High-silica rhyolites, ubiquitous features of continental volcanism, continue to evoke controversy as to their petrogenesis and evolution. We utilized the geochemical characteristics of late Vendian high-silica rhyolites erupted in the Catoctin Volcanic Province at South Mountain in Pennsylvania to probe the origin of the parental magmas and assess heterogeneities in the subsequent fractionation paths. We identified high- and low-Ti signatures within the South Mountain rhyolites, a common feature in many large igneous provinces, and these signatures are suggestive of a genetic link between basalts and rhyolites erupted in the Catoctin Volcanic Province. Two evolutionary trends are superimposed on the Ti-based subdivisions that reflect variable control of plagioclase and amphibole in the fractionating assemblage of the South Mountain rhyolites. Such distinctive evolutionary trends are evident in rhyolites from other tectonic settings (e.g., arcs), where they have been interpreted in terms of cold-wet and hot-dry conditions within the differentiating magmas. We interpret the amphibole-dominated fractionation path of the South Mountain rhyolites as following a cold-wet fractionation path compared to the hot-dry plagioclase-dominated trends. This study, which examines the geochemical implications of cryptic amphibole fractionation, has implications for assessing the role of amphibole and volatile content in the development of rhyolites in other large igneous provinces.

The driving mechanisms responsible for large igneous province formation remain a topic of considerable debate, although a mantle plume origin is now widely accepted (see Campbell, 2007, and many references therein). Other models for the genesis of large igneous provinces invoke processes such as interaction between asthenosphere-derived magma and subcontinental lithosphere (Ellam and Cox, 1991), melting of an enriched subcontinental lithosphere (Pegram, 1990; Hergt et al., 1991; Jourdan et al., 2009), upper-mantle water saturation through subduction (Ivanov et al., 2008), lithosphere dehydration (Gallagher and Hawkesworth, 1992), lateral temperature gradients associated with convective transport of hot mantle (Mutter et al., 1988; Anderson, 1994), delamination of the subcontinental lithosphere (Camp and Hanan, 2008), mixing of melts derived from different mantle reservoirs (Korenaga, 2004), and crustal contamination of mantle-derived melts (Lightfoot et al., 1990; Brandon et al., 1993). Continental flood basalts are frequently used to assess processes associated with large igneous province formation but are typically not primary in composition. Continental flood basalts experience significant fractionation and some assimilation prior to eruption (e.g., Pik et al., 1999; Kieffer et al., 2004), and, in some cases, these processes may dominate the geochemical variability of the erupted products (Thompson et al., 2007). While geochemical studies of continental flood basalt provinces therefore have typically focused on the least differentiated samples in order to probe mantle processes and heterogeneity (e.g., Pik et al., 1999), the record of plume-lithosphere interaction may be most effectively preserved in rocks with more evolved compositions. Rhyolite magmas erupted in large igneous provinces preserve an important record of these lithospheric processes, and therefore they play a key role in our understanding of the development of large igneous provinces.

Increasing awareness of the widespread occurrence of silicic magmas in large igneous provinces and the discovery of some dominantly silicic large igneous provinces (e.g., Pankhurst et al., 1998; Bryan, 2007) have highlighted the role of these magmas in probing plume-lithosphere interaction (e.g., Garland et al., 1995; Baker et al., 2000; Pankhurst et al., 2000b; Riley et al., 2001; Bryan et al., 2002). Large igneous province–related silicic magmatism, while geographically widespread, is particularly focused along rifts and the rift margins dissecting large igneous provinces such as Parana/Etendeka (Marsh et al., 2001), and Patagonia/Antarctic Peninsula (Pankhurst et al., 2000b; Riley et al., 2001). Continental extension is therefore considered to be central to the generation of large volumes of silicic magmas (Bryan et al., 2002). The significant volumes of silicic magmatism typically associated with mafic large igneous provinces (>104 km3) and silicic large igneous provinces (>106 km3) present a particular challenge in explaining how large volumes of silicic magma can be rapidly generated (Bryan et al., 2002).

