1. Introduction
The term ‘homologue’ was first formally defined as “the same organ in different animals under every variety of form and function” (Owen 1843, cited in Donoghue Reference Donoghue, Keller and Lloyd1992 and others), although the concept and variants thereof can be detected in much earlier Western thought, from Plato and Aristotle onward (Panchen Reference Panchen and Hall1994). There has since been considerable debate over the exact meaning of the terms ‘same’, ‘organ’, ‘form’, and ‘function’ in the formulation of homology concepts by biologists in different fields, but these concepts remain among the most important theoretical tools in the biological sciences.
Many continue to take a monistic approach to homology, searching for a single homology concept that is applicable to all disciplines within biology (Patterson Reference Patterson, Joysey and Friday1982; van Valen Reference van Valen1982; Roth Reference Roth and Humphries1988; Bock Reference Bock1989; Wagner Reference Wagner1989; Haszprunar Reference Haszprunar1992; Young Reference Young1993; Nelson Reference Nelson and Hall1994; Abouheif Reference Abouheif1997; Brigandt Reference Brigandt2002). That this search is ongoing evinces its lack of success to date, and has led some authors to suggest that the search for homology concepts and criteria should be abandoned entirely in favor of more urgent and relevant biological questions (Wake Reference Wake, Bock and Cardew1999). This is shortsighted, however, given that the study of similarity is fundamental to biological inquiry. The list of authors who advocate a pluralistic approach to homology is substantially and strikingly shorter (Striedter and Northcutt Reference Striedter and Northcutt1991; Sluys Reference Sluys1996; Minelli Reference Minelli1998), but also interesting in that some previous advocates of monistic approaches to homology have recently begun to suspect that pluralism may be more appropriate (Wagner Reference Wagner2000; Brigandt Reference Brigandt2003).
The goal of the present project is to argue for epistemological pluralism in understanding and describing homology. I examine the reasons for the continued failure of the monistic approach by studying three widely used accounts, and analyze a biological case study that demonstrates both the inherent pluralism of biology and the complex nature of homology. I argue that multiple concepts of homology, each explicitly defined as an integral part of the theoretical and intellectual framework of the question at hand, function as useful tools for biological research. Further, I suggest that such an approach is more compatible with the fundamentally pluralistic nature of biology (Kitcher Reference Kitcher1987; Dupré Reference Dupré1993; Mitchell Reference Mitchell2003).
2. Failure of the Monistic Approach
The term ‘homology concept’, as applied herein, is defined as a theoretical tool for the study of biological similarity, which forms an intrinsic part of the intellectual framework in which it is situated. Complete homology concepts must include a definition of what constitutes homology, criteria by which we are to recognize putative homologues, and causes for the phenomena observed. As previously stated, there have been an enormous number of homology concepts proposed, and these can be broadly grouped into phylogenetic, developmental, and hierarchical accounts. Examination of each of these accounts reveals that they are individually insufficient to explain all types of homology, although often very useful in their specific domains.
2.1. The Phylogenetic Account
This account is exemplified by the work of Colin Patterson, who defines homology as “the relation which characterizes monophyletic groups” (Reference Patterson, Joysey and Friday1982, 21). The corollary of this is that ‘homology’ and ‘synapomorphy’ are synonymous terms, an idea espoused by a number of authors (Lankester Reference Lankester1870; Bock Reference Bock1989; Nelson Reference Nelson and Hall1994). Patterson proposes a series of three tests to verify the robustness of a putative homology: the conjunction test; the similarity test; and the congruence test, each of which is discussed in detail below. This concept of homology underpins the cladistic approach to systematics (Kitching et al. Reference Kitching, Forey, Humphries and Williams1998), currently the most widely accepted method for phylogeny reconstruction and biological classification. The popularity of the method and the many well-supported phylogenies erected through its use speak to the utility of Patterson's concept in the field of phylogenetic systematics.
Patterson's concept begins to break down when extended beyond its home domain. The conjunction test, which precludes the existence of two or more homologues in the same organism, would suggest that structures such as vertebrae or insect body segments are not homologous. Intuitively these structures are similar, and it has been shown that they are patterned by duplicated genes in the homeobox family (Ruddle et al. Reference Ruddle, Bartels, Bentley, Kappen, Murtha and Pendleton1994). Thus vertebrae or insect body segments are the products of the same genes and developmental trajectories. Further, such duplication of genes leads to failure of the conjunction test at the genetic level. Patterson's concept fails to account for these phenomena.
