Differential lineage extinction, though technically a form of selection, is not enough in itself to generate progressive evolutionary change. Lineages may be ‘survivors’, but this does not make them replicators. Grains of sand are survivors. Hard grains, made of quartz or diamond, will last longer than soft grains made of chalk. But nobody has ever invoked hardness selection among sand grains as the basis for an evolutionary progression. The reason, fundamentally, is that grains of sand do not multiply. One grain may survive a long time, but it does not multiply and make copies of itself. Do species, or other groups of organisms, multiply? Do they replicate?
Alexander and Borgia (1978) assert that they do, and that they are therefore true replicators: ‘Species give rise to species; species multiply.’ The best case I can make for regarding species, or rather their gene-pools, as multiplying replicators arises from the theory of ‘species selection’ associated with the palaeontological idea of ‘punctuated equilibria’ (Eldredge & Gould 1972; Stanley 1975, 1979; Gould & Eldredge 1977; Gould 1977c, 1980a,b; Levinton & Simon 1980). I will take some time to discuss this body of theory, since ‘species selection’ is very relevant to this chapter. Another reason for taking the time is that I regard the suggestions of Eldredge and Gould as of great interest to biology generally, but I am anxious that they should not be oversold as more revolutionary than they actually are. Gould and Eldredge (1977, p. 117) are themselves conscious of this danger, though for different reasons.
My fear stems from the growing influence of a vigilant corps of lay critics of Darwinism, either religious fundamentalists or Shavian/Koestlerian Lamarckists who, for reasons that have nothing to do with science, eagerly seize upon anything that, with imperfect understanding, can be made to sound anti-Darwinian. Journalists are often only too ready to pander to the unpopularity of Darwinism in some lay circles. One of Britain’s least disreputable daily newspapers (The Guardian, 21 November 1978) served up a journalistically garbled but still just recognizable version of the Eldredge/Gould theory in a leading article, as evidence that all is not well with Darwinism. Predictably, this elicited some uncomprehending fundamentalist glee in the letter columns of the paper, some of it from disquietingly influential sources, and the public could well have been left with the impression that even ‘the scientists’ themselves now have doubts about Darwinism. Dr Gould informs me that The Guardian did not favour him with a reply to his letter of protest. Another British newspaper, The Sunday Times (8 March 1981), in a much longer article called ‘The new clues that challenge Darwin’, sensationally exaggerated the difference between the Eldredge/Gould theory and other versions of Darwinism. The British Broadcasting Corporation also got in on the act around the same time, in two separate programmes made by rival production teams. They were called The Trouble with Evolution and Did Darwin get it Wrong?, and they differed hardly at all except that one had Eldredge and the other had Gould! The second programme actually went to the lengths of digging up some fundamentalists to comment on the Eldredge/Gould theory: not surprisingly, the ill-understood appearance of dissension within the ranks of Darwinists was meat and drink to them.
Journalistic standards are not unknown in learned periodicals too. Science (Vol. 210, pp. 883–887, 1980) reported on a recent conference on macroevolution under the dramatic heading, ‘Evolutionary theory under fire’, and with the equally sensational subtitle, ‘An historic conference in Chicago challenges the four-decade long dominance of the Modern Synthesis’ (see criticisms by Futuyma et al. 1981). As Maynard Smith is quoted as saying at the same conference, ‘You are in danger of preventing understanding by suggesting that there is intellectual antagonism where none exists’ (see also Maynard Smith 1981). In the face of all the ballyhoo, I am anxious to be very clear about, to quote one of their own section headings, exactly ‘What Eldredge and Gould Did Not (And Did) Say’.
