Redefinition of the Petersburg batholith and implications for crustal inheritance in the Dinwiddie terrane, Virginia, USA

The Appalachian orogen records a protracted history of accretionary orogenesis that ultimately culminated in the closure of the Rheic Ocean basin with continent-continent collision between Laurentia and Gondwana and the amalgamation of the supercontinent Pangea by middle Permian time (Hatcher, 2002). In the southern Appalachian orogen, competing tectonic models suggest that the leading lithotectonic elements of peri-Gondwanan affinity (Fig. 1A) docked with Laurentia either by the close of the Late Ordovician–Silurian Cherokee orogeny (e.g., Hibbard et al., 2007a) or by the Late Devonian–Early Mississippian Neoacadian orogeny (e.g., Merschat et al., 2017). In both models, late Silurian–Early Devonian time is considered to be a period of tectonic quiescence along the Laurentian margin. In contrast, voluminous and widespread syn- and post-tectonic magmatism occurred throughout the crystalline core of the southern Appalachian orogen during the late Paleozoic Alleghanian orogeny (e.g., Speer et al., 1994), including terranes of the Carolinia domain (Figs. 1A and 1B), which most researchers consider to be exotic to Laurentia (e.g., Hibbard et al., 2002; Hatcher, 2010).

The central-eastern Piedmont of Virginia, USA, is situated in a critical geographic location where peri-Gondwanan terranes are currently situated adjacent to the Goochland domain that is interpreted to represent the Iapetan realm (e.g., Hibbard et al., 2007b; Fig. 1). Rocks of the late Paleozoic Peterburg Granite (sensu lato) at the northeasternmost extent of the Carolinia domain in the southern Appalachian orogen are superposed by Mesozoic rift basins and covered, in part, by Cenozoic Atlantic Coastal Plain sediments (Fig. 1B). The Petersburg Granite (sensu lato) is currently interpreted to be one of the largest batholiths in the southern Appalachian orogen (e.g., Hibbard et al., 2006). However, despite its size, the Petersburg Granite (sensu lato) has received comparatively little attention in southern Appalachian magmatic studies (e.g., Samson et al., 1995) and has been generally overlooked in orogen-scale tectonic syntheses. Lack of attention to the Petersburg Granite (sensu lato) likely stems from the assumption that it was wholly composed of igneous rocks generated during the Alleghanian orogeny and thus had no bearing on the construction of the early to middle Paleozoic Laurentian margin.

In contrast to this long-standing assumption, recent geochronology suggests that the Petersburg Granite (sensu lato) may record a protracted geologic history that spanned ca. 400–300 Ma (Buchwaldt and Owens, 2012; Owens et al., 2017, 2019), indicating that the Petersburg Granite (sensu lato) has older components and suggesting that some of the rocks were, in fact, likely not generated during Alleghanian orogenesis and may contain information on the pre-Alleghanian infrastructure that existed at this northeasternmost point of the southern Appalachian crystalline core.

In this study, we report new sensitive high-resolution ion microprobe–reverse geometry (SHRIMP-RG) and laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) U-Pb zircon crystallization and detrital zircon geochronologic analyses. We couple these data with whole-rock geochemistry, petrogenetic analysis, and detailed and reconnaissance mapping in nine 1:24,000 scale quadrangles in the central-eastern Virginia Piedmont (Fig. 2) to determine when the Petersburg Granite (sensu lato) and its associated rocks were emplaced and if the varying documented intrusive facies record protracted assembly or are from distinct, multiple events that are genetically unrelated. Our new data require a redefinition of the Petersburg Granite (sensu lato); we propose that this crustal block is not simply a singular Alleghanian batholith but instead consists of Silurian–Devonian orthogneiss and Neoproterozoic to Cambrian metasedimentary and metavolcanic rocks intruded by a late Paleozoic igneous complex (i.e., the Petersburg Granite, sensu stricto). The Petersburg Granite (sensu lato) is better modeled as an upper-amphibolite–facies, infrastructural terrane of peri-Gondwanan affinity, which we here call the Dinwiddie terrane. The Dinwiddie terrane was amalgamated to Laurentia and subsequently deformed and intruded from middle to late Paleozoic time, spanning the Acadian to Alleghanian orogenies.

The Petersburg Granite (sensu lato) was first formally defined as a geologic unit on the 1928 geologic map of Virginia as a Precambrian coarsely crystalline porphyritic biotite granite that was intruded by finely crystalline granite and cut by pegmatitic granite (Nelson, 1928). Varieties of granite, including lightgray to dark-blue equigranular granite marketed for building stone, porphyritic granite, and granite gneiss, have been recognized and described throughout the eastern Piedmont of Virginia since the early 1900s (e.g., Watson, 1906), and early workers identified foliated and gneissic granite in the quarries located in the Richmond and Petersburg, Virginia, areas (Bloomer, 1939; Steidtmann, 1945). Detailed geologic mapping during the 1960s–1970s in the Richmond area distinguished two mappable units of granite: one consisting of fine- to coarse crystalline, uniform to porphyritic, foliated to non-foliated granite, granodiorite and minor quartz monzonite, and the second consisting of porphyritic granodiorite to quartz monzonite (Goodwin, 1970, 1980, 1981; Daniels and Onuschak, 1974). Regionally, Bobyarchick (1978) separated moderately to strongly foliated gray granite at the northern and southern ends of the “eastern pluton” of the Petersburg “batholith” and described other mapped granite within the pluton as weakly to moderately foliated. Bobyarchick and Glover (1979) described mylonitic foliation in granite on the northwestern edge of the pluton adjacent to the Hylas fault zone (Fig. 2), which is part of the Eastern Piedmont fault system (Hatcher et al., 1977; Bobyarchick, 1981) that breaks the eastern Piedmont into distinct crustal blocks.

Detailed mapping in the Richmond area (Carter et al., 2007a, 2010; Carter, 2010; Bleick et al., 2011; Bondurant et al., 2011; Occhi et al., 2017; Occhi and Swanger, 2019) identified and described four mappable units assigned to the Petersburg Granite (sensu lato). (1) Subidiomorphic (massive equigranular) granite (Fig. 3A) is mostly finely crystalline, with a texture that is typically hypidiomorphic-granular. (2) Porphyritic granite (Figs. 3A and 3B) is mostly coarsely crystalline, with pale-pink potassium feldspar phenocrysts up to 6 cm in length; these phenocrysts are commonly aligned. (3) Foliated granite (Fig. 3C) is finely to coarsely crystalline, with strong foliation defined primarily by aligned biotite ± muscovite. (4) Layered granite gneiss (Fig. 3D) is distinctly compositionally layered, with compositional layering defined by alternating biotite- and quartz-feldspar–rich layers that range from 1 cm to 1 dm thick. A fifth facies, megacrystic granite, which consists of coarse- to very coarsely crystalline, prism-shaped, pale-pink potassium feldspar phenocrysts up to 3 dm in length (Fig. 3E), was recognized more recently (Carter, 2010; Occhi et al., 2015). Mappable xenoliths of amphibolite, quartz-muscovite felsic schist, and rare meta-ultramafic rocks were also mapped (Fig. 3F). Based on field relations, foliated granite and granitic gneiss were interpreted to be an older pre- to syn-orogenic intrusive facies of the Petersburg Granite, whereas massive, porphyritic, and megacrystic facies were interpreted to be syn- to mostly post-orogenic (Carter, 2011).

Wright et al. (1975) reported the first U-Pb zircon age for Petersburg Granite (sensu lato) at 330 ± 8 Ma. Regional geochronology studies on other eastern Piedmont plutons (e.g., Kish and Fullagar, 1978, using the Rb-Sr biotite and/ or whole-rock technique) narrowed the distribution of the Petersburg Granite (sensu lato) to the easternmost Piedmont. The Petersburg Granite (sensu lato) was further confined to the belt from Ashland to Dinwiddie (Fig. 2) based on reconnaissance mapping and compilation (Rader and Evans, 1993), and its age was revised based on a SHRIMP U-Pb zircon age of 314.5 ± 2.1 Ma (Lee and Williams, 1993). Owens et al. (2019) recently reported high-precision chemical abrasion, isotope dilution–thermal ionization mass spectrometry (CA-ID-TIMS) U-Pb zircon ages of 299.87 ± 0.13 Ma and 296.04 ± 0.11 Ma for porphyritic and massive granite, respectively, in the Richmond area (Fig. 2). Owens et al. (2020) also reported a ca. 300 Ma age (U-Pb zircon and titanite using the laser ablation split stream [LASS] technique) for monzodiorite from the Richmond area. Owens et al. (2017, 2019) reported an age of ca. 400 Ma for a third sample from a xenolith of foliated granite within younger granite (granite gneiss of Carter et al., 2010) but speculated that the xenolith (location shown by an “x” in Fig. 2) may represent a fragment of Goochland terrane (terrane usage of Hatcher et al., 2007) as it is similar in age to certain lithologies within the Maidens Gneiss (Owens et al., 2010a).