Patterson's test of similarity requires that putative homologues be similar in location, ontogenetic origin, and composition. It has been shown, however, that many structures considered homologous arise via different ontogenetic mechanisms, such as the anuran (frog and toad) eye lens, which is discussed further below. Patterson's concept fails to account for different combinations of genetic and epigenetic phenomena that produce putatively homologous end products.
The test of congruence with other homologues is both useful and circular at the same time. While an abundance of homologues that all suggest the same phylogeny do strengthen each other, conflicting hypotheses of homology force a decision to be made about which should be retained and which discarded as homoplasies. This decision rests on preexisting hypotheses of similarity and relationship. Further, phylogenies are generated using synapomorphies, and if these are the same as the ones being tested, then the test is not independent of the data and is meaningless. The combination of tests for similarity and congruence in Patterson's account led Donoghue (Reference Donoghue, Keller and Lloyd1992) to question whether similarity or common ancestry should have primacy in generating hypotheses of homology, a source of confusion that is further detrimental to Patterson's concept.
2.2. The Developmental Account
This account is exemplified by the work of Günther Wagner, who defines homologous structures as those that “share a set of developmental constraints, caused by locally acting self-regulatory mechanisms of organ differentiation. These structures are thus developmentally individualized parts of the phenotype” (Reference Wagner1989, 62). In this concept, a putative homologue must be distinguishable by the same norms of reaction in all taxa that share the homologue. These norms of reaction, which cause the structure to respond in the same way to environmental stimuli during development, are the cause of the homology.
This account is an improvement over Patterson's, because it allows for serial homologues such as vertebrae and insect body segments, and for sexually differentiated homologues such as testis and ovary. Wagner's account also allows for the effects of evolution. Ideas very similar to these have been suggested by Roth (Reference Roth and Humphries1988, Reference Roth and Hall1994), and both authors rely heavily on the idea that homology represents continuity of information (van Valen Reference van Valen1982). Both authors, in contrast to Patterson, emphasize causality in their homology concepts.
This account is also problematic, however, when invoked to explain instances of homology outside its native domain. How do we treat structures such as a lizard's original tail and its regenerated tail (incorporating a cartilaginous rod rather than vertebrae), when these arise via very different developmental pathways and are subject to very different developmental constraints? Wagner's original concept fails to account for this phenomenon. Wagner and Misof (Reference Wagner and Misof1993) have subsequently tackled this shortcoming with the suggestion that constraints leading to the development of homologous structures result from epigenetic interactions occurring later in development, and that early developmental variation has little effect on the outcome. This, however, requires at least partial decoupling of developmental mechanisms, and the ability of certain sets of mechanisms to completely override others, a requirement whose plausibility Wagner and Misof themselves question.
Further, at what point do we consider a structure to be sufficiently individualized, i.e., at the cell level, at the organ level, etc.? Wagner's account fails to sufficiently explicate the criteria by which homologues are recognized.
2.3. The Hierarchical Account
A third account is exemplified by the work of Ehab Abouheif (Reference Abouheif1997), who proposes that homology is best tackled in a hierarchical manner. Abouheif suggests that conflicting ideas of similarity, such as structural, developmental, or genetic, be “analyzed simultaneously within a phylogenetic framework” (Reference Abouheif1997, 405). Hypotheses of homology are generated by using appropriate similarity criteria, such as those of the previous accounts discussed here, for each hierarchical level being studied. These putative homologues are then mapped onto a phylogeny along with other traits, and those that covary with the phylogeny can be considered verified homologues. The success of this approach depends upon the quality of the phylogeny used, and upon the ability to discover putative homologues at each of the proposed hierarchical levels (Abouheif Reference Abouheif1997).
The hierarchical approach attempts to gather data from a variety of levels and unite them to discover homologous traits, instead of suggesting, as have the previous two accounts (Patterson Reference Patterson, Joysey and Friday1982; Wagner Reference Wagner1989; but see Wagner Reference Wagner2000), that certain levels should be given primacy over others. Despite this, however, the goal continues to be one overarching homology hypothesis that encompasses all levels based upon phylogenetics. Reliance upon a phylogeny, itself generated by criteria of similarity, causes this account to suffer from the same circularity and nonindependence problems inherent in Patterson's work.
