Nonetheless, the diversity-based hypothesis appears to be contradicted by the fossil record for the lineage from which Varanus descended.  The fossil record suggests that the monitor lizard lineage originated in Laurasia during the Cretaceous Period (Estes 1964; Hoffstetter 1968; Pregill et al. 1986).  Monitor lizards belong to the superfamily Varanoidea that traditionally is defined as the common ancestor of the three extant genera: Varanus, Lanthanotus, and Heloderma (McDowell and Bogert 1954; Pregill et al. 1986; Estes et al. 1988).  Varanoidea contains the families of Helodermatidae, Varanidae, Lanthanotidae, and possibly the extinct Dolichosauridae, Aigialosauridae, and Mosasauridae (McDowell and Bogert 1954; Carroll and DeBraga 1992; DeBraga and Carroll 1993; Lee 1997).  All described fossils from Varanoidea are known from locations in North America, Europe, and Asia; the only exception includes fossils from Varanus found in Tertiary and Quaternary deposits throughout much of its current distribution and the Mosasauridae found world-wide during the Late Cretaceous (see Table 1).

 However, the old theory that Serpentes is the sister taxon to mosasauroids and could belong within Varanoidea has recently been resurrected, a relationship that has long been suggested (Cope 1869; Camp 1923; McDowell and Bogert 1954), but only recently quantified using cladistics (Caldwell and Lee 1997; Lee 1997).  Mosasauroids were predatory, sea-going lizards that had highly modified flipper-like limbs and reached sizes of 15 meters (Carroll and DeBraga 1992; DeBraga and Carroll 1993).  Mosasauroids are thought to have close relations with varanoid lizards (Bauer 1892; Camp 1923; McDowell and Bogert 1954; DeBraga and Carroll 1993).  Serpentes, like Mosasauridae, achieved a world-wide distribution (Table 1); and if monitor lizards actually are closely related to these taxa, then it is possible that they could have descended from an ancestor that occurred outside of Laurasia.  Therefore, the hypothesis of a non-Laurasian origin for Varanus would appear more likely than previously thought.

 In this study, I evaluate the Laurasian fossil and Australian diversity hypotheses of varanid origin by reviewing the fossil data for varanoids and by deriving a phylogenetic hypothesis for varanoids based on a comprehensive review of the available data.  The phylogenetic hypothesis will allow testing of different biogeographic hypotheses regarding the origin of monitor lizards.  Finally, the different origin hypothesis indicated by phylogenetic studies of the extant species of Varanus are reviewed.

MATERIALS AND METHODS

 Evidence from fossil and extant taxa was collected from published studies (Pregill et al. 1986; Estes et al. 1988; Norell et al. 1992; Clos 1995; Lee 1997) and combined into a single cladistic data matrix of 308 characters, containing 290 binary and 18 multistate characters from four extant and 10 fossil taxa (Tables 2 and 3).  The two platynotan fossil genera Proplatynota and Paravaranus were assigned as the outgroups.  These two taxa have been established as belonging within Platynota, but outside of Varanoidea, making them appropriate outgroups (Estes 1964; Borsuk-Bialynicka 1984; Pregill et al. 1986; Norell et al. 1992; Lee 1997).  Proplatynota and Paravaranus were assigned to the Necrosauridae (Estes 1983; Pregill et al. 1986; Norell et al. 1992), but this family has been shown to be a nonmonophyletic assemblage and therefore has been discarded (Lee 1997; see also Estes 1983; Estes et al. 1988).

 Homologous characters found among the different studies are scored as single characters and the different states are defined to retain as much of the original information as possible.  When a particular character was used in more than one study several situations occurred.  Often, the studies were complementary, having different taxa sampled and/or additional character states.  In these cases, if the character states are deemed homologous and not in conflict then the characters are combined.  When homologous characters from different studies contain character state descriptions that are noncomplementary but not in conflict, the character states are redefined so that they are consistent with all studies.  For example, the numbers of vertebrae in characters 204 and 205, were categorized differently among some of the various studies (e.g. characters states of ‘eight or less’ and ‘seven’ combined to ‘eight or less’).  Homologous character states in direct conflict among different studies are scored as unknown, as was found only in character 6 for Estesia mongoliensis.  Otherwise, no modifications have been made to the other character codes or descriptions incorporated from the different studies.

