Subduction initiation recorded in the Dadeville Complex of Alabama and Georgia, southeastern United States

The Appalachian-Caledonian orogen formed in response to the closure of the Iapetus and Rheic Oceans and subsequent continental collisions producing the supercontinent Pangea. During convergence, sections of oceanic lithosphere were emplaced onto the continents and preserved along the >6500 km (from current-day southeastern United States to northern Norway and Sweden) Iapetan margin (Bird et al., 1971; Bird and Dewey, 1970; Hibbard et al., 2007; Pedersen and Furnes, 1991; Pedersen et al., 1988; Waldron et al., 1996). The oceanic record is more robust in the northern Appalachians and Caledonides, where well-exposed ophiolites (obducted oceanic crust) with near-complete lithospheric sections (e.g., the Bay of Islands and Betts Cove ophiolites, Canada, and the Solund-Stavfjord ophiolite, Norway) preserve a record of subduction zone processes in the Iapetus Ocean (Bédard, 1999; De Souza et al., 2008; Furnes et al., 1988; Olive et al., 1997; Oliver and McAlpine, 1998). In contrast, the southern Appalachians have a paucity of complete ophiolites, with oceanic rocks instead forming smaller complexes of mafic and ultramafic rocks (Crowley, 1976; Drake and Morgan, 1981; Guice et al., 2021; McElhaney and McSween, 1983; Misra and Conte, 1991; Mittwede, 1989; Tenthorey et al., 1996; Raymond et al., 2003, 2016; Peterson and Ryan, 2009; Spell and Norrell, 1990). The differences between the northern and southern sections of the Appalachian-Caledonian orogen have long been recognized, and studies have sought to correlate major events in the orogen’s history using ophiolites and mafic-ultramafic complexes as markers for suture zones between terranes and continents (Hibbard et al., 2007, and references therein). For these correlations to be effectively made, the specific tectonic formation settings of the southern Appalachian mafic-ultramafic complexes must first be determined.

Studies of a modern subduction zone—the Izu-Bonin-Mariana system—have resulted in a model for correlating magmatic products with their associated tectonic setting within the subduction system (Arculus et al., 2015; Barth and Gluhak, 2009; Dilek and Thy, 2009; Ishizuka et al., 2011; Ishikawa et al., 2002; Leng et al., 2012; Li et al., 2021; Pearce et al., 2015; Pearce and Reagan, 2019; Portnyagin et al., 1997; Reagan et al., 2010, 2019; Rioux et al., 2021; Shervais et al., 2004, 2019, 2021; Stern et al., 2012; Whattam and Stern, 2011; Yuan et al., 2005). In the Izu-Bonin-Mariana model, forearc basalts—formed from decompression melting of the mantle during subduction-triggered extension—are the first to erupt, forming the base of the forearc volcanic stratigraphy (Reagan et al., 2010; Pearce and Reagan, 2019), whereas boninites—formed from subsequent flux melting of the depleted mantle as the volcanic arc system is established—overlie the forearc basalts (Ishizuka et al., 2011; Reagan et al., 2019; Shervais et al., 2019, 2021). On the basis of geochemistry, suprasubduction zone ophiolites have been interpreted as representing the backarc, arc, and/or forearc regions of a subduction zone, or as capturing some combination of these settings within an evolving system (Dilek and Furnes, 2011, and references therein). Of these settings, the forearc lithosphere is the most widely recorded in Phanerozoic ophiolites (Stern et al., 2012), with recognition of this tectonic setting based on a distinctive up-stratigraphic record of volcanic evolution from forearc basalts to boninites. The Izu-Bonin-Mariana model can be utilized to identify ophiolites and mafic-ultramafic complexes that record subduction initiation and forearc spreading throughout the Appalachian-Caledonian orogen, allowing for temporal correlation of subduction initiation processes over >6500 km.

This study considered samples from the Dadeville Complex of Alabama and Georgia, the southernmost exposed sequence of mafic-ultramafic rocks in the Appalachian orogen. Whole-rock and mineral geochemical analyses coupled with U-Pb zircon geochronology were utilized to investigate the origin of the Dadeville Complex and to place it more clearly within the context of Appalachian tectonic history.

The southern Appalachian orogen can be subdivided into three domains based on differing tectonic origins: (1) the Laurentian realm, (2) the Iapetan realm, and (3) the peri-Gondwanan realm (Hibbard et al., 2007). The Laurentian realm (Rankin, 1994) encompasses the foreland and western Blue Ridge/Talladega terrane (Fig. 1A), which is composed of rocks formed on or adjacent to Laurentia that record the Grenville orogeny and the rift-to-drift sequences deposited during the breakup of eastern Rodinia (Hibbard et al., 2007). The Iapetan realm predominantly contains rocks that formed within the Iapetus Ocean—including oceanic lithosphere and island arcs. The Iapetan realm is separated into the Dunnage zone north of the New York embayment (the narrowest exposed portion of the orogen) and the Piedmont domain to the south (Hibbard et al., 2007). The peri-Gondwanan realm consists of Gondwana-derived terranes that were accreted to the Laurentian margin during the closure of the Iapetus and Rheic Oceans (Adams et al., 1995; Horton et al., 1989; Miller et al., 2006; Muller et al., 1989; Stewart et al., 1997). Notable peri-Gondwanan terranes include Ganderia, Avalonia, and Meguma in the northern Appalachians and the Carolina superterrane and Suwannee terrane in the southern Appalachians (Hibbard et al., 2007; Pollock et al., 2012; Rodgers, 1971; Williams and Hatcher, 1983).

