The advent of evolutionary theory was, in some ways, a mixed blessing for biology. On the one hand, it represented a unifying theory to chart out the course of future research. On the other hand, that unifying theory proposed that all species resulted from a single unique flow of evolutionary diversification (phylogeny). Detecting evolutionary mechanisms, and choosing among different possible mechanisms to explain particular episodes of evolution, thus required biologists to have access to, and represent explicitly, details of that unique history. In retrospect, this seems a fairly obvious, and critical, element for the growth of biology; and yet, it was not until the past generation that biology actually developed an empirical method for accessing phylogeny.
The reason for the lag between recognizing the problem and doing something about it was the question of 'homology'. Homologous characters are similar traits found in two or more species as a result of descent from a common ancestor that had the trait. Homologous traits, therefore, indicate evolutionary relationships among species (phylogeny). The problem was that if homology is both defined by phylogeny, and required to reconstruct phylogeny, then a researcher needs to adopt an Orwellian double think strategy in order to 'know the phylogeny, obtain the homologies and build the phylogeny'.
The "new systematics" of the 1940's emphasized studies of population variation within species. Under the influence of theoreticians such as Fisher, Haldane, Wright, and Dobzhansky, evolution had been re-defined as changes in gene frequencies within and among populations under different environmental conditions, with no reference to historical transformation. The work of assembling phylogenies was left largely to paleontologists, who hoped that enough fossils would be found to provide a true record of phylogeny 'carved in stone'. Failing such a complete and unambiguous record, constructing phylogenies was considered to be only a quasi-scientific pursuit. By the late 1950s a group of biologists known as 'pheneticists' or 'numerical taxonomists' suggested that evolutionary theory should not be used as a criterion for systematizing biology. Logical positivists, they argued that nothing historical could be analyzed scientifically, and thus biologists should abandon any activities they considered 'non-operational'.
Both the evolutionary taxonomists and the pheneticists believed that the problem of assessing homology and phylogeny in a non-circular manner was intractable, either for empirical or philosophical reasons. A German biologist named Willi Hennig looked at the problem from the opposite view. That is, suppose there is an empirical solution to the question of phylogeny and homology. Hennig suggested that researchers begin with the assumption that all similar features are homologous. In some cases this will lead to the incorrect, and initially undetectable, identification of non-homologous as homologous. When a phylogeny is reconstructed by grouping species according to their shared homologies, these mis-identifications will be revealed because the non-homologous characters will not co-vary with the plurality of the other characters.
The distinction between the non-Hennigian and Hennigian approaches is subtle but vital. Suppose, while describing the four different bird species, you notice that all of the males perform the same type of raised wing display during courtship. A non-Hennigian systematist would say "Since this behavior looks the same in these four species and is performed by different members of a closely related group (they are all birds), it is an homologous trait. We can use this trait to assess the phylogenetic relationships among these birds." A Hennigian systematist would say "Since this display looks the same in these four species, it is homologous. We can use this trait to assess the phylogenetic relationships among these organisms, in conjunction with other traits." In the first case, homology is assumed because ofsimilarity among characters, coupled with presumed phylogenetic relatedness among the species. Hence, there is an underlying assumption of prior knowledge of phylogeny. In the second case homology is assumed solely on the basis of similarity among characters.
Having found a starting point for identifying homology, Hennig then recognized three types of homologous characters: (1) shared general characters, which identify a collection of species as a group; (2) shared special characters, which indicate relationships among species within the group; and (3) unique characters, which identify particular species. In addition there is a separate category, homoplasy or "false homology", which tells us nothing about relationships among taxa. Only shared special characters denote particular phylogenetic relationships within a study group, so characters must be assigned to one of the three homology categories before their usefulness in a phylogenetic study can be determined. Once again, the risk of circularity is high: "If two taxa are related, then a character that they share in common is a shared special homology; therefore, this character can be used to determine if the taxa are related". Hennig suggested that this determination be made by comparing the state of each character in the study group to the state of the same characters in one or more species outside the study group (outgroups). In this way, each character is independently assigned a particular homology status (general, special, or unique) depending upon properties of species for which the phylogenetic relationships are not being assessed (the outgroups).
