1. Waterlogging Effects on Growth and Yield Components in Late-Planted Soybean
Geoffrey Linkemer, James E. Board, and Mary E. Musgrave*
ABSTRACT
irrigation and affects about 12% of the agricultural soils
in the USA (Boyer, 1982).
Inadequate oxygen supply for root respiration is the
main cause of reduced yield under waterlogged conditions (Grable, 1966; Russell, 1977). Lack of O2 inhibits
nitrogen and mineral uptake, and inhibits root growth
and nodulation in soybean (Sallam and Scott, 1987).
Thus, transport of nitrogen and/or minerals to the shoot
may be inadequate, resulting in chlorotic stunted plants
(Nathanson et al., 1984). Reduced leaf photosynthetic
rate, attributed partly to reduced stomatal conductance,
is also characteristic of waterlogged soybean (Oosterhuis et al., 1990). The cumulative effect of these events is
suboptimal CGR and reduced yield (Griffin and Saxton,
1988; Scott et al., 1989). Waterlogging effects on the
two factors controlling CGR, net assimilation rate
[NAR, g mϪ2 (leaf area) dϪ1] and leaf expansion rate
[LER, cm2 mϪ2 (land area) dϪ1] (Hunt, 1978), have not
been reported. Net assimilation rate is affected not only
by leaf photosynthetic rate, but also by light relationships within the canopy. Although waterlogging has
been reported to affect seed per pod and seed per plant
(Griffin and Saxton, 1988), yield components affecting
seed number (e.g., pod number, pods per reproductive
node, reproductive node number, and total node number) have not been studied. Greater understanding of
how waterlogging reduces yield would be gained by
identifying how growth dynamic parameters such as
CGR, NAR, and LER are linked with yield components
associated with the yield loss.
Crop growth rate has been usually affected only when
the waterlogging stress was applied for more than 2 d
(Griffin and Saxton, 1988; Scott et al., 1989). Greater
sensitivity to the stress was shown during the early reproductive (R1–R5) vs. vegetative periods (emergence–
R1). However, waterlogging sensitivities at specific developmental stages throughout soybean’s life cycle have
not been identified. Soybean may not be as tolerant of
waterlogging at late compared with optimal planting
dates, because of the shorter vegetative growth period
(emergence–R5) at late plantings (Board and Settimi,
1986). Documentation of late-planted soybean yield
losses caused by waterlogging induced by natural rain
fall is not available. Thus, our objectives were to: (i)
quantify the effects of waterlogging at different developmental stages on yield for late-planted soybean, (ii) determine waterlogging effects on growth dynamic parameters, yield components, and other physiological factors,
and (iii) assess the importance of waterlogging as a yieldrestricting factor for late-planted soybeans typically
grown in the southeastern USA.
A major agronomic problem in the southeastern USA is low yield
of late-planted soybean [Glycine max (L.) Merr.]. This problem is
aggravated by the adverse effect of waterlogging on crop growth.
Our objectives were to identify soybean growth stages sensitive to
waterlogging; identify yield components and physiological parameters
explaining yield losses induced by waterlogging; and determine the
extent of yield losses induced by waterlogging under natural field
conditions. Greenhouse and field studies were conducted during 1993
and 1994 near Baton Rouge, LA, (30؇N Lat) on a Commerce silt loam.
Waterlogging tolerance was assessed in cultivar Centennial (Maturity
Group VI) at three vegetative and five reproductive growth stages
by maintaining the water level at the soil surface in a greenhouse
study. Using the same cultivar, we evaluated the effect of drainage
in the field for late-planted soybean. Rain episodes determined the
timing of waterlogging; redox potential and oxygen concentration of
the soil were used to quantify the intensity of waterlogging stress.
Results of the greenhouse study indicated that the early vegetative
period (V2) and the early reproductive stages (R1, R3, and R5) were
most sensitive to waterlogging. Three to 5 cm of rain per day falling
on poorly drained soil was sufficient to reduce crop growth rate,
resulting in a yield decline from 2453 to 1550 kg haϪ1. Yield loss in
both field and greenhouse studies was induced primarily by decreased
pod production resulting from fewer pods per reproductive node. In
conclusion, waterlogging was determined to be an important stress
for late-planted soybean in high rainfall areas such as the Gulf
Coast Region.
I
ncreasing the yield of late-planted (after mid-June)
soybean in the southeastern USA is a major agronomic objective. Adverse weather and double cropping
after a winter cereal are the main reasons for late planting (Wallace et al., 1992). A major means of increasing
yield of late-planted soybean is to avoid environmental
stresses that slow crop grow rate [CGR, g mϪ2 (land
area) dϪ1] between emergence and R5 (Board and Harville, 1996a) (stages according to Fehr and Caviness,
1977). Stresses common to the southeastern USA that
may slow CGR during this period are drought (Muchow
et al., 1986), low soil pH (Mengel and Kamprath, 1978),
compacted soil (Smucker and Allmaras, 1993), and mineral toxicities (Hanson and Kamprath, 1979). Waterlogging is another common problem to the southeastern
USA that may adversely affect CGR, depending on
length of stress and when it occurs (Scott et al., 1989).
Waterlogging results from the ponding of water over a
poorly drained field after heavy rainfall or excessive
Geoffrey Linkemer and Mary E. Musgrave, Dep. of Plant Pathology
and Crop Physiology, and James E. Board, Dep. of Agronomy, Louisiana Agric. Exp. Stn., LSU Agric. Ctr., Baton Rouge, LA 70803. Approved for publication by the Director of the Louisiana Agric. Exp.