The Piedmont domain of the southern Appalachians includes the Inner Piedmont and the eastern Blue Ridge terranes, which predominantly consist of metamorphosed clastic lithologies with rare magmatic arc and oceanic rocks (Coler et al., 2000; Hibbard et al., 2007; Horton et al., 1998; Seal and Kish, 1990). The Brevard fault zone (Fig. 1A) separates the Piedmont domain from the eastern Blue Ridge terrane (Hibbard et al., 2007; Spell and Norrell, 1990). The Inner Piedmont is a composite terrane that contains oceanic and magmatic arc rocks with depleted incompatible element and enriched fluid-mobile element signatures interpreted to suggest a subduction zone origin (Coler et al., 2000; Hibbard et al., 2007; Horton et al., 1998; Merschat et al., 2018; Seal and Kish, 1990). The eastern Blue Ridge is predominantly composed of deep-water sedimentary units; however, several interlayered metasedimentary and mafic/bimodal volcanic suites have collectively been interpreted to represent a ca. 470–430 Ma backarc basin, the Wedowee-Emuckfaw-Dahlonega basin, which extends from Alabama to North Carolina (Barineau et al., 2015; Tull et al., 2014). The Opelika Complex—located southeast of the Dadeville Complex—was originally assigned to the Inner Piedmont but has since been correlated with units to the northwest of the Dadeville Complex and reclassified as part of the eastern Blue Ridge (Stevens, 2018).

The Dadeville Complex is situated within the Inner Piedmont at the southernmost exposed end of the Appalachians (Fig. 1A). It consists of felsic and mafic metavolcanic rocks, felsic and mafic-ultramafic intrusions, and metasedimentary units (Fig. 1B; Steltenpohl et al., 2013; Tull et al., 2018). Bordered to the northwest by the Katy Creek fault (part of the Brevard fault zone) and to the southeast by the Stonewall Line fault (Tull et al., 2018; Vandervoort, 2016), the Dadeville Complex has been interpreted as a klippe within the Tallahassee synform, structurally overlying the Wedowee-Emuckfaw-Dahlonega basin units to the northwest and the Opelika Complex to the southeast (Stevens, 2018; Tull et al., 2018).

The basal unit of the Dadeville Complex is the Ropes Creek Amphibolite, which accounts for roughly 40% of the exposed outcrop (Tull et al., 2018). The Ropes Creek Amphibolite is a layered, basaltic amphibolite with subordinate amounts of intercalated dacitic volcanics and metasedimentary units (Tull et al., 2018), and it is interpreted as metamorphosed basalt flows formed in an extensional oceanic setting (Stow et al., 1984). Zircon Hf isotope values from an intercalated dacite layer in the Ropes Creek Amphibolite suggest involvement of a depleted mantle source during formation (Tull et al., 2018). The Ropes Creek Amphibolite occurs in close association with two other units of the Dadeville Complex, the Waresville Formation—recently interpreted as synonymous with the Ropes Creek Amphibolite (Farris et al., 2017)—and the andesitic-dacitic Waverly Gneiss, which is intercalated with the Ropes Creek Amphibolite in the eastern portion of the complex (Ma et al., 2019). The Ropes Creek Amphibolite and Waverly Gneiss units are named and mapped separately on the Geologic Map of Alabama (Osborne et al., 1989), but on the Geologic Map of Georgia, they are undifferentiated and collectively mapped as “hornblende gneiss/amphibolite” (Lawton et al., 1976).

Other major units of the Dadeville Complex include the Agricola Schist, the Camp Hill Gneiss, the Chattasofka Creek Gneiss (Rock Mills Gneiss or Franklin Gneiss in Georgia), and various small occurrences of mafic-ultramafic rocks. The uppermost unit—the Agricola Schist—is a pelitic to psammitic schist that records metamorphic conditions of 5–8 kbar and 600–700 °C (Drummond et al., 1997; Tull et al., 2018), and its sedimentary deposition has been linked to either an intra-arc basin or a cover sequence (Ma et al., 2019; Tull et al., 2018). The Camp Hill Gneiss—intrusive to the Ropes Creek Amphibolite and the Agricola Schist—is a trondhjemite-tonalite pluton that is interpreted as the product of partial melting of a basaltic protolith under middle- to upper-crustal pressures (Drummond et al., 1997; Neilson et al., 1997; Sterling, 2006). The Chattasofka Creek Gneiss—intrusive to the Ropes Creek Amphibolite, the Agricola Schist, and the Doss Mountain suite—is considered to be a syncollisional granite originating from a metapelitic protolith (Davis, 2021; Drummond et al., 1997; Nielson et al., 1997; Sterling, 2006). Rocks interpreted as mafic-ultramafic intrusions into the Ropes Creek Amphibolite occur as small bodies throughout the complex (Davis, 2021; Drummond et al., 1997; Nielson et al., 1997; Sterling, 2006). The largest of these mafic-ultramafic units, the Doss Mountain suite, is comprised of pyroxenite and gabbronorite lithologies (Davis, 2021; Farris et al., 2017; Neilson, 1983; Neilson and Bittner, 1990; Neilson and Stow, 1986).

