Evaluating the role of coastal hypoxia on the transient expansion of microencruster intervals during the early Aptian

the role of coastal hypoxia on the expansion of Worldwide, a growing number of modern coastal marine ecosystems are increasingly exposed to suboxic ‐ or even anoxic conditions. Low seawater oxygen levels trigger signi ﬁ cant ecosystem changes and may result in mass mortality of oxygen ‐ sensitive biota. The applicability of observations from recent (anthropogenically in ﬂ uenced) suboxic coastal settings to fossil anoxic shallow ‐ marine environments is, however, as yet poorly explored. The test case documented here are upper Barremian to lower Aptian strata in the Lusitanian Basin (Ericeira section, Portugal). These are characterized by the tran- sient demise of rudist – coral communities and the rapid establishment of microencruster facies in the vacant ecological niches. The hypothesis is tested that the temporal expansion of the microencrusting organism Lithocodium aggregatum took place in response to platform ‐ top seawater oxygen depletion. We critically discuss the outcome of a multi ‐ proxy palaeoseawater redox approach (e.g. Rare Earth Elements (REEs), U isotopes and palaeoecology) and put the robustness of the proxies applied here to the test. This is done by considering issues with these methods in general but also empha-sizing the signi ﬁ cance of terrigenous contamination and fractionation effects. Data shown here document that evidence for coastal seawater oxygen depletion in the pre-lude of Oceanic Anoxic Event (OAE) 1a is lacking, and hence, anoxia was not the driv- ing mechanism for the demise of rudist – coral ecosystems in the proto ‐ North Atlantic platform setting studied here. In contrast, well ‐ oxygenated early Aptian platform ‐ top water masses are proposed for this site. Geologically short (decades to millennia) ﬂ uc-tuations in seawater oxygen levels cannot be excluded, however. But even if these took place, they offer no explanation for the Kyr to Myr ‐ scale patterns discussed here. The present paper is relevant as it sheds light on the complexity of mechanisms that drive punctuated Early Cretaceous coral – rudist ecosystem turnover, and assess strengths and weaknesses of redox proxies applied to ancient shallow ‐ marine platform carbon- ates. □ Anoxia, cerium anomalies, Cretaceous, Oceanic Anoxic Event 1a, redox proxies, uranium isotopes.

Oxygen is a key element for the metabolism of most marine organisms. Seawater oxygen depletion is a threat, particularly to shallow coastal areas, and indeed, the number of seasonal to permanent oxygen-depleted coasts has dramatically increased during the past 50 years (Diaz & Rosenberg 2008). The growing number of coastal hypoxic ecosystems is linked to higher rates of anthropogenic nutrient input, enhanced organic matter production and eutrophication (Rabalais et al. 2010). Hypoxia has a variety of effects on population structure and community composition (Rabalais & Turner 2001;Levin et al. 2009). Severe hypoxia can culminate in the creation of 'dead zones' (Rabalais et al. 2010) and mass mortality of many marine species. Nevertheless, organism responses differ among taxa, depending on the duration and degree of oxygen depletion (Rabalais & Turner 2001) and the mobility of the organism considered. Further, independent of the recently observed anthropogenic patterns, seawater hypoxia or even anoxia are well-known phenomena throughout Earth's history (Schlanger & Jenkyns 1976; Jenkyns 2010; Rabalais et al. 2010;Dickson et al. 2012).
The approach followed here applies tools and concepts recently tested in the Central Tethyan realm (Kanfanar section, Croatia; Hueter et al. 2019). These tools include detailed fieldwork, carbonate microfacies analysis and SEM analysis, but also involve stateof-the-art geochemical seawater redox proxies such as cerium (Ce) anomalies and uranium (U) isotope ratios. By applying these tools, we intend to critically test their robustness with regard to potential near-coastal terrigenous contamination and/or fractionation effects. In the Central Tethys, transient stages of platform-top hypoxia have been shown to act as main driving mechanism for rudist-coral decline and microencruster expansion. There, these events are coeval with OAE 1a and related organic-rich pelagic deposits (black shales) documenting basinal anoxia (Hueter et al. 2019). This is of relevance for the proto-North Atlantic case example tested here, as lower Aptian shallow-water limestones of the Lusitanian Basin (Portugal) were deposited prior to the onset of OAE 1a (Heimhofer et al. 2007;Burla et al. 2008;Huck et al. 2012) but show microencruster intervals (correlated between the Ericeira, Cresmina and Sao Juliao sections) comparable to those in the Central Tethyan realm.
Having established the relation between microencruster facies and syn-OAE 1a platform-top seawater oxygen depletion for the Central Tethyan realm (Hueter et al. 2019), the research question raised here is, if the decline of coral-rudist assemblages and the coeval mass occurrence of these microencrusters in these proto-Atlantic sections represent a more general, palaeoecological pattern that can be linked to low seawater dissolved oxygen levels in general? Moreover, if the hypothesis 'microencruster blooms are a proxy for platform-top anoxia' holds true, then the Portugal section would provide evidence that, at least in some regions, low dissolved seawater oxygen levels established prior to OAE 1a.

Geological setting
During the Early Cretaceous, the Lusitanian Basin was located along the western Iberian margin at a palaeolatitude of 25 to 30°North (van Hinsbergen et al. 2015). At this time, the tectonic regime changed from rifting to drifting and the shelf areas were affected by local subsidence (Hiscott et al. 1990;Rasmussen et al. 1998), resulting in flooding of the basin during Barremian-Aptian times (Wilson et al. 1989;Dinis et al. 2008).