Geochemical and isotopic studies have been undertaken to determine the generation, segregation, and sources of magma as well as ascent and emplacement mechanisms (e.g., Kish and Fullagar, 1978; Fullagar and Butler, 1979; Sinha and Zietz, 1982). Late Paleozoic Alleghanian plutons in the southern Appalachian orogen were formerly interpreted as remnants of a magmatic arc generated by either east-dipping subduction (Pindell and Dewey, 1982) or west-dipping subduction (Sinha and Zietz, 1982; Hatcher, 1987) of ocean crust along the eastern Laurentian margin. Speer et al. (1994) used field, petrologic, isotopic, and age data from ~60 plutons (including Petersburg Granite, sensu lato) to propose that melt generation was a result of tectonic and/or structural juxtaposition of high-temperature rocks against lower-temperature rocks, decompression melting, fluxing, or a combination of these processes due to transpressional arching or faulting. Samson et al. (1995) evaluated various tectono-petrogenetic models (e.g., subduction, crustal thickening, decompression melting, and lithospheric delamination) using bulk chemical and isotopic data from 35 southern Appalachian plutons (Petersburg Granite, sensu lato, not included). Their Nd, Sr, and Pb isotopic data demonstrated that most Alleghanian magmas were likely sourced from melting of the more juvenile terranes they intruded. Samson et al. (1995) also ruled out subduction to generate the melts, and preferred melt generation due to the insulating effect of stacked thrust sheets, or crustal anatexis caused by mantle lithospheric delamination. Coler et al. (1997) reported isotopic and geochemical data from 12 plutonic rocks (including three samples of Petersburg Granite, sensu lato) and five metamorphic rocks from eastern Piedmont terranes that are compatible with generation of magmas by anatexis of accreted juvenile crust at mid-crustal levels during continent-continent collision; their data supported east-dipping polarity for subduction during initial terrane accretion. In all of these and prior studies, Petersburg Granite (sensu lato) was treated as a singular, large batholith in the southern Appalachian central-eastern Piedmont of Virginia.

For this study, major- and trace-element geochemistry for 23 samples was determined by inductively coupled plasma–optical emission spectrometry (ICP-OES) for major elements and inductively coupled plasma–mass spectrometry (ICP-MS) or thermal desorption–inductively coupled plasma (TD-ICP) for trace and rare-earth elements (REEs) using Activation Laboratories Ltd. (Actlabs). Geochemical data from an additional 17 samples collected during previous mapping were compiled from Virginia Energy, Geology and Mineral Resources Program (VE-GMRP, formerly Virginia Department of Mines, Minerals and Energy, or VDMME). Samples were analyzed by Actlabs from 2006 to 2019 (data available upon request from VE-GMRP, Charlottesville, Virginia). Geochemical data were used to classify rock types, distinguish trends between and within map units, and define abundances of minor and trace elements (Figs. 4–7). Data are reported in Table S1 (see Supplemental Material¹). Polished thin sections were prepared for all geochemical and seven of eight geochronologic samples analyzed in this study. Mineral assemblages and textural relations (Fig. 8) for these samples were determined by optical petrography. A subset of the polished sections was analyzed by field emission–scanning electron microscope (FE-SEM) energy-dispersive X-ray spectroscopy (EDS) to determine their mineralogy. Analytical procedures of FE-SEM petrography are described in  Appendix 1. In addition, we reviewed petrographic reports from 36 VE-GMRP samples (data available upon request from VE-GMRP, Charlottesville, Virginia), including a foliated metagranitoid gneiss sample dated by the LA-ICP-MS U-Pb zircon method (sample VDMME R-11518). LA-ICP-MS U-Pb zircon analyses were conducted on two samples at the U.S. Geological Survey (USGS) G3 Plasma Laboratory at the Geology, Geophysics, and Geochemistry Science Center in Denver, Colorado, during two sessions in 2018 and 2019. Analytical procedures for sample preparation, characterization, and the LA-ICP-MS technique are described in  Appendix 2; data are presented in Table 1. ²⁰⁶Pb/²³⁸U ages are reported for zircon grains younger than 1300 Ma, and ²⁰⁷Pb/²⁰⁶Pb ages are used for zircon grains older than 1300 Ma following the recommendations of Gehrels (2012). SHRIMP-RG U-Pb zircon analyses were conducted on six samples at the USGS/Stanford SHRIMP-RG Laboratory in Stanford, California, during three sessions in 2018 and 2019. Analytical procedures for sample preparation, characterization, and the secondary ionization mass spectrometry technique are described in  Appendix 3 (analytical setup was identical for each session); weighted-average ages determined from the ²⁰⁷Pb-corrected ²⁰⁶Pb/²³⁸U spot ages are reported in Table 2, and concordia plots are constructed using ²⁰⁴Pb-corrected ²⁰⁷Pb/²³⁵U and ²⁰⁴Pb-corrected ²⁰⁶Pb/²³⁸Pb ages. For samples exhibiting overdispersion (p(χ2) >0.05), the uncertainty in weighted-average ages is expanded by the square root of the mean square of weighted deviates (MSWD)* a student’s t-factor (Vermeesch, 2018). Isotopic data for individual spots when listed in the text are ²⁰⁷Pb-corrected ²⁰⁶Pb/²³⁸U ages as reported in Table S2 (^{footnote 1}). In the text and in tables, all ratios and ages are reported at 2σ. Whole-rock geochemistry and SHRIMP and LA-ICP-MS isotopic data can also be found in an associated USGS data release (Carter et al., 2022a).

Late Paleozoic massive equigranular granite and porphyritic granite are mapped in two plutons—the northern Richmond pluton and the southern DeWitt-Sutherland pluton (Fig. 2). We assign massive equigranular granite and porphyritic granite in both plutons to the Petersburg Granite (sensu stricto). In previous work (e.g., Carter et al., 2007b), foliated granite and layered granite gneiss were also considered facies of the ~1500 km² composite late Paleozoic Petersburg Granite (sensu lato) batholith, similar to other composite late Paleozoic granites, such as the Rolesville batholith in North Carolina (Speer, 1994) and the Liberty Hill pluton in South Carolina (Speer et al., 1989). As shown in Figure 2, here we separate these rock types from the Petersburg Granite (sensu lato) and for simplicity combine them into a foliated metagranitoid gneiss unit—our informal Pocoshock Creek gneiss.

Massive equigranular granite constitutes the largest map unit by surface area and occurs as the dominant rock type in the Richmond and DeWitt-Sutherland plutons, which are separated by a septum of Pocoshock Creek gneiss (Fig. 2). Massive equigranular granite in the Richmond pluton is characterized by a typically hypidiomorphic-granular texture, with rare small (up to ~2 cm in length) potassium feldspar phenocrysts in a finely crystalline matrix of quartz, plagioclase, and biotite. Other primary minerals are hornblende and muscovite. Pyroxene is a very minor constituent of the rock, and garnet occurs sparingly. Massive equigranular granite in the DeWitt-Sutherland pluton (Fig. 2) is typically coarser crystalline than massive equigranular granite in the northern Richmond pluton, with an idiomorphic-granular texture (Fig. 3G). Potassium feldspar phenocrysts are rare, and biotite is the predominant mafic mineral phase. Quartz in the DeWitt-Sutherland pluton also tends to have a “smoky” or medium-gray (N5) color, which is less common in the Richmond pluton.

In many outcrops of massive equigranular granite in the Richmond pluton, aligned biotite plates impart weak foliation to the rock; other outcrops preserve cm-thick (0.4 in thick) layers of quartz + feldspar that are slightly coarser crystalline than surrounding quartz + feldspar + biotite granite. On the Cherry Hill 7.5-minute quadrangle (location shown in Fig. 2), foliation in massive equigranular granite of the DeWitt-Sutherland pluton consists of up to 3-cm-thick quartz and feldspar-rich layers that are separated by thin (0.5 mm) schlieren of biotite; this foliation is locally folded.

Porphyritic granite is a subordinate rock type and map unit in the Richmond and DeWitt-Sutherland plutons. In the Richmond pluton, porphyritic granite occurs primarily adjacent to the Mesozoic Richmond basin (Fig. 2). Porphyritic granite also occurs on the eastern flank of the DeWitt-Sutherland pluton farther south (Bobyarchick, 1978). Potassium feldspar phenocrysts in the DeWitt-Sutherland pluton are typically smaller than those in the Richmond pluton but are locally up to 10 cm in length (Owens et al., 2017; Occhi and Swanger, 2019). Biotite is a major mineral constituent of the rock. Muscovite is locally present as a primary magmatic mineral phase and as a reaction replacement mineral phase, typically after biotite.

