Schefferville - Phylogenetic relationships of generalists and specialists

Includes analyses with 34 focal species (including T. alpinum)

Author: Tammy L. Elliott

Date:December 15, 2015

R version 3.2.0

Three metric dataframe

g.s<-cbind(h.inv.ordered.short[,2], gen.spec.ordered.short[,2], extent.sqrt.rescale)

colnames(g.s)<-c("h.inv", "jaccard", "extent")
rownames(g.s)<-h.inv.ordered.rownames
head(g.s)

##         h.inv   jaccard    extent
## ALP 0.7259687 0.5464144 0.3694924
## ANG 0.7733402 0.9397053 0.6454640
## AQU 1.0000000 0.6922376 1.0000000
## BIG 0.7257200 0.3302116 0.4700878
## BRU 0.9121210 0.5481525 0.4761819
## CAN 0.9752128 0.9495533 0.7759281

---

Pearson’s correlation test for each two pairs

#Do Pearson's correlation test for each two pairs
extent.H.inv.cor.test<-cor.test(extent,H.inv)
extent.cooccur.cor.test<-cor.test(extent,cooccur)
H.inv.coccur.cor.test<-cor.test(H.inv, cooccur)

#Extent with habitat index
extent.H.inv.cor.test

## 
##  Pearson's product-moment correlation
## 
## data:  extent and H.inv
## t = 2.1883, df = 32, p-value = 0.03607
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  0.02575616 0.62294402
## sample estimates:
##       cor 
## 0.3607795

#Extent with co-occurrence (Jaccard) index
extent.cooccur.cor.test

## 
##  Pearson's product-moment correlation
## 
## data:  extent and cooccur
## t = 2.8002, df = 32, p-value = 0.008588
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  0.1240828 0.6798125
## sample estimates:
##       cor 
## 0.4436334

# Habitat and co-occurrence indices
H.inv.coccur.cor.test

## 
##  Pearson's product-moment correlation
## 
## data:  H.inv and cooccur
## t = 2.2104, df = 32, p-value = 0.03435
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  0.02939639 0.62516827
## sample estimates:
##       cor 
## 0.3639442

---

Blomberg’s K

#find Blomberg's 'K' value
#rownames(extent.sqrt.rescale)<-h.inv.ordered.rownames


#Extent K
extent.K<-Kcalc(extent, sedges.tree, checkdata=TRUE)
extent.K

##            [,1]
## [1,] 0.07953233

#Habitat K
H.inv.K<-Kcalc(H.inv, sedges.tree, checkdata=TRUE)
H.inv.K

##            [,1]
## [1,] 0.09785613

#Cooccur K
cooccur.K<-Kcalc(cooccur, sedges.tree, checkdata=TRUE)
cooccur.K

##            [,1]
## [1,] 0.07196142

# Use phylosignal to calculate so get p values from phylogenetic contrasts

# Extent phylosignal
extent.ps<-phylosignal(extent,sedges.tree, reps=999, checkdata=TRUE)
extent.ps

##            K PIC.variance.obs PIC.variance.rnd.mean PIC.variance.P
## 1 0.07953233       0.01485334             0.0100982          0.904
##   PIC.variance.Z
## 1       1.270905

# Habitat phylosignal
H.inv.ps<-phylosignal(H.inv,sedges.tree, reps=999, checkdata=TRUE)
H.inv.ps

##            K PIC.variance.obs PIC.variance.rnd.mean PIC.variance.P
## 1 0.09785613       0.01825359             0.0148916          0.775
##   PIC.variance.Z
## 1      0.6390491

# Co-occurrence phylosignal
cooccur.ps<-phylosignal(cooccur,sedges.tree, reps=999, checkdata=TRUE)
cooccur.ps

##            K PIC.variance.obs PIC.variance.rnd.mean PIC.variance.P
## 1 0.07196142       0.01933529            0.01146276          0.944
##   PIC.variance.Z
## 1        1.96972

#dev.new(width=3.15, height=4.2)
par(mai=c(1,0,0, 0.15))
plot(sedges.tree,x.lim=100,show.tip.label = FALSE,edge.width=4)

# Then we’ll plot line segments corresponding to each trait. The plotting parameters in here are very specific to this data set, and should be viewed as a template. The great thing is that you can change any of them yourself – you’ll need to play around with them (e.g., changing the starting positions of the segments, or their scaling) to get your own data plotted.

    nspecies <- length(sedges.tree$tip.label)

    segments(rep(65,nspecies),1:nspecies,rep(65,nspecies)+(10*H.inv), 1:nspecies,lwd=3, col="grey40")
    mtext("Habitat", at = 68, side = 1, line = 0, cex=.65,font=2)

    segments(rep(77,nspecies),1:nspecies,rep(77,nspecies)+(10*cooccur), 1:nspecies,lwd=3, col="grey10")
    mtext("Co-occurrence", srt=45,at = 82, side = 1, line = 0, cex=.65,font=2)

    segments(rep(89,nspecies),1:nspecies,rep(89,nspecies)+(10*extent), 1:nspecies,lwd=3, col="grey60")
    mtext("Extent", at = 97, side = 1, line = 0, cex=.65,font=2)
        box(which='outer',lwd=3, col = 'black')

    #textClick("Habitat", cex=1)
    #textClick("Co-occurrence", cex=1)
    #textClick("Extent", cex=1)