Review article The snowball Earth hypothesis: testing the limits of global change
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Review article The snowball Earth hypothesis: testing the limits of global change Paul F. Hoffman* and Daniel P. Schrag Department of Earth and Planetary Sciences, Harvard University, Cambridge, MA 02138, USA ABSTRACT The gradual discovery that late Neoproterozoic ice sheets alkalinity that could quantitatively account for the world-wide extended to sea level near the equator poses a palaeoenviron- occurrence of cap carbonates. The resulting high rates of mental conundrum. Was the Earth’s orbital obliquity > 60 carbonate sedimentation, coupled with the kinetic isotope (making the tropics colder than the poles) for 4.0 billion years effect of transferring the CO2 burden to the ocean, should drive following the lunar-forming impact, or did climate cool globally down the d13C of seawater, as is observed. If cap carbonates for some reason to the point at which runaway ice-albedo are the ‘smoke’ of a snowball Earth, what was the ‘gun’? In feedback created a ‘snowball’ Earth? The high-obliquity hypo- proposing the original Neoproterozoic snowball Earth hypothe- thesis does not account for major features of the Neoprotero- sis, Joe Kirschvink postulated that an unusual preponderance of zoic glacial record such as the abrupt onsets and terminations land masses in the middle and low latitudes, consistent with of discrete glacial events, their close association with large palaeomagnetic evidence, set the stage for snowball events by (> 10&) negative d13C shifts in seawater proxies, the depos- raising the planetary albedo. Others had pointed out that ition of strange carbonate layers (‘cap carbonates’) globally silicate weathering would most likely be enhanced if many during post-glacial sea-level rise, and the return of large continents were in the tropics, resulting in lower atmospheric sedimentary iron formations, after a 1.1 billion year hiatus, CO2 and a colder climate. Negative d13C shifts of 10–20& exclusively during glacial events. A snowball event, on the other precede glaciation in many regions, giving rise to speculation hand, should begin and end abruptly, particularly at lower that the climate was destabilized by a growing dependency on latitudes. It should last for millions of years, because outgas- greenhouse methane, stemming ultimately from the same sing must amass an intense greenhouse in order to overcome unusual continental distribution. Given the existing palaeomag- the ice albedo. A largely ice-covered ocean should become netic, geochemical and geological evidence for late Neoprote- anoxic and reduced iron should be widely transported in rozoic climatic shocks without parallel in the Phanerozoic, it solution and precipitated as iron formation wherever oxygenic seems inevitable that the history of life was impacted, perhaps photosynthesis occurred, or upon deglaciation. The intense profoundly so. greenhouse ensures a transient post-glacial regime of enhanced carbonate and silicate weathering, which should drive a flux of Terra Nova, 14, 129–155, 2002 depositional environment. Glacigenic fold test was positive (Sumner et al., Introduction deposits (diamictites and ice-rafted 1987), as others subsequently con- In 1987, the Caltech biomagnetist and dropstones) occur in most sections of firmed (Schmidt et al., 1991; Schmidt palaeomagnetist Joe Kirschvink gave the Elatina Formation (Lemon and and Williams, 1995; Sohl et al., 1999). undergraduate Dawn Sumner a rock Gostin, 1990), but previous palaeo- A stratigraphically consistent, polarity sample to study for her senior thesis. magnetic studies suggested that the reversal test (Sohl et al., 1999; see also The sample, collected by UCLA palae- siltstone was deposited close to the Schmidt and Williams, 1995) con- ontologist Bruce Runnegar, was a equator on the basis of unusually firmed the primary component of reddish, uncompacted, rhythmically stable remnant magnetization carried NRM in the Elatina Formation, while laminated siltstone from the Elatina by detrital haematite (Embleton and the existence of multiple reversals (Sohl Formation, a late Neoproterozoic, Williams, 1986). Kirschvink was skep- et al., 1999) suggests that the Elatina glacial and periglacial unit widely ex- tical that glaciers would ever reach sea glacial epoch lasted for several 105 to a posed in the Flinders Ranges and level in the tropics, so he instructed few 106 years. elsewhere in South Australia (Preiss, Sumner to perform a fold test (McElh- The Elatina results refocused atten- 1987; Lemon and Gostin, 1990). The inny and McFadden, 2000) on soft- tion on Neoproterozoic glaciations. A rhythmic laminations are interpreted sediment folds (Williams, 1996) in critical review of the stratigraphic, to be lunar tidal bundles (Williams, Runnegar’s sample. A positive fold geochronological and palaeomagnetic 2000), implying a shallow marine test would prove that the natural rem- constraints on virtually all late Neo- nant magnetization (NRM) is primary; proterozoic glacial deposits (LNGD) *Correspondence: Paul Hoffman, 20 a negative result would show that it is world-wide was recently published Oxford St., Cambridge, MA 02138, USA. secondary, posing no constraint on the (Evans, 2000). A total of 16 regional- E-mail: hoffman@eps.harvard.edu; fax: palaeolatitude of the Elatina glaci- scale units, known to have formed +1 617 496 0434. ation. To Kirschvink’s surprise, the near sea level, possess primary or 2002 Blackwell Science Ltd 129
The snowball Earth hypothesis • P. F. Hoffman and D. P. Schrag Terra Nova, Vol 14, No. 3, 129–155 ............................................................................................................................................................. near-primary NRM components giv- (Fig. 1). The observations are surpri- 1999). LNGD (Fig. 1) are widely dis- ing at least ‘somewhat reliable’ palaeo- sing and they argue that the Elatina tributed on all continents (Mawson, latitudes (Evans, 2000). Many were result is no fluke. 