Mantle earthquakes in the absence of subduction? Continental delamination in the Romanian Carpathians Open Access

The association of earthquakes in the upper mantle with the subduction of oceanic plates at convergent boundaries has been a fundamental tenet of plate tectonics for decades. Indeed, the recognition that Wadati-Benioff zones of localized seismicity represent the descent of oceanic lithosphere into the mantle was a significant contribution to the plate-tectonic paradigm. To date, the scientific community has yet to agree on a case where a well-defined zone of upper-mantle seismicity is unrelated to the subduction process, and yet the geologic history of the Romanian Carpathians appears to preclude such an origin for the Vrancea seismogenic zone (Knapp et al., 2005; Houseman and Gemmer, 2007; Lorinczi and Houseman, 2009).The Vrancea seismogenic zone of Romania is one of the most active seismic regions in Europe (Fig. 1), with two to three large (M_W > 7.0) events per century occurring at depths of 70–200 km, and it is widely interpreted to be the result of remnant oceanic subduction (Wortel and Spakman, 2000). Here, we present deep seismic-reflection profiles from the Eastern Carpathians that document the lithospheric architecture associated with the Vrancea seismogenic zone. Our data provide strong evidence that the Vrancea seismogenic zone seismicity is unlikely to be related to subduction, and that the geometry and position of the seismogenic volume match closely that of a large void in the lower crust beneath the Eastern Carpathians. We interpret these relationships to mean that the seismogenic body resulted from a process of delamination of the lower continental crust beneath the bend zone of the Eastern Carpathians in the absence of subduction.

As originally proposed by Bird (1979), continental delamination was envisioned as an inevitable consequence of lithospheric thickening, but largely as an aseismic process limited to the mantle lithosphere. Subsequently, Kay and Mahlburg Kay (1993) suggested that lithospheric delamination could also include density inversions in the crust, which would trigger the removal of both the lower crust and upper mantle. Our data suggest that the combination of continental delamination and introduction of lower-crustal material to mantle depths (1) is the source of mantle seismicity in the Vrancea seismogenic zone, (2) represents an important alternative process to explain mantle seismicity in the absence of subduction, and (3) suggests that lithospheric delamination may potentially be seismogenic when lower crust is involved.

Historically, geodynamic models proposed for the Vrancea seismicity have been subduction related (Fuchs et al., 1979; Girbacea and Frisch, 1998; Linzer et al., 1998; Wortel and Spakman, 2000; Sperner et al., 2001; Cloetingh et al., 2004; Schmid et al., 2008). Subsequently, much of the research conducted during the past 20 yr has utilized the various subduction models in interpreting observations of basin evolution (Tarapoanca et al., 2003; Leever et al., 2006; Matenco et al., 2007), volcanism (Seghedi et al., 2004, 2005; Downes et al., 1995; Falus et al., 2008), uplift (Sanders et al., 1999; Fugenschuh and Schmid, 2005; Krezsek and Bally, 2006), geophysical investigations of the crust (Hauser et al., 2007; Martin et al., 2005; Diehl et al., 2005), and mantle (Fan and Wallace, 1998; Weidle et al., 2005; Ismail-Zadeh et al., 2005, 2008; Ivan et al., 2008), which ultimately have influenced the prevailing ideas about the regional tectonic evolution of the Carpathian-Pannonian system (Wortel and Spakman, 2000; Ustaszewski et al., 2008; Schmid et al., 2008). However, new models for Vrancea seismicity have also been forthcoming. Knapp et al. (2005) suggested continental lithospheric delamination as a probable mechanism for the generation of seismicity beneath the East Carpathians, while Houseman and Gemmer (2007) suggested that the seismic activity resulted from gravitational instability caused by lateral thinning of the mantle lithosphere beneath the Pannonian basin, resulting in thickening and downwelling beneath the Eastern Carpathians. Lorinczi and Houseman (2009) proposed that Vrancea seismicity is the consequence of a Rayleigh-Taylor instability affecting the continental mantle lithosphere and supported this view by showing that the seismic moment release in the Vrancea seismogenic zone may explain the modeled drip instability. It should also be noted that although some models for the Vrancea seismogenic zone include both subduction and subsequent delamination events (i.e., Sperner et al., 2004; Girbacea and Frisch, 1998), we include them with the models that suggest a subduction origin for the Vrancea seismogenic zone. In contrast, the models proposed by Knapp et al. (2005), Houseman and Gemmer (2007), and Lorinczi and Houseman (2009) neither require nor involve subduction of oceanic lithosphere.

