Rheological inheritance controls the formation of segmented rifted margins in cratonic lithosphere

Observations from rifted margins reveal that significant structural and crustal variability develops through the process of continental extension and breakup. While a clear link exists between distinct margin structural domains and specific phases of rifting, the origin of strong segmentation along the length of margins remains relatively ambiguous and may reflect multiple competing factors. Given that rifting frequently initiates on heterogenous basements with a complex tectonic history, the role of structural inheritance and shear zone reactivation is frequently examined. However, the link between large-scale variations in lithospheric structure and rheology and 3-D rifted margin geometries remains relatively unconstrained. Here, we use 3-D thermo-mechanical simulations of continental rifting, constrained by observations from the Labrador Sea, to unravel the effects of inherited variable lithospheric properties on margin segmentation. The modelling results demonstrate that variations in the initial crustal and lithospheric thickness, composition, and rheology produce sharp gradients in rifted margin width, the timing of breakup and its magmatic budget, leading to strong margin segmentation.

The present-day crustal architecture, depicted by seismic data and gravity modelling, suggest an asymmetric rifting in the Labrador Sea, which displays wider shelf and necking domains on the Canadian side than on the Greenland side 1-3 . Ref. 4 suggests that the line of breakup was closer to the Greenland side (i.e., upper plate) than the Labrador side (i.e., lower plate), which means that most the syn-rift stretching is preserved on the Canadian side of the basin. Overall, stretching in the Labrador Sea appears to be characterized by slow crustal extension rates then higher mantle exhumation rates.
Various ages of continental breakup were proposed in the Labrador Sea, ranging from Turonian (ca. 92 Ma), based on a stratigraphic unconformity 5 , to Palaeocene (ca. 62 Ma), corresponding to the magnetic anomalies of Chron 27 6 . Ref. 1 used seismic reflection lines (to identify the extent of oceanic crust) combined with magnetic chrons 7 to propose a diachronous continental breakup younging northward. They suggest that oceanic accretion started ca. 65. 8-64.4 Ma in the south (i.e., Chron 29), ca. 63.3-61.1 Ma in the centre (i.e., between Chron 29 and Chron 27), and ca. 60.5-57.7 Ma in the north (i.e., Chron 26). The latter coincides with the mid to late Palaeocene (ca. 61-56 Ma) flood basalts, which are found around the Davis Strait and attributed to the Iceland plume 8,9 . Seafloor spreading in the Labrador Sea experienced a change in direction from NE-SW to N-S at ca. 60 Ma 10 . Then at ca. 50 Ma spreading rate decreased from 10 to 3-4 mm/yr (half spreading rates) before it completely ceased at the Eocene-Oligocene boundary 10 (ca. 34 Ma).

Note 2: Surface heat flow onshore the Labrador margin
Data from the east Canadian Shield, onshore the Labrador margin. show a noticeable southward increase in surface heat flow 11 . Surface heat flow values range between ca. 22-27 mW/m 2 in the Churchill and Nain provinces in the north, 27-37 mW/m 2 in the Makkovik province in the centre, and 27-47 mW/m 2 in the Grenville province in the south. Variations in heat flow can be explained by changes in crustal radiogenic heat production (i.e., differences in crustal thickness and/or composition), while regional-scale variations are related to changes in lithospheric mantle heat flow 12,13 .
The southward increase in surface heat flow onshore Labrador is consistent with the observed increase and decrease in crustal and lithospheric thickness south of the Grenville Front, respectively. Lateral variations in shear velocities across the Grenville Front at depths ranging between 80 and 150 km also suggests changes in temperature (and composition) of the deep mantle lithosphere 14,15 . The Grenville Front was a major suture zone along which the Superior province was subducted toward the SE underneath the Grenville province, giving rise to calc-alkaline arcs 16 (ca. 1.68 to 1.66 Ga). This magmatism has resulted in the depletion of the mantle lithosphere underneath the Grenville domain, such as attested by the distribution of post subduction magmatism which is found on both sides of the Labrador Sea but mostly north of the Grenville Front 17 .

Note 3: Initial thermal structure of the numerical experiments
Each of the three segments of the Labrador Sea is represented by a distinct lithospheric structure setup, which is constrained by observations from the Labrador Sea and the Canadian Shield 1 with fixed initial model geometry, temperature structure, mantle rheology, and extension rate (Fig. 2). We assume a weak (fertile) mantle in the north and central segments. which is governed by wet olivine flow low 18 , while in the southern segment we use a strong (depleted) mantle lithosphere defined by dry olivine flow law and a weak (fertile) asthenosphere defined by wet olivine flow law 18 .
The initial thermal structure of each model is calculated using the following thermal gradient equation 12 : where T(z) is the temperature at a given depth (z), Tt is temperature at the top of the layer, qt is the heat flow at the top of the layer, qb is the heat flow at the base of the layer, A is the radiogenic heat production of the layer, k is the thermal conductivity of the layer, and Δz is the thickness of the layer.  Table 2). The initial strength of the lithosphere depends on the composition of the lower crust, which is governed either by wet quartzite 19 (solid grey line) or wet anorthite 20 (dashed grey line) creep laws. Our models assume a mantle lithosphere governed by dry olivine flow law 18 in the southern segment (light green) and wet olivine flow law 18 in the central and northern segments (dark green). Geochemical evidence suggests that the continental mantle lithosphere beneath cratons may be compositionally depleted in some locations relative to the Asthenospehre 21,22 . While seismic evidence points toward a more complicated two-layer lithospheric mantle structure beneath cratons, here we represent the lithospheric mantle with a single layer that is 50 kg/m3 less buoyant than the asthenosphere at the equivalent temperature. Future work may explore 3-D simulations with a two-layer lithospheric mantle density and rheology structure, which previous 2-D modelling efforts 23 explored with depleted continental mantle lithosphere densities 50-80 kg/m3 lower than reference mantle densities.  Thermal diffusivity (m 2 s -1 ) 1.481481e-6 1.40350e-6 1.333333e-6 1.230769e-6 1.212121e-6