Portal de Periódicos da CAPES

Blasts from the past

2019; Wiley; Volume: 35; Issue: 6 Linguagem: Inglês

10.1111/cla.12378

1096-0031

Andrew V. Z. Brower,

Yersinia bacterium, plague, ectoparasites research

Biology in Transition: The Life and Lectures of Arthur Milnes Marshall. Curated and Annotated By Martin Luck. Pelagic Publishing, Exeter, UK, 2018. 423 pp. U.S. $95.00 (hardbound), ISBN 978-1-78427-166-4; ebook ISBN 978-1-78427-167-1 Phylogenetic Inference, Selection Theory, and History of Science: Selected Papers of A.W.F. Edwards with Commentaries. By Rasmus Grønfeldt Winther. Cambridge University Press, Cambridge, UK, 2018, 540 pp. U.S. $89.99 (hardbound) ISBN 978-1-107-11172-1 Resurrecting old publications as new historico-commemorative monographs brings recognition to works that are perceived to have been overlooked over the course of time, and pays tribute to their authors. Some have undertaken such exercises in honour of themselves (e.g. Mayr, 1988), and recycling previously published conference proceedings and chapters from edited volumes seems a common practice among philosophers and historians of science (e.g. Sober, 1993; Rieppel, 2016). The two books reviewed here, although both about Cantabrigian scientists, are quite different––Luck's draws attention to work from the 1880s that was published privately and is quite obscure, whereas Winther's advances a Whig history of systematics through the compilation of relatively well-known and readily available works, with adulatory commentary by a variety of maximum-likelihood afficionadoes. Marshall is long gone; Edwards has had the opportunity to frame the narrative with postscripts. First, Biology in Transition. Luck describes how he discovered two small, bound volumes in a library donated to his department at the University of Nottingham. These contained posthumously published lectures by Arthur Milnes Marshall, a late 19th Century developmental biologist. Born in 1852, Marshall was trained at Cambridge, and like many promising young biologists of his day, had the opportunity to conduct marine research at Anton Dohrn's renowned field station at Naples. After completing medical school, he became a lecturer at Owens College in Manchester. He was an ardent Darwinian, and also, by accounts, a compelling speaker, lecturing not only at the college, but also to public societies such as the Manchester Microscopical Society and the Ancoats Society (open to the general public, for its education and betterment). During his truncated career (he was killed in a climbing accident in the Lake District in 1893), Marshall only published ten scientific articles, along with a textbook on vertebrate embryology, yet he was highly enough regarded to be elected a Fellow of the Royal Society, in 1885, at the tender age of 33. The book is composed of transcripts of 21 lectures delivered in the period from 1879 to 1893. The first 14, originally bound into a book entitled Biological Lectures and Addresses introduce a variety of topics in natural history and developmental biology. Having been presented at various times and to different audiences, there is some redundancy, but in general Marshall's lectures give an interesting and entertaining overview of the state of evolutionary thinking of the late Victorian period. The other seven transcripts comprised a second volume Lectures on the Darwinian Theory. These, having been presented as a course in 1893, exhibit greater internal continuity, although Marshall does repeat some of his favourite examples and metaphors from the first series. Marshall did a good job explaining natural selection and genealogy, as well as the various sources of evidence supporting Darwin's theory. He mentioned vestigial organs, such as the atrophied but still present muscles in humans that are homologous to those that swivel the ears in other mammals. He also was very fond of Haeckel's theory of recapitulation (derived taxa proceeding through developmental stages resembling their putative ancestors), which he appears to have considered a sort of evolutionary law, even though he presented a variety of instances in which evolution seems to have gone in the opposite direction. As he quipped, “animals in their development climb up their own genealogical trees” (p.208). Man is distinctly an animal – i.e., neither a plant nor a mineral – requiring organized food, for which he is dependent on other animals and on plants. He is distinctly a vertebrate, as proved by his backbone, the relations of his nervous system, brain, heart and sense organs. Further, he clearly belongs to mammals – the presence of hairs instead of scales or feathers would alone be sufficient to show this; but he also possesses all the other characters of the group – viz., two condyles to the skull, seven cervical vertebrae, and a left aortic arch. … From the mid-20th Century onwards, common ancestry became so ingrained as “the explanation” for homologous features that empirical synapomorphies like these frequently have been viewed askance by proper-thinking evolutionists, as typological and essentialist. Luck offers a contextual introduction to each of Marshall's lectures, explaining when and to whom it was presented, as well as including extensive footnotes with details of works mentioned by Marshall, and reflections on the contemporary interpretation of some of the issues he discusses. Occasionally, Luck's understanding struck me as not quite right, particularly in the realm of palaeontology. For