The following dataset compiles the files associated with the study "Context-dependent aggression and ecological dominance in a Mediterranean ant community", submitted to Insectes Sociaux. The following files are included: Data (behavioural tests and nest monitoring datasets), Metadata (excel file that explains the data), Supplementary material (additional tables not included in the manuscript), Scripts (R scripts that read the data files and replicated the analysis) and Readme (an explanation of the dataset).

R studio, 4.4.0

METHODOLOGICAL INFORMATION

1. Description of methods used for collection-generation of data: Ant colonies in the study area were first surveyed and georeferenced, with species identified using taxonomic keys based on body morphometrics and male genitalia (Arcos and García 2023). Relative abundance was estimated, and the three most abundant species (A. senilis, M. barbarus and T. darioi) were selected for laboratory bioassays. Approximately 100 workers per species were collected from three randomly chosen colonies located 2–35 m apart. Ants were maintained temporarily in identical Fluon-coated plastic containers with soil substrate, kept at room temperature, and fed weekly with mealworms and a sugar solution, with water provided ad libitum, for the behaviour tests. Aggression tests were conducted under controlled laboratory conditions in two periods (October 2023 and March 2024) to evaluate seasonal variation in behaviour (Grangier et al. 2007). Encounters consisted of one focal ant against a group of three ants, and three types of interactions were tested: (1) within-colony controls, (2) between different colonies of the same species, and (3) between colonies of different species. For M. barbarus, assays were performed separately for minor and major worker castes.

In each trial, one ant was placed at the centre of a Fluon-coated arena, while three ants were positioned at the periphery. Individuals were separated initially and allowed to acclimate for 30 seconds before observations. Each trial lasted 120 seconds, during which the timing of the first interaction and subsequent behaviours were recorded every 10 seconds using a standardized aggression scale. Each assay type was replicated six times per colony in both seasons, generating two datapoints per trial (focal and contender). Ants were randomly selected, not reused, and all experiments were performed by a single observer blind to colony identity. This experiments resulted in the behaviour database ("behaviour_dataset.tab").

Field data were obtained by georeferencing and monitoring all nests biweekly during peak activity hours (12:00–16:00) from October 2023 to May 2024, resulting in the monitoring dataset ("datanests.tab"). Nest size was visually estimated using a standardized abundance scale based on the number of individuals at the entrance. Nest status (occupied, empty, or replaced) was recorded over time, and nest turnover was defined as occupation by another species lasting at least two weeks. Nests were individually marked to avoid duplication, and workers from newly detected nests were collected for laboratory identification. Only nests showing active occupation were included.

Bibliography: Arcos J, García F (2023) Hormigas de la península Ibérica e islas Baleares. Independently published Grangier J, Le Breton J, Dejean A, Orivel J (2007) Coexistence between Cyphomyrmex ants and dominant populations of Wasmannia auropunctata. Behav Process 74(1). https://doi.org/10.10 16/j.beproc.2006.10.009

1. Methods for processing the data: Changes in nest dynamics were analysed using multinomial logistic regression to evaluate temporal variation in the relative frequency of species’ nests. Post-hoc pairwise comparisons were performed using estimated marginal means with Tukey adjustment, and overall model significance was assessed with ANOVA. Nest turnover events were quantified monthly, and differences in mean colony size among species were tested using Kruskal–Wallis and Dunn post-hoc tests with Bonferroni correction.

Aggression levels were quantified using an aggression index (AI), calculated as the sum of aggressive interactions recorded for both focal individuals and contender groups. AI was analysed using generalized linear mixed models (GLMMs) with a negative binomial distribution. Fixed effects included focal species, group size, season, internest distance, time to first interaction, and a composite interaction combining behavioural condition and contender species. Additional interactions tested species-specific responses, while random effects included focal colony, contender colony, and trial ID to account for repeated measures. Post-hoc comparisons used estimated marginal means with Tukey-adjusted p-values, and model significance was assessed using Type III Wald chi-square tests. A second GLMM evaluated differences in latency to first interaction.

A separate multinomial model was used to estimate probabilities of behavioural categories across species pairs, including group size and season as fixed effects, and internest distance and latency as covariates. Post-hoc comparisons were corrected for false discovery rate. Data visualizations were generated using ggplot2.

1. Instrument- or software- specific information needed to interpret the data: All statistical analyses were conducted in R (v4.4.0).