Increased allocation of adult-acquired carbohydrate to egg production results in its decreased allocation to sex pheromone production in mated females of the moth Heliothis virescens

Moths, which account for ~90% of species of Lepidoptera, typically use volatile sex pheromones to locate mates over distance. These volatile, female-produced sex pheromones are a prerequisite for successful mating in most species in which males must locate females in relatively low population densities (Tamaki, 1985). The female moth typically produces and/or releases sex pheromone, a species-specific single component or blend of components, usually from a gland located between the eighth and ninth abdominal segments (Ma and Ramaswamy, 2003). Most moth sex pheromone components [so-called ‘Type I’ (Ando et al., 2004)] are biosynthesized de novo, from acetate, via fatty acid synthesis followed by several steps peculiar to the pheromone gland, including desaturation, limited chain-shortening, reduction and acetylation and/or oxidation (Blomquist et al., 2011). In contrast to the voluminous work on the steps that modify the fatty acid chain of a pheromone component (reviewed in Jurenka, 2003) and the regulation of biosynthesis by the pheromone biosynthesis activating neuropeptide (PBAN) (reviewed in Blomquist et al., 2011), the physiological pools that provide precursor for fatty acid and pheromone biosynthesis have received relatively little attention. Recent work (Foster, 2009; Foster and Johnson, 2010; Foster and Anderson, 2011) on the polyandrous moth Heliothis virescens Fabricius (Noctuidae) demonstrated that adult-acquired carbohydrate (AAC), via incorporation into hemolymph trehalose and subsequent glycolysis, is a major source of precursor (acetyl CoA) for sex pheromone production.

Carbohydrate acquired through adult feeding, usually from plant nectar, is also important in female Lepidoptera for providing metabolites for other physiologies, including flight (O'Brien, 1999); foraging for hosts for progeny to develop on and for additional adult carbohydrate; and, most notably, egg production (Boggs and Ross, 1993; Wheeler, 1996). AAC is especially important for females of longer-lived species, which mature eggs throughout adult life. In such cases, AAC can have profound effects on female fecundity (Boggs and Ross, 1993; Wheeler, 1996), with its use increasing over a female's lifetime for the production of egg metabolites, including carbohydrates, fats (Telfer, 2009) and amino acids (O'Brien et al., 2000; O'Brien et al., 2002; O'Brien et al., 2004; O'Brien et al., 2005).

Given that polyandrous moths, such as H. virescens, will often mature eggs and produce and release sex pheromone for attracting mates at the same time, reiteratively throughout their life (e.g. Ramaswamy, 1990), the common use of hemolymph trehalose suggested to us that there might be competition for this resource between these physiologies, especially if AAC were limited, as is likely in nature, where plant nectar is highly variable and widely utilized (Boggs and Ross, 1993; Wäckers et al., 2007). When AAC is limited, its allocations to egg production or pheromone production, or to both, might decrease. Although it has been postulated that production of sex pheromone might ‘cost’ female fitness (Johansson and Jones, 2007; Harari et al., 2011), the relatively small quantities of fatty acid-derived sex pheromone produced by females suggest that a direct metabolic cost on egg production is unlikely (Greenfield, 1981; Cardé and Baker, 1984; Svensson, 1996). By contrast, the large amounts of metabolites required for egg production could impact the availability of precursor for pheromone production; i.e. high usage of AAC for egg production might result in its decreased use for pheromone production. As far as we know, all moths probably have sufficient larval resources prior to adult feeding to produce sex pheromone as virgins. However, as larval-acquired nutrients are depleted, pheromone production is likely to become increasingly dependent upon AAC, which, if limited, would impact pheromone production (Foster, 2009; Foster and Johnson, 2010). This competition for AAC between the two reproductive physiologies is likely to impact mated (of polyandrous species) females more so than virgins, as mated females generally use hemolymph trehalose and other resources to mature eggs at greater rates (Ramaswamy et al., 1997).