example, footnote 24 (p.18) states, “The connection between dinosaurs and reptiles is a much more general one, in the sense that dinosaurs have many characteristics that we see in modern reptiles. Somewhere amongst the very early dinosaurs was the common ancestor of reptiles, birds and mammals, presumably an amniote.” This suggests that his concept of “dinosaurs” includes most or all extinct amniotes (pelycosaurs and so forth), rather than the monophyletic group (including birds) recognized by cladists. Later on, Luck equates the “Secondary Period” with the Devonian (footnote 444, p.307), but this is merely the Victorian term for the Mesozoic Era, as the Tertiary was for the Cenozoic (the Quaternary is still with us). In sum, the material addressed in this book is most likely to be of interest to historians of science, in the vein of Peter Bowler's books on the period (Bowler, 1988; Bowler, 1996). However, it does provide a fascinating glimpse at the state of evolutionary thinking immediately prior to the early 20th Century revolution in biology due to the elucidation of hereditary mechanisms. Much more likely to inflame the passions of this journal's readership is Rasmus Winther's collection of mostly 20+ year-old publications, and more recent commentary, by and about A.W.F. Edwards. Winther is a philosopher of science who appears to have developed an interest in Edwards’ work through the problem of genetics and human races. It is not clear whether this book stems from a meeting celebrating Edwards’ career, or is a stand-alone festschrift volume to mark his 80th year (Winther's interviews with Edwards, conducted in 2015, and transcribed towards the end of the book, are of more social-historical than scientific significance, in my view). Anthony Edwards was born in 1935. He began his training as a student of Ronald Fisher at Cambridge University in the 1950s, and earned a PhD in statistical genetics in 1961. After a peripatetic postdoctoral period, he spent most of his career at Cambridge, and was elected to the Royal Society in 2015. The book reproduces 28 of his some 250 publications (it is hard to tell exactly how many “original works” he published, because the C.V. at the end of the book includes published abstracts, multiple editions of the same books, numerous reviews or rehashes of works already published, and in some instances papers that were reprinted verbatim in more than one publication). When I say “reproduces”, I mean that literally––they are scans of original journal articles, sometimes significantly shrunk to fit the book's smaller format, and occasionally with the edges of the text clipped off. It seems odd that, well into the 21st Century, CUP would opt for such meager production values in a relatively expensive book. Perhaps some only exists as grey literature: as Winther says (p.1), much of Edwards’ work is “difficult to locate, requiring hours of effort by dedicated researchers and librarians.” Hours! Also included are laudatory essays by a number of Edwards’ colleagues and acolytes (names familiar to systematists include Joseph Felsenstein, Rasmus Nielsen, Mark Pagel and Ziheng Yang), the aforementioned interviews, and a section of odds and ends (e.g. Edwards’ C.V., a previously published paper by Winther and the transcript of a 2012 lecture about Fisher by Edwards). The book does not contain a comprehensive reference section, and the index appears to have been compiled from the commentaries and appendices only, rather than encompassing the articles that are the raison d’être of the book. Thus, although Willi Hennig's work is discussed (and dismissed) in several of Edwards’ review papers (e.g. Edwards, 1996; Edwards, 2004; see also the essay by Felsenstein), Hennig is indexed only once, to a single page of the interview section. Steve Farris, although mentioned regularly throughout the book, is not indexed at all. By contrast, Joe Felsenstein is referenced to 39 different pages. By such sleight of hand, Winther rhetorically marginalizes cladistics, the “dogmatic” is superceded by the “pragmatic”, and the narrative is distorted to favour “the best ideas” (Nielsen, p.340). Many of Edwards’ early papers address statistical issues regarding human sex ratios, and most of his later papers treat topics in the history of statistics and genetics. He also wrote a lot about Fisher's Fundamental Theorem of Natural Selection. The works of most interest to readers of Cladistics, and therefore the main subject matter here, are his articles, mainly coauthored with human geneticist Luca Cavalli-Sforza, in which he developed methods for inferring phylogenetic trees. By my count, 23 entries on Edwards’ C.V. relate to inferring phylogenetic trees, and 14 of those are included in the book, so, as may be inferred from the title, it is his supposed seminal role in the development of phylogenetics that is emphasized. The advent of computers in the late 1950s spurred interest in numerical approaches to genetic and taxonomic problems previously viewed as an “art” (Simpson, 1961). In 1961, Cavalli-Sforza, another former student of Fisher's, had compiled a dataset of human blood-group frequencies for 15 populations from around the world, and brought in Edwards to write computer programs to infer a hypothesis of relationships among them. Note that the starting point for these inferences was the phenetic conversion of each locus into an average population data point, which could be plotted into Euclidean space in relation to