TECTONIC EVENTS IN MIO-PLIOCENE TIME

As with environmental changes, a spectrum of observations suggests changes in the tectonic development of Tibet and its surroundings in late Miocene time. Some such events can be dated precisely at 7-8 Ma, and although dating of others remains less precise, many seem to have occurred near that time. If the growth of Tibet somehow affected regional climate, perhaps the most convincing test would be a demonstration that the mean elevation of Tibet changed concurrently with those climatic changes. So, let’s begin with a discussion of paleoaltimetry and then address other observations.

Paleoaltimetry of Tibet

The best test of the suggestion that most, if not all, of Tibet rose 1000-2000 m around 8 Ma would be a demonstration of lower elevations before that time and higher after it. Asserted histories of Tibetan elevation changes abound, but in my opinion, those that might address the question posed here are indefensible, if not wrong, and those that are right do not address it. The weakest inferences are based on either one or more fossil plant organs and pollen, fossil mammals, deposition rates and facies of sediment deposited near Tibet, and qualitative geomorphic observations. Let me discuss them before considering more reliable inferences.

Axelrod (1981) and Xu Ren (1978, 1981, 1984) assigned nearest living relatives to fossil plants and pollen, and they then used the environments shared by the nearest living relatives to assign paleo-elevations to southern Tibet. Xu Ren (1981), in particular, reported a rise of 3000 m since late Pliocene time. Mercier et al. (1987) subdivided these and similar data regionally to suggest that different parts of the plateau rose at different times, but they too found a rapid late Cenozoic rise for most of the plateau. Although many have criticized this approach (e.g., Chaloner and Kreber 1990; Wolfe 1971, 1979) let me use one example to illustrate how far astray this approach can go. Axelrod (1966, p. 29) associated roughly 90% of the fossil plant taxa from the Eocene Copper River Basin in northeastern Nevada with nearest living relatives either from “the Coast redwood forest [that] extends from sea level up to near 400-500 feet,” or with “the Spruce-Hemlock forest [that] has its lower margin near 4,500 feet.” The gap of 4000 feet, more than 1000 m, in the elevation ranges separating roughly half of these modern taxa from the remaining half requires that roughly half of the taxa subsequently adapted to very different present-day climates. Hence, the method not only ignores, but also is blind to evolutionary change.

Similar logic has been applied to fossil mammals, such as Hipparion (e.g., Li et al. 1981; Liu and Ding 1984). Proponents of a low Tibet in late Miocene and Pliocene time have associated such animals with warm environments that characterize regions outside of Tibet where fossil specimens have been found. Having excluded Tibet from those environments, these authors then deduced that for Hipparion and other animals to have lived on Tibet, Tibet must have been much warmer than it is today. This logic not only ignores late Cenozoic global cooling, but also denies the possibility that Hipparion, like modern horses, could have lived in a wide spectrum of environments, including Tibet.

Finally, many inferences that since 3-4 Ma parts, if not the whole, of Tibet rose from a low plateau only ~1000 m high to its present height derive from increases in sedimentation rates and in grain sizes of material deposited near Tibet (e.g., Li and Fang 1999; Li et al. 1997a, 1997b; Pares et al. 2003; Zheng et al. 2000) or from geomorphic inferences of recent down-cutting of rivers, like the Yellow River (e.g., Li et al. 1981, 1996). First, all such observations apply only to the edges of the plateau, not to the mean elevation of the large, internally drained, highest part of the plateau. Thus, even if these observations did relate to tectonic activity of Tibet, they would apply only to its outer edges. As did Will Downs, I think that most of these changes in erosion rates and grain sizes result from climate change (Molnar 2004; Molnar and England 1990; Zhang et al. 2001).

