Chemical defense and tonic immobility in early life stages of the Harlequin cabbage bug, Murgantia histrionica (Heteroptera: Pentatomidae)

Antipredation strategies contribute to the lifetime reproductive success of organisms, particularly in more vulnerable life stages that look to survive until reproduction. In insects, eggs and larval stages are often immobile or unable to rapidly flee and hide from predators. Understanding what alternative antipredation strategies they use, but also how these change over development time, is required to fully appreciate how species adapt to biotic threats. Murgantia histrionica is a stink bug, conspicuously colored from egg to adult, known to sequester defensive glucosinolates from its cruciferous hosts as adults. We sought to assess whether this chemical defense is also present in its eggs and early nymphal instars and quantified how it fluctuates among life stages. In parallel, we looked at an alternative antipredation strategy, described for the first time in this species: tonic immobility (i.e., death feigning). We also qualitatively investigated ultraviolet reflectance in eggs and adults as a proxy of conspicuousness against UV-absorbing leaves. Our results show that the eggs are significantly more chemically defended than the first two but not third mobile life stages, yet compound concentrations do not statistically differ across nymphal instars. Tonic immobility is favored by hatchlings, but less so by subsequent instars. Eggs also had obvious ultraviolet reflectance, suggesting that they would contrast against a leaf substrate and, considering their chemical load, that they may be aposematic. We argue that there are two possible interpretations of our results. One is that, throughout ontogeny, tonic immobility is a useful defensive strategy until adequate chemical protection is achieved over an extended feeding period. The other is that both aposematism and tonic immobility are used by this species, but variation in strategy use throughout ontogeny is decoupled.

Keywords: Antipredation strategies; Chemical defenses; Glucosinolates; Tonic immobility; Egg coloration; Pentatomidae; Murgantia histrionica

Animals are faced with constant threats, such as predation and parasitism, and must overcome them to survive and reproduce. Several antipredation strategies have evolved in numerous taxa to minimize predation pressures. One of them is the use of chemical compounds to deter predators through their unpalatability and/or toxicity. Species using such chemical defenses frequently advertise their unprofitability through warning signals (e.g. conspicuous body coloration), allowing predators to learn to avoid defended prey when exposed to the signals through associative learning: a phenomenon called "aposematism" (Poulton [51]; Skelhorn et al. [58]). Although usually studied in active life stages, chemical defenses in insects are also known to occur in eggs (Guerra-Grenier [27]). In their thorough book chapter on the subject, Blum and Hilker ([5]) distinguish between two types of chemical protection in insect eggs: the defensive compounds can either be produced autogenously by the parents (intrinsic origin) or can be sequestered from the parents' diet (extrinsic origin), then transferred to the eggs. These transfer mechanisms are not mutually exclusive—some species are known to use both, such as Photuris fireflies (González et al. [25]). An example of eggs protected by de novo chemicals is found in the Australian chrysomelid Paropsis atomaria Olivier 1807 (Nahrstedt and Davis [47]; Blum and Hilker [5]). All life stages of these beetles possess cyanogenic glygosides which, when under attack, can be modified to release hydrogen cyanide (HCN), a known respiratory inhibitor. As their Eucalyptus host plants are free of cyanogenic glycosides, it is believed that the chemical defense is produced by the insects themselves and then transferred into the eggs. Eggs protected through sequestration can be found for example in two orthopteran species of the genus Poecilocerus feeding on milkweed plants: P. bufonius (Klug 1832) and P. pictus (F. 1775) (Fishelson et al. [22]; Pugalenthi and Livingstone [53]; Blum and Hilker [5]). Milkweeds are well-known for their poisonous cardenolides, which are sequestered by the aposematic Poecilocerus and later incorporated in their eggs. Furthermore, whether they are of intrinsic or extrinsic origin, chemical defenses can sometimes be provided to the eggs by the fathers (Eisner et al. [18]). This is potentially advantageous to both parents since females must only pay a fraction of the metabolic costs for synthesis and/or sequestration while males with high concentrations of toxins may be favored during mate choice.

