Last year, the Zoological Society of London, like many other societies interested in taxonomic questions, celebrated the tercentenary of the birth of Carl Linnaeus. This year, it is the 250th anniversary of the publication of his Systema Naturae– an attempt to classify plants, animals and even inanimate matter, in an ordered, hierarchical form based on shared characters. A new dimension was added to this approach through the publication of Charles Darwin's On the Origin of Species by Means of Natural Selection– the 150th anniversary of which we will celebrate next year. Since then, the identification and naming of taxa has been intimately linked to the elucidation of evolutionary relationships between taxa. But how can these relationships be revealed?
Because the traditional classification, based on similarity of a few ‘key’ characteristics, is influenced by the subjectivity of the investigator, attempts were made to develop more objective approaches (for a review, see Linzey, 2001). These attempts led to the establishment of phenetic (numerical) classification and cladistic (phylogenetic) classification in the 1950s and 1960s. Both schools emphasize the need to analyze large datasets, rather than just one or a few ‘key’ characters, with each trait being treated equally. On the other hand, the two schools differ in other, important ways. Phenetics relies on overall similarity among taxonomic units and does not necessarily reflect evolutionary history. After a waning of its initial popularity, phenetic classification has been revived with the advent of molecular taxonomy. [One common method used in molecular taxonomy, neighbour-joining (Saitou & Nei, 1987), has its roots in phenetics.] Cladistics, on the other hand, is based on shared derived (apomorphic) characters, that is traits that represent rather recent adaptations in evolutionary history. Its primary goal is to form taxonomic groupings that reflect their evolutionary history.
The differences between phenetics and cladistics, as well as the implications of these differences for the reconstruction of evolutionary developments, are hardly ever discussed in recent studies. Furthermore, a common conceptual problem associated with gene expression studies is the focus on just one or two characters, which precludes a rigorous cladistic analysis. Arguably, the most serious challenge in molecular phylogenetics studies is the question of how to distinguish phylogenetic signals from random noise in molecular datasets (Wägele & Mayer, 2007). It is important to reflect on these potential problems associated with the approaches used in evo-devo studies and molecular phylogenetics when reading the paper by Thomas Stach in this issue of the Journal of Zoology (Stach, 2008).
In contrast to the interpretation of some widely publicized studies (Bourlat et al., 2003, 2006; Delsuc et al., 2006), Stach presents evidence that chordates do, indeed, represent a monophyletic group, and that the Craniata, Cephalochordata and Tunicata within the Chordata are also monophyletic. Although the estimated total number of species within the Chordata is small (∼50 000), compared with other major animal taxa, the phylogenetic relationship among these chordate groups has received particular attention due to Homo sapiens' own position within the Craniata. For a long time the prevailing notion was that the Cephalochordata are a sister taxon to the Craniata. More recently, however, the ‘Notochordata hypothesis’ was seriously questioned by studies providing molecular evidence in favour of the ‘Olfactores hypothesis,’ claiming a sister-group relationship between the Tunicata and the Craniata (Giribet et al., 2000; Delsuc et al., 2006; Swalla & Smith, 2008).
Stach addresses this issue by revisiting studies aimed at elucidating the phylogenetic relationship between the Craniata, Cephalochordata and Tunicata. He critically discusses the results and interpretation of both gene expression studies and molecular phylogenetics investigations. Stach shows that different molecular studies can come to radically different conclusions depending on the ‘model systems’ sampled or the exact approach used for analysis. As a final step in Stach's review, the various morphological characters used by the proponents of the different hypotheses are subjected to formal cladistic analysis. The outcome of this analysis is remarkable: the results overwhelmingly favour the Notochordata hypothesis, and provide little support for the Olfactores hypothesis.
This demonstration underscores the power of morphological traits. Because of their integrative nature they are the result of many processes, influenced by numerous genetic, molecular, cellular, and ontogenetic factors. Thus, it is likely that morphological traits are, in terms of the phylogenetic relevance of information, in many cases richer than molecular sequencing data.
Are there any lessons to be learned beyond the immediate significance for chordate evolution? Molecular approaches have penetrated – and enriched – almost all areas of modern biology, including taxonomy. However, relying exclusively on molecular approaches to understand complex biological processes will lead to a rather truncated, and possibly even mistaken, interpretation of biological reality. What we need are integrative approaches, incorporating information from all disciplines of biology. The broad appreciation of this seemingly trivial, but often ignored, idea will be key to the future advancement of biology.