Dietary cardenolides enhance growth and change the direction of the fecundity‐longevity trade‐off in milkweed bugs (Heteroptera: Lygaeinae)

Abstract Sequestration, that is, the accumulation of plant toxins into body tissues for defense, was predicted to incur physiological costs and may require resistance traits different from those of non‐sequestering insects. Alternatively, sequestering species could experience a cost in the absence of toxins due to selection on physiological homeostasis under permanent exposure of sequestered toxins in body tissues. Milkweed bugs (Heteroptera: Lygaeinae) sequester high amounts of plant‐derived cardenolides. Although being potent inhibitors of the ubiquitous animal enzyme Na+/K+‐ATPase, milkweed bugs can tolerate cardenolides by means of resistant Na+/K+‐ATPases. Both adaptations, resistance and sequestration, are ancestral traits of the Lygaeinae. Using four milkweed bug species (Heteroptera: Lygaeidae: Lygaeinae) and the related European firebug (Heteroptera: Pyrrhocoridae: Pyrrhocoris apterus) showing different combinations of the traits “cardenolide resistance” and “cardenolide sequestration,” we tested how the two traits affect larval growth upon exposure to dietary cardenolides in an artificial diet system. While cardenolides impaired the growth of P. apterus nymphs neither possessing a resistant Na+/K+‐ATPase nor sequestering cardenolides, growth was not affected in the non‐sequestering milkweed bug Arocatus longiceps, which possesses a resistant Na+/K+‐ATPase. Remarkably, cardenolides increased growth in the sequestering dietary specialists Caenocoris nerii and Oncopeltus fasciatus but not in the sequestering dietary generalist Spilostethus pandurus, which all possess a resistant Na+/K+‐ATPase. We furthermore assessed the effect of dietary cardenolides on additional life history parameters, including developmental speed, longevity of adults, and reproductive success in O. fasciatus. Unexpectedly, nymphs under cardenolide exposure developed substantially faster and lived longer as adults. However, fecundity of adults was reduced when maintained on cardenolide‐containing diet for their entire lifetime but not when adults were transferred to non‐toxic sunflower seeds. We speculate that the resistant Na+/K+‐ATPase of milkweed bugs is selected for working optimally in a “toxic environment,” that is, when sequestered cardenolides are stored in the body.


