A simple artificial diet for feeding and sequestration assays for the milkweed bugs Oncopeltus fasciatus and Spilostethus saxatilis

Insect artificial diets are not only an important tool for mass rearing, nutritional research, and maintaining laboratory colonies but also for studying insect‐plant interactions. For herbivorous insects able to sequester plant toxins, feeding and sequestration assays based on artificial diet allow for the investigation of physiological, ecological, and evolutionary questions which may be difficult to study using real plants representing complex chemical environments. We developed a simple artificial diet, consisting of sunflower meal pressed into pills, for the milkweed bugs Oncopeltus fasciatus (Dallas) and Spilostethus saxatilis (Scopoli) (Heteroptera: Lygaeidae), which are capable of sequestering cardenolides and colchicum alkaloids, respectively. We assessed insect performance, suitability of the diet for sequestration assays, and its shelf life. Compared to sunflower seeds which are widely used as a laboratory maintenance diet for milkweed bugs, no differences were found in terms of weight development, presence of deformities, speed of development, or mortality. Importantly, after feeding O. fasciatus and S. saxatilis sunflower pills enriched with crystalline ouabain (cardenolide) or colchicine (colchicum alkaloid), respectively, sequestration was observed in both species. Moreover, as a prerequisite to test ecological hypotheses, our method allows for adequate concentration control and homogenous distribution of toxins across the diet. Under relatively warm conditions (27 °C and 60% r.h.), the new diet was stable for up to 10 days when used for feeding assays with adult bugs. Therefore, studies focusing on the role of plant toxins in predator–prey interactions and plant defense, but also insecticide research could benefit from using this approach.

Compared to natural food sources such as plants, where multiple concomitant compounds may interfere with the effects of a substance of interest, artificial diets allow for

O R I G I N A L A R T I C L E A simple artificial diet for feeding and sequestration assays for the milkweed bugs Oncopeltus fasciatus and Spilostethus saxatilis
Laura Espinosa del Alba | Georg Petschenka improved experimental manipulation. As specific toxins, either alone or in combination, can be easily included, new research questions can be addressed that were previously out of reach. Moreover, artificial diets reduce the time, effort, space, and cost required for natural food sources (Hervet et al., 2016) and remove the need to collect or maintain a supply of demanding, sensitive, or bulky dietary resources, such as living plants.
For seed-eating milkweed bugs (Heteroptera: Lygaeidae, Lygaeinae), artificial diets have been used to address various physiological, ecological, and evolutionary questions (Ramoska & Todd, 1985;Jones et al., 1986;Cohen, 2000;Igarashi et al., 2013). One remarkable characteristic of the entire subfamily Lygaeinae is the ability to sequester plant toxins in defense against predators (Bramer et al., 2015) rendering them an important model system for the study of insect-plant coevolution (Agrawal et al., 2022;Petschenka et al., 2022). Milkweed bugs are primarily associated with plant species in the Apocynaceae, which commonly produce cardenolides (Bramer et al., 2015), potent inhibitors of the ubiquitous animal enzyme Na + /K + -ATPase (Agrawal et al., 2012). Despite a close association with Apocynaceae likely driven by the acquisition of defenses (Beran & Petschenka, 2022), species such as Oncopeltus fasciatus (Dallas) also feed on unrelated nontoxic plants (Beck et al., 1958;Ralph, 1976) which may explain why milkweed bugs can be easily reared on sunflower seeds over generations (Feir, 1974;Petschenka et al., 2022).
Remarkably, some species of milkweed bugs also sequester unrelated plant compounds such as colchicum alkaloids from Colchicum autumnale L., as reported for the palaearctic species Spilostethus saxatilis (Scopoli) . Colchicine, a compound occurring in C. autumnale, binds to tubulin and therefore interferes with microtubule dynamics, disrupting mitosis (Molad, 2002). The use of an artificial diet suitable for species sequestering different types of plant compounds allows for testing hypotheses about the specificity of sequestration, physiological costs, and trade-offs between life-history traits.
For O. fasciatus various diets including gels, powders, and semi-solid diets have been developed, but all of them resulted in suboptimal growth and higher mortality compared to the natural diet (Scheel et al., 1957). To study sequestration, Duffey et al. (1978) embedded various cardenolides into sunflower seeds by applying the compounds directly as a solution in chloroform-methanol to the surface of the seed. Jones et al. (1986) developed an agar-based artificial diet for O. fasciatus, that was adapted in our laboratory and successfully used to study the effect of dietary cardenolides on growth, sequestration, and lifehistory traits in several milkweed bug species (Pokharel et al., 2021). However, some challenges remained, as milkweed bug species like S. saxatilis showed limited performance on this diet.
Here, our objective was to develop a new, simple type of artificial diet for feeding and sequestration assays with milkweed bugs, which is suitable for O. fasciatus and S. saxatilis both sequestering chemically unrelated classes of plant toxins (i.e., cardenolides and colchicum alkaloids) and with the potential to be applied to other species with similar nutritional needs and feeding habits. Specifically, we tested the artificial diet's effectiveness on insect performance, its suitability for sequestration assays, and its shelf life.