The studied coastal Ericeira outcrop is located north of Lisbon (38°57′39.62″N/9°25′11.16″W), in the centre of the town of Ericeira (Fig. 1A, B). The Ericeira section is built by~60 m of mixed carbonate-siliciclastic sedimentary rocks of late Barremian to early Aptian age (Burla et al. 2008). Three lithostratigraphical units can be distinguished: the Regatão, the Cresmina and the Rodízio formations (Rey et al. 2003). The Cresmina Formation is subdivided into Cobre, Ponta Alta and Praia da Lagoa Members. The mixed carbonate-siliciclastic deposits of the Cobre Member (Barremian) are overlain by thickly bedded rudist-bearing carbonates of the Ponta Alta Member (Aptian), composed of massive grainstone beds with numerous rudist bivalves, nerinid gastropods, madreporid corals and abundant L. aggregatum facies.
The Lithocodium-bearing interval of the Ponta Alta Member was studied in great detail and is composed of oncoidal floatstones, columnar-patchy boundstones and a prominent, 0.4 m thick L. aggregatum patch reef facies (see 'pinnacle interval' of Huck et al. 2012). The latter grade laterally into fine-grained, quartz-rich, orbitolinid grain-to packstones and marls. The stratigraphically overlying Praia da Lagoa Member is composed of orbitolinid-rich, oyster-bearing marls and limestones, characterizing a protected lagoonal setting (Rey et al. 2003;Dinis et al. 2008;Huck et al. 2014). Above, the Praia da Lagoa Member is unconformably overlain by coarse-grained siliciclastics of the Rodízio Formation, marking the transition towards continental conditions. Based on observations of palynomorphs (Heimhofer et al. 2007), rudists and orbitolinids (Masse & Chartrousse 1997;Skelton & Masse 1998;Burla et al. 2008), the Ericeira section has been assigned to the early Aptian. This result is supported by carbon-and strontium-isotope chemostratigraphy (Burla et al. 2008;Burla et al. 2009;Huck et al. 2012

Fieldwork and thin-section microscopy
Although our study builds on the data foundation published in Huck et al. (2012), we have significantly refined the sampling and data framework. Hence, a detailed lithological log of the key interval of the stratigraphical section was generated (Fig. 2). Eleven samples (ER1 to ER11; Fig. 3) were studied with the aim to identify changes in the biotic composition and to obtain the best possible material for geochemical analysis.
Stratigraphical intervals characterized by normal marine biota composition (corals, rudists, echinoderms, gastropods) were separated from intervals yielding the abundant (partly rock-forming) microencruster facies that potentially signifies lowered seawater oxygen levels (Fig. 3). By means of thin-section analysis, skeletal components were assessed for their frequency distribution in a semiquantitative manner applying the following nomenclature: 1, very abundant; 2, abundant; 3, rare; 4, very rare; and 5, absent, in order to describe the relative volumetric significance of different components. Special focus was on the presence of L. aggregatum and its corresponding growth forms. For a discussion on the taxonomy of this organism, the reader is referred to Schlagintweit et al. (2010). For reasons of simplicity, we refer to L. aggregatum dominated carbonates as 'microencruster' facies throughout this study.

Scanning electron microscopy
A total of six samples were examined using a highresolution field emission scanning electron microscope (HR-FESEM type Zeiss Merlin Gemini 2) with a high electrical voltage of 5-10 kV at RUB. Prior, samples were cut into small cubes, glued onto glass slides and coated with gold. The freshly broken surface facing towards the top was analysed with scanning electron microscopy (SEM) and energydispersive X-ray spectroscopy (EDS) in order to detect indication for diagenetic overprint. Furthermore, we aimed to identify microbially mediated automicrite (terminology sensu Neuweiler & Reitner 1992), because the geochemical palaeoredox analyses performed here require significant portions of in situ deposited fine-grained carbonates (0.5-1 g).

Geochemical methods
For 'Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry' (LA-ICP-MS), thick sections were prepared at RUB. Rock samples were examined with a magnifying glass, and a suitable part was marked for further preparation. They were then sawn into small blocks, in order to generate glass pellets for LA-ICP-MS analyses. Analysis were performed on eleven samples with a 213 nm, Q-switched, Nd: YAG laser from New Wave, connected to a Thermo Finnigan ELEMENT2 sector field (SF) ICP-MS located at the Max Planck Institute for Chemistry in Mainz, Germany. The main focus of the analysis was on obtaining data for the calculation of the cerium anomaly (Ce/Ce*). The Ce/Ce* ratio is used as a proxy for the degree of seawater oxygenation (Tostevin et al. 2016). Measurements include Rare Earth Elements (REEs) and redox-sensitive trace elements. Every sample was measured at three spots distributed over two different portions of the thick sections.
Standard settings of the laser system were the following: (1) an energy density of 7 J/cm 2 ; (2) a pulse rate of 10 Hz; (3) the instrument was tuned to achieve low oxide production rates (ThO/Th < 0.5 %); (4) laser spot diameter size of 60 -120 µm; and (5) wash out, blank count rate and ablation times of 45 s, 20 s and 110 s, respectively. We used NIST-612 and MACS-3 for external standardization and 43 Ca as an internal standard, to calculate absolute element concentrations from signal intensities. Rare Earth Element abundances were normalized to the Post-Archaean Shale Standard (PAAS) values given in Barth et al. (2000). Ce anomaly (Ce/Ce*) was defined following Nozaki's (2008) calculation: where 'N' stands for shale-normalized concentration.
For uranium isotope analysis, sample powder was drilled from fresh carbonate sample surfaces with focus on mainly automicrite (for terminology see Hueter et al. 2019), as observed and identified during thin-section and SEM analysis. All eleven samples for U isotope analysis were weighted before digestion. For the carbonates, about 600 mg up to 1 g was digested with 6 M HCL in 15 ml Savillex beakers on a hot plate at 100°C and dried overnight. Afterwards, samples were treated with 10 ml of 1.5 M HNO 3 and centrifuged after 15 min at ultrasonic bath. The residue was quantitatively transferred into 90 ml Savillex beakers, treated with a mixture of 800 µl conc. HNO 3 and 3 ml conc. HCL (aqua regia), boiled up at 120°C for 2 hours and evaporated afterwards. All samples were dissolved in 2-5 ml of 3 M HNO 3 and subsequently spiked prior to chemical separation of U from the matrix with the IRMM 3636-A ( 236 U/ 233 U = 0.98130) mixed isotopic tracer in order to correct for any isotope fractionation on the column as well as instrumental mass bias of the MC-ICP-MS (Weyer et al. 2008;Romaniello et al. 2013;Noordmann et al. 2015).
