Physiological demands and nutrient intake modulate a trade-off between dispersal and reproduction based on age and sex of field crickets

Food availability strongly determines animal performance and fitness. Animals must acquire sufficient resources from their environment to meet the metabolic demands of life history investments, including growth, somatic maintenance, activity, and reproduction. The net energy gained from the digestion and absorption of consumed food sets an upper limit on energy available for life history investments. When resource acquisition is reduced, the physiological demands of life history cannot be fully met and life history trade-offs arise (Stearns, 1989; van Noordwijk and de Jong, 1986; Zera and Harshman, 2001). As a result of global change, animals will be increasingly threatened by reductions in the quantity and quality of resources available (Coviella and Trumble, 1999; Reich and Oleksyn, 2004; Walther, 2010), which could directly contribute to declines in population growth rates and stability (Elser et al., 2000; Lister and Garcia, 2018; Warne et al., 2010; Wetzel et al., 2016). However, fitness costs of reduced food availability are likely to vary amongst individuals and change through the lifespan along with sex- and life history strategy-specific physiological demands (Camus et al., 2017; Rapkin et al., 2018a).

Different life history traits are often maximized by diets that differ in both quantity and quality of nutrients (Harrison et al., 2014; Jang and Lee, 2018; Lee et al., 2008), reflecting the distinct physiological requirements of life history demands. For instance, somatic maintenance involves the management of long-term energy stores, primarily derived from carbohydrates. Accordingly, life span is maximized in both insects and mammals on carbohydrate-biased diets (Davies et al., 2018; Harrison et al., 2014; Jang and Lee, 2018; Le Couteur et al., 2016; Lee et al., 2008; Maklakov et al., 2008; Ng et al., 2019; Roeder and Behmer, 2014). In contrast, growth and reproduction involve building new tissues, which may require increased protein acquisition. Consequently, life history traits such as egg production rate and lifetime fecundity by females are maximized on diets with high protein content (Barragan-Fonseca et al., 2019; Clark et al., 2015; Jang and Lee, 2018; Lee et al., 2008; Ma et al., 2020; Maklakov et al., 2008; Rapkin et al., 2018a). Furthermore, for some life history investments, such as reproduction, ideal diets differ in a sex-specific manner (Harrison et al., 2014; Maklakov et al., 2008; Ng et al., 2019; Rapkin et al., 2018a, 2016; South et al., 2011). For example, carbohydrate-biased diets maximize lifetime calling effort by males, while intake of a balanced diet with equal amounts of protein and carbohydrate maximize lifetime fecundity of females in the field cricket Teleogryllus commodus (Maklakov et al., 2008). Since no one diet can maximize all life history traits, nutrient acquisition constrains life history investments (Jang and Lee, 2018). When animals allocate resources among traits that are optimized by highly divergent diets, such as lifespan versus female reproduction, nutritional conflicts are inevitable and give rise to life history trade-offs (Rapkin et al., 2018a).

In the face of nutritional conflicts, animals actively regulate feeding behavior to support concurrent allocations to multiple life history traits and maximize fitness (Simpson et al., 2004). When given a choice, intake is modulated between food sources to reach an intake target defined by the amount and balance of macro- and micronutrients consumed. Intake targets reflect an optimal balance among conflicting nutrient requirements for an individual in a given environment (Simpson and Raubenheimer, 1993; Simpson et al., 2004). While both repeatable and heritable (Camus et al., 2018; Han et al., 2016; Rapkin et al., 2018b) intake targets also change plastically along with physiological demands on short and long timescales. Following long-distance migration, carbohydrate intake of locusts drastically increases to regenerate glycogen and fat stores consumed during flight (Raubenheimer and Simpson, 1999). Similarly, to meet sex-specific reproductive demands, females but not males increase their protein intake following mating (Bowman and Tatar, 2016; Camus et al., 2018; Lee et al., 2013; Ng et al., 2019) and across the juvenile to adult life stage transition (Han and Dingemanse, 2017).

Animals frequently face environmental conditions where food choices are limited, and preferred diets are unavailable. Under these circumstances, animals can adjust either resource acquisition (feeding behavior) or allocation of resources amongst life history traits. When changing feeding behavior, animals balance the costs of excess consumption of an abundant nutrient against the costs of a deficit in a limiting nutrient (Behmer, 2009; Simpson and Raubenheimer, 1995). When changing resource allocation, animals can use long-term energy stores, or delay investments into costly life history traits such as growth and reproduction (de Jong and van Noordwijk, 1992; Boggs, 2009). An individual's need and ability to employ these strategies will depend on their current physiological demands, which are jointly determined by age, sex and life history strategy. While most studies only consider a single life stage, sex or life history strategy, these factors must be considered simultaneously in order to predict the subsets of the population most vulnerable to reduced nutrient availability.

Wing polymorphism is a widespread life history polymorphism in insects associated with a resource-based trade-off between flight and oogenesis (Harrison, 1980; Roff, 1986; Zera and Denno, 1997). Flight-capable and reproductive morphs of wing-polymorphic field crickets (Gryllus spp.) vary in the timing and amount of investment in flight or reproduction. Flight-capable long-winged (LW-f) morphs synthesize functional flight muscles as juveniles and accumulate lipid stores to fuel flight as adults (Zera and Larsen, 2001). In contrast, flightless short-winged (SW) reproductive morphs possess underdeveloped and non-functional flight muscles, and adult females achieve a greater early life fecundity compared with the LW-f morph by upregulating protein and phospholipid synthesis for oogenesis (Zera and Harshman, 2011). A third flightless morph is produced from LW-f crickets later in adulthood, when flight muscles degrade by histolysis, resulting in flightless long-winged crickets (LW-h) that are physiologically similar to the SW morph (Zera et al., 1997). Thus, flight muscle status determines current life history allocations more strongly than wing length. In early adulthood, female SW crickets select a more protein-biased diet than LW-f females to meet the demands of oogenesis (Clark et al., 2013, 2015). However, we do not know if morph differences in intake targets are life-stage and sex specific, making predictions about fitness consequences of suboptimal diets challenging.

We used a geometric framework of nutrition experimental approach to disentangle effects of diet nutrient content and life history on organismal performance (Simpson and Raubenheimer, 2012), in our focal species, the wing-polymorphic variable field cricket, Gryllus lineaticeps Stål 1858. To test the hypothesis that sensitivity to imbalanced diets is greatest when biosynthetic demands are highest, we manipulated diet macronutrient balance in juvenile and adult field crickets of both sexes, and measured investments into key tissues involved in reproduction, flight and somatic maintenance.

