Abstract

Several terranes of variable tectonic affinity and history underlie the central Appalachian Piedmont Province (eastern United States). These terranes mostly consist of widespread metasedimentary and lesser metavolcanic rocks. Intense and pervasive deformation and metamorphism have made the depositional ages and provenance of sediment in these rocks difficult to determine. The lack of tight constraints on such basic information led to a century-long debate about the tectonic significance of these rocks, particularly how they correlate to similar rocks along and across strike in the Appalachian orogen. We address these issues using U/Pb isotopic ages from single spots in 2433 zircon grains from 18 metasedimentary rock samples distributed across the Maryland Piedmont.

The resulting age signatures indicate that the Marburg Formation and Prettyboy Schist, heretofore assigned to the Westminster terrane, actually belong to the Potomac terrane, making the Hyattstown thrust the contact between the two terranes. Ediacaran Laurentia could have supplied all Potomac terrane sediment except for the detritus in one sample from the northern part of the terrane that likely came from Amazonia. This is one of the first recognitions of a Gondwana-derived terrane between Carolinia to the south and Ganderia to the north. Maximum depositional ages for Potomac terrane suprasubduction zone sedimentary rocks are latest Neoproterozoic or early Cambrian, and some may have been deposited ca. 510 Ma. Continental rifting ended ca. 560 Ma at the longitude of our study, so the transition from rifting to subduction at this location in eastern Laurentia may have lasted only 50 M.y. Lower Ordovician arc intrusions into these rocks demonstrate that the transition lasted no longer than 90 M.y. The Iapetan margin of central-eastern Laurentia was one of the shortest lived passive margins that formed in Neoproterozoic time.

INTRODUCTION

The Appalachian-Caledonian orogen is the type locality of the Wilson cycle, the first location where geologists recognized repeated creation and destruction of ocean basins between continents (Wilson, 1966; Bird and Dewey, 1970). This recognition was critical for development of the supercontinent cycle concept. Different parts of the Appalachians record two complete supercontinent cycles, from magmatic arc growth and collision that produced Rodinia and Pangea through supercontinent-destroying rifting and subsequent passive margin development. These juxtaposed and superimposed archives make the Appalachians an ideal place to compare similar processes that affected the same portion of a continent at different times. For example, in this paper we contrast the tempo of the transition between rifting cessation and subduction initiation in eastern Laurentia/North America following rifting in latest Neoproterozoic versus Late Triassic–earliest Jurassic time. Such constraints on the rates of tectonic events inform our general understanding of the supercontinent cycle and plate tectonics (e.g., Korenaga, 2006; Bradley, 2008).

Regionally, the portion of the Appalachian orogen between New York City and Virginia is of interest because it lies between Ganderia to the north and Carolinia to the south (Fig. 1; all directions are in present-day coordinates). These blocks were peri-Gondwanan terranes that accreted to Laurentia in early Paleozoic time (Hibbard et al., 2006; van Staal et al., 2009; Pollock et al., 2012). Accretion of these terranes caused orogeny on the eastern margin of Laurentia, documented by deformation, metamorphism, magmatism, and basin formation in the northern and southern Appalachians. Piedmont rocks of the central Appalachians also record early Paleozoic deformation, metamorphism, and magmatism (Drake, 1985b, 1989; Aleinikoff et al., 2002; Kunk et al., 2005; Southworth et al., 2007; Horton et al., 2010; Wintsch et al., 2010). However, current syntheses do not show Gondwanan terranes exposed in the central Appalachian Piedmont Province, raising the question, What was the tectonic cause of early Paleozoic orogeny in the central Appalachians? In this paper we test the interpretation that Gondwanan terranes do not crop out in the central Appalachians using U/Pb isotopic ages of detrital zircon to establish the provenance of the sediment that became the Piedmont rocks of Maryland, northern Virginia, and Washington, D.C.

We also use the detrital zircon ages for two other purposes. First, although most apparently are not exotic to Ediacaran Laurentia, multiple terranes with different histories compose the central Appalachian Piedmont. Detrital zircon ages allow us to probe which formations share depositional affinities, and thus where terrane boundaries lie. Second, because the depositional ages of these rocks are not well known, we use detrital zircon ages to provide constraints on their maximum possible depositional ages. Both the locations of terrane boundaries and the depositional ages are important for regional correlations and for understanding the tectonic evolution of this portion of the Appalachian orogen. For example, the Sams Creek Formation and surrounding rocks in the western Piedmont of Maryland were thought to have been deposited in the latest Neoproterozoic to early Cambrian (Southworth et al., 2007), but Graybill et al. (2012) suggested deposition in the earliest Neoproterozoic. This depositional age revision changes the interpreted tectonic setting from a basin produced by well-known rifting of eastern Laurentia ca. 570–560 Ma (Southworth, 1999; Southworth et al., 2009; Burton and Southworth, 2010) to a basin caused by putative rifting ca. 960 Ma (Graybill et al., 2012).

Similarly, the Sykesville and Laurel Formations in the eastern Piedmont were interpreted by most modern workers to be metamorphosed sedimentary rocks that were deposited in deep water at a convergent margin (Drake, 1985a; Muller et al., 1989; Pavlides, 1989; Drake and Froelich, 1997). Fleming and Self (2010) instead suggested that these rocks were mostly a thick pile of metamorphosed arc-related ignimbrites that erupted broadly coeval with intrusion of Lower Ordovician arc granitoids into the Sykesville and Laurel Formations. If this reinterpretation is correct, the succession represents a major volcanic arc terrane, whereas the conventional interpretation indicates trench or forearc basin fill. In this paper we use detrital zircon ages combined with field and thin section observations to assess the origin of the Sykesville and Laurel Formations.

GEOLOGIC SETTING

Geologists classically divide the southern and central Appalachian orogen into four tectonic provinces (Hatcher, 1989). Foreland basin strata deposited in the Paleozoic Era underlie the westernmost province, the Valley and Ridge Province; these rocks were faulted, folded, and cleaved during the Carboniferous–Permian Alleghanian orogeny (Hatcher et al., 1989; Hibbard et al., 2006). These strata were juxtaposed against the rocks that underlie the Blue Ridge Province to the east by faults; at the latitude of our study area the boundary fault is called the Keedysville Fault. The Blue Ridge Province exposes late Mesoproterozoic granitoids and related rocks, metamorphosed to conditions as high as granulite facies during latest Mesoproterozoic mountain building related to the Grenville orogeny (McLelland et al., 2010; Southworth et al., 2010). Rifting following the Grenville orogeny is represented by an initial pulse of felsic and mafic magmatism ca. 780–670 Ma (Tollo et al., 2004, 2012; Holm-Denoma et al., 2014; McClellan and Gazel, 2014) and the bimodal volcanism of the ca. 570–560 Ma Catoctin Formation, as well as felsic dikes as young as 555 ± 4 Ma (Southworth et al., 2009; Burton and Southworth, 2010). Rift or passive margin sedimentary rocks enclose the Catoctin Formation. All Blue Ridge Province rocks in the vicinity of our study area were metamorphosed to lower greenschist facies during late Paleozoic time (e.g., Kunk and Burton, 1999; Bailey et al., 2006; Southworth et al., 2007, 2009). The Piedmont Province lies directly east of the Blue Ridge Province across the Bull Run Mountain fault (Fig. 2). We discuss Piedmont rocks in detail in the next subsection. The easternmost province is the Coastal Plain, which is underlain by lower Cretaceous to Quaternary sediment deposited on the passive margin created by rifting of Africa from North America during opening of the Atlantic Ocean (Olsson et al., 1988; Edwards et al., 2010). Late Triassic to Early Jurassic rift basins superimposed on the older Piedmont rocks record initial rifting of these continents within Pangea (Weems and Olsen, 1997; Southworth et al., 2007), and Coastal Plain province strata likewise unconformably overlie Piedmont rocks (Olsson et al., 1988).