Phenocrysts in porphyritic granite are aligned locally (Figs. 3A and 3B). In thin section, mineral constituents are also aligned (Fig. 8A); Owens et al. (2017) suggest mineral alignment is a magmatic foliation. Phenocryst alignment in porphyritic rocks is co-planar to aligned biotite plates and weak compositional layering in massive equigranular granite in the Richmond pluton (Fig. 9A). At map scale, contacts between porphyritic granite and massive equigranular granite map units, and with rocks of our informal Pocoshock Creek gneiss, are sharp and concordant.

In both the Richmond and DeWitt-Sutherland plutons, myrmekite is common (Fig. 8B) but is not as abundant in porphyritic granite as in massive equigranular granite. Chlorite is a reaction replacement mineral phase that replaces biotite, and chlorite + biotite are reaction replacement mineral phases that replace hornblende in massive equigranular granite; muscovite also replaces biotite locally in porphyritic granite. Quartz crystals exhibit undulatory extinction. Allanite cores in epidote occur sparsely.

Regionally, field relations and published geochronology support that massive equigranular granite and porphyritic granite are similar in age. Crosscutting field relations between the massive equigranular granite and porphyritic granite are not always clear. For example, in the Richmond area, massive equigranular granite crystallized as dikes up to several meters (many feet) thick that crosscut porphyritic granite (Fig. 3A). U-Pb zircon ID-TIMS analyses of 299.87 ± 0.13 Ma and 296.04 ± 0.14 Ma for porphyritic and massive equigranular granite, respectively (Owens et al., 2019; locations shown in Fig. 2), confirm this observed field relation. Elsewhere, bodies of porphyritic granite are mapped within massive equigranular granite (e.g., Carter et al., 2007a). Crosscutting cm- to m- (inches to feet) thick aplite (Fig. 3A) and pegmatite dikes are also common in both plutons.

Foliated metagranitoid gneiss, or our informal Pocoshock Creek gneiss, constitutes 40% of the area formerly mapped as Petersburg Granite (sensu lato; Fig. 2). These rocks are mostly granitic and granodioritic in composition both modally and chemically, but a few are tonalitic to trondhjemitic in composition (Fig. 4). Pocoshock Creek gneiss is mainly composed of finely to coarsely crystalline quartz and feldspar and abundant biotite. Hornblende is present locally, and titanite is locally a constituent in strongly foliated rocks. Plagioclase is generally more abundant than potassium feldspar. Garnet occurs locally in gneissic rocks (Fig. 8C). Muscovite, intergrown with biotite, commonly forms foliation surfaces in strongly foliated rocks. Locally, masses of fibrous muscovite occur in the foliation; the texture is consistent with replacement of fibrous sillimanite by muscovite (Fig. 8D), but no relict sillimanite was found in these masses. Muscovite is sparse in gneissic rocks. Chlorite replaces biotite. Allanite cores in epidote are most common in strongly foliated rocks (Fig. 8E) but also occur sparsely in gneissic rocks.

The foliation in Pocoshock Creek gneiss varies from weak to strong (Fig. 3C); aligned muscovite and biotite plates define the penetrative foliation. Gneissic rocks have distinct compositional layering (Fig. 3D). The layering is defined by biotite- and quartz-feldspar–rich layers that are 1.25–5 cm thick. In some outcrops, including along Pocoshock Creek (“P” in Fig. 2), foliation is truncated by layering in gneissic rocks (Fig. 3H). Both phyllosilicate-defined foliation and compositional layering are tightly to broadly folded; in some outcrops, the folding appears to be polyphase (Fig. 3I). Foliation and compositional layering in these rocks is co-planar to biotite and phenocryst alignment in massive equigranular granite and porphyritic granite (cf. Figs. 9A and 9B).

Felsic and mafic metamorphic rocks occur as outcrop- to map-scale xenoliths mostly within Pocoshock Creek gneiss. A mappable screen of quartz-muscovite (±garnet, staurolite, kyanite, or sillimanite) felsic schist bounds the southwestern-most edge of the DeWitt-Sutherland pluton along the Nottoway River fault (Fig. 2). Quartz-muscovite felsic schist in xenoliths and wall rock are distinguishable from Pocoshock Creek gneiss by abundant muscovite and locally porphyroblastic garnet, staurolite, or kyanite (Fig. 8F) in quartz-muscovite felsic schist. Sillimanite occurs in rocks of highest metamorphic grade, but more commonly, muscovite forms very fine crystalline masses that may have replaced fibrolitic sillimanite (Fig. 8G). Sillimanite is also locally preserved as inclusions in monazite (Fig. 8H). Muscovite and biotite define a strong schistosity, which is folded, in these rocks.

Mafic xenoliths include amphibolite (Fig. 3F), hornblende-biotite gneiss, and rare meta-ultramafic rocks, which contain relict crystals of pyroxene and abundant amphibole, serpentine, talc, and chlorite. Biotite, chlorite, and epidote are common metamorphic mineral phases in amphibolite and hornblende-biotite gneiss. Some amphibolite xenoliths are coarsely crystalline and granoblastic to nematoblastic. Finer crystalline amphibolite and hornblende-biotite gneiss xenoliths are strongly foliated or compositionally layered. Discrete differences in crystalloblastic sizes, textures, and compositions define layering, which is typically ~1 cm thick. Regionally, gneissic layering in xenoliths is oriented similarly to foliation in Pocoshock Creek gneiss and in plutons of massive equigranular granite and porphyritic granite (cf. Fig. 9C with Figs. 9A and 9B), but foliation in some mafic xenoliths is discordant to foliation in Pocoshock Creek gneiss (Fig. 3F). Layering in some xenoliths is tightly to isoclinally folded; secondary foliation is axial-planar and co-planar to phyllosilicate foliation in surrounding Pocoshock Creek gneiss.

Major map units (Fig. 2), based primarily on macroscale textural differences, were identified during detailed field mapping. These units include massive equigranular granite and porphyritic granite of the Petersburg Granite (sensu stricto) and foliated metagranitoid gneiss of our informal Pocoshock Creek gneiss. Many of these rocks are mineralogically similar and have similar geochemical characteristics. Geochemistry of the units overlap, but some distinct differences exist. For example, most samples are alkali-calcic to calc-alkalic (Fig. 5A), and all but two samples are peraluminous (Fig. 5B). Most equigranular and porphyritic granite samples are classified as A-type granite, whereas most Pocoshock Creek gneiss samples fall into I- and S-type granite fields (Fig. 5C).

Harker diagrams comparing element analyses for all samples are plotted in Figure 6. Major oxides relative to SiO₂ content define moderately linear trends for most samples (e.g., Fig. 6A). However, several Pocoshock Creek gneiss samples define a distinct cluster; these samples are enriched in MnO, CaO, and MgO (Figs. 6B–6D) and depleted in K₂O (Fig. 6E). Additionally, these samples tend to have slightly lower TiO₂ content (Fig. 6F). These samples plot as tonalite, trondhjemite, and granodiorite (Fig. 4B), whereas a second subset of Pocoshock Creek gneiss samples and all massive equigranular and porphyritic granite samples plot as syeno- or monzogranite (Fig. 4A). Trace-element bivariate plots (Figs. 6I–6L) distinguish a subset of Pocoshock Creek gneiss samples from other granitic rocks.

Chondrite-normalized REE spider diagrams are plotted in Figure 7. Porphyritic granite samples have elevated light REEs relative to other rock types, no Ce anomaly, a negative Eu anomaly, and exhibit a moderate decrease in heavy REEs (Fig. 7A). Samples of porphyritic granite that have aligned phenocrysts are similar to samples of granite with randomly oriented phenocrysts. Massive equigranular granite samples generally show similar REE patterns to porphyritic samples, although two samples have a pronounced Ce anomaly, and one sample has a positive Eu anomaly (Fig. 7B). Pocoshock Creek gneiss samples (Fig. 7C) exhibit significant similarities in REE chemistry with equigranular granite samples and are enriched in light REEs, have little or no Ce anomalies, and negative Eu anomalies (cf. Figs. 7B and 7C). Of note is garnet-bearing sample M19-02-28A, which shows a marked increase in heavy REEs (Fig. 7C).

The reported Early Devonian age of a xenolith (Fig. 2) by Owens et al. (2017, 2019) motivated us to identify texturally similar rocks and determine their ages. We dated sample R-11518, a weakly foliated biotite metagranitoid assigned to our informal Pocoshock Creek gneiss, using the LA-ICP-MS U-Pb zircon method. In this sample, aligned biotite plates form a penetrative foliation. Zircon grains from this sample are generally between 150 and 300 µm long and euhedral elongate and prismatic (Fig. 10A). Most grains show distinct oscillatory zoning in scanning electron microscope–cathodoluminescence (SEM-CL) imaging (lighter cores and darker rims). Many cores are damaged or preserve inclusions. Most of the 30 analyses were discordant. After excluding discordant (more than ±5% discordant ²⁰⁷Pb/²³⁵U vs. ²⁰⁶Pb/²³⁸U) data, data that have U/Th ratios greater than 5 (Th/U < 0.2), and analyses that suggest inheritance, the remaining data yield a concordia age of 416.7 ± 4.3 Ma (Fig. 10B).

