Contrasting microfossil preservation and lake chemistries within the 1200–1000 Ma Torridonian Supergroup of NW Scotland

Abstract Oxygenation of the Proterozoic atmosphere caused the progressive build-up of dissolved sulphate on the continents and in marine environments. However, oxygen levels in the Proterozoic were low enough to allow the early burial of biological material into low redox potential environments where permineralization and the authigenic replacement of organic material, including micro-organisms, occurred by a range of minerals. Consequently, microbial sulphate reduction caused the widespread degradation of organic matter and, where iron was available, the precipitation of pyrite. By contrast, where sulphate levels were low, early preservation by other minerals (e.g. phosphate or silica) could be excellent. We show, using two Proterozoic lake sequences with low and high sulphate chemistries, but with otherwise similar characteristics, that microbial sulphate reduction caused a profound loss of morphological detail and diversity within preserved microfossils. The results could imply that there is a significant bias in the Proterozoic fossil record towards low sulphate environments, which were in reality relatively scarce.

Brasier 2012; Wacey et al. 2014). However, although acid-resistant vesicles (acritarchs) of inferred eukaryote origin are known from 1.8 Ga and became moderately diverse in nearshore habitats by 1.45 Ga (Lamb et al. 2009), overall the mid-Proterozoic record is very limited. Some researchers attribute this to evolutionary stasis in the marine realm during the 'boring billion', when high levels of hydrogen sulphide, which is toxic to eukaryotes, inhibited their evolutionary radiation , and latterly to the concept that major evolutionary innovations may have mostly been taking place on land at this time (Blank 2013;Wellman & Strother 2015).
The patchy fossil record between c. 2000 and 1000 Ma, plus the inability to assign many such fossils to crown groups, markedly influences the extent to which the Proterozoic fossil record may yet be used to calibrate molecular trees (Eme et al. 2014). There is a pressing need to identify the controls on fossil preservation and to learn what inferences we might make when the preservation is less good, so that we are not forced to interpret the lack of evidence as biological absence. Preservation depends on a range of factors, including the depositional environment, mineralogy, diagenesis and structural history. In particular, the local fluid chemistry, which, in turn, controls microbial processing and mineral precipitation, may dictate which environments are represented in the fossil record and could bias our interpretations. An excellent opportunity to test this theory is provided by two semipermanent lake basins preserved within the 1200-1000 Ma Torridonian Supergroup of NW Scotland. Remarkable preservation of organic matter has previously been reported within lake deposits of the c. 1000 Ma Torridon Group (specifically the phosphatic horizons and nodules of the Cailleach Head and Diabaig formations; Strother et al. 2011;Battison & Brasier 2012;Wacey et al. 2014). However, the preservation of organic matter has been reported to be rather poor in mudstones and carbonates of the c. 1200 Ma Stoer Group lake deposits (Cloud & Germs 1971) and microfossil recovery using traditional acid maceration approaches has met with only limited success (Strother et al. 2011).
In contrast with previous studies that concentrated on fossil material extracted from its host rock, in this work we used petrographic thin sections prepared from laminated calcareous mudstones within the Bay of Stoer Formation of the Stoer Group. We applied a wide range of high spatial resolution morphological and geochemical analytical techniques to characterize newly discovered microfossils and their associated minerals. We then compared and contrasted this style of microfossil preservation with that found in phosphatic deposits of the overlying Torridon Group. This provided a unique opportunity to compare the fossil preservation in Proterozoic lakes with similar amounts of organic matter, similar basement geochemistry and a broadly similar topographic setting (the same Archaean basement surface), but which may have developed different lake chemistries and microbial pathways. We then considered how the preservation of organic matter may be explained in terms of these differing lake biogeochemistries and implications that this may have for the Proterozoic fossil record as a whole.

