Fast Evolutionary Distinctiveness in R

Barnabas H. Daru

December 12, 2020


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! Here’s how:

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

# packages we benchmark
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),
          phyloregion=ed_phyloregion(tree), check=FALSE)
## # A tibble: 3 x 6
##   expression       min   median `itr/sec` mem_alloc `gc/sec`
##   <bch:expr>  <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
## 1 picante        11.9m    11.9m   0.00140   287.9GB    0.122
## 2 caper          21.1s    21.1s   0.0474     21.6GB    0.284
## 3 phyloregion   64.5ms   67.2ms  14.8       882.2KB    0

plot of evolutionary distinctiveness 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.