The Vrancea seismic zone is a narrow 30 × 70 × 200 km steeply NW-dipping (∼86°) body within the upper mantle beneath the East Carpathian bend zone. In the published literature, this mantle seismicity is widely attributed to the final stages of subduction (e.g., Wortel and Spakman, 2000) in the Carpathians. However, there are geologic and geophysical observations of post-Miocene geologic evolution that are spatially and temporally discordant with the subduction interpretation. These observations include the position of earthquake hypocenters as related to the inferred location of a proposed suture (Linzer et al., 1998; Wortel and Spakman, 2000). The purported suture, if existent, is thought to be located west of the current mantle seismicity beneath the flysch nappes of the East Carpathians (Girbacea and Frisch, 1998; Chalot-Prat and Girbacea, 2000; Sperner et al., 2004). The interpretation of subduction for the Vrancea seismogenic zone has led to the assumption that the mantle seismicity represents either (1) the present location of a subducted slab, or (2) a remnant slab that was subducted westward from the Vrancea seismogenic zone in pre-Miocene time, and underwent detachment and retreat toward the southeast to correspond with the zone of mantle hypocenters.

The presence of a linear Miocene–Pleistocene volcanic chain located in the East Carpathian hinterland has been interpreted as further evidence for subduction. The age of this volcanism decreases from older (11 Ma in the northwest) to younger (0.03 Ma in the southeast) toward the East Carpathian bend zone (Pécskay et al., 2006). Calc-alkaline rock types are mainly andesites and dacites, while the alkaline rocks found in the Persani Mountains near the bend zone are basalts (Peltz et al., 1973; Downes et al., 1995; Mason et al., 1996). However, the relationship of these volcanic rocks to the proposed subduction is not without complication. The most problematic geodynamic example presented by the subduction hypotheses is that the magmatism continued for >8 m.y. after the purported subduction and shortening had ceased around ca. 11 Ma. Contemporaneous alkaline basalt and andesitic eruptions in the bend region during the Pleistocene further complicate the subduction models because these extrusive magmas are >100 km west of the near-vertical purported slab. The latest alkaline eruptions in the Persani Mountains occurred ∼500,000 yr ago (Pécskay et al., 2006, and references therein), and the timing and relation of this Pliocene–Pleistocene volcanism to the inferred subduction in the southeast Carpathians are some of the many questions not fully satisfied by existing subduction models for the Vrancea seismogenic zone.

Deep seismic-reflection profiles (Fig. 2) collected in collisional orogens clearly show the suture and fabric imparted to the crust from subduction and collision. The suture between the Adriatic plate and European plate (Fig. 2A) dips steeply east through the crust and upper mantle, imparting a prominent dipping fabric (Marchant and Stampfli, 1997). Similarly, Consortium for Continental Reflection Profiling (COCORP) lines Georgia 13 and 14 (Fig. 2B) image a wide band of dipping reflectivity imparted to the crust from what is interpreted as the collision of Africa with North America (Nelson et al., 1985). Zhao et al. (1993) showed the result of the collision of the Indian plate beneath Asia where the Himalayan crust overrides the north-dipping Indian crust (Fig. 2C). In each case in Figure 2, there is a predominant crustal-scale through-going dipping fabric associated with the suture indicating collision. Accordingly, Deep Reflection Acquisition Constraining Unusual Lithospheric Activity (DRACULA 1), a deep seismic-reflection profile located in the East Carpathian hinterland, was designed to test geodynamic models that imply a subduction-related suture located at depth in crust of the East Carpathian hinterland. The DRACULA I deep seismic profile complements a previous deep seismic-reflection profile called Danube and Carpathian Integrated Action on Process in the Lithosphere and Neotectonics (DACIA-PLAN), which targeted the lithospheric architecture of the East Carpathian foreland above the Vrancea seismogenic zone (see Panea et al., 2005). Presented together the DRACULA I and reprocessed DACIA-PLAN reflection profiles provide a new, ∼120-km-deep lithospheric cross section of the southeastern Carpathians over the Vrancea seismogenic zone (Fig. 3). We processed the combined seismic profiles to enhance the crustal architecture of the Eastern Carpathians in order to test the delamination and subduction hypotheses on an orogenic scale. Specifically, the presence or absence of crustal-scale through-going dipping reflections on the seismic profile will facilitate the evaluation of the myriad geodynamic models proposed for the Vrancea seismogenic zone and evolution of the Eastern Carpathians.

