The feeding behaviour of Schistocerca gregaria, the desert locust, on ...

2 downloads 679 Views 132KB Size Report
The feeding behaviour of Schistocerca gregaria, the desert locust, on two starch mutants of Arabidopsis thaliana. Geraldine A. Wright1, David Raubenheimer2, ...
Chemoecology 10:59 –67 (2000) 0937 – 7409/00/000059–09 $1.50 +0.20 © Birkha¨user Verlag, Basel, 2000

The feeding behaviour of Schistocerca gregaria, the desert locust, on two starch mutants of Arabidopsis thaliana Geraldine A. Wright1, David Raubenheimer2, Steven Hill3 and Stephen J. Simpson2 1

Dept. Entomology, Ohio State Univ., 1735 Neil Ave., Columbus, OH 43210, USA Dept. Zoology, Oxford University, South Parks Road, Oxford, UK OX1 3PS Dept. Plant Sciences, Oxford University, South Parks Road, Oxford, UK OX1 3RB

2 3

Summary. Schistocerca gregaria, the desert locust, has been shown to regulate its dietary intake with respect to specific macronutrients in synthetic foods. This study examined the nutrients in the leaves of two starch mutants of Arabidopsis thaliana, and then compared the feeding behaviour of locusts on the two starch mutants. The high-starch mutant had c. 25 times more starch than the no-starch mutant. Newly molted 5th stadium locusts were preconditioned for 3 days on one of the mutants, and then observed for 90 min while exposed to the same or the alternative mutant. Locusts pretreated with the no-starch mutant fed longer during the first meal on high-starch mutants, spent more time feeding, and had the smaller latency to begin a meal when compared to the locusts pretreated on the highstarch mutant. The results of the study are interpreted in light of an integrative model of nutrient balancing. Key words. Arabidopsis thaliana – Schistocerca gregaria – nutritional rails – feeding behaviour – starch mutants

Introduction The past four decades have witnessed a considerable proliferation of research into the nutritional strategies of herbivorous insects (Chapman & Joern 1990; Bernays & Chapman 1994; Chapman & de Boer 1995; Schoonhoven et al. 1998; Simpson et al. 1999). Work has traditionally been concentrated largely around two methodological paradigms. On the one hand are studies of insect responses to live plant tissues. These provide the advantage of biochemical and ecological realism, but at the expense of the control that researchers have over biochemical predictor variables. On the other hand, a large number of studies have capitalised on the experimental control available through the use of chemically defined or semi-defined foods, but such studies leave unanswered questions about the generality of responses observed in unnaturally low-dimensional protocols. Correspondence to: D. Raubenheimer, e-mail: david.raubenheimer@ zoology.ox.ac.uk

More recently there have been concerted efforts to focus explicitly on complexity, through studying the interactive effects of food components on herbivorous insects (Bloem et al. 1989; Bloem & Duffy 1990; Slansky & Wheeler 1992; Raubenheimer 1992; Raubenheimer & Simpson 1993; Trumper & Simpson 1993; Stockhoff 1993). For this purpose a geometrical approach has been developed which enables the experimenter to quantify the interactive effects of food components at the levels of behaviour, physiology and insect performance (Raubenheimer & Simpson 1993, 1997, 1999; Simpson & Raubenheimer 1993Simpson & Raubenheimer 1995, 1999). Applications of this approach to date have used synthetic foods to exert absolute control over the levels of the relevant food components, the most thoroughly studied of which are accessible nitrogen and carbohydrate. Monogenic mutations that cause alterations in plant metabolism provide an alternative approach to the issue of nutritional complexity. Using such mutants the chemical composition of living plant tissues can be altered in a controlled manner. In an initial report on such studies, here we present data investigating nutritional responses of desert locusts (Schistocerca gregaria) to Arabidopsis plants which differ markedly in macronutrient composition as a result of monogenic mutations affecting enzymes of starch metabolism. One, a phosphoglucomutase mutant, is unable to synthesise starch (Caspar et al. 1985; Lin et al. 1988) and so contains low levels of digestible carbohydrates in relation to nitrogen (i.e. a high nitrogen : carbohydrate ratio), while the other, a chloroplast transporter mutant, cannot degrade starch and so contains a low nitrogen : carbohydrate ratio (Caspar et al. 1991; Trethewey & ap Rees, 1994). The study was designed to parallel experiments on Locusta migratoria, where it had been shown that pretreatment on foods with suboptimal protein : carbohydrate ratios elicited subsequent selection of foods containing high levels of the deficient nutrient (Simpson et al. 1988, 1990b). The data therefore provide an opportunity to address the question of how representative of real plant tissues are nutrient interactions observed in studies using synthetic foods.