The origin and evolution of large volumes of silicic magmas remain among the most controversial topics in modern petrology (Bachmann and Bergantz, 2004, 2008; Glazner et al., 2008; Brophy, 2009). Most models emphasize the role of assimilation, fractional crystallization, partial melting, and various combinations of these processes in the petrogenesis of such silicic magmas. Recent advances in numerical modeling (Dufek and Bergantz, 2005; Annen et al., 2006), and trace-element characterization (Bachmann and Bergantz, 2004; Davidson et al., 2007b; Bachmann and Bergantz, 2008; Glazner et al., 2008) have illustrated the importance of fractionating phases in controlling trace-element patterns in high-silica rhyolites, regardless of how the parental melt was generated (partial melting or crystal fractionation). In this contribution, we examine the trace-element characteristics of high-silica rhyolites from South Mountain, Pennsylvania, in order to probe the conditions of magma evolution in the late Vendian (564 Ma ± 9 Ma: Aleinikoff et al., 1995) Catoctin Volcanic Province, linked to the late Neoproterozoic eastern Laurentian superplume (Puffer, 2002). This study will examine parental magma heterogeneity and evolutionary paths preserved within the South Mountain rhyolite suite.

The South Mountain rhyolites occur in association with flood basalts of the Catoctin Volcanic Province in the Central Appalachians, and they are exposed along the Blue Ridge anticlinorium (Reed, 1955; Rankin, 1975; Badger and Sinha, 1988; Badger and Sinha, 2004). The Catoctin Volcanic Province had a significant areal extent of at least 11,000 km2, and has been linked with other temporally synchronous volcanic and plutonic rocks displaying an ocean-island basalt signature along the eastern margin of North America (from Newfoundland to Virginia; Puffer, 2002). It has been well established that the Catoctin flood basalts are associated with the breakup of Rodinia and the opening of the Iapetus Ocean (Rankin, 1975; Cawood et al., 2001). The significant areal extent and magnitude of volcanism along this rifted margin, combined with the recognition of the role of mantle plumes in assisting continental breakup globally (e.g., Hill, 1991), and also elsewhere in Rodinia (Li et al., 1999), prompted new models linking the rifting of Rodinia to the impingement of a superplume at the base of the continental lithosphere (Puffer, 2002; Li et al., 2008).

Within the Blue Ridge Province, silicic volcanism is most well developed at the southern and northern terminations, with more mafic volcanism dominant in the central region (Fig. 1). The Mount Rogers rhyolites (ca. 759 Ma: Aleinikoff et al., 1995) are thought to represent an initial (unsuccessful) rifting episode that was followed by the development of the Neoproterozoic Laurentian continental margin at ca. 577–550 Ma, coincident with the eruption of the Catoctin Volcanic Province (Badger and Sinha, 1988; Rankin, 1994; Aleinikoff et al., 1995; Novak and Rankin, 2004; Tollo et al., 2004). We focus on the northern portion of the Catoctin Volcanic Province in Pennsylvania, which is dominated by silicic lavas (South Mountain rhyolites) that become progressively less abundant southward (Reed, 1955; Badger and Sinha, 1992; Aleinikoff et al., 1995). The South Mountain rhyolites are interpreted to be either metamorphosed glassy flows or welded tuffs (Fauth, 1968; Aleinikoff et al., 1995) that are phenocryst poor (10%–20% of the rock volume), with feldspar (albite), quartz, biotite, ilmenite, and specular hematite (after magnetite) as phenocryst phases (Fauth, 1968; Mitra and Sinha, 2004a, 2004b).

We selected 25 samples of the least-altered metarhyolites from 255 km2 of metarhyolites exposed in the South Mountain region of southern Pennsylvania (Fig. 1; Table DR11). Samples were carefully trimmed, jaw crushed, and handpicked under a binocular microscope to avoid alteration and then powdered in a ceramic Bico flat-plate grinder. The sample powders were fused into lithium tetraborate glass disks using the procedures outlined elsewhere (Deering et al., 2008). Major elements, Zr, Sr, Rb, and Ni were analyzed by Brucker X-ray fluorescence (XRF), and the balance of the trace elements were obtained by laser ablation using Cetac LSX-200 and Micromass Platform inductively coupled plasma–mass spectrometry (ICP-MS). Trace-element reproducibility based on standard analyses was typically better than 5%. XRF analyses are presented in Table 1, and results of ICP-MS trace-element analyses are presented in Table 2.