Piedmont Province Tectonostratigraphy

The Piedmont Province consists of a collection of fault-bounded terranes of variable tectonic affinity. In Maryland, pre-Triassic rocks of the Piedmont Province are divided from west to east into the Frederick Valley synclinorium, Sugarloaf Mountain anticlinorium, Westminster terrane, Potomac terrane, and Baltimore terrane (Figs. 2 and 3). The metamorphic grade in these rocks increases from lower greenschist facies in the west to amphibolite facies in the east. In the following paragraphs, we focus on the protoliths of the rocks of these terranes.

The lower and middle Cambrian Araby Formation is the stratigraphically lowest unit in the Frederick Valley synclinorium (Reinhardt, 1974, 1977). This metasiltstone and metashale formation is depositionally overlain by an upper Cambrian to lower Ordovician succession dominated by carbonate rocks (Mathews and Grasty, 1909; Reinhardt, 1974, 1977).

The Sugarloaf Mountain anticlinorium is composed of two units. The lower Cambrian Sugarloaf Mountain Quartzite protolith was dominantly quartz arenite and the conformably overlying Urbana Formation mostly consisted of muddy sandstone and siltstone (Jonas and Stose, 1938; Southworth, 1999). The Urbana Formation also contains discontinuous map-scale pods of marble surrounded by the siliciclastic lithologies.

The Martic thrust placed the Westminster terrane on top of the Frederick Valley synclinorium and the Sugarloaf Mountain anticlinorium. Westminster terrane strata may be the deeper water equivalents of contemporaneous supracrustal rocks of the Blue Ridge Province (Rodgers, 1970; Smoot and Southworth, 2014). The Westminster terrane includes the Ijamsville Phyllite, Sams Creek Formation, and Wakefield Marble (Mathews and Grasty, 1909; Rodgers, 1970; Southworth et al., 2007). The protoliths for the Ijamsville Phyllite were dominantly shale, siltstone, and sandy mudstone, some of which was tuffaceous (Mathews and Grasty, 1909; Jonas and Stose, 1938). The Ijamsville Phyllite also contains other metamorphosed supracrustal lithologies such as sandstone and basalt (Mathews and Grasty, 1909; Southworth et al., 2007). The Sams Creek Formation protoliths included a large variety of supracrustal lithologies, including the metasandstone sampled for this study (Mathews and Grasty, 1909; Southworth et al., 2007). However, the Sams Creek Formation is dominated by mostly mafic metavolcanic rocks. The metabasalt flows contain pillows and are chemically similar to metabasalt of the Catoctin Formation in the Blue Ridge Province (Southworth, 1999). The Wakefield Marble was probably deposited near Sams Creek Formation volcanic islands, and consists of both calcitic and dolomitic marble as well as sparse interbedded metamorphosed sandstone and shale (Mathews and Grasty, 1909; Jonas and Stose, 1938; Fisher, 1978). Depositional ages for the Westminster terrane formations are not well known, but usually are assumed to be late Neoproterozoic to Cambrian (Southworth et al., 2007).

A thrust fault placed the Potomac terrane on the Westminster terrane. Previously, the Pleasant Grove fault was identified as the terrane boundary because the Marburg Formation and Prettyboy Schist were included in the Westminster terrane (Drake, 1985b, 1989; Horton et al., 1989; Southworth et al., 2007). However, based on our new detrital zircon ages, we include these two units in the Potomac terrane; this new assignment makes the Hyattstown thrust the western boundary of the terrane. Metasedimentary rocks of the Potomac terrane are divided into the Marburg, Mather Gorge, Northwest Branch, Oella, Laurel, and Sykesville Formations and the Prettyboy and Loch Raven Schists (Southworth et al., 2007). Marburg Formation protoliths were dominantly siltstone and subordinate muddy sandstone (Jonas and Stose, 1938; Drake, 1994) whereas Prettyboy Schist protoliths were finer grained (Crowley, 1976). Mather Gorge Formation protoliths were dominated by feldspathic arenite and wacke as well as sandy mudstone (Drake and Froelich, 1997); these rocks are interpreted to be turbidites (Southworth et al., 2007). The Mather Gorge Formation also contains meter-long to several kilometer–long metamorphosed mafic and ultramafic blocks that do not appear to be bounded by major faults (Drake, 1989; Southworth et al., 2007). Kunk et al. (2005) divided the Mather Gorge Formation into three domains with different 40Ar/39Ar cooling ages; from west to east these are the Blockhouse Point, Bear Island, and Stubblefield Falls domains. The Northwest Branch Formation protoliths were dominantly arenite (Drake, 1998). The Oella Formation protoliths likewise were dominated by arenite, but they were interbedded with mudstone at the centimeter to meter scale (Crowley, 1976). The protolith for the Loch Raven Schist was dominantly shale (Crowley, 1976). Protoliths of the Laurel and Sykesville Formations were dominated by diamictite consisting of pebbles to boulders with a wide range of compositions surrounded by a matrix of feldspathic muddy sandstone (Hopson, 1964; Muller et al., 1989; Pavlides, 1989). Gravel-sized clasts in both formations include quartzite, schist, and phyllite; the Sykesville Formation also contains clasts of milky quartz, granite, and metamorphosed mafic and ultramafic rocks. Depositional ages of the Potomac terrane metasedimentary rocks are not well known but are usually considered to be latest Neoproterozoic to Cambrian (Southworth et al., 2007). Unlike the more western Piedmont terranes, the Potomac terrane contains Paleozoic plutonic rocks that intruded all formations except the Marburg and Prettyboy. These felsic, arc-related magmas mostly crystallized in the Ordovician Period, with some additional Late Devonian granitic plutonism (Aleinikoff et al., 2002; Horton et al., 2010). The Potomac terrane also includes the Soldiers Delight Ultramafite, a thrust-bounded serpentinite body that separates the Sykesville Formation to the west from the Laurel Formation to the east (Drake, 1994).