We collected four additional samples of Pocoshock Creek gneiss for U-Pb isotopic analysis at the Stanford-USGS SHRIMP-RG facility. Sample M19-02-28B is a strongly foliated muscovite-biotite metagranitoid (Fig. 3C). Zircon grains from this sample are ~300–450 µm long and are elongate and prismatic (Fig. 11A). These grains exhibit oscillatory zoning in SEM-CL imaging and have distinct and narrow (10–20 µm) dark rims. Thirteen core analyses yield a ²⁰⁶Pb/²³⁸U weighted-average age of 424.6 ± 3.6 Ma, and seven analyses of rims with low Th/U (0.011 ± 0.002, typical of metamorphic zircon, e.g., Rubatto, 2017) yield reproducible ages and a ²⁰⁶Pb/²³⁸U weighted-average age of 324.4 ± 3.2 Ma (Fig. 12A).

Sample M19-02-28C is a compositionally layered granodioritic to tonalitic orthogneiss, with 2–3-cm-thick quartz-feldspar layers separated by several-mm-thick schlieren of mostly biotite, from an outcrop on Pocoshock Creek (Fig. 2). Orthogneiss is both interlayered with and truncates foliated biotite metagranitoid (Fig. 3H). Zircon grains from this sample are ~250–425 µm long and are subhedral elongate prisms (Fig. 11B). Grains show oscillatory zoning in SEM-CL imaging (lighter cores and distinct darker rims). Nineteen core analyses yield a ²⁰⁶Pb/²³⁸U weighted-average age of 418.1 ± 2.6 Ma (Fig. 12B). Two spot analyses in dark CL rims yield ages of 313 ± 10 Ma (2σ) and 316 ± 8 Ma; both overlap within error with the weighted-average age of rims from sample M19-02-28B. Eight additional analyses were located on dark CL rims. However, these analyses are discordant and/or are high in ²⁰⁴Pb (Fig. 12B; Table S2 [^{footnote 1}]). A single spot analysis in a bright core (Fig. 11B) yields an age of 449 ± 20 Ma.

Sample JR-1 is a coarsely crystalline, foliated biotite metagranitoid from the bed of the James River in Richmond. Foliation in this rock consists of several-mm-thick schlieren of mostly biotite that separate several-cm-thick layers of mostly quartz and feldspar. Subhedral to euhedral zircon grains from this sample are 300–500 µm long and are prismatic (Fig. 11C). Nineteen core analyses yield a concordant crystallization age of 403.0 ± 3.8 Ma (Fig. 12C). Six analyses of zircon rims with low Th/U (0.016 ± 0.004) yield a weighted-average age of 299 ± 14 Ma. Although relatively imprecise, this age overlaps with concordant rim ages recorded by other samples, and moreover, the crystallization age of nearby massive granite as reported by Owens et al. (2017, 2019). Two additional spot analyses, taken in bright CL zones that have vein-like geometry and are truncated by dark CL rims, yield ages of 378 ± 14 Ma and 376 ± 15 Ma (Fig. 11C).

Sample M19-02-28A is a weakly foliated and coarsely layered garnetbiotite metagranitoid. It is crosscut by several dm- to ~0.5 m-thick pegmatitic granite dikes containing cm-thick books of muscovite. These dikes crosscut the weak foliation. Subhedral and elongate zircon grains are 200–300 µm long and are heavily pitted (Fig. 11D). Very high uranium content, pitting, intragrain variance in hardness observed after polishing, and muted to absent cathodoluminescent response indicate that many grains are metamict. Several zircon grains have distinct mottled to patchy core textures and show evidence of recrystallization and subsequent zircon growth. Additionally, many analyses have very high common lead. Examining only data with relatively low common lead (²⁰⁴Pb) and U < 1650 ppm, two age clusters at ca. 400 Ma and ca. 300 Ma are consistent with other strongly foliated samples. Five concordant core analyses show overdispersion (MSWD = 3.8, p(χ2) = 0.004) and yield a ²⁰⁶Pb/²³⁸U weighted-average age of 404 ± 7 Ma. Four concordant rim analyses yield a ²⁰⁶Pb/²³⁸U weighted-average age of 308.1 ± 4.1 Ma and a Th/U = 0.019 ± 0.017. One core analysis yields an age of 631 ± 14 Ma (Fig. 11D).

We also collected two samples of massive equigranular granite from southwest of the Petersburg area (Fig. 2). Sample DW-11 is massive, coarsely crystalline, equigranular granite containing “smoky” (N5 medium-gray) quartz. Zircon grains from this sample are subhedral prisms 200–250 µm long (Fig. 11E). Grains show oscillatory zoning in SEM-CL imaging. Thirteen analyses yield a ²⁰⁶Pb/²³⁸U weighted-average age 320.5 ± 3.5 Ma (Fig. 12E). Analyses of dark CL rims with Th/U >0.2 yield ages that overlap with core ages. Sample Jack-1 is massive, coarsely crystalline granite from the Vulcan Materials Company Jack Stone Quarry. Zircon grains from this sample are euhedral prisms up to 400 µm long (Fig. 11F). The grains exhibit oscillatory zoning in SEM-CL images, and in most grains, rims are darker than cores in CL imaging. Fourteen analyses from oscillatory zoned zircon cores yield Th/U of 0.75 ± 0.15 (typical of primary igneous zircon) and a ²⁰⁶Pb/²³⁸U weighted-average age of 317.3 ± 3.6 Ma (Fig. 12F). High uranium zircon rim ages (n = 4) exhibit more scatter in age but record igneous Th/U values and overlap in age with grains cores. A few (n = 6) core spot analyses yield ages ranging from ca. 554 to ca. 376 Ma. The crystallization ages of massive equigranular granite of samples DW11 and Jack-1 are both within uncertainty of the Pennsylvanian date (314 ± 2.1 Ma) of Lee and Williams (1993) from rocks presumed to represent the Petersburg Granite (sensu lato).

In addition to obtaining the age of igneous rocks, we sampled quartz-muscovite felsic schist from a map-scale xenolith (BART in Fig. 2) for detrital zircon analysis using the LA-ICP-MS technique. Sample BonAirRRTracks is a strongly foliated quartz-muscovite felsic schist that preserves oriented sillimanite needles both as inclusions in coarse muscovite (Fig. 8G) and in porphyroblastic monazite (Fig. 8H). The most significant age mode (n = 40 of 120 analyses that form a coherent group according to the zircon age extractor “TuffZirc Age” routine in Isoplot of Ludwig, 2012) is 528.8 Ma (Fig. 13A). Minor modes are present at 450 Ma and 382 Ma, and single-grain dates were determined at 1038 Ma and 339 Ma.

Historically, the Petersburg Granite (sensu lato) has been considered to be a Late Pennsylvanian to early Permian composite batholith (e.g., Virginia Division of Mineral Resources, 1993). Strongly foliated rocks constitute a significant portion of what was mapped as Petersburg Granite (sensu lato) (e.g., Carter et al., 2007a). In the absence of modern geochronology studies, foliated and gneissic facies were interpreted to be the earliest syntectonic component of the Petersburg Granite (sensu lato) (e.g., Carter, 2011). Our new U-Pb zircon geochronology demonstrates that some strongly foliated parts of the Petersburg Granite (sensu lato) are actually late Silurian to Early Devonian in age and ~100 m.y. older than the massive equigranular and porphyritic granite of the Petersburg Granite (sensu stricto).

U/Pb zircon geochronology spot analyses of 320.5 ± 3.5 Ma (sample DW-11; Fig. 12E) and 317.3 ± 3.6 Ma (sample Jack-1; Fig. 12F) are interpreted to reflect the time of crystallization of undeformed to weakly (magmatic?) foliated granite. Rim ages that overlap with core ages in both samples (Figs. 12E and 12F) also likely reflect igneous crystallization. These rocks are 15–20 m.y. older than previously dated granite samples assigned to the Petersburg Granite (sensu stricto) in the Richmond area (Owens et al., 2017, 2019) and here help us to define the northern Richmond pluton and the DeWitt-Sutherland pluton, which together comprise the composite batholith of Petersburg Granite (sensu stricto). Sharp and concordant contacts between equigranular granite, porphyritic granite, and with rocks of our informal Pocoshock Creek gneiss suggest catazonal emplacement of the Richmond and DeWitt-Sutherland plutons. Based on a similar bimodal distribution of published ages (ca. 326–305 Ma and ca. 300–282 Ma) for other southern Appalachian orogen plutons, Speer et al. (1994) suggested that Alleghanian magmatism is represented by two pulses of Carboniferous–Permian magmatic activity, which is supported by our new data. Mineralogic observations suggest a shared history of weak metamorphism and minor post-emplacement deformation of the Richmond and DeWitt-Sutherland plutons.

