Introduction

The Late Ordovician was a time interval of continental glaciation developed in the Southern Hemisphere that coincided with a mass extinction event during the Hirnantian Age1, 2. Thus, the Hirnantian extinction event has been attributed to glaciation-induced environmental and climatic deteriorations and changes in ocean chemistry2,3,4,5,6,7,8,9,10. The Hirnantian Stage also records a concurrent positive δ13C excursion (up to ~7‰) observed globally both in carbonates and organic carbon1, 3, 8, 10,11,12,13,14,15, reflecting global perturbation of carbon cycle in Late Ordovician oceans. Two competing hypotheses have been proposed to explain the Hirnantian positive C-isotopic excursion termed HICE. It has been hypothesized that HICE may have resulted from a large increase in primary productivity and burial of organic carbon in the deep ocean3, 4. The enhanced burial of organic carbon could also have simultaneously promoted a drawdown of atmospheric pCO2 and ultimately led to the glaciation3, 4. However, organic-rich Hirnantian sedimentary rocks have not been identified12, 13, and the HICE would have predated the glacio-eustatic sea-level fall, had it been caused by enhanced burial of organic carbon and drawdown of atmospheric pCO2 11, 13. In fact, organic-rich sedimentary rocks and thus enhanced burial of organic carbon characterize uppermost Katian strata in deep marine sections in Nevada5, 11. Alternatively, Kump et al.16 hypothesized that the enhanced weathering of carbonate platforms that were exposed during glacioeustatic sea-level lowstand could have produced the HICE. According to the weathering hypothesis, a long-term drawdown of atmospheric pCO2 through increased weathering of silicate rocks could have led to the Hirnantian glaciation16. However, it has been argued that increased weathering of silicate rocks could not have led to the Hirnantian glaciation because the glaciation would first have to cause the sea-level fall that would expose silicate rocks to enhanced weathering12.

The interpretation of the HICE is important in our understanding of the causes of the Late Ordovician glaciation and paleo-environment in which the mass extinction occurred. In this study, we carry out high-resolution analyses of 87Sr/86Sr and δ13C for carbonates from the Copenhagen Canyon section in central Nevada, USA (Fig. 1). 87Sr/86Sr has been widely used to constrain continental weathering intensity in the geological past17,18,19,20,21. In particular, the relationships between marine 87Sr/86Sr ratios and glaciations have shown how glaciation, as a long considered effective cause of increasing weathering rates, could change the 87Sr/86Sr ratios of seawater22,23,24,25,26,27,28. The Ordovician 87Sr/86Sr of carbonates have provided insights into paleo-climatic changes29,30,31,32,33. However, few high-resolution 87Sr/86Sr measurements on the Hirnantian carbonates have been carried out to reconstruct continental weathering history during the glaciation. In this study, we report δ13C and 87Sr/86Sr data of carbonates from the Copenhagen Canyon section. Integrated with numerical modeling, our results can be used to test the competing hypotheses about the HICE and provide new insights into weathering history during the Hirnantian glaciation.

Figure 1
figure 1

Index map showing location of Copenhagen Canyon section in central Nevada, USA (modified from Finney et al.5).

Full size image

The Copenhagen Canyon section, situated in Nevada, was deposited in a shallow, platform margin to outer shelf setting11, 34. The biostratigraphy of the Copenhagen Canyon section is well established11, 34, and this section has been extensively studied and yielded excellent sedimentological and isotopic data11, 12, 16, 34,35,36,37. The lower part of the Copenhagen Canyon section consists of thick cherty limestone interbedded with thin calcareous mudstone and shale. On the basis of the stratigraphic patterns of lithofacies and fossil records, a sea-level curve has been reconstructed for the Hirnantian Stage at Copenhagen Canyon11. Within the P. pacificus Biozone, the lithofacies of darker gray, shaley rocks indicate a distinct deepening event, which just preceded the Hirnantian glaciation. Higher up in the section, a subaerial exposure surface is overlain by well sorted, fine quartz arenite, suggesting the maximum sea-level lowstand associated with the second advance of the ice sheet11, 36. Immediately overlying the exposure surface, wackestone, packstone, and dark gray lime mudstone with chert suggest post-glacial flooding.

