In northwestern Alberta and northeastern British Columbia, upper Cenomanian and Turonian strata of the Kaskapau Formation form a thick, mudstone-dominated and northeastward-thinning wedge that spans the foredeep of the Western Canada Foreland Basin (Varban and Plint, 2005). Well-exposed sections in the Rocky Mountain Foothills on the western margin of the foredeep can be correlated with sections in the Peace River Valley, located close to the forebulge. Correlation has been established by using abundant, publicly accessible wireline log data (Fig. 2). Detailed correlation through a grid of 756 wireline logs showed that 28 allomembers, bounded by marine flooding surfaces, could be mapped across the foredeep (Varban and Plint, 2005). In the western part of the foredeep, exemplified by the section at Mount Robert (Figs. 2, 3), nearshore and shoreface sandstones form stacked successions that prograded only 20–40 km seaward from the preserved basin margin; shoreface progradation was limited by a consistently high rate of flexural subsidence (Varban and Plint, 2005, 2008). Traced eastward from Mount Robert, successive allomembers become thinner and finer-grained, and some allomembers (e.g., allomembers 7, 9, 10) pinch out completely before reaching outcrop in the Peace River Valley, exemplified by the section at Pratts Landing (Figs. 2, 3). The section at Pratts Landing, which is the focus of this study, is located on the northern bank of the Peace River (56∘01′14′′ N, 118∘48′47′′ W; Fig. 4) and comprises stacked siltier- and sandier-upward successions, capped, at a prominent flooding surface, by weakly bioturbated, organic-rich claystones and silty claystones characterized by a very high radioactivity (i.e., boundary of allomembers 6 and 8; Figs. 3, 5). Outcrop spectral gamma ray profiles allow the Pratts Landing section to be correlated with confidence to nearby wireline logs (Fig. 3).

In the west, the Cenomanian–Turonian boundary was recognized at the top of Kaskapau allomember 6 at Mount Robert, based on the distribution of inoceramid bivalves (Fig. 3). At that section, late Cenomanian Inoceramus ex  gr. pictus (Sowerby) is widely distributed through allomembers 2 to 6, whereas Mytiloides puebloensis (Walaszczyk and Cobban) is present 2 m above the allomember 6–7 contact, indicating that the lowest zone of the Turonian is present in the lower part of allomember 7 (cf. Kennedy et al., 2000). The upper bounding surface of allomember 6 can be traced, through well logs, for 220 km eastward to Pratts Landing, where it corresponds to the sharp basal surface of a gypsum-cemented silty claystone. That sharp surface, separating allomembers 6 and 8, corresponds to an abrupt increase in radioactivity, and lies 25 cm below the first appearance of the early Turonian inoceramids Mytiloides goppelnensis (Badillet and Sornay) and Mytiloides kossmati (Heinz; Figs. 3, 5). The wireline log correlation shows that, at Pratts Landing, all of allomember 7 is missing, and the earliest Turonian zone of M. puebloensis also appears to be unrepresented, emphasizing the hiatal character of the allomember 6–8 boundary (Fig. 3).

During the Late Cretaceous, the study site was located at ∼ ±5 61∘ N (van Hinsbergen et al., 2015; paleolatitude.org), on the eastern flank of the foredeep, about 160 km from the contemporaneous western shoreline of the Western Interior Seaway (Varban and Plint, 2005, 2008). We generated carbon isotope and dinoflagellate cyst data across about 23 m of upper Cenomanian to lower Turonian strata, based on stable carbon isotope stratigraphy and inoceramid biostratigraphy (Figs. 3, 6).

The carbon isotope composition of bulk organic carbon (δ13Corg) was measured at 20 cm intervals across OAE2 in order to constrain its exact position, and at 50 cm intervals for the remainder of the section. Analyses were performed in the Stable-Isotope Biogeochemistry Laboratory of the School of Geography and Earth Sciences, McMaster University, Hamilton, Ontario, Canada. In total, 77 samples were treated with 3 N HCl to remove carbonates, rinsed with demineralized water, freeze-dried and powdered. Between 1 and 3 mg of powdered sediment sample was weighed in tin capsules and then put in a rotating carousel for subsequent combustion in an elemental analyzer. After purification of the gas sample it was passed through a SIRA II Series 2 dual-inlet isotope-ratio mass spectrometer to determine the stable carbon isotopic composition of organic matter. Carbon isotope ratios were measured against an international standard, NBS-21. The analytical reproducibility, based on replicate samples, was better than 0.1 ‰.