one another by some sort of distance metric. Initially, Cavalli-Sforza pursued a least-squares distance algorithm––also a phenetic measure––whereas Edwards conceived an approach which he referred to as “minimum evolution”, which, depending on which of his papers in the book you read, either is or is not the same thing as cladistic parsimony. Being ardent Fisherians, Cavalli-Sforza and Edwards also attempted to develop a maximum-likelihood estimator for their data, but these efforts were stymied by technical difficulties. As Ziheng Yang notes in his commentary, they also dabbled with a Bayesian approach (although, being an ardent Fisherian, Edwards is reluctant to admit it). It should be noted that Edwards’ novel contributions to the field ended with his collaboration with Cavalli-Sforza in 1972––all of the papers focused on phylogenetics published after that appear to be retrospectives of his glory days. Edwards (p.435) is unabashed about his operationalist approach to phylogenetic problems: “I, at least, as the technologist, wasn't too bothered with what these populations were. There is a difference between Luca Cavalli and me, because he really wanted to know the answers: How did the human population spread around the world? I really wanted to know the answers for philosophical and statistical problems which were worrying me.” Thus, working and reworking a single dataset, the narrative established in the book is that Cavalli-Sforza and Edwards (1964, 1967) managed to “invent” four different methods of phylogenetic inference in a fell swoop. The method of minimum evolution for reconstructing phylogenetic trees did not stem from Ockham's razor (the principle of parsimony), or from the mistaken notion that evolution proceeds parsimoniously, or from Hennig's phylogenetic systematics. As the original papers make clear, it arose as an approximation of the maximum-likelihood solution obtained on the basis of a model for random evolution (whether continuous or discrete), itself an approach justified by modern concepts of statistical inference. Edwards hastened to distance himself from systematics: “Unlike contemporary practice in taxonomy, the new methods of phylogenetic analysis belonged to the quite separate tradition of genetical model building and statistical estimation theory used, for example, in genetic linkage” (Edwards, 2009). That the conceptual foundation of this work lies in population genetics and not systematics is made abundantly clear when one observes that the data were allele frequencies in populations (not characters), and the entities being grouped were intraspecific (not taxa). This reveals Edwards’ basic failure (and that of his admirers, such as Rasmus Nielsen, whose essay “the Historic Split and Merger of Gene Trees and Species Trees” falls into the same metaphysical quagmire) to appreciate the epistemological and methodological distinctions between tokogeny and phylogeny. There is no reason to assume that conspecific populations should exhibit a strictly bifurcating hierarchy, and the fact that polymorphic populations differ from one another only in allele frequencies is rather good evidence that they do not. Richard Lewontin (1972) noticed this problem, arguing that, “Human racial classification is of no social value and is positively destructive of social and human relations. Since such racial classification is now seen to be of virtually no genetic or taxonomic significance, either, no justification can be offered for its continuance.” Responding 30 years later, Edwards (2003) called this “Lewontin's Fallacy”, arguing that even though 85% of variability occurs within human populations, differences may be discovered by multivariate statistics. Winther (2014, reproduced in the book) justly criticized Edwards’ approach as reifying racial differences through models. If one starts out treating populations as points, it is not hard to see how differences among them might be overemphasized. Another recurrent theme throughout Edwards’ phylogenetic papers is justification. He apparently misunderstood parsimony to be an evolutionary process (which he rejected but believed that others embraced) rather than an epistemological rule, and argued that the minimum evolution method “has little to support it except Occam's razor, unless it can be shown to give much the same result as a proper estimation procedure based on a specific model of evolution …” (Edwards, 1966: 440). Of course, this is now known to be the case (Rindal and Brower, 2011, although those authors did not consider the legitimacy of parsimony to depend upon its producing results that are simulacra of likelihood trees). Occam's razor has been articulated as an epistemological decision-making tool for more than 800 years, and it is clearly fundamental to learning in many nonhuman organisms, too (Schuh and Brower, 2009). Therefore, parsimony cannot be contingent on a statistical approach invented in the 1920s for its “justification”. Setting these inconvenient facts aside, one might pose the question to Edwards, if the justification for parsimony is likelihood, what is the justification for likelihood? As far as I can tell from reading this book, his sole argument seems to be that Ronald Fisher invented likelihood, and Fisher was a very clever chap. Felsenstein (this volume) seems to believe that model-based phylogenetics “needs no justification.” That's probably good, because, as I have argued elsewhere (Brower, 2016), it does not seem to have any.