Introduction

A major shift in climate and oxygen levels in Earth’s atmosphere beginning near the Emsian-Eifelian boundary (~ 395 Ma)1 and continued into the Early Frasnian when forests were expanding2,3,4. The world’s first forests were identified in the late Emsian in Spitzbergen and in Givetian strata in Gilboa, New York, USA5,6, however, Capel et al.3 identifies several major origination-extinction pulses during the Silurian-Devonian that eventually resulted in a transition to a forested terrestrial landscape during the Middle Devonian. By the end of the Givetian, root networks had deepened and by the Frasnian, aneurophyte and archaeopterid progymnosperm forests were common, resulting in thicker soil horizons starting to form; thereby increasing terrestrially-derived nutrient delivery to the marine environment2,4,7. Previous studies of these shifts in biodiversity predicted that enhanced nutrient delivery may have caused increases in productivity, oxygen stratification, deposition of organic rich black shales, and eutrophication in Frasnian epicontinental seas2,4,8,9. Middle to Late Devonian lacustrine sediments from Greenland and northern Scotland reveal a net loss in phosphorous (P), an essential biolimiting nutrient that is expected to decrease in a terrestrial environment that is undergoing plant colonization where P is liberated from minerals indirectly through the acidification of root pore spaces produced by the degradation of organic matter and release of organic exudates from roots8,10,11. A significant and long-lasting δ13C shift in the punctata conodont zone, thought to be caused by the increased delivery of liberated nutrients (e.g. P) that would enhance productivity and burial of organic carbon in the Middle to Late Devonian, is referred to punctata Event (pE) and is recognized in basins worldwide12. Suspected productivity associated with the pE may also result in amplification of biologically sourced amorphous SiO2 in areas that experienced nutrient influx through the delivery of soils formed by deeper root networks2,8. Amorphous SiO2 has been consistently underestimated in ancient sedimentary sequences, which distorts our understanding of global biogeochemical silica cycling13,14,15,16. Silica in shales was commonly interpreted as terrigenous in origin; however, Schieber17 and Schieber et al.14 demonstrated that significant proportions of quartz silt in shales could be biogenically- or diagenetically-derived, especially after the early Cambrian when radiolarians and siliceous sponges started to proliferate16. The SiO2-rich Frasnian Duvernay Formation shales displays δ13C(org) excursions characteristic of the pE, which have also been documented in the Canadian Rocky Mountains18. These basinal deposits are therefore examined in this study to determine whether SiO2 in the Duvernay shales is of biological origin and whether increases in SiO2 deposition could be linked to the significant shift in the terrestrial realm when the world’s forests were expanding.

Differentiation of SiO2 polymorphs is possible but is challenging on a macroscale. Currently the methods used to differentiate amorphous (possibly biogenic) versus crystalline SiO2 are:

  1. 1.

    Identification of siliceous microfossils or SiO2-filled cysts17,19, textures recognized in petrographic work such as irregularly shaped grains with embayments and pointy projections17,20, pyrite inclusions17,19, and quartz grains with colloform or chalcedonic textures14,17;

  2. 2.

    Scanning electron microscope cathodoluminescence imaging or energy dispersive x-ray spectroscopy14,17,19,21,22;

  3. 3.

    Oxygen isotope values14,23;

  4. 4.

    Silica excess; defined by Rowe et al.24 as the absolute value of the difference between the measured silica content and the silica versus aluminium regression line, which represents silica in the aluminosilicate phase (i.e., clay minerals)20;

  5. 5.

    A negative correlation between silica and zirconium25;

  6. 6.

    A positive correlation between silica and TOC20;

  7. 7.

    X-ray diffraction, where peak height differences are used as a crystallinity index26

  8. 8.

    Alkaline digestion27.

Each of these methodologies requires spot analysis where a specific interval is targeted through sampling. To detect SiO2 on a larger scale, our study uses longwave infrared spectroscopy (LWIR) in the 8–12 µm wavelength range28. LWIR can differentiate between amorphous and crystalline SiO2 due to asymmetric stretching of Si–O–Si bonds in amorphous SiO229,30. Using this method of detection can enhance our understanding of biogeochemical SiO2 cycling through time, since it enables in situ detection on the macroscale, thereby allowing us to map amorphous SiO2 distributions in ancient sedimentary basins and track shifts in bioproductivity that may coincide with significant climatic shifts.

