This difference underlies the two important disparities listed as lines II4a and II5a — one favoring the evolutionary capacity of organisms, the other of species. Sexual organisms can spread favorable variation to other individuals in the collectivity by recombination. But the favorable features of new species remain stuck in the species and its lineal descendants, and cannot be spread to other species in the clade — except in the infrequent circumstance of hybridization among species in a clade of multicellular forms. (By contrast, lateral transfer seems to be common in the evolution of prokaryotic lineages.) This preclusion of lateral spread puts a strong damper upon evolution within clades. The same limitation, of course, affects asexual vs. sexual organisms — and represents a standard argument for the great advantage of sex, and its evolutionary prevalence, in complex metazoans (Williams, 1975; Maynard Smith, 1978).
Species do gain advantages, on the other hand, in the necessary association of birth with change (sometimes small in extent because reproductive isolation can develop with minimal genetic change, but usually quite substantial). This input helps to offset the disadvantages of small population sizes (species in clades) for species selection. The asexual budding of a new species always yields novelty; the asexual budding of a new organism usually yields clonal identity, and only produces novelty if mutation intervenes.
Contrasting modalities of change: the basic categories
The greatest interest in this analogy lies in the third category of contrasting modalities. Since individuals vary and collectivities evolve (by cumulative changes in their contained individuals) in the standard formulation, I shall first define the three major styles of change within collectivities (populations for the organismal level, and clades for the species level).
Drive. This term has often been used for particular cases — meiotic drive, or molecular drive, for example — but deserves to be formalized and generalized. A driving process transforms a collectivity by directionally changing its contained individuals from within. Drives should be construed as opposite or at least orthogonal to selection. Drives produce change by directional transformation of relevant individuals, not by differential proliferation of some kinds of individuals over others. Thus, in pure cases of drive, change occurs without any differential proliferation. In the paradigm cases of drive, either an individual alters in the course of ontogeny and passes these modifications to offspring, while all individuals produce the same number of offspring with the same reproductive capacities (so no selection can take place); or an individual produces offspring endowed with directional differences from its own constitution — but again, no differential reproduction occurs, and no selection can take place. (As one complexity — an ineluctable consequence of the hierarchical [Page 722] perspective, but a blessing in richness rather than a nuisance in confusion — drives at one level can result from selection at a lower level. In the obvious case, anagenesis within a species — a drive at the species level — traditionally arises from selection among organisms within the species.)
Sorting (selection and drift). This descriptive term generalizes our usual notion of evolutionary change in a collectivity by differential proliferation of some kinds of individuals vs. others. Sorting, as previously defined (p. 659), is a causally neutral and purely descriptive term for any evolution by differential proliferation, whatever the mechanism involved (see original formulation in Vrba and Gould, 1986). Of the two major modes of sorting, selection, based on causal interaction of traits with environments, ranks as the canonical style of evolution, the essence of Darwin's insight, and the foundation of modern theorizing. But sorting can also proceed randomly, a process termed drift. In the hierarchical model, both selection and drift can occur at all levels, under appropriate conditions. I discussed previously, for example, how selectively-based sorting of species can occur either by upward causation from selection at the organismic level (“the effect hypothesis” of Vrba, 1980, also called “effect macroevolution”), or by selection based on irreducible fitness of species-level traits in their interaction with environments (true species selection) — see pp. 652–670.
Ontogenetic drive: the analogy of Lamarckism and anagenesis
The two categories of drive present some of the most consequential and counterintuitive pairings in the entire table (at least they stimulated my own thoughts substantially). In a first category, line IIIA, we must acknowledge as an instance of “drive” any consistently directional change that occurs during the ontogeny of an individual, and then passes by inheritance to offspring.
We do not usually include such a process in our standard account of evolution for an interesting reason based on the history of evolutionary thought and the nature of Mendelian genetics: We generally focus our causal accounts exclusively on organisms in the Darwinian tradition; at the organismic level, a drive of this character would validate the most anathematized and fallacious of alternatives to Darwinism — namely Lamarckism, with “soft” inheritance of acquired characters (see Chapter 3 on Weismann's use of hierarchical thinking to counteract Lamarckism). Thus, ontogenetic drives based on phenotypic changes that are generated by organic activity and then passed to offspring, probably don't exist at the organismal level due to the nature of DNA and the mechanics of heredity. The defeat of Lamarckism — ontogenetic drive in this context — marks one of the great episodes in the history of evolutionary thought. If evolution did proceed in the Lamarckian mode, the geological history of life would assume an entirely different appearance, primarily by enormously accelerated rates of change, and suppleness of adaptive modification. I doubt, for example, that we would find any stable higher-level entities like species in a Lamarckian world. (Human cultural change compares so poorly with Darwinian evolution primarily because our customs and technologies do evolve in this vastly more rapid and flexible Lamarckian [Page 723] mode. Whatever we invent in one generation, we pass directly to the next by emulation and instruction.)