60

G. A. Wright et al.

Materials and methods Insects and plants Locusts (Schistocerca gregaria) were reared in a continuously maintained gregarious culture on seedling wheat and wheat germ at the Department of Zoology, Oxford. For each run of the experiment, 48 newly moulted, 5th instar nymphs were collected from the culture over a period of 6 h and weighed. Seeds of the no-starch mutant pgm (Schulze et al. 1991) and the high starch mutant sex1 (Caspar et al. 1991) of Arabidopsis thaliana (Colombia background) were obtained from Dr. Sam Zeeman (John Innes Center, Norwich). They were planted in general purpose organic potting soil, and grown in a growth chamber which was set at 18–20°C and at an average light level between 80– 100 mmol quanta m − 2 s − 1 for 6–7 weeks. Inflorescences were continuously clipped off before each plant was able to produce flowers to maximise leaf production. Individual plants were collected for chemical analyses from each group of plants grown for each run of the experiment with locusts. There were 10 plants collected from each date (5 of each mutant) for a total of 60 plants.

Plant chemical methods Samples were prepared by freezing all the leaves of each individual plant without roots or influorescences and subsequently freeze drying the samples. Dried samples were ground and extracted in 80% (v/v) ethanol. Quantities of foliar sucrose, glucose, fructose, and starch were measured enzymatically as in Jones et al. (1977). Nitrogen was measured using the micro Kjeldahl method (Allen 1974). Free amino acids were measured using an HPLC method (Zanotto et al. 1996) after isolation from the extraction on a cation exchange column (Dowex 50w-X8( +H)). Glucosinolates were measured as in Magrath et al. (1993) by HPLC.

CHEMOECOLOGY al. (1993). The frass from each insect was also collected, dried and weighed. This experiment was run twice, using different sets of 48 insects. Data analysis ANOVA was performed for all the plant variables with mutant type and date of collection as factors. An estimate of consumption during the pretreatment period (frass dry wt) and measures of insect growth over the whole experiment (fat gained, growth after lipid extraction, and percentage water content) were compared between groups fed the different mutants using a Mann-Whitney t-test. These measures essentially represent the effect of the 3 day pretreatment on the body composition (performance) of the locusts, given that the test was only 90 min in duration. The amount of time spent feeding in the first meal of the test period and meal duration were calculated, and a bout criterion of 3 min and a meal criterion of 50 s were used to define meals from intermeal intervals (Simpson 1990). These values had been previously derived for S. gregaria from other detailed observations of its feeding behaviour (Raubenheimer & Simpson 1990; Simpson 1995). The total amount of time spent feeding, latency to feed and the proportion of the first meal spent feeding (time spent feeding during first meal divided by the total length of the first meal, which includes pauses) were also calculated. ANOVA was used to determine the effects of the pretreatment, test-treatment and their interaction on all of the behaviour variables during the 90 min observation. Experiment number and sex were also used as factors in the ANOVA but have been excluded from the results and discussion section. Experiment did have a significant main effect, with all of the variables being consistently higher in the first experiment than the second. There were, however, no significant interactions involving experiment, indicating that the results of the pretreatment on the observation treatment was similar for both experiments. This justified grouping the data from both experiments for the analyses. Sex had no effect on any of the behaviours, either as a main effect or in any interaction.

Experimental procedure After collection from the culture, each locust was placed in a plastic box (28×16 ×9 cm) with an individual plant of one of the starch mutants of Arabidopsis in a constant temperature room kept at 29 – 30°C. A hole was drilled in the bottom of each box, and the plant pot was sunk through the hole so that the foliage was at ground level relative to the locust. The locusts were pretreated with one or other of the starch mutants for three days. The plants were replaced twice daily, once between 10–11 am and once at 5 – 6 pm, after the first day. Following the pretreatment period, the locusts were observed in detail during a 90 min test period on the 4th day of the stadium. Sixteen of the animals were selected for each observation period. Half of the locusts in the observation were given the same mutant type as the pretreatment, and half were given the other. The pretreatment plant from the previous day was removed. Each locust was isolated from the plant in its box by upturning another, smaller (17 ×12 ×6 cm) plastic box and placing it over the locust. The pretreatment plant was removed, and the test plant was placed in the cage. After all 16 of the locusts and plants were prepared, the plastic box which was isolating each locust in its cage was moved gently to cover each plant as well as the locust, and the 90 min observation began. The locusts were monitored continuously, and their behaviour was recorded at 10 s intervals on a laptop computer programmed as an event recorder. One of three types of behaviour was recorded at each interval: feeding; contacting the plant; and quiescence or moving (i.e., anything other than feeding or contacting the plant). The observation was terminated after 90 min. Insects were weighed, killed by freezing on the 4th day after the observation and dried to constant weight then re-weighed. Total percent body water and lipids were determined as in Zanotto et

