Notes on Reproducibility:

Dryad provides an immutable repository for data. However, due to limitations in the formatting, the GitHub repository containing all scripts, input data, output data, and plots cannot be loaded directly into Dryad. Downloading the data and scripts in individual files from different sources makes reproduction difficult for the end user. To mitigate this, we included the entire GitHub repository in the associated Zenodo directory, which is linked to this Dryad DOI. Using the excellent R packages here and pacman, we have set up the repository in such a way that it can be downloaded and immediately run on any machine (once the associated R packages are installed). The GitHub repo README file contains detailed information for running each script used in this project, from the initial data wrangling to final tables and plots, and includes all intermediate files such as outputs that are used as inputs in different scripts. This Dryad repository only contains the raw data files and the intermediate output files.

The GitHub repo can also be found on my GitHub page

Data & File Overview

• The data directory contains all of the raw data used in the initial models and the data generated from the model outputs. The data on different trait types are split up into different directories. Keep reading through and this will make sense. The primary sources for the different data types are stored separately because they are called separately by the models. We also store results in these directories because every output is called later as an input, either for additional analyses, and predictions, or as input data for a plot.

Data List

1. All_Traits.csv
   • This is the big spreadsheet that contains all of the estimate, variance, nearest neighbor, and Standardized Effect Size (SES) values for each of the individual and multivariate traits, minus locomotion and phylogenetic diversity. This spreadsheet builds upon itself as you run through the repository. Each community has 4000 draws of each of the 66 trait values. All variables are scaled to a mean of 0 and a standard deviation of 1 prior to fitting any models, and the means and variances are all on the same scale. The SES scores are always reported as values centered on 0 with an sd of 1. All nearest neighbor distances are reported as Euclidean distances between n-axes of scaled variables.
     com = community , .draw = Bayesian draw number, vcCsize = Cranium centroid size, cv_wei = cranium variance weighted, vdCsize = dentary centroid size, dv_wei = dentary variance weighted, vHB = head-body length variance, vTail = tail-length variance, vHF = hind-foot variance, vEar = ear variance, vMass = mass varaince, ext_scaled = External measurement variance, vN15 = Nitrogen15 varaince, vC13 = Carbon 13 variance, sk_var = skull varaince (cranium + dentary), iso_var = isospace variance, morpho_var = all morphological measurement variance, vBsz = body size varaince, ses_vHB = SES of vHB, ses_vHF = SES of vHF, ses_vTail = SES of vTail, ses_vEar = SES of vEar, ses_vMass = SES of vMass, ses_vBsz = SES of vBsz, ses_ext = SES of ext_scaled, ses_cv_wei = SES of cv_wei, ses_vcCsize = SES of vcCsize, ses_dv_wei = SES of dv_wei, ses_vdCsize = SES of vdCsize, ses_vsk = SES of sk_var, ses_morpho = SES of morpho_var, ses_vC13 = SES of vC13, ses_vN15 = SES of vN15, ses_vIso = SES of iso_var, d_nn = Dentary Nearest Neighbor, dsize_nn = Dentary centroid size Nearest Neighbor, sk_nn = Skull shape Nearest Neighbor, morpho_nn = all morphological measurements Nearest Neighbor, Bsz_nn = body size Nearest Neighbor, nnses_dsize = SES of dsize_nn, nnses_d = SES d_nn, nnses_sk = SES of sk_nn, nnses_Bsz = SES of BSZ_nn, nnses_morpho = SES of morpho_nn, N15_nn = Nitrogen 15 Nearest Neighbor, C13_nn = Carbon 13 Nearest Neighbor, iso_nn = Isospace Nearest Neighbor, nnses_C13 = SES of C13_nn, nnses_N15 = SES of N15_nn, nnses_iso = SES of iso_nn, c_nn = Cranium Nearest Neighbor, csize_nn = Cranium centroid size Nearest Neighbor, Ear_nn = Ear length Nearest Neighbor, HB_nn = head-body length Nearest Neighbor, Mass_nn = Mass Nearest Neighbor, Tail_nn = Tail Nearest Neighbor, HF_nn = hind foot length Nearest Neighbor, ext_nn = all external measurements Nearest Neighbor, nnses_c = SES of c_NN, nnses_csize = SES of cizse_nn, nnses_Tail = SES of Tail_nn, nnses_HF = SES of HF_nn, nnses_Ear = SES of Ear_nn, nnses_HB = SES of HB_nn, nnses_Mass = SES of Mass_nn, nnses_ext = SES of ext_nn
2. Regression_Results.csv
   • Results from the Bayesian regression analyses run on each trait. These models used the number of species as the predictor and the trait values from the SES values of each trait above. Beta = mean effect size, b_lower = lower 89% of beta distribution, b_upper = upper 89% of beta distribution, Intercept = mean intercept, I_lower = lower 89% of intercept distribution, I_upper = upper 89% of intercept distribution, sigma = mean sigma from the model, s_lower = lower 89% of sigma distribution, s_upper = upper 89% of sigma distribution, R2sm = mean Bayesian R-squared value, R2sm_lower = lower 89% of Bayesian R-squared distribution, R2sm_upper = upper 89% of Bayesian R-squared distribution, var = variable, name = Description of the variable.

