Data from: Punctuated versus gradual shifts in the multivariate evolutionary process: a test with paired radiations of scincid lizards
Data files
Jan 16, 2025 version files 3.35 MB
-
README.md
14.15 KB
-
SysBio-skink-covariance-data-scripts.zip
3.33 MB
Abstract
As lineages become separated in time, they are expected to accumulate mutational (or developmental-genetic) differences that influence the macroevolutionary trajectories of those lineages even under similar environmental conditions. Here, we compare the dynamics of phenotypic evolution in radiations of scincid lizards from Australia and Madagascar that are separated by more than 100 million years of independent evolution and show rampant phenotypic parallelism. We collected linear measurements of the skull, limbs, and limb girdles from micro-CT scans of 94 Australian and 29 Malagasy species. Using multivariate comparative methods, we tested whether the underlying evolutionary covariance structure for this superficial parallelism was conserved and whether these patterns were consistent across distinct functional modules. Malagasy and most Australian skinks have similar covariance matrices for skull evolution. Results are ambiguous for limbs and limb girdles, as some trait subsets support different evolutionary processes and for other subsets, a shared covariance matrix could not be rejected. However, across most trait sets, the extremely speciose Australian genus Ctenotus exhibits a radically different covariance structure from all other lizards in these groups, including several closely related genera. The shift in Ctenotus demonstrates that the architecture of phenotypic evolutionary correlation can change at relatively shallow timescales and may explain the unique position of this clade in morphospace relative to other scincid lizards from both geographic regions. More generally, our results demonstrate that the multivariate evolutionary process can change dramatically in a relatively short period of time.
README: Data from: Punctuated versus gradual shifts in the multivariate evolutionary process: a test with paired radiations of scincid lizards
https://doi.org/10.5061/dryad.6wwpzgn65
This data package contains the data, R scripts, and supplemental materials associated with "Punctuated versus gradual shifts in the multivariate evolutionary process: a test with paired radiations of scincid lizards" by Stepanova et al. These files include museum specimen information, phenotypic data, R scripts, and results files. CT scans of UMMZ specimens are available on University of Michigan Deep Blue Data Repository (https://doi.org/10.7302/wj1k-kp72) and CT scans of non-UMMZ specimens are available on MorphoSource (https://www.morphosource.org/projects/000699053?locale=en).
Description of the data and file structure
corhmm folder contains all files related to analyses run using the R package corhmm version 2.8 (Boyko and Beaulieu 2021 Methods Ecol. Evol.). The files include:
- corhmm-limb-analyses.R -- R script to run corhmm analyses.
- covarion-mod-skinks-and-outgroups -- R data file containing the results from the best fit model. This file was used to create Supplemental Figure 19 (see SupplementalFigure19-skink-recon.R for script to create figure).
- limb-data.csv -- csv file with limb data for all species. The columns are:
- tree: valid scientific name, corresponding to name on tree.
- family: name of family species is in.
- limbs: number of limbs species has (0, 2, 4). Limb data were derived from Camaiti et al. 2022 J. Biogeogr.
- skinks-et-al-tree.phy -- trimmed phylogeny from Title et al. 2024 Science, used in corhmm analyses.
- SupplementalFigure19-skink-recon.R -- R script to create Supplemental Figure 19.
covariance analyses folder contains all files for covariance analyses using mvMORPH version 1.1.6 (Clavel et al. 2015 Methods Ecol. Evol.) and the rate/rate-shift likelihood functions. The files include:
R scripts:
- 2022-07-18-fxns.R -- secondary R script containing functions to create simmap codings for mvMORPH covariance analyses, called in main R script covariance-mvmorph-analysis.
- covariance-likelihood-analysis.R -- main R script, contains evolutionary covariance analyses using rate and rate-shift functions.
- covariance-mvmorph-analysis.R -- main R script, contains evolutionary covariance analyses using mvMORPH.
- fxns-covariance-analysis.R -- secondary R script containing functions to get all nodes for a tree, called in main R script covariance-mvmorph-analysis.
- fxns-likelihood.R -- secondary R script containing the rate and rate-shift functions, called in covariance-likelihood-analysis.R.
csv files:
- classifiers.csv -- csv file with specimen and species information, used in analyses.
- TreeName_Rabosky: valid scientific name corresponding to name on phylogeny.
- ScanName: scientific name corresponding to name of museum specimen and name on CT scan.
- MuseumNumber: specimen number.
- clade: indicates whether species is from Australian radiation or Malagasy radiation.
- pcfiggroup: group identity breaking up Ctenotus skinks (Ctenotus), non-Ctenotus Australian skinks (Australia), and Malagasy skinks (Madagascar).
- limb_reduction: indicates whether the species has limb reduction or not.