In the last few years, three studies have put bounds on middle to late Miocene paleo-elevations of three sites in southern Tibet, and all three imply that subsequent changes in height have been less than 1000-1400 m (Figure 2). Garzione et al. (2000b) measured values of ¹⁸O in modern stream water to derive a calibration of its values with height; they then measured such values in pedogenic carbonates, shells, and lacustrine micrites dating from 11 to 7 Ma in the Thakkhola graben of Nepal and used their local calibration to infer paleo-altitude (Garzione et al. 2000a). Rowley et al. (2001) used some data of Garzione et al. (2000b) and Wang et al. (1996), among other samples, to calibrate a theoretical relationship based on Rayleigh fractionation; they applied it to sedimentary rock in grabens in southernmost Tibet. They reported no change in elevation since ~10 Ma, though their results allow systemically higher estimates of paleo- than present-day elevations. Finally, Spicer et al. (2003) used leaf physiognomy and Wolfe’s (1993; Forest et al. 1999; Wolfe et al. 1998) correlations of it with environmental parameters to infer a 15-Ma paleo-elevation of the Namling basin in southernmost Tibet that is indistinguishable from that of today. These three papers rely on two different types of data (¹⁸O and fossil leaves) and three different methods. They imply that southernmost Tibet has undergone, at most, only small changes in elevation in late Cenozoic time.

As important as the studies by Garzione et al. (2000a, 2000b), Rowley et al. (2001), and Spicer et al. (2003) are in corroborating the methods used, they do not place tight bounds on the evolution of Tibet’s mean elevation. Most students of Tibetan geology recognize that an Andean margin, characterized by high elevations, and not a low terrain punctuated by a chain of volcanoes, typified southern Tibet before its collision with India (e.g., England and Searle 1986; Murphy et al. 1997). Late Cretaceous and early Cenozoic volcanic rock overlying folded, late Cretaceous sedimentary and volcanic rock in southern Tibet demonstrates significant pre-collisional crustal shortening (e.g., Chang and Cheng 1973). Thus, high paleo-elevations for southern Tibet offer no great surprise to those who imagine a rise of 1000-2000 m of the mean elevation of Tibet at ~8 Ma or a marked outward growth at that time.

Folding of the Indian Plate South of India

Because quantitative paleoaltimetry has not constrained the elevation history of more than a fraction of the plateau, we must use surrogates for changes in both the mean elevation and the lateral extent of the plateau. The best dated, if also the most remote, tectonic change occurred south of India in lithosphere beneath the Indian Ocean (Figure 2).

Excluding narrow plate boundaries, the earth’s most seismic submarine region lies near the equator south of India (e.g., Stein and Okal 1978). Sykes (1970) seems to have been the first to notice this activity; he suggested that it marked the initiation of a new plate boundary. Shortly afterward, Eittreim and Ewing (1972) described young folding and faulting of this region. Bands of gravity and geoid anomalies with a characteristic wavelength of ~200 km trend approximately east-west across this region, and positive anomalies mark zones of localized deformation (e.g., Chamot-Rooke et al. 1993; Krishna et al. 1998; Van Orman et al. 1995; Weissel et al. 1980). Drilling of one such fold (Figure 2) allowed an age of 7.5-8 Ma to be assigned to the onset of folding (e.g., Cochran 1990; Curray and Munasinghe 1989), and Krishna et al. (2001) later correlated and extrapolated sedimentary horizons to assign that age to folding over the entire region.

Using magnetic anomalies and fracture zones formed at the boundaries between the Somalia plate and the India or Capricorn plates, DeMets and Royer (2003) showed that convergence between the India and Capricorn plates was small until about 8 Ma. They could not rule out small relative motion at earlier times, and they reported small but resolvable movement between 20.1 and 17. 4 Ma, but all of the folding seems to have occurred since 7.9 Ma. Less clear is a change in relative motion between the Somalia and India plates.

Although one cannot uniquely attribute the onset and continued deformation in this region to a single change in boundary conditions on the Indian lithosphere, an obvious possibility is that some aspect of India’s convergence with Asia changed, and the force per unit length along their boundary also changed. As the “pressure gauge of Asia” (e.g., Molnar and Tapponnier 1978), Tibet applies a force per unit length to India, and if Tibet rose to a sufficient height, the force (per unit length) resisting India’s penetration into Eurasia could become large enough to deform the Indian lithosphere (e.g., Harrison et al. 1992; Molnar et al. 1993). A test of this idea can be made using simple estimates of the force per unit length needed to deform oceanic lithosphere (Martinod and Molnar 1995) and of the force per unit length necessary to maintain a high plateau at different mean elevations. Lateral support of a high plateau at a mean elevation of ~5000 m exceeds the force per unit length needed to deform oceanic lithosphere, but a plateau lower than ~4000 m can be supported by a force per unit length too small to deform oceanic lithosphere (Molnar et al. 1993). Thus, allowing for uncertainties in these estimates, an increase in elevation of ~1000-2000 m could have transformed the boundary condition on the northern edge of the Indian plate sufficiently to initiate deformation of it, and the ultimate separation of the Indo-Australian plate into the Indian and Australian plates.

