Systematics and the Fossil Record

Describing phylogenetic relationships among fossil taxa has been an important part of paleontology since the 19th century. In the last decade, the proliferation of phylogenetic estimates generated by computer-run cladistic analyses has resulted in a reappraisal of the importance of phylogenetics in paleontological studies. Andrew Smith's book Systematics and the Fossil Record represents a powerful presentation of the cladistic agenda for paleobiological studies. Because it touches on nearly every aspect of paleontology, all paleontologists should read this book soon. However, I emphasize the word "soon." Despite the claim on the back of the book by Derek Briggs (i.e., that there will be a time when no paleontologists will use taxonomic data without consulting Smith's book), this book really is more like last year's computers. That is, Systematics and the Fossil Record is extremely useful and thought provoking, but parts of it already are antiquated and much of the rest of it will become outdated in the near future, partly because of the research that it will inspire. Readers with a background in systematic theory obviously will have a much easier time appreciating Smith's book, but Smith does include a chapter describing cladistic methods. Unfortunately, this chapter conveys the impression that systematists have universally agreed on these methods when many aspects of cladistic methodology are still hotly debated. For example, there is much debate about how workers should treat multistate characters (i.e., when, if ever, they should be ordered or unordered; Hauser and Presch, 1991; Slowinski, 1993), missing data (e.g., see Maddison, 1993), polymorphic characters (including both polymorphisms within species and variation among species within higher taxa), total evidence (i.e., whether one should include different types of data, such as different molecular and morphologic characters, in the same parsimony analysis; see Swofford, 1991; de Queiroz, 1993; Eernisse and Kluge, 1993; Bull et al., 1993), and character weighting (Neff, 1986; Bryant, 1989; Chappill, 1989; Farris, 1990). Because Smith does not discuss the issues in detail (which is understandable, given that full treatments would have doubled the length of the book), his recommendations come across as assertions. Indeed, some of Smith's counsels are contradictory. For example, in his discussion of coding inapplicable data (i.e., how to code a toothless species for characters about specific tooth types), Smith advises the multistate character state schemes such as: 1: No teeth; 2: Tooth Type A; 3: Tooth Type B. However, in a previous discussion, Smith recommends treating multistate characters as unordered rather than ordered because unordered characters (supposedly) make fewer a priori assumptions. When applying this logic to multistate solutions for inapplicable characters, Smith's recommended coding essentially assumes that the different tooth types have nothing to do with one another (i.e., that they are not necessarily homologous at the level of being teeth). Is this really preferable to ordering the characters (which would assume that teeth are homologous)? Several other difficulties render many of Smith's suggestions problematic in anything other than simple situations. Consider, for example, what happens when one wishes to use several different types of tooth characters. Thus, although I think that paleontologists lacking a strong background in systematic theory will understand the book, the book is not really appropriate for learning cladistic methods. The real purpose of Smith's book is not to provide a tutorial for cookbook cladistics, but instead to present an outline of the potential implications of phylogenetic studies on previous paleobiological studies and what paleontologists can do within a phylogenetic estimate. There are some lapses, however, especially when Smith is detailing the inadequacies of the fossil record. For example, Smith suggests that the apparent tendency for orders to appear in nearshore habitats, as documented by Jablonski and Bottjer (1990a, b), is an artifact of preservational patterns. Counter-arguments to Smith's criticisms are presented elsewhere (Jablonski and Sepkoski, in press); what is notably absent is a description of how Jablonski & Bottjer's hypothesis might be better tested within a phylogenetic context or if there are any phylogenetic analyses that might contradict their hypothesis. Also, although Smith prefaces the book with the opinion that a phylogenetic approach better compensates for sampling than do other approaches, Smith gives little discussion to the effects of sampling on the robustness of cladistic analyses. Smith's discussion of higher taxa took me slightly by surprise. Instead of dismissing all higher taxa (as have some cladists; e.g., de Queiroz and Gauthier, 1990), Smith advocates retaining the Linnaean system if only for communication. Smith also acknowledges that paraphyletic taxa will corrupt diversity studies largely in the tallying of extinctions, and then only if the "extinction" of a paraphylum does not accompany the true extinction of a lineage within that taxon (see Fig. 1). Interestingly, Smith tacitly advocates a taxonomic scheme advocated (also tacitly) by Sepkoski and Kendrick (1993, a citation curiously absent from Smith's book): i.e., taxonomies in which both paraphyletic and monophyletic taxa always encompass monophyletic groups at any point in time and always terminate with the extinction of a lineage or subclade (as in Fig. 1A). What is not made clear by Smith is that where paraphyletic taxa are concerned, noise will not be created by arbitrary diagnoses of higher taxa, but by diagnoses systematically biased by stratigraphy (e.g., even though a Late