setwd("~/TTU Research/JakeLeys/2018_microbiomepaper/190128_analyses")

library(vegan)
library(plyr)
library(dplyr)
library(ggplot2)
library(rcompanion)
library(car)
library(gridExtra)

citation("vegan")
citation("plyr")
citation("dplyr")
citation("ggplot2")
citation("rcompanion")
citation("car")
citation("gridExtra")

#################################################################################################
#Beta diversity - test to determine if skin, soil, water is predictive of the microbiome#########
#################################################################################################

#read in complete dataframe including skin, soil, water samples
env.skin.dat <- read.csv("170912_LEYS_SHARED_SUB_INCLUDES_FACTORS_SupplementalFile1.csv", header=T, row.names=NULL)

#use this function to count any desired OTU read number in sample source (e.g. indicator taxa)
aggregate(Otu26160 ~ sample.source, data = env.skin.dat, sum)

#create supplemental table of study design
sup.table<-rename(count(env.skin.dat, species, state, tn.ecoregion, ecomode, Oo), Freq = n)
write.csv(sup.table, file='supplemental_table_samplecounts.csv')

#counting sample numbers for methods section
sup.table2<-rename(count(env.skin.dat, species, Oo), Freq = n)
write.csv(sup.table2, file='temp.csv')

count(env.skin.dat, tn.ecoregion)

#create otus dataframe for skin, soil, water samples
env.skin.dat.otus <- env.skin.dat[15:16026]

#create factors for skin, soil, water sample source
samplesource <- as.factor(env.skin.dat$sample.source)

#permanova to test if skin, soil, water is predictive of the microbiome
bray.env.skin.dat.otus <- vegdist(env.skin.dat.otus, method="bray",na.rm=TRUE)
adonis(bray.env.skin.dat.otus~samplesource, data=env.skin.dat.otus, permutations = 999, method='bray')

#run NMDS and plot in ggplot2
samplesource.nmds<-metaMDS(env.skin.dat.otus, distance = "bray", k=2, try=20, trymax=999)

data.scores <- as.data.frame(scores(samplesource.nmds))  #Using the scores function from vegan to extract the site scores (rows) and convert to a data.frame
data.scores
data.scores$site <- rownames(data.scores)  # create a column of site names, from the rownames of data.scores
data.scores$Sample <- samplesource #  add factor data for graphing purposes 
head(data.scores)  #look at the data
tail(data.scores)

species.scores <- as.data.frame(scores(samplesource.nmds, "species"))  #Using the scores function from vegan to extract the species scores and convert to a data.frame
species.scores$species <- rownames(species.scores)  # create a column of species, from the rownames of species.scores
head(species.scores) 

f1<-ggplot() + geom_text(data=species.scores,aes(x=NMDS1,y=NMDS2,label=species),alpha=0.5)  # add the species labels

f1<-ggplot()+geom_point(data=data.scores,aes(x=NMDS1,y=NMDS2, shape=Sample, colour=Sample),size=2) # add the point markers
#f1<-f1+geom_text(data=data.scores,aes(x=NMDS1,y=NMDS2,label=samplesource),size=4,vjust=1)   # add the site labels
f1<-f1+scale_shape_manual(values=c(15,17,3)) 
f1

#theme(axis.text.x = element_blank(),  # remove x-axis text
#axis.text.y = element_blank(), # remove y-axis text
#axis.ticks = element_blank(),  # remove axis ticks
#axis.title.x = element_text(size=18), # remove x-axis labels
#axis.title.y = element_text(size=18), # remove y-axis labels
f1<-f1+theme(panel.background = element_blank(), 
             panel.grid.major = element_blank(),  #remove major-grid labels
             panel.grid.minor = element_blank(),  #remove minor-grid labels
             plot.background = element_blank())
f1

################################################################################################
#Since sample source is predictive of the skin microbiome#######################################
#Use indicator analysis to select OTUs indicative of the snake skin microbiome##################
################################################################################################

#indicator analysis performed in mothur using cutoff criteria p < 0.005, Indicator value > 30

skin.dat.indicator <- select(env.skin.dat, sample, sample.source, species, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
skin.dat.indicator.otus <- select(skin.dat.indicator, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
write.csv(skin.dat.indicator, file='skin.dat.indicator.csv')
  
#remove rows that have OTU relative abundances that sum to zero
skin.dat.indicator.removezeros<-skin.dat.indicator[as.logical(rowSums(skin.dat.indicator.otus != 0)), ]

#Didn't remove a single sample. Interesting. 
#So, even if the OTU is predictive of the skin, it still occurs in the soil
#Ran Venn diagrams in mothur to depict this relationship

################################################################################################
#MACROSCALE COMPARISONS ACROSS STATES###########################################################
#Using indicator OTUs###########################################################################
################################################################################################

#create dataframe of only skin samples 
skinonly.dat.indicator <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
skinonly2.dat.indicator <- subset(skinonly.dat.indicator, sample.source=="skin")
skinonly2.dat.indicator.otus <- select(skinonly2.dat.indicator, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)

#create factors for macroscale skin analysis
species <- as.factor(skinonly2.dat.indicator$species)
state <- as.factor(skinonly2.dat.indicator$state)
month <- as.factor(skinonly2.dat.indicator$month)
year <- as.factor(skinonly2.dat.indicator$year)
ecomode <- as.factor(skinonly2.dat.indicator$ecomode)
oo.status <- as.factor(skinonly2.dat.indicator$Oo)

#permanova to test factors species*state*month*year*Oo status are predictive of the skin microbiome
bray.skinonly2.dat.indicator.otus <- vegdist(skinonly2.dat.indicator.otus, method="bray",na.rm=TRUE)
adonis(bray.skinonly2.dat.indicator.otus~species*state*month*year*oo.status, data=skinonly2.dat.indicator.otus, permutations = 999, method='bray')

#Removing interactions from model that are not significant
adonis(bray.skinonly2.dat.indicator.otus~species+state+month+year+oo.status+species:state+species:month+state:month+species:year+month:year+month:oo.status+species:month:oo.status, data=skinonly2.dat.indicator.otus, permutations = 999, method='bray')

#Removing interactions from model that are not significant
adonis(bray.skinonly2.dat.indicator.otus~species+state+month+year+oo.status+species:state+species:month+state:month+species:year+month:oo.status+species:month:oo.status, data=skinonly2.dat.indicator.otus, permutations = 999, method='bray')

