# example code to load a data set, compute and compare distances
#  code assumes you have saved data files in your working directory
#  you can set that working directory using File/Change Dir menu
#  or setwd('XXX') where XXX is the full path using / 
#  e.g. c:/users/netid/project


library(vegan)

# read a .csv file
lipo <- read.csv('lipo.csv', as.is=T)

# or, read a data matrix with spaces
park <- read.table('parkgrass.txt', header=T, as.is=T)

# I use as.is=T to stop automatic conversion of character strings to factors

# descriptive info about the data matrix
dim(park)
names(park)
str(park)

# convert the data frame to a matrix with ONLY the abundance info
#  usually have to remove one or more columns with labels
#  there are two of these in both lipo and parkgrass
#  often there is only one, in which case, you would use only [,-1]

park.data <- park[,c(-1,-2)]

park.m <- as.matrix(park.data)

# output and plots are more informative if the data matrix has
#   appropriate row (species/compound) and column (site / sample) names

# column names are already set using the names from the data frame
# see this:
dimnames(park.m)

# add row names to the data matrix, using the appropriate column from 
#  original data frame 
dimnames(park.m)[[1]] <- park$sppname

# check if you want to make sure
dimnames(park.m)

# if the file is 'backwards', i.e. rows are species, transpose after 
#   creating the matrix

park.m <- t(park.m)

# pretty up the site names by removing the X.  probably not necessary
#  for your data matrix
sites <- dimnames(park.m)[[1]]
sites <- substring(sites,2,nchar(sites))
dimnames(park.m)[[1]] <- sites


# Now we can have fun

# calculate distance matrix

# standardize to site total
park.m <- decostand(park.m,'total')
 
# then compute Bray-Curtis distance (default for vegdist)
park.bray <- vegdist(park.m)
# can compute many other distances
park.eucl <- vegdist(park.m, 'eucl')    # Euclidean
park.mh <- vegdist(park.m, 'horn')      # Morisita-Horn

# scatter plot matrix to compare distances
#   combine columns (cbind), then plot matrix (pairs)
pairs(cbind(park.bray,park.eucl,park.mh))

# can convert the distance object to a matrix, if need to
temp <- as.matrix(park.bray)

# which makes it easy to look at parts of the distance array
temp[1:5,1:5]

# if you want to identify individual pairs of sites
plot(park.bray,park.mh, xlab='Bray-Curtis distance', ylab='Morisita-Horn distance')
temp <- identify(park.bray, park.mh)

# which.site converts the obs. number to the pair of site names
#   defined in whichsite.r

which.site(temp, park.bray)