Aside from chemical protection, tonic immobility (TI) is another common antipredation strategy. Individuals engaging in tonic immobility enter a completely motionless state after physical contact with potential predators. TI is also referred to "thanatosis" or "death feigning" in the literature, because immobile prey are reminiscent of deceased individuals (Ruxton [55]). However, as pointed out by Humphreys and Ruxton ([34], and references therein) in their recent review (and noted by Darwin ([14])), death feigning individuals often adopt postures that are dissimilar to those of dead individuals of the same species. Henceforth, we will use the term tonic immobility. TI has been hypothesized to be an efficient anti-predation strategy through various mechanisms (not necessarily mutually exclusive) including mimicking death, the signaling of chemical defenses (Miyatake et al. [45], [46]; Ruxton [55]) and the reduction of attack rates by motion-oriented predators (Edmunds [17]; Prohammer and Wade [52]; Miyatake [44]). Although it is well known, tonic immobility is still relatively understudied and its use is most likely under-reported (Humphreys and Ruxton [34]).

A good candidate for the study of antipredation strategies is the Harlequin cabbage bug, Murgantia histrionica (Hahn 1834) (Heteroptera: Pentatomidae). This species of stink bug feeds mainly on cruciferous plants (Brassicaceae), but is known to have over 50 possible host plants (McPherson [43]). Native to Mexico, it is established in several U.S. states and is occasionally observed in Canada (McPherson [43]; Paiero et al. [49]). It is a recognized pest of economically important crops such as cabbage, collard, kale and broccoli (Ludwig and Kok [41]; Wallingford et al. [63]) and is thus studied for its management, more recently in a semiochemical context (Conti et al. [12]; Peri et al. [50]). Part of what explains the success of these bugs is their low number of natural enemies; they have a few parasitoids and virtually no predators or competitors (McPherson [43]; Amarasekare [3]), suggesting efficient anti-predation strategies. Contrary to most North American pentatomids, which are cryptic brown or green, Murgantia histrionica is easily recognizable through its orange, white and black markings (Fig. 1a). Aliabadi et al. ([2]) suggested that this conspicuous color pattern acts as a warning signal to reduce predation by birds, their most likely predators across both their native and invasive ranges. Indeed, they showed that, much like some other species feeding on cruciferous plants such as many aphids and sawflies (Opitz and Müller [48]), adult Harlequin bugs can sequester glucosinolates from their host plant into their body tissues. Glucosinolates (mustard oil glycosides) are secondary metabolites produced by cruciferous plants and their hydrolysis products (isothiocyanates, nitriles, etc.), which are toxic and used as antiherbivory defenses (Louda and Mole [40]; Opitz and Müller [48]). Although only described in Murgantia histrionica, this chemical defense could potentially be present in other members of the Strachini tribe (i.e. other Murgantia spp., Eurydema spp., Stenozygum spp., etc.) given their shared body coloration (orange, white and black markings) and use of glucosinolate-producing host plants (i.e. Brassicaceae, Capparaceae, etc.) (Exnerová et al. [20]; Samra et al. [56]). While it is still unknown whether the stink bugs sequester, along with glucosinolates, the enzyme myrosinase responsible for the hydrolysis of the compounds, or if they produce an enzyme de novo, a chemical reaction during the act of predation is necessary for mustard oils to be effective against predators (Aliabadi et al. [2]).

Graph: Fig. 1 a Adults (photo: Judy Gallagher) and b eggs (photo: USDA) of Murgantia histrionica

Like the adults of the species, M. histrionica eggs and nymphs are also quite distinctive. The eggs (Fig. 1b), usually laid in clutches of 12 (two rows of six eggs), are white with two black bands and a black spot on the external side. The operculum is also black and white, and the black portion can either form a ring or be shaped as to resemble the Yin-Yang symbol. Like in the adults, the egg color pattern is generally conserved throughout the Strachini tribe (Kalender et al. [38]; Samra et al. [56]). Nymphal coloration is somewhat similar to that of the adults in terms of hues and pattern complexity, but white markings are also found on their dorsal side. Although sequestration of glucosinolates presumably starts during immature stages, as a result of feeding on cruciferous plants, it is currently unknown whether nymphs and eggs are chemically defended against vertebrate and invertebrate predators. Furthermore, tonic immobility has been observed in hatchlings that were physically disturbed (Guerra-Grenier, personal observations). This is interesting because, like in other stink bug species, this life stage does not feed (Canerday [8]; Zahn et al. [65]), meaning that if glucosinolates are not vertically obtained through one or both parents, chemical defense cannot be achieved until later on in their development. TI could thus be an alternative line of defense, possibly used to deter motion-oriented predators.

