I argued in the last section that development establishes preferred channels of variation in two primary modes, both “positive” in their salutary contribution to a more accurate and sophisticated evolutionary theory. But by mechanistic criteria of channels as limitations or impetuses, we might deem the first mode — based on the surprising discovery of “deep homology” in the genetic basis of conserved developmental pathways among distantly related animal phyla — as “negative,” in highlighting the limitations thus imposed upon directions of change. (Nonetheless, combinatorial possibilities remain as broad as realized bilaterian diversity, so these limits may direct, but surely do not seem to throttle life — see Kirschner and Gerhart, 1998 and references therein, on flexibility and evolvability.) The second mode — based on the equally surprising discovery of common genetic pathways underlying several textbook cases of supposed convergence, thus recasting these homoplasies as parallelisms potentiated by common developmental architecture — then achieves its best explication as a set of positive impetuses for channeling adaptive change into accessible pathways. [Page 1091]
However, in Raff's larger sense, both modes express the common, cardinal feature of nonisotropic, or channeled, variation — thus imposing a preferred structure, from the “inside” of organic development, upon the raw material that external forces of Darwinian selection must utilize. Both modes also delighted (or disturbed) evolutionary biologists with the greatest surprises in the last generation of our science — based on results that were actively unexpected in theory, not merely unsuspected for lack of imagination. I shall, in this first section, discuss deep homology as the more general and fundamental of the two modes. My second section shall then extend Raff's theme of directed variation as the focus of constraint within evolutionary theory, this time through the positive channel of unanticipated parallelisms.
In a famous line from the prologue of Faust, Goethe wrote: Es irrt der Mensch, so lang er strebt — we err, so long as our struggle lasts. Goethe probably intended this celebrated statement as a romantic effusion about human striving in general, but we may apply his words to the nearly universal attitude of fellow biologists, at least since Darwin's watershed of 1859, towards Goethe's own brainchild in developmental biology, and towards the general approach to morphology — a word of Goethe's own invention — embodied within such theories.
As discussed in extenso in Chapter 4, Goethe's theory of the leaf as a botanical archetype for all lateral structures off the angiosperm stem (including cotyledons and all flower parts) presented the most famous botanical proposal among a set of archetypal theories that would soon sweep the world of animal morphology as well, culminating in the vertebral archetype advocated by Owen for all major parts of the vertebrate skeleton (including the skull and limbs) and, most extensively, by Etienne Geoffroy Saint-Hilaire for the generative basis of all animal form (first for all vertebrates, then adding arthropods, then mollusks, and no doubt proceeding further had he not then encountered the wrath and active opposition of Cuvier and his functionalist theory of adaptive form — see Chapter 4, pp. 291–312).
The argument that structural and morphological archetypes underlie, and actively generate, a basic and common architecture in taxonomically distant groups defines — both as a fact of our profession's actual history and as a dictate of the logic of our explanatory theories — the strongest kind of claim for developmental constraint as a major factor in patterns of evolutionary change and the occupation of morphospace. I suspect that the depth of this challenge has always been recognized, but the empirical case for such constraining archetypes has remained so weak, since the heyday of Geoffroy and Owen some 150 years ago, that the issue simply didn't generate much serious concern — and rightly so.
The concept of interphylum archetypes, deemed too bizarre to warrant active refutation, experienced the curt and derisive dismissal reserved for crackpot ideas in science. (Goldschmidt's saltational apostasy, on the other hand, inspired voluminous and impassioned denial because his ideas did seem sufficiently and dangerously plausible to the Modern Synthesis — see pp. 451–466). Indeed, the notion of interphylum archetypes struck most biologists as [Page 1092] so inconceivable in theory that empirical counterclaims hardly seemed necessary. After all, the notion required extensive genetic homology among phyla, and the power of natural selection, working on different paths for a minimum of 530 million years since the origin of distinct phyla in the Cambrian explosion, seemed to guarantee such thorough change at effectively every nucleotide position that the requisite common foundation could not possibly have been maintained (see Mayr, 1963, p. 609, as previously discussed on pp. 539 and 1066).