Variable field crickets, G. lineaticeps (>200 individuals), were collected at Sedgwick Reserve, University of California Natural Reserve System, UC Santa Barbara (34°41′34″N, 120°02′26″W, Santa Ynez, CA, USA) in July 2015 and transported to University of California Berkeley. Offspring of field-collected individuals were used to establish a laboratory colony composed of equal proportions of long- and short-winged adults. To generate continuous overlapping generations of crickets, newly emerged adults from the laboratory colony were added weekly into breeding colonies and provided with fresh moist substrate (mixture of soil and sand) for oviposition. Oviposition substrate was removed after 1 week and placed in individual containers to hatch. Laboratory colonies were maintained at approximately 27°C with a light:dark cycle of 16 h:8 h in age-structured large group housing at moderate densities (∼50 adults or 100 juveniles housed in 30 qt plastic bins). Prior to the start of the experiment, crickets had ad libitum access to a standard premixed and nutritionally complete dry, laboratory diet composed of wheat bran, wheat germ, milk powder and nutritional yeast (see table 1 of Zera and Larsen, 2001) and water (50 ml tubes plugged with cotton).

Within 24 h of entering either the final juvenile stage (last instar) or adulthood (for juvenile and adult treatment groups, respectively), crickets from the laboratory colony were randomly assigned to one of three nutritionally complete isocaloric artificial diets that differed only in the ratio of digestible protein to carbohydrate (P:C) (given as percentages of total diet mass): (1) protein-biased diet (P28:C14); (2) carbohydrate-biased diet (P8:C34); (3) choice between two diets, one protein and one carbohydrate-biased (Fig. 1). Diet P:C ratios were asymmetrical but selected based on the known intake targets of Gryllus firmus (a closely related species) adult females, such that individuals that feed to a compromise point between over-consuming the predominant nutrient and under-consuming the more limiting nutrient would consume equivalent calories as individuals able to feed to the intake target (Clark et al., 2013). Artificial diets were prepared as described by Behmer et al. (2002) and have been previously used in feeding studies with Gryllus spp. crickets (Clark et al., 2013, 2015, 2016). Two different carbohydrate-biased diets were used in the choice treatment diet pairings to assess the extent to which crickets actively select an intake target and diet macronutrient balance, as opposed to randomly feeding between food dishes. P28:C14 was common to both diet pairings, with the other constituent being P13:C29 (less extreme diet pairing) versus P8:C34 (more extreme diet pairing). To quantify consumption, crickets were allowed to feed ad libitum from pre-weighed (to the nearest 0.1 mg) spill-resistant Petri dishes of dried food. Fresh food and water were provided to juveniles every 5 days until the start of adulthood. The flight–oogenesis trade-off by morphs is maximized during early adulthood, so we only monitored feeding for the first 5 days of adulthood. Following feeding, food was allowed to dry for 24 h before re-weighing to calculate food intake. All accumulated solid waste present in cricket cages and Petri dishes of food was also removed and weighed. There was a strong association between total food intake and solid waste production (r²=0.94) that did not differ significantly with morph (P=0.813) or sex (P=0.375), suggesting that digestion and assimilation of food was similar across individuals.

During feeding trials, crickets were individually housed in square plastic containers (16.5×16.5×5 cm) in an incubator at 27°C with a light:dark cycle of 16 h:8 h (Percival I-36VL, Percival Scientific Inc., IA, USA). Position of crickets in the incubator was rotated daily while they were checked for survival and juvenile molt to adulthood. Cricket whole body mass to the nearest 0.1 mg was recorded both at the start and end of feeding trials using an electronic balance (Sartorius RC 250S, Sartorius, NY, USA). Over the course of the experiment, 12 individuals (9 juveniles and 3 adults) escaped or were lost from their cages. Of the remaining crickets, survival until the end of the experiment was high and similar amongst all treatments (≥95% from start of last juvenile instar to start of adulthood and ≥98% from start of adulthood to the fifth day of adulthood), suggesting that all artificial diets were sufficient to support growth and development of both juveniles and adults. All crickets that either escaped or died prior to completion of the experiment were excluded from statistical analyses (27 in total). Final sample sizes included 301 (155 female and 146 male) juvenile and 314 (158 female and 156 male) adult crickets. Upon completion of the feeding behavior assessments, all crickets were immediately frozen and stored at −80°C prior to dissection and biochemical analyses.

The expression of alternative life history strategies by crickets is associated with differential allocation of resources to key tissues that support either flight (flight muscle and somatic energy stores) or reproduction (gonads). Functional flight muscle confers flight capability and so, we used muscle status to define life history strategy (dispersal strategy=crickets with functional flight muscles; reproductive strategy=crickets with non-functional flight muscles). Visual inspections of flight muscle presence and condition were conducted for all individuals during dissections after feeding trials (Fig. 1). At the start of adulthood (adult day 0), a small proportion of SW crickets had functional flight muscles and were thus classified as expressing the dispersal strategy along with all the LW crickets (see Results and Fig. S1 for details).

To characterize biosynthesis trajectories and determine how life history allocations are altered by macronutrient imbalances, we divided body mass into three tissue components: (1) gonad (ovaries or testes), (2) flight muscle [dorsal longitudinal muscle (DLM)] and (3) somatic mass (residual body mass with flight muscle and gonads removed). All tissues were dissected from previously frozen crickets and wet tissue weight was recorded to the nearest mg with an electronic balance (Sartorius RC 250S, Sartorius, NY, USA). Tissue masses of experimental crickets were measured either at start of adulthood (adult day 0) or on the fifth day of adulthood (adult day 5) (Fig. 1). Crickets measured at the start of adulthood had fed on artificial diets as juveniles (last instar day 0 to adult day 0). Crickets measured on the fifth day of adulthood had fed on artificial diets only as adults (adult day 0 to adult day 5). In addition, tissue masses of 30 female and 30 male juvenile crickets were collected and frozen within 24 h of entering the last juvenile instar (last instar day 0) to assess sex differences in body composition that existed prior to the start of our experiment (Fig. 1). To control for body size, pronotum length was measured prior to tissue collection to the nearest micrometer using NIS-Elements D4 software (v.4.30.02) from pictures taken with a Nikon SMZ18 stereoscope (Nikon Instruments Inc., Melville, NY, USA).