##permanova to test factors ecomode*Oo status are predictive of the skin microbiome (swap out species for lifehistory "ecomode")
adonis(bray.skinonly2.dat.indicator.otus~ecomode*oo.status, data=skinonly2.dat.indicator.otus, permutations = 999, method='bray')

#betadisper to determine homogeneity of ecomode groups visualized by nmds 
macro.betadisper <- with(skinonly2.dat.indicator, betadisper(bray.skinonly2.dat.indicator.otus, ecomode))
macro.betadisper
permutest(macro.betadisper)
TukeyHSD(macro.betadisper)
plot(macro.betadisper, label = F, hull = F, ellipse = T)
boxplot(macro.betadisper)

#run NMDS and plot in ggplot2
#Stress value is 0.23 - not great but there's a pattern when graphing ecomode

skinonly2.dat.indicator.nmds<-metaMDS(skinonly2.dat.indicator.otus, distance = "bray", k=2, try=20, trymax=999)

data.scores2 <- as.data.frame(scores(skinonly2.dat.indicator.nmds))  #Using the scores function from vegan to extract the site scores (rows) and convert to a data.frame
data.scores2
data.scores2$site <- rownames(data.scores2)  # create a column of site names, from the rownames of data.scores
data.scores2$Species <- species #  add factor data for graphing purposes 
data.scores2$Month <- month #  add factor data for graphing purposes 
data.scores2$State <- state #  add factor data for graphing purposes 
data.scores2$Ecomode <- ecomode #  add factor data for graphing purposes 
data.scores2$Oo.status <- oo.status #  add factor data for graphing purposes 
head(data.scores2)  #look at the data
tail(data.scores2)

species.scores2 <- as.data.frame(scores(skinonly2.dat.indicator.nmds, "species"))  #Using the scores function from vegan to extract the species scores and convert to a data.frame
species.scores2$species <- rownames(species.scores2)  # create a column of species, from the rownames of species.scores
head(species.scores2) 

macro1<-ggplot() + geom_text(data=species.scores2,aes(x=NMDS1,y=NMDS2,label=species),alpha=0.5)  # add the species labels
macro1<-ggplot()+geom_point(data=data.scores2,aes(x=NMDS1,y=NMDS2, colour=Ecomode),size=2) # add the point markers
macro1<-macro1+theme_classic()
macro1<-macro1+theme(legend.position="none") #remove this line if you want a legend - did this for multipanel figure
macro1<-macro1+ggtitle("Macroscale - Southeastern USA") +
  theme(plot.title = element_text(hjust = 0.5))
macro1

#############################
####DISEASE - MACROSCALE#####
#Using indictors for Ophidiomyces state create dataframe of only skin samples at the MACROSCALE
#############################

macro.sfd <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00597,	Otu00628,	Otu00149,	Otu00450,	Otu00034,	Otu00452,	Otu00316,	Otu00125,	Otu00127,	Otu00124,	Otu00056,	Otu00389,	Otu00066,	Otu00258,	Otu00021,	Otu00006)
macro2.sfd <- subset(macro.sfd, sample.source=="skin")
macro2.sfd.otus <- select(macro2.sfd, Otu00597,	Otu00628,	Otu00149,	Otu00450,	Otu00034,	Otu00452,	Otu00316,	Otu00125,	Otu00127,	Otu00124,	Otu00056,	Otu00389,	Otu00066,	Otu00258,	Otu00021,	Otu00006)

#dbRDA using capscale function in vegan to model OTU data on continuous variable of Ophidiomyces copy number

plotNormalHistogram(macro2.sfd$copy.number) #Data are right skewed
macro.copy.number.sfd.cubed <- sign(macro2.sfd$copy.number) * abs(macro2.sfd$copy.number)^(1/3) #cubed root transform
plotNormalHistogram(macro.copy.number.sfd.cubed) #plot - a little better but still skewed

plotNormalHistogram(macro2.sfd.otus) #data are right skewed
macro2.sfd.otus.cubed <- sign(macro2.sfd.otus) * abs(macro2.sfd.otus)^(1/3) #cubed root transform
plotNormalHistogram(macro2.sfd.otus.cubed) #plot - a little better but still skewed

macro.sfd.capscale <- capscale(macro2.sfd.otus.cubed ~ macro.copy.number.sfd.cubed, macro2.sfd, dist="bray")
macro.sfd.capscale
plot(macro.sfd.capscale) 

anova(macro.sfd.capscale)
permutest(macro.sfd.capscale)

#DISEASE- MACROSCALE - plot capscale results with ggplot2

macro.smry <- summary(macro.sfd.capscale)
macro.smry
macro.df1  <- data.frame(macro.smry$sites[,1:2])       # CAP1 and MDS1
macro.oo <- as.factor(macro2.sfd$Oo) #create factor to graph
macro.df1$Ophidiomyces <- macro.oo
macro.df2  <- data.frame(macro.smry$species[,1:2])     # loadings for CAP1 and MDS1
macro.df1
macro.df2
macro.cap.plot <- ggplot(macro.df1, aes(x=CAP1, y=MDS1, colour=Ophidiomyces)) + 
  geom_point(aes(label=rownames(macro.df1)),size=2) +
  geom_hline(yintercept=0, linetype="dotted") +
  geom_vline(xintercept=0, linetype="dotted") +
  xlim(-3,4)+
  ylim(-2,3)
  #coord_fixed()
macro.cap.plot
macro.cap.plot<-macro.cap.plot+theme_classic()
macro.cap.plot
macro.cap.biplot <- macro.cap.plot +
   geom_text(data=macro.df2, 
            aes(x=CAP1,y=MDS1,label=rownames(macro.df2),
                hjust=0.5*(1-sign(CAP1)),vjust=0.5*(1-sign(MDS1))), 
            color="black", size=2)
macro.cap.biplot<-macro.cap.biplot+theme(legend.position="none") #remove this line if you want a legend - did this for multipanel figure
macro.cap.biplot<-macro.cap.biplot+ggtitle("Macroscale - Southeastern USA") +
  theme(plot.title = element_text(hjust = 0.5))
macro.cap.biplot

#######################################
#MACROSCALE - assessing ALPHA DIVERSITY
#Using indicator taxa for skin
#######################################

#create dataframe of only skin samples 
skinonly.dat.indicator <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
skinonly2.dat.indicator <- subset(skinonly.dat.indicator, sample.source=="skin")
skinonly2.dat.indicator.otus <- select(skinonly2.dat.indicator, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)

macro.invsimp <- diversity(skinonly2.dat.indicator.otus, "inv")
hist(macro.invsimp, main="Inverse Simpson", xlab="", breaks=10) #visualize distribution
shapiro.test(macro.invsimp) #test for normalcy statistically - run the Shapiro-Wilk test of normality.

