Plant toxicity shapes the dietary choices of herbivores. Especially when herbivores sequester plant toxins, they may experience a tradeoff between gaining protection from natural enemies and avoiding toxicity. The availability of toxins for sequestration may additionally trade off with the nutritional quality of a potential food source for sequestering herbivores. We hypothesized that diet mixing might allow a sequestering herbivore to balance nutrition and defense (via sequestration of plant toxins). Accordingly, here we address diet mixing and sequestration of large milkweed bugs (Oncopeltus fasciatus) when they have differential access to toxins (cardenolides) in their diet. In the absence of toxins from a preferred food (milkweed seeds), large milkweed bugs fed on nutritionally adequate non-toxic seeds, but supplemented their diet by feeding on nutritionally poor, but cardenolide-rich milkweed leaf and stem tissues. This dietary shift corresponded to reduced insect growth but facilitated sequestration of defensive toxins. Plant production of cardenolides was also substantially induced by bug feeding on leaf and stem tissues, perhaps benefitting this cardenolide-resistant herbivore. Thus, sequestration appears to drive diet mixing in this toxic plant generalist, even at the cost of feeding on nutritionally poor plant tissue.

Performance and sequestration of O. fasciatus on non-preferred food sources

We examined performance and sequestration of seed bugs on the tropical milkweed (Asclepias curassavica), an important host plant in the southern end of the bug’s range. All bugs used in this experiment were from a colony (30 adult bugs) of wild-collected O. fasciatus from Tompkins County, NY, USA. The colony had been reared for several generations on common milkweed (Asclepias syriaca) seeds. First instar bugs (from a pool of clutches laid by several females) were divided into five experimental treatment groups within twenty-four hours after hatching: bugs with access to milkweed seeds (nutritionally adequate, toxic), bugs with access to milkweed seeds and milkweed plant tissue (nutritionally adequate, toxic + nutritionally inadequate, toxic), bugs with access to sunflower seeds (nutritionally adequate, nontoxic), bugs with access to sunflower seeds and milkweed plant tissue (nutritionally adequate, nontoxic + nutritionally inadequate, toxic), and bugs with access to milkweed plant tissue with no seeds (nutritionally inadequate, toxic). These treatments were chosen to test all factorial combinations of nutritionally adequate toxic and nontoxic foods (both seed types), and nutritionally inadequate toxic foods (plant tissue) (Fig. 1). Plants were reared following the method outlined in Appendix I.

We conducted ten replicates of each treatment, and each replicate received five bugs. All trials were conducted in 10 cm diameter pots filled with potting soil. After bugs were added, all pots were covered with semi-transparent mesh bags to prevent bugs from escaping. We also conducted a “no herbivory” treatment, which consisted of a milkweed plant that was bagged, but received no bugs or additional seeds (n=9; one plant was removed due to infestation with thrips). For seed treatments, 0.5 g of seed was presented on the soil surface and replenished with another 0.5 g of seeds once per week. This amount of seed was determined to be well in excess of the seed required for the growth and development of five O. fasciatus nymphs. Seeds that germinated over the course of the experiment were removed and replaced with fresh, ungerminated seeds. A. curassavica seeds were used for milkweed seed treatments and organic shelled sunflower seeds from a local supermarket were used as a cardenolide-free food source. All pots were watered approximately once every two days, regardless of the presence of a milkweed plant.

The experiment consisted of 5 blocks, each consisting of two replicates of each experimental treatment (10 pots per block). Bugs were allowed to feed and develop to adulthood. Bugs were collected from pots within 24 hours of reaching adulthood, starved alone in a petri dish for 24 hours (with access to water), frozen at -80°C, and freeze-dried. n= 31 bugs (15 males and 16 females) survived to adulthood on milkweed seeds alone, n = 18 bugs (7 males and 11 females) survived to adulthood on sunflower seeds alone, n = 35 bugs (19 males and 16 females) survived to adulthood that had access to both milkweed seeds and plants, and n = 36 bugs (23 males and 13 females) survived to adulthood that had access to sunflower seeds and milkweed plants. These samples were used in analyses of development time to adulthood and adult dry mass, while a smaller subset was used for chemical analyses. The sex of adult bugs was noted, as O. fasciatus bugs are known to be sexually dimorphic in terms of adult dry mass and cardenolide sequestration (with females growing to larger sizes and sequestering higher concentrations of cardenolides) (Isman, 1977).

