全 文 ：作物学报 ACTA AGRONOMICA SINICA 2010, 36(3): 496501 http://www.chinacrops.org/zwxb/
ISSN 0496-3490; CODEN TSHPA9 E-mail: xbzw@chinajournal.net.cn

This research was partially supported by National Natural Science Foundation of China (30671315) and Heilongjiang Province Natural Science Funds
for Distinguished Young Scholar (JC2OO617), and the Massachusetts Agricultural Experiment Station, University of Massachusetts Amherst.
Corresponding author: LIU Xiao-Bing, E-mail: liuxb@neigae.ac.cn
Received(收稿日期): 2009-09-24; Accepted(接受日期): 2009-11-10.
DOI: 10.3724/SP.J.1006.2010.00496
Seed Growth Characteristics of Some Short Season Indeterminate Soybeans
Stephen J HERBERT1, LIU Xiao-Bing2,*, Gurkirat BAATH1, JIN Jian2, ZHANG Qiu-Ying2, and Masoud
HASHEMI1
1 Center for Agriculture, University of Massachusetts, Amherst 01003, MA, USA; 2 Key laboratory of Black Soil Ecology, Northeast Institute of
Geography and Agroecology, Chinese Academy of Sciences, Harbin 150081, China
Abstract: The potential yield of a soybean crop is very specific to the genotype and the environment in which it is grown. For a
given genotype, the interaction of density with arrangements of plants will play a key role in determining the competition for
available resources especially solar radiation, water and nutrients, and thus maximum yield. Inter-plant competition begins when
the immediate supply of a single necessary factor falls below the combined demands of all plants. Variation in yield among fields
or years will be related to such inter-plant competition and borne out in one or more of the seed yield components of, plant per
unit area, pods per plant, seeds per pod or weight per seed (seed size). This paper discusses the yield component of seed size in
regulating differences in seed yield with changes in density and row spacing, and differences in seed yield between different years.
Several field studies are reviewed where seed size was shown to be similar between nodes on a plant and in pods with varying
seed number. However, the basal seed in two and three-seeded pods for many varieties tested was approximately 10% smaller than
the middle or terminal seeds. Cotyledon cell number in basal seeds was 10% less than in middle or terminal seeds and all had
similar cotyledon cell size. To test this further, we changed seed size through artificially source-sink manipulation, and by light
enrichment and shading. Differences in seed size were also related to differences in cotyledon cell number. This relationship of
seed size to cell number is intriguing since with induced changes in seed size, seeds at all nodes were of similar size even though
lower nodes began filling 15 to 20 days before upper nodes.
Keywords: Pod number; Seed size; Pod and seed position; Seed growth rate; Cotyledon cell number
早熟无限结荚习性大豆子粒生长特征
Stephen J HERBERT1 刘晓冰 2,* Gurkirat BAATH1 金 剑 2 张秋英 2
Masoud HASHEMI1
1 Center for Agriculture, University of Massachusetts, Amherst 01003, MA, USA; 2 中国科学院东北地理与农业生态所黑土重点实验室,
黑龙江哈尔滨 150081
摘 要: 大豆产量潜力受到基因型和环境条件的制约。一种基因型的密度、植株分布决定其对太阳辐射、水分和养
分的利用, 进而高产的形成。当群体生长所需外界要素之一不能满足时, 植株间形成竞争。产量的区域间及年际间差
异与这种株间竞争关系密切, 最终表现为单位面积内一个或多个产量构成因子的差异, 如株荚数、荚粒数、或单粒重
(籽粒大小)。本研究探讨籽粒大小在调节不同密度、行距条件下产量差异及年际间产量差异的作用。多点试验表明, 籽
粒大小在不同节位上及不同籽粒数的荚间差异不大。然而在 2 粒或 3 粒荚内, 荚基部粒比中部及顶部粒小 10%, 而
且子叶细胞体积差异不大。在改变源库、增强光照或遮阴条件下, 籽粒大小发生变化。籽粒大小与子叶细胞数相关。
籽粒大小是可塑的, 但即使底部节位荚较顶部节位提前 15~20 d鼓粒, 籽粒大小在所有节位间差异不大, 所以籽粒大
小与子叶细胞数的关系仍值得探讨。
关键词: 荚数; 籽粒大小; 荚和籽粒部位; 籽粒生长速率; 子叶细胞数
Growing plants in crop communities introduces com-
petition and the potential yield of a crop is very specific
to the genotype and the environment in which it is grown.
The variation in yield between fields or between years
will be related to variations in growth environment, and
borne out in one or more of the seed yield components of,
plant per unit area, pods per plant, seeds per pod or
weight per seed (seed size). The intensity and quality of
第 3期 Stephen J HERBERT et al.: Seed Growth Characteristics of Some Short Season Indeterminate Soybeans 497