#test alpha for SPECIES - MACROSCALE
kruskal.test(macro.invsimp ~ species, data=skinonly2.dat.indicator) #species is statistically significant
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(macro.invsimp ~ species, data=skinonly2.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

#test alpha for ECOMODE - MACROSCALE
kruskal.test(macro.invsimp ~ ecomode, data=skinonly2.dat.indicator) #ecomode is statistically significant
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(macro.invsimp ~ ecomode, data=skinonly2.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

#test alpha for Ophidiomyces presence - MACROSCALE
kruskal.test(macro.invsimp ~ Oo, data=skinonly2.dat.indicator) #Ophidiomyces presence is NOT statistically significant
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(macro.invsimp ~ Oo, data=skinonly2.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)


################################################################################################
#MESOSCALE COMPARISONS ACROSS ECOREGIONS OF TENNESSEE###########################################
#Using indicator OTUs###########################################################################
################################################################################################

#create dataframe of only skin samples for Tennessee
tn.skinonly.dat.indicator <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
tn.skinonly2.dat.indicator <- subset(tn.skinonly.dat.indicator, sample.source=="skin")
tn.skinonly3.dat.indicator <- subset(tn.skinonly2.dat.indicator, state=="TN")
tn.skinonly3.dat.indicator.otus <- select(tn.skinonly3.dat.indicator, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)

#create factors for mesoscale skin analysis
tn.species <- as.factor(tn.skinonly3.dat.indicator$species)
tn.ecoregion <- as.factor(tn.skinonly3.dat.indicator$tn.ecoregion)
tn.month <- as.factor(tn.skinonly3.dat.indicator$month)
tn.year <- as.factor(tn.skinonly3.dat.indicator$year)
tn.ecomode <- as.factor(tn.skinonly3.dat.indicator$ecomode)
tn.oo.status <- as.factor(tn.skinonly3.dat.indicator$Oo)

#permanova to test factors species*ecoregion*month*year*Oo status are predictive of the skin microbiome in Tennessee
bray.tn.skinonly3.dat.indicator.otus <- vegdist(tn.skinonly3.dat.indicator.otus, method="bray",na.rm=TRUE)
adonis(bray.tn.skinonly3.dat.indicator.otus~tn.species*tn.ecoregion*tn.month*tn.year*tn.oo.status, data=tn.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#Remove non significant interaction factors from the model
adonis(bray.tn.skinonly3.dat.indicator.otus~tn.species+tn.ecoregion+tn.month+tn.year+tn.oo.status+tn.species:tn.ecoregion+tn.species:tn.month+tn.month:tn.year+tn.species:tn.oo.status, data=tn.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#Remove non significant interaction factors from the model
adonis(bray.tn.skinonly3.dat.indicator.otus~tn.species+tn.ecoregion+tn.month+tn.year+tn.oo.status+tn.species:tn.ecoregion+tn.species:tn.month+tn.species:tn.oo.status, data=tn.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#permanova to test factors ecomode*Oo status are predictive of the skin microbiome in Tennessee (swapping ecomode in for species)
adonis(bray.tn.skinonly3.dat.indicator.otus~tn.ecomode*tn.oo.status, data=tn.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#betadisper to determine homogeneity of ecomode groups visualized by nmds 
meso.betadisper <- with(tn.skinonly3.dat.indicator, betadisper(bray.tn.skinonly3.dat.indicator.otus, tn.ecomode))
meso.betadisper
permutest(meso.betadisper)
TukeyHSD(meso.betadisper)
plot(meso.betadisper, label = F, hull = F, ellipse = T) 
boxplot(meso.betadisper)


#run NMDS and plot in ggplot2
#Stress value is 0.22

tn.skinonly3.dat.indicator.nmds<-metaMDS(tn.skinonly3.dat.indicator.otus, distance = "bray", k=2, try=20, trymax=3000)

data.scores3 <- as.data.frame(scores(tn.skinonly3.dat.indicator.nmds))  #Using the scores function from vegan to extract the site scores (rows) and convert to a data.frame
data.scores3
data.scores3$site <- rownames(data.scores3)  # create a column of site names, from the rownames of data.scores
data.scores3$Species <- tn.species #  add factor data for graphing purposes 
data.scores3$Month <- tn.month #  add factor data for graphing purposes 
data.scores3$Ecoregion <- tn.ecoregion #  add factor data for graphing purposes 
data.scores3$Ecomode <- tn.ecomode #  add factor data for graphing purposes 
head(data.scores3)  #look at the data
tail(data.scores3)

species.scores3 <- as.data.frame(scores(tn.skinonly3.dat.indicator.nmds, "species"))  #Using the scores function from vegan to extract the species scores and convert to a data.frame
species.scores3$species <- rownames(species.scores3)  # create a column of species, from the rownames of species.scores
head(species.scores3) 

meso<-ggplot() + geom_text(data=species.scores3,aes(x=NMDS1,y=NMDS2,label=species),alpha=0.5)  # add the species labels
meso<-ggplot()+geom_point(data=data.scores3,aes(x=NMDS1,y=NMDS2, colour=Ecomode),size=2) # add the point markers
meso<-meso+theme_classic()
meso<-meso+theme(legend.position="none") #remove this line if you want a legend - did this for multipanel figure
meso<-meso+ggtitle("Mesoscale - Tennessee Ecoregion") +
  theme(plot.title = element_text(hjust = 0.5))
meso

#############################
####DISEASE - MESOSCALE######
#Using indictors for Ophidiomyces state create dataframe of only skin samples at the MACROSCALE
#############################

meso.sfd <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu01032, Otu00206, Otu00684, Otu00149, Otu00597, Otu00022, Otu00571, Otu00124, Otu00452, Otu00316,Otu00162,Otu00125,Otu00066,Otu00075,Otu00258,Otu00021,Otu00006)
meso2.sfd <- subset(meso.sfd, sample.source=="skin")
meso3.sfd <- subset(meso2.sfd, state=="TN")
meso3.sfd.otus <- select(meso3.sfd, Otu01032, Otu00206, Otu00684, Otu00149, Otu00597, Otu00022, Otu00571, Otu00124, Otu00452, Otu00316,Otu00162,Otu00125,Otu00066,Otu00075,Otu00258,Otu00021,Otu00006)