Results

The δ13C values of the upper part of the Katian Stage vary between 0‰ and 3.2‰ and are followed by a typical HICE with a large positive δ13C excursion of ~7‰ (Fig. 2, Table S1). Our C-isotopic data are consistent with the previous measurements on the same section11, 12, 16. However, we observed a second positive δ13C excursion with a peak value of 5.1‰ in the lower N. persculptus Biozone, which coincides with the second pulse of the glaciation (Fig. 2, Table S1). Higher stratigraphically, δ13C returns to the pre-excursion value of ~0.6‰ in the upper N. persculptus Biozone (Fig. 2, Table S1).

Figure 2
figure 2

Integrated carbon-strontium isotopic chemostratigraphic profile of the Hirnantian glacial interval at Copenhagen Canyon, Nevada, USA. Two shallowing periods (light blue areas) are separated by an interval of deepening (light green area)35, 36. The sea-level curve is from Finney et al.11.

Full size image

The 87Sr/86Sr profile in the upper Katian is stable with a value of ~0.70791 (Fig. 2, Table S2), which is in good agreement with the 87Sr/86Sr measured in the correlative strata elsewhere31,32,33. Within the D. mirus Biozone, 87Sr/86Sr values increase sharply from ~0.70791 to 0.70803, which correlates well with the onset of the Hirnantian glaciation as well as the rise of δ13C (Fig. 2, Table S2). The 87Sr/86Sr values decline to 0.70793 in the lower N. extraordinarius Biozone where δ13C values reach the maximum of 7.2‰ (Fig. 2, Table S2). Stratigraphically higher, the 87Sr/86Sr ratio increases abruptly to 0.70803 and is then constant in the upper N. extraordinarius Biozone, which corresponds to the interglacial deepening as well as the decrease of δ13C (Fig. 2, Table S2).

Like the isotopic signature during the first pulse of the glaciation, an increase of 87Sr/86Sr ratio to 0.70818 followed by a decrease to 0.70807 coincides with the shallowing driven by the second pulse of the glaciation as well as the second positive C-isotopic excursion (Fig. 2, Table S2). The postglacial flooding in the upper N. persculptus Biozone coincides with an abrupt 87Sr/86Sr increase to 0.70815 and subsequent decrease to 0.70802 (Fig. 2, Table S2).

Discussion

Assessment of Diagenetic Alteration of Seawater 87Sr/86Sr

A critical evaluation of diagenetic alteration of seawater 87Sr/86Sr ratios from bulk carbonates is prerequisite to interpretation of significant changes in 87Sr/86Sr. Diagenetic alteration can cause depletion in Sr and enrichment in Mn (ref. 19) and thus Sr concentration of bulk carbonate is commonly used as a reliable indicator of diagenetic alteration17, 20, 30, 38. By paired 87Sr/86Sr measurements of Ordovician bulk carbonate and well-preserved conodont apatite, Edwards et al.30 concluded that the primary seawater 87Sr/86Sr can be faithfully preserved in bulk carbonate with Sr content >300 ppm.

All the samples we analyzed yield Sr concentrations >300 ppm except one sample with Sr content of 294 ppm, and 15 out of 33 samples contain Sr >1000 ppm. In addition, Mn contents for all samples are <120 ppm (26 out of 33 samples <40 ppm) and all ratios of Sr/Mn >4.9 (23 out of 33 samples >20). The high Sr concentration of >300 ppm, low Mn contents, and the consistent 87Sr/86Sr ratios of the Katian carbonates with the previous analyses suggest that the primary Hirnantian seawater 87Sr/86Sr values are well preserved at Copenhagen Canyon.

