Soybean Stress from Defoliating Pests:
I. D. Teare, J. E. Funderburk, and L. G. Higley
North Florida Res. and Educ. Ctr., Univ. of Fla., Quincy,, Fla. 32351; Univ. of
Nebraska, Lincoln, NE 68583. Research NF-93-4.
Defoliation by insect pests is a major stress of soybean in the Southern
Region. Quantifying physiological responses of soybean to defoliation is
essential for understanding how insect defoliation reduces soybean yields.
Indeed, defoliators comprise the most abundant and diverse guild of insects that
attack soybean in the U.S. (Turnipseed and Kogan 1976). Understanding the
relationship between defoliation and plant responses, such as yield is essential
for better pest management.
Researchers have long recognized the importance of defoliating pests to
soybean, and over the past 40 years more than 50 research articles have addressed
soybean defoliation (Ostlie 1984). However, despite this volume of research,
long standing questions regarding defoliation on soybean persists. Often results
from different studies do not agree, possibly because of differences in
environment or methodology. For example, hail simulation studies, with injury
imposed on a single day, clearly do not simulate injury by defoliating insects
(Ostlie 1984). But even where actual insects or appropriate simulation
techniques are used, reported relationships between defoliation by insects and
yield vary greatly (Ostlie 1984). Nor has soybean compensatory adjustments
elucidated the understanding of insect defoliation of soybean and related soybean
Previous research often examined percent defoliation, however, 50%
defoliation of a large soybean canopy is unlikely to produce the same response
as 50% defoliation of a small canopy. We believe the light interception
hypothesis (that the principal physiological effect of defoliation is to reduce
light interception by soybean canopies) may account for much of the variability
between defoliation and yield losses reported in the literature.
Among the significant scientific benefits of this research is establishing
a physiological model for soybean defoliation that also may be applicable to
other crops. Compensatory responses to defoliation are not well documented or
understood (particularly photosynthetic responses),
The practical consequences of this situation are that calculated economic
injury levels (EILS) and economic thresholds (ETs) for soybean pests differ
widely among insects and across states. It seems unlikely that
defoliation/soybean response relationships are really as variable as existing
thresholds suggest. Rather, these thresholds probably reflect our present
uncertainty in how defoliation physiologically impacts soybean. Consequently,
to some degree our existing thresholds are suspect. Moreover, the lack of more
definitive understandings of how soybean respond to defoliation impedes the
implementation of more advanced decision tools, such as multiple species EILs
(Hutchins et al. 1988, Pedigo et al. 1989). This latter limitation is especially
important in the south where soybean defoliators usually occur as a complex of
The objectives of this study are: 1) To characterize effects of simulated
insect defoliation on soybean physiology, 2) To determine how yield responses to
defoliation can be explained by reductions in canopy light interception (through
reductions in leaf area below the critical leaf area index), 3) To determine how
response to defoliation at different reproductive stages differ, how responses
to sequential defoliation at two stages differ from defoliation at a single
stage, and how responses to simulated insect defoliation (through time) differ
from responses to defoliation on one day, 4) To characterize mechanisms of
soybean compensation to defoliation, particularly delayed leaf senescence and
altered leaf photosynthesis, 5) To compare responses at different sites (states)
and responses to determinate and indeterminate soybean growth habits), 6) To
describe defoliation/yield loss relationships for calculating single and multiple
species EILs for defoliating insect pests of soybean.
MATERIALS AND METHODS
Soybean were grown on a Norfolk loamy sand (fine, loamy, siliceous thermic
typic Kandidult) with a pH of 5.5 and 1.5 percent organic matter in 1990 and 1991
at the North Fla. Res. and Educ. Ctr. at Quincy, FL.
