The horse as an evolutionary paradox


A visual summary of horse evolution published in 1921 (derived from an earlier diagram by W.D. Matthew). From An Introduction to the Study of Fossils.

When Charles Darwin published On the Origin of Species by Means of Natural Selection in 1859 he faced a substantial problem. His evolutionary mechanism predicted that the fossil record would contain finely-graded transitions revealing what G.G. Simpson would later term the "tempo and mode" of evolution, yet such transitional creatures proved elusive. This was a major problem for paleontologists who believed that studies of European geology represented an almost complete picture of Deep Time that had already been examined carefully. If the fossil evidence for Darwin's theory had not yet been found then there was little chance it would be in the future.

Darwin, of course, did not agree. Following the lead of his friend and mentor Charles Lyell, Darwin argued that the fossil record was not as complete or well-studied as was supposed. The discovery of the skeleton of the dinosaur-like bird Archaeopteryx in 1861 alone showed that the fossil record held many secrets left to be revealed. Other less spectacular "intermediate" fossils came to light or were reinterpreted in an evolutionary light, and in 1870 T.H. Huxley was able to propose at least three evolutionary transitions among vertebrates supported by fossil evidence.

Huxley knew well enough that two of the transitions, from terrestrial carnivores to whales and reptiles to birds, could not be represented in a real ancestor-descendant series. Instead Huxley said that the fossils in these transitions represented the form we might expect "straight line" ancestors and descendants to take. The small dinosaur Compsognathus, for instance, was not a bird ancestor but a dinosaur representative of the kind of animal from which birds evolved.

Horses were another matter. The work of paleontologists like Albert Gaudry and Vladimir Kowalevsky had illustrated the linear descent of modern horses (Equus) from multi-toed ancestors like Anchitherium and Hipparion. (Palaeotherium seemed like a good candidate for the evolutionary starting point for horses, but this was debatable.) This was a perfect graded evolutionary chain that appeared to confirm Europe as the birthplace of horses.

This hypothesis was quickly scuttled by the collection of numerous fossil horses from the American west, particularly by paleontologist O.C. Marsh. Horses had evolved in North America, not Europe, and the horses that formed the backbone of the lineages outlined by Gaudry, Kowalevsky, and Huxley represented "invasions" of types that had evolved in North America first. (See my review of The People's Peking Man for some brief thoughts on fossils and national pride.)

Despite the diversity of horse types found in North America, however, they were still organized in a progressive march from small, multi-toed browsers (Eohippus) to large, one-toed grazers (Equus). The branching nature of horse evolution was recognized in technical papers but most popular discussions of horse evolution, from textbooks to museum displays, drove home the progressive narrative. This series presented undeniable evidence for the evolution of an animal familiar to all, but by what mechanism did the modern horse evolve?

As I have mentioned previously many late 19th and early 20th century paleontologists were reluctant to accept natural selection as the primary mechanism of evolution. Competing mechanisms like neo-Lamarckism and orthogenesis were more popular. This was at least partially due to the sparse nature of the fossil record; there were gaps between groups that seemed to require some sort of jump or mechanism more able to produce rapid change than selection on variation. These competing mechanisms were also more easily squared with religious views, for evolution seemed less threatening if it had a direction (i.e. fore-ordained end points various lineages were striving towards).

The linear march of horse evolution was a perfect example of the belief in a directional kind of evolution. The descendants of little Eohippus were evolving towards the Equus type and those that had died out were simply unsuccessful experiments that, if they survived, would have likewise evolved into a large, high-toothed, one-toed horse. This type of evolution could be compared to the development of an individual over a lifetime; the changes in morphology that were to occur were pre-determined. (Orthogenesis was also applied to other groups like elephants and humans. Likewise, some scientists thought that species became extinct as a result of senescence, meaning that the lineage naturally died due to a loss of evolutionary vigor.)


A visual representation of an increase in body size and reduction in toes in horses. From The Next Generation.

