Crop Science Grow Your Career with CSSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Wilson, L. J.
Right arrow Articles by Gibb, D.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Wilson, L. J.
Right arrow Articles by Gibb, D.
Agricola
Right arrow Articles by Wilson, L. J.
Right arrow Articles by Gibb, D.
Related Collections
Right arrow Crop Growth and Development
Right arrow Cotton
Right arrow Pest Management Systems
Published in Crop Sci. 43:2125-2134 (2003).
© 2003 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA

CROP ECOLOGY, MANAGEMENT & QUALITY

How to Succeed by Doing Nothing

Cotton Compensation after Simulated Early Season Pest Damage

Lewis J. Wilson*,a, Victor O. Sadrasc, Simone C. Heimoanaa and Dallas Gibbb

a CSIRO Division of Plant Industry and Australian Cotton Cooperative Research Centre, Narrabri, NSW, Australia, 2390
b NSW Agriculture and Australian Cotton Cooperative Research Centre Cotton Research Unit, Locked Bag 59, Narrabri, NSW, Australia, 2390
c CSIRO Land and Water, Private Bag No. 2, Glen Osmond, South Australia, 5064

* Corresponding author (lewis.wilson{at}csiro.au).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Clarifying the amount of pest damage that can be tolerated without justifying insecticidal control will be useful in reducing insecticide use and in development of integrated pest management (IPM) systems. This study investigated the ability of irrigated, high-yielding cotton (Gossypium hirsutum L.) to recover from artificially applied damage simulating that of early season insect pests. Six experiments were done across five cotton-growing seasons. Damage included defoliation, terminal damage, and flower bud removal in a range of timings, combinations, and intensities. Crop yield was unaffected by defoliation applied before first flower buds appeared, even when 100% of true leaves were removed on three occasions (Nodes 2, 4, and 6). Crop maturity was affected by sustained high levels of leaf loss with a peak delay of 10 d after 100% defoliation three times. Up to three light tip damage events or one heavy damage event had no effect on yield and only a slight effect on crop maturity (<5-d delay to harvest). Heavy early fruit loss (100% fruit removal from the first four fruiting branches) did not affect yield but caused a delay in maturity of {approx}7 d. Equations describing the relationship between damage type, intensity, and repetition and the yield and maturity of cotton were developed and used in sensitivity analysis to define tentative damage thresholds for IPM systems in cotton.

Abbreviations: DAS, days after sowing • DWT, dry weight • IPM, integrated pest management


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
COTTON IS ATTACKED by a range of insect pests, some of which are prevalent through the early portion of the growing cycle, such as tobacco thrips (Thrips tabaci Lindeman), tomato thrips [Frankliniella schultzei (Trybom)], the native budworm [Helicoverpa punctigera (Wallengren)], the cotton bollworm [H. armigera (Hübner)], the green mirid [Creontiades dilutus (Stål)], and the cotton tip-worm (Crocidosema plebejana Zeller) (Pyke and Brown, 1996). Application of insecticides to prevent damage caused by these pests (i) increases the risk of inadvertent environmental pollution, (ii) increases selection pressure for insecticide resistance both in target and nontarget pests, and (iii) often reduces the abundance of beneficials, thereby contributing to secondary pest outbreaks (Wilson et al., 1998). Notably, the pest thrips are also important predators of the eggs of twospotted spider mite (Tetranychus urticae Koch), which is a key secondary pest (Wilson et al., 1996). It is therefore critical to accurately assess if pests require control, taking into account the capacity of the plant to recover from some degree of damage.

Accurate estimates of the abundance of some pests are difficult to obtain because, for instance, the small size and fast movement of thrips and the elusive behavior of green mirids. Accurate estimates of Helicoverpa spp. can be made, but often plants are found damaged with no Helicoverpa spp. being observed. These issues make thresholds based on pest abundance alone less reliable, undermining grower confidence and encouraging use of preventative insecticide applications. Understanding the capacity of cotton to recover or compensate for early season pest damage will enable pest thresholds to be coupled with damage thresholds. This will allow in the development of improved IPM systems for cotton by providing a more rational basis for pest control decisions.

Cotton can recover from a degree of early pest damage, often without loss of yield or delay in crop maturity, a process known as compensation. Compensation in cotton has been reported following damage by thrips (Terry, 1992; Sadras and Wilson, 1998) and by Helicoverpa spp. (Brook et al., 1992a) and has been reported in range of other plant species (Trumble et al., 1993). Given the difficulties of manipulating populations of pests in field experiments, manually inflicted or simulated damage is often used to help understand the responses of plants to herbivore damage. Simulated damage is different from real pest damage, in part because simulated damage does not involve the saliva of the pests, which may affect plant responses. Nevertheless, because it can be inflicted more uniformly, it provides valuable insights into likely plant responses to damage. Importantly, manual removal of buds and leaves triggers morphological and physiological plant responses that closely mimic major changes induced by actual pest damage (Brook et al., 1992a,b,c; Sadras, 1996a, b).

