Oxygen-isotope evidence from ancient zircons for liquid water at the Earth's surface 4,300 Myr ago

Granitoid gneisses and supracrustal rocks that are 3,800-4,000 Myr old are the oldest recognized exposures of continental crust. To obtain insight into conditions at the Earth's surface more than 4 Gyr ago requires the analysis of yet older rocks or their mineral remnants. Such an opportunity is presented by detrital zircons more than 4 Gyr old found within 3-Gyr-old quartzitic rocks in the Murchison District of Western Australia. Here we report in situ U-Pb and oxygen isotope results for such zircons that place constraints on the age and composition of their sources and may therefore provide information about the nature of the Earth's early surface. We find that 3,910-4,280 Myr old zircons have oxygen isotope (delta18O) values ranging from 5.4+/-0.6% to 15.0+/-0.4%. On the basis of these results, we postulate that the approximately 4,300-Myr-old zircons formed from magmas containing a significant component of re-worked continental crust that formed in the presence of water near the Earth's surface. These data are therefore consistent with the presence of a hydrosphere interacting with the crust by 4,300 Myr ago.     

Absolute plate motions and true polar wander in the absence of hotspot tracks

The motion of continents relative to the Earth's spin axis may be due either to rotation of the entire Earth relative to its spin axis--true polar wander--or to the motion of individual plates. In order to distinguish between these over the past 320 Myr (since the formation of the Pangaea supercontinent), we present here computations of the global average of continental motion and rotation through time in a palaeomagnetic reference frame. Two components are identified: a steady northward motion and, during certain time intervals, clockwise and anticlockwise rotations, interpreted as evidence for true polar wander. We find approximately 18 degrees anticlockwise rotation about 250-220 Myr ago and the same amount of clockwise rotation about 195-145 Myr ago. In both cases the rotation axis is located at about 10-20 degrees W, 0 degrees N, near the site that became the North American-South American-African triple junction at the break-up of Pangaea. This was followed by approximately 10 degrees clockwise rotation about 145-135 Myr ago, followed again by the same amount of anticlockwise rotation about 110-100 Myr ago, with a rotation axis in both cases approximately 25-50 degrees E in the reconstructed area of North Africa and Arabia. These rotation axes mark the maxima of the degree-two non-hydrostatic geoid during those time intervals, and the fact that the overall net rotation since 320 Myr ago is nearly zero is an indication of long-term stability of the degree-two geoid and related mantle structure. We propose a new reference frame, based on palaeomagnetism, but corrected for the true polar wander identified in this study, appropriate for relating surface to deep mantle processes from 320 Myr ago until hotspot tracks can be used (about 130 Myr ago).     

Evidence from episodic seamount volcanism for pulsing of the Iceland plume in the past 70 Myr

The North Atlantic volcanic province has been attributed to continental rifting about 60 Myr ago over an Iceland plume head with a diameter of 1,000-2,000 km (refs 1, 2). But evidence from a few igneous centres has been used to infer that earlier plume activity occurred in this region. The three seamounts in the Rockall trough off the Atlantic coast of Scotland are among the few accessible remnants of such early plume activity. Here we present 40Ar-39Ar incremental-heating ages of samples from these seamounts, which show that volcanism began there in the late Cretaceous period (70 +/- 1 Myr ago), and then continued for the next 30 Myr in at least four discrete phases: 62, 52, 47 and 42 Myr ago. We relate this activity to pulsing of large masses (approximately 10(8) km3) of hot Iceland plume material on timescales of 5-10 Myr. This significantly extends the time span for Iceland plume activity both backwards and forwards in time, and provides a possible alternative to the 'plume head' models for the formation of continental flood basalts.     

Coupling of surface temperatures and atmospheric CO2 concentrations during the Palaeozoic era

