Paleobiology, 11(1), 1985, pp. 13-26 Rates of evolution Steven M. Stanley Abstract.—For some higher taxa, species can be identified in the fossil record with a high degree of reliability. The great geological durations of species indicate that phyletic evolution is normally so slow that litde change occurs within a lineage during 105 -107 generations. Failure to recognize sibling species in the fossil record has no bearing on this conclusion because they embody virtually no morphological change. Although slowness is the rule, we have no more precise assessment of morphological rates of phyletic evolution for any major taxon. Morphological data that have been assembled to assess rates of phyletic evolution have been meager, unrepresentative, and predominantly reflective of nothing more than body size. Net selection pressures within long segments of phylogeny—even ones that embrace large amounts of evolution—are infinitesimal and seemingly unsustainable against random fluctuations. This suggests that natural selection operates in a highly episodic fashion. Rates of adaptive radiation and extinction at the species level can be estimated for many taxa and, from them, rates of speciation in adaptive radiation. Species selection should universally tend to increase rate of speciation and decrease rate of extinction, yet these rates are positively correlated in the animal world, apparendy because they are linked by common controls: both rate of speciation and rate of extinction seem to increase with level of stereotypical behavior and to decrease with dispersal ability. Only a few "supertaxa" have been able to combine high rates of speciation with moderate rates of extinction. Steven M. Stanley. Department of Earth and Planetary Sciences, The Johns Hopkins University, Baltimore, Maryland 21218 Accepted: January 15, 1985 Introduction Although molecular clocks make a modest contribution, the fossil record is our primary source of information about rates of evolution. The notion that measurement of genotypic rates should be our foremost goal (Schopf et al. 1975) is unsupportable. Tallies of gene differences mean very little because genes interact and vary greatly in their influence over the phenotype. The phenotype must be the primary object of concern in measuring evolutionary rates. In his pioneering work on rates of phenotypic evolution, Simpson (1953) concluded that moderately rapid phyletic evolution is the norm in nature, while very fast evolution (tachytely) and very slow evolution (bradytely) are rare. This conclusion was based on an erroneous technique that yielded a curious left-skewed distribution of rates that suggested a high modal rate (Stanley 1979, pp. 127-129). Today the most widespread debate about rates of evolution focuses on the punctuational model, which holds that very slow evolution is the norm and that rapid spurts of change are rare but of great consequence (Eldredge 1971; Eldredge and Gould 1972). In this paper, I will be heavily concerned © 1985 The Paleontological Society. All rights reserved. with this issue, but with others as well. Unfortunately, limitations of length preclude discussions of many important topics and contributions, including Williamson's study of a fossil record of speciation in freshwater Mollusca, a study that may provide the first paleontological documentation of rapidly divergent geographic speciation (Williamson 1981). What I have chosen to do in the space allotted is to review the status of one general approach, some aspects of which I have developed elsewhere (Stanley 1979)—the assessment and interpretation of rates of evolution at the level of the species, rather than at the level of higher taxa, as was exemplified especially by the major works of Simpson (1944, 1953). Rates of Phyletic Evolution In evaluating our current knowledge about phyletic rates, I will reach one conclusion that I have presented before and a second that I have not: (1) phyletic rates are, in general, much slower than most evolutionists believed a decade or two ago—so slow that we must look to spurts of rapid evolution to account for the bulk of largescale transformation, and (2) although we know 0094-8373/85/1101-0002/$1.00 available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, 14 STEVEN M. STANLEY a great deal about how rates of large-scale evolution differ among higher animal taxa, we know next to nothing about how rates of phyletic evolution, which contribute in only a minor way to large-scale evolution, differ among these same higher taxa. Ideally, it is desirable to evaluate phyletic rates with comprehensive morphometric analyses. To date, little progress has been made in this area, so that before addressing this subject I will focus on taxonomic rates, which are imperfect in some ways but nonetheless revealing. Taxonomic measurement Although fossil data yield less taxonomic information than Recent data, the persistence of a lineage for a known interval without undergoing enough evolution to be regarded as a new chronospecies provides a useful indication of the rate of phyletic evolution within the lineage. Survival for a long interval without pseudoextinction represents very slow phyletic evolution. Here the perspective must be the time required for major changes to occur within the taxon to which the lineage belongs—the typical interval for a new genus or family to evolve, for example. Thus, the faa that a species of mammals has typically survived 1—2 ma, whereas many distinctive mammalian genera and families have evolved during intervals of less than 4 or 5 ma, represents strong evidence that normal phyletic evolution is much too sluggish to be the dominant factor in the origin of higher taxa (Stanley 1982). Of special importance here is the faa that for some higher taxa it is possible to estimate durations for a large sample of species, yielding a comprehensive picture of the limits of phyletic evolution. The quality of fossil data.—At issue in the taxonomic measurement of rates of phyletic evolution is whether paleontological species are meaningful entities, a status some workers question. If durations are employed to establish an upper limit for phyletic change, then the failure of paleontologists to distinguish sibling species (Levinton and Simon 1980; Schopf 1982; Levinton 1983) is not a problem. For example, it makes no difference whether populations that collectively illustrate morphological stability over a long geological interval constitute a single species or a set of sibling species; whatever their specific status, they document approximate stasis. In faa, if these populations constitute two or more at least partly contemporaneous sibling species, they provide broader evidence of stasis than would be offered by a single lineage. Also, for many higher taxa there is no question that the fossil record offers reliable data for species level taxonomy and, as will be summarized below, for species durations. Beetle species, for example, are diagnosed in the Recent by their genitalia, which are readily preserved in anoxic sediments; dental morphology serves to distinguish nearly all living species in most modern families of mammals; and seed morphology, also faithfully preserved in some sediments, is reliable for species identification of higher plants in the Recent. In the investigation of species durations, many taxa offer the advantages of having excellent fossil records in certain regions and of being closely related to taxonomically well understood living groups. I have found Lyellian percentages for Cenozoic species to be particularly useful in the estimation of mean species duration for higher taxa. A Lyellian percentage for a fossil biota is the percentage of species within that biota that survive to the present. Bivalves as an example.