R/psi.R

In alexology/biomonitoR: biomonitoring indices and metrics

#' @title Proportion of Sediment-sensitive Invertebrates index
#'
#' @description
#' \Sexpr[results=rd, stage=render]{ lifecycle::badge("maturing") }
#'
#' This function calculates the Proportion of Sediment-sensitive Invertebrates index (PSI) according to the most recent version used in UK.
#'
#' @param x Result of `aggregate_taxa()`.
#' @param method The only choice is `extence`. Users can provide their own `data.frame` (see examples) with a column called *Taxon* and the column of scores called *Score*.
#' @param abucl Log abundance categories. Treshold are set to 1, 9, 99 and 999 as in the original paper of Extence et al. (2013).
#' @param agg This option allows the composite family approach. It can be `FALSE`, `TRUE` or a `data.frame`.
#' If `FALSE` no aggregation will be performed, while if `TRUE` aggregation will be performed according to the rules described in Details.
#' A `data.frame` containing the aggregation rules can be provided by the user.
#' This `data.frame` needs a column called *Taxon* containing the taxon to aggregate and a column called *Correct_Taxon* with the aggregation specifications.
#' `agg` cannot be `TRUE` when a `data.frame` is provided as method.
#' @param exceptions Taxa that to be excluded from the calculation.
#' This option can be useful, for instance, to exclude an alien species belonging to an autochthonous family.
#' @param fssr_scores Optional, scores (fssr) for different abundance categories of taxa associated with
#'  *Fine Sediment Sensitivity Ratings*. To be used when a custom `method` is provided.
#' @param traceB if set to `TRUE` a list as specified below will be returned.
#'
#' @keywords psi
#' @details Although Extence et al 2013 did not suggest any aggregation rule, the `psi()` implementation of `biomonitoR`
#' allows for a default aggregation as specified below. Custom aggregation rules can be provided as a `data.frame`.
#'
#' \enumerate{
#'  \item Tipulidae (inc. Limoniidae, Pediciidae & Cylindrotomidae)
#'  \item Siphlonuridae (inc. Ameletidae)
#'  \item Hydrophilidae (inc. Georissidae, Helophoridae & Hydrochidae)
#'  }
#'
#' `psi()` automatically check for parent-child pairs in the scoring system, see the return section for a definition.
#' All the information used for PSI calculation can be retrieved with `show_scores()`.
#'
#' @return If `traceB` is set to `TRUE` a list with the following elements will be returned:
#' \itemize{
#'  \item `results` Results of `psi()`.
#'  \item `taxa_df` The `data.frame` used for the calculation containing the abundance of taxa receiving a score.
#'  \item `abu_df` The `data.frame` containing fssr scores and abundance classes for each site.
#'  \item `psi_df` The `data.frame` used for the calculation containing scores for each site.
#'  \item `composite_taxa` Taxa aggregated following the aggregation of the default method or set in `agg`.
#'  \item `exceptions` A `data.frame` containing the changes made by excluding the taxa included in `exceptions`.
#'  \item `parent_child_pairs` For instance in Spanish BMWP both *Ferrissia* and Planorbidae receive a score.
#'  Abundances of the higher taxonomic level need therefore to be adjusted by subtracting the abundances of the lower taxonomic level.
#' }
#'
#' @references Extence CA, Chadd RP, England J, Dunbar MJ, Wood PJ, Taylor ED. 2013. The assessment of fine sediment accumulation in rivers using macro-invertebrate community response. River Research and Applications 29, 17-55.
#' @section Acknowledgements: We thank Carol Fitzpatrick, Richard Chadd, Judy England and Rachel Stubbington for providing us with the most updated PSI scores and algorithms.
#' @importFrom dplyr '%>%' select inner_join group_by summarise rename filter
#' @importFrom tidyr pivot_longer pivot_wider
#' @importFrom tibble deframe
#' @importFrom stats aggregate
#' @export
#' @seealso [as_biomonitor]
#' @examples
#' data(macro_ex)
#' data_bio <- as_biomonitor(macro_ex)
#' data_agr <- aggregate_taxa(data_bio)
#' psi(data_agr)
#'
#' # change abundance classes
#' psi(data_agr, abucl = c(1, 9, 99, 999, 9999))
#'
#' # provide your own score system. Scores and aggregation rules are for example purpose only.
#'
#' psi_fssr <- data.frame(
#'   Taxon = c("Ephemerellidae", "Leuctridae", "Chironomidae"),
#'   FSSR_Score = c(1, 2, 3)
#' )
#'
#' psi_acc <- data.frame(Taxon = "Ephemerellidae", Correct_Taxon = "Chironomidae")
#'
#' fssr_scores <- data.frame(
#'   FSSR = rep(1:3, each = 3), ABUCLASS = rep(1:3, 3),
#'   SCORE = c(9, 10, 11, 8, 9, 10, 7, 7, 7)
#' )
#'
#' # without aggregation rules
#' psi(data_agr, method = psi_fssr, fssr_scores = fssr_scores, traceB = TRUE)
#'
#' # with aggregation
#'
#' psi(data_agr, method = psi_fssr, agg = psi_acc, fssr_scores = fssr_scores, traceB = TRUE)
psi <- function(x, method = "extence", abucl = c(1, 9, 99, 999), agg = FALSE, fssr_scores = NULL, exceptions = NULL, traceB = FALSE) {