The experimental area was bottom plowed 3 May 1990 and fertilized at a rate
of 15# N/acre, 45# P/acre and 90# K/acre on 20 May 1990. Weed control consisted
of a preplant treatment of Treflan 4E at 0.75 lb Ai/acre on 17 May 1990,
postemergence treatment of Classic at 0.5 oz Ai/acre plus Tackle at 0.5 lb
Ai/acre on 25 June 1990, and a cultivation with a rolling cultivator on 2 July
1990. The experimental area for 30 May 1991 was bottom plowed and fertilized
with 5-10-15 at 500#/acre. Weed control consisted of Treflan at 2 pts/acre and
Sencor at 3/8 AI/A in a preplant treatment on 31 May 1991.
cone planter was used both years to plant Braxton soybean in eight row plots 7
meters long and with 30 cm between rows at a rate of 26 seeds per row-m on 11
June 2 1990. Row orientation was north-south. Final soybean plant density was
18 plants/row-m in 1990 and 25 plants/row-m in 1991 (soybean were over-planted
and thinned to provide uniform plant spacing). Soybean cultivar used was Braxton
in southern states.
Insecticides used were Dimilin at 0.5 oz Ai/acre plus Dipel at 1 Ib/acre
on 9 Sept 90; Dimilin at 0.5 oz Ai/acre plus Assini at 0.025 oz Ai/acre on 16 Oct
90. The insecticides used in 1991 were Asana at .025 oz AI/A on 15 July, Asana
at .025 AI/A and Dipel 1 1/4# /A on 30 July, 8 oz Asana/A and 1# Orthene/A on 23
Aug, 8 oz Asana and 1# Orthene/A on 6 Sept, Pyrellian at 1 pt/A 19 Sept, and
finally 8 oz/A Asana and 1 pt/A Pyrellian 3 Oct.
The experimental design for soil treatments and for yield was a randomized
complete block containing four replications. Yield estimates each year were
determined for all 11 defoliation treatments. The effects of defoliation
treatments on soybean yield, light interception and photosynthesis were evaluated
by using ANOVA, subsequent treatment comparisons, and regression.
Treatments consist of 4 defoliation patterns (simulated insect defoliation
at stage R2, R4, R2R4, three levels within each pattern (defoliation to produce
a leaf area index of 3.5, 2.5, and 1.5 at late R4), and an undefoliated check.
Defoliation levels were chosen based on the measured leaf area index (LAI) at
late Rl/early R2 and the projected LAI at R4 based on the R4 measurement.
Defoliation levels were chosen to provide final LAIs after defoliation (R4) that
include values above, at, and below the critical LAI (the LAI value at which 95%
of incident light' is intercepted by a plant canopy, estimated to be ca. 3.5 for
soybean). Consequently, specific levels of defoliation may vary between
locations, but all locations will provide defoliation treatments that span the
critical LAI. Soybean was defoliated by leaflet, and leaf areas of all leaflets
removed were measured.
Defoliation was be limited to the upper two thirds of the canopy, but
extended into the lower canopy where necessary for high defoliation levels. The
center 4 row-mm of the two middle rows were defoliated. Undefoliated plots
received comparable handling (walking in plots and "fondling" plants) as
defoliated plots (to allow for compaction during defoliation and affects of
touching plants). Areas adjacent to the defoliated region (border rows and ends
of plots) were sham defoliated (stripping leaflets without quantifying area
removed) to approximately the same level as the defoliated area. Border areas
were sham defoliated during the last 8 days of the simulation (when most of the
defoliation is occurring), specifically on days 8 and 12. Sham defoliation
occurred on 17 Aug and 7 Sept in 1990, on 9 Aug and 30 Aug 1991. Treatments were
color coded with wood lath and flags to minimize potential errors in leaf
Target LAI's (3.5, 2.5, and 1.5) were chosen to provide discrete reductions
in light interception based a critical LAI of ca. 3.5. The undefoliated check
provided a treatment above the critical LAI. Because the defoliation levels
depend on these target LAIs at late R41, an estimate of the R4 LAI was needed.