—An entire curve of Lyellian percentages such as the one depicted in Fig. 1 for bivalve mollusks permits us to estimate mean species longevity. Mean longevity is approximated by the absolute value of the slope of the curve at zero—assuming a low speciation rate and the maintenance of an approximately stable age distribution (Gillespie and Ricklefs 1979). Mean species longevity can also be estimated by doubling the age of the biota whose Lyellian percentage is 50% (Stanley 1979). These two methods applied to Fig. 1 yield estimates of mean durations for bivalves of approximately 11 ma and 14 ma, respectively. Recently Levinton (1983) criticized the "50% method" of estimation, citing Kurten (1959) to the effea that a fauna representing any time plane declines by extinaion of species in an exponential fashion; this would imply that mean longevity is not about 2 but 2.9 times as long as the interval of decline to 50% (in this case, the half-life). In faa, there is no basis for a model of exponential decay, or linear decay on a log scale. It is incompatible with Van Valen's law, in which a true survivorship curve for subavailable at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, RATES OF EVOLUTION 15 NO. OF SPECIES CALIF. JAPAN 100 80 60 40 20 r r •* • A r * * 1 1 [ 1 L 1 _ J 1 • A • 1_ 12- 26- 51 - 76- 101- 126- - 25 • - 50 • - 7 5 # -loo 0 -125 0 -150 A >I50 A * * -.A < 1 1 A 1 i A 6 8 10 12 14 16 FAUNAL AGE (MILLION YEARS) 18 20 PLEIST PLIOC. U. MIOC. M. MIOC. L. MIOC. FIGURE 1. Lyellian curve for the Bivalvia, produced from data representing faunas of California and Japan (see Stanley et al. [1980] for details). The dashed diagonal line provides an estimate of mean species duration by the methodology of Gillespie and Ricklefs (1979). This line is approximately tangential to the curve at zero ago, and its slope (calculated after converting the ordinate scale from percentage to frequency) represents an estimate of mean species duration—about 11 ma. Doubling the age of faunas that contain 50% living species yields a slightly larger estimate of 14 ma. taxa within a higher taxon is purported to be log linear; here, by definition, the taxa are artificially arrayed so as to start their lives when decay begins, whereas for any slice of geologic time, all existing species already have a history. Van Valen's law does not seem to be strictly valid (Raup 1975; Sepkoski 1975; Hoffman and Kitchell 1984). Nonetheless, many survivorship curves are crudely log linear, and this implies that the decline of the species that constitute a Lyellian fauna follows quite a different pattern. Lyellian methods circumvent many of the deficiencies inherent in fossil data and are therefore generally to be preferred over direa estimation of species durations from fossil occurrences. The fundamental Lyellian assumption is simply that a fossil fauna that is employed represents a good statistical sample of the species present at the time when the fauna lived. Presumably there is some bias against species that are rare and therefore short-lived. Use of the Recent as an end point, however, offers a major advantage over similar techniques applied to extinct taxa (Kurten 1959) in that we have a relatively complete roster of living species for many taxa, which greatly enhances accuracy. Imperfect though it may be, estimation of mean species duration for extinct bivalves directly from fossil data also yields very long estimates. Data from the Devonian of New York State (Stanley 1979; based on McAlester 1963) and the Jurassic of Europe (Hallam 1976) give mean durations of about 10 ma and 20 ma, respectively. In criticizing Hallam's estimates, Schopf (1982) erroneously assumed that they were crudely calculated using stages as units. In fact, Hallam's units were zones, which provide for excellent resolution (about 1 ma). Estimates of mean duration of less than 2 ma (Koch 1980) or 3 ma (Kauffman 1978) for Cretaceous bivalves species from the Western Interior of North America are almost certainly based on incomplete ranges. The record here is so poor that Koch excluded 45 species from his study beavailable at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, 16 STEVEN M. STANLEY TABLE 1. Estimates of mean approach. species durations for a variety of biological groups, based partly or wholly on the Lyellian Biological group Estimated mean species duration (ma) Source of estimate ine bivalves ine gastropods thic foraminifers 11-14 10 13.5 20-30 >20 Planktonic foraminifers >20 Freshwater fishes Beetles Snakes Mammals Higher plants Bryophytes Marine diatoms 3 >2 >2 >1 - 2 >8 >20 » 2 0 25 (See Fig. 1) Lyellian % (Stanley 1979) Direct range estimation, including some extant species (Martinell and Hoffman 1983) Lyellian % (Stanley 1979) Age estimation (partial ranges) for extant species (Buzas and Culver, 1984). Lyellian % (Stanley 1979, after Berggren 1969); the estimate of ~ 12 ma by Levinton and Ginzburg (1984), based on a misunderstanding of the Lyellian technique, is erroneously low. The citation by Levinton of the 7.7 ma figure from Emiliani (1982) is inappropriate (figure derived from data in Stainforth et al. [1975], who included only "selected index species"). Estimates of ~5-6.5 ma (Wei and Kennett 1983; Levinton and Ginzburg 1984) are based only on extinct species; in excluding extant species, which display much greater mean longevities, these estimates are much too low. Lyellian % (Stanley 1979) All known Pleistocene species survive today (Stanley 1982; based on Coope 1970). Nearly all known Pleistocene species survive to the present (Stanley 1982; based on AufFenberg 1963) Lyellian % (Stanley 1979) Direct range estimation (Stanley 1982, based on Schankler 1980, with emendations by R. T. Bakker). Lyellian % for seed floras, diminished somewhat by Pleistocene extinction (Stanley 1982, based on data from Leopold 1967). Age estimation (partial ranges) for extant western N. American species (Stebbins 1982). Nearly all fossil species <20 ma="" old="" are="" extant="" stanley="" 1982="" after="" dickson="" 1973="" lyellian="" floras="" 13="" include="" 50="" species="" based="" on="" andrews="" 1976="" cause="" they="" were="" represented="" by="" fewer="" than="" 5="" specimens="" and="" of="" the="" 41="" included="" in="" his="" study="" only="" 2="" more="" 100="" other="" taxa="" table="" 1="" summarizes="" estimates="" mean="" durations="" for="" a="" variety="" partly="" or="" entirely="" data="" some="" stratigraphically="" useful="" fossil="" such="" high="" quality="" that="" can="" be="" estimated="" reasonably="" well="" direaly="" from="" stratigraphic="" occurrences="" examples="" ammonites="" trilobites="" graptolites="" there="" is="" no="" question="" each="" these="" groups="" duration="" much="" shorter="" bivalves="" gastropods="" foraminifers="" specific="" presented="" as="" indicated="" higher="" listed="" tables="" not="" necessarily="" homogeneous="" with="" respect="" to="" among="" unknown="" reasons="" long-ranging="" families="" characterized="" slender="" thick-walled="" siphuncles="" ward="" signor="" 1984="" conclusion="" widespread="" longer="" lived="" narrowly="" distributed="" hansen="" 