Data from "Foraging Ants as Liquid Brains: Movement Heterogeneity Shapes Collective Efficiency"

Description

The data presented here were used to generate the figures and results of the research article titled "Foraging Ants as Liquid Brains: Movement Heterogeneity Shapes Collective Efficiency" (Fernández-López et al., 2025). To fully reproduce the analyses and figures, you will need to download the related code and instructions available in the GitHub repository cited below. The data are organized into two main folders: --- Raw Data --- The raw_data folder contains tracking data from food and no-food experiments, as well as simulations using the default parameter set (see Table S1 in the manuscript and the script parameters.py in the linked GitHub repository for details). Experimental data are stored as .csv files. Each file contains: *N-ind: identity of each individual *Frame: time in frames *Xmm, Ymm: x and y coordinates in millimeters *Crossings: whether an ant is interacting (1) or not (0) *Time_sec: time in seconds (Frame / 2) *node: the closest intersection in the Y-maze to the (x, y) position Simulation data are divided into two folders, each containing 100 files, one per realization of the model: The main folder, data, contains tracking data analogous to the experiments. These files are in Parquet format and should be read using libraries such as arrow (R) or pyarrow (Python). Each file includes: *T: time in seconds *Frame: time in frames (T x 2, as integer) *N: number of ants in the arena *pos: ant positions *food_target: whether ants are being directed to a food source (0 for no, 1 for the first food patch, 2 for the second food patch) The files in the food_data folder have the following columns: *node: location of the food in the lattice *t: time of detection by an ant *origin: currently unused (contains no meaningful data) --- Processed Data --- This folder contains RData files with pre-processed data for immediate use in reproducing analyses: *det.RData: data from food experiments *nf.RData: data from no-food experiments *hex.csv: Y-maze intersection coordinates (experiment) *hex_sims.csv: Y-maze intersection coordinates (simulation) Regarding simulations: rho_eps.RData contains a summary of results across the parameter space defined by "rho" (scout proportion) and "epsilon" (social copying proportion), as used in Figure 5 of the manuscript. simulations_n contains .csv files showing the number of ants in the arena (N) over time (Frame) for each simulation. Although this can be derived from the raw data, these files are included here to directly support Figure 4 reproduction.

Steps to reproduce

These data result from the processing of video recordings of ants during foraging experiments. Full protocols are described in the related article (Fernández-López et al., 2025) and in Cristin et al. (2024): https://doi.org/10.1038/s41598-024-63307-1. Experimental Setup We conducted foraging experiments in a 2×2 meter arena, structured as a regular Y-maze (honeycomb) pattern. Each trial involved a colony of Aphaenogaster senilis ants exploring the environment for 3 hours. A total of 10 trials were conducted under experimental conditions (with food present), and 12 trials under control conditions (without food). Video Recording and Tracking All trials (with and without food) were video recorded using twelve synchronized cameras, capturing at 2 Hz and in 4K resolution. We processed the footage using a background subtraction method implemented in custom software to extract ant positions. This method involves subtracting consecutive frames to detect moving objects, in this case, the ants. The processed video tiles were then stitched together to reconstruct the full 2×2 meter maze. Tracking was performed using the Hungarian algorithm (based on distance minimization), along with additional custom heuristics to refine trajectories. Data Analysis The resulting tracking data served as the basis for all analyses. Due to the discrete structure of the Y-maze, we conducted both discrete (node-based) and continuous (position-based) analyses. Full results and relevant statistics are provided in the cited publications. For the simulations, please refer to the instructions provided in the associated GitHub repository.

Institutions

Categories

Funding

Related Links

Licence