A counterexample of elision’s efficacy over culling.

In response to “Elision: A Method for Accommodating Multiple Molecular Sequence Alignments with Alignment-Ambiguous Sites” Wheeler et al. 1995:

In most cases, elision is useful for resolving alignment-ambiguous sites in a multiple sequence alignment (MSA).  Although Wheeler et al. show that elision is better at resolving MSA ambiguities than culling, here is one counterexample in which elision and culling perform equally:

Consider two putative MSAs (as shown in the figure below). In MSA #1, taxa A and B are more homologous than taxa C. In MSA #2, taxa B and C are more homologous than taxa A. The image below illustrates this case, using a three-character alphabet {gamma, delta, epsilon}. MSA #1 and MSA #2 produce symmetrically opposite phylogenies. When we cull over these MSAs, we produce a star tree (because we cull-out both columns 1 and 2). Furthermore, when we elide over these MSAs we also produce a star tree. Consequently, in this example elision and culling both produce equal support for B = “gamma indel” and B = “indel gamma”.