Introduction
There can be dramatic changes in foraging strategies across instars in some Lepidoptera. Gaston et al. (Reference Gaston, Reavey and Valladares1991) found that 200 out of 1137 species of micro-Lepidoptera changed from leaf miners or concealed feeders as early instars, to external feeders, case bearers, or tying and leaf rolling caterpillars as late instars. Kantiki & Ampofo (Reference Kantiki and Ampofo1989), Coll & Bottrell (Reference Coll and Bottrell1991) and Kumar (Reference Kumar1992) describe foraging changes from external feeders to borers; Ostrinia nubilalis (Hübner), Eldana saccharina Walker and Chilo partellus (Swinhoe) (all Pyralidae) all feed on maize leaf sheaths or tassels until approximately the third instar, when they bore into the stem of the host. Similarly, 5–10% of Lepidoptera, across a diverse taxonomic array, are gregarious feeders in the early instars (see Stamp, Reference Stamp1980; Herbet, Reference Herbet1983). These aggregations often break up by the third or fourth instar when larvae feed alone or in small groups (Common, Reference Common1990). Other larvae may change from leaf trenchers to petiole or vein notching/snipping with instar stage (Zalucki & Brower, Reference Zalucki and Brower1992; Dussourd, Reference Dussourd, Stamp and Casey1993; Clarke & Zalucki, Reference Clarke and Zalucki2000). For most Lepidoptera, however, the various instars appear to feed in similar places and changes in behaviour, if any, are likely to be subtle (e.g. in the exposed foliage feeders).
The majority of studies on foraging behaviour of exposed foliage feeders usually only consider one instar, generally a later instar, as these are easier to observe (Scriber & Slansky, Reference Scriber and Slansky1981). The implicit inference is that early instars forage (feed and move) in a similar way to larger instars. For example Benedict et al. (Reference Benedict, Altman, Umbeck and Ring1992a, Reference Benedict, Treacy, Ring and Yenchob) and Treacy et al. (Reference Treacy, Benedict, Schmidt, Anderson and Wagner1987) studied movement and feeding of Heliothis virescens (Fabricius) (Noctuidae) using third instars but infer that the results apply to all larval stages. Since penultimate and final instars exhibit the largest weight increase and consume the bulk of plant tissue, the relative growth rate (RGR) indices of these stages are presumed indicative of the insect's development and perhaps behaviour (Scriber & Slansky, Reference Scriber and Slansky1981), although this hypothesis is rarely tested.
There is evidence to suggest that early instars are not just smaller versions of late instars. Feichtinger & Reavey (Reference Feichtinger and Reavey1989) demonstrate that there can be subtle differences in feeding strategy between instars of the same species. They found first instar Achlya flavicornis galbanus Tutt (Thyatiridae) often feed within tied leaves, while later instars always forage away from the tie. Subtle differences may be found in Heliothineae. Schmidt et al. (Reference Schmidt, Benedict and Walmsley1988) showed that resistance characteristics in cotton affected the behaviour of first instar H. virescens more than later instars. Hassan (Reference Hassan1983) showed differences in the duration of feeding events in Helicoverpa sp. larval stages in Australian cotton.
The increased emphasis on the use of genetically modified plants and behaviour-modifying chemicals to control insect pests requires a better understanding of the feeding and foraging behaviour of larvae (Scriber & Slansky, Reference Scriber and Slansky1981; Abbott & Fitt, Reference Abbott and Fitt1998; Zalucki et al., Reference Zalucki, Clark and Malcolm2002; Isaacs & Vander Werp, Reference Isaacs and Vander Werp2006). Although early instars are often the target for control techniques, studies that investigate movement and feeding (foraging) of lepidopteran early instars are infrequent (Schmidt et al., Reference Schmidt, Benedict and Walmsley1988; Eigenbrode et al., Reference Eigenbrode, Stoner, Shelton and Kain1991; Zalucki et al., Reference Zalucki, Clark and Malcolm2002). This is largely due to difficulties in applying observation methods to very small larvae. In a recent review of early stage lepidopteran survival, Zalucki et al. (Reference Zalucki, Clark and Malcolm2002) argue that the first instar needs closer consideration because mortality is frequently highest in this stage. First instars must successfully establish feeding sites and are at greater risk of succumbing to constitutive plant defences and predation by invertebrates (Titmarsh, Reference Titmarsh1992; Bernays, Reference Bernays1997).
