Stratigraphic variability in the geochemistry of sedimentary rocks provides critical data for interpreting paleoenvironmental change throughout Earth history. However, the vast majority of pre-Jurassic geochemical records derive from shallow-water carbonate platforms that may not reflect global ocean chemistry. Here, we used calcium isotope ratios (δ44Ca) in conjunction with minor-element geochemistry (Sr/Ca) and field observations to explore the links among sea-level change, carbonate mineralogy, and marine diagenesis and the expression of a globally documented interval of elevated carbon isotope ratios (δ13C; Hirnantian isotopic carbon excursion [HICE]) associated with glaciation in Upper Ordovician shallow-water carbonate strata from Anticosti Island, Canada, and the Great Basin, Nevada and Utah, USA. The HICE on Anticosti is preserved in limestones with low δ44Ca and high Sr/Ca, consistent with aragonite as a major component of primary mineralogy. Great Basin strata are characterized by lateral gradients in δ44Ca and δ13C that reflect variations in the extent of early marine diagenesis across the platform. In deep-ramp settings, deposition during synglacial sea-level lowstand and subsequent postglacial flooding increased the preservation of an aragonitic signature with elevated δ13C produced in shallow-water environments. In contrast, on the mid- and inner ramp, extensive early marine diagenesis under seawater-buffered conditions muted the magnitude of the shift in δ13C. The processes documented here provide an alternative explanation for variability in a range of geochemical proxies preserved in shallow-water carbonates at other times in Earth history, and challenge the notion that these proxies necessarily record changes in the global ocean.
Ancient shallow-water carbonate strata provide important archives of the evolution of Earth’s climate and environments. In particular, the carbon isotopic composition (13C/12C) of ancient shallow-water carbonate sediments has been widely applied as both a chronostratigraphic tool (Saltzman and Thomas, 2012) and an indicator of the partitioning of global carbon fluxes between carbonate and organic reservoirs, linking the global geochemical cycles of carbon and oxygen (Kump and Arthur, 1999; Saltzman et al., 2011). However, this interpretation of the carbon isotopic composition of shallow-water carbonates has been questioned by studies of modern analogues (Swart and Eberli, 2005; Swart, 2008; Higgins et al., 2018), where local processes, early diagenetic alteration, and changes in δ13C of carbonate minerals conspire to decouple the chemistry of shallow-water carbonate sediments from the global ocean.
Hirnantian δ13C and Sea Level
Upper Ordovician (Hirnantian) strata around the globe contain an interval of elevated δ13C, referred to as the Hirnantian isotopic carbon excursion (HICE). The HICE is associated with sea-level fall during the Hirnantian glaciation, which lasted less than 1.3 m.y. and has a geographically variable magnitude, ranging from ∼+2‰ to ∼+7‰ (Melchin et al., 2013, and references therein). Geochemical anomalies in δ15N, δ34S, 87Sr/86Sr, δ26Mg, and δ7Li (LaPorte et al., 2009; Zhang et al., 2009; Hu et al., 2017; Kimmig and Holmden, 2017; Pogge von Strandmann et al., 2017) are also found in Hirnantian strata. The origin of the HICE has been attributed to increased global organic carbon burial (Brenchley et al., 1994); synglacial changes in δ13C of the riverine weathering flux (Kump et al., 1999); and/or local carbon cycling in shallow basins partially isolated from the global ocean (Melchin and Holmden, 2006). However, these models can require unrealistic changes to carbon burial and/or weathering fluxes, or they predict cross-platform δ13C gradients (δ13C increasing with greater proximity to the coast) that contradict the variability observed in some basins.
Here, we developed a new hypothesis that explains the HICE as a record of shallow-water aragonite δ13C that evolved during sea-level change and was variably altered during early marine diagenesis. We used measurements of δ44Ca, δ13C, and Sr/Ca to identify the primary mineralogy and early diagenetic history of carbonates deposited across the Hirnantian interval. We propose a mechanism that links the observed magnitude of the HICE to glacioeustasy, depositional environment, and early diagenesis, and we discuss the implications for understanding the links between elevated δ13C and global environmental change in the geologic record.