Previous zircon U-Pb geochronology on Dadeville Complex units performed by laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) yielded crystallization dates of ca. 458 Ma, ca. 430 Ma, and ca. 465 Ma for the Ropes Creek Amphibolite/Waresville Formation, ca. 454 Ma for the Waverly Gneiss, ca. 480 Ma and ca. 448–446 Ma for the Camp Hill Gneiss, ca. 449 Ma for the Chattasofka Creek Gneiss, and ca. 394 Ma for a felsite within the Agricola Schist (Ma et al., 2019; Tull et al., 2018). Overgrowths on zircons from the Waverly Gneiss samples yielded a date of ca. 402 Ma, interpreted as the timing of peak metamorphism (Ma et al., 2019). Detrital zircon populations from the Agricola Schist and other metasedimentary units near the base of the Ropes Creek Amphibolite as well as felsic layers within other units showed a strong Grenvillian signature, indicating formation on or close to the Laurentian margin (Tull et al., 2018). Negative εNd values from the Doss Mountain suite and other mafic-ultramafic rocks suggest extraction from an evolved source and/or interaction with continental lithosphere (Tull et al., 2018).

Geochemical studies of the Doss Mountain suite, Camp Hill Gneiss, and Chattasofka Creek Gneiss indicated whole-rock major- and trace-element compositions that exhibit volcanic arc signatures (Neilson et al., 1997; Stow et al., 1984). When coupled with the similarly aged Wedowee-Emuckfaw-Dahlonega basin to its northwest, the Dadeville Complex has been hypothesized to represent the arc component of a paired arc-backarc system (Barineau et al., 2015; Tull et al., 2014). Taken with existing geochronology, the current interpretation is that the Dadeville Complex represents a dismembered volcanic arc that was accreted (with its conjugate backarc, the Wedowee-Emuckfaw-Dahlonega basin) onto Laurentia during Appalachian continental collision (Farris et al., 2017; Ma et al., 2019).

Forty-one samples were collected from the Dadeville Complex. Twenty-eight samples were collected from within the mapped regions of the Ropes Creek Amphibolite, the Waverly Gneiss, or unnamed mafic-ultramafic rocks (Fig. 1B; Neilson et al., 1997; Neilson and Stow, 1986; Stow et al., 1984). Ten samples were collected in situ from the Doss Mountain suite (Fig. 1C; Neilson and Stow, 1986), and two additional samples were collected as float. One sample was collected from the Easton Complex of Neilson and Stow (1986). The majority of the Dadeville Complex is heavily weathered, with fine- to medium-grained mafic units primarily consisting of saprolite with preserved corestones. The medium-grained Doss Mountain suite and mafic-ultramafic rocks are better preserved than the Ropes Creek Amphibolite and associated units. For additional descriptions of lithologic units and field relationships, see Neilson and Bittner (1990) and Farris et al. (2017).

Full details of the methods for bulk-rock geochemistry and geochronology are available in Supplemental Material Item B¹, with summaries presented here.

All samples had weathered materials removed and were crushed and powdered. A split of powder from each sample was sent to the Franklin and Marshall X-Ray Laboratory, where 0.4 g of powder was flux melted in the presence of lithium tetraborate and then quenched to produce a glass disc. Major-element analysis was performed on the glass disc using a Malvern PANalytical, Inc., Zetium X-ray fluorescence (XRF) spectrometer. Shards of the glass discs used for XRF analyses were mounted in 1” (2.5 cm) epoxy rounds, polished, and then analyzed for trace elements by LA-ICP-MS using a Teledyne-Cetac Analyte G2 193 nm laser coupled to an Agilent 8900 quadrupole ICP-MS in the TeMPO Laboratory, Johns Hopkins University (JHU). Data were collected using 600 µm line scans with a pre-ablation pass to remove surface contamination. Analyses were conducted using a scan speed of 10 µm/s, laser repetition rate of 10 Hz, fluence of 4 J/cm², and an analytical mask that produced a square analysis spot with dimensions 50 × 50 µm. Integration time for each isotope was 0.1 s. Prior to each line scan, baseline measurements were made for 15 s. Standard reference glasses NIST 610, NIST 612, AGV-2G, BCR-2G, and/or BHVO-2G (Jochum et al., 2005) were measured after every seven unknown analyses. Data were processed using Iolite v4 (Paton et al., 2011), employing the trace-element reduction scheme and using ⁴³Ca (determined using XRF) as the internal standard. NIST 612 was used as the primary standard for data reduction, and data accuracy was assessed using the additional standards (NIST 610, AGV-2, BCR-2G, and/or BHVO-2G). Estimates of uncertainties for measurements of all elements are reported in Supplemental Material Item A.

Samples selected for zircon U-Pb analysis were crushed using a stainless-steel ring-and-puck mill, sieved to <500 µm, and washed to remove clay-sized particles. Washed samples were subjected to magnetic separation using a Frantz magnetic separator targeting isolation of a highly nonmagnetic fraction likely to be zircon enriched. The highly nonmagnetic fraction was then subjected to density separation using a sodium polytungstate (SPT) heavy liquid medium following the method of Andò (2020). Heavy mineral separates were inspected, and zircons were picked and then annealed in a muffle furnace at 900 °C for 60 h. The annealed zircons were mounted and polished by hand to expose grain cores for analysis.

