Trace element distribution in Neoproterozoic carbonates as palaeoenvironmental indicator

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright

Author's personal copy Chemical Geology 258 (2009) 338–353

Contents lists available at ScienceDirect

Chemical Geology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c h e m g e o

Trace element distribution in Neoproterozoic carbonates as palaeoenvironmental indicator Hartwig E. Frimmel ⁎ Department of Geological Sciences, University of Cape Town, Rondebosch 7701, South Africa

a r t i c l e

i n f o

Article history: Received 28 May 2008 Received in revised form 17 October 2008 Accepted 25 October 2008 Editor: D. Rickard Keywords: Rare earth elements Carbonate Neoproterozoic Geochemistry Southern Africa

a b s t r a c t A first study of REE + Y distribution in a variety of Neoproterozoic (Cryogenian and Ediacaran) carbonates from different settings in the Saldania, Gariep, Damara and West Congo Belts in southwestern and central Africa revealed systematic differences that can be explained by varying palaeoenvironmental factors. The majority of samples display relatively unfractionated, flat shale-normalised REE + Y patterns that cannot be ascribed solely to shale contamination but are interpreted as resulting from the incorporation of near-shore colloids, possibly related to Fe-oxihydroxide scavenging. Only few carbonate units yielded trace element distributions that conform to a typical seawater composition. Those carbonates that were affected by stratiform, synsedimentary hydrothermal mineralisation are distinguished by Eu anomalies. Considering the similarity in residence time between REE and carbon, the strong influence of river-born particles on the REE + Y distribution in the analysed carbonates casts considerable doubt over the usefulness of these carbonates for stratigraphic correlation of Neoproterozoic sediment successions based on carbon isotopes. © 2008 Elsevier B.V. All rights reserved.

1. Introduction A number of recent studies have shown that chemical sedimentary rocks, such as some carbonates, banded iron formation/cherts and phosphates, can serve as useful proxies for the record of certain trace element patterns in the water from which these rocks originate. Systematic differences in the properties of the lanthanide series (REE) and Y make it possible to use them for differentiating between different types of mineral-precipitating waters (e.g. Bolhar et al., 2004; Nothdurft et al., 2004; Bolhar and Van Kranendonk, 2007). The distribution of these elements is very sensitive to water depth, salinity and oxygen level. At the same time the REE + Y distribution monitors also differences in the input sources, mainly the ratio of continental input via rivers and airborne dust versus oceanic hydrothermal input. Marine chemical sediments typically reflect a seawater REE + Y distribution that appears to be independent of age (e.g., Shields and Webb, 2004; Bolhar and Van Kranendonk, 2007). They are characterised by a uniform light REE depletion, enrichment in La, depletion in Ce, slight enrichment in Gd and marked positive Y anomaly in shalenormalised diagrams (Zhang and Nozaki, 1996). The extent of the Ce deficiency is related to oxygen level, with oxidised Ce4+ being less soluble and thus more readily adsorbed onto particles (De Baar et al., 1991; Möller et al., 1994; Alibo and Nozaki, 1999). Not surprisingly, the ⁎ Geodynamics & Geomaterials Research Division, Institute of Geography, University of Würzburg, Am Hubland, D-97074 Würzburg, Germany. Tel.: +49 9318885420; fax: +49 931 8884620. E-mail address: [email protected]. 0009-2541/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chemgeo.2008.10.033

negative Ce anomaly is absent in carbonates and banded iron formation of Archaean to earliest Proterozoic age when seawater was not sufficiently oxidising to form CeIV (Bau and Dulski, 1996; Kamber and Webb, 2001). Very different REE + Y patterns have been obtained from reducing, acidic hydrothermal fluids, which typically display a distinct positive Eu anomaly in otherwise uniform, light to middle REEenriched patterns when normalised to a shale composition. This is evident, for example, from analyses of hydrothermal discharge on the seafloor (Michard et al.,1983; Bau and Dulski,1999; Wheat et al., 2002). In contrast, river water is characterised by relatively flat REE + Y patterns with slight uniform light REE depletion and no distinct element anomalies (Goldstein and Jacobsen, 1988; Lawrence et al., 2006; García et al., 2007), which has been used successfully to distinguish between marine and lacustrine carbonates (Bolhar and Van Kranendonk, 2007). The overall REE concentration in most waters, including seawater, and their respective precipitates, is generally very low and close to the lower limit of detection of methods traditionally employed to analyse for these elements in the past. The increasing availability of inductively coupled mass spectrometric (ICPMS) methods, which have a sufficiently low detection limit for REE, and the distinct differences in the trace element geochemistry between different types of water and their respective precipitates offer an opportunity to gain new insights into likely ancient palaeoenvironmental conditions through the analyses of REE + Y distributions in chemical sedimentary rocks. The Neoproterozoic Era is famous for dramatic climatic changes that are recorded by glacial and intercalated warm-water deposits. The carbonate successions that typically overlie glaciogenic diamictite

Author's personal copy H.E. Frimmel / Chemical Geology 258 (2009) 338–353

Fig. 1. Locations of the studied carbonate samples (circles) in the various Pan-African orogenic belts in southwestern Africa, shown in an Upper Cretaceaous Gondwana breakup position.

339

deposits (cap carbonates) are, inter alia, critical for the palaeoclimatic interpretation of the Neoproterozoic rock record. One of the underlying assumptions for the palaeoclimate models is that the information obtained from chemical sediments of the post-glacial time slices is indeed representative of the global ocean water chemistry. Our understanding of these carbonate deposits is, however, still very limited, in parts because of severe problems in extrapolating the seawater composition back in time beyond the Phanerozoic Eon, in parts because of poor palaeogeographical and thus palaeoenvironmental control. A fundamental problem in our understanding of Neoproterozoic glaciations lies in the difficulty of correlation. In the absence of precisely datable units in many Neoproterozoic glacial and post-glacial successions, C isotope chemostratigraphy has become a popular method to correlate stratigraphic units on both a regional and global scale (e.g. Halverson et al., 2005). Microbial carbonates have been shown to yield some of the most reliable proxy data for the REE + Y distributions in ancient seawater (e.g. Kamber and Webb, 2001). Such microbialites are abundant in the various Neoproterozoic carbonate successions and they form, therefore, the main focus of this study. Here results are reported of trace element, including REE, distributions in carbonate samples from the Pan-African Gariep, Saldania, Damara and West Congo Belts in southwestern and central Africa (Fig. 1) in order to test the usefulness of these trace elements for better assessing the palaeoenvironmental conditions during carbonate formation. The studied examples of carbonate rocks were selected in such a way as to cover a variety of

Fig. 2. Stratigraphic correlation of the studied carbonate units across the Pan-African belts of southwestern and central Africa; source of age data: 1 — Tack et al. (2001), 2 — Jung et al. (2007), 3 — Hoffman et al. (1996), 4 — Frimmel et al. (2001b), 5 — Frimmel et al. (1996b), 6 — Fölling et al. (2000), 7 — Grotzinger et al. (1995), 8 — Barnett et al. (1997), 9 — Naidoo (2008).

Author's personal copy 340

H.E. Frimmel / Chemical Geology 258 (2009) 338–353

stratigraphic positions (i.e. Cryogenian versus Ediacaran) and depositional environments as known from regional geological and previous isotopic studies. Thus the following examples were included for this study: (i) Cryogenian carbonate rocks of the c. 750 Ma Rosh Pinah Formation in the Gariep Belt in southwestern Namibia, a post-glacial carbonate that hosts syn-sedimentary hydrothermal mineralisation; (ii) Ediacaran examples of pre-glacial limestone of the Dabie River Formation and the post-glacial Bloeddrif Member, which displays all the features that are typical of a post-glacial (“Snowball Earth”) cap carbonate in the same belt; (iii) limestones from the Cango Caves Group of the Saldania Belt in southern South Africa, representing the probably Cryogenian Nooitgedagt Member and the younger Kombuis Member, which has been interpreted as a stratigraphic equivalent of the Bloeddrif Member; (iv) carbonates from the Otavi platform in northern Namibia, which include both an older post-glacial carbonate, possibly related to the aftermath of the Sturtian glaciation, and a younger post-glacial succession above a syn-Marinoan (c. 636 Ma) glacial unit; and (v) carbonates from the West Congolian Group in the Democratic Republic of Congo, in which again carbonates above an older and a younger glaciogenic unit can be distinguished. 2. Geological settings The choice of samples investigated was dictated by the desire to obtain data from Neoproterozoic carbonates of different age, different depositional setting (proximal versus distal) and different palaeogeographic position. Thus, the analysed samples come from a variety of Neoproterozoic units from a wide area spanning from the southern tip of Africa to Central Africa (Figs. 1, 2). Although some of the details in the correlations shown in Fig. 2 may be contentious, both the Cryogenian and Ediacaran Periods are covered by the chosen samples. 2.1. Cryogenian carbonates in the Gariep Belt The Rosh Pinah Formation forms part of the Hilda Subgroup in the Port Nolloth Group, which constitutes the Neoproterozoic strata in the external part of the Pan-African Gariep Belt, the Port Nolloth Zone, in southwestern Namibia and western South Africa (Fig. 3). This part of the belt, although intensely deformed and metamorphosed at greenschist-facies conditions, still rests on its original Palaeo- to Mesoproterozoic basement. The Neoproterozoic sediment succession starts with continental siliciclastic rift deposits, the Stinkfontein Subgroup (Lekkersing and Vredefontein Formations). First marine sediments appear in the uppermost part of the subgroup as minor intercalated dolostone and limestone. Flooding of the rift shoulders was interrupted during a first glacial period, which resulted in the deposition of the largely diamictitic Kaigas Formation (Fig. 4). Riftrelated, mainly felsic volcanism occurred at the eastern basin margin towards the end of the glaciation. A horst-and-graben structure that developed during rifting was accentuated by a syn-glacial eustatic sea level fall during which an eastern failed graben, the Rosh Pinah Graben, became separated from a western half-graben. The latter eventually evolved into the Gariep Basin proper (Frimmel and Jonasson, 2003). While the volcanic activity led to the local accumulation of a thick volcanic to volcaniclastic succession, the background sedimentation further away from the volcanic centre(s) took the form of relatively thin carbonaceous argillite beds in the less ventilated parts of the Rosh Pinah Basin and of thinly laminated, partly allodapic and variably argilleceous limestone (Pickelhaube Formation) elsewhere. A series of arenitic intercalations, largely arkose and felspathic sandstone, in all of these facies reflect the overall proximity to the palaeoshore line, defined by a basement high whose outline is approximated by the current margin of the orogenic belt. The Rosh Pinah Formation contains, apart from felsic volcanic to volcaniclastic and arenitic rocks, intercalated carbonate beds that are generally only several decimetres thin. Oxygen and carbon isotopic

Fig. 3. Positions of the studied carbonate sample localities for the Rosh Pinah Formation (1), Dabie River Formation (2), and Bloeddrif Member (proximal — 3, distal — 4) in the Port Nolloth Zone of the Gariep Belt.