We also determined if feeding on imbalanced diets impacted the quality of investments into key tissues based on tissue lipid content and composition. Lipids were extracted from a subset of functional flight muscles (juvenile, n=76; adult, n=31); gonads from 5-day-old adults with the reproductive life history strategy (female, n=45; male, n=41) and 5-day-old adult somatic tissues (dispersal strategy, n=34; reproductive strategy, n=84), using a modified Folch extraction (Folch et al., 1957) in glass tubes with a 2:1:0.9 (v/v/v) mixture of chloroform:methanol:water. Prior to extraction, 0.1 mg of 1-steroyl-rac-glycerol (MAG) and l-α-lyso-phosphatidylcholine (Lyso-PC) was added to all samples as an internal standard. MAG and Lyso-PC were selected to serve as internal standards for determination of both extraction efficiency and intra-sample repeatability, since prior analyses indicated these were not present in any tissue. Samples were vortexed (1 min), incubated at room temperature with gentle shaking (10 min), and centrifuged (10 min at 500 g) to separate the organic and inorganic layers. The lower organic layer, which contained both neutral and polar lipids, was carefully removed with a glass Pasteur pipette and transferred to a new glass vial. To ensure full recovery of lipids, the extraction procedure was repeated and organic layers combined. Extracts were then dried under nitrogen (N₂) gas and reconstituted in chloroform. Based on the added internal standard, lipid recovery was between 90 and 100% for all samples. Extracts were stored at −20°C prior to analysis by thin layer chromatography. All reagents and lipid standards were of the analytical or chromatography grade and purchased from Sigma-Aldrich (St Louis, MO, USA) and Avanti Polar Lipids (Alabaster, AL, USA).

Neutral lipid classes from gonad and somatic tissue extracts and polar lipid composition of flight muscles were quantified using thin layer chromatography coupled to a flame ionization detection system (Iatroscan MK-6 TLC-FID Analyzer, Shell-USA Inc., VA, USA) as described by Williams et al. (2011). Briefly, 1.5 μl of extracted sample was spotted in triplicate onto previously blanked chromorods (Type S5, Silica; Shell-USA). On the first rod of each set, either a neutral or polar lipid standard mix was spotted [neutral lipid mix: cholesteryl palmitate (CE), stearic acid (FFA), glyceryl tristerate (TAG), cholesterol (Chol), glyceryl 1,2-distearate (DAG) and 1-steroyl-rac-glycerol (MAG); polar lipid mix: phosphatidylethanolamine (PE), phosphatidylcholine (PC), lyso-phosphatidylcholine (Lyso-PC) and cardiolipin (CL)]. To separate major classes of neutral lipids present in gonads and whole bodies, rods were developed in a TLC tank in a solvent system of benzene:chloroform:formic acid (70:30:0.5 v/v/v) until the solvent front reached the 100 mark of the rod holder (∼35 min). To separate polar lipids, present in flight muscles, rods were developed in a solvent system of chloroform:methanol:water (80:35:3 v/v/v) until the solvent front reached the 100 mark of the rod holder (∼55 min). Rods were allowed to dry (5 min) before analysis with an Iatroscan MK-6 TLC-FID analyzer (Shell-USA) (atmospheric air flow rate 2 l min⁻¹; hydrogen flow rate 160 ml min⁻¹; scanning speed 3 cm s⁻¹). A comparison of retention time of peaks from the standard mix was used to identify lipid components in samples and these components were then quantified using standard curves (0.5–2.5 μmol ml⁻¹).

All statistical analyses were conducted in R version 3.5.2 (https://www.r-project.org/). Prior to analysis, data were checked for normality and transformed if necessary. Outliers greater than three standard deviations away from the mean were removed as likely data entry errors (n=2). To control for body size, pronotum length was included as a covariate in all analyses. All averages are reported as means±s.e.m. All post hoc pairwise comparisons were performed using Tukey's HSD tests.

Intake data were analyzed separately for each life stage because of differences in feeding duration within a life stage: juveniles were allowed to feed for the entire duration of the last juvenile instar, which lasted between 10 and 22 days, but adult feeding was restricted to only the first 5 days of adulthood. One outlier was removed from all analyses involving intake by juveniles. Protein and carbohydrate intake in mg were calculated by multiplying the total amount of food a cricket ate of a particular diet by the percentage of protein or carbohydrate in the diet, respectively. Total macronutrient intake was then calculated using the sum of the protein and carbohydrate intake, as a proxy for the total calories of food consumed, since digestible protein and carbohydrate are calorically equivalent.

For crickets in the choice diet treatment group, a paired t-test was first conducted to confirm that crickets regulated intake on each of the two-choice diet pairings [P28:C14 and P8:C34 (more extreme pairing) or P28:C14 and P13:C29 (less extreme pairing)]. We also conducted a MANOVA to determine if a common intake target was reached on the two-choice diet pairings. If intake targets were significantly different, diet pair was incorporated as a random effect in subsequent models. Subsequent MANCOVAs additionally compared intake targets for crickets in the choice diet treatment group as a function of life history strategy (dispersal strategy or reproductive strategy) and sex (female or male). To determine if significant differences amongst groups were driven by differences in protein, carbohydrate, total intake, or a combination of these factors, MANCOVAs were followed up by separate ANCOVAs of the same structure.

Crickets in the imbalanced diet treatment groups could behaviorally compensate for nutrient limitation by regulating the extent to which they over- and under-consume required nutrients. Using intake targets calculated from crickets in the choice diet treatment, we estimated nutrient errors incurred on the protein- and carbohydrate-biased diet treatments. Nutrient errors were calculated as the difference of observed protein or carbohydrate intake on the protein- or carbohydrate-biased diet from predicted intake target based on body size, life history strategy, sex and age. To standardize across life stages (which had different feeding durations and thus absolute intakes), we calculated relative nutrient errors as the percentage of intake missing or consumed above the preferred total amount {[(intake on protein- or carbohydrate-biased diet/predicted intake on choice diet) – 1] × 100}. ANOVAs with fixed effects of diet (protein-biased or carbohydrate-biased), life history strategy (dispersal strategy or reproductive strategy), life stage (juvenile or adult), sex (female or male) and all possible interactions were then conducted separately on over- and under-consumption to determine if relative nutrient errors differed between experimental groups. Because palatability of imbalanced diets may differ, we also tested for effects of imbalanced diets on total intake within each life stage using an ANCOVA that included sex, life history strategy and diet as fixed factors, and body size as a covariate.

