The Structure of Evolutionary Theory
Page 148
Therefore, meaningful “homology” in this conceptual sense must generally be sought in properties that are genuinely held in common across systems and scales, and that operate to channel the different causes of these various scales into the same recognizable and distinctive pattern. Moreover, such [Page 929] homology becomes all the more interesting if the particular efficient causes at different scales — the actual pushers and movers of immediate change in each case — remain evidently disparate, thus implying, by elimination, that the observed commonality of pattern may arise from constraining channels of similar structural properties across scales. If all roads lead to Rome, then the eternal city ranks as a dominant and ineluctable attractor!
In the present case of punctuational patterns across markedly different scales of time and component entities, claims for conceptual homology rest upon an overarching hypothesis that punctuation records something quite general about the nature of change itself, and that the differing causes of punctuational change at each level — the waiting time between favorable mutations in bacterial anagenesis (see next section), the scaling of ordinary speciation as a geological moment in punctuated equilibrium, or the simultaneity of species deaths in mass extinction — must run in a common structural channel that sets and constrains the episodic nature of alteration.
If punctuated equilibrium gains this generality by conceptual homology, then both components of its name should achieve such transfer across nature's numerous scales of size and time. (The general mathematical models discussed in the last section presume such meaningful transfer as a primary rationale for their relevance.) The equilibrial component wins potential generality if active resistance to change can be validated as an important structural property of systems discrete and stable enough to be named and recognized as entities at any scale of nature (whatever the causes of stability, whether internal to self-integration or imposed from without upon an intrinsically less coherent structure — a fascinating question that should become an object of research, not the subject of prior definition). This property of active maintenance underlies our primary claim about stasis in punctuated equilibrium, and our insistence that stasis must be conceptualized and defined as a positive phenomenon, not as a disappointing or uninteresting absence of anticipated change. (Throughout this chapter, I have provided evidence, primarily in observed relative frequencies — far too high to originate either passively or randomly in a world of natural selection and genetic drift — for interpreting stasis as an actively generated property of systems, embodied in species at the scale of punctuated equilibrium, but necessarily recognized in structures of different status at scales both below and above species.)
The punctuational component, operationally measured by its short duration relative to periods of stasis within definitive structures of the same scale, would then achieve homological generality as the obverse to proposed reasons for stasis: the reinterpretation of change — at least in its usual, if not canonical, expression — as a rare and rapid event experienced by systems only when their previous stabilities have been stretched beyond any capacity for equilibrial return, and when they must therefore undertake a rapid excursion to a new position of stability under changed conditions.
Obviously, these “brave” statements about conceptual homology across disparate scales and immediate causalities must remain empty and meaningless without operational criteria for distinguishing — if I may again use the [Page 930] conventional evolutionary jargon in this wider context — meaningful similarity of genesis (homology) from misleading superficiality of appearance (analogy). As a first rule and guideline, we might look to the same basic precept of probability that regulates our general procedures in the study of overt similarity among separate phenomena: co-occurrence of substantial numbers of potentially independent parts as a sign of meaningful genetic (and conceptual) connection vs. resemblance based upon single or simple features, however visually striking, as far more likely to be unconnected, separately built, and perhaps not even meaningfully alike in any causal or functional sense (the complex and identical topology of arm bones in a bat's wing and a horse's foreleg as meaningfully homological vs. my face and the same disposition of holes on the Martian mesa as meaninglessly analogical). Thus, and in a practical sense, I focus much of the following discussion upon a search for what I will call “conjoints,” or sets of independent features whose joint occurrence predisposes us to consider meaningful conceptual homology in punctuational patterns of change produced by different immediate causes at disparate scales of size and time. (I have used the same form of argument frequently throughout this book — as in emphasizing the usual conjunction of openness to saltational change, belief in the importance of internal channeling, and suspicion about adaptationist explanation in defining the biological worldview of structuralist thinkers — see Chapters 4 and 5).
When such broad “homologies” of common structural constraint have been established across several realms of size and time, then we can most fruitfully ask some second-order questions about systematic, or “allometric,” differences (see Gould and Lloyd, 1999) in the expression of common patterns along continua ordered by increasing magnitude among scales under consideration. For example, do internal forces of cohesion among subparts set the primary basis for active stasis, or does the “fit” of a structure into a balanced and well-buffered environment, made of numerous interacting entities, prevent change in a system otherwise fully capable of continuous alteration in the absence of such externalities? Does the balance between these internalist and externalist explanations change as we mount through scales of magnitude? Is the change systematic (and therefore “allometric” in the usual sense), or capricious with respect to scale itself, having no correlation with magnitude?
