4.2.3 1400 ºC mantle potential temperature cases
Model 7: MPT = 1400ºC, SHF = 45 mW/m2, full spreading rate = 1 cm/yr. A very asymmetric rift system with a very buoyant mantle lithosphere. On the left flank, anastomosing detachment faults are large and hyperextend the sub-continental mantle for ~50 km, while the crust forms into boudins of a similar scale. On the right, the distance over which the sub- continental mantle is hyperextended is shortened to only ~30 km, with a whole-lithosphere boudins at the continent-ocean transition. Melt crystallization is again especially focused in the shear zones beneath anti-listric detachment faults. Peak partial melting is at 32%. The mechanical oceanic lithosphere is 15 km thick at the seafloor-spreading axis.
Model 8: MPT = 1400ºC, SHF = 55 mW/m2, full spreading rate = 1 cm/yr. Another very asymmetric rift, but both structurally and magmatically. Melt is widespread and production peaks at 35% partial melt and the crystallization of that melt is concentrated on the left flank. The left rift flank also hosts a ~40 km long crustal boudin. Crustal allochthons are widespread throughout but are thicker on the right rift flank. Anastomosing detachment faults present throughout the crust-mantle contact. The mechanical oceanic lithosphere is 14 km thick at the seafloor-spreading axis.
Model 14: MPT = 1400ºC, SHF = 65 mW/m2, full spreading rate = 1 cm/yr. Widespread crustal allochthons across a moderately asymmetric rift. The left flank has a greater degree of asthenospheric mantle exhumation, serpentinization, and subcontinental mantle boudinage. The right flank exhibits more magmatic accretion and possesses a large crustal block derived from the H-block. The melt fraction peaks at 28% and the mechanical oceanic lithosphere is 16 km thick.
Model 16: MPT = 1400ºC, SHF = 75 mW/m2, full spreading rate = 1 cm/yr. This is showing an asymmetric rift with extensive crustal allochthons on the left rift flank. There is less boudinage of the subcontinental mantle and of the crust, limited mostly to the right rift flank. Anastomosing detachment faults are present along the Moho on the left flank where magmatic accretion is active. Peak melt production reaches 27% melt fraction and the mechanical lithosphere is 17 km.
Model 9: MPT = 1400ºC, SHF = 55 mW/m2, full spreading rate = 2 cm/yr. This rift exhibits moderate asymmetry, with the left flank being significantly wider than the right flank and hosting a sequence of sub-continental mantle boudins and mantle domes. Anastomosing detachment faults are prominent along the hyperextended mantle. Crystallization of melt is especially focused in the shear zones beneath detachment faults in the mantle. Partial melting is widespread and reaches 35% and the mechanical oceanic lithosphere is 10 km thick at the seafloor-spreading axis.
Model 10: MPT = 1400ºC, SHF = 65 mW/m2, full spreading rate = 2 cm/yr. Structurally very similar to Model 9, but with wider margins and slightly less partial melting (32%). The mechanical oceanic lithosphere is 11 km thick, melt crystallization is focused in the shear zones of mantle detachment faults. This case is asymmetric, though some of this comes from deformation being localized in two locations due to the absence of melt extraction during sea-floor spreading to create true oceanic crust.
Model 19: MPT = 1400ºC, SHF = 75 mW/m2, full spreading rate = 2 cm/yr. The structure of this case is nearly identical to Model 17: wide margins, widespread crustal allochthons, etc.; but with an elevated melt fraction of 36% and a greater degree of melt accretion. The mechanical oceanic lithosphere is particularly thin here: only 6 km thick.
