Review of Michael Denton's
Evolution: A Theory in Crisis
Copyright © 1996-1997 by Mark I. Vuletic
Note: This document is undergoing modification. The author intends to produce a revised and expanded version of this review.
In his 1985 book Evolution: A Theory in Crisis, Michael Denton argues for a typological model of nature - a model in which "...all the variation exhibited by the individual members of a particular class [is] merely a variation on an underlying theme or design which [is] fundamentally invariant or immutable" (Denton, 1985, p. 94). This model is in direct contradiction with the evolutionary account of the history of life, in which all organisms are linked by common descent. Denton claims that while microevolution and speciation are proven phenomena, the common evolutionary descent of all organisms is "a highly speculative hypothesis entirely without direct factual support and very far from the self-evident axiom some of its more aggressive advocateswould have us believe [it is]" (Denton, 1985, p. 77); that is to say, Denton argues that there is no evidence for macroevolution.
Evolutionists - even those who agree with Theodosius Dobzhansky that "nothing in biology makes sense except in the light of evolution"(Dobzhansky, 1973) - would be surprised by Denton's suggestion that they hold macroevolution to be a "self-evident axiom." Evolutionists do not believe that macroevolution is an a priori truth; rather, they believe that the force of numerous lines of evidence must compel any good scientist to accept it a posteriori. Some of the typical lines of evidence offered in support of the common evolutionary descent of all lifeforms are (1) the gradation of organisms in systematics, (2) the biogeographical distribution of species, (3) the existence of homologous and vestigial structures as demonstrated in comparative anatomy, embryology, and molecular biology, and (4) the presence of transitional forms and gradual sequences in the fossil record. To the evolutionist's surprise, however, Denton seeks to show that these lines of evidence are either nonexistent or support typology more than they do common descent. However, Denton does not stop even there - in addition to attacking the evidence for evolution, he argues that there is no mechanism that could, even in principle, cause macroevolution.
I will argue in this paper that both of Denton's attempts to make an adequate challenge to evolutionary biology fail - neither does Denton manage to undermine the evidence for evolution, nor does he succeed in demonstrating that macroevolutionary mechanisms are inherently implausible.
Denton's attacks against evolutionary process constitute some of the worst arguments in his book. One line of argument draws an analogy between mutation and word-transformation, showing that single letter substitutions cannot change one large meaningful word or sentence to even a closely related one without passing through a whole series of meaningless intermediates. For Denton, evolution necessitates a process like:
He sat on the mat.
He stt on the mat.
He sto on the mat.
He stoo on the mat.
He stood on the mat.(Denton, 1985, p. 89)
In this example, the three intermediates are obviously nonsense, though the meaningful sentences on either end are closely related. Taking this word-transformation analogy to an extreme, Denton argues that in the natural world, two higher taxa can be connected only by a sequence that contains non-functional intermediates, and that they can thus not be connected by evolutionary means at all. Denton is attempting to use Cuvier's old argument about organisms being functional wholes, and having to change saltationally or not at all:
Every organized being forms a whole, a unique, and perfect system, the parts of which mutually correspond and concur in the same definitive action by a reciprical reaction. None of these parts can change without the whole changing. (Cuvier, 1829, p. 60)
But this is taking the analogy way to far. Language, as strange as it may seem at first, is much more restricted with regards to function than biological organisms or even mechanical entities. A small change in a word, such as adding an "x" to the end of it, will almost without fail destroy the whole meaning of the word; but a small change in a cow or a watch, such as decreasing slightly the length of a cow's hind legs or diminishing slightly the size of one cog in the watch, will not spell catastrophe. In such a way, all of the cogs in a watch could change slightly, one at a time, eventually changing the entire configuration and sizes of the cogs, without destroying its function (presuming watches can reproduce!). One can even imagine nonfunctional digital components being added until they finally begin to work, at which point the gears, springs, and cogs will fade into nothingness.
Denton, I suspect, would reply to this that yes, the word-transformation analogy is awkward, but that it applies at some level of biological change. It would not prevent small scale anatomical changes, but it certainly would prevent large ones, even through small steps. But since we have shown that small steps are possible, what is to ever stop these small steps from continuing until they have overhauled the organism entirely, as I noted in the windup-to-digital watch transition? To demonstrate that the "limits of biological change" (as some creationists refer to it) are nonexistent, let me list a few examples of observed major structural change from the biological literature.
