An Analysis of Cotton Production in California: A Model for Acala Cotton and the

Effects of Defoliators on Its Yields 1

A. P. GUTIERREZ; L. A. FALCON; W. LOEW,3

P. A. LEIPZIG; AND R. VAN DEN BOSCH'
ABSTRACT

A population model of cotton growth and development in response to weather,

agronomic factors, and insect pests has been successfuIly used to examine cotton
production in California. The impact of most arthropod pests on cotton yields is
currently under investigation. Only the effects of defoliating insects, Spodoptera exigua
(HUbner), Trichoplusia ni (HUbner), and Heliothis zea (Boddie), are reported here.
The impact of moderate defoliation by S. exigua and T. ni populations causes only
slight yield reductions. More important is the predation by S. exigua larvae on imma-
ture fruit (Le., squares). Investigations on the interaction of plant age and develop-
mental stage with defoliator damage (defoliation and predation on squares) indicates
that the greatest yield reductions occur from attack early in the squaring period. The
late season effects of H. zea are easy to measure and incorporate into the model.

Considerable controversy exists in California con- phenological events, and more flexible structure for
cerning pest control needs in cotton. The widespread enabling easier coupling of insect pest subroutines.
use of pesticides has introduced new problems (e.g., The major modifications and field validation of the
secondary pest outbreaks) and further obscured model, plus an examination of defoliation are pre-
solutions for existing ones (Falcon 1972, Smith and sented here. The model is used to examine cotton
Falcon 1973). Much of the controversy exists be- growth as influenced by weather and agronomic
cause analytical methods have not been available for factors (irrigation, nitrogen application, and timing
assessing the effects of climate, agronomic practices, of planting). Insect attack is examined with regards
and various pests on cotton yields. to timing and severity. A flow diagram for the cot-
Baker and Hesketh (1969), Hesketh et al. (1971) tOn model is presented in Fig. 1, as an aid to under-
and Baker et al. (1972) examined photosynthate standing the daily sequence of events being modelled.
production and respiratory losses in commercial up- A copy of SIMCOT UC may be obtained from the
land cotton, and formed the basis for much of the first author as it is not possible to discuss all of the
cotton modeling effort. Duncan (1971), Duncan details of the model in a paper of reasonable length.
et al. (1971) Stapleton and Meyers (1971), Wilson
et al. (1972), Stapleton et al. (1973), and McKinion California Cotton
et al. (1974) have written deterministic simulation Cotton is but one of a mosaic of crops grown in
models for cotton growth and development in re- the San Joaquin Valley. Acala, an indeterminate
sponse to weather and various agronomic factors. upland variety of cotton, is the only variety grown
A descriptive deterministic simulation model for the in the San Joaquin Valley of California. Cotton
growth and development of a single plant (SIMCOT fields are usuaIly plowed and flood irrigated (in some
II, McKinion et al. 1974) has been used with vary- cases sprinkler systems are used) during mid-winter.
ing success for estimating the effects of weather and The crop is planted from March to late May, de-
agronomic factors on the timing and severity of pending on the wetness of the season. No further
fruit shed, plant morphogenesis, and dry matter water is applied until symptoms of light moisture
production of upland cotton in the southeastern stress occur (slight leaf wilting) usuaIly mid-June
United States. A copy of SIMCOT II was made on the heavier soils. Nitrogen may be applied in
available during January 1973 and used to simulate irrigation water or by air. While planting densities
observed 1972 field data from Corcoran, Calif. The vary considerably, our cooperating farmers seed at
computer results were grossly at odds with field ob- the rate of 40 Ib/ acre which produces a population
servations. of 40-60,000 plants/ acre. The effective bloom period
A population model using the photosynthesis and occurs from early July through mid-August. The
nitrogen use subroutines from SIMCOT II was crop continues to produce new fruit until night tem-
developed (which we shaIl distinguish as SIMCOT peratures fall below the thermal threshold and/ or
UC) to accurately simulate cotton growth in Cali- soil moisture and nitrogen are depleted (see Methods
fornia. The major modifications concern its change section). Chemical defoliation of the crop nor-
to a population model, the incorporation of a more maIly occurs during October with harvest commenc-
accurate evapotranspiration routine, new criteria ing 2 wk later. Some growers apply paraquat during
for end of season, plant morphogenesis, timing of defoliation to open immature boIls and enhance
1 Received for publication 16 July 1974. Supported by NSF
yields. Fields are usually harvested twice at 2-wk
grant no. NSF-GB-34718. intervals. The simulation begins when the fields are
2 Dept. of Entomology, Univ. Calif., Davis.
3 Dept. of Entomology, Univ. Calif., Berkeley. first irrigated and runs until harvest.
125
126 ENVIRONMENTAL ENTOMOLOGY Vol. 4, no. 1

