Unofficial SJG Archive

The Unofficial Stephen Jay Gould Archive

Unofficial SJG Archive


      Horotely, Bradytely, and

t is abundantly evident that rates of evolution vary. They vary greatly from group to group, and even among closely related lineages there may be strikingly different rates. Differences in rates of evolution, and not only divergent evolution at comparable rates, are among the reasons for the great diversity of organisms on the earth. Among the living primates there are, for instance, some rather unspecialized or primitive prosimians (i.e., little changed from Eocene progenitors), a larger number of divergently specialized prosimians, many monkeys of different degrees of progression and divergence, a few apes, and the unique species of man. Important as is the purely divergent evolution, it is also clear that differential rates are involved. At the extremes, the lineages of the more primitive living prosimians have evolved less rapidly as regards the whole of their structure and adaptive position than has the lineage of man.

Anyone group, large or small, seems to have a fairly characteristic mean or modal rate of evolution and a certain range of less common rates on each side of this. The rates have a distribution pattern which is open to investigation and which is an important element in evolution. It has long been noticed, further, that there are a few lines in most large groups that seem to evolve at altogether exceptional rates. Some hardly change at all over long periods of time. Others change, over short periods of time, at rates so rapid as to be almost beyond comparison with those usual among their relatives. It must be considered whether these extraordinary rates are merely the extremes of normal variation in rates or whether they reflect special circumstances different from those in the distributions of usual rates. In either case they must involve different intensities, combinations, or both as regards


evolutionary determinants, and it would be of extreme interest, perhaps even of considerable practical value, to identify these.


In Chapters I and II the examples of how rates may be estimated and represented showed that: (a) related lines of descent commonly differ in evolutionary rates; (b) within larger taxonomic groups such as orders or classes there is generally an average or modal rate typical for the group; and (c) the average rates may differ greatly from one group to another. Since rates do vary broadly around a group mode, rate distributions of different groups tend to overlap widely even though the modes are quite distinct. Mammals seem surely to have a much higher model rate of evolution than molluscs, but some mammals have evolved more slowly than some molluscs. It even seems clear that some mammals have evolved at rates below the mode for molluscs and some molluscs at rates above the mode for mammals.

Such comparisons are hazardous because there is no absolute way to measure equivalent amounts of evolution in two groups anatomically so different and changing in such different ways. The two cannot even both be studied authoritatively by one specialist. This sort of objection is often made, and of course it carries a great deal of force. It is nevertheless not as strong or as fatal to comparisons as sometimes claimed. For instance, it is often said (e.g., again by Hutchinson, 1945, in criticism of the predecessor of this book) that molluscs may have evolved very rapidly in the soft parts without having this reflected in the shell. But in molluscs the whole skeleton is usually available and studied in fossils (which is very rarely true of, say, fossil mammals) and their skeleton reveals a great deal about the soft parts, directly (muscle scars, etc.) or indirectly (growth rates, thickness and chemical nature of shell, twisting of body, etc.). Percentagewise a mollusc shell probably tells more about the soft parts than do the usually available parts of a vertebrate. It is extremely unlikely that any marked progressive change in soft parts could occur without obvious changes in the shell.

We need not, then, bend over backward in denying validity to comparisons of rates between groups, even such very disparate groups as molluscs and mammals, and of course this applies all the more strongly to comparisons between groups that are similar, such as different sorts of molluscs or of mammals. There is, too, another sort of comparison


that is really very little affected by this particular difficulty. The rate distributions can be compared in any two groups without regard for the absolute values of the rates. It was seen in Chapter II that survivorship curves are quite comparable even when the absolute rates involved are altogether different. There is little reason to doubt that both the resemblances and differences in shapes of such curves can be significant facts that are independent of our ability (or inability) to equate individual rates in the groups compared.

Survivorship tends to be negatively correlated with rate of evolution. By and large, the longer a genus, say, endures the more slowly its included populations are evolving. The correlation cannot be exactly –1.00 and it cannot be well measured, but it is a reasonably safe approximation when we can average out fairly large bodies of data. This provides a very convenient way to approach the problems of the distributions of evolutionary rates. By assuming that evolutionary rates are the reciprocals of survivorship, it is possible to calculate from any survivorship curve (or tabular data) what percentage of the population has evolved at any particular rate. For purposes of comparing different distributions, it is further possible and desirable to divide the scale of rates not in absolute but in relative terms, e.g., in decile classes of range.

Such rate distributions for pelecypod and mammalian land carnivore genera have been calculated from the survivorship data of Figure 5, Chapter II, and are graphically presented in Figure 39. For reasons that will become clear, these rate distributions are based on extinct genera, only. The two curves have a characteristic difference: the carnivore mode is more strongly peaked and classes just below the mode are correspondingly lower. Nevertheless the two are strikingly similar in these respects: (1) both have a single mode from which frequencies drop in both directions; (2) both are more peaked (leptokurtic) than a normal curve; (3) both are strongly asymmetrical with the mode much nearer the upper (faster rate) end of the distribution (negatively skewed). It happens also that in both the mode is in the ninth decile class, and that the amount of skew is comparable in the two.