Most importantly, it has been shown repeatedly that a strictly hierarchical view of biological processes is flawed (Minelli Reference Minelli1998). Factors acting outside the organism itself, such as environmental effects on phenotype, can affect the outcome of genetic and developmental processes. Moreover, other factors acting at various levels of the hierarchy, such as gene duplication, autonomous versus nonautonomous gene expression, and ontogenetic repatterning, can disrupt hierarchical behavior of biological systems (Minelli Reference Minelli1998). There are many factors that upset the linear mapping of genes onto developmental and morphological traits. This account, therefore, will fail more often that not due to the uncoupling of some levels from others.
Although the representative accounts outlined here are very useful in attacking certain aspects of the problem, all three strive, implicitly or explicitly, to provide a definitive, superior account of homology, despite Brigandt's (Reference Brigandt2002) acknowledgement that such an account may be unknowable or nonexistent. Careful scrutiny of this monistic approach reveals that while these accounts are functional in the domains in which they were developed, attempts to extend their scope beyond their original bailiwicks are highly problematic. I argue that this failure is symptomatic of the pluralistic nature of biology, and that rather than searching for an overarching concept, these individual concepts should be retained for study in the domain to which they belong. Further, use of multiple homology concepts will aid in achieving a better understanding of variably interconnected biological systems.
3. Biological Pluralism and Homology
The pluralistic approach to homology advocated herein is founded upon both epistemological and metaphysical considerations, and as such is a cogent example of the critical, integrative pluralism advocated by Sandra Mitchell (Reference Mitchell2003). I present an overview of the epistemological and metaphysical reasons for advocating pluralism in the study of homology, and a case study that reveals the utility and suitability of this approach. While a thorough treatment of the arguments for pluralism in the biological sciences as a whole is beyond the scope of this contribution, the specific case of homology as discussed here provides further evidence in support of both epistemological and metaphysical pluralism in biology.
3.1. Epistemological Considerations
The success of many homology concepts within their domains, and their concomitant failure in other domains, indicates that epistemological pluralism is warranted. Homology concepts must be specific to the level or phenomenon that is being explained. That a particular concept does not extend across all facets of biological similarity does not detract from its value in answering specific biological questions. Rather, allowing a diversity of ideas on homology introduces a new richness of theoretical perspective.
Helen Longino's (Reference Longino2002) account of the social nature of scientific inquiry provides further support for epistemological pluralism. Workers engaged in biological research represent a wide array of expertise approach their science from divergent angles and theoretical foundations. If the scientific tools, including theories, assumptions and hypotheses, employed by such groups are explicit and available for critical review, then there is no basis for assigning primacy to one field or another with regard to ability to tackle biological questions. Rather, by encouraging inquiry in multiple directions and at multiple levels, and requiring such inquiry to proceed in a transparent fashion, simultaneously interconnected and independent biological phenomena can be plumbed for new insights. The nature of the community engaged in biological inquiry requires that multiple perspectives be entertained and multiple threads be pursued (Longino Reference Longino2002), and the deeper we probe into biological intricacy, the more such theoretical perspectives we should, indeed must, entertain.
3.2. Metaphysical Considerations
There is no fundamental characteristic of organisms or biological phenomenon that can be extracted and given precedence over any other (Kitcher Reference Kitcher1987; Dupré Reference Dupré1993). Biology is, therefore, a pluralistic science, and the degree of separation and integration of various biological entities and phenomena forces the tailoring of theories and hypotheses to specific cases. Darwin (Reference Darwin1859) advocated a pluralistic approach to the study of evolution, one of the most unified theories in biology, by stating that while natural selection may be the main mechanism by which organisms evolve, it is not the only factor at work. Rather, a plurality of factors comes into play, and often these do not act in concert with one another (Gould and Lewontin Reference Gould and Lewontin1979). Because of this, the biological world cannot be divided up into neat categories, but rather sorts into a multitude of overlapping and incongruent divisions that are dependent upon the entities and phenomena under study. To whit, one cannot have an eye without a series of developmental events, and one cannot have a series of developmental events without nucleotide sequences. Therein lies the hierarchy. At the same time, however, the same nucleotide sequence may affect a number of different developmental events leading to very different structures, or different developmental pathways may produce the same structure, and therein lies the independence. This combination of independence and interconnectedness precludes a reductionist approach to the science.
