Distribution and timing of lithospheric breakup across the Gulf of Mexico: The role of seaward-dipping reflectors, spreading propagators, and crustal shear zones Open Access

It has long been accepted that the central deep-water part of the Gulf of Mexico¹ is underlain by oceanic crust, although there is disagreement on the amount and distribution of such crust. For example, essentially all interpretations agree on oceanic crust separating the Louann and Campeche salt basins (Figs. 1A–1C; e.g., Humphris, 1979; Pindell, 1985; Sawyer et al., 1991; Marton and Buffler, 1994; Pindell and Kennan, 2009; Kneller and Johnson, 2011; Hudec et al., 2013; Christeson et al., 2014; Steier and Mann, 2019; Pindell et al., 2021a; Filina et al., 2022), and this oceanic realm is here referred to as the “narrow” oceanic basin interpretation. Some interpretations have proposed a considerably wider zone of accreted oceanic crust, limited by the large-amplitude, positive Houston magnetic anomaly and Florida magnetic anomaly on the U.S. margin and the Campeche magnetic anomaly and North Yucatan magnetic anomaly on the Mexican margin (Imbert and Philippe, 2005; Kneller and Johnson, 2011; Karner and Johnson, 2015; Lundin and Doré, 2017). These magnetic anomalies form the limits of the “wide” oceanic basin and restore when the wide oceanic basin interpretation is reconstructed.

The wide interpretation of the Gulf of Mexico is characterized by a wedge-shaped oceanic basin, bounded on the west against Mexico by the East Mexico transform (e.g., Pindell, 1985; Pindell et al., 2021a). The narrow eastern tip of the Gulf of Mexico's wedge-shaped oceanic basin lies beneath the Florida Straits. From a plate-tectonic perspective, opening of the Gulf of Mexico by seafloor spreading was initiated during breakup of Pangea in earliest Jurassic time (ca. 200–190 Ma) when the Tethys Ocean penetrated into the interior of Pangea in the vicinity of Iberia and Morocco. South America was at this time part of Gondwana and remained so until the opening of the South Atlantic Ocean in Early Cretaceous time. Yucatan constituted a large semi-independent microplate that was coupled to both Laurentia (North America and Florida) and Gondwana (South America). The uneven coupling between Yucatan and its northern and southern neighbors eventually led to significant counterclockwise rotation in a ball-bearing fashion. Schouten and Klitgord (1994) suggested that Yucatan's rotation was driven by South America pivoting Yucatan like a pinion in a rack-and-pinion gear system. The resulting dextral motion between Yucatan and North America along the East Mexico transform was appreciated early (e.g., Pindell and Dewey, 1982). With time, paleomagnetic constraints revealed a total counterclockwise (CCW) rotation of Yucatan of ~78° ± 11° (Molina-Garza et al., 1992). The kinematics of the younger spreading phase related to the narrow oceanic basin are usually constrained by restoring the edges of the Louann and Campeche salt basin polygons (e.g., Pindell, 1985; Marton and Buffler, 1994; Kneller and Johnson, 2011), by documenting the evolution of the growing space between North and South America as constrained by Central Atlantic kinematics (Kneller and Johnson, 2011), and/or by delineating Gulf of Mexico spreading axis segments and fracture zones now abandoned and recognized in satellite gravity (Sandwell et al., 2014) and magnetic data (Pindell et al., 2016; Minguez et al., 2020). Correlatable seafloor spreading magnetic anomalies have proven elusive in the Gulf of Mexico, although the abandoned spreading axis is marked by positive magnetic anomalies (Harry and Eskamani, 2013; Filina et al., 2022) that coincide with the axis revealed by the gravity data presented in Sandwell et al. (2014).

Our plate reconstructions of free-air gravity and reduced-to-pole (RTP) magnetic data provide important kinematic and geometric constraints for our proposed ~67° total CCW rotation of Yucatan. The driving force for Yucatan's rotation remains unproven, but we will argue that this motion was an inevitable consequence of accommodation of Tethyan (Central Atlantic) spreading against an impenetrable Panthalassa subduction zone, along with the cog-like coupling of Yucatan with North and South America at different stages of its evolution.

Regardless of differences in the interpreted width of oceanic crust in the Gulf of Mexico, many modern plate models for this region suggest two kinematic phases of continental divergence: (1) subaerial seafloor spreading or continental extension followed by (2) submarine seafloor spreading (e.g., Pindell, 1985; Salvador, 1991; Marton and Buffler, 1994; Imbert and Philippe, 2005; Pindell and Kennan, 2009; Kneller and Johnson, 2011; Pindell et al., 2021a). Similarly, we recognize two phases of opening, and we argue that phase 1 relates to subaerial spreading and magmatic accretion dating back to ca. 200–190 Ma, while phase 2 marks submarine seafloor spreading between ca. 164 Ma and ca. 140 Ma. Onset of phase 2 is temporally constrained by the youngest age of the evaporites, i.e., Callovian (166.1–163.5 Ma; Pindell et al., 2021b). We selected an age of 164 Ma as the youngest age of the evaporites because an absolute age was required in the plate reconstruction. The termination of spreading has been dated to late Berriasian time (here chosen to be 140 Ma) by Ocean Drilling Program Leg 77, which penetrated an unconformity in the Florida Strait that Marton and Buffler (1994) interpreted as a rift-drift hiatus.

The kinematics and timing of phase 1 are more loosely defined and hinge on constraints that rely on regional observations external to the Gulf of Mexico and on global plate models. Previous authors have made analogous comparisons between the Gulf of Mexico magnetic anomalies (Houston, Florida, Campeche, and North Yucatan magnetic anomalies) and the East Coast magnetic anomaly along the U.S. Central Atlantic margin and the West African Coast magnetic anomaly, a positive magnetic anomaly that can be traced along the West African margin and is conjugate to the East Coast magnetic anomaly. The East Coast magnetic anomaly and West African Coast magnetic anomaly are generally accepted to relate to conjugate sets of seaward-dipping reflectors associated with magma-rich opening of the Central Atlantic Ocean (e.g., Holbrook et al., 1994; Talwani et al., 1995; Roeser, 1982), but notably the seaward-dipping reflectors have not been drilled and age dated, nor can the seaward-dipping reflector flows be traced onshore to penetrated basalts (Heffner, 2008; Heffner et al., 2012). Magnetic isochrons do not exist in early Central Atlantic seafloor units and thus cannot be used to interpret the onset of seafloor spreading or emplacement age based on the magnetic time scale. Basinward-dipping reflectors in the Gulf of Mexico have similarly been interpreted as seaward-dipping reflectors with associated positive magnetic anomalies (e.g., Imbert and Philippe, 2005; Kneller and Johnson, 2011; Karner and Johnson, 2015; Lundin and Doré, 2017; Filina and Hartford, 2021), but they have also been interpreted as dipping synrift Triassic sediments without a direct relationship to breakup (e.g., Hudec et al., 2013; Pindell et al., 2014; Hudec and Norton, 2019; Steier and Mann, 2019; Kenning and Mann, 2021; Hasan and Mann, 2021; Rowan, 2022). The onset of phase 1 in the Gulf of Mexico is thus not well constrained by any age dates, per se, but our and other plate models suggest an approximately synchronous breakup of the Gulf of Mexico and the Central Atlantic at ca. 200–190 Ma (Pindell and Kennan, 2009; Kneller and Johnson, 2011).

Our plate reconstruction implies that the majority of the CCW rotation of Yucatan took place between 169 Ma and 140 Ma (~52°), and that the kinematic change at 169 Ma did not coincide with the separation of the Louann and Campeche salt bodies, per se, but started some 5 m.y. earlier, analogous to the model by Pindell et al. (2021a). The wide oceanic basin will be shown to relate to an environment of deposition during which accretion occurred: initially subaerial and subsequently possibly lacustrine prior to and during salt basin development, and fully marine following the separation of the salt bodies at ca. 164 Ma. Reconstruction of the wide oceanic basin requires an additional 15° CCW rotation of Yucatan with respect to North America.

In this paper, we present new evidence from long-offset, deep-penetrating onshore and offshore seismic reflection profiles that consistently support previous interpretations of seaward-dipping reflectors around the Gulf of Mexico and identify a number of seaward-dipping reflectors in new locations. We discuss the relationship between seaward-dipping reflectors and positive magnetic anomalies, and the plate-tectonic requirements of the crust between the seaward-dipping reflectors and the boundary of phase 2 spreading, all of which we believe strengthen the case for an early phase of subaerial oceanic accretion followed by continued seafloor spreading in a second phase characterized by a submarine environment. Our plate model predicts successive abandonment of rift/spreading propagators along the western side of Florida, leading to the final spreading configuration in the Gulf of Mexico. We propose a concept whereby rift tips to spreading axes (propagators) repeatedly attempted to breach the Florida continental crust but successively failed and were abandoned. Ultimately, the hinge zone along the somewhat older Central Atlantic margin provided a strength barrier that prevented direct linkage between spreading centers in the Gulf of Mexico and the Central Atlantic Ocean. Similarly, the final spreading axis in the Florida Straits in the Gulf of Mexico did not propagate beyond the previously formed hinge zone along the northern edge of the proto–Caribbean Ocean (see Plate Reconstruction section). A somewhat similar behavior seems also to have occurred during early breakup of the Central Atlantic and its southern propagation during the formation of proto-Caribbean oceanic crust, which spans from eastern Venezuela to northwestern Colombia; the Central Atlantic “tip” failed to breach the Panthalassa subduction zone, since that would have required the axis to jump to another plate.

We also explored a number of long-term, first-order geologic issues associated with the development of the Gulf of Mexico. We address the conspicuous absence of pre-breakup (Late Triassic) extension landward of the Houston magnetic anomaly. The Houston magnetic anomaly aligns along-strike with the Suwannee shear zone (e.g., Boote et al., 2018; Knapp and Hermann, 2019), which is a major dextral accommodation zone established during the late Paleozoic consolidation of Pangea. In the context of two phases of crustal accretion, we examine and address the tectonic implications of the undeformed Jurassic pre-salt sedimentary successions that are observed to passively drape phase 1 accreted crust. Finally, we show that, in all likelihood, both the Jurassic pre-salt successions and the Gulf of Mexico salt basin were deposited in subaerial and/or sublacustrine environments and, therefore, that the overlying prolific Jurassic hydrocarbon system of this basin was actually developed primarily upon accreted magmatic crust.

This study utilized global satellite gravity data (Sandwell et al., 2014), RTP magnetic data from the U.S. Geological Survey (Bankey et al., 2002) improved in the Gulf of Mexico (Minguez et al., 2020), General Bathymetric Chart of the Oceans (GEBCO) bathymetric data (https://www.gebco.net/data_and_products/gridded_bathymetry_data/), and TGS and Fairfield Geotechnologies seismic reflection data. We discuss 11 onshore and offshore seismic profiles that image the basement across the Gulf of Mexico. The locations of geologic boundaries and key features were used as input to the open-source GPlates plate reconstruction software of Müller et al. (2018), with which both gravity and magnetic grids were reconstructed.

We placed significant emphasis on the location, seismic geometry, and magnetic anomalies of features interpreted as seaward-dipping reflector crust. Therefore, we need to clarify our view of the tectonic significance of seaward-dipping reflectors. We will demonstrate that these reflectors represent subaerial oceanic crust formed by subaerial seafloor spreading, consisting of an upper layer of tholeiitic basalts and a lower layer of gabbros and cumulates that laterally merge structurally and petrologically into layers 2 and 3 of Penrose oceanic crust (e.g., Paton et al., 2017; Karner et al., 2021).

Following Karner et al. (2021, and references therein), we define magmatic crust as any crust formed entirely by magmatic accretion and/or eruption by the decompressive partial melting of asthenosphere at an accreting plate boundary (a spreading center). Normal oceanic or “Penrose” crust comprises a layered tholeiitic volcanic layer of sheeted dikes capped by pillow basalts (layers 2b and 2a), underlain by lower-crustal gabbros and cumulates of layer 3. Layer 2a, typically 1.0–2.5 km thick, is characterized by relatively low P-wave velocities and steep velocity gradients (increasing with depth from ~2.5 km/s to 6.2 km/s; density 2.5–2.8 g/cm³), and it consists of series of pillow basalts, breccias, and flows, interbedded with limestones or cherts. Layer 2b is a 1.0–2.0-km-thick, seismically transparent sheeted dike complex. In turn, lower oceanic crust, or layer 3, is 3.4–6.3 km thick and characterized by relatively high P-wave velocities and low gradients (6.8–7.6 km/s; density 2.92–2.97 g/cm³) and associated with a distinctive “crisscross” pattern of reflectivity, possibly related to frozen crystal mush zones and plastic deformation associated with migrating axial magma chambers (Fig. 2). Commonly, a well-imaged seismic Moho separates the oceanic crust from ultramafic peridotites of the oceanic mantle (Fig. 2). A factor of key importance is our recognition that seaward-dipping reflectors mark subaerial breakup of the continental lithosphere, after which plate divergence is focused at a narrow zone near a spreading center, and therefore no further extension of the continental lithosphere takes place.

