The root of the homology concept is the fact that individuals from different species often are composed of the same kind of building blocks, also called organs or characters. Therefore the starting point of any discussion of the homology concept has to be the idea of structurally identical parts. But what do we mean by structural identity?
For the present context the clearest concept of structural identity has been proposed recently by Mary McKitrick (1994), in a paper on the homology of bird hind-limb muscles. Consider two species with, say, characters A and B in species one and A' and B' in species two (Fig. 1). How do we decide whether A corresponds to A' or B'? McKitrick suggests a parsimony approach to this problem: character A corresponds to A' rather then to B' if it takes fewer steps to "transform" A into A' than it takes to transform A into B'. This is indeed a very good explanation of what we do as biologists comparing two species. We consider the wing of a bat to correspond to the foreleg of a cow because it would presumably take fewer steps to (mentally) transform the one into the other than it would take to transform the wing of the bat into the sternum of the cow. So we reject the possibility that the wing of the bat corresponds to the sternum of the cow but rather is comparable to the foreleg of the cow.
Figure 1:
The parsimony principle of character identification after McKitrick
(1994) states that the character A in species I corresponds to character A'
in species II rather than B' if it takes "fewer steps to transform A into A'
than it takes to transform A into B'."
The reason why this approach of recognizing structurally identical parts is so well suited for a discussion of the biological role of homologues is that it contains an implicit mechanistic assumption. If we recognize structurally identical parts between two species by "counting the number of steps" it takes to transform the one into the other, we implicitly make the assumption that these "transformation steps" are more likely to occur as natural variations than others. The parsimony principle of character identification is an implicit statement about the constraints and opportunities of evolutionary transformation. Furthermore, it assumes that the structures compared are modular units of evolutionary transformation and not just a dependent feature of some other parts of the phenotype.
This approach is naturally linked with another attempt to understand the biological significance of the homology concept. In a recent paper Louise Roth (1991) compared the formal properties of the units of selection, as defined by Lewontin (1970), with the recognized properties of homologues, such as conservation of the basic pattern with variation and individuality. A similar comparison between the gene concept and the character concept also recognized strong similarities between them (Stearns, 1992, pp14-16). Both comparisons lead to the suggestion that characters and homologues are modular units of phenotypic evolution and should be understood in the context of the mechanistic processes causing the evolution of phenotypic traits (Wagner, 1995). In the taxonomic context a recent attempt to identify the units of transformation based on species comparison has been proposed by Mary Mickevich (this volume).
It is thus suggested that morphological characters may be understood as the phenotypic units of evolutionary transformation. This suggestion also naturally relates to the origin of the homology concept in comparative anatomy and taxonomy. The heuristic purpose served by the homology concept is to make the description of variation easier by identifying units that can be compared more or less independently of the rest of the body. The fact that texts of comparative anatomy can be organized either taxonomically or around organ systems illustrates the point. One can explain the comparative anatomy of the kidney to a large extent independently of the comparative anatomy of limbs and fins. This is possible only if the described units are also units of interspecific variation, which is the same as a modular unit of evolutionary transformation.