Geochronology of the Oliverian Plutonic Suite and the Ammonoosuc Volcanics in the Bronson Hill arc: Western New Hampshire, USA

The ∼400-km-long Bronson Hill arc extends from southern Connecticut to the Maine-Québec border and is a prominent geologic feature in New England. The Bronson Hill arc consists of metamorphosed mafic and felsic volcanic rocks (Ordovician Ammonoosuc Volcanics), felsic plutonic rocks (Ordovician Oliverian Plutonic Suite) of varying composition, and a metamorphosed cover sequence of graphitic-sulfidic schist, volcanic rocks, and minor quartzite (Ordovician Partridge Formation; Fig. 1; e.g., Billings, 1956; Zen et al., 1983; Tucker and Robinson, 1990; Lyons et al., 1997; Moench and Aleinikoff, 2003; Hollocher et al., 2002; Ratcliffe et al., 2011). The Partridge Formation is overlain by the Quimby Formation in northern New Hampshire and western Maine. Felsic metatuff in the Quimby Formation yielded an age of 443 ± 4 Ma (Moench and Aleinikoff, 2003), but the Quimby Formation is not present in the study area. For some time, the Bronson Hill arc has been considered to be part of a larger peri-Gondwanan system of arcs that developed in the Iapetus Ocean outboard of peri-Laurentian arcs now located to the west (e.g., Hibbard et al., 2006). Even though the Bronson Hill arc is currently thought to be built on peri-Gondwanan crust, the position of the suture between Laurentia and the western edge of Ganderia, called the Red Indian Line, remains an open matter of debate (Dorais et al., 2012; Macdonald et al., 2014, 2017; Coish et al., 2015; Tremblay and Pinet, 2016; Karabinos et al., 2017). The Red Indian Line, defined in Newfoundland, is a major terrane-boundary mylonitic fault that separates rocks with North American faunas from rocks with Celtic brachiopods diagnostic of oceanic islands (Williams et al., 1988; Neuman, 1984). In New England, the location of the Red Indian Line has been drawn based on primary and detrital zircon age data by the research referenced above, due to a lack of fossils in the proposed sections. In southwestern New England, Cameron’s Line marks the eastern limit of the autochthonous Cambrian–Ordovician Iapetan carbonate shelf sequence and corresponds to a major Ordovician fault (Rodgers et al., 1959; Rodgers, 1971, 1985; Hatch and Stanley, 1973; Hall, 1980; Walsh et al., 2004) that is interpreted as the Iapetan suture (Stanley and Ratcliffe, 1985). The Red Indian Line and Cameron’s Line only locally coincide near the Connecticut-Massachusetts border, where the Cobble Mountain Formation and Hoosac Formation are in fault contact (Fig. 2; Zen et al., 1983; Rodgers, 1985; Stanley and Hatch, 1988; Karabinos et al., 2017), but south of that, the Red Indian Line is poorly constrained due to a lack of modern mapping and detrital zircon studies. The position and possible correlation, or lack thereof, between Cameron’s Line and the Red Indian Line in southern New England remains an important topic for future research.

Based on Nd and Pb isotopes (Aleinikoff et al., 2007; Dorais et al., 2012) and detrital zircon data that suggest a Ganderian source (Macdonald et al., 2014; Karabinos et al., 2017), the Bronson Hill arc is considered to be built on this Ganderian crust. Exposed within the southern Bronson Hill arc in Massachusetts, the Dry Hill Gneiss (dated at 613 ± 3 Ma) is crust that predates the Ordovician arc (Tucker and Robinson, 1990). It is possible the Dry Hill Gneiss represents Ganderian crust beneath the Bronson Hill arc (Aleinikoff et al., 2007). The Bronson Hill arc is just one of several Northern Appalachian volcanic arcs that were built on a peri-Gondwanan (Ganderian) crustal fragment in the Iapetus Ocean; others include the Penobscot arc-backarc system (513–482 Ma), the Tetagouche backarc (473–455 Ma), and the Popelogan-Victoria arc (475–455), the latter of which is the on-strike correlative of the Bronson Hill arc in Newfoundland, Maine, and New Brunswick (Fig. 1; Hibbard et al., 2006; van Staal and Barr, 2012; van Staal et al., 2016).

Understanding the tectonic origin of the igneous rocks that comprise the Bronson Hill arc in New England is difficult, and locating the arc rocks along the paleomargin of Laurentia or a peri-Gondwanan crustal fragment is a challenge. Recent detrital zircon studies of the Moretown Formation in Vermont led to the recognition that the formation was peri-Gondwanan and not peri-Laurentian as previously thought (Ryan-Davis, 2013; Ryan-Davis et al., 2013; Coish et al., 2013). Follow-up study confirmed this conclusion and led Macdonald et al. (2014) to place the Red Indian Line in Vermont between the Stowe (peri-Laurentian) and Moretown (peri-Gondwanan) Formations in the Ordovician accretionary complex. The Moretown Formation contains arc-derived, metamorphosed sedimentary and volcanic rocks with mafic rocks showing geochemical signatures from a suprasubduction zone setting (Coish et al., 2015). Macdonald et al. (2014, 2017) and Karabinos et al. (2017) used the name “Moretown terrane” for the substrate of the Shelburne falls arc and the Bronson Hill arc, but the distinct peri-Gondwanan Ediacaran detrital zircon provenance of the rocks in both the Moretown and Albee Formations is consistent with the peri-Gondwanan Gander terrane (Fyffe et al., 2009; van Staal et al., 2012) and may not require a new and separately named terrane. The Hawley Formation, which sits stratigraphically above the Moretown Formation, contains both Laurentian- and Gondwanan-age detrital zircons (Macdonald et al., 2014; Karabinos et al., 2017). Those authors used this data set to suggest that the “Moretown terrane” was a crustal fragment separated from Ganderia and was closer to the Laurentian margin at the beginning of the Ordovician.

All tectonic models for the evolution of the closing of the Iapetus Ocean and the development of Ordovician volcanic arcs begin with eastward subduction (Rowley and Kidd, 1981; Stanley and Ratcliffe, 1985; Karabinos et al., 1998; Ratcliffe et al., 1998). The resulting collision of the arc rocks caused the Taconic orogeny (aka “Taconian orogeny”) in New England (Stanley and Ratcliffe, 1985), which, about three decades ago, was dated at ca. 465 Ma by ⁴⁰Ar/³⁹Ar and K-Ar methods on metamorphic minerals from Vermont and Massachusetts (Sutter et al., 1985). This model was subsequently challenged with results of U-Pb zircon ages between ca. 454 Ma and 442 Ma from the Bronson Hill arc and Partridge Formation in Massachusetts, which apparently postdated Taconian metamorphic ages, implying that the Bronson Hill arc was too young to have caused the Taconic orogeny (Tucker and Robinson, 1990). In this revised model, the Bronson Hill arc represented a younger and more eastern arc that postdated an older “Ascot-Weedon-Hawley-Collinsville terrane” (Tucker and Robinson, 1990, p. 1147).

The report of older 485–470 Ma U-Pb zircon ages from volcanic arc rocks in western New England supported the idea that there were two volcanic arcs: the western and older Shelburne Falls arc in Vermont and Massachusetts (ca. 500–470 Ma), and an eastern and younger Bronson Hill arc (Fig. 2; Karabinos et al., 1998). In this scenario, the Shelburne Falls arc developed above an east-dipping subduction zone and collided with the Laurentian margin causing the Taconic orogeny at ca. 475–470 Ma (Karabinos et al., 1998). In their model, subsequent subduction reversal led to the development of the Bronson Hill arc above a west-dipping subduction zone, with subduction of a separate segment of the Iapetus Ocean beneath the newly accreted arc terranes. In the two-arc model, the Iapetus suture would lie to the east of the Bronson Hill arc (Dorais et al., 2008). Debate continued with updated Ar-Ar dating, which suggested that the Taconian metamorphism peaked at ca. 450–445 Ma in southwestern New England, rather than 465 Ma, and overlapped with younger arc ages in the Bronson Hill arc (Hames et al., 1991; Ratcliffe et al., 1998). Faunal succession, isotopic data, and paleolatitude reconstruction led Moench and Aleinikoff (2003) to suggest that the Bronson Hill arc formed off the Laurentian margin, but northwest of the Iapetan suture between ca. 470 and 460 Ma. Recent reconstructions in eastern Canada placed the Popelogan-Victoria arc and its southern correlative Bronson Hill arc on the paleowestern side of Ganderia on the leading edge of a peri-Gondwanan crustal fragment as it traversed Iapetus in the early Paleozoic (van Staal and Barr, 2012). Geochemical study of Nd-Sm, Pb, and Sr isotopes and trace elements suggested that the Ammonoosuc Volcanics are peri-Gondwanan, but that younger plutonic rocks (ca. 450 Ma) that intrude the Ammonoosuc Volcanics have Laurentian geochemical signatures (Dorais et al., 2008, 2012). These data led to a model (Dorais et al., 2012) where the Ammonoosuc Volcanics formed over an east-dipping subduction zone and were thrust westward over the Laurentian margin. The Ammonoosuc Volcanics were then intruded by the Oliverian Plutonic Suite of the Bronson Hill arc, which formed after renewed subduction over a west-dipping subduction zone, leading to inheritance of Laurentian-like Pb, Sm, and Nd geochemical signatures (Dorais et al., 2012). In this model, the Iapetus suture lies to the west of the Bronson Hill arc.