Oceanic tholeiites represent ~15%–20% melting of the upper mantle (McKenzie and Bickle, 1988). Melt volume (which translates to oceanic crustal thickness) is a function of spreading rate and upper-mantle potential temperature. Ultraslow spreading rates result in crustal thicknesses of 0.5–4.5 km; spreading rates >2.0 cm/yr tend to produce a uniform crustal thickness (6–8 km, 2–2.5 s two-way traveltime [TWTT]), termed Penrose oceanic crust. In regions characterized by elevated upper-mantle temperatures, such as might be expected for regions affected by plumes, decompressive melt thicknesses in excess of 25 km can be produced. In these cases, the spreading center will be above sea level and produce subaerial volcanic flows, ultimately resulting in seaward-dipping reflector geometries, such as on Iceland.

Seaward-dipping reflectors need not be a function of plume activity (e.g., Jacuípe-Sergipe margin) or thick magmatic crust (e.g., Sauter et al., 2023; Karner et al., 2021); they represent breakup and accretion of magmatic crust in a dominantly subaerial environment. As outlined by McDermott et al. (2019), Paton et al. (2017), and Karner et al. (2021), there are two fundamentally different magmatic systems that characterize volcanic passive margins: (1) synrift continental flood basalt flows, deformed volcanic flows, and intercalated sediments filling synrift accommodation, fed by crustal dike swarms that transect the thinned continental crust (termed type 1 seaward-dipping reflectors), and (2) post-breakup, essentially undeformed, composite seaward-dipping reflector flows (termed type II seaward-dipping reflectors) that progressively become younger toward a volcanic source/spreading center that itself is migrating away from both conjugate margins, the initial flows of which downlap earlier synrift volcanic packages and/or extended continental crust (Fig. 2).

Seismically imaged seaward-dipping reflectors represent a geometric arrangement of stacked volcanic flows with a consistent and repetitive concave-downward motif (e.g., Mutter et al., 1982; Planke and Eldholm, 1994). Individual seaward-dipping reflector packages thin in a landward direction, away from their source at the spreading center. The regularity and lateral consistency of this motif are prima-facie evidence for a constant supply of magma from the locus of extrusion, which is a spreading center that recedes from the margin as new crust is accreted. Reflector geometries (i.e., timelines) represent the chronostratigraphy of superimposed, composite volcanic flows that progressively young in the direction of spreading. Each flow downlaps onto earlier seaward-dipping reflectors (or extended continental crust) and then is flexurally rotated by subsequent volcanic loads so that the depositional dip is reversed, ultimately producing an apparent onlap geometry (e.g., Fig. 2; Palmason, 1980; Buck, 2017). Drilled seaward-dipping reflector crust consists of subaerial tholeiitic basalts with a relatively simple bulk chemistry similar to oceanic crust.

The subaerial setting of seaward-dipping reflectors can be uniquely recognized by the existence of lava deltas within the seaward-dipping reflector succession. Whenever subaerially extruded lavas intersect a body of water (lake or ocean), a lava delta is produced. If the lava flows are sourced from a spreading center where flanks are flooded, then the feather edges of the active seaward-dipping reflector flows terminate in lava deltas. Lava delta geometries are formed as lavas flow into water bodies, and as they come in contact with water, they undergo rapid quenching and fragmentation to form hyaloclastite breccias that are rapidly deposited downslope to produce inclined foresets and escarpments. The lava delta topsets provide direct information on paleoshoreline positions through time (i.e., depositional base level), while the height of the lava delta foresets is a direct measure of paleo–water depth (e.g., Abdelmalak et al., 2016).

Seaward-dipping reflector geometries are generally independent of fault-controlled accommodation, although superimposed late-stage faulting is common (e.g., Fig. 2). This deformation can occur relatively soon after eruption on the flanks of the ridge itself or much later due to ridge migrations or other tectonic events. Type II post-breakup seaward-dipping reflector packages tend to be undeformed and are regionally distributed above (i.e., stratigraphically younger than) type I synkinematic-related volcanic packages. This is a key observation that supports the interpretation that on many margins, seaward-dipping reflectors are in fact post-breakup features and the earliest manifestation of seafloor spreading. With time, the stacked volcanic seaward-dipping reflector flows that comprise the upper crust merge into pillow basalts and sheeted dikes of oceanic crust layer 2. Similarly, seaward-dipping reflector lower crust laterally merges into oceanic crust layer 3 and exhibits the same crisscross pattern that characterizes Penrose oceanic crust (Fig. 2). The change from seaward-dipping reflector upper crust to oceanic layer 2 documents the flooding of the basin and marks a transition from subaerial to subaqueous (submarine or sublacustrine) conditions at the paleo–spreading center.

The morphology of Penrose crust varies considerably between fast spreading ridges and slow to ultraslow ridges and is mainly a function of spreading rate and related magma budget. At fast ridges, exemplified by the East Pacific Rise (e.g., Karson et al., 2023), magmatic accretion keeps pace with plate separation and results in a smooth crust that generally is unfaulted. At slow to ultraslow ridges, the magma budget is insufficient to keep pace with the plate separation. Such accreted crust is characterized by a strongly block-faulted morphology and can be exemplified by the Central Atlantic Ocean (e.g., Smith et al., 2008), the Norwegian-Greenland Sea (e.g., Bruvoll et al., 2009), and the Southwest Indian Ridge (e.g., Sauter and Cannat, 2010). The thickness of the Penrose crust is on average ~6–8 km (e.g., White et al., 1992), but lower thicknesses form during ultraslow spreading (Dick et al., 2003). According to Dick et al. (2003), the spreading rate threshold between ultraslow and slow accretion is ~20 mm/yr full spreading rate. Crust generated at rates below 20 mm/yr is thinner than 6–8 km. Oceanic crust formed during slow and ultraslow spreading also often contains areas marked by mantle exhumation, so-called oceanic core complexes (e.g., Smith et al., 2008; Escartín et al., 2017; Cannat et al., 2019). In simple terms, the block-faulted morphology of slow to ultraslow ridges, a consequence of “rolling hinge” detachment faulting, reflects a time of insufficient magma supply with respect to the plate separation rate during spreading. In this paper, such accreted crust is referred to as tectonized oceanic crust. The tectonized oceanic crust morphology is typical for both phase 1 subaerial and phase 2 submarine accreted crust in the Gulf of Mexico outboard of the seaward-dipping reflector succession (described in later sections). We have been able to identify a number of oceanic core complexes on the eastern Gulf of Mexico seismic lines. It is important to emphasize that there are important structural differences between continental and oceanic core complexes (e.g., Whitney et al., 2013), especially since the oceanic crust thinned by detachment faulting becomes underplated by subsequent magmatic additions; that is, as the crust is being thinned by faulting, its thickness is quickly restored by magmatic underplating (e.g., Maffione et al., 2013). In summary, the recognizable characteristics of seaward-dipping reflectors are: (1) the convex-downward stratal morphology of the upper oceanic crust (layer 2), (2) distinctive lower crust crisscross pattern of reflectivity (layer 3), and (3) lava delta geometries.

Interpretations of seismic reflection data from TGS and Fairfield Geotechnologies and the Gulf of Mexico Basin Opening (GUMBO) refraction lines, coupled with a review of published academic Gulf of Mexico research, have been instrumental in mapping stratigraphic patterns and both the oceanic and continental crustal structure of the Gulf of Mexico. While the seismic data are clearly extremely important, there are nevertheless issues with the clear imaging of Moho topographies and, at times, even basement morphologies. This is particularly acute across the Florida shelf break and for stretched continental crust. In these situations, we augmented our crustal interpretations by using isostatic principles to predict the distribution of the Moho (termed the isostatic Moho).

Isostasy, which fundamentally represents a vertical stress balance of deformed crustal and lithospheric systems, requires a reciprocal relationship between positive and negative density contrasts as a function of depth (e.g., Watts, 2001; Karner et al., 2005). In particular, for extensional systems, the basement geometry inversely relates to the Moho topography: the deeper the basin, the thicker the sedimentary section, the shallower the Moho. When all extension-related heat has dissipated, the negative density contrast of the sediments and unfilled accommodation (air or water) needs to be balanced by the positive density variations represented by the Moho topography. However, to calculate the absolute crustal thickness, an initial crustal thickness, t₀, needs to be assumed. This relationship can be quantified by:

where t_c is the calculated crustal thickness, b is the bathymetry, S is the sediment thickness, ρ_w and ρ_m are the water and mantle densities, respectively, and ρ_s and ρ_c are the bulk sediment and crustal densities, respectively.

To demonstrate the practicality of isostasy for mapping crustal structure, we applied Equation 1 to map the isostatic Moho across seismic profile H, where a reflection Moho is well imaged in places, and compared the result with the refraction Moho determined from GUMBO line 4 (see profile descriptions, profile H). There is excellent agreement with the predicted, observed, and refraction Moho for Penrose oceanic crust and for continental crust characterized by minor extension. Consistent with GUMBO 4, we assumed that the initial (i.e., undeformed) crustal thickness was 36 km (e.g., Christeson et al., 2014). We selected nine sites to showcase the technique and then mapped a continuous Moho across the profile. For oceanic crust, bulk sediment and crustal densities were assigned values of 2.36–2.39 g/cm³ and 2.84–2.85 g/cm³, respectively. Our calculated Moho correlates well with an observed reflection. The bulk sediment density likely reflects the density of young, Mississippi River–sourced sediment. For the eastern shelfal regions, there is an excellent agreement with the predicted isostatic Moho and the refraction Moho using bulk sediment and crustal densities of 2.42–2.48 g/cm³ and 2.70–2.71 g/cm³, respectively. We suggest that the increase in bulk sediment density across the Florida shelf reflects the increased carbonate content consistent with increased sediment refraction velocities, while the bulk crustal density relates to the continental crust of Florida.

A more complicated situation presents itself for mapping the Moho across the shelf break, a region that traditionally is a problem for both seismic reflection and refraction imaging because of the scattering of sound energy. Seismic reflection profile H shows the existence of a deep basin at the base of the shelf break, which has been mapped by both us and Christeson et al. (2014). Isostatically, we would expect the crust to be relatively thin beneath the basin. The isostatic calculation predicts a compensatory shallow Moho for reasonable bulk sediment and crustal densities of 2.44–2.47 g/cm³ and 2.80–2.82 g/cm³, respectively. In contrast, the refraction Moho deepens over this zone, predicting a zone of crustal thickening under the deep basin. Further, the isostatic Moho correlates with reflection seismic reflectors. We conclude that the isostatic Moho reasonably predicts the location of the Moho under the shelf break. In fact, the predicted isostatic Moho across profile H appears to be both accurate and robust, underscoring its effectiveness for determining Moho geometries when either the seismic reflection or refraction data prove to be problematic.

Early work on sequential magnetic profiles over the U.S. East Coast and the Gulf of Mexico showed a strong similarity between the East Coast magnetic anomaly and anomalies bordering the northern Gulf of Mexico, in terms of positive response, wavelength, and amplitude (Hall, 1990). All of the anomalies were modeled in terms of induced magnetism from mafic to ultramafic susceptible basement bodies. Later, it was shown that the East Coast magnetic anomaly equates to a seaward-dipping reflector assemblage (Holbrook et al., 1994; Talwani et al., 1995), with the attendant possibility that Gulf of Mexico anomalies such as the Houston magnetic anomaly have a similar origin (e.g., Mickus et al., 2009).

Several authors have over the years interpreted the basinward-dipping events off the Middle Grounds Arch and northeast and northwest Yucatan as either seaward-dipping reflectors (e.g., Imbert and Philippe, 2005; Eddy et al., 2014; Karner and Johnson, 2015) or Triassic strata intermixed with volcanic flows (e.g., Pindell et al., 2011; Steier and Mann, 2019; Rowan, 2022). We prefer a seaward-dipping reflector interpretation due to the commonly associated positive magnetic anomalies, convex-up shape, lack of growth faults bounding the successions, generally unfaulted and laterally extensive reflector trains, undeformed overlying pre-salt succession, and the general symmetry with the conjugate margin. The identification of potential lava deltas associated with the reflectors adds weight to this interpretation (see Campeche Magnetic Anomaly and Summary of the Evidence from the Seismic Profiles sections). The association between continental breakup, seaward-dipping reflectors, and a magnetic anomaly is common. Authors advocating Triassic strata and volcanics tend to also favor magma-poor margin development with mantle exhumation along large landward-dipping detachments. We are not aware of any process that would yield positive magnetic anomalies in such a setting. The undeformed pre-salt succession is easy to explain if resting on accreted crust, but it is far more difficult to explain if situated over rifted continental crust, since overburden above extended basement blocks tends to display differential subsidence, often with passive reactivation of bounding faults. The undeformed nature of the pre-salt succession is strong evidence against a rifted substrate.

Magnetic anomalies associated with seaward-dipping reflector successions are generally positive and likely reflect induced magnetism (Davis et al., 2018). A possible exception, suggesting that remanent magnetism can generate apparent polarity reversals in seaward-dipping reflector successions, has been reported off NE Greenland (Franke et al., 2019). However, these NE Greenland magnetic reversals were interpreted by Franke et al. (2019) to reflect intrusions into continental crust and, as such, are synrift dikes and sills, which are temporally distinct from post-breakup seaward-dipping reflectors (e.g., McDermott et al., 2019). The seaward-dipping reflector geometries in the Gulf of Mexico are generally associated with positive magnetic anomalies. The correlation is not perfect, however, and conjugate landward-dipping reflections (interpreted in the Apalachicola Basin) are not everywhere associated with positive anomalies. Because seaward-dipping reflectors are a consequence of seafloor spreading, they necessarily will have a conjugate magmatic counterpart dipping in the opposite direction on the other side of the spreading center (landward-dipping reflections). It is a matter of observation (e.g., NW Africa, Namibia, western India) that, in general, the landward-dipping reflection magnetic anomalies associated with ridge jumps are not as well developed as their conjugate seaward-dipping reflector counterpart (e.g., Blaich et al., 2011, 2013). The Gulf of Mexico is no exception. The reason for this asymmetry is not clear, but in the case of the Gulf of Mexico, it may relate to interference with magnetic sources within Paleozoic volcanic suites, such as those drilled by the Raptor well (Mallis et al., 2024) in the Florida Middle Grounds Arch. Notably, other conjugate seaward-dipping reflector successions may show marked differences in both seismic and magnetic expressions. For instance, while the East Coast magnetic anomaly is a well-expressed positive magnetic anomaly, the conjugate West African Coast magnetic anomaly is far more subdued. Similarly, the South Atlantic margins reveal considerable asymmetry between major magnetic anomalies and observed seaward-dipping reflectors (Blaich et al., 2011). Further, the NE Atlantic margins likewise show variable magnetic responses to observed seaward-dipping reflector successions (Berndt et al., 2001; Hopper et al., 2014); conceivably, these differences may relate to variations in the volume of lower-crustal intrusions (Brandl et al., 2023).

We interpret the Houston, Florida, Campeche, and North Yucatan magnetic anomalies along with the smaller Sarasota Arch magnetic anomaly to reflect induced magnetism from seaward-dipping reflectors, similar to the generally accepted relationship between the East Coast magnetic anomaly and the West African Coast magnetic anomaly and seaward-dipping reflectors along the Central Atlantic margin (Davis et al., 2018; Roeser, 1982). Several authors have previously correlated the large positive-amplitude magnetic anomalies in the Gulf of Mexico with seaward-dipping reflectors (e.g., Imbert and Philippe, 2005; Eddy et al., 2014). Mickus et al. (2009) modeled the Houston magnetic anomaly as a magma-rich margin, analogous to the seaward-dipping reflector-bounded Vøring margin off Norway and to the Central Atlantic margin along the U.S. eastern seaboard (e.g., García-Reyes and Dyment, 2022). Further, the North Yucatan magnetic anomaly has been genetically linked to seaward-dipping reflectors by Filina and Hartford (2021), and similarly seaward-dipping reflectors have been reported on the Yucatan and Campeche margins by Pindell and Heyn (2022, their supplementary data), albeit in this case interpreted to overlie extended continental basement. The prominent linear Houston magnetic anomaly has been considered to be the conjugate anomaly to the Campeche magnetic anomaly off Yucatan (e.g., Imbert and Philippe, 2005), but in contrast to the Campeche magnetic anomaly, the Houston magnetic anomaly has not yet been linked to seismically observed dipping events due to imaging challenges related to the thick sedimentary succession, and the generally more difficult seismic acquisition onshore. In this paper, we provide interpretations of possible seaward-dipping reflectors on three profiles onshore Texas and Louisiana, noting that the data are ambiguous, and an association between seaward-dipping reflectors and the Houston magnetic anomaly is not straightforward.

Seaward-dipping reflectors mark a zone of subaerial crustal accretion during the early stages following Gulf of Mexico breakup and reflect the evolution of the accretion process along the margin in both dip and strike directions. At any point along the margin, the seaward-dipping reflectors are not necessarily linked to a specific time but, instead, form a marker for the initiation of subaerial seafloor spreading at that location. It is expected that in cases of diachronous breakup at propagating rifts, the ages of the associated seaward-dipping reflectors would track the local timing of this process. We will return to this topic later in the paper. The Houston–Campeche–Florida–North Yucatan magnetic anomalies are conjugate anomalies (Fig. 1C) that are restored in plate reconstructions of the wide ocean (e.g., Kneller and Johnson, 2011; Karner and Johnson, 2015; Lundin and Doré, 2017), but notably the anomalies are also restored by authors disputing the accreted nature of the crust situated between the narrow ocean and the Campeche magnetic anomaly–Houston magnetic anomaly and the Florida magnetic anomaly–North Yucatan magnetic anomaly (e.g., Pindell et al., 2021a). We will further consider this issue later in the paper. In the following sections, we provide seismic examples of candidate seaward-dipping reflector crustal geometries associated with the Houston, Florida, Campeche, North Yucatan, and Sarasota Arch magnetic anomalies in addition to profiles across the Apalachicola Basin. The profile locations are shown in Figures 1A–1C.

Profile A (Fig. 3) is a N-S onshore seismic profile in eastern Texas that does not intersect the Houston magnetic anomaly proper. The profile terminates ~25 km north of the peak of the Houston magnetic anomaly. Basinward-dipping crustal events beneath the Louann Salt that fan slightly in a basinward direction are interpreted as the updip feather edge of seaward-dipping reflectors. The profile does not image Moho but clearly shows the unfaulted nature of the margin. Little or no deformation is observed in the subsalt stratigraphy, with observed stratigraphic reflectivity subparallel to a smooth reflector at the base of salt.

Profile B (LithoSpan, Fig. 4) is a N-S profile in eastern Texas assembled from marketing documents with kind permission of the data owner, Fairfield Geotechnologies. The data provide a clear image of conformable, undeformed pre-salt strata up to 2 km thick. The pre-salt section seems to thicken basinward, although the base of the section is poorly imaged.

Consistent seaward-dipping energy is observed in the marginal-quality data below the conformable pre-salt succession, which may represent seaward-dipping reflectors. Given limitations of the data, some of the dipping reflectors, particularly in the north, could represent the enigmatically unfaulted Triassic continental clastic succession remarked on by other workers (e.g., Snedden and Galloway, 2019). Given this uncertainty, we have indicated two possibilities for the limit of the continent. Option A places the continent-ocean boundary at the Houston magnetic anomaly, and it would thus be consistent with seaward-dipping reflectors generating this anomaly. Option B extends the seaward-dipping reflector train, and hence the ocean-continent transition, farther north. The Moho is not clearly imaged, and its depth has been calculated using isostatic considerations. The estimated Moho depth seems to follow some energy in the data, which would not be obvious picks without the isostatic calculations.

Profile C (Fig. 5) is a N-S onshore seismic profile in western Louisiana. The profile reaches the Houston magnetic anomaly, but the seismic record length is not sufficient to image seaward-dipping reflectors at the Houston magnetic anomaly. However, landward of the Houston magnetic anomaly, the seismic data reveal basinward-dipping reflectors that terminate upward at the base salt horizon. These dipping events are here interpreted to represent the landward feather edge of seaward-dipping reflectors that onlap the continental margin (cf. Eldholm et al., 1989; Karner et al., 2021). The profile does not image the Moho but characterizes the postkinematic nature of the top basement and subsalt stratigraphy.

Profile D (Fig. 6) is a N-S offshore seismic profile across the Apalachicola Basin, and it shows well-expressed south-dipping seaward-dipping reflectors in the north of the profile, with possible opposing landward-dipping reflectors across the Apalachicola Basin to the south. The profile coincides with the GUMBO 3 seismic refraction profile (Eddy et al., 2014). The seaward-dipping and landward-dipping reflections ought to represent a conjugate pair with an extinct spreading center separating them along the axis of the Apalachicola Basin. The seaward-dipping reflectors are associated with a positive magnetic anomaly, but this does not everywhere appear to be the case with the landward-dipping reflections. This could argue against their magmatic nature; alternatively, as discussed earlier, it may reflect the differences in magnetic response seen on other opposing magmatic margins. The basin is well expressed on depth-to-top basement maps (e.g., Ewing and Galloway, 2019), which are mimicked by the depth to the Middle Jurassic unconformity (Pindell et al., 2015) and coincides with a positive gravity anomaly (Fig. 1B). The gravity anomaly most likely reflects both an elevated Moho and a relatively dense accreted crust beneath the basin. The basin shows a smooth depression at the base salt level in which gravity-gliding deformation of the post-salt succession has taken place. Several profiles parallel to profile C also exhibit seaward-dipping and landward-dipping reflection geometries on opposite sides of the Apalachicola Basin. In its central part, the profile crosses continental crust in the Middle Grounds Arch, an extension of western Florida's continental crust, and the Florida magnetic anomaly, marking the transition to phase 1 tectonized oceanic (accreted) crust. A step-up to normal oceanic (Penrose) crust (cf. Pindell, 2002) is shown to be roughly coincident with the southern limit of the Louann Salt beneath a large diapir.

Profile E (Fig. 7) is a NE-SW offshore seismic profile that orthogonally crosses the Florida magnetic anomaly. The profile also images the architecture of the Middle Grounds Arch. The profile shows well-imaged seaward-dipping reflector geometries, with an imaged crustal thickness of at least 8 km. The seaward-dipping reflectors terminate at the base salt reflector. The peak of the positive Florida magnetic anomaly is located ~40–50 km landward of the bulk of the seaward-dipping reflector package. Oceanward, the seaward-dipping reflectors are overlain by an ~1.5-km-thick, mildly deformed pre-salt succession that further seaward is deposited on deformed accreted crust. The Florida Platform contains sediments as old as Paleozoic age (e.g., Pindell and Heyn, 2022), which are approximately flat-lying and undeformed by either Alleghanian movements or pre-breakup extension. A well-developed pre-salt succession is observed, as is a step-up to phase 2 (Penrose) crust close to the salt limit.

Profile F (Fig. 8) is a N-S seismic profile that traverses the Middle Grounds Arch and the Apalachicola Basin, showing well-developed seaward-dipping reflectors and less well-resolved candidate landward-dipping reflections. Farther south, it traverses the Middle Grounds Arch and the Florida magnetic anomaly, showing well-expressed seaward-dipping reflectors outboard of the Florida shelf edge, the total thickness of which is at least 8 km, and the transition to tectonized phase 1 accreted crust. A well-developed pre-salt succession is observed, as is a well-resolved step-up to phase 2 (Penrose) crust close to the salt limit.

Profile G (Fig. 9) is a N-S seismic profile that again traverses the Apalachicola Basin, showing consistent seaward-dipping reflectors and less obvious landward-dipping reflections. The Moho is generally not imaged under the continent. Seaward-dipping reflectors are again observed where the profile crosses the Florida magnetic anomaly, in this case with seismic geometries that can be interpreted as lava deltas (see later discussion on this topic in the North Yucatan Magnetic Anomaly section). Imbert and Philippe (2005), Pindell et al. (2011), and Eddy et al. (2014) also reported the existence of seaward-dipping reflectors in their analyses. The step-up between phase 1 and phase 2 accreted crust is again clearly resolved, but, in this case, it is bordered by a small fault-bounded basin with counterregional dip, which may be magmatic in origin. Exhumed footwalls observed in the phase 2 crust in the southern portion of the profile may represent oceanic core complexes (e.g., Escartín et al., 2017).