Two deep seismic-reflection profiles, DRACULA I and DACIA-PLAN (originally in Panea et al., 2005), collected in 2004 and 2001, respectively, image the continental crust and upper mantle to depths of ∼120 km beneath the Transylvanian basin, the East Carpathian Mountains, and the Focsani foreland basin, Romania (Fig. 1). The DRACULA I profile extends southeast from the central Transylvanian basin across the Eastern Carpathians and terminates on the northwest side the Vrancea seismogenic zone. The DACIA-PLAN profile extends over the Vrancea seismogenic zone SE onto the Focsani foreland basin, which overlies the Moesian Platform. These two deep seismic-reflection profiles form an ∼320-km-long strike perpendicular profile of the lithosphere of the Eastern Carpathian orogen.

Both the DRACULA I and DACIA-PLAN seismic data sets utilized explosive sources. The DRACULA I profile was collected in roll-along mode, maintaining a 32 km active spread with far offsets of ∼10 km. The DRACULA I data are nominally 16-fold with a record length of 60 s and a 4 ms sample rate. The DACIA-PLAN profile was collected in three static deployments, which were merged to form the complete section (see Panea et al., 2005). Seismic data processing of DRACULA I and reprocessing of DACIA-PLAN seismic profiles included noise filtering, coherency filtering, stacking, and depth conversion. The processed seismic profiles comprise the reflection data shown in Figure 3A. Lateral and vertical variations in reflective quality on the final stacks prompted noise and amplitude analysis to test whether the reflective character was caused by geologic or acquisition effects. Amplitude analysis performed on these data concluded that seismic signal penetration regularly exceeded depths of 40 km which suggests the variation in reflectivity is likely geologic. The lateral variability of reflectivity presents a problem for robust interpretation; nevertheless, we feel these data provide an adequate basis for testing of the subduction versus delamination argument in the Eastern Carpathians.

The combined seismic profiles DRACULA I and DACIA-PLAN delineate the crustal geometry of the Transylvanian basin, East Carpathian fold-and-thrust belt, and the Focsani basin in relation to the Vrancea seismogenic zone. Vrancea seismogenic zone hypocenters projected onto the plane of the DRACULA I and DACIA-PLAN seismic profile show the relationship between the seismicity and the overlying crust (Fig. 3A). Projected hypocenters include all earthquakes M_w ≥ 3.0 recorded digitally since 1977 and were taken from the Romanian National Institute for Earth Physics (NIEP) earthquake catalogue (Grecu, 2007). The hypocenters were projected from their locations perpendicular to the plane of the seismic section. The seismicity is located on the western portion of the DACIA-PLAN profile, and the intermediate-depth seismicity is located NW of the Focsani basin below the eastern slopes of the Eastern Carpathians. For purposes of describing the location of specific features shown in Figure 3, the depth description will be followed by distances measured in kilometers from west (0 km) to east (320 km) as indicated at the top of the seismic section.