evidence points to dolomitisation of original limestone by syngenetic to early diagenetic hydrothermal fluids (Frimmel and Lane, 2005) some of which formed massive, broadly synsedimentary, stratiform sulphide ore bodies (Rosh Pinah Pb–Zn–Cu-sulfide deposit and protore of the secondary Scorpion Zn-deposit). The carbonate beds are in places microbialites. Elsewhere large slump breccias occur that indicate deposition in a seismically active region, probably the principal growth fault along the eastern margin of the Rosh Pinah Graben. Deposition of the Rosh Pinah Formation and the largely coeval Pickelhaube Formation was terminated by a hiatus of poorly constrained duration. The overlying stratigraphic unit is the Wallekraal Formation. It consists mainly of conglomerates, dolomite breccias,

Author's personal copy H.E. Frimmel / Chemical Geology 258 (2009) 338–353

341

Fig. 4. Stratigraphic subdivision of the Port Nolloth Group in the Port Nolloth Zone, external Gariep Belt; C isotope profiles from Fölling and Frimmel (2002) and Frimmel and Fölling (2004).

intercalated arenites (immature greywacke and arkose) and argillite. The contact with the underlying strata is a regionally extensive erosional unconformity, with the Wallekraal Formation sediments cutting through the entire older stratigraphy down to the pre-Gariep basement in places (Fig. 4). It is speculated that this erosion surface formed during the global Marinoan glaciation. The maximum age of rift sedimentation is given by the youngest age obtained on a basement rock, a U–Pb single zircon age of 771 ± 6 Ma for the Lekkersing granite (Frimmel et al., 2001b). U–Pb and Pb–Pb single zircon age data of 752 ± 6 (Borg et al., 2003) and 741 ± 6 Ma (Frimmel et al., 1996b), respectively, were obtained on Rosh Pinah Formation felsic volcanic rocks and they provide a minimum age for the Kaigas Formation diamictite. A minimum age for the Pickelhaube and contemporaneous Rosh Pinah Formation carbonate rocks is given by a double-spike Pb–Pb carbonate datum of 728 ± 32 Ma obtained on a limestone bed in the lower Pickelhaube Formation (Fölling et al., 2000), which has been interpreted to date early diagenesis. 2.2. Ediacaran carbonates in the Gariep Belt A younger glaciogenic diamictite (Numees Formation) within the Port Nolloth Zone is both underlain and overlain by carbonates (Fig. 4). The age of this unit is controversial. Correlation with the global Marinoan (c. 636 Ma) glaciation has been suggested (Frimmel et al., 2002) because of strong similarities in the C isotopic record and lithofacies of the bounding carbonate units. Several arguments speak, however, for a younger, Ediacaran age. These include relatively high near-primary Sr isotope ratios consistently between 0.7080 and 0.7085 (Fölling and Frimmel, 2002) and a double-spike Pb–Pb carbonate age of 555 ± 28 Ma for the overlying cap carbonate of the Bloeddrif Member (Fölling et al., 2000). More recently, micropalaeontological evidence provided further support for such an age (Gaucher et al., 2005) and correlation with the approximately 582 Ma Gaskiers (or possibly the slightly younger Moelv) glaciation is therefore favoured. The reef facies carbonates of the Dabie River Formation beneath the Numees Formation diamictite contain a marked positive δ13C anomaly but are progressively depleted in 13C with proximity to the contact with the Numees Formation (Fig. 4). This reversal in the C

isotopic trend towards lower δ13C ratios has been interpreted as signalling a change in global climate in preparation of the Numees glaciation (Fölling and Frimmel, 2002). No significant hiatus is therefore suspected between the Dabie River and Numees Formations. The Dabie River Formation, which attains a maximum thickness of 160 m, is lithologically distinguished from the other carbonatebearing successions of the Hilda Subgroup by the presence of stromatolites displaying Conophyton-like forms, several centimetres to decimetres in height. Pisolites, oolites and oncolites are also present. The formation is almost exclusively calcareous, with original limestone variably dolomitised. The carbonate rocks are typically massive, light to medium grey and, in places, intensely brecciated. Some of the carbonate breccias are interpreted as debris flow deposits, whereas others are ascribed to gravitational slumping. Cyclical emergence and submergence is indicated by desiccation cracks and by the interbedding of limestone and dolostone. A shallow-water, rimmed shelf environment, such as a barrier bar or shelf lagoon, passing seaward (westward) into a shelf margin, comprising reef build-ups and oolitic to pisolitic shoals is envisaged for the depositional environment. Reef rocks formed, or are preserved, particularly in those areas that escaped the pre-Wallekraal erosion. Consequently, the Dabie River Formation carbonates rest in many places paraconformably above dolostone of the Pickelhaube Formation and not necessarily on top of the clastic Wallekraal Formation rocks. The Bloeddrif Member (lower Holgat Formation), which attains a thickness of 100 m but thins out to only a few metres along the eastern margin of the Port Nolloth Zone, is light grey, cream or pale pink in colour and poor in organic matter. It displays the characteristics of a typical cap carbonate, i.e., vertical tube-like structures of infilled micritic sediment and cement. They are usually a few centimetres across and several decimetres high. Similar structures have been described from many other Neoproterozoic post-glacial cap carbonates, including the post-Ghaub Keilberg Member (Maieberg Formation) in the Otavi platform (Hoffman and Halverson, 2008). The Bloeddrif Member consists mainly of clean, thinly laminated limestone, but dolostone is abundant particularly in the basal and proximal sections (Frimmel and Fölling, 2004). The limestone is distinguished by very high Sr contents, reaching several thousand parts per million,

Author's personal copy 342

H.E. Frimmel / Chemical Geology 258 (2009) 338–353

Fig. 5. Simplified geology of the Saldania Belt with location of the carbonate samples from the Cango Caves Group (arrow).

from which an aragonite precursor has been inferred (Fölling and Frimmel, 2002). The calcareous Bloeddrif Member is followed by a siliciclastic metasedimentary succession (upper Holgat Formation) that comprises upwards-fining cycles of medium-bedded sandstone, greywacke and arkose with minor siltstone, mudstone and intraformational conglomerate. These rocks are interpreted to represent turbidites that were laid down in a foredeep (Frimmel and Fölling, 2004). 2.3. Carbonates of the Cango Caves Group in the Saldania Belt Neoproterozoic rocks are exposed in a number of anticlinal erosional windows within the Permotriassic Cape Fold Belt in the southwestern and southern tip of South Africa. The pre-Cape rock successions form the basement of the Palaeozoic Cape Supergroup and were deformed into folds and thrusts in the course of a latest Neoproterozoic to Cambrian accretionary orogeny that led to the formation of the Saldania Belt. Carbonate rocks occur at several localities, but in most cases their stratigraphic position is unclear because of poor exposure of stratigraphic contacts. Reasonable stratigraphic control exists only for carbonates in the Kango Inlier in the southern branch of the Saldania Belt (Fig. 5). There, the Neoproterozoic rocks are grouped together as Cango Caves Group, which is predominantly a carbonate-clastic turbidite succession. The group is subdivided into the Matjies River Formation, Groenefontein Formation and Huis Rivier Formation (Le Roux and Gresse, 1983). Of these, only the lower, former formation contains carbonates in the Nooitgedagt Member and the Kombuis Member (Fig. 2). The Nooitgedagt Member consists of shale, greywacke and limestone that represent a coarsening upward deltaic to shallow marine succession, whereas the latter is a predominantly calcareous shelf deposit. Although both members have been grouped together into the same formation (Le Roux and Gresse, 1983) and considered to represent a continuous sedimentary succession, marked differences in C and Sr isotope chemostratigraphy (Fölling and Frimmel, 2002) and a difference of 100° C in thermal overprint between the Nooitgedagt and Kombuis Members (Frimmel et al., 2001a) point to a major hiatus between the two. A distinct positive δ13C excursion (to a maximum of +10‰) and relatively low 87Sr/86Sr ratios (as low as 0.7074 in Sr-rich limestone) in the Nooitgedagt Member carbonates compare well with similar data for other Cryogenian, post-Kaigas carbonates in the Gariep Belt, whereas consistently lower δ13C (between +2 and −4‰) and significantly higher 87Sr/86Sr ratios from 0.7080 to 0.7087 in the Sr-rich (Rb/Sr b 0.0001) Kombuis Member limestone are comparable to those of the Bloeddrif Member. The latter correlation is further supported by an identical double-spike Pb–Pb carbonate age of 553 ± 30 Ma obtained

for the Kombuis Member limestone (Fölling et al., 2000) and a very similar microfossil assemblage (Gaucher and Germs, 2006).

2.4. Carbonates in the Otavi Platform (Otavi Group) The Otavi Platform in the Northern Foreland of the Damara Orogen in northern Namibia formed along the southern fringe of the Congo Craton and abuts against the continental slope facies further south and west. Thus its predominantly calcareous sedimentary successions, unified as Otavi Group, are located in a foreland position relative to the Kaoko Belt in the west and the Damara Belt in the south (Fig. 1). The Otavi Group is subdivided into three subgroups, which are separated from each other by two glaciogenic diamictite units, the lower Chuos Formation and the upper Ghaub Formation (Fig. 6). First drowning of the continental rift shoulders is evident in carbonates of the lowest subgroup, the Ombombo Subgroup. These overlie and interfinger felsic, rift-related volcanic rocks (Naauwpoort Formation). Several precise U–Pb zircon and titanite ages between 759 and 746 Ma constrain the period of magmatism (Halverson et al., 2005; Hoffman

Fig. 6. Schematic cross-section through the Otavi platform (modified after Hoffman and Halverson, 2008). For sources of the indicated U–Pb single zircon ages see text.