1949a; Cahen, 1963; Harland, 1964; apparently deposited within 10 of the While LNGD (Fig. 2) closely Hambrey and Harland, 1981; Evans, equator, and none was laid down at resemble Phanerozoic glacial deposits 2000) and they are sharply interposed a palaeolatitude greater than 60 lithologically, their distribution and in normal marine carbonate succes- (Fig. 1). Increased non-dipole compo- mode of occurrence have never fitted sions (Fig. 2c) in several regions (Har- nents in the Proterozoic geomagnetic comfortably into Phanerozoic stereo- land and Wilson, 1956; Schermerhorn field (Kent and Smethurst, 1998; Blox- types (Harland, 1964; Schermerhorn, and Stanton, 1963; Martin, 1965; ham, 2000) would not greatly affect 1974; Deynoux, 1985; Crowley and Roberts, 1976; Preiss, 1987). Distinc- those conclusions (Evans, 2000) North, 1991; Eyles, 1993; Crowell, tive ‘cap’ dolostone (and rarely lime- stone) layers sharply overlie most LNGD without significant hiatus (Fig. 2f), implying a sudden switch back to a warmer climate (Norin, 1937; Mawson, 1949b; Deynoux AND Trompette, 1976; Kröner, 1977; Wil- liams, 1979). Cap carbonates (Figs 3– 5) have unusual sedimentological, geochemical and isotopic characteris- tics not found in other Neoproterozo- ic or Phanerozoic carbonates (Aitken, 1991; Fairchild, 1993; Grotzinger and Knoll, 1995; Kennedy, 1996; James et al., 2001), and they occur even in successions otherwise lacking carbon- ate (Spencer, 1971; Deynoux, 1980; Plumb, 1981; Myrow and Kaufman, 1999). Not surprisingly, they have long served as time markers in regional and even interregional corre- lation (Dunn et al., 1971; Kennedy et al., 1998; Walter et al., 2000). The glacial origin of the Elatina Formation was first recognized by Sir Douglas Mawson – the older, thicker and more localized, Surtian LNGD in the same region were recognized much earlier by Howchin (1908), following the first described LNGD in northern Norway (Reusch, 1891) – and he discovered its cap dolostone (Maw- son, 1949b). Although conservative by nature (Sprigg, 1990), Mawson was the first (to our knowledge) to argue that late Neoproterozoic glaciation was global, with large ice sheets in the tropics (Mawson, 1949a). He went on to suggest that climatic amelior- ation paved the way for the first metazoa (Mawson, 1949a), which Fig. 1 Global distribution (a) of Neoproterozoic glaciogenic deposits with estimated palaeolatitudes based on palaeomagnetic data (modified from Evans, 2000). had been discovered by a Mawson ‘Reliability’ takes into account not only palaeomagnetic reliability but also the protegé, Reg Sprigg, in the Ediacara confidence that the deposits represent regionally significant, low-elevation ice sheets Hills west of the Flinders Ranges (Evans, 2000). Histogram (b) of the same glaciogenic deposits according to (Sprigg, 1947). Ironically (for an palaeolatitude. The discontinuous steps show the expected density function of a Adelaide resident), Mawson was uniform distribution over the sphere. Note the preponderance of low-latitude deposits opposed to continental drift, and his and absence of high-latitude deposits. This finding would not be invalidated by argument for tropical ice sheets plausible non-diplole components of the field, which would effectively raise the depended critically on the occurrence palaeolatitudes of only the mid-latitude results (Evans, 2000). The minimum in the of LNGD in tropical Africa today distribution in the subtropics may reflect the meridional variation in precipitation (Mawson, 1949a). [To his credit, minus evaporation due to the Hadley cells. Mawson urged students to read his 130 2002 Blackwell Science Ltd
Terra Nova, Vol 14, No. 3, 129–155 P. F. Hoffman and D. P. Schrag • The snowball Earth hypothesis ............................................................................................................................................................. a b Early Neoproterozoic Twitya Little Dal dolomite shale Lower Twitya cap limestone (40 m) Rapitan diamictite (115 m) c d e f Fig. 2 Neoproterozoic glacial features. Striated pavement (a) overlain by the Smalfjord diamictite (Table 1) at Bigganjargga, Varangerfjord, north Norway, the first recognized Neoproterozoic glacial deposit (Reusch, 1891) (G. P. Halverson photo). Faceted and multistriated stone (b) from the Jbéliat diamictite (Table 1) in Adrar, Mauritania. Rapitan diamictite (Table 1) sharply bounded (c) by carbonate strata, without transitional facies or intercalation at Stone Knife River, Mackenzie Mountains, north- west Canada. Dolomite dropstone in banded iron formation (d) within the Rapitan diamictite at Snake River, Mackenzie Mountains (G. A. Gross photo). Ice-rafted dropstone (e) of shelf-edge oolitic limestone in allodapic slope-facies carbonates of the Ghaub diamictite (Table 1) at east Fransfontein, north-west Namibia. Impact of dropstone on sea-bed caused the folded bicouple seen at 10 o’clock from coin. Typical abrupt conformable contact (f) between the upper dropstone unit of the Ghaub diamictite and its cap dolostone in slope facies at west Fransfontein, north-west Namibia. The coin, pen and hammer are 2, 15 and 33 cm in maximum dimension, respectively (same in Figs 4 and 5). 2002 Blackwell Science Ltd 131
The snowball Earth hypothesis • P. F. Hoffman and D. P. Schrag Terra Nova, Vol 14, No. 3, 129–155 ............................................................................................................................................................. AUSTRALIA CANADA NAMIBIA 1964) and reinforced them with Brachina Sheepbed Maieberg palaeomagnetic measurements. Low- clastic-dominated mixed carbonate-dominated inclination NRM in LNGD and asso- SB SB SB ciated strata in the North Atlantic region and elsewhere seemed initially HST HST HST to prove that ice sheets had indeed cap-carbonate sequence extended to low palaeolatitudes (Har- regressive land and Bidgood, 1959; Bidgood and Harland, 1961; Chumakov and Elston, 1989). Later, however, with the recognition of widespread, low- temperature remagnetization (McCa- be and Elmore, 1989), field tests and maximum flooding MF MF demagnetization procedures that transgressive constrain the age and subsequent MF history of magnetization became the TST TST TST cap dolostone gold-standard for acceptance of SB SB SB palaeomagnetic data (McElhinny and a b c McFadden, 2000). The Elatina For- mation was the first LNGD for which CANADA NAMIBIA MAURITANIA there were multiple field, rock-mag- Twitya Rasthof Jbeliat netic, and petrographic tests indicative base semi-truncated base truncated top truncated of a primary, low-inclination NRM, SB SB SB SB f contemporaneous with sedimentation HST HST (Embleton and Williams, 1986; Sum- SB ner et al., 1987; Schmidt et al., 1991; shale sandstone Schmidt and Williams, 1995; Sohl et al., 1999). micrite calc/dolarenite cap limestone cementstone Alternative theories for low- latitude glaciation MF MF cap dolostone SB SB Kirschvink was not the first to wrestle glacial diamictite d e with the low-latitude Elatina results. George Williams (1975, 1993, 2000) Fig. 3 Schematic variability in cap-carbonate sequences (not drawn to scale). Note had long advocated a large (> 54) distinction between ‘cap dolostone’ and ‘cap-carbonate sequence’. Workers from orbital obliquity, or tilt (the angle clastic-dominated (a) regions (e.g. Kennedy, 1996; Kennedy et al., 2001b) regard cap between the Earth’s axes of rotation dolostones as the sole representatives of the post-glacial transient. However, unusual and orbit around the sun), to account textures (e.g. crystal-fan cementstones) occur above the cap dolostone in mixed (b) for low-latitude glaciation in conjunc- and carbonate-dominated (c) regions, and the post-glacial negative d13C anomaly tion with inferred strong equatorial may encompass the entire cap-carbonate sequence (Fig. 9). The top of the cap seasonality during the Elatina glaci- dolostone does not mark the end of enhanced silicate weathering (Kennedy et al., ation (Williams and Tonkin, 1985; 2001b), but may reflect a change from carbonate-dominated to silicate-dominated Williams, 1998). In theory, obliquity weathering as carbonate platforms are inundated by rising