A published seismic profile spanning the Gulf of Mexico from outboard of the Middle Grounds Arch to Yucatan also reveals the architecture of phase 2 Penrose crust (profile 9 in Lin et al., 2019). That profile starts and ends close to the phase 1 to phase 2 boundaries on either side of the abandoned spreading axis, thus imaging mainly phase 2 Penrose crust. It is clear that this phase of accretion postdated the salt, and hence, there is no pre-salt succession over phase 2 crust. The top of the oceanic crust is well imaged, as is a strong Moho reflection, suggesting a crustal thickness of 7.5–8 km in the central part, increasing in thickness toward the Florida Platform to some 9–10 km. The abandoned spreading axis can be recognized in the seismic data, and its position is supported by gravity and magnetic data.

Profile H (Fig. 10) is a NE-SW profile running from the Tampa Embayment southwestward into Penrose crust. Dipping reflectivity beneath the salt on the carbonate shelf is interpreted as potential seaward-dipping reflectors. Tectonized accreted crust immediately west of the shelf edge accommodates a salt basin with deformation of the overlying sediments, stepping up into Penrose crust, approximately at the salt limit. Tectonic disturbances in the Penrose crust may reflect general slow extension and propagation of the spreading axis toward the Florida Strait.

Crust beneath the salt basin displays layering that may include a pre-salt basin, stacked lavas and volcaniclastics in the upper oceanic crust, or both. Imaging beneath the salt makes it impossible to confidently distinguish between these possibilities. The layering consistently dips to the southwest. Except in the deepest part of the salt basin, these dips do not appear to be fault controlled. Also, imaging is good enough to see clinoforms that prograde to the northeast. These features imply that the original depositional dip has been reversed, as is the case with seaward-dipping reflectors. The layering beneath areas southwest of the salt basin therefore may relate to a time-transgressive series of volcanic flows and volcaniclastic sediments emanating from the spreading ridge as it migrated to the southwest. At several locations immediately outboard of the salt basin, layered upper oceanic crust displays geometries that can be interpreted as lava deltas. This is potential evidence of shallow-water conditions near the spreading ridge at a time immediately following deposition of salt. The presence of clinoforms indicates that the ridge may have been subaerial at this early stage of seafloor spreading, but this will need confirmation with higher-resolution seismic data.

We chose this line to calibrate our isostatic calculations of Moho depth for two reasons: It displays a wide zone of well-imaged Moho reflections beneath the oceanic crust, which allows calibration on the oceanic side, and it is parallel to refraction line GUMBO 4 (Christeson et al., 2014), which lies ~36 km to the northwest, and therefore allows calibration in the less well-imaged areas under the continent. We plotted key surfaces derived from GUMBO 4 along with our interpretation for direct comparison: green for Moho, blue for top crust, and black for the 7 km/s velocity contour. Our isostatically calculated Moho depth matches the calibrating observations quite well: reflection Moho beneath the ocean, and refraction Moho beneath the continent. In the area where the crustal thickness is changing, our isostatic results for Moho depth do not match the GUMBO 4 Moho, with a maximum divergence of over 10 km (GUMBO 4 is deeper) beneath the major salt basin just seaward of the shelf break. We favor the shallower pick for the Moho in this area for several reasons. A shallow Moho is consistent with crustal thinning beneath a prominent basin. In the absence of significant flexure or major density variations, a deeper Moho is isostatically inconsistent with a deeper top crust. Although not as prominent as areas under the ocean crust to the southwest, faint reflectors in the area under the salt basin line up with the shallow Moho pick. Also, in the same location, the GUMBO 4 line shows that the 7 km/s crustal velocity contour diverges from their Moho pick by about the same amount relative to other areas, implying a wide area of thicker, faster velocities beneath the location of the shallow Moho pick.

Profile I (Fig. 11) is an E-W seismic profile across the Sarasota Arch magnetic anomaly that images a 50-km-wide swath of seaward-dipping reflectors and magmatic crust, but only the easternmost portion of the seaward-dipping reflector wedge beneath the carbonate shelf shows well-imaged stratigraphic layering. The original width of the seaward-dipping reflector crust may have been much greater, but it was truncated by seaward-dipping faults beneath the salt basin. The magnetic anomaly is mainly located over the platform, consistent with the location of the observed seaward-dipping reflectors. Immediately outboard of the shelf edge, the crustal architecture is overlain by debris flows and slumps associated with collapse of the shelf, obscuring imaging of seismic reflector geometries at crustal levels. The accreted crust further outboard is tectonized and may accommodate a pre-salt basin, although imaging below the salt is insufficient to distinguish pre-salt basin sediments from seaward-dipping reflectors or stacked lavas in the upper oceanic crust. Farther to the west, the profile displays an unusually thick layered upper crust, probably representing a seaward-dipping succession of tholeiitic lavas. Similar to profile H (Fig. 10), layered upper oceanic crust outboard of the salt basin displays geometries that can be interpreted as lava deltas (see later discussion on this topic in the North Yucatan Magnetic Anomaly section), constituting evidence of shallow-water deposition in post-salt time. Tectonic complexity in the western part of the profile may reflect oblique shear zones and other deformation related to adjustments in the spreading ridge location and orientation during tectonic rotation of Yucatan. A strong Moho reflection is observed, showing a crustal thickness decreasing basinward from ~16 km to ~5 km beneath the salt and then increasing slightly to a steady thickness of 7 km.

Profile J (Fig. 12) is a NW-SE seismic profile, somewhat oblique to the North Yucatan magnetic anomaly. An ~30–40-km-wide, 4-km-thick succession of seaward-dipping reflectors coincides with the positive North Yucatan magnetic anomaly. Published profiles in the same general area reveal dipping events that, like here, are interpreted as seaward-dipping reflectors (Filina and Hartford, 2021), although some authors have interpreted them as dipping Triassic sediments (Rowan, 2020, 2022; Steier and Mann, 2019; Kenning and Mann, 2021). The seaward-dipping reflector lavas terminate upward at the base salt horizon and are onlapped oceanward by an undeformed pre-salt succession that expands basinward to >3 km in thickness. The basinward edges of the pre-salt section reflectors interact, and are interbedded, with apparent landward-prograding lava deltas associated with the landward terminations of the seaward-dipping reflector packages. This observation, combined with the thickness of the pre-salt sedimentary section, indicates that production of the seaward-dipping reflector crust took a substantial amount of time to form. The Yucatan carbonate platform is faulted, but this deformation is comparatively young and postdates breakup. Slumping and a décollement above the salt on the shelf probably relate to the 66 Ma Chicxulub impact to the south.

A basement step-up is commonly observed where the seismic profiles cross the accretionary phase 1 to phase 2 boundary. Where fracture zones impinge on the seaward-dipping reflector crust, other structural complexities are observed, including basement step-downs (Rowan, 2022).

Profile K (Fig. 13) is located near the northern end of the Campeche magnetic anomaly, and it shows an ~100-km-wide, 3–5-km-thick seaward-dipping reflector succession with evidence of lava deltas (see the following section), and a transparent lower crust 3 km thick. The seaward-dipping reflectors terminate upward at the base pre-salt horizon and are onlapped oceanward and interfinger with an undeformed pre-salt stratigraphic package that expands basinward to ~4 km thickness. Landward, the profile is marked by the Yucatan carbonate shelf edge, which shows no significant deformation. The Moho is not imaged clearly in most of this profile. Line K shows some of the most clearly imaged examples of seaward-dipping reflectors in our study area. For this reason, we have provided a zoom-in of the seaward-dipping reflector train to illustrate its internal characteristics (Fig. 14). The detailed section shows good examples of what we interpret as lava deltas, implying that the landward terminations of constituent lava flows impinged on a shallow body of water; the water depth is given by the height of the lava delta foresets. If one accepts this interpretation of lava deltas, an important implication is that the basinward-dipping succession cannot be fault-bounded Triassic sediments deposited in half graben; the deltas are facing the wrong way. The presence of lava deltas also has significant implications for the Early Jurassic paleogeographic setting, which may constitute an interesting basis for future discussions. For further details, see Figure 14.

In summary, the described seismic examples reveal seaward-dipping reflector successions that can be associated with the linear positive magnetic anomalies (Houston, Florida, Campeche, and North Yucatan magnetic anomalies) and with the more local Sarasota Arch magnetic anomaly, analogous to the relationship between the East Coast magnetic anomaly and well-accepted seismically imaged seaward-dipping reflectors (e.g., Holbrook et al., 1994; Talwani et al., 1995). The Houston-Campeche magnetic anomaly and the Florida–North Yucatan magnetic anomaly are conjugate when reconstructed. However, our spreading model implies that the seaward-dipping reflectors formed diachronously during repeated failed attempts of a lengthening Gulf of Mexico spreading to link with the Central Atlantic across Florida. Thus, the Houston-Campeche magnetic anomaly seaward-dipping reflectors developed before the Florida–North Yucatan magnetic anomaly seaward-dipping reflectors, and the seaward-dipping reflectors associated with the Sarasota Arch magnetic anomaly are younger still, representing migration of the spreading axis tip toward its eventual position in the Florida Straits. The Sarasota Arch magnetic anomaly has a conjugate positive magnetic anomaly on the west side of the Florida Straits, but seaward-dipping reflectors have not been interpreted there, possibly due to overprinting by faulting.

Seaward-dipping reflector successions terminate upward against the base salt horizon or beneath undeformed pre-salt sediments. Older seaward-dipping reflector packages are onlapped oceanward and overlain by an undeformed pre-salt succession. Basinward, the seaward-dipping reflectors transition into more tectonized accreted crust that also is overlain by a pre-salt succession that generally thickens basinward. Deformation of the pre-salt succession is generally restricted to the outermost part near the phase 2 boundary. At top basement level, a structurally complex and variable step-up generally marks the transition to phase 2 Penrose crust, above which neither the pre-salt succession nor the salt was deposited. The pre-salt succession is here proposed to have been deposited in a landlocked subaerial basin (below global sea level), consistent with the observation of underlying and contemporaneous seaward-dipping reflectors, and it is interpreted to consist of fluvial and lacustrine continental sediments. The mostly undeformed nature of the pre-salt succession is expected given that the underlying accreted crust is tectonically quiescent once formed.

The boundary between phase 1 subaerially accreted crust and phase 2 Penrose crust lies close to the basinward limit of salt. The allochthonous Sigsbee salt tongue constitutes an exception. Off NE Yucatan, salt has locally been displaced up to ~30 km beyond the phase 1-2 boundary. Besides the salt limit, the phase 1-2 boundary is well expressed in large parts of the Gulf of Mexico by a sharp break from negative (landward) to positive (seaward) gravity anomalies on both Bouguer and residual Bouguer maps (e.g., Lin et al., 2019), by strong gravity gradients (Sandwell et al., 2014), and by positive magnetic anomalies (Fig 1C), which commonly coincide with a basement or structural step-up from thinner to relatively thicker crust.

The evolution of the Gulf of Mexico appears to have been closely related in timing and geometry to the development of the Central Atlantic Ocean (e.g., Huerta and Harry, 2012), being a consequence of plate kinematic linkages between the early Central Atlantic Ocean and the Gulf of Mexico during the separation of Laurentia and Gondwana (e.g., Kneller and Johnson, 2011). Both oceanic domains broadly adhere to the Wilson cycle concept (Wilson, 1966), whereby lithospheric weaknesses in an orogenic weld between two continental plates are later exploited during continental breakup. The hinterland tectonic roots of the Appalachian mountain chain, where the final closure between Laurasia and Gondwana occurred in late Carboniferous–early Permian times as part of the Alleghanian orogeny (e.g., Stampfli and Borel, 2002; Domeier and Torsvik, 2014), formed the locus for Central Atlantic opening in earliest Jurassic times (e.g., Kneller and Johnson, 2011; Thomas, 2006). The Ouachita-Marathon fold-and-thrust belt may be viewed to represent the southwesterly extension of Alleghanian closure (e.g., Huerta and Harry, 2012). Although the Ouachita-Marathon and Appalachian orogenies are of similar age and both relate to the consolidation of Pangea, they behaved quite differently during the breakup of Pangea. The Appalachian orogeny involved significant shortening (e.g., Lammie et al., 2020), while the Ouachita-Marathon orogeny has been viewed as a “soft” collision with limited crustal shortening in its central part north of the Gulf of Mexico (Huerta and Harry, 2012). The Appalachian orogen was thus both more deformed by shortening and subsequently more deformed by extension during the breakup of Pangea, when several terrane boundaries were reactivated and guided bounding faults of the Newark Supergroup extensional system (e.g., Withjack et al., 2013; Lundin et al., 2022). The Newark Basin Supergroup system and the South Georgia rift are characterized by half graben structures, which contrast to the unfaulted basement of the Eagle Mills Formation in the greater Sabine block area (Louisiana-Arkansas-Texas; Norton et al., 2018; Snedden and Galloway, 2019).