Author's personal copy H.E. Frimmel / Chemical Geology 258 (2009) 338–353

et al., 1996; Jung et al., 2007) with the latter setting a maximum constraint on the timing of the Chuos glaciation. Different stratigraphic schemes have been used for the Abenab Subgroup above the diamictitic Chuos Formation in the west (Eastern Kaoko Zone) and the east of the platform (Otavi Mountainland). As the samples for this study come from the latter region, only the stratigraphic subdivision as used for the Otavi Mountainland will be summarised here. Dark grey to black, laminated dolostone or limestone rhythmite that grades up from abiotic laminate, not more than 15 m in thickness, into sublittoral microbialaminite follows above the Chuos Formation. It forms the Berg Aukas Formation. A first generation of largely stratiform Pb–Zn sulphide mineralisation affected the formation and evidences late-rift hydrothermal activity in the basin (Pirajno and Joubert, 1993; Frimmel et al., 1996a). Overlying massive to bedded dolostone constitutes the Gauss Formation, which is followed, in turn, by stromatolite and oolite, alternating with thin-bedded limestone and argillite beds of the Auros Formation. Fault-controlled rapid changes in thickness mark the latter two formations, but the original basin configuration at that stage is not constrained in sufficient detail. Lithology and carbon isotopes presage the impending glaciation and the accompanying sea-level fall that eventually resulted in the younger glacial unit, the Ghaub Formation. The δ13C ratios range from +1 to +8‰ through most of the Abenab Subgroup but decrease to −6‰ in the uppermost part (Halverson et al., 2005). The predominantly diamictitic Ghaub Formation reaches as much as 2000 m in thickness in the western Otavi Mountainland but is laterally discontinuous. Based on a proposed correlation with a diamictite in the central Damara Belt for which Hoffmann et al. (2004) obtained a precise U–Pb zircon age of 635.5 ± 1 Ma for an intercalated ash bed, a syn-Marinoan age has been postulated for the Ghaub Formation. The cap carbonate to the Ghaub Formation glaciogenic rocks are represented by the tan to pinkish or pale grey, up to 40 m thick Keilberg Member of the Maieberg Formation (Hoffman and Halverson, 2008; Hoffmann and Prave, 1996). In the most instructive exposures, the basal meter is composed of recrystallised and cemented dolomite siltite or grainstone laminated by low-angle, metre-scale cross-beds. Above this, narrow, vertical, convex-up, stromatolitic columns about 2 to 5 cm across and with a similar spacing are characteristically developed in many places. The microbial lamination of the columns is usually poorly visible. A laminated, concave-up dolomicrite infill between the columns is generally far more apparent than the stromatolitic columns and appears as evenly spaced ‘tubes’ in outcrop. The stromatolites are overlain by a zone of giant wave ripples in peloidal dolomite with individual ripple sets reaching 150 cm in thickness and crestal spacing of 1 to 1.5 m (Allen and Hoffman, 2005). Locally, vertically standing crystal fans of calcite pseudomorphs after sea-floor aragonite are developed just above the Keilberg Member. Overall, the Maieberg Formation, which reaches a thickness of 1800 m in the Otavi Mountainland, forms a single, thick and extensive depositional sequence, initially transgressive and then gradually upward shallowing. The Elandshoek Formation (up to 1500 m thick) above the Maieberg Formation is made up of cherty, grainstone-dominated, dolomite. Several metres thick depositional cycles contain microbialaminites and are terminated by flooding surfaces. Columnar stromatolites are common in the upper half of the formation, which is better bedded. Bedding-parallel silicification is a common feature. The overlying Hüttenberg Formation begins with up to 1000 m thick, light to medium grey, bedded, in places stromatolitic dolomite with numerous chert layers. This is followed by 290 m thick light and dark grey grainstone and mudstone beds with black carbonaceous limestone intercalations and black chert layers. Above that follow massive to bedded and cyclically graded dolomitic grainstone and mudstone layers with interbedded silicified oolite beds and columnar stromatolites, altogether reaching 300 m in thickness. Evidence of a restricted basin with elevated evaporation rates exists near Tsumeb.

343

There stratiform anhydrite and gypsum were intersected in exploration drill holes through the upper Hüttenberg Formation. Furthermore, a distinct stratiform breccia zone, the so-called North Break Zone, in the Tsumeb mine has been re-interpreted as a palaeo-aquifer confined to a former evaporite bed, and fluid inclusion leachate data indicate post-diagenetic circulation of evaporitic residual brines (Chetty and Frimmel, 2000). 2.5. Carbonates of the West Congolian Group The Pan-African West Congo Belt that stretches from southwestern Gabon across western Congo-Brazzaville and the westernmost Democratic Republic of Congo into northwestern Angola contains Neoproterozoic carbonates within the West Congolian Group (Fig. 2). The group is particularly well developed in the foreland basin to the east of the belt where it is only gently folded (Fig. 7) and hardly metamorphosed. For a detailed description of the lithostratigraphy of the group see Cahen (1978). Siliciclastic continental rift deposits make up the Sansikwa Subgroup at the base of the West Congolian Group and are overlain by an older glaciogenic diamictite unit, the Lower Mixtite Formation. This is followed by a varied succession of conglomerate, argillite, calcpelite, quartz arenite, calcarenite and limestone (Haut Shiloango Subgroup) and eventually a second diamictite, the Upper Mixtite Formation. The latter is overlain by a cap carbonate sequence that develops into carbonate ramp and platform deposits with abundant stromatolite bioherms and the filamentous cyanobacterium Obruchevella (Alvarez et al., 1995). This unit is known as the Schisto-Calcaire Subgroup whose sequence stratigraphy has been described in greater detail by Alvarez (1995). Eight different lithofacies have been recognised within the Haut Shiloango Subgroup (Cahen, 1978): Conglomerate and quartzite at the base (Sh1), argillite (Sh2), partly calcareous argillite (Sh3), quartzite and argillaceous limestone (Sh4), and quartz phyllite (Sh5) comprise the Little Bembezi Formation (450 to 650 m in thickness). This is followed by the 200 to 250 m thick Sekelolo Formation, which consists of feldspathic quartzite (Sh6), argillite (Sh7), and a mixed succession of argillite, limestone (partly nodular with intercalated calcarenite), stromatolites and calcareous breccias (Sh8). Similarly, the SchistoCalcaire Subgroup is subdivided into five lithofacies: a post-glacial cap dolomite (C1), calcareous argillite and quartz arenite (C2), and partly oolitic limestone (C3) make up the Kwilu Formation (approximately 500 m in thickness). Lithofacies C3 is characterised by very high Sr concentrations (N5000 ppm; Frimmel et al., 2006). Considering the local abundance of diagenetic albite in this unit, evaporitic anhydrite has been inferred as likely precursor mineral. This is also supported by the presence of diagenetic celestite, polyhalite and clinochlore (Delpomdor, 2007). Above C3 follow limestone and dolostone, partly stromatolitic, with minor intercalated calcpelite and chert of the Lukunga Formation (C4). Lithofacies C5 (Bangu Formation, up to 270 m in thickness) comprises variably dolomitic and partly very carbonaceous dark limestone, abundant oolite, minor stromatolites and chert, with intercalations of calcpelite and talc schist. Obruchevella is present in silicified oolite. For a detailed description of the various stromatolites present in the West Congolian Group carbonates see Bertrand-Sarfati (1972). Near-primary 87Sr/86Sr ratios consistently around 0.70715 were obtained for the Haut Shiloango Subgroup limestone and these support a Cryogenian, possibly post-Sturtian age (Frimmel et al., 2006). High δ13C values of as much as +8‰ obtained in the same study are consistent with such a correlation. In contrast, limestone with very low Rb/Sr (bb0.003) of the C3, C4 and C5 lithofacies in the Schisto-Calcaire Subgroup has near-primary 87Sr/86Sr ratios that are slightly higher (0.70740–0.70753; Frimmel et al., 2006) and that correspond to those found elsewhere in post-Marinoan carbonates (Halverson et al., 2007). The δ13C ratios are negative throughout the C3 limestones with values around −1‰ but are highly variable and erratic, without noticeable trends, in the C4 and

Author's personal copy 344

H.E. Frimmel / Chemical Geology 258 (2009) 338–353

Fig. 7. Distribution of the main stratigraphic units of the central West Congo Belt and SW–NE profile (from Frimmel et al., 2006).

especially the C5 carbonates (range between −3 and +9‰). The latter variability has been explained by differences in the depositional environment, more specifically by temporarily elevated evaporation rates. Consequently, C isotopes appear not suitable for chemostratigraphic correlation of these units. 3. Samples and analytical techniques A total of 148 carbonate samples were selected for geochemical analysis. The rock samples (c. 0.5–2 kg) were reduced in size by a steel press and c. 100 g of alteration-free rock chips were handpicked and pulverised (300–400 mesh). Major element concentrations were measured by conventional X-ray fluorescence (XRF) spectrometry of fusion disks on a Phillips X'Unique II PW1480 spectrometer and trace element concentrations were determined by ICPMS using a Perkin Elmer/Sciex Elan 6000 mass spectrometer at the Department of Geological Sciences, University of Cape Town. For the latter technique, 50 mg of sample powder was first cleaned in ultrapure water and then dissolved in HF and HNO3 and eventually analysed against five-point calibration curves. For the analytical details of these techniques see Frimmel et al. (2001b). Typical lower limits of detection for the trace element concentrations reported here are b0.01 ppm. Contamination by oxides, sulphides or silicates can be a potential problem in the interpretation of the results. This problem was minimised by careful sample selection and handpicking of chips. It should be noted, however, that Nothdurft et al. (2004) tested dissolution of carbonate rocks in acids of variable strength (1 N acetic acid and 15 N HNO3) and they could not observe any significant differences in the REE recovery and patterns. The procedure followed in this study, i.e. rock dissolution in HF, invariably liberated more silicate-bound trace elements than dissolution in HNO3 only or in acetic acid. Dissolution in HF was preferred because it keeps Th and Zr in solution, whereas

HNO3 may release REE from clastic components but not Zr. As will be discussed later, Zr serves as important monitor of contamination and obtaining a correct concentration is thus crucial for the subsequent interpretation. The Rosh Pinah Formation carbonates are all dolomitised microbialaminite. Based on O and C isotopic evidence, the dolomitisation has been related to hydrothermal activity along a growth fault at the margin of a volcanic rift graben, which, in places, led to economic stratiform sulphide mineralisation (Frimmel and Lane, 2005). The analysed samples come from the farm Spitzkop 111, some 10 to 18 km north–northwest of Rosh Pinah. There five dolomite beds, each a few decimetres in thickness, occur at the top of upward-fining cycles and one such bed at the bottom of an upward-coarsening cycle (Frimmel and Lane, 2005). In each cycle, the dolomite is associated with arkose, calcareous sandstone, siltstone and mudstone. Ripple marks are common in the siliciclastic parts of the cycles as well as in tuffite. Close to a hydrothermal breccia in felsic volcanic rocks, representing a hydrothermal vent site, the dolomite is enriched in Fe and Mn (Samples KL01-36 and KL01-38 as opposed to KL01-39 to KL01-41). Magnetite is the principal Fe-phase in all samples except for one (KL01-30), which contains abundant pyrite and is also enriched in a number of minor and trace elements, such as Ti, K, Rb, Cu, Y, Zr, Nb, Cs, and Th. The pyrite-rich sample is explained as reflecting strong hydrothermal contamination by nearby felsic volcanic degassing (Frimmel and Lane, 2005). The nine analysed carbonate samples from the Dabie River Formation come from the Helskloof Pass in the Richtersveld National Park in westernmost South Africa (16.979°E, 28.3124°S). They comprise dolomitic microbialaminite (PF2, 10), syn-sedimentary limestone breccia (PF17, 18), syn-sedimentary dolomitic breccia (PF8), dolomitic stromatolite (PF4, 31, 34), and limestone from the lower part of the formation (PF121). The stromatolite sample PF34