 Phylogenetic relationships are examined using Swofford's (1993) computer program Phylogenetic Analysis Using Parsimony (PAUP) version 3.1.1 with the branch-and-bound search.  All characters are unweighted and multistate characters are unordered, which is equivalent to not assuming the presence of transformational or morphoclinal series.  Character polarization is performed by PAUP using the parsimony criterion on the shortest tree with the platynotan outgroups.  The 10,000 random-trees and the g1 statistic, which tests for the presence of significant phylogenetic signal within the data, are generated in PAUP (Hillis 1991; Hillis and Huelsenbeck 1992).  Support for individual branches is assessed using bootstrap resampling (Felsenstein 1985a; Swofford 1993) and decay indices (Bremer 1988; Bremer 1994).  The bootstrap analysis was performed using 10,000 resampled data sets with branch-and-bound searches in PAUP.  Decay indices were calculated in PAUP with constraint trees for all branches found in the parsimony tree and branch-and-bound searches keeping the shortest trees not compatible with each constraint.  Bootstrap numbers are the percentage of times each branch is present in the most parsimonious topologies from the resampled data sets (Felsenstein 1985a), whereas decay indices are the number of extra steps required from the most parsimonious topology to find an alternative topology where a particular branch is not present (Bremer 1988).

 Unfortunately, many of the fossil taxa are known only from incomplete skeletal material and thus there are large numbers of unknown character states in the data matrix (see Table 3).  Missing data have been shown to be problematic in parsimony analyses, often causing decreased branch support as indicated by the low bootstrap percentages and decay indices (Wilkinson 1995; Wilkinson and Benton 1995).  To assess the influence of the large amount of missing data and test the robustness of the phylogenetic results, a reanalysis was conducted with only characters that are parsimony informative and have more than half of the taxa scored.  This reduced the data matrix from 308 to 74 characters.

 Three alternative phylogenetic hypotheses that would support a non-Laurasian origin were investigated: Varanus is the sister taxon to the group containing Serpentes and mosasauroids; Varanus and Serpentes are sister taxa; and Varanus and the Mosasauroidea are sister taxa.  Alternative phylogenies were compared with the most parsimonious phylogeny using Wilcoxon signed rank test with two-tailed probabilities (Templeton 1983; Felsenstein 1985b).  This tests whether alternative topologies show significantly less character support than would occur by chance when compared to the most parsimonious topology (Larson 1994).  The test statistics (Ts), Z statistics, and associated two-tailed probabilities were calculated for tests with n*26 by importing the output from the “compare two trees” option in MacClade version 3.06 (Maddison and Maddison 1992) into Statview version 4.51 (Abacus Concepts) and analyzed using the nonparametric Wilcoxon signed rank test.  For tests with n?25, the test statistic (Ts) and n were used to look up the critical values for two-tailed probability levels from Table 30 in Rohlf and Sokal (1981). The alternative phylogenetic topologies were produced using constraint trees generated within MacClade and analyzed using PAUP with the branch-and-bound search.

RESULTS

 The g1 statistic of -1.571 calculated from 10,000 random trees indicates the full data matrix contains significantly more phylogenetic signal than expected by chance (g1 critical value for 15 taxa and 250 binary characters is -0.20 at the p=0.01 level, more negative g1 values indicate higher significance levels; see Table 1 in Hillis and Huelsenbeck 1992).  Of the 308 characters, 86 exhibit no variation, 113 contain variation that is uninformative, and 109 provide parsimony information.  The branch-and-bound parsimony search yielded a single most parsimonious tree at 273 steps (Figure 1).

 The reduced data matrix of 74 parsimony characters that are sampled for more than half of the taxa indicates the same single most parsimonious tree as the complete data matrix with a length of 119 steps (Figure 1).  The branch support for the reduced data matrix is also very similar to that for the full data matrix: bootstrap support deviated less than five percent and decay values changed less than four steps.

 The phylogenetic hypothesis generated in this review confirms previous findings that the extant taxa Lanthanotus and Varanus are sister taxa and that Helodermatidae is less closely related to either (McDowell and Bogert 1954; Pregill et al. 1986; Estes et al. 1988; Rieppel 1992; Lee 1997).  The Serpentes and mosasauroids form a strongly supported clade, the Pythonomorpha, as indicated by the high bootstrap and decay indices (Figure 1). This relationship was first proposed by Cope (1869), but widely overlooked until its recent resurrection (Lee 1997; Caldwell and Lee 1997).  The present study also finds that the Pythonomorpha belong outside of Varanoidea in agreement with the findings of Caldwell et al. (1995), but this relationship is weakly supported; with the addition of one step to the most parsimonious topology, the Pythonomorpha can be placed within Varanoidea, as suggested by Lee (1997).

