Plasticity in Anti-Predator Behaviour of Coral Reef Fish in the Andaman Islands

Abstract

Since the 1950s, mechanization and increased demand have drastically increased marine resource extraction, leading to the loss of top predators in coastal ecosystems. This loss has caused trophic downgrading, altering ecosystem processes and prey behavior. Prey species often implement anti-predator strategies, such as increased vigilance or foraging in safer but less resource-rich areas, at the cost of fitness. The removal of predators homogenizes risk in the landscape, disrupting foraging-safety trade-offs and cascading effects on lower trophic levels. For example, altered herbivore foraging behavior has impacted seagrass distribution in coral reefs. Additionally, human activities may create novel risks for prey species. This study investigates the behavioral effects of increased fishing pressure in coral reef ecosystems of the Andaman Islands. Using predator cue experiments, we assess whether reef herbivores respond differently to human and non-human predators and examine the impact of fishing boats on predator abundance and behavior. We also explore how these effects vary across areas with differing fishing pressures.

Data

Individuals (individuals.csv)

• Description: Sampled individuals in each plot.
• Columns:
  • ind_id: Unique individual ID.
  • species: Species identity.
  • group: Observed as part of a group (bool).
  • size_class: Approximate size class.
  • coordinates: Bounding box coordinates (list of integers).
  • file: Video file name.
  • time_in: Time first recorded (ms).
  • time_out: Time last recorded (ms).
  • remarks: Additional notes.

Observations (observations.csv)

• Description: Behavioral observations of sampled individuals.
• Columns:
  • ind_id: Unique individual ID.
  • time: Time of behavior observation (ms).
  • behaviour: Observed behavior.

Predators (predators.csv)

• Description: Predator presence in each plot.
• Columns:
  • predator_id: Unique predator observation ID.
  • species: Predator species.
  • size_class: Predator size class.
  • time: Time first recorded (ms).
  • remarks: Additional notes.

Sites (sites.csv)

• Description: Sampling site information.
• Columns:
  • date: Sampling date.
  • deployment_id: Unique deployment ID.
  • location: Sampled location name.
  • protection: Protection status (MPA or not).
  • time_in: Dive start time.
  • time_out: Dive end time.
  • depth-avg: Mean depth (m).
  • depth-max: Max depth (m).
  • visibility: Visibility (m).
  • lat, lon: Latitude and longitude.
  • crew: Boat crew names.
  • remarks: Additional notes.

Plots (plots.csv)

• Description: Plot-level information.
• Columns:
  • index: Unique plot ID.
  • time: Total sampling time (s).
  • min_vid, max_vid: Min and max video lengths (s).
  • n_videos: Number of videos.
  • path: Relative path to plot folder.

Samples (samples.csv)

• Description: Subsamples within each plot (2-minute intervals).
• Columns:
  • plot_id: Unique plot ID.
  • sample_id: Unique subsample ID.
  • start_time: Subsample start time (s).
  • video: Video file name.
  • status: Completion status.

Benthic Cover (benthic-cover.csv)

• Description: Benthic cover analysis of plots.
• Columns:
  • plot_id: Unique plot ID.
  • label: Benthic cover class label.
  • category: Cover class category.
  • subcategory: Cover class subcategory.

Rugosity (rugosit.csv)

• Description: Chain transect data from each plot.
• Columns:
  • Deployment-id: Unique deployment ID.
  • treatment: Plot treatment.
  • sample: Sample number.
  • Measured-length-cm: Chain length on benthos (cm).

Scripts

cleaning.py

• Description: Processes raw data files into cleaned datasets for analysis.

summaries.py

• Description: Summarizes the cleaned data for analysis.

generative_model.py

• Description: Generate data based on the processes described in the DAG to test the statistical model.

params.py

• Description: Parameters for the generative model.

run.py

• Description: Python script to run the stan models, make figures and summary tables.

validate.py

• Description: Validation of the statistical model.

model.stan

• Description: Bayesian multilevel heirarchial model.

functions/

• Description: Contains utility functions used in the analysis scripts.

Usage

1. Clone the repository:
   bash git clone https://github.com/cheesesnakes/antipredator-plasiticity.git
2. Install dependencies using uv (install uv if not already installed):
   

bash uv sync 3. Run scripts for data processing, analysis, and visualization. bash uv run summaries.py uv run statistical_model.py

License

This project is licensed under the MIT License.