Results Plant chemical characteristics Table 1 shows the ANOVA results of the betweenstrain comparison of tissue nutrient concentration, while means and standard errors are presented in Table 2. The no-starch mutant (pgm) had significantly more fructose, glucose, nitrogen (minus free amino acids), and water and less starch and total carbohydrates than the high-starch mutant sex1 (Table 1). Most important, the nitrogen : carbohydrate ratio in the pgm mutant was 4 times that in sex1. Fructose, free amino acids, and nitrogen, and the N/C ratio differed for both mutants across sampling dates. There was no significant date by mutant interaction for any of the variables measured; in other words, the difference between mutants was unaffected by sampling dates. Glucosinolate quantity and quality did not vary with date or mutant type (Table 3) with the exception of 4-methoxy-3-indolylmethyl, which had a mutant by date interaction. The mean range of glucosinolate concentration for each identified compound was between 0.001 and 0.013 mmol/g dry wt. The glucosinolates that were identified in the samples were: 3-methylsulphinylpropyl, 4-methylsulphinylbutyl, 4-methylthiobutyl, 3-in-

Vol. 10, 2000

The feeding behaviour of Schistocerca gregaria, the desert locust, on two starch mutants of Arabidopsis thaliana

Mutant

Fructose Glucose Sucrose Starch Nitrogen Free Amino Acids Total Carbohydrates Ratio N/C

Date

Mutant×Date

df

F

df

F

df

F

1 1 1 1 1 1 1 1

17.7*** 5.55* 0.450 884*** 28.4*** 0.02 27.2*** 183.4***

4 4 4 4 4 4 4 4

17.5*** 1.66 0.140 0.520 8.08** 20.2*** 2.14 3.10*

4 4 4 4 4 4 4 4

2.93 2.09 0.04 0.76 0.73 0.02 0.86 0.30

61

Table 1 ANOVA of the effects of mutant and date of collection on the nutrient content of Arabidopsis plants

*=PB0.05; ** = PB0.01; *** =PB0.001

Table 2 Means ( 9 SE) of measured nutrients in the foliage of two starch mutants of Arabidopsis thaliana No-starch mutant Fructose 12.6 90.6 Glucose 10.5 90.9 Sucrose 1.67 90.4 Starch 1.88 90.3 Total Carbohydrates 26.6 92 % Water 92.0 90.1 Nitrogen 38.3 92 Free Amino Acids 0.220 90.1 Ratio N/C 1.5 : 1

Locust beha6iour during the obser6ation Figure 1a shows that locusts pretreated on no-starch plants tended to take longer first meals during the test period than those pretreated on high starch plants. However, first meals were particularly long for animals pretreated on no-starch foods then transferred to high-starch foods during the test period. This effect was statistically significant, as evidenced by the pretreatment by test food interaction term in Table 5. By contrast, a comparison of the proportion of time feeding within the first meal revealed no significant effects (Fig. 1b, Table 5). This suggests that the major influence on the parameters of the first meal involved its duration, rather than the extent of feeding within the meal. Latency to feed was significantly shorter for the locusts pretreated with the no-starch mutant (Fig. 2, Table 5). There was also a tendency for latency to be shorter when the test food contained starch, but this effect did not attain statistical significance. By contrast with the duration of the first meal, there was no suggestion of a pretreatment by test food interaction for latency to feed. Total time observed feeding during the 90 min observation period was significantly greater for locusts pretreated on no-starch mutants (Fig. 3, Table 5). There was no suggestion of an effect of test food, not an interaction between pretreatment and test food.

High-starch mutant 8.76 9 1 7.68 9 0.9 2.419 0.4 49.59 1 68.3 9 2 88.2 9 0.1 24.7 9 1 0.2109 0.1 0.4 : 1

All units in mg/g dry wt. except for % water. (% Water was calculated using different samples from all the other values represented in the table)

dolylmethyl, 4-methoxy-3-indolylmethyl, 1-methoxy-3indolylmethyl, and 3-butenyl. Effects of pretreatment on locust performance Of the 4 variables which reflected consumption (frass production) and performance (body fat, non-lipid body weight, and percentage body water) during the pretreatment, two were significantly different between the pretreatments (Table 4). Locusts pretreated with the high-starch mutant had more body fat than the locusts pretreated with the no-starch mutant. Highstarch treated locusts also had a smaller percentage of body weight attributable to water compared to the no-starch treated locusts.

Mutant

3-methylsulphinylpropyl 4-methylsulphinylbutyl 4-methiobutyl 3-indolylmethyl 4-methyoxy-3-indolylmethyl 1-methoxy-3-indolylmethyl 3-butenyl

Date

Mutant×Date

df

F

df

F

df

F

4 4 4 4 4 4 4

1.80 1.65 1.27 0.440 0.050 0.300 0.680

4 4 4 4 4 4 4

1.80 0.640 3.10 0.580 0.50 0.240 0.520

4 4 4 4 4 4 4

1.81 1.12 3.71 0.930 9.40* 2.74 0.520

*= PB0.05; ** = PB0.01; *** = PB0.001

Table 3 ANOVA of Arabidopsis glucosinolates by mutant and date of collection. Glucosinolates were measured in only the first and third sampling dates for Arabidopsis

