Predation by shell-breaking crabs on a marine gastropod along a latitudinal gradient in the SW Atlantic: Influence of extrinsic and intrinsic factors
Data files
Aug 07, 2025 version files 97.17 KB
-
Dataset1.csv
72.58 KB
-
Dataset2.csv
3.54 KB
-
README.md
5.79 KB
-
Rscript_Nov2024.R
15.26 KB
Abstract
Biotic interactions—and predation in particular—are thought to follow a latitudinal gradient, increasing towards the tropics; yet empirical evidence remains contradictory and largely based on studies from the Northern Hemisphere. Moreover, the role of environmental variables shaping latitudinal gradients of predation intensity has seldom been tested. Here, we quantify predation by shell-breaking crabs on modern shells of the marine gastropod Trophon geversianus along a latitudinal gradient (40°–54°S) on the SW Atlantic coast. We further evaluate how intrinsic factors (four shell morphometric traits) and extrinsic factors (seven environmental variables and the biogeographic region) jointly influence predation patterns. Fragmentation from crushing predation affected 37% of the shells (544 out of 1,480), with the most frequent damage types being major body whorl damage (28%), deep aperture chips (11%), and extensive aperture chips (6%). When analysed by biogeographic province, fragmentation increased significantly towards the south in the Magellan province. Notably, random forest modelling revealed that intrinsic factors—particularly shell size and thickness—were stronger predictors than extrinsic factors in driving latitudinal variability of shell-breaking crab predation. By highlighting the dominant influence of intrinsic factors over extrinsic ones, this study emphasises the crucial role of species-specific traits in shaping predator-prey interactions across biogeographic regions.
Dataset DOI: 10.5061/dryad.vx0k6dk4s
Description of the data and file structure
Summary
This dataset supports a study investigating how oceanographic variables influence the frequency of shell-breaking crab traces on the marine gastropod Trophon geversianus across 34 sites in the Southwestern Atlantic. The dataset integrates biological data from 1,480 shells with environmental estimates, enabling analyses of spatial trends and ecological drivers of predation.
Contents
R Scripts
This repository includes a single R script that reproduces Figures 3 to 6 from the study, detailing:
Figure 3: PCA of oceanographic variables
Figure 4: Latitudinal trends of individual oceanographic variables and PCA scores
Figure 5: Latitudinal and biogeographic variation in crushing predation frequency
Figure 6: Intrinsic and extrinsic drivers of predation via Random Forest analyses
Files Included
File Name
Description
Dataset1.csv
Biological data: incidence of three types of shell-breaking predation traces on 1,480 Trophon geversianus shells
Dataset2.csv
Environmental data: six oceanographic variables estimated for 34 sites
Rscript_Nov2024.R
R script used to generate Figures 3–6, perform statistical analyses, and reproduce results from the manuscript
Variable Definitions
Dataset1.csv
| Variable | Description |
|---|---|
| Site | Sampling site name (matches with Dataset2) |
| Province | Biogeographic province: "Mag" (Magellan) or "Arg" (Argentinean) |
| type2 | Binary variable (0 or 1): deep aperture chip |
| type3 | Binary variable (0 or 1): extensive aperture chip |
| type5 | Binary variable (0 or 1): major body whorl damage |
| Shell.length | Shell length in mm |
| Shell.Thickness | Shell thickness in mm |
| Lat | Latitude of the site (decimal degrees, negative = South) |
| Lon | Longitude of the site (decimal degrees, negative = West) |
Dataset2.csv
| Variable | Description | Units / Notes |
|---|---|---|
| Site | Sampling site name | Matches with Dataset1 |
| lat | Latitude of the site | Decimal degrees (negative = South) |
| lon | Longitude of the site | Decimal degrees (negative = West) |
| sst | Sea surface temperature | ºC |
| salinity | Sea surface salinity | Practical Salinity Scale (PSS) |
| nflh | Normalised Fluorescence Line Height | mW·cm⁻²·µm⁻¹·sr⁻¹ |
| oxygen | Dissolved oxygen concentration | mol·m⁻³ |
| pH | pH of surface waters | Total scale |
| exposure | Wave exposure index | Arbitrary units |
Missing data: Missing values, if any, are encoded as blank cells or NA.
Software and Libraries
The following R libraries are required to run the script:
psych
FactoMineR
rcompanion
randomForestSRC
Ensure all packages are installed prior to execution.
Execution Instructions
Set your working directory in R.
Load Dataset1.csv and Dataset2.csv using the file.choose() prompt.
Run the script to generate the PDF outputs:
Fig3_nov2024.pdf
Fig4_Malve_nov2024.pdf
Fig5_nov2024.pdf
Fig6_nov2024.pdf
Data Sources
All data were derived from original field collections and public oceanographic datasets (e.g., Bio-ORACLE). Specific estimation methods are detailed in the associated manuscript.
Citation
If you use this dataset, you are encouraged to cite the associated publication:
Malve, M. et al. (2025). Predation by shell-breaking crabs on a marine gastropod along a latitudinal gradient in the SW Atlantic: influence of extrinsic and intrinsic factors. Proceedings B.
Contact
For questions about this dataset or its reuse, please contact:
Mariano Malve
[marianomalve@gmail.com]
Marcelo Rivadeneira
[marcelo.rivadeneira@ceaza.cl]
Code/software
Software and Workflow
Software
The dataset and associated analyses were conducted using R, a free and open-source statistical computing environment. All analyses were run using:
R version 4.3.1
RStudio (optional, for script editing and workflow management)
To reproduce the figures and results, users must have R installed along with the following packages:
Required R Packages
| Package | Purpose | Version (tested) |
|---|---|---|
| psych | PCA parallel analysis and exploratory stats | ≥ 2.3.9 |
| FactoMineR | Principal Component Analysis (PCA) | ≥ 2.8 |
| rcompanion | Effect size statistics (Cramér's V, prop tests) | ≥ 2.4.30 |
| randomForestSRC | Random forest classification for imbalanced data | ≥ 3.2.1 |
Access information
Data was derived from the following sources: BioOracle Database for environmental variables.