The aim of the present study was to assess the extent of antipredation defenses in eggs and early nymphal instars of the Harlequin cabbage bug. First, we wanted to assess whether eggs and nymphs have glucosinolates and, if so, whether their concentrations differ among life stages. We predicted that the concentration of glucosinolates would increase in nymphs over successive molts, reflecting an increased sequestration over feeding time. Second, we wanted to investigate the use of TI in nymphs by looking at its frequency and duration, predicting that it is used by the insects to compensate for lower chemical defenses and that TI use decreases as glucosinolate sequestration increases. Finally, we asked if the white portion of the egg color pattern, already thought to be conspicuous against a green crucifer leaf background, allows for an increased contrast by also reflecting wavelengths in the ultraviolet (UV) part of the electromagnetic spectrum. Indeed, most insect species can perceive UV light and incorporate ultraviolet coloration in their signals (Cronin and Bok [13]) and leaves are known to absorb ultraviolet radiation from the sun (Gutschick [29]). Increasing the visual contrast between the eggs and their substrate might generate a stronger warning signal, via increased saliency and/or recognizability, perhaps facilitating aversion learning following a taste of the chemically-defended prey.

M. histrionica individuals were collected in the Beltsville area (Maryland, USA) in 2016 and subsequently cultured at 28 ± 1 °C and a 16L: 8D light cycle in an incubator (VWR Chamber Diurna Growth 115V, Cornelius, USA) supplied with a jar of water for added moisture. Adults were reared continuously in 30 cm3 ventilated polyester cages (BugDorm, Taichung, Taiwan) while nymphal instars were kept in ventilated plastic containers (length: 22 cm; width: 15 cm; height: 5 cm; Rubbermaid, USA). All were fed with rinsed fresh store-bought broccoli (Brassica oleracea var. italica), replaced three times per week. Eggs were collected three times per week from the cage walls or from the broccoli stems, then placed into Petri dishes (diameter: 9 cm) lined with a filter paper until hatching, at which point they were transferred to the plastic containers.

A total of 301 individual Harlequin cabbage bugs, of a range of developmental stages, were sampled from the lab colony for glucosinolate extractions, following an adapted version of the protocol established by Doheny-Adams et al. ([16]). Since a simple qualitative (presence/absence) detection of the compound would likely be misleading for second and third instars, as its presence in the gut from previous feeding events makes it impossible to know if glucoraphanin is also present in body tissues, we opted for a quantitative analysis. Individuals were first separated into life stages (whole eggs, first instars, second instars and third instars), subsequently split into triplicates of equivalent mass (3× ~ 18.00 mg for eggs; 3× ~ 13.00 mg for first instars; 3× ~ 40.00 mg for second and third instars) and macerated in a solution of 70% methanol at a dilution ratio of 1 ml of solution per 4 mg of insect mass. The shells left behind by the first instars, which were collected before they started feeding, were also pooled (~ 3.00 mg per replicate) and treated the same way. Flowerets of broccoli (440.45 mg) from the same batch the insects fed on were ground with liquid nitrogen and then transferred into 70% methanol at the same dilution ratio mentioned above. Once prepared, samples were put into a water bath at 70 °C for 30 min for extraction.

The resulting extracts were filtered through a 0.22 mm PTFE filter and analyzed on a Shimadzu UPLC-MS system (Mandel scientific company Inc, Guelph, Canada) which contains LC30AD pumps, a CTO20A column oven, a SIL-30AC autosampler and a LCMS-2020 mass spectrometer. Briefly, 1 μl of each fraction was injected through the autosampler to a Luna omega polar C18 column (100 × 2.1 mm, 1.6 μm particle size, Phenomenex, Torrance, USA). Mobile phases were H2O and acetonitrile, with 0.1% formic acid in both. The isocratic elution method was initialed with 2% acetonitrile for 5 min. The column was then washed with 100% acetonitrile for 3 min and re-equilibrated for 5 min before the next injection. The flow rate was set at 0.5 ml/min with the column tempered at 50 °C.

