Neoarchaean (c. 2.58 Ga) halite casts: implications for palaeoceanic chemistry

In modern marginal-marine settings, halite accumulates in subaerial and subaqueous hypersaline environments including peritidal flats, and environments not affected by tides or storms but flooded by marine waters that seep through a physical barrier separating the evaporitic basin from the ocean (Handford 1991). In each of these settings, progressive evaporation of seawater leads to precipitation of calcite and gypsum followed by halite (Handford 1991). Halite precipitation takes the form of subaqueous cumulates, or subaqueous bottom or intrasediment precipitates (Lowenstein & Hardie 1985). Later influx of fresh or storm waters results in dissolution of halite such that, in the geological record, halite is rarely preserved as a mineral but rather as casts or moulds on bedding planes (Llewellyn 1968; Southgate 1982; Demicco & Hardie 1994).

In modern settings and in the geological record extending back to 1.8Ga, gypsum is a common evaporite mineral often developed in association with halite. The pre-1.8 Ga record, in contrast, is almost devoid of gypsum. The lack of gypsum is considered to have important implications for the composition of the early ocean (Grotzinger & Kasting 1993).

A number of horizons in the transition beds between the Black Reef Formation and Oaktree Formation of the Chuniespoort Group, Transvaal Supergroup, South Africa (Figs 1 and 2) contain a variety of casts that represent the oldest evidence for halite precipitation described from the geological record. The sedimentology of these and associated facies has been investigated along the eastern escarpment in Mpumulanga Province, South Africa (Fig. 1) with a view to understanding: (1) the depositional environment of halite precipitation; (2) the mode of precipitation of halite; (3) the implications for Neoarchaean ocean chemistry of halite precipitation in rock units that underlie those containing extensive aragonite crystal pseudomorphs. Critical to the interpretation of palaeoceanic chemistry is the unambiguous characterization of former evaporites on the basis of pseudomorph morphologies.

Geological setting

The Transvaal Supergroup is a late Archaean to early Palaeoproterozoic succession of siliciclastic and chemical sedimentary, and subordinate volcanic rocks that are preserved within three separate sub-basins: Transvaal, Kanye and Griqualand West (Catuneanu & Eriksson 1999). Rocks of the Transvaal Supergroup unconformably overlie either the Witwatersrand Supergroup or the Ventersdorp Supergroup and make up the floor rocks of the Bushveld Complex. In Mpumulanga Province, South Africa, the Transvaal Supergroup overlies the protobasinal WoIkberg Group, and is subdivisible into the Black Reef Formation, Chuniespoort Group and Pretoria Group (Figs 1 and 2). The Black Reef Formation is gradational into the Oaktree Formation at the base of the Chuniespoort Group. A latest Archaean age for the Black Reef-Oaktree study interval is constrained by U-Pb dating of tuffs intercalated within the Oaktree Formation NW of Johannesburg (2550 ± 3 Ma and 2588 ± 7 Ma) and in Mpumulanga Province (2583 ± 5Ma) (Walraven & Martini 1995; Martin etal. 1998).

Along the Mpumulanga escarpment, the Black Reef Formation varies in thickness from 10 to 20 m (Henry et al. 1990; Eriksson & Reckzo 1995) and consists of conglomerates, sandstones and mudstones. A lower upward-fining interval is interpreted as a braided-alluvial to floodplain transition, whereas two overlying upward-coarsening intervals are considered to represent progradational braid-delta deposits (Fig. 3; Henry et al. 1990). Braid-delta deposits (sensu McPherson et al. 1987) are confined to the northern and southern parts of the outcrop belt and the halite-bearing units are located in the northern portion of the belt (Fig. 3). The study interval overlies cross-bedded sandstones of braided-alluvial origin and is capped by large stromatolitic domes of the Oaktree Formation that are interpreted as subtidal, marine facies (Truswell & Eriksson 1975; Beukes 1987).

Facies descriptions and interpretations

Sections were measured through halite-bearing intervals in the upper Black Reef Formation and Oaktree Formation at the base of the Chuniespoort Group at three locations along the Mpumulanga escarpment: Dientje-Old Stone Bridge and Ses-I-se-Draai (Fig. 4). In addition, a section was measured through the upper Black Reef Formation and Oaktree Formation at Three Rondawels, where no halite casts were observed.

