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.

Updating the evolutionary history of Carnivora (Mammalia): a new species-level supertree complete with divergence time estimates

“…The very low bootstrap values for this part of the supermatrix tree (results not shown) indicate a surprising lack of signal within and/or a high degree of conflict among the sequence data. Interestingly, a weighted supertree analysis of the supermatrix data set as individual gene trees does reconstruct all these relationships properly (that is, in agreement with current phylogenetic opinion), again highlighting the different levels on which the supertree and supermatrix approaches operate (trees versus individual characters, respectively; see [81]) as well as the positive effect of weighting MRP analyses according to some measure of evidential support from the source trees [82,83]. …”

“…Weighting in this fashion has been shown to improve the accuracy of MRP supertree construction in simulation [82,83], which even when unweighted shows high accuracy, comparable to pure supermatrix-based analyses [83,117]. However, because this procedure requires that equivalent information is present for all nodes [118], nodes lacking this information (that is, all literature based trees) were weighted effectively neutrally according to the average bootstrap support over all nodes possessing this information (as calculated by SuperMRP.pl).…”

“…Finally, the tree corresponding to the taxonomy of Wozencraft [14] was added as a seed tree to guide the analysis and ensure sufficient overlap between all pairs of species (see [83]). To minimize its impact on the final topology, however, the nodes on the seed tree were down-weighted by a factor of 10 relative to the minimum weight assigned to any other node, meaning that the relationships present in the seed tree could be easily overruled by any real source tree.…”

Fin modules: an evolutionary perspective on appendage disparity in basal vertebrates

“…The phylogenetic framework was constructed using a supertree approach, more specifically with the matrix representation with parsimony (MRP) algorithm [70–72], which is the most commonly used method [73–76]. The MRP method is well suited for inferring topologies from diverse partially overlapping datasets [70–72, 77, 78], notably for the joint analysis of fossil and extant data [74, 79].…”

“…The MRP supertree method combines source trees with the assumption that datasets are independent [73, 76, 83, 84]. However, in this case some source trees cannot be considered as independent, particularly for the fossil groups where phylogenetic analyses often build upon previously published data matrices, adding or re-scoring taxa and characters.…”