  # check if the object x is of class "biomonitoR"
  classCheck(x)

  # useful for transforming data to 0-1 later
  if (inherits(x, "bin")) {
    BIN <- TRUE
  } else {
    BIN <- FALSE
  }


  if (!any(identical(method, "extence") | is.data.frame(method))) stop("method is not extence or a custom data.frame")
  if (!any(isFALSE(agg) | isTRUE(agg) | is.data.frame(agg))) stop("agg is not one of TRUE, FALSE or a custom data.frame")
  if (!any(is.null(fssr_scores) | is.data.frame(fssr_scores))) stop("fssr_scores is not one of NULL or a custom data.frame")

  # Store tree for searching for inconsistencies
  Tree <- x[["Tree"]][, 1:10]

  numb <- c(which(names(x) == "Tree"), which(names(x) == "Taxa")) # position of the Tree and Taxa data.frame in the biomonitoR object that need to be removed

  # remove Tree and Taxa data.frame
  x <- x[-numb]
  st.names <- names(x[[1]][-1]) # names of the sampled sites

  # create dummy variables to avoid R CMD check NOTES


  z <- w <- Taxon <- Score <- Sample <- FSSR <- FSSR_SCORE <- FSSR_Score <- SCORE <- NULL


  # the following if statement is to allow the users to provide their own psi scores and aggregation rules.
  # y represents the method to be used

  if (is.data.frame(method)) {
    if (!(isFALSE(agg) | is.data.frame(agg))) {
      stop("When method is a data.frame agg needs to be FALSE or a data.frame containing the aggregation rules")
    }
    y <- method
    if (!is.data.frame(fssr_scores)) stop("fssr_scores needed when method is a data.frame provided by the user")
    w <- fssr_scores
    if (is.data.frame(agg)) {
      z <- agg
    }
  } else {
    if (!(isTRUE(agg) | isFALSE(agg))) stop("When using the deafult method agg can only be TRUE or FALSE")
    if (!is.null(fssr_scores)) stop("When using deafult methods fssr_scores can only be NULL")

    # assign the default scores and aggregation rules as needed by the user
    if (identical(method, "extence")) {
      y <- psi_scores_fam_extence
      w <- psi_fssr
      if (isTRUE(agg)) {
        z <- epsi_acc_fam_uk
      }
    }
  }


  # the calculation of the index in biomonitoR consists in rbind all the taxonomic levels
  # in the biomonitoR object that has been previously deprived of Taxa and Tree elements and then merge
  # it with the scores data.frame.
  # The first step is to change the column name of the first column of each data.frame to
  # an unique name

  for (i in 1:length(x)) {
    colnames(x[[i]])[1] <- "Taxon"
  }

  # rbind the data.frames representing a taxonomic level each
  # aggregate is not necessary here
  DF <- do.call("rbind", x)
  rownames(DF) <- NULL
  DF <- aggregate(. ~ Taxon, DF, sum)

  if (!is.null(exceptions)) {
    DF <- manage_exceptions(DF = DF, Tree = Tree, y = y, Taxon = exceptions)
    if (!is.data.frame(DF)) {
      exce <- DF[[2]]
      DF <- DF[[1]]
    }
  }