The proposal in this paper and others (e.g., Imbert and Philippe, 2005; Kneller and Johnson, 2011; Lundin and Doré, 2017) is that initial Gulf of Mexico breakup took place north of the present coast in southern Texas and Louisiana, placing the continent-ocean boundary at, or north of, the Houston magnetic anomaly. This magnetic anomaly is an ENE-WSW linear feature some 200–500 km distance south of the arcuate Ouachita-Marathon orogenic front (Figs. 1A–1C).

Alleghanian continental accretion was characterized by large-scale dextral movements along the orogen (e.g., Berra and Angiolini, 2014; Domeier and Torsvik, 2014; Mueller et al., 2014). This motion is locally evidenced by numerous examples of deformation formed during dextral slip between the onshore Appalachian terranes (e.g., West, 1998; Hatcher, 2002; Engelder and Whitaker, 2006; Ettensohn and Lierman, 2012). The Suwannee shear zone (Mueller et al., 2014; Boote and Knapp, 2016; Knapp and Hermann, 2019) is a particularly prominent dextral terrane boundary, extending ENE-WSW from the southern Appalachians to Louisiana (Figs. 1A–1C). The Suwannee shear zone is a late Alleghanian feature, forming the suture between the largely undeformed Charleston and Suwannee terranes to the south (and Laurentian crust) and accreted Gondwanan terranes to the north within the Appalachians (Mueller et al., 2014). The Suwannee shear zone has been traced over 1200 km, and Knapp and Hermann (2019) proposed major dextral slip, consistent with dextral transport of peri-Gondwana terranes (Charleston, Suwannee, and Yucatan terranes) along the southern margin of Laurentia (Mueller et al., 2014). Most interestingly for this paper, the Suwannee shear zone, oriented slightly oblique to the Appalachian orogen, appears to align with the similarly trending Houston magnetic anomaly (Fig. 1C). This suggests that, in the NW Gulf of Mexico, the Suwannee shear zone may have been responsible for strain localization prior to initial breakup, with the Houston magnetic anomaly reflecting magmatic exploitation of a preexisting deep crustal weakness, a concept also suggested for the Central Atlantic by Lundin et al. (2022). This idea prompts a closer examination of the proposed line of breakup, and the events preceding breakup.

The Ouachita-Marathon belt is bounded to the south by a sag basin, which contains marine Permian sediments and thick Triassic continental sediments of the Eagle Mills Formation (e.g., Norton et al., 2018; Frederick et al., 2020). It is widely regarded as a successor basin to the Ouachita-Marathon orogeny (e.g., Nicholas and Wadell, 1989; Norton et al., 2018; Snedden and Galloway, 2019; Frederick et al., 2020), that is, a basin that formed in or adjacent to the orogenic belt following the cessation of tectonic activity.

The Eagle Mills Formation thickens toward the Gulf Coast but is not recorded south of the Houston magnetic anomaly (Frederick et al., 2020), which we interpret as the line marking the initial accretion of magmatic crust. Eagle Mills unit deposition was long considered to be genetically equivalent to the Late Triassic rift basins farther to the east, such as the South Georgia rift and the Newark Supergroup basins fringing the North American East Coast margin (e.g., Woods and Addington, 1973; Salvador, 1991). However, although similar in terms of being a repository for Triassic red beds, the Eagle Mills Formation is significantly different to these rift basins in many ways, with little or no fault-controlled accommodation, filling in residual topography within the underlying Paleozoic basement, and exhibiting a gentle, disconformable contact with the overlying Louann Salt (Snedden and Galloway, 2019).

From these observations and interpretations, breakup appears to have occurred without obvious preceding large-scale upper-crustal extensional faulting. However, the Eagle Mills successor basin cannot have formed without some controlling lower-crustal or mantle processes. Although our onshore profiles do not have record lengths sufficient to capture Moho reflections (Figs. 3–5), it is believed that crystalline crust south of the Ouachita-Marathon belt is thinner than the continental average, tapering southward to 15–20 km at the coast (e.g., Mickus and Keller, 1992; Harry and Londono, 2004; Huerta and Harry, 2012; Miao et al., 2022; Shen and Ritzwoller, 2016; see Fig. 15). This suggests that thinning, perhaps at the lower-crustal level, was involved in the formation of the successor basin. However, the lack of upper-crustal faulting in the Eagle Mills Formation suggests that breakup may have involved exploitation of deep-rooted Alleghanian linear weaknesses, of which the Suwannee shear zone is an example.

Subsidence in the Ouachita-Marathon belt successor basin continued after continental separation, as shown by the distribution of the Louann Salt, the northern limit of which broadly mimics the Ouachita-Marathon belt orogenic front to the north (Figs. 1A–1C). It also seems logical that Jurassic continental sedimentation should have continued after initial breakup along the Houston magnetic anomaly and Florida magnetic anomaly, including over the subaerially accreted crust prior to salt deposition. Recent stratigraphic work on limited well penetrations from below the salt in the onshore Gulf of Mexico region (see summary in Filina et al., 2022) shows a considerable time gap of ~60 m.y. between deposition of the uppermost (Triassic) Eagle Mills Formation sediments and the Bajocian salt (169 Ma). This gap could be partly explained by the poor age constraints on the Eagle Mills succession (e.g., Snedden and Galloway, 2019; Filina et al., 2022). However, it is also possible that bypassed sediments were deposited directly into a widening non-marine basin and constitute the largely undrilled pre-salt successions of the offshore Gulf of Mexico (e.g., Karner and Johnson, 2015), evidenced by the ~2 s TWTT, undeformed, pre-salt stratigraphy observed generally around the Gulf of Mexico. The paucity of onshore Eagle Mills Formation units may also simply mark a hiatus in deposition across the crown of the Sabine terrane. At this time, it is difficult to test this hypothesis due to a lack of Triassic penetrations or clear pre-salt seismic imaging south of the Sabine terrane.

A composite profile was constructed to help clarify the large-scale distribution, geometry, and spatiotemporal relations between key tectonostratigraphic units (Fig. 15). The northern portion of the profile was assembled from data within a wide corridor and is therefore marked by a white transparent box in Figures 1A–1C. The composite section traverses the seaward-dipping reflector sequences identified at the Houston magnetic anomaly and Campeche magnetic anomaly, a pre-salt succession (the thickness of which is sometimes poorly constrained, particularly onshore), and the step-up between phase 1 and phase 2 (Penrose) accreted crust. RTP magnetic data accompanying this composite section show the well-defined positive anomalies at the Houston magnetic anomaly and Campeche magnetic anomaly.

The northern section follows a path close to the N-S regional section of Jusczuk (2002), from southeastern Kansas to the Gulf of Mexico abyssal plain (labeled L in Fig. 1A). Deep internal crustal geometries of the deformed Laurentian margin and Ouachita orogenic system were derived from multiple sources (Mickus and Keller, 1992; Jusczuk, 2002; Thomas et al., 2021). Moho depth onshore was constrained by USArray receiver function data (Shen and Ritzwoller, 2016; Ma and Lowry, 2017; Miao et al., 2022) and offshore by GUMBO2 refraction and reflection seismic data (Eddy et al., 2014). North of the Ouachita area, the Moho is imaged at depths between 43 km and 46 km and rises gradually to the south toward the Gulf of Mexico (Shen and Ritzwoller, 2016). At the coastline in the vicinity of the composite section near the Texas-Louisiana border, the Moho depth varies between 29 km and 33 km. The Moho continues to get shallower offshore. Offshore, Moho depth is 26 km near the coastline of Louisiana based on refraction data (northern end of GUMBO profile 2; Eddy et al., 2014). The GUMBO data also show the Moho rising to ~20 km depth near the southern end of the composite section. Water depth and the geometries and thicknesses of sediments down to the base salt are based on Snedden and Galloway (2019, chapter 1, section 5).

We have shown that a thick pre-salt sedimentary section is commonly observed along the Gulf of Mexico margins, and similar pre-salt sediments are likely ubiquitous beneath the northern margin. Karner and Johnson (2015) suggested that pre-salt sedimentary rocks in the offshore northern Gulf of Mexico are also likely to have unusually high velocities because of their deep burial, based on similarities to interpretations in the deep section of the Salton Trough, where Fuis and Mooney (1990) observed “basement” with velocities under 6 km/s but lacking a pronounced velocity discontinuity with overlying sediments. They interpreted this layer as greenschist-facies sedimentary rocks. The equivalent section in the northern Gulf of Mexico is not imaged on seismic data, but we illustrate that it could be up to 6 km thick beneath the coast and thin to the north and south.

Coincident with the Houston magnetic anomaly, the section shows seaward-dipping reflectors at the margin up to 20 km thick and conjugate to equivalent seaward-dipping reflectors along the northwest Yucatan margin (Fig. 15). Shear wave splitting data are consistent with margin-parallel dikes at crustal levels along the northern Gulf Coast (Gao et al., 2008). The thick seaward-dipping reflectors transition into deformed oceanic crust that is thinner (5–7 km thick) and underlies the pre-salt sediments south of the shoreline. The oceanic crust changes character and becomes thicker and more tabular at the step-up near the edge of autochthonous salt.

The southern section follows a NW-SE path across the western Yucatan margin. Crustal geometries and the thicknesses and distributions of tectonostratigraphic units were constrained by depth-converted seismic reflection profile K (Fig. 13). The profile was extended to the SE beyond seismic control. In areas under the shelf where the Moho is too deep to image on the seismic data, Moho depth was estimated using isostatic calculations.

The Gulf of Mexico opening phases are here described with reference to the example seismic sections (Figs. 3–13; labeled A–K in Fig. 1), a regional cross section across the basin from the southern U.S. coast to Yucatan (Fig. 15; labeled L in Fig. 1), RTP magnetic data (Fig. 1C), the free-air gravity map (Fig. 1B), and the plate reconstructions (see Plate Reconstruction section; also shown in Fig. 16).

Kinematically, phase 1 of oceanic spreading in the Gulf of Mexico is defined by ~15° CCW rotation of Yucatan with respect to North America. Breakup is interpreted as having been magma-rich and subaerial in nature, based on the evidence of interpreted seaward-dipping reflectors coinciding with major, positive, linear magnetic anomalies (Houston, Florida, Campeche, and North Yucatan magnetic anomalies, respectively; Figs. 1C and 3–13). While initial breakup timing is not tightly constrained, it is feasible that the timing of initial breakup at the western end of the Gulf of Mexico coincided with the rapid injection of tholeiitic Central Atlantic magmatic province magmas and the related magma-rich opening of the eastern U.S. and northwest African seaboards, based on plate kinematic linkages and the general similarity between the Houston magnetic anomaly and the East Coast Atlantic magnetic anomaly (Kneller and Johnson, 2011; Kneller et al., 2012). A relationship is possible between initial breakup and the short-duration but intense Central Atlantic magmatic event, tightly constrained to 201 ± 0.5 Ma (e.g., Blackburn et al., 2013). However, it is not conclusively proven that the Central Atlantic broke up synchronously with the Central Atlantic magmatic province event, since the earliest seaward-dipping reflectors associated with the East Coast magnetic anomaly have been neither drilled nor age dated. The chronology of the oldest part of the Central Atlantic is poorly constrained (e.g., Greene et al., 2017). Stratigraphic constraints have placed the Central Atlantic breakup no later than 190 Ma (Sahabi et al., 2004; Labails et al., 2010), although it is important to note that the deeper stratigraphic sequences of the eastern U.S. seaboard are overlapped by younger units (i.e., they do not crop out anywhere on the margin). It also appears likely that it took considerably longer than the short duration of the Central Atlantic magmatic province event (600 k.y.) to generate the seaward-dipping reflector succession associated with the East Coast magnetic anomaly (Davis et al., 2018), and hence, the magnetic anomaly probably relates to a longer time span than the Central Atlantic magmatic province event. Therefore, we will consider breakup of the Central Atlantic to be ca. 201–190 Ma.

