R Under development (unstable) (2024-02-27 r85995 ucrt) -- "Unsuffered Consequences" Copyright (C) 2024 The R Foundation for Statistical Computing Platform: x86_64-w64-mingw32/x64 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > # This file is part of the standard setup for testthat. > # It is recommended that you do not modify it. > # > # Where should you do additional test configuration? > # Learn more about the roles of various files in: > # * https://r-pkgs.org/tests.html > # * https://testthat.r-lib.org/reference/test_package.html#special-files > > library(testthat) > library(DImodels) > > test_check("DImodels") Fitted model: Evenness 'E' DImodel Theta estimate: 0.8357 Fitted model: Evenness 'E' DImodel Theta estimate: 0.7603 Fitted model: Evenness 'E' DImodel Theta estimate: 0.7603 Fitted model: Evenness 'E' DImodel Fitted model: Evenness 'E' DImodel Fitted model: Evenness 'E' DImodel Fitted model: Evenness 'E' DImodel Theta estimate: 0.8045 Fitted model: Evenness 'E' DImodel Fitted model: Structural 'STR' DImodel Fitted model: Species identity 'ID' DImodel Fitted model: Average interactions 'AV' DImodel Fitted model: Evenness 'E' DImodel Fitted model: Functional group effects 'FG' DImodel Fitted model: Additive species contributions to interactions 'ADD' DImodel Fitted model: Separate pairwise interactions 'FULL' DImodel Fitted model: Structural 'STR' DImodel Fitted model: Species identity 'ID' DImodel Fitted model: Average interactions 'AV' DImodel Fitted model: Evenness 'E' DImodel Fitted model: Functional group effects 'FG' DImodel Fitted model: Additive species contributions to interactions 'ADD' DImodel Fitted model: Separate pairwise interactions 'FULL' DImodel Fitted model: Structural 'STR' DImodel Fitted model: Species identity 'ID' DImodel Fitted model: Average interactions 'AV' DImodel Fitted model: Evenness 'E' DImodel Fitted model: Functional group effects 'FG' DImodel Fitted model: Additive species contributions to interactions 'ADD' DImodel Fitted model: Separate pairwise interactions 'FULL' DImodel Fitted model: Structural 'STR' DImodel Fitted model: Species identity 'ID' DImodel Fitted model: Average interactions 'AV' DImodel Fitted model: Evenness 'E' DImodel Fitted model: Functional group effects 'FG' DImodel Fitted model: Additive species contributions to interactions 'ADD' DImodel Fitted model: Separate pairwise interactions 'FULL' DImodel Fitted model: Additive species contributions to interactions 'ADD' DImodel Theta estimate: 0.786 Fitted model: Additive species contributions to interactions 'ADD' DImodel Theta estimate: 0.856 Fitted model: Functional group effects 'FG' DImodel Theta estimate: 0.7524 Fitted model: Functional group effects 'FG' DImodel Theta estimate: 0.8299 -------------------------------------------------------------------------------- Investigating richness model and three DI alternatives AIC AICc BIC df 1 333.9517 334.3267 340.6102 2 2 329.5335 329.9085 336.1920 2 3 329.4024 330.0373 338.2804 3 4 277.0934 278.4705 292.6300 5 5 271.7793 273.6460 287.3159 6 Description 1 Model 1: Richness only 2 Model 2: Average interactions 'AV' DImodel with common identity effects and theta = 0.5 3 Model 3: Average interactions 'AV' DImodel with common identity effects and theta estimated 4 Model 4: Average interactions 'AV' DImodel with unique identity effects and theta = 0.5 5 Model 5: Average interactions 'AV' DImodel with unique identity effects and theta estimated -------------------------------------------------------------------------------- Models 1 and 2 are equivalent if all mixtures in the dataset at any given level of richness are equi-proportional. richness_vs_DI is limited in terms of model selection. Only five models are explored. See ?autoDI and ?DI for more options. -------------------------------------------------------------------------------- Fitted model: Average interactions 'AV' DImodel Theta estimate: 0.8357 -------------------------------------------------------------------------------- Investigating richness model and three DI alternatives AIC AICc BIC df 1 335.9073 336.5422 344.7853 3 2 331.4862 332.1211 340.3642 3 3 331.3535 332.3213 342.4511 4 4 278.9786 280.8452 296.7346 6 5 273.6548 276.0955 291.4109 7 Description 1 Model 1: Richness only 2 Model 2: Average interactions 'AV' DImodel with common identity effects and theta = 0.5 3 Model 3: Average interactions 'AV' DImodel with common identity effects and theta estimated 4 Model 4: Average interactions 'AV' DImodel with unique identity effects and theta = 0.5 5 Model 5: Average interactions 'AV' DImodel with unique identity