sea level (Hoffman and could have varied chaotically within Schrag, 2000). Base truncation (d,e) may occur if the local water mass fails to reach 60–90 following the early lunar-form- saturation until after post-glacial sea-level rise. Top truncation (f) occurs in regions ing impact event, but a low obliquity, with little background tectonic subsidence to create permanent accommodation once achieved, is self-stabilizing space. Abbreviations: HST, highstand systems tract; MF, marine flooding surface; through the gravitational interaction SB, sequence boundary; TST, transgressive sequence tract (van Wagoner et al., 1988). of the Moon and the Earth’s equator- ial bulge (Laskar et al., 1993). Wil- contemporary’s book (Wegener, 1922) 1971; McElhinny et al., 1974; Crowell, liams (1993) postulates a sharp and Sprigg, for one, identified the 1983, 1999; Eyles, 1993). decrease in orbital obliquity after Adelaidean succession as a pre-Meso- Most, but not all. Brian Harland of 600 Ma, with resultant moderation zoic rifted continental margin (Sprigg, Cambridge University cut his teeth in of the seasonal climate cycle giving 1952).] Mawson’s argument, however, the Arctic archipelago of Svalbard, rise to the metazoa (Williams, 1993). collapsed in the plate tectonic revolu- sentinel of the Barents Sea shelf, and With an obliquity > 54, mean tion, with most subsequent workers host to a pair of LNGD (Harland annual temperatures are lower around attributing the extent of LNGD et al., 1993; Harland, 1997). Harland the equator than at the poles (Wil- (Fig. 1) to rapid drift of different was no fixist, but he independently liams, 1975; Hunt, 1982; Oglesby and continents through polar regions at reiterated Mawson’s arguments (Har- Ogg, 1999; Jenkins, 2000), making different times (Crawford and Daily, land and Rudwick, 1964; Harland, glaciation more probable at lower 132 2002 Blackwell Science Ltd
Terra Nova, Vol 14, No. 3, 129–155 P. F. Hoffman and D. P. Schrag • The snowball Earth hypothesis ............................................................................................................................................................. rian Camb m trail F Game e ston ime shale ho ok l pbed Hay Shee ne osto dol cap ro ok Ic eB m el eF a Ke d b c e f Fig. 4 Distinctive lithofacies in the Ice Brook (Table 1) cap carbonate, Mackenzie Mountains, north-west Canada. The Ice Brook cap-carbonate sequence (a) at Shale Lake includes a very pale, flinty, cap dolostone, locally crowned by dark Hayhook limestone cementstone, then 700–900 m of black, deep-water, Sheepbed shale, shoaling finally to Gametrail dolomite, truncated here by the Lower Cambrian (photo from James et al., 2001). Low-angle cross-sets of reverse-graded peloids (b) in the cap dolostone at Gayna River. Cuspate mega-ripples, aka pseudo-tepees (c) in the cap dolostone at Arctic Red River. Axial zone of cuspate mega-ripple (d) at Gayna River, showing evidence of oscillatory (storm wave) currents sweeping across the aggrading crestline. Digitate pseudo- stromatolites composed of sea-floor barite cement (e) at the contact between the cap dolostone and Hayhook limestone at Shale Lake. The slender dark prisms growing off the tips of the barite fingers near the top of the photo are pseudomorphosed aragonite. Crystal fans of aragonite (f), replaced by calcite, and meniscus-laminated micrite in the Hayhook cap limestone at Shale Lake. latitudes (Williams, 1975). Summer sky throughout the diurnal cycle) that water (Oglesby and Ogg, 1999). The insolation is so large in the polar surface temperatures over land areas tropics, although cooler overall, regions (the sun staying high in the might exceed the boiling point of would have two hot seasons near 2002 Blackwell Science Ltd 133
The snowball Earth hypothesis • P. F. Hoffman and D. P. Schrag Terra Nova, Vol 14, No. 3, 129–155 ............................................................................................................................................................. Maieberg Fm cap carbonate begins herms inter-reef bio glacial sea-level fall fill tstone tub e cemen ite pre-glacial δ 13C drop reefs ian a eol rg mb Ke i l b e ck Ombaatjie Fm lo e tb bonat en m car Gruis Fm em tfor ba s pla d rate te 100 m lo me ot a ng r co Rasthof Fm a (truncated beneath Gruis Fm) b c e d f g 134 2002 Blackwell Science Ltd
Terra Nova, Vol 14, No. 3, 129–155 P. F. Hoffman and D. P. Schrag • The snowball Earth hypothesis ............................................................................................................................................................. et al., 2002) explanation for the exist- within middle to low latitudes, a Fig. 5 Distinctive lithofacies in the cap- ence of well-developed, periglacial, situation that has not been encoun- carbonate sequence (Maieberg Forma- sand-wedge polygons in the Elatina tered at any subsequent time in Earth tion) following the Ghaub glaciation Formation (Williams and Tonkin, history. One effect of such a palaeo- (Table 1) of the Otavi platform, north- 1985). These structures are most easily geography would be a substantially west Namibia, and possible correlatives. explained as resulting from large sea- higher albedo in the subtropics, where Section (a) near the platform margin at sonal oscillations in soil temperature clouds are least important (Kirsch- Tweelingskop, showing preglacial drop (Lachenbruch, 1962), which should vink, 1992). Any resultant glaciation in d13C, aeolian dolostone representing not occur near the equator if the would further increase the Earth’s the glacial event, mounded cap dolo- obliquity was close to its present value albedo by lowering sea level, exposing stone (Keilberg member) composed of of 23.5. On the other hand, a persist- continental shelves and inland seas stromatolites with tube-like structure, ent large obliquity does not explain (Kirschvink, 1992). Placing more and reef-like buildups of sea-floor ara- the characteristically abrupt onsets continents in the tropics would also gonite cements pseudomorphosed by and terminations of LNGD (Hambrey increase the silicate weathering rate calcite (Soffer, 1998). Tube-like struc- and Harland, 1981; Preiss, 1987; Nar- (provided tectonic uplift was not mys- tures (b) in hummocky cross-stratified cap dolostone at Khowarib Schlucht. bonne and Aitken, 1995; Hoffmann teriously absent there), leading to a Tubes initiated within the cap dolostone and Prave, 1996; Hoffman et al., colder planet with lower atmospheric and do not tap subglacial deposits as 1998a). Nor does it account for the pCO2 (Marshall et al., 1988; Worsley expected if they formed by methane close association of LNGD with car- and Kidder, 1991). In addition, the released from glacial permafrost as pro- bonates, because the ‘redistribution of meridional heat transport by the Had- posed by Kennedy et al. (2001a). Detail the radiant energy balance to polar ley cells would be weakened because of tube-like structures (c) at Khowarib latitudes should also move the car- tropical air masses would be drier on Schlucht showing meniscus-laminated bonate belts from equatorial latitudes average with greater continentality. micrite (tubes) filling circular pits within to the poles, where the glaciers (in Oceanic heat transport might also be a honeycomb of stromatolites (arched Williams’ model) should not encoun- lessened, but this is less certain. These laminae). Despite vertical continuity of ter them’ (Kirschvink, 1992). The combined effects might lead to the tube-like structures (b), synoptic relief dynamics of the postulated obliquity growth of large ice caps, nucleated on during growth was < 3 cm. Sheaves of reduction after 600 Ma remain con- islands or continents bordering the former-aragonite crystal fans (d) in jectural (Williams, 1993; Ito et al., polar seas. allodapic limestone at Tweelingskop. 