813 Paraphyletic species Michael D. Crisp and Greg T. Chandler Abstract Crisp, Michael D. and Chandler, Gregory T. (Division of Botany and Zoology, The Australian National University, Canberra ACT 0200, Australia) 1996. Paraphyletic species Telopea 6(4): 813-844. We present evidence, mainly from plants, that many recognised species and subspecies are paraphyletic. Whilst some cladists have argued that species are like other taxa, and should be monophyletic, it is clear that even cladists either implicitly or explicitly recognise non- monophyletic species. Moreover, species concepts such as the phylogenetic species concept and the composite species concept predict non-monophyly of many species. Whenever a monophyletic species is circumscribed, it is possible that a paraphyletic or metaphyletic 'residual' species is simultaneously recognised. Furthermore, attempts to place all organisms in a monophyletic taxon at every rank regress to the population level where monophyly is no longer applicable, leaving paraphyletic residuals. These groups of organisms can hardly be ignored, unless one wishes to define them out of existence (as in the monophyletic species concept). It has been argued that paraphyly is only an artifact of the Linnean system, which requires all organisms to be classified in certain ranks, e.g. species. However the phenonemon of regress shows that this is incorrect, because paraphyly is inherent in species. The solution to this conundrum is to recognise species as special taxa, which may be monophyletic or paraphyletic. (Higher taxa should always be monophyletic, and can be made so.) This requires the acceptance of a species concept that allows paraphyly, such as the phylogenetic species concept or the composite species concept. The monophyletic species concept, which does not allow paraphyly, is not acceptable. The special nature of species derives from their basal position in the phylogenetic system. Theoretically, the proportion of paraphyletic and metaphyletic species may be 50% or higher. Empirical estimates range from 20% to 50%. Use of non-monophyletic species in historical applications such as biogeography is widespread but may not be invalid, depending upon the assumptions made. Introduction In recent years, systematists have sought a species concept that is compatible with a phylogenetic system. They have rejected the biological species concept because of its reliance on the single criterion of reproduction. Entities which are distinct in many evolutionary, biological, and ecological features are nevertheless capable of interbreeding (Endler 1989, pp. 629-30). The biological species concept has never dealt satisfactorily with the conundrum of potentially (but not actually) interbreeding allopatric populations. Above all, the biological species concept is based on contemporary micro-evolutionary processes and cannot be reconciled with a phylogenetic system, in which taxa are viewed as historical units, extended in time and the units of a nested hierarchy (Rosen 1979; Donoghue 1985; Cracraft 1989; Vrana & Wheeler 1992; Frost & Kluge 1995). Species as lineages Systematists have debated whether species should be viewed as lineages or taxa (Nelson 1989b; Rieppel 1994; Frost & Kluge 1995). Recent views of species as lineages hark back to a model presented by Hennig (1966: fig. 6), showing a lineage of sexually reproducing organisms splitting into two daughter linages. Each lineage is 814 Telopea Vol. 6(4): 1996 made internally cohesive by reticulating ('tokogenetic') relationships among its component organisms, but no such connections exist between lineages •— they are mutually exclusive. Species are the internodes of a phylogenetic tree and speciation is the point at which lineages split permanently. Hennig's model has been reproduced many times, with modifications to show details or complications such as temporary versus permanent splits, reticulation and extinction (Kornet 1993a; Kornet 1993b; O'Hara 1993; Frost & Kluge 1995; Graybeal 1995). Most importantly, a lineage species is a model of evolutionary process. It is viewed as a real entity that evolves in time and space, has a definite beginning and end, and may be the ancestor of lineages comprising one or many species. It has been called the 'evolutionary species concept' (Wiley 1981; Frost & Kluge 1995) and the 'internodal species concept' (Nixon & Wheeler 1990; Kornet 1993a). Some authors have been preoccupied with 'exclusivity' of lineage species (Donoghue 1985; de Queiroz & Donoghue 1988; de Queiroz & Donoghue 1990a; de Queiroz & Donoghue 1990b; Baum 1992; Baum & Shaw 1995; Graybeal 1995). (This is often called 'monophyly' but strictly monophyly refers to a taxon diagnosed by an autapomorphy.) A lineage is exclusive only if all its members are more closely related to one another (by ancestry) than to any member of another lineage. For example, the lineage of descendants of Queen Victoria is not exclusive because some descendants are more closely related to members of other lineages (by marriage, e.g. cousins). This discussion can reduce to the absurd because any lineage may be shown to be non-exclusive if examined minutely enough, even body cells (Frost & Kluge 1995). In her formalisation of a lineage concept of species, Kornet (1993a) shows that internodal species are miitualli/ exclusive partitions of the genealogical network. Whilst this is a different notion of exclusivity from that discussed above, Kornet shows the latter problem to be irrelevant by using descent rather than ancestry as the criterion of group membership. Species as taxa A major problem with species conceived as lineages is that they have poor empirical content (Kornet 1993a). When we find two allopatric populations that are essentially similar, we have no way of judging their fate — whether they are the basis of new, historically separate lineages, whether either will become extinct, or whether they will reunite and become reproductively, tokogenetically cohesive again. Therefore, systematists have also proposed concepts of species that have an empirical basis. In this view, species are part of a pattern of similarity among organisms: the hierarchy of internested groups that are called taxa (Nelson & Platnick 1981; Nelson 1989b; Rieppel 1994). The internested groups or taxa are recognised by shared similarity in characters, known as synapomorphy or homology. This hierarchy is represented as a tree (cladogram or phylogeny), but it is an abstract representation of pattern. The branches of the tree represent taxonomic groups which are internested, static and do not evolve. Thus the stem at the base of the angiosperms represents not the ancestral species of all angiosperms, but the most inclusive set of all taxa that we call angiosperms, recognised by the set of characters that all angiosperms share, and marked on the stem. Rieppel (1994) suggests that species conceived as lineages and species conceived as taxa are 'complementary but incompatible'. (Frost & Kluge (1995) refer to this distinction as the 'scalar' hierarchy versus the 'specification' hierarchy.) If taxa are also considered to be ancestors and descendants, then we are confronted with a paradox (Nelson 1989b). For example, does the subordination of the angiosperms to the seed plants imply that the seed plants are the ancestors of the angiosperms? Surely not, because the angiosperms are also a part of seed plants. Crisp & Chandler, Paraphyletic species 815 and angiosperms are not ancestors of themselves, any more than I am part of my grandfather. However, it cannot be disputed that some member of the seed plants evolved into the first angiosperm. The resolution of this paradox is to recognise taxa as units of an hierarchical pattern, not as part of the evolutionary process. By logical extension, species belong to this hierarchy. This pattern, when reconstructed, may be used as a framework for hypotheses about the evolutionary process, e.g. that a seed plant with certain characteristics gave rise to the first angiosperm. Thus the role of ancestor is restricted to lineages and their subunits, such as individuals or populations (Rieppel 1994) or 'internodons' (Kornet f993b). If species are treated as taxa, then they are not different in kind from higher taxa. They are simply the least inclusive units in the systematic hierarchy. Recent concepts of species as phylogenetic taxa derive from Nelson & Platnick (1981: 12), who define species as 'the smallest diagnosable cluster of self-perpetuating organisms that have unique sets of characters'. This is known as the 'phylogenetic species concept', 'irreducible unit' or 'minimum diagnosable unit' (Cracraft 1989; Nixon & Wheeler 1990; Nixon & Wheeler 1992). However, a unique or diagnostic character may be either an apomorphy or a plesiomorphy, and a group diagnosed solely by the latter is not monophyletic, which is anathema to authors such as Nelson (1989a; 1989b). Such species are not simple internodal partitions of a phylogenetic tree. They 'survive' a speciation event in which an autapomorphic species branches off from the phylogenetic stem (Nixon & Wheeler 1992: fig. 4.7B). Contrast this with Hennig's (1966: fig. 6) methodological extinction of ancestral species at branch-points. Neither Cracraft nor Nixon and Wheeler confront the paraphyly issue, but instead imply that paraphyletic species are acceptable, provided that they manifest unique and fixed character combinations. Under the phylogenetic species concept, speciation is the point at which a lineage acquires an apomorphy, or more precisely when a new character is fixed (Nixon & Wheeler 1992: fig. 4.7). This is true even of species diagnosed by a plesiomorphy, because at some point earlier in history, every plesiomorphy was an apomorphy. A problem with the notion of an 'irreducible unit' is that it is prone to regress (cf. de Queiroz & Donoghue 1990b). Peripherally isolated populations in which trivia! genetic characters can easily become fixed would be diagnosed as species, contrary to general practice. One solution to the paraphyly problem is the monophyletic species concept, under which species have at least one autapomorphy (Rosen 1979; Donoghue 1985; Nelson 1989a; Nelson 1989b). However, this concept is unsatisfactory because ultimately taxa are not necessarily divisible into monophyletic sister-taxa (Smith 1994b). The smallest autapomorphic unit may have as its sister-group an unresolved symplesiomorphic cluster of organisms. The autopomorphic species concept consigns these to limbo, outside any species, but they can scarcely be ignored. Some authors have taken the pragmatic view that phenetic clusters may be treated as species. This approach has been termed the 'phenetic species concept', although it is actually an empirical criterion, free of assumptions about evolutionary process. Such units have been termed 'phena' (Mayr 1969; Smith 1994b), to distinguish them from theoretically based 'species'. In fact, the phenetic species concept is the formal equivalent of the traditional 'taxonomic' or 'morphological' species concept, under which species are circumscribed by intuitively perceived similarity among their members (Sneath & Sokal 1973: 364-5). Sometimes this concept is conflated with the phylogenetic species concept; however, clustering by overall similarity is not the same as clustering by diagnostic (fixed) characters. Clusters in phenetic space may share no diagnostic character; usually they are circumscribed by a series of partially correlating (polythetic) characters. Nevertheless, some authors have argued that phenetic clusters may be equivalent in practice to phylogenetic species (Theriot 1992; Crisp & Weston 1993). 816 Telopea Vol. 6(4): 1996 The composite species concept (Kornet 1993b), as its name implies, combines the lineage and taxon views of species. Komet first formalises the internodal species concept (Kornet 1993a) then reveals a significant drawback with it: every isolated population is a potential new lineage and it can be made permanent by extinction. Thus, internodal species tend to be trivial units compared with those that are generally recognised as species, and are more akin to populations. Moreover, the internodal species concept is operationally intractable, because the fate of an isolated population cannot be determined. Instead, Kornet defines composite species as lineages of 'internodons' which begin with the fixation of a novel character in an ancestral internodon and end with another fixation in a descendant internodon (or extinction). Composite species are parts of lineages because internodons have ancestor-descendant relationships and are mutually exclusive. Because they are diagnosed by fixed novel characters (autapomorpliies), they are also taxa and operationally equivalent to phylogenetic species. Thus the composite species concept seems to reconcile the tension between species-as-lineages and species-as-taxa (above). It should be noted that composite species are paraphyletic groups of internodons, unless they become extinct, in which case they become monophyletic (Komet 1993b: 69). Paraphyly and metaphyly Cladism has led to rejection of paraphyletic taxa on the grounds that they are not real phylogenetic units and lead to confusion about both the distribution of characters and the relationships of taxa (Donoghue & Cantino 1988; Humphries & Chappill 1988). Paraphyletic groups are considered 'artifactuaT and qualitatively different from monophyletic groups, which are 'real' taxa (Nelson 1989b). For every monophyletic taxon recognised, any of a series of paraphyletic groups may be constructed by excluding the monophyletic taxon from more inclusive (higher-level) monophyletic groups. In this way, paraphyletic groups have been treated as taxa, diagnosed by symplesiomorphies or the absence of the autapomorphies which circumscribe the excluded monophyletic groups. When taxa are discovered to be paraphyletic, systematists are inclined to divide them into several more narrowly circumscribed, monophyletic taxa (monophyly can also be achieved by amalgamation). However, this process of division may regress to the species level, where a problem is encountered: species are not divisible into monophyletic subunits. Moreover, both the phylogenetic species concept and the related composite species concept predict that many, if not most, species are not monophyletic (above). Here is a conundrum: if species are to be considered taxa, logically the sanction against paraphyly should apply to them (Cracraft 1989; Nelson 1989a; Nelson 1989b). Empirically, it has long been recognised that many accepted species are paraphyletic ('paraspedes': Ackery & Vane-Wright 1984). In a paraspecies, some (but not all) members are more closely related to members of another species than to other members of the paraspecies. Evidence for paraphyly would be a synapomorphy which some members of the paraspecies share with the other species (Fig. Ic). Some authors have pointed out that any species that lacks an evident autapomorphy is at least potentially paraphyletic; however, this is only an inference based on lack of evidence (it is also potentially monophyletic). The term 'metaspecies' has been coined (Donoghue 1985) to distinguish such species (whose phylogenetic status has not been resolved by cladistic analysis) from paraspecies (whose presumed monophyly has been tested and refuted). (Gauthier (1986) extends the metataxon concept to higher taxa but this is not relevant here.) Phylogenetic analysis of populations comprising a metaspecies may have one of three outcomes (cf. de Queiroz & Donoghue 1988: fig. 7): (i) a synapomorphy may be found for all populations, and Crisp & Chandler, Paraphyletic species 817 monophyletic monophyletic monophyletic metaspecies species species species -1 I-1 I metaspecies paraphyletic monophyletic unresolved monophyletic species species after testing species Fig. 1. Illustration of monophyletic species, paraspecies (paraphyletic) and metaspecies (unresolved), a. Initial phylogeny showing a metaspecies as sister-group to a monophyletic species. The metaspecics has no apomorphic characters except 1, which it shares with its sister species. The monophyletic species has an autapomorphy, character 2. b-d, Possible outcomes following cladistic analysis of populations in a. b, A new symapomorphy (character 3) is found for populations comprising the metaspecies, which is now recognised as a monophyletic species, c, A new synapomorphy (character 4) is found which is shared by two populations of the metaspccies and the monophyletic sister species. Tire original metaspecies is now recognised to be paraphyletic. d. No further apomorphies are found, and the metaspecies remains unresolved. These definitions apply equally to subspecies. Terminal branches represent populations. Solid bars represent original apomorplries; open bars represent additional apomorphies discovered following cladistic analysis. 818 Telopea Vol. 6(4): 1996 the 'metaspecies' is shown to be monophyletic (Fig. lb); (ii) a synapomorphy may be found (character 4, Fig. Ic) demonstrating that some populations of the metaspecies are more closely related to a recognised monophyletic species, in which case the 'metaspecies' is shown to be paraphyletic; (iii) no new apomorphy is found, and the species remains an unresolved metaspecies (Fig. Id), diagnosed only by a symplesiomorphy (character 1, Fig. Id). Both paraspecies and metaspecies are diagnosed by symplesiomorphy (character 1, respectively in Figs. Ic and Id). However, they differ in that evidence exists to show that part of the paraspecies is more closely related to another species (character 4 in Fig. Ic), whereas no such evidence is found in a metaspecies (Fig. Id). To summarise, depending upon the observed distribution of apomorphies among populations, the phylogenetic status of a species may be: unresolved (a metaspecies), non-monophyletic (a paraspecies) or monophyletic (an autapomorphic species). Note that irrespective of the phylogenetic relationship of their populations, all these species are diagnosable units consistent with the phylogenetic and composite species concepts. Therefore all are real, discoverable and corroborable entities. Moreover, the phylogenetic relationship of their parts (monophyletic, paraphyletic or metaphyletic) is also discoverable and corroborable (by the adducement of additional evidence). Objectives In this paper, we present examples of paraspecies and metaspecies and empirically estimate their proportion of all species. We show that any attempt to purge the system of these is futile, because of the asymmetric distribution of apomorphic (relatively advanced) characters among basal lineages (species). Consequently we address the conundrum of paraspecies and metaspecies in a system to which these are anathema. Finally, we consider the implications for comparative methods such as cladistic biogeography of a false assumption of species monophyly. For the purpose of this paper, we make no fundamental distinction between species and subspecies. This paper is concerned with lowest-level taxa, whether ranked as species or subspecies. The concepts monophyly, paraphyly and metaphyly apply equally to either, and to taxa of any rank. We do not consider the effects of reticulation, as this is a separate problem. Examples of paraspecies The following five examples report cladistic analyses using as terminals either populations or geographic forms that do not have evident autapomorphies and may well be paraphyletic. Are these suitable units for cladistic analysis? Some authors suggest that using paraphyletic terminals invalidates phylogenetic analysis (Cracraft 1989; de Queiroz & Donoghue 1990a; de Queiroz & Donoghue 1990b; Nixon & Wheeler 1990; Wheeler &c Nixon 1990; Vrana & Wheeler 1992). Moreover, because evolution among populations is likely to be reticulate, the strictly hierarchical model of cladistic relationships may be invalidated (Crisp & Weston 1993). However, these problems are not restricted to populations: the monophyly of most taxa (e.g. species and subspecies) is untested and thus uncertain (Nelson 1989b). Moreover, this paper shows that many such taxa are probably paraphyletic. An extensive literature attests to the frequency of reticulate evolution among recognised species (e.g.. Funk 1985; Barton & Hewitt 1989; Harrison 1991; Arnold 1992; Grant & Grant 1992; Smith 1992). Thus, problems affecting cladistic analysis of populations apply at least in part to subspecies and species. Vrana & Wheeler (1992) advocate using as terminals individual organisms, whose monophyly can (perhaps) be safely presumed. However, Crisp & Chandler, Paraphyletic species 819 their approach is likely to encounter serious sanapling problems: if the terminals comprising a data set are too sparse a sample of the variation within the study group, then an incorrect topology may be found because divergent change may confound estimates of homology. This is becoming evident in studies using molecular data (Melnick et al. 1993; Smith 1994a), as well as fossils (Donoghue et al. 1989: 444-449), By analogy, a single individual may be an inadequate sample of the character variation within a species or higher taxon. Clearly cladistic analysis cannot, by the nature of its assumptions and limitations, reconstruct all the historical events affecting populations, such as reticulation or isolation of a lineage in which no detectable apomorphy has evolved. However, it should retrieve the major divergence events as well supported clades, and on this basis we shall proceed. Daviesia ulicifolia Recently we have attempted to resolve the complex species Daviesia ulicifolia (Fabaceae: Mirbelieae). As currently circumscribed, this taxon extends from Cape York Peninsula in far north Queensland (16°S, 145°E) south to Tasmania (43°S, 147°E) and west to the Great Victoria Desert (30°S, 124°E). This is a vast range: 27° in latitude and 23° in longitude; however, the distribution is 'L'-shaped and only covers about 20% of the Australian continent. Additionally, in south-eastern mainland Australia, it extends from sea-level to tree-line at 1800 m altitude, where continuous snow cover is experienced in winter months. Not surprisingly, this is a polytypic species showing several morphological forms. Bentham (1864: 81) named four forms (under the illegitimate name D. ulicina), but neither he nor his successors have produced a satisfactory classification of the species. Our treatment used mainly morphometric characters of the leaves, stems, inflorescence and flowers to identify phenetic clusters that might be recognised as taxa. Although several morphologic- geographic clusters were evident, most of these intergraded in the regions of overlap, and we have treated them either as subspecies or informal forms (Table 1; Chandler & Crisp, in prep.). Environmental variables such as soil texture, nutrients and climatic parameters correlate with the clusters but these too overlap between forms. The only form which we are segregating at species level is the 'Yorke' form, which is autapomorphic and appears more closely related to D. arthropoda than to D. ulicifolia (Fig. 2; Chandler & Crisp, in prep.). We have made a cladistic analysis of the forms of D. ulicifolia and related species which share with it a distinctive kind of calyx with equal teeth (D. acicularis, D. areuaria, D. arthropoda and D. microcarpa: Pate et al. 1989; Crisp 1995a). At this low taxonomic level, most of the available characters are morphometric in nature, and we used the coding method of Thiele (1993a). The few qualitative characters tend to be autapomorphies for the recognised species (Table 1), for example: toothed and revolute leaf margins (D. acicularis); midrib more prominent abaxially (D. areuaria); leaves angular-terete, stems lax and unbranched (D. microcarpa). Both D. arthropoda and D. ulicifolia lack autapomorphies and should be considered a priori metaspecies. For an outgroup we used D. ivyattiana, which appears to be closely related to the D. ulicifolia group (Pate et al. 1989). Tables 2 and 3 show the character list and data matrix respectively. We used the 'branch and bound' algorithm in PAUP (Swofford 1990) to find a single most parsimonious tree of 396 steps (Fig. 2). The data set shows significant cladistic structure (PTP < 0.01: Faith & Cranston 1991). However, the low bootstrap values on most nodes indicate a weak hierarchical pattern in the characters used. Little phylogenetic structure is expected at the level of diverging geographic forms, because they are unlikely to be fully differentiated lineages due to reticulation or gene flow, even if this occurs at a reduced level compared with undifferentiated populations 820 Telopea Vol. 6(4): 1996 Table 1. Terminal taxa used in the cladistic analysis of Daviesia ulicifolia. Autapomorphies (unique defining characters) are indicated where known. Name Distribution Autapomorphies D. wyattiana Eastern Great Dividing Range Linear leaves D. acicularis NSW, mainly coast and ranges Leaf margins toothed, revolute D. arenaria Mallee, NSW, VIC, SA Midrib prominent below D. arthropoda Central Australia Minute standard petal D. microcarpa Norseman, WA Leaves angular-terete; unbranched D. ulicifolia: angustifolia East coast, N of Hunter Valley - desert Southern arid interior, Plant pruinose; WA to NSW uniflorescence racemose grampians Grampians, western Victoria - lofty Mt Lofty Range, SA Standard petal red NVP Northern plains, VIC - pilliga Pilliga scrub, NSW - rusdfolia Montane to subalpine, - VIC and NSW subumbellate mainly lowland Victoria - typical Coastal south-eastern Australia - yorke Yorke Peninsula (SA); Leaves very thick, eastern Tas rigid, wrinkled Table 2. Characters used in the cladistic analysis of Daviesia ulicifolia. All characters are continuous variables normalised by log transformation, except qualitative characters, as indicated. 1. Leaf length 2. Leaf width 3. Leaf shape (ratio distance from tip to widest point: length) 4. Leaf cross section (ratio thickness: width, at widest point) 5. Standard width 6. Inflorescence type: 0 = solitary, 1 = umbel, 2 = raceme 7. Peduncle length 8. Rachis length 9. Pedicel length 10. Midrib: 0 = more prominent above, 1 = equal, 2 = more prominent below (unordered) 11. Divaricate habit: 0 = yes, 1 = no 12. Spinescent branchlets: 0 = yes, 1 = no Crisp & Chandler, Paraphyletic species 821 Table 3. Data matrix used in the cladistic analysis of Daviesia ulicifolia. Values for each character are standardised integers over the range 0 to 30 using the method of Thiele (1994). A polymorphism is indicated by 'p' (states 1 and 2). D. wyattiana 30 20 ? 4 29 1 30 0 30 1 1 1 D. acicularis 12 13 17 6 11 0 0 0 5 0 1 1 D. arenaria 2 30 16 0 12 0 0 0 8 2 0 0 D. arthropoda 14 26 0 2 0 0 14 0 17 0 0 0 D. microcarpa 6 0 30 30 8 0 0 0 10 ? 1 1 D. uiicifolia: angustifolia 6 1 25 12 11 0 0 0 0 0 0 0 desert 7 13 22 10 5 2 7 30 8 0 0 0 grampians 7 20 16 5 30 1 6 0 9 0 0 0 lofty 9 14 20 7 27 P 7 2 8 0 0 0 NVP 0 20 26 3 3 0 0 0 2 0 0 0 pilliga 6 26 17 1 6 0 0 0 3 0 0 0 ruscifolia 3 18 28 4 26 0 0 0 7 0 0 0 subumbellate 6 13 24 7 22 1 5 0 5 0 0 0 typical 5 13 22 5 16 0 3 0 6 0 0 0 yorke 12 28 3 2 22 1 11 0 12 0 0 0 (de Queiroz & Donoghue 1988; Crisp & Weston 1993). Nevertheless one clade with a moderate bootstrap value of 67 included all forms of D. ulicifolia and the three autapomorphic species D. acicularis, D. arenaria and D. microcarpa (Fig. 2). Thus D. ulicifolia appears to be paraphyletic by exclusion of the latter species. Only the 'Yorke' form is excluded from this clade, and it shows a sister-group relationship to D. arthropoda. As the species included within the D. ulicifolia clade are well separated, no re-rooting can make D. ulicifolia appear monophyletic, even if the distinct 'Yorke' is excluded from consideration. We tried constraining monophyly of D. ulicifolia, but this increased tree-length very substantially (68 extra steps), a significant difference which was not achieved in 100 randomised data sets (T—FTP < 0.01: Faith 1991). Other manipulations, such as selectively excluding combinations of species, did not substantially alter the relationships of the forms nor alter the paraphyly of D. ulicifolia. An alternative binary encoded data set (with fewer characters) produced a very unresolved tree but again showed D. ulicifolia as paraphyletic. Banksia integrifolia Thiele (1993b) and Thiele & Ladiges (1994) made a morphological analysis of the Banksia integrifolia (Proteaceae: Banksieae) complex, which broadly overlaps with D. ulicifolia east of the Great Dividing Range. Tlieir methods were essentially similar to those described above for D. idicifolia, using either binary or morphometric characters of adult leaves, fruits and juvenile leaves. Four phenetic clusters were recognised as taxa. Three of these overlapped in distribution and morphology. Only the most northern entity {'aquilonia') was found to be both morphologically and geographically disjunct, with no intermediates. The authors would have liked to segregate this as a species, but demurred on the basis of a cladistic analysis, which nested aquilonia deep within the complex (Fig. 3). Thus, B. integrifolia would have been rendered paraphyletic by 822 Telopea Vol. 6(4): 1996 removal of aquilonia. To circumvent this problem, they might have split B. integrifolia into four species (corresponding to the four phenetic-geographic clusters), but they refrained because of the existence of intermediates between most of the clusters. Instead, they described the four forms as subspecies of B. integrifolia. D. wyattiana yorke 36 D. arthropoda gram plans lofty 67 ruscifoiia 42 subumbellate 35 typical 23 D. aclcularis 24 desert 24 29 _ angustifolia 77 D. microcarpa 6 NVP pilliga 26 41 D. arenaria Fig. 2. Shortest tree (396 steps) for geographic forms of Daviesia tilicifolia (in plain font) and related species (in italics, preceded by 'D.'). Found using branch and bound' in PAUP 3.f.l (Swofford 1990) from data in Table 3. Numbers on internodes indicate bootstrap values from 100 replicates. The tree is rooted using the outgroup D. vn/attiam. Note that D. ulicifolia is paraphyletic by inclusion of four other species. B. canei B. saxicola ssp. integrifolia ssp. monticola B. integrifolia ssp. compar ssp. aquilonia Fig. 3. Cladogram of Banksin integrifolia comprising four subspecies (.integrifolia, monticola, compar and aquilonia) and its sister-group, comprising the species B. canei and B. saxicola, from Tliicle (1993b). Note that if aquilonia were segregated as a species, the remainder of B. integrifolia would be rendered paraphyletic.