attacks very young fruit, and is the key pest of cotton

in the San Joaquin Valley (Falcon et aI. 1971).
Pesticide recommendations are made for the control
of this pest despite the fact that no definite evidence
has been published as to the extent of their economic
impact. The problem of assessing impact is com-
plicated by the fact that (1) Lygus populations are
difficult to estimate accurately (Sevacherian and
Stern 1972) and (2) little or no physical evidence
remains, proving that fruit loss is directly attributed
~ to Lygus injury. This has generated considerable
DAY LENGTIl
NET PHOTOSYNTHESIS
controversy in California as to the economic im-
PHOTOSYNTIIA TE NEED
PHOTOSYNTIIATE USED
portance of this pest. The resolution of this contro-
NITROGEN USE versy is central to the establishment of sound pest
management practices in cotton. Exceedingly low
Lygus populations occurred in the study plots dur-
ing 1972 and 1973, and for this reason their effect
is not included in this analysis.
Bollworm is predominantly an insecticide induced
pest which normally occurs during late season (Dr.
D. Gonzales, personal communication). While boll-
worm damage is easier to estimate than that caused
by Lygus, modeling of the nuances of attack inten-
sity and behavior are more difficult (Hartstack 1974,'
Stinner et al. 1974). Hence, the effects of boll de-
pletion based on bollworm moth population esti-
FIG. I.-A flow diagram of the daily pattern of cotton
mates have been hard to estimate. The model is
growth and development. used to examine the effects of observed timing and
quantity of bollworm larvae on lint yields.

Pests of Cotton in the San Joaquin Valley Other Causes of Fruit Shed
In addition to insect-caused fruit shedding, losses
The insect pests of cotton can be classified by the may also be a result of moisture, nitrogen, carbo-
principal type of damage they cause (defoliators and hydrate, and high intensity solar radiation stress
those attacking fruit) .
(Grimes et a1. 1969, McKinion et a1. 1974, Jones
Defoliators et al. 1974).
The principal defoliators are the beet armyworm
Methods
(BAW), Spodoptera exigua (HUbner), and the cab-
bage looper (CL), Trichoplusia ni (HUbner). The Plant growth and development data were collected
beet armyworm, besides being a defoliator, also in- during 1973 specifically for use in formulating and
jures the plant terminals causing developmental de- verifying the model. Data obtained from untreated
lays (Eveleens 1972)', but insignificant yield reduc- cotton during 1972 were used as validation criteria.
tions (at least at high planting densities). This spe- The cotton was grown in Tulare clay soil of the
cies also attacks squares (flower buds) causing fruit Tulare Lake Basin near Corcoran, Calif."
shed (Eveleens 1972). Falcon et aI. (1968), Falcon 1973 Data
et aI. (1971), Eveleens et al. (1974), and Ehler et
Crop development data (square, flower, boll, and
aI. (1974) have shown that pesticide applications
mainstem node counts), plant morphogenesis (phen-
suppress natural enemies and greatly increase the
ology and growth rate of various plant parts), insect
abundance of these pests. Only the effects of noc-
damage (defoliation, fruit and stem attack), insect
tuid populations on crop development and yield are
oviposition site preference as affected by plant den-
examined in this paper.
sity were collected. The experiment was composed
Pests A ttacking Fruit of 4 blocks, each consisting of 5 randomly assigned
The mirid, Lygus hesperus Knight, and bollworm, spacing treatments (2.2, 4.4, 4.4," 8.8,' and 13.8
Heliothis zea (Boddie), are the principal pests of plants/meter-row) thinned by hand after germina-
cotton fruiting parts in the San Joaquin Valley. The tion. Data obtained from the several treatments were
pink bollworm, Pectinophora gossypiella Saunders, used to estimate density related parameters for the
is found only in the desert valleys of southeastern model, with the 13.8 plants/meter-row treatment
California and was not included in this study. Lygus
• Hartstack, A. w. 1974. Model of HeUothis zea. USDA-
ARS Annual Report. Cotton Insects Res. Lab., College Station,
• Eveleens, K. G. 1972. Impact of insecticide application on Texas.
natural biological control of the beet army worm in cotton. Ph.D. "The J. G. Boswell Co., Corcoran, Calif., provided the ex-
thesis, Dept. of Entomology and Parasitology, Univ. Calif., Berke- perimental plots.
ley. 138 pp. 7 Two seeds left per drill hole.
February 1975 GUTIERREZ ET AL.: COTTON PRODUCTION MODEL 127

providing validation criteria. Each replicate con-

sisted of four 183-m North-South rows (101.6 cm
between row centers) of thinned plants (multiples
of 22.9 cm spacings). Sampling began at the south
-
Z
tI)
25

end of a replicate at the beginning of the experiment ~ 20

and moved northward as the season progressed. Four OIl
Q)
healthy plants were pulled in a systematic manner "'tJ
from each treatment replicate at 1- to 2-day inter- 0 IS
c::
vals. Plant parts were mapped as to number, age,
position, and damage. The plants were dissected by
branch (leaves, stem, and fruit), main stem, and
roots; then oven dried and weighed to determine dry
matter accumulation. Similarly, special samples were
-E
Q)
OIl

c::
10
°
13·82plants/meIer.
4·4 plants/meIer.
2·2 plan'y'm.,'er.

