Shark Predation on Dolphins in Southeast Queensland, Australia
Data and Code Repository
Study Overview
Photo-identification data examining shark bite scars on five dolphin species in southeast Queensland, Australia. The study investigates predator-prey dynamics by analysing shark bite wound prevalence across different species and habitat types (sheltered embayments vs. open coastal waters).

Data Files Required

Primary Data Files

1. BiteList.csv - Complete records of all individual dolphin observations including bite wound data
2. dat_unique.csv - Unique individual dolphins (duplicates removed) with bite presence/absence
3. AvgGroupSize.csv - Average group size and depth data for individual dolphins
4. Film.csv - Photo records indicating which side (left/right) of dolphin was photographed
5. All_Sightings.csv - Complete sighting history for all identified individuals
6. DolphinNames.csv - Individual dolphin identification codes with species and site information
   Generated Output Files (created by scripts)
   • SidesandSightings.csv
   • Dolphin_subset.csv
   • Dolphin_Subset_Sightings_Bites.csv
   • d_subset.csv
   • Dolphin_Resight_Totals.csv

Variable Descriptions
BiteList.csv

• ID: Unique identifier for each individual dolphin (3-letter code)
• Species: Dolphin species (full binomial name)
  • Sousa sahulensis (Australian humpback dolphin)
  • Orcaella heinsohni (Australian snubfin dolphin)
  • Tursiops aduncus (Indo-Pacific bottlenose dolphin)
  • Tursiops truncatus (Common bottlenose dolphin)
  • Delphinus delphis (Common dolphin)

• Shark bite: Presence of shark bite wound (Y = yes, N = no)
• Location: Survey location code
  • SC = Sunshine Coast (open coastal waters)
  • SGSS, HB combined make GSMP = sheltered embayments (Great Sandy Marine Park)
• Age.Class: Age classification
  • Calf = dependent calf
  • Noncalf = juveniles and adults combined
  • Adult
  • Juvenile
• Sex: Sex of individual (M = male, F = female, U = unknown)
• BodyArea: Anatomical location of shark bite wound
  • Dorsal fin
  • Peduncle
  • Fluke
  • Head
  • Body
• HealingStage: Healing stage of shark bite wound
• Fresh (recent injury with visible wounds)
• Healing (intermediate stage)
• Healed (scarred tissue only)
• Side: Side of body photographed (L = left, R = right)

AvgGroupSize.csv

• DolphinID: Individual dolphin identification code (matches ID from BiteList.csv)
• AverageGroupSize: Mean group size across all sightings of the individual
• AverageDepth: Mean water depth (meters) across all sightings

Film.csv

• FileNumber: Photo number in folder
• Date: date of the sighting
• Identifications: Space-separated list of dolphin IDs in photograph
• Description_Category: Photo description including side photographed (contains "L dorsal" or "R dorsal")
• Shark Bite: Yes or no of bite in photo

All_Sightings.csv

• ObservationID: Sighting number
• StartTime: Time of sighting
• DolphinID: Individual dolphin identification code
• Date: Date of sighting

DolphinNames.csv

• DolphinID: Individual dolphin identification code
• Species: Species binomial name
• Site: Primary location/site where individual was photographed
• Sex: Sex of individual (M = male, F = female, U = unknown)

## Missing Data

Empty cells occur throughout the dataset where data were not available or not applicable (e.g., HealingStage is only recorded for individuals with shark bites; Sex is empty when not determined from photographs). All R analysis scripts include appropriate handling of missing values through na.omit() and conditional filtering, so empty cells should not be replaced with placeholder text as this would interfere with script functionality.