effects and theta estimated -------------------------------------------------------------------------------- Models 1 and 2 are equivalent if all mixtures in the dataset at any given level of richness are equi-proportional. richness_vs_DI is limited in terms of model selection. Only five models are explored. See ?autoDI and ?DI for more options. -------------------------------------------------------------------------------- Fitted model: Average interactions 'AV' DImodel Theta estimate: 0.8357 Fitted model: Functional group effects 'FG' DImodel Fitted model: Custom DI model Fitted model: Average interactions 'AV' DImodel Theta estimate: 0.8357 Fitted model: Custom DI model Fitted model: Functional group effects 'FG' DImodel 'community' is a factor with 25 levels, one for each unique set of proportions. Fitted model: Functional group effects 'FG' DImodel 'community' is a factor with 25 levels, one for each unique set of proportions. Fitted model: Functional group effects 'FG' DImodel Analysis of Deviance Table Model 1: yield ~ 0 + p1_ID + p2_ID + p3_ID + p4_ID + FG_bfg_G_H + FG_wfg_G + FG_wfg_H + nitrogen50 + densityhigh Model 2: yield ~ 0 + p1_ID + p2_ID + p3_ID + p4_ID + FG_bfg_G_H + FG_wfg_G + FG_wfg_H + nitrogen50 + densityhigh + community2 + community3 + community4 + community5 + community6 + community7 + community8 + community9 + community10 + community11 + community12 + community13 + community14 + community15 + community16 + community17 + community18 + community19 Resid. Df Resid. Dev Df Deviance 1 59 136.930 2 41 80.413 18 56.517 Fitted model: Functional group effects 'FG' DImodel Theta estimate: 0.7524 'community' is a factor with 25 levels, one for each unique set of proportions. Fitted model: Functional group effects 'FG' DImodel 'community' is a factor with 25 levels, one for each unique set of proportions. Fitted model: Functional group effects 'FG' DImodel Analysis of Deviance Table Model 1: yield ~ 0 + p1_ID + p2_ID + p3_ID + p4_ID + FG_bfg_G_H + FG_wfg_G + FG_wfg_H + nitrogen50 + densityhigh Model 2: yield ~ 0 + p1_ID + p2_ID + p3_ID + p4_ID + FG_bfg_G_H + FG_wfg_G + FG_wfg_H + nitrogen50 + densityhigh + community2 + community3 + community4 + community5 + community6 + community7 + community8 + community9 + community10 + community11 + community12 + community13 + community14 + community15 + community16 + community17 + community18 + community19 Resid. Df Resid. Dev Df Deviance 1 58 127.083 2 40 80.413 18 46.67 Fitted model: Average interactions 'AV' DImodel Theta estimate: 0.7603 Fitted model: Separate pairwise interactions 'FULL' DImodel Theta estimate: 0.7706 Fitted model: Functional group effects 'FG' DImodel Fitted model: Separate pairwise interactions 'FULL' DImodel Theta estimate: 0.4525 Fitted model: Average interactions 'AV' DImodel Theta estimate: 0.4533 Fitted model: Structural 'STR' DImodel Fitted model: Custom DI model -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the AV model, including all available structures Theta estimate: 0.4533 Selection by AIC Warning: DI Model with the lowest AIC will be selected, even if the difference is very small. Please inspect other models to see differences in AIC. AIC DI_Model treat theta 1 225.4313 AV none FALSE 2 217.4542 AV none TRUE Description 1 Average interactions 'AV' DImodel 2 Average interactions 'AV' DImodel, estimating theta The test concludes that theta is significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions Selection by AIC Warning: DI Model with the lowest AIC will be selected, even if the difference is very small. Please inspect other models to see differences in AIC. AIC DI_Model treat theta 1 360.9133 STR none FALSE 2 365.5893 ID none FALSE 3 217.4542 AV none TRUE 4 220.3987 FG none TRUE 5 221.0360 ADD none TRUE 6 224.3384 FULL none TRUE Description 1 Structural 'STR' DImodel 2 Species identity 'ID' DImodel 3 Average interactions 'AV' DImodel, estimating theta 4 Functional group effects 'FG' DImodel, estimating theta 5 Additive species contributions to interactions 'ADD' DImodel, estimating theta 6 Separate pairwise interactions 'FULL' DImodel, estimating theta Selected model: Average interactions 'AV' DImodel, estimating theta -------------------------------------------------------------------------------- Step 3: No investigation of treatment effect included, since no treatment was specified (argument 'treat' omitted) -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model 'community' is a factor with 15 levels, one for each unique set of proportions. model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) DI Model 1 Selected 54 