 The PAUP search using the constraint that Varanus is most closely related to Serpentes yielded four trees at 307 steps, 34 steps longer than the most parsimonious topology, and all alternative trees are rejected at the p<0.0001 level (n=48 to 52, Ts=157.5 to 255, Z=3.952 to 4.184).  The constraint of Varanus being the sister taxon to Mosasauroidea produced two trees at 308 steps, 35 steps longer than the most parsimonious topology, and are rejected at the p<0.0001 level (n=47, Ts=161, Z=4.265).  Finally, the constraint of Varanus being the sister taxon to the group containing Serpentes and Mosasauroidea produced four trees at 282 steps, nine steps longer than the most parsimonious topology (Figure 2B).  All four alternative trees are not rejectable at the p=0.05 level for the two-tailed test (p=0.108, n=22, Ts=77, Z=1.607).  Felsenstein (1985b) showed that the one-tailed values were close to the actual probabilities but not always conservative, so two-tailed values were recommended.  Therefore, because this test is conservative and the p-values are low these alternative topologies may also be considered unlikely, but not statistically rejectable.

 The ancestral reconstructions of distributions on the most parsimonious phylogeny clearly indicate that the common ancestor of Varanus was of Laurasian origin (Figure 2A).  However, reconstructing biogeographic origin on the constraint tree in which Varanus is the sister taxon to the group containing Serpentes and mosasauroids (which was almost statistically rejected) indicates a non-Laurasian origin for Varanus is possible.

DISCUSSION

 The phylogenetic hypothesis supports a Laurasian origin for the ancestral lineage of Varanus (Figure 2A).  The genera Saniwides and Saniwa are found to be the most closely related taxa to Varanus as previously hypothesized (McDowell and Bogert 1954; Lee 1997).  The closest group to Varanus in this study is Saniwides, a Late Cretaceous fossil of Mongolia, whereas the North American Saniwa of the Early and Mid Tertiary is the next most closely related lineage.  In fact, all of the taxa placed into Varanoidea in this study are found only in Laurasian deposits except for Varanus (Figure 2A and Table 1).  Therefore, the most-parsimonious reconstruction of distributions for Varanoidea indicates that a Laurasian distribution is the ancestral condition, and the extra-Laurasian distribution of Varanus is a condition derived from a Laurasian ancestor (Figure 2A).  Taxa not included in this study are unlikely to change this finding because all of the remaining described varanoid fossils are also from Laurasian deposits.  These fossils include Paleosaniwa and Parasaniwa, both from Late Cretaceous deposits of North America and believed to be closely related to Saniwa (McDowell and Bogert 1954), and Iberovaranus from Miocene deposits of Spain (Hoffstetter 1969).  Other missing fossils that have close but ambiguous relationships with Varanoidea include dolichosaurids from Cretaceous deposits of Europe (McDowell and Bogert 1954) and many “necrosaurid” grade fossils (e.g. Necrosaurus, Eosaniwa, Provaranasaurus, Colpodontoaurus, and Parviderma) from the upper Cretaceous of Asia and the Tertiary of North America and Europe (Estes 1983; Borsuk-Bialynicka 1984).

 A possible problem with drawing any conclusions from the varanoid fossil record is that there might be a geographic bias in fossil preservation during the Cretaceous.  This bias is very real in that many of the fossil varanoids are known from Mongolian deposits from the Late Cretaceous (Table 1), and it is well established that the Mongolian Late Cretaceous faunas show unparalleled preservation and exceed all other known Late Cretaceous localities in abundance, diversity, and quality of terrestrial fossil vertebrate remains (Dashzeveg et al. 1995).  This problem with the varanoid fossil record is countered by the fact that both Serpentes and mosasauroids have an abundance of Cretaceous fossil localities outside of Mongolia and Laurasia.  In fact, the fossil record for Serpentes indicates a Gondwanaland origin for the group and shows a complete absence from Laurasia in the Cretaceous (Rage 1987).  In addition, varanoid fossils in Europe and North America are not uncommon (Estes 1963; Pregill et al. 1986; Caldwell et al. 1995).  Thus, the complete absence of varanoid fossils from localities outside of Laurasia during the Cretaceous is unlikely to be the result of a bias in the fossil record.

