Geography Reference
In-Depth Information
Several issues concerning areas of endemism should be addressed.
Crisp et al. (1995) suggested that the alternative procedures for identifying
areas of endemism were controversial, especially questioning whether the
hierarchical model of parsimony analysis of endemicity (PAE) was adequate
for that purpose. Humphries and Parenti (1999) argued that including spe-
cies that are ecologically very different can argue for a historical rather than
ecological explanation for the areas of endemism identified. Linder (2001)
proposed three optimality criteria to help choose the best estimate of the
areas of endemism: the number of areas identified, the proportion of the
species restricted to the areas of endemism, and the congruence of the dis-
tributions of the species restricted to the areas of endemism. Roig-Juñent
et al. (2002) enumerated some problems with the identification of areas of
endemism: lack of distributional data, bias toward locality data, and sub-
jectivity in drawing the exact limits of the areas of endemism.
Six methods can be applied in panbiogeography and the identification of
areas of endemism (Morrone 2004a; Morrone and Crisci 1995). I will deal
herein with the minimum spanning-tree method (Croizat 1958b, 1964), track
compatibility (Craw 1988a), PAE (Rosen 1988b), and endemicity analysis
(Szumik et al. 2002). Connectivity and incidence matrices (Page 1987),
which have never been applied empirically, and nested areas of endemism
analysis(DeoandDeSalle2006),anadaptationofthenestedcladeanalysis
(Templeton 1998), are not dealt with herein.
Croizat did not focus on explicit methods (Grehan 1991; Humphries and
Parenti 1999), but one may consider the minimum-spanning tree method as
the first formalization of panbiogeography (Craw 1988a; Page 1987). It con-
sists of delineating on maps the individual tracks of different taxa and then
superimposing them in order to find the generalized tracks. Nodes are iden-
tified in the areas where two or more generalized tracks superimpose. With
few individual tracks, the minimum-spanning tree method is easy to apply.
Algorithm
It consists of the following steps (Morrone 2004b):