Three drill cores through the Duvernay shales from different locations across the Western Canada Sedimentary Basin in Alberta were sampled at regularly spaced intervals for whole-rock geochemistry and stable isotope analyses. Shales are less susceptible than carbonate rocks to diagenetic alteration that may affect δ13C, used to identify the pE. Each core is analysed for δ13C(org) to determine if the pE excursions are recorded in each of these locations. SiO2 provenance is investigated using oxide data and geochemical proxies to determine whether there is any “excess silica”24 present in the Duvernay and if so, what the source may be. Any SiO2 that is contributed from a biological source, signalling bioproductivity associated with the pE, would likely be an amorphous polymorph of SiO2. To detect and map amorphous SiO2, the cores are imaged using a longwave infrared (LWIR) drill core hyperspectral scanner. The aims of the study are to determine whether the pE is recorded at each location in the basin, to identify intervals of amorphous SiO2, and to determine the source of SiO2 using several geochemical proxies and hyperspectral imaging. Possible sources of amorphous SiO2 include hydrothermal fluids, SiO2 produced as a by-product of clay diagenesis, or biogenic SiO2. A biogenic source of SiO2 in the Duvernay, deposited during expansion of the world’s forests, would support current theories that the development of deeper Archaeopteris tree roots resulted in increased soil genesis and riverine nutrient delivery8, ultimately increasing bioproductivity and deposition of organic material in the oceans at this time2,4,12. This study, focused on differentiation and map** of SiO2 polymorphs contributes to our understanding of silica in the world’s oceans through time, which is linked to global carbon, oxygen and climate cycles27. The novel technique used to differentiate detrital versus amorphous SiO2 may provide a critical advancement in our ability to quantify biogenic SiO2 in sediment, which is considered a high priority in develo** a better understanding of biogeochemical silica cycling8,27.

Geological background

The Duvernay Formation shales are organic-rich, calcareous to argillaceous/siliceous shales that were deposited during the Frasnian-age (382–372 Ma) Woodbend Group (Fig. 1), when the WCSB was a passive margin on the western passive margin of the North American craton31. The most up to date conodont stratigraphy from the WCSB places the lower portion of the Duvernay in the punctata conodont zone (Montagne Noire (MN) 5/6) and the upper Duvernay mostly in the hassi conodont zone (MN7-10)31. During this time, present-day Alberta was covered by a large interior seaway, with numerous carbonate buildups that grew in succession. The Leduc Formation formed on paleotopographic highs on the thickest portions of the underlying Swan Hills Formation platform in the west, and the Cooking Lake Formation platform in the east32,33. The first two stages of reef growth in the Leduc Formation are coeval with deposition of the Duvernay basinal mudstones and shales33. In the east, the Duvernay directly overlies the Cooking Lake Formation platform carbonates and in the west, the basinal equivalent Majeau Lake Formation. These western and eastern portions of the WCSB, known as the West Shale Basin and East Shale Basin respectively, are separated by the Rimbey-Meadowbrook reef trend (Fig. 1)34.

Figure 1
figure 1

(A) Map of Alberta with extent of the subsurface Duvernay Formation. modified from the Geological Atlas of the Western Canada Sedimentary Basin28. (B) Simplified stratigraphy of the Duvernay Formation in Alberta, Canada and the location of three drill cores used in this study.

Full size image

The Duvernay is highly heterogeneous both locally and across its 130,000 km2 depositional extent35. The Duvernay, or the Perdrix as it is referred to in the Rocky Mountain exposures, is comprised of ten lithofacies, defined by composition, grain size, and sedimentary structures in Knapp et al.36. Broadly, the Duvernay is composed of organic rich siliceous mudstones, carbonate, and clay-rich shales36. Based on informal litho- and chemostratigraphy and areas of production from the Duvernay, it is divided into three domains herein referred to as: Kaybob, Willesden Green (WG), and East Shale Basin (ESB) (Fig. 1). In the Kaybob area, the Duvernay has seven geochemically distinct units, and is characterized by the presence of a middle carbonate unit37,38. In our study, the Kaybob well is divided into the upper Duvernay, middle Duvernay (carbonate-rich), and the lower Duvernay. In the Willesden Green (WG) area, the Duvernay is comprised of thin, intercalated shale and fine-grained carbonate beds. In the East Shale Basin (ESB), the Duvernay is surrounded by the Leduc Formation reefs and as a result is relatively carbonate rich throughout, in comparison to the WG and the upper and lower Duvernay in the Kaybob well.