The South Mountain rhyolites are typically peraluminous, high in silica (>72% SiO2), and define two groupings based on heterogeneity in major- and trace-element concentrations and distinctive trace-element differentiation trends. These two groups are well defined in terms of variable concentrations of Al2O3, Sr, and Ba (Fig. 2). The low-Al group (<12 wt% Al2O3) has Sr (7–41 ppm) and Ba (50–628 ppm) concentrations that are lower than the high-Al group (>12 wt% Al2O3), where Sr (25–56 ppm) and Ba (386–857 ppm) are more enriched. Most samples in the low-Al group have higher values of SiO2 (75–78 wt%) in comparison to the high-Al group (SiO2 ∼72–75 wt%). Major-element variation of the South Mountain rhyolites is similar to that observed in the high-silica rhyolites erupted at Mount Rogers, with the exception of TiO2, where some overlap occurs but Mount Rogers samples have lower values (Fig. 2; Novak and Rankin, 2004). Rare earth element (REE)–SiO2 differentiation trends (in particular Y, Dy, and Yb) exhibit the most distinctive differences: REE concentrations decrease with increasing SiO2 more rapidly within the high-Al group in comparison to the low-Al group, excluding Eu (Fig. 3). For most trace elements, such as Y, Nb, Yb, and Zr (excluding Sr, Ba, and Eu—not shown), the concentrations within the high-Al group are lower than that of the low-Al group at similar SiO2 contents (Fig. 3). Neither the high- or low-Al group clusters consistently with samples from the Mount Rogers formation, but there is broad overlap in many element concentrations between the Mount Rogers and South Mountain, with the exception of Nb and Yb (Fig. 3). Chondrite-normalized REE patterns reveal further differences between the two groups: (1) the most pronounced Eu anomalies occur within the low-Al group, and (2) the middle rare earth elements (MREE) are more depleted within the high-Al group (Fig. 4). Primitive mantle–normalized trace-element patterns are broadly similar, though variations in Sr, P, and Ba are particularly apparent (Fig. 5). Within the low-Al group, further heterogeneity is observed in terms of high field strength elements (HFSE): A subset of the low-Al group plots at higher Nb for a given Zr. This variation is typical in many large igneous provinces, where rhyolites may be broadly divided into high-titanium (HT) and low-titanium (LT) groups on the basis of HFSE heterogeneity (Garland et al., 1995; Ayalew et al., 2002). It is important to note that this LT/HT division does not correlate with the previously defined variations in Al2O3; both HT and LT varieties occur within the low-Al group. While the data are limited, the Mount Rogers rhyolites typically occur at very low levels of Nb/Y, consistent with their relative depletion in Nb in comparison to South Mountain.

The observed trace-element variations are best explained as reflecting primary magmatic processes rather than secondary alteration. For example, a comparison among Y and Nb and La shows trends that are consistent with magmatic processes. Y is not decoupled from Nb and La (r2 = 0.34, r2 = 0.36, respectively), which are thought to be less mobile during secondary alteration processes (Price et al., 1991; Cotten et al., 1995). This inference is further supported by the lack of strong negative Ce anomalies in all samples (Fig. 4), which can be caused by the formation of Ce4+ under oxidizing conditions of surface environments, while other REEs remain in the trivalent state (Class and le Roex, 2008), resulting in preferential mobilization of Ce. Samples that do exhibit some Ce mobility (Fig. 4) do not, however, define any of the trends and groupings described here (Briggs et al., 2008).

Origin of Rhyolites in Mafic and Silicic Large Igneous Provinces

The ultimate origin of rhyolite magmas in mafic and silicic large igneous provinces remains a controversial topic, and there are three dominant hypotheses: (1) melting of existing crust, (2) melting of flood basalts or underplated material, and (3) open-system fractional crystallization.

Origin of Major- and Trace-Element Variations at South Mountain

We interpret the geochemical variations evident within the South Mountain rhyolites in terms of open-system crystal fractionation, consistent with most existing models for rhyolite origin in other dominantly mafic large igneous provinces. The parental magmas of the South Mountain rhyolites are inferred to have a broad subdivision into HT and LT varieties based upon the preserved HT and LT rhyolites trends evident in the South Mountain suite. Samples defined as HT (higher Nb at a given Zr concentration; Fig. 6) are restricted to the region north of 40°N, though LT varieties occur throughout the study area. The origin of LT and HT magmas within flood basalt provinces remains unresolved but may be explained in terms of a broad heterogeneous upwelling (Kieffer et al., 2004), zoned mantle plumes (Pik et al., 1999), or contributions from both an upwelling plume and the lithospheric mantle (Greene et al., 2009). Deducing the precise origin of LT and HT magmatism within the South Mountain magma suite requires further isotopic constraints and lies beyond the scope of this study. The observed division of the South Mountain rhyolites into HT and LT parental magmas does not account for the dominant geochemical heterogeneity of high- and low Al groups observed within the South Mountain rhyolites, which instead must have developed during open-system crystal fractionation processes.