Directory List

Cranial_Data

Input Data

1. Cranial_PCA_Data_36axes.csv
   • Data frame containing the PCA scores of the first 36 axes of the 3D cranial landmark data. These are the primary inputs into most models. Individuals are rows. The first column is an abbreviation. Genus = genus, Species = species, Clade = clade (see text), SubClade and SubClade_Frat = two subclade designations not used in this study, PC1:PC36 are the PC's of the Procrustes coordinates of the landmarks, Csize = centroid size. Rows are individuals. There are no relevant units for these values.

Output Data

1. Cranial_36axes_Fitted.csv
   • This is the output file from the Bayesian model that estimates a value for each landmark for each species. This contains 4000 estimates for each of the 36 landmarks for each of the 38 species. All values are unitless, scaled values with a mean of 0 and a standard deviation of 1.
2. Community_Cranial_Variance_Weightedcsv
   • This spreadsheet has the variance estimations for the cranial shape and size for each community. There are 4000 estimations for each of the 9 communities.

Dentary_Data

Input

1. Dentary_PCA_Data_20axes.csv
   • Data frame containing the PCA scores of the first 20 axes of the 3D dentary landmark data. These are the primary inputs into most models. Genus = genus, Species = species, Clade = clade (see text), SubClade and SubClade_Frat = two subclade designations not used in this study, PC1:PC20 are the PC's of the Procrustes coordinates of the landmarks, Csize = centroid size. Rows are individuals. There are no relevant units for these values.

Output Data

1. Dentary_20axes_Fitted.csv
   • This is the output file from the Bayesian model that estimates a value for each dentary landmark for each species. This contains 4000 estimates for each of the 20 landmarks for each of the 36 species. All values are unitless, scaled values with a mean of 0 and a standard deviation of 1.
2. Community_Dentary_Variance_Weighted.csv
   • This spreadsheet has the variance estimations for the dentary shape and size for each community. There are 4000 estimations for each of the 9 communities.

External_Measurement_Data

Input

1. Measurement_Data_Museum.csv
   • Data frame containing the external measurements (total length, tail length, hind-foot length, ear length, mass) for 631 specimens. Museum is the museum where the specimen is housed. See main text for abbreviations. Catalog is the specimen catalog number. Species is the species. Total = total length in millimeters, Tail = tail length in millimeters, HF = hind-foot length in millimeters, Ear = ear length in millimeters, Mass = mass in grams. These are the primary inputs into most models. Individuals are rows, and measurements are columns. Locomotor states are also included, but not called from this spreadsheet.

Output Data

1. External_Fitted.csv
   • This is the output file from the Bayesian model that estimates a value for each of the external measurements for each species. This contains 4000 estimates for each of the 5 measurements for each of the 40 species. All values are unitless, scaled values with a mean of 0 and a standard deviation of 1.
2. Community_External_Variance_Fitted.csv
   • This spreadsheet has the variance estimations for the 5 external traits for each community. There are 4000 estimations for each trait for each of the 9 communities.

Isotope_Data

Input

1. Isotope_Data.csv
   • Data frame containing the isotopic signatures for each specimen reported in Corrected 13C and Corrected 15N values. The museum and catalog number are reported for each specimen. These are the primary inputs into most models. In addition to the Corrected_13C and Corrected_15N, the values used in this paper, the hair sample mass (in micrograms) and the uncorrected carbon and nitrogen masses in micrograms)and percents are also included (though I don't use them).

Output Data

1. Isotope_Group_Level_Fitted.csv
   • This is the output file from the Bayesian model that estimates a value for each isotopic measure for each species. This contains 4000 estimates for Corrected_13C and Corrected_15N for each of the 33 species. All values are unitless, scaled values with a mean of 0 and a standard deviation of 1.
2. Isotope_Group_Level_Variance.csv
   • This spreadsheet has the variance estimations for the 2 isotopic traits for each community. There are 4000 estimations for each trait for each of the 6 communities (only 6 communities have isotopes).

Locomotion_Data

Input

1. Locomotion_List.csv
   • Data frame containing the locomotor mode for each species. 1 = Arboreal, 2 = General, 3 = Terrestrial, 4 = Amphibious.
2. Locomotion_Dummy
   • data frame of the locomotor data from Locomotion_List, but in one-shot coding.

Output Data

1. All_Locomotor_Var.csv
   • This is the output for each of the analyses run on the locomotor data. vloc = locomotor variance, com = community, ses_loc is the SES value of the locomotor variance, nn_rda is the nearest neighbor distance for each community, nnses_loc is the SES nearest neighbor value, and nsp is the number of species in each community.

Phylogenetic_Diversity

Input

1. 397_MCC.tre
   • Time-calibrated phylogenetic hypothesis of Murinae, used to trim down for a Sulawesi-only tree.
2. Clean_Sulawesi_Tree.tre
   • As the title says, it's the trimmed tree used in the analyses

Output Data

1. PD_Results.csv
   • Data frame containing the phylogenetic diversity estimates for each community. ntaxa is the number of species in each community, pd.obs is the phylogenetic diversity for each community, pd_ses is the SES PD value for each community, and com is community.
2. PD_Results_no_Hae.csv
   • Data frame containing the phylogenetic diversity estimates for each community without the Haeromys minahassae species. See text for details. ntaxa is the number of species in each community, pd.obs is the phylogenetic diversity for each community, pd_ses is the SES PD value for each community, and com is the community.