We investigated the response of cotton to the types of damage similar to that most likely to be inflicted by insect pests early in the season. This includes reduced leaf area (thrips damage); death of the apical meristem, known as tipping out (thrips, mirids, tip-worm, or Helicoverpa spp.), and damage to and loss of flower buds or squares (Helicoverpa spp. or mirids). As different pests cause different degrees of terminal damage, we also considered the effects of different severities of tip damage. While a number of studies have investigated the responses of cotton to defoliation, and loss of vegetative and reproductive buds, most reports have dealt with a single type of damage (Evenson, 1969; Bishop et al., 1977; Brook et al., 1992b; Danobrega, 1993; Longer, 1993; Sadras, 1996b). Here, we emphasized a combination of damage types (defoliation, tipping out, fruit loss), as they often coincide in the field and imposed extreme levels of damage to assess their effects on yield and timing of maturity of field-grown cotton.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Six experiments were done across five cotton-growing seasons (Exp. 1, 1994-1995; Exp. 2 and 3, 1996-1997; Exp. 4, 1997-1998; Exp. 5, 1999-2000; and Exp. 6, 2000-2001) to allow for different damage combinations and for differences between seasons. Experiments 1, 2, 4, 5, and 6 were conducted at the Australian Cotton Research Institute (30°13' S, 149°47' E), 25 km west of Narrabri in New South Wales. Experiment 3 was done on a commercial cotton property, Redmill, {approx}120 km northeast of Narrabri, NSW.

All experiments were sown into 1-m beds in early October and grown under conditions of nonlimiting N (120–150 kg N ha-1 applied presowing) or water inputs (furrow irrigated at a soil water deficit of {approx}60 mm below field capacity), in line with commercial practices. To prevent damage from pests, which could confound manually applied damage, systemic insecticides were applied at sowing with the seed [phorate (O,O-diethyl S-ethylmercaptomethyl dithiophosphate) at 600 g a.i. ha-1 in 1994-1995, aldicarb (2-methyl-2-methylsulfenylpropionaldehyde) at 450 g a.i. ha-1 for the remainder]. Experiments were also sprayed with insecticides in accordance with standard thresholds to prevent damage from pests not controlled by phorate or aldicarb and to prevent damage after the efficacy of these compounds had declined (4–6 wk after emergence). In Exp. 1, the nontransgenic cultivar Siokra V15 was used, which is an okra-leaf variety. In Exp. 2 to 6, a normal-leaf transgenic cotton cultivar, Sicala V 2i, was used, which contains a gene from Bacillus thuringiensis that produces the insecticidal protein, CryIAc, and provides control of Helicoverpa spp.

In each experiment, a randomized block design was used. Five replicates were used in Exp. 1 and four in the remainder. In Exp. 1 to 3, plots were 2 m long and 3 rows wide (3 m). The plots and blocks were laid out contiguously both end-to-end and across-the-rows. The two outer rows of each plot were buffers across the rows, and in each plot the 0.5 m on either end of the central row was a buffer. Only the middle 1 m of cotton in the central row was used for yield and maturity assessments (described below). In Exp. 4, plots were 3 rows wide but were increased in length to 6 m to allow for destructive samples to be taken (described below). As above, 1 m in the central row of each plot, with a buffer of 0.5 m at each end was preserved for yield and maturity assessments. In all experiments, damage treatments were applied to all three rows of each plot. Plant densities were in the range of 10 to 12 plants m-1 across all experiments.

Thrips damage was simulated by leaf area removal of true leaves and entailed cutting off the whole leaf at the top of the petiole (100% reduction) or cutting off portions of leaves to simulate different degrees of reduced leaf area. With the exception of Exp. 1, the cotyledons were not damaged, as we noticed they generally suffer little pest damage. Light terminal damage, simulating damage by thrips, mirids, or Helicoverpa, involved pinching out the terminal and the surrounding two unfurled leaf primordia with curved forceps. Cotton tipworm bores down through the terminal into the stem cortex, often resulting in the death of the terminal and of one or two nodes below the terminal. Such heavy tip damage was simulated by pinching off the terminal and upper two nodes, as defined by the presence of unfurled leaves. In some experiments, defoliation and tip damage were imposed repeatedly to better simulate the effect of prolonged damage to the plant by a pest population. Where terminal damage was repeated, all of the new dominant terminals were removed, which meant that for some plants, more than one terminal was removed. Helicoverpa spp. damage was simulated by removing all of the young squares from the first four sympodial branches. The growth stage of plants was defined each time damage was inflicted by counting the number of main-stem nodes of 20 undamaged plants in an area immediately adjacent to the experiment. Where damage resulted in significantly delayed growth, the timing of repeated damage was delayed as necessary to ensure that sufficient regrowth had occurred for the damage to be imposed accurately.

The effect of damage was assessed in terms of yield of cotton lint and the date of maturity of the crop. Maturity was defined as the date at which 60% of mature fruit (bolls) had matured and dehisced (Snipes and Baskin, 1994). We harvested all of the open bolls each week from each plot, beginning from the date that open bolls were first found in the experiment. Analysis of variance was used to test for differences in yield, lint weight per mature boll, and date of crop maturity between treatments and the undamaged control (Genstat Version 5, Lawes Agricultural Trust, IACR, Rothamsted, UK). When the analysis of variance was significant, Fisher's Protected LSD was used to compare treatment means with the control. The treatments imposed in each experiment and rationale is outlined below.