Rate distributions of this sort have so far been made for only a few groups, although the number has increased since 1944 when these were first published. Those that have been studied all agree in the general characteristics (1)–(3), above. It is probable that these are generaliza-


FIGURE 39. FREQUENCY DISTRUBUTIONS OF RATES OF EVOLUTION IN GENERA OF PELECYPODS AND LAND CARNIVORES. Histograms based on survivorship of extinct genera (see Figs. 5-6) and on the postulate of—1.00 correlation of survivorship and rates of evolution. Ranges (abscissal scale) divided into deciles to produce histograms strictly comparable in form although absolute rates are very different. Normal curves equal in area to the histograms drawn for comparison.

tions true of most distributions of evolutionary rates. Distributions may however, differ markedly in degree of leptokurtosis (as do the pelecypod and carnivore generic distributions to some extent) and of negative skew (as these do not to any marked degree). Figure 40 shows a distribution that is near the extreme in both respects, rates of evolution in the 1,374 extinct species of centric diatoms, calculated from raw data published by Small (1946). Kurtosis is very great, with no less than 73 percent of the total frequency in a single decile class as against about 26 percent in each of the two middle decile classes of the normal curve. Skewness is also extreme, with the mode in the tenth decile class (it is, of course, at the fifth decile, between fifth and sixth decile classes, in the normal curve). If decile classes, only, are shown, such a curve seems to be J-shaped and an exception to (1), above, but the frequency does fall off to the right of the mode, as suggested schematically at the top of the figure. Finer subdivision would show that the mode is near the


FIGURE 40. DISTRIBUTION OF EVOLUTIONARY RATES IN EXTINT SPECIES OF CENTRIC DIATOMS. (Calculated from tabulation of raw survival data compiled by Small, 1946.)

ninth decile and that frequency does fall rapidly from it to zero at the tenth decile. There is also some slight irregularity in the dropping off of frequencies to the left (in the fourth and sixth decile classes), but this is evidently mere sampling fluctuation here where frequencies are very low. The distribution is of the same type as those of Figure 39, but more extreme.

The peakedness of these distributions shows that among related lineages, most of them tend to evolve at about the same rate. In extinct pelecypod genera about a third, in extinct carnivore genera about half, and in extinct centric diatom species about three-fourths evolved at rates included in only one-tenth of the whole range of variation in rates. These concrete data confirm the vague and subjective estimation of many paleontologists that each group of organisms does have a characteristic rate of evolution approximated by most lineages in the group.

The asymmetry of the distributions shows that the characteristic (modal) rate is decidedly nearer the maximum for the group than the minimum. Rates above the modal class do not extend far beyond it


(are not much faster) and are, moreover, exhibited by relatively few lineages. In pelecypods and carnivores only 9 to 10 percent of the genera had rates faster than the modal decile class; none of the diatom species have rates above the modal decile class. Rates below the mode are more numerous than those above and also extend farther (are much slower relative to the mode than those above the mode are faster). In pelecypods more than half the rates are below the modal decile class, in carnivores more than a third are, and in centric diatoms more than a quarter. Although it was less clearly noticed before rate distributions were objectively calculated, this tendency, too, agrees with opinions occasionally expressed on the basis of subjective impression by earlier paleontologists (e.g., Matthew, 1914).

These distributions, based on taxonomic units that had run their full span and had become extinct, show that there is a definite range and pattern of such rates. The pattern is similar in different groups (so far as yet demonstrated), but it also has characteristic peculiarities in each. It gives a picture of a sort of evolutionary metabolism, one might say, for each group as a whole. (Of course I do not mean that it is an internal thing, like individual metabolism; it reflects the evolution of adaptation and the turnover of taxonomic units in the course of this.) It is this distribution that I have called "horotelic" (Simpson, 1844a). Organisms involved in it have horotelic rates, and evolution at rates so distributed is horotely.1

Further study of a number of groups, including the pelecypods and diatoms, shows that some of their lineages have evolved at rates much slower than any in the horotelic distribution as calculated from fossil records of extinct taxonomic units, only. Further studies show that these extremely low rates are part of a statistical excess of low rates in general in comparison with the horotelic distribution. This low-rate, nonnhorotelic excess is bradytely and the rates involved in it are bradytelic. Other considerations strongly suggest that some phases of evolution have also involved rates higher than any in distributions obtained in this way. That exceptionally fast evolution is tachytely and it moves at tachytelic rates.

1 When I proposed the terms horotely, bradytely, and tachytely, I apologized for adding to the technical terminology of evolution and explained that I had been quite unable to discuss rates clearly without this invention. As expected, at least one critic did object to the use of new terms, but apparently most others agreed that they make for clarity as they have come into rather general use.


Bradytely and tachytely are to be separately discussed in the remainder of this chapter. Before their detailed consideration, however, it may be pointed out that, although these concepts are derived mainly from analysis of the fossil record, it is sometimes possible to apply them with reasonable probability to groups inadequately or not known as fossils. Thus Ross (1951) has made a careful analysis of a tribe of caddis flies (Hydrobiosini) and its probable zoogeographic history. He adduces cogent evidence that generic splitting of the group began in the late Cretaceous and that of its 13 genera: (a) one has hardly changed at all since that time; (b) eleven have changed conspicuously and to varying degrees without becoming extremely different, and (c) one has undergone extreme change and reorientation of characters. His reasonable conclusion is that (a) is bradytelic, (b) represent together the horotelic distribution, and (c) is tachytelic (or it might be better to say, its lineage has been tachytelic at some time since the late Cretaceous—persistence of tachytely for so long a period is unlikely).


Evolution occurs at a great variety of rates and the lower limit of rate is obviously zero. In view of the constant flux of the environment and constant mutation and genetic recombination in organisms, a sustained rate of precisely zero would hardly be expected. It would, however, be expected that rate distributions like those of Figures 39 and 40 would tail off to the left to a point not far from zero and would include a few, rare lineages that had not changed appreciably for long periods of time or had fluctuated without appreciable additive effects. Such lineages do occur, as has long been recognized. They are of great interest and the extreme slowness of their evolution calls for explanation.