Polished mounts were carbon coated and then imaged with cathodoluminescence (CL) using a Deben Centaurus CL detector mounted on a Thermo Scientific Helios G4 UC scanning electron microscope in the Materials Characterization and Processing Facility, JHU. Mounted and CL-imaged zircons were analyzed by the aforementioned LA-ICP-MS instrumentation in the TeMPO Laboratory. Each analysis followed 25 s of washout and comprised three cleaning shots and 250 analytical shots, using a laser repetition rate of 10 Hz, fluence of 2 J/cm², and a square aperture of 20 × 20 µm or 40 × 40 µm. The ²⁰⁴Pb, ²⁰⁶Pb, ²⁰⁷Pb, ²⁰⁸Pb, ²³²Th, and ²³⁸U isotopes were measured, repeating a 1 s analytical cycle that used integration times of 0.1 s for the Th and U isotopes and 0.2 s for each of the Pb isotopes. Helium carrier gas flows were 0.38 L/min into cell and 0.35 L/min into the ablation cup, and Ar make-up gas was added to the analyte stream at a rate of 0.9 L/min prior to injection into the ICP-MS. “SQUID” tubing was used to smooth the signal at the detector. Standard reference zircons 91500 (1063.6 ± 0.3 Ma; Wiedenbeck et al., 1995; Schoene et al., 2006), Plešovice (337.13 ± 0.37 Ma; Sláma et al., 2008), Temora II (416.78 ± 0.33 Ma; Black et al., 2004), and FC-1 (1099.9 ± 1.1 Ma; Paces and Miller, 1993) were measured after every nine unknown analyses; the primary standard used for data reduction was 91500, with all others used to verify accuracy. Data reduction was performed in Iolite v4 (Paton et al., 2011) using a median fit to the standard data and including the U-Pb zircon geochronology down-hole fractionation correction (Paton et al., 2010). A long-term, laboratory-specific excess uncertainty of 2% was added in quadrature to isotope ratios obtained after data reduction to better represent inherent uncertainties in the data (method outlined in Horstwood et al., 2016). Concordia diagrams were plotted and weighted mean ²⁰⁶Pb/²³⁸U dates were calculated using IsoplotR (Vermeesch, 2018).

U-Pb dates were obtained by chemical abrasion–isotope dilution–thermal ionization mass spectrometry (CA-ID-TIMS) from analyses of single zircon grains, using the method modified after Mattinson (2005). CL images and LA-ICP-MS data were used to target zircons for TIMS analysis. Selected zircons were removed from the epoxy mounts and chemically abraded. The remaining zircon was spiked with the Boise State University mixed ²³³U-²³⁵U-²⁰⁵Pb tracer solution (BSU-1B). U and Pb were separated from the zircon matrix using an HCl-based anion-exchange chromatographic procedure (Krogh, 1973), eluted together, and dried with 2 µL of 0.05 N H₃PO₄. Pb and U were loaded on a single outgassed Re filament in 5 µL of a silica-gel/phosphoric acid mixture (Gerstenberger and Haase, 1997), and U and Pb isotopic measurements were made on a GV Isoprobe-T multicollector TIMS instrument equipped with an ion-counting Daly detector.

CA-ID-TIMS U-Pb dates and uncertainties were calculated using the algorithms of Schmitz and Schoene (2007), calibration of the BSU-1B tracer solution to ²³⁵U/²⁰⁵Pb = 77.93 and ²³³U/²³⁵U = 1.007066, U decay constants recommended by Jaffey et al. (1971), and ²³⁸U/²³⁵U = 137.818 (Hiess et al., 2012). The ²⁰⁶Pb/²³⁸U ratios and dates were corrected for initial ²³⁰Th disequilibrium using D_Th/U = 0.20 ± 0.05 (1 σ) and the algorithms of Crowley et al. (2007), resulting in an increase in the ²⁰⁶Pb/²³⁸U dates of ~0.09 Ma. A weighted mean ²⁰⁶Pb/²³⁸U date was calculated from equivalent dates (probability of fit >0.05) using Isoplot 3.0 (Ludwig, 2003) with error at the 95% confidence interval. Errors on dates from individual analyses are at 2σ. Full details of the methodology are provided in Supplemental Material Item A.

Modal mineralogy for the Dadeville Complex samples is detailed in Table 1.

Samples mapped as Ropes Creek Amphibolite and associated unnamed units in Georgia (AL20–04, AL20–05, AL20–38, AL20–41, AL20–43, AL20–47, AL22–14, AL22–18, AL22–21, AL22–27, AL22–28, AL22–29, AL22–33, AL22–34) have well-developed foliations defined by aligned amphiboles (Fig. 2A). These units comprise 30–70 modal % fine- to medium-grained amphibole and 20–55 modal % fine- to medium-grained plagioclase, with minor quartz veins and accessory epidote, clinozoisite, ilmenite/magnetite/chromite, and rutile. Samples AL20–06 and AL22–11 had low abundances of plagioclase (5 modal%) due to significant to epidotization (20 modal % epidote). Plagioclase was also variably altered to sericite, with alteration occurring as near-isotropic microcrystalline aggregates that displayed a “stringy” texture (Fig. 2A). Where present, epidote displayed a sieve texture.

The Waverly Gneiss samples (AL22–30, AL22–31, AL22–32, AL22–35) comprised 40–60 modal % amphibole, 10–30 modal % clinopyroxene (excluding AL22–31), and 20–40 modal % plagioclase. Samples AL22–30 and AL22–31 had 10–15 modal % quartz in veins. In these rocks, amphiboles exhibited chemical zonation evidenced by brown-to-green zoning from core to rim (Fig. 2B). These features were also present in three samples collected from sites mapped as Ropes Creek Amphibolite and unnamed amphibolites from Georgia (AL20–38, AL20–41, AL20–47).