The mass spectrometer with electrospray ionization (ESI) interface operated in negative selective ion monitoring (SIM) mode, the nebulizing gas flow was set at 1.5 L/min and drying gas flow was at 10 L/min. The desolvation line temperature and heat block temperature were set at 250 °C and 400 °C respectively. The detector was monitoring m/z at 436 [M–H]− with 938 µ/s scan speed. Linear calibration curves were built by injecting dilutions of glucoraphanin (Cayman chemicals, Ann Arbor, USA) which bracketed the compound concentration in the samples. Calibration curves were prepared at five concentration levels (0.2–10 ng on column) and R2 values obtained.

Eggs were sampled from the lab colony and put in clutch-specific Petri dishes (diameter: 9 cm). Individuals that hatched from those eggs were reared in the same environmental condition as the general rearing until they reached the first (86 from 8 clutches), second (81 from 10 clutches) and third instars (97 from 14 clutches). The number of clutches used increased with increasing instar as to obtain similar sample sizes while compensating for the baseline mortality rate. One at a time, every individual went through the following protocol. Bugs were first placed on a filter paper (diameter: 12.5 cm). They were then prodded by one of us (EG-G) using a paint brush. Each trial consisted of up to five sessions of five prods (25 in total), with each session separated by one minute. Trials ended when individuals either displayed TI postures (Fig. 2) for the first time during 5 s or more, or if all 25 prods unsuccessfully elicited the behavior. TI postures kept for less than 5 s were not considered given that such a short duration could be due to sensory overload or to a period of physiological recovery from impact. The presence of TI was measured as a binary variable (present or absent) and its duration (if present) was recorded. The clutch of origin of every individual was recorded to measure variation in behavior across clutches. The order at which individuals were taken from their clutch-specific container and prodded was also recorded, to look for any effect sampling might have on leftover individuals. A Petri dish lid of the same size as the filter paper was placed onto the arena between prodding sessions to stop the bugs from escaping. The soft hair of the brush was frequently moistened, twisted together and then dried by contact with another filter paper to reduce the risk of trapping the bugs among the brush hair. All trials were carried out by the same observer.

Graph: Fig. 2 Postures adopted by M. histrionica nymphs during behavioral assays. Nymphs could be active (a) or display tonic immobility when on their ventral (b) or dorsal (c) side. The pictures shown here are of an individual in its first instar. Photos: Eric Guerra-Grenier

One adult male, one adult female and 10 clutches (totaling 106 eggs) of the Harlequin cabbage bug were photographed under both ultraviolet and visible light using a Nikon D70 camera with an El-Nikkor 80 mm lens and a Baader U UV filter (Baader Planetarium, Germany: 310-390 nm UV transmission). The El-Nikkor lens is sensitive to wavelengths 320 nm and higher (Verhoeven and Schmitt [62]). A MTD70 EYE color arc bulb (70 W, 1.0A power source, www.eyelighting.co.uk) was used as the only light source and was chosen for its D65 spectrum (the same spectrum as sunlight). Eggs were glued on double sided tape on a Petri dish (diameter: 9 cm) and pictures were taken from above and from the external (banded) side. Adults were photographed on their ventral and dorsal sides.

All analyses were conducted with R version 3.3.3 (R Core Team [54]). A Kruskal–Wallis rank sum test allowed for the comparisons of glucoraphanin concentrations among eggs and the three nymphal instars. Dunn's tests were subsequently used for post-hoc pairwise comparisons between life stages using the "dunn.test" package in R (Dinno [15]). To compensate for our small sample size and lack of statistical power, we also performed a Spearman rank correlation test ("coin" package in R, Hothorn et al. [32]) comparing nymphal concentrations. Differences in population mean concentrations between hatchlings and their eggshells were analyzed by fitting a linear model following a square root transformation of the concentration data to successfully ensure homogeneity and normality (Fig. 3).