The lithospheric-scale structural features evident on the DRACULA I and DACIA-PLAN deep seismic profile include: (1) coherent 42–46-km-deep reflections beneath the Transylvanian basin between km markers 0 and 25, (2) a package of reflections dipping NW ∼20° extending from 10 to 20 km depth between km markers 60 and 75, which is bounded above and below by short bands of subhorizontal reflectivity, (3) reflectivity that gradually transitions into seismically transparent material at depths of 40–45 km between km markers 25 and 80, (4) an abrupt shallowing of transitional reflectivity beneath the Eastern Carpathians to depths of 30–33 km between markers 100 and 215, except where (5) prominent subhorizontal to slightly east-dipping (5°) reflections are present at 30–32 km depth between km markers 125 and 150, and (6) the increase in reflective depth of the transitional reflectivity to 42–45 km depth beneath the Focsani basin between km markers 215 and 320. (7) The location of the intermediate-depth Vrancea seismogenic zone hypocenters are shown to lie SE of the thin transitional reflectivity beneath the eastern East Carpathians and western Focsani basin. A larger online image of the DRACULA I and DACIA-PLAN profile is available to aid examination of the features (indicated as supplemental Figure DR1¹). For a more detailed discussion of the crustal observations associated with the DACIA-PLAN, see Panea et al. (2005).

For interpretation purposes, the seismic-reflection fabrics observed at crustal levels on the DRACULA I and DACIA-PLAN profiles show good correlation from one survey to another. As such, interpretation across the profiles is considered reliable and an accurate seismic representation of the orogen. Deep transitional reflectivity and associated coherent reflectivity across the profile are interpreted to indicate the base of the crust, or Moho. Major structural architecture within the crust is apparent on both the DRACULA I profile in the form of a west-dipping ramp beneath the Transylvanian basin and in the Focsani basin as east-dipping sediments associated with deformation in the foreland. These main features are discussed in detail next.

The Moho is commonly interpreted as the base of the crust on seismic data profiles, and it represents a subsurface boundary where the seismic velocity of P waves traveling through the planet increase sharply from 7.0 to 7.4 km/s above to greater than 8.1 km/s below. The interpreted depth of the Moho on the DRACULA I and DACIA-PLAN profiles is 31–33 km beneath the Eastern Carpathians and 42–45 km beneath the Focsani basin and corresponds to the subhorizontal and transitional deep reflections observed beneath the Eastern Carpathians and Focsani basin, respectively. These Moho depths are supported by an independent coincident refraction profile (Hauser et al., 2007), which places the Moho at similar depths for these respective locations (see Fig. 3). Below the western Transylvanian basin, the deep coherent reflections are located at 42–46 km depth, with transitional reflectivity at similar depths under eastern Transylvania, nearly 10 km deeper than the 36-km-deep Moho boundary interpreted on the refraction profile. In comparing the refraction profile and the DRACULA I profile, it is unlikely that depth conversion of the DRACULA I seismic profile overestimated the depth of observed deep reflections because processing velocities were heavily constrained by the refraction velocities reported in Hauser et al. (2007). The recognition of deep reflectivity beneath the Transylvanian basin as observed on DRACULA I is, however, not unique to this profile. Raileanu and Diaconescu (1998) identified a deep interval of coherent reflectivity at 35–46 km on reprocessed industry seismic data in the area near the western half of the DRACULA I profile, which they interpreted as a crust-mantle transition zone. Knapp et al. (2005) argued successfully that the deepest reflections at ∼46 km on the same profile referenced by Raileanu and Diaconescu (1998) represent the base of the westward extension of crust belonging to the Moesian platform. Receiver function analysis by Diehl et al. (2005) elucidated the three-dimensional (3-D) nature of the Moho in the region of the southeast Carpathians, which suggests that the Moho beneath the Transylvanian basin proximal to DRACULA I may be as deep as 38.4 km, with an error of 1.3 km. Thus, based on the preexisting evidence for deep crustal reflectivity beneath the Transylvanian basin, receiver functions proximal to the reflection profile, and the new results from DRACULA I, we propose an alternative interpretation for the Moho beneath the Transylvanian basin at ∼44–46 km depth. Regardless of the discrepancy in the thickness of the crust under the Transylvanian basin among the refraction, reflection, and receiver function data discussed here, these data sets similarly suggest that the root of the East Carpathians is missing or thin and that the crust supporting the Focsani and Transylvanian basins is up to 10 km thicker than the orogenic crust.