Hennig suggested that any trait found in one or more members of a study group that is also found in species outside the study group is a general trait. Hence, the presence of vertebrae in mammals is a general trait because there are nonmammals that also have vertebrae. Those traits occurring only within the study group are special similarities. The members of the study group are then clustered according to their special shared traits. If there are conflicting groupings, it means that some traits assumed to be homologies because they are similar are actually homoplasies. Because all homologies covary, and homoplasies do not covary, the pattern of relationships supported by the largest subset of special similarities is adopted as the working hypothesis of phylogenetic relationships. As more and more traits are sampled, there will be progressively more support for a single phylogenetic pattern. Thus, the phylogenetic systematic method works in the following way: (1) assume homology, a priori, whenever possible; (2) use outgroup comparisons to distinguish general from special homologous traits; (3) group according to shared special homologous traits; (4) in the event of conflicting evidence, choose the phylogenetic relationships supported by the largest number of traits; (5) interpret inconsistent results, post hoc, as homoplasies.
Hennig called his approach phylogenetic systematics (Hennig, 1966), but it has become known almost colloquially as cladistics. Literally meaning 'branchings', 'cladistics' was coined as a term of derision by two major authority figures in papers delivered in the same symposium for practitioners of phylogenetic systematics. Ernst Mayr, objected to the use of phylogenetic on the grounds that it suggested that other approaches were not phylogenetic. Mayr did not apply the same reasoning to the name of his preferred school of thought, evolutionary taxonomy. Both Mayr and Robert Sokal, the father of phenetics or numerical taxonomy, caricaturized phylogenetic systematic methods as being based only on discerning the divergent branching patterns without taking into account relative differences in the degree of evolutionary divergence by different lineages. Proponents of phylogenetic systematics produced a series of articles demonstrating that the degree of evolutionary divergence could be ascertained directly from their phylogenetic trees, and began calling themselves 'cladists' as a sociologically aggressive response to the sociologically aggressive attacks by the evolutionary taxonomists and pheneticists.
Armed with a method for use in formulating, testing, and refining explicit hypotheses of phylogenetic relationships that relies on the weight of evidence rather than the authority of the author, biologists began contributing detailed information about phylogenetic relationships of many groups by the late 1970s. The results indicate that as much as two-thirds of all biological classifications are inconsistent with the evidence as depictions of phylogeny; therefore, evolutionary explanations derived from those presumed phylogenies are also suspect. Not surprisingly, the disruption of traditional views promoted by practitioners of phylogenetic systematics fueled tremendous controversy within biology (see Hull, 1988).
Phylogenetic trees produced by Hennigian methods are messages about evolutionary history written in a symbolic language (Brooks and Wiley, 1988). We use the patterns of relationships among species and information about particular changes in characteristics on the tree, as well as information about the geographical and ecological context in which those various species with their various traits live, to provide us with richer evolutionary explanations and to delineate implications of those phylogenetic patterns which can be considered predictions that can be tested empirically (Brooks and McLennan, 1991).
Some North American practitioners of phylogenetic systematics were quick to realize that the basic methodology need not be restricted to use in understanding the evolution of different species in biology; it seemed to be applicable to any kind of empirical historical reconstructions. In a landmark publication Platnick and Cameron (1977) noted that similar (though not identical) methods had been developed and commonly used in linguistics (e. g Hoenigswald, 1960) and stemmatics (textual analysis) (e. g. Maas, 1958) prior to the advent of phylogenetic systematics in biology. All approaches deal with systems that show evolutionary divergence through time, and encounter problems with establishing homology and sequence among the traits used to reconstruct evolutionary history.