The LAI at R4 was likely to be 1-2 units greater than the LAI at late Rl/early
R2 (when defoliation is initiated for certain treatments); therefore, the LAI at
R4 was approximated by the measured LAI at late Rl/early R2 + 2. Treatments were
imposed by removing a given amount of leaf area (leaflets) based on the
difference in projected LAIs and the target LAIs. For the R2+4 sequential
defoliation treatments, the total leaf area removed was calculated and half
removed at R2 and half removed at R4 (so that total leaf area removed is the same
as in the R2 and R4 treatments). Although, the amount of leaf area removed
differed between states, the resulting treatments (LAI levels and corresponding
levels of light interception) were comparable. The percent defoliation per
treatment to achieve the target LAI was calculated from the formula:
[projected LAI target LAI)/projected LAI]*100
Defoliation was imposed to approximate insect injury. Ideally, daily
injury rates and duration should be based on temperature driven consumption and
development models. However, because temperature driven models would result in
substantial differences in injury rates among locations, standard durations and
injury rates were employed at all sites. Many soybean defoliators have larval
development times for the latter developmental stages (when >90% of consumption
occurs) of approximately two weeks, at temperatures commonly occurring in mid to
late summer. Consequently, defoliation was imposed over 12 days (Monday of week
1 to Friday of week 2). Defoliation occurred at soybean stages R2 and R4
(depending on treatment) which corresponds with the injury phenology for many
Daily defoliation rates depend on stage specific consumption rates. In
brief, to simulate insect feeding we need to estimate what proportion of the
total defoliation, required should occur on each day. The rationale behind the
values chosen was as follows. For this study, two aspects of development and
consumption are pertinent. First, proportion of total larval consumption in a
stage, and second, duration of developmental time in a stage. To determine the
proportion of the total defoliation that should occur in each larval stage, an
estimate of proportion of total consumption by stage was needed. Published data
on this question indicate that the proportion of total consumption by instars
are: GCW 1-2=2%, 3-4=8%, 5-6=90% (Hammond et al., 1979b); SBL 1-2=1%, 3-4=9%, 5-
6=90% (Boldt et al., 1975); and VBC 1-2=3%, 3-4=5%, 5-6=92% (Boldt et al., 1975).
Because so little defoliation occurs in the first two larval stages (<3%), for
this study we will consider defoliation only during the latter stages.
Specifically, we estimated the proportion of defoliation by stage as 3-4=10%, and
5-6=90%. The second question was duration of development time in a stage.
Literature data on green cloverworm (GCW), corn earworm (CEW), soybean looper
(SBL), and velvetbean caterpillar (VBC) were used to determine appropriate values
for this study. The proportion of time spent in various instars are: GCW 1-
2=29%, 3-4=26%, 5-6=45% (Hammond et al., 1975); and CEW 1-2=23%, 3-4=25%, 5-6=52%
(Boldt et al., 1975). Based on these values, an appropriate estimate of time
spent in each stage is 1-2=25%, 3-4=25%, and 5-6=50%. We estimated development
through stages 3-6 as requiring 12 days. Therefore, the ratio of development
times (25%:50% or 1:2) gave the number of days spent in each stage; specifically,
stages 3-4=4 days and stages 5-6=8 days. Consequently, to provide an appropriate
simulation of a lepidopteran defoliator of soybean (combining consumption and
development data), we imposed injury over 12 days with 2.5% of the total
defoliation occurring on each of the first 4 days and 11.25% occurring on each
of the last 8 days.
All calculations of total leaf area to be removed (% of total leaf area to
be removed each day per plot; conversion of leaf area to be removed to leaflets
to be removed; and defoliation summaries were provided by a computer program,
DEFOL (written by L. G. Higley for this project). To adjust for possible
discrepancies between projected and actual leaf area removed, all leaf area
removed/plot/day were quantified and entered into the program to allow for daily
adjustments. The program outputted leaf areas and numbers of leaflets to be
removed from each plot on each day. Leaf area removal was based on target
defoliation levels, appropriate injury rate, and previously removed leaf area.