solar radiation intercepted by the canopy are important
determinants of yield components and hence the yield of
soybean [1-4]. Water, nutrients, and weeds also interact
with the crop to changes in density, and spatial arrange-
ments of plants will influence the level of competition,
thereby affecting yield and individual yield components.
In a row width-density study, Herbert and Litchfield [5]
reported that an increased yield in one year of a two-year
study was mainly due to the change in seed size while
pod number per plant and seed number per pod were
similar for equivalent row widths or densities. The objec-
tive of this research was to further study the response of
seed size as an important yield component to manipula-
tion of source availability. Light enrichment and shading
were used to provide a quantitative estimate of the extent
of competition taking place in the crop community at
different stages of plant growth. The mechanism for in-
creased or decreased seed size was investigated in several
research studies and is presented and discussed.
1 Materials and methods
Field studies reported were conducted at the Univer-
sity of Massachusetts, Amherst, Crops Research and
Education Center Farm in Deerfield, MA, USA. The soil
at the experimental site was a Hadley fine sandy loam
(Typic Udifluent, coarse-silty, mixed, nonacid, mesic)
that freely drained but had good water holding capacity.
Rainfall was adequate for soybean growth and plants in
all studies did not exhibit water stress. Evans, a maturity
group 0 soybean, was used in most early studies except
as indicated in Fig. 1. Row width used was 25 cm and
density unless stated otherwise was 83 plants m–2. Soy-
bean plants showed good nodulation and good growth in
all studies. A pre-emergence mixture of 0.85 kg (ai) ha–1
Linuron [3-(3,4-dichlorophenyl)-1-methoxy-1-methylurea]
and 1.75 kg (ai) ha–1 Alachlor [2-chloro-2’, 6’-diethyl-N-
(methoxymethyl) acetanilide] supplemented with hand,
weeding was used for control of weeds in all years. Other
normal cultural practices for growing soybeans were fol-
lowed including additions of P and K when indicated as
needed by soil analysis.
In light enrichment studies, treatments were ambient
light (LE0), light enrichment initiated at late flower-early
pod formation (R3 stage) (LE2) and shading treatment
initiated at late flower-early pod formation (S2). Light
enrichment was achieved by installing 90 cm tall wire
mesh fencing (mesh hole size 4–5 cm) adjacent to the
rows bordering the center sample row, sloping away at a
45º angle from the center row, in each plot. Fences pre-
vented encroachment of plants from the neighboring
rows into the growing space, and thus increased the ra-
diation interception area of the sample row. The fences
were inspected periodically (1–3 times a week) and all
plants in rows bordering the center were pushed behind
the fences to prevent encroachment on the sample row.
Light intensity measurements, using a Li-Cor line quan-
tum sensor (LI-188B) placed parallel to, and beside the
center row plants, during the period from V5 to R3 indi-
cated control plots intercepted >95% and light enrich-
ments intercepted approximately 74% of the incoming
solar radiation. Thus, leaves at the base of the canopy in
light-enriched plots were receiving more than 25%
available light. Light intensity after R3 was always above
25% ambient light at the base of the canopy for the light
enriched treatments. Once put in place, the fences re-
mained in position until soybean maturity. Shade (S2)
was provided by using black polypropylene fabric at-
tached to a wire trellis held by wooden posts, installed
0.5 m above the soybean canopy and remained in place
until seed maturity.
Final yield for all studies for non-enriched treatments
(LE0), and for shade (S2), was determined by harvesting
plants in 4 rows with a length of 3 m. For light enrich-
ment treatments, 3 m of the central row between the
fences was harvested. For yield components analysis, 15
plants in each plot were harvested from a random starting
point at physiological maturity and data were recorded
for the whole plant as well as for each node position on
the main axis with node 1 being the unifoliate node. Data
recorded were pod number, seed number, stem dry
weight, pod dry weight and seed dry weight and from
these seed yield components were calculated.
For seed growth characteristics pods and seed samples
were collected every 5–6 days during the linear phase of
growth. Two and three seeded pods were collected to
determine the effect of position of the seed in the pod.
After the samples were collected, seed fresh mass was
recorded and seed volume was determined by a water
displacement method. Seeds were dried at approximately
70°C for 24 hours and the moisture content was calcu-
lated. Linear regression techniques were used to estimate
seed growth rates for each treatment, after eliminating the
nonlinear points from initial and final stages of seed de-
velopment. At maturity, plants were sampled to deter-
mine the final seed size. The effective filling period (EFP)
was estimated by dividing final seed size by rate of dry
matter accumulation in seeds during the linear filling
period.
The number of cotyledon cells was estimated using the
method described by Swank et al. [6] Seeds were dried for
24 hours (70°C) and then allowed to imbibe water for
8–10 hours, after which the seed coat and embryo axis
were removed. One cotyledon per seed was placed in a
fixative formalin acetic acid solution [7]. After 24 hours in
formalin acetic acid, each cotyledon was cut finely and
digested in 40 mL of chromic acid solution (80 g chromic
acid liter water–1) [8]. The chromic acid was removed after
5 days, and cotyledons were macerated before diluting
with water. An aliquot of the cell suspension was placed
on a hemacytometer, and the cells were counted under
×100 magnification. For the material not completely di-
498 作 物 学 报 第 36卷

gested, the suspension was filtered, and residues were
dried and weighed. The number of cells counted was
adjusted by the proportion of the total cotyledon mass not
digested to give total number of cells in the cotyledon.
Cell growth rate was calculated by dividing the seed
growth rate by cell number. Cell volume was obtained by
dividing maximum fresh seed volume by number of
cotyledon cells. The estimated number of cells per coty-
ledon pair was calculated using the known volumes.
2 Results
We examined 23 soybean genotypes belonging to
three different maturity groups for seed size variations
within two-seeded and three-seeded pods at six different
node positions on the main axis. In these genotypes the
size ranged from about 100 to 220 mg seed–1 (Fig. 1). It
was found that in all these genotypes the seed size re-
mained relatively constant across all node positions irre-
spective of the number of seeds per pod [20].
However, it was observed that the basal seed in both
two-seeded and three-seeded pods was smaller at maturity
than the middle and terminal seed. This trend occurred
regardless of genotype and was observed across node
positions and years. The position of seed within a pod
had an effect on the pattern of seed growth and final seed
weight at maturity (Table 1). The basal seed, which was
the closest to the vascular connection to the pod, lagged
behind in dry matter accumulation throughout seed de-
velopment. It had a lower rate of seed growth (6%) as
compared to the middle and terminal seed in the pod.
There was no significant effect of the seed position
within pod on the length of the effective filling period.
Variation in cell number of cotyledons not cell size
(weight and volume—data not shown) was responsible
for the differences in seed size within the pod (Table 1).
The cotyledon of seeds in the middle and terminal posi-
tions had a significantly greater number of cells as com-
pared to the cotyledon of basal seed. Weight and volume
of cells in the cotyledons of seeds in our study did not
differ between the three seeds within the pod, and growth
rate per cell did not increase with an increase in seed size
(Table 1).



Fig. 1 Mean weight per seed for different maturity groups of 23 soybean genotypes grown at University of Massachusetts Agronomy Re-
search Farm
Mean from 1 000 seed samples collected at harvest maturity.

Table 1 Seed growth and cotyledon characteristics of Evans soybean in relation to position in 3-seeded pods
Position Seed size (mg seed–1)
Seed growth rate
(mg seed–1 day–1)
Effective filling
period (d)
Cotyledon cell No.
(106)
Cell growth rate
(ng cell–1 d–1)
Cotyledon cell wt
(ng)
B 178.6 b 12.2 b 14.6 a 8.01 b 1.52 a 22.23 a
M 185.4 a 12.9 a 14.3 a 8.35 a 1.55 a 22.20 a
T 185.5 a 12.9 a 14.3 a 8.43 a 1.54 a 22.00 a
B, M, and T represent basal, middle and terminal seeds, respectively. Means within each column followed by the same letter are not significantly
different at the 0.05 probability level using Duncan’s multiple range test.
第 3期 Stephen J HERBERT et al.: Seed Growth Characteristics of Some Short Season Indeterminate Soybeans 499




Fig. 2 Dry weight for seeds within two-seeded and three-seeded
pods averaged for 23 soybean genotypes
B, M, and T are basal, middle, and terminal seeds.

