-
Notifications
You must be signed in to change notification settings - Fork 6
/
TS_basicSEM.R
231 lines (160 loc) · 9.11 KB
/
TS_basicSEM.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
###################################
# Basic piecewise SEM #
# for each combination of #
# phylogenetic and functional #
# diversity metric: #
# 1) FDis + eMNTD #
# 2) FRic + eMNTD #
# 3) FDis + eMPD #
# 4) FRic + eMPD #
###################################
require(dplyr)
require(piecewiseSEM)
library(semPlot)
library(lmerTest)
library(nlme)
library(car)
# Prepare data for analysis
stab<-read.delim("data.csv",sep=",",header=T)
stab<-filter(stab,Site!="BIODEPTH_GR") # should get rid of site where we didn't have good trait coverage
stab_4<-select(stab,Site,Study_length,UniqueID,SppN,eMPD,eMNTD,ePSE,sMPD,sMNTD,FDis4,FRic4,PCAdim1_4trts,SLA, LDMC, LeafN, LeafP,
Plot_TempStab,Plot_Biomassxbar, Plot_Biomasssd,Gross_synchrony,
mArid,sdAridity)
# for plots with ONLY 1 spp, we assume that a species #is perfectly synchronized with itself
stab_4$Gross_synchrony<-ifelse(is.na(stab_4$Gross_synchrony)==TRUE,1,stab_4$Gross_synchrony)
# convert synchrony metrics to different scale
stab_4$GrossAsynchrony_s<-stab_4$Gross_synchrony*-1
# further adjustments
stab_4$SppN<-as.numeric(stab_4$SppN)
stab_4$TS_lg2<-log(stab_4$Plot_TempStab,base=2)
stab_4$lg2SppN <- log(stab_4$SppN,2)
stab_4$lg2_mArid <-log(stab_4$mArid,base=2)
# Filter out NAs for Asynchrony and FRic4
stab_4<-filter(stab_4, is.na(PlotAsynchrony_s)==FALSE)
stab_444<-filter(stab_4, is.na(FRic4)==FALSE)
############################
# fit piecewise SEM models #
############################
# Control list set up for LMM in nlme
bb<-lmeControl(msMaxIter=0,msVerbose = TRUE,opt="optim",maxIter=100,optimMEthod="L-BFGS-B") ######## "msMaxIter=0" is important in here!!!
cc<-lmeControl(opt="optim")
##################
# FDis_eMNTD #####
##################
modList2=list(
lme(eMNTD~lg2SppN,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(FDis4~lg2SppN,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(GrossAsynchrony_s~lg2SppN+FDis4+eMNTD+sdAridity+lg2_mArid,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(TS_lg2~GrossAsynchrony_s+PCAdim1_4trts+lg2SppN+eMNTD+FDis4+sdAridity+lg2_mArid,random=~1+lg2SppN|Site, control=cc,data=stab_444)
)
lapply(modList2, plot)
# Explore distribution of residuals
lapply(modList2, function(i) hist(resid(i)))
# Look at variance inflation factors
lapply(modList2[3:4], vif)
sem.fit(modList2,stab_444,corr.errors=c("eMNTD~~FDis4"),conditional=T,
model.control = list(lmeControl(opt = "optim"))) #naive model
emntdfdis.fit<-sem.fit(modList2,stab_444,corr.errors=c("eMNTD~~FDis4","FDis4 ~~ PCAdim1_4trts"),conditional=T,
model.control = list(lmeControl(opt = "optim")))
emntdfdis.fit<-cbind(emntdfdis.fit$Fisher.C,emntdfdis.fit$AIC)
emntdfdis.fit$ModClass<-"FDis_eMNTD"
ts_emntd2<-sem.coefs(modList2,stab_444,standardize="scale",corr.errors=c("eMNTD~~FDis4","FDis4 ~~ PCAdim1_4trts"))
ts_emntd2$ModClass<-"FDis_eMNTD"
mf_ts_emntd<-sem.model.fits(modList2)