Dietary components
Typical artificial diets are composed of at least five ingredients (Fortes et al., 2006;Hervet et al., 2016;Pourkhatoon et al., 2016;Bajonero & Parra, 2017;Montoro et al., 2020), and the diet developed for O. fasciatus by Jones et al. (1986) contains more than 10 ingredients. However, as several species of Lygaeinae including O. fasciatus and S. saxatilis can be successfully maintained using only husked sunflower seeds as a nutritional resource (Bramer et al., 2015;Petschenka et al., 2022), we opted for a simplified approach and used sunflower meal as the only dietary ingredient of our diet. We pressed the meal into pill form to mimic the physical characteristics of a seed, which may be necessary to trigger feeding. In addition to providing a dietary resource, sunflower meal serves as an embedding matrix for the toxins of interest in a crystalline form.
In contrast to the organic sunflower seeds that we used for the maintenance of milkweed bug colonies (Alnatura, Darmstadt, Germany), the organic sunflower meal (DM-Drogerie Markt, Karlsruhe, Germany; Ölmühle Sailer, Lochau, Austria) had approximately only 2/3 of the energy and 1/3 of the fat, but approximately 1.5× more protein and 3× more fiber (based on a fresh weight basis; the carbohydrate content was similar). Specifically, the nutritional values (per 100 g fresh weight) of the organic sunflower meals used were as follows (according to the information provided by the manufacturer): 1621 kJ of energy, 13.2-15.5 g fat, 11.3-14.2 g carbohydrates, 17.3 g dietary fiber, 37.7-42.2 g protein, 0.01 g salt, 11.2 mg vitamin E, 455 mg magnesium, 17.3 mg zinc, 4.9 mg manganese, and 70 μg selenium.

Artificial diet preparation
We used handheld tablet presses (LFA Tablet Presses, Düsseldorf, Germany) to produce pills (800 mg each) in batches of six. We produced six pills at a time due to the limited capacity of the containers used for mixing sunflower meal and toxins (see below). All weight measurements were carried out using an analytical balance. For each batch, we used 5 g of organic sunflower meal (6 × 800 mg per pill + 200 mg safety margin) and added the desired quantity of toxins in a crystalline form. As sunflower meal has a water content of 10% (FEDNA, 2019), concentrations of toxins were calculated on a dry weight basis (i.e., calculations were based on a total dry weight of 4.5 g sunflower meal). After weighing the desired amounts of toxins into a 25-mL headspace screw-cap glass vial, the sunflower meal was added using a funnel. Next, we carried out a two-step mixing procedure. First, the vial was vortexed 3 × 5 s to disperse the toxins throughout the sunflower meal and the mixture was subsequently transferred into a 15-mL Falcon tube. Second, the 15-mL tube was agitated in a Fast-Prep 24 homogenizer (MP Biomedicals, Eschwege, Germany) for two 45 s cycles at 6.5 m s −1 to achieve a homogeneous mixture. Lastly, 800 mg of the mixture was poured into the tablet press using a funnel and pressed to obtain a pill (see Figure 1). We will refer to the new diet as sunflower pill(s) hereafter.

Origin of insects
Oncopeltus fasciatus came from a long-term laboratory colony (originally from the USA) obtained from the University of Hamburg in 2015. Bugs were housed in plastic boxes (19 × 19 × 19 cm) covered with gauze, raised on husked sunflower seeds and supplied with water provided in Eppendorf tubes plugged with cotton wool. In addition, boxes were equipped with pieces of cotton wool for oviposition. Colonies were maintained in environmental chambers at 28 °C and 60% r.h. at a L16:D8 cycle (Binder KBWF 240, Tuttlingen, Germany). Specimens of S. saxatilis were raised from eggs obtained from field-collected adults (Berghausen, Baden-Württemberg, Germany) maintained on husked sunflower seeds under ambient conditions. After oviposition, eggs were collected and kept under the same environmental conditions as O. fasciatus until they were used for the experiments.