4.3 Generalized numerical model results
Each of the nineteen models presented in this paper follow similar evolutionary stages, though the timing and spatial relations may be different based on extension rate or thermal conditions. As in Model 1, each rift scenario includes: 1) Symmetrical shear zones dipping a low angle (< 20) away from the rift axis in the sub-continental mantle. These shear zones facilitate the necking of the lithosphere, the upwelling of the asthenosphere, and the focusing of the melt region beneath the rift axis. 2) Seaward dipping detachment faults in the continental crust and high temperature landward dipping detachments faults in the mantle originating between 800ºC and 1000ºC. 3) An anastomosing network of crustal and mantle detachment faults that exhume both sub-continental mantle and asthenosphere. The denuded asthenosphere eventually forms a new layer in the mantle lithosphere that is intruded by pods of recrystallized melt and is bounded at its base by active melt. This layer mechanically becomes part of the oceanic lithosphere in which brittle and ductile mechanical processes occur. We name it the “mechanical oceanic lithosphere.” 4) High temperature landward dipping detachment faults initially sole under the continent and couple with crustal faults to form core complexes. 5) The region of melt production is a diffuse triangle that is focused beneath the rift axis. Increased buoyancy in this region drives the asthenosphere towards the surface, this motion is accommodated by the mantle shear zones described previously. As the asthenosphere moves upward, cools and passes through the solidus due to conductive cooling, melt crystallizes above the melt producing region and then migrates as solidified bodies laterally with the tectonic blocks. The density of crystallized melt increases towards the rift axis. 6) Below the melt production region, latent heat of fusion of the melt causes the 1300ºC isotherm to be depressed, marking what we refer to as the “thermal lithosphere” through the rest of this article. High temperature mantle shear zones are active in the upper layer forming a ultraslow spreading oceanic crust (Bickert et al., 2020).
The end-states of each model case have important structural and compositional homologies. One is asymmetry, though the degree of asymmetry is highly variable. The width of one rifted margin or flank (the distance between the start of seafloor spreading and the relatively unattenuated continental crust) is often different from its conjugate. In addition, asymmetry also exists in the structure of the rifted margin, often with regards to the presence and scale of lithospheric boudins (mantle, crust, or both) and in whether the crust or sub-continental mantle extends furthest towards the rift axis.
Another structural feature present throughout is the “shelf” of sub-continental mantle protruding below the thinned continental crust. It is visible where the lithosphere-scale mantle shear zones and detachment faults create a platform of mantle on each flank of the rift underlying the continental crust. These shelves form as the mantle is upwelled to 15-10 km depth and sub- horizontal anastomosing faults begin to accommodate hyperextension and exhumation. The shelf shallows towards the rift axis, but often forms domes or boudins of sub-continental mantle. It can also have highly variable dimensions in different rift flanks and different model cases, ranging from ~5 km to >50 km across. The mantle detachment faults that form these boudins, domes, and mantle core-complexes generally dip landward in contrast to the earlier, continental detachments that dip seaward.
Melt crystallization (also referred to as underplating, magmatic accretion, or refertilization) in our models occurs above the melting region before being migrated laterally as rift evolves. Melt crystallization is also enhanced in high temperature mantle shear zones that form the soles of anti-listric detachment faults. Sometimes this leads to underplating being concentrated on one rift flank rather than another. This magmatic accretion forms the proto- oceanic and oceanic lithosphere.
By varying mantle potential temperature (MPT), surface heat flux (SHF), and extension rate it is possible to predict specific features in each rifted margin case. Structural trends are difficult to quantify, such as degree of asymmetry or the scale of extensional duplexes/ anastomosing faults. However, such structural trends are important predictions for future geological and geophysical observations. A summary of these observable properties in each model can be found in Supplementary Table 1.
The scale of anastomosing detachment faults (quantified by measuring the maximum and minimum widths of individual extensional duplexes within the mantle detachment fault system in each model) appears inversely correlated with surface heat flux. The less heat flowing through the lithosphere, the stronger it is and the larger the individual anastomosed, sigmoidal blocks within the detachment fault systems (Fig. S3). The relationship between SHF and extensional duplexes is also clear in the correlation between SHF and the number of individual anastomosed blocks, which decreases with increased SHF (Fig. S9). However, the frequency of anastomosed blocks has a slight, positive relationship with MPT (Fig. S8).
