Data for: Born this way: Individuality is seeded before birth and robust to ecological stress
Data files
Sep 22, 2026 version files 104.25 MB
-
cichlid_feeding_data.csv
883 B
-
data_for_gallagher_et_al_2026.csv
4.57 MB
-
Model_files.zip
99.68 MB
-
README.md
4.56 KB
Abstract
Behavioral individuality is widespread, yet its developmental origins are unclear. We subjected clonal individuals to a critical ecological challenge—the threat of predation—to disentangle the developmental mechanisms of individuality. Predation may erode or amplify among-individual differences, demonstrating that individuality is developmentally plastic. If individuality is instead determined before birth, it should resist environmental influences. We tracked behaviors of genetically identical fish (Amazon mollies), reared with or without predation stress, continuously from birth for four weeks. While predation shifted mean-level behaviors, the magnitude of individuality was unaffected, indicating that individuality crystallizes robustly over time under stress and without genetic variation. Surprisingly, maternal identity strongly shaped swimming velocity but not refuge use, suggesting distinct developmental mechanisms for different ecologically relevant behaviors. We show that individuality persists despite ecological stress and is seeded before birth through non-genetic factors. Even in the face of a shared environmental challenge, the behavioral trajectories of individuals are unique.
Dataset DOI: 10.5061/dryad.5mkkwh7jx
Description of the data and file structure
We individually reared Amazon mollies in standardized environments, with and without predation stress, from birth through week four of life. We recorded individuals continuously for 12 hours every day and tracked their behavior using the machine learning framework, SLEAP. We also tracked the behavior of their mothers before they gave birth for seven days using the same methods.
Files and variables
File: data_for_gallagher_et_al_2026.csv
Description: Behavioral data for individual Amazon mollies at every hour, from birth through their first month of life.
Variables
- FishId: The unique ID of the individual.
- MomId: The unique ID of the individual's mother.
- PredatorCue: Whether the individual was reared in control or predator treatment (0 or 1).
- Quadrant: Each tracking system tank contained four separated quadrants. This indicates which quadrant the fish was in. Used for data validation.
- Hour: The hour of the day.
- Date: The date.
- ExpDay: The experimental day, also known as the age of the fish (in days).
- YearDay: The day of the entire rearing experiment, starting on the day the first fish was added to the tracking system.
- MeanVis: The mean proportion of time that the fish was visible to be tracked. Used for data validation.
- MeanAct: The mean proportion of time that the fish was moving.
- MeanCorner: The mean proportion of time that the fish was in the corner of its tank that contained the standpipe.
- MedianVel: The fish's median swimming velocity (in pixels per second).
- MeanVis_: Same as MeanVis, but using hourly means instead of daily means.
- MeanAct_: Same as MeanAct, but using hourly means instead of daily means.
- MeanCorner_: Same as MeanCorner, but using hourly means instead of daily means.
- MedianVel_: Same as MedianVel, but using hourly medians instead of daily medians.
- MedianActiveVel: The median swimming velocity when the fish is moving.
- MedianActiveVel_: Same as MedianActiveVel, but using hourly medians instead of daily medians.
- Avg_MedVel_mom: The average MedianVel of that individual's mother.
- Avg_MeanAct_mom: The average MeanAct of that individual's mother.
- Avg_MeanCorner_mom: The average MeanCorner of that individual's mother.
- avg_width: Width (in pixels) of the individual on day 28 of life.
- avg_length: Length (in pixels) of the individual on day 28 of life.
File: cichlid_feeding_data.csv
Description: A list of dates when the cichlid (fish providing predator cues) was fed an anesthetized black molly to deliver alarm cues to Amazon mollies in the predation rearing treatment.
File: Model_files.zip
Description: A compressed folder containing every model output file (.rds) associated with the results of this study. These files are required to load the exact models used when running the the analysis R script to reproduce results.
This zip contains 100 model files in .rds format that are associated with the script "code_for_gallagher_et_al_2026_ecology_letters.Rmd." These are the previously generated models used to produce the results in the manuscript. The model names correspond to the variables included and/or their purpose during analysis, along with the date. To use them in the script, make sure you place the .rds files in your working directory, as the script contains code to load them for each model. If you instead wish to rerun the models yourself, you may comment out the model loading code and instead use the full model code in the script (that is commented out by default).
File: code_for_gallagher_et_al_2026_ecology_letters.Rmd
This code is structured for flexible reproducibility of results. The code was originally run using R version 4.3.3. The model files (.rds) that contain the analysis outputs used in the publication are included in this repository. If the user wished to load the final models, they should download the .rds files before using this script. If the user would rather rerun the models themselves, the full models are also provided in the script directly following the .rds loading code for each model. To run them, simply uncomment the lines of code (remove the "#") for the chosen model. A warning that due to the complexity of the models and the large data set, many of the models may take a lengthy amount of time to run.