| INTRODUC TI ON
Chemical defenses are widespread among animals and remarkably diverse across species. Many insect herbivores accumulate secondary metabolites from their host plants and utilize them for their defense to ward off natural enemies, a phenomenon called sequestration (Opitz & Müller, 2009;Petschenka & Agrawal, 2016). For example, caterpillars of the monarch butterfly (Danaus plexippus) sequester cardenolides from milkweed plants (Asclepias spp.). On the contrary, other insects produce their toxins via de novo synthesis as observed in leaf beetles (Chrysomelidae) (Pasteels et al., 1988) or butterflies producing cyanogenic glycosides (Heliconiinae) (Brown & Francini, 1990). Sequestration of toxins from plants and de novo synthesis of toxins may trade-off evolutionarily as was suggested for Heliconius butterflies (Engler-Chaouat & Gilbert, 2007) indicating a cost of both strategies. Under an ecological view, costs of possessing defenses are assumed to be compensated by gained protection against predators (Bowers, 1992;Camara, 1997).
It was speculated that the physiological costs of de novo synthesis of defensive compounds are higher compared to the costs of sequestration of plant toxins (Fürstenberg-Hägg et al., 2014;Zvereva & Kozlov, 2016), which may explain why sequestration is a common phenomenon currently reported for more than 250 insect species acquiring toxins from at least 40 plant families (Opitz & Müller, 2009).
Nevertheless, sequestration of chemical defenses may incur physiological costs (Camara, 1997;Reudler et al., 2015) since sequestering insects are exposed to high concentrations of toxins stored within their body tissues. In accordance, physiological costs of toxin resistance (here: insecticides) such as reduced energetic resources, lifespan, and fecundity have been shown in several insects including mosquitoes (Carriere et al., 1994;Rivero et al., 2011) and bed bugs (Gordon et al., 2015). Besides physiological costs, sequestration may interfere with the insect immune system as indicated by a compromised immune response of buckeye caterpillars (Junonia coenia) dependent on the amount of iridoid glycosides in their diet (Smilanich et al., 2009).
Contrarily, dietary toxins were also suggested to interact positively with the immune system of sequestering insects such as caterpillars of the monarch butterfly (Tan et al., 2019) and the tobacco hornworm (Manduca sexta) (Garvey et al., 2021) indicating system-specific differences. However, empirical evidence on the benefits of defenses is more apparent than their costs, and the costs of chemical defenses are not always easy to detect (Lindstedt et al., 2010;Ruxton, 2014). In line with this, evidence for actual physiological costs such as negative effects on growth or other fitness parameters like longevity and fecundity is very scarce (Zvereva & Kozlov, 2016) and understanding the costs of sequestration will require reductionist comparative analyses integrating a diversity of physiological and ecological parameters.
Trade-offs play a crucial role in an organism's life history and occur when a beneficial change in one trait is linked to an unfavorable change in another trait causing a cost (Stearns, 1989). Life history theory suggests that fitness determining traits such as longevity and fecundity are negatively associated with each other (Flatt, 2011;Holliday, 1994). However, results are contradictory with studies showing positive, negative, or zero correlation between these two traits among individuals within a population (Bell, 1986;Van Noordwijk & de Jong, 1986). Generally, life history trade-offs result from compromises in resource allocation across growth, survival, maintenance, and reproduction under challenges such as predation occurring in an ecosystem (Levins, 1968;Roff, 1992;Sibly & Calow, 1986;Walsh & Reznick, 2010). Regarding chemical defense, an organism's potential physiological costs may be estimated as trade-offs between investments in defense and other physiological parameters such as growth, longevity, or fecundity (Camara, 1997;Ruxton et al., 2019).
Cardenolides are produced by more than 10 different plant families (Luckner & Wichtl, 2000;Malcolm, 1991) and are toxic to animals because they specifically inhibit the ubiquitous enzyme Na + /K + -ATPase (Emery et al., 1998;Lingrel, 1992). Na + /K + -ATPase is a cation carrier responsible for essential physiological functions such as the generation of neuronal action potentials and maintenance of an electrochemical gradient across the cell membrane (Jorgensen et al., 2003). Remarkably, insects from at least five orders, including milkweed bugs (Heteroptera: Lygaeinae), milkweed butterflies (Lepidoptera: Danaini), and leaf beetles (Coleoptera: Chrysomelidae), show common adaptations to cardenolides . Resistance in these groups is mediated by target site insensitivity due to a few amino acid substitutions in the first extracellular loop of the alpha subunit of Na + /K + -ATPase (ATPα), resulting in a high level of molecular convergence, that is, often the identical amino acid substitutions at the same positions confer resistance (Dobler et al., 2012Zhen et al., 2012).
Milkweed bugs are seed feeders primarily found on Apocynaceae species and on unrelated cardenolide-producing plants . Besides feeding on cardenolide-containing plants, milkweed bugs also sequester cardenolides to ward off predators (Evans et al., 1986;Pokharel et al., 2020). Alteration of the Na + /K + -ATPase in the Lygaeinae is probably correlated with several duplications of the ATPα1 gene resulting in four ATPα1 paralogs (A, B, C, and D) found in Oncopeltus fasciatus and Lygaeus kalmii (Dalla & Dobler, 2016;Yang et al., 2019;Zhen et al., 2012). Moreover, cardenolide-resistant Na + /K + -ATPases and the ability to sequester cardenolides are most likely synapomorphic traits of the Lygaeinae, which may account for the milkweed bugs' evolutionary success (Bramer et al., 2015).

