New palaeomagnetic data from South China
Previous palaeomagnetic data from South China tentatively suggest there may have been a rapid continental movement during the Late Ordovician to early Silurian30,32,35. However, data from the Silurian have been calculated as a mean pole for the whole Period (443.8–419 Ma)32,35 (Supplementary Fig. 1), which precludes detailed evaluation of maximum rates of continental motion during the O–S transition. Due to its importance for palaeogeographic comparison before and after the O–S boundary, the upper Telychian strata of the Huixingshao Formation (ca. 436–435 Ma) in Xiushan county, Chongqing, South China (Supplementary Figs. 1 and 2) were selected for detailed palaeomagnetic study. Standard palaeomagnetic methods were employed and are detailed in the Methods. Stepwise thermal demagnetization revealed a stable component with high unblocking temperature suggestive of a remanence carried by hematite, which is also supported by rock magnetic experiments (Fig. 1 and Supplementary Figs. 3, 4, and 6). Detailed description of the palaeomagnetic results is provided in the Supplementary information. The magnetostratigraphic record reveals at least four coherent polarity zones (Fig. 2) strongly suggesting that the high-temperature component from section Yongdong (SY) is primary and can be used for palaeogeographic reconstructions. However, the K-value of dispersion of the virtual geomagnetic poles (VGPs) of these six sites is 90.3 (Supplementary Table 1), exceeding 70, which suggests that these data may not average out the palaeosecular variation (PSV)36. To overcome this issue, we sought to combine our new data with the most reliable coeval previous data.
Fig. 1: New Silurian palaeomagnetic data from South China and compilation with previous results. Zijderveld plots (a, d, g), equal area projections (b, e, h) and normalized stepwise thermal decay curves (c, f, i) of the thermal demagnetization of representative samples from the section at Yongdong (SY) in geographic coordinates. In the Zijderveld plots, black and white dots represent horizontal and vertical projections, respectively, while in the equal area projections, they represent directions plotted in the lower and upper hemispheres respectively. j Equal area stereographic projection of site mean directions of the high-temperature components of the Huixingshao (HXS) Fm from this study in stratigraphic coordinates. k Virtual geomagnetic poles (VGPs) of HXS Fm from the SY section from this study compared with VGPs from the HXS Fm and the Rongxi (RX) Fm from Opdyke et al.35 and Huang et al.32. Resulting combined early Silurian pole (S 1 M) using all data from the RX and HXS Formations from this study and previous work is shown as red star with associated cone of 95% confidence. l The new recalculated early Silurian pole (S 1 M) is distinct from existing poles of South China71. All plots were generated with PaleoMac72. Full size image
Fig. 2: Magnetostratigraphy of the Huishingxiao Formation. Sampled section at Yongdong (SY). Directions with declinations >240° were interpreted as reversed polarity, and otherwise, as normal polarity. Full size image
We reassign the ages of existing Silurian palaeomagnetic results32,35 according to a recently updated stratigraphic timescale37,38,39 (Supplementary Fig. 1). A notable revision in these age reassignments is that the Rongxi Formation previously regarded as ca. 420 Ma in age is in fact early Telychian (ca. 438.5–437 Ma) (Supplementary Fig. 1). Again, data from these previous studies32,35 seem not to average out PSV36 (Supplementary Table 1; detailed analysis in Supplementary information). Nonetheless, after combining all data from the Rongxi and Huixingshao Formations (total 28 sites), a K-value of 48.4 is achieved, which is below 70 and suggests sufficient averaging of PSV. Furthermore, these data also pass a fold test40 at 99% confidence (k in geographic coordinates is 7.64, in stratigraphic coordinates is 31.17). This new early Silurian pole (S 1 M) calculated by averaging the VGPs from the Rongxi and Huixingshao Formations plots far from all younger poles and earns a reliability index of 6 of 7 (ref. 36, Supplementary information). Intriguingly, the new early Silurian pole (S 1 M) plots far from (≥50°) away a high-quality Late Ordovician (late Sandbian–middle Katian; 454–448 Ma, or ca. 451 Ma) pole of South China30 (Figs. 1l, 3a).
