Solid Earth forcing of Mesozoic oceanic anoxic events

The study investigates the drivers of Mesozoic oceanic anoxic events (OAEs)—transient intervals of widespread ocean deoxygenation that caused major ecological disruptions. OAEs were frequent between ~183 and 85 Ma, with mean duration ~1.4–1.5 Myr and recurrence ~7 Myr, during an interval characterized by continental breakup and abundant large igneous province (LIP) activity. It is widely accepted that OAEs relate to global warming, intensified hydrological cycle, and enhanced continental chemical weathering that increase marine productivity and organic carbon burial, leading to oxygen depletion. However, the first-order causes remain debated due to overlapping processes including volcanic/methanogenic carbon outgassing, circulation changes, and nutrient cycling. Prior hypotheses emphasize LIP CO₂ emissions, but age uncertainties and protracted LIP activity hinder definitive causation. This paper tests whether processes tied to continental breakup—especially enhanced weathering of mafic rocks during rifting and nascent seafloor spreading—could deliver phosphorus, stimulate productivity, and drive recurrent anoxia, complementing volcanic outgassing models.

Prior work links OAEs to greenhouse climates and increased nutrient delivery, recording organic-rich black shales and carbon isotope excursions. LIP emplacement has been implicated as a trigger via elevated CO₂, yet LIP age uncertainties (±1 Myr) and long activity spans (up to ~40 Myr) complicate causal attribution, and LIPs coincide with rifted margins where multiple processes co-occur. Geochemical proxies such as 87Sr/86Sr and initial 187Os/188Os capture combined influences from arc weathering, breakup, and volcanism; Sr’s ~5 Myr residence time limits temporal resolution for OAE initiation. Earth system models also highlight the role of continental configuration in shelf deoxygenation. Phosphorus is a key biolimiting nutrient; previous box-model studies showed increased continental P delivery can drive OAEs, and P recycling under anoxia can prolong events (for example, OAE 2). Submarine basalt alteration and basalt-seawater interactions have been proposed as nutrient sources coincident with ridge formation. This study builds on these insights by quantifying P fluxes from both incipient mid-ocean ridges and subaerial basalt denudation during Gondwana breakup, and testing their capacity to reproduce OAE timing and intensity in a biogeochemical framework.

