SEARCH

SEARCH BY CITATION

Keywords:

  • caloric restriction;
  • geometric framework;
  • longevity;
  • nutrition;
  • reproduction

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Caloric restriction (CR) has been widely accepted as a mechanism explaining increased lifespan (LS) in organisms subjected to dietary restriction (DR), but recent studies investigating the role of nutrients have challenged the role of CR in extending longevity. Fuelling this debate is the difficulty in experimentally disentangling CR and nutrient effects due to compensatory feeding (CF) behaviour. We quantified CF by measuring the volume of solution imbibed and determined how calories and nutrients influenced LS and fecundity in unmated females of the Queensland fruit fly, Bactocera tryoni (Diptera: Tephritidae). We restricted flies to one of 28 diets varying in carbohydrate:protein (C:P) ratios and concentrations. On imbalanced diets, flies overcame dietary dilutions, consuming similar caloric intakes for most dilutions. The response surface for LS revealed that increasing C:P ratio while keeping calories constant extended LS, with the maximum LS along C:P ratio of 21:1. In general, LS was reduced as caloric intake decreased. Lifetime egg production was maximized at a C:P ratio of 3:1. When given a choice of separate sucrose and yeast solutions, each at one of five concentrations (yielding 25 choice treatments), flies regulated their nutrient intake to match C:P ratio of 3:1. Our results (i) demonstrate that CF can overcome dietary dilutions; (ii) reveal difficulties with methods presenting fixed amounts of liquid diet; (iii) illustrate the need to measure intake to account for CF in DR studies and (iv) highlight nutrients rather than CR as a dominant influence on LS.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Extension of lifespan (LS) due to modest dietary restriction (DR) is a widely acknowledged biological phenomenon, having been empirically shown in a diverse array of organisms ranging from yeasts to primates (reviewed in Masoro, 2002, 2005; Partridge & Brand, 2005). Effects of DR on LS have mostly been attributed to caloric restriction (CR) (Masoro, 2005; Partridge & Brand, 2005; Sinclair, 2005). However, recent research has instead highlighted intake of particular nutrients as a compelling alternative mechanism that might explain DR’s effects on LS.

Recent studies suggest that specific nutrients, notably protein and certain amino acids, can play a pivotal role in extending longevity (Chippindale et al., 1993; Mair et al., 2005; Miller et al., 2005; Piper et al., 2005; Ayala et al., 2007; Carey et al., 2008; Lee et al., 2008; Maklakov et al., 2008). This nutritional view suggests that specific nutrient effects are a significant mechanism underpinning the LS extension associated with DR (Kapahi et al., 2004; Piper et al., 2005; Ayala et al., 2007; Kassi & Papavassiliou, 2008; Lee et al., 2008).

Distinguishing between CR and nutrient effects can be difficult because (i) dilutions of diet confound calories and nutrients in experimental designs that use a standard diet composition (e.g. sucrose to yeast ratio of 3:1) (Piper et al., 2005; Simpson & Raubenheimer, 2007) and (ii) compensatory feeding (CF) responses can mask the effects of calories and nutrients on LS in experimental designs that do not measure intake (Carvalho et al., 2005; Simpson & Raubenheimer, 2007; Piper & Bartke, 2008). CF can be particularly problematic as animals may adjust their feeding behaviour not only to diet concentration, but also to the nutrient composition of the diet. Several studies have shown that insects increase their consumption in response to dietary dilution (aphids, Douglas et al., 2006; Drosophila, Carvalho et al., 2005; Lee et al., 2008; locusts, Raubenheimer & Simpson, 1993). Additionally, insects may vary markedly in their nutritional requirements among species, among individual members of a species, and even through an individual’s lifetime (e.g. locusts, Raubenheimer & Simpson, 1993; Drosophila, Edgecomb et al., 1994).

Simpson & Raubenheimer (2007) proposed the Geometric Framework (GF) as a methodological tool to disentangle caloric and nutritional effects on LS. The GF views feeding as a multivariate activity, whereby an organism ingests not just energy, but nutrients (e.g. protein, carbohydrates and lipids) (Raubenheimer & Simpson, 1999). The GF emphasizes nutritional regulatory responses and measures the performance consequences of dietary intake. Implementation of the GF requires sampling of points throughout the n-dimensional dietary space (e.g. across diets varying in both nutrient composition and concentration) (Raubenheimer & Simpson, 1999; Simpson & Raubenheimer, 2007). Using the GF, Lee et al. (2008) and Maklakov et al. (2008) have recently shown that nutrient intake ratios affect LS (higher proportion of carbohydrate increased LS) in Drosophila and field crickets, respectively, but reduced caloric intake does not.

Tephritid fruit flies are an excellent system to implement the GF approach. First, the effect of DR on LS extension remains inconclusive. Medflies (Ceratitis capitata) show no increase in LS with diet dilutions (Carey et al., 2002), whereas Mexican fruit flies (Anastrepha ludens) show only a modest increase (Carey et al., 2008). Carey et al. (2005) did find extended LS for medflies fed on a stochastic diet regime compared to constant food availability, but it is not possible to quantify DR with this experimental design. Second, protein has been found to affect longevity in the Queensland fruit fly (Q-fly, Bactrocera tryoni). Q-flies live longer when allowed to self-regulate from a carbohydrate source and a protein source (Perez-Staples et al., 2007, 2008), but LS decreases when offered only a high-protein diet (Prabhu et al., 2008). Finally, reliable methods are available for measurement of intake amounts in Q-flies (Meats & Leighton, 2004; Meats & Kelly, 2008).

Given the debate over whether CR or specific nutrients underlie the LS effects of DR, we employed the GF approach to understand (i) how diet composition and concentration affect CF and (ii) the effects of CR and nutrient composition on various performance measures of Q-flies. We conducted two experiments: No-choice and Choice. For the No-choice experiments, we measured intake of 168 flies, restricted to one of 28 diet solutions that varied systemically in ratio and amount of yeast and sucrose. This experiment provided a detailed map of the relationship between nutrient intake and three performance measures – LS, lifetime egg production (LEP) and egg production rate (EPR). Lifetime egg production provides a measure of overall reproductive fitness, whereas LS and EPR describe the life history strategy leading to the observed LEP. In the Choice experiment, sucrose and yeast solutions were provided to 150 flies through separate micropipette tips at one of five concentration levels (22.5, 45, 90, 180 or 360 g L−1). The Choice experiment documents whether and how Q-flies actively regulate their nutritional intake and allows assessment of concordance between optimal nutrient ratio from the No-choice experiment and this regulated nutrient intake.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Effects of dietary restriction and diet composition on performance measures, ignoring compensatory feeding

Dietary restriction studies of invertebrate systems commonly overlook CF (Carvalho et al., 2005; Simpson & Raubenheimer, 2007). Accordingly, we first analysed our data using diet concentrations and diet composition (percent yeast) as predictors for a surface analysis of the No-choice data (i.e. assuming no compensation). Response surfaces differed for LS, LEP and EPR (Fig. 1, Supporting Table S1). Flies lived the longest on the S:Y 21:1 (95% sucrose) diet (Table 1; Fig. 1A) and LS decreased as percent yeast increased. The response surface model predicted peak longevity at a diet concentration of c. 170 g L−1, suggesting that DR increases LS. However, only a modest increase in LS was observed compared to the effect of percentage yeast (Table 1, Fig. 1A).

image

Figure 1.  Effects of the percent yeast and diet concentration on (A) lifespan (days), (B) lifetime egg production (eggs) and (C) egg production rate (eggs/day). Yeast percentage is the percent of the diet comprised of yeast. Diet concentration refers to the total amount (g) of sugar and yeast per 1 L water. Dashed vertical line represents right boundary of Carey et al. (2008) data collection (see Discussion). Surface plots are based on the raw data.

Download figure to PowerPoint

Table 1.   Summary statistics for effect of various diets on lifespan, egg production rate and lifetime egg production from No-choice experiment
Performance variableS:YConcentration (g L−1)
4590180360
  1. Values are the mean ± standard deviation for each treatment combination. Concentration refers to total amount of yeast + sucrose added to 1 L of water. S:Y ratio refers to the ratio of sucrose to yeast in the diet.