Precise ages for many of the rocks in the study area of west-central New Hampshire have been elusive due to a lack of fossils, a paucity of datable volcanic rocks, a lack of geochronology, and an abundance of faults and nappes that complicate the stratigraphy and prohibit direct correlation with rocks of more certain age. This paper presents 13 new U-Pb zircon ages from the Bronson Hill arc in west-central New Hampshire, from an area lacking geochronological control between southernmost New Hampshire and Massachusetts (Tucker and Robinson, 1990), and northern New Hampshire (Moench and Aleinikoff, 2003). The new data were collected in support of new 1:24,000 scale bedrock geologic mapping (Walsh, 2016; Walsh et al., 2019).

The bedrock geology of the Connecticut River Valley in western New Hampshire and eastern Vermont consists of highly deformed and metamorphosed Lower Paleozoic metasedimentary, metavolcanic, and metaplutonic rocks of the Bronson Hill arc and the Connecticut Valley trough (Fig. 3; Lyons et al., 1997; Ratcliffe et al., 2011; Walsh, 2016). Rocks of the Bronson Hill arc are informally considered part of the New Hampshire sequence (Billings, 1935, 1937; White and Jahns, 1950; Rankin et al., 2013). From the base up, the stratigraphy of the New Hampshire sequence consists of the Albee Formation (Late Cambrian or older), Ammonoosuc Volcanics (Upper and Middle Ordovician), Partridge Formation (Upper Ordovician), Clough Quartzite (Lower Silurian), Fitch Formation (Upper Silurian–Lower Devonian), and Littleton Formation (Lower Devonian). Billings accurately identified the regionally significant Silurian unconformity between the Partridge Formation and the Clough Quartzite. Within the Ammonoosuc Volcanics, there are numerous metavolcanic and metasedimentary units. Based on the work of Rankin et al. (2013) in the area of Billings’ type localities near Littleton, New Hampshire, the stratigraphy of the Ammonoosuc Volcanics (from oldest to youngest) includes (1) rusty sulfidic slate, felsic tuff, and other metasediments; (2) metasiltstone, phyllite, and volcaniclastic rocks; (3) metadolomite and siltstone; (4) metarhyolite tuff and siltstone; (5) meta-andesite, basaltic tuff, and pillow lavas; (6) metarhyolite tuff, lapilli tuff, and lava; (7) metafelsic and mafic volcanics, volcaniclastic rocks, and metasediments. The lower Partridge Formation overlaps with the upper Ammonoosuc Volcanics (Rankin et al., 2013) and consists of interbedded metavolcanics rocks in the lower Partridge Formation and rusty sulfidic schist and slate interlayered with metarhyolite in the upper Ammonoosuc Volcanics. Near Plainfield, New Hampshire, the contacts between the Partridge Formation and the Ammonoosuc Volcanics are generally sharp, but may be gradational over a few meters (Walsh, 2016).

In the Bronson Hill arc, the Ordovician Ammonoosuc Volcanics (Oa) and graphitic-sulfidic metapelite of the Partridge Formation are thrust over or intruded by the Ordovician plutonic rocks (e.g., Leo, 1991). The Ammonoosuc Volcanics and Partridge Formation contain chemically similar volcanic rocks (Hollocher, 1993). Mafic rocks are mostly basalt to basaltic andesite with island-arc tholeiite and backarc basin basalt signatures (Aleinikoff, 1977; Leo, 1991; Hollocher, 1993; Hollocher et al., 2002; Dorais et al., 2012). Interbedded felsic volcanic rocks are interpreted as comagmatic and derived from an oceanic-crustal or mantle source (Leo, 1991; Hollocher, 1993; Hollocher et al., 2002). The composition of the Oliverian Plutonic Suite is alkalic to granitic. In our study area, older plutonic rocks of the Bronson Hill arc are dominantly tonalites and trondhjemites. Younger intrusions are commonly granitic in composition (Leo, 1991; Hollocher et al., 2002, Dorais et al., 2008; this study).

The Connecticut Valley trough lies unconformably or in local fault contact (McWilliams et al., 2010; Karabinos et al., 2010; Ratcliffe et al., 2011) atop remnants of the early Paleozoic volcanic arc rocks called the Shelburne Falls arc to the west (Karabinos et al., 1998; Karabinos and Hepburn, 2001) and the Bronson Hill arc to the east (Stanley and Ratcliffe, 1985; Ratcliffe et al., 1998; Tucker and Robinson, 1990; Leo, 1985, 1991; Dorais et al., 2008, 2012). The Connecticut Valley trough is composed of the Silurian to Devonian Shaw Mountain, Northfield, Waits River, and Gile Mountain Formations in an unconformable autochthonous cover sequence on the pre-Silurian rocks and Mesoproterozoic Laurentian basement rocks of the Mount Holly Complex in Vermont (Fig. 3; McWilliams et al., 2010; Ratcliffe et al., 2011). The Connecticut Valley trough contains Silurian to Devonian volcanic and metasedimentary rocks with U-Pb zircon ages indicating deposition between ca. 432 and 407 Ma (Aleinikoff and Karabinos, 1990; Rankin and Tucker, 2009; McWilliams et al., 2010; Dorais et al., 2017; Perrot et al., 2018). Fossil data from the Connecticut Valley trough range from Late Silurian (Pridoli) to Early Devonian (Emsian; Boucot and Drapeau, 1968; Hueber et al., 1990; Lavoie and Asselin, 2004).

The Connecticut Valley trough evolved from an extensional tectonic setting after the Ordovician Taconic orogeny and Silurian disturbance to a foreland basin setting during the Devonian Acadian orogeny (Black et al., 2004; Tremblay and Pinet, 2005, 2016; McWilliams et al., 2010). East of the Bronson Hill arc, the Central Maine trough merges with the Connecticut Valley trough in Maine and New Brunswick (Osberg et al., 1989; Tremblay and Pinet, 2005, 2016; Hibbard et al., 2006; Rankin et al., 2007). Silurian deposition of sediments in the Connecticut Valley trough and Central Maine trough basins could have locally buried the Red Indian Line (e.g., Williams et al., 1988), obscuring the boundary between Laurentian crust to the west and peri-Gondwanan (Ganderian) crust to the east (van Staal et al., 1998; Rankin et al., 2007; Aleinikoff et al., 2007; van Staal and Barr, 2012).