Experiment 1—Defoliation and Tip Damage
Treatments were (i) defoliation by removing 50% of the area of all leaves, including the cotyledons, (ii) light tip damage (see description under Exp. 5), and (iii) the combination of (i) and (ii). Damage was imposed when plants had {approx}3, 5, and 8 nodes, either at each node individually (i.e., 3, 5, or 8), or progressively (i.e., 3, 3 plus 5, 3 plus 5 plus 8). There where 16 treatments in total, including an undamaged control (Table 1). On each damage date, the number of main-stem nodes and total leaf area of undamaged plants, measured with a leaf area meter (LI-300, LI-COR Inc., Lincoln, NE), was recorded for four samples of five plants.


View this table:
[in this window]
[in a new window]
  		Table 1. Treatments imposed in each experiment and plant growth measurements taken.

 
Experiments 2 and 3—Defoliation, Tip Damage and Fruit Loss
In these experiments, we investigated the effects of defoliation imposed alone and in combination with light tip damage and/or square removal (Table 1). The treatments were designed to simulate (i) moderate thrips damage: 50% defoliation of each leaf twice (when plants had two and four nodes) or four times (when plants had 2, 4, 6, and 8 nodes); (ii) moderate thrips damage as in (i), followed by square loss, to simulate damage by Helicoverpa spp.; (iii) heavy thrips damage: 100% defoliation twice (two and four nodes); (iv) heavy thrips damage as in (iii), with tip damage at the same times; (v) heavy thrips damage with square removal; (vi) heavy thrips damage with tip damage at the same time, followed by square removal; and (vii) square removal alone. There were 10 treatments in total, including the undamaged control (see x-axis of Fig. 1). For Exp. 2, we recorded additional details of plant growth status at each damage event by collecting four samples of five undamaged plants from similar cotton adjacent to the experimental area and counting the number of main-stem nodes and fruit for each plant and total leaf area of each group of plants. These data were analyzed separately for Exp. 2, but the yield and maturity data for both experiments were analyzed together for Exp. 2 and 3.



View larger version (40K):
[in this window]
[in a new window]
  		Fig. 1. Effect of defoliation, tip damage, and early fruit loss on maturity of cotton in Exp. 2 and 3. Values are means + SE. Asterisks indicate treatments significantly different from the control at P = 0.05. For treatments, F indicates removal of fruit from first four fruiting branches; 50L or 100L are percentages of the leaf tissue removed; number of damage events is indicated by x 2 (twice) or x 4 (four times); and tip damage is indicated by T.

 
Experiment 4
Experiment 4 investigated in more detail the plant responses to defoliation in terms of yield and maturity, but also in terms of growth and development (Table 1). Defoliation levels of 0, 40, 60, 80, and 100% true leaf damage were imposed. Each damage level was imposed at either Nodes 2 and 4 or at Nodes 2, 4, and 6. There were nine treatments in total.

Destructive harvests of 0.5 m of row were collected to monitor the amount of leaf area actually removed and the recovery of plants in terms of mass (shoot + tap root) and leaf area. Immediately following each damage event (28, 36, and 44 d after sowing, DAS) and at 62 and 104 DAS, 0.5 m of row was harvested from the central row of each plot. The number of plants in each 0.5 m was recorded and the plants partitioned into leaves, squares, green bolls, open bolls, and either stem and stems plus tap-root (first three harvests) or stems and root separately (final two harvests). Total leaf area was determined with a leaf area meter, the samples dried at 80°C and the dry weight (DWT) of each structure type recorded.

Experiments 5 and 6
These experiments investigated the effects of different numbers of events of heavy or light tip damage (Table 1). Light tip damage was repeated weekly three, five, or seven times beginning from Node 2; heavy damage was inflicted one, three, or five times beginning at Node 2. Repeated events of heavy damage were often delayed for up to an extra week to allow time for sufficient regrowth for further damage to be imposed accurately. There were seven treatments in total, including the undamaged control. These two experiments were replicates of the same experimental design and hence were analyzed together.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Experiment 1
Actual timing of damage was close to the nominal treatments of 3, 5, and 8 nodes, that is, 3.5, 5.1, and 8.0 nodes. At these stages, undamaged plants had leaf areas of 97, 222, and 662 cm2 respectively. Neither 50% defoliation nor tip damage, or the combination of the two, affected the number of harvested bolls per meter (109.6 ± 1.7, mean ± SE; F = 1.28; df = 4, 60; P = 0.24), mean boll weight (2.2 ± 0.01 g, mean ± SE; F = 1.75; df = 4, 60; P = 0.065), lint yield (240.8 ± 3.8 g m-2; F = 0.97; df = 4, 60; P = 0.5) or maturity date (DAS) (179.1 ± 0.5 d; F = 1.59; df = 4, 60; P = 0.1) of cotton, even when both defoliation and tip damage combined were repeated on three occasions.