Lifespan (days)0:15.50 ± 2.104.00 ± 0.0020.30 ± 23.6011.30 ± 9.30
1:512.20 ± 8.0016.50 ± 16.0021.50 ± 9.3015.50 ± 12.60
1:1.513.70 ± 7.5022.50 ± 9.8028.80 ± 17.3031.80 ± 22.90
1.6:121.00 ± 19.1045.50 ± 10.2039.70 ± 14.9043.00 ± 11.80
3.4:132.40 ± 15.9057.80 ± 23.0045.20 ± 19.2059.50 ± 9.80
21:163.20 ± 35.2091.50 ± 44.7087.70 ± 44.1072.80 ± 20.50
1:047.80 ± 42.0037.20 ± 21.5041.20 ± 26.9029.20 ± 9.70
Lifetime egg production (eggs per fly)0:10.00 ± 0.000.00 ± 0.0022.00 ± 38.900.00 ± 0.00
1:50.30 ± 0.5023.30 ± 57.201.70 ± 3.608.70 ± 21.20
1:1.521.50 ± 40.1016.00 ± 26.50223.5 ± 248.168.30 ± 99.10
1.6:12.50 ± 4.00193.0 ± 150.7166.5 ± 166.6220.2 ± 180.5
3.4:17.20 ± 8.00262.5 ± 403.2102.0 ± 99.30163.3 ± 199.4
21:13.30 ± 6.701.00 ± 2.407.20 ± 12.608.80 ± 20.20
1:00.00 ± 0.000.00 ± 0.000.20 ± 0.400.00 ± 0.00
Egg production rate (eggs per day per fly)0:10.00 ± 0.000.00 ± 0.000.50 ± 0.700.00 ± 0.00
1:50.00 ± 0.000.60 ± 1.400.10 ± 0.300.20 ± 0.60
1:1.50.90 ± 1.600.70 ± 1.107.60 ± 9.602.00 ± 2.80
1.6:10.10 ± 0.104.50 ± 3.403.70 ± 3.806.30 ± 6.90
3.4:10.20 ± 0.203.30 ± 4.701.90 ± 1.503.00 ± 3.90
21:10.00 ± 0.100.00 ± 0.000.10 ± 0.100.10 ± 0.30
1:00.00 ± 0.000.00 ± 0.000.00 ± 0.000.00 ± 0.00

The response surface for EPR peaked around S:Y 2:1 (67% sucrose) and at a concentration c. 200 g L−1 (Fig. 1C). In contrast, the peak of LEP was more diffuse and was closer to 2.5:1 S:Y (71% sucrose) (Fig. 1B). Similar to LS, the response surface model predicted that EPR and LEP peaked around 170 g L−1. There was no evidence that these relationships varied with body size (see Experimental procedures).

Compensatory feeding for No-choice experiment

Q-flies may alter their feeding behaviour in response to diet concentration and composition, which may affect the interpretations of DR and diet composition on the performance measures (Carvalho et al., 2005; Simpson & Raubenheimer, 2007). We found that average consumption of diet depended on the sucrose (S) and yeast (Y) concentrations of the diet in nonlinear trends (Fig. 2A; Supporting Table S2). In general, decreasing the concentration of each S:Y treatment resulted in increased volume of diet consumed. However, the degree of compensatory intake depended on the S:Y ratio. The more extreme diet imbalances (e.g. S:Y 1:0, 21:1, 1:5, 0:1) had more than a 3.5-fold increase in volume consumed as concentration decreased. There was no evidence that body size influenced total volume consumed.

image

Figure 2.  Effect of yeast and sucrose concentrations on consumption of diet for No-choice experiment. Axes represent the amount (g) of yeast or sucrose in 1 L of diet solution. Grey dashed lines are the sucrose:yeast ratios of the various diets. Volume response surface was created using predicted values from second-order regression model. Total calories, yeast and sucrose surfaces were calculated from these predicted volumes.

Download figure to PowerPoint

Compensatory feeding can be clearly seen in the results for total caloric consumption (Fig. 2B). Q-flies consumed the most calories when on a balanced diet. When on a sucrose-rich diet (S:Y 1:0, 21:1), Q-flies consumed substantially fewer calories across all concentrations and overcame the effect of diet dilution. In yeast-rich diets, flies consumed the fewest calories at the lowest and highest dilutions. For balanced diets (e.g. S:Y 3.4:1, 1.6:1, 1:1.5) total caloric intake decreased monotonically with concentration.

As a result of CF, amounts of yeast and sucrose consumed did not follow a linear decrease with concentration (Fig. 2C,D). In high yeast diets (S:Y 1:5, 0:1), yeast intake remained constant or even increased with decreasing yeast concentration, until yeast concentration dropped below 90 g L−1, at which point yeast intake decreased. Sucrose intake decreased monotonically with diet concentration, especially at S:Y 3.4:1 and 1.6:1.