Modeling 87Sr/86Sr and δ13C: Implications for perturbations of carbon and strontium cycling

The marine 87Sr/86Sr ratio is predominantly determined by fluxes from rivers and seafloor hydrothermal exchanges at mid-oceanic ridges17, 19, 39. The 87Sr/86Sr ratios of river waters are highly variable (~0.711 or higher) and dependent on the relative contributions of continental weathering sources40, 41. For example, carbonates contain high Sr concentrations (up to 1000 ppm) but low 87Sr/86Sr ratios ranging from 0.706 to 0.709 (ref. 42). In contrast, old crustal terrains with greater resistance to weathering have lower Sr concentrations but high 87Sr/86Sr ratios of >0.710 (refs 39, 42). Moreover, basaltic volcanic rocks are characterized by nonradiogenic 87Sr/86Sr values of ~0.704 (refs 33, 40) and hydrothermal flux has a nearly homogeneous 87Sr/86Sr value of ~0.703 (ref. 43).

We use a numerical box model to simulate the δ13C and 87Sr/86Sr variations against the isotopic records in the Copenhagen Canyon section (see details in Table S3). Briefly, the model is composed of a system of three fundamental equations that illustrate the mass balances and isotope mass balances for Sr and C in the ocean. The mass balance equation can be expressed as equation (1),

$$\frac{d{M}_{i}^{SW}}{dt}={F}_{in,i}-{F}_{out,i}$$
(1)

where \({M}_{i}^{SW}\) represents the mass of element i in the ocean, t is time, \({F}_{in,i}\) and \({F}_{out,i}\) are total input and output fluxes of element i respectively.

For the Sr isotopic ratio \({R}_{Sr}^{j}={({}^{87}{\rm{S}}{\rm{r}}/{}^{86}{\rm{S}}{\rm{r}})}^{j}\), the rate of change of the seawater 87Sr/86Sr is given by equation (2),

$$\frac{d{R}_{Sr}^{SW}}{dt}=\frac{{F}_{in,Sr}({R}_{Sr}^{in}-{R}_{Sr}^{SW})}{{M}_{Sr}^{SW}}$$
(2)

where \({R}_{Sr}^{SW}\) and \({R}_{Sr}^{in}\) are 87Sr/86Sr ratios of seawater and input fluxes respectively.

For the carbon isotope systematics, the rate of change of the seawater δ13C is shown in equation (3),

$$\frac{d\delta {}^{13}{C}^{SW}}{dt}=\frac{{F}_{in,C}(\delta {}^{13}{C}^{in}-\delta {}^{13}C{}^{SW})-{J}_{out,ORG}^{C}{{\rm{\Delta }}}_{ORG-SW}}{{M}_{C}^{SW}}$$
(3)

where δ13CSW and δ13Cin are the carbon isotopic compositions of seawater and input fluxes, \({J}_{out,ORG}^{C}\) is output flux as organic carbon buried in sediments, Δ ORG-SW is the carbon isotopic fractionation between output flux of organic carbon and seawater reservoir.

In our simulation, we focus on changes in fluxes and isotopic compositions of the continental inputs derived from weathering of carbonates and silicates. The large positive δ13C excursion is assumed to start at t = 0 and changes in fluxes and isotopic compositions are all applied instantaneously to obtain the first-order estimate of the impacts on marine 87Sr/86Sr and δ13C induced by the glacial-interglacial cycles.

The original δ13C of continental inputs (δ13C in, CONT ) is ~−7‰, which includes 28% contributions from weathering of organic matter (δ13C in, ORG  = −25‰) and 72% from weathered carbonates (δ13C in, CARB  = 0‰)16. However, the δ13C in, CONT would increase to ~0‰ resulting from increased exposure and weathering of carbonate platforms during the glacio-eustatically controlled sea-level drawdown. The sea-level fall also led to the restricted seawater circulation between the Martin Ridge basin and the open ocean12, 36, and thus the contribution of C-flux from the open ocean would have been dramatically diminished by inputs of continental C-fluxes. Consequently, we obtained a positive excursion in δ13C of ~7‰ at ~0.38 Ma with timing and magnitude fitting well with the δ13C excursion in the Copenhagen Canyon section (Fig. 3B and C).