The theory of punctuated equilibria suggests that evolution does not consist in continuous smooth change, ‘stately unfolding’, but goes in jerks, punctuating long periods of stasis. Lack of evolutionary change in fossil lineages should not be written off as ‘no data’, but should be recognized as the norm, and as what we should really expect to see if we take our modern synthesis, particularly its embedded ideas of allopatric speciation, seriously: ‘As a consequence of the allopatric theory, new fossil species do not originate in the place where their ancestors lived. It is extemely improbable that we shall be able to trace the gradual splitting of a lineage merely by following a certain species up through a local rock column’ (Eldredge & Gould 1972, p. 94). Of course microevolution by ordinary natural selection, what I would call genetic replicator selection, goes on, but it is largely confined to brief bursts of activity around the crisis times known as speciation events. These bursts of microevolution are usually completed too fast for palaeontologists to track them. All we can see is the state of the lineage before and after the new species is formed. It follows that ‘gaps’ in the fossil record between species, far from being the embarrassment Darwinians have sometimes taken them to be, are exactly what we should expect.
The palaeontological evidence can be argued about (Gingerich 1976; Gould & Eldredge 1977; Hallam 1978), and I am not qualified to judge it. Approaching from a non-palaeontological direction, indeed in lamentable ignorance of the whole Eldredge/Gould theory, I once found the Wrightian/Mayrian idea of buffered gene-pools resisting change, but occasionally succumbing to genetic revolutions, satisfyingly compatible with one of my own enthusiasms, Maynard Smith’s (1974) ‘evolutionarily stable strategy’ concept:
The gene pool will become an evolutionarily stable set of genes, defined as a gene pool which cannot be invaded by any new gene. Most new genes which arise, either by mutation or reassortment or immigration, are quickly penalized by natural selection: the evolutionarily stable set is restored. Occasionally … there is a transitional period of instability, terminating in a new evolutionarily stable set … a population might have more than one alternative stable point, and it might occasionally flip from one to another. Progressive evolution may be not so much a steady upward climb as a series of discrete steps from stable plateau to stable plateau [Dawkins 1976a, p. 93].
I am also quite impressed with Eldredge and Gould’s reiterated point about time-scales: ‘How can we view a steady progression yielding a 10% increase in a million years as anything but a meaningless abstraction? Can this varied world of ours possibly impose such minute selection pressures so uninterruptedly for so long?’ (Gould & Eldredge 1977). ‘… to see gradualism at all in the fossil record implies such an excruciatingly slow rate of per-generation change that we must seriously consider its invisibility to natural selection in the conventional mode—changes that confer momentary adaptive advantages’ (Gould 1980a). I suppose the following analogy might be made. If a cork floats from one side of the Atlantic to the other, travelling steadily without deviating or going backwards, we might invoke the Gulf Stream or Trade Winds in explanation. This will seem plausible if the time the cork takes to cross the ocean is of the right order of magnitude, say a few weeks or months. But if the cork should take a million years to cross the ocean, again not deviating or going backwards but steadily inching its way across, we should not be satisfied with any explanation in terms of currents and winds. Currents and winds just don’t move that slowly, or if they do they will be so weak that the cork will be overwhelmingly buffeted by other forces, backwards as much as forwards. If we found a cork steadily moving at such an extremely slow rate, we would have to seek a wholly different kind of explanation, an explanation commensurate with the time-scale of the phenomenon observed.
Incidentally, there is a mildly interesting historical irony here. One of the early arguments used against Darwin was that there wasn’t enough time for the proposed amount of evolution to have happened. It seemed hard to imagine that selection pressures were strong enough to achieve all that evolutionary change in the short time then thought to be available. The argument of Eldredge and Gou
ld just given is almost the exact opposite: it is hard to imagine a selection pressure weak enough to sustain such a slow rate of unidirectional evolution over such a long period! Perhaps we should take warning from this historical twist. Both arguments resort to the ‘hard to imagine’ style of reasoning that Darwin so wisely cautioned us against.
Although I find Eldredge and Gould’s time-scale point somewhat plausible, I am less confident about it than they are, since I do fear the limitations of my own imagination. After all, the gradualist theory does not really need to assume unidirectional evolution for long periods. In terms of my cork analogy, what if the wind is so weak that the cork takes a million years to cross the Atlantic? Waves and local currents may indeed send it backwards almost as much as forwards. But when all is added up, the net statistical direction of the cork may still be determined by a slow and relentless wind.