#remove rows that have OTU relative abundances that sum to zero
meso3.sfd.removezeros<-meso3.sfd[as.logical(rowSums(meso3.sfd.otus!= 0)),]

#dbRDA using capscale function in vegan to model OTU data on continuous variable of Ophidiomyces copy number

plotNormalHistogram(meso3.sfd$copy.number) #Data are right skewed
meso.copy.number.sfd.cubed <- sign(meso3.sfd $copy.number) * abs(meso3.sfd $copy.number)^(1/3) #cubed root transform
plotNormalHistogram(meso.copy.number.sfd.cubed) #plot - a little better but still skewed

plotNormalHistogram(meso3.sfd.otus) #data are right skewed
meso3.sfd.otus.cubed <- sign(meso3.sfd.otus) * abs(meso3.sfd.otus)^(1/3) #cubed root transform
plotNormalHistogram(meso3.sfd.otus.cubed) #plot - a little better but still skewed

meso.sfd.capscale <- capscale(meso3.sfd.otus.cubed ~ meso.copy.number.sfd.cubed, meso3.sfd, dist="bray")
meso.sfd.capscale
plot(meso.sfd.capscale) #uncertain why OTU labels aren't printing onto the plot

anova(meso.sfd.capscale)

#DISEASE- MESOSCALE - plot capscale results with ggplot2

meso.smry <- summary(meso.sfd.capscale)
meso.smry
meso.df1  <- data.frame(meso.smry$sites[,1:2])       # CAP1 and MDS1
meso.oo <- as.factor(meso3.sfd$Oo) #create factor to graph
meso.df1$Ophidiomyces <- meso.oo
meso.df2  <- data.frame(meso.smry$species[,1:2])     # loadings for CAP1 and MDS1
meso.df1
meso.df2
meso.cap.plot <- ggplot(meso.df1, aes(x=CAP1, y=MDS1, colour=Ophidiomyces)) + 
  geom_point(aes(label=rownames(meso.df1)),size=2) +
  geom_hline(yintercept=0, linetype="dotted") +
  geom_vline(xintercept=0, linetype="dotted") +
  xlim(-3.3,4)+
  ylim(-2,3)
  #coord_fixed()
meso.cap.plot

meso.cap.biplot <- meso.cap.plot +
   geom_text(data=meso.df2, 
            aes(x=CAP1,y=MDS1,label=rownames(meso.df2),
                hjust=0.5*(1-sign(CAP1)),vjust=0.5*(1-sign(MDS1))), 
            color="black", size=2)
meso.cap.biplot<-meso.cap.biplot+theme_classic()
meso.cap.biplot<-meso.cap.biplot+theme(legend.position="none") #remove this line if you want a legend - did this for multipanel figure
meso.cap.biplot<-meso.cap.biplot+ggtitle("Mesoscale - Tennessee Ecoregion") +
  theme(plot.title = element_text(hjust = 0.5))
meso.cap.biplot

#######################################
#MESOSCALE - assessing ALPHA DIVERSITY
#Using indicator taxa for skin 
#######################################

#create dataframe of only skin samples for Tennessee
tn.skinonly.dat.indicator <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
tn.skinonly2.dat.indicator <- subset(tn.skinonly.dat.indicator, sample.source=="skin")
tn.skinonly3.dat.indicator <- subset(tn.skinonly2.dat.indicator, state=="TN")
tn.skinonly3.dat.indicator.otus <- select(tn.skinonly3.dat.indicator, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)

meso.invsimp <- diversity(tn.skinonly3.dat.indicator.otus, "inv")
hist(meso.invsimp, main="Inverse Simpson", xlab="", breaks=10) #visualize distribution
shapiro.test(meso.invsimp) #test for normalcy - not normally distributed

#test alpha for SPECIES - MESOSCALE
kruskal.test(meso.invsimp ~ species, data=tn.skinonly3.dat.indicator) #species is statistically significant
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(meso.invsimp ~ species, data=tn.skinonly3.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

#test alpha for ECOMODE - MESOSCALE
kruskal.test(meso.invsimp ~ ecomode, data=tn.skinonly3.dat.indicator) #ecomode is statistically significant
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(meso.invsimp ~ ecomode, data=tn.skinonly3.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

#test alpha for Ophidiomyces presence - MESOSCALE
kruskal.test(meso.invsimp ~ Oo, data=tn.skinonly3.dat.indicator) #Ophidiomyces presence is NOT statistically significant
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(meso.invsimp ~ Oo, data=tn.skinonly3.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

################################################################################################
#MICROSCALE COMPARISONS ACROSS THE INTERIOR PLATEAU ECOREGION OF TENNESSEE######################
#Using indicator OTUs###########################################################################
################################################################################################

#create dataframe of only skin samples for Interior plateau of Tennessee
interior.skinonly.dat.indicator <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
interior.skinonly2.dat.indicator <- subset(interior.skinonly.dat.indicator, sample.source=="skin")
interior.skinonly3.dat.indicator <- subset(interior.skinonly2.dat.indicator, tn.ecoregion=="Interior plateau")
interior.skinonly3.dat.indicator.otus <- select(interior.skinonly3.dat.indicator, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)

#create factors for Interior plateau skin analysis
interior.species <- as.factor(interior.skinonly3.dat.indicator$species)
interior.site <- as.factor(interior.skinonly3.dat.indicator$site)
interior.month <- as.factor(interior.skinonly3.dat.indicator$month)
interior.year <- as.factor(interior.skinonly3.dat.indicator$year)
interior.ecomode <- as.factor(interior.skinonly3.dat.indicator$ecomode)
interior.oo <- as.factor(interior.skinonly3.dat.indicator$Oo)