0
made throughout the season to estimate the effects ~
of BAW damage on plant development; principally
via terminal damage. In addition, developing leaves
and fruit parts at specific positions were tagged and 0200 1000 2000 2600

sampled throughout their developmental period to

more precisely determine their weight and age. The Physiological time.
numbers and ages of insect species found on plant 2.-Main stem node production as a function of
FIG.
parts were also recorded. General abundance of physiological time (DO) and planting density.
Lygus was determined from sweep net counts.
60 miles away and because of generally clear skies
1972 Data
little discrepency is expected. Agronomic data on
These data were collected as part of an ongoing planting dates, irrigation, nitrogen applications, re-
pest management study. The timings of insecticide sidual soil nitrogen, organic matter in the soil, soil
sprays were made at different lunar phases (quarters) type, and other agronomic and cultural data were
with the aim of inducing different levels of worm provided by the grower.
populations (Falcon, unpublished data). The ex- A water balance accounting model developed by
periment consisted of an untreated check (a) and Ritchie (1972) was programmed and incorporated
4 lunar treatments (B=full, C=last quarter, D=new, in SIMCOT DC. Ritchie's model apportions the
E=first quarter. Data obtained from treatment E evapotranspiration potential due to climate into (1)
were similar to those from B and only B is reported that which evaporates from the soil surface; and
here. Plant growth data per meter row were taken at (2) that which transpires from the foliage of a crop
approximately weekly intervals (c.f. Falcon 1972). with changing leaf area index. This model also
The meter-row data on plant part counts are similar enables evaluation of the effects of soil type on
to that described above for 1973, but plant mapping water availability to the crop.
and dryweight data were not taken. Some discrep-
ancies in criteria for determining categories of fruit- Physiological Time
ing parts existed, but these caused little difficulty once Physiological time (DO) is used in SIMCOT DCB
recognized. The numbers and size of BAWand CL as the timing mechanism. Physiological time cal-
larvae were recorded biweekly on 32 meter-row of culated in degrees F after the met.hod of Gilbert and
cotton plants. The larvae were beaten from meter- Gutierrez (1973) can be converted to DOc by a con-
row samples of plants onto a sheet of cloth where stant (0.555). The thermal threshold (53.5°F,
they were counted by size. These data are used to 11.9°C) for cotton development was determined
assess the effects of defoliating insects on yields. from data presented by Hesketh et al. (1972) .
Developmental times for various stages of growth
Larval Food Consumption
(e.g., between mainstem nodes and fruit) were
Laboratory experiments were conducted on age- estimated from field data.
dependent dry matter consumption by BAW larvae.
Newly hatched larvae were placed on cotton plants Results
growing at 21°C and the developing larval popula- Figures 2 and 3 summarize plant developmental
tion sampled throughout their developmental period. data for 1973. Similar but less precise results were
Live and dry weights were estimated for all age lar- obtained during 1972. The effects of crowding on
vae. Consumption rates for cabbage looper larvae plant development are also shown.
are assumed to be similar. Mainstem Node Production
Weather and Agronomic Data Davidson (1973) found that mainstem node pro-
Maximum and minimum temperatures, rainfall, duction appeared to be a linear function of Julian
and relative humidity records were obtained from a time, while Wilson et al. (1972) found that the tim-
hygrothermograph situated at the test plot site. ing of plant development in cotton could be esti-
Measurements of daily solar radiation were obtained mated quite accurately in models using physiological
from the D. S. Weather Bureau Station at Fresno ca. time. Figure 2 shows that mainstem node (MSN)
128 ENVIRONMENTAL ENTOMOLOGY Vol. 4, no. 1

••
125
2·Ymeter. current squares It
•• total squares •
• total stied 0

100 •

i 9

~.
75
10 4+eter.

• •
• 13.+ter.
30
•
• •
•• .---.
•••
•
50

..
:3
~ 20

-
25

1000 1500 2000 2500 3000' 1000 1500 2000 2500

Physiologica I time.
1000 1500 2000 2500
FIG. 3.-The pattern of cumulative fruit point production (.), current square population (x), and cumulative
fruit point shed (0) per plant through time at three planting densities (2.2, 4.4, and 13.82 plants/meter-row).

production is a linear function of cumulative fruiting branch is critical, as serious error in this
DO>53.5°F and density (equation 1), until mid sum- regard would alter all subsequent events in the
mer (July 19) at which model. Figure 4 indicates that the position of the
MSN = 1 + (0.01300 - 0.000112 1st fruiting branch (FFB) can be predicted as a
Density) DDMSN, (1) function of density (meter-row). Both 1972 and
1973 data estimated the same result, and hence
time the rate of production was markedly reduced equation (2) is used in SIMCOT DC to predict the
when night temperatures fell below the develop- time delay (the number of mainstem nodes (FEB) X
mental threshold during this period (July 19, 1973). internode DO requirements (90°0» after planting
Density is expressed as plants per meter row, and
DDMSN equals the accumulated daily DO since FFB = 6.2475 + 0.1072 • Density (2)
germination. Downton and Slayter (1972) found before the 1st fruit are produced. A 50 = D° delay
in phytotron experiments that none or reduced
growth occurred when night temperatures were near
the thermal threshold. This agrees with several sets
12
of our field data. More generally, determinant types to-
of cotton are commonly observed to go through ~IO
three growth phases: Vegetative, fruiting and fruit
maturation. In our model, the rate of plant part ~ 8 ."
(/)
production is altered by carbohydrate (=Photosyn- a: 6
thate) stress (F = photosynthate demand by plant IL.

parts/ availability, O~F~ 1). The stress may be ~4

induced by various factors (e.g., low night tempera-
tures), and usually occurs when a heavy boll loao ~
o
2
z
is present. Conversely, the rate of growth increases
o J • .
when the stress is removed (e.g., when the crop
matures), and provided conditions are suitable.
5 10 15 20 25
PLANTS/METER ROW
Fruit Production
4.-The effects of planting density on the position
FIG.
Predicting the timing of appearance of the 1st of the 1st fruiting branch.
February ] 975 GUTIERREZ ET AL.: COTTON PRODUCTION MODEL ]29