Further analysis of rate distributions has, however, revealed that there is something more to explain than merely the slow rate of an occasional lineage in the tail of a falling distribution curve. In certain groups, although not in all, it turns out that rates near zero are actually more frequent than are slow rates in horotelic distributions calculated as above. It also appears that there may be a discontinuity in survivorship. In some groups, if the history of lineages arising at a given time is followed, their extinction rises to an early peak and then tapers off gradually, as expected; but there finally comes a time when no more extinctions occur over geologically long periods of time. This point


FIGURE 41. FREQUENCY DISTRUBUTIONS OF RATES OF EVOLUTION IN GENERA OF PELECYPODS AND LAND CARNIVORES. Histograms based on survivorship of extinct genera (see Figs. 5-6) and on the postulate of—1.00 correlation of survivorship and rates of evolution. Ranges (abscissal scale) divided into deciles to produce histograms strictly comparable in form although absolute rates are very different. Normal curves equal in area to the histograms drawn for comparison.

should be reached when the number of existing lineages drops to zero. But in some cases lineages do continue to survive; their number has not reached zero. Survivorship then has not dropped gradually to zero but reached a definite point in the history of each group where a discontinuity in survivorship occurred.

This odd phenomenon appears and can be analyzed in various ways. For instance, of the 54 broad genera of pelecypods first appearing in the Devonian, 30 became extinct in that period, 10 in the next, and so on down to 1 in the Jurassic. None became extinct thereafter, but 4 survive today, and will for some unmeasurable time longer (see Table 10 in Chapter II). Even when such a gap does not appear in our often rather crude data, an effect of what is surely the same phenomenon is seen in discrepancies between rate distributions calculated in different ways.

If the rates of evolution of living genera are distributed in the same way as rates of extinct genera, then a cumulative curve of ages of recent genera will be exactly the same as a survivorship curve for extinct genera, aside from fluctuations of sampling. The examples of such


curves given in Chapter II showed that this definitely is not true for curves based on pelecypod genera. Schindewolf (195Gb), although critical of those pelecypod data (and with some reason), obtained the same sort of discrepancy with better data on prosobranch gastropods. Lest it be thought that this result depends on use of genera or is limited to molluscs, it is noted also that calculation of the same curves for Small's species of centric diatoms produces the same sort of discrepancy in still greater degree (Fig. 41).

The discrepancy arises from the fact that more genera or species have survived from remote times in the past than would be expected from the horotelic rate distribution for extinct genera. In other words, there have been more lineages with rates near zero than that sort of distribution indicates. The same discrepancy is clearly shown by comparison of actual age composition of the recent fauna with expecta-

Time of Appearance (in millions of years previous to present)  A. Percentage of Expected Survival from Survivorship in Extinct Genera  B. Percentage of Realized Survival  Difference
25  92  78  –14
50  63  62  –1
100  24  39  15
150    8  30  22
200    2  24  22
250    1  19  18
300    0  13  13
350    0    8    8
400    0    4    4
450    0    0    0

tion on the basis of survivorship among extinct forms (Table 24, here, also Table 12 in Chapter II).

In carnivores actual survival is somewhat less than expectation throughout (Table 12), but this has affected genera of all ages in about the same proportion. As a result, the recent carnivore fauna is less varied than would have been expected from the Tertiary record, but its age composition, as proportions of genera surviving from various times in the past, is almost exactly as expected (Fig. 42). Rate of ex-


FIGURE 42. REALIZED AND EXPECTED AGE COMPOSITIONS OF RECENT PELECYPOD AND LAND CARNIVORE FAUNAS. Percentage of living genera arising (first appearing in fossil record) in stated epochs and periods as observed in the recent fauna (counting only genera that are known as fossils) and as would have been predicted on the basis of survivorship in extinct genera.

tinction was higher than expectation from late Pliocene to Recent, but it did not tend to affect later genera more than earlier, or earlier more than later. Moreover no living genus is older than the maximum span for extinct genera. In this group the discrepancy under consideration docs not occur, and the inference is that there are no bradytelic carnivores.

In pelecypods and gastropods, on the other hand, actual survival is greater than expected survival (Table 12, Chapter II) and these groups are today much richer, more varied, than expected on the basis of their horotelic distribution. Survival of late genera, those arising in the Tertiary, is near expectation: a little low for pelecypods and high for gastropods but in both cases as near as would be probable with sampling effects. Survival of pre-Tertiary genera, from the Ordovician for the very broad pelecypod genera and from the Triassic for the narrower gastropod genera, is, on the other hand, much greater than


horotelic expectation. The result in terms of percentage composition of the recent fauna is that there is a much smaller proportion (not absolute number) of young genera and a much higher proportion of older genera than the expectation, as is well shown in Figure 42 for pelecypods.

Moreover, even though their spans have not yet ended, some living genera have already had much longer spans than any extinct genera of the same group. Some broadly defined living genera of pelecypods have been in existence for upward of 400 million years, but the longest-lived of extinct genera, defined equally broadly, endured no more than 275 million years. In the centric diatoms, there are living species on the order of 75 million years of age but no extinct species more than about 50 million years old. It is such facts that indicate a discontinuity in survivorship. On the other hand, in the prosobranch gastropods (Schinndewolf's data as previously noted) the oldest living genera are about 200 million years old and there is at least one extinct genus that endured for 250 million years or somewhat more. This shows that there are complicating factors,2 but does not alter the fact that this group, too, shows clear-cut excess of survivorship of old genera in the recent fauna.

