Palaeos: Palaeos Evolutionary systematics
and systematics
Grade & Clade

Evolutionary systematics

Phylogeny through deep time

Grade and Clade

Horizontal and Vertical Classification

The following passage by the great paleontologist George Gaylord Simpson is worth quoting, as it elucidates very well the difficulties inherent in any system of phylogeny and systematics, and the different, indeed opposite, approaches of the Linnaean ("horizontal") and the Cladistic ("vertical") systems. This be it noted was written in 1945, many years before cladistics was founded. Here, "vertical" means phylogenetic - extending through time from ancestor to descendents. "Horizontal" means living at the same time. It is a peculiarity of geologists, and, following them, paleontologists, that time should be measured as an upward march from bottom to top. This is due obviously to the practical fact that the earliest rock strata were deposited first, and hence lie at the bottom of the sequence; the younger strata being deposited above them, leading to the traces of ancient time (as measured in successive rock deposits and the fossils they contain) as measured in terms of vertical succession. This mistaken spatial identification was only reinforced, I feel, by Haeckel and his successors constructing a "tree of life" in which the oldest and most primitive life forms are lower down, in the trunk and main branches, while the younger and later ones are higher up in the smaller branches and twigs (which is not in any way to reject the value of the tree diagram, only excessive literalism). I have however chosen to retain Simpson's jargon, if only because of the certain poetic ring, and assuming these terms are not taken too literally.

The existence of groups that are ancestral to two or more ultimately quite different phyla and the implication in classification that members of one group are more nearly related to each other than to members of other groups of the same rank give rise to the most difficult problems of classification of fossils. When the ancestral group is known, how is it to be classified? Can it be more nearly related to one than to the other of its descendent lines? In a sequence, is a group more nearly related to its ancestors, its descendants, or its contemporaries of like origin; in the human family analogy, is a man more nearly related to his father, son, or brother?

In the simplest case of an ancestral unit with two descendent lines, the usual solution in classification is either to extend the name and concept of one descendent group, the one morphologically more conservative if such a distinction is clear, to include the ancestry, or to give the ancestry a separate name and to consider it a group of the same rank as each descendant. Both systems are in common use, and it is impractical to try to use either exclusively and consistently. On lower levels of classification the second solution seems more often useful. For instance, the group immediately ancestral to the genera Pliohippus and Hipparion (among others) is also given generic rank and is called Merychippus.

Such simple cases are not particularly confusing, but analogous problems can be very intricate. It has, for instance, frequently happened that a group of, say, about subordinal scope has included numerous phyletic lines, all rather closely related but showing incipient divergence, and that one of these lines has outlived most or all of the others and has itself eventually split up into a group of about equal scope with the first. A concrete example is provided among Mammalia by the Carnivora. The early forms are quite varied, yet they are apparently derived from an immediately antecedent common ancestry. One of these early groups, the Miacoidea, survived and apparently gave rise to all the divergent phyla leading to our modern terrestrial carnivores. In such a case how is the principle of nearness of affinity to be applied?

The Miacoidea are certainly nearly related to the later carnivores, the Fissipeda, for they are ancestral to them. At the same time, they are certainly nearly allied to the more ancient and archaic Creodonta in the way often or customarily expressed by inclusion in one taxonomic unit, for they are derived from the same immediate ancestry. There are two sorts of affinity here, and following either one consistently throughout a classification is a practical impossibility. If the Miacoidea are placed in the Suborder Fissipeda, then consistency might seem to demand placing each of the other early, creodont subfamilies in a separate suborder, since their divergence from each other is equal to that of the Miacoidea, yet logical adherence to this method would make classification absurd by eventually separating virtually every ancient species as distinctive of a suborder or other large group. On the other hand if the Miacoidea are placed in the Creodonta, the implication would seem to be that they are more nearly related to, say, the hyaenodonts than to any fissipeds, which is not the case.

There are only two practical methods of dealing with such a situation, neither one of which is a clear expression of the evolutionary affinities involved. One is to place all the early forms including the Miacoidea in the Creodonta, separating the descendants of the Miacoidea as Fissipeda. The other is to place the Miacoidea in the Fissipeda and nevertheless to lump all other early lines in the Creodonta. These two solutions are equally valid, and both have been proposed and used in this particular example and also in dealing with numerous other analogous cases.