62

G. A. Wright et al.

CHEMOECOLOGY

Table 4 Mann-Whitney comparison of means 9 S.E. (g) of the effects on S. gregaria of feeding on one of two starch mutants of Arabidopsis during the pretreatment period

Frass (dry wt) Lipid wt. (lipid extracted) Lipid-free dry wt. % Water*

No-starch mutant

High-starch mutant

P-value

0.139 9 0.007 0.00190.013

0.1209 0.009 0.0089 0.011

0.185 B0.001

0.166 9 0.006 78.9 90.400

0.1629 0.006 77.39 0.400

0.661 0.020

* calculated as a percentage of dry weight, including body lipids

Discussion Biochemistry of Arabidopsis starch mutants While the two mutants of Arabidopsis used in this experiment were originally isolated to study starch biosynthesis pathways in plants (Caspar et al. 1985; Lin et al. 1988; Caspar et al. 1991; Trethewey & ap

Fig. 1 a) Duration of first meal taken by S. gregaria when transferred from a pretreatment in which they were fed one of two starch mutants of Arabidopsis to an observation period in which they were given the same or the alternate mutant. Notation on the x-axis is pretreatment food : observation food b) Proportion of first meal spent feeding

Rees 1994), our analyses of the chemical composition of their tissues demonstrated that they were well suited to the current purposes. Namely, they provided sources of plant tissue which differed in key nutritional variables and were otherwise genetically identical. Indeed, in some ways the nutritional differences between the mutants were more extreme than anticipated. For instance, we detected values for starch in the high-starch plant which were almost 50 times the quantity reported by Schulze et al. (1991), and Trethewey & ap Rees (1994). This difference might be due to the fact that plants sampled in this study were almost a month older than the ones sampled by other researchers, as starch levels are known to increase with leaf age (Goldschmidt & Huber 1992). In terms of total carbohydrate content, the high levels of starch in sex1 were partly compensated by higher levels of fructose and glucose in the no-starch pgm mutant (Table 2), presumably due to the inability of this mutant to convert these sugars to starch. However, the magnitude of these differences was trivial compared with the differences in starch content, with the result that the high-starch mutant had total digestible carbohydrate levels which were almost three times higher than the no-starch mutant. An important consequence of these differences in starch levels, is that the ratio of nitrogen to carbohydrate differed appreciably between the mutants. The no-starch mutant had a significantly higher concentration of nitrogen than the high-starch mutant (Table 2), partly due to the fact that the high starch levels in the one mutant diluted the concentration of foliar nitrogen per gram dry weight. However, if the percentage of nitrogen is calculated with respect to wet weight, the mutants had equal concentrations of nitrogen, and a four fold difference in carbohydrate (pgm: 0.6% nitrogen, 0.24% carbohydrate; sex1: 0.6% nitrogen, 0.84% carbohydrate). On a dry weight basis, the no-starch mutant had a nitrogen : carbohydrate ratio some 4 times lower than the other mutant (1.5 : 1 vs. 0.4 : 1). Research has shown that individual amino acids affect insect feeding behaviour (Simpson et al. 1990a; Bernays & Chapman 1994), so it is potentially important to identify and quantify the free amino acids present in a plant which might affect feeding behaviour (Chapman 1995). Our analyses revealed that all amino acids reported to be necessary and stimulatory for locusts (Bernays & Woodhead 1984; Simpson et al. 1990a; Bernays & Chapman 1994; Chapman 1995) were present in both mutants, and did not differ in concentration between the mutants. We also measured glucosinolates, secondary metabolites which are considered to be toxic to nonspecialist herbivores and fungi (Louda & Mole 1991; Harborne 1993). Glucosinolate composition and concentration were not different between the two mutants (Table 3) aside from 4-methoxy-3-indolylmethyl. All of the glucosinolates in the starch mutants were the same as those reported by Hogge et al. (1988) and Haughn et al. (1991) for the Colombia wildtype,

Vol. 10, 2000

The feeding behaviour of Schistocerca gregaria, the desert locust, on two starch mutants of Arabidopsis thaliana

First meal duration

Latency

Total time feeding

Prop. of time feeding in the first meal

Source

df

F

df

F

df

F

df

F

Main effects Pretreatment (P) Test treatment (O) Experiment (E)

1 1 1

11.1*** 3.47 29.9***

1 1 1

7.89** 3.26 0.21

1 1 1

16.55*** 0.00 18.77***

1 1 1

0.49 0.05 4.30*

Interactions P×E O×E P×O P×O×E

1 1 1 1

0.32 0.65 6.28** 0.70

1 1 1 1

0.14 0.13 0.02 0.97

1 1 1 1

1 1 1 1

0.04 1.33 1.10 1.01

1.61 0.19 1.16 0.65

63

Table 5 ANOVA of pretreatment effects by test treatment effects

*= PB0.05; ** = PB0.01; *** = PB0.001

except that the starch mutants had 3-butenyl. The wildtype was reported to have 5-methylsulphinylpentyl and 8-methylsulphinyloctyl; neither of these compounds was found in either starch mutant.