Some workers have embraced various forms of pluralism in addressing particular aspects of biological inquiry such as species concepts (e.g., Kitcher Reference Kitcher1987; Ereshefsky Reference Ereshefsky, Hull and Ruse1998; Mishler and Brandon Reference Mishler, Brandon, Hull and Ruse1998), and in broader accounts of biology (e.g., Dupré Reference Dupré1993; Cartwright Reference Cartwright1999; Mitchell Reference Mitchell2003). The specific case of homology, outlined here, adds further evidence for a fundamental metaphysical pluralism in biology. This case shows that in order to study independent and hierarchical biological phenomena effectively and thoroughly, we require approaches tailored to the specific biological field under investigation. Information obtained from many such targeted studies can then be compared and synthesized to the degree possible, and new fields of study may be brought to light. Just as biology is hierarchical and independent simultaneously, so must biological inquiry be in order to do the most justice to its subject matter.
3.3. Case Study: Lens Induction in Anurans
Analysis of a biological example illustrates the intricacies and complications of homology in biology. Consider the classic case of the development of the eye lens in Anura (frogs and toads). Early embryological experiments appeared to indicate that the optic vesicle is necessary for the induction of the lens in the frog Rana fusca (Spemann Reference Spemann1901). Further experimentation revealed that a number of anuran species are able to generate a lens in the absence of an optic vesicle, including species congeneric with those requiring the vesicle's presence (Jacobson and Sater Reference Jacobson and Sater1988). The genus Bufo, for example, contains species that both do (Bufo bufo) and do not (Bufo regularis) form the lens in the absence of the optic cup. It has now been shown that different series of inductive and suppressive interactions, and the influence of tissues as diverse as the neural plate and heart mesoderm, are responsible for lens formation in various anurans, and that the optic vesicle plays only a supportive role in lens development (Jacobson and Sater Reference Jacobson and Sater1988). Thus many of the developmental mechanisms inherent in the occurrence of eyes in anurans are nonhomologous, but it is difficult to defend a hypothesis of nonhomology between the eyes of two anuran species of the same genus.
Lens induction in anurans provides a classic example of the failure of individual accounts to fully explain homology, while highlighting each account's ability to contribute positively to its understanding. It is apparent that anuran eyes, while they are developmentally individualized parts of the phenotype as required by Wagner (Reference Wagner1989), do not share the same developmental constraints. They are homologous on some, but not all, hierarchical levels as required by Abouheif (Reference Abouheif1997). Finally, anuran eyes pass Patterson's (Reference Patterson, Joysey and Friday1982) test of congruence with other homologues, while failing the ontogenetic origin aspect of the test of similarity, since the eyes in each species are developing in radically different ways. Thus while it seems that the eyes of congeneric anurans must represent an obvious example of homology, no single homology concept is able to properly account for the observed similarity. However, each of the accounts of homology presented here is applicable to a particular facet of the study of anuran eyes.
Homology, referring to any and all forms of biological similarity, can be identified in many ways and in many places: the genetic code, the fossil record, extant animals, developmental pathways, complete structures, parts of structures, behaviors, and functions, to name but a few. There exist many more examples of biological phenomena that cannot be explained by reference to a single homology concept. The myriad places where homology lurks is symptomatic of the myriad systems at work in biology, and homology concepts that explain one system will not necessarily be useful in explaining another. A complete picture of biological phenomena thus requires multiple homology concepts.
3.3.1. The Right Tool for the Job
In the preceding discussion, I have identified the pluralistic nature of both the biological sciences and of homology. The numerous interconnected yet independent systems at work motivate the study of homology at numerous levels. Rather than being disheartened by the ongoing, fruitless search for an overarching concept of homology, however, I suggest that taking a pluralistic approach to homology concepts as investigative tools provides the best approach to the study of biological similarity. In this approach, a homology concept appropriate to the problem at hand is carefully and explicitly articulated as an integral part of the intellectual framework in which the investigation takes place. This articulation must include a definition of what constitutes homology, criteria by which we are to recognize putative homologues, and causes for the phenomena observed.
The adoption of a pluralistic approach to homology in the biological sciences is not intended to be blindly permissive. If a homology concept does not include an explicit definition, operational criteria, and independently testable hypotheses of causality, then the utility of the concept as a whole is compromised. Additionally, explicit articulation of all components of the homology concept allows critical scrutiny of any study involving homology. In this way, the homology concept appropriate for the question at hand becomes ‘the right tool for the job’, and the pluralistic approach is a proactive and investigatively interesting one as opposed to being, as some have suggested (e.g., Ghiselin Reference Ghiselin1987), merely defeatism and a resort to relativism and subjectivity.