#permanova to test factors species*site*month*year*Oo status are predictive of the skin microbiome in the Interior plateau of Tennessee
bray.interior.skinonly3.dat.indicator.otus <- vegdist(interior.skinonly3.dat.indicator.otus, method="bray",na.rm=TRUE)
adonis(bray.interior.skinonly3.dat.indicator.otus~interior.species*interior.month*interior.site*interior.oo*interior.year, data=interior.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#Remove non significant interaction terms
adonis(bray.interior.skinonly3.dat.indicator.otus~interior.species+interior.month+interior.site+interior.oo+interior.year+interior.species:interior.month+interior.month:interior.oo, data=interior.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#Remove non significant interaction terms
adonis(bray.interior.skinonly3.dat.indicator.otus~interior.species+interior.month+interior.site+interior.oo+interior.year+interior.species:interior.month, data=interior.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#Ecomode * Oo.status
adonis(bray.interior.skinonly3.dat.indicator.otus~interior.ecomode*interior.oo, data=interior.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#betadisper to determine homogeneity of ecomode groups visualized by nmds 
interior.betadisper <- with(interior.skinonly3.dat.indicator, betadisper(bray.interior.skinonly3.dat.indicator.otus, interior.ecomode))
interior.betadisper
permutest(interior.betadisper)
TukeyHSD(interior.betadisper)
plot(interior.betadisper, label = F, hull = F, ellipse = T)
boxplot(interior.betadisper)


#run NMDS and plot in ggplot2
#Stress value is 0.20

interior.skinonly3.dat.indicator.nmds<-metaMDS(interior.skinonly3.dat.indicator.otus, distance = "bray", k=2, try=20, trymax=999)

data.scores4 <- as.data.frame(scores(interior.skinonly3.dat.indicator.nmds))  #Using the scores function from vegan to extract the site scores (rows) and convert to a data.frame
data.scores4
data.scores4$site <- rownames(data.scores4)  # create a column of site names, from the rownames of data.scores
data.scores4$Species <- interior.species #  add factor data for graphing purposes 
data.scores4$Month <- interior.month #  add factor data for graphing purposes 
data.scores4$Ecomode <- interior.ecomode #  add factor data for graphing purposes 
head(data.scores4)  #look at the data
tail(data.scores4)

species.scores4 <- as.data.frame(scores(interior.skinonly3.dat.indicator.nmds, "species"))  #Using the scores function from vegan to extract the species scores and convert to a data.frame
species.scores4$species <- rownames(species.scores4)  # create a column of species, from the rownames of species.scores
head(species.scores4) 

#ggplot
interior<-ggplot() + geom_text(data=species.scores4,aes(x=NMDS1,y=NMDS2,label=species),alpha=0.5)  # add the species labels
interior<-ggplot()+geom_point(data=data.scores4,aes(x=NMDS1,y=NMDS2, colour=Ecomode),size=2) # add the point markers
interior<-interior+theme_classic()
interior<-interior+theme_classic()
interior<-interior+theme(legend.position="none") #remove this line if you want a legend - did this for multipanel figure
interior<-interior+ggtitle("Microscale - Interior Plateau") +
  theme(plot.title = element_text(hjust = 0.5))
interior

#######################################
####DISEASE - INTERIOR PLATEAU, TN#####
#Using indictors for Ophidiomyces state create dataframe of only skin samples for INTERIOR PLATEAU of Tennessee
######################################

interior.sfd <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00190,Otu00071,Otu00354,Otu00597,Otu00045,Otu00233,Otu00452,Otu00039,Otu00026,Otu00316,Otu00258,Otu00006)
interior2.sfd <- subset(interior.sfd, sample.source=="skin")
interior3.sfd <- subset(interior2.sfd, tn.ecoregion=="Interior plateau")
interior3.sfd.otus <- select(interior3.sfd, Otu00190,Otu00071,Otu00354,Otu00597,Otu00045,Otu00233,Otu00452,Otu00039,Otu00026,Otu00316,Otu00258,Otu00006)

interior3.sfd.otus.removezeros<-interior3.sfd[as.logical(rowSums(interior3.sfd.otus != 0)), ]
interior3.sfd.otus.removezeros.otus<-interior3.sfd.otus.removezeros[13:24]

#dbRDA using capscale function in vegan to model OTU data on continuous Ophidiomyces copy number variable

plotNormalHistogram(interior3.sfd.otus.removezeros$copy.number) #Data are right skewed
interior3.sfd.copy.number.cubed <- sign(interior3.sfd.otus.removezeros$copy.number) * abs(interior3.sfd.otus.removezeros$copy.number)^(1/3) #cubed root transform
plotNormalHistogram(interior3.sfd.copy.number.cubed) #plot - a little better than square root transformation

plotNormalHistogram(interior3.sfd.otus.removezeros.otus) #data is right skewed
interior3.sfd.otus.cubed <- sign(interior3.sfd.otus.removezeros.otus) * abs(interior3.sfd.otus.removezeros.otus)^(1/3) #cubed root transform
plotNormalHistogram(interior3.sfd.otus.cubed)

interior.sfd.capscale <- capscale(interior3.sfd.otus.cubed ~ interior3.sfd.copy.number.cubed, interior3.sfd.otus.removezeros, dist="bray")
interior.sfd.capscale
plot(interior.sfd.capscale)

anova(interior.sfd.capscale)

#DISEASE- INTERIOR - plot capscale results with ggplot2

smry <- summary(interior.sfd.capscale)
smry
df1  <- data.frame(smry$sites[,1:2])       # CAP1 and MDS1
interior.oo <- as.factor(interior3.sfd.otus.removezeros$Oo)
df1$Ophidiomyces <- interior.oo
df2  <- data.frame(smry$species[,1:2])     # loadings for CAP1 and MDS1
df1
df2
cap.plot <- ggplot(df1, aes(x=CAP1, y=MDS1, colour=Ophidiomyces)) + 
  geom_point(aes(label=rownames(df1)),size=2) +
  geom_hline(yintercept=0, linetype="dotted") +
  geom_vline(xintercept=0, linetype="dotted") +
  xlim(-3,4)+
  ylim(-2,3)
  #coord_fixed()
cap.plot

cap.biplot <- cap.plot +
  geom_text(data=df2, 
            aes(x=CAP1,y=MDS1,label=rownames(df2),
                hjust=0.5*(2-sign(CAP1)),vjust=0.5*(1-sign(MDS1))), 
            color="black", size=2)
cap.biplot<-cap.biplot+theme_classic()
cap.biplot<-cap.biplot+theme(legend.position="none") #remove this line if you want a legend - did this for multipanel figure
cap.biplot<-cap.biplot+ggtitle("Microscale - Interior Plateau") +
  theme(plot.title = element_text(hjust = 0.5))
cap.biplot

#######################################
#MICROSCALE INTERIOR PLATEAU, TN - assessing ALPHA DIVERSITY
#Using indicator taxa for skin 
#######################################