Analytical methods

Three Duvernay Formation cores were sampled and imaged in this study; 100/04-19-064-22W5/00 (Kaybob), 100/09-25-039-06W5/00 (Willesden Green = WG), and 100/08-29-031-23W4/00 (East Shale Basin = ESB; Fig. 1).

Data collection

A total of 303 samples were taken at 0.5–1 m intervals from three Duvernay drill cores. The samples were analysed for major elements using inductively coupled plasma optical emission spectrometry (ICP-OES). Samples from the Kaybob drill core were analysed for elemental chemistry using an iCAP 7000 Series ICP-OES (Thermo Fisher Scientific, Waltham, MA, USA) at Chemostrat Inc. laboratories in Houston, Texas, following the procedure outlined in Hildred et al.39. Samples from the Willesden Green and East Shale Basin cores were analysed using a Spectro ICP-OES (Perkin Elmer, Waltham, MA, USA) at Bureau Veritas Mineral Laboratories in Vancouver, British Columbia. Oxides analysed in the samples include: SiO2, TiO2, Al2O3, Fe2O3, MgO, MnO, CaO, Na2O, K2O, and P2O5. Total organic carbon (TOC) and δ13C(org) (reported relative to VPDB) were analysed at Chemostrat Inc. for the Kaybob well, and at the University of Saskatchewan Stable Isotope Laboratory for the WG and ESB. All of the samples analysed for δ13C(org) and TOC were acidified to remove carbonate material, homogenized, and the organic material oxidized to carbon dioxide, nitrogen bearing gases, and water, which are further separated for analysis. Isotopes were measured at Chemostrat Inc. using a Europa Scientific 2020 Isotope Ratio Mass Spectrometer and at the Saskatchewan Stable Isotope Laboratory using a Thermo Finnigan Flash 1112 EA coupled to a Thermo Finnigan Delta Plus XL through a Conflo III, where they were calibrated against international standards L-SVEC (δ13C(org) = − 46.6 VPDB) and IAEA-CH-6 (δ13C(org) = − 10.45 VPDB) with a lab reported precision of 0.12% (n = 18, 2σ). Fifty polished thin sections were prepared and examined from the Kaybob well. Each section was impregnated with blue epoxy to highlight pore space, and the sections were half stained with Alizarin red to recognize calcite and aragonite. The thin sections were examined using a Zeiss Axioscope A1 at MacEwan University in Edmonton, Alberta. Twenty-five samples from the Kaybob well were also analysed for mineral content using X-ray Diffraction (XRD) at the Chemostrat Inc. laboratories in Houston, Texas.

Data analysis

SiO2 and Al2O3 data were used to identify the presence of excess silica that is not considered to have been derived from continental crust, using Eq. (1)20.

$${\text{oxide}}_{{{\text{EX}}}} = {\text{oxide}}_{{{\text{sample}}}} {-}\left[ {\left( {{\text{oxide}}/{\text{Al}}_{2} {\text{O}}_{3} } \right)_{{{\text{background}}}} \times {\text{Al}}_{2} {\text{O}}_{{3\;{\text{sample}}}} } \right]$$
(1)

A value of 3.527 was used for average shale or background40. The amount of excess silica (SiEX) is reported alongside δ13C(org) data. The most recently reported conodont biostratigraphy presented in Wong et al.31 was used to correlate δ13C(org) data collected in the Duvernay with previous studies of the punctata Event12. Data analysis was performed on oxides and TOC analysed in the three wells using multivariate statistics (Principal Component Analysis; PCA) in DataDesk® 6.3.1. to determine SiO2 provenance. PCA analyzes the total variance of the oxide and TOC dataset to determine the relationship of SiO2 to clay-associated oxides (e.g. Al2O3, TiO2, K2O, MgO, Fe2O3, Na2O), carbonates (e.g. CaO, MnO), or to variables that represent a biological influence (e.g. P2O5, TOC)41. In this study, eigen vectors e1 versus e2 and e1 versus e3 were plotted for each well to account for greater than 85% of variance in the dataset.

A potential hydrothermal source for amorphous SiO2 is evaluated using an Al–Fe-Mn ternary plot. Adachi et al.42 defines a zone that implies a hydrothermal influence near the Fe-apex (up to 30% Mn) and non-hydrothermal as being Al-rich, and Mn-poor. Samples from each well in this study are plotted on this ternary diagram.