Pebbly sandstone fades

The lower Black Reef Formation along the escarpment is less than 5 m thick (Fig. 3; Henry et al. 1990) and consists of pebbly and coarse-grained sandstones containing medium- to large-scale trough cross-beds, and small-scale planar cross-beds within which the foresets are defined by grain flows (see Buck 1985). Palaeocurrent data indicate strongly unidirectional flow towards the west and SW (Fig. 5). Pebbly sandstones of the lower Black Reef Formation are sharply overlain by siltstone-shale, stromatolitic dolomite or tuff (Fig. 4). Large-scale, symmetrical ripples with wavelengths up to 20 cm (Fig. 6a) define the top of the pebbly sandstone facies at Ses-I-se-Draai (Fig. 4b).

The lower Black Reef Formation has been interpreted as a mainly braided-alluvial deposit by Henry et al. (1990) on the basis of its coarse grain size and dominance of unidirectional cross-beds. Comparable facies are developed in modern (e.g. Coleman 1969; Cant & Walker 1978), and ancient braided-alluvial deposits including those of Archaean age (e.g. Eriksson 1978; Beukes & Cairncross 1991; Els 1998). The lack of meandering-river deposits in the Black Reef Formation is consistent with an absence of bank stabilization by vegetation in the Neoarchaean landscape (see Schumm 1968). Large symmetrical ripples developed on the top of the lower Black Reef Formation reflect wave reworking associated with initial transgression. Similar ripples have been described from relict Pleistocene sediments that were reworked during the Holocene sea-level rise (Leckie 1988).

Fine- to medium-grained sandstone facies

This facies dominates the Three Rondawels section, where it is over 5 m thick (Fig. 4c). Sandstone is fine- to medium-grained and has horizontal stratification and small-scale trough crossbeds. Bedding planes exhibit a range of sedimentary structures including adhesion warts and ripples (Fig. 6b), desiccation cracks that display evidence for multiple generations of shrinkage and infill (Fig. 6c), raindrop impressions preserved as casts with crater rims (Fig. 6d), aligned, lenticular sandstone dykelets developed on desiccated mudstone polygons (Fig. 6e), and various ripples including symmetrical, asymmetrical, ladderback and interference forms; some ripples contain desiccated mudstone drapes.

The association of sedimentary structures in this facies indicates very shallow water and periodically emergent conditions. Adhesion structures are produced by wind-blown sand adhering to a wet or damp surface (Kocurek & Fielder 1982). Such conditions would also promote the formation and preservation of raindrop impressions. Alternating submergence and emergence of the depositional surface is indicated by the desiccation cracks. The symmetrical and interference ripple forms indicate an environment influenced by waves whereas the ladderback forms support an intertidal setting (see Klein 1985). Suspension settling of mud on ripples during slack-water periods was followed by exposure and desiccation. The origin of the lenticular sandstone dykelets is more problematic. Geometrically they resemble syneraesis cracks (e.g. van Straaten 1954) but such an origin would not explain their preferred alignment. A plausible explanation is that a mud layer cracked subaqueously in response to downslope creep of a semi-consolidated mud layer with the cracks subsequently infilled with sand. A similar origin has been inferred for linear shrinkage cracks in the Green River Formation (Picard 1966).

Siltstone-mudstone fades

Massive and laminated siltstone and mudstone form the upper Black Reef Formation at Ses-I-se-Draai (Fig. 4b). Traction-produced sedimentary structures are notably lacking. This facies is also interbedded with stromatolitic dolomite of the lower Oaktree Formation at the same location. Sedimentary structures are dominated by horizontal laminations, but symmetrical ripples as well as starved ripples of siltstone within mudstone are present locally. Symmetrical ripple crests vary in orientation from north-south to east-west.

Lack of traction-produced structures in this facies in the Black Reef Formation indicates slow suspension sedimentation below wave base. The presence of ripples reflects gradual shoaling into the basal Oaktree Formation.

Silicified mudstone-siltstone facies

This facies is developed in the lower Oaktree Formation at Ses-I-se-Draai and Old Stone Bridge (Fig. 4a and b) and as eight horizons interbedded with sandstone in the upper Black Reef Formation at God's Window (Fig. 1). Individual horizons range in thickness from 2 to 30 cm. Facies are dominated by silicified mudstone in laminae between 2 and 30 mm thick. Mudstone is mainly massive with rare faint parallel laminations. Intercalated within the mudstone are 1 to 4 mm thick, massive, gradedbedded and rippled, lenticular-bedded siltstone laminae (Figs 6f and 7a) that locally infill desiccation cracks and angular depressions in the underlying mudstone (Fig. 7b). Locally, scouring is present at the base of siltstone beds. Chaotic intraclast breccias are common within this facies typically in association with small-scale syndepositional faults and dismembered folds overlain by intact laminations. Other structures developed in this facies include halite casts in extensive pavements up to 5 m by 20 m, rare tepee structures (Fig. 7c), rill marks, and desiccation and prism cracks (Fig. 7a). Halite casts range from <1 cm to 2 cm in size, vary from square- to triangular- to dumbbell- to hopper-shaped, and typically are isolated from one another (Figs 7d and 8a-c). Casts commonly display internal zoning (Fig. 8d) and hopper-like pyramidal hollows on cube faces, and are commonly associated with desiccation cracks.