The Baltimore terrane (Williams and Hatcher, 1982) sits structurally beneath the Potomac terrane, juxtaposed by a series of thrusts. The oldest rocks exposed in the Baltimore terrane are felsic and minor mafic gneisses collectively called the Baltimore Gneiss (Williams, 1892); the volcanic protoliths of some of the felsic rocks crystallized ca. 1250 Ma (Aleinikoff et al., 2004). This metavolcanic and metasedimentary succession was intruded by several late Mesoproterozoic granitic plutons, including foliated biotite granite that crystallized ca. 1075 Ma (Aleinikoff et al., 2004). Seven antiformal domes on the northern and western sides of the city of Baltimore expose the Mesoproterozoic rocks in their cores (Mathews, 1907; Hopson, 1964; Fisher and Olsen, 2004). The Setters Formation was deposited nonconformably on these metaigneous rocks and partially or completely rings each dome in map view (Williams, 1891; Knopf and Jonas, 1923; Hopson, 1964). The Setters Formation is dominated by metamorphosed feldspathic muddy sandstone, siltstone, and shale (Hopson, 1964; Fisher, 1971) and is depositionally overlain by the Cockeysville Marble (Williams, 1892; Mathews and Grasty, 1909; Choquette, 1960). The Setters Formation and the underlying Mesoproterozoic rocks were intruded by granitic plutons in early to middle Paleozoic time (Fisher and Olsen, 2004). The depositional ages of the Setters Formation and Cockeysville Marble are not well known but usually are inferred to be latest Neoproterozoic or Cambrian (Williams, 1891; Higgins, 1972; Fisher and Olsen, 2004). On the southeastern side the metaigneous rocks and their metasedimentary carapace were overthrust by a metamorphosed Cambrian mafic-ultramafic complex (Sinha et al., 1997).

Summary of Depositional Settings for Piedmont Province Rocks

The Frederick Valley synclinorium and the Westminster terrane each contain carbonate successions at least several hundred meters thick, and the Urbana Formation within the Sugarloaf Mountain anticlinorium also contains calcareous rocks and beds of marble. The Sugarloaf Mountain Quartzite is predominantly metamorphosed quartz arenite and conglomerate. Similarly, the Setters Formation is depositionally overlain by the Cockeysville Marble, which had a depositional thickness of as much as 120 m (Knopf and Jonas, 1923). These lithologies support the interpretation that the depositional setting for all these rocks was shallow water covering a continental shelf and slope (Reinhardt, 1977; Southworth et al., 2007). The prevalence of mafic volcanic and volcaniclastic rocks in the Westminster terrane suggests deposition in a rift setting, likely related to the rifting of eastern Laurentia recorded by the 570–560 Ma Catoctin Formation in the Blue Ridge Province directly to the west (Southworth, 1999).

In contrast, rocks of the Potomac terrane contain nearly no carbonate strata and the Mather Gorge Formation contains turbidites, both pointing to deposition in deep water outboard of the continental shelf. The wide range of clast types in the Sykesville Formation suggests deposition near a tectonically active continental margin; furthermore, if the crystallization age of the igneous clasts was near the depositional age of the diamictite, then the active margin had magmatic activity. The scarcity of bimodal volcanic or volcaniclastic strata argues against a rift setting for deposition of any part of the Potomac terrane (cf. Alvaro et al., 2008; Ayalew and Gibson, 2009; Corti, 2009; Zhou et al., 2009). Deposition in a trench or forearc basin outboard of the continental margin would explain the wide range of clast types, the paucity of volcanic or volcaniclastic strata, the presence of the turbidites, and the nearly complete lack of bedded carbonate. Map-scale mafic and ultramafic blocks occur within the Mather Gorge Formation, and the Laurel and Sykesville Formations surround a fault-bounded ultramafic body, the Soldiers Delight Ultramafite (Fig. 2). These mafic and ultramafic blocks could have been tectonically emplaced at depth in a subduction channel (e.g., Cloos, 1982; Hernaiz Huerta et al., 2012; Aoya et al., 2013). We argue that the most likely tectonic scenario for both deposition of the Potomac terrane sedimentary rocks and emplacement of the kilometer-scale mafic and ultramafic blocks is an ocean-continent subduction zone. Other workers likewise argued for deposition of the Mather Gorge, Sykesville, and Laurel Formations at an ocean-continent convergent margin (Drake, 1985a; Muller et al., 1989; Pavlides, 1989; Drake and Froelich, 1997). The formation of the subduction zone involving oceanic lithosphere east of Laurentia did not end sedimentation in the Frederick Valley synclinorium, Blue Ridge Province, and Valley and Ridge Province during the Cambrian and into the Ordovician (Fig. 3; Reinhardt, 1974, 1977; Southworth et al., 2007; Smoot and Southworth, 2014).

METHODS

We collected samples from pre-Triassic Piedmont Province metasandstone-bearing formations exposed in Maryland (Figs. 2 and 3; Table 1). In addition, we collected two samples (810001 and 310001) in Virginia and one (909003) in Washington, D.C. Each sample came from the coarsest grained, least micaceous sandstone we identified in each outcrop. Photomicrographs of thin sections from these sandstones are shown in Figure 4 and Supplemental Figure 11. In some outcrops in the eastern Piedmont, granitoids intruded the metasedimentary rocks; in these cases we collected samples far from the intrusions and avoided rock pieces with visible veins. We did not sample the Oella Formation of the Potomac terrane because we were unable to find an outcrop that was not pervasively intruded by granite. Samples from all formations except the Araby Formation (Frederick Valley synclinorium) and Loch Raven Schist (Potomac terrane) yielded abundant zircon. Details of the zircon separation and analysis procedure are given in Appendix 1. All uncertainties are given at the two standard deviation level.

Determination of the maximum possible depositional age of a sedimentary rock based on its youngest detrital zircon grains requires careful interpretation because of the confounding effects of lead loss, multiple age zones in a single grain, growth of metamorphic zircon, analytical uncertainty, and dating a very small fraction of the zircon population present in a rock unit. Dickinson and Gehrels (2009a) showed that using the youngest single age from a suite of detrital zircon grains can yield a maximum depositional age determination that is younger than the true depositional age, so they recommended using multiple analyses with ages that overlap within uncertainty. Accordingly, we take the maximum possible depositional age to be the weighted mean of the 2 or more youngest analyses that overlap at the 1σ level plus the 2σ error on the weighted mean, rounded up to the nearest 10 M.y. interval to avoid spurious significant figures.

Laurentia consisted of an Archean core surrounded by younger terranes, some of which may have originated far from Laurentia (Whitmeyer and Karlstrom, 2007). Regardless of their origin, after accretion to the continent these terranes became part of Laurentia; therefore, we consider a Laurentian sediment source to be a source in Laurentia as the continent existed at the time of deposition. Because the sedimentary rocks considered here were deposited during or after the Ediacaran Period, we refer to the continent as Ediacaran Laurentia when discussing sediment provenance.

RESULTS

Supplemental Figure 22 shows histograms and relative probability plots for individual samples; Figure 5 summarizes the dating results from all samples. We combine the Sugarloaf Mountain anticlinorium and Westminster terrane into a western group of rocks because their detrital zircon age distributions are nearly identical. The interpretation that some Westminster terrane rocks are deeper water equivalents of Sugarloaf Mountain anticlinorium strata supports this combination (Rodgers, 1970).

Most Westminster terrane and Sugarloaf Mountain anticlinorium samples contain zircon with ages only between 1350 and 900 Ma, including uncertainties; these ages define a peak on the individual relative probability plots at 1100–1050 Ma. The Sugarloaf Mountain Quartzite sample (908001) also yielded two 1700–1600 Ma grains and the northern Sams Creek Formation sample (1010001) yielded one ca. 680 Ma and two ca. 560 Ma zircon grains. The probability density plot for the combined ages from the five western group samples displays a single spike ca. 1070 Ma (Fig. 5C).

