Fundamental concepts of Cladistics

At its heart, cladistics addresses the dichotomy between typological classification and classification which reflects common ancestry. The Linnaean classification system employed with little modification for nearly two centuries is incapable of reflecting genealogy, in that it predated the elaboration of Darwinian evolution by nearly a hundred years. The Linnaean system, which we can refer to as the system whereby organisms were classified on the basis of general similarity or similar criteria, as opposed to their phylogenetic affinities, is an artifact of Platonic essentialism (Hull 1965, Wiley 1981, Mayr 1982, 2001, Carroll 1988, de Queiroz 1988, de Queiroz & Gauthier 1990, 1992, 1994). However, since the work of Alfred Wallace and Charles Darwin and the publication of a comprehensive theory of evolution in 1859, it has become increasingly clear that taxa are not typological units or duplicates of the fundamental &#8220;essence&#8221; of the parental lineage (Dingus & Rowe 1998, Mayr 2001). Rather, taxa display phenotypic dynamism&#8212;morphological plasticity&#8212;and most crucially, are descended from common ancestors in a nested hierarchy.

However, if one has a system of classification which is founded upon the notion that just the opposite is true, it becomes evident that said system cannot adequately reflect genealogy. While the terminological infrastructure which Linne devised, and the concomitant system of binomial nomenclature will no doubt always remain a fundamental part of our taxonomic science, only by using other methods than those used by Linne himself, mapped onto that infrastructure, can we begin to reflect phylogeny in our classifications, and not trivial similarities.

Today the principal goal of phylogenetic reconstruction is delimiting taxa, which reflect genealogy. Thus, systematists seek to define taxa on grounds such that all members included within a taxon are derived from the same ancestor, and all descendants of that common ancestry are included within the taxon. Hennig used the term monophyletic to distinguish such &#8220;natural&#8221; taxa from the polyphyletic or &#8220;artificial&#8221; taxa that consist of a hodgepodge of unrelated forms (with which the classification of plants and animals remains strewn). As some systematists have used the term monophyly to denote taxa which while displaying common ancestry still exclude all forms derived from that ancestry (paraphyly), an alternative and preferable term has been coined for monophyly in the Hennigian sense, &#8220;holophyly&#8221; (Ashlock 1971, Carroll 1988). Holophyletic taxa have been termed clades. Paraphyletic taxa are referred to as grades.

Cladistics is principally concerned with the mapping of cladogenetic evolution&#8212;the branching or divergent component of evolution, whereby two lineages arising from shared ancestry, pursue different evolutionary trajectories over time. A perfect example would be Deinonychosauria and Avialae, or humans, and chimps. Anagenesis is a comparatively simple process whereby one taxon is the phyletic progenitor of another: a linear progression from one form to another without divergence between lineages. Due to this bias of cladistics, the term grade, introduced above, has also been applied to taxa, which display anagenetic change over time, and lack a branching component in their phylogeny.

Semantics aside, cladistic systematists are primarily occupied with delimiting clades, and this is done on the basis of derived traits, or apomorphies, which are uniquely shared by the members of the putatively holophyletic assemblage, thus becoming synapomorphies. Concomitantly systematists seek to avoid erecting taxa based on &#8220;primitive&#8221; characters, or plesiomorphies, as such characters bear no relevance to phylogenetic reconstruction. As one moves &#8220;up&#8221; or &#8220;down&#8221; the Linnean hierarchy, any given character takes on a different status relative to the inclusiveness of the taxon being considered. For example, while a notochord is a synapomorphy of Chordata, within higher chordates a notochord it is a shared basal trait, or symplesiomorphy, and thus irrelevant to sorting out the interrelationships of chordates themselves.

Isolating any given character in a given lineage as either an apomorphy or plesiomorphy, requires comparison with others, and this is generally accomplished through a process called outgroup analysis. Contrasting character states found in taxon A (the ingroup) to those of its nearest relations, say taxa B and C, permits the determination of character polarity and thus differentiates basal from derived traits. Characters shared between the ingroup and outgroups, are considered plesiomorphic, while those uniquely excluded from outgroups are seen as apomorphic. In other words, traits common to taxa A, B, and C are of no phylogenetic relevance in upholding or rejecting the holophyly of any of these taxa, while oppositely, characters unique to taxon A and excluded from B and C are autapomorphic of taxon A and support holophyly thereof (Hennig 1966, Kluge 1977, Patterson 1981, Wiley 1981, Carroll 1988). The choice of outgroup is, in turn, determined by previous cladistic analyses.