Most rocks from the mapped Doss Mountain suite (AL20–23, AL20–24, AL20–26, AL20–27, AL20–28, AL20–29, AL20–30) exhibited no foliation. They comprised 25–90 modal % medium- to coarse-grained (up to 5 mm) pyroxene and 30–70 modal % medium-grained (up to 2 mm) plagioclase, with accessory ilmenite, rutile, zoisite, antigorite, garnet, and spinel. Pyroxenes showed alteration of varying degrees to fine-grained (<1 mm) amphibole, and plagioclase exhibited complex polysynthetic twinning (Fig. 2C). Sample AL20–24 comprised only pyroxene with ~10 modal % alteration to amphibole. Four Doss Mountain samples (AL20–23, AL20–27, AL20–29, AL20–30) contained both orthopyroxene and clinopyroxene and had garnet coronae developed between plagioclase and amphibole. Two samples differed significantly from the rest of the Doss Mountain rocks; AL22–26 shared petrographic characteristics with the Ropes Creek Amphibolite, and AL20–45 was comparable to the Waverly Gneiss.

Twelve samples were collected from units mapped as intrusive mafic-ultramafic rocks, and all but AL20–32 displayed characteristics resembling either the Doss Mountain samples or the Ropes Creek Amphibolite and associated unnamed amphibolites. Six of the samples shared characteristics with the Doss Mountain samples (AL20–15, AL20–19, AL20–20, AL20–31, AL20–36, AL20–37), including coarse-grained plagioclase with complex polysynthetic twinning and alteration of primary pyroxenes to fine-grained amphibole. Five samples (AL20–01, AL20–35, AL22–04, AL22–05, AL22–07) displayed “stringy” plagioclase alteration and well-developed foliation defined by aligned amphiboles, which resembled the Ropes Creek Amphibolite and associated unnamed amphibolite units from Georgia. Sample AL20–32 from the Easton Complex was unique in the sample set by consisting of relict grains of pyroxene that had been significantly altered to amphibole, having biotite present as small alteration patches within the amphiboles, and having two observed grains of olivine in thin section.

As described above, the mapped units in the Dadeville Complex displayed different petrographic textures and mineralogical compositions (Section 5.1). This section therefore subdivides samples according to rock description into group A (fine- to medium-grained, foliated, mafic-to-intermediate) and group B (medium-grained, nonfoliated, mafic) samples. In the sections below, the whole-rock major- and trace-element geochemistry of the Dadeville Complex samples is compared to that of the Izu-Bonin-Mariana forearc oceanic crust lavas (Ishizuka et al., 2011; Pearce and Reagan, 2019; Reagan et al., 2010, 2015; Shervais et al., 2021). Complete whole-rock geochemistry is detailed in Table 2.

Group A samples, which contained 3–18 wt% MgO, 43–56 wt% SiO₂, 0.1–3 wt% TiO₂, 4–20 wt% Al₂O₃, 5–19 wt% Fe₂O₃, 6–17 wt% CaO, ≤3.24 wt% Na₂O, and <0.9 wt% K₂O, generally showed significant overlap with the forearc rocks from the Izu-Bonin-Mariana but typically with lower SiO₂ and Na₂O values (Fig. 3A). Relative to group A samples, group B samples showed higher concentrations of MgO (5–26 wt%), lower concentrations of TiO₂ (<1 wt%), and overlapping concentrations of all other elements. Group B samples also showed significant overlap with the Izu-Bonin-Mariana data, but they typically exhibited slightly higher MgO and CaO contents and slightly lower SiO₂, Na₂O, K₂O, and P₂O₅ contents (Fig. 3A). The Dadeville Complex samples (except one) plot along the tholeiitic trend (Fig. 3D) on the alkalis-iron-magnesium (AFM) volcanic classification diagram (Irvine and Baragar, 1971).

The high field strength elements (HFSEs; Nb, Ta, Zr, and Hf), generally considered to be immobile during metamorphism and secondary alteration, are plotted against MgO (wt%) in the bivariate diagrams of Figure 3B. The Zr concentrations of group A samples span a large range, (1.39–298.78 ppm) with most having >25 ppm, while Zr concentrations of group B samples are lower (<10 ppm), with three exceptions (AL22–34, AL22–32, and AL20–30), which range 39.42–90.78 ppm. The Hf concentrations follow the same pattern as for Zr, with the group A samples having concentrations of 0.08–7.32 ppm, with most >0.28, and the group B samples having ≤0.34 ppm, with the exception of the same samples (AL22–34, AL22–32, and AL20–30), which range 1.21–2.18 ppm. The Nb concentrations for group A samples range 0.14–4.47 ppm, except for sample AL20–45 (20.73 ppm), while group B samples contain 0.02–4.17 ppm Nb. The Ta concentrations for group A samples range 0.01–0.33 ppm, except for sample AL20–45, which has a Ta concentration of 1.13 ppm. Group B samples contain 0–0.28 ppm Ta.

Figure 3C plots ratios of selected rare earth elements (REEs) against MgO (wt%). Group A samples have positive to negative heavy REE (HREE) slopes ([Gd/Lu]_N-MORB = 0.83–1.86, where N-MORB denotes normal mid-ocean-ridge basalt), positive to negative light REE (LREE) slopes ([La/Sm]_N-MORB = 0.69–4.05), and a range from positive to negative slopes across all REEs ([La/Lu]_N-MORB = 0.63–7.51). The group B samples have positive to negative HREE slopes ([Gd/Lu]_N-MORB = 0.6–2.42), negative LREE slopes ([La/Sm]_N-MORB = 1.13–4.21), and negative slopes across all REEs ([La/Lu]_N-MORB = 1.21–6.73).