Geological context
The Torridonian Supergroup consists of two successions separated by a marked angular unconformity (Fig. 1). The lower succession, the Stoer Group, accumulated in fluviatile and lacustrine environments in an extensional basin inboard of the Grenville Orogen on the Laurentian continent (Cawood et al. 2007). It has been dated at 1177 + 17 Ma (Parnell et al. 2011). During the subsequent formation of the Rodinia supercontinent, syn-collisional successions were deposited on the Laurentian craton, including the sandstone-dominated Torridon Group, dated at about 1000 Ma (Diabaig Formation ¼ 994 + 48 Ma; Turnbull et al. 1996), which is unconformable on the Stoer Group. There is a contrast in the deduced palaeoclimate, which was dry and Mediterranean-like for the Stoer Group and more temperate for the Torridon Group (Stewart 2002). Palaeomagnetic data indicate different locations at latitudes of c. 258 N and 30 -508 S, respectively (Stewart 2002).
The Torridonian Supergroup includes several horizons interpreted to include lacustrine environments from which diverse assemblages of microfossils have been obtained (Cloud & Germs 1971;Strother et al. 2011;Battison & Brasier 2012). Their lacustrine context is suggested by packaging within red fluvial sediments, the predominance of small ripple sets, mud cracks and rain pits (Stewart 2002), the presence of varve-like lamination (Andrews et al. 2010) and by the absence of large ripples or tidal features. The Stoer Group sediments are notably evaporitic with carbonate, sulphate and pyritic deposits, including those within the black shale lake sediments of the Poll a Mhuilt Member of the Bay of Stoer Formation (Parnell et al. 2010). The determination of annual lamination in the black shale is strong evidence for seasonal lacustrine deposition. By contrast, the Torridon Group deposits are typically non-evaporitic and sulphatepoor and include the lacustrine deposits of the Cailleach Head and Diabaig formations (Fig. 1). These distinctions are evident from the high sulphur content and sulphur to organic carbon ratios (mean S : C ¼ 3.56, n ¼ 85) of the Bay of Stoer Formation, Stoer Group and the uniformly low S to C ratios (mean S : C ¼ 0.35, n ¼ 104) of the Cailleach Head and Diabaig formations of the Torridon Group ( Fig.  2), even though the carbon contents of the two formations are otherwise similar (0-0.5%, mean c. 0.2% total organic carbon; Spinks et al. 2014). The high and low S to C ratios are reflected in the abundance and paucity of pyrite, respectively. Both sequences have experienced low-grade metamorphism (Stewart 2002), but both show comparable very good preservation of sedimentation features and primary variations in geochemistry. The contrasts between the two groups are summarized in Table 1. This paper combines hitherto independent studies of sulphate chemistry and microfossil preservation to show how they are related.

Determination of sulphur to carbon ratios
The sulphur and organic carbon contents were measured to a precision of +0.5% using a LECO CS225 elemental analyser after decarbonization with hydrochloric acid.

Sulphur isotope analysis
The sulphur isotope composition of both the sulphide and sulphate components was measured in a whole rock sample of Bay of Stoer Formation shale using the chromium reduction method of Canfield et al. (1986). The H 2 S generated from the reduction of sulphide sulphur by CrCl 2 was trapped as Ag 2 S in a solution of AgNO 3 . The resulting sulphide was washed, dried and analysed by conventional procedures following the method of Robinson & Kusakabe (1975). All the sulphur isotope data are reported in d 34 S notation as per mil (‰) variations from the Vienna Canyon Diablo Troilite (V-CDT) standard.
The sulphur isotope compositions of pyritized microfossils from the Bay of Stoer Formation were determined using a CAMECA IMS 1280 ion microprobe operating in the multi-collection mode at Centre for Microscopy Characterisation and Analysis (CMCA), University of Western Australia. The analytical protocol used the rastered beam approach outlined in Farquhar et al. (2013). The external precisions of 34 S/ 32 S on the SON-3 (pyrite) standard for the two analytical sessions were 0.2‰ (2 SD; n ¼ 14) and 0.1‰ (2 SD; n ¼ 10). The sample count rates varied between c. 5 and 90% of the count rate of the standard; however, there was no observed correlation between the secondary count rate and the d 34 S value of the samples. Instrumental mass fractionation and the propagation of uncertainty also followed the protocol in Farquhar et al. (2013). The uncertainty terms included the internal uncertainty, the external uncertainty of the instrumental mass fractionation of the primary standard and the uncertainty of the standard relative to V-CDT.

Focused ion beam milling combined with scanning electron microscopy
Focused ion beam milling combined with scanning electron microscopy was performed on a Zeiss Neon 40 dual-beam instrument at the Electron Microscopy Unit, University of New South Wales. The protocol followed that reported by Wacey et al. (2012), with the final milling and image capture parameters adjusted to suit the type of sample being analysed. For both the Bay of Stoer Formation and comparative Cailleach Head Formation material, the milling depth was c. 20 mm using a 2 nA ion beam current and the electron beam images were acquired with a c. 30 s capture time at 5 kV.