Figure 3
figure 3

Sea-level curve (A), δ13C and 87Sr/86Sr data from Copenhagen Canyon section (B,E), and the numerical model simulations of marine δ13C and 87Sr/86Sr as responses to increased weathering of carbonate during the Hirnantian sea-level drawdown (C,D). The reconstructed sea-level curve is from Finney et al.11.

Full size image

The ensuing decline of HICE was driven by a reduction in δ13C in, CONT (from 0‰ to −4‰, Table S3), reflecting ice sheet retreat, sea-level rise, transgression, and the accompanied reduction of carbonate weathering. The second, higher, positive δ13C excursion (from 3.4‰ to 5.1‰, Fig. 3B and C) also resulted from an increase in δ13C in, CONT (from −4‰ to 0‰, Table S3) that was induced by the re-exposure of carbonate platforms associated with the second glacial advance. Higher in the section, δ13C returned to the pre-excursion value coincident with the post-glacial flooding.

Our modeling of δ13C excursions is apparently consistent with the δ13C records from the Copenhagen Canyon section (Fig. 3). Therefore, our data and modeling suggest that, without a change of organic carbon burial rate, an increase in δ13C of the continental input C-flux can result in large positive δ13C excursions in the epeiric sea.

We can further test the weathering hypothesis by simulating 87Sr/86Sr variations because seawater 87Sr/86Sr ratio is an effective proxy of continental weathering. Figure 3D illustrates the simulation of the Hirnantian seawater 87Sr/86Sr oscillations by changing Sr-sources and their isotopic compositions. In our simulation, a good fit with the sharp increase in 87Sr/86Sr from ~0.70791 to 0.70803 near the base of the Hirnantian Stage is achieved when the 87Sr/86Sr ratios of silicates weathering flux increase from 0.721 to 0.7315 (Fig. 3, Table S3). An increased flux of silicate weathering (\({F}_{in,SIL}^{Sr}\)) can also lead to a rise of seawater 87Sr/86Sr ratios because both flux and 87Sr/86Sr ratio (\({R}_{in,SIL}^{Sr}\)) can affect the product \({F}_{in,SIL}^{Sr}\times {R}_{in,SIL}^{Sr}\) that represents chemical weathering rate of silicates. The increase in 87Sr/86Sr ratios of silicate weathering flux in our simulation is consistent with geological processes at the beginning of the Hirnantian glaciation. The enhanced mechanical erosion driven by glacial grinding and abrasion may have produced fine-grained Rb-rich glacial till with minerals such as biotite24, 25, which possesses radiogenic 87Sr/86Sr ratios of ≥0.9245 (ref. 44). Also, global cooling can independently promote the preferential weathering of biotite26. The sudden influx of these detrital phases to epeiric seas would cause an abrupt increase of seawater 87Sr/86Sr ratios. The abrupt increase of seawater 87Sr/86Sr along with gradual increase of δ13C near the basal Hirnantian Stage suggest that silicate minerals were the dominate source of Sr at the beginning of the glaciation despite gradual exposure of carbonate platforms.

To simulate the following sharp drop in 87Sr/86Sr, we increased the proportional contribution of carbonate weathering to the total Sr input fluxes from 36.4% to 61.7%, similar to our simulation of the δ13C rise during the same interval (Table S3). Evidently, the increasing Sr contents in this interval (Fig. S2), resulting from weathering of more exposed shelf carbonates due to glacial expansion, support the modeling analyses. The modeling also indicates that the decline to a minimum of 87Sr/86Sr ratios would take ~0.38 Ma (Fig. 3D).

The following increase to 0.70803 and the steady 87Sr/86Sr trend in the middle Hirnantian corresponded with the interglacial period of ice sheet recession that would have resulted in a sea-level rise and submergence of the carbonate platforms and also would have left behind extensive, fresh, fine-grained moraines highly susceptible to weathering, especially with warmer temperature and much more melt water24, 28. The initial weathering of these fragile materials would elevate 87Sr/86Sr ratios of silicate weathering flux from 0.721 to 0.7315, the same as the onset of the glaciation (Table S3) owing to the preferential weathering of biotite, which is a common Rb-rich mineral with high 87Sr/86Sr ratios24, 25, 44. However, the 87Sr/86Sr ratios would decrease with the increasing age of the materials being weathered24, 28. As a result, there would be a weathering “spike” immediately following the deglaciation and a steady 87Sr/86Sr trend at 0.70803 in the middle Hirnantian (Fig. 3D and E).

