Since bicalyx presumably establishes its large and homeotic effect through the developmental channel of the ABC system or some homolog (see previous discussion on pp. 1063–1065), this case also illustrates the common linkage between internal channels as positive constraints and potential speed of phyletic movement down the channel — as Galton first proposed in his model of the polyhedron (see pp. 342–351). In a zoological example of the same linkage, Fitch (1997, p. 166) documents a “topological constraint” limiting the number and positions, and channeling the potential directions of evolutionary change, for rays in the male tail of the nematode C. elegans. Not only are the directions of potential change both limited and most plausibly attained by saltation, but the transitions can also be generated by single point mutations. Of this interesting correlation between constraint and saltation, Fitch concludes (1997, pp. 166-167): “Because single genetic changes can be postulated for some of the evolutionary change in the male tail, I predict that many evolutionary changes in morphology will have resulted mainly from changes at single loci . . . Because the power of selection is limited by variation, such developmental constraints could cause significant bias in the evolution of form.”
Movement four, Recapitulation and Summary: Early rules and the
inhomogeneous population of morphospace: Dobzhansky's
landscape as primarily structural and historical, not functional
and immediate.
BILATERIAN HISTORY AS TOP-DOWN BY TINKERING OF AN INITIAL SET OF RULES, NOT BOTTOM-UP BY ADDING INCREMENTS OF COMPLEXITY. As a common pattern in the history of science, great and unexpected theoretical discoveries often elicit fairly conservative theoretical interpretations [Page 1148] at first — if only because most of us can only imagine so much novelty at once. As discussed before in different contexts, when Ed Lewis (1978) recognized the Hox genes as linearly ordered in both space and time, and inferred their origin by tandem duplication from a precursor, he interpreted the evolutionary significance of these amplifications in an “obvious” and conventional manner (the proper “first pass” procedure in science, and therefore recalled here without critical intent, while noting that our later explanatory reversal only underscores the importance of Lewis's discovery). Lewis proposed that the addition of duplicated Hox genes could be directly and causally correlated with the specialization and differentiation of appendages along the arthropod AP axis, as an originally homonomous ancestor evolved into the diverse Bauplan of major arthropod taxa. In particular, Lewis argued that an addition of Hox genes allowed evolving members of the insect line to suppress the growth of legs on the abdominal segments, and (in Diptera) to convert the wings of the second thoracic segment to halteres. Lewis's original scenario matches our conventional view of evolution, and of complex systems in general — particularly in the assumption that a history of increasing elaboration in overt products (the phenotypes of complex bilaterian phyla) should be underlain by a growth in the number and intricacy of generating factors (genes regulating developmental outcomes).
In one of the most important early discoveries of evo-devo, this entirely reasonable scenario has been overturned and, in large measure, reversed. Two strong sources of evidence now indicate that a full complement of Hox genes had already evolved in the presumably homonomous common ancestor, not only of the protostome phyla, but of the entire bilaterian line (thus further exemplifying the homologies of arthropods and vertebrates). The multiple and independent evolution from homonomy towards complexly specialized and differentiated Bauplan in several phyla did not entail any increase in the number of Hox genes, but rather a regionalization and decrease in the range of action of individual genes and, especially, changes in both the regulation and content of the downstream cascades differently engaged by the various Hox genes.
1. Modern homonomous organisms share the full complement of Hox genes with closest relatives among classically differentiated invertebrates. The genome of myriapods, the homonomous sister group of insects for example, includes a full set of insect Hox homologs (Raff, 1996). At the higher level of a sister group to the entire arthropod phylum, the undoubted “standard” for highly differentiated body plans along the AP axis, the genome of the homonomous Onychophora also includes all insect Hox genes, as well as an ortholog of the pair-rule gene fushi tarazu (Grenier et al., 1997). (Modern onychophores include only a few Gondwana species, restricted to moist terrestrial habitats. The generic name of the most famous modern form, Peripatus, honors the homonomy of the numerous lobopods, the pair of leglike structures on each segment. But the Onychophora included a prominent and diverse group of marine representatives in the earliest faunas of the Cambrian period.) Grenier et al. (1997, p. 549) conclude that “the segmental [Page 1149] diversity of arthropods evolved without an increase in Hox gene number. The evolution of arthropod segmental diversity must therefore have involved regulatory changes in Hox genes and/or their targets.”