Author's personal copy H.E. Frimmel / Chemical Geology 258 (2009) 338–353

comes from the immediate contact with the overlying Numees Formation diamictite. In order to test the facies-dependence of the REE + Y distribution in carbonates within a given stratigraphic unit, two different profiles through the Bloeddrif Member were considered for this study. One profile (total of 17 samples, PF216–232) represents a proximal section sampled in the southern Richtersveld region at 17.07°E, 28.63°S. The section is characterised by large-scale early diagenetic dolomitisation and few clastic (quartz arenite) intercalations as well as a positive δ13C anomaly (Frimmel and Fölling, 2004). The second profile (total of 17 samples, PF25–30, PF89–102) represents a distal section just south of the Orange River at 16.8550°E, 28.3931°S. It comprises only clean limestone that is devoid of continental detrital input and displays a δ13C trend from negative ratios to values around 0‰ (Fölling and Frimmel, 2002). The carbonate samples from the Kombuis Member, Cango Caves Group, in the Saldania Belt look identical to those of the distal section through the Bloeddrif Member. They are all limestone devoid of significant detrital input and show neither petrographic nor geochemical evidence of significant post-depositional alteration (Fölling and Frimmel, 2002). The samples were collected near the Cango Caves, 21 km north of Outshoorn (22.2157°E, 33.3938°S). For comparison, three limestone samples from the Nooitgedagt Member were also taken from road cuts at Schoemanspoort southeast of the Cango Caves (22.236°E, 33.414°S). The carbonate samples from the Otavi Group were selected on the basis of petrographic and cathodoluminescence studies that revealed different generations of carbonate, especially in the vicinity of epigenetic hydrothermal base metal sulphide mineralisation and later supergene alteration in the course of karstification (Frimmel et al., 1996a; Verran, 1996; Chetty and Frimmel, 2000). They include the following: one dolomicrite sample, representative of the Berg Aukas Formation, from the abandoned Berg Aukas mine at the eastern end of the Otavi Mountainland (18.250°E, 19.516°S); five micritic limestone samples (DV23, 32–35) from the Keilberg Member of the lower Maieberg Formation, a dolomicrite from the middle Maieberg, and a micritic limestone (DV9) as well as a dolomicrite (KH91) from the upper Maieberg Formation, all taken from the Khusib Springs mine (18.022°E, 19.426°S); two dolomicrite samples from the upper Elandshoek Formation (2965-300, DCT2) near the abandoned Tsumeb mine (17.715°E, 19.240°S); and dolomicrite from the Hüttenberg Formation, one sample from the Tsumeb mine and another from the Kombat mine (17.702°E, 19.710°S). Carbonates from both the Haut Shiloango and the Schisto-Calcaire Subgroups of the West Congolian Group were selected for this study. All samples come from the central portion of the foreland to the east of the West Congo Belt in the Bas Congo province of the Democratic Republic of Congo (Frimmel et al., 2006). Eleven samples from the Haut Shiloango Subgroup comprise largely limestone. In addition, three calcpelite samples were included to test the influence of shale contamination. The different lithofacies of the Schisto-Calcaire Subgroup are represented by three dolomitic limestone samples from the C1, eight limestone samples from the C3, 21 variably dolomitised limestone samples from the C4, and 18 limestone, dolomitic limestone and calcpelite samples from the C5 lithofacies.

345

rich in REE, typically have distinctly different REE + Y patterns. Zirconium was used as monitor element to assess the extent of shale contamination, because Zr is abundant in shale (210 ppm in PAAS) but effectively absent in low-temperature waters. Similarly, Al concentrations were used as further proxy for small amounts of shale contamination. This was possible because of the analytical procedures employed that ensured that all clays present in the samples were also dissolved. Shale-normalised (SN) elemental anomalies were calculated on a linear scale, assuming that differences in concentration between neighbouring pairs are constant, as follows: La/La⁎ = La/(3Pr − 2Nd), Ce/ Ce⁎ = Ce/(2Pr − Nd), Eu/Eu⁎ = Eu/(0.67Sm + 0.33 Tb), and Gd/Gd⁎ = Gd/ (2Tb − Dy). As an alternative, following Lawrence et al. (2006), the anomalies were also calculated from a geometric average, assuming that the ratio between near neighbour concentrations is constant, as follows: La⁎ = Pr ⁎ (Pr/Nd)2, Ce⁎ = Pr2 ⁎ Nd, Eu⁎ = (Sm2 ⁎ Tb)0.33, Gd⁎ = (Tb2 ⁎ Sm)0.33, and Lu⁎ = Yb2/Tm. In most cases, the differences in the results obtained by these two methods are minor (b5%) or negligible. If not indicated otherwise, the elemental anomalies are reported as obtained using the linear method. 4.1. Carbonates from the Gariep and Saldania Belts The dolomitic rocks of the Rosh Pinah Formation are overall relatively rich in REE with two samples even approaching PAAS values. The REE + Y distribution deviates, however, from that of PAAS by displaying a marked relative enrichment in the middle REE (NdSN/ DySN = 0.56 ± 0.19; Fig. 8). Two samples that were taken from the immediate contact with a Fe- and Mn-oxide-rich hydrothermal breccia in underlying felsic volcanic rocks (KL01-36 and KL01-38) show higher total REE contents and a distinct negative Eu anomaly ((Eu/Eu⁎)SN = 0.51 − 0.72) compared to those samples taken from greater distance (Fig. 8a). Some of the latter have even a positive Eu anomaly with the highest (Eu/Eu⁎)SN ratios (1.7) achieved in pyrite-

4. Results Trace element concentrations are reported in Tables A1–A6. REE + Y concentrations were normalised to the Post-Archaean Australian Shale (PAAS) composite (Taylor and McLennan, 1985). Overall, the total REE contents in clean carbonate samples are expectedly very low. Elevated REE contents could be the result of contamination with oxides, sulphides, phosphates or silicates, derived mainly either from hydrothermal input and/or from the presence of terrestrial particulate matter (shale). Fortunately, these contaminants, though potentially

Fig. 8. Shale-normalised REE + Y patterns of Rosh Pinah Formation carbonates. (a) Profile away from a hydrothermal vent; (b) variably contaminated and almost uncontaminated dolomitised microbial limestone.

Author's personal copy 346

H.E. Frimmel / Chemical Geology 258 (2009) 338–353

bearing carbonates (e.g. KL01-23). Although this could be an analytical artefact of overlap with the peak for BaO, the noted positive Eu anomalies are regarded as real signals because of a lack of a systematic positive correlation between measured Ba and Eu signals (Table A1). Except for two samples, no Y anomaly can be observed in the analysed carbonates. All other carbonates from the Port Nolloth and the Cango Caves Groups have REE + Y concentrations that are one to two orders of magnitude lower than those of PAAS. The carbonate samples of the Dabie River Formation display relatively uniform shale-normalised REE + Y patterns (Fig. 9a) with a slight light REE depletion (mean (Nd/Yb)SN = 0.79). The limestone sample (PF121) is distinguished by the highest REE + Y concentrations. Otherwise there are no systematic differences in the trace element distribution between the various carbonate types analysed (limestone, dolomitic microbialaminite, stromatolite and syn-sedimentary breccia). Most samples of each lithotype display a slight positive Y anomaly ((Y/Ho)SN ≤ 1.38), whereas others have Y/Ho ratios that are close to the PAAS value of 27.3. One stromatolite and one dolomitic breccia sample show a slight positive Eu anomaly ((Eu/Eu⁎)SN = 1.5 and 1.7, respectively), and most samples also feature a positive Gd anomaly (mean (Gd/Gd⁎)SN = 1.14).

Fig. 10. Shale-normalised REE+Y patterns of carbonates from (a) the Kombuis and Nooitgedagt Members (Matjies River Formation, Cango Caves Group) and (b) the Otavi Group.

Fig. 9. Shale-normalised REE+ Y patterns of (a) Dabie River Formation carbonates, (b) carbonates of the Bloeddrif Member, lower Holgat Formation, in a proximal section south of the Kuboos Pluton, and (c) Bloeddrif Member carbonates from a distal section north of the Kuboos Pluton.

The two profiles through the Bloeddrif Member carbonates yielded very consistent results, but significant differences can be noted between the proximal section (Fig. 9b) and the distal section (Fig. 9c), i.e., between dolomitic and limestone samples. The only limestone sample in the proximal section (PF229) has the lowest REE + Y concentrations and compares well with the limestone samples from the distal section. The dolomitic samples from the proximal section display only very mild light REE depletion and weak positive Gd and Y anomalies in most samples (mean (Gd/Gd⁎)SN = 1.1, mean (Y/Ho)SN = 1.1). In contrast, the limestone samples from the distal section show a distinct positive Y anomaly with (Y/Ho)SN consistently around 1.7. They are further characterised by a significant light REE depletion ((Nd/Yb)SN = 0.41 ± 0.09). The proximal carbonates lack a La anomaly but the distal carbonates show a positive La anomaly with a mean (La/La⁎)SN of 1.80 (1.47 when using the geometric, log-linear method). A weak positive Ce anomaly is also common to the limestone samples ((Ce/Ce⁎)SN = 1.11 ± 0.10). Very low REE concentrations in some samples prevented a complete set of analyses to be obtained for the middle and heavy REE and thus the existence of a Gd anomaly could not be verified for all samples. Most samples, however, display a weak positive Gd anomaly ((Gd/Gd⁎)SN = 1.14 ± 0.05). The REE + Y patterns obtained for the Kombuis Member limestone samples show all the same characteristics as those for the distal Bloeddrif Member limestone samples (Fig.10a). The REE concentrations are so low that they are, for some samples, below the lower limit of detection for some middle and heavy REE. Overall, a clear trend towards light REE depletion is evident ((Nd/Yb)SN = 0.46 ± 0.20), as well as a distinct positive Y anomaly ((Y/Ho)SN = 1.65± 0.11). Similarly, a significant positive La anomaly ((La/La⁎)SN = 1.57 or 1.36, depending on the calculation method) and positive Gd anomaly ((Gd/Gd⁎)SN = 1.22± 0.06) are observed, whereas no significant Ce anomaly is noted ((Ce/ Ce⁎)SN = 1.07 ± 0.03). The three samples from the Nooitgedagt Member (PF200, 201, 203) show very similar patterns but a smaller La anomaly (Fig. 10a). A distinct positive Ce anomaly ((Ce/Ce⁎)SN = 1.32 ± 0.14) distinguishes them, however, from the Kombuis Member limestones.