Transmission electron microscopy
Transmission electron microscopy (TEM) data were obtained from focused ion beam milled wafers using a JEOL 2100 LaB 6 transmission electron microscope equipped with a Gatan Orius chargecoupled device camera and a Tridiem energy filter, together with an FEI Titan G2 80-200 TEM/ STEM with ChemiSTEM technology. Both instruments are located in the CMCA and were operated at 200 kV. Energy-filtered TEM elemental maps on the JEOL 2100 instrument were obtained using the conventional three-window technique (Brydson 2001) with the energy windows selected to provide the optimum signal-to-noise ratios.

Nanoscale secondary ion mass spectrometry
Ion mapping of the Bay of Stoer Formation microfossils was performed using a CAMECA  NanoSIMS 50 instrument at CMCA, with the instrument parameters optimized as described in Wacey et al. (2011). The analysis areas were 30 × 30 mm at a resolution of 256 × 256 pixels and 45 × 45 mm at a resolution of 512 × 512 pixels (each pixel measuring 117 nm in the former and 88 nm in the latter), with a dwell time of 45 ms per pixel, and a primary beam current of c. 2.5 pA. The secondary ions mapped were 16 O 2 , 24 C 2 2 , 12 C 14 N 2 , 34 S 2 and 56 Fe 32 S 2 , and charge compensation was achieved using an electron flood gun.

Laser Raman microspectrometry
Laser Raman analyses were carried out at the University of Bergen using a Horiba LabRAM HR800 integrated confocal Raman system and LabSpec5 acquisition and analysis software. Samples were standard uncovered geological thin sections, allowing the optical and chemical maps to be superimposed. All analyses were carried out using a 514.5 nm laser, 100 mm confocal hole, 1800 grating and ×50 objective lens. The laser was focused at least 1 mm below the surface of the thin sections to avoid surface polishing effects. Dual acquisitions were taken from each analysis point, each with an acquisition time of 4 s, to identify the minerals from the Raman spectra. Raman maps were acquired with a 1.5 mm spatial resolution.

Fossil preservation in the Stoer Group
Thin sections from the Bay of Stoer Formation of the Stoer Group invariably yield rather poorly preserved microfossils. Under the optical microscope microfossils appear to be confined to sparse remains, many of which have an uneven, dappled surface texture and walls that appear broken or missing in places (Fig. 3). The microfossil morphotype diversity is rather low, ranging from roughly spherical to more elongate and collapsed bag-like forms, with rare examples containing smaller interior coccoid structures (Fig. 3). The microfossils are comparatively large, commonly 20-100 mm in diameter, and are restricted to parts of the thin sections that are rich in an opaque mineral, identified (using reflected light, Raman and nanoscale secondary ion mass spectrometry (NanoSIMS); Fig. 4) as pyrite. The three-dimensional preservation of most organisms (Figs 3 & 4) attests to early fossilization prior to compaction. Pyrite occurs in distinct narrow zones and, in most of the microfossil specimens analysed, it is intimately associated with organic carbon, with both phases outlining the shape of the microfossil walls ( Fig. 4b, e, g). Organic carbon and pyrite are rarely present in the interior of the microfossils, consistent with optical images suggesting little or no preservation of cellular contents in these samples. There is a gradation in the amount of organic carbon retained within the microfossil walls (compare Figs 4 & 5) and, in rare examples, all the organic carbon has been replaced (Fig. 5). In these instances the microfossil walls are instead denoted by narrow zones of closely spaced micro-pyrite grains plus aluminosilicates (Fig. 5). Two types of aluminosilicate occur within and adjacent to the microfossil walls: the more abundant phase is magnesium-and ironrich and the minor phase is potassium-rich. These have a flaky habit and both their habit and distribution are similar to the clay minerals described in a study of microfossils from the overlying Torridon Group (Wacey et al. 2014). Other mineral phases occurring either in the microfossil interior or nearby in the sediment matrix include calcite, anatase, quartz and feldspar, broadly consistent with the known calcitic mudstone lithology in which the microfossils are found. Where the preservation is poor, all the organic matter lacks detail, rather than a patchy preservation that could be attributed to partial replacement by a mineral such as calcite, or corrosion by an aggressive fluid.