Overall, the chemical analyses of plant composition suggest that the locusts in our experiments were confronted by appreciable differences in the proportion of the key macronutrients nitrogen and carbohydrates, and the plants did not differ appreciably in the other components we measured. This difference was reflected in locust growth, since those pretreated on the high starch mutant had significantly greater body lipid content than those fed the no-starch mutant (Table 4). Similar results have been obtained for L. migratoria fed synthetic foods containing a ratio of protein : carbohydrate which is lower than the optimal dietary balance for this species (Raubenheimer 1992; Raubenheimer & Simpson 1993, 1997), and also for S. gregaria (D. Raubenheimer and S. J. Simpson, in prep.). In addition to greater body lipids, the highstarch pretreated insects retained less water in their

tissues than the no-starch pretreatment group. A negative correlation has been observed between body fat and water content in several insect and vertebrate species (Clarebrough et al. 1999); this quite possibly represents a tradeoff in the allocation of cellular and extracellular space to these body components. There are indications, however, that in both treatment groups fat levels were lower than optimal. The locusts pretreated on the high-starch mutant had 5% body fat, and the locusts pretreated on the no-starch mutant had only 1% body fat. Simpson (1982a) calculated that Locusta migratoria fed seedling wheat and wheat germ had 16% body fat at the beginning of its 5th stadium. While we are not aware of equivalent body fat measures of S. gregaria at the beginning of the 5th stadium, D. Raubenheimer and S. J. Simpson (in prep.) found that newly moulted adult L. migratoria and S. gregaria allowed to select from synthetic foods the preferred balance of protein : carbohydrate throughout the 5th stadium both had body lipid content of approximately 15%. The suggestion is thus that both groups in the current experiments lost body fat, and the loss was significantly greater in those pretreated on no-starch mutants. This indicates that both Arabidopsis

Fig. 2 Latency to feed from the beginning of the observation period

Fig. 3

The effects of pretreatment on locust performance

Total time spent feeding during the observation period

64

G. A. Wright et al.

Fig. 4 Hypothetical nutritional rails: a) Example 1, b) Example 2, c) Example 3. See text for further explanation

mutants were dilute in their carbohydrate concentrations as compared to studies using artificial foods.

Locust beha6iour during the obser6ation Interpreting the behavioural response exhibited in the observation period can only be done in the context of the nutritional composition of the two Arabidopsis mutants. The geometric framework (Raubenheimer &

CHEMOECOLOGY

Simpson 1993, 1997; Simpson & Raubenheimer 1993b) provides a means of generating hypotheses within this experimental design that can be compared with the results of the experiment. Previous applications of this approach to the feeding behaviour of S. gregaria have demonstrated that these insects will balance their intake of nutrients so that a deficit of protein will equally balance an excess of carbohydrate, and vice versa (Raubenheimer & Simpson 1997). The geometric framework has also illustrated that S. gregaria has an intake target of 89 mg of protein to 125 mg of carbohydrate during the first 3 days of its 5th stadium. Using such models to interpret the behaviour of the locusts in the experiment can be illustrated by designing hypothetical nutritional rails (sensu Raubenheimer & Simpson 1993). Three such hypothetical examples are shown in Fig. 4. In Example 1 (Fig. 4a), the two pretreatment foods are nutritionally complementary with respect to protein and carbohydrate (Raubenheimer & Simpson 1993, 1997; Simpson & Raubenheimer 1993b; Chambers et al. 1995). These foods are the same distance on either side of the intake target rail. In this case, locusts which were pretreated with one food and then given the other in a test treatment (F1 : F2 and F2 : F1) should eat much more of the test food than the control group that was given the same food in the test that they had in the pretreatment (F1 : F1 and F2 : F2) (Fig. 5a). A result closely similar to this was found in the artificial food study by Simpson et al. (1988) on L. migratoria nymphs. A locust which had been pretreated on the high protein-low carbohydrate food has lower sugar levels in its haemolymph resulting from a shortage of carbohydrate in its food and high amino acid levels, while the reverse is the case for locusts fed the high carbohydrate-low protein food. Haemolymph levels of sugars and amino acids translate into complementary selection behaviour, at least in part due to changes in gustatory receptiveness (Simpson et al. 1990a, 1991; Simpson & Raubenheimer 1993a; Zanotto et al. 1996). Example 2 (Fig. 4b) also shows two foods which lie on either side of the intake target, but they are asymmetrically positioned such that Food 2 is much closer to the intake target ratio than Food 1. Here, the locusts would be expected to eat more of the opposite food when offered it following pretreatment, relative to insects kept on the same food type, but the locusts pretreated with the more unbalanced food (F1) would be expected to eat more of F2 during the observation period than those pretreated on F2 would eat of F1 (Fig. 5b). Example 3 (Fig. 4c) is a hypothetical situation where both foods are imbalanced on the same side of the intake target. They are nutritionally non-complementary. The locusts fed with F1 during the pretreatment would be expected to eat more of F2 in the test treatment, but the same is not true of those pretreated with F2 and then given F1. Because F1 is