Given only ~10 Ma between these two ages, the 54.4° ± 6.4° arc distance between these two poles indicates a rapid APW rate of 5.4 ± 0.6 Ma−1 for South China. During this time interval, South China experienced a region tectonic movement, however it was restricted to only its southeastern part (Cathaysia terrane)41. Our early Silurian data and the Late Ordovician data are from northwestern South China (upper Yangtze terrane), which was largely unaffected by this tectonism. In addition, the regional tectonism should have only induced large differences in the declination of these data (due to potential vertical-axis rotation), but cannot explain the large inclination difference that is observed corresponding to a ~28.5° change in palaeolatitude. Non-uniformitarian magnetic fields (e.g., quadrupolar or octupolar) may also result in apparent changes in latitude33. However, in order to explain the reduced inclination of the Late Ordovician data (35°) to our Silurian data (18°), one would have to claim a same-sign octupole that was stronger than 20%, which is more extreme than any previous claims in the Phanerozoic42, and an opposite-sign octupole would increase, not decrease, inclination. Furthermore, both non-dipole cases would only affect inclination and therefore cannot explain the even larger anomaly in terms of the ~59° declination change. Lastly, an oscillation between polar and equatorial dipoles (if possible on Earth) could affect declination43, but would predict a ~90˚ change that is not observed. Therefore, we argue that this large and rapid motion of South China corroborates from an additional continent the proposed O–S TPW event27, albeit with an even larger amplitude than once thought. Nevertheless, any reproducibility test of TPW should aim to be global in scope, so we must consider the palaeomagnetic records of the other major continents.
Late Ordovician-early Silurian true polar wander
Strikingly, in addition to the large-scale 54° ± 6° APW of South China, the Late Ordovician–early Silurian palaeopoles from Tarim, Siberia, Baltica, and Gondwana also all demonstrate large arc distances of APW: 54° ± 9°, 47° ± 17°, 55° ± 14°, and 58° ± 21°, respectively (Fig. 3a, Supplementary Table 2), with associated APW rates of 5.4° ± 0.9°, 4.7° ± 1.7°, 5.5° ± 1.4°, and 5.8° ± 2.1° Ma−1, respectively. Data from Baltica and Gondwana represent recent synthetic APW paths, which consider the age error and the quality of the data34. For comparison, we also calculate the arc distances for Baltica and Gondwana from 450–430 Ma using the synthetic APW paths of Torsvik et al.44 (Supplementary Table 3), which are 51.2° ± 8.2° and 24.5° ± 18°, respectively. While the results for Baltica agree with both methods, the large difference of the two synthetic APW paths for Gondwana reflect either the larger 20 Ma age bins of Torsvik et al.44 oversmoothing the data and/or the lack of poles during this time interval which is non-ideal for synthetic methods. Nonetheless, at least four continents demonstrate similar large amplitude and synchronous polar motion. As discussed, regional tectonics and non-uniformitarian geomagnetic fields cannot explain this systematic global APW anomaly. Plate motion, driven by slab subduction and mantle convection, also cannot explain these synchronous and similar large amplitude movements of multiple isolated continents either, as it requires relative motion between different plates with different velocities (speeds and/or directions).
Fig. 3: Ordovician–Silurian apparent polar wander paths globally. a Global palaeomagnetic poles for 460–430 Ma from South China, Tarim, Siberia, Baltica, Laurentia, and Gondwana. Apparent polar wander (APW) paths all exhibit (except Laurentia) large shifts between ca. 450–440 Ma. Poles shown in present-day coordinates. Pole information is listed in Supplementary Table 2. b Poles rotated into the common reference frame of Gondwana illustrating the APW overlap. Black dot is a reference point in central Gondwana used for palaeolatitude estimates in Fig. 6e. c APW arc distances for all poles globally (in Gondwana reference frame of b) relative to the 460 Ma Gondwana pole (as an arbitrary reference point before the hypothesized true polar wander event). See text for discussion of Laurentia. Vertical bars are intervals of 95% confidence. Plots in a and b were generated with GPlates73. Full size image
TPW could explain the large and synchronous dispersions of O–S palaeopoles globally. TPW is rate-limited by the ability of the viscous mantle to deform into a reoriented hydrostatic figure45,46. TPW can occur as fast as the fastest plate motion or even comparatively faster, particularly in more ancient times when the mantle was hotter, less viscous, and thus more deformable46,47. Numerical simulations suggest that a 40–50° amplitude TPW event can occur in ~10 Ma if the viscosity of lower mantle is 1022 Pa s46. Presently lower mantle viscosity is about 3 × 1022 Pa s48, while it may be 3 times lower at 450 Ma47. Hence, considering almost all continents sped up synchronously, we propose that during Late Ordovician, most likely after the middle Katian Stage but before the Silurian early Telychian Stage, a TPW event occurred. Furthermore, the fact that all the ~50° arc distances of APW are within statistical uncertainty of each other means that the data pass the global reproducibility test of TPW.