Sulphur isotope compositions
In addition to the existing database for sulphide separates from the Stoer Group, which shows values lighter than 230‰ and a high degree of fractionation from coeval sulphates (Parnell et al. 2010), a new whole rock analysis from the Bay of Stoer Formation yielded measurements for sulphide and sulphate components of 231.7 and 235.3‰, respectively. Ion microprobe data from macroscopic pyrite within a Bay of Stoer rock chip gave d 34 S values between 234.2 and 238.7‰ with a mean of 235.5‰ (n ¼ 16; Table 2; Fig. 6). This mean value closely replicated the previously reported , showing multiple small pyrite grains aligned in narrow zones along the former vesicle wall (e.g. arrows). Note that the gaps (G) in the upper part of the vesicle wall correspond to the non-pyritized area in the optical image. (d-g) Reflected light images (parts d and f) and accompanying nanoscale secondary ion mass spectrometry (NanoSIMS) ion maps (parts e and g) of two further microfossils from the Bay of Stoer Formation. Yellow boxes indicate the areas analysed by NanoSIMS. In the NanoSIMS ion maps, pyrite (analysed as FeS 2 ) is shown in blue, with organic material (analysed as CN 2 ) shown in green. The pyrite and organic material are intimately related, but there is no continuous organic structure that would indicate pristine, non-decomposed cell or vesicle walls. Fig. 5. Nanoscale analysis of a pyritized microfossil wall using transmission electron microscopy. High-angle annular dark field (HAADF) image plus elemental maps of a pyritized microfossil wall (from the microfossil shown in Fig. 3d). The brightest correlated regions in the iron and sulphur maps indicate pyrite, whereas those in the calcium and carbon maps indicate calcite. Aluminosilicates are indicated by correlations in aluminium, oxygen, silicon and either iron, magnesium or potassium. Quartz is indicated by the correlation of silicon and oxygen, but the anti-correlation of aluminium and other elements. These data show that the organic material of the microfossil wall has been completely replaced by pyrite and aluminosilicates. Two types of aluminosilicate are present: a minor potassium-rich variety occurring very close to the pyrite and possibly replicating the original microfossil wall; and a major Mg-and Fe-rich variety that replicates the fossil wall and extends exterior to the former wall. Calcite is found interior to the former wall, whereas minor quartz is found both within and interior to the wall zone. laser ablation value (mean 235.4‰; Parnell et al. 2010), showing that the ion microprobe data are both accurate and reproducible. Ion microprobe analyses of the sulphur isotope compositions of pyritized microfossils in the Bay of Stoer Formation gave very light and consistent d 34 S values (245.8 to 248.8‰; n ¼ 8; Table 2; Fig. 7). No sulphur isotope data were collected from the Cailleach Head or Diabaig formations because these samples are pyrite-poor and pyrite has not been observed replacing microfossil organic material.

Fossil preservation in the Torridon Group
The Torridon Group shows a gradation in the fidelity of preservation of microfossils, as expected for any large population of organisms, some of which would have been alive and others at various stages of decomposition when fossilized. However, the Torridon Group is particularly notable for very high fidelity preservation within the phosphatic horizons and nodules of the Cailleach Head and Diabaig formations (Strother et al. 2011;Battison & Brasier 2012;Wacey et al. 2014). Here, microfossil preservation is among the highest quality known in the entire Precambrian rock record, with lightcoloured kerogen outlining not only intact microbial sheaths and multi-laminated or sculptured vesicles, but also clusters of cells, some with their contents partly intact (Fig. 8). Excellent three-dimensional preservation is facilitated by the encapsulation of organic material within francolite (calcium fluorapatite) and clay minerals ( Fig. 9; see also Wacey et al. 2014). We found that the proportion of organic matter identifiable in thin sections as cells, sheaths or vesicles was typically .20%. In contrast with the Stoer Group, cellular and subcellular features are composed entirely of kerogenous organic carbon (Fig. 9b -f ); mineral phases such as clays and phosphate appear to have nucleated on this material, but not grown within it or replaced it (Fig. 9c, f ). Neither the encapsulation nor replacement of cellular  Data from sample JPTOR19 were collected adjacent to those previously obtained using laser ablation gas source mass spectrometry (Parnell et al. 2010; see Fig. 6 for precise location of analysis spots). Data from sample PAR401 were collected from individual pyritized microfossils (see Fig. 7 for precise location of analysis spots).
organic material by pyrite was observed in the phosphate nodules and horizons of the Torridon Group. We therefore found a clear distinction in the fidelity of microfossil preservation between the phosphatic sediments of the Torridon Group, characterized by kerogenous cellular material preserving exquisite morphological detail and lacking pyritization, and the Stoer Group, possessing microfossils with degraded and partially replaced organic matter with a conspicuous degree of pyritization (compare Figs 4 & 5 with Fig. 9). This raises the possibility that these contrasts were related, at least in part, to the absence or presence of welldeveloped sulphur metabolic pathways in the contrasting lake ecosystems.