When, in the mid-1980's, initial studies began to discern deep homology between arthropod and vertebrate Hox genes, I well remember saying to myself (amidst my astonishment about a result so consonant with the theoretical framework that I had espoused in 1977 in my first book, Ontogeny and Phylogeny, but had not dared to view as subject to empirical validation in my lifetime): yes, perhaps for some functional commonality in the broadest construction of basic body axes (A-P in particular), but surely not for the more detailed structural homology — particularly between arthropod metameres and vertebrate somites — demanded by the old archetypal theories. But, only 15 years later, central nuggets of validity had been affirmed for nearly all the classical archetypal theories, even the most farfetched. Needless to say, the archetypes do not function as their inventors claimed. The differences between leaves and floral parts do not arise by progressive refinement of sap up the stem; and the abstract vertebra does not function as a generator for all major features of the axial skeleton (including ribs and appendages) in vertebrates and arthropods.
Moreover, at least two prominent claims for the vertebral archetype probably hold little, if any, validity. The distinctive features of the vertebrate skull and forebrain seem to arise, in large part, under the formative influence of the distinctive neural crest (see the classical statement of Gans and Northcutt, 1983), and not as a complex fusion (much like an arthropod tagma) of a definable number of rostral vertebrae (from 3 to 8 in various formulations). And although some broad homologies may set the basic axes of limbs in both arthropods and vertebrates (see pp. 1138–1142), the structures cannot be regarded as basically homologous, even in underlying developmental pathways; nor can they be derived from any particular component of a generalized vertebra. Nonetheless, all three major archetypal theories of Goethe and Geoffroy — the classical sources of ridicule for the general concept — have now been confirmed in aspects that cannot be dismissed as superficial or secondary.
MEHR LICHT (MORE LIGHT) ON GOETHE'S ANGIOSPERM ARCHETYPE. Students of the mustard Arabidopsis have discovered unexpected validity in central features of Goethe's founding theory of the archetypal leaf (see Pelaz et al., 2000). Starting at the bottom, Meinke (1992) studied the lee (leafy cotyledon) mutant that partially transforms cotyledons into leaves. He argued that the wild-type allele (LEC) activates “a wide range of embryo-specific [Page 1093] pathways in higher plants” (p. 1647), and that suppression by the lee mutation therefore causes reversion to a ground state — which, as Goethe proposed so long ago, most closely resembles a stem leaf in basic form. (Biologists with a zoomorphic bias, including the author of this book, may be confused by a claim that embryonic features might thus be conceived as departures from a ground state. The directionality of bilaterian ontogeny, with embryonic features as transient and formative, leads us to equate embryonic forms with any sensible concept of a “ground state.” But plants maintain embryonic tissues throughout life as restricted and persistently specialized regions on differentiated foundations that animal biologists might tend to regard as “adult.” Therefore, a botanical rationale for viewing these foundations as a ground state, with embryonic tissues as a specialization, can easily be defended.)
Meinke (1992, p. 1649) concludes: “The phenotype of leafy cotyledon sugge
sts that the difference between leaves and cotyledons in Arabidopsis is controlled by a single regulatory gene (LEC) expressed only during embryo-genesis.” Then, in a statement strikingly evocative of Goethe's archetypal theory, he portrays (1992, p. 1649) the ordinary stem leaf as a ground state, with all its serial homologs (to apply this zoomorphic term to cotyledons and, putatively, to flower parts) as specializations thereupon: “The preferred model is that LEC functions to activate a wide range of embryo-specific pathways in plants. Loss of gene function disrupts embryonic maturation and returns mutant cotyledons to a basal developmental state. The leafy appearance of mutant cotyledons was unexpected because there was no evidence that cotyledons defective in maturation should be transformed into foliage leaves. However, this observation is consistent with the origin of cotyledons as specialized leaves during plant evolution and the homology of embryonic cotyledons and vegetative leaves.”
