Abouheif 1997, with permission from Elsevier Science.)

example, we can show that a gene such as the homeobox gene Distal-less is a do-all functional gene, expressed along the proximal-distal axes of a wide variety of developmental systems, including vertebrates and arthropod limbs, polychaete parapo-dia, and echinoderm tube feet (Lowe and Wray 1997; Panganiban et al. 1995, 1997).

In practice, the homology of phenotypic characters is rarely traced by systema-tists to the underlying genome (except in studies of molecular evolution). Tests for homology should involve the following criteria (Patterson 1982):

1. Two homologous character states may not exist within one organism. Different character states might exist among organisms of a single taxon or species.

2. Two character states in two organisms are of the same homologous character if they occupy the same topographic or ontogenetic position in the organism.

3. A series of character states in a series of corresponding taxa belong to one homologous character if the cladistic relationship among the taxa, defined by the character, does not contradict any genealogy defined by "truly homologous" characters.

The last two criteria require qualification and suggest other criteria. Criterion two implies that homologous characters can be "located" in different taxa. This ability is strengthened to the degree that (1) evolution produces unique phenotypic sites and (2) evolution is slow. If evolution is very rapid and not unique (i.e., convergence is common), then the ability to identify the phenotypic expression of corresponding parts of the genome is erased. Location can also involve a temporal aspect, especially because ontogenetic data can be applied to systematic problems (e.g., Alberch 1985; de Beer 1958; Nelson 1978).

A special set of instances may directly link homology and polarity. Developmental anomalies often reveal seemingly ancestral states. In rare instances, whales have complete limbs similar to those used in walking ancestors, despite the fact that millions of years have elapsed since the structures related to walking were presumably lost (Andrews 1921; Lande 1978). The atavistic appearance of long-lost structures in amazing detail (e.g., Andrews 1921; Kurten 1963; Marsh 1892) seems to discredit the belief that the loss of a structure implies the loss of the genes, as a naive version of an adaptive theory of evolutionary genetics would predict (Kollar and Fisher 1980). This suggests that for whatever reason, the genome is to a degree stable and a genetic basis for homology is possible. It also suggests another criterion for homology:

4. Two states in two organisms can represent states of a homologous character if a developmental anomaly in one taxon produces an individual with a state largely similar to the other taxon, in the same topographic position.

This criterion must be used judiciously, because the simplicity of a character state might result in the evolutionary convergence of states of a nonhomologous character. For example, if two states represented different colors of a butterfly wing of two respective species, the appearance of taxon B's color, as a variant of taxon A, cannot guarantee homology. Complexity of similarity is therefore an essential element of this criterion. Unfortunately, it is difficult to define a mathematical function that relates the probability of homology to increase of similarity. One of the bithorax complex phenotypes (see Ouweneel 1976) in Drosophila melanogaster mimics the presumed ancestral state of the Diptera (i.e., two pairs of wings). There is no reason to believe, however, that this is the particular genetic route backward to the ancestral state. Appeals to strong similarity of detail, therefore, are intuitively attractive but no more than that.

Patterson's (1982) third test of homology, compatibility with other homologous characters, raises both the fundamental strength and an important weakness of the Hennigian cladistic method. This criterion implies that homology is a hypothesis, rather than a proven statement of genealogical connection among character states. The hypothesis of homology for a given set of character states is therefore corroborated if the genealogical relationships defined by synapomorphies does not contradict others defined by other characters. Homologies of several discrete sets of character states are thus reinforced to the degree that the corresponding genealogical relationships defined by the discrete sets are compatible. But what if sets of states produce incongruent inferred genealogies? How do we decide among different trees defined by different characters, and how do incongruities affect our hypotheses of homology?

The first criterion of homology, that a character not be found in two different locations on the same creature, implies that we are excluding serial homology from our discussions. Many structures - genes for example - arose in evolution by duplication. In the first descendant taxon with a duplication, there is an ambiguity in that two structures are homologous with one belonging to the ancestor. Subsequently, the ontological ambiguity of homology disappears, but there is still an epistemolog-ical confusion in the status of the relationships between the repeated structures within the same individual (e.g., two tandem genes that arose by duplication that now serve different functions). This confusion is heightened when the duplications affect the same phenotype in different ontological stages. Thus, Drosophila melanogaster has larval and adult alcohol dehydrogenases that presumably serve the same function but the same structural gene codes for both enzymes; the difference is an upstream sequence. In this case, the distinction between serial homology and evolutionary homology breaks down, without an ontogenetic criterion.

If our objective is genealogical reconstruction alone, then it is not clear whether the genotype-phenotype distinction is all that important. Characters are characters, and homologies can be established - indeed they have been established for hundreds of years - without the benefit of knowing the genetic underpinnings. Genetic data, such as nucleotide sequences, are also sources of homology, as long as some sort of criterion of location can be employed. One must be sure, for example, that one is following the same gene through a genealogy if the sequence is to mean anything. The connection between genes and phenotype becomes important when one is interested in tracing given characters in clades, particularly with regard to evolutionary mechanism. If one believes in genetic constraints, then the DNA history is as important as a phenotypic history. This connection is most crucial if we are ever to understand the relative contributions of developmental, genetic, and functional constraints to phenotypic evolution.

0 0

Post a comment