From the Department of Agronomy, University of Wisconsin, 1575 Unden Dr., Madison, WI 53706. We thank
E. T. BIngham for providing seed of the parents used
in this study, and K. K. Kldwell and S. Tavolettl for assistance with the molecular marker analysis.
The Journal of Heredity 1997*8(2)
References
Bayly IL and Craig IL, 1962. A morphological study of
the x-ray Induced cauliflower-head and single-leaf mutation In Medicago saliva. Can J Genet Cytol 4:386-397.
BIngham ET, 1966. Morphology and petiole vasculature
of five heritable leaf form* In Medicago sativa L Bot
Gaz 127221-225.
BIngham ET and McCoy TJ, 1979. Cultivated alialfa at
the dlplold level: origin, reproductive stability, and
yield of seed and forage. Crop Scl 1937-100.
Coen ES and Carpenter R, 1993. The metamorphosis of
flowers. Plant Cell 5:1175-1181.
Edit CS, Kldwell KK, Knapp SJ, Osbom TC, and McCoy
TJ, 1993. Linkage mapping In dlplold alfalfa Medicago
sativa. Genome 17:61-71.
Holland J, 1991. A survey of genetic Improvement for
yield and fertility In alfalfa (MS thesis). Madison, Wisconsin: University of Wisconsin.
Kempin SA, Savldge B, and Yanofsky MF, 1995. Molecular basis of the cauliflower phenotype In Arabidopsis.
Science 267522-525.
Kldwell KK, Woodfield DR, BIngham ET, and Osbom TC,
1994. Molecular marker diversity and yield of Isogenlc
2x and 4x single-crosses of alialfa. Crop Sci 34:784788.
Lander E, Green P, Abrahamson J, Barlow A, Daly M,
Lincoln SE, and Newburg L, 1987. MapMaker an Interactive computer package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1.174-181.
Marx GA, 1983. Developmental mutants In some annual
seed plants. Annu Rev Plant Physlol 34:389-417.
Meeks-Wagner DR, 1993. Gene expression In the early
floral meristem. Plant Cell 5:1167-1174.
Murray BE and Craig IL, 1962. A cytogenetlc study of
the x-ray Induced cauliflower-head and Single-leal mutation In Medicago saliva L. Can J Genet Cytol 4:379385.
Myers JR and Bassett MJ, 1993. Inheritance, allellsm,
and morphological characterization of unlfollate mutations In common bean. J Hered 84:17-20.
Tavoletti S, Veronesl F, and Osborn TC, 1996. RFLP linkage map of an alialfa meiotlc mutant based on an F,
population. J Hered 87:167-170.
Received August 3, 1995
Accepted July 29, 1996
Corresponding Editor Norman F. Weeden
Pod Dehiscence of Soybean:
Identification of Quantitative
Trait Loci
M. A. Bailey, M. A. R. Mian, T. E.
Carter, Jr., D. A. Ashley, and
H. R. Boerma
The dehiscence of pods (shattering) prior
to harvest is an undesirable trait of soy-
1 5 2 The Journal of Heredity 199788(2)
bean, Glycine max (L) Merr. Pod dehiscence (PD) is relatively uncommon in
modern North American soybean cultivars, but is often observed when unimproved germplasm or the wild species, G.
soja Siebold & Zucc, are used as parents
to introgress useful genes or to develop
genetically diverse breeding populations.
In light of the potential for efficient selection using DNA markers, the objective of
this study was to identify quantitative trait
loci (QTL) that condition resistance to PD.
A map of 140 linked restriction fragment
length polymorphism (RFLP) markers was
constructed using 120 F4-denved lines
from a soybean population (Young x PI
416937) that segregated for resistance to
PD. These lines were scored for PD on a
visual scale of 1 to 10 at both Athens,
Georgia, and Windblow, North Carolina, in
1994. Heritability of pod dehiscence was
92%. Associations of marker loci with QTL
that condition resistance to PD were tested using homozygous RFLP class means
in a single-factor ANOVA. A total of five
putatively independent RFLP markers
were associated with PD at both locations
and in a combined analysis over locations.