In this paper, we test whether there is physiological competition for allocation of AAC between pheromone and egg production in female H. virescens by feeding ¹³C-labeled glucose to both mated and virgin adult females, thus allowing us to quantify directly the allocations of AAC to pheromone and egg fat production. We demonstrate that, under a limited AAC scenario, allocation of AAC to egg production in mated, but not virgin, females results in its decreased proportional allocation to pheromone production.

Mated females fed unlabeled glucose daily showed a decrease in pheromone titer over the 6 days of sampling (F_1,33=17.70, P=0.0002; Fig. 1A). This was paralleled by a similar decline in hemolymph trehalose concentration (HTC) (F_1,56=10.06, P=0.0025; Fig. 1A). Virgins under the same feeding regime also showed a decline in pheromone titer over the course of the sampling period (F_1,40=26.29, P<0.0001; Fig. 1B). Considering the data qualitatively across the two different experiments, titers of virgins were typically greater than those of mated females at the same sampling time, as observed in other studies (Raina, 1989; Foster, 2009). In contrast to the mated females, HTC of virgins did not change over the 6 days of sampling (F_1,36=1.05, P=0.312; Fig. 1B).

As expected, hemolymph trehalose enrichment in mated females increased over time (Fig. 2A). Even after a single feed, trehalose enrichment was very high (ca. 0.80), demonstrating that most of the hemolymph trehalose at this time (and subsequently) was derived from AAC, consistent with the high turnover of hemolymph trehalose by mated females. Our two-factor model (F_2,47=9.782, P=0.0003) showed an effect of the number of feeds (F=15.178, P=0.0003), but not of total eggs laid (F=0.0137, P=0.907), on enrichment of hemolymph trehalose.

In our three-factor model, days fed had an effect on all three measures (pheromone, and ovarian and egg fat) of enrichment (Table 1). Total eggs laid had near-significant effects on pheromone and ovarian fat enrichments, but less so on egg fat enrichment, while trehalose enrichment had no effect on any of the three enrichments (Table 1), probably because its effects were largely accounted for in days fed. Of particular note was that while precursor enrichment of ovarian and egg fats increased with increased feeding, consistent with the increasing proportion of hemolymph trehalose derived from AAC, precursor enrichment of pheromone decreased (Fig. 2A). That is, the proportion of pheromone synthesized from AAC decreased with increased feeding, even though hemolymph trehalose was increasingly derived from AAC.

It was not clear to us what the significance of the effect of total eggs laid on precursor enrichment in ovarian (and to a lesser extent egg) fat meant in terms of an actual physiological effect, especially as the three factors in our model were all temporally related and consequently highly correlated. Therefore, we calculated the residuals from the linear relationship between the total eggs laid and the number of days fed. This gave us a sense of how many more or fewer eggs were laid by an individual female that had been fed for a certain period of time compared with the overall average across all individuals in the experiment. We then determined the relationships between these residuals and ovarian and egg fat enrichments. Both relationships were significant (F_1,58=5.107, P=0.028 for ovarian fat and F_1,49=5.993, P=0.022 for egg fat). Thus, for a given day, females that laid fewer eggs than average tended to use proportionally more AAC to synthesize new fats for eggs than did females that laid more eggs than average, consistent with the former females retaining greater amounts of AAC. Typically, across all days, precursor enrichments of ovarian and egg fats were similar, with both enrichments lower than that of the pheromone component.

We then tested whether the populational effects observed in our analyses held up at the individual level, by testing linear correlations between pheromone enrichment and egg and ovarian fats; i.e. do individuals show the same relationship between enrichments that we see when looking across all the individuals in the population? There was no correlation across individuals between pheromone enrichment and egg enrichment (F_1,32=0.014, P=0.91; data not shown), but there was a slight trend of a negative correlation (F_1,38=1.24, P=0.13) between pheromone enrichment and ovarian enrichment.