104.3104 1.9317 DI Model 2 Reference 45 98.0794 2.1795 9 6.231 0.3176 0.965 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Sequential analysis: Investigating only non-diversity experimental design structures model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) 1 Intercept only 59 1346.175 22.8165 2 treat 56 1296.957 23.1599 3 49.2178 0.7084 0.5511 -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the AV model, including all available structures Theta estimate: 0.4533 Selection by AIC Warning: DI Model with the lowest AIC will be selected, even if the difference is very small. Please inspect other models to see differences in AIC. AIC DI_Model treat theta 1 200.8260 AV 'block' FALSE 2 185.1528 AV 'block' TRUE Description 1 Average interactions 'AV' DImodel with treatment 2 Average interactions 'AV' DImodel with treatment, estimating theta The test concludes that theta is significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions Selection by AIC Warning: DI Model with the lowest AIC will be selected, even if the difference is very small. Please inspect other models to see differences in AIC. AIC DI_Model treat theta 1 364.6785 STR 'block' FALSE 2 369.3037 ID 'block' FALSE 3 185.1528 AV 'block' TRUE 4 187.1385 FG 'block' TRUE 5 186.4887 ADD 'block' TRUE 6 189.1097 FULL 'block' TRUE Description 1 Structural 'STR' DImodel with treatment 2 Species identity 'ID' DImodel with treatment 3 Average interactions 'AV' DImodel with treatment, estimating theta 4 Functional group effects 'FG' DImodel with treatment, estimating theta 5 Additive species contributions to interactions 'ADD' DImodel with treatment, estimating theta 6 Separate pairwise interactions 'FULL' DImodel with treatment, estimating theta Selected model: Average interactions 'AV' DImodel with treatment, estimating theta -------------------------------------------------------------------------------- Step 3: Investigating the treatment effect Selection by AIC Warning: DI Model with the lowest AIC will be selected, even if the difference is very small. Please inspect other models to see differences in AIC. AIC DI_Model treat theta 1 217.4542 AV none FALSE 2 185.1528 AV 'block' FALSE Description 1 Average interactions 'AV' DImodel 2 Average interactions 'AV' DImodel with treatment Selected model: Average interactions 'AV' DImodel with treatment -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model 'community' is a factor with 15 levels, one for each unique set of proportions. model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) DI Model 1 Selected 51 55.0925 1.0802 DI Model 2 Reference 42 48.8616 1.1634 9 6.231 0.5951 0.7936 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Sequential analysis: Investigating only non-diversity experimental design structures model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) 1 Intercept only 67 680.2884 10.1536 2 density 66 679.9658 10.3025 1 0.3226 0.0324 0.8578 3 density + treat 65 647.9784 9.9689 1 31.9874 3.2087 0.0779 -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the AV model, including all available structures Theta estimate: 0.7603 Selection using F tests Description DI Model 1 Average interactions 'AV' DImodel with treatment DI Model 2 Average interactions 'AV' DImodel with treatment, estimating theta DI_model treat estimate_theta Resid. Df Resid. SSq Resid. MSq DI Model 1 AV 'nitrogen' FALSE 61 174.2937 2.8573 DI Model 2 AV 'nitrogen' TRUE 60 164.8882 2.7481 Df SSq F Pr(>F) DI Model 1 DI Model 2 1 9.4055 3.4225 0.0692 The test concludes that theta is not significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions All models include density Selection using F tests Description DI Model 1 Structural 'STR' DImodel with treatment DI Model 2 Species identity 'ID' DImodel with treatment DI Model 3 Average interactions 'AV' DImodel with treatment DI Model 4 Additive species contributions to interactions 'ADD' DImodel with treatment DI Model 5 Separate pairwise interactions 'FULL' DImodel with treatment DI_model treat estimate_theta Resid. Df Resid. SSq Resid. MSq DI Model 1 STR 'nitrogen' FALSE 65 647.9784 9.9689 DI Model 2 ID 'nitrogen' FALSE 62 374.8340 6.0457 DI Model 3 AV 'nitrogen' FALSE 61 174.2937 2.8573 DI Model 4 ADD 'nitrogen' FALSE 58 156.2477 2.6939 DI Model 5 FULL 'nitrogen' FALSE 56 124.1971 2.2178 Df SSq F Pr(>F) DI Model 1 DI Model 2 3 273.1444 41.0533 <0.0001 DI Model 3 1 200.5404 90.4229 <0.0001 DI Model 4 3 18.046 2.7123 0.0535 DI Model 5 2 32.0506 7.2258 0.0016 Functional groups (argument 'FG') were not specified, and therefore not investigated. Selected model: Separate pairwise