The phenomenon here noted has of course long been discussed in a more general way or in other terms and techniques. T. H. Huxley (e.g., 1870) pointed out as early as 1862 that certain types of life are unusually "persistent." The phenomenon of "arrested evolution" (Ruedemann, 1918) has long engaged the attention of paleontologists. Their earlier analysis was, however, even more unsatisfactory than at present because the usual criterion of "persistence" was quite arbitrary and it was not shown whether the rates of evolution involved were really slow for the particular group in which they appeared. Thus Ruedemann considered as "persistent" and cases of "arrested evolution" any genera, from protistan to reptilian (there are no avian or mammalian examples) that

2 A really detailed analysis of this group cannot be attempted here, and should be done as a separate study by a specialist in fossil gastropods, but the suggestion may be made that high Triassic mortality (related to the Permian-Triassic crisis for marine vertebrates) may in this case have caused the unusual extinction of bradytelic lines. The five genera surviving from Ordovician to Triassic were probably bradytelic and the three then surviving from the Silurian were possibly so. Four of the former and two of the latter became extinct in the Triassic, and the last Ordovician and Silurian genera disappeared in the Jurassic. Except for these few genera, the maximum span for extinct genera is under 200 million years, hence somewhat, at least, less than for the oldest living genera.


appeared in more than two geological periods, in spite of the fact that this might mean survival for anything from 30 million years upward and that in some groups survival of broadly defined genera for such spans is the rule rather than the exception.

Rensch (1947) has demonstrated that the mean age of living higher categories (families to orders) is generally greater than the mean span of extinct units of the same categorical rank and in the same group, often two or three times as great. Since the long persistence of a family or an order does not preclude a fairly high rate of evolution within the group, these examples do not necessarily represent essential arrest of evolution, but the latter phenomenon is probably involved in such averages. It is quite certainly involved in the elaborate series of studies by Small (e.g., 1946), begun in the 1920's and still continuing, on "short" and "long" species, mostly in diatoms. Diatoms do show an unusually high degree of bradytely, as demonstrated above by a different analysis of some of Small's data, and Small seems to be quite right in concluding that two sorts of species, different in some quantitative way as regards rate of evolution, can be distinguished among these organisms.3

Thus there is nothing really new in the observation of what I have called "bradytely" (in print since 1944), but it does seem that rate distribution and survivorship methods put its study on a more nearly objective basis and permit clearer analysis of it.

A next step is suggested in Figure 43. In this particular group, the pelecypods, it appears that any living genus more than about 250 million years old, i.e., dating from about the Mississippian or earlier, is almost surely bradytelic. Such genera include Nucula, Nuculana (or Leda), Volsella (or Modiolus), and Pteria (or Avicula).4 This is sufficiently clear and these individual genera can be labeled "bradytelic" on a

3 It is necessary to add that Small's criterion for "shorts" and "longs" is arbitrary and does not really separate the two rate distributions. His conclusions as to the bearing of these data on evolutionary theory further seem to me beside the point and not really supported by the data. These personal opinions should not be allowed to obscure the real value of Small's laborious compilations.

4 Shimer and Shrock (1944), although reflecting fairly modern nomenclature and constriction of genera, still record Nucula and Nuclllana from the Silurian, Volsella and Pteria from the Devonian. Of course there is some difference between the earliest species and any living species in each of these genera, and sooner or later someone is sure to call that difference "generic." This will obscure but not alter the fact that the genus, as here recognized, has not shown appreciable progressive change since the Silurian or Devonian, although it has continued to undergo continual, noncumulaative speciation, as do all long-continued or widespread groups.


FIGURE 43. BRADYTELY IN PELECYPODS. Survivorship in extinct genera, continuous, unshaded curve, and analogous curve based on recent genera, broken curve, see Figure 5. The shaded area represents the excess of realized over expected survivorship caused by the nonextinction of bradytelic phyla and so is approximately proportional to the rise of bradytelic lines at the stated times in the past. (The apparent absence of bradytelic lines arising in the last 50,000,000 years may be real, but is probably an artifact caused by differential extinction of horotelic genera in this period, the extent of which cannot be estimated from the available data.)

reasonably objective basis. But the situation in, say, the Jurassic, around 150 million years ago, is different. The curves and tabular survivorship data give good reason to believe that perhaps three-fourths of the genera that survive from Jurassic to Recent and perhaps about one-fourth of the genera that arose in the Jurassic are bradytelic. But this is a statistical result, and like many such results it cannot be directly applied to the individual. There is no possible way from these data to tell which of the genera first appearing in the Jurassic are bradytelic and which are horotelic in the terms of this analysis. Such designation could only be made if some separate and individual criterion of bradytely could be found. If some Jurassic-Recent genera were found not to have changed at all and others merely to have changed very slowly, the former would include the bradytelic genera, but such a difference cannot be very clear-cut among lineages all of which have changed so little as not to warrant calling the change generic in broad, conservative


classification. Presumably some other features might particularize bradytelic genera but, as will appear in the next section, such criteria really diagnostic as to individual genera are not available. It is, indeed, likely that the statistical result does not depend on any inherent differences uniform in each individual line.

It thus appears that some of the oldest genera or species in some groups and after analysis of their rate distributions can be identified as bradytelic. Among younger forms, however, the analysis may clearly show that bradytelic lines were still arising and yet give no way of designating them individually. In other groups, as among prosobranch gastropods, the analysis indicates that bradytelic lines are present, but none can be individually designated, even among the oldest, because the same analysis shows that none is older than some horotelic lines. We can only go so far as to say that lines that have shown no appreciable change while most allied lines did show change must include the braclytelic lines, perhaps along with some that are slow horotelic, and may be more likely bradytelic than not.