Most studies on pest heliothines determine where feeding has occurred or where larvae are currently located, not how they got there (Wilson & Waite, Reference Wilson and Waite1982; Abbott & Fitt, Reference Abbott and Fitt1998). In Australian studies to date, the movement (as opposed to distribution) of Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae) has been observed only on cotton (Hassan, Reference Hassan1983; Abbott & Fitt, Reference Abbott and Fitt1998; Manners, Reference Manners2001). Similarly, studies on the closely related species (Helicoverpa zea (Boddie) and H. virescens) have concentrated on the location or movement of large larvae on cotton (Pitre, Reference Pitre1983; Hopper & King, Reference Hopper and King1984; Treacy et al., Reference Treacy, Benedict, Schmidt, Anderson and Wagner1987; Benedict et al., Reference Benedict, Altman, Umbeck and Ring1992a, Reference Benedict, Treacy, Ring and Yenchob; Jyoti et al., Reference Jyoti, Young, Johnson and McNew1999). With the exception of Schmidt et al.'s (Reference Schmidt, Benedict and Walmsley1988) study of H. zea on cotton, no studies combine feeding and movement (foraging) to test for differences between instars on whole plants.
Since understanding feeding and movement of immature stages is important for diverse aspects of caterpillar biology and ecology (e.g. likelihood of mortality, nutritional ecology, foraging theory and optimizing control strategies), we test the assumption that all instars are equivalent by determining whether differences occur in foraging behaviour between first and third instars. Helicoverpa armigera, a highly polyphagous Old World species that has been recorded on 101 native and cultivated plant species in 31 families in Australia alone (Zalucki et al., Reference Zalucki, Daglish, Firempong and Twine1986, Reference Zalucki, Murray, Gregg, Fitt, Twine and Jones1994), was used as a test system. The larvae are considered pests on many field and horticultural crops throughout the species range, yet few studies have addressed the dynamics of larval foraging behaviour per se. Larvae feed on plant reproductive structures (pollen, flowers, and fruiting structures) but also feed on leaves. Larvae feed in exposed locations and burrow into structures such as fruits, hence the common name of bollworm or corn earworm. Zalucki et al. (Reference Zalucki, Clark and Malcolm2002) described the species as having a ‘mixed’ feeding strategy. Here, foraging of first and third instar larvae is directly compared, using the crop plant Vigna radiata (mung bean) as a simple model system. Helicoverpa armigera is considered a major pest of mung bean in Australia.
Materials and methods
Mung bean (var. Emerald) plants were grown in a glasshouse at the University of Queensland, St Lucia, Australia. Mung bean were cultivated from seed in 20 cm pots using standard potting mix (Envirogreen®) with Osmocote® slow release fertilizer added to the mix. Plants were used at 4 weeks of age when they had two true sets of compound leaves. Plants were on average 14.4 cm in height with 111 cm2 of leaf area. All plants were hardened off outdoors in direct sunlight 24 h prior to use.
Observations were made in a glasshouse at the University of Queensland. The activities associated with foraging were recorded every 2–5 min over 9 h (0600–1500). Ten plants, each containing a single larva were observed simultaneously on each experimental day. Time per observation on each larva ranged from 10 to 30 s. Using Observer® Noldus software the following activities were recorded: (i) crawling fast and slow (moving behaviours); (ii) resting (larva stationary with no apparent activity); (iii) feeding (notable downward head movement into the plant surface); (iv) waving up and waving down (larva moves head from side to side with the head up in the air or down on the plant surface – interpreted as searching behaviours); (v) dropping off the plant on a silk thread; and (vi) activity unknown (usually when the larvae were hidden amongst terminal leaves). As it is difficult to observe feeding by small larvae, all recorded feeding events were checked using a×10 hand lens to confirm the larvae were actually feeding. The plant part at which each activity occurred, (growing tip of the plant, juvenile leaves (upper and lower surfaces), mature leaves (upper and lower surfaces), petioles, stems and cotyledons) and the node number were recorded.