1978="" martinell="" hoffman="" 1983="" logical="" but="" being="" ranges="" may="" exaggerated:="" poor="" records="" including="" rare="" life="" have="" geological="" narrow="" both="" horizontally="" vertically="" artifacts="" preservation="" suspension-feeding="" bivalve="" molavailable="" at="" https:="" www="" cambridge="" org="" core="" terms="" doi="" 10="" 1017="" s0094837300011362="" downloaded="" university="" sydney="" library="" 03="" jun="" 2018="" 08:00:21="" subject="" use="" rates="" evolution="" 17="" lusks="" percentages="" reveal="" nonsiphonate="" infaunal="" pectinids="" scallops="" markedly="" siphonate="" likely="" explanation="" because="" their="" great="" vulnerability="" predation="" former="" group="" small="" populations="" inherently="" susceptible="" extinction="" longevity="" indicates="" sluggish="" individual="" significantly="" plots="" histograms="" mode="" lies="" right="" jf-axis="" along="" it="" this="" pattern="" example="" mammal="" once="" established="" survive="" least="" v3="" planktonic="" 2-3="" 1979="" p="" 175="" nonetheless="" field="" evidence="" plants="" lewis="" 1966="" suggests="" incipient="" constantly="" generated="" nature="" most="" die="" out="" quickly="" scale="" time="" i="" referred="" aborted="" described="" speciation="" boom="" bust="" phenomenon="" very="" population="" size="" local="" occurrence="" must="" form="" discrete="" -axis="" histogram="" passes="" beyond="" presumably="" expanding="" its="" geographic="" range="" process="" long="" stretch="" regard="" phyletic="" also="" important="" note="" many="" disappear="" true="" rather="" pseudoextinction="" thus="" indicate="" wide="" an="" average="" will="" undergo="" little="" measurable="" change="" during="" 105="" 107="" generations="" so="" oldest="" youngest="" regarded="" competent="" taxonomist="" constituting="" same="" two="" similar="" suggestion="" localized="" evolve="" rapidly="" abundant="" mayr="" 1963="" 522="" has="" been="" cited="" loophole="" argument="" schopf="" claim="" concentrated="" too="" recognized="" traced="" heavily="" studied="" extinct="" invertebrate="" direct="" observations="" biological="" source="" estimate="" inthe="" 1-2="" creased="" values="" 6="" -="" 15="" five="" unusual="" data-mce-fragment="1"> 1 Stanley (1979), based on Palmer (1965). Graptolites 2-3 Rickards (1977); Stanley (1979). the fossil record. One strong argument against this idea is the pattern of occurrence of many living fossil taxa—those that have survived (by definition, with little evolution) in confined geographic areas at small population sizes. Especially striking here are the various living fossil taxa endemic to local deep-sea hydrothermal vents (Newman, 1985)—habitats that have certainly always been small in areal extent. If rapid evolution is the norm for rare species, it is difficult to explain how, in a small, persistent habitat like the deep-sea vent setting, very slow net change has been the rule. On the other hand, this condition conforms to the so-called test of living fossils (Stanley 1975, 1979, pp. 122- 131), which favors the punctuational model in suggesting that speciation is necessary for substantial evolutionary change: Every extant clade that in the fossil record can be seen to be long and narrow (i.e., to exhibit little branching) is represented today by living fossils. The unorthodox dismissal of living fossils as figments of scientists* imagination (Schopf 1984) is unfounded, overlooking such facts as the recognition of the insect Fannia scalaris in the Early Oligocene Baltic Amber. The identification of this 40-maold species was made by Willi Hennig (1966), of cladistic fame. As the foremost taxonomic expert on muscid flies, to which Fannia belongs, he based his conclusion on the characteristic slope of well-preserved tiny hairs. More generally, the reality of living fossils is supported by the deepsea vent faunas cited above (Newman 1985). If living fossils are artifacts of our taxonomic perceptions, what coincidence could have led several different taxonomists to be fooled by so many available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, 18 STEVEN M. STANLEY of the animals of one particular setting? Living fossils live. Morphologic measurement Morphometric evaluation of phyletic rates has great potential, yet meaningful data are at present so sparse that they yield few significant conclusions. In offering a rather severe critique of what has been done, my intention will be to suggest improved methodologies for future work. One of the chief biases of efforts to measure morphologic rates has been a misplaced emphasis on body size (or surrogate parameters for body size) as an index of overall evolutionary rates. More generally, problems have been introduced through attempts to answer general questions by reliance on single charaaers, data for single localities or regions, or data for just one or a few lineages. How important is phyletic evolution? The overuse of body size.—Phyletic evolution occurs universally. Every species changes in at least minor ways as generation follows generation. The real issue of punctuationalism is the importance of phyletic evolution: Does it account for most macroevolutionary change or play a subordinate role? Some early respondents to the punctuationalist challenge assumed that the punctuational view denied phyletic evolution altogether. Thus, Gingerich (1974), having purported to document phyletic increase in body size for the mammal genus Hyopsodus, claimed this finding as a refutation of the punctuational model. In fact, the reality of phyletic size increase for Hyopsodus is challenged by newer data (West 1979), but, even more fundamentally, we must ask whether an actual increase in body size of modest degree would refute punctuationalism. Of relevance here is the fact that no branch in the phylogeny of Hyopsodus reconstructed by Gingerich moved beyond the confines of the original genus in the course of about 3.5 ma. In fact, within the Hyopsodus clade there was apparently little morphologic evolution other than size change, which may have been punctuational: "As currently defined (Gazin 1968) species of Hyopsodus are virtually indistinguishable from one another except for size differences" (West 1979). Inasmuch as hundreds of new genera in many families of mammals were evolving during the persistence of the Hyopsodus clade, we can conclude that the sum total of evolution within Hyopsodus was trivial in the context of macroevolution. The same problem is evident in the claim that existing data leave open the possibility of statistically significant evolution within the Homo erectus lineage (Levinton [1983] in response to Rightmire [1981]). Indeed, statistically significant change may have occurred, but was such change important in human evolution? How many measurable shape parameters for H. erectus yield evolutionary trajectories that if projected forward in time would produce the modem human condition 40,000—100,000 yr ago? Evidently, virtually none. Walker (1984) concluded that phyletic change in H. erectus is largely a matter of size, finding that when sizes are standardized, the sagittal and coronal outlines of early H. erectus crania fall entirely within the envelope of shapes represented by late crania of the species—ones about 1 ma younger (Fig. 2). Thus, although metrics that primarily reflect size may indeed reveal long-term change in H. erectus, cranial shape was remarkably