dog—ace 000—000— 05 08 20000 e2 Enigma; 08 0:00.: 0020 280050 ”e 032—.

Amee v 0: 0:005:30 00: u 0: m
.0020 00x00 6 0:035:00: 0000:2020 mm H" .0020 .0023? 0aom>NoEoz 0000:w_00e mm 2.

 

mmhm Nhem nee m: 3% Sam med 0: eehm Nhem vee 0: BEA—ECU wee—

emwm evev Nee 0: vmem Nvev Ed 0: eeem veev Se 0.0: wﬁmawu—m

 

 

 

 

 

002 R: 2.0 000.0 000. 002 00.0 00 200 000_ 00.0 .00 0000.5 002
00 00 N00 090 00 00 N00 090 000 000 N00 030 0000000 00>
00000. 75100 000 00.
000< 22 3.2

 

dots—anon 00200; 05 08 20% 53> 2000200000 Wee v MM bugoaiwﬁ 0000—008 03: ”w 030,—.

41

.0020 00x00 .9 msomeoEo: 0000:3020 mm H
.0020 .eooacoz. msowzuoEo: 0803.020 v; +

 

 

 

 

 

030 wmv and Seed 3% emt.V eme _eee.e 2% wmv 3e _eee.e 8:30:00 $-30—

wvw emv Nee _eee.e 030 mmv eme .eeee e20 omv 3e .eeee eocﬁﬁoo

M030 wmv eve 586 w: mmv ewe .eeee emv wmv ewe _eee.e 9:055 .m

w: .20 wme _eee.e emv mmv one _eee.e emv emv Nme _eee.e 05303 wee—

:% emv _N.e Seed 3% 0N0 hue Seed .30 emv NNe _eee.e was—:01. .m 32

Memoa WW“ 0% Ram 7mm“; Mew—”Mm 0% RAR L-MWMom WNW w 0% kAm 0.200002 30>
wwe< 2m< vE<

 

020233 vooacov— 05 00.0 0:00:00 5000.5 53, 2000200000 Awee v 5 bag—mama 0000—008 0::

0. 2000

42

.0020 0.30 .b 0003380: 00332020 mm H”
.0020 .0003002. 00030080: 003020020 v3— 2.

 

E.— _w_ de _eee.e m: ew_ _Nd _eee.e m: N3 mmd Seed 60:50:00 neede—
eM: of emd Seed 0: mm: _Nd Seed e: 2: Rd ~eeed “00:30.80

ew_ 2: Cd meeed 2.— m2 2d meeed a: of :d meeed wﬁmﬁﬁm

 

 

 

 

 

em. N3 mmd .eeed em: e2 _~.e Seed em: m2 emd Seed 03003 002
NE M02 ee.e med e2 02 3d Seed me_ we- e_.e weed wigs .m 30—
mm MM 0% 070‘ mm MM 0m 07% 0mm DU— NK kAm 023004 30>
F.9000 we w 790000 mew 0-20000 we w

wwe< m_m< 32¢.

 

03020500 “0003002 05 00.0 0:00:00 :0 53> 2000200000 Amed v at 3:305:30 000238 ASHE H: 030...

43

protein content and 23% for oil content in the population. Furthermore, the G.
soja marker alleles were associated with greater protein content than the Kenwood
alleles. Protein content of the lines homozygous for the G. soja allele of A144
was 19 g (kg seed)’l greater than the lines homozygous for the Kenwood allele for
the combined analysis across 1997 and 1998. The protein content for Kenwood
was recorded as 405 g (kg seed)‘1 for Urbana and 420 g (kg seed)’1 for E. Lansing
in 1998. Both of these values for Kenwood were lower than the corresponding
Kenwood and G. soja class means from the population.

All three markers were signiﬁcant (P 0.05) for yield for Urbana, but none
were signiﬁcant for E. Lansing (Table 12). Lines homozygous for the G. soja
alleles of A688 had a 138 kg ha'1 lower yield compared to lines homozygous for
the Kenwood allele (Table 12) in Urbana. In 1998, the Kenwood parent yielded
5013 kg ha'1 in E. Lansing and 2029 kg ha'1 in Urbana.

The three markers were signiﬁcant (P 0.05) for seed weight in all
locations with the exception of A688 for E. Lansing in 1997 and 1998 (Table 13).
The lines homozygous for the G. soja allele of A515 had 1 g less (100 seed)’l than
the lines homozygous for the Kenwood allele across all three environments. RF LP
markers were non signiﬁcant (P - 0.05) for maturity, lodging, and height (data not

shown).