Performance measures taking account of compensatory feeding

As the Q-flies utilized CF, we repeated the DR response surface analysis for the No-choice experiment using the total amount of carbohydrate and protein consumed. LS was affected by total amount of carbohydrate and protein consumed in a nonlinear patterns (Fig. 3A, see Supporting Table S3). Following a −1 slope (isocaloric line, as protein and carbohydrate are almost calorically equal), LS increased as the proportion of protein in the diet decreased. Furthermore, LS increased monotonically with intake along individual diet rails (e.g. 47.2:1, 8:1); that is, LS consistently increased as total caloric intake increased for a given C:P ratio. Finally, we found that larger flies had shorter LS.

image

Figure 3.  Effect of protein and carbohydrate on (A) lifespan (days), (B) lifetime egg production (eggs) and (C) egg production rate (eggs/day) from No-choice experiment. Figure axes represent total consumption of protein and carbohydrate by female flies in the first 28 days of the experiment. Grey dots represent individual flies and grey dashed line shows isocaloric intake. Surface plots are based on predicted values from second-order regression models.

Download figure to PowerPoint

In the No-choice experiment, LEP increased with amount of protein and carbohydrate consumed and peaked at C:P 3.3:1. EPR showed a positive carbohydrate by protein interaction (a synergistic effect) with EPR being maximized along C:P 2.3:1. That is, EPR decreases as C:P ratio departs from the 2.3:1 nutritional trajectory.

Surface plots from the Choice experiment show trends similar to those of the No-choice experiment (Supporting Fig. S1). LS was extended for flies that consumed a sucrose rich diet (e.g. higher C:P ratios). LEP and EPR consistently increased with increased caloric intake.

Nutrient regulation in the Choice experiment

In the Choice experiment, the flies were able to regulate the intake of protein and carbohydrate separately across the diet regimes (Figs 4–5, Supporting Table S4). For sucrose concentrations above 45 g L−1, intake trajectories tracked the 3:1 ratio. However, at the lowest sucrose concentrations (22.5 g L−1), the flies appeared unable to maintain the C:P 3:1 intake trajectory and all trajectories became protein biased (the darker blue region in Fig. 5). Similarly at low yeast concentrations (22.5 and 45 g L−1), the intake trajectories became carbohydrate biased (the red region in Fig. 5). Finally, at the highest yeast concentration (360 g L−1), C:P ratios remained protein biased even at higher sucrose concentrations.

image

Figure 4.  Nutrient intake trajectories for flies in the Choice experiment. Flies were provided two feeding devices, one containing sucrose and the other yeast, both varying in one of five concentrations. Total nutrient consumed are mean values at 4 day interval over 28 days. The black dashed line represents the C:P 3:1 ratio that maximizes lifetime reproductive success. Grey solid line represents the C:P ratio if flies ate only yeast.

Download figure to PowerPoint

image

Figure 5.  Summary of nutrient ratios from the Choice experiment. Axes represent the concentration of sucrose and yeast provided to individual flies. This figure plots the natural log of the mean C consumed divided by mean P consumed [ln(C/P)]. Hence, this figure shows how yeast and sucrose concentrations affect C:P ratios. Legend shows the C:P ratio for each isocline [e.g. exp(1) = 2.73:1]. Mean values were calculated using total nutrient consumed over the first 28 days/total days alive during that period.

Download figure to PowerPoint

To maintain C:P ratios near 3:1, flies actively adjusted their yeast and sucrose intake by demonstrating strong CF responses (Supporting Fig. S2). As sucrose or yeast concentrations increased, the flies reduced intake of sucrose or yeast respectively.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Compensatory feeding

When intake is not measured, CF can confound the effects of CR and nutrients on LS in DR studies (Carvalho et al., 2005; Simpson & Raubenheimer, 2007). In this study, Q-flies adjusted their intake rates in response to both dietary dilution and composition. In the No-choice experiment, Q-flies consumed more diet when concentration decreased and when composition shifted towards more yeast. Consequently, the actual amount of yeast and sucrose consumed does not follow a smooth decreasing trend as diet concentration decreases, violating the implicit assumption of experimental designs in which diets are diluted but intake not measured. On yeast-rich diets (S:Y 1:5, 0:1), flies actually consumed more calories at the 90 and 180 g L−1 concentration than at 360 g L−1. On sugar-rich diets, caloric intake does not change with diet dilution. In contrast to the No-choice experiment, Q-flies responded predominately to just the concentration of each separate diet in the Choice experiment by increasing ingestion of the diluted diet.