For the more complex organs of inflorescence at the other end, Weigel and Meyerowitz, in their classic review (1994) of the ABC model (see pp. 1063–1065) for floral development in Arabidopsis (and many other angiosperms, though perhaps not all, see Kramer and Irish, 1999), posed a first key extension beyond the model's basic elucidation: “The ABC model left one complication, though: what happens in the absence of all organ identity activity” (p. 203). Weigel and Meyerowitz then turned to Goethe for the classic prediction based on notions of the archetypal leaf: “Goethe (1790) had proposed that floral organs represent modified leaves, suggesting that a vegetative leaf is the ground state of floral organs.”
Weigel and Meyerowitz presented striking evidence to confirm this Goethian prediction that suppression of all ABC activity should cause presumptive floral parts to approach the ground state of stem leaves. The sequential action of ABC genes permits a simple formulation of tests for this hypothesis. AC double mutants, for example, should knock out determinants for the outermost sepals of whorl 1 (triggered by A genes alone) and the innermost carpels of whorl 4 (C genes alone), but impose less effect upon the petals and stamens of whorls 2 and 3, which also require the influence of B genes (see [Page 1094] p. 1063 and Fig. 10-12). Experiments then confirmed this precise, and rather odd, prediction: “Indeed, organs in these two whorls are very much like vegetative leaves — they develop with stipules, are green and covered with branched hairs, and senesce slowly, all characteristics of leaves but not of floral organs” (Weigel and Meyerowitz, 1994, p. 203). By the same logic, triple mutants should grow all floral parts in leaf-like form — as they do: “In triple mutants that lack A, B, and C activities, all floral organs resemble leaves” (pp. 203–204 — and Fig. 10-14), thus supporting (Pelaz et al, 2000, p. 202) “the theory that flower organs are simply modified leaves.” Theissen and Saedler (2001, p. 470) add, with specific homage to Goethe: “combined loss-of-function of class A, B, and C genes results in a transformation of all floral organs to leaves, corroborating Goethe's view that leaves are a developmental ground state.”
Moreover, gain of function mutations also confirm the model by imposing inner floral expression upon outer parts, thus resembling the action, for a different symmetry of radial whorls, of classical homeotic mutations of Drosophila, expressed in a linear, anteroposterior array. Over expression of C genes, for example (1994, p. 206), represses A functions in whorls 1 and 2, “with carpels where sepals are usually found, and stamens in the places ordinarily occupied by petals” (p. 206).
Later work has revealed some of the upstream regulators of this system. For example, Pelaz et al. (2000) identified three genes (named SEP1/2/3) required for the action of B and C genus that regulate the inner three whorls of petals, stamens and carpels. In triple mutant Aribadopsis plants that suppress the action of SEP1/2/3, all floral whorls develop as sepals (which are regulated by A genes). (See Honma and Goto, 2001, for later data on the even
10-14. Mutations that delete activity of all ABC genes cause all floral organs to develop as leaves. Ordinary flower at A; triple mutant with all flower parts replaced by leaves at B. From Weigel and Meyerowitz, 1994.
[Page 1095]
broader role of SEP genes in “providing flower-specific activity” (p. 528) in combination with genes of the ABC series.)
Other studies provide additional confirmation (in modern genetic form) for Goethe's original formalist notion of leaves as a ground state. A “meristem identity factor” LEAFY (LFY) potentiates APETALA1 (API), which, in turn, activates the ABC floral genes. Wagner et al. (1999, p. 582) demonstrate that this sequence of LEAFY to API is “necessary and sufficient for this transition” (p. 582). Standard techniques for documenting the effects of both loss and gain-of-function mutants confirm this cascade. In the lfy-6 mutant, suppressing the action of LFY, “most flowers are replaced by leaves and second-order shoots”; while over expression of either LFY or API “results in formation of flowers or leaves and flowers in positions normally occupied by leaves” (Wagner et al., 1999, p. 582. See further confirmations in Busch et al., 1999).