The choice here is between so-called horizontal and vertical classification. Horizontal classification separates ancestral from descendent groups and unites contemporaneous groups, or those in a similar stage of evolution, if they are derived from a common ancestry. Vertical classification unites ancestral and descendent groups and separates contemporaneous groups that are diverging from a common ancestry. It is sometimes stated that these are mutually exclusive principles and that classification must be based on one or the other, but in fact neither can be followed consistently and any classification necessarily combines both methods. The most ardent exponent of vertical classification finds it necessary to separate ancestral and descendent genera, for instance.

Each point of view has its advantages and its dangers. Horizontal classification is usually easier and more objective. The earlier horses and tapirs, for instance, have more in common than have the earliest and latest horses, and it would be much easier to define and to recognize a horizontal group containing both primitive horses and tapirs than a vertical group containing all the horses and excluding all the tapirs. The horizontal arrangement may also be more stable, because the relative ease of recognition makes it less likely to be disturbed by later discovery, whereas most evolutionary phyla are difficult to distinguish and require considerable material for their proper recognition. Vertical groupings often require extensive rearrangement following new discoveries or new studies that modify ideas of the details of phyletic descent.

The principal advantage of vertical, and disadvantage of horizontal, classification is that the former is more in accord with the whole conception of evolutionary descent. Its principal proponents, such as the late H. F. Osborn, therefore, sometimes speak of it as evolutionary classification, as opposed to the more static and historically pre-evolutionary, more strictly Linnaean, horizontal system. Yet the horizontal affinities are just as real and are just as evolutionary as the vertical. Horizontal classification can, therefore, be as consistent with phylogeny as vertical classification, and since neither can really express phylogeny, there is really no a priori, theoretical reason for preferring one to the other even if one could be employed to the exclusion of the other, which is not possible.

-- G.G. Simpson, "The Principles of Classification and a Classification of Mammals", Bulletin of the American Museum of Natural History, vol.85, (New York, 1945) pp.17-19

Reading this passage, I have the impression that biological systematics is not unlike the wave-particle paradox of quantum physics. Is a photon or an electron a wave or a particle, energy or matter? Is a biological taxon - if indeed such an abstract entity could be considered a "real" (in the sense of objective) thing like a photon at all! (ignoring here the more extreme metaphysical speculation of any Platonizing biology) - best understood "horizontal" or more properly isochronous or evolutionary grade manner - as the Linnaean system does, or in a "vertical" or more correctly phyletic manner, as Cladistics argues it should. If we look at an electron or a photon as a wave of probability or energy we forget it is also (or rather can also appear to our instruments) as a particle of matter - but the electron or photon in itself is beyond wave and particle, energy and matter. Those are only labels we use to make sense of it with our three-dimensional minds.

The Evolutionary Systematics of Mayr, Simpson and others emphasize the horizontal grade, although the vertical dimension of monophyletic clade is accepted as well (but there is no real methodology) Cladists on the other hand have a rigorous methodology that means the rejection of the "horizontal" grade perspective. Terms like say Miacoidea, Thecodontia, Condylartha, Reptilia, or Pteridosperma, because all these groups are paraphyletic; that is they gave rise to descendents that are not included in those groups. But were we living in the Eocene epoch, the Miacoidea would then be monophyletic (and cladistically authentic); were we in the middle Triassic period Thecodontia, Reptilia and Pteridosperma would be too! Ironically the methodology that affirms phylogeny at the same time ignores the passage of time in which that very phylogeny unfolds!

images not loading? | error messages? | broken links? | suggestions? | criticism?

contact us

unless otherwise attributed, original content by M. Alan Kazlev (Creative Commons Attribution 3.0 Unported License) (Creative Commons Attribution 3.0 Unported License) . This license does not include quoted text.
page first uploaded on Kheper site 28 July 2000, on Palaeos site 26 May 2002
checked ATW021107, last modified MAK050104, 111018, edited RFVS111202
Unless otherwise noted,
the material on this page may be used under the terms of a
Creative Commons License.