Uranium separation from the host sample matrix was performed by chromatographical extraction, using Eichrom UTEVA resin (Horwitz et al. 1993;Weyer et al. 2008). Uranium isotope analyses were performed with a ThermoScientific Neptune Mass Collector-Inductively Coupled Plasma-Mass Spectrometer (MC-ICP-MS) at the Leibniz University Hannover, Germany. Analyses were performed using a Cetac Aridus II combined with a 100 µl PFA nebulizer for sample introduction and a standard Ni sampler cone and a Ni X skimmer cone. With this setup, a 50 ppb U solution was sufficient to achieve a 36 V signal on 238 U (Weyer et al. 2008;Noordmann et al. 2015). Every sample was measured three times and results represent mean values (2s.d. typically below 0.1‰). All U isotope variations are reported relative to the U isotopic composition of the CRM112a standard. The results of the isotope measurements are provided in the delta notation: The U enrichment factors (U EF ) were calculated to evaluate the terrigenous influence on the U isotope ratios in terms of a detritus correction, using the following notation: As standard U concentration, the mean concentration of the continental crust (1.3 ppm; Rudnick & Gao 2003) was used. For the standard Al, the continental crust concentration of 83.000 ppm (Greenwood & Earnshaw 1997) was used.
The lanthanum anomaly (La anomaly) was calculated following the established protocol of Bau & Dulski (1996), to test whether the Ce anomaly values are genuine or an artefact caused by elevated or depleted amounts of lanthanum. The calculation of Pr/Pr* is given by 2Pr N / (Ce N + Nd N ). The calculation of Ce/Ce* is explained in the LA-ICP-MS chapter. For the La anomaly, Ce/Ce* and Pr/Pr* are plotted against each other.
The yttrium/holmium (Y/Ho) ratio was calculated using the element concentrations of the LA-ICP-MS measurements. This ratio acts as an indicator for terrigenous contamination, where ratios close to 30 indicate contamination and ratios around 60 indicate normal seawater signals (Bau & Dulski 1999;Chen et al. 2015).
Lower normal marine facies. -The lowermost part of the section (0-0.7 m) is composed of moderately bioturbated bioclastic grainstones containing abundant bivalve and gastropod shell fragments and occasionally orbitolinid tests. Two prominent,~20 cm thick, orbitolinid packstone layers are present at section metres 0.4 and 0.7 (Figs 2, 4B). In the overlying pack-to grainstone interval (0.7-1.4 m), the degree of bioturbation increases significantly. At section metre 1.4, a karstified bedding plane with isolated madreporite corals is present (Fig. 4A, C). There, abundant Thalassinoides burrows occur and are often filled with a fine-grained orbitolinid packstone. The strongly bioturbated interval is followed by a 0.7 m thick rudist-and foraminifera-rich grainstone bed (1.5-2.2 m). Slightly above (at 2.3 m), the first nodular L. aggregatum occurrence is observed, with individual nodules around 1 cm in diameter. These rocks are also composed of rudist-, bivalve-and foraminifera-rich grainstones up to section metre 2.5.
L. aggregatum facies. -Between section metre 2.5-2.75, L. aggregatum oncoids increase in abundance and size, reaching up to 2 cm in diameter. Stratigraphically further upsection, patchy L. aggregatum morphotypes (2.8-3.4 m) embedded in higher-energy floatstones are found. This interval is furthermore characterized by abundant cm-to dm-sized, L. aggregatum oncoids often with ameboidal morphologies. Intercalated individual madreporite corals are preserved in situ and reach up to 15 cm in diameter. Between section metres 3.4 to 4.4, L. aggregatum boundstone facies with isolated occurrences of dmsized corals in growth position encrusted by L. aggregatum is present. This microencruster boundstone facies evolves gradually, with increasing abundance of L. aggregatum upwards. Furthermore, scattered pockets filled with peloidal orbitolinid grainstone and other benthic foraminifera are common. The dominant growth forms of L. aggregatum are cm-to dm-sized flat patchy fabrics. Section metres 4.4 to 4.8 are characterized by the presence of L. aggregatum pinnacles (Fig. 4D, 4E). Thin sections reveal the typical cauliflower-like growth morphologies (diameters of 1-20 cm) expanding both up-and sideward. The microencruster network shows mm-thin ferrousmanganese crusts (Fig. 5A) and is surrounded by a poorly consolidated partly dolomitized (Fe-rich) limestone matrix (Fig. 4E) (often) bearing abundant orbitolinids. Occasionally, corals serve as substrates for the formation of L. aggregatum crusts. These display well-defined layers, partly with micrite-, and sparite-filled internal cavities. Crusts of L. aggregatum, as well as other biogenic components are pervasively bored by lithophage bivalves (Gastrochaenolites), endolithic sponges (Entobia) and other, taxonomically poorly constrained, microborers. The highest abundance of automicrite facies is present in the pinnacle interval ( Fig. 5A/C). Nevertheless, facies containing patches of automicrite is also present beneath and above the intervals dominated by L. aggregatum, but is volumetrically less significant. Above the pinnacle interval (4.8-6.1 m), L. aggregatum oncoids are more complex and often irregular, showing ameboidal fabric growth forms (Fig. 2). In this part of the section, the skeletal remains of biomineralizing marine organisms are scarce.