Diagenetic Framework for Shallow-Water Marine Carbonate Sediments
The pairing of calcium isotopes (δ44Ca) and trace-element ratios (Sr/Ca and Mg/Ca) has emerged as a promising tool to identify both primary mineralogy and the style of early diagenetic alteration of marine carbonate sediments (Fantle and DePaolo, 2007; Fantle and Higgins, 2014; Higgins et al., 2018). In particular, cross-plots of δ44Ca and Sr/Ca in carbonates of all geologic ages exhibit covariation that can be quantitatively related to the style of early marine diagenesis. The covariation arises for three reasons. First, early diagenetic carbonate minerals are characterized by low Sr/Ca partition coefficients (Brand and Veizer, 1980) and small Ca isotope fractionation factors (Fantle and DePaolo, 2007; Jacobson and Holmden, 2008; Tang et al., 2008) compared to primary marine carbonate minerals (although Baker et al.  showed that deep-sea pelagic calcite can maintain high Sr/Ca during marine diagenesis). Second, δ44Ca and Sr/Ca depend on primary mineralogy, with aragonite precipitation fractionating Ca isotopes to a greater extent (Gussone et al., 2005) and incorporating more Sr (Veizer, 1983) than calcite. Third, early marine diagenesis in shallow-water carbonate sediments can occur under both fluid- and sediment-buffered conditions, depending on the extent to which pore-fluid exchange with seawater occurs through advection or diffusion (Banner and Hanson, 1990; Fantle and DePaolo, 2007; Fantle and Higgins, 2014; Higgins et al., 2018). Together, these properties of primary and diagenetic carbonate minerals and shallow-water sedimentary systems lead to a characteristic relationship between bulk δ44Ca and Sr/Ca and provide a means with which to characterize the effects of mineralogy and early marine diagenesis on the geochemistry of shallow-water carbonate sediments (Ahm et al., 2018; Higgins et al., 2018).
We generated δ44Ca (δ44/40Ca) and Sr/Ca data from 328 samples from two Upper Ordovician carbonate ramps. Limestone samples from the Vaureal and Ellis Bay Formations (Anticosti Island, Canada; Fig. 1D) were selected from those reported by Jones et al. (2011). Additional Anticosti material was sampled from the New Associated Consolidated Paper (NACP) drill core (49°37.337′N, 63°26.292′W; Desrochers et al., 2010; McLaughlin et al., 2016). Samples of the Ely Springs Dolostone (Great Basin, Nevada and Utah, USA; Fig. 2D) were selected from those reported by Jones et al. (2016) along a depth transect from shallow- to mid-ramp settings (Carpenter et al., 1986). Data from a deep-ramp section (Holmden et al., 2012) completed the Great Basin transect.
Samples were analyzed for cation ratios and calcium stable isotopic composition at Princeton University (New Jersey, USA) following the methods of Higgins et al. (2018). Ca isotope measurements are reported as the relative abundance of 44Ca to 40Ca using standard delta notation, normalized to the isotopic composition of modern seawater. The external reproducibility for SRM915b calcium carbonate standard was −1.16‰ ± 0.19‰ (2σ, N = 38), and reproducibility for an internal aragonite standard was −1.47‰ ± 0.15‰ (2σ, N = 12). New δ13C data were generated for the NACP core following the methods of Jones et al. (2011); all other δ13C data came from Jones et al. (2011, 2016).
The Anticosti Island and Great Basin data varied stratigraphically and geographically in δ13C, δ44Ca, and Sr/Ca (Figs. 1 and 2; Fig. DR1 in the GSA Data Repository1). Ca isotope values for limestones were between −0.9‰ and −1.7‰; dolostone values were between −0.5‰ and −1.5‰. Great Basin dolostones showed low Sr/Ca (0.06–0.23 mmol/mol), whereas Anticosti limestones had higher Sr/Ca that spanned a wider range (0.22–2.69 mmol/mol). Cross-plots of δ44Ca and Sr/Ca revealed covariation similar to that observed for Neogene carbonates from the Bahamas (Figs. 3A and 3C). Cross-plots of δ44Ca and δ13C showed a slight negative correlation for Ordovician and Neogene samples (Figs. 3B and 3D); Ordovician samples with elevated δ13C, including the HICE, were characterized by δ44Ca < −1.0‰ and elevated Sr/Ca.