 The alternative hypothesis, that Varanus originated in Australia, where the highest extant diversity occurs, is conceivable if this genus is most closely related to the mosasauroids, Serpentes or their common ancestor.  Mosasauridae and Serpentes both achieved world-wide distributions (Table 1), suggesting the possibility that the ancestral form also occurred in many regions (Figure 2A).  Of these alternative hypotheses, the only one that could not be rigorously rejected with the two-tailed test is that Varanus is the sister taxon to the common ancestor of mosasauroids and Serpentes, the Pythonomorpha.  However, this Varanus and Pythonomorpha sister relationship appears to be highly unlikely for a variety of reasons.  For example, the fossil record for Serpentes extends back to the Early Cretaceous of North Africa and possibly into the Late Jurassic (Rage 1987).  This would indicate that the divergence between the mosasauroids and Serpentes must have occurred at least by the Late Jurassic.  In this scenario, Varanus would have had to split from this lineage prior to this date (see Figure 3).  This is improbable because Varanus does not appear in the fossil record until the Early Miocene and possibly the Eocene (Hoffstetter 1968; Clos 1995), leaving a fossil gap of over 70 million years between the split of the Varanus lineage from Pythonomorpha and the first appearance of Varanus (Figure 3).  In addition, the Australian continent split from Gondwana during the Late Cretaceous (Metcalfe 1991), and remained geographically isolated until the Miocene when it collided with the Philippine Sea Plate approximately 25 mya (Hall 1996).  However, the first known Australian Varanus fossils are not found until the Mid-Miocene after Australia was connected with Southeast Asia (Stirton et al. 1961; Hecht 1975).  Therefore, the fossil evidence does not support the hypothesis that Varanus originated on Gondwana and found refuge on Australia during the Early Tertiary and is more consistent with the hypothesis that Varanus descended from a Laurasian ancestor and invaded Australia from Southeast Asia during the Miocene.

 Phylogenetic studies of the variation found among extant species of Varanus provide support for both the Laurasian and Australian origin hypotheses, but the Laurasian origin hypothesis appears to be more widely supported among the different studies.  Mertens (1942) was convinced that the center of origin for Varanus was in Southeast Asia, based largely on the distribution of subgenera and morphological variation.  Mertens (1942) argued that Asia and the Indonesian Archipelago have the greatest number of subgenera (eight, compared to two in Australia, of the nine recognized subgenera), indicating these areas show the greatest amount of morphological variation and represent the origin of the distribution and development of Varanus.  Although, many of the subgeneric classifications for Varanus have changed since Mertens (1942) the same conclusion can be drawn, that Asia and the Indonesian Archipelago have eight subgenera compared to two in Australia (Ziegler and Böhme 1997).  Mertens (1942) believed that from Southeast Asia varanids could have dispersed with little difficulty into the regions of their current distribution during the Early Tertiary.

 Karyotypic and protein electrophoretic data were next used to hypothesize the phylogenetic relationships among monitor species and also suggest a Laurasian origin (King and King 1975; Holmes et al. 1975; King 1990).  These studies looked at variation in karyotypes among 25 species and in the isozyme lactate dehydrogenase (LDH) among 18 species.  This evidence was used to predict that Australia was colonized from Asia in two independent invasions (King and King 1975; Holmes et al. 1975; King 1990), thus supporting the Laurasian hypothesis.

 Other research has focused mainly on morphology, particularly that of the hemipenis (Branch 1982; Böhme 1988; Card and Kluge 1995).  A cladogram constructed for 30 species of Varanus from 13 hemipenial characters, including external morphology and the internal bony elements (Böhme 1988), largely agrees with previous monitor hemipenial work (Branch 1982). Lung morphology has also been used in an analysis of 8 characters for 18 varanids (Becker et al. 1989).  The hemipenial and lung cladograms are similar in that they both suggest that the small pygmy monitors from Australia are basal and the initial radiation of Varanus was out of Australia into its present distribution.  However, due to the limited number of hemipenial and lung characters relative to the number of species, little character support is provided for the relationships within or among the major clades of Varanus, making it difficult to draw strong conclusions.

 The data from the karyotype, lysozyme, and hemipene as well as additional characters were analyzed for 23 species of Varanus (Sprackland 1991).  This analysis included 57 characters consisting of skull morphology, scalation, external morphology, and color pattern.  A single cladogram was reported by Sprackland (1991), but Card and Kluge (1995) reanalyzed Sprackland's data and found three cladograms nine steps shorter than Sprackland’s (1991) tree (211 versus 202 steps).  The consensus total-evidence tree presented by Card and Kluge (1995) indicates that the Australian species sampled are monophyletic and derived from within the Southeast Asian and Indonesian species, giving more support for the Laurasian origin hypothesis.