A single RFLP locus on linkage group J of
the USDA/lowa State University map accounted for 44% of the variation in PD
score. Epistasis was observed between
one pair of significant marker loci. These
results establish the genomic location of
one major and a few minor QTL, identify
an epistatic interaction, and indicate transgressive segregation which is plausibly
the result of susceptibility alleles contributed by the resistant parent.
Resistance to pod dehiscence (PD) prior
to harvest is required to approach the
yield potential of soybean cultivars. A level of PD that substantially reduces harvestable yield is rarely encountered in the
elite germplasm currently used for soybean cultivar development in North America. However, the use of exotic germplasm
as parents can result in a high degree of
PD in progeny. The potential for exploitation of such germplasm for the purpose of
introgressing desirable genes Into agronomically acceptable backgrounds and/or
for developing genetically diverse breeding populations lends impetus to a molecular genetic examination of PD.
Little has been reported concerning the
genetics of PD in soybean. No study has
determined the specific genes that condition PD in any soybean population, although it is generally regarded as polygenic and highly dependent on environmental
conditions (Caviness 1969; Helms 1994).
Caviness (1969), reported broad-sense
heritabillties for PD that ranged from 8998% in diverse populations. While the genetic basis of PD in soybean is largely unknown, the experience of soybean breeders indicates that visual selection for resistance is possible when PD occurs in
segregating populations. However, in
some environments field conditions are
unfavorable for expression of PD (Caviness 1969), such that the selection is either impossible or scoring is delayed until
conditions allow for expression of the
trait.
Populations developed from exotic parents almost invariably have unfavorable
alleles at many loci in addition to those
that condition PD. Several backcrosses to
elite parents are often necessary to recover desirable agronomic attributes (Carpenter and Fehr 1986). Marker-assisted selection for resistance to PD and for other desirable traits of soybean has the potential
to increase the efficiency of selecting superior progeny. Several agronomic and
seed-related traits have been mapped in
soybean (Keim et al. 1990; Lee et al. 1995;
Mansur et al. 1993), but no similar efforts
regarding PD have previously been reported for this crop. Recently common QTL
for resistance to shattering have been
identified from comparative maps of sorghum, maize, and rice (Paterson et al.
1995). While anatomical differences between the fruiting structures of cereals
and legumes almost certainly indicate fundamentally different shattering mechanisms, these results from cereal crops suggest that a similar identification of corresponding QTL among legumes will broaden the impact of soybean mapping
studies. Information on marker association with PD will be useful for practical applications such as map-based cloning and
marker-assisted selection. In view of the
potential implications for soybean breeding, the objective of this study was to
identify RFLP loci linked to QTL that condition resistance to PD in soybean.
Materials and Methods
Details of genetic mapping in this population of Young X PI 416937 have been described previously (Lee et al. 1996; Mian
et al. 1996). A total of 126 codominant and
29 dominant markers were used to construct a linkage map of 140 RFLP markers
in a population of 120 F4-derived lines. Fifteen markers were unlinked. The linkage
map was constructed using the Kosambi
Table 1. Mean pod dehUcence score for the
parents and the extreme F«-derlved lines from the
Young x PI 416937 cross
LGJ
LGE
LGL
A489-1
Score* for each location
Genotype
Combined
Athens
Wlndblow
Young
PI 416937
High progeny
Low progeny
S
03
6.9
8.0
0.1
1.1
0.5
5.3
8.0
0.0
1.2
0.1
7.8
8.3
0.0
1.4
• Score
hlscence).