Since this concept of bradytely was first published, many of those who have discussed it either have assumed that no distinction was intended between bradytely and very slow horotely or have denied that such a distinction exists. Most of these students have overlooked, or at least have made no reference to, the fact that bradytely is defined by a difference between rate distributions, a difference that does objectively exist. (There are, of course, noteworthy exceptions, e.g., Schindewolf, Carter.) It seems pointless to discuss whether a difference exists, unless it can be shown to be an artifact of the method, which has not been demonstrated or even suggested.5 It is also somewhat beside the point to designate some particular slowly evolving line as bradytelic, to show that its evolution had slowed down from a more usual, presumably horotelic line, and to deny any distinction between the two on that basis. The impossibility under some circumstances of designating particular lines as bradytelic has been pointed out above. It is also quite clear that all bradytelic lines are derived ultimately from horotelic or

5 Hutchinson (1945) did suggest an artifact of materials rather than of method: that the bradytelic genera are merely those in which evolution at normal rates affected the soft parts only, and so did not appear in fossils. This seems quite impossible, for reasons already mentioned in part. It may be added that this explanation cannot possibly apply to the diatoms, which include much bradytely in spite of the fact that living and extinct species are commonly based on identical criteria.


often from tachytelic ancestors. It is not to be assumed that they became abruptly bradytelic, although that may sometimes occur. Discontinuity in that sense, in application to change of rate within one line, is not involved in the evidence for or concept of bradytely.

It is very much to the point, however, to inquire what the difference between bradytely and horotely is, whether, for instance, it is quantitative and statistical, with intergradation and yet with mean differences and possible discontinuity of average outcome, or whether it is qualitative, with inherent discontinuity of sort and of underlying factors. Of these two alternatives, the evidence strongly suggests that the former is correct. In this sense, I agree with, for instance, Westoll (1949) that bradytely and very slow horotely are indistinguishable as seen in individual instances, but it remains true that in certain groups the prevalence of such exceedingly slow rates has produced a statistical discontinuity in rate distributions which is a collective phenomenon distinguishable from collective horotely. An opposing view is taken by Small (1946 and numerous other papers in the same series), who thinks his "shorts" and "longs" are inherently different as applied to individual cases, the short or long duration being "very largely determined by what happened to one nucleus" when the species arose by saltation. Extensive and conclusive evidence against such a view is given throughout this book, and the phenomenon of bradytely neither demands nor warrants the postulate of predetermined life span or any other factors than those normally involved in the complex evolution of adaptation.


Bradytely is essentially a statistical effect produced by the prevalence in some groups of organisms of lines with extremely low rates of evolution, or with changes fluctuating on a small scale and not appreciably cumulative. The first point, then, is to consider the circumstances of very slow or arrested evolution in general.

Certain genetic and other factors in organisms set limits to rates of evolution. It is therefore natural to suppose that slow evolution is primarily conditioned by these factors, and this has often been suggested. Nevertheless there is much evidence: (a) that the limits set by these factors in real populations are generally permissive of average or even high rates, and (b) that very slowly evolving lines are not particularly likely to have these factors in unusual degree.


Low mutation rate is one such factor often suggested (e.g., Dobzhansky, 1941, but the suggestion was practically withdrawn in 1951). But what appear to be low rates are sufficient to account for sustained and even rapid evolution (see Chapter IV). Mutation rates have not been directly determined in many slowly evolving lines, but the evidence tends to oppose the postulate that they have low rates. Drosophila, itself, may now be a very slowly evolving group, but seems to have rates of mutation comparable with those of any other group studied in this respect. The facts that very slowly evolving groups often arise by rapid evolution, that they also often split to give rise to rapidly changing branches, and that they do not have particularly low variation (below) all suggest indirectly that their mutation rates are more or less normal. Stebbins (1949, 1950) points out that slowly evolving plants may be rich in genetic variability, speciating luxuriantly and (e.g., Sequoiadendron) readily yielding horticultural varieties.

Factors that reduce genetic variability, notably apomixis, self-fertilization, and reduction of crossing over, certainly limit the possibilities for change and therefore may characterize groups that change but little ("blind alley" groups, Chapter IX). In the longer run, however, such groups generally do finally change or become unusually liable to extinction. Limitation of variation may, indeed, be a factor in bradytely of such groups as diatoms, but even in such cases it is all insufficient explanation and it does not apply at all to the majority of slowly evolving animals and plants. Most of these, even in lines surely bradytelic and with essentially no progressive evolution over very long spans, are sexually reproducing and show usual or even high degrees of variability. Opossums, which have evolved more slowly since the Cretaceous than any other living mammals, have quite as variable populations as the average of rapidly evolving mammals. Within the established adaptive type they also have undergone repeated, wide speciation, which implies a considerable store of variability. Stebbins (1949) also notes that slowly evolving plants may speciate freely. Selaginella, one of the most slowly evolving of all plants (little changed from Selaginellites in the early Mississippian) has over two hundred living species (Manton, 1950).

Long life spans or long generations, also frequently suggested as correlated with slow evolution, may be ruled out even more conclusively. As noted in Chapter V, there is surprisingly little evident


correlation between lengths of generations and rate of evolution. Here it may be added that most of the outstanding examples of slow or arrested evolution occur among organisms with relatively short generations. Bradytelic lines do not seem to have generations consistently different in length from their horotelic and tachytelic allies, and again it is pertinent that bradytelic lines may arise from and give rise to tachytelic lines. Surely change of length of generations is not a significant factor.