 Baverstock et al. (1993) used microcomplement fixation (MC'F) analysis of albumin to compute immunological distances (IDs) for 30 Varanus species.  The MC'F data indicate four distinct clades: pygmy Australian species, African species (including V. griseus), Southeast Asian species, and large Australian species.  Based on an albumin molecular clock (Wilson et al. 1977), Baverstock et al. (1993) conclude that the large Australian varanids (the V. gouldii lineage) moved to Australia from Southeast Asia no more than 13 million years ago, which again supports the fossil Laurasian origin hypothesis.   Unfortunately, this tree was not rooted with an outgroup because Heloderma would not cross-react with Varanus antisera and Lanthanotus was not available, making any conclusions tentative.

 Most recently, Fuller et al. (1998) addressed the biogeographic origins of Varanus with 12S rRNA sequences from 21 species, rooting the phylogeny with both Heloderma and Lanthanotus.  They found that the Australian species are monophyletic and nested within the Asian and Indonesian species, supporting the hypothesis that the Australian species are derived from an Asian source in agreement with the Laurasian origin based on the fossil record.

 Thus, the distribution of subgenera, karyotype, lysozyme, combined, MC’F, and 12S rRNA phylogenetic evidence are inconsistent with an Australian origin and support the Laurasian origin, whereas the studies using just hemipene and lung morphology agree more with the Australian origin.  The data sets in support of the Laurasian origin contain a much larger total number of independent characters than the two data sets that suggest an Australian origin.  Therefore, the Laurasian origin appears to be more widely supported by the phylogenetic evidence from the extant species of Varanus.

 The Australian origin hypothesis is based on the argument that the age of occupation in a geographic region is positively correlated with the species diversity, but this assumption is replete with potential problems.  Many factors that are not necessarily a function of the age of occupation can generate and maintain high species diversity, such as spatial heterogeneity, trophic complexity, resource competition, and community productivity (see chapters in Ricklefs and Schluter 1993).  For example, Simpson (1953) argues that a group must possess ecological access before it is able to shift into a new, distinct adaptive zone that could, in turn, promote speciation.  Ecological access means that the required resources in the new adaptive zone are available for exploitation (not already being used by superior competitors).  Thus, if fewer competitors were present in Australia than in other regions in which Varanus occurs, as Pianka (1969) suggested, then differences in varanid diversity would be unrelated to the age of occupation.  In addition, a number of other taxa also have extremely high diversity on Australia relative to other regions in the world.  For example, Australia has the highest diversity of lizard (Schall and Pianka 1978), termite (Whitford et al. 1992), and ant species (Greenslade and Greenslade 1983; Morton and Davidson 1988).  Many ecological and historical circumstances have been proposed for the great biotic diversity of Australia (see Morton 1993).  The high diversities of varanids, and other taxa, is thus not necessarily indicative of particularly long residence on that island continent.

 In conclusion, the most parsimonious hypothesis regarding the origin of Varanus based on fossil evidence is that the genus originated from a Laurasian ancestor and that the current distribution outside of Asia is a derived condition.  Although, the alternative hypothesis that Varanus descended from a common ancestor shared with Serpentes and mosasauroids, and thus could have originated outside of Laurasia, is not rigorously rejected, this hypothesis is unsupported by the fossil record.  Most phylogenetic studies of extant Varanus indicate that the Australian species are derived from the more basal Asian species, which is in accord with the Laurasian origin hypothesis.  Therefore, the Australian (Gondwana) origin hypothesis is largely unsupported and monitor lizards appear to have originated in Laurasia.

ACKNOWLEDGMENTS

I would like to thank Hans-Georg Horn, Allan Larson, and Jonathan Losos for providing many helpful suggestions on earlier drafts of this manuscript.  I would especially like to thank Hans-Georg Horn for kindly translating the fourth chapter (pp. 56-57) of Mertens (1942).  This work was supported by grants from the National Science Foundation, DEB-9318642 and 9701820.

REFERENCES

Bauer, G. H. C. L.  1892.  On the morphology of the skull of the Mosasauridae. Journal of
        Morphology 7(1): 1-22.