K E Y W O R D S
cardenolides, fitness costs, life history traits, milkweed bugs, Na + /K + -ATPase, sequestration,

trade-off
The goal of our study was to evaluate if cardenolide exposure and sequestration causes physiological costs or benefits, and if these effects differ across closely related milkweed bug species possessing different combinations of the traits "resistance" and "sequestration" (i.e., having resistant/sensitive Na + /K + -ATPases and sequestering/ not sequestering). Our set of species included dietary specialist and generalist milkweed bug species as well as the European linden bug Pyrrhocoris apterus (Linnaeus, 1758, Pyrrhocoride) having no adaptations to cardenolides for comparison ( Figure 1).  Stearns (1992) considers manipulating a single factor the most reliable and informative method to measure costs (or trade-offs, if any), we manipulated cardenolide concentration in the diet as a single factor. For this purpose, we established an artificial diet approach and raised larvae on increasing doses of cardenolides in the diet. We assessed growth over the course of development. Furthermore, we determined the amount of sequestered cardenolides using liquid chromatography (HPLC-DAD).
To test for effects on potential trade-offs between life history parameters due to dietary cardenolides, we investigated the influence of dietary cardenolides on the longevity and fecundity in O.
fasciatus. Trade-offs have been measured in the field (Clutton-Brock, 1982;Clutton-Brock et al., 1983) and in the laboratory (Partridge & Farquhar, 1981), for example, by genotypic studies in Drosophila melanogaster (Rose & Charlesworth, 1981a, 1981b, and by phenotypic studies in Daphnia pulex and Platyias patulus (Bell, 1984a(Bell, , 1984b. How different traits will interact depends on ecological factors (e.g. nutrition or predation) and the physiological state (e.g. developmental stage or fecundity) of an organism. Thus, trade-offs can change across different environments in different species or even within the same species. Therefore, a life history trade-off probably may only appear in a species under a particular set of conditions, such as stress (Reznick, 1985).
To explicitly investigate the influence of dietary toxins in milkweed bug species, we set out to test the following hypotheses: (i) insect species having different physiological traits (resistance and sequestration) and ecological strategies (generalist and specialist) will react differently to dietary toxins, (ii) sequestering species will experience costs, either in the presence or absence of toxins, and (iii) the fecundity-longevity trade-off will be altered by dietary toxins.

| Preparation of artificial diet
We followed Jones et al.'s method to prepare an artificial diet for Oncopeltus fasciatus (Jones et al., 1986) but used a modified approach to offer the diet to the bugs. Sunflower seeds (25 g), wheat germ (25 g), casein (25 g), sucrose (10 g), Wesson's salt (4 g), vitamins (Vanderzant Vitamin mix, 5 g), methyl 4-hydroxybenzoate (1 g), sorbic acid (0.5 g), olive oil (7.5 g), and toxins (only for the treatment groups, not for controls) were blended in 200 ml of water until the mixture was homogenous. Agar (7.5 g) was boiled separately in 300 ml of water in a microwave. After 5 min, when the agar had slightly cooled down, the agar and the mixture were combined and F I G U R E 1 Overview of the Heteroptera species used in the experiments with their key traits. We compared four species of Lygaeinae sharing cardenolide-resistant Na + /K + -ATPase (+ = resistant; − = sensitive) but differing in their ability to sequester cardenolides (+ = sequestering; − = not sequestering). Within the sequestering milkweed bug species, Oncopeltus fasciatus and Caenocoris nerii may be classified as host-plant specialists (s = specialist), while Spilostethus pandurus uses a wide variety of host plants (g = generalist). Arocatus longiceps is specialized on Platanus and Ulmus which are devoid of cardenolides, and lost its ability to sequester cardenolides in the course of evolution. Pyrrhocoris apterus belongs to the relatively closely related family Pyrrhocoridae and is specialized on cardenolide-free Malvaceae. Furthermore, it is known to possess a sensitive Na + /K + -ATPase, and, based on our recent analyses, does not sequester cardenolides. Phylogenetic relationships of Heteroptera species are based on Bramer et al. (2015) blended again. The obtained paste was poured into plastic boxes and stored at 4°C for further use. For the feeding assays, an aliquot of diet was filled into the lid of a 2 ml Eppendorf tube and sealed with a piece of stretched parafilm to be used as an "artificial seed." Using portions from the same initial bulk of the diet, we added increasing amounts of an equimolar mixture of crystalline ouabain and digitoxin (Sigma-Aldrich, Taufkirchen, Germany) to prepare diets with 2 ("low"), 6 ("medium"), or 10 ("high") mg cardenolide/g dry weight of diet. The initial diet without cardenolides added was used as a control. We used the polar ouabain and the relatively apolar digitoxin to mimic the condition that plants typically produce an array of cardenolides with a wide polarity spectrum. The concentration of cardenolides in the diet was chosen to be in the range of natural cardenolide concentrations observed in Asclepias seeds (Isman, 1977). For each batch of diet prepared, we verified the concentrations of cardenolides across all dietary treatments by using highperformance liquid chromatography (HPLC, Section 2.2).