1995; Néron de Surgy and Laskar, Ice caps create a positive feedback Interlayered micrite (light) and aragon- 1997; Williams et al., 1998; Pais et al., on climate change through ice-albedo ite (dark) cements (e) pseudomorphosed 1999). feedback (Croll, 1867; Budyko, to calcite in the Sete Lagoas cap car- Richard Sheldon (1984) postulated 1966; Manabe and Broccoli, 1985). bonate, Bambuı́ platform, south-eastern that orbiting ice rings episodically Early energy-balance climate models Brazil. Barite crust (f) and reworked collapsed into the atmosphere, tran- revealed a fundamental instability in barite-crusted clast (above pen) at sequence boundary between Jbéliat siently shielding sunlight and giving the climate system caused by this (Table 1) cap dolostone and Téniagouri rise to glaciers at low latitudes. The positive feedback (Eriksson, 1968; Group (Fig. 10) at Atar Cliffs, Adrar, theory strives to explain iron forma- Budyko, 1969; Sellers, 1969; North Mauritania. Unlike sea-floor barite tions and cap carbonates, but does et al., 1981), an effect reproduced in a (Fig. 4e), Jbéliat barite appears to be a not adequately account for low-lati- number of GCMs (Wetherald and product of phreatic diagenesis in a tude glaciation. Due to orbital Manabe, 1975; Jenkins and Smith, supratidal tepee breccia, possibly related obliquity, equatorial ice rings will 1999; Hyde et al., 2000; Pollard and to mixing of marine and meteoric cast shadows only on the winter Kasting, 2001). If more than about groundwaters. True tepee structure (g) hemisphere (see figure 2 in Sheldon, half the Earth’s surface area were to formed by layer-parallel expansion due 1984). Glaciation depends critically become ice covered, the albedo feed- to vadose cementation in supratidal on cool summers, not cold winters back would be unstoppable (Fig. 6). Jbéliat cap dolostone at Atar Cliffs. (Köppen and Wegener, 1924), as it is Surface temperatures would plummet Jbéliat tepee ridges are polygonal in contingent on a fraction of the (Fig. 7), and pack ice would quickly plan, unlike linear pseudo-tepees annual snowfall surviving the melting envelope the tropical oceans (Hyde (mega-ripples) in sublittoral cap dolo- seasons. et al., 2000; Baum and Crowley, stones (Fig. 4c,d). 2001). With reduced Neoproterozoic solar forcing, the ice-albedo instability Snowball Earth occurs in the GENESIS v2 GCM equinox, when insolation would be Kirschvink responded to the positive between 1.0 and 2.5 times modern indistinguishable from the hottest Elatina fold test with a theory he first pCO2 with different palaeogeograph- tropical seasons with low obliquity aired in 1989 (Maugh, 1989). It even- ies (Baum and Crowley, 2001; Pollard (Oglesby and Ogg, 1999). Indeed, hot tually appeared as an article, seven and Kasting, 2001). The time scale of summers and strong seasonality make paragraphs long, embedded in a 1348- the sea-ice advance is uncertain – the glaciation problematic at any latitude page book (Kirschvink, 1992). It pos- ocean’s thermal inertia resists the ice with large obliquities (Hoffman and tulated that conditions amenable to advance initially (Poulsen et al., Maloof, 1999). It does, however, pro- global glaciation were set up by an 2001), but the ocean is a finite heat vide a ready but non-unique (Maloof unusual preponderance of land masses reservoir and cooling over thousands 2002 Blackwell Science Ltd 135
The snowball Earth hypothesis • P. F. Hoffman and D. P. Schrag Terra Nova, Vol 14, No. 3, 129–155 ............................................................................................................................................................. Effective solar flux (wrt present) (Walker, 2001). Despite mean annual 1.1 1.2 1.3 temperatures well below freezing 0.9 1.0 everywhere, afternoon temperatures o 6 ice-free branch 5 in the summer hemisphere would Solar flux ca 600 Ma 90 reach the melting point (Walker, unstable range 2001). Evaporation of transient melt 0.7 meridional water would contribute, along with 1 heat transport Ice line latitude o 7 sublimation, to maintain low levels of 60 Present atmospheric water vapour, and gla- ciers would feed on daily updrafts of stable points this moisture (Walker, 2001). The o with Es=1.0 global mean thickness of sea ice 30 2 depends strongly on sea-ice albedo unstable range ( 1.4 km for albedo 0.6) and meridional variability is a complex function of solar incidence, 0o 3 4 greenhouse forcing (see below), zonal ice-covered branch albedo, ablation or precipitation, and 10-5 10-4 10-3 10-2 10-1 equatorward flowage of warm basal pCO2 (bar) ice (Warren et al., 2002). Climate physicists originally assu- Fig. 6 Ice-line latitudes (at sea level) as a function of the effective solar flux (Es), or med that no ice-albedo catastrophe equivalent pCO2 (for Es ¼ 1.0), based on a simple energy-balance model of the ever actually occurred because it Budyko-Sellers type (after Caldeira and Kasting, 1992; Ikeda and Tajika, 1999). would be permanent: the high planet- Effect of a 30% reduction in meridional heat transport is shown, as is the estimated ary albedo would be irreversible. A solar flux at 600 Ma. Of three possible stable points for Es ¼ 1.0, the Earth saviour exists, however, and Kirsch- actually lies on the partially ice-covered branch at point 1. An instability due to ice- vink (1992) identified it as the buildup albedo feedback drives any ice-line latitude < 30 onto the ice-covered branch. A of an intense atmospheric CO2 green- pCO2 » 0.12 bar is required for deglaciation of an ice-covered Earth, assuming the house through the action of plate planetary albedo is 0.6 and Es ¼ 1.0 (Caldeira and Kasting, 1992). The snowball tectonics in driving the long-term Earth hypothesis is qualitatively predicated on these findings and infers a hysteresis in carbon cycle (Walker et al., 1981; pCO2 (and consequently surface temperature) following the circuit labelled 1–7. Caldeira and Kasting, 1992; Kirsch- Starting from point 1, lowering of pCO2 causes ice lines to migrate stably to point 2, vink, 1992). On a snowball Earth, whereupon runaway ice-albedo feedback drives ice lines to the ice-covered branch at point 3. Normal volcanic outgassing over millions of years increases pCO2 to point 4, volcanoes would continue to pump initiating deglaciation. Reverse ice-albedo feedback then drives ice lines rapidly to the CO2 into the atmosphere (and ocean), ice-free branch at