Aseismic ridge subduction as a driver for the Ordovician Taconic orogeny and Utica foreland basin in New England and New York State
Published:December 28, 2018
- PDF LinkChapter PDF
Robert D. Jacobi, Charles Mitchell, 2018. "Aseismic ridge subduction as a driver for the Ordovician Taconic orogeny and Utica foreland basin in New England and New York State", Tectonics, Sedimentary Basins, and Provenance: A Celebration of the Career of William R. Dickinson, Raymond V. Ingersoll, Timothy F. Lawton, Stephan A. Graham
Download citation file:
Aseismic ridge subduction is common along modern convergent margins. We enumerate six criteria that can be used to recognize aseismic ridge subduction in orogens, including a magmatic gap with uplift followed by bimodal volcanism, which commonly includes explosive, voluminous rhyodacitic volcanism that erupts far from the trench. Features temporally linked with the explosive volcanism include retroarc thrusts and consequent thrust-loaded retroarc foreland basin development.
Using these criteria to examine features of the Taconic orogen, together with new stratigraphic and structural data from the Utica basin that constrain the basin subsidence architecture and thrust timing, we propose that at least the older units of the 456–435 Ma Oliverian Plutonic Suite in New England were generated during steepening of the downgoing slab after passage of a subducting aseismic ridge. Weakened crust from delamination and decompression melting promoted westerly directed thrusts (present-day coordinates) that loaded the Taconic retroarc foreland. The resulting Utica basin subsided rapidly and nearly synchronously over an ~150-km-wide region and contains interbedded 453–451 Ma ash layers from the Oliverian Plutonic Suite or coeval plutons to the south.
This history of basin subsidence indicates that the major thrust loads that drove development of the Utica basin were emplaced over a similarly brief interval beginning ca. 455 Ma. Thus, the Taconic thrusts, the Utica basin, the volcanic ashes, and the early Oliverian felsic magmatic units could all be related to an aseismic ridge subduction event. Because of the ubiquity of seamount chains, we expect that aseismic ridge subduction affected other segments of the Taconic orogen.
Subduction of aseismic and seismic ridges is a natural result of closure of an oceanic basin. As used here, an aseismic ridge refers to a seamount chain, ridge, or oceanic plateau located away from a spreading center, and a seismic ridge refers to a seafloor-spreading center (e.g., Vogt, 1973; McCann and Sykes, 1984). The consequent dynamics of aseismic ridge subduction (which include flat-slab subduction, arc jumps, and, late in the process, continental melting from slab steepening with asthenospheric upwelling) can be found, for example, along the western margin of South America, where aseismic ridge subduction is currently taking place and has likely affected much of the margin at some point during the Cenozoic (e.g., Ramos and Folguera, 2009). It is probable that, locally, ocean basin closure can involve strike-slip tectonics (e.g., Waldron et al., 2014), but ultimately ridge subduction must occur during complete closure of an ocean basin.
The northern Appalachian orogen of the United States and Canada has been a laboratory for the development and revision of plate-tectonic models, beginning with the seminal papers of Bird and Dewey (1970) and Dewey and Bird (1971), and continuing to the present (e.g., Macdonald et al., 2014, 2017; van Staal et al., 2016; Karabinos et al., 2017).
This paper reviews tectonic models for the Taconic orogeny and elements of aseismic ridge subduction. For the effects of seismic ridge subduction, see, for example, Bradley et al. (1993), Bourgois et al. (1996), Santosh and Kusky (2010), Tang et al. (2010), Seton et al. (2015), and van Staal et al. (2016). We present new data from the Taconic foreland basin bearing on potential elements of aseismic ridge subduction. These new stratigraphic and structural data are from the Utica black shale and stratigraphically higher, coarser clastic units in the Mohawk Valley of New York State. We also examine other potential elements of Taconic ridge subduction such as the alkalic/rhyodacitic units in the Bronson Hill arc. Finally, we modify Taconic tectonic models for the Appalachian orogen in New England and New York State by incorporating these aseismic ridge subduction elements.
Recent Taconic Plate-Tectonic Models
The most recent Taconic tectonic models for western New England rely heavily on advances in detrital zircon data and radiometric dates (Macdonald et al., 2014, 2017; Karabinos et al., 2017; for details of these and older models, see Appendix A). In these recent models, the Shelburne Falls arc initiated on the peri-Gondwanan Moretown terrane at ca. 502 Ma above an east-dipping subduction zone on the east side of the Iapetus Ocean (present-day coordinates), away from the influence of Laurentia (Fig. 1A; Macdonald et al., 2014, 2017; Karabinos et al., 2017). As Iapetus closed, the arc migrated toward a microcontinent of Laurentian affinity that lay on the west side of Iapetus. This microcontinent, represented by the Rowe belt and possibly the Green Mountain massif, provided zircon-bearing detritus to the Hawley Formation, which is interpreted to be a metavolcanic and metasedimentary unit of the Shelburne Falls arc (Macdonald et al., 2014, 2017; Karabinos et al., 2017). The Rowe belt microcontinent was separated from Laurentia by a relatively narrow oceanic tract, the “Taconic seaway.”
Closure of the Iapetus Ocean at ca. 475 Ma led to the subsequent closure of the Taconic seaway, and early Taconic thrusts began forming (Macdonald et al., 2014, 2017; Karabinos et al., 2017) in response to crustal shortening or subduction or both. Following these events, breakoff of the eastwardly subducted oceanic slab at ca. 466 Ma and attendant asthenospheric upwelling resulted in (1) late-stage mafic rocks with suprasubduction-zone signatures that intruded the remnant forearc of the Shelburne Falls arc (Kim et al., 2003; Coish et al., 2015), and (2) the 466.0 ± 0.1 Ma felsic units in the Barnard Volcanic Member of the Missisquoi Formation (a correlative of the Hawley Formation in the Shelburne Falls arc) and time-equivalent volcanic ashes in the Indian River Formation located in the Taconic thrust slices (Karabinos et al., 2017; Macdonald et al., 2017). At about the time of slab breakoff, the bimodal Ammonoosuc Volcanics developed (ca. 469–461 Ma) in the Bronson Hill arc, which lies east of the Shelburne Falls arc (Fig. 1A). The origin of the Ammonoosuc Volcanics is unclear in these models, but may have been related to slab breakoff or subduction polarity reversal.
After closure of the Taconic seaway, a new, west-dipping subduction zone (ca. 456–442 Ma) developed beneath the composite Laurentian margin. A second phase of magmatism in the Bronson Hill arc, the Quimby sequence (which includes the Quimby Formation and the ca. 456–435 Ma Oliverian and Highlandcroft plutonic suites), was generated over the west-dipping subduction zone. The Quimby Formation includes bimodal volcanics with a 443 ± 4 Ma age (Moench and Aleinikoff, 2003) or ca. 455 Ma age (Karabinos et al., 2017). The black shales of the Utica Group, which are dated at ca. 453–450 Ma, based on U-Pb zircon geochronology (Sell et al., 2013, 2015; Macdonald et al., 2017), were deposited during this orogenic phase. The final emplacement of the Taconic thrusts also occurred at this time. The resulting thrust loads and dynamic effects associated with the downgoing slab caused subsidence of the Utica foreland basin (e.g., Macdonald et al., 2014, 2017).
This tectonic model, which involves (1) Middle Ordovician collision of a Gondwanan arc above an east-dipping slab, (2) limited or no eastward subduction beneath a Laurentian microcontinent, and (3) younger westward subduction beneath a composite Laurentian margin, is consistent with (1) proposals for Late Ordovician westward subduction at ca. 450 Ma farther south in Connecticut (Sevigny and Hanson, 1993, 1995; Walsh et al., 2004; Aleinikoff et al., 2007; Chu et al., 2016), (2) the discovery of juvenile input to arc detritus in the Utica retroarc foreland from εNd studies (Macdonald et al., 2017), and (3) tilting of Laurentia (Coakley and Gurnis, 1995).
In the west-following-east subduction models, the age of final emplacement of the Taconic allochthon and the age of subsidence of the Utica basin demand that these events occurred in a retroarc position, continent-ward of the Bronson Hill arc. The questions we address here are the following: (1) What driving mechanism caused these events to occur in a retroarc location? (2) What is the relationship among the retroarc thrusts, Utica basin subsidence, K-bentonites, and the Oliverian Plutonic Suite? We suggest that aseismic ridge subduction and its effects should be considered as a causal mechanism for all four of these elements. Further, we suggest that aseismic ridge subduction should be tested and incorporated into other models and phases of the Taconic orogeny that are reviewed in Appendix A. Subduction of both seismic and aseismic ridges has been incorporated previously into Taconic tectonic models for the Canadian Appalachians and Maine, albeit without a flat-slab component (see Appendix A for details).
Utica Foreland Basin
The Utica foreland basin (Vermontian phase; sensu Rodgers, 1971), which was the site of deposition of the Utica Group, is best exposed in the Mohawk Valley of New York State (Fig. 2). There, the 453–450 Ma black shales overstep westerly the Cambrian–Ordovician “great American carbonate bank” of the Laurentian margin (e.g., Landing, 2012; see also Macdonald et al., 2017, their figure 2). In models in which Laurentia was part of the subducting lower plate, as its Iapetan margin approached and entered an east-dipping subduction zone during Utica time, the Utica basin was essentially the trench, or more specifically, the outer wall of the trench, i.e., a proforeland basin (Jacobi, 1981; Rowley and Kidd, 1981; Bradley and Kidd, 1991; following the classification of Ingersoll, 2012). In contrast, in models with a west-dipping subduction zone east of the Bronson Hill arc, the Utica basin was a retroarc foreland basin that developed on the overriding plate (e.g., Macdonald et al., 2014, 2017; following the classification of Ingersoll, 1988, 2012). Both sets of models are compatible with observed NNE/NE-striking normal faults west of the Taconic thrusts (Fig. 2) that were active both before and during deposition of the Utica Shale; these faults record 1.2 km of down-to-the-east subsidence, based on cumulative throw (Bradley and Kidd, 1991), or ~0.5 km, based on subsidence curves (Macdonald et al., 2017). For details concerning the reactivation history of these faults, see Jacobi and Mitchell (2002) and Jacobi (2010, 2011).
In the tectonic models with an east-dipping subduction zone, the westward advance of the Utica black shale over the Ordovician carbonate bank should record the passage of the Laurentian continental margin over the peripheral bulge and into the trench, since the Utica black shale was thought to represent trench deposition (e.g., Rowley and Kidd, 1981; Bradley and Kidd, 1991). Indeed, Bradley and Kusky (1986) proposed that the relative convergence rate could be calculated from the black shale overlap. On the basis of limited graptolite localities in the black shale, Bradley and Kusky (1986) calculated an approximate convergence rate of 2–3 cm/yr, assuming a steady convergence rate marked by the apparent constant rate at which the Utica black shale onlapped the subsiding Laurentian carbonate bank (Fig. 3). This rate of convergence is consistent with modern convergence rates. Bradley and Kusky (1986) and Bradley and Kidd (1991) pointed out that the cumulative throw on the Utica basin normal faults in the Mohawk Valley and the St. Lawrence Lowlands of Quebec increases eastward toward the Taconic thrust front, and they proposed that this increase was similar in magnitude to those in modern convergent margins. The Utica basin, thus, was thought to be consistent with eastward subduction and to represent a trench fill that spilled out over the Laurentian margin during final collision. We provide new evidence below that indicates the Utica basin is not an usual convergent-trench, pro-foreland basin. These new data are consistent, however, with retroarc foreland basin tectonics resulting from ridge subduction.
Elements of Aseismic Ridge Subduction and Flat-Slab Subduction
Our review of the tectonic features and geological history of the Caribbean Antilles, Middle America Trench, and Andean margin (Appendix B) indicates that aseismic ridge subduction beneath island arcs and convergent continental margins typically results in some or all of the following spatial and temporal characteristics:
(1a) uplift of the accretionary prism (with consequent unconformities) as the buoyant aseismic ridge on the downgoing oceanic slab passes beneath the upper plate, and
(1b) significantly more deformation in the accretionary prism where the aseismic ridge intersects the prism, with a possible trail of ocean-island basalt fragments ripped from the aseismic ridge and incorporated into the prism;
(2) flat-slab to moderately dipping subduction of the buoyant, aseismic ridge–bearing plate segment;
(3) significantly reduced volcanic activity in the original arc (“magmatic gap” or “volcanic gap”) above the flat slab;
(4) significant uplift of parts of the upper plate (during initial steepening of the downgoing slab after the ridge passes, as well as during flat-slab subduction in some areas);
(5a) continental delamination with alkalic/rhyolitic volcanism that displays continental partial-melt signatures during slab steepening after flat-slab subduction, and
(5b) alkalic/rhyolitic volcanism associated with large calderas that can be relatively far-removed (~550–800 km) from the trench; and
(6) increased retroarc foreland thrusting of both basement and sedimentary cover resulting from an interplay between convergence rate and crustal weakening from heating and delamination (Fig. 4).
POTENTIAL TACONIC FLAT-SLAB SUBDUCTION ELEMENTS
Several Taconic structural, stratigraphic, and magmatic features in New England and New York State potentially represent some of the aseismic ridge elements listed here (and described in detail in Appendix B). However, these elements do not all have the same probability of being observed in the regional and local geology of New England and New York State. For example, element 1a (uplift of the accretionary prism as the buoyant aseismic ridge passes, with consequent unconformities) will be difficult to recognize in New England outcrops since the prism commonly lowers after the ridge has passed (see Appendix B).
Element 1b (significantly more deformation in the accretionary prism where the aseismic ridge intersects the prism) was proposed as an explanation for the Taconic Dunnage Mélange in Newfoundland, but more recent geochemistry of intrusions into the Dunnage unit suggests that the mélange marks the passage of a seismic ridge during subduction (see Appendix A). The Hurricane mélange in Maine also has been ascribed to seismic ridge subduction, rather than aseismic ridge subduction (Schoonmaker and Kidd, 2006; see Appendix A). Element 4 (significant uplift of parts of the upper [overriding] plate) should be recognized as unconformities associated with reverse faults and (commonly) subaerial deposition. Because of the number of tectonic and metamorphic events overprinted in New England, it will be difficult to identify this element, especially if the uplifted region was relatively local. Next, we review Taconic components that may reflect other elements of flat-slab subduction.
Utica Retroarc Foreland Basin Development
For tectonic models with westward subduction at the time of Utica deposition (453–450 Ma), west-directed Taconic thrusts and the development of the Utica retroarc foreland basin might locally represent flat-slab Element 6 (retroarc thrusting and thrust-loaded basin development). As described in Appendix B, retroarc thrusting and consequent basin development are generally thought to be a result of lithosphere weakening combined with a fast relative convergence rate.
Foreland Basin Subsidence: Timing and Geometry
Fast overstep of basal Utica Shale. In the twentieth century, Appalachian geologists generally thought that the base of the Utica Shale gradually overlapped the Laurentian margin carbonate bank (e.g., Ruedemann, 1912; Kay, 1943; Kay and Colbert, 1965). Such a scenario was consistent with eastward subduction models in which the Laurentian plate margin steadily passed over the peripheral bulge and into the trench under a constant supply of trench-related muds (e.g., Jacobi, 1981; Rowley and Kidd, 1981; Bradley and Kusky, 1986; Bradley and Kidd, 1991). However, recent work that integrated biostratigraphy (graptolites), lithostratigraphy, and tephrochronology in measured outcrops and core (Fig. 3) has demonstrated that the depositional front of the black shale raced across this part of the carbonate margin—covering ~100 km in less than 200,000 yr, which is essentially instantaneously within the limits of the chronostratigraphic resolution. This rapid transgression took place in latest Climacograptus bicornis Zone time, immediately before the Corynoides americanus Zone (Fig. 3). The age of this transgression is ca. 452.7 Ma, bracketed by the Millbrig K-bentonite at 452.86 ± 0.29 Ma in the upper part of the underlying Black River Group limestones (Mitchell et al., 2004; Sell et al., 2013, 2015) and the Sherman Falls K-bentonite at 452.62 ± 0.06 Ma in the basal Utica Shale (Macdonald et al., 2017). Recalibration of the radiometric ages of the graptolite zones employed by Bradley and Kusky (1986) with the new tephrochronology brings Bradley and Kusky’s (1986) ages of the basal Utica at their three localities into approximate alignment with our new chronostratigraphy that implies a fast transgression (Fig. 3).
Our new overlap curve suggests that the depositional front of the black shale migrated westward across the basin at a rate of ~50 cm/yr or faster. Following Bradley and Kusky’s (1986) method of directly equating the rate of shale overstep to the relative rate of plate convergence (but for a more nuanced view, see Naylor and Sinclair, 2008), this rate is unrealistically high compared to known relative convergence rates at a trench, and it is especially high compared to relative convergence rates associated with continental collisions. Consequently, there must have been a significant dynamic component in the overstep signal not related directly to plate convergence.
West-thickening wedges in basal Utica Shaleand correlatives farther west: Back-rotated fault blocks. Inspection of a more comprehensive lithostratigraphic cross section (Fig. 5) allows us to generate a more complete picture of Taconic foreland subsidence. A thin C. bicornis Zone succession can be traced through the basal Utica Shale as far west as Canajoharie, New York (Figs. 2 and 5). Across this region, the basal Utica strata and the subjacent Trenton Group exhibit a series of fault-bounded wedges that each thicken to the west. This pattern is most clearly evidenced in the basal Utica strata by tracing the base of C. americanus Zone, the Sherman Falls K-bentonite bed (Fig. 5, ash #1), and the Kuyahoora K-bentonite bed (Fig. 5, ash #2; both were named by Brett and Baird  but correspond to geochemically fingerprinted beds correlated in Mitchell et al. ). These westward-thickening wedges are bounded on the west by the Hoffmans, Noses, Little Falls, and Poland faults. The wedges suggest the presence of a back-rotated or listric fault system (accompanied by subsidiary, antithetic step-downs) as the dominant architecture of the Mohawk Valley fault system. The architecture of back-rotated fault blocks is consistent with proposed west-dipping paleoslopes in the Utica Shale based on slump fold orientations (Jacobi and Mitchell, 2002).
This pattern of listric fault–driven subsidence continues west of the Little Falls fault, where the lower Utica Shale grades rapidly into the dominantly carbonate mudstone strata of the middle Trenton Group (Poland and Russia Limestones; see Mitchell et al., 1994; Brett and Baird, 2002; Brett et al., 2004). At Wolf Hollow Creek (Fig. 5, column 14), in the eastern portion of this wedge, the High Falls K-bentonite (ash #9) lies 12.1 m above the Sherman Falls K-bentonite (ash #1), whereas in the western part of the wedge at Trenton Falls (column 16), this same ash pair is separated by 28.0 m. This geometry suggests a regional-scale, back-rotated, syndepositional, east-facing listric fault block from west of the Poland fault eastward to the City Brook fault (Fig. 5).
