ASBS Newsletter – Book Review
Plant Taxonomy: The Systematic Evaluation of Comparative Data
written by Tod F. Stuessy
(From ASBS Newsletter Number 74, March 1993)
Published by: Columbia University Press, New York. 1990.
xvii+514 pp. ISBN 0-231-06784-4.
This review continues the series started in earlier issues of the Newsletter (70: 30-33, 71: 32-36, 72: 24-27), assessing the ability of the current crop of plant taxonomy textbooks to present systematics as an exciting modern science, rather than as simply a traditional scholarly exercise. Therefore, I am interested in their ability to present taxonomy as a search for general truths based on explicitly-stated testable hypotheses, rather than simply as the acceptance of pragmatic solutions that are not too inconvenient.
This particular book is explicitly aimed at university undergraduate and postgraduate students undertaking a subject covering plant systematics. It would thus be completely unsuitable for the general public, as it assumes a fair knowledge of introductory biology. It is concerned with the "philosophical and theoretical aspects of plant taxonomy" rather than the practical parts, and so there is no mention of identification techniques, there are no family descriptions, and there is no list of terminology.
The book is organized into 25 chapters in two parts (Principles of Taxonomy, Taxonomic Data) and five sections (The Meaning of Classification, Different Approaches to Biological Classification, Concepts of Categories, all in Part One; Types of Data, Handling of Data, both in Part Two), plus an Epilogue. The publication quality is generally good, with typographical errors relatively rare (although Table 25.1 needs some considerable help). The text is broken up with numerous tables and figures, either providing an overview of topics not covered in the text or providing specific examples of techniques. The subject index is relatively poor, covering major subject headings only, although there are separate taxon and author indexes. The bibliography covers just over 2,900 references, mostly from the period 1968-1986, and the illustrative examples are fairly cosmopolitan, covering all continents but focusing on North America (where Ohio gets more mentions than its intrinsic interest warrants).
In writing this book Stuessy has set himself a big task, as he unambiguously sets out to provide a replacement for Davis & Heywood's Principles of Angiosperm Taxonomy (1963), claiming that the recent books by Radford, Sivarajan, and Stace "lack the detail of coverage of most topics". Indeed, the most-cited reference in the early chapters is in fact Davis & Heywood's book, and Stuessy by and large makes a pretty good fist of updating it. This is certainly the best exposition of the principles of systematics in any of the books reviewed, and it is far more readable than the similar book by Sivarajan. However, the trade-off is that the practice of systematics (e.g. nomenclature, identification) is largely ignored.
In many ways this book is a successful attempt to present systematics as a science - Stuessy certainly has a grasp of his subject, and he has clearly thought long and hard about where it has been and where it is going. The intellectual debates are presented as fresh and stimulating, and the revolutions that have occurred in the last 40 years are seen as having advanced systematics several light years from the old-world attitudes of the first half of this century. Indeed, comments such as the following by leading systematists are given short shrift: "I would suggest that instead of having further conferences of this kind to discuss taxonomic philosophies, that we forget about taxonomic philosophy and go back to doing taxonomy." Such an appeal for the dark ages of stamp collecting has little relevance for systematics today.
Unfortunately, each chapter in the book ends with recommendations for current taxonomic practice, all of which imply that the changes are now over and that we can now settle down into a new complacency. In other words, systematics need progress no further. This may not be Stuessy's intention, and if so then it is flaw in the organization of the book. It would have been nice, however, to be left with a feeling that there are new horizons to reach for, and that a new generation of taxonomists would be attracted to the work because of its intrinsic fascination rather than because it's a cosy refuge from the world.
Furthermore, in the process of providing a detailed discussion of taxonomic principles, Stuessy inevitably (and quite consciously) comes down in favour of particular approaches to taxonomy as a scientific exercise. In particular, he is an ardent supporter of phyletic (or evolutionary) taxonomy, rather than of cladistic or phenetic taxonomy. This needs to be made clear, because such a viewpoint unavoidably colours almost everything in the book, and the book will therefore continually annoy anyone who disagrees with this stance. This is not necessarily a bad thing, because science (in practice) is basically one long argument between competing points-of-view, each of which has adherents who seek to express their ideas as convincingly as possible. However, Stuessy's attempts at "a balanced presentation" of alternative philosophies are not always successful, although he is to be commended for his laudable attempts to make it clear when he is expressing a personal opinion. Unfortunately, most of the summary and conclusion sections for each chapter are biased by personal opinions, and this is not necessarily obvious to the uninitiated.