Stn. as manuscript no. 98-38-0006. Received 20 Jan. 1998. *Corresponding author (xp3031a@lsuvm.sncc.lsu.edu).
Abbreviations: CGR, crop growth rate [g mϪ2 (land area) dϪ1]; LER,
leaf expansion rate [cm2 mϪ2 (land area) dϪ1]; NAR, net assimilation
rate [g mϪ2 (leaf area) dϪ1].
Published in Crop Sci. 38:1576–1584 (1998).
1576

2. LINKEMER ET AL.: WATERLOGGING EFFECTS ON LATE-PLANTED SOYBEAN
MATERIALS AND METHODS
Field Experiments
Field studies were hand-planted with push-cone planters
on 12 July 1993 and 16 Aug 1994 at the Ben Hur Research
Farm near Baton Rouge, LA, (30Њ N Lat). Planting in 1994
was unusually late because of constant rainfall during July
and early August of that year. Soil type was a Commerce silt
loam (fine-silty, mixed, nonacid, thermic Aeric Fluvaquent).
Seed of cultivar Centennial (Maturity Group VI) was sown
at a high density and then thinned to a plant population of
220 000 plant haϪ1 by V3. This cultivar was selected because it
was reported as waterlogging sensitive (Heatherly and Pringle,
1991) and had demonstrated good yields in previous lateplanted studies (Board and Harville, 1996a). Stand counts
made throughout the growing season verified plant population. Plot size was 6 by 6 m, consisting of one part used for
combine-harvested plot yield and the other for plant sampling
throughout the growing season. On the basis of soil test recommendations, fertilizer was applied prior to planting at a rate
of 0-0-112 kg haϪ1 (N-P-K). Weeds, diseases, and insects were
suppressed by recommended pesticides.
The experiment had three factors: years (1993 and 1994),
sites (drained and undrained), and row spacings (25- and 100cm row widths). Sites were two adjacent locations which differed in drainage properties but had the same soil type. The
drained site (nonwaterlogged) was at the high end of a sloped
field having subsurface tile drainage; whereas the undrained
site (waterlogged) was the low end of a field lacking subsurface
tile drainage. The experimental design for each site was a
randomized complete block with four replications. The analysis of variance for combined sites and years was done according
to McIntosh (1983) with years as random factors and sites and
row spacings fixed. Data were taken at maturity on plot yield,
yield components per area, and total dry matter (g mϪ2). Plot
yield (adjusted to 130 g kgϪ1 moisture) was determined on a
4.3-m2 interior plot area that had been end-trimmed by 0.9 m.
Dates of emergence, R5 and R7 were recorded on all plots
based on five randomly selected plants. Yield components
were determined from 10 plants per plot which were randomly
selected and analyzed for node number (no. mϪ2) , reproductive node number (no. mϪ2) (node bearing at least one pod
with a developed seed), percentage reproductive nodes (%
nodes that become reproductive), pods per reproductive node
(no.), pod number (no. mϪ2), seed per pod (no.), seed number
(no. mϪ2), seed size (weight per 100 seed), and sample seed
yield (g mϪ2, dried to constant weight at 60ЊC for 4 d). Areal
determinations were based on stand counts measured at maturity. Sample seed yield gain due to improved drainage was
partitioned into percentage gain from seed number and seed
size by determining yield gain attributed to greater seed number (with seed size constant) and that attributed to greater
seed size (with seed number constant). All data were analyzed
by analysis of variance and means separated by the LSD after
significant F tests. Appropriate error terms were used for
all parameters.
Crop growth rate, NAR, and LER were determined from
eight total dry matter and leaf area index samples (determined
as previously described) taken at 5-d intervals during the vegetative and early reproductive periods. Leaf area index and
total dry matter were based on 0.5-m2 random samples obtained from bordered areas of the plot. Samples consisted of
a 0.5-m length of row in the 100-cm row width and a 2-m length
of row in the 25-cm row width. Leaf area was determined by
placing 50% (by fresh weight) of the leaf blades through a
LI-COR 3000 portable leaf-area meter (LI-COR Inc., Lincoln,
NE). Total dry matter was determined after drying to constant
1577
weight at 60ЊC in a forced-air dryer. A stepwise regression
analysis (Hunt and Parsons, 1981) was used in which linear,
quadratic, and cubic components were successively tested for
significance and included if the residual sum of squares was
significantly reduced. The r 2 values for total dry matter varied
from 0.88 to 0.95, except for the 100-cm row spacing in the
undrained site in 1994 (r 2 ϭ 0.84). The r 2 values for leaf area
index ranged from 0.84 to 0.94, except for the undrained site
in 1994 (r 2 ϭ 0.70). Crop growth rate, NAR, and LER were
determined by the regression package using the predicted
levels of total dry matter and leaf area index. Significant differences were determined by t-tests using standard errors calculated by the regression program.
Soil oxygen concentrations (mmol molϪ1) were determined
at random locations within each site ϫ row spacing experimental unit at V7, R1, R2, R3, R5, R6, R6.3, and R7. Oxygen
probes were constructed of porous sintered bronze cups
attached to sampling taps as described by Dowdell et al. (1972)
and placed at a 10-cm depth to sample rhizosphere gases. Gas
was sampled from within the probes with a 50-cc syringe using
a three-way valve that prevented any influx of gas into the
probe except from the soil. Oxygen concentration of the gas
sample was determined by an oxygen analyzer (Lazar DO166, Musgrave and Ding, 1998) that had been zeroed with
nitrogen gas and then the potential (mV) of an atmospheric
air sample was compared with the potential of the rhizosphere
air sample. Determinations were made at the evaluated stages
of crop development listed above or up to 2 d later if the soil
was not sufficiently drained to allow access to the oxygen
probes. For each developmental stage, all probes were sampled on the same day. Statistical analyses for rhizosphere oxygen content were done by using the GLM and REG procedure
in SAS (SAS Inst., 1987). Significant differences are stated at
P Ͻ 0.05.