Atmospheric carbon dioxide concentrations seem to have been several times modern levels during much of the Palaeozoic era (543-248 million years ago), but decreased during the Carboniferous period to concentrations similar to that of today. Given that carbon dioxide is a greenhouse gas, it has been proposed that surface temperatures were significantly higher during the earlier portions of the Palaeozoic era. A reconstruction of tropical sea surface temperatures based on the delta18O of carbonate fossils indicates, however, that the magnitude of temperature variability throughout this period was small, suggesting that global climate may be independent of variations in atmospheric carbon dioxide concentration. Here we present estimates of sea surface temperatures that were obtained from fossil brachiopod and mollusc shells using the 'carbonate clumped isotope' method-an approach that, unlike the delta18O method, does not require independent estimates of the isotopic composition of the Palaeozoic ocean. Our results indicate that tropical sea surface temperatures were significantly higher than today during the Early Silurian period (443-423 Myr ago), when carbon dioxide concentrations are thought to have been relatively high, and were broadly similar to today during the Late Carboniferous period (314-300 Myr ago), when carbon dioxide concentrations are thought to have been similar to the present-day value. Our results are consistent with the proposal that increased atmospheric carbon dioxide concentrations drive or amplify increased global temperatures.     

Evolution of Asian monsoons and phased uplift of the Himalaya-Tibetan plateau since Late Miocene times

The climates of Asia are affected significantly by the extent and height of the Himalayan mountains and the Tibetan plateau. Uplift of this region began about 50 Myr ago, and further significant increases in altitude of the Tibetan plateau are thought to have occurred about 10-8 Myr ago, or more recently. However, the climatic consequences of this uplift remain unclear. Here we use records of aeolian sediments from China and marine sediments from the Indian and North Pacific oceans to identify three stages of evolution of Asian climates: first, enhanced aridity in the Asian interior and onset of the Indian and east Asian monsoons, about 9-8 Myr ago; next, continued intensification of the east Asian summer and winter monsoons, together with increased dust transport to the North Pacific Ocean, about 3.6-2.6 Myr ago; and last, increased variability and possible weakening of the Indian and east Asian summer monsoons and continued strengthening of the east Asian winter monsoon since about 2.6 Myr ago. The results of a numerical climate-model experiment, using idealized stepwise increases of mountain-plateau elevation, support the argument that the stages in evolution of Asian monsoons are linked to phases of Himalaya-Tibetan plateau uplift and to Northern Hemisphere glaciation.     

Evidence from detrital zircons for the existence of continental crust and oceans on the Earth 4.4 Gyr ago

No crustal rocks are known to have survived since the time of the intense meteor bombardment that affected Earth between its formation about 4,550 Myr ago and 4,030 Myr, the age of the oldest known components in the Acasta Gneiss of northwestern Canada. But evidence of an even older crust is provided by detrital zircons in metamorphosed sediments at Mt Narryer and Jack Hills in the Narryer Gneiss Terrane, Yilgarn Craton, Western Australia, where grains as old as approximately 4,276 Myr have been found. Here we report, based on a detailed micro-analytical study of Jack Hills zircons, the discovery of a detrital zircon with an age as old as 4,404+/-8 Myr--about 130 million years older than any previously identified on Earth. We found that the zircon is zoned with respect to rare earth elements and oxygen isotope ratios (delta18O values from 7.4 to 5.0%), indicating that it formed from an evolving magmatic source. The evolved chemistry, high delta18O value and micro-inclusions of SiO2 are consistent with growth from a granitic melt with a delta18O value from 8.5 to 9.5%. Magmatic oxygen isotope ratios in this range point toward the involvement of supracrustal material that has undergone low-temperature interaction with a liquid hydrosphere. This zircon thus represents the earliest evidence for continental crust and oceans on the Earth.     

Closing of the Indonesian seaway as a precursor to east African aridification around 3-4 million years ago

Global climate change around 3-4 Myr ago is thought to have influenced the evolution of hominids, via the aridification of Africa, and may have been the precursor to Pleistocene glaciation about 2.75 Myr ago. Most explanations of these climatic events involve changes in circulation of the North Atlantic Ocean due to the closing of the Isthmus of Panama. Here we suggest, instead, that closure of the Indonesian seaway 3-4 Myr ago could be responsible for these climate changes, in particular the aridification of Africa. We use simple theory and results from an ocean circulation model to show that the northward displacement of New Guinea, about 5 Myr ago, may have switched the source of flow through Indonesia-from warm South Pacific to relatively cold North Pacific waters. This would have decreased sea surface temperatures in the Indian Ocean, leading to reduced rainfall over eastern Africa. We further suggest that the changes in the equatorial Pacific may have reduced atmospheric heat transport from the tropics to higher latitudes, stimulating global cooling and the eventual growth of ice sheets.     