U/Pb zircon geochronology spot analyses of oscillatory zoned cores from five Pocoshock Creek gneiss samples (JR-1, M19-02-28A, M19-02-28B, M19-02-28C, and R-11518) yield ages ca. 425–403 Ma (Fig. 12), interpreted to reflect the time of protolith crystallization of these metaigneous rocks. These ages firmly establish that foliated and gneissic layered metagranitoids of our informal Pocoshock Creek gneiss should not be assigned to the late Paleozoic Petersburg Granite (sensu lato). Differences in trace-element concentrations (Fig. 6) between Pocoshock Creek gneiss samples and Petersburg Granite (sensu stricto) samples suggest different magma sources for these two suites of rock. Several of the Silurian–Devonian Pocoshock Creek gneiss samples fall geochemically into the tonalite-trondhjemite-granodiorite (TTG) suite of magmatic rocks (Fig. 4B), suggesting that granitic protoliths of Pocoshock Creek gneiss likely originated in a late Silurian to late Early Devonian continental-margin arc system (e.g., Moyen and Martin, 2012; Jagoutz et al., 2013).

Zircon grains separated and imaged from Pocoshock Creek gneiss samples typically have discontinuous rims with high U concentrations and Th/U of << 0.1. A small subset of these rims is sufficiently thick to allow for a U/Pb spot analysis. Rim ages exhibit scatter, and relative to grain cores rim analyses are more commonly discordant and/or high in common lead. However, nearly all analyses that are both concordant and low in common lead yield ages between 330 and 280 Ma. Additionally, rims from M19-02-28A and especially M19-02-28B, are reproducible and yield ages from ca. 324 Ma to ca. 308 Ma. In general, these ages strongly overlap with intrusive ages of massive equigranular granite of the Petersburg Granite (sensu stricto) in the DeWitt-Sutherland and Richmond plutons reported here and in Owens et al. (2017, 2019), respectively. These rims are interpreted to have formed when late Paleozoic granitic magmas intruded Pocoshock Creek gneiss (Figs. 14 and 15). Scatter in rim ages could reflect several episodes of Alleghanian metamorphism driven by pulses of magmatism or other tectonic events but could also reflect lead loss or other issues associated with analysis of these high U-bearing rims, and so no geologic interpretation is made from the scatter in rim ages. Ages of ca. 378 Ma and ca. 376 Ma on bright CL veins of zircon transect the oscillatory zoned cores but are truncated by darker rims (Figs. 11C and 12C) suggest a time of zircon fracture and metamorphic zircon growth prior to the Alleghanian orogeny (Figs. 14 and 15). For sample R-11518, we attribute the high degree of discordant zircon analyses to be related to the subsequent dynamothermal event(s) that postdated initial igneous crystallization of the metagranitoid.

Both the massive equigranular and strongly foliated suites of rock contain zircon xenocrysts that we interpret to be inherited from the crust through which they intruded. The oldest zircon core from Pocoshock Creek gneiss samples is ca. 631 Ma (sample M19-02-28A; Figs. 11D and 12D), and older zircon cores from massive equigranular granite of sample Jack-1 are ca. 554 Ma, ca. 540 Ma, ca. 502 Ma, and ca. 444 Ma (Figs. 11F and 12F). These data suggest Neoproterozoic to early Paleozoic crust supplied zircon xenocrysts to both the Silurian-Devonian and Pennsylvanian granitic melts. None of the samples contain inherited cores ca. 1 Ga that would indicate intrusion through Laurentian basement. We interpret the LA-ICP-MS detrital zircon age mode of 528.75 ± 2.85 Ma from a map-scale xenolith of quartz-muscovite felsic schist (sample BonAirRRTracks; Fig. 13) to be a reasonable maximum depositional age for the protolith of the schist. This sample does have two younger albeit minor age distributions ca. 450 Ma and ca. 380 Ma. However, we interpret those younger ages to represent metamorphic overgrowths or spots that sample igneous cores and metamorphic overgrowths—this is supported by a trend of elevated U/Th (or low Th/U) >3 in younger (<500 Ma) concordant analyses (Fig. 13B, inset). Additionally, a maximum depositional age ca. 382 Ma is not likely because field relations indicate that the quartz-muscovite felsic schist is a xenolith within Silurian–Early Devonian Pocoshock Creek gneiss. The abundance of Cambrian-age zircons is consistent with the presence of inherited Cambrian zircon grains in equigranular granite that intrudes the metasediments as is seen in sample Jack-1. The minor ages ca. 450 Ma overlap with a single inherited core age from Pocoshock Creek gneiss sample M-19-02-28C and a single inherited core age in Petersburg Granite (sensu stricto) sample Jack-1. The paucity of Mesoproterozoic zircon (n = 1) in the detrital zircon suite and lack of Mesoproterozoic xenocrysts in the intrusive rocks highlight the significant contribution of a relatively young Neoproterozoic to early Paleozoic crust in generating both sediments and granitic melts.

Horton et al. (1989, p. 214) define a tectonostratigraphic terrane as “a fault-bounded geologic entity of regional extent characterized by an internally homogeneous stratigraphy and geologic history that is different from that of contiguous terranes” (definition after Coney et al., 1980; Jones et al., 1983a, 1983b; Howell and Jones, 1984). Our new data show that Silurian–Devonian Pocoshock Creek gneiss and late Paleozoic plutons of the Petersburg Granite (sensu stricto) represent a terrane that we here call the Dinwiddie terrane, as it is separated from surrounding crustal blocks by major strands of the Eastern Piedmont fault system and has a stratigraphic and metamorphic history different from that of surrounding blocks (cf. Figs. 15 and 16). In addition to batholith-scale middle to late Paleozoic intrusive units, the Dinwiddie terrane also consists of metasedimentary and metavolcanic rocks of Neoproterozoic to at least Cambrian age; these rocks occur as outcrop- to map-scale xenoliths and a screen within Paleozoic intrusive rocks. They may be vestiges of an older volcanic arc system (e.g., Ray and Owens, 2018). The Dinwiddie terrane is bound to the west by the late Paleozoic, 0.5- to 2.4-km-wide, ductile to brittle Hylas fault zone, which separates it from the Goochland terrane (Fig. 2; see also simplified terrane map in Fig. 16). Recent detailed (Carter et al., 2022b) and reconnaissance mapping (Carter et al., 2020) along the southwestern boundary of the Dinwiddie terrane shows that it is juxtaposed against rocks of the Roanoke Rapids terrane along a 1.5-km-wide high-strain zone named the Nottoway River fault zone. The eastern boundary of the Dinwiddie terrane is under cover of Cenozoic Atlantic Coastal Plain sediments; the contact is only known from geophysical data and limited deep borehole data but is interpreted to be a suture zone (Horton et al., 1989, 1991; Lefort, 1989; Lefort and Max, 1989) separating rocks now assigned to the Dinwiddie terrane from greenschistfacies metasedimentary, metavolcanic, meta-ultramafic, and granitic plutons of the Chesapeake block (Horton et al., 1991). The Dinwiddie terrane is named for the Virginia County that hosts many of these rocks and structures.

Our new data set provides detailed field and analytical information to better understand the construction of the Appalachian crystalline core. Prior to this work, the pre-Pennsylvanian history of the Dinwiddie terrane was largely unknown because older rocks had not been identified. New mapping, geochemistry, and U-Pb zircon geochronology, coupled with an analysis of zircon textural morphology and growth stages (Fig. 14), provide important constraints on the T-t-D history of these newly recognized older rocks and the terrane that they comprise (Fig. 15).

The oldest rocks identified in the Dinwiddie terrane are the xenolith of sillimanite-bearing quartz-muscovite felsic schist in the Richmond area (sample BonAirRRTracks; Fig. 13). Lack of abundant Mesoproterozoic zircon that are typical of Laurentian margin sources (e.g., Carter et al., 2006) and Paleoproterozoic to Neoproterozoic zircon that are typical of Gondwanan sources (e.g., Pollock et al., 2012) in the detrital zircon suite is significant. In the absence of clear sedimentary input from these widespread crustal sources, we conclude that the protoliths of these metasedimentary rocks were deposited in a sedimentary basin that was proximal to its source and isolated from older basement terranes. This relation is common in forearc basin settings (e.g., Cawood et al., 2012; their figure 1). Direct evidence for quartz-muscovite felsic schist to be of volcanic arc affinity is lacking due to post-depositional, upper-amphibolite–facies metamorphism. However, the geochemistry of a nearby ~250 m × 350 m mafic xenolith is consistent with a volcanic arc origin for those rocks (Ray and Owens, 2018). Assuming the detrital zircon age mode ca. 529 Ma is close to the maximum depositional age for quartz-muscovite felsic schist and that zircon xenocrysts preserved in younger intrusive rocks are also indicative of the age of the crustal block through which they intruded, an early Cambrian volcanic arc is the inferred infrastructural terrane substrate. Neoproterozoic–Cambrian sedimentation and volcanism generally falls within published ages for peri-Gondwanan volcanic arcs and sedimentary basins of both the southern and northern Appalachian orogen (e.g., Ingle et al., 2003; Pollock et al., 2010; Fig. 17).

