Biodiversity Conservation and Phylogenetic Systematics

73

data (e.g. Blomberg et al. 2003 ; O’Meara et al. 2006 ; Harmon et al. 2010 ). It remains
possible that Brownian motion might still best capture aggregate species differences
even when individual traits diverge from a Brownian motion model, assuming traits
are evolving independently or when selective regimes fl uctuate over time (Felsenstein
1988 ). However, this expectation has rarely been evaluated using empirical data.
Finally, we note that our understanding of the distribution of phylogenetic diver-
sity across space and among communities might also be informed by further consid-
eration of evolutionary models. For example, traditional metrics of phylogenetic
diversity tend to correlate very closely with species richness (Rodrigues et al. 2005 ),
although it is possible to identify regions of greater or lower phylogenetic diversity
than predicted from species richness alone, for example, by looking at residual vari-
ation (e.g. Forest et al. 2007 ; Davies et al. 2008 ). The covariation between evolu-
tionary history and species richness might exhibit very different properties under
alternative evolutionary models, but as far as we are aware, there have not yet been
any equivalent studies exploring such models in geographical space.

●

●

●●

●

●

●●●

●●

●

● ●

●

●

●

●

●●

●

●

●

●

●

Lambda

proportional loss of phylogenetic diversity

0.08

0.10

0.12

0.14

0.16

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Fig. 4 Results from simulated extinctions with varying levels of phylogenetic clustering (Lambda)
across 100 random birth–death trees (see Fig. 1c ) assuming p(ext) = 0.75 for the top 25 % of spe-
cies. Light grey boxes = expected loss of PD for empirical branch length s (assuming phylogenetic
gradualism or a Brownian motion model of trait change); dark grey boxes = expected loss of PD
assuming equal branch lengths (matching to a punctuated model of trait change). Simulations with
Lambda = 0 are equivalent to random extinctions. This fi gure is similar to that in Davies and
Yessoufou ( 2013 ), but presents a new set of stochastic simulations

Reconsidering the Loss of Evolutionary History: How Does Non-random Extinction...