Evolution of the atmosphere and oceans

The residence times of most constituents of the atmosphere and oceans are small fractions of the age of the Earth and, in general, their rate of output has been nearly equal to their rate of input. We are disturbing a number of these dynamic equilibria quite severely. The mineralogy of marine evaporites rules out drastic changes in the composition of sea water during the last 900 Myr. The chemistry of soils formed more than 1,000 Myr ago suggests that the atmosphere then contained significantly more CO2 and less O2 than at present. Hydrogen peroxide may well have been the principal oxidant and formaldehyde the main reductant in rain water between 3,000 and 1,000 Myr ago. Major changes in atmospheric chemistry since that time are almost certainly related to the evolution of the biosphere.     

The Cenozoic palaeoenvironment of the Arctic Ocean

The history of the Arctic Ocean during the Cenozoic era (0-65 million years ago) is largely unknown from direct evidence. Here we present a Cenozoic palaeoceanographic record constructed from >400 m of sediment core from a recent drilling expedition to the Lomonosov ridge in the Arctic Ocean. Our record shows a palaeoenvironmental transition from a warm 'greenhouse' world, during the late Palaeocene and early Eocene epochs, to a colder 'icehouse' world influenced by sea ice and icebergs from the middle Eocene epoch to the present. For the most recent approximately 14 Myr, we find sedimentation rates of 1-2 cm per thousand years, in stark contrast to the substantially lower rates proposed in earlier studies; this record of the Neogene reveals cooling of the Arctic that was synchronous with the expansion of Greenland ice (approximately 3.2 Myr ago) and East Antarctic ice (approximately 14 Myr ago). We find evidence for the first occurrence of ice-rafted debris in the middle Eocene epoch (approximately 45 Myr ago), some 35 Myr earlier than previously thought; fresh surface waters were present at approximately 49 Myr ago, before the onset of ice-rafted debris. Also, the temperatures of surface waters during the Palaeocene/Eocene thermal maximum (approximately 55 Myr ago) appear to have been substantially warmer than previously estimated. The revised timing of the earliest Arctic cooling events coincides with those from Antarctica, supporting arguments for bipolar symmetry in climate change.     

Late Miocene decoupling of oceanic warmth and atmospheric carbon dioxide forcing

Deep-time palaeoclimate studies are vitally important for developing a complete understanding of climate responses to changes in the atmospheric carbon dioxide concentration (that is, the atmospheric partial pressure of CO(2), p(co(2))). Although past studies have explored these responses during portions of the Cenozoic era (the most recent 65.5 million years (Myr) of Earth history), comparatively little is known about the climate of the late Miocene (～12-5 Myr ago), an interval with p(co(2)) values of only 200-350?parts per million by volume but nearly ice-free conditions in the Northern Hemisphere and warmer-than-modern temperatures on the continents. Here we present quantitative geochemical sea surface temperature estimates from the Miocene mid-latitude North Pacific Ocean, and show that oceanic warmth persisted throughout the interval of low p(co(2)) ～12-5 Myr ago. We also present new stable isotope measurements from the western equatorial Pacific that, in conjunction with previously published data, reveal a long-term trend of thermocline shoaling in the equatorial Pacific since ～13?Myr ago. We propose that a relatively deep global thermocline, reductions in low-latitude gradients in sea surface temperature, and cloud and water vapour feedbacks may help to explain the warmth of the late Miocene. Additional shoaling of the thermocline after 5?Myr ago probably explains the stronger coupling between p(co(2)), sea surface temperatures and climate that is characteristic of the more recent Pliocene and Pleistocene epochs.     