Differences in seed size within pods occurred for all
node positions in two- and three-seeded pods (Fig. 2).
This result is intriguing since flowering and seed growth
occurs at different times for different node positions (Fig.
3). The development pattern of soybean (Fig. 3) indicates
then that cell division is occurring at different times (as
much as 15 days between node 3 and node 15) and that
some internal control is regulating the cell division pro-
cess and its duration.
In a source-sink manipulation studies (Liu et al. [16])
we were able to induce larger seeds by reducing pod load
and decrease seed size by reducing the source through
removal of the central leaflet on each main axis node (Fig.
4). Again in this study seed size was mostly similar
across node positions for each treatment despite seeds
being formed and filled at different times at each node
position.
In light enrichment studies, light intensity at the top of
canopy at beginning of pod formation was 13.06 and
7.34 μE s–1 m–2 for the unshaded and shaded treatments
respectively. At the base of canopy it was 0.19, 3.37, and
0.03 μE s–1 m–2 for LE0, LE2, and S2, respectively. Shad-
ing resulted in a 52% light reduction compared to the
control while leaves at the base of canopy in LE2 were
receiving more than 25% of the available light, compared
to less than 2% for LE0 and 0.2% for S2.
Significant differences were observed among the
treatments for pod number per plant, seed size and seed
growth rates (Table 2). Differences in pod numbers re-
flected greater pod retention for light enrichment and
greater pod loss for shading since both treatments were
imposed at early pod filling. Increased seed size with
light enrichment was similar to decreased sink (more
source per seed) and shading to a decreased source in the
source-sink study (Fig. 4). The light treatments had a
large effect on the pattern of seed growth during the fill-
ing period and final seed weight at maturity (Fig. 5). The


Fig. 3 Pods sampled from one plant at four nodes during seed
filling



Fig. 4 Dry weights of Evans soybean seeds in source limited (CL,
central leaflet removed each node), sink limited (SP, single pod each
main axis node) and untreated check (CK)
500 作 物 学 报 第 36卷

Table 2 Pod number, seed growth, and cotyledon characteristics of Evans soybean with light enrichment (LE2) and shading (S2) imposed
near the end of flowering
Treatment Pods per plant
Seed size
(mg seed–1)
Seed growth rate
(mg seed–1 day–1)
Effective filling
period (d)
Cotyledon cell
No. (106)
Cell growth rate
(ng cell–1 d–1)
Cotyledon cell wt
(ng)
LE0 13.3 b 162 b 13.8 b 13.5 b 8.6 b 1.60 b 21.5 a
LE2 16.6 a 199 a 17.3 a 11.6 b 9.4 a 1.86 a 21.6 a
S2 6.6 c 151 c 8.4 c 23.3 a 7.1 c 1.02 c 23.2 a
LE0, LE2, and S2 are control, light enriched, and shade respectively. Light treatment imposed at the R3 growth stage and remained to maturity.
Means within each column followed by the same letter are not significantly different at the 0.05 probability level using Duncan’s multiple range test.



Fig. 5 Seed growth with light enrichment and shade beginning at
the R3 growth stage

