Treatment of multiple trees in resampling analyses.

JAN DE LAET; FARRIS, JAMES S.; GOLOBOFF, PABLO.

Paris

Congreso; 23rd Annual Meeting of the Willi Hennig Society. "Phylogenetics and Evolutionary Biology"; 2004

Willi Hennig Society

When Felsenstein first used bootstrapping in phyloenetic analyses he did not discuss how to treat multipleptimal trees that might be found for a single pseudoeplicate. He subsequently introduced a method inwhich all optimal trees found for a pseudoreplicate areassed to the majority-rule summary tree calculation,ach tree being weighted by the inverse of the number ofrees found for the pseudoreplicate. While this method istill use in both PHYLIP and PAUP, no justification hasver been offered for it, and in fact it seems to have beendopted originally simply because the old PHYLIPonsensus program could not process multifurcatingrees. We point out here that this method is defective inhat it can yield high resampling frequencies for groupshat are actually unsupported by the data, and this canccur in both parsimony and likelihood analyses.