We used generalized linear models with a binomial error distribution to assess the impact of sex and diet as a categorical variable (protein-biased, carbohydrate-biased or choice) on overall frequencies of wing length (short or long) and flight muscle status, a proxy for life history strategy (functional muscle=dispersal strategy; non-functional muscle=reproductive strategy). To complement these analyses, we used continuous consumption values for protein, carbohydrate, and total intake as separate predictor variables in logistic binomial regressions with likelihood of possessing functional flight muscles as a response variable. Two separate analyses were conducted for juveniles and long-wing adults. Short-wing adults were excluded from this analysis because they rarely maintain functional flight muscles. These models also included possible interactions between intake and sex. If the interaction with sex was significant, we separated males and females and reran the models with just intake as a follow-up analysis.

Effects of imbalanced diets on development time (days in final instar) of juveniles was assessed using an accelerated failure time model (survreg function in the R ‘survival’ package: https://github.com/therneau/survival). A model with a log-logistic error distribution was selected based on having the lowest AIC score. A stepwise approach was used to simplify a fully saturated model, which included all possible interactions between life history strategy, sex, and diet (all as fixed effects). Interactive terms were removed if ΔAIC<2, until no further simplification was possible (Crawley, 2012).

For analysis of life history allocations, quantified as either tissue mass or lipid composition, we used three-factor ANCOVAs that included life history strategy, sex, diet and all possible higher order interactions between factors (all as fixed effects). These models were fitted separately for each trait and life stage. One outlier was excluded from the analyses of gonad mass in 5-day-old adults.

Finally, we sought to determine how protein or carbohydrate nutrient intake errors (described in ‘Feeding behavior and intake errors’ section, above) impacted life history allocations at the individual level. To do so, we first conducted a principal component analysis to summarize multivariate allocations to body mass, size and organ size. Principal component scores were then regressed with the values of protein and carbohydrate intake errors (in mg of food; positive values indicated over-consumption and negative values indicated under-consumption from the intake target), with different slopes fitted for each life history strategy, sex and age group.

We first discuss intake targets of crickets on the choice diet treatment, wherein crickets were each offered one of two pairings of diets (see Materials and Methods for details) from which to select their desired nutrient intake. Crickets fed non-randomly between food dishes in both combinations of paired diets, reflecting active regulation of nutrient intake (paired t-test: t=−3.35, P<0.001). Juvenile crickets ate similar amounts of carbohydrates (F_1,149=2.66, P=0.11), but slightly less protein on the more extreme compared with less extreme diet pairing (F_1,149=10.88, P=0.001), with average protein intake of 92.9 mg (95% CI, 87.1–98.7 mg) in the more extreme pairing (with P8:C34) compared with 106.5 mg (95% CI, 100.8–112.2 mg) on the less extreme pairing (with P13:C29). Adult crickets selected similar intake targets, regardless of diet pairing (F_2,155=2.77, P=0.07). Intake data from both diet pairing choices were combined into a single treatment for all subsequent analyses, and diet pairing was included as a random effect in all models concerning juveniles.

As juveniles, intake targets differed according to life history strategy (F_2,144=13.05, P<0.001) and sex (F_2,144=9.27, P<0.001), with no interaction between life history and sex (F_2,144=1.05, P=0.35) (Fig. 2A). Specifically, juvenile crickets that developed functional flight muscles consumed more carbohydrates (F_1,145=16.91, P<0.001) and more protein (F_1,145=19.91, P<0.001), resulting in a larger total intake (F_1,145=26.33, P<0.001) compared with crickets not synthesizing flight muscles. Juvenile females ate more protein (F_1,145=15.12, P<0.001) and more total food (F_1,145=4.12, P=0.04), but similar amounts of carbohydrates (F_1,145=0.02, P=0.89), compared with males.

During early adulthood, intake targets differed according to life history strategy in females but not males (Life history strategy×Sex: F_2,152=3.14, P=0.046) (Fig. 2B). The interaction between life history strategy and sex arose from a slightly elevated protein intake in females expressing a reproductive strategy compared with females with a dispersal strategy (although this fell short of statistical significance in stringent Tukey's HSD, P=0.06). Adult male intake targets showed no differences with life history strategy (Tukey's HSD, all P>0.1). In addition, irrespective of life history strategy and distinct from juveniles, all adult females ate more protein (P<0.001), carbohydrate (P<0.05) and consequently total food (P<0.01) than males, reflecting large sex differences in caloric intake that accompany the onset of adulthood (Fig. 2B).

We next compared nutrient errors on the protein-biased and carbohydrate-biased diets relative to sex, life history strategy and life stage-specific intake targets on the choice diet. Overall, crickets compromised by over-consuming the predominant nutrient (carbohydrates on the carbohydrate-biased diet and protein on the protein-biased diet) and under-consuming the limiting nutrient (Fig. 3). Nutrient errors were similar between males and females at the juvenile stage. However, in adulthood, males over-consumed the predominant nutrient to a greater extent (Sex×Age: F_1,289=4.58, P=0.03) and correspondingly under-consumed the limited nutrient to a lesser extent (Sex×Age: F_1,289=4.83, P=0.03) than females.

Among females, juveniles over-consumed the predominant nutrient by 30.0±4.4%, while adults over-consumed the predominant nutrient by only 10.0±5.8%. Correspondingly, nutrient deficits on imbalanced diets were larger in adult compared with juvenile females (under-consumed: 53.8±1.6% for juveniles versus 60.0±1.6% for adults) (Fig. 3A). For males, relative nutrient errors were consistent across life stages at approximately 50% over- and under-consumption (over-consumed: 43.0±4.7% for juveniles versus 45.0±5.2% for adults; under-consumed: 49.1±1.7% for juveniles versus 50.0±1.9% for adults) (Fig. 3B). Total food intake by males was similar across all diets, but was reduced in females in the carbohydrate-biased treatment compared with the choice diet both as juveniles (Sex×Diet: F_2,287=5.39, P=0.005) and adults (Sex×Diet: F_2,301=6.78, P=0.001) (Fig. 2). Thus, females but not males incurred caloric deficits on the carbohydrate-biased diet.

