Assessment of the Accuracy of Matrix Representation with Parsimony Analysis Supertree Construction

Abstract

Despite the growing popularity of supertree construction for combining phylogenetic information to produce more inclusive phylogenies, large-scale performance testing of this method has not been done. Through simulation, we tested the accuracy of the most widely used supertree method, matrix representation with parsimony analysis (MRP), with respect to a (maximum parsimony) total evidence solution and a known model tree. When source trees overlap completely, MRP provided a reasonable approximation of the total evidence tree; agreement was usually > 85%. Performance improved slightly when using smaller, more numerous, or more congruent source trees, and especially when elements were weighted in proportion to the bootstrap frequencies of the nodes they represented on each source tree ("weighted MRP"). Although total evidence always estimated the model tree slightly better than nonweighted MRP methods, weighted MRP in turn usually out-performed total evidence slightly. When source studies were even moderately nonoverlapping (i.e., sharing only three-quarters of the taxa), the high proportion of missing data caused a loss in resolution that severely degraded the performance for all methods, including total evidence. In such cases, even combining more trees, which had positive effects elsewhere, did not improve accuracy. Instead, "seeding" the supertree or total evidence analyses with a single largely complete study improved performance substantially. This finding could be an important strategy for any studies that seek to combine phylogenetic information. Overall, our results suggest that MRP supertree construction provides a reasonable approximation of a total evidence solution and that weighted MRP should be used whenever possible.

Plant DNA barcodes and a community phylogeny of a tropical forest dynamics plot in Panama

“…Thus, despite the sparseness of our alignment matrix (94% of the cells were missing data) the topological arrangements closely approximated expected relationships according to APG II. The importance of including in a supermatrix one gene that is present for all taxa and readily aligns across the full matrix, e.g., rbcL in our analysis, has been shown in other studies using supermatrices for building phylogenies (8,9).…”

“…In this respect a supermatrix design (8,9) is ideal for using a mixture of coding genes and intergenic spacers for phylogenetic reconstruction across the broadest evolutionary distances, as in the construction of community phylogenies (10). We define a supermatrix as a phylogenetic matrix that may contain a high incidence of missing data and the data content for any one taxon is stochastic (11) (Fig.…”

Maximum Agreement and Compatible Supertrees

“…For instance, the popular matrix representation with parsimony (MRP) method [3,35] has relatively low accuracy when the input trees overlap moderately and [12] recommends adding to the set of input trees a tree with leaves spanning most input trees, that they call a seed tree. Any supertree that is a solution of SMAST or SMCT most likely contains leaves from most, if not all, input trees (see Theorem 3) and, moreover, fully agrees with all of these trees by definition.…”

Some Desiderata for Liberal Supertrees

“…Thus weighting schemes can be used that reflect measures of support for components or triplets, such as bootstrap proportions (Felsenstein, 1985;Wilkinson, 1996) or decay indices (Bremer, 1988;Donoghue et al, 1992;Wilkinson et al, 2000). Ronquist (1996) argued the virtues of differential weighting in the context of MRP and simulations suggest that weighting in this context can improve accuracy (Bininda-Emonds and Sanderson, 2001).…”

“…The ability of any method to construct accurate supertrees under a range of readily modeled analytical conditions can be assessed by simulation (e.g. Bininda-Emonds and Sanderson, 2001;Chen et al, 2003;Lapointe and Levasseur, this volume;Ross and Rodrigo, this voume). However, the ability of methods to produce accurate trees depends very much on properties of the data, and insights from simulations notwithstanding, for the most part we do not know how accurate real supertrees are.…”