PHRYNOCEPHALUS PHYLOGENY 



9 



In both cases the consensus has the same topology (Figure 14). 

 When all characters were unordered, trees of 102 steps were pro- 

 duced which are congruent with those where characters were ordered, 

 but with less resolution in the clade consisting off! przewalskii and 

 its nearest relatives (the topology of this region of the tree is the same 

 as that shown in Figure 16.). 



When all these analyses were repeated using the 'heuristic search' 

 option ofthePAUP 3. 1.1 programme (Swafford, 1993), results were 

 identical. Bootstrapping (100 replicates), using this programme, 

 was also applied to the ordered tree rooted on Trapelus and nodes 

 with bootstrap support over 50% are indicated in Figure 14. 



Use of the successive approximations character weighting option 

 in Hennig86 produced little change in the original tree based on 

 unordered characters and rooted on Trapelus, merely resolving the 

 trichotomy in the consensus tree involving P. scutellatus and P. 

 golubevi, by making them successive branches on the main lineage 

 of Phrynocephalus. 



Principal states supporting nodes are shown in Figure 15. It will 

 be seen that some 1 3 nodes are supported by two or more conserva- 

 tive characters that show little or no homoplasy. The other nodes are 

 defined by single or noisy characters. A conservative tree recognis- 

 ing the nodes based on the former features, or with bootstrap support 

 above 50% (and in many cases both) is shown in Figure 16. 



Several nodes on the main lineage of Phrynocephalus are quite 

 well supported and a number of other subclades can be recognised. 

 Thus six species constituting a holophyletic group with marked 

 internal structure form the Phrynocephalus interscapulars group 

 consisting of P. interscapularis, P. sogdianus, P. euptilopus, P. 

 luteoguttatus, P. clarkorum and P. ornatus. The clade has geographi- 

 cal coherence, occurring in western Pakistan, Afghanistan, eastern 

 Iran and adjoining central Asia. Another well defined clade, the P. 

 theobaldi group, includes P. theobaldi, P. roborowskii and the rather 

 more different P. vlangalii. The similar tuberculated species, P. 

 helioscopus, P. persicus, P. strauchi and P. rossikowi may form 

 another unit, although it lacks marked bootstrap support. 



DISCUSSION 



Biogeography 



Ananjeva andTuniyev (1992) speculate about the history and bioge- 

 ography of Phrynocephalus in the former USSR. Their complex 

 hypothesis is difficult to assess as it is not based on an estimate of 

 phytogeny for the species concerned and does not include other 

 members of the Phrynocephalus clade. 



Phrynocephalus is a characteristic element of the deserts of 

 Palaearctic Asia, like the lacertid genus Eremias and the gecko 

 assemblage including Cyrtopedion,Agamura, Bunopus, and Crosso- 

 bamon etc. Its area cladogram is not shared with these other taxa and 

 there is substantial sympatry between species and species groups. It 

 therefore seems likely that parts of the genus dispersed into at least 

 some areas of its huge range. The estimate of phylogeny suggests that 

 the ancestor of the present species occurred in the south of the present 

 distribution of Phrynocephalus, possibly within the area running 

 from western Arabia to northwestern India. This region appears to 

 contain the primary range of Trapelus, which may be the sister of the 

 Bufoniceps + Phrynocephalus clade, and Bufoniceps itself occurs in 

 northwest India. Many of the basal branches of mainPhrynocephalus 

 lineage are found wholly or partly in this area, including/! maculatus 

 (Arabia to Pakistan), P. arabicus (Arabia), some members of the P. 

 interscapularis group (S. Afghanistan, SW. Pakistan) and/! scutellatus 

 (central and eastern Iran, S. Afghanistan and SW. Pakistan). 



From this putative source area, there may have been at least a 

 triple invasion of the presently warm and arid lowland regions of 

 central Asia (Turkmenistan, Uzbekistan, Tadzhikistan, Kirgizstan, 

 southern Khazakstan): by the P. mystaceus and P. interscapularis- 

 sogdianus lineages and by the ancestor of P. golubevi and the 

 members of its sister group (shown in Figure 15, 16). The latter 

 invasion has given rise to a series of taxa in the area ( including P. 

 golubevi, P. reticulatus, P. raddei and the P. helioscopus group). 



There was then apparently eastward spread: into the Tibetan 

 region, by the ancestor of the P. theobaldi group and perhaps P. 

 forsythii, and further north into Northwest China and southern 

 Mongolia. On the basis of morphology, it is not clear whether 

 extension into the latter region represents a single invasion and 

 radiation or independent invasion by several lineages. 



A variety of additional movements by particular lineages has also 

 occurred. For instance, although within the P. helioscopus group P. 

 strauchi and P. rossikowi have relatively small allopatric ranges, P. 

 helioscopus is widespread in former Soviet Central Asia and the 

 very similar/! persicus on the soutwestern periphery of the range of 

 this species extends into eastern Turkey and Iran. P. guttatus now has 

 a broad distribution from northwest China westwards as far as the 

 north Caspian area. 



Unfortunately, there is little or no fossil record of Phrynocephalus 

 and its immediate relatives. Material assigned to Phrynocephalus 

 has been reported from the Pliocene of eastern Turkey (Zerova & 

 Chkhikvadze, 1 984), but the precise relationships of these fossils are 

 unknown and it is not even certain whether they represent a member 

 of the clade made up of all present species of Phrynocephalus or if 

 they fall outside this grouping. 



This arrangement of branches on the main lineage of the 

 Phrynocephalus-Bufoniceps clade correlates with species distinct- 

 ness. As noted, the older southern side-branches comprise very well 

 differentiated taxa, whereas later ones in central Asia often involve 

 more similar species and this trend is especially marked among the 

 relatively recent, more terminal branches in the Northwest China- 

 Southern Mongolia region, where species are very variable, their 

 boundaries poorly defined and their taxonomy often confused. 



Structural niche 



Most members of the majority of genera in Moody's Group 6 

 (Moody, 1980) climb to some extent. This is true of Laudakia, most 

 Acanthocercus andAgama s. sir., Pseudotrapelus and most Trapelus. 

 Members of the latter genus, the likely sister-group of Bufoniceps + 

 Phrynocephalus, spend a lot of time on the ground but many of them 

 also climb in bushes. In contrast to these, Bufoniceps and 

 Phrynocephalus themselves are strictly ground-dwelling, a derived 

 condition. 



There has been dispute as to whether the ancestral spatial niche of 

 Phrynocephalus is soft, wind-blown sand. This is suggested by 

 Chernov ( 1948), Whiteman (1978) and Semenov (1987), but Golubev 

 (1989) and Ananjeva &Tuniyev (1992) consider the group arose in 

 gravel and sandstone deserts. The estimate of phylogeny presented 

 here supports the former hypothesis, with Bufoniceps and three of 

 the four basal external branches of the mainPhrynocephalus lineage 

 being found in loose-sand habitats. (References to use of soft-sand 

 habitats: P. mystaceus - Ananjeva & Tuniyev, 1992; P. arabicus - 

 Arnold, 1984, Gallagher & Arnold, 1988; P. clarkorum and P. orna- 

 tus - Clark, 1992; P. luteoguttatus - Minton, 1966; P. euptilopus - 

 Smith, 1935; P. interscapularis - Ananjeva & Tuniyev, 1992; P. 

 sogdianus -Bannikov et ah, 1979). Shifts to firmer ground occurred 

 in P. maculatus and independently in the ancestor of the clade 

 containing P. scutellatus and its sister group. There was some 