Measured δ44Ca and Sr/Ca values of limestones and dolostones from Anticosti Island and the Great Basin spanning the HICE carry a geochemical fingerprint of mineralogy, early marine diagenesis, and dolomitization that is indistinguishable from Neogene platform top and margin strata from the Bahamas.
First, the range in δ44Ca and Sr/Ca values in the Ordovician data set spans the range of values observed for the Neogene Bahamas (Fig. 3; Fig. DR2), indicating that these sediments experienced a scope of early diagenetic conditions from sediment-buffered (geochemical records set by the chemistry of the primary sediment: low δ44Ca; high Sr/Ca) to fluid-buffered (geochemical records modified by the seawater-like chemistry of fluid flushed through pore space: high δ44Ca; low Sr/Ca; Higgins et al., 2018; Ahm et al., 2018).
Second, strata containing the HICE in Monitor Range, Nevada, and in the NACP core (Figs. 1 and 2) are distinguished by δ44Ca and Sr/Ca values characteristic of a primary mineralogy dominated by aragonite, indicating that the HICE records a change in the δ13C of shallow-water aragonite and the dissolved inorganic carbon (DIC) in these environments. The δ13C of shallow-water DIC likely evolved during Hirnantian sea-level change, as Anticosti samples with high Sr/Ca and low δ44Ca span a range of δ13C from ∼1‰ to ∼4‰. This change is similar to the ∼2‰ increase in the δ13C of shallow-water aragonite in the Bahamas toward the Holocene (Swart and Eberli, 2005; Swart, 2008), albeit on a different time scale. As this Neogene increase is demonstrably decoupled from the δ13C of the global ocean, we suspect the same is true for the HICE—it reflects a change in the δ13C of shallow-water DIC and associated aragonite but not a change in the δ13C of DIC in the global ocean.
Finally, the Great Basin exhibits a distinct geographic pattern in δ44Ca across the ramp that bears a strong resemblance to observations in the Bahamas (Figs. 2B and 2C). The inner-ramp section is characterized by the highest δ44Ca; here, the HICE is missing due to nondeposition or erosion. Three mid-ramp sections preserve a modest HICE and intermediate δ44Ca, whereas deep-ramp limestones at Monitor Range have the lowest δ44Ca and the largest-magnitude HICE (Holmden et al., 2012). We interpret this cross-platform geochemical gradient to be a consequence of spatially variable pore-fluid flushing by seawater. The interpretation that flushing was highest in the landward direction is consistent with models of fluid convection on the Great Bahama Bank, in which pore-fluid velocity is typically greatest on the shallow bank-top relative to the slope (Caspard et al., 2004). Glacial-interglacial sea-level changes may have enhanced fluid flow and promoted dolomitization in both the Bahamas (Swart and Melim, 2000) and the Ordovician strata considered here. While cross-platform variations in submarine groundwater discharge and carbonate saturation state can produce geographically variable δ44Ca (Holmden et al., 2012; Shao et al., 2018), our geochemical and geological observations suggest that these factors were negligible in the Great Basin (see the Data Repository).