rates of seed growth were higher in light enrichment than
in the control, while shading resulted in significant low-
ering of seed growth rates (Table 2). These differences in
seed growth rate were significantly correlated with the
final size of the seed at maturity (r = 0.83**, Signf. 0.01
level).
However, in the present study, light enrichment had an
effect on seed growth rate but there was no effect on ef-
fective filling period, while for shade there was a reduc-
tion in seed growth rate and a partial compensating in-
crease (73%) in the effective filling period (Table 2).
Cotyledon cell number was sensitive to light condition
during flowering and seed development and differed un-
der the light treatments (Table 2). The final cell number
in the cotyledons of soybean plants under shade treat-
ments was reduced by as much as 30% in some of our
studies [2, 9, 21] and 19% in the present study. Shading of
soybean plants during seed development would have
resulted in a reduction of assimilate availability, and thus
would affect cell division in the cotyledons.
The growth rate of cells in the cotyledons differed in
all the three treatments and ranged from 1.02 to 1.86 ng
d–1 (Table 2). The rate of cell growth was significantly
correlated with seed dry weight at maturity (0.93***,
Signf. 0.001 level) and rate of seed growth (0.99***) in
this study.
3 Discussion
Seed size in soybean has been shown to be variable
between years but quite stable within a given year with
widely varying row widths, densities, across node posi-
tions, and for pods varying in seed number from one to
four [5,9]. It was found that in all these genotypes the seed
size remained relatively constant across all node posi-
tions irrespective of the number of seeds per pod [20]. The
less weight at maturity from basal seed in multi-seeded
pods than the middle and terminal seed across node posi-
tions and years in this study, indicated that seed size was
under the genetic control of the plant [5, 20-21]. Seed size in
soybean is determined by the rate of seed growth and
duration of seed filling, both of which partially are under
the genetic control (Egli et al. [17]) although the environ-
ment can modify these seed growth components [18-19].
Swank et al. [6] reported a variation in the duration of
seed filling made a significant contribution to the differ-
ences in seed size. Egli et al. [11] and Guldan and Brun [12]
found much of the variation in the seed size in soybean
genotypes, to be associated with the number of cells in
the cotyledons. Factors which affect the rate and duration
of cell division may ultimately determine the cell number
per cotyledon and consequently seed size. Differences in
growth rate of seeds correlated with number of cells in
cotyledons suggest that cotyledon cell number could be
the point of control. However, there are reports where
differences in cell size of cotyledons also contributed to
differences in seed sizes. Swank et al. [6] reported a varia-
tion in the size of the cotyledon cell to contribute to the
variation in seed size among soybean genotypes.
Hirshfield et al. [13] studied the effects of fertilizer treat-
ments on soybean seeds and found this resulted in an
increase in final seed yield which was associated with an
increase in cell size of cotyledons rather than cell number.
The relatively less cell number and lower growth rate in
basal seed, in this study, could contribute its less size in
two- or three-seeded pods.
Since a variation in the duration of seed filling made a
significant contribution to the differences in seed size
(Swank et al. [6]), a variation in light environment of the
soybean crop canopy can influence growth rates as well
as filling duration of seeds and thus both may influence
the seed size. Egli et al. [19] found that assimilate reduc-
tion treatments, when imposed during the initial stages of
seed development reduced cell division and cell number
in the cotyledons and resulted in reduced seed growth
rates. Light enrichment did increase seed growth rate and
shading slowed it, which could be the reason for the seed
第 3期 Stephen J HERBERT et al.: Seed Growth Characteristics of Some Short Season Indeterminate Soybeans 501