The degree of over-consumption did not differ based on whether the diet was protein-biased or carbohydrate-biased (F_1,289=0.130, P=0.719), suggesting that crickets regulated their intake based on the predominant macronutrient in an imbalanced diet. However, the protein-biased (P28:C14) and carbohydrate-biased (P8:C34) diets were asymmetrical to one another. As a consequence of this, protein deficits on the carbohydrate-biased diet were larger than carbohydrate deficits on the protein-biased diet for both males (protein deficits: 53.8±1.8%, carbohydrate deficits: 47.1±1.7%) and females (protein deficits: 67.2±1.7%, carbohydrate deficits: 47.4±2.0%; F_1,289=13.08, P<0.001).

Finally, nutrient errors did not differ between crickets expressing a dispersal or reproductive life history strategy (over-consumption: F_1,289=2.38, P=0.12; under-consumption: F_1,289=3.18, P=0.08). Differences in total food intake associated with life history strategy were also not affected by diet treatment in either the juvenile (Life history strategy×Diet: F_2,287=0.64, P=0.53) or adult life stage (Life history strategy×Diet: F_2,301=1.50, P=0.23).

On the first day of adulthood, short-wing (SW) crickets were more common than long-wing (LW) crickets [SW: n=194 (64%), LW: n=107 (36%)]. The overall frequencies of wing lengths did not differ as a result of juvenile diet (χ²=3.56, P=0.17) or between sexes (χ²=1.02, P=0.31). However, across all diets combined, crickets with higher total food, protein, or carbohydrate intake were more likely to become long-winged (total food: χ²=41.69, P<0.001; protein: χ²=25.08, P<0.001; carbohydrate: χ²=5.86, P=0.02). We only assessed effects of diet on wing length in juveniles because wing length is a fixed trait in adulthood. Nearly all long-wing crickets (n=106, 99%) and about one-quarter of short-wing crickets (n=55, 28%) had functional flight muscles on the first day of adulthood and were thus classified as adopting a dispersal life history strategy (Fig. S1). Mirroring findings for wing length, flight muscle status on the first day of adulthood was not related to diet treatment (χ²=0.59, P=0.75), and the likelihood of a cricket possessing functional flight muscles upon adult emergence increased with total food, carbohydrate, and protein intake (total food: χ²=26.54, P<0.001; carbohydrate: χ²=8.02, P=0.004; protein: χ²=9.34, P=0.002) (Fig. S2).

By the fifth day of adulthood, functional flight muscles were present in only 3% (n=5) of short-wing and 43% (n=67) of long-wing crickets (Fig. S1). Among long-wing adults, adult diet did not impact the overall frequency of muscle histolysis (χ²=2.23, P=0.33). However, the likelihood of muscle histolysis could be predicted based on food food, with striking differences between the sexes (Sex×Total food: χ²=13.89, P<0.001; Sex×Protein: χ²=9.60, P=0.002). Analyzing the sexes separately revealed that females that ate more total food (χ²=10.89, P<0.001) or more protein (χ²=12.47, P<0.001) were more likely to have histolyzed flight muscle (Fig. 4A,C). In contrast, males that ate more food in total were less likely to have histolyzed flight muscle (χ²=4.78, P=0.03), and there was no association between male flight muscle status and protein intake (χ²=2.33, P=0.13) (Fig. 4B,D). Adult carbohydrate intake patterns were not associated with muscle histolysis for either males or females (χ²=0.06, P=0.80) (Fig. 4E,F).

The timing of gonad synthesis differed between males and females (Fig. 5). Male testes mass increased between the start of the final juvenile instar and adulthood, but did not change during adulthood for either morph, suggesting that males completed testes synthesis as juveniles (last instar day 0: 28.14±1.26 mg; adult day 0: 45.91±0.64 mg; adult day 5: 42.65±1.55 mg) (Fig. 5D–F). In contrast, females did not initiate ovary synthesis until adulthood. Both at the start of the final juvenile instar and adulthood, females had minimal, undeveloped ovaries, which grew rapidly during the first five days of adulthood (last instar day 0: 3.74±0.18 mg; adult day 0: 8.85±0.23 mg; adult day 5: 55.20±2.33 mg) (Fig. 5A–C). Furthermore, when feeding on the choice diet, at the fifth day of adulthood, ovaries of female crickets with a reproductive strategy were significantly heavier (Tukey's HSD, P<0.001) (Fig. 5C,F) and contained more lipid stores (in the form of triacylglycerides) (Tukey's HSD, P<0.001) than testes (Fig. 6B). Gonad mass was positively associated with body size for all crickets (adult day 0: F_1,288=52.06, P<0.001; adult day 5: F_1,300=9.41, P=0.002).

Reduced protein acquisition during adulthood constrained reproductive investment by females. Adult females but not males had differences in gonad size associated with life history strategy, and ovary size on the fifth day of adulthood further depended on diet (Life history strategy×Sex×Diet: F_2,300=5.87, P=0.003). As expected, on the fifth day of adulthood, crickets with a reproductive strategy had a greater ovary mass compared with crickets with a dispersal strategy on both the choice (reproductive: 65.10±3.77 mg; dispersal: 34.92±5.94 mg) (Tukey's HSD, P<0.001) and protein-biased (reproductive: 69.88±8.03 mg; dispersal: 24.08±5.02 mg) (Tukey's HSD, P=0.005) diet treatments. However, crickets with a reproductive strategy on the carbohydrate-biased diet as adults had a similar ovary mass as the females with a dispersal strategy (reproductive: 44.74±2.32 mg; dispersal: 37.79±6.73 mg) (Tukey's HSD, P=0.92) on the fifth day of adulthood (Fig. 5C). Females with a reproductive strategy on the carbohydrate-biased diet also provisioned their ovaries with less lipid stores compared with females with a reproductive strategy on the choice diet (all lipid classes, Tukey's HSD, P<0.05) (Fig. 6).

At the start of the last juvenile instar, all crickets had small, non-functional flight muscles (1.83±0.136 mg) (Fig. 7A,D). At the start of adulthood, dispersal strategy crickets had large, functional flight muscles (10.53±0.364 mg), while flight muscles of reproductive strategy crickets remained small (2.32±0.221 mg) and non-functional (F_1,288=427.93, P<0.001) (Fig. 7B,E). Crickets with a dispersal strategy had equivalent muscle masses at the start of adulthood regardless of sex (F_1,288=0.128, P=0.721) (Fig. 7B,E), but females had heavier flight muscles than males by the fifth day of adulthood (F_1,301=5.643, P=0.018) (Fig. 7C,F). Between the first and fifth day of adulthood, flight muscle mass of females with a dispersal strategy increased from 10.65±0.51 to 17.63±0.87 mg, while in males, average muscle mass only increased from 10.40±0.52 to 14.86±0.42 mg. During early adulthood, crickets with a dispersal strategy also remodeled the phospholipid membrane composition of their flight muscles. There was an approximate four-fold increase in flight muscle phosphatidylethanolamine (PE) content between the start and the fifth day of adulthood (day 0: 92.07±5.65 nmol; day 5: 410.44±23.71 nmol), but only a slight decrease in phosphatidylcholine (PC) content (day 0: 103.78±5.81 nmol; day 5: 80.66±8.91 nmol) (Fig. S3). This resulted in an increase of the PE:PC ratio from 1:1 to 5:1. Neither juvenile nor adult diet affected flight muscle mass (juvenile: F_2,288=0.058, P=0.944; adult: F_2,301=0.594, P=0.553) (Fig. 7B,C,E,F) or phospholipid composition (ANCOVA: diet, all P>0.05). Muscle mass was positively associated with body size of all crickets (adult day 0: F_1,288=3.60, P=0.058; adult day 5: F_1,300=4.71, P=0.03).

Average somatic mass at the start of the last juvenile instar was 273.71±9.03 mg, with no differences between sexes (F_1,57=0.004, P=0.95) (Fig. 8A,D). Between the start of the last juvenile instar and start of adulthood, somatic mass increased approximately 1.5-fold in both males and females (Fig. 8A,B,D,E), reflecting a large investment into somatic growth. By the start of adulthood, crickets with a dispersal strategy had a larger somatic mass than crickets with a reproductive strategy (F_1,288=13.04, P<0.001), with no differences between sexes (F_1,288=0.003, P=0.96) (Fig. 8B,E). Juveniles on the carbohydrate-biased diet had lower somatic mass than those in choice and protein-biased diet treatments (F_2,288=7.41, P<0.001) (Fig. 8B,E). In addition, female (but not male) development time to adulthood was delayed on the carbohydrate-biased diet (Sex×Diet: χ²=13.99, P<0.001) (Fig. S4).

By the fifth day of adulthood, females on the choice and protein-biased diet were slightly heavier (F_2,301=9.13, P<0.001) and had more somatic lipid stores (in the form of triacylglycerides) compared with males (F_1,105=5.59, P=0.02; females: 20.61±1.52 nmol; males: 15.26±1.34 nmol). Despite being slightly larger on the first day of adulthood, by the fifth day of adulthood crickets with a dispersal strategy had a reduced somatic mass compared with individuals with a reproductive strategy (F_1,300=21.62, P<0.001) (Fig. 8C,F). However, in spite of their slightly lower mass, crickets with a dispersal strategy had larger lipid stores than crickets with a reproductive strategy by the fifth day of adulthood (F_1,105=18.79, P<0.001) (dispersal: 21.35±1.91 nmol; reproduction: 16.55±1.22 nmol). The effects of adult diet on somatic mass differed between sexes (Sex×Diet: F_2,300=9.34, P<0.001), with no effect of diet in males (Tukey's HSD, all P>0.05; Fig. 8F), but a strong reduction in somatic mass in females on the carbohydrate-biased diet treatment compared with the choice and protein-biased diet treatments (Tukey's HSD, all P<0.01; Fig. 8C). Finally, on the fifth day of adulthood, somatic lipid stores were reduced in all crickets on the protein-biased diet compared with the carbohydrate-biased diet (F_1,105=12.14, P=0.007) (carbohydrate-biased: 20.77±1.85 nmol; choice: 18.81±1.85 nmol; protein-biased: 14.32±1.59 nmol).

We summarized life history allocations using a principal components analysis. The first principal component (PC1) explained 45.5% of variation, with all traits loading positively, and the highest loadings were somatic mass and pronotum length (Table S1). PC1 thus primarily described differences in overall body size and was interpreted as a somatic growth index. PC2 explained 23.8% of variation, with gonad mass loading positively and muscle mass loading negatively (Table S1). We thus used PC2 scores as a flight–reproduction trade-off index, with more positive scores reflecting higher investment in reproduction relative to flight.

The somatic growth index (PC1 score) was sensitive to protein but not carbohydrate errors (Table S2). Positive associations between protein errors and the somatic growth index were present in all juveniles, except for dispersal strategy males, suggesting that when crickets consume protein in excess of their intake target, they invest more in growth during juvenile development (Table S2A). During early adulthood, the somatic growth index of the females with a reproductive strategy was still positively associated with errors in protein intake, representing a distinct sensitivity to nutrient errors amongst this cohort (Table S2A).

The sensitivity of the flight–reproduction trade-off index (PC2 score) to diet changed across the life cycle and depended on both life history strategy and sex (Table 1). Within crickets with a reproductive strategy, over-consumption of protein was associated with increased reproductive investments (higher PC2 scores) during the juvenile life stage for males, but during adulthood for females (both P<0.01) (Table 1). These life stages corresponded to the time periods during which gonad synthesis was primarily occurring for males and females, respectively. During adulthood, excess carbohydrate intake was associated with increased investment in flight muscles (lower PC2 scores), but only within males with a dispersal strategy (Table 1).

We leveraged wing-polymorphic crickets, with discrete variation in life history strategies and associated differences in the timing and demands of reproduction, to explore how sex and life history strategies influence resource acquisition and allocation across the critical transition to reproductive maturity. We found that biosynthesis of costly tissues critical for life history can drive increases in either caloric or nutrient demands, and the timing of investments in these tissues across the life cycle dictate sensitivity of particular cohorts to reductions in nutrient availability.