SQUARE BOLL rates. The criteria for senescence of root and stem
~-----.----------- -, tissue and the causes of fruit shed are similar to
FLOWER max size those estimated by McKinion et al. (1974).
~ The relationship between dry matter accumulation
~
600
•• 800
I in associated leaves and bolls (same position on a
branch) is shown in Fig. 6. Bolls begin their expo-
nential phase of growth when the associated leaf has
reached its maximum size. This point coincides with

Ix
o
'] ~ES the time the boll is no longer susceptable to abscis-
sion. Figure 7 shows individual boll weight (BW)
as a function of DO (Equation (6». Figure 8 de-
20000- picts the seasonal trend in dry matter production
for fruit,

Ix
I.D
o
ROOTS and STEMS BW = e-··"'8 + .0000DO
from DO> 700 to 2000
leaves, stems, and roots in the normal spacing (13.8
plantsl meter). Leaf, stem and root dry matter ac-
cumulation slowed July 19, ]973 because of stresses
(6)

7000-
«53.5°F). The apparent abrupt cessation of leaf
FIG. 5.-A diagrammatic representation of the devel- dry matter production shown in Fig. 8 is because
opmental times for stem, root, leaf, and fruit tissues. leaves-but not stems-abscise due to old age.
Hence, the death rate is equal to or greater than the
production rate of leaves. This reduced rate of
in fruitpoint production is made after the ] st fruit- growth continues until night temperatures are con-
ing branch because field workers cannot usually sistently below 53YF (the physiological end of
count very small fruiting parts. season occurs when the seven day average ~ 15D°).
Figure 3 depicts the per plant seasonal trend for This value accurately predicted the cessation of
cumulative fruit production and shed for 3 planting growth for ]967, '68, '72, and '73 data. Fruit dry
densities during ]973. Each point represents the matter stops accumulating at this time. This is in
mean of 4 plants. In each case the patterns are the contrast to ] 972 where the same effect was not ex-
same; only the magnitudes of the curves differs. perienced until mid September. Equations 7, 8, and
Equation 3 describes the production rate of fruit (as 9 describe potential leaf, stem, and root growth
a function of DO and planting density) until carbo- respectively as a function of time and density.
hydrate stress occurs at which time the rate is
reduced by F.
AMTLEF = 0.02]6 [e-O.••.•••7 + 0."'''] .<it• fl (Density)
(7)
FPINC = [e-O•1","_.106>
Den'IIY]
dt,
(where Mis 20Do in this case) (3)
AMTSTM = 0.0228 [e-O.'"8 + 0.09111] .<it• f2 (Density)
(8)
Age Structures and Dry Matter Accumulation
The model keeps track of time, and surviving
AMTRTS = 0.0027 [e- 108+• 0.01111] .<it• f3 (Density)
a.

(9)
tissues may be summarized according to age (e.g.,
equation [4]). (where t = DO/25.4, .<it= tll+1 - til' and density is
600 the number of plants per meter/row. If t>41, leaf
Squares = ~ age(i) (where the interval growth occurs at a rate of AMTLEF (41), and stem
i=] and root growth are proportional to stem and root
weights)
] to 600 is defined as the square stage) (4 )
As in McKinion, et al. (1974) actual growth incre-
Figure 5 depicts required developmental time for ments and priorities are determined by the ratio of
stems, roots, leaves, and fruiting points. The Ix fac- current photosynthate availability to total demand
tor is used for scaling the age dependent photosyn- by all plant parts. Daily available photosynthesis is
thetic effectiveness of leaves. This is the only arbi- calculated by subroutines derived from SIMCOT II.
trary factor in the model. Dry matter for any par-
ticular plant part is tallied according to equation (5). Moisture Requirements
This ability to keep track of plant material by age Water use data (Et = cumulative evapotrans-
is important in coupling insect pest subroutines, piration) by cotton were supplied by Dr. D. Grimes
M (University of California Research and Extension
Dry matter = 2; weight(i) Center, Parlier) and are plotted on a physiological
i=1 time scale (Fig. 9). These data indicate that water
use by the crops is also affected by low temperature.
(where M is the oldest age group) (5) The difference in the inflection points of the two
as all insect species exhibit age preference (= site curves is attributed to differing planting densities,
preference) and have age dependent consumption while the slopes of the curves during the linear phase
130 ENVIRONMENTAL ENTOMOLOGY Vol. 4, no. 1

1·25
e
e
e

-
e
e e e e
1·0
lit
~ e
e •
< e e

-
GI: e
o ·75
e e e e e

'.·~J.
e e
+-
-C
e-
m large boll.
el) ·5
~
-.J small bolls.
x = white flower.
·25 f lower stage.

large squares.
o
o 2 3 4 5 6 7 8

Fruit weight. (GRAMS.)

