This repository contains the datasets and R script for the article "Hormonal modulation of brood allocation underlying evolutionary conflicts in the supercolonial ant Tapinoma darioi", published in the journal Behavioral Ecology and Sociobiology. It contains three datasets, one regarding queen survivorship, one related to methoprene and precocene effect on larvae cannibalization, and one containing data for queen number effect on larvae cannibalization. Additionally, a file containing metadata information for the three datasets is supplemented. Finally, the R script file utilised for statystical analysis in this publication is also present.

In social insects, caste fate is typically determined by larval feeding, but in several species juvenile hormone (JH) also regulates caste development, with elevated JH levels biasing brood towards queens. In some ants, queens influence the caste fate of their offspring by depositing higher JH levels into eggs. We hypothesized that in polygynous societies, individual queens could gain fitness advantages over co-nesting reproductives by increasing JH deposition, thereby producing more sexuals that are reared by workers with low relatedness. To test this, we conducted a bioassay in the supercolonial ant Tapinoma darioi, experimentally altering queens’ circulating JH levels via topical application of methoprene (a JH analogue) and precocene II (a JH synthesis inhibitor), while monitoring brood production. Our results show that workers detect and cull the excessive production of reproductive brood, revealing for the first time the precise moment at which this occurs. These findings support the hypothesis that maternal control of JH deposition into eggs can bias caste fate to the queen’s advantage, particularly in supercolonial species where sexual larvae are often reared by unrelated workers. At the same time, our data indicate that workers are not at a loss in this evolutionary conflict of interests, since they can detect and cannibalize excess queen larvae, likely adjusting the queen/worker ratio towards the colony’s optimum from their own evolutionary perspective. Overall, our study highlights the dynamic interplay between maternal control and worker counter-adaptation in shaping caste determination, providing new insights into the mechanisms and evolutionary conflicts underlying social insect reproduction.

Description of methods used for collection-generation of data: Tapinoma darioi belongs to the recently described T. nigerrimum species complex (a group of four different cryptic species which inhabit the Mediterranean coast of Europe) (Seifert et al. 2017). Identification of the study species as T. darioi was performed based on taxonomic keys, with main traits being related to the male genitalia and worker morphometry in addition to its distribution (Seifert 2012; Seifert et al. 2017; Arcos and García 2024). A total of 385 T. darioi queens and around 8000 workers were sampled from eight different nests located in the vicinity of the Universitat Autònoma de Barcelona (NE Iberian Peninsula, 41º49’N2º10’E). Assuming a nesting strategy similar to the closely related supercolonial Argentine ant (Linepithema humile), whereby individual nests occupy an approximate distance of 3–4 m (Heller et al. 2008), we sampled nests located at least five metres apart.

We conducted toxicity tests before performing the bioassay to determine the optimal treatment solution concentrations, ensuring a biological effect while avoiding premature queen mortality. A total of 105 T. darioi queens were divided among various treatment concentrations of methoprene (SIGMA-ALDRICH), precocene II (SIGMA-ALDRICH), and an acetone-only solvent control group. Each treatment group contained three replicas, and each replica consisted of a colony of five queens and 100 workers. Queens were briefly anesthetized using carbon dioxide and then treated with a single topical application of 1 µl of solution on their abdomen. The solutions were dissolved in acetone at concentrations of 50 µg/µl, 25 µg/µl and 10 µg/µl for the JH analogue methoprene; 50 µg/µl, 25 µg/µl and 1 µg/µl for the JH synthesis inhibitor precocene II; and acetone only for the control group. Queen survival was assessed every 24 h for the next four consecutive days. No mortality was observed after four days, demonstrating that the concentrations typically used in similar experiments (Passera and Suzzoni 1979; Amsalem et al. 2014; Oliveira et al. 2017; Prato et al. 2021) had no apparent short-term toxicity in T. darioi queens.