stable, showing no significant shift toward the modern condition during more than 1 ma. Femur shape in this species was also quite stable during an interval of about 1.5 ma, according to the morphometric analysis of Kennedy (1983). Walker concluded: "H. erectus evolved relatively quickly and then survived relatively unchanged for over a million years ... . anatomically modern humans appeared with great suddenness about 40,000 years ago, in much the same rapid way as H. erectus" The studies of Hyopsodus and H. erectus, revealing little more than change of body size, illustrate a widespread bias in the measurement of rates—a favoring of parameters that are surrogates for body size. This is puzzling in light of the evidence that body size is unusually labile in evolution—evidence provided, for example, by Bergman's rule of latitudinal size variations within species and by the documentation of geologically very sudden size change in Late Pleistocene mammals (Marshall and Corruccini 1978; Davis 1981). In attempting to characterize rates of evolution for vertebrates and for invertebrates, Van Valen (1974) and Gingerich (1983) adhered to the traditional overemphasis on body size. Van Valen's compilation included 75 measured rates available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, RATES OF EVOLUTION 19 for invertebrates, of which 65 were for size parameters. Gingerich's compilation included 135 measured rates for invertebrates—those tabulated by Van Valen plus 60 taken from Hallam (1975), all of which represented body size. The bias of this overwhelming reliance on size parameters is evident in the incorporation of Hallam^ 41 estimates for Jurassic bivalves; in a second paper, Hallam (1978) observed that, although size does change significantly in Jurassic bivalve species, little else does: "The results of my analysis of 329 European Jurassic species provide, with an important exception, {Gryphaed] overwhelming support for the punctuated equilibrium model. Species whose morphology appears to persist unchanged for long periods are abruptly terminated . . . the exception .. . is phyletic size increase." Biases in the gathering of data and estimation of rates.—As illustrated in the previous section, compilers of phyletic rates have employed highly unrepresentative samples. The greatest source of bias here is the near absence of published data for static or near-static characters. Nearly half of Gingerich's invertebrate data represent bivalve mollusks. Having worked extensively with this group of animals, I could easily select hundreds of species that would collectively yield thousands of empirical millidarwin values close to zero for characters that have been nearly static for several million years, swamping the size-dominated data employed by Gingerich (1983) and drastically lowering his "mean" rate for invertebrates in general. Gould's critique of Gingerich's comparison of rates provides additional evidence of the traditional bias in data gathering. Gingerich (1983) plotted evolutionary rate, in darwins, against time. The rate in darwins is (In x2 — In xx)/ time interval (ma), where xu and x2 are the initial and final character states. Gingerich compiled rates from selection experiments, colonization events, and the fossil records of invertebrates and vertebrates. Plotting these against time on one graph, he found a negative correlation and concluded that differences in rates result from different temporal scales (an average rate for vertebrates was calculated for a span of 1.6 ma compared to a span of 7.9 ma for invertebrates). Adjusting for this alleged bias, Gingerich reached the unorthodox conclusion that invertebrates have no Australopithecus africanus Homo erectus no Homo sapiens Neanderthal FIGURE 2. Sagittal and coronal shape plots, showing great stability for hominid species (from Walker 1984). Skulls for each species were standardized to one size (calvarial length was standardized for sagittal shape and distance from the vertex of the profile to the tip of the mastoid process, for coronal shape). The envelope for each species depicts maximum and minimum values for all shapes at equidistant intervals along an axis. For H. erectus, the outer shape envelopes represent late (Olduvai, Java, and China, but not Solo) individuals; and the inner, black shape envelope, which falls within the outer one, represents two early (Koobi Fora) specimens which are more than 1 ma old. Interestingly, according to this analysis, Neanderthal resembles H. erectus more closely than modern H. sapiens, with which it overlapped in time. actually evolved more rapidly on the average than vertebrates. He also noted that the ratio of initial to final values of character states for all groups averaged about 1.2. This, as Gould observes, is a key point: Given a near constant of 1.2 as the numerator in the calculation of darwins, the plot of darwins versus measurement interval automatically has a negative slope because it is litde more than a plot of / versus 1 / /: the correlation is an artifact. Thus, there is no justification for rescaling rates or concluding that phyletic evolution is more rapid for invertebrates than for vertebrates. Gould notes still another bias. The quasi-constant value of 1.2 for the available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, 20 STEVEN M. STANLEY ratio x2/xx in Gingerich's compilation must reflea human choice: smaller differences between ancestral and descendant populations seem uninteresting, and larger differences discourage comparison by engendering doubts about ancestral-descendant relationships. In the compilations of Van Valen (1974) and Gingerich (1983) for rates of invertebrate evolution, measured in darwins, all the parameters that do not represent size are meristic in nature. Implicit in this treatment is the assumption that meristic traits are properly viewed as increasing (or decreasing) in number geometrically. For ribs that do not bifurcate ontogenetically and for most other meristic traits, however, it would seem that numerical change is fundamentally arithmetic in nature. This judgment is especially reasonable for meristic traits in which the range of variation within every species is normally nil or one unit, regardless of how many units are present. An v observation that the range of variation increases with mean number of ribs might be taken as justification for geometric treatment, but the real answer may lie in morphogenetic controls of rib number, about which we are totally ignorant. Paradoxically, even if (without a change in shell size) rib number in a scallop or similar animal is found to increase linearly with time by arithmetic measurement, rib width will decrease nonlinearly with the reciprocal of rib number. Thus, even different modes of measurement of a single evolutionary change can yield different apparent histories for rates of evolution. Still another problem has been that the large effort required for morphometric analyses has discouraged comprehensive evaluation of entire clades or higher taxa. Malmgren and Kennett (1980) conducted an exemplary multicharacter study of phyletic evolution since early Late Miocene time in a single lineage of planktonic foraminifers, but their results cannot be taken to typify the Globigerinacea. About half of all early Late Miocene species of this group have survived to the Recent with little change (Berggren 1969; Stanley 1979, p. 133). An outstanding study that places phyletic evoution in phylogenetic