I also wonder whether Eldredge and Gould give sufficient attention to the possibilities opened up by ‘arms races’ (Chapter 4). In characterizing the gradualist theory that they attack, they write: ‘The postulated mechanism for gradual uni-directional change is “orthoselection”, usually viewed as a constant adjustment to a uni-directional change in one or more features of the physical environment’ (Eldredge & Gould 1972). If the winds and currents of the physical environment pressed steadily in the same direction over geological time-scales, it might indeed seem likely that animal lineages would reach the other side of their evolutionary ocean so fast that palaeontologists could not track their passage.
But change ‘physical’ to ‘biological’ and things may look different. If each small adaptive step in one lineage calls forth a counteradaptation in another lineage—say its predators—slow, directional orthoselection looks rather more plausible. The same is true of intraspecific competition, where the optimum size, say, for an individual can be slightly larger than the present population mode, whatever the present population mode may be. ‘… in the population as a whole there is a constant tendency to favor a size slightly above the mean. The slightly larger animals have a very small but in the long run, in large populations, decisive advantage in competition … Thus, populations that are regularly evolving in this way are always well adapted as regards size in the sense that the optimum is continuously included in their normal range of variation, but a constant asymmetry in the centripetal selection favors a slow upward shift in the mean’ (Simpson 1953, p. 151). Alternatively (and shamelessly having it both ways!), if evolutionary trends really are punctuated and stepped, perhaps this in itself could be explained by the arms race concept, given time-lags between adaptive advances on one side of the arms race and responses by the other.
But let us provisionally accept the theory of punctuated equilibria as an excitingly different way of looking at familiar phenomena, and turn to the other side of Gould and Eldredge’s (1977) equation, ‘punctuated equilibria + Wright’s rule = species selection’. Wright’s Rule (not his own coining) is ‘the proposition that a set of morphologies produced by speciation events is essentially random with respect to the direction of evolutionary trends within a clade’ (Gould & Eldredge 1977). For example, even if there is an overall trend towards larger size in a set of related lineages, Wright’s Rule suggests that there is no systematic tendency for newly branched-off species to be larger than their parent species. The analogy with the ‘randomness’ of mutation is clear, and this leads directly to the right-hand side of the equation. If new species differ from their predecessors randomly with respect to major trends, major trends themselves must be due to differential extinction among those new species—‘species selection’, to use Stanley’s (1975) term.
Gould (1980a) regards ‘the testing of Wright’s rule as a major task for macroevolutionary theory and paleobiology. For the theory of species selection, in its pure form, depends upon it. Consider, for example, a lineage displaying Cope’s rule of increasing body size—horses, for example. If Wright’s rule be valid, and new species of horses arise equally often at sizes smaller and larger than their ancestors, then the trend is powered by species selection. But if new species arise preferentially at sizes larger than their ancestors, then we don’t require species selection at all, since random extinction would still yield the trend.’ Gould here simultaneously sticks his neck out and hands Occam’s Razor to his opponents! He could easily have claimed that even if his mutation-analogue (speciation) was directed, the trend might still be reinforced by species selection (Levinton & Simon 1980). Williams (1966, p. 99) in an interesting discussion which I have not seen cited in the literature on punctuated equilibria, considers a form of species selection acting in opposition to, and perhaps outweighing, the overall trend of evolution within species. He again uses the horse example, and the fact that earlier fossils tend to be smaller than later ones:
From this observation, it is tempting to conclude that, at least most of the time and on the average, a larger than mean size was an advantage to an individual horse in its reproductive competition with the rest of its population. So the component populations of the Tertiary horse-fauna are presumed to have been evolving larger size most of the time and on the average. It is conceivable, however, that precisely the opposite is true. It may be that at any given moment during the Tertiary, most of the horse populations were evolving a smaller size. To account for the trend towards larger size it is merely necessary to make the additional assumption that group selection favoured such a tendency. Thus, while only a minority of the populations may have been evolving a larger size, it could have been this minority that gave rise to most of the populations of a million years later.’
