Dataset DOI: 10.5061/dryad.hmgqnk9wn

Description of the data and file structure

Cenozoic Coral Turnover Analysis

This repository contains the raw data and supplementary material for the analyses and figures from our study:
“Trait networks reveal turnover in Caribbean corals and changes in community resilience through the Cenozoic.”.
Authors: C. G. Clay, A. Dunhill & M. Beger

The R scripts and code required to reproduce the analysis and figure can be found here and in the Git hub repository: [https://github.com/CharlotteGeorgina/CenozoicCoralTraitTurnover]

Citation

If you use this code, please cite the paper:

Clay et al. (in-review). Trait networks reveal turnover in Caribbean corals and changes in community resilience through the Cenozoic. Paleobiology

Acknowledgements

This project was supported by the Leverhulme Trust.

The analyses explore coral species and trait dynamics from the Eocene to the Pleistocene using bipartite networks, trait-trait co-occurrence networks, and dimensionality reduction techniques.

Files and variables

Caribbean_SppbySubEpoch_Eocene_Pleisto_All_new.csv - a collection by species matrix indicating which species were present or absent in each collection.

Column A indicates the epoch or sub-epoch each collection came from,

column B indicates the collection name and

columns C: IM relate to species names.

Caribbean_Reefbuilding_Traits.csv - Trait data for each of the reef building coral species.

Column A:species names,

Column B:genus name.

Column C: Family name.

Column D: whether the species is gobally extant or extinct.

Column E: whether the species is extant or extinct in the Caribbean region.

Column F: References for column D-E.

Column G: whether the species is zooxanthellae or azooxanthellae.

Column H: Reference for column G.

Column I: the numerical integration level for each species.

Column J: name for the integration level for each specues,

Column K: maximum corallite diameter.

Column L: Typical growth type.

Column M: whether the species asexually reproduced through budding and if it did was that intracalicular or extracalicular.

Column N: References for columns I-M.

All missing data represented as NA.

Supplemental information: 

C.CLay_CeonozicTraitNetworks_Supplementary_Final.docx- The supplemental material for the paper includes the tables and figures with the respective legends. 

Code/software

Repository Structure

The code is divided across six main scripts:

1. Bipartite_networks_Jaccard_analysis.R

   Constructs bipartite networks showing species presence across epochs/sub-epochs.

   Uses igraph for network projection and layout.

   Visualises species sharing via heatmaps.

   Calculates Jaccard dissimilarity between time bins to quantify species and trait turnover using the vegan package.

   Visualises species and trait dissimilarity as a line plot

2. Random_network.R

   Generates community weighted meand (CWM) for each collection within each epoch/sub-epoch

   Creates ten random samples of the data to account for differing numbers of collections between epochs/sub-epochs
   (e.g.,. 13 of the 23 Oligocene collections, as only 13 collections were available from the Eocene; early Miocene = 40,
   late Miocene = 38, late Pliocene = 36, early Pleistocene = 98).

3. Weighted_networks.R

   Constructs weighted edge networks representing all possible trait correlations.

   Calculates edge densities for comparison with filtered co-occurrence networks.

4. Trait_Trait_cooccurrenceNetworks.R

   Builds trait-trait co-occurrence networks for three key intervals of environmental change:

    Eocene–Oligocene
   
    Early Miocene–Late Miocene
   
    Late Pliocene–Early Pleistocene
   

   Based on community weighted means (CWM) calculated using the FD package.

   Significant Pearson correlations (p < 0.05) used to build networks.

   Calculates node degree, edge density, and modularity with igraph.

5. PCA.R

   Uses Principal Component Analysis (PCA) to compare trait composition (CWM) across time.

   Highlights community shifts and trait divergence across sub-epochs.

6. PCoA_.R

   Uses Principal Coordinate Analysis (PCoA) and Gower dissimilarity to assess functional richness.

   Constructs convex hulls for visualizing trait space per time bin.

   Trait vectors overlaid to indicate drivers of functional divergence.

   Uses the mFD package for trait space quality assessment.

Data & Dependencies

Data files are included in this repository.

Required R packages: igraph, vegan, FD, mFD, ggplot2, and others as listed in individual scripts.

Visualisation Tools

Final visualisation of bipartite and trait-trait networks was enhanced using Inkscape (v1.2.2).

Convex hulls and trait vectors in PCoA plots created using vegan.## Access information

Other publicly accessible locations of the data:

• https://github.com/CharlotteGeorgina/CenozoicCoralTraitTurnover

Data was derived from the following sources:

• The Paleobiology Database (The Paleobiology Database), which is made available under a CC0 International License. We gratefully acknowledge the contributions of PBDB authors and data providers. See https://paleobiodb.org for source data and contributor information.