The tri-partite nature of systematics. From Stuessy (1990).
This bias in view-point is compounded by the fact that the book is more out-of-date than its publication date would suggest. The cut-off date for new literature additions is stated in the Preface to be mid-1988, and yet the peak of the bibliographic citations is 1983-84 (see graph). This means that the coverage of topics like cladistics and genomic sequencing are quite out-of-date, and this seriously limits the utility (and fairness) of the discussions. It is also interesting to note from the graph that there are actually three peaks of reference citations, 1971-72, 1978-79 and 1983-84, although it seems unlikely that this should coincide with the appearance of important papers in the literature. The dip of citations in 1973 (a drop of 20% compared to the years either side of it) also seems odd, presumably telling us more about the behaviour of Tod Stuessy at that time than of botanists in general.
Frequency histogram of the references from 1938
onwards cited by Stuessy (1990).
There are citations for a further 200 pre-1938 references.
Stuessy's personal opinions also creep into the illustrative examples. The author index reveals that he cites his own work twice as often as his nearest rival (Vernon Heywood), which is perhaps going a bit too far even allowing for a normal degree of pride in one's own work. Stuessy also favours the use of the older (and less consistent) family names, unlike any of the other authors reviewed. Indeed, he several times points out that he is using older and/or less appropriate terminology simply because it "has been used extensively in the literature", and at least once he uses terms in ways that blatantly contradict their traditional usage in other disciplines (while complaining about other people doing the same thing to biological terms). Surely the best way to get rid of inappropriate (and therefore confusing) terminology is to eliminate it from the textbooks, so that the next generation of taxonomists is not handicapped by it?
The book starts with A Few Definitions (9 pages), which sets the scene for Part One. Unfortunately, the scene is not all good. Stuessy takes the nine pages to define only eight terms, doing so by providing a seemingly endless series of alternative definitions with very little attempt to synthesize them. This is symptomatic of the rest of this Part, where large amounts of seemingly irrelevant historical detail obscure the expected synthesis of concepts. This is not to imply that the historical reviews are not good, as indeed they are mostly excellent, but it is meant to suggest that a long list of unintegrated definitions (no matter how clear they are) is rather boring to read.
Chapter two, on The Relevance of Systematics (6 pages), provides some interesting views on what contributions we have made to the world, although its concentration on evolution is not necessarily strictly relevant to the focus on classification in the rest of the book. The ensuing chapters on The Importance and Universality of Classification (7 pages) and Characters (16 pages) are good introductions to their topics, although the discussion of cladistic characters suggests that "generalized vs. specialized" is synonymous with "general vs unique".
Section Two, on classification, begins with an unusually brief historical introduction, The Anatomy of Classification and the Artificial Approach (6 pages), although any more detailed coverage would perhaps simply duplicate Core's Plant Taxonomy (1955). The chapter on Natural and Phyletic Approaches (8 pages) covers the intuitive approaches of the last 100 years, but it tends to ignore the current practitioners of this school, which is not helpful to those students who will encounter their works.
The chapters on the Phenetic Approach (34 pages) and the Cladistic Approach (42 pages) are far and away the longest ones in the book, indicating their influence on taxonomic principles. Unfortunately, phenetics is treated as an established technique while cladistics is treated as a newcomer, in spite of the fact that their inception was very nearly contemporaneous. This means that phenetics gets a more balanced presentation as an established discipline, while cladistics is presented as still waging a "battle" for recognition. This is mainly a product of the out-of-dateness of the references - cladistics in the last few years has rapidly become the mode of choice for the presentation of phylogenetic information. Therefore, the discussion of phenetics and its impact on systematics is very good, while that of cladistics leaves something to be desired. Nevertheless, cladistics is presented as an explicit science, which is a refreshing change from most of the other books that I have reviewed.