This complex CF response to dietary dilutions and nutrient ratios is closely similar to that found by Lee et al. (2008) in similar experiments on Drosophila, and was also found in experiments on caterpillars by Lee et al. (2004). These results suggest an alternative interpretation of the results from Min & Tatar (2006). Sometimes used as evidence against the need to take account of CF in experiments on Drosophila in which intake was not measured (e.g. Piper & Bartke, 2008; Skorupa et al., 2008), Min & Tatar (2006) found that Drosophila had lower consumption on a 2% than a 16% yeast diet, and concluded that CF does not occur as Drosophila actually consumed less rather than more of the low calorie diet (2% yeast). However, since Min & Tatar only decreased yeast amounts and kept sucrose levels constant, they created a sugar-rich diet. As seen in our Fig. 2A, decreasing yeast amount while keeping sugar amounts constant caused Q-flies to consume less. Therefore, the results of Min & Tatar can be interpreted as Drosophila adjusting intake to nutrient ratios instead of a failure to show CF for dietary dilution within a given nutrient ratio.

Our CF results also delineate a potential problem with fixed allotments of liquid diets as employed by Carey et al. (2008). In this study, Carey et al. provided individual Mexican fruit flies with a fixed amount of liquid diet that varied in S:Y ratios and concentration. The fixed amount of diet was determined by estimating consumption on ad libitum diets of 25% yeast (S:Y 3:1). As seen in our Fig. 2B, this S:Y ratio has a daily caloric consumption at the highest concentration around 6.5 cal. However, for the other diets used by Carey et al. (S:Y 9:1, 24:1, and 1:0), our results indicate that flies will only consume about 4, 2 and 2 cal respectively. As Carey et al. estimate consumption for the diet with the highest consumption rate (S:Y 3:1), they may substantially overestimate projected consumption on other diets, effectively resulting in ad libitum feeding situations for many of the dilutions on the other S:Y diets. Unlike with the dry diet of rodents, unconsumed leftover liquid diet may not be observed due to evaporation. Consequently, our results highlight the need to measure ad libitum consumption on each S:Y diet instead of just one diet when using the fixed diet allotment technique as performed by Carey et al. (2008).

To demonstrate the potential effects of CF on our conclusions, we presented our response surfaces in two forms: one ignoring CF (Fig. 1) and the other using actual intakes (Fig. 3). The response surfaces that ignore CF suggested a modest increase in LS due to CR at low percent yeast and then this CR effect flattened as percent yeast increased. In contrast, actual intake amounts (Fig. 3A, using sucrose and yeast consumed instead of nutrients shows the same trend) reveals no effect of CR in low protein diets and that CR actually decreases LS at C:P ratios 2:1 and 1:1.

Role of nutrients in lifespan and egg production

Nutrient intake had a strong effect on LS in Q-flies as increases in the proportion of carbohydrate intake were associated with marked increases in LS. Furthermore, CR did not extend LS. In the LS response surface (Fig. 3A), the LS isoclines run almost perpendicular to the isocaloric line. This evidence indicates that for a constant nutrient ratio CR did not affect LS. Moreover, as the C:P nutrient ratio decreased (more protein biased), the isoclines became more parallel to the isocaloric line but LS decreased with CR. This directly opposes the hypothesis that CR underlies DR-related increases in longevity.

Our results matched remarkably well the two other studies that have employed the GF to discriminate between effects of CR and nutrients. Our experimental design was almost identical to that of Lee et al. (2008), except that we used unmated Q-flies instead of mated Drosophila. LS response surfaces from these studies show highly concordant patterns, with isoclines running perpendicular to the isocaloric line at high C:P ratios and then becoming more parallel with isocaloric lines as C:P decreases. Furthermore, we obtained similar patterns for LEP and EPR, both increasing with caloric intake and LEP peak shifted to a slightly higher C:P ratio than the EPR peak. Our surface plots also match Maklakov et al. (2008), who found similar patterns for all three performance measures with field crickets.

With medflies, Carey et al. (2002) developed a fixed diet allotment protocol similar to that often used in rodent studies. They defined ad libitum consumption as the point in which LS stops increasing with increasing diet concentration (using a S:Y 3:1 diet). Along the S:Y 3:1 ratio, our LS results with Q-flies show the same pattern as reported by Carey et al. (2002) for medflies: initial increase and the eventual plateau of LS with increasing consumption. However, when we offered ad libitum access to food Q-flies continued to consume more diet beyond the point at which LS reached a plateau and in so doing laid more eggs (see grey dots in Fig. 3A). Therefore, at least with Q-flies, this method of defining ad libitum is not valid.