Crystal fractionation is considered to be the dominant control of trace-element variations in high-silica rhyolites (Bachmann and Bergantz, 2008), and variations in the relative proportions of plagioclase and clinopyroxene and/or amphibole in fractionating assemblages of magmas evolving in similar environments is frequently invoked to account for differences in the concentration of Eu, Al2O3, and Sr (e.g., Ayalew et al., 2002). Variability in CaO, Al2O3, Ba, and Na between the high- and low-Al2O3 groups at South Mountain points to alkali-feldspar and plagioclase control, respectively. Decreasing Dy/Yb with increasing fractionation index (e.g., SiO2; Fig. 7) is observed within the high-Al South Mountain rhyolites and is commonly attributed to crystal fractionation involving amphibole (+titanite) over pyroxene (Macpherson et al., 2006; Davidson et al., 2007a; Brophy, 2009). On the basis of our observations, it is clear that the high-Al samples from South Mountain may be attributed to more amphibole and less feldspar removal than the low-Al samples. Our interpretation that the South Mountain rhyolites were derived from open-system fractional crystallization is consistent with similar interpretations made for the Mount Rogers high-silica rhyolites (Novak and Rankin, 2004).

Amphibole Fractionation in Other Large Igneous Provinces

Amphibole has been observed in both peraluminous and peralkaline large igneous province magmas (Riley et al., 2001; Ayalew et al., 2002; Peate et al., 2005), though the role of amphibole as a fractionating phase is frequently overlooked because it is generally only a minor constituent of the modal assemblage. Increasingly, however, the role of cryptic fractionation has been recognized in the geochemical evolution of arc magmas, where the correlation of Dy/Yb with SiO2 is strong evidence for amphibole fractionation (Davidson et al., 2007a). Application of these geochemical tools to rhyolites erupted in large igneous provinces may identify other instances of amphibole fractionation. The Wegel Tena, Jima, Lima Limo, and Debre Birhan ignimbrites on the Ethiopian Plateau are associated with the eruption of the African-Arabian flood basalt province and broadly form two groups divisible in terms of Sr enrichment (dividing at ∼50 ppm). The majority of the variation between the low- and high-Sr groups (which mirror the high- and low-Al groups of South Mountain) is explained in terms of variable ratios of feldspar:clinopyroxene in the fractionating assemblage (Ayalew et al., 2002); however, the strong correlation in terms of Dy/Yb versus SiO2 in the high-Sr samples (Fig. 8) suggests that in addition to the plagioclase and pyroxene, some amphibole fractionation has occurred.

Amphibole and the Implications for the Volatile Content of Silicic Large Igneous Province Magmas

The interpreted presence of amphibole fractionation from the peralkaline Ethiopian ignimbrites and the peraluminous South Mountain rhyolites (Fig. 9) has implications for the magmatic conditions during rhyolite magma evolution in large igneous provinces. The distinctive REE patterns defined by the low- and high-Al2O3 groups from South Mountain mirror those of rhyolites from rifted arcs (Bachmann and Bergantz, 2008; Deering et al., 2008). In such environments, the evolution of the REE chemistry is determined by whether the magma is either related to a hot-dry-reduced or cold-wet-oxidized source (Bachmann and Bergantz, 2008; Christiansen and McCurry, 2008; Deering et al., 2008). Hot-dry-reduced magmas fractionate olivine, plagioclase, and pyroxene, producing deep Eu anomalies, while cold-wet-oxidized magmas are dominated by amphibole and titanite fractionation, producing less significant Eu anomalies but a depletion in the MREEs (Davidson et al., 2007b; Bachmann and Bergantz, 2008; Glazner et al., 2008). The transition between these magma types has been observed within the same magmatic center in the Taupo volcanic zone (New Zealand) and is attributed to changing P-T-fO2-fH2O conditions in the intermediate mush (Deering et al., 2008). The implication of these similarities in geochemical variations is that some peraluminous magmas at South Mountain had sufficiently elevated water content (greater than ∼4%) to allow the stabilization of amphibole. While large igneous provinces and associated continental rifts are frequently thought of as ostensibly dry environments, some basalts from large igneous provinces may contain magmatic amphibole (Kieffer et al., 2004), and erupted silicic magmas may contain up to 5% H2O (Webster, 1992). The origin of the hypothesized elevated water content in the high-Al group remains unclear; however, it may relate to either heterogeneity in the volatile content of the parental basaltic magma or assimilation of hydrothermally altered crustal rocks. Oxygen isotope studies of large igneous provinces (e.g., Bindeman et al., 2008), continental extensional environments (Bindeman and Valley, 2003), and island arcs (Bindeman et al., 2001) have all shown that silicic magmas erupted in these settings can assimilate significant volumes of hydrothermally altered rock.