• Plate-tectonic reconstructions and seafloor production: Used the global plate model of Müller et al. (2016) with corrections (Torsvik et al., 2019), extracting mid-ocean ridge (MOR) kinematics via pyGPlates. Calculated seafloor production rates by summing spreading rate × segment length at 1 Myr steps (200–0 Ma), exporting time series CSVs.
• Identification of incipient MORs: Defined polygons in GPlates for nine ridge systems transitioning from continental rifts in thick lithosphere (Central Atlantic, Australia–Molucca, Africa–Madagascar, Africa–Antarctica, Iberian Atlantic, South Atlantic, India–Australia–Antarctica, Equatorial Atlantic, Madagascar–India). Polygons active for first 30 Myr of spreading (shallowest phase). Used GMT to intersect ridge datasets with polygons and build time series per basin.
• Submarine P flux modelling from incipient MORs: Adapted probabilistic models for elemental release during basaltic glass alteration. Assumptions: shallow water depths ≤1,500 m; alteration thickness decays linearly from 1,500 m to 100 m over 30 Myr due to subsidence and reduced hydrovolcanism/hydrothermal convection. Sampled uniform distributions for parameters including P2O5 loss fraction [0.002, 0.006], palagonite fraction, porosity (shallow/deep), and hyaloclastite abundance (per prior models). Standard P2O5 density 2,390 kg m−3. Accounted for ridge initiation age uncertainty by applying a random start-time offset drawn from a beta distribution (α=1, β=1.5) scaled to [0, 0.05 ti] (mean ~2%, SD ~1.3%). Conducted 10,000 Monte Carlo simulations stepping at 1 Myr; for each ridge and time step, computed hyaloclastite volume and P2O5 loss to derive moles of P released. Interpolated to regular 1 Myr intervals and summed across ridges to obtain ensemble mean and percentiles of P fluxes (180–59 Ma).
• Subaerial basalt denudation and riverine P flux: Quantified uplift-driven chemical denudation across southern Africa (Karoo-aged basalts) from ~160–80 Ma using thermochronology-derived cooling histories at 28 localities. Interpolated to 0.1 Myr spacing, computed rates of temperature drop over a symmetric 1.9 Myr moving window; scaled denudation rates linearly between Dmin=20 m Myr−1 and Dmax=175 m Myr−1 according to normalized ΔT. Assumed basalt areal extent Basalt_area=600,000 km² (based on thermochronology, kimberlite xenoliths, and feeder intrusion footprints). Converted denudation to eroded volume (km³ Myr−1). Sampled empirical P2O5 contents from GeoROC database (Karoo/Ferrar basalts; 2,495 observations, excluding 36 zeros) across 10,000 iterations to estimate dissolved/reactive P released to oceans via rivers.
• Biogeochemical modelling: Employed a five-box ocean–atmosphere COPSE-like model (Slomp & Van Cappellen; Alcott et al. 2019) tracking carbon, phosphorus, and oxygen. Configured for Mesozoic background with reduced open-ocean vertical mixing (70% of modern) to reflect a warmer, more sluggish ocean. Injected ridge-derived P as soluble reactive P into the Deep Ocean box and riverine weathering P into the proximal shelf box. Simulated individual P-pulse events (ridges and weathering) separately to avoid unrealistic cumulative atmospheric O2 feedback, and tracked deep-ocean dissolved O2 responses, generating ensembles (mean and percentile envelopes).
• Performance evaluation: Identified minima of modelled deep-ocean [O2] across separate ridge and weathering runs to construct a single predictor time series (normalized to [0,1], inverted so 1 = strongest deoxygenation). Evaluated ability to predict binary OAE occurrence using ROC analysis with bootstrap confidence intervals; reported AUC and alternative threshold metrics (Youden’s statistic and closest-to-top-left).

• Tectonic context: Identified ~26,000 km of mid-ocean ridges that initiated during the Jurassic–Cretaceous, with ~85% simultaneously active from 160–120 Ma, overlapping the principal OAE interval.
• Submarine P fluxes: Modelled phosphorus release from incipient, shallow MORs potentially exceeded 3 × 10^10 mol yr−1 (ensemble), equivalent to >75% of the present-day total reactive P flux from rivers (~3.9 × 10^10 mol yr−1).
• Subaerial P fluxes: Weathering of uplifted Karoo flood basalts in southern Africa delivered P fluxes up to ~3 × 10^10 mol yr−1, comparable to modern dissolved inorganic P input to the oceans, with peaks at ~120, 112, 104, and 95 Ma corresponding to denudation events.
• Biogeochemical response: Model ensembles show that both submarine and subaerial P pulses are sufficient to cause substantial ocean deoxygenation. Predicted minima in mean deep-ocean [O2] reach decreases as low as about −110 µmol kg−1 (~30% reduction in average marine O2), implying a rise in globally anoxic seafloor coverage from ~0.3% (modern) to ~4%—consistent with estimates for many OAEs.
• Event correspondence: Modelled deoxygenation broadly coincides with 9 of 11 confirmed OAEs. OAEs 1c and 1d occur ~2 Myr before modelled riverine peaks; OAE 1b occurs ~3 Myr before modelled deoxygenation peak. A minor deoxygenation is predicted near the Jurassic/Cretaceous boundary where no OAE is recognized. ROC analysis of the [O2]-minima predictor yields AUC = 0.705 (95% CI 0.660–0.750), indicating better-than-random predictive skill for OAE timing.
• Geochemical signals: During OAE 2, δ13C excursions and sharp declines in Os isotopes align with basalt–seawater interactions; expanded deoxygenation preceded major organic carbon burial. Mixed δ13Ccarb excursions (positive and negative) require additional sources of isotopically light carbon (e.g., volcanism, thermal metamorphism), compatible with, and complementary to, P-driven productivity feedbacks.