Upper normal marine facies. -At 6.1 m, a hiatal surface marks a change in biofacies (6.1-7.1 m; coral lagoon facies) characterized by the return of abundant corals and orbitolinids, as well as minor volumes of bivalves and echinoderm fragments (spines and ossicles). Madreporite corals in life position ( Fig. 4F; 6.4 m) are hemispherical and massive, with diameters ranging from 20 to 30 cm. The lithofacies is a coarse orbitolinid-rich pack-to grainstone. Between section metres 7.1 and 7.6, higher-energy packstones to floatstones with rudists, corals, bivalves and benthic foraminifera are present, as well as microbial facies in the form of L. aggregatum oncoids forming a volumetrically subordinate component of the rocks. Carbonates in section metres 7.6-8.5 are classified as coral lagoon facies with abundant echinoderms (7.7-7.8 m), madreporite corals in life position and minor occurrences of L. aggregatum (8-8.5 m). Dimensions of the madreporite corals gradually decrease (8-8.5 m) upsection. A regional unconformity separates the Ericeira section at its upper boundary from siliciclastic deposits of the Albian Rodízio Formation, heralding the transition towards continental conditions.

Rare Earth elements (REEs)
Values of Ce anomaly range between 0.8 and 1.0 ( Fig. 6A) throughout the Ericeira section. The Ce anomaly values are gradually decreasing from the base to the top of the section. The pinnacle interval (4.4-4.8 m) does not show any significant variation with regard to Ce/Ce* values. The Y/Ho ratios, considered an indicator for terrigenous influence, displays values between 25 and 45, with a clustering near ratios of 35 (Fig. 6B). The La anomaly diagram places most samples in field IIa, a feature that indicates a positive La anomaly resulting in an apparent negative Ce anomaly (Fig. 6C).
The concentration of all REEs ranges between 0 and 4 ppm, and their patterns show low variability. Samples from the lower normal marine facies (0-1.5 m; ER 1, ER 2 and ER 3) display a prominent europium (Eu) anomaly, whereas a lanthanum (La) anomaly, as typical for modern seawater, is absent. The ratio of Light Rare Earth Elements (LREEs) versus Heavy Rare Earth Elements (HREEs) is close to 1.0 (Fig. 6D). Overall, the patterns show a 'bell shape', most prominent for the samples with the highest Eu anomaly. The concentrations of La, Ce and Pr are rather similar in all samples; therefore, the Ce anomalies are very low or below analytical precision.

Uranium isotopes
From the base of the Ericeira section, to section metre 1.2, δ 238 U values of −0.2 to −0.3 ‰ are observed (Fig. 7A), that is values that are slightly enriched in 238 U relative to modern seawater (~−0.4 ‰; Weyer et al. 2008). Further upsection, a peak value of −0.1 ‰ (metre 1.5) is found. From this interval upsection, values are gradually decreasing to −0.7 ‰ up to section metre 3.3. Upsection, the δ 238 U values increase and reach values of −0.3 ‰ at section metre 5.3. From section metre 5.3-7.4 (the stratigraphically uppermost bed), the values remain constant and are slightly enriched in the heavy isotope relative to modern seawater (Fig. 7A). The U enrichment factors (U EF ) are not correlated with U concentrations (Fig. 7C). Below the pinnacle interval (0 to 4.4 m), the U EF range between 8 and 46, followed by an increase to a maximum of 79 within the pinnacle interval. The U enrichment factors then decrease to values close to 10 until section metre 6.5. For the uppermost sample (above the pinnacle interval; 7.4 m), the value increases and reaches a value of 46.
Plotting U/Al ratios (ppm/%) versus δ 238 U values, the outcome reveals that most samples with an δ 238 U value close to modern seawater display U/Al ratios close to 5. In contrast, the sample with the most negative δ 238 U value (ER 7; 3.4 m) reflects the highest U/Al ratio close to 15 (Fig. 7D). Furthermore, when plotting U concentration against the δ 238 U values (Fig. 7E) samples with the most negative δ 238 U values yield the lowest U concentrations and vice versa.

Interpretation and discussion
Origin and microtexture of fine-grained carbonates Automicrite, that is fine-grained, autochthonous carbonate with a homogeneous, equigranular nature is usually related to some form of mineral precipitation in biofilms or microbial mats (Neuweiler & Reitner 1992). The former presence of microbial mats or biofilms in fossil microbialites is documented by the layered nature forming crusts of automicrite that might be gravity-defying (Turpin et al. 2012(Turpin et al. , 2014. Microbial films or mats thicken until the supply of nutrients required for cell replication and/or extracellular polymeric substances (EPS) formation becomes critically low. Exopolymeric substances is an umbrella term that groups a wide range of biopolymers (Decho & Kawaguchi 1999). The biotic and abiotic degradation and alteration of EPS favours carbonate precipitation (Dupraz & Visscher 2005). In the case of active precipitation, EPS degradation can increase the alkalinity via liberation of Ca 2+ ions previously bond to polymers (Reid et al. 2000). Passive precipitation within biofilms may take place when the cation binding capacity limit is reached. In this case, the combination of local alkalinity levels and the availability of free Ca 2+ ions can lead to a spontaneous nucleation of CaCO 3 on the EPS matrix (Reid et al. 2003).