Graph: Fig. 3 Pictures of eggs and adults taken in the visual (left panels) and ultraviolet (right panels) spectra. Eggs are seen from their top (a, b) and lateral sides (c, d). The female is seen from her ventral (e, f) and dorsal (g, h) sides. The male is seen from his ventral (i, j) and dorsal (k, l) sides. UV pictures were converted in black and white images so that UV light could be perceived by human eyes on a luminance scale (the brighter the markings, the more UV-reflecting they are). Photos: Eric Guerra-Grenier

A series of generalized linear mixed models (GLMMs) and generalized linear models (GLMs) with binomial error distributions were used to regress the probability of tonic immobility as a function of all or a subset of the following factors: instar (first, second and third; fixed factor), prodding order (covariate) and clutch of origin (random factor). These models were fitted using the "lme4" package in R (Bates et al. [4]). A series of linear mixed models (LMERs) and linear models (LMs) were fitted to test for the effect of the same factors on the duration of tonic immobility, assuming normally distributed error. For both probability and duration of TI, Akaike information criteria (AIC) values finally determined which models out of all the possible factor combinations had the lowest out of sample deviance. For all statistical models that we fitted to the TI data, only one data point was taken from each individual to avoid pseudoreplication, and the significance of each of the fixed factors was determined with log likelihood ratio tests using the "car" package in R (Fox and Weisberg [24]). Simultaneous tests for general linear hypotheses with Tukey contrasts were subsequently used for post-hoc pairwise comparisons between nymphal instars using the "multcomp" package in R (Hothorn et al. [33]).

Glucoraphanin was detected in broccoli as expected, but also in the eggs and hatchlings (first instars) (Fig. 4). Concentrations of glucoraphanin were subsequently evaluated in eggs and all three instars (Fig. 5a) and were shown to differ among life stages (H = 8.7436, df = 3, p = 0.0329). Post-hoc comparisons revealed that this trend was mainly driven by the eggs having significantly higher concentrations than first ( $\overline{\text{X}}$ ̄eggs = 0.5784, seggs = 0.0833, $\overline{\text{X}}$ first = 0.1038, sfirst = 0.1011, Z = 2.6042, p = 0.0046) and second instars ( $\overline{\text{X}}$ second = 0.1167, ssecond = 0.0073, Z = 2.3778, p = 0.0087), although they did not differ from third instars ( $\overline{\text{X}}$ third = 0.2481, sthird = 0.5294, Z = 1.323, p = 0.1288). Concentrations in the first two nymphal instars did not differ significantly (Z = − 0.2265, p = 0.4104), and despite the lack of statistical difference between concentrations of eggs and third instars, we could not reject the null hypothesis that third instars had the same levels of glucoraphanin as the first (Z = − 1.4720, p = 0.0755) and second instars (Z = − 1.2455, p = 0.1065). Spearman ranks fitted only on instar data corroborated these results, indicating a positive but not technically significant correlation between life stages and glucoraphanin concentrations (rs = 1.9379, p = 0.05263), indicating that concentrations in first, second and third instars may biologically be equally low.

Graph: Fig. 4 UPLC-ESI/MS negative extracted ion chromatogram of glucoraphanin (m/z = 436 (M–H)−). a Glucoraphanin commercial standard; b broccoli flowerets; c whole eggs; d first instars (hatchlings)

Graph: Fig. 5 Boxplot representations of glucoraphanin concentrations in various life stages. Lower-case letters indicate statistically significant differences. a The concentrations of glucoraphanin (μg/mg) in whole eggs, first, second and third nymphal instars chemical defenses. The dotted lines represent concentrations of glucoraphanin in broccoli [reported by Florkiewicz et al. ([23]) and Song and Thornalley ([59])] as a means of comparison. b The concentrations of glucoraphanin (μg/mg) in hatchlings (first instars) and their shells, square-root transformed to ensure data homogeneity and normality

Since hatchlings had significantly less compound than eggs, their concentrations were compared to those of the shells they left behind when hatching (Fig. 5b). The analysis revealed that, of all the compound present in the eggs, most of it is laced within the shells (LM: F1,4 = 88.002, p < 0.001), leaving only a small portion of glucoraphanin available for integration by the developing insects.

The probability of displaying tonic immobility of a given nymph was strongly influenced by which instar it was in (LRT: χ2 = 13.0001, df = 2, p = 0.0015): hatchlings displayed the behavior significantly more frequently than second (p = 0.0304) and third instars (p = 0.0013), whereas the later two instars had similar display frequencies (p = 0.5120) (Fig. 6a). The probability of displaying TI also reduced the later nymphs were sampled from their clutch of origin (LRT: χ2 = 3.8653, df = 1, p = 0.0493) (Fig. 6b). The duration of the behavior was shorter in individuals tested later within a clutch compared to those tested earlier (LRT: χ2 = 4.2635, df = 1, p = 0.0390), but was not affected by nymphal instar. The clutch from which nymphs originated was also a key factor predicting the probability of displaying TI (LRT: χ2 = 10.051, df = 1, p = 0.001523) but not the duration of the behavior (LRT: χ2 = 1.936, df = 1, p = 0.1641).