A new species of great ape from the late Miocene epoch in Ethiopia

With the discovery of Ardipithecus, Orrorin and Sahelanthropus, our knowledge of hominid evolution before the emergence of Pliocene species of Australopithecus has significantly increased, extending the hominid fossil record back to at least 6 million years (Myr) ago. However, because of the dearth of fossil hominoid remains in sub-Saharan Africa spanning the period 12-7 Myr ago, nothing is known of the actual timing and mode of divergence of the African ape and hominid lineages. Most genomic-based studies suggest a late divergence date-5-6 Myr ago and 6-8 Myr ago for the human-chimp and human-gorilla splits, respectively-and some palaeontological and molecular analyses hypothesize a Eurasian origin of the African ape and hominid clade. We report here the discovery and recognition of a new species of great ape, Chororapithecus abyssinicus, from the 10-10.5-Myr-old deposits of the Chorora Formation at the southern margin of the Afar rift. To the best of our knowledge, these are the first fossils of a large-bodied Miocene ape from the African continent north of Kenya. They exhibit a gorilla-sized dentition that combines distinct shearing crests with thick enamel on its 'functional' side cusps. Visualization of the enamel-dentine junction by micro-computed tomography reveals shearing crest features that partly resemble the modern gorilla condition. These features represent genetically based structural modifications probably associated with an initial adaptation to a comparatively fibrous diet. The relatively flat cuspal enamel-dentine junction and thick enamel, however, suggest a concurrent adaptation to hard and/or abrasive food items. The combined evidence suggests that Chororapithecus may be a basal member of the gorilla clade, and that the latter exhibited some amount of adaptive and phyletic diversity at around 10-11 Myr ago.     

A complete insect from the Late Devonian period

After terrestrialization, the diversification of arthropods and vertebrates is thought to have occurred in two distinct phases, the first between the Silurian and the Frasnian stages (Late Devonian period) (425-385?million years (Myr) ago), and the second characterized by the emergence of numerous new major taxa, during the Late Carboniferous period (after 345?Myr ago). These two diversification periods bracket the depauperate vertebrate Romer's gap (360-345?Myr ago) and arthropod gap (385-325?Myr ago), which could be due to preservational artefact. Although a recent molecular dating has given an age of 390?Myr for the Holometabola, the record of hexapods during the Early-Middle Devonian (411.5-391?Myr ago, Pragian to Givetian stages) is exceptionally sparse and based on fragmentary remains, which hinders the timing of this diversification. Indeed, although Devonian Archaeognatha are problematic, the Pragian of Scotland has given some Collembola and the incomplete insect Rhyniognatha, with its diagnostic dicondylic, metapterygotan mandibles. The oldest, definitively winged insects are from the Serpukhovian stage (latest Early Carboniferous period). Here we report the first complete Late Devonian insect, which was probably a terrestrial species. Its 'orthopteroid' mandibles are of an omnivorous type, clearly not modified for a solely carnivorous diet. This discovery narrows the 45-Myr gap in the fossil record of Hexapoda, and demonstrates further a first Devonian phase of diversification for the Hexapoda, as in vertebrates, and suggests that the Pterygota diversified before and during Romer's gap.     

Seismic evidence for catastrophic slab loss beneath Kamchatka

In the northwest Pacific Ocean, a sharp corner in the boundary between the Pacific plate and the North American plate joins a subduction zone running along the southern half of the Kamchatka peninsula with a region of transcurrent motion along the western Aleutian arc. Here we present images of the seismic structure beneath the Aleutian-Kamchatka junction and the surrounding region, indicating that: the subducting Pacific lithosphere terminates at the Aleutian-Kamchatka junction; no relict slab underlies the extinct northern Kamchatka volcanic arc; and the upper mantle beneath northern Kamchatka has unusually slow shear wavespeeds. From the tectonic and volcanic evolution of Kamchatka over the past 10 Myr (refs 3-5) we infer that at least two episodes of catastrophic slab loss have occurred. About 5 to 10 Myr ago, catastrophic slab loss shut down island-arc volcanic activity north of the Aleutian-Kamchatka junction. A later episode of slab loss, since about 2 Myr ago, seems to be related to the activity of the world's most productive island-arc volcano, Klyuchevskoy. Removal of lithospheric mantle is commonly discussed in the context of a continental collision, but our findings imply that episodes of slab detachment and loss are also important agents in the evolution of oceanic convergent margins.     

Migration of a Late Cretaceous fish

Late Cretaceous sediments from the Western Interior of North America yield exceptionally well preserved fossils that serve as proxies for the rapidly changing climate preceding the Cretaceous/Tertiary boundary (about 67-65 Myr ago). Here we reconstruct the ontogenetic history of a Maastrichtian-age fish, Vorhisia vulpes, by using the carbon, oxygen and strontium isotope ratios of four aragonite otoliths collected from the Fox Hills Formation of South Dakota. Individuals of V. vulpes spawned in brackish water (about 70-80% seawater) and during their first year migrated to open marine waters of the Western Interior Seaway, where they remained for 3 years before returning to the estuary, presumably to spawn and die. The mean delta(18)O from the marine growth phase of V. vulpes yields a seawater temperature of 18 degrees C, which is consistent with leaf physiognomy and general-circulation-model temperature estimates for the Western Interior during the latest Maastrichtian.     