The Connecticut River Valley forms a boundary between what has been informally called the Vermont and New Hampshire sequences (Billings, 1956; Hatch, 1988; Armstrong, 1997; Thompson et al., 1968; Rankin et al., 2007; McWilliams et al., 2010). The boundary between the Connecticut Valley trough and the Bronson Hill arc is now interpreted to be the “Monroe fault” (Fig. 3; Hatch, 1988; Lyons et al., 1997; Ratcliffe et al., 2011; Spear et al., 2003, 2008; Rankin et al., 2013; Walsh, 2016). Early workers considered this contact to be an unconformity (Billings, 1956; Thompson et al., 1968; Thompson et al., 1997). Rocks of the Bronson Hill arc occur in a thrust sheet floored by the Monroe fault (Fig. 3), which carried a deformed section of plutonic rocks, Ammonoosuc Volcanics, Partridge Formation, Clough Quartzite, and the Fitch and Littleton Formations (Walsh, 2016). The Monroe thrust sheet placed the Bronson Hill arc rocks over the Connecticut Valley trough during an Acadian F₁ nappe-stage event prior to peak metamorphism at lower-amphibolite-facies conditions. Upper- and lower-plate truncations, mylonite, and local mélange characterize the Monroe fault in the area of this study (Walsh, 2016). The Monroe thrust sheet is in turn overthrust by the sillimanite-grade rocks of the Fall Mountain slice. These thrust sheets were historically interpreted as fold nappes (Thompson et al., 1968), and later as thrust nappes (Robinson et al., 1991). Rocks of the Bronson Hill arc and the Connecticut Valley trough were deformed and metamorphosed from greenschist to upper-amphibolite grade during the Devonian Acadian orogeny and to a lesser extent in the Carboniferous to Permian Alleghanian orogeny (Laird et al., 1984; Sutter et al., 1985; Harrison et al., 1989; Spear and Harrison, 1989; Spear et al., 2008; McAleer et al., 2016). F₂ doming deformed the thrust sheets and folded earlier isograds. Lower-greenschist-facies (Acadian to Alleghanian) faults truncated peak-metamorphic assemblages, isograds, and older F₁ folds and faults (Spear et al., 2008; McWilliams et al., 2013; McAleer et al., 2016). Late-stage F₃ folds show preferred left-lateral rotation sense and were probably related to late dome-stage Alleghanian deformation or motion along lower-greenschist-facies faults (Walsh, 2016). Subsequent Mesozoic brittle deformation along with the Ammonoosuc and Grantham faults, and many smaller unnamed brittle faults, had sufficient vertical or oblique-slip components to place sillimanite-grade rocks adjacent to greenschist-facies rocks and further offset the isograds (Walsh et al., 2012; McAleer et al., 2016). Apatite fission-track data support Cretaceous fault displacement and reactivation of older Paleozoic faults (Roden-Tice et al., 2009).

The Ammonoosuc Volcanics are heterogeneous, deformed, and metamorphosed volcanic and volcaniclastic rocks consisting of layered to massive greenstone, amphibolite, biotite-muscovite-chlorite-quartz-plagioclase schist and phyllite, felsic quartz-plagioclase granofels (or metafelsite), and sulfidic quartz-plagioclase schist. Biotite and chlorite appear at lower grades, while garnet and hornblende are present at higher metamorphic grade. In the garnet metamorphic zone, the rocks locally contain fascicular hornblende garbenschiefer. The Ammonoosuc Volcanics contain pods and lenses of epidote, plagioclase, and lesser quartz phenocrysts, and volcanic textures including deformed pillows (Aleinikoff, 1977), fiamme (eutaxitic texture), and volcanic breccia (Walsh, 2016).

In the study area (Fig. 3), the rocks are predominately mafic, but felsic rocks are not uncommon (Walsh, 2016). Regionally, the Ammonoosuc Volcanics are low-K bimodal amphibolites and quartz-plagioclase trondhjemitic gneisses in roughly equal proportion (Leo, 1991; Dorais et al., 2008, 2012). Metafelsites occur as medium- to very fine-grained, locally aphanitic, biotite-muscovite-chlorite-quartz-plagioclase schist or granofels with millimeter-size quartz and feldspar phenocrysts. Two samples of such metafelsite were processed, but no zircon was recovered. Zirconium concentrations in our “Oa” felsic rock samples were typically below 100 ppm.

Metafelsic rocks with definitive pyroclastic textures were mapped as the lapilli tuff member (Walsh, 2016). The lapilli tuff is a massive, pale-green to light-gray, gray-weathering muscovite-chlorite-biotite-quartz-plagioclase schist with white to light-gray felsic, flattened lapilli or lesser volcanic bombs as much as 10 cm long (Fig. 4A). The lapilli tuff also contains dark-gray-green mafic clasts. The matrix is aphanitic and contains millimeter-size quartz and feldspar phenocrysts (Fig. 4B). The lapilli tuff is a minor unit and was mapped in only four places in the North Hartland quadrangle and covers only a small area of ∼0.3 km² (Walsh, 2016). Where mapped separately, the lapilli tuff member is in contact and interlayered with mafic and felsic rocks that are typical of the undifferentiated volcanic member of the Ammonoosuc Volcanics. The individual thickness of lapilli tuff layers within mapped belts is on the order of several meters. A sample of a metarhyolite lapilli tuff was collected from a wooded outcrop ∼75 m east of the summit of Colby Hill in Plainfield, New Hampshire (Walsh, 2016). The dated rock in this study (sample NH-2089, Fig. 3) comes from an ∼3–5-m-thick layer interbedded with greenstone and green biotite-muscovite-chlorite-quartz-plagioclase schist. Chemically, the metafelsites are rhyolitic to dacitic, and the dated rock is a rhyolite. Trace- and major-element geochemistry showed that the rocks are rhyolitic to dacitic, and the dated rock is a rhyolite, from which zircon was extracted.

Felsic metatuff of the Ammonoosuc Volcanics (Oa) in the Alstead dome is a white to light-greenish gray, garnet- and muscovite-bearing, chlorite-quartz-plagioclase gneiss. The gneiss contains accessory calcite and metallic oxides, and trace monazite and zircon. The rock has a sugary, equigranular texture that is composed of a recrystallized matrix of quartz and plagioclase 0.05–0.2 mm across. Quartz and plagioclase porphyroclasts ∼1 mm across may be relict phenocrysts. Chlorite varies from 1 to 10 mm long and appears to have replaced amphibole. Flattened elliptical, polymineralic structures may represent a relict pyroclastic texture. The metatuff was mapped for 5 km along the western flank of the northern part of the Alstead dome and is best exposed at an active quarry on the northern end of Osgood Ledge, where it was sampled (Fig. 3). The metatuff is overlain by sulfidic schist of the Partridge Formation.

The term Oliverian is named for Oliverian Brook in northwestern New Hampshire (Billings, 1935). The Oliverian Plutonic Suite (Lyons et al., 1997) rocks are dominantly felsic, ranging from tonalite and trondhjemite in older rocks (older than 460 Ma, based on analysis in this study) to granite and granodiorite in younger intrusions (younger than 460 Ma; this study). Large plutons are internally massive but are intensely deformed along their margins, ranging from equigranular orthogneisses to augen gneisses. The foliation at the margins of the plutons is parallel to the contact and to the schistosity in the surrounding rocks. Plutons tend to be elongate in a north-south orientation and form a narrow belt spanning the length of the Bronson Hill arc in New England. This study examined 11 plutons from the Oliverian Plutonic Suite in southwestern New Hampshire. Each pluton is described below in order from oldest to youngest.

The Plainfield tonalite is a greenish gray, light-gray–weathering, moderately to weakly foliated epidote ± hornblende/actinolite-biotite-chlorite-quartz-plagioclase tonalite to trondhjemite gneiss. It contains characteristic small augen of quartz and lesser feldspar, up to 0.5 cm across, and accessory calcite and muscovite. Chlorite is a product of retrograded biotite and less abundant amphibole. Lyons (1955) first mapped this rock as the “gneiss east of Plainfield” and considered the Plainfield tonalite to be conformable with the surrounding volcanic rocks, but later identified the rock as trondhjemite within the Oliverian Plutonic Suite (Lyons et al., 1997). New mapping indicates that the Plainfield tonalite intruded the Ammonoosuc Volcanics in a pluton with a current surface exposure of ∼4.5 km² (Walsh, 2016; Fig. 3). The pluton contains screens and xenoliths of the host volcanic rocks and exhibits intrusive contacts along its western side (Figs. 5A and 5B). The northern end of the pluton is characterized by a zone of lit-par-lit dikes or sills. The contact along the eastern side of the pluton is poorly exposed, but felsic volcanic rocks in the adjacent Ammonoosuc Volcanics appear to be truncated. A sample of tonalite was collected at a road cut on Porter Road in the Plainfield pluton in Plainfield, New Hampshire (Fig. 3). Similar unnamed tonalitic rocks occur as lit-par-lit dikes or sills near the Plainfield tonalite, but mapped bodies are unnamed because they are not contiguous with the Plainfield pluton (Walsh, 2016).

The White River Junction tonalite, or “gneiss at White River Junction” (Lyons, 1955), intruded the low-grade belt of Ammonoosuc Volcanics west of the Mesozoic Ammonoosuc fault (Fig. 3). A sample of tonalite was collected from a large road cut on the west side of Interstate 91, at the south end of the southbound onramp at the Wilder exit (Exit 12), near mile post 71.50 (Fig. 3). The sampled rock is a pale-greenish gray tonalite with distinct quartz phenocrysts locally deformed into augen. The rock occurs as lit-par-lit layers, either as sills or dike-like bodies that intrude biotite-grade chlorite schist and greenstone of the Ammonoosuc Volcanics. Tonalite layers “range from a few inches to over 100 feet in width” (Lyons, 1955, p. 119). A sample was collected for comparison with the Plainfield tonalite (Walsh, 2016), which occurs on the high-grade, eastern side of the Ammonoosuc fault. Lyons (1955) described the rock as an albite and quartz “sodaclase tonalite” with minor biotite, zircon, clinozoisite, allanite, and pyrite, but without potash feldspar. Our observations agree with the description by Lyons and show that the White River Junction tonalite and the Plainfield tonalite are quite similar in chemistry, mineralogy, and appearance.