FIG. 6.-A plot of leaf weight on fruit weight from the 10th branch, 1st fruiting position at corresponding
periods of their development.

of water use are basically the same. The rate of and complete than the 1972 data, but lack adequate
crop water use declined dramatically at different estimates of worm numbers (small sample size).
times (2700DO vs. 3400DO), and was associated with Nonetheless, both data sets are satisfactory for
the onset of persistent low night temperatures which comparison with model simulation, provided the
occurred late in the season. data discrepancies summarized in the Methods sec-
tion are taken into account in the life table sum-
Validation of the Model
mary. The 1972 data contained the greatest amount
The 1973 plant growth data are more accurate of information on defoliator populations and were
used to estimate their affect.
1973-Without Insects
t The timing of events in the model (e.g., the first
-'":(
:(
.0 0
c
~
!~
square) was within 2 days of that observed in the
field. The pattern of mainstem node production was
< entirely accurate. The model without insects pre-
'"
.E. .~
dicts 2.41 bales of cotton, while the observed yield
t-=
J: s
>i
0
E
was 1.98 bales (21 % of total boll weight). Field
'0 counts of 311,000 mature bolls compared unfavor-
~ ~
" .!!
W
l ably with that predicted by the model 351,040. BAW
~ and alfalfa looper (Autographa cali/ornica (Speyer»
••• larvae causing severe defoliation occurred in large
0 number at the beginning of the season and probably
0 SOD 1000 1500 2000
account for most of the discrepancy (see next sec-
Physiological time. tion). In addition, bollworm caused a 4.2 % loss of
FIG. 7.-Dry matter accumulation by bolls through large bolls which must be included in these results.
time. Data from 1972 as well as laboratory studies on dry
February 1975 GUTIERREZ ET AL.: COTTON PRODUCTION MODEL 131

300
•
•• total .
stems.
250 leaves.
•••••••• "
• fruit .
•
i4( • roots. •
at: 200
C>
•••••••••
CIt•
•
•

~
C
-a.0 150

~ao
m
.- v

-I 50

.. -. •" -"••• t
•
••• • ••
1000 2500
~o
Physiological time.
FIG. B.-Dry matter accumulation by plant parts through time. The dark arrows and dates indicate when
changes in plant physiology occurred.

matter consumption rates by noctuid larvae are used age class of leaves. Field data indicate that eggs
to examine the interaction of plant growth and insect are seldom deposited on very young leaves or very
injury. Further examination of the 1973 data is old leaves, hence in the model leaves of age 840 to
deferred until insect damage is included. 1500·D are assumed to be attacked equaHy. If the
The Effects of Beet Armyworm and amount of leaf tissue required by the worm popu-
Cabbage Loopers on Cotton Yields lation is not present in this age bracket, the age leaf
bracket is enlarged in both directions until the de-
Beet armyworm and cabbage looper larvae are mand is met or the leaf tissue is exhausted. Leaves
present in cotton from early May until late summer escape attack only if they age beyond the preferred
and mid to late summer respectively, until night stage and are not required to meet additional popu-
temperatures became too cool for reproductive
lation demands.
activities (approximate night maximums lower than
AIter the eggs hatch, the larvae begin to consume
50·F). As with cotton, the life cycle of these in-
sects can be examined using physiological time, leaves at a rate which is age and time dependent.
hence the model for defoliators is merely a sub- Soohoo and Fraenkel (1966) determined that a
routine (Drymat) of the collon model. species of Prodenia converted ca. 10-30% (depend-
Figure 10 indicates stratum (main stem node ing on the host) of the dry matter they consumed to
level) where oviposition by the adult moths is dry weight body tissue. The dry weight of larvae as
occurring through time. The preference appears to determined in laboratory experiments is ca. 25% of
be for a stratum of the plant, rather than a specific wet weight, hence the wet weight of larvae is a
 132 ENVIRONMENTAL ENTOMOLOGY Vol. 4, no. 1

• 1967 MTA
lL 1968 MTA
80

70

••• 60 •
r:.:1
1&1
50
>
i= 40
~
~
:;:) 30
~
U 20

10

0 400 800 1200 1600 2000 2400 2800 :i200 3600

PHYSIOLOGICAL TIME
9.-Cumulative evapotranspiration (Ed of soil moisture in inches from a developing cotton crop during
FIG.
1967 and 1968 at Shafter, Calif. Data supplied hy Dr. D. Grimes.

reasonable approximation of the amount of leaf dry As the purpose of this section is to understand the
weight required to achieve a particular size. Equa- impact of larval feedings on cotton production, not
tion lOis used to calculate the leaf dry weight the population dynamics of worm populations, the
(LDW) required by the larval population. model is supplied an observed pattern of larval field
counts. Missing observations were interpolated from
M
observed biweekly counts. The larvae were classed
LDW = ~ N(i)' F(i) Llt
as small (~~") and large (> ~ "). These categories
i=1 (10)
correspond approximately to instars I-III and IV-
(N is the number of larvae in an age category i, VII respectively.
F is rate of food consumption by the ith aged larvae Figure 11 depicts beet armyworm and cabbage
during the daily accumulated .1t and M is the age of looper worm phenology and activity. The effects of
pupal inception when feeding stops) timing of pesticide applications on inducing pest
F(i) is calculated according to equation 11: outbreak are obvious (see also Ehler et at. 1974 and
F(i) = [F()e " + k2DO,] • k2 (11) Eveleens et al. 1974). Observed treatment yields
show insignificant differences (Table I), but perhaps
(where k1 and k2 are empirical constants, and DOl indicate a trend. The cumulative defoliation (grams)
is the mean age of the ith stage) for each treatment is shown in Fig. 12. Computer
results indicate that larval feeding alone has little
effect on yield as all treatments simulated yielded

---
......;. 25 approximately 2.8 bales of cotton for 1972. The
Z Total main Uem nodeL lack of observed differences in the simulations would
en indicate that either the model is incorrect, grossly
-
~ 2.

