Phylogenetic Definitions and Taxonomic Philosophy KEVIN?E QUEEROZ Depar@ment ??e ?eb ? a ? e ? O ? Og ? Divisiono ? Amph,p@fb ?? andRepfiles Una4edStatesNodfionarMuseum OfNaturalH ?fo ? SmL ?? nLp ? nLFr ? f ? OL Washington DC, 20560 U.S.A. ABSTRACT: An examination of he post-Darwinian history ofbiological taxonomy reveals an implicit assumption hat the definitions of taxon names consist of lists of organismal traits. That assumption re rese ts a failure to grant the concept of evolution a central ole intaxonomy, andit causes conflicts between traditional methods f efining taxon names and evolutionary concepts of axa. Phylogenetic d finitions of taxon names (de Queiroz and Gauthier 1990) gr nt the concept of common ancestry a central ole in the definitions of taxon names and thus constitute an important step inthe development of phylogenetic taxonomy. By treating phylogenetic relationships rather than organismal traits as necessary and sufficient properties, phylogenetic d finitions remove conflicts between the definitions of taxon names and evolutionary conceptsof axa. The general method of efinition represented by phylogenetic definitions of clade names can be applied to the names of other kinds ofcomposite wholes, including populations and biological species. Thatthe names of individuals (composite wholes) can be defined in terms ofnecessary and sufficient properties provides the foundation for a synthesis of seemingly incompatible positions held by contemporary individualists nd essentialists concerning the nature oftaxa nd the definitions of tax nnames KEY WORDS: Clade, class, composite whole, definition, defining property, essen- tialism, evolution, individua1, intension, name, oste sive definition, phylogeny, popula ? tion, set, species, taxon, taxonomy. INTRODUCTION Traditionally, the names of biological taxa have been treated as if they are defined by sets of organismal tr its, thatis, material traits oforganismsor heir pans, including morphology, physiology, behavior, ecology, genetics, and biochemistry. Tha view has predominated at least since the time ofAristotle (384-322 BC), persisting throu h the ages of Linnaeus (1707-1778 AD) and Darwin (1809-1882 AD) even into c ntemporary evolutionary/phylogenetic approaches to taxonomy. In the context of attempts to develop an explicitly phylogenetic system ofbiological taxonomy (e.g., Hennig 1965, 1966; Eldredge and Cracraft 1980; Wiley 1981a; Ax 1987), several authors have suggested th possibility of basing the definitions of taxonnames on phylogenetic relation- ships (e.g., Wiley 1979, 1989; Ghiselin 1984; Ridley 1986; Rowe 1987; de Queiroz 1988; Gauthier et a1. 1988a). Recently, de Queiroz and Gauthier (1990) BiologyandPh4l ? o ?? J:295-313 ? 1 ? ? O l ? 2K ?? ? A6 ? emLP ? fF ? ? ? p ?? L ?? ? LN ?? ?? 296 KEVIN DEQUEIROZ described in detail how that can beaccomplished, and they xplored someof the implications of phylogenetic def nitions for axonomic practices. Here Ireview differences between traditional and phylogenetic definitions in the context of the history of biological taxonomy. Although much of that discussion has been presented previously (de Queiroz and Gau hier 1990), itis here modified so as to be neutral with respect to a dispute, discussed later in the paper, between co temporary individualists nd e sentialists. It al o provides background for the subs quent analysis. In thatubsequent analysis, I examine the relevance of phylogenetic definitions o several issues in the philosophy of taxonomy, including general methods fo defining the names ofcomposite wholes, the categorization of such definitions, and a dispute between co tem ? porary individualists nd essentialists concerning the ature oftaxa and the definitions of taxon ames. In so doing, I hope to illustrate the importance of phylogenetic d finitions to taxonomic philosophy and consequently the coherence of the methodological system provided by the phylogenetic approach to taxonomy. I1. TRADITIONAL AND PHYLOGENETIC DEFINITIONS OF TAXON NAMES Most taxonomists have not distinguished clearly b tween taxa and taxon names (for exceptions se Hull 1965; Ghiselin 1966a), and some have proposed explicitly that definitions apply o both(e.g., Muir 1968). Biological taxa are sometimes viewed as abstract concepts rather than concrete ntities ( .g., Muir 1968; L0vtrup 1986,1987a, b), and concepts are sometimes saidtobe defined in the sense of being made distinct or clear. Alternatively, taxa are sometimes viewed as concrete ntities, and concrete ntities are sometimes saidto be defined - orof being self-defining - in the sense of marking out aboundary (e.g., Michaux et a1. 1990). In this paper, I use "define" in the sense ofspecify - ing the meaning of a symbo1, which is to say that definitions apply o words but not to the things symbolized by those words. My restricted us of "define" is not intended to ny the importance of clarifying concepts or of recognizing the boundaries of concrete ntities, but only to avoid confusing the meanings of words with either of these other things. Because they generally do not distinguish between taxa nd axonames, taxonomists have seldom addressed explicitly the issue of howtaxon names (as opposed to taxa) re to be defined. And they often use "definition" nterchan- geably with "diagnosis" (Simpson 1961). Explicit statements treating lists of organismal traits as definitions of taxon names have been made mostly by philosophers (e.g.. Hull 1965; Buck and Hull 1966; R se1973). Nevertheless, taxonomists who use "definition" as if it is asynonym of "diagnosis" generally present o ly the list of characters without an additional st tement specifying the meaning ofthe name. That is to say, either they do not distinguish between diagnoses/descriptions of taxa (i.e., lists oforganismal tr its) and definitions of taxon names, or else they avoid the issue of definition altogether. Either PHYLOGENETIC DEFINITIONS 297 alternative repr sents a failure to devise evolutionary definitions of taxon ames. Phylogenetic d finitions represent a fundamental shift in out100k concerning the role of thconcept of evolution in taxonomy. Under traditional efi itions, that concept is granted the role of an after-the-fact xplanation for the order already manifest in taxonomy instead of being taken asa central tenet from which the principles and methodsof taxonomy are to be deduced (de Queiroz 1988; de Queiroz andGauthier 1990). Denying evolution a central role in taxonomy effectively obstructs the development of a truly evolutionary or phylogenetic system oftaxonomy, and it explains the delay of the Darwinian Revolution in that discipline (e.g., O'Hara 1988; de Queiroz 1988; de Queiroz and Gauthier 1990). The persistence of traditional definitions of taxo ames exemplifies that delay, for such definitions effectively deny the concept of evolution a ce tral role in the definitions of tax names by granting organismal traits primacy overphylogenetic relationships. Traditional Definitions The traditional view that definitions of taxon names are tobe stated interms of sets of organismal traits has held regardless of whether those defining characters were viewed as manifestations of an eternal essence, as they were by Aristotle, and regardless of the precise method of definition. That situation is evident in the history of both the theory and the practice of biological taxonomy. Neverthe ? less, it is worthwhile to review certain spects of the post-Darwinian history f taxonomy in order tosupport that conclusion as well as to highlight the peri ? pheral role that the principle of evolution has layedin traditional definitions. With the acc ptance of an volutionary world view, the assumption hat definitions of taxon names wereto be based on organismal traits imposed severe constraints on theoretical propositions regarding the nature ofthose definitions As noted by Hull (1965), traditional views concerning the definitions of taxon names have their roots inthe Aristotelian method of definition.Aristotle's own theory ofdefinitions was tied to the concept of essences; however, his general method of definition can be adopted in ependent of that concept. According to Hull (1965), what Aristotle advocated, in mo em termsisdefinition by properties that are severally necessary and jointly sufficient (for taxon member ? ship), that is, definite conjunctive definitions. Taxa whose names are defin d in terms of necessary nd sufficient properties are called monothetic (Sneath 1962). Based on organismal traits, definite definitions of taxon names ? whether conjunctive or disjunctive ? proved tobe incompatible with an volutionary concept of axa (e.g.. Hull 1965; Beatty 1982; Sober 1988). Evolutionary ch nge can result in loss ormodification of he very characters that supposedly define the name of a taxon. Therefore, given that the modified d scendants re considered to belong to the taxon in question, one must conclude that the s0- called defining characters were not t uly defining, that is, they were not necessary and sufficient to define the name 298 KEVIN DE QUEIROZ As I will argue below, this problem can be solved byadopting an evolutionary concept of defining properties. That was not, however, the way in which the problem wasinitially so ved. Initially, defining properties were mplicitly assumed totake the form of organismal traits, and alternative methods of definition were advocated in or er toaccommodate n evolutionary c cept of taxa. Thus, although he later rejected this view (see below).Hull (1965) proposed thattaxon names must be defined as cluster concepts u ing indefinite disjunctive definitions (see also Beckner 1959). In such a definition, no character or set of characters is necessary nd any one of numerous ets is sufficient (for taxon membership). Taxa whose names are defined using indefinite disjunctive def nitions are called polythetic (Sneath 1962). Because no character or set of characters in an indefinite disjunctive defini- tion is necessary fortaxon membership, modification or loss of characters during the course ofevolution can occur without invalidating such adefinition. 1ndefinite definitions, however, havethe unfortunate property ofbeing indefinite (Suppe 1974, 1989), which means that the list of defining characters cannot be completed, ev n inprinciple (Hull 1965). Indefiniteness is necessary to accom ? modate the possibility of uture evolutionary change. Although that flexibility makes indefinite definitions compatible with evolution, t al o c mpromises their effectiveness as definitions. Some taxonomists and philosophers avoided the limitations of cluster concepts by adopting an alternative view oftaxa and the definitions of taxon names. Instead ofviewing taxa as sets or classes, the names of which were defined by properties of their members, Ghiselin (1969, 1974,1980,1981,1984, 1985) and Hull (1976,1977, 1978) proposed that taxa be viewed as individuals, that is, composite wholes. Under this view,the names of taxa are proper names, which supposedly cannot be defi ed interms of necessary nd sufficient properties. Proper names supposedly can only be defined by pointing to the object being amed. Lists oforganismal tr its, formerly treated as definitions, were viewed byindividualists not as definitions but as descriptions or d agnoses. 1n short, definitions taking the form of necessary and sufficient properties were abandoned altogether in order toaccommodate evolution. Constraints imposed by the tra itional method f definition are also evident in the history of axonomic practices. It has been asserted repeatedly that the theory of evolution had little mpact on axonomic practices (e.g., Dobzhansky 1937; Hopwood 1950; Himmelfarb 1968;Stevens 1984). Regardless of the truth of such a broad generalization, the principle of evolutio c rtainly did no im ? mediately take on a central role in governing those practices. That conclusion is evidenced by the long period of time during which ideas about evolutionary relationships had little bearing on the acceptance or rejection of taxa - a phenomenon related directly to prevailing concepts concerning the definitions of taxon ames. Because the meanings of taxon names were vie d asbeing embodied in organismal tr its, such traits were ffectively granted primacy over evolutionary relationships. Con equently, evolutionary relationships had little or no bearing o the acceptance or rejection of taxa. PHYLOGENETIC DEFINITIONS 299 For example, theories of polyphyletic origins for various taxa ha e been popular among evolutionists, and were especially so during the 1950s and 60s. A case in point is the history of ideas concerning the origins of Mammalia (reviewed by Hopson and Crompton 1969). Some authors (e.g., Simpson 1959; Olson 1959) believed that several differ nt line ges had evolved s0-called mammalian characters as the result of evolutionary par llelism. As I noted previously (de Queiroz 1988), acceptance of such atheory often did not lead to the rejection ofthe supposedly polyphyletic taxon, even by professed evolutionists. That situation seems to have occurred because th animals in question all shared certain characters. Given that those characters, rather than phylogenetic relationships, were viewed as efining the taxon'sname, then the precise evolutionary origins of the taxon a d its characters (i.e., whether th y were single or multiple) wereirrelevant to questions of its taxonomic validity. A similar situation exists concerning paraphyletic t xa, those consisting of an ancestor and some but not all of its descendants. Although the situation s changing (see below), paraphyletic taxa have been widely accepted by evolutionists. As in hecase of polyphyletic taxa, justifications for recognizing paraphyletic taxa have been implicitly based on shared organismal characters. This has been the case regardless of whether the argument was tieddirectly o shared characters or only indirectly through appeals to an evolutionary process (e.g., adaptation, anagenesis) intended to explain their existence and/or pattern of distribution (de Queiroz 1988). Because characters, rather than evolutionary relationships, were interpreted as he bases of taxa, the realization that axo was paraphyletic was not seen as a reason to reject i . Evolutionary considerations gained importance in the work of He nig (e.g., 1965, 1966) and his followers (e.g., E dredge and Cracraft 1980; Wiley 1981a; Ax 1987), who rejected both polyphyletic and paraphyletic taxa. Nevertheless, for the mostpart hey continued to treat the names of taxa as being defined in terms of organismal traits. This resulted in attempts to redefine the term "character" so thatonly s napomorphies, the supposedly defining characters of monophyletic taxa, would qualify (e.g., Wile1981b; Nelson a d Platnick 1981). This reformulation of the c cept of defining characters apparently was related o a constraint imposed by the traditional notion f defining properties. 1n order toreconcile the proposition that only monophyletic taxa warranted taxonomic recognition with the traditional method of definition based on organismal traits, heconcept of defining character had to be modified to exclude those characters (plesiomorphies, homoplasies) shared by the o ganisms of paraphyletic and polyphyletic taxa. 1n summary, the historical matters discussed above illustrate that defining properties of taxon names have been assumed totake the form of organismal traits. They also illustrate constraints imposed by that ssumption botho the theory oftaxonomic definitions and on taxonomic practices. As aconsequence of equating defining properties with or anismal tr its, views on the method of definition of taxon names had to be modified so that they would be compatible with evolutionary c cepts ofaxa. In order to accommodate the possibility of 3 ? KEVIN DEQUEIROZ evolutionary change, either the lists of defining characters had to be made less definite, or lse definition in terms ofnecessary and sufficient properties had to be abandoned altogether. Neither of those alternatives was ble to provide satisfactory definitions of taxon ames. 1n the case oftaxonomic pra tices, some volutionists saw no inconsistency in recognizing paraphyletic or polyphyletic taxa, presumably because the or- ganisms of those taxa shared what were judged to be significant characters. As a result, those authors a tempted to justify the recognition of such taxa with some form of after-the-fact evolutionary explanation for why organisms shared characters. The importance of phylogenetic relationships was effectively denied in that such relationships could not be used to reject taxa. Some recent taxonomists acknowledged an inconsistency in he recognition of paraphyletic and polyphyletic taxa, but because th y also implicitly assumed that defining characters ook the form of organismal traits, theywere forced to redefine the term "character" so thatit would apply only to the characters of monophyletic taxa. Phylogenetic r lationships were acknowledged as being important, but they did not become central tothe definitions of tax ames. Instead, the concept of adefining character was reformulated so that it would be compatible with modified views about taxa. Insummary, although evolutionary c nsidera ? tions have become increasingly important during the post-Darwinian history of taxonomy, they have until recent y been denied a central role in the definitions of taxon ames. p ? 70 ??? D ?? lo ? As ? a fundamental departure from traditional methods fdefinit on, e Queiroz and Gauthier (1990) described three classes of phylogen tic definitions, that i three m ans by which a name can be explicitly defined as referring to a par ? ticular clade (Figure 1). Phylogenetic d finitions were formulated in he context of an explicitly ph logenetic approach to t xonomy - one that identifies common evolutionary descent as he underlying p ocessfrom which the taxa of interest derive their existence. Common descent thus takes on the role of a central tenet from which the principles and methods of taxonomy are to be derived or educed (deQueiroz 1988). It was from this per pective that the method of defining taxon names was reformulated so that definitions would no longer be based on organismal tr its but on phylogenetic relationships. Phylogenetic d finitions are thus firmly rooted inthe concept ofevolution, that is, of common descent. In contrast withthe situation under t aditional defini ? tions, the possession of particular organismal traits is neither necessary nor sufficient foran organism to beconsidered pan ofa taxon. What is both necessary nd sufficient is being descended/rom a pa ticular ancestor. Phylogenetic def nitions, like definitions inmany sciences, serve to synonymize shorthand symbols or l bel withlonger expressions (e.g.. Popper 1950). Specifically, phy ogenetic def nitions are used togive names to clades, which otherwise mu t be identified by lengthy expressions.1 For example,