Did anoxia terminate Ediacaran benthic communities? Evidence from early diagenesis

The Ediacaran oceanic redox landscape was heterogeneous, where many basins had a shallow and highly dynamic chemocline above anoxic (ferruginous or euxinic) or low oxygen (manganous) waters. Seawater mMg/Ca ratio was also high, promoting early diagenetic dolomitisation. How the benthos responded to these conditions is fundamental to understanding their ecological dynamics. Here we utilise redox sensitive elements in early marine carbonate cements to investigate possible water column redox controls on the distribution and growth of the oldest metazoan communities. Skeletal communities in the Zaris Sub-Basin of the Nama Group, Namibia (~550547 Ma), grew in shallow waters where fine-grained carbonate sediment often shows evidence of early dolomitisation. Mid-ramp Cloudina reefs are composed of open, highly porous structures that formed multiple, successive assemblages. Each assemblage is terminated by thin (< 1 mm), layers of dolomicrite sediment and

These evolutionary step-changes occurred in a globally heterogeneous redox landscape characterised by redox-stratified oceans and a highly dynamic chemocline (e.g. Lyons et al., 2014;Kendall et al., 2015). Redox instability continued long after the first appearance of probable metazoans, skeletal hard parts and metazoan reefs .
In addition, there is evidence that early marine Fe-rich dolomite precipitation dominated deeper waters in some late Ediacaran oceans, implying that not only was much of the water column ferruginous and anoxic, but may also have contained elevated Mg/Ca ratio Hood and Wallace, 2018). Mimetic preservation by dolomite (i.e., retention of original crystallographic orientation) of original aragonite and/or high-Mg calcite (HMC) grains (Tucker, 1982;Corsetti et al., 2006), as well as the presence of well-preserved dolomite cements, suggests that early marine dolomite precipitation dominated Cryogenian to early Ediacaran oceans (ca. 740 to ca. 630 Ma; Hood and Wallace, 2015) and locally in the terminal Ediacaran (~550 Ma; Wood et al., 2016). This has been inferred to be due to widespread lowoxygen or stratified oceans and high-Mg/Ca seawater (Hardie, 2003;Hood et al., 2011;Hood and Wallace, 2018). The presence of high iron (ferroan) concentrations in early dolomite cements (Hood and Wallace, 2015), and ferroan dolomite concretions and iron speciation from shales further indicates that these oceans were anoxic and ferruginous (Canfield et al., 2008;Planavsky et al., 2011;Poulton and Canfield, 2011;Spence et al., 2016). Dramatically enhanced continental weathering occurred during the Neoproterozoic, creating a marked increase in carbonate deposition inferred to be due to a substantial input of Ca 2+ into seawater (Peters and Gaines, 2012). Fluid inclusion data confirm that seawater Ca 2+ increased markedly and Mg 2+ declined slightly during the Ediacaran to early Cambrian, so progressively lowering mMg/Ca by the early Cambrian (Brennan et al., 2004).
The Ediacaran skeletal macrobiota were restricted to probably discontinuous habitable areas of well-oxygenated waters above the chemocline Tostevin et al., 2016). The role of dynamic redox conditions in relation to the nature of the earliest metazoan benthos is therefore fundamental to understanding the ecological dynamics of the first metazoan communities. However, most bulk-rock geochemical techniques provide an indication of the dominant redox condition during sedimentation over geological rather than ecological timescales, and thus rapid fluctuations in water column redox conditions are seldom recorded. For example, in the lower Nama Group, Namibia, and the Windermere Supergroup, Canada, local palaeoredox indicators show pervasive anoxic signatures despite the presence of insitu benthic communities of inferred metazoans (Johnston et al., 2013;Sperling et al., 2015a, b;Tostevin et al., 2016). This is suggestive of opportunistic benthic colonization during periods of transient oxygenation with episodic encroachment of anoxic waters beneath a very shallow, fluctuating chemocline .
The unusual chemical state of Ediacaran oceans will be manifest in the record of early marine cements. Here we consider the petrography, trace element and stable isotope composition of varied carbonate cements in metazoan reefs from the late Ediacaran Nama Group, Namibia. The use of early marine cements allows high-resolution reconstruction of the evolution of seawater-sourced pore fluids in shallow carbonate settings (Nothdurft et al., 2004;Della Porta et al., 2015;Hood and Wallace, 2015;Wallace et al., 2017). This is of particular utility in the Ediacaran, as shallow shelf carbonates host the highest diversity of skeletal metazoans. Dolomite is also common in the Nama Group, and all dolomitic lithologies analysed via Fe speciation suggest precipitation under anoxic water column conditions .
Therefore, we also consider evidence that dolomitisation may indicate transient anoxic seawater conditions, which in turn may have structured the longevity and ecology of Ediacaran skeletal communities.