interactions 'FULL' DImodel with treatment -------------------------------------------------------------------------------- Step 3: Investigating the treatment effect Selection using F tests Description DI Model 1 Separate pairwise interactions 'FULL' DImodel DI Model 2 Separate pairwise interactions 'FULL' DImodel with treatment DI_model treat estimate_theta Resid. Df Resid. SSq Resid. MSq DI Model 1 FULL none FALSE 57 131.1550 2.3010 DI Model 2 FULL 'nitrogen' FALSE 56 124.1971 2.2178 Df SSq F Pr(>F) DI Model 1 DI Model 2 1 6.9579 3.1373 0.082 Selected model: Separate pairwise interactions 'FULL' DImodel -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model 'community' is a factor with 25 levels, one for each unique set of proportions. model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) DI Model 1 Selected 57 131.1550 2.3010 DI Model 2 Reference 42 84.6499 2.0155 15 46.5052 1.5383 0.1353 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Sequential analysis: Investigating only non-diversity experimental design structures model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) 1 Intercept only 67 680.2884 10.1536 2 block 66 679.9658 10.3025 1 0.3226 0.0324 0.8578 3 block + treat 65 647.9784 9.9689 1 31.9874 3.2087 0.0779 -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the AV model, including all available structures Theta estimate: 0.7603 Selection by BICc Warning: DI Model with the lowest BICc will be selected, even if the difference is very small. Please inspect other models to see differences in BICc. BICc DI_Model treat theta 1 295.8848 AV 'nitrogen' FALSE 2 297.7304 AV 'nitrogen' TRUE Description 1 Average interactions 'AV' DImodel with treatment 2 Average interactions 'AV' DImodel with treatment, estimating theta The test concludes that theta is not significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions All models include block Selection by BICc Warning: DI Model with the lowest BICc will be selected, even if the difference is very small. Please inspect other models to see differences in BICc. BICc DI_Model treat theta 1 364.4890 STR 'nitrogen' FALSE 2 342.5247 ID 'nitrogen' FALSE 3 295.8848 AV 'nitrogen' FALSE 4 290.9111 FG 'nitrogen' FALSE 5 305.9078 ADD 'nitrogen' FALSE 6 303.0112 FULL 'nitrogen' FALSE Description 1 Structural 'STR' DImodel with treatment 2 Species identity 'ID' DImodel with treatment 3 Average interactions 'AV' DImodel with treatment 4 Functional group effects 'FG' DImodel with treatment 5 Additive species contributions to interactions 'ADD' DImodel with treatment 6 Separate pairwise interactions 'FULL' DImodel with treatment Selected model: Functional group effects 'FG' DImodel with treatment -------------------------------------------------------------------------------- Step 3: Investigating the treatment effect Selection by BICc Warning: DI Model with the lowest BICc will be selected, even if the difference is very small. Please inspect other models to see differences in BICc. BICc DI_Model treat theta 1 288.4666 FG none FALSE 2 290.9111 FG 'nitrogen' FALSE Description 1 Functional group effects 'FG' DImodel 2 Functional group effects 'FG' DImodel with treatment Selected model: Functional group effects 'FG' DImodel -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model 'community' is a factor with 25 levels, one for each unique set of proportions. model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) DI Model 1 Selected 60 143.8881 2.3981 DI Model 2 Reference 42 84.6499 2.0155 18 59.2382 1.6329 0.0954 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Sequential analysis: Investigating only non-diversity experimental design structures model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) 1 Intercept only 67 680.2884 10.1536 2 block 66 679.9658 10.3025 1 0.3226 0.0324 0.8578 3 block + density 65 647.9784 9.9689 1 31.9874 3.2087 0.0779 -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the AV model, including all available structures Theta estimate: 0.7603 Selection by AICc Warning: DI Model with the lowest AICc will be selected, even if the difference is very small. Please inspect other models to see differences in AICc. AICc DI_Model treat theta 1 275.4202 AV none FALSE 2 274.3108 AV none TRUE Description 1 Average interactions 'AV' DImodel 2 Average interactions 'AV' DImodel, estimating theta The test concludes that theta is significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions All models include block and density Selection by AICc Warning: DI Model with the lowest AICc will be selected, even if the difference is very small. Please