LA-ICP-MS uncertainties provided below are 2s (sample standard deviation; after Horstwood et al., 2016). Three amphibolites yielded zircons for U-Pb analysis (Fig. 4). Two of the amphibolites (AL20–35 and AL22–14) were sampled ~2 km apart in the SW corner of the complex, with one collected from an area mapped as Ropes Creek Amphibolite and the other from part of the unnamed mafic-ultramafic rock unit (Fig. 1B). The zircons from AL20–35 were 100–200 µm and euhedral to subhedral, displayed oscillatory, sector, patchy, and/or spongy zoning, and had Th/U values of 0.16–0.22. LA-ICP-MS analysis of zircons from this sample yielded a 454.76 ± 3.55 Ma (2s) weighted mean ²⁰⁶Pb/²³⁸U date (n = 13; mean square weighted deviation [MSWD] = 1.1). The AL22–14 zircons were 200–400 µm, commonly fractured, and euhedral to anhedral, and they showed oscillatory, sector, cloudy, or weak zoning in CL, with some also showing thin metamorphic rims. The Th/U of the zircons for AL22–14 ranged 0.12–0.89. LA-ICP-MS analyses of zircons from this sample yielded a 464.87 ± 6.85 Ma (2σ) weighted mean ²⁰⁶Pb/²³⁸U date (n = 55/56; MSWD = 0.53). A third geochronology sample (AL22–34) was collected from the eastern side of the complex, in Georgia (Fig. 1B). The zircons from AL22–34 were 200–300 µm, fractured, and euhedral to anhedral and showed no, weak, cloudy, or patchy zoning in CL. The AL22–34 zircons had Th/U values of 0.38–1.28. LA-ICP-MS analysis of zircons from this sample yielded a 467.20 ± 16.1 Ma (2σ) weighted mean ²⁰⁶Pb/²³⁸U date (n = 10; MSWD = 0.52).

Eight zircon grains from previously analyzed (by LA-ICP-MS) sample AL20–35 were selected for CA-ID-TIMS analysis. The six oldest grains yielded a weighted mean age of 467.07 ± 0.13 Ma (95% confidence interval; MSWD = 1.5; Fig. 4). This is interpreted to date igneous crystallization. Two resolvable “younger” grains yielded dates of 466.27 ± 0.31 Ma (2σ) and 464.91 ± 0.33 Ma (2σ), which are interpreted to have retained (after chemical abrasion) domains that underwent Pb loss and/or to have small metamorphic rims.

To ensure that tectonomagmatic interpretations were made only on samples appropriate for use with common geochemical discrimination schemes, the meta-igneous Dadeville Complex samples were screened for evidence of significant alteration or compositional deviation from their original melt. This assessment included checks for cumulate effects (section 6.1.1) and postcrystallization mobility of the various elements (section 6.1.2). After screening, geochemical classifications (section 6.1.3) were based on (1) samples considered to be noncumulate in origin and (2) element groupings with demonstrably limited element mobility during metamorphism/metasomatism.

Geochemical signatures indicative of formation from cumulate processes were present in some samples collected from the Dadeville Complex. This included prominent positive Eu* anomalies (Eu* = Eu/[Sm + Gd]^0.5) and low total REE concentrations. Cumulate chemistry is not reflective of magma source; therefore, Eu* anomalies were utilized as a coarse proxy for identifying cumulate effects, and samples with Eu* anomalies <0.7 and >1.1 were not considered for the geochemical classification work outlined in section 6.1.3. Two additional samples with anomalously high Cr (>5000 ppm) and TiO₂ (>3 wt%) values—interpreted to signify accumulation of Ti- and/or Cr-rich minerals via crystal fractionation—were also excluded, leaving 21 samples interpreted as appropriate for use in tectonic interpretations.

The rocks of the southern Appalachian orogen have experienced postemplacement deformation and metamorphism up to granulite facies and may have experienced subsolidus element mobilization or open-system chemical modification. When using tectonic classification schemes, it is vital to determine whether the elements used have retained their original concentrations or if these elements were mobilized during subsequent metamorphism and/or hydrothermal alteration. To determine the extent of postcrystallization element mobilization within the Dadeville Complex, the samples were evaluated using bivariate diagrams that plot trace elements against the immobile element Y (Fig. 5; method outlined in Guice et al., 2018, 2019). The results showed that the typically fluid-mobile large ion lithophile elements (Ba, Cs, Rb, Sr) had low correlations with Y (R² = 0.33, 0.12, 0.27, and 0.29, respectively), indicating that their compositions have likely been altered by secondary processes. The HFSEs (Nb, Ta, Zr, Hf) showed strong correlations (R² = 0.89, 0.70, 0.92, and 0.77, respectively), suggesting limited mobility of these elements relative to Y. The LREEs (La, Ce, Pr, Nd, Sm, Eu) also exhibited moderate correlations with Y (R² = 0.39–0.81), while the HREEs (Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu) showed strong correlations with Y (R² > 0.81), suggesting that these elements were highly immobile relative to Y. Other elements commonly utilized as tectonic discriminators for oceanic basaltic rocks include Th, Ti, V, Mg, and Si. In the Dadeville samples, Th and Ti showed poor correlations with Y (R² = 0.36 and 0.19, respectively), and V showed no correlation with Y (R² < 0.02). Using Cr as an immobile element proxy for MgO, there was evidence that the Dadeville Complex samples have also experienced secondary Si and Mg mobility (procedure outlined in Pearce and Reagan, 2019). Based on the element mobility analysis, geochemical classification schemes for the Dadeville Complex should be limited to use of the HFSEs and REEs as discriminators of tectonomagmatic setting and evolution.