  # merge the new data.frame with the score data.frame and change
  # the names of the taxa according to the aggregation rules if needed
  DF <- merge(DF, y[, "Taxon", drop = FALSE])
  if (!is.null(z)) {
    taxa.to.change <- as.character(DF$Taxon[DF$Taxon %in% z$Taxon])
    DF <- checkBmwpFam(DF = DF, famNames = z, stNames = st.names)
  } else {
    DF <- DF
  }

  if (traceB) {
    df2 <- DF
  }

  DF <- manage_inconsistencies(DF = DF, Tree = Tree)
  if (!is.data.frame(DF)) {
    incon <- DF[[2]]
    DF <- DF[[1]]
  }

  # transform the data.frame from abundance to presence-absence if needed
  if (BIN) {
    DF <- to_bin(DF)
  }

  DF <- merge(y, DF)

  class.fun <- function(x) cut(x, breaks = c(abucl, 10^18), labels = 1:length(abucl), include.lowest = TRUE, right = TRUE)
  abu.class <- apply(apply(DF[, -c(1:2), drop = FALSE], 2, class.fun), 2, as.numeric)
  abu.class[is.na(abu.class)] <- 0
  tot.mer <- data.frame(DF[, 1:2], abu.class, check.names = FALSE)


  res.tot <- tot.mer %>%
    pivot_longer(-c(Taxon, FSSR_Score), names_to = "Sample", values_to = "ABUCLASS") %>%
    dplyr::rename("FSSR" = "FSSR_Score") %>%
    inner_join(w, by = c("FSSR", "ABUCLASS")) %>%
    select(c("Sample", "SCORE")) %>%
    group_by(Sample) %>%
    summarise(PSI = sum(SCORE)) %>%
    deframe()

  res.sen <- tot.mer %>%
    pivot_longer(-c(Taxon, FSSR_Score), names_to = "Sample", values_to = "ABUCLASS") %>%
    dplyr::filter(FSSR_Score == 1 | FSSR_Score == 2) %>%
    dplyr::rename("FSSR" = "FSSR_Score") %>%
    inner_join(w, by = c("FSSR", "ABUCLASS")) %>%
    select(c("Sample", "SCORE")) %>%
    group_by(Sample) %>%
    summarise(PSI = sum(SCORE)) %>%
    deframe()

  if (length(res.sen) == 0) (res.sen <- rep(0, length(st.names)))

  # assure to have results if all indicator taxa are missing from a sample
  if (length(res.tot) != length(st.names) | length(res.sen) != length(st.names)) {
    temp <- merge(data.frame(Sites = st.names), data.frame(Sites = names(res.tot), res_tot = res.tot), all = TRUE)
    temp <- merge(temp, data.frame(Sites = names(res.sen), res_sen = res.sen), all = TRUE)
    res <- temp$res_sen / temp$res_tot * 100
    names(res) <- temp$Sites
  } else {
    res <- res.sen / res.tot * 100
  }


  res <- res[match(st.names, names(res))]

  if (traceB == FALSE) {
    res
  } else {
    if (!exists("taxa.to.change", inherits = FALSE)) {
      df3 <- "none"
    } else {
      df3 <- taxa.to.change
    }
    if (exists("exce", inherits = FALSE)) {
      df4 <- exce
    } else {
      df4 <- "none"
    }
    if (exists("incon", inherits = FALSE)) {
      df5 <- incon
    } else {
      df5 <- "none"
    }

    temp <- tot.mer %>%
      pivot_longer(-c(Taxon, FSSR_Score), names_to = "Sample", values_to = "ABUCLASS") %>%
      dplyr::rename("FSSR" = "FSSR_Score") %>%
      inner_join(w, by = c("FSSR", "ABUCLASS")) %>%
      select(c("Taxon", "Sample", "SCORE")) %>%
      pivot_wider(names_from = Sample, values_from = SCORE, values_fill = list(SCORE = 0)) %>%
      as.data.frame()
    res <- list(results = res, taxa_df = df2, abu_cl = tot.mer, psi_df = temp, composite_taxa = df3, exceptions = df4, parent_child_pairs = df5)
    res
  }
}