Unlinked:
A725
A808n
14.8
23.1
K38S-1
B122-1
pod dehlscence) to 10 (>90% pod de20.6
map function of GMendel (Holloway and
K375-1n
Knapp 1993), assuming the data were col18.S
lected from F4-derived lines. The LOD mincr392-1r
imum of 3.0 and r maximum of 0.38 (ap0.5
proximately 50 cM) was used to construct
A233-1
the map. Markers detected by the same
4.8
enzyme and which had an identical moleccr497-1
ular weight to that of the image in SoyBase
1. RFLP markers associated with pod dehlscence (PD) In a Young x PI 416937 soybean population. The
(1995) were considered "anchor" loci and Figure
linkage group (LG) designations are according to Shoemaker and Specht (1995). Marker positions and estimated
were used to assign symbols to linkage
map distances (cM) are shown on the left side of each LG. Length of horizontal bars Indicate the R1 values for the
marker loci associated with PD In soybean based on combined data. A marker locus Is Identified by a probe
groups that corresponded to the USDA-ISU
designation and a dashed number suffix, where the latter Identifies the specific locus of the two or more loci
map (Shoemaker and Specht 1995).
detected by that probe.
Parents and 120 F4-derived lines were
grown in 1994 at two locations: Athens,
Georgia (Plant Sciences Farm) and Wind- pared for significant differences in PD us- (Shoemaker and Specht 1995), five to LG
ing an F test from the type III mean
J, and one to LG L. Two significant markblow, North Carolina (Sandhills Research
Station). The plot at Athens was two 3.66 squares obtained from the GLM procedure ers (A808n, A725) were unlinked (Figure
of SAS (SAS Institute, Cary, North Caroli1). Of these 12 marker loci, 5 were considm rows spaced 0.76 m apart, and the plot
na). A relaxed probability level of P £ .05 ered putatively independent and were deat Windblow was three 3.05 m rows
spaced 0.96 m apart. To reduce experi- was used, but only markers that were de- tected at both locations as well as in commental error due to soil heterogeneity tected at both locations and in the analy- bined analysis (Table 2). Putative indepenwithin each experimental site, the 120 sis of combined data across locations
dence was defined here as a marker locus
lines were divided into three groups of 40 were declared significant. These criteria that was greater than 50 cM from another
for significance were used to detect the marker significantly associated with the
lines based on maturity (early, medium,
consistency of putative QTL across loca- trait, and which acted in an additive manand late). The lines in each of these
tions and to minimize the probability of a
groups were placed in three separate
ner (i.e., no epistasis with other significant
tests, with three entries of Young and one type I error when a marker was not de- markers) with regards to explaining varientry of PI 416937 included in each test as tected in all three analyses. Two-way ANO- ation. It is possible that the two unlinked
reference genotypes. The experimental de- VA was used to detect epistatic interac- loci, A725 and A808n, flank a QTL on the
tions between significant markers.
sign was a randomized complete block
opposite side of another marker locus that
with two replications at Athens and three
is significantly associated with a QTL. If
replications at Windblow. The PD data
this is the case then these unlinked markResults and Discussion
were collected by visually estimating the
ers are not independent.
percentage of pods in a plot that had de- In the combined analysis, Young and PI
With the large number of comparisons
hisced on a scale of 0-10, where; 0 = <1%, 416937 differed by a PD score of 6.6 (Table tested for marker associations with PD, a
1 = 1-10%, 2 = 11-20%, 3 = 21-30%, 4 =
significance level of P ^ .05 would theo1). Transgressive segregation for greater
31-40%, 5 = 41-50%, 6 = 51-60%, 7 = 61retically result in several type I errors.
PD than PI 416937 was observed among
70%, 8 = 71-80%, 9 = 81-90%, and 10 =
However, probability of the occurrence of
the progeny in the combined analysis and
91-100%. In Athens the plots in each test
at Athens, but not at Windblow. The heri- such a type I error for the same marker at
were rated 4 weeks after the last F4-deboth locations is low (P = .0025). We
rived line in the test had matured. In Wind- tability of PD based on selection at two
therefore believe that the marker loci
locations
and
three
replications
was
92%.
blow a single rating was taken after all
identified in this study are associated with
lines had matured. In the combined ANO- There was a significant (P = .05) genotype
QTL that condition PD. Individually the
X
location
interaction,
but
the
variance
VA over locations, line and location were
five independent marker loci accounted
component
of
this
Interaction
was
only
considered random effects.