As indicated earlier, the architecture of modeled gravity and magnetic data across the Houston magnetic anomaly suggests a magma-rich margin (e.g., Mickus et al., 2009). Candidate seaward-dipping reflectors are seen landward of the Houston magnetic anomaly (profiles A–C, Figs. 3–5). These basinward-dipping reflectors terminate upward at the base of a Jurassic pre-salt section conformable with the base-salt surface, and we interpret these geometries to represent the feather edge of a larger seaward-dipping reflector package. This undrilled pre-salt succession is observed above the seaward-dipping reflectors, which were deposited during a “time gap” between Triassic sediments and the Louann Salt (e.g., Snedden and Galloway, 2019; Filina et al., 2022) mentioned earlier. Offshore, phase 1 accretionary crust is frequently seen to be overlain by an undeformed and conformable stratigraphic pre-salt succession, which probably formed in a nonmarine depositional setting (Karner and Johnson, 2015).

Seaward-dipping reflectors are a subaerial form of seafloor spreading, essentially the result of a post-breakup magmatic process that builds tholeiitic magmatic crust, where magmatism is not submarine but instead occurs at a spreading ridge that is exposed subaerially (Karner et al., 2021). Therefore, the presence of seaward-dipping reflectors implies that breakup has already occurred at that location along a margin (Karner and Johnson, 2015).

The visually attractive correlation between the East Coast magnetic anomaly and the Houston magnetic anomaly begs an important question: Why was opening geographically discontinuous between the two magnetic anomalies, and why did it not connect through Florida as a single spreading axis? The explanation may relate to global kinematics and the obliquity between the two anomalies. Both breakup trends appear to have followed inherited structural grain; the East Coast magnetic anomaly followed the NE-SW Appalachian chain, and the Houston magnetic anomaly probably exploited the ENE-WSW–trending late Alleghanian Suwannee shear zone (Boote and Knapp, 2016; Boote et al., 2018; Figs. 1A–1C). Alternatively, it is possible that initial breakup in the Central Atlantic, representing separation of Laurentia and Gondwana, formed a tract of seaward-dipping reflectors and oceanic crust that inhibited breakthrough of phase 1 Gulf of Mexico spreading to form a single continuous spreading center; that is, by the time the NE Gulf of Mexico in the vicinity of NW Florida started to open, the Central Atlantic oceanic crust was already in place, providing a rheological strength barrier (cf. Steckler and ten Brink, 1986). This could suggest a slightly later opening along the NE Gulf of Mexico compared with the Central Atlantic.

We propose a model whereby three failed arms of extension in the eastern Gulf of Mexico are recorded by the successively abandoned rift/spreading propagators of the Mississippi Salt Basin, Apalachicola Basin, and Tampa Embayment (Fig. 16). These basins are here interpreted to mark failed attempts to break through the Florida crust before the spreading axis headed southward toward the Florida Straits. Like Imbert and Philippe (2005), we interpret the strong, high-amplitude reflections in the Apalachicola Basin as seaward-dipping reflectors. However, we are aware that the seismic events also have been interpreted as intermixed continental strata and volcaniclastic flows in a fault-bounded half graben (Storey, 2020; Izquierdo-Llavall et al., 2022). The latter interpretation could imply that the basin was undergoing significant extension while subaerial lava flowed into the half graben from a spreading axis to the west. This is a viable variation to a propagator that has achieved breakup and where seaward-dipping reflectors develop. Both alternatives imply that extension diminished eastward toward the tip of the propagator or rift. A somewhat similar concept to ours, of failed breakup attempts, was suggested by Imbert and Philippe (2005). Opposing seaward- and landward-dipping high-amplitude seismic reflection geometries have been reported in the Apalachicola Basin (e.g., Eddy et al., 2014) and are here also interpreted in profiles D, F, and G (Figs. 6, 8, and 9). These opposing dipping geometries are interpreted to mark seaward-dipping and landward-dipping reflectors on either side of a failed spreading arm, now stranded on the Laurentia side after their abandonment.

The phase 1 lines of breakup coincide with major positive magnetic anomalies, such as the Florida magnetic anomaly (profiles D, E, F, and G in Figs. 6, 7, 8, and 9). Phase 1 crust may also extend along the Florida margin as far south as the Sarasota Arch (profile I in Fig. 11), i.e., the margin south of the Tampa Embayment. There, candidate seaward-dipping reflectors again coincide with a positive magnetic anomaly. The failed attempts of the propagators to transect Florida offer a possible explanation for the final curved geometry of the Gulf of Mexico spreading axis. During the final phase of opening, the Gulf of Mexico's spreading axis headed toward its final apex at the proto-Caribbean margin, which by then was a well-established ocean, and thus provided another rheological barrier preventing linkage of the Gulf of Mexico and proto-Caribbean spreading centers (see Plate Reconstructions section).

The consistent development of seaward-dipping reflectors through a succession of rift propagation and abandonment events raises an interesting question. Magma emplacement and initiation of seaward-dipping reflector development at the East Coast magnetic anomaly, and possibly the Houston magnetic anomaly, can perhaps be linked to the Central Atlantic magmatic province magmatism. However, as indicated earlier, the Central Atlantic magmatic province is recognized as an intense, short-duration event and is unlikely to account for the entire seaward-dipping reflector succession along the East Coast magnetic anomaly (Davis et al., 2018). Potential linkage between Central Atlantic magmatic province magmatism and the proposed succession of magma-rich events in the eastern Gulf of Mexico is therefore problematic (in the Apalachicola Basin, Tampa Embayment, and along the Sarasota Arch), since our model suggests that seaward-dipping reflectors developed sequentially over a time span of some 25 m.y. If this was the case, it suggests that the mantle beneath the eastern Gulf of Mexico at this time may have remained anomalously hot or hydrous for a considerable time after Central Atlantic magmatic province events and was thus predisposed to excess magmatism during breakup, perhaps due to thermal damage by the Central Atlantic magmatic province event or proximity to the Bahamas plume. Alternatively, the seaward-dipping reflector development might imply a generic connection between magma-rich margin development and the exploitation of preexisting crustal-scale dislocations, as suggested by Lundin et al. (2022), or it may simply represent the geometry that develops during subaerial spreading, regardless of whether or not the magmatism is excessive.

Between the seaward-dipping reflectors, marking boundaries to the Gulf of Mexico's continental margins and the Penrose crust in the central Gulf of Mexico, there exists a zone of deformed accreted crust. Transition from subaerial to submarine spreading is a typical evolution for magma-rich margins; domains of thinner accreted crust interposing between the seaward-dipping reflectors and Penrose crust are less common, although not unknown; a similar configuration may occur on the margin of the Campos Basin of Brazil, where seismic profiles show a transition from well-developed seaward-dipping reflectors to tectonized oceanic crust to Penrose crust (Karner et al., 2021, their fig. 20). There are two possibilities for the nature of this tectonized oceanic crust: (1) The deformation represents the common occurrence of phases of insufficient magma delivery and oceanic core complex development at the spreading center. Such is the situation in the Central Atlantic, although there, the core complexes formed in a submarine setting (e.g., Smith et al., 2008; Escartín et al., 2017; Cannat et al., 2019). (2) As breakup and spreading center reorganization propagated toward the southeast, earlier oceanic crust became involved in this migration process and resulted in superimposed deformation. The two possibilities are not mutually exclusive.

The deformed crust is characterized by low-frequency, discontinuous, high-amplitude reflections that on some profiles reveal a general basinward dip continuous with the more landward seaward-dipping reflectors (e.g., profile E, Fig. 7). This crust is unlikely to represent exhumed continental mantle because (1) it lies seaward of seaward-dipping reflectors that mark continental termination, and (2) locally on the Yucatan side, a possible Moho reflection is interpreted beneath the deformed crust (cf. Rowan, 2022, figs. 8a and 8b). The deformed crust conceivably represents slow to ultraslow spreading and may contain a mixture of block-faulted accreted crust and mantle peridotites, as observed in similar slow to ultraslow spreading regions worldwide (cf. Dick et al., 2003). Core complex development, high-angle faulting, and rugosity can characterize such oceanic crust when the spatial and temporal magma supply at the spreading center is variable, as is the case for the Central and South Atlantic regions (e.g., Smith et al., 2008; Escartín et al., 2017; Cannat et al., 2019). This tectonism is usually imparted near the spreading center to varying degrees.

The tectonized accreted crust is overlain by a largely undeformed pre-salt succession that thickens basinward, with thicknesses up to 4 km on our seismic profiles (Figs. 3–13), and locally exceeding 8 km thickness in the Rio Grande embayment, onshore south Texas (Karner and Johnson, 2015), implying concomitant subsidence of the underlying thin, tectonized crust. The pre-salt succession is thin or missing altogether onshore, in some cases lapping onto the seaward-dipping reflector succession. The lack of deformation in this unit is interpreted to mean that this section is postkinematic and was deposited in a postrift, post-breakup environment on thermally subsiding accreted crust that has remained tectonically quiescent.

The tectonized accreted crust that is seen between the seaward-dipping reflectors and Penrose crust may have formed in a shallower lacustrine environment in the central part of a landlocked basin below global sea level. The pre-salt succession may be a mix of fluvial and lacustrine sediments. A continued landlocked setting is envisioned until a transient incursion of the sea governed the development of the evaporite basin (Louann and equivalent salt). The evaporite basin lasted some 5 m.y. (Pindell et al., 2021a; Rowan, 2022) before complete final breaching of the barrier led to permanent marine conditions, sub-marine production of Penrose crust, and separation of the once-contiguous salt basin into the Louann and Campeche salt bodies. Opening of the marine gateway and the possible increase in spreading rate may have been correlated and is only schematically portrayed in our simplistic plate model (see Plate Reconstruction section).

The undeformed, presumably nonmarine or lacustrine pre-salt succession overlying phase 1 subaerial crust was succeeded by the Louann (and equivalent) evaporites, representing the first major marine incursion into the basin, and flooding of a restricted basin that previously had been below global sea level. The second phase of Gulf of Mexico opening (phase 2) in our plate model commenced by a kinematic change at ca. 169 Ma. The youngest salt age is not well defined but is of Callovian age. Thus, the kinematic change of phase 2 preceded the split of the salt basin when submarine accretion of classic Penrose oceanic crust occurred. Phase 2 involved a pronounced counterclockwise rotation of Yucatan away from Laurentia (~52° CCW). The kinematic change coincided with or slightly preceded (Pindell et al., 2016) the oldest salt in the Gulf of Mexico, which marks the initial breaching of marine barriers around the Gulf of Mexico. Production of Phase 2 accreted crust occurred once fully marine conditions were established in the Gulf of Mexico. The widening of the Florida Straits gateway was likely responsible for breaching the topographic barrier and led to the marine inundation of the Gulf of Mexico. The evaporites were divided by oceanic crust formed during continued CCW rotation of Yucatan, resulting in the Louann and Campeche evaporite provinces on the conjugate North American and Yucatan margins, respectively (Pindell and Dewey, 1982). The initial emplacement of phase 2 crust is constrained by the youngest age of these evaporites, which classically have been considered Callovian in age (e.g., Salvador, 1987) but have recently been dated to range between Bajocian, ca. 169 ± 1 Ma, and Callovian (166.1–163.5 Ma; Pindell et al., 2021a).

The transition between phase 1 and phase 2 (Penrose) crust is often marked first by gradual thinning of the tectonized accreted crust and then by an elevational step-up onto slightly shallower submarine oceanic crust (e.g., Pindell, 2002). This topographic step is here considered to mark an increased crustal thickness across the boundary with Penrose crust, causing the step by isostatic adjustment between thinner (tectonized) subaerial crust and normal-thickness Penrose crust. Overall, spreading rates were higher in phase 2, which could have produced the increased thickness of the phase 2 ocean crust across the basin. It is noteworthy that the Penrose crust still, in places, shows evidence of faulting (e.g., Figs. 6–10), perhaps related to obliquely intersected transforms and adjustments in the spreading ridge location. Potential oceanic core complexes (e.g., Escartín et al., 2017) are interpreted in the Penrose crust of profile G (Fig. 9).