The volcanic samples from the Dadeville Complex can be subdivided into three groups based on their HFSE and REE characteristics. These groupings are best illustrated with chondrite-normalized REE diagrams and N-MORB–normalized trace-element diagrams (Figs. 6A and 6B). Group 1 (n = 9) samples have features resembling Izu-Bonin-Mariana forearc basalts, with high total HREE values (∑[Gd–Lu]_N = 9.62–25.86 ppm), depleted LREEs with respect to N-MORB, positive or flat LREE slopes, and flat HREE slopes. Group 2 samples (n = 7) are geochemically comparable to Izu-Bonin-Mariana boninites, showing depleted total HREE values (∑[Gd–Lu]_N = 1.97–7.00 ppm), negative LREE slopes and flat HREE slopes, and distinctive Nb-Ta and Zr-Hf depletions. We characterize the group 3 samples (n = 2) as island-arc tholeiites, displaying high total HREE values (∑[Gd–Lu]_N = 15.64–33.69 ppm), negative LREE slopes, and negative overall REE slopes. A volcanic assemblage of forearc basalts, boninites, and island-arc tholeiites suggests that the Dadeville Complex may record subduction initiation and early protoarc development within the Iapetus Ocean (Ishizuka et al., 2011; Reagan et al., 2010, 2019; Pearce and Reagan, 2019; Shervais et al., 2019, 2021; Stern et al., 2012).

All nine of the forearc basalts samples occur exclusively in the southeastern parts of the Dadeville Complex (Fig. 7). Three of the boninites samples are in the northwestern section, adjacent to Doss Mountain, with the remaining two located in the southeastern section. Three island-arc tholeiite samples are located with the boninites, close to Doss Mountain, three are in the southeastern section, and one is in the northeastern section in Georgia. When compared to findings from the Izu-Bonin-Mariana system, the distribution of geochemistry within the Dadeville Complex volcanics—forearc basalts in the SE to boninites in the NW—suggests that volcanic rocks of the Dadeville Complex may be older (stratigraphically lower) in the southeast and younger (stratigraphically higher) in the northwest. This is consistent with previous interpretations based on structural relationships, which suggested that the northwest section of the Dadeville Complex is structurally higher than the southeast (Tull et al., 2018).

Our 467.07 ± 0.13 Ma CA-ID-TIMS U-Pb zircon age from a boninite sample, when combined with the identical LA-ICP-MS U-Pb zircon dates from the two other samples of island-arc tholeiite Ropes Creek Amphibolite (and related units), is interpreted to date forearc/protoarc volcanism in the Dadeville Complex. This 467 Ma date is younger than some previously published LA-ICP-MS U-Pb zircon dates for the Camp Hill Gneiss, Chattasofka Creek Gneiss, Waverly Gneiss, Waresville Formation, and Ropes Creek Amphibolite (Ma et al., 2019; Tull et al., 2018). Each of these previous dates were obtained from felsic units intercalated with or intruding the mafic units that were dated in this study. Previous geochronological interpretations utilized LA-ICP-MS U-Pb zircon dates, which are associated with greater uncertainty than the CA-ID-TIMS approach (e.g., Schaltegger et al., 2015). We interpret the 467.07 ± 0.13 CA-ID-TIMS date to record the early stages of tectonic convergence (subduction initiation and protoarc development in the Iapetus Ocean) in the Alabama-Georgia sector of the southern Appalachians.

Felsic plutonic and extrusive rocks of the Wedowee-Emuckfaw-Dahlonega basin (Fig. 8) are dated at ca. 470–430 Ma (Barineau et al., 2015, 2022; Holm-Denoma and Das, 2010; Ma et al., 2019; McClellan et al., 2007; Thomas, 2001; Tull et al., 2007, 2018), with only one unit—the 482 ± 7 Ma Cane Creek felsic gneiss of the Sally Free Mafic Complex—yielding an anomalously older date (Bream, 2003; Settles, 2002). The 470–430 Ma age range for units of the Wedowee-Emuckfaw-Dahlonega basin is suggestive of initiation of the Wedowee-Emuckfaw-Dahlonega basin backarc soon after the establishment of Dadeville Complex forearc/protoarc at 467 Ma. Cessation of volcanism in the Dadeville Complex and Wedowee-Emuckfaw-Dahlonega basin may have been associated with accretion of the Carolina superterrane, resulting in a regional metamorphic event at ca. 400 Ma (Hibbard, 2000; Ma et al., 2019).