From the relationships described here, it is evident that rapid subsidence took place at ca. 453 Ma, and that this subsidence was accommodated by a series of back-rotated (west-dipping) half grabens that extended ~150 km into the craton from the current location of the Taconic thrust front. From the foregoing, it is also evident that the broader, step-wise diachroneity of the shale overstep shown in the cratonic portion of the revised Bradley and Kusky (1986) overlap curve (Fig. 3) is still somewhat misleading vis-à-vis the extent and timing of regional subsidence. Rapid subsidence took place across the >150-km-wide proximal foreland basin in late C. bicornis Zone time (ca. 453 Ma), but in the western part of the basin, where both subsidence and clastic supply rates were lower, local carbonate supply rates allowed the Trenton Group lithologies to persist ~2 m.y. longer than they did east of the Little Falls fault, despite being within the domain of tectonically accelerated subsidence. Differential motions across many of the faults that cut the Utica Shale (Fig. 5) indicate that tectonically driven subsidence continued to affect the Mohawk Valley region throughout deposition of the Utica unit sediments and into the overlying (and laterally equivalent) Schenectady Formation.
Utica basin depositional geometry and previous tectonic models. The Utica basin subsidence outlined above suggests that the basin likely does not represent the outer wall of a trench at a steadily east-subducting Laurentian margin; i.e., this basin was not a long-lived pro-foreland basin that gradually encroached westwardly as the Laurentian plate slid into the trench. In particular, the presence of a persistent thin, latest C. bicornis Zone black shale interval that extended across the basin, together with the near parallelism between the overlying K-bentonites and the basal contact of the Utica Shale are difficult to reconcile with features we would expect to find had these sediments been deposited on the outer margin of a trench system. Instead, the pattern of broad, synchronous basin subsidence documented above suggests the existence of some mechanism of structural coupling that (1) linked the dominantly extensional regime in the central Mohawk Valley with the dominantly compressional regime near the Taconic front and (2) allowed nearly simultaneous activation of both fault systems and broadly synchronous subsidence over this region. In the next section, we take up the question of what tectonic setting might plausibly lead to a history of subsidence of this kind.
Foreland Basin Subsidence: Implications for Structure and Tectonic Setting
Several alternative explanations for the rapid basal black shale transgression in the Saratoga–McGregor–Little Falls fault basin are possible. One is that this fast rate of overlap suggests that the basin formed in a retroarc foreland setting, possibly driven in part by dynamics of flat-slab subduction. We explore this interpretation below.
Utica subsidence: Broken crust of a retroarc foreland basin. The fast transgression and narrow Utica basin are consistent with characteristics of Andean broken-crust retroarc foreland basins, where the rigidity of the crust has been compromised by preexisting fault systems (Appendix B; e.g., Cardozo and Jordan, 2001; Jordan et al., 2001). Such is the case with the Mohawk Valley Utica basin, where it has been proposed that Iapetan opening faults were reactivated as the NNE-trending normal faults active during basin subsidence in Utica time (Bosworth and Putman, 1986; Jacobi, 2002, 2007, 2011; but for a contradictory view, see Bradley and Kidd, 1991). Additionally, some of these same faults also sustained motion at Cambrian-Ordovician boundary time (Jacobi et al., 2006) and Knox unconformity time (e.g., Bradley and Kidd, 1991). To the north in Quebec, Taconic subsidence is proposed to have been guided by reactivated Neoproterozoic/Iapetan-opening faults, some of which were also reactivated around the time of the Cambrian-Ordovician boundary and during the hiatus associated with the Knox unconformity (Dix and Rodhan, 2006; Salad Hersi and Dix, 2006; Dix and Al-Dulami, 2011; Dix and Jolicoeur, 2011; Gbadeyan and Dix, 2013). Further, faults observed on seismic reflection data in Quebec support the contention that Iapetan-opening faults were reactivated during Taconic foreland basin subsidence, since the throw on the basal Cambrian nonconformity is greater than the throw on the Ordovician units (e.g., Séjourné et al., 2002). Similar Taconic fault geometries are observed in seismic reflection data collected to the west-southwest and southwest of the Mohawk Valley (Jacobi, 2010, 2011, 2012) and are also well documented across the Rome Trough in Pennsylvania (e.g., Ryder et al., 1992; Ryder, 2014). Thus, it is clear that the crust was already broken when the Taconic thrusts loaded the Laurentian margin. In addition to the narrow width of the basin and the rapid shale overstep, the essentially symmetrical subsidence of the basin at the onset of Utica Shale deposition is also consistent with some retroarc, flat slab–related foreland basins, such as the Bermejo Basin in the Andes, which also may have been influenced by inherited structures (see Appendix B; Jordan et al., 2001).
Utica subsidence: Pinned detachment. The sharp step in basal Utica ages across the Manheim–Little Falls fault system (Figs. 3 and 5), which in turn reflects the sharp decrease in accumulation rate and accommodation space growth west of the Little Falls fault, suggests that subsidence of the western margin of the Saratoga–McGregor–Little Falls fault basin was “pinned” by the Manheim–Little Falls fault system during basal Utica time (Fig. 6). This boundary fault zone may represent the western extent of a main detachment from which the intervening faults to the east splay (Fig. 6). Further, since these normal faults might be reactivated Iapetan-opening faults, the proposed detachment might originally have been an Iapetan-opening detachment. This hypothesis may also explain the observation that NNE-striking faults exposed in the Grenvillian crystalline rocks of the Adirondack dome (north of the Mohawk River valley) and their associated prominent topographic lineaments generally are not found in abundance farther west than the northerly continuation of the Manheim–Little Falls fault system (e.g., Kay, 1937; Isachsen and McKendree, 1977; Bradley and Kidd, 1991; Jacobi, 2011; Jacobi et al., 2015). The small-throw faults in the carbonate-rich bank west of the proposed detachment ramp at the Manheim–Little Falls fault system (Fig. 5) suggest that a splay of the master décollement extended farther west and ramped up at the Poland and Prospect faults (see Figs. 1 and 2 for locations of faults).
The pinned nature with a detachment is consistent with the interpretation that the Utica basin is a retroarc foreland basin over a west-dipping subduction zone, since detachment/décollements are proposed for retroarc foreland basins related to flat-slab subduction (Appendix B; e.g., Ramos and Folguera, 2009). In retroarc foreland basins, the prominent faulting is commonly the basin (foreland)–verging thrust faulting and related folding on the hinterland side of the basin. These retroarc thrusts and folds may have an analog in the retroarc foreland basin model for the Taconic Mohawk Valley, i.e., the thrusts of the Taconic allochthon and other associated thrusts, as well as possible thrusting of the Precambrian massifs. On the foreland side of the retroarc foreland basin, normal faults typically go unremarked or unrecognized because of their small throw compared to the thrust fault systems on the hinterland side. In fact, Bradley and Kidd (1991) suggested that the normal faults in the Mohawk Valley were evidence for a convergent margin (with an east-dipping subduction zone), because at the time of their paper, normal faults in subsiding retroarc foreland basins were generally unknown; the subsiding retroarc foreland basin was then viewed as an elastic flexed beam uncompromised by faults. Since that time, however, foreland basin faults have been recognized in such retroarc foreland basins as the Acadian and Alleghanian Appalachian basin (Jacobi et al., 2013; Jacobi and Starr, 2013) and the Cretaceous seaway in Canada (note the unidentified faults in the seismic section in figures 5 and 6ofHadlari et al., 2014), and have been proposed in the flat slab–related Bermejo Basin in the Andes (see Appendix B; Jordan et al., 2001).
The detachment model is also consistent with Bradley and Kidd’s (1991) eastward subduction model in which they proposed that the Utica basin faults might sole out on a detachment (their figure 15, model D), and is consistent with a similar model for foreland faults farther north (Hayman and Kidd, 2002a, 2002b). In the context of the eastward subduction model, the back-rotated fault blocks and the rapid black shale transgression across the regional Saratoga–McGregor–Little Falls fault block could be viewed as stick-slip dynamics related to final continent-arc collision rather than steady-state convergence (with its implication of gradual subsidence).
Utica subsidence: Strike-slip component. Strike-slip motion also may have contributed to the broadly synchronous subsidence of the Saratoga-McGregor–Little Falls fault basin. In this scenario, the NNE-striking normal faults had a strike-slip component that produced a series of rhombochasms that underwent rapid and coeval subsidence (Jacobi et al., 2015, 2016). Such oblique motion has been observed in the Andes, where oblique subduction of aseismic ridges resulted in orogen-parallel translation in the arc and foreland (e.g., Morabito et al., 2011; Margirier et al., 2015). Escape tectonics away from the New York Promontory could also have contributed to the proposed oblique slip.
Utica subsidence rates compared to other tectonic settings and potential contribution of eustasy. The amount of subsidence/throw on the Little Falls fault during the interval between the Kuyahoora and Deer River K-bentonites (~0.65 m.y., ashes #2 and #3, respectively; see Fig. 5) is on the order of 43 m compacted (~211 m decompacted), which equates to a sediment accumulation rate of 66 m/m.y. compacted and 0.32 mm/yr (325 m/m.y.) decompacted. To the limited extent that this accumulation rate can be equated to a basin subsidence rate (see, for example, Naylor and Sinclair, 2008), the rate is comparable to the subsidence rate of the Miocene Bermejo thrust-loaded retroarc foreland basin, which had a subsidence rate of 0.33 mm/yr in pre-flat-slab subduction time and 0.77–0.95 mm/yr during flat-slab subduction (see Appendix B; Ramos and Folguera, 2009). A higher Utica accumulation rate (decompacted) of ~0.50 mm/yr (492 m/m.y.) in the overlying interval (Deer River K-bentonite to the 450.7 Ma ash; Fig. 5) approaches the subsidence rate of the Bermejo foreland basin during flat-slab subduction but the Utica black shale accumulation rates may not have accurately reflected the subsidence rate, and further, the accumulation rate in the Bermejo Basin was derived from coarser sediments (Appendix B; Ramos and Folguera, 2009). For comparison, the sediment accumulation rate in the intra-arc Okinawa Trough, an extensional basin that intersects Taiwan and is related to a subduction polarity flip (see section entitled “Tectonic Setting of the Ammonoosuc and Partridge Volcanics” in Appendix A), is an astounding 3.25 m/1000 yr (3.25 km/m.y.; Salisbury et al., 2002; Clift et al., 2003).
The total amount of subsidence inferred for the Utica foreland basin is quite limited—~1.2 km down-to-the-east, based on the present cumulative fault throw across 14 (+) faults from the western margin of the foreland basin to the Taconic thrust front (Bradley and Kidd, 1991), and less for the basin from the Little Falls fault to the Saratoga-McGregor fault—763 m, down-to-the-east (Bradley and Kidd, 1991). Based on the subsidence models of Macdonald et al. (2017), the subsidence across the basin during Utica time was even less: ~0.5 km. Macdonald et al. (2017) noted that their subsidence curve calculated for the Utica basin has a convex-upward form more similar to subsidence in a proforeland basin than a retroarc foreland (Naylor and Sinclair, 2008; Xie and Heller, 2009), but they suggested that the time frame is insufficient for recognition of a concave-upward curve. The cumulative throw on the faults across the Utica basin is only about half that of the Quebec Taconic foreland basin, and less than the Timor Trench (Bradley and Kidd, 1991). These relatively low subsidence rates are in accord with the distal portions of the Bermejo broken, retroarc foreland basin, where total accumulation over 2 m.y. was less than 1 km (compacted; Ramos and Folguera, 2009). If, on the other hand, the Utica basin does represent a proforeland basin that developed during short-lived eastward subduction and collision of Laurentia with the composite arc, then the relatively small subsidence could be related to several factors, including the following: (1) The continental collision had already begun, reducing the remnant trench to a shallow trough, like the Timor Trough (Bradley and Kidd, 1991), or (2) prior to final collision, aseismic ridge subduction had reduced the trench to a shallow trough (see Appendix A, element 1a, “uplift of the accretionary prism”), or (3) the New York Promontory provided extra stability.
Although it is possible that eustatic sea-level rise (e.g., Brett et al., 2004; Haq and Schutter, 2008) could have been a factor in the basal black shale transgression, even the modest total subsidence discussed above is an order of magnitude greater than the magnitude of the relative sea-level rise observed in the interior of Laurentia on the Nashville dome. There, the backstripped relative sea-level rise was ~10 m/m.y. prior to 452 Ma and 2.6 m/m.y. after (Holland and Patzkowsky, 1997). Compare these rates to a sediment accumulation rate of 66 m/m.y. compacted (~325 m/m.y. decompacted) for the Kuyahoora–Deer River K-bentonite interval east of the Manheim fault, and 100 m/m.y. compacted (~492 m/m.y. decompacted) for the Deer River–450.7 Ma K-bentonite interval east of the Dolgeville fault. These high accumulation rates, compared to the coeval “midcontinent” rates that have been attributed to eustasy (Holland and Patzkowsky, 1997), indicate that although a rising eustatic sea-level signal could be a part of the initial black shale transgression, the nontectonic component can account for neither the significant thickening documented by the Utica Shale nor the regional differences in implied subsidence rate. Rather, the thickness must be primarily due to a tectonic component, i.e., basin subsidence, a conclusion arrived at by several workers who have examined the subsidence and sedimentation rates of the Utica strata in the Mohawk Valley (e.g., Kay, 1942, 1955; Cisne et al., 1982; Lehmann et al., 1994; Brett and Baird, 2002; Mitchell et al., 2004; Brett et al., 2004; Jacobi et al., 2016; Macdonald et al., 2017).
Utica subsidence: Summary. The new integrated tephrochronological and graptolite data suggest that the Utica basin probably was not a product of relatively long-lived, eastward subduction of the Laurentian margin as it steadily passed over a peripheral bulge and into the trench. The short-lived Utica basin, even when combined with the slightly older basin recorded by the Pawlet Formation (now in Taconic thrust sheets; Fig. 3; see Appendix D; see also Rowley and Kidd, 1981), did not accommodate a long-lived convergence between the composite arcs to the east and the oceanic tract to the west. Rather, the basin subsidence is consistent with retroarc thrust and fold belts with yoked foreland basins (as reviewed in Appendix B) and is most likely a result of geodynamics in a retroarc setting related to aseismic ridge subduction in a westward subduction model.
Timing of (Retroarc?) Foreland Taconic Thrusts
Here we take up two related questions: What constraints may be placed on the timing of thrusting that created the mélange zones (and perhaps drove the observed history of basin subsidence), and, if the timing of the thrusting can be constrained, what does this timing say about how quickly the basin responded to the arrival of the thrust loads? We then examine implications of these considerations for the timing and character of the tectonic driving forces behind these events.
The timing of the thrust faulting has long been inferred to be Taconic, based on Silurian angular unconformities above deformed Ordovician units and lack of similar thrusts and folded units along strike in the Silurian–Devonian units (e.g., Kay, 1937; Vollmer and Bosworth, 1984). However, establishing a precise age for the thrust faulting and associated mélange generation is a difficult proposition. The age of mélange formation has been constrained by graptolite age determinations from blocks within the broken formation units and from the matrix surrounding those blocks (Figs. 3 and 7; for reviews, see Vollmer and Bosworth, 1984; Bosworth, 1989; Plesch, 1994; Kidd et al., 1995). We reviewed and revised the biostratigraphic age constraints on the timing of deformation in Appendix D, and we present a summary of that effort in Figure 7. These ages, if the graptolite identifications are correct, determine a maximum age of thrusting, as do the ages of the transported units in the thrust sheets. However, these ages do not alone constrain the minimum age, and in fact Bosworth (1989) wondered whether some of the phacoidal cleavage zones record Acadian or younger deformation. Such zones could have extensive histories, and although their final motions must be as young as, or younger than, the youngest included blocks, they might have begun considerably earlier than that.
Parautochthonous Cohoes mélange zone. Series of tectonic mélange zones occur between the Taconic allochthon and ~10 km east of the Saratoga-McGregor fault; together, these zones are called the Cohoes mélange (Figs. 2 and 7; Appendix D; Bosworth and Vollmer, 1981; Vollmer and Bosworth, 1984; Bradley and Kusky, 1986; Bosworth, 1989; Plesch, 1994; Kidd et al., 1995; Landing et al., 2003). These mélanges involve phacoidally cleaved, synorogenic flysch including turbidites and mudrocks (the Austin Glen Graywacke). Exotic clasts within the mélanges include rafts of interbedded chert and shale, the Stark’s Knob pillow lavas, and limestone blocks, as well as soft-sediment-deformed graywacke clasts (e.g., Plesch, 1994; Landing et al., 2003). Our best estimate for the depositional age of the deformed sediments within the Cohoes mélange is that they were deposited synchronously with the deposition of the Utica Shale because the mélange belt includes many blocks that contain C. americanus to Orthograptus ruedemanni zone faunas, as does the enclosing matrix (Fig. 7; see also Appendix D for a detailed review and Fig. A3). Zones of deformed flysch with relatively narrow thrust fault systems that display phacoidal cleavage locally separate the mélange zones (Figs. 2 and 7); the deformed zones grade into broken formation (Bosworth and Vollmer, 1981; Vollmer and Bosworth, 1984; Bradley and Kusky, 1986; Bosworth, 1989; Plesch, 1994; Kidd et al., 1995; Landing et al., 2003). These mélange zones have been interpreted as parautochthonous thrust zones formed during the emplacement of the Taconic allochthon (e.g., Zen, 1961; Bird and Dewey, 1970; Rowley and Kidd, 1981; Vollmer and Bosworth, 1984; Plesch, 1994; Kidd et al., 1995).
