Evolutionary rate simulation
To illustrate the difference between evolutionary rate estimates based on traditional coding schemes using a continuous measure of central tendency measure (median environmental values at suitable sites) and our coding scheme based on considering the range of suitable values independent of their frequency across the landscape, we generated 4 sets of 100 stochastic birth-death phylogenies (birth rate = 1, death rate = 0) with 6, 50, 100, and 490 tip taxa using the R package phytoolsv.0.6-99 (Revell, 2012). Then, we randomly assigned simulated cool niche species to the tips of each tree. To characterize each species’ niche, we used two methods: (i) a traditional approach that uses the median cell value of suitable area within M to represent the species’ fundamental niche, and (ii) the midpoint bin value from ‘known’ bins using our new coding method (representing a best-case scenario in which all species have completely characterized niches. This same process was then repeated for the warm niche species.
These two datasets simulate the simplest evolutionary scenario: a completely conserved fundamental niche across the clade. However, estimates of niche evolution based on realized niches may differ greatly from this known value; as such, we compared rates of niche evolution estimated from our two coding schemes (traditional and known-bin). To estimate rates of niche evolution, we calculatedσ 2, the rate of Brownian motion (BM) evolution, using the fitContinuous function in the R package geigerv.2.0.6 (Harmon et al. 2008).σ 2 is equal to the rate of variance accumulation per unit of branch length (Felsenstein 1985; O’Meara et al. 2006), and has been interpreted as a measure of the rate of ecological niche evolution (Cooper et al. 2011). Smallerσ 2 values can be interpreted as representing less niche evolution, and larger σ 2 values as more niche evolution. σ 2 itself, however, does not provide information on whether niches are conserved, but rather informs about degrees of change among traits or groups (Revell et al. 2008; Cooper et al. 2011). Code for analyses can be found in Supplement 3; annoted code with tables and figures provided as an HTML document in Supplement 4.