High costs of ovarian synthesis led to elevated total resource and protein acquisition requirements for females. During the first 5 days of adulthood, when ovaries are synthesized, females consumed more total food and protein than males. For insects, lifetime egg production rate by females is maximized on a narrow range of diets that all have a large amount of protein, reflecting the high protein demands of egg production (Barragan-Fonseca et al., 2019; Camus et al., 2017; Harrison et al., 2014; Jang and Lee, 2018; Jensen et al., 2015; Kim et al., 2019; Lee et al., 2008; Maklakov et al., 2008; Rapkin et al., 2017; Roeder and Behmer, 2014). In contrast, investment into mating behaviors by males can be maximized by feeding on a broad range of diets, including relatively carbohydrate-biased diets (Bunning et al., 2015; Camus et al., 2017; Harrison et al., 2014; Jensen et al., 2015; Maklakov et al., 2008; Rapkin et al., 2017, 2016; South et al., 2011). The reproductive success of females is thus more sensitive to diet composition than males and overall fitness of females is maximized on a smaller range of diets compared with males (Camus et al., 2017). Our work thus supports the hypothesis that female preferences for more protein-biased diets are an important behavioral adaptation allowing females to meet the higher nutrient costs of reproduction (Bowman and Tatar, 2016; Camus et al., 2018; Harrison et al., 2014; Lee et al., 2013; Maklakov et al., 2008; Ng et al., 2019; Rapkin et al., 2017).

By tracking dietary preferences across life stages, we show that the elevated protein intake in females intensifies in early adulthood among flightless female crickets investing in large scale and rapid oogenesis. By the fifth day of adulthood, ovary mass of females with a reproductive strategy had increased eight-fold compared with the start of adulthood and was approximately two times greater than females with a dispersal strategy. Prior studies have shown that females but not males shift their dietary preferences and increase protein intake concurrently with the onset of gonad development either at the time of sexual maturation or following mating (Bowman and Tatar, 2016; Camus et al., 2018; Han and Dingemanse, 2017; Lee et al., 2013; Ng et al., 2019; Tsukamoto et al., 2014). Our results suggest that differences in the intensity of current investments in a particular life history trait may explain variability in the timing and magnitude of shifts in dietary preferences observed amongst individuals and across species.

Negative impacts of imbalanced diets for reproductive investment were specific to adult females expressing a reproductive life history strategy. All adult females incurred caloric and protein deficits on the carbohydrate-biased diet compared with the choice diet treatment. Females with a reproductive life history strategy suffered approximately 50% and 75% reductions in ovarian growth and energy provisioning, respectively, when constrained to the carbohydrate-biased diet, compared with their self-selected diet, and protein deficits reduced reproductive investments. This suggests that reduced protein acquisition specifically constrains reproductive investment and ovary development by flightless young adult females, rather than these being constrained by caloric intake more generally. In contrast, neither ovary mass nor flight muscle mass varied with diet on the fifth day of adulthood amongst females with functional flight muscles. Therefore, while females expressing either life history strategy incurred a similar magnitude of nutrient and caloric errors on imbalanced diets, these errors only resulted in negative life history impacts in females with a reproductive strategy. Moreover, when feeding was restricted to the carbohydrate-biased diet, ovary size of both dispersal and reproductive strategy females was similar on adult day 5. For Gryllus crickets, enhanced early life fecundity is the key fitness advantage for the flightless reproductive morphs (Mole and Zera, 1993; Roff, 1984) and we show that this is eliminated when protein acquisition is restricted. Thus, nutrient acquisition modulates the selective advantages of alternate life history strategies.

In contrast to females, males did not face an energetic or nutrient challenge due to biosynthetic demands of reproduction in either life stage. Males completed testes synthesis as juveniles, but the absence of sex differences in total food intake by juveniles suggests that caloric requirements were not elevated in males during the period of testes synthesis. Additionally, testes had ten-fold lower triglyceride content compared with ovaries, reflecting a lower energetic cost of synthesis. Males were thus able to meet costs of reproduction on imbalanced diets and their performance was insensitive to diet quality: neither testes mass nor energy provisioning differed with diet treatment or life history strategy in either life stage. Resource acquisition is, however, likely to be important for song production, which is metabolically expensive (Bailey et al., 1993; Bertram et al., 2011; Erregger et al., 2017; Hoback and Wagner, 1997; Thomson et al., 2014) and sensitive to food availability (Harrison et al., 2014; Judge et al., 2008; Maklakov et al., 2008; Wagner, 2005; Wagner and Hoback, 1999). However, male field crickets do not begin calling until at least 1 week post-eclosion to adulthood (Hunt et al., 2004; Murray and Cade, 1995; Zuk, 1988). By this point in the life cycle, most crickets with a dispersal strategy have already undergone muscle histolysis and switched life history allocations to reproduction (Zera et al., 1997). Accordingly, costs of imbalanced diets for male reproduction may be greater later in adulthood, past the point when males face the flight–reproduction trade-off.

Individuals that successfully disperse by flight colonize new habitats and locate mates, making flight capability an important contributor to the persistence and diversification of insects. However, high energetic costs of flight are hypothesized to drive the loss of flight capability when these costs offset the benefits of dispersal (Dingle, 2014; Harrison, 1980; Zera and Denno, 1997). Higher resting metabolic rates accompany the maintenance of functional flight muscles (Crnokrak and Roff, 2002; Nespolo et al., 2008; Sun et al., 2020; Zera et al., 1997) and large somatic lipid and glycogen stores must be synthesized in preparation for flight (Zera and Larsen, 2001; Zhao and Zera, 2002). Thus, during adulthood flight-capable individuals divert resources away from reproduction and into somatic maintenance to meet these energetic demands, providing the functional basis for the flight–reproduction trade-off. Total food intake of adult flight-capable and flightless female Gryllus crickets does not differ (Clark et al., 2013; Mole and Zera, 1993) and increasing total food availability does not eliminate negative associations between flight muscle and ovary mass (King et al., 2011a,b; Roff and Gélinas, 2003). Similarly to these findings, here we show that differences in total food intake associated with flight capability were absent in adulthood for G. lineaticeps. Females with a dispersal strategy had reduced ovary mass compared with crickets that were flightless at adult day 5, suggesting adult crickets with a dispersal strategy relied solely on changes to use of ingested nutrients.

In contrast to adults, juveniles adjusted resource acquisition to compensate for the energetic costs associated with the development of flight capability. Juveniles of both sexes and wing lengths that had functional flight muscles by the start of adulthood ate more compared with flightless crickets, while ending up with similar somatic and gonad masses. This suggests that the costs of flight muscle synthesis drove total resource acquisition requirements during the juvenile life stage. To our knowledge no prior study has assessed feeding behavior of juvenile wing-polymorphic crickets. Therefore, the elevated caloric requirements associated with biosynthesis of functional flight muscles reflect a previously overlooked cost of flight capability. Flight-capable juveniles will have to increase foraging efforts to meet higher caloric demands (Corrales-Carvajal et al., 2016; Csata et al., 2020; Jensen et al., 2012), and this could increase predation risk and mortality in natural environments (Lankford et al., 2001; Tessier and Woodruff, 2002).

