Plant adaptation to drought facilitates major ecological transitions, and will likely play a vital role under looming climate change. Mycorrhizas, i.e. strategic associations between plant roots and soil-borne symbiotic fungi, can exert strong influence on the tolerance to drought of extant plants. Here, I show how mycorrhizal strategy and drought adaptation have been shaping one another throughout the course of plant evolution. To characterize the evolutions of both plant characters, I applied a phylogenetic comparative method using data of 1,638 extant species globally distributed. The detected correlated evolution unveiled gains and losses of drought tolerance occurring at faster rates in lineages with ecto- or ericoid mycorrhizas, which were on average about 15 and 300 times faster than in lineages with the arbuscular mycorrhizal and naked root (non-mycorrhizal alone or with facultatively arbuscular mycorrhizal) strategy, respectively. My study suggests that mycorrhizas can play a key facilitator role in the evolutionary processes of plant adaptation to critical changes in water availability across global climates.

This dataset contains the Supplementary Data 1, Supplementary Data 2, Supplementary Data 3, Supplementary Data 4, and Supplementary Note 1 of the article (https://doi.org/10.1038/s42003-023-04722-4).

Supplementary Data 1: includes the files of the phylogenetic trees (in RDS format, which can be open in R) and data frames (in csv format) containing data on plant mycorrhizal strategy and drought adaptation assignments according to dataset v1 to v6. These dataset versions were assembled as described in the Methods of the article. These files and data frames were uploaded as supplemental information to Zenodo and can be found at https://doi.org/10.5281/zenodo.7670985.

Supplementary Data 2: provides a data frame with 112,276 geographical occurrences (latitude and longitude) for 1,066 of the species included in the analysis. These geographical occurrences were originally obtained from a Global Biodiversity Information Facility (occurrence data download https://doi.org/10.15468/dl.5mab9f) and were processed as described in the Methods of the article. The occurrences were generalized here by reducing the number of decimals to conceal the exact geographical locations of endangered and near threatened species, while preserving the integrity of the overall geographic distribution. This data frame was uploaded as supplemental information to Zenodo and can be found at https://doi.org/10.5281/zenodo.7670985.

Supplementary Data 3: provides the output files of the corHMM function resulting from the analysis. These files are deposited inside folders based on the dataset version analyzed, and the folders are named ‘Model output files based on dataset v1’ to ‘Model output files based on dataset v6’. Inside each folder, the output files are named according to their respective models as follow: ‘mode of evolution’ (+ ‘plant character’ in the case of independent evolution) + ‘model structure’ + ‘number of hidden rate categories’ + ‘_’ + ‘number of the replicated start’. For example, ‘depARD1_1.RDS’ is the output file of the model with dependent (dep) mode of evolution, ARD model structure (for details see Methods in the article), hidden rates with a single (1) category, and is the replicated start number 1. In another example, the ‘indepMycSYM3_4.RDS’ is the output file of the model with independent (indep) mode of evolution for the plant character ‘mycorrhizal strategy’ (Myc), SYM model structure, hidden rates with three (3) categories, and is the replicated start number 4. In another example, ‘indepDroER2_3.RDS’ is the output files of the model with independent mode of evolution for the plant character ‘adaptation to drought’ (Dro), ER model structure, hidden rates with two (2) categories, and is the replicated start number 3. All output files are saved in RDS format and can be opened in R with the corHMM package pre-loaded. The R code is documented in the Supplementary Note 1.

Supplementary Data 4: provides the output files containing the best estimates and their respective 95% confidence intervals (CIs) for the transition rate parameters estimated by the models best fitted to dataset v1 to v6. The best estimates and CIs result from the ComputeCI function of the R package corHMM v2.1, as described in the Methods of the article. The R code is documented in the Supplementary Note 1.

Supplementary Note 1: this note is a R script file containing the codes employed to generate the files of Supplementary Data 3, using the corHMM function of the R package corHMM v2.1 and the files of Supplementary Data 1. In addition, it contains the codes used to run the ComputeCI function of the R package corHMM v2.1 to generate the files in Supplementary Data 4, using the files of Supplementary Data 3 that correspond to the models best fitted to the six dataset versions analyzed. For details, see the Methods of the article. This file was uploaded as software to Zenodo and can be found at https://doi.org/10.5281/zenodo.7665030.