Integration of morphological and molecular data sets in estimating fungal phytogenies

Abstract

To provide a clearer picture of fungal species relationships, increased efforts are being made to include both molecular and morphological data sets in phylogenetic studies. This general practice in systematics has raised many unresolved questions and controversies regarding how to best integrate the phylogenetic information revealed by morphological and molecular characters. This is because phylogenetic trees derived using different data sets are rarely identical. Such discrepancies can be due to sampling error, to the use of an inappropriate evolutionary model for a given data set, or to different phylogenetic histories between the organisms and the molecule. Methods have been developed recently to test for heterogeneity among data sets, although none of these methods have been subjected to simulation studies. In this paper we compare three tests: a protocol described by Rodrigo et al., an adapted version of Faith's T-PTP test, and Kishino and Hasegawa's likelihood test. These tests were empirically compared using seven lichenized and nonlichenized Omphalina species and the related species Arrhenia lobata (Basidiomycota, Agaricales) for which nrDNA large subunit sequences and morphological data were gathered. The results of these three tests were inconsistent, Rodrigo's test being the only one suggesting that the two data sets could be combined. One of the three most parsimonious trees obtained from the combined data set with eight species is totally congruent with the relationships among the same eight species in an analysis restricted to the same portion of the nrDNA large subunit but extended to 26 species of Omphalina and related genera. Therefore, the results from phylogenetic analyses of this large molecular data set converged on one of the three most parsimonious topologies generated by the combined data set analysis. This topology was not recovered from either data set when analysed separately. This suggests that Rodrigo's homogeneity test might be better suited than the two other tests for determining if trees obtained from different data sets are sampling statistics of the same phylogenetic history. Key words: data sets heterogeneity, homogeneity test, lichen phylogeny, Omphalina, ribosomal DNA.