#create dataframe of only skin samples for Interior plateau of Tennessee
interior.skinonly.dat.indicator <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
interior.skinonly2.dat.indicator <- subset(interior.skinonly.dat.indicator, sample.source=="skin")
interior.skinonly3.dat.indicator <- subset(interior.skinonly2.dat.indicator, tn.ecoregion=="Interior plateau")
interior.skinonly3.dat.indicator.otus <- select(interior.skinonly3.dat.indicator, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)

interior.invsimp <- diversity(interior.skinonly3.dat.indicator.otus, "inv")
hist(interior.invsimp, main="Inverse Simpson", xlab="", breaks=10) #visualize distribution
shapiro.test(interior.invsimp) #test for normalcy - normally distributed so run anova or t-test

#test alpha for SPECIES - INTERIOR
kruskal.test(interior.invsimp ~ species, data=interior.skinonly3.dat.indicator) #Ophidiomyces presence is NOT statistically significant
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(interior.invsimp ~ species, data=interior.skinonly3.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

#test alpha for ECOMODE - INTERIOR
kruskal.test(interior.invsimp ~ ecomode, data=interior.skinonly3.dat.indicator) #Ophidiomyces presence is NOT statistically significant
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(interior.invsimp ~ ecomode, data=interior.skinonly3.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

#test alpha for Ophidiomyces presence - INTERIOR
kruskal.test(interior.invsimp ~ Oo, data=interior.skinonly3.dat.indicator) #Ophidiomyces presence is NOT statistically significant
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(interior.invsimp ~ Oo, data=interior.skinonly3.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

################################################################################################
#MICROSCALE COMPARISONS ACROSS THE Southwestern Appalachians ECOREGION OF TENNESSEE#############
#Using indicator OTUs###########################################################################
################################################################################################

#create dataframe of only skin samples for Southwestern Appalachians of Tennessee
apps.skinonly.dat.indicator <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
apps.skinonly2.dat.indicator <- subset(apps.skinonly.dat.indicator, sample.source=="skin")
apps.skinonly3.dat.indicator <- subset(apps.skinonly2.dat.indicator, tn.ecoregion=="Southwestern Appalachians")
apps.skinonly3.dat.indicator.otus <- select(apps.skinonly3.dat.indicator, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)

#create factors for Southwestern Appalachians skin analysis
apps.species <- as.factor(apps.skinonly3.dat.indicator$species)
apps.site <- as.factor(apps.skinonly3.dat.indicator$site)
apps.month <- as.factor(apps.skinonly3.dat.indicator$month)
apps.year <- as.factor(apps.skinonly3.dat.indicator$year)
apps.ecomode <- as.factor(apps.skinonly3.dat.indicator$ecomode)
apps.oo <- as.factor(apps.skinonly3.dat.indicator$Oo)

#permanova to test factors species*site*month*year*Oo status are predictive of the skin microbiome in the Southwestern Appalachians of Tennessee
bray.apps.skinonly3.dat.indicator.otus <- vegdist(apps.skinonly3.dat.indicator.otus, method="bray",na.rm=TRUE)
adonis(bray.apps.skinonly3.dat.indicator.otus~apps.species*apps.month*apps.site*apps.year*apps.oo, data=apps.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#Remove ALL interaction tests 
adonis(bray.apps.skinonly3.dat.indicator.otus~apps.species+apps.month+apps.site+apps.year+apps.oo, data=apps.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#ecomode*Oo.status
adonis(bray.apps.skinonly3.dat.indicator.otus~apps.ecomode*apps.oo, data=apps.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#ecomode+Oo.status - no significant interaction between ecomode and oo status
adonis(bray.apps.skinonly3.dat.indicator.otus~apps.ecomode+apps.oo, data=apps.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#betadisper to determine homogeneity of ecomode groups visualized by nmds 
#only two points for Arboreal factor - unbalanced sampling design skews the results - shouldn't be used in conclusions
apps.betadisper <- with(apps.skinonly3.dat.indicator, betadisper(bray.apps.skinonly3.dat.indicator.otus, apps.ecomode))
apps.betadisper
permutest(apps.betadisper)
plot(apps.betadisper, label = F, hull = F, ellipse = T)
boxplot(apps.betadisper)
TukeyHSD(apps.betadisper)

#run NMDS with ecomode as factor and plot in ggplot2
#Stress value is 0.18

apps.skinonly3.dat.indicator.nmds<-metaMDS(apps.skinonly3.dat.indicator.otus, distance = "bray", k=2, try=20, trymax=999)

data.scores5 <- as.data.frame(scores(apps.skinonly3.dat.indicator.nmds))  #Using the scores function from vegan to extract the site scores (rows) and convert to a data.frame
data.scores5
data.scores5$site <- rownames(data.scores5)  # create a column of site names, from the rownames of data.scores
data.scores5$Species <- apps.species #  add factor data for graphing purposes 
data.scores5$Month <- apps.month #  add factor data for graphing purposes 
data.scores5$Ecomode <- apps.ecomode #  add factor data for graphing purposes 
head(data.scores5)  #look at the data
tail(data.scores5)

species.scores5 <- as.data.frame(scores(apps.skinonly3.dat.indicator.nmds, "species"))  #Using the scores function from vegan to extract the species scores and convert to a data.frame
species.scores5$species <- rownames(species.scores5)  # create a column of species, from the rownames of species.scores
head(species.scores5) 

apps<-ggplot() + geom_text(data=species.scores5,aes(x=NMDS1,y=NMDS2,label=species),alpha=0.5)  # add the species labels
apps<-ggplot()+geom_point(data=data.scores5,aes(x=NMDS1,y=NMDS2, colour=Ecomode),size=2) # add the point markers
apps<-apps+theme_classic()
apps<-apps+theme(legend.position="none") #remove this line if you want a legend - did this for multipanel figure
apps<-apps+ggtitle("Microscale - Southwestern Appalachians") +
  theme(plot.title = element_text(hjust = 0.5))
apps

################################################
####DISEASE - SOUTHWESTERN APPALACHIANS, TN#####
#Using indictors for Ophidiomyces state create dataframe of only skin samples for Southwestern Appalachians of Tennessee
###############################################

apps.skinonly.dat.indicator.sfd <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00554,Otu01591,Otu14287,Otu00035,Otu10700,Otu26160,Otu00467,Otu00258,Otu00193,Otu00146,Otu00111,Otu00108)
apps.skinonly2.dat.indicator.sfd <- subset(apps.skinonly.dat.indicator.sfd, sample.source=="skin")
apps.skinonly3.dat.indicator.sfd <- subset(apps.skinonly2.dat.indicator.sfd, tn.ecoregion=="Southwestern Appalachians")
apps.skinonly3.dat.indicator.otus.sfd <- select(apps.skinonly3.dat.indicator.sfd, Otu00554,Otu01591,Otu14287,Otu00035,Otu10700,Otu26160,Otu00467,Otu00258,Otu00193,Otu00146,Otu00111,Otu00108)