Pupae of H. armigera were provided by Queensland Department of Primary Industries and Fisheries (QDPI&F), Toowoomba, Australia. The QDPI&F culture is supplemented annually with field-collected H. armigera pupae reared through one generation to avoid contamination by diseases or parasites. As they emerged, adult moths were placed in a rearing cage and provided with sucrose solution and paper towel as a laying substrate. Newly emerged moths were added to the rearing cage daily. Helicoverpa armigera eggs were obtained each morning from the rearing cage. These eggs were kept at 22°C until almost hatching. The same source of eggs was used to rear larvae to third instar. These were reared on whole mung bean plants of the same age as those used for the experiments. At the start of each recording day third instars of uniform size that had just moulted (within l h) were selected for use.
An almost equal mix of first and third instars was used for observations each day. Each larva was placed on a separate plant and a new plant used for every replicate. Early morning (0600) a single black egg (almost hatching) was placed onto the newest fully open mature leaf at the top of a plant (node 3) with a fine-haired brush. This allowed hatching to occur directly on the plant and the larvae to feed on their egg chorion which is typical behaviour for this species. Observations commenced as soon as the first instar larvae had finished feeding on the egg chorion and moved away. At the same time, third instar larvae were transferred to the upper surface of the first fully opened mature leaf of individual plants. Third instars were allowed to ‘settle’ for 10 min prior to commencement of recording behaviours. This method avoided the influence of prior feeding sites and induced plant defences. Nine hours of observation time allowed an 8 h data set to be obtained for both first and third instars. This was necessary to allow enough first instar larvae to hatch (usually within half an hour) and move away from their egg chorion.
Treatments (first and third instars) were randomized within each experimental day. Data were collected every second day over 8 days with a total of 16 replicates for first instars and 24 for third instars. When a larva went missing for 30% or more of the recording time the replicate was discarded from the data set.
Temperature and relative humidity were recorded every 10 min over the course of the experimental period. The daily temperature averaged 30.9°C (range 19.2–42.6°C) during the day, reflecting Australian summer field temperatures.
Relative growth rates
Relative growth rates (RGRs) were calculated for each day (24 h) of development from newly hatched larvae to third instar and for each 24 h of third instar development directly after moulting. We used pigeon pea plants; Cajanus cajan (L.) Millspaugh (Quest variety) grown under similar conditions to mung bean in a glasshouse at the University of Queensland. Plants were cultivated from seed in 20 cm pots using standard potting mix (Envirogreen®) with Osmocote® slow release fertilizer added to the mix. All plants were hardened off in direct sunlight 24 h prior to use.
As transferring and weighing very small larvae had the potential to damage them or alter their feeding, a separate subset of larvae was used for each time period that weights were obtained. These larvae were obtained from eggs of similar size from the same cohort. Newly hatched first instar larvae from similar sized eggs (to reduce initial variation in size) were weighed in batches of 10 to obtain an initial weight. From these, similar-sized newly hatched individual first instars were placed on leaves in glass Petri dishes (5 cm diameter) and kept in a temperature-controlled (27°C) laboratory with natural light. Leaves were changed daily. Sets of 20–25 individuals were set up to give weight gain estimates after 24, 48, 72 and 96 h. The head capsules of each were measured to confirm instar. To determine the RGR for each day within the third instar stadia, a second group of larvae (n=15–27 for each 24-h time period) were allowed to develop on plants. When these moulted to third instar, larvae were individually weighed and provided with leaves in glass Petri dishes (5 cm diameter). Leaves were changed daily and weight gain was estimated for individuals after 24, 48 and 72 h. In all these experiments individuals for both first and third instar RGR estimates were only used once. Relative growth rate was calculated as the difference in the natural log wet weights between two time periods divided by time elapsed (see Kogan & Cope, Reference Kogan and Cope1974):
where wt1 and wt0 are the wet weights at times 1 and 0.