The results support a mechanistic link between continental breakup processes and repeated Mesozoic OAEs. Enhanced weathering of mafic lithologies during the rift-to-drift transition and early ridge development released substantial pulses of bioavailable phosphorus, stimulating marine primary productivity and enhancing organic carbon burial. This nutrient-driven feedback promotes widespread ocean deoxygenation and, via sedimentary P recycling under anoxia/euxinia, can sustain anoxic conditions over Myr timescales. The quantified P sources from both submarine basalt alteration at incipient MORs and subaerial basalt denudation offer a complementary pathway to LIP CO2 outgassing hypotheses: whereas LIPs contribute greenhouse forcing and possible light carbon inputs, the phosphorus supplied by solid Earth processes associated with breakup provides a proximate mechanism to trigger and amplify anoxia. Model–data comparison indicates moderately good alignment between predicted deoxygenation pulses and observed OAEs, with an ROC AUC of 0.705. Discrepancies in exact timing (e.g., OAE 1b, 1c, 1d) are within typical age uncertainties and may reflect additional forcings (e.g., episodic volcanism, basin restriction, circulation changes). Mixed δ13C signals during OAEs, including negative excursions, imply overlapping carbon cycle perturbations that can coexist with the proposed P-driven mechanism. Overall, the study advances the view that tectonic reorganization directly modulated ocean nutrient supply and redox state, tightly coupling the solid Earth to biosphere dynamics during the Mesozoic.

Continental breakup and nascent seafloor spreading during the Mesozoic created sustained and pulsed sources of phosphorus to the oceans—from both submarine alteration of fragmental basalts at incipient MORs and subaerial weathering of uplifted flood basalts. These P inputs were sufficient to drive recurrent oceanic anoxia by boosting primary productivity and organic carbon burial, consistent with the scale and frequency of Mesozoic OAEs. Modelled deoxygenation events align with most OAEs and yield predictive skill beyond chance. The proposed framework complements volcanic outgassing-centered triggers by quantifying nutrient pathways linked to tectonomagmatic activity. The waning of breakup-related activity in the Late Cretaceous may help explain the decline in OAE frequency and intensity. Future work should integrate additional feedbacks (e.g., explicit atmospheric O2–biosphere coupling, circulation dynamics), expand denudation reconstructions beyond southern Africa, refine event chronologies, and assimilate multi-proxy records to further test and calibrate the nutrient-forcing hypothesis.

• Chronological uncertainties: LIP ages and durations (±1 Myr; activity over tens of Myr) and OAE ages (±1–2 Myr) complicate precise alignment; model includes ridge initiation age uncertainty but residual timing mismatches remain.
• Model scope: The five-box model lacks strong feedbacks for atmospheric O2 tied to the terrestrial biosphere and wildfires; therefore, combined long-term P inputs were tested as separate events to avoid unrealistic high-O2 states.
• Geographic focus: Subaerial P flux estimates emphasize southern Africa; global extrapolation is limited and estimates are considered conservative minima.
• Parameter uncertainties: Probabilistic ridge flux modelling depends on distributions for palagonitization, porosity, hyaloclastite abundance, alteration thickness decay, and P2O5 loss; submarine weathering rates and bathymetric histories introduce uncertainty.
• Proxy interpretability: Sr isotopes have long residence times; Os and C isotope records integrate multiple processes, making unique attribution challenging; mixed δ13C signals indicate additional carbon cycle perturbations.
• Model–data mismatches: Some OAEs (e.g., 1b, 1c, 1d) show offset timing; a predicted minor deoxygenation is present near the J/K boundary without a recognized OAE; ROC AUC < 1 indicates imperfect prediction.