Baverstock, P. R., D. King , M. King, J. Birrell, and M. Kreig.  1993.  The evolution of species  of the
        Varanidae: microcomplement fixation analysis of serum albumins.  Australian  Journal of Zoology
        41: 621-638.

Becker, H. O., W. Böhme, and S. F. Perry.  1989.  Die lungenmorphologie der Warane (Reptilia:
        Varanidae) und ihre Systematisch-stammesgeschichtliche bedeutung.  Bonner  Zoologische
        Beitrage 40: 27-56.

Bennett, D.  1995.  A Little Book of Monitor Lizards.  Viper Press, Aberdeen, Great Britain.

Böhme, W.  1988. Zur genitalmorphologie der Sauria: funktionelle und stammesgeschichtliche
        aspeckte.  Bonner Zoologische Monographica  27: 1-176.

Böhme, W. and T. Ziegler.  1997.  Varanus melinus sp. n., ein neuer Waran aus der V. indicus- gruppe
        von den Molukken, Indonesien.  Herpetofauna 19(111):26-34.

Borsuk-Bialynicka, M.  1984.  Anguimorphans and related lizards from the Late Cretaceous of  the Gobi
        Desert, Mongolia.  Palaeontologia Polonica 46: 5-106.

Bremer, K.  1988.  The limits of amino acid sequence data in angiosperm phylogenetic  reconstruction.
        Evolution 42: 795-803.

Bremer, K.  1994.  Branch support and tree stability.  Cladistics 10: 295-304.

Branch, W. R.  1982.  Hemipeneal morphology of platynotan lizards.  Journal of Herpetology  16:
        16-38.

Caldwell, M. W., R. L. Carroll, and H. Kaiser.  1995. The pectoral girdle and forelimb of  Carsosaurus
        marchesetti (Aigialosauridae), with a preliminary phylogenetic analysis of  mosasauroids and
        varanoids.  Journal of Vertebrate Paleontology 15(3): 516-531.

Caldwell, M. W., and M. S. Y. Lee.  1997.  A snake with legs from the marine Cretaceous of the
        Middle East.  Nature 386: 705-709.

Camp, C. L.  1923.  The classification of lizards.  Bulletin of the American Museum of Natural  History
        48: 289-481.

Card, W., and A. G. Kluge.  1995.  Hemipeneal skeleton and varanid lizard systematics.  Journal  of
        Herpetology 29(2): 275-280.

Carroll, R. L., and M. DeBraga.  1992.  Aigialosaurs: Mid-Cretaceous varanoid lizards.  Journal  of
        Vertebrate Paleontology 12(1): 66-86.

Clos, L. M.  1995.  A new species of Varanus (Reptilia: Sauria) from the Miocene of Kenya.   Journal
        of Vertebrate Paleontology 15(2): 254-267.

Cogger, H. G.  1992.  Reptiles and Amphibians of Australia.  Reed, Sydney.

Cope, E. D.  1869.  On the reptilian orders Pythonomorpha and Streptosauria.  Proceedings of  the
        Boston Society of Natural History 12: 250-266.

Dashzeveg, D., M. J. Novacek, M. A. Norell, J. M. Clark, L. M. Chiappe, A. Davidson, M. C.
        McKenna, L. Dingus, C. Swisher, and P. Altangerel.  1995.  Extraordinary preservation  in a new
        vertebrate assemblage from the Late Cretaceous of Mongolia.  Nature 374: 446- 449.

DeBraga, M., and R. L. Carroll.  1993.  The origin of mosasaurs as a model of  macroevolutionary
        patterns and processes.  Evolutionary Biology, Volume 27 (M. K.  Hecht et al., eds.), pp.
        245-322.  Plenum Press, New York.

Estes, R.  1964.  Fossil vertebrates from the Late Cretaceous Lance Formation, eastern  Wyoming.
        University California Publications in Geological Sciences 49: 1-186.

Estes, R.  1983.  Sauria Terrestria, Amphisbaenia.  Handbuch der Paläoherpetologie.  Teil 10A.
        Gustav Fischer Verlag, Stuttgart.

Estes, R., K. de Queiroz, and J. Gauthier.  1988.  Phylogenetic Relationships within the  Squamata.
        in Phylogenetic Relationships of the Lizard Families (R. Estes and G. Pregill,  eds.), pp. 119-270.
        Stanford University Press, Palo Alto, California.

Felsenstein, J.  1985a.  Confidence limits on phylogenies: an approach using the bootstrap.  Evolution
        39: 783-791.