Although the Setters Formation crops out geographically east of all the other formations, its zircon age distribution is similar, but not identical, to those from the Westminster terrane and Sugarloaf Mountain anticlinorium rocks. Zircon from the western Setters Formation sample (809002) has ages only between 1300 and 900 Ma and produces a prominent peak at 1020 Ma and a subordinate peak at 1170 Ma. The distribution of ages from the eastern Setters Formation sample (909007) is nearly identical except for the addition of two ca. 1350–1300 Ma grains and one ca. 700 Ma grain. Differences between Setters Formation and western group age distributions include the following. (1) Both Setters Formation samples produced two age peaks, at 1170 and 1020–1010 Ma, whereas all western group samples yielded only one main peak, at 1100–1050 Ma. (2) Sugarloaf Mountain Quartzite sample 908001 yielded two 1700–1600 Ma grains, but zircon of this age was not found in either Setters Formation sample. (3) The eastern Setters sample yielded zircon with moderately discordant 206Pb/207Pb ages of 882 ± 26, 881 ± 44, and 725 ± 104 Ma, but the western Setters Formation sample and all western group rocks did not produce similar ages.

Potomac terrane rocks contain zircon with a greater range of ages. All samples of these rocks have zircon in the age range 1350–900 Ma, like the western group and Setters Formation samples. However, in all Potomac terrane samples the range of ages extends beyond 1350 Ma, to 1700 or 1800 Ma in most samples and as old as 2100 Ma in the Laurel Formation sample, including uncertainties. All Potomac terrane samples except the Blockhouse Point Domain sample (810001) additionally contain populations of (1) upper Archean zircon at 2850–2600 Ma, and (2) 750–500 Ma zircon. The zircon ages from most Potomac terrane samples produce peaks on individual relative probability plots at 1480–1440 Ma and 1200–1030 Ma; on the compilation plot there are prominent peaks ca. 1470 and 1160 Ma and a minor peak ca. 1650 Ma (Fig. 5D).

The detrital zircon age distribution in the northern Mather Gorge Formation sample (1010002) is quite different from the ages in all the other Potomac terrane samples, and it is unlike those in our other Piedmont samples (Fig. 5). This sample produced a nearly continuous distribution of ages within the range 2650–520 Ma (including uncertainties), with wide gaps only between ca. 2450–2210 Ma and 1700–1550 Ma. The age distribution does not exhibit the major Mesoproterozoic age peaks found in all other Piedmont samples, but instead has a single prominent spike in age probability ca. 630 Ma.

INTERPRETATIONS AND IMPLICATIONS

Terrane Affinity of the Marburg Formation and Prettyboy Schist

Our Marburg Formation samples (910004, 910005, 909005) and our Prettyboy Schist sample (910002) yielded detrital zircon with age ranges that are nearly identical to those found from the other Potomac terrane samples and are unlike the Westminster terrane rocks (Fig. 5; Supplemental Figure 2 [see footnote 2]). We therefore include the Marburg Formation and Prettyboy Schist in the Potomac terrane, making the Hyattstown thrust the boundary between the Potomac and Westminster terranes (Fig. 2). This assignation defines the Piedmont terranes based on shared characteristics that were set at the time of initial rock formation, meaning deposition for sedimentary rocks or crystallization for igneous rocks. We do not use attributes of the rocks that formed later, during deformation or metamorphism, to group rock units into terranes.

Depositional Ages

Potomac Terrane

Maximum possible depositional ages for most Potomac terrane rocks range from 1030 to 980 Ma (Table 1). Although all Potomac terrane samples except sample 810001 from the Blockhouse Point domain of the Mather Gorge Formation yielded at least 1 analysis younger than 600 Ma, only 3 samples produced at least 2 such grains with ages that overlap at the 1σ level (Table 1; Supplemental Table 13). The Laurel (sample 808003) and northern Sykesville (sample 909001) Formations accordingly have maximum possible depositional ages of 530 and 550 Ma, respectively. The northern Mather Gorge Formation (sample 1010002) has a maximum possible depositional age of 540 Ma.

Setters Formation, Westminster Terrane, and Sugarloaf Mountain Anticlinorium

The maximum depositional ages obtained from the Setters Formation samples are 1000 and 910 Ma, slightly younger than the 1040–1020 Ma maximum depositional ages for most of the formations in the Sugarloaf Mountain anticlinorium and Westminster terrane (Table 1). Analyses from the northern Sams Creek Formation sample (1010001) produced an exception to the late Mesoproterozoic maximum depositional ages for the other Westminster terrane rocks because this sample yielded grains with 206Pb/238U ages of 681 ± 16, 563 ± 12, and 554 ± 22 Ma (Supplemental Table 1 [see footnote 3]). A metamorphic origin for the zircon in the Sams Creek Formation is unlikely because maximum metamorphic conditions reached only lower greenschist facies (Wintsch et al., 2010). Although some Sams Creek Formation zircon grains have thin rims visible in cathodoluminescence images, we did not intersect these rims with the laser beam during analysis; all analyses targeted the cores of the grains. We interpret these three young grains to be detrital. The two youngest analyses give a maximum possible depositional age of 580 Ma. These analyses are concordant and show no evidence for lead loss or mixing of multiple age zones during laser ablation (although the 206Pb/238U age and especially the 206Pb/207Pb age are imprecisely determined for the youngest analysis because of low U concentration). The ages of the two youngest grains are consistent with partial derivation of Sams Creek Formation sediment from the 570–560 Ma Catoctin Formation and related rift volcanic rocks (Southworth et al., 2009; Burton and Southworth, 2010). Southworth (1999) tied deposition of the Sams Creek and Catoctin Formations by showing matching major and trace element chemistry of metabasalt from each formation.

Our interpretation that the Sams Creek Formation was deposited after 580 Ma conflicts with the depositional age proposed by Graybill (2012) and Graybill et al. (2012), who dated 1 spot in each of 18 zircon grains from a phyllite within the Wakefield Marble adjacent to the Sams Creek Formation and found 206Pb/207Pb ages ranging from 1321 ± 20 to 956 ± 32 Ma. Graybill (2012) and Graybill et al. (2012) interpreted the phyllite to be a metamorphosed tuff and used the three youngest ages (964 ± 28, 961 ± 26, and 956 ± 32 Ma) to infer a depositional age of 970–950 Ma for the tuff. Graybill (2012) and Graybill et al. (2012) applied this depositional age to the entire Sams Creek Formation and Wakefield Marble. Deposition of the Sams Creek Formation at 970–950 Ma cannot be reconciled with the new detrital zircon results from our northern Sams Creek Formation sample. The most likely cause of the discrepancy is that the phyllite collected by Graybill (2012) was actually a metasedimentary rock, and the zircon grains in the rock were detrital. Nearly all well-studied tuffs contain little or no xenocrystic zircon (Brown and Fletcher, 1999; Reid and Coath, 2000; Brown and Smith, 2004; Charlier et al., 2005; Simon and Reid, 2005; Zhang et al., 2007; Simon et al., 2008; Zou et al., 2010; Tollo et al., 2012; but see Page and Laing, 1992), but 15 of 18 zircon grains from the Sams Creek putative metatuff crystallized tens or hundreds of millions of years before the inferred eruption age. Although such a large fraction of older zircon grains is rare for tuffs, it is common for sedimentary rocks to contain a large proportion of detrital zircon grains that crystallized hundreds of millions of years before deposition. The 1320–950 Ma age spectrum from the alleged metatuff closely matches the relative probability distributions of detrital zircon ages found in our Sugarloaf Mountain anticlinorium and Westminster terrane metasandstone samples, including the two Sams Creek metasandstones, as well as the distribution of detrital zircon ages in a metasandstone within the Wakefield Marble analyzed by Graybill et al. (2012). These comparisons with other tuffs and sandstones support our suggestion that the zircon grains from the putative metatuff sample were actually detrital grains in a metasedimentary rock.