To test the effect of experimentally manipulating queen JH levels on brood production, experimental colonies (Fig. 1c) were established by placing five queens and approximately 100 workers from the same source nest into a plastic box (20 × 10 × 5 cm) previously coated with Fluon to prevent escape. Workers were collected by partially unearthing field nests during early spring, a period when brood chambers are shallow and located near the nest entrance. This sampling procedure yielded a mixture of worker task groups, including both nurses and foragers. Workers were randomly allocated to experimental colonies while maintaining colony-of-origin matching between queens and workers. Worker number was maintained throughout the experiment; if any colony experienced a noticeable decline, workers from the same source colony were added to restore the initial colony size. Individuals were carefully transferred to avoid introducing eggs, larvae or pupae from the original nest, ensuring that all brood observed during the experimental period was laid after treatment. The experiments were conducted in two separate trials using a matched experimental design, with each trial including its own control group corresponding to the various treatment concentrations. Colonies were fed with 1:3 sugar water solution, mealworms and water ad libitum, and maintained at room temperature (~ 25 °C).

Queen JH signalling was experimentally increased via the topical application of the JH analogue methoprene, while precocene II was used to decrease circulating JH levels, as this molecule causes the destruction of the corpora allata and thereby halts JH production in insects (Amsalem et al. 2014). Treatment groups consisted of the three concentrations used in the toxicity tests, each with five replicates, and followed the same application protocol.

Methods for processing the data:

Brood developmental stage and number were recorded three times per week, on Mondays, Wednesdays, and Fridays for a total duration of three months. To ensure accurate and consistent quantification, brood counts were restricted to larvae and pupae, beginning at the first larval instar. Eggs were typically deposited in dense piles, making reliable counting difficult and highly prone to error under our experimental setup; they were therefore excluded from cumulative brood estimates. Queen numbers were established weekly to assess long-term survivorship during the experiment. All observations were conducted by the same researcher, who was not blind to treatment condition. However, because the study relied on objective quantitative measures obtained through numerical counts, observer bias was unlikely to affect the results.

Instrument- or software- specific information needed to interpret the data:

All statistical analyses were performed in R 4.3.0 (R Core Team 2024). For each colony, we calculated the total cumulative brood produced by summing all larvae and pupae recorded over the study period. Brood cannibalization was estimated from decreases in total brood counts between observations. Newly emerged adults were easily identified by their golden cuticle coloration (Hartmann et al. 2019), retained within their respective colonies, and recorded separately to ensure they were not misclassified as cannibalization events during data analysis. The effect of treatments on queen brood production was analysed with a Poisson mixed-effects model with the MASS package (Venables and Ripley 2002). The model included an interaction term between treatment and days, along with the number of queens as a covariate. Colony ID was treated as a random effect, and an observation-level random effect was used to address overdispersion. An autoregressive correlation structure (corAR1) was applied to ‘days’ to account for temporal autocorrelation. To assess the effect of treatments on the probability of brood cannibalization over time, we used a generalized linear mixed model (GLMM) with the package “glmmTMB” (Brooks et al. 2017), using a negative binomial type 2 distribution. The model was coded with the number of cannibalized larvae as the dependent variable and the interaction between time and treatment as a fixed factor. The number of queens was used as a covariate, whereas colony ID and sampling locations were included as random factors. Finally, queen survivorship throughout the experiment was assessed with a Cox proportional hazards regression model using the “survival” package (Therneau 2026). Pairwise comparisons of treatment effects were performed using estimated marginal means, and rates of change were analysed with the emtrends function of the “emmeans” package for all models (Lenth and Piaskowski 2026). As the primary objective of this study was to investigate the effect of JH titres manipulation on brood allocation outcomes and cannibalization, we focused on the main treatment effects due to a large variation and relatively low sample size per dilution. The detailed results presenting the different concentrations as well subsequent pairwise comparisons are provided in the supplementary material (tables S1-S3, and figures S1 and S2). Four of the 40 replicates were excluded due to unexplained variability, leaving 10 control, 14 methoprene, and 12 precocene replicates. A cutoff was applied to the variable days to avoid zero inflation from brood absence at the start of the experiment.