point 5, where high pCO2 combined with low planetary albedo but the sinks for CO2 – silicate weath- creates a transient ultra-greenhouse. Enhanced silicate weathering causes lowering of ering and photosynthesis – would be pCO2 to point 6, whereupon polar ice caps reform and ice lines return to the partially largely eliminated (Kirschvink, 1992). ice-covered branch at point 7. In the 1960s, Budyko was concerned with the small ice- Even if CO2 ice precipitated at the cap instability, which predicts a possible switch to the ice-free branch (e.g. poles in winter, it would likely subli- disappearance of Arctic sea ice) due to anthropogenic global warming. mate away again in summer (Walker, 2001). CO2 levels would inexorably rise and surface temperatures would of years would leave it powerless to cross the ice-albedo instability thresh- follow, most rapidly at first and more resist the ice encroachment. It is also old [but the continents are glaciated slowly later on (Fig. 7) due to the non- uncertain if the tropical ocean would because they are mostly placed in linear relation between CO2 concen- ever become entirely ice covered middle and high latitudes, contrary tration and the resultant greenhouse (Hyde et al., 2000; Baum and Crow- to palaeomagnetic evidence (Evans, forcing (Caldeira and Kasting, 1992). ley, 2001; Warren et al., 2002) 2000)]. With rising surface temperatures, sea Kirschvink (1992) speculated that Assuming an albedo runaway did ice thins but ground ice sheets expand areas of open water (polynyas) would occur, the climate would be domin- in some areas due to a stronger remain, tracking the zone of highest ated by the dry atmosphere and the hydrological cycle. If CO2 outgassing solar incidence back and forth across low heat capacity of the solid surface rates were broadly similar to today (a the equator and imparting a strongly (Walker, 2001). It would be more like reasonable assumption for 600– seasonal climate even at low latitude, Mars (Leovy, 2001) than Earth as we 700 Ma), then the time needed to consistent with geological observa- know it, except that the greater at- build up the estimated 0.12 bar CO2 tions (Williams and Tonkin, 1985). mospheric pressure would allow sur- required to begin permanent melting This is distinct from the tropical face meltwater to exist. Diurnal and at the equator, assuming a planetary ‘loophole’ model (Hyde et al., 2000; seasonal temperature oscillations albedo of 0.6, would be a few million Baum and Crowley, 2001; Crowley would be strongly amplified at all years (Caldeira and Kasting, 1992; et al., 2001), in which the ice fronts latitudes because of weak lateral heat Crowley et al., 2001). This estimate miraculously approach but never transfer and extreme ‘continentality’ (Fig. 7), while subject to large uncer- 136 2002 Blackwell Science Ltd
Terra Nova, Vol 14, No. 3, 129–155 P. F. Hoffman and D. P. Schrag • The snowball Earth hypothesis ............................................................................................................................................................. In Ta Y Au K M Si Co La 7 Am Ba WA o global mean surface temperature ( C) In -50 0 50 Ta Y Au K M Si Co La 5 Am Ba WA In Ta 5 Y elapsed time (x106 years) since ice-albedo runaway Au K M Si Co La 4 Am Ba WA In Ta Y Au K M Si Co La 3 Am Ba WA 0 In Ta Y Au K M Si Co La 2 Am Ba WA 223 243 263 283 303 323 In Ta Y global mean surface temperature (K) Au K M Si Co La 1 Am Ba WA Fig. 7 Estimated changes in global mean surface temperature, based on energy-balance calculations, and ice extent through one complete snowball event. The suggested time scale of the event of 5 Myr is conservative for an albedo ¼ 0.6, based on the estimated outgassing flux of CO2 and subsidence analysis (Hoffman et al., 1998b). The global palaeogeographical model (Powell et al., 2001) pertains to 750 Ma, 30 Myr before the ‘Sturtian’ glaciation (Table 1). Palaeocontinents: Am, Amazonia; Au, Australia; Ba, Baltica; Co, Congo; In, India; K, Kalahari; M, Mawson; Si, Siberia; Ta, Tarim; WA, West Africa; Y, South China (Yangtse). The global ice-line depictions correspond approximately to points 1–7 in Fig. 2. Note the growth of terrestrial ice sheets with rising surface temperature during the snowball event. Note also the abrupt onset and termination of glacial conditions in the low and middle latitudes, consistent with geological observations, and the saw-tooth form of the temperature curve reversed to that associated with late Quaternary glacial cycles. Note finally that the estimated surface temperatures are global mean values and give no sense of the real zonal, seasonal and diurnal ranges in temperature (Walker, 2001). tainties, is of the same order as the stratigraphic (Hoffman et al., 1998a) ciation proceeds rapidly due to reverse estimated time-scale of LNGD from evidence. Once the tropical ocean ice-albedo feedback (Caldeira and palaeomagnetic (Sohl et al., 1999) and begins to open up perennially, degla- Kasting, 1992; Crowley et al., 2001). 2002 Blackwell Science Ltd 137
The snowball Earth hypothesis • P. F. Hoffman and D. P. Schrag Terra Nova, Vol 14, No. 3, 129–155 ............................................................................................................................................................. The fall in planetary albedo occurs far curred in the 1990s, and carbon synchroneity of glacial events in faster than the excess atmospheric isotope stratigraphy was principally widely separate areas (Africa, Austra- CO2 can be consumed by silicate responsible. lia, Canada, Europe) and base global weathering, with the result that a The history of carbon isotopic vari- correlation schemes upon them. transient heat wave must follow the ation (expressed as d13C, the differ- The conventional explanation for ice retreat (Caldeira and Kasting, ence in per mil of 13C/12C with respect Neoproterozoic carbon isotopic oscil- 1992; Kirschvink, 1992). Kirschvink to standard VPDB) in marine carbon- lations and cap carbonates is oceano- (1992) concluded that a cold planet ates and covarying organic matter is a graphic (Knoll et al., 1986, 1996; with many tropical continents and robust, but ambiguous, record of bio- Kaufman et al., 1991, 1997; Derry large polar sea-ice caps ‘would be a geochemical cycling (Summons & et al., 1992; Grotzinger and Knoll, rather unstable situation, with the Hayes, 1992). Despite the fact that 1995; Grotzinger and James, 2000). potential for fluctuating rapidly significant spatial heterogeneity in The idea is roughly the following. The between the ‘‘ice house’’ and ‘‘green- d13C is dynamically maintained within ocean was physically stratified for house’’ states’. the ocean (Kroopnick, 1985), large long periods pending glaciation, but secular changes in d13C at individual switched to a turnover mode during sites are globally correlated (Veizer and particularly after glacial events. In Testing the snowball hypothesis et al., 1999; Saltzman et al., 2000), stratified mode, the biological pump Kirschvink (1992) highlighted three reflecting the residence time of carbon (descent of organic particles from the implications of his