Thereafter, the advance of the second glaciation led to the same changes of C and Sr cycling as that of the first glaciation. The rise of seawater 87Sr/86Sr at the beginning and end of the second glaciation would also be driven by increase in 87Sr/86Sr of silicate weathering flux (Table S3) associated with preferential weathering of biotite. Whereas, the glacial maximum would be consistent with positive δ13C excursion and coeval with drop of 87Sr/86Sr that was caused by enhanced carbonate weathering (Table S3).

However, the modeling results show that the increase of 87Sr/86Sr from 0.70803 to 0.70818 associated with the onset of the second glaciation would have taken ~0.17 Ma owing to the long residence time of Sr (~2.7 Ma) in seawater18. This is inconsistent with our observation that the 87Sr/86Sr dropped to 0.70807 at the same time as the peak in the second positive δ13C excursion at ~0.89 Ma (Fig. 3B and E). The only way to achieve synchronous shifts in δ13C and 87Sr/86Sr at ~0.89 Ma is to decrease the residence time of Sr by decreasing the mass of Sr in the reservoir, by increasing the input and output Sr-fluxes, or by a combination of the both. It is reasonable to significantly decrease the residence time of Sr in an epeiric sea due to smaller water masses and higher flux of weathered Sr from carbonates during the Hirnantian eustatic lowstand. Using a residence time of 0.83 Ma, a good fit of δ13C and 87Sr/86Sr shifts during the second glaciation is obtained (Fig. 3).

The second deglaciation led to a weathering “spike” of 87Sr/86Sr values increasing to 0.70815 rapidly and then declining to 0.70802 (Fig. 3E). These 87Sr/86Sr patterns are the same as those of the first glaciation advance, and are essentially controlled by the changes in the weathering rates of carbonates and biotite associated with the waxing and waning of the Hirnantian glaciations. Thus the Hirnantian glaciation within ~1 Ma correlates qualitatively with changes in δ13C and seawater 87Sr/86Sr.

The glacial advances and associated changes in weathering regime, the mass fraction and isotopic ratios of different input fluxes chosen, together with local effects of C and Sr-cycling in the epeiric sea, offer an internal consistency to the observed C and Sr isotopes records in the Copenhagen Canyon section. The C and Sr isotopic data support the hypothesis that the Hirnantian positive δ13C excursion resulted from enhanced carbonate weathering during glacioeustatic sea-level drawdown.

Methods

For δ13C analysis, approximately 150 μg samples were reacted with ~103% phosphoric acid at 70 °C in a Kiel IV carbonate device connected to a Thermo Scientific MAT 253 mass spectrometer. The carbon isotopic compositions are reported in the standard delta (δ) notation as permil (‰) deviations from Vienna PeeDee Belemnite (V-PDB), with external precision of ~0.05‰ (1σ) based on duplicate analyses of an internal standard. For analysis of 87Sr/86Sr ratio, carbonate powders of ~120 mg were dissolved in 30% acetic acid at room temperature to avoid dissolution of the non-carbonates. The solution then was centrifuged, evaporated and re-dissolved in 2.5 N HCl, standard cation-exchange procedures were performed to purify Sr from matrix ions and 87Sr/86Sr ratios were analyzed using a Thermo Scientific Triton thermal ionization mass spectrometer, following methods outlined in Lin et al.45. The reported 87Sr/86Sr ratios were corrected for instrumental mass fractionation using a ratio of 86Sr/88Sr = 0.1194. Trace metal concentrations (Mn, Sr) were measured using an ICP AES instrument with a reproducibility of ±10% (2σ).