mf_ts_emntd$ResponseVars<-c("eMNTD","FDis4","Asynchrony","Temp_Stability")
mf_ts_emntd$PredVars<-c("lg2SppN","lg2SppN","lg2SppN,eMNTD,FDis4, lg2_mArid, Arid_sd","eMNTD, FDis4,Asynchrony,lg2SppN,lg2_mArid")
mf_ts_emntd$ModClass<-"FDis_eMNTD"
# write results to file
write.table(ts_emntd2,"TS_emntd_fdis_sem_coefs_SIMPLE.csv",sep=",",row.names=F) # sem path coefficients
write.table(mf_ts_emntd,"TS_emntd_fdis_model_fits_SIMPLE.csv",sep=",",row.names=F) # model fits (marginal + conditional R2)
write.table(emntdfdis.fit,"TS_emntd_fdis_semfit_SIMPLE.csv",sep=",",row.names=F) # sem model fit
#######################
## FRic4 - eMNTD ###
#######################
modList22=list(
lme(eMNTD~lg2SppN,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(FRic4~lg2SppN,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(GrossAsynchrony_s~lg2SppN+FRic4+eMNTD+sdAridity+lg2_mArid,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(TS_lg2~GrossAsynchrony_s+PCAdim1_4trts+lg2SppN+eMNTD+FRic4+sdAridity+lg2_mArid,random=~1+lg2SppN|Site, control=cc,data=stab_444)
)
lapply(modList22, plot)
# Explore distribution of residuals
lapply(modList22, function(i) hist(resid(i)))
# Look at variance inflation factors
lapply(modList22[3:4], vif)
sem.fit(modList22,stab_444,corr.errors=c("eMNTD~~FRic4"),conditional=T,
model.control = list(lmeControl(opt = "optim"))) #naive model
emntdfric.fit<-sem.fit(modList22,stab_444,corr.errors=c("FRic4~~eMNTD","FRic4 ~~ PCAdim1_4trts"),conditional=T,
model.control = list(lmeControl(opt = "optim"))) # add eMNTD as predictor of TS
emntdfric.fit<-cbind(emntdfric.fit$Fisher.C,emntdfric.fit$AIC)
emntdfric.fit$ModClass<-"FRic_eMNTD"
ts_emntd2<-sem.coefs(modList22,stab_444,standardize="scale",corr.errors=c("eMNTD~~FRic4","FRic4 ~~ PCAdim1_4trts"))
ts_emntd2$ModClass<-"FRic_eMNTD"
mf_ts_emntd<-sem.model.fits(modList22)
mf_ts_emntd$ResponseVars<-c("eMNTD","FRic4","Asynchrony","Temp_Stability")
mf_ts_emntd$PredVars<-c("lg2SppN","lg2SppN","lg2SppN,eMNTD,FRic4,mArid,Arid_sd","Asynchrony,FRic4,eMNTD,lg2SppN,mArid,Arid_sd")
mf_ts_emntd$ModClass<-"FRic_eMNTD"
# write results to file
write.table(ts_emntd2,"TS_emntd_fric_sem_coefs_SIMPLE.csv",sep=",",row.names=F) # sem path coefficients
write.table(mf_ts_emntd,"TS_emntd_fric_model_fits_SIMPLE.csv",sep=",",row.names=F) # model fits (marginal + conditional R2)
write.table(emntdfric.fit,"TS_emntd_fric_semfit_SIMPLE.csv",sep=",",row.names=F) # sem model fit
##################
# FDis_eMPD ######
##################
modList3=list(
lme(eMPD~lg2SppN,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(FDis4~lg2SppN,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(GrossAsynchrony_s~lg2SppN+FDis4+eMPD+sdAridity+lg2_mArid,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(TS_lg2~GrossAsynchrony_s+PCAdim1_4trts+lg2SppN+FDis4+eMPD+sdAridity+lg2_mArid,random=~1+lg2SppN|Site, control=cc,data=stab_444)
)
lapply(modList3, plot)
# Explore distribution of residuals
lapply(modList3, function(i) hist(resid(i)))
# Look at variance inflation factors
lapply(modList3[3:4], vif)
