All hail King Ludd

Cladistics: a Guide to Biological Classification (third edition). By Williams, D., & Ebach, M. (2020). Systematics Association Special Volume Series). Cambridge: Cambridge University Press. https://doi.org/10.1017/9781139047678

This volume is identified as an updated third edition of the Systematics Association “Cladistics” and described as (on the back cover) “a contemporary review of cladistics”, “covers cladistics in the era of molecular data”, and as a “practical guide”. In my view, it is none of these things. The edition moniker is misleading, the book being in no way an updated extension of previous editions. Even the title “Cladistics” is misleading—at least in the way the term is used by practicing systematists today. The book would have been more descriptively titled “Three Item Analysis: A Resurrectionist Tale”. The volume offers a 1970s-centric view of systematics that bears almost no relationship to current cladistic practice, has very little to say about molecular data (other than it is “phenetic”, e.g. p 230, 269), modern computational tools (other than to say they “distort”—whatever that means—data, e.g. p 297), attempts to redefine commonly used terms, and as a “practical guide” is miscast.

I found many sections tough going requiring multiple readings of sentences and sections. This was not helped by non-standard use of terms and redefining those terms in ways that I found confusing. This includes frequent use of the word “true” in terms of analytical results, and synonymizing (p 28) taxonomy, classification, systematics, and phylogeny. This coupled with much epicycular attention paid to arguments long over, such as 1980s “meta-taxa”, do not enhance flow or intelligibility. As a general historical narrative such sections might be fine, but the authors are (to my reading) attempting to make an argument about how people should do systematics today.

There are many points that one can take issue with. I will focus on two here, first the use of the terms “parsimony” and “phenetics” and second, the general and specific tenor focused on Three-Item/Taxon-Analysis. Throughout the volume, “parsimony” as we know it in systematics is referred to as “Wagner parsimony”. Idiosyncratic, but not terrible. In its original sense, this was to signify the reversibility of character change, as opposed to models which biased change in favour of either origin or loss. The modern use of the term is simply “parsimony”. This appears to be done to remove the “unnatural” Wagner parsimony from its pedestal and replace it with the “natural” Three-Item-Analysis favoured by the authors. More jarring is the authors’ use of the term “phenetics”, which seems to apply to any procedure they do not like.

The common current use of the term phenetic applies to tree construction procedures that employ overall similarity as an optimality criterion or at least operational guide such as UPGMA and its diverse variations. The authors lump likelihood and Bayesian methods as well as some parsimony procedures as “phenetic” into the same pile. This is an obvious distortion. Each of these methods contains a specific optimality criterion that is definitely not that of overall similarity. The authors also restate the old canard that combination of data from diverse sources (i.e. “total evidence”), often in great numbers, irrespective of how you treat them, is “phenetic”. This, of course, includes any phylogenomic analysis, and for that matter—seemingly any analysis including molecular sequence data. Methods employing a data matrix (p 230) are “phenetic”, meaning “bad”. Such use abjures the intellectual content (even if pheneticists first demanded such matrices) of the term reducing it to simple name calling.

I do not have any idea what this means. I am unfamiliar with any sort of data manipulation—though it sounds bad—that parsimony-based programs are doing (and I’ve written some of them).

Yet the most important component of the book is the Three-Item-Analysis (3IA) section (Chapter 12). Most of the descriptions, definitions, and historical diversions in the entire volume are cast in the light of and build up to 3IA. For those who participated in some of these discussions in the early 1990s, 3IA was an initially interesting idea that rapidly revealed pathological behaviours and theoretical shortcomings and was discarded (some of these criticisms are cited in the volume). I have no wish to go into the specifics in this review, the papers can be dusted off for all to read. In a plea for relevance, the volume describes purportedly active areas of informatics 3IA research that have recently emerged producing LisBeth (in 2008) and Taxodium (in 2012). In the 9 years since these “recently” produced implementations have been available, LisBeth has been cited seven times and Taxodium 16 times (via Publish or Perish; https://harzing.com/resources/publish-or-perish?source=popmac_7.33.3388.7819). This is hardly consistent with an active research topic. To my knowledge, Norman Platnick himself did not employ 3IA in much (if any) of his extensive empirical work.

On the mechanical side, I found the organizational aspects of the book are a bit wanting. The chapters range from 5 to nearly 70 pages, extensive textbook-like “Box” sections do not help the flow of arguments as they can go on for pages and many would have better been incorporated into the text. As mentioned above, the writing is often difficult to follow, uses un- or semi-defined terms (e.g. “natural”, “true”) and argumentation schemes often feel contorted.

The final chapter on “Further Myths and Misunderstandings” criticizes popular books, non-systematists’ views of systematics, and discusses recent personality and other issues including the so-called “Parsimony-gate” episode. This chapter seemed a bit of a dog’s breakfast of topics the authors felt like discussing with seemingly little connection to the previous 14 chapters of the volume.

To summarize, I do not find this volume to be on “Cladistics: A Guide to Biological Classification”. The volume is a manifesto in support of 3IA and should simply have been named as such. I could not recommend this volume to students as “cladistics”, “a practical guide”, or as a valuable take on the “era of molecular data.” There is no reason that the authors should not make their arguments in the scientific literature—just that the views and arguments they express would be better served by a (possibly lengthy) peer-reviewed journal article.