Graph: Fig. 6 a The average probability of entering tonic immobility across instars. Error bars represent 95% binomial confidence intervals while letters represent differences in statistical significance. b The effect of prodding order within a clutch on an individual's probability of entering tonic immobility (left axis) and the frequency of TI in the population (right axis), both across all instars. The red trendline shows variation in the probability of entering TI as a function of prodding order. Histograms were built from TI presence/absence data and represent the behavior's frequency as a function of prodding order, with bins indicating the number of individuals that entered tonic immobility (upper histogram) or not (lower histogram)

Eggs exposed to ultraviolet light (Fig. 3b, d) showed high reflectivity in the pale, but not dark markings of the color pattern in all ten clutches photographed. The same level of contrast was present under visible light only (Fig. 3a, c). This implies that the pale markings of the eggs are perceived as true white to UV-sensitive animals as well as to those who are not (e.g., humans). Unlike the eggs, both the male and the female adults showed almost no reflection of wavelengths between 320 and 390 nm (Fig. 3e–l).

The aim of this study was to investigate two potential antipredation strategies in eggs and juvenile Murgantia histrionica: aposematism (a combination a chemical defense and a conspicuous warning signal) and tonic immobility. Our results suggest that both strategies are used by the bugs, but also that their use varies across life stages. Given that eggs are completely immobile and are "sitting ducks" to would-be predators, behavioral strategies are unavailable, leaving aposematism as a viable defense. This could explain why concentrations in the eggs are much higher than in early nymphal instars, in which glucoraphanin levels do not vary significantly. However, levels of compound in the eggs are like those of third instars. Given that first and second instars have a significantly lower concentration than eggs, one possibility is that third instars are at an intermediate level, indicating that sequestration from feeding probably starts at this stage but has not yet reached the degree seen in adults. Indeed, according to Aliabadi et al. ([2]), adults sequester glucosinolates in their tissues at concentrations 20–30 times higher than those found in the gut (i.e. about the same concentration as in cruciferous plants). However, our results also indicate that glucoraphanin concentrations did not significantly differ across all three nymphal instars, and were in fact similar to concentrations in broccoli (Song and Thornalley [59]; Florkiewicz et al. [23], although variation occurs among cultivars as shown by Brown et al. [6]). This may be because it takes time to build up the concentration in body tissues. It may also be because small nymphs of true bugs, including the Harlequin cabbage bug, are exposed to different predators than adults and bigger nymphs (mostly invertebrate vs. mostly vertebrate respectively) (Exnerová et al. [19]), and these predators likely have different susceptibilities to mustard oils. Although both vertebrates and invertebrates are usually intolerant to glucosinolates (Halkier and Gershenzon [30]), such compounds are used by plants to limit insect feeding (Louda and Mole [40]) while they can potentially be beneficial to vertebrates at low enough concentrations (Shapiro et al. [57]).

Interestingly, a large proportion of the glucosinolates in the eggs is laced within the shells and is thus not accessible to the bugs for integration into their tissues before hatching. This could explain why, even though egg predators have seldom been reported, eggs are often attacked by parasitoid wasps of the genera Trissolcus (Platigastridae) and Ooencyrtus (Encyrtidae) (McPherson [43]; Amarasekare [3]; Conti et al. [12]; Peri et al. [50]). These endoparasitoids drill through the eggshell with their ovipositor and lay their eggs directly within the host eggs, meaning that they bypass most of the chemical defense. An alternative hypothesis to explain the higher concentration of compound in the shells could be that mothers use them as toxin sinks. By doing so, females would reduce their metabolic investment in detoxification while still avoiding compromising the development of their offspring. Eggshell sequestration as a detoxification mechanism has previously been suggested in bird eggs (Kitowski et al. [39]).

