IMA Complex Systems Seminar

2:30 Wednesday, October 15, 2003


Finding supertrees using distance methods


Stephen J. Willson

Department of Mathematics

Iowa State University

Ames, IA 50011


Abstract:  Suppose that a tree T has the set S of leaves and S' is a subset of S.  Then T|S' denotes the tree obtained from T with the set S' of leaves, showing only the relationships among S'.  Suppose that for i = 1, ..., k,  T_i is a tree with the set S_i of leaves.  A "supertree" is a single tree T with the set S of leaves, where S is the union of the S_i, such that T|S_i = T_i for each i.  A major problem in phylogeny is, given such a collection of trees T_i, to identify a supertree T if one exists.  A method for solving this problem could be used to combine different phylogenetic trees from different researchers into a larger tree.


The most common method currently used to find supertrees is MRP (matrix representation with parsimony).  It is computationally intensive, it often leads to a great many possible solution trees, and it is not founded on theory.


This talk proposes a supertree method in which additional input information is required.  We require in addition that for each tree T_i there is given an additive distance function d_i on S_i which estimates the amount of evolution between any two taxa.  We then try to create a supertree T with additive distance function d which equals d_i when restricted to S_i.


Some results of simulations will be presented.