Felsenstein, J.  1985b.  Confidence limits on phylogenies with a molecular clock.  Systematic  Zoology
        34: 152-161.

Fuller, S., P. Baverstock, and D. King. 1998.  Biographic origins of goannas (Varanidae): a  molecular
        perspective.  Molecular Phylogenetics and Evolution 9(2): 294-307.

Greenslade, P. J. M., and P. Greenslade. 1983.  Ecology of soil invertebrates.  in Soils: An  Australian
        viewpoint.  (CSIRO Division of Soils, ed.), pp. 645-69.  Academic Press,  London; CSIRO,
        Melbourne.

Hall, R.  1996.  Reconstructing Cenozoic Southeast Asia.  in Tectonic Evolution of Southeast  Asia (R.
        Hall and D. Blundell, eds.), pp. 153-184.  Geological Society Special  Publications 106, London,
        England.

Harvey, M. B., and D. G. Barker.  1998.  A new species of blue-tailed monitor lizard (genus  Varanus)
        from Halmahera Island, Indonesia.  Herpetologica 54(1): 34-44.

Hecht, M. K.  1975.  The morphology and relationships of the largest known terrestrial lizard,
        Megalania prisca Owen, from the Pleistocene of Australia.  Proceedings of the Royal  Society of
        Victoria 87: 239-250.

Hillis, D. M.  1991.  Discriminating between phylogenetic signal and random noise in DNA  sequences.
        in Phylogenetic Analysis of DNA Sequences (M. M. Miyamoto and J.  Cracraft, eds.),  pp.
        278-294.  Oxford University Press, Oxford.

Hillis, D. M., and J. P. Huelsenbeck.  1992.  Signal, noise and reliability in molecular  phylogenetic
        analysis.  Journal of Heredity 83: 189-195.

Hoffstetter, R.  1968.  Presence de Varanidae (Reptilia: Sauria) dans le Miocene de catalogue.
        Considerations due l’histoire de la famille.  Bulletin du Museum Nationale d’Histoire,  Naturelle
        40: 1051-1064.

Holmes, R. S., M. King, and D. King.  1975.  Phenetic relationship among varanid lizards based  upon
        comparative electrophoretic data and karyotypic analyses.  Biochemical  Systematics and Ecology 3:
        257-262.

King, M.  1990.  Chromosomal and immunogenetic data: a new perspective on the origin of  Australia’s
        reptiles.  in Cytogenetics of Amphibians and Reptiles.  Advances of Life  Sciences. (E. Olma, ed.),
        pp. 153-180.  Birkhauser Verlag, Basel.

King, M., and D. King.  1975.  Chromosomal evolution in the lizard genus Varanus (Reptilia).
        Australian Journal of Biological Sciences 28: 89-108.

Larson, A. 1994.  The Comparison of Morphology and Molecular Data in Phylogenetic  Systematics. in
        Molecular Ecology and Evolution: Approaches and Applications (B.  Schierwater, B. Streit, G.P.
        Wagner, and R. DeSalle, ed.),  pp. 371-390.  Birkhauser  Verlag, Basel.

Lee, M. S. Y.  1997.  The phylogeny of varanoid lizards and the affinities of snakes.   Philosophical
        Transactions of the Linnean Society 352: 53-91.

Maddison, W. P., and D. R. Maddison.  1992. MacClade: Analysis of phylogeny and character
        evolution, version 3.0.  Sinauer, Sunderland, Massachusetts.

McDowell, S. B., and C. M. Bogert.  1954.  The systematic position of Lanthanotus and the  affinities
        of the anguinomorphan lizards.  Bulletin of the American Museum of Natural  History 105(1):
        1-142.

Mertens, R. 1942.  Die familie der Warane (Varanidae).  Erster teil: allgemeines.  Abhandlungen  der
        Senckenbergischen Naturfoschenden Gesellschaft 462: 1-116.

Metcalfe, I.  1991.  Southeast Asian terranes: Gondwanaland origins and evolution.  in  Gondwana
        Eight: Assembly, Evolution, Dispersal (R. H. Findlay, R. Unrug, M. R. Banks,  J. J. Veevers,
        eds.), pp. 181-200.  A. A. Balkema Publishers, Rotterdam, Netherlands.

