https://doi.org/10.5061/dryad.8w9ghx3xp

Description of the data and file structure

This dataset was generated to evaluate and enhance phylogenetic comparative methods, particularly in handling abrupt evolutionary shifts in trait data. The study aimed to develop and test the Outlier- and Distribution-Guided Correlation (OGC) method, comparing its performance with other phylogenetic methods such as PIC-MM and PGLS under diverse evolutionary scenarios. Simulated phylogenetic trees and trait data were created to reflect varying levels of noise, phylogenetic autocorrelation, and evolutionary shifts. This dataset supports robust statistical analyses and benchmarks for detecting true trait correlations and minimizing false positives in comparative studies.

Files and variables

File: Supplemental_Tables.xlsx

Description: This file includes Supplementary Tables S1–S28, which contain the data used to plot the figures presented in the main text and Supplementary materials.

colnames description:
For Supplementary_Tables.xlsx,
Table S1: ID - the index of the tips, X1 - trait X[1] value, X2 - trait X[2] value, label - labels corresponding to the two subtrees generated by the ancestral branch or labels corresponding to the PIC on shift branch or unshift branch.
Table S2: label - the simulation scenario, Method - phylogenetic comparative methods, -3-7 - Shift magnitude or error term variance
Table S3: Method - phylogenetic comparative methods, 1-7 - the depth of shift location
Table S4: PGLS_TR - PGLS accuracy on the second benchmark(Data-Pattern–Based) PGLS_NR - PGLS accuracy on the first benchmark(Structural-Coupling–Based)
Table S5-S7: Coefficient - the value of slope (or correlation coefficient), p - the p-value of the Coefficient
Table S8 : dataset - the dataset used to analysis, sample_size - the sample size of the dataset
Table S10-S19 : the columns are identical in Table S2 

Table S20: the columns are identical in Table S3

Table S21-S28: the columns are identical in Table S10-S19

measurement units description: the simulated data do not have measurement units, measurement units were occurred in empirical data, Table S5,S6 and S7.

Table S5: PIC_log_GS - 'GB' ;  PIC_log_PS - 'cm'

Table S6: PIC_SO and PIC_FG both are the rate 

Table S7: PIC_BodyMass - 'kg' ;  PIC_PopDensity - 'ind./km²'

File: RawData.zip

Description: The zip file contains the raw data for the simulated phylogenetic analyses, including trait correlation data for 16-species fixed tree, 128-species fixed tree, 128-species random trees and 256-species fixed tree across shift and non-shift scenarios and the results of shifts in different locations.

For RawData.zip - shiftlevel/ - it contains two sub-directories, Dualtrait_shifts and Singletrait_shifts, corresponding to the results of Sensitivity of Methods to the Phylogenetic Location of Evolutionary Shifts two shift scenarios in main text, the X.csv means the results of the shift in depth X.

For RawData.zip - all .csv files,
the columns XXX_est - means regression coefficients of XXX method.
XXX_p - means p value of the regression coefficients of XXX method.
for example,BM_xy_p - means the p value of PGLS(BM) model with regression formula "X1_X2",and BM_yx_p _{- means the p value of PGLS(BM) model with regression formula "X2}X1", PIC_L1_xy_p - means the p value of PIC-L1 with regression formula "X1~X2".
X1_lambda: means X1 phylogenetic signal lambda, X2_lambda: means X2 phylogenetic signal lambda, X1_K: means X1 phylogenetic signal K, X2_K: means X2 phylogenetic signal K
change_pearson_est: means Pearson correlation coefficient of traits change among the branches.change_pearson_p : means the p value of the Pearson correlation coefficient of traits change among the branches.change_spearman_est: means Spearman correlation coefficient of traits change among the branches.change_spearman_p: means p value of Spearman correlation coefficient of traits change among the branches.change_final_est: means the correlation coefficient of traits change among the branches with the selection of Pearson or Spearman.change_final_p: means the p value of traits change among the branches with the selection of Pearson or Spearman. 

Some variables pertain to regression models, such as AIC valuesOUrandom_xy_aic, likelihood valuesOUrandom_xy_loglik..., or some intermediate variables X1_lambda_p,...  These variables were not used in the final analysis, so their definitions are not provided.

Code/software

The scripts provided in this dataset were used to generate the results presented in the main text. These scripts cover all major steps of data processing and analysis, including phylogenetic tree construction, trait data simulation, outlier detection, and regression model fitting. All scripts are bundled in Code.zip. Below is a detailed description of the scripts, the required software, and the workflow:

Software Requirements:

• R: Version 4.1.3.
• R Packages: The following R packages need to be installed and loaded:
  • ape (for trait simulation, PIC calculation and Corphylo model fitting).
  • phytools (for phylogenetic tree construction and manipulation).
  • phylolm (for PGLS regression).
  • MCMCglmm (for Bayesian phylogenetic modeling).
  • ROBRT (for robust phylogenetic regression).

Scripts Overview:

• Phylogenetic Tree Construction:

  • 01Simulation_BalancedTree.R: Generates balanced tree topology.

  • 01Simulation_RandomTree.R: Creates random tree topologies.

    Traits Simulation:

  • 02Simulation_BM1&BM1+BM2.R: Simulates traits of "BM1&BM1+BM2" scenario.

  • 02Simulation_BM&BM+Norm.R: Simulates traits of "BM&BM+Norm" scenario.

  • 02Simulation_Norm&Norm+BM.R: Simulates traits of "Norm&Norm+BM" scenario.

  • 02Simulation_ShiftsinBothTraits_fixedlocation.R: Simulates traits of "shifts in both traits" scenario in root branch.

  • 02Simulation_ShiftsinOneTrait_fixedlocation.R: Simulates traits of "shifts only in one trait" scenario in root branch.

  • 02Simulation_ShiftsinBothTraits_randomlocation.R: Simulates traits of "shifts in both traits" scenario in random branch.

  • 02Simulation_ShiftsinOneTrait_randomlocation.R: Simulates traits of "shifts only in one trait" scenario in random branch.

• Outlier Detection:
  • CalculateIQR.R: Detects outliers in trait data using the IQR method.
• Regression Model Fitting:
  • 03Analysis_01_PIC.R: Implements Phylogenetic Independent Contrasts (PIC) for trait correlation analysis.
  • 03Analysis_02_PGLS_RPR.R: Performs Phylogenetic Generalized Least Squares (PGLS) regression and robust regression.
  • 03Analysis_03_Corphylo.R: Applies the Corphylo method for trait correlation analysis.
  • 03Analysis_04_MRPMM.R: Fits models using MRPMM.

Workflow:

• Tree Construction and Trait Simulation:
  • Run 01Simulation_BalancedTree.R and 01Simulation_RandomTree.R to generate phylogenetic trees and associated trait data.
  • Run 02Simulation_BM1&BM1+BM2.R , 02Simulation_BM&BM+Norm.R and 02Simulation_Norm&Norm+BM.R for traits simulation without jump
  • Run 02Simulation_ShiftsinBothTraits.R and 02Simulation_ShiftsinOneTrait.R for traits simulation with jumps.
• Outlier Detection:
  • Use CalculateIQR.R to identify outliers in trait data. Results are saved as CSV files with labeled outlier points.
• Regression Analysis:
  • Execute 03Analysis_01_PIC.R, 03Analysis_02_PGLS_RPR.R, 03Analysis_03_Corphylo.R, and 03Analysis_04_MRPMM.R sequentially to fit regression models.