Insect performance
We assessed the performance of O. fasciatus and S. saxatilis when fed pure sunflower pills (i.e., no toxins added) compared to sunflower seeds (i.e., the laboratory maintenance diet). For both species, groups of five freshly hatched nymphs were placed in a Petri dish (90 mm diameter, with vents) lined with filter paper and equipped with a 2-mL Eppendorf tube containing water (see above). In addition, we provided a small piece of tissue to provide shelter. Fifty insects were used for each treatment and each species (i.e., 10 Petri dishes per treatment and species). All Petri dishes were placed inside a growth chamber (Binder KBWF 240) at 27 °C, 60% r.h. at a L16:D8 cycle, in a spatially randomized fashion.
The following life-history traits were assessed: (1) weight development, based on the average weight of all individuals in a Petri dish (anaesthetized with CO 2 ) from first instars to recently molted adults, assessed weekly, (2) the presence of potential deformities, also assessed on a weekly basis, (3) speed of development, recorded as the day (S. saxatilis) or week (O. fasciatus) when all insects in a Petri dish had reached adulthood, and (4) cumulative mortality from first instars to recently molted adults.
We first tested O. fasciatus and included two additional diets to assess how the performance on the new diet compares to previously established approaches. Besides sunflower pills (one pill per Petri dish) an amount of sunflower seeds equivalent based on kcal (9-14 seeds, ca. 509 mg), the artificial diet by Jones et al. (1986) offered as described by Pokharel et al. (2021), and methanol-soaked seeds were tested. The latter is an adaptation of the soaking method described by Duffey et al. (1978), i.e., sunflower seeds were immersed in methanol for 24 h and subsequently dried at 60 °C for 48 h. Here, only pure methanol was used to test the diet itself and not the effect of toxins. During previous experiments, we observed that first instars had difficulties with piercing the parafilm membrane covering the diet after Jones et al. (1986;see Pokharel et al., 2021). For this reason, we included an additional test series with this diet but starting with third-fourth instars, which were raised on sunflower seeds before.
Considering the results obtained with O. fasciatus and the reduced performance of S. saxatilis with the diet adapted from Jones et al. (1986) in a preliminary trial, only the new diet (i.e., pure sunflower pills) and sunflower seeds (i.e., the laboratory maintenance diet) were tested for S. saxatilis using an identical experimental setup as described for O. fasciatus. Diets were replaced when fungal growth on the surface of the pills, desiccation (darkening and hardening), or disintegration (large cracks) occurred (three or four replacements in total). Within a species, all dietary treatments were carried out simultaneously.

Sequestration experiments
We assessed three parameters to evaluate the suitability of artificial seeds (sunflower pills) for sequestration assays: (1) sequestration of toxins from the artificial diet by milkweed bugs, (2) concentration accuracy of the incorporated toxins, and (3) toxin distribution within the pill ensuring a homogeneous toxin distribution to achieve a comparable exposure independent of the feeding location of the bugs.
We carried out the following experiment to assess sequestration of the cardiac glycosides ouabain and digitoxin (both Sigma-Aldrich, Taufkirchen, Germany) in O. fasciatus, and ouabain and the alkaloid colchicine (Roth, Karlsruhe, Germany) in S. saxatilis. Ouabain and digitoxin represent the polarity range of cardiac glycosides (Scudder & Meredith, 1982) and colchicine is the main alkaloid of C. autumnale seeds (Jung et al., 2011). For each species, recently molted adults raised on sunflower seeds at 27 °C, 60% r.h. (L16:D8; KBWF 720 for O. fasciatus and KBWF 240 for S. saxatilis) were starved overnight. Insects were then placed in groups of five in a 90-mm-diameter Petri dish lined with filter paper and equipped with one sunflower pill (one of three toxic diets or a non-toxic control, see below) and a water source (see above) under the same environmental conditions as described above. Insects were processed after 3, 7, and 10 days to quantify the amount of sequestered toxins (n = 12 for each diet at each time point).
Oncopeltus fasciatus diets were comprised of ouabain pills, pills with an equimolar concentration of digitoxin, pills containing both toxins equimolar, and control pills without toxins. Diets offered to S. saxatilis consisted of colchicine pills, pills with an equimolar concentration of ouabain (the same as in O. fasciatus), pills containing both toxins equimolar, and non-toxic control pills (concentrations shown in Table 1). Concentrations tested, besides being equimolar, fall within the natural toxin content range of both C. autumnale and Asclepias spp. seeds (Isman, 1977;Poutaraud & Girardin, 2002;Jung et al., 2011). The experiment aimed to assess competitive inhibition between ouabain and colchicine and ouabain and digitoxin, but here, we are only presenting sequestration data for O. fasciatus on ouabain and S. saxatilis on colchicine pills after 7 days as a proof of concept for the suitability of our diet for sequestration assays. Nevertheless, data on concentration accuracy and toxin distribution within pills as well as data on shelf life are included for all toxic diets (see next paragraph and section on shelf life, below).
To assess concentration accuracy and distribution of toxins within pills, 12 intact pills across all dietary treatments mentioned above were analyzed. In order to reach a minimum of seven replicates per toxin (i.e., ouabain, digitoxin, and colchicine), we included nine extra pills which had also not experienced bug feeding and covered additional concentrations (see Table 1). Each pill was split into three chunks of 100 mg that were subsequently analyzed by high-performance liquid chromatography (HPLC) individually. Pills containing two toxins were analyzed separately for each type of toxin. Concentration accuracy was calculated as the average of all pill means (three-chunk means) from the same pill type, and expressed as a percentage difference from the expected concentration. Toxin distribution within pills was calculated as the average difference between each 100 mg chunk and its respective three-chunk mean.