Fossil preservation
Pyritization is one of the processes that commonly causes the authigenic mineralization of organic matter, particularly in fine-grained organic-rich sediments, where dissolved iron reacts with the hydrogen sulphide produced by microbial sulphate reduction (MSR). A microbial origin for sulphate reduction can typically be evidenced from a large degree of sulphur isotope fractionation, as reported previously in the Stoer Group (Parnell et al. 2010) and reinforced by our new data. The precipitation of pyrite is controlled by the availability of metabolizable organic matter, plus the availability of sulphate and ferrous iron, with a balance between the decay of organic matter and its preservation by mineralization (Briggs et al. 1996;Schiffbauer et al. 2014). Pyrite formation rarely occurs early enough to preserve soft tissues (Briggs et al. 1991). In most instances, the organic matter associated with pyrite in the rock record is conspicuously degraded, which requires a timescale of only months to achieve (Sagemann et al. 1999).
The pyritization process provides a logical explanation for the contrasts in preservation between microfossils seen in the Stoer and Torridon groups. If iron and sulphate are plentiful in the early stages of heterotrophic activity, then labile organic matter will be fully consumed to precipitate pyrite. Hence the intracellular contents, delicate cells and trichomes will be the first organic structures to be degraded by sulphate-reducing microbes and there will be no trace of these as fossils, with pyrite precipitation typically occurring as framboids and micro-grains (which are plentiful in the Bay of Stoer thin sections). Later, the more resistant sheaths and vesicles will start to be consumed; we infer that iron was less plentiful by this stage (sulphate may also start to be depleted, but our d 34 S data suggest this was not the case here) and the precipitation of pyrite was proximal on and within the vesicle walls.
The high rates of sedimentation in lakes compared with oceans means that organic matter in lakes may be more rapidly and efficiently conveyed down into the anaerobic zone . Both experiments (Sagemann et al. 1999) and fossil studies (Wacey et al. 2013) reveal that MSR can result in the conspicuous heterotrophic remineralization of organic matter, resulting in the degradation of organisms and, under favourable iron-rich conditions, pyritization. There is a strong interrelationship between organic matter and MSR. The rate of MSR (if iron and sulphate are non-limiting) is controlled by the availability of organic matter and MSR causes the breakdown of the organic matter (Westrich & Berner 1984). The degree of breakdown depends on the type of organic matter, but freshly deposited micro-organisms are susceptible to rapid decomposition (Westrich & Berner 1984). As sulphate reduction is a more efficient source of energy than other anaerobic pathways, such as fermentation and methanogenesis, it can process organic material at a faster rate and to a greater degree. Studies in some modern lakes show that 80-90% of the total organic matter is decomposed by sulphate reduction during the summer when the lakes become stratified (Krevs & Kucinskiene 2012). The Bay of Stoer Formation black shales are the deposits of a stratified lake (Andrews et al. 2010), so may have similarly experienced significant MSR.
Heterotrophic activity in the absence of sulphate would still result in some degradation of organic material, as evidenced by the holes in some Torridon Group cells and vesicles (Fig. 8d), but this would not be driven by sulphate-reducing bacteria and pyrite would not precipitate. Fermentation and methanogenesis degradation pathways are significantly less efficient at processing organic material. This, combined with early encapsulation of cellular material by clay minerals and phosphate, would have limited the loss of morphological detail in the Torridon Group microfossils. An absence of iron may also result in a lack of pyrite precipitation even if sulphate is plentiful. However, the presence of syngenetic iron-rich clay minerals encapsulating the Torridon Group microfossils (Fig. 9c, e; see also Wacey et al. 2014) strongly suggests that it was sulphate, not iron, which was the missing ingredient preventing the formation of pyrite in the Torridon Group lakes. This is consistent with heavy sulphur isotope values in the Torridon Group (Parnell et al. 2010), which suggest a limited sulphate reservoir.
The rates of deposition of the Stoer and Torridon lake sediments were similar and fairly rapid . Diagenetic mineral growth around microfossils was also rapid, leading to threedimensional preservation, and was carbonate, silicate and pyrite in the Stoer lakes (Figs 4 & 5) and phosphate and phyllosilicate in the Torridon lakes ( Fig. 9; see also Strother et al. 2011;Wacey et al. 2014). If the composition of the organic matter was also similar, and there is no reason to assume otherwise, the more altered (including pyritic) state of preservation of the organic matter in the Stoer lakes may be explained by higher rates of sulphate reduction as a result of the freely available sulphate in the lake waters. Conversely, the less altered organic matter in the Torridon lake ecosystems seems to have been related to the lack of extensive sulphate reduction on or beneath the lake floor. This allowed much slower rates of organic matter degradation with the encapsulation of almost pristine organic material and the retention of exquisite anatomical detail.

