A team of seismologists has assembled the most extensive map yet of deformation deep inside Earth, revealing that the warping of rock just above the core-mantle boundary closely tracks the paths of ancient ocean plates that sank into the planet’s interior. Built from more than 16 million seismograms collected across 24 data centers, the map samples nearly 75% of the lowermost mantle and finds measurable strain in roughly two-thirds of that area. The result ties surface-level plate tectonics to processes operating nearly 2,900 kilometers below the ground, offering a new way to read the geologic record of subduction that stretches back hundreds of millions of years.
What is verified so far
The central finding comes from a study in The Seismic Record by researchers including Wolf, Li, and Romanowicz. Their approach treats seismic anisotropy, the directional dependence of seismic wave speed through rock, as a proxy for strain. When mantle minerals are squeezed or sheared over geologic time, their crystal lattices align in predictable ways, causing waves to travel faster in some directions than others. By measuring that effect across a global waveform archive, the team produced a picture of where the deepest mantle has been actively deformed and where it has not.
The deformation patterns do not appear random. They cluster in regions that line up with the expected resting places of subducted slabs, the dense remnants of oceanic lithosphere that descend at convergent plate boundaries and eventually pile up near the core. The USGS maintains a reference model called Slab2, which catalogs three-dimensional geometries of present-day subduction zones at upper-mantle scale and serves as a widely used benchmark for slab-related analyses. Although this model describes active subduction in the upper mantle rather than deep-mantle structures, the spatial correlation between its mapped slab positions and the new anisotropy data strengthens the case that material from those same zones has been sinking to the base of the mantle for tens to hundreds of millions of years.
Companion work by overlapping authors in Nature Communications zeroes in on one specific feature: the Perm Anomaly, a patch of anomalously slow seismic velocities beneath western Russia. That study interprets mantle deformation signals in the Perm region as a fossil record of convergent upwelling driven by subducted material. In other words, the strain patterns preserved in deep-mantle rock can function like geological fingerprints, recording where ancient slabs once pushed hot material upward and possibly reorganized flow at the base of the mantle.
A separate study published in Nature Geoscience adds another layer of evidence. That paper documents detections of ultralow velocity zones, or ULVZs, within high-velocity regions of the lowermost mantle. ULVZs are thin patches where seismic waves slow dramatically, often interpreted as pockets of partial melt or chemically distinct material. Finding them inside fast-velocity zones, which are typically associated with cold, dense slab remnants, reinforces the idea that subducted plates carry or generate compositionally unusual material as they settle at the core-mantle boundary. The study includes a detailed, auditable data-provenance chain that traces waveform selections back to their original recordings, setting a standard for reproducibility in deep-Earth seismology.
Taken together, these results point toward a coherent picture. Subducting plates appear not to vanish once they sink beyond the reach of conventional seismic imaging. Instead, they leave a long-lived imprint in the form of anisotropic fabrics and localized ultralow-velocity patches. The new map effectively extends the observable influence of plate tectonics from the surface down to the base of the mantle, suggesting that the history of ocean-basin closure and continental collision is encoded far deeper than previously appreciated.
What remains uncertain
Several questions remain open despite the scale of the new map. The most significant gap is temporal. Seismic anisotropy records cumulative strain, but it cannot easily distinguish deformation that occurred 50 million years ago from deformation happening now. Researchers can infer past subduction from the spatial match between anisotropy and known slab trajectories, yet the precise timing of when each slab reached the core-mantle boundary is still model-dependent. No direct observational method currently timestamps deep-mantle strain with high precision, so reconstructions rely on plate-motion histories and numerical simulations.
The interpretation of anisotropy itself carries uncertainty. Different mineral phases at extreme pressures can produce similar anisotropic signatures, and laboratory experiments on lower-mantle minerals have not fully resolved which deformation mechanisms dominate at those conditions. The Nature-hosted analyses from the Wolf, Li, and Romanowicz group acknowledge that competing mineral-physics models can shift the inferred flow directions, even when the raw seismic observations are identical. As a result, arrows drawn on deep-mantle flow maps should be viewed as hypotheses constrained by data, not as directly imaged streamlines.
Coverage gaps also matter. While nearly 75% of the lowermost mantle is sampled, the remaining quarter includes large swaths beneath oceans where seismic station density is low and earthquake paths are sparse. These unseen regions could contain anisotropy patterns that either confirm or complicate the slab-driven narrative. At present, there is insufficient data to determine whether the remaining unsampled areas follow the same statistical relationship between anisotropy and subduction history, or whether they host alternative styles of deep convection.
The Perm Anomaly study raises its own interpretive tension. Describing deformation there as evidence of “fossil convergent upwelling” is a specific dynamical claim that goes beyond simply noting strain. Whether the Perm Anomaly is a stalled plume root, a chemical pile, or something else entirely is still debated in the geophysics community. The peer-reviewed findings present one well-supported scenario, but alternative thermal and compositional models have not been ruled out. Linking that feature to past subduction episodes remains plausible yet provisional.
There are also methodological limits. Even with more than 16 million seismograms, the inversion problem (inferring three-dimensional anisotropy from finite, noisy data) is underdetermined. Choices about how to weight different seismic phases, how to parametrize the mantle, and how to regularize the solution all influence the final map. The authors attempt to quantify these effects through synthetic tests and cross-validation, but some ambiguity is inherent to global-scale seismology.
How to read the evidence
Readers evaluating this research should distinguish between three tiers of evidence at play. The first and strongest tier is observational: millions of seismograms processed through standardized methods, yielding a measurable anisotropy signal across a large fraction of the deep mantle. That data exists regardless of how it is interpreted, and it can be reanalyzed as techniques improve.
The second tier is correlational: the spatial match between anisotropy patterns, ULVZ detections, and subducted-slab positions. The alignment with models such as Slab2 is striking and consistent with long-term sinking of oceanic plates. However, correlation is not proof of causation on its own. Other processes, such as thermochemical plumes or basal chemical reservoirs, might also generate localized deformation and velocity anomalies that mimic slab signatures.
The third tier is interpretive. It includes claims about fossil upwelling, specific flow geometries, or the exact role of the Perm Anomaly in mantle convection. These inferences synthesize seismic observations, mineral physics, and geodynamic modeling, and they are necessarily more tentative. For instance, one access-controlled analysis of the Perm region argues for a particular pattern of upwelling linked to ancient convergence, while other models emphasize plume-like behavior rising from deep thermochemical piles. Both scenarios are consistent with parts of the data, and ongoing work aims to discriminate between them.
For non-specialists, a practical way to weigh these tiers is to ask what would likely survive future revisions. The existence of widespread deep-mantle anisotropy tied in broad terms to plate-tectonic history is robust, because it rests on large, reproducible datasets and basic physical principles. The precise shapes of individual anomalies, their detailed flow fields, and their links to specific tectonic events are more likely to be refined as new observations and models come online.
In that sense, the new map should be read less as a final portrait of Earth’s deep interior and more as a high-resolution draft. It extends the reach of seismology, demonstrates the value of transparent data pipelines, and frames testable hypotheses about how slabs, plumes, and chemical reservoirs interact near the core-mantle boundary. Future deployments of ocean-bottom seismometers, improvements in mineral-physics experiments, and more sophisticated inversions will sharpen the picture. But the core message is already clear: the deep mantle is not a passive backdrop. It bears the structural memory of plate tectonics, and that memory is now beginning to come into focus.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