Experiments 2 and 3
In Exp. 2, damage was imposed when undamaged plants had 2.1, 3.9, 6.4, and 6.8 nodes and leaf areas of 142, 197, 345, and 374 cm2, respectively. Square removal was done on 19 Dec. 1996 when control plants had 12.8 nodes, a leaf area of 2536 cm2, and six squares per plant. An average of 24.9 young squares were removed per meter of row. This value is less than expected (i.e., expect six squares per plant x 10 plants m-1 = 60 squares m-1 removed) because some of the squares on later fruiting branches and some very small squares were overlooked in the field but were counted on plants partitioned in the laboratory. The 100% defoliation treatments had delayed growth so removal of squares in these plots was delayed until 30 December in the 100% defoliation twice treatments, when 37 squares m-1 were removed, and until 7 January for the 100% defoliation twice plus tip damage treatment, when 42 squares m-1 were removed.

In Exp. 2, the effect of damage on plant height (cm) was assessed and both leaf damage and tip damage caused significant reductions. After the damage event at Node 4, plants with 50% (9.4 ± 0.4) and 100% (8.2 ± 0.2) defoliation and 100% defoliation plus tip damage (7.3 ± 0.3) were shorter than the control (10.3 ± 0.4) (F = 12.7; df = 4, 95; P < 0.001). After the final damage event at Node 8, plants with 50% defoliation twice (14.6 ± 0.4) were no different from the control (14.8 ± 0.7), while plants with 50% defoliation four times (13.3 ± 0.3), 100% defoliation twice (10.6 ± 0.4) or 100% defoliation twice plus tip damage (8.8 ± 0.3) were shorter than the control (F = 36.9; df = 4, 95; P < 0.001).

Yield, yield components, and maturity for Exp. 2 and 3 were analyzed together, with experiment as a treatment in the ANOVA. Neither boll number (92.9 ± 1.1 m-2; F = 1.64; df = 9, 63; P = 0.12), boll weight (2.12 ± 0.02 g; F = 1.65; df = 9, 63; P = 0.12) nor lint yield (196.0 ± 2.9 g m-2; F = 1.49; df = 9, 63; P = 0.17) were affected by early season defoliation, tipping out, or square removal, even in the treatments combining 100% leaf removal and tipping out on two occasions with early square loss.

There was a significant interaction between experiments, defoliation, and square damage (F = 2.91; df = 3, 50; P = 0.044) (Fig. 1) for crop maturity. On plants with no square damage, moderate defoliation (50% twice or four times) caused a small but significant delay of 3 to 6 d across both experiments. Heavier defoliation (100% twice) caused a significant delay, which was longer in Exp. 2 (19 d) than Exp. 1 (13 d). When square damage was combined with moderate defoliation, delays (8 to 11 d) were similar to that caused by square damage alone (8 to 9 d). This means that square damage increased the delay from moderate defoliation by {approx}4 d, indicating that the effect of square damage was less than additive (i.e., extended maturity by 4 d rather than 8 d). When square damage was combined with heavy defoliation, the increase in the delay in maturity was additive for Exp. 1; that is, 13 to 21 d, an increase of 8 d. In Exp. 2, however, addition of square damage caused little additional delay; that is, 19 to 21 d, an increase of 2 d.

Experiment 4
Damage treatments were applied at 2.6, 4.9, and 6.4 nodes, which occurred at 28, 36, and 44 DAS, respectively. As expected leaf area differed between defoliation treatments (leaf removal twice: F = 142; df = 4, 27; P < 0.001; leaf removal three times: F = 185; df = 4, 42; P < 0.001) (Fig. 2).



View larger version (28K):
[in this window]
[in a new window]
  		Fig. 2. Relative leaf area and shoot dry weight of cotton in Exp. 4. Bars indicate Fisher's Protected LSD values at P = 0.05; arrows indicate the timing of damage treatments; and numbers on each graph indicate the actual leaf area (cm2) or shoot dry weight (g) of undamaged plants at the time samples were collected. Legend refers to amount of leaf area removed (i.e., 100%) by the number of damage events (i.e., x 2 or x 3).

 
Damage resulted in reductions in total leaf area (true leaves plus cotyledons), total shoot DWT (Fig. 2), and root DWT (Table 2). Data for leaf area and shoot DWT were analyzed on a per-plant basis, as there were strong effects of plant density in earlier sample dates (28, 36, and 44 DAS). Root, square, and boll DWTs were analyzed on a per-square-meter basis, as plant density effects were generally not significant at this stage.


View this table:
[in this window]
[in a new window]
  		Table 2. Specific leaf weight, root, square, and boll dry weights (DWT) and reproductive allocation [slope of ln(reproductive DWT) against ln(vegetative DWT)], Exp. 4.

 
Leaf area and shoot DWT differed among damage treatments and the control for dates following damage (28, 36, and 44 DAS) as expected (leaf area: F = 21.7 to 38.2; df = 8, 24; P < 0.001; DWT: F = 3.4 to 30.1; df = 8, 24; P = 0.01 to P < 0.001), with the exception of 50% defoliation three times, which was not different from the control at 44 DAS (Fig. 2). At 62 DAS, the 50% defoliation treatments and the 75% defoliation twice treatment were no different from the control, but the higher defoliation treatments were different (leaf area: F = 8.7; df = 8, 24; P < 0.001; DWT: F = 10.6; df = 8, 24; P < 0.001). At 104 DAS, no treatments were different to the control (leaf area: F = 0.55 to 38.2; df = 8, 24; P = 0.81; DWT: F = 0.9; df = 8, 24; P = 0.53).

