Tectonic Reorganizations and Multistability of the Mantle-Plate System

The geological record indicates that Earth has experienced rapid and drastic tectonic reorganizations, such as the breakup of Pangea and the global event at ∼50 Ma marked by the Hawaiian–Emperor bend and synchronous kinematic shifts across all major plates (Whittaker et al., 2007). The mantle lithosphere system is a complex nonlinear dynamical system (Coltice, 2023) that can produce such tectonic transitions (Janin et al., 2025; Guerrero et al., 2025). By analogy with the climate system, which alternates between icehouse and hothouse states, a fundamental question arises: can plate tectonics exhibit multistability, and if so, does the whole mantle-lithosphere system as well?

Here we investigate dynamical transitions in mantle convection with self-consistent plate tectonics using tools from dynamical systems theory. We analyze outputs from 3D spherical mantle convection model of Coltice et al. (2019), which reproduces major tectonic features of Earth. From a 850 Myr long simulation, we construct a database of tectonic and physical variables, including plate-boundary lengths, number of plates, proportion of deforming lithosphere, global and surface root-mean-square velocities, surface and core–mantle boundary heat fluxes, mean mantle temperature, number of mantle plumes, and lithospheric net rotation rate.

We apply two complementary methods to detect dynamical transitions: (1) sample-based tests using Maximum Mean Discrepancy (MMD; Gretton et al., 2012), which identify statistical discontinuities in multidimensional distributions, and (2) Recurrence Quantification Analysis (RQA; Eckmann et al., 1987), which characterizes changes in recurrence patterns within the system’s phase space. We perform analyses separately on surface variables, mantle variables, and the combined dataset.

We identify four statistically significant transitions. Some coincide with major tectonic reorganizations, such as supercontinent assembly and breakup or global kinematic shifts, while others reflect intrinsic changes in convective or tectonic regimes. Certain transitions affect both mantle and surface dynamics synchronously, whereas others are confined to either the lithosphere or mantle flow. To interpret these transitions, we combine Principal Component Analysis (PCA) with spectral analyses of mantle thermal heterogeneity. In this framework, detected transitions correspond to shifts in one or more principal components representing distinct tectonic, thermal, and kinematic states of the system, providing quantitative evidence for multistability in mantle-plate dynamics.