Contrasting levels of β-diversity and underlying phylogenetic trends indicate different paths to chemical diversity in highland and lowland willow species

Diverse specialized metabolites contributed to the success of vascular plants in colonizing most terrestrial habitats. Understanding how distinct aspects of chemical diversity arise through heterogeneous environmental pressures can help us understand the effects of abiotic and biotic stress on plant evolution and community assembly. Here we examine highland and lowland willow species within a phylogenetic framework to test for trends in their chemical α-diversity (richness) and β-diversity (variation among species sympatric in elevation). We show that differences in chemistry among willows growing at different elevations occur mainly through shifts in chemical β-diversity and convergence or divergence among species sharing their elevation level. We also detect contrasting phylogenetic trends in concentration and α-diversity of metabolites in highland and lowland willow species. The resulting elevational patterns contribute to the chemical diversity of Salix and suggest that variable selective pressure across ecological gradients may, more generally, underpin complex changes in plant chemistry.

This repository includes the code for statistical analyses performer in R (multivariate analyses were performed in CANOCO 5) and data derived from the primary data files available from https://doi.org/10.5281/zenodo.7825538. It contains the following files:

1. READ_ME.txt – this file includes the background information and metadata for all the other files included in this submission.
2. Willow MS Analysis_R.R – this file includes the R code. The coded uses data from the files specified below.
3. Trait_data.csv –This file contains species means of the trait data.
4. Trait_data_JL.csv – This file contains species means of trait data as mentioned above but it includes Populus tremula as opposed to Trait_data.csv. This file is used for disparity through time (DTT) plots and LIPA analyses.
5. Trait_highland_data.csv – This file contains species means of trait data as mentioned above for highland willow species only.
6. Trait_lowland_data.csv – This file contains species means of trait data as mentioned above for lowland willow species only.
7. Ultrametric_tree.tre – This file contains the pruned ultrametric (excluding Populus tremula) tree used in the analyses. Willow species are marked with three letter codes.
8. Ultrametric_tree_tre1.tre – This file contains the pruned ultrametric (including Populus tremula) tree used in the analyses. Willow species are marked with three letter codes.
9. Ultrametric_tree_highland.tre – This file contains the pruned ultrametric (including Populus tremula) tree used in the analyses for highland willow species. Willow species are marked with three letter codes.
10. Ultrametric_tree_lowland.tre – This file contains the pruned ultrametric (including Populus tremula) tree used in the analyses for lowland willow species. Willow species are marked with three letter codes.
11. Not_Ultrametric_tree.tre – This file contains the pruned non-ultrametric (including Populus tremula) tree used in the analyses. Willow species are marked with three letter codes.
12. Not_UIltrametric_tree_rmTRE.tre – This file contains the pruned ultrametric (excluding Populus tremula) tree used in the analyses. Willow species are marked with three letter codes.
13. Not_Ultrametric_highland_tree_rmTRE.tre – This file contains the pruned non-ultrametric (excluding Populus tremula) tree used in the analyses for highland willow species. Willow species are marked with three letter codes.
14. Not_Ultrametric_lowland_tree_rmTRE.tre – This file contains the pruned non-ultrametric (excluding Populus tremula) tree used in the analyses for lowland willow species. Willow species are marked with three letter codes.
15. Significance.csv – This file contains p-values for uncorrected PGLS models and linear models using Abouheif´s distances.
16. AbuEffSize.csv – This file contains the R-squared values from the Linear Models using Abouheif’s distances used to calculate effect sizes for the whole dataset (R2.all), highland willows (R2.high), and lowland willows (R2.low). We only calculated effect sizes for highland and lowland willow datasets if traits for the whole dataset tested significant (Concentration of flavonoids, oxidative activity, Protein precipitation, simple richness of flavonoids, and flavonoid uniqueness.
17. IQ_distances_all.csv – This file includes mean interquartile distances among lowland vs. highland willow species on ordination axes as extracted from pPCA performed in CANOCO 5 when considering the whole metabolome.
18. IQ_distances_flavonoids.csv – This file includes mean interquartile distances among lowland vs. highland willow species on ordination axes as extracted from pPCA performed in CANOCO 5 when considering flavonoids.
19. IQ_distances_flavonoids.csv – This file includes mean interquartile distances among lowland vs. highland willow species on ordination axes as extracted from pPCA performed in CANOCO 5 when considering salicinoids.
20. LIPA_Significance.csv   – This file contains uncorrected p-values obtained with lipaMoran function. Columns are individual willow species marked with letter codes.