Evidence for the former presence of sulphates is lacking. Specifically, the following criteria are not evident on outcrop or in thin section: (1) nodules of former anhydrite replaced by quartz, calcite or dolomite; (2) pseudomorphs after swallow-tail gypsum or after anhydrite laths; (3) displacive crystallization associated with non-cubic pseudomorphs (see Demicco & Hardie 1994). Radiating crystal fans in overlying limestones that were previously interpreted as gypsum pseudomorphs (Bertrand-Sarfati 1976) are now considered to be the product of neomorphism of botryoids of aragonite, on the basis of textures and elevated strontium contents (Sumner & Grotzinger 2000).

Associations of lithologies in this facies indicate overall lowenergy conditions, favouring mud accumulation, interrupted by frequent higher-energy pulses that introduced silt to the depositional setting. Massive siltstone laminae probably reflect storm processes whereas ripple cross-laminae record weak traction reworking. Desiccation structures indicate periodic exposure of the depositional surface. The rare tepee structures may have formed as a result of the expansive growth of halite and/or early carbonate cement in the zone of evaporative pumping (see Warren 1983; Lowenstein & Hardie 1985). Chaotic breccias are mainly products of desiccation and/or incipient tepee formation.

Isolated halite casts and the presence of hoppers with concave margins support displacive growth of halite probably as a result of evaporation of capillary brines (see Shearman 1978; Gornitz & Schreiber 1981; Handford 1988). Less common casts with internal zoning support incorporative growth within the sediment (see Handford 1988). Halite formation by upward rather than downward diffusion is favoured by the shallow-water setting implied by the desiccation cracks (Handford 1988). Angular depressions at the base of siltstone laminae represent casts of halite that was dissolved by lower-salinity waters that introduced silt. Casts indicate that halite precipitation occurred at very shallow depths within the sediment or on the sediment surface.

Stromatolitic dolomite fades

Stromatolitic dolomite is the predominant facies in the Oaktree Formation and, in general, the size of Stromatolitic domes increases upwards (Fig. 4). At Ses-I-se-Draai, silicified stromatolite bearing horizons between 5 and 25 cm thick are interbedded with siltstone-mudstone, tuff and carbonate mudstone-grainstone (Fig. 4b). Stromatolites in these horizons consist of linked, low-relief domes, or vertically stacked domes that increase in width and complexity upwards (Fig. 8d). Laminations on a millimetre scale are well preserved by secondary silicification. Stromatolitic domes higher in the Oaktree Formation are up to 5 m wide, display relief of up to 50 cm, and also are characterized by well-defined millimetre-scale laminations. Secondary silicification is lacking. Stromatolites are elongated east-west to SE-NW.

All structures documented in this facies are similar to isopachously laminated stromatolites that consist of encrusting layers of former (high-magnesium?) calcite, and have been described from stratigraphic units higher up in the Chuniespoort Group and from other Archaean and Palaeoproterozoic successions (Grotzinger & James 2000; Pope & Grotzinger 2000). The encrusting layers are interpreted as products of abiotic precipitation of carbonate mud related to progressive oversaturation of seawater as a result of increasing temperature and salinity (Pope & Grotzinger 2000). Radiating crystal pseudomorphs indicative of aragonite precipitation (Sumner & Grotzinger 2000) were not identified but these are commonly destroyed by dolomitization and silicification (C. Schreiber, pers. comm.). Thus, the former presence of aragonite cannot be excluded. The upward increase in size of domical stromatolites (Fig. 4) is considered to reflect progressive deepening.

Mafic tuff

Mafic tuffs are developed in each of the measured sections and range in thickness from 50 cm to 5 m (Fig. 4). Locally, the tuffs contain glass shards up to 1 cm in length. The tuffs are mostly massive but locally display evidence of reworking in the form of horizontal stratification and symmetrical ripples. Halite casts similar in size and structure to those discussed above are developed on the top of the tuff bed at Ses-I-se-Draai.

