Conifer phylogeny based on molecular evidence: taxon sampling, gene sampling and rooting problems

ESCAPA, IGNACIO; CATALANO, SANTIAGO ANDRÉS

Congreso; XVIII International Botanical Congress; 2011

Phylogenetic analyses of gymosperms in general, and conifers in particular, have been primarily based on DNA sequences. In comparison, studies including morphological information and fossil species as terminals are extremely rare. Molecular phylogenetic analyses have resulted in notorious disagreements about phylogenetic relationships at the family and generic levels, as well as in the rooting of these groups. Most previous analyses are based on limited gene sampling and taxon sampling, which may explain some of the differences in results. Empirical and theoretical studies have repeatedly demonstrated the advantages of extensive taxon and gene sampling in phylogenetic studies, although the complexity of the analysis increases linearly with the addition of genes and exponentially with the addition of taxa. Until recently, computational limits and data availability have limited the number of phylogenetic studies including multiple genes and hundreds of species. However, development and implementation of new algorithms now allow for the analysis of large data sets using parsimony with a reasonable low computational cost. In the present study we gathered a dataset including more than 350 conifer species as ingroup taxa. Outgroup sampling includes species from other groups of extant seed plants Gene sampling includes more than 20 loci. As with many other applications of a supermatrix, ours has a high percentage of missing entries. The results show, in general terms, the monophyly of modern conifer families and other gymnosperm groups, with moderate to high support for numerous relevant clades. In order to test the effect of limited taxon and gene sampling on the results of previous phylogenetic studies we conducted several different exploratory analyses. First, in order to determine the effect of gene sampling on the monyphyly and rooting of various clades a variable number of genes was randomly deleted from the complete matrix. Similar analyses were carried out to test the effect of ingroup and outgroup sampling on the resulting phylogenetic hypotheses. Our results indicate that rooting and topology of ingroup taxa converge to the same result with increasing numbers of taxa (ingroup and outgroup) and genes. This suggests that differences among the results of previous molecular studies of conifers are most likely caused by limited taxon and gene sampling. Even when general topology among extant species can be stabilized using extensive gene and taxa sampling, there are still particular clades that show either rooting or topological problems. This is not surprising given that living species represent just a small fraction of conifer diversity known to have existed since the origins of the group (approximately 320 mya), and therefore a inclusive molecular analysis is just a first step for understanding the evolution of the group. Future studies are likely to resolve these issues by the use of a total evidence approach, including different sources of data (e.g., morphological characters) and fossil species.