Greenhouse Experiments
The study was conducted in an open-ended Quonset-type
greenhouse. Cultivar Centennial was planted on 12 July 1993
and 15 July 1994. Soil was collected from the top 30 cm of
the field test site and heat sterilized for 24 h. Soil was then
spread on benches in the greenhouse to air for two weeks and
used to fill 9-L plastic pots. Soil waterlogging was simulated
by immersing the pots in tanks equipped with float-controlled
valves to maintain the desired level of deionized water. Control conditions were achieved as the minimal water level required to maintain proper plant growth. Experimental design
was a completely randomized design with four replications
where each pot containing three plants represented a replication. Because the soil was sterilized, 50 mL of a water solution
of Nitrogin commercial inoculant (Bradyrhizobium japonicum
‘S’ Culture; LiphaTech Inc., Milwaukee, WI; 10 g LϪ1) was
applied to each pot. Fertilizer was applied prior to planting
at a rate of 0-0-112 kg haϪ1 (N-P-K).
Sixteen treatments were evaluated: a control (no waterlogging); eight transient waterlogging treatments consisting of 7 d
of waterlogging starting at V2, V3, V7, R1, R3, R5, R6, and
R6.3 (temporal midpoint of seed filling); and seven continuous
waterlogging treatments from the start of the established
stages to the end of the life cycle (no continuous treatment
was initiated at R6.3). During treatment periods, water level
was raised to just cover the soil surface. At termination of
treatment, water level was lowered to the control position.
Soil redox potential was used to quantify the waterlogging
treatment intensity. Platinum-tipped electrodes were set at a
5-cm depth in the soil with two electrodes for each treatment
(Musgrave, 1994). Each electrode was calibrated prior to use

3. 1578
CROP SCIENCE, VOL. 38, NOVEMBER–DECEMBER 1998
Table 1. Soil oxygen concentration (mmol molϪ1) at 10-cm depth
in drained and undrained sites of Centennial soybean planted
on two row spacings near Baton Rouge, LA, 1993 and 1994.
Year
Row
spacing
1993
cm
25
100
1994
Fig. 1. Daily precipitation during the crop life cycle of soybean grown
at a late planting date near Baton Rouge, LA, 1993 (A) and
1994 (B).
with quinhydrone-saturated solutions at pH ϭ 4 and pH ϭ 7.
Yield and yield component data were obtained in a manner
similar to that described for the field study. Statistical analyses
were done with the GLM and REG procedures in SAS (SAS
Inst., 1987). Significant differences were determined according
to LSD (P Ͻ 0.05).
RESULTS
Field Experiments
Heavy rainfall (5–6 cm per episode) at V2, near V7,
and during the R1 to R5 period occurred in 1993 (Fig.
1). Similar heavy rainfalls occurred in 1994 between
emergence and V2, near R1, and during the R6 to R7
period. After each period of heavy rainfall, ponding was
observed in the undrained site; whereas the drained site
was kept free of standing water (Board and Harville,
1996b). Drainage differences between sites were reflected in O2 concentrations obtained after both sites
were sufficiently drained to allow entrance (Table 1).
Because of significant year ϫ site ϫ row spacing interaction, soil O2 concentrations for each row spacing ϫ development stage treatment are presented for each year.
Significantly (P Ͻ 0.05) lower O2 concentration was
measured in the undrained vs. drained site for both row
spacings in 1993 at R2 and R3. In 1994, similar significant
reductions in O2 concentrations (P Ͻ 0.05) were measured at V7, R1, and throughout the R6 to R7 period.
Because all oxygen measurements were made after the
soil was sufficiently drained to allow access after a rain,
O2 concentrations in the undrained site directly after
heavy rain were probably much lower than those shown
in Table 1.
25
100
Developmental
stage
Soil oxygen content
Drained site
Undrained site
V7
R1
R2
R3
R5
R6
R6.3
R7
V7
R1
R2
R3
R5
R6
R6.3
R7
O2 (mmol molϪ1)
169
178
167
182
183
153*
173
170*
161
171
161
146
172
160
174
177
183
182
177
181
195
166*
180
176*
171
177
154
174
179
172
183
187
V7
R1
R2
R3
R5
R6
R6.3
R7
V7
R1
R2
R3
R5
R6
R6.3
R7
191
192
192
205
180
195
192
199
187
192
188
204
176
197
190
199
143*
165*
180
192
183
172*
154*
185*
146*
173*
181
192
182
172*
162*
186*
* Mean in the undrained site for a specific year ϫ row spacing ϫ developmental stage treatment is significantly less than in the drained site at
the 0.05 probability level.
Evidence that waterlogging stress occurred on the
undrained, but not the drained site is also documented
by the CGR patterns during the emergence to R5 period
(Fig. 2 and 3). Because patterns were similar between
row spacings, only CGR for the 100-cm row width is
presented. Waterlogging in the undrained vs. drained
sites showed the same adverse effect on biomass accumulation described in previous research (Nathanson et
al., 1984; Scott et al., 1989). Crop growth rate in the
drained site increased in a pattern indicative of normal
exponential growth for stress-free soybeans (Board and
Harville, 1996a). In contrast, the undrained site was
characterized by constant or slowly increasing CGR indicative of stress conditions. Because soil mineral content and pH were optimal at both sites, the most plausible explanation for these differences in CGR was
waterlogging stress. However, other soil properties may
have differed between sites and contributed to CGR differences.
