One of the most extensively studied aspects of phylogenetic tree shape is b
alance, which is the extent to which nodes divide a tree into clades of equ
al size. Several authors have stressed the importance of tree balance for u
nderstanding patterns of evolution. It has been remarked that paleontologic
al studies commonly produce very unbalanced trees (also called pectinate cl
adograms or "Hennigian combs"). This claim is tested here by comparing the
balance of 50 paleontological trees and 50 neontological trees, all taken f
rom the recent literature. Each tree was reanalyzed from the published data
matrix to ensure its accuracy. The results confirm that paleontological tr
ees tend to be more imbalanced than neontological trees.
That paleontological trees are more imbalanced has been represented as a sh
ortcoming of fossil data sets, but here it is argued that this is the expec
ted result. Even under a simple Markovian model in which all speciations an
d extinctions occur randomly and with equal probability in all parts of the
tree, trees based on taxa from a single time period (e.g., the present day
) are generally more balanced than trees based on all taxa that ever existe
d within the clade. Computer simulation is used to calculate the expected b
alance and standard deviation of trees for up to 40 terminal taxa over the
entire history of a model clade. The balance is measured using Coliess's in
dex. Ic, and the expected balance conforms well with published paleontologi
cal trees. The study underlines the difficulty of applying neontological tr
ee statistics in paleontology.