The important principle that meaningful similarity may reside in homology of structural constraint across scales, while particular causes that “push” phenomena through these constrained channels may vary greatly, has rarely been stated or exemplified with proper care, and has therefore usually been ignored by commentators on the role of events at one scale in the interpretation of others. Most regretfully, a frequent misunderstanding has then led to dismissal of meaningful commonality in pattern because a critic notes a strong difference in immediate causes for a pattern at two scales and then rejects, on this erroneous basis, any notion of an informative or integrative status for the similarity. Or, even worse, the critic may become intrigued by a cause just elucidated at one scale and then assume that the significance of [Page 931] such a discovery can only lie in extrapolating this particular and strongly scale bound cause to debunk a different mechanism previously proposed to explain the same pattern at another level — rather than exploring the more fruitful and integrating hypothesis that a genuine basis for meaningful similarity in pattern might reside in homologous structural constraints that channel different causes to similar outcomes at the two distinct scales.
For example, an excellent science reporter for the New York Times erroneously argued that punctuations caused by long waiting times between rapidly-sweeping, favorable mutations in bacterial anagenesis on a scale of months should lead us to reinterpret the speciational breaks of punctuated equilibrium (at geological scales) as similarly caused by quick and simple genetic changes! “The finding that all it takes is a few mutations and a little natural selection to generate punctuated evolution comes as a surprise. Researchers say numerous theories that are considerably more complex have been put forth to explain what might produce the punctuation seen in the fossil record. If bacteria are any indication, the rapid evolution documented in the fossil record might be the product of a very few simple, if quick, genetic changes” (Yoon, 1996).
But R. E. Lenski, the chief scientist in the bacterial study (Elena, Cooper, and Lenski, 1996), properly sought commonality with punctuated equilibrium in the domain of homologous reasons for punctuational patterns. Recognizing the dispari
ties in scale, and the different causes thus implied, they rightly declined to apply the term-punctuated equilibrium to their findings. Instead, they invoked the general term for the pattern itself as the title for their paper (Elena, Cooper, and Lenski, 1996): “Punctuated evolution caused by selection of rare beneficial mutations.”
PUNCTUATION BELOW THE SPECIES LEVEL. I have, at several points in this and the preceding chapter, discussed various empirical and theoretical studies that validate the pattern of substantial stability followed by rapid peak shifts in the anagenetic transformation of single populations during the microevolutionary time of potential human observation (see p. 877). I have also urged (to reiterate the theme of the preceding section) that such an important conclusion should not be read as an argument that punctuated equilibrium holds no interest for evolutionary theory because ordinary population genetics can produce patterns of stasis and punctuation — a common but erroneous claim rooted in the misinterpretation of punctuated equilibrium as a saltational theory in ecological time. Rather, this small-scale anagenetic conclusion for another domain of size and time should be read as welcome confirmation — based on causes different from the generators of punctuated equilibrium at a larger scale — for the broader claim that punctuational patterns may be common and robust across several spatial and temporal realms in nature.
But the most impressive affirmations of punctuational patterns at scales below punctuated equilibrium have emerged, in recent years, from a domain unparalleled (and unmatchable) for richness of empirical data on evolution [Page 932] over a sufficient number of generations to claim potential linkage with scales of substantial evolutionary change in nature: well-controlled experimental studies of bacterial lineages.
This field is now developing so rapidly that any particular study, as discussed here, will, no doubt, seem quite rudimentary by the time this book reaches the presses. But, as I write in 1999, an impressive case may be taken as indicative of possibilities and directions. By using strains of E. coli that pass through six generations in a single day, Elena, Cooper, and Lenski (1996; see also Lenski and Travisano, 1994) were able to study evolution in cell size for 10,000 continuous generations. By imposing constancy of environment (to limits of experimental perceptibility of course), and using a strain lacking any mechanism for genetic exchange (Elena et al., 1996), mutation becomes the sole, and experimentally well isolated, source of genetic variations.
The experimenters have followed 12 replicate populations, each founded by a single cell from an asexual clone, and each grown under the same regimen (of daily serial transfer, with growth for 24 hours in 10 ml of a glucose-limited minimal salts medium that can support ca. 5 × 107 cells per ml). At an average of 6.6 bacterial generations per cycle, the population undergoes a daily transition from lag phase following transfer, to sustained increase, to depletion of limiting glucose and subsequent starvation. At each serial transfer, a 1:100 dilution begins the next daily cycle with a minimal bacterial population of ca. 5 × 106 cells. Samples of the common ancestral population, and of selected stages in the history of each population, were stored at — 80° C, and can be revived for competition experiments with the continually evolving populations — a situation that can only fill a paleontologist with envy, and with thoughts of beautiful and utterly undoable experiments from life's multi-cellular history (neanderthals or australopithecines released in New York City; tyrannosaurs revived to compete against lions in a field of zebras, etc.).
In each of the 12 populations, both fitness and cell volume increased in a punctuational manner through the 10,000 generations of the experiment. (The experimenters measured cell volume by displacement (Lenski and Travisano, 1994, p. 6809), and mean fitness of populations by the Malthusian parameter of realized rate of increase in competition against resuscitated populations of the common ancestor.) The general path of increase followed the same trajectory in all populations, but with fascinating differences of both form and genetics in each case — a remarkable commentary, at such a small and well-controlled scale, of the roles of detailed contingency and broad predictability in evolution (see the explicit discussion of Lenski and Travisano, 1994, on this point).*
[Page 933]
9-34. Punctuation in clonal evolution below the species level — from Lenski and Travisano, 1994. In 12 replicate populations of E. coli during ten thousand generations, mean cell size increases rapidly during the first two thousand generations, and then very little during subsequent evolution.