| Quantification of cardenolides
To verify the amount of cardenolides in the artificial diet, cubes of diet (approx. 20-25 mg dry weight) were freeze dried, weighed, and added to a 2 ml screw-cap tube containing approximately 900 mg of zirconia/silica beads (ø 2.3 mm, BioSpec Products, Inc., Bartlesville, OK, US). One ml HPLC-grade methanol containing 0.01 mg/ml of oleandrin (PhytoLab GmbH & Co. KG, Vestenbergsgreuth, Germany) as an internal standard was added to the tube, and diet samples were homogenized for two cycles of 45 s at 6.5 m/s in a Fast Prep™ homogenizer (MP Biomedicals, LLC, Solon, OH, US). After centrifugation at 16,100 g for 3 min, supernatants were transferred into fresh tubes. Original samples were extracted two more times with 1 ml of pure methanol as described above. All the supernatants of a sample were combined, evaporated to dryness under a stream of nitrogen gas, and resuspended with 100 μl methanol by agitating tubes in the Fast Prep™ homogenizer without beads. Subsequently, samples were filtered into HPLC vials using Rotilabo ® syringe filters (nylon, pore size: 0.45 μm, ø 13 mm, Carl Roth GmbH & Co. KG, Karlsruhe, Germany). Finally, 15 μl of the extract was injected into an Agilent 1100 series HPLC (Agilent Technologies, Santa Clara, US) equipped with a photodiode array detector, and compounds were separated on an EC 150/4.6NUCLEODUR ® C18 Gravity column (3 µm, 150 mm × 4.6 mm, Macherey-Nagel, Düren, Germany).
Cardenolides were eluted at a constant flow rate of 0.7 ml/min at 30°C using the following acetonitrile-water gradient: 0-2 min 10% acetonitrile, 13 min 95% acetonitrile, 18 min 95% acetonitrile, 23 min 10% acetonitrile, and 5 min reconditioning at 10% acetonitrile. The same HPLC method was used for quantification of sequestered cardenolides in O. fasciatus and P. apterus, respectively. For the analysis of sequestered cardenolides in C. nerii, S. pandurus, and A. longiceps, we used a different acetonitrile-water gradient to achieve improved separation of polar cardenolides: 0-2 min 16% acetonitrile, 25 min 70% acetonitrile, 30 min 95% acetonitrile, 35 min 95% acetonitrile, 37 min 16% acetonitrile, and 10 min reconditioning at 16% acetonitrile. We interpreted peaks with symmetrical absorption maxima between 216 and 222 nm as cardenolides (Malcolm & Zalucki, 1996) and integrated peaks at 218 nm using the Agilent ChemStation software (B.04.03). The amount of cardenolides in a sample was quantified based on the peak area of the known concentration of the internal standard oleandrin.

| Insect colonies
Oncopeltus fasciatus were obtained from a long-term laboratory colony (originally from the United States) maintained on sunflower seeds. We collected specimens of P. apterus in the vicinity of linden trees (Tilia spp., Malvaceae) and specimens of A. longiceps from under the bark of plane trees (Platanus spp., Platanaceae) in Giessen, We reared all insects on organic sunflower seeds (Alnatura GmbH, Darmstadt, Germany), supplied water in cotton-plugged Eppendorf tubes, and included a piece of cotton wool as a substrate for oviposition. In addition to sunflower seeds, P. apterus was provided with approximately 10 freshly chopped mealworms twice a week.
For the experiments described below, we used first-generation offspring from field-collected P. apterus and A. longiceps (maintained as described above), whereas S. pandurus and C. nerii offspring were obtained from colonies maintained in the laboratory for more than four generations.