The development of cotton with heavier damage treatments was slower than the control. Node production of cotton with 100% defoliation twice (36 DAS: 3.4 ± 0.6; 44 DAS: 4.2 ± 0.2 nodes) or three times (36 DAS, 3.4 ± 0.4; 44 DAS, 4.5 ± 0.3) was behind that of the control (36 DAS: 5.0 ± 0.5; 44 DAS, 6.4 ± 0.3) (36 DAS: F = 2.37; df = 8, 24; P < 0.049; 44 DAS: F = 7.1; df = 8, 24; P < 0.001). Other treatments were similar to the control on all dates.

Specific leaf weight [DWT of leaves (g)/leaf area (m2)] of damaged and undamaged treatments was compared as it could indicate differences in leaf thickness. Specific leaf weight was not different among treatments at 28 or 36 DAS, which was immediately after the first and second damage event, or at 62 or 104 DAS (F = 1.88 to 2.35; df = 8, 24; P = 0.051 to 0.11). However, the specific leaf weight of all treatments, except 50% defoliation twice, was lower than the control at 44 DAS, indicating that leaves of damaged plants could be thinner than those of the control (F = 3.69; df = 8, 24; P = 0.006) (Table 2).

Tap-root DWT at 62 DAS was reduced, compared with the control, by most defoliation treatments (F = 12.0; df = 8, 23; P = 0.001) with the exception of the 50% defoliation twice treatments (Table 2). At 104 DAS, there was no difference in root DWT (F = 0.9; df = 8, 24; P = 0.56). Square biomass was lower in the more severe damage treatments at 62 DAS, as indicated by square DWT (F = 6.2; df = 8, 24; P < 0.001) (Table 2). At 104 DAS, most damage treatments had greater square biomass than the controls (F = 6.4; df = 8, 24; P = 0.001), especially the most severe damage treatment. Boll biomass at 104 DAS was also affected by treatments (F = 4.3; df = 8, 24; P = 0.002) with boll DWT being lower than the controls in the two 100% defoliation treatments (Table 2). Allometric analysis of reproductive allocation [slope of ln(reproductive DWT) against ln(vegetative DWT)] was lower in many of the heavier damage treatments (F = 4.1; df = 8, 24; P = 0.003) in the period between 62 and 104 DAS, indicating lower allocation of resources to reproductive tissue in these treatments during this period (Table 2).

Defoliation had no significant effect on final boll number (125.8 ± 3.6 m-2, F = 0.6; df = 8, 24; P = 0.73), mean boll weight (2.3 ± 0.02 g, F = 0.6; df = 8, 24; P = 0.87) or yield (288.1 ± 7.8 g m-2, F = 0.4; df = 8, 24; P = 0.88). Only the heavier defoliation treatments affected time of maturity (F = 9.2; df = 8, 24; P < 0.001) (Fig. 3). Defoliation of 87% twice, 100% twice, or 100% three times caused significant delays of {approx}4, 5, or 10 d, respectively.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 3. Effect of defoliation on crop maturity in Exp. 4. Asterisks indicate treatments significantly different from the control at P = 0.05. Values are means ± SE. Treatment labels refer to the amount of leaf area removed (i.e., 100%) by the number of damage events (i.e., x 2 or x 3).

 
Experiments 5 and 6
Tip damage was inflicted up to seven times, when the undamaged control had 3.4, 4.3, 6.1, 10.5, 12.2, 13.8, and 14.3 nodes in Exp. 5 or at 2.7, 5.9, 9.5, 11.2, 11.3, 14.9, and 15.2 nodes in Exp. 6.

Boll number did not differ between experiments (F = 0.33; df = 1, 3; P = 0.6) or among treatments (F = 0.53; df = 6, 36; P = 0.78) averaging 91.8 ± 2.1 bolls m-2 (mean ± SE) at maturity. Boll weight differed between experiments (Exp. 5, 1.86 g; Exp. 6, 2.05 g)(F = 13.3; df = 1, 3; P = 0.035) and among treatments (F = 3.5; df = 3, 36; P = 0.008). Light damage seven times or heavy damage three or five times reduced boll weight (Table 3). Yield did not differ between experiments (F = 2.44; df = 1, 3; P = 0.21) but differed among treatments (F = 3.2; df = 3, 36; P = 0.012). Light damage seven times or heavy damage five times reduced yield compared with the control (Table 3). Maturity date (DAS) differed between experiments (Exp. 5, 176.7 g; Exp. 6, 188.6 d; F = 739.8; df = 1, 3; P < 0.001) and among treatments (F = 23.6; df = 3, 36; P < 0.001). Lighter damage treatments, light damage three times or heavy damage once caused minor delays of 2 to 5 d. Intermediate damage, light damage five times or heavy damage three times caused longer delays of 8 to 9 d, while the heavier damage treatments, light damage seven times or heavy damage five times caused substantial delays of 13 to 14 d (Table 3).


View this table:
[in this window]
[in a new window]
  		Table 3. Number of bolls, boll weight, and yield of treatments in combined Exp. 5 and 6.