That foliation in some mafic xenoliths is discordant to foliation or layering in Pocoshock Creek gneiss (Fig. 3F) demonstrates that the protolith to the Pocoshock Creek gneiss intruded previously foliated rocks (Fig. 15). In the Richmond area, the map-scale xenolith of sillimanite-bearing quartz-muscovite felsic schist preserves a relict foliation defined by a preferred orientation of sillimanite needles (Fig. 8G). The presence of relict sillimanite with a preferred orientation indicates that this rock experienced multiple periods of deformation and metamorphism. Sillimanite also occurs as inclusions in muscovite, quartz, and monazite (Fig. 8H), indicating that sillimanite growth predates penetrative mica schistosity in the xenolith. We interpret the penetrative mica schistosity to be synchronous with foliation development in the Pocoshock Creek gneiss, and the relict sillimanite schistosity to predate gneissosity in host Pocoshock Creek gneiss. We cannot unequivocally rule out that sillimanite growth in the xenolith could be synchronous with development of gneissosity in Pocoshock Creek gneiss. However, we suggest that if the penetrative mica schistosity in the xenolith developed after regional foliation development in Pocoshock Creek gneiss, this deformation should have imparted a new phyllosilicate alignment in the gneiss, more so than just refolding by gentle to open folds, as in Figure 3I. Thus, we conclude that the earlier schistosity defined by sillimanite likely developed between ca. 529 Ma and ca. 425 Ma—the maximum depositional age from detrital zircon geochronology and the oldest intrusive age of Pocoshock Creek gneiss protolith, respectively.

Deformed Neoproterozoic to Cambrian sedimentary and volcanic rocks were subsequently intruded by Pocoshock Creek gneiss protolith, a suite of tonalite-trondhjemite-granodiorite and granite plutons, during the late Silurian Period to Early Devonian Epoch (Fig. 15). Geochemical data (Table S1; see ^{footnote 1}) suggest that this suite of intrusive rocks likely formed in a continental arc setting (Fig. 5), perhaps built upon the older peri-Gondwana volcanic arc system. Partial melting of the Neoproterozoic–Cambrian volcanic arc system to generate some of the Pocoshock Creek gneiss magma is consistent with the mostly I- and S-type geochemistry of the Pocoshock Creek gneiss (Fig. 5C), and especially aluminous samples, such as garnet-bearing M19-02-28A (Figs. 5C and 8C). Inherited and/or xenocrystic cores in zircon from the Pocoshock Creek gneiss, as well as Petersburg Granite (sensu stricto; Figs. 11 and 12) and detrital zircon analyses (Fig. 13), are all consistent with Neoproterozoic to Cambrian rocks underlying the exposed Dinwiddie terrane.

Although the protolith of the Pocoshock Creek gneiss crystallized between ca. 423–402 Ma, our data only constrain timing of penetrative foliation and gneissosity development in these rocks from the youngest intrusive age for the protolith ca. 402 Ma to intrusion of Petersburg Granite (sensu stricto) in the late Paleozoic (ca. 325–300 Ma; Fig. 15). Field relations show that penetrative foliation in Pocoshock Creek gneiss is regionally co-planar to what is considered magmatic flow fabric (Owens et al., 2017) in ca. 300 Ma granite in the Richmond area (cf. Figs. 9A and 9B). In outcrop, contacts between Pocoshock Creek gneiss and younger granite are sharp and concordant. Textural relations of muscovite masses in foliation of Pocoshock Creek gneiss are consistent with replacement of fibrous sillimanite by muscovite (Fig. 8D), and the presence of sillimanite would be consistent with contact metamorphism during intrusion of nearby Permian granite. Unfortunately, no relict sillimanite was found in these masses, and mesoscopic evidence for contact metamorphism is also lacking. We interpret these data to suggest that the gneissic rocks were relatively hot and dry during late Paleozoic intrusion. However, the field relations cannot clearly distinguish if the co-planarity of the gneissosity in Pocoshock Creek gneiss and magmatic layering in younger granite reflects contemporaneous formation or if the fabric in Pocoshock Creek gneiss is much older but controlled the orientation of later magmatic fabric, thus resulting in co-planarity. Additional work could help to assess contact metamorphic effects of Permian intrusions and the age of the gneissocity in the Pocoshock Creek gneiss.

Evidence for early and lower temperature deformation is preserved in the textures of zircon grains from Pocoshock Creek gneiss. Zircon grains separated from three samples (JR-1, M19-02-28C, and M19-02-28B) preserve consistent textural crosscutting relations (Fig. 14). Of note is the presence of thin, bright CL veins that transect the oscillatory zoned cores of zircon grains and are themselves truncated by ca. 320–300 Ma dark CL rims. Zircon can fracture due to internal strain related to the volume expansion of the crystal structure with increasing radiation damage (Lee and Tromp, 1995), or it can fracture due to externally induced strain (Rimsa et al., 2007; Wintsch et al., 2014). In the case of self-induced fracture, fractures typically either emanate radially from uranium-rich zones in the zircon crystal or form concentrically at the boundaries between zones of high- and low-uranium content (Lee and Tromp, 1995). These fracture patterns are not observed in the zircon grains from the Pocoshock Creek gneiss. On the contrary, veins of many orientations are observed, and some transect the entire grain across its width or length (Fig. 14A). Furthermore, typical spot analyses from zircon grains from the Pocoshock Creek gneiss have only moderate uranium (~300 ppm) and thorium content (~150 ppm), and the grains with healed fractures typically lack very dark CL bands, indicative of high radiation damage, which might drive local expansion and fracture. We suggest that brittle deformation of zircon in Pocoshock Creek gneiss resulted from externally induced strain. Because Pocoshock Creek gneiss is dominated by quartz and feldspar, these minerals likely behaved in a brittle manner such that strain was transferred to the zircon grains to create the fractures (Wintsch et al., 2014). These veins not only show these crosscutting relations, but also all have biotite inclusions, despite being collected >30 km apart. This observation suggests that zircon fracturing and healing comprised a regional event. Two analyses of these veins ca. 378 Ma and ca. 376 Ma suggest that this event occurred in the Late Devonian Epoch, Frasnian Age. At typical crustal strain rates, quartz behaves in a ductile manner above ~300 °C; therefore, we suggest that regionally, rocks of the Pocoshock Creek gneiss had cooled to approximately this or lower temperature by Late Devonian time (Fig. 15).

With the onset of the Alleghanian orogeny by Late Mississippian time, orogen-scale dextral transpression at greenschist- to amphibolite-facies metamorphism, and syn-tectonic granitic intrusion, occurred throughout the Piedmont hinterland (e.g., Bailey et al., 2004). Some re-burial likely occurred prior to the intrusion of the late Paleozoic composite Petersburg Granite (sensu stricto; Fig. 15). Although no data exist for this batholith, barometric estimates from other late Paleozoic southern Appalachian plutons yield emplacement pressures of 0.35–0.4 GPa, or depths of 13–15 km (Vyhnal and McSween, 1990; Vyhnal et al., 1991; Spear et al., 1994). Following re-burial, the first magmatic pulse (DeWitt-Sutherland pluton) of the Petersburg Granite (sensu stricto) batholith at ca. 320 (Fig. 15) Ma drove growth of high uranium concentration but low Th/U rims on zircon in the Pocoshock Creek gneiss (observed in samples M19-02-28A, M19-02-28B, M19-02-28C, and JR-1; Figs. 11A–11D). This magmatic pulse was followed by postcollisional intrusion and intrusion related to overthickened crust, with accompanying greenschist-facies retrogression, by earliest Permian time at ca. 300 Ma, which may have driven additional metamorphic zircon growth in all samples, but the data are not sufficiently precise to distinguish episodes of rim growth.

Permian granite of the Richmond pluton contains myrmekite and quartz with undulose extinction, and some rocks preserve a weak biotite foliation. These observations suggest that greenschist-facies deformation continued following the intrusion of the Petersburg Granite (sensu stricto). Ductile highstrain deformation, which began as dextral transpression and possibly evolved into postcollision orogenic relaxation and/or collapse, was confined to regional Piedmont fault zones, and lasted into middle Permian time (e.g., Durrant, 1979; Blake, 2012; Blake et al., 2012). In central-eastern Virginia, this deformation was mostly localized along the Hylas and Nottoway fault zones where mylonitic granite occurs adjacent to Silurian–Devonian Pocoshock Creek gneiss and Pennsylvanian–Permian granite. The presence of Petersburg Granite (sensu stricto) beneath sedimentary rocks of the Mesozoic Richmond basin places portions of the Dinwiddie terrane at Earth’s surface by Triassic time (Owens et al., 2017).

