In nicholasjclark/BBS.occurrences: BBS community composition and co-occurrence analysis

knitr::opts_chunk$set(echo = TRUE)
library(knitr)
opts_chunk$set(tidy.opts=list(width.cutoff=50),tidy=TRUE)

This appendix describes how to calculate species' pairwise phylogenetic and functional distance matrices. These matrices can be accessed in the BBS.occurrences library using data('Bird.phy.distances') and data('Bird.ecol.distances').

Load the species occurrence dataset, which lists the binary occurrences of 303 avian species (stored in columns) across sites

load("./BBS.Data.cleaned/BBS.2003-2009.filtered.Rdata")
BBS.occurrences <- route.presence.absence

Create a vector of avian binomial species names and remove the underscore so that names match the format needed for extracting phylogenies from Birdtree.org

sp.names <- gsub('_', ' ', colnames(BBS.occurrences))

Export the vector as a one-column .csv file to the Analysis_data folder created in Appendix S3

write.csv(sp.names, './Analysis_data/sp.names.csv', row.names = FALSE)

Once exported, copy the species names and paste into the Select species form on the Phylogeny Subsets page at Birdtree.org. We downloaded 100 trees from the Ericsson All Species dataset for our analyses. Once processed, save the resulting .nex file to the Analysis_data folder as sp.trees.nex and read in the multiphylo object using functions in the ape package

sp.trees <- ape::read.nexus("sp.trees.nex")

Now check to make sure that all of the species names are represented in the tree. Here, we have to include the underscore once again, as this is included in Birdtree.org phylogeny subsets. This call should return TRUE if there are no unmatched names

sp.names.underscore <- gsub(' ', '_', sp.names)
all(sp.names.underscore %in% sp.trees[[1]]$tip.label)

Calculate mean pairwise phylogenetic distances from the distribution of 100 avian trees and scale to range [0, 1]

sp.tree.cophenetics <- lapply(seq_along(sp.trees), function(x){
  cophenetics <- ape::cophenetic.phylo(sp.trees[[x]])
  cophenetics <- cophenetics / max(cophenetics)
})
sp.phy.distances <- Reduce(`+`, sp.tree.cophenetics) / length(sp.tree.cophenetics)

Finally, it will be useful to ensure that rows and columns of the phylogenetic distance matrix are ordered to match the column orders in the BBS species occurrence and abundance datasets.

reorder_ids <- match(sp.names.underscore, rownames(sp.phy.distances))
sp.phy.distances <- sp.phy.distances[reorder_ids, reorder_ids]

\pagebreak

Next, we gather functional trait data for each species to create a functional distance matrix. First, we need to download trait data from the EltonTraits database

  temp <- tempfile()
  download.file('https://ndownloader.figshare.com/files/5631081',
                temp)
  sp.traits <- read.table(temp, header = TRUE, 
                          fill = TRUE, 
                          quote = '"',
                          stringsAsFactors = FALSE,
                          sep = "\t")
  unlink(temp)

The species names are stored in the Scientific column and do not have an underscore. Now we filter the dataset to only include species included in our BBS observations

library(dplyr)
sp.traits <- sp.traits[which(sp.traits$Scientific %in% sp.names),] %>%
  dplyr::right_join(data.frame(Scientific = sp.names))

Gather the desired trait variables that will be used for creating ecological dendrograms. Here, we choose variables to represent species' proportional use of seven different foraging habitat categories and proportional use of ten diet categories, as these categories together should give a reasonable representation of species' local ecological niches

hab.dat <- sp.traits[,c(24:30)]
rownames(hab.dat) <- sp.traits$Scientific
diet.dat <- sp.traits[, c(10:19)]
rownames(diet.dat) <- sp.traits$Scientific

Use these diet and foraging trait datasets to calculate a Gower's distance matrix. Because these are proportional variables, we use the prep.fuzzy argument to calculate the mixed variable distance coefficient

sp.ecol.distances <- prepDendrograms(func.datasets = list(hab.dat, diet.dat),
                                       prep.types = c('prep.fuzzy', 'prep.fuzzy'))

Again, we should ensure that all species are in the same order as in the BBS datasets. We then need to add the underscore to the ecological distance row and column names so they match the remaining names

reorder_ids <- match(sp.names, rownames(sp.ecol.distances))
sp.ecol.distances <- sp.ecol.distances[reorder_ids, reorder_ids]
rownames(sp.ecol.distances) <- colnames(sp.ecol.distances) <- sp.names.underscore

Save the phylogenetic and functional trait matrices in the Analysis_data folder for downstream analysis. Note, the avian distance matrices can be directly accessed in the BBS.occurrences library using data('Bird.phy.distances') and data('Bird.ecol.distances').

save(sp.ecol.distances,
     sp.phy.distances, 
     file = './Analysis_data/
     Bird.Distance.matrices.Rdata')

nicholasjclark/BBS.occurrences documentation built on July 19, 2020, 8:31 p.m.