Author's personal copy H.E. Frimmel / Chemical Geology 258 (2009) 338–353

347

4.2. Otavi Group carbonates All analysed carbonate samples from the Otavi Group have one to two orders of magnitude less REE + Y compared to PAAS, with the limestone samples generally yielding higher concentrations than the dolomitic samples (Fig. 10b). The Berg Aukas Formation dolomicrite shows light REE depletion ((Nd/Yb)SN = 0.37) and a marked positive Eu anomaly ((Eu/ Eu⁎)SN = 1.84). All the Maieberg Formation carbonates, both limestone and dolostone, independent of stratigraphic position within the formation, display remarkably uniform REE + Y patterns without any significant element anomalies. An exception is one dolomicrite that shows a slight positive Y anomaly. The light REE depletion is less pronounced than in the other carbonates ((Nd/Yb)SN = 0.63 − 1.09). Similarly, the Elandshoek Formation dolomicrite yielded relatively uniform REE + Y patterns but with a stronger light REE depletion ((Nd/ Yb)SN = 0.24 − 0.74). In contrast, the Hüttenberg Formation dolomicrite shows a distinct positive Y anomaly ((Y/Ho)SN = 1.40− 1.47) but otherwise a pattern similar to the other samples. Noteworthy is a general absence of a Ce anomaly in any of the Otavi Group carbonates. 4.3. West Congolian Group carbonates The limestone samples of the Haut Shiloango Subgroup show very consistent, uniform REE+Y patterns (Fig.11a), independent of total REE+Y content. No distinct element anomalies are present. Light REE depletion, omnipresent in almost all other carbonates analysed in this study, is only very weakly developed ((Nd/Yb)SN =0.82±0.07). All but two samples lack a positive Y anomaly. To assess the possible influence of shale contamination on the overall REE+Y patterns, three calcpelite samples were investigated as well and they yielded indistinguishable patterns but overall higher REE+Y concentrations that are close to PAAS values. Similarly, the REE + Y patterns obtained for the dolomitic limestone of the C1 and the limestone of the C3 lithofacies within the Schisto-Calcaire Subgroup are very uniform and lack significant element anomalies, except for one limestone sample that displays a weak positive Yanomaly (Fig. 11b). The C1 dolomitic limestones have overall higher (but still an order of magnitude less than PAAS) REE + Y contents but show no significant difference in the distribution of these trace elements. Amongst the samples from lithofacies C4, dolomitic limestone has generally lower total REE+Y concentrations than the pure limestone; in some cases reaching the lower limit of detection for several elements. Most of the limestone samples, which still have one to two orders of magnitude less total REE contents than PAAS, the shale-normalised REE+Y patterns are flat and uniform (Fig. 11c). In some samples, especially the partly dolomitised ones, an enrichment in light REE (particularly La) can be noted. Two samples yielded a positive Y anomaly ((Y/Ho)SN =2.81), whereas the remainder lacks such an anomaly. The shale-normalised REE + Y patterns obtained for most of the C5 carbonates are again flat and uniform (Fig. 11d) with no element anomalies. Overall, the dolomite-bearing carbonates have higher total REE + Y concentrations than pure limestone but display similar patterns. Three calcpelite samples, for comparison, expectedly contain one to two orders of magnitude more REE + Y (corresponding to PAAS). They show a trace element distribution that is analogous to that of the limestone and variably dolomitised carbonate rocks. In few carbonate samples, a slight positive Eu anomaly is observed and those limestone samples with the lowest total REE content seemingly have a negative Yb anomaly. As the concentrations of the latter are close to the analytical capabilities, this Y anomaly might not be real. 5. Interpretation and discussion 5.1. Contamination and alteration Many of the REE + Y patterns obtained in this study do not conform to typical seawater patterns. Therefore the possibility of contamination

Fig. 11. Shale-normalised REE + Y patterns of carbonates from the West Congolian Group: (a) Haut Shiloango Subgroup, (b) lithofacies C1, (c) lithofacies C4, and (d) lithofacies C5 of the Schisto-Calcaire Subgroup.

needs to be assessed carefully before deducing chemical peculiarities of the precipitating waters and thus depositional environment. The most critical sources of contamination are continent-derived detrital material, notably clay minerals, Fe–Mn-oxides and sulphides. Nothdurft et al. (2004) showed that as little as 2% shale contamination can alter the REE + Y pattern of marine carbonates to such an extent that elemental anomalies become effectively eradicated, resulting in flat, uniform shale-normalised patterns. Useful monitors for the extent of terrestrial particulate matter are the concentrations of elements, such as Zr, Th and Al. They are concentrated in different detrital minerals,

Author's personal copy 348

H.E. Frimmel / Chemical Geology 258 (2009) 338–353

such as zircon and clay minerals. Consequently, a positive correlation is to be expected between REE and Zr, Th as well as Al concentrations. Such a relationship is exemplified by the samples from the West Congolian Belt, which include calcpelites, as depicted in Figs. 12a and b. A very good positive correlation exists also between Al and Th (R2 = 0.99; not shown). Those samples with high REE concentrations are clearly contaminated by a considerable shale component and are thus not used any further for palaeoenvironmental reconstructions. It should be noted, however, that even those carbonate samples that have very low REE contents and less than 2% of the Zr concentration of PAAS (e.g. sample HFWC67) yielded the same overall REE + Y pattern as the others (see Fig. 11). Terrestrial detrital contamination should follow the trends shown in Figs. 12a and b. Plotting all data into a total REE versus Zr space (Fig. 12c) reveals, however, two distinctly different trends. The steep trend on this diagram is defined essentially by the West Congolian samples and reflects contamination by terrestrial material. The other trend, following only a marginal increase in Zr with increasing REE concentration, is defined by the Rosh Pinah Formation carbonates. For the latter, both field relationships and geochemical evidence for a strong hydrothermal influence have been presented previously (Frimmel and Lane, 2005) and the elevated REE contents in some samples can be explained by hydrothermal alteration. This is supported by REE + Y patterns that deviate markedly from PAAS (Fig. 8) and by the fact that those samples that were most intensely affected by hydrothermal fluid–rock interaction display distinct Eu anomalies. They can be either positive or negative, depending on the redox state of the hydrothermal fluid and its total sulphur activity, with positive Eu anomaly observed in pyritebearing carbonate and negative Eu anomaly in Fe-oxide rich domains. These examples illustrate that hydrothermal contamination of carbonates can be readily distinguished from terrestrial-detrital contamination. In the following interpretation of the REE + Y patterns obtained for the various Neoproterozoic carbonates, an upper threshold value of

4 ppm Zr will be applied, which corresponds to about 2% shale contamination. This restriction should ensure that the observed REE+ Y trends can be considered to approximate those of the contemporaneous carbonate-forming water unless the original REE concentration in the carbonate was very low — an effect that may be seen, for example, in some of the West Congolian carbonates (Fig. 11). The majority of the studied samples conform to the above requirement, but for some stratigraphic units, such as the Haut Shiloango Subgroup, only very few samples meet this criterion. All shale-normalised diagrams and values presented in this study are based on the PAAS values. If other shale composites are used as reference, such as the North American Shale Composite (NASC), this would not change the overall trends and critical element anomalies (Alibo and Nozaki, 1999). The latter authors showed that seawater has a positive Ho anomaly when normalised against NASC, but this anomaly is not seen when normalisation is carried out against other reference values, including PAAS. The same was noted for Lu/Lu⁎. Strong positive La, negative Ce and even the weak positive Gd anomaly in seawater seem to be, however, independent of the normalisation (De Baar et al., 1985; Alibo and Nozaki, 1999). For a more detailed discussion of normalisation artefacts in REE + Y patterns see Kamber et al. (2005). Comparison of the various shale-normalised REE + Y patterns obtained in this study reveals that Eu anomalies are independent of the total REE content and thus unlikely to be related to shale contamination. The Eu3+/Eu2+ redox potential in aqueous solutions depends mainly on temperature and to a lesser extent on pressure, pH and REE speciation (e.g., Bau, 1991), which explains the positive Eu anomalies typically found in acidic, reducing hydrothermal fluids. Thus the positive Eu anomalies observed in some samples may be the product of either admixture of hydrothermal fluids or co-precipitation of hydrothermal Fe-sulphide. The carbonates from the Rosh Pinah and Berg Aukas Formations provide good examples of precipitation from hydrothermally influenced waters. Both formations are characterised by syn-sedimentary exhalative base metal sulphide mineralisation and both contain carbonates with distinct positive Eu anomalies. In addition, selective Eu mobilisation by diagenetic fluids may cause Eu anomalies. It is also possible that Eu is preferentially released during weathering if present as Eu2+ in the weathered source (Nozaki et al., 2000), especially if it is rich in feldspar (Kamber et al., 2005). This may be the case in some samples from the Schisto-Calcaire Subgroup, notably the C5 lithofacies, which display positive Eu anomalies. The Eu anomalies do not seem to be an artefact of insufficient correction for BaO+ because no significant correlation exists between Ba and Eu concentrations (Table A6). Consequently, Eu seems to be of little use for assessing the original depositional setting. Many of the analysed carbonates are dolomitic and thus the question arises as to the effect of dolomitisation on the REE + Y patterns. Comparison of dolostone and limestone data from the various units, such as the Dabie River Formation (Fig. 9a), the proximal section through the Bloeddrif Member (Fig. 9b), the Maieberg Formation (Fig. 10b), and the C4 and C5 lithofacies of the SchistoCalcaire Subgroup (Fig. 11c,d), reveals that there are no systematic differences in the REE + Y patterns between dolomitised and nondolomitised samples. No relation can be observed between the degree of dolomitisation and REE abundance, which confirms the observation by Banner et al. (1988), who found that dolomitisation of Mississippian limestones did not significantly affect their REE signatures. 5.2. Marine versus non-marine origin

Fig. 12. Positive correlation between the total REE concentration and respectively Al2O3 (a) and Zr (b) in the carbonate rocks and calcpelites of the West Congolian Group reflects variable shale contamination. (c) Total REE versus Zr concentration plot for all analysed samples shows two trends, one depicting shale contamination (see Fig. 11b), the other hydrothermal contamination (dominated by Rosh Pinah Formation samples).

A conspicuous feature common to the majority of the analysed carbonate rocks is the deviation of their REE + Y distribution from a normal seawater composition. Only the carbonates from the distal section through the Bloeddrif Member, the Kombuis Member, and from the Hüttenberg Formation have positive Y anomalies with Y/Ho ratios (47 ± 3, 45 ± 3, and 39 ± 1, respectively) that are markedly higher than that of the upper continental crust. Estimates of the latter are 27.5

Author's personal copy H.E. Frimmel / Chemical Geology 258 (2009) 338–353

(Taylor and McLennan, 1985) and 26.24 (based on more recent work on sediments from Queensland by Kamber et al., 2005). For comparison, the Y/Ho ratio in open seawater is typically between 60 and 90 but is strongly dependent on salinity (Lawrence et al., 2006). The relatively unfractionated shale-normalised REE + Y patterns displayed by the majority of Neoproterozoic carbonates could be explained by formation in a meteoric environment. A lacustrine setting for most of the analysed carbonates is, however, very unlikely on the basis of field relationships and geochemical/isotopic data. Only in the case of the Rosh Pinah Formation, a lacustrine setting cannot be excluded and would be in agreement with the field evidence of thin carbonate beds in cycles that are dominated by siliciclastic deposits with graded bedding (Alchin et al., 2005). Except for the lack of a redox-controlled Ce anomaly, the Rosh Pinah patterns resemble those obtained for Lake Tanganyika, which is not only one of the largest and deepest lakes in the world but is also affected by sub-lacustrine hydrothermal activity (Barrat et al., 2000). A similar hydrothermal influence into a restricted basin (near-shore) is also suggested for the Berg Aukas Formation, which hosts syn-sedimentary exhalative mineralisation. Although Y and Ho have similar ionic radii, identical charge, and thus similar geochemical behaviour, Ho is removed from seawater twice as fast as Y because of differences in the surface complexation behaviour (Nozaki et al., 1997). This makes the Y/Ho ratio a particularly useful monitor for the differentiation between marine and non-marine deposits (Bau, 1996; Nothdurft et al., 2004). Amongst all those studied stratigraphic units that have Y/Ho close to the upper continental crust value, samples exist that lack a shale-normalised Y anomaly in spite of no shale contamination being recognisable. This lack of a significant Y anomaly is therefore likely to be a primary feature of the precipitating water. Yttrium/Ho ratios in river waters are equal to, or slightly higher than, the PAAS value but well below seawater values, i.e., below 60 (e.g., Lawrence et al., 2006). The slight variability indicates limited fractionation of Y/Ho in the fluvial environment, depending on weathering reactions of individual REE + Y-bearing minerals and the pH of the water (Elderfield et al., 1990; Lawrence et al., 2006). The Y/Ho ratios for almost all analysed carbonate samples, except for the distal Bloeddrif Member, the Kombuis Member, and the Hüttenberg Formation, are within the range given for river waters. A considerable freshwater influence may be, therefore, inferred for the depositional environments of these carbonates. General light REE depletion, marked positive La and negative Ce anomalies as well as a weak positive Gd anomaly are further features typically ascribed to a seawater origin. The Gd anomaly, however, does not seem to be unique to seawater. It has been described also from modern river waters in Australia (Lawrence et al., 2006). Consequently, the weak positive Gd anomalies that are present in almost all of the analysed samples are not considered diagnostic of a specific depositional environment. Those samples with little light REE depletion or even enrichment are characterised by Y/Ho ratios close to PAAS (Fig. 13a). The latter relationship is in agreement with a strong freshwater component because dissolved river load typically displays uniform, largely unfractionated REE + Y patterns with only mild light and heavy REE depletions (Goldstein and Jacobsen, 1988; Lawrence et al., 2006). A positive La anomaly is present in almost all samples, but it is generally much smaller than expected for a normal marine carbonate. Expectedly, those samples with a positive La anomaly also tend to display a stronger Y anomaly, in agreement with a marine origin (Fig. 13b). Contamination, be it hydrothermal or detrital (shale), results in the disappearance of the La anomaly and this is clearly seen in the (La/La⁎)SN versus Zr diagram (Fig. 13c). The same diagram also illustrates, however, that not all of those samples that show very little or no signs of contamination (Zr b1 ppm) display a distinct positive La anomaly. Consequently, the lack of a La anomaly in these samples is taken as a near-

