Function to extract accepted names and taxonomy from GBIF Backbone Taxonomy

get_gbif_taxonomy.R

#' Get accepted canonical names and taxonomy for a given species name
#'
#' @param x a character string or vector of species names.
#' @param infraspecies logical. If TRUE, the given name is resolved to infraspecies taxon, i.e. subspecies names will not be pooled (not working!)
#' @param fuzzy logical. If TRUE, function tries fuzzy matching for species requests. May produce output, if normal matching fails. (not working!)
#' @param verbose logical. If FALSE, warnings and messages are suppressed.
#'
#' @return a data.frame of accepted taxon names and higher taxa.
#' @import taxize
#' @import data.table
#' @export
#'
#' @examples
#'
#' get_gbif_taxonomy(c("Chorthippus albomarginatus", "Chorthippus apricarius", "Chorthippus biguttulus", "Chorthippus dorsatus", "Chorthippus montanus", "Chorthippus parallelus", "Chrysochraon dispar", "Conocephalus dorsalis", "Conocephalus fuscus", "Decticus verrucivorus", "Euthystira brachyptera", "Gomphocerippus rufus", "Gryllus campestris", "Metrioptera roeselii", "Omocestus viridulus", "Phaneroptera falcata", "Platycleis albopunctata", "Spec", "Stenobothrus lineatus", "Stenobothrus stigmaticus", "Stethophyma grossum", "Tetrix kraussi", "Tetrix subulata", "Tetrix tenuicornis", "Tetrix undulata", "Tettigonia cantans", "Tettigonia viridissima"))
#'

get_gbif_taxonomy <- function(x, infraspecies = FALSE, fuzzy = FALSE, verbose = TRUE) {

  if(length(x) > 1) { # recursive wrapping for vectorized input
    out <- lapply(x, get_gbif_taxonomy)
    out <- data.table::rbindlist(out, fill = TRUE) # combine into data.frame

  } else {

    # spellchecking: resolve names using data source 11 (GBIF Backbone Taxonomy)
    resolved <- taxize::gnr_resolve(x,
                                    preferred_data_sources = c(11),
                                    best_match_only = TRUE,
                                    canonical = TRUE)

    # return NA for unsuccessful matches
    if(is.null(resolved$matched_name2) ) {
      out <- data.frame(user_supplied_name = x)
      attributes(out)$warning <- paste("No matching species name found!")
    } else {

    # get gbif ID and detailled information, e.g. synonym status
    temp <- taxize::get_gbifid_(resolved$matched_name2)[[1]]

    # switch for allowing for fuzzy matching
    if(!fuzzy) temp <- subset(temp, matchtype == "EXACT")

    # eliminate infraspecies (!! Needs improvement: this will make the function remove supraspecies taxa, too!)
    if(!infraspecies) temp <- subset(temp, rank == "species") #ask for rank according to name resolution provided

    # if given name is a synonym, do a new request for the accepted species name
    if(all(temp$status == "SYNONYM")) {
      if(length(unique(temp$species)) == 1) {
        out <- get_gbif_taxonomy(unique(temp$species))
        out$synonym = TRUE
        out$user_supplied_name = x

      }
      if(length(unique(temp$species)) > 1) out <- get_gbif_taxonomy(temp$species[which.max(temp$rank)])
      out$synonym = TRUE
      out$user_supplied_name = x
    } else {  # if given name is an accepted name, return result into 'out'
      if(any(temp$status == "ACCEPTED")) {
        temp <- subset(temp, status == "ACCEPTED")
        out <- temp
        # add choice for fuzzy matching which returns warning
            }

      out <- cbind(user_supplied_name = x, synonym = FALSE,   scientificName = out$species,fullname = out$scientificname, out[,c("rank", "confidence", "kingdom",  "phylum", "class","order", "family", "genus")], taxonomy = "GBIF Backbone Taxonomy",  taxonID = out$usagekey)

      if(out$synonym & verbose) warning(paste("Synonym provided! Automatically set ScientificName to accepted species Name!"))
    }
  }

  }

  class(out) <- c("data.frame", "taxonomy")
  return(out)
}

Hi Florian, I've come across what may be another 'bug' in the function. To be fair, I'm using it on a pretty tricky list-- most of them are synonyms and many of them are old. The issue seems to be that when get_gbifid_() can't find a particular synonym and reverts it to a HIGHERRANK match, then the resolve_synonyms step, which takes place before the step that takes care of higher rank matches, doesn't have a species column to rerun taxize on... it only has a genus column. Looks like get_gbifid_() automatically throws away columns that are pointless in its output. This causes the resolve synonyms step to return an empty list, which subsequently causes the function to crash. The current tricky name I'm getting an error on is 'Dromicia frontalis.' It is supposed to be a synonym of Acrobates pygmaeus, but I would be happy if the function simply threw out this synonym, because it is not actually in the gbif taxonomy, and is just a case of someone mis-identifying an existing species as a new one in the 1800s. I will try to debug it on my end, but would appreciate any pointers! Cheers, Anikó

…

On Mon, Jun 17, 2019 at 7:35 AM Aniko Toth ***@***.***> wrote: Thanks, I’ll check it out! The doubtful synonym I tried was “Antechinus unicolour” if I remember right. On Thu, 13 Jun 2019 at 9:19 pm, Florian Schneider < ***@***.***> wrote: > Oh, sorry! I just figured I need to point out that this gist is an older > version of the function shipped with the 'traitdataform' package (including > full documentation)! That newer version should already deal with DOUBTFUL > cases. You might want to try that instead. Just install from CRAN via > install.packages('traitdataform'). The parameter options have changed a > little, but main input is still a vector of species names. > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <https://gist.github.com/69e61b14c12ccdda780fbc1c5f0a4f1c?email_source=notifications&email_token=AHD6WCE5RBKPSX4T7R7NKO3P2IULBA5CNFSM4HXFUCBKYY3PNVWWK3TUL52HS4DFVNDWS43UINXW23LFNZ2KUY3PNVWWK3TUL5UWJTQAFTTIG#gistcomment-2942595>, > or mute the thread > <https://github.com/notifications/unsubscribe-auth/AHD6WCBOQD62DA7FV737RMDP2IULBANCNFSM4HXFUCBA> > . >