Westernmost thrust zone and its timing. The westernmost system of Taconic thrusts is located ~5 km west of the Cohoes mélange zone along the Mohawk River near Vischer Ferry (Fig. 2) and involves flysch of the Schenectady Formation with Diplacanthograptus spiniferus Zone graptolites. The thrusts are slightly west of the original western boundary of the Vischer Ferry zone and Ruedemann’s Line (Figs. 2, 8, and 9; Plesch, 1994; Kidd et al., 1995). The individual fault zones are characterized by phacoidally cleaved, interbedded graywacke sandstone/siltstones and mudstones. The zones strike north to northeast and dip moderately to the east (Fig. 8). Asymmetric minor folds in the fault zones strike uniformly north, plunge nearly horizontally, and indicate top-to-the-west transport (Fig. 8). Floating hinge lines of phacoidal cleavage and sediment layering are common (Fig. 9). The exception to the shallowly plunging folds are soft-sediment-deformed (“wild-folded”) graywacke clasts restricted to the mélange (Fig. 9) that have a variety of plunges, some relatively steep. The strike divergence between the asymmetric folds and the faults suggests a small component of left-lateral oblique motion on the NNE-striking thrusts. Recumbent folds are common, and overturned, gently dipping bedding extends eastward from the western fault zones, forming the intermediate limb of a nappe (Fig. 8).
The maximum age of thrusting at this site is unambiguously constrained to be younger than the D. spiniferus Zone sediments that are here deformed; however, the minimum age is more difficult to determine precisely. We propose that this thrusting most likely occurred relatively soon (geologically speaking) after deposition. The soft-sediment-deformed graywacke clasts that are restricted to the thrust zones and the phacoidal cleavage are both consistent with thrusting while the sediments were water-rich, and with rapid dewatering (although for an alternate view for phacoidal cleavage, see Vollmer, 1981; Bosworth, 1989; Kidd et al., 1995). Dewatering during burial leads to increased stiffness and a shift from ductile to brittle behavior over a relatively broad range of depths that vary with sediment composition and strain rate. We estimate from compaction curves and observed brittle behavior in core (e.g., Revil et al., 2002; Bolås et al., 2004; Bjorlykke et al., 2009) that the observed soft-sediment deformation is most likely to have taken place at burial depths less than about 1 km. Extrapolating the subsidence curve for the Mohawk Valley region provided in Macdonald et al. (2017), 1 km of subsidence suggests an upper age limit of ca. 445 Ma.
This minimum age estimate of ca. 445 Ma is consistent with regional structural analysis. Epstein and Lyttle (1993, 2001) argued that autochthonous and parautochthonous deformation zones exposed in the Mohawk Valley (discussed above and see Appendix D) can be traced along strike into northwestern New Jersey and northeastern Pennsylvania. There, the deformation zones are beveled and overstepped by the Lower Silurian Shawangunk Conglomerate along a major angular unconformity, the Taconic unconformity. Exposures along the Delaware River corridor show that the basal contact of the Shawangunk Conglomerate also truncates the slaty cleavage within the Taconian mélange (Epstein and Lyttle, 2001). The Taconian flysch immediately beneath the Taconic unconformity in this region contains a D. spiniferus Zone graptolite fauna (Berry, 1970) coeval with that at the Vischer Ferry thrust site. The age of the overlying Shawangunk Conglomerate is more difficult to determine precisely. Regional lithostratigraphic correlations suggest that this unit is laterally equivalent to the Tuscarora Formation farther south, where that unit exhibits a gradational contact with the underlying red beds of the Juniata Formation, which are generally considered Late Ordovician, and with the Medina Group in New York (Johnson, 1985; Brett et al., 2006), which also appears to straddle the Ordovician-Silurian boundary (Bergström et al., 2011). The lithostratigraphically equivalent Clinch Mountain Formation in Tennessee contains a suite of K-bentonites (the Thorn Hill complex; Bergström et al., 1998) that are associated with lowermost Silurian conodonts (Distomodus kentuckyensis Zone; Manzo, 2002). From the foregoing, the youngest plausible age for the Vischer Ferry thrust is about ca. 441 Ma.
Local tectonostratigraphic relationships may permit a tighter bracketing of the thrust timing. Hanson (2010) and Hanson et al. (2010) documented the presence of five soft-sediment-deformation zones in the D. spiniferus Zone rocks within core 75NY-2 (column 12, Fig. 5), which is located on the upthrown side of the Saratoga-McGregor fault, ~10 km west of the Vischer Ferry zone (core location in Fig. 2). The deformed zones in the core, given their close proximity in stratigraphic position and geographic location, may represent the effects of the Vischer Ferry thrusts. The age of these deformed zones in 75NY-2 are similar to the age of the D. spiniferus Zone–aged slump folding episodes and sediment slide scar represented by the Thruway unconformity (Jacobi and Mitchell, 2002). If these distal deformational events correspond to compressional events recorded in the Vischer Ferry thrust zone, then they suggest that the principal shortening and mélange formation recorded in that zone may have developed within the interval of the D. spiniferus Zone (451.6–450.9 Ma).
It seems likely that the Snake Hill olistolith (which is dominantly composed of relatively coarse clastic sediments that bear a shelly fauna deposited during the O. ruedemanni Zone; English et al., 2006) came to rest within the D. spiniferus Zone sediments of the Vischer Ferry graywacke zone somewhere around this time as well. Similarly, Riva (1987, p. 931) suggested that stratigraphic relations in the Quebec reentrant indicate that “the final emplacement of the frontal units of the Taconic allochthon took place in late C. spiniferus Zone time or in early C. pygmaeus Zone time at the latest.”1
Although we have established that this thrust zone, and associated folding, are most likely Taconian, the temporal resolution is insufficient to distinguish whether the thrusting occurred before the dominant motion on the normal faults, as might be expected in a simple retroarc foreland basin model where subsidence results from thrust loading. However, if the thrusting occurred soon after deposition in D. spiniferus Zone time, as we suspect it did, then the age does predate final motion on the normal faults, as determined from growth fault geometries on the far western faults such as the Little Falls fault and the Herkimer fault (Fig. 5).
Timing of thrusts in the Taconic allochthon and parautochthon and its implications. The Taconic allochthon thrusts must have initiated after deposition of units in the allochthonous slices—the Indian River Formation (ca. 466–464 Ma and younger; Macdonald et al., 2017) and the Mount Merino Formation (ca. 457 Ma based on C. bicornis Zone graptolites; for details, see Appendix D), but the thrusts perhaps initiated during deposition of the Austin Glen and Pawlet formations, which are most likely of late C. bicornis Zone age (e.g., Rowley and Kidd, 1981; Appendix D). In that case, thrust initiation took place between 455 Ma (base of Pawlet Formation in the Taconic allochthon) and perhaps 453 Ma (near the top of the Pawlet Formation; Fig. 10; e.g., Macdonald et al., 2017; Appendix D and Fig. A3). In contrast, biostratigraphic consideration of the mélange zones immediately west of the Taconic allochthon suggest that the leading edge of the Taconic allochthon, including the frontal thrusts of the Giddings Brook slice, arrived at their present position by late C. bicornis to early C. americanus Zone time (Figs. 2, 7, and 10; ca. 454 Ma to 452.7 Ma; Appendix D and Fig. A3). Farther west, in the parautochthonous belt between the Vischer Ferry thrust zone and the mélanges associated with the Taconic allochthon (Figs. 2 and 7), biostratigraphy indicates a D. spiniferus Zone age (ca. 451 Ma) for deposition of the faulted flysch of the Austin Glen Formation—an age that appears to postdate the proposed thrusting phase of the Taconic allochthon (Fig. 10: Appendix D, Fig. A3). We suggest that the timing of thrusting marked by the discrete mélange zones in the parautochthonous belt was relatively soon after deposition (perhaps as early as ca. 451.6 Ma to a minimum age of >450.9 Ma). Like the Vischer Ferry thrust zone, we base the older age on soft-sediment-deformed graywacke clasts that are restricted to the thrust zones, and the younger age on the observation that rocks above the highest deformed zones in core 75NY-2 are of the Geniculograptus pygmaeus Zone. The phacoidal cleavage, characteristic of the mélange zones, is also consistent with thrusting while the sediments were water-rich and undergoing rapid dewatering.
Several conclusions can be drawn from these thrust ages, if the assumptions and biostratigraphy are correct. (1) The Taconic thrusts probably had at least two major phases of motion, as documented by the western parts of the Taconic allochthon and the younger mélange/thrust zones west of the Taconic allochthon (Fig. 10). (2) Significantly, the thrusting duration of the Taconic allochthon was relatively short: a maximum length ~4 m.y. (455–451 Ma) and perhaps less. The duration of thrusting in the Cohoes mélange zone (parautochthonous belt) is not determinable, but it is assumed to be of similar duration. The thrusting at the far western reaches of the Taconic thrusts (the Vischer Ferry thrust zone) also had a short duration (a maximum from 451.6 to 450.9 Ma). Thus, the interval during which the Taconic allochthon and thrusts to the west were emplaced was extremely limited; these thrust events do not document a long-lived tectonic episode such as would be expected for an accretionary prism resulting from subduction of a large-scale ocean basin.
The timing of the Taconic allochthon thrusts (455 Ma, maximum, to 451 Ma) coincides closely with the timing of subsidence in the Saratoga-McGregor–Little Falls fault basin to the west (Fig. 10). The broad zone of nearly synchronous initial subsidence in the Saratoga-McGregor–Little Falls fault basin is dated at 453 Ma, as recorded by the rapid overstep of the Utica black shale (Figs. 3 and 5). This subsidence occurred about the time the farthest-traveled slices of the Taconic allochthon had advanced half of their total travel distance (if they initiated at 455 Ma and were moving relatively steadily) or about a third of their total travel distance if they initiated at 454 Ma. In the second alternative, the time lag between thrust initiation and subsidence is ~1 m.y., similar to that proposed for the subsidence time lag of the Andean foreland basin at 32°S (Fig. 4) after thrust loading.
The eastern boundary of the ~100-km-wide, Saratoga-McGregor–Little Falls fault basin is located ~10 km west of the fontal thrusts of the Taconic allochthon. Slightly older black shale/flysch basins occur east of the Saratoga-McGregor fault and extend under the present position of the Taconic allochthon (Fig. 3). These basins also may have subsided relatively synchronously across their width, with sharp steps in basal age across the master bounding faults, as can be inferred from cross sections of Taconic allochthon sections (e.g., Rowley and Kidd, 1981, their figure 2). If the basin were on a scale comparable to the Saratoga-McGregor–Little Falls fault basin (~100 km wide), then at least one more basin should exist east of the Saratoga-McGregor fault. The interpretation that two basins probably existed during emplacement of the Taconic allochthon implies that the response time from thrust loading is less than 1 m.y., if the Taconic allochthon thrusted for only 2 m.y. The rapid response to thrust loading may be a signature of previously “broken” continents, such as in the Andean foreland basins (e.g., Ramos and Folguera, 2009), but other factors can offset the effect of already fractured continental crust (e.g., Cardozo and Jordan, 2001; Jordan et al., 2001; Appendix B). The Laurentian margin was “broken” by faulting associated with the Iapetan opening, and, because it had been a passive margin until the time of thrust loading, it had not been “healed” and buttressed by later convergent-margin batholiths and tectonics.
Assuming that the farthest-traveled slices of the Taconic allochthon advanced on the order of 200 km or more (Bradley, 1989; Kidd et al., 1995), then the propagation rate was on the order of 100 km/m.y. (if the thrusts moved for only 2 m.y.). Similarly, if the thrusts in the parautochthonous belt record a total of ~60 km of westward motion (Vollmer and Bosworth, 1984) over ~1 m.y., then the propagation rate for these thrusts was on the order of 6 cm/yr. The 10 cm/yr propagation rate is an order of magnitude higher than the 1.3 cm/yr propagation rate of the retroarc thrust front related to aseismic ridge subduction in the Aconcagua fold-and-thrust belt in the Andes (Ramos and Folguera, 2009). The 10 cm/yr rate is, however, comparable to relative plate convergence rates.
Bronson Hill Alkalic/Rhyodacitic Magmas
Flat-slab subduction elements 5a (continental delamination with alkalic/rhyolitic volcanism that displays continental partial-melt signatures during slab steepening after flat-slab subduction) and 5b (alkalic/rhyolitic volcanism far removed from the trench) may apply to selected magmatism in the Bronson Hill arc(s). The Bronson Hill magmatism consists of the 467–461 Ma Ammonoosuc-Partridge bimodal (but dominantly tholeiitic) volcanism and the Quimby sequence, which includes poorly dated (443 ± 4 Ma and ca. 455 Ma) bimodal volcanics (including a 60-m-thick metatuff in the Quimby Formation), the 456–435 Ma Oliverian Plutonic Suite with granite, granodiorite, trondhjemite, and rhyolite porphyry, and the 454–435 Ma Highlandcroft Suite with bimodal, but generally granitic and granodioritic composition granites and granodiorites (e.g., Moench and Aleinikoff, 2003; Karabinos et al., 2008, 2017; Dorais et al., 2011; Macdonald et al., 2014, 2017).
There is little question that, in a general sense, at least some of the acidic volcanics/magmas of the Highlandcroft and Oliverian plutonic suites have mineralogical and geochemical signatures that indicate a contribution from a continental partial-melt source. For example, over 35 years ago, Rowley and Kidd (1981, p. 214) suggested that acidic volcanics in the “upper part of the Ammonoosuc sequence … may be related to anatectic melting of the basement of the volcanic arc due to shortening and thickening of the crust during collision….” Similarly, based on trace elements, rare earth elements (REEs), and isotopes (87Sr/86Sr and εNd), Hollocher et al. (2002, p. 38) suggested that the Highlandcroft Suite (449 Ma to 440 Ma, his assigned ages) and other late plutons such as the Cortland Complex were generated as a result of “decompression melting and heating of the lower crust” that stemmed from asthenospheric upwelling during slab detachment or delamination during trench rollback of an east-dipping subduction zone. High initial 87Sr/86Sr ratios (0.7045–0.711) and εNd values of –5.8 to +1.0 for felsic Bronson Hill magmas in the Oliverian Plutonic Suite also suggested involvement of an older continental crust (as well as a mantle component) in the generation of these magmas (Samson, 1994; Andersen and Samson, 1995; Samson et al., 1995; Hollocher et al., 2002). Dorais et al. (2008, 2011) suggested that this continental crust was probably Laurentia, based on low 207Pb/204Pb (15.54–15.6) and εNd values (–7.9 to –0.5) that lie between inferred mantle compositions and Laurentian crust. Dorais et al. (2011) proposed that the Laurentian crustal melt component of the Oliverian Plutonic Suite was acquired when a continental arc developed over a west-dipping subducting zone that initiated beneath the Laurentian margin after that margin had obducted the older Ammonoosuc arc (which was in turn built on peri-Gondwanan crust during eastward subduction). Modeling suggests the northern granitic plutons (now gneisses), located in central and northern New Hampshire, were derived from melting of garnet-free, intermediate/felsic crust, whereas the southern tonalitic and granodioritic plutons in the western part of the Bronson Hill arc from Connecticut to central New Hampshire were derived from melting of garnet-free mafic crust, and the high-Sr, low-Y, heavy (H) REE–depleted felsic rocks in the eastern part of the southern Bronson Hill arc were derived from melting of garnet-bearing mafic crust (Hollocher at al., 2002).
All of the examples of crustal melting discussed here could result from either downgoing slab detachment, as previously proposed for the Highlandcroft Plutonic Suite by Hollocher et al. (2002), or from steepening of the subducting slab following flat-slab subduction and passage of a subducting aseismic ridge (Appendix B). The delamination with asthenospheric rise and decompression melting hypotheses are essentially the same basic model in both cases. It appears that the tectonic framework perhaps can be used to evaluate and differentiate these alternatives. For example, the 3–5 m.y. volcanic gap represented by the Partridge Formation (Moench and Aleinikoff, 2003) could represent the magmatic gap that occurs when aseismic ridge subduction flattens the subducting plate. In that case, the Quimby sequence volcanics could represent the effects of post-flat-slab subduction that included asthenospheric upwelling and decompression melting (see Appendix A for further discussion concerning the origin of the Ammonoosuc and Partridge volcanics).
Relation of the Highlandcroft and Oliverian Plutonic Suites to the Ash Layers in the Utica Shale and Trenton Group, and Ash Layer Origin
Some of the Highlandcroft Plutonic Suite felsic plutons, such as the 452 ± 4 Ma Adamstown pluton (Lyons et al., 1986; Moench and Aleinikoff, 2003), have ages very similar to the ash layers (K-bentonites) in the Utica Shale, such as 452.6 and 451.9 Ma (Fig. 4; e.g., Sell et al., 2013, 2015; Macdonald et al., 2017). This similarity in ages has led to the proposal that felsic plutons in the Quimby sequence in New Hampshire, or similar plutons in Connecticut, could be a source for (some of) the ash layers in the Utica Shale (Moench and Aleinikoff, 2003; Jacobi et al., 2016; Macdonald et al., 2017). The Deicke and Millbrig ash layers (e.g., Mitchell et al., 2004) are slightly older than the Utica ashes, based on stratigraphy (they occur in the Trenton Group) and radiometric dates of 453.7 ± 0.2 Ma and 452.9 ± 0.2 Ma, respectively (Sell et al., 2013; or 454.5 ± 0.5 Ma and 453.1 ± 1.3 Ma, respectively, inTucker and McKerrow, 1995). These ages are comparable to the 456–435 Ma Oliverian Plutonic Suite, although the Deicke-Millbrig ash layers appear to have a more southeasterly source (for a review, see Samson et al., 1989; Kolata et al., 1996).
The geochemistry of phenocrysts, glass inclusions, and xenocrysts in the K-bentonites in both the Trenton Group and the Utica Shale is also consistent with the Oliverian Plutonic Suite and the Highlandcroft Plutonic Suite as sources (e.g., Jacobi et al., 2016). The Deicke-Millbrig ashes and the volcanic ashes in the Utica Shale both have elevated 87Sr/86Sr ratios (calculated for 450 Ma) that range from 0.710 to 0.712 for the Deicke-Millbrig ashes and range from 0.706 in C. americanus Zone Utica Shale ashes to 0.709 in C. spiniferus Zone Utica Shale ashes (Samson et al., 1989; Samson, 1996). The Oliverian Plutonic Suite has similar 87Sr/86Sr ratios, ranging from 0.7045 to 0.711 (Hollocher et al., 2002) and from 0.706 to 0.715 (Dorais et al., 2008). Based on 87Sr/86Sr, εNd values, and xenocrysts, the source of the Deicke-Millbrig ash layers was neither a typical island-arc tholeiite (IAT) nor a mid-ocean-ridge basalt (MORB) setting; rather, the source was an anatectic melt of an evolved continental crust, or a mantle source followed by significant interaction with an evolved continental crust (Delano et al., 1990; Samson et al., 1989; Samson, 1996). Further, εNd values indicate an Adirondack Grenvillian-like source, and the cooling history of hornblende phenocrysts (determined from 40Ar/39Ar plateaus) is also very similar to that of the Adirondack Grenville Province (Samson, 1996), suggesting a Laurentian continental source. Zircon grains with inherited Grenville ages in Utica ash samples suggest that these ashes also erupted through Laurentian crust (Macdonald et al., 2017), consistent with garnet inherited from a Precambrian terrane (Delano et al., 1990). The presence of components apparently derived from a Laurentian source within the Utica ashes is consistent with the suggestion that the Oliverian Plutonic Suite was derived at least in part from partial melting of Laurentian continental crust (e.g., Dorais et al., 2011).