 
Relationships between Damage and Crop Maturity or Yield
Regression was used to explore the relationship between the frequency of damage, the severity of damage, and the reduction in yield or delay in maturity of cotton. Data for each treatment in each experiment were compiled into a dataset which included identifiers for the amount of leaf area removed (DR, 0–100%), the duration of the damage, expressed as the final node at which defoliation was imposed (DD, control = 0), the number of tip damage events (TE, 0–7 events), and the severity of tip damage events, expressed as phyllochrons (time between exsertion of leaves) removed, assuming that in the terminal there are four phyllochrons, two visible and two embryonic, (TS: none = 0; light = 4 phyllochrons; heavy = 6 phyllochrons).

Our expectation was that effects on yield or maturity would be because of the interaction between the severity of the damage and its duration or repetition. Hence, we calculated interaction terms for defoliation (DRDD) and tip damage (TETS). These were regressed against the delay in maturity (days) of each treatment compared with the control for each experiment and the crop yield, expressed as the % reduction in yield compared with the control, that is, (treatment yield x 100/control yield) - 100. The interaction terms were fitted additively and also in interaction to test for the possibility of interaction between defoliation and tip damage, that is, DRDD + TETS, and DRDD x TETS. A term denoting each experiment was also fitted to test for differences in responses between experiments.

Crop maturity, expressed as days earlier (-ve) or later (+ve) than the control, was well explained by the equation

[1]
where Exp. 1 = -2.56, Exp. 2 = 1.82, Exp. 3 = 5.36, Exp. 4 = -2.48, Exp. 5 = -0.94, and Exp. 6 = 1.46. The interaction term DRDD x TETS was not significant (P = 0.35), indicating no interaction between defoliation and tip damage in this data. A simplified version excluding the experiment term was also fitted for use in estimating general responses:

[2]

Crop yield was less well explained:

[3]
where Exp. 1 = 7.26, Exp. 2 = 0.78, Exp. 3 = -4.22, Exp. 4 = 2.76, Exp. 5 = 4.78, and Exp. 6 = 3.47. The interaction term DRDD x TETS was not significant (P = 0.13), indicating no interaction between defoliation and tip damage. Similarly, a simplified version excluding the experiment term was also fitted for use in estimating general responses:

[4]


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
Defoliation
Damage by pests such as thrips often results in reduced leaf area and size of cotton plants which is visually very striking (Sadras and Wilson, 1998). In field experiments involving actual thrips damage, the dynamics of leaf area, expressed as the ratio between crops damaged by thrips and insecticide-protected crops, showed a typical biphasic pattern, including an initial phase when reductions in leaf area reached a maximum of 20 to 50% {approx}40 DAS, and a second phase when leaf area in crops damaged by thrips increased faster than in controls (Sadras and Wilson, 1998). A similar biphasic pattern that was displaced by {approx}10 d was observed for shoot growth (Sadras and Wilson, 1998). Here, we were able to reproduce very similar patterns of leaf area and shoot dry matter reduction and recovery, summarized in Fig. 2. The overall similarity between the patterns of leaf area, shoot dry matter, crop yield, and maturity found in previous studies involving actual insect damage (i.e., Sadras and Wilson, 1998), and the responses generated with manual leaf removal reinforces the confidence in this technique.

We found that moderate to intense levels of damage, up to 50 to 87.5% loss of true leaf area twice, had transient effects on plant growth. Plants recovered from such damage rapidly with little effect on the onset of fruiting, or ultimately on the yield or maturity of cotton. For example, in Exp. 4, plants with 75% leaf area removed twice had recovered in leaf area and DWT by 62 DAS (34 d after damage). Earlier studies have similarly found that large reductions in leaf area in excess of 75% are required before cotton yield was affected. Lane (1959) simulated loss of 25, 50, 75, or 100% of leaf area at the seedling, squaring, flowering, or boll-filling stages. At the seedling and squaring stages, only the 100% leaf removal treatments consistently reduced yield. Kerby and Keely (1987) also found that manual removal of the first two true leaves had no significant effect on early plant growth.

Longer et al. (1993) and Kerby and Keely (1987) found that severe artificial defoliation damage could reduce the growth rate of cotton. We also found a similar exponential response to loss of leaf area in the seedling stages (up to eight true leaves) (Eq. [1] and [2]), i.e., defoliation had little effect on either yield or maturity until high levels, in excess of 90% removal, regardless of how often imposed. For instance, at 44 DAS, plants with 100% true leaf removal three times lagged behind the controls by 1.7 nodes and at 62 DAS their leaf area and leaf DWT were below that of the control. This delay in recovery probably explains the delays in both square and boll production. Recovery of leaf area and plant DWT was complete by 104 DAS, and yield was not affected, though maturity was, as would be expected from the delay in fruit production. These findings suggest that complete, or almost-complete loss of leaf area results in an extreme shortage of assimilate as new leaf area developed during the early stages of recovery is effectively a sink rather than a significant exporter of assimilate. This shortage of assimilate may delay growth, as indicated by delays in node production in Exp. 4, and hence ultimately result in delayed fruit development with eventual delayed crop maturity and reduced yield if the delay pushes boll maturation into unfavorable conditions.

The mechanisms of recovery from such damage are not clear; however, our results provide some support for the hypothesis of Sadras and Wilson (1998), that reduced specific leaf weight may contribute to plants recovering from defoliation. That is, defoliated plants are able to increase their leaf area, and hence light interception, by making leaves thinner. This hypothesis would need to be tested, however, with assessment of the specific leaf weight of individual leaves on damaged vs. undamaged leaves. In this study, the younger average age of leaves on damaged plants may have caused a bias toward lower specific leaf weight that we cannot exclude.