Consider, first, the wildly varying plants of the genus Brassica (broccoli, cauliflower, cabbages, kale, and brussel sprouts), which have been derived through artificial selection from a single species of wild mustard (Campbell, 1990, p.432). Anyone who has compared kale with brussel sprouts can see that their structures are extremely different.1 Another example of major structural change is the set of widely variant dog varieties that has resulted from selective breeding.2 Just look at the different varieties of dogs that exist today, and ask whether a Chihuahua and St. Bernard can be connected only by saltations? If such profound change can occur and speciation does, as Denton conceded, occur, then what is to theoretically stop a remote ancestor from evolving into all of the primates? What is theoretically to stop an ancient ungulate from evolving into a whale, or an ancient fish into an amphibian? The fossil record, as we will see later on, provides even more evidence for major structural change, both in skeletal and soft organ characteristics. Yet Denton persists:
While sentences, machines, and other sorts of complex systems can undergo a certain degree of gradual functional change, there is invariably a limit beyond which the system cannot undergo further gradual change. To cross as it were from one "type" to another necessitates a relatively massive reorganization involving the redesign or respecification of all or most of the interacting component subsystems. Systems can undergo gradual microevolution through a succession of minor changes in their component structures but macro-evolution invariably involves a sudden "saltational" change. (Denton, 1985, p. 91)
In no place does he offer support for this characterization of macroevolution, and for his claim concerning the limits of gradual functional change. Rather, he seems to be taking as self-evident the assertions of Cuvier. Denton does discuss at great length in Chapters 9 and 14 some of the more complex organs in various vertebrates, such as the avian pulmonary system and its lack of a remote counterpart among the reptiles, but these long discussions seem to amount to little more than saying: "We haven't told an evolutionary story for it yet, and it seems difficult to me to believe that such a story could exist, therefore there is not one." This is the type of reasoning that Richard Dawkins has dubbed the "Argument from Personal Incredulity"(Dawkins, 1987, p. 38). Dawkins notes as well as any evolutionist that "anti- evolution propaganda is full of alleged examples of complex systems that 'could not possibly' have passed through a gradual series of intermediates"(Dawkins, 1987, p. 86), and he goes to great lengths to show that accounts can definitely be formulated that explain the development of structures and processes as sophisticated as the human lung and bat echolocation. Perhaps if more evolutionary theorists were around, we would have accounts about every single structure Denton could incredulously point to. But Denton really should be able to formulate such stories himself. The point is, pointing out how impossible it seems, at first glance, for a structure to have evolved gradually, does not constitute evidence that gradual macroevolution is impossible or improbable - it says something rather about one's failure to give hard thought to the possible means whereby complex structures could be generated.
At the molecular level, Denton discredits himself by quoting Emile Zuckerkandl to show that "it is now generally conceded by protein chemists that most functional proteins would be difficult to reach or interconvert through a series of successive individual amino acid mutations"(Denton, 1985, p. 320). Zuckerkandl's quote (Zuckerkandl, 1975, p. 21) seems quite damning to the casual reader, but when one reads the entire article, one finds out that Zuckerkandl largely contradicts Denton. By Zuckerkandl's analysis, most advanced functional proteins cannot interconvert directly, and cannot be reached by some saltational mechanisms, but that they certainly can each be reached through gradual evolution from a common ancestor.
Denton seems to harbor much personal confusion about what non-saltational evolution actually is, and it is from this confusion that another one of his erroneous claims against evolution derives. He believes evolution is a random search process, that somehow mutation plus natural selection yields results about as random as macromutation:
Ultimately, Darwin's theory implied that all evolution had come about by the interactions of two basic processes, random mutation and natural selection, and it meant that the ends arrived at were entirely the result of a succession of chance events. Evolution by natural selection is therefore, in essence, strictly analogous to problem solving by trial and error, and it leads to the immense claim that all the design in the biosphere is ultimately the fortuitous outcome of an entirely blind random process - a giant lottery. (Denton, 1987, p. 43)
This is clearly an incorrect way of looking at things. To illustrate, we will examine a modern variant of Cuvier's "animal space", a multi-dimensional space in which all possible phenotypes exist, and are arranged next to one another according to the amount of difference between their genotypes (this is a very high-dimensional space, as anyone familiar with the number of genes and number of alleles per gene in a typical organism well knows). Absolutely nonfunctional phenotypes - ones that will not survive in any environment - are represented as blank spaces. Now take a functional point A and a functional point B that are separated by a few hundred thousand points in this space. The probability of a macromutation, which is somewhat like making a lottery drawing (although not exactly), changing A to B is astronomically small. Likewise the probability of A changing stepwise to B through a series of random events, is astronomically small. Denton seems to think that stepwise evolution is like the latter. But this is not so. The rate of mutation in organisms, "one mutation per locus per 10^5 to 10^6 gametes"(Campbell, 1990, p. 445) is sufficient to give natural selection an immense amount of variation to work with at any functional starting point. It is probable that one mutant of A will be a functional point closer to B. If this descendant is selected for, it will have a large number of descendants, whose gametes will undergo the same rate of mutation, making it extremely likely that another functional point closer to B will be produced. Without natural selection, the process would be random, and a connection from A to B would require an extremely improbable sequence of events to occur. But if thousands, hundreds of thousands, or millions of mutants are created at each locus, and selective pressures choose those forms that are closer to B, then it is virtually inevitable that B will arise.