I do not find it hard to believe that some of the major macroevolutionary trends, of the Cope’s Rule type (but see Hallam 1978), observed by palaeontologists are due to species selection, in the sense of this passage from Williams, which I think is the same as Eldredge and Gould’s sense. As I believe all three authors would agree, this is a very different matter from accepting group selection as an explanation for individual self-sacrifice: adaptations for the good of the species. There we are talking about another kind of group selection model, where the group is really seen not as a replicator but as a vehicle for replicators. I shall come to this second kind of group selection later. Meanwhile I shall argue that a belief in the power of species selection to shape simple major trends is not the same as a belief in its power to put together complex adaptations such as eyes and brains.
The major trends of palaeontology, simple increases in absolute size, or of relative sizes of different parts of the body, are important and interesting, but they are, above all, simple. Accepting Eldredge and Gould’s belief that natural selection is a general theory that can be phrased on many levels, the putting together of a certain quantity of evolutionary change demands a certain minimum number of selective replicator-eliminations. Whether the replicators that are selectively eliminated are genes or species, a simple evolutionary change requires only a few replicator substitutions. A large number of replicator substitutions, however, are needed for the evolution of a complex adaptation. The minimum replacement cycle time when we consider the gene as replicator is one individual generation, from zygote to zygote. It is measured in years or months, or smaller time units. Even in the largest organisms it is measured in only tens of years. When we consider the species as replicator, on the other hand, the replacement cycle time is the interval from speciation event to speciation event, and may be measured in thousands of years, tens of thousands, hundreds of thousands. In any given period of geological time, the number of selective species extinctions that can have taken place is many orders of magnitude less than the number of selective allele replacements that can have taken place.
Again it may be just a limitation of the imagination, but while I can easily see species selection shaping a simple size trend such as the Tertiary elongation of horse legs, I cannot see such slow replicator-elimination putting together a suite of adaptations such as those of whales to aquatic life. Now,
it may be said, surely that is unfair. However complex the aquatic adaptations of whales may be when taken as a whole, can they not be broken down into a set of simple size trends with allometric constants of varying magnitude and sign in different parts of the body? If you can stomach one-dimensional elongation in horses’ legs as due to species selection, why not a whole set of equally simple size trends advancing in parallel, and each one driven by species selection? The weakness in this argument is a statistical one. Suppose there are ten such parallel trends, which is surely a highly conservative estimate for the evolution of aquatic adaptations in whales. If Wright’s Rule is to apply to all ten, in any one speciation event each of the ten trends is as likely to be reversed as to be advanced. The chance that all ten will be advanced in any one speciation event is one half to the power ten, which is less than one in a thousand. If there are twenty parallel trends, the chance of any one speciation event advancing all twenty simultaneously is less than one in a million.
Admittedly, some advancement towards complex multidimensional adaptation might be achieved through species selection even if not all ten (or twenty) trends were advanced together in any one selective event. After all, much the same critical point can be made about the selective deaths of individual organisms: it is rare to find one individual animal that is optimal in all the different dimensions of measurement. The argument finally returns to the difference in cycle times. We shall have to make a quantitative judgement taking into account the vastly greater cycle time between replicator deaths in the species selection case than in the gene selection case, and also taking into account the combinatorial problem raised above. I have neither the data nor the mathematical skills to undertake this quantitative judgement, though I have a dim feeling for the kind of methodology that would be involved in setting up an appropriate null hypothesis: it would fall under the general heading which I like to think of as ‘What D’Arcy Thompson might have done with a computer’, and programs of the appropriate type have already been written (Raup et al. 1973). My provisional guess is that species selection will not be found to be a generally satisfactory explanation of complex adaptation.