My impression is that this folding is the most precisely dated large-scale tectonic event in the region that includes Tibet and its surroundings and that apparently occurred simultaneously with environmental changes in the region.

Deformation of Intracontinental Regions North and East of Tibet

Although India collided with Tibet at ca. 45-50 Ma (e.g., Garzanti and Van Haver 1988; Najman et al. 2001, 2002; Rowley 1996, 1998; Searle et al. 1987), a variety of observations suggest that currently high terrain north and east of the high plateau developed long after that time.

Tien Shan. Magnetostratigraphy of sediment accumulating in intermontane basins suggests that they formed at ~10-13 Ma in much of the western Tien Shan (Abdrakhmatov et al. 2001; Figure 2). The tightest constraints come from the Chu Basin on the northern edge of the Tien Shan in Kyrgyzstan and the adjacent Kyrgyz Range (Figure 2). Using magnetostratigraphy, Bullen et al. (2001) showed that sediment accumulation had begun before ~9 Ma, though at a low rate. Cooling ages using both fission-track and (U-Th)/He dating suggest an abrupt onset of erosion of the Kyrgyz Range at 11 ± 1 Ma, from which Bullen et al. (2001, 2003) inferred an emergence of this Range and the formation of the Chu Basin. Magnetostratigraphy from intermontane basins farther south suggests comparable ages, if slightly older (12-13 Ma) for the oldest sediment (Abdrakhmatov et al. 2001). Fission-track dates of ~20-25 Ma from other parts of the Tien Shan suggest that erosion was not negligible (e.g., Sobel and Dumitru 1997; Yin et al. 1998), and therefore that elevated terrain existed before late Miocene time. The filling of basins beginning at 10-13 Ma, however, suggests that mountain building became important at this later interval.

A late onset of mountain building gains some support from estimates of the total shortening across the Tien Shan and its current rate of shortening. Avouac et al. (1993) estimated ~200 km of north-south shortening across the western part of the belt between the Tarim Basin and the Kazakh Platform, although more recent work suggests that the amount may be less (e.g., Abdrakhmatov et al. 2001). The current rate across the same region as measured using GPS is ~15-20 mm/yr (Abdrakhmatov et al. 1996; Reigber et al. 2001), and in Kyrgyzstan, late Quaternary slip rates on the major faults match the GPS rates for that portion of a transect (Thompson et al. 2002). Thus, if the rate were constant, it would imply that all of the shortening occurred since ~10 Ma (Abdrakhmatov et al. 1996). In any case, the combination of present-day rates with the total shortening requires an acceleration of convergence since ~20 Ma, if not since 10 Ma.

Gobi-Altay of Mongolia. Evidence for timing of the rise of the Gobi-Altay (Figure 2) is far less convincing than that for the Tien Shan. Using the ratio of the height of the broad, flat summit plateau on Ih Bogd, the highest mountain in the Gobi-Altay, above the adjacent basin to the vertical component of current rate of slip on the main oblique strike-slip fault north of Ih Bogd (Ritz et al. 1995), Kurushin et al. (1997) inferred that the summit plateau emerged since 5 to 10 Ma.

Baikal Rift Zone. Deposition in basins and volcanic activity suggest that rifting in the Baikal area began in Oligocene time but then accelerated in late Cenozoic time (e.g., Kaz'min et al. 1995; Kuzmin et al. 2000; Logatchev 1974; Mats 1993; Mats et al. 2000; Rasskazov et al. 2003). Kaz'min et al. (1995) argued that most sediment was deposited since Late Miocene time, and that faulting began at that time. Indeed, a rift seems to have been present at 5 Ma, for drill cores have penetrated sediment of that age (e.g., Kuzmin et al. 2000). Relying in part on the offset of a dike by a thrust fault dated at 12-14 Ma in the Baikal region (Ruzhich et al. 1972), Delvaux et al. (1997) inferred that thrust faulting and northeast-southwest shortening occurred until late Miocene time, when rifting began, if at a slow rate. Moreover, dating of volcanic rock in northern Mongolia suggests that grabens southwest of Lake Baikal itself began to form at 10-8 Ma (Rasskazov et al. 2003). Thus, although a precise date for when rifting began cannot be given, much of the crustal extension associated with that process seems to have occurred since 10 Ma or so.