- skink-morphometrics-full.csv -- csv file containing skull, limb, and girdle measurements for all skink species in study. Measurements are in mm and were done in VGStudio. See Appendix 1 and Supplemental Figures 1 + 2 for details about which traits were measured.
other files in folder:
- mvmorph-covariance-matrices folder -- folder containing 7 txt files with results (including both statistical results and covariance matrices) from mvMORPH analyses. Each txt file has results from a particular partition scheme for each trait subset. A summary of the statistical results is located in Supplemental Tables 4 and 5.
- skinktree.phy -- phylogeny used for analyses with all skinks in study. Trimmed from Title et al. 2024 Science.
figures folder contains scripts and corresponding files to create all main text figures (except Fig. 1).
R scripts:
- Figure2-phylogeny.R -- R script to create Figure 2.
- Figure3-kde-pca.R -- R script to create Figure 3.
- Figure4-akaike-weight.R -- R script to create Figure 4.
- Figure5a-covariance-shifts-on-tree.R -- R script to create Figure 5 (covariance shifts on phylogeny).
- Figure5b-covariance-matrix-plots.R -- R script to create Figure 5 (ellipses of the eigenvectors).
- Figure6-covariance-correlation-matrices.R -- R script to create Figure 6.
- Figure7-tip-rates.R -- R script to create Figure 6.
- phylo_funx.R -- secondary R script containing functions to create breaks in tree branches, for plotting purposes (called to create Fig. 5a).
csv files:
- aic-table.csv -- table of AIC values from mvMORPH analyses, used for graphing Figure 4. Each row corresponds to a single trait subset. The first column (TraitSubset) provides the name of the trait subset. The remaining columns list the 16 models tested.
- classifiers -- csv file with specimen and species information. The columns are:
- TreeName_Rabosky: valid scientific name corresponding to name on phylogeny.
- ScanName: scientific name corresponding to name of museum specimen and name on CT scan.
- MuseumNumber: specimen number.
- clade: indicates whether species is from Australian radiation or Malagasy radiation.
- pcfiggroup: group identity breaking up Ctenotus skinks (Ctenotus), non-Ctenotus Australian skinks (Australia), and Malagasy skinks (Madagascar).
- limb_reduction: indicates whether the species has limb reduction or not.
- cov-matrix-cranial-bones-ctenotus -- covariance matrix for Ctenotus from mvMORPH analyses for the cranial bones trait subset, unrelated eigenvectors model. Trait names are written in the first row and first column. Used to graph Figure 5b and Figure 6.
- cov-matrix-cranial-bones-others -- covariance matrix for non-Ctenotus skinks from mvMORPH analyses for the cranial bones trait subset, unrelated eigenvectors model. Trait names are written in the first row and first column. Used to graph Figure 5b and Figure 6
- cov-matrix-limbs-girdles-ctenotus -- covariance matrix for Ctenotus from mvMORPH analyses for the limbs and girdles trait subset, unrelated eigenvectors model. Trait names are written in the first row and first column. Used to graph Figure 5b
- cov-matrix-libms-girdles-others -- covariance matrix for non-Ctenotus skinks from mvMORPH analyses for the limbs and girdles trait subset, unrelated eigenvectors model. Trait names are written in the first row and first column. Used to graph Figure 5b
- cov-matrix-limbs-only-australia -- covariance matrix for non-Ctenotus Australian skinks from mvMORPH analyses for the limbs only trait subset, unrelated eigenvectors model. Trait names are written in the first row and first column. Used to graph Figure 6.
- cov-matrix-limbs-only-ctenotus -- covariance matrix for Ctenotus from mvMORPH analyses for the limbs only trait subset, unrelated eigenvectors model. Trait names are written in the first row and first column. Used to graph Figure 6
- cov-matrix-limbs-only-madagascar -- covariance matrix for Malagasy skinks from mvMORPH analyses for the limbs only trait subset, unrelated eigenvectors model. Trait names are written in the first row and first column. Used to graph Figure 6
- skink-morphometrics-full.csv -- csv file containing skull, limb, and girdle measurements for all skink species in study. Measurements are in mm and were done in VGStudio. See Appendix 1 and Supplemental Figures 1 + 2 for details about which traits were measured.
- skink-radiation.csv -- csv file used to plot Figure 2. The columns are:
- tree: valid scientific name corresponding to name on phylogeny.
- limbs: number of limbs the species has (0, 2, 4)
- digits: number of digits the species has on one side (ranges from 0-10)
- clade: category used to separate out the Australian sphenomorphine and Malagasy scincine radiations from the other skinks for Figure 2. Values are Australia (Australian sphenomorphine radiation), Madagascar (Malagasy scincine radiation), and blank (not part of either radiation in study).
- limbReduction: indicates whether species is limb-reduced or not.
other files:
- ozmadskinktree.phy -- trimmed phylogeny from Title et al. 2024 Science containing the Australian and Malagasy radiations only
- radskinktree.phy -- trimmed phylogeny from Title et al. 2024 Science for Scincidae
- supplemental-covariance-matrix-figures folder contains all scripts and matrices to create Supplemental Figures 6-18
- supplemental-cov-matrices.R -- R script to create Supplemental Figures 6-18, which show the best-fit covariance matrices for each trait subset.