Carey et al. (2008) conducted an experiment similar to ours using the Mexican fruit fly, except that they used the fixed diet allotment method (discussed above) and S:Y ratio only ranged from 1:0 to 3:1. Our surface plots based on diet composition (Fig. 1) do match very well with the Mexican fruit fly plots where data ranges from the two studies overlap (to the left of the grey dashed line in Fig. 1). Lifespan for females peaked at high S:Y ratio and decreasing concentration resulted in a modest increase in LS until peaking at 25% of ad libitum concentration. Similarly, increases in percent yeast raised LEP and above 25% restriction level diet concentration had no effect on LEP. It is likely the Mexican fruit fly utilized CF and if Carey et al. had measured intakes they might have reached conclusions more similar to those of our study.

Our LEP response surface illustrates substantial fitness costs if flies cannot regulate nutrient intake. When offered the chance to self-select their diet in the Choice experiments, Q-flies actively regulated nutrient intake along the C:P 3:1 trajectory, which maximized LEP in No-choice trials. Limits to regulation ability were evident when the sucrose or yeast concentrations fell to 45 g L−1. At this level, CF was insufficient to maintain nutrient intake and intake trajectories became biased toward the nutrient in the more concentrated solution. Similar abilities to regulate nutrients have been shown for other species using the GF methodology (Raubenheimer & Simpson, 1999; Lee et al., 2008; Maklakov et al., 2008).

It has often been proposed that there is a trade-off between fecundity and LS caused by competing demands for resources (Williams, 1966; Kirkwood & Holliday, 1979; de Jong, 1993; Partridge et al., 2005). The GF results discussed above have important implications for this perspective. Whereas resources have been traditionally viewed as unitary (i.e. a single currency such as energy rather than as multiple nutrient currencies), the GF results indicate that carbohydrates and protein are not equivalent. Thus, LS increased smoothly as the ratio of carbohydrate to protein in the diet increased, independently of calorie intake, and egg production was maximal at an intermediate nutrient ratio, falling as diets became either more protein or carbohydrate biased. These patterns could be interpreted as carbohydrates being essential to somatic maintenance and reproduction, but ingested protein being important primarily for reproduction. Hence, a carbohydrate-rich diet results in high levels of somatic maintenance but no reproduction. As C:P ratio falls and protein becomes more available, reproduction is activated and carbohydrates are diverted from somatic maintenance to fuel the energy demands of reproduction and to build lipid reserves in eggs. Eventually, at excessively low C:P ratios carbohydrate becomes scarce and both LS and reproduction decrease. This model would imply that deamination of protein to fuel metabolism and build fat is limited.

An alternative explanation for the different response surfaces for LS and LEP is that there are costs to protein ingestion, both direct and/or associated with egg production (Lee et al., 2008). Q-flies need to ingest protein for egg production and increasing protein intake leads to higher EPR, up to a limit (Drew, 1987; Meats & Leighton, 2004). Increasing protein intake has been shown to increase mitochondrial production of reactive oxygen species, causing more oxidative damage to cells (Ayala et al., 2007). Additionally, protein intake may affect the TOR signalling pathway, which is sensitive to amino acids and has been shown to affect LS in Drosophila (Kapahi et al., 2004). Thus, our response surfaces may reflect the positive effects of protein ingestion on egg production and the negative effects of protein on LS.

Conclusion

This study illustrates how CF can be problematic to DR studies and it adds to a growing recognition that nutrients play a central role in determination of LS. Our results demonstrate that CF can alter conclusions when intake is not measured, and that for imbalanced diets CF can even overcome dietary dilutions. We illustrate how a study taken as evidence against CF in Drosophila could be reinterpreted as support for CF. We also highlight potential difficulties with the current implementation of the method of using fixed amounts with liquid diets. Although only three species have been studied so far using the GF, all of them insects, a consistent pattern is emerging that (i) increased consumption of protein is associated with a marked decreased in LS and (ii) CR does not increase LS. These results come at an interesting time when it is known that many gene products involved in energy metabolism (e.g. TOR signalling pathway) also affect LS (Kapahi et al., 2004; Curtis et al., 2006; Kassi & Papavassiliou, 2008), but little is known about the interaction between these genes and their nutritional environment.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Study species and housing

