Analysis of the Pax6/Eyeless gene has indicated that the mechanism of eye induction may be conserved across the animal kingdom. However, the dramatic variation of the eye structure not only between vertebrates and invertebrates, but also within the vertebrate lineage, has suggested that events downstream of eye induction may have evolved independently. [Page 1130] Our results show that the role played by Hh signalling in retinal differentiation is conserved between flies and fish. This suggests that Hh was already used to pattern a primordial eye structure before vertebrate and invertebrate eye lineages diverged, and thus supports a common evolutionary origin of the animal eye.
A question OF priority. This emerging story of Pax-6 homologies directly engages one of the classic conundrums of macroevolutionary theory, an issue that troubled Darwin himself, and that elicited a famous treatment — based, in a fascinating but not really surprising coincidence, upon the evolution of eyes! — in one of the Origin's most brilliant passages (in Chapter 6, entitled “Difficulties on Theory”): how can “organs of extreme perfection” ever arise if crucial components of the final product could not have functioned in their current manner in any conceivable ancestral form of simpler design? Darwin's general answer established the important evolutionary principle of cooptation: the component in question must have originally functioned in another, perhaps related, manner, and then been coopted for its current role (see pp. 1218–1224 for full treatment).
But this general solution then engendered a second problem of even broader import: how can a trend towards a highly complex organ ever get started at all, if the initial stages can bear so little structural or functional similarity to the final product? In this case, how could eyes ever form if the simplest incipient state in the founding member of a trend couldn't function for anything even roughly analogous to vision? (How, in other words, can evolution ever take the first step to a simple light-sensing organ, not to mention the much later development of image-forming devices?) How can evolution “know” where to start when faced with millions of potentially alterable molecules and processes, none manifesting even the first selected step of a forthcoming trend? (Natural selection may power the trend after step one has been reached, but how can this initial entrance be effected?)
To resolve this deeper problem, Darwin advanced the brilliant hypothesis — in the sense of a wonderfully simple idea once formulated, but quite nonobvious beforehand — that first steps must rely upon purely fortuitous variation, or fortuitous cooptability, in the favorable direction. Writing of Batesian mimicry in butterflies, for example, Darwin notes that the adaptive value of a tasty mimic to a noxious model cannot be gainsaid, but what, he then asks, can get the process started? Why, in particular, did the ancestor of the mimic choose this particular model among scores of other noxious species in the same fauna? Darwin answers that the first step must rely upon a slight fortuitous resemblance to one particular model — thus setting an initial (and accidental) tiny advantage that natural selection can “notice” and thenceforth enhance.
In a famous passage, Darwin uses this argument to defend the evolution of complex lens-eyes by natural selection:
To suppose that the eye, with all its inimitable contrivances . . . could have been formed by natural selection, seems, I freely confess, absurd in [Page 1131] the highest possible degree. Yet reason tells me, that if numerous gradations from a perfect and complex eye to one very imperfect and simple, each grade being useful to its possessor, can be shown to exist . . . then the difficulty of believing that a perfect and complex eye could be formed by natural selection, though insuperable by our imagination, can hardly be considered real (1859, pp. 186-187).
Expanding his discussion in later editions, Darwin specifies a potential starting point of maximal simplicity in structure and function: “The simplest organ which can be called an eye consists of an optic nerve, surrounded by pigment-cells and covered by translucent skin, but without any lens or other refractive body. We may, however, ... descend even a step lower and find aggregates of pigment-cells, apparently serving as organs of vision, without any nerves, and resting merely on sarcodic tissue. Eyes of the above simple nature are not capable of distinct vision, and serve only to distinguish light from darkness” (1872b, p. 135).
Therefore, following this theme, if we wish to develop a complete evolutionary explanation for the role of Pax-6 homologies in regulating the formation of complex structures in the lens-eyes of several phyla, we also need to understand its ancestral role in species with much simpler organs of vision, or without eyes at all. Why, in short, did Pax-6, rather than some other molecule, become the homologous “master control gene” of such complex structures, especially if the common ancestor of modern phyla with lens-eyes had only evolved eyes of much simpler form and function?
Fortunately, even at our current embryonic stage of research, some intriguing hints exist for a resolution, thus completing the intellectual structure of an evolutionary argument for important parallelism in the evolution of eyes, as regulated by the positive channel of Pax-6 homologies. Pax-6 homologs have been cloned from three cnidarian genera — from a jellyfish and a hydra (Sun et al., 1997, though questioned by Catmull et al., 1998). Catmull et al. speculate (1998, p. 355) that “the capture of a homeobox by an ancestral Pax gene probably permitted a transition from functions in cell-fate specification to roles in anterior patterning,” and later to still more specialized roles in the development of the central nervous system and finally in the specification of eyes. Because Acropora lacks eyes, despite showing sensitivity to light, Catmull et al. suspect that the Pax-6 homolog of this cnidarian may regulate anterior (and distal) patterning of the nervous system. (They also conjecture, on the same grounds, that the Pax-6 homolog in the blind nematode C. elegans may operate as a plesiomorphic regulator of the head region, rather than as a sign of heritage from an eyed ancestor.)