As in Experiment 1, precursor enrichment of hexadecanoate (16:Acyl) in eggs laid in the preceding 20 h increased over time (Fig. 2B). Our model showed an effect of number of feeds (F=5.74, P=0.031), but not of number of eggs laid (F=0.064, P=0.864), on precursor enrichment of 16:Acyl in eggs. Precursor enrichment of the stored pheromone precursor acid, (Z)-11-hexadecenoate (Z11-16:Acyl), did not change over time; consequently, there were no effects of either number of feeds (F=3.69, P=0.075) or number of eggs laid (F=1.53, P=0.236). By contrast, precursor enrichment of the stored pheromone precursor acid, 16:Acyl, decreased over time, with an effect of number of feeds (F=8.20, P=0.012), but not of total eggs laid (F=1.62, P=0.223). Overall, precursor enrichment of 16:Acyl in the pheromone gland tended to be less than that of Z11-16:Acyl, but both were typically greater than precursor enrichment of egg fat. Although strictly not comparable, it was apparent that precursor enrichment of Z11-16:Acyl was substantially lower at the start of the experiment (i.e. after one or two feeds) than that of the pheromone component in Experiment 1.

In Experiment 3, precursor enrichment of pheromone produced by virgins fed U-¹³C-glucose daily increased (F_1,29=30.56, P<0.0001; Fig. 3A) with increasing number of feeds, such that by the end of the experiment (when females were 6 days old) pheromone precursor enrichment had more than doubled. This demonstrates the proportional increase of AAC allocated to pheromone production in virgins, and is consistent with an increase in hemolymph trehalose derived from AAC, expected from repeated adult feeding on (labeled) glucose.

Similarly, precursor enrichment of Z11-16:Acyl also increased (F_1,18=6.97, P=0.017) over time in Experiment 4, although precursor enrichment of 16:Acyl did not change (F_1,18=1.06, P=0.317; Fig. 3B). As in Experiment 2, precursor enrichment of 16:Acyl tended to be less than that of Z11-16:Acyl. If we again, make the (not strictly valid) cross-experimental comparison of precursor enrichment of pheromone and pheromone precursor acid, it is apparent that, in contrast to what we observed for mated females, precursor enrichments in pheromone and Z11-16:Acyl were similar at the start of the experiment.

With limited intake of AAC, female H. virescens could reduce its allocation to all reproductive physiologies or selectively reduce its allocation to one (or more) physiology. By feeding females successive, but limited, amounts of ¹³C-labeled glucose, we tested how allocation of AAC to pheromone and egg production changed with successive adult feeds over time. In our experiments, females were housed individually in small containers that precluded flight; i.e. their activity and respiratory levels were relatively low and, we assume, similar between virgin and mated females. Therefore, we could observe relative allocation differences to the two reproductive physiologies between virgin and mated females without complications from other metabolically demanding activities, such as flight (Candy et al., 1997; O'Brien, 1999). However, it is important to recognize that our results and interpretations could be modified in natural conditions by high demands for the same metabolite by other physiological activities, such as flight.

With successive feeds, the proportion of hemolymph trehalose derived from AAC increased, as expected. If allocation of this pool to the two reproductive physiologies were to follow the same pattern, then precursor enrichment of pheromone and egg fats should also increase over time. While this was the case for virgins, it was not so for mated females, for which ovarian and egg fat precursor enrichments increased, but pheromone enrichment decreased. The most likely explanation for this effect is that increased egg maturation rate in mated females, mediated by higher juvenile hormone titers controlling ovarial patency (Ramaswamy et al., 1997), causes increased demand for hemolymph trehalose (Foster, 2009; O'Brien et al., 2000), resulting in a deficit of this metabolite for pheromone production. By contrast, virgins maintained or increased their proportional allocation of AAC to pheromone because of their lower egg maturation rate and its consequent lower demand for hemolymph trehalose (Foster, 2009; Ramaswamy et al., 1997). That demand for AAC was lower in virgins than in mated females was apparent by HTC being maintained over the duration of the experiment in the former, while decreasing substantially in the latter, under the same feeding regimen. Thus, for mated females in particular, sufficient AAC is important for maintaining allocation of hemolymph trehalose to both reproductive physiologies, but when AAC is limited, egg production limits allocation of this resource to pheromone production.