The proper species-level analog for ontogenetic drive, or Lamarckian evolution, sounds a bit bizarre at first — but probably only for the irrelevant psychological reason that we have so firmly rejected the organismic example, while promoting the species-level version as a standard mode of change. With species as individuals and organisms as parts, the gradual transformation (without branching) of an entire species by organismal selection — the standard, canonical description of “evolution” itself — becomes the legitimate analog, at the species level, of heritable ontogenetic alteration, or Lamarckian change, at the organismal level! If, as tradition used to hold, such ontogenetic drive dominates macroevolution, then we must record this striking difference in pattern between levels.
I would argue, however (and under my admittedly partisan commitment to punctuated equilibrium), that this standard impression is fictional, and that ontogenetic drive occurs only rarely at the species level. Differences in frequency will, of course, persist — for mechanisms of inheritance preclude ontogenetic drive in theory at the organismic level, while the analogous process remains possible in principle, though rare in fact, given the nature of populations and their modes of change, at the species level. Thus, small importance remains a common theme at both levels. Most species originate in a geological moment, and persist in stasis thereafter (with, at most, mild fluctuation about an unvarying mean, but no directional change, as the concept of drive requires — see Chapter 9).
I would also venture an analogy to the organismal level in support of inherent reasonableness for the rarity of anagenesis (ontogenetic drive) at the species level. As argued above, the Lamarckian mode works with extraordinary rapidity and efficiency: if organisms changed in this way, we could not fail to notice, because evolution would then operate so differently. I would suggest that we approach macroevolution at the species level in the same way. If species changed gradually most of the time, the pageant of life's history, as shown by the fossil record, would present an entirely different appeara
nce. The most extensive transformations would occur in a few million years at most. (Many hypothetical calculations have been made to illustrate this point — for example, that a small, four-footed, terrestrial mammal can evolve into the largest whale in a fraction of Tertiary time, so long as a single population in transformation maintains the smallest effectively measurable selection coefficient, unabated and without change in direction.)* Stable [Page 724] clades could not dominate the history of life, as they manifestly do, particularly in marine invertebrates (clams, snails, horseshoe crabs, brachiopods, all from the Paleozoic to the present); nor, among more rapidly changing terrestrial clades, could dinosaurs (not to mention the more stable insects) persist and rule for so long in a world where most species evolved continually by the analog of ontogenetic drive in the Lamarckian mode.
Of course, we could posit other reasons for braking the rapidity and efficiency of change by ontogenetic drive in macroevolution — disruption of trends by mass extinction; high frequency of trends that benefit organisms but harm species (peacock's tails), for example. But I suspect that the simplest of all reasons will explain the evident pattern: the species-level analog of ontogenetic drive — gradual transformation within a species — just doesn't occur very often.
Finally, I note that R. A. Fisher's classic argument for the impotence of species selection rests on the standard assumption that this mode of driving does prevail in evolution. For, if most species, most of the time, changed gradually from within (see pp. 644–646), then selection among species would be, as Fisher rightly noted, an operative but impotent process, capable of generating only an insignificant amount of change relative to the dominant and ubiquitous drives of anagenesis. But if anagenesis rarely occurs, Fisher's argument collapses. I wonder if Fisher ever explicitly realized that anagenesis would trump species-selection because anagensis is Lamarckian at the species level, while species selection is Darwinian at the same level — for Lamarckian processes can always overwhelm the much weaker force of Darwinian change if both operate generally and in an unimpeded manner at the same level.
Reproductive drive: directional speciation as an important
and irreducible macroevolutionary mode separate from
species selection
Thus, the first category of ontogenetic drive illustrated interesting differences in style between levels, but little variation in effect — for I conclude that this mode has scant impact upon evolution at either level. But when we consider the second category of reproductive drive (biased production of offspring that vary in a given direction from parents), we encounter one of the chart's most striking disparities — a crucial, yet almost entirely unrecognized and unexplored difference in basic pattern between micro- and macroevolution. To choose a hypothetical example of simplified form but maximal clarity: Suppose that each collectivity (a population for the organismal level, or a clade for the species level) contains ten individuals (organisms or species, for the two levels). Each individual gives birth to a single offspring, and all offspring have identical life spans and reproductive capacities. Thus, no selection at all [Page 725] can take place. Now suppose that a strong bias exists in production of offspring, so that 80 percent arise with smaller bodies than their parent (lower weight for the offspring organisms, lower average body weight for the offspring species). Suppose also that this pattern continues from generation to generation. This driving process would generate a strong trend to smaller bodies in the collectivities at both levels — a gradual trend to decreased body size in the population at the organismal level; and to species with smaller average body sizes within the clade at the species level.
