The Martian surface is characterized by a dichotomy in elevation, crustal thickness and morphology between the northern lowlands and southern highlands (Platz et al., 2013;Watters et al., 2007). The highlands are highly cratered and incised by many old river valleys (Tanaka et al., 2014). The oldest terrains outcrop in the Southern hemisphere (Platz et al., 2013) which also concentrates observations of volcanic constructs and felsic rocks (Carter & Poulet, 2013;Wray et al., 2013). In contrast, the upper portion of the lowlands crust is made of vast and smooth basaltic plains and sedimentary deposits derived from the erosion of the highlands (Platz et al., 2013;Tanaka et al., 2014). The dichotomy is perhaps the most evident feature of the crust (Figure 1). Superposed on the dichotomy boundary is the Tharsis bulge, whose origin may (Andrews-Hanna et al., 2008), or may not (Neumann et al., 2004), be related to it. Prior to the InSight mission, inversions of gravity and topography data were used to constrain the crustal thickness of Mars and its lateral variations, though these models depended on assumptions such as crustal density and minimum crustal thickness (Neumann et al., 2004;.

<p>The Martian crust is made up of sedimentary and volcanic rocks that are mainly mafic in composition. Nevertheless, orbital and in-situ observations have revealed the presence of felsic rocks (Payré et al, 2022), all located in the southern hemisphere, where the crust is thicker. These rocks likely formed by differentiation of a basic protolith. On Earth, this process occurs at plate boundaries and is linked to active plate tectonics. But on Mars, we have no evidence of active or ancient plate tectonics.</p> <p>On one-plate planets, there exists a positive feedback mechanism on crustal growth: the crust being enriched in heat-producing elements, the lithosphere is hotter and thinner where the crust is thicker, which implies a larger melt fraction at depth and therefore a larger extraction rate and a larger crustal thickening where the crust is thicker. We proposed that this mechanism could have been at the origin of the Martian dichotomy (Bonnet Gibet et al, 2022). This mechanism further implies that regions of thicker crusts, characterized by a larger amount of heat sources, a thinner lithosphere and an increased magmatism, are also marked by higher temperatures. Here we investigate whether crustal temperatures in regions of thick crust may be maintained above the basalt solidus temperature during crust construction, which would allow for the formation of partially molten zones in the crust and hence differentiated rocks by extraction of the melt enriched in water and silica. In this scenario, felsic rock formation would be concomitant to crustal construction and dichotomy formation on Mars.</p> <p>We use a bi-hemispheric parameterized thermal evolution model with a well-mixed mantle topped by two different lithospheres (North and South) and we account for crustal extraction and magmatism in these two hemispheres. We formulate a Bayesian inverse problem in order to estimate the possible scenarios of thermal evolution that are compatible with constraints on crustal thickness and dichotomy amplitude derived from the InSight NASA mission. The solution is represented by a probability distribution representing the distribution on the model parameters and evolution scenarios. This distribution is sampled with a Markov chain Monte Carlo algorithm, and shows that a non-negligible range of scenarios allows for partial melting at the base of the Southern crust below the Highlands during the first Gyr of Mars' evolution. On the contrary, partial melting of the base of the northern crust is insignificant. Models that fit InSight constraints and allow for differentiation of a fraction of the Southern crust point to a relatively low reference viscosity (~10<sup>20</sup> Pa.s) that can be explained by a wet mantle at the time of crust extraction.</p>