52 • May 2003: 295–298 Nelson & al.

• Brummitt on paraphyly: a response

P O I N TS O F V I E W

Brummitt on paraphyly: a response

Gareth Nelson, Daniel J. Murphy & Pauline Y. Ladiges

School of Botany, The University of Melbourne, Victoria 3010, Australia. g.nelson@botany.unimelb.edu.au

(author for correspondence); d.murphy@unimelb.edu.au; p.ladiges@unimelb.edu.au

In a dozen oral presentations and five publications species of a group, such as they would be if the species
over the past eight years, Richard Brummitt (1996, were the result of dichotomous speciation…. Having
1997a, b, 2002; Brummitt & Sosef, 1968) has doggedly before us 32 terminal species, represented by the black
defended “paraphyletic” taxa, claiming that they are log- dots above, we would be able to make four groups (such
ically required by traditional Linnaean classification. He as genera): A, B, C, D.
claims also that “classification into Linnaean taxa with- “It is clear that, even without paleontological knowl-
out any being paraphyletic is a logical absurdity [impos- edge of the connections, an adequate knowledge of the
sibility]” (2002: 33, 40): “If we are classifying all the morphology of these species would suffice to indicate
products of evolution, i.e., the whole evolutionary tree of that genus B is more closely related to genus A than to
life, every taxon we recognise must make another taxon genus C; and that, before grouping the 32 species into
paraphyletic. That is a simple logical fact”. four genera, it would be necessary to group them into
To illustrate this fact, Brummitt offered a phyloge- two ‘supergenera’ or subfamilies: AB and CD. And with-
netic tree (2002; Fig. 1, top). He did not explain the cir- in each genus it would be possible also to recognise sub-
cles and the lines that interconnect them, but each circle genera and even smaller groups of more closely related
may be seen as a taxon, and each line as an ancestor- species”.
descendant relationship between two taxa: “The whole “If this scheme corresponds to reality, one may con-
diagram is one clade, and the dark circles together form clude that the distinction between groups of equal taxo-
a…lesser clade within the larger clade. It doesn’t matter nomic rank cannot be arbitrary; and also that the distinc-
what rank we are talking about, so, just for the sake of tions are not caused by gaps in the system, gaps produced
convenience, let us think in terms of genera. If I call all by extinction. Even in the absence of extinction, the dis-
the dark circles one genus defined by the characters at 1, tinctions would be quite clear. And while it might be
then the open circles must be a different genus…. But arbitrary to consider group A a genus, no good systema-
then the genus of open circles is paraphyletic...” tist would ever combine some species of group C with
(Brummitt, 2002: 33). He refers to some circles as ances- AB, and the other species of group C with D”.
tors and others as descendants, and the whole as “a dia- Rosa’s scheme may be used to illustrate the classifi-
gram of what actually happened in evolution” (p. 37). He cation of the eucalypt group, which includes the genus
does not consider the possibility that the open circles Angophora Cavanilles, 1797 (13 species)—correspon-
might comprise more than one genus, but he asserts that ding to Rosa’s group B. Most remaining eucalypts have
the circles “have remained comparatively little changed traditionally been placed in the genus Eucalyptus
or unchanged—a situation that has arisen millions of L’Héritier, 1788 (700+ species)—Rosa’s groups A+C+D.
times in evolution” (p. 33). Eucalyptus was eventually found paraphyletic, with
Part of Brummitt’s repeated message (Brummitt, bloodwoods and ghost gums (100+ species), subgenera
2002) is a challenge: “Until somebody can draw for us a Corymbia and Blakella (Pryor & Johnson, 1971), most
phylogenetic tree…divided fully into Linnaean taxa closely related to Angophora (Ladiges & al., 1995;
without any being paraphyletic, we will continue to Udovicic & al., 1995). The paraphyly was rectified (Hill
believe that our arguments on the inevitability of para- & Johnson, 1995) simply by recognising bloodwoods
phyletic taxa are correct (Brummitt & Sosef, 1998). That and ghost gums as one genus, Corymbia—Rosa’s group
was three years ago. We have had no takers. The offer is A—leaving Eucalyptus with fewer species (600+)—
still open”. Rosa’s group C+D. Recognising Corymbia as a genus
In 1918 such a phylogenetic tree of this type was does not create paraphyly but rather eliminates it by
published by Daniele Rosa (pp. 137–138) with the fol- making Eucalyptus monophyletic. For Brummitt , how-
lowing remarks: “The following scheme [Fig. 1, bottom] ever (Brummitt, 2002: 33), “As soon as we assign a rank
…represents the connections of affinity between the to a group, we create paraphyly”.

295
Nelson & al. • Brummitt on paraphyly: a response 52 • May 2003: 295–298

A B C D

Fig. 1. Phylogenetic trees showing patterns of dichotomous speciation. Above, after Brummitt (2002: fig. 1; “Stylised
diagram of a phylogenetic tree in which a major character change has occured at point 1, giving rise to descendants
indicated by darkened circles”). Below, after Rosa [1918: 137–138, “connections of affinity between the species of a
group, such as they would be if the species were the result of dichotomous speciation.... Having before us 32 terminal
species, represented by the black dots above, we woud be able to make four groups (such as genera): A, B, C, D”].

