Dataset associated with the manuscript "Stingray spine diversity reflects performance trade-offs linked to puncture and breakability" in Proceedings B (Publication DOI: 10.1098/rspb.2026.0545)

Data and R Code Associated with the Paper

Stingray spine diversity reflects performance trade-offs linked to puncture and breakability

Authors

• Emily Poulin: Department of Ecology and Evolutionary Biology, University of California Irvine
• Matthew A. Kolmann: Department of Biology, University of Louisville
• Melanie L.J. Stiassny: Department of Ichthyology, American Museum of Natural History
• Karly E. Cohen: Friday Harbor Labs, University of Washington
• Jonathan M. Huie: Department of Ecology and Evolutionary Biology, University of California Irvine
• Jules Chabain: Department of Evolution, Ecology, and Behavior, University of Illinois Urbana-Champaign
• Christopher M. Martinez: Department of Ecology and Evolutionary Biology, University of California Irvine

Contact (Corresponding Author): - epoulin@uci.edu

---

Study Summary

We examine the functional morphology of defensive spines in 30 stingray species spanning five families and diverse habitats. By integrating morphological data, biomechanical modeling, and performance tests, we find that spine evolution reflects a trade-off between puncture performance, anchoring ability, and durability. These trade-offs are likely central to the effectiveness of stingray defenses in different ecological contexts.

---

Software

Analyses were done in the R statistical environment, version 4.4.1.

R Packages Used

• geomorph, version 4.0.10
• geiger, version 2.0.11
• ggplot2, version 3.5.2
• phytools, version 2.4.4
• dplyr, version 1.1.4
• nlme, version 3.1-168
• ape, version 5.8-1

---

Description of the Data and File Structure

1. AllTaxa.csv

Table of length equal to number of sampled stingray species (30), with the following columns:

Variable Definitions and Units

Unless otherwise noted, all measurements are unitless ratios derived from micro-CT scan data.

• Habitat: Habitat classification coded as:
  • marine = marine species
  • freshwater = freshwater species
• Family: Taxonomic family designation.
• PhyloSpp: Species names (formatted to match phylogenetic tree tip labels).
• Dorsoventral: Average first principal stress under dorsoventral bending simulations from finite element analysis (FEA) (units: MPa).
• Lateral: Average first principal stress under lateral bending simulations from finite element analysis (FEA) (units: MPa).
• SM50: Second moment of area measured at 50% of spine length from the base (units: mm^4).
• IncludedAngle: Spine tip included angle representing tip sharpness (units: degrees).
• SerrationAngle: Angle of serration orientation relative to the primary spine axis (units: degrees).
• PropSerrated: Proportion of total spine length containing serrations (unitless proportion, ranging from 0–1).
• SerrationLength: Length of an individual serration at spine midpoint divided by total spine length (unitless ratio).
• SA.Vol: Surface area-to-volume ratio of the spine (units: mm^-1).
• SpineAR: Spine aspect ratio calculated as spine length divided by maximum spine width (unitless ratio).

2. DistributionOfTrees.nex

Full distribution of 10,000 pruned and time-calibrated phylogenetic trees from Stein et al. (2018). A random subset of 500 trees was used to assess sensitivity to phylogenetic uncertainty.

3. Subset_trees.nex

Random subset of 500 trees used to assess sensitivity to phylogenetic uncertainty.

4. MolecularTaxaOnly.csv

Trait data for subset of species with molecular data available in the MolecularTaxaOnly.tre phylogeny. Same columns as AllTaxa.csv.

5. MolecularTaxaOnly.tre

Phylogenetic tree from Stein et al. (2018) containing only the 22 sampled stingray species with molecular data available.

6. DynamicPuncture.csv

Dynamic performance data for idealized spine physical models:

• Model: Identifier (1–5), corresponding to estimated trait values along PC1
• Puncture_Distance: Depth of dynamic puncture into silicone (mm)

7. SummaryTree

Summary phylogenetic tree from a distribution of 10,000 time-calibrated trees generated by Stein et al. (2018) for all sampled species.

8. DynamicRemoval.csv

Dynamic performance data for idealized spine model removal tests:

• Model: Identifier (1–5), corresponding to estimated trait values along PC1
• Removal_Force: Average force required to extract spine model from silicone (N)

9. SlowPuncture.csv

Quasi-static performance data for idealized spine model puncture tests:

• Model: Identifier (1–5), corresponding to estimated trait values along PC1
• Max_Force: Peak force required to puncture silicone (N)

10. SlowRemoval.csv

Quasi-static performance data for idealized spine model removal tests:

• Model: Identifier (1–5), corresponding to estimated trait values along PC1
• Min_Force: Peak force required to extract spine model from silicone (N)

---

Code / Software

1. SummaryTree.R

R code containing main phylogenetic comparative analyses. Includes PCA, PGLS regressions, phylomorphospace plots, MANOVAs, disparity tests, and estimation of trait values along PC1 for physical model design.

2. DistributionOfTrees.R

R code for re-running MANOVA, ANOVA, and PGLS analyses across 500 trees sampled from DistributionOfTrees.nex.

3. MolecularTaxaOnly.R

R code for re-running MANOVA, ANOVA, and PGLS analyses across stingray species with molecular data, using tree MolecularTaxaOnly.tre.

4. Puncture_Removal.R

R code analyzing dynamic and quasi-static performance of physical models in puncture and removal tests. Includes ANOVA, Tukey HSD post-hoc tests, and figure generation.