Qilian Shan. Similarly, although deformation seems to have occurred on the northeast margin of Tibet shortly after the collision between India and Eurasia, Métivier et al. (1998) inferred that the mountain ranges comprising the Qilian Shan and adjacent high terrain rose at ~5.3 Ma (Figure 2). They relied on an apparently abrupt increase in sedimentation rate at that time in the Qaidam Basin, southwest of the high terrain. As others have emphasized from the distribution of Oligocene and Miocene sedimentary rock and differing grain sizes (e.g., Dupont-Nivet et al. 2004; Fang et al. 2003; Horton et al. 2004; Ritts et al. 2004; Wang et al. 2003b; Yin et al. 2002), eroding terrain must have lain near the various sedimentary basins in northeastern Tibet when that deposition occurred. Thus, the arguments of Métivier et al. (1998) do not imply that prior to ~5.3 Ma relief was absent, but rather that mountains of the present size seem unlikely. A weakness in the suggestion of a rapid rise at 5.3 Ma is that that age corresponds to the boundary between the Miocene and Pliocene Epochs; it requires that erosion rates, attributed to the emergence of adjacent terrain, accelerated precisely when marine microorganisms evolved, which seems causally impossible and therefore improbable. I do not doubt that sedimentation increased within a few million years of 5.3 Ma, but assigning a different date for the onset of rapid sedimentation also requires revising the sedimentation rates. Regardless of my doubts about the precise date when the Qilian Shan emerged, the observation that sedimentation increased rapidly at ~5 Ma in a basin dammed by high terrain to the north and northeast is consistent with a late Miocene emergence of that terrain then.

Linxia Basin and Adjacent Edge of the Tibetan Plateau. Using magnetostratigraphy, Fang et al. (2003) showed that sediment began accumulating in the Linxia Basin (Figure 2) at ~29 Ma and slowly accelerated, as if within a foreland basin adjacent to a northeastward propagating fold-and-thrust belt. The accumulation rate then decreased at ~6 Ma. Fang et al. (2003) suggested that flexure ceased at that time, presumably because thrust slip on the fault creating the load flexing the basin slowed or stopped. Accordingly, they inferred that the locus of thrust faulting stepped northeastward.

In support of this timing, Zheng et al. (2003) measured detrital fission-track ages from the Linxia Basin, and using Brandon’s (1992) method for isolating populations of ages, they showed an abrupt change near 8 Ma, though possibly as early as 14 Ma, in the difference between the mean age of detritus and the depositional age (Fang et al. 2003). Zheng et al. (2003) interpret this change to document the emergence of the range just west of the basin, Laji Shan, where Proterozoic rock crops out widely.

Emergence of the Liupan Shan. The Liupan Shan, which lies ~200 km northeast of the Linxia Basin and is marked by a single ridge of high terrain, forms the northeasternmost zone of crustal shortening and thickening associated with the formation and outward growth of Tibet (Zhang et al. 1991). Using reset fission-track ages on detrital material deposited since Cretaceous time, Zheng et al. (2005) found that the older material had been heated sufficiently to anneal tracks and to reset ages, and then subsequently cooled since ~8 Ma. They inferred that thrust faulting, growth of the mountain range, erosion of it, and the resulting cooling of buried sediment began then.

Min Shan and Longmen Shan. Using both fission-track and (U-Th)/He dates from different elevations in these regions, Kirby et al. (2002) found slow cooling, at <1ºC/Myr, from Jurassic to mid-Miocene time, and then rapid cooling at 30-50ºC/Myr (Figure 2). For the Longmen Shan, rapid cooling began no earlier than 12-13 Ma and perhaps as recently as 5-6 Ma. For the Min Shan, it began no earlier than 6-7 Ma and perhaps as recently as 4-5 Ma. Kirby et al. (2002) suggested that mountain building began when cooling accelerated.