Queensland flies (Q-flies, Bactrocera tryoni) were obtained as pupae from the Fruit Fly Production Facility at Elizabeth Macarthur Agricultural Institute (EMAI, New South Wales, Australia). The EMAI fly stock is maintained on a larval diet of lucerne chaff, torula yeast, and sucrose and an adult diet of 4:1 ratio of sucrose to hydrolysed yeast (S:Y). All flies emerged in a laboratory at Macquarie University, and within 24 h of emergence females were transferred to individual cages. We used unmated flies to avoid potential effects of mating on feeding, longevity and egg production. The cages comprised two stacked 480 mL clear plastic cups, with the bottom cut out of the inner cup. Mesh was secured over the top of the inner cup and an oviposition substrate was placed on the floor of the intact outer cup. The oviposition substrate was a parafilm-covered Petri dish (55 mm) containing 7 mL of 0.7% lemon essence solution (Queen Fine Foods Pty Ltd, Alderly, Queensland, Australia). The parafilm was pierced eight times with an insect pin. Temperature and humidity were maintained at c. 25 °C and c. 70% respectively.

Experimental diets

Treatment diets consisted of varying ratios and total concentrations of hydrolysed yeast (MP Biomedicals, Aurora, Ohio, USA, no. 103304: 45% protein, 24% carbohydrate, 21% indigestible fibre, 8% water and 2% other) and sucrose dissolved in distilled water. All macronutrient calculations were based on these values. Diets and water were dispensed using pipette tips (Eppendorf 2–200 μL tip) containing 100 or 120 μL of diet for No-choice or Choice experiments respectively. Pipette tips were capped with plasticine to decrease evaporation. Diet dispensers were replaced every 4 days and were checked daily to ensure adequate diet remained.

Nutrient intake was calculated by measuring length from pipette tip to liquid level using digital callipers. These lengths were converted to volumes by using a mathematical function relating length and volume for the pipette tip (fitted using piecewise regression). These volumes were further corrected for evaporation. For all experiments, each treatment diet had multiple controls (2 or 3) to measure evaporative loss of that diet. These controls consisted of the same cage setup and protocol as the treatment cages, but no flies.

No-choice experiment

To test the effect of protein/carbohydrate amounts on LS and fecundity, flies were maintained on one of 28 different treatment diets varying in amounts of sucrose and yeast. These 28 diets differed in the ratio of sucrose:yeast (S:Y; 1:0, 21:1, 3.4:1, 1.6:1, 1:1.5, 1:5, or 0:1) and total S+Y concentration (45, 90, 180, 360 g L−1 of water). These S:Y ratios provide an even sampling of possible diet ratios (see Fig 2) and resulted in carbohydrate:protein ratios (C:P) of 1:0, 47.2:1, 8:1, 4:1, 2:1, 1:1, 1:1.9. We assume that yeast is only uniquely supplying protein and not some other key nutrient(s). The fact that our results closely match those of Maklakov et al. (2008), who used chemically defined diets, substantiates our claim that protein and carbohydrates from yeast are the main nutrients affecting the performance measures in Q-flies.

Each diet treatment had three replicate flies, resulting in 84 female flies per experimental run, plus two controls for each diet treatment to measure evaporation rates. Diet and water consumption, as well as egg counts, were recorded every 4 days, at which time pipette tips and oviposition substrates were replaced. Mortality was recorded daily. As a measure of fly size, we photographed the right wing of each dead fly at standard magnification through a stereoscope phototube and measured distance (mm) from the intersection of the anal and median band to the margin of the costal band and the R4 + 5 vein using imagej software (Perez-Staples et al., 2007). We conducted two sequential experimental runs (n = 168).

Choice experiment

To understand the nutritional strategies of female Q-flies, we measured intake trajectories of individual flies provided two pipette tips, one containing yeast and the other sucrose. Each tip had a concentration of 22.5, 45, 90, 180, or 360 g L−1, resulting in 25 choice treatments (5 yeast × 5 sucrose concentrations). Each experimental run had three replicates for each diet treatment and two experimental runs were performed (total n = 150). Each diet had three controls. As with the No-choice experiment, diet data and egg counts were recorded every 4 days, mortality checked daily, and wing measurements taken postmortem.