Based on these arguments, we offer the following as an alternate interpretation of crustal thickness of the Eastern Carpathians as observed on the DRACULA I and DACIA-PLAN reflection profiles. Beneath the Transylvanian basin, the crustal thickness is interpreted to be as deep as ∼46 km, until it abruptly thins to 30–33 km below the western East Carpathian fold-and-thrust belt. The thin portion of the crust continues SE onto the DACIA-PLAN profile and over the Vrancea seismogenic zone with a thickness varying from ∼33 to 34 km. Under the eastern portion of the orogens, the crustal thickness increases from 33–34 km to 41–46 km beneath the Focsani basin. This interpretation shows an orogen absent of a crustal root.

The Vrancea seismogenic zone intermediate-depth seismicity lies SE of the apparent thinned crust of the orogen. Crustal-depth earthquakes extend from shallow crustal depths (5–10 km) to the base of the reflections at ∼35 km beneath the Focsani basin. Viewed on Figure 3A, the distance between the top of the intermediate seismicity (∼70 km) and deepest events is 130 km. This distance corresponds well to the 125–130 km horizontal dimension of the thin orogenic lower crust interpreted here. We interpret this relationship as possible evidence for the former location of the lithosphere now represented by the Vrancea seismogenic zone. Girbacea and Frisch (1998), Chalot-Prat and Girbacea (2000), and Sperner et al. (2004) also proposed in their models that the Vrancea seismogenic zone restored horizontally beneath the Eastern Carpathians and was the former mantle lithospheric root beneath the Eastern Carpathians. The interpretation we present here suggests that continental mantle lithosphere removed from beneath the orogen also may include the removal of a portion of the continental lower crust.

Dipping seismic reflections at crust or mantle depth that would indicate subduction geometries are not manifest in any form on the seismic profile presented here. On the DRACULA I and DACIA-PLAN section, there are only two locations where significant dipping reflections exist, albeit not on a scale at which the whole crust is affected. One set of dipping reflections in the eastern Transylvanian basin is located in the middle crust and is more than 100 km NW of the observed seismicity. The second is in the shallow tilted sediments associated with the Focsani foreland basin. The dipping feature beneath the Transylvanian basin could mark flexure of the crust in response to subduction. However, subhorizontal reflections below and to the SE of this feature preclude flexure because they would likely be dipping as well in the case of subduction. Furthermore, this would suggest the proposed suture to be located beneath the Transylvanian basin, a result not predicted by any current Vrancea seismogenic zone model (i.e., Girbacea and Frisch, 1998; Chalot-Prat and Girbacea, 2000; Gvirtzman, 2002; Sperner et al., 2004; Heidbach et al., 2007; Schmid et al., 2008). If this dipping feature is a suture, it is not traceable to the surface and would preclude an interpretation that suggests a Miocene suture exists in the East Carpathian hinterland. Therefore, there is little, if any, evidence for a NW-dipping fabric in the Transylvanian or Carpathian crust to appeal for a remnant suture zone indicating a former plate boundary according to these data.

On the seismic section, the NW-dipping reflections in the crust under the Transylvanian basin are interpreted to follow a crustal ramp over which the allochthonous crystalline and sedimentary rocks comprising the East Carpathian mountain belt were transported during the Neogene. Though the reflections that indicate this feature are discontinuous from the top of the ramp to the SE, we interpret them as a detachment beneath the East Carpathians traced from the ramp to the foreland deformation front. Palinspastic restorations addressing estimated Miocene shortening of the East Carpathians have reported a number of values, ranging from 116 km (Burchfiel, 1976), 125 km (Knapp et al., 2005), and 130 km (Roure et al., 1993) to 220 km (Morley, 1996). The distance from the top of the ramp to the tip of the blind thrust deforming the Quaternary Focsani basin sediments is ∼140 km. This estimated shortening measurement is greater than, but conforms to, the estimates of shortening of 125 km (Knapp et al., 2005) and 130 km (Roure et al., 1993).