Data analysis
Data were originally recorded as durations in seconds. Because temperatures differed slightly over days, all data were converted to physiological time units (PTU) as is often done in these situations when working with insects (Gilbert et al., Reference Gilbert, Gutierrez, Frazer and Jones1976; Johnson & Zalucki, Reference Johnson and Zalucki2005). Our PTUs equate to degree minutes rather than degree days. Calculations of PTUs were based on the development model described in Room (Reference Room1983) and using temperature recordings made every 10 min over the experimental time.
As activities within individuals were not independent, mean time budgets (the proportion of an 8-h day spent on an activity), the mean number of events (frequency of an activity) and mean duration of a single bout within each activity were calculated for every individual. These data were analysed using Generalized Linear Models (SAS GLM procedure) to determine differences between larval stages with respect to the time spent on activities or positions on the plant. Day was used as a blocking effect. All proportional data were arcsine square root transformed before analysis. If durational data were not normally distributed they were log transformed and tested for normality. Where data could not be normalized a Kruskal-Wallis non-parametric test was used. Count data were analysed using a Kruskal-Wallis non-parametric test. Data that tested decision-making and the proportion of the population doing each activity were analysed using a log likelihood ratio (G) test. Differences in RGR were plotted against instar and compared using simple pair-wise two sample t-tests.
Results
All larvae started on the first fully mature leaf on the plant (at node 3). Analysis of the choice to move up or down the plant upon leaving the first mature leaf showed that there was no significant difference between first and third instars (likelihood ratio G=0.5987, d.f.=1, P=0.4391, based on 12 first and 15 third instars). In both instar stages the majority of larvae moved up (75% and 86% for first and third instars respectively). This applied to the second choice if the larvae returned to a mature leaf after moving around the plant (likelihood ratio G=0.6334, d.f.=1, P=0.4261, based on 10 first and 12 third instars). When larvae reached the top of the plant (the tip or first juvenile leaves) their decision to stay or move back down the plant was recorded. When comparing instar stage the analysis of the choice they made showed there was no significant difference between instar (likelihood ratio G=0.0271, d.f.=1, P=0.8692, based on 33 and 22 decisions for firsts and thirds, respectively). In both instars (first and third) the majority of larvae chose to stay at the top of the plant once they got there (88% and 86%, respectively).
The proportion of a day allocated to plant sites and doing key activities was determined for each larva. With the exception of the terminal (growing tip) of the plant and the lower side of mature leaves, the proportion of time spent at a site (juvenile leaves, mature leaves, stems, petioles or cotyledons) did not differ between instars. First instars spent a greater proportion of the day at the terminal than third instars (F=3.86, d.f.=1, 30, P=0.0226). Conversely, first instars spent less time on the lower side of mature leaves than third instars (F=4.01, d.f.=1, 30, P=0.0197).
The proportion of the 8-h day (time budget) spent doing activities associated with searching and feeding differed between instars (fig. 1). First instars spent more of their day waving up and down (F=4.23, d.f.=1, 30, P=0.0488) than third instars. The proportion of time spent crawling appeared different between instars but was not statistically significant (F=3.89, d.f.=1, 30, P=0.0582). The proportion of the day apportioned to feeding was significantly greater in third instars (32%) than in firsts (12%) (F=39.10, d.f.=1, 30, P=0.0001). There were no other differences in the proportion of the day spent doing key behaviours (drop off, resting, waving up or down, crawling fast or slow). The proportion of the day where larvae could not be seen (i.e. unknown and almost always when hidden amongst terminal leaves) was less than 5% of the total time in both instars.
Fig. 1. Time budgets for behavioural activities of Helicoverpa armigera; n=12 and 19 for first (□) and third (
) instars respectively. Bars represent ±standard errors of the mean. * Indicates significance within each activity at P<0.05 level.
The time spent on a single bout (i.e. its duration) differed between instars (fig. 2). First instar feeding bouts were shorter (Kruskal-Wallis H=7.8306, d.f.=1, 30, P=0.0051) while resting bouts were longer (Kruskal-Wallis H=5.1579, d.f.=1, 30, P=0.0231) than for third instars.