In the absence of predation, juvenile crickets with a dispersal strategy were able to meet the caloric requirements necessary to support flight muscle development irrespective of diet composition, mitigating potential costs of imbalanced diets for flight capability. Therefore, imbalanced diets during the juvenile stage do not impact the likelihood of functional flight muscles at the start of adulthood, nor the quality of flight muscles. Similarly to our results, neither wing morph frequencies nor flight performance of the Australian ground cricket were impacted when reared with unlimited access to a poor-quality diet (Hall et al., 2008). This highlights the importance of considering the behavioral feeding strategy of a given life stage when determining susceptibility to imbalanced diets.

Because juveniles require more calories to support flight muscle synthesis, flight-capable adults should be more prevalent in field cricket populations whenever environmental resources are abundant. This prediction is consistent with prior findings in Gryllus crickets that prevalence of flight-capable adults is reduced when reared in either over-crowded or stressful laboratory conditions (Roff, 1990b; Zera and Tiebel, 1988) and contrasts to other wing-polymorphic insects, such as planthoppers and aphids, among which, higher frequencies of flight-capable morphs accompany poor-quality habitats (Braendle et al., 2006; Denno et al., 1986; Denno and Grissell, 1979). These contrasting patterns may be explained by differences in the physiological demands of flight: when demands of flight are high, development of migratory phenotypes will only be supported by high quality habitats (Cease et al., 2017). Costs of dispersal are expected to be higher for crickets compared with planthoppers and aphids, because as larger-bodied insects, crickets will not be able to as effectively use winds to conserve energy during long-distance flight (Dingle, 2014). Complete loss of flight capability by North American Gryllus crickets is common (22 of 41 species are flightless; Weissman and Gray, 2019). Elevated total resource requirements likely contribute to this loss of flight capability by eliminating ecological advantages of dispersal, such as escape from poor quality habitats.

Animals use nutrient sensing to appropriately time life history transitions to environmental conditions (Smykal and Raikhel, 2015; Templeman and Murphy, 2017). We have demonstrated that wing-polymorphic crickets use nutrient sensing to modulate the timing of wing muscle histolysis, and moreover show that the nutrient sensing is sexually dimorphic. Flight-capable females delayed flight-muscle histolysis (and thus the onset of large scale oogenesis) when caloric and protein intake was low during the first 5 days of adulthood, whereas flight-capable males delayed flight muscle histolysis and preserved flight capability when caloric and carbohydrate intake was high. Protein-limited females may gain a fitness benefit from retaining dispersal ability when protein availability is insufficient to support oogenesis and migrate to a more suitable habitat. The response in males was consistent with our general conclusions about flight being promoted by high quality nutritional environments (also see Cease et al., 2017). Accumulation of somatic lipid stores during adulthood was reduced when crickets fed on the protein-biased diet. Carbohydrate limitation may compromise flight ability because carbohydrates are important metabolic precursors for flight fuel synthesis (Zera et al., 2016, 1999). Dietary regulation of reproduction is sex specific in Drosophila melanogaster (Camus et al., 2019), and our work suggests that this may be a general phenomenon in insects that is driven by life history evolution.

Sex differences in responses to environmental conditions may also play an underappreciated role in the maintenance of stable wing polymorphisms over evolutionary timescales (Roff, 1990a). When food is abundant and high quality, males can maintain flight muscles and exploit the benefits of dispersal, while females may forgo flight in order to exploit conditions that are suitable for rapid ovary synthesis. Thus, selective pressures favoring the maintenance of flight ability in males will counter selective pressures favoring loss of flight in females. Gryllus crickets are generalist feeders and consume a primarily plant-based diet in the field (Alexander and Otte, 2009). As the climate becomes warmer, nitrogen content of plants is predicted to decline, suggesting that protein acquisition may become increasingly constrained (Reich and Oleksyn, 2004). Animals can over-consume nutrient-poor resources to avoid deficits of a limiting nutrient (Behmer, 2009; Simpson and Raubenheimer, 1995). However, limits on storage capacities and metabolic costs of processing and eliminating excess nutrients make over-consuming a predominant nutrient costly (Clark et al., 2013). Our findings suggest that adult females may limit over-consumption on imbalanced diets and instead respond to these declines by delaying reproduction and extending the dispersal window, thus buffering populations from changes in resource quality and potentially slowing the evolutionary loss of flight.

We have illustrated that the costs of imbalanced diets depend on the interaction between sex, life stage and life history strategy. Although this context dependence makes simple predictions challenging, some general principles have emerged that can aid in identifying individuals that will be most sensitive to low quality diets. Both the timing and magnitude of biosynthesizing energetically costly tissues determine how much and what type of food an animal will need to acquire. For crickets, synthesis of flight muscles during the juvenile stage and ovaries during early adulthood resulted in elevated resource demands. This resulted in temporal specificity to how sex and life history strategy relate to resource requirements; life history strategy was a primary determinant of juvenile caloric requirements, while sex became more important in determining caloric requirements once ovarian development was initiated in adulthood. Despite elevated resource demands during adulthood, females avoided excessive over-consumption of imbalanced diets, which left reproductive female crickets uniquely sensitive to fitness reductions on low quality (protein-deficient) diets. Testes synthesis is relatively inexpensive compared with ovary production, leaving males robust to performance costs of imbalanced diets. In response to changing nutrient availability, crickets can adapt by using sex-specific nutrient signaling to appropriately adjust relative investments into costly physiological traits and thereby alter their life history strategy, even on relatively short time scales. Together, this work shows that changes in nutrient acquisition can modulate life history trade-offs, shifting the costs and benefits of alternate life history strategies across life stages.

We would like to thank Anthony J. Zera and Bill Wagner for introducing us to the cricket system, Kate McCurdy and the staff at Sedgwick Reserve for support in the field, Kevin Roberts, Emily King, Chris Huebner, Katelyn Adam, Tony Huynh, Annie Sompayrac, Gina Kotos, Makayla Arcara, Rose Kang and Liana Williams for help in the lab and field, George Brooks, José Vázquez-Medina, and George Roderick for discussion and comments on interpretation of these results, and two anonymous reviewers for their helpful comments and suggestions.