Data analysis

We implemented a two-stage approach for our analyses. First, we explored factors affecting the quantity of diet consumed. For the No-choice experiment, we conducted a general linear model testing the effects of wing length (WL), yeast (Y) and sucrose (S) concentrations (and their interactions) on diet volume consumed. For the Choice experiment, we used a manova model with the same predictor variables to explore how they affected S and Y consumption. For these models, all response variables were transformed to meet model assumptions.

Second, for the No-choice experiment, we conducted a second-order surface analysis to test the effects of carbohydrate (C) and protein (P) consumption (evaporation corrected) on longevity and fecundity measures. For these models, we used only the first 28 days of consumption data, so that the confounding factor of LS was mitigated. Highly similar patterns were obtained when all data were included. A second-order surface analysis included the main factors (sucrose and protein consumption), quadratic effects of the main factors, and the interaction between the main effects (Montgomery, 2001). Additionally, we included WL and first order interactions with C and P to control for any body size effects, although any nonsignificant interactions with WL were removed in the final model.

In all surface models, we initially tested for different response surfaces between replicates by including the replicate interaction terms. Information criteria (AIC and BIC) indicated a significantly better fit for models with no replicate terms (ΔBIC>17) and hence all replicate interactions were removed.

Finally, for the performance measures in the Choice experiment, we fitted a nonparametric thin-plate spline to the data. We did not conduct a parametric response surface because the experimental design resulted in data being clumped making a narrower surface and caused a strong correlation between protein and carbohydrate consumption. As response surfaces are best visualized, we created surface plots using the FIELDS package in R (v2.7.1).

All statistical analyses were checked for violations of model assumptions through standard residual analyses and any violations were ameliorated using variable transformation. Backtransformed values are used for plots. All statistical analyses were performed in SAS 9.1, except for the nonparametric thin-plate spline, which was performed in R.

Acknowledgments

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

We would like to thank staff of New South Wales Department of Primary Industries, especially Laura Jiang and Selliah Sandaralingam, who generously provided us Queensland fruit flies from their mass-rearing facility. This project was facilitated by Horticulture Australia Limited (HAL) in partnership with Australian Citrus Growers and was funded by the Citrus levy (project code: CT05002). The Australian Government provides matched funding for all HAL R&D activities. Additional financial support was provided through a Macquarie University Research Development Grant (PWT). Stephen Simpson was supported by an Australian Research Council Federation Fellowship, and Diana Pérez-Staples was supported by a UNESCO-L’ORÉAL Co-sponsored Fellowship for Young Women in Life Sciences. We thank Fiona Clissold and Kwang Lee for discussion on experimental treatments, Ximena Nelson and Maria Castillo-Pando for help in experimental set up and Ken Cheng, John Prenter, two anonymous reviewers, and James Carey for insightful comments on the manuscript.

Author contributions

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Benjamin G. Fanson – data/statistical analysis and manuscript preparation. Christopher W. Weldon – conducted No-choice experiments. Diana Pérez-Staples – conducted Choice experiments. Stephen J. Simpson – project conceptual framework and experimental design. Phillip W. Taylor – project conceptual framework, logistics and experimental design.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Fig. S1 Effect of protein and carbohydrate on a) lifespan (days), b) lifetime egg production (eggs) and c) egg production rate (eggs/day) from Choice experiment. Figure axes represent total consumption of protein and carbohydrate by female flies in the first 28 days of the experiment. Gray dots represent individual flies and gray dashed line shows isocaloric intake. Surface plots are based on raw data.

Fig. S2 Yeast (a) and sucrose (b) consumption in response to varying yeast and sucrose concentrations from Choice experiment. Each figure represents the average 4-day consumption values (μL). Axes represent the amount (g) of yeast and sucrose in 1L of diet solution. Surface plots are based on predicted values from second-order regression models.

Table S1 Parameter estimates for effects of diet concentration and proportion of yeast on lifespan, lifetime egg production, and egg production rate from No-choice experiment.

Table S2 Parameter estimates for effects of sucrose and yeast concentrations on total intake volume from second order regression for No-choice experiment.

Table S3 Parameter estimates for effects of nutrient intake on lifespan, lifetime egg production, and egg production rate for No-choice experiment.

Table S4 Parameter estimates for effects of yeast and sucrose concentration on yeast and sucrose consumption for Choice experiment.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
ACEL_497_sm_Tables_S1-S4.doc100KSupporting info item
ACEL_497_sm_Fig.S1_small.tif508KSupporting info item
ACEL_497_sm_Fig.S2_small.tif455KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.