theory which 150 times the ocean mixing time surface) drove the d13C of dissolved ‘might lend themselves to geological (Kump and Arthur, 1999). The d13C inorganic carbon (DIC) in the surface tests’. First, ‘glacial units should be of carbonate rocks is relatively waters to higher values. Partial rem- more or less synchronous Harland, immune from large diagenetic changes ineralization of the organic ‘rain’ in 1964)’. Second, ‘strata from widely because pore fluids are effectively conjunction with bacterial sulphate separate areas which preserve a record buffered by the large rock carbon reduction in the water column pro- of these [rapid] climatic fluctuations reservoir (Veizer et al., 1999). duced deep water laden with 13C- [between ‘‘ice house’’ and ‘‘green- The range of secular d13C variabil- depleted DIC. With prolonged strati- house’’ states] might bear an overall ity of > 10& in the late Neoprotero- fication, the contrast in d13C values similarity in lithologic character, zoic (Jacobsen and Kaufman, 1999; between the surface and deep water which would be a result of the global Walter et al., 2000) is markedly great- DIC reservoirs (2–3& in the present scale of the climatic fluctuations’. er than for the Phanerozoic (Veizer ocean, Kroopnick, 1985) increased to Third, ‘the presence of floating pack et al., 1999) or the 1.5 billion years > 10&, consistent with the preglacial ice should reduce evaporation, act to before the late Neoproterozoic (Bra- high d13C values (Fig. 8). When turn- decouple oceanic currents from wind sier and Lindsay, 1998; Kah et al., over began, anoxic, alkalinity-laden, patterns and, by inhibiting oceanic to 2001). The largest Neoproterozoic deep water upwelled to the surface, atmospheric exchange of O2, would d13C anomalies consistently bracket releasing CO2 and precipitating cap enable the oceanic bottom waters to LNGD (Figs 8 and 9), with enriched carbonates with low d13C values. Like stagnate and become anoxic. Over values (d13Ccarb > 5&) observed the snowball hypothesis, the ocean time, ferrous iron generated at the prior to glaciation and depleted values ‘turnover’ model (Grotzinger and mid-oceanic ridges or leached from (d13Ccarb < 0&) afterwards. A steep Knoll, 1995; Knoll et al., 1996) seeks the bottom sediments would build up decline in d13C values of 10–15& to explain many of the salient obser- in solution and, when circulation precedes any physical evidence of vations, but the basic premise is became reestablished toward the end glaciation or sea-level fall in many faulty. First, it is physically difficult of the glacial period, the iron would areas (Narbonne and Aitken, 1995; if not impossible to shut down the oxidize to form a ‘‘last-gasp’’ blanket Hoffman et al., 1998b; Brasier and overturning circulation for periods of banded iron formation in upwelling Shields, 2000; McKirdy et al., 2001; > 1.5 kyr (Zhang et al., 2001). Sec- areas.’ Here is how these tests turned Xiao et al., 2001; Halverson et al., ond, organic productivity would crash out. 2002). Cap carbonates are universally if the nutrient-rich upwelling flux were depleted, in some cases approaching removed (Hotinski et al., 2001). Con- mantle values of )6 ± 1& (Knoll and sequently, stratification would not Test 1: carbon isotopes Walter, 1992; Kaufman and Knoll, raise the d13C in surface waters to 5– Direct radiometric testing of predic- 1995; Kennedy, 1996; Kaufman et al., 10&, as is observed. Third, cap car- ted glacial synchroneity has not 1997; Hoffman et al., 1998b; Prave, bonates formed during post-glacial proved possible (Table 1) because 1999a; Brasier and Shields, 2000; sea-level rise (see below), when the of an apparent dearth of suitable Walter et al., 2000; James et al., injection of glacial meltwater would material for dating (Evans, 2000). 2001). Although interpretations of suppressed not enhance, ocean turn- Low diversity and turnover rates of these remarkable trans-glacial isotopic over. Kennedy (1996) attributes cap microfloral and protistan remains shifts differ widely (see below), there is carbonates and their low d13C values pose severe limits on biostratigraphic consensus that they are supra-regional simply to post-glacial sea-level rise, resolution until after the major Neo- in nature. Even those opposed to the but his model also depends on an 8– proterozoic glacial events (Knoll, snowball hypothesis (Kennedy et al., 10& difference in d13C across the 2000). Regardless, a sea change in 1998, 2001a,b; Prave, 1999a,b; Con- thermocline, compared with 2.5& favour of synchronous glaciation oc- don and Prave, 2000) agree on the for Mesozoic–Cenozoic oceans. 138 2002 Blackwell Science Ltd
Terra Nova, Vol 14, No. 3, 129–155 P. F. Hoffman and D. P. Schrag • The snowball Earth hypothesis ............................................................................................................................................................. (1999); 7. Clauer (1987); 8. Brasier et al. (2000); 9. Norin (1937); 10. Evans (2000); 11. Walter et al. (2000); 12. Grey and Corkeron (1998), Li (2000); 13. Frimmel et al. (2002) [McMillan (1968) would place Kaigas post-741±6 Ma]; 14. Hoffman et al. (1996); 15. Corsetti and Kaufman (in press); 16. Narbonne and Aitken (1995). Correlation of glaciations by column is rather arbitrary due to meagre radiometric age constraints. Glacials in bold type are All ages U–Pb zircon, except Rb–Sr fine-mica ages (*). For primary sources see: 1. Gorokhov et al. (2001); 2. Harland (1997); 3. Dempster et al. (2002); 4. Evans (2000); 5. Thompson and Bowring (2000); 6. Myrow and Kaufman In 1998, we invoked the snowball Age (Ma) hypothesis, then in eclipse, as a 777 ± 7 772 ± 5 778 ± 2 possible explanation for the d13C pro- files (Figs 8 and 9) of carbonates bounding glacial horizons in northern Glaciation? Namibia (Table 1). Hoffman et al. (1998b) gave three, mutually compat- ible explanations for the 12C enrich- Kaigas ment observed in cap carbonates. First, organic productivity on snow- 723 + 16/-10 ball Earth would be greatly reduced Age (Ma) 755 ± 18 748 ± 12 for 106–107 years, which should cause 746 ± 2 742 ± 2 741 ± 6 d13C values to shift towards the hydrothermal CO2 input ()6 ± 1& VPDB, Des Marais and Moore, 1984) Glaciation Konnarock Tereeken at mid-ocean ridges (Kump, 1991). ‘Sturtian’ Sturtian Rapitan Surprise Jiankou Gubrah Second, if carbonate sedimentation Chuos Walsh rates globally were very high, due to the alkalinity flux from weathering in the greenhouse transient, the organic Age (Ma) fraction of the burial flux would be 700–730 negatively effected. Consequently, 630* 630* carbonate would have low d13C values even if organic productivity had fully recovered from the snowball event Hankalchoug Glaciation Port Askaig ‘Marinoan’ Ice Brook Elbobreen Landrigan (Hoffman and Schrag, 1999). Third, Smalfjord Wildrose Numees Elatina Shuram Nantuo Jbéliat Ghaub the transfer of atmospheric CO2 to bicarbonate (via the weathering cycle) involves an 8& isotopic fraction- ation favouring 13C. In transferring 604 + 4/-3 Age (Ma) 601 ± 4 595 ± 2 > 98% of the built-up atmospheric 595* CO2 to the ocean and ultimately the sedimentary reservoir, the kinetic iso- tope effect (Rayleigh distillation) will drive the atmospheric CO2 reservoir to ever lower d13C values (Hoffman Loch na Cille Wilsonbreen Glaciation Mortensnes et al., 1998b). To the degree that ‘Varanger’ Squantum Rainstorm Gaskiers atmospheric CO2 remains a major Table 1 Radiometric age constraints on Neoproterozoic glacial events Egan source of DIC, this would be reflected by cap-carbonate d13C values that decline with time (Fig. 8). However, the nadir in d13C (Fig. 9) does ne- Palaeocontinent Age (Ma) cessarily correspond to the comple- 565 ± 3 tion of atmospheric CO2 drawdown, 560* 570 570 as Kennedy et al. (2001b) argue, but reflects the net effect of many fac- tors, including those already men- South China West Africa tioned, as well as methane hydrate Aus-Maws Aus-Maws Laurentia Laurentia Laurentia Laurentia Laurentia Kalahari Avalaon destabilization during deglaciation Avalon Baltica Arabia Congo Tarim (Kaufman et al., 1997; Kennedy et al., 2001a; but see Hoffman et al., 2002a) and dolomite–calcite isotopic equilibrium. 