The geochemistry of biotite phenocrysts in the Deicke and Millbrig ash beds also supports a continental arc source (Haynes et al., 2011). Haynes et al. (2011) pointed out that the geochemistry of the Deicke biotites closely resembles that of the La Pacana ignimbrites and the Cerro Chascun (sic, usually spelled Chascon) rhyolites (among others). The ashes resulted from explosive eruptions of hydrous metaluminous to peraluminous magmas from a large vent with post-eruption caldera formation (Haynes et al., 2011). The volumes of the Deicke and Millbrig ash beds are each a minimum of 330 km3 and may have been three times that much (Huff et al., 1996; Samson et al., 1989). These volumes compare well to the 5–3 Ma voluminous ignimbrites in the northern Puna volcanic field of the La Pacana caldera in the Andean inner arc (see Appendix B), such as the >500 km3 Pujsa ignimbrite, the >100 km3 Toconao ignimbrite, and the >1600 km3 Atana ignimbrite (e.g., Kay and Coira, 2009). The La Pacana caldera lies above an anomalously shallow low-velocity zone that is interpreted to indicate decompression melting of the mantle wedge in a region of asthenospheric upwelling and lithosphere delamination that resulted in crustal partial melting and the production of the ignimbrites (e.g., Kay and Coira, 2009; Appendix B). The asthenospheric upwelling and decompression melting occurred during steepening of the flat slab after passage of the Juan Fernandez aseismic ridge (Appendix B). Interestingly, the second stage of sub-Andean retroarc thrusting commenced ca. 4.5 Ma, overlapping the time (4.2–3.8 Ma) of voluminous ignimbrites, including those at the La Pacana caldera.
The geochemistry of biotite phenocrysts, the 87Sr/86Sr and εNd values, and the cooling history of hornblende phenocrysts that have a Grenvillian component, and the volume of the ashes are all consistent with partial melting of the Laurentian crust during slab steepening after flat-slab subduction of an aseismic ridge. We suggest a similar origin for the overlying ashes in the Utica Group, since they too have elevated 87Sr/86Sr ratios. An inferred southeasterly source for the Deicke-Millbrig ashes (Kolata et al., 1998) suggests that they were not derived from the Oliverian Plutonic Suite, but from some other comparable suite, yet unidentified, farther south in the Appalachians (e.g., Samson, 1996). Such a source does not, however, preclude an Oliverian source for some of the other ashes in the Deicke-Millbrig suite of ashes. The possibility of multiple sources related to flat-slab subduction suggests that multiple episodes of aseismic ridge subduction occurred along the western margin of the Taconic seas, comparable to the situation along the present Andean margin.
TACONIC TECTONIC MODELS WITH INCORPORATED FLAT-SLAB SUBDUCTION ELEMENTS
West-Following-East Subduction Models
A subduction polarity flip from eastward subduction to westward subduction is proposed to have occurred after generation of the Shelburne Falls arc and before the bulk of the Quimby sequence (Fig. 11; e.g., Karabinos et al., 1998, 2017; Moench and Aleinikoff, 2003; Dorais et al., 2011; Macdonald et al., 2014, 2017). In the Karabinos et al. (2017) model, the Moretown terrane and Shelburne Falls arc collided with Laurentian crustal fragments ca. 475 Ma, based on such considerations as 471–460 Ma 40Ar/39Ar cooling dates in the Laurentian-affinity Rowe Formation (e.g., Laird et al., 1984; Tremblay et al., 2000; Castonguay et al., 2012) and detrital zircon ages and provenance. The possibly protracted collision set up the dynamics for the subduction polarity reversal. The 466.0 ± 0.1 Ma Barnard Volcanic Member of the Missisquoi Formation (a correlative of the Hawley Formation, part of the Shelburne Falls arc) and roughly coeval volcanic ashes in the Indian River Formation in the Taconic allochthon (466.2 ± 0.1 Ma and 464.2 ± 0.1 Ma; Macdonald et al., 2017) are thought to mark either slab breakoff following the collision (Fig. 11B) or subduction zone reversal (Karabinos et al., 2017; Macdonald et al., 2017). In the Karabinos et al. (2017) model, the Ammonoosuc Volcanics straddle the time of slab breakoff and subduction polarity reversal. In the Moench and Aleinikoff (2003) and Dorais et al. (2011) models, the polarity flip occurs after the generation of the Ammonoosuc Volcanics. Moench and Aleinikoff (2003) suggested that the subduction reversal occurred during a magmatic gap of some 3–5 m.y. in part of the Partridge Formation (that overlies the Ammonoosuc Volcanics). All these models propose that westward subduction resulted in Bronson Hill arc volcanics of the Quimby Formation (456–435 Ma), the younger Oliverian and Highlandcroft Plutonic Suites.
We propose that the Oliverian Plutonic Suite (and the Highlandcroft Plutonic Suite) included volcanism that was related to effects of slab steepening after flat-slab subduction of an aseismic ridge (Figs. 11A and 11B). In this model, the 452–450 Ma Utica volcanic ashes (Macdonald et al. 2017) represent ash falls from the explosive events in the Bronson Hill arc, probably from the Oliverian Plutonic Suite (and/or similar units to the south).
Another consequence of flat-slab subduction (and relatively high convergence rates) is both thick-skinned and thin-skinned retroarc thrusting that in turn can cause retroarc foreland basin subsidence from loading by the thrusts (Figs. 11A and 11B; Appendix B). The thrusting results from an interplay between relatively high convergence rates and a lithosphere weakened by delamination and asthenospheric upwelling. Late motion on the thrusts that bound the western extent of the Green Mountain massif in southern Vermont (Karabinos, 1988) and the Berkshire massif in Massachusetts (Ratcliffe and Harwood, 1975; Ratcliffe, 1979), as well as the deeper reaches of the Champlain thrust (eg., Stanley and Ratcliffe, 1985), may represent such thick-skinned thrusting. Mélange zone thrusts west of the Taconic allochthon, as well as the thrust slices within the Taconic allochthon, could correspond to the expected thin-skinned thrusts. Although 470–460 Ma 40Ar/39Ar cooling ages in western Vermont and Massachusetts (Sutter et al., 1985; Tucker and Robinson, 1990) have been used to suggest uplift at this time (e.g., Macdonald et al., 2014), additional motion on the thick-skinned thrusts could have continued into the interval of final Taconic allochthon and parautochthon thrust motion discussed above.
The overlap in timing of sub-Andean retroarc thrusting and explosive volcanism (Appendix B; e.g., Kay and Coira, 2009) appears to fit well with the broadly synchronous development of thrusts west of the Taconic allochthon, Utica basin subsidence, tephra beds, and the early Oliverian Plutonic Suite. As discussed in section entitled Parautochthonous Cohoes Mélange zone,” the initial age of motion on the westernmost, and perhaps youngest, thrust in the Utica basin in the Mohawk Valley may have been syndepositional, i.e., on the order of 451 Ma. The maximum age of older thrust systems represented by wider tectonic mélange zones west of the Taconic allochthon may have a range of ages, based on matrix and included clasts ages (Fig. 7). The age of initial tectonic mélange formation west of the Taconic allochthon may date from ca. 453–451 Ma. As Bosworth (1989) noted, the minimum, perhaps reactivated, age of thrust motion is not well constrained, and it is less well constrained for the eastern mélange zones, in particular. The narrow age range of initial thrusting (453–451 Ma) is consistent with the rapid spread of black shale across much of the Utica basin west of the Saratoga-McGregor fault. The relatively narrow Utica basin, and the proposed rapid response of subsidence to thrust loading are both consistent with a preexisting broken continental crust of anomalously lower strength, as proposed for the Laurentian margin with preexisting Iapetan-opening faults (some of which sustained Late Cambrian–Middle Ordovician motion). Those faults were reactivated during basin subsidence as the “Taconic” NNE-striking normal faults mapped west of the Taconic thrusts in the Mohawk Valley region. The episodic pulses of shortening events represented by, for example, mélange formation followed by the late-stage thrusts, and the episodes of rapid spread of black shale deposition might signal a non-steady-state contractional framework, similar to that found in the Canadian Rocky Mountain fold-and-thrust belt (e.g., Pană and van der Pluijm, 2015) and in the Andean thrusts where reactivation is common (e.g., Kay and Coira, 2009).
In these west-following-east subduction models, the origin(s) of the 469–458 Ma Ammonoosuc Volcanics remains equivocal. Slab detachment, aseismic ridge subduction, and subduction polarity reversal models all can involve asthenospheric upwelling that could result in the bimodal nature of the Ammonoosuc Volcanics (see Appendix A for further discussion).
Continuous Eastward Subduction Models
In the continuous eastward subduction model of Hollocher et al. (2002), the Shelburne Falls arc is followed by back-arc extension that is recorded by the Ammonoosuc Volcanics. Slab detachment results in the Highlandcroft Plutonic Suite. The simplicity of this model (and those versions that followed; e.g., Valley et al., 2015) is attractive, but the new provenance studies (Macdonald et al., 2014, 2017; Karabinos et al., 2017) make the simple eastern subduction model less attractive than the more recent models. For this model, we would again propose an aseismic ridge/flat-slab subduction model for generation of at least some of the Oliverian and Highlandcroft Plutonic Suites. An attractive component of the continuous eastward model is that the boninites in the Shelburne Falls arc could signify that a second arc—the Ammonoosuc arc phase of the Bronson Hill arc—was developing behind the boninitic arc, as is commonly the case for a boninitic terrane (e.g., Kim and Jacobi, 1996). In this model, parts of the Ammonoosuc Bronson Hill arc also may record an aseismic ridge subduction event not related to the younger Oliverian Plutonic Suite event (see “Discussion”). The Quimby bimodal volcanics, which follow a 3–5 m.y. “magmatic hiatus” in the Partridge Formation between ca. 461 and 455 Ma (Moench and Aleinikoff, 2003), could also represent a linkage related to aseismic ridge subduction.
In the continuous eastward subduction model, the post-flat-slab magmatism of the Oliverian Plutonic Suite is not linked to the development of the Utica basin. Rather, the Utica basin subsidence records the approach of the Laurentian margin to the combined arcs. These arcs could then vent volcanic ash into the Utica basin, as a record of aseismic ridge subduction associated with the Oliverian Plutonic Suite. A problem with the continuous eastward subduction model is that the Taconic thrusts we have discussed do not record a long-established contractional event, such as what one would expect for an accretionary prism constructed during continuous eastward subduction. However, the short history of Taconic thrusts could be compatible with a limited eastward subduction event (such as the destruction of the Taconic seaway).
It has been proposed that almost the entire Andean margin has undergone aseismic ridge subduction at some time in its subduction history (e.g., Ramos and Folguera, 2009). Our review of aseismic ridge subduction (see Appendix B) allows us to establish a series of six elements that can used to recognize aseismic ridge subduction. Although individual features may not be distinctive of this process, together they do provide an effective means by which to identify aseismic ridge subduction episodes. Nonetheless, aseismic events have not been recognized in the southern New England Taconic Appalachians in the past. Recognition of some of these five elements can be difficult. For example, element 1a (uplift of the accretionary prism with consequent unconformities) is a transient feature, since the uplift decreases as the aseismic ridge passes from the region. Nevertheless, local unconformities should remain, and could be recognized, although it may be difficult to distinguish this cause of the local unconformity from other potential causes.
Element 1b (significantly more deformation in the accretionary prism where the aseismic ridge intersects the prism, with a possible trail of ocean island basalt fragments ripped from the aseismic ridge and incorporated into the prism) shares a similar fate as element 1a. Since exposures of tectonic mélange associated with the Taconic accretionary prisms are relatively rare, opportunities to map regional differences in the severity of deformation that the sediments underwent may be limited at best. It might be thought that, in the simplest case, if the tectonic mélange zones parallel late-stage faults, piggyback basin fills, and the inferred trench location, then these mélange zones may not be related to aseismic ridge subduction. As documented in Appendix B, small aseismic ridges will not deflect the strike of faults on the walls of the trench. In contrast, if the strike of the tectonic mélange swings obliquely to the regional trench, aseismic ridge subduction may have occurred there, as documented in the Middle America Trench (Appendix B). Finally, if the mélange contains blocks of ocean-island basalt (as originally proposed for the Dunnage mélange in Newfoundland, Appendix A) or other units with enriched MORB (E-MORB) affinities (e.g., Galapagos; Appendix B), then the mélange probably records the passage of an aseismic ridge.
Unlike the first two elements of aseismic ridge subduction, the remaining four elements, arranged in a set of temporal and spatial linkages, can provide powerful tools for recognition of aseismic ridge subduction. This sequence involves a temporal and spatial gap in the occurrence of magmatic activity (“magmatic quiescence”) associated with uplift (recognized by a local unconformity) that is followed by volcanism (“volcanic flareups”). This volcanism may include voluminous explosive felsic volcanism with partial-crustal-melt signatures that erupted far removed from the trench, plus bimodal volcanism, followed later by arc volcanism closer to the trench. If the relative convergent rate is sufficient to exert horizontal stress across the delaminated, weakened crust (in the Andes about >4 cm/yr; Appendix B), then retroarc thrusting and consequent thrust-loaded basin development will occur in the retroarc foreland during roughly the same interval as the explosive volcanism.
The critical points in this identification stratagem—that a volcanic gap is followed by explosive volcanism with anatectic signatures, and that volcanism occurred coeval with the thrusting and basin development—are becoming possible to recognize only because of extensive, high-resolution dating programs, as well as geochemical research (e.g., Tucker and Robinson, 1990; Aleinikoff et al., 2007; Sell et al., 2013; van Staal et al., 2016; Karabinos et al., 2104, 2017; Macdonald et al., 2017). In the past, the large uncertainties in radiometric dates for the Ordovician combined with the paucity of reliable dates meant that relatively short periods of local magmatic gaps and flareups, especially if shoshonites, adakites, rhyolites, and dacites were involved, were difficult to recognize. It was not possible to definitely recognize a 3–5 m.y. magmatic gap, as compared to all the other gaps, or even recognize the coeval nature of the Oliverian Plutonic Suite and the volcanic ash layers preserved in the Utica basin.
Using this set of linkages, we propose that at least part of the Oliverian Plutonic Suite represents the explosive volcanism related to slab steepening after flat-slab subduction that was caused by aseismic ridge subduction. The probable attendant delamination weakened the crust, promoting contractional thrust systems to develop west of the arc. This retroarc thrusting included both thick-skinned thrusting of the crust (represented perhaps by late thrusting along the western boundary of the Green Mountain and Berkshire massifs, as well as the deeper reaches of the Champlain thrust) and the thin-skinned thrusts of the Taconic allochthon and parautochthon between the Green Mountains and the normal faults of the Utica basin west of Albany, New York. The thrusts loaded the retroarc foreland, and the foreland responded almost instantaneously, geologically, partly because the Laurentian margin was already weakened by Iapetan-opening faults (and later reactivations of those faults). The narrowness of the basin, and the relatively short-lived nature of the basin (a few million years, as opposed to the sort of geologically persistent convergence present around the Pacific Rim) are compatible with the proposal that the basin was ultimately the result of aseismic ridge subduction.
Can we determine the spatial extent of the aseismic ridge that was subducted? If we assume that the orthogonal component of the relative convergence rate was >4 cm/yr, then we can estimate a crude dimension of the aseismic ridge, based on that rate being the lower limit for thrust effects in the Andes (Maloney et al., 2013), and the observation that the Utica basin was loaded for ~3–5 m.y. Taking an average of 4 m.y. for basin loading, and the convergence rate, the component of the aseismic ridge dimension that was orthogonal to the trench far east of the Mohawk Valley was at least ~160 km. This crude calculation results in a plausible areal extent for the subducted aseismic ridge. Accordingly, the basic assumptions of the aseismic ridge subduction model also are reasonable.
In the Andes, the along-strike basins are developed as the aseismic ridge plows past the margin on a very oblique trajectory. This oblique trajectory has two implications relevant to the present discussion. One is that the retroarc foreland basin that develops as a result of aseismic ridge subduction should be relatively restricted along strike if the relative subduction vector is orthogonal to the trench, and if the long dimension of the aseismic ridge is parallel to the convergence vector. However, as is the case in the central Andes, convergence commonly is oblique (Appendix B), and commonly the aseismic ridge was built with a long dimension not exactly parallel to the relative convergence vector. Interplay between these two factors can lead to a wider along-strike retroarc foreland basin as the aseismic ridge sweeps along the margin. The resulting retroarc basin(s) will necessarily be diachronous along strike. In the Andean margin, the retroarc foreland basins are not continuous, and they appear to reflect local influence of preexisting structure on the general basin form (e.g., Jordan et al., 2001). Alternatively, a significant aseismic ridge may locally pin the subduction zone, so that during later trench rollback, the trench-seamount chain junction will not migrate along the trench strike (e.g., the Emperor Seamount Chain in the Pacific; Vogt, 1973).
In the case of the Utica basin, our mapping of the farthest west thrust west of Albany, New York (Figs. 2, 8, and 9), suggests that the relative convergence vector was oblique, with a left-lateral component, based on the trend of the encapsulating thrust boundaries versus the asymmetric fold axes in the thrust zone. This oblique motion is consistent with transport directions of thrusts and mélange zones farther east (e.g., Vollmer and Bosworth, 1984), but it is not consistent with proposed right-lateral motion farther to the west-southwest based on three-dimensional seismic surveys (e.g., Jacobi, 2011, 2012). The divergence in sense of motion could be related to escape tectonics around the New York Promontory (e.g., Jacobi, 2011, 2012). In any case, an oblique component of motion characterized the Taconic convergence, which would lead to a retroarc foreland basin that was laterally extensive and time-transgressive along strike. This is consistent with what we know generally of the Taconic foreland basin, which in its broadest sense extends from Alabama to western Newfoundland, but which appears to have consisted of several domains with distinct subsidence histories (compare, for instance, reconstructions of Taconic foreland history in western Newfoundland [Waldron et al., 1993] and Pennsylvania [Ganis and Wise, 2008] with that given herein). These details, however, are beyond the scope of the present paper.