But Sheng et al. (1997), in an intriguing discovery that might link Pax-6 to an ancestral function tied more closely to vision, found that Drosophila Pax-6 directly regulates the expression of the visual pigment rhodopsin in photoreceptor cells. Sheng et al. (1997, p. 1122) therefore propose that “the evolutionarily ancient role of Pax-6 was to regulate structural genes (e.g. rhodopsin) in primitive photoreceptors, and only later did it expand its function [Page 1132] to regulate the morphogenesis of divergent and complex eye structures.” “Pax-6,” they continue (p. 1129), “is locked in the regulatory pathway of eye development because of its more ancient function in the direct regulation of terminal photoreception genes like rh. Later in evolution, genes specific for each type of eye may have been added to this regulatory pathway to specify divergent and complex eye structures.”
This appealing hypothesis, if validated, would address both issues in Darwin's dilemma, as described above, for the origin of organs of extreme complexity. First, Pax-6 would manifest a plesiomorphic function related to vision in much simpler ancestral structures that can detect light or motion, but do not form images — for rhodopsin operates as a major visual pigment in such organs. Second, and proceeding phylogenetically further back to a potential utility even before the origin of vision (analogous to the initial choice of a model in the evolution of mimicry), rhodopsin operates in sensitivity towards light in all three multicellular kingdoms — suggesting a symplesiomorphy of great phyletic depth! — even when the physiological basis of response cannot be meaningfully compared with vision in animals. Rhodopsin, for example, acts in phototaxis to guide the swimming of green algae towards or away from light. Moreover, Saranak and Foster (1997) show that rhodopsin also guides the zoospores of the fungus Allomyces reticulatus towards light — suggesting (Saranak and Foster, 1997, p. 465) “the origin of vision might have been the phototaxis of their unicellular ancestors.”
PARALLELISM IN THE SMALL: THE ORIGIN OF CRUSTACEAN FEEDING ORGANS. Although interphylum parallelisms, based on homologies of developmental pathways, may provide greater eclat for t
heir status as both utterly unanticipated in traditional Darwinian theory, and also a bit “weird” to boot, the greater importance and transformative power of this principle for the ordinary practice of evolutionary research will surely reside in the far more numerous and precisely defined cases of parallel evolution within much smaller monophyletic clades. In these instances, a parallel rather than convergent basis for similar adaptations does not provoke the same sort of surprise (for this alternative had always been plausible in theory for taxa of shared Bauplan and relatively recent common ancestry), but the value of parallelism becomes greatly increased by the operational basis thus granted to firm and testable explanations — by moving away from adaptationist scenarios in the largely speculative mode, and towards morphogenetic rules with specifiable, even predictable, realizations.
Ultimately, I suspect that the major reformatory significance in such accumulating examples of parallelism “in the small” will lie primarily in their capacity to resuscitate, and place upon center stage, the once derided formalist concept that taxonomic order largely represents the realized manifestations of more general developmental rules and pathways (“laws of form” in the archaic, but not entirely invalid, terminology of Geoffroy's biology), rather than the adaptive nuclei where environmental advantage reins in a much more promiscuous range of possibilities. (In this formalist or structuralist view, adaptation by natural selection surely sets the actual points of occupation [Page 1133] along potential pathways of realizable form, but basic taxonomic order reflects the limits and preferred channels of internal potential as much, or more, than the happenstances of immediate selective advantage.)
To cite just one impressive case of extensive parallelism in the taxonomic order of a substantial clade within a phylum, Averof and Patel (1997) have studied the action of the Hox genes Ubx and abdA in specifying the form of gnathal and thoracic appendages in Crustacea. In most arthropods, the gnathal appendages (maxillae) are specialized for feeding, and the larger thoracic appendages for locomotion. But in numerous and phyletically varied crustacean taxa, appendages of the anterior thoracic segments have been reduced in size and specialized for primary utility in feeding. These thoracic feeding limbs are called maxillipeds, and phyletic analysis clearly illustrates their multiple independent evolution within the Crustacea.
As a general, and presumably ancestral, pattern in Crustacea, the transition between gnathal and thoracic segments marks the anterior expression boundary for Ubx and abdA, and these genes do not operate in the gnathal region, where smaller feeding appendages (maxillae) develop. The branchiopods, for example, follow this basic scheme: Ubx and abdA are expressed throughout the thorax, and no maxillipeds form (Averof and Akam, 1995). These genes do not operate in the anterior gnathal segments, which develop the smaller maxillae. In other groups, the generation of maxillipeds on anterior thoracic segments correlates precisely with the suppression of Ubx and abdA in these segments alone, and these specialized appendages then grow to resemble the smaller maxillae of the adjacent anterior (gnathal) region of the AP axis, where Ubx and abdA are not expressed in normal development.