#remove rows that have OTU relative abundances that sum to zero
apps.skinonly3.dat.indicator.sfd.removezeros<-apps.skinonly3.dat.indicator.sfd[as.logical(rowSums(apps.skinonly3.dat.indicator.otus.sfd != 0)), ]
apps.skinonly3.dat.indicator.otus.sfd <- select(apps.skinonly3.dat.indicator.sfd.removezeros, Otu00554,Otu01591,Otu14287,Otu00035,Otu10700,Otu26160,Otu00467,Otu00258,Otu00193,Otu00146,Otu00111,Otu00108)

#dbRDA using capscale function in vegan to model OTU data on continuous Ophidiomyces copy number variable

copy.number.sfd.cubed <- sign(apps.skinonly3.dat.indicator.sfd.removezeros$copy.number) * abs(apps.skinonly3.dat.indicator.sfd.removezeros$copy.number)^(1/3) #cubed root transform
plotNormalHistogram(copy.number.sfd.cubed) #plot - a little better than square root transformation

plotNormalHistogram(apps.skinonly3.dat.indicator.otus.sfd) #data is right skewed
apps.skinonly3.dat.indicator.otus.sfd.cubed <- sign(apps.skinonly3.dat.indicator.otus.sfd) * abs(apps.skinonly3.dat.indicator.otus.sfd)^(1/3) #cubed root transform
plotNormalHistogram(apps.skinonly3.dat.indicator.otus.sfd.cubed)

apps.sfd.capscale <- capscale(apps.skinonly3.dat.indicator.otus.sfd.cubed ~ copy.number.sfd.cubed, apps.skinonly3.dat.indicator.sfd.removezeros, dist="bray" )
apps.sfd.capscale
plot(apps.sfd.capscale)

anova(apps.sfd.capscale)

#DISEASE- SOUTHWESTERN APPALACHIANS - plot capscale results with ggplot2

apps.smry <- summary(apps.sfd.capscale)
apps.smry
apps.df1  <- data.frame(apps.smry$sites[,1:2])       # CAP1 and MDS1
apps.oo <- as.factor(apps.skinonly3.dat.indicator.sfd.removezeros$Oo) #create factor to graph
apps.df1$Ophidiomyces <- apps.oo
apps.df2  <- data.frame(apps.smry$species[,1:2])     # loadings for CAP1 and MDS1
apps.df1
apps.df2
apps.cap.plot <- ggplot(apps.df1, aes(x=CAP1, y=MDS1, colour=Ophidiomyces)) + 
  geom_point(aes(label=rownames(apps.df1)),size=2) +
  geom_hline(yintercept=0, linetype="dotted") +
  geom_vline(xintercept=0, linetype="dotted") +
  xlim(-3.5,4.5)+
  ylim(-2,3)
  #coord_fixed()
apps.cap.plot

apps.cap.biplot <- apps.cap.plot +
  geom_text(data=apps.df2, 
            aes(x=CAP1,y=MDS1,label=rownames(apps.df2),
                hjust=0.5*(1-sign(CAP1)),vjust=0.5*(1-sign(MDS1))), 
            color="black", size=2)
apps.cap.biplot<-apps.cap.biplot+theme_classic()
apps.cap.biplot<-apps.cap.biplot+theme(legend.position="none") #remove this line if you want a legend - did this for multipanel figure
apps.cap.biplot<-apps.cap.biplot+ggtitle("Microscale - Southwestern Appalachians") +
  theme(plot.title = element_text(hjust = 0.5))
apps.cap.biplot

#######################################
#MICROSCALE SOUTHWESTERN APPALACHIANS, TN - assessing ALPHA DIVERSITY
#Using indicator taxa for skin 
#######################################

#create dataframe of only skin samples for Southwestern Appalachians of Tennessee
apps.skinonly.dat.indicator <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
apps.skinonly2.dat.indicator <- subset(apps.skinonly.dat.indicator, sample.source=="skin")
apps.skinonly3.dat.indicator <- subset(apps.skinonly2.dat.indicator, tn.ecoregion=="Southwestern Appalachians")
apps.skinonly3.dat.indicator.otus <- select(apps.skinonly3.dat.indicator, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)

apps.invsimp <- diversity(apps.skinonly3.dat.indicator.otus, "inv")
hist(apps.invsimp, main="Inverse Simpson", xlab="", breaks=10) #visualize distribution
shapiro.test(apps.invsimp) #test for normalcy - normally distributed so run anova or t-test
apps.invsimp

#test alpha for SPECIES - SOUTHWESTERN APPALACHIANS, TN
kruskal.test(apps.invsimp ~ species, data=apps.skinonly3.dat.indicator) 
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(apps.invsimp ~ species, data=apps.skinonly3.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

#test alpha for ECOMODE - SOUTHWESTERN APPALACHIANS, TN
kruskal.test(apps.invsimp ~ ecomode, data=apps.skinonly3.dat.indicator) 
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(apps.invsimp ~ ecomode, data=apps.skinonly3.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

#test alpha for Ophidiomyces presence - SOUTHWESTERN APPALACHIANS, TN

kruskal.test(apps.invsimp ~ Oo, data=apps.skinonly3.dat.indicator) 
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(apps.invsimp ~ Oo, data=apps.skinonly3.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

################################################################################################
#MICROSCALE COMPARISONS ACROSS THE Blue Ridge Mountains ECOREGION OF TENNESSEE#############
#Using indicator OTUs###########################################################################
################################################################################################

#create dataframe of only skin samples for Blue Ridge Mountains of Tennessee
blue.skinonly.dat.indicator <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
blue.skinonly2.dat.indicator <- subset(blue.skinonly.dat.indicator, sample.source=="skin")
blue.skinonly3.dat.indicator <- subset(blue.skinonly2.dat.indicator, tn.ecoregion=="Blue Ridge Mountains")
blue.skinonly3.dat.indicator.otus <- select(blue.skinonly3.dat.indicator, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)