Insect and pill extraction
Freeze-dried insects without digestive tracts (removed by dissection before freezing) were weighed and extracted individually using 1 mL of methanol and approximately 30 zirconia/silica beads (2.3 mm diameter; BioSpec Products, Bartlesville, OK, USA) in 2-mL plastic screw-cap vials (Sarstedt, Nümbrecht, Germany). Samples were homogenized twice for 45 s at a speed of 6.5 m s −1 in a Fast Prep homogenizer (MP Biomedicals, Solon, OH, USA), followed by centrifugation at 16100 g for 3 min at 22 °C. The supernatant was transferred into a new 2-mL screw-cap vial, and a second extraction cycle following the same procedure was carried out. Pooled Colchicine (n = 2) 2 Digitoxin (n = 2) 3.83 Ouabain + colchicine (n = 2) 3.65 + 2 Digitoxin + ouabain (n = 2) 3.83 + 3.65 Digitoxin + ouabain (n = 3) 3.9 + 3.9* Colchicine (n = 3) 8* Colchicine (n = 3) 24* supernatants were evaporated at room temperature under a flow of nitrogen gas. Dried samples were then re-suspended in 250 μL methanol, agitated once for 45 s at a speed of 6.5 m s −1 in the FastPrep homogenizer, and centrifuged at 16100 g for 3 min at 22 °C. Finally, samples were filtered through a 0.45 μm syringe nylon filter (Roth) into an HPLC vial. Each individual freeze-dried pill was crushed, and three separate chunks of 100 mg each were taken from different areas of the pill and individually transferred to 15-mL Falcon tubes. Approximately 90 zirconia/silica beads and 15 mL of methanol were added to each tube, and samples were homogenized as described above, followed by centrifugation at 4000 g for 12 min at 22 °C. An aliquot of 150 μL was taken from the supernatant of each tube, transferred to a plastic 2-mL screw-cap vial, and subsequently evaporated and prepared for HPLC analysis as described above.

HPLC analysis
Cardenolides were analyzed on an Agilent 1260 Infinity II HPLC equipped with an EC 150/4.6 NUCLEODUR C18 Gravity column (3 μm, 150 × 4.6 mm; Macherey-Nagel, Düren, Germany) using diode-array detection. Toxins were eluted at a constant flow of 0.7 mL/min with a gradient of acetonitrile and water as follows: for 0-2 min at 16% acetonitrile, 2-25 min from 16 to 70%, 25-30 min from 70 to 95%, and 30-35 min at 95%, followed by 10 min of reconditioning at 16% acetonitrile. Ultraviolet absorbance spectra were recorded between 210 and 350 nm. Peaks showing a characteristic single symmetrical absorption maximum between 216 and 222 nm, corresponding to an unsaturated lactone functional group, were considered cardenolides and quantified at 218 nm. Peaks in different samples with retention time differences of <0.2 min were considered the same compound. Concentrations of cardenolides were calculated using an external digitoxin and ouabain octahydrate calibration curve at 5, 10,25,50,100,150,200,250,300, and 350 μg mL −1 . Chromatograms were analyzed with the Agilent OpenLab software v.3.4.0.
Colchicine and related metabolites (colchicoids) were analyzed on an Agilent 1100 series HPLC using the same column as above. Toxins were eluted at a constant flow of 0.7 mL/min with a gradient of acetonitrile and water containing 0.25% of phosphoric acid as follows: for 0-2 min at 10% acetonitrile, 10 min at 40% acetonitrile, 15 min at 80% acetonitrile, and 16 min at 10% acetonitrile, with 5 min of reconditioning at 10% acetonitrile. Ultraviolet absorbance spectra were recorded between 190 and 400 nm. Peaks showing a characteristic absorption maximum between 230-245 and 350-355 nm, resembling the absorption spectra of colchicine, were considered colchicoids and quantified at 350 nm. Peaks in different samples with retention time differences of <0.2 min were considered the same compound. Colchicoid concentrations were calculated using an external colchicine calibration curve at 5, 10,25,50,100,150,200,250,300, and 350 μg mL −1 . Chromatograms were analyzed with the Agilent ChemStation software v.B.04.03.