Code Files

1. 1_Sharkbite_Graphs.R
   Purpose: Data preparation, summary statistics, and visualization of shark bite patterns
   Key Functions:
   • Loads and cleans BiteList.csv
   • Creates binary Sharkbite variable (0/1)
   • Derives Habitat_type from Location
   • Removes duplicate individuals to create dat_unique
   • Generates summary statistics by species and location
   • Creates visualizations of bite percentages by species and habitat
   • Filters data to create subset for analysis (excludes Orcaella heinsohni and Delphinus delphis due to small sample sizes)
   • Calculates power analyses for ANOVA tests
   Outputs:
   • Summary tables of bite prevalence
   • Publication-quality figures (TIFF format):
   o Shark_bite_percentage_by_habitat.tiff
   o Bite_count_no_zeros.tiff
   o Bite_count_bothsides.tiff
   • dat_unique.csv
   • Dolphin_subset.csv
   • d_subset.csv
   Dependencies: tidyverse, tidyr, pwr
2. 1b_Both_Sides_Photographed.R
   Purpose: Determines which dolphins have been photographed on both left and right sides
   Key Functions:
   • Processes Film.csv to identify which photos show left vs. right dorsal fin
   • Merges with sighting data to create comprehensive record
   • Adds Both_sides variable (Y/N) to individual records
   • Creates prepare_dolphin_data.R function (sourced by other scripts)
   Outputs:
   • SidesandSightings.csv
   • Both_sides column added to dat_unique
   Dependencies: dplyr, stringr
3. 2_Resights.R
   Purpose: Calculates resighting frequencies for individual dolphins
   Key Functions:
   • Counts unique sighting dates per individual
   • Removes duplicate date entries
   • Merges with dolphin names and species data
   • Calculates mean, SD, and range of resighting frequencies
   Outputs:
   • resightings_no_repeats.csv
   • Dolphin_Resight_Totals.csv
   Dependencies: dplyr
4. 3_Chisquared.R
   Purpose: Chi-squared tests for categorical associations
5. 4_GLM.R
   Purpose: Primary statistical analyses using generalized linear models
   Key Analyses:
   5a. GLM for Shark Bite Presence/Absence (Binomial family)
   • Full model: Sharkbite ~ SpeciesAge.ClassHabitat_typeAverageGroupSize + AverageDepth
   • Model selection using AICc (via dredge function from MuMIn package)
   • Analysis performed on:
   o All data (dat_unique1)
   o Subset excluding small sample species (dolphin_subset)
   • Generates predicted probability plots with confidence intervals
   5b. GLM for Bite Count
   • Full model: Number_of_Bites ~ SpeciesAge.ClassHabitat_typeAverageGroupSize + AverageDepth
   • Analyzes number of bite wounds per individual (for those with both sides photographed)
   • Tests for overdispersion
   • Creates visualization of mean bite count by species
   Outputs:
   • combined_plots.tiff (predicted probabilities by species, age class, and habitat)
   • Total bites species difference.tiff
   • Model summaries and diagnostics
   Dependencies: tidyverse, car, performance, pROC, MuMIn, marginaleffects, boot, pwr, partykit, patchwork, ggpubr, GGally
6. 5_Avg_depth_group_size.R
   Purpose: Calculates and prepares average depth and group size covariates
   Note: Creates AvgGroupSize.csv used in GLM analyses

Software Requirements

R Version
Analyses were conducted in R version 4.4.3

Required R Packages

• tidyverse (data manipulation and visualization)
• stringr (string processing)
• ggplot2 (visualization, part of tidyverse)
• car (regression diagnostics)
  • performance (model diagnostics)
• pROC (ROC curve analysis)
• MuMIn (model selection and averaging)
• marginaleffects (predicted values and marginal effects)
• boot (bootstrap methods)
• pwr (power analysis)
• partykit (classification trees)
• patchwork (combining plots)
• ggpubr (publication-ready plots)
• GGally (correlation matrices)

Installation

install.packages(c("tidyverse", "car", "performance", "pROC", "MuMIn",
"marginaleffects", "boot", "pwr", "partykit",
"patchwork", "ggpubr", "GGally"))