Incision of Southeast Tibet. Also using fission-track and (U-Th)/He dates from elevation transects into deep gorges cut into southeastern Tibet, Clark et al. (2005; Clark 2003) inferred slow cooling at <1ºC/Myr in Cretaceous and early Cenozoic time, followed by rapid cooling beginning between 9 and 13 Ma (Figure 2). They reported an average erosion rate during the period of slow cooling of <0.02 mm/yr (< 20 m/Myr) on the interfluves between deep gorges, and they concluded that the present-day gentle surface of low relief almost surely was flat and low when cooling was slow. Thus, since 9-13 Ma, the surface defined by this low-relief terrain rose ~2000 m. They buttressed this inference with a discussion of the present-day drainage and the possibility that rapidly rising terrain led to river capture (Clark et al. 2004).

Summary. The preceding discussion describes regions where tectonic activity seems to have begun long after India collided with Eurasia. In the context of tectonic processes that reflect the growth of high terrain concurrent with regional climate change, the mention of a few caveats might help readers.

First, notice that the formation of basins and ranges within the Tien Shan began before 8 Ma. Thus, if that tectonic activity relates to growth of Tibet, it preceded most (but not all) of the environmental changes discussed above, if by no more than a few million years. Of course, a rise of Tibet by as much as 1 km surely requires a finite amount of time measured in millions of years. Conversely, if these tectonic developments in the Tien Shan reflect a change in the structure of Tibet, they imply that such changes began before the atmosphere responded to them, presumably because exceeding a threshold required a finite change in Tibet’s mean elevation.

Second, with the uncertainties in ages, perhaps none of the inferred emergence of mountain ranges and higher elevations occurred simultaneously. Nor do any of these observations require a concurrent increase in elevation of the high part of Tibet. Rather, current uncertainties merely permit such a possibility.

Finally, all of these inferences depend on changes in erosion, incision, or sedimentation; none documents changes in mean elevations and none quantifies amounts of vertical movement. Tectonic activity is not the only trigger for increased erosion, and changes in erosion, incision, or sedimentation do not necessarily imply concurrent tectonically created relief (e.g., Molnar and England 1990; Zhang et al. 2001). From a converse view, however, if one hypothesizes that the surface of Tibet rose 1000-2000 m beginning a few million years before 8 Ma, and that the associated gain in potential energy per unit area induced deformation in surrounding regions, then these hypotheses pass tests set by the evidence presented above for late Miocene increases in erosion, incision, or sedimentation.

Normal Faulting and East-West Extension of Tibet

When Harrison et al. (1992) suggested that Tibet had reached its maximum elevation at ~8 Ma, and regional climate changed at the same time, one fact that they used, and that Molnar et al. (1993) also exploited, was the date of ~8 Ma for the onset of slip on one major normal fault in Tibet. Dating along two transects of material in the fault zone on the southeast side of the Nyainqentanghla in Tibet (Figure 2), supported by calculations of conductive heat transport, implies an initiation of faulting at 8 ± 1 Ma (Harrison et al. 1995; Pan and Kidd 1992). Moreover, from the cooling ages on the two sides of the fault, Harrison et al. (1995) inferred 15-20 km of slip on the fault, which shows that this fault is not minor. Harrison et al. (1995) also reported, without giving details, that two other normal faults in southern Tibet formed at 8 ± 1 Ma and 9 ± 1 Ma (Figure 2). Stockli et al. (2002) also obtained dates elsewhere in southern Tibet that corroborate this timing, but subsequent to Harrison’s work, others have measured dates, from which they inferred an earlier onset of normal faulting.

Coleman and Hodges (1995) dated hydrothermally deposited mica in north-south extension fractures in the Greater Himalaya near the Thakkhola graben (Figure 2). They used their date of ~14 Ma to demonstrate normal faulting and east-west extension at that time. Later Garzione et al. (2000a, 2003) showed that sediment had accumulated in the Thakkhola graben by 11 Ma, and like Coleman and Hodges (1995), they questioned an 8-Ma onset of east-west extension via normal faulting. I do not think that normal faulting in this setting necessarily bears on the question of when east-west extension of Tibet began; in some regions of oblique subduction, analogous normal faulting occurs within the overriding plate (e.g., Ekström and Engdahl 1989; McCaffrey 1992). If the faulting associated with the Thakkhola graben bears a similar relationship to oblique convergence, it need not reflect the deformation field farther north within Tibet.