Tuffaceous horizons are widespread in the lower part of the Chuniespoort Group (Walraven & Martini 1995; Martin et al. 1998) but the location(s) of the explosive volcanic centres is not known. Also unclear is whether the tuffaceous horizons in different parts of the basin reflect a single or multiple explosive events. In the study area, deposition of the tuffs occurred in shallow water, as implied by the traction-produced structures. In addition, the local presence of halite casts supports a shallowwater, evaporitic setting.

Palaeolatitudinal constraints

Because certain rock types such as evaporites (halite, gypsum and anhydrite), carbonates, coals and tillites are climatically sensitive sediments, and tend to be deposited under restricted conditions, they are useful in palaeoclimatic studies. Evaporites occur in the subtropics, where it is dry, and where evaporation exceeds the total of precipitation plus inflow of surface water (Gordon 1975). Carbonates, in particular those of the Bahamian type, occur in equatorial, subtropical, and warm temperate regions, where it is warm and where there is adequate sunlight penetration (Scotese & Barrett 1990).

The distribution of climatically sensitive sedimentary rocks has been used to independently test and verify the palaeolatitudes calculated from palaeomagnetic studies assuming a geocentric axial dipole model for the Earth's magnetic field in the past (Irving & Briden 1962; Opdyke 1962; Briden 1968, 1970). For the Mesozoic and Cenozoic, palaeolatitudinal positions of the major continents are known with great precision through the use of seafloor magnetic anomalies (Ziegler et al. 1983). Detailed palaeomagnetic studies have also allowed for accurate reconstructions of Palaeozoic palaeogeography (Cocks & Torsvik 2002; Torsvik & Cocks 2004), but palaeomagnetic reconstructions for the Precambrian are much less certain, because of the paucity of well-defined ages. For the Mesozoic and Cenozoic, Scotese & Barrett (1990) plotted the latitudinal distribution of known climatically sensitive sedimentary rocks (evaporites, carbonates, coals and tillites) in the form of pole-to-pole histograms. They successfully used these histograms and their associated probability functions to calculate a Palaeozoic Apparent Polar Wander (APW) path for the Gondwana Supercontinent, which is in fairly good agreement with the palaeomagnetically determined APW of Bachtadze & Briden (1990). Scotese & Barrett (1990) showed that climatically sensitive sedimentary rocks can be used to successfully predict the location of palaeolatitudes, and hence of palaeopoles, assuming that the zonal distribution of climate patterns was the same throughout the Phanerozoic eon as it is today. This assumption can be used for most of Precambrian Earth history, except for those periods when the equator-to-pole temperature gradient was very different from now, such as during the global periods of glaciation during the Palaeoproterozoic and Neoproterozoic eras.

The analytical results of Scotese & Barrett (1990), based partly on the data of Parrish et al. (1982), show that evaporites are restricted mainly to latitudes (N or S) of between 5° and 35°, with an occurrence probability of 0.72, and most modern carbonates are restricted to latitudes (N or S) of between 10° and 30°, with an occurrence probability of 0.55. Thus, for the Transvaal Supergroup halite-carbonate association, there is a maximum probability that this pair of climatically sensitive lithologies was formed at between 10° and 30° palaeolatitude at c. 2.58 Ga. This subequatorial palaeolatitude result fills an important gap in our palaeogeographical knowledge of the Transvaal Supergroup, because the only well-constrained palaeomagnetic palaeopole from this sequence is from the much younger 2222 ±13 Ma Ongeluk lavas (Cornell et al. 1996). Evans et al. (1997) inferred an equatorial palaeolatitude (11° ± 5°) for the Ongeluk lavas; these occur in the Postmasburg Group that overlies the Ghaap Group, a correlative of the Chuniespoort Group. The preserved rocks of the Transvaal Supergroup thus appear to have been deposited while the Kaapvaal Craton was situated in low palaeolatitudes, <30° N or S. Because of the absence of palaeomagnetic data, little can be said about the movement of the Kaapvaal Craton between the period of deposition of the Black Reef Quartzite Formation and outpouring of the Ongeluk lavas. However, a 64.5 ± 17.5° palaeolatitude for the 2782 ± 5 Ma Derdepoort basalt (Wingate 1998) suggests a northward migration of the Kaapvaal Craton between 2.8 and 2.2 Ga.

Discussion

Vertical successions of facies (Fig. 4) suggest progressive deepening from alluvial settings in the lower Black Reef Formation to a subtidal environment at the time of deposition of the Oaktree Formation. The presence of stromatolites produced by precipitation from oversaturated seawater supports a marine rather than a lacustrine depositional environment. In the Ses-I-se-Draai location, wave ripples developed above alluvial facies of the lower Black Reef Formation define a ravinement surface and, together with the overlying facies, indicate rapid deepening. The upper Black Reef Formation at this locality consists of a progradational parasequence that was followed by deepening into the basal Oaktree Formation. A similar vertical transition (except for the later deepening phase) from continental sandstone to thin transgressive-phase deposits followed by thicker subtidal to supratidal, regressive-phase deposits is well documented from sabkhas of Abu Dhabi (Kinsman & Park 1976; Wright 1984).