perspective is the monograph of Hayami (1984) for the genus Cryptopecten. Hayami documented the appearance of discontinuous morphologic variation in one lineage about 0.5 ma ago and an increase in the frequency of the new morphotype to about 0.4 in the Recent. He nonetheless found no significant evolutionary change for many other morphometric characters of the lineage or for other Cryptopecten lineages in general. His conclusion was that "morphological change within each lineage of Cryptopecten is, in every way, much less significant in comparison with the morphological difference between distinct lineages." Thus, he invoked rapidly divergent speciation to account for distinctness of the lineages and reconstructed a predominantly punctuational pattern of evolution for the genus. I see no basis at present for comparing phyletic rates among higher taxa. We do know that phyletic evolution in many invertebrate taxa is typically too slow to cause pseudoextinction in 1 or 2 ma, but much of the mammalian extinction yielding species durations of this length is clearly true extinction. Thus, phyletic evolution could be as slow in the Mammalia as in invertebrate taxa. There is no panacea for all the problems described above, but there would seem to be two important areas for improvement. First, it is important to attempt to assess morphology comprehensively and objectively. This calls for the employment of multivariate statistics, or at least for the indpendent assessment of many traits per lineage. Second, if phyletic rates are to be compared for higher taxa or other large groups of species, then each group must be characterized by a large, representative set of lineages. In a general way, both of these conditions are met by the previously outlined taxonomic methods, including the Lyellian techniques, for comparing phyletic rates among higher taxa: most species are distinguished by many traits, and large samples of species are employed in comparisons. Unfortunately, it will take many years for more quantitative multivariate morphometric analyses to yield data of comparable scope. Stratigraphic and geographic control.—Bookstein et al. (1977) evaluate the uncertainties that plague attempts to distinguish phyletic evolution from punctuational change when we are faced with incomplete fossil sequences of the sort recorded by Gingerich (1974, 1978) for Eocene mammals. If the pattern happens to be punctuational, there is sometimes an easy way out of such predicaments. New data can reveal an uncertain evolutionary pattern or an alleged phyavailable at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, RATES OF EVOLUTION 21 letic transformation to be a more complex segment of phylogeny in which two temporally stable species are separated by a speciation event. All that is needed is the discovery that two species thought possibly to intergrade actually overlap in time. As an example, the alleged evolution of the Eocene primate Tetonius into Pseudotetonius, though interpreted as phyletic (Rose and Bown 1984), cannot be, because teeth virtually identical to those of the ancestral form have been found at a slightly different geographic location in sediments about 1.5 ma younger than the alleged interval of transition (R. T. Bakker, pers. comm. 1984). Reconstructions of phyletic change based on data from single stratigraphic sections or a few local sections are open to question even when control in the vertical dimension is excellent. Raup and Crick (1981) reanalyzed the voluminous biometric data that Brinkmann (1929) gathered through centimeter-by-centimeter collecting of the Jurassic ammonite Kosmoceras sensu lato. They obtained equivocal results as to whether zigzag trends represent significant evolution. Brinkmann's data were from a single British brick pit, however, raising a question whether the patterns may represent nonevolutionary biogeographic oscillations in the populations of a variable species (and some ammonite species are highly variable [Reeside and Cobban I960]). Evolutionary reversals.—Raup and Crick's (1981) analysis of Brinkmann's Kosmoceras data elucidates the biases governed by the vertical interval of sampling employed in the assessment of phyletic evolution. They found that the outcome of runs tests and a statistical assessment of random fluctuations were strongly dependent on vertical sampling intervals. The data of this study reveal how an insignificant temporal fluctuation that is quickly reversed would seem significant if considered in the absence of data from higher in the section. On a broader scale, there is another kind of problem. In the section studied, Raup and Crick found the mean diameter of Zugokosmoceras to be significantly larger for the highest 20 specimens than for the lowest 20 specimens. In the absence of intermediate samples, or in the presence of one or two, this information might be employed to calculate a rate of phyletic evolution under the assumption that it resulted from reasonably consistent selection pressure. Given intermediate populations, however, Raup and Crick found a fluctuating pattern of change. What is more, for a 100-cm sampling interval, the possibility that the actual pattern could result from a random walk could not be ruled out at the 99% level. The paradox of infinitessimal net selection pressures.—In a study of large-scale phenotypic change in Cenozoic mammals, including horses and creodonts, Lande (1976) found net rates to be so slow that if produced by constant selection pressures they would have been brought about by only about one selective death per million individuals per generation. Lande's conclusion was that such selection pressures are so weak that random drift cannot a priori be excluded as the primary agent of change. My view (Stanley 1979, pp. 56-57), seconded by Stebbins (1982), is (1) that the simple faa that the net changes were substantial and seemingly adaptive rules out genetic drift and (2) that selection pressures as small as the net ones calculated by Lande are too small to be maintained against stochastic fluctuations. The implication is that selection has operated in spurts, either (1) in rapidly divergent speciation events or (2) within single species, which have thereby evolved in a stepwise or "staircase" pattern. This leads to my next topic. Staircase evolution?—If we find that species often change rather little for long stretches of geologic time and conclude that evolution does much or most of its work in spurts, the question remains whether these spurts are associated with multiplicative speciation (the original punctuational model) or transform species in toto to yield a pattern of staircase evolution (a modified punctuational model). The most definitive evidence here is documentation of temporal overlap between distinctive lineages; then if they are connected, it must be via a branching event. The problem of resolution arises if a speciation (branching) event is immediately followed by the competitive extinction of the ancestral species by the daughter species. (This is by no means the rule, even under sympatric conditions, first because many speciation events entail ecological divergence, and second because interspecific competition is characteristically weak within some higher taxa.) Rapid competitive exclusion may available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, 22 STEVEN M. STANLEY have been the rule within the genus Homo because of great niche breadth and conscious efforts of one