Allometric analysis further indicated that treatments with more severe damage had a reduced allocation of dry matter to reproductive structures, enhancing recovery of vegetative structures and, hence, leaf area index. This enabled plants to recover without loss of yield though with delayed maturity, suggesting that the allometric ratio of heavily damaged plants eventually achieved that of undamaged plants. The postponement of allocation to reproductive structures partially explains the delay in crop maturity of heavily damaged treatments. More frequent dry matter harvests would allow greater discrimination between more and less severe damage treatments in the rate and mechanism of recovery following defoliation.

Tip Damage
Our results show that cotton is able to compensate well for repeated tip damage events up to a point. Cotton tolerated up to seven light tip damage events without affecting yield and up to three without affecting maturity. Lower amounts of heavy tip damage could be tolerated; for example, three events did not reduce yield and one event caused only a minor delay in maturity. Others have similarly reported a high tolerance of cotton for terminal damage. Brook et al. (1992b) reported that tip damage applied at four or six true leaves or at first-square did not reduce yield and caused delays of 1 to 4 d. They found in one experiment that early (Node 4) or late (first square) damage resulted in a significant yield increase.

The main effect of tip damage is to cause production of vegetative branches (monopodia), which in turn produce fruiting branches (sympodia). Tip damage could therefore be expected to cause an initial delay in the development of fruiting branches, but thereafter a higher number of vegetative branches would potentially lead to a faster rate of production of fruiting branches and therefore fruiting sites (Lei and Gaff, 2003). The delay in reestablishment of vegetative branch growth and hence fruiting could be expected to be related to the severity and repetition of damage, as our results show. Light tip damage could be expected to delay the onset of fruiting less than heavy damage of a similar frequency. The delay in fruiting caused by tip damage would be expected to have a greater effect on maturity than yield, because provided growing conditions are adequate the plant should have time to mature a similar fruit load, and our results support this assertion. At some point, extensive tip damage could be expected to result in yield loss as well as delayed maturity, as a plant's growth is delayed sufficiently to push the fruit maturation period into less favorable growing conditions at the end of the season or because of the plant's capacity to develop a full canopy is curtailed, thereby reducing light interception and the plant's assimilate supply and hence yield potential. Intraspecific variation in cotton responses to tip damage has been identified and also needs to be considered, as the recovery of some varieties is faster following tip damage than others (Sadras and Fitt, 1997).

Interaction between Damage Types
There was no significant interaction between defoliation, tip damage, and fruit damage. The effects of each type of damage were additive (see Eq. [1] and [2]). Tip damage might be expected to promote recovery from other forms of damage by increasing branching and, hence, leaf area and fruiting sites. However, others have similarly found that tip damage interacts additively with other types of damage, such as fruit loss. For instance, Brook et al. (1992b) found that early tip damage in the variety Siokra 1-1 increased yield by 211 kg lint ha-1 compared with undamaged cotton, whereas heavy early fruit loss resulted in a loss of 63 kg lint ha-1. The combination of these treatments resulted in a gain of 100 kg lint ha-1, hence Brook et al. (1992b) found that tip and fruit damage were statistically additive. Nevertheless, in Exp. 2 and 3 there was a nonsignificant trend for plants with both defoliation and tip damage to have higher yields than plants with defoliation alone, suggesting that further investigation of the interactions between tip damage and other types of damage may be justified.

Evenness of Damage
In our studies, plants within a treatment were uniformly damaged; that is, all plants were damaged. However, pest distributions in cotton fields are rarely uniform, often showing patchy or aggregated distributions (Wilson and Room, 1983; Wilson and Morton, 1993). It is therefore likely that the distribution of damage is less uniform than in our experiments. Our studies examine the capacity of plants suffering a simulated level of damage to recover, that is, plant level responses. Sadras (1996d), however, showed that the uneven distribution of damage allows for population level compensation; that is, when "herbivore attack on one individual allows another to grow more rapidly" (Crawley, 1983). Sadras (1996d) compared crops that were uniformly tip damaged, uniformly undamaged, or unevenly damaged, where every second plant was tip damaged, and found evidence of strong plant–plant interactions. It is likely that damage from thrips is uneven and this should be considered in future research.

Implications for Pest Management
The results presented here therefore confirm the conclusion of Sadras and Wilson (1998), that early season pest damage is often largely cosmetic, with little effect on crop yield or maturity, despite its highly visual and dramatic appearance. These experiments were all done in fully irrigated, well-fertilized crops in the lower Namoi Valley, which is a full season region. Season length, as limited by temperature or water availability, and nutrient availability could constrain the potential ability of plants to compensate for herbivory (Oesterheld and McNaughton, 1991; Sadras, 1996c). Weather could also affect recovery; for instance, in cooler regions there may not be sufficient time for recovery before crop growth is limited by temperature. In some of the cooler regions in the more north-easterly cotton production zones of the USA, thrips damage has often been shown to stunt growth, delay maturity, and reduce yield (Hawkins et al., 1966; Johnson et al., 1988). Similarly, soil type could also be important; for instance, poorer soils with low water-holding capacity may not support as vigorous growth or recovery compared with the soils in this study. Our results therefore cannot be reliably extrapolated to dryland crops, or to cooler or shorter season regions, which require further research.