Taken together, the Ca and C isotope records from Anticosti Island and the Great Basin provide a new framework with which to interpret the origin, preservation, and diagenetic alteration of positive δ13C values associated with icehouse climates. In this model, platform carbonate sediments are dominated by aragonite. While the Ordovician ocean is commonly considered to have been a “calcite sea” (Hardie, 1996), low-latitude shallow platforms were warm even during the Hirnantian (Finnegan et al., 2011) and would have readily precipitated aragonite (Balthasar and Cusack, 2015; see also the Data Repository). Schematically, we envision the following scenario (Fig. 4): Before glaciation, aragonite δ13C hovered around 0‰, with increases in shallower environments. Rapid flushing of seawater shifted the δ13C signal toward seawater values (Fig. 4A). Sea-level fall associated with the Hirnantian glaciation moved shallow-water environments to more distal settings that had previously existed as deeper-ramp environments (Fig. 4B). Changes in the carbon cycle on the platform during the sea-level lowstand resulted in elevated aragonite δ13C on the ramp (Fig. 4B). The mechanism for increased δ13C in platform waters remains enigmatic (as it is for the ∼2‰ increase in the Bahamas over the Neogene), but an increase in platform productivity linked to a decline in water depth and reduced exchange with the open ocean is one possibility (Panchuk et al., 2006). After rapid ice melting and re-flooding, ramp δ13C declined, and distal shallow-water deposits were buried by offshore highstand deposits (Fig. 4C).
Because pore-fluid circulation on carbonate platforms is most intense toward the basin margin (i.e., landward) and most sluggish in the basin center (i.e., offshore), the effects of early marine diagenesis were most pronounced in updip settings, where intense fluid flow caused fluid-buffered alteration (Fig. 4C) and, in the Great Basin, regional dolomitization. Those updip deposits acquired geochemical signatures reflecting the chemistry of seawater-like pore fluid (δ13C and δ44Ca both close to 0‰). In contrast, the synglacial aragonitic lowstand deposits were subjected to less advective flux and were more protected from this seawater-buffered alteration (at Monitor Range), generally retaining their primary geochemical signatures representing shallow-ramp environments, even after neomorphism to calcite (Fig. 4C).
Our analysis does not constrain the magnitude of the HICE for the global ocean and does not require that any component of the HICE represents changes in the partitioning of global carbon fluxes between carbonate and organic carbon (Kump and Arthur, 1999). Nevertheless, many deep basin sections record a small-magnitude HICE in organic carbon isotope ratios (Gorjan et al., 2012). Whereas these records may themselves be sensitive to platform δ13C (Oehlert and Swart, 2014), an ∼1.5‰ global HICE has been suggested based on analysis of an open-ocean–facing section in Nevada (Ahm et al., 2017). Any such global signal would have been significantly modified in shallow carbonate settings by the processes considered here.
Paired analyses of δ44Ca and Sr/Ca in ancient shallow-water carbonate sediments provide a novel framework with which to interpret carbon isotope excursions in the geologic record. This framework provides a way to identify primary versus diagenetic geochemical signals and to understand how early marine diagenesis varied in both space and time. Records of δ44Ca and Sr/Ca of the HICE from the Great Basin and Anticosti Island are consistent with this framework; strata deposited in outboard settings prior to sea-level rise preserved the geochemistry of the primary sediment—sediment that reflects the chemical conditions (e.g., δ13C) on the local carbonate platform and not the global ocean (Swart, 2008). In contrast, strata deposited in shallow proximal settings during sea-level highs experienced early marine diagenesis under fluid-buffered conditions and recorded the geochemistry of these pore fluids (e.g., modified seawater). Many intervals of elevated δ13C in the Paleozoic are characterized by spatially varying magnitudes, are short in duration, and are associated with sea-level changes (Farkaš et al., 2016). As such, the mechanisms invoked here for the Late Ordovician are likely to be widespread over Earth history, with broad implications for interpreting the environmental significance of δ13C excursions and associated geochemical perturbations recorded in ancient platform carbonate rocks.
Jones acknowledges the Donors of the American Chemical Society Petroleum Research Fund for partial support of this research. We are grateful to Ministère de l’Energie et des Ressources Naturelles du Québec for access to the NACP core. Ahm acknowledges support from the Simons Foundation (New York, SCOL 611878). Elizabeth Lundstom, Danielle Santiago Ramos, and Stephanie Moore provided laboratory support. We thank the three anonymous reviewers for constructive comments.