Amphibole and Volatiles in Peralkaline Magmas

The existence of amphibole in large igneous province or rift environments is frequently linked to peralkaline magmas (MacDonald et al., 2008; Marshall et al., 2009). Variation in terms of amphibole stability and volatile content in peralkaline magmas differs markedly from peraluminous and metaluminous magmas. Fluorine has long been recognized as a key volatile in controlling the stability of amphibole in water-poor magmatic systems (e.g., Grigoriev and Iskull, 1937; Wones and Gilbert, 1982) and is an important volatile in large igneous provinces (e.g., Yirgu et al., 1999). For mildly peralkaline magmas, F-rich amphibole replaces clinopyroxene at low fH2O, contrary to phase relations observed in peraluminous and metaluminous magmas (Scaillet and MacDonald, 2001). Peralkaline magmas also exhibit an inhibition of plagioclase crystallization in favor of alkali feldspar, regardless of CaO or volatile content of the magma (Scaillet and MacDonald, 2001). These phase relations have significant consequences for the trace-element characteristics of the magmas erupted in large igneous province and rift settings; mildly peralkaline magmas with low fO2 may exhibit less pronounced Eu anomalies (e.g., Ethiopia; Ayalew et al., 2002). The consequences for amphibole fractionation in these systems is less clear, while peralkaline Ethiopian ignimbrites indicate a progressive decline of Dy/Yb with fractionation (Fig. 8), the currently limited partition coefficient data for amphibole in peralkaline rocks suggest that it may not fractionate Dy from Yb (Marshall et al., 2009). Further trace-element investigation of these systems is necessary to examine this discrepancy. Importantly, the appearance of amphibole in peralkaline systems is indicative of low water content in these magmas (Scaillet and MacDonald, 2001; Ayalew et al., 2002). These observations suggest that for high-silica rhyolites erupted at South Mountain (and in other large igneous provinces), the volatile content and peralkalinity may have significant control over the fractionating assemblages and resultant trace-element patterns. Consequently, the presence of amphibole does not necessarily require a high aH2O value, as discussed previously. However, without available exposures of the intermediate progenitor containing the amphibole, determination of the amphibole composition and, hence, the F contents, remains an unresolved aspect of our model.

Rhyolites from the South Mountain region of Pennsylvania record processes active in the lithosphere during the eruption of the late Vendian Catoctin Volcanic Province, part of the eastern Laurentian superplume (e.g., Puffer, 2002). The presence of large volumes of rhyolitic magmatism in large igneous provinces is generally attributed to the commencement of rifting (Pankhurst et al., 2000a; Marsh et al., 2001; Peate et al., 2005), and the abundant silicic volcanism associated with the Catoctin Volcanic Province along the paleo–continental margin is consistent with the successful rifting of Laurentia at this time.

The broad division of the South Mountain rhyolites into low- and high-Ti varieties mirrors similar classifications noted in other younger flood basalt provinces (e.g., Paraná; Garland et al., 1995). We suggest that this primary division of the rhyolites is related to distinct parental basaltic magmas derived from either a heterogeneous eastern Laurentian superplume, or from the interaction of these plume magmas with the continental lithosphere. These basaltic magmas subsequently evolved toward rhyolitic compositions through open-system fractionation. We have recognized two distinctive fractionation paths that are independent of this Ti subdivision, revealing significant variation in terms of the residual mineral assemblages. The dominant South Mountain rhyolite group exhibits major- and trace-element characteristics typical of substantial plagioclase removal. However, a smaller subset of the South Mountain rhyolites follows an alternate path that is consistent with the presence of amphibole in the source mush. We interpret these heterogeneous fractionation paths as proxies for the volatile content of the evolving magmas, and suggest that such techniques applied to rhyolites from other large igneous provinces may yield new constraints on the volatile content of large igneous province magmas.

We thank Tom Vogel for reading and commenting on an earlier version of this manuscript, Sheldon Turner for his help with drafting Figure 1, and Robert Smith for discussions about the Catoctin in Pennsylvania. The manuscript benefited from the comments of an anonymous reviewer and the editorial handling of R. Russo. This project was in part funded by a grant to Briggs by the Michigan State University Center for Undergraduate Research in Earth System Science (CURESS).

1GSA Data Repository Item 2010100, Table DR1, location information for South Mountain samples, is available at www.geosociety.org/pubs/ft2010.htm, or on request from editing@geosociety.org, Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301-9140, USA.