Shelf life
We assessed macroscopic signs of microbial growth, desiccation, and disintegration after 3, 7, and 10 days at 27 °C and 60% r.h. Four pills were evaluated per species: for S. saxatilis (1) control, (2) 3.65 mg g −1 ouabain octahydrate, (3) 2 mg g −1 colchicine, and (4) 3.65 mg g −1 ouabain octahydrate + 2 mg g −1 colchicine; for O. fasciatus (1) control, (2) 3.65 mg g −1 ouabain octahydrate, (3) 3.83 mg g −1 digitoxin, and (4) 3.83 mg g −1 digitoxin + 3.65 mg ouabain octahydrate. The level of replication was n = 10 after 3 days, n = 7 after 7 days and n = 4 after 10 days per species, sample size decreasing due to insects being processed for toxin quantification and subsequent disposal of pills. Ouabain octahydrate will be referred to as ouabain hereafter. The pills accessed by S. saxatilis were made from the organic sunflower meal from DM-Drogerie Markt and placed in a KBWF 240 growth chamber. The pills accessed by O. fasciatus consisted of sunflower meal produced by Ölmühle Sailer and were placed in a KBWF 720 growth chamber.

Statistical analysis
Statistical analyses were carried out using R software v.4.1.2. To assess weight development from first instar to adulthood, we ran an LME model that investigates the effect of diet on weight from the nlme package (R Core Team, 2022). We used a 'corA1' correlation structure for equally spaced measurements and included the interaction of diet with time, time as a fixed effect, and subject (i.e., Petri dish) as a random effect, followed by a least-squares means post hoc analysis (LSM; function emtrends) with Holm P-value adjustment method in the case of O. fasciatus (five diet approaches tested). For comparing the speed of development across diets, we run a Kruskal Wallis test followed by a post hoc analysis using a Dunn test with the Benjamini & Hochberg P-value adjustment method for O. fasciatus and a t-test for S. saxatilis (only two diets compared). As deformities and cumulative mortality were count data, we performed a GLM logistic regression for both species followed by a Tukey honestly significant difference (HSD) post hoc analysis in the case of O. fasciatus. For all parameters, the level of replication was n = 10 Petri dishes per treatment and species, with each Petri dish containing five insects initially. All data presented are untransformed raw data.

Insect performance
There were no differences in weight development for both milkweed bug species between sunflower pills and sunflower seeds. For O. fasciatus, weight development at any time point was not different between sunflower pills and sunflower seeds (t = −0.19-1.72, d.f. = 318, P ≤ 0.85, LSM: pill vs. sunflower seeds, P = 0.76; Figure 2A). Regarding the other three dietary approaches -diet after Jones since first instar, diet after Jones since third-fourth instar, and methanol-soaked seeds -, weights were lower compared to sunflower seeds (LSM: all three dietary approaches vs. sunflower seeds: P < 0.0001). We obtained identical results for S. saxatilis, which grew equally well on sunflower seeds and sunflower pills (t = −0.35, d.f. = 160, P = 0.73; Figure 2B).
Regarding speed of development and cumulative mortality, no differences were found between bugs of either species raised on sunflower pills vs. sunflower seeds. For O. fasciatus, speed of development (Kruskal-Wallis χ 2 = 14.877, d.f. = 2, P < 0.001; Dunn test: pill vs. sunflower seeds: z = 0.66, P = 0.51) and cumulative mortality (z = 0.62-6.23, P ≤ 0.54; Tukey test: pill vs. sunflower seeds, z = 0.62, P = 0.97) did not differ between the two diets ( Table 2). None of the individuals raised on the diet after Jones et al. (1986) since the first instar reached adulthood, and only 62.1% raised on this diet since the third-fourth instar. Insects raised on methanolsoaked seeds reached adulthood after week 7 but needed much longer than sunflower-seed-and pill-reared insects (Dunn test: methanol-soaked vs. sunflower seeds, z = −2.96, P < 0.01; methanol-soaked vs. pill, z = 3.62, P < 0.001). In line with these results, the cumulative mortality of the additional three dietary approaches was higher than on sunflower seeds (Tukey test: vs. diet after Jones since first instar, z = 6.23, P < 0.001; vs. methanol-soaked seeds, z = 2.90, P = 0.03; vs. diet after Jones since third-fourth instar, z = 2.73, P = 0.049).
For S. saxatilis, speed of development (t = −0.32, d.f. = 16.41, P = 0.76) and cumulative mortality (z = −1.46, P = 0.14) did not differ between sunflower pills and sunflower seeds (Table 2). No deformities were observed for O. fasciatus in any treatment except in four insects from methanol-soaked seeds (Table 2). In S. saxatilis, we observed a few cases of deformities (Table 2) in sunflower-pillraised insects, but there was no significant difference from individuals raised on sunflower seeds (z = 0.62, P = 0.54).