By itself with eucalypts aside, Rosa’s scheme satis- responding to the two subapical open circles of
fies, at least in part, Brummitt’s criterion of a Brummitt’s tree. Similarly for the entire family imagine
“tree…divided fully into Linnaean taxa”, but it lacks a a taxon, (ABCD)e, corresponding to the open circle at
taxon for each of its 31 nodes of common ancestry. In a the apex of his tree.
case such as this Brummitt suggests that “paraphyly” is In Rosa’s scheme, (AB)e would be a third genus in
most relevant when extinct taxa come into play (p. 38): subfamily AB; (CD)e, a third genus in subfamily CD;
“the situation in my Fig. 1 could easily appear…as two and (ABCD)e, a genus in a third, monotypic, subfamily
sister taxa [Rosa’s two subfamilies?], with neither para- in family ABCD. With addition of these three hypotheti-
phyletic, if only extant taxa are analysed”. cal taxa, Rosa’s scheme would satisfy Brummitt’s crite-
Consider, then, Rosa’s two subfamilies of extant rion of a fully divided tree, complete with imaginary
taxa, AB and CD, and the hypothetical ancestry implied “fossil ancestors”, as far as the basic divisions are con-
by the first few dichotomies—the basal nodes of his dia- cerned:
gram. For each subfamily, imagine an extinct taxon, family: ABCD
(AB)e and (CD)e, potentially knowable from fossils, cor- subfamily: (ABCD)e

296
 52 • May 2003: 295–298 Nelson & al. • Brummitt on paraphyly: a response

subfamily: AB in some cases difficult to sort out. True, but this is to be

genus: (AB)e expected—and even hoped for if the order (Primates) is
genus: A a natural one; early members of two such groups should
genus: B be convergent, no matter how far later members of the
subfamily: CD two have diverged”.
genus: (CD)e Brummitt’s “paraphyly” is perhaps better, even best,
genus: C exemplified by evidence-free assertion such as that of
genus: D Lewontin (1981): “Birds arose from nonbirds and
Additional taxa, subgenera and sections, could be simi- humans from nonhumans. No person who pretends to
larly imagined for the remaining 28 nodes of common any understanding of the natural world can deny these
ancestry. facts”. Indeed (Pinto-Correia, 1997: 36, from Chandogya
In practice, extinct ancestral taxa are seldom of con- Upanishad 3.19.1): “At the beginning there was nothing
cern, because organisms credibly representing them are but a non-Being. It became the Being”. For Colin
seldom if ever in hand. For this reason Brummitt’s “fully Patterson (2002: 19), such nonbird, nonhuman, or non-
divided tree,” if it presupposes a complete fossil record being of paraphyletic groups in general is: “...an abstrac-
of “what actually happened in evolution,” is a theoretical tion that is beyond criticism…a statement that has the
construction, rather than a practical one constrained by appearance of knowledge but in fact contains none—a
empirical data. Even so, as a representation of history’s piece of antiknowledge, derived from evolutionary theo-
continuum, what credibility is there in a few circles and ry”. Nonbirds, nonhumans, and other such “paraphylet-
interconnecting lines? ic” notions do not integrate in classification. They do not
In a world where all things are possible, one may relate to taxa because they are not mutually exclusive,
consider any taxon, extinct or extant, as the ancestor of nor is one wholly inclusive of another. Most nonbirds are
another. In Rosa’s scheme, consider genus A the ancestor also nonhumans and vice versa.
of genus B (or B of A), and there is no need of the hypo- Ancestral taxa may be argued to imply paraphyly
thetical genus (AB)e. Similarly, consider C the ancestor (Nelson & Platnick, 1984), even by logical necessity if a
of D (or D of C), and there is no need of (CD)e. Consider person be so disposed. In practice, however, organisms
subfamily AB the ancestor of subfamily CD (or CD of attributed to an ancestral taxon need not represent a para-
AB), and there is no need of (ABCD)e. With these con- phyletic assemblage of two or more taxa, as shown by
siderations extended to subgenera and sections, Rosa’s the available evidence of character data. With further
original scheme would completely satisfy Brummitt’s study, organisms attributed to any taxon might prove to
criterion. represent two or more taxa that together form a mono-
None of this is novel, being exactly described by phyletic, paraphyletic, or even polyphyletic assemblage.
George Simpson (1945: 17–18): “In the simplest case of Any such discovery arises only through evidence found
an ancestral unit with two descendant lines, the usual relevant to a particular case.
solution in classification is either to extend the name and Alleged ancestral taxa in the history of paleontology,
concept of one descendant group, the one morphologi- not mentioned by Brummitt, often disappear once their
cally more conservative if such a distinction is clear, to internal relationships become evident—as shown by
include the ancestry, or to give the ancestry a separate actual evidence in specific cases. Everett Olson (1971:
name and to consider it a group of the same rank as each 347) commented on the fate of one such group,
descendant”. Thecodontia, “an array of primitive archosaurian rep-
Simpson’s views are similar to those of Brummitt, tiles” that: “...fits well into the ‘horizontal’ pattern of
particularly as they concern horizontal classification and classification that is often imposed on a ‘basal complex’
its necessity—the deliberate creation of paraphyletic taxa in which lines are difficult to sort out. With increased
based on evidence of relationship of their subtaxa— knowledge, this sorting may take place, and if fully suc-
which for Simpson is “usually easier and more objec- cessful, the primitive complex may disappear from clas-
tive” than vertical classification (1945: 18). sification”. Brummitt opposes disappearance of a basal
A history of controversy exists over this matter, e.g., complex, as if evidence of paraphyly should really be
over Simpson’s Prosimii, non-anthropoid primates seen and valued as evidence of ancestry, which it is not.
(1945: 183). Vertebrate paleontologist Alfred Romer For Brummitt, “Evolution is paraphyly all the way”:
stated (1968: 183): “I do not believe this procedure is the birds from nonbirds, being from nothingness, ex nihilo
proper one; to lump all forms below the monkey-ape omnia. Persons not of his persuasion in this respect he
level as Prosimii is comparable to dividing all animals terms “special creationists”. While distancing himself
into Vertebrata and Invertebrata. Simpson justifies it pri- from creationism, he nevertheless adopts its philosophy
marily on the fact that early lemurs and the tarsioids are and its antiknowledge. His stance is at odds with tradi-