Although this effect of egg production on pheromone production in mated females was apparent in our full model by the effect of days fed on the various enrichments, for individual moths there was no correlation between egg fat and pheromone enrichment and only a weak negative correlation (consistent with our trend from the full model) between ovarian fat and pheromone enrichment. Because time, in general, had a strong effect on enrichments, the correlation between ovarian and pheromone enrichments is likely to be more appropriate, as these enrichments were sampled at the same time, whereas egg fat enrichment was from eggs laid roughly 24 h prior. The weaker correlation between ovarian fat and pheromone enrichment, in comparison to the significant trends over time in our full model, likely resulted from the fewer points available for the correlation compared with the full model. In particular, pheromone enrichment of older females (i.e. when both pheromone titer and enrichment were declining) became difficult to determine because of the smaller amounts of pheromone and incorporation of label into isotopomers.

This physiological competition for AAC between pheromone and egg production could be widespread across moths, especially given the biosynthetic similarity of many moth sex pheromone components (Blomquist et al., 2011), but is likely to be most significant for species such as H. virescens, which feed as adults, mature eggs following eclosion and are polyandrous. However, even in species that do not fit these criteria, it is conceivable that starvation during the larval stage could result in reduced carbohydrate reserves available for pheromone production at eclosion. In such cases, lower production of pheromone could make females less competitive for attracting mates (Johansson and Jones, 2007), with adult feeding being crucial to compensate for this.

The decline in pheromone enrichment over time in mated females means that AAC (hemolymph trehalose) is used proportionally less for producing precursor for synthesis of pheromone and, consequently, proportionally more larval-derived (unlabeled) sources of precursor are used. The most likely pool for this is fat stored in glycerolipids in the pheromone gland, with this relatively large pool (Foster, 2005) providing acetate precursor for de novo biosynthesis, as well as, potentially, preformed stored acids (especially 16:Acyl and Z11-16:Acyl) for pheromone biosynthesis. However, this decrease in proportional allocation of AAC to pheromone biosynthesis was accompanied by a very substantial decline in pheromone titer, suggesting that the increased proportional use of larval carbon is unable to compensate fully for the decreased availability of AAC. The low absolute amount of pheromone produced by these mated females is unlikely to be sufficient to attract conspecific males over distance (Foster and Johnson, 2011). Thus, with limited AAC intake, the reproductive cost of egg production in mated females is no further mating and lesser direct and indirect benefits that might accrue from such polyandry (Jennions and Petrie, 2000). With an adequate supply of AAC throughout the lifetime, mated female H. virescens continue to produce sufficient quantity of pheromone to attract males for further mating (Foster, 2009; Foster and Johnson, 2011). Thus, availability of AAC is a proximate mechanism underlying, at least in part, the frequency of mating in this and perhaps other polyandrous moth species.

Comparison of precursor enrichment in the pheromone component and its precursor acid, Z11-16:Acyl, showed apparent differences between mated and virgin females, especially at the start of experiments: mated females had substantially higher precursor enrichment of the pheromone component than of the precursor acid, while virgins had similar values for both moieties. Although this acid is the immediate pheromone precursor, most of it is rapidly reduced and oxidized to pheromone; i.e. the Z11-16:Acyl pool analyzed consists largely of acid stored in glycerolipids (Foster and Anderson, 2012). Exactly why stored Z11-16:Acyl in mated females is lower in AAC-derived precursor than the pheromone component is unclear, although the fact that the two are similar in virgins suggests it is a function of the greater demand for hemolymph trehalose in mated females. The generally lower precursor enrichment of stored 16:Acyl, compared with both pheromone and Z11-16:Acyl, suggests that (at least a portion of) this pool of 16:Acyl may be compartmentally distinct from that of the other two moieties, especially given its ubiquity and the possibility of contamination from non-glandular tissue. Regardless, the rate of synthesis of this stored acid is slow and it consequently represents a relatively small allocation of AAC in the gland (Foster and Anderson, 2012).