In both years, waterlogging in the undrained site significantly (P Ͻ 0.05) reduced CGR during most of the
vegetative and early reproductive periods (Fig. 2 and
3). Significant (P Ͻ 0.05) waterlogging effects on LER
and NAR also occurred. Differences in LER between
sites occurred concomitantly with CGR changes, indicating that LER may have accelerated CGR; or perhaps

4. LINKEMER ET AL.: WATERLOGGING EFFECTS ON LATE-PLANTED SOYBEAN
1579
Fig. 2. Crop growth rate, net assimilation rate, and leaf expansion rate for Centennial soybean planted in 100-cm rows at a late planting date
near Baton Rouge, LA, 1993. *,**,*** Indicates that mean in drained site is significantly greater compared with the undrained site at the
0.05, 0.01, and 0.001 probability levels, respectively.
LER was stimulated by CGR. Net assimilation rate
showed striking differences between sites at the earliest
sampling dates, indicating it played an important role
in initiating CGR differences between sites. Previous
research showed that leaf photosynthetic rate was decreased by waterlogging (Oosterhuis et al., 1990). This
probably explained the NAR differences between sites.
Drainage site had significant effects (Table 2) on plot
yield (P Ͻ 0.0001), seed number (P Ͻ 0.01), seed size
(P Ͻ 0.0001), pod number (P Ͻ 0.01), and pods per
reproductive node (P Ͻ 0.001). Because year ϫ site,
row spacing ϫ site, and year ϫ site ϫ row spacing
interactions were not significant (P Ͻ 0.05) for plot yield
or the yield components affected by site (except seed
size), data were averaged across row spacings and years
in Table 3. Significant year ϫ site interaction for seed
size was created by greater waterlogging-induced seed
size reduction in 1994 compared with 1993. However,
seed size was significantly (P Ͻ 0.05) reduced in un-
drained vs. drained sites in both years. Within years,
plot yield was significantly (P Ͻ 0.05) reduced on the
undrained vs. drained sites. Across years, drainage increased plot yield by 58%. Sample yield was affected
similarly to plot yield by improved drainage. Partition
of sample yield gain on the drained vs. undrained sites
into seed number and seed size indicated that seed number contributed to 63% of the sample yield gain with
seed size contributing 37%. Greater seed number, in
turn, was entirely caused by increased pod number since
seed per pod was unaffected by site (Table 2). More
pods in the drained vs. undrained sites were related
to greater pods per reproductive node (Table 3) since
reproductive node number was not affected by sites
(Table 2).
Greenhouse Experiments
To characterize the soil conditions, redox potential
was measured beginning at the end of each transient

5. 1580
CROP SCIENCE, VOL. 38, NOVEMBER–DECEMBER 1998
Fig. 3. Crop growth rate, net assimilation rate, and leaf expansion rate for Centennial soybean planted in 100-cm rows at a late planting date
near Baton Rouge, LA, 1994. *,**,*** Indicates that mean in the drained site is significantly greater compared with the undrained site at the
0.05, 0.01, and 0.001 probability levels, respectively.
waterlogging treatment and continued to the end of the
experiment to see the change in soil conditions over
time. Because the year ϫ treatment interaction was not
significant, data for a particular treatment were averaged across years. The redox potential values for the
control were constantly in a range between 400 and 550
mV (Fig. 4 and 5). At this range oxygen is available for
the plant and for the soil microorganisms. Reduction in
soil redox potential was observed when waterlogging
was applied any time during the life cycle, but the lower
limit of the values and the speed of recovery were different for each treatment. The process of recovery of redox
potential varied with the stage of the crop at which
waterlogging was applied. When waterlogging was ap-
Table 2. ANOVA of plot yield and yield components for Centennial soybean grown on drained and undrained sites near Baton Rouge,
LA, 1993 and 1994.
Source of variation
df
Year (Y)
Site (S)
YϫS
Row space (RS)
Y ϫ RS
S ϫ RS
Y ϫ S ϫ RS
C.V. (%)
1
1
1
1
1
1
1
Plot seed
yield
Seed
no.
Seed
size
Seed
per pod
Pod
no.
Pods per
rep. node
Rep.
node no.
Total dry
matter (R7)
****
****
NS
****
NS
NS
NS
16.3
****
**
NS
NS
NS
NS
NS
12.5
****
****
****
NS
*
**
NS
3.3
NS
NS
NS
NS
**
*
NS
3.0
****
**
NS
NS
NS
NS
NS
13.4
NS
****
NS
*
NS
NS
NS
7.9
****
NS
NS
NS
*
NS
NS
8.3
****
***
NS
NS
NS
NS
NS
14.4
*, **, ***, **** Mean square is statistically significant at the 0.05, 0.01, 0.001, and 0.0001 probability levels, respectively.
NS ϭ not significant.

6. 1581
LINKEMER ET AL.: WATERLOGGING EFFECTS ON LATE-PLANTED SOYBEAN
Table 3. Plot yield, sample yield, seed size, seed number, pod number, pods per reproductive node, and total dry matter (R7) for
Centennial soybean planted on drained and undrained sites near Baton Rouge, LA, 1993 and 1994. Data are averaged across row
spacings and years.
Site
Drained
Undrained
LSD(0.05) to compare site means
Plot yield
Sample yield
Seed size
Seed no.
Pod no.
Pods per rep. node
Total dry matter (R7)
kg haϪ1
2452
1550
255
g mϪ2
295
219
38
g per 100 seed
13.43
11.88
0.24
no. mϪ2
2249
1842
263
no. mϪ2
1129
912
121
no.