Sykesville Formation Protolith

Fleming and Self (2010) argued that the Sykesville and Laurel Formations, as well as the correlative Indian Run Formation in northern Virginia, are dominated by metamorphosed ignimbrites and that deposition of these formations occurred adjacent to major silicic volcanic calderas. One of the main pieces of evidence presented by Fleming and Self (2010) is their reinterpretation of foliated, pebble-sized clasts with large aspect ratios as metamorphosed flattened pumice lapilli with flame structures (Fig. 4A). All of these clasts that we examined are dominantly muscovite (Fig. 4B). Features so rich in aluminum and potassium are more consistent with an origin as mudstone clasts than as dacitic or rhyolitic pyroclasts (e.g., Hess, 1989), supporting the conventional interpretation that the Sykesville Formation is dominated by metasedimentary rocks. Furthermore, only 1 of 133 zircon grains from the Laurel Formation sample (808003) and 1 of 329 zircon grains from the 2 Sykesville Formation samples (909003, 909001) yielded U/Pb ages within 20 M.y. of the inferred ignimbrite eruption age of 475–450 Ma, but nearly all well-studied tuffs are dominated by zircon that crystallized within a few million years prior to eruption (references in previous subsection). In contrast, the distributions of zircon ages in the Laurel and Sykesville Formations closely match the distributions in other samples from the Potomac terrane that are interpreted to be metasedimentary rocks, such as the Marburg and Mather Gorge Formations and the Prettyboy Schist (Supplemental Figure 2 [see footnote 2]; also see Horton et al., 2010). These comparisons with both tuffs and other metasedimentary rocks in the Potomac terrane likewise support the conventional interpretation that the protolith of the Sykesville Formation was sedimentary, not volcanic.

Sources of Sediment

Possible Source Continents: Laurentia, Amazonia, Rio de la Plata, and West Africa

According to Li et al. (2008), the Amazonia and Rio de la Plata cratons were positioned against eastern Laurentia following the middle Neoproterozoic breakup of the supercontinent Rodinia. Because we do not address other blocks within Pannotia (Powell, 1995; Nance et al., 2014), we refer to this joined Laurentia–Amazonia–Rio de la Plata group of continental blocks as the LAR group. Alternatively, only a sliver of the Amazonia craton, the Arequipa-Antofalla block, may have remained adjacent to Laurentia after middle Neoproterozoic time (Escayola et al., 2011). Following latest Neoproterozoic to earliest Cambrian rifting, the Iapetus Ocean separated Laurentia from the Amazonia (plus Arequipa-Antofalla), Rio de la Plata, and West Africa cratons (Pollock et al., 2012). Accreted peri-Gondwanan blocks along strike to the north and south of the Maryland Piedmont were derived from the margins of Amazonia (Ganderia, Avalonia, and Carolinia) and West Africa (Meguma and Suwanee) (van Staal et al., 2009; Pollock et al., 2012). Thus it is possible that these Gondwanan cratons and/or peri-Gondwanan terranes, in addition to Ediacaran Laurentia, provided sediment to the Maryland Piedmont basins in latest Neoproterozoic and early Paleozoic time.

Figure 6 shows that Ediacaran Laurentia contained igneous rocks with a wide range of crystallization ages, and zircon growth during metamorphism overlaps and extends beyond the igneous crystallization ages. Three post-Archean gaps stand out: 2500–2000 Ma was a period of little metamorphism or felsic magmatism in Laurentia except for the Wopmay orogen on the northwestern corner of the continent (Hildebrand et al., 2010; Hoffman et al., 2011); 950–780 and 670–580 Ma also were times of little metamorphism or felsic magmatism in Laurentia. The only well-known Laurentian exceptions are two Iapetus rift-related plutons in western Newfoundland that crystallized at 617 ± 8 and 602 ± 10 Ma (Williams et al., 1985; van Berkel and Currie, 1988). The Long Range dikes of western Newfoundland and eastern Labrador also crystallized ca. 615 Ma, but they are dominantly mafic and contain little zircon (Kamo et al., 1989). Smith (2003) reported a crystallization age of 602 ± 2 Ma for a felsite dike from the Reading Prong of eastern Pennsylvania. However, this age determination came from thermal ionization mass spectrometry of a single mechanically abraded zircon grain, and multiple grains are required to ensure repeatability and thus a robust age. Furthermore, the volume of lithologically similar felsite in this region is small. Although the Goochland terrane in the central Virginia Piedmont contains several small granitic plutons that crystallized ca. 660–580 Ma (Owens and Tucker, 2003), it is unknown whether this tiny continental fragment was part of Laurentia by the beginning of the Cambrian Period.

The Amazonia, Rio de la Plata, and West Africa cratons likewise contain igneous and metamorphic rocks with a wide range of Archean and Proterozoic crystallization ages, but important differences in their geologic history allow recognition of distinctive potential sediment sources. Unlike Laurentia, the Amazonia and Rio de la Plata cratons contain abundant felsic igneous rocks with crystallization ages between ca. 2250 and 2020 Ma (Rapela et al., 2007; Buenano Macambira et al., 2009; Brito Neves, 2011). The Goias magmatic arc lies in the early Cambrian collision zone between Amazonia and the Sao Francisco craton to the southeast. Igneous crystallization ages in the Goias arc range from ca. 930 to 610 Ma, with important peaks in tectonic activity near 900 and 630 Ma (Laux et al., 2005; Moura et al., 2008; Matteini et al., 2010), although the Sao Francisco craton and superjacent Goias arc did not join with Amazonia until ca. 550–510 Ma (Trindade et al., 2006; Moura et al., 2008; McGee et al., 2012; Tohver et al., 2012). The Rio de la Plata craton is bounded by rocks that record 850–750 Ma rifting and 650–600 Ma metamorphism (Rapela et al., 2011; Tohver et al., 2012). Post-Mesoarchean West Africa craton rocks crystallized throughout the periods 2750–1750 and 760–550 Ma, with a notable gap between ca. 1700 and 1000 Ma (summarized in Abati et al., 2010).

Sediment Sources for Maryland Piedmont Rocks

The discussion in the following subsections concerns the ultimate sources of zircon in Piedmont metaclastic rocks, not actual sediment transport pathways at the time of deposition. That is, some detrital zircon grains in the Piedmont rocks could have been stored as clasts in sedimentary rocks or xenocrysts in igneous rocks for a period between initial zircon crystallization and subsequent deposition in Piedmont basins.