inspect other models to see differences in AICc. AICc DI_Model treat theta 1 354.9063 STR none FALSE 2 324.9167 ID none FALSE 3 274.3108 AV none TRUE 4 262.2169 FG none TRUE 5 275.7806 ADD none TRUE 6 264.9828 FULL none TRUE Description 1 Structural 'STR' DImodel 2 Species identity 'ID' DImodel 3 Average interactions 'AV' DImodel, estimating theta 4 Functional group effects 'FG' DImodel, estimating theta 5 Additive species contributions to interactions 'ADD' DImodel, estimating theta 6 Separate pairwise interactions 'FULL' DImodel, estimating theta Selected model: Functional group effects 'FG' DImodel, estimating theta -------------------------------------------------------------------------------- Step 3: No investigation of treatment effect included, since no treatment was specified (argument 'treat' omitted) -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model 'community' is a factor with 25 levels, one for each unique set of proportions. model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) DI Model 1 Selected 58 127.0827 2.1911 DI Model 2 Reference 41 80.4127 1.9613 17 46.67 1.3997 0.1865 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Sequential analysis: Investigating only non-diversity experimental design structures model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) 1 Intercept only 67 680.2884 10.1536 2 treat 66 648.3010 9.8227 1 31.9874 3.2565 0.0757 -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the AV model, including all available structures Theta estimate: 0.7603 Selection by BIC Warning: DI Model with the lowest BIC will be selected, even if the difference is very small. Please inspect other models to see differences in BIC. BIC DI_Model treat theta 1 286.6419 AV 'nitrogen' FALSE 2 287.0963 AV 'nitrogen' TRUE Description 1 Average interactions 'AV' DImodel with treatment 2 Average interactions 'AV' DImodel with treatment, estimating theta The test concludes that theta is not significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions Selection by BIC Warning: DI Model with the lowest BIC will be selected, even if the difference is very small. Please inspect other models to see differences in BIC. BIC DI_Model treat theta 1 358.9638 STR 'nitrogen' FALSE 2 334.4255 ID 'nitrogen' FALSE 3 286.6419 AV 'nitrogen' FALSE 4 278.7087 FG 'nitrogen' FALSE 5 291.8826 ADD 'nitrogen' FALSE 6 284.7467 FULL 'nitrogen' FALSE Description 1 Structural 'STR' DImodel with treatment 2 Species identity 'ID' DImodel with treatment 3 Average interactions 'AV' DImodel with treatment 4 Functional group effects 'FG' DImodel with treatment 5 Additive species contributions to interactions 'ADD' DImodel with treatment 6 Separate pairwise interactions 'FULL' DImodel with treatment Selected model: Functional group effects 'FG' DImodel with treatment -------------------------------------------------------------------------------- Step 3: Investigating the treatment effect Selection by BIC Warning: DI Model with the lowest BIC will be selected, even if the difference is very small. Please inspect other models to see differences in BIC. BIC DI_Model treat theta 1 277.8519 FG none FALSE 2 278.7087 FG 'nitrogen' FALSE Description 1 Functional group effects 'FG' DImodel 2 Functional group effects 'FG' DImodel with treatment Selected model: Functional group effects 'FG' DImodel -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model 'community' is a factor with 25 levels, one for each unique set of proportions. model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) DI Model 1 Selected 61 144.2107 2.3641 DI Model 2 Reference 43 84.9725 1.9761 18 59.2382 1.6654 0.0858 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Sequential analysis: Investigating only non-diversity experimental design structures model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) 1 Intercept only 67 680.2884 10.1536 2 density 66 648.3010 9.8227 1 31.9874 3.2565 0.0757 -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the AV model, including all available structures Theta estimate: 0.7603 Selection by BIC Warning: DI Model with the lowest BIC will be selected, even if the difference is very small. Please inspect other models to see differences in BIC. BIC DI_Model treat theta 1 286.6419 AV none FALSE 2 287.0963 AV none TRUE Description 1 Average interactions 'AV' DImodel 2 Average interactions 'AV' DImodel, estimating theta The test concludes that theta is not significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions All models include density Selection by BIC Warning: DI Model with the lowest BIC will be selected, even if the