Although there is no direct evidence for an eclogite root in the East Carpathians, the phase transformation of granulite-facies lower crust to eclogite is thought to be an important step in orogenic evolution and necessary for delamination (Austrheim, 1991; Nelson, 1991; Kay and Mahlburg Kay, 1991; Bousquet et al., 1997; Jackson et al., 2004). Jackson et al. (2004) applied the reasoning that fluids interacting with granulite lower-crustal rocks in the eclogite stability field (T > 500 °C and P > 1.2 GPa, approximately >40 km) are necessary to catalyze the phase change to eclogite. Further, these authors suggested that the absorption of fluids into granulite would be a slow process if only diffusion occurs, but would be accelerated if fluids moved through fracture pathways formed by brittle deformation. Hydrous mineral phases present in basement rocks of the Moesian Platform involved in shortening in the Eastern Carpathians likely contain sufficient water to drive metamorphism to eclogite facies. Moreover, fluid or additional fluids could be derived from mantle metasomatized prior to Miocene shortening (Morogan et al., 2000; Chalot-Prat and Girbacea, 2000). Miocene shortening in the Eastern Carpathians was caused as a result of Alpine collision leading the eastward escape and clockwise rotation of the Tiza-Dacia microplate during collision with the East European and Moesian Platforms (Csontos, 1995; Csontos and Voros, 2004; Roure et al., 1993; Royden, 1993). This collision shortened the crust and depressed it into the eclogite stability field and likely provided the mechanical process needed to accelerate fluid mobility in the East Carpathian root, leading to eclogitization. Therefore, we hypothesize that the formation of the East Carpathian root during Miocene shortening may have undergone a metamorphic phase change to eclogite facies, which later drove delamination.

Given the thin orogenic root presented in this work and the apparent lack of evidence for the presence of a subduction boundary within these data, we suggest the following sequence for the Miocene to present tectonic evolution of the Eastern Carpathian Mountains (Fig. 4). Beginning in the early Miocene and terminating at ca. 11 Ma, the Eastern Carpathians underwent ∼140 km of shortening due to the eastward escape of the Tiza-Dacia microplate in response to the Alpine-Tethys closure and collision. The shortening observed in the Eastern Carpathians is largely thought to have been accomplished by the consumption of a marine basin floored by oceanic lithosphere (Sperner et al., 2004; Ustaszewski et al., 2008). The Vrancea seismogenic zone is often interpreted as the last remnant of this putative ocean basin (Wortel and Spakman, 2000). However, unequivocal evidence that the Vrancea seismogenic zone was directly part of a subducted ocean basin has never been substantiated, although numerous studies commonly interpret it as such, perhaps owing to early geophysical interpretations of Vrancea seismicity (i.e., Fuchs et al., 1979). Sandulescu (1988) argued that the consumed basin was underlain by attenuated continental crust, which would seem likely, since Panea et al. (2005) and Mucuta et al. (2006) described grabens of possibly Mesozoic age beneath the Focsani basin. Based on reprocessed petroleum industry seismic data, Knapp et al. (2005) argued that the East Carpathian foreland is floored by the Moesian basement that extends west beneath the Transylvanian basin. Due to the lack of subduction fabrics on the DRACULA I and DACIA-PLAN seismic profiles, we favor the latter two arguments by Sandulescu and Knapp, that the basin accommodating shortening during Miocene time in the Eastern Carpathians was floored by basement of continental affinity and that the shortening was intracontinental in nature. By the end of the Miocene–early Pliocene, the orogenic crust, thickened by nappe emplacement, underwent a metamorphic phase change to eclogite facies, which caused densification of the orogenic root and began delamination. Gravity and isostatic modeling by Sperner et al. (2004) show that thickening of the crust and lithosphere beneath the orogen as a result of collision would cause subsidence above the thickened orogenic root, and that following delamination of the continental lithosphere, the locus of subsidence would migrate eastward in response to the subcrustal load. Matenco et al. (2007) and Leever et al. (2006) reported that from Pliocene–Pleistocene time, there was a SE shift in the accommodation and subsidence of the Focsani basin. While these authors attributed this phenomenon to lithospheric buckling, this same subsidence response may well have occurred and was predicted by Sperner et al. (2004) as a result of detachment and sinking of delaminated continental lithosphere. In Pliocene time, and continuing into the Pleistocene, the delaminated root descended into the upper mantle toward its current location. We suggest that the delaminated orogenic root is now represented by the Vrancea seismogenic zone and that this delamination was initiated prior to the onset of Pleistocene alkalic and calc-alkaline volcanism in the southern East Carpathians, dated at 2.25 and 2.1 Ma, respectively (Downes et al., 1995; Pécskay et al., 2006), and was contemporaneous with uplift of the southeast Carpathians and subsidence in the Focsani foreland basin (Sanders et al., 1999; Krezsek and Bally, 2006; Leever et al., 2006; Matenco et al., 2007).