All units except the northern Mather Gorge Formation.Figure 6 shows that known sources in southern and central Ediacaran Laurentia could have produced all 2303 detrital zircon grains analyzed from our 17 samples (not including the northern Mather Gorge Formation sample). The main age peak in all these samples is 1250–950 Ma. The ca. 1250–1020 Ma part of this range closely matches the crystallization ages of the abundant felsic plutons that compose the Grenville Province now exposed in inliers directly to the west and the ca. 1050–950 Ma part corresponds to the age of the final Grenvillian metamorphism of these rocks (McLelland et al., 2010; Southworth et al., 2010; Tollo et al., 2010, 2012). The northern Appalachian portion of the Grenville Province also experienced granitic intrusion between ca. 1050 and 950 Ma (Rivers, 1997; McLelland et al., 2010). In Potomac terrane rocks, there is another important age group between 1500 and 1300 Ma with a peak ca. 1470 Ma, but few analyses between 1500 and 1600 Ma. These ages closely match the crystallization ages of both the Granite-Rhyolite Province rocks and the 1450–1350 Ma granitoids that intruded much of central Laurentia through eastern Canada. The main age peaks in our Piedmont Province samples are similar to age signatures from upper Neoproterozoic and lower Cambrian metasandstone deposited on Blue Ridge Province Proterozoic rocks (Fig. 7; Carter et al., 2006; Hebert et al., 2010; Satkoski et al., 2012; Satkoski, 2013), supporting our interpretation that Ediacaran Laurentia supplied the Piedmont sediment (not including the northern Mather Gorge Formation sample).

Laurel Formation sample 808003 contained two grains that yielded 206Pb/238U dates of 518 ± 10 and 511 ± 10 Ma. The uncertainties from these two low U/Th analyses overlap at the 1σ level, suggesting that the dates could reflect predeposition crystallization ages rather than lead loss or metamorphic zircon growth after deposition of the Laurel Formation (cf. Dickinson and Gehrels, 2009a). If so, the possible sources for these zircon grains may be limited to central New Mexico or south-central Colorado (western United States). Felsic igneous rocks with crystallization ages between 525 and 510 Ma are rare in southern Laurentia, but granite and syenite of this age are exposed in central New Mexico and south-central Colorado (Schoene and Bowring, 2006; Amato and Mack, 2012). Spencer et al. (2014) found abundant ca. 520 Ma detrital zircon in the upper part of the Cambrian Van Horn Sandstone in nearby western Texas, supporting the interpretation that the ca. 518 and 510 Ma grains in the Laurel Formation came from this part of Laurentia. Alternatively, these two young grains could have been derived from the ca. 580–530 Ma rift-related magmatic rocks in eastern Ediacaran Laurentia; the detrital grain ages could be slightly younger than the magmatic ages due to a small amount of lead loss from the detrital zircon.

Of the 2303 detrital zircon grains analyzed, only one yielded an age between 2600 and 2100 Ma: a grain from Prettyboy Schist sample 910002 gave a 206Pb/207Pb age of 2367 ± 74 Ma. Three other grains yielded ages in or near the range 2100–2000 Ma: 2025 ± 62 Ma from Laurel Formation sample 808003, 2005 ± 14 Ma from northern Sykesville Formation sample 909001, and 1994 ± 20 Ma from southern Sykesville Formation sample 909003. Similarly, the 17 samples yielded only 12 of 2303 zircon grains that crystallized between 900 and 600 Ma. The paucity of detrital zircon that crystallized in the periods 2600–2000 and 900–600 Ma is compatible with derivation from Ediacaran Laurentia but inconsistent with derivation of abundant sediment from Amazonia, Rio de la Plata, or West Africa. The predominance of Mesoproterozoic zircon in all of our Piedmont samples rules out West Africa as a major source of sediment.

Although the sediment for all the Piedmont rocks except the northern Mather Gorge Formation could have been derived from Ediacaran Laurentia, Potomac terrane sources within Ediacaran Laurentia were different than the sediment sources to the Westminster terrane, Sugarloaf Mountain anticlinorium, and Setters Formation basins. The sources of detritus to these latter basins were nearly completely restricted to the Grenville Province and its overlying rift deposits. The only exceptions are two grains from Sugarloaf Mountain Quartzite sample 908001 that yielded 206Pb/207Pb ages of 1691 ± 44 and 1638 ± 66 Ma. In contrast, the ultimate sources for Potomac terrane sediment were nearly every pre-Ordovician rock suite in southern and central Laurentia (Fig. 6), plus possibly the Wopmay orogen in northwestern Laurentia. The detrital zircon ages from the Piedmont rocks also differ sharply from the age populations in Ediacaran and Cambrian sandstone deposited on the northern margin of Laurentia, which commonly contain a prominent detrital zircon age peak ca. 1850 Ma (Beranek et al., 2013).

Northern Mather Gorge Formation. The distribution of detrital zircon ages acquired from northern Mather Gorge Formation sample 1010002 is unlike the detrital zircon ages from all of our other Piedmont Province samples (Figs. 5 and 7). Unlike the other samples, the northern Mather Gorge sample produced a prominent peak in ages ca. 630 Ma and a nearly continuous distribution of ages within the range 2650–520 Ma (including uncertainties), with wide gaps only between ca. 2450–2210 and 1700–1550 Ma. 52% of the grains from this sample crystallized in the range 800–500 Ma, whereas in our other Piedmont samples only 1% crystallized during this interval. These age disparities indicate very different sediment sources for the northern Mather Gorge Formation compared to the remainder of the Maryland Piedmont. Two end-member scenarios can explain the provenance differences. (1) Northern Mather Gorge Formation detritus was derived from Ediacaran Laurentia, but from different parts of the continent than the sediment in the remainder of the Maryland Piedmont rocks. The 2500–2000 Ma zircon in the northern Mather Gorge Formation came from the Wopmay orogen and the 800–500 Ma grains came from rift-related rocks. The many detrital zircon ages that are outside the ranges of crystallization ages of rift-related igneous rocks reflect small magnitude lead loss from grains that actually crystallized at 780–670 or 580–530 Ma, or possibly ca. 617–602 Ma. In this scenario, the rift rocks contributed ∼50% of the detritus in the northern Mather Gorge Formation but only 1% in the other Maryland Piedmont rocks. (2) Unlike the rest of the Maryland Piedmont units, northern Mather Gorge Formation detritus was not derived from Ediacaran Laurentia. The most likely source continent in this scenario is Amazonia. The 1700–1000 Ma zircon in the northern Mather Gorge Formation rules out West Africa as a sediment source, and the Rio de la Plata and other Gondwanan cratons were within or on another side of the Gondwana supercontinent, not on the margin bordering the Iapetus Ocean (Pollock et al., 2012). Amazonia (Li et al., 2008) and Arequipa-Antofalla (Escayola et al., 2011) had separated from Laurentia by the time of deposition, so sediment transport alone cannot explain the unusual detrital zircon age signature of the northern Mather Gorge Formation. Instead, if option 2 is correct, the northern Mather Gorge Formation would be a fragment of an exotic terrane that originated near Amazonia and was tectonically emplaced on Laurentia in the Paleozoic.