The West Lebanon trondhjemite is a weakly foliated, tan-weathering, white to gray biotite-muscovite-perthite-quartz-plagioclase trondhjemite. The trondhjemite designation is based on mineral chemistry. The West Lebanon trondhjemite crops out in the southern portion of the Hanover quadrangle in New Hampshire, just south of Interstate 89 at exit 20 in the town of West Lebanon (Figs. 3 and 5). A sample of trondhjemite was collected behind a large shopping plaza in Lebanon, New Hampshire, southwest of the junction of Interstate 89 and New Hampshire (NH) Route 12A (Fig. 3). The unit is similar to the “gneiss at White River Junction, Vermont” mapped by Lyons (1955), which is exposed on the west side of the Connecticut River near White River Junction, Vermont (White River Junction tonalite, this study). Lyons described “pink gneissic granite” containing quartz and albite that was intruded by lit-par-lit injection into chlorite schist. In this study, exposed contacts with amphibolites of the Ammonoosuc Volcanics indicate that the trondhjemite is intrusive. Moderate grain-size coarsening and rare garnet were observed in amphibolites near intrusive contacts (Figs. 5C and 5D). Foliation trends in the trondhjemite are parallel to the foliation in the adjacent amphibolite.

The Sugar River dome trondhjemite is a variably foliated, gray- to white-weathering, ±garnet-muscovite-biotite-quartz-plagioclase trondhjemite. Chlorite replaces biotite. The sample of trondhjemite was collected from a wooded outcrop 500 m south of East Green Mountain Road in Claremont, New Hampshire, at the northern end of the Sugar River dome (Fig. 3). The Sugar River trondhjemite is exposed south of Green Mountain to just south of the Sugar River near Claremont, New Hampshire (Fig. 3). Sills or dikes of trondhjemite intrude amphibolite of the Ammonoosuc Volcanics along the margin of the pluton, and screens and xenoliths of amphibolite occur within the trondhjemite. The margins of the trondhjemite are strongly foliated and parallel to the foliation in the overlying metavolcanic rocks. Away from the contact, the trondhjemite is weakly foliated.

The Alstead dome gneisses are gray to white, moderately to strongly foliated, and medium to coarse grained. The gneisses are generally low in potassium and are largely trondhjemitic (Leo, 1991); however, Kruger (1946) also described monzogranite from the Alstead dome. The contact relationships are poorly exposed, but trondhjemitic to granitic dikes cut the Ammonoosuc Volcanics near the contact and suggest that the Oliverian Plutonic Suite gneisses were intrusive, as proposed by Leo (1991). The Alstead dome occurs as a northern and southern lobe (Fig. 3). Most of the southern body is in contact with sulfidic schist of the Partridge Formation. On the east side, the contact is intrusive, and on the west side, the units are juxtaposed by late faulting. A sample of monzogranite gneiss (A1065, from the northern part of the southern lobe) was collected from a recently excavated road cut along Cobb Hill Road (aka Forristall Road on the 1998 topographic map of the Alstead quadrangle). The rock is light-gray, porphyroclastic garnet-muscovite-biotite granite gneiss with a prominent mineral lineation at 125° defined by quartz ribbons, porphyroclast tails, and muscovite and biotite aggregates. At the sample location, contacts with adjacent rock units are not exposed, and xenoliths are not common, but the southern lobe intrudes both the Ammonoosuc Volcanics and the Partridge Formation. Coarse mono crystalline and polycrystalline quartz-eyes are present and are interpreted as deformed phenocrysts common in other Oliverian Plutonic Suite rocks in the Bronson Hill arc (Leo, 1991).

The Croydon pluton, exposed in the core of the Croydon dome, is a white- to gray-weathering, variably foliated chlorite-hornblende-biotite-quartz-plagioclase–potassium feldspar granodiorite with minor quartz monzonite and quartz diorite (Chapman, 1952). The granodiorite contains mafic enclaves. Structurally, the Croydon pluton forms the core of a dome that is overturned on its southwest side. The southwest margin of the Croydon pluton is in contact with amphibolites of the Ammonoosuc Volcanics along an ∼200-m-wide shear zone, where deformation has transformed the granodiorite into an augen gneiss of the same composition (hornblende-biotite-quartz-plagioclase–potassium feldspar; Fig. 5E). The shear zone is too wide to be related to emplacement. Away from the shear zone and within the pluton, where the geochronology sample was collected (Fig. 3), the granodiorite is mineralogically similar to the augen gneiss, but it contains only a weak fabric defined by aligned biotite and hornblende. Part of the eastern side of the pluton is in contact with the amphibolite and schist of the Ammonoosuc Volcanics and is not sheared like the western border. Little of this contact is exposed, as most of its eastern margin is truncated by the Mesozoic Grantham fault, which juxtaposes the Croydon pluton with the Devonian Bethlehem Gneiss and metasedimentary rocks of the Silurian Rangeley and Ordovician Partridge Formations (Fig. 3; Lyons et al., 1997).

The Mascoma dome granite is primarily a pink to tan, weakly foliated to massive biotite-plagioclase-quartz-microcline/microperthite granite, but it locally transitions to a quartz monzonite or quartz diorite in the northwest part of the pluton (Naylor, 1969). Chapman (1939) described a range in compositions from quartz diorite to granite, with the most abundant rock type being granodiorite. Based on the work of Naylor (1969), Lyons et al. (1997) showed the Mascoma dome cored by pink biotite granite and rimmed by granodiorite to tonalite. Compositional changes are gradational within the pluton (Chapman, 1939; Naylor, 1969). A moderate foliation increases toward the margins of the pluton (Naylor, 1969). The Mascoma dome granite is in contact with the Holts Ledge gneiss along a sharp contact, but it is unclear if this unit is a border phase of the Mascoma pluton or the lower part of the of the Ammonoosuc Volcanics (Naylor, 1969; Leo, 1991).

A sample of granite was collected from the western side of the Mascoma pluton, at a roadside outcrop along May Street Extension in Canaan, New Hampshire (Fig. 3). The sample location plots within the border phase granodiorite to tonalite as shown by Lyons et al. (1997) but matches the description of the “unstratified core-rock granite” of Naylor (1969). Because Chapman (1939) did not subdivide the rock types on his map, we are uncertain if the sampled rock is part of the core granite or a granite within the border phase.

The Lebanon dome granite is a pink to tan, equigranular, weakly foliated to nonfoliated, medium- to coarse-grained, blocky-weathering muscovite-biotite-plagioclase-quartz-microcline granite. Biotite is locally retrograded to chlorite. The granite makes up the interior of the Lebanon dome pluton, and a sample of granite was collected ∼40 m south of the northern end of an ∼160-m-long road cut on the east side of Route 120 in the city of Lebanon, New Hampshire (Fig. 3). The granite has long been recognized as the plutonic core of the Lebanon pluton (Hitchcock, 1908; Merritt, 1921; Goldthwait, 1925; Kaiser, 1938; Lyons et al., 1997). The granite is in contact with the marginal quartz diorite (see below). The contact varies from sharp to gradational and locally contains mutually intrusive contacts (Chapman, 1939; Lyons, 1955; Leo, 1991).

A sample of quartz diorite was collected at the western margin of the Lebanon pluton, at a road cut on the south side of U.S. Route 4 at Interstate 89 Exit 19 in Lebanon, New Hampshire (Fig. 3). The quartz diorite is a gray, equigranular, moderately to well-foliated, medium-grained, blocky-weathering epidote-quartz-hornblende-biotite-plagioclase quartz diorite gneiss. The dated rock is a border phase to the core granite of the Lebanon dome pluton (Merritt, 1921; Kaiser, 1938; Lyons, 1955). The quartz diorite is in contact with graphitic schist of the Partridge Formation (Fig. 5F). A rusty sulfidic zone due to metasomatism occurs in the surrounding schist (Kaiser, 1938; Lyons 1955). This altered zone was confirmed near the south end of the Lebanon dome by Walsh (2016) but was not recognized at the sample locality. The metasomatized rock occurs in the Partridge Formation discontinuously along the contact with the quartz diorite as a quartz-biotite-epidote-plagioclase schist studded with abundant plagioclase porphyroblasts (Walsh et al., 2012; Walsh, 2016). Kohn et al. (1992) interpreted the contact between the border gneiss and the Partridge Formation as a fault, and although the contact is locally strained, mapping shows sharp contacts between the two units and map-scale xenoliths of Partridge Formation within the quartz diorite (Walsh, 2016). At the sample collection locality, the contact is sharp and roughly parallel to the dominant foliation (Fig. 5F), and we conclude the intrusive contact is locally sheared, but it is not a significant fault.