I now revisit the case study of the anuran eye lens in the context of this approach to homology. A homology concept designed as an investigative tool to study the development of the eyes in anurans might be formulated as follows:
Definition. Morphological structures whose overall forms, as well as their component parts, closely resemble each other in physical appearance, are homologous. Thus the eyes of Bufo bufo and Bufo regularis are homologous structures.
Criteria. The resemblance among structures is sufficient to exclude other structures from similar comparisons; the structures are located in the same topographical position in each specimen; the structures function in a broadly similar fashion (i.e., photoreception). Thus we can recognize the eyes of Bufo bufo and Bufo regularis as homologous.
Cause. Common ancestry. Thus we understand the reason for the presence of the homologues in both species.
Using this homology concept as an investigative tool, we can study the reasons behind the divergence of developmental pathways leading to anuran eyes, without being encumbered by the lack of homology at the developmental level.
A different homology concept allows the study of a different aspect of anuran eyes, namely their ancestry. Such a concept could be formulated as follows:
Definition. Structures that have arisen via identical sequences of developmental events are homologous. Thus the eyes of Bufo bufo and Bufo regularis are not homologous.
Criteria. Developmental pathways involve the same timing, onset, and offset of events; signaling molecules and mechanisms are the same; embryos are developing under identical environmental conditions. Thus we can recognize that eyes of Bufo bufo and Bufo regularis are not homologous.
Cause. Common ancestry. Thus the nonhomologous eyes of Bufo bufo and Bufo regularis arose separately.
Using this homology concept we can study the evolution of eyes, perhaps through selective pressure generated by a need to capture moving prey, and we have the impetus for reevaluating the validity of the genus Bufo.
These examples show that an explicitly stated homology concept can be a very valuable tool in studying biological questions, and demonstrate how the use of multiple homology concepts provides the freedom to investigate the complexity of the biological world. Careful articulation of definitions, criteria, and causes will make the homology concept being used ‘the right tool for the job’, and will allow exploration of multiple aspects of biological similarity.
4. Objections and Responses
Arguments for the use of pluralism in science often generate negative responses from the monistic camp (e.g., Ghiselin Reference Ghiselin1987; Hull Reference Hull1987). The standard objections to pluralism are that it is relativistic and subjective, and as such can lead only to confusion, ambiguity and problems of communication. In effect, the concern is that pluralism is far too much of an ‘anything goes’ approach to science (Ghiselin Reference Ghiselin1987; Hull Reference Hull1987), with some going so far as to suggest that pluralism is actually destructive in science (Hull Reference Hull1987). This is not the case according to the account presented here. As other advocates of pluralistic approaches to the biological sciences have stated (e.g., Ereshefsky Reference Ereshefsky, Hull and Ruse1998; Longino Reference Longino2002), it is the responsibility of each worker to clearly define the framework wherein he or she is operating. Rather than creating a communication barrier, this explicitness will in fact bring increased clarity and represent an improvement over the present state of affairs, in which homology concepts are seldom clearly articulated in conjunction with the investigation at hand. The approach to homology presented here is emphatically not an ‘anything goes’ proposal, and requires explicit and responsible articulation of the definition, criteria, and causes associated with any homology concept. These must be made available and defended in an open forum.
5. Conclusions and Prospectus
Biology is pluralistic (Kitcher Reference Kitcher1987; Dupré Reference Dupré1993; Mitchell Reference Mitchell2003), and attempts to interpret biological similarity phenomena using a monistic, reductionist approach are encumbered by the incompatibility of this approach with the complexity of the discipline. Investigation of representative homology concepts from several disciplines within biology reveals that while useful in their native domains, such concepts fail when attempts are made to extend their scope. Study of a biological example reveals the complexity of the subject and the ability of biological systems to function both hierarchically and independently at the same time. Epistemological pluralism in the study of homology in the biological sciences best conforms to the metaphysical pluralism that underlies the science, by drawing upon multiple homology concepts to explain biological phenomena.
Homology concepts should be viewed as important investigative tools, and explicit articulation of the definition, criteria, and causes associated with these concepts should accompany any investigation that makes reference to the phenomenon of homology. Such an explicit articulation forms an integral part of the theoretical and intellectual framework of the investigation at hand, and makes the homology concept ‘the right tool for the job’. A pluralistic approach, rather than being relativistic, subjective, and generative of chaos, will bring a new richness of perspective and will open new areas and directions for biological inquiry.