| Growth assay
We carried out feeding assays to investigate the influence of increasing doses of dietary toxins on the growth of larvae of four species of milkweed bugs (O. fasciatus, C. nerii, S. pandurus, and A. longiceps) and an outgroup, P. apterus. These species either lack the ability to tolerate and sequester cardenolides (P. apterus), possess a cardenolide-resistant Na + /K + -ATPase, and can sequester cardenolides (O. fasciatus, C. nerii, and S. pandurus) or possess a cardenolide-resistant Na + /K + -ATPase but lost the ability to sequester (A. longiceps) (Figure 1). We placed three second instar (L2) larvae from the stock colonies in a Petri dish (60 mm × 15 mm, with vents, Greiner Bio-One, Frickenhausen, Germany) lined with filter paper (Rotilabo ® round filters, Carl Roth GmbH & Co. KG, Karlsruhe, Germany) that was supplied with an artificial seed (either being devoid of cardenolides, or possessing a cardenolide concentration of 2, 6, or 10 µg/mg dry weight) and a water source (a 0.5 ml Eppendorf tube plugged with cotton wool). The artificial seeds were replaced once in 2 weeks. All Petri dishes were spatially randomized and maintained in a controlled environment (KBWF 240 climate chamber, Binder, Tuttlingen, Germany) at 21°C, 60% humidity, and a day/night cycle of 16/8 h under artificial light. The growth of larvae was assessed twice a week over a period of 3 weeks by sedating all bugs of a Petri dish with CO 2 and weighing them jointly. After reaching adulthood, at least one bug per Petri dish was transferred to a toxin-free diet for 10 days to avoid a potential bias from toxins remaining in the gut (i.e., not being sequestered) by purging. Finally, bugs were killed by freezing, freeze-dried, weighed, extracted and analyzed by HPLC to estimate the amount of sequestered cardenolides (Section 2.2).
We also estimated the amount of excretion products on filter paper that may provide an indication of food intake (Appendix S1). For O. fasciatus, growth assays were carried out in batches.
Altogether three experiments (n = 10 per treatment for experiments I and II; n = 5 per treatment for experiment III) were carried out. Additionally, we also carried out growth assays using a different O. fasciatus strain ( Figure S3). Sequestration of cardenolides in O. fasciatus was only evaluated in specimens from the experiments I and II.

| Developmental time
Since the effects of dietary cardenolides on growth were most pronounced in O. fasciatus, we carried out a separate experiment to assess additional life history parameters including duration of larval development, adult lifespan, and body size under the influence of dietary cardenolides in this species. We chose the medium-dose cardenolide (6 µg/mg dry weight) because we observed that O. fasciatus showed maximal growth on this diet in our previous experiment. The experimental setup was similar to that of the growth assay. However, here only one L2 larva was placed in each Petri dish to monitor the time of larval development. Petri dishes lined with filter paper either containing medium-dose diet or control diet without toxins and a water source (Section 2.4) were spatially randomized and kept in a climate chamber (Fitotron ® SGC 120, Weiss Technik, Loughborough, UK) at 27°C, 60% humidity, and a day/night cycle of 16/8 h under artificial light. We checked the Petri dishes every day for dead individuals, raised the bugs until adulthood, and observed them until they died. We also measured the body length of adult males and females raised on the two different diets using a Vernier caliper.