Sequestration of toxins
Insects were not only able to sequester but also appeared to upconcentrate toxins from the new artificial diet (Figure 3). Oncopeltus fasciatus feeding on sunflower pills with an aimed concentration of 3.65 mg g −1 dry weight sequestered 4.43 μg ouabain mg −1 dry weight (1.2× more; Figure 3A) whereas S. saxatilis feeding on sunflower pills with an aimed concentration of 2 μg colchicine mg −1 dry weight sequestered 2.84 μg mg −1 dry weight of colchicine and colchicine-like metabolites (1.4× more; Figure 3B). Spilostethus saxatilis feeding on non-toxic sunflower pills contained 0.28 μg colchicine mg −1 dry weight due to maternal transfer from field-collected females via the eggs.

Concentration accuracy and toxin distribution
Chemical analyses of individual pills revealed total amounts of digitoxin, and especially ouabain, closely matching the expected concentrations (Table S2, Figure S1). An average difference of 3.8% compared to the expected concentration was observed for ouabain and a 12.3% average difference compared to the expected concentration was observed for digitoxin (see Figure S1 for details). For colchicine, we tested a much larger range of concentrations according to experimental needs (dose range: 2-24 mg g −1 ), and a 27.6% average difference between the final concentration and the expected concentration was obtained. Specifically, at low concentrations (2 mg g −1 dry diet), the difference compared to the expected concentration was 40.7%, whereas for higher concentrations (8 and 24 mg g −1 of the dry diet), the difference was 14.8 and 14.2%, respectively. Finally, when considering toxin distribution, the distribution difference within pills compared to the final concentration F I G U R E 2 Mean (± SE; n = 10) weight (mg) of Oncopeltus fasciatus and Spilostethus saxatilis on various artificial diets and sunflower seeds. (A) Weight development of O. fasciatus raised on sunflower seeds (blue), sunflower pills (red), diet after Jones et al. (1986) since the first instar (N1; green) and since the third-fourth instar (N3-N4; yellow), or sunflower seeds soaked in methanol after evaporation (orange). (B) Weight development of S. saxatilis raised on sunflower seeds (blue) or sunflower pills (red). was 16.4% for colchicine, and 6.4 and 2.6% for ouabain and digitoxin, respectively (Tables S1 and S3, Figure S1).

Shelf life
We assessed the shelf life of sunflower pills made from two different brands of sunflower meal over 10 days at 27 °C and 60% r.h (Table S4). Pills made out of sunflower meal from DM-Drogerie Markt that were accessed by S. saxatilis and placed in a KBWF 240 growth chamber showed the following results: no macroscopic microbial growth was observed after 7 days in any of the four pill groups (control, ouabain, colchicine, and ouabain + colchicine; n = 7 each). After 10 days, microbial growth was only evident in 50% of the colchicine pills and 37.5% of the ouabain + colchicine pills (n = 4 each). Furthermore, no disintegration was observed in any pill group, and desiccation was only observed after 10 days, appearing as a slight darkening in all pill groups (Table S4).
Pills made from sunflower meal from Ölmühle Sailer that were offered to O. fasciatus and placed in a KBWF 720 growth chamber showed the following results: no macroscopic microbial growth was observed after 3 days in any of the four pill groups tested (control, ouabain, digitoxin, and digitoxin + ouabain; n = 10 each). After 7 days, macroscopic microbial growth was observed in 42.9-57.1% of the control, digitoxin, and digitoxin + ouabain pills, but not in the pills containing ouabain only (n = 7 each). After 10 days, 25-50% of all pill groups showed microbial growth (n = 4 each). In terms of disintegration and desiccation, cracks or border damage were present after 3 days in all pill groups, and a slight darkening also appeared after 3 days. Moreover, size reduction began after 7 days in all pill groups (Table S4).

