R/count_taxa.R

In helixcn/plantlist: Looking Up the Status of Plant Scientific Names based on The Plant List Database, Searching the Chinese Names and Making checklists of plants

#' Get the number of taxa under each taxonomic level for a dataframe generated
#' by the function CTPL
#'
#' Compute the number of species under each family, each genus, and each group.
#' See the value secion.
#'
#' Simple function summarizing the number of taxa under each taxonomic
#' level.
#'
#' @param checklist_dat A data.frame generated by CTPL or in the same format.
#' @return \item{NO_OF_FAMILIES}{Number of families in the data.frame}
#'
#' \item{NO_OF_GENERA}{Number of genera in the data.frame}
#'
#' \item{NO_OF_SPECIES}{Number of species in the data.frame}
#'
#' \item{NO_OF_FAMILIES_BY_GROUP}{Number of families under each group of higher
#' plants, namely, Moss, Ferns and Lycophytes, Gymnosperms, Angiosperms}
#'
#' \item{NO_OF_GENERA_BY_GROUP }{Number of genera under each group, namely,
#' Moss, Ferns and lycophytes, Gymnosperms, Angiosperms}
#'
#' \item{NO_OF_SPECIES_BY_GROUP }{Number of species under each group, namely,
#' Moss, Ferns and lycophytes, Gymnosperms, Angiosperms}
#'
#' \item{NO_OF_GENERA_BY_FAMILY }{The name is self-explanatory}
#'
#' \item{NO_OF_SPECIES_BY_FAMILY}{The name is self-explanatory}
#'
#' \item{NO_OF_SPECIES_BY_GENUS }{The name is self-explanatory}
#' @author Jinlong Zhang
#' @seealso \code{\link{make_checklist}}, \code{\link{CTPL}}
#' @references None
#' @examples
#'
#' ## Do not run
#' ## See the vignette
#'
#' @export count_taxa
count_taxa <-
  function(checklist_dat) {
    checklist_dat <-
      subset(
        checklist_dat,
        select = c(
          "YOUR_SEARCH",
          "GROUP",
          "SPECIES_FULL",
          "FAMILY_NUMBER",
          "FAMILY",
          "GENUS",
          "SPECIES"
        )
      )
    if (any(is.na(checklist_dat))) {
      warning(paste(
        "Taxa: '",
        paste(checklist_dat$YOUR_SEARCH[is.na(checklist_dat$SPECIES)],
              collapse = ", "),
        "' do(es) not have full scientific name, ignored\n",
        sep = ""
      ))
    }
    
    checklist_dat <- stats::na.omit(checklist_dat)
    
    # Add a number to each group
    checklist_dat$GROUP <-
      ifelse(
        checklist_dat$GROUP == "Bryophytes",
        paste("1", checklist_dat$GROUP),
        checklist_dat$GROUP
      )
    
    checklist_dat$GROUP <-
      ifelse(
        checklist_dat$GROUP == "Ferns and lycophytes",
        paste("2", checklist_dat$GROUP),
        checklist_dat$GROUP
      )
    checklist_dat$GROUP <- ifelse(is.na(checklist_dat$GROUP),
                                  "", checklist_dat$GROUP)
    checklist_dat$GROUP <-
      ifelse(
        checklist_dat$GROUP == "Gymnosperms",
        paste("3", checklist_dat$GROUP),
        checklist_dat$GROUP
      )
    checklist_dat$GROUP <-
      ifelse(
        checklist_dat$GROUP == "Angiosperms",
        paste("4", checklist_dat$GROUP),
        checklist_dat$GROUP
      )
    
    checklist_dat$FAMILY_NUMBER <-
      gsub("[^0-9]", "", checklist_dat$FAMILY_NUMBER)
    
    # Sorting by GROUP, FAMILY_NUMBER and SPECIES_FULL
    checklist_dat[is.na(checklist_dat)] <- ""
    
    checklist_dat3 <- checklist_dat[order(checklist_dat$GROUP,
                                          checklist_dat$FAMILY_NUMBER,
                                          checklist_dat$SPECIES_FULL),]
    
    no_genera_by_family_dat <-
      unique(data.frame(GENUS = checklist_dat3$GENUS, 
                        FAMILY = checklist_dat3$FAMILY))
    no_genera_by_family <-
      stats::aggregate(no_genera_by_family_dat$GENUS,
                list(no_genera_by_family_dat$FAMILY),
                "length")
    colnames(no_genera_by_family) <- c("family", "no_of_genera")
    
    no_species_by_family_dat <-
      unique(data.frame(SPECIES = checklist_dat3$SPECIES, 
                        FAMILY = checklist_dat3$FAMILY))
    no_species_by_family <-
      stats::aggregate(
        no_species_by_family_dat$SPECIES,
        list(no_species_by_family_dat$FAMILY),
        "length"
      )
    colnames(no_species_by_family) <-
      c("family", "no_of_species")
    
    no_genera_by_group_dat <-
      unique(data.frame(GENUS = checklist_dat3$GENUS, 
                        GROUP = checklist_dat3$GROUP))
    no_genera_by_group <-
      stats::aggregate(no_genera_by_group_dat$GENUS,
                list(no_genera_by_group_dat$GROUP),
                "length")
    colnames(no_genera_by_group) <- c("group", "no_of_genera")
    
    no_species_by_group_dat <-
      unique(data.frame(SPECIES = checklist_dat3$SPECIES, 
                        GROUP = checklist_dat3$GROUP))
    no_species_by_group <-
      stats::aggregate(no_species_by_group_dat$SPECIES,
                list(no_species_by_group_dat$GROUP),
                "length")
    colnames(no_species_by_group) <- c("group", "no_of_species")
    
    no_families_by_group_dat <-
      unique(data.frame(FAMILY = checklist_dat3$FAMILY, 
                        GROUP = checklist_dat3$GROUP))
    no_families_by_group <-
      stats::aggregate(
        no_families_by_group_dat$FAMILY,
        list(no_families_by_group_dat$GROUP),
        "length"
      )
    colnames(no_families_by_group) <-
      c("group", "no_of_families")
    
    no_species_by_genus_dat <-
      unique(data.frame(SPECIES = checklist_dat3$SPECIES, 
                        GENUS = checklist_dat3$GENUS))
    no_species_by_genus <-
      stats::aggregate(no_species_by_genus_dat$SPECIES,
                list(no_species_by_genus_dat$GENUS),
                "length")
    colnames(no_species_by_genus) <- c("genus", "no_of_species")
    
    no_families <- length(unique(checklist_dat3$FAMILY))
    no_genera <- length(unique(checklist_dat3$GENUS))
    no_species <- length(unique(checklist_dat3$SPECIES))
    
    return(
      list(
        NO_OF_FAMILIES          = no_families,
        NO_OF_GENERA            = no_genera,
        NO_OF_SPECIES           = no_species,
        NO_OF_FAMILIES_BY_GROUP = no_families_by_group,
        NO_OF_GENERA_BY_GROUP   = no_genera_by_group,
        NO_OF_SPECIES_BY_GROUP  = no_species_by_group,
        NO_OF_GENERA_BY_FAMILY  = no_genera_by_family,
        NO_OF_SPECIES_BY_FAMILY = no_species_by_family,
        NO_OF_SPECIES_BY_GENUS  = no_species_by_genus
      )
    )
  }