Leaflets to be removed were calculated from leaf area to be removed and a user-
supplied estimate of average leaflet size on the first day. Subsequently, the
program calculates the average leaflet size based on number of leaflets removed
and measured leaf areas. Because the defoliation levels were based on projected
LAIs at R4, it is important to have an idea of actual LAIs during defoliation so
that adjustments can be made if the projections are greatly in error. Measures
of plant leaf area were taken immediately before the defoliation period, which
provide a measure of the actual LAIs. (To convert a mean plant leaf area into
an LAI for 76 cm rows and 25 plants/row-m, multiply the plant leaf area [in cm2)
by 0.00329). Records of total leaf area actually removed (by plot) were
maintained to calculate actual defoliation at end of the defoliation period.
Light interception was measured in the plant canopy weekly from R3 to R6
to include measures at each reproductive stage. A line quantum sensor, 76 cm
long, was centered across the row to obtain a measure of photosynthetically
active radiation (PAR) will be obtained. Measurements were made for each of the
two center rows of each plot. Additionally, a measurement was made outside the
plots, in full sun, for each block, to indicate PAR with 0% light interception.
All measurements were taken within one hour of solar noon.
Individual plant samples (for growth analysis) were taken at ca. weekly
intervals at R2, R3, R4, and R6. For each sample date 3 plants/plot were
removed, using a stratified random sampling procedure. Plants were bagged,
labeled by treatment/block, and returned to the laboratory for measurements.
Immediately prior to defoliation at R2, plants were randomly selected from any
of the 1.5 m areas on either end of the two center rows. On all other dates
plants will be selected randomly from any of the 1 m sections at the end the 4
m defoliated regions of the two center rows. Once a section was sampled
additional samples were not taken from the same region, and the center 2 M of the
2 center rows were not sampled for growth analysis. The R3 sample were taken
immediately after R2 defoliation and the R5 sample immediately after the R4
defoliation. At harvest, individual plant samples (3 plants/plot) were taken
from the center 2 m of the middle 2 rows. After individual plant samples were
obtained, all remaining plants in the defoliated, 4 m, middle two rows were
harvested to provide plot yield. The appropriate sampling pattern is indicated
below (note that sections B and C are the defoliated areas):
Individual Plant Growth Samples R2 section A; R3, R4, R5, R6 section B1-B4
Individual Plant Yield Samples Section C; Plot Yield section B & C
-- A (1.5 m) -:- B (1 m) --:-- C (2 m) --:-- B2 (1 m) --:-- A (1.5 m) --
A (1.5 m) --:-- B3 (1 m) -:- C (2 m) --:-- B4 (1 m) --:-- A 1.5 m ) -
Variables measured for growth analysis were: height (measure from
cotyledonary node), vegetative stage, reproductive stage, branches, nodes, lowest
leaf-bearing node (cotyledonary node=1, unifoliate node = 2, etc), leaves,
flowers, pods, leaf area, leaf dry weight, support (stem and petiole) dry weight,
and pod dry weight.
Variables measured for yield analysis (individual plant yield) were: 0
seeded pods, 1 seeded pods, 2 seeded pods, 3 seeded pods, 4 seeded pods, pod dry
weight (with seeds), seed dry weight, and support (stem) dry weight.
Variables measured for plot yield were yield and percent moisture.
Additional data were agronomic practices herbicide treatments,
tillage, fungicides; soil factors soil type, soil pH, % organic matter; weather
data daily maximum and minimum temperatures, daily rainfall; and important
dates planting date, emergence date (80% emergence), canopy closure, sampling
dates for growth, light, individual yield, and plot yield.
In specific states (Arkansas and Florida in the south, Iowa and Nebraska
in the north) photosynthesis measurements were taken weekly, from R2, to examine
the soybean compensation to defoliation through altered leaf photosynthesis.
