Crisp & Chandler, Paraphyletic species 
829 
comparison between eastern Australia and western Australia, excluding the central arid 
region and central-northern inonsoonal region (Top End')- Thus the comparison is 
between the humid and subhumid south-west on the one hand, and the humid and 
subhumid east coast and ranges on the other. These regions were divided by onset of 
aridity during the late Miocene-Pliocene, particularly the formation of the Nullarbor 
Plain. Interestingly, the proportion of metataxa in the west (16%) is less than half that in 
the east (35%). This difference is significant (G = 7.736,1 d.f., P < 0.01), and suggests that 
the history of speciation or subsequent differentiation has differed in the two regions. 
Discussion 
Species concepts predict paraphyly 
The examples given in this paper represent a common (perhaps universal) pattern of 
asymmetry between sister species. Hennig (1966: 59) recognised this asymmetry, and 
described it as the 'deviation rule'. A divergent, autapomorphic species is either sister 
taxon to a metataxon, or nested within a paraspecies. In this view, branch-points in 
phylogenetic trees are like a bush of twigs (metataxa) from which leaders (autapomorphic 
taxa) emerge (cf. de Queiroz & Donoghue 1990a: fig. 6). In fact, virtually all concepts 
that treat species as historical entities make an implicit or explicit prediction that some 
or all species are not monophyletic. Evolutionary and related species concepts (Wiley 
1981; Frost & Kluge 1995) treat permanently split lineages as species. It is not surprising 
that proponents of the lineage notion of species are preoccupied with exclusivity 
(de Queiroz & Donoghue 1988; de Queiroz & Donoghue 1990a; de Queiroz & Donoghue 
1990b; Baum & Shaw 1995; Graybeal 1995) because all newly established lineages probably 
are not exclusive (sensu Donoghue 1985), and thus either paraphyletic, metaphyletic, or 
even polyphyletic if reticulation continues (cf. Frost & Kluge 1995: fig. 3). The expectation 
is that they eventually become exclusive (by extinction of part of the paraphyletic or 
metaphyletic residue) and monophyletic (by acquiring new autapomorphies) (Rieseberg 
& Brouillet 1994; Baum & Shaw 1995; Frost & Kluge 1995; Graybeal 1995), but the 
prediction of initial non-monophyly discomfits many authors. Frost & Kluge (1995) 
reject exclusivity arguments as invalid reductionism from one 'scalar' level of explanation 
(species) to another (populations or individuals). However, to us it seems a logical 
extension to analyse species at the level of their subunits, such as populations or 
internodons (Kornet 1993a). To regard species as indivisible seems absurd. 
Under the phylogenetic species concept, any species that is diagnosed by at least 
one autapomorphy is expected to be monophyletic, but any species that is diagnosed 
only by plesiomorphies may be either metaphyletic or paraphyletic. The composite 
species concept goes further: it explicitly predicts that all species are paraphyletic 
groups of their subunits (internodons), unless they have spawned no descendant 
species (Kornet 1993b: fig. 5a). Moreover, Kornet's model predicts that most species 
give rise to descendant species, except a few that become extinct before they can do 
so. This is because composite species are viewed as branches from a main limb, 
rather than equally splitting branches, emphasizing the same asymmetry at speciation 
as recognised in Hennig's deviation rule (Kornet 1993a: 87). 
Only the monophyletic (autapomorphic) species concept disallows the possibility of 
non-monophyletic species, but at a high cost. Either species would not be mutually 
exclusive and ancestral species would include descendant species (Kornet 1993b: 
71), or it would be necessary to reject all species that lack autapomorphies, which 
would leave many organisms permanently unassigned to species. Both alternatives 
would be unacceptable to most biologists. 