Extending the model to other angiosperm clades, Hofer et al. (1997) studied PEAFLO, the pea homolog of LFY. They performed several experiments to extend Goethe's formalist concept of morphological serial homology, now abetted by new data on genetic and developmental homology, between leaves and flower parts. They state: “A striking comparison can be made between the similar developmental units of compound leaves and flowers: both arise laterally from primordia derived from the shoot apical meristem; both produce lateral, leaf-like organs; and both are determinate.” Hofer et al. (1997) then affirmed and extended the evidence for developmental homology by (1) identifying pleiotropic mutants that affect both leaf and floral development in similar ways, and (2) by studying homeotic mutations that “result in the conversion of floral organs to leaf-like structures” (p. 581). Their concluding remark, reinforced by a later observation of Theissen and Saedler (2001, p. 469), might have caused Faust to lose his bet with Mephistopheles — by inducing such delight that this restless, archetypal romantic might finally have savored a present moment with sufficient gusto to blurt out the fateful phrase that would seal his doom: “verweile doch, du bist so schon” (stay awhile, thou art so beautiful). Hofer et al. write (1997, p. 586): “Compound leaves and flowers can thus be considered to be derivatives of the same ancestral structure.” Theissen and Saedler simply conclude: “Goethe was right when he proposed that flowers are modified leaves.”
HOXOLOGY AND GEOFFROY'S FIRST ARCHETYPAL THEORY OF SEGMENTAL HOMOLOGY
An epitome and capsule history of hoxology. These Goethian confirmations extend, at least for now, little beyond the serial homology of apparently disparate parts on the same plant. But archetypal claims for homology across distantly related phyla raise far more serious theoretical problems. No Shockwaves attended the discovery of common genetic and developmental pathways for the serial array of arthropod appendages, despite their functional differentiation as antennae, mouthparts, legs, genital claspers, etc. [Page 1096] But the discovery that homological pathways also persist among animal phyla that have evolved independently since the Cambrian explosion has reversed previous certainties and brought Geoffroy's despised archetypal theories into renewed respectability.
The roots of this great discovery extend back (at least terminologically) to another key figure of this book, the English geneticist William Bateson (see Chapter 5, pp. 396–415). Bateson became fascinated by a class of mutations with the peculiar, and often large, effect of causing the characteristic form of one member in a serial array to develop in a different location usually occupied by another member of the same array. Bateson called such mutations “homeotic,” and their peculiar forms, almost humorous in some cases, gave them a special salience among geneticists. Unsurprisingly — for arthropods are serial organisms par excellence, while this particular insect became the lynchpin of genetics — the homeotic mutations of Drosophila became classics of the genre, famous for their oddness as well as t
heir utility (for geneticists, not for the afflicted flies!).
We all remember our undergraduate textbook pictures — and the attendant, inevitable thoughts of Hollywood monster movies — of flies with such mutations as antennapedia (legs where antennae “ought” to be), bithorax (with another pair of wings rather than halteres on the third thoracic segment, thus seeming to “revert” the fly — a false interpretation as we shall see — to the ancestral four winged condition), and bithoraxoid (with a supernumerary pair of legs on the first abdominal segment, thus giving eight legs in toto and seeming to mock the very definition of the class Hexapoda). In my favorite example, a homeotic mutation in mosquitoes actually replaces the biting stylets with a pair of legs, thus rendering the creature “ouchless.” I entertained various fantasies about breeding these lovely mutants, introducing them into natural populations, and destroying this scourge of humanity from within. But, alas and unsurprisingly, the scheme would never work, and I couldn't interest a single venture capitalist — for the mutation is effectively lethal; a mosquito that cannot bite to draw blood cannot feed at all.
The Structure of Evolutionary Theory Page 174