species to exterminate another, but coexistence of Neanderthal and modern Homo sapiens (Leveque and Vandermeersch 1981) is proof that the latter did not evolve by wholesale phyletic transformation of the former. Thus far, evidence of staircase evolution has been published for only a few lineages, such as the planktonic radiolarian Eudocubus vema (Kellogg 1975). One might expect this pattern to occur more frequently within planktonic marine taxa than in other ecological groups because of their relatively unrestricted gene flow. A high incidence of staircase evolution, if documented for life in general, would have implications with regard to the causation of stasis. I have suggested that one explanation for stasis may be that gene flow in many species is too restricted geographically to permit the panmictic spread of new genetic elements but strong enough locally to prevent most small populations, each exposed to unique selection pressures, from diverging (Stanley 1979, pp. 48-51). If much evolution takes place by the sudden restructuring of species without a population bottleneck, then panmictic gene flow must generally be too strong for my hypothesis to be valid. The implication would be that morphogenetic constraints play a dominant role in stabilizing species (Lerner 1954; Mayr 1963; Eldredge and Gould 1972; Alberch 1980; Stanley 1982). Rates of Speciation and Extinction The species is the fundamental unit of phylogenetic branching and extinction. In this section I will examine a few issues that relate to the use of species as units in the measurement of what Simpson (1944, 1953) termed taxonomic frequency rates. Kates of adaptive radiation Rate of extinction is the reciprocal of mean species duration (Stanley 1975), estimates of which were listed in Tables 1 and 2. For a monophyletic adaptive radiation that follows an approximately exponential course, the fractional rate of inaease (/?) is approximately In N/t, where N is number of species after time /. The fractional rate of increase during adaptive radiation equals the speciation rate (5) minus the extinction rate (£), so that the former can be estimated from the relationship S = R + E. Actually a calculated extinction rate includes pseudoextinction, but various incidences of pseudoextinction can be plugged into the calculation to establish boundary conditions. When this is done, there is no question that there are enormous differences among taxa; it could hardly be otherwise, given the great disparities in the various rates of R among higher taxa: the R value that typifies radiating marine bivalve families yields only 3 or 4 species in 20 ma, whereas the value that typifies mammalian families yields roughly 80 species during such an interval (Stanley 1979, p. 108). In applying this approach to various taxa (Stanley 1979, ch. 9), I have by no means meant to imply that adaptive radiation is perfectly exponential and have, in fact, noted that it should generally follow a modified exponential (sigmoid) curve. I have simply suggested that the early stages of adaptive radiation should not depart radically from exponential paths, and I have then employed estimates of R for rapidly radiating taxa to estimate S. The results are quite striking in that different taxa display quite different maximum rates of increase (values of R). This point is illustrated in Fig. 3, which shows that the net rate of increase for many families and subfamilies of the Insecta has greatly outstripped the most rapidly radiating modem clades of marine Mollusca. A key point, overlooked by Wilson (1983), who derogated the approach, is that one must employ only the most rapidly radiated subgroups to estimate R for a higher taxon. Numerous additional data points could be added to those plotted for the Mollusca in Fig. 3, to yield a triangular distribution, but these would represent taxa that expanded slowly (we all know of such groups) or that expanded rapidly and then slowed down. In particular, Wilson argued that my use of 7 taxa to characterize maximum rates of adaptive radiation in the Insecta was improper because numerous other families have undergone much slower net rates of expansion. I have plotted his data for such families with my data in Fig. 3. I originally estimated a mean R value of 0.19 ma- 1 for the 7 rapidly radiating taxa (Stanley 1979, p. 256). As illustrated in Fig. 3, this value does indeed approximate rampant rates of radiation (the available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, RATES OF EVOLUTION 23 100,000 0,0001- I OOOtN lOOfe50 100 AGE OF TAXON (t) FIGURE 3. Evidence of markedly different rates of rapid adaptive radiation for extant taxa of insects (circles) and marine bivalve and gastropod mollusks (x's). For each taxon, logarithm of number of species (N) is plotted against age (/) [ma]). The molluscan taxa (genera and families) were chosen for study (without precise knowledge of values of N or /) as seeming to constitute relatively young clades currently expanding rapidly; the faa that the points representing them form a roughly linear array indicates (1) that they are indeed in the midst of approximately exponential expansion and (2) that their mean rate of increase (r = 0.069 ma-1) represents the typical rate of rampant radiation for these groups (for details, see Stanley and Newman 1980). Data for clades (families and subfamilies) of insects are from Stanley (1979) (solid circles) and Wilson (1983) (hollow circles). Points representing different data for the same families are identified by letters (A = Asilidae, B = Bombyliidae, S = Syrphidae, T = Tipulidae). The data from Stanley represent taxa recommended by F. M. Carpenter as being in the midst of rapid radiation. The data from Wilson (1983) represent a much wider assortment of extant families. The faa that the most rapidly radiating taxa in Wilson's compilation are radiating at approximately the mean rate for those compiled by Stanley (1979) indicates that this rate (r = 0.19 ma-1) is a good estimate for rampant radiation of an insea family or subfamily. Insea clades that have undergone less rapid net expansion either (1) radiated slowly from the start or (2) radiated rapidly at first and then slowed their rate of expansion, perhaps even losing diversity. Not surprisingly, all insea clades older than 50 ma seem to be of the second type. Clearly, radiation is much more rapid in the Inseaa than in the marine Bivalvia or Gastropoda. charaaeristic R value) for the Inseaa. Thus, Wilson's data support my choice of taxa to charaaerize R for the Inseaa. Furthermore, the important heuristic point is that the general technique reveals enormous differences among higher taxa in rates of adaptive radiation. For example, no sizable extant dade of marine Bivalvia or Gastropoda has radiated as rapidly as the rates for numerous huge clades of Cenozoic inseas. One of the interesting outcomes of this analysis is that genera and families of mollusks radiate at approximately the same rate. In contrast, many extant mammalian genera of recent origin have radiated at extraordinarily high rates. The implication is that the latter represent bursts of radiation with gaps in between. This pattern may reflea the importance of interspecific competition in the Mammalia (Stanley 1979, pp. 287— 289). It would seem that periodic adaptive breakthroughs yield new genera, which then radiate rapidly within the available geographic area until competition interferes. Within most families of Mollusca, in contrast, genera are not usually ecologically disaete units and competition does not brake their adaptive radiations so suddenly. Weaker competition in the Bivalvia than in the Mammalia seems to be indicated by the faa that bivalve families have continued to grow available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, 24 STEVEN M. STANLEY 120 11 0: 100: 90: 80: 70: 60; 50: 40: 30: 20: 10- 0 : 'NUMBER OF MAMMALIAN FAMILIES PRESENT r Excludin g bots and odontocete plus mysticet e whales I ' M ' '—r—i—r-|— i 1— | ■ . | ^| | Late Paleoc. Eoc. Olig. Mio. Plio. 2R«c . 