sem.fit(modList3,stab_444,corr.errors=c("eMPD~~FDis4"),conditional=T,
model.control = list(lmeControl(opt = "optim"))) #naive model
empdfdis.fit<-sem.fit(modList3,stab_444,corr.errors=c("eMPD~~FDis4","FDis4 ~~ PCAdim1_4trts","eMPD ~~ PCAdim1_4trts"),conditional=T,
model.control = list(lmeControl(opt = "optim"))) #
empdfdis.fit<-cbind(empdfdis.fit$Fisher.C,empdfdis.fit$AIC)
empdfdis.fit$ModClass<-"FDis_eMPD"
ts_empd2<-sem.coefs(modList3,stab_444,standardize="scale",corr.errors=c("eMPD~~FDis4","FDis4 ~~ PCAdim1_4trts","eMPD ~~ PCAdim1_4trts"))
ts_empd2$ModClass<-"FDis_eMPD"
mf_ts_empd<-sem.model.fits(modList3)
mf_ts_empd$ResponseVars<-c("eMPD","FDis4","Asynchrony","Temp_Stability")
mf_ts_empd$PredVars<-c("lg2SppN","lg2SppN","lg2SppN,eMPD,FDis4,mArid,Arid_sd","Asynchrony,FDis, eMPD,lg2SppN,mArid,Arid_sd")
mf_ts_empd$ModClass<-"FDis_eMPD"
# write results to file
write.table(ts_empd2,"TS_empd_fdis_sem_coefs_SIMPLE.csv",sep=",",row.names=F) # sem path coefficients
write.table(mf_ts_empd,"TS_empd_fdis_model_fits_SIMPLE.csv",sep=",",row.names=F) # model fits (marginal + conditional R2)
write.table(empdfdis.fit,"TS_empd_fdis_semfit_SIMPLE.csv",sep=",",row.names=F) # sem model fit
#######################
## FRic - eMPD #######
#######################
modList33=list(
lme(eMPD~lg2SppN,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(FRic4~lg2SppN,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(GrossAsynchrony_s~lg2SppN+FRic4+eMPD+sdAridity+lg2_mArid,random=~1+lg2SppN|Site,control=cc,data=stab_444),
lme(TS_lg2~GrossAsynchrony_s+PCAdim1_4trts+lg2SppN+FRic4+eMPD+sdAridity+lg2_mArid,random=~1+lg2SppN|Site, control=cc,data=stab_444)
)
lapply(modList33, plot)
# Explore distribution of residuals
lapply(modList33, function(i) hist(resid(i)))
# Look at variance inflation factors
lapply(modList33[3:4], vif)
sem.fit(modList33,stab_444,corr.errors=c("eMPD~~FRic4"),conditional=T,
model.control = list(lmeControl(opt = "optim"))) #naive model
smpdfdis.fit<-sem.fit(modList33,stab_444,corr.errors=c("eMPD~~FRic4","FRic4 ~~ PCAdim1_4trts"," eMPD ~~ PCAdim1_4trts"),conditional=T,
model.control = list(lmeControl(opt = "optim")))
smpdfdis.fit<-cbind(smpdfdis.fit$Fisher.C,smpdfdis.fit$AIC)
smpdfdis.fit$ModClass<-"FRic4_eMPD"
ts_smpd2<-sem.coefs(modList33,stab_444,standardize="scale",corr.errors=c("eMPD~~FRic4","FRic4 ~~ PCAdim1_4trts"," eMPD ~~ PCAdim1_4trts"))
ts_smpd2$ModClass<-"FRic4"
mf_ts_smpd<-sem.model.fits(modList33)
mf_ts_smpd$ResponseVars<-c("eMPD","FRic4","Asynchrony","Temp_Stability")
mf_ts_smpd$PredVars<-c("lg2SppN","lg2SppN","lg2SppN,eMPD,FRic4,mArid,Arid_sd","Asynchrony,F-S,FRic, eMPD,lg2SppN,mArid,Arid_sd")
mf_ts_smpd$ModClass<-"FRic4_eMPD"
# write results to file
write.table(ts_smpd2,"TS_empd_fric_sem_coefs_SIMPLE.csv",sep=",",row.names=F) # sem path coefficients
write.table(mf_ts_smpd,"TS_empd_fric_model_fits_SIMPLE_April2018.csv",sep=",",row.names=F) # model fits (marginal + conditional R2)
write.table(smpdfdis.fit,"TS_empd_fric_semfit_SIMPLE_April2018.csv",sep=",",row.names=F) # sem model fit