We advocated a first point several years ago (Cogné et al., 1999). Should a compaction induced inclination flattening mechanism in sediments follow a simple law, it would depend not only on the flattening amount in sediments, but also on the initial inclination of the magnetic field. Therefore, the anomaly should be related to the paleolatitudes at which sediments were formed. A very simple law of the form:

With a volume loss in sandstones of ∼0.2–0.4 (Baldwin and Butler, 1985; Lemée and Guéguen, 1996) and an a coefficient of 1.3 (Arason and Levi, 1990a,b), values of f = 0.3 to f = 0.5 were judged as reasonable bounds of a compaction induced ΔI in continental Meso-Cenozoic sediments from central Asia. The resulting curves of ΔI as a function of latitude in Asia (e.g., Cogné et al., 1999; Fig. 2a) are in disagreement with Cretaceous experimental values obtained in localities spread from Tibet to Central Asia localities (e.g., from south to north; points #1 to 8 and yellow qualitative patch on Fig. 2a). As a matter of fact, in the southernmost localities, ΔI is the largest where it should have been the smallest, whereas it is the contrary in the northern sites. Thus, our experimental data of Mesozoic sedimentary localities from Tibet to central Asia fail to demonstrate this latitudinal dependence of the inclination anomaly, at least in these examples.

A second line of evidence further arguing against sedimentary processes emerges from the recent analysis of the paleomagnetic database (Cogné et al., 2013). A significant number of paleomagnetic studies in Asia come from igneous rocks, such as basalts, in which no significant compaction or sedimentary effects on magnetic recording are usually suspected. The inclination anomaly also exists in these studies, no more but no less than in sedimentary localities of eastern Asia. In effect, the Cenozoic inclination anomaly has been obtained in basalts from Mongolia (Hankard et al., 2007). Furthermore, this anomaly is demonstrated in all stable blocks of eastern Asia, including the North and South China Blocks (NCB and SCB), Korea and Amuria terranes, and in the Siberian craton (e.g., see the review by Cogné et al., 2013), regardless of the igneous or sedimentary nature of sampled localities.

In order to check this observation, we have computed a new APWP solely based on paleomagnetic data from East Asia (e.g., North and South China blocks, Amuria terrane, Korean Peninsula, and Siberian Craton), still based on the dipolar field hypothesis. We have then re-computed the Δλ differences between measured and modeled paleolatitudes at each locality (Fig. 2b), using this new East Asia APWP (e.g., Cogné et al., 2013). The anomaly disappears in these eastern Asia terranes (Fig. 2b). The residual Δλ differences are statistically identical for data from sediments (−0.8° ± 6.7°, n = 80; Fig. 2b, left) and from igneous rocks (0.2° ± 6.2°, n = 66; Fig. 2b, right). Moreover, these values are not significantly different from zero.

This observation also applies to data from the Japanese islands only, which were not included in the computation of the East Asia APWP. In effect, both sedimentary and effusive paleomagnetic results from Japan demonstrate an “inclination anomaly” since 20 Myr (Fig. 2c, left). We notice that the anomaly is of the same order of magnitude for sedimentary and igneous rock formations. This is again confirmed when computing the residual Δλ anomaly using the new East Asia APWP (Fig. 2c, right): the residual discrepancy between predicted and observed paleolatitudes averages −2.0° ± 5.2° for sediments and −0.2° ± 4.3° for igneous rocks. Here again, these values are not statistically different either from each other or from zero. Therefore, this Japanese example also demonstrates that, within uncertainties, no statistically significant bias in magnetic field recording can be evidenced in sedimentary beds with respect to effusive rocks.

In our analyses and compilations, these studies from East Asia show that residual Δλ values for sediments are always lower than for igneous rocks. This may be interpreted as the sign of a slightly shallower inclination recorded in sediments, in the order of ∼1–2° in terms of paleolatitudes. However, this is still well under the 2σ error limits of ∼5° on these statistics; thus it carries no robust statistical significance.

Altogether, this comparison of paleomagnetic results from sedimentary vs. igneous rocks, based on the analysis of the paleomagnetic database from Asia, leads us to conclude that sedimentary processes during diagenesis and/or deposition are not the first-order mechanism responsible for the widespread Cenozoic inclination anomaly in Asia (Fig. 1a), though in some peculiar cases, with some peculiar lithologies, such a mechanism can actually be advocated.