2.43
1.90
0.14
g mϪ2
493
362
63
plied at the early vegetative stages (V2 and V3) redox
was slower to recover than when waterlogging was applied at V7 (Fig. 4). When waterlogging was applied at
reproductive stages, soil redox potential stayed below
300 mV, the level at which free oxygen disappears
(Fig. 5).
The sensitivity of the soybean cultivar Centennial to
waterlogging at different stages is shown in Table 4.
When transient waterlogging stresses were applied at
V2 and V3 a significant reduction in height of the plants
was observed. Except at R1, transient waterlogging did
not affect plant height when applied during the reproductive period. Branch number was significantly reduced when transient waterlogging was applied at V2,
V7, and R1. Reproductive nodes per branch did not
differ between treatments (data not shown). The number of pod per reproductive node was significantly (P Ͻ
0.05) affected by transient waterlogging at R1 and R3.
Seed size was significantly (P Ͻ 0.05) reduced when
transient waterlogging was applied at R3 and R5;
whereas no waterlogging treatment significantly (P Ͻ
0.05) reduced seed per pod below control levels.
The effect of waterlogging on the yield components
is expressed cumulatively in the total yield per experimental unit. No reduction in yield was observed when
waterlogging was applied at V3, V7, R6, or R6.3. On
the other hand, waterlogging significantly reduced the
yield when applied at V2, R1, R3, and R5. Greatest
yield depression occurred with waterlogging at R3,
whereas the least yield reduction was at V2. Intermediate yield losses occurred when stress was applied at R1
and R5. Yield component mechanisms for yield loss
varied with timing of the stress. Stress at V2 caused
yield loss through reduced seed and pod number (since
seed size and seed per pod were similar) which was
associated with a significant reduction in branches per
plant. Reduced branches per plant and pod per reproductive node contributed to the yield loss created by
waterlogging at R1. Both reduced seed size and pod per
reproductive node accounted for much of the yield loss
Fig. 4. Soil redox potential of Centennial soybean for temporary (A)
and continuous (B) waterlogging commencing during different
stages of the vegetative period. The first measurements for temporary or continuous treatments commencing at a given developmental stage were made at the end of the 7-d transient-stress period.
* Indicates that treatment mean is significantly less than control
at the 0.05 probability level.
Fig. 5. Soil redox potential of Centennial soybean for temporary (A)
and continuous (B) waterlogging commencing during different
stages of the reproductive period. The first measurements for temporary or continuous treatments commencing at a given developmental stage were made at the end of the 7-d transient-stress period.
* Indicates that treatment mean is significantly less than the control
at the 0.05 probability level.

7. 1582
CROP SCIENCE, VOL. 38, NOVEMBER–DECEMBER 1998
Table 4. Yield components for transient waterlogging stresses and yield for transient and continuous stresses for Centennial soybean
grown in a greenhouse study. Stresses were initiated at the specific developmental stage and averaged across 1993 and 1994.
Devel.
stage
Plant
height
Branch
no.
Pods per
reprod. node
Seed per
pod
Seed size
Yield for 7-d
transient stress
Yield for
continuous stress
Control
V2
V3
V7
R1
R3
R5
R6
R6.3
cm per plant
65.5ab*
40.8h
48.1fg
59.2bcde
51.9ef
70.8a
64.0ab
69.5ab
64.0ab
no. per plant
4.3a
3.0bcd
4.3a
3.5bcd
2.8d
3.7abc
4.0ab
3.6ab
4.4a
no.
4.3a
4.5a
4.5a
4.3a
2.4b
0.6d
5.0a
4.3a
4.8a
no.
2.0abcd
1.9abc
2.3a
2.0abc
1.9abcd
1.3bcd
1.3bcd
2.0ab
2.0abc
g per 100 seed
13.0a
12.0abc
11.0abcd
12.0abc
12.0abc
5.0g
6.0efg
10.0abcde
16.0abcd
g per plant
32.9A
23.4B
29.8AB
26.9AB
11.8CD
2.3E
11.0CD
25.2AB
27.1AB
g per plant
32.9A
1.3F
0.8F
10.2CD
9.1CDE
6.8DEF
15.2C
24.4B
* Means not followed by the same lower-case letter are significantly different at the 0.05 probability level. Yields for transient and continuous stresses
commencing at the same developmental stage not followed by the same capital letter are significantly different at the 0.05 probability level.
for the R3 treatment, whereas stress at R5 was entirely
due to reduced seed size.
Comparison of effects of continuous vs. transient waterlogging can also aid in identifying sensitive stages.
Lack of significant differences between transient and
continuous waterlogging indicates the developmental
period for the transient stress is waterlogging sensitive.
Transient stresses at R1, R3, and R5 gave similar yields
(P Ͻ 0.05) compared with continuous stresses commencing at the same time. Thus, the R1 to R5 period showed
greater waterlogging sensitivity than the vegetative period (emergence–R1) or the late reproductive period
(R5–R7). Continuous waterlogging from V2, V3, and
V7 caused a greater yield loss than transient stress at
this time, probably because the stress occurred during
the sensitive R1 to R5 period.