Although contacts are poorly exposed and generally inferred, the medium-grained, nonfoliated, mafic samples in the Doss Mountain suite and throughout the rest of the Dadeville Complex have previously been interpreted as intrusive to the Ropes Creek Amphibolite (Neilson and Stow, 1986). Petrographic analysis of the Doss Mountain rocks revealed that four of the seven samples contained co-occurring orthopyroxene, clinopyroxene, and coronitic garnet, features suggestive of metamorphic reaction at granulite facies (St-Onge and Ijewliw, 1996). Peak metamorphic conditions in the Dadeville Complex—based on metamorphic mineral assemblages—have been reported as 5–8 kbar and 600–700 °C for the Agricola Schist and 10 kbar and 750–800 °C for the Ropes Creek Amphibolite (Drummond et al., 1997). Microprobe major-element analysis and backscattered-electron (BSE) imaging were performed on six samples that had potential granulite-facies textures (Fig. 2E). Two-pyroxene thermometry using the calibration of Brey and Köhler (1990) provided multiple ranges of temperatures for the samples (see Supplemental Material Item B). The calculations had large errors, and the partition coefficients indicated that the compositions may not be in equilibrium (KD_[Fe-Mg] = 0.4–0.8), precluding precise determination of the temperatures reached by the Doss Mountain samples; however, the data indicated that peak pressure-temperature (P-T) conditions recorded in the Doss Mountain suite exceeded 7 kbar and 760 °C, consistent with estimates for other units of the Dadeville Complex (Drummond et al., 1997). BSE imaging revealed that the garnet coronae formed at the interface between plagioclase and the amphibolitized rims of clinopyroxene, a texture associated with rehydration of granulite-facies rocks during retrogression to amphibolite-facies assemblages (St-Onge and Ijewliw, 1996). Additionally, garnet was present in some amphibolites sampled from the Ropes Creek Amphibolite, indicating metamorphic pressures exceeding 7 kbar, i.e., the lower estimate of pressure for garnet stability in metabasic rocks (Green et al., 2016; Wei and Duan, 2019). Further work is needed to fully constrain the metamorphic history of the Dadeville Complex rocks and may provide important insights into the tectonic evolution of the southern Appalachians.

Prevailing tectonic models for the northern Appalachians (Hibbard et al., 2007; van Staal and Barr, 2012) suggest that subduction initiated with eastward dip (present-day reference) at ca. 500–490 Ma, followed by terrane accretion and subsequent subduction polarity reversal at ca. 480 Ma. These models can account for more complete ophiolite sequences in the northern Appalachians, including preservation of mantle tectonites, layered ultramafics, a gabbroic section, sheeted dikes, and a metamorphic sole. The geometry of this model—where the forearc section formed during subduction initiation and was primed for obduction onto a peri-Laurentian terrane along the subduction thrust of the eastward-dipping subduction zone—can account for such complete preservation in the northern Appalachians (Stern, 2004; Stern et al., 2012). In the northern Appalachians, obduction (overthrusting) of the complexes also resulted in only greenschist-facies to lower-amphibolite-facies conditions, differing significantly from the higher P-T conditions typically recorded by mafic-ultramafic complexes of the southern Appalachians (Anderson and Moecher, 2009).

To explain the lesser abundance, inferior preservation, and higher metamorphic grade of the southern Appalachian mafic-ultramafic oceanic rocks, we propose a model wherein portions of forearc lithosphere were tectonically eroded, underthrusted, and/or carried to depth via subduction, to be exhumed during later tectonic events. According to this model, the divergent fates of forearc lithosphere from the northern and southern Appalachians can be reconciled by the polarity of the initiating subduction zone: The southern Appalachian mafic-ultramafic rocks record forearc lithosphere formed above a westward-dipping (rather than eastward-dipping) subduction zone in the Iapetus Ocean. This model (illustrated in Fig. 9) is consistent with previous interpretations of the Dadeville Complex having formed on the Laurentian (continental) side of the subduction trench, above a westward-dipping subduction zone (Barineau et al., 2015; Tull et al., 2014, 2018). The development of the similarly aged Wedowee-Emuckfaw-Dahlonega basin backarc in the overriding plate is also consistent with westward subduction of the Iapetan lithosphere under continental lithosphere of the Laurentian margin (Barineau et al., 2015, 2022; Tull et al., 2014, 2018).

1. The Dadeville Complex represents a sequence of forearc basalts, boninites, and island-arc tholeiites.

2. The up-stratigraphic-section evolution in geochemistry from forearc basalt to boninite is consistent with the forearc/protoarc volcanic rocks found in the Izu-Bonin-Mariana forearc and other suprasubduction zone ophiolites.

3. A CA-ID-TIMS U-Pb zircon date of 467 ± 0.13 Ma (2σ) for the Dadeville Complex is interpreted to date subduction initiation (and forearc/protoarc spreading in the Iapetus Ocean).

4. Granulite-facies conditions of >7 kbar and >760 °C are recorded by the Dadeville Complex rocks, suggesting deep underthrusting of the forearc/protoarc sequence (Dadeville Complex) during the subsequent convergence history.

5. We propose that earliest subduction in the Alabama-Georgia sector of the Appalachians was westward-dipping (current coordinates) and developed proximal to the Laurentian margin.

We thank Wentao Cao, Chong Ma, and James Tull for their thorough reviews, which improved the manuscript. Dana Brenner, Josh Garber, and Stanley Mertzman are thanked for their assistance with whole-rock laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) and X-ray fluorescence (XRF) analysis. Mark Caddick is thanked for providing access to the Virginia Tech laboratory for additional major-element XRF analysis.

Funding for this project was provided by a National Science Foundation CAREER Grant (NSF-2042631), Johns Hopkins University (JHU) Catalyst Grant, Hopkins Extreme Materials Institute Seed Grant, JHU Department of Earth and Planetary Sciences Field Research Grant, Explorers Club Washington Group Exploration and Field Research Grant, Geologic Society of America Graduate Student Research Grant, and JHU Krieger School of Arts and Sciences J. Brien Key Fund Award.

We acknowledge the historical and continuing connection between Indigenous people and their native lands and specifically acknowledge that our research in the Dadeville Complex was performed on the stolen ancestral lands of the peoples of the Muscogee Nation (https://www.muscogeenation.com). We recognize their historical and ongoing resistance to settler occupation and offer this land acknowledgment to promote reflection and support of Indigenous movements for identity, freedom, and self-determination.