#### Diversity Standardization: Coverage-based and size-based approach ####


## OTU richness(q=0)/Shannon's(q=1)/Simpson's(q=2) index

library(iNEXT)

Result_coverage=estimateD(otutable, datatype = "abundance", base = "coverage", level = fixed_coverage)

Result_size=estimateD(otutable, datatype = "abundance", base = "size", level = fixed_size)

## Phylogenetic diversity
library(ape)
library(devtools)
library(ade4)
library(iNextPD)

tree=readChar("tree.txt",file.info("tree.txt")$size)
phy=newick2phylog(tree)

otutable=read.table("otutable.txt",header=T)
PD=iNextPD(otutable[,2:ncol(otutable)],otutable[,1],phy, q=0, knots=knots, endpoint=endpoint)   

# Picking fixed diversity value (coverage-based or size-based) 
for (i in 1:ncol(otutable)){
  for(k in 1:knots){
    pd=PD$iNextPDEst[[[i]]$m[k]==Fixed_Coverage/size
    if(pd==TRUE){
      Result[i,1]=PD$iNextPDEst[[i]]$SC[k] #Fixed coverage or size
      Result[[i,2]=PD$iNextPDEst[[[i]]$m[k] #number of reads of fixed PD
      Result[[i,3]=PD$iNextPDEst[[[i]]$qPD[k] #PD estimates
      stop
    }
  }
}

rownames(Result)=names(PD$iNextPDEst)
colnames(Result)=c("Coverage/Size","Numb_of_reads","PD")



#### GLMM analysis and regression plot ####


library(gdata)
library(lavaan)
library(lme4)
library(nlme)
library(MASS)

R=Data$Season
Summer=which(R == "Summer")
Spring=which(R == "Spring")

par(mfrow = c(3, 2))
par(oma = c(1, 1, 2, 2)) 
par(mar = c(4, 4, 1, 1)) 

# Testing Hypothesis I 

X=Data$C_HNF_Sh
Y=Data$HBA_biomass
glmm_plot(X,Y,R)
mtext('(a)', side = 3, line = 0, at=1, outer = F,cex=1)
mtext('Predator diversity', side = 1, outer = F,cex=0.8,line=2.4)
mtext('Prey biomass (ngC)', side = 2,  outer = F,cex=0.8,line=2.5)

X=Data$C_HNF_Sh
Y=Data$PPBR_H
glmm_plot(X,Y,R)
mtext('(b)', side = 3, line = 0, at=1, outer = F,cex=1)
mtext('Predator diversity', side = 1,  outer = F,cex=0.8,line=2.5)
mtext('PPBR', side = 2, outer = F,cex=0.8,line=2.4)


#Testing Hypothesis II

X=Data$C_HB_Sh
Y=Data$HNF_biomass
glmm_plot(X,Y,R)
mtext('(c)', side = 3, line = 0, at=5, outer = F,cex=1)
mtext('Prey diversity', side = 1,  outer = F,cex=0.8,line=2.5)
mtext('Predator biomass (ngC)', side = 2, outer = F, cex=0.8, line=2.4)

X=a$C_NonArchaeaCyano_Sh
Y=a$PPBR_H
points(X[Spring],Y[Spring],pch=2,lwd = 1.5,cex=1.1)
glmm_plot(X,Y,R)
mtext('(d)', side = 3, line = 0, at=5, outer = F,cex=1)
mtext('Prey diversity', side = 1,  outer = F,cex=0.8, line=2.5)
mtext('PPBR', side = 2, outer = F, cex=0.8, line=2.4)


#testing Hypothesis III

X=Data$C_HB_Sh
Y=Data$C_HNF_Sh
glmm_plot_III(X,Y,R)
mtext('(e)', side = 3, line = 0, at=1, outer = F,cex=1)
mtext('Prey diversity', side = 1,  outer = F,cex=0.8,line=2.5)
mtext('Predator diversity', side = 2, outer = F,cex=0.8,line=2.4)


## Function: glmm_plot()

glmm_plot=function(X,Y,R){

  plot(Y~X,type = "n", log = "y", xlab='',ylab='',las=1)
  points(X[Summer],Y[Summer],ylab ='',xlab ='',pch=1,lwd = 1.5,cex=1.1, xlim=c(min(X),max(X)),ylim=c(min(Y),max(Y)))
  points(X[Spring],Y[Spring],pch=2,lwd = 1.5,cex=1.1)
	