The Alstead dome trondhjemite consists of buff to very light-gray, moderately to strongly foliated, medium- to coarse-grained, equigranular trondhjemite gneiss. The rock described here is from the northern lobe of the Alstead dome (see Alstead granite gneiss description above and Fig. 3). It contains plagioclase, quartz, and biotite with accessory muscovite, epidote, garnet, and allanite, and trace zircon and monazite. In the northern lobe, dikes of trondhjemite occur in amphibolite of the Ammonoosuc Volcanics along the contact, and the northern lobe of trondhjemite is contained entirely within the Ammonoosuc Volcanics and is interpreted as intrusive. The dated sample is moderately foliated and was collected from outcrops near the north end of the Alstead dome in Alstead, New Hampshire (Fig. 3).

The Unity dome granite sample is a gray, gray- to orange-weathering, fine- to medium-grained, ±garnet ± hornblende-chlorite-biotite–K-feldspar–plagioclase-quartz rock. The pluton varies in composition from granite and granodiorite to trondhjemite and tonalite. Accessory minerals in the granite sample include magnetite and apatite, and the rock is locally enriched in magnetite near the contact with the Ammonoosuc Volcanics, suggesting contact metasomatism. The granite is moderately to weakly foliated. The sample was collected on Stage Road along the Little Sugar River in Unity, New Hampshire (Fig. 3). Outcrops in the core of the Unity pluton and the area of the sample locality are sparse, so the bulk of the pluton could not be described (Fig. 3).

Zircon U-Pb analyses were conducted on the U.S. Geological Survey (USGS)/Stanford sensitive high-resolution ion microprobe–reverse geometry (SHRIMP-RG) at Stanford University. Samples were analyzed by the authors on four different visits to the Stanford laboratory over 3 yr. Zircon crystals were handpicked and mounted on double-sided tape on glass slides in ∼1 × 6 mm rows, cast in a 25-mm-diameter by 4-mm-thick epoxy disc, ground to half-thickness, and polished with 6 µm and 1 µm diamond suspensions. Crystals were imaged using cathodoluminescence (CL) to identify internal structure/zoning and inclusions. Prior to analysis, the mounted zircons were washed with a 1 N HCl solution, rinsed in distilled water, and dried in a vacuum oven. The mount surface was coated with an ∼100-Å-thick layer of Au for conductivity. The mounts were stored overnight at low pressure (10–7 torr) in the upper sample chamber of the ion microprobe before being moved down onto the sample stage in the source chamber.

Secondary ions were sputtered from the zircon using an O₂⁻ primary ion beam, which was accelerated at 10 kV and had an intensity varying from 3.0 to 4.0 nA. The primary ion beam spot had a diameter of ∼25 µm and sputtered to a depth of ∼1–3 µm. Before each analysis, the sample surface was rastered for 60–120 s to remove surface contamination. All peaks were measured using a 5 scan peak-hopping routine on a single discrete-dynode electron multiplier operated in pulse counting mode.

Calculated ages were calibrated to R33 zircon standard (²⁰⁶Pb/²³⁸U age = 419 Ma; Black et al., 2004), which was analyzed repeatedly throughout the analytical sessions. Age calculations followed the methods described by Williams (1997) and Ireland and Williams (2003), using Squid 2.51 and Isoplot 3.76 programs (Ludwig, 2001, 2003, 2012). Age errors were calculated using Isoplot 3.76 and were calculated by propagating only the assigned data-point errors, without considering the scatter of the data points from one another or from the concordia curve (Ludwig, 2012). Individual data points are plotted as 2σ error ellipses or error bars, whereas calculated weighted averages included 2σ uncertainties. The ²⁰⁶Pb/²³⁸U ages were corrected for common Pb using a ²⁰⁷Pb-correction, whereas ²⁰⁷Pb/²⁰⁶Pb ages were corrected using the ²⁰⁴Pb-correction method. These corrections use common Pb compositions from the Stacey and Kramers (1975) model. Zircon concentration data for U, Th, and the measured trace elements were standardized against in-house zircon standard MAD-green (Barth and Wooden, 2010), with concentrations calculated from secondary ion yields normalized to 90Zr216O+. The U and Th concentrations are believed to be accurate to at least ±20%. U-Pb analytical data are presented in Table 1.

Only data that were <10% discordant and showed no inheritance or Pb loss were used in the calculation of ages. Care was taken to avoid any inclusions or defects in the zircon crystal. Data points with >10% discordance were carefully examined to determine the cause of discordance. Data points were only eliminated if it was clear that discordance was the result of crystal defects (fractures, radiation damage), accidental analysis of inclusions, or data points that were erroneous due to machine irregularities (i.e., sample charging, drift, mass fractionation, etc.). Where zircon analyses produced data that were >10% discordant or reversely discordant, the ²⁰⁶Pb/²³⁸U weighted average zircon ages were calculated. Reverse discordance is the result of either (1) the ²⁰⁷Pb peak being off-center during the analyses or (2) an instrumental mass fractionation occurring for ²⁰⁷Pb/²⁰⁶Pb. Reversely discordant analyses came from multiple sessions, so it is unclear as to which of the two issues was responsible.

The SHRIMP-RG U-Pb zircon geochronology for 13 samples is presented below in order of decreasing age. Data for all samples are included in Table 1, and the ages are summarized in Table 2. Representative zircon images are shown in Figure 6, and data plots are shown in Figure 7. All errors are reported as 2σ.

Analyzed zircon grains from the sample of Plainfield tonalite were 100–150 µm long with aspect ratios (ratio of length to width) of 2:1, but the population included irregular “chips” and square grains. In CL, all grains showed oscillatory zoning, dark inclusions, and cracks and rims that were apparently low in uranium (light color, Fig. 6A). No obvious cores were observed. Analyses of oscillatory-zoned zircon interiors yielded a concordia age of 474.8 ± 5.2 Ma (2σ; Fig. 7A).

Zircon grains from the White River Junction tonalite were 100–200 µm long with an aspect ratio of 2:1, champagne to dark pink, and euhedral to subhedral. In CL images, the zircon grains typically showed large dark cores with minor zonation, and bright oscillatory-zoned rims (Fig. 6B). Analyses were consistently reversely discordant but yielded a ²⁰⁶Pb/²³⁸U weighted mean age of 471.2 ± 6.2 Ma (2σ; Fig. 7B).

Analyzed zircon grains were 150–200 µm long, inclusion poor, euhedral, and clear to pink, and had typical aspect ratios of 2:1. In CL, all grains exhibited oscillatory zoning (Figs. 6C and 6D). Clear grains had luminescent cores and dark rims, while pink grains had a higher percentage of dark overgrowths. Dark zones correlated with high U (>1500 ppm) and rare earth element (REE) concentrations. Analyses from low-U luminescent zones exhibited less age scatter than analyses from dark zones, but both CL populations resolved a statistically indistinguishable concordia age of 466.0 ± 3.9 Ma (2σ). One pink grain had a rounded luminescent core that gave a concordant age of 573 ± 20 Ma (Table 1), which is interpreted to reflect inheritance (Fig. 7C).

Zircon grains from the Sugar River dome trondhjemite were typically 50–150 µm in length with an aspect ratio of ∼2:1, well faceted, and pink to brown in color. Zircon grains exhibited faint oscillatory growth zoning, and some contained irregular-shaped cores (Fig. 6E). Cores and rims of zircon grains yielded a ²⁰⁶Pb/²³⁸U weighted mean age of 460.2 ± 3.4 Ma (2σ; Fig. 7D).

Analyzed zircon grains were 150–200 µm long with aspect ratios of 3:1 and typically exhibited oscillatory zoning in CL images; some contained high-U cores (Fig. 6F). Grains were pink to light brown and euhedral to subhedral. Zircon grains yielded a concordia age of 459.6 ± 2.4 Ma (2σ; Fig. 7E).