349

Fig. 13. (a) Relationship between positive Y anomaly and light REE depletion, expressed as (Nd/Yb)SN; (b) positive correlation between Y and La anomalies for all samples analysed; (c) Relationship between La anomaly and contamination, expressed in terms of Zr concentration.

primary signal of the depositional environment, which is unlikely to be open marine but may be strongly influenced by freshwater. As oxidised Ce4+ is less soluble and more readily adsorbed onto particles than Ce3+, the extent of Ce depletion reflects the oxygenation state of the water. Not surprisingly, Ce oxidation takes place preferentially at shallow water depths (e.g., Alibo and Nozaki, 1999), and evidence of Ce oxidation in seawater exists for the time since the Palaeoproterozoic, but not from the Archaean (Kamber and Webb, 2001; Kamber et al., 2004; Bolhar and Van Kranendonk, 2007). Cerium oxidation is more sensitive to pH than to oxygen fugacity and favoured by alkaline solutions (Elderfield and Sholkovitz, 1987). For this reason, and for a shift of the Ce4+/Ce3+ redox equilibrium towards higher oxygen fugacity with increasing temperature, negative Ce anomalies

Author's personal copy 350

H.E. Frimmel / Chemical Geology 258 (2009) 338–353

are typically absent in low-pH and/or hydrothermal precipitates. Effectively all of the analysed carbonate samples lack a negative Ce anomaly. This phenomenon is independent of total REE concentration and Zr or Al content and thus seems to be a primary feature that could imply only poorly oxygenated and/or relatively acidic waters. The most extreme samples, a dolostone from the Dabie River Formation (PF10) and a limestone sample from the Nooitgedagt Member (PF201) both of which are characterised by very low total REE contents, light REE depletion and a positive Y anomaly, are distinguished by even a marked positive Ce anomaly ((Ce/Ce⁎)SN = 1.42 and 1.48, respectively). An explanation for this behaviour of Ce may be obtained from the comparison of data from proximal and distal sections through the same stratigraphic unit. 5.3. Near-shore versus open marine environments One of the main findings of this study is the observation that, in general, almost all of the analysed carbonates are richer in trace elements (including REE) than Archaean or Proterozoic carbonates. To find an explanation for this difference, it is advantageous to assess the significance of depositional environment, i.e., the proximity to palaeoshore lines, on the carbonate geochemistry. The data obtained on the two sections through the Bloeddrif Member at the bottom of the Holgat Formation neatly illustrate the differences in trace element behaviour between near-shore and distal environments. At the same time they might explain the non-marine signature observed in the majority of the analysed carbonates. The two studied sections through the same stratigraphic unit, a cap carbonate succession above the glaciogenic Numees Formation diamictite, differ in terms of lithology, chemistry and isotopic composition (Frimmel and Fölling, 2004). One section consists entirely of limestone with high Sr concentrations (1078–2483 ppm) and low Mn/Sr, Fe/Sr and Ca/Sr ratios indicative of only very limited post-depositional alteration. Low SiO2 (0.4–3.1 wt.%), Y (≤2 ppm), Zr (≤16 ppm) and low Rb (≤5 ppm) contents reflect only minimal continental detrital input, which is in agreement with a distal depositional environment. In contrast, the other section is largely dolomitic, has thin sandstone beds intercalated, and the carbonates therein contain an order of magnitude more Y, Zr, and Rb, reflecting a larger continental detrital component in a proximal, near-shore environment. Of particular importance is the observation that the C isotopic composition of the least altered carbonates follows different trends in the two sections (Frimmel and Fölling, 2004; Fig. 4). The bottom of both sections contains a negative δ13C excursion (b − 3‰) but further upwards, recovery of the δ13C ratios to around 0‰ is noted in the distal section, whereas a marked positive δ13C excursion (+5‰) is contained in the proximal section. This enrichment in 13C within the proximal section has been interpreted as evidence of a stronger salinity stratification in a poorly circulated sea (Frimmel and Fölling, 2004). If such a suggested strong palaeo-environmental control on δ13C exists, it would cast doubt on a series of chemostratigraphic correlations of Neoproterozoic strata many of which are typically based on C isotopic anomalies. Since it has been shown that salinity exerts a strong control on the distribution of REE + Y (Lawrence et al., 2006), the data obtained in this study might help in better interpreting the noted difference in δ13C in distal and proximal profiles. The REE + Y patterns obtained for the various limestone samples from the distal section are all essentially identical, irrespective of total REE concentration. Contamination is negligible. The patterns bear features typical of seawater, such as a positive Y anomaly but the La and Ce anomalies are not as pronounced as expected for open marine chemical sediments. The continental detrital influx in the proximal section is also reflected by overall higher REE, Zr, Th, and Al contents as well as a tendency towards uniform, flat shale-normalised REE + Y patterns. Most representative of the original water composition are

those samples with the lowest total REE (about 1% of PAAS values) and these differ from the distal limestone samples by having a less pronounced Y anomaly, lack of a La anomaly, and overall less light REE depletion. This, together with the observed lack of a negative Ce anomaly in all of the proximal carbonates, points to a considerable riverine influx in the near-shore depositional section of the Bloeddrif Member. As the mean oceanic residence time of C is even shorter than that of most REE (exceptions are Ce and possibly Pr and Eu; Alibo and Nozaki, 1999), it is doubtful whether near-coastal carbonate rocks are useful proxies for the C isotopic composition of ancient ocean waters. Similarly as the REE + Y patterns in these rocks do not reflect open marine conditions, their C isotopic composition is unlikely to be representative when and where local changes in the ocean water chemistry occurred, be it due to enhanced evaporation or dilution by river water. This explains the discrepancy in the C isotope profiles through the distal and proximal sections of the Bloeddrif Member. The difference in δ13C, with a positive excursion in the proximal section and the lack thereof in the distal section, is not necessarily just a function of diachronous sediment deposition from the slope up to the shelf, as suggested by Hoffman et al. (2007), but is likely to be controlled by local palaeoenvironmental circumstances. Consequently, the C isotope ratios are not representative of ambient seawater composition and caution must be taken when using C isotope profiles from proximal sections for chemostratigraphic correlation. 5.4. Freshwater versus near-shore colloids Many of the relatively flat shale-normalised REE + Y patterns cannot simply be explained as being contaminated and thus not representative of ambient water chemistry. This is best seen from the wide range in La anomalies at low Zr concentrations (Fig. 13c). Based on the REE + Y distribution alone, carbonate formation in either freshwater or brackish water in an estuary would be possible explanations for the relatively flat, unfractionated REE + Y patterns. As discussed above, a lacustrine origin of the carbonates can be excluded except for maybe the Rosh Pinah Formation carbonates. Strontium isotope ratios may help to distinguish between a coastal fringing and a lacustrine environment. In contrast to C and REE, whose oceanic residence time is on the order of hundreds to a few thousand years (Alibo and Nozaki, 1999), Sr has a very long oceanic residence time of about 2.4 myr (e.g. Jones and Jenkyns, 2001) and is, therefore, a more reliable proxy. Most of the analysed carbonates yielded least altered 87Sr/86Sr ratios that are in the range expected for contemporaneous Cryogenian or Ediacaran seawater (Fölling and Frimmel, 2002; Frimmel et al., 2006). Although this argument may be circumstantial because the chronostratigraphic position has been indirectly inferred in some cases from Sr isotope ratios, the least altered 87Sr/86Sr ratios are in all cases well below values that are expected for river waters, i.e. 0.711 to 0.712 (Palmer and Edmond, 1989). A freshwater source for the carbonates can thus be excluded. As shown by Lawrence and Kamber (2006), marine REE + Y signals form already at very low salinities when mixing river water with seawater. Moreover, mixing with freshwater should lead to a dilution of several other elements whose concentrations are generally elevated in the studied carbonates. Consequently, precipitation from estuarine waters can be also excluded. An alternative and preferred explanation is the incorporation of near-shore colloids, possibly related to Fe-oxihydroxide scavenging. This is maybe best illustrated by the observed lack of a negative Ce anomaly in all of the proximal Bloeddrif Member carbonates. The marine origin of these samples would require the presence of such an anomaly. A highly acidic and reducing depositional environment is unlikely and thus the absence of that anomaly would imply the introduction of some material with a positive shale-normalised Ce

Author's personal copy H.E. Frimmel / Chemical Geology 258 (2009) 338–353

Fig. 14. Total REE concentration versus Fe diagram for the proximal section through the Bloeddrif Member carbonates (square — limestone, diamond — dolostone).

anomaly. Sholkovitz (1992) showed that Fe-colloids in river water make an excellent candidate for such a material and such a derivation seems indicated by a positive correlation between Fe and total REE contents as illustrated in Fig. 14. The analysed samples follow two trends on that figure. One trend passes almost through the origin, which shows that essentially all REE in these carbonate rocks were introduced via Fe-bearing phases. The only limestone sample, which represents the protolith prior to dolomitisation, contains no detectable Fe and only 1.55 ppm total REE. A second trend intersects the xaxis at about 7 ppm REE, which indicates a small variation in Feindependent total REE contents in the dolomitised limestone. The good correlation between Fe and total REE in these samples highlights that riverine Fe-colloids have a strong control over the REE distribution in these carbonates. The same kind of trace element distribution as found in the proximal Bloeddrif Member carbonates seems to be a common feature in most of the analysed carbonates, especially the various microbialaminites and stromatolites (Rosh Pinah Formation, Dabie River Formation, Maieberg Formation, most of the West Congolian carbonates). Many of them represent former microbial reefs. Based on the elevated incorporation of near-shore colloids as inferred from the REE + Y patterns obtained in this study, a coastal fringing reef environment is therefore suggested.

