Improvements to implied weighting

GOLOBOFF, PABLO

Sao Jose do Rio Preto

Congreso; XXX International Meeting of the Willi Hennig Society; 2011

Wili Hennig Society

This presentation will discuss some improvements to implied character weighting, already outlined in Goloboff et al. (2008. Cladistics 24:758 ? 773). The first is the possibility to apply implied weighting to a subset of the characters, giving a constant weight to the rest of the characters. The second enhancement allows weighting sets of characters, based on the average homoplasy of the set; within each set, all the characters have the same weight. This could already be done easily under successive weighting, but under implied weighting special precautions are required to eliminate effects resulting from different numbers of characters in the weighting sets. This can be used to collectively give weights to all first, second, and third positions in a gene. Or, alternatively, to define regions of the molecule which could be expected to have differences in their reliability as indicators of phylogenetic relationship (e.g. stems/loops). The third enhancement is reducing the effect of missing entries on the implied weights. As missing entries for a character increase, the homoplasy for that character decreases, and thus characters with many missing entries tend to have higher implied weights. This is solved by using a weighting function which decreases the implied weight more rapidly for those characters with more missing entries. All the methods discussed have been implemented (with the "xpiwe" command) in TNT.