Teleseimic tomography used to image the mantle structure beneath the East Carpathians shows a seismically fast (cold and/or higher density) volume in the mantle beneath the Focsani basin, with the Vrancea seismicity located in or on its NW edge (Wortel and Spakman, 2000; Weidle et al., 2005; Martin et al., 2005). This observation is consistent with the interpretation presented here in Figure 3B, that the seismogenic body may represent a seismically fast portion of crust removed due to delamination. Delamination of the orogenic root beneath the Eastern Carpathians also provides means for asthenospheric mantle to rise to depths as shallow as 30–35 km in place of the “missing” lower crust on the seismic section. Asthenospheric upwelling as a result of delamination may be (1) the cause of uplift of the southeast Carpathians during the Pliocene–Pleistocene (Sanders et al., 1999), (2) the source of the deep-seated alkaline volcanism in the Persani Mountains (Downes et al., 1995; Pécskay et al., 2006), (3) the cause of the low-seismic-velocity anomaly located NW of the Vrancea seismogenic zone in tomographic studies of the region (Wortel and Spakman, 2000; Weilde et al., 2005; Martin et al., 2005), and (4) the cause of attenuation of seismic waves generated in the Vrancea seismogenic zone as recorded in Transylvania (Russo et al., 2005).

From the deep seismic-reflection data shown here, we cannot advocate the presence of a suture zone indicating the location where an ocean basin was consumed beneath the East Carpathians. These data suggest that the root of the southeast Carpathians is absent, as indicated by ∼32–33-km-thick crust beneath the orogen and two-dimensional geometry of the missing crust that matches closely that of the intermediate-depth Vrancea seismogenic zone hypocenters. Though we also cannot unequivocally state that delamination of lower crust and mantle lithosphere is the cause of the Vrancea seismogenic zone and source of the geodynamic processes observed in the Eastern Carpathians, we propose that numerous geologic and geophysical observations support our interpretation for crust-inclusive continental lithospheric delamination. The horizontal length of the missing lower crust corresponds to the depth dimension of the intermediate Vrancea earthquakes and suggests a direct link between the seismic body and the thin East Carpathian crust. Due to the two-dimensional nature of this study, the extent of the continental delamination beneath the Eastern Carpathians can only be postulated. Nevertheless, the delamination hypothesis proposed here for the origin of the Vrancea seismogenic zone may represent the final stages of continental lithospheric delamination beneath the length of the Eastern Carpathians. The model presented here suggests that mantle-depth seismicity associated with the Vrancea seismogenic zone is unlikely to be the result of subduction and that continental lithospheric delamination may be an alternate process capable of producing mantle-depth earthquakes in the absence of subduction tectonics.

We thank Greg Houseman and two anonymous reviewers for comments that strengthened this manuscript. We thank Liviu Matenco for insightful discussion and suggestions that have also improved this work. The authors would like to acknowledge the National Science Foundation for providing the funding to complete this research through the Tectonics Program grant, EAR-0310118. We thank Incorporated Research Institutions for Seismology (IRIS) and the Passcal instrument pool for providing recording instrumentation and field support for data acquisition. Fieldwork was carried out through collaboration with the University of Bucharest, Romania, the University of South Carolina, and the National Institute for Earth Physics, Bucharest-Magurele, whose faculty, staff, and students comprised team DRACULA. Data processing was completed using ProMAX software supplied to the University of South Carolina on a grant from Landmark Graphics Corporation.