A response to D. McCarthy (2005a) Biogeographical and geological evidence for a smaller, completely-enclosed Pacific basin in the Late Cretaceous. Journal of Biogeography, 32, 2161–2177 The recent paper by McCarthy (2005a) follows McCarthy (2003) in explaining apparently problematic biotic distributions from the Pacific-bordering continents using a somewhat extreme version of the expanding Earth hypothesis. By shrinking the Earth to remove the entire global ocean floor, the oldest extant pieces having formed > 180 Ma in the Early Jurassic, the hypothesis brings what are now quite distal locations into apparent close proximity (Fig. 1, which is a redraft of Fig. 3 in McCarthy, 2003). It is pleasing to see alternative viewpoints being aired in the scientific literature, but, I believe, the reconstructions are geometrically impossible and many of the underpinning arguments are either flawed or simply wrong – hence the comment. Note also that the biological aspects of McCarthy's (2003) paper were eloquently countered by Briggs (2004). Apparent pan-Pacific faunal linkages as shown in McCarthy (2003). (a) Earth reduced in size to remove all of the ocean-floor crust in line with the proposal of McCarthy (2003, 2005a). (b) ‘Classic’ Gondwana plotted on an orthogonal projection (= distal satellite view) using the GMAP software package (Torsvik & Smethurst, 1999). As with (a), the focal point of the image is set at 40° E, 20° S. The continents are placed in their pre-breakup positions (c. 160 Ma), with Africa plotted in its present-day position. To simplify matters, no attempt has been made to portray the other crustal blocks occupying Tethys, nor the form of Greater India (Ali & Aitchison, 2005). McCarthy's reconstructions are based on the ocean crust age poster of Mueller et al. (1993) and Müller et al. (1997) (first author is the same in both works). On this map, various features (continents, ocean fracture zones, hot-spot traces, mid-ocean ridges and, of course, ocean floor ages) are shown on a Mercator projection (Fig. 2; Fig. 1 in McCarthy, 2005a). From the image, McCarthy has taken the continental edges of eastern Australia and East Asia (China) and placed them jigsaw-like fashion against central west South America and the Pacific coast of Mexico–Alaska. This configuration is McCarthy's view of Late Triassic (228–200 Ma) Earth: no oceans, all continent. By placing Australia–Asia against the Americas, the surface that the geological community conventionally believes was occupied by the vast Panthalassa Ocean (e.g. Barron et al., 1981; Schettino & Scotese, 2005) is eliminated. Ocean crust age map (Mueller et al., 1993; Müller et al., 1997) as shown in McCarthy (2003, 2005a). At the bottom, the three globe views (which physically expand with time) are McCarthy's reconstructions for the Pacific Basin in the Late Triassic, Late Cretaceous and Present-day. Having worked on Gondwana reconstructions (Ali & Aitchison, 2005), as well as plate modelling of the Tethys–SE Asia–W/SW Pacific regions for some time (e.g. Ali & Hall, 1995; Hall et al., 1995; Ali & Aitchison, 2000, 2004; Abrajevitch et al., 2005; Whattam et al., 2005), I found the proposal stimulating, principally because of the questions that immediately sprang to mind. For instance, if all of the ocean floor is eliminated and the Earth's surface is shrunk to c. 40% of its present value (sea occupies c. 70% of the spherical plane, but for this analysis we are concerned with the ocean basins proper and not the epicontinental seas, for example the North Sea, Hudson Bay, southwest South China Sea, Sea of Okhotsk, etc.), the planet's radius must have been c. 63% of today's value (6378 km) (Fig. 3a). As a consequence, the Earth must have ballooned by about four times since the Late Triassic (the cube of 0.63 is 0.25). On this point alone the lines of discussion are many. For example, how was the volume increase achieved and did it involve a mass gain? If the Earth's core 200 Ma had its present-day radius (c. 3500 km), what was the thermal gradient across the planet's mantle and crust (the two shells together would then have been less than 400 km thick)? It is noteworthy that if today's Africa was placed on the Late Triassic globe, the S–N length of the continent along 20° E (Cape Argulhas, c. 34° S, to north-east Libya, c. 33° N = 7440 km) would occupy c. 106° of latitude. Furthermore, without plate tectonics and one of its key associated processes, ophiolite/island arc obduction prior to 200 Ma, how did ocean floor volcanic rocks with their fossil-rich pelagic cover sequences find their way into the cores of the pre-Late Triassic mountain belts (e.g. Appalachian–Caledonian, Ural and Central Asian)? Of more interest, however, is the Late Triassic reconstruction, as it is possible to examine its validity. McCarthy's positioning of eastern Australia against western South America is fascinating, because the geologists modelling Gondwana have for a long time connected eastern South America and southern–western Australia via Africa–Antarctica–India (Smith & Hallam, 1970; Powell et al., 1988; Schettino & Scotese, 2005) (see Fig. 3b). Indeed the basic form of the super-continent has been around since Alfred Wegner's groundbreaking publication in 1912 (see Holmes, 1965, Chapter 31, which reviews the contributions of