Automicrites are the materials of choice for the geochemical analyses applied here as the autochthonous, often mono-mineralic nature of these carbonates (Fig. 5C) can be established and proxy data have a high potential to represent the local seawater/marine pore water properties. In contrast, fine-grained detrital carbonate (Fig. 5B) is often a transported material with individual particles displaying variable geochemical signatures typifying a variety of biogenic and abiogenic sources (Turpin et al. 2012(Turpin et al. , 2014. Hence, detrital fine-grained carbonate was not further considered here. Oxygen depletion: strategies of recent marine organisms and application to palaeoecology Many marine organisms have developed strategies to cope with hypoxic conditions. Studies dealing with recent settings document that many organisms are able to detect decreasing dissolved oxygen concentrations on a molecular level (Wu 2002). Pending that these organisms are highly mobile (e.g. fish or portunid crabs), these biotas move into shallower settings to find more oxygen-rich waters and to escape the oxygen-depleted bottom water layer (Breitburg 1992;Wu 2002;Ekau et al. 2009). Less mobile species, such as lobsters and echinoderms, crawl to the top of structures rising above the seafloor to avoid hypoxic to anoxic bottom waters (Burd & Brinkhurst 1984). Some echinoderms are known to temporarily change their metabolism to an anaerobic mode, enabling them to survive transient periods of severe hypoxia (Vaquer-Sunyer & Duarte 2008). Burrowing organisms (e.g. worms and shrimp) will emerge onto the sediment surface to escape the rising anoxic horizon (Baden et al. 1990;Nilsson & Rosenberg 1994).
With respect to sessile organisms such as for example bivalves, these are able to change their body postureat least to some degreeto reach higher into the water column by extending their siphon (Jørgensen 1980). Another strategy is to increase the water flow over respiratory structures and to maintain the flux of oxygen by increased ventilation and pumping rates (Warren 1984;Petersen & Petersen 1990). This is, however, coupled with a higher-energy demand and exerts significant stress. Some species reduce their energy demand by either slowing down their metabolic activity, reducing locomotion or increasing the efficiency of metabolic processes (Gracey et al. 2001). In general, specific responses to hypoxia are dependent on the remaining oxygen concentration and the duration of oxygen depletion (Vaquer-Sunyer & Duarte 2008), as well as the presence of several previous periods of hypoxia (Altieri 2006).
With reference to Cretaceous oceans, Hueter et al. (2019) studied the lower Aptian Kanfanar shallowwater carbonate platform section in Istria (Croatia) characterized by a biological turnover that shares many similarities with that observed in the Ericeira section discusses here. In the Kanfanar section, normal marine biota such as benthic foraminifera, bivalves, gastropods and corals decline significantly in abundance or disappear during an interval dominated by the microencrusting organisms B. irregularis and L. aggregatum. Hueter et al. (2019) demonstrated that platform-top hypoxic water masses triggered this biological turnover, albeit in a depositional setting (isolated platform lacking siliciclastic influx) that differed from the Ericeira one in several aspects.
As a starting point, the presence of a diverse normal marine fauna, as documented from the Ericeira section, is here taken as an indicator of normal oxygenated seawater. The abundance of orbitolinids and other benthic foraminifera, bivalves, corals, echinoderms and gastropods in the lower normal marine facies (0-4.4 m) implies well-oxygenated waters. This observation is supported by pervasive Thalassinoides bioturbation (section metre 0.75-1.4 m; Fig. 2), indicating an oxygenated upper sediment layer (Wetzel 1991;Bromley 1996). The presence of Thalassinoides is also in agreement with the coarse and carbonaterich sediments (0.75-1.4 m) typifying the restricted lagoonal facies, deposited under low sedimentation rates (Löwemark et al. 2004).
In the Lusitanian Basin, the first occurrence of L. aggregatum (2.25-3.2 m; Fig. 2) is accompanied by a significant decrease in bioturbation, here interpreted as a withdrawal of bioturbating organisms to the sediment surface, and therefore the upward movement of the redox boundary layer to the sediment-water interface (Baden et al. 1990;Nilsson & Rosenberg 1994). The prominent L. aggregatum mass occurrence as reflected by the pinnacle interval is characterized by a dense packaging of cauliflower growth forms often encrusting coral substrates. The presence of corals, organisms that are bound to normal oxygenated seawater conditions (Laboy-Nieves et al. 2001), even within the L. aggregatum dominated facies, is indicating variations in the prevailing dissolved oxygen concentrations that control the distribution of organisms. Generally, a decrease in the abundance and diversity of normal marine biota is observed whereas endolithic bivalves (Gastrochaenolites) with diameters of <1 cm and boring galleries of sponges (Entobia) are still abundant (Huck et al. 2012).
An interesting feature is that cavities in the pinnacle architecture are filled with (often dolomitized) orbitolinid wackestones and packstones. Orbitolinids are organisms that are assumed to require normal marine seawater oxygen levels. Nevertheless, it is of importance that the pinnacle interval laterally grades into orbitolinid packstones (São Julião;~3 km South of Ericeira). Given the winnowed nature of foraminifera in these cavities, it seems likely that they represent hydrodynamically transported fossil assemblages (Huck et al. 2012). Several interpretations seem likely.
Orbitolinid-rich sediments formed in more proximal settings that were not affected by whatever environmental parameters triggering the microencruster facies at the study site and were occasionally washed in during storms. These might include more distal, deeper, or coastal environments exposed to more vigorous water circulation. Alternatively, the orbitolinid facies represents slightly older (but not yet consolidated) deposits that formed along the coast prior to the pinnacle intervals. These older sediments were entrained by currents and waves, transported and deposited in cavities of the pinnacle interval. Similar patterns are observed along present-day suboxic coasts and seem rather plausible (Filatoff & Hughes 1996;Nicholas et al. 2011;Chen et al. 2016).
