Why ontogenetic homology criteria can be misleading: lessons from digit identity transformations.

As the basis for comparative biology, correctly assigning character homology is critical. Yet, identifying homologous characters in practice is often challenging. Among the major roadblocks is that the mechanistic bases of character homology remain in question. Thus, investigators must rely on several independent lines of evidence (e.g., character anatomy, phylogenetic distribution, or embryological position); however, these distinct sources of evidence often lead to conflicting diagnoses of character homology. What is more, there is no consensus regarding the relative importance of distinct lines of evidence for determining character homology. Here, we review the difficulties that have hindered the search for the mechanistic bases of character identity, and relate these issues to a recently proposed mechanistic hypothesis of character identity--the Character Identity Network Hypothesis. Next, using two well-studied cases of homology conflict (i.e., avian and skink digit identity), we assess the utility of different lines of evidence in diagnosing homology. We conclude that, when comparing adult structures, because anatomical characters more closely reflect the actions of the developmental genetic mechanisms of character individuation they are more reliable than embryological homology criteria.