Punctuational patterns occur at the two different scales of overall trajectory and detailed dynamics, even within the limited scope of this study. In the 1994 paper, Lenski and Travisano sampled each of the 12 populations once every 500 generations. They noted a rapid increase in mean cell size, well fit by a hyperbolic model (see Fig. 9-34), for the first 2000 generations in each population, followed by several thousand subsequent generations of little or no further increase — a pattern that they described as punctuational in one of the major conclusions of their paper (1994, p. 6809): “For ca. 2000 generations after its introduction into the experimental environment, cell size increased quite rapidly. But after the environment was unchanged for several thousand generations more, any further evolution in cell size was imperceptible ... These data therefore indicate a rapid bout of morphological evolution after the population was placed in the experimental environment, followed by evolutionary stasis (or near stasis).”
But, as reported in a later paper (Elena, Cooper and Lenski, 1996), they then sampled each population at a much finer scale — every 100 generations for the first 3000 generations of the experiment. Now they found clear evidence of a punctuational “step pattern” (see Fig. 9-11, discussed previously) within the initial phase of rapid increase that they had previously fit with a simple hyperbolic model. The authors noted: “When mean fitness was measured every 100 generations over the period of faster change ... a step function model, in which periods of stasis were interrupted by episodes of rapid change, gave a better fit to the data than did the hyperbolic model. Evidently, [Page 934] it was necessary to make measurements at sufficiently high frequency to resolve the punctuated dynamics.”
This gain of insight by finer sampling raises the important methodological theme that proper choice for a scale of inquiry depends crucially upon the resolution needed to identify and characterize the underlying causal process of the observed pattern — in particular, to specify the natural unit or entity of change in the given system. (Such studies often face the paradox that, whereas the recognition of this principle requires no act of genius, empirical adequacy often founders upon a conceptual dilemma: We can specify a proper scale if we know the causal basis beforehand. But, more often than not, we undertake such research in order to discover an unresolved causal basis — thus bringing upon ourselves the classical problem of a single equation with two unknowns: the causal basis and the scale required for its identification, to complete the analogy.)
At the macroevolutionary scale of punctuated equilibrium sensu stricto, events of speciation represent the natural unit, and geological resolution must be sufficient to identify the occurrence and timing (relative to stasis, or any other pattern, in the species's subsequent geological history) of origination for these macroevolutionary “atoms.” Several published studies have been fatally marred by the basic flaw of using a scale so coarse that a trend generated by multiple events of staircase speciation could not be distinguished in principle from the same result achieved by smooth anagenesis in an unbranched lineage. In the most widely discussed fallacy thus engendered, Cronin et al. (1980) claimed gradualism (explicitly against punctuated equilibrium) for major trends in hominid evolution because a temporal sequence of mean values moved in the same direction. But the successive points were so widely separated in time and morphology that the authors could not determine whether they had measured mean values of successive species during their periods of stasis, or had sampled points in an a
nagenetic continuum. (Punctuated equilibrium, after all, was proposed as an alternative explanation for phyletic trends of this kind, not as a denial of their existence!) The scale of measurement used by Cronin et al. may be compared with Lenski's first procedure of sampling every 500 generations. Both schemes are too coarse to “catch” the underlying causal unit of change — speciation, if punctuated equilibrium holds, for the macroevolutionary case of hominids; the infrequent origin and sweep of favorable mutations in bacterial anagenesis.
In a fascinating study, extending (to an utterly different realm of inquiry) the generality of this important point about appropriate scale of measurement for the recognition of punctuations, Lampl et al. (1992) note that human growth in body length has generally, and for centuries of study, been regarded as smooth (albeit highly variable in rate at different states of ontogeny) because “individuals have been traditionally measured at quarterly intervals during infancy, and annually or biannually during childhood and adolescence. Physiological data are mathematically smoothed and growth is represented as a continuous curve” (1992, p. 801). But by measuring a sample of 31 “clinically normal” Caucasian children at intervals ranging from [Page 935] daily to weekly between the ages of 3 days and 21 months, Lampl et al., using the language and concepts of punctuated equilibrium, found that “90 to 95 percent of normal development during infancy is growth-free and length accretion is a distinctly saltatory process of incremental bursts punctuating background stasis” (p. 801). In fact, Lampl et al. did not detect the pattern of quickness in change and prevalence of stasis until they measured subjects at their finest daily scales (for even the semi-weekly and weekly measurements smoothed out punctuations over intervals of stasis). They conclude (Lampl et al., 1992, p. 802): “Human length growth during the first two years occurs during short (less than 24 hours) intervals that punctuate a background of stasis. Contrary to the previous assumption that the absence of growth in developing organisms is necessarily pathological, we postulate that stasis may be part of the normal temporal structure of growth and development.”