DISCUSSION
Previous studies indicated that soybean was more sensitive during the early reproductive than during the vegetative periods (Griffin and Saxton, 1988; Scott et al.,
1989). The current study identified some vegetative
stages as sensitive to waterlogging and demonstrated
that some reproductive stages were more sensitive than
others. Greatest sensitivity to waterlogging occurred
during a 7-d period starting at R3 (Table 4). Stress
applied at this time reduced yield by 93%. The second
most vulnerable periods for waterlogging were 7-d periods commencing at R1 or R5. Yield losses of 67% occurred for these treatments. Significant yield loss (30%)
also occurred when plants were waterlogged at V2. Soybean was tolerant to waterlogging during the rapid seed
filling period commencing near R6 and also during the
latter half of the vegetative period. Thus, the effects of
waterlogging on soybean yield are strongly dependent
on when the stress occurs. This difference in sensitivity
suggests reports of waterlogging tolerance in some other
cultivars should be interpreted with caution (Van Toai
et al., 1994). Since waterlogging can occur under field
conditions at any time, emphasis should be placed on
selecting those cultivars that are tolerant to stress at all
the sensitive periods. When late-planted soybean are
flood irrigated, caution should be used to apply the
water during a tolerant period.
In the field study, rainfall totaling over 3 to 5 cm fell
near V2 and the R1 to R5 period in both years (Fig.
1). As evidenced by CGR responses and soil oxygen
concentration, these rain events were sufficient to cause
waterlogging which resulted in a 37% yield loss. We
posited that if 3 to 5 cm of rain falling over 1 to 2 d of
a sensitive period could cause significant yield loss, then
10 cm of rain in a single week during a sensitive period
falling on a poorly drained soil would probably result
in waterlogging stress (rainfall probabilities in Muller
et al., 1989 are calculated for intervals of 1 wk or longer).
Using rainfall probabilities calculated from 1951 to 1980,
the driest part of Louisiana (northern third) would have
a 19% chance of receiving a waterlogging stress severe
enough to reduce yield of late-planted soybean. Probability of yield loss would be greatest in the southern
third of Louisiana (54%), the region of greatest rainfall;
and intermediate (24%) in the central third of Louisiana. Thus, waterlogging is an important environmental
stress affecting yield of late-planted soybean in areas of
the Gulf Coast Region having rainfall similar to southern Louisiana. Because the probability of receiving a
waterlogging stress for late-planted soybean is so high
in these areas, soybean should not be planted after 15
June unless some provision is made for drainage (e.g.,
sloping the field or ditching) or growing waterloggingtolerant cultivars.
Reduced CGR induced by waterlogging occurred
through effects on NAR and LER (Fig. 2 and 3). The
initial adverse waterlogging effect appeared to be on
NAR; as it was usually depressed prior to, or concomitant with, decreased CGR. Because LER and CGR
declined concomitantly, LER also may have played a
role in decreasing CGR in the waterlogged site. Waterlogging depressed NAR probably through effects on
leaf photosynthetic rate and conductance (Oosterhuis et
al., 1990). Data from the current study confirm previous
findings that growth dynamic factors during the vegetative period have a large effect on yield at late planting
dates (Board and Harville, 1996a).
Decreased CGR for the undrained vs. drained field
site led to decreased pods per reproductive node resulting in decreased yield (Table 3). Results differ from
previous research which showed similar pod number
between waterlogging treatments (Griffin and Saxton,
1988). Waterlogging affected plot yield mainly through
the same yield component mechanism shown by row

8. LINKEMER ET AL.: WATERLOGGING EFFECTS ON LATE-PLANTED SOYBEAN
spacing, defoliation, and shading (Board et al., 1992;
Board and Harville, 1993; Board et al., 1995). Stresses
which adversely affect CGR appear to have a consistent
mechanism for reducing soybean yield; that is, lower
pod production through control of reproductive node
number and/or pods per reproductive node. Because
both yield components are influenced by CGR during
the vegetative and/or early reproductive periods, cultural and genetic strategies to improve yield should aim
at optimizing CGR during these critical periods.
Seed size played a secondary role in explaining yield
differences. Seed size was affected more by waterlogging
in 1994 compared with 1993 probably because of greater
rainfall during R3 to R6 (Fig. 1), the period when cotyledonary cell number is determined (Egli, 1994; Peterson
et al., 1992). Reduced assimilatory capacity during this
time has been shown to decrease seed size (Egli et al.,
1989) and was probably responsible for lower seed size
in the undrained site (Table 3). Previous reports indicated that stress (nonoptimal row spacing, partial defoliation, and shade) did not alter seed size (Board et al.,
1992; Board and Harville, 1993; Board et al., 1995). Seed
size was reduced in the current study by stress, probably
because waterlogging-induced CGR reductions (R1–
R5) were greater compared to stress-induced CGR (R1–
R5) reductions in the previous studies. Thus, it appears
that when CGR is reduced about 30% during the early
reproductive period (as was the case for these previous
studies), yield loss occurs entirely through reduced pod
number; whereas with greater CGR reductions (50–60%
in the current study), seed size is also reduced.
Yield component responses to waterlogging in the
greenhouse study paralleled responses in the field study.
Depending on when the stress was applied, loss was
attributed to either reduced seed size and/or reduced
pod number, with the latter usually having greater importance. Yield reduction caused by stress at R5 was
due entirely to decreased seed size (Table 4), whereas
reduced yield due to stress at R3 was mainly caused by
fewer pods per reproductive node. Yield loss due to
stress at V2 was entirely related to lower branch number
resulting in fewer reproductive nodes (reproductive
nodes per branch were constant). Yield loss due to stress
at R1 was due to a combination of reduced branch
number and pods per reproductive node.
Thus, yield components differed in sensitivity to waterlogging. Seed per pod and reproductive nodes per
branch were unaffected by waterlogging stress. Seed
size was affected when stress was applied at R3 and R5,
but not throughout the remainder of the reproductive
period. Because seed filling periods were similar between treatments (data not shown), seed size reduction
must have occurred through reduced seed growth rate.
