In the present day ocean, most of the water column is well ventilated as a consequence of thermohaline circulation processes that lead to subduction of cold, oxygen rich and dense water masses in high northern and southern latitudes (e.g. Kuhlbrodt et al., 2007). Exceptions are restricted basins, in which the limited exchange with the oxygen rich water masses of the open ocean is not sufficient to counteract oxygen consumption by organic matter respiration such as in the Black Sea (Murray et al., 1989). In the open ocean, strongly oxygen depleted water bodies occur underlying highly productive surface waters such as in the major upwelling areas off the western continental margins of Africa and the Americas or below the monsoon-driven upwelling of the Arabian Sea (Helly and Levin, 2004). In the geological past, regional or global ventilation of the ocean underwent significant changes on different time scales due to a variety of reasons, including changes in atmospheric and ocean circulation, stratification, temperature or tectonic processes. It is, however, difficult to quantify the past spatial extent and intensity of oxygen minima because the oxygen concentration of the water column is not directly recorded in the sediments. As a consequence, derivative proxies have been applied to reconstruct past ocean oxygenation.

A characteristic feature of marine low-oxygen environments on various time scales are black, organic-rich, and laminated sediments (Kemp, 1996; Meyer and Kump, 2008). They are known to date back to the late Precambrian (Tucker, 1983). Widespread and contemporaneous occurrences of these deposits in Devonian, Permian, early Jurassic, early and late Cretaceous, and mid-Miocene successions depict periods of sluggish ocean circulation or extensive highly productive seas (Schlanger and Jenkyns, 1976; Buggisch, 1991; Flower and Kennett, 1993; Wignall and Twitchett, 1996; Trabucho-Alexandre et al., 2010). The question of whether these laminated sediments were formed due to enhanced primary production or due to restricted ventilation of near-bottom waters has fuelled a long-lasting debate (e.g. Calvert, 1987). Yet the discovery of laminated sediments in the Arabian Sea during the International Indian Ocean Expedition in 1965 revealed that this sedimentary facies is confined to oxygen minimum zones (OMZs) at mid-depth (Schott et al., 1970). Laminated sediments at the south-west African, Peruvian and Californian margin provided further evidence for their association with today's OMZs (van Andel, 1964; Reimers and Suess, 1983; Struck et al., 2002). In contrast, basin-wide stagnation events resulting in the deposition of organic-rich, at least partially laminated sediments were recorded during short time intervals with specific environmental settings from the Pliocene to early Holocene in the eastern Mediterranean (sapropels) and in the Sea of Japan (Stein and Stax, 1990; Rohling and Hilgen, 1991). They are, however, not considered potential analogues for the extensively occurring black, laminated shales of the Mesozoic including the Cretaceous ocean anoxic events (OAEs; Erbacher et al., 2001).

Stable carbon isotope data obtained from Devonian, Toarcian, Aptian and Cenomanian–Turonian successions revealed that the organic-rich beds recorded profound perturbations of the global biogeochemical cycles, of which the Cenomanian–Turonian boundary interval (OAE 2) was probably the most extensive event (e.g. Jenkyns et al., 1994; Hesselbo et al., 2000; Joachimski et al., 2002; Herrle et al., 2004). Detailed investigations of the geochemistry, microfossil assemblages, and sedimentary structures of both recent and fossil strata were performed to unravel the interplay of local, regional and global processes driving their formation, and to enforce a mutual understanding of late Quaternary OMZs and Cretaceous OAEs (Thiede and Suess, 1983; Anbar and Rouxel, 2007; Dale et al., 2012; Owens et al., 2013). These studies were complemented by oceanographic, biological and biochemical studies in Recent upwelling systems and OMZs. However, this actualistic approach has been hampered by long periods of burial, diagenesis, and evolution of the biosphere since their deposition in Mesozoic times.

Geochemical redox proxies were extensively explored on Cretaceous black shales in order to constrain past ocean oxygenation, in particular trace metals (Brumsack, 2006; van Bentum et al., 2009; Dale et al., 2012), sulfur isotopes (Hetzel et al., 2009; Owens et al., 2013), and iron isotopes (Owens et al., 2012). Some of these proxies have also been measured on surface sediments and sediment cores from the Peruvian OMZ, in particular U/Mo ratios (Böning et al., 2004, Scholz et al., 2011, 2014a, b). However, only a very few data points are available for a regional U/Mo – bottom-water oxygen calibration in the Peruvian OMZ (Scholz et al., 2011). They strongly differ from corresponding data obtained from other OMZs (McManus et al., 2006). This hampers a quantitative reconstruction of past oxygenation with U/Mo ratios for the Peruvian OMZ as well as for black shales from the Cretaceous OAE 2.

Besides geochemical redox indicators, there are only a few other reliable parameters that have been sufficiently explored to investigate palaeo low-oxygen conditions in the Mesozoic and Cenozoic, which are trace fossils, laminations, and organic carbon accumulation rates. Their potential, constraints, and implications for an assessment of past water column oxygenation are addressed in this review. Particular emphasis is put on the comparison of Holocene OMZ sediments from the upwelling area off Peru with deposits of Cretaceous OAE 2 from the Moroccan shelf.

The Peruvian Margin study is based on stratigraphic and sedimentological data from 136 sediment cores within and below today's OMZ off the western South American continental margin. They are located between the Equator and 18∘ S and were retrieved from water depths between 180 and 2200 m (Fig. 1). Data of 94 cores were taken from the literature and 42 new cores recovered during R/V METEOR cruises M77-1 and M77-2 in 2008 were assessed as part of this study (Table A1 in the Appendix). The cruises were performed in the framework of Collaborative Research Centre (SFB) 754 “Climate Biogeochemistry Interactions in the Tropical Ocean”, through which supplementary data for the environmental interpretation of the sedimentary records are available.

In particular, oxygen concentrations along the Peruvian continental margin were measured during R/V METEOR cruises M77-1, M77-2 (Krahmann, 2012) and M77-3 (Kalvelage et al., 2013). We considered 159 CTD stations with a maximum water depth of 1750 m and a maximum distance of 175 km to the shore (online Supplement M77-1-3_CTD_Data.xls). Other CTD casts further offshore were not included because they already showed significantly elevated oxygen concentrations compared to proximal locations at the same latitude. Kalvelage et al. (2013) observed an offset of ∼ 2 µmol kg-1 between the CTD attached optode oxygen sensors and the more sensitive STOX sensors, which are based on a Clark-type oxygen sensor, and corrected the optode data by 2 µmol kg-1 during the cruises M77-3 and M77-4. The oxygen data presented in our study were not corrected for a 2 µmol kg-1 offset since the STOX sensors were not deployed during M77-1 and M77-2.

We considered visual core descriptions, physical property data, in particular dry bulk densities, sand content and abundances of biogenic, terrigenous and diagenetic components, and organic carbon contents. The chronostratigraphy of the cores was established with radiocarbon datings on monospecific samples of planktonic or benthic foraminifera, or bulk sedimentary organic carbon.

The age models of the cores from M77-1 and M77-2 cruises are based on Mollier-Vogel et al. (2013). Otherwise, published, conventional radiocarbon ages and new 14C accelerator mass spectrometer (AMS) datings were calibrated using the software “Calib 7.0” (Stuiver and Reimer, 1993) and by applying the marine calibration set “Marine13” (Reimer et al., 2013). Reservoir age corrections (ΔR) were carried out using the Marine Reservoir Correction Database (http://calib.qub.ac.uk/marine/). Regionally weighted mean ΔR values ranged from 89 to 338 years for the eastern Pacific off Peru. The uncertainties of the source data ranged from ±31 to ±82 years (2-sigma). For the pre-Holocene part of the records, the radiocarbon-based chronologies were supplemented with planktonic and benthic oxygen isotope curves correlated to stacked reference records (e.g. Liesicki and Raymo, 2005) or Antarctic ice cores (e.g. EPICA Community Members, 2006). Subrecent sedimentation rates were constrained by 210Pb excess activity profiles (Reimers and Suess, 1983; Mosch et al., 2012). All ages are given in calendar years before 1950 AD (abbreviated as cal. ka). Organic carbon and bulk sediment accumulation rates (g cm-2 kyr-1) were calculated from linear sedimentation rates (cm 10-3 years) and bulk dry densities (g cm-3) following van Andel et al. (1975).

The M77-1 and M77-2 cores included in this study were described immediately after opening aboard R/V METEOR (Pfannkuche et al., 2011). Two parallel series of volume-defined samples were taken in 5 or 10 cm intervals with cut-off syringes. One series of 10-cc samples was freeze-dried and physical properties were determined from wet sample volumes and the weight loss after drying applying standard protocols and a pore-water density of 1.026 g cm-3 (Boyce, 1976). The other series of 20-cc samples dedicated to isotopic measurements, microfossil, and sand-fraction examination was washed gently with tap water through a 63 µm sieve within a few hours after sampling. Washing of fresh, wet samples facilitates a better preservation of delicate calcareous microfossils, which otherwise may have been corroded or even dissolved by oxidation products of ferrosulfides and labile organic matter (Schnitker et al., 1980). The residues were dried at 50 ∘C and weighed. For stable oxygen and carbon isotope analyses, about 30 specimens of the planktonic foraminiferal species Globigerinoides ruber (white), Neogloboquadrina dutertrei or 3–6 specimens of the benthic species Uvigerina striata, U. peregrina or Globobulimina pacifica were picked from the size fractions 250–355 µm or > 63 µm, respectively. We used these species because they were abundant in the cores studied and incorporate stable oxygen isotopes in equilibrium with the surrounding pore and supernatant bottom waters (e.g. McCorkle et al., 1990). Oxygen and carbon isotopes were measured with a Thermo Fisher Scientific 253 Mass Spectrometer coupled to a CARBO KIEL automated carbonate preparation device at GEOMAR, Kiel. The long-term analytical precision (1-sigma) for δ18O and δ13C was better than 0.06 and 0.03 ‰ on the VPDB scale, respectively, based on more than 1000 measurements of an in-house carbonate standard during the respective measurement sessions. Replicate measurements of benthic foraminifera from the same sample showed an external reproducibility of ±0.1 ‰ for δ18O. For radiocarbon analyses, 33–179 specimens of Planulina limbata or 229–250 specimens of Neogloboquadrina dutertrei were picked from the size fraction > 63 µm or 5–20 mg of ground bulk sediment was prepared. AMS measurements were performed at the Leibniz Laboratory for Radiometric Dating and Stable Isotope Research, University of Kiel (CAU) and at Beta Analytic Inc. Dried samples used for physical property measurements were ground with an agate mortar. Aliquot subsamples of 3–20 mg were analysed for total carbon and organic carbon content with a Carlo-Erba Element Analyzer (NA1500) at GEOMAR, Kiel. The long-term precision was ±0.6 % of the measured values as revealed by repeated measurements of two internal carbon standards.

Since the pioneering work of Einsele and Wiedmann (1975), Cenomanian to Lower Campanian organic-rich marlstones of the Tarfaya Basin in southern Morocco have been studied as a type locality of Cretaceous upwelling-related sediments at the eastern margin of the central North Atlantic (Wiedmann et al., 1978; Leine et al., 1986; Kuhnt et al., 1997, 2001, 2005; Kolonic et al., 2005; Aquit et al., 2013). Numerical climate and circulation models of the mid-Cretaceous Atlantic support a prevalence of cool and nutrient-rich intermediate deep water masses in this area along the north-west African margin (Poulsen et al., 2001; Topper et al., 2011). The late Cenomanian to early Turonian OAE 2 sediments discussed here were examined in outcrop sections during five field expeditions of the Kiel Micropaleontology Group in 1997, 1998, 2000, 2003 and 2009. In addition, core material from two commercial wells (S13 and S75), and a 350 m deep research well drilled in October–December 2009 (Tarfaya SN∘ 4) were considered in this study (Fig. 1).

Analytical methods applied to samples from outcrops and drill cores were detailed in Kuhnt et al. (2005) and Aquit et al. (2013). Core sections from the new exploration well Tarfaya SN∘ 4 were cut lengthwise and described. Line scan measurements and photographs were acquired with a Ja CVL 1073 CCD colour line scan camera with 3 sensors of 2048 pixels and a dichroic RGB beam splitter prism (RGB channels at 630, 535 and 450 nm) at the Institute of Geosciences, Kiel University. Colour measurement in L*a*b* units are from RGB digital images. Scanning was performed (resolution of 143 pixel per 70 micron) on the polished surface of oriented cores. Intensity of lamination vs. bioturbational homogenisation of the sediment was estimated using high-resolution lightness ( L*) measurements for cores of SN∘ 4. We calculated a lamination index based on a moving window standard deviation of the lightness values, similar to the method previously applied on core S75 (Kuhnt et al., 2005). Organic carbon and carbonate contents of SN∘ 4 core samples were measured with a Carlo-Erba Element Analyzer (NA1500) at GEOMAR, and with a conventional carbonate bomb at the Institute of Geosciences, Kiel University.

The Pleistocene to Holocene and late Cenomanian to early Turonian stages are more than 94 million years apart in Earth's history. A direct comparison of their sedimentary record and environmental processes is hampered by burial diagenesis, evolution of marine biota, different continental and ocean configurations, and different climates, ocean circulation, and biogeochemical cycles. The late Cenomanian was marked by the onset of OAE 2. Sedimentological, faunistic and biogeochemical parameters suggested that large parts of the water column were devoid of dissolved oxygen, but the absolute levels are less well constrained. In an actualistic approach, we compared deposits of OAE 2 from the Moroccan shelf close to Tarfaya with deglacial and Holocene OMZ sediments from the upwelling area off Peru and found only a few parameters for a reliable investigation of palaeo low-oxygen conditions in both records, i.e. trace fossils, laminations and organic carbon accumulation rates.

The millimetre-sized trace fossil Chondrites was common in Cretaceous strata, in particular in the beds directly underlying OAE 2 black shales. It was also found in modern oxygen-depleted environments, where it is created by a nematode pursuing chemotrophy in anaerobic conditions. The burrow diameter increased with oxygen level from 5 to 45 µmol kg-1 in the San Pedro Basin, California. However, Chondrites has never been reported from Peruvian OMZ sediments. The oxygen–burrow size relationship is challenged by centimetre-sized crab burrows appearing at oxygen levels around 10 µmol kg-1 below the OMZ already. Crab burrows are also common in Cretaceous sediments. Their appearance in OAE 2 sediments may therefore indicate that a threshold of approximately 10 µmol kg-1 bottom-water oxygen has been exceeded.

Laminations are a more reliable indicator, but they display only one, very low oxygen level. In the Peruvian, north-eastern Pacific and Pakistan OMZs, depositional laminae created by seasonal or multi-annual variations in sediment supply or composition were preserved at bottom-water oxygen concentrations of less than 5 µmol kg-1. Coherent occurrences of laminated beds and biogeochemical indicators for oxygen drawdown in Tarfaya OAE 2 sediments supported the applicability of this feature for bottom-water oxygen estimates. The cyclic pattern of laminated and non-laminated intervals in Tarfaya sections and in sediment cores from the eastern Pacific suggested the impact of climatic variations with direct linkages to the high-latitude Southern Ocean as a source of nutrients and better ventilated intermediate waters. This regular cyclic pattern is blurred in Peruvian OMZ sediments by erosion, omission and redeposition due to near-bottom currents and breaking internal waves, making the preservation of laminated sediments an exception rather than the rule. Redeposition features were also observed in Tarfaya outcrop sections and reveal episodic, strong currents on the Cretaceous shelf and upper slope as an important process that was likely responsible for many observed unconformities in upper Cenomanian and lower Turonian formations.

Organic carbon accumulation rates of late Holocene sediments off Peru displayed a disjunct pattern. They showed a high scatter and a broad abundance maximum between 0.8 and 2.8, mode value at 1.3 µmol kg-1, in laminated sediments under a Recent bottom-water oxygenation of < 5 µmol kg-1. If we compare the carbon accumulation rates of the Tarfaya OAE 2 laminated sediments with late Holocene to Recent ones from the Peruvian OMZ, the Cretaceous rates between 0.7 and 2.8 g C cm-2 kyr-1 match the data range of the majority of late Holocene sediments very well. Taking into account the high burial efficiency of organic carbon deposited in OMZs, and calculating deposition fluxes from the photic zone, the maximum Cretaceous values would account for only half of the present-day export production under the Peruvian upwelling. Thermal maturation or the loss of volatile hydrocarbons from Tarfaya black shales may well account for this difference. Maximum Holocene carbon accumulation rates off Peru compare well to the present-day export production. This agreement is, however, valid only for sediments with a continuous, laminated record. All other cores exhibiting average carbon accumulation rates have most likely been subjected to instant winnowing and redeposition of organic detritus.

At higher oxygen levels, organic carbon accumulation rates showed an inverse exponential relationship with oxygen concentrations. This mirrors the successive bioirrigation and concomitant decomposition of organic matter through increasingly better ventilation below the Peruvian OMZ. Such a relationship has not been described before. Few available data from the Arabian Sea suggested a similar covariance conferring credibility to the pattern observed at the Peruvian Margin (Koho et al., 2013). The relationship has been used to assign a palaeo oxygen level to a well constrained, intermittently oxygenated interval at the base of cycle 0 (named the Plenus Cold Event) in the Tarfaya sections. The estimate of 38 µ mol kg-1 disagrees overall with the composition of the Cretaceous and Recent benthic foraminiferal assemblages prevailing at this oxygen level.

In summary, close similarities and distinct differences between the two periods of low oxygenation in the sedimentary record of the Cretaceous OAE 2 and the late Quaternary OMZs were recognised. More data are required to further constrain the organic carbon accumulation–oxygen relationship. This emerging palaeoproxy has to be complemented and corroborated by other, advanced bottom-water ventilation proxies, e.g. molybdenum isotopes or I / Ca ratios in foraminiferal shells in order to achieve more quantitative reconstructions of past oxygen levels and their controlling factors.