One of the most important and most obvious trends in the data is the correlation between mechanical oceanic lithosphere thickness and full spreading rate. Of the two extension rates tested using GeoFLAC, the 1 cm/yr cases consistently produce thicker oceanic lithosphere than their 2 cm/yr counterparts (Fig. S4). The thickness of the oceanic lithosphere doesn’t have a clear correlation to either MPT or SHF (Figs. S5 & S6), though this may just be a limit of our models, which don’t include the effects of melt extraction and volcanism on building oceanic crust. Relationships between MPT, SHF, and oceanic lithosphere thickness should be consistent with the observations of oceanic lithosphere at fast- and slow- spreading ridges (Langmuir & Forsyth, 2007; Macdonald, 2001; Searle, 2013), and thus melt extraction is an important caveat regarding this work. Mantle exhumation, quantified here by the number of mantle core complexes (domes of mantle material) that develop, is observed to decrease as SHF increases (Fig. S9). However, colder MPT cases exhibit larger core complexes that are not associated with the initiation of seafloor spreading (Fig. 5) Given that melt extraction has not been implemented, this result reflects a bias towards physical processes associated with amagmatic extension in our models.
The degree of partial melting and of melt crystallization is also observed to be a function of MPT and of extension. In the 1300ºC MPT cases, the percentage of partial melt ranges from 3% to 20%. For cases with a MPT of 1400ºC the percentage of partial melt ranges from 27% to 36%, a smaller range but with greater percentage melt (Fig. S5). A similar relationship holds for extension rate, where range of melt fractions is 3-35% for 1 cm/yr full spreading rates and 19-36% for 2 cm/yr full spreading rates (Fig. S4). The crystallization of this melt in the form of underplating or magmatic accretion also increases with MPT and SHF, as seen from the general increase in crystallization from the top to bottom and from left to right of Fig. 4. The relatively high percentage of partial melt in some of these cases may be a result of our lacking of melt extraction; fractionation, melt conduits, and heterogeneities in the mantle among other necessary simplifications in our numerical models may change the parameters that determine melt percentage. However, the trends and structures observed in each model appear to reflect observed features of real rifted margins, especially the Ivorian margin.
5 Discussion
The experiments in GeoFLAC give new insight into the rifting processes. Our numerical modeling results provide a framework for the kinematic and magmatic development of rifted margins. Extension of the mantle lithosphere during the necking phase is accommodated by large, lithosphere-scale shear zones in the mantle that dip under the continental lithosphere and away from the rift axis. As per Ruh et al. (2022), these shear zones are regions where grain-size reduction enhances diffusion creep and weakens the lithosphere to the point where it can rupture. In previous work, models that don’t incorporate dynamic grain recrystallization lack some key features shown in this article. Without dynamic grain recrystallization, there is no change in fault orientation during the emplacement of oceanic lithosphere, no doming of the mantle in core-complexes, and no lithospheric-scale mantle shear zones (e.g., Lavier et al., 2019). Since the Ivorian, Alpine Tethys, Uralide, and other margins (Clerc et al., 2018) show evidence of out-of-sequence detachment faults, mantle core complexes, and anastomosing shear zones it can be surmised that dynamic recrystallization plays a major role in continental rifting. Serpentinization, while present in some models, does not play an important role and is limited to the shallow crust at low temperature, where H2O is available. Decompression melting in the mantle under the rift axis increases the buoyancy of the mantle and enhances upwelling of the asthenosphere along the lithosphere-scale mantle shear zones. This is a mechanism by which the mantle lithosphere is attenuated and asthenospheric material is brought towards the surface as part of lithospheric breakup.