Clay diagenesis (e.g. K-metasomatism) can produce amorphous SiO2 as a by-product43. To determine whether the clays in the Duvernay have undergone significant diagenetic alteration that may have introduced amorphous SiO2, oxide data is used to first determine the chemical index of alteration (CIA), which assesses the degree of weathering that the sediments have undergone from their source (Eq. 2)44,45.

$${\text{CIA}} =_{{{\text{molar}}}} \left[ {\left( {{\text{Al}}_{2} {\text{O}}_{3} } \right)/\left( {{\text{Al}}_{2} {\text{O}}_{3} + {\text{CaO}}* + {\text{ Na}}_{2} {\text{O}} + {\text{K}}_{2} {\text{O}}} \right)} \right] \, \times 100$$
(2)

CaO* should represent Ca only in the silicate portion44,45. Following McLennan44, CaO was corrected for using phosphate data where CaO* is equal to the moles of CaO minus moles of P2O5 × 10/3. This value is compared to the moles of Na2O, and if the corrected value of CaO* is less than moles of Na2O, than this CaO* value is used in Eq. 2, otherwise CaO* is equal to Na2O.

CIA values calculated for sediments that have experienced diagenesis through K-metasomatism need to be corrected using an A-CN-K (Al2O3–CaO* + Na2O–K2O) plot where molar values of Al2O3, (CaO* + Na2O), and K2O are plotted on a ternary diagram. A typical weathering trend for sediment parallels the A-CN line, where the sediment loses Ca, Na, and K as it becomes more weathered from the original source rock. The level of K-enrichment, reflecting K-metasomatism and possible amorphous silica by-product, can be estimated by projecting each sample plotted on the A-CN-K ternary diagram back to the assumed, pre-metasomatized, original position (CIAcorr) along the weathering trend line (A-CN parallel line)27. Early plant colonization of landscapes in the Middle to Late Devonian increased delivery of P to the marine environment, likely increasing productivity in the upper water column and localized bottom water anoxia, but with eventual stabilization of P liberation as the vegetation becomes established, it is unknown whether these shifts cause sustained basin-wide changes in the marine environment8. Map** intervals of increased productivity (biogenic SiO2) on a basin scale will aid our understanding of terrestrial-marine teleconnections during periods of significant climatic shifts.

Conclusions

This study examines the distribution of amorphous SiO2 in the Duvernay Formation in three wells across a basin transect to determine whether an increase in productivity (e.g. radiolarians) is associated with the emergence of deeper root networks and thicker soil horizons contributing to terrestrially-derived nutrients in the oceans during the punctata Event (pE). Previous investigation of modern and ancient records for intervals of amorphous SiO2 that may be attributed to biogenic sources have relied on geochemical proxies or separated aliquots that have not captured all of the amorphous SiO227. LWIR hyperspectral imaging allows us to detect opal-A SiO2 on a macroscale, as a continuous dataset, using a non-destructive instrument. This ability to quickly and efficiently detect intervals of opal-A SiO2 means that we may start to map these intervals on a basin scale to improve paleogeographic reconstruction and map past oceanic circulation. Opal-A SiO2 identified by LWIR imaging was determined to have two sources: 1) a by-product of clay diagenesis identified using an A-CN-K plot, and 2) biogenic silica, which is the primary source of SiO2 in two of the three wells. In the case of the Duvernay in the Western Canada Sedimentary Basin, the effect of the pE resulted in increased paleoproductivity in two areas of Duvernay deposition basinward of a large barrier reef. The third area represented by the ESB well, which is located on the carbonate platform and behind several large elongated barrier reefs, shows only minor amorphous SiO2. While the A-CN-K plot suggests there may have been a minor contribution of SiEX or opal-A SiO2 from clay diagenesis, the basinward increase in SiEX and opal-A SiO2, correlation of SiO2 to P2O5, TOC, and the δ13C(org) shift associated with the pE, and detection of possible radiolarian tests or SiO2-filled algal cysts all point to a dominantly biogenic SiO2 source in the Duvernay. This finding supports theories that oceans experienced an increase in productivity as the world’s forests expanded and developed deeper root networks which increased soil genesis and delivery of terrestrially-derived nutrients into the marine realm. However, the highest concentration of SiEX, biogenic opal-A SiO2 and TOC is in the upper Duvernay of the most distal well and does not correspond to a δ13C(org) excursion. This study demonstrates the need for more studies that examine the correlation of δ13C excursions and reactions recorded in the sedimentary record and the need for differentiation of amorphous from crystalline SiO2 so that we may better understand global silica cycling through terrestrial-marine teleconnections.