Fast Evolutionary Distinctiveness

Introduction

Evolutionary distinctiveness is a metric that quantifies how isolated a species is on a phylogenetic tree – some species have few or no close living relatives.

The standard method for calculating evolutionary distinctiveness is either by using the R packages picante or caper. For very large trees, such calculation is a memory-intensive operation and a bottle-neck for these algorithms.

Because of these challenges, we developed a new method in our phyloregion package that speeds up the process significantly to produce results in seconds! See how:

Let’s try computing evolutionary distinctiveness for a tree with 5,000 species:

library(ape)
library(ggplot2)
# packages we benchmark
library(phyloregion)
library(picante)
library(caper)

tree <- ape::rcoal(5000)

ed_picante <- function(x) picante::evol.distinct(x, type="fair.proportion")
ed_caper <- function(x) caper::ed.calc(x)
ed_phyloregion <- function(x) phyloregion::evol_distinct(x, type="fair.proportion")

res <- bench::mark(picante=ed_picante(tree),
          caper=ed_caper(tree),
          phyloregion=ed_phyloregion(tree), check=FALSE)
summary(res)

## # A tibble: 3 × 6
##   expression       min   median `itr/sec` mem_alloc `gc/sec`
##   <bch:expr>  <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
## 1 picante        4.02m    4.02m   0.00415        NA    18.9 
## 2 caper          9.17s    9.17s   0.109          NA     5.45
## 3 phyloregion   65.3ms  70.56ms  13.9            NA     1.98

autoplot(res)

Here, phyloregion is several orders of magnitude faster and efficient in memory allocation than the other packages.

The function in phyloreegion is called evol_distinct and it is used as follows:

evol_distinct(tree, type = c("equal.splits", "fair.proportion"), ...)

If you find this vignette tutorial useful, please cite in publications as:

Daru, B.H., Karunarathne, P. & Schliep, K. (2020) phyloregion: R package for biogeographic regionalization and macroecology. Methods in Ecology and Evolution 11: 1483-1491. doi: 10.1111/2041-210X.13478.