Egli et al. (1989) have demonstrated that seed growth
rate can be affected by cotyledonary cell number when
assimilate supply is reduced during the cell formation
period. Apparently, reduced seed size in the current
study occurred through the same mechanism. Waterlogging stress applied during the effective filling period
(period of rapid seed filling, mainly R6–R7) had no
effect on seed growth rate or seed size. Thus, the period
1583
of cotyledonary cell number formation had greater sensitivity to waterlogging compared with the rapid seed
filling period. These results agree with previous studies
showing that the rapid seed filling period is buffered
from transient stress (Westgate et al., 1989).
Pods per reproductive node was the yield component
most affected by waterlogging (Table 4), except for
stress initiated at R5. Previous yield formation studies
indicated that this yield component is very responsive
to changes in assimilatory capacity during the period in
which it is formed (R1–shortly after R5, Board and
Harville, 1993). Greatest waterlogging sensitivity at R3
can thus be explained because stress at this time occurred when two yield components were being formed
(pods per reproductive node and cotyledonary cell number), which are strongly influenced by canopy assimilatory capacity. Branch number, which is associated with
branch dry matter (Board et al., 1990), was also sensitive
to waterlogging in the current study (Table 4). As with
the yield components described above, branch number
is responsive to alterations in canopy assimilatory capacity induced by row spacing (Board et al., 1990), shade
(Board et al., 1995), and defoliation (Board and Tan,
1995). Because branch number is largely determined
between R1 and R5 (Board and Settimi, 1986), waterlogging stress at R1 sufficiently reduced assimilatory
capacity enough to reduce branch number to where it
made a major contribution to yield loss. Reduced branch
number caused by stress at V2 probably resulted from
reductions in plant height (Table 4) and a lower number
of main stem nodes from which branches could be produced (Board, 1985).
Results demonstrate that pods per reproductive node
and indicators of branch development [branch number,
branch dry weight, or total branch length (Board et al.,
1990)] can be used as criteria to quantify genetic and
environmental responses to waterlogging. Such criteria
could be used to identify genotypes/cultivars resistant to
waterlogging or to assess the effects of environmentalcultural factors on waterlogging-induced phenomena in
soybean. Since branch development and pods per reproductive node respond to greater light interception during the emergence to R5 period (Board et al., 1990;
Board and Harville, 1993), cultural methods such as
increased plant population and reduced row spacing
would ameliorate the adverse effects of waterlogging
on yield of late-planted soybean. In the current study,
reduced row spacing resulted in a 37% increase in yield
when averaged across years and sites (1687–2318 kg
haϪ1, P Ͻ 0.05). However, percentage narrow-row yield
increases were greater in the undrained site (1216–1888
kg haϪ1, 55%, P Ͻ 0.05) compared with drained site
(2164–2748 kg haϪ1, 28%, P Ͻ 0.05). Thus, narrow-row
yield increases in our study were greater under adverse
growing conditions. In summary, waterlogging influenced yield through yield formation mechanisms similar
to those shown by other stresses (nonoptimal row spacing, reduced light, partial defoliation) whose main effect
is to reduce crop growth rate. It appears that regardless
of the type of stress that influences crop growth rate,
effects on yield components are similar.

9. 1584
CROP SCIENCE, VOL. 38, NOVEMBER–DECEMBER 1998
CONCLUSIONS
Waterlogging for 7 d caused greatest yield loss for
late-planted soybean when stress was applied at R3.
Significant (P Ͻ 0.05) yield loss also occurred when
stress was applied at R1, R5, and V2. All other developmental periods were unaffected by waterlogging. Thus,
waterlogging must be avoided during the early reproductive period (R1–R5) and the early vegetative period
(V2) to obtain optimal yield. Based on these findings
and rainfall probability data, waterlogging is an important stress affecting yield of late-planted soybean in
southern Louisiana and in other parts of the Gulf Coast
Region having a similar rainfall pattern. The adverse
effect of waterlogging on CGR was initiated through
effects on NAR. Reduced crop growth rate, in turn,
affected yield mainly through pod number by regulation
of pods per reproductive node and branch number; a
pattern similar to that shown by previous row spacing,
shade, and defoliation studies. Waterlogging commencing at R3 or R5 can also have adverse effects on seed
size.
ACKNOWLEDGMENTS
We thank the United States Agency for International Development, Mission for Costa Rica, for supporting Mr. Geoffrey Linkemer, and we thank Mr. Peter M. Guillot for technical
assistance. Supported in part by funding from the Louisiana
Small Grain Research and Promotion Board. We also express
gratitude to Dr. Lynn LaMotte of the Department of Experimental Statistics, Louisiana State University, for reviewing
the experimental design and statistical procedures in the
manuscript.
REFERENCES
Board, J.E. 1985. Yield components associated with soybean yield
reductions at nonoptimal planting dates. Agron. J. 77:135–140
Board, J.E., and B.G. Harville. 1993. Soybean yield component responses to a light interception gradient during the reproductive
period. Crop Sci. 33:772–777.
Board, J.E., and B.G. Harville. 1996a. Growth dynamics during the
vegetative period affects yield of narrow-row, late-planted soybean.
Agron. J. 88:567–572.
Board, J.E., and B.G. Harville. 1996b. Strategies for increasing yields
of late-planted soybean. Louisiana Agric. 39(2):22–25.
Board, J.E., B.G. Harville, and A.M. Saxton. 1990. Branch dry weight
in relation to yield increases in narrow-row soybean. Agron. J.
82:540–544.
Board, J.E., M. Kamal, and B.G. Harville. 1992. Temporal importance
of greater light interception to increased yield in narrow-row soybean. Agron. J. 84:575–579.