The use of both hemolymph trehalose, increasingly derived from AAC, and stored fats, largely derived from larval-acquired nutrients, as two pools supplying precursor to pheromone production in H. virescens is similar to the situation for carbon allocation to egg production reported in other Lepidoptera [and indeed in other insects in which adults feed on sugars (e.g. Rivero et al., 2001; Min et al., 2006)]. In studies on various moth and butterfly species fed larval and adult diets with different carbon isotopic signatures, carbon allocation to eggs was found to fit a two-compartment (pool) model, with one pool composed principally of larval-acquired carbon and the other a mixture of larval- and adult-acquired carbon (O'Brien et al., 2000; O'Brien et al., 2004). The increasing allocation of AAC in the mixed pool to ovarian and egg fats in our study was consistent with other studies on allocation of AAC to egg metabolites in other species of Lepidoptera (O'Brien et al., 2000; O'Brien et al., 2002; O'Brien et al., 2004; O'Brien et al., 2005), thereby supporting the validity of our fat proxy for the pattern of allocation of ACC to egg production.

In summary, AAC supplies a common pool, hemolymph trehalose, of precursor for pheromone and egg production in the moth H. virescens. When AAC is limited for mated females, physiological competition for this metabolite results in its allocation to egg production occurring at the expense of its allocation to pheromone production, thus limiting the ability of females to attract further mates. Thus, availability of AAC to adult females may be a proximate mechanism underlying the incidence of polyandry in this and other species of moths.

Heliothis virescens were from a colony maintained at the Department of Entomology, NDSU, for 9 years, with the original insects obtained from a colony at USDA-ARS-BRL, Fargo, ND, USA. Larvae were reared individually in small plastic cups (Solo Cup Co., Lake Forest, IL, USA) containing a wheat germ–casein diet, at 25°C under a 16 h:8 h light:dark photoperiod. Insects were sexed at the pupal stage, with females and males placed in separate containers. Adults were collected daily, and maintained under the same temperature and light conditions as larvae, until used in experiments.

Females were maintained individually in 140 ml plastic containers that precluded flight throughout an experiment. This ensured that females were relatively inactive and that respiratory differences between virgin and mated females were minimized. At the end of the scotophase on the day following eclosion (i.e. when females were 24–48 h old, hereafter referred to as 1 day old), each female was given a 37.5 μl drop (20.1 μmol) of 10% (w/w) U-¹³C-glucose (99%; Cambridge Isotope Laboratories, Andover, MA, USA) on a watch glass to feed upon; 37.5 μl was the maximum amount of liquid that females consumed reliably. Only females that consumed the entire drop were used in experiments. Each subsequent day, at the end of the scotophase, females were given another 37.5 μl drop of labeled glucose, until they were analyzed for pheromone or fat (see below). Females were not given access to any liquids (water, sugar solutions, etc.) between feeds. Eggs were not collected from virgins, because the number laid by females was low and erratic (some days, virgins did not lay any eggs).

Females were mated by placing 1-day-old virgins with 1-day-old virgin males, just prior to the start of scotophase. Pairs in copula were removed and left to finish copulation, after which the female was removed and placed in a small plastic container. At the end of scotophase, a female was given a 37.5 μl drop of 10% (w/w) U-¹³C-glucose on which to feed. Feeding was repeated each day until females were analyzed. Each day following mating, either at the time of feeding or analysis, the number of eggs a female laid was counted. On the day a female was analyzed, the eggs she laid in the prior 20 h were collected and analyzed for isotopic enrichment of fats (see below).