Petrographic evidence for Neoproterozoic marine dolomite
In order to distinguish cements as marine, they must fulfil criteria that indicate precipitation from seawater-sourced fluids (Davies, 1977;Wallace, 2012, 2018). These include 1) an isopachous, pore-lining fabric of fibrous and inclusion-rich crystals oriented normal to cavity walls, 2) cements directly intergrown with or succeeded by marine geopetal sediments, and 3) cements truncated by other earlyforming features such as neptunian fractures.
That some late Neoproterozoic carbonate cements represent marine dolomite is based upon crystallographic character and how dolomite replaces pre-existing carbonate fabrics (Tucker, 1983;Hood and Wallace, 2012. Carbonate cements precipitated in seawater will be length-fast (with a low angle between the crystal c-axis and the fastest growth vector), and grow rapidly either as acicular, aragonite (orthorhombic) crystals with blunt terminations, or more slowly as calcite (trigonal) crystals with either acute rhombohedral, or scalenohedral form (Folk, 1974;Berner, 1975). By contrast, sparry calcite will be length-slow (with a high angle between the crystal c-axis and the fastest growth vector) and is most commonly precipitated as late stage (burial) pore-occluding cement where low Mg/Ca solutions and negligible contamination by Mg 2+ ions does not inhibit growth perpendicular to the c-axis (Folk, 1974;Folk and Land, 1975). Marine dolomite cements (which do not naturally occur in seawater today) always show a length-slow character and commonly conform to three crystal forms termed 'radial slow', 'radiaxial slow' and 'fascicular slow' dolomite (Hood et al., 2011;Hood and Wallace, 2012. Dolomitisation of a calcite cement precursor will be syntaxial and preserve the original character and optical continuity of calcite as a consequence of the common crystal form (trigonal rhombohedra) shared by both dolomite and calcite (Hood and Wallace, 2012). By contrast, dolomitisation of an aragonite precursor will only crudely preserve the original aragonitic fabric due to the contrast in crystal form between aragonite (orthorhombic) and dolomite (trigonal) (Hood and Wallace, 2012).
As such, primary marine dolomite is represented by an isopachous and inclusion-rich cavity-lining cement, which retains its original length-slow character Wallace, 2012, 2018).

Geological Setting
The Nama Group (~550 -541 Ma) is a terminal Ediacaran succession of highly fossiliferous mixed clastics and carbonates ranging from supratidal to outer ramp settings with varying hydrodynamic conditions (Grotzinger and Miller, 2008). The Nama Group was deposited across the Zaris and Witputs sub-basins separated by the Osis Arch (Fig. 1A), and the strata have been correlated using sequence stratigraphy, chemostratigraphy, and dated ash beds (Saylor et al., , 1998Grotzinger et al., 1995).
We consider in-situ Cloudina reefs at Driedoornvlagte reef complex (Penny et al., 2014;, from the far north of the Zaris subbasin (Fig. 1A), which formed during deposition of from the Omkyk Member, Kuibis Subgroup (Fig.   1B). Driedoornvlagte is an isolated carbonate platform approximately 10 km in length and ~500 m thick that grew in a high-energy, mid-ramp setting (Smith, 1998;Adams et al., 2004). Uranium-lead chronology of an ash bed in the immediately overlying lower Hoogland Member (upper Kuibis Subgroup) yields an age of 547.32 ± 0.65 Ma Schmitz, 2012).

