The Tharsis bulge is the largest volcano-tectonic centre on Mars. Its growth started during the Noachian epoch (>3.7 billion years ago)2 and the associated enormous transfer of mass, energy and release of volatiles from the mantle had implications for the planet’s evolution, including its climate, surface environment and mantle dynamics. The earliest signs of activity are limited to Noachian extensional tectonics observed around Claritas Fossae in the ancient Syria Planum, Thaumasia and Tempe Terra regions2. Assuming a 100-km-thick elastic spherical shell, models of the topographic effect of the Tharsis load have suggested that most of the bulge was largely in place by the end of the Noachian epoch. If so, it could have influenced the orientation of valley networks1. However, the elastic lithosphere thickness at the Noachian according to modern estimates3 (<30 km) corresponds to weak support of the Tharsis load by lithospheric flexure. In fact, early completion of the Tharsis bulge is questionable in light of recent studies: the formation of radial and concentric structures around Tharsis (including Valles Marineris) indicates complex and multi-stage growth of the bulge from the late Noachian to the Amazonian2; the inferred crust thickness at the time of formation of the Coracis Fossae rift is evidence of late (<3.5 billion years ago) crustal growth in the Tharsis region4; and tectonic structures initially assigned to the Noachian era2 eventually also affect Hesperian units (Extended Data Fig. 1). Thaumasia and Claritas Fossae are the only Noachian units of Tharsis above the average elevation of the southern hemisphere of Mars (Extended Data Fig. 1). However, these regions at the southern edge of Tharsis have remnant crustal magnetization, in contrast with the rest of the magmatic province, and their formation has been related to a pre-Tharsis orogenic event5.

The rotation axis orientation of a planet is constrained by its degree-2 gravity field, which is directly related to non-spherical contributions to the inertia tensor. Before Tharsis, the hemispheric dichotomy of the topography controlled the orientation of the planet relative to its spin axis6. The growth of Tharsis must have induced a reorientation of the planet with respect to its spin axis (TPW) moving the mass anomaly to its current position close to the equator6,7,8,9,10. The rotation poles before Tharsis’ formation and TPW (palaeo poles) can be constrained by removing the Tharsis and remnant rotational figure contributions to the degree-2 gravity field. This yields a palaeo north pole at (100.5° ± 49.5° W, N) and a palaeo south pole at (79.5° ± 49.5° E, S)10, near the centre of the crustal dichotomy (80° E, 60° S)9.

In addition to a rotation of the geographic reference, TPW induces an adjustment of the topography as a result of a new equilibrium rotational figure10. A large TPW event (20°–25°) such as the one triggered by the formation of Tharsis may have had global implications. For instance, the distribution of surface landforms influenced by local slopes or early Mars climate simulations should be calculated within a reference frame corresponding to the planet’s configuration at the time of the formation of such landforms. Valley networks, incised during an early climate distinct from the present climate, are mainly located on the highlands11,12 within a domain of latitudes ranging from −60° S to +30° N (Fig. 1a). In the present reference frame, the density and spread in latitude of Early Hesperian/Noachian valley networks show large variations with longitude12. The valley network distribution is dominated by local or regional slopes rather than by elevations12,13. However, no combination of topographic parameters is able to explain the lack of valley networks between 30° S and 60° S from the Argyre basin to the east of the Hellas basin and the very low density in Arabia Terra (Fig. 1). In fact, the valley networks occur in a band that appears to follow a small circle tilted with respect to the present equator. We have quantitatively confirmed this observation by least-squares adjustment of the valley network density map to a small circle (Methods). The normal vector of the plane containing the small circle defines a north pole located north of Tharsis along the meridian passing through the centre of the Tharsis bulge (118° W, 69° N) and a south pole south of the Hellas basin (62° E, 69° S). This rotation axis is consistent with the calculated palaeo north pole location before the formation of Tharsis and TPW (100.5° ± 49.5° W, N; ref. 10). The proximity of the palaeo north pole inferred from the valley network distribution and the one inferred from degree-2 spherical harmonic gravity without Tharsis10 is remarkable because these results are independent.

Figure 1: Noachian/Early Hesperian valley networks distribution and density12 before and after TPW. a, In the present reference frame. The palaeo pole positions are indicated with a diamond, with the spread in longitude and latitude for solutions shaded from black (14°) to red (15°), corresponding to the associated root-mean-square value for each solution, as given by equation (16) (see Methods). The white dashed line is the pre-TPW equator. b, In the pre-TPW configuration. Valley networks occur within a latitudinal band of ±14° centred at 24° S. The black dashed line is the present equator. Full size image Download PowerPoint slide

In the pre-TPW configuration, the valley network distribution defines a latitudinal band of ±14° centred at 24° S, that is, in the south tropical regions (Fig. 1b). The incision of valley networks spans the Hesperian–Noachian (>3.5 billion years ago) period14,15 and it is unclear whether the valley network formation or the TPW event caused by the Tharsis rise occurred first. To investigate whether the Tharsis bulge controlled the direction of the valley network, we modelled the stream network in a latitudinal band (from 40° S to the dichotomy) in the pre-TPW reference frame for a topography without Tharsis and before TPW at 1° per pixel (available in the Supplementary Information; see also Methods) and for a present-day topography with Tharsis at the same resolution. The directions of large-scale valley networks in the pre-TPW and in the present configuration are both compatible with the observed directions of valley network (Extended Data Figs 1 and 2). In both configurations valley networks are oriented towards the north, reflecting the topographic dichotomy of the Martian surface. This result indicates that the presence of the Tharsis bulge is not necessary to explain their orientation. However, the occurrence of valley networks within a palaeo tropical band, between the equator and 40° S, subject to precipitation (ice, snow or rainfall), supports a post-incision Tharsis-driven TPW during the Early/Late Hesperian period.