2. Phylogenetic reconstruction affirms a full Hox complement for the bilaterian common ancestor, with restriction and occasional unemployment far more prominent than addition in subsequent evolution. The analysis of De Rosa et al. (1999) indicates at least 8 Hox genes for the protostome common ancestor, and at least 7 for the bilaterian progenitor (see Fig. 10-24). Moreover, comparisons at greater detail support the growing consensus (see Halanych et al., 1995; and Aguinaldo et al., 1997, based on 18S ribosomal RNA) that the protostome phyla split into two great genealogical groups, the ecdysozoans or molting phyla (including arthropods, nematodes and priapulids), and the lophotrochozoans (including annelids, mollusks, brachiopods, platyhelminths, and nemerteans). The ecdysozoan genome includes the posterior Hox gene Abd-B, whereas lophotochozoans have two “Abd-B-like”
10-24. By this analysis and cladogram, the protostome common ancestor must have possessed at least eight Hox genes, whereas the bilaterian progenitor must have had seven. From De Rosa et al., 1999.
[Page 1150]
genes named Post1 and Post2. In addition, ecdysozoan genomes include Ubx in the central cluster, whereas lophotrochozonas share the different, but closely related, Lox2.
Although De Rosa et al. (1999) cite 7 as a minimum for the common bilaterian ancestor (lab/Hox1, pb/Hox2, Hox3, Dfd/Hox4, Scr/Hox5, one additional central gene and one posterior gene), Figure 10-24 also indicates at least 10 shared Hox genes for the stem lophotrochozoan and 8 for the stem ecdysozoan. Using a more generous estimate based on a hypothesis that “most or all of the Hox genes that are present in extant bilaterians may have been present in the common ancestor, but that some orthology relationships have become obscured” (De Rosa et al., 1999, p. 775), the protostome common ancestor might have possessed ten Hox genes (the 7 listed above plus 2 central and 1 posterior), or even more if the deuterostome situation of multiple posterior Hox genes is primitive rather than derived. In any case, either a minimum of 7 or a maximum of 10 or more provides ample support for the key conclusion that a full Hox complex had already evolved before the establishment of distinctive features of the major bilaterian Bauplan. De Rosa et al. (1999, p. 775) conclude their article by stating: “The subsequent bilaterian history of Hox genes would have been primarily one of functional divergence and gene loss, rather than gene duplication. Regardless of the exact number of Hox genes in the bilaterian ancestor, the major period of progressive expansion of the Hox cluster due to tandem duplication events predated the radiation that generated the bilaterian crown phyla, concurrent with radical evolutionary changes in body architecture and development.”
As a fascinating footnote to the rich phyletic information contained in the conservation of Hox genes, the Mesozoa have long presented a deep puzzle in the study of animal phylogeny. These creatures lack body cavities and effectively all the characteristic organs of animals, including a gut or a nervous system. Their maximally simplified development even proceeds without gastrulation or the differentiation of germ layers. Many zoologists have therefore
considered their organization as primitive, and have even regarded the Mesozoa as a surviving key to the phyletic transition between unicellularity and the evolution of truly multicellular organization with differentiation of tissues and organs — hence their name, Mesozoa (from the Greek meso, meaning “middle”), as a potential intermediate between the protistans, formerly called Protozoa, and the true Metazoa. But the mesozoans are parasites of metazoans, and parasites often become extremely simplified in phenotype. Thus, the opposite interpretation of descent from an ordinary and complex metazoan ancestor has remained entirely plausible. Unfortunately, the highly simplified and autapomorphic anatomy of mesozoans has provided no clues about ancestry, despite more than a century of extensive study.