Water column redox
Low total iron (FeT) in carbonates, normal marine Fe/Al in shales and carbonate, and the presence of negative cerium anomalies in carbonates indicates an oxic environment, suggesting that reef growth at Driedoornvlagte was into a dominantly oxic water column Tostevin et al., 2016;Bowyer et al., 2017).
Sequence stratigraphy shows that reef growth took place during a long-lived transgression (Adams et al, 2004). However, adjacent deeper waters were characterised by either low-oxygen manganous conditions, or anoxic ferruginous conditions Tostevin et al., 2016;Bowyer et al., 2017). A very shallow oxycline persisted throughout deposition of shelf grainstones during the Lower Omkyk Member (OS1), as evident from anoxic ferruginous waters during transgression, to manganous conditions during the highstand systems tract . By contrast, early Upper Omkyk Member (OS2) sediments record a transition to well-oxygenated conditions, and reef growth at Driedoornvlagte took place during an interval of stable oxia. Similarly, oxic water column conditions dominated time-equivalent deposition at more proximal mid-ramp sites, including Zebra River where smaller, biostromal reefs also formed .
By contrast, pulsed ferruginous conditions are recorded in inner ramp settings, which may reflect either development of sluggish circulation and elevated productivity, or upwelling of anoxic deep water during transgression Bowyer et al., 2017). Ce anomaly data indicate episodic incursion of manganous waters at shallow water depths at inner ramp sites, and at intermediate Fe T > 0.5 wt%, Fe HR /Fe T from 0.29 to 1. These are strongly suggestive of formation within an anoxic and iron-rich water column .
Marine transgressions show repetitive shoaling of the oxycline, and general reef growth may have been terminated by clastic influx and the development of anoxic ferruginous conditions with deposition of the overlying transgressive Urikos Member shales on the deep, outer shelf at Driedoornvlagte (Adams et al., 2004;Bowyer et al., 2017). A broadly synchronous, but short-lived, phase of anoxia is documented across the mid to inner ramp at the maximum flooding surface of the large scale transgressive systems tract (at the base of parasequence 4 of Adams et al., 2004Adams et al., , 2005, contemporaneous with the Urikos shale .

Architecture and fabric of the Driedoornvlagte Reef Complex
The notably expanded stratigraphic thickness of OS2 at Driedoornvlagte relative to shallower sections is inferred to be due to foreland basin subsidence in proximity to the Damara orogen (Smith, 1998;Grotzinger and Miller, 2008). Driedoornvlagte carbonate platform growth kept-up with subsidence-induced relative sea level rise during deposition of the OS2, but was eventually drowned by fine, basinal siliciclastics of the Urikos Member, equivalent to carbonate production of OS2 Units 4 and 5 in shallower sections to the south (Smith, 1998;Adams et al., 2004Adams et al., , 2005Dibenedetto and Grotzinger, 2005).
The carbonate build-up of OS2 is interpreted to record 3 cycles (Units 1 -3) of relative sea level change, each of which is represented by sequentially expanded stratigraphic horizons as a consequence of increasing subsidence during deposition (Figs. 1b and 2) (Adams et al., 2004). Microbial-metazoan reefs at Driedoornvlagte grew in association with coalesced thrombolite mounds of the final cycle of reef growth (Unit 3m), representing a ~50m thick transgressive succession which formed in a shallow subtidal setting immediately prior to drowning Adams et al. 2004). There have been no recorded occurrences of evaporite pseudomorphs, karstification or other features indicative of subaerial exposure or evaporitic conditions. The inferred accommodation increase during deposition of Unit 3m is based upon an observed increase in size of coalesced thrombolite-stromatolite mounds, and the formation of synsedimentary fissures and collapse breccias (Adams et al., 2004).
In Unit 3m, Cloudina reefs are broadly associated with Namacalathus communities, and Namapoikia, which preferentially inhabited synsedimentary fissures (Wood et al, 2002;. Although the presence of Cloudina reefs has been disputed (Mehra and Maloof, 2018), in-situ reef growth is confirmed by: a) the mutual attachment of Cloudina individuals by 'meniscus' skeletal structures in life position together to form frameworks, and b) decimeter-scale cement-filled, reef framework cavities (Penny et al., 2014;. Dolomite cement increases in volume from the base to the top of each assemblage, and field observations suggest that its occurrence was controlled by the available pore space for precipitation after growth of acicular cement. In a ~1m transect taken at 20 mm intervals though a Cloudina reef ( Abundance (density), however, shows no correlation with position within a community, but does show an overall general decrease through the transect (Fig. 4B).
Environmental controls on Cloudina size are unknown, and may include cyclical changes in nutrient levels, water depth, hydrodynamic energy, space availability, or carbonate supersaturation.
In the Nama Group, marine transgressions led to a shoaling of the oxycline . Here, we hypothesize that termination of Ediacaran skeletal communities across the ramp may be due to the flooding of shoaling, anoxic waters.
We test this by considering the succession of calcite and dolomite cements, and sediments.