The pinnacle interval is characterized by conspicuous, mm-thick, reddish to brownish ferrous-manganese crusts coating microencruster surfaces (Fig. 5A). The origin of these crusts allows for multiple interpretations: Generally, the presence of iron and manganese oxide-hydroxides points to abundant dissolved oxygen in the ambient seawater. In modern oceans, manganeseand iron-rich oxide crusts are known from a series of environments that include hydrothermal sources (Toth 1980) in cold pelagic, but also warm, shallow-water environment (Tazaki 2000). Given the coastal setting of the depositional environments investigated here, hydrogenous deposition of metals from a terrigenous source is also a likely interpretation. Examples for these processes are found in the present-day Black Sea where concentration of manganese in seawater takes place under anoxic conditions, is advected towards the redox interface, and then precipitating under oxic conditions (Roy 1992). Trace-element contents in these deposits result from adsorption from seawater onto the iron and manganese colloids during advection. Last but not least, iron and manganese crusts may also form due to clay mineral diagenesis, iron and manganese remobilization, transport and deposition during burial diagenesis . Important circumstantial evidence is found when taking the orbitolinid sands, filling cavities in the microencruster bindstone fabric, into consideration. Judging from field and thin-section evidence, orbitolinid sands filled cavities after the formation of ironmanganese encrustations and provide clear evidence for a marine origin of these features, that is excluding a late diagenetic origin. Besides the presence of oxygen-dependent corals, the development of Fe-Mn crusts additionally indicates episodes of normal oxygenated seawater. Upsection of the pinnacle interval (4.8-8.5 m), the decrease in the abundance of L. aggregatum and the return of bivalves, gastropods, benthic foraminifera and oysters (especially above the transgressive surface at section metre 6.1) documents the return of normal marine oxygenated seawater conditions. Geochemical redox proxies applied to modern carbonates: implications for palaeoseawater redox reconstructions Geochemical redox proxies applied to near-coastal shallow-water carbonates are very helpful tools but, unfortunately, susceptible to several factors other than redox levels (Ling et al. 2013). Below we discuss these proxies based on features observed in Recent settings and critically compare these observations with evidence collected from the Cretaceous section in Portugal. Examples for the complexity of these proxies are found in studies documenting uranium enrichment and 238 U/ 235 U isotope ratios of Recent organic-rich surface sediments across continental margins (Abshire et al. 2020).
Uranium enrichment and U isotope ratios. -Uranium has a long seawater residence time (~0.5 Myr; Dunk et al. 2002), occurs in two redox states, U (IV) and U (VI) and is rather soluble in oxygenated water due to the predominant uranyl ion U VI O 2 2− stabilized by the formation of non-reactive carbonate complexes (Calvert & Pedersen 1993). In suboxic and anoxic environments, the reduction of U (VI) to U (IV) is important for the removal of U from the oceans (Morford & Emerson 1999;McManus et al. 2006). Modern seawater has δ 238 U values in the order of −0.4‰ (Weyer et al. 2008). In any ideal scenario, the δ 238 U values of modern primary carbonate precipitates and well-preserved corals (aragonitic) with an age of up to 600 ka are undistinguishable from the δ 238 U of modern seawater (Stirling et al. 2007;Tissot & Dauphas 2015). Recent bulk carbonate sediments are usually slightly offset from seawater (up to 0.2‰) due to the incorporation of minor amounts of U (IV) into the carbonate lattice (Chen et al. 2018). Hence, different carbonate particles within the same sample might display variable δ 238 U values (Hood et al. 2018). Variations in δ 238 U values are further induced by microbial reduction of U (VI) to U (IV) under anoxic conditions at the sedimentwater interface (Zheng et al. 2002). During the incorporation of U into carbonate precipitated in the upper water column, the U isotope composition will fractionate to heavier values, pending that the seawater is anoxic or hypoxic. In case of an oxic upper water column and anoxic deep-water masses, significant reduction-driven U isotope fractionation takes place (Romaniello et al. 2013;Chen et al. 2018). Furthermore, terrigenous input can influence the δ 238 U signature and samples must be screened for detrital Al concentrations and the U EF must be corrected for a detrital signal.
In the case of the Ericeira carbonates, δ 238 U values in the first two section metres are slightly enriched in 238 U relative to modern seawater. Chen et al. (2018), however, proposed that carbonates may indeed show an average offset of about~0.2 ‰ relative to seawater. Assuming that this offset can be applied to the data shown here, δ 238 U values reflect carbonate sediments deposited in oxic seawater with U values of −0.4 ‰ (Fig. 7A). With reference to samples taken from the Lithocodium facies, δ 238 U values decrease and reach a minimum of −0.7 ‰. Possible explanations for these, relative to modern seawater, depleted δ 238 U values are: (1) a light detrital source of U (Holmden et al. 2015) resulting in an incorrect correction for terrigenous U; (2) a U isotope fractionation towards isotopically depleted δ 238 U during carbonate deposition; (3) a local source of isotopically light U in a restricted setting; and (4) a globally lower seawater δ 238 U at the time of deposition. Given that the Ericeira carbonate section was deposited prior to the onset of OAE 1a, the notion of a globally lower seawater δ 238 U signature, expected during the acme of OAE 1a (Brennecka et al. 2011), seems unlikely. Moreover, a local source of isotopically light U in a restricted basin, due to the connection to the North Atlantic is equally unlikely. In contrast, iron-rich dolomites and ferrous-manganese crusts, both characteristic features of the pinnacle interval, potentially act as sources of light U during diagenesis and dissolution. The most likely interpretation is perhaps that of a riverine light detrital U source and furthermore, fractionation towards isotopically depleted δ 238 U during carbonate deposition.
The U enrichment factors beneath and above the pinnacle interval are generally higher than 5, a feature that suggests only moderate terrigenous influence. Any given value larger than 1.0 points to an enrichment relative to the average crustal abundance. Enrichment factors >3 represent a statistically relevant enrichment, and values >10 represent a moderate to strong degree of enrichment (Algeo & Tribovillard 2009).
In modern normal marine systems, enrichment of U in sediments can be classified as follows: (1) oxic environments are typified by a minor enrichment; (2) suboxic environments typically display a moderate enrichment (U EF < 10); and (3) anoxic environments show characteristically high levels of enrichments (U EF > 10; Algeo & Tribovillard 2009;Abshire et al. 2020). Two samples taken from the here studied microencruster interval provide U EF values that are significantly higher (45 to 80) (section metres 3.4 and 4.8; Fig. 7C), which might therefore point to anoxic conditions. The high U EF values could be consistent with the low δ 238 U and the low U concentration of seawater, if assuming a global expansion of seafloor anoxia. In case of a fast development of anoxia, U becomes more non-conservative and features a shorter ocean residence time, allowing a change of seawater δ 238 U at time scales of 10 k.y. (Brennecka et al. 2011;Lau et al. 2016;Elrick et al. 2017). Nevertheless, regarding the pre-OAE 1a setting studied here, a fast global expansion of seafloor anoxia is unlikely.