. '.
insensitive or that larval feeding has a greater impact
than feedbacks due merely to dry matter loss. Stimu-
- ~._~.
.,
'
......••
r • •
.. .• .- - - -.
._\-,1 '-' -- - .;- lation of plant growth from light insect damage, as
well as retardation from heavier damage is well
"....... Highe5t node of defoliation. known. Davidson (1973)8 demonstrated both effects
.' in cotton. He further showed that leaf punching,
simulating larval feeding, caused a greater reduction
in yield than simply removing an equivalent amount
15ClO 2000 2500
of leaf tissue at the petiole. The model can accom-
'000 3000

Physiological time. • Davidson, A. 1973. Computerized plant growth analysis

FIG. 10.-The observed highest node where insect de- of-the interactions of arthropod pests and other factors with the
cotton plant. Ph.D. thesis, Dept. of Entomological Sciences.
foliation has occurred through time. Univ. Calif., Berkeley. 192 pp.
 February 1975 GUTIERREZ ET AL.: COTTON PRODUCTION MODEL 133

~j 'A
100
Hewn '''100 A

~ ..- -
•..
oft

.!- 'DO

I
..•..
w
!! B vi Q

N ~ Ql v
M
~oo ::: Ql 'DO ~
0 ! E ~
•..
> N
M
..sz ~ ~
E•.. a ro __ I_Ill

0 ~ ___ rv-'111 } INSTAlS

~ ~
E
>. -
_ - _
I-III}
IV-VI INS'''U •..
Ql
<200 0.._
0
Qj ...•0
C
~
-0
•..
.B
'00

.B
.J:I
Ql
01 'DO

~
d
Z
E
:J
- 0
•..
Ql
200 .J:I •••
E
:J 0
Z
IDO
lDO

~
/ /' \
V
...• -..• ~
~
-
~
- ~-- -~~--,~~ /
/- ...•...•.• - /
/

J J A J J A
FIG. ll.-Observed phenologies and patterns of beet armyworm and cabbage looper larvae in four chemical
application experiments during 1972 at Corcoran, Calif. The large arrows indicate when insecticides were applied.

modate the negative effects of leaf loss, but not the age squares which results in additional square shed-
slight stimulation of a small amount of defoliation. ding. In all treatments, a square mortality factor
Eveleens (1972),' Davidson (1973),· and this (IL)
study have observed that beet armyworm larvae dam- III
(IL) = B' ~ BAWl' where B equal 0.2
i=I (12)
~

1· ~'4

",a..
G
• 8
v C
A Table I.-Observed versus simulated cotton yields for
the different pesticide treatments at Corcoran, California
during 1972 (see Fig. 11). A reference bale is 479 pounds
8.. '3
• D
of Iint.a

-
.!
~
·2

Observed
Simula-
tion with
BAW
and CL
Boll-
worm
Com-
plete
~ yield effects damage simula-
'"0 ·1 Treatment (bales) (bales) (bales) tion

R A untreated
.2 check 2.78 2.77 0.0004 2.766
10 20 30 10 20 30 10 B 2.74 2.75 .019 2.731
1 1
C 2.618 2.67 .014 2.656
JULY. AUGUST. SEPTEMBER.
D 2.626 2.74 .062 2.678
FIG. 12.-Cumulative dry matter consumed by the
noctuid larval populations depicted in Fig. II. • Only replicates planted on the same day were lIsed.
134 ENVIRONMENTAL ENTOMOLOGY Vol. 4, no. 1

~ per plant X plants per acre X % wt. of boll

oc: which is actual lint].
The observed discrepancies are small and the
...
E
yields for the various treatments are in approxi-
.! mately the right order. The final results (Table 1)
show that the observed worm populations caused
"o Q)
C)
insignificant damage (0.162 bales maximum, treat-
ment C).
E • 1 planVdrill hole. If the same pattern of larvae, as observed in treat-
o
Q 20
o 2 plantydrill hole. ment C, are moved forwards or backwards in time
.~ by lO-day increments, it is possible to examine the
effects of the same defoliation (BA Wand CL) and
~ o BAW attack on squares during different stages of
I 5 m B plant development (Table 2). The greatest effect
on yield occurs when defoliators (BAW principally)
Plenty meter. attack during the early fruiting period; the most
important effect being square depletion and only
secondly direct defoliation. Defoliator activity late
Fro. 13.-The observed effects of beet armyworm
damage on plant terminals as affected by planting in the season appears to have a rather minimal
density. effect.
The estimates of larval numbers experienced early
in the season during 1973 were inadequate, but they
proportional to the number of larvae in instars I-III did provide a good fix on the phenology of worm
(equation 12) was applied each Ilt to all age squares. infestation. Hence the first larval peak observed in
The value of B was estimated from 1973 field data, treatment C-1972 and the observed 1973 worm
and is equal to 0.02. The expression phenology were used to examine the 1973 yield dis-
III crepancy. Only 43% of the defoliation observed in
2; BAWl C and the observed 1973 bollworm damage were re-
i=I quired to account for the yield reduction. The com-
puted number of green plus mature bolls (318,000)
is the summation of all beet armyworm larvae in
compared favorably with observed values (311,000),
instars I-III. Larger larvae are normally not found
while the simulated lint yield was 2.02 bales versus
inside fruit bracts, but rather on the leaves. When 1.98 bales observed.
I" was incorporated into the model, the simulated
The 0.43 bale difference between the 1.98 bales
results for A, B, C, but not D compared very favor-
observed and 2.41 simulated occurs when insects
ably with observed yields (Table 1). were not included. A 0.1 bale reduction due to
Bollworm damage to bolls (lint reduction) occurs
bollworm was observed, and the 0.3 bales were
late in the season and was estimated from samples
probably lost to defoliators (e.g. as in Table 2). In
and subtracted from the computed yields. It is addition, the general pattern of fruit developed was
assumed that the average weight of attacked mature highly reasonable and the dry matter production of
bolls is ca. 8 g. Yield reduction due to H. zea was leaves, stems, and roots predicted by the model was
estimated from counts made on 80 plants per treat-
within 5% of that observed in the field.
ment. The yield reduction per acre caused by H. Two observations also merit discussion here. First,
zea was calculated according to the following beet armyworm attacks on plant terminals cause
equation: delays in plant development. As a result fewer
lint loss = [wt. boll X number bolls attacked nodes are produced and the number and age struc-