The topographic step-up has a relief of 1–3 km (Pindell et al., 2014; Lin et al., 2019). The pre-salt sedimentary succession terminates basinward against the step-up. The phase 1-2 crustal boundary is delineated by the Campeche and Louann Salt limits (except the allochthonous Sigsbee salt tongue), by small positive magnetic anomalies, and by an inflection in the gravity anomaly.

Our plate reconstruction for the evolution of the Gulf of Mexico is illustrated by a series of maps spanning the interval 200–140 Ma (Fig. 16), generated using the GPlates software of Müller et al. (2018), reconstructing satellite free-air gravity anomalies (Sandwell et al., 2014) and magnetic anomalies (Bankey et al., 2002) as a function of time. The reconstructions show the general sequence of tectonic events responsible for the development of the Gulf of Mexico. Although there are some general constraints on timing (for example, the beginning and end of phases 1 and 2), ages of intermediate steps are not well constrained (for example, the ages of the failed propagators along the western margin of Florida). Large parts of the western margin of North America (and Mexico) were accreted along the Panthalassa subduction zone after the opening of Gulf of Mexico, although constraining the evolution of these western arc-related terranes is beyond the scope of the paper, so these features have been masked in the reconstructions of Figure 16.

Like many plate models for the opening of Gulf of Mexico (e.g., Pindell, 1985; Salvador, 1991; Marton and Buffler, 1994; Imbert and Philippe, 2005; Pindell and Kennan, 2009; Kneller and Johnson, 2011; Pindell et al., 2021a), our model also involves two kinematic phases. To a first order, our model is kinematically similar to the one by, e.g., Pindell et al. (2021a), although our total rotation of Yucatan is larger. A more fundamental difference between our model and that presented by Pindell et al. (2021a) lies in the interpretation of crustal type between the Houston magnetic anomaly/Campeche magnetic anomaly (our phase 1 boundary) and the onset of submarine seafloor spreading (our phase 2 boundary). We propose that the magnetic anomalies reflect conjugate seaward-dipping reflectors that mark conjugate continental boundaries. Thus, the anomalies (continental boundaries) should reconstruct on top of one another in plate reconstructions of the pre-breakup fit. Other workers, such as Pindell et al. (2021a), interpreted extended continental crust in the mostly sub-salt domain between these magnetic anomalies and the generally accepted boundary of Penrose crust (our phase 2 boundary), implying that the magnetic anomalies are located within extended continental crust. Such an interpretation is problematic because the distance between the Houston magnetic anomaly and Campeche magnetic anomaly at the onset of phase 2 opening was in the order of 500 km (cf. our 164 Ma reconstruction, fig. 13C in Pindell et al., 2021a, and fig. E1B in Pindell and Heyn, 2022, their supplementary data). Reconstruction of an ~500-km-wide zone of thinned continental crust would necessarily result in a wide slab of pre-rift continental crust that would prevent reconstructing the anomalies; however, note that such a reconstruction is still shown in the Pindell et al. (2021a) reconstruction.

In any event, key boundaries in a plate reconstruction are those between continent and accreted crust, here considered to largely be marked by seaward-dipping reflectors and the associated Houston, Florida, Campeche, and North Yucatan magnetic anomalies, and the Sarasota Arch magnetic anomaly. The East Mexico transform, at the distal (western) end of the wedge-shaped Gulf of Mexico, marks a continental transform boundary (e.g., Pindell and Kennan, 2009; Román Ramos et al., 2009). Magnetic isochrons are not observed in the Gulf of Mexico, and therefore the plate reconstructions rely on indirect constraints on timing and kinematics. However, in the eastern Gulf of Mexico, the interpreted phase 1 to phase 2 boundary is highlighted by positive magnetic anomalies (Fig. 1C), by a marked free-air gravity gradient (e.g., Nguyen and Mann, 2016), by seismically observed basement step-ups, and by the limit of subsequent salt deposits. In the U.S. west-central Gulf of Mexico, the thick post-breakup sedimentary apron generally masks the phase 1 boundary, while on the Mexican side (in Sureste Basin), severe halokinesis generally interferes with clear imaging of the crustal architecture and pre-salt sedimentary geometries. Overall, the Louann and Campeche Salt polygons coincide closely with the phase 1 to phase 2 boundary, except for the allochthonous Sigsbee salt tongue, and minor extrusion of salt canopies over phase 2 crust off NE Yucatan (Rowan, 2022). The phase 1 reconstruction of Yucatan and Florida relies on the global reconstruction of Gondwana and Laurentia, while the phase 2 reconstruction is geometrically and kinematically constrained by fracture zones revealed by the vertical gradient of satellite free-air gravity data (Sandwell et al., 2014), and by the coupling between Yucatan and both Laurentia and Gondwana as it rotated (Schouten and Klitgord, 1994; Kneller and Johnson, 2011). Phase 1 is thus least constrained both in time and kinematics, and although the start and end of phase 2 are fairly well constrained in time, changes in the rates of extension remain uncertain. Nevertheless, calculation of average angles of rotation reveal rotation rates three times higher rates during phase 2 opening.

Generally recognized issues that must be addressed in plate reconstructions of Pangea are overlaps between the Blake Plateau–Bahamas Platform and Africa, and between southern Florida and the Demerara/Guinea Plateaus (e.g., Bullard et al., 1965; Klitgord et al., 1984; Pindell and Kennan, 2009; Kneller and Johnson, 2011; Pindell et al., 2021b; Erlich and Pindell, 2021). The Blake Plateau–Bahamas Platform is covered by some 10 km of carbonates, effectively preventing imaging of the underlying crust, but providing evidence that the underlying crust, if continental, must have been substantially stretched and/or thinned. Modeling of gravity and magnetic data does not conclusively constrain crustal geometries of this platform, although Dale (2013) interpreted the platform to be underpinned by volcanic accreted crust. Two general approaches have been most commonly applied to address the overlap issue. One is to rotate all of Florida clockwise (CW) back through time (e.g., Kneller and Johnson, 2011). The other is to translate the southern tip of Florida northward along a loosely defined sinistral boundary called the Bahamas fracture zone (Klitgord et al., 1984; Marton and Buffler, 1994), the Jay fault (Pindell, 1985), or the Florida transfer zone (e.g., Erlich and Pindell, 2021).

The three Jurassic basins along the west side of Florida (Mississippi Salt Basin, Apalachicola Basin, and Tampa Embayment) have been interpreted by Pindell et al. (2021a) as Jurassic rifts, limited on their eastward ends by the Florida transfer zone, along which 300–400 km of sinistral slip is invoked. Although the presence of the sinistral transform has figured in a number of papers since Walper and Rowett (1972), its existence has been questioned on several grounds (e.g., Ball, 1991; Kneller and Johnson, 2011; Beaman et al., 2017). This paper diverges from earlier work and interprets these three basins as failed propagators, with significant implications for the timing of events in the eastern Gulf of Mexico. If the dipping reflectors identified in the Apalachicola Basin and alongside the Sarasota Arch are indeed seaward-dipping reflectors and landward-dipping reflections, then our model suggests that the geometry and arrangement of these failed breakup axes requires that they become successively younger from north to south, representing a gradual stepwise southward shift and lengthening of the main phase 1 spreading axis. Sequential development of the proposed failed propagators, regardless if they contain accreted crust or not, would have governed a CCW bending of Florida during opening of the Central Atlantic and Gulf of Mexico, thereby reducing the overlap against the Demerara-Guinea Plateau. We acknowledge that some degree of distributed strike-slip deformation may have occurred at the terminations of the Mississippi Salt Basin, Apalachicola Basin, and Tampa Embayment against the western Florida margin. However, we see no need for—or direct evidence of—a single major strike-slip fault with 300–400 km of sinistral slip traversing the Florida mainland, as proposed by Pindell et al. (2021a). At the Sarasota Arch, the candidate seaward-dipping reflectors are succeeded seaward in close geographic proximity by (successively) thin tectonized phase 1 crust (overlain by the pre-salt succession and the mid-Jurassic salt) and a step-up to phase 2 Penrose crust (profile G, Fig. 9). This would be expected given the proximity to the apex of the subsequent phase 2 spreading axis and would suggest that the timing of breakup at this location would have been late in phase 1, perhaps during late Early Jurassic or early Middle Jurassic times.

In our plate reconstruction, the overlap problem between southern Florida and the Demerara and Guinea Plateaus was addressed in three ways:

(1) By a small CW bending (rotation of segments) of northern Florida into the South Georgia rift and Mississippi Salt Basin, and closure of the Apalachicola Basin and Tampa Embayment;

(2) By allowing the Demerara/Guinea Plateau seaward-dipping reflectors to build gradually through Early Jurassic time because age dating of dredge samples from the Demerara Plateau has yielded an age of 173.4 ± 1.6 Ma (Basile et al., 2020), thus supporting the idea that the seaward-dipping reflectors along the western sides of the Demerara and Guinea Plateaus developed over some time and are not restricted to Central Atlantic magmatic province age; and

(3) By allowing ~30 km slip along a proposed ENE-trending sinistral shear zone across central Florida, informally called the Osceola shear zone. This lineament coincides with a positive magnetic anomaly across Florida, which in turn coincides with the boundary between the Osceola granitic and arc terranes (Erlich and Pindell, 2021). A Jurassic volcanic zone is located on the south side of the boundary (Erlich and Pindell, 2021), further testifying to the potential significance of the shear zone. Westward, the Osceola shear zone extends into the Tampa Embayment, and eastward, it marks a basement elevation step beneath the Blake Plateau carbonate platform (Dale, 2013). The gap between southern Florida and Yucatan in the Pindell et al. (2021a) full closure configuration is avoided by the proposed movement along the Osceola shear zone.

The early Paleozoic Suwannee Basin overlies the Suwannee terrane, the south-dipping Brunswick suture zone, and the Charleston terrane, and it provides a minimum age for the amalgamation of the Suwannee and Charleston terranes in the late Neoproterozoic to early Cambrian (Boote et al., 2018). The subhorizontal Suwannee Basin sediments are essentially undeformed, implying that the basin and the underlying terranes remained intact during the Alleghanian collision. The ENE-trending Suwannee shear zone, located north of the Suwannee and Charleston terranes (Figs. 1A–1C), has been traced for ~1400 km and has been proposed to represent a major dextral transform boundary between the peri-Gondwanan terranes (Suwannee, Charleston, and Yucatan) and Laurentia (Knapp and Hermann, 2019). In this paper, we suggest that the Suwannee shear zone continues westward along the Houston magnetic anomaly, extending its length to ~2500 km. The western prolongation of the Suwannee shear zone along the Houston magnetic anomaly originally bounded the Yucatan and Sabine terranes. Our postulated Osceola shear zone is parallel to the Suwannee shear zone. The southern boundary to the Suwannee and Yucatan terranes against Gondwana (South America) has also been proposed to be a major shear zone, with a sinistral displacement (Mueller et al., 2014). These three shear zones are subparallel (Figs. 16 [top panel] and 17), and we follow Mueller et al. (2014) in that they formed during the Gondwana-Laurentia collision.

An unnamed linear magnetic anomaly of interest to this study is here informally termed the Appalachian frontal anomaly on Figure 17. This NNE-trending positive magnetic anomaly can be mapped from southern Alabama to southern Kentucky, over a distance of ~750 km. The anomaly is located near the front of the Appalachian deformation, and at the surface, it coincides with the Sequatchie anticline and other Appalachian foreland structures (Steltenpohl et al., 2013). However, the anomaly must be related to a deep-seated crustal feature, the nature of which currently has not been recognized. A similar ~500-km-long linear anomaly in Yucatan aligns parallel to the Appalachian frontal anomaly when Yucatan is restored to its pre-breakup location (Fig. 17). It is tempting to correlate the anomalies and use them as geologic piercing points, but the alignment may be fortuitous. If correlated, the anomaly trend parallelism upon full closure of the Gulf of Mexico supports the 67° CW rotation required to restore Yucatan to its original position and suggests ~200 km of dextral slip along the Suwannee shear zone prior to the opening of Gulf of Mexico.