11% of the genotypic variance component. for 5.1% (A808n) to 44.4% (B122-1) of the
A total of 12 marker loci were associated variation in PD. When added together (asThe mean PD score from the F4-derived
suming independence) these five loci aclines were compared with the RFLP data. with PD in the combined analysis across
counted for 69% of the total variation. It
For each of the 155 marker loci, the ho- locations (Figure 1). Four of the markers
was impractical to place these five markmozygous RFLP class means were com- mapped to USDA linkage group (LG) E
Brief Communlcatioris 1 5 3
Table 2. Patatlve Independent RFLP loci associated with variation In soybean pod deblscence over two
locations
Combined
Allellc means
RFLP
locus
Linkage
group"
P
IP
(%)
B122-1
A489-1
cr274-l
A725
A808n
J
L
E
Unlinked
Unlinked
.0001
.02
.008
.01
.03
44.4
5.7
7.3
6.6
5.1
PI
Young 416937
Score*
1.1
1.9
2.8
3.3
2.7
3.9
2.9
1.7
2.1
1.8
Athens
Wlndblow
R1
P
R>
(%)
P
TO
.0001
.02
.02
.01
.02
39.1
5.6
6.0
12
5.7
.0001
.02
.009
.03
.05
425
5.00
7.1
53
4.1
• USDA linkage group (Shoemaker and Specht 1995).
'Score = 0 (<1% pod dehiscence) to 10 (90-100% pod dehiscence).
ers in a multiple regression model because
of the considerable loss of data (loss of
12.5% heterozygotes/marker).
The Young allele was associated with resistance to PD for the most important locus (B122-1; R2 = 44.4%), but at three of
the five putatlvely independent marker
loci PI 416937 alleles conditioned resistance (Table 2). Transgressive segregation
of progeny lines for greater PD than PI
416937 in Athens and in the combined
data supports the contribution of susceptibility alleles from both parents. It is
worth noting that Young has shown a
greater tendency for PD in some environments than most other contemporary cultivars of similar maturity (Burton J, personal communication). The fact that alleles of Young at marker loci cr274-l, A725,
and A808n condition increased PD provides a genetic explanation for the occasional tendency for PD of Young and for
the transgressive segregation observed in
progeny of Young X PI 416937.
Of the possible two-way epistatic interactions between independent markers
that were associated with PD in the combined analysis, one (B122-1 X cr274-l)
was significant (P < .01). This epistatic interaction was also observed in the analysis of each location individually. When a
line contained the Young allele at cr274-l
and the PI 416937 allele at B122-1, it had a
greater PD score than predicted by additive gene action. The R2 for the ANOVA including the main effects and the interaction of B122-1 X cr274-l was 49%. Individually these markers accounted for 44%
(B122-1) and 7% (cr274-l) of the variation
in PD (Table 2). Thus the magnitude of the
interaction was small.
We have previously reported on the
identification of QTL that condition maturity in this population (Lee et al. 1996). A
common assumption among experienced
observers is that the longer a soybean
plant remains in the field after reaching
1 5 4 The Journal of Heredity 1997 88(2)
maturity the more likely the expression of
PD. Ratings of PD in Athens were normalized for maturity by scoring plots 4 weeks
after the last line had matured in each of
the early, medium, and late maturing tests.
In contrast, a single rating of PD was taken
at Windblow, so plots differed in the time
spent in the field after reaching maturity.
Accordingly, a significant (/> = .05), negative phenotypic correlation coefficient (r
= —.33*; i.e., earlier maturity correlated
with higher PD score) between PD and maturity was observed at Wlndblow, but not
at Athens.
