Earth system instability amplified biogeochemical oscillations following the end-Permian mass extinction

Abstract

After the end-Permian mass extinction, the Earth system underwent extreme ecological and environmental fluctuations, including high temperatures, recurrent oceanic anoxia, and carbon cycle oscillations as demonstrated by the geochemical isotope proxy records. However, the underlying mechanism behind these oscillations remains poorly understood. Here we propose that they were produced by a coupled oscillation mode of marine phosphorus (P) and atmosphere-ocean carbon (A), driven by nonlinear redox controls on marine phosphorus burial. Our modeling demonstrates that the initial emplacement of the Siberian Traps and the mass extinction (on land and in the ocean) directly led to an early Triassic greenhouse. More importantly, it homogenized the ocean floor redox condition towards anoxia, activating amplifying feedbacks and destabilizing the system. The internal dynamics of an unstable system-rather than recurrent volcanic shocks-triggered the periodic oscillations (limit cycles) of serial excursions in carbonate carbon and uranium isotopes during the early Triassic.

Fig. 1. Environmental perturbations and delayed recovery of global marine and terrestrial biodiversities after EPME.

See Supplementary Information (Table S1) for datum sources, or find data from Supplementary Dataset 1. a Marine carbonate carbon isotope (δ¹³C_carb) profile. Red and black lines are the moving average, minimum, and maximum with a moving window = 0.1 Myr. b Marine redox proxy, carbonate uranium isotope (δ²³⁸U_carb) profile. Four negative Uranium isotope excursions (NUIE) at around 252 Ma, 250.5 Ma, 248.5 Ma, and 247 Ma represent multiple expansion peaks of ocean anoxia. c Sea surface temperature proxy, apatite oxygen isotope (δ¹⁸O_apatite) profile, showing the early Triassic hothouse regimes (252–247 Ma). d End-Permian to late early Triassic terrestrial plant and marine invertebrate (from PBDB) diversities, showing delayed biotic recovery^,.

Fig. 2. Feedback diagram illustrating the plausible key processes leading to the system’s destabilization.

Solid and dashed arrows show positive and negative effects, respectively. The boxes show three key processes that contribute to the system’s destabilization. a The emplacement of the Siberian Traps resulted in elevated tectonic uplift, which subsequently enhanced silicate weathering and then phosphorus weathering. b The collapse of the terrestrial ecosystem halted the burial of phosphorus and organic carbon on land. c The collapse of marine ecosystems, combined with the above two processes, led to a homogenization of ocean floor redox conditions towards an anoxic state, and facilitated the anoxia-phosphorus-burial feedback. The latter includes detailed positive and negative feedback loops within redox-sensitive phosphorus cycles. The positive feedback loop highlights how the elevated oceanic phosphate concentrations stimulate new production, leading to anoxia and the recycling of phosphorus from sediments, effectively trapping the system in anoxic conditions, with a timescale of ~0.1 Myr. The negative feedback mechanism, characterized by enhanced organic carbon burial, results in oxygen accumulation over a longer timescale of ~1 Myr, alleviating marine anoxia.

Fig. 3. Homogeneity of ocean floor redox conditions controls the stability of the phosphorus-carbon-oxygen system.

a The phase plane for an idealized stable system, showing the dP/dt = 0 nullcline surface (plasma) where oceanic phosphorus level is invariant in time, and the dO/dt = 0 (red) and dA/dt = 0 (black) surfaces, and their intersections with the P surface. b, c Time series of the normalized P–O–A levels and δ¹³C_carb. d–f The phase plane and time series for an idealized “fast” P–O–A unstable system, showing limit cycle oscillations, omitting the adjustment process, with the P surface being folded with both unstable (repelling) and stable (attracting) regions. A complete cycle can be subdivided into four parts, the fast accumulation and consumption of the P are marked by gray bars in panel e. Note the different scales of the tmodel axis are given in various cases. X_norm represents the X level compared to the modern values.

Fig. 4. Biogeochemical modeling results (time series).

a Tectonic Uplift (U) and Vegetation (V) forcings, representing the long-term effect of the volcanic eruption of the Siberian Traps and collapse of terrestrial ecosystem. b System stability forcing (S), representing the degree of homogeneity of the ocean floor redox conditions. Two groups of model runs: lower S for heterogeneous ocean floor redox condition, hence a stable system (blue lines and bars), and higher S for homogenous ocean floor redox condition, hence an unstable system (orange lines and bars). c–e Comparison between modeled result and raw data of δ¹³C_carb, δ¹⁸O_apatite vs temperature (K) and δ²³⁸U_carb, respectively. f Degree of marine anoxia. Note that the first negative uranium isotope excursion is not captured here, see extended model runs in Fig. S14, where the results are more consistent with data when short-term forcings are considered.

Fig. 5. Modeling results (phase planes) of biogeochemical cycles corresponding to the time series analyses (shown in Fig. 4).

Note that the oxygen and carbon nullcline surfaces are omitted, but the intersection lines (red and black lines on P surface) are shown. a Phase plane of the adjustment process during the end-Permian. Changing trajectories during 254‒252 Ma (bold red line). b, c Phase planes of the control model run, changing trajectories during 252‒248 Ma (bold green lines) and 248‒244 Ma (bold blue lines). d Corresponding time series of reservoir P–O–A levels. e, f, g Phase planes of the treatment model run and corresponding time series of reservoir P–O–A levels. Note the control and treatment model runs both initialized from panel (a), and different scales in it are set to show the adjustment process from the end-Permian in detail. See the definition of the abbreviations in Fig. 3.