The parasequence at Ses-1-se-Draai provides constraints on the depositional setting of the halite-cast hosting facies and by implication the environment of halite precipitation. Lack of traction-produced structures in the siltstone and mudstone facies at the base of the parasequence implies a sub-wave base environment, whereas evidence for exposure throughout the upper half of the parasequence supports a peritidal setting in which small-scale stromatolitic domes and carbonate muds and sands together with evaporites were deposited. The preponderance of evidence for intrasediment growth and dissolution of evaporites supports a sabkha and, in particular, a saline mudflat setting (see Handford 1991; Demicco & Hardie 1994). In such an environment, evaporite crystals close to the surface commonly dissolve during floods, resulting in crystal moulds, although a decrease in salinity could also be related to influx of marine waters during storms.

The geomorphology of the sabkha setting can be evaluated with reference to the cross-section of the Black Reef Formation (Fig. 3). This section utilized the top of the last coarse-grained facies as the datum. If the cross-section is hung from the base of the first occurrence of large-scale, stromatolitic domes, it would demonstrate that the halite-bearing and associated facies occupy the northern margin of the fan-delta deposits. In a palaeogeographical sense, this depression may have taken the form of an embayment between a fan delta to the south and an undocumented fan delta further north. If so, it is likely that halite-bearing facies may also be present between the two fan deltas shown on the cross-section (Fig. 3), but poor outcrop precludes testing of this model.

The lack of evidence for sulphate minerals in the studied sections has important implications for Neoarchaean palaeoceanic and/or palaeoatmospheric chemistry. Rare gypsum pseudomorphs reported from the Neoarchaean Carawine Dolomite in the Hamersley Basin, Australia (Simonson et al. 1993) represent the oldest evidence of gypsum precipitation. Mesoarchaean barite from the Warrawoona Group in the Pilbara Block of Australia, previously interpreted as a replacement of gypsum (e.g. Buick & Dunlop 1990), is now considered to represent a primary hydrothermal precipitate (Runnegar el al. 2001). Evaporite pseudomorphs from the Onverwacht Group in the Barberton Greenstone Belt, South Africa, represent silicified nahcolite, a sodium bicarbonate (Lowe & Worrell 1999). The virtual absence of gypsum from the early Earth record is attributed to low sulphate concentrations in early Precambrian oceans related to the anoxic state of the atmosphere or to a high bicarbonate-to-carbonate ratio in early Precambrian oceans such that during progressive evaporation calcium would have been exhausted before the gypsum field was reached (Grotzinger & Kasting 1993). The results of this study support the model of Grotzinger & Kasting (1993) but do not resolve the alternative interpretations for the absence of gypsum. The presence of aragonite crystal pseudomorphs that make up as much as 50% of Neoarchaean carbonate successions including the Malmani Dolomite indicates over-saturation of the sea-water with respect to calcium carbonate (Grotzinger & James 2000; Sumner & Grotzinger 2000). However, the lack of any ferric iron pigmentation in the Black Reef alluvial facies (Twist & Cheney 1986) and in older Archaean alluvial facies such as the Moodies Group and Pongola and Witwatersrand Supergroups (Eriksson 1978; Beukes & Cairncross 1991; Els 1998) indicates overall anoxic atmospheric conditions favouring a low sulphate content of the oceans.

Conclusions

(1) Halite cast-bearing beds in the 2.58 Ga upper Black Reef and lower Oaktree formations accumulated in supratidal or sabkha palaeoenvironments.

(2) Halite precipitation occurred by displacive growth within the host sediment.

(3) Evidence for the former presence of sulphates is lacking, thereby supporting previous hypotheses that the Neoarchaean ocean was deficient in sulphate or contained anomalously high bicarbonate contents.

(4) A subequatorial palaeolatitude for the Transvaal Basin at 2.58 Ga is implied by the association of carbonate and halite, and provides a new data point for palaeogeographical reconstructions.

Field work on which this paper is based was funded by National Geographic Society grant 6003-97. We thank C. Schreiber and P. Turner for their insightful reviews, and M. Fowler for his constructive comments.

© 2005 Geological Society Publishing House Provided by ProQuest LLC. All Rights Reserved.