Given the number of seamounts, seamount chains, seamount provinces, and oceanic plateaus in the present oceans, and the probability that almost the entire Andean margin has undergone aseismic ridge subduction at one time or another, it is probable that more than one aseismic subduction event occurred during the course of the Taconic orogeny. We suggest other sequences should be examined in light of the aseismic ridge model, including the bimodal Quimby volcanism, which follows the 3–5 m.y. magmatic gap of the Partridge Formation, and the bimodal Ammonoosuc Volcanics.
The subduction of aseismic ridges can lead to a segmentation of the arc, both in temporal terms and in structural terms, such as in the Middle America Trench region (Appendix B). Thus, an orogen may exhibit significant cross-strike structures related to multiple seamounts passing below the upper plate, and the resulting magmatic processes may be strongly diachronous along strike. Such a situation may explain the apparent difference in age of the Ammonoosuc Volcanics between Massachusetts and New Hampshire. The oblique passage of a single aseismic ridge beneath the margin could also contribute to along-strike diachroneity (and to strike-slip motion in the foreland).
Aseismic ridge subduction is a common occurrence. Almost the entire Andean margin has undergone aseismic ridge subduction at some time in its subduction history. Using the Andean margin and the Middle America Trench as models, we developed a series of six elements that can be used to recognize aseismic ridge subduction in an orogen. Four of these elements, arranged in a set of temporal and spatial linkages, are particularly distinctive. The four-element sequence involves the following: (1) A spatiotemporal magmatic gap (“magmatic quiescence”) is associated with uplift. Flat-slab subduction leads to a shutdown of magmatism and localized uplift as the aseismic ridge subducts. (2) Volcanism following the magmatic gap (“volcanic flareup”). Steepening of the flat slab after the aseismic ridge passes initiates asthenospheric upwelling and decompression melting, which can produce partial melting of the overlying lithosphere and a volcanic flareup. The volcanism can be bimodal and commonly includes explosive, voluminous rhyodacitic volcanism that exhibits a partial-crustal-melt signature and that is erupted at great distances from the trench (up to 850 km). Broadly temporally linked with the explosive volcanism are (3) retroarc thrusting, and (4) consequent thrust-loaded basin development that occurs in the retroarc foreland. Basinal sediments will contain ash layers from the explosive volcanism that have partial-melt signatures. Retroarc foreland thrusting can include both thick-skinned and thin-skinned thrusts. Thrusting and basin subsidence will develop if the relative convergent rate is sufficient to exert horizontal stress across the delaminated, weakened crust (in the Andes ~4 cm/yr or greater). This critical sequence of aseismic ridge subduction effects may be recognizable in ancient orogens now that research in those orogens is beginning to benefit from extensive, high-resolution dating programs coupled with geochemical and isotopic analyses, along with the development of a similarly high-resolution stratigraphy in the foreland basins.
Using these criteria to examine the Taconic orogen in western New England, we propose that at least the older units in the 456–435 Ma Oliverian Plutonic Suite, which consists of mildly alkalic granites and rhyolite, were generated during steepening of the downgoing slab after passage of a subducting aseismic ridge. The continental partial-melt signatures of the Oliverian Plutonic Suite are consistent with this interpretation. In the model with passage of the subducting aseismic ridge, the probable attendant delamination weakened the crust, promoting contractional thrust systems to develop west of the arc, in the retroarc foreland. This retroarc thrusting included both thick-skinned thrusting of the crust (perhaps represented by late thrusting along the western boundary of the Green Mountain and Berkshire massifs, as well as the deeper reaches of the Champlain thrust), and the thin-skinned thrusting associated with the Taconic allochthon and parautochthonous mélange zones in the Taconic system between the Green Mountains and the normal faults of the Utica basin west of Albany, New York. The thrusts loaded the retroarc foreland, and the foreland responded almost instantaneously, geologically, partly because the Laurentian margin was already weakened by Iapetan-opening faults and reactivations of those faults. The narrowness of the basin, and the short-lived nature of both the basin and the dominant motion on the thrusts of the Taconic allochthon and parautochthon (a few million years), as opposed to a typical Pacific convergence history, are compatible with the proposal that the basin was ultimately the result of aseismic ridge subduction. Volcanic ash layers in the Utica basin document continental partial-melt volcanics similar in age (453–450 Ma) and composition to the Oliverian Plutonic Suite. Oblique subduction of the aseismic ridge will extend the foreland basin along strike, but an along-strike transgressive nature should be observed in detail. It is possible that the characteristics of the short-lived thrusts, as well as the yoked short-lived basin subsidence, are also consistent with eastward closure of a narrow ocean basin or extended crust with stick-slip contraction during final collision.
Because of the ubiquity of seamounts, seamount chains, and oceanic plateaus, we expect there were other instances of aseismic ridge subduction during the Taconic orogeny. We suggest other sequences should be examined in light of the aseismic ridge model, such as the Partridge Formation 3–5 m.y. magmatic gap followed by the bimodal Quimby volcanism, and some of the bimodal Ammonoosuc Volcanics, especially the younger 453 Ma felsic Ammonoosuc volcanics. Different timings along the length of the subduction complex should be expected for the effects of aseismic ridge subduction, based on different times of impingement for different aseismic ridges, or, the oblique passage of a single aseismic ridge beneath the margin, depending upon the obliquity of subduction of the incoming aseismic ridge.
We recognize that not all bimodal volcanic suites are related to aseismic ridge subduction, and, in fact, the same asthenospheric rise and decompression melting proposed for slab steepening has been proposed for other convergent settings, including subduction-zone polarity reversals and slab detachment. Nevertheless, we suggest that the sequence of effects of aseismic ridge subduction should be in the arsenal for tectonic interpretations of coeval magmatic, contractional, and basinal elements, and that in the Taconic region of western New England, at least one such example exists: the coeval Oliverian Plutonic suite, westerly thrusting between the Green Mountain massifs and the normal faults of the Utica basin, and the Utica basin subsidence with anatectic-melt ash layers.
We thank Marjorie Gale, Daniel Goldman, Jim Hibbard, Jon Kim, Paul Karabinos, Steve Leslie, Francis Macdonald, John Martin, Rich Nyahay, Nick Ratcliffe, Scott Samson, Rich Schweickert, Bruce Selleck, Cees van Staal, and the Albany crew from the 1970s (Kidd, Dewey, Rowley, Delano, and others) for the years we have spent discussing Northern Appalachian geology—for some (Jim Hibbard for example), it has been over 40 years. Jacobi acknowledges Marshall Kay and Bill Dickinson, who were truly inspirational, insightful, and foresighted scientists; they were a pleasure to be around. Bill Kidd guided us over a period of several days to many of the Cohoes mélange sites discussed in the text, and we are indebted to him for this valuable assistance. The research presented here has benefited from the efforts of several graduate students who conducted projects in these strata under our supervision (including Paul Agle, Stephanie Amodeo, Gareth Cross, Richard Frieman, Stacey Hanson, Anna Hrywnak, Kyle Jones, Todd Marsh, Alex O’Hara, Erin Richley, Melissa Roloson, Steve Saboda, and Tayler Schweigel). We thank Francis Macdonald and Bob Hatcher for careful, perceptive reviews that considerably improved the manuscript, and Tim Lawton, the volume editor, for his constructive suggestions. This research was supported by various grants to Jacobi from the New York State Energy Research and Development Authority and to Mitchell from the National Science Foundation (in collaboration with John Delano, Scott Samson, Steve Leslie, and Pete Sadler).
APPENDIX A: PLATE-TECTONIC MODELS
Taconic Plate-Tectonic Models for New England
Early tectonic models of the Taconic orogeny involved a single eastward-dipping subduction zone (present coordinates) under an arc (Bronson Hill arc; e.g., Jacobi, 1981; Rowley and Kidd, 1981; Stanley and Ratcliffe, 1985; Bradley and Kidd, 1991; Ratcliffe et al., 1998, 1999; Hollocher et al., 2002; Schoonmaker et al., 2016). In these models, the west-directed Taconic thrusts represented the Taconic accretionary prism, and the northerly striking normal faults west of the thrusts in eastern New York State represented the Laurentian continent flexing and stretching as it passed into the trench. The Upper Ordovician Utica Group black shales and the overlying coarser clastics were thought to mark a trench fill that spilled over onto the craton.
More recently, recognition of Ordovician boninitic volcanics in the Hawley Formation, which lies west of the Bronson Hill arc, and the possibility of two ages of Taconic volcanism, led to the concept that two Taconic arcs developed in western New England (e.g., Kim and Jacobi, 1996, 2002; Karabinos et al., 1998; Kim et al., 2003; Moench and Aleinikoff, 2003; Dorais et al., 2011; Macdonald et al., 2014, 2017). In the two-arc model, the older Shelburne Falls arc, including the Hawley Formation, was built over an east-subducting oceanic plate (“Neo-Iapetus” of Karabinos et al., 1998; or Iapetus of Macdonald et al., 2014), whereas the Bronson Hill arc (sensu lato) was constructed over a west-dipping subduction zone; this subduction zone developed along the western margin of Iapetus after the collision of the Shelburne Falls arc with Laurentia destroyed “Neo-Iapetus.” U-Pb zircon dates on the Shelburne Falls arc rocks peak at about 475 Ma, but older rocks with dates between 502 and 486 Ma also occur (e.g., Karabinos et al., 1998; Macdonald et al., 2014, 2017). In these original subduction polarity-flip models, the Bronson Hill arc formed approximately 460–442 Ma (e.g., Karabinos et al., 1998).
Modifications based on new data have considerably complicated the original subduction polarity-flip models. Moench and Aleinikoff (2003) proposed that the Bronson Hill arc is actually a composite of two arcs: an older arc (469–458 Ma) and a younger arc (456–443 Ma). The older arc is represented by the tholeiitic basalts and thin felsic metatuffs of the Ammonoosuc Volcanics, which are overlain by the Partridge Formation, a sequence of predominantly black sulfidic slate and schist with bimodal volcanics including rhyolites and dacites (e.g., Moench and Aleinikoff, 2003; Dorais et al. 2011). The Chickwolnepy Intrusions (for location, see Fig. 1) are thought to be a part of this older arc, and they include sheeted dikes that may have been feeders for the Ammonoosuc Volcanics (Fitz, 2002; Moench and Aleinikoff, 2003). The reported ages of the Chickwolnepy Intrusions (467 ± 4 Ma—Aleinikoff and Moench, 1992; 458 ± 6 Ma—Aleinikoff et al., 2015) are broadly compatible with the older arc. The age of the Partridge Formation is determined by C. bicornis Zone graptolites (ca. 457—452.5 Ma; Harwood and Berry, 1967), which Karabinos et al. (2017) noted would belong to Riva’s (1974) interpretation of the N. gracilis Zone (but see Appendix D for a discussion of the overlapping age implications of these usages). Additionally, the radiometric age of the Partridge Formation was established by a U-Pb zircon date of a volcanic unit in Massachusetts of 449 +3/–2 Ma (Tucker and Robinson, 1990) and by detrital zircons, which indicate that part of the Partridge Formation has a maximum age of 452 Ma (Merschat et al., 2016). Recently, an older belt of volcanics was identified in the “Ammonoosuc Volcanics” along the western margin of the Bronson Hill arc in New Hampshire that has dates of 475 ± 9 and 477 ± 7 Ma (Aleinikoff et al., 2015), which overlap those of the Shelburne Falls arc. Similar dates also have been obtained from intrusions into the Ammonoosuc Volcanics in New Hampshire (475 ± 5 Ma and 466 ± 8 Ma; Valley et al., 2015). These newly obtained dates extend the age range of the Ammonoosuc arc. Younger ages of Ammonoosuc Volcanics in a belt immediately east of the older Ammonoosuc Volcanics yielded dates of 457 ± 9 Ma, 452 ± 13 Ma, and 449 ± 7 Ma (Aleinikoff et al., 2015), consistent with the ages of the younger Bronson Hill arc. Correlatives of the Ammonoosuc Volcanics to the north in Quebec include parts of the Ascot complex (Fig. 1A), including 462–460 Ma rhyolites and granites (Tremblay et al., 2000; Moench and Aleinikoff, 2003).
The younger arc in the composite Bronson Hill arc (ca. 456 Ma to ca. 443 Ma, or possibly as young as 435 Ma) is essentially the Bronson Hill arc of Karabinos et al. (1998). This arc is represented by the Quimby Formation (443 Ma and 455 Ma, generally graywackes and slates with bimodal volcanics), the Oliverian Plutonic Suite (mildly alkalic granites and rhyolite, 456–435 Ma), and the Highlandcroft Plutonic Suite (granitic to mafic plutons, 454–436 Ma; e.g., Moench and Aleinikoff, 2003; Karabinos et al., 2017; Macdonald et al., 2014, 2017). Based on Nd and Pb isotopic signatures, Dorais at al. (2011) suggested that the ca. 470 Ma Ammonoosuc portion of the Bronson Hill arc has a Gondwanan-like crustal component, whereas the younger Quimby Formation and Oliverian Plutonic Suite have Laurentian Nd and Pb isotopic signatures. These data implied to Dorais et al. (2011) that the Oliverian melts rose through the eastward-subducted leading edge of Laurentia (which could include accreted microcontinents of Laurentian affinity such as the proposed Rowe belt of Karabinos et al., 2017; Macdonald et al., 2017).
Macdonald et al. (2014, 2017) and Karabinos et al. (2017) found that detrital zircon U-Pb age normalized probability density plots indicated that part of the base into which the Shelburne Falls arc intruded, the Moretown Formation (Fig. 1A, now designated “terrane”), has a Ganderian (Gondwanan) signature (like the Ammonoosuc portion of the Bronson Hill), but surprisingly, the Hawley Formation, which includes the Shelburne Falls arc boninites, has both Gondwanan and Laurentian detrital zircon signatures. A tectonic model to explain the detrital zircon data and extensive new radiometric dates has been proposed by Macdonald et al. (2014, 2017) and Karabinos et al. (2017) (see also review in section entitled “Recent Taconic Plate-Tectonic Models”). In this model, the Moretown terrane was a microcontinent of Gondwanan-derived crust on the east side of Iapetus. The Moretown terrane was intruded by the Hawley boninites above an east-dipping subduction zone when Iapetus had closed sufficiently to allow a Laurentian source to provide detritus to the Hawley Formation. Such a source could be the Rowe Schist, which lay on the west side of Iapetus (and west of the trench and suture) and represents either a Laurentian microcontinent or extended Laurentian crust. After a subduction polarity flip from eastward to westward subduction at 466 Ma, the west-dipping subduction resulted in the Oliverian and Highlandcroft plutonic suites.
Even after the recognition of multiple arcs (Shelburne, older and younger Bronson Hill arcs) and the tectonic models that incorporated a subduction polarity flip to westward subduction for the eastern, youngest arc, some researchers have maintained that eastward subduction could have resulted in all of the described arcs. Hollocher et al. (2002) maintained that the Shelburne Falls arc and both parts of the Bronson Hill arc were generated over an east-dipping subduction zone. They suggested that slab breakoff of an east-dipping oceanic plate resulted in the younger Bronson Hill arc rocks such as the Oliverian Plutonic Suite. Fitz (2002) also suggested that the Chickwolnepy Intrusions developed over an east-dipping subduction zone during slab detachment or rhombochasm growth. The recent dates in the Bronson Hill arc that have a Shelburne Falls age are also compatible with a long-lived, east-dipping subduction zone (Valley et al., 2015), but slab breakoff and subduction polarity reversal could also result in spatially overlapping volcanics in two phases (Karabinos et al., 2017). Van Staal et al. (1998, 2009, 2016) proposed that the 476–453 Ma Popelogan arc in Maine and New Brunswick also developed over an east-dipping subduction zone, one that was affected by significant trench rollback with concomitant back-arc development. The western, youngest belt of the Popelogan arc appears to be the along-strike correlative of the Bronson Hill arc (van Staal et al., 2016; Karabinos et al., 2017; Fig. 1A). Van Staal et al. (2016) proposed that after collision of the Popelogan arc with Laurentia at about 455 Ma, a west-dipping subduction zone (which they called the “Salinic” or “Silurian” subduction zone) initiated at ca. 450 Ma in a former backarc of the Popelogan arc (the Tetagouche basin). This subduction zone dipped westerly beneath the composite Laurentia and accreted terranes, including the Bronson Hill arc.
Tectonic Setting of the Ammonoosuc and Partridge Volcanics
The bimodal nature of the Ammonoosuc and Partridge volcanics and the wide range of tectonic settings that can be inferred from their geochemistry (island-arc tholeiite, MORB, calc-alkaline basalt, back-arc basin basalt) have led to multiple suggested origins. Based on major, trace, and REE geochemistry, the Ammonoosuc Volcanics (Hollocher, 1993; Dorais et al., 2011) and the geochemically similar Partridge Formation volcanics (Hollocher, 1993) were proposed to have formed in a backarc basin over an east-dipping subduction zone. Additionally, Dorais et al. (2011) suggested that the Ammonoosuc Volcanics have a peri-Gondwanan crustal signature from incorporation of Ganderian sediment, based on Nd and Pb isotopes. Moench and Aleinikoff (2003) argued for volcanic generation in a suprasubduction-zone region over an east-dipping subduction zone, but their tectonic model (Moench and Aleinikoff, 2003, their figure 7) shows the Chickwolnepy intrusions (partly feeders for the Ammonoosuc Volcanics; Fitz, 2002; Moench and Aleinikoff, 2003) in a backarc position.
More recent models for the Ammonoosuc Volcanics have centered on two tectonic associations: subduction polarity flip, following the model of Teng (1996) for Taiwan (e.g., Karabinos et al., 2003), or slab breakoff of an east-dipping oceanic slab. Slab detachment would result in asthenospheric upwelling and decompression melting (Fig. 11B), which initially would produce bimodal volcanism similar to the Ammonoosuc Volcanics. In the subduction polarity reversal model, an intra-arc basin, such as the Okinawa Trough north of Taiwan, forms in an extensional regime (e.g., Teng, 1996; Shinjo et al., 1999; Teng et al., 2000; Clift et al., 2003). The Okinawa Trough is characterized by bimodal volcanism (e.g., Shinjo et al., 1999) that has subduction, rising asthenospheric, and partial-melt signatures (Shinjo et al., 1999; Teng et al., 2000; Clift et al., 2003). A third possible origin is that the Ammonoosuc Volcanics are an effect of aseismic ridge subduction when the oceanic slab began to steepen after passage of the subducted ridge. For example, asthenospheric upwelling and decompression melting related to flat-slab subduction (Appendix B; Fig. A1) are proposed to have resulted in the 25–22 Ma Tambo-Tambillo bimodal volcanics-basalts and shoshonites in the Andes (Appendix B; Fig. A2; e.g., Kay and Coira, 2009). The models of Karabinos et al. (2017) and Macdonald et al. (2017) suggest that the Ammonoosuc Volcanics developed during the time of both slab breakoff and subduction polarity reversal.
