Skip to main content

Archived Comments for: Long branch attraction, taxon sampling, and the earliest angiosperms: Amborella or monocots?

Back to article

  1. Darwin's abominable mistery solved? Not so fast

    Vadim Goremykin, Institut fuer Spezielle Botanik, Universitaet Jena, Jena, Germany

    12 October 2004

    Vadim V Goremykin1, Karen I Hirsch-Ernst2, Stefan Wolfl3 and Frank H Hellwig 1

    1 Institut fuer Spezielle Botanik, Universitaet Jena, Jena, Germany

    2 Zentrum Pharmakologie und Toxikologie, Universitat Goettingen, Goettingen, Germany

    3 Klinik fur Innere Medizin, Universitaet Jena, Jena, Germany

    We were interested to read the revised version of the manuscript by Stefanovic et al. [1] who conclude that some of our earlier reported findings are a result of inadequate taxon sampling, inappropriate phylogenetic methodology and rapid evolution in the grass lineage. They emphasize the crucial importance of using phylogenetic methods that best model the underlying molecular evolutionary processes, and hypothesize that by analyzing the Goremykin et al. (2003) dataset [2] with a focus on rate heterogeneity and taxon sampling of monocots, the Amborella topology will be recovered. The authors portray us as (a) having carried out ill considered phylogenetic analyses and as (b) vehement advocates of an evolutionary hypothesis with grasses as the most basal angiosperm lineage.

    It concerns us that numerous aspects of the paper by Stefanovic et al. continue to be deliberately misleading. As known by Stefanovic et al. we have published two papers concerning the question of basal angiosperm lineages [2,3], where we report analyses of complete chloroplast genome data and discuss issues of phylogenetic reconstruction that we consider are relevant for developing a better understanding of the nature of basal angiosperm lineages.

    We reported analyses using a number of different substitution models and methods of phylogenetic reconstruction [2]. Like Stefanovic et al. [1] we too used Modeltest to examine 56 hierarchical nested models and identified that the best fitting symmetrical model (one assuming positional rate heterogeneity) also gave an Amborella basal topology. However, there were many models that did not give this result, and given the poor fit between the best models and real sequence data we raised the question of the appropriateness of fitting symmetric sequence substitution models to our data. The doubt about Amborella being basal was raised in our minds because prior to DNA sequence analyses no one had ever suggested that Amborella might be a basal angiosperm lineage. It is not a finding that is suggested from morphological analyses (see [2] for a review of classical views). Thus, it is the interpretation of the data, not its analysis which differs between the findings of Stefanovic et al. and ours. However, Stefanovic et al. do not acknowledge this. Rather their arguments are based on the presentation of an earlier published finding as if it were a novel observation.

    We discussed concerns about the analyses of genome sequences relevant to the issue of angiosperm origins [2,3]. In particular we have commented on the issue of concatenation of genome data and the need to better sample genomes before strong conclusions can be drawn in respect to identification of the most basal angiosperms. This statement is in direct contrast to the suggestion by Stefanovic et al. that we are ignorant of the importance of taxon sampling. Stefanovic et al. repeatedly stress that they have sufficient taxon sampling to resolve relationships amongst the angiosperms, whilst our work stands in sharp contrast. However, rather than making assertive statements, it would be more helpful if Stefanovic et al. simply explained the large body of previously accumulated results that support their views, taking into account the well-known issue of tree instability with character-wise small data sets (as shown by Rokas et al. 2003 [4] and by Goremykin et al. 2004 for the chloroplast data [3]).

    Given the limitations of available data at the moment, we acknowledge that it may be incorrect to define monocots as the most basal angiosperm lineage. Indeed we tried to emphasize this point in our second manuscript [3] (an observation that obviously has been overlooked by Stefanovic et al). At the same time, we are also not confident in the conclusions of Stefanovic et al. concerning angiosperm origins, particularly, since these are based on analyses in which they removed grasses from their alignment, leaving Acorus as the only representative of monocots. Analyses of their complete data, however, are contradictory. They do not allow to establish the monophyly of Acorus and grasses (i.e. to provide the rationale for the removal of grasses) with any confidence. One cannot help but notice that authors criticise our insufficient sampling of angiosperms and at the same time they delete an important line of angiosperms (see also below) in order to receive the desired result.

    We suspect that the issues of long branch attraction (LBA) are more complex than envisaged by Stefanovic et al. We report a number of observations relevant to this issue in our second paper, but these points are ignored by Stefanovic et al. For example, the observation that the numbers of synapomorphies in our data supporting monophyly of the dicots is eightfold (amino acid alignment) and fivefold (nucleotide alignment) higher than the numbers of synapomorphies supporting the alternative clade of monocots+eudicots (presented by the authors as the correct one) is not mentioned or discussed by Stefanovic et al.

    Our finding that the grasses share by far the largest number of same character states with Pinus to the exclusion of all other angiosperm lines is also not mentioned in their manuscript. If the appearance of grasses next to Pinus is due to LBA, one can expect that the above phenomenon is due to homoplasy. Yet we have shown in our paper on Nymphaea chloroplast DNA that these positions are not homoplastic.

    In summary, while the paper by Stefanovic et al. provides some new data and some useful observations relevant to the issue of angiosperm origins, these are presented in a way that deliberately misrepresents our published findings and also our views. We also feel that Stefanovic et al. could make their important points without the need to falsely suggest novelty in order to artificially elevate the significance of their own contribution.

    References

    [1] Stefanovic S, Rice DW, Palmer JD: Long branch attraction, taxon sampling, and the earliest angiosperms: Amborella or monocots? BMC Evol Biol 2004, 4:35

    [2] Goremykin VV, Hirsch-Ernst KI, Wolfl S, Hellwig FH: Analysis of the Amborella trichopoda chloroplast genome sequence suggests that amborella is not a basal angiosperm. Mol Biol Evol 2003, 20(9):1499-505.

    [3] Goremykin VV, Hirsch-Ernst KI, Wolfl S, Hellwig FH: The chloroplast genome of Nymphaea alba: whole-genome analyses and the problem of identifying the most basal angiosperm. Mol Biol Evol 2004 21(7):1445-54.

    [4] Rokas A, Williams BL, King N, Carroll SB: Genome-scale approaches to resolving incongruence in molecular phylogenies. Nature 2003, 425(6960):798-804.

    Competing interests

    None declared

Advertisement