Fig. 2. The mean duration of single bouts in Helicoverpa armigera larvae; n=12 and 19 for first (□) and third () instars respectively. Bars represent ±standard errors of the mean. * Indicates significance within each activity at P<0.05 level.
In the majority of cases there were no significant differences in the amount of time larvae spent on a single visit (a bout) to any particular part of the plant. However, in the case of plant terminals first instar larvae spent more time on a visit than third instars (F=4.37, d.f.=1, 30, P=0.0142). That is, on any given visit they stayed there longer. When considering activities at different plant parts, feeding bouts by third instars were longer on the upper surface of a mature leaf than those of first instars (Kruskal-Wallis H=5.0, d.f.=1, 30, P=0.0253) (fig. 3). The same trend appeared for the lower surface of mature leaves but was not significantly different (Kruskal-Wallis H=2.1429, d.f.=1, 30, P=0.1432). A feeding bout was also longer for third instars when feeding on juvenile leaves (Kruskal-Wallis H=4.0333, d.f.=1, 30, P=0.0446) than for first instars.
Fig. 3. The mean duration of a single feeding bout by Helicoverpa armigera larvae when visiting different plant parts; n=12 and 19 for first (□) and third (
) instars respectively. Bars represent ±standard errors of the mean. * Indicates significance for each position at P<0.05 level.
In addition to the duration of a single bout, the number of times (or frequency with which) larvae went to a site or did a key activity was calculated. The number of times larvae fed and rested differed between instars. Third instars rested more often (Kruskal-Wallis H=8.5130, d.f.=1, 30, P=0.0035) and fed more often (Kruskal-Wallis H=16.2875, d.f.=1, 30, P=0.0001) than first instars (fig. 4).
Fig. 4. The mean frequency of an activity by first and third Helicoverpa armigera instars; n=12 and 19 for first (□) and third () instars respectively. Bars represent ±standard errors of the mean. * Indicates significance within each activity at P<0.05 level.
If a larva was in the first stadium then there was a greater likelihood of it being found in the tip of the plant than a third instar (likelihood ratio G=4.2804, d.f.=1, P=0.0386, based on 12 first and 19 third instars). In a given population there was only a 50% chance of finding a third instar in the tip of the plant compared to 100% of first instar larvae. The only behavioural activity found to be less likely to occur in third instars, compared to first instars, was crawling slowly (likelihood ratio G=14.7775, d.f.=1, P=0.0001 based on 12 first and 19 third instars). The majority (16 out of 19) of third instars were not recorded as doing this activity. This was opposite to first instars where most (10 out 12) were recorded doing this activity.
Initial weights of first instar larvae for all experiments differed very little (all c. 0.06–0.07 mg). There was a decline in relative growth rate (RGR) over time (days) and instar (fig. 5): RGR was significantly highest on the first two days in neonates and lowest in instar III (fig. 5). Relative growth rates did not change for each day within the third instar stadium (fig. 5).
Fig. 5. Relative growth rate (mg−1mg−1day−1) of Helicoverpa armigera larvae by day and instar based on wet weights. Each histogram is the mean (±standard error) for an independent group of larvae for each 24-h period. Sample sizes and letters indicating significant differences at P<0.01 level are shown in the top bar.
Discussion
In the present study the major differences between instar stages were attributed to feeding and moving. Our comparison of first and third instars was based on c. 320 ‘caterpillar hours’ of observation and represents about a third of the time first instars require to complete their stadium; 6 degree days (dd) out of 19 dd (for instar durations see Twine (Reference Twine1978)). We observed third instars for about one-fifth of the time they spend in the instar (6 dd out of 30 dd). We believe our observations are representative of foraging behaviour within the first portion of each stage. We did not observe behaviour associated with moulting at the end of either stage.
Both instars traversed much of the relatively small plants (on average 111 cm2 in area) in the time they were observed and both showed a tendency to move up plants to terminal and juvenile leaves. Larvae varied in the time they spent at different locations but rarely did they stay at one location for the entire observation period. Larvae continued to move, feed and rest throughout the time they were observed.