| Reproductive fitness
We conducted two additional experiments to assess the effect of dietary cardenolides on reproductive fitness. We raised L2 larvae in bulk (around 50 individuals) until adulthood in plastic boxes (19 × 19 × 19 cm) covered with gauze either on two artificial seeds of medium-dose or control diet and a water source (four Eppendorf tubes of 2 ml plugged with cotton wool). Boxes were kept in a climate chamber (Fitotron ® SGC 120, Weiss Technik, Loughborough, UK) at 27°C, 60% humidity, and a day/night cycle of 16/8 h under artificial light. Artificial seeds were replaced once in 2 weeks. At least 3 (but not older than 6)-day-old males and females from the same treatment were paired in Petri dishes (9 cm × 1.5 cm, with vents, Greiner Bio-One, Frickenhausen, Germany) lined with filter paper and a water source (a 2 ml Eppendorf tube plugged with cotton wool). Additionally, we included a piece of cotton wool as a substrate for oviposition. Petri dishes were spatially randomized and kept under the same conditions as described above. In a first experiment, adult bugs were supplied with the same type of artificial diet that they were raised upon (i.e., either control or mediumdose cardenolide artificial seeds). In nature, adults of O. fasciatus disperse after reaching adulthood and forage for other seeds besides Asclepias spp. (Feir, 1974). Therefore, we carried out a second experiment under the same conditions as described above in which pairs of bugs were supplied with approx. 20 sunflower seeds instead of artificial seeds. Since a substantial portion of eggs in both experiments were unviable (possibly due to the use of an artificial diet), we counted only hatchlings and not the total number of eggs produced by each female over its entire lifespan. Especially after transfer to sunflower seeds, viable eggs were only produced by

| Statistical analysis
Statistical analyses were computed using JMP ® Pro 15 statistical software (SAS Institute, Cary, NC, US). All data were 1og 10 transformed to achieve homogeneity of variances and normality of residuals. For feeding experiments, we analyzed sequential data on larval masses with repeated measures ANOVA followed by LSMeans Tukey HSD test to assess potential differences across treatments. We compared the amounts of sequestered cardenolides across treatments and milkweed bug species by ANOVA followed by LSMeans Tukey HSD test and included bug species and treatment as model effects.
Additionally, we estimated Pearson's correlation coefficients be-

| Influence of dietary cardenolides on growth
We examined the influence of dietary toxins on the growth of P. apterus, O. fasciatus, C. nerii, S. pandurus, and A. longiceps (Figure 2) using an artificial diet containing increasing doses of cardenolides ( Figure S1). Growth of P. apterus was compromised substantially
We found substantial differences regarding the concentration of sequestered cardenolides which is consistent with earlier findings (Bramer et al., 2015).  Besides the total amounts of cardenolides sequestered, we also compared the number of structurally different cardenolides across the sequestering species ( Figure S2). Based on retention time comparison using an authentic standard, ouabain was sequestered as such. However, a peak with the retention time of digitoxin was not detected, but we found up to three compounds with a cardenolide spectrum and increased polarity probably representing digitoxin metabolites. Remarkably, we found more than one and up to three putative digitoxin metabolites in S. pandurus and C. nerii. In O. fasciatus, we did not find any digitoxin metabolites.

| Influence of dietary cardenolides on reproductive fitness of Oncopeltus fasciatus
We observed substantial effects on the reproductive fitness of fe-   (Chaplin & Chaplin, 1981). Additionally, the African danaid butterfly Danaus chrysippus, having a cardenolide-resistant Na + /K + -ATPase (Petschenka et al., 2013), developed faster and produced larger adults when reared on Calotropis procera containing cardenolides compared to caterpillars raised on Tylophora spp. lacking cardenolides (Rothschild et al., 1975). Furthermore, Zygaena filipendulae larvae reared on the plant Lotus corniculatus containing cyanogenic glucosides developed faster, and the larvae showed decelerated development when reared on transgenic L. corniculatus free of cyanogenic glucosides (Zagrobelny et al., 2007).