We note that palaeomagnetic poles from Laurentia during this time are characterized, in contrast, by much less APW, and almost essentially a stillstand (Fig. 3). At face value, one continent with a statistically different arc distance of APW compared with those of other continents does not invalidate the TPW hypothesis49. This point of caution is particularly relevant here because during this time Laurentia was an isolated plate with its own tectonic motion vector. In the Paleozoic, the Iapetus and Rheic oceans that existed in between Laurentia and West Gondwana rapidly expanded and vanished12,34, which certainly would have resulted in fast tectonic movements of Laurentia and may seem at odds with its small amount of APW. As the tectonic motion of Laurentia during the closure of the Iapetus would have been mostly opposite to its sense of motion due to TPW, the effect of TPW would be partially offset and thus should appear as a relative stillstand, where APW = plate motion + TPW. In this sense, as Laurentia would have undergone large tectonic motion during this time, its palaeomagnetic stillstand can only be reconciled if TPW in the opposite direction is invoked. Thus, the TPW event inferred from all other continents provides a convenient way to explain the prior paradox of a Laurentian APW stillstand during the closure of the Iapetus Ocean.
We should also note that, strictly speaking, Laurentia may not exhibit a total stillstand. The circles of 95% confidence of the youngest and oldest Laurentian poles (460 and 430 Ma) only very slightly overlap, and the results of an F test50 demonstrate that the poles are distinct from each other at the 99% confidence level (F = 10.6). This test indicates that the 18.9° ± 19.3° arc distance between the two O–S poles is statistically significant. Therefore, while the presumably considerable tectonic motion of Laurentia partially masks the ~50° TPW event, Laurentia nonetheless does indeed record a statistically significant portion of the TPW amplitude that, in reconstructed coordinates, is consistent with the sense of TPW rotation more clearly recorded on the other continents. Otherwise, this relative stillstand may be an artifact of the large age errors of these poles used for APW comparison34.
As defined as the migration of the maximum moment of inertia (I max ) to align with Earth’s spin axis, TPW occurs as a rotation about an Euler pole controlled by the minimum moment of inertia (I min ) that is equatorial and is therefore predicted to circumscribe a great-circle APW path. Identifying TPW as a great-circle APW path also assumes that plate motion of the continent relative to the mantle is negligible, the change in the orientation of the principal axes of non-hydrostatic moment of inertia is instantaneous, and those subsequently do not change at all. The similar amplitude and synchronicity of these five continents indicate their individual plate motions are negligible relative to the shared TPW motion. Numerical simulations indicate such a change in the orientation of the principal axes of non-hydrostatic moment of inertia can be completed within 10 Ma46. There is also a notable absence of poles in between the before/after poles recording the TPW shift (Fig. 3). These systematic gaps in the APW paths of all continents are consistent with the stroboscopic effect expected for TPW, which is a non-linear process that speeds up and slows down, thus rendering it less likely for rocks to form (making them available for palaeomagnetic sampling) during the peak rate of TPW in the middle of the event. A simple simulation (Methods) demonstrates that it is 20 times less likely to sample TPW “in action” than the endpoints largely before/after the TPW event (Fig. 4). This inherent bias can explain why the O–S TPW event is sampled exclusively by endpoints for all continents. We therefore confirm and refine the original proposal27 of a large amplitude ~50° TPW event occurring across the O–S boundary.
Fig. 4: Probability of sampling true polar wander. a True polar wander (TPW) angle as a function of time with an initial condition of 25°. b TPW speed (black line) and probability function (shaded gray) as a function of time. Full size image
Given this was a time of major plate tectonic reorganization in between assembly of megacontinent Gondwana and its larger supercontinent Pangaea51, there is no shortage of potential sources of subduction-related mass anomalies that might have provided the excitation for the large-scale TPW event across the O–S boundary. The Australian Tasmanides, the Laurentian Appalachians, and the Baltic Caledonides were all active at this time; however, provided their positions relatively close the TPW axis (I min ), their influence on Earth’s rotation would have been dampened compared to mass anomalies elsewhere. In contrast, both the Proto-Tethyan and Terra Australis subduction systems on either side of Gondwana were ~90° away from I min and thus in the plane of TPW containing I max and I int would have been ideally positioned relative to Earth’s prolate non-hydrostatic figure to have excited large-scale TPW.