Materials and Methods
Large hand samples were taken from multiple Cloudina reefs from the ~ 50m thick mixed microbial-metazoan Unit 3m on the southeast flank of the carbonate build-up on Farm Driedoornvlagte (Fig. 2). The palaeo-horizontal was defined by micritic geopetal sediments within and around Cloudina individuals.
Thin sections (75 x 50 mm) were etched with 0.2 M hydrochloric acid for 10s followed by staining with a mixed solution of Alizarin Red S and Potassium ferricyanide in 0.2 M HCl for approximately 30s before rinsing with deionised water (Dickson, 1965).
We examined polished thin sections under a Cathodoluminescence Cold Cathode CITL 8200 MK3A mounted on a Nikon optiphot microscope to identify differing luminescence (Habermann et al. 1996;1998;Mason and Mariano, 1990 (Lumsden and Chimahusky, 1980), using a weighted mean incident wavelength for CuKα radiation of 1.541838Å and the d 104 peak at a 2θ angle of 30.9°.

Distribution of cements, petrography and elemental geochemistry
Driedoornvlagte Cloudina reefs show a consistent diagenetic paragenetic sequence of six successive cement generations (Fig. 5), each with distinct petrographic and cathodoluminescence characteristics, and Fe, Mn and Sr concentrations (Table 1 and Supplementary Data) as described below. Large (up to 40 mm radius), volumetrically abundant acicular cements with square crystal terminations form botryoids (AC; Fig. 3D), which are pseudomorphed from aragonite (Grant 1990). These grow dominantly from Cloudina skeletons (Fig. 6B).
AC also intergrew with geopetal sediment (Fig. 6A). Acicular cements are mainly non-luminescent ( At the top of each Cloudina assemblages, thin (< 1 mm) layers of geopetal, internal dolomicritc sediment (DS) appear, commonly with a nonplanar texture (Sibley and Gregg, 1987;Figs. 6A, E). This sediment appears only in the upper parts of each assemblage, and often, but not always, terminates acicular cement growth (Fig. 6A).
Geopetal sediments also commonly infill Cloudina individuals (Fig. 7A), and the elemental composition ranges of these sediments is indistinguishable from both the sediment that covers acicular cements and Cloudina skeletons. Dolomicritic sediment fill has elevated Fe (up to 6850 ppm, mean 1980 ppm, n= 52), and slightly elevated Mn (up to 1000 ppm, mean 220 ppm, n= 52). Sr reaches up to 720 ppm, mean 150 ppm (n= 52).
The microcrystalline skeletal structure of some Cloudina individuals has been partially or completely dolomitised (D1). D1 is mid-brown and inclusion-rich, with poorly preserved, often patchy, CL zonation ( indicative of a length-fast character (Fig. 9F). This difference in inferred crystallographic character may simply be a consequence of perspective whereby the plane of the thin section bisects length fast crystals at angles divergent from the c-axis (Dickson, 1983). In some rare instances, D2 cements contain cubic crystals with a maximum diameter of 20 µm, which display blue-violet luminescence under CL Individual crystals exhibit an average length-width ratio of 5:1 when the plane of the thin section bisects the crystal parallel to the c-axis. C1 cement is not isopachous but may grow either vertically upward or downwards from cavity ceilings and directly nucleate upon D2, skeletal Cloudina material or AC cements. C1 columnar scalenohedral cements invariably radiate from single points, forming bundles which, when cut perpendicular to the plane of the thin section, appear as rosettes with nonluminescent to patchy bright luminescent or non-luminescent columnar cores (C1a) and brightly luminescing rims (C1b) (Figs. 9G,H).
Rarely, C1a/b is overgrown by a limpid euhedral calcite cement of moderate luminescence (C1c) (brighter than D1 and D2 but duller than C1b) which lacks inclusions but preserves fine CL zonation with only very minor variability in luminescence between growth zones (Fig. 7F). This cement sometimes appears as an epitaxial overgrowth of C1 (Fig. 7F)

