Data for: Maternal provisioning of offspring with defence chemicals in a facultatively parthenogenetic stick insect

Parents can invest in offspring by transferring environmental factors, such as nutrients or diet-derived defence chemicals, into eggs or embryos. However, in systems where females can reproduce facultatively without a male (facultative parthenogenesis), it is not known how reproductive mode and maternal environment affect offspring provisioning. The facultatively parthenogenetic stick insect Megacrania batesii sprays a defensive fluid from paired prothoracic glands. Here, we report that some hatchlings of M. batesii can spray even prior to their first feeding, and provide evidence that both eggs and hatchlings contain the same diet-derived chemical (the alkaloid actinidine) that is present in adult defensive spray. We also explored potential causes of variation among hatchlings in the capacity to spray, using a fully crossed experiment to investigate how offspring provisioning is affected by sexual vs. parthenogenetic reproduction and high vs. low maternal diet. We found that high maternal diet resulted in increased egg size but slower egg development, and maternal diet interacted with genotype to affect hatchling body size. Eggs laid by male-paired females were larger, developed more quickly, and had higher hatching success by comparison with eggs laid by unpaired females, suggesting that mating and fertilization enhance some aspects of offspring performance. However, hatchlings produced by unpaired females had larger prothoracic glands relative to body size than did hatchlings produced by male-paired females, suggesting that mating is associated with reduced provisioning of offspring with defensive chemicals. Our results reveal a novel example of maternal transfer of food-derived defence chemicals to offspring, and suggest that offspring provisioning with defence chemicals is affected by female reproductive mode.

Methods
MATERIALS AND METHODS

Animal maintenance

Eggs of M. batesii were collected from mixed-sex populations of the Northern genotype and all-female populations from the Southern genotype between Cow Bay and Cape Tribulation, Queensland (Miller et al. 2024a), in February and March of 2020 and raised in the lab at UNSW Sydney. Eggs were maintained in 125 mL plastic containers with moist cocopeat in a room with controlled temperature of 27°C and watered periodically to prevent desiccation. Hatchlings (F0) were housed in full-sib same-sex pairs in cylindric plastic cages (20 × 40 cm) and fed leaves of Pandanus tectorius until they become adults. The F1 descendants of those individuals were used in the experiment described below, including 20 females produced by Northern genotype mothers and 24 females produced by Southern genotype mothers. Reproductive mode was manipulated for F1 females descended from both Northern and Southern populations. Twenty days after the females started to lay eggs, 22 females (14 Northern genotype, 8 Southern genotype) were housed with males in separate cages (sexual treatment), while 22 females (6 Northern genotype, 16 Southern genotype) were kept alone (parthenogenetic treatment). The reproductive treatments were set up in the most ecologically relevant way for M. batesii, since adult females in all-female populations are rarely found in close proximity with other adult females, but many adult females in mixed-sex populations are guarded continuously by males (Boldbaatar et al. 2024). Consistent with previous reproductive mode manipulations on this species (Wilner et al. 2025), females in the sexual treatment laid mostly fertilized eggs as indicated by the near-even sex ratio of their offspring (34 males, 48 females, 41% male offspring), whereas females in the parthenogenetic treatment laid exclusively unfertilized eggs and produced exclusively female offspring (66 females).

Chemical analyses

To investigate whether mothers of M. batesii were transferring their defensive chemicals to offspring, we carried out gas-chromatography mass spectrometry analyses (GC-MS). We compared the chemical composition of egg yolk and hatchlings’ prothoracic glands (offspring of experimental adults chosen randomly and frozen at -80°C until dissection) with the defence fluid sprayed by adults. We extracted chemicals from 18 eggs, 20 hatchlings and 2 adults (one of each sex) to analyse their chemical composition. The eggs were washed with ethanol and had their shell removed to expose the yolk. We dissected the prothoracic glands of hatchlings by cutting the ventral upper portion of the prothorax and removing the glands with a pair of fine forceps. Defence fluid was collected from live adults by tapping the prothorax with an insertion glass vial until the adults sprayed their defence fluid inside the vial. Yolk, prothoracic glands, and defence fluid were collected in insertion glass vials (300 μL). The amount of yolk, prothoracic glands, and defence fluid were not consistent among samples, precluding quantitative comparisons of chemical concentrations. Extractions were performed on ice to avoid loss of volatile compounds. Each sample was transferred to a 1.5 mL glass vial, and 220 μL of chloroform was added and allowed to evaporate overnight at room temperature, leaving a small quantity of precipitate on the bottom of the vial. We then added 100 μL of acetonitrile to each sample and vortexed each vial. The vials with the extracts were placed immediately into a GC-MS instrument for the analyses of the chemical composition of the samples.