But Kobayashi et al. (1999) have isolated a Hox gene, DoxC, from a dicyemid mesozoan (parasites of cephalopod renal sacs). PCR analysis shows that DoxC is an ortholog of the “middle group” Hox series. The middle group Hox genes have only been found in triploblasts, and do not exist in Cnidaria. Hence, these data would seem to validate the hypothesis that mesozoans [Page 1151] are secondarily simplified bilaterians, and not the sole survivors of the vaunted intermediary group between protists and metazoans. Moreover, further study of DoxC affirms its orthology with the lophotrochozoan LoxS, rather than with a middle Hox gene of the ecdysozoan clade. Therefore, not only are the mesozoans revealed as simplified metazoan parasites, but we may also place the ancestry of these formerly enigmatic forms after the major split in protostome phylogeny, and into one of the two great groups as a relative of platyhelminths, brachiopods, and annelids, rather than arthropods or onychophores.
These revolutionary discoveries have inspired a growing literature on the hypothetical phenotype, or at least the shared developmental architecture, of a stem bilaterian, or even a stem animal. Slack et al. (1993) tried to define a “zootype” as the “defining character, or synapomorphy” of the kingdom Animalia (p. 491), with maximal expression in ontogeny at a “phylotopic stage ... at which all major body parts are represented in their final positions as undifferentiated cell condensations ... or the stage at which all members of the phylum show the maximum degree of similarity” (loc. cit.). They base this concept on common possession of “a system of gene expression patterns, comprising the Hox cluster type genes and some others [encoding] relative position in all animals” (loc. cit.). As discussed just below, such a concept may apply to all triploblasts, but probably not to diploblasts (whose Hox homologs show some common properties of individual action, but not the integrated spatial and temporal colinearity found in most triploblasts studied so far).
The less ambitious attempt to define the phenotype and organization of a common bilaterian ancestor (named Urbilateria by de Robertis and Sasai, 1996; see also Kimmel, 1996; de Robertis, 1997; and Pennisi and Roush, 1997) may be more tractable, but specific arguments about whether this common ancestor had already developed recognizably modern versions of segments (Kimmel, 1996), antennae, photoreceptors, or a heart (see Fig. 10-25 for a cartoon of alternative possibilities from Pennisi and Roush, 1997) — with obvious implications for views about the homology of adult phenotypes, beyond the already established homology of underlying generators — are, in my opinion, premature.
For example, in a challenging proposal, Arendt et al., 2001, propose a homologous origin for tube-shaped guts of primary larvae in both protostomes and deuterostromes — structures long granted an independent origin in conventional evolutionary thinking. They base their argument upon “the shared, and very specific, expression of brachyury in ventral developing foreguts of the starfish bipinnaria, echinoid pluteus, enteropneust tornaria and polychaete trochophore larva,” suggesting “common ancestry (homology) of larval foreguts in Protostomia and Deuterostomia, despite the different developmental origin of these structures” (p. 84). The authors then claim additional support from equally specific actions of two other developmental genes — a common expression of goosecoid in the foreguts of various bilaterians, and of otx along pre- and postoral ciliary bands. As an obvious point of contention, [Page 1152] these putative homologies, if expressive of common ancestry, require that primary larvae be regarded as the basal anatomical form of bilaterian animals — a hypothesis supported by some (Peterson and Davidson, 2000, for example), but vigorously rejected by others (Valentine and Collins, 2000, for example).
We cannot yet determine — if genetic and developmental data of modern organisms could allow us, in principle, to resolve such questions at all — whether a hypothetical urbilaterian already possessed highly developed phenotypic structures (in either larvae or adults) acting like the Corinthian columns of my metaphor on pages 1134–1142, or whether this common ancestor had only expressed its extensive genetic and developmental pathways, preserved forever after as homologies in all bilaterian phyla, as phenotypic Pharaonic bricks of limited specification and extensive flexibility. The solution to such puzzles requires paleontological data (not yet available, but eminently attainable in principle). In any case, I regard this issue as a largely speculative sidelight that does not affect — and must not lead us to forget or put aside — the striking reformulation of evolutionary theory implied by the well-documented genetic and developmental homologies alone. De Robertis expresses this key argument in the final line of his 1997 article on the ancestry of segmentation: “The realization that all Bilateria are derived from a complex ancestor represents a major change in evolutionary thinking, suggesting that the constraints imposed by the previous history of species played a greater role in the outcome of animal evolution than anyone would have predicted until recently.”