One of the biggest difficulties in the cladistic chapter is that Stuessy tends to treat cladistics as synonymous solely with the parsimony approach, which may have once been true but is no longer very accurate. He thus produces long discussions of material that is now irrelevant (or at least less important than it once was). For example, all cladograms are treated as necessarily being rooted, so that character polarities must be decided first, and this is treated as being a major problem. However, functional outgroup analysis can be used, where an unrooted tree is first produced using unordered characters, and then the character polarities are determined by rooting the tree with an outgroup. This possibility is treated as an aside towards the end of the chapter, whereas it is the fundamental basis of almost all analyses of molecular sequence data. The discussion of character polarities is thus nothing more than a long self-justification for the ideas presented in the paper by Crisci & Stuessy (1980), which moves well outside generally-accepted cladistic practice.
The balance of topics is less than felicitous elsewhere in the chapter, as well. For example, the longest section is on clique techniques (which are rarely used these days), whilst distance tree techniques and maximum likelihood techniques are hardly mentioned at all. The discussion of the efficacy of the tree-building algorithms is thus rather weak. A better framework for this section would be to compare the likelihood that the model on which each algorithm is based is in fact real. No technique in science will ever be a final answer, and therefore we must make conscious decisions about what aspects of reality we are trying to incorporate into our reconstruction of the world.
The section on cladistic classification is also rather brief and inaccurate. The comment is made that "paraphyletic groups to some cladists are not useful for classification", which seems to be a little disingenuous, since most cladists seem to accept this. The argument for accepting paraphyletic groups is based on the use of both patristic (degree of divergence) and cladistic (branching pattern) information. If the patristic data are included, then some of the descendants of an ancestor may be placed in a separate group in the classification (because they have diverged sufficiently from the other descendants), leaving the remainder of the descendants as a paraphyletic group. This classification will then contain more information (patristic + cladistic) than the equivalent purely cladistic one. However, the point is not which classification contains more information, but which one allows you to retrieve that information. A cladistic classification allows you to retrieve all of the cladistic information that has gone into it, but the alternative classification does not allow you to retrieve any information, since you don't know whether any one part of the classification was based on the patristic or the cladistic information. Consequently, it is difficult to be a cladist and to accept paraphyletic groups at the same time.
It is also worth noting that not all of the branching pattern needs necessarily to be incorporated into the classification, as implied by Stuessy. Thus, not all of the dichotomies of the cladogram have to represent groups in the classification; however, all groups in the classification should represent a branch on the cladogram. Similarly, parts of the cladogram that are not fully resolved into dichotomous branches allow the formation of so-called metaphyletic groups, which are not resolvable into either monophyletic or paraphyletic groups given the current data set.
The assessment of the impact of cladistics on taxonomy is marred by a confusion of phylogeny and classification. Cladistics is a method of reconstructing the phylogeny of a group of organisms, and this phylogeny may (or may not) then be used to construct a classification. Consequently, cladistics produces an explicitly-stated and testable hypothesis concerning the phylogeny - it does not produce a testable classification, as claimed by Stuessy. The production of a classification is thus a by-product of the phylogeny, rather than an integral part of it (other potential by-products include, for example, the analysis of evolutionary patterns and the analysis of biogeography). This same confusion of trees and classifications carries over into the next chapter as well, to its considerable detriment. Stuessy also, strangely enough, considers that cladistic analysis is "most useful" at the specific and generic levels, although there is no clear justification for this conclusion.
The final chapter in this section, Evaluation of the Three Major Approaches: The New Phyletics (19 pages), is an attempt to synthesize the best bits of the phenetic, cladistic, and phyletic approaches to classification. As far as I can make out, the New Phyletics is something like a cladistic analysis but in which plesiomorphies turn out to be useful after all. Unfortunately, determining precisely when these plesiomorphies are useful does not appear to be able to be made explicit, thus leaving a very large subjective component in an otherwise laudable attempt to be objective.