size difference in varied environmental conditions. Ma-
nipulation of the source-sink ratio by shading or defolia-
tion during initial stages of seed development consis-
tently resulted in a reduction in cotyledon cell number in
all genotypes tested. On increasing the source-sink ratio
by fruit removal, there was an increase in cotyledon cell
number. Similar changes in cell number presumably oc-
curred in our study with induced changes in the source-
sink ratio. Swank et al. [6] reported significant differences
in cotyledon cell number in one soybean genotype across
years and suggested that environment may have influ-
enced the number of cells in developing cotyledons of
seeds. Variations in the growth rate per cell probably re-
flect variation in the supply of assimilates to the seed
since this affects the growth rates of seeds in soybean [18].
Previous authors have reported a close association
between cotyledon cell number and growth rates in soy-
bean [11-12]. Although cell number in soybean is geneti-
cally determined [11] research presented above indicates
cell number can be influenced by the physiological envi-
ronment of the seed during the cell division phase of seed
development. Thus, growth rate of seeds in soybean is a
major function of the number of cells in the cotyledons
and is influenced by the supply of assimilates to the de-
veloping cotyledons.
In contrast to these results, Swank et al. [6] reported
variations in cotyledon cell size were associated with
differences in seed size in differing soybean genotypes.
As the seed size increased the number of cells per unit
mass of seed decreased, showing cell size to mainly con-
tribute to variation in seed size. However, the results of
our research in diverse studies indicate that variation in
seed size within one genotype was due to differences in
the cotyledon cell number and not cell size. Also, envi-
ronmental factors influencing cell division which may
explain much of the variation in seed size, and thus seed
yield, need further investigation.
4 Conclusion
Differences in seed size were related to differences in
cotyledon cell number, which was regulated by the
changes in light regime. A similar seed size was found
across nodes on a plant and in pods with varying seed
number. The basal seed in two and three-seeded pods was
approximately 10% smaller than the others, attributing to
cotyledon cell number rather than cotyledon cell size.
This relationship of seed size to cell number is intriguing
since with induced changes in seed size, seeds at all
nodes were of similar size even though lower nodes be-
gan filling 15 to 20 days before upper nodes. The stabil-
ity of seed size across node positions and how it is con-
trolled in the plant merit further research.
References
[1] Taylor H M, Mason W K, Bennie A T P, Rowse H R. Response of
soybeans to two row spacing and two soil water levels: I. An
analysis of biomass accumulation, canopy development, solar ra-
diation interception and components of seed yield. Field Crops
Res, 1982, 5: 1–14
[2] Willcott J, Herbert S J, Liu Z Y. Leaf area display and light inter-
ception in short season soybeans. Field Crops Res, 1984, 9:
173–182
[3] Myers R L, Brun W A, Brenner M L. Effect of raceme-localized
supplemental light on soybean reproductive abscission. Crop Sci,
1987, 27: 273–277
[4] Board J E, Harvelle B G. Explanations for greater light intercep-
tion in narrow vs. wide-row soybean. Crop Sci, 1992, 32:
198–202
[5] Herbert S J, Litchfield G V. Partitioning soybean seed yield
components. Crop Sci, 1982, 22: 1074–1079
[6] Swank J C, Egli D B, Pfeiffer T W. Seed growth characteristics of
soybean genotypes differing in duration of seed fill. Crop Sci,
1987, 27: 85–89
[7] Berlyn G P, Miksche J P. Botanical Microtechnique and Cyto-
chemistry. Ames, IA: Iowa State University Press, 1976
[8] Reinert J, Yeoman M M. Plant Cell and Tissue Culture. New York:
Springer-Verlag, 1982
[9] Mathew J P, Herbert S J, Zhang S H, Rautenkranz A F, Litchfield
G V. Differential response of soybean yield components to the
timing of light enrichment. Agron J, 2000, 92: 1156–1161
[10] Egli D B. Seed growth and development in soybean. Buenos
Aries, Argentina: World Soybean Conference IV. 1989. pp 256–261
[11] Egli D B, Fraser J, Legget J E, Poneleit C G. Control of seed
growth in soybeans [Glycine max (L.) Merrill]. Ann Bot, 1981, 48:
171–176
[12] Guldan S J, Brun W A. Relationship of cotyledon cell number
and seed respiration to soybean seed growth. Crop Sci, 1985, 25:
815–819
[13] Hirshfield K M, Flannery R L, Dale J. Cotyledon cell number and
cell size in relation to seed size and seed yield of soybean. Plant
Physiol Biochem, 1992, 31: 395–400
[14] Davis L A, Addicott F T. Abscisic acid: Correlation with abscis-
sion and with development in the cotton fruit. Plant Physiol,
1972, 49: 644–648
[15] Dunphy E J, Hanway J J, Green D E. Soybean yield in relation to
days between specific development stages. Agron J, 1979, 71:
917–920
[16] Liu X B, Herbert S J, Hashemi A M, Litchfield G V, Zhang Q Y.
Yield and yield components response of old and new soybean
cultivars to source-sink manipulation. Plant Soil Environ, 2006,
52: 150–158
[17] Egli D B, Orf J H, Pfeiffer T W. Genotypic variation for duration
of seed filling in soybean. Crop Sci, 1984, 24: 587–592
[18] Egli D B, Guffy R D, Meckel W, Legget J E. The effect of
source-sink alterations on soybean seed growth. Ann Bot, 1985,
55: 395–402
[19] Egli D B, Ramseur E L, Yu Z W, Sullivan C H. Source-sink al-
terations affect the number of cells in soybean cotyledons. Crop
Sci, 1989, 29: 732–735
[20] Herbert S J, Litchfield G V, Putnam D H. Seed size variation in
short season soybean. Mass Agron Res Rpt, 1985, 7: 8–10
[21] Herbert S J, Liu Z Y, Litchfield G V, Willcott J. Seed growth pat-
terns and seed size relationships in short season soybeans. Mass
Agron Res Rpt, 1984, 6: 45–49