  PQL=glmmPQL(Y~X, ~1 | R, family = gaussian(link = "log"),verbose = FALSE)
  t=summary(PQL)[]$tTable[2,5]
  rp = vector('expression',1)
  if(t<0.01){
    rp[1]=substitute(expression(paste(italic("P")," < 0.01",sep="")))[2]
  }else{
    rp[1]=substitute(expression(italic(P)==MYOTHERVALUE), list(MYOTHERVALUE =format(t, digits = 2)))[2]    
  }
  legend('topright', legend = rp, bty = 'n',cex=1)

 if(t<0.0500){
    abline(lm(log10(Y)~X),lty=1)
	}
  }


## Function: glmm_plot_III()

glmm_plot_III=function(X,Y,R){

  plot(Y~X,type = "n", log = "y", xlab='',ylab='',las=1)
  points(X[Summer],Y[Summer],ylab ='',xlab ='',pch=1,lwd = 1.5,cex=1.1, xlim=c(min(X),max(X)),ylim=c(min(Y),max(Y)))
  points(X[Spring],Y[Spring],pch=2,lwd = 1.5,cex=1.1)

  PQL=glmmPQL(Y~X, ~1 | R, family = gaussian,
                 verbose = FALSE)
  t=summary(PQL)[]$tTable[2,5]
      rp = vector('expression',1)
  if(t<0.01){
    rp[1]=substitute(expression(paste(italic("P")," < 0.01",sep="")))[2]
  }else{
    rp[1]=substitute(expression(italic(P)==MYOTHERVALUE), list(MYOTHERVALUE =format(t, digits = 2)))[2]    
  }
  legend('topright', legend = rp, bty = 'n',cex=1)

  if(t<0.0500){
    abline(lm(Y~X),lty=1)
	}
  }


####PLS_SEM analysis and plot####


PLS=function(Data){
##Do scalping
  sa=matrix(0,nrow(Data),ncol(Data))
  for (i in 1:ncol(Data)){
    sData[,i]=scale(Data[,i])
  }
  colnames(sData)=colnames(Data)
  rownames(sData)=rownames(Data)
  ##model setting
  Tem= c(0,0,0,0,0,0)
  Nut= c(0,0,0,0,0,0)
  HBD= c(1,1,0,0,0,0)
  HNFD= c(1,1,1,0,0,0)
  HNFB= c(1,1,1,1,0,0)
  HBB= c(1,1,1,1,1,0)
  sat.inner = rbind(Tem,Nut,HBD,HNFD,HNFB,HBB)
  sat_mod=rep("A", 6)
  sat.outer = list(1,2:4,5,6,7,8)
  satpls = plspm(sData, sat.inner, sat.outer,scheme="centroid", br=1000,mode=sat_mod)
  
  r=satpls$path
  
  txt=matrix(0,nrow(r),nrow(r))
  for(j in 1:nrow(r)){
    for (i in 1:ncol(r)){
      if ( abs(r[i,j])>0.26) {
        txt[i,j]=1
      } else if (r[i,j]==0) {
        txt[i,j]=0
      } else
        txt[i,j]=0.5
    }
  }
  
  line.width=matrix(0,nrow(r),nrow(r))
  for(j in 1:nrow(r)){
    for (i in 1:ncol(r)){
      if (abs(r[i,j])>0.26) {
        line.width[i,j]=2.6
      } else if (r[i,j]==0) {
        line.width[i,j]=0
      } else
        line.width[i,j]=1
    }
  }
  
  line.color=matrix(0,nrow(r),nrow(r))
  for(j in 1:nrow(r)){
    for (i in 1:ncol(r)){
      if (abs(r[i,j])>=0.26) {
        line.color[i,j]="black"
      } else if (r[i,j]==0) {
        line.color[i,j]=0
      } else
        line.color[i,j]="grey"
    }
  }
  
  D=round(satpls$path,digits=2)
  d=format(D)
  dd=as.data.frame(d)
  
  plotmat(D,arr.type="triangle", arr.length=0.3, shadow.size =0,
          box.type="ellipse", box.col="grey", box.prop =0.6, box.size=0.08,
          cex.txt =txt, arr.lcol=line.color, arr.col=line.color,
          arr.lwd =line.width,curve =0, arr.pos =0.65)
 
 t=summary(satpls)$gof
  rp = vector('expression',1)
  rp[1]=substitute(expression(italic(gof)==MYOTHERVALUE), list(MYOTHERVALUE =format(t, digits = 3)))[2]    
  legend('bottomright', legend = rp, bty = 'n',cex=0.8)
}


#### Pearson's correlation coefficient and P-value####

Result=matrix(0, ncol(Data), ncol(Data))
Result_pvalue=matrix(0, ncol(Data), ncol(Data))
for(i in 1:ncol(Data)){
  for(j in 1:ncol(a)){    
    Result[i,j]=cor(as.numeric(Data[,i]),as.numeric(Data[,j]),use= "na.or.complete")
    Result_pvalue[i,j]=cor.test(as.numeric(Data[,i]),as.numeric(Data[,j]))$p.value
  }
}
colnames(Result)=colnames(Data)
rownames(Result)=colnames(Data)
colnames(Result_pvalue)=colnames(Data)
rownames(Result_pvalue)=colnames(Data)