For treatments in which bugs had access to a milkweed plant, all aboveground biomass was collected once all bugs had reached adulthood and the crown of the plant (the top six fully expanded leaves and all intervening stem) was freeze-dried for chemical analysis. The remaining aboveground biomass was dried for 3-5 days at 45 oC in a drying-oven. All bugs with access to milkweed plants but without seeds died before reaching adulthood. For plants with no adult bugs (i.e., the “no herbivory” treatment, which received no bugs and the treatment where bugs had access to milkweed plants only), collection of the plants of a given treatment was timed with the collection of the first and last plants in the same block on which all bugs reached adulthood.

Impacts of O. fasciatus feeding on milkweed performance and defensive chemistry

Due to low bug survivorship when feeding on milkweed plant tissue alone, a separate experiment was conducted to assess the impacts of bug feeding on milkweed growth and defensive chemistry. Eighteen milkweed plants were placed in bags at the 10-12 leaf stage as above and separated into three blocks of six plants each. Half the plants within each block received five first-instar milkweed bugs, and control plants received no bugs. One control and one herbivore-damaged plant was collected from each block at one, two, and three weeks after the start of treatments. As in the previous experiment, the crown of each plant was freeze-dried for chemical analysis, and the remaining aboveground biomass was dried in a drying oven. Height, aboveground dry mass, number of leaves damaged by bugs, and total number of leaves were recorded for all plants.

Analysis of plant and insect cardenolides

To analyze cardenolides, we used high performance liquid chromatography (HPLC) on freeze-dried leaves, adult bugs (n=10 per treatment, 5 male and 5 female; except for the treatment which only received milkweed seeds, for which 3 male and 3 female bugs were run), and seeds (n=5 sunflower and A. curassavica). Cardenolides were extracted from samples and analyzed following the method established by Züst et al. (2019). For seeds and leaves, 50 mg (+/- 1 mg) of powdered sample was used (except for two small leaf samples with 35 mg), while whole single adults were used for bug samples. To extract cardenolides, we ground each sample to a powder and added 1.6 mL of methanol, spiked with 20 μg of the cardenolide digitoxin (Sigma-Aldrich, St. Louis, MO), as an internal standard, along with approximately 30 FastPrep beads. A separate vial (containing only methanol and digitoxin) was run to generate a standard curve. Samples were agitated twice for 45s at 6.5 m/s (FastPrep-24 homogenizer), centrifugated at 12,000 rpm for 15 min, and solids were discarded. Bug and seed samples were defatted by adding 1.5 mL of hexanes and vortexing three times for 10s. The top (hexane) layer was removed and discarded. Supernatant was dried in a centrifugal concentrator at 35oC, resuspended in methanol (250 μL), and filtered with a 0.45 μm low-binding hydrophobic filter plate. Following the method established by Züst et al. (2019), 15 μL of each sample was injected into an Agilent 1100 series HPLC equipped with a diode-array detector and a Gemini-NX C18 reversed-phase column 3 μm, 150 mm × 4.6 mm (Phenomenex, Torrance, CA, USA). The injected sample was eluted at 0.7 mL/min across an acetonitrile-water gradient (0-2 min 16% acetonitrile, 2-25 min 16-70% acetonitrile, 25-30 min 70-95% acetonitrile, 30-35 min 95% acetonitrile, and a 10 min post-sample run with 16% acetonitrile).

The commercially-available cardenolide digitoxin, which is not found in milkweed, was used as an internal standard. HPLC chromatogram peaks with a single absorption maximum between 214 and 222 nm were considered to be cardenolides (Malcolm & Zalucki, 1996), and cardenolide concentrations were calculated from peak areas at 218 nm and were standardized by the concentration of the internal standard digitoxin in each sample and sample dry mass.

Statistical analyses

For each sample, the Shannon-Wiener diversity index (H’) was calculated by taking the natural logarithm of the proportion each cardenolide comprised of the total cardenolides in a sample and summing this across all cardenolides in the sample. Additionally, a polarity index was calculated by expressing each cardenolide as a proportion of the total cardenolides in a sample, weighting this by the cardenolide’s retention time, and summing this across all cardenolides in the sample (Jones et al., 2019). Higher polarity index corresponds to a more nonpolar assemblage of cardenolides. Statistics were performed in R version 4.1.2.  Total cardenolide concentration, polarity, diversity, and metrics of bug performance were compared among bugs using fixed effects ANOVAs (response variable ~ seed type*plant presence + bug sex). Bugs that only fed on sunflower seeds were excluded from polarity and diversity analyses, as these bugs contained only trace amounts of cardenolides. Plant growth and cardenolide diversity in the second experiment (examining plant induction over time) were also analyzed using ANOVAs (response variable ~ week*bug presence). Pairwise comparisons were performed using Student’s t-tests. The relationships between bug feeding and plant cardenolide concentration, polarity, and diversity were evaluated using Pearson correlations.

Raw data are attached as .csv files. Analyses were performed in R version 4.1.2.