An anomaly of the Tertiary dipolar geomagnetic field in Asia was first proposed by Westphal (1993) from an analysis of Eocene paleomagnetic data from Iberia to Afghanistan. Based on information available at that time, the author proposed that a transitory dipole, inclined with respect to the Earth rotation axis, could have accounted for a ∼18° latitude anomaly in Eurasia during the Eocene. Shortly after, Chauvin et al. (1996) proposed a map of the Tertiary inclination anomaly over Asia with a common mechanism for all studied formations. Disputing the hypothesis of an inclined dipole, they proposed that the persistence of long-lasting non-dipolar magnetic field components could actually account for the overall Eurasian anomaly. In particular, they suggested that the contribution of an axial octupolar component G3 = g₃⁰/g₁⁰, as high as ∼10%–15%, could account for the ∼10°–14° anomaly of inclination at 35°N latitude. However, the authors pointed out that such high values of non-dipolar components should be considered as abnormal over several tens of millions of years. Si and Van der Voo (2001) pursued the same idea and, through different tests over Asian data, reached the same conclusion that a long-lasting G3 = g₃⁰/g₁⁰ contribution on the order of ∼10%–15% could explain a large part of the Cenozoic inclination anomaly. These authors, however, still did not provide explanations for such long-lasting persistence of a non-dipolar magnetic field component in the Cenozoic at the global scale.

We illustrate this hypothesis with the paleolatitude anomaly map of Fig. 1b. We have constructed this map using the reference European APWP of Besse and Courtillot (2002), and introducing a non-dipolar G3 = g₃⁰/g₁⁰ contribution in the computation of paleolatitude differences Δλ (Livermore et al., 1983; Merrill et al., 1998). Since the paleolatitude anomaly appears to have a maximum at ∼50–40 Ma, progressively diminishing through time (e.g., 20 Ma) then vanishing at ∼10 Ma (e.g., Hankard et al., 2007), we have modeled a 15% contribution of G3 for 30–50 Ma-old data, 7.5% for 10–30 Ma-old data, and 0% for 0–10 Ma-old data. The map on Fig. 1b suggests that a long-lasting octupolar contribution to the global magnetic field over the last 50 Myr can indeed account for the anomaly over all of eastern Asia. In effect, except for the Tibetan and Central Asian terranes, the East Asian regions are colored light yellow to light green, i.e. has a zero (± 5°) paleolatitude anomaly (Fig. 1b).

Although this non-dipolar model could account for the inclination anomaly, the idea of a long-lasting g₃⁰ term in the geomagnetic field is still far from having been proven on independent lines of evidence. In particular, we emphasize that Courtillot and Besse (2004) proposed that this octupolar term was, on the average, not in excess of a statistically insignificant value of 3 ± 8% of the dipolar field during the last 200 Myr. Still based on the dipolar hypothesis, we therefore consider that the European APWP is not the proper reference.

We propose to interpret the paleolatitude anomaly of eastern Asia as due to the inappropriateness of the European reference APWP in predicting the paleolatitude of the eastern end of the Eurasian plate in Cenozoic times. We have produced an independent APWP of eastern Asia from 130 Ma to the Present (Cogné et al., 2013; Fig. 3a), based on paleomagnetic results from the North and South China blocks, the Korean peninsula, Amuria terrane and Siberia (NCB, SCB, KOR, AMU and SIB, Fig. 3c), assumed to be the reasonably stable components of this highly mobile system.

This APWP (closed symbols on Fig. 3a) is obviously different from the European APWP of Besse and Courtillot (2002; open symbols on Fig. 3a). More precisely, this difference resides in a far-sided location of eastern Asia Cenozoic paleopoles with respect to the poles of the European reference APWP, as seen from Asian site locations (grey sector on the right side of Fig. 3a). On the average, this difference amounts to 9.2° ± 1.8° for the 10 Ma–60 Ma period, quite similar to the ∼10° paleolatitude anomaly observed over eastern Asia in the map of Fig. 1a. As a consequence, computing the Δλ paleolatitude using the East Asian APWP as a reference, while still assuming a dipolar magnetic field hypothesis, resolves the paleolatitude anomaly of the eastern regions of Asia (Fig. 1c). In effect, as was the case for the map computed under the non-dipolar hypothesis (Fig. 1b), all are covered with light yellow/green colors (Fig. 1c), depicting a zero (± 5°) paleolatitude anomaly.