In the Late Ordovician, the Proto-Tethyan system experienced a fundamental shift from subduction to collision52. Both the timing (pre-TPW) and the sense of this change in slab dynamics—with the foundering oceanic slab likely ponding at the mantle transition zone, thus causing a positive anomaly in the geoid kernel driving TPW for this region equatorward53—are consistent with the observed palaeogeographic shift of the Tethyan subduction zone from mid-latitudes into the tropics (Fig. 5). Also, in the Terra Australis system on the other side of Gondwana, an intriguing coincidence is that the new position of the South Pole (post-TPW) becomes centered on the Antarctica–South America segment of the subduction system (Fig. 5) that experienced a dramatic shift from negative to positive hafnium isotopes at this age54. Such a shift due to increased mantle-derived magmatism in the arc indicates slab retreat, which can occur before slab break-off as a slab meets resistance to subduction after impinging the mantle transition zone55. Because of the time lag between slab subduction in the upper mantle and its penetration into the lower mantle, a dramatic slab avalanche from the upper into the lower mantle after stagnation at the mantle transition zone could thus conveniently explain the new pole position assumed in the Silurian as the geodynamic change in the Terra Australis would have driven TPW for this region poleward53. Thus, the dramatic changes in slab dynamics of both subduction systems on either side of Gondwana could have contributed to the collective forcing behind the largest TPW event in the past 500 million years.
Fig. 5: Palaeogeographic reconstructions based on Ordovician–Silurian true polar wander. Reconstructions for: (a) 460–450 Ma, (b) 445 Ma, and (c) 440 Ma. Palaeomagnetic poles are color-coded as in Fig. 3a. I min , minimum moment of inertia (equatorial true polar wander axis of rotation). Palaeomagnetic poles (and associated continents) of each age are rotated to coincide with the South Pole, although a 5–10° range of flexibility is occasional adopted as is common praxis in ancient palaeomagnetic reconstructions. The latitudinal band between 15° north and south of the Equator is assigned as the humid tropical zone with intensive chemical weathering. Maps are shown in Mollweide projection. For better displaying their distribution, we fixed the continents and rotate the Mollweide projection to fit the Palaeo-south pole. All plots were generated with GPlates73. Full size image
It is also possible that the waxing and waning of ice sheets across Gondwana contributed to the mass anomalies driving O–S TPW, or there was some feedback between TPW and glaciation. In particular, there is a migration of glacial centers from northern Africa to southern Africa–South America, where glacial and periglacial strata in the former region are predominantly Ordovician and those in the latter neighboring regions are predominantly latest Ordovician or Silurian15. That is, the mass load associated with incipient Ordovician glaciation applied in northern Africa could have been driven to the equator by TPW, causing southern Africa–South America to move to the pole and thus moving the glacial center there in the earliest Silurian (Fig. 5). This hypothesis, by extension, would also predict ensuing oscillatory Silurian–Devonian TPW back in the direction of northern Africa (in order to drive the glacial center in southern Africa–South America equatorward), which has indeed been previously hypothesized27, but the assessment of which is beyond the scope of our study on O–S TPW. In the Cenozoic, however, glaciation is typically regarded more as an effect of TPW rather than a cause of it56, as the amplitude of glacially induced TPW is smaller than TPW driven by mass reorganizations in the mantle56. Nevertheless, given the larger size of the Paleozoic pan-Gondwanan ice sheet, and thus its presumably larger mass load, glacial loading deserves further investigation for potentially driving the O–S TPW event. If valid, such an interpretation—the incipient glacial load causing TPW, which then led to more severe glaciation as Gondwana became centered over the South Pole—presents a fascinating potential feedback between TPW and glaciation.