Improvements in IPM in Australian cotton need to take into account the early season compensatory capacity of cotton. This period is critically important in Australia, as it corresponds with the movements of beneficial populations into cotton from other hosts. Disruption of these populations by broad-spectrum insecticides can increase the risk of outbreaks of secondary pests such as spider mites (Wilson et al., 1998) or aphids (Wilson et al., 1999), as well as reducing the effect of beneficials on primary pests such as Helicoverpa spp. In terms of IPM, therefore, the results suggest that reasonably high levels of defoliation and/or tipping-out can be tolerated without the need to spray, thereby reducing costs, environmental pollution, and helping to conserve beneficial insect populations. Significantly, amongst the pests often targeted early season, the phytophagous thrips are also important predators of mite eggs (Wilson et al., 1996).

A limitation of many current pest thresholds is that they are developed assuming an average level of damage from given pest density. However, a given pest density can be associated with a range of levels of plant damage, depending on earlier pest numbers and on plant growing conditions. Thresholds that took into account both pest abundance and plant damage levels would allow for the possibility that a pest may exceed an abundance threshold but plant damage does not exceed a damage threshold; therefore, control could be avoided or delayed.

Simple sensitivity analyses were done with Eq. [2] and [4] to help derive potential thresholds for management of defoliation or tip damage (Tables 4, 5). As an indication of a link between damage and commercial practice, we assumed that delay was significant if it was longer than 5 d, which is the level that normally begins to cause concern for cotton growers. We assumed yield loss was economically important if it was >4%. This was based on a grower wanting to do more than cover the cost of control, that is, double his money, assuming that the crop is valued at $2800 U.S. ha-1 (7 bales, 227 kg per bale, $400 per bale) and control costs of $50 ha-1 ($40 insecticide + $10 application), hence a yield loss of {approx}2% is required just to recoup control costs. Both crop yield and maturity are relatively insensitive to defoliation, because of the power nature of their response to the proportion of leaf area removed. This can be seen for crop maturity in Fig. 3. Defoliation up to 70% continuing as late as six true leaves has no economic effect on yield or maturity. Single terminal damage events had no effect on maturity or yield but multiple events affected both. For instance, three light damage events or two heavy damage events caused a delay of >5 d or yield loss exceeding 5%. In the field, plants are often exposed to combinations of both tip damage and defoliation, and it is possible to derive estimates of delay or yield loss combining both types of damage (Tables 4, 5).


View this table:
[in this window]
[in a new window]
  		Table 4. Effect of defoliation of true leaves and of tip damage on maturity of cotton, from Eq. [2]. Underscoring indicates damage combinations resulting in a delay of 5 d or less.

 

View this table:
[in this window]
[in a new window]
  		Table 5. Effect of defoliation of true leaves and of tip damage on yield of cotton, expressed as yield relative to undamaged cotton, from Eq. [4]. Underscoring indicates damage combinations resulting in yield reduction of 4% or less.

 
The values shown in Tables 4 and 5 can serve as tentative thresholds for plant damage that can be used in conjunction with pest abundance thresholds to allow better decisions. This is provided the growing conditions, soil types, crop nutrition, and irrigation are similar to those in these experiments. As an example, modified thresholds for thrips and Helicoverpa spp. in Australia now incorporate assessment of both pest abundance and plant damage and emphasize that both must be over threshold before pest control is justified (Deutscher and Wilson, 1999; Mensah and Wilson, 1999). The actual damage thresholds used are based on the studies reported here as well as results of real pest damage studies (Brook et al., 1992a; Sadras and Wilson, 1998). In the future, the information obtained in Exp. 4, where the recovery of damaged plants was monitored, may be used to link the effects of reduced leaf area from pests such as thrips with crop simulation models via their effects on plant growth.


   ACKNOWLEDGMENTS
 
We thank Dee Hamilton, Mark Laird, Allison Wales, Sally Kennedy, Lesley Burke, Deanne Johnson, Deirdre Lally, Les Bauer, Lyn Gett, Trudy Staines, Kelly Scott, Mike Mennell, and Kym Bush for technical assistance under trying conditions. Thanks to Greg Constable, Stephen Milroy, and Tom Lei (CSIRO Plant Industry) for critical review of the manuscript. This research was funded by the Cotton Research and Development Corporation (Grants CSP46C, CSP74C).

Received for publication December 21, 2002.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 




This article has been cited by other articles:


Home page
 ANN BOT (LOND)Home page
T. T. LEI and L. J. WILSON
Recovery of Leaf Area through Accelerated Shoot Ontogeny in Thrips-damaged Cotton Seedlings
Ann. Bot., July 1, 2004; 94(1): 179 - 186.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Wilson, L. J.
Right arrow Articles by Gibb, D.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Wilson, L. J.
Right arrow Articles by Gibb, D.
Agricola
Right arrow Articles by Wilson, L. J.
Right arrow Articles by Gibb, D.
Related Collections
Right arrow Crop Growth and Development
Right arrow Cotton
Right arrow Pest Management Systems

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome