Devonian sharks from Antarctica 
301 
depends partly on whether the Xenacanthida is 
defined as a 'stem-based' or 'apomorphy-based' 
clade in the sense of de Queiroz and Gauthier 
(1990; also Doyle and Donoghue 1993). More 
important, however, is the issue of whether 
Antarctilamna is better placed within the 
Phoebodontidae, as Zidek (1990) has suggested. 
This decision must be based on synapomorphies, 
and there is a need to clarify the characters on 
which the various families are based, using a 
cladistic framework which distinguishes 
symplesiomorphy, synapomorphy, and 
autapomorphy. From this perspective we may 
consider 'crown-group' and 'stem-group' 
xenacanths separately, before examining the 
validity of the family Phoebodontidae. 
Regarding 'crown group' xenacanths, Zangerl 
(1981) recognised two xenacanth families: 
Diplodoselachidae and Xenacanthidae Fritsch, 
1889, the latter including three genera: 
Orthacanthus, with a straight spine of circular cross- 
section attached to the shoulder girdle, and 
Xenacanthus and Pleuracanthus with a cranial spine 
transversely elliptical in cross-section, but which 
differed in dentition and pectoral fin structure 
(lepidotrichia present in Xenacanthus). Zidek (1990, 
in press) provided new information on the spine of 
Orthacanthus, which is cephalic in the type (O. 
cylindricus Agassiz, 1843), but positioned above the 
shoulder girdle in the species 'O.' senketibergianus 
Fritsch, which perhaps may be closely related to 
Diplodoselache. Zidek otherwise recognised four 
valid xenacanth genera: Expleurocanthus, 
Orthacanthus, Xenacanthus, and Triodus. Details of 
tooth morphology for some of these was reported 
by Hampe (1988a) ( Xenacanthus ; 1988b, 1991; 
Orthacanthus; 1989, Triodus). These genera are 
distinguished by such dental characters as the 
height of the central cusp, the presence or absence 
of striations and serrated edges on the cusps, and 
the number of nutritive foramina on the base, as 
well as by numerous histological features (Hampe 
1991). These are similar features to those used to 
differentiate phoebodont teeth from xenacanth 
teeth in the Devonian, so there is little point in 
including these late Palaeozoic forms in 
comparisons of stem group taxa. However, these 
studies are instructive in giving an indication of 
variability in tooth morphology. Thus, in 
Xenacanthus there are about 16 teeth in each jaw 
ramus, with the largest teeth located in the middle 
region of the jaw, and small posterior teeth lacking 
the central cusp (Hampe 1988a: figure 2). Apart 
from this 'gradient' heterodonty (Duffin and Ward 
1983), there is little variation in tooth morphology. 
Regarding stem-group xenacanths, Antarctilamna 
was considered by Zidek (1990) to be of 
phoebodontid (ctenacanthoid) affinity because its 
teeth have bilobed bases. The Family 
Phoebodontidae was placed within the Superfamily 
Ctenacanthoidea by Zangerl (1981), and the latter 
was united with hybodonts and neoselachians on 
the possession of two dorsal fin spines of 
neoselachian morphology. However, as noted by 
Ginter and Ivanov (1992), the tooth character (outer 
pair of cusps as high or higher than main cusp) by 
which the family was diagnosed by Williams (1985) 
does not occur in the articulated specimen 
'Phoebodus' heslerorum. Ginter and Ivanov (1992) 
considered this family to contain three genera: 
Phoebodus, Omalodus, and Thrinacodus, but the last 
genus was earlier interpreted as a xenacanth 
(Johnson 1984), although subsequently included 
with Phoebodus by Turner (1982). Long (1990: 62) 
included Thrinacodus within the Phoebodontidae 
and noted similarities between the expanded root 
of Thrinacodus with that of Phoebodus gothicus 
(Ginter 1990). 
Omalodus Ginter and Ivanov, 1992 resembles 
Portalodus in the labial projection of the base, and 
the absence of a lingual torus. If this is the 
important taxonomic character, then the diplodont 
condition of Portalodus must have evolved 
separately from that in other forms (e.g., 
Antarctilamna, crown xenacanths), implying that 
the family Phoebodontidae containing their three 
genera is a paraphyletic grouping. It is clear that 
phoebodontid monophyly is not well established, 
and there is a need to assess the polarity for a 
range of characters concerned with fin-spines and 
teeth, as analysed below. 
Analysis of Morphology 
Fin-spines 
Zangerl (1981: figure 51) used a branching 
diagram ('cladogram') to place Desmiodontida, 
Xenacanthida, Symmoriida and some other groups 
together with a basal node representing the 
absence of spines of neoselachian morphology, 
which he regarded as the primitive condition for 
Elasmobranchii. However, outgroup comparison to 
holocephalans, placoderms, acanthodians and 
osteichthyans would indicate that at least one 
dorsal fin-spine could be primitive, as previously 
argued by Young (1982) and followed by Maisey 
(1984) and Lund (1985). The single dorsal spine of 
crown-group xenacanths, which attached to the 
shoulder girdle or the back of the cranium, is an 
autapomorphy, with the more posterior position of 
the spine supporting the dorsal fin in Diplodoselache 
(Dick 1981), and a specimen ascribed to 
Orthacanthus (see above), showing that this 
specialisation evolved within the group. Together 
with the unique combination of diplodont teeth 
and a ctenacanth-like spine, as demonstrated in 
Antarctilamna, Zangerl's hypothesis is difficult to 
sustain on the grounds of parsimony. Furthermore, 