6. Conclusions The majority of a variety of Cryogenian and Ediacaran marine carbonate rocks from the Pan-African Saldania, Gariep, Damara and West Congo Belts do not display a normal marine REE+ Y signature. They show relatively flat, unfractionated shale-normalised REE + Y patterns without the positive La, negative Ce and positive Y anomalies that are typical of marine precipitates. Comparison of samples with variable shale contamination revealed the same patterns also for effectively uncontaminated carbonates. Moreover, comparison of the same calcareous stratigraphic unit, an Ediacaran post-glacial cap carbonate (Bloeddrif Member), in an open marine, distal position and in a nearshore, proximal position illustrated a strong dependence of the REE+ Y distribution on riverine particle influx. This reflects the overall relatively short residence time of these elements in seawater. River-born Fecolloids are recognised as the particles that have the strongest effect on the REE distribution in the mixing zone between pure seawater and river water. This effect is seen most prominently in the proximal Bloeddrif section in the external Gariep Belt, the Maieberg and Elandshoek Formations of the Otavi platform in the northern Damara Belt as well as in most of the West Congolian carbonates. A coastal fringing reef environment is envisaged for these largely microbialaminitic to stromatolitic carbonates. Early diagenetic dolomitisation, which is more widespread in the proximal environments, seems to have little influence on the overall REE + Y pattern.

351

Apart from the distal Bloeddrif Member carbonates, those of the Kombuis Member and the Hüttenberg Formation yielded typical marine REE + Y signatures as well. The identical results obtained for the limestone of both the Bloeddrif and the Kombuis Members confirm the previously suggested correlation of these two units (Fölling and Frimmel, 2002; Gaucher and Germs, 2006). A similar marine signature is observed for the Nooitgedagt Member limestone, but the Y anomaly is less pronounced and Ce displays a positive anomaly. The latter is interpreted, analogous to the above conclusions, as indicative of a near-coastal depositional environment. This finding supports the contention that the Nooitgedagt and Kombuis Members, though traditionally grouped into the same formation, were not deposited in the same environment (Frimmel et al., 2001a; Fölling and Frimmel, 2002). Finally, those carbonates that were affected by syn-sedimentary hydrothermal activity, evident from the presence of stratiform, synsedimentary Fe–Mn-oxide and/or base metal sulphide mineralisation, are distinguished in their REE + Y patterns, particularly by showing a distinct positive or negative Eu anomaly in the cases of hydrothermal sulphide or Fe-oxide precipitation, respectively. This is evident especially in the Rosh Pinah Formation of the external Gariep Belt and the Berg Aukas Formation in the lower Otavi Group in the northeastern Damara Belt. Overall, this study has confirmed that REE + Y distributions in carbonates can be useful monitors of the depositional palaeoenvironment. After using REE + Y to better interpret Archaean (Kamber and Webb, 2001; Bolhar et al., 2004) as well as Phanerozoic carbonates (e.g. Nothdurft et al., 2004), this study is a first attempt to obtain a better understanding of Neoproterozoic carbonates from the distribution of these trace elements. Although the Neoproterozoic seawater composition remains relatively poorly constrained, a distinction seems to be possible with the aid of REE + Y between open marine (distal), near-coastal and possibly lacustrine deposits, as well as hydrothermally influenced deposits. The results obtained emphasise the presence of concentration gradients in REE + Y away from shore in agreement with previously noted concentration gradients in other trace elements. Shen and Boyle (1988), for example, described elevated Ba concentrations in nearshore corals compared to distal, oceanic corals — an enrichment that has been explained by terrestrial run-off (Shen and Sanford, 1990). More recently, Sinclair (2005) pointed out, however, that a series of environmental and/or biological factors can influence the Ba distribution in modern corals. Similarly, it can be expected that future geochemical studies of ancient carbonates will reveal that a number of hitherto poorly understood factors control the trace element distribution in these rocks. Considering that even Ba, with a residence time of approximately 10,000 years, follows a steep concentration gradient in near-shore environments, an even stronger gradient is to be expected for trace elements, such as the REE and Y, whose residence time is, on average, an order of magnitude shorter. While these elements may serve as useful monitors of the local depositional environmental, their use for regional or global chemostratigraphic correlation is very limited. This limitation is even more strongly applicable to C whose residence time is even shorter than that of most REE. Overall the conclusion can be drawn that many Neoproterozoic carbonate successions developed in near-shore environments in which C isotopes cannot be used as reliable chemostratigraphic indicator. Great caution should be applied, therefore, when using δ13C ratios, especially positive excursions, from such succession for inter-basin correlation. Acknowledgements A. Späth and F. Rawoort are thanked for assistance with the ICPMS analyses, D. Reid for XRF analyses. Thanks also go to L. Tack for providing samples from the West Congo Belt. K.-P. Kelber helped with

Author's personal copy 352

H.E. Frimmel / Chemical Geology 258 (2009) 338–353

the drafting of maps. Very constructive comments by an anonymous reviewer are much appreciated. Financial support from the South African National Research Foundation (grant 2069090) is gratefully acknowledged. This is a contribution to IGCP478 (“NeoproterozoicEarly Palaeozoic Events in southwestern Gondwana”) and IGCP512 (“Neoproterozoic Ice Ages”). Appendix A. Supplementary data Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.chemgeo.2008.10.033. References Alchin, D.J., Frimmel, H.E., Jacobs, L.E., 2005. Stratigraphic setting of the metalliferous Rosh Pinah Formation and the Spitzkop and Koivib Suites in the Pan-African Gariep Belt, southwestern Namibia. South African Journal of Geology 108, 19–34. Alibo, D.S., Nozaki, Y., 1999. Rare earth elements in seawater: particle association, shalenormalization, and Ce oxidation. Geochimica et Cosmochimica Acta 63, 363–372. Allen, P.A., Hoffman, P.F., 2005. Extreme winds and waves in the aftermath of a Neoproterozoic glaciation. Nature 433, 123–127. Alvarez, P., 1995. Les facteurs de contrôle de la sédimentation du Supergroupe OuestCongolien (Sud-Congo). Rampe carbonatée et activité biologique au Protérozoique supérieur. Géologie régionale et générale. Documents du BRGM, 239, 1–270. Alvarez, P., Chauvel, J.-J., Van Viet-Lanoë, B., 1995. Obruchevella, cyanobactérie fossile du Protérozoïque supérieur du Congo. Implications sur l'âge du groupe Schisto-calcaire et de la glaciation fini-Protérozoïque. Comptes Rendus Académie Sciences Paris, série IIa, 320, 639–646. Banner, J.L., Hanson, G.N., Meyers, W.J., 1988. Rare earth element and Nd isotopic variations in regionally extensive dolomites from the Burlington-Keokuk Formation (Mississippian): implications for REE mobility during carbonate diagenesis. Journal of Sedimentary Petrology 58, 415–432. Barnett, W., Armstrong, R.A., de Wit, M.J., 1997. Stratigraphy of the upper Neoproterozoic Kango and lower Palaeozoic Table Mountain Groups of the Cape Fold Belt revisited. South African Journal of Geology 100, 237–250. Barrat, J.A., Boulegue, J., Tiercelin, J.J., Lesourd, M., 2000. Strontium isotopes and rareearth element geochemistry of hydrothermal carbonate deposits from Lake Tanganyika, East Africa. Geochimica et Cosmochimica Acta 64, 287–298. Bau, M., 1991. Rare-earth element mobility during hydrothermal and metamorphic fluid–rock interaction and the significance of the oxidation state of europium. Chemical Geology 93, 219–230. Bau, M., 1996. Controls on fractionation of isovalent trace elements in magmatic and aqueous systems: evidence from Y/Ho, Zr/Hf, and lanthanide tetrad effect. Contributions to Mineralogy and Petrology 123, 323–333. Bau, M., Dulski, P., 1996. Distribution of yttrium and rare-earth elements in the Penge and Kuruman iron-formations, Transvaal Supergroup, South Africa. Precambrian Research 79, 37–55. Bau, M., Dulski, P., 1999. Comparing yttrium and rare earths in hydrothermal fluids from the Mid-Atlantic Ridge: implications for Y and REE behaviour during near-vent mixing and for the Y/Ho ratio of Proterozoic seawater. Chemical Geology 155, 77–90. Bertrand-Sarfati, J., 1972. Stromatolites columnaires de certaines formations carbonatées du Précambrien supérieur du Bassin Congolais (Bushimay, Lindien, OuestCongolien). Musée Royal de l'Afrique centrale, Tervuren, Annales, série in-8o, Sciences Géologiques 74, 1–45. Bolhar, R., Kamber, B.S., Moorbath, S., Fedo, C.M., Whitehouse, M.J., 2004. Characterisation of early Archaean chemical sediments by trace element signatures. Earth and Planetary Science Letters 222, 43–60. Bolhar, R., Van Kranendonk, M.J., 2007. A non-marine depositional setting for the northern Fortescue Group, Pilbara Craton, inferred from trace element geochemistry of stromatolitic carbonates. Precambrian Research 155, 229–250. Borg, G., Kärner, K., Buxton, M., Armstrong, R., Van der Merwe, S.W., 2003. Geology of the Skorpion zinc deposit, southern Namibia. Economic Geology 98, 749–771. Cahen, L., 1978. La stratigraphie et al tectonique du Supergroupe Quest-Congolien dans les zones médiane et externe de l'orogène Quest-Congolien (Pan-Africain) au BasZaire et dans les régions voisines. Annals of the Royal Museum for Central Africa, Tervuren, in-8°, Sci. Géol. 83, 1–150. Chetty, D., Frimmel, H.E., 2000. The role of evaporites in the genesis of base metal sulphide mineralisation in the Northern Platform of the Pan-African Damara Belt, Namibia: geochemical and fluid inclusion evidence from carbonate wall rock alteration. Mineralium Deposita 35, 364–376. De Baar, H.J.W., Brewer, P.G., Bacon, M.P., 1985. Anomalies in rare earth distribution in seawater: Gd and Tb. Geochimica et Cosmochimica Acta 49, 1961–1969. De Baar, H.J.W., Schijf, J., Byrne, R.H., 1991. Solution chemistry of the rare earth elements in seawater. European Journal of Solid State Inorganic Chemistry 28, 357–373. Delpomdor, F., 2007. Lithostratigraphie et sédimentologie de la chaîne Quest Congolienne du Néoprotérozoïque supérieur, Bas-Congo, République Démocratique du Congo, Unpubl. MSc thesis, Free University of Brussels, 138 pp. Elderfield, H., Sholkovitz, E.R., 1987. Rare earth elements in the pore waters of reducing nearshore sediments. Earth and Planetary Science Letters 82, 280–288.