Early Mars climate simulations16,17,18 assuming the present topography predict patchy ice/water accumulation in a south tropical band for a cold/icy scenario17,18 and precipitation around Hellas basin and in Tharsis region for warm/wet conditions18. Both kinds of simulation fail to explain the occurrence of valley networks down to 45° S at the east of Hellas and down to 60° S at the south of Tharsis. They also predict substantial rain/snowfall in the west of Tharsis. This prediction does not match the lack of valley networks in this region. We performed new simulations using the same model and conditions (see Methods) using a pre-TPW topography. We found that ice tends to stabilize and accumulate in a tropical band (Fig. 2) similar to the distribution of valley networks (Fig. 1b), as a result of enhanced precipitation induced by adiabatic cooling when the atmospheric circulation transports water vapour southward, up to the highlands. An icy patch is also predicted in the Tharsis region but recent volcanic flows preclude the preservation of ancient morphologies in this region. Intensity of drainage is affected by several factors and is not only related to the intensity of precipitation. The geological history subsequent to valley network formation may also be responsible for heterogeneous modifications of the palaeo drainage intensities.

Figure 2: Permanent ice deposits predicted by the global climate model for early Mars, with obliquity 45°, a circular orbit and mean surface pressure ~0.2 bar. a, With present-day topography. b, With topography before TPW, without the Tharsis bulge. These maps show the yearly minimum of surface ice in simulations with just enough ice in the system to simulate the location where the ice stabilizes under a specific configuration. Elevations are given in kilometres, with solid lines representing isoaltitudes equal to or higher than 0 km and dashed lines representing isoaltitudes lower than 0 km. Full size image Download PowerPoint slide

Can we find any geological clues of a past polar climate at the location of the palaeo poles? The palaeo north pole is located in Scandia Colles (Extended Data Fig. 4), a knobby terrain that probably represents one of the rare Noachian units in the north polar region19. Interestingly, some of the landforms of this region have been attributed to Late Hesperian polar ice retreat or melting20,21, which could be the result of the displacement of the pole at this time. This palaeo north pole is located within the unique region of the northern plains where the SHARAD (Shallow Subsurface Radar) instrument has detected the base of ground ice at a depth of several tens of metres22 that could correspond to the remnant of an ancient ice cap. Similarly, the southern hemisphere radar interface detected in the Dorsa Argentea region has been attributed to the remnant of an Hesperian-age buried ice cap23. The palaeo north pole is also surrounded by an area of anomalous neutron counting rates24 associated with high hydrogen content (indicating water) in the first metre below the surface (Extended Data Fig. 5). Although the formation of this reservoir is likely to be recent, it may be related to the slow sublimation of deeper and older ice reservoirs favouring the stability of ice near the surface.

The palaeo south pole is located in the Noachian volcanic unit, Malea Planum. Some deposits in this region (Pityusa Patera) are possible remnants of an extensive polar ice sheet emplaced in the late Noachian to mid-Hesperian25. Erosional lineated terrains26 and faintly dentritic Hesperian valley networks (Axius Valles and Mad Vallis)25 are other features expected in a context of erosion by meltwater associated with retreating ice sheets25,27. A higher concentration of double-layered ejecta in this unit28 also suggests the presence of an ice-rich substratum28,29 that might be related to the polar palaeo location of this unit.

In summary, we consider that the planet was initially oriented such that the dichotomy was parallel to the equator (Fig. 3a) during the Noachian and the Early Hesperian periods6. Precipitation occurred in a tropical band during these periods. The incision of valley networks was contemporaneous with an initial phase of magmato-tectonic activity at Tharsis (Fig. 3b). The prolonged magmatic activity during the Hesperian moved Tharsis close to the present equator after the end of most of the fluvial activity (Fig. 3c). This TPW event may have generated stresses that were large enough to produce a global tectonic pattern (Extended Data Fig. 6). Global stress and tectonic pattern calculations show that extensional features oriented roughly north–south should be observed around the palaeo poles7,30. These fault patterns are not observed7. However, the identification of the expected global tectonic pattern may be complicated by additional stresses generated by Tharsis loading or the possibility of stress relaxation30. Stress relaxation may arise as a result of the expected slow TPW speed of about 1° per million years31,32.

Figure 3: Scenario for a TPW driven by a late growth of Tharsis contemporaneous with valley network incision. a, During the Early Noachian period, the planet was initially oriented such that the dichotomy was parallel to the equator6. b, Precipitation occurred in a tropical band during the Noachian and the Early Hesperian periods and was contemporaneous with the growth of Tharsis. c, The prolonged magmatic activity at Tharsis during the Hesperian period drove a slow TPW and moved Tharsis close to the present equator after the end of most of the fluvial activity. Full size image Download PowerPoint slide

Causal links between volcanic outgassing, build-up of atmospheric pressure and precipitation have already been considered, but the valley networks were thought to have formed after the emplacement of most of the Tharsis load33. In light of our results, we conclude that precipitation occurred during the birth and growth of the Tharsis bulge rather than when its activity was declining. Such a scenario is more plausible than the post-Tharsis formation of valley networks, considering early degassing and progressive depletion of volatiles in the mantle source. The calculated pre-Tharsis topographic map of Mars provides a framework within which to examine the first billion years of its geological history.