Another common approach, exemplified by Ruedemann (1918, 1922a, 19220), who devoted much attention to this subject a generation ago, is to seek particular morphological and physiological characteristics associated with low-rate lines. Small size, nocturnal habits, and carrion-feeding are among the many suggestions of this sort that have been made. Some such characters may have a bearing (although it is not likely that those just mentioned have much if any), but obviously they are not really explanatory. Among related lines fully comparable in such respects, some may evolve very slowly and others rapidly. Groups almost wholly different in characters of this sort, such as some plants and some animals, may have very similar slow—or fast—rates of evolution. It is perhaps true that sessile animals have evolved more slowly on an average than mobile animals, but this is a slender and probably misleading clue in view of the facts that similarly sessile animals have evolved at very different rates and that most plants are sessile but that they have run as broad a gamut of rates as have animals.

In such comparisons as that of sessile and mobile animals a certain confusion may arise. It is one problem that whole groups (say, corals) may have an average rate that is lower than that of another group (say, fishes) and quite a different problem that one line (among corals or among fishes) may have a very much lower rate than another allied and similar line. Both problems call for study, and they are not wholly unrelated, but it is stultifying to confuse the two. The frequently repeated generalization that primitive organisms have evolved more slowly than advanced ones lends itself to that confusion and also involves a howlingly circular argument as applied to groups now living: we call some of them "more primitive" precisely because they have evolved more slowly. If the statement is applied more correctly to older, hence literally more primitive, and younger members of the same group, it simply is not true. There is no definite tendency for


average rates within given groups to accelerate. Changes in such averages are highly variable and if there is an over-all balance it may be on the side of deceleration of rates. For instance, the later lungfishes evolved much more slowly than the primitive members of the group (Chapter II, and Westoll, 1949), and this seems to be a rather common pattern in long-lived groups.

As applied to individual lines, the idea that more primitive forms evolve more slowly is also a misconception. It is an interesting fact that bradytelic lines, as far as identifiable as such, or any extremely slowly evolving lines are usually seen to have been progressive, relatively advanced types up to the time when their evolution was arrested. It is their long persistence without much change after their allies have gone on to explore other avenues of modification that gives them an archaic aspect or makes them primitive in comparison with contemporaries of later phases of their history. Lingulids had a fairly high type of invertebrate structure for the Ordovician and limulids were advanced for the Triassic. Sphenodonts were about as advanced as any reptiles in the Triassic and so were crocodiles in the Cretaceous. Lingula, Limulus, Sphenodon, and Crocodylus are primitive as of now just because they are (with little question in these cases) bradytelic and have been for long periods. But such groups have been passed in the race; they did not have an initial handicap. Stebbins (1949) has pointed out that the same generalization is true of plants: the lines with arrested evolution were specialized and advanced when their evolution was arrested.

Another approach has been to designate certain places or broad environmental situations as conducive to slow evolution: oceans, islands, tropical forests, and so on. It is probably correct that sustained, slow, or arrested evolution is more likely to be found in some such environments than others. (Islands happen to be one of the places where it is less likely, contrary to some statements, see Chapter IX.) But the approach to the problem is still obviously inadequate, for very rapid evolution may occur in exactly the same situations (many marine groups, epiphytes in tropical forests, etc.).

It is, in fact, clear once more that the key to this evolutionary problem is not in the organisms or in their environments but in the relationship between the two. When lines persist with arrested evolution, they have maintained adaptation without appreciable change. For this


to occur, there must be a balance between breadth of adaptive zone and its variation or deviation in time. If organisms are very broadly adapted, if they can, for instance, live on a variety of foods or grow on a variety of soils and in a variety of climates, environmental factors may fluctuate greatly without requiring any adaptive change for survival, as long as some part of the broad zone persists. If adaptation is narrow, the zone must be a very uniform and stable one in space and time for evolution in it to be arrested and its occupying populations to persist.

Arrested evolution may thus occur in a great variety of environments and organisms and with quite different sorts of adaptation and specialization. It is, however, more likely to occur with broad adaptation than with narrow, because adaptation in a broad zone is more likely to persist than in a narrow one. Both broad and narrow may persist, however. The lungfish zone would seem to be quite narrow, and yet it has persisted without marked change since the Devonian and all lungfishes have apparently been bradytelic in it for perhaps 200 million years. It may well be that if a major zone ("major" in being quite distinctive from all others) is rather narrow, only bradytelic groups will eventually be left in it. On the other hand, the zone of the opossums is quite wide, and the opossums have survived repeated and radical environmental change without themselves changing much.

Evolutionary change is so nearly the universal rule that a state of motion is, figuratively, normal in evolving populations. The state of rest, as in bradytely, is the exception and it seems that some restraint or force must be required to maintain it. This force undoubtedly exists and is the same as the force that usually orients evolutionary change: selection. Selection on groups with arrested evolution must be centripetal and must be more effective than mutation, cyclic or fluctuating environmental change, or any other factor making for change. Whether adaptation is broad, as it usually is, or narrow, such a group must be perfectly adapted to its zone and almost any change must be quickly detrimental and effectively checked by selection.

The most slowly evolving groups do seem all to be very highly and specifically adapted to a particular zone. Their typical history is one of rather rapid shift into a new, stable and persistent zone, still rapid adjustment (postadaptation) in the zone, with weeding out of less adaptive characters and lineages, and then a long, eventless course of relatively unchanging continuation. That they are actually held by


force, so to speak, is confirmed by the fact that many of them (not all) have given off branches that rapidly evolved into different zones when the opportunity, or perhaps it is better to say the occasion, arose. Thus bradytelic Ostrea (sensu lato) repeatedly gave rise to relatively very rapidly evolving coiled (Gryphaea) and spiral (Exogyra) lineages, among others. Opossums gave rise rapidly and early to a great variety of divergent groups in South America: Caroloameghiniidae, Borhyaenidae, Polydolopidae, Caenolestidae, each family except the first (as far as known) highly varied. Typically these rapid side-branches, or most of them, become extinct while the parent group continues unchanged.