difference is very small. Please inspect other models to see differences in BIC. BIC DI_Model treat theta 1 358.9638 STR none FALSE 2 334.4255 ID none FALSE 3 286.6419 AV none FALSE 4 278.7087 FG none FALSE 5 291.8826 ADD none FALSE 6 284.7467 FULL none FALSE Description 1 Structural 'STR' DImodel 2 Species identity 'ID' DImodel 3 Average interactions 'AV' DImodel 4 Functional group effects 'FG' DImodel 5 Additive species contributions to interactions 'ADD' DImodel 6 Separate pairwise interactions 'FULL' DImodel Selected model: Functional group effects 'FG' DImodel -------------------------------------------------------------------------------- Step 3: No investigation of treatment effect included, since no treatment was specified (argument 'treat' omitted) -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model 'community' is a factor with 25 levels, one for each unique set of proportions. model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) DI Model 1 Selected 60 137.2527 2.2875 DI Model 2 Reference 42 80.7353 1.9223 18 56.5175 1.6334 0.0952 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Sequential analysis: Investigating only non-diversity experimental design structures model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) 1 Intercept only 67 680.2884 10.1536 2 block 66 679.9658 10.3025 1 0.3226 0.0313 0.8601 -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the AV model, including all available structures Theta estimate: 0.8357 Selection by BIC Warning: DI Model with the lowest BIC will be selected, even if the difference is very small. Please inspect other models to see differences in BIC. BIC DI_Model treat theta 1 289.1779 AV none FALSE 2 291.4109 AV none TRUE Description 1 Average interactions 'AV' DImodel 2 Average interactions 'AV' DImodel, estimating theta The test concludes that theta is not significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions All models include block Selection by BIC Warning: DI Model with the lowest BIC will be selected, even if the difference is very small. Please inspect other models to see differences in BIC. BIC DI_Model treat theta 1 362.2065 STR none FALSE 2 339.9356 ID none FALSE 3 289.1779 AV none FALSE 4 281.9191 FG none FALSE 5 294.7050 ADD none FALSE 6 288.2771 FULL none FALSE Description 1 Structural 'STR' DImodel 2 Species identity 'ID' DImodel 3 Average interactions 'AV' DImodel 4 Functional group effects 'FG' DImodel 5 Additive species contributions to interactions 'ADD' DImodel 6 Separate pairwise interactions 'FULL' DImodel Selected model: Functional group effects 'FG' DImodel -------------------------------------------------------------------------------- Step 3: No investigation of treatment effect included, since no treatment was specified (argument 'treat' omitted) -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model 'community' is a factor with 25 levels, one for each unique set of proportions. model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) DI Model 1 Selected 60 143.8881 2.3981 DI Model 2 Reference 42 84.6499 2.0155 18 59.2382 1.6329 0.0954 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Sequential analysis: Investigating only non-diversity experimental design structures model Resid. Df Resid. SSq Resid. MSq Df SSq F 1 Intercept only 203 2040.865 10.0535 2 block 202 1915.455 9.4825 1 125.4102 13.1363 3 block + density 201 1911.058 9.5078 1 4.3968 0.4605 4 block + density + treat 200 1909.368 9.5468 1 1.6904 0.1771 Pr(>F) 1 2 4e-04 3 0.4982 4 0.6744 -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the AV model, including all available structures Theta estimate: 0.7218 Selection by BIC Warning: DI Model with the lowest BIC will be selected, even if the difference is very small. Please inspect other models to see differences in BIC. BIC DI_Model treat theta 1 818.6809 AV 'nitrogen' FALSE 2 809.1482 AV 'nitrogen' TRUE Description 1 Average interactions 'AV' DImodel with treatment 2 Average interactions 'AV' DImodel with treatment, estimating theta The test concludes that theta is significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions All models include block and density Selection by BIC Warning: DI Model with the lowest BIC will be selected, even if the difference is very small. Please inspect other models to see differences in BIC. BIC DI_Model treat theta 1 1061.7447 STR 'nitrogen' FALSE 2 958.1681 ID 'nitrogen' FALSE 3 809.1482 AV 'nitrogen' TRUE 4 764.8121 FG 'nitrogen' TRUE 5 802.9692 ADD 'nitrogen' TRUE 6 760.8406 FULL 'nitrogen' TRUE Description 1 Structural 'STR' DImodel with treatment 2 Species identity 