This scheme is a bit simplistic, because it assumes a constant environment and constant selective forces, making the result almost teleological in character, which is not an accurate account of evolutionary processes. In actuality, the environment fluctuates, so selective pressures vary constantly, and lifeforms must constantly change to keep up. In a fluctuating environment, it is less probable that a path will be traced to B. But this is hardly a problem, because evolution is not teleological. Nothing in evolution says that Homo sapiens had to arise from the long march from bacteria, or that it was even probable that it would arise. But even in a fluctuating environment, it is clear that we could expect something very structurally different and reproductively isolated from A to eventually arise. As Carl Sagan likes to point out, without whatever fortuitous cataclysm caused the downfall of the dinosaurs, there might be intelligent lizards - instead of us - around now pondering the stars and wondering where they came from. To repeat, nothing in evolution claims that the history of life had to unravel exactly the way it did, but evolution is by the far the best, and certainly a plausible, explanation of why life did in fact develop as it did. The predictive power of evolution at the macro-level is low, but the retrodictive, explanatory power of evolution is immense. And to summarize then, it is clear that Denton completely mischaracterizes evolutionary processes when he refers to them as matters of blind chance.
Denton has failed in his direct attack on evolutionary process. Thus, not having a better theory, he must now show that there is an extraordinary degree of typological character in the data - a degree that evolution could not possibly have produced - if he is to disprove evolution.
Way back at the beginning of this paper, I mentioned the various lines of evidence that evolutionists list in support of evolution. Denton seeks to refute all of them except for biogeography, which is very unfortunate, since biogeography is a very powerful tool for establishing the plausibility of transitional forms and distinguishing homology from analogy. Let us examine Denton's arguments each in turn, and see how they fare.
First of all, in a response to systematics, Denton cites the increasing popularity of cladism and lists quotes by those who use it to the effect that we can only determine sister relationships between organisms, not ancestral relationships. The problem with this is that cladism obviously does not contradict evolution, and in fact the data underlying the arrangement of organisms on the cladogram supports evolution more than it does an underlying typological process. It interprets classified species as being descendants of unspecifiable common ancestors. The reason for this, however, is an epistemic one and not an ontological one. Cladists are suggesting that we cannot know what form the common ancestor between two other species took, not that there are no common ancestors. Systematics, being a way of arranging data, fails to be extraordinary evidence in favor of typology. In my opinion, an appeal to taxonomy is not a very good way to support evolution, either. Taxonomy has helped us to systematically (no pun intended) summarize the data that reveals an evolutionary pattern. As such, it should be appeals to the underlying data, not to the classification system, that will count as evidence. This takes us right into the next aspects that Denton seeks to criticize.
Homologous structures are frequently cited as evidence for evolution. If organisms do not share common ancestry, why do they share so many anatomical, embryological, and molecular similarities? Denton denies that they do. Instead, he seems to propose that many of the similarities are the result of convergent evolution, and that the rest are insufficient to prove evolution. For instance, he cites embryological dissimilarities, particularly between the mammalian and reptilian blastula, ignoring all of the similarities that do exist and are displayed in just about every biology textbook (the famous vonBaer chart). This is problematic for Denton, because while evolution obviously can account for and even predicts differences between organisms at various stages in their development, typology does not allow for more than superficial similarities.