However, one of the main justifications for the New Phyletics is that the resulting classifications contain more information (patristic + cladistic, as discussed above), and so they suffer the problem of inadequate information retrieval. They are thus less predictive than the equivalent cladistic classification, not more predictive as claimed by Stuessy. Phyletics is also claimed to have more "efficiency" when producing classifications compared to cladistics, although this concept is equated solely with the speed of intuitive judgements. This is like preferring Newton's equations for planetary motion to those of Einstein - they produce answers faster, even if these answers are known to be inaccurate. Much is also made in this chapter of the number of characters used to construct classifications (more is better), but it is clear that it is the congruence of the characters rather than their quantity that is responsible for the stability of the resulting classifications.
Unfortunately, Stuessy's attempt to produce a new science of taxonomy are thus weakened because he apparently cannot give up the traditional aesthetic subjective judgements that he presumably learnt to use during his own training. His comment that "the majority of practicing taxonomists, despite the furor of past decades over phenetics and cladistics, still make classifications phyletically" is more a reflection of the general demographic aging of the taxonomic population than a justification for current taxonomic practice. The New Phyletics thus turns out to be a way for traditional taxonomists to pretend that they are keeping up-to-date without actually changing their behaviour in any substantive way. This chapter is therefore the weakest part of the book by far.
The third section covers the categories used in taxonomy. It begins with The Taxonomic Hierarchy (5 pages), followed by chapters on Species (21 pages), Subspecies, Variety and Form (12 pages), Genus (13 pages), and Family and Higher Categories (7 pages). Each chapter starts with an historical overview, followed by a discussion of relevant points, ending with a recommendation for usage. Most of the discussion is very good, but there are a number of points where issue can be taken. For example, Stuessy suggests that classifications based on ordination analyses do not have the "predictive characteristics of classes that result from hierarchical classification". However, this will not be true if evolution is reticulate, as he claims elsewhere in the book (dichotomous speciation is referred to as "a most dubious assumption") - under these circumstances then ordinations will be more predictive.
The discussion of species is suitably detailed given the amount of time that biologists have devoted to the concept, but it is marred by loose thinking in places. For example, the discussion of whether species are classes or individuals shows considerable confusion of transformational versus variational evolution (Charles Darwin's contribution to biology was to point out that individuals undergo transformational evolution while classes undergo variational evolution). Furthermore, the chapter ends with a recommendation for using the biological species concept, which is at odds with the claim that species "have evolutionary reality". As Ernst Mayr has repeatedly pointed out, the biological species concept is atemporal, and thus has no evolutionary derivation. Its use in a phylogenetic context is thus meaningless, and it is therefore of doubtful utility if the taxonomic hierarchy is to reflect phylogeny. The biological species concept is also different depending on whether it is defined in a negative manner (species do not interbreed with each other) or in a positive manner (species are collections of interbreeding populations) - biological species are thus units of convenience based on pragmatism.
The discussion of genera is another area that suffers from out-of-date references, notably in the claim that "only a few studies so far of plant groups have dealt with determining generic limits using cladistics". Stuessy disparagingly refers to the "dogmatic cladistic view" that regards genera as natural only if they are based on synapomorphies, preferring to give formal taxonomic recognition to groups that are "conservative in size and shape of certain features reflecting clear morphological discontinuities". At the higher taxonomic levels he also refers to "the evolutionary value of paraphyletic groups", although he never presents any "value" other than pragmatic utility. He also mis-represents the arguments about the paraphyletic nature of montypic genera, since their sister groups are not necessarily paraphyletic.
Part Two of the book discusses taxonomic data. This part has no real explanation of the terminology used (unlike Part One), and there is no systematic survey. It is mainly a brief introduction to each of the types of data used by taxonomists, with comprehensive references to other works that provide more details, followed by many illustrative examples of their uses and usefulness in classification. It covers Morphology (15 pages), Anatomy (18 pages), Embryology (16 pages), Palynology (21 pages), Cytology (25 pages), Genetics and Cytogenetics (16 pages), Chemistry (22 pages), Reproductive Biology (13 pages), and Ecology (16 pages), the relative chapter lengths accurately reflecting the amount of detail presented.