'ID' DImodel with treatment 3 Average interactions 'AV' DImodel with treatment, estimating theta 4 Functional group effects 'FG' DImodel with treatment, estimating theta 5 Additive species contributions to interactions 'ADD' DImodel with treatment, estimating theta 6 Separate pairwise interactions 'FULL' DImodel with treatment, estimating theta Selected model: Separate pairwise interactions 'FULL' DImodel with treatment, estimating theta -------------------------------------------------------------------------------- Step 3: Investigating the treatment effect Selection by BIC Warning: DI Model with the lowest BIC will be selected, even if the difference is very small. Please inspect other models to see differences in BIC. BIC DI_Model treat theta 1 775.4057 FULL none FALSE 2 760.8309 FULL 'nitrogen' FALSE Description 1 Separate pairwise interactions 'FULL' DImodel 2 Separate pairwise interactions 'FULL' DImodel with treatment Selected model: Separate pairwise interactions 'FULL' DImodel with treatment -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model 'community' is a factor with 25 levels, one for each unique set of proportions. model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) DI Model 1 Selected 190 336.5572 1.7714 DI Model 2 Reference 176 228.7338 1.2996 14 107.8234 5.9261 <0.0001 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the AV model, including all available structures Theta estimate: 1.0331 Selection by AIC Warning: DI Model with the lowest AIC will be selected, even if the difference is very small. Please inspect other models to see differences in AIC. AIC DI_Model treat theta 1 80.22723 AV none FALSE 2 82.19915 AV none TRUE Description 1 Average interactions 'AV' DImodel 2 Average interactions 'AV' DImodel, estimating theta The test concludes that theta is not significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions Selection by AIC Warning: DI Model with the lowest AIC will be selected, even if the difference is very small. Please inspect other models to see differences in AIC. AIC DI_Model treat theta Description 1 100.39643 STR none FALSE Structural 'STR' DImodel 2 100.67327 ID none FALSE Species identity 'ID' DImodel 3 80.22723 AV none FALSE Average interactions 'AV' DImodel 4 53.36836 FULL none FALSE Separate pairwise interactions 'FULL' DImodel Functional groups (argument 'FG') were not specified, and therefore not investigated. The 'ADD' variables are only computed for > 3 species cases as the 'ADD' model is not informative for the 2 or 3 species case. Selected model: Separate pairwise interactions 'FULL' DImodel -------------------------------------------------------------------------------- Step 3: No investigation of treatment effect included, since no treatment was specified (argument 'treat' omitted) -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model Lack of fit test is not possible as there are no repititions of communities in the data. Skipping step 4 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the FULL model, including all available structures Theta estimate: 1.4017 Selection by AIC Warning: DI Model with the lowest AIC will be selected, even if the difference is very small. Please inspect other models to see differences in AIC. AIC DI_Model treat theta 1 363.9277 FULL none FALSE 2 365.5317 FULL none TRUE Description 1 Separate pairwise interactions 'FULL' DImodel 2 Separate pairwise interactions 'FULL' DImodel, estimating theta The test concludes that theta is not significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions Selection by AIC Warning: DI Model with the lowest AIC will be selected, even if the difference is very small. Please inspect other models to see differences in AIC. AIC DI_Model treat theta Description 1 385.8681 STR none FALSE Structural 'STR' DImodel 2 379.6018 ID none FALSE Species identity 'ID' DImodel 3 363.9277 FULL none FALSE Separate pairwise interactions 'FULL' DImodel Functional groups (argument 'FG') were not specified, and therefore not investigated. The 'ADD' variables are only computed for > 3 species cases as the 'ADD' model is not informative for the 2 or 3 species case. Selected model: Separate pairwise interactions 'FULL' DImodel -------------------------------------------------------------------------------- Step 3: No investigation of treatment effect included, since no treatment was specified (argument 'treat' omitted) -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model 'community' is a factor with 7 levels, one for each unique set of proportions. model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) DI