Directly south of the Suwannee shear zone, there lies the South Georgia rift, which partly overlies the Suwannee Basin (Figs. 1A–1C). Approximately 250 wells have penetrated the sedimentary succession of the South Georgia rift and have encountered primarily Late Triassic red beds (Heffner, 2008; Heffner et al., 2012). Several wells reveal minor thicknesses of Jurassic sediments (Heffner, 2008; Heffner et al., 2012). Two N-S refraction profiles, SUGAR profile 1 and profile 2, across the South Georgia rift reveal moderate crustal thinning (β_max ranging from 1.4 to 1.8) and regional underplating of ~1.5 km interpreted to be Late Triassic in age (Marzen et al., 2020). Erosion of Jurassic sediments cannot be excluded, however, and the age of the underplating is equivocal, although demonstrably postrift. We therefore allow some minor Jurassic extension in the South Georgia rift, increasing westward and connecting with the Mississippi Salt Basin. This permits a limited CW rotation of northern Florida during the reconstruction. The model restores the wedge-shaped Mississippi Salt Basin–South Georgia rift, Apalachicola Basin, and Tampa Embayment by ~35 km, ~50 km, and ~35 km, respectively (at their western ends).

As well as the seaward-dipping reflector geometries and associated magnetic anomalies, the phase 1 observations detailed in this paper have the following common characteristics: (1) the feather edges of seaward-dipping reflectors emplaced across thinned continental crust, with minor or no visible pre-breakup extensional faulting of the upper crust landward of the seaward-dipping reflectors; (2) a seaward transition from seaward-dipping reflectors to thinner tectonized accretionary crust; (3) an undeformed pre-salt succession overlying the latter; and (4) a basement “step-up” from thinner phase 1 subaerially accreted crust to thicker phase 2 Penrose oceanic crust.

Relatively abrupt ocean-continent transitions appear to be a feature of magma-rich margins worldwide (e.g., Doré and Lundin, 2015). In the Gulf of Mexico phase 1 examples described earlier, the break may have been facilitated by weaknesses related to subvertical anastomosing crustal shear zones such as the Suwannee shear zone. The alignment between the Suwannee shear zone and the Houston magnetic anomaly is intriguing, but there is less evidence for this mechanism in the Apalachicola Basin. The Tampa Embayment arguably can be linked to the proposed ENE-trending Osceola shear zone across central Florida, which may have been significant in accommodating phase 1 plate movement, although utilization of this shear zone did not proceed to opening, arguably due to its proximity to the Central Atlantic oceanic lithosphere. Breakup on the margin along the Sarasota Arch and southward may have been related to transforms that developed during the early phases of phase 1 opening when Yucatan moved southeastward, a concept proposed more than 30 yr ago by Pindell (1985) and Salvador (1987), or to oblique propagation and abandonment of successive aborted rift segments.

From a wider plate-tectonic perspective, opening of the Central Atlantic Ocean, the proto–Caribbean Ocean, and the Gulf of Mexico initiated during breakup of Pangea in the earliest Jurassic (ca. 200–190 Ma) and can be viewed as a result of the Tethys Ocean encroaching on Pangea. These oceanic domains all utilized the Alleghanian suture and associated transform weaknesses, broadly consistent with the classic Wilson cycle concept (Wilson, 1966). Pangean breakup was initiated in the Central Atlantic Ocean and its southern prolongation, the proto–Caribbean Ocean (now largely consumed by the Caribbean plate; e.g., Braszus et al., 2021). The tip of the proto–Caribbean Ocean must somehow have linked with the Panthalassa subduction zone, perhaps in some diffuse manner, since a subduction zone–ridge–subduction zone is an unstable triple junction configuration (McKenzie and Morgan, 1973).

Thus, the effective termination of the tip of the proto–Caribbean Ocean by the Panthalassa subduction zone would have required the southern tip of the Central Atlantic/proto-Caribbean spreading system to have been accommodated in some other way. We propose that Gulf of Mexico developed as an opposed wedge-shaped ocean to compensate for the tip of the proto–Caribbean Ocean, and that the East Mexico transform, bounding the Gulf of Mexico opening against Mexico (e.g., Pindell and Kennan, 2009; Román Ramos et al., 2009; Pindell et al., 2021a), formed as a geometric necessity. While the opposing, wedge-shaped Gulf of Mexico and proto–Caribbean Ocean spreading systems were both active (until ca. 140 Ma), the majority of the extension toward the SW end of this spreading system was taken up along the East Mexico transform in the Gulf of Mexico. This suggests that any attempted linkage between the tip of the proto–Caribbean Ocean and the upper plate of the Panthalassa subduction zone was ineffective or nonexistent (Fig. 18).

The apparent influence of preexisting strong oceanic lithosphere on propagating zones of extension is a repeated theme in the Gulf of Mexico. Other examples, whereby a young ocean is impeded by a preexisting ocean, include the Red Sea–Gulf of Suez rift versus the hinge zone to the Tethyan oceanic plate, leading to the development of the Dead Sea transform (Steckler and ten Brink, 1986), and the Labrador Sea–Baffin Bay–Lancaster Sound and Jones Sound rifts versus the Amerasia Basin hinge zone, leading to the development of the broader Wegener fault transform system (Lundin and Doré, 2019).

Strong oceanic lithosphere appears to resist being broken by intersecting propagators; conversely, preexisting deep-seated weaknesses in continental lithosphere probably facilitate breakup without significant pre-breakup extension (Lundin et al., 2022). The Alleghanian Suwannee shear zone and the subparallel shear zone along the northern margin of South America both arguably guided breakup of the Gulf of Mexico and the proto–Caribbean Ocean.

The superregional scenario of Tethyan impingement on Pangea and the resulting opening of the Central Atlantic focused the tip of the proto-Caribbean Ocean spreading center to interact with the Panthalassa subduction zone. The amount of extension along the length of the Central Atlantic is comparatively even, meaning that the proto-Caribbean tip had to be accommodated by an opposed wedge-shaped area of similar extension magnitude: namely, the Gulf of Mexico (Pindell and Dewey, 1982). Similarly, the diachronous migration of failed propagators along the western Florida margin can be seen as a function of the strength barrier formed by the developing Central Atlantic and, finally, by the proto–Caribbean Ocean. The arcuate-shaped Gulf of Mexico phase 2 area of Penrose crust is a highly unusual oceanic configuration and can be viewed as a result of the successive earlier propagator abandonments in the Gulf of Mexico caused by significant and prolonged rotation of the Yucatan microplate.

Our interpretation places much of the prolific hydrocarbon system of the Gulf of Mexico over phase 1 accreted crust. For many years, it has been a general assumption that hydrocarbon systems do not develop on oceanic crust, but these ideas have required significant revision and rethinking after recent discoveries in the South and Equatorial Atlantic. Recent major hydrocarbon discoveries off Namibia, in the Graf and Venus fields (Hedley et al., 2022; Winter et al., 2022), along the Sergipe margin in NE Brazil, in the Outer Basin High of the Santos Basin (Karner et al., 2021), and along the Guyana margin (Trude et al., 2022) demonstrate that petroleum systems do indeed develop on accreted and seaward-dipping reflector crusts if favorable depositional environments and sufficient overburden are present. Presumably, previous objections to such hydrocarbon systems were based mainly on heat-flow arguments, and/or accreted crust was necessarily undeformed and associated with deep-water paleobathymetry. Contemporary accreted crust will have lower heat flow than thinned continental crust. In the Gulf of Mexico, the main Kimmeridgian–Lower Berriasian marine source rocks (e.g., Cunningham et al., 2016) were deposited before the main Gulf of Mexico basin fill in the latest Cretaceous and Cenozoic. These source rocks are overmature beneath the thicker northern sedimentary apron of the Gulf of Mexico but are in the oil window in the distal basin where overburden is thinner (e.g., Ewing and Galloway, 2019). Conversely, not all accreted crust is a suitable substrate for a hydrocarbon system per se, because most such oceanic basins do not share the prolonged and rather unusual, restricted, and isolated early history.

1. The Gulf of Mexico is underlain by oceanic crust that was initially accreted subaerially with seaward-dipping reflector development in a landlocked basin below global sea level (phase 1).

2. We interpret the Houston, Florida, North Yucatan, Campeche, and Sarasota Arch magnetic anomalies to be related to seaward-dipping reflectors, similar to the relationship between the East Coast magnetic anomaly and associated seaward-dipping reflectors along the Central Atlantic margin. The seaward-dipping reflectors mark a zone of subaerial crustal accretion formed during the earliest stage of oceanic spreading that followed breakup between Yucatan and North America.

3. Phase 1 continental breakup and accretion were diachronous and marked by seaward-dipping reflectors in successively younger locations of breakup. The gradual breakup spanned Early and Middle Jurassic time (ca. 200/190–164 Ma). Continental sediments, probably fluvial and lacustrine, were shed into the landlocked basin during phase 1 and were deposited on the subaerially accreted crust.

4. Transient breaching of land bridges allowed temporary marine invasions in the Bajocian (ca. 169 Ma) and initiated the evaporite basin in the Gulf of Mexico, which expanded in width until the Callovian (ca. 164 Ma), when permanent breaching established fully marine conditions, terminated evaporation, and initiated submarine accretion of Penrose crust (phase 2).

5. Post-salt spreading (beginning ca. 164 Ma) continued to widen the Gulf of Mexico and separated the salt basin into two parts, the Louann and Campeche Salt Giants.

6. As well as marking the change from nonmarine to marine conditions, the division between the phase 1 and phase 2 modes of accretion was associated with plate kinematic changes and a probable increase in spreading rate, creating the distinct step-up observed between the two types of accreted crust.

7. At a superregional scale, initial breakup of Pangea in the Central Atlantic Ocean can be viewed as an outcome of impingement of the Tethys Ocean on Pangea. The Gulf of Mexico and the proto–Caribbean Ocean represent the southern distal end of the Central Atlantic Ocean, and these paired wedge-shaped oceans probably formed as a consequence of the evolving ocean approaching the Panthalassa subduction zone.

8. During early breakup of Pangea, Gondwana separated from Laurentia along the Alleghanian Iapetus suture zone parallel to the Appalachians, forming the Central Atlantic Ocean, while the proto–Caribbean Ocean and Gulf of Mexico probably utilized major Alleghanian shear zone weaknesses on either side of the peri-Gondwana terranes (particularly Yucatan).

9. In terms of plate kinematics, Yucatan initially traveled southeastward with South America (Gondwana), but from ca. 169 Ma onward, Yucatan acted as a microcontinent and rotated markedly counterclockwise. This kinematic change governed transient breaching of tectonically maintained land bridges that had isolated the Gulf of Mexico from the global ocean, and following ~5 m.y. of continued rotation, these barriers were permanently breached.

10. Yucatan's rotation and the Gulf of Mexico's arcuate final spreading axis were outcomes of successive failed attempts of the Gulf of Mexico spreading axis to cut across Florida. Three easterly oriented arms of extension (propagators) on the west side of Florida (the Mississippi Salt Basin, Apalachicola Basin, and Tampa Embayment) all failed and collectively reshaped the spreading axis, which finally focused southeastward into the Florida Straits. All propagators failed due to encountering strength barriers related to hinges of oceanic lithosphere on the Atlantic margin of Florida.

11. In the plate reconstruction proposed here, overlap problems between southern Florida and the Demerara and Guinea Plateaus were addressed by minor CW rotation of Florida, moderate strike-slip accommodation along the Osceola shear zone in central Florida, and taking into account the gradual build-out of the oceanic plateaus through the Early Jurassic. Large-scale sinistral movement along a conceptual Florida transfer zone is not required in this model.

12. The major Upper Jurassic hydrocarbon source rocks in Gulf of Mexico (Oxfordian and Tithonian) were deposited soon after fully marine conditions had been established. These organic-rich deposits are essentially basinwide. An implication of the proposed accretionary substrate beneath the Gulf of Mexico is that most of the hydrocarbons discovered in the Gulf of Mexico (~60 billion barrels of oil equivalent) represent a petroleum system over oceanic crust. Such petroleum systems have been considered unusual, but they are increasingly being recognized elsewhere, e.g., on the deep-water Namibian margin.

We thank TGS and Fairfield Geotechnologies for their kind permission to publish selected seismic profiles, and we thank Irina Filina for sharing the reduced-to-pole magnetic grid. Grateful thanks go to Jim Pindell and Ian Norton for very effective and useful critiques at the review stage. E.R. Lundin and G.D. Karner respectively thank Equinor and ExxonMobil management for permission to publish. The four of us would each like to thank our spouses, Barbro, Pilar, Karen, and Barbara, for their support and understanding during the long gestation of this paper.