Data Collection Methods

Photo-identification surveys were conducted in southeast Queensland coastal waters from 2022 to 2025. Shark bite wounds were identified from high-quality photographs based on characteristic scarring patterns.
Geographic Extent
Study Area: Southeast Queensland, Australia

Survey Locations:

• Open coastal: Sunshine Coast waters
• Sheltered embayment: Great Sandy Marine Park including the embayment of Hervey Bay and waters of the Great Sandy Straits
• Coordinate System: WGS84 decimal degrees

Statistical Methods

Primary Analyses

Generalized Linear Models (GLMs) with binomial error distribution for bite presence/absence and Poisson distribution for bite count data. Full factorial models were constructed including:

• Fixed effects: Species, Age.Class, Habitat_type, AverageGroupSize, AverageDepth
• Interactions: Species × Age.Class × Habitat_type × AverageGroupSize

Model selection performed using Akaike Information Criterion corrected for small sample sizes (AICc) via multi-model inference (dredge function). Best-fitting models selected based on lowest AICc values.

Power Analyses

A priori power calculations conducted using pwr.anova.test with:

• Medium effect size (f = 0.5)
• Significance level α = 0.05
• Desired power = 0.80
  Species with insufficient sample sizes (n < 14) were excluded from final analyses.

Data Subsetting Decisions

Species Exclusions

• Orcaella heinsohni (Australian snubfin dolphin): Small sample size
• Delphinus delphis (Common dolphin): Small sample size
• Tursiops truncatus (Common bottlenose dolphin): Excluded from some analyses due to limited sample size with both sides photographed

Photo Coverage Requirements

For bite count analyses, only individuals photographed on both left and right dorsal fins were included to ensure complete wound detection and avoid underestimation of bite frequency.

Ethical and Permit Information

We acknowledge the Gubbi Gubbi/Kabi Kabi and Butchulla peoples, the traditional custodians of the lands and waters on which this work was conducted and pay our respects to Elders past, present and emerging. All research was conducted under permits P-MPP-100189673 and WA0046948 and under ethics approval from the University of the Sunshine Coast Animal Ethics Committee (ANS 2287).

Data Availability and Reuse

License

This dataset is released under Creative Commons Zero (CC0) Public Domain Dedication.

Citation

When reusing these data, please cite: Hume, G.V., Levengood, A.L., Webster, G.W., Townsend, K.A., Peddemors, V.M., Holmes, B.J.. (2026). Data from: Investigating the predation risk of coastal dolphins via the presence of shark bite scars across south-east Queensland, Australia. Dryad Digital Repository. https://doi.org/10.5061/dryad.76hdr7t9t

Contact Information

Corresponding Author: Georgina V. Hume georginahume95@gmail.com

Repository Structure

├── README.md (this file)

├── data/

│ ├── BiteList.csv

│ ├── AvgGroupSize.csv

│ ├── Film.csv

│ ├── All_Sightings.csv

│ └── DolphinNames.csv

├── code/

│ ├── 1.Sharkbite_Graphs.R

│ ├── 1b.Both_Sides_Photographed.R

│ ├── 2.Resights.R

│ ├── 3.Chisquared.R

│ ├── 4.GLMs.R

│ └── 5.Avg_depth_group_size.R

Acknowledgements

We want to thank numerous field assistants and volunteers who have contributed to data collection in the field for the Dolphin Research Project south-east Queensland. We are grateful for the funding support provided by the Holsworth Wildlife Research Endowment & Ecological Society of Australia, the Ethel Mary Read Research Grant, the Oceania Project, CID foundation, the Winifred Violet Scott Charitable Trust and UniSC HDR funds that have allowed us to carry out this work.

Version History

• v1.0 (2026-01-08): Initial release accompanying manuscript submission to Ecology and Evolution

• v1.1 (2026-04-29): Updated README to accompany manuscript edits during revisions

Last Updated: 22/05/2026