Board, J.E., and Q. Tan. 1995. Assimilatory capacity effects on soybean yield components and on pod number. Crop Sci. 35:846–851.
Board, J.E., and J.R. Settimi. 1986. Photoperiod effect before and
after flowering on branch development in determinate soybean.
Agron. J. 78:995–1002.
Board, J.E., A.T. Wier, and D.J. Boethel. 1995. Source strength influence on soybean yield formation during early and late reproductive
development. Crop Sci. 35:1104–1110.
Boyer, J.S. 1982. Plant productivity and environment. Science 218:
443–448.
Dowdell, R.J., R.A. Smith, R. Crees, and S.W.F. Restall. 1972. Field
studies of ethylene in the soil atmosphere—equipment and preliminary results. Soil Biol. Biochem. 4:1472–1480.
Egli, D.B. 1994. Seed growth and development. p. 127–148. In K.J.
Boote et al. (ed.) Physiology and determination of crop yield. ASA,
Madison, WI.
Egli, D.B., E.L. Ramseur, Y. Zhen-wen, and C.H. Sullivan. 1989.
Source-sink alterations affect the number of cells in soybean cotyledons. Crop Sci. 29:732–735.
Fehr, W.R., and C.E. Caviness. 1977. Stages of soybean development.
Iowa Agric. Exp. Stn. Spec. Rep. 80.
Grable, A.R. 1966. Soil aeration and plant growth. Adv. Agron. 18:
57–106.
Griffin, J.L., and A.M. Saxton. 1988. Response of solid-seeded soybean to flood irrigation. II. Flood duration. Agron. J. 80:885–888
Hanson, W.D., and E.J. Kamprath. 1979. Selection of aluminum tolerance in soybeans based on seedling-root growth. Agron. J. 71:
581–586.
Heatherly, L.G., and H.C. Pringle III. 1991. Soybean cultivars’ response to flood irrigation of clay soil. Agron. J. 83:231–236.
Hunt, R. 1978. Plant growth analysis. Inst. Biol. Stud. Biol., 96. Camelot Press, Southampton, UK.
Hunt, R., and E.T. Parsons. 1981. Plant growth analysis. In Users
instructions for the stepwise and spline programs. Unit of comparative ecology. Univ. Of Sheffield, Sheffield, UK.
McIntosh, M.S. 1983. Analysis of combined experiments. Agron. J.
75:153–155.
Mengel, D.B., and E.J. Kamprath. 1978. Effect of soil pH and liming
on growth and nodulation of soybeans in histosols. Agron. J.
70:959–963.
Muchow, R.C., T.R. Sinclair, J.M. Bennett, and L.C. Hammond. 1986.
Response of leaf growth, leaf nitrogen, and stomatal conductance
to water deficits during vegetative growth of field-grown soybean.
Crop Sci. 26:1190–1195.
Muller, R.A., J.D. McLaughlin, and G.J. McCabe. 1989. Probabilities
and climatic patterns of precipitation in Louisiana 1951–1980. Louisiana Agric. Exp. Stn. Bull. no. 799.
Musgrave, M.E. 1994. Waterlogging effects on yield and photosynthesis in wheat varieties grown in Louisiana. Crop Sci. 34:1314–1318.
Musgrave, M.E., and N. Ding. 1998. Evaluating wheat cultivars for
waterlogging tolerance. Crop Sci. 38:90–97.
Musgrave, M.E., A.G. Hopkins, Jr., and C.J. Daugherty. 1991. Oxygen
insensitivity of photosynthesis by waterlogged Apios americana.
Environ. Exp. Bot. 31:117–124.
Nathanson, K., R.J. Lawn, P.L.M. DeJabrun, and D.E. Byth. 1984.
Growth, nodulation and nitrogen accumulation by soybean in saturated soil culture. Field Crops Res. 8:73–92.
Oosterhuis, D.M., H.D. Scott, R.E. Hampton, and S.D. Wullschleger.
1990. Physiological responses of two soybean [Glycine max (L.)
Merr] cultivars to short-term flooding. Environ. Exp. Bot. 30:85–92.
Peterson, C.M., C.O’H. Musjidis, R.R. Dute, and W.E. Westgate.
1992. A flower and pod staging system for soybean. Ann. Bot.
(London) 69:59–67.
Russell, R.S. 1977. Plant root systems: Their function and interaction
with soil. McGraw-Hill, Maidenhead, Berks., UK.
Sallam, A., and H.D. Scott. 1987. Effects of prolonged flooding on
soybeans during early vegetative growth. Soil Sci. 144:61–66.
SAS Institute. 1987. SAS/STAT guide for personal computers. 6th
ed. SAS Inst., Cary, NC.
Scott, H.D., J. De Angulo, M.B. Daniels, and L.S. Wood. 1989. Flood
duration effects on soybean growth and yield. Agron. J. 81:631–636.
Smucker, A.J.M., and R.R. Allmaras. 1993. Whole plant responses
to soil compaction. p. 727–731. In Buxton et al. (ed.) International
crop science I. CSSA, Madison, WI.
Van Toai, T.T., J.E. Beuerlein, A.F. Schmitthenner, and S.K. St.
Martin. 1994. Genetic variability for flooding tolerance in soybeans.
Crop Sci. 34:1112–1115.
Wallace, S.U., T. Whitwell, J.H. Palmer, C.E. Hood, and S.A. Hull.
1992. Growth of relay intercropped soybean. Agron. J. 84:968–973.
Westgate, M.E., J.R. Schussler, D.C. Reicosky, and M.L. Brenner.
1989. Effect of water deficits on seed development in soybean. II.
Conservation of seed growth rate. Plant Physiol. 91:980–985.