A comparison of the newly derived temperature-time path of the Dinwiddie terrane (Fig. 15) with existing data from adjacent terranes further distinguishes the Dinwiddie terrane and permits speculation on the timing of terrane juxtaposition. To the northwest of the Dinwiddie terrane across the Hylas fault zone is the Goochland terrane (Fig. 2, and simplified terrane map in Fig. 16), which is composed of Mesoproterozoic basement (State Farm Gneiss and Montpelier Anorthosite) and Mesoproterozoic(?) to Paleozoic cover (Maidens Gneiss) that are intruded by Neoproterozoic to Pennsylvanian plutons (Farrar, 1984; Aleinikoff et al., 1996; Owens and Tucker, 2003; Owens et al., 2010a; Martin et al., 2020). Geochemistry suggests that ca. 552 Ma Sabot Amphibolite is a metamorphosed rift or mid-ocean ridge basalt (Martin et al., 2020), unlike amphibolite xenoliths in Pocoshock Creek gneiss and Petersburg Granite (sensu stricto), which are likely arc-related (Ray and Owens, 2018). A-type Neoproterozoic felsic magmatism present in the Goochland terrane has not been identified in the Dinwiddie terrane. Devonian plutonic rocks ca. 392 Ma included in the Maidens Gneiss (Owens et al., 2010a) are slightly younger than ca. 425–403 Ma Pocoshock Creek gneiss. All of these characteristics indicate that the infrastructure of these two terranes is distinct. Furthermore, Devonian plutonic rocks of the Maidens Gneiss appear to be partial melts of a plutonic substructure and preserve two-pyroxene granulite-facies mineral assemblages (Farrar, 1984) as young as Devonian in age (Shirvell et al., 2004; Owens et al., 2010a). This is in stark contrast to the exposed Dinwiddie terrane rocks, which were at lower greenschist-facies at the same time (Fig. 16A).

The Dinwiddie terrane is bound to the southwest by the Nottaway River fault zone, which exhibits top-to-the-NE, shear-sense indictors that place rocks of the Roanoke Rapids terrane structurally above rocks of the Dinwiddie terrane (Carter et al., 2020; Carter et al., 2022b). Rocks of the Roanoke Rapids terrane include felsic and mafic metavolcanic rocks, some metasedimentary rocks (including lower greenschist-facies bedded metasiltstone recently mapped on the Cherry Hill quadrangle by Carter et al., 2022b), and felsic and mafic metaplutonic rocks (Sacks, 1996, 1999). Reported ages for all Roanoke Rapids terrane rocks are overwhelmingly Neoproterozoic: in Virginia, felsic metavolcanic rocks are ca. 673 Ma (Owens and Hamilton, 2018); intrusive granitoids range from ca. 629 Ma to ca. 615 Ma, and tonalite and metagabbro are ca. 614 Ma (Owens et al., 2010b; Owens et al., 2011) and ca. 610 Ma (Dearborn et al., 2016), respectively. Reported ages for meta-igneous rocks of the Roanoke Rapids terrane in North Carolina overlap in age with the older Virginia rocks and range from ca. 672 Ma to ca. 668 Ma (Horton and Stern, 1994; Coler and Samson, 2000). Additionally, a sample of metaconglomerate from North Carolina yielded a maximum depositional age of ca. 630 Ma, with a peak age mode of ca. 670 Ma, a single grain at ca. 480 Ma, and scattering of Mesoproterozoic grains (Pollock et al., 2010). Both the igneous and detrital signatures are distinctly different from those in the Dinwiddie terrane. Roanoke Rapids terrane rocks also yield dominantly negative initial ε_Nd values and Mesoproterozoic depleted mantle model ages, indicating involvement of isotopically evolved, older crust in their petrogenesis (Pollock et al., 2010; Owens and Hamilton, 2018; Fig. 16B). Dinwiddie terrane rocks have no apparent evidence of Mesoproterozoic crustal involvement. Again, many characteristics distinguish that the infrastructure of these two terranes is different. Regional metamorphism in the Roanoke Rapids terrane is greenschist facies in the core of a major regional syncline, to epidote-amphibolite facies on its flanks (Carter et al., 2020), but the timing or character of this event is not currently resolved (Fig. 16B). In contrast, multiple metamorphic events affected rocks of the Dinwiddie terrane, including an amphibolite-facies (sillimanite zone) event between ca. 529 Ma and ca. 425 Ma. Lastly, lower greenschist-facies rocks of the Roanoke Rapids terrane are juxtaposed against staurolite schists of the Dinwiddie terrane along the Nottaway River fault (Figs. 16B and 16C). For these reasons, it is unlikely that any correlation between Dinwiddie terrane and Roanoke Rapids terrane existed prior to adjacency along the Nottaway River fault zone in late Paleozoic time.

The above data strongly indicate that the Dinwiddie terrane is distinct from adjacent terranes to the west and south. Specifically, both the Goochland and Roanoke Rapids terranes exhibit evidence for a role of Mesoproterozoic crust in their genesis—the Dinwiddie terrane does not. Instead, the Dinwiddie appears more similar to the Chesapeake block to the east (but buried under coastal plain sediments), where all current data indicate that igneous rocks are Neoproterozoic or younger in age (Horton et al., 2016, and references therein), and no Mesoproterozoic components have been identified. In this regard, the Dinwiddie terrane seems to mark a key east-west change in basement age and genesis. Given the above-mentioned differences in T-t-D histories between the Dinwiddie and terranes to the west and south, and the modern difference in metamorphic grade across the bounding faults, we suggest that the Dinwiddie terrane was not juxtaposed against Goochland and Roanoke Rapids terranes until the Alleghanian orogeny. The history of the newly defined Dinwiddie terrane could be further refined with additional field, geochemical, and isotopic data.

Our re-evaluation of the Petersburg Granite (sensu lato) of the centraleastern Piedmont province of Virginia reveals that foliated and gneissic rocks formerly considered facies of a composite late Paleozoic batholith are 100 m.y. older than previously thought and range from ca. 425 Ma to ca. 403 Ma. These Silurian–Devonian strongly foliated and layered gneisses, which we refer to as our informal Pocoshock Creek gneiss, are geochemically distinct from Pennsylvanian–Permian equigranular granite, porphyritic granite, and monzodiorite Petersburg Granite (sensu stricto). Protoliths of Pocoshock Creek gneiss are part of a TTG suite of magmatic rocks that may have originated in a late Silurian to late Early Devonian continental arc system. Petersburg Granite (sensu stricto) is mostly syncollisional A-type granite, indicating a different magma source, and orogenesis, for this later magmatic event. Petersburg Granite (sensu stricto) includes the ca. 300 Ma northern Richmond pluton and the ca. 320 Ma southern DeWitt-Sutherland pluton; despite ~20 m.y. age difference, both plutons are still considered to be part of the Petersburg Granite (sensu stricto) batholith.

Collectively, our data indicate that some rocks formerly assigned to the Petersburg Granite (sensu lato) are part of a distinct terrane—the Dinwiddie terrane—in the central-eastern Piedmont of the southern Appalachian orogen. The Dinwiddie terrane is plausibly an exotic fragment of peri-Gondwanan affinity preserved in the southern Appalachian crystalline core. Metasedimentary and metavolcanic rocks in the Dinwiddie terrane that are preserved in xenoliths are of likely volcanic arc origin. Detrital zircon analyses of metasedimentary rocks indicate minimal contributions from Mesoproterozoic sources and a likely early Cambrian maximum depositional age. The Silurian–Devonian Pocoshock Creek gneiss intruded these Cambrian metasedimentary and metavolcanic rocks. Lastly, the Pennsylvanian–Permian Petersburg Granite (sensu stricto) intruded the terrane. Inherited zircon grains from both the Pocoshock Creek gneiss and Petersburg Granite (sensu stricto) range from Neoproterozoic to Devonian, but most are Cambrian in age and support the interpreted age of the oldest metasedimentary rocks of the terrane. Zircon grains in Pocoshock Creek gneiss record fracturing and metamorphic zircon growth at ca. 377 Ma as well as significant zircon rim growth coincident with late Paleozoic intrusion of the Petersburg Granite (sensu stricto). The Dinwiddie terrane is unlike the Goochland and Roanoke Rapids terranes to the west and may be more similar to the Chesapeake block to the east, which is buried beneath Atlantic Coastal Plain sediments. Thus, the Dinwiddie terrane may mark a significant east-west change in the crustal makeup of the southern Appalachian orogen at this latitude.