Analyzed zircon grains from granite gneiss in the northern part of the southern lobe of the Alstead dome were 50–100 µm wide, inclusion-rich, subhedral to euhedral, and pink, and typically had aspect ratios of ∼3:1. In CL, all grains exhibited oscillatory zoning near their rims, but turbid, patchy, and inclusion-rich cores were common. Additionally, thin (<10 µm) dark rims surrounded the oscillatory portion of most grains (Fig. 6G). Analyses that clearly fell within portions of crystals that exhibited oscillatory zoning were concordant and produced a concordia age at 456.1 ± 6.7 Ma (Fig. 7F). Analyses that were from (or overlapped) dark rims generally showed a high percent discordancy, had high REE concentrations, and were anomalously young. No reproducible age could be resolved from these rims.

Sparse zircon from the metatuff was light champagne color to almost colorless and 150–200 µm long with aspect ratios of 2:1–3:1. Zircons were often “ragged” or cracked with oscillatory zoning (Fig. 6H). Although several analyses were reversely discordant, these zircon grains yielded a ²⁰⁶Pb/²³⁸U weighted mean age of 455.0 ± 11.0 Ma (2σ; Fig. 7G).

Analyzed zircon grains were light brown to clear and 100–300 µm long with aspect ratios of ∼4:1. Zircon grain zonation was variable; some grains had rims with oscillatory zoning and cores with patchy or irregular zoning in CL, but others showed the reverse relationship (Fig. 6I). Oscillatory zoning occurred rarely throughout the entire grain. Core and rim analyses yielded a concordia age of 453.8 ± 3.4 Ma (2σ; Fig. 7H).

Analyzed zircon grains were euhedral, pink to light brown, and 150–300 µm long and had aspect ratios of ∼3:1. CL imaging showed distinct cores, and both zircon cores and rims had oscillatory zoning in images (Fig. 6J). Zircon cores and rims produced a concordia age of 450.1 ± 4.1 Ma (2σ; Fig. 7I).

The sample contained zircons that were euhedral, pink to light brown, and 150–300 µm long, and had aspect ratios around 3:1. Narrow high-U outer rims and distinct cores were observed, but rare, and oscillatory zoning was common in CL images (Fig. 6K). Cores and rims yielded a concordia age of 448.0 ± 5.1 Ma (2σ; Fig. 7J).

Zircons from a garnet-bearing trondhjemite in the northern lobe of the Alstead dome were light-brown to champagne colored with aspect ratios around 1:2–1:3. In CL images, the zircons were oscillatory zoned with dark rims, and some zircons contained inclusions (Fig. 6L). Zircon analyses were reversely discordant but yielded a ²⁰⁶Pb/²³⁸U weighted mean age of 447.5 ± 4.9 Ma (2σ; Fig. 7K).

Zircon grains were clear to dark pink and euhedral, with typical aspect ratios of around 3:1. Dark inclusions were common. CL imaging showed cores that ranged from dark and unzoned to oscillatory zoned (Fig. 6M). Zircon rims always contained oscillatory zoning. Zircon analyses were reversely discordant but yielded a ²⁰⁶Pb/²³⁸U weighted mean age of 445.9 ± 6.0 Ma (2σ; Fig. 7L).

Analyzed zircon grains were brown, subhedral, and 50–100 µm long and had aspect ratios of 2:1. Zircon grains were typically dark in CL and lacked oscillatory zoning (Fig. 6N). Zoning was typically patchy or nonexistent. Analyses yielded a ²⁰⁶Pb/²³⁸U weighted mean age of 445.2 ± 6.7 Ma (2σ; Fig. 7N).

The new U-Pb zircon ages for the Bronson Hill arc in New Hampshire (presented here) range from ca. 475 Ma to ca. 440 Ma. These ages overlap with U-Pb zircon ages from the Shelburne Falls arc, which have an age of range of ca. 500 Ma (north) to ca. 430 Ma (south; Fig. 8; Table 2, and references therein). In the Bronson Hill arc, Rankin et al. (2013) reported an age of 492.5 ± 7.8 Ma from a tonalite that intrudes the Albee Formation in northern New Hampshire, and other relatively old ages have been reported from the Joslin Turn pluton (ca. 469 Ma) in New Hampshire, and the Boil Mountain Complex (ca. 477 Ma) in Maine (Moench and Aleinikoff, 2003). Although our new data and the previously reported ages show that, statistically, the Shelburne Falls arc has an older median age, the age range overlaps between the Bronson Hill arc and the Shelburne Falls arc. This overlap questions the conclusion that the arc segments can be distinguished by age (Karabinos et al., 1998, 2017; Ratcliffe et al., 1998). The collisional events related to the Ordovician Taconic orogeny culminated by ca. 455–445 Ma in southwestern New England (Hames et al., 1991; Ratcliffe et al., 1998). Recent ⁴⁰Ar/³⁹Ar data from detrital white mica from low-grade rocks in the Connecticut Valley trough of southern Quebec preserve an eroded record of the Taconic event at ca. 445–443 Ma (Perrot et al., 2018). In southern Quebec, ⁴⁰Ar/³⁹Ar cooling ages from amphibole and muscovite range from 469 to 455 Ma (Tremblay and Pinet, 2016). The new ages of volcanic and plutonic rocks from this study of the Bronson Hill arc predate or are synchronous with this time interval. Earlier arguments by Tucker and Robinson (1990) and Karabinos et al. (1998) suggested that Bronson Hill arc rocks were too young to have been involved in the Taconic orogeny. However, as described by Ratcliffe et al. (1998), this idea was based on the original hornblende cooling age (465 Ma) of Sutter et al. (1985), which was old, not on the younger cooling ages (ca. 445 Ma) of Hames et al. (1991) from southwestern New England, or the range of cooling ages (472–455 Ma) from Tremblay and Pinet (2016). In a revised model, Karabinos et al. (2017) and Macdonald et al. (2017) called for reversal of subduction polarity from eastward- to westward-dipping between ca. 466 and 455 Ma, accompanied by development of the Bronson Hill arc at ca. 455–440 Ma. Our new data, combined with previous ages, show that the Bronson Hill arc was active well before 455 Ma and can also be explained without the need for polarity reversal, in agreement with Stanley and Ratcliffe (1985), Leo (1991), and Hollocher et al. (2002). Westward subduction was also proposed by Sevigny and Hanson (1995) in western Connecticut. Moench and Aleinikoff (2003) also called on a subduction polarity reversal, citing Karabinos et al. (1998), and a perceived 3–5 m.y. magmatic hiatus between ca. 460 and 456 Ma; that hiatus no longer exists in the data.

The general evolution of chemical composition of rocks shows a trend toward more calc-alkaline granitic compositions, decreasing with age, and consistent with evolution of an arc system (Leo, 1991). Older intrusions tend to be more tonalitic or trondhjemitic, while younger rocks typically are granitic (Fig. 9; Table 2). Work by Leo (1991), Dorais et al. (2008, 2012), and Hollocher et al. (2002) showed that there was also a spatial component to compositional variation in the Bronson Hill arc, from tonalitic and granodioritic in the south to mostly granitic in the north. The compositional trend by age is recognized in both the Bronson Hill arc and Shelburne Falls arc. We suggest that the compositional trend by age was caused by the arrival of the Shelburne Falls and Bronson Hill arc system on the Laurentian margin. Granitic compositions are the likely result of greater input of continental detritus into the subduction zone during Taconian collisional tectonics and partial subduction of the Laurentian margin rocks during the obduction of the Shelburne Falls arc–Bronson Hill arc onto Laurentia. This is consistent with the findings of Tucker and Robinson (1990), who reported that younger Bronson Hill arc rocks in Massachusetts were built on or near the Laurentian margin. Taconian sedimentary rocks in New York contain ca. 450 Ma ash layers (Sell et al., 2015; Macdonald et al., 2017) with inherited Laurentian zircons. The magmas of the erupted rocks may have migrated through Laurentian crust (Macdonald et al., 2017). Alternatively, due to the refractory nature of zircon, detrital zircon in sediments can survive subduction and be inherited in newly generated igneous rocks (Xu et al., 2018). If this latter interpretation remains viable, it obviates the need for westward subduction beneath Laurentian crust. The deposition of flysch of the Walloomsac Formation along the Middle Ordovician unconformity further constrains the timing of Taconian collisional events (Zen, 1961; Zen and Ratcliffe, 1971; Rodgers, 1971, 1985; Jacobi, 1981; Walsh et al., 2004). Late Middle to early Late Ordovician fossils constrain the age of the Walloomsac Formation in New York and correlative Ira Formation in Vermont (Zen, 1961; Potter, 1972; Ratcliffe, 1974; Zen et al., 1983; Finney, 1986; Ratcliffe et al., 1999). In western Connecticut, the age of thrust emplacement is constrained by the Blackriveran age (ca. 454 Ma) of the Walloomsac Formation (Ratcliffe et al., 1999), which lies tectonically beneath the allochthonous rocks (Walsh et al., 2004). The Brookfield Gneiss, located just east of the Taconic root zone (Cameron’s Line), has yielded intrusive ages of 453 ± 3 Ma (Sevigny and Hanson, 1995) and 453 ± 6 Ma (Walsh et al., 2004). Stitching plutons of the Cortlandt-Beemerville magmatic belt in this area yielded similar U-Pb zircon ages, including the Hodges Complex (446 ± 7 Ma), Mount Prospect Complex (451 ± 5 Ma), Bedford Complex (452 ± 5 Ma), Peach Lake Complex (450 ± 4 Ma), and Cortlandt Complex (446 ± 2 Ma; Ratcliffe, 1968, 1981; Ratcliffe et al., 1982, 2012). Thus, the paleontologic and geochronologic data suggest that the final transport of the composite accreted rocks, at least in western Connecticut, occurred after ca. 454 Ma, and may have been completed by ca. 453–450 Ma (Walsh et al., 2004). In Figure 9, we note this as the approximate time of terminal crustal thickening, after which the plutons were either more evolved (granites and granodiorites) or mafic, with the latter perhaps related to delamination or sublithospheric tears and mantle upwelling (Ratcliffe et al., 2012). Earlier, more juvenile magmas were produced when the Shelburne Falls arc–Bronson Hill arc was still outboard of Laurentia, and largely oceanic crust was being subducted (e.g., Leo, 1991; Hollocher et al., 2002). Supported by isotope geochemistry from the Bronson Hill arc, Dorais et al. (2008, 2012) concluded that observed chemical trends were the result of a subduction polarity switch; however, Nd and Pb isotope compositions reflect backarc geochemistry in the Ammonoosuc Volcanics, while younger plutons of the Oliverian Plutonic Suite (ca. 450 Ma) have a Laurentian signature. We suggest that these chemical trends may have been inherited from the Laurentian margin and its derived sediments, which were partially subducted and participated in melt production of younger Bronson Hill arc magmas (Hollocher et al., 2002). In modern arc environments, subducted sediment contributions may have a significant impact on the isotopic compositions of generated melts (Labanieh et al., 2012).