4. US Appalachians (Blue Ridge) The swing to negative d13C values 5. New England, USA (Boston) 6. Newfoundland (Conception) 16. NW Canada (Windermere) 11. South Australia (Adelaide) 12. NW Australia (Kimberley) 2. NE Svalbard (Hecla Hoek) observed beneath the glacial deposits 15. SW USA (Death Valley) 3. SW Scotland (Dalradian) 1. N Norway (Verstertana) 9. NW China (Quruqtagh) 13. SW Namibia (Gariep) 10. South China (Sinian) 14. NW Namibia (Otavi) in many regions (Figs 8 and 9) must Region (succession) 7. Mauritania (Jbéliat) discussed in the text. originate differently (Kennedy et al., 2001b; Halverson et al., 2002; Schrag 8. Oman (Huqf) et al., 2002) as they are separated in time from the cap-carbonate anomal- ies by the glacial periods which were of long duration (Hoffman et al., 2002 Blackwell Science Ltd 139
The snowball Earth hypothesis • P. F. Hoffman and D. P. Schrag Terra Nova, Vol 14, No. 3, 129–155 ............................................................................................................................................................. CANADA NAMIBIA AUSTRALIA m -5 0 5 10 m -5 0 5 10 m -5 0 5 10 50 50 50 Sheepbed a b c Maieberg Brachina 40 40 40 no δ13Ccarb data in shale no δ13Ccarb data in shale 30 30 30 Hayhook 20 20 Ravens- Keilberg 10 10 throat Nuccaleena 0 0 0 ICE BROOK GLACIATION GHAUB GLACIATION ELATINA GLACIATION 0 0 0 -20 d -20 e Trezona f -40 13 no δ Ccarb data in clastics -40 -500 -60 Enorama -60 Ombaatjie -80 -80 Keele -100 -1000 -120 Etina -140 -120 -160 -1500 -140 -180 -160 -200 -2000 -220 -180 SECTION NOT SHOWN SECTION NOT SHOWN SECTION NOT SHOWN 80 80 80 g h i Tindelpina 13 60 no δ Ccarb data in shale 60 60 Rasthof Twitya 40 40 40 20 20 20 0 0 0 RAPITAN GLACIATION CHUOS GLACIATION STURTIAN GLACIATION -5 0 5 10 -100 no δ13Ccarb data in clastics j -5 0 5 10 δ13C (o/oo VPDB) -200 subaerial exposure surface Ombombo -300 cap limestone rhythmite cap limestone (cementstone) cap dolostone -500 aeolian sandstone intertidal microbialaminite -600 cross-bedded grainstone terrigenous sandstone -700 columnar stromatolite wavy-laminated micrite -5 0 5 10 flat-laminated micrite shale, siltstone δ13C (o/oo VPDB) Fig. 8 Representative d13C profiles for carbonates bounding the older and younger Neoproterozoic glacials (Table 1) in the northern Canadian Cordillera (left), the Otavi platform in Namibia (centre), and the Adelaide succession in South Australia (right). Note high values (> 5&) before glaciation and low values (< 0&) afterwards, and the drop in values that leads the younger glaciation. Correlation follows Kennedy et al. (1998). Sources of data are (a) James et al. (2001) Redstone section, (b) unpublished Khowarib Schlucht section, (c) unpublished Parachilna Gorge section, (d) unpublished Stone Knife River section, (e) Hoffman et al. (1998b) and unpublished Keiserfontein section, (f) McKirdy et al. (2001), (g) unpublished Stone Knife River section, (h) unpublished Ombonde River section, (i) McKirdy et al. (2001), (j) unpublished Omutirapo section. Note non-uniform thickness scales in panels d–f. 1998a; Sohl et al., 1999). We will Spencer, 1972; Williams, 1979; Dey- Test 2: cap carbonates discuss the preglacial anomalies, both noux, 1980; Fairchild and Hambrey, positive and negative (Fig. 8), and the Most LNGD are capped by continu- 1984; Miller, 1985; Preiss, 1987; Ait- larger problem of how snowball ous layers (Figs 3–5) of pure dolo- ken, 1991; Fairchild, 1993; Brookfield, events originate, in the penultimate stone (and locally limestone), metres 1994; Grotzinger and Knoll, 1995; section of the paper. to tens of metres thick (Spencer and Kennedy, 1996; Hoffman et al., 140 2002 Blackwell Science Ltd
Terra Nova, Vol 14, No. 3, 129–155 P. F. Hoffman and D. P. Schrag • The snowball Earth hypothesis ............................................................................................................................................................. Elandshoek platform 400 HST Maieberg cap-carbonate sequence 300 Limestone: δ O18 δ13C Dolomite: δ O18 δ13C (alteration zone) 200 100 MFS maximum flooding TST Keilberg cap dolostone 0 Ombaatjie platform -50 ( -10 -5 0 5 subaerial exposure surface ) o/oo VPDB Fig. 9 New representative d18O and d13C profiles for carbonates bounding the Ghaub (Table 1) glacial surface on the Otavi platform (Khowarib Schlucht section) in northern Namibia. Pockets of diamictite occur sparingly on the glacial surface overlain by the Keilberg cap dolostone. High d13C values > 5& prevail for 460 m beneath the preglacial drop, which is erosionally truncated in this section compared with that shown in Fig. 8(e). Section was measured and collected on the back side of the ridge shown, where the recessive interval of maximum flooding is continuously exposed. Change in scale at the datum is necessitated by the photographic perspective. Lithofacies as in Fig. 8. 1998a; Corkeron and George, 2001; preglacial rocks disconformably ‘Cap-carbonate sequences’ (see below James et al., 2001). Their basal con- (Wright et al., 1978; Preiss, 1987; Ait- and Fig. 3) share a distinct set of tacts are characteristically abrupt ken, 1991; Hegenberger, 1993; Hoff- unusual sedimentary structures (Figs 2c, 2f, 4a and 5a), with little mann and Prave, 1996). They range (Cloud et al., 1974; Peryt et al., 1990; evidence of reworking or significant from thin, allodapic, deep-water Aitken, 1991; Fairchild, 1993; Grotz- hiatus (Norin, 1937; Williams, 1979; deposits (Kennedy, 1996), through inger and Knoll, 1995; Kennedy, 1996; Eisbacher, 1981; Deynoux, 1982; Pre- agitated shelfal facies (James et al., Hoffman et al., 1998a; James et al., iss, 1987; Hoffman et al., 1998a; Bra- 2001), near-shore grainstones and 2001), and they are consistently sier et al., 2000). Cap dolostones are stromatolites (Spencer and Spencer, depleted in 13C relative to other Neo- transgressive and they typically extend 1972; Corkeron and George, 2001), to proterozoic carbonates (Knoll and far beyond the confines of their ante- supratidal tepee beccias (Fig. 5g; Walter, 1992; Kaufman and Knoll, cedent glacial deposits, blanketing Deynoux and Trompette, 1976). 1995; Kaufman et al., 1997; Hoffman 2002 Blackwell Science Ltd 141
