possiblesites=1:100                                           #possible site values
probabilities=seq(.8,.1,length.out=100)                       #vector of probabilities to use in sampling to ensure unequal replicates
site=sort(sample(possiblesites,500,replace=T,prob=probabilities))  #sample sites to make non-uniform replicates
quercus=rbinom(length(site),1,prob=0.5)                    #generate binary data to simulate species presence - absence
d=data.frame(site,quercus)                              #combine in data.frame and overwrite prdb
head(d)
                                                  

site.freq=table(d$site)					#table with site frequency

dups=names(site.freq[site.freq>1])             #sites with duplicates
loners=names(site.freq[site.freq==1])          #sites with no duplicates
d.new=d[d$site%in%loners,]            #create new table with sites that only have one plot
d.new$repID=NA                              #create new field to hold which replicate was selected
for(i in 1:length(dups)){                      #start the for loop for each element of dup
  code=dups[i]                                #set code=name of site for this iteration
  x=d[d$site==code,]                                     #make x a new table subsetted to site of interest
  presence=x$quercus                       #get vector of presence/absence. 
  if (sum(presence)==0) select=sample(1:nrow(x),1)               #if all zeros, choose one
  if (sum(presence)>0) select=sample(which(x$quercus==1),1)   #if some ones, choose a one
  d.new=rbind(d.new,c(x[select,],repID=select))            #add to new table with ID of which replicate chosen
}

length(unique(d$site))                                       #number of original sites

cbind(table(table(d$site)))                                  #number of old sites and the number of times they occur
table(table(d.new$site))                                     #number of new sites and the number of times they occur