Denton is more resourceful in his attack on anatomical similarities. He states:
...the validity of the evolutionary interpretation of homology would have been greatly strengthened if embryological and genetic research could have shown that homologous structures were specified by homologous genes and followed homologous patterns of embryological development...but it has become clear that the principle cannot be extended in such a way. (Denton, 1985, p. 145)
His reason for that last conclusion is that
...the way the gastrula is formed and particularly the positions in the blastula of the cells which give rise to the germ layers and theirs migration patterns during gastrulation differ markedly in the different vertebrate classes... homologous structures are arrived at by different routes. (Denton, 1985, p. 145-146)
Then follows mention of species-specific pleiotropic genes which influence the development of homologous structures. How could homology point to common ancestry if the homologous structures are constructed in part due to the instructions of genes not found in other species?
The answer to Denton's assertions about embryological development is that he simply is unaware of a set of facts that reveal the differences in migration patterns and positions to be superficial: the differing amounts of yolk in the eggs of varying classes of vertebrates imposes constraints upon embryological development that make the process seem quite different from class to class. Scott Gilbert's discussion of avian gastrulation gives the general idea:
Cleavage in avian embryos creates a blastodisc above an enormous volume of yolk. This inert, underlying yolk mass imposes severe constraints on cell movements, and avian gastrulation appears at first glance to be very different from that of a sea urchin or frog. We shall soon see, though, that there are numerous similarities between avian gastrulation and those gastrulations we have already studied. Moreover, we shall see that mammalian embryos - which do not have yolk - retain gastrulation movements very similar to those of bird and reptile embryos. (Gilbert, 1991, p. 138)3
As for pleiotropy, it demonstrates yet another case of the argument from personal incredulity. There is no reason at all to presume that the set of genes controlling the development of a structure could not change in number over time from species to species.
Is it plausible to believe that most homologous structures developed through convergent evolution? Only if the organisms wielding the structures either (a) lived in very similar environments having very similar selective pressures, or (b) the structure would confer a definite advantage in almost any environment (structures such as eyes and wings). But regarding both of these, it is necessary that the two classes of organisms that supposedly acquired their "homologous structures" through convergence not be linked to ancestors which share the same structure. It is a stretch to believe that the pentadactyl form arose separately and out of necessity in fish, amphibians, reptiles, birds, and mammals, especially since the classes share ancestral links to one another as demonstrated by fossil evidence and embryological and molecular correlations. Of course, Denton does not believe the fossil record is adequate to make evolutionary inferences from, so I must deal with that next, but first a few notes on the embryological and molecular correlations I mentioned.
Embryos do not undergo convergent evolution. There is no reason for them to do so. In fact, the embryo is a repository for useless vestigial structures like gill arches, and teeth that are reabsorbed by toothless creatures such as whales before birth. Plenty of junk from our evolutionary histories accumulate in embryonic stages, and it requires many ad hoc assumptions to explain them within the framework of typology. As for molecular data, Denton makes a show of cytochrome-c amino acid substitution numbers appearing to divide all life into typologically distinct classes. But this kind of division is to be expected, considering that we derived the numbers from contemporary organisms, and not from fossilized organisms. If we were able to extract DNA from enough fossils, the typological boundaries would break down. In no way is the molecular data incompatible with evolution. Denton charges that neutral amino acid replacement would not give such perfectly typological results because different organisms have different rates of mutation per generation. In response:
There are many factors affecting mutation rates. They are not simply dependent on "number of generations". You see, it a species reproduces every twenty five years, as opposed to every two hours, you could suggest that the two hour species evolves "faster" or that it has a higher mutation rate. But that is patently a misconception. During the 25 years that the other species is not reproducing, its gametes are nonetheless (well its germ cells) subject to mutations; indeed a lot of mutations could accumulate in its 25 yrs generation time. So most emphatically, no: mutation rates are not understood to be dependent on the number of generations. (Ladomery, 1995)
Given this, evolutionary processes would be expected to yield the molecular data shown. Moreover, there are sequences of DNA (called "introns") that are excised from RNA copies before translation, and these introns, despite never being subject to selective effects, give the same general pattern as cytochrome-c amino acid sequences - namely the degree of similarity that evolutionists have come to expect. The fact that this pattern is maintained for a sequence of nucleotides that can mutate freely without consequence indicates that the organisms containing them have shared ancestry (because before they split off, their intron nucleotide sequences could not possibly diverge!). The fact that introns reveal the same "distance" between classes as functional sequences rather than giving wildly different distances may well be damaging to typology, unless we presume the archetype contains archetypal non-functional sequences. In any case, without a typological process proposed to support the typological interpretation of the molecular evidence, the evolutionary mechanism wins over and the molecular evidence is shown to reveal an evolutionary pattern.