We favor scenario 2 for the following reasons. (1) The northern Mather Gorge Formation zircon ages are very similar to detrital zircon age distributions from some rocks from the known peri-Gondwanan terranes Avalonia, Meguma, Carolinia, and Arequipa-Antofalla (Fig. 7). The distinctive features of the age distributions from all these samples are a dominant peak at 630–600 Ma and the absence of a major peak at 1200–1000 Ma. (2) In southern and eastern Laurentia, 670–580 Ma felsic igneous rocks were rare (Cawood et al., 2001; Whitmeyer and Karlstrom, 2007; Burton and Southworth, 2010), but they were common in West Gondwana (Laux et al., 2005; Moura et al., 2008; Abati et al., 2010; Matteini et al., 2010; Rapela et al., 2011; Tohver et al., 2012). (3) The small-volume Laurentian felsic rocks that crystallized ca. 617 and 602 Ma cannot have been the source for the abundant older detrital zircon that crystallized at 680–630 Ma. Furthermore, rift metasandstones in western Newfoundland with sediment sources partially in the 617–602 Ma granite bodies have their most prominent detrital zircon age peak at or before ca. 1000 Ma (Cawood and Nemchin, 2001), not ca. 630 Ma as for our northern Mather Gorge Formation sample (Fig. 5). One Newfoundland rift deposit, the South Brook formation, was deposited unconformably on the 602 ± 10 Ma Round Pond granite, yet only 1 of 52 zircon grains from a metasandstone from the South Brook formation yielded an age younger than 900 Ma (Cawood and Nemchin, 2001). Because 617–602 Ma granite contributed only a small fraction to the total zircon population in the proximal metasandstone samples, it is unlikely that these plutons would produce a major age peak in a sample 1500 km along strike. (4) For similar reasons, the Goochland terrane seems an unlikely source for the ca. 630 Ma detrital zircon grains. Ca. 1050 Ma intrusions are the volumetrically dominant Precambrian rocks in the Goochland terrane, and the areally most extensive Precambrian rock type, the ca. 1050 State Farm Gneiss, contains abundant zircon (Owens and Tucker, 2003). Therefore, we expect a major age peak ca. 1050 Ma in zircon derived from the Goochland terrane, as found in sediment largely derived from Blue Ridge Province Proterozoic rocks (Fig. 7; Carter et al., 2006; Satkoski et al., 2012; Satkoski, 2013).

If scenario 2 is correct, the northern Mather Gorge Formation is one of the first units in the mid-Atlantic sector of the Appalachian Piedmont identified as having received sediment from a Gondwanan source (see also Bailey et al., 2008; Bosbyshell et al., 2012; Hughes et al., 2014). As such, it is one of the first candidates for an exotic terrane in this region. Further data on the depositional age, provenance, and structural setting of this unit will aid understanding of its tectonic significance.

Comparison to Blue Ridge Cambrian Sandstone and Tectonic Setting During Deposition

U/Pb dating of detrital zircon from late Neoproterozoic and early Cambrian sandstone throughout the Blue Ridge Province shows that zircon ages from central Blue Ridge basins in Maryland, West Virginia, and northern Virginia are similar to our western group and Setters Formation spectra, whereas zircon ages from southern Blue Ridge basins are like our Potomac terrane ages (not including the northern Mather Gorge Formation; Fig. 7; Carter et al., 2006; Hebert et al., 2010; Satkoski et al., 2012; Satkoski, 2013). Figure 8 shows three end-member models to explain the different sediment sources and depositional settings for the latest Neoproterozoic to Cambrian basins of the Blue Ridge and Maryland Piedmont. Figure 8A shows deposition of the Piedmont terranes across strike of one another but at different times to allow different sediment sources. Figure 8B depicts deposition at approximately the same time but with the Potomac terrane along strike from the other basins, not directly outboard of the western group. This option implies later along-strike translation of the Potomac terrane to insert it between the Westminster and Baltimore terranes. A challenge to both models is the transport of sediment into relatively deep water past the continental shelf and then into the shallow water of the Setters basin on the Baltimore terrane. Figure 8C presents a solution to this problem by positing eastern sources for at least some Setters detritus. This option also explains the minor differences between the Setters Formation and western group detrital zircon age spectra: these sets of detrital zircon ages are broadly similar but differ in detail because different blocks of crust of similar age produced the sediment for the two basins.

Status of the Mather Gorge Formation

The three Mather Gorge Formation samples yielded disparate detrital zircon age spectra (Supplemental Figure 2 [see footnote 2]). The Bear Island domain ages are nearly identical to the ages from Potomac terrane rocks outside the Mather Gorge Formation. The Blockhouse Point domain age distribution is similar to that from the Bear Island domain in that both contain an age peak near 1450 Ma and a continuous distribution of ages between ca. 1700 and 900, but the spectra are dissimilar in that the Blockhouse Point domain sample did not yield ages outside this range. As described herein, detrital zircon from the northern Mather Gorge Formation produced an age spectrum that is so unlike the age distributions from all other Maryland Piedmont rocks that we conclude that the northern Mather Gorge Formation is part of an exotic terrane. These provenance differences indicate that the sampled units in the Mather Gorge Formation likely were not depositionally correlative and thus should not be combined into a single formation. Kunk et al. (2005) reached the same conclusion using muscovite 40Ar/39Ar ages from these rocks.

Tempo of Transition from Rifting to Subduction

Zircon U/Pb dating over the past ∼12 years has illuminated an important difference in the timing of subduction initiation following breakup of Pangea versus the LAR group. In the mid-Atlantic region, rifting of Pangea began in the Late Triassic Epoch (ca. 220 Ma) and basalt erupted in latest Triassic and earliest Jurassic time (ca. 200 Ma; Manspeizer et al., 1989; Weems and Olsen, 1997; Blackburn et al., 2013). Subduction has not yet started at this latitude, so the interval between the end of rifting and the beginning of subduction is at least 200 M.y. In contrast, flood basalt in the Catoctin Formation records final rifting of this portion of the LAR group ca. 570–560 Ma (Southworth et al., 2009; Burton and Southworth, 2010) and arc-related granitoids intruded the Sykesville Formation at 478 ± 6 and 472 ± 4 Ma (Drake and Fleming, 1994; Aleinikoff et al., 2002), giving a maximum interval of 90 M.y. between rifting cessation and subduction initiation.

The Potomac terrane metasedimentary rocks offer the opportunity to further refine constraints on the pace of this transition because the Mather Gorge, Sykesville, and Laurel Formations likely were sourced from Ediacaran Laurentia and deposited in an oceanic trench or forearc basin. If these rocks were deposited in Laurentian suprasubduction zone basins, a tighter constraint on the age of subduction initiation comes from their depositional ages than from the crystallization ages of the granitoids that intruded them. Our Laurel Formation sample (808003) yielded one 518 ± 10 and one 511 ± 10 Ma zircon, and it is possible that these grains reflect detrital input into the Laurel Formation rather than lead loss or metamorphic zircon growth. In many cases on convergent margins the crystallization age of the youngest detrital zircon is near the true depositional age (e.g., Grove et al., 2008; Dickinson and Gehrels, 2009a), suggesting that the period of transition from the end of rifting to the beginning of subduction actually may have lasted only 50 M.y.

Whether or not this last speculation is correct, it is clear that the tempo of the transition from the cessation of rifting to subduction initiation at the study latitude was much faster following breakup of the LAR group than Pangea (see also Waldron et al., 2014). Furthermore, the post-LAR group transition in eastern Laurentia at the study latitude is among the quickest known examples of conversion to subduction following rifting in the Neoproterozoic Era (Bradley, 2008). The unusual celerity of the eastern Laurentia case suggests a tectonic situation that was not present in most Neoproterozoic–early Paleozoic settings around the globe. More rapid subduction initiation in eastern Laurentia following Neoproterozoic compared to Late Triassic rifting also is opposite the global trend of slower transitions to subduction in the Proterozoic compared to the Phanerozoic eons (Bradley, 2008). Better timing constraints on Paleozoic events recorded in Piedmont rocks in Maryland and beyond are one avenue that may lead toward a fuller understanding of these differences, and insights on their causes will contribute to our knowledge of the processes involved in subduction initiation.