DISCUSSION
Apart from its support of general insect performance, suitability for experimental purposes, and having sufficient T A B L E 2 Additional performance parameters (%) of Oncopeltus fasciatus and Spilostethus saxatilis on artificial diets and sunflower seeds. For both species n = 10 per treatment.

F I G U R E 3
Sequestration of toxins from the new artificial diet (sunflower pills): mean (± SE) concentration (μg toxin per mg dry weight) of (A) ouabain sequestered by adult Oncopeltus fasciatus, and (B) colchicine sequestered by adult Spilostethus saxatilis. All insects were raised on sunflower seeds and fed for 7 days on the corresponding sunflower pill. Dotted yellow lines represent the toxin concentration for each type of pill (3.65 mg g −1 ouabain for O. fasciatus; 2 mg g −1 colchicine for S. saxatilis). The traces of colchicine in S. saxatilis fed with non-toxic pills (control insects) resulted from the maternal transfer of the toxin from field-collected specimens. As O. fasciatus came from a long-term, toxin-free laboratory colony, no maternal transfer of the toxin was possible.
shelf-life, an ideal insect artificial diet should be easy to prepare. Whereas artificial diets are normally composed of at least five ingredients (Fortes et al., 2006;Hervet et al., 2016;Pourkhatoon et al., 2016;Bajonero & Parra, 2017;Montoro et al., 2020), the new diet presented here only consists of a single ingredient (commercially available sunflower meal) and can be prepared with standard laboratory equipment and hand-held tablet presses. We tried to keep our diet as simple as possible to avoid introducing compounds which potentially interfere with toxins added for experimental reasons.
We assessed performance of O. fasciatus and S. saxatilis on the newly developed sunflower pills over their entire development. We used sunflower seeds as a control because they represent a common laboratory maintenance diet devoid of toxins, a prerequisite for sequestration assays that is not met in the natural diets of both species (i.e., Asclepias spp. and C. autumnale). Furthermore, sunflower seeds have been successfully used to maintain O. fasciatus for at least 4 years with approximately 10-12 generations a year in our laboratory, and it was shown before that they represent a suitable diet for S. saxatilis and additional milkweed bug species . Although the new diet differs remarkably from sunflower seeds regarding energy and nutrients (Alnatura, 2022;Ölmühle Sailer, 2022), weight development, cumulative mortality, deformities, and speed of development were not different when both milkweed bug species were raised on the new diet compared to sunflower seeds. Altogether, milkweed bugs raised on the new artificial diet did not show impaired performance as was reported for other artificial diets (Scheel et al., 1957;Jones et al., 1986).
In other heteropterans, the use of dry diets containing sunflower seeds, sunflower oil, and ground lyophilized green beans provided advantages such as continuous rearing throughout the year (avoiding diapause; Pourkhatoon et al., 2016), enhanced performance compared to other artificial diets (Fortes et al., 2006), and higher mean fecundity (Mendoza et al., 2016), respectively. Nevertheless, problems remained, such as longer developmental time, lower fertility (Fortes et al., 2006;Pourkhatoon et al., 2016), and lower adult weight (Mendoza et al., 2016).
Preparing sunflower pills and including toxins mostly relies on standard laboratory equipment. We demonstrated that O. fasciatus and S. saxatilis can sequester ouabain and colchicine included into our diet in a crystalline form. The accumulation of toxins above diet concentrations that our results seem to indicate in both cases has also been reported in the same model species (i.e., O. fasciatus feeding on Asclepias syriaca L. seeds (Agrawal et al., 2022) as well as in other sequestration systems such as Longitarsus sp. beetles sequestering pyrrolizidine alkaloids (Dobler et al., 2000) and the lepidopteran Zygaena filipendulae (L.) sequestering cyanogenic glucosides (Zagrobelny et al., 2014). The ability to sequester crystalline toxins is probably due to extra-oral dissolution of toxins by bug saliva injected into the diet during feeding, as the continued secretion of saliva and reingestion of salivary -emulsified -material is essential to feeding (Miles, 1959;Feir & Beck, 1963;Bongers, 1969). Besides cardenolides and colchicum alkaloids, our diet probably allows for incorporating any kind of crystalline toxin as well as combinations of toxins and might therefore be used to approach a variety of physiological, evolutionary, and ecological questions including the effect of structurally different sequestered defenses on predator-prey interactions and physiological responses to non-natural toxins.
Concentrations of toxins in the pills can be adjusted relatively precisely and toxins are evenly distributed within pills. A possible explanation for the less precise concentration and toxin distribution observed in colchicine pills might be an extraction bias due to the physicochemical properties of colchicine which might be different from cardenolides. Apart from adequate concentration control, amounts can greatly deviate from natural concentrations, allowing for the study of dose-dependent effects.
Under relatively warm conditions (27 °C and 60% r.h.), the shelf life of our artificial diet varied remarkably between two setups, which could be due to the involvement of different brands of sunflower meal or different insect species. Pills offered to S. saxatilis had a better shelf life than the pills offered to O. fasciatus, both in terms of microbial growth, disintegration, and desiccation. If these differences were not due to the use of different sunflower meals or different climate chambers, they could be mediated by differences between the two insect species such as the bigger size and higher metabolic activity of O. fasciatus compared to S. saxatilis. Nevertheless, our new diet is suitable for short-term (10 days) toxin feeding assays involving adult insects without replacing the diet. Shelf life of artificial diets normally ranges from 2-3 days (Mendoza et al., 2016;Pourkhatoon et al., 2016;Hayashida et al., 2018) to 5-7 days (Wang et al., 2013;Hervet et al., 2016).
In conclusion, we developed a simple artificial diet for toxin feeding assays for the milkweed bugs O. fasciatus and S. saxatilis that has the potential to be applied to other species of sequestering milkweed bugs with similar nutritional needs and feeding habits. The benefits of our diet include its high quality as a dietary resource, its suitability for experimental purposes, its acceptable concentration accuracy, and the homogenous distribution of toxins. Furthermore, the diet is sufficiently stable to carry out short-term feeding assays for up to 10 days with adult insect specimens without exchanging the diet. Given the wide variety of different meals on the market, we suggest that additional artificial diets can be developed for other insects that feed on the fresh products from which meals are produced. Studies of plant defense, sequestration, predator-prey interaction, and insecticide assays are potential fields where our diet model could be applied.