| D ISCUSS I ON
In contrast to these studies, caterpillar growth of the milkweed butterfly species Euploe core, D. plexippus, and D. gilippus was unaffected by cardenolides across eight Asclepias species ranging from very low to very high cardenolide contents (Petschenka & Agrawal, 2015). Notably, all the studies mentioned above focused on insect feeding performance on intact plants or plant organs such as leaves or seeds that naturally represent highly complex diets (but see Bowers, 1984). This could be one reason why contradicting outcomes in response to the same class of chemical compounds were observed even within related plant species. Here, we used an artificial diet to control for variation across dietary treatments rigorously.
We showed that cardenolides had a positive effect on growth in O. fasciatus and C. nerii, both of which may be categorized as dietary specialists feeding on seeds of Asclepias spp. and seeds of Nerium oleander, respectively. We speculate that the positive impact on growth upon exposure of toxins may be due to selection on the resistant Na + /K + -ATPases to function optimally in a "toxic environment," that F I G U R E 5 Reproductive success of Oncopeltus fasciatus in the presence or absence of dietary cardenolides. Each horizontal bar represents the mean (±SE) total number of hatchlings produced by O. fasciatus females until death. Larvae were either raised on toxic diet or control diet until adulthood. A group of bugs stayed on the respective diets after reaching adulthood (left; n for control = 13, n for toxic = 12), while the other group was transferred to sunflower seeds (right; n = 13 for control, n = 10 for toxic). Females producing no viable eggs at all were excluded from the analysis. Within the same group, different letters indicate significant differences. Dots and triangles represent jittered raw data is, in the body tissues of a milkweed bug storing large amounts of cardenolides. In other words, there could be functional trade-offs of cardenolide-adapted Na + /K + -ATPases in a physiological environment that is devoid of cardenolides, a phenomenon that could be called "evolutionary addiction." Alternatively, better growth could be due to increased consumption of the larvae mediated by cardenolides as phagostimulants (Pantle & Feir, 1976). Nevertheless, our excretion data hint toward equal consumption of diet regardless of dietary cardenolide concentration ( Figure S4). Since endogenous and sequestered defenses may trade-off (Engler-Chaouat & Gilbert, 2007), the lack of cardenolides for sequestration could lead to a higher investment towards endogenous defenses (i.e., defensive secretions) and therefore impair growth on the cardenolide-free control diet. Although such a trade-off has been suggested to occur in the milkweed bug Lygaeus equestris (Havlikova et al., 2020), metathoracic scent glands were shown to be reduced in O. fasciatus and other milkweed bugs (Aldrich, 1988;Schaefer, 1972) making this explanation rather unlikely.
The milkweed bug species A. longiceps is specialized on plants producing no cardenolides such as Platanus spp. or Ulmus spp.
However, due to its evolutionary history, A. longiceps possesses resistant Na + /K + -ATPases, but has lost the ability to sequester cardenolides (Bramer et al., 2015). In contrast, S. pandurus possesses resistant Na + /K + -ATPases, and feeds on a wide array of host plants (Péricart, 1998;Vivas, 2012), including cardenolide producing species such as Nerium oleander and species of Calotropis from which it sequesters cardenolides (Abushama & Ahmed, 1976;Von Euw et al., 1971). For both species, dietary cardenolides did not influence growth, that is, they grew equally well on all diets.
The lack of a positive effect of dietary cardenolides on growth in A. longiceps may be associated with the inability of this species to sequester cardenolides. Accordingly, its Na + /K + -ATPases may have undergone a different selection regime compared to the sequestering species. In other words, their putative suite of Na + /K + -ATPases may have adapted to the absence of cardenolides in the body tissues secondarily. Alternatively, or in addition, there could be further physiological mechanisms involved mediating between sequestration and growth such as gaining energy from metabolizing sequestered toxins. Nevertheless, the latter seems rather unlikely as an explanation for increased growth in O. fasciatus and C. nerii, since the sugar moieties in digitoxin are dideoxy sugars not known to be easily metabolized as an energy resource (Liu & Thorson, 1994) and ouabain (carrying a rhamnose moiety) was found to be sequestered as such (Scudder & Meredith, 1982 pressures in this species and may thus interfere with the pattern of amino acid substitutions across the different Na + /K + -ATPases or lead to differential expression of the Na + /K + -ATPase genes across different tissues. The outgroup species, P. apterus, belongs to a different family (Pyrrhocoridae) and is not adapted to cardenolides, that is, it has a cardenolide-sensitive Na + /K + -ATPase and is not able to sequester cardenolides (Bramer et al., 2015). The lack of a resistant Na + /K + -ATPase most likely explains why larval growth in this species was compromised substantially by dietary cardenolides. Alternatively, or in addition, reduced growth could be due to feeding deterrence mediated by dietary cardenolides as indicated by the reduced amount of excretion products observed during our feeding trial.