#create factors for Blue Ridge Mountains skin analysis
blue.species <- as.factor(blue.skinonly3.dat.indicator$species)
blue.site <- as.factor(blue.skinonly3.dat.indicator$site)
blue.month <- as.factor(blue.skinonly3.dat.indicator$month)
blue.year <- as.factor(blue.skinonly3.dat.indicator$year)
blue.ecomode <- as.factor(blue.skinonly3.dat.indicator$ecomode)
blue.oo <- as.factor(blue.skinonly3.dat.indicator$Oo) #only a single record of Oo in this ecoregion. Do not model for disease as in other spatial comparisons
blue.oo

#permanova to test factors species*site*month*year are predictive of the skin microbiome in the Blue Ridge Mountains of Tennessee
bray.blue.skinonly3.dat.indicator.otus <- vegdist(blue.skinonly3.dat.indicator.otus, method="bray",na.rm=TRUE)
adonis(bray.blue.skinonly3.dat.indicator.otus~blue.species*blue.month*blue.site*blue.year, data=blue.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#no significant interaction between any factors - drop from the model and run independently
adonis(bray.blue.skinonly3.dat.indicator.otus~blue.species+blue.month+blue.site+blue.year, data=blue.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#permanova to test factors ecomode*site*month are predictive of the skin microbiome in the Blue Ridge Mountains of Tennessee (swapping ecomode for species)
adonis(bray.blue.skinonly3.dat.indicator.otus~blue.ecomode*blue.month*blue.site, data=blue.skinonly3.dat.indicator.otus, permutations = 999, method='bray')
#no significant interactions - run ecomode by itself

#ecomode Blue Ridge Mountains
adonis(bray.blue.skinonly3.dat.indicator.otus~blue.ecomode, data=blue.skinonly3.dat.indicator.otus, permutations = 999, method='bray')

#run NMDS and plot in ggplot2
#Stress value is 0.15

blue.skinonly3.dat.indicator.nmds<-metaMDS(blue.skinonly3.dat.indicator.otus, distance = "bray", k=2, try=20, trymax=999)

data.scores6 <- as.data.frame(scores(blue.skinonly3.dat.indicator.nmds))  #Using the scores function from vegan to extract the site scores (rows) and convert to a data.frame
data.scores6
data.scores6$site <- rownames(data.scores6)  # create a column of site names, from the rownames of data.scores
data.scores6$Species <- blue.species #  add factor data for graphing purposes 
data.scores6$Month <- blue.month #  add factor data for graphing purposes 
data.scores6$Ecomode <- blue.ecomode #  add factor data for graphing purposes 
head(data.scores6)  #look at the data
tail(data.scores6)

species.scores6 <- as.data.frame(scores(blue.skinonly3.dat.indicator.nmds, "species"))  #Using the scores function from vegan to extract the species scores and convert to a data.frame
species.scores6$species <- rownames(species.scores6)  # create a column of species, from the rownames of species.scores
head(species.scores6) 

#ggplot
blue<-ggplot() + geom_text(data=species.scores6,aes(x=NMDS1,y=NMDS2,label=species),alpha=0.5)  # add the species labels
blue<-ggplot()+geom_point(data=data.scores6,aes(x=NMDS1,y=NMDS2, colour=Ecomode),size=2) # add the point markers
blue<-blue+theme_classic()
blue<-blue+ggtitle("Microscale - Blue Ridge Mtns") +
  theme(plot.title = element_text(hjust = 0.5))
blue

#######################################
#MICROSCALE - BLUE RIDGE MOUNTAINS, TN - assessing ALPHA DIVERSITY
#Using indicator taxa for skin 
#######################################

#create dataframe of only skin samples for Blue Ridge Mountains of Tennessee
blue.skinonly.dat.indicator <- select(env.skin.dat, sample, species, sample.source, Oo, state, site, tn.ecoregion, month, year, ct, copy.number, ecomode, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)
blue.skinonly2.dat.indicator <- subset(blue.skinonly.dat.indicator, sample.source=="skin")
blue.skinonly3.dat.indicator <- subset(blue.skinonly2.dat.indicator, tn.ecoregion=="Blue Ridge Mountains")
blue.skinonly3.dat.indicator.otus <- select(blue.skinonly3.dat.indicator, Otu00169,	Otu00318,	Otu00557,	Otu00279,	Otu01239,	Otu00355,	Otu01976,	Otu00192,	Otu00278,	Otu00058,	Otu00467,	Otu00142,	Otu00097,	Otu00149,	Otu00094,	Otu00573,	Otu00178,	Otu00259,	Otu00073,	Otu00193,	Otu00116,	Otu00160,	Otu00098,	Otu00571,	Otu00389,	Otu00022,	Otu00066,	Otu00124,	Otu00258,	Otu00043,	Otu00226,	Otu00309,	Otu00105,	Otu00089,	Otu00075,	Otu00392,	Otu00191,	Otu00083,	Otu00235,	Otu00049,	Otu00125,	Otu00099,	Otu00120,	Otu00021,	Otu00146,	Otu00048,	Otu00046,	Otu00029,	Otu00041,	Otu00031,	Otu00011,	Otu00019,	Otu00012,	Otu00111,	Otu00007,	Otu00006)

blue.invsimp <- diversity(blue.skinonly3.dat.indicator.otus, "inv")
hist(blue.invsimp, main="Inverse Simpson", xlab="", breaks=10) #visualize distribution
shapiro.test(blue.invsimp) #test for normalcy - normally distributed so run anova or t-test

#test alpha for SPECIES - BLUE RIDGE MOUNTAINS, TN
kruskal.test(blue.invsimp ~ species, data=blue.skinonly3.dat.indicator) 
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(blue.invsimp ~ species, data=blue.skinonly3.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

#test alpha for ECOMODE - BLUE RIDGE MOUNTAINS, TN
kruskal.test(blue.invsimp ~ ecomode, data=blue.skinonly3.dat.indicator) 
#Create 1x1 plot environment
par(mfrow = c(1, 1))
#Plot
boxplot(blue.invsimp ~ ecomode, data=blue.skinonly3.dat.indicator, ylab="Inverse Simpson", las=2, cex.axis=0.8)

#test alpha for Ophidiomyces presence - BLUE RIDGE MOUNTAINS, TN
#Do not test - only 1 record of Ophidiomyces in this region

#multiplot ecomode nmds figure in ggplot
grid.arrange(macro1, meso, apps, interior, blue, ncol =2)

#multiplot capscale analysis figure in ggplot
grid.arrange(macro.cap.biplot, meso.cap.biplot, apps.cap.biplot, cap.biplot, nrow=2)