Our aim was to determine how allocation of AAC to both mating and egg production changed with increased number of feeds (i.e. effectively over time). We gave adult females a daily feed (see above) of highly ¹³C-enriched (99%) glucose and calculated precursor enrichment (see below) in the moieties that we analyzed. For pheromone production, we analyzed both the major pheromone component, Z11-16:Ald, and the stored pheromone precursor acids, Z11-16:Acyl and 16:Acyl (Choi et al., 2005; Foster, 2005). We only analyzed the major pheromone component (and not the other pheromone components) because it constitutes over 90% of the mass of total pheromone produced and released by females (Teal et al., 1986), and because determination of enrichment in abundant compounds is inherently easier than determination of enrichment in trace compounds (i.e. the other components). Regardless, because the other pheromone components are biosynthesized from the same acetate precursor (Choi et al., 2005), precursor enrichments are likely to be similar to that of the major component. We analyzed the pheromone precursor acids because surplus (i.e. not converted to pheromone) quantities of these are stored in glandular glycerolipids (Foster, 2005). While the net biosynthetic rates of both stored acids (especially 16:Acyl) are considerably lower than that of the pheromone component (Foster and Anderson, 2011; Foster and Anderson, 2012), we wished to check that precursor enrichment patterns in these acids were consistent with that of the pheromone component.

In Lepidoptera, AAC is allocated to amino acid, carbohydrate and lipid in eggs (O'Brien et al., 2002; Telfer, 2009). We used allocation to de novo-synthesized fatty acids in eggs (and ovaries) as a proxy for AAC allocation to eggs because: (1) fats are a major component of H. virescens eggs (Cohen and Patana, 1985) and therefore constitute a major allocation; (2) fats and pheromone are biosynthesized by similar routes, with common precursors (Blomquist et al., 2011); and (3) the allocation patterns for AAC to amino acids in eggs have been determined previously for a number of species of Lepidoptera (O'Brien et al., 2004; O'Brien et al., 2005; O'Brien et al., 2002), allowing comparison with our patterns for egg fat. In the first experiment, we measured enrichment in ovarian as well as egg fats, because of the latency between when eggs were laid and when pheromone and hemolymph were sampled; the eggs sampled could have been laid ca. 24 h prior to sampling of the other parameters, whereas ovaries were sampled for enrichment at the same time as hemolymph and pheromone. For allocation to ovarian and egg fats, we measured precursor enrichment of 16:Acyl, because this is, by far, the most abundant de novo-biosynthesized acid in H. virescens eggs (Cohen and Patana, 1985); lineolate and linolenate are also highly abundant, but these are from larval dietary sources, as Lepidoptera cannot biosynthesize them (Blomquist et al., 1991). Other de novo-biosynthesized acids, such as octadecanoate and oleate, are incorporated into insect eggs, but these are likely synthesized from the same precursor pool and thus their precursor enrichments will be the same as that of 16:Acyl (Wolfe and Chinkes, 2005).

We determined precursor enrichment in fats and pheromone by mass isotopomer distribution analysis (MIDA) (Hellerstein and Neese, 1992; Chinkes et al., 1996). MIDA is a combinatorial solution for determining the enrichment of a monomeric precursor pool by measuring the distribution of mass isotopomers in the resultant polymeric product(s). Assuming that the polymers are randomly assembled, a given precursor enrichment will give a specific pattern of isotopomers in a polymer. If the isotopomer pattern is measured, then precursor enrichment can be determined. In practice, only two labeled mass isotopomers (usually those with one and two labeled precursor units are the most intense and, therefore, the easiest to quantify) need be determined (Wolfe and Chinkes, 2005). Precursor enrichment is the proportion of labeled (tracer) precursor of the true precursor pool (including labeled and unlabeled) in a polymeric product and its calculation is not affected by isotopic fractionation effects (Hellerstein and Neese, 1999). However, in our study, the true precursor pool is acetyl CoA, with labeled precursor derived from the labeled glucose fed to adults. Hence isotopic fractionation may occur during glycolysis and subsequent pyruvate decarboxylation; i.e. unlabeled glucose may be converted to acetyl CoA faster than labeled glucose, and thus may comprise proportionally more of the true precursor pool. Regardless, our aim was not to determine absolute amounts of AAC converted to pheromone or fatty acids, but rather to compare how AAC allocation to the various moieties changed with increased feeding.