Now to the fossil record. Similar to the creationists, Denton proposes that there are not enough transitional forms in the fossil record, and that what transitional forms do exist are rather dubious in nature because they do not show soft organ changes (since organs aren't fossilized) and are not intermediate in every single characteristic. Thus, Archaeopteryx lithographica, perhaps the most famous transitional form of all time, is inadequate because its wings and feathers were fully formed. Never mind that Archaeopteryx sported more reptilian skeletal characteristics than it did avian ones, and ignore the fact that its skeletal characteristics very closely match a class of wingless reptiles called therodonts that existed around the same time and in the same geographical location. It is apparent, then, that Denton's standards for labeling a fossil a "transitional form" are unreasonably restrictive.
As far as soft organ characteristics go, researchers can indeed also obtain information on a fossilized organism's soft anatomy and behavior by paying attention to structural characteristics. Thus, the recent find of Ambulocetus natans (Berta, 1994) lends further credence to the ungulate-to-whale transition that Denton finds so incredulous, because the skeletal arrangement of the creature is such that it could undulate its spine to produce a motion not unlike that of the whale's tail motion (not to mention that Ambulocetus was found geographically and temporally just about where evolutionists hoped to find it). Similarly, examination of the configuration of basal ridges in fossilized reptile-mammal transitional form skulls shows how endothermy developed gradually, even though the evolution of the soft, complex endothermic apparatus could not be directly observed (Hillenius, 1994).
Denton's denial of the fish-to-amphibian transition as demonstrated by Eusthenopteron (a late Devonian fish) and Icthyostega (a late Devonian amphibian) is a striking example of the excessive demands he makes on transitional forms. He echoes creationist Duane Gish's criticism that Icthyostega has well developed limbs for terrestrial movement, while Eusthenopteron has mere fins. In the first place, it is unreasonable to expect a transitional form to be transitional in every single skeletal characteristic it exhibits. Secondly, the correlation between the skull and vertebral characteristics of the two creatures is impossible to account for in a framework of typology:
The crossopterygian fish Eusthenopteron is linked to the early amphibian Icthyostega by a number of characteristics: (1) same pattern of skull bones as Icthyostega, (2) internal nostrils (found only in land animals and sarcopterygians - a taxonomic group encompassing lungfish and crossopterygians), (3) teeth like amphibians', (4) a two-part cranium (icthyostegids are the only other vertebrates that have this characteristic), and (5) same vertebral structure. (derived from McGowan, 1984, 152-153)
Every once in a while, a new transitional form turns up and strengthens the evolutionary pattern inherent in the fossil record . But in any case, to deny that the ones that already exist are what they appear to be is a sure indication of unreasonably strict criteria for transitional forms.
We have seen that Denton's assault on evolutionary pattern has failed. The "problems" he points out with the evolutionary interpretation of the data have turned out to not be problems at all. It has additionally been shown - thanks to transitional forms, true homologous structures, and the various novelties that appear in embryological and molecular realms not subject to selective pressures - that deriving a typological pattern from the data is itself a problematic undertaking.
Evolutionary pattern and process stands vindicated from Denton's assault. It does not win out by default, being implausible but socially established and lacking a superior alternative - rather, it is a plausible process with no contenders, and is backed up strongly by empirical evidence. There is debate within evolutionist circles about systematics, tempo, and the roles of genetic drift and preadaptation, and still plenty of work to be done fleshing out stories about the development of certain structures, but none of this in any way puts macroevolution and the pattern of non-teleological common descent in a crisis situation. Rather, they are indicators that evolutionary biology is still a field which offers work to be done, just like any other field.
1My thanks to Elizabeth Hart for bringing Brassica to my attention.
2A creationist critic responded to me at this point by saying that all dogs are members of the same species, and that their variation thus does not constitute evidence for evolution. However, this critic misses my point entirely. What I was attempting to show was that large structural change could occur naturally. The fact that such change can occur within the same species makes the case for macroevolution stronger, not weaker.
3My thanks to Chris Colby for pointing out this resource to me.
Home Page |
The FAQ | Must-Read Files | Index | Creationism | Evolution | Age of the Earth | Flood Geology | Catastrophism | Debates