Model 1 Selected 61 974.9396 15.9826 DI Model 2 Reference 57 889.4896 15.6051 4 85.4501 1.3689 0.2561 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Step 1: Investigating whether theta is equal to 1 or not for the AV model, including all available structures Theta estimate: 0.9768 Selection using F tests Description DI Model 1 Average interactions 'AV' DImodel DI Model 2 Average interactions 'AV' DImodel, estimating theta DI_model treat estimate_theta Resid. Df Resid. SSq Resid. MSq Df DI Model 1 AV none FALSE 60 281.4708 4.6912 DI Model 2 AV none TRUE 59 281.2109 4.7663 1 SSq F Pr(>F) DI Model 1 DI Model 2 0.2599 0.0545 0.8162 The test concludes that theta is not significantly different from 1. -------------------------------------------------------------------------------- Step 2: Investigating the interactions Selection using F tests Description DI Model 1 Structural 'STR' DImodel DI Model 2 Species identity 'ID' DImodel DI Model 3 Average interactions 'AV' DImodel DI Model 4 Functional group effects 'FG' DImodel DI Model 5 Separate pairwise interactions 'FULL' DImodel DI_model treat estimate_theta Resid. Df Resid. SSq Resid. MSq Df DI Model 1 STR none FALSE 63 1462.1967 23.2095 DI Model 2 ID none FALSE 61 1165.7315 19.1104 2 DI Model 3 AV none FALSE 60 281.4708 4.6912 1 DI Model 4 FG none FALSE 59 280.4582 4.7535 1 DI Model 5 FULL none FALSE 58 205.4254 3.5418 1 SSq F Pr(>F) DI Model 1 DI Model 2 296.4652 41.8521 <0.0001 DI Model 3 884.2608 249.663 <0.0001 DI Model 4 1.0126 0.2859 0.5949 DI Model 5 75.0328 21.1848 <0.0001 The 'ADD' variables are only computed for > 3 species cases as the 'ADD' model is not informative for the 2 or 3 species case. Selected model: Separate pairwise interactions 'FULL' DImodel -------------------------------------------------------------------------------- Step 3: No investigation of treatment effect included, since no treatment was specified (argument 'treat' omitted) -------------------------------------------------------------------------------- Step 4: Comparing the final selected model with the reference (community) model 'community' is a factor with 16 levels, one for each unique set of proportions. model Resid. Df Resid. SSq Resid. MSq Df SSq F Pr(>F) DI Model 1 Selected 58 205.4254 3.5418 DI Model 2 Reference 48 179.8704 3.7473 10 25.555 0.682 0.7356 -------------------------------------------------------------------------------- autoDI is limited in terms of model selection. Exercise caution when choosing your final model. -------------------------------------------------------------------------------- Fitted model: Average interactions 'AV' DImodel Theta estimate: 0.7603 Fitted model: Custom DI model Fitted model: Separate pairwise interactions 'FULL' DImodel Fitted model: Additive species contributions to interactions 'ADD' DImodel Fitted model: Evenness 'E' DImodel Fitted model: Average interactions 'AV' DImodel Fitted model: Additive species contributions to interactions 'ADD' DImodel Fitted model: Functional group effects 'FG' DImodel Fitted model: Separate pairwise interactions 'FULL' DImodel Fitted model: Species identity 'ID' DImodel Fitted model: Evenness 'E' DImodel Fitted model: Average interactions 'AV' DImodel Theta estimate: 0.7603 Fitted model: Additive species contributions to interactions 'ADD' DImodel Theta estimate: 0.786 Fitted model: Functional group effects 'FG' DImodel Theta estimate: 0.7524 Fitted model: Separate pairwise interactions 'FULL' DImodel Theta estimate: 0.602 Fitted model: Species identity 'ID' DImodel Theta estimate: NA Fitted model: Functional group effects 'FG' DImodel Fitted model: Functional group effects 'FG' DImodel Fitted model: Separate pairwise interactions 'FULL' DImodel Fitted model: Average interactions 'AV' DImodel Theta estimate: 0.7603 Generated contrast matrix: p1_ID p2_ID p3_ID p4_ID AV nitrogen50 densityhigh Test 1 0 0 0 0 0 0 0 `nitrogen150:densityhigh` Test 1 0 Generated contrast matrix: p1_ID p2_ID p3_ID p4_ID AV nitrogen50 densityhigh Test 1 1 1 -1 -1 0 0 0 `nitrogen150:densityhigh` Test 1 0 Generated contrast matrix: p1_ID p2_ID p3_ID p4_ID AV nitrogen50 densityhigh Test 1 1 1 -1 -1 0 0 0 `nitrogen150:densityhigh` Test 1 0 Generated contrast matrix: p1_ID p2_ID p3_ID p4_ID AV nitrogen50 densityhigh Test 1 1 1 -1 -1 0 0 0 `nitrogen150:densityhigh` Test 1 0 Generated contrast matrix: p1_ID p2_ID p3_ID p4_ID AV nitrogen50 densityhigh 50v150 0 0 0 0 0 1 0 `nitrogen150:densityhigh` 50v150 0 Fitted model: Functional group effects 'FG' DImodel [ FAIL 0 | WARN 0 | SKIP 0 | PASS 166 ] > > proc.time() user system elapsed 25.09 0.71 25.81