8 Cret. • Sub-R«c. TIME 50MY FIGURE 4. Histories of family diversity for the Mammalia and Bivalvia. Diversification in the Mammalia leveled off" permanently early in Oligocene time, after only about 30 ma of adaptive radiation. Diversification in the Bivalvia has continued to the present. Maximum body size for the terrestrial Mammalia was also attained about 30 ma after the adaptive radiation began (Indrichotherium of Oligocene age being the largest known land mammal), whereas gigantic bivalves did not evolve until Cretaceous time, some 300 ma after the bivalve radiation began. (After Stanley 1973.) in number since early in the Paleozoic, whereas number of mammalian families leveled off during the Oligocene (Fig. 4). I am doubtful that we can ever develop more particularized models of adaptive radiation than the exponential approximation. The logistic curve was selected by population biologists for its simplicity, but it is only one of many models that depict the damping of population growth by crowding. Furthermore, nature is not a petri dish. For some radiating groups, a contraction of habitats or the expansion of predatory taxa may slow diversification as much or more than ecological crowding. Species selection There can be no question that to some extent differential rates of speciation and extinction drive large-scale evolutionary trends. If evolution is concentrated heavily in speciation events, selection at this level is especially important because phyletic trends are relegated to a minor role. In any case, the accumulation of long-lived species will progressively change the age distribution of species within a clade undergoing species selection (Slatkin 1981). Contrary to the reading of my work by Gilinsky (1981) and Gould (1982), in general discussion of species selection I have always given equal attention to rates of speciation and rates of extinction. The striking fact is that, whereas species selection, by definition, tends to maximize S and minimize E, the two tend to be correlated in the animal world, perhaps because certain biological traits serve as important controls for both (Stanley 1979, pp. 231-235, 258- 268). For example, it would seem that rates of speciation and extinction increase with level of stereotypical behavior (which is an isolating mechanism and also a parameter of niche breadth) and decrease with dispersal ability (a force that opposes isolation but promotes survival). There might still be a tendency for a taxon to benefit or suffer by being locked into high or low rates of S and E. If the punctuational model is valid, an important benefit of having a high turnover rate is that it yields a high rate of largescale evolution. Interval of time required for the evolution of large body size is useful as a test for comparing overall evolutionary rates for unrelated groups, and in keeping with the punctuational prediction, the rapidly radiating Mammalia attained their maximum body size much more rapidly than the Bivalvia (Fig. 4). On the other hand, when mass extinctions must be weathered, slow and steady wins the race: high turnover rates carry the burden of instability. This can easily be seen by considering what would happen to a taxon with a normally high turnover rate compared to one with a low rate if speciation in each were shut down altogether. The first taxon would dwindle in diversity much more quickly than the second, which could thereby withstand a much longer hostile interval (Stanley 1979, pp. 272-278). Thus, it would appear to be no accident that the trilobites, graptolites, and ammonoids—all invertebrate taxa with exceptionally high rates of speciation and extinction—suffered repeated mass extinctions and ultimate demise. Only a few "supertaxa" such as the numeriDIVERSITY (NO. FAMILIES) DIVERSITY (NO. FAMILIES) available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, RATES OF EVOLUTION 25 cally successful insects, rodents, and snakes have stretched the linkage between S and E, possessing very high rates of speciation and only moderately high rates of extinction. These traits may reflect poor dispersal but large, stable populations for species within the confines of their narrow ranges. Unfortunately, there would appear to be good reason why vermin are inheriting the earth. Literature Cited ALBERCH, P. 1980. Ontogenesis and morphological diversification. Amer. Zool. 20:653-667. ANDREWS, G. W. 1976 Miocene marine diatoms from the Choptank Formation, Calvert County, Maryland. U.S. Geol. Surv. Prof. Paper 910. AUFFENBERG, W. 1963. The fossil snakes of Florida. Tulane Stud. Zool. 10:131-216. BERGGREN, W. A. 1969. Rates of evolution in some Cenozoic planktonic foraminifera. Micropaleont. 7.5:351-365. BOOKSTEIN, F. L., P. D. GINGERICH, AND A. G. KLUGE. 1978. Hierarchical linear modeling of the tempo and mode of evolution. Paleobiology. 4:120-134. BRINKMANN, R. 1929. Statistisch-biostratigraphische Untersuchungen an mitteljurassischen Ammoniten uber Artbegrift und Stammesenwicklung. Abh. Ges. Wiss. Gottingen, Math-Phys. Kl., N. F. 23:1-249. BUZAS, M. A. AND S. J. CULVER. 1984. Species duration and evolution: benthic Foraminifera on the Atlantic continental margin of North America. Science. 22.5:829-830. COOPE, G. R. 1970. Interpretations of Quaternary insect fossils. Ann. Rev. Entomol. 25:97-120. DAVIS, S. J. M. 1981. The effects of temperature change and domestication on the body size of Late Pleistocene to Holocene mammals of Israel. Paleobiology. 7:101-114. DICKSEN, J. H. 1973. Bryophytes of the Pleistocene: The British Record and Its Chronological and Ecological Implications. Cambridge Univ. Press; Cambridge. ELDREDGE, N. 1971. The allopatric model and phylogeny in Paleozoic invertebrates. Evolution. 25:156-167. ELDREDGE, N. AND S. J. GOULD. 1972. Punctuated equilibria: an alternative to phyletic gradualism. Pp. 82-115. In: Schopf, T. J. M., ed. Models in Paleobiology. Freeman, Cooper & Co.; San Francisco. EMILIANI, C. 1982. Extinctive evolution. J. Theor. Biol. 97:13- 33. GAZIN, C. L. 1968. A study of the Eocene condylarthran mammal Hyopsodus. Smithson. Misc. Coll. 253:1-90. GILINSKY, N. L. 1981. Stabilizing species selection in the Archaeogastropoda. Paleobiology. 7:316-331. GILLESPIE, J. H. AND R. E. RICKLEFS. 1979. A note on the estimation of species duration distributions. Paleobiology. 5:60-62. GINGERICH, P. D. 1974. Stratigraphic record of Early Eocene Hyopsodus and the geometry of mammalian phylogeny. Nature. 248:107-109. GINGERICH, P. D. 1976. Paleontology and phylogeny: patterns of evolution at the species level in early Tertiary mammals. Amer. Jour. Sci. 276:1-28. GINGERICH, P. D. 1983. Rates of evolution: effects of time and temporal scaling. Science. 222:159-161. GOULD, S. J. 1982. The meaning of punctuated equilibrium and its role in validating a hierarchical approach to macroevolution. Pp. 83-104. In: Milkman, R., ed. Perspectives on Evolution. Sinauer; Sunderland, Mass. GOULD, S. J. 1984. Gingerich's smooth curve of evolutionary rate: a psychological and mathematical artifact. Science. 