Table 2.- The effect of timing of defoliator activity 00 cotton yield. Observed defoliator. populations from treat-
ment C have been moved at ten day intervals ahead (+) or back (-) ••

Bolls
Yields Total
Treatment (Bales) Mature Green bolls

C+ 40 (early square period) 1.92 63.7 14.7 78.4

C+ 30 1.91 62.4 15.6 78.0
C+ 20 2.28 76.7 14.7 91.4
C+ 10 2.68 94.5 12.1 106.6
C (late square period 2.68 94.0 10.1 104.1
C- 20 (early boll period) 2.83 99.4 9.9 109.3

• (+)
(-) = later
= earlier in the season than
in the season.
the observed phenology for treatment C (Fig. 11 and 12).
February 1975 GUTIERREZ ET AL.: COTTON PRODUCTION MODEL 135

25

CORCORAN, CALIF.
o
10.24.13 .
...:
c: • 2 plantydrill hole.
C
- Q.
15

-o
~ 10

..c
c: 5
C 22·2%

~
o 202. 4.4 4.4'l.4' 13082.

Open. Closed. Damaged.

FIG. 14.-The number of open, closed and bollworm damaged bolls observed in five density plantings during
1973 at Corcoran. Calif. The numbers at the bottom of the histograms for open bolls identify them as to plants/
meter row.

ture of fruiting parts may be greatly influenced. interaction of crop system components. Accurate
Figure 13 shows the relationship between planting prediction of crop development is hampered by the
density and percent terminals damaged. Low plant- inability to predict climate.
ing densities suffer a greater degree of damage. This model shows that yields are limited by the
During 1973, plants with damaged terminals yielded plants' ability to provide nutrients for all of its
fewer large green bolls at the end of the season developing parts. The amount of nutrient which the
(28% at the lowest density). The higher percent- plant can provide is greatly influenced by weather
age of terminals damaged at low densities is offset (solar radiation, moisture, temperature, etc.), avail-
by the greater number of green bolls. The plant is able inorganic nutrients, competition between plants,
less able to compensate for damage at high densities. pestiferous arthropods and its genetic potential.
Second, bollworm damage appears to' be much Acala SJ-l is highly sensitive to all these factors.
greater in low density plantings than in high (Fig. However, if manageable factors (water, fertilizer)
14). This observation is extremely important, be- are kept optimum, it is possible to observe how this
cause the crop cannot compensate for boll damage variety responds to weather and its pests.
occurring late in the season, and cotton tends to the Fruit are retained until nutrient supply is out-
same yield per unit area over a wide range of plant- stripped by demand, at which time considerable
ing densities in the absence of insect pests. Planting shedding occurs. Fruit production continues until
densities greater than 8 plants per meter row appear carbohydrate stress occurs (O~F~l). Leaf, stem and
more suitable for reducing bollworm losses. root growth are slowed considerably at this time
and the plant switches to a boll maturation phase.
Discussion In tropical areas, new fruit may be produced after
Cotton plant growth and development is influ- the bolls mature (stress removed), while in temperate
enced by a number of factors, many of which are areas growth continues until frequent low night tem-
not amenable to management (e.g., weather). The peratures stop all growth (e.g., California). The
function of the model is to help understand the length of the fruiting phase determines the potential
136 ENVIRONMENTAL
ENTOMOLOGY Vol. 4, no. 1