Vol. 10, 2000

The feeding behaviour of Schistocerca gregaria, the desert locust, on two starch mutants of Arabidopsis thaliana

even further away than is F2 from being optimal, insects should eat a smaller meal of F1 after pretreatment on F2 than if kept on F2 (Fig. 5c). The locusts’ behaviour on the plants in the Arabidopsis experiment can be compared to predictions of their behaviour made by the above hypothetical models. Fig. 6a shows how the mutants appear as nutritional rails for estimated protein (%N× 6.25, Allen 1974) and total carbohydrate. When compared to the hypothetical examples, the rails are most similar to Example 3 (Fig. 4c). However, the behaviour of the locusts on the starch mutants during their first meal (Fig. 1a) is most similar to Example 2 (Fig. 5b), and not to what was predicted for Example 3 (Fig. 5c). In both

Fig. 5 Hypothetical feeding behaviour for nutritional rails: a) Example 1, b) Example 2, c) Example 3. See text for further explanation

65

treatments where locusts were given the other mutant during the observation period, they took longer first meals than did the control treatment locusts, but the locusts pretreated with the no-starch mutant and then given the high-starch mutant took considerably longer meals than their control group kept on the no-starch mutant (Table 5, Fig. 1a). Locusts switched from the high-starch to the no-starch mutant showed a far less pronounced increase in first meal duration relative to those kept on the no-starch mutant: there was no indication of a decrease in meal duration as expected from Example 3 above (Fig. 5c). Why is there a discrepancy between the actual behavioral outcome and that expected given the nutritional composition of the two mutants? There are two possible explanations: either some critical factor has been omitted from the models (i.e. there is a critical hidden dimension – Simpson & Raubenheimer 1995), or else the nutritional rails for the plants are misrepresented. While the former is unlikely given the range of variables measured in our study, it is quite possible that the nutritional rails for the plants have been misrepresented. The assumption when calculating the rails was that the nutrients in the plants were equally utilizable by the locusts as the nutrients found in artificial foods. While it is almost certain that the soluble carbohydrates and the starch measured in Arabidopsis were readily utilizable by the locusts, the same is not true of the protein. Protein was estimated from percentage nitrogen using micro Kjeldahl. The Kjeldahl method measures organic sources of nitrogen (Allen 1974). However, not all the nitrogen that was measured may have been utilizable by the locusts, so the estimation of protein based on %N may be an overestimation of the protein available. In the earlier experiments using artificial foods, the protein source was a mixture of casein, peptone, and albumen, which has been shown to be highly utilizable by locusts (Zanotto et al. 1993; Simpson & Raubenheimer 1995). This is in contrast to other studies of insects feeding on their host plants, which have shown that the nitrogen in plant material is utilized at anywhere from 12 to 60% efficiency, depending on which insect species was feeding on which plant (Waldbauer 1968). Slansky & Feeny (1977) showed that Pieris rapae utilized nitrogen from different host plants with efficiencies of 30–56% depending on the plant species. L. migratoria was found to have a nitrogen utilization efficiency (NUE) of approximately 50% when feeding on highly nutritious wheat seedlings (Simpson 1982a). The index used as a measure of utilizable nitrogen (NUE) is calculated as the amount of weight gained due to nitrogen divided by the amount of nitrogen present in the food ingested (Waldbauer 1968; Slansky & Feeny 1977). To find out whether all the nitrogen available in the Arabidopsis was actually utilized by the locusts, an estimate of NUE was calculated. This involved making some assumptions about variables that were not measured in the present experiment, based on previous studies of locusts. Locust nitrogen growth was estimated from final carcass nitro-

66

G. A. Wright et al.

CHEMOECOLOGY

emphasised by our geometric framework, is the relationship between utilizable nutrients in the food and the current nutrient requirements (intake target) of the insect.