Wegner, Alexander du Toit, Lester King, Sam Carey, J. Tuzo Wilson, as well as his own proposal). McCarthy's Gondwana, shown on the website at http://www.4threvolt.com (McCarthy, 2005b), follows the conventional configuration. The GMAP computer program (Torsvik & Smethurst, 1999) makes it possible to draw properly crustal blocks on an Earth-like surface and to display them in a variety of projections, although it is clearly acknowledged that the software does not allow the Earth's radius to be changed. As a starting point, refer to Fig. 3(b), which shows the accepted Gondwana reconstruction (which in this case has the core continent Africa placed in its present-day position, with Australia out towards the east) plotted on an orthogonal projection (i.e. as the Earth would be seen from the Moon). Figure 4 builds on this image, although it is necessary to switch to a Galls (cylindrical) projection to show McCarthy's full continental linkage. Following McCarthy (2003, 2005a), a second Australia(-2) is shown, positioned just to the west of central South America; central eastern North America is located to the west of northern Africa (the standard position); the eastern end of Eurasia is placed against Mexico–Alaska as McCarthy (2003, 2005a) proposed. Curiously, if one accepts the expanding Earth hypothesis and the Australia–South America connection, an equally good fit can be achieved by placing the Great Australian Bight (south coast of the continent) against north-west South America (Fig. 5a). Indeed, there are several particularly tight plate configurations that can be made by simply matching continental outlines: South Australia–West Africa (Fig. 5b); and, as a logical consequence, eastern North America–eastern Antarctica, eastern North America–north-west South America. Of course, in all of these examples (which are certainly not proposals), other geological information would argue against such correlations. Reconstruction of McCarthy's proposal ‘properly’ drafted using the GMAP software using the various stencils for Gondwana, Australia, North America (one with Greenland), and Eurasia. As in Fig. 3, no attempt has been made to portray the Tethyan crustal blocks nor the form of Greater India. Examples of exotic continental linkages based simply upon the matching of continental shelf lines: (a) southern Australia and north-western South America, and (b) South Australia and western Africa. The figures were drafted using the GMAP software. At this point it is worth mentioning an important issue that is not explained by McCarthy. It is widely accepted that the present-day western edges of North and South America and the eastern sides of Australia and East Asia looked somewhat different in the Late Triassic. Since 200 Ma, all four margins have been substantially modified through: (1) the accretion of micro-continental blocks and intra-oceanic island arc terranes, and/or (2) the generation of continental arc material (e.g. Schermer et al., 1984; Howell, 1985, 1995; Vaughan et al., 2005). This explains why northern Central America actually sits on top of northern South America. In fact, the cratonic edge of North America (shown on the figures) is a more useful guide for the western edge of the continent in the Late Triassic. A second North America(-2) can also be shown, and is linked in the customary manner (e.g. Bullard et al., 1965; Torsvik et al., 2001) to north-west Europe via Greenland; the Labrador Sea and north-east Atlantic Ocean were both closed in the Late Triassic, and did not begin opening until the early Cenozoic. The positions of Australia-1 and Australia-2 are telling. If the orientations of a great circle drawn between equivalent points on the west coasts of the two Australias are compared, there is an angular difference of 68°. Critically, wrapping the Gondwana continents around a much smaller globe would greatly increase this angle. Asia's positioning creates very real problems owing to ‘overlap’ of its south-eastern part as western south-east Asia (Indochina south to Sumatra = western Sundaland) extends across New Guinea–north-east Australia. Eastern Sundaland (Borneo–western Java), which is not part of the GMAP ‘Eurasiayoung’ stencil, would sit over Peru and western Bolivia. On McCarthy's much-reduced Late Triassic globe, this part of Eurasia has to have sat farther east and north, probably in the Amazon Basin. The addition of a second North America(-2) to McCarthy's string of continents, western Eurasia being linked via Greenland, is in accordance with traditional modelling. It is impossible to describe this misfit of North America-1 and -2. It simply cannot be accommodated by any surface resembling that associated with a sphere. For both Australia and North America we are necessarily compelled to have both continents at two places at once. Whilst it is possible for quantum physicists to make such arguments regarding some of the subatomic particles, it is a little more difficult with continents. Similarly, Asia, specifically the south-eastern part, cannot be fitted into McCarthy's model. Perhaps the best analogy for McCarthy's Late Triassic reconstruction is the Reutersvärd–Penrose Impossible Triangle (Fig. 6) (Draper, 1978) and its derivative, M.C. Escher's ‘Waterfall’ (e.g. Locher, 1972). Sectors of the triangle and watercourse appear