Elderfield, H., Upstillgoddard, R., Sholkovitz, E.R., 1990. The rare-earth elements in rivers, estuaries, and coastal seas and their significance to the composition of ocean waters. Geochimica et Cosmochimica Acta 54, 971–991. Fölling, P.G., Frimmel, H.E., 2002. Chemostratigraphic correlation of carbonate successions in the Gariep and Saldania Belts, Namibia and South Africa. Basin Research 14, 69–88. Fölling, P.G., Zartman, R.E., Frimmel, H.E., 2000. A novel approach to double-spike Pb–Pb dating of carbonate rocks: examples from Neoproterozoic sequences in southern Africa. Chemical Geology 171, 97–122. Frimmel, H.E., Fölling, P.G., 2004. Late Vendian closure of the Adamastor Ocean: timing of tectonic inversion and syn-orogenic sedimentation in the Gariep Basin. Gondwana Research 7, 685–699. Frimmel, H.E., Jonasson, I., 2003. The controls on Neoproterozoic base metal sulphide mineralization. In: Eliopoulos, D.G., et al. (Ed.), Proceedings of the 7th Biennial SGA Meeting, 24–28 August 2003, Athens. Mineral Exploration and Sustainable Development, vol. 2. Millpress, Rotterdam, pp. 661–664. Frimmel, H.E., Lane, K., 2005. Geochemistry of carbonate beds in the Neoproterozoic Rosh Pinah Formation, Namibia: implications on depositional setting and hydrothermal ore formation. South African Journal of Geology 108, 5–18. Frimmel, H.E., Deane, J.G., Chadwick, P.J., 1996a. Pan-African tectonism and the genesis of base metal sulfide deposits in the northern foreland of the Damara Orogen, Namibia. In: Sangster, D.F. (Ed.), Carbonate-hosted lead–zinc deposits. Society of Economic Geologists, Littleton, Special Publication, vol. 4, pp. 204–217. Frimmel, H.E., Klötzli, U., Siegfried, P., 1996b. New Pb–Pb single zircon age constraints on the timing of Neoproterozoic glaciation and continental break-up in Namibia. Journal of Geology 104, 459–469. Frimmel, H.E., Fölling, P.G., Diamond, R., 2001a. Metamorphism of the Permo-Triassic Cape Fold Belt and its basement, South Africa. Mineralogy and Petrology 73, 325–346. Frimmel, H.E., Zartman, R.E., Späth, A., 2001b. Dating Neoproterozoic continental breakup in the Richtersveld Igneous Complex, South Africa. Journal of Geology 109, 493–508. Frimmel, H.E., Fölling, P.G., Eriksson, P., 2002. Neoproterozoic tectonic and climatic evolution recorded in the Gariep Belt, Namibia and South Africa. Basin Research 14, 55–67. Frimmel, H.E., Tack, L., Basei, M.S., Nutman, A.P., 2006. Provenance and chemostratigraphy of the Neoproterozoic West Congolian Group in the Democratic Republic of Congo. Journal of African Earth Sciences 46, 221–239. García, M.G., Lecomte, K.L., Pasquini, A.I., Formica, S.M., Depetris, P.J., 2007. Sources of dissolved REE in mountainous streams draining granitic rocks, Sierras Pampeanas (Córdoba, Argentina). Geochimica et Cosmochimica Acta 71, 5355–5368. Gaucher, C., Germs, G.J.B., 2006. Recent advances in South African Neoproterozoic to early Palaeozoic biostratigraphy: correlation of the Cango Caves and Gamtoos Groups and acritarchs of the Sardinia Bay Formation, Saldania Belt. South African Journal of Geology 109, 193–214. Gaucher, C., Frimmel, H.E., Germs, G.J.B., 2005. Organic-walled microfossils and biostratigraphy of the upper Port Nolloth Group (Namibia): implications for the latest Neoproterozoic glaciations. Geological Magazine 142, 539–559. Goldstein, S.J., Jacobsen, S.B., 1988. Rare-earth elements in river waters. Earth and Planetary Science Letters 89, 35–47. Grotzinger, J.P., Bowring, S.A., Saylor, B.Z., Kaufman, A.J., 1995. Biostratigraphic and geochronologic constraints on early animal evolution. Science 270, 598–604. Halverson, G.P., Hoffman, P.F., Schrag, D.P., Maloof, A.C., Rice, A.H.N., 2005. Towards a Neoproterozoic composite carbon isotope record. Geological Society of America, Bulletin 117, 1181–1207. Halverson, G.P., Dudas, F.Ö., Maloof, A.C., Bowring, S.A., 2007. Evolution of the 87Sr/86Sr composition of Neoproterozoic seawater. Palaeogeography Palaeoclimate Palaeoecology 256, 103–129. Hoffman, P.F., Halverson, G.P., 2008. The Otavi Group of the western Northern Platform. The Eastern Kaoko Zone and the western Northern Margin Zone. In: Miller, R.M. (Ed.), The Geology of Namibia. Geological Survey of Namibia, Windhoek, pp. 13–70–13-137. Hoffmann, K.H., Prave, A.R., 1996. A preliminary note on a revised subdivision and regional correlation of the Otavi Group based on glaciogenic diamictites and associated cap dolostones. Communications of the geological survey of Namibia 11, 77–82. Hoffman, P.F., Hawkins, D.P., Isachsen, C.E., Bowring, S.A., 1996. Precise U–Pb zircon ages for early Damaran magmatism in the Summas Mountains and Welwitschia inlier, northern Damara belt, Namibia. Communications of the geological survey of Namibia 11, 47–52. Hoffman, P.F., Halverson, G.P., Domack, E.W., Husson, J.M., Higgins, J.A., Schrag, D.P., 2007. Are basal Ediacaran (635 Ma) post-glacial “cap dolostones” diachronous? Earth and Planetary Science Letters 258, 114–131. Hoffmann, K.-H., Condon, D.J., Bowring, S.A., Crowley, J.L., 2004. A U–Pb zircon date from the Neoproterozoic Ghaub Formation, Namibia: constraints on Marinoan glaciation. Geology 32, 817–820. Jones, C.E., Jenkyns, H.C., 2001. Seawater strontium isotopes, oceanic anoxic events, and seafloor hydrothermal activity in the Jurassic and Cretaceous. American Journal of Science 301, 112–149. Jung, S., Hoffer, E., Hoernes, S., 2007. Neo-Proterozoic rift-related syenites (Northern Damara Belt, Namibia): geochemical and Nd–Sr–Pb–O isotope constraints for mantle sources and petrogenesis. Lithos 96, 415–435. Kamber, B.S., Webb, G.E., 2001. The geochemistry of late Archaean microbial carbonate: implications for ocean chemistry and continental erosion history. Geochimica et Cosmochimica Acta 65, 2509–2525.

Author's personal copy H.E. Frimmel / Chemical Geology 258 (2009) 338–353 Kamber, B.S., Bolhar, R., Webb, G.E., 2004. Geochemistry of late Archaean stromatolites from Zimbabwe: evidence for microbial life in restricted epicontinental seas. Precambrian Research 132, 379–399. Kamber, B.S., Greig, A., Collerson, K.D., 2005. A new estimate for the composition of weathered young upper continental crust from alluvial sediments, Queensland, Australia. Geochimica et Cosmochimica Acta 69, 1041–1058. Lawrence, M.G., Kamber, B.S., 2006. The behaviour of the rare earth elements during estuarine mixing - revisited. Marine Chemistry 100, 147–167. Lawrence, M.G., Greig, A., Collerson, K.D., Kamber, B.S., 2006. Rare earth element and yttrium variability in South East Queensland waterways. Aquatic Geochemistry 12, 39–72. Le Roux, J.P., Gresse, P.G., 1983. The sedimentary-tectonic realm of the Kango Group. In: Söhnge, A.P.G., Hälbich, I.W. (Eds.), Geodynamics of the Cape Fold Belt. Special Publication of the Geological Society of South Africa, vol. 12, pp. 33–46. Michard, A., Albarede, F., Michard, G., Minster, J.F., Charlou, J.L., 1983. Rare-earth elements and uranium in high-temperature solutions from East Pacific Rise hydrothermal vent field (13 Degrees North). Nature 303, 795–797. Möller, P., Dulski, P., Bau, M., 1994. Rare-earth element adsorption in a seawater profile above the East Pacific Rise. Chemie der Erde (Geochemistry) 54, 129–149. Naidoo, T., 2008. Provenance of the Neoproterozoic to early Palaeozoic successions of the Kango Inlier, Saldania Belt, South Africa, University of Johannesburg, Johannesburg. 260 pp. Nothdurft, L.D., Webb, G.E., Kamber, B.S., 2004. Rare earth element geochemistry of Late Devonian reefal carbonates, Canning Basin, Western Australia: confirmation of a seawater REE proxy in ancient limestones. Geochimica et Cosmochimica Acta 68, 263–283. Nozaki, Y., Zhang, J., Amakawa, H., 1997. The fractionation between Y and Ho in the marine environment. Earth and Planetary Science Letters 148, 329–340. Nozaki, Y., Lerche, D., Alibo, D.S., Snidvongs, A., 2000. The estuarine geochemistry of rare earth elements and indium in the Chao Phraya River, Thailand. Geochimica et Cosmochimica Acta 64, 3983–3994. Palmer, M.R., Edmond, J.M., 1989. The strontium isotope budget of the modern ocean. Earth and Planetary Science Letters 92, 11–26.

353

Pirajno, F., Joubert, B.D., 1993. An overview of carbonate-hosted mineral deposits in the Otavi Mountain Land, Namibia: implications for ore genesis. Journal of African Earth Sciences 16, 265–272. Shen, G.T., Boyle, E.A., 1988. Determination of lead, cadmium, and other trace metals in annually-banded corals. Chemical Geology 67, 47–62. Shen, G.T., Sanford, C.L., 1990. Trace-element indicators of climate variability in reefbuilding corals. In: Glynn, P.W. (Ed.), Global Ecological Consequences of the 1982– 83 El Niño-Southern Oscillation. Elsevier, Amsterdam, pp. 255–283. Shields, G.A., Webb, G.E., 2004. Has the REE composition of seawater changed over geological time? Chemical Geology 204, 103–107. Sholkovitz, E.R., 1992. Chemical evolution of rare earth elements: fractionation between colloidal and solution phases of filtered river water. Earth and Planetary Science Letters 114, 77–84. Sinclair, D.J., 2005. Non-river flood barium signals in the skeletons of corals from coastal Queensland, Australia. Earth and Planetary Science Letters 237, 354–369. Tack, L., Wingate, M.T.D., Liégeois, J.-P., Fernandez-Alonso, M., Deblond, A., 2001. Early Neoproterozoic magmatism (1000–910 Ma) of the Zadinian and Mayumbian Groups (Bas-Congo): onset of Rodinia rifting at the western edge of the Congo craton. Precambrian Research 110, 277–306. Taylor, S.R., McLennan, S.M., 1985. The Continental Crust: Its Composition and Evolution. Blackwell, Oxford. 312 pp. Verran, D., 1996. Genesis of the Khusib Springs Cu–Pb–Zn–(Ag) deposit, Otavi Mountainland, Namibia. Unpubl. BSc (Hons) thesis, Dept. of Geological Sciences, University of Cape Town, 36 pp. Wheat, C.G., Mottl, M.J., Rudnicki, M., 2002. Trace element and REE composition of a low-temperature ridge-flank hydrothermal spring. Geochimica et Cosmochimica Acta 66, 3693–3705. Zhang, J., Nozaki, Y., 1996. Rare earth elements and yttrium in seawater: ICP-MS determinations in the East Caroline, Coral Sea, and South Fiji basins of the western South Pacific Ocean. Geochimica et Cosmochimica Acta 60, 4631–4644.