Sulphur isotope compositions
The inference of MSR in the Bay of Stoer Formation is strongly supported by very light sulphur isotope compositions. The sulphur isotope data from sulphide nodules in the Bay of Stoer shales show a marked fractionation (up to 55‰) relative to contemporary sulphate sampled in red beds and are interpreted to reflect MSR in the lacustrine sediments (Parnell et al. 2010). The new data, from the separated sulphate and sulphide fractions of a whole rock shale sample, show that both fractions are characterized by light isotopic compositions (lighter than 230‰). Petrographic data show that the sulphate post-dated the sulphide . This implies that the sulphate fraction in the shale was oxidized from the microbially reduced sulphide. Sulphide oxidation can also be microbially mediated, with negligible isotopic fractionation (Brabec et al. 2012), consistent with the measured Stoer Group d 34 S data.
Pyrite precipitated on and around decaying micro-organisms records very light and rather consistent d 34 S values, even lighter than the values reported from the early diagenetic pyrite nodules and the whole rock data. This may indicate the control of MSR by a different population of bacteria and/or the increased cycling of sulphur in these microenvironments. However, this very light pyrite represents a very limited proportion of the total pyrite and therefore does not affect the bulk isotopic composition.

Implications for the fossil record
The data from the Torridonian Supergroup indicate that the high-fidelity morphological preservation of microscopic organic cellular material is favoured in environments conducive to phosphatization and clay mineral precipitation, as opposed to pyritization. Given that the presence of sulphate, rather than iron, appears to determine whether pyritization occurs in these lakes, this raises the intriguing possibility that sulphate-poor environments may have better preserved fossil records than sulphateenriched environments. Could this mean that Proterozoic shallow marine and terrestrial environments, where much of the fossil record of this age is found, have records that are biased by the pathways of decomposition?
There was an increased probability of sulphaterich conditions in shallow marine and terrestrial environments after the Great Oxidation Event, at least by the latter part of the Mesoproterozoic (Kah et al. 2004). Evidence for sulphate evaporites in the Mesoproterozoic is emerging on several palaeocontinents (Parnell et al. 2015). There are a number of examples in shallow water/terrestrial Proterozoic fossil assemblages where microfossils show a lower fidelity of preservation when pyritized than the adjacent non-pyritized specimens. These include the 1878 Ma Gunflint Formation, where organic microfossils encapsulated and permineralized in silica retain additional morphological features to those only a few micrometres away that have been pyritized (Wacey et al. 2013). In the c. 1650 Ma Amelia Dolomite from the Northern Territory of Australia, pseudomorphs of gypsum (calcium sulphate, a source of sulphate for MSR) occur in close proximity to abundant microfossils (Muir 1976). Furthermore, cells assigned to Sphaerophycus reticulatum that occur with abundant pyrite have been noted as being particularly decomposed compared with other genera in the same silicified horizon (Muir 1976, fig. 6) and examples of pyrite growing along the outlines of cell walls at advanced stages of decomposition have also been reported (Muir 1976, fig . 7).
In the c. 1450 Ma Wumishan Formation of northern China, gypsum crystals co-occur with pyrite crystals and very poorly preserved microfossils in thrombolites, although the quality of microfossil preservation may also have been adversely affected by carbonate precipitation in this setting (Tang et al. 2013). In the c. 1200 Ma Angmaat Formation of Baffin Island, gypsum is again commonly associated with the chert nodules that preserve microfossils (Knoll et al. 2013). Here, silicified microfossils show high-fidelity preservation (e.g. Knoll et al. 2013, figs 3 and 4), but microfossils are rare (and poorly preserved) to absent in zones where the organic material experienced early diagenetic pyritization. In the 750 Ma Chuar Group from the Grand Canyon, vase-shaped microfossils preserved as casts coated with pyrite provide less wall definition that those preserved as siliceous or organic-coated casts (Porter & Knoll 2000). By contrast, in mid-Proterozoic sediments where preservation is particularly good, pyrite is conspicuously lacking -for example, in the Changchang Group, north China (Lamb et al. 2009). These relationships are consistent with those observed in the Torridonian Supergroup and the combined evidence