This research is supported by the U.S. Geological Survey (USGS) National Cooperative Geologic Mapping Program. Geologic mapping by Virginia Energy conducted through USGS, National Cooperative Geologic Mapping Program STATEMAP awards 05HQAG0088, 06HQAG0039, 08HQAG0091, G10AC0042, GC14AC00331, G17AC00310, C18AC00294, and G16AC00300. We thank Heather Bleick, Amy Bondurant, and Rick Berquist (Virginia Energy) for mapping assistance in the Richmond area from 2005 to 2009 through the USGS STATEMAP Program. Lorne Powell, Jonathan Johnson, and Ashley Lynn (USGS), David Spears and Patrick Finnerty (Virginia Energy), and Jack Nolan (University of North Carolina Wilmington) assisted in mapping along the Nottaway River and on the Cherry Hill 7.5-minute quadrangle from 2019 to 2020 through the USGS FEDMAP Program. We thank Kristian Price and Thomas Strong for sample preparation and Laura Pianowski for assistance in geochronology analyses and data management. We thank Eric Angel, Brad Ito, and Marsha Lidzbarski for assistance in the SHRIMP-RG laboratory. We appreciate the thoughtful reviews by Ryan Deasy and Randy Orndorff (USGS); David Spears (Virginia Energy); David Blake and Todd LaMaskin (University of North Carolina Wilmington); and Jamie Levine (Appalachian State University). Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Electron images of polished thin sections coated with carbon were taken on a Hitachi SU-5000 field emission–scanning electron microscope (FE-SEM) operating in high vacuum mode at U.S. Geological Survey in Reston, Virginia. The FE-SEM is equipped with a 30mm2 EDAX energydispersive spectrometer–silicon drift detector (EDS-SDD). Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Zircon was concentrated from samples for LA-ICP-MS analyses using standard mineral separation methods (crushing, sieving, water density separation, magnetic separation, and heavy liquids) as detailed in Strong and Driscoll (2016). Zircon was mounted in epoxy, ground to approximately half thickness of the grains, and polished prior to cathodoluminescence (CL) imaging using a JEOL 5800 LV scanning electron microscope (SEM).

For detrital zircon geochronology analyses (e.g., sample BonAirRRTracks), zircon was ablated with a Teledyne-Photon Machines Excite 193 nanometer argon-fluoride (ArF) excimer laser in spot mode. The laser spot sizes for zircon were ~25 µm. Each analysis consisted of 150 total bursts with a repetition rate of 5 Hz, laser energy of ~3 millijoules and an energy density of 4.11 Joules per square centimeter. Pit depths are typically less than 10 μm. The rate of helium carrier gas flow from the HelEx cell of the laser was ~0.6 L per minute (L/min). Make-up argon gas (~0.6 L/min) was added to the sample stream prior to its introduction into the plasma. Nitrogen with flow rate of 5.5 mL per minute was added to the sample stream to allow for significant reduction in ThO+/Th+ (<0.5%) and improved the ionization of refractory thorium (Hu et al., 2008). With the magnet centered at a constant mass, the flat tops of the isotope peaks of ²⁰²Hg, ²⁰⁴(Hg + Pb), ²⁰⁶Pb, ²⁰⁷Pb, ²⁰⁸Pb, ²³²Th, ²³⁵U, and ²³⁸U were measured by rapidly deflecting the ion beam with a 30-second on-peak background measured prior to each 30-second analysis. Raw data were reduced off-line using the Iolite 2.5 program (Paton et al., 2011) to subtract on-peak background signals, correct for U-Pb downhole fractionation, and normalize the instrumental mass bias using external mineral reference materials, the ages of which had previously been determined by isotope dilution–thermal ionization mass spectrometry. Ages were corrected by standard sample bracketing with the primary zircon reference material Temora2 (417 Ma; Black et al., 2004) and secondary reference materials FC-1 (1099 Ma; Paces and Miller, 1993) and Plešovice (337 Ma, Sláma et al., 2008). Raw data are shown in Table 1. For detrital zircon samples, reduced data were compiled into probability density plots using Isoplot 4.15 (Ludwig, 2012). ²⁰⁶Pb/²³⁸U ages are reported for zircons younger than ca. 1300 Ma and ²⁰⁷Pb/²⁰⁶Pb ages are used for older zircons following the recommendations of Gehrels (2012). Analyses that were greater than 10% discordant or more than 5% reverse discordant (²⁰⁶Pb/²³⁸U vs. ²⁰⁷Pb/²⁰⁶Pb) were excluded from interpretive analyses.

For igneous rock samples (e.g., Virginia Energy, Geology and Mineral Resources program sample R-11518; foliated metagranitoid), 30 spot analyses (25 μm) were conducted on zircon grains (after extraction, selection, and mounting). Spot analyses were selected based on zoning relationships determined in scanning electron microscope (SEM) cathodoluminescence (CL); imaging conducted at the USGS Microbeam facility in Denver, Colorado. Spots were selected based on zoning relationships determined in SEM imaging. Core and rim relationships were determined in several of the zircons analyzed. In some cases, both zircon cores and rims from the same zircon grain were analyzed to compare ages. Zircon analyses from this sample that were greater than 5% discordant or reverse discordant (²⁰⁷Pb/²³⁵U vs. ²⁰⁶Pb/²³⁸U) were excluded from interpretive analyses. Additionally, analyses that had U/Th ratios greater than 5 were excluded from interpretive analyses due to the possibility that they are metamorphic overgrowths. Raw data reduction was completed using Iolite v. 2.5 (Paton et al., 2011). Reduced data were then interpreted using the Excel macro Isoplot v. 4.15 (Ludwig, 2012). Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Zircon grains were extracted from six, ~5–10 kg samples at the USGS in Reston, Virginia. Samples were crushed and ground in a Sturtevant jaw crusher and Bico direct-drive disk mill, respectively, and then sieved to less than 250 μm (60 mesh). Particles larger than 250 μm were given a second pass through the disk mill. The fraction smaller than 250 μm was then passed over a Wilfley table to concentrate heavy minerals. For samples with abundant heavy minerals, the heavy fraction was sent over the Wilfley table a second time to reduce the volume for further processing. The heavy minerals were dried immediately on a hotplate. Following removal of the most magnetic material with a hand magnet, the sample was sent through a Frantz L1 magnetic separator and lithium heteropolytungstate (LST) heavy liquid (ρ = 2.85 g per cubic centimeter [g/cm³]) to further concentrate zircon. The order of these two steps varied depending on the mineral assemblage of the sample. The nonmagnetic fraction heavier than LST was then put through methylene iodide (ρ = 3.3 g/cm³) to produce a final concentrate. Zircons were hand-picked from this concentrate under incident and transmitted light using a Leica Z16 binocular microscope.

Selected zircon grains were mounted on double-sided Kapton tape and fixed with Struers EpoFix epoxy in a 1-inch round cylinder. The resulting 1-inch round mount was polished with 2500-grit wet-dry sandpaper and polished sequentially on a Struers LaboPol polisher using 6 µm and 1 µm diamond suspensions to expose internal textural features with a goal of polishing halfway into the grains (for more information, see https://shrimprg.stanford.edu/sample-types-and-sam-ple-preparation).

Prior to isotopic analysis, all grains were simultaneously imaged in backscattered electron (BSE) and panchromatic cathodoluminescence (CL) modes using a Hitachi SU-5000 FE-SEM scanning electron microscope at the USGS in Reston, Virginia. The panchromatic CL images were obtained using a Hitachi UVD detector with the bias turned off. Following sensitive highresolution ion microprobe-reverse geometry (SHRIMP-RG) isotopic analysis, the analyzed grains were re-imaged using a Delmic-Sparc CL detector fitted to the SU-5000 to confirm analysis locations using high-resolution CL imagery (Fig. 11).

Zircon grains from the six samples were analyzed by secondary ionization mass spectrometry on the U.S. Geological Survey (USGS)/Stanford SHRIMP-RG during three separate analytical sessions in 2018 and 2019. Sample DW-11 was analyzed in October of 2018; M19-02-28A, M19-02-28B, and M19-02-28C were analyzed in June of 2019; and samples JR-1-1 and Jack-1 were analyzed in October 2019. A similar analytical setup was used in each session and measured nine peak locations in five cycles on the single-collector SHRIMP-RG. The spot size for all analyses was ~20 μm in diameter and ~1 μm in depth, and the primary beam was ~3–6 nA. Zircon standard R33 (419 Ma; Black et al., 2004) mounted with the unknowns was used to correct ²⁰⁶Pb/²³⁸U ages for elemental fractionation and was run after every fourth unknown analysis. Raw data were reduced using Squid 2 (Ludwig, 2009) and plotted using Isoplot 3.75 (Ludwig, 2012). Calculated Pb/ U ages for these Paleozoic aged samples are reported at 2σ and are either concordia ages (Ludwig, 1980, 2003) or weighted-average ages; these data are shown in Table 2. Spot ages were considered concordant if the 2σ error ellipse overlapped concordia on a Wetherill plot (Spencer et al., 2016). Uncertainties on single spot analyses, when stated, are at 1σ. Analytical isotopic data for individual spot analyses are given in Table S2 (text ^{footnote 1}). Uranium and thorium concentrations are also reported for each analysis and are relative to analyses of the concentration standard MAD-559 (3940 ppm U, 483 ppm Th; Coble et al., 2018) that was mounted with each set of unknowns. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.