Leaflets at ca. nodes 6, 9, and 12 were marked, and photosynthetic rates
monitored before, during, and after defoliation. Leaflets on at least two plants
per plot were measured. Measurements were made in full sunlight at comparable
times for each measurement. Measurments were taken at each R-stage, R2 thru R6
10 Aug, 16 Aug, 24 Aug, 5 Sept in 1990 and 2 Aug, 7 Aug, 20 Aug, 5 Sept, 12 sept,
in 1991 respectively. In 1991 the photosynthesis measurements were greatly
hindered by excessive rainfall and cloudy weather.
Analysis included calculation of variables for classical growth analysis
and for yield component analysis. Statistical procedures used will include
analysis of variance and regression techniques. All data were subjected to
analysis of variance. When the F test was significant, multiple range tests were
RESULTS AND DISCUSSION
RAINFALL AND DROUGHT
Weather must be discussed in relation to defoliation results for 1990 at
Quincy, FL. The summer of 1990 was one of the driest on record. The average
rainfall for the soybean growing of the previous 10 year period was 26 inches
compared to 12 inches for 1990. Soybean were definitely stressed during this
period, particularly from the R4 to R6 stage (Fig. 1).
The 1991 rainfall for the soybean growing season was 28 inches. Rainfall
distribution was heavy in the early part of the season, but was dry during the
R5 and R6 reproductive stage (Fig. 1).
The quantitative effect of drought on soybean in relation to defoliation
period is shown as a function of 1990 stomatal resistance (RS) in Fig. 2).
Stomatal resistances were statistically significant for defoliation treatments
(F=2.57; df=7,21; P<0.05). Stomatal resistances (RS) were significantly
different in 1991. Defoliation at the R2 stage resulted in RS differences at the
R5 sampling date and defoliation at the R4 stage resulted in RS differences at
the R4 and R5 stage.
SEED, POD, AND STEM WEIGHT vs DEFOLIATION TREATMENT
Seed weight in 1990 was highly correlated with pod weight (R2 = 0.997) in
relation to defoliation treatment. Defoliation intensity is ranked in relation
to 1990 seed weight in Fig. 3 to order the integrated effect of defoliation
treatments [defoliation intensity = () LAI, soybean physiological stage,
duration] for that year. One check was not significantly different from R2-3.5,
R4-3.5, R2R4-3.5 as would be expected if LAI 3.5 is truly the critical LAI. The
1991 defoliation intensity needs to be reordered in relation to 1991 seed weight.
The 1991 data seems to follow original hypothesis and 1990 and 1991
illustrations need to be reordered in relation to 1991. Note the 1991 R2
defoliation intensities order themselves correctly and are the least affected by
Ranking the LAI intensity [LAI intensity = () LAI, soybean physiological
stage, and duration] in relation to seed yield for each soybean physiological
stage where defoliation occurred should indicate the severity of defoliation
treatments (R2, R4, R2R4) to soybean yield. R2 ranking of 1990 LAI intensity was
3.5=2.5=1.5, R4 ranking was 2.5=3.5>1.5, and R2R4 is 3.5>2.5=1.5; where > and =
shows significance and nonsignificance, respectively, at the 5% level of
Seed weight for 1990 was less highly correlated with stem weight (R2=0.702)
at the more severe defoliation intensities (based on ranking in relation to seed
weight). The stem weights of the three most severe defoliation intensities (R4-
1.5, R2R4-1.5, R2-1.5) increased in contrast to a reduction in seed weight (data
NUMBER 0, 1, 2, 3 SEEDED PODS/PLANT VS DEFOLIATION TREATMENT
Number of one, two and three seeded pods/plant for 1990 and 1991 are shown
in Fig. 4. In 1990 and 1991, total number of seeds/plant was most closely
related to number of two seeded pods. Total number of seeds/plant was less
affected by one seeded pods/plant and three seeded pods/plant. The number of
pods with zero seed (Fig. 5) are of less interest. The controls in 1990 had
significantly more pods with zero seed than all the defoliation treatments, which
seems to indicate that limited defoliation to the critical LAI promotes pod
abortion in stressed soybean. The 1991 need to be reordered the same as in Fig.