Caterpillars need to locate suitable feeding sites and consume sufficient food to complete a stage, whilst avoiding being eaten or parasitized. Their behavioural repertoire is limited to activities associated with finding food (crawling and head waving that we associate with searching for and locating a suitable site to feed), feeding, resting and dropping off. Resting may be associated with avoidance of predation or parasitism (staying still) as well as digestion of food. Dropping off was rarely observed and is most likely a predator avoidance mechanism but could provide a means of rapid searching by getting to lower plant parts or moving to another plant (Stamp & Bowers, Reference Stamp and Bowers2000). Although the sequence of these behaviours was basically similar in both instars, the time allocated to some behaviours (such as feeding), and their location differed between the two instars.
There are studies that suggest differences in feeding and foraging behaviour of some heliothine larval stages. Schmidt et al. (Reference Schmidt, Benedict and Walmsley1988) found first instar H. zea on cotton rested more than they fed over a 10-h period. Although no statistical comparison was made between instars, their results suggest that feeding and resting time budgets are different for first instars compared to thirds and fifths. There appears to be a difference on bolling cotton plants in time spent by caterpillars of different stages of H. zea in different locations on plants. First instar larvae were in the terminal and leaf areas more than third instar larvae. Although the study by Schmidt et al. (Reference Schmidt, Benedict and Walmsley1988) was based on a comparison of cultivars rather than instars, our results are similar. Wilson et al. (Reference Wilson, Gutierrez and Leigh1980) showed that the first three larval instars of H. zea tended to move up cotton plants whilst the last two instars moved down. They suggest the movement up was associated with the need to search for fruiting structures, which are more likely to be found on the upper part of the plant. This need decreased with larval age, as larger fruits were more acceptable. Although the cues used by H. armigera to locate suitable feeding/resting sites are not known, first and third instars on vegetative mung beans also moved upward on the plant suggesting this tendency is general in the Heliothineae (Hassan, Reference Hassan1983; Pitre, Reference Pitre1983), and is not necessarily only related to the presence of flowering or fruiting structures.
First instar H. armigera move more frequently than thirds and spend more time on or near the terminal part of the plant. Compared to third instars, first instars move often and rest for long periods (853 versus 626 PTUs). In moving, they traversed all areas of a 14 cm high plant in a relatively short time. They effectively made an average of 19.3 (minimum 7, maximum 37) intra-plant part movements (not including leaf to leaf movement on compound leaves). Either the first instars could not find suitable places to feed for sustained periods and were repeatedly sampling the plant, avoiding induced plant defences, or they are avoiding the potential risk of predation by staying on the move and then resting.
First instar H. armigera feed for less time (only 12% of the time observed and for shorter bouts (351 versus 504 PTUs) than third instars. Feeding in caterpillars may be constrained by the risk of predation (Heinrich, Reference Heinrich, Stamp and Casey1993). Bernays (Reference Bernays1997) found that the risk of predation during feeding was very high and suggested strong selection pressure for rapid food intake in early instars. Small, poorly defended first instars may need to find food quickly, feed rapidly for short bouts and stay still for longer. First instar H. armigera feeding sites are small and rarely cross large veins (personal observation). When first instars reached terminal leaves they tended to stay for longer. Jyoti et al. (Reference Jyoti, Young, Johnson and McNew1999) suggest neonate H. virescens were found more often on the underside of leaves than the upper surface as a means of concealment. First instars did not stay in terminal parts in our study. The need to move from this location may be driven by the behaviour of some predators that move to higher parts of the plant. Neussly & Sterling (Reference Neussly and Sterling1994) showed Geocoris punctipes (Say) (Hemiptera: Lygaeidae) and Pseudatomoscelis seriatus Reuter (Hemiptera: Miridae) spend more time in the higher part of the plant.
Induced localized changes in plant part suitability (Inbar et al., Reference Inbar, Doostdar, Leibee and Mayer1999) may also drive the need to move. Frelichowski & Juvik (Reference Frelichowski and Juvik2001) showed secondary metabolites and trichome density were more likely to affect continuation of feeding in first instars than third instars. Third instars of H. armigera on the other hand are some three times longer than first instars (9–15 mm in length) and about nine times more massive (on average 13.1 vs. 0.7 mg for thirds and firsts respectively). This stage fed more often and for longer periods, crawled less and fed more often in exposed locations on mature older leaves. Third instars are not always hindered by plant chemistry or morphology (but see Van Dam et al., Reference Van Dam, Hermenau and Baldwin2001). They are better defended (spitting behaviour, biting, size) and perhaps can afford the risk of feeding in the open.