Large anomalies with hot colors (orange to red, Fig. 1c) remain in Tibet and central Asia, and cold colors (green to blue) over the Indochina peninsula, with a total magnitude between them of ca. ± 15° (Fig. 1c). These anomalies can confidently be attributed to:

• • the ongoing indentation of India into Asia, resulting in intracontinental shortening between Tibet, Central Asia and Siberia;
• • the southeastward extrusion of Indochina Peninsula.

This, however, is not the focus of the present paper and will no longer be discussed here.

Therefore, under the hypothesis of a dipolar geomagnetic field holding through geological times, a specific APWP for East Asia (Fig. 3a) can actually account for the main features of the Cenozoic paleolatitude anomaly over East Asia. Thus, we turn to the discussion of a potential tectonic decoupling between western and eastern Eurasia during at least the Cenozoic.

Cenozoic differential motions between Europe and Siberia/East Asia would result in relative movements that would explain the differences between the reference APWPs of Fig. 3a. On Fig. 3 (b and c), we propose a scenario in which eastern Eurasia (e.g., Siberia and East Asia) follows the new East Asia APWP, locating it ∼10° further south than predicted from both the European APWP Cenozoic poles and the rigid Eurasia plate hypothesis. The bold contour with yellow background in these figures (Fig. 3b) underlines the discrepancy of Eurasian paleolatitudes predicted by the European APWP, to be compared to the paleolatitudes computed following the East Asian APWP (contours with grey background), assumed to be the stable parts of the Asian mobile system in the Cenozoic. Indeed, the inescapable consequence of this hypothesis is that Eurasia can no longer be considered to have been a single rigid plate since at least the beginning of the Cenozoic.

A non-rigid Eurasian plate was put forward by Cogné et al. (1999), but, at that time, no precise tectonic constraints were available as to the locus and ages of tectonic movements accounting for Cenozoic relative movements. In the frame of this hypothesis, relative north–south movements between Europe and Asia could have been concentrated along two discrete zones, the Tornquist–Tesseyre line (TTL) and the Ural Mountains (UM). However, field evidence of Cenozoic deformation along the Tornquist–Tesseyre zone or the Urals Mountains was quite scarce at that time. Since then, however, several studies have documented Cenozoic tectonics within both TTL and UM structures. They can be summarized as follows.

Kopp (1999, 2005, 2007) and Bergerat et al. (2007) identified an Eocene NE–SW shortening across the TTL, which, accounting for the relative azimuth of the TTL, results in dextral strike-slip faulting in this area. The movements we predict in this area (large red and thin black arrows over TTL on Fig. 3 b) are coherent with these field observations.

Geological and structural studies in the southern (Kopp, 2005; Verzhbitskii and Kopp, 2004, 2005) and northern (Kopp, 2007) Urals documented a Cenozoic east–west shortening between the central Kazakhstan block and the East European craton (Puchkov, 1997; Verzhbitskii and Kopp, 2005). Considering the Ural orientation, a sinistral strike-slip component is predicted by our model (e.g., large red arrows and black left-lateral shear arrows over UM on Fig. 3b), as actually described by geological field studies (Kopp, 2005, 2007; Puchkov, 1997).

Although the situation in central Asia is complicated by the Miocene Arabian collision in the south (e.g., Verzhbitskii and Kopp, 2004, hatched grey areas on Fig. 3b), the present-day topography of the Ural Mountains also suggests that it results from a Cenozoic east transpression with a left-lateral shear component.

A last point remains to be clarified: the amounts of continental shortening and respective lateral movements implied by this tectonic model. No precise evaluations, based on either field measurements or paleomagnetic estimates, are yet available. The inferred mid-Eocene to present-day shortening across the Ural is likely more important than across the Tornquist zone, as suggests the difference in relief. Our reconstructions (e.g., Fig. 3b) suggest that intracontinental displacements of a few tens of kilometers (up to ∼200 km?) in shortening and transcurrent movements are localized in those discrete zones, but it might be possible that diffuse continental deformations also exist over Eurasia, from East Europe to Siberia.

We emphasize that such moderate intracontinental movements could account for the large inclination or paleolatitude anomaly observed in the remote parts of eastern Eurasia, implying a more than 1000 km latitude discrepancy. This can be regarded as a huge lever-arm effect and one can understand why it has long remained undetected. Only more precise tectonic and paleomagnetic studies in the middle-eastern Eurasian regions, but also a careful check of the kinematic consistency between East Eurasia and North America in the Arctic, will help fully solving the plate circuit over these northern regions and produce the required amounts of intracontinental shortening across Eurasia.