Since the publication of Tucker and Robinson (1990), it has been suggested that the Bronson Hill arc plutons are too young to have been a part of the Taconic orogeny (e.g., Karabinos et al., 1998). This conclusion led to a variety of models that used the Shelburne Falls arc as the driver of the Taconic orogeny and relied on a subduction zone flip to explain the presence of the Bronson Hill arc (Karabinos et al., 1998, 2017; Moench and Aleinikoff, 2003; Dorais et al., 2008, 2012; Macdonald et al., 2014, 2017). With the discovery that Taconic deformation was in places 10–15 m.y. younger than originally proposed, the subduction polarity flip and development of two separate arcs may no longer be required. Tremblay and Pinet (2016) reported evidence for a potential subduction polarity reversal in eastern Quebec, but the timing of this event is <425 Ma, after the magmatic activity in the Bronson Hill arc and Shelburne Falls arc. Zagorevski et al. (2010) concluded that the leading edge of Ganderia (the Popelogan-Victoria arc) did not dock with Laurentia until 455 Ma, and that Gander proper did not reach the Laurentian margin until ca. 430 Ma. We present new data that show most pluton ages are older than the age of Taconic peak metamorphism (ca. 445 Ma; Hames et al., 1991), negating the interpretation that the Bronson Hill arc could not have been part of the Taconic orogeny. However, it is also true that the overlap in ages of the Bronson Hill arc and Shelburne Falls arc on its own does not necessitate that they constitute a single composite arc that formed over the same east-dipping subduction zone (Stanley and Ratcliffe, 1985; Leo, 1991; Ratcliffe et al., 1998; Hollocher et al., 2002; Tremblay and Pinet, 2016), and so the subduction zone polarity reversal model remains plausible. Additional data are necessary to test the competing interpretations, and Dorais et al. (2012) made an important contribution to the isotopic study of the Bronson Hill arc rocks; however, more isotopic data are clearly needed (see Labanieh et al., 2012 for example). Below, we discuss the existing models and our own.

In our model, the Bronson Hill arc and Shelburne Falls arc make up a modified composite arc system that is potentially continuous at depth, buried beneath the Silurian–Devonian rocks of the Connecticut Valley trough and Mesozoic Hartford basin (Figs. 2 and 3). Eastward subduction produced the Ammonoosuc Volcanics and older tonalite and trondhjemite intrusions in the Shelburne Falls arc and Bronson Hill arc (Leo, 1991; Hollocher et al., 2002). The present geographic distribution of ages suggests that subduction may have been toward the southeast, because the Shelburne Falls arc–Bronson Hill arc system appears to young in that direction (Fig. 8; Karabinos et al., 2017). This may partially account for the apparently older average age of the Shelburne Falls arc when compared to the average age of the Bronson Hill arc (Figs. 9B and 9C); however, we urge caution when interpreting a lack of data. However, younger ages are present in western Connecticut. It is possible that these younger ages in the Shelburne Falls arc of Connecticut are part of the Bronson Hill arc that was pulled to the west by the opening of the Connecticut Valley trough or Hartford basin (see Fig. 2), that they represent late- to posttectonic stitching plutons that postdate arc formation (Ratcliffe, 1968, 1981; Ratcliffe et al., 1982, 2012), or that they represent plutonic events above a west-dipping subduction zone (Sevigny and Hanson, 1993, 1995; Karabinos et al., 1998, 2017). We note, too, that in adjacent Canada, the debate continues regarding the role and timing of eastward or westward subduction near the end of the Taconic orogeny (De Souza et al., 2014; van Staal et al., 2015; Tremblay and Pinet, 2016), and resolution of this debate remains for future work.

We propose that slab subduction shallowed during the Taconic orogeny, with greater input of continental detritus and the partial subduction of the Laurentian margin (Fig. 10). Crustal thickening would have led to a cessation of arc magmatism (Karlstrom et al., 2014) closest to the Laurentian margin and a migration of granitic plutonism southeastward, toward the Bronson Hill arc, where more-evolved granite plutons are found. This interpretation is consistent with the chemical change observed from largely island-arc and backarc signatures in the Ammonoosuc Volcanics to continental-arc signatures by ca. 443 Ma (Moench and Aleinikoff, 2003; Dorais et al., 2008, 2012). By the time of Silurian–Devonian sedimentation, the accretionary complex, including the Moretown Formation, the Shelburne Falls arc, and the foreshortened basement slices, had already been transported westward over Mesoproterozoic basement and Ediacaran to Lower Paleozoic cover rocks on the Laurentian margin (Ratcliffe et al., 1998, 2011, 2012). The Connecticut Valley trough developed after the Taconic orogeny and during Silurian extension and transition to a foreland basin setting in the Devonian Acadian orogeny (e.g., McWilliams et al., 2010; Perrot et al., 2018). The basin formation led to the present appearance of two partially buried, separate volcanic arcs on either side of the Connecticut Valley trough, especially in northern New England and Quebec.