226:994-995. HALLAM, A. 1975. Evolutionary size increase and longevity in Jurassic bivalves and ammonites. Nature. 258:493-496. HALLAM, A. 1978. How rare is phyletic gradualism and what is its evolutionary significance? Evidence from Jurassic bivalves. Paleobiology. 4:16-25. HANSEN, T. A. 1978. Larval dispersal and species longevity in lower Tertiary gastropods. Science. 299:885-887. HAYAMI, I. 1984. Natural history and evolution of Cryptopecten (a Cenzoic-Recent pectinid genus). Univ. Museum, Univ. of Tokyo, Bull. 24. 149 pp. HENNIG, W. 1966. Fannia seal arts Fabricius, eine rezente Art in Baltischen Bernstein. Stuttgarter Beitrage zur Naturkunde. 150: 1-12. HOFFMAN, A. AND J. A. KITCHELL. 1984. Evolution in a pelagic planktic system: a paleobiologic test of models of multispecies evolution. Paleobiology. 20:9-33. KAUFFMAN, E. G. 1978. Evolutionary rates and patterns among Cretaceous bivalves. R. Soc. Lond. Philos. Trans. 284B:277-304. KELLOGG, D. E. 1975. The role of phyletic change in the evolution of Pseudocubus vema (Radiolaria). Paleobiology. 2:359-370. KENNEDY, G. E. 1983. A morphometric and taxonomic assessment of a hominine femur from the Lower Member, Koobi Fora, Lake Turkana. Amer. J. Phys. Anthropol. 62:429-436. KOCH, C. F. 1980. Bivalve species duration, areal extent and population size in a Cretaceous sea. Paleobiology. 6:184-192. KURTEN, B. 1959. On the longevity of mammalian species in the Tertiary. Soc. Sci. Fenn. Commentat. Biol. 22:1-14. LANDE, R. 1976. Natural selection and random genetic drift in phenotypic evolution. Evolution. 30:314-334. LEOPOLD, E. B. 1967. Late Cenozoic patterns of plant extinction. Pp. 203-246. In: Martin, P. S. and H. E. Wright, Jr., eds. Pleistocene Extinctions: The Search for a Cause. Yale Univ. Press; New Haven, Conn. LERNER, I. M. 1954. Genetic Homeostasis. Oliver & Boyd; Edinburgh. LEVEQUE, F. AND VANDERMEERSCH, B. 1981. Le Neandertalien de Sainte-Cesaire. La Recherche. 22:241-244. LEVINTON, J. S. 1983. Stasis in progress: the empirical basis of macroevolution. Amer. Rev. Ecol. Syst. 24:103-137. LEVINTON, J. S. AND GINZBURG, L. 1984. Repeatability of taxon longevity in successive foraminifera radiations and a theory of random appearance and extinction. Proc. Nat. Acad. Sci. U.S.A. 82:5478-5481. LEVINTON, J. S. AND C. M. SIMON. 1980. A critique of the punctuated equilibria model and implications for the detection of speciation in the fossil record. Syst. Zool. 29:130-142. LEWIS, H. 1966. Speciation in flowering plants. Science. 252:167- 172. MCALESTER, A. L. 1963. Pelecypods as stratigraphic guides in the Appalachian Upper Devonian. Geol. Soc. Amer. Bull. 74:1209- 1224. MALMGREN, B. A. AND J. P. KENNETT. 1981. Phyletic gradualism in a late Cenozoic planktonic foraminifera! lineage. Paleobiology. 7:230-240. MARSHALL, L. G. AND R. S. CORRUCCINI. 1978. Variability, evolutionary rates, and allometry in dwarfing lineages. Paleobiology. 4:101-119. MARTINELL, T. AND A. HOFFMAN. 1983. Species duration patterns in the Pliocene gastropod fauna of Emporda (northeast Spain). N. Jahrb. Geol. Palaontol. Mh. 22:698-704. MAYR, E. 1963. Animal Species and Evolution. 797 pp. Harvard Univ. Press; Cambridge, Mass. NEWMAN, W. A. 1985. The abyssal hydrothermal vent invertebrate fauna: a glimpse of antiquity? In: Jones, M. L., ed. The Hydrothermal Vents of the Eastern Pacific: An Overview. Bull. Biol. Soc. Washington. 6. in press. available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use, 26 STEVEN M. STANLEY PALMER, A. R. 1965. Trilobites of the Late Cambrian Pterocephaliid Biomere in the Great Basin, United States. U.S. Geol. Surv. Prof. Paper 493. RAUP, D. M. 1975. Taxonomic survivorship curves and Van Valen's Law. Paleobiology. 7:82-96. RAUP, D. M. AND R. E. CRICK. 1981. Evolution of single characters in the Jurassic ammonite Kosmoceras. Paleobiology. 7:200- 215. REESIDE, J. B. AND W. A. COBBAN. I960. Studies of The Mowry Shale (Cretaceous) and contemporary formations in the United States and Canada. U.S. Geol. Surv. Prof. Paper 355. RICHARDS, R. B. 1977. Patterns of evolution in the graptolites. Pp. 333-358. In: Hallam, A., ed. Patterns of Evolution, as Illustrated by the Fossil Record. Elsevier; Amsterdam. RIGHTMERE, G. P. 1981. Patterns in the evolution of Homo erectus. Paleobiology. 71:241-246. ROSE, K. D. AND T. M. BOWN. 1984. Gradual phyletic evolution at the generic level in early Eocene omomyid primates. Nature. 309:250-252. SCHANKLER, D. 1980. Faunal zonation of the Willwood Formation in the central Big Horn Basin, Wyoming. Univ. Michigan Pap. Paleontol. 24:99-114. SCHOPF, T. J. M. 1982. A critical assessment of punctuated equilibria. I. Duration of taxa. Evolution. 36:1144-1157. SCHOPF, T. J. M. 1984. Rates of evolution and the notion of living fossils. Ann. Rev. Earth Planetary Sci. 72:245-292. SCHOPF, T. J. M., D. M. RAUP, S. J. GOULD, AND D. S. SIMBERLOFF. 1975. Genomic vs. morphologic rates of evolution: influence of morphologic complexity. Paleobiology 2:63-70. SEPKOSKI, J. J. 1975. Stratigraphic biases in the analysis of taxonomic survivorship. Paleobiology. 2:343-355. SIMPSON, G. G. 1944. Tempo and Mode in Evolution. 237 pp. Columbia Univ. Press; New York. SIMPSON, G. G. 1953. The Major Features of Evolution. 434 pp. Columbia Univ. Press; New York. SLATKIN, M. 1981. A diffusion model of species selection. Paleobiology. 7:421-425. STAINFORTH, R. M., J. L. LAMB, H. LUTERBACHER, J. H. BEARD, AND R. M. JEFFORDS. 1975. Cenozoic Planktonic Foraminiferal Zonation and Characteristics of Index Forms. Univ. Kansas Paleont. Contrib. Art. 62. STANLEY, S. M. 1973. Effects of competition on rates of evolution, with special reference to bivalve mollusks and mammals. Syst. Zool. 22:486-506. STANLEY, S. M. 1975. A theory of evolution above the species level. Proc. Nat. Acad. Sci. U.S.A. 72:646-650. STANLEY, S. M. 1979. Maaoevolution: Pattern and Process. 332 pp. W. H. Freeman; San Francisco. STANLEY, S. M. 1982. Macroevolution and the fossil record. Evolution. 36:460-473. STANLEY, S. M., W. O. ADDICOTT, AND K. CHINZEI. 1980. Lyellian curves: possibilities and limitations. Geology. 5:422-426. STANLEY, S. M. AND W. A. NEWMAN. 1980. Competitive exclusion in evolutionary time: the case of the acorn barnacles. Paleobiology. 6:173-183. STEBBINS, G. L. 1982. Perspectives in evolutionary theory. Evolution. 36:1109-1119. VAN VALEN, L. 1974. Two modes of evolution. Nature. 252: 290-300. WALKER, A. 1984. Extinction in hominid evolution. Pp. 119— 152. In: Nitecki, M. H., ed. Extinctions. Univ. Chicago Press; Chicago. WARD, P. D. AND P. W. SIGNOR. 1983. Evolutionary tempo in Jurassic and Cretaceous ammonites. Paleobiology. 9:183-198. WEI, K.-Y. AND J. P. KENNETT. 1983. Nonconstant extinction rates of Neogene planktonic foraminifera. Nature. 30.5:218-220. WEST, R. M. 1979. Apparent prolonged evolutionary stasis in the primitive Eocene hoofed mammal Hyopsodus. Paleobiology. 5: 252-260. WILLIAMSON, P. G. 1981. Paleontological documentation of speciation in Cenozoic molluscs from Turkana Basin. Nature. 293: 437-443. WILSON, M. V. H. 1983. Is there a characteristic rate of radiation for the insects? Paleobiology. 9:79-85. available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0094837300011362 Downloaded from https://www.cambridge.org/core. University of Sydney Library, on 03 Jun 2018 at 08:00:21, subject to the Cambridge Core terms of use,