References

Fig. 6 Nutritional rails for starch mutants of Arabidopsis: a) Rails derived from measured nutrients, b) Rails for starch mutants revised to take into account estimated utilization efficiencies of S. gregaria

gen and an estimate of initial carcass nitrogen based upon assumptions that carcass water, lipid and nitrogen contents at the start of the stadium were 80, 18 and 14%, respectively (Simpson 1982b; Zanotto et al. 1993). The amount consumed of Arabidopsis leaf material was estimated from frass produced by assuming that 75% of the leaves of Arabidopsis is undigestible materials (cellulose and lignins), based on a measured mean of 25% dry weight being starch, sugar and nitrogenous compounds. Based on these calculations, NUE was estimated to be in the region of 30%, providing re-calculated ratios of 2.7 : 1 utilizable protein to carbohydrate for the no-starch mutant and 0.7 : 1 usable protein to carbohydrate for the high-starch mutant (Fig. 6b). These ‘‘corrected’’ nutrient rails for the mutants are considerably more similar to Example 2 (Fig. 4b) than Example 3 (Fig. 4c). Using these recalculalated ratios, the behaviour of the locusts is in accordance with what was predicted by the models in Fig. 4b and 5b. Overall, these data demonstrate that similar phenomena are observed in insect responses to nutritionally manipulated plant tissue as in experiments using synthetic foods. They also demonstrate, however, that insect responses may not be interpretable on the basis of food nutrient content alone. The key variable, as

Allen SE (1974) Chemical Analysis of Ecological Materials. GB-Oxford: Blackwell Bernays EA and Chapman RF (1994) Host Plant Selection by Phytophagous Insects. New York: Chapman and Hall Bernays EA, Woodhead S (1984) The need for high levels of phenylalanine in the diet of Schistocerca gregaria nymphs. J Ins Phys 30:489 – 493 Bloem KA, Duffy SS (1990) Interactive effect of protein and rutin on larval Heliothis zea and the endoparasitoid Hyposoter exiguae. Entomol exp app 54:149 – 160 Bloem KA, Kelley KC, Duffey SS (1989) Differential effect of tomatine and its alleviation by cholesterol on larval growth and efficiency of food utilization in Heliothis zea and Spodoptera exigua. J Chem Ecol 15:387 – 398 Caspar T, Huber SC, Somerville CR (1985) Alterations in growth, photosynthesis, and respiration in a starchless mutant of Arabidopsis thaliana deficient in chloroplast phosphoglucomutase activity. Plant Phys 79:11 – 17 Caspar T, Lin TP, Kakefuda G, Benbow L, Preiss J, Somerville CR (1991) Mutants of Arabidopsis with altered regulation of starch degradation. Plant Phys 95:1181 – 1188 Chambers PG, Simpson SJ, Raubenheimer D (1995) A behavioural analysis of the defense of nutrient intake in Locusta migratoria nymphs. Anim Beh 50:1513 – 1523 Chapman RF (1995) Chemosensory regulation of feeding. Pp 101– 136 in Chapman RF, de Boer G (eds) Regulatory Mechanisms in Insect Feeding. New York: Chapman and Hall Chapman RF and de Boer G (eds) (1995) Regulatory Mechanisms in Insect Feeding. Chapman & Hall, New York Chapman RF and Joern T (eds) (1990) Biology of Grasshoppers. New York: John Wiley & Sons Clarebrough C, Mira A, Raubenheimer D (1999) Sex-specific differences in nitrogen intake and investment by feral and laboratory – cultured cockroaches. J Ins Phys 46:677 – 684 Goldschmidt E, Huber T (1992) Regulation of photosynthesis by end-product accumulation in leaves of plants storing starch, sucrose, and hexose sugars. Plant Phys 99:1443 – 1448 Harborne JB (1993) Introduction to Ecological Biochemistry. GBLondon: Academic Press Haughn GW, Davin L, Giblin M, Underhill EW (1991) Biochemical genetics of plant secondary metabolites in Arabidopsis thaliana: the glucosinolates. Plant Phys 97:217 – 226 Hogge LR, Reed DW, Underhill EW, Haughn GW (1988) HPLC separation of glucosinolates from leaves and seeds of Arabidopsis thaliana and their identification using thermospray liquid chromatography-mass spectrometry. J Chromat Sci 26:552 – 556 Jones MG, Outlaw Jr. WH, Lowry OH (1977) Enzymic assay of 10 − 7 to 10 − 14 moles of sucrose in plant tissues. Plant Phys 60:379 – 383 Lin TP, Caspar T, Somerville CR, Preiss J (1988) Isolation and characterization of starchless mutant of Arabidopsis thaliana lacking ADPglucose pyrophosphorylase activity. Plant Phys 86:1131 – 1135 Louda S and Mole S (1991) Glucosinolates: chemistry and ecology. Pp 123 – 164 in Rosenthal G, Berenbaum M (eds) Herbivores: Their Interactions with Secondary Plant Metabolites. GB-London: Academic Press, 2nd ed. vol. 1 Magrath R, Herron C, Giamoustaris A, Mithen R (1993) The inheiritance of aliphatic glucosinolates in Brassica napus. Plant Breeding 111:55 – 72 Raubenheimer D (1992) Tannic acid, protein, and digestible carbohydrate: dietary imbalance and nutritional compensation in locusts. Ecol 73:1012 – 1027

Vol. 10, 2000

The feeding behaviour of Schistocerca gregaria, the desert locust, on two starch mutants of Arabidopsis thaliana