The geochemistry of the Ammonoosuc Volcanics, such as the enriched large ion lithophile elements (LILEs) and depleted high field strength elements (HFSEs), is comparable to the Okinawa Trough basalt geochemistry, although the 87Sr/86Sr ratios are much higher in some of the Ammonoosuc Volcanics (compare Shinjo et al.  data to those from Hollocher , Hollocher et al. , and Dorais et al. ). The bimodal volcanism in the Okinawa Trough north of Taiwan has both a subduction signature and a rising asthenosphere signature (Shinjo et al., 1999; Teng et al., 2000; Clift et al., 2003). Basalts closer to Taiwan display varying Sr isotope ratios and incompatible trace-element compositions, suggesting local variations in interactions with the extensional trough system, the colliding plates (Shinjo et al., 1999), and sediment flux.
Bimodal volcanism is also a hallmark of aseismic ridge subduction when the downgoing slab begins to steepen after flat-slab subduction (e.g., the 25–22 Ma Tambo-Tambillo bimodal volcanics in the Andean margin; Appendix B; Fig. A2; e.g., Kay and Coira, 2009). Since the slab breakoff and polarity reversal models both involve asthenospheric upwelling and decompression melting, it is difficult to differentiate between these models and an aseismic ridge subduction model, which also involves asthenospheric upwelling.
The Ammonoosuc Volcanics in New Hampshire where Moench and Aleinikoff (2003) worked appear to be older than the volcanics in the central and southern Bronson Hill arc (see reviews by Hollocher et al., 2002; Moench and Aleinikoff, 2003). Moench and Aleinikoff (2003) correlated Hollocher’s (1993) bimodal volcanism in the Partridge Formation with their Quimby sequence. If the Ammonoosuc Volcanics resulted from a subduction zone polarity flip, and if the volcanics are diachronous along strike, then the subduction polarity flip probably migrated along the arc, as proposed for Taiwan (Clift et al., 2003). In contrast, if some of the bimodal volcanism represents bimodal volcanism typical of early post-flat-slab subduction, and if the ages of the volcanics are diachronous along strike, then the strong variability in ages could reflect either oblique subduction of an aseismic ridge (Appendix B; see Kay and Coira, 2009) or subduction of multiple aseismic ridges at different times.
In the recent tectonic model of Karabinos et al. (2017), the Partridge volcanics postdate both the reversal and the slab breakoff, as marked by a volcanic “flareup” in the Barnard volcanics at 466.0 ± 0.14 Ma. We suggest that the 3–5 m.y. “magmatic hiatus” in the Partridge Formation between about 461 and 455 Ma (Moench and Aleinikoff, 2003), followed by the bimodal Quimby volcanism, is the sort of linkage that might indicate aseismic ridge subduction. In the recent models by Karabinos et al. (2017) and Macdonald et al. (2017), the magmatic quiescence in the Partridge Formation is too young to mark the time of subduction polarity flip as originally proposed (Moench and Aleinikoff, 2003).
Previously Proposed Taconic Plate-Tectonic Models That Involve Ridge Subduction in Newfoundland and Maine
Subduction of both seismic and aseismic ridges has been incorporated into Taconic (sensu lato) plate-tectonic models for the Canadian Appalachians and Maine. In Newfoundland (Fig. 1B), Jacobi and Wasowski (1985) and Wasowski and Jacobi (1985) proposed that the Ordovician tholeiitic to alkalic volcanics of the Summerford Group represent a seamount chain/oceanic plateau that had partly subducted northwesterly (present-day coordinates) under a peri-Laurentian arc and transferred into an accretionary prism located on the northwestern margin of Iapetus. They also proposed that the seamount subduction process left a trail of highly deformed units as it passed northwestwardly through the accretionary prism, resulting in the Dunnage Mélange. The seamount chain hypothesis was based on (1) trace-element and limited REE abundances that indicated the volcanics were within-plate basalts (oceanic-island basalt) or possibly E-MORB, (2) associated carbonates and chert that suggested the volcanic edifices were tall enough to pass through the carbonate compensation depth, and (3) the long-lived (Tremadocian through Caradocian) nature of the volcanic complexes (Caradocian is approximately equivalent to Sandbian in modern usage).
Continued field work and further geochemical analyses suggested that the Iapetan suture (Red Indian Line) lies west of the Summerford Group and the Dunnage mélange (e.g., Williams et al., 1988; van Staal, 2007). In this tectonic framework, the Summerford seamounts/oceanic plateau would have accreted to the peri-Gondwanan Victoria arc (and the stratigraphically underlying Penobscot arc), which lay east of the Iapetan suture (Red Indian Line) above an east-dipping subduction zone (van Staal et al., 1998; Kim and Jacobi, 2002; Zagorevski et al., 2007). The Summerford Group may have arrived at the Victoria arc-trench region about the time of Dunnage mélange generation (ca. 469 Ma; van Staal et al., 1998). However, Zagorevski et al. (2007) speculated that arrival of the Summerford seamounts/oceanic plateau might have been earlier and might have precipitated the ca. 480 Ma closing of the Penobscot backarc basin, with obduction of the oceanic crust onto Ganderia, although they could not eliminate subduction of a spreading center as a causal mechanism. Subsequently, during the collapse of Iapetus, thought to be marked by an approximately 455 Ma black shale (e.g., Zagorevski et al., 2008), the Victoria arc collided with, and subducted beneath, the peri-Laurentian Annieopsquotch accretionary tract (which includes the Red Indian Lake Group arc), which lay on the west side of Iapetus (e.g., Zagorevski et al., 2008), and the Summerford seamounts were ultimately transferred to the peri-Laurentian margin (van Staal et al., 1998).
In contrast to the seamount subduction model for the Summerford Group, more recent research suggested the Summerford Group volcanics, the Dunnage Mélange, and the Coaker porphyry intrusions into the Dunnage Mélange are all the result of spreading center (oceanic seismic ridge) subduction (Zagorevski et al., 2012). With additional trace-element and REE analyses, Zagorevski et al. (2012) reinterpreted the within-plate to transitional arc affinities of the Summerford Group volcanics to represent part of an extensional phase of the Victoria arc. Zagorevski et al. (2010, 2012) suggested that the approximately 472 Ma to 467 Ma anomalous rift magmatism in the Victoria arc may reflect ridge subduction beneath the arc. The 469 Ma Coaker porphyry intrusions into the Dunnage Mélange have also been suggested to originate from seismic ridge (spreading center) subduction under an accretionary prism (Kidd et al., 1977; van Staal et al., 1998; Zagorevski et al., 2012). In the present tectonic model, this subduction took place beneath the forearc region of the Victoria arc (Zagorevski et al., 2012). The model of Coaker porphyry resulting from seismic ridge subduction is based on trace-element and REE geochemical analyses, zircon inheritance data from the porphyry, and geochemical, isotopic, and spinel characteristics of xenoliths in the Coaker porphyry. Zagorevski et al. (2012) suggested that the deformation in the Dunnage mélange also resulted from the seismic ridge subduction beneath the arc-trench gap.
In New Brunswick, Maine, and northernmost New Hampshire (Fig. 1A), van Staal et al. (2016) proposed that seismic ridge subduction may have resulted in intra-arc/backarc spreading in the Popelogan arc, a southwestern analog of the Victoria arc. Compilation of existing radiometric dates and new age determinations of volcanic systems exposed in three subparallel outcrop belts through New Brunswick and Maine suggested that the west-facing Popelogan arc and trench migrated northwest (present coordinates) from about 476 Ma to 460 Ma. Van Staal et al. (2016) proposed that this migration was the result of trench retreat/rollback associated with eastward subduction. In each of the outcrop belts, an up-section switch occurs from island-arc tholeiites and calc-alkaline arc-related volcanics to tholeiitic/alkalic volcanics and dikes with within-plate basalt affinities. In the two southeastern, older outcrop belts, the up-section evolution to within-plate basalt is ascribed to the development of backarc basins, backarc spreading within these basins, and associated significant trench rollback/retreat to the northwest. In Maine, for the Munsungan-Winterville Inlier, the northwesternmost outcrop belt with the youngest volcanics (ca. 467 Ma to 453 Ma of the Balmoral phase), van Staal et al. (2016) suggested that the within-plate basalt might record the passage of the seismic ridge beneath the arc. They proposed that eastward subduction of the seismic ridge began in New Brunswick between 459 and 455 Ma, just before Iapetus closure and collision of the Popelogan arc with Laurentia at about 455 Ma. The seismic ridge subduction might have resulted in the ca. 458–453 Ma within-plate alkalic volcanics observed in the youngest and most northwesterly outcrop belt of the Popelogan arc in Maine, the Balmoral phase (Fig. 1A). Van Staal et al. (2016) regarded the present-day regions between the Ordovician volcanic outcrop belts as parts of the intra-arc/backarc rifts with local spreading centers. Van Staal et al. (2016) and Karabinos et al. (2017) proposed that the Popelogan arc in Maine (Balmoral phase for van Staal et al., 2017) is on strike with, and correlates with, the Bronson Hill arc. Consistent with this correlation, Karabinos et al. (2017) suggested that in the Exploits lithotectonic zone of Newfoundland (Fig. 1B), the Popelogan/Victoria arc formed on the trailing (eastern) margin of the Dashwoods block, not the leading (western) margin of Ganderia, as previously portrayed (e.g., Zagorevski et al., 2007, 2012). After the Popelogan-Laurentia collision, westward subduction initiated in the Tetagouche basin (formerly a backarc of the Popelogan arc) at about 450 Ma and underthrust the composite Laurentian margin and accreted/obducted terranes.
In Maine, Schoonmaker and Kidd (2006) proposed that Ordovician E-MORBs and within-plate basalts of the Bean Brook Gabbro (and associated Dry Way volcanics) that intruded into continentally derived sediments and the Hurricane Mountain mélange (Fig. 1A) represented spreading-center ridge subduction under an accretionary prism. They proposed west-dipping subduction based on the present-day location of the Chain Lakes massif microcontinent (Fig. 1B) west of the mélange and intrusions. In contrast, van Staal et al. (2016) suggested that although the geochemistry does indicate ridge subduction, the subduction under the composite Popelogan arc was to the east, since the arc is younger to the west.
APPENDIX B: ELEMENTS OF ASEISMIC RIDGE SUBDUCTION AND FLAT-SLAB SUBDUCTION
We focus here on the effects of shallow-dipping subduction of a buoyant oceanic slab: i.e., flat slab subduction to moderately dipping slab subduction associated with aseismic ridges. These conditions lead to the set of features summarized in the following paragraphs. The elements discussed below are keyed to the list of effects from aseismic ridge subduction in the main body of the paper. These discussion points are based primarily on examples from the Caribbean Antilles, Middle America Trench, and Andean margin.
Element 1A: Uplift of the Accretionary Prism
In the Lesser Antilles island arc, seismic reflection profiles indicate that the accretionary prism is uplifted (compared to elevations along arc-strike) by as much as 3+ km where large aseismic ridges intersect, such as the Barracuda and Tiburon Ridges (Bouysse and Westercamp, 1990; Bangs et al., 2003; Laigle et al., 2013). Additionally, outcrop geology suggests that the Caribbean upper (overriding) plate is tilted and uplifted by as much as 2 km where a large oceanic plateau is proposed to have subducted (e.g., Bouysse and Westercamp, 1990).
In the Middle America Trench, along the continental margin of Costa Rica, the forearc displays variations in uplift along the length of the arc that correspond to subduction of aseismic oceanic plateaus, seamounts, and oceanic “rough” topography (e.g., von Huene et al., 1995; Fisher et al., 1998). The larger seamounts, plateau, and the Cocos Ridge are related to the Galapagos hotspot track on the Cocos plate (e.g., von Huene et al., 2000). Lücke and Arroyo (2015) warned, however, that the timing of Cocos Ridge impingement, and the links between aseismic ridge subduction and observed tectonic effects, such as uplift, may be too simplified with yet-unrecognized causal mechanisms.
Isolated seamounts appear to “tunnel through” (von Huene et al., 2000) the accretionary prism, with localized doming and uplift of the preexisting prism on the order of the seamount relief (1.5–2.5 km). The structural relief diminishes up-section and upslope as the seamounts penetrate the accretionary prism. In contrast, the larger oceanic plateaus appear to influence and build more permanent structures and uplift across the entire width of the accretionary prism. For example, the Quepos Plateau extension affected the offshore and onshore portions of the accretionary prism in Costa Rica (von Huene et al., 2000), and Meschede et al. (1999) suggested that the subduction extension of the Cocos Ridge uplifted the presently onshore Osa mélange from the base of the accretionary prism, which here is between 4 and 8 km deep (von Huene et al., 2000).
The size of the impinging seamount/plateau appears to exert considerable influence on the structural response of the trench and accretionary prism/outer arc. In the Mariana Trench, Fryer and Smoot (1985) found that seamounts larger than about 100 km in diameter do not exhibit the horsts and grabens typical of the outer wall of the trench, and these large seamounts can remain complete in the accretionary prism. Similarly, von Huene et al. (2000) showed that the larger oceanic plateaus, such as the 30-km-wide, 2-km-high Quepos Plateau (which is located along the border of the Cocos Ridge complex), also do not exhibit graben and horst structures. Additionally, the trench in the region of the crest of the ~200-km-wide, 3-km-high Cocos Ridge is only about 1 km deep, compared to 5+ km deep in the region of smooth oceanic crust to the northwest.
The along-strike wavelength of the structural uplifts in the Costa Rica accretionary prism is similar to that of the bathymetry of the impinging “rough” topography and the Cocos Ridge (Fisher et al., 1998; von Huene et al., 2000). This variance in along-strike uplift indicates, along with faults transverse to the strike of the accretionary prism, that the accretionary prism/arc is highly segmented along its length by seamount/ridge/plateau intersections with the arc, with blocks on the order of 10–35 km wide, measured along arc strike (Fisher et al., 1998).
Although the localized uplift can be significant in the accretionary prism, the doming above subducting seamounts and the trail behind, appear to be transient for the relatively small seamounts, with as short a duration as 0.5 m.y. in the Costa Rica accretionary prism (von Huene et al., 2000). Slumping off the structural high, deposition in the trail behind the seamount, and tectonic accommodation relatively quickly bring the slopes back into the regional tapers (von Huene et al., 2000). In fact, Laigle et al. (2013) suggested that local accretionary prism uplift is followed by subsidence as the ridge/seamount/plateau subducts obliquely beneath the accretionary prism in the Lesser Antilles.
These uplifts can result in local unconformities. For example, von Huene et al. (2000) noted the similar timing (late Miocene) of the arrival of the Cocos Ridge at the accretionary prism and an unconformity observed in seismic reflection data and in Deep Sea Drilling Project cores, although others have suggested younger ages of impingement (Morell et al., 2012; Vannucchi et al., 2013). Similarly, an unconformity is presently developing in the Lesser Antilles where the uplifted (by 2 km) Caribbean plate exposes Lower Cretaceous arc volcanics as a result of an oceanic ridge subduction in the Eocene–Miocene (e.g., Bouysse and Westercamp, 1990).
Element 1B: Deformation Results from the Aseismic Ridge Penetration into the Accretionary Prism
In the Lesser Antilles, increased folding and faulting in the accretionary prism are observed on seismic reflection data along the convergent path where the Barracuda and Tiburon Ridges, and other aseismic ridges, or their extensions, intersected the accretionary prism/forearc (e.g., Bangs et al., 2003; Laigle et al., 2013). In this particular case, it is hypothesized that a buoyant aseismic ridge was transferred to the upper (overriding) plate and began to function as the backstop (e.g., Bangs et al., 2003). McCann and Sykes (1984) also proposed trench rollback as a result of oceanic platform suturing onto the overriding plate in the Puerto Rico–Hispaniola region.
In the Middle America Trench of Costa Rica, the extensions of two large oceanic ridges, the Cocos Ridge and the Quepos Plateau, affected both the offshore and onshore portions of the accretionary prism in Costa Rica (von Huene et al., 1995, 2000), unlike the short-term nature of the seamount effects (as discussed above). The “tunneling” of the seamounts and plateaus into the accretionary prism/outer arc must cause localized extreme deformation of the accretionary sediments, forming mélange, as proposed by Jacobi and Wasowski (1985) for the Dunnage Mélange in Newfoundland, and perhaps as evidenced by the exposed Osa mélange above the Cocos Ridge extension in Costa Rica (Meschede et al., 1999).
Element 2: Flat-Slab to Moderately Dipping Slab Subduction
The dip of the subducting oceanic plate shallows under the upper plate where “buoyant” aseismic ridges on the downgoing plate inhibit slab sinking. This shallowing is commonly called “flat-slab subduction.” Classic examples of flat/shallow-slab subduction can be observed under the Andes (e.g., Gutscher et al., 1999; for in-depth reviews, see Kay and Coira, 2009; Ramos and Folguera, 2009) and east of the Middle America Trench (e.g., Lücke and Arroyo, 2015).
Along the Andean margin, two large flat-slab subduction segments are observed that are associated with magmatic gaps, the Pampean or Chilean flat-slab segment, which is caused by the oblique subduction of the Juan Fernandez Ridge (Fig. 1A), and the Peruvian flat-slab segment, which is related to oblique subduction of the Nazca Ridge (e.g., Kay and Coira, 2009; Ramos and Folguera, 2009; Baudino and Hermoza, 2014), and perhaps a second aseismic ridge, the proposed (and presently totally subducted) Inca Plateau (Gutscher et al., 1999). Additionally, a third magmatic gap, the Patagonian volcanic gap, occurs where the Chile Rise spreading center intersects the South American margin. The scale of these flat-slab segments and volcanic gaps is immense; the Peruvian flat-slab segment is about 1000 km long, measured parallel to the Andean margin, and the Pampean or Chilean segment is about 620 km (e.g., Ramos and Folguera, 2009). Yet, the Nazca Ridge is only about 260 km wide (e.g., Gutscher et al., 1999), and the Juan Fernandez Ridge varies from about 16 to 88 km wide (Ramos and Folguera, 2009). The great trench-parallel length of the Peruvian flat-slab segment can be partially explained by oblique subduction, with the Nazca Ridge sweeping along the margin during oblique subduction, but Gutscher et al. (1999) proposed a second subducted plateau, the Inca Plateau, to explain the long trench-parallel length of Peruvian flat-slab segment and the significant trench-parallel dips and high in the subducting slab depth inferred from earthquakes. The flat-slab lengths along the margin are impressive when compared to the scale of outcrop geology in New England.