Palaeogeographic reconstructions based on true polar wander
Traditionally, the superposition of APW paths is used to reconstruct the configuration of different continents during time intervals of supercontinentality57. However, during times of plate tectonic reorganization in between supercontinents, this method cannot be used to reconstruct isolated continents that are in relative motion, which is most likely how the end-Ordovician world was kinematically configured12. Nonetheless, when APW is predominantly driven not by plate motion but by TPW, then the superposition of APW paths can be used to determine the relative positions of different continents whether they are united or isolated because the TPW motion is shared by all continents and thus provides a common global reference frame26,58. Such an APW comparison only requires a minimum of two poles from before and after the TPW event. Therefore, we can accurately reconstruct global paleogeography of the major continents across the O–S boundary by leveraging TPW.
To make our reconstructions, northwest Africa is fixed and all the other continents are rotated into northwest African coordinates (Euler rotation parameters listed in Supplementary Table 4). We first fitted a great circle to the palaeopoles from Gondwana, of which northwest Africa is a part (Fig. 5a). Poles from all the other continents were then rotated to overlap the Gondwanan poles at their corresponding ages. The TPW-based reconstructions constrain the relative positions of all these continents during 460–440 Ma, not only including palaeolatitude constraints, but also commonly unconstrained relative palaeolongitude. As mentioned, the essentially opposite tectonic motion of Laurentia effectively cancels out some of the TPW rotation for Laurentia, therefore its position relative to other continents changes over time.
Three high-resolution, TPW-based palaeogeographic reconstructions are provided at 460–450, 445, and 440 Ma (Fig. 5). The 445 Ma reconstruction is an interpolated position between 450 and 440 Ma. Subduction zones and the evolution of Avalonia is simplified from Cocks and Torsvik12. A salient difference between our reconstructions and previous ones5,6,8,12,59 is that Gondwana rapidly swept over the South Pole (Figs. 5, 6e). Meanwhile, during 460–450 Ma, the Niger–Chad zone was located at the South Pole rather than the Morocco–Algeria zone (Fig. 5a). At 460–450 Ma, Gondwana was distributed from the South Pole to the Equator, with the majority of the landmass located at high-to-mid latitudes (Fig. 5a). Laurentia straddled the equator, with its east coast (present coordinates) outside of the tropics. The positions of Baltica and Siberia are similar to previous reconstructions5,6,12,59. Constrained using the APW path of Tarim, Turan–Karakum–Tarim–North China60 is constrained to a position between South China and Siberia. Most of South China was in the tropics, which is consistent with the palaeoequatorial setting suggested by the mega-nodular limestone, a time-specific carbonate facies61.
Fig. 6: Late Ordovician–early Silurian global change and true polar wander. a Temporal distribution of the palaeomagnetic sampling horizons by Formation (Fm) from South China. Snowflake above the Hirnantian stage indicates the age of short, sharp glacial advance3. b δ13C record (black) and sea surface temperature variation (red) from Rasmussen et al.7 c Biodiversity during the Sanbian–Telychian stages from Deng et al.11 (purple line) and Rasmussen et al.7 (blue line). CR capture–recapture modeling. d Rates of origination (blue) and extinction (magenta) with 1 million year age binning from Deng et al.11. e Palaeolatitudinal variation of a reference point in central Gondwana (12°S, 10°E) calculated by using 460–430 Ma Gondwana palaeopoles listed in Supplementary Table 2. Note that the 430 Ma palaeolatitude is not displayed. Orange vertical bars are intervals of 95% confidence. Full size image
After 450 Ma, TPW initiated a dramatic change in palaeogeography. At 445 Ma, in the middle of the TPW event, northern Africa moved off the South Pole, where it was replaced by southern Africa and South America (Fig. 5b). (In terms of tectonic motions, Laurentia moved closer to Baltica, but farther from Gondwana because of the fast opening of the Rheic Ocean.) During the TPW event, Baltica and Avalonia moved into low latitudes, and Siberia, Turan–Karakum–Tarim–North China, and South China ended up straddling the equator and were nearly all located within the tropics (Fig. 5b). After the TPW event was over by 440 Ma (the Silurian), northern Africa and Arabia occupied low latitudes, while South America and southern Africa were located around the South Pole (Figs. 5c, 6e). Siberia and Turan–Karakum–Tarim–North China all moved out of the tropics, while Baltica moved into the tropics and South China moved into the tropics of the northern hemisphere. By the Silurian, more continents were positioned at mid-to-low latitudes (more than ~14,000,000 km2; Fig. 5c) than before (Fig. 5a, b).