Williams et al. (2001) dated north-south trending dikes in southern Tibet (Figure 2) and obtained ages of 13.8 ± 0.8 to 18.3 ± 2.7 Ma. They argued that these dates imply that east-west extension of Tibet had begun, and thus Tibet had reached its maximum elevation by this time. As Nakamura (1977) showed for island-arc volcanoes, dikes tend to form orthogonal to arcs, presumably because the maximum compressive stress is aligned in that direction, and hence the least compressive horizontal stress is oriented parallel to the arc. Dike injection does not imply that significant strain occurs parallel to the arcs, but simply that horizontal compressive stresses resist intrusion of dikes oriented perpendicular to the arcs less than dikes oriented parallel to the arcs. I consider the existence of potassium-rich basaltic dikes in southern Tibet (see below) more of a surprise than their orientations, and that the orientations of dikes do not demonstrate the attainment of a maximum elevation or the onset of significant east-west extension.

Among normal faults dated in Tibet, that for the Shuang Hu graben in central Tibet (Figure 2) poses the biggest problem for those of us who cling to the view that Tibet reached its maximum elevation after, not before, ~10 Ma. Using Rb-Sr and ⁴⁰Ar/³⁹Ar, Blisniuk et al. (2001) determined ages of mineralization in a fault zone as 13.5 Ma, which they consider a minimum age for the onset of faulting. Noting that Yin et al. (1999) had inferred a vertical offset of ~7 km for the faulting on the west side of the graben, Blisniuk et al. (2001) inferred that the faulting that they dated was significant. They argued further that this faulting called into question proposed relationships between normal faulting and tectonic processes on the margin of the plateau, relationships that motivate this paper. Yet, perhaps one should recall that normal faulting often occurs at discontinuities in strike-slip faults, where one strand steps to another parallel strand. A good example is along the Xianshuihe fault in eastern Tibet, where fault-plane solutions of earthquakes show normal faulting in the pull-apart zone between parallel strike-slip strands (e.g., Molnar and Lyon-Caen 1989; Zhou et al. 1983). Many of the grabens in central Tibet lie along strike-slip faults (e.g., Taylor et al. 2003), and thus the relationship of the Shuang Hu graben to strike-slip faulting may serve as a reminder that normal faulting can occur in local pull-apart basins along strike-slip faults and not be associated with regional extension. (Near Death Valley in the Basin and Range Province, thrust faulting in the Avawatz Mountains at the east end of the Garlock fault provides a spectacular exception to the rule that normal faulting dominates late Cenozoic deformation of that Province.)

The preceding paragraphs on normal faulting may read like a political argument trying to refute claims by opponents. When Pan and Kidd (1992) first presented evidence that the Nyainqentanghla faulting dated from approximately 8 Ma, and then Harrison et al. (1995) trimmed its uncertainty to only 1 Myr, the result seemed too good to be true. Perhaps those of us who used that single date to assign a major role to Tibet and its underlying mantle (e.g., Harrison et al. 1992; Molnar et al. 1993) strained its applicability. At the same time, it seems to me that the few reliable dates of faulting from the interior of the plateau leave open the timing of Tibet’s attaining its maximum mean elevation.

Cenozoic Potassium-Rich Basaltic Volcanism

Turner et al. (1993, 1996) interpreted late Cenozoic, potassium-rich volcanism in northern Tibet to reflect melting of mantle lithosphere. They assumed that potassium and other large-ion lithophile elements had seeped into the lithosphere from an underlying partially molten asthenosphere, and their presence in erupted lavas implied melting of the lithosphere. Following England and Houseman (1989), they suggested that such lithosphere had thickened in response to India’s penetration into Eurasia, and then removal of the lower part of that thickened lithosphere exposed its upper parts to temperatures high enough to remelt it, as potassium-rich volcanism.