Philip Piccoli enabled our use of the electron microprobe. We acknowledge the support of the Maryland NanoCenter and its NispLab (Nanoscale Imaging, Spectroscopy, and Properties Laboratory), which is supported in part by the National Science Foundation (NSF) as a MRSEC (Materials Research Science and Engineering Center) shared experimental facility. We thank George Gehrels, Victor Valencia, Mark Pecha, and the staff of the Arizona LaserChron Center for facilitating our zircon analyses in their laboratory. The LaserChron Center is supported by NSF grant EAR-1032156. Irene Kadel-Harder and Rebecca Ohly provided able laboratory assistance. We thank Steven Whitmeyer for generously providing the digital files to produce Figure 6. Reviews by Luke Beranek, Todd Lamaskin, and an anonymous reviewer greatly improved the manuscript, and we thank associate editor Todd Lamaskin as well as science editor Raymond Russo for editorial handling.

APPENDIX 1. ZIRCON DATING METHODS

For all aspects of zircon dating we followed the procedures described in Gehrels et al. (2008) and Dickinson and Gehrels (2009a, 2009b). We isolated zircon using conventional mineral separation techniques, including rock pulverization by hand using a mortar and pestle, removal of silt and clay by hand-panning in water, removal of magnetic grains using a Frantz magnetic barrier separator, and density separation using methylene iodide. We then poured the zircon grains onto double-sided tape and cast them in an epoxy disk along with ∼6 shards of the Sri Lanka zircon standard (564 ± 3 Ma; Gehrels et al., 2008). After hand-polishing to expose the interiors of the grains, we produced backscattered electron and cathodoluminescence images using the JEOL JXA-8900R electron probe microanalyzer at the University of Maryland.

We dated the cores of ∼200 zircon grains from each sample by laser ablation–inductively coupled plasma–mass spectrometry in the Arizona LaserChron Center at the University of Arizona, taking care to avoid multiple cathodoluminescence zones, inclusions, and cracks. We used two different mass spectrometers for this dating. Samples dated in 2009 (see Table 1) were analyzed using a GV Instruments Isoprobe following the protocols given by Gehrels et al. (2008). The laser spot diameter was 35 μm. The other samples were analyzed using a Nu Plasma high-resolution multicollector mass spectrometer. Ablation of the zircon was performed using a New Wave UP193HE excimer laser and a spot diameter of 40 μm for the 2010 analyses and 30 μm for the 2011 analyses. The ablated zircon was carried in helium into the plasma source of the mass spectrometer, which is equipped with a flight tube of sufficient width that U, Th, and Pb isotopes were measured simultaneously. All measurements were made in static mode using Faraday detectors with 3 × 1011 ohm resistors for 238U, 232Th, 208Pb, 207Pb, and 206Pb and discrete dynode ion counters for 204Pb and 202Hg. Each analysis consisted of background measurement via a 15 s integration on peaks with the laser off followed by fifteen 1 s integrations with the laser firing. There was then a 30 s delay to purge the previous sample and prepare for the next analysis. Interference of 204Hg with 204Pb was addressed by measurement of 202Hg and subtraction of 204Hg using the natural 202Hg/204Hg ratio of 4.35. This mercury correction was not significant for most analyses because mercury backgrounds in the Nu Plasma mass spectrometer generally are low.

During analysis, inter-element fractionation of Pb/U usually is ∼5% ,whereas the apparent fractionation of lead isotopes typically is <0.2%. For analyses using both the Isoprobe and the Nu Plasma mass spectrometers, in-run analyses of shards of the Sri Lanka zircon standard were used to correct for this fractionation. The uncertainty from the calibration correction was 1%–2% (2s) for both 206Pb/207Pb and 206Pb/238U ages. Results of our standard analyses are provided in Supplemental Table 1 (see footnote 3) and a summary is available in Supplemental Table 24. Analyses of the Sri Lanka zircon standard also were used to determine the concentrations of uranium and thorium in each sample zircon. The standard zircon contains ∼518 ppm uranium and 68 ppm thorium.

For each analysis a correction for common lead was made using the 204Pb measurement (mercury corrected for the Nu Plasma analyses) and assuming an initial lead composition from Stacey and Kramers (1975). Uncertainties of 1.5 for 206Pb/204Pb and 0.3 for 207Pb/204Pb were applied to these lead composition values based on the variation in lead isotopic composition in modern crustal rocks.

These corrections and other data reduction were performed off-line using an Excel program developed at the Arizona LaserChron Center. We removed from further consideration analyses with: (1) high 204Pb, (2) low 206Pb/204Pb ratio, (3) >5% error on the 206Pb/207Pb date, (4) >10% error on the 206Pb/238U date, (5) >25% normal discordance or 5% reverse discordance, (6) high U concentration, or (7) high U/Th ratio. We use the remaining analyses for our interpretations (Supplemental Table 1 [see footnote 3]; Figs. 5 and 7; Supplemental Fig. 2 [see footnote 2]). We used Isoplot to calculate weighted means and to produce concordia plots (Ludwig, 2008).

206Pb/238U dates usually are more precise than 206Pb/207Pb dates for zircon younger than ca. 1.4 Ga, whereas the reverse is true for older grains. However, 206Pb/207Pb dates are only minimally affected by recent lead loss, so in most cases they more closely indicate the time of crystallization for zircon older than ca. 1 Ga. Thus during interpretation it generally is preferable to use 206Pb/238U dates for grains younger than 1 Ga and 206Pb/207Pb dates for older zircon. However, almost all of our samples contain zircon with a range of ages that spans this 1 Ga preferred cutoff value. Accordingly, we used 700 Ma as the cutoff in order to avoid an artificial break in the interpreted ages.

1Supplemental Figure 1. Photomicrographs of thin sections from each sample except 810001 from the Blockhouse Point Domain of the Mather Gorge Formation. Samples 211001 of the Loch Raven Formation and 909006 of the Araby Formation did not yield zircon. Thin sections were cut perpendicular to foliation and, when present, parallel to lineation. All images were acquired using transmitted, cross-polarized light. The scale is the same for all images. Mineral abbreviations: grt—garnet, hem—hematite, ms—muscovite, q—quartz. Please visit http://dx.doi.org/10.1130/GES01140.S1 or the full-text article on www.gsapubs.org to view Supplemental Figure 1.
2Supplemental Figure 2. Plots showing the spectrum of detrital zircon U/Pb ages from each sample. Please visit http://dx.doi.org/10.1130/GES01140.S2 or the full-text article on www.gsapubs.org to view Supplemental Figure 2.
3Supplemental Table 1. Uranium and lead isotopic data for the zircon grains from each sample. Standard analyses are appended to the end of the analyses from each sample. Please visit http://dx.doi.org/10.1130/GES01140.S3 or the full-text article on www.gsapubs.org to view Supplemental Table 1.
4Supplemental Table 2. Summary of uranium and lead isotopic data for the shards of the standard that were analyzed during dating of zircon from each sample. Please visit http://dx.doi.org/10.1130/GES01140.S4 or the full-text article on www.gsapubs.org to view Supplemental Table 2.