In practice, the simple conceptual basis of outgroup analysis is muddled by ontogeny which does not often conform to our view of how it should appear. Such difficulties are most apparent in paleontology where the vagaries of the fossil record compound the matter, and for this reason paleontological systematists have classically used stratigraphy as a &#8220;yardstick&#8221; by which to quantify character polarity in the face of ontogenetic ambiguities (Carroll 1988). A sterling example of the usefulness of this method can be found in E. D. Cope&#8217;s work on establishing the character polarity of molar cusps in Cenozoic mammals (summarized in Butler 1982 and Carroll 1988).

And yet using stratigraphic data in this fashion to help tease out polarity of characters is fraught with problems. Foremost among the difficulties includes the inability to accurately isolate a character as plesiomorphic for a taxon simply because it appears at an earlier geologic horizon, as evidenced by the longevity of many genera and species and the fact that contrary to Hennig&#8217;s initial assertion in 1966, ancestral taxa need not suffer extinction following cladogenesis (Wiley 1981). Thus, one cannot solely rely on the geological occurrence of taxa for calculating polarity (Schaeffer, Hecht & Eldredge 1972, Carroll 1988), which in turn has led some practitioners of cladistic analysis to argue for reducing the role of the fossil record in establishing polarity (Hennig 1981, Wiley 1981, Patterson 1981).

In addition to a hodgepodge of other methods for establishing character polarity (summarized in Carroll 1988), one of the most useful criteria employed by systematists are structural morphoclines, characters, which vary quantitatively throughout a taxon. It is a general pattern of evolution that such morphoclines are directed more or less one way, for example, once lost, digits are rarely recapitulated. However, even this method for helping to elucidate polarity is prey to the vagaries of evolution, and like any other method, cannot be exclusively relied upon.

Once clades have been delimited, the inferred phylogeny can be presented in graphical form. Cladistics has employed a concise graphic to represent hypothetical phylogenies, a cladogram. Cladograms are constructed on the supposition that cladogenesis follows a bifurcating pattern, allying sister clades -- clades sharing most recent common ancestry -- in the context of holophyletic containing clades. While some researchers have argued that such a bifurcating pattern is not in fact reflective of the majority of cladogenetic events, there remains no convincing data to suggest otherwise.

A cladogram is constructed through statistical analysis of a data set (characters), which seeks to optimize the number of synapomorphies underwriting the holophyly of each node of the cladogram. Due to the size and complexity of the data set to be analyzed, computer programs such as PAUP (Phylogenetic Analysis Using Parsimony) are employed to more efficiently collate (Swafford 1991, Paul 2002). The resultant cladograms are then compared on the basis of parsimony&#8212;the postulate that the simplest cladogram requiring the least number of reversals and convergences will be most accurate (more or less Occam&#8217;s Razor applied to phylogeny).

Thus formulated, a cladogram represents a quantified hypothesis as to the evolutionary history of any given group, which can be tested against the data provided by multiple sources, including the paleontological record of whatever group was being analyzed. This ability to put together a coherent phylogenetic hypothesis, express it simply, and then subsequently test it against further data is perhaps the single greatest advantage offered by cladograms over other methods of phylogenetic mapping.

It is important to bear in mind, however, that cladograms are not absolute and immutable representations of phylogeny. Several researchers have rightly criticized the tendency of some to interpret cladograms too literally, in the process failing to realize that the utility of cladograms is their ability to present phylogenetic hypotheses (Halstead 1982, Carroll 1988). As happens all to often in science, certain researchers have in turn further distorted the situation by conversely claiming that cladograms are nothing more than mere speculation, produced through fairly meaningless &#8220;number crunching&#8221; and that they are adhered to dogmatically (see especially Feduccia 1996, 1999). This is as inaccurate as the opposing fallacy that cladograms are unassailably correct. The actual fact of the matter is that cladograms subsequently shown to be incorrect in their phylogenetic hypotheses have been abandoned, while those, which remain well substantiated by the data at hand, continue to be used. Like other scientific hypotheses, those generated cladistically are both open to further corroboration, or falsification in light of new data.

This general review of cladistic methodology already hints at the tremendous acrimony amongst systematists as to the utility of cladistics, and the extent to which it can be accurately used in phylogenetic reconstruction. Moreover, misrepresentations of cladistics abound. For instance, cladistics is not synonymous with phylogenetic reconstruction; rather it is a means towards that end. Cladistics is furthermore not synonymous with systematics or taxonomy, and indeed its goals are far more restricted than either of those broad disciplines (e.g., cladistics does not set out to define or redefine the species concept). Cladistics was explicitly outlined as a procedure for mapping the patterns of evolution in higher operational taxonomic units, and that is where its greatest strength will always lie.

In the final analysis, almost all researchers agree that taxa must be delimited on the basis of synapomorphies and not symplesiomorphies (Sereno 1990). The consensus ends there, however. Some of the more contentious issues pertaining to the utility of cladistics are summarized below.