Morton, S. R.  1993.  Determinants of diversity in animal communities of arid Australia.  in  Species
        Diversity in Ecological Communities: Historical and Geographical Perspectives  (R. E. Ricklefs and
        D. Schluter, ed.), pp. 159-169.  University of Chicago Press,  Chicago, London.

Morton, S. R., and C. D. Davidson.  1988.  Comparative structures of harvester ant communities  in
        arid Australia: A new hypothesis.  American Naturalist 132: 237-256.

Norell, M. A., M. C. McKenna, and M. J. Novacek.  1995.  Estesia mongoliensis, a new fossil
        varanoid from the Late Cretaceous Barun Goyot Formation of Mongolia.  American  Museum
        Novitates 3045: 1-24.

Pianka, E.  1969.  Habitat specificity, speciation, and species density in Australian desert lizards.
        Ecology 50:498-502.

Pregill, G., J. Gauthier, and H. Greene.  1986.  The evolution of helodermatid squamates, with
        description of a new taxon and an overview of Varanoidea.  Transactions of the San  Diego Society
        of Natural History 21: 167-202.

Rage, J. C. 1987.  Fossil History.  in Snakes: Ecology and Evolutionary Biology (R. A. Seigel, J.  T.
        Collins, and S. S. Novak, eds.), pp. 49-76.  MacMillan, New York.

Ricklefs, R. E., and D. Schluter.  1993.  Species Diversity in Ecological Communities: Historical  and
        Geographical Perspectives.   University of Chicago Press, Chicago, London.

Rieppel, O. 1992.  The skeleton of a juvenile Lanthanotus (Varanoidea).  Amphibia-Reptilia 13:  27-34.

Rohlf, F. L., and R. R. Sokal.  1981.  Statistical Tables.  Freeman WH, San Francisco.

Schall, J. J., and E. R. Pianka.  1978.  Geographical trends in the numbers of species.  Science  201:
        679-686.

Simpson, G. G.  1953.  The major features of evolution.  Columbia University Press, New York.

Sprackland, R. G.  1991.  The origin and zoogeography of monitor lizards of the subgenus  Odatria
        Gray (Sauria: Varanidae): a re-evaluation.  Mertensiella 2: 240-252.

Stirton, R. A., R. T. Tedford, and R. H. Miller.  1961.  Cenozoic stratigraphy and vertebrate
        paleontology of the Tirari desert, South Australia.  Records of the South Australian  Museum 14:
        19-61.

Storr, G. M., L. A. Smith, and R. E. Johnstone.  1983.  Lizards of Western Australia II.  Dragon  and
        Monitors.  Western Australian Press, Australia.

Swofford, D. L.  1993.  PAUP: Phylogenetic analysis using parsimony, version 3.1. Illinois  Natural
        History Survey, Champaign.

Templeton, A. R.  1983. Phylogenetic inference from restriction endonuclease cleavage site  maps with
        particular reference to the evolution of humans and the apes.  Evolution 37:  221-244.

Whitford, W. G., J. A. Ludwig, and J. C. Noble.  1992.  The importance of subterranean termites  in
        semi-arid ecosystems in south-eastern Australia.  Journal of Arid Environments 22: 87- 91.

Wilkinson, M.  1995.  Coping with abundant missing entries in phylogenetic inference using
        parsimony.  Systematic Biology 44: 501-514.

Wilkinson, M., and M. J. Benton.  1995.  Missing data and rhynchosaur phylogeny.  Historical
        Biology 10: 137-150.

Wilson, A. C., S. S. Carlson, and T. J. White. 1977.  Biochemical evolution.  Annual Review of
        Biochemistry 46: 573-639.

Ziegler, T. and W. Böhme. 1997.  Genitalstrukturen und paarungsbiologie bei squamaten  Reptilien,
        speziall den Platynota, mit bemerkungen zur systematik.  Mertensiella 8: 1- 210.

FIGURE 1.  The most parsimonious topology estimated from both the full data matrix (273 steps; Consistency Index (CI)=0.813; Retention Index (RI)=0.710) and the reduced data matrix of 74 parsimony informative characters that are sampled for more the half of the taxa (119 steps; CI= 0.681; RI=0.719).  Numbers above branches in bold are the bootstrap and decay values for the full data matrix, while the numbers below the branches are for the reduced data matrix.  The bootstrap percentages, listed first, are from 10,000 replicates with the branch-and-bound search in PAUP 3.1.1 (Swofford 1993).  Only bootstraps above 50 percent are shown.  Decay values are represented by the numbers in brackets.  # identifies fossil taxa.