Syst. Biol. 47(4):719–726,1998 Phylogenetic Uncertainty, Molecular Sequences, and the Denition of Taxon Names MICHAEL S. Y. LEE School of Biological Sciences, University of Sydney, Sydney, New South Wales 2006, Australia One of the most important recent developments in taxonomy has been the increasing acceptance that biological taxa should be dened in terms of phylogenetic relationships, rather than essential characters (de Queiroz, 1994). Three types of phylogenetic denitions of taxon names have been proposed: Node-based. The most recent common ancestor of A and B, and all its descendants. The least inclusive clade containing A and B. Stem-based. All taxa that share a more recent ancestor with A than with taxon B. The most inclusive clade containing A but not B. Apomorphy-based. The rst ancestor with trait X, and all its descendants. The clade diagnosed by trait X. In each case the original formulation (de Queiroz and Gauthier, 1990, 1992) is presented, along with the recently proposed modied phrasing (Schander and Thollesson, 1995) that removes reference to ancestors. The latter reformulations blunt the criticism that phylogenetic denitions are problematic because they implicate the “discarded concept” of ancestors (Dominguez and Wheeler, 1997:372). The use of such phylogenetic denitions is becoming increasingly prevalent, especially in studies dealing with higher vertebrates (e.g., Estes et al., 1988; Gauthier et al., 1988; Rowe, 1988; Sereno, 1991; Ford and Cannatella, 1993; Padian and May, 1993; Wolsan, 1993; Wyss and Flynn, 1993; Holtz, 1994, 1996; Meng and Wyss, 1994; Laurin and Reisz, 1995; Lee, 1998; Lee and Caldwell, 1998; Dilkes, 1998) and has generated considerable debate (e.g., Lucas, 1992; Rowe and Gauthier, 1992; Bryant, 1994,1996, 1997; Sundberg and Pleijel, 1994; Schander and Thollesson, 1995; Lee, 1996a, b; Lidén and Oxelman, 1996; Wyss and Meng, 1996; Cantino et al., 1997; Dominguez and Wheeler, 1997; de Queiroz, 1997). For inDirect correspondence to the author’s current address: Department of Biological Sciences, Monash University, Clayton, Victoria 3168, Australia; E-mail: [email protected] stance, some have suggested that ranksignifying sufxes in conventional Linnaean names convey information about inclusiveness of taxa, whereas names in the phylogenetic system do not (e.g., Lidén and Oxelman, 1996; Dominguez and Wheeler, 1997). However, rank-signifying sufxes can lead to conict between associating a name with a particular clade or with a particular Linnaean category (de Queiroz, 1997) and encourage unjustied assumptions about the “equivalence” of clades sharing the same sufx (Sundberg and Pleijel, 1994; Knox, 1998). Furthermore, Linnaean sufxes usually do not convey information about inclusiveness of taxa (Lee, 1996b). Dominguez and Wheeler (1997) claim, for instance, that sufxes show that Dasycerinae is nested within Staphylinidae. No such information is actually conveyed, however; all that can be acertained from the spelling of the names is that Dasycerinae is nested within (or possibly redundant with) some taxon ending in -idae. Without external information such as a reference cladogram, it is impossible to infer that the more-inclusive -idae taxon is Staphylinidae (Lee, 1996b). Nesting of taxa would be conveyed only in the rare instances where taxa have the same root but different Linnaean sufxes (e.g., Iguanidae, Iguaninae). Another area of debate concerns the stability of taxon names with phylogenetic denitions compared with those having conventional denitions based on types, characters, and Linnaean categories. Some have argued that phylogenetic denitions of taxon names are undesirable because the contents of the named clade can change if accepted phylogenetic relationships change (Lid én and Oxelman, 1996; Dominguez and Wheeler, 1997). However, despite assertions that phylogenetic taxonomy is “less stable than the Linnaean system by any meaningful measure” (Dominguez and Wheeler, 1997:367), these workers have not yet attempted to 719 720 SYSTEMATIC BIOLOGY demonstrate how a monophyletic Linnaean system would afford greater stability in such circumstances (e.g., Lee, 1996b). This is because such instability pervades any monophyletic system of nomenclature (e.g., Nelson, 1973; Gaffney, 1979). Here, I discuss how such instability might be minimized in the context of a particular monophyletic system, phylogenetic taxonomy. Previous writers have not emphasized that clades named by the three types of phylogenetic denitions are not always equally stable in content, or that one definition is not always superior to the others. If conciseness of denitions is valued, a node-based denition is preferable when certain areas of a phylogeny are poorly corroborated, whereas a stem-based denition is superior if other areas are poorly corroborated. The recommended context for apomorphy-based denitions, and the relevance of the three types of denitions to molecular phylogenies, will also be discussed. If these suggestions are correct, the stability of named clades, and conciseness of denitions of taxon names, can be improved by knowing when to employ each type of phylogenetic denition. Conversely, the practice of applying node-based names to all clades (e.g., Laurin and Reisz, 1995) is not optimal but can result in named clades with unstable contents, cumbersome denitions, or both in many situations. Similarly, the practice of always applying node-based names to clades bounded by extant forms (“crown-clades”) and applying stem-based names to more-inclusive clades containing fossil forms, without considering which areas of the reference phylogeny are poorly corroborated (e.g., Wolsan, 1993), can result in similar problems (Lee, 1996a; Cantino et al., 1997). The preceding approach does place more emphasis on the (verbal) denitions of names than on the concepts (clades) described by those denitions (see de Queiroz, 1994). Thus, where a taxon name can be applied to two slightly different clades (e.g., a stem-based or a node-based clade) with the same known contents, the clade that can be described more succinctly is preferred. However, if one instead emphasizes the con- VOL. 47 cept (clade) more than the verbal denition, then attaching a name to a particular clade has primacy; whether or not the chosen clade can be described succinctly becomes largely irrelevant. If this perspective is adopted, the recommendations of this paper may not be acceptable. Longer denitions also have advantages, some of which are discussed at the end of this paper. Most phylogenetic analyses produce cladograms containing both highly and poorly corroborated groupings. A typical result is shown in Figure 1a, where certain clades are well supported, having Bremer indices of 5 or more and bootstrap frequencies of at least 90%. The other clades are only weakly supported. A conservative cladogram for the group, with the dubious clades collapsed into polytomies, is shown in Figure 1b. I will focus on the properties of stemand node-based phylogenetic denitions as applied to a particular grouping on the conservative cladogram (Fig. 1b): the wellcorroborated clade FGHIJ, whose internal relationships are poorly supported. The taxon name Zeligidae will be dened so that it includes all ve taxa, and only these ve. (This name refers to Leonard Zelig, the human doppelganger in the eponymous Woody Allen lm who so desires acceptance that he continually metamorphoses to resemble those around him, thus creating difculties for those trying to associate his name with a recognizable biological entity.) A node-based denition of Zeligidae might then be, “the least-inclusive clade containing F and J.” If all the poorly corroborated clades on the most-parsimonious cladogram (Fig. 1a) are later supported by additional characters, then this denition would indeed include F, G, H, I, and J, as desired. However, the contents of Zeligidae as dened are highly unstable, given the current uncertainty regarding relationships among F, G, H, I, and J. If, based on additional characters, F and J are later identied as closest relatives (Fig. 1c), then the taxon Zeligidae would suddenly exclude G, H, and I. The same uncertainty would apply to any nodebased denition that referred to only two, three, or four of the ve taxa. There will always be the possibility that the taxa men- 1998 POINTS OF VIEW 721 FIGURE 1. (a) A hypothetical reference phylogeny, with both well-corroborated and poorly corroborated clades. The taxa to be included in the clade to be named Zeligidae are shown in bold font. For each clade, the rst number is the Bremer index; the second number, the bootstrapping percentage value. (b) A conservative phylogeny, with poorly corroborated clades in cladogram (a) collapsed into polytomies. (c) A new resolution of the uncertainty in cladogram (b), where F and J form a clade. A node-based denition of Zeligidae using F and J as reference taxa differs widely in content between cladogram (a) and the new arrangement. (d) A hypothetical reference phylogeny with the same topology as cladogram (a) but differing degrees of support for each clade. The taxa to be included in the clade to be named Zeligidae are shown in bold font. (e) A conservative phylogeny, with poorly corroborated clades in cladogram (d) collapsed into polytomies. (f) A new resolution of the uncertainty in (e), where C forms the sister-group to all other taxa. A stem-based denition of Zeligidae that uses C and H as reference taxa differs widely in content between cladogram (d) and the new arrangement. tioned in the denition will form a clade to the exclusion of the remaining taxa, which would thus be excluded from Zeligidae. The only way to ensure that a node-based definition of Zeligidae will include F, G, H, I, and J, regardless of how the uncertainty in Figure 1b is eventually resolved, is to include all ve taxa in the denition (Cantino et al., 1997), which then becomes “the leastinclusive clade containing F, G, H, I, and J.” However, such a denition is now rather long and cumbersome, because all ve included taxa (which in a real situation might be long binomials) have to be listed. One of the advantages of phylogenetic denitions is that taxa can be dened by simple reference to clades, rather than by exhaustive enumeration of “essential” characters or included taxa (e.g., de Queiroz, 1994). This advantage partly disappears if a complete list of included forms has to be included in a phylogenetic denition to make the name stable. However, the name Zeligidae can be associated with a clade containing FGHIJ much more succinctly by using a stem-based definition. The denition “the most-inclusive clade containing H but not C” will always include F, G, H, I, and J—and only these taxa. The uncertainty in Figure 1b might eventually be resolved in one of several ways; it is very unlikely, however, that the clade FGHIJ will be dismantled. Thus, pending a drastic revision of phylogenetic relationships, F, G, I, and J will be more closely related to H than to C. Furthermore, the external relationships of the named clade are well-corroborated: The nearest outgroup is the clade CDE, and the monophyly of this outgroup is well-corroborated. Thus, there is little possibility of D or E being more closely related than C to clade FGHIJ and thus of 722 SYSTEMATIC BIOLOGY being inadvertently “captured” by such a stem-based denition. Thus, if brevity is desirable, stem-based rather than node-based denitions should be used if relationships within the named clade are uncertain but the external relationships are well-corroborated. In these situations stem-based denitions are unaffected by uncertain relationships within the named clade. Stem-modied node-based denitions have recently been proposed (Wyss and Meng, 1996): If it is assumed that all organisms in Figure 1b are extant, such a definition of Zeligidae would be “the leastinclusive clade containing H and all living organisms more closely related to H than to C.” The phylogenetic relationships invoked in this type of denition are therefore the same as those of stem-based denitions. They should perhaps more appropriately be called “node-modied stem-based denitions,” since the basic structure of denition is similar to a stem-based denition but is altered slightly (by the addition of “least inclusive” and “living”) to apply to a morerestricted clade. In the context of phylogenetic uncertainty, therefore, stem-modied node-based denitions behave like normal stem-based denitions and (not surprisingly) are also appropriate for clades with uncertain basal relationships (see Wyss and Meng, 1996). A different situation applies to the example illustrated in Figure 1d–f. The topology of the phylogeny in Figure 1d is identical to that in Figure 1a, but support for each clade is different. A conservative cladogram for the group, with the dubious clades collapsed into polytomies, is shown in Figure 1e. Again, we wish to attach the taxon name Zeligidae to a clade including taxa F, G, H, I, and J, and only these taxa. Here, a node-based denition is more appropriate. The denition of Zeligidae as “the least-inclusive clade containing F and J” will always refer to the clade FGHIJ, regardless of how the relationships among the other taxa (A–E) eventually resolve. Because the clades FG and HIJ are both strongly supported, any two taxa (one from each clade) can be used to form a stable and succinct node-based denition that will include taxa F, G, H, I, and J, and only these taxa. VOL. 47 In contrast, stem-based denitions in this situation result in names being attached to unstable clades, unless the denitions are made long and cumbersome. The denition of Zeligidae as “the most-inclusive clade containing H but not C” will contain only FGHIJ if the poorly corroborated nodes on the most-parsimonious cladogram (Figure 1d) are later further supported on the basis of additional evidence. However, if later studies suggest that any or all of A, B, D, and E are more closely related to H than is C, then the former four taxa will be included (inadvertently) in Zeligidae (Fig. 1f). Thus, the only stem-based denition of Zeligidae that will be stable in content despite this phylogenetic uncertainty would have to be phrased as “the most-inclusive clade containing H but not A, B, C, D, or E.” Any taxon in the well-corroborated clade FGHIJ can be used instead of H. However, because the relationships among A, B, C, D, and E to the clade of interest (FGHIJ) are poorly corroborated, the only way to guard against inadvertently “capturing” one or more of them in a stembased denition is to mention and exclude all of them in the denition. As a result, six taxa have to be included in the phrasing of the denition. Thus, if brevity is desirable, node-based rather than stem-based denitions should be used if relationships within the named clade are well-corroborated but external relationships are poorly supported. In these situations, node-based denitions are unaffected by uncertain external relationships of the named clade. As mentioned before, stem-modied node-based denitions have properties similar to those of stem-based denitions and would thus also be unstable in this context. The stability of clades named by apomorphy-based denitions also needs to be discussed. If a unique and unreversed apomorphy diagnoses a well-corroborated clade (e.g., AB in Fig. 1a), a denition of a name that uses that apomorphy will (obviously) refer to a well-corroborated clade and be stable in content. In particular, that putative apomorphy is almost certainly homologous in all the taxa possessing it, and the denition will refer to a monophyletic group. If, on the other hand, a unique and unreversed apomorphy diagnoses a poorly 1998 POINTS OF VIEW corroborated clade (e.g., HIJ in Fig. 1a), a denition using that apomorphy will refer to a dubious clade and have unstable contents. There is usually little justication for naming such clades in any case. In particular, there is a real possibility that addition of new taxa and/or characters to the analysis will reveal that the putative apomorphy is actually convergent in the taxa that possess it, and the denition will refer to a polyphyletic group (Bryant, 1994; see below). In such cases, the denition will have to be either abandoned or substantially revised. Stability of a clade named by an apomorphy-based denition therefore depends on the support provided by other characters. Homoplastic apomorphies should be avoided in denitions, even if one of the convergent acquisitions or reversals helps diagnose a well-corroborated clade. Such clades should be named with use of other, nonhomoplastic apomorphies. However, certain types of homoplasy (i.e., convergence) are more problematic than others (i.e., reversal). For instance, the denition “the taxon diagnosed by wings” will refer to a polyphyletic assemblage since wings have evolved convergently in insects, sh, pterosaurs, birds, bats, and plant seeds. However, the denition “the taxon diagnosed by the tetrapod limb” will still refer to a discrete clade, even though subsequent reversals have occurred in snakes, amphisbaenians, caecilians, and other taxa. Because apomorphy-based phylogenetic denitions refer to the origin of a trait but not to its subsequent history, the origin of the trait must be a unique and clearly identiable event; the subsequent history of the trait is less relevant. Use of an apomorphy that exhibits reversals will therefore not result in the recognition of a paraphyletic group (contra Bryant, 1994). Thus, apomorphy-based denitions should only be used if there is strong evidence that apomorphy has a single origin, helps diagnose a well-corroborated clade, and can therefore be condently assumed to be homologous in all taxa possessing it. Dominguez and Wheeler (1997) have criticized apomorphy-based denitions for focusing on the origin of a trait rather than on its subsequent history. Because the di- 723 agnostic character might be lost (absent) in some members of a named clade, a reference cladogram is required to identify the contents of that named clade. They propose an alternative system, which they claim obviates the need for a reference cladogram, i.e., basing a taxon name on a “dening” character and interpreting organisms with “subsequent modications” (including total loss) of that character as somehow still “possessing” it. However, it is still difcult to see how such denitions can be applied in the absence of a reference cladogram. Suppose Tetrapoda is dened as “all organisms with four pentadactyl limbs;” without some cladogram, however, one could not possibly infer that bird wings and snake “nothings” are modications of pentadactyl limbs and that birds and snakes are thus tetrapods. Node-based and stem-based denitions can be framed with reference to molecular phylogenies. However, apomorphybased denitions and traditional characterbased denitions cannot readily accommodate molecular sequence information, although certain genetic traits can be used (e.g., gene duplications and rearrangements). Any character-based denitions expressed in the context of nucleotide or amino acid sequences will be extremely cumbersome and never achieve widespread usage. Node-based and stem-based phylogenetic denitions therefore have the advantage that they can readily incorporate the rapidly burgeoning information from such data. In particular, novel clades discovered by using such methods (e.g., the hippopotamus– cetacean clade: Gatesy et al., 1996; Hasegawa and Adachi, 1996; Buntjer et. al., 1997; Gatesy, 1997) cannot be dened by using apomorphy-based or traditional denitions unless one can also nd uniquely derived morphological apomorphies congruent with the molecular results. However, as such clades were not previously suspected on the basis of conventional (morphological, behavioral, etc.) analyses, the search for unambiguous nonmolecular apomorphies to use in such denitions might be fruitless. This observation blunts the argument that denitions of taxon names “should be related to the evidence [= characters] from 724 SYSTEMATIC BIOLOGY which they are derived” (Dominguez and Wheeler, 1997:369): In some situations, this will be highly impractical if not impossible. Thus, stem-based and node-based denitions are more widely applicable because, unlike apomorphybased denitions (and traditional character-based denitions), they can be used to name clades identied by molecular sequence data. The above recommendations are not hard rules; rather, they should be considered along with the suggestions of de Queiroz and Gauthier (1992), Bryant (1996), and Cantino et al. (1997). In some cases particular conventions may conict (Cantino et al. 1997). For instance, de Queiroz and Gauthier (1992) and Bryant (1996) both recommended standardizing widely known taxon names by giving them node-based, crownclade denitions. However, if basal relationships within a clade to be named are poorly corroborated (as in Fig. 1b), a node-based denition will be less stable or concise than a stem-based denition. Well-known taxon names, if attached to such clades via nodebased denitions, will therefore have unstable contents or cumbersome denitions (Lee, 1996a; Cantino et al., 1997). The alternative of attaching well-known taxon names to such clades by using stem-based denitions might not be acceptable to some biologists. Even in a stem-based denition, all known fossils that fall outside the diversity of the living clade can be excluded by using as a reference taxon the nearest known fossil outgroup to the living clade. However, future work might unearth even closer fossil relatives of the living clade. A well-known taxon name attached to this living clade through the use of a stem-based denition, therefore, would “capture” such new fossils. In this situation, taxonomists trying to dene a wellknown taxon name will have to weigh the stability of the content, and brevity of the denition, of a stem-based denition against the greater exclusivity (with respect to future fossil discoveries) of a node-based definition. If the latter approach is preferred, a long and complex denition would be required. A stem-modied node-based denition (Wyss and Meng, 1996; see above) might be the most appropriate choice in this context. VOL. 47 Bryant (1997) has recently suggested that “designated phylogenetic contexts” should be added to denitions of taxon names. Denitions of taxon names are deemed valid only in the context of particular reference phylogenies, thus stabilizing their usage and content. For instance, a denition of Haematothermia might be “the least-inclusive clade including birds and mammals.” This name, which was intended to refer to a bird–mammal clade, would be valid only in the context of heterodox phylogenies that group birds and mammals as extant sister groups (e.g., Gardiner, 1982; Lovtrup, 1985; Gardiner, 1993). However, this approach does not solve completely the problem of nomenclatural instability caused by uncertain phylogenetic relationships. Under previous schemes, phylogenetic instability results in changes to the contents (inclusiveness) of taxon names, whereas under Bryant’s scheme, phylogenetic instability would result in discarding certain names and coining new ones.This might be seen as replacing one kind of nomenclatural instability with another. The above recommendations remain relevant even if designated phylogenetic contexts are adopted, since they will result in application of names to clades such that designated phylogenetic contexts are unlikely to be violated. For instance, in Fig. 1a, applying a node-based name to the clade FGHIJ (“the least-inclusive clade containing F and J”) requires a designated phylogenetic context where FGHIJ form a clade, with F and J on opposite branches of its basal dichotomy. If basal relationships within FGHIJ are poorly corroborated (Fig. 1a), the designated phylogenetic context might be violated (Fig. 1b, c). In contrast, applying a stem-based name to the clade FGHIJ (“the most-inclusive clade including H but not C”) requires a designated phylogenetic context where FGHIJ form a clade and C is part of the sister group of that clade, an arrangement that is unlikely to be violated (Fig. 1b). Conversely, if basal relationships within FGHIJ are well-corroborated, but external relationships are poorly supported (Fig. 1d), then the reverse is true. The designated phylogenetic context of a node-based 1998 725 POINTS OF VIEW name for FGHIJ is likely to be satised (Fig. 1e), while that of the stem-based name might be violated (Fig. 1f). Similarly, applying an apomorphy-based name to a clade wellcorroborated by other characters obviously means thatthe designated phylogenetic context is likely to remain satised. As discussed above, instead of using a brief stem-based denition if basal relationships within the named clade are uncertain, and a brief node-based denition if external relationships of the named clade are uncertain, one can use either type of denition but include detailed (and thus long) lists of relevant taxa. Such denitions have the possible advantage of including designated phylogenetic contexts in their phrasing. Under Bryant’s proposals, Zeligidae can be attached to the clade FGHIJ in Figure 1b by dening it as “the most-inclusive clade containing H and not C,” and appending a designated phylogenetic context such as ABCDE(FGHIJ). However, if long denitions are not problematic, the designated phylogenetic context can be incorporated into the denition itself, which would thus read, “the least-inclusive clade containing F, G, H, I, and J but excluding A, B, C, D, and E.” This denition would make sense only in the context of phylogenetic hypotheses consistent with the arrangement ABCDE(FGHIJ); it would be nonsensical (and thus automatically invalid) in the context of alternative hypotheses. The need for mentioning numerous taxa in such denitions might be alleviated by reference to standard lists (e.g., Ford and Cannatella, 1993; Bryant, 1996). However, it is unlikely that standard lists will always contain groupings of taxa that coincide neatly with those required by many definitions. Furthermore, this approach partly shifts rather than addresses the problem of long denitions, as the denitions would then not stand alone but be partly contained in the relevant lists. Thus, if long denitions are not problematic, both stem-based and node-based denitions can be phrased to include designated phylogenetic contexts. Similarly, apomorphy-based denitions can also be elaborated to restrict their applicability to phylogenetic hypotheses where the apo- morphy is uniquely derived: “the clade diagnosed by the unique origin of apomorphy X.” Such denitions will be invalid in phylogenetic contexts where apomorphy X originates multiple times. Stem-, node-, and apomorphy-based denitions that include a designated phylogenetic context also have the advantage that their contents are at least partly identiable without reference to a cladogram. This refutes the argument that phylogenetic denitions are undesirable because they are cannot stand alone but rather must be interpreted in the context of extraneous information such as a reference cladogram (e.g., Dominguez and Wheeler, 1997). A recent case study has suggested that use of phylogenetic denitions of taxon names increases their stability and explicitness and thus results in superior taxonomies (Cantino et al., 1997). However, the full potential of the phylogenetic system to improve nomenclatural stability will not be realized if only one type of denition (e.g., node-based) is used exclusively, or if the three types of definitions are applied uncritically. Rather, in some contexts one type of denition will result in stable named clades and concise definitions, whereas in other contexts it will result in unstable named clades, cumbersome denitions, or both. In certain situations (e.g., molecular phylogenies) some types of denition are inapplicable. Knowing when to use each type of denition will help maximize the stability, conciseness, and accessibility of phylogenetic taxonomies and will enable systematists to communicate their work much more clearly and effectively. ACKNOWLEDGMENTS I thank R. Shine and colleagues for logistical support; K. de Queiroz, D. Cannatella, H. Bryant, G. Rouse, and two anonymous referees for helpful discussion and comments; and the Australian Research Council for funding. REFERENCES BRYANT , H. N. 1994. Comments on the phylogenetic denition of taxon names and conventions regarding the naming of crown clades. Syst. Biol. 43:124–130. BRYANT , H. N. 1996. Explicitness, stability, and universality in the phylogenetic denition and usage of taxon names: A case study of the phylogenetic taxonomy of the Carnivora (Mammalia). Syst. Biol. 45:174– 189. 726 SYSTEMATIC BIOLOGY BRYANT, H. N. 1997. Cladistic information in phylogenetic denitions and designated phylogenetic contexts for the use of taxon names. Biol. J. Linn. Soc. 62:495–503. BUNTJER , J. R., I. A. HOFF, AND J. A. LENSTRA . 1997. Artiodactyl interspersed DNA repeats in cetacean genomes. J. Mol. Evol. 45:66–69. CANTINO, P. D., R. G. OLMSTEAD , AND S. J. W AGSTAFF . 1997. A comparison of phylogenetic nomenclature with the current system: A botanical case study. Syst. Biol. 46:313–331. DE QUEIROZ, K. 1994. Replacement of an essentialistic perspective on taxonomic denitions as exemplied by the denition of “Mammalia.” Syst. Biol. 43:497– 510. DE QUEIROZ, K. 1997. Misunderstandings about the phylogenetic approach to biological nomenclature: A reply to Lid én and Oxelman. Zool. Scr. 26:67–70. DE QUEIROZ, K., AND J. GAUTHIER. 1990. Phylogeny as a central principle in taxonomy: Phylogenetic denitions of taxon names. Syst. Zool. 39:307–322. DE QUEIROZ, K., AND J. GAUTHIER. 1992. Phylogenetic taxonomy. Annu. Rev. Ecol. Syst. 23:449–480. DILKES, D. W. 1998. The early Triassic rhynchosaur Mesosuchus browni and the interrelationships of basal archosauromorph reptiles. Philos. Trans. R. Soc. Lond. B 353:501–541. DOMINGUEZ , E., AND WHEELER , Q. D. 1997. Taxonomic stability is ignorance. Cladistics 13:367–372. ESTES , R., K. DE QUEIROZ, AND J. GAUTHIER . 1988. Phylogenetic relationships within Squamata. Pages 119– 281 in Phylogenetic relationships of the lizard families (R. Estes and G. K. Pregill, eds.). Stanford Univ. Press, Stanford, California. FORD, L. S., AND D. C. CANNATELLA . 1993. The major clades of frogs. Herpetol. Monogr. 7:94–117. GAFFNEY, E. S. 1979. An introduction to the logic of phylogeny reconstruction. Pages 79–111 in Phylogenetic analysis and paleontology (J. Cracraft and N. Eldredge, eds.). Columbia Univ. Press, New York. GARDINER , B. G. 1982. Tetrapod classication. Zool. J. Linn. Soc. 74:207–232. GARDINER , B. G. 1993. Haematothermia: Warm-blooded amniotes. Cladistics 9:369–395. GATESY , J. 1997. More DNA support for a Cetacea/ Hippopotamidae clade: The blood-clotting protein gene g -brinogen. Mol. Biol. Evol. 14:537–543. GATESY , J., C. HAYASHI , M. A. CRONIN , AND P. ARCTAN DER . 1996. Evidence from milk casein genes that cetaceans are close relatives of hippopotamid artiodactyls. Mol. Biol. Evol. 13:954–963. GAUTHIER , J., R. ESTES , AND K. DE QUEIROZ. 1988. A phylogenetic analysis of Lepidosauromorpha. Pages 15–98 in Phylogenetic relationships of the lizard families (R. Estes and G. K. Pregill, eds.). Stanford Univ. Press, Stanford, California. HASEGAWA , M., AND J. ADACHI . 1996. Phylogenetic position of cetaceans relative to artiodactyls: Reanalysis of mitochondrial and nuclear sequences. Mol. Biol. Evol. 13:710–717. HOLTZ, T. R. J R. 1994. The phylogenetic position of the Tyrannosauridae: Implications for theropod systematics. J. Paleontol. 68:1100–1117. VOL. 47 HOLTZ, T. R. JR. 1996. Phylogenetic taxonomy of the Coelurosauria (Dinosauria: Theropoda). J. Paleontol. 70:536–538. KNOX , E. B. 1998. The use of hierarchies as organizational models in systematics. Biol. J. Linn. Soc. 63:1– 49. LAURIN , M., AND R. R. REISZ. 1995. A reevaluation of early amniote phylogeny. Zool. J. Linn. Soc. 113:165– 223. LEE , M. S. Y. 1996a. Stability in meaning and content of taxon names: An evaluation of crown-clade denitions. Proc. R. Soc. Lond. B 263:1103–1109. LEE , M. S. Y. 1996b. The phylogenetic approach to biological taxonomy: Practical aspects. Zool. Scr. 25:187– 190. LEE , M. S. Y. 1998. Convergent evolution and character correlation in burrowing reptiles: Squamate phylogeny revisited. Biol. J. Linn. Soc. (in press). LEE , M. S. Y., AND M. W. CALDWELL . 1998. The anatomy and relationships of Pachyrhachis problematicus, a primitive snake with hindlimbs. Philos. Trans. R. Soc. Lond. B (in press). LID ÉN , M., AND B. OXELMAN . 1996. Do we need phylogenetic taxonomy? Zool. Scr. 25:183–185. LOVTRUP, S. 1985. On the classication of the taxon Tetrapoda. Syst. Biol. 34:463–470. LUCAS , S. G. 1992. Extinction and the denition of class Mammalia. Syst. Biol. 41:370–371. MENG, J., AND A. R. WYSS. 1994. Enamel microstructure of Tribosphenomys (Mammalia, Glires): Character analysis and systematic implications. J. Mammal. Evol. 2:185–203. NELSON , G. 1973. Classication as an expression of phylogenetic relationships. Syst. Zool. 21:344–359. PADIAN , K., AND C. L. MAY. 1993. The earliest dinosaurs. N. M. Mus. Nat. Hist. Sci. Bull. 3:379–381. ROWE , T. 1988. Denition, diagnosis, and origin of Mammalia. J. Vert. Paleontol. 8:241–264. ROWE , T., AND J. GAUTHIER . 1992. Ancestry, paleontology, and the denition of the name Mammalia. Syst. Biol. 41:372–378. SCHANDER , C., AND M. THOLLESSON . 1995. Phylogenetic taxonomy—some comments. Zool. Scr. 24:263– 268. SERENO , P. C. 1991. Basal archosaurs: Phylogenetic relationships and functional implications. Mem. Soc. Vert. Paleont. 2:1–53. SUNDBERG , P., AND F. PLEIJEL . 1994. Phylogenetic classication and the denition of taxon names. Zool. Scr. 23:19–25. WOLSAN , M. 1993. Phylogeny and classication of early European Mustelida (Mammalia: Carnivora). Acta Theriol. 38:345–384. WYSS, A. R., AND J. J. FLYNN . 1993. A phylogenetic analysis and denition of the Carnivora. Pages 32–52 in Mammal phylogeny: Placentals (F. S. Szalay, M. J. Novacek, and M. C. Novacek, eds.). Springer-Verlag, New York. WYSS, A. R., AND J. MENG . 1996. Application of phylogenetic taxonomy to poorly resolved crown-clades: A stem-modied node-based denition of Rodentia. Syst. Biol. 45:559–568. Received 6 November 1997; accepted 1 June 1998 Associate Editor: D. Cannatella
© Copyright 2025 Paperzz