In comparing languages, for example, one would begin with the assumption that 'pollo', 'pollo', and 'poulet', all meaning 'chicken' in Italian, Spanish, and French are similar because they are 'homologous', i. e. , derived from a common ancestral language. Furthermore, one would conclude that 'pollo' is plesiomorphic to 'poulet'. Applying these criteria to a large number of words and grammatical constructions would corroborate the initial hypotheses. This would provide a clue about mechanisms of linguistic change in the evolution of French, based on the homologous transformation of pollo to poulet. Based on the large number of words and grammatical constructions, we would also find that Portuguese would be included in the group of languages including Italian, Spanish and French. Recognizing that the word for chicken in Portuguese is 'frango' would lead us to predict that an older word similar to pollo had been replaced historically, and to search for that particular historical event. This would lead us to the Moorish occupation of Iberia, and the homoplasious introduction of various Arabic words into Portuguese (and Spanish, where, for example, the Arabic 'wad' becomes 'guad', as in Guadalajara). We could further use the linguistic phylogeny obtained to examine the history of changes in pronunciation. In Italian, the pronunciation of 'pollo' is similar to the English 'polo', whereas in Spain 'pollo' is pronounced something like 'pol-yo', and in parts of Latin America something like 'poyo'. Each component of language added to the comparative framework provided by this type of approach enhances our understanding of linguistic evolution.
This is particularly evident in more recent efforts toward the global reconstruction of language families and macro-families (e.g. Nichols 1992) or in ambitious attempts to correlate, from an evolutionary point of view, world wide linguistic and genetic mappings such as Cavalli-Sforza et al. (1994), who critically discuss the comparative merits of cladograms and the methods upon which they are based. Referring to Hennig (1966), Farris (1973) and Sober (1988), Cavalli-Sforza et al. note the recent interest of zoologists and botanists in "developing robust methods for reconstruction of evolution based on a careful choice of characters that are most informative for this goal" (1994: 31), and underline the potential relevance of cladistics beyond the disciplinary domain within which it was developed. In a paper given at the 1995 meeting of the International Federation of Rock Art Study Organizations (Torino), palaeontologist Shirley Chesney raised the question: "Can cladistics be applied to the classification of palaeolithic representation?" and pointed out that cladistics had been successfully applied to problems ranging from emotional development following Darwin's early insight to the structural relationship among dinosaurs. She invited those participating in a symposium on "New Approaches" to consider whether cladistics could prove productive if applied to the abundant and chaotic rock art data that evades chronological classification as a slowly evolving development toward complexity.
Similarly, in textual analysis copying texts by different people can result in mistakes (typographical errors), embellishments (such as marginal illuminations or the addition of text), or reductions (elimination of passages from the original) that can serve to establish the chronology of textual evolution. From such a chronology, a phylogeny of the evolution of a piece of literature/writing, we can often gain clues about the social and political context of the various copiers.
Nearly 20 years ago, Platnick and Cameron closed their discussion of cladistic methods in biology, linguistics and stemmatics thusly, ". . . biologists may be able to gain some historical perspective on current controversies in systematic theory by comparison with stemmatics and linguistics. In contrast to biology, where a prevalent and relatively unspecific "evolutionary" methodology has been successively challenged by phenetics and cladistics, stemmatics and linguistics have used cladistic methods essentially since their inception as discrete fields of inquiry. . . We may ask why it is then that cladistic methods have not always been used in systematics, just as they have been in stemmatics and linguistics." Their call for increased communication between biologists and linguists and textual analysts has gone largely unheeded, and groups of scholars with similar interests remain (to paraphrase) two groups separated by a common methodology.References Brooks, Daniel R. and and Deborah A. McLennan. 1991. Phylogeny, Ecology and Behavior: A Research Program in Comparative Biology. Chicago: University of Chicago Press.
Daniel R. Brooks is Professor of Zoology at the University of Toronto and Research Associate in the Royal Ontario Museum. He is the author, with E.O. Wiley, of Evolution as Entropy (University of Chicago Press, 1986, 1988), and with Deborah A. McLennan, of Phylogeny, Ecology and Behavior: A Research Program in Comparative Biology (University of Chicago Press, 1991) and Parascript: Parasites and the Language of Evolution (Smithsonian Institution Press, 1993). He has been interested in thedevelopment of quantitative applications of phylogenetic systematics, and the applications of the approach in comparative evolutionary studies and basic evolutionary theory. His field research on the phylogeny, historical biogeography and coevolution of parasitic worms inhabiting vertebrates has been centred in Latin America. He is currently involved, as the coordinator for the parasite working group, in the All-Taxon Biodiversity Inventory project in the Guanacaste Conservation Area of Northwestern Costa Rica.