However, the arrangement of the information among the chapters is not always reasonable. For example, the discussion of DNA sequencing is in the chapter on chemistry, while allozymes are discussed in the chapter on genetics. The argument presented for this arrangement is an example of the out-of-date nature of a number of the concepts in this section. This is highlighted by the extremely inadequate nature of the description of DNA sequencing techniques, with polymerase chain reactions not even being mentioned. Furthermore, Stuessy tries to convince us that "phenetic algorithms are best [for] base pair sequence data", which must come as a shock to those of you who have been happily doing cladistic analyses on these data for years. His discussion of the utility of cytogenetics in cladistic analyses is also very confused, because he fails to acknowledge that cladistics is concerned with cladistic information not necessarily with patristic information (knowledge of inter-taxon crossability is useful for the latter but not for the former).
The chapter on ecology also has a few minor inadequacies. Stuessy manages to list several works on plant population ecology without reference to Harper's Population Biology of Plants (1977) or to Silvertown's Introduction to Plant Population Ecology (1987); and Greig-Smith's Quantitative Plant Ecology (1983) is also ignored completely. He also reproduces what looks like the most outrageous vegetation map of the world that I have ever seen (the entire east coast of Australia is occupied by "Temperate rain forests", in spite of the fact that "Sclerophyllous vegetation" is also a map unit); and he suggests that hybridization between introduced weeds and native plants is not "natural" (thus ignoring the dynamic nature of the relationships between species). His (brief) discussion of biogeography also presents the vicariance versus dispersal debate as a discussion of "whether one or the other is the all-inclusive general explanation" rather than a discussion of whether one or the other produces testable hypotheses explaining plant distribution. The same confusion mars the equally-brief discussion in the chapter on cladistics, where dispersal is referred to as "non-informative" rather than "untestable".
The book ends with two practical chapters on Gathering of Data (8 pages) and Presentation of Data (15 pages), which are unique to this book among all of those reviewed. These chapters open up a whole new area of offering advice on the practical aspects of being a taxonomist, which could be usefully expanded into a book in its own right. However, I don't think that I'd recommend Sokal & Rohlf's Biometry (1981) as a "starter" for data analysis, as it is too detailed; perhaps their Introduction to Biostatistics (1988) would be more appropriate.
All in all, I enjoyed this book, in spite of my criticisms. This is partly because Stuessy has presented enough of each concept to stimulate debate while having the forthrightness to make his own opinions clear, and partly because it avoids most of the "boring" bits of taxonomy. However, I can't help being disappointed with the conservative nature of the pragmatism that creeps into all of the conclusions that are reached. In the final analysis, the book is Stuessy's magnum opus, where he provides a justification for the attitudes that he holds and the work that he has done. As a university text-book, therefore, it has serious flaws, because it still points towards the recent past rather than the recent future. Anyone who is determined to play an active part in the future development of systematics will thus find this book very annoying, since it has the overall flavour of a current pause in progress rather than of a continuing advance.
So, what conclusions can I come to about all of the books reviewed so far in this series? The book by Charles Jeffrey is the most "traditional" of them all, concentrating on nomenclature rather than on principles, although it is in the process of being usefully updated. The book by Clive Stace has rather too much of a hangover from the interest in biosystematics that grew in the 1960s and 1970s, and rather too little about modern principles. The book by Samuel Jones & Arlene Luchsinger is the most readable general introduction to taxonomic principles and especially practice, but its almost exclusive focus on American plants means that at least half of the book is of dubious value. The book by Albert Radford is far and away the best compendium of taxonomic principles and practice, but it is almost unreadable as a textbook and so is more useful as a reference source. The book by V.V. Sivarajan is a good introduction to the principles of taxonomy, but no-one should pay good money for such poor publishing quality. Finally, the book by Tod Stuessy is the most readable introduction to the principles (but not the practice), provided you realize (and to some extent agree with) his personal viewpoint. So, if you're looking for a book to convince someone else that systematics is a modern science, then Stuessy's book comes closest to what you want, but I don't think you'll really convince anyone.
Reviewer: David Morrison
Department of Applied Biology
University of Technology, Sydney