Origin of calcite cements
Iron and manganese concentrations influence cathodoluminescence in marine carbonate cements, so providing constraints on the redox conditions for the pore fluids (Barnaby and Rimstidt, 1989;Hood and Wallace, 2015 decreasing Eh that are in equilibrium with poorly crystalline Fe-Mn oxides (Barnaby and Rimstidt, 1989). The overall Fe-Mn and cathodoluminescence behaviour of the Driedoornvlagte reef cements are similar to that observed in burial cement successions from Phanerozoic carbonates (e.g. Grover and Read, 1983).
Cathodoluminescence may also reveal fine growth zones within cements, which correspond to variable redox chemistry (Barnaby and Rimstidt, 1989).
Recrystallisation or replacement by another mineral will destroy original cathodoluminescence zonation. The preservation of such delicate zonation, which does not extend beyond crystal boundaries, therefore indicates that cements are a product of primary growth. Zonation patterns are controlled by the crystal form of each cement type, and so reveal the direction of pore-filling crystal growth away from a substrate Wallace, 2012, 2018).
The inclusion-rich, acicular cement (AC) is interpreted to represent an early marine precipitate of originally aragonitic mineralogy. The interpretation of an originally aragonitic mineralogy for AC is supported by the elevated Sr content (up to 6900 ppm), as previously reported at Driedoornvlagte by Grant (1990). The acicular, aragonite-precursor cement (AC) is commonly overlain by both geopetal dolomicrite sediment (DS) inferred to be of marine origin, and by Cloudina individuals, which either grew against or were deformed by acicular cement (Fig. 8A).
The general absence of clear CL zonation in the dolomite cements (D1 and D2) suggests recrystallisation after a precursor, but these cements are nonetheless distinct from both burial cements and coarsely recrystallised, dolomitic replacements (Whittaker et al., 1994;Corsetti et al., 2006). However, there are localised instances where the dolomite cement shows fine zonation, revealing a crystal form very similar to shorter, wedge-shaped marine dolomite cements of the older Karibib Formation, Northern Namibia, which also show greater definition of growth zonation towards rhombic terminations (Hood and Wallace, 2018).
The columnar length-fast character and scalenohedral form of C1 cement crystals when bisected parallel to the c-axis (e.g. Fig. 9H)