plausible, but when examined as a whole, both constructs are logically impossible. At this stage, the onus really is on McCarthy to provide proper figures, preferably on orthogonal and cylindrical projections for an Earth with a radius c. 0.63 times that of the present-day value, thereby allowing the community to assess his Late Triassic configuration. If the basic puzzle does not work, it is unlikely that the Earth Scientists will even register the radical ideas he is proposing. The Reutersvärd–Penrose Impossible Triangle (drawn by the author) is a possible analogy for McCarthy's Late Triassic reconstruction. Small elements of the figure appear physically plausible, but the entire construct is not logically viable. No attempt has been made to counter the various geological ‘arguments’ and, indeed, misrepresentations proffered by McCarthy (2003, 2005a). There simply is not enough space, and it would only involve summarizing the contents of various textbooks (e.g. Moores & Twiss, 1995; Windley, 1995; Kearey & Vine, 1996) and innumerable journal papers. However, issue is taken with statements relating to work I have been associated, specifically the ‘Stable Asia Shallow Inclination Problem’ (McCarthy, 2005a, referring to Ali et al., 2003; Ali & Aitchison, 2004). It is a fact that anomalously shallow inclinations are regularly reported from palaeomagnetic studies of Upper Cretaceous and Cenozoic rocks in south Central Asia (and elsewhere and for other intervals), but all of the disturbed results are from sedimentary formations (e.g. Chauvin et al., 1996; Gilder et al., 2001; Dupont-Nivet et al., 2002). As Ali & Aitchison (2004) have shown, the limited data that exist for volcanic rocks in the region indicate that the ‘problem’ arises from sediment compaction and/or the flow regime at the time of deposition (see also Tauxe, 2005, and Yan et al., 2005). Volcanic (and intrusive) rocks are, however, different. First, their magnetization is acquired some time after crystallization (800 to > 1000 °C), when the temperature of the magnetic grains falls below the ‘blocking’ temperature (c. 575 °C for magnetite; c. 675 °C for haematite). Second, because they are crystalline, igneous rocks cannot easily be squashed unless huge forces are applied. Hence, the magnetic remanences of volcanic rocks do not show the inclination shallowing effect (e.g. Bazhenov & Mikolaichuk, 2002; Huang et al., 2005; Tauxe, 2005). Critically, the ‘Stable Asia Shallow Inclination Problem’ can be worked around (Ali & Aitchison, 2004). This is contrary to what is implied by McCarthy's somewhat slanted discussion of the issue, which uses the phenomenon as a basis for dismissing the idea that some crustal blocks have moved huge angular distances relative to one another (e.g., India and Eurasia), a feature that cannot easily be accommodated in the expanding Earth hypothesis. Plate-tectonic theory, which has now been around for forty-odd years, currently offers the best explanation for how the Earth's shell has evolved since the ‘Explosion of Life’c. 540 Ma (marked by the start of the Cambrian), and for a substantial interval before (e.g. Li & Powell, 2001; Torsvik, 2003). Since its introduction in the 1960s, new techniques [for instance seismic tomography, which has enabled subducted lithospheric slabs in the mantle to be imaged (e.g. Van der Hilst et al., 1997; Van der Voo et al., 1999; Hall & Spakman, 2002; Miller et al., 2006) and present-day plate motions to be assessed using GPS studies (e.g. Michel et al., 2001; Beavan et al., 2002)], together with a mass of investigations (well over 200 two-month ocean drilling legs, innumerable geophysical and geological studies) have reaffirmed the basic power of the paradigm. In the author's opinion, when placed in an early 1970s perspective, today's global plate-tectonic models for the Mesozoic and Cenozoic (last 250 Myr) are largely finalized. Modifications will involve refinements, but not much more. The questions the biogeographers may (or may not) be asking following McCarthy's recent papers need to be examined in that light. Briggs’ (2004) comments regarding McCarthy's assertions about the pan-Pacific biotic record suggest that they may not carry much weight. Thus, the geoscientists can, for the time being at least, continue interpreting practically all of the geological phenomena on the surface of the Earth using plate-tectonic theory as the overarching mechanism. The adoption of an expanding Earth model is probably not going to be that helpful. Jonathan Aitchison, Dei Faustino and Iain Ross provided comments on a late-stage version of the manuscript. Jason R. Ali works as a geologist at the University of Hong Kong. His research includes plate modelling of the India–Asia collision system and E-SE Asia/W-SW Pacific, stratigraphic studies of the SW China Emeishan large igneous province and its possible link with the end-Guadalupian mass extinction c. 260 Ma, and the orientation of important religious buildings, cities, temples, etc. Despite the stand taken in this piece, he is/has been involved in controversial plate-modelling projects (Philippine Sea Plate in the mid-1990s, India–Asia collision today). Future work will address the apparently contentious Late Cretaceous–early Cenozoic India–Asia–Africa biological connection. Editor: Chris Humphries