The snowball Earth hypothesis • P. F. Hoffman and D. P. Schrag Terra Nova, Vol 14, No. 3, 129–155 ............................................................................................................................................................. et al., 1998b; Kennedy et al., 1998; (1998a) suggested that that the snow- 1998a; Corkeron and George, 2001; Myrow and Kaufman, 1999; Brasier ball event occupied the vast majority James et al., 2001). In Namibia and and Shields, 2000; Walter et al., 2000). of this time, when the average sedi- the North American Cordillera, stro- They share certain features with Ar- mentation rate on the platform was matolites in cap dolostones of the chean and some pre-Neoproterozoic very low. By the end of the event, Ghaub, Numees (and correlative Bil- carbonates (Grotzinger and Knoll, substantial net subsidence had dah), Wildrose, and Ice Brook glaci- 1995; Grotzinger and James, 2000; occurred, although masked at that ations (Table 1) are distinguished by Sumner, in press), but they are with- time by relative sea-level fall. Upon internal, vertically orientated, tube- out parallel in the Phanerozoic. They deglaciation, sea level rose and chem- like or sheet-like pockets of infilled have long been considered paradox- ical and/or detrital sediments rapidly micritic sediment and cement (Cloud ical because they suggest that an filled in the accommodation space et al., 1974; Hegenberger, 1987, 1993; abrupt transformation from glacial provided, making up for ‘lost time’, Hoffman et al., 1998a, 2002a). The to tropical conditions took place vir- so to speak. High sedimentation rates tube-like variety (Fig. 5b) has been tually everywhere (Spencer, 1971; in cap dolostones and overlying lime- attributed to gas escape (Cloud et al., Schermerhorn, 1974; Williams, 1979; stones are consistent with their low 1974; Hegenberger, 1987; Kennedy Fairchild, 1993). Kirschvink (1992) d13C values (see above) and with et al., 2001a), but close inspection did not foresee cap carbonates as their petrological characteristics (see of well-preserved examples (Fig. 5c) consequential to snowball events – he below). reveals a stromatolitic (microbial) ori- was unaware of their existence (perso- Petrologists, being fascinated by the gin (Hoffman et al., 1998a, 2002a; nal communication, 1999) – but they unusual, have had a field day with cap James et al., 2001). The micritic sedi- are a striking affirmation of the sec- carbonates (Peryt et al., 1990; Aitken, ment gathered in shallow pits or ond testable implication (see above) of 1991; Fairchild, 1993; Grotzinger and gutters on the stromatolitic surface, his theory. In cap carbonates, we see Knoll, 1995; Kennedy, 1996; Hoffman which aggraded with remarkably little the ‘smoke’, if not the ‘gun’, of the et al., 1998b; James et al., 2001; Ken- change in morphology (Fig. 5b). As a snowball Earth. nedy et al., 2001a; Corsetti and Kauf- result, the sediment-filled pits and ‘Cap dolostones’ (Fig. 3) are the man in press; Sumner, in press). The gutters became vertical tubes and transgressive systems tracts (Van Elatina, Ghaub, Ice Brook, Landrigan sheets, respectively. The pitted stro- Wagoner et al., 1988) of depositional and Numees glacial deposits (Table 1) matolite surfaces must have resembled sequences (‘cap-carbonate sequences’) are directly overlain by pale, flinty, giant golf balls (so designed to reduce associated with post-glacial sea-level very pure, cap dolostones (Fig. 3), drag). Cap dolostones have much rise. They typically grade across a generally 3–30 m thick (Aitken, 1991; more to reveal (Hoffman et al., marine flooding surface (Fig. 3) into Kennedy, 1996; James et al., 2001). 2002b), but the whole ensemble of deeper water limestone or shale, which They are invariably well laminated structures suggests rapid aggradation gradually shoals upward through car- and, at proper palaeodepths, hum- on storm-dominated ramps. bonate or siliciclastic lutites and are- mocky cross-laminated with laminae Locally, cap dolostones are overlain nites to an exposure surface 100s of defined by sets of reverse-graded by limestone buildups composed lar- metres above the base of the sequence peloids (Fig. 4b) (Aitken, 1991; Ken- gely of sea-floor cements (Peryt et al., (Figs 4a and 9). The great thickness nedy, 1996; James et al., 2001). Struc- 1990; Aitken, 1991; Grotzinger and and deep-water aspect of many cap- tures referred to as ‘tepees’ (Fig. 4c) Knoll, 1995; Kennedy, 1996; Soffer, carbonate sequences (von der Borch occur in all the above but they have 1998; James et al., 2001). The cements et al., 1988; Narbonne and Aitken, neither the geometry nor associations consist of upward-fanning, prismatic 1995; Dalrymple and Narbonne, 1996; of conventional supratidal tepees crystals pseudomorphosed by calcite Hoffman et al., 1998b) is anomalous (Kendall and Warren, 1987). In plan, (Figs 4f and 5e). The square-tipped, within their overall stratigraphic con- they are linear and parallel, orientated pseudo-hexagonal habit of the crystals text (Fig. 9). This is commonly subparallel to the regional palaeoslope indicates that the primary phase was ascribed to post-glacial sea-level rise, (Williams, 1979; Eisbacher, 1981; Ait- aragonite. Although individual crystal but long-term tectonic subsidence is ken, 1991; Kennedy, 1996; James fans are small (< 1 cm), they aggre- required to preserve the sequence et al., 2001), not polygonal like con- gate into reef-like masses (pseudo- from isostatically driven erosion, be- ventional tepees (Fig. 5g), which re- stromatolites) that range from metres cause the glacio-eustatic cycle alone sult from lateral expansion due to the to decametres in scale (Fig. 5a,d). The creates no permanent accommodation force of crystallization (Kendall and cements did not precipitate slowly in space. On the Otavi carbonate plat- Warren, 1987). Inspection of their the absence of particulate sediment, form in northern Namibia, an estima- cuspate axial zones (Fig. 4d) reveals but grew rapidly in a race against ted background tectonic subsidence that they formed by continuous micritic burial (Figs 4f and 5d,e). The rate equivalent to 0.065 mm yr)1 aggradation in a highly oscillatory cementstone buildups are up to 50 m of isostatically adjusted carbonate flow regime. Microbial bioherms thick, but eventually they give way to (Halverson et al., 2002) implies a (Fig. 5a–c) and biostromes (stromat- flaggy, allodapic, pink limestone with timescale on the order of 5 million olites) occur within some cap dolo- large-scale hummocky cross-bedding years to create the accommodation stones (Spencer and Spencer, 1972; (Maieberg Formation, Namibia), or space occupied by the younger cap Cloud et al., 1974; Wright et al., 1978; to black, organic-rich shale (Sheepbed carbonate sequence (Fig. 9), which is Walter and Bauld, 1983; Bertrand- Formation, Canada). The cements 300–400 m thick. Hoffman et al. Sarfati et al., 1997; Hoffman et al., imply a high degree of carbonate 142 2002 Blackwell Science Ltd
You can also read