Origin of dolomite cements
We infer that the elevated iron concentrations found in dolomitised Cloudina individuals (D1), geopetal dolomicrite sediment, and dolomite cements D2 were introduced during the dolomitisation process and were not a pre-existing feature of any of the precursors. Therefore, constraining the timing of dolomitisation is key to understanding the evolution of redox conditions.
As a result of the slow kinetics of dolomite formation at temperatures less than 40°C (Arvidson and Mackenzie, 1999), it has been proposed that under near-normal sea water conditions unstable, precursor, Ca-Mg carbonate phases form, such as very high-Mg calcite (VHMC). Prior to dolomite formation VHMC, over decades or more, may become more ordered dolomite through dissolution and re-precipitation (Kaczmarek et al., 2017). This progression is common in natural settings where higher burial temperatures and long time periods overcome kinetic barriers to dolomite formation (Kaczmarek et al., 2017).
High magnesium calcite (HMC) has previously been inferred for the mineralogy of original Cloudina shell material at Driedoornvlagte and has similarly been suggested to provide a localised source of Mg 2+ required during subsequent diagenetic dolomitisation (Grant, 1990), forming the D1 cement. Early dolomitisation of a HMC precursor cement is similarly inferred for D2, as neither the preceding acicular cement, nor later calcite cements, are dolomitised. This is supported by the observation that some of the D2 cement crystals exhibit the classic HMC form with a length-fast character and acute rhombic crystal terminations (Figs. 7D, 8F).
The presence of dissolution suggests that precipitation took place under variable conditions which may have included multiple fluid alteration stages accompanied by possible changes in carbonate saturation state. Dolomite cement increases in volume from the base to the top of each assemblage, and is controlled by the available pore space for precipitation after synsedimentary acicular cement growth.
There is a complete overlap in the δ 18 O carb of dolomitic and calcitic components, but dolomitic components reach slightly lower values. The most depleted δ 13 C carb values are also associated with dolomite (from -0.29 to 2.62‰) compared to the calcitic components (from 1.15 to 3.02‰) (Fig. 10). There is a notable positive correlation between δ 13 C carb and δ 18 O carb within the dolomite samples, which strongly suggests the influence of diagenetic alteration which progressively lowered values.
Low oceanic oxygen levels in the Neoproterozoic might provide a control on marine dolomite precipitation by promoting microbial reactions such as methanogenesis or sulphate reduction (Burns et al., 2000). The carbon isotopic composition of dolomite may also reveal the degree to which dolomitization was controlled by bacterial processes within anoxic pore fluids. Using carbon isotope signatures, both sulphate reduction and methanogenesis have been evoked to explain precipitation of transient dolomite cements in Holocene shallow marine carbonate platforms (Teal et al., 2000) and for marine dolomite cements of the Cryogenian Oodnaminta Reef complex (Hood et al., 2011;Hood and Wallace, 2012).
Notwithstanding the very low quantities of pyrite (< 0.1 wt%), the light carbon isotope signature of some combined dolomite sediment and dolomite cement samples relative to all other components supports a role for bicarbonate produced by a variety of early diagenetic organic matter remineralisation reactions.
Dolomitisation and dolomite formation in the modern is largely restricted to shallow sediments that are frequently subject to evaporative conditions (Machel, 2004). Though reef growth of Unit 3m is inferred to have occurred above fair weather wave base, there are no features of subaerial exposure or evaporative conditions but rather transgression and deepening leading to eventual swamping by basinal shales of the Urikos Member (Adams et al., 2004). Given the high porosity of the reef system, pore spaces were likely connected to the overlying water column prior to inundation by Urikos sediments. Circulation through the sediment of considerable volumes of seawater is required for dolomite formation, driven by tidal plumbing, wind generated currents and storms. We infer that sediment porosity was sufficiently high to have maintained interstitial seawater circulation and dolomite precipitation. Furthermore, precipitation of the dolomite cement was likely promoted by sustained, elevated pore water alkalinity, where Mg was sourced from seawater and possibly aragonite neomorphism (although this would most likely have occurred mainly during burial), as well as localised sulphate reduction. Elevated Fe also supports a role for anoxia.
These conditions, coupled with unstable and reactive precursor phases such as HMC, would further enhance the dolomitization potential of fluids sourced from seawater (Corsetti et al., 2006).
The geopetal sediment likely acted as a route for dolomitising fluids due to its relatively high permeability and small grain size. But the timing of dolomitisation of the geopetal sediment, HMC Cloudina individuals, and inferred HMC precursor cement is difficult to constrain. However the fact that D2 cement crystals largely show a unimodal size distribution is suggestive of a single cementation event (Machel, 2004). Fluorite is a common accessory mineral in dolomite as a consequence of increasing Ca 2+ and decreasing Mg 2+ (resulting in decay of MgF + complexes) in pore solutions during dolomitisation (Dickson, 1980;Möller et al., 1980). Iron speciation and Ce/Ce* data strongly suggest local water column conditions to have been variably ferruginous, manganous and oxic Tostevin et al., 2016b;Bowyer et al., 2017). The simplest explanation for the sequence of inferred redox conditions and associated crystallographic character of carbonate cements is the effect of seawater of changing reduction potential prior to pore water disconnect by deposition of the Urikos Member. In this case, dolomitisation is inferred to have been early, and was catalysed in the presence of ferruginous waters of elevated Mg/Ca, in addition to anoxic microbial activity.
We interpret such a succession of early marine cements in these open reef systems to represent precipitation at either successively deeper positions in the water column or through changing pore water chemistry due to transgression and/or upwelling, from shallow oxic (AC, C1a) through to manganous (C1b). The dolomitized sediment (DS) and associated dolomite crust (D1 and 2) may have formed as a result of a temporary incursion of Mg-rich, anoxic ferruginous waters, combined with sulphate reduction, but this hypothesis remains to be tested. Dissolution is common, suggesting possible changes in the mean carbonate saturation state. These changes in sea water and/or pore fluid chemistry may have been produced by upwelling related to high-frequency cycles of transgression which led to shoaling of the oxycline .

Conclusions
Ediacaran oceans were probably of high Mg/Ca with a shallow and fluctuating oxic chemocline compared to most of the Phanerozoic. Our cement data are consistent with the interpretation that Nama Group Ediacaran skeletal communities grew in oxic waters but close to the chemocline, with growth terminated by low oxygen, manganous waters that shoaled during late transgressive to early highstand    Adams et al., 2004) with carbon isotope chemostratigraphy (Smith, 1998) and metazoan fossil distribution.