The origin of the out-of-sequence (continentward-dipping), anastomosing detachment faults observed in the Ivorian continent-ocean transition, and in ophiolites like Nurali and Lanzo is also shown by our numerical modeling results. As the lithospheric shear zones shallow, they couple with crustal detachment faults and initiate the exhumation phase of rifting. The shear zones then transition into the anastomosing faults described previously (the distal or out-of-sequence detachment fault system). There is a weak inverse correlation between surface heat flux and the width of anastomosing faults (the size of each extensional duplex). A possible explanation for this is that the feedback loop between grain damage and dislocation creep which mechanically weakens the lithosphere is inhibited by a colder, stronger lithosphere (de Bresser et al., 2001; Tullis & Yund, 1985). The orientation of these out-of-sequence detachment faults is important. Under the stress regime imposed by buoyant, melt-rich mantle, they dip away from the rift axis and towards the continent. This interpretation is supported by the unusual strike of the out-of- sequence faults at the Ivorian margin. There, the out-of-sequence faults strike NNE while the rift axis trends NNW. If the anomalous orientation of these faults were a solely function of the seafloor spreading stress environment, they should be parallel to the rift axis. If they were a function of the dextral stress imposed by the neighboring Romanche and St. Paul transform faults, then their orientation should represent en echelon faults and strike WNW, nearly perpendicular to the actual orientation. We hypothesize that this orientation is a result of the upward push of irregular magma bodies beneath the rift. This idea is consistent with the nearby presence of a mantle core-complex that could have concentrated decompression melting underneath it (before migrating eastward relative to the rift axis due to extension) as seen especially in models 1, 6, 9, and 10.
Thus, transition from continental rifting to seafloor spreading is strongly influenced by how tectonism and magmatism interact. In the Ivory Coast margin, the earliest evidence of melt is the volcaniclastic-to-pillow basalt that comprises Unit 6. Near the continent, this unit starts out thin, only a few tens of meters thick. However, as the rift develops and the rift axis migrates seaward relative to the continent, Unit 6 thickens to ~1 km and is characterized by more chaotic reflectors (pillow lavas). This increase in volcanism correlates to the shallowing and growth of the melting region of the asthenosphere as modeled by GeoFLAC. Dunite channels (Kelemen et al., 1995; Liang et al., 2010) and faults, such as documented in the Lanzo Massif (Müntener & Piccardo, 2003), provide conduits for this melt to erupt sub-aerially. The shallowing melt region also creates sheared gabbroic bodies that appear in the footwall of the anastomosing detachment faults as Unit 8 (gabbroic bodies also appear as isolated features in the sub-continental mantle of Unit 5). This is consistent with magmatic accretion seen in our GeoFLAC results. Magma that migrates past the solidus as it advects crystallizes above the melt triangle, accreting new crust or refertilizes ancient mantle lithosphere. Because the out-of-sequence detachment faults facilitate mantle upwelling, accretion is concentrated in footwalls of the detachment faults. The out-of-sequence system emplaces these gabbroic magma bodies, which source from (via faults or dunite channels as conduits) (Liang et al., 2010) sheeted dikes and pillow basalts of Unit 6. Thus, the network of detachment faults and the attenuation of the lithosphere above the decompression melting region control the onset of seafloor spreading and determine the structure of the continent-ocean transition.
New oceanic lithosphere in our GeoFLAC formulation comes in two flavors. The mechanical oceanic lithosphere forms above the melt triangle from upwelled asthenosphere and crystallized magma. In cases with a faster extension rate, the mechanical oceanic lithosphere is thinner as the high temperature asthenosphere is upwelling faster than it dissipates heat. Melt crystallization is focused under anti-listric shear zones and detachment faults because these are regions where the melt-rich asthenosphere passed through the solidus with greater frequency than elsewhere. The thermal oceanic lithosphere is defined by the 1300ºC isotherm, which sits beneath the melt triangle due to the latent heat of fusion depressing the geotherm at those depths. Because melt extraction is not part of our GeoFLAC formulation, oceanic crust is not included in our simulated oceanic lithosphere.
6 Conclusions
Synthesizing results from the Alpine Tethys margin and from the Ivorian margin with the GeoFLAC modeling provides a framework to understand rifted margin evolution at magma-poor margins. All three lines of evidence point towards homologous processes that control rifted margin evolution: high-temperature, anastomosing detachment faults that facilitate hyperextension; shear zones in the mantle lithosphere that facilitate the upwelling of asthenosphere driven by buoyancy; and increasing melt production as continents migrate away from the rift axis leading to magmatic accretion, seafloor spreading, and a change in the local stress environment (Fig. 8).