The magnitude of the correlation coefficient (r = .33*) at Windblow indicates
that maturity is not the overwhelming factor that conditions PD at this location.
Furthermore, none of the five QTL identified for PD corresponded to any of five
QTL that were identified for maturity from
data derived from the same plots (Lee et
al. 1995).
Regardless of the effects of maturity on
PD, a striking result of this study was the
consistency of QTL across locations. We
have shown that PD in this soybean population is highly heritable and is conditioned by one major and a few minor QTL.
Our data also indicate the contribution of
alleles for susceptibility to PD from a resistant parent, a finding that is not surprising in view of the propensity of Young
to dehisce in conducive environments and
the transgressive segregation for PD.
This study will facilitate efforts to further map PD genes in soybean and perhaps provide a basis for comparative mapping of the trait in other legumes. Such an
approach may facilitate map-based cloning of QTL. With the advent of improved
marker technology, our results will also allow breeders to utilize marker-assisted
strategies to efficiently select for resistance to PD.
From Pioneer Hl-Bred International, Inc., Johnston,
Iowa (Bailey), the Department of Crop and Soil Science,
University of Georgia, Athens, GA 30602-7272 (Mian,
Ashley, and Boerma), and the Department of Crop Science, North Carolina State University, Raleigh, North
Carolina. This research was funded by state and Hatch
funds allocated to the Georgia Agricultural Experiment
Stations and by a grant from the United Soybean Board.
The Journal of Heredity 1997:88(2)
References
Carpenter JA and Fehr WR, 1986. Genetic variability for
desirable agronomic traits in population containing
Gfycine soja germplasm. Crop Scl 26:681-686.
Cavlness CE, 1969. Herltablllty of pod dehiscence and
Its association with some agronomic characters In soybean. Crop Scl 9:207-209.
Helms TC, 1994. Greenhouse and field evaluation of
pod dehiscence In soybean. Can J Plant Scl 74:699-701
Holioway JL and Knapp SJ, 1993. GMendel 3.0 users
guide. CorvalUs, Oregon: Oregon State University.
Kelm P, Dlers BW, Olson TC, and Shoemaker RC, 1990.
RFLP mapping In soybean: association between marker
loci and variation In quantitative traits. Genetics 126:
735-742.
Lee SH, Bailey MA, Mian MAR, Carter TE Jr, Ashley DA,
Hussey RS, Parrott WA, and Boerma HR, 1996 Molecular markers associated with soybean plant height,
lodging, and maturity across locations. Crop Scl 36:
728-735.
Mansur LM, Lark KG, Kross H, and Ollveira A, 1993.
Interval mapping of quantitative trait loci for reproductive, morphological, and seed traits of soybean (Glycme max L). Theor Appl Genet 86:907-913.
Mian MAR, Bailey MA, Ashley DA, Wells R, Carter TE
Jr, Parrott WA, and Boerma HR, 1996. Molecular markers associated with water use efficiency and leaf ash In
soybean. Crop Scl 36:1252-1257.
Paterson PH, Un YR, U Z, Schertz KF, Doebley JF, Pinson SRM, Liu SC, Stansel JW, and Irvine JE, 1995. Convergent domestication of cereal crops by Independent
mutation at corresponding genetic loci. Science 29:
1714-1718.
Shoemaker RC and Specht JE, 1995. Integration of the
soybean molecular and classical genetic linkage
groups. Crop Scl 35:436-446.
SoyBase, 1995. A soybean genome database. Ames,
Iowa: Iowa State University.
Received December 20, 1995
Accepted July 29, 1996
Corresponding Editor. James L Hamrtck
Natural Outcrossing in
Grasspea
M. A. Chowdhury and A. E.
Sllnkard
The outcrossing rate of a species is important in designing experiments for inheritance and linkage studies and selection of appropriate breeding methods for
crop improvement. Though predominantly
self-pollinated, frequent heterozygosity
was found in isozyme studies of grasspea
(Lathyrus sativus L.). We established a