The Hox genes of diploblasts apparently do not show the intergenic organization of colinearity that defines the key developmental homology of bilaterians, but cnidarians do possess Hox genes with some commonality of action to their bilaterian homologs. Martinez et al. (1998) found four Hox
10-25. A cartoon from Pennisi and Roush, 1997, on our minimal and maximal homologies with an ancestral bilaterian. The maximal version is segmented, and has evolved a prototypical heart, eyes, a mouth, and antennae.
[Page 1153]
genes among various Cnidaria, corresponding to one medial group precursor, the anterior Drosophila genes labial and proboscipedia, and the posterior Abdominal-B (see also Miller and Miles, 1993). But even though these cnidarian genes follow the same 3' to 5' order as their bilaterian homologs, no evidence exists for colinearity of action in the development of any body structure (particularly the oralaboral axis). Moreover, cnidarians apparently lack several key bilaterian Hox elements. Martinez et al. (1998, p. 748) write that “the genes in the middle of the [bilaterian] Hox clusters form a monophyletic group that includes no cnidarian genes. This is most readily explained by derivation of these genes through duplication of a single precursor after the origin of Cnidaria.”
A fascinating study by Cartwright et al. (1999), however, does affirm some general similarity of Hox action in cnidarians and bilaterians by demonstrating a formative role for cnidarian Hox in specification of the oral-aboral axis in two distantly related hydrozoans, Hydra itself and Hydractinia symbiolongicarpus. (This single cnidarian Hox gene, Cnox-2, specifies full differentiation along the oral-aboral axis of polyps, whereas the sequential colinear activation of the full bilaterian Hox suite specifies differentiation along the bilaterian AP axis — thus illustrating once again that the primary novelty of bilaterian origins resides in the spatial sequence of Hox genes and the evolution of their coordinated action. We have, in any case, no reason to view the cnidarian oral-aboral axis as homologous to the AP axis of bilaterians.)
The study of Cartwright et al. gains strength from the multiple possibilities for natural and laboratory experiments inherent in the fourfold polymorphism of Hydractinia polyps, and in the ease of experimental transformation of one type into another. The “normal” feeding polyp of Hydractinia, the gastrozooid, corresponds to that of Hydra, and shows full oral-aboral
differentiation from the distal mouth and hypostome to the body column and foot at the proximal end. In both Hydra and Hydractinia, Cnox-2 is expressed at high levels in the foot and body column and at successively lower levels up the axis towards the head, which shows very weak Cnox-2 expression.
But Hydractinia symbiolongicarpus also develops three polymorphic variants; clearly interpretable as intensifications of either the oral or aboral ends of the main polypary axis (see Fig. 10-26). Gonozooids and dactylozooids are specialized, respectively, for sexual reproduction and for capturing eggs of the colony's hermit crab host. Both lack a hypostome and tentacles and seem to represent “an expansion of the body column to the exclusion of oral regions” (Cartwright et al., 1999, p. 2183). The authors found “no detectable difference in Cnox-2 expression along the aboral-oral axis in either the gonozooid or dactylozooid” (p. 2185), and general levels of expression equaled those found at the base of the gastrozooid — thus affirming the anatomical inference that both polymorphs develop by extending the specialized aboral end of the axis to the full length of the polyp, and suppressing the head region entirely.
In satisfying contrast, a fourth polymorph, the tentaculozooid that plays a role in defending the colony, resembles a single gastrozooid tentacle, and [Page 1154] therefore appears to represent “all head” — an expanded part of the oral end only, with the aboral end suppressed. Cartwright et al. (1999) found very weak expression of Cnox-2 throughout the full length of tentaculozooids “at approximately the same level seen in the tentacles of gastrozooids” (p. 2185).
In an additional affirmation by experimental manipulation, dactylozooids can be removed from the colony and induced thereby to transform into gastrozooids. Cnox-2 expression initially decreased in the developing hypostome of the transforming polyp, and then in the tentacle region, but not at the aboral end, “until ultimately, dactylozooids that fully transformed into gastrozooids displayed aboral-oral Cnox-2 expression patterns indistinguishable from that of normal gastrozooids” (p. 2185).
The Structure of Evolutionary Theory Page 183