Functions to compute Simpson's and Sorensen's betas from taxa lists from different sites

functions_betadiversity.R

#returns biodiversity components a, b, and c from two vectors with taxa names.
#x: character vector, taxa list of one site
#y: character vector, taxa list of another site
abc <- function(x, y){
  
  #list to store output
  out <- list()
  
  #filling the list
  out$a <- length(intersect(x, y))
  out$b <- length(setdiff(x, y))
  out$c <- length(setdiff(y, x))
  
  #returning the output
  out
}


#returns the Sorensen's beta from the output of abc()
sorensen_beta <- function(x){
  
  x$bsor <- round(2 * x$a / (2 * x$a + x$b + x$c), 3)
  
  x
  
}

#returns the Simpson's beta from the output of abc()
simpson_beta <- function(x){
  
  x$bsim <- round(min(x$b, x$c) / (min(x$b, x$c) + x$a), 3)
  
  x
  
}

#returns the biodiversity components a, b, and c and the betadiversity indices of Sorensen and Simpson
#x: character vector, taxa list of one site
#y: character vector, taxa list of another site
betadiversity <- function(x, y){
  
  require(magrittr)
  
  abc(x, y) %>%
    sorensen_beta() %>%
    simpson_beta()
  
}

#computes betadiversity indices for a set of sites of an arbitrary size stored in a long-format data frame
#x: data frame with at least two columns: site.column, with the site name, and taxa.column, with the taxon name. The data frame ought to have one row per site and taxon.
#site.column: character, name of the column in x with the site names.
#taxa.column: character, name of the column in x with the taxa names.
betadiversity_multisite <- function(
  x, 
  site.column, #column with site names
  taxa.column #column with taxa names
){
  
  #get site combinations
  site.combinations <- utils::combn(
    x = unique(x[, site.column]),
    m = 2
  )
  
  #iterating through site pairs
  betadiversity.df <- foreach::foreach(
    i = 1:ncol(site.combinations),
    .combine = 'rbind'
  ) %do% {
    
    #site names
    site.one <- site.combinations[1, i]
    site.two <- site.combinations[2, i]
    
    #getting taxa lists
    taxa.list.one <- x[x[, site.column] %in% site.one, taxa.column]
    taxa.list.two <- x[x[, site.column] %in% site.two, taxa.column]
    
    #betadiversity
    beta <- betadiversity(
      x = taxa.list.one,
      y = taxa.list.two
    )
    
    #adding site names
    beta$site.one <- site.one
    beta$site.two <- site.two
    
    #returning output
    beta
    
  }
  
  #remove bad rownames
  rownames(betadiversity.df) <- NULL
  
  #reordering columns
  betadiversity.df <- betadiversity.df[, c(
    "site.one",
    "site.two",
    "a",
    "b",
    "c",
    "bsor",
    "bsim"
    )]
  
  #returning output
  betadiversity.df
  
}