The flat part of the oceanic slab under Peru extends from about 250 km to about 550 km away from the trench, at about 100 km depth, before the slab begins sinking down to 150 km depth at 700 km away from the trench (e.g., Ramos and Folguera, 2009). At the maximum extent of the present Chilean flat slab, the oceanic slab reaches 100 km depth about 250 km from the Peru-Chile Trench and dips gently from there to 150 km depth about 650 km from the trench, where it then begins to dip more steeply (e.g., Cahill and Isacks, 1992; Kay and Coira, 2009).
In Costa Rica, earthquake seismic tomography (e.g., Syracuse et al., 2008; Arroyo et al., 2009) integrated with gravity data and modeling (Lücke and Arroyo, 2015) show a relatively shallowly dipping subduction zone segment below the “Seamount Province.” The oceanic plate (Cocos plate) dips at 58° between 80 and 200 km depth, compared to a dip of 71° for the same depth to the north where smooth ocean crust is being subducted (Lücke and Arroyo, 2015). This shallower subduction translates to about a 48 km offset away from the trench for the 100 km depth structural contour on the top of the subducting slab. The shallow-dipping portion of the subducting oceanic plate is only about 75 km wide (measured parallel to the strike of the arc), with relatively narrow boundaries (Lücke and Arroyo, 2015).
In the Cocos Ridge region of the Middle America Trench, several effects result from the increased rigidity and resistance to bending caused by the extra thickness of the seamount province, aseismic plateaus, and ridges. These manifestations include reduced relief on the peripheral bulge, a shallower trench (by as much as 4 km), less relief on trench-parallel grabens and horsts in the seamount province, no grabens and horsts related to oceanic-plate flexing in the Cocos Ridge region, and a dip only half the usual under the near-trench part of the accretionary prism (e.g., von Huene et al., 2000; Ranero et al., 2003; Grevemeyer et al., 2007). Further, the strikes of the usually trench-parallel horst and grabens curve toward parallelism with the Cocos Ridge, some 45° away from the trench axis, in response to the lack of flexing and descent of the Cocos Ridge at the outer wall of the trench, compared to regions to the north (see figure 2invon Huene et al., 2000).
Element 3: Volcanic Gap above the Flat-Slab Subduction Segment
Magmatic/volcanic gaps (measured along arc strike) occur above subducted aseismic ridges (Fig. A1). For example, along the Andean margin, both the Peruvian and Chilean (Pampean) flat-slab segments display a distinct lack of Quaternary volcanoes above the flat-slab segments (e.g., Isacks and Barazangi, 1977; Ramos and Folguera, 2009), i.e., on the order of 1000 km (measured parallel to the trench) for the Peruvian flat-slab segment and 600 km for the Pampean or Chilean segment (e.g., Ramos and Folguera, 2009). Smaller Quaternary volcanic gaps also occur, such as the Pica Gap in northern Chile, where the Iquique aseismic ridge subducted (Kay and Coira, 2009). Similarly, Quaternary volcanoes are absent above the landward extension of the subducted Cocos Ridge (e.g., Lücke and Arroyo, 2015).
Island arcs display similar volcanic gaps. For example, in the Lesser Antilles, the northern part of the arc experienced a 10 m.y. hiatus in volcanic activity, which may have been the result of a shallow-dipping subducting slab related to a series of seamount chains that intersected the arc during the late Oligocene (e.g., McCann and Sykes, 1984; Bouysse and Westercamp, 1990).
Adakites are dispersed across the volcanic gap above the subducted Cocos Ridge (e.g., Drummond et al., 1995). The proposed origin of these adakites is uncertain and includes (1) partial melting of the downgoing slab from contact with the mantle/asthenosphere, either at a slab window (e.g., Abratis and Wörner, 2001) or a detached slab (Gazel et al., 2011), or (2) melting of basalt or cumulate mafics near the base of the upper plate (Bindeman et al., 2005), or (3) fluid-induced mantle melting (Hidalgo and Rooney, 2014).
Volcanic centers can develop far removed from the trench during flattening of the subducting plate. For example, volcanism occurred as far as 850 km from the Peru-Chile Trench at 32°S, where the flattening Nazca plate with the Juan Fernandez Ridge finally dipped below 200 km depth (e.g., Ramos and Folguera, 2009)!
Element 4: Significant Uplift of Parts of the Upper (Overriding) Plate (during Initial Steepening of the Downgoing Slab after Passage of the Ridge)
Along the Andean margin, one of the most dramatic elements related to flat-slab subduction is represented by the uplifted portions of the western continental regions, including the Altiplano (Bolivia, Fig. A1), the Puna region (Chile and Argentina, Fig. A1), and the Fitzcarrald arch (Peru). The Altiplano is a 400-km-wide (trench-normal) region with an elevation of about 3750 m that is located south of the Bolivia orocline and extends some 200 km to the east of the trench. The plateau is estimated to have uplifted between 2500 and 3500 m in ~3.5 m.y. (10.3 Ma to 6.8 Ma; Garzione et al., 2006; Ghosh et al., 2006). A range of contrasting models have been proposed to explain the Altiplano, from those that suggest the uplift is related to slab steepening after passage of a subducted ridge (e.g., Kay and Coira, 2009) to those that do not call upon ridge subduction as a causal mechanism (e.g., Molnar and Garzione, 2007). Complicating the ability to construct a realistic tectonic model is the likelihood that either crustal thickening or removal of dense mantle (and dense lower crust) can result in isostatic uplift (e.g., Garzione et al., 2008). Based in part on the observation of low seismic velocities that imply no mantle presently remaining under the Altiplano (Myers et al., 1998), Molnar and Garzione (2007) and Garzione et al. (2008) proposed that rapid removal (delamination) of the eclogitic lower crust and lithosphere during a relatively short period (3.5 m.y.) drove Altiplano uplift. In this model, the lower crust and deeper lithosphere removal was accomplished perhaps by sinking drips of dense crust/lithosphere (Garzione et al., 2008). The mantle removal would allow lighter asthenospheric inflow that in turn would result in isostatic uplift in late Miocene time (e.g., Molnar and Garzione, 2007; Garzione et al., 2008). Molnar and Garzione (2007) proposed that the extended ~30 m.y. period of crustal shortening and thickening (and possibly even underthrusting by the Brazilian Shield; Myers et al., 1998) did not result in significant uplift before delamination because of a dense eclogitic lower crust.
Perceived problems with this model of uplift from delamination (see, for example, Kay and Coira, 2009) can be resolved with a model that suggests that mid- to lower-crustal (silicic) flow from regions on either side of the present Altiplano (the Cordillera Occidental and the Cordillera Oriental, which were both elevated before the Altiplano) depressed the lower crust beneath the future Altiplano, causing the lower crust of the future Altiplano to eclogitize and then delaminate, resulting in Altiplano uplift (e.g., Husson and Sempere, 2003; Kay and Coira, 2009). Kay and Coira (2009), among others, suggested that slab steepening after Juan Fernandez Ridge passage resulted in decompression melting (Fig. A1). This melting gave rise to mafic magmas that ponded at the base of the crust. The resistate (residuum) of the differentiating mafic magmas, along with the eclogitized crust, resulted in a lower crust that was denser than the mantle, which in turn caused delamination and sinking of the lower crust. This delamination and sinking would have been accompanied by a rise of asthenosphere, as proposed for the Altiplano and other parts and times of the Andean margin, such as the Carboniferous–Permian magmatism and tectonics in the Pampean region of Argentina (e.g., Ramos and Folguera, 2009).
Along the Andean margin in Peru, north of the Bolivia orocline, the flat-slab subduction of the Nazca Ridge extension may have caused the uplift of the Fitzcarrald arch (Ramos and Folguera, 2009), about 550 km from the trench (measured parallel to the relative convergence orientation), as well as some 400 km inland from the coastline at Lima. In the Maranon Basin, which is located in northeastern Peru east of the Andes (the western margin of the basin is about 500 km from the trench), an estimated 800 to 1200 m section is missing, based on seismic reflection profiles, well logs, and thermal history modeling (Baudino and Hermoza, 2014). The timing of the uplift and erosion represented by the regional unconformity is consistent with the timing (Pliocene, ca. 4 Ma) of the arrival of the “Inca Plateau” at the basinal margin (Baudino and Hermoza, 2014), which Gutscher et al. (1999) proposed presently lies immediately below the basin.
Element 5: Post-Flat Slab Magmatism: Alkalic and Rhyodacitic Volcanism with Continental Partial-Melt Signatures in Large Calderas Far Removed from the Trench
If the trajectory of the subducting aseismic ridge is oblique to the relative convergence vector, or if the aseismic ridge is discontinuous, at some point, the subducting ridge will move away from a particular region on the upper plate that was formerly above the aseismic ridge. The result is steepening of the downgoing slab in that area, with consequent decompression melting and asthenospheric upflow into the newly created wedge between the downgoing slab and the upper plate (Fig. A1). These effects can result in continental delamination and partial melting of the lithosphere. The result is magmatic provinces with mantle and continental partial-melt signatures that are stepped significantly back from the trench (as much as 500 km to 800 km), compared to the location of the usual frontal volcanic arc (Fig. A1; Kay and Kay, 1993; Kay and Coira, 2009; Ramos and Folguera, 2009). This is the “inner arc” that defines the “arc-jump” compared to location of the arc in times preceding flat-slab subduction.
Classic examples for magmatism that postdates flat-slab subduction occur in the Andes; Kay and Coira (2009) developed three transects that stretch across the Andes (southern Altiplano and farther south) in order to demonstrate the magmatic effects of flat-slab subduction (two of which are displayed in Fig. A1). Below we review the southern Altiplano transect (19°S to 20°S) in order to establish the general components of the post-flat-slab magmatism (Fig. A1).
The Andean Altiplano-Puna Plateau, located just south of the Bolivia orocline in Chile and Bolivia, records the effects of slab steepening after the southward passage of the aseismic Juan Fernandez Ridge. For the southern Altiplano region at 19°S to 20°S, volcanic inactivity and uplift occurred from about 38 Ma to 27 Ma (e.g., James and Sacks, 1999; Kay and Coira, 2009). James and Sacks (1999) proposed that the volcanic inactivity and uplift are indicative of flat-slab subduction as the Juan Fernandez Ridge passed by, although Kay and Coira (2009) questioned whether the subduction had been flat or merely moderately dipping. In the region of the southern Altiplano transect, magmatism that postdates flat-slab subduction developed in three bands: the “frontal arc,” which is about 200 km from the trench, the “inner arc,” which has a maximum extent of 300 km to 500 km away from the trench, and the Tambo-Tambillo volcanics, which separate the two bands and are about 300–400 km behind the trench (Figs. A1 and A2; e.g., Hoke and Lamb, 2007; Kay and Coira, 2009).
The first magmatic events to occur when the slab began to steepen (but while the slab was still dipping relatively shallowly; Kay and Coira, 2009) were the 25–22 Ma (Oligocene–Miocene) Tambo-Tambillo bimodal volcanics, including basalts and shoshonites (Fig. A2). The source of the volcanics was spinel lherzolite mantle at <90 km depth, based on REE modeling (Hoke and Lamb, 2007).
The inner arc, which lies east of the Tambo-Tambillo volcanics, is a long-lived series (ca. 25 Ma to <1 Ma) of andesitic to dacitic domes, calderas, shoshonite flows, and ignimbrites (Fig. A2; Kay and Coira, 2009). The dacitic ash flows from one complex alone, the 9 Ma to 5 Ma Los Frailes complex, cover about 2000 km3 (Kay and Coira, 2009). A low-velocity zone below the Los Frailes complex in the continental crust at 15 km depth is thought to indicate a zone where the lithosphere has either been detached or significantly transformed by partial melting (e.g., Beck and Zandt, 2002; Kay and Coira, 2009). Helium isotope data also suggest that mantle melting occurred under the Los Frailes complex, as well as other complexes in the inner arc (Hoke and Lamb, 2007). The Los Frailes complex is associated with proposed decompression melting and continental delamination during slab steepening that occurred between 10 and 6 Ma (Figs. A1 and A2; Kay and Coira, 2009). Ignimbrites at another complex (the 8.4 to 6.5 Morococala center; Fig. A2) include two-mica rhyolitic tuffs (commonly thought of as continentally derived), and biotite-quartz latite tuffs (Kay and Coira, 2009). Pliocene to Pleistocene (<5 Ma) basaltic to high-K basaltic andesites also erupted in the same region.
After the subducting aseismic ridge has moved past a particular region of the trench/convergent margin, the frontal arc is re-established. The frontal arc, about 200 km from the trench, also produced early ignimbrites: the voluminous 22.7 to 19.4 Ma Oxaya rhyodacites, which cover over 3000 km3 (Fig. A2; e.g., Wörner et al., 2000; Kay and Coira, 2009), are centered about 200 km from the trench. Large andesite shield volcanoes were built from 20 to 9 Ma. Late Miocene and Pliocene to Holocene volcanism in the frontal arc show about a 40 km migration cratonward from large Miocene arc centers and the early ignimbrite fields in the Oxaya field (Fig. A2; Kay and Coira, 2009). Arc migration away from the trench after the aseismic ridge passes has also been proposed for several island-arc systems. For example, in the Lesser Antilles, the northern part of the arc jumped 50 km farther from the trench after 10 m.y. of volcanic inactivity (e.g., Bouysse and Westercamp, 1990).
Element 6: Retroarc Thrusting of Both Basement and Sedimentary Cover
Retroarc thrusting is a hallmark of flat-slab subduction, and it can involve both basement and cover sequence units (Figs. 4 and A1). As Ramos and Folguera (2009) summed up, as the flat-slab (or moderately dipping) subduction steepens, the resulting heating and partial delamination of the crust/lithosphere significantly weaken the continental crust. Since the continent is under horizontal compression, the weakened crust shortens and fails in a series of thrusts and folds that involve basement. The retroarc thrusting loads the foreland plate, resulting in a rapidly responding and rapidly subsiding foreland basin that is yoked to the thrusts from the hinterland (Fig. 4). Additionally, retroarc thrusting with a basement component occurs during flat-slab subduction as well, with the same results of a loaded foreland developing a foreland basin (Fig. 4; e.g., Hilley et al., 2004; Ramos and Folguera, 2009; Kay and Coira, 2009). These retroarc thrust systems may have an additional causal factor. Maloney et al. (2013) found that most compressional events along the Andean margin (including the fold-and-thrust belts east of the Altiplano-Puna Plateau in the central Andes) correlate with trench-normal convergence rates of >4 cm/yr.
The retroarc region of the Andes provides an example of the structural elements that develop during and after flat-slab subduction, coupled with a relatively high trench-normal convergence rate (e.g., Kay and Coira, 2009; Ramos and Folguera, 2009; Maloney et al., 2013). One example is the Aconcagua fold-and-thrust belt above the Pampean flat-slab segment in the central Andes at 32.5°S (Fig. 4), which contains the highest peak in the western hemisphere. As the Juan Fernandez aseismic ridge arrived at the margin in this region at about 9 Ma, thin-skinned shortening in the Principal Cordillera (located immediately west of the Chile-Argentina boundary) transformed into intracontinental thrusts (“thick-skinned” thrusts that involved basement rocks) in the Frontal Cordillera east of the Chile-Argentina boundary about 260 to 300 km east of the trench (Fig. 4; e.g., Hilley et al., 2004; Ramos and Folguera, 2009; Giambiagi et al., 2012). The eastward migration of the thrust front increased from ~2.5 mm/yr in pre-flat-slab time to ~13.3 mm/yr during shallow slab subduction (Ramos and Folguera, 2009).
From 5 Ma to 2 Ma, east-directed thrusting migrated farther east, forming the Precordillera thrust belt between 300 and 360 km from the trench (Fig. 4). This east-directed thrusting involved Paleozoic sequences, minor Mesozoic sequences, and the Cenozoic sedimentary section of the western part of the Bermejo foreland basin. In some interpretations, Precordilleran thrusting involved continental basement as well (e.g., Hilley et al., 2004, see their figure 13), but in other models generally did not (Zapata and Allmendinger, 1996; Ramos and Folguera, 2009). In contrast to a simple model of basin development in response to thrust loading on its western margin, west-directed reverse faults involving basement also occur along the eastern margin of the Bermejo foreland basin (Zapata and Allmendinger, 1996; Jordan et al., 2001); these thrusts are related to basin and range uplift of the Sierra Pampeanas in response to the flat-slab subduction.
Thrust loading resulted in an increase in the Miocene Bermejo foreland basin subsidence rate, from 0.33 mm/yr in pre-flat-slab time (20–9 Ma) to 0.77–0.95 mm/yr during initiation of flat-slab subduction (9–6 Ma; Ramos and Folguera, 2009). Ramos and Folguera (2009) estimated that the foreland basin response in subsidence rate to thrust loading is very rapid in broken continental crust, on the order of 1 m.y. Modeling of the subsidence in the foreland basins, including the Bermejo Basin, from about 29°S to 31.5°S showed that from 20 to 9 Ma, the thin-skinned thrust loading did not have a significant effect on the Bermejo Basin (Cardozo and Jordan, 2001). Cardozo and Jordan (2001) and Jordan et al. (2001) found that the response of basins to loading is unsteady and spatially variable, based on the rigidity of the basin crust, including such factors as amount of faulting and terrane boundaries, and thickness of the crust.
A second example of retroarc thrusting along the Andean margin is the Agrio fold-and-thrust belt in the Neuquen Basin at ~37°S. The downgoing Nazca plate presently has a moderate dip of about 30° (e.g., Vera et al., 2015), but late Miocene rhyodacitic magmas up to 550 km from the trench, high gravity anomalies, and tomographic analyses that suggest shallow asthenosphere together indicate that this region (the Payenia flat-slab segment) underwent flat-slab subduction from 15 to 5 Ma (Kay and Copeland, 2006; Kay et al., 2006a, 2006b; Folguera et al., 2007; Ramos and Folguera, 2009). Fault systems that developed during continental breakup (detachments and listric faults) were reactivated in several phases, including during Eocene and the flat-slab Miocene compressional events. Basement-involved thrusts accommodated shortening in the west, whereas suprastructure thrusts are common in the Agrio fold-and-thrust belt farther east (e.g., Ramos and Folguera, 2009; Giambiagi et al., 2012).