Dinocyst abundances were determined for 21 samples, covering the entire section, using standard palynological methods. About 5 g of freeze-dried sediment was processed following a standardized quantitative method (e.g., Sluijs et al., 2003), which involves the addition of a known amount of Lycopodium marker spores (Stockmar, 1971). To dissolve carbonates and silicates, HCl (∼ 30 %) and HF (∼ 38 %) were added, respectively. After centrifugation, acids were discarded. The remaining residues were sieved over a 15 µm nylon mesh and the > 15 µm fraction was mounted on slides for analysis by light microscopy. Samples were counted to a minimum of 250 dinocysts, which were identified to genus or species level at 500× magnification, following the taxonomy of Fensome and Williams (2004). All samples and slides are stored in the collection of the Laboratory of Palaeobotany and Palynology, Utrecht University, the Netherlands. All data (δ13Corg and palynology) are listed in the Supplement (Table S1).

Originally the cysts Cyclonephelium membraniphorum (Cookson and Eisenack, 1962), which was renamed Cauveridinium membraniphorum (Masure in Fauconnier and Masure, 2004), were differentiated from Cyclonephelium compactum (Deflandre and Cookson, 1955), based on the generally higher and structurally ordered crests and membranes of C. membraniphorum. Additionally, cysts of C. membraniphorum form a series of funnel-shaped structures bordering unornamented mid-dorsal and mid-ventral areas. However, the apparent morphological variation regarding ornamentation within the two species exceeds the defined difference between the two species. Therefore, it was proposed to refer to the dinocyst morphological complex Cyclonephelium compactum–membraniphorum, rather than separating both species (Marshall and Batten, 1988). We agree that the two species are members of a morphological continuum and therefore group all these morphotypes of this continuum from our study site and the literature under the Cyclonephelium compactum–membraniphorum morphological plexus (Ccm) (Fig. 7; Table 1). For the compilation of the global biogeographical distribution of Ccm prior to, during and after OAE2, a literature survey was conducted.

At Pratts Landing the OAE2 interval is recorded between 10.2 and 16.8 m, based on a 2 ‰ positive shift in δ13Corg (Fig. 6). The Cenomanian–Turonian boundary is placed at 15.3 m, at the sharp base of a 20 cm thick, heavily gypsum-cemented silty claystone lacking macrofauna (Fig. 5). The base Turonian marker inoceramid species Mytiloides puebloensis was not found, but the succeeding inoceramid zone, characterized by M. goppelnensis and M. kossmati, starts approximately 25 cm above the basal surface (Fig. 3). Ccm is a general constituent (1–4 %) of the dinocyst assemblage at Pratts Landing throughout the section, i.e., also below the onset of OAE2 (Fig. 6).

All localities (n=35) with reported cysts of Ccm (i.e., Cauveridinium membraniphorum, Cyclonephelium membraniphorum, Cyclonephelium compactum and/or Cyclonephelium compactum–membraniphorum) are listed in Table 1 and shown in Fig. 8. The first common presence (FCP) of Ccm could only be determined for 20 of the localities as a result of poor stratigraphic constraints and only qualitative reporting of Ccm at the other 15 localities.

Recent dinocyst biostratigraphic studies from the East Coast Basin, New Zealand, show that the FCP of Ccm was ca. 500 kyr before the onset of OAE2 (Schiøler and Crampton, 2014). At northern high latitudes, notably Pratts Landing and the Norwegian Sea (Radmacher et al., 2015), Ccm is a consistent constituent of the dinocyst assemblage throughout the late Cenomanian. In contrast, at most Northern Hemisphere mid-latitude sites, Ccm has not been reported before OAE2, with the exception of a few spot occurrences at Eastbourne and Iona-1 (Pearce et al., 2009; Eldrett et al., 2014). Crucially, Ccm was never a consistent constituent of mid-latitude dinocyst assemblages before OAE2. This indicates that Ccm had a high-latitude biogeographical distribution in both hemispheres before OAE2.

Five Northern Hemisphere shelf sites in Europe and North America, namely Pratts Landing, Iona-1 (southwestern Texas, USA), Bass River (New Jersey, USA), Eastbourne (East Sussex, UK) and Wunstorf (Lower Saxony, Germany), were selected to compare established biozonation, high-resolution records of δ13C, and the relative abundances of Ccm (Fig. 9; Pearce et al., 2009; Eldrett et al., 2014; van Helmond et al., 2014a, 2015). Maximum relative abundances of Ccm (i.e., > 10 %) are recorded during the first maximum in the OAE2 characterizing carbon isotope excursion (point “A” – cf. Voigt et al., 2008), at Pratts Landing (Figs. 6, 9). At the same stratigraphic position, Ccm becomes abundant at several other Northern Hemisphere mid-latitude sites, for example the southern part of the Western Interior Seaway, the proto-North Atlantic shelf, the European shelf, and the Tethys (Figs. 8, 9; Table 1). Despite a spot occurrence at point “A”, the FCP of Ccm seems somewhat delayed at Eastbourne (i.e., Plenus Marl Bed 7 – Fig. 9; Pearce et al., 2009); this is a local phenomenon, because in other English Chalk sections (e.g., Dodsworth, 2000) the FCP of Ccm coincides with that of other Northern Hemisphere mid-latitude sites.