Ages alone do not conclusively prove the existence of only a single arc. Northern Appalachian volcanic arcs built on a Ganderian crustal fragment in the Iapetus Ocean have overlapping ages, including the Penobscot arc-backarc system (513–482 Ma), the Tetagouche backarc (473–455 Ma), and the Popelogan-Victoria arc (475–455 Ma), the latter of which is the on-strike correlative of the Bronson Hill arc (Fig. 1; van Staal and Barr, 2012; van Staal et al., 2016). Our zircon study focused on obtaining reliable igneous ages, and suspected inherited cores were intentionally not targeted. Our data do include, however, a single zircon grain at ca. 573 Ma. The provenance of this inherited grain is unknown, and it may have been derived from either a peri-Gondwanan (Ganderian) or Laurentian source rock. The age is common in the peri-Gondwanan terranes (Meinhold et al., 2014) and overlaps with the age of Laurentian rift volcanics from the Pinney Hollow Formation in Vermont, dated at 571 ± 5 Ma (Walsh and Aleinikoff, 1999), and it is close to the age of the Pound Ridge Granite Gneiss (562 ± 5 Ma) and Yonkers Gneiss (563 ± 2 Ma) in the Manhattan prong (Tollo et al., 2004). Similarly, the inherited zircon grain in this study corresponds to detrital zircon ages in the Moretown Formation of Vermont and other Ganderian rocks in New Brunswick and Maine (Macdonald et al., 2014; Fyffe et al., 2009). Based on new detrital zircon data, Macdonald et al. (2014) presented evidence that the Iapetan suture, or Red Indian Line, is located west of the Shelburne Falls arc between the Moretown and Rowe/Stowe Formations. It should be noted that the Red Indian Line is almost certainly a zone with variations and complications along its length. In the absence of older ages from the Bronson Hill arc, this model still relied on the collision of the Shelburne Falls arc as the major driver of the Taconic orogeny. Our data are consistent with the interpretation that the suture between peri-Gondwanan–sourced rocks and Laurentian-sourced rocks is below the Moretown Formation and to the west of both the Shelburne Falls arc and the Bronson Hill arc (Macdonald et al., 2014). The Shelburne Falls arc–Bronson Hill arc system shares a similar history that can most easily be explained by an arc system that was built above an east-dipping subduction zone (Stanley and Ratcliffe, 1985; Leo, 1991; Hollocher et al., 2002). The polarity reversal model (Karabinos et al., 1998, 2017) may still be plausible; however, additional data, such as inherited zircon cores and comprehensive isotope geochemistry, especially from the Shelburne Falls arc, remain for future work. In southern Vermont, the position of the Red Indian Line may correlate with the contact between the Rowe Schist and the Moretown Formation, called the East Dover fault (Ratcliffe and Armstrong, 1999). In central and northern Vermont, the fault between the Stowe and Moretown Formations was not named (Ratcliffe et al., 2011). Walsh and Ratcliffe (1994) called the fault between the Stowe and Moretown Formations the Raymond Hill fault. It has also been mapped as the Townshend thrust (Ratcliffe et al., 1997), and the Keyes Mountain thrust at the north end of the Chester dome (Fig. 3; Ratcliffe, 2000).

Moench and Aleinikoff (2003) proposed that magmatism was not continuous, but that there was quiescence between 460 and 455 Ma. The chemistry of the ca. 443 Ma Quimby Formation intrusive and volcanic rocks changed to a continental-arc signature from the largely island-arc and backarc signatures in the underlying Ammonoosuc Volcanics (Moench and Aleinikoff, 2003; Dorais et al., 2008, 2012). In this model, magmatism in the Ammonoosuc Volcanics ceased, and the subduction polarity switched, as evidenced by the magmatic hiatus. In the two-arc model, subduction reversal from east to west was needed to build a supposedly younger Bronson Hill arc. Our new data set, plus published results, now shows that there is no gap in ages during this time period (Fig. 9) and that, with the discovery of older ages in the Bronson Hill arc, there may be no need for subduction reversal. In a simpler model, the Shelburne Falls arc–Bronson Hill arc system was built above an east-dipping subduction zone over a period of at least 70 m.y., with most ages spanning ∼26 m.y. between 475 Ma and 449 Ma (Fig. 9C). Most magmatism had ceased prior to ca. 430 Ma (428 ± 2 Ma Pumpkin Ground orthogneiss [Sevigny and Hanson, 1993]; 435 ± 3 Ma Tunnel Brook granite [Moench and Aleinikoff, 2003]; Table 3). After that, Silurian extension began (Dorais et al., 2017) and continued to at least 419 Ma, based on the age of the Comerford Intrusive Complex (Rankin et al., 2007) and the bimodal Braintree Intrusive Complex (Black et al., 2004; Ratcliffe et al., 2011). These mafic and felsic dikes and plutons intruded the Ammonoosuc Volcanics east of the Connecticut Valley trough (Walsh, 2016) and the Moretown and Cram Hill Formations west of the Connecticut Valley trough (Ratcliffe et al., 2011). The Connecticut Valley trough in Vermont and New Hampshire contains Silurian to Devonian volcanic and metasedimentary rocks with U-Pb zircon ages indicating deposition between ca. 432 and 407 Ma (Aleinikoff and Karabinos, 1990; Rankin and Tucker, 2009; McWilliams et al., 2010; Dorais et al., 2017). Extension led to the formation of the Connecticut Valley trough and split the Bronson Hill and Shelburne Falls arcs, perhaps as the result of a pull-apart basin over lithospheric delamination and asthenospheric upwelling (Black et al., 2004; Tremblay and Pinet, 2005, 2016; McWilliams et al., 2010). Given the current extent of geochronologic data in New England, it appears that widespread volcanism in the Ammonoosuc Volcanics and overlying Partridge and Quimby Formations was ending by around 445 Ma (Tucker and Robinson, 1990; Moench and Aleinikoff, 2003). This is the same time as postcollisional stitching plutons from the Cortlandt-Beemerville magmatic belt in New York and Connecticut (Ratcliffe, 1968,1981; Ratcliffe et al., 1982, 2012).

The precise age range of the Ammonoosuc Volcanics has some uncertainty. Zircon is not present in every sample, and some collected samples were barren (Table 2). Separated zircon grains tend to be small and poorly formed in the Ammonoosuc Volcanics (Figs. 6F and 6H). This leads to difficulty in SHRIMP analysis and may produce larger errors for U-Pb ages (Fig. 7G). In our study area, Oliverian Plutonic Suite rocks intruded both the upper and lower Ammonoosuc Volcanics. Parts of the Ammonoosuc Volcanics must be older than the Plainfield tonalite (474.8 ± 5.2 Ma) and as young as the Alstead dome metavolcanic sample (A400; 455.0 ± 11 Ma). The Plainfield tonalite is the oldest documented plutonic rock to intrude the Ammonoosuc Volcanics, predating the Joslin Turn tonalite by ∼5 m.y. (469 ± 2 Ma; Moench and Aleinikoff, 2003). Much of the Ammonoosuc Volcanics remains undated, and this age range may change with further analysis. Ages in the Ammonoosuc Volcanics now overlap with ages of the Oliverian Plutonic Suite. This supports the conclusion that much of the plutonism and volcanism was contemporaneous over an extended time (Leo, 1991). SHRIMP spot analyses on single CL zones yielded overlapping ages between our uppermost Ammonoosuc sample (A400) and at least the Alstead dome pluton (A1065). These data confirm the results of multigrain thermal ionization mass spectrometry data from Massachusetts and New Hampshire (Tucker and Robinson, 1990), which showed that the plutonic suite and Ammonoosuc Volcanics were coeval. These overlapping ages have been interpreted to suggest the existence of a fault between the plutons and the volcanic carapace based on metamorphic pressure and temperature differences between the Oliverian Plutonic Suite and the Ammonoosuc Volcanics (Kohn and Spear, 1999). However, our 1:24,000 scale mapping showed no evidence of this hypothetical fault in the Alstead dome, where granitoid dikes intrude the host rocks along the margin of the pluton. We suggest that the Alstead pluton is a subvolcanic intrusive rock, and the contemporaneity of the pluton and the metatuff do not require major fault offset. Regardless, the conclusion is that plutonism and volcanism overlap within the Bronson Hill arc.

New U-Pb zircon ages from the Oliverian Plutonic Suite and the Ammonoosuc Volcanics in the Bronson Hill arc overlap with ages from the Shelburne Falls arc. Zircon ages, rock types, and geochemical evolution of the Bronson Hill arc and Shelburne Falls arc are similar. The simplest conclusion is that the Shelburne Falls arc and Bronson Hill arc are parts of one long-lived composite arc system, and so the tectonic evolution of New England does not require a change in subduction polarity. Our data are consistent with the interpretation that the main suture (Red Indian Line) between the Laurentian margin and a peri-Gondwanan volcanic arc lies to the west of the Shelburne Falls–Bronson Hill composite arc, at the base of the Moretown Formation (Macdonald et al., 2014). Collision of the composite Shelburne Falls–Bronson Hill arc system with Laurentia was responsible for the Taconic orogeny. Magmatic activity continued beyond the end of collisional tectonics until ca. 430 Ma (428 ± 2 Ma Pumpkin Ground orthogneiss [Sevigny and Hanson, 1993]; 435 ± 3 Ma Tunnel Brook granite [Moench and Aleinikoff, 2003]; Table 3). Subsequent basin formation of the Connecticut Valley trough led to the present appearance of two partially buried, apparently separate, volcanic arcs on either side of the Connecticut Valley trough.

We thank John Aleinikoff, Peter Thompson, Peter Robinson, Rory McFaddon, and Robert Wintsch for helpful discussions. This manuscript benefited from U.S. Geological Survey reviews by Nicholas Ratcliffe, John Aleinikoff, and Randall Orndorff. We thank the associate editor and three reviewers for constructive review comments for this journal. Jorge Vazquez and Matthew Coble assisted with the sensitive high-resolution ion microprobe–reverse geometry (SHRIMP-RG) analyses at Stanford University. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. government.