The nutritional ecology and relevant efficiencies for early instars of H. armigera have not been determined but may explain some of the behavioural differences found between first and third instars. As predicted, first instars were found to have significantly higher RGR than second or third instars (fig. 5). As with most Lepidoptera the values used below are based in some cases on large instars (value for ECI and RCR) (Scriber & Slansky, Reference Scriber and Slansky1981); but see Barton Browne & Raubenheimer (Reference Barton Browne and Raubenheimer2003) for at least a comparative study of fourth vs. fifth instars.
Helicoverpa armigera have an average egg weight of 0.1 mg and an average first instar weight of 0.7 mg (M.-L. Johnson, unpublished data). This equates to first instars gaining about 0.6 mg in biomass (B) during the stadium. For third instars the equivalent weight gain is about 10.7 mg; based on mean weight for second instars of 2.4 mg (the average of heliothis individuals developing in five and six instars) and 13.1 mg for third instars. Given a developmental period of 19 dd for first instars and 30 dd for third instars (Twine, Reference Twine1978), the proportion of time spent feeding (this study) and assuming an efficiency of conversion of ingested food (ECI) of 0.35 for both instars (see Scriber & Slansky, Reference Scriber and Slansky1981), then first instars would need to consume 0.75 mg of leaf material dd−1 of time spent feeding (0.6/0.35/(19∗0.12)). For third instars this would be 3.18 mg dd−1 (10.7/0.35/(30∗0.32)). Relative to the weight gain during the instar this gives an estimated RCR (relative consumption rate) of 1.1 mg mg−1 dd−1 for first instars and 0.24 mg mg−1 dd−1 for third instars. Assuming equivalent ECI values for each instar, first instars have to consume food at five times the rate of a third instar.
We used a value for ECI that is about the average for Lepidoptera. Assuming first instars had ECI values around the maximum for Lepidoptera (0.6, see fig. 4 in Scriber & Slansky, Reference Scriber and Slansky1981) this would give them a theoretical RCR of around 0.62 mg mg−1 dd−1, or nearly three times the value for third instars. This suggests first instars have to eat fast (a great deal of food in a short space of time). They can achieve this with particular mouthparts (Bernays & Janzen, Reference Bernays and Janzen1988) and by being selective as to where they feed; on softer tissue with higher water and nitrogen content, i.e. terminal parts. This could give them a higher RGR (relative growth rate) (Scriber & Slansky, Reference Scriber and Slansky1981). First instar H. armigera larvae have the mouthparts of a rapid soft tissue feeder, namely incisor/toothed mandibles such as those described in Bernays (Reference Bernays1991). However, third instar H. armigera also have similar toothed mandibles (B.W. Cribb, personal communication). The constraint of predation pressure may well have selected for a feeding apparatus that can shorten food-handling time. Behaviourally first instars do restrict feeding time and forage so as to locate themselves in areas that are both secluded and provide soft, moist tissue that would aid feeding efficiency. Third instars are less selective as to where they feed, spending more time feeding on larger older leaves, and consume a higher proportion of indigestible food.
Although similar behavioural patterns occur between first and third instar H. armigera, there are significant differences in their foraging behaviour. We suggest that studies that infer all heliothine larvae are the same should be interpreted with caution. In addition to the findings of this study, differences between instars may be greater in plants with fruiting structures of varying toughness and with the inclusion of predators that only target early instars.
Acknowledgements
Thanks go to D. Green for expert statistical advice, M. Keller for behavioural matrix software, C. Wiley and F. Rowheder for behavioural observations, and M. Furlong and A. Reeson for helpful comments on the manuscript. W. Rochester provided advice on conversion of data to physiological time units. Helicoverpa mouthpart data was provided by B.W. Cribb.