At Bass River and Wunstorf the FCP of Ccm precisely correlates with a drop in sea surface temperature (van Helmond et al., 2014a, 2015), leading to the suggestion that the dinoflagellate taxon that produced Ccm migrated to these sites in response to climatic cooling. We therefore suggest that sea surface temperature was the primary control on the biogeographical distribution of Ccm outside high-latitude regions. For the Shell Iona-1 core the FCP of Ccm coincides with a minimum in organic carbon, redox-sensitive elements and relatively high abundances of benthic foraminifera and trace fossils indicative of a period of improved oxygenation of bottom waters (Eldrett et al., 2014). This is in agreement with previous observations for the interval showing PCE-related cooling of sea surface temperature in the proto-North Atlantic (Forster et al., 2007; Sinninghe Damsté et al., 2010; van Helmond et al., 2014b). The sustained presence of Ccm after the PCE at all sites, except Bass River (Fig. 9), suggests that, in addition to sea surface temperature, other environmental and paleoceanographic factors became dominant in determining the distribution of Ccm once it had occupied niches at lower latitudes. For example, salinity, (enhanced) nutrient availability and proximity to the shoreline may have been important (Harris and Tocher, 2003). Preservation of palynomorphs, e.g., dinocysts, is variable within sections and between sections but is unrelated to the occurrences of certain species.

The migration of Ccm towards lower latitudes in response to cooling resembles dinoflagellate migration events during other periods of marked climatic change. Dinocysts referable to the Arctic Paleogene taxon Svalbardella were encountered in mid- and low latitudes during the most pronounced Oligocene glaciations (ca. 30–25 Ma; van Simaeys et al., 2005). In contrast, during the Paleocene–Eocene Thermal Maximum, tropical species of the dinocyst genus Apectodinium moved from low toward high latitudes in response to peak warmth (Crouch et al., 2003; Sluijs et al., 2007). Studies across the Cretaceous–Paleogene boundary indicate initial high latitude to equatorial dinoflagellate migration at the boundary, followed by a reverse migration. This presumably took place in response to impact-related initial climatic cooling followed by a return to warmer conditions (Brinkhuis et al., 1998; Galeotti et al., 2004; Vellekoop et al., 2014).

The biogeographical expansion of Ccm towards the Equator seems to be a relatively strong response to a moderate change in sea surface temperature (ca. 3–5 ∘C). The southward migration of Ccm over relatively large distances, i.e., 20–30∘ of latitude southwards, may have been amplified by the flatter meridional temperature gradients across OAE2 (e.g., Sinninghe Damsté et al., 2010). Compared to the present day, which is characterized by a much steeper meridional temperature gradient, relatively small changes in temperature in the mid-Cretaceous and early Paleogene may have had a much larger impact on the distribution of marine organisms.

Most of the Cretaceous is covered by the Normal Superchron C34n (ca. 126–84 Ma; Gradstein et al., 2012), hampering application of magnetostratigraphy. Stratigraphic correlation for the Cenomanian–Turonian boundary interval therefore relies on biostratigraphy and carbon isotope stratigraphy (Gale et al., 2005) as well as on recent advances in astrochronology (e.g., Meyers et al., 2012; Eldrett et al., 2015). Pelagic sediments are often carbonate-poor, because the calcite compensation depth was relatively shallow during OAE2, complicating planktonic foraminifer and calcareous nannofossil biostratigraphy (e.g., Erba, 2004). Consequently, carbon isotope stratigraphy is the main stratigraphic tool for OAE2 because the positive carbon isotope excursion is recognized in all active carbon reservoirs (Tsikos et al., 2004). Calibration of carbon isotope stratigraphy with bioevents is, however, essential to establish detailed stratigraphic frameworks.

The coincidence of the FCP of Ccm with the base of the W. archaeocretacea and the upper part of the M. geslinianum zones close to the first maximum in the positive carbon isotope excursion (point “A”; Fig. 9) suggests that dinoflagellate migration probably occurred within thousands to 10 000 years. The FCP of Ccm thus represents a useful biostratigraphic marker, being, to date, the only widely found microfossil to mark the PCE, except at high latitudes.