Chemical analyses of all extracts were carried out on a Focus DSQ GC-MS equipped with a Triplus autosampler (Thermo Fisher Scientific, Germany). Separations were carried out using a HP-5MS capillary column with 30 m x 0.25 mm of internal diameter and 0.25 mm of film thickness (19091S-433, J&W Scientific, USA). A spitless injection with a 4 μl  injection volume was used, with the injector temperature set to 230°C. The source temperature was set at 200°C, and the start time (solvent delay) was set at 5 min. Oven temperature was as follows:  initial temperature 60°C (held for 2 min), then from 60°C to 115°C (held for 2 min) with a heating rate of 6.5°C/min, then from 115°C to 125°C (held for 1 min) with a heating rate of 1°C/min, and from 125°C to 325°C (held for 9 min) with a heating rate of 7.2°C/min. GC-MS data were processed on the qual browser in Xcalibur software (version 2.1, Thermo Scientific), and spectra matched against the Wiley NIST library (NIST wiley 9 NIST 11) using the NIST Search 2.0 programme for identifications.

Diet manipulation

We implemented two feeding treatments (high food and low food) and two reproduction treatments (sexual and parthenogenetic) in a full factorial design (Figure 2). Each female or pair experienced both feeding treatments sequentially, with each treatment lasting ten days. Half of the experimental adults from each reproduction treatment were assigned initially to the high food treatment and half to the low food treatment (first diet treatment). After the first diet treatment was completed, adults that were in the high food treatment were switched to the low food treatment and vice versa (second diet treatment). In the high food treatment, female-male pairs or single females received two pieces of leaf (~10 cm in length) every second day (five feedings over 10 days) (Figure 2). At each feeding, the sizes of remaining leaves from the previous feeding were recorded and the old leaves were then discarded. The high food treatment can be considered as ad libitum feeding (75% of the total leaf area was eaten). Leaves were eaten mostly by the females as adult males eat much less than females (Boldbaatar 2022). In the low food treatment, female-male pairs or single females received one piece of leaf (~ 10 cm in length) on the third and seventh days of treatment (two feedings over 10 days) (Figure 2). The size of the leaf was recorded, and the leaf was discarded after two days from the day of feeding. Insects in the low food treatment ate 88% of total leaf area, with the unconsumed part of the leaf often dry or covered by the clip used to hang the leaf.

Prior to the experiment, several females were dissected to estimate the length of time required for eggs to develop fully in the ovaries. We observed that eggs in different stages of development, including five to six eggs almost fully yolked, were present in the females’ ovaries and oviduct. Therefore, we opted for discarding the eggs laid over the first five days after the start of the feeding manipulation. Females of M. batesii lay approximately one egg per day, and those first five laid eggs would have been fully or almost fully yolked prior to the diet manipulation and therefore unlikely to be affected by the treatments. Eggs laid in the second half of the treatment (last five days) were collected, photographed, and placed into separate containers with a layer of moist cocopeat at the bottom. The eggs were separated into individual containers so that hatchlings could be frozen without provoking the defence spray behaviour. We recorded the number of eggs laid, eggs size, egg development time (days from laying to hatching), and egg hatching success (proportion of eggs that hatched). The eggs, hatchling thorax and hatchlings’ dissected prothoracic glands were photographed using a camera (Leica MC170HD) mounted on a stereoscope (Leica MZ16A, Wetzlar, Germany) with lighting standardized by using a consistent light source and blocking out ambient light. Traits were then measured from the images using ImageJ (Schneider, Rasband & Eliceiri 2012). We measured the length and width of the eggs (mm), the length of the thorax (cm), and area of the prothoracic glands (calculated as the product of length and maximum width in pixels) as a proxy of the amount of defence fluid present in the glands.

Statistical analyses

Feeding treatment (high vs. low food), reproductive mode (pairing with a male vs. no pairing), maternal genotype (Northern vs. Southern genetic cluster), order of feeding treatment application (low→high vs. high→low), and all interactions among diet, reproduction and genotype, were modelled as fixed effects using the R package glmmTMB (Brooks et al. 2017). Gaussian error distribution was used to model egg development time, egg size (arbitrary units), hatchling size, and prothoracic gland size (log-transformed, arbitrary units), poisson error distribution was used to model the number of eggs laid, and binomial error distribution was used to model egg hatching success (hatched vs. unhatched outcome for each egg). Mother identity was modelled as a random effect in all models to account for repeated measures on individual focal females, but paternal identity was not modelled because parthenogenetically produced females had no father. We modelled prothoracic gland size with hatchling body size (thorax length) included as an additional fixed covariate to account for variation in body size among hatchlings. Because female and male hatchlings are very similar in body size and hatchling sex did not explain variation in any measures of hatchling performance, we did not include hatchling sex in the models.

Akaike’s Information Criterion corrected for small samples (AICc) was used to determine support for covariates and interactions by comparing models containing all possible subsets of the fixed effects from the above glmmTMB models (but retaining maternal identity as a random effect in all models) using the “dredge” function in the MuMIn package (Bartoń 2024). The null (intercept-only) model was included among the models compared for each response variable. The top model and other models with ΔAICc < 2 were considered to have statistical support (Burnham & Anderson 2002; Johnson & Omland 2004; Burnham, Anderson & Huyvaert 2011; Symonds & Moussalli 2011), and model-averaged effect coefficients were calculated based on all supported models using the “model.avg” function. As a measure of model fit, the marginal R2 (proportion of total variance explained by the fixed effects) and condition R2 (proportion of total variance explained by both fixed and random effects) were calculated for the full model using the “r.squaredGLMM” function. All statistical analyses were conducted using R version 4.1.0 (R_Core_Team 2021).