The rise and growth of Tibet Andreas Mulch and C. Page Chamberlain It is not difficult to be impressed by the grandeur of high mountainous regions, but it is difficult to reconstruct how the elevation of such regions evolved. A study of the Tibetan plateau does just that. Is Everest now at its highest point, or was it once even loftier? What was the greatest height attained by the vast highlands of the Tibetan plateau, and when did this occur? As described elsewhere in this issue by Rowley and Currie (page 677)1, these questions can be tackled — if not yet answered definitively — by analysing the isotopic composition of ancient raindrops. With this approach, the authors show that Tibet continuously grew northward over millions of years in response to the thickening of Earth’s crust associated with the collision of the Indian and Asian continental plates. The driving forces for this collision are generated deep in Earth’s mantle. But the key to unravelling the uplifting history of the central Tibetan plateau is found in lake sediments on the plateau, some of which formed as long ago as 40 million years. In these lakes and their surrounds, changes in the oxygenisotope composition of surface water (which is controlled by regional climate and elevation) are recorded in sediments. Systematic variations in oxygen-isotope composition across the plateau reveal that spatially variable uplift of the plateau to 4,000 metres or more above sea level was intimately linked to the timing and rates of convergence of India and Asia (Fig. 1). Rowley and Currie1 estimate that uplift to 4,000 metres was initiated as long ago as 40 million to 50 million years, in the early stages of that convergence. The evolution of mountain topography reflects the balance between tectonic forces in Earth’s crust and upper mantle, and climatically driven erosion at Earth’s surface. Their relative role in controlling the rise of mountains remains unclear2,3, but the problem can be approached by reconstructing the elevation history of large continental plateaux. Such studies can improve our understanding of the coupling between tectonics and long-term climate change. For example, the Tibetan plateau — the largest continental highland on Earth — is a major barrier to air flow in the atmosphere, and it has been suggested that uplift of the plateau triggered the onset of the Indian summer monsoon3. The chemical fingerprint of rain and snow that precipitated on the Tibetan plateau is found in the oxygen-isotope composition of calcareous minerals in Tibetan lake sediments. This is expressed as _18O, which is the 18O/16O ratio in the soils or sediments relative to the 18O/16O ratio in sea water. For example, the _18O of calcite formed in soils is related to the_18O of soilwater or groundwater by a temperature- dependent fractionation factor; so _18O in calcite formed in soil is a sensitive tracer of surface water that stems from precipitation. The underlying principle of oxygen-isotope altimetry is that water that precipitates as rain or snow becomes increasingly depleted in 18O the higher up a mountain range that it falls4; systematic changes in _18O with elevation can then be used to infer relative elevation differences between the water source in the ocean and the elevation at which the rain or snow fell5,6. In the context of earlier isotopic studies6,7, the low _18O values of carbonates found by Rowley and Currie1 across the central Tibetan plateau reflect the south–north migration of high terrain in response to crustal thickening and the buoyant rise of Earth’s surface. This finding agrees with the results of thermomechanical modelling for the growth and uplift history of the Tibetan plateau8. It was proposed previously that the present stature of Tibet reflects tectonic processes involving thinning and delamination of Earth’s crust and mantle during the Miocene (10 million to 8 million years ago)3. But it has since been suggested9,10 that the sequential rise and northward growth of Tibet started much earlier than that, in Eocene times, about 50 million years ago. The new results1 support this idea that the Tibetan plateau is a long-standing topographic feature that arose from the collision between India and Asia, and is not the more recent product of other, deeper-seated processes. The results also show that stable isotopic data from sedimentary sequences do indeed record the long-term, high-elevation history of the plateau, and can provide absolute elevation constraints at various times in the geological history of such a topographic feature. Moreover, taken in conjunction with independent estimates of palaeoelevation11, this approach overcomes some of the limitations in deciphering the competing effects of climate and elevation change in the past. Only slowly are we making progress in understanding how the interactions of surface uplift, bedrock erosion and sediment transport create dynamic feedbacks between the biosphere, the atmosphere and the crust and upper mantle. However, new techniques for estimating palaeoaltimetry that relate processes at various levels of Earth’s crust, from the surface to deeper regions where high-temperature deformation causes rock flow, will allow us to develop a more dynamic view of how mountain ranges change their shape and stature. Such techniques exploit elevation- dependent changes in concentration of carbon dioxide in the atmosphere12, or the isotopic composition of surface waters that circulate deep in Earth’s interior during the late tectonic evolution of mountain ranges13. Deciphering the oxygen-isotope record in lake and soil deposits requires careful consideration of the competing effects of climate change and changes in surface elevation. The application of multi-proxy isotopic systems, which take account of both surface and deeper Earth environments, can complement such studies and greatly enhance our predictive capabilities in such a task. ■ Andreas Mulch and C. Page Chamberlain are in the Department of Geological and Environmental Sciences, Stanford University, Stanford, California 94305, USA. e-mail: mulch@pangea.stanford.edu