The precision of this correlation is impressive, and presumably causal. Among the malacostracans, for example, the leptostracans also develop no maxillipeds, and Ubx and abdA are expressed throughout the thorax. But in peracarids, the first, and sometimes the second, of eight thoracic appendages develop as maxillipeds. Averof and Patel (1997) document the suppression of both Ubx and abdA in these anterior thoracic segments with maxillipeds. In Mysidium colombiae, for example, Tl generates a maxilliped, whereas the appendage of T2 remains primarily a swimming organ, but develops gnathal features at its distal end. Averof and Patel found that Ubx and abdA are entirely repressed in Tl, but expressed in the proximal portion of the T2 endopod, while being excluded from the distal portion that acquires the gnathal features of a maxilliped.
The familiar decapods (lobsters, crabs, shrimp) generally bear eight thoracic segments, the anterior three with maxillipeds and the posterior five with walking legs (hence the name of the group, meaning 10-footed). This situation correlates perfectly with the suppression of Ubx and abdA in the first three thoracic segments by backward shifting of their joint anterior expression boundary. The finer scale variations within the clade also follow the same developmental rule. For example, although adult lobsters of the most familiar (and edible) Homarus americanus bear the usual five pairs of large thoracic limbs and three pairs of anterior maxillipeds, only T1 and T2 show [Page 1134] limb reduction at hatching, while the appendage of T3, at this early stage, continues to resemble a walking leg in size and form. Averof and Patel found that, at this intermediate point in development, Ubx and abdA are repressed only in T1 and T2. (The repression presumably extends to T3 during later molts, but Averof and Patel do not present data for these later stages.)
Maxillipeds also develop in several other crustacean groups, widely dispersed within the taxonomic space of the clade. In a non-exhaustive compilation, Averof and Patel found no exceptions to the rule that maxillipeds develop instead of walking legs when the anterior expression boundary of Ubx and abdA shifts back, thus suppressing the action of these Hox genes in a specified number of anterior thoracic segments. For example, Averof and Patel (1997) studied two copepod species with maxillipeds only on Tl. They found (unsurprisingly by now) that the anterior expression boundary of Ubx and abdA had shifted back only one segment, with activity beginning inT2.
The important evolutionary message of these findings follows from the clear implication, based on cladistic analysis, that maxillipeds have arisen several times, and independently, in crustacean phylogeny — but always, as Averof and Patel's data illustrate so impressively, under control of the same homologous developmental rule, presumably a plesiomorphic trait of the clade. Thus, this striking example of clearly adaptive, multiply repeated, and effectively identical, transformations of anterior thoracic walking legs to feeding appendages represents a striking case of parallel evolution based on frequent evocation of a homologous developmental pathway, and not a demonstration of convergent evolution rooted in similar pressures of natural selection acting upon unconstrained and “random” variation in each case.
Averof and Patel (1997, p. 686) affirm this interpretation, but unfortunately introduce some terminological confusion (albeit minor, and easily correctable) in summarizing their splendid study: “Our findings indicate that such convergent changes may have been achieved by similar developmental changes (involving similar posterior shifts in the expression boundary of Ubx-abdA) on several independent occasions. This suggests that, given a particular developmental system, there may be limited ways for achieving a particular morphological result.”
But these changes are fully parallel, and not convergent, in both developmental pathway and phenotypic result because maxillipeds arise by independent recruitment and expression of the same, homologically retained developmental rule among the taxa that independently evolve appendages of the same basic form and anatomical structure along a strongly positive and clearly adaptive internal channel of constraint. (For lens eyes of squid and vertebrates, on the other hand, homologous generators build similar structures from different tissues — thus making the eyes largely convergent as adult phenotypes and largely parallel in developmental architecture.)
PHARAONIC BRICKS AND CORINTHIAN COLUMNS. In both gastronomy and the academy, too much of a good thing can quickly pall. A concept, to [Page 1135] be useful and interesting, must make distinctions and define categories of exclusion. A rubric for all possible cases explains nothing — as Gilbert and Sullivan's Grand Inquisitor Don Alhambra explained to the naively egalitarian Gondolier Kings of Barataria: “When everyone is somebody, then no one's anybody.”
It is certainly understandable, and probably psychologically inevitable, that exciting discoveries tend to become overextended in the first flush of reformatory application. Our recently acquired ability to identify genetic ho
mologies in the developmental pathways of homoplastic final structures has sometimes engendered a misplaced enthusiasm for reinterpreting similarities previously ascribed to pure convergence as examples of parallel evolution (defined as homoplastic results based on homology of underlying generators).
But, as I have argued throughout this book, concepts only become interesting in contexts set by the logic and the history of theoretical issues thus addressed. The primary significance in recasting convergence as parallelism lies in the very different implications of the two processes both within Darwinian theory and for the larger question of the relative weights that should be assigned to internal structural constraint and functional adaptation in populating the morphospace of life's history. Pure convergence stands at a Darwinian functional extreme, where uncanny similarities of phyletically distant taxa arise from entirely different starting points, propelled by selective pressures alone, without any boost from internal channeling.
Parallelism, by contrast and with reversed evolutionary meaning, attributes the identical result, at least in large part, to a homologous generating channel that guides two independent sequences of selection down the same path from within.
The Structure of Evolutionary Theory Page 180