Continuing Colorado plateau uplift by delamination-style convective lithospheric downwelling

The Colorado plateau is a large, tectonically intact, physiographic province in the southwestern North American Cordillera that stands at ～1,800-2,000?m elevation and has long been thought to be in isostatic equilibrium. The origin of these high elevations is unclear because unlike the surrounding provinces, which have undergone significant Cretaceous-Palaeogene compressional deformation followed by Neogene extensional deformation, the Colorado plateau is largely internally undeformed. Here we combine new seismic tomography and receiver function images to resolve a vertical high-seismic-velocity anomaly beneath the west-central plateau that extends more than 200?km in depth. The upper surface of this anomaly is seismically defined by a dipping interface extending from the lower crust to depths of 70-90?km. The base of the continental crust above the anomaly has a similar shape, with an elevated Moho. We interpret these seismic structures as a continuing regional, delamination-style foundering of lower crust and continental lithosphere. This implies that Pliocene (2.6-5.3?Myr ago) uplift of the plateau and the magmatism on its margins are intimately tied to continuing deep lithospheric processes. Petrologic and geochemical observations indicate that late Cretaceous-Palaeogene (～90-40?Myr ago) low-angle subduction hydrated and probably weakened much of the Proterozoic tectospheric mantle beneath the Colorado plateau. We suggest that mid-Cenozoic (～35-25?Myr ago) to Recent magmatic infiltration subsequently imparted negative compositional buoyancy to the base and sides of the Colorado plateau upper mantle, triggering downwelling. The patterns of magmatic activity suggest that previous such events have progressively removed the Colorado plateau lithosphere inward from its margins, and have driven uplift. Using Grand Canyon incision rates and Pliocene basaltic volcanism patterns, we suggest that this particular event has been active over the past ～6?Myr.     

Tracing the stepwise oxygenation of the Proterozoic ocean

Biogeochemical signatures preserved in ancient sedimentary rocks provide clues to the nature and timing of the oxygenation of the Earth's atmosphere. Geochemical data suggest that oxygenation proceeded in two broad steps near the beginning and end of the Proterozoic eon (2,500 to 542 million years ago). The oxidation state of the Proterozoic ocean between these two steps and the timing of deep-ocean oxygenation have important implications for the evolutionary course of life on Earth but remain poorly known. Here we present a new perspective on ocean oxygenation based on the authigenic accumulation of the redox-sensitive transition element molybdenum in sulphidic black shales. Accumulation of authigenic molybdenum from sea water is already seen in shales by 2,650 Myr ago; however, the small magnitudes of these enrichments reflect weak or transient sources of dissolved molybdenum before about 2,200 Myr ago, consistent with minimal oxidative weathering of the continents. Enrichments indicative of persistent and vigorous oxidative weathering appear in shales deposited at roughly 2,150 Myr ago, more than 200 million years after the initial rise in atmospheric oxygen. Subsequent expansion of sulphidic conditions after about 1,800 Myr ago (refs 8, 9) maintained a mid-Proterozoic molybdenum reservoir below 20 per cent of the modern inventory, which in turn may have acted as a nutrient feedback limiting the spatiotemporal distribution of euxinic (sulphidic) bottom waters and perhaps the evolutionary and ecological expansion of eukaryotic organisms. By 551 Myr ago, molybdenum contents reflect a greatly expanded oceanic reservoir due to oxygenation of the deep ocean and corresponding decrease in sulphidic conditions in the sediments and water column.     

A late Miocene dust shower from the break-up of an asteroid in the main belt

Throughout the history of the Solar System, Earth has been bombarded by interplanetary dust particles (IDPs), which are asteroid and comet fragments of diameter approximately 1-1,000 microm. The IDP flux is believed to be in quasi-steady state: particles created by episodic main belt collisions or cometary fragmentation replace those removed by comminution, dynamical ejection, and planetary or solar impact. Because IDPs are rich in 3He, seafloor sediment 3He concentrations provide a unique means of probing the major events that have affected the IDP flux and its source bodies over geological timescales. Here we report that collisional disruption of the >150-km-diameter asteroid that created the Veritas family 8.3 +/- 0.5 Myr ago also produced a transient increase in the flux of interplanetary dust-derived 3He. The increase began at 8.2 +/- 0.1 Myr ago, reached a maximum of approximately 4 times pre-event levels, and dissipated over approximately 1.5 Myr. The terrestrial IDP accretion rate was overwhelmingly dominated by Veritas family fragments during the late Miocene. No other event of this magnitude over the past approximately 10(8) yr has been deduced from main belt asteroid orbits. One remarkably similar event is present in the 3He record 35 Myr ago, but its origin by comet shower or asteroid collision remains uncertain.     