For sex pheromone, the pheromone gland of a female was dissected under a binocular microscope at the middle of the scotophase, and allowed to extract in 5 μl of n-heptane for at least 3 h. For pheromone titers, 50 ng of (Z)-11-tetradecenal (Z11-14:Ald) was added as an internal standard. For pheromone gland fats, the dissected gland was placed in 2:1 dichloromethane:methanol and left at −15°C for at least 18 h. After this, the solvent was decanted from the tissue and evaporated under a gentle stream of nitrogen. The residue was reacted with 0.5 mol l⁻¹ methanolic KOH for 1 h before aqueous 1.0 mol l⁻¹ HCl was added (Bjostad et al., 1987). The resultant fatty acid methyl esters (FAMEs) were extracted with n-heptane after vigorous shaking.

Ovaries were dissected in physiological saline under a binocular microscope, just after dissection of the pheromone gland. After patting dry on absorbent paper, ovaries were placed in a 1.5 ml Eppendorf tube, along with ca. 500 μl of 2:1 dichloromethane:methanol, and ground with a disposable plastic pestle before centrifugation at 2000×g for 5 min. The supernatant was decanted from the pellet and the solvent was evaporated under a stream of nitrogen. Fatty acids from eggs were extracted by the same approach. FAMEs were generated from the extract residues as for the pheromone gland extract.

Hemolymph was sampled from females just prior to dissection of the pheromone gland. A small hole was made in the cuticle on the intersegmental membrane between the sixth and seventh abdominal segments, and the hemolymph droplet was collected in a calibrated 5 μl glass capillary. After measurement of the volume (typically 2–5 μl), hemolymph was transferred to a glass vial, along with 50 μg of mannitol as internal standard, and water removed under vacuum. Trehalose was acetylated by reaction with acetic anhydride in pyridine at 100°C (Foster, 2009).

Pheromone and FAMEs were analyzed by gas chromatography/mass spectrometry (GC/MS) on a Hewlett-Packard 5890/5972 (Palo Alto, CA, USA). The GC was fitted with a 30 m×0.25 mm i.d. ZB-Wax column (Phenomenex, Torrance, CA, USA) and a splitless injector, and used helium as carrier gas at a constant flow of 1 ml min⁻¹. The GC column oven temperature was programmed from 80 to 180°C at 15°C min⁻¹, following an initial delay of 1 min, and then to 220°C at 3°C min⁻¹.

For MIDA, we used the MS in the selected ion monitoring mode. For Z11-16:Ald, we recorded m/z 220 (M+0; unlabeled), 222 (M+1; one ¹³C₂-labeled acetate) and 224 (M+2; two ¹³C₂-labeled acetates). These ions correspond to the molecular ion with loss of a water molecule, a reasonably intense ion with an intact carbon skeleton. For methyl hexadecanoate (16:Me) and methyl (Z)-11-hexadecenoate (Z11-16:Me), we recorded, respectively, m/z 270, 272 and 274, and 268, 270 and 274. These were the molecular ions of the M+0 (unlabeled), M+1 (one labeled acetate) and M+2 (two labeled acetates) isotopomers of the FAMEs.

Pheromone of females fed only unlabeled glucose, for determination of titers, was quantified in SIM mode, monitoring m/z 220 (for Z11-16:Ald) and 192 (for the internal standard, Z11-14:Ald), after calibrating the MS's response with quantitative standards.

