01_DATA_load_and_clean — Step 7

Compute community-level FRic for all species, combine all PCA objects into a named list, export PCoA coordinates, species–IUCN table, TPD lists, and an eigenvalue summary table for supplementary material.

export FRic

Outputs: data/processed/PCA_Birds.rds data/processed/pcoa_coords.rds data/processed/tpd_lists.rds data/processed/species_table.rds results/tables/Annexe_Table_Eigenvalues.csv

phenoBird <- read.csv("data/processed/phenoBirdsImputedREADY.csv", stringsAsFactors = FALSE)
colnames(phenoBird)[1] <- "scientificNameStd"
rownames(phenoBird)    <- phenoBird[, 1]

# Taxonomy table
taxo <- phenoBird[, c(2, 3, 4)]
taxo$Genus        <- vapply(strsplit(rownames(taxo), "_"), function(x) x[1], character(1L))
taxo$GenusSpecies <- rownames(taxo)
colnames(taxo)    <- c("class", "order", "family", "genus", "genusspecies")
write.csv(taxo, "data/processed/Taxo_Birds.csv", row.names = TRUE)

PCA_file    <- c("data/processed/PCA_Birds_MLD.rds","data/processed/PCA_Birds_M.rds",
                 "data/processed/PCA_Birds_D.rds","data/processed/PCA_Birds_L.rds")
PCA_file_OK <- sub("\\.rds$", "_OK.rds", PCA_file)
TPD_file    <- c("data/processed/TPD_Birds_LMD.rds","data/processed/TPD_Birds_M.rds",
                 "data/processed/TPD_Birds_D.rds","data/processed/TPD_Birds_L.rds")
TPD_file_bio <- c("data/processed/Birds_TPDs_sdggs7_MLD.rds","data/processed/Birds_TPDs_sdggs7_M.rds",
                  "data/processed/Birds_TPDs_sdggs7_D.rds","data/processed/Birds_TPDs_sdggs7_L.rds")

for (i in seq_along(PCA_file)) {
  PCA     <- readRDS(PCA_file[i])
  TPDsAux <- readRDS(TPD_file[i])
  species <- rownames(PCA$PCoA$vectors)
  species <- species[species %in% TPDsAux$data$species]
  occurences <- matrix(1, ncol = length(species), nrow = 1, dimnames = list("ALL", species))
  TPDc_occ          <- TPD::TPDc(TPDsAux, occurences)
  PCA$ALLFRic       <- TPDRichness(TPDc_occ)$communities$FRichness
  PCA$ALLDensity    <- densityProfileTPD(TPDc_occ)
  TPDs_sdggs7       <- readRDS(TPD_file_bio[i])
  TPDc_occ_bio      <- TPD::TPDc(TPDs_sdggs7, occurences)
  PCA$ALLFRicBiogeo <- TPDRichness(TPDc_occ_bio)$communities$FRichness
  saveRDS(PCA, file = PCA_file_OK[i])
}

type      <- c("MLD", "M", "D", "L")
PCA_Birds <- setNames(lapply(PCA_file_OK, readRDS), type)
saveRDS(PCA_Birds, "data/processed/PCA_Birds.rds")

PCA    <- readRDS("data/processed/PCA_Birds.rds")
coords <- list(locomotion   = PCA$M$PCoA$vectors,
               diet         = PCA$D$PCoA$vectors,
               reproduction = PCA$L$PCoA$vectors)
coords <- lapply(coords, function(x) {
  colnames(x) <- paste0("PC", seq_len(ncol(x)))
  x <- as.data.frame(x); x$species <- rownames(x); x
})
saveRDS(coords, "data/processed/pcoa_coords.rds")

species_df <- phenoBird[, c("scientificNameStd", "category")]
colnames(species_df) <- c("species", "iucn")
species_df <- na.omit(species_df[!species_df$iucn %in% c("DD","EW","EX","RE"), ])
saveRDS(species_df, "data/processed/species_table.rds")

tpd_lists <- list(locomotion   = readRDS("data/processed/TPD_Birds_M.rds"),
                  diet         = readRDS("data/processed/TPD_Birds_D.rds"),
                  reproduction = readRDS("data/processed/TPD_Birds_L.rds"))
saveRDS(tpd_lists, "data/processed/tpd_lists.rds")

table_eigen_clean <- do.call(rbind, lapply(names(PCA), function(name) {
  tab <- PCA[[name]]$PCoA$values[, c(1, 2, 4)]
  tab[, 2:3] <- 100 * tab[, 2:3]
  tab <- round(tab, 3)
  tab <- cbind(Space = name, Axis = rownames(tab), tab)
  colnames(tab) <- c("TraitSpace","Axis","Eigenvalue","%Variance","%Cumulative")
  tab
}))
dir.create("results/tables", showWarnings = FALSE, recursive = TRUE)
write.csv(table_eigen_clean, "results/tables/Annexe_Table_Eigenvalues.csv", row.names = FALSE)