number of bolls that can be produced, while the Eveleens, K. G., R. van den Bosch, and L. E. Ehler.
length of the maturation phase and the favourabiIity 1974. Secondary outbreak induction of beet army-
of weather determine how much dry matter the worm by experimental insecticide application in cot-
bolls accumulate. During the early season, before ton in California. Ibid., 2: 497-503.
any significant nutrient demand is made by maturing Falcon, L. A., R. van den Bosch, C. A. Ferris, L K.
Strombert, L K. Etzel, R. E. Stinner, and T. F.
fruit there is little effect from low temperatures Leigh. 1968. A comparison of season-long cotton
above freezing. pest-control programs in California during 1966.
Insect pests serve as an additional factor decreas- J. Econ. Entom. 61: 633-42.
ing yields by attacking the plant parts. The instan- Falcon, L A. 1972. Integrated control of cotton pests
taneous manufacture and distribution of photosyn- in the far west. Pages 30-2 in Beltwide Cotton
thate is commonly described by the expression, dW / Prod. Res. Conf. Proceedings. Memphis.
dt = P-RW, where W is dry weight, R the respira- Falcon, L. A., R. van den Bosch, J. Gallagher, and A.
tion rate per unit dry weight, P is the rate of photo- Davidson. 1971. Investigation of the pest status
synthesis and t is time. The effects of insects add of Lygus hesperus in cotton in central California. J.
Econ. 64: 56-61.
new dimensions and can be summarized in the ex-
Gilbert, N., and A. P. Gutierrez. 1973. A plant-aphid-
pression dW/dt = P-RW-C+I, where C is the parasite relationship. J. Anim. Eco!' 42: 323-340.
photosynthetic or previous growth dry matter con- Grimes, D. W., H. Yamada, and W. L. Dickens. 1969.
sumed directly and I (+ or -) is the photosynthetic Functions for cotton (Gossypium hirsutum L.) pro-
dry matter accrued from growth stimulation (+ at duction from irrigation and nitrogen fertilization
low level) or depression due to wound healing (- variables. I. Yield and evapotranspiration. Agron.
for high rates of plant injury). Observed patterns J. 61: 769-73.
of defoliation indicate that dry matter loss (C + I) Hesketh, J. D., D. N. Baker, and W. G. Duncan. 1971.
results in minimal yield reductions. More important Simulation of growth and yield on cotton: respira-
is the predation effects of BAW larvae on squares. tion and the carbon balance. Crop Sci. II: 394-8.
1971. II. Simulation of growth and yield in cotton-
In this case, the earlier the timing of BAW attack
environmental control of morphogenesis. Ibid., 12:
during the squaring period, the more severe the 436-9.
effect. Thus, predation on squares which have a Jones, J. W., A. C. Thompson, and J. D. Hesketh. 1974.
high probability of maturing (those set early) have Analysis of Simcot: Nitrogen and growth. Pages
the greatest effect on yield during a short season. 111-6 ill Beltwide Cotton Prod. Res. Conf. Proc.
This may not be as important during a longer sea- Memphis.
son, because the plant has some ability to compensate. McKinion, J. M., J. W. Jones, and J. D. Hesketh. 1974.
Analysis of Simcot: photosynthesis and growth.
Acknowledgment Pages 117-24 in Beltwide Cotton Prod. Res. Conf.
Special thanks are due Dr. C. M. Merritt, Division Proc. Memphis.
Ritchie, J. T. 1971. Model for predicting evaporation
of Biological Control, Berkeley, for her kind help.
from a row crop with incomplete cover. Water
The advice of Dr. D. N. Baker was invaluable. Resour. Res. 8: 1204-13.
Sevacherian, V., and V. M. Stern. 1971. Sequential
REFERENCES CITED
sampling plan for lygus bugs in California cotton
Baker, D. N., and J. D. Hesketh. 1969. Respiration fields. Environ. Entomol. I: 704-9.
and carbon balance in cotton (Gossypium hirsutum Smith, R. F., and L. A. Falcon. 1973. Insect control
L.). Pages 60-4 in 1969 Proc. Beltwide Cotton for cotton in California. Cotton Grow. Rev. 50:
Prod. Res. Conf. Jan. 7-8, 1969, New Orleans. 15-27.
Baker, D. N., J. D. Hesketh, and W. G. Duncan. 1972. Soohoo, C. F., and G. Fraenkel. 1966. The consump-
Simulation of growth and yield in cotton. Crop Sci. tion, digestion and utilization of food plants by a
II: 431-5. polyphagous insect, Prodenia eridamia (Cramer).
Downton, J., and R. O. Slayter. 1972. Temperature J. Insect Physiol. 12: 711-30.
dependence of photosynthesis in cotton. Plant Stapleton, H. N., D. R. Buxton, F. L. Watson, D. L.
Physio\. 50: 518-22. Nolting, and D. N. Baker. 1973. Cotton: a com-
Duncan, W. G. 1971. Simcot, a simulator of cotton puter simulation of cotton growth. Tech. Bull. Ariz.
growth and yield. Pages 115-8 in Proc. of work- Agric. Exp. Stn.
shop on tree growth dynamics and modeling. Duke Stapleton, H. N., and R. P. Meyers. 1971. Modeling
Univ., Oct. I1-12, 1971. subsystems for cotton-the cotton plant simulation.
Duncan, W. G., D. N. Baker, and J. D. Hesketh. 1971. Trans. ASAE 14: 950-3.
Simulation of growth and yield in cotton. III. A Stinner, R. E., R. L. Rabb, and J. R. Bradley. 1974.
computer analysis of the nutritional theory. Page Population dynamics of He/iolhis zea (Boddie) and
78 in Proc. 1971 BeItwide Cotton Prod. Res. Conf. H. virescells (F.) in North Carolina: a simulation
Ehler, L. E., K. G. Eveleens, and R. van den Bosch. model. Environ. Entomol. 3: 163-8.
1974. An evaluation of some natural enemies of Wilson, A. G., R. D. Hughes, and N. Gilbert. 1972.
cabbage looper in cotton in California. Environ. The response of cotton to pest attack. Bull. Ento-
Entom.2: 1009-15. mol. Res. 61: 405-14.