Summing up, for the interpretation of an anoxic setting (similar to the Kanfanar section, Croatia; Hueter et al. 2019), the most pronounced seawater oxygen depletion and the most prominent peak in the δ 238 U values in the Ericeira section would be expected in the microencruster-dominated facies (2.5-6.1 m) and especially within section metres 4.4 and 4.8 (pinnacle interval). In the Ericeira section, this δ 238 U peak (−0.7‰) is present at section metre 3.4 (Fig. 7A), that is below the pinnacle interval. Furthermore, the δ 238 U values return to 'normal oxidized seawater' U values within the microencrusterdominated facies and below the hiatal surface (6.1 m). In an anoxic environment, such a return to normal values would be likely found above the microencruster-dominated facies, given that a change in facies and biota is an indicator for a change in seawater redox conditions (trend to more oxygenated conditions). Given the long residence time of U in seawater (~0.5 Myr; Dunk et al. 2002), short-term patterns recorded here do not reflect the U isotope composition of seawater. Given the pre-OAE 1a age of this section, it therefore seems more plausible that iron-rich dolomites and Fe-Mn crusts acted as source of light U or, even more likely, riverine input acted as light U source and caused a fractionation towards isotopically depleted U during the carbonate deposition. This interpretation is in agreement with the extremely high U EF (U EF > 10 indicate substantial authigenic enrichment; Algeo & Tribovillard 2009) suggesting terrigenous influence (connection to the hinterland) as controlling factor for the U isotopic signal (same for REEs), rather than genuine changes in seawater redox conditions.
Rare Earth Elements and cerium anomaly. -Rare Earth Elements and specifically Ce anomalies are among the most widely applied redox proxies (German et al. 1991;Sholkovitz & Schneider 1991;Bellanca et al. 1997;Bodin et al. 2013;Della Porta et al. 2015). Cerium, present in its tetravalent state in oxic environments will be preferentially removed from the water column by scavenging processes. In open marine, oxidized environments, Ce oxidation is mediated through bacterial activity and takes place in the upper oceanic water masses. Hence, Ce anomalies serve as tracers of palaeoredox conditions in the upper water column. When these water masses become anoxic, Ce turns to its reduced state and behaves chemically similar to its neighbours La and Pr. Specifically, the Ce/Ce* ratio tends then to be closer to 1.
As material of choice for the palaeoredox reconstruction with Ce anomalies, microbial automicrite has been proven to record the ambient seawater REE geochemistry and can serve as a tracer of palaeooxygen levels (Olivier & Boyet 2006;Della Porta et al. 2015). Analysis requires comparably large sample volumes (0.5-1 g), a prerequisite that is given in samples taken from the microbial crusts associated with the pinnacle interval (Fig. 5C). Care must be taken to not combine data from automicrites and such from detrital micrite. This issue was circumvented by comparing the Si content of detrital micrite and automicrite in EDS spectra (Guido et al. 2011). Pinnacle interval automicrites display very low Si contents providing support to the notion as in situ, non-detrital materials (Fig. 5B, C).
Moreover, the concentration of specific REEs has a direct influence on the Ce anomaly values. This is because the Ce anomaly is calculated from the concentration differences of La, Ce and Pr. To exclude this bias, the La anomaly diagram of Bau & Dulski (1996) is commonly applied (Fig. 6C). In the case of the Ericeira section, most values plot in field IIa of Figure 6, indicating a positive La anomaly. The implication of this is that the Ce pattern observed in the Portugal section is the product of an anomaly in La, and therefore does not reflect a true depletion in the seawater oxygen content.
REE concentrations behave comparably conservative during diagenesis (Nothdurft et al. 2004) whereas Ce anomalies are susceptible to a variety of influences (Sholkovitz & Shen 1995) including: terrigenous influence, high aluminium (Al) and scandium (Sc) concentrations, and anomalies in lanthanum concentrations. Ling et al. (2013) place an upper Al concentration limit at 0.35 % (3500 ppm) and at 2 ppm for Sc. In the Ericeira section, the concentrations of both elements are significantly lower (Al ≤ 1500 and Sc ≤ 1 ppm). Moreover, REE patterns measured in the Portugal section ( Fig. 6D) take up the typical 'bell-shaped' pattern, interpreted as an indicator for contamination and/or riverine influence (Della Porta et al. 2015). This interpretation is supported by the LREEs to HREEs ratio, which is close to, or even above 1.
The yttrium/holmium (Y/Ho) ratio is a widely used screening tool to detect terrigenous contamination (Fig. 6B). Chen et al. (2015) place the minimum Y/Ho value of seawater signal at 35. Modern seawater Y/Ho ratios are close to 60 or higher whereas values of 35 or lower must be treated with care and most likely reflect contamination. Y/Ho ratios measured in samples taken from the Ericeira section range between 25 and 45 (mean of 35), a feature that points to a pronounced terrigenous influence.
Summing up, the Ce anomalies observed in the Portugal section suggests that this proxy is biased by terrigenous influence (La anomaly), a notion that seems likely in these coastal sections, and should not be used as evidence for seawater anoxia. Uranium isotopes represent the most intricate proxy but suggest well-oxygenated conditions, close to the recent δ 238 U values, throughout the section studied. Minor variations of the δ 238 U values can be explained by local changes in seawater redox conditions and do not reflect strong oxygen-depleted conditions. Therefore, proxy data must be handled with care and oxygen depletion seems unlikely.