As discussed previously (p. 691), reproductive drives of this kind can occur at the organismal level, and a variety of names for such processes exist, including mutation pressure and meiotic drive. But the Darwinian tradition has always regarded such phenomena as insignificant as a consequence of their rarity. Indeed, these processes must be rare in a fully Darwinian world, because reproductive drives violate the necessary precondition of undirected variability for natural selection (see pp. 144–146). Darwinians did not win this debate by simple logic or evident factuality, but only by a great intellectual struggle marking a crucial episode in the history of evolutionary thought. The classical debate about orthogenesis, for example (see Chapter 5), centered upon the Darwinian denial of such reproductive drives, which, as the competing orthogeneticists all realized, would overwhelm selection by higher efficacy — if they existed. Perhaps such reproductive drives rarely occur at this level in nature because, having no known basis for inherent adaptivity, they have been actively suppressed by organismal selection — another potential example of the most distinctive feature of organismic individuality: the power evolved by functional integrity to suppress lower-level selection from within.
However, when we move to the species level, the analogous driving phenomenon of directional speciation suffers no constraint or suppression — and may represent one of the most common modes of macroevolution. Two major reasons underlie the high potential frequency for directional speciation (as opposed to the rarity of its analog at the organismal level — see line III2a on the chart). First, as noted in several other contexts, the species-individual does not maintain integrity (as the organism does) by suppressing differential proliferation of some parts over others. Since drives at an upper level arise by differential proliferation of lower-level units, this absence of suppression leaves a large open field for driving processes to operate at the birth of new species. Second, since new species-individuals must arise with sufficient heritable novelty to win reproductive isolation from their parent (whereas children of asexual organisms may be clonally identical with parents), all species births include genetic change as an automatic consequence. Any statistical directionality in such changes among species in a clade will produce a trend by drive.* [Page 726]
We may postulate any number of plausible circumstances that would generate directional biases in the origin of new species — thus producing a cladal trend without any contribution from species selection. Moreover, potential causes for directional bias exist at all levels — organismic, demic, or species — thus greatly expanding the scope of the phenomenon. As a central theoretical point, directional speciation, when based on irreducible species-level properties, represents a style of independent and causal macroevolution not based on species selection. Thus, the claim for an independent body of macroevolutionary theory does not depend upon the validity and high relative frequency of the Darwinian analog most often discussed as a paradigm case, namely species selection. Directional speciation, when based on irreducible species-level traits or processes, designates another category of intrinsically macroevolutionary change.
To continue in the hypothetical mode with the example cited previously, one can easily imagine how a cladal trend, attributable entirely to reproductive drive (and not at all to selection), and leading to decreasing average (organismal) body size, might be caused at either the organismal, demic, or species level. At the conventional organismic level, a pervasive environmental change over the entire region of a clade's occupancy might favor natural selection for smaller bodies. (Perhaps, to choose a somewhat cardboard example, a temperate region has become tropical, and smaller organisms now gain advantages within each species of a clade by the adaptive correlates of Bergmann's Rule.) Each species produces a single daughter species and then dies out — so no selection can occur at the species level. But if most species, in the new climatic regime, originate at smaller average body size because natural selection favors this trait among the organisms in each species, then the cladal trend arises by directional speciation with a cause based on selection at the organismal level — the classic case of a drive at a higher level produced by directional selection among contained parts at a lower level.
For a hypothetical case based on interdemic selection, suppose that each species in a clade develops ten small and
isolated demes at the periphery of the parental range. Suppose that average body size in these peripheral isolates varies randomly around the parental mean. Suppose further that, for each species, only one of the ten peripheral demes survives, intensifies its differences, and eventually becomes a new species — while the parent and the other nine peripheral isolates all die. Again, no species selection can take place, for [Page 727] each parent spawns one and only one daughter. (I realize, of course, that these strictures sound absurd if construed as actual and coordinated occurrences in nature; I am only following the time-honored heuristic method in science of constructing “pure” end-member hypothetical to help clarify our thoughts.)
The Structure of Evolutionary Theory Page 116