In contrast to chemical protection, which did not vary markedly across nymphal stages, Harlequin bug nymphs showed life stage-specific variation in their probability, but not duration, of entering tonic immobility. Indeed, first instars displayed the behavior significantly more frequently than second and third instars. As we cannot reject the null hypothesis that glucoraphanin content was invariant across nymphal stages, given the low power of our statistical analysis, we have no evidence to support our hypothesis that TI is present in younger life stages to compensate for the lack of chemical defense. A higher probability of TI in hatchlings could however be explained by variation in life history traits. M. histrionica is not really active until its second instar. Zahn et al. ([65]) reported that neonate bugs do not feed and remain aggregated with their siblings to acquire endosymbionts from the hatched clutch deposited there by the mother; only after the first molt did they start to feed and disperse. In other words, if attacked, individuals in their first instars are too small and probably not active enough to successfully flee. TI thus becomes the optimal behavioral strategy, in addition to chemical protection. It is also conserved at a lower degree in second and third instars. Yet again, variation in life history traits could explain the lower frequency in older nymphs: fleeing becomes a safer option as size and speed increase, leading up to an almost zero probability of entering TI in the final nymphal instars and adults (Guerra-Grenier, personal observations).

Regardless of instar, TI use was also influenced by the order in which we prodded individuals of a given clutch. Those tested first had a higher probability of entering tonic immobility than those tested last and kept the posture for a longer period. This may be because of an alarm cue perceived by the nymphs. Such a cue could be the visual detection of conspecifics being "attacked" but is most likely olfactory in nature. There is a reason why stink bugs are named this way: they release pungent volatile chemicals when disturbed, and these chemicals are known to elicit dispersal in neighboring conspecifics (Ishiwatari [36]). In Murgantia histrionica specifically, some of these compounds are thiocyanides derived from the glucosinolates they absorb from their host plants (Aldrich et al. [1]; Aliabadi et al. [2]). It would make sense for individuals reacting to these chemical cues to engage less frequently and in shorter TI periods if they are stimulated to flee. It should however be noted that ecological factors cannot fully explain the use of TI. Inter-clutch variation in the probability of displaying TI suggests that there are other factors affecting tonic immobility in M. histrionica. One possibility is that the propensity to engage in TI is partly hereditary, which would not be surprising considering that genetic variation in TI is known to occur in other taxa (Miyatake et al. [45]).

As mentioned in the introduction, TI is sometimes positively correlated with chemical defenses and has been hypothesized to act as a warning display (Miyatake et al. [45], [46]; Ruxton [55]). Such a correlation is not apparent in Murgantia histrionica based on our results: the frequency of TI diminishes from first to second and third instar, but we cannot reject the null hypothesis that glucosinolate levels remain the same across these three life stages. With that said, given that adults have much higher concentrations compared to what we observed in early nymphs (Aliabadi et al. [2]) and that they do not seem to use TI as an antipredation strategy (Guerra-Grenier, personal observation), there may be a negative correlation between chemical defense and tonic immobility across all life stages, from hatchlings to adults. If that were true, then TI would not serve as a warning signal but rather simply as an alternative defense strategy, potentially deterring motion-oriented predators. Further investigations of the relationship between chemical defense and tonic immobility are warranted to thoroughly test this hypothesis. More data on TI and glucosinolate content in fourth and fifth instar as well as in adults should also be collected to expand the test of such a relationship across all life stages of this stink bug.

In addition to the description of their chemical defense, we have demonstrated that M. histrionica eggs are most likely highly conspicuous to small, visually-oriented predators, whether they are UV-sensitive or not. Indeed, the white markings of the eggs reflect ultraviolet light, meaning that they appear white (i.e., they reflect all the wavelengths that a given visual system is sensitive to) to any potential predator, regardless of whether they are sensitive to wavelengths between 300 and 400 nm. While our qualitative dataset cannot help distinguish between a peak in UV reflectance and a simply broader reflectance pattern across the UV–VIS spectrum, such a color pattern would likely contrasts strongly against a leaf background, which usually absorbs UV-light (Gutschick [29]). Even if eggs were to be laid under a leaf, thus limiting the amount of UV light that might make it to the clutch (e.g., if the wind blows the leaf underside towards the sky, or if UV light bounces off the ground), the overall chromatic and achromatic contrast between egg parts and between eggs and their substrate should still be biologically significant without the ultraviolet component. By combining the visual contrast between eggs and oviposition sites with the presence of a chemical load, we argue that M. histrionica eggs are most likely aposematic, in that they are both conspicuous to would-be predators and unprofitable; as such, the black and UV-white pattern probably acts as a warning signal. Interestingly, white colors displayed by adults were not UV-white, suggesting that the features (i.e., pigments and/or refracting nanostructures) responsible for the white coloration perceived by humans in adults and eggs are different to some extent. Future studies should aim to collect spectral data with a UV–VIS spectrophotometer to properly quantify the reflectance patterns of both eggs and adults.