suggests that the heterotrophic consumption of cellular organic material by sulphate-reducing bacteria may have restricted the quality of the Proterozoic fossil record. The age range over which these contrasts are observed implies that the c. 200 myr age difference between the two deposits in the Torridonian Supergroup is not a critical factor. The examples where pyritized and non-pyritized fossils occur in close proximity serve to emphasize the relevance of pyritization to preservation. However, the more general examples are of either good (nonpyritized) or poor (pyritized) preservation. This potentially leads to bias, where the best-preserved fossils are from environments with anomalously low sulphate contents or very early permineralization by other anomalous pore water chemistries.
The norm, which in Mesoproterozoic marine rocks may be sulphate concentrations of a few mM (Kah et al. 2004), may not engender good preservation. The Torridonian example of a sulphaterich deposit is lacustrine, although the majority of the surviving Mesoproterozoic record is marine. However, the relevance of sulphate-rich pore waters to fossil preservation is independent of environment. Any perception that lake deposits are predominantly freshwater may also be erroneously biased by the current environment because the geological record shows that periods of widespread continental sedimentation commonly contain sulphate-rich terrestrial deposits (Warren 2010).
There would have been more rapid changes in redox conditions near the surface during the early stages of burial in the Proterozoic, with the precipitation of organic-stabilizing minerals more likely than during the Phanerozoic. Lower levels of atmospheric oxygen meant less penetration of oxygen into the surface sediments. There was a lack of bioturbation and scavenging in marine environments before the Cambrian (Canfield & Farquhar 2009). Changes in the redox potential therefore took place at shallower sediment depths than in the Phanerozoic. In lakes, in particular, because of their greater mean sedimentation rate , sediments rapidly passed through the zone of near-surface oxidation and into a realm where MSR and the precipitation of pyrite, phosphate and other minerals occurred. The contrasting history between sulphate-rich and sulphate-poor environments was therefore especially pronounced in these Proterozoic continental successions. The sulphate level in the Phanerozoic ocean was higher than in the Proterozoic, but this should not be a critical factor. Sulphate levels were already sufficiently high in the Mesoproterozoic ocean to cause massive gypsum precipitation and the presence of sulphate-saturated pore waters. With such a high availability of sulphate, the control on the degree of sulphate reduction to form pyrite is more likely to be the availability of organic matter or iron.
We interpret the Stoer lakes to have formed under conditions of interior, closed drainage under a warm, semi-arid climate conducive to evaporation and the formation of calcium carbonate and calcium sulphate. Abundant sulphate ions allowed for a fully expressed range of sulphur isotope fractionation during bacterial sulphate reduction and may also have led to episodically euxinic conditions in the deeper lake. By contrast, the younger Diabaig and Cailleach Head lakes formed under conditions with open rather than enclosed drainage and the climate was conceivably more temperate. Neither carbonate nor sulphate grew as authigenic precipitates on the lake floor. Instead, phosphate ions were concentrated during the degradation of organic matter and calcium phosphate readily precipitated when the redox conditions were favourable.

Conclusion
An important development during the Mesoproterozoic was the expansion of environments with sulphate evaporites as a consequence of the oxidative weathering of sulphide minerals in continental rocks. These environments were fundamental habitats for sulphate-reducing bacteria, which could utilize much or all of the available organic matter. Ironically, this expanding habitat was not conducive to providing a good fossil record. The whole fossil record is controlled by what preservation has allowed, but the Mesoproterozoic record may have a distinct bias as a result of the global rise in sulphate.
We acknowledge the Australian Microscopy & Microanalysis Research Facility at the Centre for Microscopy, Characterisation and Analysis, The University of Western Australia, a facility funded by the University, State and Commonwealth Governments. DW acknowledges funding from the European Commission and the Australian Research Council. This is publication number 838 from the Australian Research Council Centre of Excellence for Core to Crust Fluid Systems.