Homology and causes

A critique of homology as a morphological concept

Two sequences of nucleotides are homologous if they are descended through a chain of replication from a common precursor molecule. Since organs are not copies or transcriptions of organs, the concept of morphological homology has no such simple and unambiguous definition. The theoretical vagueness of morphological homology is reflected in its many and inconsistent criteria of identification. Structures may be conventionally deemed homologous even though they are radically dissimilar in form, relationships, or function, or develop via dissimilar ontogenetic processes, or originate from nonhomologous embryological precursors. Hypothesis of homology are conventionally rejected when they are contradicted by known patterns of phylogenetic relationships, even if the structures in question are minutely similar in their form, function, and development. The dependence of interspecific homology on phylogeny is often expressed by saying that two structures are homologous if they are inherited from corresponding structures in a common ancestor. However, this is a circular definition (what counts as a "corresponding" structure is itself a question of homology), and it falsely assumes that structures can be inherited. At bottom, homology is an essentialist concept; two things are homologous only if they are in some essential sense the "same" thing and properly called by the same word. The concept can be made intelligible in an evolutionary context only by giving it a cladistic interpretation that makes homology judgments dependent on the outcome of a phylogenetic analysis. It follows that such judgments cannot play a role in evaluating conflicting phylogenetic hypotheses.

[Limbs of tetrapods--an actual problem of evolution]

The tetrapod limb is among the most intensively studied developmental systems in tetrapods. Earlier models based on a symmetrical determination of the early limb bud were recently questioned by new observations documenting a predominance of the postaxial axis during limb development. These findings are at variance with observations on limb reduction in a number of fossil reptile clades. Recent investigations corroborate the hypothesis of a decoupling of the patterns and sequence of chondrification and ossification in limb development. Ontogenetic repatterning may establish new topological patterns of connectivity among constituent elements in a biological structure as cartilage is replaced by bone. To view chondrification and subsequent ossification of cartilaginous precursors as decoupled processes bears directly on conjectures of homology of limb elements, and also requires the distinction of two separate mechanisms of limb reduction: truncation of morphogenesis and failure of ossification.

Euclidean distance matrix analysis: a coordinate-free approach for comparing biological shapes using landmark data

For problems of classification and comparison in biological research, the primary focus is on the similarity of forms. A biological form can be conveniently defined as consisting of size and shape. Several approaches for comparing biological shapes using landmark data are available. Lele (1991a) critically discusses these approaches and proposes a new method based on the Euclidean distance matrix representation of the form of an object. The purpose of this paper is to extend this new methodology to the comparison of groups of objects. We develop the statistical versions of various concepts introduced by Lele (1991a) and use them for developing statistical procedures for testing the hypothesis of shape difference between biological forms. We illustrate the use of this method by studying morphological differences between normal children and those affected with Crouzon and Apert syndromes and craniofacial sexual dimorphism in Cebus apella.

Molecular homology and DNA hybridization

We reviewed the concept of homology, which can broadly be defined as a correspondence between characteristics that is caused by continuity of information (Van Valen 1982). The concept applies widely in molecular biology when correspondence is taken to mean a genetic relationship resulting from a unique heritable modification of a feature at some previous point in time. Such correspondence can be established for features within a single organism as well as between organisms, making paralogy a valid form of molecular homology under this definition. Molecular homology can be recognized at a variety of organizational levels, which are interdependent. For example, the recognition of homology at the site level involves a statement of homology at the sequence level, and vice versa. This hierarchy, the potential for nonhomologous identity at the site level, and such processes as sequence transposition combine to yield a molecular equivalent to complex structural homology at the anatomical level. As a result, statements of homology between heritable units can involve a valid sense of percent homology. We analyzed DNA hybridization with respect to the problems of recognizing homology and using it in phylogenetic inference. Under a model requiring continuous divergence among compared sequences, DNA hybridization distances embed evolutionary hierarchy, and groups inferred using pairwise methods of tree reconstruction are based on underlying patterns of apomorphic homology. Thus, symple-siomorphic homology will not confound DNA hybridization phylogenies. However, nonhomologous identities that act like apomorphic homologies can lead to inaccurate reconstructions. The main difference between methods of phylogenetic analysis of DNA sequences is that parsimony methods permit hypotheses of nonhomology, whereas distance methods do not.

A multivariate phenetic approach to neuronal nuclei resemblances across species, with examples from three regions of cerebral cortex of a mammal, a bird and a reptile

In order to test the assumed homology between three examples of cortical cerebral architecture corresponding to three species of mammals, birds and reptiles respectively, a method arising from the quantitative standpoint of the concept of homology has been utilized. For this purpose, data corresponding to randomly selected neurons were analysed by means of multivariate techniques such as principal component, principal co-ordinates and hierarchical cluster analyses since on each neuron a set of nineteen variables, chosen as a definition of its dendritic morphology, was measured. This method has made it possible both to discern the neuron forms which are responsible for the discrimination between the three cortical nuclei studied and to show by setting up a discriminant index, the degree of overall similarity between such nuclei. These two findings provide evidence against both the fibrous connection criterion with which some modern neurobiologists ascertain homologs and the categorical concept of homology.