Getting and cleaning data
A common hypothesis tested in comparative methods is whether habitat shifts drive rates of diversification in various groups. For example, Santini et al. (2013) tested, among other things, whether reef-associated pufferfishes enjoyed faster rates of speciation compared to their non-reef relatives.
First, load the fishtree package and download the subset of the Fish Tree of Life corresponding to the taxon of interest. To make things more interesting we’ll work on the entire order (Tetraodontiformes) rather than the family in the 2013 study.
library(fishtree)
tree <- fishtree_phylogeny(rank = "Tetraodontiformes")
plot(tree, show.tip.label = FALSE, no.margin = TRUE)
Next, we need to get habitat data and associate it with our phylogeny. The rfishbase package (Boettiger et al. 2012) has a variety of convenient functions to access data recorded by the Fishbase editors. Load the rfishbase package and retrieve the relevant data in the DemersPelag field, which identifies whether a species is reef-associated or not, among other things.
library(rfishbase)
tips <- gsub("_", " ", tree$tip.label, fixed = TRUE)
fb_results <- species(species_list = tips, fields = c("Species", "DemersPelag"))
fb_results <- fb_results[!is.na(fb_results$Species), ]
head(fb_results)
#> # A tibble: 6 x 2
#> Species DemersPelag
#> <chr> <chr>
#> 1 Carinotetraodon lorteti benthopelagic
#> 2 Carinotetraodon borneensis benthopelagic
#> 3 Carinotetraodon irrubesco benthopelagic
#> 4 Carinotetraodon salivator benthopelagic
#> 5 Carinotetraodon travancoricus demersal
#> 6 Carinotetraodon imitator benthopelagicNote that we had to replace the underscores in the tip labels with spaces. This is a common source of errors, so if your analyses don’t seem to work correctly always check whether the functions you’re using expect underscores or spaces.
There’s a lot of data in the DemersPelag field, but we only want to know if the species is reef-associated or not.
reef <- data.frame(tip = gsub(" ", "_", fb_results$Species),
is_reef = as.numeric(fb_results$DemersPelag == "reef-associated"))
head(reef)
#> tip is_reef
#> 1 Carinotetraodon_lorteti 0
#> 2 Carinotetraodon_borneensis 0
#> 3 Carinotetraodon_irrubesco 0
#> 4 Carinotetraodon_salivator 0
#> 5 Carinotetraodon_travancoricus 0
#> 6 Carinotetraodon_imitator 0We’ve also converted the tip labels back to underscores, since we need to ensure that the tip labels on our phylogeny match the labels on our trait data. The geiger package (Pennell et al. 2014) provides a convenient function that will perform this check. The name.check function expects row names on our data object, so we will do that as well.
library(geiger)
rownames(reef) <- reef$tip
nc <- geiger::name.check(tree, reef)
nc
#> $tree_not_data
#> [1] "Lagocephalus_lagocephalus_lagocephalus"
#> [2] "Monotrete_cochinchinensis"
#> [3] "Monotrete_leiurus"
#> [4] "Sphoeroides_cheesemanii"
#> [5] "Takifugu_fasciatus"
#> [6] "Tetraodon_abei"
#> [7] "Tetraodon_baileyi"
#> [8] "Tetraodon_cambodgiensis"
#> [9] "Tetraodon_palembangensis"
#> [10] "Tetraodon_suvattii"
#> [11] "Tetraodon_turgidus"
#> [12] "Tetrosomus_fornasini"
#>
#> $data_not_tree
#> character(0)We’ve identified a mismatch between the tree and the data. We’ll exclude the tips lacking trait data using drop.tip:
If we also had data that was not in the tree, we could exclude that using the following command, but it isn’t necessary in this case:
Confirm that we have the same number of observations in the tree and the data:
