Data from: Lungless tadpoles breathe fresh air into hypotheses for tetrapod lung loss and trait regain
Data files
Nov 20, 2025 version files 247.42 MB
-
bayestraits_dryad.zip
247.41 MB
-
README.md
3.86 KB
Abstract
The environmental factors associated with adaptive trait loss and the extent to which lost traits can be regained have been subject to much speculation and debate in evolutionary biology. We use tadpole lungs to test if previously proposed environmental factors, such as a stream habitat, are associated with larval lung loss and whether lungs can be regained following loss. We assembled a dataset of lung presence for the larvae of 529 anurans, finding 28 instances of larval lung loss, and developed a methodological framework to test the evolutionary associations between lung loss, general habitat type, terrestriality, and stream specialization, finding strong support for the final two factors in adaptive lung loss. The likelihood of regain is thought to depend on whether developmental pathways are preserved over time; accordingly, we predicted larval lung loss to be highly reversible. And yet, we found that larval lungs were never regained, despite lungless tadpoles evolving to live in habitats that favor lung use and all lungless tadpoles developing into lunged frogs. Traditional explanations of irreversibility do not easily explain why stage-specific trait loss would be irreversible, prompting us to examine alternative explanations for natural patterns of irreversibility observed across the tree of life.
https://doi.org/10.5061/dryad.3tx95x6sk
Description of the data and file structure
This data was collected for phylogenetic analysis of lung and habitat evolution in frog tadpoles. Missing data are represented as NAs.
Files and variables
File: Supp_table_1.csv
Description: Data table used in all analyses
Variables
- sort: numeric identifier for sorting data by family
- Family: Taxonomic Family
- Subfamily: Taxonomic subfamily when relevant
- Genus:
- Taxa:
- lung data source: Note on where lung data originated (dissection, literature, etc.)
- specimen numbers: Museum specimen identifier
- Tip in Portik et al. 2023: If different name is used in that tree, that alternative name is provided here.
- phylogeny notes: Notes for how and why alternative names were used.
- lung: binary lung status (0 = lungless, 1 = lunged)
- guild: Ecomorphological group
- ecology: binary aquatic habitat status (0 = lotic, 1 = lentic)
- terrestrial: binary overall habitat status (0 = aquatic, 1 = terrestrial or semiterrestrial)
- Spec_lotic: binary specialized lotic status (1 = lotic suctorial/gastromyzophorous, 0 = others)
- ecology_source: Citation for habitat information
- Lung_source: Citation for source of lung data if appropriate
File: lung_loss_git.zip
Description: R environment with scripts for processing, analyzing, and visualizing all data. This includes preparing raw data and trees for BayesTraits and then full statistical analysis, including both direct phylogenetic analysis via phylogenetic generalized linear modeling (phylo_glm) and markov model estimation (corHMM) and analysis of BayesTraits outputs. This file also includes exact code to reproduce all figures appearing in the paper and supplement under the /figures/ directory, as well as several functions used throughout.
File: bayestraits_dryad.zip
Description: Windows directories with BayesTraits (BT) datasets, command files, posterior outputs, and shell scripts for processing raw BT output files into .csv files that are easily loaded into an R environment.
File structure: One main directory with a series of subdirectories including different aspects of two sets of BT analyses. One is the packaged four-state analysis, which uses the "discrete" mode of BT and takes two binary variables (lunged/lungless and lotic/lentic). The other is the eight-state analysis, which uses the "multistate" mode in BT to build a custom 8x8 Q matrix (see supplemental methods).
- Data: trait data files formatted for BT and created in lungloss_git/scripts/set_up/BT_data_setup.R script in the lungloss_git folder.
- Trees: Nexus files formatted for BT and created in the set_up.R script in lungloss_git/scripts/set_up/Initial_set_up. These files include a translation block and then a series of phylogenetic trees from the posterior distribution of Portik et al. (2023) (see supplemental materials for full citation).
- Command_files: Series of BT command files that include all instructions for creating different evolutionary models. See main text for comprehensive list of different model tests.
- Outputs: BT analysis outputs, including a file for raw outputs and processed posteriors and stepping stones (log marginal likelihoods). outputs/shells includes shell scripts and use instructions that can be used to transform raw BT output files into files that are easily read by R and similar programs.
Code/software
R Studio, BayesTraits
Access information
Other publicly accessible locations of the data:
This dataset includes a few different sources. Table 1 includes tadpole lung presence/absence data from a variety of sources listed in the table, including dissection, microCT scanning, and literature searches. Table 1 also includes tadpole habitat data from literature searches (citations embedded within the table). We also include BayesTraits command files and posterior outputs.