The problem was that all the major changes in horses that paleontologists fixated upon (size, tooth height/type, and number of toes) could be understood as adaptations to changing environments. Either horse evolution was truly being driven by natural selection or directional forces were mimicking true adaptation. This paradox was pointed out by T.D.A. Cockerell in his 1920 textbook Zoology;

Naturalists ... have sometimes postulated what they called orthogenesis, the first part of the word meaning "straight" or "regular," as in orthodox. This implies that evolution follows a predetermined path, which was laid out for it in the beginning. Thus, the horse group was to increase in size, decrease the number of its toes, etc. It actually behaved as if following out a program planned in advance. The idea is not inherently absurd, since this is the course of individual development ; and it may well be imagined that there is something in the nature of a particular kind of protoplasm, that will lead it to vary in a certain direction. Indeed, we know that it does not vary in all directions ; thus we cannot get a genuinely blue rose. It is to be noted, however, that the evolution of the horse group is also strictly along the lines of adaptation. The climate became cooler and drier; the animal became an inhabitant of the plains. The solid hoof is adapted for running on hard ground, for receiving the impact of the heavier body; also for kicking the carnivorous enemies which had in the meanwhile evolved to prey upon the horse. The long-crowned, hard teeth are adapted for feeding on the vegetation to be found in open, dry places, and what might be regarded as a difficulty has been so completely overcome that the animal now- needs the type of food for which it is specially fitted. The whole history is one of adjustment to conditions, and the evolutionary process could not have taken place in the Eohippus environment, for the simple reason that the changes would all have been detrimental, leading eventually to extinction.

It is not explicit, but in this passage Cockerell proposes a compromise between orthogenesis and adpatation. An organism may contain variations to allow the next stage of development to take place but the conditions have to be right for those variation to be viable. If Eohippus had grazing dentition, then, according to Cockerell's model it would die. There was a barrier to further development that had to be removed. Thus we have an image of lineages striving "upwards" but perhaps blocked by environmental conditions. Only when the environment changes in the right way does the organism become "released" to evolve to the next stage.

Orthogenesis was ultimately crushed by the development of the modern evolutionary synthesis during the 1940's and 1950's. (G.G. Simpson was instrumental in extirpating it from paleontology.) The presence of a natural force driving organisms to particular ends was an inference that had no solid evidence to support it. Rather than being a true evolutionary mechanism it was more of a framework for understanding the pattern of evolution over millions of years. (For a recent summary of our present understanding of horse evolution see this article by Bruce MacFadden.)

Even today the evolution of the horse is often presented in a straight-line fashion with little attention to the previous diversity of the group. Why should this be so? There is a sort of "pull of the recent" in science where we wish to know the linear story of how we got to where we are now.* This is the desire that has constructed faulty phylogenies and promulgated "textbook cardboard" versions of the history of science. It seems that we can readily appreciate the diversity of groups that have left no living descendants, but we are more comfortable telling linear stories to explain the presence of living groups. It may indeed be possible to do this but it also requires some pruning and pushing extinct branches off to the sidelines. This is why we tip our hats to robust australopithecines in summaries of human evolution but rarely discuss them in detail.

*[I have borrowed this phrase from paleontology. Its regular usage refers to the fact that the fossil record is generally more complete the closer it is to the present.]

It is unfortunate, and frustrating, that we appear to be stuck with images of evolution that imply progress. We are still grappling with paradoxical images that are supposed to reveal adaptation but more strongly imply some kind of predestined direction. I was reminded of this last night when I picked up Stephen Jay Gould's 1989 book Wonderful Life. Gould opens with a discussion of images of progress in evolution and the strange relationship between scientists and illustration. One would think that scientists would think carefully about the images they use, particularly in paleontology where restorations and diagrams are so often used, but this is not necessarily so. Tradition seems to bind us to particular modes of illustration even after they have become outdated.

Even 20 years after Wonderful Life was published I still see images of evolution that infer progress (especially the erroneous trajectory from chimpanzee to Australopithecus afarensis to Homo sapiens). Are the traditional modes of illustration truly the best, or are we just unimaginative? I am leaning more towards the latter, but I suspect it will always be a problem. There seems to be a continual fascination with how the present world came to be as it is; species or entire lineages that have been lost along the way are thus seem of little relevance. This is absolutely untrue, but it is hard to fight the pull of the recent.