44

.33 v e 6.3%? s: u a. w.
.320 exam .9 303882 woes—Hwﬁoe mm H
.320 .eooaaoM. msowaﬁoEo: moumcwmmoe MM .r

 

 

 

 

 

 

mewm emem ee. we ﬂew :em nee m: eewm even ee.e m: 359800 wee—
mm; on? mee m: we; Nmmv Se 8 S; we? ee.e am: wEmwa—m
$2 32 mod mee $2 eat me.e me.e MKS 32 bed ee.e «SKID wei
mm MM N2 “Tm mm MM mm kAm ”mm *MM -~ RAK 5:804 30>
was wo— Lh: me. was wo—
wwe< 2m< v3<

 

dorm—smog eooBﬁM 05 Se 20% .23 33683 Ame.e v at mega—mama Eco—SE kind ”2 035.

45

.63 v me «59%? 8: n a: a
.820 3.8». .b msowaoEo; moemcwaoe mm H .320 .eooacoM. meowboﬁoc mueazwsoe MM ea

 

5.3 v.2 cod 3 e.3 WC Ed neeed m.3 Wm— e_.e eeed cop—5:80 wedde—
e.3 Num— ned m: w.3 0.2 2d Seed v.3 w.m_ m_.e Need 3:55.00
N: e6. med 2. ed— 0.: wed eeed wd— mg.— ned ee.e wﬁmcwqw
Num— h.m_ Cd eeeed ed— e.m_ vmd Seed e.N_ ed— e_.e _eee.e «Swat: mod—

N.: 3.. cod mm: a: a: :.o 83. 2: :2 .3 wood 383m 32
mm MM am hi mm MM am $43 ”mm Q: E “7: 8:83 a;

 

 

 

7803 8: w M33 8: w 7803 8: w

wme< 2 m< v3<
dose—ageleooBEM 2e 8e meoom ee_ SWEEPS Se 282: 330 33250 ”2 03m...

46

C1914

The three RFLP markers were not signiﬁcant (P ‘ 0.05) for seed protein
and oil content at any location or over locations for the C1914 population (Table
14). The G. soja genotypic class means for protein and oil content were similar to
the experimental line C1914. The means of both were 472 g (kg seed)’1 for the
combined analyses for 1997 and 1998 for all three markers. Furthermore, oil
contents for all three markers were also similar when the C1914 alleles were
compared to G. soja alleles. Yield was not signiﬁcantly different (P ' 0.05) at
any location or across locations (Table 15), however, a 57 kg ha'1 reduction in
yield was found when G. soja genotypic class means were compared with the
C1914 class means for Urbana for the marker A144. A 202 kg ha"1 reduction in

yield of the G. soja alleles for E. Lansing in 1998 was observed.

47

.320 exam .9 msowMNoEo—M 33:333. mm H
.320 .32 U. mzowzuoﬁon moﬁcwﬁoe 00 H

 

even even mwem wam deem Sam wag—«Am

 

 

 

$2 $2 $2 $2 82 def «5&3 weﬁ
mm DU mm 00 “mm e00 5:301— 30>
.23 we. .23 wx 7a: wx
mwe< 2m< v3<

 

dorm—aoqv—e— U 05 E 8320 238% 2: do 282: 22> ”m3 2an

.320 32. .U meowzono: 3:2—wave mm H .320 .320. msowaoEo: 3:2—wave DU .3.

 

 

 

 

 

 

S. of em. 02 on. S— N? NS N3 N? N: N: BEAM—~80 waged—
?! 3%— ve_ me. v2 v2 e? m? e: m5. 5: w: 322.80

mm: Nb— we. S. we_ $3 e3 e3 e: e: Ev e: maﬁa—Sm

N2 em: e2 e2 e2 of New ewv Sv New New ewv «5&3 M33
.2 e2 e2 e3 .2 e2 mew dew new Ev new dew wﬁmqﬁm nee—
mm UU mm 00 mm 00 mm 00 mm DU ”mm 200 5:802 50>

vacuum wvew
wwe< 2m< v3< wwe< 2m< v3<
=0 £88m

 

dorm—ace 320 05 2 $320 238% 05 mo £82: 28:8 mo e5 2805 ”3 052—.