Further experiments measuring prey (eggs, but also nymphs) acceptance by ecologically relevant predators over repeated exposures are necessary to confirm that predators learn to avoid consumption of the bugs based on recognition of their color pattern (although an innate aversion to egg and nymph stripes is also possible (Halpin et al. [31])). Such tests should also compare predation levels by chewing vs. piercing-sucking predators to test for the relative importance of the shell (which composes most of the chemical defense) in the learning process. Confirmation of efficacy is also required for the TI strategy: although the posture adopted by Harlequin bug nymphs when disturbed fits with the defining characteristics of TI (Humphreys and Ruxton [34]), it is still unknown whether it reduces predation in this species. An alternative explanation for the behavior could be that it would allow the bugs to fall off their host plant when threatened in nature, a defensive strategy described in other insect taxa (Gross [26]; Chaboo [11]; Humphreys and Ruxton [35]). While we did occasionally observe younger nymphs fall from their host plant during colony maintenance, we did not test that hypothesis specifically. Further testing of TI in M. histrionica thus requires using bugs placed on their host plants to confirm whether immobilized individuals fall off the plant or remain close to predators on the leaves.

Complex color patterns often play a role in multiple defensive and/or communication strategies (Marshall [42]; Tullberg et al. [61]; Caro et al. [9], [10]). Specifically for striped black and white patterns, known functions vary from crypsis (e.g. disruptive camouflage, countershading, etc.) to warning signals and can interfere with a predator's ability to attack a prey through a dazzling effect (Stevens et al. [60]; Izzo et al. [37]; Feltwell [21]). Aside from aposematism, the complex M. histrionica egg color pattern may thus have alternative adaptive functions, and future studies should look into those in this species as well as in taxa with similar egg coloration such as Eurydema spp. and Stenozygum spp. (plesiomorphy: Kalender et al. [38]; Samra et al. [56]), but also in Piezodorus spp. (convergent evolution: Bundy and McPherson [7]). For example, the stripes and dot may convey an intraspecific signal influencing the ovipositional behaviors of conspecific females, provide disruptive camouflage through distance-dependent crypsis, protect against harmful solar radiation and/or reduce desiccation (Guerra-Grenier [27]). Additionally, because eggs vary in their proportion of black, pigment variation may help with heat absorption in colder periods. Such a thermoregulatory function has actually been shown in the mobile life stages of the species, where late nymphal instars exposed to colder temperatures will metamorphose into adults with a higher degree of melanisation than those exposed to warmer temperatures (White and Olson [64]).

We would like to thank Paul K. Abram, Sophie Potter, Changku Kang, Felipe Dargent, Casey Peet-Paré, Tammy Duong, Yolanda Yip, Ian Dewan, Lauren Efford, Karl Loeffler-Henry, Gustavo L. Rezende, Greg Bulté, Naomi Cappuccino, Mark Forbes, and Andrew Simons for helpful discussions and/or technical assistance. We also thank Donald C. Weber for sending us eggs of Murgantia histrionica to build our lab colony, as well as two anonymous reviewers for suggestions that improved this manuscript. A previous version of this paper has previously been available on BioRxiv (https://doi.org/10.1101/2021.01.29.428818).

Conceptualization, EG-G, RL, JTA, TNS; Methodology, EG-G, RL, JTA, TNS; Investigation, EG-G, RL; Formal Analysis, EG-G, RL, TNS; Writing—Original Draft, EG-G, RL; Writing—Review and Editing, EG-G, RL, JTA, TNS; Supervision, JTA, TNS; Funding Acquisition, JTA, TNS.

This research was supported by a FRQNT postgraduate scholarship to EG-G, NSERC grants to RL, JTA and an NSERC Discovery grant to TNS.

Data is available on Dryad (Guerra-Grenier et al. [28], https://doi.org/10.5061/dryad.tmpg4f4zg).

The authors declare no conflict of interest.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.