The discovery of horses like Anchitherium should have blown the idea of orthogenesis out of the water a long time ago. Granted, when it was discovered, it was the first three-toed horse known, but as time went on and we found that there were numerous species of horses living alongside each other, several of them not exactly direct ancestors of the modern horse (compare Anchitherium and Parahippus, the latter is thought to have given rise to the modern equids and the former is thought to have kicked the bucket).

Another flaw in the idea of orthogenesis comes from one of the most famous orthogenesists, Henry Osborn's, favorite taxonomic group, the brontotheres, and ironically among the closest relatives to the horse (for the most part among the perissodactyls, closer than rhinos and tapirs are, to say the least). While there was a trend to get bigger and bulkier during the Eocene for brontotheres, there are also little scurrying species like Nanotitanops, and hippo-like brontotheres running around. The problem is to keep public scientific tomes simple, much of the family tree has to be pruned or mentioned only in passing, giving the erroneous idea of orthogenesis to the public.

By Metalraptor (not verified) on 16 Mar 2009 #permalink

ironically among the closest relatives to the horse (for the most part among the perissodactyls, closer than rhinos and tapirs are, to say the least)

Last time I checked, the brontotheres (and the chalicotheres) were stem-perissodactyls. Have I missed something?

By David MarjanoviÄ (not verified) on 16 Mar 2009 #permalink

In the field of knowledge visualisation, it's known that one dimension is so much easier to depict than two or more that it is much preferred by almost all of us. We cheerfully collapse the multivariate positions of politics into "left" and "right", show two bar charts where one scatter chart would be richer and more informative, create lists like telephone directories when lists are the least insightful way to deal with the data (who cares that "Alabama" is alphabetically first among US states? that's no reason to show it first in the list)

Faced with a daunting collection of fossils, how tempting it must be to line them up left to right, and then study no further?

Derek; Good point. Such visualizations are convenient, and perhaps my objection to them is not so much that they are continually used but they seem to be the primary way evolution is visualized. If readers of a popular book, for instance, were introduced to the branching diversity of a group and then shown the tracing of a particular lineage, that would not necessarily be bad. It just seems to be that there is less effort put into visualizing the pattern of evolution than in presenting straight-line stories that imply progression towards a particular end.

Metal; The problem is that fossils do not speak for themselves. I can comfortably criticize orthogenesis now, but it was a compelling idea once upon a time. As I said I think it was more of a framework used to understanding convergence and evolutionary "progress" and that is why it hung on so long.

David; I don't think anything has changed in that regard; as far as I am aware they are stem perissodactyls.

If brontotheres are *stem* perrisodactyls, than they would not be closer to horses than to rhinos and tapirs, because rhinos and tapirs are crown perrisodactyls.

Although orthogenesis may not be accurate anymore, there are a few evolutionary trends that mimics its effects. There are some animals, like Thalassocnus, that go through a very direct, linear evolutionary path with very few (if any) sidebranches. And then there's Dollo's Law, which states that you can't re-evolve a feature you've previously lost, which is why some birds evolve "false teeth" instead of just re-activating the dormant genes responsible for tooth expression. Once you've started down a certain track, there are only so many options from there.

I foresee an evolutionary step coming in the evolution of evolutionary theory: psychoevolutionary illustration, the science of the psychology of evolution illustrations. But will it be progress, or merely an adaptation to the current needs of scientists?

Zach, that reminds me of Steven Jay Gould's famous quip in "Life's Little Joke", that those who would show evolutionary success as a march of triumph along a line are ironically forced to use as their examples the least successful of groups: those whose options are getting narrower and narrower.

"It is unfortunate, and frustrating, that we appear to be stuck with images of evolution that imply progress." I think this relates to the Victorian view of history as a march of progress. We're still Victorian in many cultural assumptions and I'd suggest that this is one of them.

In many analysises I have seen, brontotheres are put as a sister group to horses and sometimes paleotheres, and chalicotheres are put as a sister group to the clade of rhinos+tapirs.

But you are right, orthogenesis can indeed be the "fruit in the tree" so to speak, the biologist's temptation to try to use an easier and more appealing explanation that the actual methodology

By Metalraptor (not verified) on 19 Mar 2009 #permalink