How to deal with conflicting palaeoecological and geochemical evidence?
Assessing the above-discussed geochemical data from the Portugal section, oxygen-depleted seawater as main stressor driving the transient decline of a diverse marine fauna and the rise of the microencruster buildup facies seems unlikely. This outcome stands in clear contrast to data documented in Hueter et al. (2019) from a Central Tethyan shallowwater setting, showing evidence for a correlation between oxygen-depleted platform-top water masses and the transient mass occurrence of (bacinelloid) microencrusters. Clearly, the response of Cretaceous carbonate platform ecosystems to environmental change is complex and might be regionally and temporally different even where similar facies patterns are observed.
Following Schmitt et al. (2019), we tentatively suggest that a combination of several stressors induced the palaeoecological trends observed. Previous authors (Immenhauser et al. 2005;Waite et al. 2007;Rameil et al. 2010;Schlagintweit et al. 2010;Bover-Arnal et al. 2011;Huck et al. 2012) argued that increased trophic levels and water-quality decline (algal bloom and turbid waters) are relevant factors to be considered. In the (relatively proximal) coastal Ericeira section, due to its connection to the hinterland, an increase in riverine run-off and nutrient availability is likely and documented in the geochemical proxies applied here. Furthermore, high-amplitude relative sea-level fluctuations, resulting from the opening of the proto-North Atlantic (Rey et al. 2003;Dinis et al. 2008) might have played a role.
Observations from Recent nearshore sediment-impacted reefal systems might be, at least to some degree, relevant in this context. Many classical studies suggested that sediment particles and related nutrient influx smother reefal organisms, stunt and kills corals and reduces illumination relevant for photosynthesis (Rodgers 1990). Lokier et al. (2009) demonstrated the presence of a clear relationship between the volume of siliciclastic/volcanoclastic sediment input and the composition and nature of carbonate-producing organisms. On the other hand, it has been demonstrated that some sessile reefal organisms tolerate short-term burial and can survive in high-energy environments where strong currents or wave activity exhumes sediment-covered organisms Wolanski et al. 2005). Despite the ability of many sessile benthic organisms to clear small amounts of sediment from their surfaces (Rodgers 1983;Rosen et al. 2002;Lirman & Manzello 2009), becoming temporarily covered by sediments is still a major stressor and potentially lethal, either due to a reduced efficient photosynthesis or because of physically blocked feeding mechanisms . With references to Holocene scleractinian coral reefs, significant sediment and nutrient input into the reefal domain is related to a reduced species number, less live corals, lower growth rates and decreased calcification, greater abundance of branching forms, reduced coral recruitment and slower rates of reef accretion (Rodgers 1990;Erftemeijer et al. 2012;Jones et al. 2015). The presence of turbid, nutrient-rich waters, however, does not mean that coral growth is entirely prohibited and a somewhat more refined view has been proposed in more recent work (Lokier et al. 2009;Perry et al. 2009Perry et al. , 2012Ryan et al. 2016;Santodomingo et al. 2016;Johnson et al. 2017). In some cases, suboptimal habitats are now considered as important in their role of reefal diversification by hosting a pool of species tolerant to stressed conditions (Ryan et al. 2016;Santodomingo et al. 2016). Evidence suggests the existence of well-developed Holocene reefs with high accretion rates in turbid shallow-water habitats, characterized by high sedimentation rates and siliciclastic input (Perry et al. 2009(Perry et al. , 2012Ryan et al. 2016). It is proposed that studies documenting the response of Recent nearshore reefal ecosystems to waterquality decline might form an interesting starting point for future work. For the time being, and with reference to the Cretaceous case example studied here, we conclude that we find no evidence that significant amounts of (transient) clastic influx suffocated the coral-rudist facies in the Ericeira section, and by this created an ecological niche that was subsequently occupied by microencruster facies. Hence, if patterns in clastic influx and, related to this, trophic levels, were significant factors in the Cretaceous case study discussed here, then amount of clastic influx was not comparable with that of sediment-impacted reefs as described from the modern world. Nevertheless, we consider the lessons from the Recent basically meaningful, and suggest that they should form a starting point for future research aiming at a better understanding of these fossil faunal turnover events.

Conclusions
This study documents, compares and discusses a palaeoecological and a geochemical data set across a conspicuous upper Barremian-lower Aptian shallowmarine carbonate platform succession in Portugal. The most prominent feature in the section studied is the decline of what is considered a normal marine rudist-coral ecosystem and the transient establishment of a microencruster facies forming metre-sized, morphologically complex buildups. Geochemical palaeoredox proxies (Ce/Ce* and δ 238 U isotope data) are applied to autochthonous microbial (auto)micrite samples collected from this section in order to assess the significance of changes in dissolved seawater oxygen levels (coastal anoxia). The outcome documents the complexity of these proxies applied to nearshore shallow-marine sediments. Chemostratigraphical anomalies in Ce/Ce* ratios are observed, but likely reflect terrigenous influence (La anomaly) rather than changes in seawater dissolved oxygen levels. The interpretation of uranium isotope data is not straightforward but suggest well-oxygenated conditions, close to recent seawater δ 238 U values, throughout the section studied. Excluding coastal seawater anoxia as main stressor, it seems likely that a set of environmental stressors including perhaps trophic levels and high-frequency relative sea-level change, among other factors, caused the stratigraphical patterns observed in Portugal, a scenario that is in agreement with previous work (Bover-Arnal et al. 2011;Huck et al. 2012). The outcome of this study contrasts recent work by Hueter et al. (2019) documenting a clear relation between Central Tethyan OAE1a-time equivalent microencruster intervals that formed as response to time intervals when shallowmarine seawater anoxia established. Clearly, the interaction of middle Cretaceous carbonate platform ecosystems with environmental stressors and waterquality decline is highly complex. Even when palaeoecological patterns in different sections share important similarities, evidence for directly comparable causal relations is as yet lacking.