3,. but R4-2.5, R4-1.5, and R2R4-1.5 had significantly more 0-seeded pods.
PROPORTION INTERCEPTED PHOTO-ACTIVE-RADIATION
Defoliation treatments reduced the PIPAR absorbed by the soybean canopy
(Fig. 6). Even with drought between R4 and R6, soybean plants increased leaf
area after defoliation between R4 and R5 and increased PIPAR by canopy.
Defoliation treatment in 1991 for R2-R4 and R4 defoliation reduced PIRAR
after defoliation at the R5 and R6 stage.
PHOTOSYNTHESIS vs. DEFOLIATION TREATMENT
Defoliation in 1990 at R2 seemed to increase photosynthesis (Fig. 7, top).
Photosynthesis was greatest for 1.5>2.5>3.5
photosynthesis of defoliation treatments below the controls (Fig. 3, bottom) on
Julian day 236. For that day, photosynthesis was greatest for the
control>2.5>3.5=1.5. Following drought at Julian day 247, photosynthesis for R4
defoliation was greatest for 1.5>3.5>2.5>control. Photosynthesis for R2
defoliation on Julian day 247 was similar, 1.5>2.5=3.5>control.
Photosynthesis in 1991 decreased with time when defoliated at the R2 stage
because of the high leaf area over time that covered the preselected
photosynthesis leaves. The R4 defoliation had some reduction in photosynthesis
Our thanks to E. Brown, Agric. Tech. IV; North Fla. Res. and Educ. Ctr.,
Univ. of Fla., Quincy, FL; for data collection, computer processing, and data
Boldt, P. E., K. D. Biever, and C. M. Ignoffo. 1975. Lepidopteran pests of
soybean: consumption of soybean foliage and pods and development time. J.
Econ. Entomol. 68:480-482.
Hammond, R. B., L. P. Pedigo, and F. L. Poston. 1979a. Green cloverworm leaf
consumption on greenhouse and field soybean leaves and development of a
leaf-consumption model. J. Econ. Entomol. 72:714-717.
Hammond, R. B., F. L. Poston, and L. P. Pedigo. 1979b. Growth of the green
cloverworm and a thermal-unit system for development. Environ. Entomol.
Hutchins, S. H., L. G. Higley, and L. P. Pedigo. 1988. Injury-equivalency as
a basis for developing multiple-species economic injury levels. J. Econ.
Ingram, K. T., D. C. Herzog, K. J. Boote, J. W. Jones, and C. S. Barfield.
1981. Effects of defoliating pests on soybean canopy CO, exchange and
reproductive growth. Crop Sci. 21:961-968.
Johnson, K. B. 1987. Defoliation, disease, and growth: a reply. Phytopathol.
Ostlie, K. R. 1984. Soybean transpiration, vegetative morphology, and yield
requirements following simulated and actual insect defoliation. Ph.D.
dissertation. Iowa State University, Ames, IA.
Pedigo, L. P., L. G. Higley, and P. M. Davis. 1989. Concepts and advances in
economic thresholds for soybean entomology. p. 1487-1493 in A. J. Pascale
(ed.) Proc. World Soybean Res. Conf. IV. Vol. III.
Shibles, R., I. C. Anderson, and A. H. Gibson. 1975. Soybean. p. 164-165 in
L. T. Evans (ed.) Crop Physiology, some case histories. Cambridge Univ.
Turnipseed, S. G., and M. Kogan. 1976. Soybean entomology. Annu. Rev.
Waggoner, P. E., and R. D. Berger. 1987. Defoliation, disease, and growth.
4 I .