A light carbon reservoir recorded in zircon-hosted diamond from the Jack Hills

The recent discovery of diamond-graphite inclusions in the Earth's oldest zircon grains (formed up to 4,252 Myr ago) from the Jack Hills metasediments in Western Australia provides a unique opportunity to investigate Earth's earliest known carbon reservoir. Here we report ion microprobe analyses of the carbon isotope composition of these diamond-graphite inclusions. The observed delta(13)C(PDB) values (expressed using the PeeDee Belemnite standard) range between -5 per mil and -58 per mil with a median of -31 per mil. This extends beyond typical mantle values of around -6 per mil to values observed in metamorphic and some eclogitic diamonds that are interpreted to reflect deep subduction of low-delta(13)C(PDB) biogenic surface carbon. Low delta(13)C(PDB) values may also be produced by inorganic chemical reactions, and therefore are not unambiguous evidence for life on Earth as early as 4,250 Myr ago. Regardless, our results suggest that a low-delta(13)C(PDB) reservoir may have existed on the early Earth.     

Normal faulting in central Tibet since at least 13.5 Myr ago

Tectonic models for the evolution of the Tibetan plateau interpret observed east-west thinning of the upper crust to be the result of either increased potential energy of elevated crust or geodynamic processes that may be unrelated to plateau formation. A key piece of information needed to evaluate these models is the timing of deformation within the plateau. The onset of normal faulting has been estimated to have commenced in southern Tibet between about 14 Myr ago and about 8 Myr ago and, in central Tibet, about 4 Myr ago. Here, however, we report a minimum age of approximately 13.5 Myr for the onset of graben formation in central Tibet, based on mineralization ages determined with Rb-Sr and 40Ar-39Ar data that post-date a major graben-bounding normal fault. These data, along with evidence for prolonged activity of normal faulting in this and other Tibetan grabens, support models that relate normal faulting to processes occurring beneath the plateau. Thinning of the upper crust is most plausibly the result of potential-energy increases resulting from spatially and temporally heterogeneous changes in thermal structure and density distribution within the crust and upper mantle beneath Tibet. This is supported by recent geophysical and geological data, which indicate that spatial heterogeneity exists in both the Tibetan crust and lithospheric mantle.     

Iron meteorite evidence for early formation and catastrophic disruption of protoplanets

In our Solar System, the planets formed by collisional growth from smaller bodies. Planetesimals collided to form Moon-to-Mars-sized protoplanets in the inner Solar System in 0.1-1 Myr, and these collided more energetically to form planets. Insights into the timing and nature of collisions during planetary accretion can be gained from meteorite studies. In particular, iron meteorites offer the best constraints on early stages of planetary accretion because most are remnants of the oldest bodies, which accreted and melted in <1.5 Myr, forming silicate mantles and iron-nickel metallic cores. Cooling rates for various groups of iron meteorites suggest that if the irons cooled isothermally in the cores of differentiated bodies, as conventionally assumed, these bodies were 5-200 km in diameter. This picture is incompatible, however, with the diverse cooling rates observed within certain groups, most notably the IVA group, but the large uncertainties associated with the measurements do not preclude it. Here we report cooling rates for group IVA iron meteorites that range from 100 to 6,000 K Myr(-1), increasing with decreasing bulk Ni. Improvements in the cooling rate model, smaller error bars, and new data from an independent cooling rate indicator show that the conventional interpretation is no longer viable. Our results require that the IVA meteorites cooled in a 300-km-diameter metallic body that lacked an insulating mantle. This body probably formed approximately 4,500 Myr ago in a 'hit-and-run' collision between Moon-to-Mars-sized protoplanets. This demonstrates that protoplanets of approximately 10(3) km size accreted within the first 1.5 Myr, as proposed by theory, and that fragments of these bodies survived as asteroids.