48

Conclusions

The objective of this study was to test if a QTL allele from G. soja would
increase protein content in three different genetic backgrounds. This G. soja allele
increased protein content in the Kenwood and Parker backgrounds, but not in the
C1914 background. This suggests that there is a high protein allele that is allelic
with the G. soja allele in C1914 but not in Parker or Kenwood. Seed yields in E.
Lansing were twice as great as in the Urbana location. The low yields in Urbana
were at least partly the result of the populations being grown outside their range of
adaptation and that seed ﬁlling occurred during a dry period.

The high protein allele from G. soja was associated with less yield in
crosses with Kenwood and Parker but not in C1914. The data provide additional
evidence that the allele that increases protein content also lowers yield. If the high
protein gene was associated with less yield due to a coupling linkage with a yield
reducing gene, this coupling linkage would have to be present for both the gene
from G. soja and C1914. This is less likely than the high protein gene causing the
yield reduction.

If there was a coupling linkage between the genes that connibuted to
decrease seed yields and the allele for higher protein content, eventually, through
recombination, these effects could be separated. To do this, a larger population
size would need to be studied and/or the number of backcrosses to the recurrent
parent would have to be increased. If it were a pleiotrophic effect, then these traits
would never be separated.

49

 

In nearly all studies conducted to increase seed protein content, yield
reductions were found when protein content was increased. Not only should the
question of pleiotrophy be considered when deciding whether to continue with this
research, but also calculations for protein per hectare. Despite the signiﬁcant
increase in seed protein content, protein per hectare was only marginally greater in
lines homozygous for the G. soja QTL for increased protein content because of the
associated decrease in seed yield. For example, lines homozygous for the G. soja
allele for A144 produced 1754 kg ha’1 protein while lines homozygous for the
Parker allele produced 1748 kg ha'1 protein in E. Lansing during 1998.
Unquestionably, because protein per hectare for the Parker parent was only
slightly lower than lines associated with the G. soja homozygous class, it would
not be advantageous to continue research if the high protein gene caused the yield

reduction.

50

REFERENCES

Brim, C.A., and J .W. Burton. 1979. Recurrent selection in soybean. II.
Selection for increased percent protein in seeds. Crop Sci. 19:494-498.

Diers, B.W., P. Keim, W.R. Fehr, and R.C. Shoemaker. 1992. RF LP
analysis of soybean seed protein and oil content. Theor. Appl. Genet.
83:608-612.

Diers, B.W., and T.C. Osborn. 1994. Genetic diversity of oilseed Brassica
napus germplasm based on restriction fragment length polymorphisms.
Theor. Appl. Genet. 88:662-668.

Fehr, W.R., C.E. Caviness, D.T. Burmood, and J .S. Pennington. 1971.
Stage of development descriptions for soybeans, Glycine max (L.) Merrill.
Crop Sci. 112929-931.

Hymowitz, T., and R.J. Singh. 1987. Taxonomy and speciation. In:
Wilcox JR (ed) Soybean: improvement, productions, and uses, 2nd edn.
Agronomy 16:23-48.

Kisha, T.J., C.H. Sneller, and B.W. Diers. 1997. Relationship between
genetic distance among parents and genetic variance in populations of
soybean. Crop. Sci. 37:1317-1325.

Miller, J.E., and W.R. Fehr. 1979. Direct and indirectselection for protein
in soybean. Crop Sci. 19:101-106.

Palmer, R.G., and T.C. Kilen. 1987. Qualitative genetics and cytogenetics.
p. 23-48. In J .R. Wilcox (ed) Soybean: improvement, production, and uses.
2"d ed. Agron. Monogr. 16. ASA, CSSA, and SSSA, Madison, WI.

SAS Institute. 1987. SAS/STAT guide for personal computers, Version 6
ed. SAS Institute, Cary, NC.

Sebolt, A.M., and B.W. Diers. 1998. Evaluation of genes from the wild

Glycine soja that increase protein content in soybean. p. 69. In 1998
Agronomy abstracts. ASA, Madison, WI.

5l

 

Smith, K.J., and W. Huyser. 1987. World distribution and signiﬁcance of
soybean. p. 1-22. In J .R. Wilcox (ed) Soybean: improvement, productions, and
uses. 2Ind edn. Agron. Monogr. l6. ASA, CSSA, and SSSA, Madison, WI.

52

"111111111111111111111111111?
31293017876693