Island arcs commonly display retroarc thrusts, first observed over 40 yr ago (e.g., Hamilton, 1979; Silver et al., 1983) in the Banda/Sunda arc. However, Silver et al. (1983) suggested that the retroarc thrusting there was the result of arc-continent collision. Researchers ten Brink et al. (2009) catalogued several retroarc thrust systems around the globe that have various proposed origins, including incipient subduction polarity flips or subduction-related mantle flow. However, detailed study of faults revealed in multibeam bathymetric data of the Muertos thrust belt, which is a retro-thrust belt on the south side of Hispaniola and Puerto Rico, as well as sandbox experiments, suggested to ten Brink et al. (2009) that retroarc thrusting in island arcs can form without incipient subduction polarity flip or mantle flow. Rather, rigid behavior of an arc will transmit stress into the retroarc side of the arc. It appears that retroarc thrusts at an island arc are thus not definitive in terms of ridge subduction.
APPENDIX C: KEY TO BIOSTRATIGRAPHIC RANGES IN FIGURE 7
For Undeformed Flysch Zone: (1) Riva (1974) and Fisher (1977). (*2) Mitchell, herein, Diplacanthograptus spiniferus, Geniculograptus typicalis, and Orthograptus quadrimucronatus recovered from Schenectady Formation strata immediately beneath the western limit of the Vischer Ferry thrust zone in bluffs on north side of the Mohawk River (GPS: 42.822593°N, 73.858149°W on private property; 42.824676°N, 73.860769°W, accessible from the Mohawk Landing Nature Preserve).
Vischer Ferry Thrust Zone (new): (*3) Mitchell, herein; (3a, b) D. spiniferus from siltstones within strongly folded, thin-bedded sandstone between mélange zones in bluffs on north side of the Mohawk River east of the western limit of the Vischer Ferry thrust zone (GPS: 42.820786°N, 73.856315°W, on private property) and along access road to Lock 7, south side of Mohawk River (GPS: 42.79986307°N, 73.84593118°W); (3c) D. spiniferus, Dicranograptus nicholsoni, Amplexograptus praetypicalis, and Cryptograptus insectiformis from folded, silty black shales beneath prominent mélange zone, exposures adjacent to Lock 7 Powerhouse, north side of Mohawk River (GPS: 42.80174447°N, 73.84220098°W).
Vischer Ferry Folded and Faulted Flysch Zone: (*4) Berry (1963a) Sites S1 and S2, D. spiniferus Zone faunas. Re-collection from bluffs on south side of Saratoga Lake by Mitchell also yielded D. spiniferus Zone faunas. (*5a) Berry (1963a), Orthograptus truncatus intermedius Zone in silty black shales region between Round Lake and Saratoga Lake; Berry sites S5, S6, and S8 lie within this zone and contain D. spiniferus Zone fauna. S3–S4 lie west of Saratoga fault; S1, S2, S7 lie within Vischer Ferry zone (see below). (*5b) D. spiniferus Zone faunas from recent road-cut exposures on Round Lake Bypass (GPS: 42.948825°N, 73.793396°W) and roadside exposure on Ruhle Road North (GPS: 42.960475°N, 73.816947°W). (6) Snake Hill olistolith, from sandstones of Snake Hill Formation, Berry (1963a) and English et al. (2006) reported “Trenton” trilobites and brachiopods, and Mitchell (inEnglish et al., 2006) reported Normalograptus mohawkensis, indicative of the O. ruedemanni Zone.
Western Exotic Mélange: (7) Ages poorly controlled; Ruedemann (1930) provided a joint list for a few poor outcrops but not for individual sites; Plesch (1994) concluded age range is probably similar to Vischer Ferry zone.
Halfmoon Graywacke Zone: (8) Western locality, the “old quarry” site of Berry (1962, 1977), Rickard and Fisher (1973), Plesch (1994), lower Corynoides americanus Zone. (*9) Interstate-87 (Riva fide Kidd inPlesch, 1994), Nemagraptus gracilis Zone of Riva (1974), but likely upper part, i.e., within the lower Climacograptus bicornis Zone of Berry (1962). (10) Eastern locality, C. bicornis Zone (Ruedemann, 1912).
Eastern Exotic Mélange: (11) Blocks at Cohoes Falls locality, Riva fide Kidd inPlesch (1994), N. gracilis Zone of Riva (1974), but likely as at site 9, through C. americanus Zone. (*12) Stark’s Knob olistolith, “Trenton”-age pillow lava (Landing et al., 2003), age based on presence of Liospira? sp. (misspelled as Leiospira in Landing et al.), possibly representing a gastropod from mid-Sandbian (Turinian; C. bicornis Zone) and younger rocks in North America and Europe (Paleobiology Database, 2017). (13) Black mudstone block near contact with N. gracilis Zone fauna (Riva fide Kidd inPlesch 1994), but likely as at site 9.
Flysch Mélange Zone and Waterford Shale Zone: (14) Waterford Dam, Peebles Island, and Green Island, C. americanus to O. ruedemanni zones (Ruedemann, 1901, 1912, 1930; Riva fide Kidd inPlesch, 1994; Mitchell inEnglish et al., 2006). (15) “Type” Normanskill Group, Mount Merino Formation, cherty argillite blocks at classic exposures along Normans Kill at Glenmont and Kenwood rail road cut, C. bicornis Zone (Ruedemann, 1901, 1912, 1930; Berry, 1962; Plesch, 1994; Kidd et al., 1995). (16) Albany Rural Cemetery, C. americanus Zone (Ruedemann, 1908, 1930; Berry, 1963a; Riva, 1974; Goldman, 1995).
Frontal Exotic Mélange: (*17) O. truncatus intermedius Zone, Austin Glen–like sandstone blocks and shale matrix of wildflysch at Moordener Kill (Berry, 1962; Berry inZen, 1967, cited erroneously by Bradley and Kusky [1986, p. 677] as indicating a D. spiniferus Zone age), and also from sandstone blocks “1½ miles east of Rensselaer” (Berry, 1962, 1977). (18) Rysedorph Hill, wildflysch pebbles, Lower Cambrian–“Trenton” (Ruedemann, 1930) and limestone blocks with “Snake Hill” shelly fauna (Zen, 1967), too restrictively interpreted as D. spiniferus Zone by Bradley and Kusky (1986), as both units include equivalents of C. bicornis to D. spiniferus zone faunas. (19) Black chert-bearing, Mount Merino–like blocks at four Ruedemann (1930) localities, C. bicornis Zone, summarized in Plesch (1994).
APPENDIX D: BIOSTRATIGRAPHIC AGE CONSTRAINTS ON THE TIMING OF DEFORMATION
Figure 7 reviews and updates the biostratigraphic control on the units involved in the several mélange zones and deformed flysch zones west of the Taconic allochthon exposed along the Mohawk River. To accurately assess the biostratigraphic age of these rocks, however, it is necessary that we touch briefly upon certain details of the biostratigraphic zones employed. Graptolite faunas from these rocks have been referred to different zonal schemes: those of Berry (1960) and Riva (1969, 1974), and one or another of these schemes has been used subsequently by other authors (Fig. A3). However, confusion about the meaning and scope of these zones has led to mistaken assertions about the age of the rocks. In particular, both zonal schemes include a N. gracilis Zone, but these zones differ greatly in duration and concept (Figs. 7, A3). Berry (1960) employed two zones for rocks in this interval of concern: the N. gracilis and C. bicornis zones, distinguishing the latter (younger) zone by the first appearance of the eponymous species, among others. Riva (1974) contended that these zones had the same fauna and, relying on an assemblage zone approach, combined them into a single, longer N. gracilis Zone. Following the British tradition, he recognized a Diplograptus multidens Zone for the upper part of the interval that Berry referred to the C. bicornis Zone. This upper C. bicornis Zone interval lacks N. gracilis and generally has a lower-diversity fauna. More recent work in the southern and central Appalachians by Finney et al. (1996), as well as work on Sandbian rocks elsewhere in Laurentia (e.g., Lenz and Chen, 1985; Mitchell et al., 2003), Australasia (Vandenberg and Cooper, 1992), China (Chen et al., 2017), and Wales (Bettley et al., 2001), demonstrates that, contrary to Riva’s (1974) assertion, a distinct N. gracilis–bearing interval with a unique fauna precedes the appearance of C. bicornis and the associated species of the C. bicornis Zone. N. gracilis continues upward into the lower half of the C. bicornis Zone, which we may recognize as an informal lower C. bicornis Zone, along with an upper subzone that is roughly equivalent to Riva’s D. multidens Zone, which lacks N. gracilis but includes a few new species.
Turning to the age of the classic Normanskill graptolite faunas, the collections tabulated by Berry (1962) from the Mount Merino and Austin Glen formations are nearly all referable to the lower C. bicornis Zone as defined above. They contain both N. gracilis and C. bicornis together with a diverse assemblage of other graptolites. Nine of the samples reviewed by Berry (1962), mostly low-diversity collections from near the base of the Mount Merino Formation, contain N. gracilis without C. bicornis, but nearly all of these collections contain one or another of the usual C. bicornis Zone species or lack many of the other species expected in an N. gracilis Zone fauna, or both. Thus, they either are from very high in the N. gracilis Zone or are simply poor samples. Riva provided identifications of N. gracilis Zone collections as personal communication to several workers, which they subsequently quoted in their published work, but none of these sources provided detailed fauna lists, and so we cannot determine the precise zonal assignment for these collections. Based on the available data, there is no convincing evidence that any of the Mount Merino Formation is early Sandbian in age, and, in the absence of positive evidence to the contrary, we assume that all reports of N. gracilis Zone faunas from Normanskill and related rocks in New York State and nearby regions in New England actually represent the C. bicornis Zone strata and are mid- to late Sandbian, or ca. 456.6–453 Ma, in age.
Macdonald et al. (2017) presented new geochronological data that demonstrate that the Indian River Formation is early Darriwilian in age, based on dates on volcanic ashes of 466.1 ± 0.2 Ma and 464.2 ± 0.1 Ma. In their figures 2 and 11, they showed continuity between this now-very-old Indian River Formation and the overlying Mount Merino Formation and placed the bulk of the duration of the latter unit within the mid- to late Darriwilian (as old as ca. 463 Ma). This old age assigned for much of the Mount Merino conflicts with the analysis presented above. Rather, we suspect that the Indian River strata and dark slates of a similar mid-Darriwilian age also known from the Taconic allochthons (Ash Hill Quarry, Mount Merino, New York; Ruedemann, 1904) may be equivalents of the Dauphin Formation in Pennsylvania (Ganis, 2005), which formed prior to initiation of the Taconic foreland basin and which are likewise separated from the overlying mid-Sandbian and younger foreland basin succession by an unconformity with an ~5-m.y.-long hiatus in deposition (Ganis and Wise, 2008; Fig. A3).
In Figure 7, we attempt to distinguish determinations that represent intact strata or mélange matrix from those based on clasts. The matrix and blocks of mélange both indicate the maximum age of mélange formation and associated thrusting. The blocks either were plucked from the hanging wall and footwall during thrusting or were incorporated as the thrust overrode slumps associated with the thrust regime (such as the slumps on the Middle America Trench wall; von Huene et al., 2000). Soft-sediment deformation of the mélange blocks can provide a limit to the minimum age of the first stage of mélange formation, and therefore thrusting, assuming the soft-sediment deformation resulted from thrusting and not earlier slumping. The significance of the age of the youngest blocks is difficult to gauge; it could represent the minimum age of thrusting, i.e., the time the thrust ramp breached the surface and no sediments were yet deposited that were younger, or the age of the blocks could mean merely that the thrust had not sampled strata of a younger age.
The Cohoes Mélange consists of a series of tectonic mélange zones between the frontal thrusts of the Taconic allochthon and a fault zone about 10 km east of the Saratoga-McGregor fault (Figs. 2 and 7). In the central part of this domain, the mélange zones are separated by regions of less-deformed, generally openly folded strata. Exotic clasts are rare in the western exotic mélange belt (Fig. 7) and most consist of deformed remnants of the beds surrounding the mélange. In situ strata exposed at the surface in the Vischer Ferry Zone and farther west in the Undeformed Flysch Zone contain a D. spiniferus Zone graptolite assemblage that has a D. spiniferus Zone age (sites 1–5, Fig. 7). Four new graptolite localities in the western extent of the Vischer Ferry Zone establish that the sediments in this part of the belt include rocks of D. spiniferus Zone age and that these rocks were thrust onto rocks of this same age (Figs. 7, A3). The type locality of the Snake Hill Formation is an olistolith (roughly 350 m in width) on the eastern shore of Saratoga Lake, within the Vischer Ferry zone, and its sandy tempestite-dominated and shelly fossil–bearing succession has produced graptolites of the Orthograptus ruedemanni Zone (Mitchell inEnglish et al., 2006). Berry (1963a) recovered D. spiniferus Zone graptolites from black shales on the west side of the lake, and we have obtained additional D. spiniferus Zone assemblages from new exposures near Round Lake (Appendix C; Fig. 7). Older Utica strata most likely underlie both the Vischer Ferry zone and the undeformed flysch to the west, but no strata are known to crop out in that region.
The matrix of the Western Exotic Mélange and of the zones farther to the east are poorly constrained. Where the data are clearly derived from matrix, those data in all the belts east of the Vischer Ferry Zone suggest C. americanus to D. spiniferus zone ages. In the intervening Halfmoon graywacke zone, Plesch (1994) reported an N. gracilis Zone age based on personal communication from J. Riva to W.S.F. Kidd in 1983, but it is unclear whether this was from an olistolith or intact formation. Parautochthonous rocks from the eastern zones (especially those in the Waterford Shale Zone, variously referred to as Normanskill Shale and Snake Hill Shale) contain mainly C. americanus Zone faunas, although some may be of late C. bicornis Zone age (Rickard and Fisher, 1973; Fisher and Warthin, 1976). On the other hand, several of the classic Normanskill sites of Ruedemann (1901, 1930) are within the Waterford Shale Zone and are blocks of Mount Merino strata (Plesch, 1994; Kidd et al., 1995). Thus, the Halfmoon Graywacke Zone and the mélange zones to the east of that domain (the Eastern Exotic Mélange, the Waterford Shale Zone, and the Frontal Exotic Mélange Zone; Fig. 7) potentially have older ages (up to ~4 m.y. older), or incorporate older recycled material, compared to the mélange zones to the west of the Halfmoon Graywacke Zone.
The Frontal Exotic Mélange, especially the conglomeratic, wildflysch-like units at Rysedorph Hill and Moordener Kill, have a stunning range of clast ages, from mid-Cambrian to possible D. spiniferus Zone age. The older age ranges in the mélange zones east of the Halfmoon Graywacke Zone are consistent with the C. bicornis Zone graptolites found in the Pawlet flysch (Fig. 3; including the Sandbian Ps-5 collection that Berry  included with a set of Tremadocian to lower Floian strata in the Poultney Slate), which is part of the Taconic allochthon farther east, and which contains clasts that appear to have been derived themselves from “Taconic” thrust slices (Rowley and Kidd, 1981).
The matrix ages reported from the wildflysch rocks in the Frontal Exotic Mélange are C. americanus to D. spiniferus (Berry, 1962, 1977; Berry inZen, 1967). Note that Bradley and Kusky (1986, p. 677) referred to the Berry in Zen information, correctly noting that Berry referred faunas from Moordener Kill wildflysch to his Orthograptus truncatus intermedius Zone but then mistakenly equated this with the D. spiniferus Zone. Berry’s zone corresponds to the entire C. americanus–D. spiniferus zone interval of Riva (1969, 1974). Berry cited personal communication from Elam, who considered the Moordener Kill wildflysch matrix to be “the same as those in the Canajoharie Shale” (Berry, 1962, p. 713), which, if correct, indicates that the fauna is older than that of the D. spiniferus Zone.
These wildflysch units are thought to represent olistostrome deposits that were overridden by the thrusts, but the wildflysch may also incorporate clasts plucked from the footwall and hanging wall in addition to those contributed by debris flows. However, the possibility that the phacoidal cleavage in mélange with wildflysch may document hard-rock shearing as well as tectonic dewatering (e.g., Bosworth, 1989) complicates the simple assumption of matrix dating the thrust generation. Nonetheless, the matrix age of these rocks commonly was thought to represent the time at which a thrust bearing the contained blocks intersected the olistostromes at the seafloor (e.g., Zen, 1967; Stanley and Ratcliffe, 1985; Vollmer and Bosworth, 1984). Broken formations adjacent to mélanges of the frontal exotic mélange zone exhibit soft-sediment deformation and phacoidal cleavage as well (e.g., Plesch, 1994), which also suggest that they experienced thrusting soon after matrix deposition, i.e., most likely during the C. americanus to O. ruedemanni Zone interval (e.g., Vollmer and Bosworth, 1984).
This arrival timing is further constrained by reports of Normanskill rocks thrust over conformable sequences of shelf carbonate and black shale along the eastern margin of the parautochthon. Offield (1967), Rickard and Fisher (1973), and Fisher and Warthin (1976) reported conformable contacts between the Balmville Limestone (or equivalents, which are shelly grainstones with a lower Trenton Group fauna) and an overlying black shale unit. These shales are reported to be of late C. bicornis Zone age (Berry, 1962, 1963b, 1970; Berry inOffield, 1967; Riva inRickard and Fisher, 1973).
The formation or matrix age of the rocks in the Frontal Exotic Mélange Zone overlaps extensively with the age of the youngest rocks in the Taconic thrust sequence. Berry (1962, 1963b, 1977) reported C. americanus Zone equivalent ages from the upper part of the Austin Glen graywacke at a site west of Troy within the Giddings Brook slice. Furthermore, the Giddings Brook succession may include gradational contacts between Balmville-like shelly grainstones (which contain the conodont Phragmodus undatus, of late Sandbian or younger age) and the overlying Walloomsac black shales, which are also of C. bicornis Zone age (Potter, 1959; Zen and Bird, 1963; Zen, 1967). Finally, wildflysch units such as that at Whipstock Hill (Zen, 1967) again consist of a vast range of clast lithologies, including Austin Glen–like sandstones along with Trenton-like carbonates within a matrix of Walloomsac lithology. Accordingly, the Frontal Exotic Mélange Zone appears to form part of a continuum of deformation that extends into the lower slice of the Taconic allochthon and frontal thrusts of the Giddings Brook slice of the Taconic allochthon, and it appears to have arrived near its present locale by late C. bicornis to C. americanus Zone time.