Although Arnaud et al. (1992) associated such volcanism with intracontinental subduction, Molnar et al. (1993) used both the logic given by Turner et al. (1993, 1996) for convective removal of mantle lithosphere and the relatively young ages of such volcanism, mostly <10 Ma, as further support for late Cenozoic removal of mantle lithosphere. Subsequently, others have reported older potassium-rich volcanism elsewhere in Tibet (e.g., Chung et al. 1998), and in particular in southern and southwestern Tibet as early as ~25 Ma (Chung et al. 2003; Miller et al. 1999; Nomade et al. 2004; Williams et al. 2001) to 42 Ma in southeastern Tibet (Wang et al. 2001). Thus, using young basaltic volcanism in northern Tibet as an argument for a late Cenozoic rise of Tibet seems to have been a mistake, and I omit it from Figure 2.

Do Tectonic (and Other) Observations Imply a Rapid Change in Tibet's Growth at ~8Ma?

When Harrison et al. (1992) proposed that Tibet reached its maximum elevation near 8 Ma, they relied on the dating of one normal fault (Nyainqentanghla), one fold south of India, and some of the environmental changes discussed above. Molnar et al. (1993) both developed these arguments further and added late Cenozoic potassium-rich volcanism to them. Since that time, much new data have been added, with the 8-Ma onset of normal faulting and crustal thinning within the plateau becoming questionable, and volcanism becoming irrelevant to changes in Tibet’s elevation. The initiation of folding south of India provides the only major tectonic event that can be both associated with Tibet’s growth and shown to have occurred simultaneously with the environmental changes discussed above. This reduction in supporting evidence might seem to deny Tibet’s growth a relationship to regional environmental change.

As summarized above, the dating of many features in the areas surrounding Tibet over the past 10 years can be interpreted to suggest that the high terrain north and northeast of Tibet and on the northeastern and eastern margins of Tibet developed in late Miocene time. Several aspects of these dated phenomena, however, make tying them all to a late Miocene pulse in the growth of Tibet a bit like a house-of-cards. First, in most cases, dating is not sufficiently precise to require them to be simultaneous, and for one region, the Tien Shan, the onset of rapid erosion of mountainous terrain and deposition of adjacent sediment preceded 8 Ma by a few million years. Moreover, virtually all of the phenomena dated—erosion, incision, cooling of deeply buried rock, and accumulation of sediment—do not by themselves require a change in tectonic processes. The one clear exception is the ratio of total shortening across the western Tien Shan of at least 100 km (Abdrakhmatov et al. 2001) and perhaps 200 km (Avouac et al. 1993) to the current rate of shortening of 15-20 mm/yr (Abdrakhmatov et al. 1996; Reigber et al. 2001), which requires a marked acceleration of shortening in Late Cenozoic time. Skeptics, therefore, cannot be faulted if they reject the assertion that Tibet rose near 8 Ma, and its rise effected regional climate change.

It seems to me that alternative views of the data presented here have comparable validity with that offered above to a skeptic. First, if Tibet did rise 1000-2000 m just before 8 Ma, it would have imposed a higher deviatoric compressive stress on its surrounding regions. Accordingly those surroundings might undergo accelerated horizontal shortening and mountain building. Such shortening and associated crustal thickening would in turn create additional potential energy of erosive agents. Thus, the evidence suggesting an abrupt change in erosion, or incision, of Tibet’s surroundings allows an abrupt rise of Tibet to pass one test made of it.

From another perspective, suppose that the arguments used above do call for crustal shortening, and not accelerated erosion due to climate change. Then, recall that India collided with southern Asia at ~45-50 Ma and has penetrated into Eurasia at an essentially constant rate since that time. The apparently rapid onset of such deformation, in the period since 15 and before 5 Ma and across a wide region surrounding Tibet implies a change in processes within Tibet, even if the evidence is insufficient to define the specific processes that changed.

Thus, the answer to the question – do tectonic (and other) observations imply a rapid change in Tibet’s growth at ~8 Ma? – lies somewhere between a cautious no and a risky yes. Like Will Downs, I like taking risks in science. Accordingly, the replacement of the word “imply” with “suggest” would elicit a stronger yes, but the word “require” would demand a no.