With this background, let us turn back briefly to the characteristics of the populations and environments involved. As was found true of extinction, the balance maintained here, lost in extinction, is compounded of such extremely numerous, varied, and complex factors That it may he quite impossible to designate them explicitly in a specific instance. For the populations, the prime essentials are that these be extremely well integrated genetically and that any deviation be subject to effective counter-selection. It might be expected that such groups would have especially elaborate systems of modifiers or polygenes, but as far as I know no concrete comparisons pertinent to this relationship have been made or are yet possible.

If selection, and apparently quite feeble selection in some cases, is to be effective, gene drift must be excluded. This requires population above a minimal size, but that size need not be very large in all cases. Westoll (1949) has given some reason to believe that early lungfishes, evolving rapidly, had larger populations than the bradytelic living forms, but the population size in the latter is apparently quite large enough to enable selection to hold them in a narrow zone—the fact that the zone is narrow probably means that centripetal selection is relatively strong and so is still effective in relatively small populations. Most groups with arrested evolution have large, even extremely large, breeding populations. Outstanding exceptions seem to be Sphenodon and Latimeria, but Sphenodon apparently had large populations until human disturbance banished it from the mainland of New Zealand and no one has any idea how large a breeding population Latimeria may have.6

6 It has also been remarked (e.g., Carter, 1951) that the monotremes and Peripatus have small populations and so do not agree with requirements for brady-


As Stebbins (1949, 1950) has emphasized for plants, there is no reason to think that large population size in itself promotes slow evolution. It makes for more effective selection, and also within limits for longer duration, and hence is propitious for arrested evolution when selection is centripetal. That selection may tend not toward change but toward stabilization is a thesis developed with skill and at great length by Schmalhausen,7 whose work cannot be summarized here (but see also brief comment in Chapter VI). Others, e.g., Heuts (1949), have also given specific examples of mechanisms developed by selection that inhibit rather than promote change. One additional point that might be mentioned is that intraspecific selection must also be centripetal if evolution is to be arrested while a lineage continues. The group must be so organized as to minimize genetically effective competition within it, or always to favor the modal type, or both. Strong dimorphism in secondary sexual characters would not be expected in such a group among animals and in fact does not occur in any example known to me.

As to the environment, the principal requirement is that it persist and do so with less fluctuation than the maximum that can be met by existing and expressed, not potential, variation in the populations conncerned. How much fluctuation may be so met depends of course on the amount and sort of expressed variation and the details of the adapptation involved but, in general and in terms of the environment itself, the more stringent it is in requirements for life, the less variation can occur without requiring evolutionary change or causing extinction. Arrested evolution is therefore less likely to occur in very difficult environments, such as deserts, impermanent environments, such as salt lakes, or highly variable environments, such as the alpine zone. It is more likely to occur (although of course rapid evolution may also occur) in the easier, more permanent, and less variable environments

tely. It is now clear that a large population is not a requirement but only a probability. What some students would consider a small population in subjective and relative terms may yet be such as to make centripetal selection effective, which is the essential point. As regards the monotremes, they are extremely specialized in many respects. It is improbable that they are bradytelic and their evolution may not even have been unusually slow—they merely happen to have some reptilian characters lost in true mammals. Their breeding populations were also moderate if not large before they felt the effects of man and his works. Peripatus is quite likely to be bradytelic, but I know of no estimates as to its breeding populations; forms so obscure in habitat are likely to appear rare when not so in fact.

7This is the same man occasionally cited as "Shmalgausen," which is an English transliteration of the Russian transliteration of his originally German name.


such as the ocean, its strand, the shifting but long enduring major lowland rivers, the more slowly shifting and also long enduring great forest belts, and particularly in the most nearly permanent of climatic zones, subtropical to tropical.

It now seems to me possible to explain the phenomenon of bradytely in general terms and in the light of these factors of slow or arrested evolution. Any series of related lineages or even one lineage in the course of its history is subjected to a great many environmental changes and to many changes in its own populations. Some of these changes are concomitant with or in a sense produce progressive adaptive changes in the populations. Others shunt development off into another directions, or move it into a different adaptive zone. Still others lead to loss of adaptation, hence extinction. In the case of single lineages, in most instances one of these three results will ensue. Sometimes, however, early change may, in a manner of speaking, have oriented and aimed the group down a line where adaptation is retained indefinitely without further change.

In terms of multiple lineages the process might be visualized as movement of the various sorts of organisms down a series of corridors with many baffles. Most corridors turn, forcing the organisms to turn, too (evolve new adaptation). In a straight corridor the baffles (changes in environment and population) deflect some into side corridors (new adaptive zones) and stop some literally dead (extinct). After a certain time (a long time) all those seriously liable to deflection or stopping have been weeded out. There remain only those sorts of organisms that have run through the whole repertory of baffles and are no longer subject to deflection or stopping. Suddenly at the last baffle that does stop or deflect, there is a discontinuity of outcome. The remaining sorts of organisms now persist indefinitely.

The statistical result actually observed would arise from such a situation. It is seen, too, that one of the survivors, the bradytelic lines, need not be evolving more slowly than some of the others, the horotelic lines, and might have no characters that would reveal its bradytelic nature, before the final event, the disappearance of the last horotelic line of the same age as the bradytelic line. The distinction of the bradytelic line is an adaptive relationship that makes it miss the baffles, that permits indefinite survival without change in the face of all vicissitudes that do eventuate in its adaptive zone. Bradytelic lines are merely the residuum of a process that regularly reduces the percentage


of unchanged groups but that stops short of reduction to zero, if bradytely does occur.