Although specialist insects can successfully feed on toxic host plants, it is generally expected that the underlying resistance traits incur costs via trade-offs, which are expected to depend on the ecological context and the molecular mechanisms involved (Peterson et al., 2016;Smilanich et al., 2016). As O. fasciatus raised on cardenolide-containing diet developed faster into adults and had a longer lifespan compared to those raised on cardenolide-free diet, cardenolide exposure clearly influences the fecundity-longevity trade-off. A faster development and an extended lifespan both indicate higher fitness. Nevertheless, individuals exposed to dietary cardenolides produced a lower number of hatchlings contradicting higher fitness. This apparent disadvantage, however, could be alleviated by maternal transfer of cardenolides to the eggs mediating protection against predators (Newcombe et al., 2013). Moreover, we found that feeding on a non-toxic diet (i.e., sunflower seeds) as adults can compensate for this effect. This likely resembles the natural situation since O. fasciatus larvae feed and cluster around on Asclepias seeds or seedpods, while adults disperse and forage on various plants (Feir, 1974). Contrary to our findings, it was shown that male O. fasciatus fed with A. syriaca seeds invested in reproduction at the expense of survival when compared to those fed with sunflower seeds (Attisano et al., 2012). In this study, however, only the diet of male specimens was manipulated and female O. fasciatus were exclusively fed with sunflower seeds. Moreover, due to the use of milkweed seeds, it is not possible to directly attribute the observed effects to cardenolides. In conclusion, it seems likely that cardenolides exert a positive effect on overall fitness in O. fasciatus, which is in disagreement with theory predicting costs of sequestration.
Dealing with xenobiotics can require energy allocation towards metabolism or can cause pleiotropic effects due to particular molecular mutations that might confer a selective advantage in the presence of xenobiotics but incur a cost in their absence (Coustau et al., 2000;Mauro & Ghalambor, 2020). Although it is unclear how the function of Na + /K + -ATPase could be related to growth and longevity in milkweed bugs, the enzyme is involved in many physiological processes apart from being a cation carrier suggesting many unknown non-canonical functions of Na + /K + -ATPase (Liang et al., 2007), which could provide a mechanistic link. In O. fasciatus, the three gene copies (α1A, α1B, and α1C) encoding α-subunit of cardenolideresistant Na + /K + -ATPases have diverse functions (Lohr et al., 2017) and duplicated α1 gene copies do not only vary in number and identity but also show specific expression patterns in different body tissues (Dobler et al., 2019;Yang et al., 2019;Zhen et al., 2012) allowing for complex regulation.
In conclusion, reduced growth in the absence of cardenolides in highly specialized milkweed bugs with cardenolide-resistant Na + /K + -ATPases suggests a novel type of physiological cost arising in the absence of plant toxins. Mechanistically, this cost could be due to negative pleiotropic effects mediated by resistant Na + /K + -ATPases not functioning optimally in a physiological environment lacking cardenolides. Furthermore, the observed effects of cardenolides on the fecundity-longevity trade-off probably leading to increased fitness in O. fasciatus may be due to optimized resource allocation under the influence of sequestered cardenolides. Our study suggests a concept that is contradictory to the general assumption that sequestered plant toxins produce physiological costs and our results indicate a further level of coevolutionary escalation.

ACK N OWLED G M ENT
We thank Sabrina Stiehler for technical support, Martin Kaltenpoth for advice on rearing Pyrrhocoris apterus, and Tobias Züst for discussion. We are greatly indebted to Andreas Vilcinskas and the Justus Liebig University Giessen for infrastructural support. This research was funded by DFG grant PE 2059/3-1 to G.P. and the LOEWE Program of the State of Hesse by funding the LOEWE Center for Insect Biotechnology and Bioresources.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
All raw data reported in this manuscript were archived in Dryad: https://doi.org/10.5061/dryad.q2bvq 83m3.