We tested AAC allocations in both mated and virgin females, because they represent two different physiological demands for AAC: mated females mature eggs at a higher rate and, therefore, have a higher demand for AAC, while virgins mature eggs at a lower rate and have a lower demand for AAC (Ramaswamy et al., 1997; Foster, 2009). We tested a scenario in which AAC was limited: feeding was restricted to once a day and the concentration of carbohydrate was considerably less (up to one-fifth) than that typically found in plant nectar (Perret et al., 2001). Moreover, this amount appeared sufficient to maintain HTC in virgins, but not in mated females (see Results). We tested this AAC-limited scenario because we wanted to create physiological competition for AAC; a scenario with access to unlimited AAC would likely have produced a moot result, with mated females able to maintain AAC allocation to pheromone (e.g. Foster, 2009; Foster and Johnson, 2011), and therefore capable of producing enough pheromone for attracting males and remating, while maturing eggs. Separate experiments were required for analyses of pheromone and glandular fatty acids, as methanolic KOH (to produce FAMEs) reacts with aldehydes.

We fed mated females U-¹³C-glucose daily and analyzed individual females for each of HTC, and enrichment of trehalose, pheromone, ovarian (16:Acyl) and egg (16:Acyl) fats, for 0–9 days following mating. We analyzed four to eight different females on each day.

Mated females fed U-¹³C-glucose daily were analyzed for precursor enrichment of pheromone gland (16:Acyl and Z11-16:Acyl) and egg fats (16:Acyl) for 1–4 days after mating. We analyzed four to five different females on each day.

Virgin females, fed U-¹³C-glucose daily, were analyzed for precursor enrichment of pheromone (Experiment 3) and pheromone gland fats (16:Acyl and Z11-16:Acyl; Experiment 4) for 1–6 days and 1–4 days, respectively, after the initial feeding. For each enrichment measurement, four to six different females were analyzed on each day.

Pheromone titers of mated and virgin females were determined. Females were fed a drop of 37.5 μl of unlabeled 10% glucose daily, as in the enrichment experiments, until analyzed. Five to eight females of each type were analyzed on each day.

Changes in HTC and pheromone titers over time, for both mated and virgin females, were analyzed by linear regression, after first checking for normality and heteroscedasticity of data. Note that the HTC data for mated females were from Experiment 1, while new females (i.e. not used in Experiments 3 or 4) were used for the virgin female data.

For Experiment 1, we initially analyzed data by a structural equation modeling/path analysis approach (SAS Institute Inc., 2012), in order to investigate the relationships, simultaneously, among days fed, enrichment of pheromone, ovaries and eggs, and total numbers of eggs laid. Multiple iterations of the model were investigated, but the core model included factors in which the number of days fed could influence all other variables, all enrichments were allowed to influence each other, and the enrichment of eggs and ovaries could relate with the number of total eggs. However, no insights were gained that were not apparent with simpler and more common statistical analyses. Therefore, we used a general linear model (GLM) approach, with trehalose enrichment, days fed and total eggs laid as factors, and precursor enrichment of pheromone, ovaries (16:Acyl) and eggs (16:Acyl) as responses. Data for enrichment of trehalose and eggs were arcsine and log transformed, respectively, to ensure normality. We also tested linear correlations between pheromone enrichment and egg and ovarian fat enrichments to test whether any trade-off in metabolite allocation was apparent at the individual level.

In Experiment 2, we used linear models to test the relationships among days fed and eggs laid and precursor enrichments of pheromone gland fats (Z11-16:Acyl) and egg fat (16:Acyl), while in Experiments 3 and 4, we used the same approach to test for the relationship between days fed and pheromone and between days fed and pheromone gland fats (Z11-16:Acyl and 16:Acyl), respectively. All GLMs were performed in JMP (JMP, 2012).

We are grateful for the constructive comments of the two reviewers.