297
Nelson & al. • Brummitt on paraphyly: a response 52 • May 2003: 295–298

i–xvi, 1–350.
tional science and its fundamental maxim, ex nihilo Udovicic, F., McFadden, G. I. & Ladiges, P. Y. 1995.
nihil—from nothing, nothing comes. By our reading he Phylogeny of Eucalyptus and Angophora based on 5S
is at odds also with, and isolated from, the history of sys- rDNA spacer sequence data. Molec. Phyl. Evol. 4:
tematics, from Aristotle to the present, and its common 247–256.
thread of, or preoccupation with, natural classification
and the nature of the evidence for it.

ACKNOWLEDGEMENTS
For helpful comment and literature, we are grateful to Pietro
Passerin d’Entrèves and David Williams.

LITERATURE CITED
Brummitt, R. K. 1996. In defence of paraphyletic taxa. Pp.
371–384 in: van der Maesen, L. J. G., van der Burgt, X. M.
& van Medenbach de Rooy, J. M. (eds.), The Biodiversity
of African Plants, Proceedings XIVth AETFAT Congress,
22–27 August 1994, Wageningen, The Netherlands.
Kluwer, Dordrecht.
Brummitt, R. K. 1997a. Taxonomy versus cladonomy, a fun-
damental controversy in biological systematics. Taxon 46:
723–734.
Brummitt, R. K. 1997b. Proposing the motion “that this
house believes that Linnaean classification without para-
phyletic taxa is nonsensical”: the theoretical case.
Published by the author, Kew.
Brummitt, R. K. 2002. How to chop up a tree. Taxon 51: 1–41.
Brummitt, R. K. & Sosef, M. S. M. 1998. Paraphyletic taxa
are inherent in Linnaean classification—a reply to
Freudenstein. Taxon 47: 411–412.
Hill, K. D. & Johnson, L. A. S. 1995. Systematic studies in the
eucalypts. 7. A revision of the bloodwoods, genus
Corymbia (Myrtaceae). Telopea 6: 185–504.
Ladiges, P. Y., Udovicic, F. & Drinnan, A. N. 1995. Eucalypt
phylogeny—molecules and morphology. Aust. Syst. Bot.
8: 483–497.
Lewontin, R. 1981. Evolution/creation debate: a time for truth.
BioScience 31(8): 559.
Nelson, G. & Platnick, N. 1984. Systematics and evolution.
Pp. 143–158 in: Ho, M.-W. & Saunders, P. T. (eds.),
Beyond Neodarwinism. Academic Press, London.
Olson, E. C. 1971. Vertebrate Paleozoology. Wiley Inter-sci-
ence, New York.
Patterson, C. 2002. Evolutionism and creationism. The
Linnean 18(2): 15–33.
Pinto-Correia, C. 1997. The Ovary of Eve: Egg and Sperm
and Preformation. Univ. Chicago Press, Chicago.
Pryor, L. D. & Johnson, L. A. S. 1971. A Classification of the
Eucalypts. Australian National Univ. Press, Canberra.
Romer, A. S. 1968. Notes and Comments on Vertebrate
Paleontology. Univ. Chicago Press, Chicago.
Rosa, D. 1918. Ologenesi. R. Bemporad, Firenze. [Reprint:
Vergata, A. La (ed.). 2001. Daniele Rosa, Ologenesi.
Biblioteca della Scienze Italiana, no. 32. Giunti, Firenze.]
Simpson, G. G. 1945. The principles of classification and a
classification of mammals. Bull. Amer. Mus. Nat. Hist. 85:

298