The baffles are so numerous in nature that one would expect bradytely to be in general less common than horotely and to be absent altogether in some groups. That expectation is confirmed by observations, such as the examples discussed earlier in this chapter.


It is my opinion that tachytely is a usual element in the origin of higher categories and that it helps to explain systematic deficiencies of the paleontological record. It will therefore be discussed in some detail in the following chapter, which is devoted to those topics. At this point, the need is only to try to clarify the relationships of tachytely to horotely and bradytely (a subject that I personally find both abstruse and difficult but highly significant and enlightening as regards major evolutionary processes).

It is in the nature of the method that the study of survivorship cannot reveal whether there is also a nonstandard distribution of rates generally faster than horotelic, analogous to the nonstandard distribution of low rates that it does reveal. Evolution at exceptionally high rates cannot long endure. A tachytelic line must soon become horotelic, bradytelic, or extinct. For this and other reasons, which will later be suggested, tachytelic lines, or tachytelic phases in the evolution of lines horotelic or bradytelic at other times, are poorly recorded by fossils and often not recorded at all. There are occasional glimpses in the form of fossils that happen to be found at some point in the tachytelic phase (e.g., the Devonian amphibians from Greenland) and some complete sequences of a rather marginal sort, but for the most part a given tachytelic phase is an inference from ancestral and descendant forms in a record with important gaps. Often the inference is perfectly clear and its correctness is confirmed by the scattered records that do show it directly, but the incidence of tachytely cannot be counted or calculated, even on a sampling basis, as can that of bradytely.

What distinguishes tachytely from horotely is not only that it occurs at exceptionally fast rates but also that it occurs while populations are shifting from one major adaptive zone to another, and especially when a threshold is crossed (see Chapter VI). Such a phenomenon differs from horotely only quantitatively and not sharply. It grades into horotely


both in the sense that a tachytelic line gradually becomes horotelic and in the sense that the differences in degree of adaptive change involved, in rate of evolution reached, and in clarity of threshold are all on continuous scales. There is no sharp point where one can say, "Below this is horotely and above it is tachytely." Nevertheless there is a difference that may be very appreciable and significant in less marginal cases.

Well-recorded but marginal instances occur in such sequences as that of the Equidae (Chapter VIII). The changeover from browsing to grazing has a threshold effect and involves a definite, temporary increase in evolutionary rate. This verges, at least, on tachytely. The even more rapid changes from one stable type of foot mechanism to another are still more clear-cut examples of this sort of phenomenon and might well be designated as rather small-scale tachytely. Another well-recorded example of small-scale tachytely, leading to extinction in this case as tachytely often does, is that of Gryphaea (Chapter IX). One of the most remarkable known examples is that of the snail Valenciennesia, formerly classified in another suborder than the notably different Limnaea, but shown by Gorjanovic-Kramberger (190l, 1923) to have evolved from the latter so rapidly that the whole process occurred while a horse, Hipparion gracile, on adjacent lands showed no appreciable change. Here, too, change to a distinctly different adaptive zone is involved: from clear and fresh to muddy and brackish water.8

As Stebbins (H)50) has suggested, tachytelic lines are likely to be, although they are not necessarily, narrowly adapted. They include the forms that are forced into rapid change by changes in their adaptive zones. Their adaptive evolution tends to be the opposite of that seen in bradytely, and the factors involved tend to be inverse. In the figure of speech utilized above, tachytelic lines hit the baffles and are deflected into sharply divergent corridors.

Tachytely is defined as a phylogenetic phenomenon and is to be distinguished from rapid speciation or splitting. Rapid evolution in the sense of high origination rates, such as occur on invasion of new and open habitats (e.g., in the Galápagos or Hawaiian Islands, as discussed before, or rapid development of endemics in African lakes, see Worthington, 1937, 1940) is a phenomenon quite distinct from tachytely. Nevertheless, when adaptive radiation occurs, and especially if this

8 Basse (1938) has cast doubt on Gorjanovic-Kramberger's interpretation, but Basse's opposing evidence is weak.


FIGURE 44. DIAGRAMMATIC REPRESENTATION OF CASES OF HOROTELY, BRADYTELY, TACHYTELY. This particular sequence, one of many that occur, shows a bradytelic group from which is split off a tachytelic branch which in turn gives rise to a horotelic group.

is of a basic or major sort, the shift of each divergent line into the zone it comes to occupy is commonly tachytelic.

Horotely represents a sort of normal or average turnover or metabolism in the evolution of a group of organisms. To the extent that evolution is defined as change, bradytely is a cessation of evolution without extinction. Tachytely is an episodic acceleration, figuratively a brief fever imposed on the normal metabolism. Tachytelic lines arise from horotelic or bradytelic lines, in the latter case usually (perhaps always) by splitting. A tachytelic line must soon become horotelic, bradytelic, or extinct. Horotelic lines usually arise from other horotelic lines, occasionally from tachytelic lines (but these occasional events are unusually important) or from bradytelic lines, in the latter case by splitting and usually through a tachytelic phase. Horotelic lines may remain horotelic, or become bradytelic or tachytelic. Bradytelic lines develop from horotelic or often rather directly from tachytelic lines. The bradytelic lines themselves remain bradytelic and rarely if ever become horotelic or tachytelic again, but it is rather common for them to give rise to tachytelic branches, which in turn become horotelic or extinct. A simplified pattern of one of several possible relationships between horotely, bradytely, and tachytely is shown in Figure 44.

George G. Simpson, "Horotely, Bradytely, and Tachytely," from The Major Features of Evolution, New York: Columbia University Press, 1953, pp. 313-337.