- 37 cov-matrix...csv files -- covariance matrices for the best-fit models for each trait subset. Name of file indicates the corresponding trait subset and the group. Trait names are written in the first row and first column. Used to graph supplemental figures 6-18.
- Supplemental Figures 6-18 -- plots of the covariance matrices for the best-supported model for each trait subset.
Also included in the zipped data folder:
- Appendix1.doc -- descriptions of each trait measured.
- pca_limb_scores -- csv file containing all the principal component (PC) scores for the limb and girdle traits. Each row is a species.
- pca_skull scores -- csv file containing all the principal component (PC) scores for the skull traits. Each row is a species.
- Supplemental-Figures -- .doc file with all supplemental figures
Supplemental Tables:
- supplemental-table-1-specimen-list -- list of all specimens used in study. The columns are:
- TreeName: valid scientific name matching the name on the phylogeny used in study.
- ScanName: scientific name of specimen matching the museum record and the name on CT scan.
- MuseumNumber: specimen number with museum code.
- clade: indicates whether species is from Australia or Madagascar.
- supplemental-table-2-pca-skull-loadings -- PC loadings for skull trait. First column has the name of all 20 measurements in the study. Every other column corresponds to a principal component and contains the loading for that trait.
- supplemental-table-3-pca-limb-loadings -- PC loadings for limb/girdle traits. First column has the name of all 19 measurements in the study. Every other column corresponds to a principal component and contains the loading for that trait.
- supplemental-table-4-mvmorph-skull-results -- Excel file with model results from mvMORPH analyses of skull traits, with different tabs for each trait subset. The columns for each tab include:
- ModelName: lists the name of the model.
- LogLikelihood: log likelihood score of model
- AIC: AIC score of model
- Nparameters: number of parameters of model.
- supplemental-table-5-mvmorph-limbs-results -- Excel file with model results from mvMORPH analyses of limb and girdle traits, with different tabs for each trait subset. The columns for each tab include:
- ModelName: lists the name of the model.
- LogLikelihood: log likelihood score of model
- AIC: AIC score of model
- Nparameters: number of parameters of model.
- supplemental-table-6-mvmorph-limbs-limbed-species-only -- Excel file with model results from mvMORPH, comparing Ctenotus and skinks with PC1 > 0 for limb traits.
- supplemental-table-7-rate-rate-shift-skull-results -- Excel file with model results from the rate and rate shift analyses for skull traits. Sp1 and Sp2 indicate the bounding species for one of the partitions. The columns are:
- Partitions: number of partitions in model. 1 indicates that all species have the same matrix, 2 indicates that there are 2 groups (bounded by species 1 and 2), and 3 indicates 3 groups (Ctenotus, non-Ctenotus Australian skinks, and Malagasy skinks).
- LogLikelihood: reports the log likelihood score of model.
- AIC: reports the AIC score of model.
- NParameters: reports the number of parameters estimated in the model.
- ModelType: indicates whether the best-supported model is a constant model, rate-shift model, or rate-state-shift model.
- supplemental-table-8-rate-rate-shift-limb-results -- Excel file with model results from the rate and rate shift analyses for limb and girdle traits. Sp1 and Sp2 indicate the bounding species for one of the partitions. The columns are:
- Partitions: number of partitions in model. 1 indicates that all species have the same matrix, 2 indicates that there are 2 groups (bounded by species 1 and 2), and 3 indicates 3 groups (Ctenotus, non-Ctenotus Australian skinks, and Malagasy skinks).
- LogLikelihood: reports the log likelihood score of model.
- AIC: reports the AIC score of model.
- NParameters: reports the number of parameters estimated in the model.
- ModelType: indicates whether the best-supported model is a constant model, rate-shift model, or rate-state-shift model
- supplemental-table-9-corhmm-results -- results from corhmm analyses. The columns are:
- ModelName: name of model tested. See manuscript for descriptions of each model.
- RateCategories: specifies whether the model had 1 rate category or 2 rate categories.
- logLike: log likelihood score of model.
- AIC: AIC score of model.
- AICc: corrected AIC score of model.
- deltaAIC: the difference between the AIC score of the model and the best fit model (the model with the lowest AIC score which was the covarion_uneven model).
Code and software
R version 4.3.1 was used to run all analyses and create all figures. Annotations are provided in all scripts to guide data loading, analyses, and figure creation. The following packages are required:
- ape version 5.7.1
- bm version 1.0 (built from GitHub, script contains code for install)
- corhmm version 2.8
- MASS version 7.3.60
- mvMORPH version 1.1.6
- phylo version 1.0 (built from GitHub, script contains code for install)
- phytools version 1.9.16