Raubenheimer D, Simpson SJ (1990) The effects of simulatenous variation in protein, digestible carbohydrate and tannic acid on the feeding behaviour of larval L. migratoria and S. gregaria. I. Short term studies. Phys Entomol 15:219–223 Raubenheimer D, Simpson SJ (1993) The geometry of compensatory feeding in the locust. Anim Beh 45:953–964 Raubenheimer D, Simpson SJ (1997) Integrative models of nutrient balancing: application to insects and vertebrates. Nutrit Res Rev 10:151 – 179 Raubenheimer D, Simpson SJ (1999) Integrating nutrition: a geometrical approach. Entomol exp app 91:67–82 Schoonhoven LM, Jermy T and van Loon JJA (1998) Insect-Plant Biology: From Physiology to Evolution. New York: Chapman and Hall Schulze W, Stitt M, Schulze ED, Neuhaus H, Fichtner K (1991) A quantification of the significance of assimilatory starch for growth of Arabidopsis thaliana. Plant Phys 95:890–895 Simpson CL, Simpson SJ, Abisgold JD (1990a) The role of various amino acids in the protein compensatory response of Locusta migratoria. Symp Biol Hung 39:39–46 Simpson SJ (1982a) Changes in the efficiency of utilization of food throughout the fifth-instar nymphs of Locusta migratoria. Entomol exp app 31:265–275 Simpson SJ (1982b) The Control of Food Intake in 5th instar Locusta migratoria nymphs. PhD. thesis, Univ. of London Simpson SJ (1990) The pattern of feeding. Pp 73–104 in Chapman RF, Joern A (eds) Biology of Grasshoppers. New York: John Wiley and Sons Simpson SJ (1995) Regulation of a meal: chewing insects. Pp 137 – 156 in Chapman RF, de Boer G (eds) Regulatory Mechanisms in Insect Feeding. New York: Chapman and Hall Simpson SJ, Mordue JA and Hardie J (eds) (1999) Proceedings of the 10th International Symposium on Insect-Plant Relationships. Dordrecht: Kluwer Academic Publishers Simpson SJ, Raubenheimer D (1993a) A multi-level analysis of feeding behavior: the geometry of nutritional decisions. Phil Trans Roy Soc London B 342:381–402 Simpson SJ, Raubenheimer D (1993b) The central role of the haemolymph in the regulation of nutrient intake in insects. Phys Entomol 18:395 –403 Received 16 November 1999; accepted 23 December 1999.

.

67

Simpson SJ, Raubenheimer D (1995) The geometric analysis of feeding and nutrition: a user’s guide. J Ins Phys 41:545–553 Simpson SJ and Raubenheimer D (1999) Geometric models of macronutrient selection. Pp 29 – 42 in Berthoud H-R, Seeley RJ (eds) Neural Control of Macronutrient Selection. Boca Raton: CRC Press Simpson SJ, James S, Simmonds MSJ, Blaney WM (1991) Variation in chemosensitivity and the control of dietary selection behaviour in the locust. Appetite 17:141 – 154 Simpson SJ, Simmonds MSJ, Blaney WM (1988) A comparison of dietary selection behaviour in larval Locusta migratoria and Spodoptera littoralis. Phys Entomol 13:325 – 332 Simpson SJ, Simmonds MSJ, Blaney WM (1990b) Compensatory dietary selection occurs in larval Locusta migratoria but not Spodoptera littoralis after a single deficient meal during ad libitum feeding. Phys Entomol 15:235 – 242 Slansky F, Feeny PP (1977) Stabilization of the rate of nitrogen accumulation by larvae of the cabbage butterfly on wild and cultivated food plants. Ecol Monog 47:209 – 228 Slansky F, Wheeler S (1992) Caterpillars’ compensatory feeding response to diluted nutrients leads to toxic allelochemical dose. Entomol exp app 65:171 – 186 Stockhoff BA (1993) Ontogenetic changes in dietary selection for protein and lipid by gypsy moth larvae. Phys Entomol 39:677– 686 Trethewey RN, ap Rees T (1994) A mutant of Arabidopsis thaliana lacking the ability to transport glucose across the chloroplast envelope. Biochem J 301:449 – 454 Trumper S, Simpson SJ (1993) Regulation of salt intake by nymphs of Locusta migratoria. J Ins Phys 39:857 – 864 Waldbauer GP (1968) The consumption and utilization of food by insects. Adv Ins Phys 5:229 – 288 Zanotto FP, Raubenheimer D, Simpson SJ (1996) Haemolymph amino acid and sugar levels in locusts fed nutritionally unbalanced diets. J Comp Physiol B 166:223 – 229 Zanotto FP, Simpson SJ, Raubenheimer D (1993) The regulation of growth by locusts through post-ingestive compensation for variation in the levels of dietary protein and carbohydrate. Phys Ent 18:425 – 434