AC K N O W L E D G M E N T S
We would like to thank Madlen Prang for her assistance with R analysis, Prayan Pokharel for valuable scientific exchange and support, Prof. Dr. Hans-Peter Piepho for his guidance and advice on statistical analyses, Sabrina Stiehler for the time and effort invested in showing LE bug rearing and maintenance and her help with HPLC techniques and analysis, and Sarah Rißmann for technical support. This work was supported by DFG grant PE 2059/3-1 to GP and the LOEWE (Landes Offensive zur Entwicklung Wissenschaftlich-ökonomischer Exzellenz) program of the State of Hesse via funding the LOEWE Center for Insect Biotechnology and Bioresources. Open Access funding enabled and organized by Projekt DEAL.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in Dryad at https://doi.org/10.5061/dryad.cc2fq z68c.

R E F E R E N C E S S U P P O R T I N G I N FO R M AT I O N
Additional supporting information can be found online in the Supporting Information section at the end of this article. Table S1. Detailed toxin quantification of sunflower pills. Each pill was divided in three 100-mg chunks, used to obtain the mean toxin concentration. The difference between each chunk and its corresponding mean and the mean difference was calculated per each pill type. Table S2. Concentration accuracy (mg toxin/g dry weight) of sunflower pills. The mean toxin concentrations per pill from Table S1 were used to obtain the final mean (± SE) concentration per pill type. The difference between the final concentration and the expected concentration was calculated and transformed into a percentage per pill type and per toxin. Table S3. Toxin distribution (μg toxin/mg dry weight) within sunflower pills. The mean difference per pill type from Table S1 was used to calculate the percentage difference from the final concentration per pill type and per toxin. Table S4. Detailed shelf life of sunflower pills per insect species. Desiccation, microbial growth, and disintegration were assessed in sunflower pills accessed by Oncopeltus fasciatus and Spilotethus saxatilis. Figure S1. Concentration accuracy and toxin distribution within sunflower pills. The dots indicate jittered raw data, each column represents three 100 mg chunks of one individual pill. Ouabain pills are represented in purple, digitoxin pills in orange and colchicine pills in pink. Black horizontal lines stand for the mean derived from the three 100 mg chunks. In red is shown the average (± SE) toxin concentration of each pill type (final concentration). Grey horizontal dashed lines represent the expected concentration.